Chemosphere 212 (2018) 56e66

Contents lists available at ScienceDirect

Chemosphere
journal homepage: www.elsevier.com/locate/chemosphere

Prednisolone degradation by UV/chlorine process: Influence factors,

transformation products and mechanism
Kai Yin a, 1, Qunying He a, 1, Chengbin Liu a, *, Yongxiu Deng a, Yuanfeng Wei a, Shuo Chen a,
Tongcai Liu a, Shenglian Luo a, b, **
a
State Key Laboratory of Chem/Biosensing and Chemometrics, Hunan University, Changsha 410082, PR China
b
Department of Environmental and Chemical Engineering, Nanchang Hangkong University, Nanchang 330063, China

h i g h l i g h t s g r a p h i c a l a b s t r a c t

UV and chlorine decomposed pred-

nisolone synergistically in UV/chlo-
rine system.

Anions (Cl, Br, HCO 2
3 and SO4 ),
NHþ4 and FA inhibited the predniso-
lone degradation.

Relative contribution of
OH and RCS
for prednisolone degradation was
calculated.

a, b-unsaturated ketone moiety for
prednisolone was preferential reac-
tion moiety confirmed by theoretical
calculations and experiments.

Some intermediates of prednisolone
by UV/chlorine showed higher
toxicity than itself based on the
toxicological tests.

a r t i c l e i n f o a b s t r a c t

Article history: Prednisolone (PDNN) as an emergent micropollutant directly influences the regional ecological security.
Received 8 March 2018 In this study, the degradation of PDNN by ultraviolet activated chlorine (UV/chlorine) oxidation process
Received in revised form was comprehensively evaluated. The quenching experiment suggested that the PDNN degradation in UV/
6 August 2018
chlorine process was involved in the participation of hydroxyl radical (
OH) and reactive chlorine species
Accepted 8 August 2018
Available online 13 August 2018
(RCS). Influence factors including chlorine dosage, pH, common anion and cation, fulvic acid (FA) on
PDNN degradation via UV/chlorine process were investigated. A low chlorine (7.1 mg L1) promoted the
Handling Editor: W Mitch PDNN degradation, while a high chlorine dosage (>7.1 mg L1) was adverse. The pH (4.0e10.0) showed
negligible effect, while the investigated anions (Cl, Br, HCO 2 þ
3 and SO4 ), NH4 and FA exerted negative
Keywords: impact on PDNN degradation. An efficient process to minimize pharmaceutical micropollutants was the
Prednisolone disposal of human urine containing a high concentration of pharmaceutical and potential toxic metab-
UV/Chlorine olites. An inhibitory effect was observed in the synthetic urine (fresh urine and hydrolyzed urine). The
Degradation kinetics intermediates/products were identified and the mechanism of PDNN degradation was proposed. PDNN
Degradation mechanism
gone through three degradation routes, involving the direct addition of a, b-unsaturated ketone at C1 or
Acute toxicity
C5, the photolysis of C17 and H-abstraction of C11. The main reactive sites were further determined by
comparison of the frontier orbitals calculation and the proposed mechanism. Based on the toxicological

* Corresponding author. State Key Laboratory of Chem/Biosensing and Chemometrics, Hunan University, Changsha 410082, PR China.
** Corresponding author. State Key Laboratory of Chem/Biosensing and Chemometrics, Hunan University, Changsha 410082, PR China.
E-mail addresses: chem_cbliu@hnu.edu.cn (C. Liu), sllou@hnu.edu.cn (S. Luo).
1
The authors contributed equally to the work.

https://doi.org/10.1016/j.chemosphere.2018.08.032
0045-6535/© 2018 Elsevier Ltd. All rights reserved.
 K. Yin et al. / Chemosphere 212 (2018) 56e66
tests for PDNN degradation, TP396 (TP396-C1Cl and TP396-C5Cl) and TP414-2-1 (TP414-C1CleC5OH)
exhibited much higher toxicity than PDNN, and prolonging reaction time was necessary to achieve PDNN
detoxification.
1. Introduction
Prednisolone (PDNN), one kind of glucocorticoids (GCs), is
frequently and extensively applied in pharmaceuticals exhibiting
anti-inflammatory effect (Herrlich and Go € ttlicher, 2002; Lu et al.,
2006). As the part of “Essential Medicines” (WHO, 2017), PDNN is
consumed increasingly. PDNN is metabolically stable in the body
and can be discharged rapidly via urine with unchanged forms or
metabolite products (Chang et al., 2007; Garg and Jusko, 1991; Jia
et al., 2016). Furthermore, prednisone is extensively metabolized
to PDNN and therefore is likely to increase the environmental loads
of PDNN (Besse and Garric, 2008). Consequently, PDNN is ubiqui-
tous in the aquatic environment worldwide at concentrations
ranging from ng L1 to mg L1 (Chang et al., 2007; Jia et al., 2016).
The concentration of PDNN in river water samples from China was
0.03e0.64 ng L1 (Chang et al., 2007), and PDNN presented in
treated sewage treatment plants (WWTP) effluents from
Switzerland and Czech Republic with the concentration range of
5e24 ng L1 (Macikova et al., 2014). The loads of PDNN in untreated
hospital wastewater ranged between 315 and 1918 ng L1 (Schriks
et al., 2010). The removal efficiency of PDNN was very low (<20%)
by traditional sewage treatment plants (Chang et al., 2007).
PDNN release into the environment primarily via mammalian
excretion. Moreover, urine contains pharmaceuticals at 2e3 orders
of magnitude higher concentrations compared to municipal
wastewater and is considered as a promising aquatic system to
minimize the harm of the pharmaceuticals (Winker et al., 2008).
Urine is usually classified as fresh and hydrolyzed urine (Landry and
Boyer, 2013). Fresh urine has just been discharged from the body
(Landry and Boyer, 2013). Hydrolyzed urine is quite different from
fresh urine for the changes of the original composition due to hy-
drolysis of urea (Landry and Boyer, 2013). To utilize urine as fer-
tilizer efficiently, it is necessary to remove these pharmaceuticals
that might accumulate in the soil and plants and exert hazard (Etter
et al., 2011). Herein, the modified urine was premeditated as the
aquatic system for PDNN degradation. And there is little knowledge
of PDNN degradation in the modified or real water system.
Several studies have demonstrated the deleterious effects (e.g.,
endocrine disrupting effect) of PDNN to the aquatic fauna
(Kugathas and Sumpter, 2011; McNeil et al., 2016). Plasma glucose
concentrations of adult fathead minnows were increased after
treated with 1 mg PDNN L1 for 21 days, and leucocytes were
reduced significantly compared to the control groups (Kugathas
and Sumpter, 2011). Egg masses of Physa acuta exposed to PDNN
(125 mg L1 ~ 1000 mg L1) led to a significant decline in growth,
survival, and heart rate, as well as notable abnormalities in em-
bryonic development (Bal et al., 2016). The continual exposure of
PDNN (7.8 mg L1 ~ 2000 mg L1) may show multigenerational ef-
fects on life-history parameters (e.g., fecundity, average time to first
brood, mean brood size, body length) of crustacean Ceriodaphnia
dubia (Bal et al., 2017). Therefore, it is of great importance to seek an
efficient technique for PDNN elimination from the water.
Various approaches have been developed for PDNN elimination.
During ferrate (VI) treatment, only 15.5e29.4% of PDNN was
removed when increasing ferrate (VI) exposure from 10.7 to
44.6 mg L1 min1 (Yang et al., 2012). Pflug team found O3 can
expeditiously remove PDNN in 3 min, while nearly 60% PDNN will
57
© 2018 Elsevier Ltd. All rights reserved.
be slowly eliminated after 96 h of chlorine treatment with
100 mg L1 Cl2 (Pflug et al., 2017). UV/H2O2 have been reported to
be an efficient method for PDNN elimination (Wols et al., 2013).
Photocatalytic oxidation (P25 and sulphurdoped TiO2) could ach-
ieve the complete PDNN removal at low concentrations
(10e25 mg L1), and the observed oxidation effciency up to
40 mg W1 h1 (Klauson et al., 2013). Another two methods, solar
UVB-induced photodegradation (Cacciari et al., 2017) and a solar
process (DellaGreca et al., 2004) were focus on the mechanism of
PDNN degradation. Those studies suggested that PDNN
(ε242nm ¼ 15000 M1 cm1) was susceptible to UV irradiation, thus
photo-based treatment will be a brilliant method for PDNN elimi-
nation. However, little reports about PDNN degradation in UV/
chlorine treatment which is extensively applied in drinking water
plants or recreational water environment (Sun et al., 2016; Wang
et al., 2016). As an alternative for advanced oxidation processes
(AOPs), UV/chlorine is of great efficiency for various micro-
pollutants elimination (Sun et al., 2016; Yin et al., 2018a). The
diverse reactive species such as hydroxyl radical (
OH) and reactive
chlorine species (RCS) are generated in UV/chlorine process for
contaminant degradation (Fang et al., 2014; Feng et al., 2010). Also,
the residual chlorine can protect the drinking water distribution
systems from pathogenic microorganisms (Xiang et al., 2016).
The main objectives of this study were to: (1) investigate the
degradation kinetics of PDNN by UV/chlorine process under oper-
ational variables, (2) calculate the relative contributions of
OH and
RCS on PDNN degradation to obtain the second-order rate con-
stants of
OH with PDNN, (3) assess the efficacy of UV/chlorine
process in eliminating PDNN in synthetic human urine, (4) propose
the degradation pathways of PDNN by UV/chlorine treatment, and
(5) employ toxicological test to evaluate the acute toxicity and
chronic toxicity of PDNN or its products.
2. Materials and methods
2.1. Chemicals
Prednisolone (PDNN, purity >98%), sodium hypochlorite (NaClO,
containing 10% free chlorine as Cl2), hydrogen peroxide (H2O2, 30%),
tert-butanol (tBuOH), HPLC grade methanol, nitrobenzene (NB) and
benzoic acid (BA) were purchased from Sigma-Aldrich. Other
analytically pure chemicals were bought from Sinopharm Group
Chemical Reagent Co. Ltd. (Beijing, China). Fulvic acid (FA) was
obtained from J&K Scientific, Ltd. (Beijing, China). All solutions
were prepared using Millipore ultrapure water. Fresh stock solu-
tions of free chlorine and Na2SO3 were made every time for the
experiments. The modified urine was prepared according to
Table S1 in the Supporting Information (Landry and Boyer, 2013;
Zhang et al., 2015). Fig. S1 showned the UV absorbance spectrum of
four water types, including unbuffered ultrapure water (UW),
phosphate buffer solution (PBS, 10 mM, pH ¼ 7.0), artificial fresh
(AFU, pH ¼ 6.0) and hydrolyzed (AHU, pH ¼ 9.0) urine. Cl2 repre-
sented chlorine used in the experiment.
2.2. Experimental procedures
All photolysis experiments were conducted in a 150-mL
58 K. Yin et al. / Chemosphere 212 (2018) 56e66
cylindrical quartz reactor (6.3 cm diameter  5 cm deep) equipped
with a collimated beam apparatus using a 15 W low-pressure
mercury lamp with an emission wavelength of 254 nm (Fig. S2).
The average photonic intensity per volume (Iv, 2.12  106 E L1
s1) and the effective path length (L, 2.31 cm) were determined by
H2O2 chemical actinometry (Beltran et al., 1995; Xiao et al., 2015).
The corresponding average UV irradiance (Eavg) was about
2.31 mW cm2 (see Text S1 for details). The temperature was kept
at 25 ± 0.1  C using an air circulating system. The light source was
stabilized at least 30 min before experiments to ensure a relatively
stable radiation intensity.
The degradation of PDNN ([PDNN]0 ¼ 10 mM, the characteristics
of PDNN was shown in Table S2) by UV/chlorine was conducted in a
70-mL solution containing PBS (10 mM, pH ¼ 7.0) and Cl2
(0e56.7 mg L1). [Cl2]0 was the initial concentration. The degra-
dation of PDNN by UV photolysis or chlorine alone was carried out
in a similar manner. Samples (0.5 mL) were withdrawn at specific
time intervals, quenched with 15 mM Na2SO3 at a molar ratio of
sulfite/chlorine of 1.5:1 and filtered through 0.45 mm filter mem-
brane before HPLC analysis for residual PDNN, or LC-MS analysis for
degradation products.
The effects of environmental factors on the photodegradation
kinetics of UV/chlorine were studied by mixing PDNN with ions or
FA. The influence of pH was also researched in 10 mM phosphate
solution with pH adjusted from 4.0 to 10.0 by 1 M H3PO4 or NaOH.
Two types of artificial urine (AFU and AHU) were used to interpret
the degradation of PDNN by UV/chlorine in the complex matrix
(Zhang et al., 2015). The pH of artificial fresh was also adjusted by
1 M H3PO4 or NaOH.
To determine the pseudo-second-order rate constants between
PDNN and
OH or RCS, NB and BA were chosen as competitors in the
competition experiment (Text S1). NB was used as
OH probe due to
its high reactivity with
OH and BA preferred to react with
OH or Cl

(Sun et al., 2016; Wang et al., 2016). The fluence-based rate con-
stants were also listed, and the derivation of fluence-based rate
constants from the rate constants in time units displayed in Text S2.
The reported results were the mean values of at least three ex-
periments, and the error bars in the figures represented the stan-
dard error of the mean.
2.3. Analytical methods
A free chlorine stock solution (about 3.55 g L1 as Cl2) was
prepared by diluting from the 10% sodium hypochlorite (NaClO)
solution in ultrapure deionized water and periodically standardized
by DPD/FAS UV-spectrophotometry (Wang et al., 2016). A sodium
sulfite solution (15 mM) was prepared every time for the experi-
ments. pH was determined with a pH meter (FE20, Mettler Toledo).
The residual concentrations of PDNN, BA and NB were quanti-
fied by a HITACHI L-2130 HPLC coupled with an Inert Sustain C18
column (4.6  250 mm, 5 mm particle size) and L-2420 UVevis de-
tector with the wavelength set at 247 nm, 226 nm and 262 nm,
respectively. The isocratic mobile phase was a mixture of methanol
and water (60:40, v/v), along with a flow rate of 0.5 mL min1
during the quantification of residual PDNN. The sample injection
volume was 10 mL and the column temperature was fixed at 35  C.
Samples used for the identification of degradation in-
termediates/products were pretreated by OASIS@HLB Extraction
Cartridge for the possible contamination of mass spectrometer by
the inorganic agent in the samples obtained from the UV/chlorine
treatment. After pretreatment, HPLC-ESI-tqMS (Agilent Technology
1290/6460 Triple Quad LC/MS) was applied to distinguish the in-
termediates/products. A ZORBAX Eclipse Plus C18 (2.1  50 mm,
1.8-mm and 600 Bar) was used for separation with an isocratic
mobile phase composed of methanol and water (60:40, v/v) at a
flow rate of 0.3 mL min1 and the column temperature was held at
35  C. HPLC-ESI-tqMS was used under positive ESI (þESI) mode
(more details shown in Table S3).
2.4. Calculation of frontier electron densities (FEDs) of PDNN
The Gaussian 09 is often applied for the study of pollutant
degradation (Frisch et al., 2009; Ohko et al., 2002; Zhang et al.,
2017). The frontier electron densities (FEDs) calculations of the
highest occupied molecular orbital (HOMO) and the lowest unoc-
cupied molecular orbital (LUMO) based on the optimized geometry
of PDNN was operated by the Gaussian 09 with hybrid density
functional theory (DFT) on a basis set of B3LYP/6e31 (d, p) (Xiao
et al., 2015). And the values of 2FED2HOMO and
FED2HOMO þ FED2LUMO were obtained to predict the reaction sites for
electron transfer and radical attack, respectively (Ohko et al., 2002).
2.5. Toxicity tests
The Microtox test and Quantitative structure-activity relation-
ship (QSAR) based Ecological Structure-Activity Relationship Model
(ECOSAR) program were employed to evaluate the acute toxicity
and chronic toxicity of PDNN or its products (Zhang et al., 2016a).
The Microtox test is performed with the marine bacterium Vibrio
fischeri. Substances that interfere with the cellular energy meta-
bolism cause a decrease in light emission, which is indicative of
general toxicity (Yin et al., 2017, 2018b). To evaluate this non-
specific toxicity, the level of bacteria bioluminescence was
measured before exposing to sample and after 25-min exposure to
sample.
3. Results and discussion
3.1. PDNN degradation in PBS
The PDNN degradation in UV and UV/chlorine process followed
the pseudo-first-order kinetics (Fig. 1A). The observed rate constant
(kobs,PDNN) by UV/chlorine was 5.53 (±0.13)  103 s1 corre-
sponding to a fluence-based rate constant of 2.39 (±0.54) cm2 J1.
While PDNN degradation by chlorination alone was very poor
(Fig. 1B), and PDNN was susceptible to UV irradiation with a
kobs,PDNN of 4.20 (±0.085)  103 s1 (i.e., 1.82 (±0.037) cm2 J1)
(Cacciari et al., 2017; Wols et al., 2015). These results indicated that
oxidation of PDNN by chlorine alone could be neglected. Hence,
establishing a kinetic model of PDNN degradation (Eq. (1)) were
involved in the combination of UV direct and radical indirect
photolysis for UV/chlorine process.
There existed several reactive species when degrading PDNN by
UV/chlorine. It reported that the photolysis of PDNN under UV
irradiation could produce reactive oxygen species (ROS, including

OH, singlet oxygen (1O ), O
- and H O ) (Cacciari et al., 2017), and
2 2 2 2
the common reactive species in UV/chlorine process at pH 7.0 were

OH and RCS (including Cl
, Cl
- and ClO
) (Fang et al., 2014). The
2
participation of ROS except
OH could be ruled out (Cacciari et al.,
2017).
To identify the reactive species involving in PDNN elimination,
tBuOH was employed. tBuOH could rapidly react with
OH and Cl

with the rate constants as high as 6.0  108 M1 s1 and
1.9  108 M1 s1, respectively (Gilbert et al., 1988), while reacting
with Cl2
- with a relatively low rate constant of 7  102 M1 s1
(Hasegawa and Neta, 1978). 1 mM tBuOH was enough to scavenge

OH and Cl
in UV/chlorine process (Fig. S5), and 99.8%
OH was
quenched by 1 mM tBuOH according to the calculation (Wang et al.,
2016). After the addition of 1 mM tBuOH, the kobs,PDNN was
decreased by 11.57% (from 5.53 to 4.89  103 s1). There existed a
 K. Yin et al. / Chemosphere 212 (2018) 56e66 59

Fig. 1. The pesudo first-order kinetics of PDNN degradation by UV/chlorine, UV and UV/chlorine þ1 mM tBuOH (A) and effect of chlorine dosage on PDNN degradation by
chlorination alone (B). [Cl2]0 ¼ 7.1 mg L1 for (A), [PDNN]0 ¼ 10 mM, pH ¼ 7.0, and Eavg ¼ 2.31 mW cm2.

difference between rate constant of UV/chlorine þ1 mM tBuOH and
UV alone (Fig. 1A) indicating the existence of other contributors in
UV/chlorine process, and it most likely to be Cl2
- and ClO
. And, Cl
,
Cl2
- and ClO
were treated as RCS to determinate the relative
contribution of
OH and RCS.
Summarily,
OH and RCS synergistically degraded PDNN in UV/
chlorine process. And the time-based degradation in UV/chlorine
system can be expressed as Eqs. (1)e(3).


½PNDDt
ln ¼ kobs;PDNN *t ¼ ðkdirect þ kindirect Þ*t
½PDNN0
can selectively react with
OH (3.9  109 M1 s1) (Buxton et al.,
1988). BA is prone to react with
OH and Cl
with the reaction
rate constants value of 5.9  109 M1 s1 (Buxton et al., 1988) and
1.8  1010 M1 s1 (Ma rtire et al., 2001), respectively. The steady-
state concentration of
OH in UV/H2O2 system ([
OH]ss1) was
calculated based on Eq. (4) (Buxton et al., 1988; Sun et al., 2016).
PDNN was difficult to be degraded by H2O2 alone (Cacciari et al.,
2017), therefore, k
OH,PDNN was obtained in UV/H2O2 process
(Fig. S6) according to Eqs. (4) and (5) (Sun et al., 2016).
(1) (4)

P
Kindirect¼K
OH,PDNN*[
OH]þ kRCS,PDNN*[RCS]. (2)

(3)

where kdirect (s1) represents the pesudo-first-order rate constant of

PDNN by UV direct photolysis, kindirect (s1) represents the pesudo-
first-order rate constant of PDNN by indirect photolysis attributing (5)
to the reactive species, such as
OH and RCS. The k
OH,PDNN (M1 s1)
and kRCS,PDNN (M1 s1) are the second-order rate constant of PDNN According to the obtained [
OH]ss1 and known k
OH,NB, k
OH,PDNN
P was determined to be 1.10  109 M1 s1 (in Text S3). In the co-
with
OH and RCS, respectively kRCS;PDNN *½RCS. presents the sum
of the product between the concentration of each chlorine radical presence of PDNN, NB and BA for UV/chlorine treatment, the
and its second-order rate constant with PDNN. similar method can be used to calculate the steady-state concen-
The contribution of the radicals (
OH and RCS) was only 24.05% tration of
OH ([
OH]ss2), and the steady-state concentration of Cl

for PDNN degradation, which indicated that the dominant degra- ([Cl
]ss) from Eqs. (6) and (7), as shown in Fig. S7.
dation way of PDNN by UV/chlorine process was the direct
photolysis. (6)

3.1.1. Determined the relative contribution of
OH and RCS

To investigate the reactivity of PDNN with
OH and RCS, the (7)
competing kinetic method with NB and BA probe was adopted. NB

(8)
60 K. Yin et al. / Chemosphere 212 (2018) 56e66
In Eq. (8), [Cl2
-]ss and [ClO
]ss represented the steady-state
concentration of Cl2
- and ClO
, respectively. kCl
;PDNN , kCl
 or
2 ;PDNN
kClO
;PDNN standards for the reaction rate constants of Cl
, Cl2
- or
ClO
with PDNN, respectively. [RCS]ss represented the steady-state
concentration of RCS. The values of [
OH]ss2 and [Cl
]ss were
calculated to be 8.97 (±0.039)  1013 M and 1.97
14
(±0.86)  10 M, respectively. Based on Eq. (9), the contribution
of
OH and RCS for PDNN degradation in UV/chlorine process were
calculated to be 17.84% and 6.21%, respectively. However, there
were no reports about the determination of the coexistence of Cl2
-
and ClO
, hence it was hard to estimate or kClO
;PDNN .
3.1.2. Economic analysis among UV, UV/chlorine and UV/H2O2
To seek a cost-effective AOP method for PDNN elimination,
specific kinetics parameters for UV photolysis, UV/chlorine and UV/
H2O2 were calculated and listed in Table 1 (the detailed calculation
was shown in Text S4). The EE/OT (kWhm3 order1) is defined as
the electrical energy (kWh) required to degrade the contaminants
by one order magnitude in m3 of contaminated water, which was
usually applied to optimize contaminants removal by economically
(Bolton and Cater, 1994). k1 (s1) is the time-based pseudo-first
order rate constant of the pollutant decay in the UV/chlorine sys-
tem, k1’ (m2 J1) is the fluence-based pseudo-first order rate con-
stants of the pollutant decay in the collimated beam apparatus for
the same water quality parameters as used in the UV/chlorine
system.
At a specific degradation conditions of PDNN removal (PBS, pH
7.0), k1 and k1’ in different degradation method follows the order:
UV/chlorine ＞ UV/H2O2 ＞ UV. Moreover, EE/OT among those
system follows the order: UV/chlorine ＜ UV/H2O2 ＜ UV. Those
results suggested that UV/chlorine and UV/H2O2 were more cost-
effective than UV alone.
During water treatment, effluents require a chlorine residual,
the UV/chlorine AOP avoids the additional chlorine demand
required to quench residual other oxidants (e.g., H2O2, persulfate or
peroxylmonosulfate) (Xiang et al., 2016). Furthermore, radical
production from free chlorine was more efficient due to high molar
absorption coefficients and quantum yields for chlorine photolysis
(Fang et al., 2014). Thus, UV/chlorine was regard as an effective and
novel method for PDNN elimination.
3.2. Impacts of chlorine dosage and pH
Fig. 2A displayed the kobs,PDNN under different chlorine dosages
(0e56.7 mg L1) at pH 7.0 in UV/chlorine process. The kobs,PDNN
increased from 4.10 (±0.096)  103 to 5.53 (±0.13)  103 s1 (i.e.,
1.77 (±0.041) to 2.39 (±0.054) cm2 J1) as increasing chlorine
dosage from 0 to 7.1 mg L1. Under a certain amount of UV
Table 1
Specific kinetics parameters for UV photolysis, UV/chlorine and UV/H2O2 of PDNN in
PBS at 254 nm.
k1 （s1） k1’ （cm2 J1） EE/OT
（kWh m3 order1）
UV/chlorine 5.53  103 2.39 0.54
UV/H2O2 5.14  103 2.23 0.53
UV 4.20  103 1.82 0.61
(9)
irradiation (Eavg ¼ 2.31 mW cm2), kobs,PDNN gradually reached pla-
teaus at chlorine dosage of 7.1 mg L1 for the full photon absor-
bance was reached, and then the production of reactive species
remained unchanged with increasing chlorine dosage above
7.1 mg L1 (Wang et al., 2016). And the excess chlorine dosage
above 7.1 mg L1 can scavenge the reactive species as radical
quencher (Table S4) (Zhang et al., 2015), so the value of kobs,PDNN
decreased gradually. Hence, the optimum dosage of chlorine was
7.1 mg L1 for the following experiments.
The effect of pH (4.0e10.0) on kobs,PDNN was shown in Fig. 2B.
There appeared no significant pH-dependence effect with kobs,PDNN
value close to 5.84 (±0.32)  103 s1 (i.e., 2.53 (±0.014) cm2 J1).
The two possible reasons were listed below. (i) The contribution
from ClO
and ClOH
- might exactly compensate the decrease of

OH, thus leading to little change in the degradation rate of PDNN
(Sun et al., 2016). pH increase could promote the scale of OCl, and
a large quantity of ClO
formed due to that the rate constants of
OH
and Cl
reacting with OCl are 4.4 and 2.7 times faster than that
with HOCl, respectively (Table S4). Moreover, the reaction to form
ClOH
- also accelerated with more OCl, and the O
- produced by
the photolysis of OCl could also be rapidly transformed into
OH at
pH below 11.9 (Fang et al., 2014). (ii) Though pH could affect the
dissociation of HOCl/OCl (pKa ¼ 7.5) with the production of free
radicals (Feng et al., 2010) available in the solution, there is no
obvious difference between 4HOCl (0.62) and 4OCl (0.55) at 254 nm
which incorporated subsequent free chlorine or chloramine scav-
enging by the
Cl and
OH daughter radicals (Chuang et al., 2017).
The pH-independence effect also existed during PDNN treated by
UV/H2O2 (Fig. S8), and the kobs,PDNN value in UV/chlorine and UV/
H2O2 system at different pHs were nearly the same. The results
strongly confirmed the dominant role of direct photolysis.
3.3. Effects of common anions, cations and FA
Fig. 3A showed the effects of common anions, including Cl, Br,
HCO 2
3 and SO4 , on the degradation of PDNN in UV/chlorine system.
The kobs,PDNN decreased after addition of different concentration of
anions. Cl could react rapidly with
OH and Cl
to produce ClOH
-
and Cl2
-, respectively (Table S4) (Fang et al., 2014; Sun et al., 2016).
The produced ClOH
- could be reversibly decomposed into
OH with
a similar formation rate constant of ClOH
- under the neutral/basic
condition (Xiao et al., 2016), which indicated that the reaction be-
tween
OH and Cl could be negligible. When increasing Cl con-
centration from 0.05 to 5 mM, the concentration of Cl2
- could be
increased by 2 orders of magnitude (Wu et al., 2016). And the
radical scavenging reactions between Cl2
- and
OH, Cl
or Cl2
-
could become obvious (Table S4), where the radical recombination
HOCl/Cl2 would compete with PDNN for photons. The reformation
of HOCl/Cl2 can be found in Chuang team’ work, which concluded
that radical reactions (indirect photolysis) accounted for ~80% of
free chlorine loss or reformation for the UV/free chlorine AOP
(Chuang et al., 2017). Even so, increasing Cl concentration from 3
to 5 mM did not exert the negative effect in this case. Likewise, an
addition of Br reduced the kobs,PDNN for the low reactive species
generated (ClBr
- and BrOH
-) (Table S4) (Yang et al., 2014).
A common range of HCO 1
3 (50e200 mg L ) in natural water
slightly lowered down kobs,PDNN in UV/chlorine process, as shown in
Fig. 3A. In UV/chlorine system, HCO

3 could react with OH and Cl


 K. Yin et al. / Chemosphere 212 (2018) 56e66 61

Fig. 2. Effect of chlorine dosage (A) and pH (B) on the kobs,PDNN in UV/chlorine process. [PDNN]0 ¼ 10 mM and Eavg ¼ 2.31 mW cm2, pH ¼ 7.0 for (A) and [Cl2]0 ¼ 7.1 mg L1 for (B).

Fig. 3. Influence of anions (A) and FA (B) on the kobs,PDNN in UV/chlorine process. [PDNN]0 ¼ 10 mM, [Cl2]0 ¼ 7.1 mg L1, pH ¼ 7.0 and Eavg ¼ 2.31 mW cm2 [FA]0 represents the
concentrations of FA (mg L1) at time zero.

with rate constants of 8.5  106 M1 s1 and 2.4  109 M1 s1
(Buxton et al., 1988, 2000), respectively. And the converted product
was CO3
-, a less oxidation potential radical (Yang et al., 2014),
which has low reactivity with electron-poor pollutants. The
kobs,PDNN gradually decreased from 5.53 (±0.13)  103 to 4.53
(±0.093)  103 s1 (i.e., 2.39 (±0.054) to 1.96 (±0.040) cm2 J1)
with the addition of SO2 2
4 from 0 to 5 mM. SO4 scavenged both OH

other hand, NH3 acted as radical scavenger rather than contributor
(k ¼ 1.0  106 M1 s1) and Cl
(k ¼ 2.5  108 M1 s1) to produce
SO4
- (Table S4), which is a more selective oxidation specie than
OH
(Hasegawa and Neta, 1978). And, SO4
- usually prefers to react with
contaminants containing carboxyl groups, amino groups, and
benzene rings (Zhang et al., 2016c). Then, a self-scavenging effect of
SO4
- would continue (k ¼ 7.0  108 M1 s1, Table S4). Hence, the
existence of anions could decrease the kobs,PDNN in UV/chlorine
process.
The effect of cations (Naþ, Kþ, Mg2þ and NHþ 4 ) on the PDNN
degradation were investigated as well. Except for NHþ 4 (Fig. S9), all
of other cations exhibited the negligible impacts on kobs,PDNN.
Generally, ammonia-nitrogen (NH3/NHþ 4 ) will react with HOCl/OCl


to produce chloramines (Table S4), including NH2Cl, NHCl2 and
NCl3, a type of disinfectant widely used in drinking water treatment
plants. The produced chloramine in solution at pH < 3 is NCl3, at pH
3e5, NHCl2, at pH 5e8, a mixture of NH2Cl and NHCl2, and at
pH > 8, NH2Cl (Kovacic et al., 1970). A low molar ratio of chlorine
dose (Cl2) to NH3/NHþ 4 (Cl/N ¼ 0.02 < 1) will lead to a slow oxidation
of NH3/NHþ 4 . Thus, at pH 7, the dominant chloramine was NH2Cl
with a low concentration. The inhibition of NH3/NHþ 4 to PDNN
degradation could be explained by two reasons. On the one hand,
the quantum yields for radical generation (Cl
and NH2
) by UV
photolysis of NH2Cl is 0.20, which lower than that of HOCl/OCl
(0.62/0.55) (Chuang et al., 2017). The photolysis of chloramines
generates the amine radical (NH2
), which was reported not to react
quickly with organic compounds in general (NIST, 2016). On the
to PDNN elimination by consuming
OH (k ¼ 9.0  107 M1 s1) and
generated less reactive nitrogen species (NH2
,
NO2,
NO, NH2O2

and ONOOH/ONOO) (Table S4) (Zhang et al., 2015).
FA was used as an alternative for natural organic matter (NOM)
(Zhang et al., 2016b). The kobs,PDNN decreased from 5.53 (±0.13) to
3.37 (±0.51)  103 s1 when increasing FA dosage from 0 to
20 mg L1 (Fig. 3B). The FA inhibition effect on PDNN degradation
should be ascribed to the fact: (i) FA worked as a UV inner filter (UV
filter effect calculated to be 22.66%, shown in Fig. S10), and absor-
bed competitively photon with PDNN and chlorine (Luo et al.,
2012); (ii) FA played a role of radical scavenger consuming Cl
and
OH in UV/chlorine process (Table S4) (Fang et al., 2014).
3.4. PDNN degradation in different water matrices
The degradation of PDNN was suppressed in two synthetic urine
including AFU and AHU, while the degradation of PDNN was
enhanced in UW (Fig. 4) by compared with that in PBS. Large
amounts of anions in AFU could account for the decrease of kobs,PDNN
(as discussed in 3.4). In addition, the second-order rate constants of
62 K. Yin et al. / Chemosphere 212 (2018) 56e66

tend to happen at that part consistenting with the literature

(DellaGreca et al., 2003; Klauson et al., 2013).
Additionally, the evaluated values of FEDs (2FED2HOMO and
FED2HOMO þ FED2LUMO) based on the optimized structure (Fig. S11)
were listed in Table S5. The relatively high 2FED2HOMO values of the
a, b-unsaturated ketone-moiety, especially at the C2, C4 and O22
atoms, meant the easier electron transfer by RCS or
OH radical.
Additionally, a, b-unsaturated ketone-moiety, especially at the C1,
C5 and O22 atoms with higher values of FED2HOMO þ FED2LUMO, a
higher density of the two electrons (one occupying the highest
orbital and the other occupying the lowest unoccupied orbital of
the ground state), is more susceptible to attack by a radical reagent
(Fukui et al., 1954). Consequently, it allowed to temporarily suppose
that C2 or C4 atoms should be the sites of electron transfer, and C1,
C5 or O22 was involved in addition reaction.
Based on the theoretical calculation, the experiments to acquire
the degradation products and mechanism of PDNN in UV/chlorine
process were carried out. However, it is almost no mineralization of
PDNN by UV/chlorine (Fig. S12), which indicated that the formed
intermediates/products were still present in solution. The extracted
ion chromatograph (EIC) and time-dependent profile on these
Fig. 4. Comparison of PDNN degradation by UV/chlorine in different water matrices.
[PDNN]0 ¼ 10 mM, [Cl2]0 ¼ 7.1 mg L1 and Eavg ¼ 2.31 mW cm2.
detected byproducts are shown in Fig. S13 and Fig. 6, respectively.
The peak areas of TP298, TP300, TP318, TP396, TP406 and TP415 (TP

urea and citrate with
OH were 7.9  105 M1 s1 and
2.4  108 M1 s1, respectively (Buxton et al., 1988; Zepp et al.,
1992). Hence, citrate and urea in AFU were also responsible for
the suppression effect. As for AHU, the inhibitory effect resulted
from the high concentration of anions and NHþ 4 (as discussed in
3.4). Additionally, the kobs,PDNN in AHU was small as that in AFU,
which might be ascribed to the high concentration of HCO 3
(0.25 M) and ammonia (0.5 M) in AHU (Zhang et al., 2015). After the
addition of PDNN and chlorine, there represented nearly no
changes of pH in UW. So, the higher kobs,PDNN in UW than PBS could
be due to H2PO 2
-
4 /HPO4 that could produce HPO4 by reacting with
the free radicals (Table S4).

3.5. Degradation mechanism of PDNN in UV/chlorine process

To determine the preferential sites, the frontier molecular or-

bitals of PDNN was calculated, as shown in Fig. 5. The HOMO, a
pebonding orbital, is mainly located on the part of a, b-unsaturated
ketone-moiety of PDNN. The frontier orbital theory elaborated that
the electrophilic reactions of radicals most likely occur in these
electron rich regions while the nucleophilic reactions most likely
take place at the moiety of LUMO regions (Fukui, 1982). Therefore,
the most possible reaction moiety was located on the a, b-unsat- Fig. 6. Evolution on transformation products of PDNN degradation by UV/chlorine.
urated ketone-moiety of PDNN. And the direct photolysis of PDNN [PDNN]0 ¼ 10 mM, [Cl2]0 ¼ 7.1 mg L1, pH ¼ 7.0, and Eavg ¼ 2.31 mW cm2.

Fig. 5. Atomic orbital compositions of frontier molecular orbitals (A, HOMO; B, LUMO) of PDNN, which were calculated using Gaussian 09 program at the B3LYP/6e31 (d, p) level.
The red and green colors represent the positive and negative phases of the molecular orbital of PDNN, respectively, and the blue arrow indicated the possible reaction site (a, b-
unsaturated ketone-moiety). (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
 K. Yin et al. / Chemosphere 212 (2018) 56e66
Scheme 1. Proposed pathway of PDNN degradation in UV/chlorine process.
and number represented the transformation product and its mo-
lecular weight, respectively) increased with reaction progress,
while the peaks of TP336, TP334, TP352 and TP387 arose around
60 s (Fig. 6). Some photolysis products of PDNN in UV/chlorine
produced, including TP318-1, TP318-2, TP298, TP334-2, TP336-2,
TP300-1, TP316-1 and TP300-2, were also founded quite different
from PDNN photodegradation under a solar simulator (Scheme S1)
(DellaGreca et al., 2004).
The
OH or RCS reacted with organic pollutants by H-abstrac-
tion, addition and electron-transfer (Stefan, 2017). The single
electron transfer of
OH or RCS was commonly in electron-rich
organic compounds, such as phenol, toluene, aniline, and hexam-
ethylbenzene (Hasegawa and Neta, 1978). Moreover, the product of
PDNN by chlorine alone was not considered in the scheme of PDNN
by UV/chlorine for the poor elimination of PDNN by chlorine. So,
the proposed mechanism will mainly concentrate on the addition,
H-abstraction reaction and photolysis (Stefan, 2017).
According to these detected intermediates/products, the
63
proposed reaction sites by calculation and known literature
(Cacciari et al., 2017; Pflug et al., 2017), a possible degradation
mechanism of PDNN treated by UV/chlorine was proposed (Scheme
1) and the possible reactive positions in UV/chlorine process were
shown in Fig. S14, which contained three possible degradation
pathways.
For the route 1, monochlorinated PDNN (TP396) was formed by
the direct addition of Cl
to C1 or C5, as Cl
or Cl2
- will ready to react
with the C]C bond of unsaturated aliphatic compounds by Cl

addition (Stefan, 2017) rather than H-abstraction. And the similar
mechanism was proposed in clofibric acid degradation by UV/
chlorine (Tang et al., 2018). Afterwards, TP414 derived from TP396
by the deep addition of
OH of C5 or C1 just like the reason for Cl

addition. There are three supposed sites of addition reaction,
however,
OH addition was preferentially to C5 or C1 site in the
unsaturated bonds instead of addition to the electron-deficient C
atom in C]O22 (Stefan, 2017). Moreover, the product of H-
abstraction from C2 or C4 was not detected possibly due to the
64 K. Yin et al. / Chemosphere 212 (2018) 56e66

conjugation effect of a, b-unsaturated ketone-moiety which will
decrease the stabilization of the C-centered radical generated from
H-abstraction of C2 or C4 (Stefan, 2017).
Route 2 was initiated by the photolysis of PDNN. Firstly, TP300-
1, androgen 11b-hydroxyboldione produced by the cleavage of C17
side-chain cleavage under UV irradiation (DellaGreca et al., 2003),
which was consistent with the photolysis of PDNN by UV (Fig. S15,
Scheme S2). Then it reacted with
OH and Cl
by radical addition to
yield the hydroxylated androgen 11b-hydroxyboldione (TP318-1)
and chlorinated androgen 11b-hydroxyboldione (TP336-1). Mean-
while, TP318-1 also appeared under the UV photolysis. Simulta-
neously, TP334-1 and TP316-1 were produced by H-abstraction of
C9 due to the neighboring electron-donating substituents (e.g.,
eCH3, eOR, eNR2), which will stabilize the resulting C-centered
radical and enhance the H-abstraction by
OH or RCS (Hasegawa
and Neta, 1978; Stefan, 2017). Moreover, TP300-1 might undergo
the UV photolysis to produce TP316-1 by
OH attacking. Afterwards,
the formation of multiple hydroxylated and chlorinated products
(TP336-2, TP334-2, TP352-1, TP352-2 and TP406) was attributed to
the further attack of
OH and Cl
. Additionally, a photo rearrange-
ment reaction occurred between TP300-1 and TP300-2, because
the cross-conjugated ketone in TP300-1 tended to happen the
lumiketone rearrangement between C1 and C3, which also called 1, 3
(p-p*) reaction (Ricci et al., 2003). TP318-2 was formed from the
hydrolysis of TP300-2 (DellaGreca et al., 2004).
As for route 3, the photo-product androgen d1-adrenosterone
(TP298) generated under UV irradiation (DellaGreca et al., 2003).
Likewise, hydroxylation (TP316-2) and chloro-hydroxylated
androgen d1-adrenosterone derivatives (TP352-3 and TP387)
were attributed to the further attack of
OH and Cl
. Meanwhile, the
UV irradiation will activate the C]C and then
OH addition more
easily to happen. Some inter transformation may exist among these
products, especially the H-abstraction of alcohols (such as the
transformation from TP300-1 to TP298) (Pflug et al., 2017).
Generally, based on the kinetics, the removal of PDNN was
primarily attributable to direct UV photolysis other than radicals
such as
OH and RCS in UV/chlorine. However, the proposed
mechanism reflected that the significant role of
OH and RCS in the
transformation pathways. It founded that the reaction sites were
mainly concentrated on a, b-unsaturated ketone-moiety by com-
parison with the orbital calculation and the proposed mechanism.
showed almost no changes, which suggested that formed photo-
byproducts have similar toxicity with PDNN. Meanwhile, the dif-
ference of IL evolution between UV and UV/chlorine could stem
from the chlorinated byproducts formed. So, it is significantly to
assess the products in UV/chlorine system.
3.6.2. Ecotoxicity
To comprehensively evaluate the toxicity of the products, QSAR
based ECOSAR program was applied to predict the ecotoxicity of
PDNN and its products. The prediction of the ECOSAR program is
based on the definite chemical structures and the basic character-
istics of PDNN. The optimum predication would be to compare the
chemical class definition with the structural features of PDNN. The
class ketone alcohols had the closest corresponding toxicity value
and thus was selected for prediction. Because of the structural
similarity of PDNN and its products, the same class was selected for
its products. The toxicities for fish, daphnid and green algae were
calculated and the results were shown in Table S6. TP396 (TP396-
C1Cl and TP396-C5Cl) and TP414-2-1 (TP414-C1CleC5OH) exhibi-
ted much higher toxicity than PDNN in both acute toxicity and
chronic toxicity. Moreover, it founded if reaction happens on C2 or
C4 of a, b-unsaturated ketone, TP396 (TP396-C2Cl and TP396-C4Cl)
and TP414-2-1 (TP414-C4CleC2OH) also showed much higher
toxicity than PDNN (Table S7). Making a comparison between those
products from reactions on C1 (or C5) and C2 (or C4), the later one
displayed more harmful to aquatic organisms.
This result of toxicological test indicated that prolonging treat-
ment time of PDNN by UV/chlorine could achieve not only the
degradation but also detoxification. And, the derivatives of PDNN
should be addressed.
4. Conclusions
The degradation of PDNN in UV/chlorine process was preceded
than chlorination and UV photolysis and followed the pseudo-first-
order kinetics. The contribution of
OH and RCS (Cl
, Cl2
- and ClO
)
for PDNN degradation in UV/chlorine process was 17.84% and 6.21%,
respectively. The pH showed a negligible effect on the degradation

3.6. Toxicological test

3.6.1. Acute toxicity tests

The acute toxicity tests with luminescent bacterium Vibrio
fischeri on PDNN and degradation samples withdrawn at different
time were determined. Though there are some standards of
byproducts commercially unavailable, the acute toxicity tests of a
mixture of byproducts and residual PDNN is highly valuable. The
inhibitory rates of luminescence (IL, %) of bacterium Vibrio fischeri
were calculated according to Eq. (10).

I0  It
IL ¼ *100% (10)
I0

where I0 represents the luminescence intensity of negative control

sample and It represents the luminescence intensity of the sample
at reaction time.
With prolonged treatment time under UV/chlorine condition,
the IL increased to two peaks (P1 and P2), which indicated the in-
termediate products of PDNN had toxicity higher than the parental
drug (Fig. 7). The IL value of the sample after treatment for 5 min Fig. 7. Acute toxicity of luminescent bacterium Vibrio fischeri in samples during
was 8.42% and approximately close to PDNN itself. However, the IL different reaction period. [PDNN]0 ¼ 10 mM, [Cl2]0 ¼ 7.1 mg L1, pH ¼ 7.0 and
value of degradation samples withdrawn from UV photolysis Eavg ¼ 2.31 mW cm2.
 K. Yin et al. / Chemosphere 212 (2018) 56e66
of PDNN in UV/chlorine process. Anions (Cl, Br, HCO 
3 and SO4 ),
NHþ4 and FA exhibited inhibition effect on PDNN degradation. The
suppression effect also occurred in the synthetic urine. The trans-
formation products of PDNN degradation mainly included the
chlorinated PDNN, androgen 11b-hydroxyboldione, androgen d1-
adrenosterone and their derivatives. a, b-unsaturated ketone-
moiety was firmly determined to be the preferential reaction
moiety by comparison of the frontier orbitals calculation and the
proposed mechanism. The acute toxicity tests for luminescent
bacterium Vibrio fischeri indicated that higher toxic byproducts
were generated during UV/chlorine process, which could account
for the containing chlorinated byproducts. Even so, prolonging
treatment time (~5 min) for UV/chlorine in this study, the IL value
could be reduced to the level as with UV irradiation. Hence, UV/
chlorine process not only increase the rate of PDNN degradation,
but also achieve PDNN detoxification under appropriate treatment
time.
Acknowledgements
This work was supported by the National Natural Science
Foundation of China (51478171 and 51778218).
Appendix A. Supplementary data
Supplementary data related to this article can be found at
https://doi.org/10.1016/j.chemosphere.2018.08.032.
References
Bal, N., Kumar, A., Du, J., Nugegoda, D., 2016. Prednisolone impairs embryonic and
posthatching development and shell formation of the freshwater snail, Physa
acuta. Environ. Toxicol. Chem. 35, 2339e2348.
Bal, N., Kumar, A., Du, J., Nugegoda, D., 2017. Multigenerational effects of two glu-
cocorticoids (prednisolone and dexamethasone) on life-history parameters of
crustacean Ceriodaphnia dubia (Cladocera). Environ. Pollut. 225, 569e578.
Beltran, F.J., Ovejero, G., Garciaaraya, J.F., Rivas, J., 1995. Oxidation of polynuclear
aromatic-hydrocarbons in water .2. UV-Radiation and ozonation in the presence
of UV-radiation. Ind. Eng. Chem. Res. 34, 1607e1615.
Besse, J.P., Garric, J., 2008. Human pharmaceuticals in surface waters. Toxicol. Lett.
176, 104e123.
Buxton, George V., Greenstock, C.L., Helman, W.P., Ross, A.B., 1988. Critical review of
rate constants for reactions of hydrated electrons, hydrogen atoms and hy-
droxyl radicals (
OH/
O-) in aqueous solution. J. Phys. Chem. Ref. Data 17,
513e886.
Bolton, J.R., Cater, S.R., 1994. Homogeneous photodegradation of pollutants in
contaminated water: an introduction. Aq. Surf. Photochem 467e490.
Buxton, George V., Bydder, Mark, Salmon, G.A., Williams, J.E., 2000. The reactivity of
chlorine atoms in aqueous solution. Part III. The reactions of Cl
with solutes.
Phys. Chem. Chem. Phys. 2, 237e245.
Cacciari, R.D., Reynoso, E., Montejano, H.A., Biasutti, M.A., 2017. Photodegradation of
prednisolone under UVB solar irradiation. Role of photogenerated ROS in the
degradation mechanism. Photochem. Photobiol. Sci. 16, 1717e1726.
Chang, H., Hu, J., Shao, B., 2007. Occurrence of natural and synthetic glucocorticoids
in sewage treatment plants and receiving river waters. Environ. Sci. Technol 41,
3462e3468.
Chuang, Y.H., Chen, S., Chinn, C.J., Mitch, W.A., 2017. Comparing the UV/mono-
chloramine and UV/free chlorine advanced oxidation processes (AOPs) to the
UV/hydrogen peroxide AOP under scenarios relevant to potable reuse. Environ.
Sci. Technol. 51, 13859e13868.
DellaGreca, M., Fiorentino, A., Iesce, M.R., Isidori, M., Nardelli, A., Previtera, L.,
Temussi, F., 2003. Identification of phototransformation products of prednisone
by sunlight: toxicity of the drug and its derivatives on aquatic organisms. En-
viron. Toxicol. Chem. 22, 534e539.
DellaGreca, M., Fiorentino, A., Isidori, M., Lavorgna, M., Previtera, L., Rubino, M.,
Temussi, F., 2004. Toxicity of prednisolone, dexamethasone and their photo-
chemical derivatives on aquatic organisms. Chemosphere 54, 629e637.
Etter, B., Tilley, E., Khadka, R., Udert, K., 2011. Low-cost struvite production using
source-separated urine in Nepal. Water Res. 45, 852e862.
Fang, J., Fu, Y., Shang, C., 2014. The roles of reactive species in micropollutant
degradation in the UV/free chlorine system. Environ. Sci. Technol. 48,
1859e1868.
Feng, Y., Smith, D.W., Bolton, J.R., 2010. Corrigendum: photolysis of aqueous free
chlorine species (HOCl and OCl-) with 254 nm ultraviolet light. J. Environ. Eng.
Sci. 37, 179e180.
65
Frisch, M.J., Trucks, G.W., Schlegel, H.B., et al., 2009. Gaussian 09, A.01 (Wallingford,
CT).
Fukui, K., 1982. Role of frontier orbitals in chemical reactions. Science 218, 747e754.
Fukui, K., Yonezawa, T., Nagata, C., Shingu, H., 1954. Molecular orbital theory of
orientation in aromatic, heteroaromatic, and other conjugated molecules.
J. Chem. Phys. 22, 1433e1442.
Garg, V., Jusko, W.J., 1991. Simultaneous analysis of prednisone, prednisolone and
their major hydroxylated metabolites in urine by high-performance liquid
chromatography. J. Chromatogr. A 567, 39e47.
Gilbert, B.C., Stell, J.K., Peet, W.J., Radford, K.J., 1988. Generation and reactions of the
chlorine atom in aqueous solution. J. Chem. Soc. Faraday. Trans. 1 (84),
3319e3330.
Hasegawa, K., Neta, P., 1978. Rate constants and mechanisms of reaction of chloride
(Cl-2) radicals. J. Phys. Chem. 82, 854e857.
Herrlich, P., Go € ttlicher, M., 2002. The anti-inflammatory action of glucocorticoid
hormones. In: Cato, A.C.B., Sch€ acke, H., Asadullah, K. (Eds.), Recent Advances in
Glucocorticoid Receptor Action, vol. 40. Ernst Schering Research Foundation
Workshop, Springer, Berlin, Heidelberg, pp. 297e304.
Jia, A., Wu, S., Daniels, K.D., Snyder, S.A., 2016. Balancing the budget: accounting for
glucocorticoid bioactivity and fate during water treatment. Environ. Sci. Tech-
nol. 50, 2870e2880.
Klauson, D., Pilnik-Sudareva, J., Pronina, N., Budarnaja, O., Krichevskaya, M.,
Ka€kinen, A., Juganson, K., Preis, S., 2013. Aqueous photocatalytic oxidation of
prednisolone. Cent. Eur. J. Chem. 1620e1633.
Kovacic, P., Lowery, M.K., Field, K.W., 1970. Chemistry of N-bromamines and N-
chloramines. Chem. Rev. 70, 639e665.
Kugathas, S., Sumpter, J.P., 2011. Synthetic glucocorticoids in the environment: first
results on their potential impacts on fish. Environ. Sci. Technol. 45, 2377e2383.
Landry, K.A., Boyer, T.H., 2013. Diclofenac removal in urine using strong-base anion
exchange polymer resins. Water Res. 47, 6432e6444.
Lu, N.Z., Wardell, S.E., Burnstein, K.L., Defranco, D., Fuller, P.J., Giguere, V.,
Hochberg, R.B., McKay, L., Renoir, J.-M., Weigel, N.L., 2006. International Union
of Pharmacology. LXV. The pharmacology and classification of the nuclear re-
ceptor superfamily: glucocorticoid, mineralocorticoid, progesterone, and
androgen receptors. Pharmacol. Rev. 58, 782e797.
Luo, X., Zheng, Z., Greaves, J., Cooper, W.J., Song, W., 2012. Trimethoprim: kinetic
and mechanistic considerations in photochemical environmental fate and AOP
treatment. Water Res. 46, 1327e1336.
Macikova, P., Groh, K.J., Ammann, A.A., Schirmer, K., Suter, M.J., 2014. Endocrine
disrupting compounds affecting corticosteroid signaling pathways in Czech and
swiss waters: potential impact on fish. Environ. Sci. Technol. 48, 12902.
Ma rtire, D.O., Rosso, J.A., Bertolotti, S.G., Roux, G.A., Braun, A.A., Gonzalez, M.C.,
2001. Kinetic study of the reactions of chlorine atoms and Cl
- 2 radical anions in
aqueous solutions. II. toluene, benzoic acid, and chlorobenzene. J. Phy. Chem. A.
105, 5385e5392.
McNeil, P.L., Nebot, C., Sloman, K.A., 2016. Physiological and behavioral effects of
exposure to environmentally relevant concentrations of prednisolone during
zebrafish (Danio rerio) embryogenesis. Environ. Sci. Technol. 50, 5294e5304.
NIST Standard Reference Database 40, 2016. NDRL/NIST Solutions Kinetics Database.
National Institute of Standards and Technology, Gaithersburg, MD, Version 3.0.
Ohko, Y., Iuchi, K.I., Niwa, C., Tatsuma, T., Nakashima, T., Iguchi, T., Kubota, K.,
Fujishima, A., 2002. 17b-Estradiol degradation by TiO2 photocatalysis as a
means of reducing estrogenic activity. Environ. Sci. Technol. 36, 4175e4181.
Pflug, N.C., Kupsco, A., Kolodziej, E.P., Schlenk, D., Teesch, L.M., Gloer, J.B.,
Cwiertny, D.M., 2017. Formation of bioactive transformation products during
glucocorticoid chlorination. Environ. Sci. Water. Res. Technol. 3, 450e461.
Ricci, A., Fasani, E., Mella, M., Albini, A., 2003. General patterns in the photo-
chemistry of pregna-1, 4-dien-3, 20-diones. J. Org. Chem. 68, 4361e4366.
Schriks, M., Leerdam, J.A.V., Linden, S.C.V.D., Burg, B.V.D., Wezel, A.P.V., Voogt, P.D.,
2010. High-resolution mass spectrometric identification and quantification of
glucocorticoid compounds in various wastewaters in The Netherlands. Environ.
Sci. Technol. 44, 4766e4774.
Stefan, M.I., 2017. Advanced Oxidation Processes for Water Treatment - Funda-
mentals and Applications. IWA Publishing, London, pp. 7e424.
Sun, P., Lee, W.-N., Zhang, R., Huang, C.-H., 2016. Degradation of DEET and caffeine
under UV/chlorine and simulated sunlight/chlorine conditions. Environ. Sci.
Technol. 50, 13265e13273.
Tang, Y., Shi, X., Liu, Y., Feng, L., Zhang, L., 2018. Degradation of clofibric acid in UV/
chlorine disinfection process: kinetics, reactive species contribution and path-
ways. Roy. Soc. Open Sci. 5, 171372.
Wang, W.-L., Wu, Q.-Y., Huang, N., Wang, T., Hu, H.-Y., 2016. Synergistic effect be-
tween UV and chlorine (UV/chlorine) on the degradation of carbamazepine:
influence factors and radical species. Water Res. 98, 190e198.
WHO, 2017. The Selection and Use of Essential Medicines: Report of the WHO
Expert Committee, (Including the 20th WHO Model List of Essential Medicines
and the 6th Model List of Essential Medicines for Children). World Health Or-
ganization, pp. 183e574.
Winker, M., Tettenborn, F., Faika, D., Gulyas, H., Otterpohl, R., 2008. Comparison of
analytical and theoretical pharmaceutical concentrations in human urine in
Germany. Water Res. 42, 3633e3640.
Wols, B., Hofman-Caris, C., Harmsen, D., Beerendonk, E., 2013. Degradation of 40
selected pharmaceuticals by UV/H2O2. Water Res. 47, 5876e5888.
Wols, B., Harmsen, D., Beerendonk, E., Hofman-Caris, C., 2015. Predicting pharma-
ceutical degradation by UV (MP)/H2O2 processes: a kinetic model. Chem. Eng. J.
263, 336e345.
66 K. Yin et al. / Chemosphere 212 (2018) 56e66
Wu, Z., Fang, J., Xiang, Y., Shang, C., Li, X., Meng, F., Yang, X., 2016. Roles of reactive
chlorine species in trimethoprim degradation in the UV/chlorine process: ki-
netics and transformation pathways. Water Res. 104, 272e282.
Xiang, Y., Fang, J., Shang, C., 2016. Kinetics and pathways of ibuprofen degradation
by the UV/chlorine advanced oxidation process. Water Res. 90, 301e308.
Xiao, Y., Zhang, L., Yue, J., Webster, R.D., Lim, T.-T., 2015. Kinetic modeling and en-
ergy efficiency of UV/H2O2 treatment of iodinated trihalomethanes. Water Res.
75, 259e269.
Xiao, Y., Zhang, L., Zhang, W., Lim, K.-Y., Webster, R.D., Lim, T.-T., 2016. Comparative
evaluation of iodoacids removal by UV/persulfate and UV/H2O2 processes.
Water Res. 102, 629e639.
Yang, B., Ying, G.-G., Zhao, J.-L., Liu, S., Zhou, L.-J., Chen, F., 2012. Removal of selected
endocrine disrupting chemicals (EDCs) and pharmaceuticals and personal care
products (PPCPs) during ferrate (VI) treatment of secondary wastewater efflu-
ents. Water Res. 46, 2194e2204.
Yang, Y., Pignatello, J.J., Ma, J., Mitch, W.A., 2014. Comparison of halide impacts on
the efficiency of contaminant degradation by sulfate and hydroxyl radical-based
advanced oxidation processes (AOPs). Environ. Sci. Technol. 48, 2344e2351.
Yin, K., Li, F., Wang, Y., He, Q., Deng, Y., Chen, S., Liu, C., 2017. Oxidative trans-
formation of artificial sweetener acesulfame by permanganate: reaction ki-
netics, transformation products and pathways, and ecotoxicity. J. Hazard Mater.
330, 52e60.
Yin, K., Deng, Y., Liu, C., He, Q., Wei, Y., Chen, S., Liu, T., Luo, S., 2018a. Kinetics,
pathways and toxicity evaluation of neocicotinoid insecticides degradation via
UV/chlorine process. Chem. Eng. J. 346, 298e306.
Yin, K., Deng, L., Luo, J., Crittenden, J., Liu, C., Wei, Y., Wang, L., 2018b. Destruction of
phenicol antibiotics using the UV/H2O2 process: kinetics, byproducts, toxicity
evaluation and trichloromethane formation potential. Chem. Eng. J. 351,
867e877.
Zepp, R.G., Faust, B.C., Hoigne, J., 1992. Hydroxyl radical formation in aqueous re-
actions (pH 3-8) of iron (II) with hydrogen peroxide: the photo-Fenton reaction.
Environ. Sci. Technol. 26, 313e319.
Zhang, R., Sun, P., Boyer, T.H., Zhao, L., Huang, C.-H., 2015. Degradation of phar-
maceuticals and metabolite in synthetic human urine by UV, UV/H2O2, and UV/
PDS. Environ. Sci. Technol. 49, 3056e3066.
Zhang, R., Yang, Y., Huang, C.H., Li, N., Liu, H., Zhao, L., Sun, P., 2016a. UV/H2O2 and
UV/PDS treatment of trimethoprim and sulfamethoxazole in synthetic human
urine: transformation products and toxicity. Environ. Sci. Technol. 50,
2573e2583.
Zhang, R., Yang, Y., Huang, C.-H., Zhao, L., Sun, P., 2016b. Kinetics and modeling of
sulfonamide antibiotic degradation in wastewater and human urine by UV/
H2O2 and UV/PDS. Water Res. 103, 283e292.
Zhang, T., Chen, Y., Leiknes, T., 2016c. Oxidation of refractory benzothiazoles with
PMS/CuFe2O4: kinetics and transformation intermediates. Environ. Sci. Technol.
50, 5864e5873.
Zhang, S., Wang, L., Liu, C., Luo, J., Crittenden, J., Liu, X., Cai, T., Yuan, J., Pei, Y., Liu, Y.,
2017. Photocatalytic wastewater purification with simultaneous hydrogen
production using MoS2 QD-decorated hierarchical assembly of ZnIn2S4 on
reduced graphene oxide photocatalyst. Water Res. 121, 11e19.