Evaluation of the toxicity of textile effluent treated by electrocoagulation

The complex degradation of textile processing effluents is worrying in relation to this type of effluent's characteristics. These effluents usually present high chemical oxygen demand (COD) (Ribeiro et al. 2014) and dye concentrations, as well as other, highly stable and toxic compounds (Azzaz et al. 2018), which affect the efficiency of conventional treatment, such as biological and chemical coagulation, adversely. Most dyes and their byproducts have carcinogenic potential (Ribeiro et al. 2014) that can cause severe risks to aquatic biota and human health. Therefore, treating these effluents effectively before discharge to water bodies is essential.

Several technologies have been studied for treating textile effluents, including ozonation (Wijannarong et al. 2013), sonication (Rayaroth et al. 2015), photocatalysis (Rosa et al. 2015), membrane filtration (Dasgupta et al. 2015), and electrocoagulation (EC) (Ribeiro et al. 2014).

EC is promising in treating various industrial wastewaters (Phalakornkule et al. 2010; Oliveira et al. 2020). In EC, electrode dissolution of the anode yields hydrolysis products, such as metal hydroxides, that are effective in destabilizing contaminants and/or forming particles of reduced solubility that trap pollutants (Behbahani et al. 2011). With the generation of coagulating agents in situ, the addition of chemicals and the production of excess sludge can be avoided. Mollah et al. (2001) reported that EC requires simple and easy-to-operate equipment, which produces an effluent with a lower concentration of total dissolved solids (TDS) than chemical treatments.

Most studies have reported using direct current (DC) in EC. However, DC can cause oxidative corrosion at the anode, as well as the formation of an oxide layer at the cathode, reducing cathode/anode current flow, and hence treatment efficiency (Khandegar & Saroha 2013). Electrode passivation also increases electrical energy consumption. The pulsed current (PC) mode has been studied in electrochemical methods in water/wastewater treatment (Lu et al. 2015). In addition to avoiding electrode passivation, PC has the advantage of lower energy consumption than DC (Rocha et al. 2018; Oliveira et al. 2020).

While most research involving textile effluent treatment does not address the issue of toxicity because, in some situations, even treated effluent can present high toxicity, some studies have included the use of seeds for effluent toxicity tests. Lettuce seeds (Lactuca sativa) in effluent, soil, or sediment are used for toxicity tests, due to the low energy reserve required for germination and their ensuing rapid growth (Dutka 1989). Tests using L. sativa evaluate the samples' toxic effects on germination and elongation of the root, which can show sensitivity to different compounds at different levels. In this paper, the aim was to assess color and turbidity removal, and the toxicity of the textile effluent after EC treatment by PC.

The electrolytic cell configuration was set up according to Martins et al. (2017). The cylindrical EC reactor (60.0 cm height and 10.0 cm I.D) was constructed in acrylic, with four parallel plate electrodes (5.0 × 40.0 × 0.3 cm) of 304 stainless steel connected by a bipolar array (total surface area 1,200 cm²). A magnetic stirrer was used to mix liquids in the electrolytic medium (Tecnal TE-0851, Piracicaba/Brazil).

Aiming to obtain the optimum operating conditions for the process, experimental planning was carried out using statistical and mathematical resources:

• 1: Select factors (stirring rate, pulse frequency, and inter-electrode distance) and levels for the construction of the Box–Behnken design (BBD);

• 2: Obtain models from the BBD experimental results and statistical analysis;

• 3: Construct response surface graphics, demonstrating the optimal operating conditions for removing color and turbidity.

The validity of the assumption of normality of the residuals was evaluated using least square estimators and regression analysis was also based on which errors followed a normal distribution. The experimental data were processed using R software (R Development Core Team 2014).

Before color and turbidity analysis, the samples were left for 1 h to allow floc particles (suspended solids) to settle. Turbidity analysis was performed using a portable turbidimeter (Hach, 2100P). Color analysis was performed on the supernatants after separating the suspended solids by centrifugation at 4,000 rpm for 5 min. Absorbance was measured by spectrophotometry (Shimadzu, UV-1800) at 400 nm and the UV–Vis spectral for color removal evaluation. Sample pH was measured immediately after sampling using a pH meter (Tecnal Tec 5). A thermometer 20 cm from the reactor top measured the temperature in situ.

The tests were conducted using lettuce seeds (L. sativa), which were exposed to the different concentrations (1, 10, 25, 75, and 100% v/v) of the textile effluent before and after EC, within the optimum operating conditions (Martins et al. 2017), to evaluate both lethal effects through non-germination and sublethal impacts through the development of radicles. The seeds were seeded using Petri dishes (∅ = 9.5 cm) and duplicated for each concentration tested. Each plate was prepared by placing a filter paper in the base, to which 4 mL of effluent sample was added, and, with the aid of forceps, 20 seeds were disposed of equidistantly.

After this, the plates were closed to prevent moisture loss, covered in dark paper for protection against light, and conditioned at 22 ± 2 °C for 120 h (EPA 1989). For each test, distilled water was used as a control, considering acceptability criteria: seed germination ≥90% and root length variability ≤30%. Sensitivity tests were performed with sodium chloride (NaCl), at 9.21 g–NaCl L⁻¹ concentration, representing the positive control. Dilutions were also made to give 10, 25, and 50% (v/v) concentrations, following the same procedure as the sample tests.

ARS and ARC are the average radicular lengths in the sample and negative control, respectively, and AGS and AGC are the average numbers of germinated seeds in the sample and negative control, respectively. RGI values were classified into three categories according to the toxicity effects observed (Young et al. 2012):

• (a)

  Inhibition of root elongation (I): 0 < x < 0.8

• (b)

  No significant effects (NSE): 0.8 ≤ x ≥ 1.2

• (c)

  Stimulation of root elongation (S): x > 1.2

where x is the value obtained for RGI.

After total sedimentation, the residue (scum and sludge) generated by EC under optimum operating conditions (Martins et al. 2017) was collected and conditioned in a drying oven at 105 °C until a constant weight was attained. The dry mass was subjected to energy-dispersive X-ray fluorescence (XFR) analysis (Rigaku-ZSX Mini II) for elemental identification.

In addition to color and turbidity removals, EEC must be quantified, as it will indicate the cost of treatment. EEC was calculated for the experiments, Table 2, and the values obtained were between 2.29 and 3.33 kWh m⁻³.

Table 3 shows the mathematical model coefficients estimated by the software, the coefficients' standard deviations, the respective t-Student values, and the significance of each regression coefficient through the ‘p’ value. Statistically significant coefficients, considering a 95% confidence interval (p < 0.05), are highlighted in bold.

Table 3

Coefficients of linear regression, coefficient standard deviations, t-values, and probability of model coefficients' statistical significance (p) regarding color and turbidity removal

Response .	Factor .	Coefficient .	Standard deviation .	t-value .	p-Value .
Color	Intercept	98.62250	0.077956	1,265.1104	<2.2 × 10−16
	X1	0.0325	0.055123	0.5896	0.576976
	X2	− 0.16375	0.055123	− 2.9706	0.024937
	X3	− 0.38125	0.055123	− 6.9164	0.000452
	X1:X2	−0.08	0.077956	−1.0262	0.344367
	X1:X3	− 0.2025	0.077956	− 2.5976	0.040792
	X2:X3	0.0175	0.077956	0.2245	0.829828
		− 0.22625	0.077956	− 2.9023	0.027253
		0.04875	0.077956	0.6254	0.554762
		−0.01125	0.077956	−0.1443	0.889979
Turbidity	Intercept	83.50875	0.43149	193.5359	1,284 × 10−12
	X1	1.80250	0.30511	5.9077	0.0010461
	X2	− 2.13062	0.30511	− 6.9832	0.0004291
	X3	− 1.83188	0.30511	− 6.0040	0.0009612
	X1:X2	1.48750	0.43149	3.4474	0.013677
	X1:X3	−0.22000	0.43149	−0.5099	0.6283545
	X2:X3	0.42125	0.43149	0.9763	0.3666344
		− 3.0325	0.43149	− 7.0280	0.0004144
		0.18625	0.43149	0.4316	0.6810792
		0.48875	0.43149	1.1327	0.3005640

Response .	Factor .	Coefficient .	Standard deviation .	t-value .	p-Value .
Color	Intercept	98.62250	0.077956	1,265.1104	<2.2 × 10−16
	X1	0.0325	0.055123	0.5896	0.576976
	X2	− 0.16375	0.055123	− 2.9706	0.024937
	X3	− 0.38125	0.055123	− 6.9164	0.000452
	X1:X2	−0.08	0.077956	−1.0262	0.344367
	X1:X3	− 0.2025	0.077956	− 2.5976	0.040792
	X2:X3	0.0175	0.077956	0.2245	0.829828
		− 0.22625	0.077956	− 2.9023	0.027253
		0.04875	0.077956	0.6254	0.554762
		−0.01125	0.077956	−0.1443	0.889979
Turbidity	Intercept	83.50875	0.43149	193.5359	1,284 × 10−12
	X1	1.80250	0.30511	5.9077	0.0010461
	X2	− 2.13062	0.30511	− 6.9832	0.0004291
	X3	− 1.83188	0.30511	− 6.0040	0.0009612
	X1:X2	1.48750	0.43149	3.4474	0.013677
	X1:X3	−0.22000	0.43149	−0.5099	0.6283545
	X2:X3	0.42125	0.43149	0.9763	0.3666344
		− 3.0325	0.43149	− 7.0280	0.0004144
		0.18625	0.43149	0.4316	0.6810792
		0.48875	0.43149	1.1327	0.3005640

View Large

Positive coefficient values indicate an increase in the response value when the variable moves toward its maximum studied level. At the same time, negative values indicate an effect of increased response when the variable moves toward its minimum level. With respect to interactions, positive values indicate that the response will increase if the two variables are moving toward the same level, lower or higher, and negative values indicate an increase in the response if the variables are moving in opposite directions, that is, one variable moves toward the upper level and the other toward the lower level.

As seen in Table 3, the three independent variables were significant in some way (in linear, quadratic, or interaction terms) for the removal of color and turbidity. However, for color removal, the inter-electrode distance in the linear term (X₃) stood out compared to the other variables (‘p’ values = 4.52 × 10⁻⁴), influencing the removal of both parameters inversely (the higher the level, the lower the removal efficiency).

Regarding turbidity, the frequency of the pulses (X₂) and the inter-electrode distance (X₃) were highlighted, the linear term with ‘p’ values equal to 4.291 and 9.612 × 10⁻⁴, respectively, and the agitation speed in the quadratic term (⁠⁠), with a ‘p’ value equal to (4.144 × 10⁻⁴). The coefficients in these three cases were negative, indicating that turbidity removal will tend to decrease when the X₂, X₃, and values increase.

Removals (%) tend to increase when the pulse frequency is reduced to the lowest level studied (600 Hz), the stirring rates were between 150 and 350 rpm, 100 and 320 rpm, for color and turbidity, respectively.

Table 4 shows the optimal values of the operating variables and the removal proportions calculated from the response surface methodology for color and turbidity.

The optimization study showed that for color and turbidity removal:

• (a)

  The optimum stirring rate was close to the midpoint studied (200 rpm), indicating that, at lower stirring rates, ion mobility within the reactor may not favor floc formation. At the highest stirring rates, however, flocs can collide with one another due to the high turbulence (Modirshahla et al. 2008), being able to provide larger numbers of smaller suspended and dissolved particles in the medium, increasing turbidity and color levels.

• (b)

  The optimum inter-electrode distance was closer to the lowest level studied (1 mm); similar results were also found by Modirshahla et al. (2007). Low inter-electrode distance might make electron flow easier (as shown by the higher current densities, Table 2), increasing the amount of coagulating agent.

• (c)

  The optimum pulse frequencies were around 1,000 Hz (between the lowest and midpoint levels). The exact frequency was found by Lu et al. (2015) and Shu et al. (2016) when studying sulfide removal in sewage and manganese and ammoniacal-nitrogen removals in wastewater, respectively. Xu et al. (2017) found 5,000 Hz for zinc and manganese removal efficiencies in smelting wastewater. The pulse frequency affects the ion migration rate and reduces the electrode's concentration polarization. Still, the mass transfer may be limited by higher pulse frequencies (Shu et al. 2016).

Following optimization, a sample of raw water was submitted to EC treatment using a 200 rpm stirring rate, 1,000 Hz frequency, and 1 mm inter-electrode distance. Process efficiency was evaluated using parameters including pH, conductivity, turbidity, COD, color, total suspended solids (TSS), sulfate, chloride, etc. (Martins et al. 2017).

Removal was satisfactory for COD, turbidity, color and sulfate, reaching 81, 86, 99, and 99%, respectively. The process reached 46% removal of TSS (Martins et al. 2017). It is known that EC is not efficient for chloride removal.

Treated effluent obtained under optimal conditions was used in the toxicity tests.

The results proved the seeds' sensitivity to NaCl, with a concentration-dependent toxicity response in which both seed RGI and GI (%) decreased with increasing NaCl concentration, causing radicle growth inhibition at a concentration of only 10% (v/v). When 100% (v/v) NaCl concentration was tested, complete inhibition of seed germination occurred.

Table 6 and Figure 13 show that as the effluent concentrations increased, both RGI and GI (%) decreased. The full concentration of the treated effluent also reached a low GI (0.72%). Both raw and treated effluent inhibited radicle growth from a concentration of about 50% (v/v).

Due to the textile effluent's complexity, i.e., the variety of chemicals used, including dyes from different grades, synthetic gums, salts, surfactants, etc., it is difficult to identify the leading cause of toxicity. It is believed that one cause of the treated effluent's remaining toxicity is the NaCl it contains, as neither the chloride concentration nor the electrical conductivity decreased during treatment (Table 4).

In the seed sensitivity test, which considered the approximate concentration of NaCl in the raw effluent, positive effects of L. sativa on germination and root growth inhibition were presented against this salt, and similar results were found by Young et al. (2012).

Although the total nitrogen concentration decreased after EC (Table 2), toxic nitrogen species may have been formed from the oxidative process, as shown by the decrease in oxidation–reduction potential (ORP) (Table 4). This phenomenon is also observed by Palácio et al. (2009) who used iron electrodes in the EC treatment of textile dye effluent.

The residue generated during EC was weighed and a total mass of 4.41 g was reported. As the treated effluent volume was 2.60 L, EC treatment generated approximately 1.7 kg of solid residue/m³ treated effluent, which was subjected to energy-dispersive XFR analysis – Table 7.

Iron, chromium, nickel, manganese, and sulfur were expected in the residue because they were components of the 304 stainless steel electrodes used. The three main electrode components (Fe, Cr, and Ni) were also those found in the greatest proportions in the residue mass. The presence of chlorine in the residue is also relevant because of the high concentration of chloride found in the effluent.

EC by PC using stainless steel electrodes was investigated in relation to textile wastewater in this study, leading to four conclusions:

• The process's color and turbidity removal efficiencies were 98–99% and 74–85%, respectively.

• Statistical modeling and response surface data defined the optimum values for stirring rate, inter-electrode distance, and pulse frequency. The optimum stirring rate was close to 200 rpm. The best inter-electrode space was 1 mm and the optimum pulse frequency was 1,000 Hz.

• Based on the optimal independent variable values, toxicity tests with L. sativa seeds were a simple and efficient tool to evaluate the quality of EC-treated effluent. These tests with L. sativa showed that the EC-treated effluent was still highly toxic, probably due to the high NaCl concentration remaining. This indicates that EC could form part of an effluent treatment system, but it is not recommended that it be used alone.

• The solid residue generated had significant contents of Fe, Cr, Ni, and Cl.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.