Degradation of tris(1-chloro-2-propanyl) phosphate by the synergistic effect of persulfate and zero-valent iron during a mechanochemical process

This study revealed a dual pathway for the degradation of tris(1-chloro-2-propanyl) phosphate (TCPP) by zero-valent iron (ZVI) and persulfate as co-milling agents in a mechanochemical (MC) process. Persulfate was activated with ZVI to degrade TCPP in a planetary ball mill. After milling for 2 h, 96.5% of the TCPP was degraded with the release of 63.16, 50.39, and 42.01% of the Cl−, SO42−, and PO43−, respectively. In the first degradation pathway, persulfate was activated with ZVI to produce hydroxyl (·OH) radicals, and ZVI is oxidized to Fe(II) and Fe(III). A substitution reaction occurred as a result of the attack of ·OH on the P–O–C bonds, leading to the successive breakage of the three P–O–C bonds in TCPP to produce PO43−. In the second pathway, a C–Cl bond in part of the TCPP molecule was oxidized by SO4·− to carbonyl and carboxyl groups. The P–O–C bonds continued to react with ·OH to produce PO43−. Finally, the intermediate organochloride products were further reductively dechlorinated by ZVI. However, the synergistic effect of the oxidation (·OH and SO4·−) and the reduction reaction (ZVI) did not completely degrade TCPP to CO2, resulting in a low mineralization rate (35.87%). Moreover, the intermediate products still showed the toxicities in LD50 and developmental toxicant. In addition, the method was applied for the degradation of TCPP in soil, and high degradations (> 83.83%) were achieved in different types of soils.


Introduction
Brominated flame retardants (BFRs) have been banned in Europe on the grounds of their toxicity (Veen and Boer 2012). They have been replaced by phosphorus flame retardants (PFRs), which are used to prevent the spread of flames after combustion and to delay ignition. PFRs are widely used in the textile and electronics industries, in transportation, and in home improvement materials (Veen and Boer 2012). However, PFRs can be released to the environment via volatilization, product wear, and leakage during production and use. Although PFRs are not listed as persistent organic pollutants (POPs), they are long-lived in the environment (Quintana et al. 2008). As a typical PFRs, tris(1-chloro-2-propanyl) phosphate (TCPP) has a potential toxicity which is similar with BFRs ( Björklund et al. 2004). Thus, TCPP should be removed from the environment as a matter of urgency.
TCPP has been detected in water, air, soil, and organisms (Un-Jung et al. 2017;Wang et al. 2017). The levels of TCPP in surface water range from 0.05 to 10 μg/L, and those in river sediments were up to 165 μg/kg dw. Previous reports have shown that TCPP is both carcinogenic and teratogenic and accumulates in the liver and kidneys, resulting in a decrease of the number of cells and a change in neurotransmitters (Hoffman et al. 2014). Mechanochemical (MC) destruction is an effective and Responsible Editor: Ricardo A. Torres-Palma Speciality: Phosphorus flame retardants; Advanced oxidation process; Ball milling environmentally friendly method of removing chlorinated organic pollutants (Yan et al. 2017). Unlike photocatalytic and microbial degradation (Jurgens et al. 2014;Ruan et al. 2013), MC degradation is inexpensive, rapid, efficient, produces harmless degradation products (Cagnetta et al. 2016), and has been used for the degradation of persistent organic pollutants (Veen and Boer 2012). However, the MC degradation of TCPP has not been reported previously.
In the MC process, chemical reagents are activated by the high mechanical energy in a ball-milling reactor, and the bonds in the target pollutants are destroyed as a result of the change in the morphology and crystal structure of the pollutant (Dubinskaya 2010). TCPP contains P-O-C and C-Cl bonds, which both have important roles in maintaining the stability and toxicity of POPs. However, the use of MC processes to simultaneously degrade pollutants containing both phosphorus and chlorine has not been reported previously. We suggest that the mechanisms of the MC degradation of TCPP are different from those of other POPs.
Metals and metal oxides have been used as co-milling reagents to enhance the efficiency of the MC process in ball-milling (Fan et al. 2018;Yan et al. 2017). As a strong oxide, persulfate is widely used in the degradation of organic matter. It can be activated by heat, alkalis, ultraviolet light, microwave irradiation, and transition metals to form highly active sulfate radicals (SO 4 · − ) that react with organic molecules (Oh et al. 2009;Wei et al. 2016). Also, persulfate can be activated by MC process with NaOH or CaO to degrade organic pollutants (Fan et al. 2020;Yan et al. 2015). Previous studies have shown that zero-valent iron (ZVI), Fe(II), and Fe(III) can activate persulfate (Oh et al. 2009;Wei et al. 2016;Yan et al. 2017). Under the action of free radicals, the C-Cl bonds of organic pollutants are broken, and organic substances gradually become inorganic substances, eventually resulting in mineralization (Cagnetta et al. 2017). In the MC process, zero-valent metals or metal oxides have been used as co-milling reagents with persulfate to enhance the MC degradation of pollutants (Liu et al. 2016a;Wang et al. 2019). However, the degradation of chlorinated organic compounds via an MC process with both ZVI and persulfate as co-milling agents has not yet been reported, and the degradation mechanisms remain unclear. Especially, the synergistic effect of the reduction of ZVI and the oxidation reaction of SO 4 · − in MC process on pollutants degradation need to be further explored.
ZVI and persulfate were used as co-milling reagents to remove TCPP in a planetary ball mill. The objectives of this study were to reveal the mechanism by which a persulfate + Fe + mechanochemical (PS + Fe 0 + MC) process can use activated persulfate to degrade TCPP. The degradation intermediates and final products of TCPP were to be identified. The aim was to determine the pathway for the degradation of TCPP by this MC method to develop a new method for the remediation of chlorinated organic compounds.
Three soils were collected from Liaoning (red soil), Jiangsu (yellow soil), and Heilongjiang (black soil) in China, respectively. All the soils were air-dried for one week ground and passed through a 1-mm-mesh screen. The selected soil was stored at 4 ℃ and used as natural soil.

Ball-milling experiment
The MC experiment was performed in a planetary ball mill (QM-1SP2, Nanjing University Instrument Corp., Nanjing, China). Different ratios of iron powder and persulfate were mixed with the TCPP as co-milling reagents to give different weight ratios. The Fe 0 + persulfate + TCPP mixture weighed 5 g. And 150 g of stainless-steel balls with diameters of 5 and 8 mm were placed into two 500-mL stainless-steel pots. The planetary ball mill was operated at a speed of 300⁓500 rpm, and the direction of rotation changed automatically every 45 min, including a 15 min pause for cooling. All the experiments were performed in triplicate. The values of the mean and standard deviation were calculated by SPSS 19.0 software.

MC degradation of TCPP in soil
A 4.5 g of different types of soil with 200 mg/kg of TCPP, 1 g of iron powder, and 0.25 g of persulfate were mixed and added into the ball milling tank with 30:1 of the ball material mass ratio (BMR). The planetary ball mill was operated at a speed of 400 rpm, and the rotation direction changed automatically every 45 min, including 15-min pause cooling time. All experiments were performed in triplicate.

Determination of TCPP and the degradation product
To determine the concentration of TCPP and the degradation product, 0.05 g of each ground sample was extracted with 10 mL of ethyl acetate and concentrated to 1 mL in a rotary evaporator (Rotavapor R-200, Buchi, Flawil, Switzerland) (Liu et al. 2016b). The detailed extraction method is provided in Supplementary Information (SI). The concentration of TCPP was determined by GC/MS/MS using an Agilent 7000B system equipped with a DB-1 column (30 m × 0.25 mm × 1.00 μm) (Agilent Technologies Inc., USA). The column temperature was increased from 60 to 280℃ (held for 1 min) at 15 ℃ min −1 , and the residence time of the solvent was 5 min. The carrier gas was 99.999% He. The mass spectrometer transmission line and the ion source temperature were both 250 ℃, and the mass scan range was 30 ~ 400.
The degradation products of TCPP were qualitatively determined by liquid chromatography-quadrupole time-of-f light mass spectrometry (LC-QTOF-MS) using a Shimadzu LC 30A + AB Sciex Triple TOF 5600 + system equipped with a Agilent ZORBAX SB-C18 column (100 mm × 2.1 mm × 3.5 μm). The column temperature was 40 ℃. The measurements were performed in the positive ion mode by gradient elution using water (containing 0.1% formic acid) and acetonitrile at a flow rate of 0.3 mL/min. The mass scan range was 50⁓1000 (Yu et al. 2019). After MC treatment, 0.01 g of each sample was dissolved in 20 mL of Milli-Q water. The mixture was stirred with a magnetic stirrer for 20 min at 80 ℃ and then treated ultrasonically for 30 min before filtration through a 0.47 μm fiber filter (Zhang et al. 2011). Cl − , SO 4 2− , and PO 4 3− were determined in the collected filtrate using an ICS900 ion chromatography system (Dionex, USA) equipped with a IonPac AG23 Guard Column (4 × 50 mm). The column temperature was 25 ℃ with a flow rate of 1 mL/min, and current intensity of the anion was 35 mA. The change in ion concentration is calculated by the release rate:

Determination of Cl
where C t is the concentration of the ion at t (min) and C 0 is the concentration of the ion at 0 min.

Instrumental analysis
Different methods of analysis were used to characterize the ball-milling samples collected at different times in the ballmilling experiment to clarify the mechanism of TCPP degradation by the MC process. Raman spectra were recorded using a DXR532 microscopic confocal Raman spectrometer (Thermo Fisher Scientific, USA) with an Ar laser beam at 514.5 nm and a scan range of 400⁓2500 cm −1 (Liu et al. 2016b). The infrared spectra were measured with the KBr disk method using a Thermo Fisher Scientific Nicolet IS5 Fourier transform infrared (FTIR) spectrometer in the range 400⁓4000 cm −1 (Dong et al. 2019).
The crystal structure of the ball-milling sample was analyzed with an Ultima IV X-ray diffraction (XRD) spectrometer (Rigaku Electric Co., Ltd, Japan) in the range 2θ = 10⁓90° at a step size of 0.02° (Sui et al. 2017). The elemental distribution and micro-morphology of the ballmilling samples were analyzed by field-emission scanning electron microscopy using an FEI Quanta 400 FEG (Quanta, USA). The changes in the valence states of the elements in the ball-milling samples were analyzed by X-ray photoelectron spectrometry (XPS) using a ESCALAB 250Xi spectrometer (Thermo Fisher Scientific, USA), and the binding energy scale was calibrated by the C1s peak at 284.8 eV (Dong et al. 2019). The active species and free radical were determined by an electron spin resonance (ESR) using Bruker ESR A320 (Germany). After ball-milling with Fe, PS, and soil, the powder was washed with distilled water, and the radical capture experiment was conducted using the filtrate.

Statistical analysis
All the data were mean ± standard deviation values from three different experiments obtained in triplicate. An analysis of variance (ANOVA) was used to test the significance of the results using SPSS 19.0 software, and p < 0.05 was considered to be statistically significant.

Degradation of TCPP under different conditions in the PS + Fe 0 + MC treatment process
The degradation of TCPP by the PS + Fe 0 + MC process was studied, and the efficiency of degradation of TCPP by this method was compared with degradation by Fe 0 + persulfate and a mechanochemical process + persulfate. Figure 1 shows that the PS + Fe 0 + MC process can degrade TCPP effectively. The material ratio (the mass of co-milling regent to the mass of the target pollutant) is one of the most important factors affecting the MC method (Liu et al. 2016b). Figure 1a shows that the degradation efficiency of TCPP increased as the material ratio increased when the ratio of ZVI to persulfate was 4:1. When the material ratio reached 20:1 and 30:1, the degradation efficiencies of TCPP 1 and TCPP 2 were > 90% after ball-milling for 2 h. When the material ratio was 30:1, the degradation efficiencies of TCPP 1 and TCPP 2 were as high as 97.10 and 98.17%, respectively. However, when the material ratio was 10:1, the degradation efficiencies of TCPP 1 and TCPP 2 were only 77.76 and 70.01%, respectively. These results show that as the material ratio increases, an increase in the ZVI and persulfate content increases the degradation efficiency of TCPP. ZVI provides electrons during the ball-milling process, which activate persulfate to generate free radicals, which, in turn, attack the chlorine atoms in pollutants and accelerate the process of dechlorination.
The BMR is another important factor affecting MC degradation. Three BMRs (10:1, 20:1, and 30:1) were selected. Figure 1b shows that the efficiencies of the MC degradation of TCPP increased with an increase in the BMR. When the BMR was 30:1, the degradation efficiencies of TCPP 1 and TCPP 2 were 93.78 and 94.30%, respectively, but the degradation efficiencies of TCPP 1 and TCPP 2 were only 81.24 and 62.87% when the BMR was 10:1. This is because the collision energy increases as the mass of the milling balls increases, which favors the degradation of pollutants. The removal of pollutants is also related to the collision power. The collision power includes the collision energy and the collision frequency (Abdellaoui and Gaffet 1995). The collision frequency in the ball mill tank increases as the number of milling balls increases. However, too many milling balls reduce the space available for the movement of balls in the ball-milling tank, which decreases both the collision speed and energy (Sui et al. 2017). Figure 1c shows the effect of milling speed on the degradation of TCPP and indicates that the degradation efficiency increases with an increase in the milling speed. The degradation efficiencies of TCPP 1 and TCPP 2 reached 93.79 and 94.30%, respectively, after 1 h of milling time at 500 rpm. The kinetic energy in a planetary ball mill is transferred during the ball-milling process, and the speed determines the total energy input from the ball mill (Abdellaoui and Gaffet 1995). As the rotation speed increases, the collision speed and frequency of the milling ball and the target pollutant increase, which results in a high mechanical energy of the mixture in the tank, thereby converting it into chemical energy. The MC reaction is therefore accelerated, resulting in the rapid degradation of the pollutant. The removal efficiency of TCPP was > 97% at 400 and 500 rpm after 2 h of milling time. Figure 1d shows the degradation of 0.16 g TCPP with persulfate, ZVI, or a mixture of ZVI and persulfate as the co-milling regents (4.84 g). After milling for 2 h, the Fig. 1 Degradation efficiency of TCPP by the PS + Fe 0 + MC process under different a material ratios (ratio of ZVI and persulfate 4:1, with ball-milling for 2 h), b ball to material ratios, and c milling speeds. d Comparison with different co-milling reagents. Different letters in (a), (b), and (c) indicate a significant difference at p < 0.05 degradation efficiencies of TCPP 1 were 20.21, 80.15, and 95.50% in the MC + persulfate (MC + PS), MC + Fe 0 , and PS + Fe 0 + MC experiments, respectively. The degradation efficiencies of TCPP 2 were 20.21, 75.93, and 96.50% in the MC + PS, MC + Fe 0 , and PS + Fe 0 + MC experiments, respectively. The degradation efficiency of TCPP did not change when persulfate alone was used as the co-milling reagent. The degradation efficiency increased with an increase in the milling time for the Fe 0 + MC and the PS + Fe 0 + MC systems. After 45 min of ball-milling, the PS + Fe 0 + MC system had the highest degradation efficiency of TCPP. In the initial phase of reaction (within 45 min), when only Fe 0 is used for degradation system, TCCP 1 and TCCP 2 display a better degradation efficiency than that of Fe 0 + PS. The result implies that there are two different degradation mechanisms for Fe 0 and Fe 0 + PS. When only Fe 0 is applied, TCCP 1 and TCCP 2 as the electron acceptors conduct dechlorinate; when Fe 0 and PS are coexisting, Fe0 reacts with PS due to the acidity of PS. After being consumed, Fe 0 continues to participate in the degradation reaction.
A combination of ZVI and persulfate has been used previously as an advanced oxidation method to remove organic pollutants from water (Oh et al. 2009;Wei et al. 2016). As a strong oxidant, persulfate can be activated by ultraviolet light, heating, or transition metals to generate sulfate radicals that can oxidize and decompose pollutants. Metals can produce lattice defects after ball-milling and then generate a large number of electrons, which favors electron transfer in the dechlorination process (Cagnetta et al. 2017). Therefore, the Fe(II) produced by the electron transfer of ZVI can activate persulfate to generate SO 4 · − to promote the removal of TCPP.
The particle shape and size of the co-milling agents also can affect the efficiency of the degradation of pollutants. In this study, small particles were adsorbed onto the surface of large particles and large agglomerates formed, suggesting that the large particles would break and form a new surface after ball-milling (Fig. S3). The particle size of the co-milling agents decreased with increasing milling time (Fig. S4). After milling for 60 min, the number of particles > 45 μm decreased. After milling for 120 min, particles > 45 μm disappeared completely, showing that the degradation of TCPP continued over time (Wang et al. 2020). The smaller the particle size and the larger the specific surface area of ZVI, the stronger the reaction activity. Small size was beneficial for ZVI to stimulate PS to produce SO 4 · − and reduce organic pollutants.

Dechlorination and mineralization of TCPP
The molecular structure of TCPP contains a phosphoric acid center and three chlorinated alkyl chains. When the free radicals generated during the ball-milling process attack the TCPP molecule, Cl − and PO 4 3− are released, and S 2 O 8 2− is converted to SO 4 2− . Figure 2 shows that the release efficiencies of Cl − , SO 4 2− , and PO 4 3− increased as the ball-milling time increased. The release of SO 4 2− was higher than that of Cl − and PO 4 3− until a milling time of 90 min. After ballmilling for 2 h, the release efficiencies of Cl − , SO 4 2− , and PO 4 3− were 63.16, 50.39, and 42.01%, respectively. TCPP can be dechlorinated to phosphorus-containing organic species, the organic group of which is removed to produce PO 4 3− . In addition, PO 4 3− is not completely released, indicating that the phosphoric acid center of TCPP is difficult to destroy completely. When compared with the high degradation efficiency of TCPP, the dechlorination of TCPP is not complete, which indicates that a large number of chlorine-containing degradation intermediates are produced during the MC degradation of TCPP. Figure 2 also shows that the removal efficiency of total organic carbon only reaches 35.87%, indicating that TCPP is not completely mineralized, consistent with the results of Liu et al. (2017). According to the literature, PS-based advanced oxidation processes a strong oxidizing ability, while having a poor mineralization capacity, which can be attributed to the inherent property of SO 4 · − . By increasing the PS dosage could only slight improve the mineralization rate, and excess SO 4 2− generation could result in secondary pollution.

Characterization of milled samples
FTIR, XRD, and Raman analyses were carried out on the ball-milled samples to clarify the mechanism of MC degradation of TCPP. FTIR spectrometry was used to determine the changes in the characteristic FTIR peaks of the functional groups in the TCPP molecules during ball-milling (Table S2).

Fig. 2
Ion release rate and removal rate of total organic carbon from ball-milled samples at different times Figure 3a shows the XRD spectra of the milled samples for different milling times. After milling for 30 min, the characteristic peak for persulfate weakened, indicating that PS was consumed or decomposed in during the ball-milling process. After milling for 2 h, the Fe peak was greatly increased and the persulfate peak completely disappeared, indicating that the increased crystallinity of Fe activated persulfate, which has a major role in the degradation of TCPP in the ball-milling process. In addition, the Fe peak was broadened with the increasing milling time, indicating that the distortion of Fe lattice increased, which provided more free electrons for the redox process in the degradation of TCPP (Sui et al. 2017). Figure 3b shows that the C-Cl peak at 695 cm −1 decreased with an increase in the ball-milling time, indicating that the Cl atom in TCPP was gradually removed to form Cl − . The peaks at 1049, 1281, and 1428 cm −1 , which represent the P-O-C and C-C bonds and the CH 2 group, respectively, decreased with an increase in the milling time (Yan et al. 2017;Zhang et al. 2020), which shows that the chloropropane group linked to the ether bond was removed from TCPP. The fracture of the C-C bond means that the chloropropane side-chain was broken down to form smaller molecules. The peaks at about 1100 cm −1 , characteristic of the P-O bond, increased after ball-milling (Siow et al. 2014), which indicates that the number of phosphoric acid groups increased. The peak at 1642 cm −1 related to the C = O bond appeared during ball-milling, which means that carbonyl groups formed in the product.
As the milling time increased, the color of the MC degradation product deepened, and the product turned completely black after milling for 2 h (Fig. S5), suggesting that the samples were carbonized, which was confirmed by the Raman spectra (Fig. 3c). The peak of the initial TCPP was unclear after 1 h of ball-milling. After milling for 2 h, two peaks at appeared at the D-band (1330⁓1380 cm −1 ) and the G-band (1540⁓1580 cm −1 ), which correspond to the typical Raman spectra of amorphous carbon and graphite, respectively (Wakayama et al. 1999). The appearance of the D and G peaks in the Raman spectra suggests that TCPP was carbonized to carbon materials with defective and disordered structures (Xu et al. 2013). The sequence of peaks D and G showed that the carbonization process first generated amorphous carbon and then part of the carbon product was converted to graphite. Zhang et al. (2011) reported that pentachloronitrobenzene can be converted to amorphous carbon and graphite during milling with Fe. The intensity of the D and G peaks was not high, which showed that only part of the TCPP was mineralized through carbonization. This shows that it is difficult to fully mineralized TCPP and that mineralization was not the only pathway used to degrade TCPP.
ESR was used to identify the active species in the PS + Fe 0 + MC system for selected samples ball-milled for  Figure 3d shows that the characteristic peaks of ·OH and SO 4 · − appeared, indicating that ·OH and SO 4 · − were produced during the ball-milling process (Yin et al. 2020). This result confirmed that the degradation of TCPP by the MC process was a result of oxidation by ·OH and SO 4 · − .

Mechanism of activation of persulfate
XPS was used to analyze the chemical composition and changes in the valence of the elements on the surface of the sample to investigate the degradation mechanism of TCPP. All the XPS peaks were calibrated using the C1s peak at 284.8 eV. Table S3 gives the specific binding energy positions of the fitted peaks of ZVI in the initial sample (Deng et al. 2020). The changes in the peak areas of samples Fe 0 , Fe(II), and Fe(III) at different milling times were compared. Figure 4a shows that the peak area of ZVI decreased with an increase in the milling time, whereas the peak areas of Fe(II) and Fe(III) increased. This suggests that ZVI was oxidized to Fe(II) by persulfate during the ball-milling process with the formation of SO 4 2− (Eq. (1)) (Al-Shamsi and Thomson 2013). Some of the Fe(II) was oxidized further to Fe(III). The persulfate (S 2 O 8 2− ) was converted into SO 4 2− and SO 4 · − (Eq. (2)) (Oh et al. 2009). Some of the Fe(II) reacted with the sulfate radicals generated in Eq. (2) to produce Fe(III) and SO 4 2− (Eq. (3)) (Kishi and Ikeda 1973). However, as shown in Fig. S6, the concentration of the produced Fe(II) was far higher than that of Fe(III), showing that Fe(III) is easy to be reduced to Fe(II) in MC process (Eq. (5)). The results show that the degradation of TCPP by the PS + Fe 0 + MC process was accompanied by the generation of SO 4 2− , consistent with the increase in the release rate of SO 4 2− shown in Fig. 2 (Deng et al. 2020). The generated SO 4 · − in Eq. (2) reacted with water (present in humid air and in additives) to generate hydroxyl radicals (·OH), SO 4 2− , and H + (Eq. (4)). These results all confirm that persulfate can be activated by ZVI to generate SO 4 · − and ·OH, thereby promoting the degradation of TCPP. Finally, the reduction reaction of ZVI could remove chlorine in the micromolecular chlorinated organics, which were the intermediate products during the MC process (Eq. (6)).
(1)  Figure 4b shows the change in valency of chlorine. In the initial stage, the characteristic peaks of the C-Cl bond appeared at about 200.7 and 202.5 eV, and a weak Cl − peak was seen at about 198.2 eV (Moulder et al. 1992) (Table S4). After ball-milling for 1 h, the characteristic peak of the C-Cl bond gradually weakened until it disappeared at 2 h, whereas the characteristic peak of Cl − gradually strengthened. This shows that TCPP was destroyed and the organic chlorine in the TCPP molecule was transformed into inorganic chlorine.

Degradation pathways of TCPP in PS + Fe 0 + MC treatment
The degradation products of TCPP1 were analyzed by LC-QTOF-MS and five intermediate products were detected: C 6 H 13 Cl 2 O 4 P (product A, m/z 251.0); C 3 H 8 ClO 4 P (product B, m/z 175.0); C 9 H 17 Cl 2 O 5 P (product C, m/z 307.0); C 9 H 17 Cl 2 O 6 P (product D, m/z 323.0); and C 6 H 12 Cl 2 O 6 P (product E, m/z 247.0). Table S5 and Fig. S7 show the corresponding information and mass spectra.
Two degradation pathways of TCPP are proposed (Fig. 5). The first pathway is the P-O-C cleavage pathway, which is similar to the oxidative degradation of organic phosphorus compounds (Cheng et al. 2020). The ·OH generated by the activation of persulfate attacks the phosphate center of TCPP, and TCPP is cleaved at the P-O-C bond to form the hydroxylated product A and 1-chloropropane. Product A is oxidized by the same process to form product B, producing a third 1-chloropropane and PO 4 3− . In the second pathway, the C-Cl at the end of the alkyl chain of TCPP is oxidized by SO 4 · − to form a carbonyl compound (product C), which, in turn, is oxidized to produce a carboxylic acid (product D). The C-Cl bond in product D cannot be oxidized further, but the phosphate center in product D is attacked by the same route as the first pathway. Product D is removed from one of the 1-chloropropane molecules to form product E, the P-O-C bonds of which are broken, followed by the formation of propionic acid through substitution, addition, and SO 4 · − cleavage (Yu et al. 2019). The oxidative dechlorination in the second pathway confirms that the role of ZVI is not to reduce TCPP, but to activate persulfate to oxidize TCPP in the PS + Fe 0 + MC system (Li et al. 2021).
The degradation mechanism includes oxidation, dechlorination, hydroxylation, and dealkylation. ·OH and SO 4 · − mainly attack the phosphate center and the C-Cl bond at the end of the alkyl chain through substitution, addition, and a series of electron transfer reactions. TCPP was converted into Cl − , PO 4 3− , and some intermediate products. However, SO 4 · − could not oxidize C-Cl bond in (6) Fe 0 + R − Cl → Fe(II) + Cl − + RH micromolecular organochlorides, while ZVI activated by MC could play the role in reductive dechlorination.

Toxicities of TCPP and its MC degradation intermediates
The Environmental Protection Agency Toxicity Estimation Software Tool (TEST) version 5.1 was used to evaluate the toxicity of TCPP and its degradation intermediates. In the TEST software, the 50% lethal dose (LD 50 ) of oral rats was selected as an index to predict the acute toxicities of the degraded intermediate products. As shown in Fig. 6a, the LD 50 of TCPP was 409.97 mg/kg, which is considered "very toxic." After ball milling, the LD 50 values of some intermediates were converted into "toxic" areas, while others still maintained in the "very toxic" areas. The results show that PS + Fe 0 + MC process reduced the toxicity of LD 50 and TCPP can be converted into some low-toxic products. However, most of intermediates showed developmental toxicant more than that of TCPP (Fig. 6b). Although TCPP could be almost completely degraded via PS + Fe 0 + MC process, the toxicity of the intermediates was not effectively removed because TCPP could not be completely mineralized by MC.

Degradation of TCPP in soils by MC with ZVI and PS
As shown in Fig. 7, after milling for 2 h, the degradation rates of TCPP 1 in red soil, yellow soil, and black soil were 92.13%, 85.76%, and 83.83%, respectively, and the degradation rates of TCPP 2 were 90.80%, 87.38%, and 87.01%, respectively. To clarify the relation between the degradation rates of TCPP and characteristics of soils, the pH, iron contents, and organic matter contents of three soils were detected (Table S6). The results showed that the degradation of TCPP was high under acid conditions. The previous studies have shown that PS is easily activated to generate SO 4 · − and ·OH under low pH conditions, and increasing pH inhibited the formation of SO 4 · − and ·OH. Iron concentration is another important influencing factor (Fan et al. 2018). The degradation of TCPP in red soil was faster than that of the other types of soils because iron concentration in red soil was higher than that of other two soils. The higher the iron concentration in soil, the higher the degradation of TCPP. This is because the presence of iron can promote the production of SO 4 − · from PS to degrade pollutants rapidly. In addition, the order of organic matter contents in different soils was as follows: red soil < yellow soil < black soil. After ball milling for 2 h, the degradation rates of TCPP in the soil were as follows: red soil > yellow soil > black soil. The results showed that the content of organic matter in soils affected negatively the degradation rates, which confirmed the previous research (Fan et al. 2018). In addition, during the first 15 min of MC degradation, the removal efficiencies of TCPP in soil were higher than that of TCPP without soil mixture, which was because that metal oxide in soil might promote the degradation rates of MC to TCPP. Therefore, we suppose that PS + Fe 0 + MC process is a promising method for the removal of organic pollutants in soil.

Conclusions
This study shows that the PS + Fe 0 + MC process is an effective method for the degradation of TCPP. The degradation efficiency was as high as 96.5% after milling for 2 h with a material ratio of 30:1, a BMR of 30:1, and a milling speed of 400 rpm. The ·OH formed in the combination process attacked the P-O-C bonds in phosphate center of TCPP, meanwhile a C-Cl bond in part of the TCPP molecule was Effect of soil types on TCPP degradation. The concentration of TCPP, 200 mg/kg; mass ratio of soil to PMS, 18:1; BMR, 30:1; rotation speed, 400 rpm oxidized by SO 4 · − to carbonyl and carboxyl groups, resulting in dechlorination via oxidation of the C-Cl bond. Finally, ZVI reductively dechlorinated the intermediate organochloride products, resulting in the mineralization of TCPP due to the synergistic effect of the reduction of ZVI and the oxidation reaction of ·OH and SO 4 · − in MC process although LD 50 and developmental toxicant of the intermediate products were not completely removed. These results show that persulfate can be used in the MC degradation of organic pollutants; moreover, pollutants in soils can be degraded via the MC methods. The PS + Fe 0 + MC system is an efficient and environmentally friendly method that is widely used to remove organic phosphorus pollutants.
Author contribution W. Qiao and Q. Yang designed the work, performed MC degradation and sampling, and performed all parameters' measurements used in this project. Y. Qian and Z. Zhang analyzed data and participated in the interpretation of data. W. Qiao and Q. Yang contributed to drafting and critically revising of the paper. All authors gave final approval of the version to be published and agreed to be accountable for all aspects of the work.
Funding This work was financially supported by the Natural Science Foundation of Jiangsu Province (BK20201388), the Industry Prospect and Common Key Technologies in Jiangsu Province (BE2018015), and the Priority Academic Program Development of Jiangsu Higher Education Institutions (PAPD).

Data availability
The datasets used or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations
Ethics approval Not applicable.

Competing interests
The authors declare no competing interests.