Electrochemical oxidation process on palm oil mill effluent waste activated sludge: optimization by response surface methodology

The palm oil industry has always been a very popular and crucial contributor to the economy of Malaysia. Over the years, the number of palm oil mills has increased gradually, consequently generating more and more by-products and waste from the palm oil mill effluent (POME). By-products have had significant impacts upon the environment as they generate excessive biological oxygen demand (BOD) and chemical oxygen demand (COD). Furthermore, POME contains excessive concentrations of organic nitrogen, phosphorus and different supplement substances. These have caused the need for treatment for POME to ensure that it stays within the threshold limit value (TLV) based on the standards set by the local environmental organisations, for instance Department of Environment (DOE), Malaysia before discharge into water bodies (Kamyab et al. 2018).

Biological treatment is the most common treatment technology being practiced to treat POME. However, waste activated sludge (WAS), an unwanted residue that contains high organic and solid contents with poor dewaterability will be generated after the biological treatment. Landfilling and incineration are both common ways to handle WAS, however these methods have their downside by either jeopardizing the environment or threating the living population. Anaerobic digestion (AD) has always been the conventional method for treating the sewage treatment plant (STP) WAS, whereby it could reduce its volume, to ensure that it is safe to be treated as bio-solids. In addition, AD has the capability to release biogas that can be used as a source of renewable energy. Nevertheless, due to the presence of hard-to-manage substances within the POME WAS, AD could only do so much for instance, the elimination of the biodegradable part (Nurhayati et al. 2020). Hydrolysis in AD has always been a drawback that limits the treatment's efficiency due to the presence of extracellular polymeric substances (EPS) like lipids, polysaccharides, fatty acids and nuclei acids that have high water binding ability. For that reason, pre-treatment is required to achieve a better performance on the overall treatment system for POME WAS.

The nature of the electrodes used plays a crucial role as it could influence the performance of the EO. Bare metal as electrodes is treatable but corrosion could occur. These problems were able to be resolved by introducing a coating of transition metal oxides (TMO) and mixed metal oxides (MMO) (Ken & Sinha 2021). MMO helps to reduce the percentage of corrosion current through the coating which then increases the shelf life of the electrodes. MMO coated on metal plates are also termed as dimensionally stable anodes (DSA), they are known to have an extensive service life over a vast range of current density (Johnson & Kumar 2020). Lead dioxide (PbO₂) is able to excel in the treatment, but due to its lead dissolution properties it endangers the treatment system. Tin(IV) oxide (SnO₂) possesses a shorter life span when compared with others and iridium (IV) oxide (IrO₂) lacks O₂ over-potential that causes it to be less efficient. As for ruthenium (IV) oxide (RuO₂), it has been listed as one of the most versatile and finest oxidation catalyst due to its high stability and ohmic conductivity (Ajab et al. 2020). For excellent chlorine evolution properties (∼2.0 V), mechanical and chemical resistance in high pH conditions is needed (Johnson & Kumar 2020). Comparing MMO, RuO₂ showed a promising performance. It covers a vast range of treatments for wastewater from pharmaceutical to oil mill and even dye removal with the assistance of chlorine generated during the treatment process (Jiang et al. 2021). With its stability, it tends to have a longer life span when compared with other DSA. Pechini's method is an effective process used in the fabrication of DSA as the chelated metal cation and glycol undergoes polyesterification, forming covalent and coordinative bonds (Kang et al. 2020).

The aim of this study was to identify the optimum operating conditions for EO leading to a pre-treated sludge that is well disintegrated and dewatered for the subsequent AD treatment. The efficiencies of the pre-treatment were tested based on four dependent variables i.e., mixed liquor volatile suspended solids (MLVSS) removal, capillary suction timer (CST) reduction, EPS increment and soluble COD (sCOD) and increment on three independent variables which were current density, electrolyte concentration and electrolysis time. Current density is known to be one of the most important operating variable in EO, as higher current density would allow more electrons to pass by generating more oxidizing species (•OH radicals) to biodegrade the organic pollutants (Moteshaker et al. 2020). One study showed that EO promoted no biodegradability in terms of sCOD at 30–40 mA/cm² (Pasalari et al. 2021). Electrolysis time determines the reaction rate and removal efficiency of the pre-treatment (Heidari et al. 2020). Stagnant sCOD solubilization would gradually start to occur at 90 minutes due to the accumulation of soluble organic matter, which builds up the resistance of the system thus reducing the current (Ye et al. 2016). Lastly, the type of electrolyte and its concentration affect the conductivity of the EO process. For instance, sodium chloride (NaCl) at the range of 1.8–20 g/L, is able to generate oxidant species that have the quality of showing a higher reaction rate with organic compounds (Fajardo et al. 2017). The chloride ions from NaCl would minimize the side effects of anions such as bicarbonate (HCO₃⁻) and sulfate (SO₄²⁻), which would cause resistance to the overall treatment efficiencies. Electrolyte concentration should be limited, as excessive active chlorine species would lead to the annihilation of organic matter (Zhu et al. 2019). The conventional methods of determining the relationship between dependent and independent variable are rather sophisticated and time consuming. Hence, the response surface methodology (RSM), a statistical and mathematical method, was implemented due to its reliability and accuracy. RSM is popular in experiment design, model buildings, the relationship of variables evaluation and condition optimization. The most well-known technique in RSM is central composite design (CCD) for identifying the optimum operating conditions for pre-treatment by making traits need fewer experiment runs (Darvishmotevalli et al. 2019).

The POME WAS were collected from the local palm oil mill located at Ayer Kuning, Perak, Malaysia. The collected WAS were thickened and refrigerated at 4 °C prior to any analysis and pre-treatment. The characteristics of raw POME WAS are listed in Table 1. Different stages of the experiments required different types of chemicals, together with its purposes, which are concluded in Table 2 below. The electrodes used were titanium plates with the dimensions of 150 mm × 30 mm and a thickness of 2 mm.

Pechini's method was chosen for the fabrication of titanium (Ti) electrodes coated with ruthenium oxide (RuO₂). RuO₂ was chosen due to its distinction of chlorine evolution over-potential and weakening of oxygen evolution, therefore generating free radicals for organic compound oxidation (Johnson & Kumar 2020). Three stages are involved in the fabrication process. Firstly, pre-treatment of electrodes using sandpaper and isopropyl alcohol for washing purposes then boiling with 20% chloride acid and 10% oxalic acid in order to remove the titanium's oxide hydroxide coating and improve its roughness at the same time. The second stage is the preparation of precursor solution, which was EG:CA:RuCl₃ with the molar ratio of 10:3:1. The final stage is the coating process, 20 layers of coating was done and completed with the final calcination at 400–500 °C for 1 hour. The morphology of the coated electrodes was analyzed using field emission scanning electron microscopy (FESEM) and energy dispersive X-ray (EDX) spectroscopy.

Electrochemical oxidation was carried out in a 500-mL beaker with an electrode distance of 5 cm throughout the experiment in a fixed parallel manner with an effective electrode area of ∼55 cm², to achieve high solubility of sludge (Ye et al. 2016). A constant speed of 300 rpm was applied throughout the experiment to ensure that the process was maintained at a homogeneous manner and to prevent any settlement of the sludge. The electrolyte used in this study was NaCl, where it has a trait in generating oxidant species that had the quality of showing higher reaction rates with organic compounds (Fajardo et al. 2017).

Standard method 2540E was applied for the measurement of MLVSS and its reduction was calculated. Standard method 5220D was used for the measurement of sCOD level after pre-treatment (APHA 2005). Furthermore, a CST apparatus (model: 304M Triton) was used to measure the dewaterability of the pre-treated sludge in seconds. Protein and polysaccharide are both common indicators for the quantification of EPS. The EPS increment was calculated by the increase in the summation of both protein and polysaccharide. The BCA method and the sulfuric-UV method were used to identify the protein and polysaccharide, respectively. The detection methods for EPS and sCOD were through a UV-vis spectrophotometer (model: JASCO V-730) and were calculated with the equations generated from the calibration curve.

In this study, RSM type CCD was applied for optimization that generated 20 runs (14 non-center points and six center points), the experimental design data together with its responses are listed in Table 4.

The ANOVA outputs were tabulated in Table 5. Coefficient of determination (R²), level of significant (p-value) and the lack of fit test are the most representative indicators in proving the model's significance (Isam et al. 2019). R² and Adjusted R² (R²_adj) were given more attention whereby R² is the extent of the level of response variability reduced through the independent variable of the model. R² reaching 1 is the most desirable value, to have a balanced match between experimental and predicted data. However, a minimum of 0.8 is sufficient (Darvishmotevalli et al. 2019). All of the dependent variables were able to achieve the minimum suggested value of R² i.e. R²_MLVSS = 0.9773, R²_CST = 0.9610, R²_EPS = 0.9230 and R²_sCOD = 0.9044. On top of that, the closeness between R² and R²_adj, i.e. R²_adjMLVSS = 0.9568, R²_adjCST = 0.9259, R²_adjEPS = 0.8537 and R²_adjsCOD = 0.8183 indicated that there were no chances of insignificance within the model (Ken & Sinha 2021).

Moreover, a p-value <0.05 should be assured as it serves as evidence that the model is significant (Darvishmotevalli et al. 2019). The RSM type CCD ANOVA output tables for the four variables (linear, two-way interactions and quadratic) are attached in Tables S1–S4 as supplementary material. From the table, all variables were able to achieve a relatively low p-value where the p-values for MLVSS and CST were <0.0001, whereas the p-values for EPS and sCOD were 0.0002 and 0.0005, respectively. Furthermore, a lack of fit of test should be ‘not significant’ as it proves that the model is able to be fitted. All four dependent variables were able to achieve, however, a probability lack of fit (PLOF) of EPS that showed a lower percentage among the other three variables. This portrays that there was a lower chance that EPS will not be fitted due to noise. Lastly, the measurement of the ratio between signal to noise is termed as the adequate precision (AP) (Moteshaker et al. 2020). It shows the comparison between the ranges of values predicted at design points with the mean prediction error. All of the AP exceeded the minimum value of 4, means that the maneuvering of design space could take place through the models (Isam et al. 2019).

From the surface plots (Figure 3(a)–3(c)), MLVSS removal, CST reduction and EPS increment were shown to be increasing as the current density increased. This can be explained due to the ion species that were formed that possessed the robust affinity to treat the organic matter within the WAS. Higher current promotes more electrons flowing and thus enhancies the rate of oxidation. High resistance within the reaction is not the only factor to be affected. To further explain this, secondary oxidizing agents will be created due to the oxidation of organic matter. These secondary oxidizing agents tend to be weaker and are unable to perform as well as •OH radicals, which will cause mineralization to take place and influence the post-treatment (Pérez-Rodríguez et al. 2019). As mentioned, pre-treatment efficiency is dependent on the amount of oxidizing agents being created, the increase in current density only affects direct oxidation as strong oxidizing agents are formed at the surface of the electrodes. Although, the higher the current density the better the efficiency, particularly in removing the WAS solid content. However, the amount of voltage being applied needs to be taken into consideration as it would be less economical or possibly side reactions might occur that would deteriorate the overall performance, especially on the sCOD solubilization beyond ∼23 mA/cm², as shown in Figure 3(d). This is in close agreement with Pasalari and his co-researchers who found that no biodegradability would be promoted in terms of sCOD in the EO process when the current density reached 30–40 mA/cm² (Pasalari et al. 2021). It is important to understand that the restriction of current density is necessary as voltage increases, heat would be dissipated leading to other side effects (Hu et al. 2021). Not only that, the operating conditions of the EO pre-treatment should be applied to help in solubilizing the organic matter in order to enhance biodegradability and overcome the hydrolysis problem in AD. High current density would lead to the mineralization of organic molecules instead of solubilizing them (Barrios et al. 2021).

Electrolysis time is considered as one of the few vital parameters that would affect the performance of EO. From the surface plots, MLVSS removal and CST reduction were directly proportional towards the electrolysis time, as shown in Figure 3(a) and 3(b), due to the increased level of oxidizing agents (•OH radicals). The electrolysis time had more effect with higher current density, in terms of MLVSS removal. Both of the independent variables had inseparable relationship whereby a balance needed to be achieved when the voltage of the system was higher, and a lower electrolysis time is needed to achieve a certain level of solubilization (Rodriguez et al. 2018). However, there would be decay in organic matter if the duration is too long. The sCOD increment was enhanced by electrolysis time at the beginning of the EO and dropped further after the peak at 30 min, as shown in Figure 3(d). According to Pasalari et al. (2021), sCOD concentration would increase sharply at the beginning of the pre-treatment, which showed that intracellular materials were released into the aqueous phase, thus enhancing the biodegradability of the WAS. This was due to the electromigration of ions at the beginning of the EO. Inorganic matter is released into the liquid phase through solubilization. The accumulation of ions or organic matters takes place at both electrodes and hence causes a blockage that increases the current resistance and restricts the process's ability to take place correctly (Heidari et al. 2020). Due to the increase in current resistance, the current drops gradually throughout the entire process, which affected the performance of the sCOD increment. The decrease in concentration of sCOD most likely occurred at extremely high current density at 44.86 mA/cm² or due to the high electrolysis time at 90 min from the results. This could be explained by the rapid sCOD mineralization of dissolved organic substances compared with solubilization.

Electrolyte (NaCl) concentration was fixed at the middle level set due to its least significant level. NaCl helps in preventing the formation of oxidizing layers by other anions and causes ohmic resistance within the cell and thus prevents the decline in COD removal. However, depletion in sCOD due to mineralization is not desired as it would further affect the biogas generation in the subsequent AD. From the results, a slight drop in sCOD occurred as the NaCl concentration increased, as well as the EPS concentrations. This could be due to the presence of negative charges at the bacteria surfaces that generate a high electrostatic repulsive force that leads to the occurrence of sludge disintegration (Heng et al. 2021). MLVSS removal and CST reduction rate showed no effect on the increase in NaCl concentration.

Pre-treatments were performed aiming to enhance the digestibility of WAS, which in turn improved the performance of AD. The best way to indicate AD performance was through biogas yield and different studies have been carried out showing that pre-treatment was able to improve biogas yield. Huang et al. (2021) showed that an sCOD increment from 29 to 350 mg/L successfully increased the methane yield (from 105 to 134 mL CH₄/g VSS) using an electrochemical pre-treatment (Ti/RuO₂) on municipal WAS. Another study proved that biogas generation from 109 N-L CH₄/kg.VS in non-pretreated WAS increased to 311.9 ± 6 N-L CH₄/kg.VS using electrochemical pre-treatment at a current density of 28.6 mA/cm² (Barrios et al. 2021).

The generation of POME WAS has gradually increased over many years due to the development of the palm oil industry. It is important that the generated sludge is treated before being released into the environment. The EO process using Ti/RuO₂ was used in this study. FESEM surface images and EDS elemental analysis of the Ti/RuO₂ electrodes indicated that the preparation following Pechini's method can produce a uniform and homogeneous coating throughout the titanium plate. An RSM type CCD was successfully applied to describe the EO process using a second-order model. The most significant effects on WAS disintegration and dewaterability were found to be due to current density and electrolysis time, while the influence of NaCl concentration was relatively insignificant. The model exhibited that the suggested model agreed well with experimental results with high R² values (0.9044–0.9773) for all the studied dependent variables. The optimum operating conditions were in the range of current density 17–27 mA/cm² and electrolysis time 55–75 minutes, at a fixed concentration of NaCl (10 g/L), resulting in MLVSS removal >20%, CST reduction >43%, EPS increment <19% and sCOD increment >25%. Consequently, these results proved that Ti/RuO₂ could play an important role in the pre-treatment of POME WAS in sludge disintegration and adequate dewaterability, for the subsequent AD to further enhance the solid removal and biogas generation.

The authors are thankful to Universiti Tunku Abdul Rahman (UTAR), Malaysia for providing facilities and support for this research. This work was funded by the UTAR Research Fund under project number IPSR/RMC/UTARRF/2019-C1/G01.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.