- Email: [email protected]
Occurrence of eight household micropollutants in urban wastewater and their fate in a wastewater treatment plant. Statistical evaluation
Science of the Total Environment 481 (2014) 459–468
Contents lists available at ScienceDirect
Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv
Occurrence of eight household micropollutants in urban wastewater and their fate in a wastewater treatment plant. Statistical evaluation Laure Pasquini a,⁎, Jean-François Munoz b, Marie-Noëlle Pons c, Jacques Yvon a, Xavier Dauchy b, Xavier France d, Nang Dinh Le c, Christian France-Lanord e, Tatiana Görner a a
Laboratoire Interdisciplinaire des Environnements Continentaux, CNRS, Université de Lorraine, 15 Avenue du Charmois, 54501 Vandœuvre-lès-Nancy cedex, France Laboratoire d'Hydrologie de Nancy, ANSES, 40 rue Lionnois, 54000 Nancy, France Laboratoire Réactions et Génie des Procédés, CNRS, Université de Lorraine, 1 rue Grandville, 54001 Nancy cedex, France d GEMCEA, 149 rue Gabriel Péri, 54500 Vandœuvre-lès-Nancy, France e Centre de Recherches Pétrographiques et Géochimiques, CNRS, Université de Lorraine, 15 rue Notre Dame des Pauvres, 54501 Vandœuvre-lès-Nancy cedex, France b c
H I G H L I G H T S • • • •
Presence of eight household micropollutants in two different urban catchments was assessed. Concentration levels of the target compounds in sewage influent, treated wastewater and sludge were compared. The efficiency of a conventional activated sludge process was assessed. Statistical analyses have related the micropollution and macropollution in the WWTP.
a r t i c l e
i n f o
Article history: Received 30 December 2013 Received in revised form 14 February 2014 Accepted 15 February 2014 Available online 12 March 2014 Keywords: Household micropollutants Urban wastewater Wastewater treatment plant Sludge Statistical analysis Modeling
a b s t r a c t The occurrence in urban wastewater of eight micropollutants (erythromycin, ibuprofen, 4-nonylphenol (4-NP), ofloxacin, sucralose, triclosan, perfluorooctanoic acid (PFOA) and perfluorooctane sulfonate (PFOS)) originating from household activities and their fate in a biological wastewater treatment plant (WWTP) were investigated. Their concentrations were assessed in the liquid and solid phases (sewage particulate matter and wasted activated sludge (WAS)) by liquid chromatography–tandem mass spectrometry. The analysis of sewage from two different urban catchments connected to the WWTP showed a specific use of ofloxacin in the mixed catchment due to the presence of a hospital, and higher concentrations of sucralose in the residential area. The WWTP process removed over 90% of ibuprofen and triclosan from wastewater, while only 25% of ofloxacin was eliminated. Erythromycin, sucralose and PFOA were not removed from wastewater, the influent and effluent concentrations remaining at about 0.7 μg/L, 3 μg/L and 10 ng/L respectively. The behavior of PFOS and 4-nonylphenol was singular, as concentrations were higher at the WWTP outlet than at its inlet. This was probably related to the degradation of some of their precursors (such as alkylphenol ethoxylates and polyfluorinated compounds resulting in 4-NP and PFOS, respectively) during biological treatment. 4-NP, ofloxacin, triclosan and perfluorinated compounds were found adsorbed on WAS (from 5 ng/kg for PFOA to 1.0 mg/kg for triclosan). The statistical methods (principal component analysis and multiple linear regressions) were applied to examine relationships among the concentrations of micropollutants and macropollutants (COD, ammonium, turbidity) entering and leaving the WWTP. A strong relationship with ammonium indicated that some micropollutants enter wastewater via human urine. A statistical analysis of WWTP operation gave a model for estimating micropollutant output from the WWTP based on a measurement of macropollution parameters. © 2014 Elsevier B.V. All rights reserved.
1. Introduction
⁎ Corresponding author. Tel.: +33 3 83 59 62 64; fax: +33 3 83 59 62 55. E-mail address: [email protected] (L. Pasquini).
http://dx.doi.org/10.1016/j.scitotenv.2014.02.075 0048-9697/© 2014 Elsevier B.V. All rights reserved.
In the last decades, the use of pharmaceuticals and various chemicals present in many products used in our everyday life has become a source of environmental pollution transported by urban wastewater. Even though removal mechanisms for conventional macropollution (organic
460
L. Pasquini et al. / Science of the Total Environment 481 (2014) 459–468
matter, nitrogen and phosphorus) in wastewater treatment plants (WWTPs) are understood and effective, this is not the case for most micropollutants (Kümmerer, 2001; Radjenović et al., 2009; Suarez et al., 2010; Richardson and Ternes, 2011). The removal of compounds such as pharmaceuticals, personal care products or detergents is usually not checked by plant managers or authorities because few or no regulations exist concerning their release into water bodies. The scientific interest in the environmental impact of these molecules, their behavior in WWTPs, and their occurrence in water bodies first emerged some ten years ago as showed in the literature review by Pasquini et al. (2013). Many of them are thought to be a possible threat to environmental health and safety. Our research focused on eight micropollutants originating from different activities of our everyday life: erythromycin and ofloxacin (antibiotics), ibuprofen (anti-inflammatory drug), triclosan (biocide), 4-nonylphenol (4-NP is a mixture of branched isomers resulting from detergent degradation), sucralose (sweetener), and two perfluoroalkyl acids (PFAAs) (Buck et al., 2011)—perfluorooctanoic acid (PFOA) and perfluorooctane sulfonate (PFOS). PFOA and PFOS are employed in industrial and domestic products as floor cleaning agents and paints, but also in the manufacture of emulsifying agents, non-adhesive surfacing and waterproofing agent for textiles, leather and food packaging (Prevedouros et al., 2006). The target compounds were chosen according to their: i) occurrence in domestic use, ii) physico-chemical properties in order to observe their possible partitioning between water and sludge in the WWTP. Erythromycin, 4-NP, triclosan and PFAAs which are hydrophobic compounds, and ofloxacin, are known to adsorb on sludge during biological treatment (Lindström et al., 2002; Picó and Andreu, 2007; Ochoa-Herrera and Sierra-Alvarez, 2008; Soares et al., 2008; Lillenberg et al., 2009). Ibuprofen and sucralose are hydrophilic compounds that should mostly remain in the aqueous phase, iii) social and scientific interest: the French General Directorate for Health initiated in 2009 national campaigns to assess the occurrence of pharmaceuticals and perfluorinated compounds in water intended for drinking water production (surface water and ground water). Our study focused on quantifying the target compounds in wastewater (in both the liquid and solid phases) from two urban areas and in the WWTP receiving wastewater from different catchments. The objective was i) to assess the occurrence of micropollutants released in two different types of urban areas, ii) to evaluate the removal (biodegradation or adsorption on sludge) of these compounds from wastewater by biological treatment in an urban WWTP, and iii) to examine by statistical methods if there is any relationship between micropollution and macropollution entering and leaving the studied WWTP. 2. Materials and methods 2.1. Sampling sites Wastewater samples from two urban catchments in the area of Nancy, in the North-East of France, were studied. The first catchment is residential, with about 2100 lodgings and a sanitary sewage network. The second one is a mixed catchment composed of a hospital (about 1200 hospital beds), houses and administrations and with a combined sewage network. In both catchments, sampling was performed directly in the sewer. The studied WWTP has a capacity of 500,000 p.e. (population equivalent). It treats urban wastewater from 21 municipalities (400,000 p.e.) including the two previously described catchments, and industrial wastewater from a brewery (100,000 p.e.). Urban wastewater is subject to pretreatment (grit, sand and oil removal followed by primary
settling), biological treatment (pre- and post-denitrification combined with Biolift® technology), final clarification and then tertiary treatment (phosphorus precipitation). The sludge wasted from the final clarifier and the primary sludge are combined to get digested and dried after thickening. Industrial wastewater represents less than 10% of the WWTP's daily inlet flow. It is pretreated separately to reduce its load to the same level as the urban wastewater and joins urban flows in the same biological treatment. Samples were taken at the inlet (urban wastewater before pretreatment) and outlet of the plant (treated wastewater) and in the recycle line from the final clarifier. Table S1, in Supplementary material, summarizes the sampling locations studied and the matrices analyzed. 2.2. Sample collection Several sampling campaigns were performed in winter and summer. The wastewater and treated water samples were collected in dry weather with an automatic sampler (ISCO 3700, Teledyne ISCO, USA) over 24 h (1 l per hour and per bottle, time proportionnal). The 24 samples were always grouped by two consecutive; thus 12 samples were analyzed to determine concentrations of the target compounds. The sludge samples were manually collected. Before applying extraction procedures (described in Section 2.5.), wastewater and sludge samples were stored in two different bottles: part in HDPE (high density polyethylene) bottles for perfluorinated compounds analysis and part in amber glass bottles for the other six compounds. It was previously checked that the studied molecules did not adsorb on glass bottles. 2.3. Analysis of conventional wastewater parameters Several conventional wastewater pollution parameters were assessed. Soluble chemical oxygen demand (COD) was measured by spectro-photocolorimetry (US EPA, 1993) following filtration on 0.45 μm pore size and chemical oxidation by a sulfochromic mixture. Ammonium content was determined by the Hach Nessler Method 8038 and measured on a Hach DR/2400 spectrophotometer (Hach Co., Colorado, USA) (error ± 0.5 mg/L N-NH4 +). Turbidity was measured by spectrophotometry at 450 nm and expressed in terms of formazin units. pH and conductivity were measured on a MeterLab ION 450 ion analyzer with two probes (Radiometer Analytical SAS, France). 2.4. Isotopic tracing of water sources The infiltration of rainwater or groundwater into sanitary sewers can modify sewage chemistry by simple dilution of wastewater components, by addition of supplementary pollutants, or by a change of the deuterium/hydrogen ratio analysis (Houhou et al., 2010). Therefore the infiltration of rainwater or groundwater was checked in all sampling campaigns by ensuring that they were performed at a constant D/H ratio (more information and example in Supplementary material, Fig. S1). The hydrogen isotopic composition (D/H ratio) of water was measured with an Isoprime mass spectrometer coupled with an elemental analyzer using a chromium reducing reactor (Morrison et al., 2001). The results were reported in δ-notation (δD) which represents the permil deviation of the measured isotopic ratio (D/Hsample) relative to a reference material (D/HSMOW). Standard materials were V-SMOW (Vienna Standard Mean Ocean Water).
δD ¼
D=Hsample −1 1000: D=HSMOW
ð1Þ
L. Pasquini et al. / Science of the Total Environment 481 (2014) 459–468
2.5. Extraction of micropollutants, analysis and validation of analytical methods To extract target compounds from liquid samples, two solid-phase extractions (SPE) were performed (Fig. 1). Before extraction, wastewater was centrifuged (20,000 g, 40 min, 20 °C) and filtered through a cellulose nitrate filter (pore size 0.45 μm, Sartorius Stedim Biotech GmbH, Gottingen, Germany) and pH was adjusted with NaOH (1 mol/L) or HCl (1 mol/L). The method developed for the extraction of PFOA and PFOS from wastewater was based on the protocol of Boiteux et al. (2012). The other six compounds (three pharmaceuticals, 4-nonylphenol, triclosan and sucralose) were extracted as described previously (Pasquini et al., 2013). The target compounds were extracted from solid samples according to the method of Yoo et al. (2009), adapted and validated on our molecules and samples (particulate matter and sludge) (Pasquini et al., 2013). Supplementary information on the extraction methods are available in Supplementary data, document 1. The extracted target compounds were then separated and analyzed by using LC–MS/MS (Table S2 of Supplementary data) and analytes were detected using a multiple reaction monitoring mode, monitoring two parent ion mass transitions. A seven-point calibration curve was generated for each compound using standard solutions. Quantitative analysis of the target compounds was performed using an internal standard method with their isotopemarked molecules, except for ofloxacin and 4-nonylphenol (external calibration). Table S3 gives the method detection limit (MDL) and method quantification limit (MQL) of each compound in liquid and solid phases. To determine the recoveries of extraction methods for the eight compounds from liquid and solid phases, spiking procedures were performed on wastewater and sludge. Two levels of concentration were tested in wastewater, and the spiking experiments were repeated 10 times for each concentration. Sludge was spiked at one level of concentration, close to the limit of quantification of each compound, and the experiment was performed in triplicate. Non-spiked wastewater and
461
sludge samples were also analyzed to determine any background concentrations of these compounds. In the liquid phase, recoveries ranging from 78 to 108% were obtained for all the compounds except 4-NP, which presented lower recoveries (30–60%). In the solid phase (suspended matter and sludge), all of the target compounds presented good recoveries, ranging from 79 to 132%, as presented in Table S4 (Supplementary data). To avoid false positive, false negative and contamination of the samples, quality control samples were injected every 10 injections, and sampling blanks and extraction blanks were performed. 2.6. Statistical analyses XLSTAT software (Addinsoft, Paris, France) was used for data analysis and statistics. First, principal component analysis was made to visually indicate any relationships between micropollution and macropollution parameters. The correlations were calculated according to the Pearson's coefficient. Then, multiple linear regressions (based on least squares method (Lebart et al., 1979)) were used to quantitatively describe variables of interest. 3. Results and discussion Two sampling campaigns were performed in the sewers of both urban catchments (mixed and residential) in summer and winter 2011. Three campaigns were performed at the WWTP, in summer and winter 2011 and in summer 2012. 3.1. Analysis of conventional wastewater parameters The measurement of conventional parameters (Table S5) showed that wastewater from the mixed and residential catchments was very similar in terms of organic pollution (COD) and ammonium concentrations. Turbidity was slightly higher in the residential catchment (sanitary sewage network) in summer 2011 than in the mixed one (combined sewage network). However, no seasonal effect on wastewater composition was observed. Wastewater at the WWTP inlet was less
Fig. 1. Schematic view of the extraction and analysis procedures for target compounds in wastewater and solid phases. SPE: solid phase extraction; LC–MS/MS: liquid chromatography– tandem mass spectrometry; UPLC–MS/MS: ultra performant liquid chromatography–tandem mass spectrometry; ES+, ES−: positive and negative electrospray ionization respectively.
462
L. Pasquini et al. / Science of the Total Environment 481 (2014) 459–468
loaded, with lower concentrations of ammonium and COD and lower turbidity than in the sewers. This could be explained by dilution due to water from non-domestic uses (commercial use, street washing …) from other catchments arriving at the WWTP. The WWTP outlet COD and turbidity values comply with that of the French legislation; the decree of June 22nd 2007 imposes COD concentrations below 125 mg/L at the outlet, and 90% removal for suspended matter. These parameters indicate that the studied WWTP was operating efficiently during the three sampling campaigns. We also checked in the WWTP database outlet parameters measured one week before and one week after sampling campaigns, to be sure that the WWTP was working under normal conditions and with stable performance. The measured values can thus be considered representative of WWTP characteristics and performance when operating normally. Hydrogen isotopic analyses of wastewater proved that all samples were taken without infiltration of groundwater or rainwater. 3.2. Occurrence of household micropollutants in urban wastewater Fig. 2 presents the daily mean concentrations and their standard deviation. A large standard deviation results from extreme concentration peaks at certain hours, with intense human activity (between 6 am and 10 am, and then at 6 pm and 10 pm., an increase of COD and ammonium concentration was observed) (Butler et al., 1995; Campos and von Sperling, 1996). Concentrations of erythromycin, ibuprofen, triclosan, PFOA and PFOS were very similar in both urban catchments. However, in some cases the wastewater reflected specific human activity with a release of specific micropollutants. In the mixed catchment, higher concentrations of ofloxacin were released than in the residential area. This is probably related to the presence of the hospital, where ofloxacin is used in the postoperative prevention of respiratory infections and in serious urinary or bone infections. In contrast, pharmaceuticals such as erythromycin and ibuprofen are frequently used also in home treatments. Sucralose was released in higher concentrations in the residential wastewater in both campaigns, which could be linked to the more frequent use of low-calorie (sugar-free) food and mainly in summer (related to more frequent use of sodas or to slim body challenge period) (Chiu and Westerhoff, 2010). 3.3. Fate of household micropollutants in the WWTP The concentrations of the eight micropollutants at WWTP inlet and outlet were measured over 24 h (12 measurements over 24 h). The average concentrations in the liquid phase are shown in Fig. 2. Inlet concentrations of micropollutants were similar in winter and summer seasons, and removal efficiency was similar in all campaigns, except for PFOS. Micropollutant removal efficiency was assessed using the assumption that brewery wastewater did not contain pharmaceuticals, sucralose, triclosan and PFAAs. Measurements taken by the brewery itself showed no trace of 4-NP. Three cases of removal efficiency for the studied micropollutants were observed: a) good or mean removal, b) no removal, and c) special behavior. a) More than 90% of ibuprofen and 75% of triclosan were removed from wastewater during treatment. Good removal efficiencies (up to 99%) were also found elsewhere (Buser et al., 1999; Vanderford and Snyder, 2006). Ofloxacin concentrations varied slightly over the seasons (from about 0.7 μg/L in summer 2011 to about 0.3 μg/L in summer 2012) but the removal efficiency remained systematically low, around 25%. Low rates for ofloxacin removal in biological treatment processes, from 25 to 50%, have been reported previously (Lee et al., 2007; Xu et al., 2007).
b) No elimination was observed for erythromycin, sucralose and PFOA. Concentrations of these compounds in water at plant inlet and outlet were approximately the same: on average 0.2 μg/L of erythromycin, 3 μg/L of sucralose and 10 ng/L of PFOA were released in the outlet treated water. Other works (Gobel et al., 2004; Miège et al., 2008; Yu et al., 2009; Shivakoti et al., 2010; Buerge et al., 2011; Campo et al., 2014) have also reported little or no removal of these compounds by WWTP's processes. This is not surprising in the cases of sucralose and PFOA since neither molecule is biodegradable (Kissa, 2001; Brorström-Lundén et al., 2008). The case of erythromycin is often discussed in literature because contradictory removal efficiencies have been observed. In some studies, erythromycin was not eliminated from the water phases in WWTPs (Gobel et al., 2005; Miege et al., 2008), but it was found to be highly biodegradable under aerobic conditions (and persistent in anoxic reactors) (Suarez et al., 2010). Pasquini et al. (2013) reported good biodegradation of erythromycin (ranging from 50 to 80%) in a reactor with aerobic and anoxic periods. c) The behavior of 4-NP and PFOS deserves special attention. In all three sampling campaigns, the concentrations of 4-NP were higher in the treated outlet water than in the inlet wastewater. The same phenomena was observed for PFOS during the summer 2011 campaign. The measured concentrations were above the method quantification limits, so it cannot be an analytical random error. It has been reported (Rhoads et al., 2008; Vega Morales et al., 2009; Campo et al., 2014) that the degradation of nonylphenol ethoxylates into nonylphenol and polyfluorinated compounds such as N-ethylperfluorooctan sulfonamidoethanol (N-EtFOSE) into PFOS can take place during the biological process. These reactions can explain the increase of 4-nonylphenol and PFOS concentrations through the biological treatment process. This study allowed to obtain data about the influent and effluent concentrations of micropollutants in water, and to assess the elimination of target compounds from the liquid phase. To answer the question of whether the removal of micropollutants in the liquid phase is due to biological degradation or to the displacement of pollution to the solid phase, measurements were performed on particulate matter and sludge. Table 1 presents the concentrations of target compounds in the particulate matter of the inlet wastewater, in activated sludge, and in dehydrated outlet sludge. The analysis showed that ibuprofen was not adsorbed on solid matter (particulate matter in the inlet wastewater and sludge), which suggests that ibuprofen was eliminated from the liquid phase due to degradation by biological treatment and not by adsorption on the sludge. Erythromycin and sucralose were neither adsorbed on the particulate matter or the sludge nor eliminated from the liquid phase and consequently they were released in the river by the WWTP effluent. They appear to pass through the treatment process in the liquid phase without degradation. Ofloxacin, 4-nonylphenol, triclosan, PFOA and PFOS were adsorbed on the inlet particulate matter and the outlet sludge. The inlet and outlet concentrations on the solids were quite similar (Table 1) keeping in mind that the dehydrated sludge results from codigestion of primary sludge and wasted activated sludge. According to the literature, triclosan is essentially biodegraded (Singer et al., 2002) and ofloxacin is first quickly removed from the liquid phase by adsorption on solid phases (Cha et al., 2006; Lindberg et al., 2006), and then degraded by biological treatment (Radjenovic et al., 2009; Li and Zhang, 2010). 4-Nonylphenol is known to be adsorbed on solid matter and biodegraded in aerobic conditions (Topp and Starratt, 2000; De Weert et al., 2008). The concentrations measured in our study were low (0.1–1.3 mg/kg), and comparable to some reported in the literature, which can range from 1.0 mg/kg (Pothitou and Voutsa, 2008) to 500 mg/kg (Lee and Peart, 1995).
L. Pasquini et al. / Science of the Total Environment 481 (2014) 459–468
463
Fig. 2. Daily mean concentrations of the target compounds in the liquid phase of wastewater sampled over sampling campaigns in the sewers of the mixed and residential areas, and at WWTP inlet and outlet. Mean values were calculated from 12 samples over 24 h.
464
L. Pasquini et al. / Science of the Total Environment 481 (2014) 459–468
Table 1 Concentration levels of the target compounds in the solid phases: in particulate matter of the inlet wastewater, in activated sludge and in outlet sludge. Particulate matter (μg/kg)
Activated sludge (μg/kg)
Dehydrated sludge (μg/kg)
Summer 2011 4-Nonylphenol Ofloxacin Triclosan PFOA PFOS Erythromycin Ibuprofen Sucralose
95 507 265 bMQL bMQL bMDL bMDL bMDL
166 400 383 bMQL 97 10−3 bMDL bMDL bMDL
n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a.
Winter 2011 4-Nonylphenol Ofloxacin Triclosan PFOA PFOS Erythromycin Ibuprofen Sucralose
793 170 1002 bMQL 8 10−3 bMDL bMDL bMDL
115 183 415 bMQL 18 10−3 bMDL bMDL bMDL
1278 233 974 bMQL 25 10−3 bMDL bMDL bMDL
Summer 2012 4-Nonylphenol Ofloxacin Triclosan Erythromycin Ibuprofen Sucralose
160 171 368 bMDL bMDL bMDL
228 68 433 bMDL bMDL bMDL
184 24 748 bMDL bMDL bMDL
MDL: method detection limit. MQL: method quantification limit; under MQL compounds are detected but not quantifiable. n.a.: not analyzed.
PFOA inlet and outlet concentrations remained identical (10 ng/L) in the liquid phase and it was detected in the solid phase at very low concentrations (under MQL = 8 ng/g). Thus PFOA seems to pass through the WWTP without being affected by the treatment process. PFOS concentrations in the solid phase appeared to increase in the outlet sludge and during one campaign it also increased in the liquid phase. PFOS concentration may increase due to the degradation of its precursors, so no conclusions may be made about the fate of PFOS in the WWTP. PFOS, more hydrophobic than PFOA, was adsorbed on sludge at higher concentrations than PFOA.
3.4. Multivariate statistical analysis In complex systems with many variables, statistical methods can be a useful tool to seek empirical relationships to reveal or describe behavior, properties or phenomena. It is then possible to look for a physical explanation of observed facts. For example, statistical analysis was used to identify the most relevant operating parameters relating to process performance in three papermill WWTPs (Avella et al., 2011). A statistical approach has also been used to detect a change in WWTP operation: a change in correlation coefficients or unsuitability of new measurements with the previously established model could be a relevant way to detect certain problems (Kiss et al., 2011). In this study, micropollutant and macropollutant concentrations measured at the WWTP inlet and outlet were submitted to statistical analysis. A preliminary examination by principal component analysis (PCA) (Lebart et al., 1979) revealed linear relationships between micropollution and macropollution entering the WWTP (76.59% of data variability was explained by the first three principal components) (Supplementary data, document S2). Linear relationships were also observed when considering inlet and outlet concentrations of micropollutants and macropollutants. 57.59% of data variability was explained by the first two principal components; to better explain data variability, five principal components should be considered (78.07%). Once correlations between micropollution and macropollution were observed on correlation circles, multiple linear regression (MLR) was
used to quantitatively describe a variable of interest: a micropollutant concentration as a function of macropollution parameters. First, we tried to describe the pollution entering the WWTP, seeking relationships between micropollution and macropollution. We next looked for any relationships between the inlet and outlet micropollution and outlet macropollution. Finally, we investigated the suitability of describing the outlet micropollution in relation to the inlet and outlet macropollution commonly measured at the WWTP. 3.4.1. Relationships between inlet micropollutant and inlet macropollutant concentrations MLR was used to examine if specific links exist between inlet concentrations of micropollutants and three commonly-measured inlet macropollution variables: COD, ammonium concentration (NH+ 4 ) and turbidity: C microin
calc
¼f
M C macroin
ð2Þ
calc
where C microin is the calculated inlet concentration of a micropollutant (in μg/L and ng/L for PFAAs), and C macroin M is the measured macropollution at the WWTP inlet, expressed in terms of COD (mg/L), NH+ 4 (mg/L) and turbidity (NTU). Considering the complexity of the system, only simple types of regressions were examined: linear, square, logarithm, and inverse functions were considered. The accuracy of numerical expressions to describe the parameters (the intensity of the relationships) was estimated by the correlation coefficient (r) between the measured micropollutant concentrations and the values given by its regression model. The use of transformations (square, logarithm, and inverse) did not significantly increase the degree of correlation. Thus, only coefficients of linear expressions are presented in Table 2. For example, ibuprofen inlet concentration can be expressed as:
Ibuprofenin ¼ 1:509 þ 0:00225 CODin þ 0:08460 NH4 þ 0:00087 Turbidityin :
þ
in
ð3Þ
L. Pasquini et al. / Science of the Total Environment 481 (2014) 459–468 Table 2 Structures of the linear regressions describing the inlet micropollutant concentrations as a function of macropollution measured at theWWTP inlet, and correlation coefficients (r) of the models.
Ibuprofenin Erythromycinin Ofloxacinin Triclosanin Sucralosein 4-NPin PFOAin PFOSin
Constant
CODin
NH4+in
Turbidityin
r
1.509 0.026 0.078 −0.022 0.721 0.807 10.026 29.438
0.00225 0.00003 0.00076 0.00017 0.00358 0.00177 −0.02430 −0.08608
0.08460 0.00213 0.00371 −0.00035 0.10180 −0.00092 −0.02191 −0.38320
0.00087 0.00023 0.00074 0.00146 −0.00608 −0.00188 0.02138 0.11262
0.744 0.673 0.594 0.702 0.748 0.862 0.583 0.772
Table 3 Correlations between micropollution and macropollution parameters. Significance level = 0.05.
Ibuprofenin Erythromycinin Ofloxacinin Triclosanin Sucralosein 4-NPin PFOAin PFOSin
CODin
NH4+in
Turbidityin
0.407 0.241 0.398 0.080 0.502 0.546 −0.410 −0.503
0.734 0.607 0.492 0.368 0.631 −0.127 −0.014 −0.174
0.368 0.551 0.422 0.698 −0.006 −0.666 0.382 0.487
A satisfactory reliability was obtained for all of the expressions: the calculated values followed the same tendency as the measured values in the 95% confidence interval (see Fig. S4 in Supplementary material). Ibuprofen, PFOA and PFOS concentrations were essentially determined by a constant (first column of Table 2). These compounds enter the studied WWTP in near-constant concentrations. Thus they seem to present low relations with the measured macropollution parameters (that vary over the day). Table 3 shows correlation coefficients between the micropollution and macropollution parameters (correlation matrix of Lebart et al. (1979)). It is interesting to note that micropollutants released in wastewater by the human body (erythromycin, ofloxacin, ibuprofen and sucralose) are strongly and mainly correlated to ammonium concentration (essentially issued from urine (Larsen and Gujer, 1996)). No correlation with ammonium is observed for 4-NP, PFOA and PFOS, used in many applications (detergents, textiles, packaging) not intended to pass through the human body. Triclosan, used in bodycare and toothpaste, for example, correlates mainly with turbidity and to a lesser extent with ammonium. 3.4.2. Relationships between outlet concentrations of micropollutants and measured inlet micropollution and outlet macropollution We then investigated whether a general performance of the WWTP (macropollution removal) has an impact on the efficiency of micropollution removal. Thus, MLR was used to examine if there are any relationships among outlet concentrations of one micropollutant and its inlet concentration
465
and the macropollution variables (COD, ammonium concentration and turbidity) measured at the outlet: C microout
calc
¼f
M M C microin ; C macroout :
ð4Þ
C microout calc is the calculated concentration of micropollutant (in μg/L, and ng/L for PFAAs) at the WWTP outlet, C macroout M is expressed in terms of COD (mg/L), NH+ 4 (mg/L) and turbidity (NTU), and is measured at the WWTP outlet, and C microin M is the measured concentration of micropollutant (in μg/L or ng/L) at the WWTP inlet. Table 4 shows the coefficients of the linear numerical expressions of these models and correlations (r) between the measured micropollutant concentrations and the values given by its regression model. Again, a satisfactory reliability (r) was obtained by all of the expressions (except for PFOA): according to the binary diagrams of Fig. S5 (measured and calculated concentrations), all the points lay within the 95% confidence interval. Low regression coefficients of sucralose (r = 0.535) and PFOA (r = 0.295) concentrations indicate weak links between these molecule concentrations and the macropollution and micropollution parameters used. Indeed, both molecules are not biodegradable and were not eliminated by biological treatment. The macropollution parameters are irrelevant for the modeling of these micropollutants which present little concentration variability. The following observations on micropollutant behavior are issued from statistical analysis (Table 4): - Ibuprofen and triclosan (which were well removed from the liquid phase) outlet concentrations were not dependent on their inlet concentrations. Their constants (first column of Table 4) correspond approximately to their mean outlet concentrations. It means that the WWTP treatment is efficient enough to remove both compounds present at concentration levels currently measured in wastewater up to residual values of approximately 0.05 μg/L and 0.11 μg/L, respectively. - Ofloxacin and 4-NP outlet concentrations were predominantly dependent on their inlet concentrations (a feature also observed in the PCA correlation circle). However, their behavior in the wastewater treatment process is radically different: approximately 25% of ofloxacin was always eliminated from wastewater, indicating its partial removal whatever were the inlet concentrations. It raises the question as to why the biological treatment does not exceed that value. 4-Nonylphenol concentration in the liquid phase increased at the WWTP outlet by about 50%, whatever its inlet concentration. This increase is probably due to the degradation of precursors whose concentrations are not known. - Erythromycin, sucralose, PFOA and PFOS, which the treatment process did not remove from wastewater, were essentially determined by constants (respectively 0.10, 3.0, 12.2, and 97.1) which correspond approximately to their mean inlet (and outlet) concentrations. Their output concentrations are only slightly related to output macropollution parameters. The modeling showed that, in spite of the fact that the WWTP was efficient concerning
Table 4 Structures of the linear regressions modeling the outlet micropollutant concentrations as a function of macropollution measured at the WWTP outlet and micropollution measured at the inlet, and correlation coefficients (r).
Ibuprofenout Erythromycinout Ofloxacinout Triclosanout Sucraloseout 4-NPout PFOAout PFOSout
Constant
Microin
CODout
NH4+out
Turbidityout
r
0.054 0.106 0.051 0.111 3.012 1.114 12.226 97.127
−0.00216 0.01395 0.25100 0.00575 0.03256 0.41776 −0.05450 −0.18616
0.00059 −0.00032 0.00202 −0.00035 0.00137 −0.00487 −0.01250 0.12052
−0.00585 0.00298 −0.01614 0.00481 −0.03874 −0.04359 0.12173 −3.07105
0.00563 0.00461 0.01550 −0.00096 −0.00853 0.02423 −0.05018 −1.52455
0.600 0.776 0.847 0.794 0.535 0.772 0.295 0.764
466
L. Pasquini et al. / Science of the Total Environment 481 (2014) 459–468
Fig. 3. Measured and calculated micropollutant concentrations at the WWTP outlet. Calculated concentrations were obtained by MLR on the measured inlet and outlet macropollution parameters (relationships presented in Table S6).
L. Pasquini et al. / Science of the Total Environment 481 (2014) 459–468
macropollution and satisfactory when removing easily biodegradable micropollutants (ibuprofen and triclosan), non-biodegradable micropollutants (erythromycin, sucralose, PFOA and PFOS) were not impacted by the biological treatment. 3.4.3. Relationships between outlet micropollutant concentrations and measured inlet and outlet macropollution In the last step, we tried to assess if it would be possible to estimate the micropollution released into water bodies only by measuring inlet and outlet macropollution concentrations (measured daily in a WWTP). In the wastewater entering the WWTP, relationships between the inlet micropollution and macropollution were successfully established by relationship (2). The outlet micropollution concentrations were described by relationship (4). Combining relationships (4) and (2), the outgoing micropollutant concentrations can be related to macropollution entering and leaving the WWTP: C microout
est
M M ¼ f C macroout ; C macroin :
ð5Þ
Linear numerical expressions are given in Table S6 (Supplementary data). Fig. 3 shows the measured and calculated micropollutant concentrations at the WWTP outlet. Reliability was satisfactory for all of the expressions on measurements made over two years in summer and winter periods: the calculated micropollutant concentrations follow the same tendency as the measured values (high regression coefficient, r). This statistical analysis allowed to establish correlations in the studied WWTP between the micropollutant outlet concentrations and inlet and outlet macropollution parameters. The model obviously requires validation through other sampling campaigns. However, this approach could be an operating method to estimate the micropollutant outlet concentrations released into the receiving body by measuring only macropollution parameters. WWTP operators or researchers often seek tools to estimate parameters without measuring them. This is even more important for micropollutants, which are difficult to analyze, requiring special equipment and skills. Each WWTP works with a specific technology and specific removals. Therefore, in order to establish models for a given WWTP, it is unavoidable to make preliminary measurements in this one. Afterwards estimation could be possible under two quite plausible conditions: that the studied WWTP will work in a stationary regime (as was the case during our measurements) and that no major change should occur in the behavior of the population in the urban area draining wastewaters into the WWTP. If there are any technological changes at the WWTP, or changes in human behavior new correlations will need to be sought. 4. Conclusions The presence of eight micropollutants from domestic activity in the sewage of two urban areas reflected some specific human activities (higher concentrations of ofloxacin in an area with a hospital, and of sucralose in a residential area). The assessment of their fate in an urban WWTP allowed to draw up an inventory of released micropollutant concentrations. The biological treatment process satisfactorily removed most macropollution, but only three micropollutants — ibuprofen, triclosan and partially ofloxacin — were removed from the liquid phase. Five micropollutants — erythromycin, 4-nonylphenol, sucralose and PFOA and PFOS — were released into the receiving water body. Their outlet concentrations were low (several ng up to μg/L) and some of them were more or less transferred to the solid phase and found in the outlet WWTP sludge. In this work, beyond the information about the fate of micropollutants in the WWTP process, the statistical analysis (MLR) mathematically described the observed micropollutant removal (or not) and brought
467
out that some micropollutants are released into wastewater by the human body (relationship with ammonium). The statistical evaluation of data enabled to diagnose the studied WWTP's operation relating the WWTP inlet and outlet micropollutants and outlet macropollution parameters. We did not have enough measurements to validate the model on a new series of measurements, but it may be a potential tool for characterizing any WWTP and estimating the micropollution leaving the WWTP without having to systematically measure it (which is not easy and beyond the scope of a WWTP laboratory). Conflict of interest The authors declare no conflict of interest. Acknowledgments We would like to thank the Veolia group, operators of the WWTP in Nancy-Maxéville, for allowing us access to the treatment plant, professor José Ragot from the CRAN (Centre de Recherche en Automatique de Nancy) for his valuable advice on statistical analysis, as well as the Greater Nancy, the Zone Atelier Moselle and the Lorraine Region. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.scitotenv.2014.02.075. References Avella AC, Gorner T, Yvon J, Chappe P, Guinot-Thomas P, de Donato P. A combined approach for a better understanding of wastewater treatment plants operation: statistical analysis of monitoring database and sludge physico-chemical characterization. Water Res 2011;45:981–92. Boiteux V, Dauchy X, Rosin C, Munoz JF. National screening study on 10 perfluorinated compounds in raw and treated tap water in France. Arch Environ Contam Toxicol 2012;63:1–12. Brorström-Lundén E, Svenson A, Viktor T, Woldegiorgis A, Remberger M, Kaj L. Measurement of surcalose in the Swedish screening program 2007. Swedish Environmental Research Institute; 2008. Buck RC, Franklin J, Berger U, Conder JM, Cousins IT, de Voogt P, et al. Perfluoroalkyl and polyfluoroalkyl substances in the environment: terminology, classification, and origins. Integr Environ Assess Manag 2011;7:513–41. Buerge IJ, Keller M, Buser HR, Müller MD, Poiger T. Saccharin and other artificial sweeteners in soils: estimated inputs from agriculture and households, degradation, and leaching to groundwater. Environ Sci Technol 2011;45:615–21. Buser HR, Poiger T, Muller MD. Occurrence and environmental behavior of the chiral pharmaceutical drug ibuprofen in surface waters and in wastewater. Environ Sci Technol 1999;33:2529–35. Butler D, Friedler E, Gatt K. Characterizing the quantity and quality of domestic waste-water inflows. Water Sci Technol 1995;31:13–24. Campo J, Masiá A, Picó Y, Farré M, Barceló D. Distribution and fate of perfluoroalkyl substances in Mediterranean Spanish sewage treatment plants. Sci Total Environ 2014;472:912–22. Campos HM, von Sperling M. Estimation of domestic wastewater characteristics in a developing country based on socio-economic variables. Water Sci Technol 1996;34: 71–7. Cha JM, Yang S, Carlson KH. Trace determination of beta-lactam antibiotics in surface water and urban wastewater using liquid chromatography combined with electrospray tandem mass spectrometry. J Chromatogr A 2006;1115:46–57. Chiu C, Westerhoff PK. Trace organics in Arizona surface and wastewaters. In: Halden RU, editor. Contaminants of emerging concern in the environment: ecological and human health considerations, 1048. ; 2010. p. 81–117. De Weert J, De La Cal A, Van den Berg H, Murk A, Langenhoff A, Rijnaarts H, et al. Bioavailability and biodegradation of nonylphenol in sediment determined with chemical and bioanalysis. Environ Toxicol Chem 2008;27:778–85. Gobel A, McArdell CS, Suter MJF, Giger W. Trace determination of macrolide and sulfonamide antimicrobials, a human sulfonamide metabolite, and trimethoprim in wastewater using liquid chromatography coupled to electrospray tandem mass spectrometry. Anal Chem 2004;76:4756–64. Gobel A, Thomsen A, McArdell CS, Alder AC, Giger W, Theiss N, et al. Extraction and determination of sulfonamides, macrolides, and trimethoprim in sewage sludge. J Chromatogr A 2005;1085:179–89. Houhou J, Lartiges BS, France-Lanord C, Guilmette C, Poix S, Mustin C. Isotopic tracing of clear water sources in an urban sewer: a combined water and dissolved sulfate stable isotope approach. Water Res 2010;44:256–66.
468
L. Pasquini et al. / Science of the Total Environment 481 (2014) 459–468
Kiss AMN, Marx B, Mourot G, Schutz G, Ragot J. Observers design for uncertain Takagi– Sugeno systems with unmeasurable premise variables and unknown inputs. Application to a wastewater treatment plant. J Process Control 2011;21:1105–14. Kissa E. Fluorinated surfactants and repellents. New York: Marcel Dekker; 2001. Kümmerer K. Drugs in the environment: emission of drugs, diagnostic aids and disinfectants into wastewater by hospitals in relation to other sources — a review. Chemosphere 2001;45:957–69. Larsen TA, Gujer W. Separate management of anthropogenic nutrient solutions (human urine). Water Sci Technol 1996;34:87–94. Lebart L, Morineau A, Fénelon JP. Traitement des données statistiques. Dunod, Paris: Méthodes et programmes; 1979 [518 pp.]. Lee HB, Peart TE. Determination of 4-nonylphenol in effluent and sludge from sewage treatment plants. Anal Chem 1995;67:1976–80. Lee HB, Peart TE, Svoboda ML. Determination of ofloxacin, norfloxacin, and ciprofloxacin in sewage by selective solid-phase extraction, liquid chromatography with fluorescence detection, and liquid chromatography–tandem mass spectrometry. J Chromatogr A 2007;1139:45–52. Li B, Zhang T. Biodegradation and adsorption of antibiotics in the activated sludge process. Environ Sci Technol 2010;44:3468–73. Lillenberg M, Yurchenko S, Kipper K, Herodes K, Pihl V, Sepp K, et al. Simultaneous determination of fluoroquinolones, sulfonamides and tetracyclines in sewage sludge by pressurized liquid extraction and liquid chromatography electrospray ionization– mass spectrometry. J Chromatogr A 2009;1216:5949–54. Lindberg RH, Olofsson U, Rendahl P, Johansson MI, Tysklind M, Andersson BAV. Behavior of fluoroquinolones and trimethoprim during mechanical, chemical, and active sludge treatment of sewage water and digestion of sludge. Environ Sci Technol 2006;40:1042–8. Lindström A, Buerge IJ, Poiger T, Bergqvist PA, Müller MD, Buser HR. Occurrence and environmental behavior of the bactericide triclosan and its methyl derivative in surface waters and in wastewater. Environ Sci Technol 2002;36:2322–9. Miège C, Choubert JM, Ribeiro C, Eusèbe M, Coquery M. Removal efficiency of pharmaceuticals and personal care products with varying wastewater treatment processes and operating conditions — conception of a database and first results. Water Sci Technol 2008;57(1):49–56. Miege C, Choubert JM, Ribeiro L, Eusebe M, Coquery M. Removal efficiency of pharmaceuticals and personal care products with varying wastewater treatment processes and operating conditions — conception of a database and first results. Water Sci Technol 2008;57:49–56. Morrison J, Brockwell T, Merren T, Fourel F, Phillips AM. On-line high-precision stable hydrogen isotopic analyses on nanoliter water samples. Anal Chem 2001;73:3570–5. Ochoa-Herrera V, Sierra-Alvarez R. Removal of perfluorinated surfactants by sorption onto granular activated carbon, zeolite and sludge. Chemosphere 2008;72:1588–93. Pasquini L, Munoz JF, Rimlinger N, Dauchy X, France X, Pons MN, et al. Assessment of the fate of some household micropollutants in urban wastewater treatment plant. Chem Pap 2013;67:601–12.
Picó Y, Andreu V. Fluoroquinolones in soil—risks and challenges. Anal Bioanal Chem 2007;387:1287–99. Pothitou P, Voutsa D. Endocrine disrupting compounds in municipal and industrial wastewater treatment plants in Northern Greece. Chemosphere 2008;73:1716–23. Prevedouros K, Cousins IT, Buck RC, Korzeniowski SH. Sources, fate and transport of perfluorocarboxylates. Environ Sci Technol 2006;40:32–44. Radjenovic J, Jelic A, Petrovic M, Barcelo D. Determination of pharmaceuticals in sewage sludge by pressurized liquid extraction (PLE) coupled to liquid chromatography– tandem mass spectrometry (LC–MS/MS). Anal Bioanal Chem 2009;393:1685–95. Radjenović J, Petrović M, Barceló D. Fate and distribution of pharmaceuticals in wastewater and sewage sludge of the conventional activated sludge (CAS) and advanced membrane bioreactor (MBR) treatment. Water Res 2009;43:831–41. Rhoads KR, Janssen EML, Luthy RG, Criddle CS. Aerobic biotransformation and fate of N-ethyl perfluorooctane sulfonamidoethanol (N-EtFOSE) in activated sludge. Environ Sci Technol 2008;42:2873–8. Richardson SD, Ternes TA. Water analysis: emerging contaminants and current issues. Anal Chem 2011;83:4616–48. Shivakoti BR, Tanaka S, Fujii S, Kunacheva C, Boontanon SK, Musirat C, et al. Occurrences and behavior of perfluorinated compounds (PFCs) in several wastewater treatment plants (WWTPs) in Japan and Thailand. J Environ Monit 2010;12:1255–64. Singer H, Müller S, Tixier C, Pillonel L. Triclosan: occurrence and fate of a widely used biocide in the aquatic environment: field measurements in wastewater treatment plants, surface waters, and lake sediments. Environ Sci Technol 2002;36:4998–5004. Soares A, Guieysse B, Jefferson B, Cartmell E, Lester JN. Nonylphenol in the environment: a critical review on occurrence, fate, toxicity and treatment in wastewaters. Environ Int 2008;34:1033–49. Suarez S, Lema JM, Omil F. Removal of Pharmaceutical and Personal Care Products (PPCPs) under nitrifying and denitrifying conditions. Water Res 2010;44:3214–24. Topp E, Starratt A. Rapid mineralization of the endocrine-disrupting chemical 4-nonylphenol in soil. Environ Toxicol Chem 2000;19:313–8. US EPA. Method 410.4 The determination of chemical oxygen demand by semi-automated colorimetry; 1993. Vanderford BJ, Snyder SA. Analysis of pharmaceuticals in water by isotope dilution liquid chromatography/tandem mass spectrometry. Environ Sci Technol 2006;40:7312–20. Vega Morales T, Torres Padron ME, Sosa Ferrera Z, Santana Rodriguez JJ. Determination of alkylphenol ethoxylates and their degradation products in liquid and solid samples. TrAC Trends Anal Chem 2009;28:1186–200. Xu WH, Zhang G, Li XD, Zou SC, Li P, Hu ZH, et al. Occurrence and elimination of antibiotics at four sewage treatment plants in the Pearl River Delta (PRD), South China. Water Res 2007;41:4526–34. Yoo H, Washington JW, Jenkins TM, Laurence Libelo E. Analysis of perfluorinated chemicals in sludge: method development and initial results. J Chromatogr A 2009;1216:7831–9. Yu J, Hu J, Tanaka S, Fujii S. Perfluorooctane sulfonate (PFOS) and perfluorooctanoic acid (PFOA) in sewage treatment plants. Water Res 2009;43:2399–408.
Contents lists available at ScienceDirect
Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv
Occurrence of eight household micropollutants in urban wastewater and their fate in a wastewater treatment plant. Statistical evaluation Laure Pasquini a,⁎, Jean-François Munoz b, Marie-Noëlle Pons c, Jacques Yvon a, Xavier Dauchy b, Xavier France d, Nang Dinh Le c, Christian France-Lanord e, Tatiana Görner a a
Laboratoire Interdisciplinaire des Environnements Continentaux, CNRS, Université de Lorraine, 15 Avenue du Charmois, 54501 Vandœuvre-lès-Nancy cedex, France Laboratoire d'Hydrologie de Nancy, ANSES, 40 rue Lionnois, 54000 Nancy, France Laboratoire Réactions et Génie des Procédés, CNRS, Université de Lorraine, 1 rue Grandville, 54001 Nancy cedex, France d GEMCEA, 149 rue Gabriel Péri, 54500 Vandœuvre-lès-Nancy, France e Centre de Recherches Pétrographiques et Géochimiques, CNRS, Université de Lorraine, 15 rue Notre Dame des Pauvres, 54501 Vandœuvre-lès-Nancy cedex, France b c
H I G H L I G H T S • • • •
Presence of eight household micropollutants in two different urban catchments was assessed. Concentration levels of the target compounds in sewage influent, treated wastewater and sludge were compared. The efficiency of a conventional activated sludge process was assessed. Statistical analyses have related the micropollution and macropollution in the WWTP.
a r t i c l e
i n f o
Article history: Received 30 December 2013 Received in revised form 14 February 2014 Accepted 15 February 2014 Available online 12 March 2014 Keywords: Household micropollutants Urban wastewater Wastewater treatment plant Sludge Statistical analysis Modeling
a b s t r a c t The occurrence in urban wastewater of eight micropollutants (erythromycin, ibuprofen, 4-nonylphenol (4-NP), ofloxacin, sucralose, triclosan, perfluorooctanoic acid (PFOA) and perfluorooctane sulfonate (PFOS)) originating from household activities and their fate in a biological wastewater treatment plant (WWTP) were investigated. Their concentrations were assessed in the liquid and solid phases (sewage particulate matter and wasted activated sludge (WAS)) by liquid chromatography–tandem mass spectrometry. The analysis of sewage from two different urban catchments connected to the WWTP showed a specific use of ofloxacin in the mixed catchment due to the presence of a hospital, and higher concentrations of sucralose in the residential area. The WWTP process removed over 90% of ibuprofen and triclosan from wastewater, while only 25% of ofloxacin was eliminated. Erythromycin, sucralose and PFOA were not removed from wastewater, the influent and effluent concentrations remaining at about 0.7 μg/L, 3 μg/L and 10 ng/L respectively. The behavior of PFOS and 4-nonylphenol was singular, as concentrations were higher at the WWTP outlet than at its inlet. This was probably related to the degradation of some of their precursors (such as alkylphenol ethoxylates and polyfluorinated compounds resulting in 4-NP and PFOS, respectively) during biological treatment. 4-NP, ofloxacin, triclosan and perfluorinated compounds were found adsorbed on WAS (from 5 ng/kg for PFOA to 1.0 mg/kg for triclosan). The statistical methods (principal component analysis and multiple linear regressions) were applied to examine relationships among the concentrations of micropollutants and macropollutants (COD, ammonium, turbidity) entering and leaving the WWTP. A strong relationship with ammonium indicated that some micropollutants enter wastewater via human urine. A statistical analysis of WWTP operation gave a model for estimating micropollutant output from the WWTP based on a measurement of macropollution parameters. © 2014 Elsevier B.V. All rights reserved.
1. Introduction
⁎ Corresponding author. Tel.: +33 3 83 59 62 64; fax: +33 3 83 59 62 55. E-mail address: [email protected] (L. Pasquini).
http://dx.doi.org/10.1016/j.scitotenv.2014.02.075 0048-9697/© 2014 Elsevier B.V. All rights reserved.
In the last decades, the use of pharmaceuticals and various chemicals present in many products used in our everyday life has become a source of environmental pollution transported by urban wastewater. Even though removal mechanisms for conventional macropollution (organic
460
L. Pasquini et al. / Science of the Total Environment 481 (2014) 459–468
matter, nitrogen and phosphorus) in wastewater treatment plants (WWTPs) are understood and effective, this is not the case for most micropollutants (Kümmerer, 2001; Radjenović et al., 2009; Suarez et al., 2010; Richardson and Ternes, 2011). The removal of compounds such as pharmaceuticals, personal care products or detergents is usually not checked by plant managers or authorities because few or no regulations exist concerning their release into water bodies. The scientific interest in the environmental impact of these molecules, their behavior in WWTPs, and their occurrence in water bodies first emerged some ten years ago as showed in the literature review by Pasquini et al. (2013). Many of them are thought to be a possible threat to environmental health and safety. Our research focused on eight micropollutants originating from different activities of our everyday life: erythromycin and ofloxacin (antibiotics), ibuprofen (anti-inflammatory drug), triclosan (biocide), 4-nonylphenol (4-NP is a mixture of branched isomers resulting from detergent degradation), sucralose (sweetener), and two perfluoroalkyl acids (PFAAs) (Buck et al., 2011)—perfluorooctanoic acid (PFOA) and perfluorooctane sulfonate (PFOS). PFOA and PFOS are employed in industrial and domestic products as floor cleaning agents and paints, but also in the manufacture of emulsifying agents, non-adhesive surfacing and waterproofing agent for textiles, leather and food packaging (Prevedouros et al., 2006). The target compounds were chosen according to their: i) occurrence in domestic use, ii) physico-chemical properties in order to observe their possible partitioning between water and sludge in the WWTP. Erythromycin, 4-NP, triclosan and PFAAs which are hydrophobic compounds, and ofloxacin, are known to adsorb on sludge during biological treatment (Lindström et al., 2002; Picó and Andreu, 2007; Ochoa-Herrera and Sierra-Alvarez, 2008; Soares et al., 2008; Lillenberg et al., 2009). Ibuprofen and sucralose are hydrophilic compounds that should mostly remain in the aqueous phase, iii) social and scientific interest: the French General Directorate for Health initiated in 2009 national campaigns to assess the occurrence of pharmaceuticals and perfluorinated compounds in water intended for drinking water production (surface water and ground water). Our study focused on quantifying the target compounds in wastewater (in both the liquid and solid phases) from two urban areas and in the WWTP receiving wastewater from different catchments. The objective was i) to assess the occurrence of micropollutants released in two different types of urban areas, ii) to evaluate the removal (biodegradation or adsorption on sludge) of these compounds from wastewater by biological treatment in an urban WWTP, and iii) to examine by statistical methods if there is any relationship between micropollution and macropollution entering and leaving the studied WWTP. 2. Materials and methods 2.1. Sampling sites Wastewater samples from two urban catchments in the area of Nancy, in the North-East of France, were studied. The first catchment is residential, with about 2100 lodgings and a sanitary sewage network. The second one is a mixed catchment composed of a hospital (about 1200 hospital beds), houses and administrations and with a combined sewage network. In both catchments, sampling was performed directly in the sewer. The studied WWTP has a capacity of 500,000 p.e. (population equivalent). It treats urban wastewater from 21 municipalities (400,000 p.e.) including the two previously described catchments, and industrial wastewater from a brewery (100,000 p.e.). Urban wastewater is subject to pretreatment (grit, sand and oil removal followed by primary
settling), biological treatment (pre- and post-denitrification combined with Biolift® technology), final clarification and then tertiary treatment (phosphorus precipitation). The sludge wasted from the final clarifier and the primary sludge are combined to get digested and dried after thickening. Industrial wastewater represents less than 10% of the WWTP's daily inlet flow. It is pretreated separately to reduce its load to the same level as the urban wastewater and joins urban flows in the same biological treatment. Samples were taken at the inlet (urban wastewater before pretreatment) and outlet of the plant (treated wastewater) and in the recycle line from the final clarifier. Table S1, in Supplementary material, summarizes the sampling locations studied and the matrices analyzed. 2.2. Sample collection Several sampling campaigns were performed in winter and summer. The wastewater and treated water samples were collected in dry weather with an automatic sampler (ISCO 3700, Teledyne ISCO, USA) over 24 h (1 l per hour and per bottle, time proportionnal). The 24 samples were always grouped by two consecutive; thus 12 samples were analyzed to determine concentrations of the target compounds. The sludge samples were manually collected. Before applying extraction procedures (described in Section 2.5.), wastewater and sludge samples were stored in two different bottles: part in HDPE (high density polyethylene) bottles for perfluorinated compounds analysis and part in amber glass bottles for the other six compounds. It was previously checked that the studied molecules did not adsorb on glass bottles. 2.3. Analysis of conventional wastewater parameters Several conventional wastewater pollution parameters were assessed. Soluble chemical oxygen demand (COD) was measured by spectro-photocolorimetry (US EPA, 1993) following filtration on 0.45 μm pore size and chemical oxidation by a sulfochromic mixture. Ammonium content was determined by the Hach Nessler Method 8038 and measured on a Hach DR/2400 spectrophotometer (Hach Co., Colorado, USA) (error ± 0.5 mg/L N-NH4 +). Turbidity was measured by spectrophotometry at 450 nm and expressed in terms of formazin units. pH and conductivity were measured on a MeterLab ION 450 ion analyzer with two probes (Radiometer Analytical SAS, France). 2.4. Isotopic tracing of water sources The infiltration of rainwater or groundwater into sanitary sewers can modify sewage chemistry by simple dilution of wastewater components, by addition of supplementary pollutants, or by a change of the deuterium/hydrogen ratio analysis (Houhou et al., 2010). Therefore the infiltration of rainwater or groundwater was checked in all sampling campaigns by ensuring that they were performed at a constant D/H ratio (more information and example in Supplementary material, Fig. S1). The hydrogen isotopic composition (D/H ratio) of water was measured with an Isoprime mass spectrometer coupled with an elemental analyzer using a chromium reducing reactor (Morrison et al., 2001). The results were reported in δ-notation (δD) which represents the permil deviation of the measured isotopic ratio (D/Hsample) relative to a reference material (D/HSMOW). Standard materials were V-SMOW (Vienna Standard Mean Ocean Water).
δD ¼
D=Hsample −1 1000: D=HSMOW
ð1Þ
L. Pasquini et al. / Science of the Total Environment 481 (2014) 459–468
2.5. Extraction of micropollutants, analysis and validation of analytical methods To extract target compounds from liquid samples, two solid-phase extractions (SPE) were performed (Fig. 1). Before extraction, wastewater was centrifuged (20,000 g, 40 min, 20 °C) and filtered through a cellulose nitrate filter (pore size 0.45 μm, Sartorius Stedim Biotech GmbH, Gottingen, Germany) and pH was adjusted with NaOH (1 mol/L) or HCl (1 mol/L). The method developed for the extraction of PFOA and PFOS from wastewater was based on the protocol of Boiteux et al. (2012). The other six compounds (three pharmaceuticals, 4-nonylphenol, triclosan and sucralose) were extracted as described previously (Pasquini et al., 2013). The target compounds were extracted from solid samples according to the method of Yoo et al. (2009), adapted and validated on our molecules and samples (particulate matter and sludge) (Pasquini et al., 2013). Supplementary information on the extraction methods are available in Supplementary data, document 1. The extracted target compounds were then separated and analyzed by using LC–MS/MS (Table S2 of Supplementary data) and analytes were detected using a multiple reaction monitoring mode, monitoring two parent ion mass transitions. A seven-point calibration curve was generated for each compound using standard solutions. Quantitative analysis of the target compounds was performed using an internal standard method with their isotopemarked molecules, except for ofloxacin and 4-nonylphenol (external calibration). Table S3 gives the method detection limit (MDL) and method quantification limit (MQL) of each compound in liquid and solid phases. To determine the recoveries of extraction methods for the eight compounds from liquid and solid phases, spiking procedures were performed on wastewater and sludge. Two levels of concentration were tested in wastewater, and the spiking experiments were repeated 10 times for each concentration. Sludge was spiked at one level of concentration, close to the limit of quantification of each compound, and the experiment was performed in triplicate. Non-spiked wastewater and
461
sludge samples were also analyzed to determine any background concentrations of these compounds. In the liquid phase, recoveries ranging from 78 to 108% were obtained for all the compounds except 4-NP, which presented lower recoveries (30–60%). In the solid phase (suspended matter and sludge), all of the target compounds presented good recoveries, ranging from 79 to 132%, as presented in Table S4 (Supplementary data). To avoid false positive, false negative and contamination of the samples, quality control samples were injected every 10 injections, and sampling blanks and extraction blanks were performed. 2.6. Statistical analyses XLSTAT software (Addinsoft, Paris, France) was used for data analysis and statistics. First, principal component analysis was made to visually indicate any relationships between micropollution and macropollution parameters. The correlations were calculated according to the Pearson's coefficient. Then, multiple linear regressions (based on least squares method (Lebart et al., 1979)) were used to quantitatively describe variables of interest. 3. Results and discussion Two sampling campaigns were performed in the sewers of both urban catchments (mixed and residential) in summer and winter 2011. Three campaigns were performed at the WWTP, in summer and winter 2011 and in summer 2012. 3.1. Analysis of conventional wastewater parameters The measurement of conventional parameters (Table S5) showed that wastewater from the mixed and residential catchments was very similar in terms of organic pollution (COD) and ammonium concentrations. Turbidity was slightly higher in the residential catchment (sanitary sewage network) in summer 2011 than in the mixed one (combined sewage network). However, no seasonal effect on wastewater composition was observed. Wastewater at the WWTP inlet was less
Fig. 1. Schematic view of the extraction and analysis procedures for target compounds in wastewater and solid phases. SPE: solid phase extraction; LC–MS/MS: liquid chromatography– tandem mass spectrometry; UPLC–MS/MS: ultra performant liquid chromatography–tandem mass spectrometry; ES+, ES−: positive and negative electrospray ionization respectively.
462
L. Pasquini et al. / Science of the Total Environment 481 (2014) 459–468
loaded, with lower concentrations of ammonium and COD and lower turbidity than in the sewers. This could be explained by dilution due to water from non-domestic uses (commercial use, street washing …) from other catchments arriving at the WWTP. The WWTP outlet COD and turbidity values comply with that of the French legislation; the decree of June 22nd 2007 imposes COD concentrations below 125 mg/L at the outlet, and 90% removal for suspended matter. These parameters indicate that the studied WWTP was operating efficiently during the three sampling campaigns. We also checked in the WWTP database outlet parameters measured one week before and one week after sampling campaigns, to be sure that the WWTP was working under normal conditions and with stable performance. The measured values can thus be considered representative of WWTP characteristics and performance when operating normally. Hydrogen isotopic analyses of wastewater proved that all samples were taken without infiltration of groundwater or rainwater. 3.2. Occurrence of household micropollutants in urban wastewater Fig. 2 presents the daily mean concentrations and their standard deviation. A large standard deviation results from extreme concentration peaks at certain hours, with intense human activity (between 6 am and 10 am, and then at 6 pm and 10 pm., an increase of COD and ammonium concentration was observed) (Butler et al., 1995; Campos and von Sperling, 1996). Concentrations of erythromycin, ibuprofen, triclosan, PFOA and PFOS were very similar in both urban catchments. However, in some cases the wastewater reflected specific human activity with a release of specific micropollutants. In the mixed catchment, higher concentrations of ofloxacin were released than in the residential area. This is probably related to the presence of the hospital, where ofloxacin is used in the postoperative prevention of respiratory infections and in serious urinary or bone infections. In contrast, pharmaceuticals such as erythromycin and ibuprofen are frequently used also in home treatments. Sucralose was released in higher concentrations in the residential wastewater in both campaigns, which could be linked to the more frequent use of low-calorie (sugar-free) food and mainly in summer (related to more frequent use of sodas or to slim body challenge period) (Chiu and Westerhoff, 2010). 3.3. Fate of household micropollutants in the WWTP The concentrations of the eight micropollutants at WWTP inlet and outlet were measured over 24 h (12 measurements over 24 h). The average concentrations in the liquid phase are shown in Fig. 2. Inlet concentrations of micropollutants were similar in winter and summer seasons, and removal efficiency was similar in all campaigns, except for PFOS. Micropollutant removal efficiency was assessed using the assumption that brewery wastewater did not contain pharmaceuticals, sucralose, triclosan and PFAAs. Measurements taken by the brewery itself showed no trace of 4-NP. Three cases of removal efficiency for the studied micropollutants were observed: a) good or mean removal, b) no removal, and c) special behavior. a) More than 90% of ibuprofen and 75% of triclosan were removed from wastewater during treatment. Good removal efficiencies (up to 99%) were also found elsewhere (Buser et al., 1999; Vanderford and Snyder, 2006). Ofloxacin concentrations varied slightly over the seasons (from about 0.7 μg/L in summer 2011 to about 0.3 μg/L in summer 2012) but the removal efficiency remained systematically low, around 25%. Low rates for ofloxacin removal in biological treatment processes, from 25 to 50%, have been reported previously (Lee et al., 2007; Xu et al., 2007).
b) No elimination was observed for erythromycin, sucralose and PFOA. Concentrations of these compounds in water at plant inlet and outlet were approximately the same: on average 0.2 μg/L of erythromycin, 3 μg/L of sucralose and 10 ng/L of PFOA were released in the outlet treated water. Other works (Gobel et al., 2004; Miège et al., 2008; Yu et al., 2009; Shivakoti et al., 2010; Buerge et al., 2011; Campo et al., 2014) have also reported little or no removal of these compounds by WWTP's processes. This is not surprising in the cases of sucralose and PFOA since neither molecule is biodegradable (Kissa, 2001; Brorström-Lundén et al., 2008). The case of erythromycin is often discussed in literature because contradictory removal efficiencies have been observed. In some studies, erythromycin was not eliminated from the water phases in WWTPs (Gobel et al., 2005; Miege et al., 2008), but it was found to be highly biodegradable under aerobic conditions (and persistent in anoxic reactors) (Suarez et al., 2010). Pasquini et al. (2013) reported good biodegradation of erythromycin (ranging from 50 to 80%) in a reactor with aerobic and anoxic periods. c) The behavior of 4-NP and PFOS deserves special attention. In all three sampling campaigns, the concentrations of 4-NP were higher in the treated outlet water than in the inlet wastewater. The same phenomena was observed for PFOS during the summer 2011 campaign. The measured concentrations were above the method quantification limits, so it cannot be an analytical random error. It has been reported (Rhoads et al., 2008; Vega Morales et al., 2009; Campo et al., 2014) that the degradation of nonylphenol ethoxylates into nonylphenol and polyfluorinated compounds such as N-ethylperfluorooctan sulfonamidoethanol (N-EtFOSE) into PFOS can take place during the biological process. These reactions can explain the increase of 4-nonylphenol and PFOS concentrations through the biological treatment process. This study allowed to obtain data about the influent and effluent concentrations of micropollutants in water, and to assess the elimination of target compounds from the liquid phase. To answer the question of whether the removal of micropollutants in the liquid phase is due to biological degradation or to the displacement of pollution to the solid phase, measurements were performed on particulate matter and sludge. Table 1 presents the concentrations of target compounds in the particulate matter of the inlet wastewater, in activated sludge, and in dehydrated outlet sludge. The analysis showed that ibuprofen was not adsorbed on solid matter (particulate matter in the inlet wastewater and sludge), which suggests that ibuprofen was eliminated from the liquid phase due to degradation by biological treatment and not by adsorption on the sludge. Erythromycin and sucralose were neither adsorbed on the particulate matter or the sludge nor eliminated from the liquid phase and consequently they were released in the river by the WWTP effluent. They appear to pass through the treatment process in the liquid phase without degradation. Ofloxacin, 4-nonylphenol, triclosan, PFOA and PFOS were adsorbed on the inlet particulate matter and the outlet sludge. The inlet and outlet concentrations on the solids were quite similar (Table 1) keeping in mind that the dehydrated sludge results from codigestion of primary sludge and wasted activated sludge. According to the literature, triclosan is essentially biodegraded (Singer et al., 2002) and ofloxacin is first quickly removed from the liquid phase by adsorption on solid phases (Cha et al., 2006; Lindberg et al., 2006), and then degraded by biological treatment (Radjenovic et al., 2009; Li and Zhang, 2010). 4-Nonylphenol is known to be adsorbed on solid matter and biodegraded in aerobic conditions (Topp and Starratt, 2000; De Weert et al., 2008). The concentrations measured in our study were low (0.1–1.3 mg/kg), and comparable to some reported in the literature, which can range from 1.0 mg/kg (Pothitou and Voutsa, 2008) to 500 mg/kg (Lee and Peart, 1995).
L. Pasquini et al. / Science of the Total Environment 481 (2014) 459–468
463
Fig. 2. Daily mean concentrations of the target compounds in the liquid phase of wastewater sampled over sampling campaigns in the sewers of the mixed and residential areas, and at WWTP inlet and outlet. Mean values were calculated from 12 samples over 24 h.
464
L. Pasquini et al. / Science of the Total Environment 481 (2014) 459–468
Table 1 Concentration levels of the target compounds in the solid phases: in particulate matter of the inlet wastewater, in activated sludge and in outlet sludge. Particulate matter (μg/kg)
Activated sludge (μg/kg)
Dehydrated sludge (μg/kg)
Summer 2011 4-Nonylphenol Ofloxacin Triclosan PFOA PFOS Erythromycin Ibuprofen Sucralose
95 507 265 bMQL bMQL bMDL bMDL bMDL
166 400 383 bMQL 97 10−3 bMDL bMDL bMDL
n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a.
Winter 2011 4-Nonylphenol Ofloxacin Triclosan PFOA PFOS Erythromycin Ibuprofen Sucralose
793 170 1002 bMQL 8 10−3 bMDL bMDL bMDL
115 183 415 bMQL 18 10−3 bMDL bMDL bMDL
1278 233 974 bMQL 25 10−3 bMDL bMDL bMDL
Summer 2012 4-Nonylphenol Ofloxacin Triclosan Erythromycin Ibuprofen Sucralose
160 171 368 bMDL bMDL bMDL
228 68 433 bMDL bMDL bMDL
184 24 748 bMDL bMDL bMDL
MDL: method detection limit. MQL: method quantification limit; under MQL compounds are detected but not quantifiable. n.a.: not analyzed.
PFOA inlet and outlet concentrations remained identical (10 ng/L) in the liquid phase and it was detected in the solid phase at very low concentrations (under MQL = 8 ng/g). Thus PFOA seems to pass through the WWTP without being affected by the treatment process. PFOS concentrations in the solid phase appeared to increase in the outlet sludge and during one campaign it also increased in the liquid phase. PFOS concentration may increase due to the degradation of its precursors, so no conclusions may be made about the fate of PFOS in the WWTP. PFOS, more hydrophobic than PFOA, was adsorbed on sludge at higher concentrations than PFOA.
3.4. Multivariate statistical analysis In complex systems with many variables, statistical methods can be a useful tool to seek empirical relationships to reveal or describe behavior, properties or phenomena. It is then possible to look for a physical explanation of observed facts. For example, statistical analysis was used to identify the most relevant operating parameters relating to process performance in three papermill WWTPs (Avella et al., 2011). A statistical approach has also been used to detect a change in WWTP operation: a change in correlation coefficients or unsuitability of new measurements with the previously established model could be a relevant way to detect certain problems (Kiss et al., 2011). In this study, micropollutant and macropollutant concentrations measured at the WWTP inlet and outlet were submitted to statistical analysis. A preliminary examination by principal component analysis (PCA) (Lebart et al., 1979) revealed linear relationships between micropollution and macropollution entering the WWTP (76.59% of data variability was explained by the first three principal components) (Supplementary data, document S2). Linear relationships were also observed when considering inlet and outlet concentrations of micropollutants and macropollutants. 57.59% of data variability was explained by the first two principal components; to better explain data variability, five principal components should be considered (78.07%). Once correlations between micropollution and macropollution were observed on correlation circles, multiple linear regression (MLR) was
used to quantitatively describe a variable of interest: a micropollutant concentration as a function of macropollution parameters. First, we tried to describe the pollution entering the WWTP, seeking relationships between micropollution and macropollution. We next looked for any relationships between the inlet and outlet micropollution and outlet macropollution. Finally, we investigated the suitability of describing the outlet micropollution in relation to the inlet and outlet macropollution commonly measured at the WWTP. 3.4.1. Relationships between inlet micropollutant and inlet macropollutant concentrations MLR was used to examine if specific links exist between inlet concentrations of micropollutants and three commonly-measured inlet macropollution variables: COD, ammonium concentration (NH+ 4 ) and turbidity: C microin
calc
¼f
M C macroin
ð2Þ
calc
where C microin is the calculated inlet concentration of a micropollutant (in μg/L and ng/L for PFAAs), and C macroin M is the measured macropollution at the WWTP inlet, expressed in terms of COD (mg/L), NH+ 4 (mg/L) and turbidity (NTU). Considering the complexity of the system, only simple types of regressions were examined: linear, square, logarithm, and inverse functions were considered. The accuracy of numerical expressions to describe the parameters (the intensity of the relationships) was estimated by the correlation coefficient (r) between the measured micropollutant concentrations and the values given by its regression model. The use of transformations (square, logarithm, and inverse) did not significantly increase the degree of correlation. Thus, only coefficients of linear expressions are presented in Table 2. For example, ibuprofen inlet concentration can be expressed as:
Ibuprofenin ¼ 1:509 þ 0:00225 CODin þ 0:08460 NH4 þ 0:00087 Turbidityin :
þ
in
ð3Þ
L. Pasquini et al. / Science of the Total Environment 481 (2014) 459–468 Table 2 Structures of the linear regressions describing the inlet micropollutant concentrations as a function of macropollution measured at theWWTP inlet, and correlation coefficients (r) of the models.
Ibuprofenin Erythromycinin Ofloxacinin Triclosanin Sucralosein 4-NPin PFOAin PFOSin
Constant
CODin
NH4+in
Turbidityin
r
1.509 0.026 0.078 −0.022 0.721 0.807 10.026 29.438
0.00225 0.00003 0.00076 0.00017 0.00358 0.00177 −0.02430 −0.08608
0.08460 0.00213 0.00371 −0.00035 0.10180 −0.00092 −0.02191 −0.38320
0.00087 0.00023 0.00074 0.00146 −0.00608 −0.00188 0.02138 0.11262
0.744 0.673 0.594 0.702 0.748 0.862 0.583 0.772
Table 3 Correlations between micropollution and macropollution parameters. Significance level = 0.05.
Ibuprofenin Erythromycinin Ofloxacinin Triclosanin Sucralosein 4-NPin PFOAin PFOSin
CODin
NH4+in
Turbidityin
0.407 0.241 0.398 0.080 0.502 0.546 −0.410 −0.503
0.734 0.607 0.492 0.368 0.631 −0.127 −0.014 −0.174
0.368 0.551 0.422 0.698 −0.006 −0.666 0.382 0.487
A satisfactory reliability was obtained for all of the expressions: the calculated values followed the same tendency as the measured values in the 95% confidence interval (see Fig. S4 in Supplementary material). Ibuprofen, PFOA and PFOS concentrations were essentially determined by a constant (first column of Table 2). These compounds enter the studied WWTP in near-constant concentrations. Thus they seem to present low relations with the measured macropollution parameters (that vary over the day). Table 3 shows correlation coefficients between the micropollution and macropollution parameters (correlation matrix of Lebart et al. (1979)). It is interesting to note that micropollutants released in wastewater by the human body (erythromycin, ofloxacin, ibuprofen and sucralose) are strongly and mainly correlated to ammonium concentration (essentially issued from urine (Larsen and Gujer, 1996)). No correlation with ammonium is observed for 4-NP, PFOA and PFOS, used in many applications (detergents, textiles, packaging) not intended to pass through the human body. Triclosan, used in bodycare and toothpaste, for example, correlates mainly with turbidity and to a lesser extent with ammonium. 3.4.2. Relationships between outlet concentrations of micropollutants and measured inlet micropollution and outlet macropollution We then investigated whether a general performance of the WWTP (macropollution removal) has an impact on the efficiency of micropollution removal. Thus, MLR was used to examine if there are any relationships among outlet concentrations of one micropollutant and its inlet concentration
465
and the macropollution variables (COD, ammonium concentration and turbidity) measured at the outlet: C microout
calc
¼f
M M C microin ; C macroout :
ð4Þ
C microout calc is the calculated concentration of micropollutant (in μg/L, and ng/L for PFAAs) at the WWTP outlet, C macroout M is expressed in terms of COD (mg/L), NH+ 4 (mg/L) and turbidity (NTU), and is measured at the WWTP outlet, and C microin M is the measured concentration of micropollutant (in μg/L or ng/L) at the WWTP inlet. Table 4 shows the coefficients of the linear numerical expressions of these models and correlations (r) between the measured micropollutant concentrations and the values given by its regression model. Again, a satisfactory reliability (r) was obtained by all of the expressions (except for PFOA): according to the binary diagrams of Fig. S5 (measured and calculated concentrations), all the points lay within the 95% confidence interval. Low regression coefficients of sucralose (r = 0.535) and PFOA (r = 0.295) concentrations indicate weak links between these molecule concentrations and the macropollution and micropollution parameters used. Indeed, both molecules are not biodegradable and were not eliminated by biological treatment. The macropollution parameters are irrelevant for the modeling of these micropollutants which present little concentration variability. The following observations on micropollutant behavior are issued from statistical analysis (Table 4): - Ibuprofen and triclosan (which were well removed from the liquid phase) outlet concentrations were not dependent on their inlet concentrations. Their constants (first column of Table 4) correspond approximately to their mean outlet concentrations. It means that the WWTP treatment is efficient enough to remove both compounds present at concentration levels currently measured in wastewater up to residual values of approximately 0.05 μg/L and 0.11 μg/L, respectively. - Ofloxacin and 4-NP outlet concentrations were predominantly dependent on their inlet concentrations (a feature also observed in the PCA correlation circle). However, their behavior in the wastewater treatment process is radically different: approximately 25% of ofloxacin was always eliminated from wastewater, indicating its partial removal whatever were the inlet concentrations. It raises the question as to why the biological treatment does not exceed that value. 4-Nonylphenol concentration in the liquid phase increased at the WWTP outlet by about 50%, whatever its inlet concentration. This increase is probably due to the degradation of precursors whose concentrations are not known. - Erythromycin, sucralose, PFOA and PFOS, which the treatment process did not remove from wastewater, were essentially determined by constants (respectively 0.10, 3.0, 12.2, and 97.1) which correspond approximately to their mean inlet (and outlet) concentrations. Their output concentrations are only slightly related to output macropollution parameters. The modeling showed that, in spite of the fact that the WWTP was efficient concerning
Table 4 Structures of the linear regressions modeling the outlet micropollutant concentrations as a function of macropollution measured at the WWTP outlet and micropollution measured at the inlet, and correlation coefficients (r).
Ibuprofenout Erythromycinout Ofloxacinout Triclosanout Sucraloseout 4-NPout PFOAout PFOSout
Constant
Microin
CODout
NH4+out
Turbidityout
r
0.054 0.106 0.051 0.111 3.012 1.114 12.226 97.127
−0.00216 0.01395 0.25100 0.00575 0.03256 0.41776 −0.05450 −0.18616
0.00059 −0.00032 0.00202 −0.00035 0.00137 −0.00487 −0.01250 0.12052
−0.00585 0.00298 −0.01614 0.00481 −0.03874 −0.04359 0.12173 −3.07105
0.00563 0.00461 0.01550 −0.00096 −0.00853 0.02423 −0.05018 −1.52455
0.600 0.776 0.847 0.794 0.535 0.772 0.295 0.764
466
L. Pasquini et al. / Science of the Total Environment 481 (2014) 459–468
Fig. 3. Measured and calculated micropollutant concentrations at the WWTP outlet. Calculated concentrations were obtained by MLR on the measured inlet and outlet macropollution parameters (relationships presented in Table S6).
L. Pasquini et al. / Science of the Total Environment 481 (2014) 459–468
macropollution and satisfactory when removing easily biodegradable micropollutants (ibuprofen and triclosan), non-biodegradable micropollutants (erythromycin, sucralose, PFOA and PFOS) were not impacted by the biological treatment. 3.4.3. Relationships between outlet micropollutant concentrations and measured inlet and outlet macropollution In the last step, we tried to assess if it would be possible to estimate the micropollution released into water bodies only by measuring inlet and outlet macropollution concentrations (measured daily in a WWTP). In the wastewater entering the WWTP, relationships between the inlet micropollution and macropollution were successfully established by relationship (2). The outlet micropollution concentrations were described by relationship (4). Combining relationships (4) and (2), the outgoing micropollutant concentrations can be related to macropollution entering and leaving the WWTP: C microout
est
M M ¼ f C macroout ; C macroin :
ð5Þ
Linear numerical expressions are given in Table S6 (Supplementary data). Fig. 3 shows the measured and calculated micropollutant concentrations at the WWTP outlet. Reliability was satisfactory for all of the expressions on measurements made over two years in summer and winter periods: the calculated micropollutant concentrations follow the same tendency as the measured values (high regression coefficient, r). This statistical analysis allowed to establish correlations in the studied WWTP between the micropollutant outlet concentrations and inlet and outlet macropollution parameters. The model obviously requires validation through other sampling campaigns. However, this approach could be an operating method to estimate the micropollutant outlet concentrations released into the receiving body by measuring only macropollution parameters. WWTP operators or researchers often seek tools to estimate parameters without measuring them. This is even more important for micropollutants, which are difficult to analyze, requiring special equipment and skills. Each WWTP works with a specific technology and specific removals. Therefore, in order to establish models for a given WWTP, it is unavoidable to make preliminary measurements in this one. Afterwards estimation could be possible under two quite plausible conditions: that the studied WWTP will work in a stationary regime (as was the case during our measurements) and that no major change should occur in the behavior of the population in the urban area draining wastewaters into the WWTP. If there are any technological changes at the WWTP, or changes in human behavior new correlations will need to be sought. 4. Conclusions The presence of eight micropollutants from domestic activity in the sewage of two urban areas reflected some specific human activities (higher concentrations of ofloxacin in an area with a hospital, and of sucralose in a residential area). The assessment of their fate in an urban WWTP allowed to draw up an inventory of released micropollutant concentrations. The biological treatment process satisfactorily removed most macropollution, but only three micropollutants — ibuprofen, triclosan and partially ofloxacin — were removed from the liquid phase. Five micropollutants — erythromycin, 4-nonylphenol, sucralose and PFOA and PFOS — were released into the receiving water body. Their outlet concentrations were low (several ng up to μg/L) and some of them were more or less transferred to the solid phase and found in the outlet WWTP sludge. In this work, beyond the information about the fate of micropollutants in the WWTP process, the statistical analysis (MLR) mathematically described the observed micropollutant removal (or not) and brought
467
out that some micropollutants are released into wastewater by the human body (relationship with ammonium). The statistical evaluation of data enabled to diagnose the studied WWTP's operation relating the WWTP inlet and outlet micropollutants and outlet macropollution parameters. We did not have enough measurements to validate the model on a new series of measurements, but it may be a potential tool for characterizing any WWTP and estimating the micropollution leaving the WWTP without having to systematically measure it (which is not easy and beyond the scope of a WWTP laboratory). Conflict of interest The authors declare no conflict of interest. Acknowledgments We would like to thank the Veolia group, operators of the WWTP in Nancy-Maxéville, for allowing us access to the treatment plant, professor José Ragot from the CRAN (Centre de Recherche en Automatique de Nancy) for his valuable advice on statistical analysis, as well as the Greater Nancy, the Zone Atelier Moselle and the Lorraine Region. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.scitotenv.2014.02.075. References Avella AC, Gorner T, Yvon J, Chappe P, Guinot-Thomas P, de Donato P. A combined approach for a better understanding of wastewater treatment plants operation: statistical analysis of monitoring database and sludge physico-chemical characterization. Water Res 2011;45:981–92. Boiteux V, Dauchy X, Rosin C, Munoz JF. National screening study on 10 perfluorinated compounds in raw and treated tap water in France. Arch Environ Contam Toxicol 2012;63:1–12. Brorström-Lundén E, Svenson A, Viktor T, Woldegiorgis A, Remberger M, Kaj L. Measurement of surcalose in the Swedish screening program 2007. Swedish Environmental Research Institute; 2008. Buck RC, Franklin J, Berger U, Conder JM, Cousins IT, de Voogt P, et al. Perfluoroalkyl and polyfluoroalkyl substances in the environment: terminology, classification, and origins. Integr Environ Assess Manag 2011;7:513–41. Buerge IJ, Keller M, Buser HR, Müller MD, Poiger T. Saccharin and other artificial sweeteners in soils: estimated inputs from agriculture and households, degradation, and leaching to groundwater. Environ Sci Technol 2011;45:615–21. Buser HR, Poiger T, Muller MD. Occurrence and environmental behavior of the chiral pharmaceutical drug ibuprofen in surface waters and in wastewater. Environ Sci Technol 1999;33:2529–35. Butler D, Friedler E, Gatt K. Characterizing the quantity and quality of domestic waste-water inflows. Water Sci Technol 1995;31:13–24. Campo J, Masiá A, Picó Y, Farré M, Barceló D. Distribution and fate of perfluoroalkyl substances in Mediterranean Spanish sewage treatment plants. Sci Total Environ 2014;472:912–22. Campos HM, von Sperling M. Estimation of domestic wastewater characteristics in a developing country based on socio-economic variables. Water Sci Technol 1996;34: 71–7. Cha JM, Yang S, Carlson KH. Trace determination of beta-lactam antibiotics in surface water and urban wastewater using liquid chromatography combined with electrospray tandem mass spectrometry. J Chromatogr A 2006;1115:46–57. Chiu C, Westerhoff PK. Trace organics in Arizona surface and wastewaters. In: Halden RU, editor. Contaminants of emerging concern in the environment: ecological and human health considerations, 1048. ; 2010. p. 81–117. De Weert J, De La Cal A, Van den Berg H, Murk A, Langenhoff A, Rijnaarts H, et al. Bioavailability and biodegradation of nonylphenol in sediment determined with chemical and bioanalysis. Environ Toxicol Chem 2008;27:778–85. Gobel A, McArdell CS, Suter MJF, Giger W. Trace determination of macrolide and sulfonamide antimicrobials, a human sulfonamide metabolite, and trimethoprim in wastewater using liquid chromatography coupled to electrospray tandem mass spectrometry. Anal Chem 2004;76:4756–64. Gobel A, Thomsen A, McArdell CS, Alder AC, Giger W, Theiss N, et al. Extraction and determination of sulfonamides, macrolides, and trimethoprim in sewage sludge. J Chromatogr A 2005;1085:179–89. Houhou J, Lartiges BS, France-Lanord C, Guilmette C, Poix S, Mustin C. Isotopic tracing of clear water sources in an urban sewer: a combined water and dissolved sulfate stable isotope approach. Water Res 2010;44:256–66.
468
L. Pasquini et al. / Science of the Total Environment 481 (2014) 459–468
Kiss AMN, Marx B, Mourot G, Schutz G, Ragot J. Observers design for uncertain Takagi– Sugeno systems with unmeasurable premise variables and unknown inputs. Application to a wastewater treatment plant. J Process Control 2011;21:1105–14. Kissa E. Fluorinated surfactants and repellents. New York: Marcel Dekker; 2001. Kümmerer K. Drugs in the environment: emission of drugs, diagnostic aids and disinfectants into wastewater by hospitals in relation to other sources — a review. Chemosphere 2001;45:957–69. Larsen TA, Gujer W. Separate management of anthropogenic nutrient solutions (human urine). Water Sci Technol 1996;34:87–94. Lebart L, Morineau A, Fénelon JP. Traitement des données statistiques. Dunod, Paris: Méthodes et programmes; 1979 [518 pp.]. Lee HB, Peart TE. Determination of 4-nonylphenol in effluent and sludge from sewage treatment plants. Anal Chem 1995;67:1976–80. Lee HB, Peart TE, Svoboda ML. Determination of ofloxacin, norfloxacin, and ciprofloxacin in sewage by selective solid-phase extraction, liquid chromatography with fluorescence detection, and liquid chromatography–tandem mass spectrometry. J Chromatogr A 2007;1139:45–52. Li B, Zhang T. Biodegradation and adsorption of antibiotics in the activated sludge process. Environ Sci Technol 2010;44:3468–73. Lillenberg M, Yurchenko S, Kipper K, Herodes K, Pihl V, Sepp K, et al. Simultaneous determination of fluoroquinolones, sulfonamides and tetracyclines in sewage sludge by pressurized liquid extraction and liquid chromatography electrospray ionization– mass spectrometry. J Chromatogr A 2009;1216:5949–54. Lindberg RH, Olofsson U, Rendahl P, Johansson MI, Tysklind M, Andersson BAV. Behavior of fluoroquinolones and trimethoprim during mechanical, chemical, and active sludge treatment of sewage water and digestion of sludge. Environ Sci Technol 2006;40:1042–8. Lindström A, Buerge IJ, Poiger T, Bergqvist PA, Müller MD, Buser HR. Occurrence and environmental behavior of the bactericide triclosan and its methyl derivative in surface waters and in wastewater. Environ Sci Technol 2002;36:2322–9. Miège C, Choubert JM, Ribeiro C, Eusèbe M, Coquery M. Removal efficiency of pharmaceuticals and personal care products with varying wastewater treatment processes and operating conditions — conception of a database and first results. Water Sci Technol 2008;57(1):49–56. Miege C, Choubert JM, Ribeiro L, Eusebe M, Coquery M. Removal efficiency of pharmaceuticals and personal care products with varying wastewater treatment processes and operating conditions — conception of a database and first results. Water Sci Technol 2008;57:49–56. Morrison J, Brockwell T, Merren T, Fourel F, Phillips AM. On-line high-precision stable hydrogen isotopic analyses on nanoliter water samples. Anal Chem 2001;73:3570–5. Ochoa-Herrera V, Sierra-Alvarez R. Removal of perfluorinated surfactants by sorption onto granular activated carbon, zeolite and sludge. Chemosphere 2008;72:1588–93. Pasquini L, Munoz JF, Rimlinger N, Dauchy X, France X, Pons MN, et al. Assessment of the fate of some household micropollutants in urban wastewater treatment plant. Chem Pap 2013;67:601–12.
Picó Y, Andreu V. Fluoroquinolones in soil—risks and challenges. Anal Bioanal Chem 2007;387:1287–99. Pothitou P, Voutsa D. Endocrine disrupting compounds in municipal and industrial wastewater treatment plants in Northern Greece. Chemosphere 2008;73:1716–23. Prevedouros K, Cousins IT, Buck RC, Korzeniowski SH. Sources, fate and transport of perfluorocarboxylates. Environ Sci Technol 2006;40:32–44. Radjenovic J, Jelic A, Petrovic M, Barcelo D. Determination of pharmaceuticals in sewage sludge by pressurized liquid extraction (PLE) coupled to liquid chromatography– tandem mass spectrometry (LC–MS/MS). Anal Bioanal Chem 2009;393:1685–95. Radjenović J, Petrović M, Barceló D. Fate and distribution of pharmaceuticals in wastewater and sewage sludge of the conventional activated sludge (CAS) and advanced membrane bioreactor (MBR) treatment. Water Res 2009;43:831–41. Rhoads KR, Janssen EML, Luthy RG, Criddle CS. Aerobic biotransformation and fate of N-ethyl perfluorooctane sulfonamidoethanol (N-EtFOSE) in activated sludge. Environ Sci Technol 2008;42:2873–8. Richardson SD, Ternes TA. Water analysis: emerging contaminants and current issues. Anal Chem 2011;83:4616–48. Shivakoti BR, Tanaka S, Fujii S, Kunacheva C, Boontanon SK, Musirat C, et al. Occurrences and behavior of perfluorinated compounds (PFCs) in several wastewater treatment plants (WWTPs) in Japan and Thailand. J Environ Monit 2010;12:1255–64. Singer H, Müller S, Tixier C, Pillonel L. Triclosan: occurrence and fate of a widely used biocide in the aquatic environment: field measurements in wastewater treatment plants, surface waters, and lake sediments. Environ Sci Technol 2002;36:4998–5004. Soares A, Guieysse B, Jefferson B, Cartmell E, Lester JN. Nonylphenol in the environment: a critical review on occurrence, fate, toxicity and treatment in wastewaters. Environ Int 2008;34:1033–49. Suarez S, Lema JM, Omil F. Removal of Pharmaceutical and Personal Care Products (PPCPs) under nitrifying and denitrifying conditions. Water Res 2010;44:3214–24. Topp E, Starratt A. Rapid mineralization of the endocrine-disrupting chemical 4-nonylphenol in soil. Environ Toxicol Chem 2000;19:313–8. US EPA. Method 410.4 The determination of chemical oxygen demand by semi-automated colorimetry; 1993. Vanderford BJ, Snyder SA. Analysis of pharmaceuticals in water by isotope dilution liquid chromatography/tandem mass spectrometry. Environ Sci Technol 2006;40:7312–20. Vega Morales T, Torres Padron ME, Sosa Ferrera Z, Santana Rodriguez JJ. Determination of alkylphenol ethoxylates and their degradation products in liquid and solid samples. TrAC Trends Anal Chem 2009;28:1186–200. Xu WH, Zhang G, Li XD, Zou SC, Li P, Hu ZH, et al. Occurrence and elimination of antibiotics at four sewage treatment plants in the Pearl River Delta (PRD), South China. Water Res 2007;41:4526–34. Yoo H, Washington JW, Jenkins TM, Laurence Libelo E. Analysis of perfluorinated chemicals in sludge: method development and initial results. J Chromatogr A 2009;1216:7831–9. Yu J, Hu J, Tanaka S, Fujii S. Perfluorooctane sulfonate (PFOS) and perfluorooctanoic acid (PFOA) in sewage treatment plants. Water Res 2009;43:2399–408.