1 Introduction

Biochar, due to its redox activity in contaminant removal in water and soil systems, has attracted considerable interest (Fei et al. 2022; Zhu et al. 2020; Yuan et al. 2017; Klüpfel et al. 2014). Through its electron transfer and mediation capabilities, biochar can facilitate the degradation and transformation of environmental contaminants, thereby influencing their biogeochemical cycling and environmental fate (Yuan et al. 2017; Klüpfel et al. 2014). Generally, phenolic moieties and conjugated π-electrons in biochar, as well as persistent free radicals (PFRs) on biochar, contribute to its electron transfer capability (Cuong et al. 2021; Klüpfel et al. 2014).

PFRs, as a class of organic free radicals on biochar, have great potential for applications in micropollutant transformation in the environment (Fang et al. 2014, 2015; Zhu et al. 2020). PFRs can trigger sequential single-electron transfer to oxidants (i.e., H2O2 and O2) to generate reactive oxygen species (ROS, ·OH, ·O2, etc.) for the degradation of organic pollutants (Fang et al. 2014, 2015). The direct reductive capacity of PFRs has also been demonstrated for p-nitrophenol degradation and Cr(VI) remediation (Zhu et al. 2020; Yang et al. 2016a). For example, PFRs on biochar can reduce Cr(VI) to less toxic Cr(III) through direct single-electron transfer and indirect formation of ·O2 (Xu et al. 2020a; Zhu et al. 2020). Our recent study showed that PFRs on iron-modified biochar can promote Fe(III)/Fe(II) cycling and accelerate the reduction of Cr(VI) to Cr(III) (Hu et al. 2024). The transformation reactions of PFRs to contaminants in these studies were predominantly executed under aerobic conditions. The reactivity of PFRs under anaerobic environments, such as inundated paddy soil, is rarely reported (Fang et al. 2014, 2015; Xu et al. 2020a; Zhu et al. 2020). To date, only two recent studies have focused on the reactivity of PFRs under anaerobic conditions (Guan et al. 2023; Xu et al. 2023). Based on solution color development after a reaction, Guan et al. (2023) reported that biochar-based PFRs could react with tertiary butanol under anaerobic conditions, but no relevant mechanism was investigated. Xu et al. (2023) revealed that PFRs could transfer electrons to H2O to form ·O2, which promoted the abiotic oxidation of tire antioxidants. Therefore, further investigations are needed on the reaction mechanisms of PFRs under anaerobic environments. Anaerobic environments are more prevalent in deep soils or flooded paddy soils, which are commonly contaminated with hexavalent Cr(VI) near urban industrial areas, threatening the safety of agricultural products and human health (Liu et al. 2021). However, the mechanism of Cr(VI) reduction by biochar PFRs under anaerobic conditions is not clear.

Dissolved organic matter (DOM) also contains these “long-lived” PFRs (Jacobs et al. 2012), and the reactivity of PFRs is higher in biochar-derived DOM than in bulk biochar (Chen et al. 2021; Dong et al. 2014; Lian et al. 2020). The PFRs present in DOM can increase the number of contact sites with pollutants (Lian et al. 2020). Here, these PFRs in the DOM are considered to be dissolved PFRs. In addition, lignin in river water DOM and saturated compounds in soil DOM were the main species promoting sulfamethoxazole degradation and arsenic oxidation under aerobic conditions, as determined by Fourier transform ion cyclotron resonance mass spectrometry (FTICR-MS) (Liang et al. 2023; Zeng et al. 2021). In our previous investigation, we focused exclusively on the bulk PFRs in biochar (Hu et al. 2024). However, the key active compounds of DOM for Cr(VI) reduction under anaerobic conditions, especially the essential reductive species in biochar-derived dissolved PFRs, remain undetermined. A combination of excitation-emission matrix (EEM) spectroscopy and ultrahigh-resolution FTICR-MS (Jia et al. 2021; Zeng et al. 2021) could allow the identification of key dissolved PFRs active species involved in Cr(VI) reduction and thus improve our understanding of the redox processes occurring in dissolved PFRs under anaerobic soil conditions.

Therefore, in this study, iron oxide (α-Fe2O3) modified rice husk biochar with high PFRs contents was prepared to investigate the reactivity and major active species of biochar-derived PFRs under anaerobic conditions. This research can help to provide a new perspective on the reductive species released by iron-biochar, and improve the understanding of the molecular reduction mechanism of Cr(VI) by iron-biochar.

2 Materials and methods

2.1 Preparation of biochar and iron-biochar

Rice husk was selected as the biomass. Since in our prior study, it was reported that the content of PFRs was highest for biochar prepared at a 500 °C pyrolysis temperature (Hu et al. 2024), the subsequent iron-modified biochar was synthesized at 500 °C. Nanometer-sized α-Fe2O3 powder was mixed with rice husk (Fe/biomass = 5%, w/w) by using a Pulverisette-6 Planetary Mill (FRITSCH, Germany) for 1 h. The mixed sample was slowly pyrolyzed at 500 °C in a tube furnace (GSL-1100-X, HF-Kejing, China) under a N2 atmosphere for 2 h, yielding MBC. Biochar without added α-Fe2O3 was included as a control sample and designated BC.

2.2 Characterization of biochar and iron-biochar

The C, H, O, N, and S contents, pH, zeta potential, microporosity, surface morphologies, surface elemental compositions (C, O, S, Fe, and Cr), oxygen functional groups, and crystalline iron in BC and MBC were determined as reported in our previous study (Hu et al. 2024) and can be referred to in Additional file 1: Text S1. The dissolved sulfate in biochar was extracted with deionized water (1:20 w/v) for 1 h, and sulfate concentrations were determined by ion chromatography (ICS-5000+, Thermo Fisher Scientific, USA).

2.3 Batch sorption kinetics and isothermal sorption experiments

Anaerobic Cr(VI) sorption kinetic and isothermal sorption experiments were conducted with MBC and BC in a glovebox filled with 3–5% H2-balanced N2. The Cr(VI) solution was first purged with high purity N2 (99.999%) for 2 h to remove dissolved oxygen. The biochar samples were placed in a glovebox for 2 days prior to the experiment for removal of adsorbed O2. Batch Cr(VI) sorption by MBC and BC was performed as described in our previous study (Hu et al. 2024). All sorption experiments were conducted in triplicate. Briefly, 2.0 g of the biochar sample was added to 200 mL of 50 mg L−1 Cr(VI) solution in a 250 mL glass bottle. The pH was adjusted to 3.0 ± 0.1 with the addition of 0.1 M HCl or NaOH solution. No buffer was added for this experiment. Kinetic experiments were performed in the dark at 25 ± 1 °C and 500 rpm. Samples were collected from the solution at different reaction times (0, 0.5, 1, 2, 3, 4, 8, 12, and 24 h) and then filtered through a 0.45 μm membrane (PTFE). At the end of the kinetic experiments, biochar particles were collected, rinsed with deionized water, and freeze-dried prior to X-ray photoelectron spectroscopy (XPS) and electron paramagnetic resonance (EPR) analysis. For subsequent DOM analysis, the filtered solution was collected. Concentrations of total Cr, Cr(VI), Cr(III), total Fe, Fe(II), and Fe(III) in solution were measured and are detailed in Additional file 1: Text S2. Sorption kinetic models of pseudo-first-order (PFO) and pseudo-second-order (PSO)  are provided in Additional file 1: Text S3.

For isothermal sorption experiments, Cr(VI) solutions ranged from 5 to 1000 mg L−1. Cr(VI) solution and biochar were mixed at a ratio of 100:1 in 15 mL centrifuge tubes. Centrifuge tubes were shaken in the dark at 25 ± 1 °C and 250 rpm for 48 h. Total Cr, Cr(VI), total Fe, and Fe(II) concentrations in solution were then determined. Sorption isotherm models of Freundlich and Langmuir are listed in Additional file 1: Text S3. The experiments for aerobic Cr(VI) sorption kinetics and isothermal sorption by MBC and BC were carried out under conditions without oxygen controlling, and the other sorption conditions were consistent with anaerobic experiments.

2.4 EPR and quenching experiments

To quantify the intensity of bulk PFRs, 20 mg of BC or MBC was injected into a micropipette in an EPR instrument (Bruker A300 spectrometer, Germany). The specific parameters were as follows: center field 3510 G, sweep width 300 G, microwave frequency 9.85 GHz, microwave power 19.23 mW, modulation frequency 100 kHz, modulation amplitude 1.0 G, and sweep time 60.42 s (Liang et al. 2020; Zhu et al. 2020). As the dissolved PFRs cannot be isolated from DOM, we characterized the DOM in this study instead of the dissolved PFRs. DOM was collected through a 0.45 μm pre-combusted glass fiber filter membrane (Whatman), freeze-dried, and measured by EPR in subsequent kinetic experiments.

ROS identification was performed using EPR coupled with 5,5-dimethyl-l-pyrroline-N-oxide (DMPO) as a spin-trapping agent to characterize ·O2 and ·OH (Han et al. 2022). 4-Amino-2,2,6,6-tetramethylpiperidine (TEPM) was used to capture 1O2. Briefly, each 10 g L−1 biochar suspension was filtered after 10 min or 30 min dark shock. The filtered solution was then mixed with 100 mM scavenger in a 1:1 volume ratio to obtain 1 mL of mixed solution. Then, 30 μL of the mixed solution was absorbed into a capillary glass tube for the EPR assay.

Moreover, to identify the impacts of ROS on Cr(VI) reduction, different scavengers were used as electron competitors during the quenching tests. ρ-Benzoquinone (BQ, 100 mM) was added to probe ·O2, isopropanol (IPA, 100 mM) and terephthalic acid (TPA, 50 µM) were used to probe ·OH, and furfuryl alcohol (FFA, 100 mM) was selected to act as the scavenger of 1O2 (Zhu et al. 2020, 2022; Liu et al. 2022).

2.5 UVvis absorption spectra and EEM fluorescence spectra

Absorption and fluorescence spectra of biochar-derived DOM before and after Cr(VI) removal were obtained simultaneously using an Aqualog fluorometer (HORIBA Scientific, USA). The detailed UV–vis and fluorescence properties are listed in Additional file 1: Text S4. A total of 40 EEMs from MBC and BC were corrected and decomposed using parallel factor analysis (PARAFAC) to extract components using the drEEM toolbox version 2.0 with MATLAB software (http://models.life.ku.dk/drEEM, last accessed: June 2014) (Murphy et al. 2013), and the details were described in our previous studies (Tang et al. 2020, 2021). A two- to seven-component PARAFAC model was explored in the context of core consistency and residual analysis (Additional file 1: Figs. S1 and S2). A four-component PARAFAC model was identified, which explained 95.8% of the variance and successfully passed the split-half validation test with a split style of S4C6T3 (Additional file 1: Fig. S3), indicating that the model was stable.

2.6 Ultrahigh-resolution electrospray ionization FTICR-MS analysis

Eight DOM samples of MBC and BC were selected for FTICR-MS analysis. These samples were taken before and after 24 h of reaction with Cr(VI) under anaerobic and aerobic conditions. Desalting and concentration of anaerobic DOM were performed, and the samples were kept in a glovebox to prevent O2 exposure until FTICR-MS analysis. The DOM samples were desalted and concentrated using Varian Bond Elute PPL cartridges (1 g per 6 mL) (Lv et al. 2016; Spencer et al. 2014; Dittmar et al. 2008). Our previous study showed that the average extraction efficiency for DOM routinely obtained with this procedure yielded an average value of 71.8% (on a carbon basis, unpublished data). Ultrahigh-resolution mass spectra for the purified DOM fractions were acquired using a solari X XR FTICR-MS (Bruker Daltonik GmbH, Bremen, Germany) equipped with a 9.4 T refrigerated actively shielded superconducting magnet (Bruker Biospin, Wissembourg, France) and a Paracell analyzer cell. Samples were ionized by electrospray ionization (ESI) (Bruker Daltonik GmbH, Bremen, Germany) in negative ion mode. The detection mass range was set to m/z 100–800. Elemental ratios of O/C ≤ 1.2 and H/C ≤ 2.5 were used to further constrain the formula calculations. The O/C ratio, H/C ratio, molecular weight (MW), nominal oxidation state of carbon (NOSC), double bond equivalent (DBE), and modified aromaticity index (AImod) were used to describe the molecular properties of DOM, which are detailed in Additional file 1: Text S5. Seven molecular groups were further delineated by the AImod and H/C cutoffs (Lv et al. 2016; Spencer et al. 2014): (1) saturated compounds (H/C > 2), (2) peptides (2 > H/C ≥ 1.5 and N > 0), (3) unsaturated aliphatic compounds (2 > H/C ≥ 1.5 and N = 0), (4) carbohydrates (H/C ≤ 2, O/C > 0.9), (5) lignin-like/phenolic formulas (AImod ≤ 0.5 and H/C < 1.5), (6) aromatic formulas (0.5 < AImod ≤ 0.66), and (7) condensed aromatic formulas (AImod > 0.66).

3 Results and discussion

3.1 Cr(VI) removal and transformation

Iron modification significantly improved the conversation of Cr(VI). The amount of Cr(VI) removed and the amount of Cr(III) generated by MBC were 1.69–1.76 and 1.49–3.18 times higher than those by BC, respectively (Fig. 1a, b and Additional file 1: Fig. S4a, b). The amount of Cr(VI) removed by MBC was 48.6 mg L−1 within 24 h under anaerobic conditions, much higher than the 34.8 mg L−1 of Cr(VI) removed under aerobic conditions (Fig. 1a). The models for the kinetics of Cr(VI) removal and sorption isotherms showed higher values under anaerobic conditions compared to aerobic conditions (Additional file 1: Tables S1 and S2). Further discussion can be found in Additional file 1: Text S6. However, the amount of Cr(III) produced by MBC was lower under anaerobic conditions (0.72–2.78 mg L−1) than under aerobic conditions (1.31–7.77 mg L−1) (Fig. 1b). These results suggest that O2 has a dual effect on Cr(VI) removal and Cr(III) generation by MBC. Cr(VI) is more efficiently removed under anaerobic conditions, while Cr(III) is predominantly formed under aerobic conditions. This disparity is due to an electron transfer mechanism under aerobic conditions where electrons from PFRs are transferred to O2, producing ·O2 which facilitates Cr(VI) reduction, resulting in a higher concentration of Cr(III). Conversely, aerobic environments harbor an abundance of ROS (such as ·OH and 1O2) with higher redox potentials that have the ability to re-oxidize the reduced Cr(III), resulting in a diminished Cr(VI) removal efficiency compared to anaerobic conditions (a detailed discussion is provided in the subsequent section). In addition, the concentrations of Cr(III) for MBC initially increased rapidly and then decreased with increasing reaction time due to the electrostatic attraction between MBC and Cr(III) (Fig. 1b and Additional file 1: Figs. S5a and S6). However, the decreases in the concentrations of Cr(III) were much greater under aerobic conditions than under anaerobic conditions (Fig. 1b), suggesting the reoxidation of Cr(III) under aerobic conditions.

Fig. 1
figure 1

a Cr(VI) removal kinetics; b Cr(III) generation kinetics; c Cr(VI) adsorption isotherms; and d Cr(VI) removal and reduction efficiencies of MBC

The species of Cr adsorbed on biochar and the changes in surface functional groups of biochar after Cr(VI) removal were investigated using XPS analysis. To avoid interferences from oxygen, sample preparation was performed in an anaerobic glovebox. The samples used for XPS analysis were collected at 24 h from the anaerobic/aerobic kinetics experiments. Approximately 92.2% and 88.9% of the Cr sorbed on MBC was in the form of Cr(III) under anaerobic and aerobic conditions, respectively, greater than that on BC by 89.3% and 76.6% (Fig. 2a, b and Additional file 1: Fig. S7a, b). Given that the quantity of Cr(III) generated in the solution is less than the amount of Cr(VI) removed (Fig. 1a, b), it can be inferred that Cr(VI) is more susceptible to adsorption followed by reduction, as demonstrated by a prior study (Zhong et al. 2018). In addition, the efficiency of Cr(VI) reduction was investigated by estimating the content of Cr(III) adsorbed on the biochar surface and the Cr(III) generated in the sorbent solution. The proportion of Cr(III) adsorbed by biochar was determined by XPS. The Cr(VI) reduction efficiencies on MBC were 89.8% and 62.2% under anaerobic and aerobic conditions, respectively, greater than those on BC by 32.7% and 49.3%, accounting for 92.5% and 89.4% of Cr(VI) removal (Fig. 1d and Additional file 1: Fig. S4d). The above results indicated that a higher reduction efficiency of Cr(VI) was observed on MBC, especially under anaerobic conditions. Additionally, the micropore volume and pore size distribution of MBC were essentially the same as those of BC (Additional file 1: Fig. S5e, f and Table S3), suggesting that the active species on MBC, rather than the micropores, mainly facilitated Cr(VI) transformation, as discussed below.

Fig. 2
figure 2

XPS spectra of MBC. a, b Cr 2p XPS spectra under aerobic and anaerobic conditions; c C1s XPS spectra; and d O1s XPS spectra. OL, OA and OS in panel d represent iron-bound lattice oxygen, surface-adsorbed oxygen, and oxygen banded to the carbon surface, respectively

O-containing functional groups on the biochar surface can adsorb or reduce Cr(VI) via complexation or redox processes (Hu et al. 2024; Fei et al. 2022; Zhang et al. 2020). In the O1s spectra, the peaks at 530.4, 531.6 and 533.5 eV are iron-bound lattice oxygen (OL), surface-adsorbed oxygen (OA), and oxygen banded to carbon surface (OS), respectively (Mian et al. 2018; Yang et al. 2016b). The OA and total O content of MBC after Cr(VI) removal increased from 2.12% to 8.44% and 13.2% to 25.4%, respectively (Fig. 2d and Additional file 1: Table S4), indicating that chromium oxides formed on the MBC surface. In addition, the content of surface reductive C–O groups of MBC decreased by 10.9% after Cr(VI) removal under anaerobic conditions, 4.95 times higher than that of 2.2% under aerobic conditions (Fig. 2c and Additional file 1: Table S4). However, the content of surface oxidative C=O groups also decreased by 0.51% under anaerobic conditions (Fig. 2c and Additional file 1: Table S4). These results suggested that the reduction of Cr(VI) by surface C–O is limited because phenolics are theoretically oxidized to carboxyl groups, leading to an increase in C=O content (Liu et al. 2020; Zhang et al. 2020). Therefore, these reduced C–O groups were probably released into solution, consistent with the higher C–O intensity of MBC-derived DOM observed in the FTIR spectra (Additional file 1: Fig. S5d).

Magnetite (Fe3O4) was successfully loaded onto biochar after α-Fe2O3 doping, as shown by energy dispersive spectrometry (EDS), XPS, FTIR, and XRD analyses (Fig. 3 and Additional file 1: Figs. S5 and S8). Before Cr(VI) removal, Fe(II) (710.7 and 724.3 eV), Fe(III) (712.1 and 725.7 eV), and satellite peaks of Fe(II) (715.1 eV) were observed in the Fe 2p3/2 XPS spectra of MBC (Fig. 3a). Previous studies have reported that Fe(II) atoms can transfer electrons or diffuse from the internal Fe3O4 or Fe3S4 to the iron-based surface of the magnetically modified bulk biochar, thus reducing the adsorbed Cr(VI) to Cr(III) (Zhong et al. 2018; Wang et al. 2020). Electron transfer between solid Fe(II) and adsorbed Cr(VI)  was obvious under aerobic conditions, with the satellite Fe(II) peaks apparently disappearing and the Fe(III) peaks increasing from 42.6% to 63.7% after Cr(VI) removal (Fig. 3b). However, the Fe(III) peaks increased slightly from 42.6% to 46.2% after Cr(VI) removal under anaerobic conditions (Fig. 3c), which may be due to a re-reduction of Fe(III) to Fe(II). In addition, MBC released a small amount of ionic Fe(II) (4.33 mg L−1) in the control group [without Cr(VI)], and its surface Fe content decreased from 1.32% to 0.78% after Cr(VI) removal (Additional file 1: Fig. S9 and Table S4). However, no Fe(II) or Fe(III) was detected in the experimental group [with Cr(VI)]. This indicated that the released Fe(II) was immediately oxidized by Cr(VI), and insoluble hydroxide precipitates may have formed via the hydrolysis of Fe(III) or the mixing of Fe(III) oxide with ions, as reported in our recent study (Hu et al. 2024). The stoichiometric ratio between Cr(VI) and Fe(II) without biochar is 1:3 (Xu et al. 2020b). However, the released concentration of Fe(II) in this study (4.33 mg L−1) was much lower than the concentration of Cr(VI) reduced (44.9 mg L−1). Therefore, other reductive substances were involved in the conversion of Cr(VI), because the Fe(II) loaded or released from MBC in this study was not sufficient to reduce such a large amount of Cr(VI).

Fig. 3
figure 3

Fe 2p XPS spectra of MBC. a Before Cr(VI) adsorption; b after Cr(VI) adsorption under aerobic conditions; c after Cr(VI) adsorption under anaerobic conditions

3.2 Influence of bulk PFRs on Cr(VI) transformation

PFRs with delocalized unpaired electrons can engage in direct reactions with contaminants through single-electron transfer. The concentration and type of these radicals predominantly influence the redox reactivity of biochar (Zhu et al. 2020; Ruan et al. 2019). An increase in the concentration of PFRs enhances the electron transfer capability of biochar (Zhu et al. 2020). PFRs are typically classified into three categories: oxygen-centered radicals, carbon-centered radicals, and oxygenated carbon-centered radicals, based on their spectroscopic splitting factors (g values) (Vejerano et al. 2011). Oxygen-centered radicals are represented by semiquinone radicals (g > 2.0040), oxygenated carbon-centered radicals by phenoxyl radicals (g = 2.0030–2.0040), and carbon-centered radicals by cyclopentadienyls (g < 2.0030). Prior studies have demonstrated that a higher g value of PFRs corresponds to a lower electron donation ability (Ruan et al. 2019; Zhong et al. 2018). For instance, carbon-centered radicals (i.e., cyclopentadienyls) possess a potent electron-donating ability to reduce pollutants [such as Cr(VI)] with high redox potential (Zhong et al. 2018), while oxygen-centered radicals (i.e., semiquinone radicals) can function as reversible electron carriers in redox reactions (Xu et al. 2019). In the present study, the bulk PFRs concentration of MBC increased dramatically by  5.6 times compared to that of unmodified BC (Fig. 4a, d and Table 1), indicating that the addition of transition metals significantly promoted bulk PFRs formation (Fang et al. 2015; Vejerano et al. 2011). The g values of MBC and BC were 2.0035, suggesting a class of carbon–oxygen-centered PFRs (phenoxyl radicals). This finding implies that α-Fe2O3 doping only altered the concentration of the bulk PFRs on biochar but did not modify the type of PFRs, consistent with results from FeCl3-modified biochar in previous studies (Hu et al. 2024; Vejerano et al. 2011). Furthermore, PFRs exhibit properties akin to free electrons and display a high electron donor capacity, as the g value of 2.0035 is close to the g factor of free-electron (2.0023) and carbon–oxygen-centered PFRs have low Mulliken charge (0.150 e) (Zhong et al. 2018).

Fig. 4
figure 4

EPR signals of MBC and BC. Bulk PFRs signals of a, b BC and d, e MBC under aerobic and anaerobic conditions; ·O2 signals of c BC and f MBC under aerobic conditions after 10 min of sharking in the dark

Table 1 The PFRs properties of BC and MBC

The bulk PFRs of MBC and BC showed significantly different EPR signals under anaerobic/aerobic conditions. The concentrations of bulk PFRs for MBC decreased sharply from 3.56 × 1018 spins g−1 to 0.448 × 1018 spins g−1 in systems without Cr(VI) and increased slightly to 0.516 × 1018 spins g−1 after Cr(VI) treatment under anaerobic conditions (Fig. 4e and Table 1), indicating that these bulk PFRs did not directly participate in Cr(VI) reduction. In addition, the g values of MBC in these systems increased from 2.0035 to 2.0045 both in the presence and absence of Cr(VI), indicating that the PFRs changed from carbon–oxygen-centered (phenoxyl radicals) to oxygen-centered (semiquinone radicals) species under anaerobic conditions (Xu et al. 2023). These results suggested that the bulk PFRs on MBC had a lower ability to donate electrons to Cr(VI) under anaerobic conditions. However, the PFRs signals in MBC under aerobic conditions were completely different from those under anaerobic conditions. Under aerobic conditions, the concentration of bulk PFRs for MBC decreased slightly from 3.56 × 1018 spins g−1 to 2.47 × 1018 spins g−1 in systems without Cr(VI). After Cr(VI) removal, the concentration decreased significantly to 0.608 × 1018 spins g−1 (Fig. 4d and Table 1). This suggests that PFRs were consumed by providing electrons to Cr(VI).

In addition to the direct reduction of bulk PFRs, indirect reduction of Cr(VI) can also occur for MBC under aerobic conditions. Superoxide radical (O2) signals were detected in the MBC solutions (Fig. 4f) and disappeared following the addition of BQ (Additional file 1: Fig. S11a). The neutralization of ·O2 with BQ (100 mM) led to a decrease in Cr(VI) removal efficiency from  69.6% to 60.9% within 24 h (Additional file 1: Fig. S11d), suggesting an indirect reduction of Cr(VI) by PFRs transferring electrons to O2 to generate ·O2, consistent with a previous study (Xu et al. 2020a). Moreover, hydroxyl radical (·OH) and singlet oxygen (1O2) signals were also detected in the MBC solutions under aerobic conditions and disappeared following the addition of IPA and FFA, respectively (Additional file 1: Figs. S10 and S11). These ·OH (2.8 eV) and 1O2 (1.88 eV) are capable of re-oxidizing the generated Cr(III) to Cr(VI). However, quenching ·OH with IPA (100 mM) decreased the Cr(VI) removal efficiency from  69.6% to 65.8% (Additional file 1: Fig. S11d), potentially due to the fact that IPA can also neutralize the PFRs, thus resulting in a lower removal of Cr(VI). A prior study reported that PFRs on biochar were easily extracted by polar solvents such as methanol, dichloromethane, and IPA (Fang et al. 2014). Terephthalic acid (TPA, 50 µM) was further used to neutralized ·OH (Liu et al. 2022), and marginally increased the removal of Cr(VI) to 70.9% (Additional file 1: Fig. S11d). Similarly, FFA (100 mM) was used to neutralize 1O2 (Liu et al. 2022), and the Cr(VI) removal efficiency increased to 74.6% (Additional file 1: Fig. S11d). These quenching results indicated that ·OH and 1O2 can re-oxidize Cr(III) under aerobic conditions, resulting in low Cr(VI) removal efficiency. This could explain the phenomenon that the decrease in Cr(III) concentrations was higher under aerobic conditions than under anaerobic conditions. Similar results were observed for BC (Fig. 4a–c). The above results indicated that the bulk PFRs on MBC and BC were directly or indirectly involved in Cr(VI) reduction under aerobic conditions, while the generation of ·OH and 1O2 could also hinder the reduction of Cr(VI). However, the bulk PFRs on MBC and BC had little effect on Cr(VI) reduction under anaerobic conditions.

Organic functional groups containing unpaired electrons (such as PFRs) in biochar could be present in the form of DOM that participated in Cr(VI) reduction (Dong et al. 2014). Evidently, phenoxyl PFRs was detected in DOM derived from MBC under anaerobic conditions (Fig. 4e). Although the concentration of PFRs in the DOM (1.06 × 1018 spins g−1) was lower than bulk biochar (3.56 × 1018 spins g−1), its carbon content (2.97%) was also significantly less than bulk biochar (45.3%) (Table 1 and Additional file 1: Table S3). Given that PFRs are organic radicals, we correlated the PFRs concentration to its carbon content. The PFRs/C in DOM was 35.7 × 1018 spins g−1 C−1 (Table 1), much higher than in bulk MBC (7.86 × 1018 spins g−1 C−1), indicating a superior electron transfer capacity of DOM. Previous research also suggested that the active intermediates in the DOM electron transfer process primarily consist of electron-rich phenoxyl radicals (Remke et al. 2022, 2023). Additionally, compared to bulk PFRs, these dissolved PFRs increased contact sites with contaminants and thus resulted in higher reactivity (Lian et al. 2020). Moreover, humic acids with high aromatic carbon content are more susceptible than fulvic acids to forming these dissolved PFRs (Jacobs et al. 2012). Biochar produced at 500 °C contains a high aromatic carbon content of over 90%, according to our prior studies (Hu et al. 2019, 2021). Therefore, these abundant dissolved PFRs derived from MBC may play an important role in the reduction of Cr(VI). We further analyzed these dissolved PFRs by EEM and FTICR-MS.

3.3 UV–vis and fluorescence evidence of dissolved PFRs involved in Cr(VI) transformation

PFRs in DOM released by biochar can directly transfer electrons to reduce Cr(VI) to Cr(III), resulting in changes in the photochemical properties of the solution (Dong et al. 2014; Chen et al. 2021). Since the dissolved PFRs could not be isolated from the DOM, we characterized the DOM instead of the dissolved PFRs. UV–vis spectroscopy can be used to reflect the chemical composition of DOM, and the Napierian absorption coefficient (m−1) was calculated (detailed in Additional file 1: Text S4). The UV–vis absorbance intensities of the MBC solution in systems with Cr(VI) increased under both anaerobic and aerobic conditions and were much higher than those of systems without Cr(VI) (Fig. 5a, b). Two spectral regions, 250–300 nm and 320–450 nm, showed significant changes in systems with Cr(VI). Similar results were observed for the absorbance of BC-derived DOM (Additional file 1: Fig. S12). It is noted that these two regions were more involved in DOM-Cr(VI) binding interactions (Mu et al. 2022). Generally, the 230–300 nm spectral region is associated with conjugated carboxylic groups, while the 300–450 nm spectral region reflects the absorption characteristics of both phenolic and carboxylic groups (Habibul and Chen 2018).

Fig. 5
figure 5

UV–vis spectra of MBC under a aerobic and b anaerobic conditions. c The relative contributions of the PARAFAC components of MBC

Compared to the significant increase in DOM absorbance at 250–450 nm for MBC under aerobic conditions, the value at 250–450 nm was slightly increased under anaerobic conditions. The low concentration of electron acceptors (Cr(VI), 1.45 mg L−1) in the MBC system under anaerobic conditions can explain the small change in absorbance, as Mu et al. (2022) reported that the absorbance of a solution was highly related to Cr(VI) concentration. In addition, with increasing reaction time, the wavelengths of the highest peaks in the 250–300 nm and 320–450 nm regions of MBC increased from 254 to 272 nm and 347 nm to 371 nm, respectively. These observations suggested that DOM interacted with Cr(VI) to generate compounds with higher degrees of unsaturation or greater amounts of conjugation, which contributed to the absorbance redshift (Mu et al. 2022; Worms et al. 2019). In addition, the formation of DOM-Fe complexes was probably responsible for a slight increase in the absorbance of MBC under anaerobic conditions in the Cr(VI)-free systems (Fig. 5b and Additional file 1: Fig. S13) (Jia et al. 2021; Kim et al. 2019). Bianco et al. (2014) revealed that electron transfer of L-tryptophan , L-tyrosine, and 4-phenoxyphenol occurred after irradiation ≥ 8 h, resulting in an increase in the spectrum above 330–350 nm. Phillips and Smith (2015) reported that the adsorption spectrum of fulvic acid decreased at 250–400 nm after reduction by NaBH4. Therefore, the absorbance spectrum of DOM changes when electron transfer occurs.

Visual and quantitative estimation of DOM components was performed by EEM fluorescence spectroscopy and PARAFAC analysis (Jia et al. 2021; Zhang et al. 2019). The fluorescence spectra of MBC solutions showed significant quenching after Cr(VI) removal (Additional file 1: Figs. S14 and S15), indicating that DOM has a strong capacity to bind Cr(VI) (Mu et al. 2022; Zhang et al. 2021). Four components (referred to as C1, C2, C3, and C4) were identified by PARAFAC analysis (Additional file 1: Fig. S16). The C1, C2, and C3 components (Em/Ex = 284/263 nm, 315/275 nm, 306/263 nm, respectively) were classified as protein-like substances (Coble 2007). The C4 component (Em/Ex = 411/< 250, 302 nm) was characterized as a low molecular weight UVA humic-like component (Stedmon and Markager 2005). In systems without Cr(VI), the relative intensity of the protein-like components (C1–C3) was higher in MBC than in BC (Additional file 1: Fig. S17), suggesting that the addition of α-Fe2O3 changed the composition of DOM (Jia et al. 2021). In systems with Cr(VI), the relative intensities of C1 and C4 for MBC decreased significantly, and those of C1, C2, and C4 for BC dropped markedly after the removal of Cr(VI) (Fig. 5c and Additional file 1: Fig. S17). This suggested that these components tend to bind Cr(VI) and may participate in Cr(VI) reduction. The protein-like components contain reducing groups, such as amino groups and phenolic groups, which allow the reduction of Cr(VI) to Cr(III) (Mu et al. 2022). The absorption and fluorescence spectra of DOM indicated that protein-like and humic-like components of DOM were involved in Cr(VI) reduction, and these components may be the main species of PFRs, including conjugated carboxylic groups and phenolic groups. Subsequent FTICR-MS analysis verified the specific components of these dissolved PFRs.

3.4 Molecular evidence of dissolved PFRs involved in Cr(VI) transformation

The molecular composition of MBC-derived DOM changed significantly under anaerobic conditions, with the average DOM formula changing from C21.1H26.0O5.25N0.31S0.47 to C22.3H28.3O6.79N0.91S0.45 after Cr(VI) removal (Additional file 1: Table S5). In addition, a significant increase in molecular weight (MWw) and double-bond equivalents (DBEw) indicated the formation of molecules of larger sizes and more unsaturated compounds, consistent with the increased absorbance shown in Fig. 5a, b. Three categories were established: the formulas removed from DOM after Cr(VI) removal (F1), the formulas remaining before and after Cr(VI) removal (F2), and newly produced formulas after Cr(VI) removal (F3). The Cw, MWw, DBEw, and modified aromaticity index (AImod,w) values of MBC increased from 22.9 to 29.2, 423 to 541, 9.40 to 13.5, and 0.27 to 0.37 (F1 vs. F3), respectively (Additional file 1: Table S5), indicating that DOM compounds with higher MW and degrees of unsaturation were formed; additionally, a higher aromatic index was obtained. Moreover, the proportions of CHO and CHOS compounds for MBC-derived DOM decreased after Cr(VI) removal, while the proportions of CHON and CHONS compounds increased (Additional file 1: Table S5). These results suggested that CHO and CHOS compounds were mainly involved in Cr(VI) transformation, while CHON and CHONS compounds were likely retained in DOM after Cr(VI) removal.

Reductive soluble sulfur compounds can effectively reduce Cr(VI) to Cr(III) in soil remediation (Yang et al. 2021). Higher proportions of CHOS compounds and dissolvable sulfate content were observed in MBC-derived DOM than in BC-derived DOM, especially under anaerobic conditions (Additional file 1: Fig. S18a and Table S5). These results showed that α-Fe2O3 doping increased the release of S-containing compounds. In addition, the proportions of F1 and F3 for CHOS compounds for MBC were higher than those of CHO compounds (Table 2), suggesting that CHOS compounds were mainly involved in Cr(VI) reduction. Among these S-containing compounds, 31.6% of the CHOS formulas in F1 for MBC showed O/S ratios < 4 (Fig. 6a). In addition, C17H28O3S and C18H30O3S were the dominant molecules in the CHOS compounds of MBC before Cr(VI) removal (Additional file 1: Fig. S18b). The above findings indicated that MBC contained a high proportion of highly reduced-S compounds (Song et al. 2018). After Cr(VI) removal, these highly reduced-S compounds decreased to 27.6% in F3 for MBC, suggesting the oxidation of low O/S CHOS compounds to highly oxygenated sulfides, especially those with O/S ratios above 6 (Fig. 6a). The nominal carbon oxidation state (NOSC) describes the average C oxidation state per formula and is expected to respond to DOM redox changes (Wang et al. 2019; Patel et al. 2021). The NOSCw value of F3 categories for CHOS compounds of MBC under anaerobic conditions was  −0.33, and was higher than   −0.43 of F1 categories, further indicating that the newly formed compounds in DOM were more oxidized as a result of electron donating to Cr(VI) (Table 2).

Table 2 Molecular parameters for CHO and CHOS compounds of MBC-derived DOM
Fig. 6
figure 6

a Classification of CHOS species in the removed formulas (F1) and newly produced formulas (F3) categories of MBC-derived DOM according to the ratios of O/S in their molecules. b Relative changes in compound classes between the MBC-derived DOM before and after reaction with Cr(VI). The relative change was calculated by subtracting the percentage of a compound class in the MBC-derived DOM from the percentage of the compound class in the DOM after reactions with Cr(VI). COO-based Kendrick mass defect (KMD) vs. DBE for the c unsaturated aliphatic compounds and d lignin-like compounds in CHO compounds of MBC-derived DOM

The changes in the composition of the major compound classes were further analyzed by calculating the relative changes in the percentages of the compound classes after the reactions (Fig. 6b and Additional file 1: Fig. S19). Under anaerobic conditions, a significant decrease in the unsaturated aliphatic compounds and lignin-like compounds was observed after Cr(VI) removal by MBC-derived DOM, indicating that these compounds could react with Cr(VI) (Fig. 6b). Unsaturated aliphatic compounds contain a number of unsaturated acids, such as monocarboxylic acids and binary carboxylic acids, which can promote the conversion of Cr(VI) by electron transfer (Xu et al. 2019; Fei et al. 2022). Similarly, lignin-like compounds were oxidized by Cr(VI). Unsaturated and hydroxylated aliphatic carboxylic acids are involved in the first step in lignin oxidation, and the generated conjugated unsaturated (hydroxylated) acids can undergo aromatization and condensation to form condensed aromatic structures (Waggoner et al. 2015; Waggoner and Hatcher 2017). As expected, a significant increase in condensed aromatic compounds was observed in MBC-derived DOM after Cr(VI) removal (Fig. 6b).

A series of Kendrick Mass Defect (KMD) analyses were performed to identify the major active species reducing Cr(VI). According to the KMD definition, formulas with the same KMD value differ only in multiples of the selected mass corresponding to atoms or functional groups, such as O, H2, or COOH (Zeng et al. 2020). As shown in Fig. 6c, d, the regular –COO spacing pattern in unsaturated aliphatic compounds and lignin-like compounds reflected the presence of abundant polycarboxylic, conjugated carboxylic, and/or carboxylic acid ester groups in DOM prior to Cr(VI) removal. However, the amounts of –COO compounds with low DBE values decreased significantly, while some –COO compounds with high DBE values were produced after Cr(VI) removal. These results indicated that most of the molecules were carboxyl-substituted and that decarboxylation may be an important reaction between DOM and Cr(VI). A loss of carboxyl groups can lead to the formation of organic radicals that can sacrifice hydrogen and undergo aromatization and condensation to form condensed aromatic structures (Waggoner et al. 2015; Waggoner and Hatcher 2017), consistent with the increased absorbance at 250–450 nm (Fig. 5). Moreover, single electron transfer reactions of organic molecules with strong reductivity are more favorable under anaerobic conditions due to the lower Eh values of the solution (Additional file 1: Fig. S6).

In contrast, under aerobic conditions, after reaction with Cr(VI), MBC-derived DOM decreased the content of unsaturated aliphatic compounds and condensed aromatic compounds and increased the content of saturated compounds, lignin-like compounds, and aromatic/polyphenol compounds (Fig. 6b). Phenols are expected to promote Cr(VI) reduction because of their strong correlation with electron donor capacity (Lv et al. 2018). In this study, a portion of the condensed hydrocarbons was ionized, and these compounds with high DBE values were more prone to attack by ·OH or 1O2 in solution (Additional file 1: Fig. S10), leading to the formation of lignin-like or aromatic compounds by ring-opening. Westerhoff et al. (1999) first reported that ·OH tends to attack the aromatic carbon structure in DOM. Our recent study also showed that ·OH prefers to react with unsaturated bonds through addition reactions during oxidation (Zhuo et al. 2019). Additionally, aliphatic and saturated compounds can promote 1O2 formation (Zeng et al. 2020), consistent with the higher EPR signals of 1O2 in solution after Cr(VI) removal (Additional file 1: Fig. S10). Moreover, in DOM-only systems where the dissolved organic carbon content  was approximately 8.0 mg L−1, the Cr(VI) removal efficiency by MBC-derived DOM under anaerobic conditions was 3.93 times higher than that under aerobic conditions (Fig. 7a). Furthermore, DOM derived from MBC under anaerobic conditions contained higher S/C ratio and oxygen-rich functional groups (Additional file 1: Table S3 and Fig. 7b, c). These results further demonstrated the high electron transfer capability of dissolved PFRs for Cr(VI) reduction under anaerobic conditions.

Fig. 7
figure 7

a Cr(VI) removal efficiencies for biochar-derived DOM. The initial Cr(VI) concentration was 50 mg L−1. The initial pH was 3.0 ± 0.1. b C1s and c O1s XPS spectra of MBC-derived DOM

3.5 Mechanistic insights

The mechanisms of Cr(VI) transformation by MBC-derived PFRs are shown in Scheme 1. The bulk PFRs on MBC showed little effect on Cr(VI) reduction under anaerobic conditions, as the PFRs changed from a high concentration of phenoxyl radicals with high reducing activity to a low concentration of semiquinone radicals with low reducing activity. The dissolved PFRs containing high contents of phenoxyl radicals released by MBC promoted Cr(VI) reduction. Additional FT-ICR MS analysis showed that low O/S (O/S < 4) reductive S-containing compounds and –COO in the unsaturated aliphatic and lignin-like compounds of the dissolved PFRs dominated the reduction of Cr(VI). In contrast, the bulk PFRs on MBC promoted the reduction of Cr(VI) by direct electron transfer or indirect formation of ·O2 under aerobic conditions. However, the reduced Cr(III) could be re-oxidized due to the presence of large amounts of ROS (·OH and 1O2) in solution.

Scheme 1
scheme 1

Schematic illustration of Cr(VI) reduction pathways by MBC

4 Conclusion

This study highlights the importance of α-Fe2O3-modified biochar in remediating Cr(VI) in anaerobic environments and the potential impact of dissolved PFRs on Cr(VI) reduction. Compared to unmodified biochar, α-Fe2O3 incorporation increased the concentration of phenoxyl radicals  by 5.6 times, thereby enhancing Cr(VI) transformation. In addition, α-Fe2O3-doped biochar was more efficient in reducing Cr(VI) under anaerobic conditions than under aerobic conditions due to the release of dissolved PFRs, which have high reactivity and accessibility. Additional FTICR-MS analysis indicated that reduced S compounds (O/S < 4) and carboxylic acid (–COO) groups in the unsaturated aliphatic and lignin-like compounds in dissolved PFRs were the potential major active species accelerating Cr(VI) reduction under anaerobic conditions. In contrast, under aerobic conditions, these dissolved PFRs promoted the generation of ·OH and 1O2, which in turn re-oxidized the reduced Cr(III). Based on our work, we recommend that the reactivity of dissolved PFRs in iron-biochar be considered in anaerobic environments such as paddy soils. This study provides new insights into the role of dissolved PFRs from iron-modified biochar in promoting Cr(VI) reduction in flooded paddy soils.