Water, Air, & Soil Pollution

, 224:1397

Electrochemical Degradation of the Reactive Red 141 Dye Using a Boron-Doped Diamond Anode


  • José Mario Aquino
    • Department of ChemistryUniversidade Federal de São Carlos
    • Department of ChemistryUniversidade Federal de São Carlos
  • Manuel A. Rodrigo
    • Department of Chemical Engineering, Facultad de Ciencias QuímicasUniversidad de Castilla La Mancha
  • Cristina Sáez
    • Department of Chemical Engineering, Facultad de Ciencias QuímicasUniversidad de Castilla La Mancha
  • Pablo Cañizares
    • Department of Chemical Engineering, Facultad de Ciencias QuímicasUniversidad de Castilla La Mancha

DOI: 10.1007/s11270-012-1397-9

Cite this article as:
Aquino, J.M., Rocha-Filho, R.C., Rodrigo, M.A. et al. Water Air Soil Pollut (2013) 224: 1397. doi:10.1007/s11270-012-1397-9


The electrochemical degradation of the Reactive Red 141 azo dye was done using a one-compartment filter-press flow cell with a boron-doped diamond anode. The response surface methodology (with a central composite design) was used to investigate the effect of current density (10–50 mA cm−2), pH (3–11), NaCl concentration ([NaCl]) (0–2.34 g L–1), and temperature (15–55 °C) on the system’s performance. The charge required for 90 % decolorization (Q90), the fraction of chemical oxygen demand removal after 6 min of electrolysis (COD6), and the fraction of total organic carbon removal after 90 min of electrolysis (TOC90) were used to model the obtained results. The lowest values of Q90 were attained at pH <4 in the presence of higher values of [NaCl] (>1.5 g L−1), due to the electrogeneration of active chlorine, present mainly as HClO. The value of COD6 was not affected by the solution pH, but increased with [NaCl] up to 1.5 g L−1. Higher temperatures (>40 °C) led to a decrease in COD6, as a consequence of side reactions. Higher values of TOC90, which can be reached only with strong oxidants (such as ·OH and Cl·), were efficiently attained at low [NaCl] values (<0.7 g L−1) in acidic solutions that inhibit the formation of ClO3 and ClO4. Finally, the obtained results allow inferring that most probably the mineralization of the dye starts with an attack on the chromophore group, followed by the degradation of intermediate species.


Response surface methodologyConductive-diamond anodeElectrochemical oxidationDye mineralizationDye electrooxidationChloride mediated oxidation

1 Introduction

Synthetic dyes, whose main application is in the textile industry, may stay in the environment for long periods due to their high chemical stability, thus presenting the potential of significant damage to aquatic and human lives (Umbuzeiro et al. 2005; Carneiro et al. 2010). Hence, and also because of increasingly rigid environmental regulations (Hessel et al. 2007), any dye-containing effluent needs to be properly treated before its disposal. The common treatment methods of precipitation and coagulation are inefficient to remove these organic pollutants, due to the high solubility of the dyes, which is promoted mainly by sulfonic groups (Muthukumar et al. 2005). The biological method is of great interest due to economical aspects; however, when treating dyes with azo bonds (−N = N–) in the aerobic mode, the removal efficiency is very low due to the generation of aromatic amines (Pandey et al. 2007; Chen et al. 2011). Moreover, limitations such as pH, salt concentration, carbon sources, and long treatment times are the main drawbacks of the biological treatment. On the other hand, electrochemical methods, particularly electrochemical oxidation, can be successfully applied, possibly with high organic removal rates and practically without operational limitations (Chen 2004; Martínez-Huitle and Brillas 2009; Peralta-Hernández et al. 2012). The choice of adequate electrode materials and operational conditions is required to achieve these characteristics (Rodrigo et al. 2010; Jara and Fino 2010). Moreover, the combination of different methods to remove the organic load can be successfully applied (Cañizares et al. 2008).

Boron-doped diamond (BDD) electrodes are being extensively used in many electrochemical areas, such as electrochemical degradation (Pereira et al. 2012b; Cañizares et al. 2005, 2006a, b, 2007, 2008; Sáez et al. 2007; Andrade et al. 2009; Rodriguez et al. 2009; Rodrigo et al. 2010; Aquino et al. 2011, 2012; Peralta-Hernández et al. 2012), synthesis of oxidants (Cañizares et al. 2009; Sánchez-Carretero et al. 2011), and electroanalysis (Pereira et al. 2012a), due to some intrinsic favorable characteristics (Panizza and Cerisola 2005). The main drawbacks of these electrodes are their high cost and the choice of adequate substrates for the BDD film deposition (Aquino et al. 2011); however, the use of BDD electrodes in the remediation of organic pollutants has resulted in high organic removal rates, with high current efficiencies and low energy consumption, as for instance in the electrooxidation of 3,4,5-trihydroxybenzoic acid (Panizza et al. 2008). This good performance is mainly due to the efficient oxidation of organics by electrogenerated hydroxyl radicals (·OH), which, as a consequence of their high oxidation potential and quasi-free nature on the BDD film, promptly oxidize organic pollutants (Kapalka et al. 2009). Furthermore, organic oxidation might also be mediated by other oxidants, such as Cl2/HClO/ClO or S2O82−, generated when the electrooxidation of the organic pollutant is carried out in the presence of chloride or sulfate ions (Panizza and Cerisola 2009), respectively.

The operational conditions of electrochemical methods needed to achieve high levels of oxidant production and, consequently, high removal rates depend on several variables, such as pH, current density, NaCl concentration ([NaCl]), and temperature. In order to analyze the effect of different variables at distinct levels (or values) in the electrooxidation of a pollutant, the experimental design methodology can be a powerful tool to find and analyze the level of the variables that lead to the best responses, i.e., the desired results (Prasad et al. 2008; Myers et al. 2009; Férnandez et al. 2010). The central composite design (CCD) is a well-known class of second-order designs (or also second-order models) used for sequential experimentation that involves a limited number of experiments in comparison to the most common methodology of investigating one variable at a time (often referred as factorial experiment). CCD is composed of factorial points (related to the estimation of linear and interaction terms), axial points (related to the estimation of quadratic terms), and center runs (allowing the estimation of the pure error) that can be used for optimization and prediction of possible synergistic effects between variables, which can only be visualized at certain levels (Anglada et al. 2011; Domínguez et al. 2012). The response surface methodology (RSM) is another powerful statistical tool that can be used to model, predict, and also find the best operational conditions to practically any experimental design, including CCD. Thus, these two statistical methodologies enable the development of empirical models for a single set of responses that were obtained under specific conditions previously determined by experimental design (often variable screening procedures are carried out). These methodologies need to be carefully used to obtain empirical models that describe real systems as accurately as possible. So, the main purpose of these methods is to match the region of operation to the region of interest during a process optimization (Myers et al. 2009).

The aim of the present work is to study the electrochemical oxidation of the Reactive Red 141 (RR 141) azo dye on a Si/BDD anode in a filter-press flow cell, in the presence or absence of chloride ions in solution. The effect of pH, current density, NaCl concentration, and temperature on the system’s electrochemical performance with respect to color, chemical oxygen demand, and total organic carbon removals is assessed using RSM based on a CCD matrix. The data generated by the use of these statistical tools on the degradation of the RR 141 dye are compared to results from the literature obtained using the one variable at a time methodology. A significant contribution of this work is to show that similar tendencies are observed with these different methodologies; however, the statistical methodology needs only a limited number of experiments and yields mathematical empirical modeling equations that allow to predict behaviors in the whole region of operation.

2 Experimental

2.1 Chemicals

All chemicals, including NaCl (a.r., Panreac), Na2SO4 (a.r., Panreac), H2SO4 (a.r., Sigma-Aldrich), NaOH (a.r., Panreac), and RR 141 (Dystar), were used as received. Doubly deionized water (Millipore Milli-Q system, resistivity ≥18.2 MΩ cm) was used for the preparation of all solutions.

2.2 Electrochemical Degradation Experiments

The electrochemical experiments were carried out in a one-compartment filter-press flow cell in the batch mode. Si/BDD and stainless steel (AISI 304) plates were used as anode and cathode, respectively. Both electrodes were circular (100 mm diameter, geometric area of 78 cm2) and the interelectrode distance was kept at 9 mm. The BDD film (Adamant Technologies) was prepared through the hot filament chemical vapor deposition technique on a monocrystalline silicon (p-doped) substrate and the specified boron content was 500 ppm. The Si/BDD anode was assembled on a circular stainless steel plate using a silver paste for electric contact.

A full 24 CCD coupled with RSM was used in the electrochemical oxidation experiments of the RR 141 dye. The investigated variables were current density (j), pH, [NaCl], and temperature (θ) at five different levels (−2, −1, 0, 1, 2); Table 1 shows the used ranges and coded levels for these variables. In order to evaluate the pure error and the lack of fit, three replications were carried out at the design center. All the experiments were carried out randomly using 0.6 L of 100 mg L−1 RR 141 in an aqueous 0.1 mol L−1 Na2SO4 solution, at a flow rate of 400 L h−1. The pH value was fixed and constantly monitored at desired levels by additions of a concentrated solution of H2SO4 or NaOH.
Table 1

Range and codification of the independent variables (Xi) used in the experimental design

Independent variables

Coded levels






Current density, X1 (mA cm−2)






pH, X2






[NaCl], X3 (g L−1)






Temperature, X4 (°C)






2.3 Analyses and Modeling

In order to monitor the color of the RR 141 dye solution, spectra were obtained from 200 to 800 nm using an UV–vis spectrophotometer (UV-1603, from Shimadzu) at certain time intervals, until complete decolorization was accomplished. The chemical oxygen demand (COD) of the dye solution was also monitored at certain time intervals until its total removal by sampling 2 mL of electrolyzed solution. These samples were mixed in glass tubes filled with a dichromate oxidizing solution from Merck (Spectroquant®). These tubes were left for 2 h at 150 °C in a digestion block to completely oxidize the organic matter, according to the closed reflux colorimetric method (Eaton et al. 2005). Subsequently, after the samples reached ambient temperature, their absorbance was read at 420 nm using a DR 2010 spectrophotometer. The organic load of the dye solution was analyzed through its total organic carbon (TOC) content by sampling 5 mL of electrolyzed solution at certain time intervals, using a Shimadzu TOC-5050 analyzer. The TOC content was analyzed by subtraction of the measured values of inorganic and total carbon using a high-temperature combustion method with a nondispersive infrared detector (Urbansky 2001). Changes in the color (monitored at 520 nm), COD, and TOC give different information about the progress of the electrolyses. To use them properly in an experimental design methodology, single outputs should be defined from these time-changing parameters, because responses of the experimental design should consist of single values for each set of inputs. Taking this into account, the responses used to model the system’s electrochemical performance were (1) the charge per unit volume of the electrolyzed solution required to attain 90 % decolorization (Q90), (2) the fraction of COD removal after 6 min of electrolysis (COD6), and (3) the fraction of TOC removal after 90 min of electrolysis (TOC90). Decolorization is a consequence of specific attack on the dye’s chromophores. Since these groups use to be easily oxidizable, it is expected that they be rapidly attacked, even during the very first stages of the electrolysis (Aquino et al. 2012). Q90 focuses specifically on the effect of the electrolysis on these chromophores and quantifies the charge necessary to get a 90 % removal of these groups. This is a value high enough to assess the effectiveness of the oxidation of these groups in a single parameter (the higher the efficiency, the lower the value of Q90) without reaching values in which the concentration of chromophores turns them into limiting reagents (because 10 % of chromophores still remain in solution). COD changes express the progress of the oxidation of the organic molecules contained in the solution. This progress starts with the oxidation of the dye’s functional groups, continues with the degradation of the dye molecules to simpler molecules, and finishes with the formation of inorganic carbon. For this reason, it is very interesting to assess the changes of this parameter during the initial stage, and 6 min of electrolysis means an arbitrary time not long enough to assure a significant mineralization of the organics, but significantly long to quantify oxidation trends different of those associated to chromophores. Finally, TOC informs about the conversion of organic carbon into CO2; that is, it provides information about the end point of the oxidation. An electrolysis time of 90 min is a period long enough to assure a significant progress of the oxidation reaction because it is longer than the time that would be required to completely mineralize the dye if the efficiency of the process was 100 %.

The linear and quadratic equations used to model the responses were \( Y={\beta_0}+\Sigma {\beta_{\mathrm{i}}}{X_{\mathrm{i}}} \) and \( Y={\beta_0}+\varSigma {\beta_{\mathrm{i}}}{X_{\mathrm{i}}}+\varSigma {\beta_{\mathrm{i}\mathrm{i}}}{X_{\mathrm{i}}}^2+\varSigma {\beta_{\mathrm{i}\mathrm{j}}}{X_{\mathrm{i}}}{X_{\mathrm{j}}} \), respectively, where β0,i,ii,ij is the model coefficient and Xi,j is the independent variable. The MS-Excel software was used for the mathematical calculation of the model coefficients as well as the analysis of variance (ANOVA). Then, the response surfaces were constructed with the Origin 8.0 software using the best fitted equations.

3 Results and Discussion

3.1 Fitted Empirical Models and Their Significance

The observed and predicted values obtained for the Q90, COD6, and TOC90 empirical modeling, along with the CCD matrix and the energy consumption associated to the observed Q90, can be seen in this article’s Electronic supplementary material (Table ESM-1). The fitted equations that best describe the Q90, COD6, and TOC90 responses are given below—nonsignificant coefficients were excluded according to ANOVA (for details, see Table ESM-2 in the Electronic supplementary material) and Student’s t test (at 95 % confidence level).
$$ \begin{array}{*{20}c} {Q^{90 }}=1.337-0.127{X_1}+0.297{X_2}+0.101{X_3}+0.080{X_4}-0.165{X_1}^2-0.048{X_2}^2-0.092{X_3}^2-0.111{X_4}^2- \hfill \\ 0.048{X_1}{X_2}+0.049{X_1}{X_3}+0.081{X_1}{X_4}+0.159{X_2}{X_3}+0.058{X_2}{X_4}-0.120{X_3}{X_4} \hfill \\\end{array} $$
$$ \begin{array}{*{20}c} \mathrm{CO}{{\mathrm{D}}^6}=93.61+16.96{X_1}+11.27{X_3}+2.64{X_4}-5.90{X_1}^2-1.97{X_2}^2-13.27{X_3}^2-4.53{X_4}^2+1.53{X_1}{X_3}+7.45{X_1}{X_4}- \hfill \\ 2.74{X_2}{X_3}-2.37{X_3}{X_4} \hfill \\\end{array} $$
$$ \begin{array}{*{20}c} \mathrm{TO}{{\mathrm{C}}^{90 }}=77.09+9.61{X_1}-6.30{X_2}-8.51{X_3}-2.72{X_4}-1.69{X_1}^2-2.20{X_2}^2-2.11{X_3}^2-1.19{X_4}^2-1.34{X_1}{X_2}+ \hfill \\ 4.84{X_1}{X_4}+3.95{X_2}{X_4}+3.96{X_3}{X_4} \hfill \\\end{array} $$

A good correlation was obtained between the predicted (calculated using the quadratic empirical equations) and observed values—see Fig. ESM-1 in the Electronic supplementary material. No lack of fit was observed in the models for the Q90, COD6, and TOC90 responses, according to Fischer’s F test (at 95 % confidence level—see Table ESM-2 in the Electronic supplementary material).

3.2 Color Removal

Figure 1 shows some of the response surfaces for Q90 (the higher this parameter, the less efficient is the attack on the chromophores) as a function of the independent variables pH, [NaCl], and θ. Among them, pH is the most important variable when electrooxidation is carried out in the presence of chloride ions, due to the generation of distinct pH-dependent chloro oxidant species (Cheng and Kelsall 2007).
Fig. 1

Response surfaces for the charge required for 90 % decolorization (Q90) as a function of a pH and [NaCl] (at 35 °C, using 30 mA cm–2), b pH and temperature (using 30 mA cm−2 and 1.17 g L−1 NaCl), and c pH and current density (at 35 °C, using 1.17 g L−1 NaCl)

Figure 1a (data obtained at 35 °C, using 30 mA cm−2) shows the Q90 behavior as a function of pH and [NaCl]. The best operational region, in which Q90 is lowest, occurs at acidic conditions (pH <4) and [NaCl] >1.5 g L−1, due to the generation of the HClO species. Basic conditions (pH >7.5) favor the generation of the OCl species, which is a weaker oxidant than HClO. Additionally, other chloro oxidants in higher oxidation states (like ClO3 and ClO4) can be generated at basic conditions, but these species are inefficient to oxidize organic molecules (Cañizares et al. 2006b) due to their very low oxidation potential (Ghernaout et al. 2011). Furthermore, the generation of these chloro oxidants can result in decreased availability of ·OH for organic oxidation on the BDD film surface, as pointed out by Polcaro et al. (2009):
$$ \mathrm{Cl}{{\mathrm{O}}^{-}}\left( {\mathrm{aq}} \right)+2\cdot \mathrm{OH}\to \mathrm{Cl}{{\mathrm{O}}_3}^{-}\left( {\mathrm{aq}} \right)+2{{\mathrm{H}}^{+}}\left( {\mathrm{aq}} \right)+2{{\mathrm{e}}^{-}} $$
$$ \mathrm{Cl}{{\mathrm{O}}_3}^{-}\left( {\mathrm{aq}} \right)+\cdot \mathrm{OH}\to \mathrm{Cl}{{\mathrm{O}}_4}^{-}\left( {\mathrm{aq}} \right)+{{\mathrm{H}}^{+}}\left( {\mathrm{aq}} \right)+{{\mathrm{e}}^{-}} $$

An interesting behavior is that Q90 in the absence of chloride ions is almost constant and practically independent of pH. Moreover, the Q90 levels attained under these conditions are lower than the ones obtained under basic conditions in the presence of chloride ions; this is a consequence of the generation of the ClO3 and ClO4 species, which are inefficient to attack the dye’s chromophore group in comparison to the ·OH or S2O82− species (Cañizares et al. 2006b; Ghernaout et al. 2011).

Figure 1b (data obtained using 30 mA cm−2 and 1.17 g L−1 NaCl) shows a similar behavior with pH: acidic solutions are promptly decolorized due to the HClO species’ attack on the unsaturated bonds of the dye’s chromophore group, as discussed above. The temperature did not have a significant influence on the color removal, since quite similar results were obtained in the whole range of temperatures studied; however, one would expect that high temperatures would lead to higher Q90 values due to a decrease of the concentration of the gaseous HClO species, as reported by Montanaro and Petrucci (2009), and the chemical decomposition of the S2O82− species, as observed by Panizza and Cerisola (2008).

Figure 1c (data obtained at 35 °C, using 1.17 g L−1 NaCl) shows that practically any applied current density could be efficiently used at acidic solutions, due to the formation of the HClO species that mediates indirect oxidation in the bulk of the solution.

The lowest value attained for the energy consumption associated to Q90 was 1.78 kW h m−3 (experiment 22: 30 mA cm–2, pH 3, 1.17 g L−1 NaCl, and 35 °C, see Table ESM-1 in the Electronic supplementary material). This value is higher in comparison to the one previously reported by Aquino et al. (2010b), who, using a Ti-Pt/β-PbO2 anode at strongly acidic conditions (75 mA cm−2, pH 1, 1.17 g L−1 NaCl, and 35 °C), attained a value of only 0.78 kW h m−3. On the other hand, Rajkumar and Kim (2006), using a dimensionally stable anode (36 mA cm−2, initial pH 6.2–6.5, 1.5 g L−1 NaCl, and 25 °C), reported a value of 3.45 kW h m−3 to obtain at least 95 % decolorization. Clearly, if the goal was only to decolorize the dye solution, Ti-Pt/β-PbO2 could be the anode of choice.

3.3 COD Removal

Figure 2 shows some of the response surfaces obtained for COD6 (the higher this parameter, the more efficient is the oxidation of organics during the first stages) as a function of pH, [NaCl], and θ. In contrast to what was described for color removal, the solution pH does not exhibit any significant influence on COD6, as can be observed in Fig. 2a (data obtained at 35 °C, using 30 mA cm−2). This behavior, which was also reported by Costa et al. (2010) based on conventional (one variable at a time) experiments, indicates that all the chloro oxidant species (Cl2, HClO, or OCl) can readily oxidize the RR 141 molecule. The insignificant COD6 variation with solution pH is actually surprising, since these chloro oxidant species have different oxidation potentials and are also present in different concentrations. Thus, one would expect distinct oxidation rates, as reported in the literature using a Ti-Pt/β-PbO2 anode (Aquino et al. 2010; Aquino et al. 2010a; Aquino et al. 2010b). Additionally, the oxidation rate of the RR 141 dye significantly increased with [NaCl] up to 1.2 g L−1, in comparison to the condition without chloride ions; this shows the importance of the indirect oxidation mediated by the chloro oxidants. On the other hand, high values of [NaCl] may enhance side reactions (Eqs. 4 and 5), with the consequent diminishment of the availability of ·OH for the generation of the chloro oxidant species and oxidation of the organic pollutant.
Fig. 2

Response surfaces for the COD removal percentage after a 6 min electrolysis (COD6) as a function of a pH and NaCl concentration (at 35 °C, using 75 mA cm−2), b pH and temperature (using 75 mA cm−2 and 1.17 g L−1 NaCl), and c pH and current density (at 35 °C, using 1.17 g L−1 NaCl)

Figure 2b (data obtained using 30 mA cm−2 and 1.17 g L−1 NaCl) shows that intermediate temperatures (between 30 and 40 °C) considerably enhance the value of COD6, practically at any solution pH, due to the increasing oxidation power of the oxidant species (·OH, Cl2, HClO, and OCl). However, at higher temperatures, it seems that side reactions that eliminate these species start to predominate, leading to a diminishment of COD6 above 40 °C; a similar behavior (regarding peroxodisulfate elimination) was reported by Panizza and Cerisola (2008) for the electrooxidation of the Acid Blue 22 dye using a BDD anode. Figure 2c (data obtained at 35 °C, using 1.17 g L−1 NaCl) shows that COD6 values of 100 % were attained at high current densities, independently of the solution pH, due to the increase in the production of oxidants; however, these current conditions lead to high energy consumptions and low current efficiencies.

The COD removal here reported for the RR 141 dye using the Si/BDD anode was significantly higher than the ones previously obtained using Ti-Pt/β-PbO2 (Aquino et al. 2010b) and DSA® (Rajkumar and Kim 2006) anodes. This is a consequence of the superior oxidation power of the BDD anode that enables the formation of highly oxidizing species such as ·OH and S2O82−.

Another interesting behavior is that total COD removal is attained before complete decolorization is observed (see Fig. ESM-2 in the Electronic supplementary material). This puzzling behavior, noticed during almost all experiments, might indicate that the RR 141 dye molecule undergoes addition reactions in the azo bonds and the aromatic rings (Deborde and von Gunten 2008), promoted by the chloro oxidants. The last reaction does not lead to color alterations, but it results in organochloride compounds that might not be oxidizable by dichromate ions, as pointed out previously (Aquino et al. 2012); thus, the measured COD value may actually differ from the real one. In fact, Baker et al. (1999) reported on a variety of compounds, including organochloride compounds, whose theoretical and experimental COD values were very different. More evidence on the formation of intermediate compounds will be given below. Additionally, the rapid COD decrease in both pH conditions (5 and 9) is in agreement with the response surface shown in Fig. 2a and might indicate that there is no specificity towards the addition of chloro oxidants to the aromatic ring.

3.4 TOC Removal

Figure 3 shows some of the response surfaces obtained for TOC90 (the higher this parameter, the more efficient is the mineralization of organics) as a function of the independent variables described earlier for Q90 and COD6. Figure 3a (data obtained at 35 °C, using j = 30 mA cm−2) shows that [NaCl] has a negative effect on TOC90: the increase of the chloride ion concentration leads to a decrease in TOC90. According to Deborde and von Gunten (2008), the chloro oxidant species can react with organic matter in three different ways: oxidation, addition, and substitution reactions. Although these reactions lead to structural modifications in the RR 141 dye molecules, with consequent changes in color and COD, they are not capable of mineralizing (transformation to CO2 and H2O) the RR 141 dye, due to the low oxidation potential of the chloro oxidants; consequently, the formation of organic intermediates is expected. Total mineralization can only be accomplished by ·OH radicals, due to their high oxidation power. The TOC90 decrease as [NaCl] increases might be a consequence of ·OH radical depletion on the Si/BDD surface, due to the reaction with chloride ions to generate the OCl or Cl · species (Grebel et al. 2010). On the other hand, even though mineralization is favored in the absence of chloride ions, small NaCl concentrations might contribute to that process, probably due to the formation of the chloro radical. This behavior was also observed during the electrochemical mineralization of the Acid Black 210 (Costa et al. 2010) and Reactive Blue 19 (Montanaro and Petrucci 2009) dyes, using a BDD anode, and the Reactive Red 195 dye, using a β-PbO2 anode (Song et al. 2010). Another possibility could be the formation of intermediate organochlorides, as discussed above, which are recalcitrant and consequently demand extended reaction times. Strong evidence that might support this speculation is the total COD removal before complete decolorization, as discussed above. Figure 3b (data obtained using 30 mA cm−2 and 1.17 g L−1 NaCl) shows that acidic solutions and low temperatures favor the mineralization of the RR 141 dye, because these conditions inhibit the formation of the ClO3 and ClO4 species. It is important to highlight that the high values of TOC90 reported in Fig. 3b were obtained in the presence of 1.17 g L−1 NaCl. Thus, the effect of the presence of NaCl on the mineralization degree depends on concentration, solution pH, temperature, and current density. Figure 3c (data obtained at 35 °C, using 1.17 g L−1 NaCl) shows that, independent of the applied current density, the highest values of TOC90 are attained at acidic conditions, which inhibit the formation of ClO3 and ClO4 and the consequent diminishment of the amount of ·OH on the BDD film. Moreover, high current densities lead to an increase in the mineralization rate due to the higher rate of ·OH formation; however, lower current efficiencies are obtained under this condition.
Fig. 3

Response surfaces for the TOC removal percentage after a 90-min electrolysis (TOC90) as a function of a pH and NaCl concentration (at 35 °C, using 75 mA cm−2), b pH and temperature (using 75 mA cm−2 and 1.17 g L−1 NaCl), and c pH and current density (at 35 °C, using 1.17 g L−1 NaCl)

The long time required for TOC removal in comparison to the ones required for color and COD abatement suggests that the oxidative attack starts on the chromophore groups of the RR 141 dye molecules, with the consequent generation of intermediates. These intermediates are associated to COD abatement, but not to mineralization, as was discussed by Sáez et al. (2007). Thus, the higher stability and complexity of the electrolyzed compounds lead to longer electrolysis times and, consequently, higher electric energy consumptions to attain complete mineralization.

4 Conclusions

The use of the response surface methodology (with a central composite design) enabled to study the color, COD, and TOC removals as a function of current density, pH, [NaCl], and temperature. The obtained mathematical empirical models fitted well the experimental values, whereas the resulting 3D surfaces showed similar color, COD, and TOC removal behaviors as the ones reported in the literature for one variable at a time experiments (Aquino et al. 2010, 2012; Costa et al. 2010; Montanaro and Petrucci 2009; Rajkumar and Kim 2006; Sáez et al. 2007; Song et al. 2010). This behavior confirms that statistical tools are a good choice for optimization, modeling, and studying of the electrochemical degradation process of organic compounds, such as synthetic dyes. The color removal showed a significant dependence on pH, due to the generation of the HClO species at acidic solutions (pH >3). On the other hand, the COD removal was almost pH independent, indicating that the RR 141 dye molecule is oxidized by any chloro oxidant species, with the possible formation of organochloride by-products. Higher values of TOC removal were attained at low chloride ion concentrations (more hydroxyl radicals available), due to the possible formation of chloro radicals and diminishment of the formation of organochloride compounds. Additionally, acidic solutions (pH >3) led to high mineralization rates, due to the inhibition of the side reactions that yield the ClO3 and ClO4 species. The higher electrolysis time required to attain mineralization, in comparison to color and COD removal, indicates that the degradation of the RR 141 dye starts with an attack on the chromophore group, followed by the oxidation of the generated intermediates.


This work was supported by the Junta de Comunidades de Castilla La Mancha, Spain (project PEII11-0097-2026). The Brazilian agencies CNPq and CAPES (scholarship for J. M. Aquino) are gratefully acknowledged. Dystar is also acknowledged for supplying the dye sample.

Supplementary material

11270_2012_1397_MOESM1_ESM.pdf (2.1 mb)
ESM 1(PDF 2.10 mb)

Copyright information

© Springer Science+Business Media Dordrecht 2012