Universal method to determine acidic licit and illicit drugs and personal care products in water by liquid chromatography quadrupole time-of-flight

Graphical abstract


Method details
Many different types of pollutants have been found in environmental compartments as water. Licit and illicit drugs or personal care products are some of the so-called emerging contaminants extensively used by humans [1,2]. A number of analytical methods are already available [3][4][5][6][7][8][9] to determine emerging contaminants in environmental matrices at low concentrations. However, these methods are only reported for one type of instrument. In this study, we proposed a procedure to analyse pharmaceuticals, illicit drugs, personal care products and others contaminants on different water matrices through a common method for a triple quadrupole (QqQ) and a quadrupole time-offlight (QqTOF) mass spectrometers.
Water used for preparation of calibration standards and LC-MS mobile phase was purified by an Elix Milli-Q system (Millipore, Billerica, MA, USA). Methanol was purchased from Panreac (Castellar del Vallès, Barcelona, Spain) and formic acid was purchased from Amresco (Solon, OH, USA). Ammonium fluoride was acquired from Alfa Aesar GmbH & Co KG (Karlsruhe, Germany).

UHPLC-QqTOF-MS/MS conditions
The chromatography was performed with an Agilent 1260 Infinity (Agilent, Waldbronn, Germany) using an Agilent Poroshell EC-C18 maintained at temperature of 30 C. A constant flow rate of 0.2 mL min À1 was used. The mobile phase consists of two solvents, 2.5 mM ammonium fluoride in methanol (as organic solvent) and 2.5 mM ammonium fluoride in water (as aqueous solvent). The UHPLC system was coupled to a hybrid QqTOF ABSciex Triple TOF TM 5600 (Framingham, MA, USA). The MS acquisition was performed using negative ionization (NI) and scan mass spectra between m/z 100-700 with the Turbo Ionspray source. The MS parameters were: ion spray voltage, 5000 V; declustering potential (DP), 120 V; collision energy (CE), 10; temperature 400 C with curtain gas (CUR) 25 (arbitrary units); ion source gas 1 (GS1) 50 and ion source gas 2 (GS2) 50. The QqTOF-MS/MS instrument was calibrated after every three samples using external reference compounds. The MS/MS acquisition was also performed using information-dependent acquisition (IDA) following operating parameters: declustering potential two (DP2), 110 V; ion release delay (IRD), 67 V; ion release width (IRW), 25 V; IDA MS/MS was performed at a fixed CE of 40 V, ions that exceeded 100 cps and ion tolerance of 50 mDa (isotopes higher than 4 Da were excluded). Data acquisition and processing was carried out using software Analyst (Framingham, MA, USA), Peak View 1.0 with the application XIC manager and MultiQuant 2.0.

Sampling
The developed method was applied to 21 influent and 21 effluent samples collected from three wastewater treatment plants (WWTPs) of metropolitan area of Valencia and 25 surface waters from Túria River. Wastewater samples were 24-h composite samples and river samples were grab ones. All samples were stored in polyethylene terephthalate (PET) bottles and once arrived at the laboratory, immediately frozen at À20 C until analysis to prevent degradation of contaminants.

Validation of the analytical method
Validation of the analytical method was performed partly according to the Commission Decision 2002/657/EC [11] and partly to the Eurachem guide [12] on that subject since none of them has a binding nature for water contaminants. Table 1 shows limit of quantification (LOQ), matrix effect (ME), recovery and relative standard deviation (RSD) obtained by UHPLC-QqTOF determination. The method provides LOQ between 1 and 150 ng L À1 , recoveries from 39% to 115%, matrix effects ranged from 6 to À52% and relative standard deviations (RSD) lower than 21%. The linearity was determined by calibration curves from LOQ-5000 ng L À1 in water-methanol (70:30) or as a matrix matched standards, with linear coefficients of determination (R 2 ) ! 0.99, except for salicylic acid (R 2 ) ! 0.98. Table S1 in Supplementary information depicts these parameters for UHPLC-QqQ. Table 2 shows the quantification of the selected analytes in the different water samples, as mean value AE RSD using QqQ and QqTOF instruments. The quantification of the detected compounds in the three matrices with QqQ was carried out according to the instrumental conditions previously reported [1] (see Table S2 in Supplementary information). The quantification of detected compounds with QqTOF was performed using MultiQuant 2.0 software. The results of QqQ and QqTOF were very similar, which confirms that the method is valid for both. Table 3 presents, mass (Da), adduct, extraction mass (Da), mass error (ppm), retention time (RT) and intensity of the selected compounds (spiked Milli-Q water with 100 ng L À1 ). The identification of target and non-target was carried out against the XIC manager Table with data of 1212 pharmaceuticals, 546 pesticides, 378 polyphenols and 233 mycotoxins. Furthermore, a total of 86 AE 9 pharmaceuticals, 2 AE 1 pesticides and 14 AE 3 other compounds were detected in influent samples; 45 AE 14 pharmaceuticals, 1 AE1 pesticides and 7 AE 3 other compounds were detected in effluent samples, and 20 AE 6 pharmaceuticals, 1 AE1 pesticides and 5 AE 3 other compounds in river water samples. Fig. 1 illustrates the identification of acetaminophen (paracetamol) and Fig. 2 of the non-selected hydrochlorothiazide to show the identification system capabilities. Fig. S1 in Supplementary information shows the extracted ion chromatogram of all substances present in water and the non-target compound identification of theophylline in influent wastewater sample.

Background
There are hundreds, even thousands of emerging contaminants that can occur in water. Traditionally, the scheme used for their determination involves generic sample preparation procedures able to extract almost any of them, and target determination for the unique and highly specific detection of the selected contaminant(s) [3][4][5]. This scheme is time-consuming (ca. 30 min each chromatographic run for a specific group of contaminants) and do not have versatility to detect unexpected emerging contaminants not selected for the target analysis. Currently, there are some reports of non-target detection through high resolution mass spectrometry that provide full scan Table 1 Method performance parameters: limit of quantification (LOQ, ng L À1 ), absolute recoveries (%), method repeatability (RSD, %) and matrix effect (ME, %) using QqTOF for effluent, influent and river water samples. Linearity: linear coefficients (R 2 ) were ! 0.99 in all cases, except for salicylic acid (R 2 ! 0.98); LOQ was established as the concentration that, after extraction, gives a UHPLC peak height value 1.0 Â 10 4 ; Recoveries and relative standard deviations (RSDs) of selected compounds were calculated in samples spiked at 100 ng L À1 subtracting the peak areas corresponding to native analytes in the sample and tested in quintuplicate; Matrix effect was evaluated by comparing the slope of the calibration curves obtained for spiked influent, effluent or surface water extracts with the slope of that obtained for standard prepared in water-methanol (70:30, v/v) spiked at the same level.
information as well as compound fragmentation (any m/z signal from the sample extract) [2,8]. However, high resolution mass spectrometer can provide inaccurate quantification [8] or enough Table 2 Comparison of the quantitative results obtained using the ABSciex TripleTOF TM 5600 (QqTOF) and a more traditional triple quadrupole (QqQ) for influent, effluent and river water samples.  sensitivity [2]. Latest generation instruments have improved their quantification possibilities as well as the identification capabilities of any unexpected substance by the application information dependent acquisition (IDA) modes that automatically provide MS/MS spectra of the most intense precursor ions (without previous selection) as an additional confirmation of the detected compounds [2].  The few examples of these broad screening systems are mostly focus on the positive ionization mode because there are more contaminants that ionized in positive mode and their MS sensitivity is higher. When mass spectrometry is combined with liquid chromatography (recommended for polar compounds as the emerging contaminants) the commonly used additives of the mobile phases (volatile salts and acids) enhanced the ionization in the positive ionization mode and inhibited it in the negative ionization one. Acidic contaminants, commonly better ionized by negative ionization are more difficult to detect and frequently the sensitivity does not reach the low levels emerging contaminants are present in water. Recently, Petrie et al. [9] demonstrated a substantial improvement of ionization efficiency in negative ionization mode by using NH 4 F enriched mobile phase to metabolomics studies. Our previously reported method using NH 4 F as mobile phase additive instead of more conventional substances also improved the ionization efficiency of the 21 selected compounds in a reproducible way using a triple quad instrument [1]. These results were recently confirmed for wide range of compounds [10]. Our current study proves that the addition of NH 4 F to the mobile phase instead of more conventional ammonium formate is also successful for the simultaneous determination of acidic contaminants in water by UHPLC-QqTOF [13,14] increasing sensitivity and quantification capabilities. The strong basicity of the fluoride anion (F À ) in the gas phase increases deprotonation of basic analytes.
The results showed good agreement between both systems for the analysed samples. For QqQ, naproxen was the pharmaceutical at highest concentration (3327 ng L À1 ) at the influent of the WWTPs which was in a lower concentration at the effluent (10 ng L À1 ). Indomethacin, clofibric acid and triclocarban were the lowest detected with 7 ng L À1 in influent samples. Regarding effluent samples, the highest detected concentration was diclofenac with 173 ng L À1 , being the gemfibrozil the compound with the lowest (5 ng L À1 ). Finally, for river waters, the concentration of target analytes was, in general, lower than WWTPs samples being the compound in major concentration the acetaminophen with 177 ng L À1 and ibuprofen with 153 ng L À1 . Concerning the concentration calculated with QqTOF, the mean concentration levels detected in influent samples ranged from 12 ng L À1 (clofibric acid) to 2963 ng L À1 (naproxen) being naproxen the most detected compound as in the case of QqQ. In the effluent the highest concentrations were methylparaben (121 ng L À1 ) followed by diclofenac (109 ng L À1 ). In river waters the concentration levels ranged from 7 ng L À1 (butylparaben) to 159 ng L À1 (ibuprofen). These results show a good correlation between both techniques as in our previous paper [3].