Influence of the reactor configuration and the supporting electrolyte concentration on the electrochemical oxidation of Atenolol using BDD and SnO 2 ceramic electrodes

Electrochemical oxidation of β-blocker atenolol (ATL, 100 ppm) at different applied current densities (33, 50 and 83 mA·cm) using a reactor divided by an ion-exchange membrane and an undivided one was investigated. Two types of anodes were used for this purpose: a boron-doped diamond (BDD) anode and new low-cost ceramic electrodes made of tin dioxide doped with antimony (Sbdoped SnO2). Degradation was assessed using a high performance liquid chromatography, while mineralization by measuring total organic carbon (TOC) dissolved in sample. Except for the lowest current density, ATL was completely


Introduction
The continuous introduction of pharmaceutical and personal care products (PPCPs) to the environment and the increasingly restrictive legislation, make PPCPs be considered emerging contaminants. Many of these compounds are not removed in conventional wastewater treatment plants (WWTPs) and pass into the environment [1][2][3]. This can suppose a significant risk to the health of the ecosystem.
For this reason, in recent years, many efforts are being dedicated to the investigation of different processes for the elimination of these emerging compounds. Advanced oxidation processes (AOPs) are attractive methods as a solution to this problem, specifically, the electrochemical advanced oxidation processes (EAOPs) [4][5][6][7] due to their low cost, high effectiveness and their kindness to the environment, since no chemical additives are used. In the group of EAOPs, one of the most prominent technique is the anodic oxidation. This technique is based on the destruction of the contaminant by: (i) direct oxidation, i.e. transfer of electrons from the compound to the anode or (ii) indirect oxidation by electrogenerated species on the surface of the anode, such as hydroxyl radicals ( • OH): where M is the anodic surface and M( • OH) the hydroxyl radicals sorbed on the anodic surface. However, undesirable reactions that consume these radicals and benefit the formation of oxygen can occur: 2 ( • ) → 2 + 2 + 2 + + 2 − To avoid these reactions, the anode must possess a high overpotential towards the formation of O2. In this sense, there are two types of anodes [5]: non-active and active ones, whose main difference is based on whether the hydroxyl radicals are physisorbed or chemisorbed on the surface. In the latter case, the radicals are oxidized and a covalent bond of oxygen with the metal is formed according to Equation 4: Active anodes, such as Pt [8], graphite [9] and dimensionally stable anode (DSA) [10], are efficient for the oxidation of short-chain molecules, but not for complex molecules. In contrast, physisorbed radicals (non-active anodes) have a greater oxidant power and they are capable of mineralizing complex molecules to CO2 and H2O. As an example of these electrodes are boron-doped diamond (BDD) and oxides of Sn and Pb [11,12].
Nowadays, one of the most efficient anode is the BDD electrode [11,13]. This electrode has a high reactivity, a great oxygen overpotential and a high chemical stability. However, it is not viable on an industrial scale due to its high cost and its difficulty in the manufacture.
As an alternative, ceramic electrodes capable of competing with the BDD are being developed [14,15]. The advantages offered by the ceramic electrodes are their low cost, the increased active area due to its porosity, and the easiness of manufacture. In a previous study [16], new ceramic electrodes composed of SnO2 doped with Sb were studied for the electrooxidation of the antibiotic Norfloxacin.
In that study, it was found that this emerging compound could be effectively oxidized by these new electrodes. In every case, the electro-oxidation efficiency is considerably increased by means the use of an electrochemical reactor separated by an ion-exchange membrane, since the reduction reactions of the oxidant species and the intermediate compounds generated on the anode are avoided. Besides, with this reactor configuration, the more acidic pH reached in the anodic compartment enhances the electro-oxidation of the organic compounds of interest [16][17][18].
In this paper, the electrochemical study of these new ceramic electrodes for the degradation of another pharmaceutical compound recently found in the effluents of WWTPs, Atenolol (ATL), is carried out. ATL is a drug belonging to the group of beta-blockers used for the treatment of cardiovascular diseases such as hypertension, coronary heart diseases, arrhythmia, and myocardial infarction after the acute event. In patients is practically eliminated by the kidneys. The problem related to this compound is the low elimination rate in WWWTPs [19].
A lot of studies have recently been carried out using the photocatalysis technique to remove ATL, in which the addition of a catalyst is necessary [20][21][22]. Several investigations have been conducted on the removal of Atenolol by electrooxidation. Sirés et al. [23] studied the anodic oxidation of ATL using BDD and Pt as anodes at constant current at a pH of 3. They proved that BDD electrode is more effective than the Pt one due to the great amount of active • OH generated on the surface of BDD, and the minimization of the parasitic reactions.
Murugananthan et al. [24] compared different supporting electrolytes (Na2SO4, NaCl and NaNO3) for the degradation of ATL with BDD and Pt electrodes. They showed that the worst electrolyte for the mineralization was NaNO3, and the best conditions were obtained with the BDD electrode in Na2SO4 medium.
The objective of this work is to study the electrochemical oxidation of ATL with the new ceramic electrodes and thus prove their versatility for the removal of different compounds. Moreover, the influence of the reactor configuration (onecompartment and two-compartment reactor in the presence of a cation-exchange membrane) on the ATL electro-oxidation has also been evaluated. Under all the experimental conditions texted, the toxicity of the solutions submitted to the electro-oxidation is evaluated by Vibrio fisheri bacteria.

Stability of the new Sb-doped SnO2 ceramic electrodes
Electrodes based on SnO2 doped with Sb present good characteristics to be used as anodes in EAOPs. However, their low stability to anodic polarization in aqueous media limits their use [25,26]. In this work, accelerated service life tests were carried out in 0.5 M H2SO4 solution at 100 mA·cm -2 , considering that the electrode was deactivated when the potential overcomes 5 V from its initial value [27,28]. The tests were performed in a conventional three electrode cell, where the working electrode was the new Sb-SnO2 ceramic electrode, and as counter and reference electrodes a Pt and an Ag/AgCl, encapsulated with 3 M KCl, electrodes were used, respectively.

Electrochemical oxidation
The electrochemical oxidation processes were carried out for a solution composed of 100 ppm Atenolol (ATL, Sigma-Aldrich) and 0.014M of Na2SO4 (Sigma-Aldrich) as supporting electrolyte. In order to study the effect of the supporting electrolyte, tests at different concentrations of Na2SO4 (0.014, 0.05 and 0.1 M) were also carried out.
The ATL molecule contains two reactive sites: the aromatic ring and the amino group. The first site does not depend on the solution pH, but the latter does.
Atenolol is very hydrophilic and soluble in water, having a pKa value of 9.6.

8
The divided reactor, whose schematic representations is shown in Figure 1, is formed by two glass chambers of 250 cm 3 each, separated by a Nafion 117 cation-exchange membrane (from Dupont) [16]. The membrane contact surface was 11.34 cm 2 and it was equilibrated with the working solution for 24 h before each experiment.   [29,30] were calculated: The mineralization current efficiency (MCE) defined as a function of the organic matter eliminated (Equation 6) [31,32], was also determined to quantify the efficiency of the process.
where Δ[TOC]t (mg·L -1 ) is the removal of TOC after a certain time t(min), n is the number of exchanged electrons in the oxidation reaction, F is the Faraday constant (96485 C·mol -1 ), V is the volume of the electrolytic cell (L), m is the number of carbon atoms in the ATL molecule (14), I the applied current (A) and 7.2x10 5 is a conversion factor (60 s·min -1 x12000 mg·mol -1 ). Assuming that all the nitrogen in the molecule passes to ammonium ions ( 4 + ), as explained in other study [23], the number of exchanged electrons (n) was taken as 66 according to the following reaction:

Iodometry tests
Iodometry tests were carried out to determine the total oxidants present in the samples. Specifically, in this study the oxidants may only be persulfates and hydrogen peroxide. The hydroxyl radicals cannot be determined by this method due to their low lifetime [33,34]. To carry out these tests, 2 grams of KI were added in the sample. The Iions reacted with the oxidants to form I2 resulting in a yellow-orange solution.
The I2 formed was titrated with 0.01 M sodium thiosulfate (Na2S2O3) until the solution had a light-yellow colour. At this time a few drops of 1% starch (w/v) were added, and the solution turned dark brown-bluish coloured. Then, the titration was continued until the solution was transparent.

12
The concentration of H2O2 was determined using a colorimetric method [35]. And the difference between total oxidants and the amount of H2O2 generated was the amount of persulfates generated.

Toxicity measurements
In some cases, the oxidation process of the organic matter can lead to byproducts more toxic than the initial compound [36,37]. Ecotoxicity tests were carried out by bioluminescence measurements of the Vibrio Fischeri bacterium to evaluate this phenomenon.
This study was carried out by the Microtox M-500 equipment (Microbics, 1989).
The Microtox® system is a bioassay that examines the acute toxicity of environmental samples and pure compounds based on the reduction of the natural bioluminescence of the marine bacterium Vibrio Fischeri after 15 minutes of exposure at 15 °C [36]. Before each measurement, the samples were adjusted to a pH range between 6 and 8, with NaOH or H2SO4.  On the other hand, according to Ding et al. [27], the method of preparation for the same electrode composition also has a great influence on the stability, since the service life tests (100 mA·cm -2 in 1 M H2SO4) of SnO2 anodes obtained from electrodeposition reached a higher lifetime (15 h) that those obtained for the electrodes prepared from dip-coating (10 min). From these results, it is inferred that the ceramic electrodes of the present paper have a greatly improved stability in relation to other SnO2 electrodes. On the other hand, for the divided reactor, the anode type has a strong influence on the results, as observed in Figure 3c and 3d. In this case, the BDD electrode degraded more ATL than the ceramic one in both graphs. In addition, it was again obtained that the higher the applied current, the greater the degradation velocity of the ATL for both electrodes. For the BDD electrode, the ATL was completely removed at 120, 60 and 40 min for the corresponding current densities of 33, 50

Comparison of the electrochemical reactor and electrodes
and 83 mA·cm -2 . In the case of the ceramic electrode, for 50 and 83 mA·cm -2 , the atenolol was completely eliminated at 150 and 120 minutes respectively, whereas for the lowest applied current value the degradation achieved at the end of the electrolysis was 95%. Figure 3d shows that although the curves were similar for each electrode as the processes were under mass transport control in both cases, the degradation was faster for the BDD electrode. This was due to the fact that the BDD electrode generated more oxidizing species, as persulfates, hydrogen peroxide and sulfate radicals (Equations from 8 to 11), in addition to the hydroxyl radicals. This trend was not observed in the undivided reactor ( Figure 3Figure 3b) because these oxidizing species electrogenerated could be reduced at the cathode according to Equation (12). To verify this statement, iodometry tests were performed (Table 1), and it was demonstrated that the BDD electrode produced more persulfates than the ceramic one and this difference was more noticeable in the divided reactor. The presence of hydrogen peroxide was not observed, and radicals cannot be measured by this technique due to their low lifetime, as previously mentioned. In addition, from Figure 3 it is inferred that a greater degradation of ATL was achieved in the divided reactor since the membrane prevented the reduction of the intermediate products, and also due to the resulting acid pH value which enhances electro-oxidation. The exponential decrease of the relative ATL concentration observed for both reactors corroborates that the electrochemical system was controlled by mass transport with a typical process of pseudo-firstorder. This trend was already observed for the degradation of other organic compounds such as Norfloxacin under the same experimental conditions [16].
Therefore, the velocity of the ATL electro-oxidation reaction can be written as follows: where k is the kinetic constant, [ • OH] and [ATL] correspond to the concentration of hydroxyl radicals and Atenolol, respectively, and r is the reaction rate. For a given current density, the concentration of hydroxyl radicals is constant, resulting the velocity equation [43,44]: where kapp is the apparent kinetic constant (k·[ • OH]).
The decay of the relative ATL concentration with respect to the degradation time can be calculated by integrating the previous equation as follows: This equation allows the calculation of the apparent kinetic constants (kapp), which are represented as a function of the current density for both types of reactor and electrodes in Figure 4. As can be seen, a linear trend of kapp with the current density was observed for all conditions. This fact indicates that the formation of • OH and other oxidizing species, which reacts with ATL, were proportional to the applied current density for all cases. In the membrane reactor, higher kapp values were reached since the oxidizing agents and oxidized species could not be reduced at the cathode, therefore the degradation of the ATL was faster in this reactor. In this figure it was verified again that no significant difference was observed between both electrodes in the undivided reactor. By contrast, in the divided reactor, a dependency with the anodic material was observed, since the BDD presented a higher ATL degradation rate than the ceramic electrode under the same conditions because more oxidants were generated as previously mentioned.
Contrasting the previous figures, it can be inferred that ATL degradation was faster than its mineralization. In order to compare both phenomena, the extent of total electrochemical combustion (Φ) was calculated (Figure 6a and 6b) according to Equation 5. The fact that Φ was less than unity indicates that Atenolol was not directly mineralized to CO2, but the formation of intermediates took place, which could also be transformed into CO2 or other shorter-chain intermediates [47]. As can be observed, Φ rose with the electrolysis time and then, especially for the BDD electrode at higher current densities, Φ values remained practically constant. This last observation means that the organic byproducts generated in solution were degrading to CO2, indicative that the organic matter present was short-chain acids such as formic or oxalic acids [23,48]. The Φ values were also higher at higher i. The trends for both electrodes were the same, although the Φ values for the ceramic electrodes were slightly lower due to the lower amount of oxidizing species generated. The higher values of Φ obtained in the divided reactor were due to the cation-exchange membrane which prevented the oxidized products from being reduced, as previously commented.
Regarding the mineralization current efficiency (Figures 6c and 6d), this parameter was greater for the divided reactor for both electrodes since for this reactor it was possible to eliminate more organic matter. In addition, the electrode that presented better values of MCE was the BDD one. This fact was due to the interaction of the • OH radicals with the surface of each type of electrode [16]. In all cases, MCE values were low but these values are typical of these kind of processes [49,50]. In the undivided reactor (Figure 6c), this parameter remained practically constant with the time, for this reason in Figure 3b the curves were overlapped. For the divided one, it can also be observed that the MCE value decreased with time. This fact was due to the decrease of organic matter in the reactor and the increase of parasitic reactions.

23
Comparing the MCE values with the applied current density, it is observed that at lower current density the MCE was higher. This may be due to the increase of parasitic reactions such as the formation of O2 at higher values of applied current density (Equations 2 and 3) and the medium oxidation (Equations from 8 to 11) [51][52][53][54].

Influence of the supporting electrolyte concentration
In order to study whether the concentration of sodium sulfate influences the oxidation process, tests with different concentrations of Na2SO4 were carried out in the undivided reactor.   [58][59][60]. Regarding the ceramic electrode for these ranges of sodium sulfate and current density, at higher concentration of supporting electrolyte the mineralization of the ATL was slower. This may be due to either the fact that sulfate ions of higher concentrations could block the active sites of the electrode surface or to the enhancement of the parasitic reactions as was also observed in other studies [61].
The evolution of the apparent kinetic constant as a function of supporting electrolyte concentration at different applied current densities was also studied (not shown). As shown in Figure 4, when the current density increased, kapp was higher due to the increase in the formation of oxidizing species. Regarding the type of electrode, for the BDD, an increase in the concentration of sodium sulfate caused an increase in the kapp. This fact is due to a greater formation of S2O8 2- (Table 1)

Toxicity
Biological tests are the most appropriate when measuring the real effects, on the organism being studied, of any physical or chemical agent. In addition, these tests are very diverse, and these effects may occur at different levels, from effects on subcellular structures to effects on whole populations of one type of organism. In this work, luminescent marine bacteria Vibrio Fischeri were used to carry out ecotoxicity tests on the initial and final samples of each test.   Table 1 (0.014 M). In the undivided reactor although, the BDD electrode formed persulfates, they were not enough to show toxicity. Comparing the concentration of the supporting electrolyte for the highest applied current density (since more persulfates were generated), the final solution was more toxic at the highest concentration of Na2SO4 and using the BDD electrode. According to different studies [62,63], a sample is considered toxic when the TU value is equal to or greater than 10, therefore, any of the tested solutions can be considered toxic.

Conclusions
New low-cost ceramic electrodes made of tin dioxide doped with antimony have been prepared to carry out EAOPs. The ceramic substrate considerably increased the service lifetime of Sb-doped SnO2 electrodes.
Experimental results indicate that the electrochemical oxidation is very effective for the complete elimination of Atenolol (ATL). This technique strongly depends on the type of anode used, being the BDD electrode the most efficient due to the low interaction between the hydroxyl radicals (•OH) formed and the electrode surface. The Sb-doped SnO2 ceramic electrode degrades the organic compound more slowly, but also reaches high values of degradation and mineralization, which demonstrates its viability for this process.
Regarding the type of reactor, the use of a membrane improves the degradation rate of the ATL, the degree of mineralization and the mineralization current efficiency, since it avoids the reduction of the by-products formed during the oxidation and the oxidizing species electrogenerated. In addition, the pH decrease taking place in the anodic compartment of the electrochemical reactor enhances the electro-oxidation kinetics of ATL since the redox potential of the •OH radicals formed in the anode, is greater at lower pHs. Therefore, degradation of the organic compounds takes place faster in the presence of the cationexchange membrane.

32
It was also shown that the concentration of sodium sulfate affects the oxidation of ATL for both electrodes. For the BDD electrode, a higher degree of mineralization is achieved with the higher concentration of the supporting electrolyte. On the other hand, the opposite effect occurs for the ceramic electrode.
Finally, regarding the toxicity, it is verified that, although some final solutions were more toxic than the initial one, mainly due to the generation of persulfates, the solutions were not considered toxic in any case, which demonstrates that the electro-oxidation technique is compatible with the environment.