*2.3. Analytical Methodology*

To extract and preconcentrate the target pharmaceuticals in wastewater samples, a methodology based on SPE previously optimized [19] was used. Briefly, 250 mL of filtered wastewater at pH 7 were extracted using 500 mg Oasis HLB cartridges (Waters, Barcelona, Spain) in a Varian SPE manifold (Madrid, Spain). After the loading of the samples, the cartridges were washed with 5 mL of Milli-Q water and after that, the retained compounds were eluted with 5 mL of methanol. To achieve a grea<sup>t</sup> preconcentration factor, the solvent was evaporated under a gentle nitrogen steam and reconstituted with 1 mL of Milli-Q water with 100 <sup>μ</sup>g·L−<sup>1</sup> of internal standards. Before analysis the extracts were filtered using Chromafil Xtra PET-20/25 syringe filters with a pore size of 0.20 μm from Machery-Nagel (Düren, Germany).

**Figure 1.** Layout of wastewater treatment plants (WWTPs) surveyed. (**a**) Conventional Wastewater Treatment Plant 1 (C-WWTP1); (**b**) Conventional Wastewater Treatment Plant 2 (C-WWTP2); (**c**) Natural Wastewater Treatment Plant (N-WWTP). Figure adapted from [19].

After the extraction, ultra-high performance liquid chromatography tandem mass spectrometry (UHPLC–MS/MS) was used as a separation and detection technique. The separation of pharmaceuticals was performed using an ACQUITY UPLC BEH Waters C18 column (50 mm × 2.1 mm, 1.7 μm) from Waters Chromatography (Barcelona, Spain). The mobile phase used consisted of LC–MS grade water and methanol both with 0.5% acetic acid and the separation of target compounds was done in gradient mode (Table S1). The chromatographic system used was an AQCUITY UPLC system and consisted of a binary solvent manager pump, an autosampler, a column manager, and a triple quadrupole detector (TQD) controlled by Masslynx software, all of them from Waters Chromatography (Barcelona, Spain). The detection of the target compounds was carried out using electrospray ionization (ESI) in both positive and negative mode and the mass spectrometer parameters had a capillary voltage of 3.00 kV in positive mode and –2.50 kV in negative mode, source temperature of 120 ◦C and desolvation temperature of 450 ◦C. The conditions of fragmentation and the ions of the pharmaceuticals under study are presented in Table S2.

This analytical methodology presented grea<sup>t</sup> recoveries, between 52.9% and 123.6%, and appropriate detection limits, between 15.3 ng·L−<sup>1</sup> and 13.3 <sup>μ</sup>g·L−1. The method also showed grea<sup>t</sup> linearity, with correlation coefficients (r2) over 0.99 in all cases and good intra-day and inter-day repeatability, with relative standard deviation (RSD) values below 22% [19].

#### **3. Results and Discussion**

In this study, solid phase extraction (SPE) coupled to ultra-high-performance liquid chromatography tandem mass spectrometry (UHPLC-MS/MS) method was applied for analysis of water samples from three WWTPs. After the analysis, a statical study was performed in order to establish the median concentrations of the detected pharmaceuticals. The median concentration was used instead of the average concentration due to the dispersion of the concentrations detected during the two years of study. Moreover, the frequency of detection and the removal efficiency of each purification process were also calculated by dividing the number of samples with detected concentrations of a pharmaceutical by the total number of analyzed samples (positive results in Table 2).

#### *3.1. Occurrence and Concentrations of Target Pharmaceuticals*

#### 3.1.1. Conventional Treatment WWTPs

Conventional wastewater purification treatments are based on a primary treatment which is focused on the elimination of fats, sand, and coarse solids. After that, a secondary treatment, usually based on biological processes, is used to degrade organic matter from wastewater as well as to eliminate suspended solids and other organic pollutants. These processes could be aerobic or non-aerobic, and the most used biological process is activated sludge. Finally, if the wastewater will be re-used, tertiary processes must be performed in order to obtain a water with grea<sup>t</sup> quality. There are many tertiary treatment technologies and most used are microfiltration, nanofiltration or reverse osmosis [20–23].

C-WWTP1 is based in activated sludge technology and as can be seen in Table 2, the stimulants under study (nicotine, caffeine, and paraxanthine) present the highest concentrations at influent samples, probably because these compounds are used in pharmaceutical formulations but also are excreted by smokers in the case of nicotine and after drinking beverages like coffee in the case of caffeine and its metabolite, paraxanthine. In fact, the median concentrations of these compounds reached values between 45.8 and 95.6 <sup>μ</sup>g·L−<sup>1</sup> during monitoring. This behavior was observed in other studies in which the highest concentrations of pharmaceuticals studied were obtained for these three stimulants as well [24–27]. Moreover, two anti-inflammatory drugs, naproxen and ibuprofen, also presented median concentrations in the range of <sup>μ</sup>g·L−<sup>1</sup> (between 19.9 and 27.3 <sup>μ</sup>g·L−1) in influent samples and naproxen showed the highest concentration of all the monitored compounds (521.7 <sup>μ</sup>g·L−<sup>1</sup> in an influent sample). Anti-inflammatory compounds show a high rate of consumption in Spain [28] and the high concentrations of ibuprofen can be explained because this compound does not require a medical prescription as some authors have stated [29]. In addition, two pharmaceuticals related to cardiovascular problems and diseases namely, atenolol and gemfibrozil, presented low influent median concentrations of <sup>μ</sup>g·L−<sup>1</sup> (1.13 and 3.35 <sup>μ</sup>g·L−1, respectively). The median concentrations of the rest of the target compounds were in the range of ng·L−1. As the purification process was performed, the median concentration of the compounds decreased. In fact, the sum of the median concentrations of the target compounds ranged from 150.9 <sup>μ</sup>g·L−<sup>1</sup> in the influent, to 0.90 <sup>μ</sup>g·L−<sup>1</sup> after the osmosis treatment, which is indicative of grea<sup>t</sup> elimination in the wastewater purification system. Figure 2a shows the changes in the distribution of pharmaceutical median concentrations in the different sampling points. In this sense, it can be observed that the contributed total concentration of stimulants decreased during the purification process, while the contribution of some compounds, like carbamazepine, increased; this is because the concentrations during the purification process remain stable. In fact, carbamazepine's median concentration was similar in points A1 and A2 (0.146 and 0.052 <sup>μ</sup>g·L−1, respectively), but the contribution to the total concentration changed from 0.1% to 5.7% after secondary and tertiary treatment. The composition of the final effluent obtained in the WWTP is a key factor, because it will be useful to predict the possible effects of effluents used in agriculture or discharged into the environment.


*Molecules* **2020**, *25*, 1639



## *Molecules* **2020**, *25*,1639

**Table 2.** *Cont.*

(**a**) 

(**b**) 

**Figure 2.** *Cont.*

**Figure 2.** Distribution of pharmaceuticals from Table 1. (**a**) C-WWTP1, (**b**) C-WWTP2, (**c**) N-WWTP.

C-WWTP2 has an activated sludge treatment too, but in this WWTP the study was conducted in the tertiary process which is based on electrodialysis reversal. This technology has been demonstrated as an effective way to remove some emerging pollutants from drinking water [30] but studies in wastewaters are scarce and for this reason, this tertiary process was evaluated. In this wastewater treatment plant, the samples were taken after the secondary process (point B1) and after the electrodialysis technology (point B2). Before this treatment, an ultrafiltration process was also performed as pretreatment. Ten of the 11 pharmaceuticals under study were detected at concentrations that ranged from 0.055 to 0.723 <sup>μ</sup>g·L−<sup>1</sup> in point B1 and from 0.043 to 0.701 <sup>μ</sup>g·L−<sup>1</sup> in point B2. The highest concentrations were of gemfibrozil, caffeine, and paraxanthine, which ranged from 0.241 to 0.723 <sup>μ</sup>g·L−1; this coincides with the results of C-WWTP1. This same behavior can be explained because both WWTPs treat the water of big urban areas with similar characteristics. In the same way, the lowest concentrations after secondary treatments (points A2 and B1) coincide in the two conventional WWTPs and correspond to erythromycin, nicotine, and trimethoprim. Figure 2b shows a similar distribution of pharmaceuticals before and after the purification process which means that this technology had a similar impact in the removal of the target compounds.

#### 3.1.2. Natural Treatment WWTP

An alternative to conventional WWTPs is natural WWTPs. From the different types of natural treatment technologies, constructed wetlands have revealed themselves as a grea<sup>t</sup> alternative to treat municipal wastewaters from small communities or isolated areas in both vertical flow and horizontal flow layouts [31,32]. In this system, the purification of wastewater is partially done in the vertical flow wetland and the horizontal flow wetland improves the water quality. The highest concentrations of target pharmaceuticals match with C-WWTP1, and were from caffeine, paraxanthine, nicotine, ibuprofen, and naproxen. All of these compounds presented median concentrations that ranged from 9.60 to 40.05 <sup>μ</sup>g·L−1, while the rest of the target compounds were in the range of ng·L−1. The highest concentrations detected in the influent samples of this WWTP were from naproxen (320.07 <sup>μ</sup>g·L−1) and caffeine (126.40 <sup>μ</sup>g·L−1). As in the previous WWTPs, in this WWTP the concentrations decreased as the purification process was conducted. This can be observed in the total median concentrations of the WWTP that were 117.02 <sup>μ</sup>g·L−<sup>1</sup> in the influent (point C1), 104.14 <sup>μ</sup>g·L−<sup>1</sup> after Imhoff treatment (point C2), and 51.61 and 19.06 <sup>μ</sup>g·L−<sup>1</sup> after vertical flow (point C3) and horizontal flow (point C4) wetlands, respectively. Figure 2c shows the distribution of pharmaceuticals in the sampling points and

it can be observed that the distribution in the influent of this WWTP is similar to C-WWTP1. In the influents of both WWTPs, the majority of compounds found were stimulants and anti-inflammatories. Nevertheless, the pharmaceutical profiles were different in the rest of the sampling points due to the removal efficiency of each WWTP. In the final effluent of the N-WWTP we observed a large contribution of ibuprofen to the total amount of pharmaceuticals; this was not observed for the other WWTPs.

#### *3.2. Removal of Target Pharmaceuticals*

As previously stated, pharmaceutical compounds have become a concerning group of emerging pollutants to the scientific community and their removal from wastewaters is essential to ensure the environmental quality of recipient ecosystems. Biologically-based WWTPs produce effluents that maintain water quality standards in order to reuse them or dispose into the environment at a reasonable cost, but these WWTPs have limited capability to remove pharmaceuticals [33]. For this reason, it is essential to evaluate the removal efficiency of different technologies, even more so if the purified wastewaters will be re-used in agriculture. To calculate the removal of the different treatments, the following equation was used in the different samplings and WWTPs surveyed.

$$RE\ (\%) = 100 - \left(\frac{[Effl]}{[Inf]} \ast 100\right)$$

where *RE* is removal efficiency, [*E*ffl] is the measured concentration of the pharmaceutical in the effluent of the treatment, and [*Inf*] is the measured concentration of the pharmaceutical in the influent of the treatment.

#### 3.2.1. Conventional Treatment WWTPs

In C-WWTP1, the secondary process provides median removals over 98% for the compounds with the highest concentrations (nicotine, caffeine, paraxanthine, ibuprofen, and naproxen). For the pharmaceuticals used in cardiovascular diseases, namely atenolol and gemfibrozil, the removal efficiency of the biological treatment was also great, with median removals over 72.5%. Diclofenac and trimethoprim showed slightly lower removal efficiencies of 42.9% and 66.5%, respectively. Only carbamazepine showed a negative value of removal. This means that the concentrations after the biological treatment were higher than in the influent. The observed persistence of carbamazepine in treated wastewaters has been the topic of many studies around the world, and for this reason, some authors like Hai et al. have proposed its use as an anthropogenic marker in water [34]. Other authors have attributed the poor removal of carbamazepine to its molecular structure and hydrophilicity [35–37]. In this WWTP, tertiary processes were also evaluated. The microfiltration process was not effective; in fact, the median removals were between –52.9% and 19.1%. Nevertheless, reverse osmosis was an effective method to remove these compounds. The removals of pharmaceuticals in this treatment were high, in all cases over 55%, even for persistent compounds such as carbamazepine. By comparing the concentrations of pharmaceuticals in influent and final effluent, the combination of biological treatment and tertiary processes based on reverse osmosis was effective in the removal of pharmaceutical residues. The median recoveries of the overall purification process were over 90% for all compounds under study, except carbamazepine, which showed an overall removal of 73.7%. This removal could be considered as very satisfactory in comparison with other studies in which the removal of this compound only reached 10–30% [38].

In C-WWTP2, another tertiary process was evaluated; in this case, the combination of an ultrafiltration and electrodialysis reversal process. There is very little literature about the efficiency of this treatment process in wastewaters. In our study, the monitoring was conducted for one year and the results showed that this technology has a moderate efficiency. Only four compounds (atenolol, naproxen, ibuprofen, and diclofenac) showed a median removal between 48% and 58%. The electrodialysis technology was not efficient with the rest of the compounds under study. The

target stimulants, nicotine, ca ffeine, and paraxanthine, showed removal e fficiencies between 13.8% and 25.5% while the e fficiencies for trimethoprim, carbamazepine, and gemfibrozil were 20.4%, 12.3%, and 5.4%, respectively. These results complement the study of Arola et al. which established that using electrodialysis, diclofenac and ibuprofen were preferentially retained in the diluent [22]. However, the study of Arola et al. was done in a pilot plant and the authors established that the study must be confirmed using real wastewaters, like in this study.

#### 3.2.2. Natural Treatment WWTP

In this WWTP, the Imho ff process showed limited removal e fficiency for pharmaceutical residues. In this sense, only two compounds showed 100% elimination during this process, but these two compounds, erythromycin and diclofenac, were detected in less than 20% of the surveyed samples; thus, it is not possible to ensure that this type of purification system is appropriate for the elimination of these two pharmaceuticals. Regarding the anti-inflammatories (naproxen and ibuprofen) and gemfibrozil, they showed negative elimination ratios, which means that the concentrations after the Imho ff process were higher than before. These negative elimination ratios could be explained by daily fluctuations in the concentrations of these compounds because the samples were taken at the same time in each sampling point. Furthermore, some deconjugation processes, in which conjugated compounds are converted into free compounds during the purification process, could explain the negative removal ratios obtained [7,39]. For the rest of the compounds, the eliminations were not high, and the median removals were between 2.9% and 41.8%. Regarding the constructed wetland processes, in most cases both configurations (vertical and horizontal flows) showed similar behaviors. In the vertical flow system, removals over 60% were achieved for the three stimulants, nicotine, ca ffeine and paraxanthine. Atenolol showed a medium removal of 51.4% and poor elimination e fficiencies (between 10.6% and 33.1%) were obtained for trimethoprim, and the three anti-inflammatory compounds. Carbamazepine showed a trend similar to conventional secondary processes and the elimination was negative, which means that the concentration after the purification process was higher, as observed in other studies. In this vertical flow process, the concentrations of gemfibrozil were also higher after treatment which was also observed in a previous study on this WWTP [19]. Finally, the horizontal flow treatment provided slightly higher removal e fficiencies than vertical flow treatment. In this sense, five compounds presented median eliminations over 75% (nicotine, ca ffeine, paraxanthine, atenolol, and naproxen). For trimethoprim and ibuprofen, the removals were low (30.0% and 26.4%, respectively), but higher than those obtained in the vertical flow systems. For diclofenac, a di fferent trend was observed, because the median removal for this compound was –162.5%, which indicates an increase in the concentrations after this treatment which could be explained by the daily fluctuations or by deconjugation processes as in previous treatments. Finally, carbamazepine and gemfibrozil showed negative removals too, but the increase of the concentrations after the treatment was lower than in the vertical flow treatment. Overall, the natural treatment processes performed in this WWTP showed grea<sup>t</sup> eliminations for stimulants (over 97.5%), atenolol (90.9%), naproxen (79.4%), and trimethoprim (64.0%). Three compounds (carbamazepine, diclofenac, and gemfibrozil) showed negative removals after the whole purification system and this trend coincided with a previous study performed at this WWTP [19].

#### 3.2.3. Comparison between Conventional and Natural Purification Treatments

To perform comparisons of the treatment technologies, between the two systems of conventional and natural purification, it is necessary to compare the same stages of purification. As can be seen in Figure 3a, both natural and conventional systems provide similar removal e fficiencies for target compounds after secondary treatment. The highest elimination rates in both WWTPs were obtained for stimulants, with median removals over 97% and atenolol, which showed a median removal over 90% in both systems. Regarding anti-inflammatory compounds, the trends in the two surveyed WWTPs were di fferent. Conventional treatments showed large removals for naproxen and ibuprofen (between 98.7% and 99.5%), while in the natural WWTP, the removals for these compounds were between 36.1%

and 79.4%. For the third anti-inflammatory compound, diclofenac, C-WWTP1 showed a positive elimination ratio (42.9%) while in the N-WWTP, this efficiency was negative, which implies greater concentration of the free compound in the effluent of the system. In this sense, effluent concentrations of diclofenac were 10 times higher than influent concentrations, but in all cases below 0.5 <sup>μ</sup>g·L−1. This trend was the same for gemfibrozil which showed a median elimination of 72.5% in C-WWTP1 and more than –300% in the N-WWTP. Finally, for trimethoprim, very similar median removals were achieved (in both wastewater treatment facilities it was over 60%) and this trend was also seen for carbamazepine which reflected the recalcitrant behavior of this compound with median removals near –50% in both WWTPs.

\* not determined 

**Figure 3.** Median removal efficiencies for target compounds using (**a**) secondary treatments and (**b**) tertiary treatments.

Regarding tertiary processes, it can be seen in Figure 3b that the reverse osmosis process was the most e fficient purification process. The stimulants showed good removals for reverse osmosis process and not-satisfactory removals for microfiltration and electrodialysis reversal technologies. Regarding anti-inflammatories, similar removal e fficiencies were obtained for naproxen and ibuprofen using reverse osmosis and electrodialysis reversal (between 54.1% and 58.3%) and for this family of compounds, it was stated that microfiltration technology was not appropriate for the elimination of them. For the rest of the compounds under study, in all cases the removal e fficiencies were significantly better using reverse osmosis while electrodialysis reversal showed a better elimination capacity than microfiltration, but poor e fficiency (below 30%).
