Electrochemical degradation of DCF by boron-doped diamond anode: degradation mechanism, pathways and influencing factors

Diclofenac (2-[(2,6-dichlorophenyl) amino] phenyl acetic acid, DCF), is one of most common nonsteroidal anti-inflammatory drugs (NSAIDs) effective in relieving pain and inflammation, and is widely used all over the world, with a yearly consumption of approximately 940 tons (Zhang et al. 2008). As an organic weak acid with relatively strong polarity, DCF is not likely to hydrolyze or volatilize, and it has been detected in surface water, groundwater, and wastewater concentrations ranging from ng/L to μg/L (Stuer-Lauridsen et al. 2000; Tauxe-Wuersch et al. 2005; Zorita et al. 2009). Such an invasive presence indicates that it cannot be effectively removed from wastewater treatment plants (WWTPs) (Lindqvist et al. 2005; Wu et al. 2019). Lishman et al. (2006) found 12 out of 26 samples where DCF concentrations were higher in WWTP effluent compared to influent in Ontario. Kreuzinger et al. (2004) reported that at least 8 days of sludge retention time were needed to significantly biodegrade DCF. The average removal rate of DCF was very low (26%) based on the sewage treatment process reported by Lindqvist et al. (2005). Some researchers focus on its potential toxicity to the environment (Fent et al. 2006; Aslan et al. 2020; Duarte et al. 2020). Oaks et al. (2004) referred to DCF residues as the cause of vulture population decline in Pakistan. Its testicular and epididymal toxicity also attracts attention (Owumi et al. 2020). Fu et al. (2020) found DCF residue in water to be biotransformed by Gammarus pulex and Hyalella azteca, representing a 25 to 110-fold increase of bioconcentration factors. The safe and effective removal of DCF residue has been a growing concern. Adsorption and advanced oxidation process such as photodegaradation are commonly known methods for the removal of pharmaceutical drugs (Ali et al. 2016; Sharma et al. 2018; Balasubramani et al. 2020; Dhiman et al. 2021). Ultrasonic, ozonation, the Fenton process, photocatalysis, chlorine oxidation, and electrolysis have been studied to examine and compare the removal effect on DCF (Huber et al. 2003; Brillas et al. 2010; Nie et al. 2014; Wang et al. 2014; Wang et al. 2015a; Kumar et al. 2020; Li et al. 2020a, 2020b).

In recent years, the use of boron-doped diamond (BDD) anodes in environmental protection areas have been widely investigated (Murugananthan et al. 2008; Pacheco et al. 2011; Rabaaoui et al. 2013; Fernandes et al. 2014; Antonin et al. 2015; Yang et al. 2016; Fudala-Ksiazek et al. 2018; Garcia-Espinoza et al. 2018). BDD anodes show high chemical stability in harsh environments including high voltage and current densities (Chen et al. 2003; Yang et al. 2016). During the electrolysis process, BDD anodes are able to produce weakly adsorbed hydroxyl radicals (•OH) from water oxidation. The producing •OH is the most oxidizing species after fluorine (E° = 2.8 V VERSUS the standard hydrogen electrode (SHE)), and it can unselectively mineralize many organic compounds (Pacheco et al. 2011), thereby showing advantages in mineralization efficiency compared with other anode materials, such as platinum (Pt), graphite felt, and dimensionally stable anodes (DSA) i.e., titanium and ruthenium oxide–iridium(IV) oxide (Ti/RuO₂-IrO₂)) (Sopaj et al. 2016). Several studies have used BDD anodes to degrade DCF. Brillas et al. (2010) reported that the use of BDD anodes with a lower current could significantly promote the total organic carbon (TOC) abatement of DCF compared with Pt anodes in a neutral buffer medium. Another study found that a TOC removal of about 90% could be obtained for all studied DCF concentrations using BDD electrodes (Ihos et al. 2015). But these studies focused more on comparing BDD with other anodes on TOC removal without systematic research on the influencing factors and degradation pathways. Further study on the influencing factors, degradation products, degradation pathways, and degradation mechanism of DCF implemented by BDD anodes is still needed.

In this study, a series of comparative experiments was carried out to achieve three primary objectives: (1) to investigate the degradation characteristics of DCF in the electrochemical advanced oxidation process using BDD anodes; (2) to examine the effects of water quality parameters and operating parameters (i.e., humic acid (HA), bicarbonate (HCO₃⁻), chloride ion (Cl⁻), pH, electrolyte activity, current density, electrode spacing) on the electrochemical degradation; and (3) to identify the main degradation products and the DCF degradation pathway in electrochemical process using BDD anodes. The results contribute to understanding the mechanism and influencing factors of DCF degradation by niobium (Nb)/BDD anodes, and provide reference for the practical application of electrochemical treatment on pharmaceutical wastewater.

Diclofenac sodium (C₁₄H₁₀Cl₂NNaO₂) and HA were purchased from Sigma Aldrich (St. Louis, MO, USA). High-performance liquid chromatography (HPLC) grade methanol (MeOH) was supplied by Merck (Darmstadt, Germany). Sodium sulfate (Na₂SO₄), sodium chloride (NaCl), sodium bicarbonate (NaHCO₃), sodium nitrate (NaNO₃), disodium hydrogen phosphate (Na₂HPO₄), sodium dihydrogen phosphate (NaH₂PO₄) and tertiary butanol (TBA) were analytical grade and obtained from Sinopharm Chemical Reagent Co. Ltd (Shanghai, China). All solutions were prepared with ultrapure water, with a resistivity of at least 18.2 MΩ cm⁻¹, produced from a Millipore purification system (Billerica, CA, USA), and stored at 4 °C in the dark.

Electrolysis experiments were carried out in an undivided cylindrical cell made of polymethyl methacrylate. Two pieces of bipolar Nb/BDD electrode (10 cm × 5 cm × 2 mm, 6 μm coating thickness with 2,500 ppm boron doping) were purchased from NeoCoat company (La Chaux-de-Fonds, Switzerland). The effective electrode surface area submerged in the solution was 40 cm². Solutions containing 30 mmol L⁻¹ Na₂SO₄ were used as the electrolyte. Constant current was supplied by an eTM-K8011SPL DC power (eTommens, China). A stir bar was placed inside the cell and a magnetic stirring apparatus (AS ONE, Japan) was used to ensure that the solution was well mixed.

The initial solution pH was adapted by 0.1 mol L⁻¹ sodium hydroxide (NaOH) or sulfuric acid (H₂SO₄). The buffer solution was not used because it may interfere with the electrolysis process. Two electrodes were placed face to face with a gap of 2 cm. 50 μmol L⁻¹ DCF and electrolyte were added to 1,000 mL of reaction solution. To compare the effects of different electrolytes, NaNO₃ and NaH₂PO₄ were selected and added to the same initial conductivity as 30 mmol L⁻¹ Na₂SO₄. After the reaction solution was thoroughly mixed, experiments were initiated by applying a certain constant current. At predetermined time intervals, samples were withdrawn and filtered through a 0.22-μm polytetra fluoroethylene membrane before analysis. All experiments were performed at room temperature and conducted at least twice to ensure consistency.

The pH, temperature, and conductivity values of the reaction solution were measured using a HQ40-d multi-parameter water analyzer (Hach, Loveland, CO, USA). DCF was analyzed using an Agilent 1260 HPLC (Agilent Technologies, Santa Clara, CA, USA) equipped with an Eclipse XDB-C18 analytical column (4.6 × 150 mm, 5 μm particle size) and a photodiode array detector. The mobile phase with a flow rate of 1 mL min⁻¹ consisted of 30% formic acid hydroxide solution (0.1%, v/v) and 70% MeOH. The detection wavelength was 276 nm. Scanning electron microscopy (SEM) and Raman spectra analysis were performed using a Hitachi Regulus 8100 SEM (Hitachi, Ltd, Japan) and Thermo Fisher DXR2 Raman spectrometer (Thermo Fisher Scientific, USA).

The transformation products (TPs) were purified and concentrated by a solid phase extraction (SPE) workstation (Supelco Analytical, Bellefonte, PA, USA) with a 3 cc/60 mg Waters HLB cartridge (Waters, Corporation, Milford, MA, USA), and were identified through analysis using HPLC coupled with electrospray ionization (ESI) and quadrupole time-of-flight (qTOF) mass spectrometry (MS). Equipment used was the Ultimate 3000 HPLC (Thermo Fisher Scientific, Waltham, MA, USA), and the maXis Q-TOF-MS (Bruker AXS, Karlsruhe, Germany) with an ESI interface operated in the positive ion mode (ESI+)). Detailed procedures of SPE, further information about the mobile phase operation condition, and operating parameters of HPLC-ESI-qTOF-MS were noted in Text S1, Table S1, and Text S2 in supplementary material.

Early experiments show that there was no obvious loss of DCF for several hours in the 30 mmol L⁻¹ of SO₄²⁻ solution without current. With the direct current power supplied, the decline of DCF followed the pseudo-first-order kinetics with an obvious kinetics constant of 0.0429 min⁻¹ (C₀ = 50 μmol L⁻¹, Table 1). Two other types of electrolytes which could not be oxidized or reduced in this electrochemical process, nitrate (NO₃⁻) and phosphate (PO₄³⁻) which had the same conductivity, were compared with SO₄²⁻ (Figure S1). Results show that the kinetics constants dropped to 0.0263 min⁻¹ (NO₃⁻) and 0.0111 min⁻¹ (PO₄³⁻), which were 61.3% and 25.9% of the kinetics constants in the reaction using SO₄²⁻ as the electrolyte. After 30 min reaction, the removal rates of DCF were 73.0% (SO₄²⁻), 52.7% (NO₃⁻), and 28.6% (PO₄³⁻), respectively (see Table 1).

The degradation of DCF with SO₄²⁻ was much faster than with PO₄³⁻ or NO₃⁻. PO₄³⁻ may act as a scavenger of •OH and the resulting radicals or products which were unable to oxidize DCF. Chen et al. (2018) also found that the degradation of 2,4-dichlorophenol (2,4-DCP) was much faster in Na₂SO₄ than in the NaNO₃ medium. The presence of SO₄²⁻ led to the formation of the sulfate radical (SO₄^•–) by direct electron transfer or oxidation by •OH produced during the electrochemical process. The presence of •OH and SO₄^•– in the electrolysis process with BDD anodes by electron spin-resonance spectroscopy experiments using 5,5-dimethyl-1-pyrroline-1-oxide as a trapping agent has been confirmed (Chen et al. 2018). SO₄^•– has a high standard oxidation–reduction potential (2.6 V versus normal hydrogen electrode) and a longer half-life (Avetta et al. 2015; Luo et al. 2019). Aniline-based pharmaceuticals (DCF and sulfamethoxazole) can quickly react with the SO₄^•– with second-order kinetic rate constants in the 9–10 × 10⁹ M⁻¹ s⁻¹ range (Ahmed et al. 2012).

To further clarify the contributions of different radical species, quenching experiments were also carried out. MeOH and TBA are widely used to distinguish the contribution of •OH and SO₄^•–. The kinetics constants that MeOH reacts with •OH and SO₄^•– were 9.7 × 10⁸ M⁻¹s⁻¹ and 3.2 × 10⁶ M⁻¹s⁻¹, respectively (Eibenberger et al. 1978; Buxton et al. 1988). For TBA, the kinetics constants were 3.8–7.6 × 10⁸ M⁻¹s⁻¹ (•OH) and 4–9.1 × 10⁵ M⁻¹s⁻¹(SO₄^•–) (Buxton et al. 1988; Neta et al. 1988). The constant of TBA reacting with •OH was almost three orders of a magnitude higher than that with SO₄^•–. Therefore, MeOH served as an effective quencher for both •OH and SO₄^•–, while TBA served as an effective quencher for •OH.

The total ion current chromatogram is shown in Figure S2 at different reaction times (0, 15, 30, and 40 min) of the SPE-concentrated sample from DCF degradation by BDD electrolysis process. Figure S2 (0 min) shows the retention time of DCF as 14.4 min. Figures S3 and S4 shows the mass spectrum and proposed structures of DCF and its main products detected by HPLC-ESI-qTOF-MS (ESI+). Eight TPs were identified, namely TP341, TP330, TP328, TP314, TP310 (two isomers), and TP266 (two isomers). The natural isotopic density of ³⁵Cl and ³⁷Cl satisfied the 3:1 ratio, which can help to identify the number of chlorine atoms in the TPs. The isotopic peak relative intensity of DCF (m/z 296), TP310, TP328, TP341, and TP266 show a ratio 9:6:1, which verifies the presence of two chlorine atoms. Moreover, the isotopic peak relative density of ³⁵Cl and ³⁷Cl of TP330 and TP314 satisfied the ratio 27:27:9:1, demonstrating the presence of three chlorine atoms in these TPs. The production of TP328 was also detected in the oxidation of DCF by aqueous chlorine dioxide (Wang et al. 2014, 2015b). Given there is no extra chlorine atom addition in the electrolysis process, the emergence of TPs containing three chlorine atoms suggests dechlorination and chlorine oxidation processes in the BDD electrolysis. The m/z of TP341 was 45 Da more than DCF, possibly caused by the substituent of –NO₂ on the benzene ring, which was also detected in DCF abiotic and biotic processes in enriched nitrifying sludge (Wu et al. 2019). The substituent of –NO₂ may suggest the release of N from DCF cleavage because there was no extra N addition in this process. Two types of TP310 with different retention times suggests two different replacement positions of O on the benzene ring, which have also been detected in other research (Ahmed et al. 2012; Nie et al. 2014; Loos et al. 2018). The m/z of TP328 were 32 Da (2 −OH) more than DCF, and three different retention times indicate the presence of three isomers of different substituent modes of −OH on the benzene ring. The same phenomenon was also present in TP341, TP310, and TP266. According to the mass spectrum information and other reference research, the DCF degradation pathway in BDD electrolysis proposed is shown in Figure 3. According to the degradation products detected by liquid chromatography–mass spectrometry, hydroxylation, and decarboxylation are main pathways in DCF degradation, and dechlorination, chlorination, and nitro substitution were also part of this process.

The kinetic constants and removal rate of DCF degradation with different initial pH values are listed in Table 2. Figure 4(a) shows the effects of the initial pH on DCF degradation. DCF degradation was accelerated with the increase of initial pH values when pH values were between 6 and 10. The kinetics constants were 0.1087 min⁻¹ (pH 10) > 0.0766 min⁻¹ (pH 9) > 0.0507 min⁻¹ (pH 8)> 0.0429 min⁻¹ (pH 7) > 0.0384 min⁻¹ (pH 6). When the initial pH was 5, the kinetics constant was 0.0964 min⁻¹ (between the constants at pH 9 and pH 10). The removal of DCF after a 30 min reaction period reflected the same tendency as 99.8% (pH 10) > 93.6% (pH 5) > 90.2% (pH 9) > 77.7% (pH 8) > 73.0% (pH 7) > 68.3% (pH 6). DCF degradation was faster in an acidic or alkaline environment than in a neutral one, and the optimal pH for degradation was pH 10.

The effect of electrolyte concentration on DCF degradation is shown in Figure 4(b). The degradation of DCF accelerates with the increase of electrolyte concentration (10 mmol L⁻¹ to 50 mmol L⁻¹). Table 2 lists the kinetics constants as 0.0633 min⁻¹ (50 mmol L⁻¹) > 0.0429 min⁻¹ (30 mmol L⁻¹) > 0.0306 min⁻¹ (20 mmol L⁻¹) > 0.0166 min⁻¹ (10 mmol L⁻¹). The removal of DCF after a 30 min reaction reflects the same tendency as 88.0% (50 mmol L⁻¹) > 73.0% (30 mmol L⁻¹) > 57.5% (20 mmol L⁻¹) > 37.9% (10 mmol L⁻¹). The increase of electrolyte concentration increases the conductivity, as well as promotes the production of SO₄^•–, which accelerates the electrodegradation of DCF. Chen et al. (2018) also found that along with the SO₄²⁻ concentration increasing from 0.01 to 0.2 M, the apparent rate constants of 2,4-DCP electrolysis increased from 0.48 to 1.38 h⁻¹.

The effect of current density on DCF degradation is shown in Figure 4(c). The increase of current density promotes the degradation of DCF (20 mA cm⁻² to 100 mA cm⁻²). The kinetic constants are 0.1965 min⁻¹ (100 mA cm⁻²) > 0.0429 min⁻¹ (50 mA cm⁻²) > 0.0108 min⁻¹ (30 mA cm⁻²) > 0.0090 min⁻¹ (20 mA cm⁻²) (Table 2). After a 30 min reaction, the DCF degradation reached 73.0% (50 mA cm⁻²), 27.6% (30 mA cm⁻²), and 22.6% (20 mA cm⁻²) and was completely removed when the current density was 100 mA cm⁻².

The increase of current density could improve the generation of radicals via electron transfer. This phenomenon has also been found in electrochemical degradation processes of losartan (Salazar et al. 2016), sulfamethazine (Barhoumi et al. 2016; Sopaj et al. 2016), ifosfamide (Fabiańska et al. 2015), and nitrobenzene (Rabaaoui et al. 2013). The increase of current density resulted in a higher concentration of •OH in the bulk solution (Komtchou et al. 2017). In addition, the BDD anode can extend the range of current densities and significantly raise the most optimal current density compared to Pt and DSA anodes due to its very high O₂ evolution over potential and weak adsorption of BDD(•OH) (Sopaj et al. 2016). A significant temperature increase (24.6 °C) was observed after 30 min of electrolysis when the current density was set at 100 mA cm⁻², which was more than for other temperature changes (+2.3 °C (20 mA cm⁻²), +2.9 °C (30 mA cm⁻²), and +9.1 ºC (50 mA cm–2)). Some researchers also found that the higher temperature significantly enhances the removal rate of contaminants, including DCF, by accelerating the diffusion and transfer of radicals, contaminants, and byproducts in electrolysis (Coledam et al. 2016; Liu et al. 2018; Loos et al. 2018) and the ClO₂ oxidation process (Wang et al. 2015b). In this paper, the increase of temperature at high current density may further promote the degradation of DCF.

Figure 4(d) and Table 2 show the effects of electrode spacing on DCF degradation. Notably, degradation kinetic constants were 0.0339 min⁻¹ (0.5 cm), 0.0678 min⁻¹ (1 cm), and 0.0429 min⁻¹ (2 cm), and the removal rates of DCF after 30 min reaction were 66.3% (0.5 cm), 89.5% (1 cm), and 73.0% (2 cm). The best spacing was 1 cm, which suggests that spacing electrodes too close together may inhibit the diffusion of free radicals and DCF in the bulk solution, and influence the following reaction.

HCO₃⁻, Cl⁻, and HA are typical dissolved substances in water which may interface with the free radical oxidation process. Figure 5(a) shows the effect of HCO₃⁻ on DCF degradation. The addition of HCO₃⁻ significantly inhibits its degradation as the kinetic constants are reduced from 0.0507 min⁻¹ to 0.0312 min⁻¹(0.1 mmol L⁻¹) and 0.161 min⁻¹ (1 mmol L⁻¹) (Table 3). The removal rates reduced from 77.7% to 59.6% (0.1 mmol L⁻¹) and 38.9% (1 mmol L⁻¹). The addition of HCO₃⁻ resulted in the consumption of •OH to produce •CO₃⁻, which acts as a selective oxidation species, but with a much lower reaction rate constant than that of SO₄^•– and •OH for the oxidation process of organic pollutants (Orellana-García et al. 2015; Barazesh et al. 2016; Xiao et al. 2016; Fu et al. 2019). The reaction rate constants of •CO₃⁻ (a one-electron oxidant, E₀ = 1.78 V at pH 7) with several typical organic compounds are in the range of 10⁶–10⁷ M^–1s^–1 (e.g. atrazine, fluometuron, tertbutryn, and diurona) and thus, on average, slower compared to 10⁸–10¹⁰ M^–1s^–1 for SO₄^•– and/or •OH (Arciprete et al. 2012; Luo et al. 2019). HCO₃⁻ therefore mainly scavenges SO₄^•– and •OH, resulting in a decreased oxidation rate of organic compounds (Fabiańska et al. 2015), as well as DCF. Li et al. (2020c) also found that HCO₃⁻ showed an inhibitory effect on electrocatalytic hydrodechlorination of DCF using polypyrrole-reduced graphene oxide (PPy-rGO) modified Pd-Ni bimetallic electrodes. When 5 mmol L⁻¹ of HCO₃⁻ were added, the kinetic constant and removal rate were lower than with 0.1 mmol L⁻¹ of HCO₃⁻ added but higher than with 1 mmol L⁻¹ of HCO₃⁻. This addition of 5 mmol L⁻¹ HCO₃⁻ also increased conductivity, which could promote the electrolysis process. The weakened inhibition of 5 mmol L⁻¹ HCO₃⁻ may therefore be a combination of these two different effects.

The addition of HA significantly inhibited the removal of DCF (Figure 5(b), Table 3). The kinetic constants were 0.0330 min⁻¹(0.1 mg L⁻¹), 0.0187 min⁻¹ (1 mg L⁻¹), and 0.0170 min⁻¹ (5 mg L⁻¹). The DCF removal rates reduced to 61.7% (0.1 mg L⁻¹), 42.2% (1 mg L⁻¹), and 38.0% (5 mg L⁻¹). Generally, natural organic matter (NOM) reacts with free radicals and inhibits the degradation of contaminants (Barazesh et al. 2016; Xiao et al. 2016; Ji et al. 2018; Li et al. 2020b). Ahn et al. (2017) experimentally determined the rate constants of Suwanner River NOM with SO₄^•– (k_SO4^•–/NOM = 4.50 × 10⁶ M^–1s^–1) and •OH (k_•OH/NOM = 3.30 × 10⁸ M^–1s^–1). NOM affects the oxidation reaction based on radicals through scavenging radicals and reversing the radicals intermediates to the parent compound (Lutze et al. 2015; Feng et al. 2017). Komtchou et al. (2017) also found that HA competed with the contaminant and its by-products during water treatment. Thus, the inhibited degradation of DCF was expected in this study.

The reusability of the Nb/BDD electrode was also studied. The reusability capacity of the Nb/BDD electrode for DCF degradation was investigated for five consecutive cycles (Figure 7). The degradation capacity remained stable, showing a slightly decrease of DCF removal from 73% to 69% after five cycles. SEM and Raman analyses before and after degradation cycles were carried out and are shown in Figure S6. The BDD film remained dense without obvious damage after five cycles (Figure S6(b)) compared to the film before reaction (Figure S6(a)). Raman spectra after five cycles indicated that the small amount of graphite on the surface was oxidized or removed during the anodizing process, with the absence of the peak at 1,580 cm⁻¹, while the signal for diamond was still strong (Figure S6(c) and (d)), demonstrating good stability of BDD.

The degradation of DCF by BDD anodes have proven a good fit for the pseudo-first-order degradation kinetics. Both acidic and alkaline conditions promoted the DCF degradation. The degradation was also promoted by the increase of electrolyte concentration and current density. HA and HCO₃⁻ significantly inhibited the degradation of DCF, whereas Cl⁻ accelerated the degradation of DCF. •OH, SO₄^•–, and others contributed 76.5%, 6.5%, and 17.0% of the degradation kinetics of DCF, respectively. •OH contributed more than SO₄^•– in this process. Major TPs of DCF were identified. According to the degradation products detected by liquid chromatography–mass spectrometry, hydroxylation, dechlorination, and decarboxylation are the main pathways to DCF degradation, and chlorination and nitro substitution are also included in degradation process.

This work was supported by the Shandong Provincial Natural Science Foundation (No. ZR2017BB037, 2018ZRLD007) and the National Natural Science Foundation of China (No. 31300414 and No. U2006209).

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