Improved Dewaterability of Waste Activated Sludge by Fe(II)-Activated Potassium Periodate Oxidation

Fe(II)-activated potassium periodate (KIO4) oxidation was used to improve the dewaterability of waste-activated sludge for the first time. Compared with those of raw sludge, the capillary suction time (CST), specific resistance filtration (SRF), and water content of filter cake (WC) of sludge treated using the Fe(II)/KIO4 process under the optimal conditions (i.e., the initial pH = 6.8, KIO4 dose = 1.4 mmol/g volatile suspended solids, Fe(II)/KIO4 molar ratio = 1.2) decreased by 64.34%, 84.13%, and 6.69%, respectively. For conditioned sludge flocs, the Zeta potential and particle size were increased, and hydrophilic proteins in extracellular polymeric substances (EPS) were partly degraded, accompanied by the transformation of tightly bound EPS into soluble EPS and the conversion of dense sludge flocs into loose and porous ones. During Fe(II)/KIO4 oxidation, Fe(IV) and the accompanying •OH were determined as the predominant reactive species and the underlying mechanism of sludge EPS degradation was proposed. This work provides a prospective method for conditioning the sludge dewaterability.


Introduction
Wastewater treatment plants produce massive waste activated sludge (WAS) with moisture content higher than 99%. Sludge dewatering is an imperative procedure to reduce the sludge volume and subsequently to save on the cost of sludge transportation and disposal. The efficiency of sludge dewatering is largely dependent on the extracellular polymeric substances (EPS) [1][2][3][4][5][6]. Destroying the EPS was deemed to be capable of improving the sludge dewatering performance. Herein, advanced oxidation processes (AOPs) have received much attention due to their efficiency in degrading EPS [7].
As a new strategy to improve sludge dewaterability, there are two reasons for requiring an in-depth and systematic study. One point is the optimal operational conditions of the Fe(II)/KIO 4 oxidation system. Operational conditions influence the oxidation extent of the EPS, which is directly related to sludge dewatering performance. It is known that  4 ), bovine albumin (BSA), Folin-phenol reagent, ascorbic acid (AA), L-histidine, nitro blue tetrazolium chloride (NBT), phenol, and methanol (MeOH) were obtained from Shanghai Aladdin Biological Technology Co., Ltd. (Shanghai, China). All chemicals were analytically pure and used without further purification.

Experimental Design
Fe(II)/KIO 4 oxidation was designed to improve the dewaterability of WAS. A series of batch tests were carried out in 300 mL beakers. The beaker was continuously stirred in a thermostatic magnetic stirrer at 25 • C and 150 rpm. CST, SRF, and W C were selected as representative indicators for sludge dewatering performance. To explore the effect of the initial pH on the sludge dewatering performance, the initial pH was adjusted to 3.0, 4.0, 5.0, 6.8, 8.0, and 9.0 while keeping the KIO 4 dose (at 1.4 mmol/g VSS) and Fe(II)/KIO 4 molar ratio (1.2) unchanged. Similarly, with the Fe(II)/KIO 4 molar ratio fixed at 1.2 and the initial pH maintained at 6.8, the effect of KIO 4 dose (0.6, 0.8 1.0, 1.2, 1.4, 1.6, and 1.8 mmol/g VSS) on the sludge dewatering performance was investigated. To determine the effect of Fe(II)/KIO 4 molar ratio on the sludge dewaterability, the Fe(II)/KIO 4 molar ratio was set as 0.0, 0.3, 0.6, 0.9, 1.2, 1.5, and 1.8 under the conditions of a KIO 4 dose of 1.4 mmol/g VSS and an initial pH of 6.8. For all experiments, the reaction time lasted for 20 min. All experiments were repeated three times.

Determination of Sludge Dewaterability
SRF and CST are both evaluation indexes of sludge filtration and dewatering performance, while W C is an index reflecting the effectiveness of sludge dewatering. The smaller the values of SRF, CST and W C , the better the dewatering and filtration performance of the sludge. The CST of sludge was measured by CST instrument (DP-MT) equipment. The SRF was measured by the vacuum filtration method: the Buchner funnel was filled with the sludge suspension of 100 mL, and a constant pressure of 0.06 MPa was applied using a vacuum pump. First, start the stopwatch then, when the filtration starts, the corresponding filtration volume in the measuring cylinder is recorded every 10 s. As the filtration speed slows down, the interval time of the record is increased, and the filter volume is recorded every 1 min. The filtration was stopped when there was no filter drop. The vacuum valve needs to be continuously adjusted throughout the experiment to ensure that the experimental pressure is constant. The SRF (m/kg) of the sludge was calculated by Equation (1) [29]: where P (kg/m 2 ) is the pressure applied; A (m 2 ) is the filter area; µ (kg·s/m 2 ) is the KV, ω (kg/m 3 ) denotes dry solid weight per unit volume sludge on the filtrate media; and b is the time-to-filtration ratio, which is the slope of the curve that is obtained by plotting the ratio of the time of filtration to the volume of filtrate (t/V) versus the filtrate volume (V). W C obtained during SRF test was determined by standard methods [30].

Sludge Properties
The pH value was measured with a pH meter (pHs-3C). Particle size and particle size distribution were determined by a laser particle size analyzer (Malvern Master sizer 2000). The Zeta potential of the sludge supernatant was measured using a Malvern potential analyzer (Malvern Zeta sizer Nano ZS90). In addition, the physical surface morphology changes of sludge flocs samples were observed by a scanning electron microscope (SEM) (ZEISS Sigma 300). The surface functional groups of sludge flocs were determined by Fourier transform infrared spectroscopy (FT-IR) (Nicolet IS5, Thermo Fisher Scientific, Waltham, MA, USA).
EPS extraction was conducted according to a thermal extraction method [31]. A concise description is as follows: A 15 mL sludge sample was centrifuged at 4000× g at 4 • C for 15 min. The supernatant was collected and denoted SB-EPS. The pellet was resuspended in 15 mL of 0.05% NaCl using a vortex mixer, heated at 70 • C for 1 min, and subsequently centrifuged at 4000× g at 4 • C for 10 min. The collected supernatant was considered LB-EPS. The residual pellet was resuspended in a 0.05% NaCl solution to its original volume, incubated at 60 • C in a water bath for 30 min, and centrifuged at 4000× g at 4 • C for 15 min to collect TB-EPS. The protein and polysaccharide concentration in EPS were determined using the Lowry method or the phenol-sulfuric acid method [32]. The initial pH value always plays a key role in AOPs, which will affect the sludge dewaterability adjustment by Fe(II) activated KIO 4 oxidation. Figure 1 depicts the effects of the initial pH on the performance indices of sludge dewatering. changes of sludge flocs samples were observed by a scanning electron microscope (SEM) (ZEISS Sigma 300). The surface functional groups of sludge flocs were determined by Fourier transform infrared spectroscopy (FT-IR) (Nicolet IS5, Thermo Fisher Scientific, Waltham, MA, USA).

Results and Discussion
EPS extraction was conducted according to a thermal extraction method [31]. A concise description is as follows: A 15 mL sludge sample was centrifuged at 4000 g at 4 °C for 15 min. The supernatant was collected and denoted SB-EPS. The pellet was resuspended in 15 mL of 0.05% NaCl using a vortex mixer, heated at 70 °C for 1 min, and subsequently centrifuged at 4000× g at 4 °C for 10 min. The collected supernatant was considered LB-EPS. The residual pellet was resuspended in a 0.05% NaCl solution to its original volume, incubated at 60 °C in a water bath for 30 min, and centrifuged at 4000× g at 4 °C for 15 min to collect TB-EPS. The protein and polysaccharide concentration in EPS were determined using the Lowry method or the phenol-sulfuric acid method [32].

Effects of Initial pH on Sludge Dewaterability
The initial pH value always plays a key role in AOPs, which will affect the sludge dewaterability adjustment by Fe(Ⅱ) activated KIO4 oxidation. Figure 1 depicts the effects of the initial pH on the performance indices of sludge dewatering.  As shown in Figure 1a, the CST of the treated sludge received its minimum value (47.55 s) at the initial pH of 6.8, which was the pH of the raw sludge. Similarly, the SRF and Wc of treated sludge with no pH adjustment gained minimum values of 0.99 × 10 12 m/kg (Figure 1b), and 83.75% (Figure 1c), respectively. Interestingly, as the initial pH decreased from 6.8 to 3.0, the CST, SRF, and Wc increased to 101.65 s, 4.05 × 10 12 m/kg, and 84.47%, respectively, which signified deterioration in the sludge dewaterability. As it is known that a lower pH in AOPs always results in a higher production of active species, and, consequently, a higher degradation of organic contaminants. During recent decades, EPS has been reckoned as a key factor in hampering sludge dewatering. Many studies endeavored to degrade EPS as much as possible so as to improve sludge dewaterability. However, in this study, more degradation of EPS at an initial pH lower than 6.8 exhibited negative effects on sludge dewaterability. A plausible explanation is the layered structure of EPS and its hydrophilic/hydrophobic characteristics. As Yu et al. [6] reported, EPS were degraded layer by layer in sludge oxidation, and different EPS layers showed quite different hydrophilic/hydrophobic organic distributions. The surface EPS layer possesses a relatively higher hydrophobicity, whereas the inner EPS layer has a relatively higher hydrophilicity. Excessive destruction of the surface hydrophobic EPS and exposure of the inner hydrophilic EPS elevated the energy barrier among sludge flocs, which increased the bound water content of the sludge and resultantly deteriorated the sludge dewaterability [6,33,34]. While increasing the initial pH from 6.8 to 9.0, the sludge dewatering performance also showed a downward trend. The CST, SRF, and Wc increased to 107.25 s, 3.61 × 10 12 m/kg, and 84.25%, respectively. This can be explained as follows: In alkaline conditions, Fe(II) easily reacts with OH − and reduces the amount of effective activator in the reaction system. Moreover, Fe 2+ and Fe 3+ react easily with OH − to form Fe(OH)2 and Fe(OH)3 colloids, which increase the viscosity of sludge and affect its filtration ability. As a result, sludge dewatering becomes more difficult [32]. To sum up, an initial pH that is too low or too high is not conducive to improving the sludge dewaterability, and the initial pH of 6.8 is optimal. This result implied that there is no need to adjust the initial pH during sludge conditioning by Fe(Ⅱ)/KIO4 oxidation.  As shown in Figure 1a, the CST of the treated sludge received its minimum value (47.55 s) at the initial pH of 6.8, which was the pH of the raw sludge. Similarly, the SRF and W C of treated sludge with no pH adjustment gained minimum values of 0.99 × 10 12 m/kg (Figure 1b), and 83.75% (Figure 1c), respectively. Interestingly, as the initial pH decreased from 6.8 to 3.0, the CST, SRF, and W C increased to 101.65 s, 4.05 × 10 12 m/kg, and 84.47%, respectively, which signified deterioration in the sludge dewaterability. As it is known that a lower pH in AOPs always results in a higher production of active species, and, consequently, a higher degradation of organic contaminants. During recent decades, EPS has been reckoned as a key factor in hampering sludge dewatering. Many studies endeavored to degrade EPS as much as possible so as to improve sludge dewaterability. However, in this study, more degradation of EPS at an initial pH lower than 6.8 exhibited negative effects on sludge dewaterability. A plausible explanation is the layered structure of EPS and its hydrophilic/hydrophobic characteristics. As Yu et al. [6] reported, EPS were degraded layer by layer in sludge oxidation, and different EPS layers showed quite different hydrophilic/hydrophobic organic distributions. The surface EPS layer possesses a relatively higher hydrophobicity, whereas the inner EPS layer has a relatively higher hydrophilicity. Excessive destruction of the surface hydrophobic EPS and exposure of the inner hydrophilic EPS elevated the energy barrier among sludge flocs, which increased the bound water content of the sludge and resultantly deteriorated the sludge dewaterability [6,33,34]. While increasing the initial pH from 6.8 to 9.0, the sludge dewatering performance also showed a downward trend. The CST, SRF, and W C increased to 107.25 s, 3.61 × 10 12 m/kg, and 84.25%, respectively. This can be explained as follows: In alkaline conditions, Fe(II) easily reacts with OH − and reduces the amount of effective activator in the reaction system. Moreover, Fe 2+ and Fe 3+ react easily with OH − to form Fe(OH) 2 and Fe(OH) 3 colloids, which increase the viscosity of sludge and affect its filtration ability. As a result, sludge dewatering becomes more difficult [32]. To sum up, an initial pH that is too low or too high is not conducive to improving the sludge dewaterability, and the initial pH of 6.8 is optimal. This result implied that there is no need to adjust the initial pH during sludge conditioning by Fe(II)/KIO 4 oxidation. With the KIO4 dose increased from 0.6 to 1.4 mmol/g VSS, the CST, SRF, and Wc obviously decreased from 88.95 s, 4.14 × 10 12 m/kg and 87.43%, to 47.55 s, 0.99 × 10 12 m/kg With the KIO 4 dose increased from 0.6 to 1.4 mmol/g VSS, the CST, SRF, and W C obviously decreased from 88.95 s, 4.14 × 10 12 m/kg and 87.43%, to 47.55 s, 0.99 × 10 12 m/kg and 83.75%, respectively. This implied that more reactive species were generated with the increase in KIO 4 dose, which led to an enhanced oxidation of sludge flocs [13]. As the KIO 4 dose further increased to 1.8 mmol/g VSS, CST and SRF increased instead; meanwhile, W C maintained stable. On the one hand, an excess of IO 4 − will react with reactive free radicals (Equation (2)) [35]. On the other hand, due to the decrease in flocculation activity of overoxidized sludge, the sludge filterability is reduced and the dehydration is deteriorated [36].  and 83.75%, respectively. This implied that more reactive species were generated with the increase in KIO4 dose, which led to an enhanced oxidation of sludge flocs [13]. As the KIO4 dose further increased to 1.8 mmol/g VSS, CST and SRF increased instead; meanwhile, WC maintained stable. On the one hand, an excess of IO4 − will react with reactive free radicals (Equation (2)) [35]. On the other hand, due to the decrease in flocculation activity of overoxidized sludge, the sludge filterability is reduced and the dehydration is deteriorated [36]. As the Fe(Ⅱ)/KIO4 molar ratio increased from 0 to 1.2, the CST, SRF, and Wc obviously decreased from 133.35 s, 6.24 × 10 12 m/kg and 89.75%, to 47.55 s, 0.99 × 10 12 m/kg and 83.75%, respectively. The reason for the improvement of sludge dewatering performance is that Fe 2+ , as an initiator, activates IO4 − to generate free radicals, which accelerates the disintegration of sludge flocs. When the Fe(Ⅱ)/KIO4 molar ratio was lower than 1.2, insufficient ferrous ions were not capable of forming enough reactive species and, therefore, the oxidation of sludge flocs was greatly limited [37]. Whereas Fe(Ⅱ)/KIO4 molar ratio exceeded 1.2, the dewatering performance of the sludge became poor, which was mainly due to the fact that the excess ferrous consumed the reactive species produced. For instance, excessive Fe(II) would result in a competitive consumption of Fe(IV) (k(Fe(IV)+Fe(II)) = 1.4 × 10 5 M −1 s −1 ) [17].

Zeta Potential and Particle Size Distribution
Zeta potential can reflect the stability of the sludge colloid matrix to a certain extent and, therefore, is listed as one of the key indicators to evaluate the dewatering performance of sludge [38].
As shown in Figure 4a, the Zeta potential of raw sludge was −10.60 mV, which was due to the existence of carboxylic acid and phosphate groups in the EPS [39]. As a comparison, the Zeta potential of the sludge treated by KIO4 alone changed to −11.92 mV, which brought about an increased electrostatic repulsion and, hence, was detrimental to the improvement of dewatering performance. After treatment with ferrous iron alone, the Zeta potential increased to −7.91 mV, which may be caused by the neutralization of negative charges incurred by Fe 2+ and its oxidization product (Fe 3+ ). Meanwhile, the Zeta potential of the sludge treated by the Fe(Ⅱ)/KIO4 oxidation system increased to −1.87 mV, which boosted the coagulation of sludge flocs and significantly improved the dewatering performance of sludge. The increase in Zeta potential may be ascribed to the decomposition of negatively charged organic matter in sludge EPS and neutralization of negative charges by positive-charged iron ions. As the Fe(II)/KIO 4 molar ratio increased from 0 to 1.2, the CST, SRF, and W C obviously decreased from 133.35 s, 6.24 × 10 12 m/kg and 89.75%, to 47.55 s, 0.99 × 10 12 m/kg and 83.75%, respectively. The reason for the improvement of sludge dewatering performance is that Fe 2+ , as an initiator, activates IO 4 − to generate free radicals, which accelerates the disintegration of sludge flocs. When the Fe(II)/KIO 4 molar ratio was lower than 1.2, insufficient ferrous ions were not capable of forming enough reactive species and, therefore, the oxidation of sludge flocs was greatly limited [37]. Whereas Fe(II)/KIO 4 molar ratio exceeded 1.2, the dewatering performance of the sludge became poor, which was mainly due to the fact that the excess ferrous consumed the reactive species produced. For instance, excessive Fe(II) would result in a competitive consumption of Fe(IV) (k (Fe(IV)+Fe(II)) = 1.4 × 10 5 M −1 s −1 ) [17].

Zeta Potential and Particle Size Distribution
Zeta potential can reflect the stability of the sludge colloid matrix to a certain extent and, therefore, is listed as one of the key indicators to evaluate the dewatering performance of sludge [38].
As shown in Figure 4a, the Zeta potential of raw sludge was −10.60 mV, which was due to the existence of carboxylic acid and phosphate groups in the EPS [39]. As a comparison, the Zeta potential of the sludge treated by KIO 4 alone changed to −11.92 mV, which brought about an increased electrostatic repulsion and, hence, was detrimental to the improvement of dewatering performance. After treatment with ferrous iron alone, the Zeta potential increased to −7.91 mV, which may be caused by the neutralization of negative charges incurred by Fe 2+ and its oxidization product (Fe 3+ ). Meanwhile, the Zeta potential of the sludge treated by the Fe(II)/KIO 4 oxidation system increased to −1.87 mV, which boosted the coagulation of sludge flocs and significantly improved the dewatering performance of sludge. The increase in Zeta potential may be ascribed to the decomposition of negatively charged organic matter in sludge EPS and neutralization of negative charges by positive-charged iron ions.
The particle size is usually considered another key factor affecting sludge dewatering. As displayed in Figure 4b, the medium value of particle size (D 50 ) of sludge flocs was reduced from 52.85 to 51.08 µm after treatment by KIO 4 alone. The reduction in D 50 can be explained by the release of some hydrophilic groups in the sludge EPS subjected to the oxidation of KIO 4 . One accompanying negative effect is the blockage of the filtration channel in sludge flocs. Once Fe 2+ was added, the negative charges on the surface of the sludge flocs were neutralized, and the sludge flocs quickly aggregated into larger sludge particles with a D 50 being 64.7 µm. In regards to the Fe(II)/KIO 4 process, the oxidation effect tended to make the sludge flocs smaller. The oxidation of active species led to a decrease in hydrophilic organic compounds such as proteins. On the other hand, the flocculation effect of iron ions drove the sludge flocs to become larger. As a compromise, the D 50 of sludge flocs treated by Fe(II)/KIO 4 process became 60.07 µm, which was larger than the D 50 of raw sludge (52.85 µm) and helped to improve the sludge dewaterability. The particle size is usually considered another key factor affecting sludge dewatering. As displayed in Figure 4b, the medium value of particle size (D50) of sludge flocs was reduced from 52.85 to 51.08 μm after treatment by KIO4 alone. The reduction in D50 can be explained by the release of some hydrophilic groups in the sludge EPS subjected to the oxidation of KIO4. One accompanying negative effect is the blockage of the filtration channel in sludge flocs. Once Fe 2+ was added, the negative charges on the surface of the sludge flocs were neutralized, and the sludge flocs quickly aggregated into larger sludge particles with a D50 being 64.7 μm. In regards to the Fe(Ⅱ)/KIO4 process, the oxidation effect tended to make the sludge flocs smaller. The oxidation of active species led to a decrease in hydrophilic organic compounds such as proteins. On the other hand, the flocculation effect of iron ions drove the sludge flocs to become larger. As a compromise, the D50 of sludge flocs treated by Fe(Ⅱ)/KIO4 process became 60.07 μm, which was larger than the D50 of raw sludge (52.85 μm) and helped to improve the sludge dewaterability.

FT-IR Spectra and SEM Analyses
To further understand the impact of Fe(Ⅱ)/KIO4 oxidation on sludge dewaterability, the changes in functional groups on sludge flocs were analyzed by FT-IR spectra.

FT-IR Spectra and SEM Analyses
To further understand the impact of Fe(II)/KIO 4 oxidation on sludge dewaterability, the changes in functional groups on sludge flocs were analyzed by FT-IR spectra.
As exhibited in Figure 5a, the peaks around 3415 cm −1 and 3164 cm −1 , correspond to O-H stretching vibrations (mainly present in polysaccharides and phenols [40]). The peaks around 2926 cm −1 and 2852 cm −1 , are related to the symmetric and asymmetric stretching vibrations of CH 2 in lipids. The amide I band at 1657 cm −1 is mainly related to C=O and C-N stretching associated with proteins. The swelling vibration of N-H and C-N on amide II in the protein is related to the peak near 1533 cm −1 . The peaks near 1401 cm −1 , 1235 cm −1 , and 1035 cm −1 , corresponds to the bending vibration and C-O stretching vibration in the face of phenolic O-H, the asymmetric stretching vibration of C-O-C in ester, and the stretching vibration of C-O in polysaccharide, respectively [41]. As presented in Figure 5b, after treatment by Fe(II)/KIO 4 , the peak at 1235 cm −1 almost disappeared, suggesting the decomposition of the corresponding organic matter. Meanwhile, the relative intensities of typical peaks located around 1657 cm −1 , 1533 cm −1 , and 1035 cm −1 , decreased, indicating that proteins and polysaccharides on the surface of sludge flocs were partly degraded. and 1035 cm −1 , corresponds to the bending vibration and C-O stretching vibration in the face of phenolic O-H, the asymmetric stretching vibration of C-O-C in ester, and the stretching vibration of C-O in polysaccharide, respectively [41]. As presented in Figure 5b, after treatment by Fe(Ⅱ)/KIO4, the peak at 1235 cm −1 almost disappeared, suggesting the decomposition of the corresponding organic matter. Meanwhile, the relative intensities of typical peaks located around 1657 cm −1 , 1533 cm −1 , and 1035 cm −1 , decreased, indicating that proteins and polysaccharides on the surface of sludge flocs were partly degraded.  To intuitively analyze the influence of sludge treatment on the structure of sludge flocs, SEM analyses were conducted. As shown in Figure 6a, the raw sludge flocs were dense with few pores. This structural characteristic was embodied with low porosity and few filter channels. As a result, water is difficult to separate from the sludge [42]. Therefore, the SRF and CST of raw sludge were maintained at a high level. After treatment by the Fe(II)/KIO 4 process, the dense structure of the sludge flocs transformed to become loose and porous (Figure 6b). The sludge drainage passages for more water permeation were thus reconstructed, which was in favor of promotion of sludge filterability [39,43]. To intuitively analyze the influence of sludge treatment on the structure of sludge flocs, SEM analyses were conducted. As shown in Figure 6a, the raw sludge flocs were dense with few pores. This structural characteristic was embodied with low porosity and few filter channels. As a result, water is difficult to separate from the sludge [42]. Therefore, the SRF and CST of raw sludge were maintained at a high level. After treatment by the Fe(Ⅱ)/KIO4 process, the dense structure of the sludge flocs transformed to become loose and porous (Figure 6b). The sludge drainage passages for more water permeation were thus reconstructed, which was in favor of promotion of sludge filterability [39,43].

Polysaccharide and Protein in EPS
EPS consists of high hydrophilic substances, and those substances mainly include proteins and polysaccharides [6]. EPS is highly charged polymers that interact with water molecules to form gels, which can be divided into three parts from outside to inside, namely, the soluble EPS (S-EPS), the loosely bound EPS (LB-EPS), and the tightly bound EPS (TB-EPS) [44]. The S-EPS is weakly bound to cells or dissolved in solutions. The LB-EPS is a loose slime layer without an obvious end. The TB-EPS is bound to the cell surface

Polysaccharide and Protein in EPS
EPS consists of high hydrophilic substances, and those substances mainly include proteins and polysaccharides [6]. EPS is highly charged polymers that interact with water molecules to form gels, which can be divided into three parts from outside to inside, namely, the soluble EPS (S-EPS), the loosely bound EPS (LB-EPS), and the tightly bound EPS (TB-EPS) [44]. The S-EPS is weakly bound to cells or dissolved in solutions. The LB-EPS is a loose slime layer without an obvious end. The TB-EPS is bound to the cell surface tightly and stably [45].
As shown in Figure 7, for raw sludge, the proteins concentrations in S-EPS, LB-EPS, and TB-EPS were 53.85, 161.21, and 690.56 mg/L, respectively. In regard to the treated sludge, the proteins concentration in S-EPS greatly increased to 158.85 mg/L, whereas the protein concentration in TB-EPS sharply decreased to 395.26 mg/L. This result indicated that LB-EPS was transformed into S-EPS. Similarly, it has been reported that the oxidative dissolution of EPS leads to the gradual transformation of TB-EPS to LB-EPS and subsequently to S-EPS, and the increased protein content in SB-EPS improved sludge dewatering [46,47]. As a whole, the concentration of proteins in EPS of treated sludge showed a 21.5% decline in comparison to that of raw sludge. In this study, along with the degradation of proteins, the CST and SRF decreased. This result was parallel to the finding by Wang et al. [48], who reported that proteins (including aromatic protein substances and tryptophan protein substances) are negatively correlated with sludge dewaterability. Moreover, the degradation of protein-like substances has been proven to be the key organics benefitting the improvement in sludge dewaterability [49,50]. In addition, the concentration of polysaccharides in the EPS of treated sludge exhibited a slight increase instead of decrease, which is in accordance with Liu et al. [32], who concluded that polysaccharides showed little influence on sludge dewaterability in comparison with proteins while pretreating WAS by Fe(II)-activated PMS oxidation. Similarly, Wu et al. [51] reported that high hydrophilic proteins played a more important role in sludge dewaterability than polysaccharides. In addition, the concentration of polysaccharides in the EPS of treated sludge exhibited a slight increase instead of decrease, which is in accordance with Liu et al. [32], who concluded that polysaccharides showed little influence on sludge dewaterability in comparison with proteins while pretreating WAS by Fe(II)-activated PMS oxidation. Similarly, Wu et al. [51] reported that high hydrophilic proteins played a more important role in sludge dewaterability than polysaccharides.

Reactive Species
The identification of reactive species is of great importance to understand the underlying mechanism of the Fe(II)/KIO 4 process. The possible reactive species include • OH, singlet oxygen ( 1 O 2 ), superoxide anion radical ( • O 2 − ), iodate radical ( • IO 3 ), and Fe(IV) [12,17]. It has been reported that ascorbic acid (AA) has the ability of non-selective scavenging free radicals, which can be used to preliminarily test whether radical species were generated in the reaction system [52]. Moreover, TBA [53], L-histidine [54], NBT, and phenol [19] have been proven to possess strong scavenging ability on • OH, 1 O 2 , • O 2 − , and • IO 3 , respectively. Therefore, AA, TBA, L-histidine, NBT, and phenol were used in this study to identify and differentiate free radicals. Significantly, Zong et al. [17] have provided conclusive evidence for the generation of high-valent iron−oxo species (Fe(IV)) in the Fe(II)/KIO 4 (5)). Note that Fe(IV) undergoes self-decay and subsequently generates H 2 O 2 and Fe(III) (Equation (6)) [55]; thus, • OH can be considered to be indirectly triggered by the self-decomposition of Fe(IV) and the subsequent Fenton reaction (Equation (7)). In addition, it is inferred that • O 2 − was generated according to Equation (8). As depicted in Figure 8, the values of three indices (i.e., CST, SRF, and W C ) increased with the addition of AA, indicating that radical species were involved in this system. After the addition of TBA, the values of the three indices also greatly increased, suggesting that scavenging of • OH produced in the Fe(II)/KIO 4 process worsened the sludge dewaterability. It was noteworthy that the additions of NBT and phenol had a slight impact on sludge dewatering performance, indicating that • O 2 − and • IO 3 played a small role. In fact, the redox potential of • O 2 − is only −0.28 V [56,57], and its oxidative contribution to the degradation of sludge EPS was thus expected to be negligible [58]. Moreover, • IO 3 was excluded as a reactive species in the Fe(II)/KIO 4 process [17]. The seemingly quenching effect that arose from the addition of phenol may be ascribed to the adverse impact of phenol on other reactive species. In regards to the addition of L-histidine, its effect on sludge dewaterability was slightly more obvious compared to the additions of NBT and phenol. However, 1 O 2 was excluded as a reactive species in the Fe(II)/KIO 4 process based on irrefutable proofs [17]. Herein, whether L-histidine will react with Fe(IV) or KIO 4 and resultantly poses negative influence on the oxidizing power of Fe(IV) or the generation of • OH merits further investigations.

Environmental Implications
Taking the effects of initial pH, KIO4 dose, and Fe(Ⅱ)/KIO4 molar ratio on sludge dewaterability into comprehensive considerations, the optimal conditions are as follows: the initial pH = 6.8, the KIO4 dose = 1.4 mmol/g VSS, and the Fe(Ⅱ)/KIO4 molar ratio = 1.2. Compared to those of the untreated raw sludge, the CST (47.55 s) decreased by 64.34%, SRF (0.99 × 10 12 m/kg) decreased by 84.13%, and Wc (83.75%) decreased by 6.69% under the optimal conditions. It can be concluded that the dewaterability of WAS can be effectively improved by Fe(Ⅱ)/KIO4 oxidation. As for the underlying mechanism, the role of Fe(IV) as a reaction intermediate needs to be emphasized. Although the reactivity of Fe(IV) toward organics is generally weaker than that of radical species (i.e., • OH and SO4 •− ), the steady-state concentration of Fe(IV) (∼10 −9 M) could be several orders of magnitude higher than those of • OH and SO4 •− (∼10 −12 M). By in situ generating the high-spin Fe(IV) species and accompanying • OH, the Fe(II)/KIO4 process could achieve a better

Environmental Implications
Taking the effects of initial pH, KIO 4 dose, and Fe(II)/KIO 4 molar ratio on sludge dewaterability into comprehensive considerations, the optimal conditions are as follows: the initial pH = 6.8, the KIO 4 dose = 1.4 mmol/g VSS, and the Fe(II)/KIO 4 molar ratio = 1.2. Compared to those of the untreated raw sludge, the CST (47.55 s) decreased by 64.34%, SRF (0.99 × 10 12 m/kg) decreased by 84.13%, and W C (83.75%) decreased by 6.69% under the optimal conditions. It can be concluded that the dewaterability of WAS can be effectively improved by Fe(II)/KIO 4 oxidation. As for the underlying mechanism, the role of Fe(IV) as a reaction intermediate needs to be emphasized. Although the reactivity of Fe(IV) toward organics is generally weaker than that of radical species (i.e., • OH and SO 4 •− ), the steadystate concentration of Fe(IV) (∼10 −9 M) could be several orders of magnitude higher than those of • OH and SO 4 •− (∼10 −12 M). By in situ generating the high-spin Fe(IV) species and accompanying • OH, the Fe(II)/KIO 4 process could achieve a better performance on the degradation of representative pollutants in comparison with the Fe(II)/PMS and Fe(II)/PDS processes [17]. In this work, no comparisons in terms of EPS degradation degree between Fe(II)/KIO 4 and Fe(II)/PMS or Fe(II)/PDS were conducted; hence, we cannot claim that Fe(II)/KIO 4 has a higher oxidizing capacity towards EPS. Additionally, excessive degradation of EPS may instead deteriorate the sludge dewatering performance [6]. It can be inferred that the Fe(II)/KIO 4 process achieved a moderate degree of oxidation towards sludge EPS under the optimal conditions. Apart from the effectiveness in improving sludge dewatering performance, the Fe(II)/KIO 4 process has other attractive advantages. Compared with liquid-form oxidants (e.g., H 2 O 2 , hypochlorous acid, and peracetic acid), KIO 4 , as a solid-form oxidant, is relatively stable and carries less risk during transport and storage. More importantly, there is no requirement for regulation of the initial pH, which can reduce the consumption of agents and simplify the operation. Attributed to the abovementioned advantages, Fe(II)-activated KIO 4 oxidation has great prospect and practical value in sludge conditioning for dewatering.

Conclusions
In this study, Fe(II)-activated KIO 4 oxidation was successfully applied to improve the sludge dewatering performance, which is closely relevant to the evolution of sludge properties. Under the optimal conditions, the CST can decrease by 64.34%, SRF can be reduced by 84.13%, and W C can be lowered by 6.69%. A moderate degree of oxidation towards sludge EPS is crucial for enhancing sludge dewaterability. Herein, regulation of the initial pH of raw sludge is not essential, contributing to less consumption of agents and a simpler operation. High hydrophilic proteins played a major role, whereas polysaccharides played a negligible part in sludge conditioning. In situ generated Fe(IV) should receive special attention, not only due to its own oxidation capacity, but also due to its self-decay that can trigger the generation of • OH. Fe(IV) and accompanying • OH are predominant reactive species responsible for the oxidation of EPS. This work proposes a facile strategy to improve the sludge dewaterability.

Conflicts of Interest:
The authors declare that they have no known competing financial interest or personal relationships that could have influenced the work reported in this paper.