Aquacultural Engineering 94 (2021) 102174

Contents lists available at ScienceDirect

Aquacultural Engineering
journal homepage: www.elsevier.com/locate/aque

Practical use of response surface methodology for optimization of

veterinary antibiotic removal using UV/H2O2 process
K. Jutarvutikul a, C. Sakulthaew b, C. Chokejaroenrat a, *, A. Pattanateeradetch a, S. Imman c,
N. Suriyachai c, T. Satapanajaru a, T. Kreetachat c
a
Department of Environmental Technology and Management, Faculty of Environment, Kasetsart University, Bangkok, 10900, Thailand
b
Department of Veterinary Technology, Faculty of Veterinary Technology, Kasetsart University, Bangkok, 10900, Thailand
c
Integrated Biorefinery Excellent Center (IBC), School of Energy and Environment, University of Phayao, Phayao, 56000, Thailand

A R T I C L E I N F O A B S T R A C T

Keywords: Three of the most commonly used veterinary antibiotics—enrofloxacin (ENR), sulfamethoxazole (SMX), and
Antibiotic removal oxytetracycline (OTC)—were chosen as representative antibiotics for UV/H2O2 treatments. The objective was to
Advance oxidation process determine the optimization of UV/H2O2 to remove antibiotics from aquaculture discharge water using response
H2O2 sequential addition
surface methodology. The degradation of the antibiotics was investigated under varying UV/H2O2 conditions in
Response surface methodology
environments with different levels of pH, water matrices, humic acid, and constituent ions. The degradation
UV-assisted chemical oxidation
results demonstrated that increasing the H2O2 dosage facilitated ENR degradation at a neutral pH while facili­
tating degradation of SMX and OTC at a slightly acidic pH. The optimum removal conditions for ENR, which was
used in all influential effect experiments and the contact tank experiments, was obtained at 10 mM H2O2, a
pretreated COD of 87.51 mg L− 1, and an initial pH of 6.15. Among the tested anions, only the presence of Cl-
showed slight positive effects on ENR degradation, due to the generation of secondary active radicals. During the
reaction, the hydroxyl radical ( OH) was present at a higher pH while singlet oxygen (1O2) was slightly present at
•

a lower pH. The experimental results from H2O2 sequential addition indicated that freshly added H2O2 could
quench the recently generated OH and therefore a high H2O2 concentration with frequent adding was not
•

necessary. Our contact system reduced the ENR concentration in both the effluent reservoir and in the UV
irradiation zone. The overall results supported the use of the UV/H2O2 system to treat remnant antibiotics in the
discharge water.

1. Introduction detection of antibiotic residues in animal muscles from aquaculture

farms located close to the contaminated area (Zhong et al., 2018). Some
The development of medicinal technology combined with the antibiotics are even present in bioactive metabolite forms as well as the
development of intensive animal farming has exponentially increased parent compounds such as enrofloxacin (ENR), and ciprofloxacin (CIP)
the use of veterinary antibiotics. Unlike human antibiotics, veterinary (Parpounas et al., 2017). Antibiotic residues can negatively impact on
antibiotics tend to be misused because the farmers use antibiotics not the microorganism community and nutrient cycles as its long-term
only therapeutically but also prophylactically to improve productivity exposure has been considered responsible for the proliferation of
(Spellberg et al., 2013), especially in a developing country like Thailand antibiotic-resistance genes which can be self-developed and transferred
(Holmström et al., 2003). More than 30–90 % of initial active forms of among local microorganisms in the environment (Bergeron et al., 2017).
antibiotics were reported to be released to the nearby watersheds This occurrence usually involves incomplete antibiotic removal from
through livestock and aquaculture farming discharge water that usually conventional wastewater treatment plants (Watkinson et al., 2007).
contained animal waste excretion without any pretreatment (Khodadadi Although chlorination is usually applied following conventional water
et al., 2018; Chen et al., 2018). Such discharge has resulted in the treatment, its production of toxic byproducts is a considerable concern

* Corresponding author.
E-mail addresses: kornkanok25.j@gmail.com (K. Jutarvutikul), cvtcns@ku.ac.th (C. Sakulthaew), chanat.c@ku.ac.th (C. Chokejaroenrat), apiladda.pa@ku.th
(A. Pattanateeradetch), saksit.im@up.ac.th (S. Imman), nopparat.suri26@gmail.com (N. Suriyachai), fscitus@ku.ac.th (T. Satapanajaru), torpong.kr@up.ac.th
(T. Kreetachat).

https://doi.org/10.1016/j.aquaeng.2021.102174
Received 11 February 2021; Received in revised form 6 June 2021; Accepted 9 June 2021
Available online 13 June 2021
0144-8609/© 2021 Elsevier B.V. All rights reserved.
K. Jutarvutikul et al. Aquacultural Engineering 94 (2021) 102174

for secondary contamination. alcohol and p-benzoquinone were obtained from Acros Organics (Wal­
The UV/H2O2 oxidation process is an efficient, advanced oxidation tham, MA, USA). Meat extract was purchased from Himedia (Mumbai,
process (AOP) that generates nonselective and highly oxidative hy­ India). Peptone and hydrochloride acid (HCl, 37 %) were purchased
droxyl radicals. Although a UV/H2O2 system has been proven very from RCI Labscan Ltd. (Bangkok, Thailand). Phosphoric acid (H3PO4, 80
effective in removing antibiotics, there has been only limited informa­ %) was purchased from QrëC (Auckland, New Zealand). Sodium per­
tion available on its use to successfully process antibiotic-contaminated oxodisulfate (Na2S2O8) was purchased from Applichem GmbH (Darm­
discharge water from animal farms. The efficiency of the process mostly stadt, Germany).
relies on using an oxidative source such as H2O2 and UV. However, Synthetic wastewater was prepared using chemicals with modifica­
increasing the UV irradiation level is not economical and fails to reach tion to Klamerth et al. (2010), as listed in Table S1. This was used as the
acceptable concentrations, and a high dose of H2O2 could result in COD initial stock solution equivalent to a COD level of 52 mg L− 1,
corrosion in equipment (Ngumba et al., 2020; Sun et al., 2019). excluding the initial concentration of tested antibiotics. In the optimal
Interactions between most operating variables need to be investi­ experiments, COD was diluted to the desired concentration if needed.
gated to evaluate optimization of the UV/H2O2 process, but this requires
a large number of runs, is time-consuming, and is not economically
2.2. Chemical analyses
feasible. Therefore, response surface methodology (RSM) is usually
selected to evaluate optimization of the UV/H2O2 process using different
Temporal changes in the antibiotics were analyzed using high per­
interactions of variables. This involves varying independent variables
formance liquid chromatography (HPLC; 600E, Waters, Milford, MA,
depending upon the influential effects while other parameters are kept
USA) equipped with a UV detector (2487, Waters, Milford, MA). A
constant to determine the optimum treatment conditions. In most
mixture of 0.1 % (v/v) acetic acid and acetonitrile (80:20) was used as
related RSM studies for catalytic processes, H2O2 dose, UV dose, contact
the mobile phase for the ENR and OTC analysis and 60:40 for the SMX
time, and pH, are usually selected as the individual effects of the process
analysis. To obtain better peak intensity, the injection volumes were set
variables. However, COD has been an overlooked independent variable
at 20 μL for SMX and OTC and at 10 μL for ENR, with detection wave­
as it can be controlled from previous conventional treatment and often
lengths at 265, 354, and 280 nm, respectively. In addition to the changes
COD has become an adverse effect in decreasing contaminant degra­
in antibiotics, COD was measured using the closed reflux method
dation performance. Ranges in COD in aquaculture farming systems can
(APHA, 1998).
vary from 22 to 111 mg L− 1 or more depending on the animal culture
intensity (Pedersen et al., 2012). A COD higher than 100 mg L− 1 can
even provide a suitable niche for local bacterial communities, thus 2.3. Experimental setup and sampling
resulting in a positive impact on removing a target contaminant (Alju­
boury et al., 2017; Wei et al., 2020). Antibiotic degradation experiments were conducted in a custom-
Quinolone, sulfonamide, and tetracycline are the three most- made stainless-steel housing with four UV Lamps (G4T5, Eiko Global,
commonly used antibiotic groups in aquatic veterinary practice in Shawnee, KS, USA; 4 W; 254 nm) and magnetic stirrers (C-MAG HS7,
many countries, especially in Asia, and thus are contaminants wide­ IKA®-Werke GmbH & Co. KG, Staufen, Germany) which allowed us to
spread throughout the region (Suzuki and Hoa, 2012). A report in 2013 conduct experiments in quadruplicate (see inset in Fig. 1). We used a
showed that among these three well-known antibiotics (ENR; 69 %, 150 mL quartz cell as an experimental unit containing 100 mL of 0.1 mM
SMX; 44 %, OTC; 30 %), ENR was the most used and most frequently antibiotic (ENR, SMX, or OTC). At each preselected sampling time, 700
found antibiotic in Asian aquaculture-receiving watersheds and in nat­ μL of aliquots were filtered through a 0.45 μm PTFE syringe filter,
ural flowing water (Rico et al., 2013). However, the practical use of RSM quenched with 100 μL of 3000 mg L− 1 sodium thiosulfate, and trans­
has been overlooked for determining the optimal condition for treating ferred to an HPLC vial. Then, samples were stored at 4 ◦ C before HPLC
antibiotics. In the current study, ENR, SMX, and OTC were selected as analysis.
representative veterinary antibiotics to be tested with a UV/H2O2 sys­ The UV catalytic tests were conducted in the above set-up with 0.1
tem. The objective of this study was to use the RSM method to optimize mM ENR in DI water. We used S2O2− 8 or H2O2 at 0.5 mM for the ho­
the performance of the UV/H2O2 system for the removal of well-known mogeneous catalytic system, and we used UV/TiO2 or UV/ZnO at 0.1 g
veterinary antibiotics. Accordingly, optimal condition results were uti­ 100 mL− 1 solution for the heterogeneous catalytic system. Samples were
lized in influential effect testing experiments and in a contact tank sys­ collected and quenched as discussed earlier. However, here, we moni­
tem that was designed to comply with the discharging event and water tored changes in COD using sacrificing samples at each collection.
conditions. In all experiments, we used both the antibiotic removal efficiency
and relative concentration C/Co to evaluate the system performance
2. Materials and methods under different varying conditions. All experiments were conducted in
triplicate unless stated otherwise. After completing each treatment,
2.1. Chemicals temporal changes in the antibiotic or COD were determined and the
antibiotic removal efficiency was calculated [Eq. 1]. The antibiotic
Enrofloxacin (C19H22FN3O3; ENR), sulfamethoxazole (C10H11N3O3S; degradation observed rate constant (kobs) was mostly used throughout
SMX), oxytetracycline (C22H24N2O9; OTC), hydrogen peroxide (H2O2) the study to compare the system performance at different condition [Eq.
(30 %), sodium hydroxide (NaOH), and humic acid (a mimic natural 2–3]:
organic matter; NOM), were purchased from Merck (Darmstadt, Ger­
[Co − Ct ]
many). Acetonitrile (CH3CN) of HPLC grade and acetic acid (CH3COOH) Antibiotic removal efficiency (%) = × 100 (1)
were purchased from Honeywell Burdick & Jackson (Ulsan, Korea). Ct
Sodium chloride (NaCl), magnesium sulfate heptahydrate Ct
(MgSO4.7H2O), calcium chloride (CaCl2), potassium chloride (KCl), Relative concentration; = e− kobs t
(2)
Co
dipotassium phosphate (K2HPO4), calcium sulfate dihydrate
(CaSO4.2H2O), and tert-butanol were purchased from Carlo Erba (Milan, Ct
Italy). Barium chloride (BaCl2), sodium thiosulfate (NaS2O3), and urea ln = − kobs t (3)
Co
((NH2)2CO) were purchased from Ajax Finechem (Auckland, New Zea­
land). Ethyl alcohol anhydrous (99.9 %) and isopropyl alcohol were where Co and Ct are the initial concentration and the concentration at
purchased from Daejung chemicals (Siheung, South Korea). Furfuryl time t (min), respectively.

2
K. Jutarvutikul et al. Aquacultural Engineering 94 (2021) 102174

Fig. 1. Schematic diagram of UV/H2O2 contact tank.

2.4. Response surface methodology experiment

Y = β0 + β1 X1 + β2 X2 + β3 X3 + β12 X1 X2 + β13 X1 X3 + β23 X2 X3 + β11 X12
We performed the RSM experiment to determine the optimization of + β22 X22 + β33 X32
three independent variables—H2O2 concentration (X1 ; mM), initial pH (4)
(X2 ) and COD concentration (X3 ; mg L− 1)—of the UV/H2O2 system using
antibiotic removal percentage as a response variable. To effectively where Y is the predicted response or antibiotic removal efficiency, β0 is
design the range for each variable, we tested each effect individually: the constant, β1 , β2 , and β3 are linear coefficients, β12 , β13 , and β23 are
H2O2 (0–10 mM), initial pH (3–11) and COD concentration (0–116 mg cross factor interaction coefficients, and β11 , β22 , and β33 are quadratic
L− 1). The experiment for the control of H2O2 variation was set with no coefficients. By conducting RSM experiments and the analysis of vari­
H2O2 addition in the DI water, while the experiments for the control of ance (ANOVA) analysis at the 0.05 significance level, we obtained the
pH variation and COD variation were set with a constant H2O2 con­ optimum conditions for ENR, OTC, and SMX removal. Because ENR is
centration of 7.5 mM, also in the DI water. The Minitab 18 statistical considered the major veterinary antibiotic and is frequently found in
software (Minitab, LLC, State College, PA, USA) was used including a natural receiving water (Wei et al., 2012; Zhong et al., 2018), we
Box-Behnken design based on the results from previous experiments to selected ENR for further treatability and applicability experiments.
delineate 15 combinations of the three reaction parameters for one
antibiotic (Table 1). Samples were periodically collected at 0, 5, 15, 30, 2.5. Active radical experiments
45, 60, and 90 min and the sampling protocol was followed. The in­
teractions between the independent and response variables were pre­ To test the predominant radical species in the UV/H2O2 system with
sented graphically and fitted to a quadratic model [Eq. 4]: varying pH (3–11), we spiked either tert-butanol (TBA), furfuryl alcohol
(FA), or p-benzoquinone (PBQ) in each unit containing 100 mL ENR
solution to maintain the final concentration of alcohol at 500 mM. A
consistent sampling protocol was followed as detailed earlier. The ENR
temporal changes were determined at preselected times; then, the
Table 1 apparent first-order rate constants of each conditions were compara­
Codified and uncodified experimental values according to Box-Behnken design tively evaluated.
for enrofloxacin (ENR) removal.
Codified value Actual value ENR removal at 90 min. 2.6. Influential effect experiments
(%)
Set
X1 X2 X3 H2O2 pH COD Actual Predicted To evaluate the influence of constituent anions on ENR degradation,
(mM) (mg/L) value value we individually varied anion concentrations (Cl− , NO−3 , CO2- 2-
3 , and SO4 )
A − 1 − 1 0 1.25 3 88.5 30.43 36.02 to the final concentration of 0.024 or 5 mM, which was based on typical
B 1 − 1 0 10 3 88.5 89.20 88.12 concentrations frequently detected in receiving water. Anions and H2O2
C − 1 1 0 1.25 11 88.5 46.50 47.58 were added simultaneously to the designated values in the UV system
D 1 1 0 10 11 88.5 70.25 64.68
E − 1 0 − 1 1.25 7 69.0 71.05 69.29
during the stirring processes.
F 1 0 − 1 10 7 69.0 93.91 98.84 Because real wastewater usually exists in the presence of a high
G − 1 0 1 1.25 7 108.0 63.04 58.12 loading of organic matter, the influential effect experiments were car­
H 1 0 1 10 7 108.0 96.00 97.77 ried out in different types of water matrices (DI: deionized water; TW:
I 0 − 1 − 1 5.625 3 69.0 67.72 63.93
tap water, SWW: synthetic wastewater; DW: discharge water, or RW:
J 0 1 − 1 5.625 11 69.0 55.91 56.59
K 0 − 1 1 5.625 3 108.0 57.09 56.41 river water) or with the presence of humic acid concentrations (HA; 20
L 0 1 1 5.625 11 108.0 48.09 51.88 mg L− 1). The SWW was prepared as described earlier and was equivalent
M 0 0 0 5.625 7 69.0 88.57 88.89 to 52 mg L− 1 COD (Table 1). The DW was collected from fishery farms
N 0 0 0 5.625 7 69.0 96.70 88.89 housing domestic freshwater fish. The RW was collected from natural
O 0 0 0 5.625 7 69.0 94.23 88.89
flowing water which was also used for raising fish in floating cages on

3
K. Jutarvutikul et al. Aquacultural Engineering 94 (2021) 102174

the Chao Phraya River in Chai Nat province, Thailand (15.184147 N, (equivalent to 325 mL), the second pump was operated at similar
100.119988E). Water quality characteristics were based on the records influent and effluent flow rates to maintain a constant volume in the
of the Pollution Control Department during Aug 2019–Aug 2020 at a quartz cell.
nearby sampling station (Table S1). The degradation experiments were Three individual experiments were conducted with different
conducted using the optimized conditions for ENR obtained from the amounts of added H2O2 dose to sum to a total H2O2 concentration of 10
RSM analysis. mM, 40 mM, 60 mM, or 120 mM. The temporal changes of ENR were
monitored from two locations: in the UV reactor and in the effluent
2.7. Sequential addition experiment reservoir.

To determine whether the sequential addition of an H2O2 dose 3. Results and discussion
altered the kinetic degradation of the antibiotic, an aqueous solution of
100 mL of 0.1 mM ENR was treated using the UV/H2O2 system. The 3.1. Antibiotic degradation experiment
experiments were conducted with different amounts of final H2O2 con­
centration: 5, 7.5, 10, 40, or 120 mM. Dosages of H2O2 were divided into The relative COD concentrations of the tested ENR with various types
either one-half, one-quarter, or one-sixth proportions and added into the of UV catalytic processes were compared at the same H2O2 concentra­
ENR solution while exposed to UV irradiation for the entire time. tion and are presented in Fig. 2. The degradation rate changed with a
Samples were collected every 7.5 min for the total 90 min and analyzed change from the heterogeneous to the homogenous UV system. The
for ENR residue. homogeneous process had better treatment efficiency than the hetero­
geneous process. Furthermore, at the same pH, the COD removal was
faster in S2O2−
8 followed by H2O2 for the homogeneous process and even
2.8. Contact tank experiment
faster in ZnO followed by TiO2 for the heterogeneous process. This was
due to the radical oxidation that was only prominent near the catalytic
The tank experiments were conducted in the UV reactor as shown
surface compared to the evenly-distributed radicals in the solution in the
earlier (Fig. 1). Two 1 L reservoirs (influent and effluent) and non-
homogeneous process. While the advantage of using a catalyst is its ease
adsorbing tubing (Masterflex Viton® tubing; Coleparmer, IL, USA)
of separation from water, a very small particle size (<10-9 m) and the
were placed outside the reaction tank. The first peristaltic pump (Mas­
generation of reactive oxygen species (ROS) under light irradiation can
terflex, Cole-Parmer, IL, USA) was designed to introduce ENR-
impact the aquatic environments (Wiesner et al., 2006). Therefore, un­
contaminated water from the first reservoir. The pH of the 0.1 mM
less a regeneration system is provided to separate the used catalysts from
ENR solution was adjusted to 6.8 ± 0.1 which was obtained from the
the treated discharge water, using a synthetic catalyst may not be
optimization experiments; then, the solution was introduced into the
entirely suitable for continuously flowing discharge water from aqua­
tank using a flow rate of 5 mL min− 1. We also conducted a similar
culture farms. Although S2O2−8 was proven to be better at removing ENR,
experiment with SWW or the real discharge water (DW) from the
the increases in salt content (SO2−
4 ) following treatment may still be of
aquaculture farm to ensure applicability. The UV lamp was turned on
concern, especially for successive farms nearby. After delineating other
simultaneously with the quartz cell receiving the first drop of the solu­
oxidant drawbacks, UV/H2O2 can overcome most, if not all, oxidant
tion. The first sampling occurred at 20 min, equivalent to 100 mL of ENR
usage concerns because its absence of secondary contaminant genera­
solution prior to receiving the first dose of H2O2. Although the selected
tion and its better treatment performance for antibiotic-contaminated
H2O2 addition timings obtained from the sequential addition experi­
water (Cuerda-Correa et al., 2020).
ments were the ones that had proven better performance, we also con­
To evaluate the temporal changes of the antibiotic during degrada­
ducted several H2O2-adding conditions to compare their overall
tion using the UV/H2O2 system, experimental units were designed to
removing efficiency and to ensure that our H2O2 condition worked on
determine the most effective variation compared to the control (no
both DI water and any organic constituent containing water. At 65 min

Fig. 2. COD relevant concentration (COD/CODo) following ENR degradation in UV-assisted chemical oxidation system (H2O2, S2O2−
8 , TiO2, ZnO).

4
K. Jutarvutikul et al. Aquacultural Engineering 94 (2021) 102174

variation). We selected H2O2 concentration (1.25–10 mM), pH solution H2O2 concentration above 7.5 mM due to the reaction of excess H2O2
(3–11) as independent variables for the variation test. Because COD was with OH that generated peroxyl radicals (HO2) which could automati­
• •

likely controllable to maintain the treatment system functionality, we, cally react with any produced OH [Eq. 5–7] (Liu et al., 2015). During
•

selected COD concentration as the last independent variable. Therefore, this propagation process, dimerization can also occur, thus minimizing
the total COD concentration was slightly different between the groups of OH molecules [Eq. 8]:
•

antibiotics. In addition, the COD levels of each antibiotic were 108–69

(5)
• •

mg− COD L− 1, 116–77 mg− COD L− 1, and 88–49 mg− COD L− 1 for ENR, H2O2 + OH → HO2+ H2O
SMX, and OTC, respectively. The observed removal percentages of each • •
H2O2 + HO2 → OH + H2O + O2 (6)
antibiotic for different variables at 90 min are reorganized and shown in
Figure S1. Initially the H2O2 concentration was an influent factor, (7)
• •
OH + HO2 → H2O + O2
especially with OTC, while neutral pH and COD had only a slight effect
(8)
• •
OH + OH → H2O2
on OTC removal.
The results showed a positive correlation between H2O2 and the These chain reactions caused the inverse relation between the
antibiotic removal efficiency, as it tentatively increased with the OH degradation rates of the target compounds and the H2O2 concentration.
•

concentration. Hydrolysis and thermal degradation of these antibiotics Therefore, increasing the H2O2 concentration in the UV/H2O2 system
was observed in separate experiments under the absence of light irra­ would be unnecessary for ENR or OTC contaminated water, unless such
diation. By using UV alone, ENR and SMX were degraded by up to 40 % effluent contained high amounts of other oxidizable carbon compounds.
and 70 %, respectively, while OTC was only degraded by 20 %, indi­ The initial pH had a mixed influence on the antibiotic degradation as
cating that OTC was less sensitive to UV-photolysis (Fig. 3A–3C). Our pH played an important role in the remnant charges on the antibiotic
results also indicated that SMX molecules may have already been more molecules. Under varying pH treatments, The ENR and OTC degradation
sensitive to UV alone as it was degraded by 70 % at 90 min following rates were highest in the neutral solution, while that of SMX was better
ENR and OTC. This SMX degradation process easily occurred because of in acidic solutions (Fig. 3D–3F). SMX is mostly negatively charged for
the presence of aniline (a strong electron donating group) and the pH > 7 (pKa1SMX = 1.6, pKa2SMX = 5.7), resulting in better removal
cleavage of isoxazole ring, which is a main degradation pathway of SMX performance at a higher pH as the OH preferentially attacks molecules
•

(Niu et al., 2013). On the other hand, OTC was more persistent when in the deprotonated state (Fig. S2; Ngouyap Mouamfon et al., 2011). The
subjected to the UV irradiation than either SMX or ENR, indicating that ENR degradation rate would have shown similar result following
either a higher UV dose, a longer contact time, or UV-assisted oxidation UV/H2O2 reaction as both of its pKa values were relatively neutral
may be required to obtain better removal of this pharmaceutical. (pKa1ENR = 6.2, pKa2ENR = 7.75) (Sayen et al., 2018), but as it is a zwitter
However, there was no significant degradation with increases in the ion, at a neutral pH it is better at absorbing UV photons in this pH range

Fig. 3. Antibiotic removal (enrofloxacin, ENR; sulfamethoxazole, SMX; oxytetracycline, OTC) for varying conditions of (A–C) H2O2 concentration (0–10 mM), (D–F)
pH (3–11), and (G–I) COD concentration (69–116 mgL− 1) during UV/H2O2 oxidation. Bars indicate ± sample standard deviations (n = 3).

5
K. Jutarvutikul et al. Aquacultural Engineering 94 (2021) 102174

(Fig. S3; Shokrolahi et al., 2019). On the other hand, the OTC degra­ cause the inhibition effect to degrade organic compounds, with COD
dation results indicated only slight differences in the degradation rate competing for the UV irradiation and scavenging any available active
with varying pH compared to the neutral state (pKa1OTC = 3.6, pKa2OTC = radicals. This lends credence to a more efficient treatment in which most
7.5, pKa2OTC = 7.9), as shown in Fig. S4 (Liu et al., 2015). This was due of the COD should be removed prior to entering our proposed system
to the sensitive reaction between OTC and OH. Ferradini et al. (1978) (Fig. 4).
•

proposed that H2O2 decomposition may lead to the generation of both

OH and O − that may enhance the radical reaction on OTC.
• •

3.2. RSM optimization

It was noted that our H2O2 concentration range was quite low (0–10
mM) compared to many other studies (such as Qiu et al., 2019 and The Box-Behnken design consisted of 15 experimental sets (Experi­
Sciscenko et al., 2020). These researchers suggested that in most cases, a ments A–O) for the predicted results of antibiotic removal at 90 min. of
higher H2O2 concentration was not necessary because it would lead to oxidative reaction for three individual antibiotics (Table 1). Multiple
the loss of radicals [Eq. 5, 7]. analyses of variance revealed an empirical relationship between the
We believed that the formation of O − may have been so slight that it
•

response (antibiotic removal; Y) and the three variables (X1 – X3 ). Both

was not enough to react with the ENR molecules. In addition, the re­ individual, square, and 2-way interaction contributions of all variables
action rate of O − with OH was 4-fold less than that of the ENR- OH
• • •

10 − 1 − 1 10 − 1 − 1
were expressed in the following polynomial second order model for ENR
reaction (2.0 × 10 M s versus 7.9 × 10 M s (Buxton et al., [Eq. 9]:
1988; Santoke et al., 2009). Another possible reason supporting the OTC
degradation was that at high pH, the OTC molar extinction coefficient YENR = − 144.1 + 7.67X1 + 26.32X2 + 2.743X3 − 0.500X1 X2 + 0.0296X1 X3
was at 254 nm and direct UV photolysis might have occurred prior to the + 0.0090X2 X3 − 0.252X12 − 1.789X22 − 0.01768X32
oxidation of OH on the ENR molecules.
•

(9)
When the COD concentration was high, available UV photons were
easily absorbed on the organic molecules, reducing the antibiotic The SMX and OTC models are shown in the Supporting Information
removal efficiency. Fig. 3G–3I shows that the control condition provided [Fig. S5, S6]. These experimental expressions indicated mixed results.
the best removal efficiency. The presence of natural constituents can Increasing most variables resulted in increasing antibiotic removal. Only
the pH for SMX degradation and the COD concentration for OTC

Fig. 4. Response surface plot, described by calculated model fitted from experimental results of ENR removal during UV/H2O2 oxidation (A–C) and comparison of
predicted and experimental values of each experiment (D).

6
K. Jutarvutikul et al. Aquacultural Engineering 94 (2021) 102174

degradation represented antagonistic effects, implying that increasing for H2O2 concentration:pH:COD: =10 mM:6.15:87.51 mg L− 1 (Table 2;
these two variables had a negative impact on these two antibiotics. The ENR and Table S4; SMX and OTC). These optimized conditions were
effects of the H2O2 concentration were notable in the weak acidic so­ further selected to assess the feasibility of applying this system in
lution for OTC and SMX compared to a basic solution and was most different environmental conditions because ENR is extensively used in
suitable with neutral pH for ENR. At 10 mM H2O2 concentration, the veterinary applications (Wei et al., 2012) and its degradation pathway is
antibiotic degradation was best, as discussed earlier. The high removal toward the formation of ciprofloxacin, another well-known licensed
percentage at greater than 95 % (at 90 min) was observed in all ex­ human antibiotic.
periments with high H2O2 concentrations (10 mM) and neutral pH (7).
Similar levels of removal efficiency were observed in acidic solution and 3.3. Radical scavenger experiments at varying pH
in any composition of COD, such as Experiments B, F, and H (Table 1).
However, the alkaline solution had less removal efficiency, such as in Because larger coefficients indicate a greater response [Eq. 9], the
Experiment D; in this run, the high pH may have facilitated OH scav­ ENR degradation experimental expression indicated that pH had a large
•

enging with a higher H2O2 concentration. When the condition was set at effect on the response (Y). Other researchers had shown that the
the highest and lowest codified values (H2O2 1.25, 10 mg L− 1), the effect degradation of contaminant under an UV/H2O2 system was possibly
of the H2O2 concentration was notable in the acidic condition (Experi­ from: (1) direct photolysis, (2) OH oxidation, or (3) singlet oxygen
•

1
ments A and B) compared to the basic condition (Experiments C and D). oxidation ( O2) (Li et al., 2011), but limited information has been
So, by increasing the H2O2 concentration at the lower pH, a larger dif­ available for the active radical species at different pH levels. Therefore,
ference in removal efficiency was observed compared to at a higher pH we varied the pH in for further testing the principal radical species using
(59 % versus 24 %), as shown in Table 1. ENR optimized condition.
The more frequent use of ENR compared to the other tested antibi­ From here on, the ENR normalized concentration (C/Co) was sepa­
otics is further discussed using the statistical analyses in the Supporting rately plotted against reaction time to calculate the observed pseudo-
Information (Table S3). The significance of the model parameters was first-order kinetic degradation rate (kobs) which was plotted against in­
classified using the probability value (P-value) and the index for influ­ dependent variables for comparison (Fig. 5). We used three different
ence factor (F-value). At a P-value less than 0.05 at the 95 % confidence chemicals in this experiment. Tert-butanol (TBA) served as OH scav­
•

8 − 1 − 1
interval, these models were considered significant while at a P-value enger (k = 6.0 × 10 M s ), but it was inert to other types of active
greater than 0.1, these models were considered not significant. The re­ species (Li et al., 2011), while furfuryl alcohol (FA) served as a singlet
sults showed that H2O2 and pH were both significant for ENR treatment, oxygen (1O2) scavenger (k = 1.2 × 108 M− 1 s− 1) (Appiani et al., 2017).
while the interactions of other parameters were not significant for both Unlike these alcohols, p-benzoquinone (PBQ) was supposed to serve as a
the SMX and OTC treatments (P > 0.1; Table S3). scavenger for the super oxide radical (O2− ). However, we found that
•

Based on the Pareto analysis of the standardized effects to classify the spiking PBQ into the ENR solution resulted in darkening of the ENR
level of significance of each variable, it was confirmed that the H2O2 solution which interfered with the ENR analysis. Although O2− can be
•

concentration and the relations between variables were the most sig­ generated from the reaction of OH and excess H2O2 (k = 2.7 × 107 M− 1
•

nificant factors for ENR and SMX, while pH was the most significant s− 1), this phenomenon was unlikely in our study because of the prefer­
factor for OTC (Fig. S7-S9). However, the relationships between the able reaction of ENR and OH (k = 7.9 × 1010 M− 1 s− 1) (López-Peñalver
•

variables were significant for ENR and not significant for SMX. The re­ et al., 2010; Santoke et al., 2009). Hence, O2− measurement is not
•

sidual COD concentration was significant for ENR but not significant for needed in this study.
SMX and OTC. Although the COD concentration was not very high, it Further results also confirmed that the ENR degradation rates were
could represent the COD concentration in natural flowing water which impeded in the presence of any alcohols, indicating that OH is the main
•

has been reported as low as 5 mg L− 1 and can be regulated to be up to 60 reactive species in the solution for all ranges of pH (Fig. 5). Under the
mg L− 1 in China (Frimmel and Abbt-Braun, 2011; Zhou et al., 2018). neutral pH condition, it was clear that ENR degradation was from both
The ENR ANOVA analysis is shown in Table S3 using a confidence OH and 1O2, resulting in the highest ENR degradation rate. Other
•

interval (CI) of 95 %. The suitability of the polynomial equations was research showed that pH is a major parameter responsible for the exis­
evaluated using the determination coefficient (R2) which was 0.92, tence of the predominant radical species (Guan et al., 2011). The overall
indicating that the ENR removal was well-defined and a good fit of the results suggested that antibiotic degradation using this system can be
model. This high R2 value provided proof that the response surface retarded under extreme pH conditions and it must be neutralized to be
model for predicted values of ENR removal efficiency was coherent. more suitable in this oxidative reaction.
According to the removal efficiencies, the ENR degradation relied on
a high H2O2 dosage and a neutral pH condition. By aligning maximum 3.4. Effects of anions
antibiotic degradation, the ENR optimized experimental condition was
The optimal condition for ENR treatment obtained from the 3D
Table 2 surface response was used to further investigate the influence of ions as
Summary of optimal conditions for enrofloxacin (ENR) another important environmental constituent. The selected anions were
degradation by UV/H2O2 oxidation system obtained Cl− , NO−3 , CO2− 2−
3 , and SO4 as they are ubiquitous in natural receiving
from response surface methodology and summary of water at concentrations as low as 1 mg L− 1 for NO− or up to 1000 mg L− 1
coefficient of determination (R2) of model. for Cl− , and SO2− 4 in some areas (Mallick, 2017; Varden and
Parameter Value Bou-Abdallah, 2017). Among other anions, positive effects were found
H2O2 (mM) 10
from the presence of Cl− which was the most sensitive anion that can
easily convert to Cl and simultaneous reaction with HOCl − , even in
• •
pH 6.15
COD (mg L− 1) 87.51 very low amounts, resulting in an increase in the ENR degradation rate
Fit 106.23 (Jayson et al., 1973).
SE Fit 2.35
In this optimized condition, the pH was set at 6.15 so that both HCO−3
95 % CI 101.45,111.01
and H2CO3 were predominant, with OH being slightly stronger than in
•

95 % PI 92.30,120.16
S 6.4459 alkaline conditions (Fig. 5) which was consistent with other research
R2 0.9252 that reported the highest OH concentration was under acidic conditions
•

R2 (adj) 0.9059 (pH2) (Liao et al., 2001). The reaction between HCO−3 and OH was not
•

R2 (pred) 0.8710
very important as its degradation rate was lower than that of CO2− 3 by

7
K. Jutarvutikul et al. Aquacultural Engineering 94 (2021) 102174

Fig. 5. Observed kinetic rate constant (kobs) for various pH levels of ENR degradation under different radical scavengers (tert-butanol, TBA; furfuryl alcohol, FA)
using calculated optimal condition of UV/H2O2 oxidation system. Inset graph shows temporal changes of ENR of corresponding kobs. Bars indicate ± sample standard
deviations (n = 3).

50-folds (kCO32− = 8.5 × 106 M− 1 s− 1). On the other hand, the presence concentration made up for the SWW, similar to the constituent ions,
of CO2− 3 decreased the degrading performance by up to 42 %. This dissolved organic matter (DOM), and suspended solids in the RW and
decrease resulted from the working pH rather than the reaction between DW. Aside from the UV irradiation impediment, secondary radicals may
OH and CO2−3 (kCO32- = 3.9 × 10 M
8 − 1 − 1
s ) which was relatively low be generated. Most of the natural water constituents were incapable of
•

compared to the other constituent ions (Acero and Gunten, 2000). Un­ absorbing the short wavelength range (200–290 nm) which would
like with SO2− 2−
4 alone, UV irradiation on SO4 is known to produce the compete with target contaminants, especially with OH with humic acid,
•

sulfate radical which may cause radical scavenging with available OH the major component in DOMs. As presented in Fig. 7, HA at 20 mg L− 1
•

2−
(Wang and Wang, 2018). In addition, SO4 is also very reactive to the could subdue the ENR removal efficiency. This can be explained from
hydroxyl radical (kSO42− = 10.0 × 109 M− 1 s− 1) (Buxton et al., 1988). the humic substance property itself that showed similar turbidity,
Notably, SO4− is a very strong OH quencher that can also be formed indicating that the humic acid can effectively absorb UV radiation rather
• •

from the reaction of OH and SO2− 4 . Yet this latter phenomenon with than initiate the active radical species. Furthermore, humic acid can also
•

SO4− is unlikely and requires a much longer reaction time because of its act as a radical scavenger as it mostly contains electron-rich moiety
•

relatively low reaction rate (kSO4− = 0.0015 × 109 M− 1 s− 1) (Moussavi compounds on the surface. However, it is noteworthy that at a low
•

et al., 2018). concentration of HA (<5 mg L− 1), we observed a slight increase in kobs

The NO−3 concentration has been reported to be as low as 20 mg L− 1, compared to the control. Sharma et al. (2015) elaborated that such a
resulting in a positive impact on SO4− oxidation (Angkaew et al., 2019). case was possible from the specific characteristics of humic substance
•

On the contrary, our results showed a decrease in the ENR degradation surface which contains various types of organic constituents that can
efficiency at an NO−3 concentration of 0.024–0.5 mM (Fig. 6). The facilitate other reactive species against available oxidative radicals.
presence of excess NO−3 concentration following UV irradiation can form Because varying HA may produce different results, this information
both reactive nitrogen species (NO3) as well as another strong reactive suggested that our treatment would probably more effective if the fil­
•

radical, the oxygen radical (O ) (Li et al., 2011). Although the NO−3 trating system or some kind of maturation ponds were applied to allow
•−

concentration is not naturally high, the ENR oxidation from UV/H2O2 separation of humic colloids from the water.
was mostly from OH. These results indicated that these subsequent As discussed earlier, the formation of Cl and Cl2− may occur from the
• • •

radicals suppressed OH activity and later reduced removal efficiency, reaction between Cl and available Cl− . A similar reason can also be
• •

thus minimizing the degradation rate (kobs). Unless these constituent applied to the tap water (TW) which were reported by Metropolitan
ions were in extreme conditions, which rare for our studied working Waterwork Authority of Thailand (MWA, 2017) to contain free chlorine
conditions, a high concentration or strong UV source may not be residue in the range 0.2–2 % and some other physico-chemical param­
necessary. eters under regulated maximum concentration levels (MCL). For
example, the end-pipe tap water quality reported from the Bureau of
3.5. Effects of different water matrices Food and Water Sanitation contained 42 mg L− 1 sulfate, 36 mg L− 1
chlorine, and less than 1.8 MPN/100 mL of coliform (BFWS, 2018).
Treating the ENR optimized conditions with different water matrices These may also influence the formation of secondary radicals to simul­
taneously quench the OH as well as possessing the ability to absorb UV
•

decreased the ENR degradation rate by 6.8–29.4 % compared to

deionized water (Fig. 7). All tested water samples had a neutral pH as discussed earlier.
condition, indicating that pH did not have large influence on ENR The reason for the decrease in ENR degradation in SWW, DW and RW
degradation compared to other water constituents. The results showed matrices was the presence of organic constituents such as DOM, organic
that the ENR degradation rates were the highest in DI followed by TW, colloids, and suspended solids (SS). The SWW containing oxidizable
SWW, HA, DW, and RW. This was due to several types of salt compounds such as meat extract, peptone, and urea, was equivalent to

8
K. Jutarvutikul et al. Aquacultural Engineering 94 (2021) 102174

Fig. 6. (A) Observed kinetic rate constant (kobs) of ENR degradation with different constituent ions (Cl− , CO2- 2-
3 , SO4 , and NO3 ) and (B) temporal changes of ENR of
−

corresponding kobs. Bars indicate ± sample standard deviations (n = 3).

60 mg L− 1, which was lower than in general RW, resulting in retardation 3.6. Sequential addition experiment
of the ENR degradation rates. The water quality analyses of DW and RW
showed that RW was low in BOD concentration, but both were quite The effect of H2O2 addition was systematically investigated to
high in the coliform amount (Table S2). The latter has also been proven further improve the oxidation process to reflect a real-world application.
to have a negative impact on the oxidative reaction (Vargas et al., 2012). The results from our batch experiments where final H2O2 concentrations
The authors provided proof that E. coli can scavenge for active radical of 5 and 7.5 mM were initially added into the ENR solution showed that
species such as OH and O2− by damaging both their DNA and a part of ENR degradation reached a plateau at 60–70 min, resulting in 85 % of
• •

the cell membrane. These results revealed that if this treatment were to ENR residues, and only 30 min for the final H2O2 concentration of 120
be used with target contaminant in real-world applications, more oxi­ mM (Fig. 8A–B, 8E). When the H2O2 addition was split into 2, 3, and 6
dants or a longer UV contact time would have been necessitated and that times while maintaining the final H2O2 concentration at 5 mM, we found
sequential addition of H2O2 may need to be involved. The difference in that the ENR removal efficiency decreased by 20–60 %, especially at the
the optimum conditions for the three antibiotics as discussed earlier more frequent additions in the beginning phase (<20 min), as shown in
indicated that various independent variables were needed for different Fig. 8A. A similar result was evident at the final concentrations of H2O2
types of target contaminants. The fact that the water constituents were at 7.5 (Fig. 8B). The control (one addition) resulted in lower efficiency
completely different from place to place, a little higher H2O2 concen­ when the total H2O2 concentration was 10 mM. These results implied
tration from the optimal conditions may be suggested in actual water that the H2O2 concentration of 1.6 mM may yield the same efficiency as
treatment unless the ~10 % differences in system efficiency was the total H2O2 concentration of 5 mM and that a higher concentration
acceptable. may have retarded ENR degradation.
As a practical use of this system in a real-world application, we

9
K. Jutarvutikul et al. Aquacultural Engineering 94 (2021) 102174

Fig. 7. Observed kinetic rate constant (kobs) of ENR degradation with different water matrices (DI = deionized water; TW = tap water; SWW = synthetic wastewater;
HA = water containing 20 mg L− 1 humic acid; RW = river water; and DW = aquaculture discharge water). Inset graph shows temporal changes of ENR of corre­
sponding kobs. Bars indicate ± sample standard deviations (n = 3).

anticipated that a higher amount of the total H2O2 could be added due to 3.7. Contact tank experiment
the presence of other organic and ion constituents. Therefore, the 40 mM
and 120 mM experiments were tested (Fig. 8D–8E). The results showed Treated wastewater is not entirely clean as it can be contaminated
that adding H2O2 a few times was better than adding only one time and 6 with emerging contaminants or some other persistent organic pollutants
times for the 40 mM experiment. When the 120 mM was tested, the from conventional treatments. By using the optimization results for the
control was much better in overall degradation efficiency and the ENR ENR treatment from our RSM experiments, we tested our contact system
removal reached a plateau as earlier as 30 min. This can be explained with the ENR solution using also the sequential addition from earlier
from the discrepancy of the molar ratios of added H2O2 at each time and experiments. This flow-through experiment was designed to serve as a
the ENR concentration. When this ratio was maintained at 0.5–1.0, the real-world application and we imitated the sequential addition of H2O2
ENR removal was better than the other addition conditions. However, as it achieved the best performance in earlier experiments. The addition
when the ratio was very high (120 mM H2O2:10 mM ENR) the available of H2O2 was at 20 min and every 45 min afterwards until the total
ENR could be instantly treated prior to starting the radical self- concentration of H2O2 reached the designated concentration (10, 40, or
scavenging, resulting in the fastest removal at the control condition 120 mM). Temporal changes in the ENR concentrations showed that
(Fig. 8E). Therefore, we selected a high total dose (>10 mM) for the ENR was reduced in the solution to as low as 5% within 240 min for all
successive contact tank experiments. types of H2O2 addition and water matrices (Fig. 9). Based on the ENR
Although the overall removal of all sequential additions was in the removal percentage, the contact tank experienced 15 % less removal
range 80–99 %, adding more times did not always produce the highest than in the batch experiment (Fig. 3A versus Fig. 9A). This may have
removal efficiency, indicating a loss of OH may have occurred. In been due to the sizing discrepancy and UV alone was possibly not suf­
•

addition, the removal efficiency trends were less steep compared to ficient for a larger-scale application.
adding H2O2 fewer times (Fig. 8C). Possible reasons included the reac­ A comparison between 10, 40, and 120 mM total H2O2 concentration
tion of generated OH and the freshly added H2O2 [Eq. 5]. Once it was conducted to determine whether the oxidant dose had an effect on
•

happened, localized radicals such as OH, HO2− , and O2− were available the ENR degradation. Although the high H2O2 concentration was proven
• • •

for H2O2 to undergo radical recombination, resulting in more H2O2 to be unnecessary, other researchers found that up to 29.4 mM of H2O2
molecules in the solution. The kinetic rate of OH production using the concentration, the removal rate started to reduce from the radical
•

UV/H2O2 process, was slower than the recombination reaction. How­ scavenging process (Borowska et al., 2015). It was also clear that
ever, this pace could enhance the self-scavenging reaction of available ENR-contaminated water was continuously introduced into the system,
radicals, causing the disappearance of radicals to the organic molecules. resulting in a slower ENR degradation rate. Although the total H2O2
These reactions explained the reduced removal efficiency with the more concentraftion was different by 3-folds, the results showed almost the
frequent addition of H2O2. This was also confirmed by our previous same trend in ENR reduction (Fig. 9). In the beginning phase of the
experiments where the recombination effects were minimized as the larger H2O2 dose experiment (120 mM), ENR removal was slightly better
H2O2 concentration was not as high (<10 mM), as shown in Fig. 8A–8B. until further addition of H2O2 causing ENR removal to remain similar to
Although a higher content of H2O2 can result in radical scavenging and that from the smaller total H2O2 dose experiment (Fig. 9). Here, for­
reproduce H2O2 via dimerization of OH (Qiu et al., 2019), a large dif­ mation of OH radicals was responsible for the reduction of ENR as well
• •

ference in ENR and H2O2 could also result as an adverse effect, as dis­ as the reduction of radicals due to self-scavenging, indicating that this
cussed earlier. higher dose of H2O2 may not be necessary unless sufficient scavenger

10
K. Jutarvutikul et al. Aquacultural Engineering 94 (2021) 102174

the UV reactor at the same time that the H2O2 was sequentially added.
Interestingly, the ENR concentration increasingly vacillated once the
ENR-contaminated DW was introduced and continually decreased. As
we used the discharge water received from the farms, the total sus­
pended solids (SS) was quite high (Table S2). These SS may serve as
adsorption sites for ENR, which could, in turn, cause a sudden increase
in the ENR concentration. The ENR concentration eventually diminished
to almost 10 %. This result also confirmed that the initial removal of SS
using maturation ponds would be appropriate if using the UV-assisted
H2O2 system.
Overall, we proved that sequential addition was more effective than
a single addition, providing that the total H2O2 concentration was
ascertained to treat the available organic constituents. An undesirable
removal rate due to insufficient stirring that resulted in an unevenly
distributed OH radicals was an unexpected reverse effect from adding
•

more than one H2O2 dose into the system (Zhou et al., 2016). However,
we observed that ENR continuously decreased both in the contact tank
and in the effluent reservoir. Given that this reverse phenomenon did
not occur in our system, a sizing difference in a larger-scale application
may reveal such results. The laboratory-scale results may need to be
further optimized for the hydrological and environmental conditions to
sufficiently treat contaminant residuals from discharging wastewater. In
addition, we evaluated our bench-scale, photo-assisted H2O2 operating
cost to be approximately USD 2.91 to treat 1 m3. While this calculation
was based on the price of a kilowatt hour in Thailand and that the real
operation cost may be a little higher, our cost was approximately 2-folds
lower than other advanced oxidation processes (Yonar et al., 2015).

4. Conclusion

The present study showed that UV-assisted chemical oxidation with

H2O2 can be used to treat antibiotic-contaminated discharge water from
an aquaculture farm. Compared with other UV assisted systems, our
results demonstrated that the use of UV in a homogeneous oxidative
system was more effective than in a heterogeneous system. The tested
veterinary antibiotics were enrofloxacin (ENR), sulfamethoxazole
(SMX), and oxytetracycline (OTC). Response surface methodology was
used to determine the optimal conditions of the UV/H2O2 system using
H2O2, pH, and remnant COD as independent variables. The antibiotic
degradation revealed that a high H2O2 dose at low pH was preferable
except for ENR degradation that was more pronounced with a neutral
condition, and that the presence of COD in the treated effluent should be
minimized to prevent competition with available radicals and UV irra­
diation. However, a high level of H2O2 is not recommended as it may
result in self-scavenging of radicals. The effects on ENR degradation
were investigated of parameters such as pH, active radical scavengers,
constituent ions, and different water matrices. The results showed slight
positive effects with the presence of Cl− and showed negative effects for
the other tested constituent anions. These phenomena were due to the
generation of secondary active radicals that could also enhance anti­
biotic degradation and suppress OH activity. Although H2O2 sequential
•

Fig. 8. Effect of sequential addition of H2O2 amount on ENR removal per­

addition was beneficial in oxidizing contaminants, a higher dose and the
centage for different time intervals for up to total of (A) 5 mM, (B) 7.5 mM, (C)
frequent addition of H2O2 may accelerate the reaction between freshly
10 mM, (D) 40 mM, and (E) 120 mM under optimized experimental condition.
added H2O2 and available radicals before the H2O2 is distributed. Since
the water constituents may vary, at least 10 % more H2O2, or perhaps
was present.
even more if applied at a much larger scale, and a longer UV contact time
To prove this, we further tested the system efficiency with the same
may be necessary. The ratio of the singly added H2O2 to the ENR con­
total H2O2 dose at 40 mM with SWW and DW, where ions and organic
centration was recommended to be maintained in the range 0.5–1.0 to
matter were available. The results showed that the total removal effi­
provide a faster ENR removal rate. These treatment schemes implied
ciency of SWW decreased by 40 % in the UV reactor (Fig. 9), but the
that the UV/H2O2 process is a potential method for treating aquacultural
overall removal efficiency almost reached 100 % removal in the effluent
discharge water contaminated with antibiotic residuals and its appli­
reservoir with time. After testing with the DW, the removal efficiency
cation can certainly reduce the risks of antibiotic-contaminated water in
was better than that for SWW by 15 % at 240 min. This may have been
the receiving discharge water and the water entering downstream
due to the difference in constituent ions in the SWW matrix that could
aquaculture farms. However, as the water would probably be site and
have less effect on the ENR-contaminated DW. It was noted that the ENR
contaminant specific, future investigations on actual water matrices
concentration slightly increased when new water was introduced into
may be necessary.

11
K. Jutarvutikul et al. Aquacultural Engineering 94 (2021) 102174

Fig. 9. Temporal changes in ENR concentration (C/Co) observed in contact tank following treatment at different total H2O2 concentrations and with different water
matrices. Arrows indicate H2O2 addition at designated concentration and x-axes represent both flow-through volume (mL) and time (min).

CRediT authorship contribution statement References

K. Jutarvutikul: Formal analysis, Methodology, Data curation, Acero, J.L., Gunten, Uv., 2000. Influence of carbonate on the Ozone/Hydrogen peroxide
based advanced oxidation process for drinking water treatment. Ozone Sci. Eng. 22
Investigation. C. Sakulthaew: Conceptualization, Formal analysis, Re­ (3), 305–328. https://doi.org/10.1080/01919510008547213.
sources, Writing - original draft, Writing - review & editing. C. Choke­ Aljuboury, D.D., Palaniandy, P., Aziz, H., Feroz, S., 2017. Degradation of Total organic
jaroenrat: Conceptualization, Formal analysis, Data curation, carbon (TOC) and Chemical oxygen demand (COD) in Petroleum wastewater by
solar Photo-Fenton process. Global Nest J. 19 (3), 430–438. https://doi.org/
Methodology, Writing - original draft, Writing - review & editing, Su­ 10.30955/gnj.002238.
pervision, Validation, Visualization, Project administration, Funding Angkaew, A., Sakulthaew, C., Satapanajaru, T., Poapolathep, A., Chokejaroenrat, C.,
acquisition. A. Pattanateeradetch: Writing - review & editing. S. 2019. UV-activated persulfate oxidation of 17β-estradiol: implications for discharge
water remediation. J. Environ. Chem. Eng. 7 (2), 102858. https://doi.org/10.1016/
Imman: Writing - review & editing. N. Suriyachai: Writing - review & j.jece.2018.102858.
editing. T. Satapanajaru: Writing - review & editing, Methodology, APHA, 1998. Standard Methods for the Examination of Water and Waste Water, 23 ed.
Funding acquisition. T. Kreetachat: Writing - review & editing. American Public Health Association.
Appiani, E., Ossola, R., Latch, D.E., Erickson, P.R., McNeill, K., 2017. Aqueous singlet
oxygen reaction kinetics of furfuryl alcohol: effect of temperature, pH, and salt
Declaration of Competing Interest content. Environ. Sci. Process. Impacts 19 (4), 507–516. https://doi.org/10.1039/
C6EM00646A.
Bergeron, S., Raj, B., Nathaniel, R., Corbin, A., LaFleur, G., 2017. Presence of antibiotic
The authors declare that they have no known competing financial resistance genes in raw source water of a drinking water treatment plant in a rural
interests or personal relationships that could have appeared to influence community of USA. Int. Biodeterior. Biodegradation 124, 3–9. https://doi.org/
10.1016/j.ibiod.2017.05.024.
the work reported in this paper.
BFWS, 2018. Monitoring Report on the Metropolitan Waterwork Authority Tap Water
Quality. http://foodsan.anamai.moph.go.th/download/D_Water/Surveillance/surv
Acknowledgments eillance_MWA.pdf.
Borowska, E., Felis, E., Miksch, K., 2015. Degradation of sulfamethoxazole using UV and
UV/H2O2 processes. J. Adv. Oxid. Technol. 18 https://doi.org/10.1515/jaots-2015-
This research was financially supported by a Graduate Program 0109.
Scholarship from The Graduate School, Kasetsart University and The Buxton, G.V., Greenstock, C.L., Helman, W., et al., 1988. Critical review of rate constants
National Research Council of Thailand (NRCT)-Ministry of Higher Ed­ for reactions of hydrated electrons, hydrogen atoms and hydroxyl radicals (⋅OH/⋅O-)
in aqueous solution. J. Phys. Chem. Ref. Data 17, 513. https://doi.org/10.1063/
ucation, Science, Research and Innovation (MHESI) (No. 191/2563), 1.555805.
Bangkok, Thailand. The Faculty of Environment and the Faculty of Chen, Y., Chen, H., Zhang, L., Jiang, Y., Gin, K.Y.-H., He, Y., 2018. Occurrence,
Veterinary Technology, Kasetsart University provided facility support, distribution, and risk assessment of antibiotics in a subtropical river-reservoir
system. Water 10 (2), 104. https://doi.org/10.3390/w10020104.
special research unit fund, and access to analytical instruments. Dr. Cuerda-Correa, E.M., Alexandre-Franco, M.F., Fernández-González, C., 2020. Advanced
Pawee Klongvessa provided statistical assistance and Mr. Athaphon Oxidation Processes for the Removal of Antibiotics from Water. An Overview. Water
Angkaew, Ms. Kanidrawee Techauaey, and Ms. Ratakul Luangaram 12, 102. https://doi.org/10.3390/w12010102.
Ferradini, C., Foos, J., Houee, C., Pucheault, J., 1978. Reaction between superoxide
provided technical assistance. The Pollution Control Department sup­ anion and hydrogen-peroxide. Photochem. Photobiol. 28, 697–700. https://doi.org/
plied the Chao Phraya River water characteristics and quality data re­ 10.1111/j.1751-1097.1978.tb07002.x.
ports at selected water sampling stations in Chai Nat province, Thailand. Frimmel, F., Abbt-Braun, G., 2011. Sum parameters: potential and limitations. Treatise
on Water Science 3, 3–29. https://doi.org/10.1016/B978-0-444-53199-5.00048-8.
Guan, Y.-H., Ma, J., Li, X.-C., Fang, J.-Y., et al., 2011. Influence of pH on the formation of
Appendix A. Supplementary data sulfate and hydroxyl radicals in the UV/Peroxymonosulfate system. Environ. Sci.
Technol. 45 (21), 9308–9314. https://doi.org/10.1021/es2017363.
Holmström, K., Gräslund, S., Wahlström, A., Poungshompoo, S., et al., 2003. Antibiotic
Supplementary material related to this article can be found, in the use in shrimp farming and implications for environmental impacts and human
online version, at doi:https://doi.org/10.1016/j.aquaeng.2021.102174.

12
K. Jutarvutikul et al. Aquacultural Engineering 94 (2021) 102174

health. Int. J. Food Sci. Technol. 38 (3), 255–266. https://doi.org/10.1046/j.1365- Aquaculture 412-413, 231–243. https://doi.org/10.1016/j.
2621.2003.00671.x. aquaculture.2013.07.028.
Jayson, G.G., Parsons, B.J., Swallow, A.J., 1973. Some simple, highly reactive, inorganic Santoke, H., Song, W., Cooper, W.J., et al., 2009. Free-radical-Induced oxidative and
chlorine derivatives in aqueous solution. Their formation using pulses of radiation reductive degradation of fluoroquinolone pharmaceuticals: kinetic studies and
and their role in the mechanism of the Fricke dosimeter. Journal of the Chemical degradation mechanism. J. Phys. Chem. A 113 (27), 7846–7851. https://doi.org/
Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases 69 (0), 10.1021/jp9029453.
1597–1607. https://doi.org/10.1039/F19736901597. Sayen, S., Ortenbach-López, M., Guillon, E., 2018. Sorptive removal of enrofloxacin
Khodadadi, M., Ehrampoush, M.H., Ghaneian, M.T., et al., 2018. Synthesis and antibiotic from aqueous solution using a ligno-cellulosic substrate from wheat bran.
characterizations of FeNi3@SiO2@TiO2 nanocomposite and its application in photo- J. Environ. Chem. Eng. 6 https://doi.org/10.1016/j.jece.2018.08.012.
catalytic degradation of tetracycline in simulated wastewater. J. Mol. Liq. 255, Sciscenko, I., García Ballesteros, S., Sabater, C., Castillo, M.-A., Escudero, C., Oller, I.,
224–232. https://doi.org/10.1016/j.molliq.2017.11.137. Arques, A., 2020. Monitoring photolysis and (solar photo)-Fenton of enrofloxacin by
Klamerth, N., Rizzo, L., Malato, S., Maldonado, M.I., et al., 2010. Degradation of fifteen a methodology involving EEM-PARAFAC and bioassays: role of pH and water matrix.
emerging contaminants at μgL− 1 initial concentrations by mild solar photo-Fenton Sci. Total Environ. 719, 137331 https://doi.org/10.1016/j.scitotenv.2020.137331.
in MWTP effluents. Water Res. 44 (2), 545–554. https://doi.org/10.1016/j. Sharma, J., Mishra, I.M., Kumar, V., 2015. Degradation and mineralization of Bisphenol
watres.2009.09.059. A (BPA) in aqueous solution using advanced oxidation processes: UV/H2O2 and UV/
Li, Y., Niu, J., Wang, W., 2011. Photolysis of Enrofloxacin in aqueous systems under S2O82− oxidation systems. J. Environ. Manage. 156, 266–275. https://doi.org/
simulated sunlight irradiation: kinetics, mechanism and toxicity of photolysis 10.1016/j.jenvman.2015.03.048.
products. Chemosphere 85 (5), 892–897. https://doi.org/10.1016/j. Shokrolahi, S., Farhadian, M., Davari, N., 2019. Degradation of Enrofloxacin antibiotic in
chemosphere.2011.07.008. contaminated water by ZnO/Fe2O3/Zeolite nanophotocatalyst. Journal of Applied
Liao, C.-H., Kang, S.-F., Wu, F.-A., 2001. Hydroxyl radical scavenging role of chloride and Research in Water and Wastewater 6 (2), 150–155. https://doi.org/10.22126/
bicarbonate ions in the H2O2/UV process. Chemosphere 44, 1193–1200. https:// arww.2019.1415.
doi.org/10.1016/S0045-6535(00)00278-2. Spellberg, B., Bartlett, J.G., Gilbert, D.N., 2013. The future of antibiotics and resistance.
Liu, N., Sijak, S., Zheng, M., Tang, L., Xu, G., et al., 2015. Aquatic photolysis of florfenicol N. Engl. J. Med. 368 (4), 299–302. https://doi.org/10.1056/NEJMp1215093.
and thiamphenicol under direct UV irradiation, UV/H2O2 and UV/Fe(II) processes. Sun, Y., Cho, D.-W., Graham, N.J.D., Hou, D., Yip, A.C.K., Khan, E., Song, H., et al., 2019.
Chem. Eng. J. 260, 826–834. https://doi.org/10.1016/j.cej.2014.09.055. Degradation of antibiotics by modified vacuum-UV based processes: mechanistic
López-Peñalver, J.J., Sánchez-Polo, M., Gómez-Pacheco, C., et al., 2010. consequences of H2O2 and K2S2O8 in the presence of halide ions. Sci. Total Environ.
Photodegradation of tetracyclines in aqueous solution by using UV and UV/H2O2 664, 312–321. https://doi.org/10.1016/j.scitotenv.2019.02.006.
oxidation processes. J. Chem. Technol. Biotechnol. 85 (10), 1325–1333. https://doi. Suzuki, S., Hoa, P.T.P., 2012. Distribution of quinolones, sulfonamides, tetracyclines in
org/10.1002/jctb.2435. aquatic environment and antibiotic resistance in Indochina. Front. Microbiol. 3 (67)
Mallick, J., 2017. Hydrogeochemical characteristics and assessment of water quality in https://doi.org/10.3389/fmicb.2012.00067.
the Al-Saad Lake, Abha Saudi Arabia. Appl. Water Sci. https://doi.org/10.1007/ Varden, L., Bou-Abdallah, F., 2017. Detection and separation of inorganic cations in
s13201-017-0553-1. natural, potable, and wastewater samples using capillary zone electrophoresis with
Moussavi, G., Pourakbar, M., Shekoohiyan, S., Satari, M., 2018. The photochemical indirect UV detection. Am. J. Analyt. Chem. 08, 81–94. https://doi.org/10.4236/
decomposition and detoxification of bisphenol A in the VUV/H2O2 process: ajac.2017.81007.
degradation, mineralization, and cytotoxicity assessment. Chem. Eng. J. 331, Vargas, G., Moreira, R., Spricigo, D., José, H., 2012. Treated domestic sewage: kinetics of
755–764. https://doi.org/10.1016/j.cej.2017.09.009. Escherichia coli and total coliform inactivation by oxidation with hydrogen
MWA, 2017. MWA Specification for Water Supply Quality (accessed Date 28 December peroxide. Química Nova 36, 252–256. https://doi.org/10.1590/S0100-
2020). https://web.mwa.co.th/ewtadmin/ewt/mwa_internet_eng/main.php?filen 40422013000200009.
ame=Water_Quality. Wang, J., Wang, S., 2018. Activation of persulfate (PS) and peroxymonosulfate (PMS)
Ngouyap Mouamfon, M.V., Li, W., Lu, S., Chen, N., Qiu, Z., Lin, K., 2011. and application for the degradation of emerging contaminants. Chem. Eng. J. 334,
Photodegradation of sulfamethoxazole applying UV- and VUV-Based processes. 1502–1517. https://doi.org/10.1016/j.cej.2017.11.059.
Water Air Soil Pollut. 218 (1), 265–274. https://doi.org/10.1007/s11270-010-0639- Watkinson, A.J., Murby, E.J., Costanzo, S.D., 2007. Removal of antibiotics in
y. conventional and advanced wastewater treatment: implications for environmental
Ngumba, E., Gachanja, A., Tuhkanen, T., 2020. Removal of selected antibiotics and discharge and wastewater recycling. Water Res. 41 (18), 4164–4176. https://doi.
antiretroviral drugs during post-treatment of municipal wastewater with UV, UV/ org/10.1016/j.watres.2007.04.005.
chlorine and UV/hydrogen peroxide. Water Environ. J. 34 https://doi.org/10.1111/ Wei, R., Ge, F., Chen, M., Wang, R., 2012. Occurrence of ciprofloxacin, Enrofloxacin, and
wej.12612. Florfenicol in animal wastewater and water resources. J. Environ. Qual. 41,
Niu, J., Zhang, L., Li, Y., Zhao, J., et al., 2013. Effects of environmental factors on 1481–1486. https://doi.org/10.2134/jeq2012.0014.
sulfamethoxazole photodegradation under simulated sunlight irradiation: kinetics Wei, Y., Jin, Y., Zhang, W., 2020. Treatment of high-concentration wastewater from an
and mechanism. J. Environ. Sci. 25 (6), 1098–1106. https://doi.org/10.1016/ oil and gas field via a paired sequencing batch and ceramic membrane reactor. Int. J.
S1001-0742(12)60167-3. Environ. Res. Public Health 17 (6), 1953. https://doi.org/10.3390/ijerph17061953.
Parpounas, A., Litskas, V., Hapeshi, E., et al., 2017. Assessing the presence of Wiesner, M., Lowry, G., Alvarez, P., Dionysiou, D., Biswas, P., 2006. Assessing the risks of
enrofloxacin and ciprofloxacin in piggery wastewater and their adsorption behaviour manufactured nanomaterials. Environ. Sci. Technol. 40, 4336–4345. https://doi.
onto solid materials, with a newly developed chromatographic method. Environ. Sci. org/10.1021/es062726m.
Pollut. Res. - Int. 24 (29), 23371–23381. https://doi.org/10.1007/s11356-017- Yonar, M., Yonar, S., Pala, A., Silici, S., Saglam, N., 2015. Trichlorfon-induced
9849-9. haematological and biochemical changes in Cyprinus carpio: ameliorative effect of
Pedersen, L.-F., Good, C.M., Pedersen, P.B., 2012. Low-dose hydrogen peroxide propolis. Dis. Aquat. Org. 114 https://doi.org/10.3354/dao02866.
application in closed recirculating aquaculture systems. N. Am. J. Aquac. 74 (1), Zhong, Y., Chen, Z.F., Dai, X., Liu, S.S., Zheng, G., Zhu, X., Liu, S., Yin, Y., Liu, G., Cai, Z.,
100–106. https://doi.org/10.1080/15222055.2011.651562. 2018. Investigation of the interaction between the fate of antibiotics in aquafarms
Qiu, W., Zheng, M., Sun, J., Tian, Y., Fang, M., Zheng, Y., Zhang, T., Zheng, C., 2019. and their level in the environment. J. Environ. Manage. 207, 219–229. https://doi.
Photolysis of enrofloxacin, pefloxacin and sulfaquinoxaline in aqueous solution by org/10.1016/j.jenvman.2017.11.030.
UV/H2O2, UV/Fe(II), and UV/H2O2/Fe(II) and the toxicity of the final reaction Zhou, W., Zhao, H., Gao, J., Meng, X., Wu, S., Qin, Y., 2016. Influence of a reagents
solutions on zebrafish embryos. Sci. Total Environ. 651, 1457–1468. https://doi. addition strategy on the Fenton oxidation of rhodamine B: control of the competitive
org/10.1016/j.scitotenv.2018.09.315. reaction of ⋅OH. RSC Adv. 6 (110), 108791–108800. https://doi.org/10.1039/
Rico, A., Phu, T.M., Satapornvanit, K., Min, J., Shahabuddin, A.M., Henriksson, P.J.G., C6RA20242J.
Murray, F.J., Little, D.C., et al., 2013. Use of veterinary medicines, feed additives and Zhou, Y., Duan, N., Wu, X., Fang, H., 2018. COD discharge limits for urban wastewater
probiotics in four major internationally traded aquaculture species farmed in Asia. treatment plants in China based on statistical methods. Water 10, 777. https://doi.
org/10.3390/w10060777.

13