Walnut Shell Biomass Triggered Formation of Fe₃C-Biochar Composite for Removal of Diclofenac by Activating Percarbonate

Abstract

Percarbonate (SPC) as a promising substitute for liquid H₂O₂ has many advantages in the application of in situ chemical oxidation (ISCO). Developing efficient, cost effective and environmentally friendly catalysts for SPC activation plays the key role in promoting the development of SPC-based ISCO. Herein, the walnut shell biomass was combined with ferric nitrate for the catalytic synthesis of Fe₃C@biochar composite (Fe₃C@WSB), which demonstrated high efficiency in activating SPC for the removal of diclofenac (DCF). The Fe₃C showed average crystallite size of 32.6 nm and the composite Fe₃C@WSB demonstrated strong adsorptivity. The prepared Fe₃C@WSB could activate both SPC and H₂O₂ with high efficiency at ca. pH 3 with extremely low leaching of iron, while in a weak acidic condition, higher efficiency of DCF removal was obtained in the Fe₃C@WSB/SPC process than in the Fe₃C@WSB/H₂O₂ process. Moreover, the Fe₃C@WSB/SPC and Fe₃C@WSB/H₂O₂ processes did not show significant differences when supplied with varying amounts of catalyst or oxidant, but the Fe₃C@WSB/SPC process exhibited stronger capability in dealing with relatively highly concentrated DCF solution. Based on quenching experiments and electron spin resonance (ESR) analysis, heterogeneous activation of SPC was assumed as the dominant route for DCF degradation, and both the oxidation by radicals, including •OH, •O₂⁻ and CO₃^•−, combined with electron transfer pathway contributed to DCF degradation in the Fe₃C@WSB/SPC process. The cycling experiment results also revealed the stability of Fe₃C@WSB. This work may cast some light on the development of efficient catalysts for the activation of SPC.

1. Introduction

The merits of percarbonate (Na₂CO₃·1.5H₂O₂, SPC) have attracted increasing attention in recent years, which holds great promise for the degradation of refractory organic contaminants. The long-term stability, low price, and anti-explosion property as compared to liquid H₂O₂ are the dominant advantages of the solid state SPC [1]. Additionally, the coexisting carbonate ions can convert •OH radicals into other reactive species with the steady state concentration higher than •OH to compensate the quenching effect induced by CO₃^2–/HCO₃^– [2]. The final decomposition by-products of SPC, including inorganic carbons and H₂O, are highly biocompatible, providing SPC with the potential to serve as the reagent for in situ chemical oxidation (ISCO).

The SPC-based advanced oxidation processes (SPC-AOPs) have emerged as promising alternative technologies in water purification [3]. Normally, the methods that can effectively activate liquid H₂O₂ are always effective for SPC, like Fenton-like oxidation, photo-assisted oxidation, and discharge plasma-involved oxidation [1]. Even more multiple active species, radicals and nonradicals, can be produced by different SPC-activated approaches than in H₂O₂-activated processes. Except for the frequently detected radicals (e.g., •OH, O₂^•−) in traditional Fenton-like processes, carbonate radicals (CO₃^•−) and peroxymononcarbonate (HCO₄^–) were also reported to contribute to the degradation of target pollutants [4].

Among various SPC-AOPs, the Fe-based catalysts, homogeneous or heterogeneous, are the most intensively investigated activating ways due to the environmental friendliness, energy-saving feature and the convenience in ISCO application. The Lv Shuguang Group have conducted a lot of research on homogeneous Fe(II) or Fe(III)-catalyzing SPC processes [3,4]. Different chelating agents were tried to promote the cycling of Fe(II)/Fe(III). The heterogeneous Fe-based SPC-AOPs can relieve the dissolution of iron and increase the sustainability of the catalyst by reclaiming and recycling, especially for the ferromagnetic ones [5]. The Iron carbide (Fe₃C) has attracted increasing attention in recent years due to the multiple iron valence states within Fe₃C, which is beneficial to the internal adjustment of the chemical state of iron via Fe-C bonds [6]. The attempts to apply Fe₃C in the activation of peroxides have been made, e.g., to activate peroxymonosulfate [7], persulfate [8], and hydrogen peroxide [9]. Both radical and nonradical pathways were frequently reported for Fe₃C/peroxides processes, and low leaching of iron species was also mentioned as a distinct advantage for Fe₃C-based catalysts. So far, hardly any studies can be found using Fe₃C as an activator for SPC, and the differences between Fe₃C/H₂O₂ and Fe₃C/SPC processes are also unclear, which is helpful for the establishment of efficient ISCO process.

Moreover, the carbon source for preparing Fe₃C is mainly dominated by some commercial organic compounds, like melamine [10], tannic acid [11], etc. Problems of high cost and environmental pollution may arise from these chemicals in practical application. Using biomass as the green carbon source for synthesizing Fe₃C can not only cut down the production cost but also realize the reuse of solid wastes. Although several kinds of biomass were tried to successfully produce Fe₃C-based composites, such as pig blood [12], soybean straw [13], and pomelo peel [14], the pure phase of Fe₃C was still difficult to obtain due to the complexity of biomass composition. Unlike the organic reagent, the pyrolysis products and pyrolysis behaviors of various biomass were very unpredictable, and therefore the iron–carbon composites containing multi-phases like Fe₃C, iron oxides, and zero-valent iron were the commonly seen products. Therefore, searching for appropriate biomass sources for preparing relatively pure Fe₃C composite is still needed for understanding the exact properties of Fe₃C during the activation of peroxides. Compared to animal biomass, the composition of agricultural and forestry biomass (AFB) was relatively stable, mainly consisting of cellulose, hemicellulose and lignin. Using AFB as the carbon source may provide a promising and repeatable method to prepare relatively pure phase of Fe₃C.

In this study, walnut shells (WSs) were employed as the carbon source to successfully synthesize Fe₃C/biochar composite (Fe₃C@WSB), which was applied in the activation of SPC and H₂O₂ to compare the performance of two processes. Diclofenac (DCF), as a typical endocrine disrupting compound, was applied as the target pollutant in view of its high risk to the ecological environment and the ubiquity in the effluent of wastewater treatment plants [15]. The advantages of SPC were revealed and the properties of Fe₃C in the catalytic process were also examined. The mechanisms of DCF degradation in the Fe₃C@WSB/SPC process were also proposed accordingly.

2. Results and Discussion

2.1. Characterization

2.1.1. XRD of the Synthesized Catalyst

The crystal structure of the prepared catalyst was characterized first by XRD (Figure 1). It was observed that the prepared catalyst had a group of distinct diffraction peaks. One main peak located at ~24° was associated with the (002) plane of the graphitic carbon, which indicated that the walnut shell precursor was converted to biochar with graphitic structure after the pyrolysis treatment. Additionally, a group of strong sharp peaks well indexed to Fe₃C (JCPDS No.85-1317 [16]) were also observed from 35~70°, and hardly any other iron phase could be identified from XRD, suggesting that the composite Fe₃C@WSB was well prepared.

The average crystallite size of Fe₃C@WSB was calculated by Debye–Scherrer formula based on Equation (1) [17], which was calculated as 32.6 nm.

$$\mathrm{D}=\mathrm{K}\mathsf{\lambda }/(\mathrm{B} \cos \mathsf{\theta })$$

(1)

where D is the crystallite size, K is a constant of 0.89, λ is the X-ray wavelength (0.1542 nm), θ is the Bragg diffraction angle, and B is the full width at half maximum.

2.1.2. Morphology and Porosity Characterization

The porous property of Fe₃C@WSB was revealed from the N₂ adsorption–desorption test, with the results shown in Figure 2. The Fe₃C@WSB exhibited a typical type-IV isotherm with an obvious H4 hysteresis loop, indicating that Fe₃C@WSB was classified as a mesoporous material, which could be further confirmed by the pore size distribution curve. Additionally, the BET specific surface area (SSA) and corresponding pore volume of Fe₃C@WSB were determined to be 184.04 m²/g and 0.28 cm³/g, respectively. The promising porosity of the prepared Fe₃C@WSB was favorable for pollutant enrichment and the exposure of abundant active sites during the pollutant removal process.

The morphology of Fe₃C@WSB was investigated by SEM and TEM, with results shown in Figure 3. It was found in Figure 3a that WS-dereived biochar (WSB) had abundant pores within its skeleton, indicating that the produced gaseous substances during the pyrolysis of walnut shell could effectively activate the biochar with rich porosity. After iron salt was hybridized with walnut shell, the obtained Fe₃C@WSB also maintained the porous structure (Figure 3b), which was consistent with the N₂ adsorption/desorption results. It was also observed that the Fe₃C@WSB had much rougher appearance than pristine WSB, which was ascribed to the attachment of Fe₃C particles on the WSB surface. The EDS elemental analysis indicated that Fe₃C@WSB contained the elements of C, O, and Fe with the atomic ratio of 80.6%, 6.2%, and 13.2%, respectively (Figure 3c). Based on the EDS elemental mapping results shown in Figure 3d–f, these three elements were uniformly distributed on Fe₃C@WSB. From the TEM images shown in Figure 3g,h, it was observed that the Fe₃C particles demonstrated spherical shape and were embedded uniformly in the biochar layer.

2.1.3. The Chemical State Analysis

The chemical valence and bonding features of Fe₃C@WSB were further investigated by XPS technique. It was found from the wide-scan XPS spectrum shown in Figure 4a that the C, O, and Fe elements could be detected from Fe₃C@WSB, which was in accordance with the EDS results. The core-level C 1s spectrum shown in Figure 4b could be divided into three peaks with the binding energies located at 284.8, 286.5 and 289.0 eV, which could be assigned to the C-C/C=C (63.89%), C-O (29.96%), and C=O (6.15%) bonding, respectively [18]. The high C-C/C=C bonding content indicated the formation of graphitic carbon in WSB. Furthermore, as seen in Figure 4c, the Fe 2p spectrum contained two main areas which were associated with the Fe 2p3/2 and Fe 2p1/2 bands. Specifically, the Fe 2p3/2 area could be deconvoluted to four peaks at 709.1, 712.5, 713.8, and 718.0 eV, corresponding to the Fe₃C, Fe²⁺, Fe³⁺ species and the satellite peak, respectively [19,20,21]. Meanwhile, the bonding ratios of the above iron species were calculated to be 16.57%, 48.72%, and 34.71%, demonstrating the multi-valence nature of iron species in Fe₃C@WSB.

2.2. DCF Removal Performance of Fe₃C@WSB

As shown from the XRD pattern, a graphitic structure was formed from the walnut biochar, which may be beneficial to the adsorption of target pollutants. Herein, the adsorption performance of Fe₃C@WSB toward DCF was first examined. It was found (Figure 5) that the adsorption extent of DCF onto the Fe₃C@WSB surface was strongly dependent on solution pH. Stronger adsorption was obtained under more acidic conditions, in which more than 80% DCF could be adsorbed within 90 min at pH 3.18, implying the relatively strong adsorptivity of the prepared Fe₃C@WSB. The enhancement of adsorption under lower pH conditions was most likely due the dissociative property of DCF. The pKa value of DCF is 4.15 [22], which means the molecular form DCF dominates at pH < 4.15 and the ionic form DCF dominates at pH > 4.15. Moreover, the hydrophobicity of molecular DCF (logP = 4.51) was much stronger than the ionic form of DCF (logP = 0.7) [23], which easily resulted in the improvement of adsorption.

The catalytic performance of Fe₃C@WSB was compared to the Fe₃C@WSB/SPC and Fe₃C@WSB/H₂O₂ processes under varied pH conditions (Figure 6a,b). It can be seen that under acidic conditions, DCF could be rapidly removed and the removal rate was higher in both processes compared to the adsorption rate, which implied the effective degradation of DCF. Under neutral and alkaline conditions, not significant improvement of DCF removal was observed compared to adsorption alone. The consumption of H₂O₂ in both Fe₃C@WSB/SPC and Fe₃C@WSB/H₂O₂ processes was also measured, which coincided with the removal performance of DCF that obvious consumption of H₂O₂ was only detected in acidic conditions. Generally, the H₂O₂ consumption in Fe₃C@WSB/SPC process was higher than that in the Fe₃C@WSB/H₂O₂ process, which was likely due to the auxiliary activation of H₂O₂ by co-existing HCO₃^–/CO₃^2– to form the HCO₄^– (Equations (2) and (3)) [24]. It was also observed, Figure 6a,b, that in weak acidic conditions (pH₀ ≈ 4), DCF could be completely removed in the Fe₃C@WSB/SPC process within 20 min, but a lag phase appeared in the Fe₃C@WSB/H₂O₂ process. This probably indicated the differences in the mechanisms for DCF degradation in two processes under pH₀ ≈ 4. Specifically, the •OH radicals in both processes were mainly produced around the catalyst surface via heterogeneous activation of H₂O₂ in view of the extremely low amount of iron leaching from Fe₃C@WSB during the catalytic process (Figure 7). Additionally, under pH₀ > 4, more than half the amount of DCF existed in ionic form, which more likely stayed in the aqueous phase far from the solid catalyst. As a result, the produced •OH may have difficulties in contacting a fraction of DCF in the bulk liquid, which may also lead to the recombination of •OH to form H₂O₂ (Equation (4)). However, in the Fe₃C@WSB/SPC process, the •OH radicals were easily converted to CO₃^•− in the presence of carbonate/bicarbonate ions (Equations (5) and (6)). The steady state concentration of CO₃^•− was reported to be four orders of magnitude higher than •OH in advanced oxidation processes [25], and the reaction between CO₃^•− and DCF was also rapid (k = 7.8 × 10⁷ M⁻¹ s⁻¹ [1]), which was probably the reason for the better performance of DCF degradation in the Fe₃C@WSB/SPC process.

Table 1. Comparison of different Fe-based catalysts for the removal of DCF in different studies.

Composite	Pore Feature	DCF Removal Rate (mM∙min−1)	Preparation Conditions	Ref.
Fe3C@WSB	SSA = 184.04 m2 g−1, mesoporous	ca. 0.0250	walnut shell, Fe(NO3)3∙9H2O, 900 °C, 3 h	this study
Fe@biochar	SSA = 175.8 m2 g−1, mesoporous	ca. 3.47 × 10−5	pitch pine sawdust, K2FeO4, 900 °C, 2 h	[26]
Fe0-pencil graphite	SSA = 88.56 m2 g−1, mesoporous	ca. 0.0005	pencil lead graphite, FeCl3∙6H2O, NaBH4	[27]
Fe3O4/MoS2	SSA = 77.1 m2 g−1, mesoporous	ca. 0.0031	(NH4)6Mo7O24·4H2O, CH4N2S, 220 °C, 18 h	[28]
bio-Pd/Fe@Fe3O4	--	ca. 0.0016	Na2PdCl, FeCl3∙6H2O, E. faecalis living cell, FeSO4·7H2O, NaBH4, −50 °C, 48 h	[29]
LaFeO3	SSA = 27.4 m2 g−1, mesoporous	ca. 0.0008	La(NO3)3∙6H2O, Fe(NO3)3∙9H2O, C6H8O7∙H2O, 500 °C, 4 h	[30]
CuO@FexOy	SSA = 8.68 m2 g−1, mesoporous	ca. 0.0156	CuSO4·5H2O, EDTA-2Na, FeS, 450 °C, 3 h	[31]

summarizes various Fe-based composites for the removal (adsorption or degradation) of DCF in different studies. It is clearly seen that the Fe₃C@WSB composite prepared in this study showed significant advantage in removing DCF.

Na₂CO₃·1.5H₂O₂ → Na₂CO₃ + 1.5H₂O₂

(2)

H₂O₂ + HCO₃⁻ ↔ HCO₄⁻ + H₂O

(3)

2•OH → H₂O₂

(4)

CO₃²⁻ + •OH → CO₃^•− + OH

(5)

HCO₃⁻ + •OH → CO₃^•− + H₂O

(6)

2.3. Influence of Different Parameters

The influence of different parameters, including the concentration of oxidant, the dosage of catalyst, and the initial concentration of DCF, was examined in both processes at the optimum pH condition of 3.2 ± 0.1.

It was found, Figure 8a,b, that the degradation of DCF was accelerated by increasing the concentration of SPC or H₂O₂ to some extent, while further increase in the oxidant concentration could decelerate the removal rate of DCF, which implied the dual role of SPC or H₂O₂ in both processes. An optimum concentration was obtained as 1.0 mM and 1.125 mM for SPC and H₂O₂, respectively. An overly high concentration of H₂O₂ may react with the produced •OH radicals to disturb the effective removal of DCF (Equations (7) and (8)) [32]. The effects of catalyst dosage are shown in Figure 8c,d. Similar influencing trends were obtained in both processes that higher DCF removal rate was achieved by increasing the catalyst dosage due to the increased number of active sites. Generally, the Fe₃C@WSB/SPC and Fe₃C@WSB/H₂O₂ processes did not show significant differences when supplied with varying amounts of catalyst or oxidant.

•OH + H₂O₂ → H₂O + HO₂•

(7)

HO₂• + HO₂• → H₂O₂ + O₂

(8)

However, the influence of the initial DCF concentration showed remarkable differences in two processes. Although a similar decreasing trend of the DCF removal rate was obtained as elevating the DCF concentration in both processes, the Fe₃C@WSB/SPC process exhibited stronger capability in dealing with relatively highly concentrated DCF solution. In comparison, 0.08 mM DCF could be completely removed in the Fe₃C@WSB/SPC process within 60 min, while the removal rate was ca. 80% in the Fe₃C@WSB/H₂O₂ process. This was also probably caused by the presence of more abundant CO₃^•− during the activation of SPC. The conversion from •OH to CO₃^•− via reactions 4~5 could relieve the recombination of •OH and increase the contact probability between DCF molecules and radicals, which was beneficial to treating the DCF with higher concentration.

The influence of coexisting inorganic ions (10 mM) was also compared, including Cl⁻, NO₃⁻, H₂PO₄⁻, and SO₄²⁻ (Figure 8g,h). Except for H₂PO₄⁻, the other ions showed weak influence on DCF removal. Significant inhibition was observed in both processes with the addition of H₂PO₄⁻, which may be because H₂PO₄⁻ reacted with •OH to form the weaker H₂PO₄• (Equation (9)) [33].

H₂PO₄⁻ + •OH → H₂PO₄• + OH⁻

(9)

2.4. Activation Mechanism of SPC

In SPC-based AOPs, active species such as •O₂⁻, •OH, ¹O₂, and CO₃^•− are commonly detected. DMPO and TMP were used as the trapping agents for ESR analysis. As shown in Figure 9b,c, obvious signals of •OH and •O₂⁻ could be detected at pH ≈ 3.0 using DMPO as the trapping agent [34,35], but the above signals were not recognizable at pH ≈ 6.0, which confirmed the efficient generation of •OH and •O₂⁻ in the Fe₃C@WSB/SPC process at pH ≈ 3.0. It is worth mentioning that CO₃^•− was not captured in the process, which may be because that the capture of •OH by DMPO in the aqueous phase hindered the evolution of •OH to form CO₃^•− via reactions 4~5 in the ESR analysis. Moreover, hardly any ¹O₂ could be detected using TMP as the trapping agent, which implied that the produced •O₂⁻ (Equation (10)) mainly reacted with other compounds instead of forming ¹O₂ (Equation (11) [36]).

O₂ + e⁻ → •O₂⁻

(10)

•O₂⁻ + 2H⁺ → H₂O + ¹O₂

(11)

To further determine the contribution of these active species to DCF removal, a series of competitive scavenging experiments was conducted (Figure 9a). TBA was used to quench •OH (k_{(•OH + TBA)} = 5.2 × 10¹⁰ M⁻¹s⁻¹ [37]) in the aqueous phase, in which an 11.11% reduction in the removal of DCF was achieved by adding 500 mM TBA. This indicated that •OH in the aqueous phase was involved in the degradation of DCF. The introduction of 10 mM p-BQ to quench •O₂⁻ (k_{(•O2− + p-BQ)} = (0.9 ~ 1.0) × 10⁹ M⁻¹s⁻¹ [38,39]) caused 32.51% inhibition of the DCF removal rate, indicating that •O₂⁻ was also one of the major contributors to DCF removal. Moreover, since a large amount of produced •OH may be surface bound, phenol was applied to quench the surface-bound radicals (k_{(•OH + Phenol)} = 6.0 × 10⁸ M⁻¹s⁻¹ [5,40]. It was observed that the inhibition rate was 20.79% by adding 10 mM phenol, which supported the assumption of the contribution from surface-bound radicals to DCF removal. In addition, although the CO₃^•− radicals could not be detected on ESR, they may exist and contribute to DCF degradation. Therefore, DMA (10 mM) was applied as an effective quencher for CO₃^•− radicals to determine the contribution of CO₃^•− (k_{(CO3•− + DMA)} = 1.8 × 10⁹ M⁻¹s⁻¹ [41]). As observed, the removal of DCF was significantly reduced by 45.52% with the addition of 10 mM DMA, which confirmed that CO₃^•− played an important role in DCF degradation.

In addition to the reactive oxygen species (ROS), other nonradical mechanisms were also investigated. The possibility of high-valent iron formation was examined by using PMSO as the probe, which could be oxidized by high-valent iron via oxygen atom transfer to generate the unique product of PMSO₂ [42]. It was found (Figure 10a) that, in the absence of Fe₃C@WSB, hardly any variation of [PMSO] and [PMSO₂] could be detected, which suggested that SPC could not oxidize PMSO. However, the coexistence of SPC and Fe₃C@WSB (without DCF) could oxidize about 20% PMSO without the formation of PMSO₂, which excluded the presence of high-valent iron. The transformation of PMSO may be caused by other mechanisms (e.g., radical pathway) during the catalytic reaction between Fe₃C@WSB and SPC. The deceleration of the PMSO transformation in the Fe₃C@WSB/SPC process by the addition of DCF was probably due to the competition between PMSO and DCF for the radical species. The electron transfer process was also a commonly seen mechanism for the degradation of organic pollutants in the heterogeneous activation of peroxides. The good conduction characteristics could be disclosed by the electrochemical impedance spectroscopy (EIS) of Fe₃C@WSB (Figure 10b), which showed a small semicircle radius in the high-frequency region and a relatively vertical oblique line in the low-frequency region. Good conductivity and strong adsorptivity are important factors for efficient electron transfer in the catalytic processes. The linear sweep voltammetry (LSV) results demonstrated in Figure 10c showed gradual increase in the response current with the sequential addition of SPC and DCF, which could reveal the existence of electron transfer from DCF to SPC in the Fe₃C@WSB/SPC process. From the I-t curve in Figure 10d, the sudden drop of current could be detected upon the addition of SPC, which also verified the electron transfer between SPC and Fe₃C@WSB. The addition of DCF could result in more significant change of current, indicating that the catalyst could also obtain electrons from DCF.

Based on the above analysis, both radical and nonradical pathways were involved in the degradation of DCF in the Fe₃C@WSB/SPC process.

2.5. The Reuse Performance of Fe₃C@WSB

The reusability of Fe₃C@WSB was evaluated by conducting cycling experiments. Results are shown in Figure 11a, where after four consecutive cycles, the DCF removal rate could still maintain 85% within 60 min, indicating the good reusability of Fe₃C@WSB. The used Fe₃C@WSB was further characterized to examine the possible change in structure. It is observed in Figure 11b that the graphite structure and Fe₃C pattern were still clear after the cycling experiments, which revealed the stability of Fe₃C@WSB in the catalytic reactions. The C 1s and Fe 2p XPS spectra were also analyzed (Figure 11c,d) and it was found that compared to Figure 4, the ratio of Fe₃C slightly decreased from 16.61% to 10.01%, but the relative ratio of Fe²⁺/Fe³⁺ almost kept the same as that of the fresh Fe₃C@WSB, which could suggest the strong stability of the chemical state of iron by the regulation from Fe-C bonds. The relative ratio of different C-related peaks did not demonstrate significant change, which may imply that the functional group of Fe₃C@WSB was not the dominant mechanism for SPC activation.

3. Materials and Methods

3.1. Chemicals

Walnut shells were collected from a food market, Nanjing, China. Iron nitrate nonahydrate (98.5–101.0%), L(+)-ascorbic acid (≥99.7%), sodium sulfate (≥99.0%), sodium nitrate (≥99.0%), sodium dihydrogen phosphate (≥99%), 1,10-phenanthroline monohydrate (≥99.0%), sodium acetate anhydrous (99.0%), sodium thiosulfate pentahydrate (≥99.0%), sodium chloride (99.5%), glacial acetic acid (≥99.5%), sulfuric acid (95–98%), phenol (≥99.0%), sodium hydroxide (≥96.0%), ethanol (≥99.7%) were purchased from Sinopharm Chemical Reagent Co. Ltd., Shanghai, China. Sodium percarbonate (SPC, ≥13% active oxygen), p-benzoquinone (p-BQ), phosphoric acid (85.0–90.0%), 2,2,6,6-tetramethyl-4-piperidone (TMP, ≥98.0%), methyl phenyl sulfoxide (PMSO, 98.0%), methyl phenyl sulfone (PMSO₂, 98.0%), and 5,5-dimethyl-1-pyrroline-N-oxide (DMPO, ≥97.0%) were obtained from Aladdin, Shanghai, China. Diclofenac sodium salt (DCF, ≥98.0%) and tert-butanol (TBA, ≥99.0%) were obtained from Sigma Aldrich, Louis, MO, USA. The N, N-dimethylaniline (DMA, 99.0%) was obtained from Macklin, Shanghai, China. Hydrogen peroxide (H₂O₂, ≥30.0%) was purchased from Nanjing Chemical Reagent Co. Ltd., Nanjing, China.

Ultra-pure water produced by HHitech water purification system was used (HHitech Instruments Co., Ltd, Shanghai, China). The solution pH adjustment was performed using 1.0 M H₂SO₄ or 1.0 M NaOH.

3.2. Synthesis of the Catalyst

The feedstock of walnut shells was washed several times with deionized water and dried at 70 °C in the oven. The dried walnut shells were mechanically pulverized into powders using a stainless steel grinding machine. After sieving from a 100 mesh sieve, the fine powders of walnut shells were obtained, which were mixed with Fe(NO₃)₃∙9H₂O with the mass ratio of 2:1 (biomass:Fe(NO₃)₃∙9H₂O) in ultrapure water. The above mixture was stirred in water bath at 70 °C until a pasty mixture was obtained, which was then transferred into the oven to completely evaporate the water. The obtained yellowish solid was ground into fine powders as the precursor for preparing Fe₃C@WSB. The precursor was pyrolyzed in the tube furnace at 900 °C for 3 h under N₂ atmosphere with an elevated rate of 5 °C/min. The final catalyst was obtained after cooling down to ambient temperature and thorough grinding. When the precursors were pyrolyzed at 900 °C, the species of Fe may have changed as follows: Fe(NO₃)₃ → Fe₂O₃ →Fe₃O₄ → Fe⁰ → Fe₃C, and the reaction equations are shown in Equations (12)–(15) [43,44,45].

3Fe₂O₃ + C → 2Fe₃O₄ + CO↑

(12)

2Fe₂O₃ + 3C → 4Fe⁰ + 3CO₂↑

(13)

Fe₃O₄ + 2C → 3Fe⁰ + 2CO₂↑

(14)

3Fe⁰ + C → Fe₃C

(15)

3.3. Characterization Methods

The morphology of the obtained catalysts was characterized by scanning electron microscopy (SEM, FEI Quanta 200, Hillsboro, OR, USA) equipped with an energy dispersive spectroscopy (EDS) detector. The X-ray diffraction (XRD) patterns were detected on Rigaku Ultima IV X-ray diffractometer with a Cu-Kα radiation source (1.541841 Å). Scanning rate and 2θ angles were set at 10°/min and 20~70°, respectively. Brunauer–Emmett–Teller (BET) specific surface area and pore size distribution were examined by nitrogen (N₂) adsorption–desorption on a Micromeritics ASAP 2020 HD88 instrument (Micromeritics, Norcross, GA, USA). The surface chemical composition and element valence states were investigated using X-ray photoelectron spectroscopy (Kratos Axis Ultra DLD, Tokyo, Japan). The transmission electron microscopy (TEM) characterization was conducted on JEOL-2100 F, JEOL, Tokyo, Japan.

3.4. Experimental Conditions

All experiments were carried out in 250 mL beakers at room temperature under mechanical stirring. In a typical experiment, the catalyst with a final concentration of 0.1 g/L was added and uniformly dispersed in 200 mL deionized water. Another 50 mL solution dissolving the desired amount of DCF stock solution and H₂O₂ (or solid SPC) was prepared separately, named as solution B. The catalytic reaction was initiated immediately by mixing the catalyst mixture and solution B. Unless otherwise stated, the initial concentrations of DCF and H₂O₂ were fixed as 0.02 mM and 1.5 mM, respectively. When SPC was used as the oxidant, 1.0 mM SPC stoichiometrically containing 1.5 mM H₂O₂ was added in solid form, which could be dissolved within 10 s. At certain time intervals, 2.0 mL of reaction solution was withdrawn, which was immediately filtered through a 0.22 μm membrane into vials containing 0.5 mL ethanol for further quantification of DCF.

For experiments needing the adjustment of initial solution pH, the required amount of 1.0 M H₂SO₄ or 1.0 M NaOH was added immediately as the catalyst mixture and solution B was mixed. The required amount of acid or alkaline was predetermined by several trials in preliminary experiments and the pH₀ values denoted the condition at the beginning of the catalytic reaction. To evaluate the reusability of the catalyst, the catalyst was washed five times with ultrapure water after use, filtered and dried at 60 °C for the next cycle.

The batch experiments were conducted in duplicates at least twice until the errors were below 5%, and the average values obtained were used for plotting.

3.5. Analytical Methods

The DCF concentration was analyzed by High-Performance Liquid Chromatography (HPLC; Dionex Ultimate 3000, Thermo Fisher Scientific, Waltham, MA, USA) equipped with a reverse-phase C18 column (250 mm × 4.6 mm × 5.0 μm) and an ultraviolet and visible (UV-Vis) spectrophotometry detector with the detection wavelength at 278 nm. The mobile phase was a mixture of 80/20% (v/v) methanol and 0.1% phosphoric acid water solution at a flow rate of 1.0 mL/min. The concentration of SMX was also analyzed by HPLC with the detection wavelength at 266 nm and the mobile phase mixture of 65/32% (v/v) methanol and 0.1% phosphoric acid water solution. The concentrations of PMSO and PMSO₂ were detected by HPLC at wavelengths of 215 nm. The mobile phase was a mixture of 20/80% (v/v) methanol and 0.1% phosphoric acid water solution. All of the column temperatures were set at 35 °C. The electron spin resonance (ESR) spectra were obtained by Bruker EMX-10/12 device with X-band field scanning (Bruker, Billerica, MA, USA). The concentration of total dissolved iron and dissolved Fe²⁺ in the solution was determined by a spectrophotometric method at 510 nm via forming the complex with 1, 10-phenanthroline, and L(+)-ascorbic acid was added into the sample vials to reduce the Fe³⁺ to Fe²⁺ for further color development [46]. Electrochemical tests were carried out using an electrochemical workstation (CH1660E) with a three-electrode system, including a platinum wire (counter electrode), Ag/AgCl (reference electrode), and prepared Fe₃C@WSB (working electrodes).

4. Conclusions

The pyrolysis process of walnut shell biomass and ferric nitrate could form the stable composite of Fe₃C@WSB, which could be applied as an efficient catalyst for the activation of SPC. The Fe₃C showed average crystallite size of 32.6 nm and the composite Fe₃C@WSB demonstrated strong adsorptivity toward DCF. Acidic conditions were favorable for the Fe₃C@WSB/SPC process. Although the prepared Fe₃C@WSB could activate both SPC and H₂O₂ with high efficiency at ca. pH 3 with extremely low leaching of iron, at weak acidic conditions, higher efficiency of DCF removal was obtained in the Fe₃C@WSB/SPC process than in the Fe₃C@WSB/H₂O₂ process. The variation in catalyst dosage and H₂O₂ (or SPC) concentration made similar influences on DCF removal in the Fe₃C@WSB/SPC and Fe₃C@WSB/H₂O₂ processes. However, the Fe₃C@WSB/SPC process exhibited stronger capability to remove DCF with high concentration. Both the oxidation by radicals, including •OH, •O₂⁻ and CO₃^•−, combined with the electron transfer pathway, contributed to DCF degradation in the Fe₃C@WSB/SPC process. The composite Fe₃C@WSB also demonstrated good reusability and strong stability during the catalytic reactions.