About 3.0 million people living under a typical tropical savannah climate in the Brazilian Federal District (FD) have faced an unprecedented water crisis. Considering the need for indirect reuse of wastewater for public supply, this work aimed to investigate FD water sources regarding the presence and risks of three contaminants of emerging concern: caffeine, carbamazepine, and atrazine. Samples from two current water sources (Descoberto and Santa-Maria Lakes) and two future water sources of the FD (Paranoá and Corumbá Lakes) were analyzed by solid-phase extraction followed by liquid chromatography coupled to hybrid quadrupole-time-of-flight mass spectrometry (UPLC-QTOF/MS). Method precision and accuracy were satisfactory and limits of quantification ranged from 0.37 to 0.54 ng/L. Higher concentrations were observed for caffeine in the future water sources (39 to 180 ng/L) followed by carbamazepine (5.4 to 25 ng/L) and atrazine (3.9 to 15 ng/L). The less-impacted water sources, in current use in the FD, present caffeine concentrations ranging from 4.8 to 32 ng/L and atrazine levels varying between 2.4 and 5.5 ng/L. Carbamazepine was not detected in these reservoirs. Environmental risk assessment indicates a possible risk for carbamazepine and atrazine, evidencing the need for further studies. No human health risk was depicted within the results.
The capital of Brazil, Brasília, is located in the Brazilian Federal District (FD) under a typical tropical savannah climate with distinct periods of precipitation and humidity. The winter is dry with approximately 120 days without rainfall, resulting in severe problems related to water scarcity and rationing. The most important drinking water systems of the FD (Descoberto Lake and Santa-Maria Lake production systems) have become insufficient to supply about 3.0 million people living in the region. Thus, several actions have already been taken by the Environmental Sanitation Company to improve water availability, such as the use of alternative low-flow water intakes, the constant policing of the water sources, and the minimization of water losses during production processes.
As a result of low rainfall rates for three consecutive years, combined with a lower water recharge and an intense water use, the region is experiencing the largest water crisis in its history. To alleviate this problem, several long-term alternatives were evaluated, two of which were selected for the expansion of the water supply system: the use of Corumbá and Paranoá Lakes as water sources. The former is located beyond the borders of the FD and receives effluents from wastewater treatment plants (WWTPs), either directly or through its tributaries, while Paranoá Lake is an urban water system that receives effluents from two important WWTPs of the FD, as well as urban drainage waters and contaminated waters from tributaries, some of them running through densely populated areas.
Under this new reality, the indirect reuse of water is significant [
As pointed out by Snyder and Bennoti [
Although Descoberto and Santa-Maria Lakes are the main water sources in the FD, most of the work involving the presence of emerging contaminants has been carried out in the waters of Paranoá Lake considering its historical importance and future multiple-use possibilities. Abbt-Braun et al. [
Descoberto Lake was previously investigated for the presence of caffeine, atrazine, atenolol, and DEET, an active ingredient in insect repellents [
In Brazil, the monitoring of contaminants in drinking water became mandatory only in 1977, with the publication of BSB Ordinance No. 56/1977, which recommends periodic determinations of 10 inorganic contaminants, 12 pesticides, and 14 organoleptic parameters. After successive revisions over time, water quality standards were gradually increased. Nowadays, determinations of 15 inorganic contaminants, 15 organic substances, 27 pesticides, 7 disinfectants and their by-products, and more than 21 organoleptic parameters are required, through Annex XX of Consolidation Ordinance No. 05/2017, published by the Brazilian Ministry of Health. Within the contaminants investigated in the present work, only atrazine is considered in the Brazilian legislation for raw and treated waters. For the other substances, i.e., caffeine and carbamazepine, there are still no initiatives in Brazil for their inclusion in a monitoring program of national proportions.
This work was motivated by the importance of emerging contaminants in situations of indirect water reuse and by the limited amount of information regarding this class of contaminants in Brazil, especially in the FD. Thus, the present work aimed to develop and apply a method based on the solid-phase extraction of caffeine, atrazine, and carbamazepine followed by the quantification by ultra-efficiency liquid chromatography coupled to a high-resolution hybrid mass spectrometer (quadrupole-time-of-flight). These chemicals were selected due to their use as tracers of anthropogenic discharges in surface waters [
Methanol and acetone (HPLC grade) were obtained from Scharlau Chemie SA (Spain). Formic acid, acquired from Sigma-Aldrich (USA), was used as mobile phase additive. Ultrapure water was produced in a Milli-Q Academic system (Millipore, USA). Caffeine (98%, CAS 58-08-2) and atrazine (99%, CAS 1912-24-9) were purchased from Fluka Analytical (USA), whereas carbamazepine (99%, CAS 298-46-4) was obtained from Sigma-Aldrich (USA). Stock solutions (200 mg L−1), prepared by the solubilization of appropriate amounts of each solid standard in methanol, were kept in amber glass bottles at -10°C. A mixed stock solution containing 400
Figure
Map showing the Federal District in Brazil and the sampling points selected. DL: Descoberto Lake, SL: Santa-Maria Lake, SR: Santa-Maria River, VG: Vargem-Grande Stream, MC: Milho-Cozido Stream, CL: Corumbá Lake, PL-C: Paranoá Lake (conventional uptake), and PL-E: Paranoá Lake (emergency uptake).
Although located in an environmental protection area, the Descoberto Lake basin suffers from several problems such as invasions of protected areas, high population densities, agricultural activities, and siltation. Santa-Maria Lake is considered the most protected water source of the FD due to its restricted access through the National Park of Brasília. Sampling points were also established in the tributaries of Santa-Maria Lake: Santa-Maria River (SR), Milho-Cozido Stream (MC), and Vargem-Grande Stream (VG).
Two sampling points were established in Paranoá Lake. This urban artificial lake was built in 1959 to generate electric power and to improve the microclimate of the future capital of Brazil, Brasília, but is currently used for recreation, sports, tourism, and fishing. Paranoá Lake also receives urban drainage waters, effluents from two wastewater treatment plants, and other diffuse contributions [
Finally, the sampling point CL is located in one of the branches of Corumbá Lake, an artificial lake formed for electric power generation. This lake faces similar problems to those suffered by Descoberto Lake.
A total of 25 samples were collected in different sampling campaigns. Most of them (60%) were from Paranoá Lake. In this case, nine samples were obtained from the conventional water intake point whereas six were obtained at the emergency intake point. Three samples were from Descoberto Lake, three were from Santa-Maria Lake, and one sample was from Corumbá Lake. The remaining three samples were collected in three tributaries of the Santa-Maria Lake. Table
Characteristics of the investigated samples and acronyms used in the present work.
Aquatic system | Acronym | Coordinates | Sampling depth (m) | Season |
---|---|---|---|---|
Paranoá Lake (Conventional) | PL-C | 15°47'36.9"S 47°47'22.9"W | 1, 5 and 10 | Rainy and dry |
Paranoá Lake (Emergency) | PL-E | 15°44'37.2"S 47°49'51.9"W | 1, 5 and 10 | Rainy |
Corumbá Lake | CL | 16°12'26.7"S 48°09'55.2"W | 0 | Dry |
Descoberto Lake | DL | 15°46'41.5"S 48°13'52.9"W | 9 and 16 | Rainy (only 16 m) and dry |
Santa-Maria Lake | SL | 15°40'33.2"S 47°57'19.6"W | 9 and 16 | Rainy (only 16 m) and dry |
Santa-Maria River | SR | 15°40'58.1"S 48°01'09.8"W | 0 | Dry |
Vargem-Grande Stream | VG | 15°40'23.4"S 48°01'11.3"W | 0 | Dry |
Milho-Cozido Stream | MC | 15°40'11.8"S 48°00'24.0"W | 0 | Dry |
Samples from different depths were collected using a Van Dorn water sampler and transferred to amber glass bottles (1 L). Surface samples were obtained directly into glass bottles. All bottles were previously cleaned in the laboratory and rinsed with the sampled water on site. Samples were transported to the laboratory on ice and preserved at 4oC until further preparation steps.
In the laboratory, samples were first passed through one or more glass fiber filters (GF-3, 0.7
Analyses were carried out using an Expert Ultra LC100 chromatographic system (Eksigent Technologies, USA) consisting of a binary pump, a vacuum degasser, a thermostated autosampler (LC100-XL), and a thermostated column oven, coupled to a hybrid quadrupole-time-of-flight tandem mass spectrometer (TripleTOF 5600+, Sciex, Canada) with a DuoSpray ion source interface operated in the electrospray ionization (ESI) mode. Nitrogen was produced by a high-purity generator (Genius 3010, Peak Scientific, USA) and used as source gas.
Separation was performed using a C18 column (Kinetex 2,6
Successive injections of a 100
Data acquisition was performed using the high-resolution multiple reaction monitoring (HR-MRM) mode. Firstly, precursor ions were selected by direct infusions of a 0.1% formic acid solution prepared in methanol containing 400
Acquisition parameters used in the UPLC-QTO/MS system.
Analyte | Formula | Exact mass (Da) | Precursor ion (Da) | Product-ion (Da) | DP (V) | CE (eV) | RT (min) |
---|---|---|---|---|---|---|---|
CAF | | 194.08037 | 195.0877 | 138.066 | 100 | 25 | 1.74 |
195.0877 | |||||||
110.0349 | |||||||
CMZ | | 236.094963 | 237.1022 | 194.096 | 70 | 30 | 2.27 |
237.1022 | |||||||
192.0808 | |||||||
ATZ | | 215.093773 | 216.1010 | 174.054 | 100 | 30 | 2.55 |
216.1010 | |||||||
104.0010 |
The mass spectrometry system was firstly calibrated using the CsI/ALILTLVS solution under direct infusion. Then, a CMZ solution was used for routine mass tuning on a daily basis. During the analyses, a mix solution was injected every five chromatographic runs for mass calibration. An error up to 2 ppm was considered acceptable. Formic acid was added to all working solutions to favor positive ionization of the target compounds.
Analytical curves were tested for the homogeneity of variances by the Cochran test. Outliers were verified by the Grubbs test. For the heteroscedastic data, a weighted least squares regression method was performed using weighting factors that produced the lowest sum of the relative errors [
The environmental risk was assessed by calculating the risk quotient (
For all analytes, the five-point analytical curves were heteroscedastic according to the Cochran test. Also, no outliers were depicted using the Grubbs test. The best weighting factor for caffeine and atrazine was
Work range and linearity of the external calibration curves.
Analyte | Work range ( | Weight | R2 | LOQ ( | S/N at LOQ |
---|---|---|---|---|---|
CAF | 0.48 to 300 | | 0,99 | 0.48 | 13.7 |
CMZ | 0.48 to 300 | | 0,99 | 0.48 | 20.6 |
ATZ | 0.48 to 300 | | 0,99 | 0.48 | 16.6 |
The limit of quantification (on-column) was admitted as the lowest concentration of the analytical curves for all analytes, i.e., 0.48
Table
Coefficients of variance for intraday analysis of mixed solutions of the analytes.
Concentration | Precision (%) | ||
---|---|---|---|
Atrazine | Caffeine | Carbamazepine | |
0.48 | 1 | 5 | 3 |
2.4 | 4 | 2 | 4 |
12 | 3 | 5 | 2 |
60 | 2 | 6 | 5 |
300 | 1 | 6 | 5 |
Precision was considered satisfactory for all analytes since coefficients of variance below 5% were observed, with the exception of caffeine, where values of 6% were obtained for the highest concentration levels. During the experiments, it was noticed that the standard deviation of the analytical curves was influenced by the ambient temperature, the cleaning and maintenance of the mass spectrometer orifice plate, the stabilization of the analytical signals, and mainly the constant calibration of the exact mass, which must be done frequently throughout the analysis. Thus, results shown in Table
The matrix effect (ME) was evaluated by plotting two curves: the first one obtained by the analysis of three solutions containing increasing concentrations of the analytes in methanol and the second one with the same concentrations of the analytes in a sample matrix, i.e., an extract of a Paranoá Lake sample collected at 1 m depth. Both curves were plotted with three concentration levels due to the small amount of sample extract available. Figure
Matrix effects on the analytical response of the investigated contaminants.
For all analytes, the matrix effect was manifested in order to attenuate the analyte response with a tendency to underestimate higher concentrations. Caffeine suffered less influence of the matrix (14% attenuation), followed by atrazine (18%) and carbamazepine (19%). The slopes were statically compared using Student’s
Percentage of recovery obtained for spiked extracts and for a Paranoá Lake sample spiked with all analytes.
Samples | Recovery (%) | ||
---|---|---|---|
Caffeine | Carbamazepine | Atrazine | |
Extract | 83±11 | 79±9 | 107±6 |
Extract | 81±8 | 77±9 | 74±7 |
Extract | 80±8 | 62±8 | 64±6 |
Natural water | 102±6 | 78±5 | 112±7 |
It is observed in Figure
Accuracy was also assessed by a recovery test for extraction efficiency. In this case, a sample of Paranoá Lake was enriched with 55 ng/L of each analyte and submitted to extraction and analysis using the weighted analytical curves described in Table
Method limits of quantification (LOQm) were expressed by the instrument LOQ (Table
Table
Concentrations of caffeine, carbamazepine, and atrazine in the current water sources of the Brazilian Federal District and in selected tributaries.
Analytes | Concentration (ng/L) | ||||||
---|---|---|---|---|---|---|---|
DL | SL | SR | VG | MC | |||
| | | | | | | |
CAF | 13 (D) | 32 (R) | 4.8 (D) | 10 (R) | 83 (D) | 75 (D) | 123 (D) |
CMZ | ND (D) | ND (R) | ND (D) | ND (R) | ND (D) | ND (D) | ND (D) |
ATZ | 5.5 (D) | 2.8 (R) | 3.4 (D) | 2.4 (R) | ND (D) | ND (D) | ND (D) |
CAF: Caffeine, CMZ: Carbamazepine, ATZ: Atrazine, DL: Descoberto Lake, SL: Santa-Maria Lake, SR: Santa-Maria River, VG: Vargem-Grande Stream, MC: Milho-Cozido Stream, ND: Not detected, R: Rainy season, D: Dry season
Only caffeine and atrazine were detected in Descoberto and Santa-Maria Lakes in all samples investigated. The concentration of the former was higher varying between 10 and 32 ng/L in Descoberto Lake and from 4.8 to 10 ng/L in Santa-Maria Lake.
Higher levels of caffeine in the Descoberto Lake are expected due to the occupation of adjacent areas by condominiums and by the increasing population density observed in the region in the last years. These factors may contribute to the presence of caffeine, a known tracer of human activities [
Atrazine levels varied between 2.4 and 5.5 ng/L in both compartments and probably arise due to minor diffuse sources related to agricultural activities in the surrounding areas. No significant differences were observed when results from different seasons were compared indicating that pollution processes may be stable over the year.
Table
Concentrations of caffeine, carbamazepine, and atrazine in the future water sources of the Brazilian Federal District.
Analyte | Concentration (ng/L) | ||||||
---|---|---|---|---|---|---|---|
CL | PL-C | PL-E | |||||
| | | | | | | |
CAF | 149 (D) | 77 (R) | 77 (R) | 50 (R) | 49 (R) | 121 (R) | 180 (R) |
CMZ | 8.5 (D) | 17 (R) | 25 (R) | 15 (R) | 5.4 (R) | 21 (R) | 9.0 (R) |
ATZ | 9.3 (D) | 9.4 (R) | 13 (R) | 5.8 (R) | 3.9 (R) | 15 (R) | 7.6 (R) |
CAF: Caffeine, CMZ: Carbamazepine, ATZ: Atrazine, CL: Corumbá Lake, PL-C: Paranoá Lake (Conventional uptake), PL-E: Paranoá Lake (Emergency uptake), R: Rainy season, D: Dry season
The three analytes investigated were found in all samples of the Corumbá and Paranoá Lakes. Concentrations of caffeine were higher in both lakes, followed by carbamazepine and atrazine. As expected, concentrations were also consistently higher in these reservoirs compared to the current water sources of the FD. The presence of contaminants in samples from different depths indicates the vertical mixture of the waters of the Paranoá Lake. However, no further tendency was depicted within the samples investigated.
No significant differences were observed in the results considering both sampling points of Paranoá Lake. Caffeine Levels varied between 39 and 180 ng/L in Paranoá Lake considering all investigated samples, corroborating previous reports regarding such contaminant in the lake. In the sampling point PL-C, Abbt-Braun et al. [
Carbamazepine concentrations ranged from 5.4 to 25 ng/L in the samples from Paranoá Lake. For Corumbá Lake, a concentration of 8.5 ng/L was obtained. Again, no significative differences were observed between sampling points, seasons, and depths investigated. A previous report also found carbamazepine in both sampling points of Paranoá Lake in concentrations varying from <5 to 16 ng/L. Atrazine levels, varying from 3.9 to 15 ng/L, were also similar to previous results corroborating with the scenario of contamination that has remained stable since 2010 in the FD.
Figure
Environmental and human risk assessments of selected emerging contaminants in future and current water sources of the Brazilian Federal District. Dark and light grey regions indicate risk and possible risk, respectively. Blank regions indicate no risk.
For the environmental risk assessment, risk quotients (
As PNEC is an estimate, a restrictive demarcation of what is “acceptable” or “not acceptable” is not possible for MEC values below or above this parameter, respectively. Therefore, for a more realistic risk assessment, it is considered that RQ greater than 1 may imply risk while values lower than 0.1 indicate no risk. Intermediate values indicate possible risk as well as the need for further studies [
For human risk assessment, it was also considered a worst-case scenario where removal efficiency during drinking water treatment processes was null. Water quality criteria were derived considering ADI data available in the literature for carbamazepine (300 ng/kg) [
A method based on the solid-phase extraction followed by quantification using liquid chromatography coupled to high-resolution hybrid mass spectrometry (UPLC-QTOF/MS) was developed and applied for the quantification of caffeine, carbamazepine, and atrazine in water sources of the Brazilian Federal District.
Accuracy was considered satisfactory considering matrix effects, as well as recoveries experiments, carried out with samples collected in Paranoá Lake. Precision was also satisfactory under weighted least squares regressions using the most appropriated weighting factors.
Concentrations of the investigated analytes were consistently higher in the future water sources as they receive urban drainage waters, effluents from wastewater treatment plants, and other diffuse contributions. As a result, possible environmental risks were depicted for carbamazepine in the future water sources. Atrazine levels in all water sources were also in a range of possible environmental risks. No risk for human health was estimated based on the worst-case scenario where removal in water treatment plants is not achieved.
Our results point towards a crucial role of indirect water reuse in situations of water scarcity and rationing. Receiving waters may contain several contaminants of recent concern that should be investigated to ensure the safe use of water for different purposes. Although risks to human health have not been evidenced in this work, our results may be useful for constructing a reliable contamination scenario to other alternative and more complete risk assessment models.
The data used to support the findings of this study are available from the corresponding author upon request.
The authors declare that there are no conflicts of interest regarding the publication of this scientific paper.
This research was supported by the Brazilian Funding for Innovation and Research (FINEP 01.13.0470.00) and by the Federal District Research Foundation (FAPDF 193.000.714/2016). The authors thank the Environmental Sanitation Company of the Federal District (CAESB) for the operational support during the sampling and the Analytical Center of the Institute of Chemistry at the University of Brasília for providing access to the UPLC-QTOF/MS used in this work.