Kinetic Analysis of Haloacetonitrile Stability in Drinking Waters

Subscriber access provided by UNIV OF CALIFORNIA SAN DIEGO LIBRARIES

Article

Kinetic Analysis of Haloacetonitrile Stability in Drinking Waters Yun Yu, and David A. Reckhow Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.5b02772 • Publication Date (Web): 14 Aug 2015 Downloaded from http://pubs.acs.org on August 24, 2015

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Environmental Science & Technology is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 26

Environmental Science & Technology

1

Kinetic Analysis of Haloacetonitrile Stability in Drinking Waters

2

Authors: Yun Yu*1, David A. Reckhow2

3 4 5

1. PhD Candidate, 18 Marston Hall, 130 Natural Resources Rd., Dept. of Civil and Environmental

6

Engineering, University of Massachusetts Amherst, Amherst, MA, 01003-9293. E-mail:

7

[email protected] Phone: 413-362-4918 (Corresponding Author).

8 9 10

2. Professor, 18 Marston Hall, 130 Natural Resources Rd., Dept. of Civil and Environmental Engineering, University of Massachusetts Amherst, Amherst, MA 01003-9293. E-mail: [email protected]. Phone: 413-545-5392

1 ACS Paragon Plus Environment


11

12

ABSTRACT Haloacetonitriles (HANs) are an important class of drinking water disinfection byproducts (DBPs)

13

that are reactive and can undergo considerable transformation on time scales relevant to system

14

distribution (i.e., from a few hours to a week or more). The stability of seven mono-, di- and

15

trihaloacetonitriles was examined under a variety of conditions including different pH levels and

16

disinfectant doses that are typical of drinking water distribution systems. Results indicated that hydroxide,

17

hypochlorite, and their protonated forms could react with HANs via nucleophilic attacks on the nitrile

18

carbon, forming the corresponding haloacetamides (HAMs) and haloacetic acids (HAAs) as major

19

reaction intermediates and end products, respectively. Other stable intermediate products, such as the N-

20

chloro-haloacetamides (N-Cl-HAMs) may form during the course of HAN chlorination. A scheme of

21

pathways for the HAN reactions was proposed and the rate constants for individual reactions were

22

estimated. Under slightly basic conditions, hydroxide and hypochlorite are primary reactants and their

23

associated second-order reaction rate constants were estimated to be 6 to 9 orders of magnitude higher

24

than those of their protonated conjugates (i.e., neutral water and hypochlorous acid), which are much

25

weaker but more predominant nucleophiles at neutral and acidic pHs. Developed using the estimated

26

reaction rate constants, the linear free energy relationships (LFERs) summarized the nucleophilic nature

27

of HAN reactions and demonstrated an activating effect of the electron withdrawing halogens on nitrile

28

reactivity, leading to decreasing HAN stability with increasing degree of halogenation of the substituents,

29

while subsequent shift from chlorine to bromine atoms has a contrary stabilizing effect on HANs. The

30

chemical kinetic model together with the reaction rate constants that were determined in this work can be

31

used for quantitative predictions of HAN concentrations depending on pH and free chlorine contact times

32

(CTs), which can be applied as an informative tool by drinking water treatment and system management

33

engineers to better control these emerging nitrogenous DBPs, and can also be significant in a regulatory

34

context.


Page 2 of 26

Page 3 of 26


35



36

37

INTRODUCTION The presence of halogenated nitrogenous disinfection byproducts (N-DBPs) in drinking

38

waters, including haloacetonitriles (HANs) and haloacetamides (HAMs), is of increasing concern

39

to the public health due to their substantially higher cytotoxic and genotoxic potencies as well as

40

greater developmental toxicity than the currently regulated DBPs.1-5 Haloacetonitriles (HANs) are

41

an important class of N-DBPs that are nearly ubiquitous in drinking waters, with the total mass

42

typically representing approximately 10% of the trihalomethanes (THMs).6,7 Regardless of their

43

relatively low occurrence, the significance of HANs as emerging N-DBPs is offset by their

44

perceived high cytotoxicity and genotoxicity, which are up to two orders of magnitude greater

45

than those of the regulated haloacetic acids (HAAs).1 HANs were first identified in US tap water

46

in 19758 and their formation can be attributed to the chlorination or chloramination of free amino

47

acids,9-12 combined amino acids that are bound to humic structures,7,10 and to a lesser extent,

48

heterocyclic nitrogen in nucleic acids.12,13 Simultaneous to the discovery of HANs, it was noticed

49

that this group of compounds was absent in finished waters with high pHs.9,14 It was revealed by

50

subsequent studies that HANs are metastable and can undergo considerable degradation through

51

neutral and base-catalyzed hydrolysis on time scales relevant to distribution system residence

52

times,17 and the hydrolysis rates increase with increasing pH.9,14-17

53

In addition to pH as an important facilitator, the presence of chlorine is also known to have an

54

accelerating effect on the rate of loss for dihaloacetonitriles (DHANs).7 Compared to the

55

hydrolysis rate of dichloroacetonitrile (DCAN), its accelerated rate of loss in the presence of

56

chlorine indicates the existence of certain independent chlorination pathways.16,17 Peters and co-

57

workers16 proposed that DCAN reacts with chlorine through either direct addition of HOCl onto

58

the cyano group or through hypochlorite catalyzed hydrolysis, and both pathways lead to the

59

formation of dichloroacetamide (DCAM), N-chloro-dichloroacetamide (N-chloro-DCAM) as

60

reaction intermediates, and dichloroacetic acid (DCAA) as final product. Since it is a common 4 ACS Paragon Plus Environment

Page 4 of 26

Page 5 of 26


61

practice for drinking water systems to maintain at least 0.2 mg/L free chlorine residual throughout

62

distribution, and concentrations near the point of entry can be as high as 2 mg/L, it is crucial to

63

understand the impact of chlorine on the stability of a more complete set of HANs, especially the

64

brominated species. Furthermore, the formation and lifetime of the intermediate products have

65

not been assessed and reconciled with the prevailing reaction pathways.

66

On the other hand, there are a large number of US drinking water utilities that have switched

67

from free to combined chlorine, particularly chloramines, to minimize the formation of THMs

68

and HAAs,18 as regulations for these two groups of DBPs are becoming more and more stringent

69

(e.g., the Stage II Disinfectants/Disinfection Byproducts Rule). In addition to the regulatory

70

drivers, there are concerns that chloramines could otherwise enhance the formation of N-DBPs,

71

because the nitrogen incorporated in those N-DBPs can be derived either from the organic

72

precursors, or in the case of chloramination, from the disinfectant.19 Based on the collected data

73

required by the Information Collection Rule (ICR), a higher level of total HANs (i.e., DCAN,

74

BCAN, DBAN and TCAN) was detected in large US surface water plants that used chloramines

75

(both with and without chlorine) than those that only used chlorine.20 However, such elevated

76

HAN occurrence may be attributable to the higher level of precursors in the source water, and not

77

necessarily to an inherent tendency of chloramines to form more HANs.20 In fact, laboratory

78

research has shown a higher formation potential of DCAN during free chlorination than during

79

chloramination regardless whether chloramines were pre-formed or formed in-situ.21 It is

80

noteworthy that the stability of HANs under conditions that are typical of those used by systems

81

practicing chloramination has not been reported to clarify whether the relatively higher

82

occurrence of HANs is due to their greater stability in the presence of chloramines or to the

83

higher tendency of chloramines in the formation of HANs as an additional nitrogen source.

84 85

Despite the general understanding that HANs actively transform into secondary byproducts under a wide range of pH conditions with and without the presence of chlorine, most of the



86

previous work is focused on DHANs, especially on DCAN as the most prevalent HAN

87

species.7,9,14,17 Moreover, the second-order reaction rate constants with regard to relevant drinking

88

water constituents (e.g., hydroxide, free chlorine, etc.) have not been systematically reported or

89

determined for most HANs, the corresponding reaction mechanisms have not been elucidated,

90

and intermediate reaction products have not been quantified and reconciled with the prevailing

91

reaction pathways. For these reasons, a more comprehensive kinetic analysis is necessary to

92

understand the reaction kinetics, pathways, and products for the other monohalogenated,

93

trihalogenated, and particularly the brominated HANs, considering that the brominated DBPs are

94

demonstrated to be more toxic than their chlorinated analogs.3 Perhaps most importantly, kinetic-

95

based decomposition models have not been developed for most of the chlorinated and brominated

96

HANs to adequately characterize their reactions and to allow for quantitative estimation of their

97

concentrations in drinking water distribution systems based on pH and chlorine residual as two

98

predominant influencing factors.

99

The purpose of this study was to evaluate the chemical stability of a more complete set of

100

HANs under a wide range of pH conditions (i.e., pH 6-9) with and without disinfectant doses (i.e.,

101

free chlorine and chloramines), and to obtain a fundamental understanding of HAN reaction

102

mechanisms as well as the nature of consequent reaction products. Another key objective of this

103

investigation was to quantitatively characterize HAN reactions by developing a mathematical

104

kinetic model and determining the corresponding rate constants for individual HAN reactions.

105

The resulting model can be used in quantitative predictions of HAN temporal concentration

106

profiles in drinking water distribution systems with or without simple modifications depending on

107

the dynamics of chlorine residual.

108

109

MATERIALS AND METHODS Chemicals


Page 6 of 26

Page 7 of 26

110


Unless otherwise noted, all chemicals were purchased from Fisher Scientific Co. and were of

111

analytical grade. Purified DBP standard compounds including monochloroacetonitrile (MCAN),

112

monobromoacetonitrile (MBAN), dichloroacetonitrile (DCAN) and trichloroacetonitrile (TCAN)

113

were purchased from Sigma-Aldrich. Bromochloroacetonitrile (BCAN) and dibromoacetonitrile

114

(DBAN) were supplied by Crescent Chemical. Bromodichloroacetonitrile (BDCAN) and three of

115

the brominated HAMs were synthesized by CanSyn Chem. Corp. in Canada. The haloacetic acids

116

mix was obtained from Sigma-Aldrich. Sources and purities of all the standard compounds are

117

available in Table S1.

118

Experimental Conditions

119

All solutions were prepared in ultra-pure Milli-Q water (EMD Millipore Corp.) containing 10

120

mM phosphate buffer and were adjusted to the desired pH with sodium hydroxide or hydrochloric

121

acid. One milliliter of mixed HAN stock solution (1 mg/mL in methanol) was introduced into 4 L

122

buffered solutions at the start of each experiment, so that the initial concentration for individual

123

HANs was approximately 250 µg/L. Chlorination of HANs was conducted by adding small

124

volumes of acidified sodium hypochlorite stock solution (5.65-6%, pH 5.2) to reach the target

125

doses, prior to which, the actual concentration of the chlorine stock was standardized using the

126

DPD-FAS titrimetric method (EPA Method 330.4). Chloramination was carried out by adding

127

small amounts of a 40 mM chloramine stock solution to each sample, and the chloramines were

128

pre-formed by mixing aqueous ammonium sulfate and sodium hypochlorite at a Cl2/N ratio of 0.8

129

M/M, with pH of both solutions adjusted to 8.5 before mixing. After dosing with chlorine or

130

chloramines, samples were partitioned off into 300 mL BOD bottles and were stored headspace-

131

free in a dark 20℃ constant temperature chamber for a maximum of 19 days. At the prescribed

132

reaction times, one bottle of sample would be sacrificed and analyzed immediately for

133

disinfectant residual and DBP concentrations. Six sample replicates were analyzed in this study

134

for the estimation of measurement uncertainties. 7 ACS Paragon Plus Environment


135

Sample Preparation and Chromatographic Analysis

136

The extraction and analysis of HANs was based on EPA Method 551.1. After the prescribed

137

reaction time, 20 mL aliquots of sample were first acidified using 100 µL of 6N hydrochloric acid.

138

In the case of chlorination and chloramination of HANs, residual oxidant was quenched by 20

139

mg/L ascorbic acid after sample acidification. HANs were extracted by adding 4 mL of pentane

140

with an internal standard (1,2-dibromopropane) into each sample, together with 15 g of

141

anhydrous sodium sulfate. The samples were shaken at 300 rpm for 15 minutes and the upper

142

organic layer was collected for chromatographic analysis. Haloacetic acids were quantified

143

following the EPA 552.2 method. The standard operating procedures include pH adjustment and

144

quenching of the disinfectant residual, acidification of 30 mL sample using 1.5 mL of 95.0-98.0%

145

W/W sulfuric acid, and extraction with methyl tert-butyl ether, followed by methylation using 5%

146

acidic methanol. Analysis of the HAMs was conducted via a solid-phase extraction/gas

147

chromatography-mass spectrometry (SPE/GC-MS) method that was developed by the authors for

148

this study (Yu & Reckhow, unpublished method). The SPE procedure involves initial

149

conditioning of the extraction cartridges (Bond Elut PPL, 200 mg, 3 mL, Agilent Technologies)

150

using 9 mL of methanol followed by 6 mL of Milli-Q water, sample loading (100 mL at ~2

151

mL/min), nitrogen drying of the cartridges for 30 minutes, and final elution with 2 mL of ethyl

152

acetate. HANs and the derivatized methyl haloacetates were analyzed using an Agilent 6980 gas

153

chromatography with a linearized micro-electron capture detector (µ-ECD). HAMs were

154

separated and detected by a Varian CP-3800 gas chromatography coupled with a Varian Saturn

155

2200 ion-trap mass spectrometer using chemical ionization. Detailed information about the

156

capillary GC columns and oven temperature programs are provided in Table S2.

157

RESULTS AND DISCUSSION

158

Hydrolysis of Haloacetonitriles


Page 8 of 26

Page 9 of 26


159

The hydrolysis of seven HANs (MCAN, MBAN, DCAN, BCAN, DBAN, TCAN and

160

BDCAN) was investigated at pH 6, 7, 8, 8.5, and 9 in phosphate buffered solutions for reaction

161

times of a few minutes to a total of 19 days (456 hours). Residual HAN concentrations were

162

consistent with a rate law that is first-order in HANs (Figure 1). All seven HANs are most stable

163

at pH 6 and the rate of loss increases with both increasing pH and the number of halogens, which

164

is in general agreement with previous observations regarding HAN hydrolytic stability.7,9,15,17 The

165

instantaneous hydrolysis of trihaloacetonitriles (THANs) even under slightly acidic and neutral

166

pH conditions (i.e., pH 6-7) explains their overall absence in most drinking water systems as

167

noted from the ICR database.20 In sharp contrast, concentrations of the monohalogenated

168

acetonitriles (MHANs) remained nearly constant during the entire period of the hydrolysis

169

experiment regardless of pH (Figure S1). Furthermore, with the same number of halogens in the

170

substituents, HAN hydrolysis rate decreased as the halogens shifted from chlorine to bromine,

171

resulting in the following hierarchy of HAN hydrolytic stability:

172

MBAN>MCAN>DBAN>BCAN>DCAN>BDCAN>TCAN.

173 174

Figure 1. Semi-logarithmic plots of residual HAN concentrations versus reaction time under five

175

hydrolysis pH conditions.

176

To verify the prevailing HAN hydrolysis pathways,17 two putative intermediates (i.e., HAM

177

and HAA) were quantified during the course of HAN hydrolysis. In general, results demonstrated

178

that the loss of HANs was accompanied by a rapid increase in the corresponding HAM 9 ACS Paragon Plus Environment


179

concentrations, followed by a slower formation of the HAAs. Unlike the HAAs, HAMs are

180

metastable intermediates and are also subject to possible hydrolysis depending on pH and the

181

number of halogens in the substituent (Yu & Reckhow, unpublished work). Figure 2 shows the

182

formation of DCAM (and TCAM) and DCAA (and TCAA) during DCAN (and TCAN)

183

hydrolysis under four different pH conditions. It is obvious in Figure 2 that DCAM tended to

184

hydrolyze when pH was above 8, causing its concentration to first increase and then decrease at

185

pH 9. Compared to DCAM, TCAM started to hydrolyze at a lower pH (i.e., pH 8) due to the

186

higher degree of halogenation, and its concentration profile was characterized by a distinct peak

187

at pH 8 and only by its decomposition at pH 9. In spite of the temporal changes in individual DBP

188

concentrations, the molar sum of the three HAN, HAM and HAA species remained constant over

189

reaction time for all the hydrolysis experiments, and this mass balance substantiates the previous

190

hypothesis that hydrolysis of HANs only produces HAMs and HAAs as major reaction

191

products.15,17

192


Page 10 of 26

Page 11 of 26


193

Figure 2. Intermediates formation during the course of DCAN (top row) and TCAN (bottom row)

194

hydrolysis .The dashed lines represent the initial DCAN and TCAN molar concentrations spiked

195

at the beginning of each experiment.

196 197

Based on the 1st-order HAN kinetics that is evident in Figure 1, the full 2nd-order hydrolysis rate law is proposed to be the following: = − ∙ = − ∙ − ∙ (. 1)

198

In the above equation, kH2O and kOH are the neutral and basic hydrolysis rate constants,

199

respectively. Although it has been acknowledged that neutral water is about nine orders of

200

magnitude less reactive to HANs than the anionic hydroxide,17 the neutral hydrolysis rate

201

constant (i.e., ) and the product of can be similar in magnitude, and thus equally

202

contributing to the hydrolysis rate of HAN when pH is below or close to 5. For this reason, the

203

proposal of a neutral hydrolysis pathway and the estimation of the corresponding reaction rate

204

constant (i.e., ) are necessary for the assessment of HAN hydrolysis rates at slightly acidic

205

pHs. Both of the neutral and basic hydrolysis rate constants for the seven HANs were estimated

206

using a Bayesian modeling approach,22 and the details of this statistic estimation method and the

207

resulting rate constant estimates will be addressed in the later section.

208 209

In many cases, this hydrolysis model is further stratified into a hierarchical structure22 by parsing it into a 1st-order observed rate constant Kobs, as shown in Equation 2: = + ∙ (. 2)

210

Given that HAN hydrolysis has been previously investigated by several teams of

211

researchers,7,9,14-17 it is important to reconcile our results with those that have been reported.

212

Generally, when pH was below or equal to 8, the 1st-order observed rate constants (i.e., Kobs)

213

determined in this work were in agreement with literature values (Figure S2). Certain



214

disagreements were noted at higher pHs and the possible explanations for those inter-laboratory

215

differences are addressed in the Supporting Information.

216

Reaction of Haloacetonitriles in the Presence of Free Chlorine

217

The HAN reaction kinetics were further investigated across three pH levels (i.e., pH 5, 6, and

218

7) in the presence of free chlorine (initial chlorine dose: 0.5 mg Cl2/L ~ 4.0 mg Cl2/L). It is

219

evident in Figure 3 that the presence of free chlorine caused rapid loss of HANs, particularly of

220

the THANs (i.e., TCAN and BDCAN), and the rate of loss accelerated with both increasing pH

221

and increasing chlorine dose. This chlorine-assisted reaction followed many of the trends noted

222

for HAN hydrolysis, with THANs having the highest rate of loss followed by DHANs and finally

223

MHANs under all investigated conditions. Furthermore, within each of the three groups, the

224

greater the extent of bromination, the longer the HAN persisted. As a result, the stability

225

hierarchy for the seven HANs remained the same both with and without the presence of chlorine

226

residual. Perhaps more importantly, such a similarity between HAN hydrolysis and chlorination

227

behavior implies that the underlying reaction mechanisms might be analogous for these two

228

pathways, which will be further explored using liner free energy relationships (LFERs) below.


Page 12 of 26

Page 13 of 26


229 230

Figure 3. Semi-logarithmic plots of residual HAN concentrations versus reaction time under three

231

chlorination pH conditions (i.e., pH 5, 6, and 7) with four different initial free chlorine doses (i.e.,

232

0.5 mg Cl2/L, 1.0 mg Cl2/L, 2.0 mg Cl2/L, and 4.0 mg Cl2/L). The lines indicate the predicted

233

concentrations based on the HAN kinetic model.

234

Despite the fact that theoretical HAN chlorination pathways have already been proposed,16,17

235

the actual reaction intermediates and end products have not to be verified with quantitative

236

laboratory evidence. For this reason, the two exclusively hypothesized intermediates, HAM and

237

HAA, were for the first time, quantified during the course of HAN chlorination at a moderate

238

chlorination pH (i.e., pH 6) in this study. Results are shown in Figure 4 for DCAN and TCAN as

239

two representative compounds of the entire group.

240

Formation of both of the two dichloro- and trichloro-intermediates was observed,

241

compensating the loss of DCAN and TCAN under all chlorination conditions. Mainly due to

242

more substantial HAN loss that was resulted, the higher was the initial chlorine dose, the more

243

was the intermediate formation, especially for DCAA and TCAA. On the other hand, the 13 ACS Paragon Plus Environment


244

concentration of HAMs exhibited a slight decrease at longer reaction times (particularly, TCAM

245

concentration at pH 6 with 4.0 mg Cl2/L initial chlorine dose), which can be ascribed to their own

246

decomposition through reactions with residual chlorine (Yu & Reckhow, unpublished work).

247

More importantly, there was a substantial discrepancy between the molar sum of the three HAN,

248

HAM and HAA species and the initial HAN doses (Figure 4). Such a negative deviation from the

249

mass balance is indicative of the formation of some other intermediates that were not quantified

250

in this study. In a companion study (Yu & Reckhow, unpublished work), we recognized that the

251

intermediate HAMs can be further N-chlorinated by HOCl/OCl-, forming the N-chloro-HAMs,

252

which will tend to deprotonate and stabilize within the pH range typical of drinking water

253

distribution systems (pKa,N-Cl-DCAM=3.71 and pKa,N-Cl-TCAM=2.91).23 Considering the scope of this

254

kinetic study, a full description of the byproduct analysis including the relevant chromatographic

255

and mass spectroscopic evidence for the identification and quantification of this group of

256

halogenated nitrogenous compounds will be presented in a companion paper (Yu & Reckhow,

257

2016; manuscript in preparation). However, it is hypothesized that the deprotonated N-chloro-

258

HAMs other than the HAMs and HAAs may form as the major reaction intermediates during

259

HAN chlorination due to their relatively high stability.


Page 14 of 26

Page 15 of 26


260 261

Figure 4. Formation of DCAM and DCAA (top row), TCAM and TCAA (bottom row) during

262

DCAN and TCAN chlorination at pH 6. Purple diamonds represent the intermediates that were

263

not identified and quantified in this investigation. The dashed lines indicate the initial DCAN and

264

TCAN molar concentrations spiked at the beginning of individual chlorination experiments. Analogous to HAN hydrolysis, the 2nd-order chlorination kinetics can be proposed as follows

265 266

by assuming significant HAN reaction rates with both hypochlorous acid and hypochlorite:

= − − − !" − !" (. 3)

267

To account for the pH-dependent HOCl/OCl- speciation, Equation 3 is reformulated using

268

total free chlorine concentration (i.e., Ct) and dissociation constant Ka24 for hypochlorous acid

269

with corrections for ionic strength (i.e., I): = −( + + $% !& + $' !& ) ∙ (. 4) $% =

) *,, ; $' = ) (. 5) ) + *,, + *,, 15 ACS Paragon Plus Environment


Page 16 of 26

270

In some of the chlorination experiments, significant depletion of chlorine occurred,

271

particularly when the initial chlorine doses were low. As a result, its concentration cannot be

272

treated as constant without introducing substantial error for the estimation of the chlorination

273

reaction rate constants. For this reason, we numerically integrated residual chlorine over time

274

(i.e., /% !& ), which is defined as chlorine contact time (CT) and the final kinetic model can be

275

formulated as Equation 6. The estimation of the individual reaction rate constants using the

276

Bayesian estimation framework will be addressed below.

&

&

ln = ln% − ( + ) ∙ − ($% + $' ) ∙ 2 !& (. 6) %

277

Stability of Haloacetonitriles in the Presence of Chloramines

278

Because of the continuous reaction with chlorine as demonstrated in the above section, HANs

279

are often detected at lower levels in systems that only use free chlorine,20 nevertheless, the total

280

amount of HANs that initially form in those chlorination systems may be substantially greater.17,21

281

However, chloramines have not been assessed for their reactivity with HANs to clarify whether

282

the relatively higher occurrence of HANs in most chloramination systems20 is attributed to their

283

greater stability in the presence of chloramines or to the tendency of chloramines to form more

284

HANs as an additional nitrogen source. Therefore, a set of experiments was conducted for the

285

evaluation of HAN stability at a typical chloramination pH (i.e., pH 8.5) with varying doses of

286

preformed chloramines. The use of preformed chloramines instead of forming chloramines in-situ

287

via ammonia addition was done to prevent HANs from reacting with transient free chlorine

288

before the latter had a chance to fully combine with ammonia. Results indicated that there was no

289

significant difference in HAN stability with or without the presence of chloramines at doses up to

290

4 mg/L (as Cl2) (Figure S3), implying that no reactions between HANs and chloramines were

291

significant enough to be detectable under the investigated conditions.


Page 17 of 26


292

Estimation of Reaction Rate Constants Using Bayesian Framework

293

In this study, the four reaction rate constants (i.e., kH2O, kOH, kHOCl, and kOCl in Equation 6)

294

were estimated in a Bayesian framework, which is an alternative statistic method to the classic

295

least squares regression for the estimation of model parameters. The main benefits of Bayesian

296

estimation are the ability to relax distributional assumptions on parameters, entertain nonlinear

297

model structure and most importantly, pool information between experiments in order to reduce

298

the influence of outliers and thus providing more robust estimations.22 Moreover, the Bayesian

299

framework also allows the inclusion of prior information or expert knowledge when available to

300

reduce the uncertainty in model estimation.25,26

301

The final distribution of all model parameters, known as the joint posterior distribution

302

(Figure S4) shows that the basic hydrolysis rate constant (i.e., kOH) and the hypochlorite

303

chlorination rate constant (i.e., kOCl) for individual HANs were in the same order of magnitude.

304

Moreover, both of the two reaction rate constants ranked in reverse order to the HAN stability

305

hierarchy, increasing with increasing number of halogens, in particular, with the number of

306

chlorine atoms (Table S3). On the contrary, the neutral hydrolysis rate constants (i.e., kH2O) and

307

the hypochlorous acid chlorination rate constants (i.e., kHOCl) were not only estimated to be

308

several orders of magnitude smaller, but also had some fluctuation in terms of following the

309

halogenation pattern. Perhaps most importantly, for all the seven HANs, the kHOCl estimates were

310

normally distributed around zero, suggesting that given the size of the dataset collected in this

311

study, this reaction rate constant proposed in Equation 6 is not estimated to be statistically

312

different from zero. As a consequence, the HAN kinetic model can be reduced to Equation 7 by

313

dropping the HOCl chlorination term, which will in turn leave the data with higher degrees of

314

freedom to allow for more precise estimation of the remaining three reaction rate constants. The

315

resulting joint posterior distribution of kH2O, kOH and kOCl (Figure S5 and Table S4) shows that all

316

of the three reaction rate constants for MBAN were not statistically different from zero due to its 17 ACS Paragon Plus Environment


Page 18 of 26

317

remarkably high stability regardless of pH and chlorine doses. For the other MHAN (i.e., MCAN),

318

only kOH is statistically significant and therefore can be estimated with sufficient precision.

319

Furthermore, the neutral hydrolysis rate constants (i.e., kH2O) for the two THANs (i.e., BDCAN

320

and TCAN) were also noted to be trivial compared to kOH and kOCl. Following the previous

321

methodology, all the reaction rate constants were re-estimated by dropping the insignificant terms

322

to leave the dataset with more freedom. The final posterior estimates of kH2O, kOH, and kHOCl for

323

the seven HANs are listed in Table 1 with 95% confidence intervals. &

ln = ln% − ( + ) ∙ − $' ∙ 2 !& (. 7) %

324

Table 1. Estimates of neutral, basic hydrolysis rate constants (kH2O and kOH), and hypochlorite

325

chlorination rate constant (kOCl) through Bayesian estimation. kH2O (hr-1) Median 95% C.I.

kOH (M-1hr-1) Median 95% C.I.

MBAN NS MCAN NS DBAN 1.38E-04 (0.46, 2.31) E-04 BCAN 1.36E-04 (0.42, 2.35) E-04 DCAN 1.68E-04 (0.66, 2.70) E-04 BDCAN NS TCAN NS 326 *NS – not significant 327

4.14E+01 1.09E+03 2.57E+03 5.60E+03 4.45E+04 1.23E+05

NS (0.89, 7.35) E+01 (1.03, 1.16) E+03 (2.43, 2.72) E+03 (5.29, 5.91) E+03 (4.20, 4.71) E+04 (1.17, 1.31) E+05

kOCl (M-1hr-1) Median 95% C.I.

1.54E+02 3.24E+02 6.85E+02 1.36E+04 3.91E+04

NS NS (1.23, 1.86) E+02 (2.91, 3.58) E+02 (6.40, 7.30) E+02 (1.30, 1.42) E+04 (3.77, 4.06) E+04

Taft Linear Free Energy Relationships (LFERs)

328

The structural impact of reactant on the thermodynamic and kinetic properties of the

329

participating reaction is usually assessed using LFERs. Establishing LFERs also helps in the

330

understanding of reaction mechanisms and allows the prediction of reaction rates on the

331

assumption that compounds with structural similarities behave alike.27-30 The Taft equation

332

(Equation 8) was selected for this dataset because it has been previously used in the evaluation of

333

the substituent inductive and steric effect on the reactivity of aliphatic acetonitriles.15,31,32 In the 18 ACS Paragon Plus Environment

Page 19 of 26


334

Taft equation, K0 is the reaction rate constant for unsubstituted acetonitrile, k is the pathway-

335

specific reaction rate constant for a particular HAN with substituent R. σ* and Es are Taft’s polar

336

and steric substituent constants (the detailed calculations of these two constants are listed in the

337

Supporting Information). The polar and steric sensitivity factors, ρ and δ, are resulting model

338

parameters representing the sensitivity of the reaction rate to the substituent polar and steric

339

properties across the entire group of HANs. Figure 5 shows the ρ and δ estimates based on the

340

estimated median rate constants (Table S4) for two major HAN reactions (i.e., basic hydrolysis

341

and chlorination by hypochlorite), via standard nonlinear least squares regression. log

342

= 78 ∗ + : (. 8) %

Since the product of ρσ* is always greater than that of δEs, it can be inferred that both

343

reactions are more sensitive to the polar than to the steric property of the halogenated substituents.

344

For this reason, the higher hydrolysis and chlorination rates for more halogenated HANs can be

345

explained by the higher electron-withdrawing effect from the halogen aggregate, which activates

346

the nitrile carbon and renders it more electrophilic, even though the steric hindrance of the

347

aggregate also increases with increasing number of halogens. Perhaps most importantly, positive

348

ρ estimates for both of the two LFERs reveal the same nucleophilic nature of HAN hydrolysis

349

and chlorination reactions; that is, HANs react with hydroxide and hypochlorite through

350

nucleophilic attacks on the nitrile carbon. This also explains the absence of certain HAN

351

chloramination reactions since chloramines are, in most cases, strong electrophiles.



352 353

Figure 5. Taft LFERs based on median kOH and kOCl estimates.

354

Implications of HAN Reaction Kinetics with Respect to Drinking Water Treatment and System

355

Management

356

With the understanding of HAN reaction kinetics, the persistence of this group of compounds

357

in drinking water distribution systems can be predicted based on distributed water pH and

358

disinfectant residual. When exact chlorine exposure (i.e., CT) during drinking water distribution

359

is not readily available, an averaged chlorine residual !

Kinetic Analysis of Haloacetonitrile Stability in Drinking Waters

Recommend Documents