Advanced Oxidation Processes and Nanoﬁltration to Reduce the Color and Chemical Oxygen Demand of Waste Soy Sauce

: Currently, the ozone (O 3 ) oxidation efﬁciency in the treatment of waste soy sauce provides 34.2% color removal and a 27.4% reduction in its chemical oxygen demand (COD). To improve the O 3 oxidation efﬁciency, hydrogen peroxide (H 2 O 2 ) is used to cause a H 2 O 2 /O 3 process. In H 2 O 2 /O 3 process experiments, a previously optimized pH of 11 and applied O 3 dose of 50 mg L − 1 were used and the H 2 O 2 /O 3 ratio was varied between 0.1 and 0.9 in intervals of 0.2. The results show that an H 2 O 2 /O 3 ratio of 0.3 results in the highest efﬁciencies in terms of color removal (51.6%) and COD reduction (33.8%). Nanoﬁltration (NF) was used to pretreat the waste soy sauce to improve color removal and COD reduction. The results showed that NF with an NE-70 membrane results in 80.8% color removal and 79.6% COD reduction. Finally, the combination of NF and H 2 O 2 /O 3 process resulted in the best treatment efﬁciency: 98.1% color removal and 98.2% COD reduction. Thus, NF & H 2 O 2 /O 3 process can be considered as one of the best treatment methods for waste soy sauce, which requires high intrinsic color removal and COD reduction efﬁciencies.


Introduction
Waste soy sauce has a high chemical oxygen demand (COD). It has an intense dark brown color due to the presence of caramel pigments and the occurrence of the Maillard reactions producing melanin or melanoidin [1,2]. Several processes are typically used to treat soy sauce wastewater including nitrification and denitrification, an activated sludge process, and biological treatments. However, it is difficult to control the stability of the soy sauce treatment processes because of annual fluctuations in the amount of waste soy sauce generated (47,914 tons were produced in 2016 compared to 37,232 tons in 2015 and 40,461 tons in 2014) [3]. Biological treatment removes only a small fraction of the organic matter that contributes to the color of the effluents [4][5][6]. Therefore, the wastewater discharged to the environment is colored, which is aesthetically displeasing; moreover, the wastewater could affect the ecosystem due to its non-biodegradability and recalcitrance [7]. Therefore, it is necessary to decolorize waste soy sauce before it is discharged.
In our earlier study, overcoming the limitations of ozone (O 3 ) oxidation to reduce the color and COD of waste soy sauce was intensively investigated [8]. Up to 34.2% color removal and 27.4% COD reduction were achieved by O 3 oxidation at a pH of 11 with an applied O 3 dose of 50 mg L −1 [8]. However, the color removal and COD reduction were not thoroughly investigated as the salt water Molecules that were larger than the pore size of the packing material were excluded and eluted first, at the void volume. Smaller molecules were able to penetrate through the porous infrastructure and were attenuated, corresponding to a higher retention time [16].

H 2 O 2 /O 3 Process
A schematic of the benchtop-scale reactor system used herein is shown in Figure 1. Experimental runs were performed in a 1 L Pyrex reactor with 500 mL of sample. The reactor was filled with 500 mL of waste soy sauce and agitated with a magnetic stirrer at 150 rpm. The optimum conditions for color removal and COD reduction by O 3 -based oxidation were a pH of 11 and an applied O 3 concentration of 50 mg L −1 [8]. The O 3 -containing gas was supplied continuously for 30 min through a gas diffuser at the bottom of the reactor. An O 3 trap containing a 2% potassium iodide solution was connected in series with the reactor to verify the O 3 gas concentration in the outlet gas stream. A 0.1 N Na 2 S 2 O 3 (sodium thiosulfate) solution was used as the reducing agent for the reverse titration in the trap. Subsequently, 0.1 N H 2 SO 4 (sulfuric acid) was used to facilitate the reaction of the O 3 in the liquid phase with the I 2 [8]. All experiments were performed in duplicate at room temperature under a fume hood (for safety due to the presence of O 3 gas). The pH, applied O 3 concentration, and reaction time were kept constant while the H 2 O 2 /O 3 ratio was varied between 0.1 and 0.9 in intervals of 0.2. Table 2

Nanofiltration (NF) System
The NF membranes used in this study were thin-film composite NF membranes, NE-70 and NE-90 (Toray Chemical, Korea). These membranes have different top layers, zeta potentials, wettabilities, and roughnesses as detailed in Table 4. The characteristics of the NF membrane were investigated in previous study [17]. A laboratory-scale dead-end NF membrane was used in this experiment. The membrane surface area was 0.015 m 2 . Waste soy sauce was fed into the filtration module by a gear pump. The filtration experiments were performed with a commercial NF module. Figure 2 shows a schematic of the NF system. In all experiments, a pressure of 1.5 MPa (15 bar) was applied at room temperature. The evolution of flux and rejection progressed over a period of 6 hours. For analysis and comparison, the measurements were taken after 1 h of filtration.

Nanofiltration (NF) System
The NF membranes used in this study were thin-film composite NF membranes, NE-70 and NE-90 (Toray Chemical, Korea). These membranes have different top layers, zeta potentials, wettabilities, and roughnesses as detailed in Table 4. The characteristics of the NF membrane were investigated in previous study [17]. A laboratory-scale dead-end NF membrane was used in this experiment. The membrane surface area was 0.015 m 2 . Waste soy sauce was fed into the filtration module by a gear pump. The filtration experiments were performed with a commercial NF module. Figure 2 shows a schematic of the NF system. In all experiments, a pressure of 1.5 MPa (15 bar) was applied at room temperature. The evolution of flux and rejection progressed over a period of 6 hours. For analysis and comparison, the measurements were taken after 1 h of filtration.

Characteristics of Waste Soy Sauce
The chemical properties of the sample were evaluated according to the guidelines for testing water pollution developed by the Ministry of Environment of Korea [18]. Table 5 shows the characteristics of the waste soy sauce. The pH value of 4.6 indicates that the waste soy sauce was acidic. The color was found to be 3810 TCU and the COD was measured to be 231.5 g/L. These results are similar to those reported in a previous study: 4038 TCU color and 229.1 g/L COD [8]. The T-N and T-P concentrations were measured to be 10.4 and 2.8 g/L, respectively. The salinity, 16.4%, was much higher than that found in other types of organic wastewater. In general, the high-salinity wastewater does not exhibit high removal efficiencies with biological treatment systems. This is because the performances of biological treatment processes are adversely affected by the negative effects of salt on microbial flora [19]. Moreover, the TOC, which is used as an indicator of water pollution, was found to be 57.6 g/L.

Characteristics of Waste Soy Sauce
The chemical properties of the sample were evaluated according to the guidelines for testing water pollution developed by the Ministry of Environment of Korea [18]. Table 5 shows the characteristics of the waste soy sauce. The pH value of 4.6 indicates that the waste soy sauce was acidic. The color was found to be 3810 TCU and the COD was measured to be 231.5 g/L. These results are similar to those reported in a previous study: 4038 TCU color and 229.1 g/L COD [8]. The T-N and T-P concentrations were measured to be 10.4 and 2.8 g/L, respectively. The salinity, 16.4%, was much higher than that found in other types of organic wastewater. In general, the high-salinity wastewater does not exhibit high removal efficiencies with biological treatment systems. This is because the performances of biological treatment processes are adversely affected by the negative effects of salt on microbial flora [19]. Moreover, the TOC, which is used as an indicator of water pollution, was found to be 57.6 g/L.

Optimization of H 2 O 2 /O 3 Ratio
Our previous study revealed that a pH of 11 and an applied O 3 concentration of 50 mg L −1 are optimal for color removal and COD reduction; these conditions were used in this study. H 2 O 2 was combined with O 3 to accelerate the oxidation of the organic molecules present in the wastewater. The H 2 O 2 /O 3 ratio was varied between 0.1 and 0.9 to optimize the color removal by H 2 O 2 /O 3 process; the results of this experiment are shown in Figure 3.

Optimization of H2O2/O3 Ratio
Our previous study revealed that a pH of 11 and an applied O3 concentration of 50 mg L −1 are optimal for color removal and COD reduction; these conditions were used in this study. H2O2 was combined with O3 to accelerate the oxidation of the organic molecules present in the wastewater. The H2O2/O3 ratio was varied between 0.1 and 0.9 to optimize the color removal by H2O2/O3 process; the results of this experiment are shown in Figure 3. As the H2O2/O3 ratio increased from 0.1 to 0.3, the color removal increased gradually from 41.2 to 51.6%. The H2O2/O3 ratio of 0.3 resulted in the greatest color removal. The color removal decreased sharply from 51.6 to 17.1% as the H2O2/O3 ratio increased above 0.3.
The lower residual color after oxidation with H2O2/O3 as the H2O2/O3 ratio increased was attributed to the increased formation of OH radicals [20][21][22][23]. Other studies have reported increased H2O2/O3 ratios and biodegradability after wastewater treatment with O3 and H2O2 [24,25]. The decreased color removal with H2O2/O3 ratios above 0.3 was attributed to the strong scavenging effects of carbonate (CO3 2− ) and bicarbonate (HCO3 − ). The results showed that an H2O2/O3 ratio of 0.3 was optimal for color removal (Yielding 51.6% color reduction) because of the high oxygenation capacity resulting from the suitable amount of hydroxyl radicals. Thus, it was confirmed that the H2O2/O3 process are more efficient (51.6%) in removing color than in O3-based oxidation (34.2%) [8].  Figure 4 shows the influence of the H2O2/O3 ratio between 0.1 and 0.9 on the COD reduction due to the O3 treatment of waste soy sauce. The results show that the COD reduction as a function of the H2O2/O3 ratio followed a similar trend to that of the color removal shown in Figure 3. The COD reduction was optimal (a 73.6 g/L reduction from 217.73 g/L) at an H2O2/O3 ratio of 0.3. This result confirms that a sufficient amount of OH radicals were produced at an H2O2/O3 ratio of 0.3. The H2O2 addition facilitated the production of OH radicals, resulting in a synergistic effect between the applied O3 concentration and the COD reduction [26]. The COD reduction effect decreased distinctly from 33.8 to 16.6% as the H2O2/O3 ratio increased above 0.3 as observed with the color removal. This was likely due to the OH radicals being consumed by the excessive amounts of H2O2 [27][28][29].
The results shown in Figures 3 and 4 confirm that the color removal and COD reduction strongly depend on the H2O2/O3 ratio. However, the H2O2/O3 ratio influences color removal more than COD reduction. Moreover, an H2O2/O3 ratio of 0.3 was the most effective in terms of both color removal and COD reduction.  Figure 4 shows the influence of the H 2 O 2 /O 3 ratio between 0.1 and 0.9 on the COD reduction due to the O 3 treatment of waste soy sauce. The results show that the COD reduction as a function of the H 2 O 2 /O 3 ratio followed a similar trend to that of the color removal shown in Figure 3. The COD reduction was optimal (a 73.6 g/L reduction from 217.73 g/L) at an H 2 O 2 /O 3 ratio of 0.3. This result confirms that a sufficient amount of OH radicals were produced at an H 2 O 2 /O 3 ratio of 0.3. The H 2 O 2 addition facilitated the production of OH radicals, resulting in a synergistic effect between the applied O 3 concentration and the COD reduction [26]. The COD reduction effect decreased distinctly from 33.8 to 16.6% as the H 2 O 2 /O 3 ratio increased above 0.3 as observed with the color removal. This was likely due to the OH radicals being consumed by the excessive amounts of H 2 O 2 [27][28][29].
The results shown in Figures 3 and 4 confirm that the color removal and COD reduction strongly depend on the H 2 O 2 /O 3 ratio. However, the H 2 O 2 /O 3 ratio influences color removal more than COD reduction. Moreover, an H 2 O 2 /O 3 ratio of 0.3 was the most effective in terms of both color removal and COD reduction.

Size Exclusion Chromatography Analysis
The COD reduction was not significant despite the use of a H2O2/O3 process because the COD of waste soy sauce is considerably higher than that of other types of waste and wastewater. In addition, there is a high IOD due to the high COD of the waste soy sauce. Thus, the initial COD of waste soy sauce makes it unsuitable for the oxidation treatment [30]. In addition, it is known that H2O2 interferes with the COD reduction by this process by consuming oxidation agents including potassium dichromate [31]. Thus, the molecular weight distribution of the total organic carbon was measured to determine the best membrane-based filtration pretreatment in the interest of improving the COD reduction effect of the H2O2/O3 process. Figure 5 shows the molecular weight distribution in the waste soy sauce as measured via sizeexclusion chromatography (SEC) using a method for detecting organic matter wherein the adsorption of the sample was measured at an ultraviolet (254 nm) wavelength [32]. The main peak shows the molecular weight of the organic matter ranged 750-2200 dalton. From this data, the appropriate molecular weight cutoff (MWCO) for the NF membranes (NE-70 and NE-90) was selected to approve the COD reduction effect by H2O2/O3 process. To facilitate stable treatment by oxidation, the COD reduction by the membrane pretreatment was calculated to

Size Exclusion Chromatography Analysis
The COD reduction was not significant despite the use of a H 2 O 2 /O 3 process because the COD of waste soy sauce is considerably higher than that of other types of waste and wastewater. In addition, there is a high IOD due to the high COD of the waste soy sauce. Thus, the initial COD of waste soy sauce makes it unsuitable for the oxidation treatment [30]. In addition, it is known that H 2 O 2 interferes with the COD reduction by this process by consuming oxidation agents including potassium dichromate [31]. Thus, the molecular weight distribution of the total organic carbon was measured to determine the best membrane-based filtration pretreatment in the interest of improving the COD reduction effect of the H 2 O 2 /O 3 process. Figure 5 shows the molecular weight distribution in the waste soy sauce as measured via size-exclusion chromatography (SEC) using a method for detecting organic matter wherein the adsorption of the sample was measured at an ultraviolet (254 nm) wavelength [32]. The main peak shows the molecular weight of the organic matter ranged 750-2200 dalton.

Size Exclusion Chromatography Analysis
The COD reduction was not significant despite the use of a H2O2/O3 process because the COD of waste soy sauce is considerably higher than that of other types of waste and wastewater. In addition, there is a high IOD due to the high COD of the waste soy sauce. Thus, the initial COD of waste soy sauce makes it unsuitable for the oxidation treatment [30]. In addition, it is known that H2O2 interferes with the COD reduction by this process by consuming oxidation agents including potassium dichromate [31]. Thus, the molecular weight distribution of the total organic carbon was measured to determine the best membrane-based filtration pretreatment in the interest of improving the COD reduction effect of the H2O2/O3 process. Figure 5 shows the molecular weight distribution in the waste soy sauce as measured via sizeexclusion chromatography (SEC) using a method for detecting organic matter wherein the adsorption of the sample was measured at an ultraviolet (254 nm) wavelength [32]. The main peak shows the molecular weight of the organic matter ranged 750-2200 dalton. From this data, the appropriate molecular weight cutoff (MWCO) for the NF membranes (NE-70 and NE-90) was selected to approve the COD reduction effect by H2O2/O3 process. To facilitate stable treatment by oxidation, the COD reduction by the membrane pretreatment was calculated to From this data, the appropriate molecular weight cutoff (MWCO) for the NF membranes (NE-70 and NE-90) was selected to approve the COD reduction effect by H 2 O 2 /O 3 process. To facilitate stable treatment by oxidation, the COD reduction by the membrane pretreatment was calculated to be over 70%. Thus, experiments were conducted using two membranes, NE-70 (71.8% of the expected COD removal) and NE-90 (73.2% of expected COD removal), confirming the effectiveness of the pretreatment in terms of color removal and COD reduction.

Color Removal and COD Reduction by Nanofiltration (NF)
In our earlier study, O 3 -based oxidation exhibited low color removal (34.2%) and COD reduction (27.4%) efficiencies at pH 11 with an O 3 injection dose of 50 mg L −1 [8]. In this study, we used H 2 O 2 /O 3 process and found that the removal efficiency was higher than that with than O 3 -based oxidation under the same conditions. However, the color removal and COD reduction were not complete even with H 2 O 2 /O 3 process. Thus, we applied NF as a pretreatment to enhance the color removal and COD reduction by oxidation method. Table 6 shows that color removal and COD reduction by NF was similar with the NE-70 and NE-90 membranes. The results show that the color removal and COD reduction were similar with both membranes even though the MWCO of the NE-90 membrane (210 daltons) is lower than that of the NE-70 membrane (350 daltons). The NE-90 membrane, which yielded a color reduction of 81.3% and a COD reduction of 80.7%, was slightly more effective for removing color and reducing the COD than the NE-70 membrane, which yielded a color reduction of 80.8% and a COD of 79.6%. However, the NE-70 is of sustainability usable than NE-90 for experiment. It can further be concluded that the NE-70 membrane is suitable as a pretreatment for waste soy sauce treatment as it improves the color removal and COD reduction.

Flux Variation
To predict the lifetime of the nanofiltration (NF) membranes for waste soy sauce treatment, flux experiments were conducted. Figure 6 shows the flux decline of the two NF membranes during 4 hours of operation. The flux decreased sharply initially then steadily decreased after 45 min of operation. The flux of the NE-70 membrane was lower than that of the NE-90 membrane. This was attributed to the properties of the membranes: NE-90 has a higher contact angle (41.5 ± 3.7 • ) than NE-70 (22.6 ± 1.9 • ), showing that its surface is more hydrophobic [17,33,34]; this would cause hydrophobic organic compounds to adsorb onto the surface [17]. In addition, NE-90 has a higher surface roughness than NE-70, which results in a greater repulsive force between the membrane surface and the solute and, thus, a lower solute permeability [17].
These results show that NE-70 is more suitable for this application as its flux decline is lesser during the waste soy sauce treatment than that of NE-90 although the color removal and COD reduction were similar for the two membranes.
operation. The flux of the NE-70 membrane was lower than that of the NE-90 membrane. This was attributed to the properties of the membranes: NE-90 has a higher contact angle (41.5 ± 3.7°) than NE-70 (22.6 ± 1.9°), showing that its surface is more hydrophobic [17,33,34]; this would cause hydrophobic organic compounds to adsorb onto the surface [17]. In addition, NE-90 has a higher surface roughness than NE-70, which results in a greater repulsive force between the membrane surface and the solute and, thus, a lower solute permeability [17].

Comparison of Treatment Methods
To enhance the effectiveness of the O 3 oxidation, we aimed to increase the OH radical production by adding H 2 O 2 and applying NF as a pretreatment to overcome the limitation (i.e., the high COD concentration) of the H 2 O 2 /O 3 process. Figure 7 and Table 7 show comparisons of color removal and COD reduction obtained by various treatment methods for waste soy sauce treatment. The results show that the NF & H 2 O 2 /O 3 process resulted in the greatest color removal and COD reduction (98.1% and 98.2%, respectively) of the five methods considered. However, the residual color was 74.4 TCU and the residual COD was 4.2 g/L, which are considerably higher than those in ordinary wastewater. Therefore, additional treatment methods should be investigated in order to further improve the process. These results show that NE-70 is more suitable for this application as its flux decline is lesser during the waste soy sauce treatment than that of NE-90 although the color removal and COD reduction were similar for the two membranes.

Comparison of Treatment Methods
To enhance the effectiveness of the O3 oxidation, we aimed to increase the OH radical production by adding H2O2 and applying NF as a pretreatment to overcome the limitation (i.e., the high COD concentration) of the H2O2/O3 process. Figure 7 and Table 7 show comparisons of color removal and COD reduction obtained by various treatment methods for waste soy sauce treatment. The results show that the NF & H2O2/O3 process resulted in the greatest color removal and COD reduction (98.1% and 98.2%, respectively) of the five methods considered. However, the residual color was 74.4 TCU and the residual COD was 4.2 g/L, which are considerably higher than those in ordinary wastewater. Therefore, additional treatment methods should be investigated in order to further improve the process.

Conclusions
In this study, the H2O2/O3 process were optimized for removing the color and reducing the COD of waste soy sauce. The H2O2/O3 process were conducted under optimized conditions (H2O2/O3 ratio or 0.3, pH of 11.0, and applied O3 dose of 50 mg L −1 ), resulting in 51.6% color removal and 33.8% COD reduction. This was primarily due to the high oxidation capability of O3 in the presence of the hydroxyl radicals introduced by the addition of H2O2. Moreover, an appropriate membrane was selected for waste soy sauce pretreatment based on the molecular weight distribution of the

Conclusions
In this study, the H 2 O 2 /O 3 process were optimized for removing the color and reducing the COD of waste soy sauce. The H 2 O 2 /O 3 process were conducted under optimized conditions (H 2 O 2 /O 3 ratio or 0.3, pH of 11.0, and applied O 3 dose of 50 mg L −1 ), resulting in 51.6% color removal and 33.8% COD reduction. This was primarily due to the high oxidation capability of O 3 in the presence of the hydroxyl radicals introduced by the addition of H 2 O 2 . Moreover, an appropriate membrane was selected for waste soy sauce pretreatment based on the molecular weight distribution of the wastewater. The addition of this pretreatment resulted in 98.1% color removal and 98.2% COD reduction. Comparing with alternative methods, NF & H 2 O 2 /O 3 process can be considered one of the best treatment methods for waste soy sauce, which requires particularly high degrees of color removal and COD reduction. These results can ultimately guide future research into the best wastewater treatment techniques that would allow the wastewater to be reused and mitigate the environmental impacts when it is discharged. Funding: This research was funded by the Gyeongnam Green Environment Center (16-4-20-22-8).