On-the-Fly Kinetic Monte Carlo Simulation of Aqueous Phase

Subscriber access provided by UNIV OF MISSISSIPPI

Article

On-the-fly Kinetic Monte Carlo Simulation of Aqueous Phase Advanced Oxidation Processes Xin Guo, Daisuke Minakata, and John Crittenden Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.5b02034 • Publication Date (Web): 01 Jul 2015 Downloaded from http://pubs.acs.org on July 13, 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 27

Environmental Science & Technology

1

On-the-fly Kinetic Monte Carlo Simulation of Aqueous Phase Advanced Oxidation

2

Processes

3 4

Prepared for Environmental Science and Technology

5

6

7 8

Xin Guo1

9

Daisuke Minakata2

10

John Crittenden1*

11 12 13 14 15 16

1. School of Civil and Environmental Engineering, Georgia Institute of Technology, 828 West Peachtree Street, Atlanta, GA 30332 2. Department of Civil and Environmental Engineering, Michigan Technological University, 1400 Townsend Drive, Houghton, MI, 49931

17

18

*Corresponding

19

email: [email protected]

author phone: 404-894-5676

fax: 404-894-7896

20

ACS Paragon Plus Environment


21

Abstract

22

We have developed an on-the-fly kinetic Monte Carlo (KMC) model to predict the degradation

23

mechanisms and fates of intermediates and byproducts that are produced during aqueous phase advanced

24

oxidation processes (AOPs). The on-the-fly KMC model is comprised of a reaction pathway generator, a

25

reaction rate constant estimator, a mechanistic reduction module, and a KMC solver. The novelty of this

26

work is that we develop the pathway as we march forward in time rather than developing the pathway

27

before we use the KMC method to solve the equations. As a result, we have fewer reactions to consider

28

and we have greater computational efficiency.

29

We have verified this on-the-fly KMC model for the degradation of polyacrylamide (PAM) using

30

UV light and titanium dioxide (i.e., UV/TiO2). Using the on-the-fly KMC model, we were able to predict

31

the time-dependent profiles of the average molecular weight for PAM. The model provided detailed and

32

quantitative insights into the time evolution of the molecular weight distribution and reaction mechanism.

33

We also verified our on-the-fly KMC model for the destruction of (1) acetone, (2) trichloroethylene

34

(TCE), and (3) polyethylene glycol (PEG) for the ultraviolet light/hydrogen peroxide AOP. We

35

demonstrated that the on-the-fly KMC model can achieve the same accuracy as the computer-based first-

36

principles KMC (CF-KMC) model, which has already been validated in our earlier work. The on-the-fly

37

KMC is particularly suitable for molecules with large molecular weights (e.g., polymers) because the

38

degradation mechanisms for large molecules can result is 100’s of thousands to even millions of reactions.

39

And the ordinary differential equations (ODEs) that describe the degradation pathways cannot be solved

40

using traditional numerical methods, but the KMC can solve these equations.

41

Introduction

42

Advanced oxidation processes (AOPs) have been used for degrading recalcitrant contaminants

43

into biodegradable organic compounds or into carbon dioxide and mineral acids in water.1 However,

44

AOPs are mechanistically complex in nature, and numerous intermediates and byproducts are produced

45

during the treatment processes. Developing method to predict the formation of intermediates and


Page 2 of 27

Page 3 of 27


46

byproducts are important for engineering AOPs that are more effective because some of these

47

intermediates and byproducts [e.g., monochloroacetic acid and dichloroacetic acid produced by the

48

degradation of trichloroethylene (TCE)] may pose potential risks to human health.2,3 In addition, some of

49

these intermediates and byproducts (e.g., volatile acids) may require longer reaction times to destroy.

50

Consequently, we need to understand the detailed degradation mechanisms and fate of intermediates and

51

byproducts during AOPs so that we can design AOPs that reduce toxicity.

52

Various studies have investigated the degradation mechanisms of AOPs.4-10 Although these

53

studies have shed light on the detailed elementary reactions and the radical pathways in AOPs, these

54

studies are limited in the following aspects. First, experimental studies that determine the degradation

55

mechanisms are time consuming, especially for larger molecules. In addition, these experimental studies

56

would be cost prohibitive if we were to investigate the degradation pathways of all compounds that are

57

used in commerce (e.g., thousands of organic chemical compounds are produced annually and they could

58

end up in the environtment).11 Second, the kinetic models that were developed in these studies used

59

lumped reactions for simplicity, which prevent us from obtaining a detailed insight into the degradation

60

process and predicting the degradation mechanisms for newly discovered organic compounds in the

61

environment. Third, these kinetic models also require numerical methods to solve the ordinary differential

62

equations (ODEs), which might be too stiff to be solved for complicated reaction pathways. For example,

63

the degradation mechanism of polyethylene glycol (PEG with a MW of 3600 Dalton) in the UV/H2O2

64

process includes 522,057 species and 696,183 reactions, which might not be solved by most of the ODE

65

solvers.12

66

To overcome these limitations, Guo et al.12 have developed a computer-based first-principles

67

kinetic Monte Carlo (CF-KMC) model to simulate organic compound degradation in AOPs. The CF-

68

KMC model uses a computer algorithm that can predict the degradation pathways in AOPs for a given

69

parent compound. These predicted degradation pathways consist of elementary reactions, in contrast to

70

the lumped reactions that are used by traditional kinetic models. In addition, the CF-KMC model uses a

71

KMC solver to solve the degradation mechanisms without solving ODEs. Hence, difficulties such as



72

stiffness encountered in traditional ODE-based kinetic models are avoided. The CF-KMC model has

73

successfully simulated the degradation of various parent compounds, including low molecular weight

74

contaminants (e.g., acetone and TCE) and large contaminants (e.g., PEG). However, since the

75

computational time for the CF-KMC model is proportional to the number of reactions that are included in

76

the mechanism,13 the computational efficiency of the CF-KMC model for simulating large degradation

77

mechanism can be improved by reducing the number of reactions using an on-the-fly mechanism

78

generation.

79

The on-the-fly KMC model can predict and solve the degradation mechanism simultaneously,

80

rather than elucidating the full mechanism before calling the KMC solver to solve the equations. This on-

81

the-fly strategy can significantly decrease the number of ODEs that are solved by the KMC solver, and,

82

hence, it can increase the computational efficiency. In addition, the on-the-fly KMC model also uses a

83

mechanistic reduction module that can remove unimportant reactions that do not contribute significantly

84

to the overall degradation rate, which can further increase the computational efficiency.

85

In this study, an on-the-fly KMC model is developed and verified by comparing model

86

predictions to the degradation products of polyacrylamide (PAM) using the UV/TiO2 process. Detailed

87

mechanisms for the degradation of PAM and fates of intermediates and byproducts are generated and

88

time-dependent profiles of number averaged molecular weight (Mn) and molecular weight distribution

89

(MWD) were predicted. The on-the-fly KMC model is validated by comparing the model predictions to

90

experimental data. We also verified that the on-the-fly KMC model agrees with the model results that

91

were obtained with the KMC method and a numerical solution to the ODEs for the degradation of acetone,

92

TCE, and PEG in the UV/H2O2 process. These simulation results achieved the same accuracy as

93

compared with the simulation results obtained by the CF-KMC method and numerical solution to the

94

ODEs. The on-the-fly KMC model developed in this study represents a major step forward in predicting

95

the degradation mechanisms and fates of degradation products for a wide range of organic contaminants

96

with various sizes and functional groups.


Page 4 of 27

Page 5 of 27

97


Methods

98

Figure 1 shows the general structure of the on-the-fly KMC model. For the first time step there is

99

only one parent component. For subsequent time steps, the by-products that are formed during the given

100

time step are added to the parent compound list or species pool, which is the collection of all species that

101

exist at some point in time. Then, the on-the-fly KMC model iteratively runs the following four modules

102

for each time step.

103

The first module is a pathway generator14 that generates the elementary reactions that are

104

included in the degradation pathway. For each time step, the pathway generator predicts the products

105

from the existing species pool (i.e., products that can be directly produced from the existing species pool

106

for the next elementary reaction step). Then, the pathway generator adds all newly generated products and

107

reactions into the species pool and reaction list.

108

The second module estimates the reaction rate constant for each newly predicted reaction. In this

109

study, we used the Group Contribution Method (GCM)15 to estimate the second order reaction rate

110

constants for hydroxyl radical reactions. For other reactions, the rate constants were either obtained from

111

literature-reported values or estimated based on similar reactions reported in the literature (see SI),

112

because there are a limited number of experimentally determined rate constants in literature. Quantum

113

chemical calculations have been frequently and successfully employed to predict the degradation

114

pathways and reaction kinetics for HO•-initiated reactions in aqueous phase but we did not use them in

115

this study because they can not be used for large molecules and linear free energy relationships need to be

116

developed for a wider variety of compounds and reacions. .16-19 Several robust tools based on the quantum

117

chemical calculation, such as Linear Free Energy Relationships (LFERs),20-22 have also been developed to

118

estimate reaction rate constants for various reactions for aqueous phase AOPs, including hydroxyl radical

119

reactions, oxygen addition to carbon-centered radicals, disproportionation of peroxyl radicals, and

120

unimolecular decay of peroxyl radicals. The LFERs can overcome the limitations of the GCM, such as

121

availability of data for all possible functional groups, averaging of the impact of functional groups,



122

additivity of rate constants, disregard of electronic intra- and inter-molecular effects, hydrogen bonding

123

effects, steric effects, and solvation effects.20 However, we did not use them in this study because the

124

LFERs for some reactions (e.g., oxyl radicals) could not be developed due to the little availability of the

125

experimentally determined rate constants. Accordingly, LFERs will be integrated into the on-the-fly

126

KMC model in the future.

127

The third module is a mechanistic reduction algorithm that can eliminate unimportant reactions

128

and species from the newly generated pathways and improve the computational efficiency of the on-the-

129

fly KMC model. Briefly speaking, the unimportant reactions are those that do not significantly contribute

130

to the overall reaction rate loss of a given species. In this study, we used the Directed Relation Graph

131

(DRG) method with a criterion of 10-3 for the mechanistic reduction. This means that any reactions that

132

contribute less than 0.1% to the degradation of a compound will be eliminated. The detailed description

133

about DRG can be found in Lu et al.19 and Guo et al.24

134

The last module is a KMC solver that can predict species concentrations at the next time step

135

without generating and solving ODEs that describe the degradation pathway. At each time point, the

136

KMC solver selects a given reaction to occur from the reaction list and updates the concentrations of

137

species that are involved in this selected reaction. Then, the KMC solver calculates the time interval for

138

the next time point. The detailed description of the KMC solver can be found in Gillespie25 and Guo et

139

al.12

140

The four modules described above will be executed iteratively for each time step until the desired

141

simulation time is reached. Then the stored values of the concentrations at various times can be plotted.

142

Results and Discussion

143

Validation of the on-the-fly KMC model

144

In the Supporting Information, we validated our on-the-fly KMC model by comparing simulation

145

results that were obtained by the on-the-fly KMC model and the CF-KMC model for the degradation of

146

acetone, TCE, PEG in a UV/H2O2 process. We demonstrated that the on-the-fly KMC model can achieve


Page 6 of 27

Page 7 of 27


147

the same accuracy as the CF-KMC model in terms of concentration profiles of degradation products.

148

Previously, we demonstrated that the CF-KMC gave identical results to the solution of the ODEs using

149

numerical methods for the degradation of acetone and TCE in a UV/H2O2 process.12 Table 1 compares the

150

computational efficiency of the on-the-fly KMC model with the CF-KMC model. We found that for large

151

parent compounds (i.e., PEG and polyacrylamide (PAM)), the on-the-fly KMC model generally saves

152

60%-70% CPU time as compared with the CF-KMC model. This is because that the on-the-fly KMC

153

model solves only the degradation mechanism for each time step and does not have to solve the equations

154

for the complete mechanism. Hence, the on-the-fly KMC model solves a smaller reaction pathway

155

network as compared with the full reaction pathway network that is solved by the CF-KMC model.

156

However, for small parent compounds (i.e., acetone and TCE), the computational efficiencies of both on-

157

the-fly KMC model and CF-KMC model are about the same. This is because the number of reactions in

158

the degradation mechanisms for small parent compounds is not large. As a result, the on-the-fly KMC

159

model actually solves the same-sized mechanism as compared with the full mechanism solved by the CF-

160

KMC model for most of the simulation time. Consequently, the on-the-fly KMC model cannot

161

significantly increase the computational efficiency in this case.

162

Mechanism generation for the degradation of PAM

163

The degradation mechanism of PAM in the UV/TiO2 process was generated by the on-the-fly

164

KMC model. The experimental conditions are described by Vijayalakshmi et al.26 The initial

165

concentration and averaged molecular weight of PAM were 0.012 mM and 1.64 × 105 g/mol, respectively.

166

The wavelength of UV light was predominantly 365 nm and the light intensity was 1.19 × 10-5

167

Einstein/L·s. Commercial Degussa P-25 TiO2 was used as the photocatalyst and the quantum yield for

168

HO• formation was reported to be 0.04.27 The reactor was a completely mixed batch reactor (CMBR).

169

The elementary reactions in the degradation mechanism of PAM were generated by the pathway

170

generator as shown in Table SI 1 of the Supporting Information. These elementary reactions cover most

171

of reaction types that have been discovered to occur during aqueous phase AOPs, including: (1) hydroxyl



172

radical H-atom abstraction that produces carbon-centered radicals, (2) oxygen addition to carbon-centered

173

radicals, (3) beta(β) scission of oxyl radicals, and (4) hydrolysis of aldehydes. The pathway also includes

174

some overall reactions, such as bimolecular decays of peroxyl radical, because experimental studies have

175

not elucidated the elementary reactions. In addition, the current version of the pathway generator does not

176

include the electron-transfer reactions between hydroxyl radicals and amide groups. As a result, this

177

reaction was manually added into the pathway generator. The current version of the pathway generator

178

also does not include the cross-disproportionation of peroxyl radicals since the reaction rates of these

179

reactions are much slower than competing reactions (e.g. self-disproportionation of peroxyl radicals).28 In

180

addition, we did not consider the adsorption of PAM on TiO2 particles since it was not observed by the

181

experiments conducted by Vijayalakshmi et al.26

182

The reaction rate constants were obtained primarily by two methods: 1) estimation based on the

183

literature reported experimental values and 2) calculations from the GCM. The GCM can estimate

184

hydroxyl radical reaction rate constants with an uncertainty within 0.5-2 times of the experimental values.

185

The reaction rate constants that could not be obtained by the above two methods were estimated based on

186

similar reactions that have been experimentally observed. Table SI 1 in the Supporting Information

187

contains the values of all the reaction rate constants and how they were obtained or estimated. We

188

investigated the impact of the uncertainty of reaction rate constants on the on-the-fly KMC model

189

simulations. Accordingly, we conducted a sensitivity analysis and found that reaction rate constants with

190

large uncertainties did not have significant impact on the accuracy of the on-the-fly KMC model

191

simulations. The detailed description about the sensitivity analysis process can be found in the Supporting

192

Information.

193

The pathway was reduced by the DRG method with a criterion of 0.1%, which means the DRG

194

method removed reactions that had rates smaller than 0.1% of overall consumption rate of a species of

195

interest for each time step. The reactions removed were mainly H-atom abstractions by carbon-centered

196

radicals from a C-H bond. The reduced mechanism was solved using the KMC solver with an initial


Page 8 of 27

Page 9 of 27


197

population of 108. Details of DRG method is discussed in the Supporting Information. The process of

198

determining initial population for the KMC solver can be found in authors’ previous articles.12,24

199

Predicted Reaction Pathways of PAM Degradation

200

Initial stage of the degradation of PAM

201

Figure 2 shows the predicted degradation pathway of PAM in the UV/TiO2 process. As the first

202

step, the backbone of PAM molecule has two positions that can be attacked by hydroxyl radicals via H-

203

atom abstraction reaction: 1) the hydrogen atom on the α-carbon (i.e. H-abstraction A in Figure 2) and 2)

204

the hydrogen atom on the β-carbon (i.e. H-abstraction B in Figure 2). The ratio of the contributions of

205

these two pathways is 1:3 as estimated by the GCM. These two pathways produce two different carbon-

206

centered radicals, which have radical sites located at α-carbon and β-carbon, respectively. In addition, the

207

H-atom abstraction of hydrogen atom from the –CO-NH2 group with hydroxyl radical is negligible as

208

reported by Karpel Vel Leitner et al.29

209

Degradation of α-carbon centered radical

210

The α-carbon centered radical that is generated by the H-atom abstraction of PAM (i.e., H-

211

abstraction A in Figure 2) produces oxyl radical, ─ CH2─C(CONH2)(O•)─CH2─CH(CONH2)─, through

212

the oxygen addition followed by the bimolecular decay of peroxyl radicals. This oxyl radical undergoes

213

the β-scission reaction and generates two sub-chains ending with a O=C(CONH2)─ group and carbon-

214

centered radical group, namely,, O=C(CONH2)─CH2─CH(CONH2)─ and ─CH(CONH2)─CH2•. On one

215

hand, O=C(CONH2)─CH2─CH(CONH2)─ is attacked by hydroxyl radicals and is further degraded into

216

low molecular weight products (LMWPs). On the other hand, ─CH(CONH2)─CH2• rapidly reacts with

217

O2 to form a end-peroxyl radical, ─CH(CONH2)─CH2OO•.

218

The end-peroxyl radical, ─CH(CONH2)─CH2OO•, is consumed through three bimolecular decay

219

channels (i.e., bimolecular decays that are shown as A, B, and C in Figure 2). First,

220

─CH(CONH2)─CH2OO• goes through the Russell reaction30 (i.e., bimolecular decay C in Figure 2) to

221

produce end-hydroxyl group, which is further degraded to end-aldehyde group via the reaction with HO•.



Page 10 of 27

222

Second, ─CH(CONH2)─CH2OO• also directly produces end-aldehyde group (i.e., bimolecular decay B

223

in Figure 2), which is hydrolyzed later to form a carboxylic acid group at the end of the molecule. The

224

end-carboxylic acid group undergoes random attack by HO• again, which finally leads to the formation of

225

various LMWPs. Third, ─CH(CONH2)─CH2OO• produces end-oxyl radical, ─CH(CONH2)─CH2O• (i.e.,

226

bimolecular decay A in Figure 2). The contributions of these three bimolecular decay channels are same.

227

The end-oxyl radical, ─CH(CONH2)─CH2O•, decays through unimolecular fragmentation by

228

C─C bond scission to form formaldehyde and end-carbon-centered radical, ─CH2─•CH(CONH2), which

229

is further degraded to end-peroxyl radical, ─CH2─CH(CONH2)OO•. ─CH2─CH(CONH2)OO• is

230

consumed by three bimolecular decay channels (i.e., bimolecular decay D, E, and F in Figure 2) to

231

produce end-hydroxyl group (─CH2─CH(CONH2)(OH)), end-acyl group (─CH2─CO─ (CONH2)), and

232

end-oxyl radical group (─CH2─CH(CONH2)(O•)). On one hand, ─CH2─CH(CONH2)(OH) and

233

─CH2─CO─ (CONH2) are attacked by hydroxyl radicals and finally degraded to LMWPs. On the other

234

hand, ─CH2─CH(CONH2)(O•) undergoes β-scission to generate OHCCONH2 and ─CH(CONH2) ─CH2•.

235

The degradation pathway of ─CH(CONH2) ─CH2• has been described above.

236

Degradation of β-carbon centered radical

237

The β-carbon centered radical, ─CH2─CH(CONH2)─•CH─CH(CONH2)─, that is formed by H-

238

atom abstraction of PAM (i.e., H-abstraction B in Figure 2) reacts with oxygen to produce peroxyl radical,

239

─CH2─CH(CONH2)─CH(OO•)─CH(CONH2)─. This peroxyl radical undergoes three degradation

240

channels, which forms three products:

241

(1) inner-hydroxyl group, ─CH2─CH(CONH2)─CH(OH)─CH(CONH2)─,

242

(2) inner-ketone group, ─CH2─CH(CONH2)─C(=O)─CH(CONH2)─, and

243

(3) inner-oxyl radical, ─CH2─CH(CONH2)─CH(O•)─CH(CONH2)─.

244

The inner-hydroxyl group is further degraded to inner-ketone group, which is attacked by hydroxyl

245

radicals again to finally form various LMWPs. The inner-oxyl radical undergoes β-scission to produce


Page 11 of 27


246

─CH2─•CH(CONH2) and OHC─CH(CONH2)─, whose degradation pathways have already been

247

described above.

248

Simulation Results of PAM Degradation

249

Figure 3 compares the calculated profile of number averaged molecular weight (M n) with

250

experimental data reported by Vijayalakshmi et al.25 The simulation results are consistent with the

251

experimental data. The results showed that Mn decreases from 1.64 × 105 g/mol to 5 × 104 g/mol after 200

252

minutes, which indicates the long chain PAM molecules are degraded into short chain oligomers during

253

the UV/TiO2 degradation process.

254

To investigate into the degradation process and associated mechanisms, we simulated the time

255

evolution of the molecular weight distribution (MWD) of the PAM degradation as shown in Figure 4. The

256

results from the simulation indicate a uniform and large molecular weight at the beginning (i.e., t = 0). As

257

the degradation proceeds, the peak of the MWD shifts from large molecular weight to small molecular

258

weight, which is consistent with the trend that is indicated by the experimental Mn profile.

259

Figure 5 shows the time evolution of the total number of reactions that are included. At the initial

260

stage (i.e., first 25 min), the number of reactions increases rapidly. Most of the reactions that are

261

generated at this stage are H-atom abstraction reactions occurred on the backbone of PAM. At the second

262

stage (i.e., 25 min to 100 min), the growth rate of the number of reactions slows down and the number of

263

reactions finally converges to around 106. During this stage, the generated reactions are majorly consisted

264

of oxygen addition reaction, bimolecular decay reaction, β scission reaction and so forth. At the final

265

stage (i.e., 100 min to 200 min), the cumulative number of reactions evaluated keep constant around 106,

266

which indicates the degradation pathway is nearly complete.

267

The major low molecular weight products (LMWPs) predicted by the on-the-fly KMC model

268

included formic acid and oxamic acid. The on-the-fly KMC model also predicted minor LMWPs

269

including formaldehyde and glyoxylamid. We also calculated the concentration profiles of major low



270

molecular weight products (i.e., formic acid and oxamic acid) and these profiles are shown in the

271

Supporting Information.

272

Environmental implications

273

For simulating AOPs in natural waters, the on-the-fly KMC model is able to predict the impact

274

of pH and alkalinity on the degradation process. The detailed information about how we implement the

275

impact of pH and alkalinity can be found in authors’ previous article.31 In addition, the current version of

276

the on-the-fly KMC simulates the impact of natural organic matter (NOM) on AOPs by accounting for

277

NOM quenching of hydroxyl radical and UV light absorption. However, the reactions between NOM and

278

radical species are complex32 and more NOM reactions may be included in the future if knowledge of the

279

structural and chemical characteristics of NOM become available.

Page 12 of 27

280

The current version of the pathway generator in the on-the-fly KMC model can predict

281

degradation mechanisms for a wide range of organic contaminants based on the known reaction rules

282

discovered from the past experimental observations. These reaction rules include hydrogen-atom

283

abstraction reaction from a C-H bond or O-H bond, HO• addition reaction to a C=C bond of an aliphatic

284

compound, oxygen addition reaction to organic radicals, bimolecular decay of peroxyl radical reaction,

285

HO2• elimination reaction, β scission reaction, 1,2-H shift reaction, hydrolysis reaction and so forth.

286

However, the pathway generator still does not include some reaction rules that are specific to certain

287

functional groups (e.g. S-, N-, or P-atom-containing groups, amide group, and benzene ring). In the future,

288

the on-the-fly KMC model could predict the degradation mechanisms of more organic compounds by

289

adding more reaction rules into the pathway generator, once these reaction rules have been experimentally

290

determined.

291

Acknowledgement

292

This work was supported by National Science Foundation Award 0854416. The authors also

293

appreciate support from the Brook Byers Institute for Sustainable Systems, Hightower Chair and Georgia


Page 13 of 27


294

Research Alliance at Georgia Institute of Technology, and the startup funds from the Michigan

295

Technological University.

296

Supporting Information

297

The supporting information provides tables listing all reactions and corresponding rate constants

298

included in the generated degradation mechanism of PAM. The detailed description of the sensitivity

299

analysis results, mass balance analysis, and validation of the on-the-fly KMC model are also included.

300

This material is available free of charge via the Internet at http://pubs.acs.org.

301



302

TOC art

303


Page 14 of 27

Page 15 of 27


304

Table Captions

305

Table 1. Comparison of CPU times for the on-the-fly KMC model and the CF-KMC model. The

306

processor used is 64-bit 2.4 GHz Intel® Core™ 2 Duo CPU

307 308 309



310

Table 1. Parent Compound Acetone TCE PEG PAM

CPU Time for onthe-fly KMC model 18 sec 13 sec 12 min 20 min

CPU Time for CFKMC model 17 sec 13 sec 30 min 65 min

311


Page 16 of 27

Page 17 of 27


312

Figure Captions

313

Figure 1. Overall structure of the on-the-fly KMC model.

314 315

Figure 2. Generated degradation mechanism of PAM in the UV/TiO2 process.

316 317

Figure 3. Comparison of the on-the-fly KMC model calculations to experimental data for the number

318

averaged molecular weight.

319 320

Figure 4. Calculated time evolution of molecular weight distribution for the degradation of PAM during

321

the UV/TiO2 process. The bars are the calculated mass fractions for polymers with various molecular

322

weights.

323 324

Figure 5. Time evolution of the cumulative number of reactions evaluated for the degradation of PAM

325

during the UV/TiO2 process.



Input of parent compound(s) or Species Pool at time t

Reaction pathway generator

Reaction rate constant estimator

Mechanistic reduction module

KMC solver If t < simulation time If t =simulation time

Plot concentration versus time profiles for all species 326 327

Figure 1


Page 18 of 27

Page 19 of 27


328 329

Figure 2



Page 20 of 27

1.8 Experiment

Averaged molecular weight (105 g/mol)

1.6

On-the-fly KMC 1.4 1.2 1 0.8 0.6 0.4 0.2 0 0

330 331

20

40

60

80 100 120 Time (min)

Figure 3


140

160

180

200

Page 21 of 27


Mass fraction

1

t=0

0.8 0.6 0.4 0.2 0 0

0.15

0.3

0.45

0.6

0.75

0.9

1.05

1.2

1.35

1.5

1.65

Molecular weight (105 g/mol)

332

Mass fraction

0.4

t = 50 min

0.3 0.2 0.1 0 0

0.15

0.3

0.45

0.6

0.75

0.9

1.05

1.2

1.35

1.5

1.65


333

Mass fraction

0.1

t = 150 min

0.08 0.06 0.04 0.02 0 0

0.15

0.3

0.45

0.6

0.75

0.9

1.05

1.2

1.35

1.5

1.65


334

Mass fraction

0.05

t = 200 min

0.04 0.03 0.02 0.01 0 0

0.3

0.45

0.6

0.75

0.9

1.05

Molecular weight

335 336

0.15

1.2

(105

1.35

g/mol)

Figure 4


1.5

1.65


Page 22 of 27

1.2

Number of reactions (106)

1 0.8 0.6

0.4 0.2 0

0 337 338

50

100 Time (min)

Figure 5


150

200

Page 23 of 27

339 340 341 342


Literature Cited 1. Glaze, W. H.; Kang, J. W.; Chapin, D. H. The chemistry of water treatment processes involving ozone, hydrogen peroxide and UV radiation. Ozone: Sci. Eng. 1987, 9, 335–352. 2. Rosenfeldt, E.J.; Linden, K.G. Degradation of endocrine disrupting chemicals bisphenol

343

A, ethinyl estradiol, and estradiol during UV photolysis and advanced oxidation

344

processes. Environ. Sci. Technol. 2004, 38, 5476-5483.

345

3. Huber, M.M.; Canonica S.; Park G-Y.; von Gunten U. Oxidation of pharmaceuticals

346

during ozonation and advanced oxidation processes. Environ. Sci. Technol. 2003, 37,

347

1016-1024.

348

4. Li, K.; Stefan, M.I.; Crittenden, J.C. Trichloroethene degradation by UV/H2O2 advanced

349

oxidation process: product study and kinetic modeling. Environ. Sci. Technol. 2007, 41,

350

1696-1703.

351

5. Stefan, M.I.; Hoy, A.R.; Bolton, J.R. Kinetics and mechanism of the degradation of

352

acetone in dilute aqueous solution sensitized by the UV photolysis of hydrogen peroxide.

353

Environ. Sci. Technol. 1996, 30, 2382-2390.

354

6. Stefan, M.I.; Bolton, J.R. Reinvestigation of the acetone degradation mechanism in dilute

355

aqueous solution by the UV/H2O2 process. Environ. Sci. Technol. 1999, 33, 870-873.

356

7. Stefan, M.I.; Bolton, J.R. Mechanism of the degradation of 1,4-dioxane in dilute aqueous

357

solution using the UV/Hydrogen peroxide process. Environ. Sci. Technol. 1998, 32,

358

1588-1595.

359 360

8. Stefan, M.I.; Mack, J.; Bolton, J.R. Degradation pathways during the treatment of methyl tert-butyl ether by the UV/H2O2 process. Environ. Sci. Technol. 2000, 34, 650-658.



361

9. Cooper, W.J.; Cramer, C.J.; Martin, N.H.; Mezyk, S.P.; O’Shea, K.E.; von Sonntag, C.

362

Free radical mechanisms for the treatment of methyl tert-butyl ether (MTBE) via

363

advanced oxidation/reductive processes in aqueous solutions. Chem. Rev. 2009, 109,

364

1302-1345.

365

10. Santos, L.C.; Poli, A.L.; Cavalheiro, C.C.S.; Neumann, M.G. The UV/H2O2 –

366

photodegradation of poly(ethyleneglycol) and model compounds. J. Braz. Chem. Soc.

367

2009, 20, 1467-1472.

368 369 370

11. Richardson, S. D. Water Analysis: Emerging contaminants and current issues. Anal. Chem. 2009, 81, 4645–4677. 12. Guo, X.; Minakata, D.; Crittenden, J. Computer-based first-principles kinetic Monte Carlo

371

simulation of polyethylene glycol degradation in aqueous phase UV/H2O2 advanced

372

oxidation processes. Environ. Sci. Technol. 2014, 48, 10813-10820.

373 374 375

13. Yang, J; Hlavacek, W.S. The efficiency of reactant site sampling in network-free simulation of rule-based models for biochemical systems. Phys. Biol. 2011, 8, 055009. 14. Li, K.; Crittenden, J. Computerized pathway elucidation for hydroxyl radical-induced

376

chain reaction mechanisms in aqueous phase advanced oxidation processes. Environ. Sci.

377

Technol. 2009, 43, 2831-2837.

378

15. Minakata, D.; Li, K.; Westerhoff, P.; Crittenden, J. Development of a group contribution

379

method to predict aqueous phase hydroxyl radical (HO•) reaction rate constants. Environ.

380

Sci. Technol. 2009, 43, 6220-6227.

381 382

Page 24 of 27

16. Fang, H.; Gao, Y.; Li, G.; An, J.; Wong, P.; Fu, H.; Yao, S.; Nie, X.; An, T. Advanced oxidation kinetics and mechanism of preservative propylparaben degradation in aqueous


Page 25 of 27


383

suspension of TiO2 and risk assessment of its degradation products. Environ. Sci. Technol.,

384

2013, 47, 2704-2712.

385

17. Gao, Y.; Ji, Y.; Li, G.; An, T. Mechanism, kinetics and toxicity assessment of OH-

386

initiated transformation of triclosan in aquatic environments. Water Res., 2014, 49, 360-

387

370.

388

18. An, T.; Gao, Y.; Li, G.; Kamat, P.; Peller, J.; Joyce, M. Kinetics and mechanism of •OH

389

mediated degradation of dimethyl phthalate in aqueous solution: experimental and

390

theoretical studies. Environ. Sci. Technol. 2014, 48, 641-648.

391

19. Leopoldini, M.; Chiodo, S.; Russo, N.; Toscano, M. Detailed investigation of the OH

392

radical quenching by natural antioxidant caffeic acid studied by quantum mechanical

393

models. J. Chem. Theory and Comput. 2011,

394

20. Minakata, D.; Crittenden, J. Linear free energy relationships between aqueous phase

395

hydroxyl radical reaction rate constants and free energy of activation. Environ. Sci.

396

Technol. 2011, 45, 3479-3486.

397

21. Minakata, D.; Song, W.; Crittenden, J. Reactivity of aqueous phase hydroxyl radical with

398

halogenated carboxylate anions: experimental and theoretical studies. Environ. Sci.

399

Technol. 2011, 45, 6057-6065.

400

22. Minakata, D.; Mezyk, S.P.; Jones, J.W.; Daws, B.R.; Crittenden, J.C. Development of

401

linear free energy relationships for aqueous phase radical-involved chemical reactions.

402

Environ. Sci. Technol. 2014, 48, 13925-13932.

403 404

23. Lu, T.; Law, C.K. A directed relation graph method for mechanism reduction. Proc. Combust. Inst. 2005, 30, 1333-1341.



405

24. Guo, X.; Minakata, D.; Niu, J.; Crittenden, J. Computer-based first-principles kinetic

406

modeling of degradation pathways and byproduct fates in aqueous-phase advanced

407

oxidation processes. Environ. Sci. Technol. 2014, 48, 5718-5725.

408 409 410 411 412

Page 26 of 27

25. Gillespie, D.T. Exact stochastic simulation of coupled chemical reactions. J. Phys. Chem. 1977, 81, 2340-2361. 26. Vijayalakshmi, S.P.; Madras, G. Photocatalytic degradation of poly(ethylene oxide) and polyacrylamide. J. Appl. Polym. Sci. 2006, 100, 3997-4003. 27. Sun, L.; Bolton, J.R. Determination of the quantum yield for the photochemical generation

413

of hydroxyl radicals in TiO2 suspensions. J. Phys. Chem. 1996, 100, 4127-4134.

414

28. Von Sonntag, C.; Schuchmann, H-P. The elucidation of peroxyl radical reactions in

415

aqueous solution with the help of radiation-chemical methods. Angew. Chem. Int. Ed. Engl.

416

1991, 30, 1229-1253.

417

29. Karpel Vel Leitner, N.; Berger, P.; Legube, B. Oxidation of amino groups by hydroxyl

418

radicals in relation to the oxidation degree of the α-carbon. Environ. Sci. Technol. 2002,

419

36, 3083-3089.

420

30. Russell, G.A. Deuterium-isotope effects in the autoxidation of aralkyl hydrocarbons –

421

mechanism of the interaction of peroxy radicals. J. Am. Chem. Soc. 1957, 79, 3871-3877.

422

31. Crittenden, J. C.; Hu, S.; Hand, D. W.; Green, S. A kinetic model for H2O2/UV process in

423

a completely mixed batch reactor. Water Res. 1999, 33, 2315-2328.

424

32. Westerhoff, P.; Aiken, G.; Amy, G.; Debroux, J. Relationships between the structure of

425

natural organic matter and its reactivity towards molecular ozone and hydroxyl radicals.

426

Wat. Res. 1999, 33, 2265-2276.


Page 27 of 27


82x43mm (300 x 300 DPI)


On-the-Fly Kinetic Monte Carlo Simulation of Aqueous Phase

Recommend Documents