Abstract

The elimination of total and fecal coliforms, from raw surface water, was carried out by electrochemical oxidation using either boron doped diamond (BDD/Ti) or graphite (GP) anodes, in a chloride-free medium. The optimal values of the operation parameters, maximizing the coliform elimination percentage, were determined using statistical experimental design. The current density (j: 2–20 mA/cm2), the conductivity (σ: 500–900 μS/cm) and the anode materials (An) were considered as variables to perform the Box-Behnken experimental design together with the response surface methodology analysis for optimization. The statistical analysis indicated that, in the evaluated range, the disinfection efficiency increased with an increase in j and decreased with an increase in σ. The following optimal conditions for the elimination of total and fecal coliforms were found: j: 10 mA/cm2, σ: 500 μS/cm and BDD/Ti used as anode material. The BDD/Ti electrode let to achieve complete coliform elimination after ca. 20 min of reaction while the GP one needed ca. 27 min. In water treated with both BDD/Ti and GP anode, after 7 days, any coliforms growth was observed. As a result of the oxidation process, the total organic carbon and nitrite concentration decreased while nitrate concentration increased.

INTRODUCTION

The purpose of disinfection processes is to eliminate pathogens as bacteria and viruses responsible for waterborne diseases. Nevertheless, according to statistics of the Water Project Foundation, 783 million people do not have access to clean and safe water and almost 2.5 billion do not have access to adequate sanitation, being exposed to a permanent health risk. Children are especially vulnerable to untreated water and it is estimated that one out of every five deaths in the population under 5 years old is due to a water related disease. The situation is even more complex in developing countries where around 80% of illnesses are linked to both poor quality of water and sanitation conditions. Consequently, it is of high priority to develop and evaluate new low cost disinfection systems accessible for rural communities (Hunter et al. 2009; GilPavas et al. 2014; Kostyla et al. 2015).

Different technologies have been used to eliminate pathogens from water. Chlorination is the most common and effective way to remove a wide range of microbial pathogens (Barashkov et al. 2009). However, the addition of chlorine induces the formation of potentially harmful disinfection by-products (DBPs). Over 600 DBPs can be formed as a result of the contact of chemical oxidants with organic matter present in water (Castro-Hermida et al. 2008; Park et al. 2016). Moreover, in some rural areas, where continuous transportation of chemical oxidants is unviable, it is necessary to develop and implement processes using in-situ formation of disinfection agents. Searching for alternative treatment processes that can overcome these drawbacks, electrochemical disinfection of water has emerged as a promising option. It has shown to be cost efficient, environmentally friendly and competent in the elimination of microorganism from water (Bergmann & Rollin 2007; Ndounla et al. 2014). This technique implies in-situ formation of oxidizing species (for instance, OH, H2O2, O3, peroxodisulfate, peroxodicarbonate and peroxodiphosphate) by the pass of electric current through a suitable electrode, avoiding transportation and storage of hazardous chemicals (Hussain et al. 2014).

The effectiveness of electrochemical disinfection depends on different parameters including: electrolyte-cell configuration, electrode material, electrolyte type, solution composition, type of microorganisms, mass transfer conditions and current density/electrode potential ratio (Kerwick et al. 2005). In relation to the electrode material, the usefulness of different anodes (such as Pt, graphite (GP), activated carbon fibers, IrO2, RuO2, PbO2, SnO2, TiO2, boron doped diamond (BDD), etc.) has been reported in open literature (Kerwick et al. 2005; Polcaro et al. 2007; Jeong et al. 2009; Akhavan & Ghaderi 2010; Bruguera-Casamada et al. 2016). However, some of them (for instance, Pt and RuO2) require the presence of chloride ions or/and chlorine containing species (for instance, Cl2, HClO and/or ClO). Consequently, the formation of harmful disinfection by-products (named DBPs) can be expected. In contrast, the application of BDD anode let to produce the OH radical, which is very reactive and highly appropriated for environmental applications. Moreover, it is very effective in the absence of Cl ion, preventing the accumulation of active chlorine and the possible formation of DBPs (Bruguera-Casamada et al. 2016). However, the cost of this type of electrodes is significantly higher than the other materials (for example, GP) (Martínez-Huitle 2007). The GP has also been reported to present an inherent antibacterial activity, attributed to membrane stress induced by sharp edges. It can result in physical damages of cell membranes, leading to the loss of bacterial membrane integrity and the leakage of its RNA (Akhavan & Ghaderi 2010). Liu et al. (2011) have reported that GP dispersion exhibits a moderate cytotoxicity with the Escherichia coli cell (inactivation percentage of 26.1 ± 4.8%). On the other hand, Qi et al. (2015) have detected that any inactivation of BDD anode was observed when used without current density. Therefore, the disinfection occurs mainly due to the formation of oxidizing species on the anode material.

Thus, the aim of this work is: (1) to define a simple, low cost system for the electrochemical treatment of water available in rural areas of developing countries, able to eliminate pathogenic microorganisms; and (2) to assess its disinfection efficiency on both total and fecal coliform bacteria. Two different anodes materials (GP and BDD/Ti) were tested separately under different operational conditions (in the absence of chloride ions). The Box-Behnken experimental design (BBD) and the response surface methodology (RSM) were used to optimize the electro-oxidation process.

MATERIALS AND METHODS

Reagents and water samples

Water samples were taken directly from a creek passing through EAFIT University, in Medellín-Colombia, and processed right after being collected to avoid bacterial death during storage. The commercial medium Petrifilm (EC Petrifilm™, 3 M Microbiology, St Paul, MN, USA) was used to detect and count E. coli (blue colonies) and total coliform (violet colonies) bacterias. Macherey-Nagel Nano-Color reagent kits (references: 0–26, 0–94, 0–68, 0–65) were used to quantify chemical oxygen demand (COD), total organic carbon (TOC), nitrites (NO2) and nitrates (NO3). Additionally, sodium thiosulfate (Na2S2O3, 99.5% pure, Aldrich Co.) was used as support electrolyte. Table 1 presents the characterization of the raw water used in this study.

Table 1

Physicochemical and bacteriological properties of studied water samples

ParameterValue
pH 6.6 
Conductivity, μS/cm 500 
Turbidity, NTU 
COD, mg/L 46.5 
TOC, mg/L 18 
NO3, mg/L 0.870 
NO2, mg/L 0.183 
Total coliforms (TC), CFU/mL 6,750 
Fecal coliforms (FC), CFU/mL 5,600 
ParameterValue
pH 6.6 
Conductivity, μS/cm 500 
Turbidity, NTU 
COD, mg/L 46.5 
TOC, mg/L 18 
NO3, mg/L 0.870 
NO2, mg/L 0.183 
Total coliforms (TC), CFU/mL 6,750 
Fecal coliforms (FC), CFU/mL 5,600 

Analytical methods

For each sample of water, the number of microorganisms was stated as colony forming units (CFU). The concentration of viable cells was evaluated using the plate count method. The growth medium contained Violet Red Bile, a gelling agent soluble in cold water, and an indicator of beta-glucuronidase activity that facilitates colony quantification. Plates were incubated at 30 °C for 24 h. Next, the number of colonies on the plates was counted. For all counts, at least three replicate plates were used and the results were reproducible within an average relative error of ±10%. During the experiments, at each sampling time, 1 mL of suspension was withdrawn and immediately quenched with excess of Na2S2O3 (10 mM) to eliminate the residual disinfectants in the sample. Then, each sample was kept in a sterile container, in darkness, at room temperature (25 ± 1 °C). The COD analyses were performed following the closed reflux method with colorimetric measurements (method 5220D). The TOC measurements were carried out following the method 5310D. The NO3 and NO2 concentrations where determined according to the Standard methods 4500B and 4500D, respectively. For NO2, NO3, COD and TOC measurements, a Nano-Color thermo-reactor and spectrophotometer were used. Turbidity was determined following the 2130B standard method. Additionally, conductivity and pH were determined with a Thermo Scientific Orion 5 start multiparameter analyzer. In all cases, the average value of the measurements is reported. Error bars (indicating the error or uncertainty in a reported measurement) present the variability of data graphically.

Disinfection process and electrolytic system

The electro-oxidation experiments were carried out in a Plexiglas, continuously stirred, batch jacketed reactor (80 mL). Two vertically-placed plates, with 1 cm separation between them, were used as electrodes. Two different anode materials were evaluated: (i) a graphite rectangular plate (GP) (with dimensions of 34.1 × 24 × 6 mm, and a total surface area of 21.9 cm2) and (ii) BDD film deposited on a titanium substrate (BDD/Ti) (BDD coating: p-doped, poly-crystalline, ≈3 micrometer thick; doping level: 700–800 ppm boron, with dimensions 15 × 50 × 5 mm. Supplier: Fraunhofer, USA). The cathode consisted of a titanium electrode (Ti (99. 4%) with dimensions of 3.6 × 0.11 × 2.7 cm) or a BDD/Si electrode with the same dimensions of BDD/Ti one. The voltage was regulated with a BK-Precision source (0–30 V, 0–5 A). Test's temperature was maintained at 25 °C using a Polyscience 712 thermostat connected to the reactor jacket. For disinfection experiments, the total reaction time was fixed at 12 min, the stirring intensity at 350 RPM and the anode available area was equal to 4 cm2. At each sampling time, in order to avoid any contamination, the coliform measurements were performed immediately inside a sterile chamber. The possible bacterial growth was monitored with coliform measurements after 24 h, 94 h and 7 days.

Experimental design and statistical analysis

According to the literature (Brillas & Martínez-Huitle 2015), several operational factors can affect the efficiency of electrochemical disinfection process. The following ones were taken into consideration in this study: current density (j), conductivity (σ), electrode surface area, type of anode material and stirring velocity. The other variables, such as pH and bacterial population, corresponded to the natural conditions of the raw water. In order to evaluate the effect of process variables on the efficiency of coliform elimination and to establish their optimal values, two statistical experimental designs were randomly programmed, to avoid any systematic error, using Statgraphics Centurion XVI. They were performed as follows:

  • The first group of experiments, corresponding to a fractional factorial design, 25-1 with three central points, consisted of 19 experimental runs with two replicas (not shown here). It determined that j and σ present, statistically, the most significant effect on the disinfection process. Similar results were reported previously (Mascia et al. 2012; Mascia et al. 2013; Long et al. 2015; Qi et al. 2015). Notice that an increase in j enhanced the process efficiency and increased the energy cost. Therefore, j, σ and anode material were chosen as operational variables to be optimized in the next experimental design.

  • In the second group of experiments, the BBD coupled with the RSM was applied to evaluate the individual and synergetic effects of the most significant operation variables. It consisted of 15 tests with two replicas. It let to: (i) reduce the number of experimental trials, comparing to a complete factorial design, (ii) adjust a second order polynomial model and predict the response variable from the experimental factors, (iii) establish limitations of the fitted model with regression coefficient determination, and (iv) determine the optimal conditions of the process. It is important to remark that RSM was not used to understand the mechanism of coliforms inactivation, but to determine the optimal operating conditions at certain operating specifications. Thus, the following independent variables (factors) and their operational ranges were selected, considering also the results of the first experimental design: (i) current density (j: 2–20 mA/cm2); (ii) conductivity (σ: 500–900 μS/cm); and (iii) anode material (An: GP or BDD/Ti). Since An is a discrete variable, the following three levels were defined to understand the influence of the anode and cathode material on the electrochemical disinfection process: level 1: corresponding to BDD/Ti anode and Ti cathode; level 2: BDD/Ti used as anode and BDD/Si as cathode; and level 3: corresponding to GP/Ti anode and Ti cathode. When required, the initial conductivity was adjusted adding the corresponding volume of Na2SO4 (1 mM). The upper conductivity value in the studied range was fixed according to Colombian regulation for drinking water (1,000 μS/cm).

The efficiency of the electrochemical process was defined based on total coliform (%ETC) and fecal coliform (%EFC) elimination percentages. They were calculated according to Equation (1), where Zi and Z account for the initial and final values of FC and TC measured concentrations.  
formula
(1)
For the experimental design, the independent variables and their levels, summarized in Table 2, were coded according to Equation (2). The experimental results were adjusted to a second-order multi-variable polynomial model (Equation (3)):  
formula
(2)
 
formula
(3)
Table 2

Variables (factors) and their levels for experimental design

VariableCoded factors, X
−101
Level 1Level 2Level 3
A: Current density (j) mA/cm2 11 20 
B: Conductivity (σ), μS/cm 500 700 900 
C: Electrodes' material (anode–cathode)a 
VariableCoded factors, X
−101
Level 1Level 2Level 3
A: Current density (j) mA/cm2 11 20 
B: Conductivity (σ), μS/cm 500 700 900 
C: Electrodes' material (anode–cathode)a 

a1: BDD/Ti-Ti, 2: BDD/Ti-BDD/Si, 3:GP-Ti.

where Yi is the predicted response variable, β0, βi, βii, βij are the regression coefficients for: the intercept, lineal, square, and interaction terms, respectively; and Xi and XJ are independent variables. The quality of the fitted model and its prediction capacity were judged from the variation coefficient, R2. Determination of the significant main and interaction effects of factors, influencing the %EFC and %ETC, were followed by analysis of variance (ANOVA), Pareto diagram, response surface plot and variation coefficients. More details on the applied methodology have been reported elsewhere (GilPavas et al. 2014).

The energy consumption (EC) during the EO process (expressed in kWh/m3) was calculated using Equation (4):  
formula
(4)
where I = average applied current (A); V = average cell potential (V); t = electrolysis time (h); VR = solution volume (m3).
The operational costs (OC, USD/m3) were calculated as follows:  
formula
(5)
where Celectrolyte is the amount of electrolyte consumed (kg/m3), a and b correspond to the cost of electrical energy (0.18 USD/kWh) and NaSO4 electrolyte (1.23 USD/kg), respectively, matching to the Colombian market prices in October 2016.

RESULTS AND DISCUSSION

To examine the combined effect of the three independent process variables (j, σ and electrode material) on %ETC and %EFC, 15 experiments with two replicas were performed. Independent variables, their experimental ranges and the corresponding experimental data (their average values) are given in Table 3. The %ETC and %EFC values ranged from 68% to 100% and from 74% to 100%, respectively. The following changes in the OC, varying j in the range of 2–20 mA/cm2, were determined: 0.014–1.17 USD/m3 and 0.022–1.176 USD/m3 for BDD/Ti and GP electrodes, respectively. The operational conditions that maximized both the %ETC and %EFC correspond to the run 7 (Table 3). However, in this case, the OC are one of the highest (ca. 1.17 USD/m3). Notice that using j as low as 2 mA/cm2, it is still possible to achieve ca. 90% of ETC with the OC of 0.025 USD/m3 (run 6, Table 3). These results imply the possibility of total coliform elimination from water without the necessity to induce high electric current, indicating lower operational costs and higher reaction time. The use of GP-Ti electrodes also obtained very high %ETC (ca. 96%) for intermediate values of j (11 mA/cm2) and σ (500 μS/cm) (see run 5 in Table 3), representing intermediate OC (0.054 USD/m3). Considering the difference in price between BDD (25 USD/cm2, Fraunhofer USA) and GP ($0.84 USD/cm2 available at Medellin's local market) electrodes, the GP one becomes an interesting alternative for bacteria elimination.

Table 3

The experimental and predicted results for the three factors (j, σ and An) and their corresponding values

Runj (mA/cm2)σ (μS/cm)An%ETC
%EFC
OC (USD/m3)
YExpYPredYExpYPredYExpYPred
11 900 88 87.7 81 83.47 0.357 0.38 
900 75 73.8 77 73.44 0.014 0.018 
11 700 84 82.3 82 80.67 0.462 0.450 
11 900 89 89.0 82 83.35 0.340 0.370 
11 500 96 93.0 94 92.65 0.540 0.510 
500 89 87.20 87 87.76 0.025 0.030 
20 500 100 100 100 100 1.172 1.220 
11 700 83 82.30 81 80.67 0.455 0.450 
20 900 100 100 100 99.04 1.040 1.010 
10 20 700 95 93.21 99 98.41 1.170 1.150 
11 11 500 96 96.57 96 93.02 0.540 0.510 
12 700 68 69.79 77 77.59 0.022 0.026 
13 11 700 80 82.33 79 80.67 0.045 0.045 
14 700 75 76.21 74 76.21 0.021 0.025 
15 20 700 96 94.79 99 96.79 1.176 1.15 
Runj (mA/cm2)σ (μS/cm)An%ETC
%EFC
OC (USD/m3)
YExpYPredYExpYPredYExpYPred
11 900 88 87.7 81 83.47 0.357 0.38 
900 75 73.8 77 73.44 0.014 0.018 
11 700 84 82.3 82 80.67 0.462 0.450 
11 900 89 89.0 82 83.35 0.340 0.370 
11 500 96 93.0 94 92.65 0.540 0.510 
500 89 87.20 87 87.76 0.025 0.030 
20 500 100 100 100 100 1.172 1.220 
11 700 83 82.30 81 80.67 0.455 0.450 
20 900 100 100 100 99.04 1.040 1.010 
10 20 700 95 93.21 99 98.41 1.170 1.150 
11 11 500 96 96.57 96 93.02 0.540 0.510 
12 700 68 69.79 77 77.59 0.022 0.026 
13 11 700 80 82.33 79 80.67 0.045 0.045 
14 700 75 76.21 74 76.21 0.021 0.025 
15 20 700 96 94.79 99 96.79 1.176 1.15 
In order to determine the main and double-interaction effects of factors influencing the disinfection process, an ANOVA was performed. It consists of classifying and cross-classifying statistical results, decomposing the contribution of each variable (factors) and double-interactions in the variance of each response variable. Table 4 presents the ANOVA for %ETC and %EFC and the p statistical parameter. The p-values were used to identify experimental parameters that present statistical influence on the particular response. The value of p ≤ 0.05 means that the studied variable is statistically significant with the 95% confidence level. As shown in Table 4, the j and σ present p-values ≤ 0.05 for the three response variables, suggesting that they influence directly the disinfection efficiency and the operational costs. Moreover, the following double-interactions of different factors also fulfil the statistical constraint: for %ETC: σ – σ and j-σ; for %EFC: σ – σ; and for OC: j-j. In numerical terms, it implies that their corresponding coefficients in the regression model present significant weight in the polynomial equation. Thus, three second order polynomial equations (Equations (6)–(8)) were fitted to the experimental data, as functions of the three independent process variables (j, An and σ).  
formula
(6)
Table 4

ANOVA results for the %ETC and %EFC as a function of j, An and σ

Sum of squaresDegrees of freedomMean squareF-ratiop-value
Factor (%ETC) 
A:j 882 882 145.84 0.0001 
B:σ 102.961 102.961 17.02 0.0091 
C:An 5.61125 5.61125 0.93 0.3797 
AA 0.023141 0.023141 0.9531 
AB 49 49 8.1 0.036 
AC 16 16 2.65 0.1648 
BB 282.423 282.423 46.7 0.001 
BC 0.1225 0.1225 0.02 0.8924 
CC 5.73083 5.73083 0.95 0.375 
Error total 30.2392 6.04783   
Total (corr.) 1371.92 14    
R2 = 97.8%= 93.8% 
Factor (%EFC) 
A:j 856.98 856.98 67.97 0.0004 
B:σ 177.661 177.661 14.09 0.0132 
C:An 0.03125 0.03125 0.9622 
AA 81.7078 81.7078 6.48 0.0515 
AB 24.01 24.01 1.9 0.2261 
AC 2.25 2.25 0.18 0.6903 
BB 114.931 114.931 9.12 0.0294 
BC 0.0625 0.0625 0.9466 
CC 13.0385 13.0385 1.03 0.3558 
Error total 63.0392 12.6078   
Total (corr.) 1311.74 14    
R2 = 95.19%= 86.54% 
Factor (OC)       
A:j 2.46642 2.4664 1435.52 0.0000 
B:σ 0.03726 0.0373 21.69 0.0055 
C:An 0.00002 0.0000 0.01 0.9096 
AA 0.05905 0.0590 34.37 0.002 
AB 0.00526 0.0053 3.06 0.1407 
BB 0.00213 0.00213 1.24 0.3157 
BC 0.00004 0.00004 0.02 0.8815 
CC 0.00094 0.00094 0.55 0.4927 
Error total 0.00859 0.00172   
Total (corr.) 2.58116 14    
R2 = 99.75%= 98.90% 
Sum of squaresDegrees of freedomMean squareF-ratiop-value
Factor (%ETC) 
A:j 882 882 145.84 0.0001 
B:σ 102.961 102.961 17.02 0.0091 
C:An 5.61125 5.61125 0.93 0.3797 
AA 0.023141 0.023141 0.9531 
AB 49 49 8.1 0.036 
AC 16 16 2.65 0.1648 
BB 282.423 282.423 46.7 0.001 
BC 0.1225 0.1225 0.02 0.8924 
CC 5.73083 5.73083 0.95 0.375 
Error total 30.2392 6.04783   
Total (corr.) 1371.92 14    
R2 = 97.8%= 93.8% 
Factor (%EFC) 
A:j 856.98 856.98 67.97 0.0004 
B:σ 177.661 177.661 14.09 0.0132 
C:An 0.03125 0.03125 0.9622 
AA 81.7078 81.7078 6.48 0.0515 
AB 24.01 24.01 1.9 0.2261 
AC 2.25 2.25 0.18 0.6903 
BB 114.931 114.931 9.12 0.0294 
BC 0.0625 0.0625 0.9466 
CC 13.0385 13.0385 1.03 0.3558 
Error total 63.0392 12.6078   
Total (corr.) 1311.74 14    
R2 = 95.19%= 86.54% 
Factor (OC)       
A:j 2.46642 2.4664 1435.52 0.0000 
B:σ 0.03726 0.0373 21.69 0.0055 
C:An 0.00002 0.0000 0.01 0.9096 
AA 0.05905 0.0590 34.37 0.002 
AB 0.00526 0.0053 3.06 0.1407 
BB 0.00213 0.00213 1.24 0.3157 
BC 0.00004 0.00004 0.02 0.8815 
CC 0.00094 0.00094 0.55 0.4927 
Error total 0.00859 0.00172   
Total (corr.) 2.58116 14    
R2 = 99.75%= 98.90% 
 
formula
(7)
 
formula
(8)

The adjustment of developed models was evaluated based on the variation coefficients (R2 and ). The is a more suitable parameter, because it takes into account the size of the data set. Both coefficients were very close to unity, indicating that RSM was very suitable to describe the electro-disinfection process using BDD anode containing system. According to Table 4, the following values of were found: 93.8, 86.54% and 98.90% for %ETC, %EFC and OC, respectively. Indeed, good agreements between the experimental and predicted values were observed (see Table 3). Next, the response surface plots were drafted with the regression models (Equations (4) and (5)) to see the simultaneous effects of two factors on a response variable (Figure 1). Thus, it was possible to determine the optimal operating conditions as follows.

Figure 1

Response surface plots for the interactive effect of j and σ on %EFC and %ETC. Reaction time = 12 min, T = 25 °C. (a) %ETC with BDD/Ti; (b) %ETC with GP; (c) %EFC with BDD/Ti; (d) %EFC with GP.

Figure 1

Response surface plots for the interactive effect of j and σ on %EFC and %ETC. Reaction time = 12 min, T = 25 °C. (a) %ETC with BDD/Ti; (b) %ETC with GP; (c) %EFC with BDD/Ti; (d) %EFC with GP.

Interactive effect of current density (j) and conductivity (σ) on %ETC and %EFC

The obtained surfaces indicate that, for both BDD (An: 1) and GP (An: 3), an increase in j resulted in an increase in the disinfection efficiency. It is in agreement with the results previously reported by Qi et al. (2015) for disinfection of solution containing 107 UFC/mL of E. coli (using a BDD anode and 0.05 M of Na2SO4). They found that the increase in current density from 15 to 20 mA/cm achieved almost complete microbial elimination, reducing the reaction time from ca. 30 min to 20 min. It occurs due to the more intensive production of oxidizing species, mainly the OH, responsible for coliform membrane disruption. Figure 1 shows that j presents similar influence on disinfection efficiency independent of types of anode material. Jeong et al. (2006) and Gómez-López et al. (2013) also have found that, in the disinfection of water containing E. coli, high j values are beneficial for fast microbial inactivation. Nevertheless, it is less efficient from the point of view of EC. Moreover, σ presents the highest positive effect on the process efficiency at its minimal value (500 μS/cm). According to the literature (Flox et al. 2006), it can be caused by an increase in Na2SO4 concentration promoting the formation of . However, it has any significant effect on the electrochemical oxidation process due to its lower oxidation potential, comparing to that of OH radical. Additionally, if Na2SO4 is used at high concentrations, it can act as culture media for bacterial growth (Muralitharan et al. 1991), being even self-defeating for the elimination process. The other studies have also reported a decrease in the disinfection efficiency under increasing Na2SO4 concentration (Qi et al. 2015). It implies that electro-generated plays a role in combination with the OH radical decreasing the disinfection efficiency.

Effect of electrode material (An) on %ETC and %EFC

As it can be seen, for the elimination of total coliforms, BDD/Ti anode gives better results than GP one (for the reaction time of 12 min, Figure 1). For σ value of 500 μS/cm, it was possible to achieve near 100%ETC and %EFC, with 20 mA/cm2, using both BDD/Ti and GP anode. However, high current combined with low conductivity implies the usage of high voltage and consequently higher energy cost. When j = 10 mA/cm2, the %ETC reaches 96 and 92% using BDD/Ti and GP, respectively. Similar behavior was observed for j = 2 mA/cm2, with efficiencies of 92 and 84%ETC using BDD/Ti and GP anodes. These results imply that BDD/Ti anode produces larger amount of the OH radical than GP one. The inherent cytotoxicity of GP would compensate the lower OH production comparing to that of BDD/Ti anode. Therefore similar efficiencies are obtained with both materials, as it was shown by the ANOVA results.

Interactive effect of current density (j) and conductivity (σ) on OC

Figure 2 presents the response surface plots for OC, using: (a) BDD/Ti and (b) GP electrodes. The σ seems to present slightly ‘positive’ effect on the OC, since there is a small decrease in the cost at higher conductivities values. This is due to the lower electricity resistance involved in the use of lower voltages. However, high conductivity may significantly reduce the process efficiency. From this reason, 500 mS/cm was selected as the optimal condition to carry out the process. On the other hand, the OC rapidly increased with an increase in j from 2 to 20 mA/cm2 (Figure 2). For example, when j = 10 mA/cm2, the operational cost accounts for ca. 0.45 USD/m3, which is a reasonable value for this type of processes. Additionally, at these conditions (Figure 1), coliform elimination efficiencies reaches ca. 90% after 12 min of treatment. Therefore, σ = 500 μS/cm and j = 10 mA/cm2 were selected as the most appropriate conditions for water treatment.

Figure 2

Response surface plots for the interactive effect of j and σ on OC. Reaction time = 12 min, temperature = 25 °C. (a) OC with BDD/Ti; (b) OC with GP.

Figure 2

Response surface plots for the interactive effect of j and σ on OC. Reaction time = 12 min, temperature = 25 °C. (a) OC with BDD/Ti; (b) OC with GP.

Evolution of the %ETC and %EFC of the wastewater at selected conditions

At selected conditions, the evolution of the electrochemical disinfection process was monitored as function of time (Figure 3). As it can be seen, in the elimination of both FC and TC, BDD/Ti electrode presents slightly better performance than GP one. Using GP electrode, it is possible to achieve complete elimination of coliforms after ca. 27 min of reaction. The application of BDD/Ti one requires ca. 20 min. Notice that at the beginning of the disinfection process, GP based treatment is faster than the BDD/Ti one. However, after the reaction have progressed a few minutes (ca. 5 minutes), BDD/Ti reaches higher degree of coliform elimination. It can be related to a two-step mechanism of coliform elimination using GP electrode as follows: (i) the first one caused by the cytotoxicity of GP, which have been reported to inactivate ca. 26% of E. coli (Liu et al. 2011) and (ii) the second one caused by the formation of oxidizing species (Polcaro et al. 2007). On the other hand, BDD/Ti has not been reported to present inherent cytotoxicity, even if it is more efficient in the production of OH radicals than GP (Brillas & Martínez-Huitle 2015).

Figure 3

Evolution of FC, TC and OC with GP and BDD electrodes at j = 10 mA/cm2, σ = 500 μS/cm, temperature = 25 °C.

Figure 3

Evolution of FC, TC and OC with GP and BDD electrodes at j = 10 mA/cm2, σ = 500 μS/cm, temperature = 25 °C.

Once reaction was finished, the microorganism evolution was monitored analyzing samples after 24 h, 94 h and 7 days. Any coliforms growth was observed for water treated using both BDD/Ti and GP electrode, indicating that the cell damage was enough to the complete elimination of coliforms. Thus, from the experimental results presented in Figure 3, the following observations can be stated for GP and BDD/Ti electrodes, respectively: (i) the 13% and 25% of reduction in TOC; (ii) the decrease in nitrites concentration from 0.183 mg/L to 0.096 and to 0.038 mg/L; (iii) the increase in nitrates concentration from 0.87 mg/L to 1.23 and to 1.44 mg/L; (iv) the increase in OC with an increase in electrolysis time, for both electrodes. Notice that lower OC, using GP electrode, can be related to its higher electrical conductivity compared with that of the BDD film; (v) the required time for total removal of TC and FC was of ca. 30 min for GP electrode and 20 min for BDD/Ti one.

CONCLUSIONS

Electro-oxidation using BDD/Ti or GP anodes and Na2S2O3 as supporting electrolyte let to achieved complete elimination of total and fecal coliform of raw surface water (without any additional chemical substances). This method can be considered as a suitable chlorine-free disinfection treatment. The statistical analysis of the experimental data indicated that the operational factors presenting the highest influence on coliform elimination efficiency are conductivity and current density. From the RSM, the following optimal operational conditions were established: j = 10 mA/cm2, 500 μS/cm and temperature = 25 °C. The variation of FC and TC with time indicated that BDD/Ti electrode presents slightly higher efficiencies that GP one. However, the GP electrode cost is only a fraction of the BDD/Ti one. Thus, GP electrode become a very promising material for the electrochemical disinfection of raw surface water. The preliminary operational cost analysis for water disinfection treatment using an electro-oxidation system led to a total unitary cost, aiming to achieve complete disinfection, of 1.1 USD/m3 for GP anode and 1 USD/m3 for BDD anode.

ACKNOWLEDGEMENTS

The authors acknowledge ‘Dirección de Investigación de la Universidad EAFIT, Medellín, Colombia’ for financial support of this research. The staff of the ‘Laboratorio de Ingeniería de Procesos’ is also recognized for their participation.

REFERENCES

REFERENCES
Barashkov
,
N. N.
,
Eisenberg
,
D.
,
Eisenberg
,
S.
,
Shegebaeva
,
G. Sh.
,
Irgibaeva
,
I. S.
&
Barashkova
,
I. I.
2009
Electrochemical chlorine free AC disinfection of water contaminated with Salmonella typhimurium bacteria
.
Russian Journal of Electrochemistry
46
(
3
),
306
311
.
Brillas
,
E.
&
Martínez-Huitle
,
C.
2015
Decontamination of waste waters containing synthetic organic dyes by electrochemical methods. An updated review
.
Applied Catalysis B: Environmental
166–167
,
603
643
.
Castro-Hermida
,
J. A.
,
García-Presedo
,
I.
,
Almeida
,
A.
,
González-Warleta
,
M.
,
Correia
,
J. M.
&
Mezo
,
M.
2008
Presence of Cryptosporidium spp. and Giardia duodenalis through drinking water
.
Science of the Total Environment
405
,
45
53
.
Flox
,
C.
,
Cabot
,
P. L.
,
Centellas
,
F.
,
Garrido
,
J. A.
,
Rodríguez
,
R. M.
,
Arias
,
C.
&
Brillas
,
E.
2006
Electrochemical combustion of herbicide mecoprop in aqueous medium using a flow reactor with a boron-doped diamond anode
.
Chemosphere
64
,
892
902
.
GilPavas
,
E.
,
Acevedo
,
J.
,
López
,
L. F.
,
Dobrosz-Gómez
,
I.
&
Gómez-García
,
M. A.
2014
Solar and artificial UV inactivation of bacterial microbes by Ca-alginate immobilized TiO2 assisted by H2O2 using fluidized bed photoreactors
.
Journal of Advanced Oxidation Technologies
17
(
2
),
343
351
.
Hunter
,
P. R.
,
Zmirou-Navier
,
D.
&
Hartemann
,
P.
2009
Estimating the impact on health of poor reliability of drinking water interventions in developing countries
.
Science of the Total Environment
407
,
2621
2624
.
Hussain
,
S. N.
,
Heras
,
N.
,
Asghar
,
H. M. A.
,
Brown
,
N. W.
&
Roberts
,
E. P. L.
2014
Disinfection of water by adsorption combined with electrochemical treatment
.
Water Research
54
,
170
178
.
Jeong
,
J.
,
Kim
,
J.
&
Yoon
,
J.
2006
The role of reactive oxygen species in the electrochemical inactivation of microorganisms
.
Environmental Science & Technology
40
,
6117
6122
.
Kerwick
,
M. I.
,
Reddy
,
S. M.
,
Chamberlain
,
A. H. L.
&
Holt
,
D. M.
2005
Electrochemical disinfection an environmentally acceptable method of drinking water disinfection
?
Electrochimica Acta
50
,
5270
5277
.
Kostyla
,
C.
,
Bain
,
R.
,
Cronk
,
R.
&
Bartram
,
J.
2015
Seasonal variation of fecal contamination in drinking water sources in developing countries: a systematic review
.
Science of the Total Environment
514
,
333
343
.
Liu
,
S.
,
Helen
,
T.
,
Hofmann
,
M.
,
Burcombe
,
E.
,
Wei
,
J.
,
Jiang
,
R.
,
Kong
,
J.
&
Chen
,
Y.
2011
Antibacterial activity of graphite, graphite oxide, graphene oxide and reduced graphene oxide: membrane and oxidative stress
.
ACS Nano
5
(
9
),
6971
6980
.
Martínez-Huitle
,
C. A.
2007
Conductive diamond electrodes for water purification
.
Journal of Material Research
10
(
4
),
419
424
.
Mascia
,
M.
,
Vacca
,
A.
&
Palmas
,
S.
2012
Fixed bed reactors with three dimensional electrodes for electrochemical treatment of waters for disinfection
.
Chemical Engineering Journal
211–212
,
479
487
.
Muralitharan
,
M. S.
,
Chandler
,
S. F.
&
Van Steveninck
,
R. F. M.
1991
Effects of Na2SO4. K2SO4 and KCl on growth and ion uptake of callus cultures of Vaccinium corymbosum L. cv. Blue Crop
.
Annals of Botany-Oxford Journals
69
(
5
),
459
465
.
Polcaro
,
A. M.
,
Vacca
,
A.
,
Mascia
,
M.
,
Palmas
,
S.
,
Pompei
,
R.
&
Laconi
,
S.
2007
Characterization of a stirred tank electrochemical cell for water disinfection processes
.
Electrochimica Acta
52
(
7
),
2595
2602
.