Efficiency of an up-flow Anaerobic Sludge Blanket reactor coupled with an electrochemical system to remove chloramphenicol in swine wastewater

The application and design of treatment systems in wastewater are necessary due to antibiotics’ potential toxicity and resistant genes on residual effluent. This work evaluated a coupled bio-electrochemical system to reduce chloramphenicol (CAP) and chemical oxygen demand (COD) on swine wastewater (SWW). SWW characterization found CAP of,10 μg/L and 17,434 mg/L of COD. The coupled system consisted of preliminary use of an Up-flow Anaerobic Sludge Blanket Reactor (UASB) followed by electrooxidation (EO). The UASB reactor (primary stage) was operated for three months at an organic load of 8.76 kg of COD/md and 50 mg CAP/L as initial concentration. In EO, we carried out a 2 (time operation and intensity) factorial design with a central composite design; we tried two Ti cathodes and one anode of Ti/PbO2. Optimal conditions obtained in the EO process were 240 min of operation time and 1.51 A of current intensity. It was possible to eliminate 44% of COD and 64.2% of CAP in the preliminary stage. On bio-electrochemicals, total COD and CAP removal were 82.35 and .99.99%, respectively. This coupled system can be applied to eliminate antibiotics and other organic pollutants in agricultural, industrial, municipal, and other wastewaters.

antibiotic in many countries, but it is still used and found in wastewater due to its low cost, which leads to bacterial resistance problems (Tan et al. 2018).
Various efforts have been made to eliminate antibiotics by applying biological treatments, some with low yields because antibiotics kill or inhibit bacterial growth, affecting these systems (Huang et al. 2021). Likewise, eliminating these compounds becomes more complex in the case of wastewater in the presence of high concentrations of organic matter (Huang et al. 2021). Zheng et al. (2018) used a sequential anaerobic digestion system with intermittent aeration for the elimination of 11 antibiotics in swine wastewater, finding 87.9% elimination at a load low of 0.17 + 0.041 kg COD/m 3 d, of which 30.4% was due to the absorption of sludge and 57.5% to biodegradation. Cheng et al. (2018) used a combined biological filter (aerated/anaerobic) to eliminate nine antibiotics detected in pig wastewater with concentrations up to 0.192 mg/L, reaching eliminations of .82%. On the other hand, the up-flow Anaerobic Sludge Blanket reactor (UASB) is widely used to treat swine wastewater. It can eliminate concentrated effluents of organic matter to high-rate (Torkian et al. 2003;Kim et al. 2013;Pérez-Pérez et al. 2016;Mainardis & Goi 2019;Mainardis et al. 2020;Oliveira et al. 2020;Vassalle et al. 2020) and antibiotics in different grades. Antibiotic removal depends on the initial concentrations, antibiotics classes, types of bioreactors, and operating conditions (Cheng et al. 2018). Sorption and biodegradation are two of the most important mechanisms for eliminating antibiotics from wastewater (Cheng et al. 2018).
Due to the incomplete degradation of antibiotics in the biological phase, it is necessary to add a post-treatment to achieve their mineralization; this post-treatment can be electrooxidation, which is easily automatized and does not require the addition of chemical compounds (Moreira et al. 2017;Garcia-Segura et al. 2018;Romero-Soto et al. 2018). On the electrooxidation process, organic matter is directly oxidized through the •OH radicals that are generated on the anode surface (more than 90%); likewise, other oxidizing agents are generated indirectly, such as: HClO, H 2 S 2 O 8 , H 2 O 2, and organic matter is mineralized to CO 2 and H 2 O (Drogui et al. 2007;García-Gómez et al. 2014;Romero-Soto et al. 2018). Existing different electrodes materials are used for electrochemical oxidation; notably, the Ti/PbO2 electrodes are highly efficient, economical and lead release during electrolysis is negligible (Li et al. 2021). Hou et al. (2019) investigated antibiotics elimination in UASB reactor coupled anoxic-oxic tank and advanced oxidation technologies (UV irradiation, ozonation, Fenton, and Fenton/UV, separately) for the elimination of antibiotics in pharmaceutical wastewater with an organic load of 6.5 kg/m 3 d. In this work, the elimination of CAP and COD from swine wastewater in the presence of a high load of organic matter (8.76 kg/m 3 d) was investigated using a system integrated by a UASB reactor coupled to electrooxidation to achieve complete degradation of the pollutants present. For this reason, the objective of this research was to evaluate the degradation efficiency of CAP and COD by a UASB system coupled to electrooxidation.

Experimental unit and operation conditions
Preliminary treatment was conducted in an 800 mL cylindrical UASB, made of Plexiglass material, as shown in Figure 1. The dimensions of the anaerobic reactor were: diameter of 7 cm, height of 23 cm, and a conical base of 5 cm in height. For experimental development, the anaerobic reactor was inoculated with 180 g of anaerobic biomass from a region brewing at the south of Sonora, Mexico. A peristaltic pump (Masterflex ® ) was used to hold a load of 4.38 kg of COD/m 3 d (HRT of 35.6 h); during acclimatization for 66 days (at this stage, the reactor was not monitored) and 8.76 kg of COD/m 3 d (HRT of 17.8 h) during monitoring (for 97 days). The same batch of water sampled was used in the influent during experimentation to maintain the equal organic load. Likewise, prior COD analyzes were carried out on the influent, and if this decreased, the HRT was adjusted. In the second stage, a cylindrical electrochemical reactor ( Figure 1) made of plexiglas with a diameter of 9 cm, height of 22 cm, and a conical base of 5 cm in height with a working volume of 1,000 mL and three mesh electrodes were used. An anode of titanium/lead dioxide (Ti/TiPbO 2 ) and two cathodes of titanium (Ti) placed cathode-anode-cathode, with an inter-electrode distance of 1 cm, were used. Ti/TiPbO 2 anode was used because it is economical material and can achieve efficiencies as high as the boron-doped diamond (BDD) and other materials such as Ti/SnO 2 have been reported that presented passivation on the anode surface during the oxidation of some organic pollutants (García-Gómez et al. 2014). Electrodes dimensions were all 15 cm in length and 6.5, 4.5, and 2.5 cm in diameter, with 888, 706 and 342 cm 2 of active surface area, respectively. The current intensity was applied using a BK Precision ® of Triple Output DC Power Supplies, model 1673 (Yorba Linda, California, USA). The flow was operated by peristaltic pumps (Masterflex ® ) in ascendant recirculation (45 mL/min). All experiments were conducted at room temperature at 25 + 2°C.

Swine effluent
Swine wastewater was collected in a general effluent from a farm in the north of Obregon city in the state of Sonora, Mexico; it was kept at a temperature of 4°C before being used. The swine farm does not have a treatment system and the effluents are discharged into the nearest drain. For the UASB reactor, the residual water was characterized previously and enriched with chloramphenicol (firstly diluted and subsequently raw swine wastewater was used). During the acclimatization stage, the solution was prepared using a proportion of 1:2 of swine wastewater and distilled water at pH of 7, enriched with different doses of chloramphenicol (CAP, !99.0% Sigma Aldrich, USA) weekly, from 5 to 50 mg/L (during two months) hold a load of 4.38 kg of COD/m 3 d since its installation. The suspension was agitated with a stirring bar for at least 1 hour to ensure complete dissolution. After the acclimatization, raw wastewater was added to an anaerobic reactor with a CAP concentration of 50 mg/L (intermediate concentrations from Tan et al. (2018) and Chen et al. (2015) studies), a load of 8.76 kg of COD/m 3 d. The effluent of this stage was collected and preserved at a temperature of 4°C for later use as an influent of the electrooxidation process (EO) under optimized operating conditions. Samples were collected in the influent and effluent of the biological system and effluent of the EO treatment for further analysis.

Experimental design for EO
CAP degradation in the EO process with raw wastewater was performed using a response surface methodology (RSM). Treatment time (X 1 ) and current intensity (X 2 ) were the independent variables of the model. X 1 and X 2 were selected because preliminary studies were the variables with a major effect on response variables. Likewise, in a real wastewater treatment application, it is difficult to change the temperature and pH to be treated when there are large volumes of water and when working with biological systems; for this reason, these factors were not evaluated (Romero-Soto et al. 2018). For the time range, values below and above the Chen et al. (2015) study were taken with a current intensity between 4 and 7.5 times lower than that investigation (1 A (2.7 mA/cm 2 ) and 2.5 A (6.8 mA/cm 2 )) to verify if the antibiotic could be removed at a lower current intensity. The domain for X 1 was 60-240 min (U i,0 ¼ 150 min), and for X 2 was 1-2.5 A (U i,0 ¼ 1.75 A). Two levels had been assigned to each factor (2 2 plan), leading to 13 experiments comprised of four runs for the factorial design and nine runs for the central composite design, including five replicates at the center point and four runs for the extreme high and extreme low (Table 1). CAP and COD removal (percentage) and energy consumption were the three investigated responses (response variables). Design Expert ® 7 (version 7.0.0) was used to generate the quadratic polynomial model. The optimal operating conditions were applied in the residual effluent of the biological treatment as post-treatment. CAP removal (R) and energy consumption (E) were calculated using Equations (1) and (2): where C o is the initial concentration of CAP (mg/L), C f is the final concentration of CAP (mg/L), I is current intensity (A), U is electrical potential (V), t is treatment time (h), and V treated water volume (m 3 ).
The following second-order equation gives the predicted response in all experimental fields (Equation (3) where Y is the experimental response and b o is the average of the experimental response. Coefficients b i , b ii , and b ij are the linear, quadratic, and interaction effects between factors i and j for the response Y.

Analytical details
Chloramphenicol analysis was performed by HPLC (Agilent™, 1260, USA). Liquid chromatography separation was carried out in a ZORBAX 300-extender-C18 column 4.6 Â 150 mm, 3.5 μm (Agilent™, USA). For CAP, methanol: water, 65:35 (v/v) with a flow rate of 1 mL min À1 at λ 280 and 225 nm in a retention time (RT) of 1.99 min was used. Two mL of sample and 2 mL of buffer solution (methanol: water, 65:35 (v/v)) were mixed, stirred and after 30 min of repose, 1 mL of the supernatant was taken and left to evaporate over 12 hours, the dried sample was reconstituted with 1 mL of methanol HPLC (JT Baker™) and filtered with a membrane of 0.22 μm (Merck Millipore, USA). The antibiotic detection limit was 10 μg/L. Chemical oxygen demand (COD), NO x (nitrate NO 3 À and nitrite NO 2 À ), ammonium (NH 4 þ ), and orthophosphate (PO 4 3À ) were determined using the standard methods (APHA 1999). For the colorimetric analysis, a brand spectrophotometer (Thermo Scientific ® Waltham, MA) was used. The pH was measured by a pH meter (HI 2550, HANNA ® Instruments).

Physicochemical characterization of swine wastewater
Swine wastewater was characterized before treatment; Table 2 shows these results. Nutrient concentration in swine wastewater depends on the animal weight, livestock practices (feeding, frequency of cleaning, etc.), and treatment applied before being discharged to receiving sources (Garcia- Sanchez et al. 2016). In this study, the ammonia nitrogen and total COD concentration were 386.76 + 46 mg-N/L and 17,000 + 450 mg/L, respectively. The sampled swine farm has a complete cycle; it has bellies, obtains piglets and stallions, and fattens them until sent to the slaughterhouse (closed cycle); which means that there is the presence of different types of organic waste (drugs) and nutrients. Likewise, the high concentrations of organic matter show that the biodigester used to treat the general farm effluents is not working properly (is not wholly under anaerobic conditions and hydraulic residence times have not been established). Then, the farm wastewater effluent is directly deposited in the closest receiving sources without treatment. However, the values obtained are low compared with those found by Zhao et al. (2016), who found 14,000-20,000 mg/L and 2,500-4,000 mg-N/L of COD and total ammonia nitrogen, and the results of Garcia-Sanchez et al. Oxytetracycline was the antibiotic mainly detected, with a concentration of 114 ng/L, which is a kind of tetracycline antibiotic, and had been detected in other studies in a concentration ranging from 2 ng/L to 68 mg/L in stream waters in a small catchment area with livestock farms (Matsui et al. 2008). This study focused on CAP only because this antibiotic is already prohibited in many countries due to its adverse health effects, such as aplastic anemia, and because it is bioaccumulative and has been found in the tissues of animals destined for human consumption (Zhang et (2 2 ), and four other experiments carried out around the experimental field (low and high extremities). The corresponding second-order polynomial equations models are given by Equations (4)-(6) for CAP and COD removal and energy consumption, respectively: Y 1 ¼ 83:05 þ 17:21X 1 þ 12:20X 2 À 4:84X 1 X 2 À 3:69X 2 1 À 2:26X 2 2 (4) Y 2 ¼ 39:14 þ 10:30X 1 þ 4:11X 2 þ 1:42X 1 X 2 þ 5:42X 2 1 þ 1:30X 2 2 (5) Y 3 ¼ 17:32 þ 10:65X 1 þ 9:34X 2 þ 5:70X 1 X 2 À 0:019X 2 1 þ 0:91X 2 2 (6) Coefficients of equation models were calculated using the half difference between the arithmetic average of the response values when the variable is code at the levels À1 and þ1 (García-Gómez et al. 2014; Romero-Soto et al. 2018). Thus, a positive coefficient will positively affect the response, while a negative coefficient will have a negative effect on the response.
Coefficient b o ¼ 83.05 represents the average CAP removal obtained from all the experiments. The coefficient b 1 ¼ þ17.21 corresponding to the operation time (X 1 ) indicates that the CAP removal increased on average by 34.42% (2 Â 17.21) when the experimental time varied from 60 to 240 min; likewise, it is the most influential variable on the antibiotic removal. The second one is the current intensity (X 2 ). According to coefficient b 2 ¼ þ12.20, the elimination of the antibiotic increased on average by 24.40% (2 Â 12.20) when the intensity passed from 1 to 2.5 A.
For the COD removal, b o ¼ 17.32 was the average removal obtained from all the experiments. The coefficient b 1 ¼ þ10.30 indicates only 20.60% COD removal increases when the experimental time varied from 60 to 240 min, and b 2 ¼ þ4.11 suggests that 8.22% of the organic matter is eliminated when the current increased from 1 to 2.5 A.
For energy consumption, b o ¼ 39.14 was the average energy consumed in all the experiments. The coefficient b 1 ¼ þ10.65 indicates that 21.3 kWh/m 3 of energy consumption increases when the experimental time is varied from 60 to 240 min; while b2 ¼ þ9.34, means an energy consumption of 18.68 kWh/m 3 when the current increased from 1 to 2.5 A.
The CAP and COD removal and energy consumption is shown by the response surface plot depicted in Figure 2. As observed, the removals could reach !99.99% if the times in both response variables increased. The maximum reduction achieved was 99.75 and 67.84% and the minimum was 48.26 and 29.012% for the CAP and COD removal, respectively. The best result for CAP removal (99.75%) was: 150 min and 2.81 A with an energy consumption of 32.32 kWh/m 3 . In comparison, the removal of COD (67.84%) was at 240 min and 2.5 A with an energy consumption of 44 kWh/m 3 (experiment 13 and 4, Table 2). Analysis of variance (ANOVA), given in Table 3, shows that CAP F-value of 37.11 implies the model is significant. On CAP removal A and B are substantial variables because the p-value is ,0.0001, and its F value is more meaningful (58.57 and 116.62) than the F value of the model (37.11). On COD removal, only time (B) is significant because the p-value is 0.0022 (,0.05), and the F value (22.09) is higher than the F value of the model (6.24). Intensity (A) is not significant (P.α). On energy consumption, A and B are significant, p-value is ,0.0001.
The coefficient correlation value (R 2 ¼ 0.9636) means that the empirical model could not explain only 3.64% of the total variation. 'Prob . F' less than 0.05 indicates model terms are significant. A low dispersion of data observed in Figure 3 means a good fit of the removals obtained and the predicted. For COD removal, the coefficient correlation value (R 2 ¼ 0.8168) implies that the empirical model could not explain 18.32% of the total variation, which means the Lack of Fit is significant. R 2 should be at least 0.80 for a good fit of a model (Joglekar & May 1987). High dispersion of the data is observed in Figure 3(b) due to the low correlation coefficient obtained. For energy consumption, the F-value of 2485.38 implies the model is significant. There is only a 0.01% chance that an F-value this large could occur due to noise. The coefficient correlation value (R 2 ¼ 0.9994) means that the empirical model could not explain 0.06% of the total variation. It implies low dispersion of data, observed in Figure 3(c), which means a good fit for the obtained and predicted energy consumption.

Optimization of electrochemical system
The importance of the variables and their interactions was calculated using Pareto analysis. This analysis estimates the effect (percentage) of each variable studied on the response (García-Gómez et al. 2014). For this study, treatment time was the most  significant contribution to the CAP and COD removal with 75.16%, followed by the current intensity with 18.70%. The high contribution for the time and current intensity is due to these parameters controlling hydroxyl radicals and other oxidants produced during the experimentation, such as HClO, H 2 S 2 O 8 , H 2 O 2 , etc. (Drogui et al. 2007;García-Gómez et al. 2014;Moreira et al. 2017;Romero-Soto et al. 2018). Chemical reactions are shown in Equations (7)-(9): A particular compromise was established to determine the optimum conditions of the process, CAP and COD removal, and the energy consumption (priority for CAP removal, followed by COD elimination and finally reduction energy consumption). The objective is more likely to maximize the CAP and COD removal than to minimize energy consumption. For optimization, there are five levels of importance in the statistical package (Design-Expert software), which were selected based on the order of priority of the response variables. Five points were assigned to CAP degradation, four to COD degradation, and three to energy consumption. Based on these preferences and other parameters of the model, the Design-Expert software could generate the following optimum conditions: treatment time ¼ 240 min and current intensity ¼ 1.51 A. The predicted CAP and COD removal and energy consumption were 92.64 and 52.23%, and 23.16 kWh/m 3 , respectively. Therefore, the overall desirability for this work is 0.667 as all responses are predicted to be within the desired limits. These conditions were applied in the effluent of the biological system.
There are different works where this technology has been applied to remove other contaminants (Table 4). CAP and COD removals achieved are high; however, the time and energy consumption is required too. Because swine wastewater had high organic matter concentration, which competes in removing other compounds, this proves that an increase in organic load may negatively impact antibiotic reduction (McAdam et al. 2011;Tran et al. 2013;Yoo et al. 2020).

Degradation of CAP and COD in the coupled biological-electrochemical system
A UASB reactor was used as a pretreatment, which was monitored for three months after acclimatization. The porcine wastewater was treated through an anaerobic stage and after by electrooxidation under optimal conditions (240 min and 1.51 A); this coupling was made to eliminate the high organic matter concentration found in the studied effluent using a UASB reactor followed by an electrooxidation system to complete antibiotics oxidation. Figure 4 shows the removals achieved by the UASB reactor after 97 days of operation. The initial concentration of CAP was 50 mg/L and 17,000 + 75 mg/L of COD. CAP concentration in this study was high; however, we wanted to verify that biological systems could cope with high loads of antibiotics and organic matter and because these compounds were in the order of several mg/L in effluents of hospitals, pharmaceutical production, and aquaculture farms (Kummerer 2001;Chelliapan et al. 2006;Meng et al. 2015;Hou et al. 2019). This combined system can be used to treat other wastewater effluents.
The maximum COD removal in the UASB reactor was 44 + 1.2% at an organic load of 8.76 kg of COD/m 3 d (shown in Figure 4), which is low compared with the results found by Zhao et al. (2016), who reached 83.6% in swine wastewater at organic load (VOL) of 2.29 kg/m 3 d (almost four times less than this work). Lo et al. (1994) obtained removal efficiency of COD from 95% in swine manure at a VOL of 1.65 kg/m 3 d. However, the COD removal decreased to 57% at VOL of 3.5 kg/m 3 d. Campos et al. (2005), with the same configuration at 20 h HRT and VOL of 1.42 kg COD/m 3 d, obtained removal efficiency of COD of 84.0%. The low efficiency obtained in the biological system may be because only 60% is a biodegradable matter (Andreadakis 1992). Likewise, a high concentration of ammonia can inhibit the activity of microorganisms and hence hinder the maximum organic loading rates in the anaerobic processes (Zhao et al. 2016). Garcia-Sanchez et al. (2016) mentioned that the presence of antibiotics does not affect COD removal if the reactor is acclimated. Zaiat et al. (2001) and Sanchez et al. (2005) recommended increasing the solids retention time (SRT) to improve the organic matter removal due to the UASB not being suitable for the treatment of pig manure, based on poor yield for low HRT and high VOL. The main mechanisms of removal of veterinary antibiotics in anaerobic conditions are sludge sorption and biodegradation. Greater than 60% of antibiotics in the influent are biodegraded, 24% are adsorbed by sludge, and 15% of the antibiotics remain in the effluent (Zheng et al. 2018). The combined system of UASB and EO reached 82.35 and .99.9% of COD and CAP removal, respectively (shown in Figure 5). No studies have coupled this type of biological system with EO to  eliminate antibiotics in swine wastewater. However, the most significant degradation of organic matter was carried out in the biological stage; likewise, the CAP was eliminated mainly during the EO because microorganisms tend to utilize the easily degradable organic substances rather than the refractory antibiotics maintaining their metabolism (McAdam et al. 2011;Zheng et al. 2018). The perspectives of the combined system are that it can be applied in the elimination of antibiotics  and other pollutants in wastewater effluents of different activities with high organic loads such as aquaculture, pig farm, pharmaceutical, hospital, and others. Likewise, the coupling of the UASB system with an electrochemical reactor can contribute to reducing energy consumption through the use of biogas produced in the anaerobic digester. In addition, UASB allows reducing organic loads before being treated in an electrochemical system and is recommended to use two UASB reactors before the electrochemical system if the wastewater has a COD ! 6,500 mg/L, as in this study. Another alternative to reduce energy consumption in advanced treatment is solar panels use as an energy supply, especially in zones with a larger area such as aquaculture and pig farms.

CONCLUSION
A coupled biological-electrochemical system for the elimination of CAP and COD in swine wastewater was used. The swine effluent sampled presented a CAP concentration below to limit detection, for which the antibiotic was added (50 mg/L of CAP). The UASB reactor (at an organic load of 8.76 kg of COD/m 3 d) reached only 44 and 64.2% of COD and CAP removal. Likewise, the bio-electrochemical combination system increased elimination to 82.35 and .99.9% of COD and CAP, respectively (using optimal conditions in EO process: treatment time ¼ 240 min and current intensity ¼ 1.51 A). Electrolysis time and current intensity are more significant variables on this type of effluent treatment due to the high organic matter. The methane capture from the anaerobic system and its conversion into electricity can be applied to an electrochemical system to reduce operating costs. This study underlines the effectiveness of UASB and electrooxidation for chloramphenicol removal in swine wastewater. It could treat other wastewater types such as municipal, agriculture, aquaculture, and industrial.

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