Quality Issues in Water Sampling, Sample Pre

Part III Analytical Strategies

9 Quality Issues in Water Sampling, Sample Pre-Treatment and Monitoring Sara Bogialli1, Stefano Polesello2 and Sara Valsecchi2 1

Department of Chemistry, University of Padua, Italy 2 Water Research Institute, IRSA-CNR, Italy

9.1 Introduction One of the most important aims of environmental monitoring is to provide spatial and temporal trends of concentrations of chemical substances in a certain matrix: this goal can be achieved only if quality and comparability of analytical results is ensured. There is a general consensus on the need for a quality assurance (QA) framework in the analytical laboratories which should establish their own quality control (QC) program. As regards the chemical monitoring under the European legislation, Commission Directive 2009/90/EC represents the legal basis for the performance of the analytical methods and gives technical specifications for chemical monitoring. Based on the requirements of this directive, the application of internal and external quality control measures, such as the use of blanks, standards, (certified) reference materials or regular participation in laboratory inter-comparison, is mandatory. According to this Directive (Art. 3) European Member States shall ensure that all methods of analysis, including laboratory, field and online methods, used for the purposes of chemical monitoring programmes carried out under Directive 2000/60/EC are validated and documented in accordance with EN ISO/IEC-17025 standard or other equivalent standards accepted at international level. In environmental monitoring the quality assurance program must include all the steps that are necessary to collect the samples in the field, transport to the laboratory and store them before the extraction and analysis. The QC tools for controlling the first steps of the analytical procedures are currently less developed than those related to the analytical steps [1].

Transformation Products of Emerging Contaminants in the Environment: Analysis, Processes, Occurrence, Effects and Risks, First Edition. Edited by Dimitra A. Lambropoulou and Leo M. L. Nollet. # 2014 John Wiley & Sons, Ltd. Published 2014 by John Wiley & Sons, Ltd.

284 Transformation Products of Emerging Contaminants in the Environment

ISO Standards 5667 series provide guidance on sampling techniques: two of them are devoted to the preservation and handling of water, sediment and sludge samples [2,3]. A guidance for the evaluation of the contribution of the sampling procedures to the total uncertainty has been published by the Eurachem network [4]. Recently, intercalibration exercises on sampling have been carried out in the European countries [5,6]. Contradictory results on the relevance of the sampling step to the total uncertainty have been shown. As an example, the conclusions of three on-site intercalibration campaigns carried out under the framework of the EU expert group on Chemical Monitoring and Analysis (CMA) [5] are that no clear effect of the contribution of the sampling to the total variance can be evidenced in the analysis of PAH and PBDE, while the contribution of the analytical steps is still prevailing at river concentrations. On the contrary, a French collaborative trial on lake sampling [6] showed that for a few metals (cobalt, nickel) there is a far more predominant (>90%) sampling uncertainty, as compared to the analytical uncertainty. For other metals, the initial blank tests demonstrated that the main contribution to the total uncertainty is the contamination of sampling systems. These exercises have been carried out with compounds for which certified standards and materials and validated methods of analysis are available. When, on the contrary, transformation products (TPs) have to be monitored, the main sources of uncertainty are the lack of validated analytical methods, certified reference standards and, in most cases, even the commercial standards. These deficiencies represent a hindrance in the application of the usual QA/QC protocols which must be adapted to the specific cases and substances. The aim of this chapter is to review the application of the QA/QC principles to the monitoring of TPs in the water bodies with a special focus on the fulfilment of regulatory requirements for compound identification.

9.2 Monitoring of Transformation Products in Water Bodies Though the introduction of stable and/or toxic TPs in risk assessment is mentioned in all relevant regulatory assessment schemes ([7] and references therein), the introduction of TPs into the regular and systematic water monitoring programs is still rare if we exclude the monitoring of stable metabolites of pesticides. Considering the main European lists of substances to be monitored in the waters, such as the list of priority substances (PS) defined by Environmental Quality Standard (EQS) Directive 2008/105/EC for surface and coastal waters and the list of substances in the Drinking Water Directive, none of them can be considered as TPs, even if some compounds, for example, alkylphenols, can be considered a TP because they can be generated by degradative processes in the treatment plants or in the aquatic environment. In the recent proposal for the revision of the Environmental Quality Standard Directive [8] the new list of PS includes perfluorooctanesulfonate (PFOS) which can be the final and persistent TP of perfluorooctane sulfonyl fluoride (POSF)-derived substances, generally known as PFOS-precursors. Apart from the aforementioned compounds, which are very diffused and already included in regulatory monitoring programs, most of the current works on TPs dealt with their formation in studies representing single chemical or biological transformation processes or specific environmental compartments. The importance of taking TPs into consideration in the regulatory assessment schemes is, however, widely recognized [7]. For example, according to the

Quality Issues in Water Sampling, Sample Pre-Treatment and Monitoring 285

European chemicals regulation REACH [9] “consideration should be given to whether the substance being assessed can be degraded to give stable and/or toxic degradation products. Where such degradation products can occur, the assessment should give due consideration to the properties (including toxic effects and bioaccumulation potential) of the products that might arise” [10]. Furthermore VICH guidelines for veterinary medicines [11] and the European Medicines Agency (EMEA) guideline for human medicines [12] in the higher tiers of the assessment procedures include degradation studies for the relevant environmental compartments in order to identify the most important TPs. However, most studies, and also regulatory assessment schemes, with the exception of the EU Pesticide Directive [13,14], which defines the concept of relevant metabolite, do not provide a more comprehensive assessment of which TP(s) might significantly contribute to the overall environmental risk posed by a given chemical or which, for that reason, might need to be included in environmental monitoring programs. General criteria for the choice of the TPs of ECs in monitoring programs are generally linked to the availability of stable standards. This criterion has been adopted in most existing monitoring programs around Europe and the USA. The first wide monitoring program on pharmaceuticals, hormones and other organic wastewater contaminants in US streams, carried out in 1999 and 2000, [15] included some stable pharmaceutical metabolites, such as metabolites of erythromycin (erythromycin-H2O), digoxin (digoxigenin), enalapril maleate (enaprilat), paroxetine, nicotine (cotinine), and caffeine (1,7-dimethylxanthine). A more recent survey of the occurrence of pharmaceuticals and endocrine disrupting compounds in US drinking waters [16] included metabolites of atorvastatin, fluoxetine, and simvastatin which were selected according to the availability of analytical standards. In fact this latter requirement was considered necessary to establish a recognized QA/QC program for monitoring. The same criterion of choice has been adopted, for example, in a recent monitoring of Dutch waters [17], which determined 17 common pharmaceuticals and 9 TPs, that were available as standards, in surface waters, pre-treated surface waters, groundwaters and drinking waters. The discovery of the “sewage epidemiology” [18] has given a new impulse to the monitoring of drugs and pharmaceutical metabolites and TPs in the output of treatment plants. These studies can give reliable results because both native and labeled standards of drugs and their metabolites are generally available, since they are widely used in clinical and pharmacokinetic studies. This novel approach was proposed to estimate drug consumption in a community by wastewater analysis [18–20]. The correspondence between the pattern of drug metabolites detected in wastewater and their human excretion profile is one of the main assumptions on which the present method is based [20]. This assumption has been recently verified by a study [21] which concluded that the profiles of cocaine metabolites in wastewater matched those in human urine reported in the literature, suggesting that measures in wastewater reflect the real human excretion and that wastewater analysis is suitable for assessing drug consumption, provided that a proper storage of samples has been adopted in order to avoid the degradation of the analytes that should be used for drug consumption estimation [21]. The key point for assuring the quality of a monitoring study which involves TPs is the possibility to control the degradation processes active in the sample after the sampling. If we want to capture a picture of a sample at the sampling moment, assuring the sample representativeness, it is necessary to avoid the degradation of the parent compound, generating TPs


which should be considered sample artifacts. This issue, which is of general interest but assumes particular relevance in TP monitoring, will be thoroughly discussed in the following section. In the current monitoring plans for river and drinking water, the approach based on a target list of stable TPs neglects numerous organic contaminants, including compounds with potential harmful effects. Screening analyses should be established as principal tools to improve and complement the substance spectra for monitoring purposes. Kern et al. [22] developed an analytical procedure for screening a wide target list of possibly present TPs, in complex environmental samples in the absence of commercially available reference standards. This procedure, also called “suspect screening” [23], is based on a solid phase extraction (SPE) with broad enrichment efficiency, followed by liquid chromatography and tandem mass spectrometry (Linear Trap Quadrupole-Orbitrap) with high mass resolution and accuracy. The target list of plausible TPs (about 1800) has been assembled using the University of Minnesota Pathway Prediction System (UM-PPS) for the computer-aided prediction of products of microbial metabolism and an extensive search for TPs reported in the scientific literature. The identification of “suspect” TPs consisted of extracting the exact mass from the chromatogram, selecting peaks of sufficient intensity, checking the plausibility of the retention time, and interpreting mass spectra. The procedure was used to screen for TPs of 52 pesticides, biocides, and pharmaceuticals in seven representative surface water samples from different regions in Switzerland. Notwithstanding the authors screened a very wide list, only 19 TPs were detected and identified in real samples, including both some well-known and commonly detected TPs and some rarely reported ones (e.g., biotransformation products of the pharmaceuticals venlafaxine and verapamil, or of the pesticide azoxystrobin) [22]. The alternative approach is the so-called “non-target” or unknown screening which, unlike the suspects screening, starts without any a priori information on the compounds to be detected [23]. The introduction of high resolution mass analysers (HRMS), such as time of flight (TOF) and Orbitrap, hyphenated to high resolution chromatographic techniques (gas chromatography, GC, and liquid chromatography, LC), with commercial software and database available for the compound identification [23–26], allowed the application of non-target screening of ECs, including their TPs, in environmental investigative monitoring. Some examples of the identification of TPs by non-target and suspect screening in different aquatic matrices have been recently reported: a rapid automated LC–QTOFMS screening of Spanish rivers and wastewaters [27] allowed the identification of TPs of acetaminophen and azithromycin, based on the combination of an accurate-mass database with the predicted fragmentation pathways of the parent species. The non-target LC-QTOF screening method has also been applied to freshwater sediment samples after exhaustive Soxhlet extraction [28] and allowed the determination of pharmaceuticals, together with their synthetic intermediates and TPs, discharged by pharmaceutical industries. Multidimensional gas chromatography coupled to high resolution mass spectrometry (GCxGC-TOF-MS) has also been employed for the non-target screening monitoring of semivolatile and non-polar organic contaminants and their TPs in the Henares river basin (central Spain), a representative basin with a continental Mediterranean climate [29]: TPs detected and identified in this monitoring campaign were galaxolide-lactone, 4-amino musk xylene and 9,10-anthracenodione. Notwithstanding the positive and promising aforementioned examples, the application of LC-MS techniques for non-target screening of environmental contaminants in extensive and regulatory monitoring is still rare [28]. One of the reasons is the high cost of the HRMS


equipment capable of determining the accurate mass of detected compounds. Moreover, well-known difficulties associated with the structural elucidation of a true unknown from its empirical formula [30], a rather limited availability of comprehensive mass spectral libraries [31], and the reliability of commercial identification software working in non-target and post-target analysis [26] represent important obstacles for a more widespread application of LC-HRMS techniques for identification purposes. The reliability of the identification of TPs in water monitoring, which is the main quality issue in monitoring programs involving screening techniques, is widely discussed in Section 9.4.

9.3 Sample Representativeness and Stability Issues Obtaining representative samples of the aqueous system being investigated is of primary importance for a relevant description of the studied environment. A representative sample is one that typifies (“represents”) in time and space that part of the aqueous system to be studied, and is delineated by the objectives and scope of the study. Representativeness of the samples is a very important issue in monitoring environmental pollutants in general, but it is even more important for the TPs which are not directly released into the environment but that are formed during the residence time of the precursors in the environmental matrices. In order to collect and analyze a representative sample that will yield the information required, artifacts during sampling and during storage of the collected water samples must be minimized. Possible formation of artifacts during water sampling and sample storage has been already widely addressed by several authors for the most common pollutants (see e.g., [1]) and the same quality criteria can be apply to the monitoring of TPs, including the use of quality assurance/quality control (QA/QC) samples, which must be taken, prepared and analyzed in the same manner as the environmental samples. During the field operation, some sources of contamination error associated with field operations and shipping can be taken account of by the use of field blanks, which are prepared in the field by filling the appropriate sample container with certified distilled/deionized water or clean sand or soil. Field blanks are collected to check for cross-contamination that might occur during sample collection. Additionally, if volatile compounds are a concern trip blanks can be required. They are prepared in the laboratory, taken to the field, and then brought back to the laboratory with the environmental samples and analyzed. Trip blanks consist of distilled/deionized water or certified clean sand or soil handled, transported, and analyzed in the same manner as the other samples acquired. Field blanks and trip blanks are useful to control volatilization of pollutants as well as absorption of contaminants from air. For instance, in an Italian laboratory it was observed that eluents for LC-MS (methanol and ammonium acetate buffer) absorbed caffeine and cotinine, which is a metabolite of nicotine, from the laboratory air in a few days (Valsecchi, unpublished data). In the case of TPs an “ad hoc” matrix-spiked sample could be used in order to take account of the possible degradation of the parent compounds during shipping and storage: the sample can be spiked in the field with stable isotope labeled internal standard (SIL-IS), not only for isotopic dilution correction, but also in order to study the degradation and the possible formation of TPs, provided that they retain the labeled moiety in their structures. Ideally, environmental samples should be analyzed immediately after sample collection. However, it is often necessary to store them before processing because of logistic constraints.


Sample containers used in transport and storage can lead to contamination or changes in the relevant chemical properties of the sample but some precautions, depending on the nature of the contaminants to be analyzed, can be taken to avoid this drawback. Plastic materials, except polytetrafluoroethylene (PTFE), must not be used for the samples which are to be analyzed for hydrophobic organic contaminants; these samples must be stored in glass, PTFE or stainless steel containers. On the contrary, plastic should be used to store water samples for more polar ECs, such as pharmaceuticals and PCP, as these analytes can potentially be adsorbed on glass surfaces. However, care must be taken because some TPs can derive from the degradation of additives and plasticizers [32]. Even nonylphenol, the degradation product of the non-ionic surfactants nonylphenol polyethoxylates, can be leached out from the plastic material present in the laboratory [33,34]. The type of containers should always be selected after consulting the laboratory performing the chemical analyses, or the containers should be supplied by the laboratory. Moreover, depending on the parameter to be determined, specific conditioning and/or cleaning of sample containers prior to use may be required. For example in the case of perfluorinated alkyl acids all glassware must be meticulously cleaned by rinsing with reagent water or solvent to avoid sample contamination [35]. When specific information on the appropriate container is lacking, such as for most TPs, silanized glass is recommended [36]. The aforementioned quality issues are common to all compounds; specificity of quality issues for TPs is mainly related to the need to minimize unavoidable chemical or biological changes that occur in samples after collection. As for the parent pollutants, the TPs can degrade and disappear from the collected samples because of the biotic and abiotic processes. Moreover, TPs can be formed by the degradation of active parent compounds taking place in the collected samples. Therefore stabilization and preservation of TPs in collected samples are crucial, otherwise significant under- or overestimation of the actual amount of residue originally present at the sampling moment could result. Storage begins when the samples are taken. Because the first few hours after sampling are the most critical for changes to occur in the sample, preservation steps should be taken, where possible, immediately upon sample collection. The methods of preservation are, more generally, intended to (i) retard biological action, (ii) retard hydrolysis of chemical compounds and complexes, (iii) reduce volatility of constituents, and (iv) reduce absorption effects. The most common preservation methods are usually limited to pH control, chemical addition, refrigeration, and freezing. Usually, no particular preservation techniques have been employed for the monitoring programs which included TPs, and the sampling and storage procedures are similar to those assessed for stable organic compounds. Acidification is commonly used to stabilize collected samples due to its ability to limit microbial activity [37]. For example, acidification to pH 3 with 3.5 M sulfuric acid has been used to inhibit microbial activity [38] for the determination of nine antiviral drugs (acyclovir, abacavir, lamivudine, nevirapine, oseltamivir, penciclovir, ribavirin, stavudine, zidovudine) and one active metabolite (oseltamivir carboxylate) in raw and treated wastewater. In order to improve stability in wastewater composite sampling, acidification is the optimal preservation methodology which is carried out by adding small volumes of acid over regular intervals during the composite sampling time. However, it has to be remembered that acidification might influence partitioning of certain chemicals between aqueous solution and (suspended) solids. In this case both aqueous solution and solids have to be analyzed for the presence of the studied compounds [36].


Acidification is not always effective to preserve degradation products, as in the case of TPs of a widely used explosive, hexahydro-1,3,5-trinitro-1,3,5-triazine (RDX), measured in groundwaters impacted by US military sites [39]. Storing samples at 4 C increased significantly the stability of some of its TPs but did not completely prevent degradation of MEDINA, the most unstable TP of RDX. Mild alkaline conditions, obtained by addition of sea salts, were necessary for long term preservation in aqueous solution, thus allowing accurate detection after sampling and transport from the field to the laboratory. A further overlooked example is the artifactual formation of monomethyl mercury (MMHg) due to the decomposition of dimethyl mercury (DMHg) under acidic conditions, the most commonly used preservation techniques employed for MMHg samples [40]. This facile conversion of DMHg to MMHg at low pH needs an alternative preservation method to acidification for samples to be analyzed for MMHg when appreciable DMHg concentrations are possible. Biological and chemical degradation can also be interrupted by adding chemical preservatives: drinking water samples are preserved with sodium azide after having quenched the disinfection residuals with ascorbic acid [16]. Preservation of chloro-s-triazine herbicides and their TPs in finished chlorinated drinking water samples was ensured by buffering and dechlorinating samples with 20 mM ammonium acetate as well as adding sodium omadine as an antimicrobial reagent [41]. Cooling between þ4 and 0 C (for a few days) and freezing at 20 C are the most common preservation techniques for organic samples that cannot be analyzed soon after sampling [42]. The allowable storage period for organic contaminant samples may vary from a few days to several months, depending on the stability and chemical and physical properties (e.g., vapor pressure) of the analytes. Therefore, no general recommendations can be given, but the effect of storage conditions on the measured concentrations should be tested in advance. In general, if samples are analyzed within 24 h and stored in the dark at 1–5 C, biological activity in the sample is not arrested but measured concentrations are not significantly impacted. Using storage temperatures below 20 C may allow storage of samples for longer time periods. However, freezing is not appropriate for volatile components. It is also necessary to remove suspended matter, algae and other micro-organisms by filtering the sample on 0.22 mm filters before freezing to avoid changes in dissolved concentrations of substances caused by precipitation or cell disruption. In fact the risk of precipitation of, for example, calcium carbonate at low temperatures, and other processes such as co-precipitation and colloid coagulation during freezing should be considered. Also, interruption of the native microbial activity obtained by freezing can be reversible. It is known that when bacteria are frozen and then thawed, their survival is dependent on many parameters, such as cooling and warming rates and holding temperature [43]. In addition to other preservation methodologies the removal of suspended wastewater particulates by filtering samples before the storage may improve the stability of analytes as it prevents potential adsorption onto particulates [36]. Sampling protocols often prescribe minimal exposure of the sample to light. This is a proper practice, although photophysical degradation during the actual sampling will generally be negligible. Even the photolysis half lives of compounds that are sensitive toward photodegradation are of the order of days [44]. Specific stability tests are now implemented in EU-wide monitoring activities for EC but this monitoring program at the moment does not include TPs [45]. To assure the quality of monitoring data of TPs in environmental matrices uniform ways of sample collection and


handling specifically studied for TPs have still to be comprehensively addressed. Preservation methods commonly used for non-TPs chemicals have to be tested in order to evaluate their suitability for TPs, but few studies, up to now, have been conducted to test the stability of TPs in environmental matrices. Particular attention has been devoted to the stability of metabolic residues of illicit drugs and their active parent compounds in wastewater in order to apply the sewage epidemiology approach. For these studies it is mandatory to prevent the analyte degradation since the pattern of illicit drugs metabolites detected in wastewater must reflect the human excretion profile in urine in order to back-calculate drugs consumption in a community by wastewater concentrations. Degradation or formation of illicit drugs and their metabolites during the storage of wastewater samples have been assessed by several experiments ([46] and references therein). Castiglioni et al. [21] studied the stability of cocaine and its metabolites in untreated wastewaters after a freeze-and-thaw cycle and after one and three storage days at 4 C. The concentrations of some analytes declined substantially with a concurrent increase in several demethylated TPs after three days of refrigerated storage (Figure 9.1). On the contrary, after freeze-and-thaw cycles all analytes were stable, except cocaine and ecgonine methyl ester which showed a slight decrease (Figure 9.1). Sometime discrepancies, probably due to differences in the composition of the wastewaters, can be found in the literature, as in the case of amphetamine: some studies [46–48] observed a high stability at 4 C for 72 h, while others [36,49] claimed a clear upsurge of AMP within 12 to 72 h at 2 and 19 C. Although large differences exist in the set-up of the various experiments, their main findings are that the best storage protocol provides for samples to be frozen immediately after collection and kept frozen until analysis. Baker and Kasprzyk-Horden [36] presented a comprehensive review of the stability of illicit drugs, pharmaceuticals and their metabolites and some TPs in wastewater. They

Figure 9.1 Stability of the compounds in wastewater immediately after drug spiking (T0), after a freeze-and-thaw cycle (F/T), and after 1 d (T1) and 3 d (T3) of storage at 4 C. Benzoylecgonine (BE), norbenzoylecgonine (NorBE), Cocaine (COC), norcocaine (NorCO), cocaethylene (CE), ecgonine (ECG), ecgonine methyl ester (EME), anhydroecgonine (AE), anhydroecgonine methyl ester (AEME) (reproduced with permission from Ref. [21] Copyright (2011) Elesevier Ltd.).


evaluated the stability of compounds in the environmental matrix after short term (72 h) and long term (weeks) storage. The short term stability study is of considerable importance as certain compounds may degrade significantly over a period of only several hours, a factor that must be considered if samples must be shipped, or a 24-h composite sampler is used or, even more, if a passive sampling campaign is carried out. Furthermore, logistic constraints often oblige operators to store samples for even a few weeks before the analysis. In short term stability studies [36] pH resulted to be the most important factor, compared to other tested factors, such as filtration and temperature, with a significant stability improvement after acidification. After 24 h, nearly half of all the compounds monitored showed a stability change (>15%) at neutral pH, while acidified wastewater (pH 2) led to only 8% degraded analytes, regardless of the temperature. Nevertheless, the positive effect of acidification depends also on the compound characteristics: for instance, cotinine, a TP of nicotine, showed high stability in filtered wastewater over a 72-h period at pH 2 while, on the contrary, other compounds, such as benzodiazepines, temazepam and oxazepam, showed lower stability in acidic wastewaters than in the same samples at pH 7.4. The storage temperature of samples was also found to affect stability, although to a lesser extent than pH. After three days in raw (unfiltered) wastewater at temperatures of 2 and 19 C the percentages of compounds reporting a significant stability change were 54 and 69%, respectively. The influence of suspended solids in the samples was found to decrease the concentrations of some analytes in the aqueous matrix: 1,7-dimethylxanthine, a TP of caffeine, showed a significant concentration change in raw (unfiltered) wastewater at pH 7.4 after 12 h for both the investigated temperatures (2 and 19 C), while good stability for 1,7-dimethylxanthine was found, under the same conditions, in filtered wastewater. For long term storage, freezing samples is the best choice to achieve the preservation of active parent compounds, metabolites and TPs [46,47,50], with the drawbacks of needing large freezers and time to thaw the samples before analysis. Gonzales-Marino et al. [48] and Baker and Kasprzyk-Horden [36] investigated the storage of drugs, pharmaceuticals and their metabolites and some TPs stored at 20 C for up to 6 weeks, after their extraction onto Oasis HLB or MCX cartridges, and no degradation of compounds was experienced. Therefore, the storage of SPE cartridges is an ideal alternative to the freezing of large volumes of aqueous samples, also allowing easy shipping of the samples. Another issue to be considered in the stability framework is that conjugated metabolites have the potential in field samples to undergo deconjugation and transform back to the original form of the parent drug, which can explain the observed increase in the concentration of morphine with a concurrent decrease in the stability of morphine-3b-glucuronide [36]. Deconjugation processes are usually quick, starting in a few hours and completing within a few days. In the case of pharmaceuticals and drugs, if the associated TPs are conjugate metabolites there is no absolute way of discriminating among the TPs that were originally present and those formed during hydrolysis of conjugate metabolites, and overestimation of actual environmental TP concentrations is attained. On the other hand, risk assessment based only on the environmental concentrations of the non-coniugated TP could underestimate the true risk of TP compounds in the environment [51]. If TPs have to be analyzed in soil or sediment samples, preservation techniques for assuring sample stability are different. Soil and sediment are not as homogeneous matrices as water but, similarly, in order to avoid undesirable bio-transformation of analytes in the soil/sediment after sampling, microbial and enzymatic activities must be eliminated [52]. Whenever possible, freezing to preserve soil or sediment samples should be avoided because it can change the


grain size distribution of the sediment. The most employed technique used to store sediment and soil samples is lyophilization, which is not a sterilization process. If any bacteria are present in the product and if it is lyophilized, when the product is reconstituted with a solvent or water most times the bacteria will grow back. A review on the methods to actually sterilize soil or inhibit microbial activity in soil has been published by Trevors [53]: the objective of soil sterilization is to destroy or eliminate microbial cells while, at the same time, not significantly altering the chemical and physical characteristics of the soil [54]. Nevertheless, most of these methods requires thermal treatments (autoclaving or microwaving) and should induce degradation of labile compounds: on the other hand the most efficient method for sterilization is g-irradiation, but this process can start a radical degradation process by producing hydroxyl radicals. Alternatively, chemical preservatives, such as mercuric chloride or sodium azide, can be added to the solid sample. To conclude, in general all storage methods will affect the sample to some extent, and the choice of preservation technique depends mainly on the objective of the sample collection. No recommendations can be given for a universal preservation or storage technique. An ideal technique for one class of compounds may interfere with the analysis of other substances. To overcome this problem, a sufficient sample volume should be collected to allow specific preservation or storage techniques for each specific group of analytes.

9.4 Identification of Transformation Products and Legislative Requirements 9.4.1 Basic Principles The analytical challenge related to the identification of TPs of ECs is affected by the availability of the certified standards. While the unquestionable identification of organic contaminants requires a reference standard, generally the combination of mass spectrometric (MS) and NMR detection is used to propose a reliable assignment of an unknown compound. Anyway, direct NMR analysis of water samples is not feasible without several isolation steps, while liquid chromatography (LC)-NMR is currently expensive and not sufficiently sensitive. Thus, the monitoring of TPs of ECs in the environment is nowadays performed mainly by chromatographic separation followed by MS detection. The general principles requested by International Agencies for confirming the presence of an organic compound are well established and they are valid for both standard and standardless determinations. The main themes have been summarized in one of the most comprehensive and exhaustive guidelines on confirmatory assays, the European Commission Decision 2002/657/EC [55] concerning residues analysis of regulated veterinary drugs (with relative analytical standards available) in food. This guideline has introduced new definitions and criteria for confirmatory analysis, where chromatographic separation coupled to MS plays the leading role. In particular, analysis with LC-MS has become increasingly important, especially for chemicals in aquatic environments, as this technique is suited to the measurement of polar and thermolabile compounds. However, whereas LC-MS seems to be a fitfor-purpose detection system for monitoring TPs, since TPs are often more polar than the parent chemicals, nevertheless, the lack of a standardized database of mass spectra is a great drawback. For this reason, GC-MS analysis remains attractive because of large and reliable libraries based on the EI ionization technique, suitable for identifying non-target compounds.

Quality Issues in Water Sampling, Sample Pre-Treatment and Monitoring 293 Table 9.1 Relationship between mass ions and IPs with examples of IPs obtained with different MS techniques [55]. MS Technique

Identification Points Earned Per Ion

Examples Techniques

Low resolution mass spectrometry Low resolution -MSn transition products

1.0 1.5

High resolution MS

2.0

High resolution MSn transition products

2.5

a

GC-MS (EI or CI), LC-MS GC-MSn, LC-MSn

LC-HRMS, GC-HRMSa LC-HRMS, GC-HRMSa

Number of ions

IPs

N

n

1 precursor and 2 daughters 1 precursor, 1 daughter and 2 granddaughters N

4

1 precursor and 1 daughter

5 2n 4.5

Resolution greater than 10 000 for the entire mass range at 10% valley.

According to 2002/657/EC Decision, an unambiguous determination is based on the socalled identification points (IPs), that rely both on the resolution power of the different mass analyzers and on the multiple-stage of MS selection. Low resolution MS concurs with one IP for each ion and 1.5 IP for each transition from precursor ion>daughter ion, while high resolution MS (HRMS) with two IP for each ion and 2.5 IP for each transition. Triple quadrupole (QqQ), ion trap (IT) or linear IT are considered low resolution mass analyzers, whereas high resolution mass spectrometers (HRMS) are those with a resolution typically greater than 10 000 for the entire mass range at 10% valley (20 000 FWHM, full width at half maximum). According to the strictest criteria for forbidden substances, four IPs are necessary to confirm the presence of a target compound. If a certified standard is available, the permitted tolerances of both relative retention times (0.5% for GC and 2.5% for LC) and intensities of the detected ions have also been set. Table 9.1 reports some examples of IPs obtained with different MS techniques, while Table 9.2 shows maximum permitted tolerances for relative ion intensities, as requested by 2002/657/EC. Anyway, according to this definition, FT ion cyclotron resonance (FTICR) and OrbitrapTM are always considered HRMS instruments, while time of flight (TOF) and the hybrid Q-TOF remain borderline systems, depending on the mass range of the analytes, whilst they are currently the most widespread MS techniques proposed for the twofold screening–confirmatory purpose [56–59]. Chromatographic separation followed by the HRMS-based full-scan method is an attractive option, because the possibility to obtain exact mass signals and isotopic patterns with high accuracy for both molecular and fragment ions often allows a selective detection and characterization of target and non-target residues in complex matrices. An inadequate resolving power of the MS can result in false negative [60] or false positive results, if ions of the analytes cannot be separated from isobaric co-eluting ions related to the matrix compounds. Moreover, HRMS full scan methods do not need previous compound-specific instrument tuning, and they offer the possibility to detect non-“a priori” selected analytes, so that data can be evaluated retrospectively.

294 Transformation Products of Emerging Contaminants in the Environment Table 9.2 Maximum permitted tolerances for relative ion intensities using a range of mass spectrometric techniques [55]. Relative intensity (% of base peak) >50 >20–50 >10–20 10

EI-GC-MS (relative) (%)

CI-GC-MS, GC-MSn LC-MS, LC-MSn (relative) (%)

10 15 20 50

20 25 30 50

The alternative use of mass accuracy (expressed in mDa) instead of mass resolution has been considered in a future revision of the European guideline for a reliable confirmation with HRMS technologies [61], even considering that resolution should be expressed for a specific mass. 9.4.2 Quality Control in Qualitative Analysis of TPs According to the IPs criteria reported in Table 9.1, the identification of a compound in MS is always assured by selecting two ions in HRMS or two precursor ion–daughter ion transitions in low resolution MS, with the correct ion ratios (Table 9.2) and a signal to noise ratio S/ N 3. The ion/transition with the worst S/N is used for confirmation purposes and affects the limit of detection (LOD), while quantitative analysis is generally carried out from the best one, in order to increase the precision of the results. As regards the significance of the fragmentation pattern, it is widely accepted that multiple fragmentations should not occur in the same part of the molecule and that general neutral losses, like water and carbon dioxide, do not provide an adequate selectivity. However, it is not so easy to comply with these indications and several examples of not specific or redundant transitions are common in environmental analysis. For example, losses of water and ammonia are generally experienced for tetracycline antibiotics [62], loss of carboxylic acid is typical of perfluorinated compounds [63], while the pesticide pentachlorophenol exhibits only the loss of hydrochloric acid [64]. This matter is not enforceable when the fragmentation pattern can be compared with the relative standard one, but it becomes crucial for standardless detection without further support of NMR analysis. Conversely, neutral loss scan can be helpfully performed for a survey of TPs [65,66]. The presence of halogens or sulfur [66,67] in a TP structure can assist data interpretation because of their peculiar isotopic pattern. This property is very helpful for identifying TPs produced by a chlorination process in water treatment plants (WTPs) [7] or generated from degradations of polybrominated diphenyl ethers (PBDEs). When HRMS is used for screening and confirmation of compounds with reference standards, the positive candidate has to comply with general queries, that is, matching retention times with acceptable tolerance (see previous section) and a signal or S/N higher than a certain threshold. The last query is also used for non-target screening, together with the accuracy expressed in mDa or ppm. As a consequence, a huge number of possible positive results is generated and the data interpretation remains the real bottleneck of the analysis. Isotopic pattern matching has to be considered to reduce the candidate list coming from the mass accuracy and false positives.


Most observations can be summarized in seven rules useful for interpreting molecular ion spectra and reducing results in the “generating formula” step, as reported for the metabolomic approach [68]:

Restriction on the number of elements Adherence to the LEWIS and SENIOR chemical rules Isotopic patterns Ratio of hydrogens/carbons Ratio of additional elements such as nitrogen, oxygen, phosphorus, and sulfur to carbon Adherence to the element ratio probabilities Inclusion of trimethylsilyl groups (for derivatized compounds in GC-MS analysis).

The use of databases, libraries or software devoted to suggesting molecular structures on the basis of the mass defect and isotopic pattern reduce this laborious process, but results have to be carefully checked. A certain number of databases and libraries for the prediction of substances structure, other than those furnished by manufacturers, are accessible: ChemSpider (www.chemspider.com), DAIOS (http://www.daios-online.de/daios/), KNApSAcK (http:// kanaya.naist.jp/knapsack_jsp/top.html), Massbank (http://www.massbank.jp/index.html), Merck Index (https://themerckindex.cambridgesoft.com/), Metlin (http://metlin.scripps.edu/), NIST (http://webbook.nist.gov/chemistry/form-ser.html), PubChem (http://pubchem.ncbi.nlm.nih. gov/), SDBS (http://riodb01.ibase.aist.go.jp/sdbs/cgi-bin/cre_index.cgi) are examples of free databases. Some information about biodegradation routes and compounds are shared by the University of Minnesota (University of Minnesota Pathway Prediction System, UM-PPS available at http://umbbd.ethz.ch/index.html). Another chance to plan a non-target analysis has to be considered for those substances that exhibit a fragment ion typical of a class of contaminants. In this case, a “data-dependent acquisition mode” can supply molecular ion and daughter ion in HRMS of a TP during the same chromatographic run, achieving the IPs necessary for the correct identification. HRMS in multiple-stage fragmentation mode is needed for the identification of unknown substances, but a high accuracy of measures (range of mass error 2–5 ppm) has to be assured. Moreover, ion fragments generated from an EC parent compound can be used to predict possible TPs for screening purposes, since the bonds prone to be cleaved are the same involved in the MS fragmentation with soft ionization as those that might be broken in the reaction to form the TPs [69]. However, the mass accuracy for non-target methods (standardless and/or unknown and/or screening analysis), depends on several factors, such as the analyte concentration, type of matrix and sample preparation. In fact, a generic sample preparation and multiresidue procedures can affect the accuracy and selectivity of the mass measurement, increasing the amount of interfering endogenous compounds. Another sort of “matrix effect” can be experienced if internal calibration procedures are employed: in this case a quite high concentration of the calibrant solution can reduce the analyte’s response and accuracy. Blank samples analysis, even by background subtraction tools, is always necessary to avoid a tedious characterization of a peak derived from impurities or produced by analytical procedures [70]. In a similar way, a comparative analysis of influent and effluent water samples could be performed with the aim to isolate only the “odd peaks” in the processed waters [71]. Finally, in HRMS analysis, LODs should not be assessed only by S/N of the chromatographic peaks, because the huge capabilities of HRMS can support the reconstruction of


highly selective signals of target compounds with a virtually infinite S/N [72]. Thus, an acceptable accuracy of the related mass spectra should be taken into account for the estimation of LODs. 9.4.3 Applications Examples of identification of TPs belong to two groups: (i) analysis of compounds with available standards, and (ii) tentative characterization of compounds arising from non-target screening without available standards. The first approach guarantees the undoubted confirmation of a contaminant defined emergent, and it could be performed with low resolution tandem MS systems or with full scan-based methods with HRMS instrumentation, according to criteria set by the European guideline [55]. The last choice is limited by the creation of databases as large as possible, with retention times and accurate HR mass spectra of both molecular and fragment ions related to organic compounds. Confirmatory analysis of TPs in waters, by conventional GC or LC separation coupled to QqQ or IT mass analyzers has been extensively reviewed [73–75]. Conversely, some examples of qualitative analysis of ECs and TPs, carried out with HRMS systems, are presented here in detail with the intent of emphasizing some quality control issues typical of different identification protocols. An in-house database with approximately 400 pharmaceuticals and pesticides [27,76] was used for the identification of several compounds in wastewater and river waters through MS and MS/MS full scan with a LC-QTOF-MS system. Many more substances (about 3000) were included in a library for wastewater analysis with an OrbitrapTM instrument [70]. In this case, other relevant signals, not identified by the in-house library, were further processed with the manufacturer’s software to generate a number of molecular formulae, which were then, with other web-available databases, assigned a possible structure that was finally confirmed with HRMS fragmentation. This approach in four steps is followed by several researchers [22,23,26,77] with different LC-HRMS systems, but it is time-consuming both for generating libraries and interpreting all relevant MS signals. Anyway, this sequence with final confirmation through mass fragmentation is the highest degree of reliability that can be reached for standardless analysis of TPs. The screening of TPs of 52 pesticides, biocides, and pharmaceuticals in surface water samples from Switzerland was performed with an in-house library, assembled on the basis of the UM-PPS for the computer-aided prediction of products of microbial metabolism and the literature [22]. Comparison with blank samples, signal threshold, mass error