Desalination 280 (2011) 197–202

Contents lists available at ScienceDirect

Desalination

j ourna l homepage: www.e lsev ie r.com/ locate /desa l

Development and optimization of the BDD-electrochemical oxidation of theantibiotic trimethoprim in aqueous solution

Teresa González a, Joaquín R. Domínguez a,⁎, Patricia Palo a,Jesús Sánchez-Martín a, Eduardo M. Cuerda-Correa b

a Department of Chemical Engineering and Physical Chemistry, Area of Chemical Engineering, Faculty of Sciences, University of Extremadura, Avda. de Elvas, s/n, E-06006 Badajoz, Spainb Department of Organic and Inorganic Chemistry, Faculty of Sciences, University of Extremadura, Avda. de Elvas, s/n, E-06071 Badajoz, Spain

⁎ Corresponding author. Tel.: +34 924289300x86888E-mail address: jrdoming@unex.es (J.R. Domínguez)

0011-9164/$ – see front matter © 2011 Elsevier B.V. Aldoi:10.1016/j.desal.2011.07.012

a b s t r a c t

a r t i c l e i n f o

Article history:Received 16 September 2010Received in revised form 2 June 2011Accepted 3 July 2011Available online 3 September 2011

Keywords:Electrochemical advanced oxidationprocessesElectro-oxidationAnodic oxidationBoron-doped diamond electrodesPharmaceuticalsTrimethoprim

The occurrence of antibiotics in the aquatic environment has led to an increasing concern about the potentialenvironmental risks and the maintenance and spread of antibacterial resistance among microorganisms.Electrochemical oxidation processes are promising technologies to treat low contents of toxic andbiorefractory pollutants in water. Anodic oxidation of Trimethoprim, the most frequently detected antibioticin surface waters, was carried out using boron-doped diamond electrodes at galvanostatic mode. A statisticaldesign of experiments has been used to study the influence of the different operating variables: pH (in therange 3–11), current intensity (from 0 to 320 mA cm−2), supporting electrolyte concentration Na2SO4 in therange 0–0.5 mol L−1, and solution flow rate between 1.25 and 10.80 cm3 min−1. Response SurfaceMethodology technique was used to optimize Trimethoprim degradation. Current intensity resulted to bethe main variable influencing Trimethoprim degradation, followed by salt concentration and pH. ANOVA testreported significance for five of the fourteen involved variables. An optimum Trimethoprim degradation of100% was found at pH 3, under a flow rate equal to 1.25 cm3 min−1, and with a current density equal to207 mA cm−2, using a supporting electrolyte concentration equal to 0.49 mol L−1.

; fax: +34 924289385..

l rights reserved.

© 2011 Elsevier B.V. All rights reserved.

1. Introduction

Considering the wide variety and different applications ofantibiotics and the ever increasing tendency to use them, it is nosurprise that these drugs are among the most ubiquitous environ-mental contaminants [1,2]. The occurrence of antibiotics in theaquatic environment has led to an increasing concern about thepotential environmental risks as well as about the maintenance andspread of antibacterial resistance among microorganisms. Develop-ment of antibiotic resistance of bacteria was demonstrated in surfaceand groundwater affected by the sewage treatment plant effluent [3].Whether antibiotics transported in the environment from human andlivestock sources lead to increased antibiotic resistant bacteria or havedeleterious effects on water quality is an important but largelyunresolved issue. The use of antibiotics may also accelerate thedevelopment of antibiotic resistance genes and bacteria, which shadehealth risks to humans and animals [4].

The main source of antibiotics in the environment is the excretionof incompletely metabolized drugs by humans and animals. Othersources may include the disposal of unused antibiotics and wastefrom pharmaceutical manufacturing processes [5]. Residential care

facilities and hospitals are known contributors of antibiotics tomunicipal wastewater [6]. Other contributors to surface andgroundwater are effluent from wastewater treatment plants [6]and industrial facilities (including pharmaceutical plants), andsurface run-off from concentrated animal feeding operations [1].

Recently, scientists have raised concerns about pharmaceuticalresidues detected in surface waters and wastewaters and theirpotential to cause adverse effects in humans and aquatic species[7,8]. Wastewater typically contains a number of medications andantibiotics that people have either used or discarded. Many of thesechemical compounds remain biologically active and some of them,especially antibiotics, stimulate the development of antibioticresistant bacteria.

Recent studies have confirmed the presence of low levels ofpharmaceuticals in estuaries, rivers, streams, and ground water aswell as in sediments [9,10]. The ecological effects of these contaminantsin coastal waters are largely unknown. However, there is growingevidence that some of these chemicals may have negative effects onreproduction in aquatic species.

Trimethoprim (TMP) [2,4-diamino-5-(3′,4′,5′-trimethoxybenzyl-pyrimidine)] (molecular formula C14H18N4O3, molecular weight290.32, CAS Number 738-70-5, pKa=7.2) is a very good antifolatedrug. It selectively inhibits the bacterial form of the dihydrofolatereductase enzyme [11,12]. TMP is among the most importantsynthetic antibiotics used worldwide in human and veterinary

198 T. González et al. / Desalination 280 (2011) 197–202

medicine acting as an inhibitor in the chemotherapy treatment due toits antifolate effect by interaction with dihydrofolate coenzymes [13–15]. TMP is mainly used in the prophylaxis and treatment of urinarytract infections, as well as for prevention and treatment of respiratoryor gastro-intestinal tract infections in cattle, swine and poultry [16].

In light of these facts, advanced drinking and wastewatertreatment options should be considered for the removal of thesedrugs. Between these options, Electrochemical Advanced OxidationProcesses (EAOP) [17–19] are very attractive for water and waste-water decontamination due to their low cost and high effectiveness,without needing addition of toxic chemical reagents and withoutproducing dangerous wastes. Among EAOP, Anodic Oxidation (AO) isthe most effective technique. It consists in the destruction ofpollutants in an electrolytic cell under the action of hydroxyl radicalsformed as intermediate from water oxidation at the surface of theanode [20,21].

M þ H2O→Mð·OHÞ þ Hþ þ e

� ð1Þwhere M(•OH) denotes the hydroxyl radical adsorbed on the anodeMor remaining near its surface.

Recently, AO has attracted great attention for wastewatertreatment due to manufacture of a new anode material. The use of aboron-doped diamond (BDD) thin-film anode brings technologicallyimportant characteristics such as an inert surface with low adsorptionproperties, remarkable corrosion stability and a wide potentialwindow in aqueous medium [22,23]. These properties confer to theBDD-anode a much greater O2 overvoltage than a conventional anodesuch as Pt, allowing the production of larger amounts of highlyreactive M(•OH) product from reaction (1). This material is consid-ered the best “non-active” anode. The material interacts so weaklywith •OH radicals that allows the direct reaction of organics with M(•OH) to give fully oxidized reaction products such as CO2 as follows:

42Mð·OHÞ þ C14H18N4O3→42M þ 14CO2 þ 9H2O þ 4NO2: ð2ÞDifferent authors have shown that several pollutants have been

completely mineralized by AO with BDD-anodes, whereas the use of aPt-anode under comparable conditions leads to weak decontamina-tion because of the formation of carboxylic acids that are hardly everoxidized with this material [24–28].

When BDD-anode is used, other weaker oxidizing species likeperoxodisulfate, (if sodium sulfate is used as supporting electrolyte, asin this work) can also be competitively formed with reactive oxygenspecies (see Eq. (3)):

2SO2−4 →S2O

2−8 þ 2e

−: ð3Þ

This investigation presents the results obtained from the BDD-AOof TMP. A statistical design of experiments (DOE) was used to studythe influence of the different operating variables: pH (in the range 3–11), current density (J) from 0 to 320 mA cm−2 which is equivalent toa current intensity, I, comprised between 0 an 4 A, given the fact thatthe electrode surface is equal to 12.5 cm2, supporting electrolyteconcentration (SEC), Na2SO4 in the range 0–0.5 mol L−1, and solutionflow rate (Qv) between 1.25 and 10.80 cm3 min−1. Response SurfaceMethodology (RSM) technique was used to optimize TMP degrada-tion (XT,%). So, the objective of the present study was to analyze theinfluence of the different operating variables and find out theoptimum values for this decontamination process.

Table 1Operating levels.

pH Qv (cm3 min−1) I (A) SEC (mol L−1)

Lowlevel(−1)

Highlevel(+1)

Lowlevel(−1)

Highlevel(+1)

Lowlevel(−1)

Highlevel(+1)

Lowlevel(−1)

Highlevel(+1)

5 9 3.64 8.42 1 3 0.125 0.375

2. Materials and methods

2.1. Experimental procedures

TMP of the highest purity available (> 98% TLC) was provided bySigma-Aldrich Spain. TMP stock solution was prepared with high

purity water obtained from a Millipore Milli-Q® system. All reagentsand solvents were of analytical grade and were purchased fromPanreac (Spain). The solution pH was adjusted with sodiumhydroxide and orthophosphoric acid (0.01/0.01 mol L−1) buffersolution. The supporting electrolyte used in this investigation,Na2SO4, was provided by Panreac in analytical purity grade.

A detailed description of the experimental procedure in terms ofelectrolytic cell, power supply and degradation experiments has beenprovided in a previous paper [29].

The initial concentration of TMP for all experiments was 1.7210−4 mol L−1 and temperatures was kept constant at 25±0.1 °C.TMP concentration was determined in all cases by HPLC with the aidof a Waters Chromatograph equipped with a 996 Photodiode ArrayDetector and a Waters Nova-Pak C18 Column (5 μm 150×3.9 mm). Itdisplayed a well defined peak for TMP at a retention time (tr) of2.42 min at 277 nm. For these analyses, samples of 50 μl were injectedinto the chromatograph and a 30:70 (v/v) methanol/water (10−2 Morthophosphoric acid) mixture was passed at a flow rate of1 mL min−1 as mobile phase.

Total Organic Carbon content (TOC) was determined as describedelsewhere [29].

2.2. Mathematical and statistical procedures

Typically, the one-factor-at-a-time approach has been used tostudy the influence of several operation variables (factors) on aparameter (response). Nevertheless, this classical methodology in-volves a large number of experiments and some important conclu-sions on the interaction among factors can be missed. To avoid theseimportant disadvantages, statistical design of experiment (DOE)provides an efficient alternative since it offers the possibility ofanalyzing the effect of variables and their responses minimizing thenumber of experiments [30].

In the present study, a routine DOE previously optimized by ourresearch group [29] has been used. StatGraphics® software provides auseful and powerful mathematical and statistical tool in order todevelop the experimental planning, to analyze the results and to drawconclusions on a rigorous basis. StatGraphics® Plus was used todesign and analyze the experimental results. A factorial centralcomposite orthogonal and rotatable (CCORD) design was used withtwelve replicates of the central experiment. This design involves theuse of a two-level factorial design with 2k points combined with 2kaxial points and n replicates of the central experiment, k being thenumber of factors. N, the total number of experiments with k factors isobtained according to Eq. (4):

N = 2k + 2·k + n: ð4Þ

In this study k was equal to four and n – as indicated above – wasconsidered to be twelve. The axial distance was 2 in order toguarantee an orthogonal and rotatable design. Hence, according toEq. (4), the total number of experiments resulted to be 36.

To evaluate the influence of pH, I, SEC and Qv and theirconcomitant effects on the effectiveness of the BDD electrochemicaloxidation of TMP, different levels were set for this pharmaceutical.

Table 3ANOVA. Results.

Factor Sum of squares Degrees of freedom F-ratio p-value

A: Coded pH 553.9 1 6.95 0.0154B: coded Qv 84.75 1 1.06 0.3142C: coded I 5602 1 70.2 0.0000D: codec SEC 1185 1 14.8 0.0009AA 15.82 1 0.20 0.6605AB 216.8 1 2.72 0.1140AC 187.0 1 2.35 0.1405AD 13.87 1 0.17 0.6808BB 11.88 1 0.15 0.7033

199T. González et al. / Desalination 280 (2011) 197–202

The selection of such levels was performed in antecedence (data notshown). Table 1 summarizes the operating levels used in the presentstudy.

In addition to the operating levels indicated in Table 1, in order toguarantee the rotatability of the design, a set of experiments withcoded values of variables equal to ±2 was performed. Hence, thisspecific study involves a range of 3–11 in pH, 1.25–10.80 cm3 min−1

in flow rate, 0–4 A in current and 0–0.5 M in the concentration ofsupporting electrolyte. Table 2 shows the experimental planning inDOE, and the response obtained in each of the experiments performedin this study.

BC 13.51 1 0.17 0.6848BD 60.45 1 0.76 0.3937CC 3874 1 48.6 0.0000CD 415.1 1 5.21 0.0330DD 164.2 1 2.06 0.1659Total error 1674 21

3. Results and discussion

As indicated above, Table 2 shows the experimental planning inDOE as well as the response obtained in each experiment in terms ofremoval efficiency of TMP (XT).

3.1. Numerical analyses. ANOVA report

The ANOVA analysis reports on the significance of the differentparameters. The results of the ANOVA tests are summarized inTable 3. According to the RSM, fourteen factors have beenconsidered and only five of them exhibit a p-value below a limitof significance of 0.05, so they are statistically significant. This latterconfirms the consistency of the model used to represent thebehavior of the interactive factors.

Table 2Experimental planning in DOE and obtained response in each experiment.

Run Coded pH Coded Qv Coded I Coded SEC XT(%)

1 1 1 −1 −1 83.82 1 −1 −1 1 99.83 −1 −1 −1 −1 62.74 1 −1 1 −1 99.85 −1 1 −1 1 68.26 0 0 0 −2 69.77 0 −2 0 0 99.98 1 −1 1 1 1009 0 0 −2 0 0.0010 −1 −1 1 1 10011 −1 1 1 −1 89.612 0 0 0 0 98.913 1 1 1 −1 99.714 0 0 0 0 98.515 −1 1 −1 −1 50.516 0 0 0 0 98.817 0 0 0 0 99.518 0 0 0 2 10019 2 0 0 0 99.520 0 0 0 0 99.321 1 −1 −1 −1 64.422 1 1 −1 1 94.923 0 0 0 0 97.924 −1 −1 −1 1 93.325 0 0 0 0 97.826 0 0 0 0 98.827 −1 −1 1 −1 96.628 0 2 0 0 92.829 0 0 2 0 99.830 0 0 0 0 98.931 −1 1 1 1 99.432 0 0 0 0 98.733 0 0 0 0 98.034 0 0 0 0 98.835 −2 0 0 0 82.736 1 1 1 1 99.6

Non-linear polynomial regression was carried out and thefollowing expression was obtained:

XT %ð Þ = 98:64 + 4:80·A−1:88B + 15:28·C + 7:03·D−0:703·A2 ++ 3:68·AB−3:42·AC−0:93·AD + 0:61·B2 + 0:92·BC−1:94·BD−−11:00·C2−5:09 CD−2:26·D2

ð5Þ

where the values of A (pH), B (flow rate, Qv), C (current intensity, I)and D (Na2SO4 concentration) should be coded. The values of TMPremoval, XT, are given in %. The correlation factor r2 is equal to 0.88.This regression predicts an optimumXT equal to 100% at pH equal to 3,operating under a flow rate of 1.25 cm3 min−1 (retention time, η,equal to 0.996 min), current intensity equal to 2.59 A (i.e., equivalentto a current density of 207 mA cm−2 provided that the electrodesurface is 12.5 cm2) and a salt concentration equal to 0.493 mol L−1.For each experiment, the difference between experimental XT andcalculated XT (according to Eq. (5)) is represented in Fig. 1 versus thespecific run number. It may be easily concluded from Fig. 1 thatresiduals are located in a random order at both the sides of the 0 axis.This fact suggests that no correlation can be appreciated and, hence,the randomization of the design is fully effective and no accumulationof experimental error is observed.

3.2. Graphical analyses and significance of variables

The modelization of results was made on the basis of the fourteenfactors indicated in Eq. (5). The Pareto plot (Fig. 2) is considered as agraphical expression of the ANOVA test. In this kind of plots, barsrepresent the standardized effects of each of the factors involved inthe analysis, considering them as the pH, current (I), supportingelectrolyte concentration (SEC), flow rate (Qv) and combinations of all

Fig. 1. Residuals versus run number.

Fig. 2. Pareto plot: standardized effects.

200 T. González et al. / Desalination 280 (2011) 197–202

of them. Nonfilled bars are a graphical representation of negative-affecting factors. In this case the Pareto plot indicates that flow rate,the squared of current, pH and SEC, and the combinations of I-SEC,pH-SEC, pH-I and Qv-SEC affect in a negative manner the removalefficiency of TMP. The mathematical expression of the negativeinfluence of these factors is the fact that all of them appear inexpression (5) behind a negative sign (−). On the contrary, filled barswithin the Pareto plot represent positive-affecting factors. In this caseI, SEC, pH, the squared of Qv, and the combination of pH-Qv and Qv-Iexert a positive effect on the removal efficiency of TMP, XT. It is worthnoting that a vertical rule stands near to 2 in the Pareto plot. In thisconnection it must be born in one's mind that this rule is closelyrelated to the signification level of ANOVA test, which has been fixedin a 95% of confidence. Accordingly, bars overcoming the vertical rulein the Pareto plot must be considered as inside the significance region(i.e., factors A, C, D, CC and CD). On the contrary, bars located behindthe vertical rule are not statistically significant (i.e., B, AA, BB, DD, AB,AC, AD, BC and BD). The Pareto plot also provides interestinginformation on how factors influence the final response, XT. Positive –

filled – bars indicate that an increase in a given variable makes XT

increase too. Negative – nonfilled – bars indicate exactly the opposite.Hence, as an example, it may be observed that as rate flow level risesthe value of XT decreases whereas if current intensity and saltconcentration rise, XT increases as well.

3.3. Influence of variables. Possible interactions

The use of a CCORD model makes it also possible to analyze themain effects of the involved variables. In particular, the influence ofsingle variables can be observed in Fig. 3 (left). In this figure four plotshave been drawn representing the effect of varying each of thevariables while the other ones are kept constant. It may be easilyobserved that current intensity, SEC and pH exhibit a quasi-linearpositive-affecting tendency, whereas Qv presents a slightly linearnegative-tendency, which suggests a scarcely negative influence ofthis latter variable. Current intensity appears to be the variable

Fig. 3. Single effects (left) and interactions (right) of flo

exerting a more remarkable effect on the removal efficiency of TMP,followed by the supporting electrolyte concentration (SEC) and pH.These results are in accordance with those depicted in the Pareto plot(Fig. 2). However, the fact that the optimal point is set in the lowestpH level (−2.00 in coded values) may be somewhat surprising. Thismay be attributable to the occurrence of an interaction involving pHand flow rate as well as to the fact that the squared pH factor (A2)exhibits a negative sign in Eq. (5). As a consequence of this specificcircumstance, the optimal point is located at a negative value of pH.Finally, the influence of the fourth variable – flow rate – resulted to benot significant.

Fig. 3 (right) shows the interaction between each two of the fourvariables studied. Each pair of curves represents the evolution of XT byvarying one variable in the extremes of the CCORDmodel, that is, withits pair variable equal to 1 (upper value) and equal to−1 (low value).If parallel lines are obtained it may be concluded that there is nointeraction between the variables under study. The occurrence ofcrossing lines indicates the opposite. The level of interaction of onevariable with the other is represented as an intermediate betweenthese two situations.

According to this latter, no interaction may be assumed in the caseof the pairs of variables AD, as the curves exhibit a parallel behavior.Hence, the modification of one of the variables does not significantlyaffect the other one. In contrast, for the pairs of variables AB thecorresponding lines clearly intersect as shown in Fig. 3 (right). Hence,the interaction between pH (A) and flow rate (B) is evident. Anintermediate situation is observed in the case of the pairs AC, BC, BDand CD, where plots cannot be considered as parallel but they do notintersect, either. Since these plots do not exhibit a parallel behavior, itmay be assumed that there is interaction, and consequently themodification of one of the variables affects the other one. However, itmust be assumed that the intersection takes place outside the regionanalyzed in this particular CCORD model.

3.4. Response surface and contour plot

The surface plot is a graphical representation that allowsevaluating the behavior of the whole system under analysis from aqualitative point of view. It is perhaps the most important graphicalrepresentation in the response surface methodology approach.

Fig. 4 represents the surface plot corresponding to Eq. (5). Thisfigure clearly shows that the response surface plot is a quite convexsurface within the whole studied region. Additionally, a maximum isapparent in the positive middle region of the coded I (+0.59). Fig. 4also includes the contour plot.

From a quantitative point of view, a statistical analysis of themodel yields to an optimum at the coded values equal to −2 for pH,−2 for QV, +0.59 for I and +1.94 for SEC, which is equivalent topH=3, flow rate Qv=1.25 cm3 min−1 (retention time, η, equal to0.996 min), current intensity I=2.59 A (current density, J, equal to

w rate, pH, dosage of Na2SO4 and current intensity.

Fig. 4. Response surface and contour plot model CCORD: current-SEC.

201T. González et al. / Desalination 280 (2011) 197–202

207 mA cm−2) and supporting electrolyte concentration SEC=0.493 mol L−1 of Na2SO4.

Given the fact that three of the four factors under study are insidethe significant region of the model (that is, all three of them arestatistically significant as their p-values are under 0.05), this optimummay be considered as statistically different from other close points.Under the optimal conditions a maximum value of TMP removal, XT,equal to 100% is predicted by the model.

In order to corroborate this assertion, an experiment wasperformed under the optimal conditions provided by the model(i.e., pH=3, Qv=1.25 cm3 min−1, J=207 mA cm−2 and SEC=0.493 mol L−1). The results obtained for TMP removal, COD andTOC, were 100%, 20.1% and 50.9%, respectively.

3.5. Rationalization of the results

As indicated above, current intensity is the most importantvariable influencing the removal of TMP. This fact may be explainedtaking into consideration Eq. (1). According to this equation, thegeneration of the very reactive hydroxyl radical is remarkablyinfluenced by the current intensity. However, Fig. 3 (left) suggeststhat current tends to present an optimum. Thus, it may be concludedthat above a certain value of current density (in this caseJ=207 mA cm−2) the TMP removal efficiency does not increasewith current. In this sense, the generation of large amounts ofhydrogen bubbles at the cathode may result in the hindering of theoxidation process in aqueous phase.

The second most important variable influencing the TMP removalwas the supporting electrolyte concentration. The influence of thisparameter is positive within the whole interval here studied. This factmay be justified by taking into account two factors. Firstly, an increasein the concentration of salt favors the electrical conductivity insolution. Secondly, the formation of such an oxidizing species likeperoxodisulfate (see Eq. (3)) is also favored by the increase of SO4

2−

ions in solution. As a consequence, the model predicts an optimumvalue outside this range (for +1.94).

The fact that flow rate exerts a negative influence on the removalefficiency of TMP was expectable since the larger the flow rate, theshorter is the residence time of TMP molecules inside the reactor.

Finally, for pH, the optimum point is set in the lowest pH level(−2.00 in coded values). This fact is attributable to the protonation ofamino groups of TMP, hence increasing their reactivity toward thehighly nucleophilic hydroxyl radicals.

4. Conclusions

Electrochemical Advanced Oxidation of Trimethoprim on Boron-Doped Diamond Electrodes in aqueous solution has been studied. Inorder to establish the influence of operating variables, an orthogonal,

rotatable factorial central composite statistical design of experimentswas performed. The results obtained clearly show that currentintensity is the variable exerting the most remarkable influence onthe removal efficiency of trimethoprim, followed by the supportingelectrolyte concentration and pH. On the other hand, the fourthvariable – flow rate – does not appear to exert a significant influencewithin the region of the CCORD model. The use of the ANOVA testreported significance for five of the fourteen variables involved. Themodel predicted an optimum oxidation point at pH=3, Qv=1.25 cm3 min− 1 (η=0.996 min), J=207 mA cm− 2, using aSEC=0.493 mol L−1. These optimal conditions were confirmedexperimentally and a value of XT=100% was achieved.

Acknowledgments

This investigation has been supported by the Comisión Inter-ministerial de Ciencia y Tecnología (CICYT) CTQ 2007-60255/PPQproject, as well as the Junta de Extremadura under PRI-07A031project.

References

[1] R. Hirsch, T. Ternes, K. Haberer, K.L. Kratz, Occurrence of antibiotics in the aquaticenvironment, Science of the Total Environment 225 (1999) 109.

[2] V.J. Harwood, J. Whitlock, V. Withington, Classification of antibiotic resistancepatterns of indicator bacteria by discriminant analysis: use in predicting thesource of fecal contamination in subtropical waters, Applied and EnvironmentalMicrobiology 66 (2000) 3698.

[3] B. Halling-Sorensen, S.N. Nielsen, P.F. Lanzky, F. Ingerslev, H.C. Holten Liitzhofl, S.E.Jorgensen, Occurrence, fate and effects of pharmaceutical substances in theenvironment—a review, Chemosphere 36 (1998) 357.

[4] N. Kemper, Veterinary antibiotics in the aquatic and terrestrial environment,Ecological Indicators 8 (2008) 1.

[5] D.B. Kathryn, K. Jerzy, T. Bruce, H.C. Timothy, B.M. Douglas, Occurrence ofantibiotics in hospital, residential, and dairy effluent, municipal wastewater, andthe Rio Grande in NewMexico, Science of the Total Environment 366 (2005) 772.

[6] A.C. Alder, C.S. McArdell, E.M. Golet, E.M. Molnar, H.M. Kuch, W. Giger,Environmental exposure of antibacterial agents in hospital and municipalwastewater and river water in Switzerland [Abs], Proceedings of 3rd InternationalConference on Pharmaceuticals and Endocrine Disrupting Chemicals in Water,March 19–21 2003, p. 236, (Minneapolis, MN).

[7] L. Ouattara, M.M. Chowdhry, C. Comninellis, Electrochemical treatment ofindustrial wastewater, New Diamond and Frontier Carbon Technology 14(2004) 239.

[8] M. Mascia, A. Vacca, A.M. Polcaro, S. Palmas, J.R. Ruiz, A. Da Pozzo, Electrochemicaltreatment of phenolic waters in presence of chloride with boron-doped diamond(BDD) anodes: experimental study and mathematical model, Journal ofHazardous Materials 174 (2010) 314.

[9] P. Cañizares, R. Paz, J. Lobato, C. Sáez, M.A. Rodrigo, Electrochemical treatment ofthe effluent of a fine chemical manufacturing plant, Journal of HazardousMaterials 138 (2006) 173.

[10] O. Scialdone, A. Galia, C. Guarisco, S. Randazzo, G. Filardo, Electrochemicalincineration of oxalic acid at boron doped diamond anodes: role of operativeparameters, Electrochimica Acta 53 (2008) 2095.

[11] G.H. Hitching, L.F. Kuyper, D.P. Baccananari, in: M. Sandler, H.J. Smith (Eds.),Design of Enzyme Inhibitors as Drugs, Oxford University Press, NY, 1998, p. 343.

[12] S.B. Raj, N. Stanley, P.T. Muthiah, G. Bocelli, R. Ollá, A. Cantoni, Crystal engineeringof organic salts: hydrogen-bonded supramolecular motifs in trimethoprimsorbate dihydrate and trimethoprim. o-nitrobenzoate, Crystal Growth and Design4 (2003) 567.

[13] Z. Bekçi, Y. Seki, M.K. Yurdakoç, Equilibrium studies for trimethoprim adsorptionon montmorillonite KSF, Journal of Hazardous Materials B133 (2006) 233.

[14] M.S. Diaz-Cruz, M.J. Lopez de Alda, D. Barcelo, Environmental behavior andanalysis of veterinary and human drugs in soils, sediments and sludge, Trends inAnalytical Chemistry 22 (2003) 340.

[15] A.L. Batt, D.S. Aga, Simultaneous analysis of multiple classes of antibiotics by iontrap LC/MS/MS for assessing surface water and groundwater contamination,Analytical Chemistry 77 (2005) 2940.

[16] F.C.C.R. De Paula, A.C. de Pietro, Q.B.J. Cass, Simultaneous quantification ofsulfamethoxazole and trimethoprim in whole egg samples by column-switchinghigh-performance liquid chromatography using restricted access media columnfor on-line sample clean-up, Journal of Chromatography. A 1189 (2008) 221.

[17] N. Bensalah, M.A. Quiroz Alfaro, C.A. Martínez-Huitle, Electrochemical treatmentof synthetic wastewaters containing Alphazurine A dye, Chemical EngineeringJournal 149 (2009) 348.

[18] G. Güven, A. Perendeci, A. Tanyolaç, Electrochemical treatment of simulated beetsugar factory wastewater, Chemical Engineering Journal 151 (2009) 149.

[19] C. Ahmed Basha, E. Chithra, N.K. Sripriyalakshmi, Electro-degradation andbiological oxidation of non-biodegradable organic contaminants, ChemicalEngineering Journal 149 (2009) 25.

202 T. González et al. / Desalination 280 (2011) 197–202

[20] B. Marselli, J. García-Gomez, P.A. Michaud, M.A. Rodrigo, Ch. Comninellis,Electrogeneration of hydroxyl radicals on boron-doped diamond electrodes,Journal of the Electrochemical Society 150 (2003) D79.

[21] M. Panizza, G. Cerisola, Application of diamond electrodes to electrochemicalprocesses, Electrochimica Acta 51 (2005) 191.

[22] E. Brillas, J.C. Calpe, J. Casado, Mineralization of 2,4-D by advanced electrochemicaloxidation processes, Water Research 34 (2000) 2253.

[23] C.A. Martínez-Huitle, S. Ferro, Electrochemical oxidation of organic pollutants forthe wastewater treatment: direct and indirect processes, Chemical SocietyReviews 35 (2006) 1324.

[24] E. Brillas, I. Sirés, C. Arias, P.L. Cabot, F. Centellas, R.M. Rodríguez, J.A. Garrido,Mineralization of paracetamol in aqueous medium by anodic oxidation with aboron-doped diamond electrode, Chemosphere 58 (2005) 399.

[25] P. Cañizares, J. Lobato, R. Paz, M.A. Rodrigo, C. Sáez, Electrochemical oxidation ofphenolic wastes with boron-doped diamond anodes, Water Research 39 (2005)2687.

[26] L. Boulbaba, B. Nasr, G. Abdellatif, Electrochemical oxidation of benzoic acidderivatives on boron doped diamond: voltammetric study and galvanostaticelectrolyses, Chemical Engineering and Technology 29 (2006) 944.

[27] C. Flox, S. Ammar, C. Arias, E. Brillas, A.V. Vargas-Zavala, R. Abdelhedi, Electro-Fenton and photoelectro-Fenton degradation of indigo carmine in acidic aqueousmedium, Applied Catalysis B: Environmental 67 (2006) 93.

[28] I. Sirés, F. Centellas, J.A. Garrido, R.M. Rodríguez, C. Arias, P.L. Cabot, E. Brillas,Mineralization of clofibric acid by electrochemical advanced oxidation processesusing a boron-doped diamond anode and Fe2+ and UVA light as catalysts,Applied Catalysis B: Environmental 72 (2007) 373.

[29] J.R. Domínguez, T. González, P. Palo, J. Sánchez-Martín, Anodic oxidation ofketoprofen on boron-doped diamond (BDD) electrodes. Role of operativeparameters, Chemical Engineering Journal 162 (2010) 1012.