Rate Constants and Mechanisms of the Reactions of Cl• and Cl2•−

with Trace Organic ContaminantsYu Lei,† Shuangshuang Cheng,† Na Luo,‡ Xin Yang,*,† and Taicheng An‡

†School of Environmental Science and Engineering, Guangdong Provincial Key Laboratory of Environmental Pollution Control andRemediation Technology, Sun Yat-sen University, Guangzhou 510275, China‡Guangzhou Key Laboratory Environmental Catalysis and Pollution Control, School of Environmental Science and Engineering,Institute of Environmental Health and Pollution Control, Guangdong University of Technology, Guangzhou 510006, China

*S Supporting Information

ABSTRACT: Cl• and Cl2•− radicals contribute to the

degradation of trace organic contaminants (TrOCs) such aspharmaceutical and personal care products and endocrine-disrupting chemicals. However, little is known about theirreaction rate constants and mechanisms. In this study, thereaction rate constants of Cl• and Cl2

•− with 88 targetcompounds were determined using laser flash photolysis.Decay kinetics, product buildup kinetics, and competitionkinetics were applied to track the changes in their transientspectra. Cl• exhibited quite high reactivity toward TrOCs withreaction rate constants ranging from 3.10 × 109 to 4.08 × 1010

M−1 s−1. Cl2•− was less reactive but more selective, with

reaction rate constants varying from <1 × 106 to 2.78 × 109

M−1 s−1. Three QSAR models were developed, which were capable of predicting the reaction rate constants of Cl2•− with

TrOCs bearing phenol, alkoxy benzene, and aniline groups. The detection of Cl•-adducts of many TrOCs suggested that Cl•

addition was an important reaction mechanism. Single electron transfer (SET) predominated in reactions of Cl• with TrOCsbearing electron-rich moieties (e.g., sulfonamides), and their cation radicals were observed. Cl• might also abstract hydrogenatoms from phenolic compounds to generate phenoxyl radicals. Moreover, Cl• could react with TrOCs through multiplepathways since more than one transient intermediate was detected simultaneously. SET was the major reaction mechanism ofCl2

•− reactions with TrOCs bearing phenols, alkoxy benzenes, and anilines groups. Cl2•− was found to play an important role in

TrOC degradation, though it has been often neglected in previous studies. The results improve the understanding of halogenradical-involved chemistry in TrOC degradation.

■ INTRODUCTION

Halogen radicals such as chlorine radicals (Cl•) and dichlorineradicals (Cl2

•−) play an important role in photochemicalreactions in sunlit estuarine and coastal waters rich in chlorideions (Cl−) and also in engineered treatment systems (e.g.,chlorine photolysis in drinking water treatment and advancedoxidation processes (AOPs) in brine and wastewater treat-ment).1−5 Halogen radicals participate in the transformation oftrace organic contaminants (TrOCs) such as pharmaceuticaland personal care products (PPCPs) and endocrine-disruptingchemicals (EDCs) and make substantial contributions to theirdegradation.6,7

Cl• and Cl2•− radicals can be formed in environmentally

relevant waters in the reactions such as sulfate radical reactionswith Cl−, free chlorine photolysis, and natural organic matterphotosensitized reactions in the presence of Cl− (eqs 1−3).Cl• (E0 = 2.55 V vs SHE) and Cl2

•− (E0 = 2.13 V vs SHE) arehighly reactive in transforming a variety of TrOCs.8 In thechlorine photolysis under UV irradiation, more than 60% ofbenzoic acid degradation was attributed to Cl•, partially

because of the high reaction rate constant of benzoic acid withCl• (1.8 × 1010 M−1 s−1).9 Cl2

•− reacts more slowly than Cl•

and is often neglected.9−11 However, Cl2•− can become

important in AOP treatment of compounds with high reactionrate constants.12 For example, the two pesticides, methidathionand dimethoate, reacted with Cl2

•− with rate constants of 1.3 ×108 and 1.1 × 108 M−1 s−1, respectively and Cl2

•− contributedmore than 98% of their removal at a Cl− concentrationexceeding 10 mM in a UV/persulfate AOP.12 A recent workalso revealed the important role of Cl• and Cl2

•− indegradation of organic micropollutants in the UV/NH2Clprocess.13 It is therefore important to obtain the reaction rateconstants of Cl• and Cl2

•− with TrOCs to improve theunderstanding of the chlorine radical-involved chemistry.

Received: April 23, 2019Revised: August 3, 2019Accepted: September 4, 2019Published: September 4, 2019

Article

pubs.acs.org/estCite This: Environ. Sci. Technol. 2019, 53, 11170−11182

© 2019 American Chemical Society 11170 DOI: 10.1021/acs.est.9b02462Environ. Sci. Technol. 2019, 53, 11170−11182

Dow

nloa

ded

via

GU

AN

GD

ON

G U

NIV

OF

TE

CH

NO

LO

GY

on

Oct

ober

9, 2

019

at 0

3:20

:01

(UT

C).

See

http

s://p

ubs.

acs.

org/

shar

ingg

uide

lines

for

opt

ions

on

how

to le

gitim

atel

y sh

are

publ

ishe

d ar

ticle

s.

+ → + = ו− − − • − −kSO Cl SO Cl 3.1 10 M s4 42

18 1 1

(1)

+ → +• •hvHOCl HO Cl (2)

+ = × = ו − •− − −−

−F k kCl Cl Cl 8.5 10 M s 6.0 10 s2 39 1 1

34 1

(3)

* + → +− − •− • •−NOM Cl /2Cl NOM Cl /Cl32 (4)

Despite the recognition of the important role of chlorine-containing radicals, the reaction rate constants of Cl• and Cl2

•−

with TrOCs remain largely unknown, which hinders under-standing the fate of TrOCs. Halogen radicals play an importantrole in the photo-oxidation of algal toxins (e.g., microsystin-LRand domoic acid) in sunlit seawater.3,14 In UV/chlorine AOPs,the degradation of TrOCs was also largely attributed to Cl•

and Cl2•− and ClO•, but their contributions were indirectly

obtained by subtracting the contribution of UV irradiation andoxidation by free chlorine and HO• from the overalldegradation observed.15,16 The contributions of Cl• andCl2

•− were unable to be directly calculated due to the missingreaction rate constants. Kinetic models predicting TrOCsdegradation rates generally underestimate the observed valuesunless second-order reaction rate constants of TrOCs withhalogen radicals are included.16,17 Although the reaction rateconstants of Cl• and Cl2

•− with alcohols, carboxylic acids,alkanes, and aromatic compounds have been reported,18−21 thedata of reaction rate constants with TrOCs are very limited.Cl• and Cl2

•− can react with organic compounds via singleelectron transfer (SET), H-abstraction, and addition pathways.The current understanding of the Cl• reaction mechanism wasmainly concentrated on saturated aliphatic compounds andbenzenes.18,20,22 Addition was suggested as the majormechanism of reactions of Cl• with substituted benzenes,20,22

and both SET and H-abstraction were found to contribute theCl• reactions with oxygenated hydrocarbons using electronspin-resonance spectroscopy or laser flash photolysis techni-que.18,23,24 A recent study using quantum mechanicalcalculations suggested that H-abstraction and Cl• additionwere the major Cl• reaction mechanisms from the observationof the linear correlations between the free energies ofactivation and 31 Cl• reaction rate constants with organiccompounds, mainly saturated aliphatic compounds.25 Cl2

•−

was found to react via H-abstraction for aliphatic compounds,addition for olefins, and SET for aromatics bearing electron-donating groups.1,19 These previous studies suggested thestructure-dependent reaction mechanisms for Cl• and Cl2

•−

reactions. TrOCs generally have complicated and diversestructures, and their reaction mechanisms with Cl• and Cl2

•−

are not well understood yet. Tracking the formation oftransient intermediates using time-resolved pulse radiolysis orlaser flash photolysis can be a powerful approach to explore theinherent reaction mechanisms.26−28 Meanwhile, quantitativestructure activity relationship (QSAR) approaches have beenused to determine the reaction rate constants (e.g., with HOCl,O3, and ClO2) based on the molecular features (e.g., Hammettconstants) of various compounds when their reactionsproceeded via the similar reaction pathways.29 Thus, develop-ment of QSAR models can also assist in understanding thereaction mechanisms.As such, the objective of this study was to obtain the

reaction rate constants and explore the reaction mechanisms ofCl• and Cl2

•− with TrOCs, primarily PPCPs and EDCs due to

their wide occurrence. Laser flash photolysis was applied totrack the decay of Cl• and Cl2

•− and the formation of transientintermediates. The reaction rate constants of TrOCs with Cl•

and Cl2•− were evaluated using three kinetic methods,

including the radical decay kinetics, product buildup kinetics,and competition kinetics. The reaction rate constants werestatistically analyzed to better understand the role of functionalgroups in influencing the reactivity of organics toward Cl• andCl2

•−. QSAR models were developed for phenolic compounds,anilines, and alkoxy benzene derivates. The reaction mecha-nisms of TrOCs with Cl• and Cl2

•− were proposed based onthe observation of generated transient intermediates. More-over, the reaction rate constants of Cl• and Cl2

•− werecompared with those of HO•, highlighting their important rolein AOPs.

■ MATERIALS AND METHODS

Chemicals. A total of 88 target compounds are involved inthis study. Their physicochemical properties and sources arelisted in Table S1 of the Supporting Information. The targetcompounds included 68 TrOCs and 20 model compounds.TrOCs investigated in this study included six antipyretic

analgesics, six β-lactams ( and ),

six sulfonamides ( ), four fluoroquino-

lones ( ), four nitroimidazoles ( ),

four tetracyclines ( ), four macrolides,

three β-blockers, three lipid regulators, three xanthines

( ), two H2 antagonists and fourteen other

PPCPs, three phthalate esters ( ), three estrogens

( ), and three other EDCs. Chloroace-

tone (≥98%) was obtained from Adamas Reagent, Ltd.(China). Na2S2O8 (≥99.99%), NaCl (≥99.99%), andpotassium thiocyanate (KSCN, ≥99.99%) were purchasedfrom Aladdin Bio-Chem Technology (China). NaOH (2 M inwater) was purchased from Tokyo Chemical Industry Co., Ltd(Japan). All of the solutions were prepared in ultrapure water(≥18.2 MΩ). The pH of the solutions was adjusted to 7.0 ±0.3 by adding NaOH.

Laser Flash Photolysis Experiments. The second-orderrate constants of reactions of TrOCs with Cl• and Cl2

•− weremeasured using a laser flash photolysis system. All experimentswere conducted at room temperature (25 ± 2 °C). Duplicate

Environmental Science & Technology Article

DOI: 10.1021/acs.est.9b02462Environ. Sci. Technol. 2019, 53, 11170−11182

11171

Figure 1. Decay kinetics of Cl•: (a) transient decay traces of Cl• at 320 nm with different concentrations of roxithromycin, (b) plot of the first-order decay rate constant kobs vs roxithromycin concentration. Product buildup kinetics of Cl•: (c) typical growth traces at 420 nm for the productgenerated from bisphenol A reacting with Cl•, (d) first-order buildup rate constant kobs vs bisphenol A concentration. Thiocyanate anion (SCN−)competition kinetics of Cl•: (e) (SCN)2

•− formation kinetics at 472 nm, (f) competition kinetics plot for Cl• reacting with amoxicillin using SCN−

as a reference compound; the slope represents the second-order rate constant of reaction of Cl• with amoxicillin. Decay kinetics of Cl2•−: (g)

transient decay traces of Cl2•− at 340 nm with different concentrations of metoprolol, (h) plot of the first-order decay rate constant kobs vs

metoprolol concentration.

Environmental Science & Technology Article

DOI: 10.1021/acs.est.9b02462Environ. Sci. Technol. 2019, 53, 11170−11182

11172

Table 1. Rate Constants Determined for the Reactions of TrOCs and Model Compounds with Cl• and Cl2•− at pH 7e,f,g

kCl• (1010 M−1 s−1) kCl2•− (108 M−1 s−1) kHO• (10

9 M−1 s−1)

no. compound C.k.a D.k.b ref ref

PPCPsAntipyretic Analgesics

1 acetaminophen 1.33 ± 0.19 4.32 ± 0.39 - 1.711

1.24 ± 0.26c

2 aspirin 0.68 ± 0.14 0.23 ± 0.04 - -3 diclofenac 3.77 ± 0.65 11.54 ± 0.52 - 8.38 ± 1.2450

4 ibuprofen 2.77 ± 0.35 <0.05 - 7.04 ± 0.5250

5 indomethacin - 4.99 ± 0.51 - 6.751

6 naproxen 2.01 ± 0.15 6.57 ± 0.43 - 8.6150

β-Blockers7 atenolol 2.29 ± 0.23 4.11 ± 0.24 7.05 ± 0.2727

8 metoprolol 1.71 ± 0.31 5.07 ± 0.38 2.2 ± 0.428 8.39 ± 0.0627

9 propranolol - 17.81 ± 0.60 19 ± 128 10.7 ± 0.227

β-Lactams10 amoxicillin 1.27 ± 0.08 4.20 ± 0.07 - 6.94 ± 0.4452

0.79 ± 0.06c

11 cefotaxime 2.30 ± 0.12 4.91 ± 0.16 - -12 cephalexin 2.17 ± 0.25 5.06 ± 0.29 - 8.5 ± 0.741

13 cefaclor 1.59 ± 0.20 3.68 ± 0.09 - 6.00 ± 0.1353

14 penicillin G 1.25 ± 0.09 3.30 ± 0.30 - 7.97 ± 0.1152

15 penicillin V 1.31 ± 0.09 3.36 ± 0.41 - 8.76 ± 0.2852

Fluoroquinolones16 ciprofloxacin 1.39 ± 0.35 2.19 ± 0.08 <0.528 5.94 ± 1.7250

17 enrofloxacin 1.53 ± 0.40 3.27 ± 0.15 - 7.95 ± 0.2354

18 flumequine 0.77 ± 0.23 0.94 ± 0.07 - 8.26 ± 0.2854

19 ofloxacin 1.54 ± 0.25 3.48 ± 0.39 - 4.2 ± 0.555

Macrolides20 azithromycin 0.78 ± 0.04 <0.05 2.9 ± 0.641

0.83 ± 0.03b

21 erythromycin 0.68 ± 0.03 <0.05 3.3315

0.70 ± 0.03b

22 roxithromycin 0.72 ± 0.07b <0.05 5.4 ± 0.341

23 tylosin - 0.46 ± 0.03 8.2 ± 0.141

Nitroimidazoles24 dimetridazole 0.42 ± 0.03 0.84 ± 0.05 3.6517

25 metronidazole 0.31 ± 0.05 1.24 ± 0.08 3.5417

26 ornidazole - 0.93 ± 0.05 2.6617

27 ronidazole - 1.55 ± 0.08 -Sulfonamides

28 sulfanilamide 3.12 ± 0.40 4.32 ± 0.12 - 8.256

29 sulfadimethoxine 4.08 ± 0.24 4.46 ± 0.50 - -30 sulfadiazine 3.35 ± 0.22 4.27 ± 0.37 - 4.5 ± 1.1350

31 sulfamethazine 3.21 ± 0.11 4.85 ± 0.28 - 8.3 ± 0.856

32 sulfamethoxazole 3.64 ± 0.21 4.72 ± 0.39 - 8.5 ± 0.356

3.4 ± 0.413

33 sulfathiazole 3.78 ± 0.49 5.08 ± 0.36 - 7.9 ± 0.456

Tetracyclines34 chlortetracycline - 8.50 ± 0.49 - -35 doxycycline - 11.35 ± 0.69 - 7.7450

36 oxytetracycline 2.36 ± 0.56 9.36 ± 0.29 - 6.9650

37 tetracycline 1.98 ± 0.42 11.80 ± 0.79 - 7.7050

Lipid Regulators38 bezafibrate 1.04 ± 0.09 (3.24) - 8.00 ± 0.2257

39 clofibric acid 0.55 ± 0.13 (1.41) - 6.98 ± 0.1257

40 gemfibrozil 2.14 ± 0.17 2.87 ± 0.12 - 10.0 ± 0.657

Xanthines41 caffeine 3.87 ± 0.35 9.28 ± 0.52 - 6.4 ± 0.7150

42 theophylline 3.98 ± 0.42 8.78 ± 0.34 - 8.22 ± 0.0358

43 xanthine 3.81 ± 0.40 1.85 ± 0.11 - 5.242

Environmental Science & Technology Article

DOI: 10.1021/acs.est.9b02462Environ. Sci. Technol. 2019, 53, 11170−11182

11173

Table 1. continued

kCl• (1010 M−1 s−1) kCl2•− (108 M−1 s−1) kHO• (10

9 M−1 s−1)

no. compound C.k.a D.k.b ref ref

H2 Antagonists44 cimetidine 0.43 ± 0.11 27.78 ± 1.64 30 ± 228 6.559

45 famotidine 1.72 ± 0.26 16.52 ± 0.43 - 14.6 ± 2.560

Other PPCPs46 carbamazepine 3.30 ± 0.26c 0.43 ± 0.03 <0.528 8.8 ± 1.217

5.6 ± 1.661

3.7 ± 0.313

47 clenbuterol - 5.54 ± 0.14 - 9.4815

48 iopromide 2.75 ± 0.39 20.54 ± 0.96 - 3.350

49 mabuterol - 3.48 ± 0.27 - -50 mesalazine 2.23 ± 0.25 14.24 ± 0.28 - 6.762

51 metformin - 0.21 ± 0.01 - -52 primidone 0.62 ± 0.10 1.58 ± 0.02 - 6.6315

53 salbutamol - 3.02 ± 0.20 - 6.1815

54 salicylic acid - 2.10 ± 0.22 - 1742

55 sucralose 1.11 ± 0.16 <0.01 - 1.663

56 terbutaline - 12.05 ± 0.72 - 6.87 ± 0.4315

57 triclosan 2.76 ± 0.44 2.48 ± 0.14 - 4.4364

58 trimethoprim 2.11 ± 0.12 18.78 ± 0.23 24 ± 328 6.3 ± 0.8550

1.69 ± 0.63c

59 venlafaxine - 3.58 ± 0.14 - 8.815

EDCsPhthalate Esters

60 dimethyl phthalate 1.81 ± 0.18 0.14 ± 0.03 - 3.265

61 diethyl phthalate 1.97 ± 0.14 0.11 ± 0.02 - 4.266

62 dibutyl phthalate 1.96 ± 0.22 0.11 ± 0.02 - 4.767

Estrogens63 estrone (E1) 2.06 ± 0.21 3.66 ± 0.24 - 2664

64 estradiol (E2) 2.01 ± 0.30 (3.96) - 14.150

0.8 ± 0.0213

65 ethinyl estradiol (EE2) 2.56 ± 0.11 (3.96) - 10.3 ± 0.750

0.21 ± 0.0213

Other EDCs66 bisphenol A 1.82 ± 0.23 5.82 ± 0.62 - 8.7715

1.45 ± 0.08c

67 methylparaben 1.52 ± 0.13 1.61 ± 0.06 - 5.0168

68 nonylphenol 1.00 ± 0.07 (3.65) - 11 ± 269

Model Compounds69 4-chloroaniline 2.17 ± 0.14 5.23 ± 0.23 - -70 4-methylcatechol 2.49 ± 0.14 11.84 ± 0.33 - -71 4-nitroanisole - 0.25 ± 0.03 - -72 6-aminopenicillanic 0.34 ± 0.03 3.27 ± 0.31 - 2.4 ± 0.0552

73 7-aminocephalosporanic 1.14 ± 0.07 2.29 ± 0.11 - -74 acetylacetone 0.29 ± 0.03 1.42 ± 0.08 - 9.942

75 aniline 2.74 ± 0.31 6.79 ± 0.45 - 1742

76 anisole - 1.62 ± 0.09 - 5.442

77 benzoic acid 1.35 ± 0.15 0.02 ± 0.003 - 4.342

1.80 ± 0.322

78 gallic acid 1.83 ± 0.27 7.53 ± 0.43 - -79 imidazole - 1.82 ± 0.17 - 5.242

80 phenol 1.12 ± 0.09 2.20 ± 0.12 2.519 6.642

0.90 ± 0.12c

81 protocatechuic acid - 5.90 ± 0.31 - -82 p-toluidine 2.73 ± 0.56 9.47 ± 0.52 - -83 pyrimidine 0.05 ± 0.01b <0.01 - 0.1642

84 pyrocatechol 2.82 ± 0.33 5.66 ± 0.41 - 1142

85 TMBd (1,3,5-trimethoxybenzene) 1.33 ± 0.08 26.06 ± 1.72 - 8.147

0.83 ± 0.18c 28.71 ± 2.55c

86 2,4-dimethoxypyrimidine - 0.30 ± 0.02 - -

Environmental Science & Technology Article

DOI: 10.1021/acs.est.9b02462Environ. Sci. Technol. 2019, 53, 11170−11182

11174

or triplicate tests were performed, and the error range of rateconstants was given at 95% confidential intervals (details inText S1).Determining the Second-Order Rate Constants of Cl•.

Laser flash photolysis of chloroacetone represents a suitablemethod to generate Cl• for its kinetic investigation becauseCl2

•− is not present at the beginning of the reaction (eq 5).18,21

+ → +• •hvCH COCH Cl CH COCH Cl3 2 3 2 (5)

The Cl• absorption band ranges from 280 to about 380 nmwith a maximum at 320 nm (ε320 nm ≈ 4500 M−1 cm−1, FigureS2).21 The potential effects of CH3COCH2

• on the spectra andkinetics could be excluded since it exhibits only weakabsorption at around 295 nm and has very low reactivity (k< 5 × 107 M−1 s−1 for TrOCs, details in Text S2 and FigureS3).18,30 Three methods were applied to determine thereaction rate constants of Cl• with the target compounds(T), including radical decay kinetics (monitoring the Cl• decayrate at 320 nm) (Figure 1a,b), product buildup kinetics (Figure1c,d), and competition kinetics using SCN− as a referencecompound (Figure 1e,f). Details are shown in Text S3.Determining the Second-Order Rate Constants of

Cl2•−. Cl2

•− was generated by reacting Cl− with SO4•−, which

was generated from the photolysis of S2O82− (eqs 1, 3, and

6).31,32 A high chloride concentration (0.5 M) was used tominimize the competition of the reactants for SO4

•− and Cl•.Cl2

•− exhibited maximum absorption at 340 nm (ε340nm = 8800M−1 cm−1, Figure S4).19,31

+ →− •−hvS O 2SO2 82

4 (6)

+ → + = ו− − − • − −kSO Cl SO Cl 3.1 10 M s4 42

18 1 1

(1)

+ = × = ו − •− − −−

−F k kCl Cl Cl 8.5 10 M s 6.0 10 s2 39 1 1

34 1

(3)

The reaction rate constants of Cl2•− with all TrOCs were

measured by monitoring the decay trace of Cl2•− at 340 nm

(Figure 1g,h). Further details are shown in Text S4 and TextS5. Due to the low solubility or low reactivity of some targetcompounds, only upper limits of rate constants for them arereported (e.g., <5 × 106 M−1 s−1 for erythromycin). Ionicstrength corrections to the measured reaction rate constants ofCl2

•− were applied, as shown in Text S6.QSAR Analysis. Quantitative structure−activity relation-

ship models were quantified to explore the relationshipbetween the structures of aromatic TrOCs (phenols, alkoxybenzenes, and anilines) and their reactivity toward Cl2

•−.Hammett σ+ constants were used as the substituent descriptorsin this study because of their wide usage, easy accessibility, andgood applicability.29,33 Furthermore, some mechanistic in-

formation could be obtained from the Hammett-constant-based QSAR (details in Text S7).

Modeling Prediction of Elimination of TrOCs duringWater Treatment. Based on the modeled steady-stateconcentrations of radicals and the reaction rate constantsobtained in this study, a UV/chlorine process was evaluatedand the degradation rates of TrOCs attributed by HO•, Cl•,and Cl2

•− were estimated. The details are provided in Text S8.

■ RESULTS AND DISCUSSIONValidation of the Method. The reaction rate constants

obtained from this study were compared with the limited dataavailable in the literature using different analytical methods,and good consistency was achieved. For example, the reactionrate constant of methanol with Cl• was determined as (9.0 ±0.5) × 108 M−1 s−1 in this study using the decay kineticsmethod (Figure S5), which is consistent with 1.0 × 109 M−1

s−1 reported in water.18,21 The reaction rate constant of SCN−

with Cl• was determined to be (5.3 ± 0.2) × 109 M−1 s−1 inthis study (Figure S6), which is consistent with the value of(5.3 ± 0.3) × 109 M−1 s−1 reported by Buxton.21 The rateconstants of Cl• with water and chloroacetone were alsoverified as depicted in Text S9. The rate constant of Cl• withbenzoic acid was determined as (1.35 ± 0.15) × 1010 M−1 s−1

using competition kinetics. Martire’s group has reported thatCl• reacts with benzoic acid, chlorobenzene, and toluene with acommon rate constant of (1.8 ± 0.3) × 1010 M−1 s−1,22 whichis slightly higher than the one obtained in this work. Thedifference may be due to the different reaction pH values: pH2−3 in their work and pH 7 in this study. The Cl2

•− reactionrate constants are also in agreement with those of previousstudies. For example, Cl2

•− reacted with phenol at (2.2 ± 0.12)× 108 M−1 s−1 as determined in this study, close to thereported values of (2.5−3.2) × 108 M−1 s−1 in aqueoussolution.19,34 The rate constants of Cl2

•− with propranolol andtrimethoprim were determined as (1.78 ± 0.60) × 109 and(1.88 ± 0.23) × 109 M−1 s−1, respectively, in agreement withthe values of (1.9 ± 0.1) × 109 and (2.4 ± 0.3) × 109 M−1 s−1

determined by Hoffmann’s group (Table 1).28

As there are three methods for determining the reaction rateconstants of Cl• with target compounds, the rate constants forthe same compound determined by different methods werealso compared. Rate constants determined using decay kineticsagree very well with the data from competition kinetics, withina factor of ±20% (e.g., azithromycin and erythromycin), asshown in Table 1. Rate constants determined using productbuildup kinetics are generally lower than those fromcompetition kinetics; some differences would even reach afactor of 2. For example, reaction rate constants of Cl• withbisphenol A were determined to be (1.45 ± 0.08) × 1010 M−1

s−1 using product buildup kinetics, accounting for 80% of thatobtained using competition kinetics [(1.82 ± 0.23) × 1010 M−1

Table 1. continued

kCl• (1010 M−1 s−1) kCl2•− (108 M−1 s−1) kHO• (10

9 M−1 s−1)

no. compound C.k.a D.k.b ref ref

Model Compounds87 4,6-dimethyl pyrimidine - <0.05 - -88 thiazole - 0.39 ± 0.02 - -

aData determined using competition kinetics (C.k.). bData determined using decay kinetics (D.k.). cData determined using the product buildupkinetics method (P.b.k.). dData of TMB were all determined at pH 3.0. eref: previously published data with reference. f(): The data in parentheseswere predicted from a QSAR model. g-: Not available.

Environmental Science & Technology Article

DOI: 10.1021/acs.est.9b02462Environ. Sci. Technol. 2019, 53, 11170−11182

11175

s−1]. The reason may lie in that more than one pathway getsinvolved in the reaction of Cl• with T at the same time and theproduct tracked reflects only part of the reactions.35,36

Second-Order Rate Constants of Reactions of TrOCswith Cl• and Cl2

•−. The second-order rate constants forreactions of TrOCs with Cl• and Cl2

•− together with themethod used are summarized in Table 1. The rate constantsfor reactions of Cl• with 54 TrOCs and reactions of Cl2

•− with68 TrOCs are presented. For some TrOCs, only their salts canbe purchased (e.g., doxycycline hydrochloride), so theirreaction rate constants with Cl• could not be determineddue to the generation of Cl2

•− from Cl• reacting with Cl− inthe chemicals. The available rate constants for reactions of Cl•,Cl2

•−, and HO• from the literature together with those for 20model compounds are also listed in Table 1 for comparison.Cl•: The reaction rate constants of Cl• with TrOCs vary

from 3.10 × 109 to 4.08 × 1010 M−1 s−1 with an average of 1.98× 1010 M−1 s−1 for the compounds tested. The Cl• reactionrate constants are high and many exceed the rate constants ofHO•. Literature reports also show that Cl• has high reactivitytoward many substances such as alcohols (6.5 × 108 to 6.0 ×109 M−1 s−1)18 and benzenes (6 × 109 to 1.8 × 1010 M−1 s−1)in water.20,22

Reaction rate constants of Cl• are classified into groups(Figure 2a). Sulfonamides exhibit the highest reactivity towardCl• with rate constants varying from 3.12 × 1010 to 4.08 × 1010

M−1 s−1. Estrogens (2.01 × 1010 to 2.56 × 1010 M−1 s−1) andphthalate esters (1.81 × 1010 to 1.97 × 1010 M−1 s−1) also reactwith Cl• quickly. In contrast, the rate constants of nitro-imidazoles (3.1 × 109 to 4.2 × 109 M−1 s−1) and macrolides

(6.8 × 109 to 8.3 × 109 M−1 s−1) are significantly lower thanthose of the other groups of TrOCs shown in Figure 2a.Although TrOCs in the same group tend to have similar rateconstants in most cases, substituents on the core structure canaffect the reaction rate constants greatly. Using lipid regulatorsas examples, gemfibrozil has much higher reaction rateconstants with Cl• (2.14 × 1010 M−1 s−1) than clofibric acid(5.5 × 109 M−1 s−1). This may result from the electron-donating methyl group on the alkoxy benzene ring ofgemfibrozil and the electron-withdrawing chlorine atom onclofibric acid.Some of the measured rate constants approach and even

exceed the diffusion-control limits (∼2 × 1010 M−1 s−1),including those of the sulfonamides (3.12 × 1010 to 4.08 ×1010 M−1 s−1), xanthines (3.81 × 1010 to 3.98 × 1010 M−1 s−1),diclofenac (3.77 × 1010 M−1 s−1), and others. The rateconstants reported in the literature for substituted benzenes(1.8 × 1010 M−1 s−1) in aqueous solution also approach thediffusion-control limits.22 The rate constant of Cl• with anilinein dichloromethane also had an unexpectedly high value of 4.0× 1010 M−1 s−1.37 One possible explanation is that Cl• couldreact with target compounds via single electron transfer (SET),by which two molecules do not necessarily need to diffuse andencounter in the solvent cage. Further proof is described laterin the discussion of Cl• reaction mechanisms. Anotherexplanation is that if diverse reactive sites are involved, thesum of them may exceed the diffusion-controlled rateconstant.38 This phenomenon is especially noticeable forTrOCs with complex structures. For instance, the reaction ofCl• with oxytetracycline may involve reacting at the phenolmoiety (1.12 × 1010 M−1 s−1), the β-diketone group (2.9 × 109

M−1 s−1), the tricarbonyl group (rate constant unavailable),and the amino group (rate constant unavailable).39,40 Theoverall reaction rate constant (2.36 × 1010 M−1 s−1) may thusexceed the diffusion-controlled rate constant at a single site(Table S5).Cl2

•−: In contrast to the high reactivity of Cl•, the reactivityof Cl2

•− toward TrOCs is much lower. The reaction rateconstants of Cl2

•− with TrOCs vary between <1 × 106 and2.78 × 109 M−1 s−1 with an average of 5.08 × 108 M−1 s−1. Thevalues are all below the diffusion-control limit, indicating thatthe reactions of Cl2

•− with TrOCs are activation-controlled.The rate constants of reactions of Cl2

•− with TrOCs spanseveral orders of magnitude. The reaction rate constants forsome macrolides are below the detection limit (5 × 106 M−1

s−1), whereas they exceed 109 M−1 s−1 for tetracyclines and H2antagonists. Cl2

•− is thus a more selective oxidant thanCl•.12,19,26

Reaction rate constants of Cl2•− with different groups of

TrOCs are shown in Figure 2b. The rate constants dependstrongly on the functional groups involved and the corestructures. The difference among various groups is much moreobvious than with Cl•. Among the 12 groups of TrOCs shownin Figure 2b, H2 antagonists show the highest reactivity towardCl2

•− (1.65 × 109 to 2.78 × 109 M−1 s−1), along with the β-blockers (4.11 × 108 to 1.78 × 109 M−1 s−1) and tetracyclines(8.5 × 109 to 1.18 × 109 M−1 s−1). Electron-rich structuressuch as 5-member heterocycles, alkoxy benzene rings, andphenolic rings seem to be responsible for the relatively highreactivity of these compounds. The TrOCs demonstratingreaction rate constants exceeding 1 × 109 M−1 s−1 includedoxycycline, diclofenac, tetracycline, terbutaline, mesalazine,famotidine, propranolol, trimethoprim, iopromide, and cime-

Figure 2. (a) Second-order rate constants of reactions of Cl• withvarious classes of TrOCs. (b) Grouped rate constants for reactions ofCl2

•− with TrOCs.

Environmental Science & Technology Article

DOI: 10.1021/acs.est.9b02462Environ. Sci. Technol. 2019, 53, 11170−11182

11176

tidine. TrOCs with an electron-withdrawing ester group on thearomatic ring or those with no aromatic ring in their structureexhibit very low reactivity toward Cl2

•−. The phthalate esters(1.1 × 107 to 1.4 × 107 M−1 s−1) and macrolides (<5 × 106 to4.6 × 107 M−1 s−1) are examples. The reaction rate constantsof Cl2

•− also vary in the same group of TrOCs. For example,among the four macrolides, the reactivity of tylosin towardCl2

•− (4.6 × 107 M−1 s−1) is significantly higher than that ofazithromycin, erythromycin, or roxithromycin (all <5 × 106

M−1 s−1). This can be attributed to the high reactivity of theconjugated diene moiety in tylosin’s macrolactone ring.41

QSAR Analysis. No linear correlation was observedbetween Hammett σ+ constants and the reaction rate constantsof aromatic TrOCs with Cl•. For example, toluene,chlorobenzene, and phthalate esters have σ+ values of −0.31,0.11, and 0.48, respectively, but their reaction rate constantswith Cl• are all in the range 1.8−2.0 × 1010 M−1 s−1.22,33

TrOCs that show relatively much lower reactivity toward Cl•

(kCl• < 6 × 109 M−1 s−1) are typically small sized (<200 Da)and contain an electron-poor saturated carbon skeleton or anelectron-withdrawing group. Chloroform (kCl• = 1.2 × 107 M−1

s−1) and methyl tert-butyl ether (kCl• = 1.3 × 109 M−1 s−1) aretypical examples.18,42 Most large TrOC molecules (>200 Da)or those bearing electron-rich moieties exhibit high reactivitytoward Cl• (kCl• ranging from ∼6 × 109 to ∼4 × 1010 M−1

s−1). Advanced estimation methods such as the groupcontribution method could be developed for predicting Cl•

reaction rate constants in the future.29,43

Figure 3 shows the QSAR models developed for phenols,alkoxy benzenes, and anilines in their reactions with Cl2

•−.Both the data determined in this work and that reported byHasegawa and his colleagues were used to develop themodels.19 The QSAR models for these three groups of TrOCsall show good results (all of the k values were predicted by theresulting linear correlations within a factor of 1/2−2 and R2 ≥0.881); see below

∑ σ= − = =+k n Rlog( ) 8.369 0.666 15, 0.924o m pPhOH , ,2

(7)

∑ σ= − = =+k n Rlog( ) 8.246 0.881 9, 0.951o m pABD , ,2

(8)

∑ σ= − = =+k n Rlog( ) 8.826 0.476 7, 0.881o m pArNH , ,2

2

(9)

kPhOH, kABD, and kArNH2here are second-order rate constants for

reactions of Cl2•− with nondissociated phenols, alkoxy

benzenes, and anilines, respectively.Each of these types of compounds produced a negative

Hammett slope, suggesting an electrophilic reaction mecha-nism (e.g., electron transfer). The rate constants increased withthe electron-donating capacity of the substituents (Figure 3).The highest slope was −0.881 for the alkoxy benzenes,indicating that their reaction with Cl2

•− is most stronglyimpacted by any substituents involved. The highest interceptvalue of 8.826 for the anilines suggests that they are particularlyreactive with Cl2

•−. Equations 7−9 can, with due caution, beused to predict rate constants for TrOCs bearing thesefunctional groups. The predicted kCl2•− values for somerepresentative TrOCs are listed in Table S3 and shown asred dots in Figure 3. These include substituted phenols,estrogens, and aniline dyes, which all show considerablereactivity toward Cl2

•−. The QSARs for benzene derivatives

cannot be developed since some electron-withdrawingsubstituents such as −Br, −Cl, and −CN on benzene ringsenhance the reaction rate constants by factors of up to 3.5 forsubstituted benzoic acids.19 Different reaction mechanismsmight then be involved, such as chlorine addition to thearomatic rings.

Reaction Mechanisms of Cl•. To explore the reactionmechanisms of Cl• with TrOCs, the transient intermediatesgenerated were tracked.Cation radicals were first tracked to explore the SET

reaction pathway (eq 10). Due to their extremely short lifetimeand fast dissociation at pH 7, acidic pH (i.e., pH = 3) was usedfor the experiments.44−46

+ → +• •+ −Cl TrOC TrOC Cl (10)

1,3,5-Trimethoxybenzene (TMB) was used as a modelcompound as its cation radical (TMB•+) has been reported

Figure 3. Correlations between the second-order rate constants (k)for the reactions of Cl2

•− with (a) phenols, (b) alkoxy benzenes, and(c) anilines vs ∑σo,m,p

+. The solid lines represent the fitted linearrelationships and the dashed lines represent the predicted error rangesof a factor of 1/2 and 2, respectively.

Environmental Science & Technology Article

DOI: 10.1021/acs.est.9b02462Environ. Sci. Technol. 2019, 53, 11170−11182

11177

to show peak absorption at 580 nm and has known absorptivity(ε580nm = 5710 M−1 cm−1).47 TMB•+ was observed from thereaction of TMB with Cl• at pH 3.0 (Figure 4a). Both its decayrate and absorbance decreased with increasing pH (pH > 3.0),and the signal became too weak to analyze at pH 6.8 (FigureS7). The signal recorded at pH 3.0 has no difference with thatrecorded at pH 2.0 but decreased significantly at pH 5.0,indicating that TMB•+ had a pKa between 3.0 and 5.0.45 Usingthe product buildup kinetics to track TMB•+ formation, thespecific rate constant through SET reactions was determinedto be 8.3 ×109 M−1 s−1, which was 62.4% of the overallreaction rate constant of Cl• with TMB determined usingcompetition kinetics method (1.33 ×1010 M−1 s−1) at pH 3.0(Figure S8). Cation radicals of many TrOCs were alsodetected in Cl•-induced reactions, including sulfonamides(400−440 nm) and mesalazine (420 nm) (Figure S9). Thestrong absorption in the range of 400−440 nm was due to thegeneration of aniline cation radicals (C6H5NH2

•+).37 Cationradicals at 440 and 500 nm were also observed from Cl•

reactions with atenolol and carbamazepine, respectively. Theirsignals had similar pH dependence as TMB•+, which furtherproved that these signals were attributed to the generation ofcation radicals. The specific rate constant through SETreactions and overall rate constant (pH 3.0) were alsocompared for each above-mentioned TrOC (Table S4).Results show the specific rate constants for the SET channelaccounted for more than 80% of the overall rate constants(e.g., 90.4% for sulfamethazine). The dominant SET reaction

pathways also explained the very fast reaction rate constants ofCl• with some TrOCs (e.g., sulfonamides), which canapproach and even exceed the diffusion-control limits.The reaction mechanism of Cl• addition to an aromatic ring

has been previously proved by detection of Cl•-adducts as wellas quantum mechanical calculations.20,22,25 Cl• addition wasalso involved in the reaction of Cl• with many TrOCs in thisstudy. In the reaction of Cl• with penicillin G, only a broadabsorption centered at 320 nm was observed, whereas thekinetic trace recorded at 320 nm decayed very slowly (Figure4b). It seemed that some long-lived transient species weregenerated, which overlapped with Cl• absorption in the sameregion. Their kinetic traces were isolated by kinetic modeling(the inset in Figure 4b). The long-lived transient species canbe attributed to the Cl•-adducts according to their character-istic absorption region (∼300−350 nm).20,22 As only Cl•-adducts were observed in the reaction of Cl• with penicillin G,this indicated that addition was the major reaction mechanism.

+ → −• •Cl TrOC (Cl TrOC) (11)

The kinetic traces recorded at 320 nm increased withincreasing TrOC concentrations, which was because theabsorption of Cl•-adducts increased with increasing TrOCconcentrations (Figure S10). It is hard to capture themseparately from the time-resolved spectra since theirabsorption bands were overlapped by the absorption of Cl•.The Cl•-addition pathways were rather common in reactions

Figure 4. (a) Time-resolved spectra of the reaction of TMB with Cl•. Inset: kinetic traces recorded at 320 nm (the absorbance multiplied by 1/5)and 580 nm. Conditions: [Chloroacetone] = 10 mM, [TMB] = 0.4 mM, and pH = 3.0. (b) Time-resolved spectra of reaction of penicillin G withCl•. Inset: kinetic traces recorded at 320 nm; the decay trend of Cl• and buildup trend of Cl•-adducts were isolated by kinetic modeling.Conditions: [Chloroacetone] = 10 mM, [penicillin G] = 0.2 mM, and pH = 3.0. (c) Time-resolved spectra of reaction of BPA with Cl•. Inset:kinetic traces recorded at 320 and 420 nm. [Chloroacetone] = 10 mM, [BPA] = 0.189 mM, and pH = 7.0. (d) Time-resolved spectra of thereaction of TMB with Cl2

•−. Inset: kinetic traces recorded at 340 nm (the absorbance multiplied by 1/10) and 580 nm. Conditions: [S2O82−] = 10

mM, [Cl−] = 0.5 M, [TMB] = 0.4 mM, and pH = 3.0.

Environmental Science & Technology Article

DOI: 10.1021/acs.est.9b02462Environ. Sci. Technol. 2019, 53, 11170−11182

11178

of Cl• with TrOCs, as Cl•-adducts of many TrOCs wereobserved in this work.The H-abstraction reaction pathway was observed for Cl•

reactions with some phenol-containing TrOCs.

+ → +• •Cl PhOH PhO HCl (12)

For example, bisphenol A (BPA), amoxicillin, and acetamino-phen generated absorption bands centered at around 390−420nm during their reactions with Cl• (Figures 4c and S11). Thisis the characteristic absorption band of phenoxyl radicals,which may be formed by H-abstraction from phenols byCl•.37,48,49 Using the product buildup kinetics method to trackphenoxyl radical formation from reactions of Cl• with BPA, thespecific rate constant through the H-abstraction reactionpathway was 1.45 × 1010 M−1 s−1, which was 79.0% of theoverall reaction rate constant determined by tracking Cl• decay(1.82 × 1010 M−1 s−1) at pH 7.0 (Table 1). Similarly, H-abstraction dominated the reactions of Cl• with acetamino-phen (93%), amoxicillin (62.2%), and phenol (80%).It should be noted that Cl• might react with TrOCs through

multiple pathways. The failure to develop QSAR for Cl• rateconstants using the Hammett constant also proved that theelectron-donating/withdrawing neighboring functional groupsdid not affect the overall reaction and multiple reactionpathways other than SET were involved. For example, the Cl•-adduct of TMB was also observed other than TMB•+ (FigureS10g), suggesting that SET and Cl•-addition both contributedto the reaction of Cl• with TMB (with SET accounting for62.4% overall reaction kinetics as mentioned above). SET hasbeen proved as the major reaction channel of Cl• with sulfadrugs, but Cl•-adduct was also observed in the reaction of Cl•

with sulfamethoxazole (Figure S10f). Both Cl•-adducts andphenoxyl radicals were observed in the reaction of Cl• withamoxicillin and BPA, suggesting the involvement of bothaddition and H-abstraction (Figures S10 and S11).Reaction Mechanisms of Cl2

•−. The great variation in thereaction rate constants with Cl2

•− reflects the different reactionpathways involved. The literature reported that the SETpathway generally had higher reaction rate constants than H-abstraction and Cl2

•− addition pathways.19 Using TMB as amodel compound and tracking its transient spectra, TMB•+

was also observed in the reaction of TMB with Cl2•− (Figure

4d). The consistency between the specific rate constant fromthe SET pathway (2.87 × 109 M−1 s−1, determined from theformation of cation radicals) and the overall rate constant(2.61 × 109 M−1 s−1, determined from the Cl2

•− decaykinetics) indicated that SET predominated the reaction ofCl2

•− with TMB (Figure S12). Phenols, alkoxy benzenes, andanilines were proposed to mainly undergo SET with Cl2

•− asthe QSAR models using the Hammett constants indicated theelectron-donating and -withdrawing neighboring functionalgroups in these compounds affected the overall reactions withCl2

•−. Cl2•− is also known to react with oxygenated

hydrocarbons mainly through H-abstraction with rate con-stants ranging from ∼103 to ∼105 M−1 s−1, and it reacts witholefins mainly via addition with rate constants ranging from∼105 to ∼108 M−1 s−1.19,26 The rate constants of aspirin,carbamazepine, sucralose, phthalate esters, and macrolides arearound 107 M−1 s−1 or even lower, implying that addition orhydrogen abstraction might be their principal reactionpathways with Cl2

•−. The intermediates spectra were nottracked further.

Reaction Sites. A detailed discussion for the reaction sitesof TrOCs reacting with Cl• and Cl2

•− can be found in TextS10 and Table S5.

Comparison with HO•. Cl• and Cl2•− are often coexistent

with HO• in AOP treatment, so their reaction rate constantsshould be compared. The reaction rate constants of HO• withthe TrOCs in this study varied from 1.0 × 107 to 1.7 × 1010

M−1 s−1 (Table 1). The ratios RCl•/HO

• (i.e., kCl•/kHO•) andRCl2

•−/HO

• were calculated, with the results displayed in FigureS13. A total of 44 of the 51 TrOCs studied (86%) have Cl•

reaction rate constants higher or comparable to their HO• rateconstants. All, however, had Cl2

•− reaction rate constants lowerthan their HO• constants [in 47 of 61 cases (77%), less thanone-tenth of those with HO•]. The reaction rate constants ofCl2

•− with cimetidine and iopromide were, however, close tothose with HO•.

Modeling Prediction of Elimination of TrOCs duringWater Treatment. As an example, the degradation of TrOCsattributable to HO•, Cl•, and Cl2

•− in UV/chlorine treatmentwas estimated assuming typical treatment conditions: freechlorine dosed at 20 μM, 1 mgC L−1 Suwannee River naturalorganic matter, 1 mM Cl−, a UV flux of 0.78 mW cm−2, andpH 7. The predicted steady-state concentrations of HO•, Cl•,and Cl2

•− were 7.52 × 10−14, 5.25 × 10−15, and 7.05 × 10−13

M, respectively. The contributions of HO•, Cl•, and Cl2•− to

TrOC removal are presented in Figure S14. Some compoundssuch as dimetridazole, roxithromycin, and estrone werepredominantly degraded by HO• reactions. The contributionof Cl• was generally low due to its low concentration. Cl2

•−

contributed strongly to the degradation of many TrOCs andeven became the dominant oxidant in degrading acetamino-phen, trimethoprim, iopromide, and cimetidine. Thus, the roleof Cl2

•− in TrOC degradation cannot be neglected. Indeed, itmay be particularly important in saline waters rich in chlorideions.

Implications. The reaction rate constants with TrOCsevaluated in this study ranged from 3.10 × 109 to 4.08 × 1010

M−1 s−1 for Cl• and from <1 × 106 to 2.78 × 109 M−1 s−1 forCl2

•−. Cl• was very reactive with many TrOCs having highervalues than HO•. Cl2

•− was more selective, and TrOCs withelectron-rich moieties showed high reactivity toward Cl2

•−

(>109 M−1 s−1). Cl2•− was often considered to be negligible in

pollutant removal due to its low reactivity. However, the resultsin this study suggested that Cl2

•− can play non-negligible oreven dominant roles in the degradation of some TrOCs(Figure S14) and its role may be more pronounced whentreating salty waters with high Cl− concentrations. On theother hand, QSAR models established for reactions of Cl2

•−

with phenols, alkoxy benzenes, and anilines can be used topredict the unknown reaction rate constants that have not beenexperimentally obtained. Additionally, the inclusion of reactionrate constants of Cl• and Cl2

•− with TrOCs in the kineticmodels may be helpful to obtain more accurate concentrationsof Cl• and Cl2

•−.This work also provides mechanistic insight of reactions of

Cl• and Cl2•− with TrOCs. Multiple reaction mechanisms

including addition, H-abstraction, and single electron transferwere identified, which would be helpful to understand TrOCtransformation pathways by Cl• and Cl2

•− oxidation in naturaland engineered aquatic systems. As Cl•-adducts have beenfound in the reaction of Cl• with a variety of TrOCs, chlorine-containing products may be formed. Their formation and

Environmental Science & Technology Article

DOI: 10.1021/acs.est.9b02462Environ. Sci. Technol. 2019, 53, 11170−11182

11179

associated toxicity need to be paid attention when usingchlorine-derived radicals to treat pollutants.

■ ASSOCIATED CONTENT*S Supporting InformationThe Supporting Information is available free of charge on theACS Publications website at DOI: 10.1021/acs.est.9b02462.

Laser flash photolysis experiments; determination ofreaction rate constants of Cl• and Cl2

•−; reactivity ofCH3COCH2

•; ionic strength corrections; QSAR anal-ysis; estimated removal of TrOCs during water treat-ment; method validation and proposed reaction sites;physicochemical properties and sources of selectedtarget compounds; structural approximation; predictedrate constants; comparison of kSET with koverall for Cl•

reactions; Kintecus model for radical concentrationprediction; time-resolved transient absorption spectraand kinetic traces of Cl•, Cl2

•−, and cation radicals, Cl•-adducts, and phenoxy radicals; ratio of k values of Cl•,Cl2

•− to those of HO•; and prediction of isolatedcontributions of Cl•, Cl2

•−, and HO• in UV/chlorinetreatment (PDF)

■ AUTHOR INFORMATIONCorresponding Author*E-mail: yangx36@mail.sysu.edu.cn. Tel: +86-2039332690.ORCIDXin Yang: 0000-0001-9860-423XTaicheng An: 0000-0001-6918-8070NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSWe would like to acknowledge the financial support from theNational Key Research and Development Program of China(2017YFE0133200), the National Natural Science Foundationof China (grants 21577178, 21622706, 21876210, and41425015), and the Guangdong Province Science andTechnology Planning Project (2019A050503006), and theResearch Fund Program of Guangdong Key Laboratory ofEnvironmental Catalysis and Health Risk Control(GKECHRC-11).

■ REFERENCES(1) Zhang, K.; Parker, K. M. Halogen radical oxidants in natural andengineered aquatic systems. Environ. Sci. Technol. 2018, 52, 9579−9594.(2) Leri, A. C.; Mayer, L. M.; Thornton, K. R.; Northrup, P. A.;Dunigan, M. R.; Ness, K. J.; Gellis, A. B. A marine sink for chlorine innatural organic matter. Nat. Geosci. 2015, 8, 620−624.(3) Parker, K. M.; Mitch, W. A. Halogen radicals contribute tophotooxidation in coastal and estuarine waters. Proc. Natl. Acad. Sci.U.S.A. 2016, 113, 5868−5873.(4) Cheng, S.; Zhang, X.; Yang, X.; Shang, C.; Song, W.; Fang, J.;Pan, Y. The multiple role of bromide ion in PPCPs degradation underUV/chlorine treatment. Environ. Sci. Technol. 2018, 52, 1806−1816.(5) Chuang, Y. H.; Chen, S.; Chinn, C. J.; Mitch, W. A. Comparingthe UV/monochloramine and UV/free chlorine advanced oxidationprocesses (AOPs) to the UV/hydrogen peroxide AOP underscenarios relevant to potable reuse. Environ. Sci. Technol. 2017, 51,13859−13868.(6) Zhang, R.; Meng, T.; Huang, C. H.; Ben, W.; Yao, H.; Liu, R.;Sun, P. PPCP degradation by chlorine-UV processes in ammoniacal

water: New reaction insights, kinetic modeling, and DBP formation.Environ. Sci. Technol. 2018, 52, 7833−7841.(7) Kong, X.; Wu, Z.; Ren, Z.; Guo, K.; Hou, S.; Hua, Z.; Li, X.;Fang, J. Degradation of lipid regulators by the UV/chlorine process:Radical mechanisms, chlorine oxide radical (ClO•)-mediated trans-formation pathways and toxicity changes. Water Res. 2018, 137, 242−250.(8) Wardman, P. Reduction potentials of one-electron couplesinvolving free radicals in aqueous solution. J. Phys. Chem. Ref. Data1989, 18, 1637−1755.(9) Fang, J.; Fu, Y.; Shang, C. The roles of reactive species inmicropollutant degradation in the UV/free chlorine system. Environ.Sci. Technol. 2014, 48, 1859−1868.(10) Sun, P.; Lee, W. N.; Zhang, R.; Huang, C. H. Degradation ofDEET and caffeine under UV/chlorine and simulated sunlight/chlorine conditions. Environ. Sci. Technol. 2016, 50, 13265−13273.(11) Li, Y. Y.; Song, W. H.; Fu, W. J.; Tsang, D. C. W.; Yang, X. Theroles of halides in the acetaminophen degradation by UV/H2O2

treatment: Kinetics, mechanisms, and products analysis. Chem. Eng. J.2015, 271, 214−222.(12) Caregnato, P.; Rosso, J. A.; Soler, J. M.; Arques, A.; Martire, D.O.; Gonzalez, M. C. Chloride anion effect on the advanced oxidationprocesses of methidathion and dimethoate: role of Cl2

•− radical.Water Res. 2013, 47, 351−362.(13) Sun, P.; Meng, T.; Wang, Z.; Zhang, R.; Yao, H.; Yang, Y.;Zhao, L. Degradation of Organic Micropollutants in UV/NH2ClAdvanced Oxidation Process. Environ. Sci. Technol. 2019, 53, 9024−9033.(14) Parker, K. M.; Reichwaldt, E. S.; Ghadouani, A.; Mitch, W. A.Halogen Radicals Promote the Photodegradation of Microcystins inEstuarine Systems. Environ. Sci. Technol. 2016, 50, 8505−8513.(15) Guo, K.; Wu, Z.; Shang, C.; Yao, B.; Hou, S.; Yang, X.; Song,W.; Fang, J. Radical chemistry and structural relationships of PPCPdegradation by UV/chlorine treatment in simulated drinking water.Environ. Sci. Technol. 2017, 51, 10431−10439.(16) Guo, K.; Wu, Z.; Yan, S.; Yao, B.; Song, W.; Hua, Z.; Zhang, X.;Kong, X.; Li, X.; Fang, J. Comparison of the UV/chlorine and UV/H2O2 processes in the degradation of PPCPs in simulated drinkingwater and wastewater: Kinetics, radical mechanism and energyrequirements. Water Res. 2018, 147, 184−194.(17) Lian, L.; Yao, B.; Hou, S.; Fang, J.; Yan, S.; Song, W. Kineticstudy of hydroxyl and sulfate radical-mediated oxidation ofpharmaceuticals in wastewater effluents. Environ. Sci. Technol. 2017,51, 2954−2962.(18) Wicktor, F.; Donati, A.; Herrmann, H.; Zellner, R. Laser basedspectroscopic and kinetic investigations of reactions of the Cl atomwith oxygenated hydrocarbons in aqueous solution. Phys. Chem.Chem. Phys. 2003, 5, 2562−2572.(19) Hasegawa, K.; Neta, P. Rate constants and mechanisms ofreaction of Cl2

•− radicals. J. Phys. Chem. A 1978, 82, 854−857.(20) Alegre, M. L.; Gerones, M.; Rosso, J. A.; Bertolotti, S. G.;Braun, A. M.; Martire, D. O.; Gonzalez, M. C. Kinetic study of thereactions of chlorine atoms and Cl2

•− radical anions in aqueoussolutions. 1. Reaction with benzene. J. Phys. Chem. A 2000, 104,3117−3125.(21) Buxton, G. V.; Bydder, M.; Salmon, G. A.; Williams, J. E. Thereactivity of chlorine atoms in aqueous solution Part III. The reactionsof Cl• with solutes. Phys. Chem. Chem. Phys. 2000, 2, 237−245.(22) Martire, D. O.; Rosso, J. A.; Bertolotti, S.; Roux, G. C. L.;Braun, A. M.; Gonzalez, M. C. Kinetic study of the reactions ofchlorine atoms and Cl2

•− radical anions in aqueous solutions. II.Toluene, benzoic acid, and chlorobenzene. J. Phys. Chem. A 2001,105, 5385−5392.(23) Gilbert, B. C.; Stell, J. K.; Peet, W. J.; Radford, K. J. Generationand reactions of the chlorine atom in aqueous solution. J. Chem. Soc.,Faraday Trans. 1 1988, 84, 3319−3330.(24) Herrmann, H. Kinetics of aqueous phase reactions relevant foratmospheric chemistry. Chem. Rev. 2003, 103, 4691−4716.

Environmental Science & Technology Article

DOI: 10.1021/acs.est.9b02462Environ. Sci. Technol. 2019, 53, 11170−11182

11180

(25) Minakata, D.; Kamath, D.; Maetzold, S. Mechanistic Insightinto the Reactivity of Chlorine-Derived Radicals in the Aqueous-Phase UV-Chlorine Advanced Oxidation Process: Quantum Mechan-ical Calculations. Environ. Sci. Technol. 2017, 51, 6918−6926.(26) Neta, P.; Huie, R. E.; Ross, A. B. Rate constants for reactions ofinorganic radicals in aqueous solution. J. Phys. Chem. Ref. Data 1988,17, 1027−1284.(27) Song, W.; Cooper, W. J.; Mezyk, S. P.; Greaves, J.; Peake, B. M.Free radical destruction of β-blockers in aqueous solution. Environ.Sci. Technol. 2008, 42, 1256−1261.(28) Jasper, J. T.; Shafaat, O. S.; Hoffmann, M. R. Electrochemicaltransformation of trace organic contaminants in latrine wastewater.Environ. Sci. Technol. 2016, 50, 10198−10208.(29) Lee, Y.; von Gunten, U. Quantitative structure-activityrelationships (QSARs) for the transformation of organic micro-pollutants during oxidative water treatment. Water Res. 2012, 46,6177−6195.(30) Zegota, H.; Schuchmann, M. N.; Schulz, D.; Sonntag, C. V.Acetonylperoxyl radicals, CH3COCH2O2

•: A study on the γ-radiolysisand pulse radiolysis of acetone in oxygenated aqueous solutions. Z.Naturforsch. 1986, 41, 1015−1022.(31) Buxton, G. V.; Bydder, M.; Salmon, G. A. Reactivity of chlorineatoms in aqueous solution Part 1 The equilibrium Cl•+Cl−→Cl2

•−. J.Chem. Soc. Faraday Trans. 1998, 94, 653−657.(32) Buxton, G. V.; Bydder, M.; Salmon, G. A. The reactivity ofchlorine atoms in aqueous solution Part II.The equilibriumSO4

•−+Cl− → Cl•+SO42−. Phys. Chem. Chem. Phys. 1999, 1, 269−273.

(33) Hansch, C.; Leo, A.; Taft, R. W. A survey of Hammettsubstituent constants and resonance and field parameters. Chem. Rev.1991, 91, 165−195.(34) Alfassi, Z. B.; Huie, R. E.; Neta, P.; Shoute, L. C. T.Temperature dependence of the rate constants for reaction ofinorganic radicals with organic reductants. J. Phys. Chem. A 1990, 94,8800−8805.(35) Song, W.; Yan, S.; Cooper, W. J.; Dionysiou, D. D.; O’Shea, K.E. Hydroxyl radical oxidation of cylindrospermopsin (cyanobacterialtoxin) and its role in the photochemical transformation. Environ. Sci.Technol. 2012, 46, 12608−12615.(36) Fang, H. S.; Gao, Y. P.; Li, G. Y.; An, J. B.; Wong, P. K.; Fu, H.Y.; Yao, S. D.; Nie, X. P.; An, T. C. Advanced oxidation kinetics andmechanism of preservative propylparaben degradation in aqueoussuspension of TiO2 and risk assessment of its degradation products.Environ. Sci. Technol. 2013, 47, 2704−2712.(37) Alfassi, Z. B.; Mosseri, S.; Neta, P. Reactivities of chlorineatoms and peroxyl radicals formed in the radiolysis of dichloro-methane. J. Phys. Chem. A 1989, 93, 1380−1385.(38) Tentscher, P. R.; Lee, M.; von Gunten, U. MicropollutantOxidation Studied by Quantum Chemical Computations: Method-ology and Applications to Thermodynamics, Kinetics, and ReactionMechanisms. Acc. Chem. Res. 2019, 52, 605−614.(39) Chateauneuf, J. E. Direct measurement of the absolute kineticsof chlorine atom in carbon tetrachloride. J. Am. Chem. Soc. 1990, 112,442−444.(40) Song, W.; Xu, T.; Cooper, W. J.; Dionysiou, D. D.; De La Cruz,A. A.; O’Shea, K. E. Radiolysis studies on the destruction ofmicrocystin-LR in aqueous solution by hydroxyl radicals. Environ. Sci.Technol. 2009, 43, 1487−1492.(41) Dodd, M. C.; Buffle, M. O.; von Gunten, U. Oxidation ofantibacterial molecules by aqueous ozone: Moiety-specific reactionkinetics and application to ozone-based wastewater treatment.Environ. Sci. Technol. 2006, 40, 1969−1977.(42) National Institute of Standards and Technology. NDRL/NISTSolution Kinetics Database on the Web. http://kinetics.nist (accessedFeb 25, 2019).(43) Minakata, D.; Li, K.; Westerhoff, P.; Crittenden, J. Develop-ment of a group contribution method to predict aqueous phasehydroxyl radical (HO•) reaction rate constants. Environ. Sci. Technol.2009, 43, 6220−6227.

(44) Jie, J.; Liu, K.; Wu, L.; Zhao, H.; Song, D.; Su, H. Capturing theradical ion-pair intermediate in DNA guanine oxidation. Sci. Adv.2017, 3, No. e1700171.(45) Kobayashi, K.; Tagawa, S. Direct observation of guanine radicalcation deprotonation in duplex DNA using pulse radiolysis. J. Am.Chem. Soc. 2003, 125, 10213−10218.(46) Bietti, M.; Baciocchi, E.; Steenken, S. Lifetime, ReductionPotential and Base-Induced Fragmentation of the Veratryl AlcoholRadical Cation in Aqueous Solution. Pulse Radiolysis Studies on aLigninase “Mediator”. J. Phys. Chem. A 1998, 102, 7337−7342.(47) O’Neill, P.; Steenken, S.; Schulte-Frohllnde, D. Formation ofRadical Cations of Methoxylated Benzenes by Reaction with OHRadicals, Tl2+, Ag2+, and S04*

− in Aqueous Solution. An Optical andConductometric Pulse Radiolysis and in situ Radiolysis Electron SpinResonance Study. J. Phys. Chem. A 1975, 79, 2773−2779.(48) Field, R. J.; Raghavan, N. V.; Brummer, J. G. A pulse radiolysisinvestigation of the reactions of bromine dioxide radical (BrO2

•) withFe(CN)6

4−, Mn(II), phenoxide ion, and phenol. J. Phys. Chem. A1982, 86, 2443−2449.(49) Spanget-Larsen, J.; Gil, M.; Gorski, A.; Blake, D. M.; Waluk, J.;Radziszewski, J. G. Vibrations of the phenoxyl radical. J. Am. Chem.Soc. 2001, 123, 11253−11261.(50) Wols, B. A.; Hofman-Caris, C. H. Review of photochemicalreaction constants of organic micropollutants required for UVadvanced oxidation processes in water. Water Res. 2012, 46, 2815−2827.(51) Zhao, Y.; Kuang, J. M.; Zhang, S. Y.; Li, X.; Wang, B.; Huang,J.; Deng, S. B.; Wang, Y. J.; Yu, G. Ozonation of indomethacin:Kinetics, mechanisms and toxicity. J. Hazard. Mater. 2017, 323, 460−470.(52) Song, W.; Chen, W.; Cooper, W. J.; Greaves, J.; Miller, G. E.Free-radical destruction of β-lactam antibiotics in aqueous solution. J.Phys. Chem. A 2008, 112, 7411−7417.(53) Dail, M. K.; Mezyk, S. P. Hydroxyl-radical-induced degradativeoxidation of β-lactam antibiotics in water: absolute rate constantmeasurements. J. Phys. Chem. A 2010, 114, 8391−8395.(54) Santoke, H.; Song, W.; Cooper, W. J.; Greaves, J.; Miller, G. E.Free-radical-induced oxidative and reductive degradation of fluo-roquinolone pharmaceuticals: Kinetic studies and degradationmechanism. J. Phys. Chem. A 2009, 113, 7846−7851.(55) Marquez, G.; Rodriguez, E. M.; Beltran, F. J.; Alvarez, P. M.Determination of rate constants for ozonation of ofloxacin in aqueoussolution. Ozone: Sci. Eng. 2013, 35, 186−195.(56) Mezyk, S. P.; Neubauer, T. J.; Cooper, W. J.; Peller, J. R. Free-radical-induced oxidative and reductive degradation of sulfa drugs inwater: absolute kinetics and efficiencies of hydroxyl radical andhydrated electron reactions. J. Phys. Chem. A 2007, 111, 9019−9024.(57) Razavi, B.; Song, W.; Cooper, W. J.; Greaves, J.; Jeong, J. Free-radical-induced oxidative and reductive degradation of fibratepharmaceuticals: Kinetic studies and degradation mechanisms. J.Phys. Chem. A 2009, 113, 1287−1294.(58) Sunil Paul, M. M.; Aravind, U. K.; Pramod, G.; Saha, A.;Aravindakumar, C. T. Hydroxyl radical induced oxidation oftheophylline in water: a kinetic and mechanistic study. Org. Biomol.Chem. 2014, 12, 5611−5620.(59) Latch, D. E.; Stender, B. L.; Packer, J. L.; Arnold, W. A.;McNeill, K. Photochemical fate of pharmaceuticals in the environ-ment: Cimetidine and ranitidine. Environ. Sci. Technol. 2003, 37,3342−3350.(60) Ching, T. L.; Haenen, G. R. M. M.; Bast, A. Cimetidine andother H2 receptor antagonists as powerful hydroxyl radicalscavengers. Chem.-Biol. Interact. 1993, 86, 119−127.(61) Wang, W. L.; Wu, Q. Y.; Huang, N.; Wang, T.; Hu, H. Y.Synergistic effect between UV and chlorine (UV/chlorine) on thedegradation of carbamazepine: Influence factors and radical species.Water Res. 2016, 98, 190−198.(62) Joshi, R.; Kumar, S.; Unnikrishnan, M.; Mukherjee, T. Freeradical scavenging reactions of sulfasalazine, 5-aminosalicylic acid and

Environmental Science & Technology Article

DOI: 10.1021/acs.est.9b02462Environ. Sci. Technol. 2019, 53, 11170−11182

11181

sulfapyridine: Mechanistic aspects and antioxidant activity. FreeRadical Res. 2005, 39, 1163−1172.(63) Bourgin, M.; Borowska, E.; Helbing, J.; Hollender, J.; Kaiser, H.P.; Kienle, C.; McArdell, C. S.; Simon, E.; von Gunten, U. Effect ofoperational and water quality parameters on conventional ozonationand the advanced oxidation process O3/H2O2: Kinetics of micro-pollutant abatement, transformation product and bromate formationin a surface water. Water Res. 2017, 122, 234−245.(64) Huang, Y.; Kong, M. H.; Westerman, D.; Xu, E. G.; Coffin, S.;Cochran, K. H.; Liu, Y. Q.; Richardson, S. D.; Schlenk, D.; Dionysiou,D. D. Effects of HCO3

− on degradation of toxic contaminants ofemerging concern by UV/NO3

−. Environ. Sci. Technol. 2018, 52,12697−12707.(65) An, T. C.; Gao, Y. P.; Li, G. Y.; Kamat, P. V.; Peller, J.; Joyce,M. V. Kinetics and mechanism of •OH mediated degradation ofdimethyl phthalate in aqueous solution: Experimental and theoreticalstudies. Environ. Sci. Technol. 2014, 48, 641−648.(66) Lei, Y.; Zhu, C. Z.; Lu, J.; Zhu, Y. C.; Zhu, M. Y.; Chen, T. H.;Peng, S. C. Photochemical reaction kinetics and mechanisms ofdiethyl phthalate with N (III) in the atmospheric aqueousenvironment. J. Photochem. Photobiol., A 2018, 362, 21−30.(67) Wen, G.; Ma, J.; Liu, Z. Q.; Zhao, L. Ozonation kinetics for thedegradation of phthalate esters in water and the reduction of toxicityin the process of O3/H2O2. J. Hazard. Mater. 2011, 195, 371−377.(68) Gao, Y.; Ji, Y. M.; Li, G. Y.; An, T. C. Theoretical investigationon the kinetics and mechanisms of hydroxyl radical-inducedtransformation of parabens and its consequences for toxicity:Influence of alkyl-chain length. Water Res. 2016, 91, 77−85.(69) Ning, B.; Graham, N. J. D.; Zhang, Y. P. Degradation ofoctylphenol and nonylphenol by ozone - Part II: Indirect reaction.Chemosphere 2007, 68, 1173−1179.

Environmental Science & Technology Article

DOI: 10.1021/acs.est.9b02462Environ. Sci. Technol. 2019, 53, 11170−11182

11182