In the past few decades, zero-valent iron (Fe⁰) has been wildly employed for various contaminants removal including halogenated organics, nitro-aromatics, heavy metals, and oxy-anions due to the advantages of low cost, environment-friendly and low redox potential (E⁰ = ‒0.44 eV) [1]. As we all know, there are various iron corrosion products formed during the corrosion of Fe⁰, which can degrade the contaminants via adsorption, reduction, and co-precipitation [2-4]. Up to now, despite the extensive research on Fe⁰, the traditional Fe⁰ still suffering from several challenges including its essential passive layer, narrow working pH, and reactivity loss after long-term operation due to the reaction of Fe⁰ with water and oxygen to form iron (hydr)oxides [3, 5]. However, our previous work found that the degradation efficiencies of pollutants by micron-scale Fe⁰ (mFe⁰) were elevated with the successive treatment [6]. This phenomenon was diverse from the previous consequence in the aging process of Fe⁰.

It's reported that the Fe⁰ particles will form an oxide film in the air, which is mainly composed of the inner shell (magnetite) and the outer shell (hematite) [7-9]. Electrons can transfer freely in the magnetite layer, but the outer layer of hematite has a great hindrance to the transmission of electrons [10]. Besides, long-term batch studies have believed that the primary or secondary mineral precipitation such as magnetite (Fe₃O₄), lepidocrocite (γ-FeOOH) and green rust can sustain the Fe⁰ reactivity, but the hematite (α-Fe₂O₃), goethite (α-FeOOH) and magnetite (γ-Fe₂O₃)) can decrease the performance of Fe° for contaminants removal [11-13]. The various corrosion products have opposite effects on the activity of Fe⁰. Besides, some researchers found that the reaction of Fe⁰ with O₂ can produce reactive oxygen species (ROS) to oxidize both inorganic and organic compounds [14, 15]. Thus, besides structural transformation of Fe⁰, the free radicals produced in the presence of oxygen will certainly make some contribution in long-term reaction. As mentioned above, numerous investigations on the oxidation/passivation and free radicals of the Fe⁰ system were reported. However, there are rare studies on the influence of structural revolution and free radical transformation of Fe⁰ in successive reactions on the pollutant removal process, and the relative mechanism is not always clear.

Therefore, the primary objective of this work is focus on the changes of reactivity, structure and free radicals of mFe⁰ in continuous operation process. Presently, most of the researches were concerning the nanoscale zero-valent iron (nFe⁰). However, we have limited insight into the short or medium-term corrosion processes on mFe⁰ surface nor do we know the key parameters in cyclic experiments. We choose p-nitrophenol (PNP) as the model pollutant because of its highly toxic, carcinogenic, bioaccumulation and easily detected intermediates [16-18]. The purposes of this study are to (ⅰ) explore the degradation of PNP by mFe⁰ in the cyclic process, (ⅱ) reveal the morphological, structural, and compositional evolution of mFe⁰ during the continuous reactions, (ⅲ) clarify the influence concerning to PNP removal by free radicals in mFe⁰ system.

To investigate thoroughly the reactivity and recyclability of mFe⁰ in the consecutive process, the experiment was conducted for 20 cycles (the detailed information of materials and methods is provided in Supporting information). As shown in Fig. 1a, the removal rate of PNP in the whole process can be divided into three periods: (ⅰ) The removal rates increased with the cycle (from the 1^st cycle to the 5^th cycle), (ⅱ) the removal rate gradually and persistently increased and then reached a relative plateau (from the 6^th cycle to the 12^th cycle), (ⅲ) the ability of reduction decreased with the aging time (from the 13^th cycle to the 20^th cycle). In the period (ⅰ), the removal rate of PNP almost displays a gradient increase. In the next stage, the removal rate of PNP maintained at 100.0% after 30 min. However, the removal rate of PNP gradually decreased from 100.0% to 69.0% in the period of (ⅲ). Additionally, to give a clear comparison of the removal performance, the PNP removal kinetics were fitted by the pseudo-first-order kinetic reaction (Fig. S1 in Supporting information). The kinetic analysis indicates that the degradation efficiencies of PNP with different cycles all were described by the pseudo-first-order and the correlation of determinations (R²) were all greater than 0.95, as shown in Fig. S1 (Supporting information). It was observed that the k_obs for PNP removal of the 5^th cycle (0.1779 min^-1) was 6.74-fold greater than the first time (0.0264 min^-1) in Figs. S1a and S2 (Supporting information). This result was mainly attributed that the pre-existing oxidation film of mFe⁰ was destructed in the process of contact with water. The core of mFe⁰ was exposed and more active sites were exposed, which enhanced the corrosion effect of mFe⁰ core and promoted the degradation rates of PNP swiftly [10, 19, 20].

Besides, it can be found in Fig. S3 (Supporting information) that the amount of precipitate in the effluent (the mFe⁰ particles were fixed at the bottom of the reactor by a magnet and then take out the effluent sample from the homogeneous phase) decreased gradually with the process of continuous reactions. Interestingly, there was almost no precipitate in the effluent after the 5^th cycle. Furthermore, as shown in Fig. 1b, the concentration of dissolved iron rapidly increased to the highest level for 5 min, but there were little dissolved iron left in the solution after 30 min of treatment in the initial three cycle reactions. As usual, the concentration of total iron was at a high level (Fig. S4 in Supporting information). This phenomenon was perhaps attributed to that the released Fe²⁺ was rapidly oxidized to Fe³⁺ and formed iron oxides (hydroxides) following Eqs.1 and 2. Meanwhile, the rapid increase of pH (Fig. S5 in Supporting information) suppressed the release of iron ions, but the iron oxides and hydroxides continued to grow [21]. Surprisingly, after the former reactions, the concentration of dissolved iron couldn't be detected, and even the total iron concentration of the solution was plummeted (from 73.81 mg/L to 18.85 mg/L), as shown in Figs. 1b and S4. Subsequently, both of the dissolved iron concentrations and the total iron concentration remained at a low level. It can be seen that there is a relationship between the transformation of iron ion and the removal of PNP. In the period (ⅰ), a large number of iron ions were released into the solution in the first 5 min due to the rapid corrosion of the iron core under acidic condition. Then, the released Fe²⁺ was oxidized to oxides dispersed in the solution or deposited on the surface of iron to form an oxidation film. Dissolved iron was almost not detected in the period (ⅱ), the total iron was much lower than that in the period (ⅰ), and the removal rate of PNP reached a peak and remained stable. On the one hand, the formation of oxide film not only increase the absorption capacity of mFe⁰, but also accelerates the electronic transfer from the core to the surface, so the reduction of pollutants is greatly improved [10]. On the other hand, these fluctuations in reactivity connected with the rapid destruction of the original Fe³⁺ oxide film on mFe⁰ in the process of PNP treatment, and the subsequent formation of a new passivating mixed-valence Fe²⁺-Fe³⁺ oxide shell [10, 22, 23]. Also, there was a rapid destruction of outer film during each washing process, which can enhance the reactivity of mFe⁰ in the following cycles. Hence, we assumed that the structure of mFe⁰ changes with the cycle reaction process, which played a significant role in the mass transfer tempo of the pollutants, intermediates, and corrosion products between the mFe⁰ and the liquid phase. When the disruptions of the preexisting oxide film, the corrosion of the iron core were strengthened. Simultaneously, with the release of Fe²⁺ and the consumption of H⁺, the solution pH would increase. As shown in Fig. S5, the solution pH during mFe⁰ reaction exceeded 7.0 from initial pH of 5.6 and then reached 9.6 maximally in the whole 20 cycle reaction processes, which were following the previous results of Fe⁰ catalysts [24, 25]. Furthermore, the reason why the pH of the solution was declined is that the Fe³⁺ was hydrolysis following Eq. (3) and the oxidation rate of Fe²⁺ is higher than the corrosion rate of Fe⁰. Notably, when solution pH increased to the range of 7.0-8.0, the mFe⁰ corrosion would alter their pathway to Eq. (4) where no proton consumption or generation took place, so as the pH kept almost constant [26, 27]. This explained why there were a lot of reddish-brown flocculent precipitates in the initial three cycle reactions of effluent, as shown in Fig. S3.

The aforementioned results reveal that the reactivity of mFe⁰ could be profoundly improved. We speculate that the change of mFe⁰ surface structure and radicals are responsible for the enhancement of the PNP removal rate. Therefore, additional experiments should be performed to investigate the effects of details of surface structure and radicals.

Scanning electron microscope (SEM) and energy dispersive spectrometry (EDS) were utilized to characterize the change of morphology and the weight ratio of elements on the surface of aged mFe⁰ particles including 1^st, 2^nd, 3^rd, 4^th, 5^th, 6^th, 12^th, 13^th, 17^th, 20^th cycles, which had a strong transformation of the removal rate of PNP. The SEM images were shown in Figs. 2 and S6 (Supporting information), the shape of mFe⁰ surface appeared different from each other in the whole aging process. For fresh mFe⁰, the surface was smooth and dense [28, 29]. The surface of the mFe⁰ particle was smooth at low magnification after 30 min of reaction (Fig. 2A), while some flaky substances can be observed on the surface of mFe⁰ at high magnification (Fig. 2a). Then, with the increase of cycles, it can be observed that the amount of submicron scale spherical corrosion products (most of the corrosion products were smaller than 1 μm) were formed on the surface of mFe⁰, and the flaky corrosion products were almost invisible (Figs. 2b-j). It also confirms that the accumulation of oxides on the surface gradually increases and the core-shell structure is formed. Besides, it can be found that the globular materials covered on the surface of mFe⁰, the reaction rate increased significantly, which implied that this oxide layer can promote the removal efficiency of PNP. However, this promotion effect does not last as the growth of surface products. When it reached a certain degree, as shown in Figs. 2G-J, its reactivity declined with the increase of surface deposit. The results indicated that a medium thickness of the shell has a high-level conductivity, but the over-thick oxide shell weakened the reduction capacity. It can be seen from the SEM-EDS results (Fig. S7 and Table S1 in Supporting information) that the weight ratio of Fe and O in mFe⁰ was 15.26% and 84.74% respectively at the end of the first reaction. It indicated that the surface of mFe⁰ was oxidized in the PNP at initial pH 5.6. Table S1 shows that the mass fraction proportion of oxygen was increasing, which indicated that the iron oxides accumulated gradually in the successive cycle processes.

To further confirm the ingredient of these corrosion products, the mFe⁰ particles were further analyzed by X-ray diffraction (XRD) based on SEM-EDS analysis. The peaks at 45° and 65° should be assigned to Fe⁰ [30]. Fig. 3 revealed that the XRD patterns of 1st aged mFe⁰ particles (after the 1st reaction) were consistent with the fresh one and there is no new characteristic peak. However, the result of SEM-EDS shows that amorphous and/or crystalline iron (hydro)oxides generate on the mFe⁰ surface. Hence, the corrosion products were maybe amorphous iron (hydr) oxides [6]. Besides, XRD analysis reveals that characteristic peaks matched with magnetite (Fe₃O₄) became more distinct and stronger with the progress of cycle time. When mixed with water, the mFe⁰ reacted with H₂O/O₂ to form hydroxides or oxyhydroxides as shown in Eqs. 5-7 [31].

As we can see from Fig. 3, the intensity of mFe⁰ peaks gradually weakens, and the peaks of Fe₃O₄ start to emerge and strengthen. The results indicatedthat the formation of Fe₃O₄ in a certainperiod (from the 1^st cycle to 12^th cycle) does enhance the removal rate of PNP. The formation of magnetic layer forms a thin and stable semiconductor film under a certain thickness, which enhances the ability of the internal iron to release electrons to the outside, thereby directly reducing pollutants [10, 32]. Besides, the transformation of poorly crystalline iron oxides to highly crystalline iron oxides due to the electron transfer between Fe²⁺ and Fe³⁺ and the corrosion of internal iron core [33]. Dong et al. also reported that after 60 days of aging, the magnetite and/or maghemite were the dominant corrosion products for the bare nFe⁰ [30]. On the one hand, it's reported that the inverted spinel structure was the main existing form of Fe₃O₄, inwhich Fe²⁺ and the half of Fe³⁺ were in an octahedral oxygen environment, while another half of Fe³⁺ existed in a tetrahedral oxygen environment [31, 34]. In the presence of oxygen, the Fe²⁺ present in Fe₃O₄ was oxidized to Fe³⁺, resulting in the formation of γ-Fe₂O₃, without a change in the crystal structure [31]. Hence, the decrease in the reactivity of mFe⁰ after the 12^th cycle can be attributed to the Fe₃O₄ semiconductor layer on the surface of mFe⁰ gradually transformed to γ-Fe₂O₃ passivation layer, which hindered the electron transfer from the iron core [13]. However, the characteristic peaks of γ-Fe₂O₃ cannot be found in XRD might due to the peaks of γ-Fe₂O₃ and Fe₃O₄ are very close in XRD patterns [35]. On the other hand, the decrease of PNP reduction efficiency attributed to the degree of corrosion, which means that the layer of Fe₃O₄ is too thick to transport electrons from mFe⁰ to PNP. To confirm the interactions between mFe⁰ core and Fe₃O₄ shell, pure Fe₃O₄ particles were used to compare with the mFe⁰ after the 20^th cycle for PNP removal. As shown in Fig. S8 (Supporting information), k_obs of mFe⁰ after the 20^th cycle was 4.12 times than that of Fe₃O₄. Hence, the consequence of the interactions between mFe⁰ and Fe₃O₄ coatings is sensible.

To further identify the composition and chemical state of aged mFe⁰, X-ray photoelectron spectroscopy (XPS) was employed to characterize the aged mFe⁰ particles. The survey scan (Fig. S9 in Supporting information) showed the presence of Fe 2p, O 1s, and C 1s. The C 1s signal at 284.8 eV is used to correct the binding energy so that the acquired spectra are calibrated. High resolution of Fe 2p spectra were used to analyze the different valence states of iron (Fig. S10 in Supporting information). The results illustrated that four peaks could be observed in fresh mFe⁰ particles, the peak at 710.5 and 723.8 eV correspond to Fe²⁺ species, the peaks located at 712.6 and 726.1eV are characteristic of Fe³⁺, and the peak at 706.2eV proves the presence of Fe⁰, which indicates that an oxide film formed on the surface of fresh mFe⁰ [5, 36, 37]. Moreover, the position of the peaks at 718.9eV and 732.9eV aresatellite peaks for Fe³⁺. The similar peak positions corresponding to Fe³⁺ and Fe²⁺ were found in Fe 2p spectra of the aged mFe⁰ samples (Fig. 4), which testifying that the surface of mFe⁰ was covered with oxides film [5, 38, 39]. However, the peak of Fe⁰ almost disappeared in the following XPS analyses. A reasonable explanation is the sampling depth of XPS for metal oxide is less than 10 nm [5, 40]. In addition, to further identify the transformation of iron oxides in cycling processes, the molar ratios of the ferrous iron to the total iron (Fe²⁺/Fe_total) and ferric iron to the total iron (Fe³⁺/Fe_total) were calculated by the fitting peak areas of Fe 2p core level spectra, respectively. Table S2 (Supporting information) showed a general trend that the fraction of Fe³⁺ increased progressively while that of Fe²⁺ declined with the process of continuous reactions. It proves that the bond Fe²⁺ participated in the reduction of PNP. Interestingly, the molar ratios of Fe²⁺/Fe_total for the 4^th (0.3012), 5^th (0.3110), and 6^th (0.3153) are close to the theoretical value of Fe₃O₄ (0.3333) [37]. The molar ratio of Fe²⁺/Fe_total maintained a relative plateau (0.3441-0.3531) from the 7^th to the 11^th cycle, and then gradually decreased to 0.1705 at the end of 20^th cycle. The results further suggest that when the surface composition is closed to Fe₃O₄, it has the most significant effect on PNP removal. In addition, the ratios for Fe²⁺/Fe_total was fluctuation, suggesting that fresh Fe²⁺ continuously be produced during the cycle (Table S2). Therefore, it is considered that there is a cycle of Fe²⁺ and Fe³⁺ on the surface of mFe⁰ [41]. At the same time, the surface Fe²⁺ was gradually oxidized to Fe³⁺ by dissolved oxygen (Eq. 8) [31, 42]. The proportion of Fe³⁺ also began to increase from the 12th cycle, which indirectly confirmed that γ-Fe₂O₃ was generated. Besides, it is reported that 718.9 eV satellite peaks prove the existence of γ-Fe₂O₃ [39].

Besides, High resolution of O 1s spectra can be divided into three peaks (Figs. S11and S12 in Supporting information). The peak at 530.0 eV is assigned to the lattice oxygen (O₂^-), which confirmed the existence of Fe₃O₄ and/or γ-Fe₂O₃ [6, 43]. Another peak at 531.6 eV was interpreted as oxygen in a single bond of the hydroxyl group adsorbed at the surface of mFe⁰ [43]. And the peak at 533.6 eV belongs to the chemically or physically adsorbed water [5, 44-46]. Table S3 (Supporting information) illustrated that the ratio of O₂^-/O_total increased with the cycling numbers. The results indicate that the main corrosion products were Fe₃O₄ and/or γ-Fe₂O₃. Concurrently, the increase of O₂^- also proved that the oxidation products were continually accumulating.

As result verified above, the formation of more crystalline magnetite in partially oxidized mFe⁰ upon aerobic condition. The previous study reported that the oxide film mediates electron transfer from Fe⁰ to the contaminant by acting as a semiconductor [5]. The electron transfer is bound up with the oxidation of iron core, being reflected by its free corrosion potential. The more negative free corrosion potential value corresponds to a higher electron transfer rate [47-49]. To further clarify the promoting effect of oxide shell on electron transfer, we investigated the electron transfer activity of aged mFe⁰ particles by measuring their free corrosion potential with Tafel polarization diagrams. The Tafel plots show that the free corrosion potential of aged mFe⁰ (after 1^st reaction) was ‒0.72 V, indicating that the electron transfer ratio was enhanced compared with the fresh one (‒0.44 V), as shown in Fig. S13 (Supporting information). We continuously measured the corrosion potential of following aged mFe⁰ particles including 1^st, 2^nd, 3^rd, 4^th, 5^th, 6^th, 12^th, 13^th, 17^th and 20^th. With the increase of cycling rounds and the accumulation of Fe₃O₄ on the surface of mFe⁰, the absolute value of free corrosion potentials increases gradually (2^nd: ‒0.78 V, 3^rd: ‒0.79 V, 4^th: ‒0.81 V and 5^th: ‒0.83 V) in the period (ⅰ). During the period (ⅱ), the potential shifted from ‒ 0.82 V to ‒0.80 V, suggesting that the appropriate thickness of the oxide shell can maintain the electron transfer rate at a high level. It indicated that a large number of electrons released from the iron core, which can directly reduce PNP and maintain the cycle of Fe²⁺ and Fe³⁺ on the surface of mFe⁰ through the thin and stable oxidation shell. As we assumed that the electron transfer ratio began to decrease in the period (ⅲ). The reason is that the transformation of γ-Fe₂O₃ hindered the electronic transmission ability. Nevertheless, the free corrosionpotential of aging mFe⁰ was still more negative than ‒0.44 V after 20 cycles. Hence, it is rational to conclude that the formation of Fe₃O₄ on the mFe⁰ surface to form a core-shell structure can enhance the electron transfer efficiency to reduce the PNP, but the overwrapped oxide shell and the transformation of γ-Fe₂O₃ can decline the ability of aged mFe⁰.

To reveal the possible existent effect of ROS on PNP removal by mFe⁰ under the aerobic conditions, we conducted a series of quenching experiments with the addition of scavengers. The tert-butyl alcohol (TBA) and isopropanol (IPA) were carried out to detect the existence of hydroxyl radical (^·OH) [25, 50]. The results were shown in Fig. S14 (Supporting information), which shows different trends as we previously divided the whole process into three periods. In the period (ⅰ) and (ⅱ), the PNP removal rates were slightly depressed in the presence of IPA and TBA, which revealed the little contribution of ^·OH in mFe⁰ system. However, it is observed in the period (ⅲ) that the addition of TBA and IPA could enhance the removal efficiency of PNP by nearly 10%-15%. It is reported that ^·OH yield is closely related to Fe²⁺ species [51]. To make sure that the reaction system produced ^·OH and how ^·OH to changed, ^·OH production and surface oxidation degree of Sca-mFe⁰(the mFe⁰ particles were taken after the quenching experiments with addition of scavengers) were employed to further evaluate the reason for the enhancement of PNP removal efficiency. As shown in Fig. S15 (Supporting information), the accumulated ^·OH is over 29.8 μmol/L at 5 min for all 20 rounds. The trend of ^·OH accumulation is consistent with the variation of solution pH (Fig. S5). Because the original solution was acidic and contained hydrogen ions (H⁺), some H⁺ were adsorbed on the surface of mFe⁰ to form atomic hydrogen (H^·) following Eq. 9, but the generated ^· could be quickly scavenged by dissolved O₂ to produce HO₂^· and H₂O₂ (Eqs. 10-12), which then would convert to ^·OH [25].

Besides, the XRD patterns (Fig. S16 in Supporting information) show that the aged Sca-mFe⁰ particles have the typical diffraction peaks for magnetite and Fe⁰. To further justify the influence of the addition of scavengers to the surface structure of Sca-mFe⁰, ScamFe⁰ particles after 13^th, 17^th, and 20^th cycle were further determined by XPS analysis. The surface Fe²⁺/Fe_total of Sca-mFe⁰ increased from 0.2759, 0.1891 and 0.1705 to 0.3759, 0.3584, and 0.3537 compared to the mFe⁰ after the 13^th, 17^th, and 20^th reactions, respectively (Fig. S17 in Supporting information, Tables S2 and S4). XPS spectra of O 1s also indicate a slight decrease of surface O₂^- compared to that without the addition of TBA and IPA, as shown in Fig. S17 and Table S5 (Supporting information).

Collectively, the reduction effect of mFe⁰ to PNP is enhanced with the addition of TBA and IPA. The rapid quenching of ^·OH shifted the Eq. 12 to the positive direction. Thus, the generation of Fe²⁺ would be promoted by the Eqs. 13, 14 [50]. From the analysis of XRD and XPS, it is considered that the generation rate Fe²⁺ is higher than that of Fe³⁺ in the Sca-mFe⁰ system. At the same time, it is reported that the surface bond Fe²⁺ has a strong reduction ability and can effectively remove the adsorbed contaminants. XPS results also happen to show that the amount of Fe²⁺ bonded on the surface is also increased compared with the components without scavengers. Thus, it is reasonable that the addition of TBA and IPA slowed down the accumulation of iron oxide (e.g., γ-Fe₂O₃) on the outer surface, which kept the oxidation shell in a state of high electron transport capacity. In addition, ^·OH is indeed produced in the whole process of PNP removal by mFe⁰, but it did not play a leading role in the removal of PNP. Generally, because the reaction is exposed to the air, active oxides are produced, which have a small negative feedback effect on the reduction reaction. The addition of TBA and IPA suppresses these active oxides and enhances the reduction of PNP.

According to the above results and analysis, the reduction of PNP is a dynamic process by mFe⁰, under the condition of continuous cycle reaction. Scheme 1 thoroughly illustrates the dynamic reaction mechanisms of the cyclic process. The results for the aforementioned are summarized as four aspects. Firstly, the formation of iron (hydrogen) oxides can increase the specific surface area of these particles, thus improving the connection between particles and solutions [28]. Secondly, the oxide film transfers electrons from mFe⁰ to pollutants in the form of the semiconductor. However, there is a certain relationship between the electron transfer efficiency and the stacking degree of the oxide film. In the short run, the main corrosion production of mFe⁰ is Fe₃O₄. When the accumulation of Fe₃O₄ reaches an intermediate value (relative to the aging process of 20 cycles), suggesting that a thin and stable core-shell structure is formed, the ability of electron transfer could be greatly enhanced, and the cycle of the Fe²⁺ and Fe³⁺ is improved. Otherwise, the excessive accumulation of Fe₃O₄ forms a thicker oxide shell and partial Fe₃O₄ transforms to γ-Fe₂O₃, which weakens the electron transfer efficiency. Further-more, the formation of the outer shell can prevent dissolved Fe²⁺ to release from the iron core to the solution. The scavenger experiments show that the Fe²⁺ increase compared with the experiments without scavengers, and the removal efficiency of PNP is improved. Finally, the presence of ^·OH might be not conducive to PNP reduction. The presence of scavengers can restrain the detrimental effect of ^·OH, and then accelerate the removal rate of PNP by aged mFe⁰.

In this study, we explored the dynamic changes of the reaction activity of mFe⁰ in the cyclic reaction process. Several complementary characterization methods expounded the effects of structural transformation and radical production on the removal of PNP. We found that the structure and composition of mFe⁰ change with the cyclic numbers when exposed to water. The continuous reactions led to the formation of mainly Fe₃O₄ and a little γ-Fe₂O₃. These oxides were uniformly distributed on the surface of mFe⁰ in the form of microspheres to form a core-shell structure, which enhanced the electron transfer rate from the iron core to the outside due to the mediation of the oxidation film. However, the alleviation of the molar ratios of Fe²⁺/Fe_total would decline the removal rate of PNP. Besides, one of the most probable accountabilities for the influence of radical is the consumption of surface reactive Fe²⁺ species. While the surface Fe²⁺/Fe_total of ScamFe⁰ could be recovered in the presence of scavengers, which enhanced the reduction of PNP. Therefore, the reactivity of aged mFe⁰ is a dynamic process with the increasing of reaction cycles. If this process can be regulated reasonably, mFe⁰ will be more prominent in wastewater treatment.

The authors report no declarations of interest.

The authors would like to acknowledge the financial support from China Postdoctoral Science Foundation (No. 2019T120843), and Sichuan Science and Technology Program (No. 2019YJ0091).

Supplementary material related to this article can be found, in the online version, at doi:https://doi.org/10.1016/j.cclet.2020.08.007.