Chlorination and bromination of olefins: Kinetic and mechanistic aspects

Hypochlorous acid (HOCl) is typically assumed to be the primary reactive species in free available chlorine (FAC) solutions. Lately, it has been shown that less abundant chlorine species such as chlorine monoxide (Cl 2 O) and chlorine (Cl 2 ) can also influence the kinetics of the abatement of certain organic compounds during chlorination. In this study, the chlorination as well as bromination kinetics and mechanisms of 12 olefins (including 3 aliphatic and 9 aromatic olefins) with different structures were explored. HOCl shows a low reactivity towards the selected olefins with species-specific second-order rate constants < 1.0 M −1 s −1 , about 4-6 orders of magnitude lower than those of Cl 2 O and Cl 2 . HOCl is the dominant chlorine species during chlorination of olefins under typical drinking water conditions, while Cl 2 O and Cl 2 are likely to play important roles at high FAC concentration near circum-neutral pH (for Cl 2 O) or at high Cl − concentration under acidic conditions (for Cl 2 ). Bromination of the 12 olefins suggests that HOBr and Br 2 O are the major reactive species at pH 7.5 with species-specific second-order rate constants of Br 2 O nearly 3-4 orders of magnitude higher than of HOBr (ranging from < 0.01 to > 10 3 M −1 s −1 ). The reactivities of chlorine and bromine species towards olefins follow the order of HOCl < HOBr < Br 2 O < Cl 2 O ≈ Cl 2 . Generally, electron-donating groups (e.g., CH 2 OHand CH 3 -) enhances the reactivities of olefins towards chlorine and bromine species by a factor of 3-10 2 , while electron-withdrawing groups (e.g., Cl-, Br-, NO 2 -, COOH-, CHO-, -COOR, and CN-) reduce the reactivities by a factor of 3-10 4 . A reasonable linear free energy relationship (LFER) between the species-specific second-order rate constants of Br 2 O or Cl 2 O reactions with aromatic olefins and their Hammett σ+ was established with a more negative ρ value for Br 2 O than for Cl 2 O, indicating that Br 2 O is more sensitive to substitution effects. Chlorinated products including HOCl-adducts and decarboxylated Cl-adduct were identified during chlorination of cinnamic acid by high-performance liquid chromatography/high resolution mass spectrometry (HPLC/HRMS). © 2020 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license ( http://creativecommons.org/licenses/by/4.0/ )


Introduction
Chemical oxidants such as chlorine and ozone have been applied for water disinfection since the 20th century ( Le Paulouë and Langlais, 1999 ;McGuire, 2006 ;von Sonntag and von Gunten, 2012 ). Consecutively, they have been used for the transformation/abatement of inorganic and/or organic micropollutants detected in water ( von Gunten, 2018 ). However, undesirable dis-infection by-products (DBPs) or harmful transformation products can be generated from the reactions of oxidants with water matrix components and micropollutants under certain conditions ( Boorman, 1999 ;Gallard and von Gunten, 2002a , b;Richardson et al., 2007 ;Sedlak and von Gunten, 2011 ).
The aim of this study was to establish kinetic and mechanistic data on chlorine or bromine reactions with olefins with various substituents. Both aliphatic and aromatic olefins were investigated in this study. The effect of FAC and bromine speciation on the apparent second-order rate constants was investigated by pH variations and/or concentration variations of FAC and bromine. Based on these systematic variations of the reaction conditions, speciesspecific second-order rate constants for various chlorine (mainly HOCl, Cl 2 O, and Cl 2 ) and bromine species (mainly HOBr and Br 2 O) were determined to assess their importance during chlorination or bromination of various olefinic compounds. Furthermore, products from the chlorination of cinnamic acid were identified.

Chemicals and reagents
The 12 selected olefins ( Fig. 1 ) were purchased from Sigma Aldrich and used without further purification (Table S1 (SI) for more information on purities and CAS numbers). Stock solutions of the olefins (4 mM) were prepared by dissolving the compounds in methanol due to their low solubility in water. Spiking solutions of olefins were prepared by diluting the methanolic stock solutions to 1 mM by ultrapurified water (18.2 M/cm) obtained from a Millipore Milli-Q water purification system. The final concentrations of methanol introduced in the reaction solutions were < 2.5 ‰ (v/v), which was demonstrated in control experiments to have a negligible effect on the FAC and bromine concentrations for the relevant time scales in this study. FAC stock solutions (~45.0 mM) were obtained by diluting a commercial sodium hypochlorite (NaOCl) solution (Sigma Aldrich, reactive chlorine 10-15%) and quantified spectrophotometrically by measuring the OCl − absorbance at 292 nm ( ε = 350 M −1 cm −1 ) ( Kumar and Margerum, 1987 ) every week (concentration changed ≤ 5.0% within one week).
Stock solutions of bromine were obtained by oxidation of potassium bromide (KBr) solutions with a 1.2 times higher ozone concentration (1.0-1.1 mM) at pH 4.0 in 2 mM phosphate buffer to guarantee full oxidation of bromide ( Pinkernell et al., 20 0 0 ). After 24 h, the HOBr solution was purged with nitrogen for 15 min to remove residual ozone and then the solution pH was elevated to 12.0 using NaOH to avoid the disproportionation of HOBr ( Beckwith and Margerum, 1997 ). The HOBr stock solution (0.8-0.9 mM) was stored at 4 °C in brown bottles and standard-ized spectrophotometrically by measuring the OBr − absorbance at 329 nm ( ε = 332 M −1 cm −1 ) at pH 12 before use ( Kumar and Margerum, 1987 ). Fresh HOBr solutions were prepared every week.

Kinetic experiments
All experiments were performed in duplicates at room temperature (23 ± 2 °C) either using amber glass vials with PTFE caps or by stopped-flow. For slow reactions, to apply pseudo first-order conditions, excess FAC or bromine ([FAC] 0 /[olefin] 0 or [bromine] 0 /[olefin] 0 ≥10) working solutions were introduced into pH buffered solutions (10 mM acetic acid for pH 5.0-6.5, 10 mM phosphate buffer for pH 7.0 and 7.5, and 10 mM borate for pH 8.0-9.0). Particularly, in experiments investigating the impacts of Cl − on the chlorination kinetics of CA and Dien-COOH ( Fig. 1 ), KCl (0-20 mM) was also added to acetic acid (10 mM) buffered solutions at pH 5.0. Thereafter, the solution pH was measured and re-adjusted (addition of FAC or bromine slightly increased the pH) by H 2 SO 4 (1 M) or NaOH (1 M) to maintain the reaction solution at the desired pH. Finally, a selected olefin (3-5 μM) was added to the vigorously stirred solution to initiate the reaction and the vials were immediately capped. 1 mL samples were withdrawn after pre-determined reaction times and quenched with thiosulfate (

Analyses of olefins, oxidants, and transformation products
A Thermo Scientific HPLC (UltiMate 30 0 0) equipped with a Symmetry C18 column (125 × 3.0 mm, 5 μm particle size) and a Diode Array Detector (UltiMate 30 0 0) was used for the determination of the selected olefins. The isocratic mobile phase consisted of acetonitrile (phase A) and 10 mM phosphoric acid (pH ~2.3, phase B) at a flow rate of 0.8 mL/min (SI , Table S1 and Figure S3 for more information). UV-vis spectra were measured by a Shimadzu UV-1800 spectrophotometer. The initial and final concentrations of FAC and bromine in working solutions were measured by the ABTS method ( Pinkernell et al., 20 0 0 ), and the results indicated that the oxidant concentrations varied < 10% for all the kinetic runs. The Cl − concentration in the FAC solution was analyzed by ion chromatography (IC) (ICS30 0 0, Thermo Scientific) with a limit of quantification (LOQ) of 0.05 mg/L.
For the identification of transformation products of CA ( Fig. 1 ) by FAC, samples were measured by a Thermo Fisher Scientific Q Exactive Plus high-resolution mass spectrometer (HRMS) coupled to a Dionex UltiMate 30 0 0 LC Pump and a Thermo Pal autosam- pler. An Atlantis® T3 C18 column (3.0 × 150 mm, 3 μm particle size) was used for separation. The mobile phase consisted of water and methanol with both of them containing 1% formic acid at a flow rate of 0.3 mL/min. An initial MS full-scan (mass resolution 140 0 0 0 at 20 0 Da) followed by five data-dependent fragmentation MS/MS experiments (mass resolution 17500 at m/z 200) were performed in both positive and negative electrospray ionization mode (i.e., ESI( + ) and ESI(-)).

Data modeling
Pseudo-first-order rate constants ( k obs ) for the reactions of olefins with FAC or bromine were determined from linear regressions of the experimental ln([olefins] t /[olefins] 0 ) vs time data. Second-order rate constants ( k app ) for the reactions of chlorine species (HOCl, Cl 2 O, and Cl 2 ) and bromine species (HOBr and Br 2 O) with olefins were calculated by nonlinear least-squares regressions in Origin 8.5 ( Seifert, 2014 ). Figure S4 (SI) shows the first-order plots for the oxidation of olefins (5.0 μM) by FAC (0.5-4.5 mM) at pH 7.5. The good linear correlations suggest that the reactions are first-order with respect to olefins. The reaction rates for chlorination of olefins in the presence of excess FAC can be described by Eqs. (10) and (11)

Kinetics of FAC reactions with olefins
where k represents the second-order rate constant for the reaction of FAC with an olefin; k obs is the observed pseudo-first-order rate constant for reaction of FAC with an olefin; [FAC] 0 represents the initial FAC concentration; and n represents the reaction order in FAC. A log transformation of Eq. (11) yields Eq. (12) lo For most of the olefins, n values are ≥1.4. Similar trends were also observed for reactions of aromatic ethers ( Sivey et al., 2012 ) and dimethenamid ( Sivey et al., 2010 ) with FAC, which can be explained by the combined contributions of different chlorine species to the overall chlorine reactivity. The where k HOCl , k C l 2 O , and k C l 2 represent the species-specific secondorder rate constants (M −1 s −1 ) for the reactions of HOCl, Cl 2 O, and Cl 2 with olefins, respectively.
Cl 2 contributes < 5% at [FAC] 0 of 0.1 mM, while it becomes increasingly important with increasing [FAC] 0 for each compound. f(Cl 2 ) remains < 20% for the majority of the compounds (except for two chlorophenols and one cyclic olefin) at pH 7.5 even at [FAC] 0 of 6.0 mM (the maximum [FAC] 0 used in this study). Even though the contribution of Cl 2 seems to be generally low, its relevance for chlorination of olefins at pH 7.5 was investigated for Dien-COOH and CA as representative aliphatic and aromatic olefins, respectively.
Determination of kinetic parameters for chlorination of Dien-COOH and CA . The impact of the Cl − concentration (0-20 mM) on the chlorination kinetics of Dien-COOH and CA at pH 5 were investigated. As shown in Fig. 3 , k obs for both Dien-COOH and CA increase linearly with increasing Cl − concentrations. k obs for Dien-COOH in the presence of 10 mM Cl − is almost 40 times higher than the value without addition of Cl − ([Cl − ] baseline = 1.7 [FAC] 0 , i.e., 0.31 and 1.02 mM for Dien-COOH and CA, respectively). Similar observations were also reported during chlorination of phenols ( Lau et al., 2016 ), alkenes ( Lau et al., 2019 ), and p -xylene ( Voudrias et al., 1988a , b), which was attributed to the reaction of Cl 2 . Based on Eq. (3) , the formation of Cl 2 is favored in the presence of Cl − at lower pH. k C l 2 and k C l 2 O for Dien-COOH and CA could be obtained by fitting the experimental data in Fig. 2 (2(a) for Dien-COOH and 2(d) for CA) and Fig. 3 with Eq. (13) following a similar protocol as reported in the literature ( Lau et al., 2019 ) (SI, Text S3 and Figure S7 for more information). The experimental data could be well fitted (lines in Fig. 2 (a), 2(d), and Fig. 3 ) and the obtained k C l 2 O and k C l 2 were (9.9 ±1.5) × 10 5 M −1 s −1 and (1.1 ±0.2) × 10 6 M −1 s −1 for Dien-COOH; and (2.6 ±0.8) × 10 4 M −1 s −1 and (4.2 ±0.1) × 10 4 M −1 s −1 for CA, respectively ( Table 1 ).   Table 1 . Overall, Cl 2 O and Cl 2 have comparable reactivities towards Dien-COOH and CA (within a factor of 2). The concentrations of HOCl, Cl 2 O, and Cl 2 at pH 7.5 were calculated as a function of the initial FAC concentrations (Text S1, SI) and are shown in Figure S1b (SI). Within the investigated FAC concentration ranges (0.1-6.0 mM), HOCl is by far the dominant chlorine species, followed by Cl 2 O with a concentration nearly 35 times higher than that of Cl 2 . Therefore, the low calculated concentration of Cl 2 leads to only a small contribution (f(Cl 2 ) < 5%) to the chlorination of Dien-COOH and CA at pH 7.5 in Fig. 2 (Figures S8a-S8b, SI).
pH-dependence of chlorination of Dien-COOH and CA. The kinetics of the chlorination of Dien-COOH and CA were investigated in the pH range of 5.0-8.5. k obs for Dien-COOH and CA showed a strong pH-dependence, with a decreasing trend as the pH increased ( Fig. 4 ). The pH dependence of k obs for Dien-COOH and CA was fitted by Eq. (13) with the obtained values for k HOCl , k C l 2 O , and k C l 2 ( Table 1 ). A fairly good agreement between the experimental (squares in Fig. 4 ) and calculated results (lines in Fig. 4 ) was obtained for both Dien-COOH and CA. f(HOCl), f(Cl 2 O), and f(Cl 2 ) for the chlorination of Dien-COOH and CA in the investigated pH range are shown in Figures S8c-S8d (SI). f(Cl 2 ) was ~60% at pH 5 for both Dien-COOH and CA, and decreased to < 5% at pH 7. Cl 2 O played an important role at circum-neutral pH with f(Cl 2 O) ≤20% for both Dien-COOH and CA, while f(Cl 2 O) decreased significantly at lower and higher pH. f(HOCl), f(Cl 2 O), and f(Cl 2 ) for Dien-COOH and CA as a function of pH under typical drinking water chlorination conditions were also calculated and are shown in Figures S8e-S8f (SI). The importance of Cl 2 decreased as the pH increased with f(Cl 2 ) decreasing from > 30% at pH 5 to < 2% at pH 7.5 for both olefins.
Though f(Cl 2 O) reached the maximum at neutral pH, it was < 5% in the pH range investigated. These results suggest that Cl 2 tends to contribute to the chlorination kinetics of olefins in the presence of Cl − at lower pH (pH ≤6.0), while Cl 2 O plays a more significant role in the pH range of 6.0 < pH ≤7.5 at high FAC concentrations. However, for conditions relevant for drinking water, the contributions of Cl 2 and Cl 2 O for the reactions with Dien-COOH and CA are typically low (Figures S8e-S8f, SI).

Contributions of HOCl and Cl 2 O to the oxidation of selected olefins by FAC
Determination of kinetic parameters . The above results suggest that Cl 2 contributes only slightly to the chlorination kinetics of olefins at pH > 6.0. Therefore, to simplify the modeling process, Eq. (13) was adapted to Eq. (15) by excluding the Cl 2 reactions for pH > 6.0.
Second-order rate constants for the reactions of HOCl ( k HOCl ) and Cl 2 O ( k C l 2 O ) with olefins were computed by fitting data of k obs ( Fig. 2 ) with nonlinear least-squares regressions based on Eq. (15) . The k obs data for each olefin at pH 7.5 is well predicted (lines in Fig. 2 ) and the corresponding values for k HOCl and k C l 2 O are compiled in Table 1 . For each of the selected olefins (except for MCA), k C l 2 O was 4-6 orders of magnitude higher than k HOCl . f(Cl 2 O) and f(HOCl) to the overall transformation of olefins by FAC under our experimental conditions were calculated by the second-order rate constants in Table 1 and are presented in Figure S9 (SI). Overall, f(Cl 2 O) increases gradually with increasing [FAC] 0 . At pH 7.5, Table 1 Species-specific second-order rate constants (M −1 s −1 ) for the reactions of chlorine (HOCl and Cl 2 O) (based on Eq. (15) with data from Fig. 2 ) and bromine (HOBr and Br 2 O) (based on Eq. (17) with data from Fig. 6 ) with olefins as well as the half-lives (t 1/2 ) of olefins for typical drinking water chlorination conditions. The values for k C l 2 of Dien-COOH and CA were calculated with Eq. (13) from the data in Fig.s 2 and    Chlorination under realistic conditions . f(Cl 2 O) and f(HOCl) for the transformation of selected olefins by FAC as well as the halflives (t 1/2 ) of olefins at pH 7.5 for typical drinking water chlorination conditions were calculated and are presented in Table 1 . HOCl contributes the most to the transformation of all the selected olefins with f(HOCl) generally > 85%, while Cl 2 O contributes < 10% (except for Dien-OH (10.4%) and 4-CH 3 CA (13.7%)). These observations are different from the results obtained under the laboratory conditions above, for which Cl 2 O plays an equal or more significant role compared to HOCl ( Figure S9, SI). This discrepancy is due to the lower concentration of FAC applied for typical drinking water chlorination conditions (i.e., [FAC] = 2.0 mg/L as Cl 2 ≈28 μM), leading to much lower Cl 2 O concentrations ( Figure S1, SI). Overall, f(Cl 2 O) to the transformation of olefins by chlorine is mainly determined by two factors, (i) the relative reactivities of HOCl and Cl 2 O towards olefins (  HOCl ] (16) f(Cl 2 O) were calculated as a function of k HOCl / k C l 2 O (10 −8 -10 −3 ) for various FAC concentrations ([FAC] = 0.028-6.0 mM) ( Fig. 5 ). f(Cl 2 O) increases with increasing reactivity of Cl 2 O (i.e., decreasing k HOCl / k C l 2 O ) for each HOCl concentration. A higher f(Cl 2 O) can be calculated at higher HOCl concentrations for a fixed k HOCl / k C l 2 O .
The t 1/2 during chlorination ([FAC] = 2.0 mg/L as Cl 2 ≈28 μM) of the selected olefins varies from < 40 min (for Dien-OH) to > 126 days (for 4-NO 2 CA) with values ≥1.1 days for most of the olefins. This result indicates that chlorination will only lead to a partial abatement of the selected olefins in drinking water treatment and distribution systems.
These results indicate that bromine species other than HOBr contribute to the transformation of olefins.

Contributions of HOBr and Br 2 O
Based on the assessment above and the production of HOBr solutions by reaction of Br − with excess ozone, the presence of Br 2 can be excluded for our experimental systems. Therefore, only HOBr and Br 2 O will be considered in this manuscript. Br 2 O which is in equilibrium with HOBr ( Eq. (6) ) generally exhibits higher reactivities towards organic compounds than HOBr, and it can play important roles in the transformation of organic compounds ( Sivey et al., 2013( Sivey et al., , 2015. The high reaction order (n ≥1.4 except for 4-Cl-α-CNCA) in bromine under our experimental conditions is likely due to the contributions of Br 2 O because this species is proportional to [HOBr] 2 ( Eq. (6) ). Accordingly, the observed first-order rate constants for the reactions of bromine with olefins can be interpreted by Eq. (17) : k HOBr and k B r 2 O represent the species-specific second-order rate constants for reactions of HOBr and Br 2 O with an olefin, respectively.
k HOBr and k B r 2 O values were obtained by fitting k obs in Fig. 6 with Eq. (17) via non-linear least-squares regressions. The experimental data were well fitted (lines in Fig. 6 ) and the obtained values for k HOBr and k B r 2 O are compiled in Table 1 . For most of the selected olefins (except for 4-Cl-α-CNCA), k B r 2 O is 3-4 orders of magnitude higher than k HOBr ( < 0.01-> 10 3 M −1 s −1 ). The fractions of HOBr (f(HOBr)) and Br 2 O (f(Br 2 O)) to the overall transformation of olefins under the investigated conditions are shown in Figure  S13 (SI). The relative importance of Br 2 O to the overall reactivity increased with increasing [HOBr] 0 . Br 2 O was the dominant reactive species for the bromination of Dien-OH, CA, 4-ClCA, 4-BrCA, 4-CH 3 CA, 2-CH 3 CA, and α -CH 3 CA with f(Br 2 O) > 50%, in accordance with the higher reaction order in bromine (n > 1.5) therein. For Dien-CHO and 4-NO 2 CA, HOBr also played a non-negligible role in their transformation, while for 4-Cl-α-CNCA, HOBr was the primary bromine species.

Effects of substituents on the kinetics of FAC reactions with olefins
Aliphatic olefins. The species-specific second-order rate constants in Table 1 demonstrate that substituents play a significant role in the chlorination of olefins. For aliphatic olefins, the second-order rate constants for the reactions with Cl 2 O decreased in the order of Dien-OH ((1.8 ±0.1) × 10 7 M −1 s −1 ) > Dien-COOH ((9.9 ±1.5) × 10 5 M −1 s −1 ) > Dien-CHO ((6.8 ±0.9) × 10 3 M −1 s −1 ). A similar trend was observed for k HOCl with Dien-OH (19.0 ±1.5 M −1 s −1 ) > Dien-COOH (4.0 ±0.6 M −1 s −1 ) > Dien-CHO (0.1 ±0.01 M −1 s −1 ). The smaller electron-withdrawing effect of the OH-than the COOH-group leads to a higher electron density in the conjugated double bonds of Dien-OH compared to Dien-COOH ( Lee and von Gunten, 2012 ), which enhances its reactivity with electrophiles such as Cl 2 O (a factor of 10) and HOCl (a factor of 5). A substitution by an aldehyde as in Dien-CHO leads to an electron deficiency and results in a lower chlorine reactivity than Dien-COOH (a factor of 10 2 ), because -CHO is a stronger electron-withdrawing group than -COO − (deprotonated form of Dien-COOH is the major species (p K a = 4.75 ( Arya, 1980 )) at pH 7.5) as suggested by their Taft constants ( σ * = 2.15 and -1.06 for -CHO and -COO − , respectively ( Lee and von Gunten, 2012 )).
Aromatic olefins . For aromatic olefins including CA and its derivatives, substituents at both the benzene ring and at the olefin bond were investigated ( Fig. 1 and Table 1 ). Furthermore, a cinnamic acid methyl ester was investigated. HOCl shows very low reactivities towards CA and its derivatives with k HOCl < 1.0 M −1 s −1 , wherefore, the substituent effects are mainly discussed based on olefin reactions with Cl 2 O. k C l 2 O for CA derivatives (with substituents at the benzene ring) decrease in the order of 4-CH 3 CA ≈ 2-CH 3 CA > CA ≈ 4-ClCA ≈ 4-BrCA >> 4-NO 2 CA. The methyl group as an electron-donor (i.e., Hammet constants σ + for 4-CH 3and 2-CH 3 -groups are -0.306 and -0.210, respectively) ( Lee and von Gunten, 2012 ) increases the electron density on the benzene ring, which may activate the double bond via an inductive effect. Therefore, a higher k app for the reactions of Cl 2 O with 4-CH 3 CA and 2-CH 3 CA compared to the unsubstituted CA (a factor of 3-10) were obtained. The substitution of electron-withdrawing halogen substituents (Cl-and Br-) at the para position of the benzene ring of CA displays slight effects on its reactivity towards Cl 2 O. Comparatively, substitution by a NO 2 -group remarkably decreased the second-order rate constant for the reaction of 4-NO 2 CA with Cl 2 O by a factor of 100 compared to CA. These results are likely due to the stronger electron-withdrawing properties of the NO 2 -group as indicated by its larger Hammet constant (i.e., σ + for NO 2 -, Cl-, and Br-groups are 0.777, 0.112, and 0.148, respectively) ( Lee and von Gunten, 2012 ), which leads to a significantly greater impact on the olefin bond than the Cl-and Br-groups ( Fang et al., 1958 ;Butt and Topsom, 1980 ;Wang and Chen, 2020 ). Similar results were also reported by Aruna and Manikyamba (1995) , where 4-ClCA and CA showed comparable reactivities towards quinolinium dichromate with both of them reacting much more readily than 4-NO 2 CA.
Moreover, substituents on the CA double bond also showed significant impact on the reactivity of aromatic olefins towards Cl 2 O. k C l 2 O for the reaction of α-CH 3 CA with Cl 2 O was 10 times higher compared to CA, while k C l 2 O for the reaction of 4-Cl-α-CNCA with Cl 2 O was nearly 10 times lower compared to 4-ClCA ( Table 1 ). These observations can be explained by the electron-donating effect of the methyl group in α-CH 3 CA, which increases the electron density of the double bond. In contrast, the cyano group in 4-Cl-α-CNCA reduced the electron density of the double bond due to its electron-withdrawing character ( Lee and von Gunten, 2012 ). For MCA, Cl 2 O displayed a much lower reactivity compared to CA due to the ester group in MCA, which is expected to decrease the electron density via its electron-withdrawing effect ( Lee and von Gunten, 2012 ).

LFER
A LFER for the reactions of CA and its derivatives with Cl 2 O/Br 2 O was developed. A previous study showed that Taft constants σ * were the most suitable descriptors for establishing LFER relationships for olefin reactions with ozone ( Lee and von Gun-ten, 2012 ). An attempt was made to establish a relationship between k C l 2 O or k B r 2 O for selected olefins vs Taft σ * but it was impossible due to the lack of Taft σ * for most of the substituents on the olefins. In contrast, a reasonable correlation between log ( k C l 2 O ) or log ( k B r 2 O ) and Hammett σ + for the reaction of Cl 2 O or Br 2 O with CA derivatives (with substituents at the benzene ring) could be established (Figures S14a-S14b, SI). Negative slopes ( ρ) were obtained for both cases with -2.6 ±0.3 (R 2 = 0.95) for Cl 2 O and -3.8 ±0.3 (R 2 = 0.96) for Br 2 O. The more negative ρ value for Br 2 O than for Cl 2 O indicates that Br 2 O is more sensitive to substituent effects. Moreover, the LFER between log ( k C l 2 O ) or log ( k B r 2 O ) vs Hammett σ + by excluding 4-NO 2 CA was also assessed. Acceptable correlations (R 2 = 0.76 and 0.84 for Cl 2 O and Br 2 O, respectively, Figures S14c-S14d, SI) were obtained with a similar trend as observed by including 4-NO 2 CA (Figure S14a-S14b, SI). These results confirm that though 4-NO 2 CA as an end member tends to dominate the correlation (Figures S14a-S14b, SI), it does not affect the final conclusion.

Transformation products
The transformation products from the reaction of FAC (600 μM) with CA (5 μM) at pH 5.0 were analyzed by HPLC/HRMS in both (ESI ( + )) and (ESI (-)) mode. Three transformation products (TPs Fig. 7. Proposed mechanism for the reaction of HOCl with CA at pH 5.0. For a discussion of the transformation products see text and Figures S15-S17 in SI. 1-3) were detected with elution times of 12.92, 14.38, and 19.10 min for TP-1, TP-2, and TP-3, respectively (Figures S15a-S15b, SI). TP-1 and TP-2 identified at ESI(-) mode have the same molecular ion of m/z 199.0168, suggesting that they are structural isomers.  Figure S16, SI) and their chlorine isotopic patterns (i.e., 35 Cl/ 37 Cl = 3/1, Figure S17, SI), corresponding to HOCl-adduct products of CA (C 9 H 8 O 2 ). The higher response of peak TP-2 compared to TP-1 (Figure S15a, SI) indicated that TP-2 might be the major product. TP-3 with a molecular ion of m/z 139.0314 ([M + H] + ) was identified at ESI ( + ) with a minor peak (Figure S15b, SI) and was assigned to C 8 H 7 Cl, corresponding to a Cl-adduct product of the decarboxylated CA. It has been reported that addition of HOCl or HOBr at the olefin β-carbon relative to the benzene ring is favored due to the polarization of the π -electron of CA ( Yamada et al., 1985 ). Hence, TP-1, TP-2, and TP-3 are proposed as 2-hydroxyl-3chlorophenyl propionic acid, 2-chloro-3-hydroxylphenyl propionic acid, and 1-chloro-2-phenylethylene, respectively ( Fig. 7 ). Brominated analogues of TPs 1-3 were previously detected from HOBr reaction with CA involving the formation of a bromonium ion intermediate ( Yamada et al., 1985 ), the corresponding chloronium ion is likely also formed in the case of HOCl ( Fig. 7 ). HOCl/Cladduct products were also reported for chlorination of carbamazepine and CA derivatives ( Norwood et al., 1980 ;Soufan et al., 2013 ). TP-1 and TP-2 were also detected from the CA (5 μM) reaction with HOCl (2.4 mM) at pH 7.0 with a similarly higher response of peak TP-2 than TP-1 ( Figure S18a, SI). TP-3 was not detected at pH 7.0, while a new peak (TP-4) with m/z of 163.0401 appeared in ESI(-) mode ( Figure S18, SI).

Conclusion
The kinetics and mechanisms for the reactions of selected aliphatic and aromatic olefins with chlorine and bromine species were investigated. The following conclusions can be drawn: -HOCl has a low reactivity towards selected olefins with speciesspecific second-order rate constants < 1.0 M −1 s −1 , nearly 4-6 orders of magnitude lower than for Cl 2 O and Cl 2 . -Cl 2 is non-negligible for chlorination of olefins at lower pH (pH ≤6.0) in the presence of excess Cl − , while Cl 2 O plays an important role near neutral pH (6.0 < pH ≤7.5) at high FAC concentration. Nevertheless, HOCl is the dominant chlorine species for the oxidation of the selected olefins under typical drinking water chlorination conditions. -HOBr has a large variation of reactivities towards the selected olefins with species-specific second-order rate constants ranging from < 0.01 to > 10 3 M −1 s −1 , about 3-4 orders of magnitude lower than for Br 2 O. -The reactivities of chlorine and bromine species towards olefins increases in the order of HOCl < HOBr < Br 2 O < Cl 2 O ≈ Cl 2 . -Electron-donating groups (e.g., CH 2 OH-and CH 3 -) enhance the reactivities of cinnamic-type olefins towards chlorine and bromine species by a factor of 3-100, while electronwithdrawing groups (e.g., Cl-, Br-, NO 2 -, COOH-, CHO-, -COOR, and CN-) reduce the reactivities by a factor of 3-10 0 0 0. -HOCl-and/or Cl-adducts are generated during chlorination of cinnamic acid at pH 5.0 and 7.0.