Analytical and Bioanalytical Chemistry

, Volume 387, Issue 4, pp 1469–1478

Polar herbicides, pharmaceutical products, perfluorooctanesulfonate (PFOS), perfluorooctanoate (PFOA), and nonylphenol and its carboxylates and ethoxylates in surface and tap waters around Lake Maggiore in Northern Italy

Authors

    • European Commission – DG Joint Research CentreInstitute for Environment and Sustainability
  • Jan Wollgast
    • European Commission – DG Joint Research CentreInstitute for Environment and Sustainability
  • Tania Huber
    • European Commission – DG Joint Research CentreInstitute for Environment and Sustainability
  • Georg Hanke
    • European Commission – DG Joint Research CentreInstitute for Environment and Sustainability
Original Paper

DOI: 10.1007/s00216-006-1036-7

Cite this article as:
Loos, R., Wollgast, J., Huber, T. et al. Anal Bioanal Chem (2007) 387: 1469. doi:10.1007/s00216-006-1036-7

Abstract

A survey of contamination of surface and drinking waters around Lake Maggiore in Northern Italy with polar anthropogenic environmental pollutants has been conducted. The target analytes were polar herbicides, pharmaceuticals (including antibiotics), steroid estrogens, perfluorooctanesulfonate (PFOS), perfluoroalkyl carboxylates (including perfluorooctanoate PFOA), nonylphenol and its carboxylates and ethoxylates (NPEO surfactants), and triclosan, a bactericide used in personal-care products. Analysis of water samples was performed by solid-phase extraction (SPE) then liquid chromatography–triple-quadrupole (tandem) mass spectrometry (LC–MS–MS). By extraction of 1-L water samples and concentration of the extract to 100 μL, method detection limits (MDLs) as low as 0.05–0.1 ng L−1 were achieved for most compounds. Lake-water samples from seven different locations in the Southern part of Lake Maggiore and eleven samples from different tributary rivers and creeks were investigated. Rain water was also analyzed to investigate atmospheric input of the contaminants. Compounds regularly detected at very low concentrations in the lake water included: caffeine (max. concentration 124 ng L−1), the herbicides terbutylazine (7 ng L−1), atrazine (5 ng L−1), simazine (16 ng L−1), diuron (11 ng L−1), and atrazine-desethyl (11 ng L−1), the pharmaceuticals carbamazepine (9 ng L−1), sulfamethoxazole (10 ng L−1), gemfibrozil (1.7 ng L−1), and benzafibrate (1.2 ng L−1), the surfactant metabolite nonylphenol (15 ng L−1), its carboxylates (NPE1C 120 ng L−1, NPE2C 7 ng L−1, NPE3C 15 ng L−1) and ethoxylates (NPEnOs, n = 3-17; 300 ng L−1), perfluorinated surfactants (PFOS 9 ng L−1, PFOA 3 ng L−1), and estrone (0.4 ng L−1). Levels of these compounds in drinking water produced from Lake Maggiore were almost identical with those found in the lake itself, revealing the poor performance of sand filtration and chlorination applied by the local waterworks.

Keywords

Polar herbicidesPharmaceuticalsPerfluorooctanesulfonate (PFOS)NonylphenolDrinking waterSolid-phase extractionLiquid chromatography–tandem mass spectrometry (LC–MS–MS)

Introduction

In many countries drinking water is produced from rivers which are affected by waste-water-treatment-plant effluents. This multifunctional use of rivers raises serious concern about drinking-water quality when contaminants are not removed by water-treatment plants. Polar compounds, especially, because they are readily soluble in water, pose problems for fresh-water systems and the supply of clean drinking water, because they are difficult to remove during water-treatment processes. In contrast, less polar substances are usually well removed during treatment of waste water and production of drinking water, because of their adsorption by solid organic matter.

Several studies have identified polar anthropogenic contaminants, for example pesticides, pharmaceutical products, estrogens, surfactants, and personal-care products (fragrances, sunscreens, antimicrobial compounds) in finished drinking water [16]. From 1992 to 1994 clofibric acid, a pharmaceutical metabolite, and N-(phenylsulfonyl)sarcosin were detected in Berlin (Germany) tap water at maximum concentrations of 270 and 105 ng L−1, respectively [1]. Stackelberg et al. [5] detected eighteen anthropogenic contaminants in finished drinking water in the low μg L−1 range (in the USA). These compounds included pharmaceutical compounds and their metabolites, fragrance compounds, flame retardants, plastizicers, cosmetic compounds, and a solvent. In Southern California, thirteen polar organic pharmaceutical and personal-care products (including clofibric acid, ibuprofen, triclosan, and DEET) were detected at low μg L−1 concentrations in finished drinking water from water filtration plants [4]. Phthalate esters have been reported in drinking water and bottled water [7, 8]. Sacher et al. identified eight pharmaceutical products (including diclofenac, carbamazepine and sulfamethoxazole) in groundwater wells in Germany, at concentrations up to 1 μg L−1 [9].

Sand filtration and flocculation are very inefficient at removing trace amounts of polar anthropogenic compounds from water. If the raw water of waterworks is contaminated by such polar contaminants, removal can only be ensured by use of more advanced techniques, for example ozonization, activated carbon, or membrane filtration. Ozonization, for example, has been shown to be very effective at oxidizing carbamazepine and diclofenac [10].

In the finished drinking water of Barcelona (Spain), however, herbicides and alkylphenols have been detected even after ozone treatment and carbon filtration; nonylphenol and the nonylphenol ethoxycarboxylates NPE1C and NPE2C were found at concentrations up to 90, 105, and 215 ng L−1, respectively [3], and maximum concentrations of the herbicides atrazine and simazine and the plasticizer bisphenol A were 18, 32, and 5 ng L−1, respectively [2].

Although no adverse health effects have yet been attributed to consumption of these compounds at such low concentrations, in a recent study of the effects on human cells of a drug mixture at environmental levels it was found in laboratory tests that a combination of pharmaceutical compounds can inhibit the growth of embryonic kidney cells [11]. Following the precautionary principle, drinking water should be free from hazardous contaminants, so data are needed to characterize the status of the aquatic environment and investigate contamination trends. In the European Union the maximum allowable concentration of individual pesticides in drinking water is 0.1 μg L−1, and the sum for all pesticides should not exceed 0.5 μg L−1, as laid down in Council Directive 80/778/EC [12], as amended by Directive 98/83/EC. In contrast, the occurrence of pharmaceutical products, surfactants, or personal-care products in drinking water is not regulated. The European Water Framework Directive from the year 2000 [13], amended by the new proposed Priority Substances Directive [14], only sets environmental quality standards for surface waters.

The objective of this study was to investigate the occurrence of polar emerging anthropogenic contaminants (herbicides, pharmaceutical products, perfluorinated surfactants (PFOS, PFOA), and nonylphenol and its carboxylates and ethoxylates) in river and lake water and in tap (drinking) water produced from Lake Maggiore. This lake, the second largest in Italy, has suffered from mercury and DDT contamination in the past [1517]. The quality of its water decreased substantially from the 1950s to the 1970s because of the effects of domestic and industrial contamination. Although nutrients and some non-polar contaminants have attracted widespread attention, polar substances have been investigated less often.

The region of Lake Maggiore can be characterized as a mountainous and tourist-affected area without much agricultural land use; there are many developed smaller urban centers and some industrial impact on the west side (in Piemonte). In most places around Lake Maggiore people receive their tap water from mountain spring or ground water sources. In some areas however lake water is used for production of drinking water.

The survey presented here is related to the chemical monitoring activity foreseen in the WFD—starting from 2007 all member states of the EU will have to monitor their water bodies for 33 selected priority compounds [18] and for other river basin-specific pollutants. According to this directive, member states “shall identify within each river basin district all bodies of water used for abstraction of water intended for human consumption” and “shall monitor those bodies of water which provide more than 100 m3 drinking water a day for all priority substances discharged and all other substances discharged in significant quantities which could affect the status of the body of water and which are controlled under the provisions of the Drinking Water Directive”. The monitoring frequency for small communities (<10,000 inhabitants) should be at least four times a year [18].

Experimental

Chemicals and reagents

Analytical standards were obtained from the Sigma–Aldrich group, Dr Ehrenstorfer (Augsburg, Germany; http://www.analytical-standards.com/), or Wellington Laboratories (Guelph, Ontario, Canada). Long-chain nonylphenol ethoxylates NPEnOs (n = 3–17) were available only as a commercial surfactant mixture of different oligomers (Kao, Barcelona, Spain) with an average of nine ethoxy groups [3]. Standards for NPE2C and NPE3C were not available. Methanol (SupraSolv), ethyl acetate (SupraSolv), acetonitrile (LiChrosolv for HPLC), and acetone (SupraSolv) were obtained from Merck (Darmstadt, Germany). All the target compounds are listed in Table 2.

Single standard stock solutions of the analytes in the high mg L−1 range were prepared by weighing milligram amounts of the compounds and dissolution in 10 mL methanol or Milli-Q water. Working standard solutions were prepared by further diluting the stock standard solutions with methanol. All compounds were combined in the same standard solution.

Samples and sample pretreatment

Water samples from Lake Maggiore and from some of its tributary rivers and creeks in the southern part of the lake were investigated, as also were some tap waters (Fig. 1). The samples (grab samples) were taken between February and April 2006. The lake-water samples were taken from the shore at Arolo, Ispra, Angera, Sesto Calende, Arona, Stresa, and Verbania (Villa Taranto). The rivers and creeks investigated were: Creek Ballarate (in the village Arolo), River Bardello (Bozza), Creek Acqua Nera (Ispra), Creek Vévera (Arona), Creek Tiasca (Meina), Creek Erno (Lesa), Creek S. Spessa (Baveno), River Strona (Gravellona Toce), River Toce (Gravellona Toce), Creek San Bernadino (Verbania), and River Toce (Villadossola). In Table 3 the rivers are categorized as “affected” rivers and mountain rivers, which were relatively clean (Creek San Bernadino (Verbania), Creek S. Spessa (Baveno), River Toce (Villadossola). Rain water was collected in by the Joint Research Centre in Ispra by using glass funnels and 2 and 5-L glass bottles. The sampling points around Lake Maggiore are depicted in Fig. 1. Four rain events were studied: 24 March 2006 (1.5 L collected), 28 March (1.35 L), 3–4 April (400 mL), and 9–10 April (6 L). Six tap-water samples were analyzed from places were lake water is used as the source. Two mountain spring waters were also investigated, and served as blank water samples. They were taken in Omegna (Lake Orta) and in Premia (close to Crodo in Val Formazza, north of Domodossola). Water samples were collected in 1-L glass bottles and stored at 4 °C; they were not filtered. The glass bottles were washed in a laboratory dish washer and then baked in an oven at 450 °C for 5 h; non-amber Schott-Duran (Wertheim/Main, Germany) glass bottles were used for sampling; extraction was performed within 48 h; no preservatives were added to the samples.
https://static-content.springer.com/image/art%3A10.1007%2Fs00216-006-1036-7/MediaObjects/216_2006_1036_Fig1_HTML.gif
Fig. 1

Map of Lake Maggiore with sampling points. F, fiume=river; T, torrente=creek

Solid-phase extraction

The 1-L water samples were extracted by solid-phase extraction (SPE). The SPE procedure for extraction and clean-up of the water samples was performed automatically using an AutoTrace SPE workstation (Caliper Life Sciences) and Oasis HLB cartridges (Waters; 200 mg, 6 mL). The cartridges were activated and conditioned with 5 mL methanol and 5 mL water at a flow-rate of 5 mL min−1. The water samples were subsequently passed through the wet cartridges at a flow-rate of 5 mL min−1, the columns rinsed with 2 mL water at 3 mL min−1 and the cartridges were dried for 30 min using nitrogen at 0.6 bar. Elution was performed with 2:2:1 methanol–acetone–ethyl acetate (2 × 3 mL). Evaporation of the extracts to 100 μL, with nitrogen, was performed at 35 °C in a water bath, by use of a TurboVap II Concentration Workstation (Zymark, Hopkinton, MA, USA) and a Liebisch Labortechnik (Bielefeld, Germany) evaporator; extracts were not evaporated to dryness.

Liquid chromatography–tandem mass spectrometry (LC–MS–MS)

Analysis was performed by reversed-phase liquid chromatography (RP-LC) with electrospray ionization (ESI) mass spectrometric (MS) detection, by atmospheric-pressure ionization (API) in the negative or positive ionization modes, using a triple-quadrupole MS–MS system. LC was performed with an Agilent 1100 Series LC system comprising a binary pump, vacuum degasser, autosampler, and a thermostatted column compartment. LC separations were performed on a 100 mm × 2.1 mm, 3-μm particle, Hypersil Gold column (Thermo Electron). Tandem mass spectrometry was performed with a bench-top triple-quadrupole Quattro Micro spectrometer from Waters–Micromass (Manchester, UK) equipped with an electrospray probe and a Z-spray interface.

Compounds in different classes were analyzed in five different LC–MS–MS runs because compounds amenable to positive or negative ionization conditions should not be analyzed in the same run. Table 1 shows the different HPLC gradient conditions used for the different compound classes. The mobile phases were prepared from water and acetonitrile. The aqueous phase was either acidified with 0.1% acetic acid (pH 3.5) or used without buffer (for alkylphenolic compounds and steroid estrogens). Nonylphenol is not separated from its carboxylates under acidic solvent conditions and the steroid estrogens better ionized under neutral conditions [19]. Pharmaceutical compounds, herbicides, and perfluorinated surfactants were separated using 0.1% acetic acid as the aqueous phase. Nonylphenol ethoxylates (NPEOs) were separated using 20 mmol L−1 ammonium acetate in both the water and acetonitrile. They were detected as their \( NH^{ + }_{4} \) adduct ions [M+NH4]+; the MRM transitions for the NPEnOs (n = 3–17) are not given in Table 2 but can be found elsewhere [20, 21]. The flow-rate was always 0.25 mL min−1; amounts of water at the starts of the gradients in the different methods are given in Table 1. The gradient always proceeded to 90% acetonitrile in 25 min. This composition was held for 5 min then returned to the starting conditions over 5 min, followed by 5 min equilibration. The injection volume was 5 μL; injection was performed with an autosampler.
Table 1

Liquid chromatography gradient conditions for the separation of the different classes of compounds

Compounds

LC solvents

Gradient start

Herbicides and pharmaceutical products (+)

Water (0.1% acetic acid) and acetonitrile

95% water

Pharmaceutical products and triclosan (−)

Water (0.1% acetic acid) and acetonitrile

95% water

Perfluorinated surfactants (−)

Water (0.1% acetic acid) and acetonitrile

50% water

Alkylphenolic compounds and steroid estrogens (−)

Pure water and acetonitrile

70% water

Nonylphenol ethoxylates (+)

20 mmol L−1 ammonium acetate in both the aqueous and acetonitrile phases

55% water

Gradient always up to 90% acetonitrile within 25 min

(+) positive ionization, (−) negative ionization

Table 2

LC–MS–MS conditions, retention times, SPE recovery, and MDLs

Compound

MRM

Cone

Coll.

Ret. Time (min)

Recovery (%)

MDL (ng L−1)

Alkylphenolic compounds

      

Bisphenol A

227→133

45

30

8.5

89 ± 3

0.2

Nonylphenol

219→133 (147)

45

32

20.6

72 ± 5

10a

4-n-Nonylphenol

219→106

40

22

23.0

37 ± 5

n.a.

Octylphenol

205→106

45

20

21.2

52 ± 5

0.1

Nonylphenoxyacetic acid NPE1C

277→219 (133)

35

18

15.4

68 ± 12

2a

NPE1C (d2)

279→133

35

40

15.4

76 ± 9

n.a.

NPE2C

321→219

35

40

16.8

n.a.

0.05

NPE3C

365→219

35

40

17.4

n.a.

0.05

Octylphenoxyacetic acid OPE1C

263→205 (106)

35

20

16.5

70 ± 8

0.05

OPE2C

307→205

35

20

14.7

n.a.

0.05

Octylphenoldiethoxylate OPE2O

312→183

25

12

19.8

62 ± 4

3

Nonylphenoldiethoxylate NPE2O

326→183

23

12

23.3

48 ± 6

3

Steroid estrogens

      

Estriol

287→145 (171)

50

40

6.3

51 ± 16

0.1

Estradiol

271→145 (183)

50

45

8.8

54 ± 15

0.1

Estrone

269→145 (143)

50

45

10.6

63 ± 13

0.05

17α-Ethinylestradiol

295→145 (159)

50

40

10.2

44 ± 1

0.1

Diethylstilbestrol

267→222 (237)

30

35

12.4

78 ± 14

0.1

Pharmaceuticals

      

Ketoprofen

253→209

25

10

16.9

66 ± 8

0.1

Benzafibrate

360→274

32

17

17.3

66 ± 12

0.1

Diclofenac

294→250

25

13

19.7

65 ± 6

0.3

Ibuprofen

205→161

25

8

20.0

62 ± 12

0.1

Gemfibrozil

249→121

27

15

21.4

62 ± 9

0.05

Triclosan (bactericide)

287→35

25

12

22.4

68 ± 11

2

Ciprofloxacin

332→231 (314)

30

23

8.7

8 ± 11

2

Enrofloxacin

360→316 (245)

30

24

9.2

7 ± 12

1

Sarafloxacin

386→299 (368)

25

27

10.0

8 ± 3

0.3

Sulfamethoxazole

254→156

30

17

11.4

64 ± 9

0.3

Erythromycin

734→158 (576)

30

20

12.6

28 ± 10

0.2

Carbamazepine

237→194

35

18

14.3

68 ± 11

0.1

Roxithromycin

838→158 (680)

30

22

14.4

29 ± 10

0.1

Caffeine (stimulant)

195→138 (110)

25

19

7.5

75 ± 8

5a

Herbicides

      

Atrazine-desethyl

188→146

35

16

10.3

71 ± 10

0.1

Simazine

202→132

40

18

13.0

70 ± 11

0.2

Simazine (IS, d5)

207→137

26

18

13.0

72 ± 9

n.a.

Atrazine

216→174 (132)

35

17

15.1

72 ± 13

0.05

Atrazine (IS, d5)

221→179 (137)

25

18

15.1

73 ± 8

n.a.

Isoproturon

207→72

30

17

15.8

72 ± 13

0.05

Isoproturon (IS, d6)

213→78

30

18

15.8

75 ± 6

n.a.

Diuron

233→72

30

17

16.0

72 ± 12

0.02

Terbutylazine

230→174 (132)

35

17

17.7

69 ± 8

0.3a

Terbutylazine (IS, d5)

235→179 (137)

22

17

17.7

70 ± 8

n.a.

Linuron

249→160

33

18

18.2

75 ± 9

0.5

Perfluorinated surfactants

      

PFHpA; perfluoroheptanoate

363→319

14

10

6.5

61 ± 14

0.5a

PFOA; perfluorooctanoate

413→369

14

10

8.9

60 ± 11

1a

PFOA (13C4)

417→372

14

10

8.9

62 ± 9

n.a.

PFOS; perfluorooctanesulfonate

499→80, 99

60

47

16.7

56 ± 10

0.1

PFOS (13C4)

503→80, 99

60

47

16.7

58 ± 12

n.a.

PFNA; perfluorononanoate

463→419

14

10

12.0

56 ± 9

0.1

PFDA; perfluorodecanoate

513→469

14

11

15.4

60 ± 8

0.05

PFUnA; perfluoroundecanoate

563→519

14

11

19.0

55 ± 13

0.5a

PFDoA; perfluorododecanoate

613→569

14

11

22.3

38 ± 12

0.05

SPE recovery was from 1 L water spiked at 10 ng L−1 using 200 mg Oasis HLB cartridges

Normal font, positive ionization; italic font, negative ionization

n.a., not applicable

MRM, multiple reaction monitoring; IS, internal standard; coll., collision energy; MDL, method detection limit

aBlank value-determined MDLs

Instrument control and data acquisition and evaluation (integration and quantification) were performed with MassLynx software. Nitrogen was used as the nebulizer gas and argon as the collision gas. The capillary potential was 3.0 kV in positive-ion mode and −2.8 kV in negative-ion mode, the extractor lens and RF lens potentials were 1.0 V and 0.1 V, respectively. The source and desolvation temperatures were 120 and 350 °C. Cone and desolvation gas flows were 50 and 600 L h−1, respectively. The analyzer conditions used for MRM analysis were: LM 1 and HM 1 resolution 11.0, ion energy 1 1.0, entrance −1 (negative-ion mode), 2 (positive-ion mode), exit 1, LM 2 and HM 2 resolution 10.0, ion energy 2 2.0, multiplier 600 V. The MRM inter-channel delay was 0.05 and the inter-scan delay 0.15.

Quantitative LC–MS–MS analysis was performed in the multiple-reaction-monitoring (MRM) mode. Collision-induced dissociation (CID) was performed using argon at approximately 3.5 × 10−3 mbar as collision gas at collision energies of 7–40 eV. The optimized characteristic MRM precursor→product ion pairs monitored for the quantification of the compounds, with the cone potential and collision energy, are given in Table 2. Not all compounds have two sensitive transitions which can be used at such low trace concentrations. They were obtained by continuous injection of single standard solutions.

The compound-dependent method detection limits (MDLs) for the SPE–LC–MS–MS procedure were determined by analysis of real water samples (Table 2), at a signal-to-noise ratio of three, with 1 L of water extracted and concentrated to 100 μL (concentration factor 10,000). For most of the compounds MDLs as low as 0.05–0.1 ng L−1 were achieved. The injection volume was 5 μL. Blank values for some compounds were in the low ng L−1 range in the laboratory Milli-Q water but not in clean mountain spring water (except for caffeine, nonylphenol, and the nonylphenol carboxylates (NPECs) and ethoxylates (NPEnOs), terbutylazine, and the perfluorinated surfactants perfluorooctanoate (PFOA) and perfluoroundecanoate; see below).

Quantification and quality assurance

The compounds were identified by retention-time matching and by monitoring their specific MRM transitions. External quantification with standard solution mixtures was usually used. Amounts were measured by use of regression equations for the dependence of peak area on concentration using six-point external calibration plots after chromatography of standard solutions in the range 0.01–10 mg L−1. Some analytes were quantified by use of internal standards, by using the ratio of peak area of the unknown to that of an internal standard (see below). Standards were injected after every 6th real sample to monitor decreases in sensitivity because of MS source contamination; after injection of ∼30 samples the sensitivity decrease was typically in the range ∼10–20%. Methanol blanks were also injected after every 6th real sample to clean the column and for investigation of memory effects. Recoveries were determined by spiking samples of Milli-Q water with the compounds at concentrations of 10 and 100 ng L−1 (n = 6); they were usually in the range 50–90% (Table 2). Concentrations (Table 3) were corrected for SPE recovery but not for signal-ion suppression because of co-eluting matrix components. Standard deviations of recoveries from SPE were typically in the range 5–15%. Additional quality assurance is currently being performed by participation in laboratory inter-comparison studies for PFOA/S and pharmaceutical products.
Table 3

Concentrations (ng L−1) of polar contaminants in lake, tap, rain, and river water

 

Lake Water (N = 8)

Tap Water (N = 6)

Affected rivers (N = 9)

Mountain rivers (N = 3)

Rain water (N = 4)

Mountain spring water (N = 2)

 

Average±SD

Range

Average±SD

Range

Range

n

Range

n

 

n

 

n

Carbamazepine

6.4 ± 2.3 (n = 8)

2.4–9.3

5.0 ± 2.2 (n = 6)

3.0–9.1

7.0-345

9

n.d.–2.9

2

n.d.

 

n.d.

 

Sulfamethoxazole

6.8 ± 1.9 (n = 8)

4.4–9.5

 

n.d.

3.5–89

5

n.d.

 

n.d.

 

n.d.

 

Gemfibrozil

0.7 ± 0.5 (n = 8)

0.2–1.7

0.4 ± 0.3 (n = 6)

0.1–0.8

0.4–44

9

n.d.–0.4

2

n.d.

 

n.d.

 

Benzafibrate

0.7 ± 0.3 (n = 8)

0.3–1.2

0.7 ± 0.6 (n = 6)

0.2–1.9

0.3–38

9

n.d.–0.2

1

n.d.–0.8

2

n.d.

 

Ibuprofen

1.2 ± 0.6 (n = 3)

n.d.–1.6

 

n.d.

1.3–73

9

n.d.–3.0

2

n.d.

 

n.d.

 

Diclofenac

1.4 ± 0.2 (n = 2)

n.d.–1.6

 

n.d.

1.7–158

9

n.d.–2.2

1

n.d.

 

n.d.

 

Estrone

0.1 ± 0.1 (n = 7)

n.d.–0.4

 

n.d.

n.d.–2.0

6

n.d.–0.2

2

n.d.

 

n.d.

 

Estradiol

 

n.d.

 

n.d.

n.d. - 0.9

1

n.d.

 

n.d.

 

n.d.

 

Triclosan

2.6 ± 1.7 (n = 3)

n.d.–4.1

 

n.d.

n.d.–15

6

n.d.

 

n.d.

 

n.d.

 

Atrazine

3.3 ± 1.0 (n = 8)

2.6–5.5

5.3 ± 5.2 (n = 6)

3.0–16

n.d.–3.2

7

n.d.–0.3

1

n.d.–2.7

1

n.d.

 

Isoproturon

0.1 ± 0.0 (n = 5)

n.d.–0.13

0.1 ± 0.0 (n = 2)

n.d.–0.14

n.d.–0.5

3

n.d.

 

0.3–1.5

4

n.d.

 

Diuron

3.9 ± 3.0 (n = 8)

0.9–10.6

2.7 ± 1.7 (n = 6)

1.4–5.9

n.d.–46

8

n.d.

 

0.7–8.4

4

n.d.

 

Terbutylazine

5.9 ± 1.2 (n = 8)

4.0–7.1

9.5 ± 11.8 (n = 6)

3.3–33.5

n.d.–71

8

0.9–1.2

3

7.4-206

4

0.1

2

Atrazine-desethyl

5.5 ± 2.5 (n = 8)

3.6–11.1

7.0 ± 5.2 (n = 6)

3.2–17.0

n.d.

 

n.d.

 

n.d.

 

n.d.

 

Simazine

7.1 ± 3.7 (n = 8)

4.1–15.9

11.7 ± 11.5 (n = 6)

3.0–31.2

n.d.–8.5

2

n.d.

 

1.5–9.2

4

n.d.

 

Caffeine

58 ± 33 (n = 8)

23–124

23.7 ± 15.6 (n = 6)

10.5–53

0.6–1056

9

3.3-44

3

30–182

4

0.5–1.4

2

Nonylphenol

< MDL

 

< MDL

 

MDL–140

9

< MDL

 

< MDL

 

< MDL

 

NPE1C

64.2 ± 27.5 (n = 8)

18–120

38.3 ± 17.2 (n = 6)

11.1–60

4.3-1836

9

1.1–12.6

3

0.8–9.0

4

0.4

2

NPE2C

4.3 ± 1.9 (n = 8)

0.7–6.6

2.0 ± 1.7 (n = 6)

0.4–4.9

1.1–920

9

n.d.–0.8

2

0.4–0.9

4

n.d.

 

NPE3C

9.2 ± 3.5 (n = 8)

2.6–15.1

5.6 ± 3.3 (n = 6)

1.4–11.2

1.0-716

9

0.2–1.7

3

0.3–1.6

4

n.d.

 

Bisphenol A

0.9 ± 0.3 (n = 5)

n.d.–1.2

 

n.d.

0.9–61

4

n.d.

 

n.d.–5.5

1

n.d.

 

NPE2O

3.0 ± 2.2 (n = 4)

n.d.–3.0

4.1 (n = 1)

n.d.–4.1

3.1–41

9

n.d.–2.4

1

4.4–18.3

4

n.d.–2.8

2

Σ NPEnO (n = 3-17)

83 ± 96 (n = 8)

15-307

81 ± 99 (n = 6)

24–281

21-5114

9

126-189

3

441-1891

4

172-191

2

PFHpA

0.6 ± 0.06 (n = 8)

0.5–0.7

0.5 ± 0.2 (n = 6)

0.3–0.8

0.3–2.3

9

0.1–0.3

3

0.6–1.1

4

n.d.–0.8

1

PFOA

2.4 ± 0.4 (n = 8)

1.8–2.9

2.4 ± 0.7 (n = 6)

1.0–2.9

0.6–15.9

9

< MDL

3

2.4–7.9

4

< MDL

2

PFOS

7.8 ± 0.6 (n = 8)

7.2–8.6

8.1 ± 1.2 (n = 6)

6.2–9.7

n.d.–38.5

4

n.d.–0.3

1

3.3–16.7

4

n.d.

 

PFNA

0.6 ± 0.1 (n = 8)

0.4–0.8

0.5 ± 0.5 (n = 6)

0.3–0.7

0.2–16.2

9

0.2–0.2

3

1.1–1.8

4

n.d.

 

PFDA

0.3 ± 0.1 (n = 8)

0.1–0.4

0.2 ± 0.1 (n = 6)

0.1–0.3

n.d.–10.8

8

n.d.–0.1

2

0.5–1.0

4

n.d.

 

PFUnA

0.5 ± 0.6 (n = 8)

0.1–1.9

0.2 ± 0.1 (n = 6)

0.1–0.4

0.1–38

9

0.1–0.2

3

0.5–0.6

4

0.1

2

PFDoA

0.8 ± 0.7 (n = 8)

0.1–2.1

1.0 ± 1.0 (n = 6)

0.1–2.8

0.1–14.1

9

n.d.–0.1

2

n.d.–1.1

2

n.d.

 

SD, standard deviation; N, number of samples; n, number of positive samples

Average values were calculated for lake and tap water only

Values are corrected for recovery but not for signal ion suppression

MDL for nonylphenol ∼10 ng L−1 and for PFOA ∼1 ng L−1

Perfluorooctanesulfonate (PFOS) and perfluorooctanoate (PFOA) were quantified by use of the internal surrogate standards perfluoro-1-[1,2,3,4-13C4]octanesulfonate and perfluoro-n-[1,2,3,4-13C4]octanoate, respectively (Wellington Laboratories, Guelph, Ontario, Canada). These internal standards were added to the water samples, before SPE, at a concentration of 1 or 10 ng L−1. For quantification of the herbicides simazine, atrazine, isoproturon, and terbutylazine samples were spiked with the deuterated internal standards simazine (d5), atrazine (d5), isoproturon (d6), and terbutylazine (d5) at a concentration of 10 ng L−1. NPE2C and NPE3C were quantified with the NPE1C standard, assuming similar response factors.

Results and discussion

Method optimization

Figure 2 shows typical LC–MS–MS chromatograms obtained from a tap-water extract in which pesticides, pharmaceuticals and perfluorinated surfactants (PFOA, PFOS) were detected.
https://static-content.springer.com/image/art%3A10.1007%2Fs00216-006-1036-7/MediaObjects/216_2006_1036_Fig2_HTML.gif
Fig. 2

LC–MS–MS chromatograms obtained from a tap water extract

SPE recovery (Table 2) was determined by spiking 1 L Milli-Q water (n = 6) with 10 and 100 ng L−1 of the compounds and extracting as described above. Recovery for most of the compounds ranged from 50 to 90%. Lower recoveries were obtained for ciprofloxacin, enrofloxacin, sarafloxacin, erythromycin, and roxythromycin, because, according to the literature, antibiotics should be extracted by use of special SPE procedures using EDTA or cation-exchange adsorption [2224]. Other SPE materials were not tested in this study.

Procedural blank values were investigated for all the compounds studied, but determined only for caffeine, nonylphenol, NPE1C, NPEnOs, terbutylazine, and two perfluorinated surfactants [25], because these compounds were detected in all the samples analyzed, even in Milli-Q, rain, and mountain spring water. No blank problems were caused by the Teflon tubes of the LC–MS–MS system [25]. Because Milli-Q water contained higher levels of the compounds than mountain spring or rain water it was not used for blank level determination. Blank values determined in mountain spring water were: terbutylazine (0.1 ng L−1), caffeine (1.4 ng L−1), nonylphenol (5.5–8.8 ng L−1), NPE1C (0.4 ng L−1), NPEnOs (172–191 ng L−1), PFOA (0.5 ng L−1), and perfluoroundecanoate (0.1 ng L−1).

A recent publication includes a good discussion of the problem of blanks in nonylphenol analysis [26]. The paper reports that alkylphenol (and phthalate) blanks are quite low for residue-analysis grade solvents but higher in laboratory indoor air, because they are leached from material or instruments present in the laboratory. This explains our nonylphenol blanks. It is probably leached from the plastic SPE cartridges.

Ion suppression

Ion suppression is a well known LC–MS phenomenon, but has rarely been studied for environmental water samples (because it is difficult and work-intensive). Quantitative LC–MS–MS results may be adversely affected by ion suppression caused by co-eluting sample matrix components, which can have dramatic effects on the intensity of the analyte signal because of suppression of ionization efficiency in the electrospray source [27].

Ion suppression was checked, by standard addition, for the pharmaceutical compounds gemfibrozil, ibuprofen, diclofenac, benzafibrate, and carbamazepine, and for PFOS and nonylphenol, in extracts from a contaminated river water sample (Creek Vevera), a lake water, and a tap water. For the extract from the river-water sample, suppression reduced the signal for these pharmaceuticals by approximately 30–40%; for the lake-water extract it was ∼20% and for the tap water extract ∼10%. For PFOS and nonylphenol however, no signal suppression was observed for the three extracts studied. The real environmental water concentrations of the pharmaceuticals are therefore higher than the values given in Table 3. No signal enhancement was observed.

Field samples

Lake, river, tap (drinking), and rain water samples in the southern part of Lake Maggiore were analyzed. In this way we could assess patterns of contamination of polar anthropogenic chemicals in this area.

All results for lake, tap, river, rain, and mountain spring waters are listed in Table 3. Because concentrations in the low ng L−1 range were quite constant at the different monitoring points along the lake and in the tap water produced from lake water, average concentrations and their standard deviations were calculated for lake and tap waters. For the other types of water only the concentration range is given.

All substances listed in Table 3 except ibuprofen (number of positive results n = 3), diclofenac (n = 2), estradiol (not detected), triclosan (n = 3), isoproturon (n = 5), bisphenol A (n = 5), and NPE2O (n = 4) were regularly detected in the lake water. The most abundant compounds were caffeine, the herbicides terbutylazine, atrazine, simazine, diuron, and atrazine-desethyl, the pharmaceuticals carbamazepine and sulfamethoxazole, and NPE1C, NPE2C, NPE3C, PFOS, and PFOA (Table 3). These compounds are relatively equally distributed in the lake water. The most constant values were found for PFOS and PFOA, confirming the persistent character of these substances. The antibiotics ciprofloxacin, enrofloxacin, sarafloxacin, erythromycin, and roxythromycin were also detected in river, lake and tap waters. Quantification of these compounds was difficult, however, because of their low recovery by SPE (Table 2). Nonylphenol was not quantifiable at levels below 10 ng L−1, because of high laboratory blanks (see above).

Figure 3 shows the average concentrations of the polar contaminants detected in lake and tap water. It is apparent from this figure that many of the compounds were detected at almost the same concentrations in the tap water as in the Lake Maggiore water, possibly because of their insufficient removal by the sand filtration and chlorination used in the waterworks at Lake Maggiore for production of tap (drinking) water. In tap water produced from ground water less of these substances were detected (carbamazepine (<1 ng L−1), atrazine (12 ng L−1), terbutylazine (5 ng L−1), atrazine-desethyl (20 ng L−1), simazine (6 ng L−1), and no PFOS (not in Table 3). These measurements show that pesticides are penetrating into ground water.
https://static-content.springer.com/image/art%3A10.1007%2Fs00216-006-1036-7/MediaObjects/216_2006_1036_Fig3_HTML.gif
Fig. 3

Average concentrations of polar contaminants in lake (n = 8) and tap (n = 6) water

Diclofenac was detected only twice in the lake water, and ibuprofen and triclosan three times (Table 3). This, and their absence from the tap water, are indicative of the good biodegradability of these three compounds (it is reported in the literature that diclofenac and ibuprofen are rapidly degraded by direct photolysis under normal environmental conditions [28, 29], and that triclosan also is a non-persistent substance [30]).

In rivers the highest concentrations were detected for NPE1C (up to 1830 ng L−1), caffeine (1050 ng L−1), carbamazepine (345 ng L−1), diclofenac (158 ng L−1), and nonylphenol (140 ng L−1). Ketoprofen (not included in Table 3) was detected in one river water sample only, the Creek Vevera (concentration ∼80 ng L−1). PFOS was detected in two river water samples (Creek Vevera and River Strona) at concentrations >20 ng L−1; River Strona (38.5 ng L−1) is affected by the metal industry (in Omegna/Gravellona Toce). Figure 4 shows results from monitoring of the most important contaminants (except NPE1C and caffeine) in the river and creek water samples. In this figure the difference between affected and uncontaminated mountain rivers is readily apparent.
https://static-content.springer.com/image/art%3A10.1007%2Fs00216-006-1036-7/MediaObjects/216_2006_1036_Fig4_HTML.gif
Fig. 4

Concentrations of the most important contaminants in the different rivers and creeks around Lake Maggiore

Four rain water samples were analyzed. The most abundant compounds were caffeine (up to 182 ng L−1), NPE2O (18 ng L−1), and terbutylazine (206 ng L−1). The presence of polar pesticides in rain water, at concentrations up to 730 ng L−1, has also been reported in Denmark, where the most abundant pesticides were atrazine, isoproturon, terbutylazine, and MCPA [31]. High NPEnO levels (between 0.5 and 1.8 μg L−1; sum of all isomers, n = 3–17) were also found in rain water.

Lake Maggiore receives municipal, agricultural, and industrial discharges, directly or via its tributary rivers. Detection of some compounds (atrazine, isoproturon, diuron, terbutylazine, simazine, caffeine, PFOA, PFOS, and the NPEO surfactants) at high concentrations in rain water suggests that atmospheric deposition contributes to the contamination of the lake by these substances.

Conclusions

In Lake Maggiore and in drinking water produced from the lake water several polar anthropogenic contaminants—herbicides, pharmaceuticals, personal-care products, surfactants, or their metabolites—were detected at low ng L−1 concentrations. Although natural processes often provide an effective means of removing many micro-contaminants, some effluent-derived contaminants seem to be resistant to biodegradation, and can therefore be characterized as polar persistent pollutants. Lake Maggiore receives municipal, agricultural, and industrial discharges, directly or via its tributary rivers. The lake water is used in some areas for production of drinking water. Most of the compounds detected are unregulated “emerging” pollutants. For the organic contaminants studied, European drinking water standards have been established for the pesticides only. Little is known about the fate of these contaminants in the aquatic environment and their possible human health impacts in drinking water. Concentrations of polar priority contaminants (herbicides and nonylphenol) in Lake Maggiore are far below proposed environmental quality standards (EQS) of the Water Framework Directive (WFD), which are 0.3 μg L−1 for nonylphenol and between 0.2 and 1 μg L−1 for polar herbicides [14]. Introduction of wastewater effluents into surface waters and drinking water aquifers, and thus the water re-use, is becoming more common throughout the world, because the continuing growth of the human population creates a corresponding increase in the demand for the supply of fresh water. It is, therefore, necessary to identify the chemicals likely to be present in recycled water, evaluate their potential effects on humans and aquatic ecosystems, and assess approaches for minimizing their release.

Acknowledgement

We would like to thank Gunther Umlauf for his helpful comments on the manuscript. Three anonymous reviewers are thanked for their revision.

Copyright information

© Springer-Verlag 2007