Development of Micellar HPLC-UV Method for Determination of Pharmaceuticals in Water Samples

Method for extraction and determination of amoxicillin, caffeine, ciprofloxacin, norfloxacin, tetracycline, diclofenac, ibuprofen, nimesulide, levonorgestrel, and 17α-ethynylestradiol exploiting micellar liquid chromatography with PDA detector and solid-phase extraction was proposed. The usage of toxic solvents was low; the chromatographic separation of the medicaments was performed using a C18 column and mobile phases A and B containing 15.0% (v/v) ethanol, 3.0% (m/v) sodium dodecyl sulfate (SDS), and 0.02 mol·L−1 phosphate at pHs 7.0 and 8.0, respectively. The method is simple, selective, and fast, and the analytes were separated in 23.0 min. For extraction, 1000 mL of sample containing 2.0% (v/v) ethanol and 0.002 mol·L−1 citric acid at pH 2.50 was loaded through a 1000 mg of C18 cartridge. The analytes were eluted using 3.0 mL of ethanol, which were evaporated and redissolved in 0.5 mL of mobile phase. Concentration factors better than 1200, except amoxicillin (224), were obtained. The analytical curves were linear (R 2 better than 0.992); LOD and LOQ (n=10) presented values in the range of 0.019–0.247 and 0.058–0.752 mg·L−1, respectively. Recoveries of 99% were obtained, and the results are in agreement with those obtained by the comparative methods.


Introduction
For many years, the analysis of environmental contaminants was done to determine the presence of pesticides, air pollutants, petroleum residues, and many other substances designated as conventional pollutants [1,2]. Recently, a large number of compounds not regulated by countries legislation have been identi ed as potential pollutants, which have been designated as emerging contaminants [3]. e pharmaceuticals and their metabolites belong to this new class of pollutants, although their presence in the environment is not new [4,5].
Many tons of these drugs are produced annually and used in human and veterinary medicine. Generally, the exact amount of pharmaceuticals produced is not published in the literature [6], but it is known that Brazil, United States, France, and Germany belong to the group of the world's largest consumers of these drugs [7]. us, the monitoring of these compounds in the environment has gained great interest as they are often found in e uents from wastewater treatment plants and natural waters [4,8,9]. In Brazil, the population self-medication is common because part of the population does not have access to adequate medical care and has easy access to medicines due to the high number of drugstores, including those with unethical business practices [8,10]. After the administration, many pharmaceuticals are transformed into one or more metabolites and excreted in the urine and feces, causing serious problems to the environment [4,8,10]. It is known that the rampant use of pharmaceuticals such as antibiotics can make some microorganisms resistant to these drugs as some bacteria have the ability to modify their genetic material [6,11,12].
Furthermore, some drugs are used in animal treatment in the rearing of livestock, pigs, and chickens, and the waste generated from these activities has become a major source of environmental contamination due to its use as fertilizer in farmland. us, the pharmaceutical compounds are not metabolized and their metabolites can pass into groundwater and eventually to watercourses such as rivers and lakes, a ecting aquatic life [6,8,12]. Another source of environmental contamination by these drugs is associated with the disposal of waste from pharmaceutical companies and hospitals in land lls that can contaminate underground waters [4,6]. e presence of hormones in the environment has been indicated as responsible for causing endocrine disturbances in human and animal organisms and endocrine disruptors [13,14]. ere is evidence that reproductive system of certain terrestrial and aquatic organisms is a ected by estrogen, resulting in the development of abnormalities and reproductive impairment in exposed organisms, even when these drugs are present at low concentrations [6]. Ca eine is a natural stimulant and the most widely consumed psychoactive drug in the world as it is present in soft drinks, co ee, tea, cocoa, and chocolate and is used concomitantly with various medications as a stimulant [15]. Due to its widespread consumption, ca eine has been used as a potential indicator of anthropogenic pollution of surface water resulted from human activity [16,17].
Due to the growing concern of the presence of antibiotics, anti-in ammatories and hormones in water intended for public supply a number of methodologies devoted to identify and quantify these compounds in various samples that have been proposed. e most common are those involving separation techniques such as gas and liquid chromatography and capillary electrophoresis coupled with several detectors such as MS, UV-Vis, FID, ECD, and others [18][19][20][21][22][23][24][25][26][27] associated with SPE with solvents such as formic acid/water [20], methanol/methyl tertiary-butyl ether [21], n-hexane/ethyl acetate/methanol [22], or more modern methods such as ultrasonic-assisted extraction/centrifugation and puri cation with SPE [18] and the pressurized liquid extraction followed by extract puri cation using SPE [19].
As it is not an easy task the development of extraction and determination methods to chemical compounds with very di erent characteristics, few methods permit the determination of the analytes present in this paper in a single run.
us, environmental researchers and laboratories dedicated to the analysis of drugs should use several methods for the determination of drugs from di erent classes in environmental samples, which increase the cost and time of the analysis. e exception is the method based on MS detector that can furnish adequate results to water analysis of emergent pollutants, but it should be considered that a lot of laboratories dedicated to routine water analysis did not have mass spectrometer, mainly due to the cost to acquire the equipment or lack of skilled professionals capable of properly using the equipment.
us, to ll this gap, in the present work is proposed a robust, simple, and green HPLC-UV method associated with SPE extraction for the determination of antibiotics (amoxicillin, cipro oxacin, nor oxacin, and tetracycline), anti-in ammatories (diclofenac, ibuprofen, and nimesulide), hormones (17α-ethynylestradiol and levonorgestrel), and ca eine in water samples, providing the laboratories dedicated to water analysis with a tool able to determine the more representative pharmaceuticals from classes of antibiotics, anti-in ammatories, and hormones used in Brazil in a single chromatographic run and using a simple HPLC-UV; a common equipment in a number of laboratories devoted to contaminant analysis, contributing to environmental researchers involving emerging pollutants. e task can be done exploiting the micellar chromatography as the surfactant can modify the C18 and the aqueous phase, increasing the possibilities of interactions of the analytes with both phases, permitting the separation of substances with di erent polarities in the same chromatographic run. e selected compounds are considered toxic, are among the most used in Brazil, present biological activity, and their presence in the aquatic environment has already been demonstrated, which justify their determination.

Experimental
2.1. Reagents. All reagents and solvents were of analytical grade (purity higher than 98%); the aqueous solutions were prepared using deionized water and were ultrasonically degassed and vacuum-ltered through a cellulose acetate membrane of 0.45 μm before chromatographic use.
Solutions containing ethanol or n-propanol or n-butanol of HPLC-grade (Tedia, Fair eld, USA), anhydrous dibasic sodium phosphate (Na 2 HPO 4 ), SDS, or CTAB (cetyl trimethyl ammonium bromide), or SDBS (sodium dodecyl benzene sulfonate), or Triton X-100 (polyoxyethylene (9-10) p-phenol tertoctyl) (both from Sigma-Aldrich, St. Louis, USA) at di erent concentrations and pH values were tested as mobile phase. e main mobile phases containing 3.0% (m/v) SDS, 20.0 mmol·L −1 Na 2 HPO 4 , and 15.0% (v/v) ethanol at pH 7.0 (phase A) and pH 8.0 (phase B) were prepared by dissolving 3.046 g of SDS, 0.284 g of Na 2 HPO 4 , and 15.0 mL of ethanol at approximately 70.0 mL of water and leaving the solution under stirring until complete homogenization. Later, the pH was adjusted by adding HCl or NaOH 1.0 mol·L −1 , and the nal volume was made up to 100.0 mL in a volumetric ask. All mobile phases used in this work were prepared following the same procedures and just modifying the percentages of reagents or changing the organic solvent to n-propanol or n-butanol or the surfactant.
Amoxicillin, cipro oxacin, nor oxacin, tetracycline, diclofenac, ibuprofen, nimesulide, 17α-ethynylestradiol, levonorgestrel, and ca eine were supplied by Sigma-Aldrich (St. Louis, USA), and 20.0 mg·L −1 stock solutions were prepared by dissolving 1.0 mg of each drug in the mobile phase A, the solution was submitted to sonication for 3.0 min to ensure complete dissolution of analytes, and the volume was completed to 50.0 mL in a volumetric ask. e solutions were stored in amber asks, protected from light, and were frozen. Just before analysis, the solutions were left to attain thermal equilibrium with the room temperature.
To verify the stability of each drug in the mobile phase and the wavelength to monitor them, 50 mL of 10.0 mg·L −1 of each drug standard solution was prepared in the mobile phase A and divided into two portions of 25.0 mL. e rst set was frozen and protected from light, and the second set was kept at 25°C and immediately used to obtain the UV-Vis spectra (200 to 600 nm). e procedure was repeated each 60 min for 3 h during the rst day and then the solution was stored in the refrigerator at 10°C. Later, the solutions (two sets) were analyzed every day during 4 days.

Sampling, Sample
Treatments, and Optimization of the SPE Procedure. A total of 7 river water samples were collected from rivers owing in the center of Maringá (Paraná, Brazil) in zones with di erent population densities and industrial activities. Samples 1-3 were collected from the south of Maringá in Moscados stream (ca. 2.9 km from its source in the Inga Park), Cleópatra stream (ca. 2.3 km from its source, located inside the Pioneiros Forest Park), and Borba Gato stream (ca. of 3.3 km from its source located in the Horto Florestal Park), respectively. Samples 4-6 were collected in the north of Maringá in Mandacarú stream (ca. 2.0 km from the source), Morangueiro stream (ca. 4.0 km from the source), and Maringá stream (ca. 2.0 km from the search), respectively; sample 7 was tap water collected in the analytical chemistry laboratory at Maringá State University.
Before sample collection, each bottle was prerinsed with the sample for three times. e samples were sent in boxes packed with ice to the laboratory at Maringá State University. Immediately upon reception, the samples were vacuum ltered through Millipore 0.45 µm membrane; then, to 1.0 L of each sample was added 2.0% (v/v) ethanol and 2.0 mmol·L −1 citric acid, and the pH was adjusted to 2.5 with HCl 1.0 mol·L −1 .
For sampling, 4 L of each water sample was collected according to the standard protocol established by the Water Resources Company Management [28], which considers sampling timing, sampling point, sampling tools and containers, sampling operation, eld records, labeling, transport, and storage of samples. Immediately upon reception, the samples were vacuum ltered through Millipore 0.45 μm membrane and to 1.0 L of each sample was added 2.0% (v/v) ethanol and 2.0 mmol·L −1 citric acid, and the pH was adjusted to 2.5 with HCl 1.0 mol·L −1 . Later, the samples were submitted to the following extraction procedure: before the extraction, the C18 stationary phase was preconditioned by passing 5 mL of methanol, 5 mL of pure water, and 5 mL of an aqueous solution containing 2.0% (v/v) ethanol and 2.0 mmol·L −1 citric acid at pH 2.5; then, 1000 mL of each sample was passed through the cartridge at a ow rate of 7.0 mL·min −1 . e solid phase was dried under vacuum for 30 s, and the analytes were eluted using 3.0 mL of ethanol at 1.0 mL·min −1 . e sample was ltered through a te on membrane, evaporated, and redissolved in 0.5 mL with the mobile phase A and injected into the chromatograph. To obtain the extraction procedure able to be applied to the ten analytes, the following variables were tested: pH (2.0-8.7), sample (1.0-10.0 mL·min −1 ) and eluent (0.5-3.0 mL·min −1 ) ow rates, sample (25.0-1000 mL) and eluent (1.0-10 mL) volumes, amount of stationary phase (500 and 1000 mg), and chemical nature of the eluent (methanol, ethanol, acetone, and acetonitrile).

Equipment and Separation Conditions.
e chromatographic separation was performed using an HPLC from ermo Electron Corporation (Waltham, USA) containing quaternary pump model Surveyor LC Plus, manual injector valve of 20 μL Rheodyne Model 8096, UV-Vis photodiode array detector model Surveyor PDA with quartz cell with optical path of 5.0 cm, ChromQuest software version 4.2 (Macherey-Nagel, Germany) for acquisition and signal recording and a column RP-18 ODS o base 250 × 4.6 mm (id) equipped with a guard column RP-18 ODS 10 × 4.0 mm (id) both with particles of 5-micron pore size of 100Å and carbon content of 15.5%.
Before the rst and after the last injection of the day, the column was cleaned with ultrapure water for 30.0 min at a ow rate of 0.5 mL·min −1 . e initial conditioning of the stationary phase was performed by passing mobile phase A through the column for 20.0 min at a ow rate of 1.0 mL·min −1 . After standard/sample injection (20.0 μL), the separation process was carried out as follows: 0.0-2.0 min with 100.0% of phase A with the instantaneous change to the 100.0% of phase B and keeping it until 25 min always at a ow rate of 1.0 mL·min −1 . e temperature was xed at 25°C, and the analytes were monitored at 220 (amoxicillin, nor oxacin, tetracycline, diclofenac, ibuprofen, nimesulide, 17α-ethynylestradiol and ca eine), 240 (levonorgestrel), and 280 nm (cipro oxacin), simultaneously. After each analysis, the column was reconditioned for 10.0 min using phase A at a ow rate of 1.0 mL·min −1 .

Optimization of the Chromatographic Method.
Preliminary experiment was done to identify the main variables a ecting the chromatographic separation such as pH (5.0, 6.0, 7.0, and 8.0), percentage of the organic modi er in the mobile phase (15.0, 20.0, and 25.0% (v/v)), nature of the organic modi er (n-butanol, n-propanol, and ethanol), SDS concentration (2.5, 3.0, and 3.5% (m/v)), and ow rate (0.8, 1.0, 1.2, and 1.5 mL·min −1 ). erefore, a factorial design experiment 2 3 was carried out in duplicate in order to optimize the separation conditions and the contribution of the variables (Na 2 PO 4 , ethanol, and SDS concentrations) in the chromatographic separation (Table 1).

Method Calibration, Characterization, and Sample
Analysis. For quantitative determination of pharmaceuticals and ca eine in the water samples, calibration curves were plotted using the peak area (y) versus concentration of the analytes in the following ranges: amoxicillin, ciprooxacin, diclofenac, ibuprofen, tetracycline, levonorgestrel, nimesulide, and nor oxacin from 0.1 up to 25.0 mg·L −1 ; 17α-ethynylestradiol from 0.5 up to 25.0 mg·L −1 ; and caffeine from 0.08 up to 25.0 mg·L −1 . e calibration curves were used to determine the analyte concentrations in the samples and in the blank with and without spiking. e LOD and LOQ (n � 10) were estimated using the signal-to-noise ratio of 3.3 and 10.0, respectively [9], and 7 river water samples were analyzed by the proposed method and the obtained results were compared to those obtained by the six comparative HPLC-DAD methods involving SPE [18,24,29,30], stir-bar sorptive and liquid desorption [25], and liquid-liquid extraction [31] methods to the following analytes: tetracycline [18]; ibuprofen, diclofenac, and nimesulide [24]; levonorgestrel and 17α-ethynylestradiol [25]; ca eine [29]; cipro oxacin and nor oxacin [30]; and amoxicillin [31].
Sample recovery tests were done spiking the samples with the following: 5.0 µg·L −1 of ibuprofen, 17α-ethynylestradiol, diclofenac, nimesulide, levonorgestrel, cipro oxacin, and noroxacin; 10 µg·L −1 of tetracycline and ca eine; and 60 µg·L −1 of amoxicillin. After extraction procedure, the extracts were evaporated completely and stored in the refrigerator; just before analysis, the samples were redissolved in 0.5 mL of the mobile phase A. e spiked samples were analyzed (n � 5) by the proposed and by comparative methods during ve consecutive days (n � 5).

Spectra and Stability of the Analytes.
e UV-Vis spectra of analytes showed electromagnetic radiation absorption in the region between 200 and 600 nm with higher intensity in the UV region. It is observed that signal overlaps, which makes it di cult for the simultaneous determination of the analytes without prior separation. It was decided to monitor the analytical signals at 220, 240, and 280 nm, because the molar absorptivity is high to the majority of the analytes and the mobile phase absorption is low; the exceptions are levonorgestrel (λ max � 240 nm) and cipro oxacin (λ max � 280 nm) that presented low molar absorptivity at 220 nm. For this reason, the signals can be monitored at 220 nm by laboratories that do not have chromatograph with PDA detector, but the sensitivity to levonorgestrel and cipro oxacin will be poor. e UV-Vis spectra also indicated that the analytes were stable in the mobile phase A for a long period, except tetracycline that presented degradation of 4.1, 8.5, 13.4, and 16.0% after 24, 48, 72, and 96 h, respectively. When the samples were kept frozen and in the absence of light before analysis, the tetracycline degradation decreased to lower than 2.0%, indicating that the standards and samples should be kept under these conditions until analysis.

Extraction of the Analytes.
Initially, it was decided to use the C18 phase, control the pH, and use methanol as eluent, as 10 mL of this solvent permitted to get the elution of ten analytes from the cartridge. e sample pH was varied from 2.0 up to 8.5, and pH 2.5 was chosen as the better condition due to the optimum extraction percentage for most of the analytes. Amoxicillin, high polar molecule, presented low a nity by the solid phase, and its recuperation was always lower than 51%, whereas ibuprofen, diclofenac, and nimesulide presented high interaction with the solid phase and could only be eluted around pH 2.0 with recoveries of 83.6, 59.0, and 75.5%, respectively. Cipro oxacin, nor oxacin, and tetracycline have two positive charges at pH 2.5, and the increase in the pH reduced their recuperation from 92.0, 89.0, and 100.0% for pH 2.5 to 36.3, 19.4, and 42.2% for pH 8.5, respectively; and ca eine, 17α-ethynylestradiol, and levonorgestrel did not su er a signi cant change in their interaction with the solid phase with the pH and presented recuperation percentages of 94.7, 92.0, and 80.0%, respectively.
A ow rate of 7.0 mL·min −1 was chosen as the better compromise between time and e ciency of extraction; for this condition, recoveries around 95% were obtained for the analytes; the exception was amoxicillin (50%). Variations in the citric acid concentration (2.0, 3.0, and 5.0 mmol·L −1 ) did not change the extraction e ciencies; thus, 2.0 mmol·L −1 of citric acid was maintained. e loading sample volume should be as high as possible to get high concentration factors. When solutions of the analytes in the concentrations of 1.0, 0.5, 0.25, 0.05, and 0.025 mg·L −1 were associated with extraction sample volumes of 25, 50, 100, 500, and 1000 mL (to keeping the nal concentration after elution in 2.5 mg·L −1 ), it was noted an improvement in the extraction with the increase of sample volume; however, for the more polar analytes (amoxicillin, tetracycline, ca eine, and diclofenac), a reduction in the recovery percentages was observed. For sample volume of 500 mL, the best recoveries for most analytes were obtained with values higher than 85%, except for amoxicillin (15.3%); however, in order to obtain high concentration factor to levonorgestrel and 17α-ethynylestradiol hormones, that could be present at very low concentrations in the samples, the condition of 1000 mL was selected, even with a reduction in the extraction e ciency to amoxicillin, tetracycline, caffeine, and diclofenac. e amount of solid phase in the cartridge was also studied, and it was noted that reducing the amount of solid phase led to lower extraction e ciency to tetracycline and amoxicillin.
us, the cartridge with 1000 mg of C18 was selected.
In order to achieve the highest possible concentration factor with minimum use of organic solvents (methanol, ethanol, acetone, and acetonitrile), the volume of the solvent was varied (1.5, 3.0, 5.0, 7.0, and 10.0 mL) and it was not observed signi cant variations in the analytes recoveries when the eluent was methanol or ethanol and the eluent volume was varied from 3.0 to 10.0 mL; thus, 3.0 mL of ethanol was elected as the better condition due to its low toxicity. To acetone and acetonitrile, the recoveries were low to all analytes, mainly to tetracycline, nor oxacin, and ethynylestradiol. e eluent ow rate was varied, and it was observed a decrease in the extraction e ciencies for all the Journal of Analytical Methods in Chemistry analytes to ow rates higher than 1.0 mL·min −1 ; then, the eluent ow rate was xed at 1.0 mL·min −1 . us, the nal extraction conditions for simultaneous extraction and concentration of the ten analytes were as follows: 1000 mL of the sample containing 2.0% (v/v) ethanol, 2.0 mmol·L −1 citric acid at pH 2.50, C18 cartridge with 1000 mg of the solid phase, and ow rate of 7.0 mL·min −1 . After analytes' retention, the adsorbent was dried under vacuum 30 s and the analytes were eluted with only 3.0 mL of ethanol at a ow rate of 1.0 mL·min −1 . e solvent was evaporated, and the samples were redissolved in 0.5 mL of a solution with 15.0% (v/v) ethanol, 3.0% (m/v) SDS, and 20.0 mmol·L −1 phosphate bu er at pH 7.0 (mobile phase A). Under these conditions, it was yielded recovery percentage values of higher than 95%, except to tetracycline (64.3%) and ca eine (66.0%), and concentration factors higher than 1200, except to amoxicillin (224).

Preliminary Tests.
e developed chromatographic method should be as green as possible, and taking into account the di erent polarities of the analytes, it was decided to exploit the micellar chromatography [32,33] due to the possibility to get the solubilization of organic compounds of low polarity in the aqueous medium and, at the same time, change the polarity of the C18 stationary phase with the surfactants. For the task, SDS, CTAB, SDBS, and Triton X-100 surfactants were tested and the anionic surfactant SDS was chosen because of its low critical micellar concentration (cmc of 0.0082 mol·L −1 in water), which permits to use lower surfactant concentration. Furthermore, SDS presents low light absorption and scattering e ect in the UV region, where the analytes should be monitored, allowing to get lower baseline signals and better sensitivity.
Low toxic organic modi er solvent to assist the solubilization of the ten analytes in the micellar medium was chosen after solubility tests with several proportions of ethanol or n-propanol or n-butanol with water and SDS. It was observed that solutions containing 15% of ethanol or butanol or n-propanol together with at least 3.0% SDS and 20 mmol·L −1 of phosphate bu er at pH 7.0 were able to solubilize all the analytes as well as elute them from the C18 column ( Figure 1). e di erences in the chromatograms for the di erent organic modi ers can be attributed to the di erent polarities of each organic solvent aliphatic chain, which explains the highest retention time obtained when ethanol was used as an organic modi er (Figure 1). Due to this e ect, ethanol was considered the better organic modi er because it improved the symmetry and chromatographic resolution of some peaks (except nimesulide, cipro oxacin, and nor oxacin) and is low toxic than n-butanol and n-propanol. In addition, the use of ethanol permitted the identi cation of the tetracycline peak, which could not be done when it was used n-propanol or n-butanol.

pH E ect in the Chromatographic
Separation. pH is an important variable to be studied when the analytes have ionizable groups, as variations in pH can promote changes in the solubility and in the ionic interactions of analytes with the micellar medium and with the stationary phase. When the pH was varied from 6.0 up to 8.0 (Figure 2), ca eine (peak 2, pKa 2.19), 17α-ethynylestradiol (peak 6, pKa 9.44), and levonorgestrel (peak 7, pKa 1.05) did not show signi cant variation in their retention time as these compounds are neutral in that pH range; the di erent retention time periods were attributed to their di erent hydrophobicities. On the other hand, the density of negative charges increased with pH to diclofenac (peak 4, pKa 5.35), ibuprofen (peak 5, pKa 5.82), and nimesulide (peak 8, pKa 7.15), causing large reduction in their retention time due to the high a nity of the charged analytes by the micellar aqueous mobile phase and their repulsion by the negatively charged groups of the surfactants adsorbed on the stationary phase.
Tetracycline (pKa's 3.94, 7.62, and 9.19), cipro oxacin (pKa's 3.32, 7.12, and 8.42), and nor oxacin (pKa's 3.38, 7.16, e e ect also caused peak broadening, and at pH 5.0, it was not possible to identify the tetracycline peak due to excessive molecule retention. Amoxicillin (pKa's 3.01, 7.32, and 9.70, pI 5.20) is also an amphoteric molecule and with the pH reduction from 7.0 to 6.0 (Figures 2(a) and 2(b)), its retention time increased because the molecule almost reached the neutrality and became more hydrophobic.
In none of the tested pH conditions was possible to obtain the chromatographic separation of the ten analytes with resolutions adequate to their analytical determination. Due to the di erent analyte characteristics, they could be divided into two groups: those with symmetric and thin peaks and presenting low retention time and those presenting broad peaks, high retention time, and low e ciency and chromatographic resolution. en, it was decided to do a mobile phase pH gradient starting with 100% of phase A changing gradually (7.0, 5.0, and 2.0 min) to 100% of phase B (same composition of phase A at pH 8.0). e phase with pH 6.0 was not chosen because some analytes were highly retained in the stationary phase under this condition (Figure 2).
us, it was concluded that the better pH gradient would be 100% of phase A until 2.0 min with a sudden change to 100% of phase B 100%. Under this condition, it was possible to get the separation of the ten compounds with adequate chromatographic resolution in only 23.0 min (Figure 3(d)).

Variables' Interactions.
e preliminary experiments demonstrated that the use of ethanol as an organic modi er and gradient of phases A and B were adequate to achieve the analyte separation; thus, a factorial design 2 3 (Table 1) with SDS, ethanol, and phosphate concentrations as variables in both phases was carried out to verify the main and the interaction e ects of variables and their importance in the analytes separation.
Reducing ethanol concentration in the mobile phase from 15.0% to 12.0% (v/v) promotes an increase in the analyte retention time and in time of analysis (experiments 4 and 3, Figure 4). e e ect can be explained by the reduction of nonpolar characteristic of the mobile phase, which is di cult for the elution of the low polar analytes from the C18 phase, which led to an overlap between tetracycline (peak 3) and diclofenac (peak 4), nimesulide (peak 8) and ciprooxacin (peak 9), and nor oxacin (peak 10) and levonorgestrel (peak 7), decreasing their chromatographic resolutions ( Table 2). e reduction in SDS concentration from 3.0% to 2.5% (m/v), experiments 4 and 2 (Figure 4), also increased the retention time for peaks 3, 4, 6, 9, 10, and 7. e reduction in the number of micelles in the mobile phase increased the hydrophobic interaction of 17α-ethynylestradiol (peak 6) and levonorgestrel (peak 7) by C18 phase, causing the coelution of 17α-ethynylestradiol (peak 6) with nimesulide (peak 8). e compounds 3, 4, 9, 10, and 7 increased their a nities by the stationary phase due to electrostatic interactions between analytes, partially positive charged, and the negative ionized sulfonic acid groups adsorbed on C18 phase. is factor improved the separation between nimesulide (peak 8) and cipro oxacin (peak 9) and ibuprofen (peak 9) and nor oxacin (peak 10) ( Table 2).
Analyzing the results furnished by experiments 4 and 8, it was noted that the increase in the phosphate concentration (from 20 to 30 mmol·L −1 ), and therefore Na + concentration, reduced the a nity of ibuprofen (peak 5), nimesulide (peak 8), nor oxacin (peak 10), and cipro oxacin (peak 9) by the stationary phase due to two reasons. First, because ion-exchange competition between Na + and the positively charged analytes by the ionized sulfonic acid groups absorbed on C18 phase and, second, due to the partial stabilization of the analytes charged by the phosphate ions. us, the chromatographic resolutions between diclofenac (peak 4) and ibuprofen (peak 5) and 17α-ethynylestradiol (peak 6) and nimesulide (peak 8) decreased from 1.68 and 2.10 to 0.51 and 1.42, respectively (Table 2). e variations in ethanol and SDS concentration practically did not a ect the amoxicillin peak, whereas ca eine experienced a slight increase in its retention time when ethanol concentration was decreased. e association of lower percentages of SDS and ethanol led to higher time of analysis (30 min), and despite an improvement in the resolution between peaks 6 and 10, the resolution between diclofenac (peak 4) and ibuprofen (peak 5) decreased ( Figure 4, experiment 1, Table 2). Furthermore, when these factors were associated with higher phosphate concentrations (Figure 4, experiment 5), it was observed the coelution of diclofenac (peak 4) and ibuprofen (peak 5) and 17α-ethynylestradiol (peak 6) and nimesulide (peak 8).
From the main e ects (Table 3), it was possible to note that ethanol factor improved the chromatographic resolution between the analytes, especially between peaks 4 and 5, the exception was between peaks 6 and 8. e SDS factor improved the resolution between peaks 6 and 8 and worsened it to peaks 8 and 9. e phosphate factor reduced the resolution between peaks 6 and 8 and increased it to peaks 3 and 4. SDS promoted the higher increase in the time of analysis followed by ethanol, whereas the phosphate concentration presented a contrary e ect and induced the reduction in the time of analysis. e second-order (AB, AC, and BC) and third-order (ABC) interaction e ects indicated intense variable interactions (Table 3), showing that the variables could not be studied in an independent form. e ethanol-SDS interaction was signi cant to increase the resolution between peaks 3 and 4 that was not critical but decreased the chromatographic resolution between peaks 4, 5, 6, and 8 and increased the time of analysis. e ethanol-phosphate interaction was important to increase the resolution between peaks 6 and 8 and contributed to decreasing the resolution between peaks 3 and 4 and 8 and 9. e SDS-phosphate interaction increased the resolution between peaks 8 and 9 and decreased the resolution between peaks 4 and 5 and 6 and 8. e ethanol-SDS-phosphate interaction increased the time of analysis as well as the resolution between peaks 8 and 9 but decreased the resolution between peaks 4 and 5 and 6 and 8 (Table 3).
Considering all variable e ects, it was concluded that experiment 4 presented the best characteristics; under this condition, it was possible to separate the ten analytes in only 23.0 min always with chromatographic resolution better than 1.45. e e ect of the ow rate in the analyte separation showed that, for ow rate values higher than 1.0 mL·min −1 , the chromatographic resolutions worsened, and the separation between tetracycline and diclofenac, diclofenac and ibuprofen, and cipro oxacin and nor oxacin was poor, then 1.0 mL·min −1 was selected. us, the nal separation conditions were as follows: mobile phase with 15.0% (v/v) ethanol, 3.0% (m/v) SDS, 20.0 mmol·L −1 phosphate bu er at pH 7.0 to phase A and 8.0 to phase B; maintaining 100% of phase A until 2.0 min with abrupt change to 100% of phase B; ow rate of 1.0 mL·min −1 ; 25°C; injected volume of 20 μL; monitoring the signals at 220, 240, and 280 nm; and time of analysis of 23.0 min (Figure 4, experiment 4).

Calibration, Characterization, and Sample Analysis.
e analytical curves for the ten analytes in the concentration range of 0.10 up to 25.0 mg·L −1 for amoxicillin, diclofenac, ibuprofen, levonorgestrel, nimesulide, tetracycline, cipro oxacin, and nor oxacin; 0.5 up to 25.0 mg·L −1 for When the proposed method was applied to the analysis of the 7 water samples, the standard deviation was ca. 2% and the peaks showed to be free of interferences. e results furnished by the proposed method are in agreement with those obtained by the comparative chromatographic methods (Table 4). Furthermore, the obtained results to sample recovery tests were consistent with high recovery values and low standard deviations (Table 5).
Considering all the obtained results, the proposed method presented analytes average recovery of 99.12% and intraday (n � 3) and interday (n � 5) precision of 1.11% and 2.30%, respectively, whereas the comparative methods presented analytes average recovery of 98.64% and intraday (n � 3) and interday (n � 5) precision of 1.34% and 2.97%, respectively, indicating that the proposed method furnished results similar to those obtained by the comparative methods, but with the di erence that it was necessary to carry out more than one method for the determination of the ten analytes. e analysis of river water samples indicated the presence of ca eine in all the samples with concentrations ranging from 0.071 mg·L −1 to 1.204 mg·L −1 , probably due to (3) (1)  Journal of Analytical Methods in Chemistry the human activities. It was veri ed the presence of antibiotics in three streams: tetracycline and nor oxacin in Cleópatra stream, nor oxacin in the Mandacarú stream, and cipro oxacin in the Morangueiro stream (Table 4). e proposed chromatographic and extraction method presented high accuracy and precision for the analysis of water samples, allowing the quanti cation of the ten analytes with only one extraction and chromatographic method.

Conflicts of Interest
e authors declare that they have no con icts of interest. Data were obtained for water samples spiked with 5.0 µg·L −1 of ibuprofen, 17α-ethynylestradiol, diclofenac, nimesulide, levonorgestrel, cipro oxacin, and nor oxacin; 10 µg·L −1 of ca eine; and 60 µg·L −1 of amoxicillin. e conventional methods were the same as used in Table 4.