The Electrooxidation of Synthetic Bipyridyl Herbicide Wastewaters with Boron-Doped Diamond Electrodes: A Technical and Economic Study to Boost Their Application for Pollution Prevention in the Agricultural Sector

Abstract

Boron-doped diamond electrodes (BDDEs) offer a highly efficient pathway to mineralize recalcitrant compounds due to their reduced energy requirements, fewer chemical inputs, and mechanical stability. In this work, the electrochemical degradation of paraquat (PQ) and diquat (DQ) was studied using an undivided cell (Condiacell^®-type) at circumneutral pH, and under galvanostatic control. The roles of applied current density, volumetric flow rate, and herbicide concentration were systematically studied through a central composite design (CCD) using a closed-flow reaction setup. Under the best operating conditions (i.e., for PQ: 1.6 mA/cm², 80 mL/min, and 70 mL/min, and 70 mg/L; and for DQ: 1.5 mA/cm², 80 mL/min, and 73 mg/L), a spectrophotometric analysis evidenced that the herbicides were satisfactorily removed (ca. 100%) while mineralization degrees were above 90%. Furthermore, the produced effluents yielded significant increases in seed germination and root length, which suggest a reduction in toxicity. Energy consumptions of 0.13 and 0.18 kWh/g of TOC are reported with the electrochemical cells for the PQ and DQ treatments, respectively. The PQ and DQ treatments by electrooxidation are estimated to emit nearly 2.7 and 38.9 kg CO₂/m³ of water treated, with a cost around USD 250/m³. Carbon emissions could be greatly decreased for PQ (0.28 kg CO₂/m³) and DQ (0.40 kg CO₂/m³) if electricity were generated from renewable resources. Although this study suggests that the use of BDDE can be considered as a green alternative for agrochemical removal due to lower carbon emissions, the environmental profile of the process is determined by the degree of renewability of the electrical grid of each country or region.

1. Introduction

In recent years, the high demand for agricultural products has led to the excessive use of herbicides to protect crops and improve their quality, thereby increasing global food production. However, these compounds can reach aquatic ecosystems through surface runoff, leaching, fumigation, deposition, and soil erosion [1,2]. Notable herbicides include paraquat and diquat, both of which have a quaternary ammonium group in their structures [3] responsible for their herbicidal and toxicological properties [4]. Paraquat (1,1′-dimethyl-4,4′-bipyridylium dichloride; PQ,

Table 1. Chemical structures and absorption maxima of paraquat dichloride and diquat dibromide.

Name	Chemical Structure	λmax (nm)
Diquat C 12H12Br2N2, 6,7-dihydrodipyrido [1,2-a:2′,1′-c] pyrazinedium dibromide		257
Paraquat C12H14Cl2N2, 1,1′-dimethyl-4,4′-bipyridylium dichloride		309

) is one of the most widely used herbicides due to its high efficiency, low cost, and non-selectivity in controlling weed and grass growth [5]. Diquat (6,7-dihydrodipyrido [1,2-a:2′,1′-c] pyrazinedium dibromide, DQ, Table 1), is a desiccant, defoliant, preharvest, and non-selective herbicide belonging to the bipyridinium class [6].

PQ and DQ herbicides are considered persistent organic pollutants and highly mobile. Due to their high solubility in water (PQ = 620 g/L, DQ = 700 g/L), they become dispersed in the environment and can consequently be found in water bodies (canals, groundwater, rivers, and lakes) as well as in food sources [2,7,8].

Due to their high environmental and human toxicity, these herbicides have been banned in over 50 countries, including China, the United States, and throughout the European Union [9]. In Colombia, paraquat is banned for aerial application, and in Uruguay, it is the only herbicide requiring a professional prescription. Paraquat was reevaluated in Brazil, leading to a phase restriction, culminating in a permanent ban in 2020 [10]. The toxicity of paraquat in humans has been documented across various organs, including the liver, brain, kidneys, heart, adrenal glands, muscles, and most severely, the lungs, where it induces respiratory failure by transforming oxygen into free radicals, possibly resulting in death [11]. Diquat has also been shown to cause damage to the heart, central nervous system, kidneys, liver, and lungs [12].

The high toxicity of PQ and DQ, along with their low biodegradability, has necessitated the development of advanced oxidation processes (AOPs) such as Fenton processes [13], solar photo-Fenton [14], solar corrosion Fenton [15] solar-Fenton pilot lagoon [16], Fenton, electro-Fenton, photoelectro-oxidation, and photoelectro-Fenton [9], as well as solar-assisted heterogeneous photo-Fenton-like [17] ZnO/sunlight [18], and TiO₂/UV light [19], along with electrochemical processes [20,21,22], to mitigate unwanted effects on humans and ecosystems.

AOPs like Fenton rely on ferrous iron-derived compounds to catalyze the activation of oxidants for the generation of reactive oxygen species. Nevertheless, the application potential of homogeneous catalysts is limited by some disadvantages, like a limited pH range, particle agglomeration, sludge production, and low mass transfer of reactive species. Nonetheless, heterogeneous catalysts have been developed to address these issues [23].

Within electrochemical AOP, anodic oxidation is an electrochemical oxidation process (EAOP) known to generate high amounts of hydroxyl radicals, which are highly oxidizing and capable of achieving the complete mineralization of contaminants. Anodic oxidation is one of the most effective methods due to its satisfactory performance in destroying persistent compounds. The advantages that distinguish these processes include the sole requirement of electrical energy, the absence of chemical reagents, minimal production of secondary waste, limited or no need for catalysts, and simple equipment that can be scaled up for industrial use [2,23,24]. This makes them suitable for water treatment due to their compatibility, efficiency, and environmental safety. Boron-doped diamond (BDD) electrode technology is an environmentally friendly EAOP, and has become particularly attractive for industrial-scale wastewater treatment, as it has demonstrated high efficiency—up to 100%—in fully mineralizing synthetic and real effluents regarding the degradation of various contaminants. This effectiveness is primarily due to its long service life, and its high chemical and electrochemical stability compensates for the initial investment [2,25,26].

At BDD, the direct electrooxidation (EO) of organic compounds (R) occurs according to Reaction 1, while indirect oxidation is facilitated by highly oxidizing reactive species generated at the anode, such as hydroxyl radicals (^•OH) that are physisorbed on BDD (Reactions 2 and 3), along with other highly oxidizing reactive species, as illustrated in the following reactions [27]:

BDD + R_red → BDD + ze⁻ + R_ox

(1)

BDD + H₂O ↔ BDD (^•OH) + H⁺ + e⁻

(2)

R + BDD (^•OH) → BDD + xCO₂ + yH₂O + z (inorganic ions)

(3)

Anodic oxidation (AO) is a highly efficient process in which organic contaminants can be oxidized through the direct transfer of electrons to the anode surface, resulting in the formation of reactive species as oxidation intermediates, converting water to hydroxyl radicals (^•OH) in the absence of chemical byproducts. Most organic pollutants are non-selectively destroyed by ^•OH radicals until they are completely mineralized into CO₂, inorganic ions, and water [2,21,28].

The electrochemical degradation of contaminants with boron-doped diamond (BDD) electrodes presents several innovations compared to Fenton methods for herbicide degradation: BDD electrodes generate highly reactive hydroxyl radicals, just like Fenton methods, but allow for the more efficient oxidation of complex contaminants that are typically resistant to Fenton treatments, such as some herbicides and persistent organic compounds. Fenton methods, by requiring the use of hydrogen peroxide and iron salts, generate metal sludge and may be less safe. In contrast, BDD technology avoids these reagents and secondary residues [29,30].

BDD electrodes have been studied for the treatment of recalcitrant contaminants such as herbicides [22,25], pharmaceuticals [31], phenol and its derivatives [32], as well as dyes including various dyes [33], Direct Red 80 azo dye [34], perfluorooctanoic acid [35] and Rhodamine B dye [36]. Efficient DQ mineralization with BDDEs was achieved in our laboratories using both an undivided electrochemical cell (Condiacell^®-type, Condias GmbH, Itzehoe, Germany) and an H-type cell under acidic conditions (pH = 2) [22].

The primary objectives of this research were to evaluate the electrochemical degradation of PQ and DQ at circumneutral pH using a commercial BDDE film and to assess the phytotoxicity of raw and treated effluents on cucumber seeds (Cucumis sativus L.). We determined the influence of applied current density on total organic carbon (TOC) removal and the removal rates of PQ and DQ. Experiments designed using central composite design (CCD) and response surface methodology (RSM) identified the best operating conditions for TOC removal.

2. Materials and Methods

2.1. Reagents

Paraquat dichloride x-hydrate PESTANAL^® (standard, 99% pure, Sigma-Aldrich, St. Louis, MO, USA, Table 1), diquat dibromide monohydrate PESTANAL^® (standard, 99% pure, Sigma-Aldrich, St. Louis, MO, USA, Table 1), HgSO₄ (Baker Analyzed, A.C.S., J.T. Baker, Phillipsburg, NJ, USA, 99.5%), Ag₂SO₄ (Baker Analyzed, 99.5%), H₂SO₄ (Meyer 95–98%, A.C.S.) and H₂O₂ (Meyer 29–32%, A.C.S certified). The supporting electrolyte was 0.05 M Na₂SO₄ (Baker Analyzed, A.C.S., 99.5%). All experiments were carried out with deionized water.

2.2. Analytical Techniques

The mineralization of each herbicide was estimated from its TOC decrease. The samples were analyzed by a Shimadzu (TOC-V CPH) organic carbon analyzer, which uses the 680 °C combustion catalytic oxidation method and achieves the total combustion of samples by heating them to 680 °C in an oxygen-rich environment inside total carbon combustion tubes filled with a platinum catalyst (Shimadzu TOC-V CPH analyzer, Shimadzu Corp., Kyoto, Japan). Herbicide concentrations were monitored with a UV–Vis spectrophotometer (Cary 60, 10 mm quartz cuvettes; 4800 nm/min, Agilent Technologies, Tokyo, Japan). PQ and DQ removals were estimated based on height variations at the maximum absorbance (λ = 309 and 257 nm, respectively). Aqueous solutions of PQ and DQ at various concentrations (0–30 mg/L) yielded the linear calibration plots shown in Figure S1 (Supplementary Materials). Thereafter, samples were taken every 30 min during the entire degradation experiments (300 min) and diluted with 0.05 M Na₂SO₄ solution to reach suitable optical densities. The solution pH was measured with a Conductronic pH-120 potentiometer. COD measurements were made with an H₂O₂-modified method [22,37].

2.3. Experimental Setup

The experiments were conducted at room temperature in a Condiacell^® type cell with a 3-BDDE array (Figure 1), where one electrode was the anode (100 cm²) and the other two served as cathodes (200 cm²). The separation between electrodes was 5 mm. The current was controlled with an AMEL Potentiostat–Galvanostat model 2051 (AMEL, Italy). DiaChem^® BDDEs (with 1–10 μm conductive diamond layer, 500–8000 ppm boron concentration, and 0.1 Ω cm resistivity) were purchased from Condias GmbH (Itzehoe, Germany). Samples were taken for analysis every 30 min during the degradation time (300 min) and diluted as necessary with electrolyte solution to achieve adequate optical densities.

2.4. Experimental Design

Response surface methodology (RSM) and central composite design (CCD) were used to optimize the effect of herbicide concentration (mg/L), current density (mA/cm²), and volumetric flow rate (cm³/min) on the mineralization of PQ and DQ. Each mineralization percentage was measured by evaluating TOC decay after 300 min of reaction with respect to the initial TOC value. This ratio was selected as the response factor. The CCD star-type method [38,39] consisted of three series of experiments: (a) a 2^k factorial design (i.e., all possible combinations of codified values between +1 and –1); for k = 3 variables, this consisted of eight experiments; (b) axial or star points (+1.68 and −1.68), and 0 (values for rotatable central composite designs) for three variables (i.e., six experiments); and (c) six replicates of the central point (i.e., 0). Therefore, 20 experiments were performed in which the three variables were codified at six levels within the following ranges: herbicide concentration (9.5–110.5 mg/L), current density (0.159–1.841 mA/cm²) and volumetric flow (79.55–920.45 cm³/min). Such values were selected from our earlier results [22].

A Pareto chart plot was built to discern the most significant method variables, which were modeled doing the analysis of variance (ANOVA). The Pareto chart depicts the t-values of the effects for each of the model terms [40]. Model terms exceeding the Bonferroni limits were treated as the most significant terms contributing to the response. The model was selected based on the p values with non-significant lack of fit. The response surfaces were generated by the MINITAB 20 and DESIGN EXPERT 11 software to determine the best conditions for PQ and DQ degradation. The polynomial models were statistically confirmed using ANOVA based on the F test, and their quality of fit deduced from the corresponding determination coefficients, R².

2.5. Kinetic Experiments

The kinetic experiments were conducted under the best conditions obtained. Samples were collected at intervals from 30 to 300 min, and the concentrations of PQ, DQ, TOC and COD were monitored. The experiments were conducted in triplicate.

2.6. Phytotoxicity Assays

Phytotoxicity bioassays were conducted with the samples before and after the process using cucumber seeds (Cucumis sativus L.), as recommended by the Ecological Effects Guide [41]. Ten seeds were placed on a Whatman^® filter paper in 55 mm diameter Petri dishes and placed in contact with 1 mL of each of the following solutions: control with distilled water, untreated, and treated sample. Subsequently, the plates with cucumber seeds were incubated at 20 °C for 72 h in the absence of light. Two repetitions were conducted per treatment. The germination index, GI, was determined using the number of germinated seeds in each sample, GSA, the number of seeds that germinated in the control, GSC, and the average root lengths for essays and controls, RLS and RLC, respectively, according to the following Equation (4) [42,43]:

$$\mathrm{GI}\left(\%\right)=\left({\displaystyle \frac{{\mathrm{GS}}_{\mathrm{A}}}{{\mathrm{GS}}_{\mathrm{C}}}}\right)\times \left({\displaystyle \frac{{\mathrm{RL}}_{\mathrm{S}}}{{\mathrm{RL}}_{\mathrm{C}}}}\right)\times 100$$

(4)

3. Results and Discussion

TOC, COD, and herbicide removal were evaluated through factors that affect the performance of the electrochemical process, such as current density, j (mA/cm²), volumetric flow rate, Q (cm³/min), the initial concentration of herbicide, [herb] (mg/L), and pH (see

Table 2. Coded levels and real values for CCD and RSM analyses of the electrochemical treatments of PQ and DQ solutions in 0.05 M Na₂SO₄.

Herbicide	Experiment	Coded Value			Real Values			Observed Responses
					Herbicide mg/L	j mA/cm2	Q cm3/min	% Removal
		A	B	C				COT	Herbicide	DQO
PQ	1	−1	−1	−1	30	0.5	250	35.65	72.92	26.13
	2	1	−1	−1	90	0.5	250	36.97	65.38	65.04
	3	−1	1	−1	30	1.5	250	81.60	98.77	97.84
	4	1	1	−1	90	1.5	250	84.87	94.06	94.69
	5	−1	−1	1	30	0.5	750	40.42	87.59	92.71
	6	1	−1	1	90	0.5	750	39.75	65.69	65.71
	7	−1	1	1	30	1.5	750	82.10	100	100
	8	1	1	1	90	1.5	750	87.08	95.94	94.49
	9	−1.68	0	0	9.55	1	500	76.32	100	100
	10	1.68	0	0	110.46	1	500	75.04	91.65	92.68
	11	0	−1.68	0	60	0.16	500	4.17	9.90	11.99
	12	0	1.68	0	60	1.84	500	92.41	98.68	100
	13	0	0	−1.68	60	1	79.55	82.07	94.18	91.91
	14	0	0	1.68	60	1	920.45	81.32	98.92	97.01
	15	0	0	0	60	1	500	71.75	92.90	95.93
	16	0	0	0	60	1	500	80.45	97.13	95.39
	17	0	0	0	60	1	500	81.31	96.40	95.16
	18	0	0	0	60	1	500	76.10	95.02	95.66
	19	0	0	0	60	1	500	77.99	95.30	91.49
	20	0	0	0	60	1	500	74.67	94.21	87.82
DQ	1	−1	−1	−1	30	0.5	250	39.41	92.30	28.56
	2	1	−1	−1	90	0.5	250	32.23	97.84	5.23
	3	−1	1	−1	30	1.5	250	82.44	100	53.28
	4	1	1	−1	90	1.5	250	86.02	100	47.09
	5	−1	−1	1	30	0.5	750	37.87	100	46.03
	6	1	−1	1	90	0.5	750	7.43	66.69	17.89
	7	−1	1	1	30	1.5	750	71.38	100	49.10
	8	1	1	1	90	1.5	750	86.57	100	57.15
	9	−1.68	0	0	9.55	1	500	60.84	100	29.08
	10	1.68	0	0	110.46	1	500	72.61	98.63	57.60
	11	0	−1.68	0	60	0.16	500	25.98	35.90	29.08
	12	0	1.68	0	60	1.84	500	91.48	99.62	70.62
	13	0	0	−1.68	60	1	79.55	80.24	100	58.81
	14	0	0	1.68	60	1	920.45	57.91	100	57.90
	15	0	0	0	60	1	500	70.50	100	48.06
	16	0	0	0	60	1	500	70.61	100	92.94
	17	0	0	0	60	1	500	82.07	100	73.59
	18	0	0	0	60	1	500	75.59	100	70.28
	19	0	0	0	60	1	500	70.56	100	68.44
	20	0	0	0	60	1	500	77.43	100	82.29

).

3.1. The Influence of Independent Experimental Variables on Electrochemical Processes

Electrochemical experiments were carried out with PQ and DQ solutions at circumneutral pH (pH = 6.5) using the values according to the experimental design matrix using CCD and RSM. The results observed in Table 2 generated the following TOC removal equations.

$${\mathrm{PQ}}_{\mathrm{Removal}}^{\mathrm{TOC}}\left(\%\right)=-26+0.216\left[\mathrm{PQ}\right]+140.3 \mathrm{j}+0.005 \mathrm{Q}-0.00217 {\left[\mathrm{PQ}\right]}^2-46.56{ \mathrm{j}}^2+0.063\left[\mathrm{PQ}\right]\mathrm{j}-0.0048 \mathrm{jQ}$$

(5)

$${\mathrm{DQ}}_{\mathrm{Removal}}^{\mathrm{TOC}}\left(\%\right)=5.8+0.243\left[\mathrm{DQ}\right]+70.11\mathrm{j}+0.0343 \mathrm{Q}-0.005112 {\left[\mathrm{DQ}\right]}^2-29.71{\mathrm{j}}^2+0.47 \left[\mathrm{DQ}\right]\mathrm{j}-0.00019\left[\mathrm{DQ}\right]\mathrm{Q}+0.0158 \mathrm{jQ}$$

(6)

From these equations, the positive linear coefficients of PQ and DQ, j and Q, and the positive coefficient found for the interaction between herbicide concentrations and j, reveal that mineralization is favored by an increase in current density (despite the negative quadratic coefficients obtained for herbicide concentration and current density). These equations in terms of real factors are used to make predictions about the responses for the given levels of each factor.

The overall effects can be visualized through the polynomial plots for PQ (Figure 2) and DQ degradation (Figure 3) constructed from Table 2. Surface analysis (Figure 2) indicates that removals of 91%, 100%, and 100% of TOC, PQ, and COD, respectively, were obtained by applying a current density of 1.55 mA/cm² with a volumetric flow rate of 79.55 mL/min for the degradation of PQ. In the case of DQ, the response surface (Figure 3) shows that 92%, 100%, and 77% removals of TOC, DQ, and COD were obtained, respectively, with a current density of 1.84 mA/cm² and a volumetric flow rate of 495.75 mL/min, for the degradation of DQ (Table 2). The pH of the solution decreased during the process (i.e., from 6.7 to 4). This can be attributed to the intermediate degradation products of herbicide on the anode surface, whereby ^•OH radicals can oxidize organic compounds to acids. There is a residual COD at the end of the electrolytic process that reveals remaining organic matter present in the electrolyte; even though the DQ and PQ molecules are no longer observed in the UV–Vis analyses, there is still a remnant of around 20% organic matter (for the case of higher current density) which can be attributed to intermediates still present at the end of the electrolysis.

Table 3. Optimized values for the variables under study.

	Optimal Conditions			% Removal
Herbicide	Herbicide mg/L	j mA/cm2	Q mL/min	Herbicide	TOC	COD
PQ	72.74	1.55	79.95	100	91	100
DQ	67.64	1.47	495.75	100	92	77

summarizes the optimal conditions and efficiencies obtained with the variables studied for both DQ and PQ.

Validation of the Quadratic Models

The F-values for lack of fit in the previous models were 3.22 for PQ and 5.96 for DQ, both lower than the critical value at a 95% confidence level. This indicates that the model adequately describes the experimental data, with R² values of 0.97 and 0.93, respectively (

Table 4. ANOVA statistics for quadratic models from CCD for PQ and DQ generated the following TOC removal.

	Sum of Squares	Mean Square	F-Value	p-Value Prob > F
PQ
Model	10,045.26	1116.14	40.52	<0.0001
A = [PQ]	3.33	3.33	0.1210	0.7351
B = j	8035.09	8035.09	291.70	<0.0001
C = Q	5.93	5.93	0.2153	0.6526
A2	55.17	55.17	2.00	0.1874
B2	1952.72	1952.72	70.89	<0.0001
C2	0.4166	0.4166	0.0151	0.9046
Residual	275.46	27.55
Lack of Fit	210.21	42.04	3.22	0.1125
R2 = 0.9733    Adjusted R2 = 0.9493    Predicted R2 = 0.8367
DQ
Model	9449.62	1049.96	13.26	0.0002
A = [DQ]	0.0653	0.0653	0.0008	0.9776
B = j	7480.62	7480.62	94.47	<0.0001
C = Q	405.37	405.37	5.12	0.0472
A2	305.00	305.00	3.85	0.0781
B2	794.95	794.95	10.04	0.0100
C2	204.78	204.78	2.59	0.1389
Residual	791.88	79.19
Lack of Fit	678.16	135.63	5.96	0.0361
R2 = 0.9227    Adjusted R2 = 0.8531    Predicted R2 = 0.4589

). The proximity of the determination coefficients to unity further confirms the statistical significance of both quadratic models [44,45].

p-values below 0.05 indicate significant model terms; in this case, j and j² are significant. Values greater than 0.1000 suggest non-significant terms. The lack-of-fit F-value of 3.22 for PQ indicates that the lack of fit is not significant relative to pure error. A Pareto chart analysis was used to derive more meaningful conclusions regarding the magnitude and importance of different effects (i.e., variables and interactions) (Figure 4). This chart, in terms of t-values with two limits (i.e., t-value and Bonferroni limit), shows that j exceeds both, reflecting the statistical significance of this study and further supporting the above conclusions. From the height of the bars, the main factor (B = j) has the greatest effect on the response (i.e., TOC removal efficiency, %), compared to the other factors (A = herbicide concentration, and C = volumetric flow) and their interaction effects. In summary, the significance analysis shows that only j is statistically significant.

3.2. Phytotoxicity Assay

An important consideration in the application of advanced oxidation processes (AOPs) is the potential formation of toxic intermediates or byproducts, which could be even more toxic than the original compounds present in the samples. This makes conducting phytotoxicity tests crucial. Both pollutants, PQ and DQ, exhibited significant phytotoxicity on the cucumber seeds (see

Table 5. Phytotoxicity of PQ and DQ solutions for cucumber seeds.

Treatment	Seed Germination, GS (%)	Root Length, RL (%)	Germination Index, GI (%)
PARAQUAT
Control (distilled water)	100	100	100
Raw effluent	90	19.20	17.28
Treated effluent	100	92.82	92.82
DIQUAT
Control (distilled water)	100	100	100
Raw effluent	90	14.97	13.47
Treated effluent	100	39.91	39.91

). The germination rate of the cucumber seeds in contact with the untreated samples was lower than that of both the treated samples and the control group. There was a clear difference in germination between the cucumber seeds exposed to the treated effluent and those in the control group.

The results indicate that both untreated and treated samples were phytotoxic, as the GI values for the seeds were 93 and 40% for PQ and DQ, respectively. Specifically, the GI increased from 17.28% and 13.47% (untreated) to approximately 93% and 40% (treated). The degradation of DQ and PQ by a solar photo-Fenton process mediated by Fe (III) -oxalate complexes reported a GI increase for DQ from 4.7 to 55.8%, and for PQ from 16.5% to 59.7%, using Cucumis sativus seeds [14]. This indicates that the electrochemical process was more efficient for PQ compared to DQ with respect to reducing effluent toxicity for each herbicide.

3.3. Kinetic Study

The kinetics of the reaction between PQ and DQ and the ^•OH generated during electrochemical oxidation were monitored using UV–Visible spectrophotometry.

Table 6. Pseudo-first-order rates for the electrochemical degradation of PQ and DQ.

Herbicide	Equation	Rate Constants	R2
PQ	C = e−0.0087t	0.0087 min−1	0.9916
DQ	C = e−0.0105t	0.0105 min−1	0.9669

shows that the pseudo-first-order model accurately describes both systems. The experimental data were fitted to this model using nonlinear regression analysis with Statistica 10 software. Figure 5 illustrates the decrease in PQ and DQ concentrations at a similar rate under optimal conditions, with 70 mg/L of DQ and 73 mg/L of PQ. PQ was essentially eliminated after 300 min, while DQ disappeared in 240 min. From this analysis, rate constants of 0.0087 min⁻¹ (R² = 0.99) for PQ and 0.0105 min⁻¹ (R² = 0.97) for DQ were determined. This behavior indicates the production of a constant amount of ^•OH during treatment, at least while the herbicides are being degraded.

Ghavi et al. [46] report kinetic studies on PQ degradation that fit a pseudo-first-order kinetic model, with rate constant of 0.0299 min⁻¹ for the UV/sodium persulfate/nanoparticles of TiO₂ process (77% degradation efficiency) and 0.0604 min⁻¹ for a UV/sodium periodate/nanoparticles of TiO₂ process (90% degradation efficiency) under AOP hybrid conditions. These experiments were conducted with an initial herbicide concentration of 30 mg/L at 25 °C over 40 min. Similarly, Valenzuela et al. [22] report pseudo-first-order rate constants for DQ degradation at different initial concentrations, for advanced electrochemical oxidation processes using boron-doped diamond (BDD) anodes. Their results showed 100% DQ degradation in a three-electrode cell with a rate constant of 0.013 min⁻¹ at an initial concentration of 50 mg/L and a current density of 1.5 mA/cm². In an H-type cell, they achieved 92% degradation, with a rate constant of 0.009 min⁻¹ for the same initial concentration and a current density of 7.5 mA/cm² at pH 2 over 5 h. The rate constants reported here are similar to those from the three- electrode BDD cell, even at circumneutral pH, indicating that the process is efficient without the need for water acidification.

3.4. Parameters Analyzed

Another important parameter is the Instantaneous Current Efficiency (ICE), which indicates the fraction of the applied current that is used to reduce the initial chemical oxygen demand (COD) over a given period of time. In other words, this parameter gives an estimate of the overall efficiency of the process and is calculated using the following expression [47]:

$${\mathrm{ICE}}_{\mathrm{COD}}={\displaystyle \frac{\mathrm{FV}}{8\mathrm{I}}}\left({\displaystyle \frac{\left[{\left(\mathrm{COD}\right)}_{\mathrm{t}}-{\left(\mathrm{COD}\right)}_{\mathrm{t}+\Delta \mathrm{t}}\right]}{\Delta \mathrm{t}}}\right)$$

(7)

where (COD)_t and (COD)_{t + Δt} are the chemical oxygen demands (in g O₂/L) at times t and t + Δt (in s), respectively; F is Faraday’s constant (i.e., 96,487 C/mole), I is the applied current (in A), and V is the volume of the electrolyte (in L).

The ICE (%) values obtained are displayed in Figure 6. It can be observed that the ICE is significantly lower for DQ compared to PQ. For both herbicides, the ICE decreases over time, as the primary reactions involve the degradation of the reactive groups in the herbicide molecules. As their concentrations decrease, the degradation of their products becomes more complex. Additionally, water electrolysis becomes more significant, further reducing overall efficiency. Simpler molecular structures are known to result in higher ICE values [48], as seen in the case of PQ.

Under mass transport control, ICE decreases as the process progresses, and the COD for PQ decays exponentially. The ICE values at 0.5 h were 100% for PQ and 9% for DQ, indicating that a very low concentration of the organic compound was involved in the system for DQ. Similar ICE characteristics have been reported for the oxidation of other organic molecules, such as bentazon [49], glyphosate [50], diquat [22], and herbicide wastewater [51].

On the other hand, energy consumption (kWh) per gram of TOC removed (ΔTOC) during the degradation period is given by the following expression [52]:

$${\mathrm{EC}}_{\mathrm{TOC}}={\displaystyle \frac{{\mathrm{E}}_{\mathrm{c}}\mathrm{It}}{\left(\Delta \mathrm{TOC}\right){\mathrm{V}}_{\mathrm{s}}}}$$

(8)

where E_c is the cell potential (V), I is the applied current (A), t is the electrolysis time (h), Vs. is the solution volume (L), and (ΔTOC) is the corresponding experimental abatement of TOC (in mg/L).

Figure 7 shows the energy consumption per unit of TOC mass. The most economical process was achieved at 1.55 mA/cm² for PQ, with an energy consumption of 0.1 kWh/gTOC at 300 min, during which PQ concentration was reduced by 100% and TOC by 91%. For DQ, the energy consumption was 0.24 kWh/gTOC at 300 min, with a 100% reduction in concentration and a 92% reduction in TOC. These values are lower than those reported by Valenzuela et al. [22], where the EC values were 0.3 kWh/gTOC for an undivided cell and 1.8 kWh/gTOC for an H-type cell, making the EC obtained in this study more efficient, particularly for DQ.

The energy cost of the process can be calculated using the following equation [53]:

$$\mathrm{Cost} = \mathrm{Energy} \mathrm{consumption} \times \mathrm{Cost} \mathrm{per} \mathrm{kWh}$$

(9)

Considering basic consumption, the price of electricity in Mexico is USD 0.047 kWh, making the energy cost per gram of TOC USD 0.0047 kWh for PQ and USD 0.0113 kWh for DQ.

3.5. Factors Influencing PQ and DQ Electrochemical Degradation

3.5.1. Effect of Current Density

The degradation mechanism of contaminants during electrochemical oxidation using BDD electrodes involves both direct and indirect oxidation. In indirect oxidation, hydroxyl radicals generated by the electrode selectively attack and mineralize organic substances. As shown in Figure 8, the removal rates of COD, TOC, PQ, and DQ increased with a rise in current density from 0.16 mA/cm² to 1.84 mA/cm², indicating that current density directly impacts the efficiency of hydroxyl radical generation [54].

High PQ removal (99%) was achieved with a maximum current density of 1.84 mA/cm² after 300 min of electrolysis, and TOC removal increased from 37% (at 50 mA) to 87% (at 184 mA). Complete DQ removal (100%) occurred at a current density of 1.0 mA/cm² after 200 min, with TOC removal improving from 26% (at 16 mA) to 91% (at 184 mA). This is because an increased current density produces more ^•OH radicals, accelerating the electron transfer rate for direct oxidation at the electrode surface, as well as indirect oxidation by ^•OH [55]. Therefore, the rate of electrochemical degradation accelerates, and higher current densities generate more hydroxyl radicals, enhancing herbicide mineralization [56].

However, as shown in Figure 8, when the current exceeds 1.5 mA/cm², the COD and TOC removal rates for PQ slightly decrease. This suggests that excessive current density can negatively affect electrolysis efficiency. Higher current densities do not necessarily improve removal rates, as hydroxyl radical concentrations may decrease due to secondary reactions [57], as well as by the recombination of ^•OH radicals.

3.5.2. Effect of Initial Concentration

The initial concentration of PQ and DQ was varied between 9.55 and 110.46 mg/L to examine its impact on removal efficiency. As shown in Figure 9, the removal efficiency of PQ decreased from 100% to 89.6% after 270 min of electrolysis when the initial concentration increased from 9.55 mg/L to 110.46 mg/L. Similarly, the removal rate of DQ dropped from 100% to 54.8% after 180 min of electrolysis with the same increase in initial concentration. The initial concentration of PQ had a lesser impact on degradation efficiency compared to DQ.

As the concentration of PQ and DQ increases, the number of organic molecules present rises, but the hydroxyl radicals initially generated are insufficient to degrade all the herbicide molecules, leading to a reduction in degradation efficiency [54].

The concentrations used in this study are higher than those typically reported in the literature, as the focus is to assess the potential for eliminating high levels (9–110 mg/L) of herbicides, which are usually produced during the washing of containers or packaging used for these chemical substances [58]. The triple rinse method recommended by the EPA [59] can remove up to 99.99% of pesticides from containers; however, it produces large volumes of contaminated water with highly variable agrochemical concentrations (100–500 mg/L) [58,60]. In terms of chemical oxygen demand (COD, mg O₂/L), some studies report values ranging from 200 to 2000 mg O₂/L, indicating that these waters contain higher organic matter concentrations than typically found at the inflows of urban and industrial treatment systems [60,61,62,63]. Finally, it is important to note that field samples often exceed limits set by government agencies. Therefore, developing in situ treatment systems to prevent these contaminants from reaching surface water bodies is of significant interest to the scientific community [60,63].

3.6. Factors Influencing PQ and DQ Electrochemical Degradation: Concentration and Current Density

The initial concentrations of PQ and DQ, ranging from 10 to 110 mg/L, were selected to study their influence on removal efficiency. Increasing PQ concentration did not significantly affect its removal rate, which remained around 100%. Additionally, TOC removal (i.e., mineralization) increased proportionally with the concentration up to 90 mg/L (Figure 10a,b). However, concentrations beyond this threshold resulted in diminished TOC removal, with a 12% reduction (from 83 to 71%, Figure 10a).

For DQ and PQ, deeper removal (>95%) and higher mineralization (TOC removal above 90%) were observed at lower concentrations (10–90 mg/L) (Figure 10b). During degradation, the two substances exhibited different behaviors, likely due to structural differences that influence the susceptibility to radical attack [64].

Previous studies on the degradation of DQ and PQ by advanced oxidation processes have shown that DQ is more recalcitrant than PQ. Florêncio et al. [57] suggested that the formation of a redox pair Br⁻/Br in diquat solutions might be responsible for this behavior. Furthermore, other studies highlight that advanced oxidation processes often result in partial removal (70–90%), producing a partially transformed effluent rather than one that is deeply mineralized [20,22,65].

While electrooxidation at lower herbicide concentrations is effective, key performance indicators (such as EC_TOC, carbon emissions, and cost) rise dramatically. For instance, decreasing the herbicide concentration from 90 mg/L to 10 mg/L results in an approximately 1190% increase in these figures of merit. At higher concentrations, the availability of hydroxyl radicals becomes insufficient to fully degrade/mineralize the herbicide, leading to a reduction in overall performance [54]. However, a lower carbon footprint is seen at the highest concentrations due to reduced hydroxyl radical recombination reactions, thanks to the abundance of organic molecules (Figure 11b).

Complete herbicide removal (100%) and high mineralization levels (~87–90%) were achieved at the highest current density (1.84 mA/cm²) after 5 h of treatment (Figure 3). The mechanism behind this is straightforward: as the current density increases, more ·OH are produced, accelerating the electrochemical degradation and resulting in the near-complete mineralization of the treated solution [55]. Increasing the current density from 1.5 to 1.84 mA/cm² improves process efficiency by shortening the time needed to remove 80% of the initial herbicide concentration (PQ: from 165 to 90 min, DQ: from 130 to 80 min).

Additionally, highly mineralized effluents (~90%) were achieved after 300 min of treatment. However, the primary characteristic of higher current densities is their impact on profitability and sustainability [66]. For instance, carbon emissions increased 113% for DQ and 75% for PQ, while costs rose substantially—by 113% for PQ (from USD 3.8 to 8.1 per m³) and by 74% for DQ (from USD 3.4 to 5.9 per m³). As shown in Figure 11, the best results were obtained at 1.5 mA/cm², which aligns with the best operating conditions suggested by the experimental design (i.e., PQ_[j]: 1.55 mA/cm² and DQ_[j]: 1.47 mA/cm²).

As shown in Figure 11, when the current exceeds 1.5 mA/cm², the COD and TOC removal rate decreases slightly. This phenomenon suggests that when the current density surpasses a certain threshold, it negatively affects electrolysis efficiency. Excessive current density leads to secondary reactions, such as the recombination of hydroxyl radicals, reducing the overall effectiveness of the process.

3.7. Environmental and Economic Performance of the Electrochemical Process

For over 40 years, electrooxidation processes have been used effectively to mineralize 90–99% of recalcitrant and toxic compounds under various operating conditions (current density, anode material, electrolyte, pH, organic load, and solar- and UV-assisted treatments) [66,67,68]. However, concerns regarding energy costs and environmental impact remain barriers to the widespread adoption of this technology at the industrial level [66,67]. Therefore, we analyzed the effects of different energy sources, current density, and herbicide concentration on carbon emissions and treatment cost.

The best operating conditions led to a total removal (100%) and deep mineralization (>90%) of herbicides after 5 h of treatment (Table 2). The cost of herbicide treatment can vary greatly due to fluctuating electricity prices, ranging from as low as 0.002 up to 0.53 USD/kWh. While many factors impact electricity costs—including fuels, plant costs, weather, regulations, location, consumers, and the transmission and distribution system—the major drivers are generation, distribution, and transmission costs [69].

This study considered electricity prices of 0.1 USD/kWh (low), 0.25 USD/kWh (average), and 0.5 USD/kWh (high). Extremely low prices (e.g., <0.05 USD/kWh) reported by coal, oil, and gas exporters such as Iran, Iraq, Algeria, Russia, and Libya were excluded. The highest treatment costs (max cost) for PQ (Figure 12a) and DQ (Figure 12b) are expected in countries or regions with high state and local taxes, as well as elevated transmission and distribution costs (e.g., Ireland, United Kingdom, Italy, Belgium, and Germany).

Conversely, the lowest treatment costs (min cost) will be seen in countries with cheaper electricity generation options, where electricity prices fall below 0.2 USD/kWh. These low prices are often driven by local subsidies and/or abundant natural resources like rivers, wind, solar irradiation, and non-renewable reserves (e.g., coal, oil, gas), which are used to produce electricity. Examples include Mexico, Canada, Colombia, Brazil, the USA, and Sweden.

Lowest carbon emissions (<20 kg CO₂/kg TOC removed) are achieved in countries like Brazil, Sweden, and Norway, where electricity primarily comes from renewable sources such as hydro, solar, or wind. These sources, along with power generation methods with lower carbon intensities like nuclear energy, contribute to lower emissions (Figure 12c). On average, around 60% of the world’s electricity is generated from coal and gas. As a result, countries heavily reliant on non-renewable energy sources may see carbon emissions up to five times higher, reaching as much as 100 kg CO₂/kg TOC removed.

Interestingly, the carbon footprint of the electrooxidation process could be reduced by 50–90% (to around 10–12 kg CO₂/kg TOC removed) if the electricity grid is powered by green resources like wind or photovoltaics instead of non-renewables [60,63,64]. Furthermore, electrooxidation technologies avoid using catalysts (such as iron solutions), oxidants (like hydrogen peroxide), and intensifiers (such as persulfate), which would otherwise increase both costs and the carbon footprint of the treatment process [70,71].

Considering the global average for electricity generation (61% from fossil fuels, 9% from nuclear, and 30% from renewables), the electrooxidation treatment of PQ and DQ could result in emissions of around 2.7 and 38.9 kg CO₂/m³ of water treated, respectively. These values align with the literature reports of emissions ranging from 2 to 50 kg CO₂/m³ and treatment costs between USD 2 and USD 50 per cubic meter [72]. If renewable energy sources are used, the emissions could be significantly reduced to 0.28 kg CO₂/m³ for PQ and 0.40 kg CO₂/m³ for DQ.

Higher emissions are usually reported for more complex wastewaters, where factors like reactor configuration (batch versus continuous), operating conditions (basic or acidic), and the presence of radical scavengers increase the use of reagents and extend reaction times [63,65]. Moreover, the nature of the target pollutant and kinetic considerations (such as reaction order, phase reaction, and reaction mechanisms) can also affect the overall operating costs and carbon emissions of the process [71,73].

Boron-doped diamond (BDD) electrodes possess unique properties, including high oxidative power, exceptional chemical and physical stability, low fouling even under extreme operating conditions, and a reduced background current [67,74,75]. These characteristics result in higher mineralization efficiencies and shorter reaction times, ultimately leading to lower electrical energy consumption per gram of substance removed [67,74]. Unlike other advanced oxidation processes, it does not require catalysts (such as iron or copper salts) or additional oxidizing agents (e.g., H₂O₂, O₃) to generate free radicals, thereby avoiding potential interference with subsequent treatments and preventing increased conductivity in the discharged effluents [74,76]. These systems can be designed to be compact and modular, allowing for easy transportation to agricultural centers, thus eliminating the need for storing and transporting potentially hazardous substances. Additionally, these electrodes can be paired with renewable energy sources, such as wind and solar power, to mitigate one of their primary drawbacks: the carbon dioxide emissions associated with electricity production.

In this sense, Fernández-Marchante et al. [77] conclude that it is possible to achieve carbon footprints as low as 0.002 kg CO₂-eqv./L of treated water. The direct use of electrical energy generated from renewable sources, without relying on batteries or storage systems, has recently demonstrated promising results in the treatment of herbicides such as 2,4-dichlorophenoxyacetic acid [67,78] and clopyralid (3,6-dichloro-2-pyridinecarboxylic acid) [78]. However, the variability in energy supply often impacts the current density, which notably affects the production of oxidants in these electrodes [67,74,76].

4. Conclusions

The degradation of PQ and DQ was achieved through electrochemical treatment using boron-doped diamond electrodes (BDDEs). A central composite design and response surface methodology were employed to identify optimal conditions for maximizing herbicide removal at circumneutral pH. The best conditions were a current density of 1.5 (±0.05) mA/cm² and a treatment time of 300 min, resulting in 100% removal efficiency for both herbicides. Additionally, a 92% TOC removal for DQ and 91% for PQ was achieved, while COD removal reached 77% for DQ and 100% for PQ.

The experimental data under these optimal conditions closely followed a pseudo-first-order kinetic model. The instantaneous current efficiency (ICE) was higher for PQ than for DQ, consistent with the higher applied current density (1.55 mA/cm²). The energy consumption (EC) values obtained were 0.1 kWh/g TOC for PQ and 0.24 kWh/g TOC for DQ, indicating that this process is an effective, energy-efficient, and user-friendly alternative for treating herbicide-contaminated water.

Regarding carbon emissions, treatment of PQ and DQ by electrooxidation resulted in emissions between 2.7 and 38.9 kg CO₂/m³ of treated water, with associated costs ranging from USD 2 to USD 50 per cubic meter. These emissions could be significantly reduced to 0.28 kg CO₂/m³ for PQ and 0.40 kg CO₂/m³ for DQ if electricity is sourced from renewable energy. In terms of toxicity, the process was more effective for PQ than DQ, particularly in reducing effluent phytotoxicity.