Chemical Engineering Journal 335 (2018) 352–361

Contents lists available at ScienceDirect

Chemical Engineering Journal

journal homepage: www.elsevier.com/locate/cej

Comparison of drinking water treatment processes combinations for the T

minimization of subsequent disinfection by-products formation during
chlorination and chloramination
⁎
Jianglin Hu, Wenhai Chu , Minghao Sui, Bin Xu, Naiyun Gao, Shunke Ding
State Key Laboratory of Pollution Control and Resources Reuse, College of Environmental Science and Engineering, Tongji University, Shanghai 200092, China

G RA P H I C A L AB S T R A C T

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

Keywords: Micro-ozone, permanganate, ferrate, and chlorine dioxide were applied as pre-oxidants to meet conventional
Drinking water treatment requirements for water quality and algal control. The formation of typical carbonaceous disinfection by-products
DBPs (C-DBPs) and emerging nitrogenous disinfection by-products (N-DBPs) during both chlorination and chlor-
Ozonation amination were investigated with combinations of different processes: (i) coagulation-sedimentation-filtration
Dichloroacetonitrile
(CSF), (ii) pre-KMnO4/O3/K2FeO4/ClO2 matched with CSF, and (iii) pre-KMnO4/O3/K2FeO4/ClO2 coupled with
Toxicity
CSF and followed by granular activated carbon (GAC)/O3-GAC advanced treatment. All conventional water
quality indexes met the requirements of the national standards for drinking water quality of China (GB5749-
2006) when the system was under stable operation; only NH4+-N, which only met the requirement with either
pre-oxidation or post-ozonation, failed to do so. Micro-ozone pre-treatment coupled with CSF and O3-GAC
showed the best performance in removing chloroform (CF) and dichloracetonitrile (DCAN) precursors. Pre-
oxidation had a negative impact on chloropicrin (TCNM) and chloral hydrate formation, but a positive effect on
DCAN. The trade-off analysis based on the toxicity of the selected DBPs indicated that pre-oxidation (KMnO4, O3,
K2FeO4, and ClO2) had a positive influence on the control of overall cytotoxicity, and O3 pre-oxidation coupled
with CSF followed by O3-GAC performed the best. However, for the control of overall genotoxicity, the ClO2 pre-
treatment showed best results, whether or not GAC or O3-GAC was involved. These results are important for
optimizing the design of the treatment processes in drinking water treatment plants using eutrophic lake water
where precursors for regulated C-DBP and unregulated N-DBP formation are prevalent.

Abbreviations: C-DBPs, carbonaceous disinfection by-products; CF, chloroform; HALs, haloaldehydes; CSF, coagulation-sedimentation-filtration; DBPs, disinfection by-products;
DCAcAm, dichloroacetamide; DOC, dissolved organic carbon; DON, dissolved organic nitrogen; DWTPs, drinking water treatment plants; FPs, formation potentials; HAAs, haloacetic
acids; HAcAms, haloacetamides; N-DBPs, nitrogen-containing disinfection by-products; TCAcAm, trichloroacetamide; THMs, trihalomethanes
⁎
Corresponding author.
E-mail addresses: [email protected], [email protected] (W. Chu).

http://dx.doi.org/10.1016/j.cej.2017.10.144
Received 21 July 2017; Received in revised form 21 September 2017; Accepted 22 October 2017
Available online 23 October 2017
1385-8947/ © 2017 Elsevier B.V. All rights reserved.
J. Hu et al. Chemical Engineering Journal 335 (2018) 352–361

1. Introduction formation of typical C-DBPs and emerging N-DBPs during both chlor-
ination and chloramination was investigated with three different pro-
Disinfection of drinking water has been an indispensable treatment cess combinations: (i) coagulation-sedimentation-filtration (CSF), (ii)
process since the early 20th century as it has reduced the incidence of pre-KMnO4/O3/K2FeO4/ClO2 matched with CSF, and (iii) pre-KMnO4/
waterborne diseases including cholera, typhoid fever, and amoebic O3/K2FeO4/ClO2 coupled with CSF and followed by GAC/O3-GAC ad-
dysentery [1,2]. However, there are numerous organic materials (i.e. vanced treatment. Conventional water quality indexes (turbidity, per-
humus, amino acids, proteins, environmental pollutants) in natural manganate index (CODMn), UV254 and TOC), the formation potentials
water that can react with free chlorine to yield halogenated organics (FPs) of typical DBPs (THMs, CH, HANs, HNMs, and HAcAms) and the
[3,4]. Ever since disinfection by-products (DBPs) were first related to reduction of DBP FPs were all examined.
chronic health concerns, rather than acute effects, in the early 1970s
[5], approximately hundreds of DBPs have been reported in the lit- 2. Materials and methods
erature [4,6–8]. More attention has been focused on nitrogen-con-
taining DBPs (N-DBPs), including haloacetonitriles (HANs), haloni- 2.1. Materials
tromethanes (HNMs), and haloacetamides (HAcAms), because they are
actually more cytotoxic and genotoxic in mammalian cell assays than The pilot experiment was carried out in October 2016. All the raw
the regulated carbonaceous DBPs (C-DBPs), including trihalomethanes water was sampled through the intake of a drinking water plant located
(THMs), haloacetic acids (HAAs), and chloral hydrate (CH) [8–10]. in Hefei Province (China) which draws water from Lake Chao. This lake
When raw water contains bromine or/and iodine, brominated and io- has long been considered one of the five largest freshwater lakes in
dinated analogues with much higher cytotoxicity and genotoxicity than China (with maximum capacity 4.81 billion cubic meters and an
chlorinated DBPs may also occur [3]. average water depth of 3 m), and is a major source of the potable water
Removing precursors of DBPs using pre-treatment or advanced post- used in Hefei Province. However, the lake has been subjected to eu-
treatment technologies before disinfection is a relatively effective way trophication and frequent algal blooms for years, due to the enrichment
to control the formation of DBPs [11,12]. Dissolved organic nitrogen of the nitrogen and phosphorus in the Chao Lake from the residents' raw
(DON) is a major precursor of N-DBPs. Compared to dissolved organic sewage, agricultural non-point source pollution and waste water from
carbon (DOC), the main precursors of C-DBPs [13–15], DON is greatly the factory [40–42]. In order to deal with the water quality problems
impacted by algal blooms and fertilizer runoff [12,16]. A median DON caused by algae, the DWTPs using eutrophic lake water usually add pre-
concentration of 0.37 mg/L of N was reported in surface water and treatment and/or advanced treatment (e.g., O3-GAC) to ensure the
0.24 mg/L of N in shallow groundwater [14]. The DOC concentration is safety of drinking water quality.
usually much higher than DON, and in earlier studies DOC/DON This study measured THMs, CH, HANs, HNMs, and HAMs. DCAcAm
averaged 18 mg DOC per mg DON [14,17]. (98.5%) and TCAcAm (98.5%) standards were purchased from Alfa
Previous studies indicated that Al and Fe salts used as coagulants do Aesar (Karlsruhe, Germany). Free chlorine stock solution was prepared
not efficiently coagulate organic nitrogen compounds or proteins using sodium hypochlorite solution (active chlorine > 5%, Sinopharm
[17–19]. Algal-derived DOC enriched in nitrogen, in particular, is re- Chemical Reagent Co., Ltd., China). All other chemical reagents were of
portedly difficult to be removed during coagulation [20]. Many studies analytical grade at least and were obtained from Sinopharm Chemical
have shown that the use of ozone, permanganate, ferrate or chlorine Reagent Co., Ltd. (Shanghai, China) unless noted. Ultrapure water was
dioxide as pre-oxidants can enhance the removal of algae [21,22]. obtained from a Millipore Milli-Q Gradient water purification system
Apart from its application as a disinfectant, ozone has been used as a (18 MΩ·cm, Billerica, MA, USA).
pre-oxidant, or in advanced treatment, due to its high oxidation po-
tential [23,24]. It can oxidize a wide range of nuisance or potential 2.2. Pilot-plant process flows
toxicants, including substances with natural colour and compounds
associated with unpleasant tastes or odours [23,25–28]. Ozone, in- The pilot plant is located in a typical DWTP near Lake Chao. The
tegrated with granular activated carbon (GAC) filtration, is commonly process flows included pre-treatment, coagulation, sedimentation, sand
used to degrade toxic micro-pollutants, remove precursors of THMs, filtration, post-ozonation, and GAC filtration. Pre-treatment included
and increase bio-degradability in GAC filtration [26,29–31]. Taking the four different oxidants: KMnO4, O3, K2FeO4, and ClO2 pre-oxidation.
cost and practicability into consideration, low doses of ozone (micro- The design flow velocity of the pilot plant used in this study was 1 m3/
ozone) are also used as pre-oxidants. Pre-oxidation with potassium h.
permanganate, despite not being a cure-all technology, has long been
considered an alternative to pre-chlorination, not merely for lowering 2.2.1. Pre-treatment
THM concentrations but also for eliminating compounds that produce The pre-treatment mainly included ozone, potassium permanga-
tastes and odours [32,33]. Ferrate is a strong oxidant; its reduction nate, ferrate, and chlorine dioxide. Ozone is generated by a domestic
potential can reach as high as +2.2 V in an acidic environment, which ozone generator (GuoLin, Qingdao, China). A pre-ozone reaction
makes it even more powerful than ozone and potassium permanganate column (0.16 m [Φ] × 2.75 m [H]) was used to ensure sufficient con-
[34]. Studies have also shown that pre-treatment with potassium ferrate tact and dissolution of ozone in the water. The hydraulic retention time
could enhance coagulation efficiency and increase algae removal (HRT) was 10 min and co-flow ambient of ozone and water was se-
[35–37]. Chlorine dioxide is becoming a more popular agent in lected. Except for ozone, each oxidant (potassium permanganate, fer-
drinking water treatment. As reported before, chlorine dioxide used as rate, or chlorine dioxide) was auto-added from a barrel under control of
pre-oxidant can reduce the formation of THMs and HAAs [38,39]. a metering pump and controlled at about 1 mg/L in the water.
Taking into account the environment impacts, low consumption,
and low operation costs, low doses of these oxidants were re- 2.2.2. Coagulation-sedimentation-filtration (CSF)
commended for pre-oxidation processes in many drinking water treat- With poly aluminium ferric sulphate dosed at 10 mg/L, the coagu-
ment plants (DWTPs) of China, besides the quality of drinking water lation process was achieved in a three-stage series flocculation chamber
must meet the requirements of national standards for drinking water (1.25 m [L] × 0.4 m [W] × 0.6 m [H]). The average velocity gradients
quality of China (GB5749-2006). However, little is known on the for- in the three chambers, which had speed-regulating stirring devices,
mation of typical C-DBPs and N-DBPs during chlorination and chlor- were 50, 20, and 8 s−1, respectively. A sedimentation tank (1.25 m
amination after these pre-treatments and conventional treatments, used [L] × 0.73 m [W] × 1.65 m [H]) with up-flow inclined-tube settler was
alone or followed by O3-GAC. Therefore, in the present study, the in front of the filter. The filtration column (0.4 m [Φ] × 1.7 m [H]) was

353
J. Hu et al. Chemical Engineering Journal 335 (2018) 352–361

filled (bed depth 1.2 m) with quartz sand (grain diameter 0.8–1.2 mm) Error bars in all the figures represent the relative standard deviation of
of which the uniformity coefficient was less than 1.5. The filtration rate the three replicates.
was 8 m/h while the HRTs of the coagulation chamber, sedimentation
tank, and filtration column were 18, 90, and 9 min, respectively.
3. Results and discussion

2.2.3. Post-advanced treatment (O3-GAC) 3.1. Results for turbidity, UV254, DOC, NH4+-N and CODMn
Three stainless steel reaction columns in series were used to achieve
ozonation. The details of the ozonation columns were the as for the pre- The average turbidity reached 55 Nephelometric Turbidity Units,
ozonation columns mentioned before. An ozone generator was used to and the initial UV254 and DOC were about 0.54 cm−1 and 6.0 mg/L,
obtain ozone. Subsequent to ozonation, a granular activated carbon respectively (Table SM2). As shown in Fig. SM1, after the conventional
(GAC) filter was applied (0.4 m [Φ] × 2.5 m [H], bed depth 1.85 m, treatment (coagulation-sedimentation-filtration, namely CSF), the re-
HRT 8 min). Coal GAC was used as filler material and its characteristics moval efficiency reached 98.8%, 85.7%, and 31.2% (sampling 2), re-
are shown in Table SM1. spectively, showing that the treatment combination CSF had a rela-
tively high degree of turbidity removal. When the pre-treatment
2.2.4. Disinfection processes (KMnO4, O3, K2FeO4, and ClO2) were introduced, the tur-
Free chlorine stock solution was prepared using sodium hypo- bidity of the effluent had a slight decline. The removal efficiencies of
chlorite solution (active chlorine > 5%, Sinopharm Chemical Reagent UV254 showed a similar tendency. Ozone reacted easily with olefinic or
Co., Ltd., China). Portions of water samples were chlorinated or carbonyl bonds and destroyed benzene rings; therefore, aromatic or-
chloraminated in 40 mL amber glass bottles at 25 °C, in the dark, for ganics, which belonged to unsaturated organic matter with strong ab-
24 h. Before disinfection, solutions were buffered using sodium dihy- sorption at 254 nm (the fraction indicated by UV254), were removed
drogen phosphate and sodium monohydrogen phosphate to achieve with remarkable efficiency. As shown in Fig. SM1b, the pre-ozonation
reaction conditions at pH 7.5 or 6.5. The DBP FP tests were conducted treatment matched with CSF and O3-GAC process performed its best
following the method developed by Krasner et al. [43], and chlorine/ (removal rate 95.5%). Similarly, ozone pre-oxidation with subsequent
chloramine dosage were determined by both DOC and NH4+. The de- CSF and O3-GAC (pre-O3 + CSF+O3 + GAC) had the highest DOC re-
tails on the DBP FP tests are presented elsewhere [3]. moval efficiency, followed by ferrate pre-treatment matched with CSF
and O3-GAC (pre-K2FeO4 + CSF+O3 + GAC). Ozone facilitated coa-
gulation by increasing the concentration of oxygenated functional
2.3. Sampling and analytical methods
groups or through the bridging reactions [46], and ferrate enhanced
coagulation efficiency by inducing cellular components as coagulate aid
As soon as collected after each process (shown in Fig. 1), all water
and increase particle concentration through precipitant iron species
samples were immediately sent back to the laboratory via refrigerated
[35]. Details on the removal of NH4+-N and the change in the CODMn
transport and refrigerated at 4 °C until use. Some water quality para-
values are available in the supplementary materials (Fig. SM2). CODMn
meters were measured promptly after collection, including turbidity,
of the effluent were all below 2.2 mg/L, reaching the national standards
UV254, CODMn, NH4+-N, pH, and temperature.
for drinking water quality of China (Table SM3). However, the con-
A UV/Vis double beam spectrophotometer (Unico4802, USA) was
centration of NH4+-N was above the required 0.5 mg/L when neither
used to determine the UV254. Turbidity was measured using a portable
pre-oxidation nor post-ozonation was adopted, but met the requirement
turbidity meter (HACH 2100N, USA). The acidic potassium perman-
if either of them was included in the treatment process.
ganate method (GB11892-1989) was applied to CODMn analysis. NH4+-
N was determined using a spectrophotometer (HACH DR6000, USA).
DOC was measured using a TOC analyser (Shimadzu TOC-VCPH, 3.2. Effects on THMs FPs
Japan). Details of the measurement and analysis of turbidity, pH,
temperature, dissolved oxygen, and dissolved organic carbon (DOC) are CF was the dominant species of THMs detected in the water sam-
available elsewhere [44]. ples. As the raw water was not rich in bromide, only a light con-
THMs, including chloroform (CF) and bromodichloromethane centration of BDCM was above the detection limit (Fig. SM3). For some
(BDCM), were analysed using gas chromatography-mass spectrometry reason, conventional treatments (CSF) were not efficient at THM FP
(GC/MS, Shimadzu-QP2010, Japan) with a purge and trap sample reduction. When pre-oxidation was not involved, a CF FP reduction of
concentrator (eclipse4660, OI, USA), based on the United States only 11.1% was achieved (Table SM4). In general, pre-oxidation had a
Environment Protection Agency method 524.2 [45]. HAN and CH positive influence on CF FP reduction (Fig. 2), which is in agreement
analysis were done using another GC/MS (GCMS-QP2020, Shimadzu with results from the previous studies [11,42,47]. When O3, K2FeO4,
Corporation, Japan) with liquid-liquid extraction using methyl tert- and ClO2 were used before the CSF, the formation of CF was enhanced
butyl ether. HAcAms were detected by a gas chromatography with compared with CSF only. Of all the processes, the highest reduction
electron capture detector (Shimadzu-QP2010plus, Japan) and specific efficiency (nearly 56%) of CF FPs resulted from the combination of low-
details about HAcAm detection can be found in a previous study [46]. dose ozone for pre-oxidation, matched with conventional treatment and
Further details of the DBP analyses are presented elsewhere [6,46]. post-ozonation (Pre-O3 + CSF+O3 + GAC, Fig. 2). A previous study

(i) KMnO4 (ii) O3 (iii) K2FeO4 (iv) ClO2

Post- GAC
Coagulation Sand ozonation Filtration
Pre-treatment Sedimentation Filtration Sampling 3
Sampling 1 Sampling 2 GAC
Sampling 4
Filtration
Lake Raw Coagulation Sand
Chaohu Watertank Sedimentation Filtration
Post- GAC
ozonation Filtration

Fig. 1. Flow chart showing processes at the pilot water treatment plant.

354
J. Hu et al. Chemical Engineering Journal 335 (2018) 352–361

Fig. 2. The formation potentials (FPs) of CF of the water samples derived from different pre-oxidation processes. (Pre-KMnO4, Pre-O3, Pre-K2FeO4, and Pre-ClO2 represent pre-oxidation
by KMnO4, O3, K2FeO4, and ClO2, respectively. CSF refers to conventional treatment (coagulation-sedimentation-filtration); pH of all samples was adjusted to 7.5 ± 0.2 to detect CF
using phosphate buffer solution. Reaction time = 24 h).

[11] has shown that the combination of ozonation and biological ac- 3.4. Effects on HANs and TCNM
tivated carbon (BAC) resulted in more efficient (about 89%) reduction
in the formation of CF. Those outcomes demonstrated that microbial As depicted in Fig. 4, TCNM and DCAN were detected. KMnO4, O3,
action (i.e. biodegradation of CF precursors) played an indispensable K2FeO4, and ClO2, as pre-oxidants, have the ability to reduce the for-
role in the O3-BAC process, while the O3-GAC without substantial mation of HANs [11,42,44,49–52]. Fig. 4a shows the DCAN detected in
biological action in the study had only a limited effect on reducing CF water samples derived from different processes. The blank control (CSF)
formation. Overall, when chloramination was adopted for disinfection, process had a lower reduction rate of DCAN formation potential (about
the CF yields decreased sharply, and all pre-oxidation processes, when 11.9%) (Table SM5). Correspondingly, KMnO4, O3, K2FeO4, and ClO2
combined with O3-GAC treatment, achieved a relatively good FP re- pre-treatments with conventional treatment (CSF) achieved reduction
duction (> 50%) (Table SM4). rates of 17.6%, 23.6%, 19.6%, and 14.5%, respectively, and these rose
to 45.3%, 51.1%, 49.3%, and 46.1% when O3-GAC treatment was
3.3. Effects on CH FPs adopted (Pre-oxidation + CSF + O3 + GAC, Fig. 4a). A similar ten-
dency was observed during chloramination. The FP of DCAN during
The third major class of DBPs, haloaldehydes (HALs) , usually reach chlorination was slightly higher than during chloramination. In short,
high concentrations [6]. Chloral hydrate (CH) is the most common form micro-ozone pre-treatment, combined with CSF-O3-GAC processes, was
of HALs and was the dominant species detected. Fig. 3 demonstrates that best for DCAN FP reduction.
the median concentration of CH in the chlorinated raw water reached as Some previous studies indicated that using O3 as pre-oxidant would
high as 19 μg/L in weak acidic environment, after 24 h. Regardless of dramatically increase the formation of TCNM [51,53]. Mitch et al. [54]
chlorination and chloramination, although the CH FPs exhibited a sig- found the increased median TCNM formation FP reached ∼226% and
nificant decline after CSF only, pre-oxidation by KMnO4, O3, K2FeO4, and 133% when pre-ozonation was followed by chlorination and chlor-
ClO2 appeared more or less to promote the formation of CH, of which amination, respectively. As shown in Fig. 4b, when pre-ozonation was
pre-ozonation matched with CSF exacerbated the most (Pre-O3 + CSF, involved in the conventional treatment (Pre-O3 + CSF), there was a
Fig. 3). After the GAC adsorption, the FPs of CH declined. However, when soaring increase in TCNM FP after the treatment, which is even higher
post ozonation was added before GAC, the CH FPs exhibited noticeable than TCNM formation in the raw water. Similarly, ferrate would in-
improvement, especially with the pre-KMnO4 process (pre-KMnO4 + CSF crease the formation of TCNM (as ozone did) during subsequent
+O3 + GAC, Fig. 3). Yang et al. [22] found that when chlorine dioxide chlorination/chloramination [51]. Pre-ClO2 oxidation and subsequent
and ferrate doses were both 1 mg/L, the formation CH was enhanced, but Cl2/NH2Cl increased the FPs of TCNM (Pre-ClO2 + CSF, Fig. 4b).
inverted results were obtained when ferrate dose increased to 20 mg/L. However, chlorine dioxide pre-treatment matched with O3-GAC (Pre-
With ClO2 pre-oxidation, the formation of CH was caused by an inter- ClO2 + CSF + O3 + GAC, Fig. 4b) achieved a decrease in TCNM for-
action between the oxidation effect of ClO2 to produce aldehydes, which mation. Yang et al. [22,39] found that ClO2 pre-treatment of polluted
are precursors of CH, and its destructive effect on hydrophobic organic water achieved mixed results probably due to the difference of water
matter, which is also the main precursor of CH [48]. CH FP during quality. In addition, the nitrogen in TCNM can originate from both
chloramination was much lower than during chlorination, although a organic nitrogen and inorganic nitrogen (NH2Cl). In this study, the
similar tendency in variation was exhibited. concentration of TCNM formed during chlorination and chloramination

355
J. Hu et al. Chemical Engineering Journal 335 (2018) 352–361

Fig. 3. The formation potentials (FPs) of CH of the water samples derived from different pre-oxidation processes. (Pre-KMnO4, Pre-O3, Pre-K2FeO4, and Pre-ClO2 represent pre-oxidation
by KMnO4, O3, K2FeO4, and ClO2, respectively. CSF refers to conventional treatment (coagulation-sedimentation-filtration); pH of all samples was adjusted to 6.5 ± 0.2 for CH detection
using phosphate buffer solution. Reaction time = 24 h).

at almost the same level, which is consistent with the previous studies cytotoxicity index (CTI) and genotoxicity index (GTI), which were tried
[53,54]. for the synthetic toxicity evaluations of a process in connection with
several DBP species, were calculated by making a sum of each value
3.5. Effects on HAMs calculated by multiplying the toxicity value of each DBP by its FP. The
calculation via Eqs. (1) and (2) are shown below. The toxicity values
Dichloroacetamide (DCAcAm) and trichloroacetamide (TCAcAm) include the cytotoxicity value (%C1/2X) and genotoxicity value (Geno-
were detected, while the concentration of monochloroacetamide was toxic potency) of different DBPs, which is put forward by Plewa and
under the detection limit. Previous studies indicated that protein-like other researchers [6,8,58].
organic matter in certain hydrophilic fractions played a critical role in 1
DCAcAm formation [10,55]. The raw water from Chao Lake was rich in CTI = Σ ⎡ × CX ⎤
⎣ %C1/2X ⎦ (1)
algal exudates and protein-like organic materials, which were of par-
ticular importance when pre-oxidation processes were involved. Some
1
pre-oxidation processes can result in the rupture of algal cells and en- GTI = Σ ⎡ × CX ⎤
⎢ Genotoxic potency ⎥ (2)
hanced release of intracellular matter. For example, permanganate pre- ⎣ X ⎦
oxidation cannot result in the algal cell lysis, but this was not the case 1
%C /2X and Genotoxic potencyx Expressed forms of cytotoxicity and
with pre-oxidation using ozone [56,57]. The oxidizability difference genotoxicity, defined elsewhere [8]; CX Formation potential (nM−1) of
probably can explain why pre-treatment with some oxidants increased each DBP; X CF, BDCM, CH, DCAN, TCNM, DCAcAm and TCAcAm.
the formation of DCAcAm immediately and tremendously, but some did As shown in Fig. 6a, the CTI of all the processes (that used chlorine
not. Together with the removal effect of the CSF, DCAcAm FP was re- as disinfectant) were calculated. In general, the pre-oxidation (in-
duced compared to that in raw water (Fig. 5a). Among the four pre- cluding KMnO4, O3, K2FeO4, and ClO2) matched with CSF, all achieved
treatment processes, pre-ClO2 with CSF performed the best. Moreover, CTI reduction to some extent. When GAC was introduced, due to its
when the pre-treatment processes were matched with O3-GAC, in- adsorption of precursors, a general decline of CTI was observed. These
creases were observed in the FPs of DCAcAm and TCAcAm (Fig. 5). As demonstrated that all these processes had a positive effect on risk
for chloramination, because chloramines can react with organic pre- control of overall cytotoxicity. Post-ozonation and GAC processes fur-
cursors to form more DCAcAm, and also can serve as the nitrogen ther contributed to the decline of CTI. Among all the processes in-
source in the formation of DCAcAm, thus a higher DCAcAm level was vestigated, micro-ozone pre-oxidation, coupled with both CSF and O3-
seen here (Fig. 5a). GAC, performed the best. The GTI results are shown in Fig. 6b. The
most striking bar is the pre-ozonation and CSF process, with a GTI much
3.6. Trade-off analysis based on the toxicity and formation of selected DBPs higher than for the raw water. Through data analysis of the computa-
tional procedure (Tables SM6 and SM7), it was found that this phe-
Based on the research and analysis above, the increase and decrease nomenon could mainly be ascribed to TCNM, which played a leading
of different DBPs indicated that a trade-off analysis should be added to role among all DBPs being investigated in this study in the GTI calcu-
provide a comprehensive toxicity assessment based on all investigated lation. Taking all the processes into consideration, pre-ClO2 was the
DBPs formed from the processes mentioned above. Therefore, a method most preferred to control the overall genotoxicity, whether or

356
J. Hu et al. Chemical Engineering Journal 335 (2018) 352–361

Fig. 4. The formation potentials (FPs) of DCAN (a) and TCNM (b) of the water samples derived from different pre-oxidation processes. (Pre-KMnO4, Pre-O3, Pre-K2FeO4, and Pre-ClO2
represent pre-oxidation by KMnO4, O3, K2FeO4, and ClO2, respectively. CSF refers to conventional treatment (coagulation-sedimentation-filtration); pH of all samples was adjusted to
6.5 ± 0.2 using phosphate buffer solution. Reaction time = 24 h).

357
J. Hu et al. Chemical Engineering Journal 335 (2018) 352–361

Fig. 5. The formation potentials (FPs) of DCAcAm (a) and TCAcAm (b) of the water samples derived from different pre-oxidation processes. (Pre-KMnO4, Pre-O3, Pre-K2FeO4, and Pre-
ClO2 represent pre-oxidation by KMnO4, O3, K2FeO4, and ClO2, respectively. CSF refers to conventional treatment (coagulation-sedimentation-filtration); pH of all samples was adjusted
to 6.5 ± 0.2 using phosphate buffer solution. Reaction time = 24 h).

358
J. Hu et al. Chemical Engineering Journal 335 (2018) 352–361

Fig. 6. The cytotoxicity index (a, CTI) and genotoxicity index (b, GTI) of the selected disinfection by-products (DBPs) formed during different processes.

not GAC or O3-GAC was involved. The CTI and GTI results of all pro- 4. Conclusions
cesses using chloramine as disinfectant are also present in Fig. 6. An
outright decline of both cytotoxicity and genotoxicity risk can be seen Application of four different oxidants (KMnO4, O3, K2FeO4, and
during chloramination when compared to chlorination. ClO2) slightly enhanced the reduction of turbidity, with the highest
efficiency achieved with the Pre-K2FeO4 process. UV254 and DOC

359
J. Hu et al. Chemical Engineering Journal 335 (2018) 352–361

showed a similar tendency while micro-ozone pre-treatment with post- [10] M.J. Plewa, M.G. Muellner, S.D. Richardson, F. Fasano, K.M. Buettner, Y.T. Woo,
A.B. McKague, E.D. Wagner, Occurrence, synthesis, and mammalian cell cytotoxi-
ozonation showed the best performance. Among all 13 different pro- city and genotoxicity of haloacetamides: an emerging class of nitrogenous drinking
cesses flows, low-dose ozone for pre-oxidation, matched with conven- water disinfection byproducts, Environ. Sci. Technol. 42 (2008) 955–961.
tional treatment and post-ozonation, was most efficient at lowering the [11] W.H. Chu, N.Y. Gao, Y. Deng, M.R. Templeton, Ozone–biological activated carbon
integrated treatment for removal of precursors of halogenated nitrogenous disin-
CF FP. The investigated O3-GAC without substantial biological action fection by-products, Chemosphere 86 (2012) 1087–1091.
had a limited effect on the reduction of CF, because the removal of CF [12] W.H. Chu, N.Y. Gao, S.W. Krasner, M.R. Templeton, D.Q. Yin, Formation of halo-
precursors by GAC adsorption (instead of biodegradation) is very lim- genated C-, N-DBPs from chlor(am)ination and UV irradiation of tyrosine in
drinking water, Environ. Pollut. 161 (2012) 8–14.
ited. Pre-oxidation had a negative effect on the removal of TCNM and [13] D.A. Reckhow, T.L. Platt, A.L. MacNeill, J.N. McClellan, Formation and degradation
CH precursors. The conventional treatment (CSF) processes had a low of dichloroacetonitrile in drinking waters, J. Water Supply Res. Technol. Aqua 50
reduction of DCAN FP (∼13.9%) When the pre-treatment was asso- (2001) 1–13.
[14] P. Westerhoff, H. Mash, Dissolved organic nitrogen in drinking water supplies: a
ciated with CSF processes, the KMnO4, O3, K2FeO4, and ClO2 pre-
review, J. Water Supply Res. Technol. Aqua 51 (2002) 415–448.
treatments achieved 26.8%, 31.9%, 30.2%, and 24.0% removal effi- [15] P.C. Chiang, E.E. Chang, C.C. Chuang, C.H. Liang, C.P. Huang, Evaluating and
ciency, respectively, which increased to 48.6%, 53.0%, 51.0%, and elucidating the formation of nitrogen-contained disinfection by-products during
48.5% when post-ozonation and GAC were added to the process flows. pre-ozonation and chlorination, Chemosphere 80 (2010) 327–333.
[16] A.D. Shah, W.A. Mitch, Halonitroalkanes, halonitriles, haloamides, and ni-
The DCAcAm FP depended to a large extent on the oxidizability of the trosamines: a critical review of nitrogenous disinfection byproduct formation
different oxidants. pathways, Environ. Sci. Technol. 46 (2012) 119–131.
Through the trade-off analysis of different water treatment pro- [17] W. Lee, P. Westerhoff, Dissolved organic nitrogen removal during water treatment
by aluminum sulfate and cationic polymer coagulation, Water Res. 40 (2006)
cesses based on the toxicity and formation of selected DBPs, it was 3767–3774.
concluded that pre-oxidation had a positive influence on the control of [18] R.A. Vilge, A. Masion, T. Boulangé, D. Rybacki, J.Y. Bottero, Removal of natural
overall cytotoxicity from these investigated DBPs. Among all the pro- organic matter by coagulation-flocculation: a pyrolysis-GC-MS Study, Environ. Sci.
Technol. 33 (1999) 3027–3032.
cesses, micro-ozone pre-oxidation coupled with conventional treat- [19] J. Pietsch, F. Sacher, W. Schmidt, H.J. Brauch, Polar nitrogen compounds and their
ments (CSF) and O3-GAC performed the best in the control of cyto- behaviour in the drinking water treatment process, Water Res. 35 (2001)
toxicity from these investigated DBPs. However, pre-oxidation with 3537–3544.
[20] D.L. Widrig, K.A. Gray, K.S. McAuliffe, Removal of algal-derived organic material
chlorine dioxide, matched with GAC or O3-GAC was the preferred
by preozonation and coagulation: monitoring changes in organic quality by pyr-
choice at controlling the genotoxicity from these investigated DBPs. olysis-GC-MS, Water Res. 30 (1996) 2621–2632.
Moreover, an outright decline of both cytotoxicity and genotoxicity risk [21] J.D. Plummer, J.K. Edzwald, Effects of chlorine and ozone on algal cell properties
and removal of algae by coagulation, J. Water Supply Res. Technol. Aqua 51 (2002)
can be seen during chloramination when compared to chlorination.
307.
Future research would need to evaluate the range of DBPs controlled by [22] X. Yang, W. Guo, X. Zhang, F. Chen, T. Ye, W. Liu, Formation of disinfection by-
these commonly used pre-treatments, conventional treatments and products after pre-oxidation with chlorine dioxide or ferrate, Water Res. 47 (2013)
advanced treatment in a range of water matrices and the energy im- 5856–5864.
[23] Y. Lee, U. von Gunten, Oxidative transformation of micropollutants during muni-
plications for the selection of oxidant doses. cipal wastewater treatment: comparison of kinetic aspects of selective chlorine,
chlorine dioxide, ferrate, and ozone) and non-selective oxidants (hydroxyl radical,
Acknowledgements Water Res. 44 (2010) 555–566.
[24] U. von Gunten, Ozonation of drinking water: Part I Oxidation kinetics and product
formation, Water Res. 37 (2003) 1443–1467.
This work was supported by the National Natural Science [25] W.H. Glaze, Drinking-water treatment with ozone, Environ. Sci. Technol. 21 (1987)
Foundation of China (Nos. 51778445, 51578389), the National Major 224–230.
[26] J. Hollender, S.G. Zimmermann, S. Koepke, M. Krauss, C.S. McArdell, C. Ort,
Science and Technology Project of China (Nos. 2014ZX07405003, H. Siegrist, Elimination of organic micropollutants in a municipal wastewater
2017ZX07201005), the Shanghai Rising-Star Program (No. treatment plant upgraded with a full-scale post-ozonation followed by sand filtra-
17QA1404400) and State Key Laboratory of Pollution Control and tion, Environ. Sci. Technol. 43 (2009) 7862–7869.
[27] C. Gottschalk, J.A. Libra, A. Saupe, Ozonation of water and waste water: a practical
Reuse Foundation (PCRRE16009). guide to understand ozone and its application, Int. J. Environ. Stud. 5 (2010)
795–796.
Appendix A. Supplementary data [28] P.W. Prendiville, Ozonation at the 900 cfs Los Angeles Water Purification Plant,
Ozone Sci. Eng. 8 (1986) 77–93.
[29] T.A. Kurniawan, W.H. Lo, G.Y. Chan, Degradation of recalcitrant compounds from
Supplementary data associated with this article can be found, in the stabilized landfill leachate using a combination of ozone-GAC adsorption treatment,
online version, at http://dx.doi.org/10.1016/j.cej.2017.10.144. J. Hazard. Mater. 137 (2006) 443–455.
[30] A.D. Shah, Z.Q. Liu, E. Salhi, T. Höfer, B. Werschkun, U. Von Gunten, Formation of
disinfection by-products during ballast water treatment with ozone, chlorine, and
References peracetic acid: influence of water quality parameters, Environ Sci: Water Res.
Technol. 1 (2015) 465–480.
[1] C. Anderson, Cholera epidemic traced to risk miscalculation, Nature 354 (1991) [31] E.E. Chang, C.H. Liang, Y.W. Ko, P.C. Chiang, Effect of ozone dosage for removal of
255. model compounds by ozone/GAC treatment, Ozone Sci Eng. 24 (2002) 357–367.
[2] S. Baldursson, P. Karanis, Waterborne transmission of protozoan parasites: review [32] J. Fan, R. Daly, P. Hobson, L. Ho, J. Brookes, Impact of potassium permanganate on
of worldwide outbreaks – an update 2004–2010, Water Res. 45 (2011) 6603–6614. cyanobacterial cell integrity and toxin release and degradation, Chemosphere 92
[3] W.H. Chu, N.Y. Gao, Y. Deng, S.W. Krasner, Precursors of dichloroacetamide, an (2013) 529–534.
emerging nitrogenous DBP formed during chlorination or chloramination, Environ. [33] Y. Jiang, J.E. Goodwill, J.E. Tobiason, D.A. Reckhow, Comparison of the effects of
Sci. Technol. 44 (2010) 3908–3912. ferrate, ozone, and permanganate pre-oxidation on disinfection byproduct forma-
[4] G.H. Hua, D.A. Reckhow, Comparison of disinfection byproduct formation from tion from chlorination, Ferrites and Ferrates: Chemistry and Applications in
chlorine and alternative disinfectants, Water Res. 41 (2007) 1667–1678. Sustainable Energy and Environmental Remediation, American Chemical Society,
[5] J.J. Rook, Formation of haloforms during chlorination of natural waters, Water 2016, pp. 421–437.
Treat. Exam. 23 (1974) 234–239. [34] V.K. Sharma, Potassium ferrate(VI): an environmentally friendly oxidant, Adv.
[6] S.W. Krasner, H.S. Weinberg, S.D. Richardson, S.J. Pastor, R. Chinn, M.J. Sclimenti, Environ. Res. 6 (2002) 143–156.
G.D. Onstad, A.D. Thruston, Occurrence of a new generation of disinfection by- [35] J. Ma, W. Liu, Effectiveness and mechanism of potassium ferrate(VI) preoxidation
products, Environ. Sci. Technol. 40 (2006) 7175–7185. for algae removal by coagulation, Water Res. 36 (2002) 871–878.
[7] I. Voukkali, A.A. Zorpas, Disinfections methods and by-products formation, Desalin. [36] H. Aslani, R. Nabizadeh, S. Nasseri, A. Mesdaghinia, M. Alimohammadi,
Water Treat. 5 (2015) 1150–1161. A.H. Mahvi, N. Rastkari, S. Nazmara, Application of response surface methodology
[8] M.J. Plewa, E.D. Wagner, M.G. Muellner, K.M. Hsu, S.D. Richardson, Comparative for modeling and optimization of trichloroacetic acid and turbidity removal using
mammalian cell toxicity of N-DBPs and C-DBPs, ACS Symposium Series, Oxford potassium ferrate(VI), Desalin. Water Treat. 57 (2016) 25317–25328.
University Press, 2007, pp. 36–50 Chapter 3. [37] J.Q. Jiang, B. Lloyd, Progress in the development and use of ferrate(VI) salt as an
[9] S.D. Richardson, M.J. Plewa, E.D. Wagner, R. Schoeny, D.M. DeMarini, Occurrence, oxidant and coagulant for water and wastewater treatment, Water Res. 36 (2002)
genotoxicity, and carcinogenicity of regulated and emerging disinfection by-pro- 1397–1408.
ducts in drinking water: a review and roadmap for research, Mutat. Res. 636 (2007) [38] K. Linder, J. Lew, B. Carter, R. Brauer, Avoiding chlorite: chlorine and ClO2 together
178–242. form fewer DBPs, Opflow 32 (2006) 24–26.

360
J. Hu et al. Chemical Engineering Journal 335 (2018) 352–361

[39] X. Yang, W. Guo, W. Lee, Formation of disinfection byproducts upon chlorine di- trichloronitromethane and dichloroacetonitrile during chlorination/chloramina-
oxide preoxidation followed by chlorination or chloramination of natural organic tion, Chemosphere 88 (2012) 25–32.
matter, Chemosphere 91 (2013) 1477–1485. [50] B.Y. Song, X. Sun, Y. Jiang, Y. Li, L. Che, Formation of disinfection by-products
[40] G.P. Shang, J.C. Shang, Spatial and temporal variations of eutrophication in western during chlorine dioxide pre-oxidation of chironomid larvae metabolites followed by
Chaohu Lake, China, Environ. Monit. Assess. 130 (2007) 99–109. chlorination, Desalin. Water Treat. 57 (2016) 28698–28704.
[41] L. Yang, K. Lei, W. Yan, Y. Li, Internal loads of nutrients in lake chaohu of china: [51] Y. Jiang, J.E. Goodwill, J.E. Tobiason, D.A. Reckhow, Impacts of ferrate oxidation
implications for lake eutrophication, Int. J. Environ. Res. 7 (2013) 1021–1028. on natural organic matter and disinfection byproduct precursors, Water Res. 96
[42] Y. Zhang, R. Ma, M. Zhang, H. Duan, S. Loiselle, J. Xu, Fourteen-year record (2016) 114–125.
(2000–2013) of the spatial and temporal dynamics of floating algae blooms in Lake [52] J.Q. Jiang, S. Wang, A. Panagoulopoulos, The exploration of potassium ferrate (VI)
Chaohu, observed from time series of MODIS images, Remote Sens. 7 (2015) as a disinfectant/coagulant in water and wastewater treatment, Chemosphere 63
10523–10542. (2006) 212–219.
[43] S.W. Krasner, M.J. Sclimenti, W. Mitch, P. Westerhoff, A. Dotson, Using formation [53] T. Bond, J. Huang, M.R. Templeton, N. Graham, Occurrence and control of ni-
potential tests to elucidate the reactivity of DBP precursors with chlorine versus trogenous disinfection by-products in drinking water – a review, Water Res. 45
with chloramines. In: AWWA Water Quality Technology Conference, 2007, (2011) 4341–4354.
Denver, CO. [54] W.A. Mitch, S.W. Krasner, P. Westerhoff, A. Dotson, Occurrence and Formation of
[44] W.H. Chu, N.Y. Gao, Y. Deng, M.R. Templeton, D.Q. Yin, Formation of nitrogenous Nitrogenous Disinfection By-Products, Water Research Foundation, D., CO, USA,
disinfection by-products from pre-chloramination, Chemosphere 85 (2011) 2009.
1187–1191. [55] H. Huang, B.Y. Chen, Z.R. Zhu, Formation and speciation of haloacetamides and
[45] USEPA, Method 524.2. Measurement of purgeable organic compounds in water by haloacetonitriles for chlorination, chloramination, and chlorination followed by
capillary column gas chromatography/mass spectrometry (Revision). chloramination, Chemosphere 166 (2017) 126–134.
Environmental Protection Agency. 1992, Washington, DC. [56] S. Zhou, Y. Shao, N. Gao, L. Li, J. Deng, M. Zhu, S. Zhu, Effect of chlorine dioxide on
[46] W.H. Chu, N.Y. Gao, D.Q. Yin, S.W. Krasner, W.A. Mitch, Impact of UV/H2O2 Pre- cyanobacterial cell integrity, toxin degradation and disinfection by-product for-
oxidation on the formation of haloacetamides and other nitrogenous disinfection mation, Sci. Total. Environ. 482–483 (2014) 208–213.
byproducts during chlorination, Environ. Sci. Technol. 48 (2014) 12190–12198. [57] P. Xie, J. Ma, J. Fang, Y. Guan, S. Yue, X. Li, L. Chen, Comparison of permanganate
[47] W.H. Chu, X. Li, T. Bond, N.Y. Gao, D.Q. Yin, The formation of haloacetamides and preoxidation and preozonation on algae containing water: cell integrity, char-
other disinfection by-products from non-nitrogenous low-molecular weight organic acteristics, and chlorinated disinfection byproduct formation, Environ. Sci. Technol.
acids during chloramination, Chem. Eng. J. 285 (2016) 164–171. 47 (2013) 14051–14061.
[48] J. Świetlik, E. Sikorska, Application of fluorescence spectroscopy in the studies of [58] M.J. Plewa, E.D. Wagner, Charting a new path to resolve the adverse health effects
natural organic matter fractions reactivity with chlorine dioxide and ozone, Water of DBPs, Chapter 1, in: Tanju Karanfil, Bill Mitch, Paul Westerhoff, Yuefeng Xie
Res. 38 (2004) 3791–3799. (Eds.), Recent Advances in Disinfection By-products, American Chemical Society,
[49] X. Yang, Q. Shen, W. Guo, J. Peng, Y. Liang, Precursors and nitrogen origins of Washington, DC, 2015, pp. 3–23.

361