ARTICLE pubs.acs.org/ac
Automatic Searching and Evaluation of Priority and Emerging Contaminants in Wastewater and River Water by Stir Bar Sorptive Extraction followed by Comprehensive Two-Dimensional Gas Chromatography-Time-of-Flight Mass Spectrometry María Jose Gomez,*,† Sonia Herrera,† David Sole,† Eloy García-Calvo,† and Amadeo R. Fernandez-Alba†,‡ †
IMDEA-Agua (Instituto Madrile~ no De Estudios Avanzados-Agua), Punto Net, Edificio ZYE 2°, Parque Científico Tecnol ogico de la Universidad de Alcala, 28805 Alcala de Henares, Madrid, Spain ‡ Pesticide Residue Research Group, University of Almería, 04120 Almería, Spain
bS Supporting Information ABSTRACT:
A new analytical method based on stir bar sorptive extraction (SBSE), followed by comprehensive two-dimensional gas chromatography (GCxGC-TOF-MS), has been developed for the automatic searching and evaluation of nonpolar or semipolar contaminants in wastewater and river water. The target compounds selected were 13 personal care products (PCPs), 15 polycyclic aromatic hydrocarbons (PAHs) and 27 pesticides. Excellent results have been obtained in terms of separation efficiency and also in terms of compound identification. Exceptional method detection limits were achieved applying the optimized method, at or below 1 ng/L for most of the compounds in real samples. The reliable confirmation of analyte identity was possible at this low concentration level, even for typically troublesome compounds such as the PAHs. The other validation parameters were good. In addition to obtaining analytical information such as identification and quantification of target analytes, it is also possible to screen for nontarget compounds or unknowns. New contaminants have been identified in the wastewater effluents and river water samples, such as cholesterol and its degradation products, pharmaceuticals, industrial products, other pesticides, and PCPs. The multidimensional information generated by the instrument can also be used by the researchers for contrasting samples and identifying, much more easily, the major differences between samples. We have used this feature to propose studies of comparison between the fingerprinting of different water samples, such as the contamination variation along a river affected by the discharge of urban wastewaters and also the contamination variation over a period of time in the effluent. Results show that the most frequently detected contaminants (and the contaminants detected at higher concentrations) were the PCPs. The musk fragrances galaxolide and tonalid were the most concentrated compounds in the samples. The pesticides and PAHs were present at much lower concentration than PCPs.
T
he main tool of the European Water Policy to reduce chemical pollution of surface water bodies is the Water Framework Directive (WFD).1,2 The WFD’s strategy against pollution of water involves establishing a list of substances that pose a significant risk to or via the aquatic environment. Nevertheless, priority pollutants constitute only part of the large chemical pollution puzzle3 the number of potentially hazardous chemicals that can reach the environment is very large, and new r 2011 American Chemical Society
substances are constantly being developed and released. There is a diverse group of unregulated pollutants, many times called “emerging” contaminants, including pharmaceuticals and personal care products (PPCPs); This raises an increasing concern, as Received: November 22, 2010 Accepted: February 17, 2011 Published: March 09, 2011 2638
dx.doi.org/10.1021/ac102909g | Anal. Chem. 2011, 83, 2638–2647
Analytical Chemistry they are released on a continuous basis to aquatic ecosystems, where undetectable or unnoticed effects may occur.4-6 Wastewaters are the main route of emission of organic contaminants to the environment.4,7-19 Multiresidue analytical methodologies are becoming essential tools for a detailed chemical characterization related to identification of priority and emerging pollutants in waters.8-11 Papers related to multiresidue analytical methodologies have increased over recent years although most of them are focused on target analysis methods.8-10 This means that a large number of compounds and their degradation products fall outside of any control. Therefore, there is a need for methods offering rapid and reliable screening of a large number of compounds. To assess the pollution in waters a target analysis supplemented by a screening analysis using the same data file is a promising approach. Up to date, nonpolar or semipolar compounds residue analysis in waters has been accomplished by gas chromatography mass spectrometry (GC-MS) or tandem mass spectrometry (GC-MS/ MS).9-15 Quadrupole and ion trap mass spectrometers are commonly used for targeted analysis. The need to select ions (SIM) or MS/MS transitions limits the number of compounds that can be analyzed. The mass spectrometers can be operated in scan mode in the screening or as an untargeted approach to overcome the restrictions encountered with target analysis, though the reduction in acquisition speed, poor response and interferences limit the suitability of such an approach. GCxGC-TOF-MS is a powerful technique that allows the separation of many constituents of previously unresolved complex mixtures of contaminants in environmental samples; this technique allows simultaneous determination of hundreds if not thousands of pollutants at low levels in a single analysis. A number of reviews on GCxGC have been published.16-19 There are a variety of studies in which the GCxGC technique has become a valuable analytical tool. GCxGC has been applied mainly in petrochemical and geochemical analysis,20,21 flavor and fragrance,22,23 food,24,25 and pollutants in the environment.26-28 The major benefits of GCxGC in environmental analysis are improvements in trace analysis by increasing sensitivity and the resolution of target compounds from coextracted impurities and interferences. The benefit of coupling GCxGC with TOFMS includes acquiring full range nonskewed mass spectral information for all peaks.Until now, GCxGC-TOFMS has scarcely been used for environmental analysis. Nevertheless studies have demonstrated the power of the GCxGC technique for the full separation of complex mixtures, such as PAHs, PCBs,28,29 and surface water contaminants.26 In a recent publication, Matamoros et al.27 determine, for the first time, pharmaceuticals, pesticides, and related organic contaminants in river water by SPE, followed by GCxGC-TOF-MS. One challenge for the analysis of organic micropollutants in water samples is the low concentrations in which these analytes are present; therefore, a preconcentration step is necessary. Solid phase extraction (SPE) has been the method of choice for the analysis of micropollutants in waters,8-11,14,27 however low recovery rates for very lipophilic compounds are experienced when using SPE on natural waters and wastewaters.11,30 Among the different preconcentration procedures found in the recent literature and the importance that miniaturization has acquired in chemical analysis, the use of small and handy devices are gaining significance. Stir bar sorptive extraction (SBSE) shows an increasing demand for the analysis of micropollutants in water and has been successfully applied in preconcentration of PPCPs,
ARTICLE
PAHs, and pesticides.13,15,31-33 The use of SBSE minimizes the amount of organic solvent used, is rapid, useful for multiresidue analysis and increases the detection limits significantly.13 A disadvantage of this technique is that, until now, only PDMS-coated stir-bars are commercially available, and because of the nonpolarity of the PDMS polymer, polar compounds are poorly extracted. Solid-phase microextraction (SPME) is another technique that miniaturize sample preparation, however, lower detection limits (sub-ng/L to ng/L) can be achieved with SBSE, since extraction is performed with a larger amount of PDMS. The aim of this study is to present SBSE-GCxGC-TOFMS as a powerful tool for the automatic screening, identification and quantification of priority and emerging organic micropollutants in wastewater effluent and river water samples. In addition to analytical information, such as identification and quantification of target analytes, with important advantages respect the classical GC-MS/MS analysis, the advantages of the GCxGC technique are used to propose studies of comparison between the fingerprinting of different water samples.
’ EXPERIMENTAL SECTION Chemicals and Reagents. All the analytical grade chemicals (purity >95%) included in this study were purchased from Sigma-Aldrich (Steinheim, Germany) or from Dr. Ehrenstorfer (Augsburg, Germany). The selection of the analytes and working solution preparation is presented in the Supporting Information. Analytical-grade ethyl acetate, methanol, and sodium chloride (purity, 99.5%) were supplied by J.T. Baker (Deventer, Holland). Analytical-grade water was purchased from Fluka (Buchs, Switzerland). Effluent Wastewater and River Water Sampling. Treated wastewater samples were collected from the effluent of the Alcala de Henares wastewater treatment plant (WWTP), a town with about 200 000 inhabitants located in the Community of Madrid (in central Spain). This WWTP discharges directly into the Henares River. The river samples analyzed in this study were collected at different points along the Henares River. Sampling was done at 20 and 1150 m downstream from the Alcala de Henares WWTP emission point and 35.000, 24.000, and 6.000 m upstream of the WWTP. Grab water samples (2 L) were collected in clean amber glass bottles. Once in the laboratory, all the samples were stored at 4 °C prior to analysis, which was performed within 24 h. Sample Preparation. Water samples were used directly, without preliminary filtration. The extractions were performed with 20 mm 0.5 mm (length film thickness) PDMS commercial stir bars (47 μL; Twister, a magnetic stirring rod is placed inside a glass jacket and coated with PDMS) obtained from Gerstel (M€ulheim a/d Ruhr, Germany). For sample preparation details see Supporting Information. Instrumental Analysis. The coated stir bars were thermally desorbed using a commercial thermal desorption unit TDU (Gerstel) connected to a programmed temperature vaporisation (PTV) system injector CIS-4 (Gerstel) by a heated transfer line at 300 °C. The PTV injector was installed in a GC GC-TOFMS system which consisted of an Agilent 7890A (Agilent Technologies, Palo Alto, CA, USA) gas chromatograph, equipped with a secondary oven to fit the secondary column, and a quad-jets modulator (two cold jets and two hot jets). Liquid nitrogen was used to cool down the nitrogen gas for cold pulses and automatically filled from a Dewar using a liquid leveller, 2639
dx.doi.org/10.1021/ac102909g |Anal. Chem. 2011, 83, 2638–2647
Analytical Chemistry
ARTICLE
Table 1. Validation Data Obtained with the SBSE-GCxGC-TOF-MS Method for the Analysis of Target Compounds in Effluent Wastewater and River Watera precision RSD (%) N=5
linearity (ng/L)
compound
log Kow
recoveries in
MDLs
MQLs
matrix (%) N = 3
(ng/L)
(ng/L)
theoretical
WWE
recoveries (%) WWE
diluted 1:4
river
WWE river WWE river
range
R2
range
WWE
R2
river
rep.
repr.
WWE
WWE
2-EHMC
5.8
100
43
81
109
0.10 0.02
0.2
0.07
1-500
0.9917
1-500
0.9887
15
28
4-MBC
5.1
98
142
141
142
0.04 0.18
0.14
0.60
1-500
0.9950
1-500
0.9916
19
18
acenaphthene
3.92
79
79
99
134
0.10 0.19
0.34
0.62
1-500
0.9908
1-500
0.9961
7
5
acenaphthylene alachlor
4.07 3.52
84 60
17 157
67 124
136 134
0.16 0.24 1.74 0.75
0.52 5.82
0.81 2.51
1-500 5-500
0.9865 0.9835
1-500 1-500
0.9996 0.9856
4 21
5 13
aldrin
6.5
100
15
30
123
0.89 0.46
2.96
1.53
1-500
0.9860
1-500
0.9911
18
17
anthracene
4.45
93
121
110
139
0.02 0.04
0.06
0.14
1-500
0.9956
1-500
0.9830
17
16
atrazine
2.61
16
149
148
115
benzo[a]anthracene
5.79
100
81
97
98
3.07 0.98 10.23
3.28
5-500
0.9938
1-500
0.9880
15
9
0.30 0.10
0.32
1-500
0.9959
1-500
0.9900
27
15
1.01
benzo[a]pyrene
5.97
100
21
57
67
0.18 0.27
0.61
0.89
1-500
0.9950
1-500
0.9921
17
17
benzo[ghi]perylene
6.63
100
17
29
60
0.99 1.30
3.29
4.32
5-500
0.9980
5-500
0.9694
11
26
benzo[k]fluoranthene BHT
6.84 5.1
100 98
22 192
61 149
69 101
0.18 0.14 0.08 0.03
0.61 0.26
0.48 0.10
1-500 1-500
0.9988 0.9838
1-500 1-500
0.9910 0.9893
18 24
17 26
biphenylol-2
3.09
36
130
116
131
1.77 1.08
5.9
3.61
5-500
0.9912
5-500
0.9856
9
21
chlorfenvinphos
3.81
75
124
125
121
0.52 0.09
1.75
0.3
1-500
0.9890
1-500
0.9890
18
12 10
(trans,cis) chlorpyrifos-ethyl
4.96
98
82
80
137
0.44 0.58
1.46
1.93
5-500
0.9927
5-500
0.9886
15
chlorpyrifos-methyl
4.31
90
148
110
138
0.55 0.54
1.82
1.79
1-500
0.9918
10-500
0.9890
11
8
chrysene
5.73
100
54
95
92
0.13 0.07
0.43
0.22
1-500
0.9854
1-500
0.9930
21
11
clorophene DDE (o,p)
3.6 6.51
65 100
137 171
121 138
132 114
0.08 0.55 0.76 0.60
0.26 2.55
1.82 2.00
1-500 1-500
0.9914 0.9990
1-500 1-500
0.9809 0.9957
15 18
8 28
DDT (p,p)
6.91
100
25
60
99
1.82 1.96
6.06
6.53
5-500
0.9954
5-500
0.9935
16
24
diazinon
3.81
75
155
108
141
0.10 0.46
0.34
1.53
1-500
0.9879
1-500
0.9872
17
13
dibenz[a,h]anthracene
6.5
100
18
34
59
1.16 0.75
3.86
2.52
5-500
0.9934
5-500
0.9886
17
19
dieldrin
5.4
99
58
57
79
0.33 0.12
1.11
0.40
1-500
0.9890
1-500
0.9943
14
12
endosulfan i
3.83
76
157
147
140
1.38 1.15
4.60
3.85
5-500
0.9907
5-500
0.9940
14
12
endosulfan ii
3.83
76
163
149
110
0.68 0.17
2.28
0.56
1-500
0.9897
1-500
0.9919
26
24
endosulfan sulfate endrin
3.66 5.2
68 99
139 181
121 120
129 134
1.79 1.60 0.82 0.82
5.96 2.72
5.32 2.73
5-500 1-500
0.9915 0.9997
5-500 1-500
0.9910 0.9941
17 14
13 10 11
fluoranthene
5.16
99
141
123
140
2.02 0.05
6.73
0.18
10-500
0.9962
1-500
0.9900
16
fluorene
4.18
87
138
106
128
0.15 0.04
0.51
0.14
1-500
0.9959
1-500
0.9913
10
7
galaxolide
5.9
100
194
161
153
0.02 0.08
0.06
0.27
1-500
0.9810
1-500
0.9920
24
17
hexachlorobenzene
5.73
100
63
72
142
1.46 0.90
4.86
3.01
5-500
0.9903
5-500
0.9854
21
25
indeno[1,2.3-cd]
6.7
100
17
39
55
1.06 1.49
3.53
4.98
5-500
0.9917
5-500
0.9960
17
26
3 3.72
32 71
49 107
56 105
53 103
8.57 2.15 28.57 0.40 0.02 1.32
7.17 0.07
10-500 1-500
0.9990 0.9935
10-500 1-500
0.9820 0.9883
27 10
26 12
18.42 10.02
10-500
0.9950
10-500
0.9869
28
20
1-500
0.9967
1-500
0.9945
13
13
pyrene iprodione lindane methidathion
2.2
methoxychlor
5..08
7
168
122
137
5.53 3
98
83
91
102
0.45 0.41
1.51
1.37
metolachlor
3.13
38
160
139
133
0.53 0.01
1.78
0.05
1-500
0.9878
1-500
0.9872
23
15
molinate
3.21
43
68
98
97
0.72 0.81
2.39
2.7
1-500
0.9956
1-500
0.9850
11
20
musk ketone
4.3
90
176
136
137
0.06 1.07
0.19
3.56
1-500
0.9870
10-500
0.9829
18
14
musk xylene
4.4
92
147
114
123
2.54 1.86
8.45
6.19
10-500
0.9980
10-500
0.9910
21
23
naphthalene oxybenzone
3.3 3.79
48 99
132 181
131 136
103 146
0.15 0.02 0.07 0.04
0.51 0.23
0.06 0.14
1-500 1-500
0.9881 0.9857
1-500 1-500
0.9948 0.9898
13 15
16 9
2640
dx.doi.org/10.1021/ac102909g |Anal. Chem. 2011, 83, 2638–2647
Analytical Chemistry
ARTICLE
Table 1. Continued precision RSD (%) N=5
linearity (ng/L)
compound
log Kow
recoveries in
MDLs
MQLs
matrix (%) N = 3
(ng/L)
(ng/L)
theoretical
WWE
recoveries
diluted
range
R2
range
R2
rep.
repr.
WWE
WWE
(%)
WWE
1:4
river
phantolide
4.9
97
143
141
149
1.08 0.98
3.59
3.27
5-500
0.9920
5-500
0.9990
27
25
phenanthrene
4.46
93
150
111
133
0.02 0.01
0.06
0.05
1-500
0.9943
1-500
0.9860
14
12
0.28 0.06
0.93
0.19
1-500
0.9887
1-500
0.9910
22
20
0.11 0.03 0.37 0.11 10.2 8.37 33.4 27.9
1-500 20-500
0.9834 0.9841
1-500 10-500
0.9880 0.9891
15 22
10 17
procymidone
3.08
36
173
146
124
pyrene simazine
4.88 2.18
97 7
135 152
120 108
132 131
WWE river WWE river
WWE
river
terbutryn
3.74
72
42
54
138
0.16 0.45
0.52
1.5
1-500
0.9978
1-500
0.9930
18
9
terbutylazine
3.21
43
150
130
133
0.64 0.10
2.13
0.32
1-500
0.9763
1-500
0.9980
23
11
tonalid
5.7
100
184
106
142
0.07 0.04
0.22
0.12
1-500
0.999
1-500
0.9860
17
14
traseolide
5.4
99
134
101
139
0.30 0.29
0.99
0.98
1-500
0.9932
1-500
0.9900
24
23
triclosan
4.76
96
170
147
144
0.12 0.06
0.39
0.21
1-500
0.9947
1-500
0.9909
14
5
trifluralin
5.34
99
76
95
130
2.25 1.20
7.5
3.99
10-500
0.9888
5-500
0.9897
15
13
vinclozoline
3.1
37
194
168
108
1.14 0.95
3.8
3.18
5-500
0.9861
5-500
0.9885
15
14
a
BHT, butylated hydroxytoluene; 4-MBC, 4-methylbenzyledene camphor; 2-EHMC, 2-ethylhexyl methoxycinnamate; WWE, wastewater effluent; rep., repeatability; repr., reproducibility
which accessed to a 60 L liquid nitrogen storage tank. The first column was a 10 m x 0.18 mm i.d., 0.2 μm film thickness Rtx-5 coated with 5% diphenyl 95% dimethylpolysiloxane from Restek, (max. prog. temp.: 350 °C). As second column a 1 m x 0.1 mm i. d., 0.10 μm film thickness Rxi-17 coated with 50% diphenyl 50% dimethyl polysiloxane from Restek (max. prog. temp.: 320 °C) and a 1 m x 0.1 mm i.d., 0.10 μm film thickness coated with a liquid crystalline phase Rt-LC50 (max. prog. temp.: 270 °C), also from Restek, were used. The MS system was a Pegasus 4D TOF from LECO Corporation (St. Joseph, MI, USA). For instrumental analysis details see Supporting Information. Method Performance. Recoveries, linearity range, sensitivity, precision and matrix effects were all calculated. For method performance details see Supporting Information. Blank Issues. Blank contamination is a common problem observed in the determination of UV filters, musk fragrances, antioxidants and PAHs at trace levels.11,15,30,34,35 Thus, precautions were taken to prevent contamination from personnel, organic solvents, equipment and glassware.10 Blank assays were performed routinely using the same procedure as above, employing MiliQ water samples, to check for laboratory background levels of the studied compounds. Though the detected amounts of the target compounds were low (below 5 ng/L), it was considered necessary to subtract the quantitative values of the compounds found in the blanks.
’ RESULTS AND DISCUSSION SBSE Extraction. Three parameters affecting the extraction conditions were studied: the addition of NaCl, the addition of MeOH and the extraction time. Polar solutes (Ko/w < 3.3) normally result in very low recoveries in SBSE using PDMS fibers. Salting out is a technique usually employed in SBSE extraction to achieve higher polar
solutes recoveries; NaCl is added to reduce the analytes’ water solubility and increase the partitioning coefficient between PDMS fiber and water.13,15,31,32,36 In this study, the effect of NaCl addition to the sample on SBSE efficiency was evaluated at three different levels: no salt addition, at 10 g, and at 20 gr. The greater ionic strength, the addition of 20 g NaCl, increases the signal intensities for more polar compounds, especially for simazine, atrazine, terbutylazine, vinclozoline, iprodione, procymidone, and naphthalene. On the other hand, most hydrophobic compounds show lower extraction efficiency when ionic strength increases, because of their adsorption onto the glass walls and other surfaces, as has been previously verified.13,15,31,36 The use of methanol to avoid the adsorption phenomenon has shown excellent results and was evaluated for this application. The influence of MeOH addition on extraction efficiency was tested at three levels: 5, 10, and 20%. As expected, the addition of methanol to the water samples increased the signal intensities of the most hydrophobic compounds, as was the case with the pesticides aldrin, DDT, DDE or trifluralin, and the most apolar PAHs and PCPs. The best results were obtained with the addition of 10% MeOH - higher MeOH concentrations decrease the response of most hydrophilic compounds, and no significant enhancement in extraction efficiency was observed for the hydrophobic compounds. With regard to extraction time, some studies have reported that extraction periods of 11-14 h can be considered as optimum for most of the analytes to reach the steady state.13,31 Shorter extraction periods can also be considered without an important decrease in detectability, but if all the analytes are measured at equilibrium conditions, better precision is obtained.13,15 In our case, a stirring time of 14 h (overnight) and a speed of 900 rpm were selected as optimum conditions for increasing method sensitivity and achieving sorption equilibrium 2641
dx.doi.org/10.1021/ac102909g |Anal. Chem. 2011, 83, 2638–2647
Analytical Chemistry
ARTICLE
Figure 1. GCxGC-TOFMS contour plots obtained using the different column sets and different oven temperature programs.
for all analytes. With longer extraction periods (24 h) an improvement in the response of most of the compounds was not observed. Performance of the SBSE-GCxGC-TOF-MS Method. Validation results, determined in wastewater effluent and river water extracts, are presented in Table 1. By applying the SBSE-GCxGC-TOF-MS technique, the method detection limits (MDLs) for more than 75% of the compounds tested were lower than 1 ng/L in both wastewater and river water. Values obtained ranged from 0.02 to 2.5 ng/L and 0.01 to 2.15 in wastewater effluent and river water, respectively; except for the most polar analytes: atrazine, iprodione, methidation, and simazine in wastewater effluent, which exhibited higher but still very good detection limits, 3, 8.6, 5.5, and 10.2 ng/L, respectively, and simazine in river water (8.4 ng/L). The detection limits achieved were lower to those obtained in a previously published method using SPE, followed by GCxGC-TOF-MS for the analysis of contaminants in waters,27 and similar, or lower in many cases, to those obtained in others published methods for these contaminants in waters.11,12,14 It should be underlined that, even at this low level, the analytes can be reliably determined in environmental matrices and unambiguously identified by means of their full mass spectra. The method quantification limits (MQLs) were from 0.06 to 8.45 ng/L and 0.05 to 7.17 in wastewater effluent and river water, respectively, except for the four analytes commented on previously that were slightly higher. The calibration curves obtained in the present study were linear over the entire range studied; with correlation coefficients in the range of 0.981 to 0.999 (see Table 1). Precision in the chromatographic response was determined in terms of repeatability and reproducibility. The RSDs were, in all cases, lower than 28%. In Table 1, the RSDs obtained for wastewater effluent
(the most complex matrix) are summarized. The RSDs in river matrix were between 3% and 23%, in all cases were similar or lower than in wastewater effluent matrix. Table 1 shows the log Kow and theoretical recoveries of the compounds investigated in this work (see Supporting Information). In general, it can be observed that for hydrophobic analytes (log Kow > 3.5), the extraction efficiency of SBSE is higher than 70%. On the other hand, for the most polar analytes, the recoveries are lower. However, as we have mentioned in the previous section, with the addition of NaCl the signal intensities of these polar compounds increase. As with many other sample preparation techniques for trace analysis, SBSE efficiency can be greatly affected by the complexity of the matrix involved.11 For this reason, recoveries have been calculated in wastewater effluent and river water by comparing them with the area of standard compounds from Milli-Q water. Two phenomena were observed: (i) a decrease in signal intensity compared to the MilliQ water samples and (ii) an increase in the signal intensity. (i) This signal suppression can be explained by the presence of other components in the sample, which may compete with the target compounds for adsorption sites of the stir bar, thus potentially reducing the extraction efficiency. As Table 1 shows, in river water matrix, no significant matrix suppression was observed. Only six compounds have mean recoveries in the range 53-69%. These compounds were all PAHs and were among the most hydrophobic target compounds, except for iprodione. This effect has also been observed in other studies.12,37,38 Some authors explain this phenomenon as the increased affinity of the analytes for lipophilic substances in the matrix.37 With respect to the more complex matrix, wastewater effluent, we can see a recovery decrease of more than 75% for eight compounds. These compounds were also among the most hydrophobic target 2642
dx.doi.org/10.1021/ac102909g |Anal. Chem. 2011, 83, 2638–2647
Analytical Chemistry compounds. (ii) On the other hand, in gas chromatography, the sample matrix can cause an enhancement in the observed chromatographic response for analyte residues in matrix extract compared to the same concentration in a matrix-free solution;11,39,40 this phenomenon is called “matrix-induced chromatographic response enhancement”. The problem originates from the irreversible adsorption of certain sample components on active sites in the GC system, especially in the injection linear. The compounds prone to matrix-induced chromatographic enhancement effects are either thermolabile or rather polar, and they are typically capable of hydrogen bonding.10,32 As can be observed in Table 1, in river water, the signal enhancement percentage was lower than 35% for most of the compounds. However, for wastewater effluent, the enhancement in the chromatographic response for some of the analytes was higher, in the range from 20% to 74%. A number of approaches have been developed to compensate for matrix effects in SBSEGC-MS. The use of surrogates from the beginning of the analytical procedure has been especially useful, but finding a suitable isotopically labeled surrogate for each analyte can be a difficult task. In addition to this, in certain cases, this approach does not compensate for matrix effects.13 Another option is the time-consuming and laborious standard addition method.38 The most direct means is through reduction of matrix components. A simple solution to this problem is dilution of the sample. In this work, we have sequentially diluted, with Milli-Q water, a spiked wastewater effluent extract (1:2, 1:3, 1:4, and 1:10), and the signal intensities were compared to those obtained for spiked Milli-Q water at the same concentration. A dilution of 1:4 was shown to be sufficient to minimize the matrix effects for most of the compounds (see Table 1); hence we have used this dilution factor for the analysis of the wastewater effluent samples. In addition, matrix-matched calibration was used for quantitative purposes in order to minimize these effects.11,39,40 GCxGC Separation. Compared with 1D-GC, the application of GCxGC in this type of analysis resulted in improved chromatographic resolution both in terms of (i) separation of individual pollutants from each other and (ii) separation of pollutants from matrix components. To obtain a satisfactory separation for reliable monitoring of the waters, the GCxGC separation conditions have to be well-chosen. Therefore, it is important that the target compounds can be distributed over almost the entire GCxGC plane. To fulfill the orthogonality principle,19 it is necessary to choose columns that provide independent separation mechanisms in the first and second dimensions. Normally, in GCxGC, the analytes are separated by volatility in the first dimension and by polarity in the second dimension, combining a nonpolar column and a polar column. For this application, we chose a nonpolar primary column (an Rtx-5 column) for the volatility-based separation. The choice of the second column was more critical, a very polar column (LC-50) and a medium polar column (Rtxi-17) were evaluated and compared. In Figure 1, the contour plot chromatograms are shown, obtained using the different column sets and different oven temperature programmes. Despite good selectivity and wide spreading of spots on the 2D separation map achieved with the most polar phase (see Figure 1a), this phase was not stable enough to withstand the high temperatures required to elute the heaviest compounds even though it withstood high temperatures to be a polar column. Furthermore, with this column set, the analytes elute at high retention times in the second dimension, and when using a five-
ARTICLE
second modulation time (which is the one selected to draw a good peak shape and obtain the best sensitivity) not all compounds elute from the chromatographic system in their own modulation time, but rather in the next modulation cycle, thus causing so-called wrap-around.19 This effect is illustrated in Figure 1a, which shows the PAHs phenanthrene, pyrene and pyrene D10, which should elute at the highest second dimension retention time, but instead elute in the next modulation cycle at the beginning of the chromatogram. Consequently, the experiments were carried out with a more temperature-resistant phase, the Rtxi-17: this is less polar than the LC-50 phase, but a good separation can be achieved selecting the appropriate separation conditions, as is shown in Figure 1b and 1c. The oven temperature control allows easy fine-tuning of the GCxGC system. Changing the oven temperature program, it is possible to increase separation between the analytes; in Figure 1b, the analytes do not occupy the entire region of the chromatogram, instead with the oven temperature program selected in Figure 1c, the analytes are distributed over almost the entire GCxGC plane. These separation conditions appear to have been well-chosen: the analytes are all separated from one another and all the compounds can elute, avoiding wrap-around - at a final temperature in the seconddimension column of 315 °C. In Figure 1, the contour plots show the ability of GCxGC to separate and resolve additional components in the second dimension that would otherwise be coelutions. This is the case with PAHs, a group of isomers that are almost always indistinguishable, and numerous coelutions are observed during the 1D-GC separation. Analysis of Target Compounds. Once the method is developed for the automatic identification of the studied compounds, the software generates a peak table at the end of the analysis with information about the target compounds. In addition to this, on selecting each compound, its deconvoluted mass spectra and mass spectral library results are shown. The positive identification criterion of the target compounds was based on: (a) first dimension retention time deviation (2 s for most of the analytes and second dimension retention time deviation (0.5 s, compared to that of a standard and (b) a match factor better than 800 and a similarity factor better than 600. The match factor is an indication of how well the acquired mass spectrum matches the analyte’s calibration/reference mass spectrum and the similarity factor is a MS match factor provided by the library search algorithm, which describes how well the library hit matches the peak using all masses (the number of both factors is between 0 and 999, where higher numbers mean a better fit) In Table S1 (Supporting Information), the first and second dimension retention times, the qualifier ions and the quantification masses for the target compounds are listed. From this table, it can be seen that excellent repeatability of retention times is achieved. It is interesting to add that, compared to 1D-GC with conventional MS detection, where identification is usually based on two to four qualifier ions, GCxGC-TOF-MS instead uses the complete mass spectrum for identification, resulting in improved detectability. Figure 2a shows the GCxGC-TOF-MS contour plot of the target compounds identified in a river water sample collected 20 m downstream of the Alcala de Henares WWTP emission point. This figure also shows the peak table obtained for this sample. A filter has been used so as to highlight only the target compounds. Notice that the software provides the means for creating group classifications. The peak table displays the name of the group assigned to each compound and, in the contour plot, compounds 2643
dx.doi.org/10.1021/ac102909g |Anal. Chem. 2011, 83, 2638–2647
Analytical Chemistry
ARTICLE
Figure 2. (a) Analysis of target compounds. GCxGC-TOFMS contour plot and peak table of a river water sample collected at 20 m downstream of the point of emission of the Alcala de Henares WWTP. It has been used a filter to see only the target compounds. The figure also illustrate the identification of diazinon and 2-ethylhexyl trans-4-methoxycinnamate (2-EHMC) with the library and reference spectrum. (b) Identification of nontarget compounds. GCxGC-TOFMS contour plot of a river water sample collected at 20 m downstream of the point of emission of the Alcala de Henares WWTP, showing over 2700 peaks at S/N = 10. It can be observe some examples of nontarget compounds identified in the sample. In addition, in the figure is shown the surface plot zoom and the spectra library searches of 4-chlorophenol and loratadine. 2644
dx.doi.org/10.1021/ac102909g |Anal. Chem. 2011, 83, 2638–2647
Analytical Chemistry
ARTICLE
Figure 3. Contamination status. Automatic searching of temporal and spatial contamination variation of organic contaminants.
belonging to the same group are marked with the same color. In this study, the compounds are classified into three different groups; pesticides (marked with a purple spot in the chromatogram), PCPs (with yellow spots) and PAHs (in red). The chosen pesticides and PCPs have varied physic-chemical properties. For this reason they do not possess ordered structure of the chromatograms as in case of compounds with similar physicochemical characteristic, they are distributed over entire chromatogram and not easily recognized. Thereby the identification of analytes using mass spectrometry becomes essential. For this river sample, sixteen target compounds were identified and quantified: five pesticides, eight PCPs and three PAHs. The figure illustrates how the pesticide diazinon and the U.Vfilter 2-ethylhexyl trans-4-methoxycinnamate are correctly identified in the river sample using the reference and library full mass range spectra; it can be observed that the deconvoluted mass spectra of the compounds in the acquired sample match perfectly with the reference and library spectrum. They have a match and library score higher than 900. It is important to add that the resolution reached with the GCXGC technique allows the separation of the target compounds from matrix components and this feature is an advantage for their easy identification in TOF-MS. In addition to the identification of the target compounds, the software of the instrument calculates the total area of the compounds and the identified target analytes are quantified automatically against the reference calibration curves. A manual review was made of all integrations and identifications to ensure accuracy. Screening of Nontarget Compounds. In target analysis, the increased number of peaks resulting from the sample matrix can be largely ignored during the data review. Therefore, a screening (nontarget) analysis is necessary to obtain an overview, as real as possible, of the sample constituents. Obviously, this approach
requires adequate instrumentation, able to acquire full massrange spectra with adequate sensitivity, and software to process a large amount of data in as automated a way as possible. The GCxGC-TOFMS software’s library search function automatically searches for analyte matches. All the peaks found are present in the peak table, which contains, among other parameters, the library search results, which include the compound’s name and mass spectral similarity factors. So, in addition to the identification and confirmation of the target compounds, all the data generated in the peak table are used to simultaneously highlight potential target compounds to be further added to the method. The compounds that will become visible and identified depends on the analyst. The library(ies) from which the software will identify the unknown spectra, the setting of S/N and also the minimal similarity factor all rest on the decision of the analyst. In this study, we set the minimum required similarity factor to 600; we found that, in almost all cases, it was the lowest similarity factor value for a compound to be identified. The signal-to-noise ratio setting is critically important when real samples are analyzed as the chromatograms have many interfering peaks laid down upon them. For river water samples, we have determined that the S/N ratio must be set to 10, otherwise we are not able to detect the target compounds and we also lose information regarding unknowns. For wastewater effluent samples, the S/N could be set to 30 or 50, since this is a matrix with more interfering peaks and the compounds are present at higher concentrations. Figure 2b displays the contour plot with all the compounds detected in the river water sample collected 20 m downstream from the Alcala de Henares WWTP emission point. Over 2700 compounds were found in this sample at S/N = 10. Reviewing the peak table and the two-dimensional chromatogram, we identified new compounds of interest for the assessment of river water pollution. Figure 2b shows examples of the identified contaminants in this river water sample. They were insecticides, 2645
dx.doi.org/10.1021/ac102909g |Anal. Chem. 2011, 83, 2638–2647
Analytical Chemistry pesticides, industrial products, antioxidants, antiseptics, pharmaceuticals, illegal drugs, degradation products, etc. The figure illustrates the potential of the GCxGC-TOF-MS technique to locate and correctly identify trace concentrations of two of these nontarget compounds: 4-chlorophenol and loratadine. Figure 2b shows a 3D contour plot zoom of the area where these two compounds elute along with the spectra library searches. In the case of 4-chlorophenol, this compound elutes at the beginning of the chromatogram, which is an area of the chromatogram where many matrix compounds coelute. Notice that the 4-chlorophenol peak is submerged among matrix interference compounds. However, the resolution reached with the GCXGC technique permits us to resolve this compound, which would otherwise overlap in 1D-GC. The figure shows that the deconvoluted mass spectra of 4-chlorophenol fit very well with the library mass spectra (a similarity factor of 954). Another example is the pharmaceutical loratadine. It is obvious that the area where this compound elutes (Figure 2b) contains a number of matrix components at high concentrations. However, the surface plot illustrates the ability of GCxGC to achieve a clear separation of loratadine from the matrix constituents, due to the varying selectivity of the compounds on the second-dimension column. The mass spectrum obtained was of very good quality and loratadine was identified with a high similarity match factor. Similar results were obtained for the other nontarget compounds identified in the sample. These compounds can be verified with the analytical standard and added to the list of target compounds to be quantified in future analysis. Automatic Evaluation of Pollution Variation in Space and Time. As well as obtaining a great deal of information regarding the contaminants present in a water sample, the fact that GCxGC is able to produce a two-dimensional “picture” or “fingerprint” of a sample opens up the opportunity for sample comparison protocols. The direct comparison of 2-D GCxGC “pictures”, or bubble plots, may be sufficient in showing differences between samples. The illustration in Figure 3 displays a comparison of the contamination status of different water samples in space and time. To visualize peak location and intensity in a two-dimensional image, a bubble plot was used. The bubble radius corresponds to the relative area of the peak represented. It should be noted that the bubble radius corresponds to the relative area of the peaks represented in a sample. For a real comparison between samples, we introduced a number of maximum radius sizes for the bubbles in each sample, which is proportional to the most concentrate compound in the sample. In addition, as we have seen previously, each class of compounds can be assigned a different color, allowing fast visual analysis of group-type classification. With this example, we can observe that the 2D-chromatogram facilitates analyte identification and the bubble plots are very helpful to rapidly detect which compound classes are present in the sample in greater proportion and concentration. Therefore, from this study it is evident that PCPs are the most abundant contaminants both in terms of concentration as well presence in all water samples. The pesticides and PAHs are present at much lower concentrations. Individually, the fragrance galaxolide was the compound found at the highest concentration in most of the cases, followed by another fragrance, tonalid. On the left of the illustration, four bubble plots are shown, corresponding to water samples collected from the WWTP effluent and three different zones along the river; 35 km upstream from the WWTP effluent emission point and 20 and 1150 m downstream from the WWTP effluent emission point. The fingerprinting of these
ARTICLE
samples is useful in evaluating the spatial contamination variation of a river. In this case, it is evident that the river is affected by the discharge of urban wastewaters. It can be seen that, as we expected, the contamination decreases the further one goes along the river from the point of emission. On the right of the illustration, we can observe that all the multidimensional information generated by the instrument can also be used by the researchers for contrasting samples and identifying, much more easily, the major differences between samples taken on different days. In this example, the temporal contamination variation of the WWTP effluent is evaluated. This information is very useful to quickly check the main sources of contamination of river water, the effectiveness operation of the WWTP, to indentify abnormal peaks contamination or to highlight the months of highest contamination. Furthermore, all this data provided the information necessary to recognize far more easily the chemical construction of samples and helped to rapidly select chemical markers of water contamination.
’ CONCLUSIONS In this study, SBSE-GCxGC-TOF-MS has been demonstrated for the first time to be a powerful tool in the analysis of priority and emerging contaminants in waters. Excellent results have been obtained in terms of separation efficiency and also in compound identification. The results have shown the high sensitivity achieved with SBSE-GCxGC-TOF-MS. The method detection limits for most compounds were well below 1 ng/L in the MS full scan mode, using only 100 mL of river water sample and 25 mL of wastewater effluent sample. TOF-MS data processing provides automatic peak findings using MS deconvolution and spectral searching against reference mass spectrum and different libraries. This is very useful for the analysis of target compounds and to screen for nontarget compounds or unknowns. Thus, besides the identification and quantification of the target compounds, other new contaminants have been identified in the water samples, such as cholesterol and its degradation products, pharmaceuticals, illegal drugs, industrial products as well as other pesticides and PCPs. In addition to this, we have used the GCxGC features to propose studies of comparison between the fingerprinting of different water samples. These kinds of analysis give valuable information about the contamination status of rivers and wastewaters. ’ ASSOCIATED CONTENT
bS
Supporting Information. Additional information about the analytes selected, sample preparation, instrumental analysis, and methods and a table showing first and second dimension retention times, qualifier ions, and quantification mass of the target compounds. This material is available free of charge via the Internet at http://pubs.acs.org.
’ ACKNOWLEDGMENT The authors acknowledge the Spanish Ministry of Education and Science (Programa Consolider Ingenio 2010 CE-CSD200600044) for economical support. María Jose Gomez acknowledges the “Juan de la Cierva” research contract from The Spanish Ministry of Science and Technology. 2646
dx.doi.org/10.1021/ac102909g |Anal. Chem. 2011, 83, 2638–2647
Analytical Chemistry
’ REFERENCES (1) European Parliament and Council 2000/60/EC establishing a framework for Community action in the field of water policy. Official Journal of the European Union L327, 1-72. (2) European Council, 2008. Common Position (EC) No. 3/2008. Official Journal of the European Communities, 1-15. (3) Blasco, C.; Pico, Y. TrAC, Trends Anal. Chem. 2009, 28, 745– 757. (4) Daughton, C. G.; Ternes, T. A. Environ. Health Perspect. 1999, 107, 907–938. (5) Plagellat, C.; Kupper, T.; Furrer, R; Alencastro, L. F.; Grandjean, D.; Tarradellas, J. Chemosphere 2006, 62, 915–925. (6) Andersen, J. A.; Muir, D.; Ueno, D.; Darling, C.; Theobald, N.; Bester, K. Environ. Toxicol. Chem. 2007, 26, 1081–1089. (7) Ternes, T. A.; Joss, A.; Siegrist, H. Environ. Sci. Technol. 2004, 38, 392A–399A. (8) Gros, M.; Petrovic, M.; Barcelo, D. Anal. Bioanal. Chem. 2006, 386, 941–952. (9) Gomez, M. J.; Ag€uera, A.; Mezcua, M.; Hurtado, J.; Mocholí, F.; Fernandez-Alba, A. R. Talanta 2007, 73, 314–320. (10) Gomez, M. J.; Martinez-Bueno, M. J.; Lacorte, S.; FernandezAlba, A. R.; Ag€uera, A. Chemosphere 2007, 66, 993–1002. (11) Gomez, M. J.; G omez-Ramos, M. M.; Ag€uera, A.; Mezcua, M.; Herrera, S.; Fernandez-Alba, A. R. J. Chromatogr. A 2009, 1216, 4071–4082. (12) Prieto, A.; Basauri, O.; Rodil, R.; Usobiaga, A.; Fernandez, L. A.; Etxebarria, N.; Zuloaga, O. J. Chromatogr. A 2010, 1217, 2642–2666. (13) Prieto, A.; Zuloaga, O.; Usobiaga, A.; Etxebarria, N.; Fernandez, L. A. J. Chromatogr. A 2007, 1174, 40–49. (14) Guitart, C.; Readman, J. W. Anal. Chim. Acta 2010, 658, 32–40. (15) Rodil, R.; Moeder, M. J. Chromatogr. A 2008, 1179, 81–88. (16) Marriott, P.; Shellie, R. Trends Anal.Chem. 2002, 21, 573–583. (17) Panic, O.; Gorecki, T. Anal. Bioanal. Chem 2006, 386, 1013– 1023. (18) Cortes, H. J.; Winniford, B.; Luong, J.; Pursch, M. J. Sep. Sci 2009, 32, 883–904. (19) Dall€uge, J.; Beens, J.; Brinkman, U. A. Th. J. Chromatogr. A 2003, 1000, 69–108. (20) Seeley, J. V.; Seelye, S. K.; Libby, E. K.; McCurry, J. D. J. Chromatogr. A 2007, 45, 650–656. (21) Pedroso, M. P.; Fonseca, L. A.; Ferreira, E. C.; Poppi, R. J.; Augusto, F. J. Chromatogr. A 2008, 1201, 176–182. (22) Vial, J.; Thiebaut, D.; Sassiat, P.; Beldean-Galea, M. S.; Gomez, M. J.; Gognon, G.; Mallipattu, S.; Teillet, B.; Bouzige, M. J. Chromatogr. Sci. 2010, 48, 267–273. (23) Oezel, M. Z.; Gogus, F.; Lewis, A. C. J. Chromatogr. A 2008, 46, 157–161. (24) Mondello, L.; Casilli, A.; Tranchida, P. Q.; Presti, M.; Dugo, P.; Dugo, G. Anal Bioanal Chem. 2007, 389, 1755–1763. (25) Banerjee, K.; Patil, S. H.; Dasgupta, S.; Oulkar, D. P. J. Chromatogr. A 2008, 1190, 350–357. (26) Semard, G.; Bruchet, A.; Cardina€el, P.; Bouillon, J.-P. Water Sci. Technol. 2008, 57, 1983–1989. (27) Matamoros, V.; Jover, E.; Bayona, J. M. Anal. Chem. 2010, 82, 699–706. (28) Ong, R.; Lundstedt, S.; Haglund, P.; Marriott, P. J. Chromatogr. A 2003, 1019, 221–232. (29) Korytar, P.; Leonards, P. E. G.; Boer, J.; Brinkman, U.A.Th. J. Chromatogr. A 2005, 1086, 29–44. (30) Bester, K. J. Chromatogr. A 2009, 1216, 470–480. (31) Sanchez-Avila, J.; Quintana, J.; Ventura, F.; Tauler, R.; Duarte, C. M.; Lacorte, S. Mar. Pollut. Bull. 2010, 60, 103–112. (32) Perez-Carrera, E.; Leon, V. M.; Gomez, A.; Gonzalez-Mazo, E. J. Chromatogr. A 2007, 1170, 82–90. (33) Giordano, M.; Fernandez-Franzon, M. J.; Ruiz, G. F.; Pico, Y. Anal. Bioanal. Chem. 2009, 393, 1733–1743. (34) Kawaguchi, M.; Ishii, Y; Sakui, N.; Okanouchi, N.; Ito, R.; Inoue, K.; Saito, K.; Nakazawa, H. J. Chromatogr. A 2004, 1049, 1–8.
ARTICLE
(35) Berset, J. D.; Kupper, T.; Etter, R.; Tarradellas, J. Chemosphere 2004, 57, 987–996. (36) Llorca-Porcel, J.; Matínez-Soriano, E.; Valor, I. J. Sep. Sci. 2009, 32, 1425–1429. (37) Benijts, T.; Vercammen, J.; Dams, R.; Tuan, H. P.; Lambert, W.; Sandra, P. J. Chromatogr. B 2001, 755, 137–142. (38) Serodio, P.; Cabral, M. S.; Nogueira, J. M. F. J. Chromatogr. A 2007, 1141, 259–270. (39) Saari, E.; Per€am€aki, P.; Jalonen, J. Microchem. J. 2007, 87, 113–118. (40) Poole, C. F. J. Chromatogr. A 2007, 1158, 241–250.
2647
dx.doi.org/10.1021/ac102909g |Anal. Chem. 2011, 83, 2638–2647