Development of QuEChERS-based extraction and liquid chromatography-tandem mass spectrometry method for simultaneous quantification of bisphenol A and tetrabromobisphenol A in seafood: fish, bivalves, and seaweeds
- First Online:
- Cite this article as:
- Cunha, S.C., Oliveira, C. & Fernandes, J.O. Anal Bioanal Chem (2017) 409: 151. doi:10.1007/s00216-016-9980-3
- 265 Downloads
A liquid chromatography-tandem mass spectrometry (LC-MS/MS) method for the simultaneous determination of bisphenol A (BPA) and tetrabromobisphenol A (TBBPA) in different seafood samples was developed and validated. Sample preparation was based on a quick, easy, cheap, effective, rugged, and safe (QuEChERS)-based procedure through an extraction of target analytes with acidified acetonitrile (MeCN) added with inorganic salts (MgSO4, NaCl) followed by a liquid–liquid extraction (LLE) using hexane-tertbutylmethyl ether/hexane-benzene to eliminate matrix co-extracts. The developed method promotes a better removal of interferences than that achieved with the classic QuEChERS procedure. The method was validated following the guidelines of the European Union (EU) for relevant seafood matrices such as fish, mussel, and seaweed. Accuracy (81 % average of recovery), reproducibility (12 % average relative standard deviation for both intra-day and inter-day repeatability), and sensitivity for the target analytes (method detection limits of 0.07 ng/g wet weight (ww) and 0.06 ng/g ww for BPA and TBBPA, respectively) were evaluated for all the matrices studied.
KeywordsBisphenol A Tetrabromobisphenol A LC-MS/MS QuEChERS Fish Contaminants
Global consumption of seafood has increased since the 1950s and, despite nature’s limitations, it is expected to keep growing . At the same time, public awareness and concern about seafood safety has also shown a marked increment, most of which related to health risks derived from the possible presence of emerging contaminants. Currently, more than 700 emerging contaminants, categorized into 20 classes, are listed as present in the European aquatic environment . These are chemicals that are not commonly monitored in the environment but which have the potential to enter the environment and cause adverse ecological and human health effects , like bisphenol A (BPA) and tetrabromobisphenol A (TBBPA).
BPA is an industrial chemical used for more than 50 y to produce polycarbonate plastics, epoxy resins, and coatings used to line metallic food and beverage cans. Owing to the massive use and emissions of BPA derived products, BPA (log kow 2.2–3.4) is widespread in the global environment. It is commonly found in sediment , sludge , and environmental water samples , as well as canned beverages and milk  and other canned food products [8, 9]. Despite being extensively studied, information regarding BPA data in biota is scarce , probably due to the absence of legislation regarding their presence on biotic material. Two recent surveys performed by Liao and Kannan in the United States  and China  reported mean levels of BPA in seafood of 3.23 ng/g (n = 23) and 14.1 ng/g (n = 11), respectively. Toxicological studies on animals showed that BPA has estrogen activity that can disrupt physiological functions of animals and humans, especially in the reproductive and endocrine systems [13, 14].
TBBPA is a derivative of BPA produced by the bromination of BPA in the presence of a halocarbon solvent with water or 50 % hydrobomic acid or alkyl monoethers, methanol, and acetic acid . TBBPA can be used as flame retardant to improve fire safety of electrical and electronic equipment such as printed circuit boards, computers, and TVs. Once released into the environment, TBBPA (log kow 3.2–6.4) will likely distribute into sediment and soil, retained to the organic fraction of particulate matter, and to the lipid fraction of biota . Although it has been found in samples of soil , sediment , sludge , and biological samples , only few scientific works have been reported on TBBPA presence in seafood . Ashizuka et al.  reported quantified values ranging from 0.01 to 0.11 ng/g wet weight (ww) for TBBPA in 45 samples of seafood. TBBPA had been reported to affect thyroid hormones and neurological function . The structural similarity of TBBPA with the thyroid hormone thyroxine (T4) suggested that TBBPA can act as potent competitor for T4 binding to human transthyretin and so it may have thyroid-disrupting effects .
The growing release of emerging contaminants into the aquatic environment requires the development and validation of suitable and accurate analytical methods, in order to enable their correct monitoring and the adoption of appropriate measures when levels of concern are found. So far, only few methods have been developed for the simultaneous analysis of TBBPA and BPA in fish samples and other seafood samples such as bivalves and seaweeds [4, 15]. Bivalve ability to filter incredible amounts of water may significantly increase its risk of contamination with harmful compounds. Seaweeds are also considered good biomarkers of environmental contamination. From an analytical perspective, determining TBBPA and BPA in seafood represents a challenging task not only because of the complex chemical composition of the matrices, with a high content of lipids that can be difficult to remove through sample preparation and further cause chromatographic disturbances and instrumental damage, but also because only trace levels of target analytes are likely to be present. An additional difficulty might be observed when a common pretreatment process (preservation, extraction, and cleanup) is considered for different matrices. Therefore, efficient extraction and cleanup procedures are necessary to overcome the matrix effect and to allow the concentration of the analytes in the final extracts.
Typically, TBBPA is isolated from biotic samples through Soxhlet extraction [22, 24], pressurized liquid extraction (PLE) , or accelerated solvent extraction (ASE) , with a variety of organic solvents, typically hexane, acetone, dichloromethane, or their mixtures. These extraction methods are usually followed by a cleanup step using solid-phase extraction (SPE) , gel permeation chromatography , neutral or acidified silica , or sulphuric acid treatment . Compared with TBBPA, considerably fewer analytical methods have been published for determination of BPA in aquatic biota; once again, solvent extraction followed by SPE is often used in these cases [28, 29]. Most recently, an alternative extraction derived from the QuEChERS procedure, originally developed for the analysis of pesticide residues in low fat-high moisture matrices, has been employed for extraction of TBBPA in fish  and BPA in canned seafood . The technique involves, in both cases, a micro scale extraction using acetonitrile in the presence of inorganic salts (magnesium sulphate and sodium chloride) followed by a dispersive SPE clean-up and, only in the BPA work, by a dispersive liquid–liquid microextraction (DLLME). The main advantages of this procedure are the simplicity, high throughput sample handling, and good recoveries obtained. Moreover, QuEChERS is a very versatile procedure that allows working at different pH, and several modifications can be performed at the cleanup step.
Concerning the detection techniques, most of the methods for TBBPA analysis are based on liquid chromatography (LC) coupled to tandem mass spectrometry (MS/MS) . In contrast, BPA is often analyzed by gas chromatography coupled to mass spectrometry (GC-MS) after a derivatization step [8, 9, 31], although some authors have also used LC-MS/MS  or LC with fluorescence detection .
All the procedures mentioned above provide acceptable recoveries for a wide range of analyte/matrix combinations, but the achievable limit of quantification (LOQ) is typically in the range 0.1–2.0 ng/g ww, which can be restrictive when looking into the environmental distribution and fate of the low amounts of BPA and TBBPA.
The main aim of the presented study was to develop an innovative LC-MS/MS method for accurate determination of both TBBPA and BPA in different biota matrices at ultra-trace level (LOQs ≤ 0.1 ng/g ww), enabling the formation of database needed to estimate dietary intake as well as to assess the distribution/bioaccumulation of emerging contaminants in the aquatic biota.
Standards and reagents
Bisphenol A (BPA; 99 % purity) and tetrabromobisphenol A (TBBPA; 99 % purity) were purchased from Sigma-Aldrich (West Chester, PA, USA). Tetrabromobisphenol A ring-13C12 (TBBPA13C12; 99 % purity) and d16-bisphenol A (BPAd16; 98 atom % D) used as internal standard (I.S.) were purchased from Sigma-Aldrich and Cambridge Isotope Laboratories, Inc. (Tewksbury, MA, USA), respectively. Individual standard solutions of the BPA and TBBPA were prepared in methanol (MeOH, HPLC grade from Sigma-Aldrich) at concentrations of 2000 μg/L. Working mixture solutions of 1000 μg/L were prepared in ammonium acetate:methanol (10:90), a solvent used in the extraction. Individual working solutions of BPAd16 and TBBPA13C12 (1000 μg/L) were also prepared in ammonium acetate: methanol (10:90). All the solutions were stored at –20 °C.
QuEChERS solvents acetonitrile (MeCN, gradient grade for HPLC; 78.6 % purity) and anhydrous magnesium sulfate (anhydrous MgSO4; 99.5 % purity) were purchased from Sigma-Aldrich; formic acid, hydrochloric acid (HCl), sodium chloride (NaCl; 99.5 % purity), and ammonium acetate (97 % purity) were purchased from AppliChem Panreac ITW Co. (Barcelona, Spain). To ensure efficient removal of phthalates and residual water, anhydrous MgSO4 was treated for 5 h at 500 °C in a muffle furnace. LLE solvents n-hexane (gradient grade for HPLC), tert-butyl methyl ether (MTBE, pro-analysis), and benzene (pro-analysis) were purchased from Merck (Darmstadt, Germany). Formic acid (99–100 %) and chloride acid both analytical grade were purchased VWR Int. (Radnor, PA, USA). Ultra-pure Milli-Q water was obtained using a Millipore Milli-Q system (Millipore, Bedford, MA, USA) and MeOH (MeOH, for HPLC LC-MS grade) was purchased from VWR.
Instrument and analytical conditions
A high-performance liquid chromatography (HPLC) system Waters Alliance 2695 (Waters, Milford, MA, USA) was interfaced to a Quattro Micro triple quadrupole mass spectrometer (Waters, Manchester, UK). Chromatographic separation was achieved using a Kinetex C18 2.6 μ particle size analytical column (150 × 4.6 mm) with pre-column from Phenomenex (Tecnocroma, Portugal), at a flow-rate of 200 μL/min. The column was kept at 30 °C and the autosampler was maintained at ambient temperature (±25 °C). Mobile phase consisted of 90 % MeOH and 10 % aqueous solution of 5 mM ammonium acetate (pH 5), isocratic. Total run time was 15 min. The sample injection volume was 20 μL.
MS/MS acquisition was operated in negative-ion mode with multiple reaction monitoring (MRM), the collision gas was argon 99.995 % (Gasin, Portugal) with a pressure of 2.9 × 10−3 mbar in the collision cell. Capillary voltages of 3.0 KV were used in the negative ionization mode. Nitrogen was used as desolvation gas and cone gas was the flow of 350 and 60 L/h, respectively. The desolvation temperature was set to 350 °C and the source temperature to 150 °C. Dwell times of 0.1 s/scan were selected. The data were collected using the software program MassLynx4.1.
LC-MS/MS parameters to analysis of BPA and TBBPA, with the IS BPAd16 for BPA and TBBPA13C12 for TBBPA determination
Retention time (min)
Precursor ion (Da)
Product ions (Da)
Cone energy (V)
Collision energy (kV)
Dwell time (ms)
Sardine, mussel, canned tuna, canned sardine, and canned mackerel samples were randomly collected in local markets in the city of Porto, Portugal. All edible content of 25 specimens (or around 400 g) were pooled, homogenized with a grinder (Retasch Grindomix GM200; Germany), and frozen at –80 °C. Then the samples were freeze-dried for 48 h at –80 °C and low pressure (around 0.010 mBar, Telstar Cryodos; Lane Bristol, UK), homogenized as above, and maintained at 4 °C until analysis. Dried seaweeds (Laminaria digitata) were purchased from organic grocery stories in the city of Porto, Portugal.
Sample preparation entailed the following steps for all the samples: (1) weigh 2 g of freeze-dried sample into a 40 mL amber glass vial tube, add 80 μL of TBBPAc13 (IS, 1000 μg/L), and 80 μL of BPAd16 (IS, 1000 μg/L); (2) add 7 mL of deionized water and 10 mL of 1 % HCl in MeCN, vortex, and place it on a wrist action shaker for 10 min; (3) add 4 g of anhydrous MgSO4 and 1 g of NaCl; (4) shake vigorously by hand for 5 min; (5) centrifuge the tube at 4736 g for 3 min; (6) transfer 3 mL of the MeCN extract to a 20 mL vial tube and add 7 mL of deionized water; (7) add 4 mL of hexane:tertbutylmethylether (3:1 v/v); (8) shake gently by hand for 30 s; (9) centrifuge at 4736 g for 1 min removing the organic phase; (10) add 4 mL of hexane:benzene (3:1 v/v). Then, for fish samples (11), the organic phases were combined and evaporated to dryness using a gentle nitrogen stream at room temperature; for mussel and seaweed samples (11) the organic phases were combined, adding 500 mg of C18, 500 mg of PSA, and 1500 mg of MgSO4, vortexed 1 min, centrifuged at 4736 g for 3 min, and the top layer was evaporated to dryness using a gentle nitrogen stream at room temperature. Finally, the dry extracts were re-dissolved in 1 mL of mobile phase (100 μL of 5 mM ammonium acetate: 900 μL of MeOH pH <5) and analyzed by LC-MS/MS.
Quality assurance/quality control
In order to avoid any kind of contamination, nitrile plastic gloves were used throughout the analytical work and the use of plastic materials was avoided. Amber glass vials were heated (400 °C) overnight prior to use. Using these precautions, no problems concerning levels in analytical blank samples were observed. Additionally, together with each batch of samples a procedural blank (i.e., water sample extracted as described above) was analyzed to assure the lack of background contamination.
To ensure correct identification and quantification of the target analyte, the following quality criteria were applied: (1) retention times should match those of the standard analytes within a window of ±0.05 min; and (2) relative intensity ratios of the selected ions should be within ±15 % of the expected/theoretical value.
Linearity was evaluated in seven calibration levels using blank sardine, mussel, and seaweed samples. The pool of blank samples (without the analytes of interest) used in calibration were obtained after a previous analysis of several commercial samples. Recovery and intra-day precision expressed as relative standard deviations (RSDs) were performed by adding to blank samples (sardine, mussel, and seaweed) standard solutions of the analytes, left at the room temperature to equilibrate during 2 h, and then submitted to the entire procedure. Results obtained for each analyte were compared with values obtained from similar samples added with known amounts of analytes after extraction. In both cases, the ISs were added after extracting just before LC-MS/MS analysis. For recovery testing, the spike concentrations were 1 and 10 ng/g dry weight (dw) to sardine samples and for both mussel and seaweed samples they were 10 and 40 ng/g dw; each level was analyzed six times. The lowest spiking level was selected to cover the concentrations close to the LOQs. Inter-day precision was assessed for two levels of concentration. Therefore, six spiked samples were extracted and analyzed on three different days for a period of 3 wk. The LOQs were estimated as the lowest matrix-matched standard, which provided a signal-to-noise ratio (S/N) higher than 10 for the quantification ion and higher than three for qualification ions. Limits of detection (LODs) were estimated as the lowest matrix-matched standard, which provided a signal-to-noise ratio (S/N) higher than three for both quantification ion and qualification ions.
Normal distribution of the residuals was evaluated through the Shapiro-Wilk’s test (sample size <50), and afterwards the dependent variable was studied using the Kruskal-Wallis H test. Furthermore, if a statistical significant effect was verified, Mann-Whitney U test was applied for means of comparison of more than two independent samples.
Statistical analyses were performed at a 5 % significance level, using SPSS software, ver. 22.0 (IBM Corp., New York, NY, USA).
Results and discussion
Individual standard solutions of 100 mg/L were prepared in methanol:5 mM ammonium acetate (pH <5) 90:10 v/v for MS optimization by infusion experiments. Full-scan mass spectra and MS/MS spectra were acquired in order to obtain the maximum number of available transitions for each analyte. For all the analytes, the best results in terms of sensitivity were always obtained with ESI in the negative ionization mode, using the [M – H]– as precursor ion. The two most sensitive selected reaction monitoring (SRM) transitions were selected for each analyte: the most abundant was used for quantification (Q), whereas the second one used was for confirmation (q). This method allowed us to reach the minimum number of identification points (IPs) required (three IPs for legally registered analytes) for a safe confirmation .
Selection of suitable mobile phase plays an important role in the ionization efficiency before the analytes enter the MS/MS system. Mixtures of MeCN/water or MeOH/water acidified or not with acetic or formic acid, and adding a buffer such as ammonium acetate, are common candidates as the mobile phase in LC analysis. Since TBBPA has pKa values of 7.5 and 8.5, the mobile phase was acidified with 5 mM of ammonium acetate (pH 5) to avoid its ionization. Moreover, acidification has a positive effect regarding the prevention of microbial growth in the aqueous mobile phase. Methanol–5 mM ammonium acetate and acetonitrile–5 mM ammonium acetate were investigated as mobile phases. Results showed that when using methanol–ammonium acetate 5 mM, the analytical response of the target analytes was higher than when using acetonitrile–ammonium acetate 5 mM. Several mobile phase compositions and gradient elution programs were tested, and the best separation was obtained using a mixture of 90 % methanol and 10 % of an aqueous solution of 5 mM ammonium acetate (pH 5) in isocratic mode at a flow rate of 200 μL/min.
Sample preparation optimization
In the original QuEChERS procedure, extraction is achieved through the use of MeCN and the addition of anhydrous MgSO4 and NaCl, followed by a dispersive solid phase (d-SPE) cleanup step using C18 and PSA as sorbents. This procedure with some modifications was recently used by Lankova et al.  to extract TBBPA and other compounds from fish followed by LC-MS/MS analysis. The authors used half of the sample (5 g) compared with the original QuEChERS and added formic acid to improve TBBPA recovery. Despite the acceptable LOQ obtained, 0.5 ng/g ww for TBBPA, lower levels are always desirable due to the extreme toxicity of this analyte even at very low levels (e.g., crucian carp revealed irreversible damage when exposed to 1.01 mg/L ). Regarding BPA, it was demonstrated that amounts as low as 10–100 μg/L are able to reduce the masculinization effect of trembolone, a potent toxicant that binds fish androgen receptors . Thus, to achieve the performance characteristics that we set initially (LOQs ≤ 0.1 ng/g for both TBBPA and BPA), the extraction procedure based on QuEChERS was optimized. In this study, the influence of several parameters on extraction efficacy, including H2O:MeCN ratio and the addition of formic acid or 10 M chloride acid, were tested. Experiments were achieved using mixtures of H2O:MeCN and freeze dried samples (sardine, mussel, and seaweed) spiked at a level of 500 ng/mL for each analyte in the final extract. In addition, aliquots spiked after the extraction process were used to compare the efficacy of the extraction. Water content in a sample is known as an important parameter, which considerably affects recovery of the extraction step . Commonly, for samples with less of 80 % moisture, the amount used for analysis had to be reduced and water added to accomplish more than 80 % by volume of added MeCN. Acetonitrile has been found to provide considerably better extraction efficiencies for apolar analytes (TBBPA log kow 3.2–6.4 and BPA log kow 2.2–3.4) [8, 30]. On the other hand, acetonitrile was capable of precipitating a significant fraction of the proteins, which is desirable. In this study, 1 g of freeze-dried sample (containing around 2 % of moisture) was augmented with 7 mL of H20 containing different % of formic acid (1, 5, and 10 %) and HCl 10 M acid (0.1, 1, and 5 %). Acidification was considered essential due to the pKa of TBBPA. It was shown that addition of formic or 10 M HCl acids significantly improved efficiency of the partition step for TBBPA without negatively affecting the BPA extraction. The best recoveries were achieved with 1 % HCl 10 M.
Average of recovery (%, n = 6), repeatability (%RSD, n = 6), inter-day repeatability, method limit of detection (LOD), and limit of quantification (LOQ)
Inter-day repeatability (RSD)
LOQ μg/kg dw
LOD μg/kg dw
Inter-day repeatability (RSD)
LOQ μg/kg dw
LOD μg/kg dw
1 (0. 1)
Inter-day repeatability (RSD)
LOQ μg/kg dw
LOD μg/kg dw
The linearity of the method using the matrix-matched standard (standards added to blank samples) was prepared as described in a section “Quality assurance/quality control”. Calibration curves were constructed by plotting the analyte/IS peak area ratio obtained against the concentration values. A linearity concentration range from 1 to 600 ng/g dw for both BPA and TBBPA in sardine, with correlation coefficients of ≥0.99 were obtained. The linear ranges for both BPA and TBBPA were 2.5–600 ng/g in mussel and seaweed matrix, with correlation coefficients of ≥0.99.
Table 2 shows the results obtained from the recovery experiments for two levels of concentration (for the concentration measured, see a section “Quality assurance/quality control”). Slight differences can be observed between the levels assayed and the matrix studied. Generally, better recoveries were obtained from fish than from mussel or seaweed. Overall, BPA recovery ranged from 67 to 82 %, and from 77 to 107 % for TBBPA. Inter-day repeatability values for BPA ranged from 9 to 18 %, and from 6 to 13 % for TBBPA. Intra-day repeatability ranged from 8 to 17 % for both BPA and TBBPA.
The results reported provide evidence that the optimized method achieves acceptable recoveries in line with criteria sets by European Union (EU) guidelines [38, 39]; average recoveries were in the range 70–120 %, with the only exception of BPA in seaweeds (67 %).
Methods LOQ and LOD obtained for the three matrices in the study are presented in Table 2. LOQ for BPA ranged from 0.5 to 25 ng/g dw (0.09 to 0.3 ng/g ww), whereas LOQ for TBPPA ranged from 0.25 to 2 ng/g ww (0.04 to 0.2 ng/g ww). LOD ranged from 0.2 to 0.8 ng/g dw (0.03 to 0.09 ng/g ww) for BPA and from 0.1 to 0.7 ng/g dw (0.01 to 0.06 g/g ww) for TBBPA. Overall, trace levels LOQ ≤ 0.1 ng/g ww were achieved for both TBBPA and BPA in fish and mussel, values that are lower than those reported in the literature [31, 40].
The developed method was applied to analyze different wild and canned samples collected in Porto city. Neither BPA nor TBBPA was detected in any of the wild samples analyzed, namely sardine, mussel, and Laminaria digitata. These results are consistent with those reported in literature  for the wild species under study, with the exception of the work by Ashizuka et al. , which found 0.11 ng/g ww of TBBPA in sardine. BPA traces were also detected in other wild marine species, with levels ranging from 0.5 (yellow seafin) to 1.1 (tongue sole) ng/g ww .
In canned samples of tuna and mackerel, both analytes were detected at levels of 23.93 ng/g ww and 15.25 ng/g ww for BPA, respectively, and 5.18 ng/g ww and 2.03 ng/g ww for TBBPA, respectively. In canned sardine, only TBBPA was detected at a level of 2.12 ng/g ww. The levels of BPA found are similar to those previously reported in canned tuna samples by Podlipna and Cichna-Markl , ranging from 6 to 67 μg/kg, and by us in a previous work, ranging from not detected to 99.9 μg/kg . However, slightly higher values were obtained by Thomson and Grounds , Munguía-López et al. , and Lim et al. , with levels ranging from <7.1 to 102.7 μg/kg, <20 to 109 μg/kg, and “not detected” to 116.9 μg/ kg, respectively, for canned tuna samples. None of the canned samples analyzed exceeded the migration level of BPA in food (600 μg/kg) established by European Food Safety Authority (EFSA) .
The presence of both BPA and TBBPA in canned seafood could be related to the migration of these monomers from cans coated with resins containing these compounds. According to the literature, these compounds can leach out of canned food during usage (e.g., high temperature, acidic or basic conditions, and physical damage can lead to the hydrolysis of plastic polymers into the estrogenic monomer) [40, 44].
The main conclusion of this study is that the use of a modified QuEChERS followed by liquid–liquid partition with hexane/tertbutylmethyl ether and hexane/benzene can provide an excellent method to extract TBBPA and BPA from different seafood matrices. The acidification of the acceptor solvent (acetonitrile) and the addition of inorganic salts to induce phase separation in QuEChERS allowed an efficient extraction of analytes from the solid donor phase. The cleanup capability of the liquid–liquid system used was superior to d-SPE, usually applied in the QuEChERS procedure, regarding the recovery rates of the analytes. Furthermore, organic phase resulting from the clean-up step was easily concentrated, which allowed better sensitivity compared with d-SPE. The performance of the whole QuEChERS/LLE–LC-MS/MS method showed suitability for an accurate quantification of both BPA and TBBPA at trace levels (LOQ ≤ 0.1 ng/g ww) in wild or canned fish, mussel, and seaweed samples. It should be emphasized that as far as we know, none of the previously published LC-MS/MS methods allowed such accurate analysis of BPA and TBBPA simultaneously. As mussel and seaweed are commonly used as bio-indicators of environmental contamination, this new method could be useful for elucidating the effect of distribution and fate of BPA and TBBPA in aquatic biota.
The research leading to these results has received funding from the European Union Seventh Framework Program (FP7/2007-2013) under the ECsafeSEAFOOD project (grant agreement no. 311820). S.C.C., C.O. and J.O.F. thank REQUIMTE, FCT (Fundação para a Ciência e a Tecnologia) and FEDER through the project UID/QUI/50006/2013 - POCI/01/0145/FEDER/007265 with financial support from FCT/MEC through national funds and cofinanced by FEDER, under the Partnership Agreement PT2020. S.C.C. acknowledges the FCT for the IF/01616/2015 contract.
Compliance with ethical standards
Conflict of interest
The authors declare that there is no conflict of interest.