development and evaluation of photocatalytic linear ... · stephanie gora doctor of philosophy...

Development and Evaluation of Photocatalytic Linear Engineered Titanium Dioxide Nanomaterials for the Removal of Disinfection

Byproduct Precursors from Drinking Water

by

Stephanie Gora

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy

Department of Civil Engineering University of Toronto

© Copyright by Stephanie Gora 2017

ii

Development and Evaluation of Photocatalytic Linear Engineered

Titanium Dioxide Nanomaterials for the Removal of Disinfection

Byproduct Precursors from Drinking Water

Stephanie Gora

Doctor of Philosophy

Department of Civil Engineering

University of Toronto

2017

Abstract

Photocatalysis has long been touted as a potential drinking water treatment technology but has

proven difficult to implement at full scale. This project aimed to address two of the perennial

challenges preventing the use of photocatalysis for drinking water treatment: the need to safely

remove the photocatalyst from the water after treatment and the danger that incomplete

mineralization of contaminants will lead to the formation of intermediate compounds that are

more reactive or toxic than their parent compounds. A suite of titanium dioxide-based linear

engineered nanomaterials (LENs) was synthesized and compared to standard commercial

titanium dioxide nanoparticles in terms of filterability, settleability, surface characteristics,

crystal phase structure, available surface area, photonic efficiency, and propensity to form

hydroxyl radicals. The LENs were also evaluated in terms of their ability to remove disinfection

byproduct (DBP) precursors from two natural surface water matrices via adsorption and

photocatalysis. DBPs, which form when naturally occurring organic precursor compounds

interact with chemical disinfectants used in drinking water treatment, are suspected carcinogens

and are widely regulated throughout the world. In this study, photocatalysis increased the DBP

iii

formation potential of both water matrices at short irradiation times. Longer treatment times

resulted in decreased in DBP formation potential. Adsorption removed DBP precursors from the

water without transforming them. The surface area and crystal phase structure of the

nanomaterials were identified as important drivers of photocatalytic treatment effectiveness and

regenerability. Adsorption efficacy was mainly impacted by surface area, agglomeration, and

charge interactions. The effects of both adsorption and photocatalysis on DBP formation

potential were strongly influenced by the composition of the water matrix being treated. The

results of this project have informed the conceptual design of two titanium dioxide-based water

treatment processes for DBP precursor removal: a single step photocatalytic system and a two-

step adsorption and regeneration system.

iv

Acknowledgments

This thesis is dedicated to my grandmother, Simone Francoeur, who was my first teacher, best

friend, and greatest supporter. Although she passed away during the early stages of this project, I

know that she would have been proud of my achievements as a researcher, mentor, and teacher.

This project was funded through scholarships provided by the Natural Sciences and Engineering

Research Council of Canada (NSERC), the Ontario Ministry of Training, and Engineers Canada

and Manulife.

The last five years have been the most challenging and rewarding of my life so far. My

supervisor, Professor Susan Andrews, has been a constant source of support throughout the

process and I would like to thank her for her patience and her insightful feedback over the past

five years. I would also like to acknowledge my other committee members, Professor Robert

Andrews, Professor Ron Hofmann, Professor Elodie Passeport, and Professor Benoit Barbeau,

who have provided helpful commentary throughout my project. This project wouldn’t have been

possible without the assistance and good company provided by our summer students and interns,

including Tassia Brito Andrade, Adrielle Costa Souza, Leonardo Furtado, Yijun (Jessie) Gai,

Michelli Park, Wan-Ying (Jenny) Yue, Chuqiao (Kaya) Yuan, Kennedy Santos, Katherine

Dritsas, and Jingyi Han. The various members of the Drinking Water Research Group, in

particular Aleksandra (Ola) Sokolowski and Jim Wang, have also been of great help.

Although they were not directly involved in this project, Pasquale Cirone, Margaret Walsh, and

the members of the Process Engineering Department at CBCL Limited in Halifax have all made

important contributions to my growth as a scientist and an engineer and I am thankful for the

efforts that they have made on my behalf over the years. I would also like to thank my family

and friends, including my parents, John and Sandra Gora, my sister, Jill Gora, along with Jen

Hill, Katherine Perrott, Sarah Jane Payne, and Seamus for their time, love, and patience as well

as the plants, animals, and caretakers of High Park in Toronto, which has been my refuge and

inspiration during my time in Toronto.

Finally, I would like to acknowledge the love and support of my partner, Andrew Sinclair, who

has been an excellent companion throughout this adventure and who will no doubt be very

pleased to see me graduate!

v

Table of Contents

Abstract ........................................................................................................................................... ii

Acknowledgments.......................................................................................................................... iv

Table of Contents .............................................................................................................................v

List of Tables ............................................................................................................................... xiii

List of Figures ............................................................................................................................. xvii

Nomenclature ............................................................................................................................. xxvi

Introduction .................................................................................................................................1

Background ..........................................................................................................................1

1.1.1 Light Sources ...........................................................................................................2

1.1.2 Formation of Undesirable Intermediates or Byproducts ..........................................2

1.1.3 Removal of Nanomaterials After Treatment............................................................3

Specific Research Objectives ...............................................................................................3

Associated Publications .......................................................................................................6

References ............................................................................................................................7

Background Literature Review ...................................................................................................9

Heterogeneous Photocatalysis .............................................................................................9

2.1.1 Photocatalysts ..........................................................................................................9

2.1.2 Photocatalytic Degradation of Organic Contaminants in Aqueous Media ............10

Titanium Dioxide ...............................................................................................................12

2.2.1 TiO2 Structure, Polymorphs, and Behaviour .........................................................12

2.2.2 Probing the Behaviour of TiO2 Photocatalysts ......................................................13

2.2.3 Effects of TiO2 Nanomaterials on Human and Environmental Health ..................15

2.2.4 TiO2 Photocatalysis for Drinking Water Treatment ..............................................17

vi

Engineered TiO2 Nanomaterials ........................................................................................19

2.3.1 Linear Engineered TiO2 Nanomaterials (LENs) ....................................................20

2.3.2 Alkaline Hydrothermal Synthesis of LENs ...........................................................20

2.3.3 Effects of Synthesis Parameters on Nanomaterial Properties ................................23

Natural Organic Matter ......................................................................................................23

2.4.1 Health, Operational, and Aesthetic Effects of Natural Organic Matter .................23

2.4.2 NOM Removal in Drinking Water Treatment Plants ............................................24

2.4.3 Photocatalytic Degradation of Natural Organic Matter .........................................25

Adsorption..........................................................................................................................27

2.5.1 Adsorption Theory .................................................................................................27

2.5.2 Adsorption of NOM to TiO2 ..................................................................................30

2.5.3 Nanoparticle Agglomeration ..................................................................................32

References ..........................................................................................................................35

Materials and Methods ..............................................................................................................50

Synthesis and Characterization of Linear Engineered Nanomaterials ...............................50

3.1.1 Alkaline Hydrothermal Synthesis Procedure .........................................................50

3.1.2 Characterization of LENs ......................................................................................52

Water Matrices ...................................................................................................................54

Experimental Apparatus.....................................................................................................57

3.3.1 Light Sources .........................................................................................................57

3.3.2 Additional Apparatus .............................................................................................57

Sample Preparation and Experimental Design ..................................................................57

3.4.1 Photocatalysis Tests ...............................................................................................57

3.4.2 Adsorption Tests ....................................................................................................58

3.4.3 Nanomaterial Regeneration ...................................................................................58

3.4.4 Settling Tests ..........................................................................................................58

vii

3.4.5 Filtration Tests .......................................................................................................59

Sample Analysis.................................................................................................................60

3.5.1 Dyes .......................................................................................................................60

3.5.2 Disinfection Byproduct Surrogates ........................................................................60

3.5.3 Disinfection Byproduct Formation and Analysis ..................................................60

3.5.4 Other Analyses .......................................................................................................61

Quality Control ..................................................................................................................61

References ..........................................................................................................................62

Preliminary Experimental Findings and Concept Development ...............................................64

Methods and Materials .......................................................................................................65

4.1.1 Experimental Design ..............................................................................................65

4.1.2 Materials ................................................................................................................66

4.1.3 Light Sources .........................................................................................................67

4.1.4 LEN Synthesis .......................................................................................................68

Results and Discussion ......................................................................................................69

4.2.1 Effect of Time, TiO2 Dose, Water Type, and Light Source on NOM

Adsorption and Degradation ..................................................................................69

4.2.2 Solar Photocatalysis with LENs for NOM Removal .............................................82

4.2.3 LENs for Dye Removal .........................................................................................91

Summary, Conclusions, and Implications for Future Experiments ...................................94

4.3.1 Light Source ...........................................................................................................94

4.3.2 Selection of Optimal LENs ....................................................................................96

4.3.3 Natural vs. Synthetic Water Matrices ....................................................................96

References ..........................................................................................................................97

Adsorption of Natural Organic Matter and Disinfection Byproduct Precursors from

Surface Water onto TiO2 Nanoparticles: pH Effects, Isotherm Modeling, and Implications

for the Use of TiO2 for Drinking Water Treatment.................................................................100

viii

Abstract ...................................................................................................................................100

Introduction ......................................................................................................................101

5.1.1 Titanium Dioxide for Drinking Water Treatment ................................................101

5.1.2 Natural Organic Matter ........................................................................................102

5.1.3 Adsorption of NOM to TiO2 ................................................................................102

5.1.4 Adsorption Models...............................................................................................103

5.1.5 Potential Risks and Opportunities Associated with the Use of TiO2

Nanoparticles for Water Treatment .....................................................................104

Materials and Methods .....................................................................................................105

5.2.1 Materials ..............................................................................................................105

5.2.2 Analytical Methods ..............................................................................................106

5.2.3 Sample Preparation ..............................................................................................107

Results and Discussion ....................................................................................................108

5.3.1 Disinfection Byproduct Formation During Photocatalysis ..................................108

5.3.2 NOM Removal via Adsorption – Time Series Experiments ...............................109

5.3.3 Effects of pH and TiO2 Dose on Adsorption ........................................................110

5.3.4 Modeling of Adsorption Isotherms ......................................................................113

5.3.5 Adsorption of DBP Precursors.............................................................................116

5.3.6 Effect of pH on Adsorption of LC-OCD Fractions .............................................118

Conclusions ......................................................................................................................119

References ........................................................................................................................121

Supplementary Material for Chapter 5 ............................................................................125

Development of Settleable Engineered Titanium Dioxide Nanomaterials for the Safe

Removal of Disinfection Byproduct Precursors from Drinking Water ..................................132

Abstract ...................................................................................................................................132

Introduction ......................................................................................................................132

Experimental ....................................................................................................................136

ix

6.2.1 Materials ..............................................................................................................136

6.2.2 Synthesis of Nanostructured Materials ................................................................137

6.2.3 Characterization of Nanomaterials ......................................................................138

6.2.4 Settling Tests ........................................................................................................139

6.2.5 Formation of ·OH Radicals ..................................................................................140

6.2.6 Characterization of NOM ....................................................................................140

6.2.7 Adsorption and Photocatalytic Degradation Under UVA Light ..........................140

6.2.8 Electrical Energy per Order Calculations ............................................................141


6.3.1 Nanomaterial Characterization ............................................................................142

6.3.2 Hydroxyl Radical Formation ...............................................................................149

6.3.3 Settling Experiments and Modeling.....................................................................149

6.3.4 Photocatalytic Degradation of Methylene Blue Dye Over Time .........................156

6.3.5 Removal of DBP Precursor Surrogates via Adsorption and Photocatalysis ........159

6.3.6 Electrical Energy per Order .................................................................................162

6.3.7 Comparison of Reaction Rate Constants and Implications for Degradation

Pathways ..............................................................................................................164

Conclusions ......................................................................................................................165

References ........................................................................................................................167

Supplementary Material ...............................................................................................................172

Photocatalysis with Engineered TiO2 Nanomaterials to Prevent the Formation of

Disinfection Byproducts in Drinking Water ...........................................................................180

Abstract ...................................................................................................................................180

Introduction ......................................................................................................................181

Materials and Methods .....................................................................................................184

7.2.1 Materials ..............................................................................................................184

7.2.2 Apparatus .............................................................................................................185

x

7.2.3 Synthesis and Characterization of Engineered TiO2 Nanomaterials ...................186

7.2.4 Settling and Filtration ..........................................................................................187

7.2.5 NOM and Dye Degradation Experiments ............................................................188

7.2.6 Calculations..........................................................................................................188

Results ..............................................................................................................................189

7.3.1 Characterization of Engineered TiO2 Nanomaterials ..........................................189

7.3.2 Filtration and Settling ..........................................................................................193

7.3.3 Degradation of Methylene Blue Dye ...................................................................196

7.3.4 Degradation of Natural Organic Matter (Dissolved Organic Carbon and

UV254) ................................................................................................................197

7.3.5 Removal and Degradation of Disinfection Byproduct Precursors .......................201

7.3.6 Alternative Measures of System Efficiency: Applied UV Dose and Power per

Volume .................................................................................................................207

7.3.7 Correlation Between Methylene Blue Degradation, NOM Degradation, and

DBPfp ..................................................................................................................209

Summary and Conclusions ..............................................................................................211

References ........................................................................................................................213


Removal of NOM and Disinfection Byproducts from Drinking Water Using Regenerable

Nanoscale Engineered TiO2 Adsorbents .................................................................................219

Abstract ...................................................................................................................................219

Introduction ......................................................................................................................220

Methods and Materials .....................................................................................................222

8.2.1 Materials ..............................................................................................................222

8.2.2 Raw Water Quality ..............................................................................................223

8.2.3 Synthesis and Characterization of Engineered Nanomaterials ............................225

8.2.4 Adsorption Experiments ......................................................................................225

8.2.5 Regeneration Experiments ...................................................................................226

xi

8.2.6 Filtration and Settling Tests .................................................................................227

8.2.7 Analysis of AO24, DOC, UV Absorbance, and DBP Formation Potential .........228

8.2.8 Isotherm Modeling and Other Statistical Analyses .............................................228


8.3.1 Characterization of Engineered Nanomaterials ...................................................229

8.3.2 Acid Orange 24 Adsorption to TiO2 ....................................................................230

8.3.3 NOM Adsorption to TiO2 Nanomaterials ............................................................233

8.3.4 Regeneration of Engineered Nanomaterials After NOM Adsorption ..................244

8.3.5 Removal of Nanomaterials from Treated Water ..................................................246

Conclusions ......................................................................................................................252

References ........................................................................................................................253


Summary, Conclusions, Engineering Significance, and Implications of Research ................265

Summary of Findings .......................................................................................................265

Overall Conclusions .........................................................................................................269

9.2.1 TiO2 Removes Disinfection Byproduct Precursors via Adsorption and

Degradation ..........................................................................................................269

9.2.2 Material Synthesis Conditions Determine the NOM Adsorption and

Degradation Behaviour of LENs .........................................................................270

9.2.3 Filtration is the Most Practical Option for Nanomaterial Removal .....................270

Engineering Significance of Findings ..............................................................................271

Implications for Future Research .....................................................................................272

References ........................................................................................................................274

Appendix A: Effects of Synthesis Conditions on LEN Characteristics......................................276

Appendix B: Matrix Impacts on Adsorption and Photocatalytic Degradation of NOM by

TiO2 .........................................................................................................................................280

Appendix C: Calibration Curves ................................................................................................284

xii

Appendix D: Quality Control .....................................................................................................291

Appendix E: Proposed TiO2-based Treatment Systems .............................................................294

Appendix F: Cost Comparison of Proposed TiO2-based Treatment Systems to Existing

Water Treatment Processes .....................................................................................................297

Appendix G: Irradiance Considerations .....................................................................................313

Appendix H: Sedimentation Analysis ........................................................................................317

Appendix I: Statistical Analysis of Regeneration Results ..........................................................329

Appendix J: Evaluating and Modeling System Performance .....................................................331

xiii

List of Tables

Table 3.1 Precursor materials and temperature setpoints employed during the alkaline

hydrothermal synthesis of LENs in this project .................................................... 52

Table 3.2 Characteristics of four water sources (in lab measurements, variable n, error

values represent standard deviation from the mean) ............................................. 54

Table 3.3 Percentages of different LC-OCD fractions present in raw water matrices used in

this study (n = 2) ................................................................................................... 55

Table 3.4 Additional water quality data for three natural water matrices used in this study

(DWSP 2010-2012) .............................................................................................. 56

Table 4.1 Experimental conditions used in previous studies ................................................ 65

Table 4.2 Summary of synthetic water quality ..................................................................... 67

Table 4.3 Light sources used for preliminary photocatalysis experiments ........................... 68

Table 4.4 Summary of synthesis parameters for first generation LENs ............................... 68

Table 4.5 Pseudo-first order reaction rate constants and fits for DOC and UV254 removal

from synthetic river water by different doses of TiO2 P25 nanoparticles irradiated

by simulated solar light ......................................................................................... 73


from Otonabee River water by different doses of TiO2 P25 nanoparticles

irradiated by simulated solar light......................................................................... 78


from synthetic water by 0.15 g/L of TiO2 P25 nanoparticles irradiated by

simulated solar light or high intensity UVA light ................................................. 81

Table 4.8 Summary of percent removal and kinetic parameters - DOC ............................... 86

Table 4.9 Summary of percent removal and kinetic parameters – UV254 ........................... 87

xiv

Table 4.10 Summary of characteristics of UVA light sources ............................................... 94

Table 4.11 Second generation LENs synthesis conditions ..................................................... 96

Table 5.1 Summary of isotherm parameters for the adsorption of NOM from Otonabee

River water onto P25 TiO2 nanoparticles at pH 4, pH 6, and pH 8. Error values

represent the 95% confidence interval on the mean. .......................................... 114

Table 5.S.1 Raw water quality ............................................................................................... 125

Table 5.S.2 Freundlich isotherm parameters for DOC ........................................................... 125

Table 6.1 Summary of raw water quality ............................................................................ 136

Table 6.2 Summary of nanomaterial synthesis conditions and percent degradation of

methylene blue dye during quality control tests ................................................. 138

Table 6.3 Characteristics of P25 and four linear engineered nanomaterials ...................... 143

Table 6.4 Removal, reaction rate constants, and R2 values for the pseudo-first-order

degradation of methylene blue dye by UV light, P25 nanoparticles, and four LENs

(error values represent the 95% confidence interval) ......................................... 158

Table 6.5 Reaction rate constants and R2 values for the pseudo-first-order degradation of

DOC by UV light, P25 nanoparticles, and four LENs (error values represent the

95% confidence interval of the rate constant) ..................................................... 161

Table 6.6 EEO values provided by Collins and Bolton (2016) for methylene blue

degradation by UV/H2O2 and UV/TiO2 and EEO values for the degradation of

methylene blue by P25 and second generation LENs irradiated by UVA LEDs 162

Table 6.S.1 Terminal settling velocities for sand particles and alum flocs (Crittenden et al.,

2012) ................................................................................................................... 174

Table 6.S.2 Predicted terminal settling velocities for nanomaterials in distilled water ......... 175

Table 6.S.3 Predicted terminal settling velocities for nanomaterials in a river water ............ 176

xv

Table 6.S.4 Predicted and actual settling time for nanomaterial suspensions prepared in

distilled water ...................................................................................................... 177

Table 6.S.5 Predicted and actual settling time for nanomaterial suspensions prepared in river

water .................................................................................................................... 177

Table 6.S.6 Predicted effective density of agglomerates formed by four nanomaterials in

distilled water and river water............................................................................. 178

Table 6.S.7 Fit of linear correlation between hydroxyl radical formation rate constants and

NOM degradation rate constants ........................................................................ 179

Table 7.1 Summary of raw water quality ........................................................................... 185

Table 7.2 Shape, size, and surface characteristics of LENs ............................................... 190

Table 7.S.1 NOM adsorption normalized to available surface area ...................................... 218

Table 8.1 Summary of raw water quality ........................................................................... 224

Table 8.2 Isotherm parameters for the adsorption of AO24 by P25 and two LENs ........... 231

Table 8.3 Modified Freundlich model isotherm parameters for the removal of DOC, UV254,

THM precursors, and HAA precursors from Otonabee River (OTB) and Ottawa

River (OTW) water by P25 nanoparticles and two LENs .................................. 240

Table A.1 Summary of LEN synthesis studies .................................................................... 276

Table A.2 Summary of the effects of LEN synthesis conditions on LEN characteristics ... 278

Table B.1 Matrix effects on NOM adsorption onto TiO2 surface ....................................... 280

Table B.2 Matrix effects on the photocatalytic degradation of NOM by TiO2 ................... 282

Table C.1 Parameters and fits of calibration curves (THMs) .............................................. 286

Table C.2 Parameters and fits of calibration curves (HAAs) .............................................. 287

Table C.3 Parameters and fits of calibration curves (TiO2 vs. turbidity) ............................ 288

xvi

Table C.4 Parameters and fits of calibration curves (TiO2 vs. UV375) .............................. 289

Table F.1 Recent capital costs for small water treatment systems in Canada ..................... 298

Table F.2 Cost of commercially available TiO2 LENs ....................................................... 299

Table F.3 Assumptions for energy cost analysis – single step treatment process ............... 299

Table F.4 Assumptions for material cost analysis – two-step treatment process ................ 305

Table F.5 Assumptions for energy cost analysis – two-step process .................................. 307

Table G.1 Irradiance of LED 1 ............................................................................................ 315



Table H.1 Settling time required for P25 and two LENs in MilliQ water ........................... 318

Table H.2 Predicted and actual time required to remove 10%, 50%, and 90% of TiO2 from

various water matrices ........................................................................................ 322

Table H.3 Effects of nanomaterial addition, time, and pH adjustment on the pH of MilliQ

water .................................................................................................................... 325

Table I.2 Statistical analysis of regeneration data – NOM (UV254) experiments............. 330

Table J.1 EEO values provided by Collins and Bolton (2016) for methylene blue

degradation by UV/H2O2 and UV/TiO2 and EEO values for the degradation of

methylene blue by P25 and second and third generation LENs irradiated by UVA

LEDs ................................................................................................................... 333

Table J.2 Cost to reduce the THMfp and HAAfp of OTW water via photocatalysis with P25

and NB 700 ......................................................................................................... 339

xvii

List of Figures

Figure 1.1 Organizational framework of research project ....................................................... 6

Figure 2.1 Alkaline hydrothermal synthesis method as described by Kasuga et al. (1999),

Wong et al. (2011), Liang (2014), and others ....................................................... 21

Figure 4.1 DOC of synthetic water matrix after adsorption by different doses of P25

nanoparticles ......................................................................................................... 69

Figure 4.2 Change in DOC content of synthetic water treated different doses of P25 TiO2

nanoparticles and irradiated by simulated solar light ........................................... 70

Figure 4.3 Change in the UV254 of synthetic river water treated with different doses of P25

TiO2 nanoparticles irradiated by simulated solar light ......................................... 72

Figure 4.4 Removal of DOC from Lake Ontario water by different doses of P25 TiO2

nanoparticles irradiated by simulated solar light .................................................. 75

Figure 4.5 Removal of UV254 from Lake Ontario water by different doses of P25 TiO2


Figure 4.6 Removal of DOC from Otonabee River water by different doses of P25 TiO2


Figure 4.7 Removal of UV254 from Otonabee River water by different doses of P25 TiO2


Figure 4.8 Removal of NOM from Otonabee River water by 0.15 g/L of P25 TiO2

nanoparticles irradiated by simulated solar light or high intensity UVA light as a

function of irradiation time ................................................................................... 80


nanoparticles irradiated by simulated solar light and high intensity UVA light as a

function of UVA dose ........................................................................................... 81

Figure 4.10 SEM images of nanobelts (NBs)........................................................................... 83

xviii

Figure 4.11 SEM images of nanowires (NW) and nanotubes (NT) ......................................... 84

Figure 4.12 XRD results for (a) nanobelts, (b) nanowires, and (c) nanotubes ......................... 85

Figure 4.13 Distribution of NOM fractions in Otonabee River water samples dosed with 0.1

g/L of P25, NB, NW, or NT and mixed in the dark for 1 minute ......................... 88

Figure 4.14 Effect of 60 minutes of photocatalysis with four LENs irradiated with simulated

solar light on the distribution of NOM fractions in the sample ............................ 89

Figure 4.15 Adsorption and decolourization of methylene blue dye by 0.1 g/L of P25

nanoparticles or one of three LENs after 0, 15, 30, 45, and 60 minutes of

irradiation with simulated solar light .................................................................... 92

Figure 4.16 Adsorption and decolourization of AO24 by 0.1 g/L of P25 nanoparticles or one

of three LENs after 0, 15, 30, 45, and 60 minutes of irradiation with simulated

solar light .............................................................................................................. 93

Figure 5.1 THMfp and HAAfp of Otonabee River water treated with 0.25 g/L and irradiated

by high intensity UVA LED light ....................................................................... 108

Figure 5.2 Adsorption of DOC from raw unchlorinated water from Otonabee River water (A)

and Lake Ontario water (B) adjusted to pH 4, pH 6, and pH 8 and mixed with 0.5

g/L of P25 TiO2 nanoparticles for four hours ..................................................... 111

Figure 5.3 Size distribution of agglomerates of P25 nanoparticles in Otonabee River water

adjusted to pH 4, pH 6, and pH 8 ........................................................................ 113

Figure 5.4 DOC data from Otonabee River water tests (A) and Lake Ontario water tests (B)

fitted to the modified Freundlich model ............................................................. 115

Figure 5.5 THMfp (A) and HAAfp (B) of Otonabee River water treated with increasing

concentrations of P25 TiO2 nanoparticles at pH 4, pH 6, and pH 8 ................... 117

Figure 5.6 LC-OCD fractions present in raw unchlorinated Otonabee River water and water

adjusted to pH 4, pH 6, and pH 8 and mixed with 0.5 g/L of P25 TiO2

nanoparticles for four hours ................................................................................ 119

xix

Figure 5.S.1 DOC of Otonabee River water (A) and Lake Ontario water (B) dosed with 0.5 g/L

of P25 TiO2 nanoparticles and allowed to mix in the dark for between 0 and 480

minutes ................................................................................................................ 126

Figure 5.S.2 UV254 of Otonabee River water (A) and Lake Ontario water (B) dosed with 0.5

g/L of P25 TiO2 nanoparticles and allowed to mix in the dark for between 0 and

480 minutes ......................................................................................................... 127

Figure 5.S.3 SUVA of Otonabee River water (A) and Lake Ontario water (B) dosed with 0.5

g/L of P25 TiO2 nanoparticles and allowed to mix in the dark for between 0 and

480 minutes ......................................................................................................... 128

Figure S.5.4 Adsorption of UV254 from raw unchlorinated water from Otonabee River water

(A) and Lake Ontario water (B) adjusted to pH 4, pH 6, and pH 8 and mixed with

0.5 g/L of P25 TiO2 nanoparticles for four hours ............................................... 129

Fig. 5.S.5 DOC data from Otonabee River water tests (A) and Lake Ontario water tests (B)

fitted to the Freundlich isotherm model .............................................................. 130

Figure S.5.6 THMfp (A) and HAAfp (B) data from Otonabee River water tests fitted to the

linearized modified Freundlich model ................................................................ 131

Figure 6.1 TEM images of A: NB 130/550, B: NB 130/700, C: NB 240/550, and D: NB

240/700 ............................................................................................................... 144

Figure 6.2 TEM images with SAED indexed regions (yellow) and HRTEM images with

corresponding FT image of LEN samples (figure created by Robert Liang at the

University of Waterloo using results obtained at McMaster University) ........... 146

Figure 6.3 Determination of isoelectric point of NB 130/550 (A), NB 130/700 (B), NB

240/550 (C), NB 240/700 (D), and P25 nanoparticles (E) using zeta potential at

various pH conditions ......................................................................................... 148

Figure 6.4 Settling of P25 nanoparticles and four engineered nanomaterials in purified water

and raw Otonabee River water (n = 3, error bars represent the standard deviation

from the mean) .................................................................................................... 151

xx

Figure 6.5 Photographs of P25 (A), NB 130/550 (B), NB 130/700 (C), NB 240/550 (D), and

NB 240/700 (E) settling in purified water .......................................................... 152

Figure 6.6 Photocatalytic degradation of methylene blue dye by P25 nanoparticles and four

LENs (error bars represent the standard deviation from the mean) .................... 157

Figure 6.7 Photocatalytic degradation of DOC by P25 nanoparticles and four LENs ......... 160

Figure 6.8 Removal of UV254 by photocatalysis with P25 nanoparticles and four LENs .. 160

Figure 6.9 EEO values for DOC removal from synthetic water via UV/H2O2 treatment with a

low pressure UV lamp (Yen and Yen, 2015) and DOC removal from raw surface

water via UV/TiO2 treatment with P25 and four lab synthesized LENs irradiated

with UVA LEDs ................................................................................................. 163

Figure 6.S.1 HTPA / ·OH radical formation by P25 and four linear engineered nanomaterials

irradiated with UVA LED light (n = 3, error bars represent standard deviation

from the mean) .................................................................................................... 172

Figure 6.S.2 Particle size distribution for P25 nanoparticles, NB 130/550, NB 240/550, and NB

240/700 suspended in distilled water .................................................................. 173

Figure 6.S.3 Particle size distribution for P25 nanoparticles, NB 130/550, NB 240/550, and NB

240/700 suspended in river water ....................................................................... 173

Figure 6.S.4 Linear correlation between normalized DOC and UV254 degradation rate

constants and HTPA/hydroxyl radical formation rate constants for four linear

engineered nanomaterials.................................................................................... 178

Figure 7.1 Characterization of NB 550 (A) and NB 700 (B) via TEM and SAED. Figure

created by Robert Liang from the University of Waterloo using results obtained at

McMaster University .......................................................................................... 191

Figure 7.2 Hydroxyl radical (·OH) radical production by P25, NB 550, and NB 700. Error

bars represent the standard deviation from the mean (n = 3). ............................. 193

xxi

Figure 7.3 Filtration indexes of three TiO2 nanomaterials suspended in purified (MQ) water,

Otonabee River (OTB) water, and Ottawa River (OTW) water ......................... 194

Figure 7.4 Average settling rates of TiO2 nanomaterials suspended in MilliQ (MQ) water,

Otonabee River (OTB) water, and Ottawa River water (OTW) ......................... 195

Figure 7.5 Degradation of methylene blue dye by P25 nanoparticles and two LENs ......... 196

Figure 7.6 Degradation of DOC and UV254 from (A) Otonabee River water and (B) Ottawa

River water (B) by P25 nanoparticles and two LENs ......................................... 198

Figure 7.7 Reduction in the formation of trihalomethanes in two water matrices after

treatment by (A) P25 in OTB water, (B) NB 550 in OTB water, (C) NB 700 in

OTB water, (D) P25 in OTW water, (E) NB 550 in OTW water, (F) NB 700 in

OTW water. Error bars represent the 95% confidence interval of the mean. ..... 202

Figure 7.8 Reduction in the formation of haloacetic acids in two water matrices after



OTW water. Error bars represent the 95% confidence interval of the mean ...... 206

Figure 7.9 Reduction of THMfp in OTB and OTW water via photocatalysis by 0.25 g/L of

NB 700 irradiated with UVA LEDs (365 nm) as a function of irradiation time

(min), UV dose (J/cm2), and power per treated volume (kWh/m3) .................... 209

Figure 8.1 TEM images of (A) NB 550 and (B) NB 700 ..................................................... 230

Figure 8.2 AO24 adsorption data fitted to the Freundlich isotherm model ......................... 232

Figure 8.3 AO24 adsorption by virgin and regenerated NB 550 and NB 700. All samples

were prepared in duplicate and error bars represent the 95% confidence interval

on the mean. Legend numbers correspond to the number of regeneration cycles.

............................................................................................................................. 233

xxii

Figure 8.4 Adsorption of DOC from Otonabee River water (OTB) and Ottawa River water

(OTW) by P25 nanoparticles and two LENs. Error bars represent the 95%

confidence interval on the mean. ........................................................................ 235

Figure 8.5 Adsorption of UV254 from Otonabee River water (OTB) and Ottawa River water

(OTW) by P25 nanoparticles and two LENs nanomaterials. Error bars represent

the 95% confidence interval on the mean. .......................................................... 236

Figure 8.6 Adsorption of THM precursors from Otonabee River water (OTB) and Ottawa

River water (OTW) by P25 nanoparticles and two LENs. Error bars represent the

95% confidence interval on the mean. ................................................................ 237

Figure 8.7 Adsorption of HAA precursors from Otonabee River water (OTB) and Ottawa

River water (OTW) by P25 nanoparticles and two LENs. Error bars represent the

95% confidence interval on the mean. ................................................................ 238

Figure 8.8 DOC adsorption data sets from experiments conducted in (A) Otonabee River

(OTB) water and (B) Ottawa River (OTW) water fitted to a modified Freundlich

isotherm model.................................................................................................... 239

Figure 8.9 THMfp adsorption data sets from experiments conducted in (A) Otonabee River

(OTB) water and (B) Ottawa River (OTW) water fitted to a modified Freundlich

isotherm model.................................................................................................... 242

Figure 8.10 Adsorption of aromatic NOM (UV254 absorbing NOM) by virgin and

regenerated NB 550 and NB 700. Error bars represent the 95% confidence

interval on the mean and legend numbers correspond to the number of

regeneration cycles.............................................................................................. 245

Figure 8.11 Filtration indexes of raw water and three TiO2 nanomaterials suspended in MilliQ

water at pH 6 and pH 8 and two raw surface water samples .............................. 247

Figure 8.12 Percent removal of turbidity via sedimentation for three TiO2 nanomaterials

suspended in two raw surface water samples ..................................................... 249

xxiii

Figure 8.13 Zeta potential of P25 nanoparticles and two LENs in two natural water matrices

250

Figure 8.S.1 Zeta potential as a function of pH for two TiO2 LENs ....................................... 259

Figure 8.S.2 Time series data for AO24 removal by P25 nanoparticles and LENs ................. 260

Figure 8.S.3 Time series data for DOC removal by P25 nanoparticles and two LENs from

Otonabee River (OTB) and Ottawa River water (OTW) .................................... 260

Figure 8.S.4 UV254 isotherms in (A) OTB water and OTW water ........................................ 261

Figure 8.S.5 HAAfp isotherms in OTB water and OTW water .............................................. 262

Figure 8.S.6 Concentration of AO24 in water treated with virgin and regenerated LENs ..... 263

Figure 8.S.7 UV254 of OTB and OTW water treated with virgin and regenerated LENs ..... 263

Figure 8.S.8 Ratio of TCM to BDCM in surface water treated with TiO2 .............................. 264

Figure 8.S.9 Ratio of DCAA to TCAA in surface water treated with of TiO2........................ 264

Figure 9.1 Framework for the development of a prototype of a two-step adsorption and

photocatalytic process for drinking water treatment ........................................... 273

Figure C.1 Representative calibration curve for methylene blue dye ................................... 284

Figure C.2 Representative calibration curve for Acid Orange 24 ......................................... 285

Figure C.3 Representative calibration curve for TOC .......................................................... 285

Figure C.4 Calibration curves for four THMs (Summer 2016) ............................................ 286

Figure C.5 Calibration for nine HAA species (Summer 2016) ............................................. 287

Figure C.7 Calibration curves for P25 and LENs vs. UV absorbance at 375 nm ................. 289

Figure C.8 Calibration curves for HTPA vs. fluorescence (Ex: 315 nm, Em: 425 nm) ....... 290

Figure D.1 Quality control results for batches of second generation LENs ......................... 291

xxiv

Figure D.2 Quality control results for batches of third generation LENs ............................. 291

Figure D.3 Quality control chart for TOC/DOC ................................................................... 292

Figure D.4 QC chart for TCM Figure D.5 QC chart for BDCM ........................................ 293

Figure D.6 QC chart for DCAA Figure D.7 QC chart for TCAA ....................................... 293

Figure E.1 Single step photocatalytic system with membrane filtration for separation ....... 294

Figure E.2 Two-step adsorption and regeneration system with membrane filtration ........... 295

Figure E.3 Two-step adsorption and regeneration system with sedimentation .................... 296

Figure F.1 Estimated annual energy cost for the single step treatment process option as a

function of plant capacity.................................................................................... 300

Figure F.2 Number of reuses required for the single tank system to be competitive with

existing water treatment systems as a standalone option .................................... 302

Table F.4 Assumptions for material cost analysis – two-step treatment process ................ 305

Figure F.3 Effects of TiO2 LEN dose and plant capacity on estimated energy costs ........... 306

Figure F.4 Effects of TiO2 LEN unit cost and plant capacity on estimated annual materials

cost ...................................................................................................................... 306

Figure F.5 O&M costs as a function of plant capacity for existing water treatment processes

used for NOM removal at large or small scale ................................................... 308

Figure G.1 Absorbance at 365 nm and average irradiance through the volume of the sample

for 50 mL samples of distilled water dosed with varying concentrations of P25

TiO2 nanoparticles ............................................................................................... 314

Figure G.2 Average irradiance through different volumes of sample at different doses of P25

TiO2 nanoparticles in MilliQ distilled water ....................................................... 314

Figure H.1 Effect of particle/agglomerate size on time required to settle ............................ 318

xxv

Figure H.2 Particle size distributions for P25 nanoparticles (A), NB 550 (B), and NB 700 (C)

in MilliQ water (natural pH) and two natural water matrices. ............................ 320

Figure H.3 Time required to settle as a function of particle/agglomerate size and

particle/agglomerate density ............................................................................... 321

Figure H.4 Percent removal of turbidity over time via settling in real water matrices ......... 324

Figure H.5 Filtration indexes of raw water and three TiO2 nanomaterials suspended in MilliQ

water at pH 6 and pH 8 and two raw surface water samples .............................. 325

Figure H.6 Percent removal of turbidity from suspensions made with TiO2 nanomaterials in

MilliQ water at pH 6 and pH 8 ........................................................................... 327

Figure J.1 Reduction of THMfp in OTB and OTW water matrices via photocatalysis by NB

700....................................................................................................................... 331

Figure J.2 Comparison of EEOs for DOC and THM precursor degradation by UV/H2O2 and

UV/TiO2 with P25 and third generation LENs ................................................... 334

Figure J.3 Reduction of THMfp in OTB and OTW water matrices via photocatalysis with

NB 700 ................................................................................................................ 335

Figure J.4 Power required to remove 90% of DOC and THMfp from different water matrices

using UV/TiO2-based treatment processes ......................................................... 337

Figure J.5 Reduction of THMfp in OTB and OTW water via photocatalysis with NB 700 as

a function of UV dose (fluence) .......................................................................... 341

Figure J.6 Transmittance of light through 10 mg/L methylene blue solution and the two raw

water matrices used in this project ...................................................................... 344

Figure J.7 Absorbance of UV and visible light by 0.05 g/L and 0.1 g/L of P25 nanoparticles

and 0.1 g/L of NB 550 and NB 700 suspended in MilliQ water ........................ 345

Figure J.8 Simplified model describing the degradation of an organic contaminant via TiO2

photocatalysis ...................................................................................................... 351

xxvi

Nomenclature

AO24 Acid Orange 24

AOP Advanced oxidation process

BDCAA Bromodichloroacetic acid

BDCM Bromodichloromethane

BET Brunauer–Emmett–Teller

CCC Critical coagulation concentration

CDBM Chlorodibromomethane

DBAA Dibromoacetic acid

DBP Disinfection byproduct

DBPfp Disinfection byproduct formation potential

DCAA Dichloroacetic acid

DOC Dissolved organic carbon

DWRG Drinking Water Research Group

DWSP Drinking Water Surveillance Program

EDL Electrical double layer

GAC Granular activated carbon

HAA Haloacetic acid

HAAfp Haloacetic acid formation potential

GC Gas chromatography

HRTEM High resolution transmission electron microscope

IEP Isoelectric point

IHSS International Humic Substances Society

LC-OCD Liquid chromatography with organic carbon detection

xxvii

LED Light emitting diode

LEN Linear engineered nanomaterial

LP Low pressure

MB Methylene blue

MBAA Monobromoacetic acid

MCAA Monochloroacetic acid

MF Microfiltration

MilliQ Ultrapure water produced by a Millipore water purification system

MP Medium pressure UV lamp

NB Nanobelt

NB 130/550 Second generation nanobelt synthesized at 130oC and calcined at 550oC




NB 550 Third generation nanobelt calcined at 550oC

NB 700 Third generation nanobelt calcined at 700oC

NOM Natural organic matter

NT Nanotube

NW Nanowire

OTB Otonabee River

OTW Ottawa River

P25 Standard TiO2 nanoparticle manufactured by Evonik Degussa

PAC Powdered activated carbon

PES Polyethersulfone

xxviii

QC Quality control

ROS Reactive oxygen species

SAED Selected area electron diffraction

SEM Scanning electron microscope

SRNOM Suwannee River NOM

SUVA Specific UV254 absorbance

TBAA Tribromoacetic acid

TBM Tribromomethane (bromoform)

TC Calcination temperature

TCAA Trichloroacetic acid

TCM Trichloromethane (chloroform)

TEM Transmission electron microscope

TH Hydrothermal synthesis temperature

THAAfp Total haloacetic formation potential

THM Trihalomethane

THMfp Trihalomethane formation potential

TiO2 Titanium dioxide

TOC Total organic carbon

TTHMfp Total trihalomethane formation potential

UV Ultraviolet

UV254 Ultraviolet light absorbance at 245 nm

UVA Ultraviolet light between 315 nm and 400 nm

WTP Water treatment plant

XRD X-ray diffraction

1

Introduction

Background

Titanium dioxide (TiO2) is a semiconductor photocatalyst that has occasionally been employed

for water and wastewater treatment but has yet to be widely adopted for these purposes.

Photocatalysts, like other catalysts, are materials that participate in chemical reactions without

being consumed by them. Unlike other catalysts, however, photocatalysts are only active when

irradiated with light of the proper wavelength. TiO2 is activated by light at or below 385 nm,

which is within the UVA light spectrum. In aqueous media and in the presence of oxygen the

irradiation of TiO2 with UV light results in the formation of photoinduced electrons and electron

holes on the surface of the photocatalyst. These active species can interact directly with

contaminants or react with water and oxygen to form reactive oxygen species (ROS). The

following chemical equation describes the simplified oxidative organic degradation pathway in

photocatalytic water treatment systems:

𝑂𝑟𝑔𝑎𝑛𝑖𝑐 𝐶𝑜𝑛𝑡𝑎𝑚𝑖𝑛𝑎𝑛𝑡 + 𝑂2ℎ𝑣>𝐸𝑏,𝑠𝑒𝑚𝑖𝑐𝑜𝑛𝑑𝑢𝑐𝑡𝑜𝑟→ 𝑀𝑖𝑛𝑒𝑟𝑎𝑙𝑠 (𝑒. 𝑔. 𝐶𝑂2, 𝑚𝑖𝑛𝑒𝑟𝑎𝑙 𝑎𝑐𝑖𝑑𝑠) (1.1)

Depending on the original form of the organic species under consideration, intermediate

compounds can be formed during the reaction between the parent compound and the

photoinduced oxidative holes and/or various ROS formed upon irradiation (Malato et al., 2009).

Most research on drinking water treatment with TiO2 has focused on its use as an advanced

oxidation process (AOP). Like other AOPs, TiO2 photocatalysis involves the addition of a

relatively inert chemical to the water. The added chemical is then activated via a second force, in

this case UVA light. Unlike other AOPs such as UV/H2O2, the original material is a nanosized

solid that is not consumed during the oxidation process.

The three main challenges that prevent the use of TiO2 for drinking water treatment are:

1. How do we provide light of the appropriate wavelength (<385 nm) and intensity in a

reliable and energy efficient manner?

2. How do we avoid the formation of potentially dangerous intermediate compounds?

3. How do we remove the photocatalyst from the water after treatment?

2

The goal of this research was to address these challenges and lay the groundwork for the

eventual development of a TiO2-based drinking water treatment system. The following

subsections provide a quick introduction to the current body of knowledge about TiO2

photocatalysis for drinking water treatment.

1.1.1 Light Sources

TiO2 is only activated by light with wavelengths at or below 385 nm, which is in the UVA range.

The vast majority of TiO2 photocatalysis studies have made use of high intensity UVA lamps,

usually with a maximum irradiance at 365 nm. These lamps are widely available, emit light in

the required range and come in a variety of configurations, making it easy to adapt them to

bench-scale apparatus. The sun emits a wide spectrum of irradiation, some of which passes

through the Earth’s atmosphere to the surface, including small amounts in the UVA and UVB

ranges. Efforts have been made to harness this light to drive photocatalytic processes, however,

the low UV intensity and unpredictability of solar irradiation have limited the use of these

processes for drinking water treatment. The low pressure (LP) and medium pressure (MP) UV

lamps commonly used for drinking water disinfection also emit light below 385 nm and are

therefore able to activate TiO2. These lamps require substantial energy input and can only be

used in a limited number of configurations but they can potentially provide concurrent

disinfection. UVA light emitting diodes (LEDs) are an attractive alternative to existing light

sources because they are inexpensive, long lasting, and less energy intensive than the other UV

lamps. They are also small and easy to integrate into different reactor configurations. UVA LEDs

with a maximum irradiance at 365 nm were used for the majority of the photocatalytic

degradation and regeneration experiments in this project.

1.1.2 Formation of Undesirable Intermediates or Byproducts

The degradation of complex organic molecules, including natural organic matter (NOM), via

TiO2 photocatalysis is a multistep process and full mineralization may not be achieved within an

acceptable treatment time frame. For example, other researchers, including Liu et al. (2008),

Huang et al. (2008), and Gerrity et al. (2010) have observed that at short treatment times

photocatalysis can increase the overall disinfection byproduct formation potential (DBPfp) of

3

water by breaking down large NOM molecules into smaller ones that are more reactive towards

disinfectants such as chlorine.

TiO2 has both adsorptive and oxidative abilities and in this project it was hypothesized that the

risk of increased DBPfp after photocatalysis could be avoided by instead removing DBP

precursors via adsorption to TiO2. Adsorption is a well established water treatment process used

to remove organic contaminants, including disinfection DBP precursors. Existing adsorbents

such as powdered activated carbon (PAC) are effective but difficult to regenerate. In theory, a

photocatalytic TiO2 adsorbent could be regenerated via photocatalysis and reused for adsorption

indefinitely, or at least multiple times.

1.1.3 Removal of Nanomaterials After Treatment

Systems that employ TiO2 are challenging to operate in part because it can be difficult to

separate the suspended particles or nanoparticles from the water after treatment (Ochiai and

Fujishima, 2013). Much research has been conducted to engineer TiO2 materials that are

immobilized on solid supports but which maintain their photocatalytic properties and a few

researchers have had success with magnetic TiO2 nanomaterials (Ng et al., 2014) and TiO2-

covered zeolites (Liu et al., 2014). An alternative to immobilization may be the development of

engineered TiO2 nanomaterials that are large, heavy, or buoyant enough to be removed via

common clarification processes such as filtration, sedimentation, and flotation.

Specific Research Objectives

The overall objective of this research project was to develop a conceptual treatment methodology

that would address some or all of the three main challenges preventing the adoption of TiO2 for

drinking water treatment. Conceptual schematics of the two potential treatment processes are

provided in Appendix E. The first conceptual treatment system (single step process) uses

photocatalysis for DBP precursor degradation. In the second treatment concept (two-step

process) TiO2 is used to remove DBP precursors via adsorption and then regenerated via

photocatalysis. Both concepts rely on the existence of a TiO2 material that is small enough to

maintain the desirable photocatalytic properties of TiO2 nanoparticles but large enough to be

4

removed via sedimentation or filtration. Four specific research objectives were developed to

guide the project. They are summarized in the subsections that follow as well as in Figure 1.2,

which presents a visual framework for the overall research project.

1. Explore the use of standard TiO2 nanoparticles for NOM and DBP precursor removal via

adsorption and photocatalytic degradation.

The use of TiO2 photocatalysis for DBP precursor removal has been studied by researchers such

as Liu et al. (2008), Huang et al. (2008), and Gerrity et al. (2010), all of whom raised concerns

that while photocatalysis is effective for DBP precursor degradation at longer treatment times, it

can actually increase DBP formation potential at shorter treatment times. Although only a few

researchers have proposed the use of NOM adsorption to TiO2 for water treatment (Liu et al.,

2014; Ng et al., 2015), the phenomenon has been explored by numerous materials science and

environmental chemistry researchers in a contaminant fate and transport context (Liu et al.,

2013; Mwaanga et al., 2014; Erhayem and Sohn, 2015). These studies have consistently shown

that TiO2 adsorbs NOM, particularly large and aromatic fractions, and that the extent of NOM

adsorption is impacted by the pH, ionic strength, and calcium content of the water matrix.

Preliminary experiments relating the removal of DBP precursor surrogates such as DOC and

UV254 by commercial P25 TiO2 nanoparticles to process parameters such as TiO2 dose,

adsorption time, irradiation time, and light source are described in Chapter 4. A more in-depth

exploration of the use of P25 nanoparticles for DBP precursor removal via photocatalytic

degradation and adsorption at different pH levels is described in Chapter 5.

2. Develop engineered nanomaterials that are easy to remove from the water via conventional

water treatment clarification processes but retain the adsorptive and photocatalytic properties of

standard TiO2 nanoparticles.

Linear engineered nanomaterials (LENs), which are nanosized in at least one dimension but

larger than commercial nanoparticles, retain many of the properties of smaller nanomaterials but

promise to be easier to remove from water after treatment due to their relatively larger size. In

this study, three sets, or generations, of LENs were synthesized using a simple hydrothermal

method originally proposed by Kasuga et al. (1999). The size, shape, crystallinity, and surface

characteristics of the LENs were manipulated by varying the precursor materials and

5

temperatures used at different points in the synthesis procedure. The development of the first

generation of LENs, which is described in Chapter 4, was based on modifications to the standard

hydrothermal method suggested by Yuan and Su (2004). The results of these experiments

informed the design of the second generation of LENs, which underwent extensive materials

characterization and were evaluated in terms of their ability to degrade methylene blue dye and

DBP precursor surrogates when irradiated with UVA light as well as their recoverability from

the water via sedimentation. The results and implications of these experiments are described in

Chapter 6.

3. Evaluate the use of the linear engineered nanomaterials for DBP precursor removal from real

water matrices via photocatalytic degradation.

Although TiO2 photocatalysis is a promising technology for the degradation of organic

contaminants, including NOM, its effects on overall DBPfp are dependent on treatment time (Liu

et al., 2008) and others have suggested that water matrix characteristics such as ionic strength

and NOM content and character can also influence the extent of NOM degradation and the

resulting DBPfp of the treated water. Chapter 7 describes a study in which the third generation

LENs, which represent a subtly modified subset of the second generation of LENs, were

characterized and evaluated in terms of their effects on the DBPfp of two real surface water

matrices via photocatalysis upon irradiation with UVA light.

4. Evaluate the use of the linear engineered nanomaterials for DBP precursor removal from real

water matrices via adsorption.

Adsorption may prove to be a safer way to incorporate TiO2 into drinking water treatment

because it avoids the risk of increasing the DBPfp of the water. Chapter 8 presents an in-depth

study of the adsorption of an indicator dye and DBP precursors by commercial nanoparticles and

the third generation LENs. Adsorption isotherms were used to characterize and compare the

removal of an DOC, UV254, THM precursors, and HAA precursors from two water matrices

that differed in terms of ionic content and NOM content and character. The LENs were also

evaluated in terms of their removability via sedimentation and filtration at TiO2 doses relevant to

adsorption and their reusability after regeneration under UVA light.

6

Figure 1.1 Organizational framework of research project

Associated Publications

The following publications are associated with this project:

Gora, S. and Andrews, S. (2017) Adsorption of natural organic matter and disinfection byproduct

precursors from surface water onto TiO2 nanoparticles: pH effects, isotherm modelling and

implications for using TiO2 for drinking water treatment, Chemosphere, 174, 363-370

https://doi.org/10.1016/j.chemosphere.2017.01.125

Gora, S., Liang, R., Zhou, Y.N., Andrews, S. (2017) Settleable engineered titanium dioxide

nanomaterials for the removal of natural organic matter from drinking water (in review –

Chemical Engineering Journal)

Gora, S., Liang, R., Zhou, Y.N., Andrews, S. (2017) Photocatalysis with engineered TiO2

nanomaterials to prevent the formation of disinfection byproducts in drinking water (drafted)

https://doi.org/10.1016/j.chemosphere.2017.01.125

7

Gora, S. and Andrews, S. (2017) Removal of NOM and disinfection byproduct precursors from

drinking water using regenerable nanoscale engineered TiO2 adsorbents (drafted)

References

Bavykin, D.V. and Walsh, F.C. (2009) Titanate and Titania Nanotubes: Synthesis, Preparation,

and Application, RSC Publishing

Erhayem, M. and Sohn, M. (2014) Stability studies for titanium dioxide nanoparticles upon

adsorption of Suwannee River humic and fulvic acids and natural organic matter, Science of the

Total Environment, 468-469, pp. 249-257

Gerrity, D., Mayer, B., Ryu, H., Crittenden, J., and Abbaszadegan, M. (2009) A comparison of

pilot-scale photocatalysis and enhanced coagulation for disinfection byproduct mitigation, Water

Research, 43, pp. 1597-1610

Huang, X., Leal, M., and Li, Q. (2008) Degradation of natural organic matter by TiO2

photocatalytic oxidation and its effect on fouling of low-pressure membranes, Water Research,

pp. 1142-1150

Kasuga, T., Hiramatsu, M., Hoson, A., Sekino, T., and Niihara, K. (1999) Titania nanotubes

prepared by chemical processing, Advanced Materials, 11 (15), pp. 1307-1311

Liu, S., Lim, M., Fabris, R., Chow, C., Drikas, M., Amal, R. (2008A) TiO2 photocatalysis of

natural organic matter in surface water: Impact on trihalomethane and haloacetic acid formation

potential, Environmental Science and Technology, 42, 6218-6223

Liu, S., Lim, M., Fabris, R., Chow, C., Drikas, M., Korshin, G., and Amal, R. (2010) Multi-

wavelength spectroscopic and chromatography study on the photocatalytic oxidation of natural

organic matter, Water Research, 44, pp. 2525-2532

Liu, S., Lim, M., and Amal, R. (2014) TiO2-coated natural zeolite: Rapid humic acid adsorption

and effective photocatalytic regeneration, Chemical Engineering Science, 105 pp. 46-52

8

Liu, W., Sun, W., Borthwick, A., and Ni, J. (2013) Comparison on aggregation and

sedimentation of titanium dioxide titanate nanotubes and titanate nanotubes-TiO2: Influence of

pH, ionic strength, and natural organic matter, Colloids and Surfaces A: Physicochemical

Engineering Aspects, 434, pp 319-328

Malato, S., Fernandez-Ibanez, P., Maldonado, M.I., Blanco, J., and Gernjak, W. (2009)

Decontamination and disinfection of water by solar photocatalysis: Recent overview and trends,

Catalysis Today, 147, 1-59

Mwaanga, P., Carraway, E.R., and Schlautman, M.A. (2014) Preferential sorption of some

natural organic matter fractions to titanium dioxide nanoparticles: influence of pH and ionic

strength, Environmental Monitoring and Assessment, 186, pp. 8833-8844

Ng, M., Kho, E.T., Liu, S., Lim, M., and Amal, R. (2014) Highly adsorptive and regenerative

magnetic TiO2 for natural organic matter (NOM) removal in water, Chemical Engineering

Journal, 246, pp. 196-203

Philippe, K.K., Hans, C., MacAdam, J., Jefferson, B., Hart, J., Parsons, S.A. (2010B)

Photocatalytic oxidation of natural organic matter surrogates and the impact on trihalomethane

formation potential, Chemosphere, 81, 1509-1516

Yuan, Z-Y and Su B-L (2004) Titanium oxide nanotubes, nanofibres, and nanowires, Colloids

and Surfaces A: Physicochem. Eng. Aspects, 241, pp. 173-183

9

Background Literature Review

Heterogeneous Photocatalysis

2.1.1 Photocatalysts

Heterogeneous photocatalysis is the process of using light to activate a solid photocatalytic

semiconductor material such that it can drive desirable oxidation and reduction reactions.

Photocatalytic semiconductors have electronic structures characterized by a filled valence band

and an empty conduction band. The gap between these two bands is called the band gap and the

energy required to promote an electron from the valence band to the conduction band is the band

gap energy (Eb). The limiting wavelength is directly related to the width of the band gap:

Ephoton = hc/ (2.1)

Where h is Planck’s constant (6.626 x 10-34 m2.k/s), c is the speed of light (299,792,458 m/s),

and is the wavelength in meters (Malato et al., 2009). Activation will only occur if the photons

that reach the photocatalyst surface are sufficiently energetic to bridge the band gap. The specific

formulation of the catalyst will have an effect on the band gap energy and thus on the range of

wavelengths that can be used to excite it. For example, TiO2 has an Eb value between 3 eV and

3.2 eV depending on the types and proportions of semiconductor phases present and thus can

only be activated by wavelengths below 385 nm (UVA light) whereas CdS has an Eh of 2.5 eV

and a corresponding maximum activation wavelength of 497 nm, which is within the visible light

spectrum.

When a photon with energy equal to or greater than the bandgap energy is absorbed by the

catalyst, an electron/hole pair is formed.

𝑃ℎ𝑜𝑡𝑜𝑐𝑎𝑡𝑎𝑙𝑦𝑠𝑡 + ℎ𝑣 → 𝑒− + ℎ+ (2.2)

The photogenerated hole and electron then undergo one of the following (Liang, 2014):

1. Recombination resulting in the release of heat

2. Stabilization (trapping) by metastable surface states

3. Reaction with species adsorbed to the photocatalyst surface

10

2.1.2 Photocatalytic Degradation of Organic Contaminants in Aqueous Media

Photocatalysts are like other catalysts in that their interactions with other chemical species follow

the following five steps:

1. Transfer of reactants from the bulk matrix to the surface of the catalyst

2. Adsorption of reactants onto catalyst surface

3. Reaction of adsorbed species at the surface

4. Desorption of reaction products from the surface

5. Transfer of reaction products back to the bulk matrix

Unlike standard catalysts, however, photocatalysts are activated by light instead of temperature

(Hermann, 2010). Further complicating matters is the fact that in aqueous media the

photocatalyst will interact not only with contaminants but also with water and oxygen. Water

adsorbed to the surface of the photocatalyst is oxidized by the photogenerated hole to form H2O2

while oxygen is reduced by the coexisting electron, forming the superoxide radical. These

processes and the subsequent reactions eventually result in the formation of the highly oxidative

•OH, or hydroxide, radical (Nosaka and Nosaka, 2013). Many researchers also assume that the

photogenerated holes can interact with H2O in a single electron transfer reaction to directly

produce •OH along with a hydrogen ion (Jenks et al.,2013; Hermann et al., 2012; Malato et al.,

2009), but this process has not been experimentally confirmed and may only occur within certain

pH ranges (Nosaka and Nosaka, 2013). The hydroxide radical may remain adsorbed to the

catalyst surface (•OH(a)) or be present in the bulk solution surrounding the catalyst (Murakami

et al., 2007). The hole itself as well as the other reactive oxygen species (ROS) present can also

react directly with chemical species adsorbed on the surface of the photocatalyst.

The •OH radical, which may be adsorbed to the catalyst surface (•OH(a)) or even present in the

bulk solution surrounding the catalyst (Murakami et al., 2007), is a non-specific oxidant that is

able to degrade most organic compounds, including the components of cell walls and cell

membranes. The oxidative capacity of the different ROS varies, but recent studies suggest that in

some systems, their role may be as or more important than that of •OH (Nosaka and Nosaka,

2013). The dominant oxidation mechanism (hole, •OH, or other ROS) will vary depending on the

characteristics of the target molecule (Henderson, 2011) and the characteristics of the

11

photocatalyst and will affect the distribution of intermediate products formed upon oxidation

(Jenks et al., 2013) as well as the optimum conditions for the photocatalytic process.

The following chemical equation describes the simplified oxidative organic degradation pathway

in photocatalytic water treatment systems:

𝑂𝑟𝑔𝑎𝑛𝑖𝑐 𝐶𝑜𝑛𝑡𝑎𝑚𝑖𝑛𝑎𝑛𝑡 + 𝑂2ℎ𝑣>𝐸𝑏,𝑠𝑒𝑚𝑖𝑐𝑜𝑛𝑑𝑢𝑐𝑡𝑜𝑟→ 𝑀𝑖𝑛𝑒𝑟𝑎𝑙𝑠 (𝑒𝑔. 𝐶𝑂2, 𝑚𝑖𝑛𝑒𝑟𝑎𝑙 𝑎𝑐𝑖𝑑𝑠) (2.3)

Depending on the original form of the organic species under consideration, a variety of

intermediates can be formed during the reaction between it and the photoinduced oxidative holes

and the various ROS species formed upon irradiation (Malato et al., 2009). Eventually, these

intermediates are degraded first to carboxylic acids and later to CO2 and water.

Photocatalytic degradation is usually modeled using either pseudo-first-order kinetics or the

Langmuir-Hinshelwood model. Pseudo-first-order kinetics imply that the reaction rate (r) is

equal to the rate of disappearance of one of the reactants:

𝑟 =−𝑑𝐶

𝑑𝑡= −𝑘𝑟𝐶 (2.4)

The reaction constant, kr, can be determined by plotting ln C vs. t.

In reality, however, heterogeneous catalytic reactions consist of two steps, adsorption of the

reactants to the catalyst and a catalyzed reaction. The first is assumed to be quick, while the rate

of the second is slower and will vary depending on reaction conditions (Ollis, 2013).

𝐴 + 𝑆 𝑘1→𝐴𝑆

𝑘−1→ 𝐴 + 𝑆 (2.5)

𝐴𝑆𝑘𝑟→ 𝑃𝑟𝑜𝑑𝑢𝑐𝑡𝑠 (2.6)

The Langmuir-Hinshelwood model, which accounts for both of these steps suggests that the

photocatalytic reaction rate (r) is proportional to the fraction of surface coverage by the organic

contaminant (), the reaction rate constant (kr), the concentration of the contaminant (C), and the

Langmuir adsorption constant (K):

rdC

dtkr

krKC

1KC (2.7)

12

The Langmuir-Hinshelwood model can often be simplified to the pseudo-first order model when

operating at low contaminant concentrations because adsorption is less likely to have an impact

on overall removal rates under these conditions (Huang et al., 2008). The rate constants for

reactions between •OH radicals and most organic compounds have been found to be on the order

of 106 1/M.s to 109 1/M.s (Malato et al., 2009).

The initial concentrations of the reactant(s), the mass of catalyst added, the wavelength of light

applied, the characteristics of the water matrix, pH, temperature, irradiance (i.e. radiant flux), and

oxygen levels affect the rate of photocatalytic degradation (Malato et al., 2009; Herrmann,

2012). pH is also important as it impacts the zero point charge of the catalyst surface and the

target compounds, which in turn affects adsorption efficiency and thus reaction rate (Malato et

al., 2009).

Titanium Dioxide

Titanium dioxide (TiO2) is a widely studied semiconductor that has become the default catalyst

used in solar and UV photocatalytic water treatment applications because it can be used

repeatedly, is chemically and thermally stable, and has strong mechanical properties (Malato et

al., 2009). Fine and nanosize TiO2 particles are also widely used in industries ranging from

personal care products to industrial coatings to textiles (Yang and Westerhoff, 2014). The

behaviour of TiO2 nanomaterials in these and other applications is determined by their size and

crystal structure as well as the medium in which they are suspended.

2.2.1 TiO2 Structure, Polymorphs, and Behaviour

TiO2 naturally occurs in three phases, also referred to as crystal structures or polymorphs,

anatase, brookite, and rutile. Anatase and rutile are the most common and best characterized

polymorphs. Despite its higher band gap energy, anatase is widely held to be the more

photoactive form of TiO2. Luttrell et al. (2014) summarized the most widely accepted reasons for

this phenomenon:

13

1. Anatase’s higher band gap energy results in a higher valence band that may be above the

redox potential of adsorbed molecules. This facilitates electron transfer and overall

reactivity.

2. The crystalline structures of the different polymorphs might influence the chemical and

physical characteristics of the nanoparticle surface and thus its ability to adsorb

molecules and trap photogenerated holes and electrons.

3. Anatase has an indirect band gap that is smaller than its direct band gap and the direct

and indirect band gaps of rutile. This makes it easier for anatase to maintain charge

separation, thus increasing the change of reaction between charge carriers and adsorbed

species.

4. The anatase structure is more conducive to the transport of charge carriers than that of

rutile, so charge carriers are more likely to diffuse from the bulk of the particle to the

surface before recombination.

Luttrell et al.’s own results support the hypothesis that charge carriers (“excitons”) generated

within the bulk of anatase particles are able to travel to the surface more easily than those

generated within the bulk of rutile particles because the crystalline structure of the former

facilitates exciton transport. Mixed phase materials such as the ubiquitous P25 nanoparticles

contain multiple TiO2 polymorphs. The ratio of anatase to rutile in P25 has been reported to

range from 70:30 to 80:20. Many researchers believe that the coexistence of the two polymorphs

results in synergistic increases in overall reactivity, but others, notably Ohtani et al. (2010), have

denied the existence of this synergistic effect. Zheng et al. (2010) synthesized mixed phase linear

TiO2 nanomaterials (see Section 2.2.3) containing anatase and TiO2(B), a metastable polymorph

of TiO2, and found that these were more photocatalytically active than pure anatase or pure

TiO2(B) materials.

2.2.2 Probing the Behaviour of TiO2 Photocatalysts

Photocatalysts vary in terms of photoactivity and degradation ability based on their structure and

component atoms and probe compounds, including dyes and simple organic compounds, are

widely applied to evaluate the photoactive and degradative characteristics of novel

photocatalysts (Jenks, 2013). Methylene blue dye is by far the most common probe compound

used to assess the activity of TiO2 nanomaterials (Mills and MacFarlane, 2007), though stearic

14

acid and azo dyes are also popular. Indeed, the degradation of methylene blue dye forms the

basis of the ISO methodology Determination of photocatalytic activity of surfaces in an aqueous

medium by degradation of methylene blue (ISO 10678:2010), which is described in detail by

Mills (2012). Unfortunately, this test does not differentiate between different dye degradation

mechanisms and its results cannot be used to quantify the extent of •OH radical formation in a

photocatalytic system.

Until recently, the •OH radical has been assumed to be the dominant active species in TiO2

photocatalysis (Nosaka and Nosaka, 2013), and many researchers have proposed alternative,

hydroxyl radical-specific, testing methods for TiO2 nanomaterials. These include a method

proposed by Sun and Bolton (1996) in which the degradation of methanol to formaldehyde is

tracked using high pressure liquid chromatography and compared to the delivered UV dose,

determined using a UV-Vis analyzer equipped with an integrating sphere, to determine a

quantum yield relating the number of photons added to the system to the extent of formation of

hydroxyl radicals. Other methods that have been developed for the detection of •OH radicals

include electron paramagnetic (spin) resonance, p-chlorobenzoic acid degradation and detection

of the resulting compounds with LC-MS/MS, laser induced fluorescence with emission

monitoring at 310 nm, or degradation of terephthalic acid with fluorescence monitoring (Han et

al., 2002; Nosaka and Nosaka, 2013). Recent studies have suggested that •OH radicals are not in

fact the dominant oxidative species in many photocatalytic systems (Nosaka and Nosaka, 2013)

and as such the methods discussed above, which focus specifically on the •OH radical, are

unlikely to be sufficient to characterize the behaviour of TiO2 photocatalytic systems because

they do not account for the presence and activity of other ROS (e.g. superoxide radical) or

surface phenomena.

Unlike other advanced oxidation processes, which rely on one or a small set of active ROS

species generated in situ, photocatalytic systems are complicated by their reliance on two surface

phenomena: Adsorption and the presence and formation of highly oxidative electron holes when

the material is irradiated with light of the proper wavelength. Adsorption, which must take place

before a contaminant or probe compound can be degraded, can result in apparent reduction of the

reactant even in the absence of light. Electron holes further complicate analysis because, like

ROS, they are highly oxidative and contribute to the overall degradation of organic contaminants

in solution. Different degradation mechanisms can lead to the formation of different intermediate

15

products and the relative contributions of •OH radicals, other ROS, and surface phenomena can

be elucidated with careful experimentation (Jenks, 2013). This process is necessarily complex

and rarely attempted except in detailed mechanistic studies. In applied studies, where the focus is

on the development of useable materials and the pathway(s) contributing to degradation are of

secondary interest, dyes and other simple probes provide a reasonable estimate of photocatalytic

activity and a point of comparison to the work of other researchers (Mills and MacFarlane,

2007). These simplified methods, however, can be complicated by photolytic degradation of the

probe molecule and/or non-oxidative contributions to probe molecule transformation (Jenks,

2013) and should therefore be viewed as starting points rather than endpoints.

2.2.3 Effects of TiO2 Nanomaterials on Human and Environmental Health

Recent studies have raised concerns that photocatalytic nanomaterials, including TiO2

nanoparticles, have potential negative human health and environmental impacts. The anatase

crystal phase, which is highly photoactive and therefore has a higher oxidizing potential, has

been identified as the most dangerous form of TiO2 (Love et al., 2012). The oxidative properties

of TiO2 are not the only source of concern, however, as nanosize material present a risk based

simply on their small size (Shi et al., 2013), which impacts their transport through both the

environment and the human body as well as their reactivity in both situations.

TiO2 is widely used in food, medicine, and commercial products and there are therefore

numerous routes through which humans can be exposed to TiO2 nanomaterials. A thorough

review by Love et al. (2012) identified four routes of human exposure to TiO2 nanomaterials:

Ingestion

Inhalation

Transdermal delivery

Injection

Of these, inhalation is currently the route of exposure of greatest concern, though this is still up

for debate because of conflicting results in animal studies (Shi et al., 2013). Particles with

aerodynamic diameters of approximately 0.1 m are the least likely to be deposited in the human

respiratory tract, but smaller particles can travel to the trachea, bronchial tubes, and alveoli

(Kuempel et al., 2015). The likelihood of deposition, clearing, and negative human health

16

impacts is also affected by the shape of the particles. Fibrous particles with diameters below 1 to

3 m and aspect ratios equal to or greater than 3:1 are of particular concern (WHO, 1999; CDC,

2005), though the actual human health implications of nanosized spherical and fibrous particles

are still being explored (Kuempel et al., 2015). Although ingestion has not been linked to any

lethal health effects, a recent study identified sub-lethal impacts of TiO2 nanoparticles on

intestinal cells, suggesting that this route of exposure may lead to chronic health effects

(Koeneman et al., 2014) and dermal exposure to TiO2 nanoparticles has been linked to chronic

sub-lethal health effects in animal models (Wu et al., 2009).

TiO2 nanoparticles enter the municipal wastewater system when people wash off or dispose of

body care products that contain TiO2 (e.g. sunscreen, toothpaste) and when they wash textiles

impregnated with TiO2 (Yang and Westerhoff, 2014). Although the majority of TiO2 materials

that enter wastewater treatment plants are removed by the treatment processes and eventually

disposed along with other residual solids (Kiser et al., 2009), a substantial portion can also end

up in the plant’s effluent and eventually in the receiving water body (Yang and Westerhoff,

2014), where they may be ingested or otherwise interact with aquatic organisms. For example,

TiO2 nanoparticles have been shown to accumulate in the internal organs of rainbow trout after

ingestion (Ramsden et al., 2009), though the authors of this study noted that the resulting effects

on the health of the trout were comparable to those caused by other potentially toxic metals.

Once TiO2 nanomaterials have entered the aquatic environment, water matrix components such

as ions (e.g. calcium) and natural organic matter (NOM) can affect their eventual transport and

toxicity (Hotze et al., 2010). A comprehensive study by Liu et al. (2013) demonstrated that

agglomeration was always favoured when the pH was close to the isoelectric point (IEP) of the

nanomaterial in question, that the presence of calcium ions increased the likelihood of

nanoparticle agglomeration and subsequent settling by compressing the electrical double layer

surrounding the nanoparticles, that NOM decreased agglomeration and settling by increasing the

energy barrier preventing nanoparticle aggregation. Liu et al. also demonstrated that LENs and

nanoparticles behaved differently under similar environmental conditions, specifically that high

concentrations of calcium could actually stabilize LEN suspensions by shifting the apparent IEP

of the LENs. Seitz et al. (2016) found that NOM, particularly aromatic NOM, decreased the

toxicity of TiO2 nanoparticles towards Daphnia magna, a common indicator organism. The study

of the toxicity of nanomaterials (nanotoxicity) is a field that is still in its infancy, and the lack of

17

definitive evidence on the health and environmental effects of nanomaterials, including TiO2

nanoparticles, should not be taken as proof that these materials do not pose a risk to humans or

other organisms.

Given the potential negative impacts of TiO2 nanomaterials on human and environmental health,

it is imperative that any novel process employing such materials for water or wastewater

treatment ensure the complete removal of the materials from the treated water before it is sent to

the drinking water distribution system or returned to the environment. Treatment facilities and

personal safety equipment will also need to be designed to prevent operators from being exposed

to potentially toxic levels of nanoparticles via inhalation in the treatment plant.

2.2.4 TiO2 Photocatalysis for Drinking Water Treatment

To-date, most research into the use of TiO2 photocatalysis in water and wastewater treatment has

been for disinfection and the degradation of organic compounds such as dyes, pesticides,

pharmaceuticals, and other trace organics. Full-scale application of the technology remains

vanishingly rare, however, for the reasons outlined in Chapter 1 of this document.

2.2.4.1 Existing TiO2-based Water Treatment Systems

TiO2-based water and wastewater treatment systems have been installed in various parts of the

world, though almost exclusively at small scale. One of these, the Photo-Cat Water Purification

System by Purifics, was developed in Ontario and has been studied by numerous research groups

in North America including Gerrity et al. (2009), who explored its use for DBP precursor

destruction, and Benotti et al. (2009), who evaluated its ability to remove pharmaceuticals and

estrogenic activity from water. The process consists of a photocatalytic reactor illuminated by a

low pressure mercury UV lamp and a ceramic ultrafiltration membrane to separate the

photocatalyst from the treated water. Purifics currently has at least two small (0.1 – 1.9 MLD),

self contained Photo-Cat installations in the Southern United States for chromium IV removal or

1,4 dioxane removal from groundwater and have also been awarded contracts for leachate

treatment systems in Southern Ontario and Europe (Purifics, n.d).

A few manufacturers have also developed TiO2-based adsorbents. These are primarily applied

for heavy metals removal and, to the author’s knowledge, none make use of TiO2’s

18

photocatalytic properties. These include the AdsorbsiaTM from DOW and MetSorbTM from

Graver Technologies. Both are formulated to work in a pressurized packed bed configuration and

most of the installations listed on their websites are for small communities, schools, and

commercial operations (Dow, n.d.; Graver Technologies, 2015).

The Plataforma Solar de Almeria, a pilot testing facility for solar-based energy and water and

wastewater technologies in Spain, has been operating since the 1980s. Dozens, if not hundreds,

of studies have been published based on the findings of solar photocatalysis-based projects

completed at the facility but to the author’s knowledge, none of these technologies have been

commercialized for point of use, small, or full-scale drinking water treatment.

2.2.4.2 Hybrid TiO2 Water Treatment Systems

Photocatalytic technologies can provide concurrent disinfection and NOM reduction, and as

such, may make excellent pre- or post-treatment options for existing water treatment processes

such as biologically active carbon filtration (BAC), enhanced coagulation, chlorine disinfection,

or membrane filtration. The oxidation of NOM via photocatalysis results in the formation of

smaller compounds, some of which may be more or less biodegradable (Toor and Mohseni,

2007; Philippe et al., 2010A; Metz et al., 2012), more or less adsorbable by activated carbon,

(Philippe et al., 2010A), more or less likely to cause membrane fouling (Huang et al., 2008), or

more or less reactive towards chlorine (Wisznioski et al., 2002; Toor and Mohseni, 2007; Bond

et al., 2009; Gerrity et al., 2009; Philippe et al., 2010B; Liu et al., 2010) than the original

compounds.

2.2.4.3 Reactor Design

The central role of surface reactions in photocatalytic degradation means that the reactor

configuration will necessarily be determined by the configuration and placement of the light

source and the characteristics of the photocatalyst itself. TiO2 is usually applied to water as

nanoparticles because their high surface area/mass ratio (specific surface area) maximizes the

total surface area available for photon and reactant adsorption (Nakata and Fujishima, 2012).

Systems that employ TiO2 can be challenging to operate because the nanoparticles can prevent

the passage of light through the suspension and also because it can be difficult to separate the

suspended particles or nanoparticles from the water after treatment (Ochiai and Fujishima, 2013).

19

Separation is usually achieved using membrane filtration after photocatalytic treatment (Malato

et al., 2009) and commercially available TiO2 nanoparticles (e.g. Evonik Degussa P25

nanoparticles), which usually have diameters in the tens of nanometers, can plug the membrane

filters by depositing in the membrane pores, which restricts the flow of water through the

membrane (Zhang et al., 2009).

The immobilization of TiO2 on solid supports, including magnetic particles (Ng et al., 2014),

zeolites (Liu et al., 2014), solid and hollow glass beads (Daneshvar et al., 2015; Wang et al.,

2013; Denny et al., 2008; Kim and Lee, 2005), the internal surface of tubular borosilicate glass

reactors (Alousan et al., 2012) and many other surfaces as described by Robert et al. (2013), has

been widely reported over the past two decades, however, only a few of these have been scaled

up to pilot or full-scale. In immobilized TiO2 designs the main driver, irrespective of light

source, is the need to minimize mass transfer resistance by maximizing the illuminated surface

area and decreasing the thickness of the water layer washing over that surface. Proposed full-

scale reactor designs that have been built at the bench- or pilot-scale include the rotating disk

photocatalytic reactor, the annular flow photoreactor, the packed bed photoreactor, and the fixed

bed sloping filter (as described in Dionysiou et al., 2000). All of these aimed to achieve

maximum contact between contaminants in the water and the catalyst surface while allowing for

adequate illumination. Another important challenge has been the ever present risk of TiO2

becoming detached from the support and traveling into the treated water (Robert et al., 2013).

An alternative to immobilization may be the development of engineered TiO2 nanomaterials that

are large, heavy, or buoyant enough to be removed via common clarification processes such as

filtration, sedimentation, and flotation. Such materials can be engineered using different

synthesis processes including various hydrothermal methods, electrochemical/anodic deposition,

and a variety of template-based methods (Bavykin and Walsh, 2009).

Engineered TiO2 Nanomaterials

Engineered nanomaterials such as carbon nanotubes have been a mainstay of the nanotechnology

industry for many years. TiO2-based engineered nanomaterials including nanotubes, nanowires,

and nanospheres, though less well known, have been synthesized and characterized by research

20

groups around the world in recent years and hold promise for numerous applications including

sensors and solar cells (Bavykin and Walsh, 2010).

2.3.1 Linear Engineered TiO2 Nanomaterials (LENs)

The foundational paper on linear engineered TiO2 nanomaterials (LENs) was published by

Kasuga et al. in 1998. In this paper, the authors describe the synthesis of TiO2 nanotubes using a

two-step hydrothermal method. Since this paper was published, researchers throughout the world

have developed numerous other LENs including nanobelts (large, flatter nanotubes), nanowires

(longer, thinner nanotubes), and nanorods (solid, cylindrical nanotubes).

Like nanoparticles, LENs have a high specific surface area but, unlike nanoparticles, they may

be large enough to remove via filtration and/or sedimentation. At least one research group has

coupled LENs with membrane filtration for water treatment and found that the LENs were less

likely to restrict the flow of water through the membrane over time (Zhang et al., 2009). LENs

can also be further engineered to form mats or membranes that are more easily incorporated into

environmental applications (Nakata and Fujishima, 2012). For example, Liu et al. (2012)

constructed a TiO2 nanofibre membrane for drinking water disinfection that was able to remove

bacteria through a combination of dead-end filtration and cell oxidation. Hu et al. (2011)

developed nanowire membranes (NWMs) with a nominal pore size of 100 nm, which is

comparable to that of a MF membrane. The NWMs lacked mechanical strength but were able to

degrade a selection of pharmaceutical products in batch mode under UV irradiation. A more

recent paper by Hu et al. (2013) reported on the development of more complex 3-D membrane

structures constructed of TiO2 nanobelts overlaid with TiO2 nanoparticles. LENs can also be

“grown” on solid surfaces via electrochemical/anodic deposition, avoiding the complications of

immobilizing nanoparticles onto a solid support (Bavykin and Walsh, 2009).

2.3.2 Alkaline Hydrothermal Synthesis of LENs

The alkaline hydrothermal method for TiO2 LEN synthesis is well established, does not require

highly specialized laboratory equipment or expensive reagents, and is easily manipulated to yield

nanosize materials with different morphological and chemical characteristics (Bavykin and

Walsh, 2010). The three main steps of this method are depicted in Figure 2.3 and summarized in

the subsections that follow.

21

Figure 2.1 Alkaline hydrothermal synthesis method as described by Kasuga et al.

(1999), Wong et al. (2011), Liang (2014), and others

2.3.2.1 Step 1: Hydrothermal Synthesis

LENs can be synthesized from many different TiO2 precursors, but anatase powders and

commercial P25 nanoparticles are the most commonly used. The precursor is added to a Teflon-

lined container along with a small volume of strong alkaline solution, usually 10 M NaOH. The

container is secured inside a steel reactor, heated to between 100oC and 250oC, and allowed to

react for times ranging from 20 h to multiple days. The overall reaction that occurs during the

hydrothermal synthesis step can be summarized as shown below:

2 𝑇𝑖𝑂2 + 3 𝑁𝑎𝑂𝐻 → 𝑁𝑎2𝑇𝑖3𝑂7 + 𝐻2𝑂 (2.8)

Essentially, the hydrothermal step breaks the existing Ti-O-Ti bonds and replaces them with Ti-

O-Na bonds and Ti-O-OH bonds (Kasuga et al., 1999).

22

2.3.2.2 Step 2: Ion Exchange

The products of the hydrothermal process are removed from the autoclave reactor and rinsed

numerous times with purified water. This rinsing process cause Na+ ions to be exchanged for H+

ions, resulting in the formation of Ti-O-H bonds (Kasuga et al., 1999). Rinsing continues until a

predetermined point (pH ~ 7, conductivity = 70 S/cm, 1.2 L of water, etc.). The rinsed materials

are then immersed in 0.1 N HCl and the H+ ions released by HCl dehydrate the Ti-OH bonds,

resulting in the reformation of Ti-O-Ti or Ti-O-H-O-Ti bonds. After one hour the materials are

removed from the acid bath via filtration or centrifugation and then rinsed again to remove any

remaining Na+ ions.

The ion exchange process is summarized in the following chemical equation:

𝑁𝑎2𝑇𝑖3𝑂7 + 2 𝐻𝐶𝑙 → 𝐻2𝑇𝑖3𝑂7 + 𝑁𝑎𝐶𝑙 (2.9)

Some debate remains as the importance of rinsing and ion exchange as well as the mechanisms

underlying the gradual formation of LENs during these steps (Wong, 2009). More recent studies

have suggested that sheet formation occurs during the hydrothermal step, partially explaining the

dependence of LEN length on the conditions of the hydrothermal step, while the substitution of

H+ for Na+ ions occurs during the ion exchange step (Liang, 2014).

2.3.2.3 Step 3: Calcination

Calcination is the process of heating the materials to temperatures above 300oC to modify their

crystal phase structure. The rinsed and dried materials are crushed to increase the overall surface

area available for reaction and then placed in a furnace capable of reaching temperatures between

300oC and 1,000oC. Depending on the temperature chosen and the crystalline structures

remaining after ion exchange and rinsing, the LENs will be converted to various titanate

structures, TiO2(B) (a metastable form of TiO2), anatase, rutile, or some combination thereof

upon calcination. For example, the conversion of trititanate to anatase occurs at 700oC as shown

in the equation below:

𝐻2𝑇𝑖3𝑂7 700𝑜𝐶→ 3 𝑇𝑖𝑂2 + 𝐻2𝑂 (2.10)

23

2.3.3 Effects of Synthesis Parameters on Nanomaterial Properties

Survey studies by Yuan and Su (2004), Wong et al. (2011), Qamar et al. (2008), and others have

established that the precursor materials, hydrothermal synthesis temperature, extent and method

of post synthesis cleaning and ion exchange, and calcination temperature have important effects

on the final products. The choice of precursor materials and hydrothermal temperature affects the

overall size and aspect ratio of the linear nanomaterials, with higher temperatures generally

resulting in larger materials (Yuan and Su, 2004). The washing, ion exchange, and calcination

steps affect the surface and crystalline structures of the materials, and thus their photocatalytic

properties (Qamar et al., 2008; Ali et al., 2016). Appendix A of this documents contains a

detailed review of the effects of these different parameters on the characteristics of TiO2 LENs.

Natural Organic Matter

2.4.1 Health, Operational, and Aesthetic Effects of Natural Organic Matter

The term NOM refers to a heterogenous mix of compounds formed during the degradation of

plants and other detritus within the watershed. It is ubiquitous in natural surface water and one of

the main targets of modern water treatment systems. Although NOM does not present any risk to

human health in its natural form, some NOM compounds can react with chlorine and other

oxidants used to disinfect and purify drinking water to yield disinfection byproducts (DBPs).

Studies conducted the 1970s established that halogenated compounds were formed when NOM

present in treated water came into contact with chlorine used for disinfection (Rook et al., 1974;

Bellar et al., 1974) and numerous studies since then have identified potential human health risks

related to these DBPs based on in vitro and animal experiments (Richardson et al., 2007). Recent

re-evaluation of these studies in light of data collected in the intervening decades has cast some

doubt on the well accepted theory that regulated DBPs such as THMs and HAAs present a risk to

human health at the levels commonly detected in drinking water (Hrudey, 2009). Nonetheless,

some unregulated DBPs have been shown to be highly genotoxic (Richardson et al., 2007;

Krasner et al., 2009) and these can be formed under the same conditions that encourage the

formation of regulated DBPs (Zheng et al., 2015; McKie et al., 2015). DBPs are also widely

24

regulated across the world, and as a result, much effort is expended in designing and operating

water treatment plants to remove NOM ahead of disinfection.

There are also non-health related concerns related to NOM in drinking water, including

operational effects on water treatment unit processes. Membrane filters are easily fouled by

NOM, which reduces their permeability and increases the frequency of backwashing and

chemical cleaning and the attendant power and chemical costs. NOM also interferes with

primary disinfection processes such as germicidal UV and chlorination by decreasing the UV

transmittance of the water and exerting a chlorine demand. It also exerts an oxidant demand

during ozonation and other oxidation-based processes used for contaminant degradation. Finally,

NOM frequently imparts a brown or yellow colour to the water that is an aesthetic concern for

drinking water consumers.

2.4.2 NOM Removal in Drinking Water Treatment Plants

NOM can be removed from drinking water via coagulation (with or without pH control),

adsorption onto activated carbon, and in some cases, via size exclusion using high pressure

membrane filters. It can also be transformed via oxidation processes into compounds that are

more easily removed by downstream processes like biofiltration or less likely to form DBPs

upon chlorination. Coagulation usually removes between 10 and 70 percent of NOM from raw

surface water (White et al., 1997) and the extent of removal is strongly impacted by the

characteristics of the overall water matrix including NOM type and concentration (Jacangelo et

al., 1995; White et al., 1997; Marhaba and Pipida, 2000; Wassink et al., 2011; Zheng et al., 2015)

as well as the operational parameters of the treatment process including pH and alkalinity

adjustment, coagulant dose, and mixing time (Jacangelo et al., 1995; Edzwald and Tobiason,

1999). The residual metal salts used for coagulation can be recycled within the plant (Tobiason et

al., 1999; Gottfried et al., 2008), however, in practice they are usually disposed of after only a

single use. Activated carbon in granular (GAC) form has also been used for NOM removal for

many decades in North America, though PAC has also been explored as a NOM removal option.

GAC is used in fixed bed reactors while PAC is applied to the water before or during coagulation

(Chowdhury et al., 2013). Both processes can achieve high percent removals of NOM, including

DBP precursors, under some conditions (Jacangelo et al., 1995), however, GAC’s ability to

remove NOM declines over time and high PAC doses are often required to achieve treatment

25

objectives (Chowdhury et al., 2013). Nanofiltration, a high pressure, energy intensive membrane

filtration process, can achieve nearly complete removal of NOM via size exclusion (Jacangelo et

al., 1995; Itoh et al., 2001; de la Rubia et al., 2008) and has been successfully coupled to

ultrafiltration membranes and employed for DBP precursor removal in small systems in Canada

(Lamsal et al., 2012) and the United States (Lozier et al., 1997). Oxidation via ozonation or one

of the more recently developed advanced oxidation processes (AOPs) can enhance DBP

precursor removal across existing biologically active filters (Toor and Mohseni, 2007) and AOPs

such as UV/H2O2 and UV/O3 can also be used on their own for DBP precursor removal (Lamsal

et al., 2011). There is, however, some concern that under some conditions incomplete

mineralization during AOP treatment could result in an overall increase in DBPfp (Toor and

Mohseni, 2007; Bond et al., 2009). All of these methods have attendant challenges, mostly

related to high chemical or energy costs, and there is demand for alternative NOM removal

processes, for both small and large communities, that minimize these costs while still providing

adequate treatment.

2.4.3 Photocatalytic Degradation of Natural Organic Matter

Like other organic compounds, NOM is vulnerable to oxidation by ROS. Rate constants ranging

from 1 to 5 x 108 M-1s-1 have been observed between •OH radicals and NOM isolates. Although

it is often assumed that NOM removal is exclusively due to photocatalytic degradation by ROS

in TiO2-based systems, many researchers have observed additional removal in the absence of

light. This suggests that adsorption of NOM onto the photocatalyst is another important removal

mechanism in these systems (Wiszniowski et al., 2002; Liu et al., 2008). Adsorbed compounds

are then degraded upon illumination of the catalyst.

A study by Liu et al. (2008) determined that pH of the solution and the irradiation time have

important roles in the degradation of large molecules into smaller ones, with higher reductions

observed at higher pHs and longer irradiation times. Other matrix components also affect

degradation rates. For example, the presence of bicarbonate has been shown to reduce the rate of

NOM degradation. This reduction has been attributed to •OH radical scavenging as observed in

other AOP studies (Liao et al., 2001) or, alternatively, to an increase in particle agglomeration

leading to an overall reduction in available surface area for reaction (Autin et al., 2014). The

presence of Fe(III) and Cu(II) has been linked to increases in the degradation rates of model

26

compounds in some photocatalytic systems, though these effects were concentration and pH

dependent (Butler and Davis, 1993). Indeed, other researchers have observed catalyst fouling in

systems containing iron and manganese (Burns et al., 1999). Other ions, including phosphate,

nitrate, and chloride have also been found to inhibit the degradation of model organic

compounds by TiO2 photocatalysis (Chen et al., 1997; Burns et al., 1999). A table describing the

effects of common matrix components on the degradation of organic contaminants by TiO2 is

provided in Appendix B.

Liu et al. (2008B) investigated the effect of UVA/TiO2 treatment on the size and characteristics

of humic acids using high pressure size exclusion chromatography (HPSEC) and resin

fractionation. Their results suggested that UV/TiO2 treatment degraded large humic acid

molecules into smaller ones. Other researchers, including Huang et al. (2008) and Philippe et al.

(2010A) have also found that UV/TiO2 breaks large molecular weight compounds down to

smaller ones. Results from Huang et al. (2008) and Liu et al. (2008B) suggest that the resulting

intermediate(s) are more resistant to degradation than their parent compounds. In the latter study,

the concentration of charged and neutral hydrophilic compounds increased with increasing

irradiation time, suggesting that the hydrophobic compounds were being broken down to

hydrophilic intermediates by the photocatalytic process. The concentration of charged

hydrophilic acids eventually decreased, indicating that some portion of the original humic acid

content was fully mineralized. A 2010 study by the same research group (Liu et al., 2010)

employed a different fractionation method (LC-OCD) but found similar results.

In the Liu et al. study (2010) the researchers also observed that the DBPfp of the water initially

increased at shorter irradiation times (< 60 min) before decreasing at longer irradiation times,

suggesting that incomplete mineralization of larger or more recalcitrant NOM compounds at

shorter treatment times was causing a temporary overall increase in the concentration of DBP

precursors present in solution. The increase in DBPfp was only observed in one water source,

suggesting that this effect was matrix dependent. Other researchers have also observed this

increase in the overall DBPfp at short UV/TiO2 treatment times (Gerrity et al., 2009),

highlighting the risk of using photocatalysis for water treatment. Many of the researchers cited

above observed some degree of NOM removal via adsorption to TiO2 in their experiments. In

some cases, this removal was substantial (e.g. Huang et al., 2008), suggesting that adsorption

might be a viable alternative to photocatalysis for removal of NOM, including DBP precursors,

27

by TiO2. In such a system it might theoretically also be possible to harness the photocatalytic

properties of TiO2 to regenerate the adsorbent, resulting in a multi-use product that could be a

viable alternative to existing NOM and DBP precursor removal technologies.

Adsorption

2.5.1 Adsorption Theory

Adsorption is a process whereby one species becomes associated with a solid surface via

chemical or physical interactions. Çeçen and Aktaş (2011) describe the adsorption of aqueous

species, including NOM, as being controlled by two driving forces. The first is the affinity of the

target species toward water. Hydrophobic species are more likely to adsorb to a solid substrate in

water than hydrophilic species because they are seeking to get away from water molecules. The

other driving force is related to the chemical characteristics of the adsorbate and the adsorbent.

Adsorption has often been described using the following reversible chemical equation:

𝐴 + 𝑆𝑘↔𝐴𝑆 (2.11)

Where A represents the species being adsorbed (adsorbate) and S represents an adsorption site on

the solid surface (adsorbent).

The amount of adsorbate taken up by the adsorbent and rate at which the two become associated

are functions of the diffusion of the adsorbate from the bulk water through the hydrodynamic

layer and to the surface of the adsorbent (external mass transfer) and the actual adsorption

(physical or chemical) of the adsorbate to the surface of the adsorbate (Fogler, 2002). When the

adsorbent is porous, internal mass transfer must also be considered. In the case of TiO2

nanoparticles, which are not porous, internal mass transfer is not an important factor in

adsorption. More complex TiO2 nanostructures such as nanobelts and nanospheres are often

porous, however, so internal mass transfer is likely to affect the overall adsorption kinetics.

Depending on the characteristics of the system being investigated, any of these three processes

can be rate limiting. External mass transfer resistance can be minimized by mixing, and most

28

preliminary adsorption experiments are carried out in batch reactors mixed constantly at a

consistent rate to minimize the overall resistance to adsorption and make it easier to observe the

surface adsorption kinetics (Chowdhury et al., 2013).

Over time, the overall adsorption process tends towards a steady state known as the point of

equilibrium. Adsorption at the point at and beyond equilibrium is quantified using the parameter

qe, which is the mass of adsorbate adsorbed per unit of adsorbent (mg/g or mg/mg). qe is

calculated as follows:

𝑞𝑒 =𝐶𝑜−𝐶𝑒

𝐷 (2.12)

Where Co is the original concentration of the adsorbate in solution (mg/L), Ce is the

concentration of adsorbate in solution once the system has achieved equilibrium (mg/L), and D is

the dose of adsorbent added to the system (mg or g).

Adsorption processes are usually evaluated in the laboratory using adsorption isotherm models

developed theoretically based on the principles described above or empirically. The most

common adsorption models are the Langmuir and Freundlich isotherm models.

2.5.1.1 Langmuir Isotherm

The Langmuir isotherm is a well understood and accepted empirical adsorption model that is

widely used in water treatment applications. It can be written as follows:

𝑞𝑒 =𝑏𝐶𝑒𝑞𝑚𝑎𝑥

1+𝐶𝑒𝑏 (2.13)

Where qe is the mass ratio of adsorbed adsorbate to adsorbent at equilibrium (mg/g), qmax is the

maximum possible mass of adsorbate that can be adsorbed (mg/g), Ce is the concentration of the

adsorbate in the treated water (mg/L), and b (L/mg) is a constant related to the energy of

adsorption (Shahbeig et al., 2013). Higher values of b and qmax imply better adsorption.

As shown below, a plot of 1/qe vs. 1/Ce will be linear and have a slope of 1/KLqmax and a y

intercept of 1/qmax.

1

𝑞𝑒=

1

𝑞𝑚𝑎𝑥+

1

𝑏𝑞𝑚𝑎𝑥

1

𝐶𝑒 (2.14)

29

Assumptions of the Langmuir isotherm (Shahbeig et al., 2013) include monolayer adsorption,

irreversible adsorption, a homogeneous adsorbent interacting with homogeneous adsorption

sites, an adsorbent with a finite adsorption capacity for the adsorbate, and no interaction between

molecules adsorbed to neighbouring sites.

2.5.1.2 Freundlich Isotherm

A common alternative to the Langmuir isotherm is the Freundlich isotherm. The Freundlich

isotherm often fits well to empirical data and is often used to model heterogeneous systems such

as the adsorption of organic molecules to activated carbon (Summers et al., 1988; Çeçen and

Aktaş, 2011). Unlike the Langmuir isotherm, it can be used to describe multilayer adsorption,

reversible adsorption, and adsorbents with non-uniform adsorption sites (Shahbeig et al., 2013).

𝑞𝑒 = 𝐾𝐹𝐶𝑒1/𝑛

(2.15)

Where KF is the Freundlich adsorption constant ([mg/g]/[mg/L]1/n) and 1/n is the Freundlich

slope, an empirical parameter that describes the shape of the adsorption curve (Chowdhury et al.,

2013). The linearized form of the Freundlich isotherm is shown below.

log 𝑞𝑒 = log𝐾𝐹 +1

𝑛log 𝐶𝑒 (2.16)

To evaluate the fit of the Freundlich isotherm, log qe is plotted against log Ce. 1/n is the slope of

a line fitted to the data and KF is calculated from the y intercept of that line. When 1/n is constant

or nearly so, KF can be used as an indicator of adsorption capacity, with higher KF indicating

greater adsorption capacity (Chowdhury et al., 2013).

2.5.1.3 Modification of the Freundlich Isotherm Model to Account for a Fixed NOM Concentration

The adsorption behaviour of natural organic matter (NOM) onto drinking water adsorbents has

proven difficult to characterize because it is a dilute heterogenous mixture made up of polymeric

molecules of varying molecular weights and chemical characteristics. The composition of this

mixture varies based on source water characteristics and seasonal changes, making it difficult to

establish a representative Co value. Additionally, it is difficult to concentrate or dilute natural

water matrices without also modifying the concentrations of other water components, which

30

complicates the execution of standard adsorption experiments, which are usually conducted

using a constant adsorbent dose and changing initial concentration of adsorbate.

Summers and Roberts (1988) conducted a series of experiments to determine whether the

adsorption of NOM to activated carbon could be characterized if the NOM concentration was

held constant and the dose of activated carbon was varied. They found that a modified version of

the Freundlich isotherm could be used to describe the adsorption of NOM to activated carbon

when experiments were conducted with a constant initial concentration of NOM and changing

doses (D) of activated carbon. They developed the following equation to express this

relationship:

𝑞𝑒 = 𝐾𝐹(𝐶𝑒/𝐷)1/𝑛 (2.17)

Where D has units mg/L or g/L. This model can be linearized as follows:

log 𝑞𝑒 = log𝐾𝐹 +1

𝑛log(𝐶𝑒/𝐷) (2.18)

The modified Freundlich isotherm was developed to analyze the adsorption of NOM to activated

carbon, and assumes that the adsorbent interacts with a heterogeneous mixture of potential

adsorbates. It has been used to model the removal of NOM from drinking water by activated

carbon (e.g. Karanfil and Kitis 1999; Li et al., 2002), as well as nanoscale carbon adsorbents

(Hyung and Kim, 2008).

2.5.2 Adsorption of NOM to TiO2

TiO2 photocatalysis, particularly with standard P25 nanoparticles, is a well-studied phenomenon

but the adsorption of NOM to TiO2 materials is less well understood, particularly in terms of its

utility as a treatment technology. Most of the studies that have explored NOM adsorption to

TiO2, including those by Wiszniowski et al. (2002), Erhayem and Sohn (2014), Liu et al. (2013),

Thio et al. (2011), Domingos et al. (2009) and Mwaanga et al. (2014) have been conducted by

environmental chemistry researchers whose main concern was the transport of nanoparticles

through the natural environment.

31

2.5.2.1 Adsorption Time Studies

Many researchers operating in both the contaminant fate and transport context and the treatment

context have conducted kinetics studies to establish the amount of time required to reach

adsorption equilibrium. For example, Mwaanga et al. (2014) and Erhayem and Sohn (2014), who

were exploring the effects of NOM on nanparticle transport in the environment, both found that

adsorption initially occurred quickly but also noted a very small amount desorption occurring as

time went on. Both studies used UV254 as their response parameter. As a result, their adsorption

isotherm experiments were run for 48 h (Erhayem and Sohn, 2014) or 72 h (Mwaanga et al.,

2014). Other researchers, particularly those exploring the use of TiO2 for NOM removal from

drinking water, have come to different conclusions regarding the time to equilibrium. For

example, Huang et al. (2008) used a five minute adsorption period, which they chose based on

time to adsorption studies while Ng et al. (2014) allowed for 15 to 30 minutes of adsorption and

Liu et al. (2014) allowed for a full 24 hours of adsorption.

2.5.2.2 Adsorption Isotherm Fitting

Adsorption isotherms are widely used to describe and predict the adsorption of NOM to activated

carbon and other adsorbents used in drinking water treatment as well as to explain the

interactions between NOM and various adsorbing media present in the natural environment. A

subset of the TiO2 researchers cited in this thesis made efforts to fit their findings to common

adsorption isotherm models. The overall photocatalytic reaction between TiO2 and organic

contaminants under irradiation is usually modeled using the Langmuir-Hinshelwood model,

which assumes that adsorption precedes degradation and that the former occurs according to the

Langmuir model (Malato et al., 2009), at least during photocatalysis. Dark adsorption

(adsorption in the absence of photocatalysis) of NOM by TiO2 has also been successfully

modeled using the Langmuir isotherm by Liu et al. (2014). Results presented by Mwaanga et al.

(2014) were analyzed as part of this study and also determined to be a good fit to the Langmuir

isotherm model. Given the heterogeneity of NOM, other models, including the Freundlich

isotherm model and the modified Freundlich model, might also be appropriate. The results of Liu

et al. (2013) were equally well described by the Langmuir and Freundlich isotherms, those of

Sun and Lee (2012) were best described by the combined Langmuir-Freundlich isotherm model,

32

and Erhayem and Sohn (2014) successfully fitted their data to the modified Freundlich isotherm

model described in Section 2.5.1.3.

2.5.2.3 Matrix Effects on Adsorption

Studies by Mwaanga et al. (2014) and Erhayem and Sohn (2014) established that pH and ionic

strength had effects on adsorption and noted that larger, more aromatic NOM compounds were

adsorbed preferentially. In Mwaanga’s study the pH of the solution had an impact on the amount

of adsorption as well as the fit of the model. Erhayem (2013) observed a similar relationship

between pH and the mass of NOM adsorbed from the water and also noted that phosphate,

nitrate, and bicarbonate inhibited adsorption. At high pH, divalent ions (Mg2+, Ca2+) encouraged

more adsorption than monovalent ions. Agglomeration reduces the overall surface area available

for adsorption in the system and thus more of the adsorbate is likely to remain in solution when

the materials are agglomerated than if they are fully dispersed. Numerous researchers including

Liu et al. (2013), Loosli et al., (2015), Domingos et al. (2009), and Thio et al. (2011), have

explored the agglomeration of TiO2 nanoparticles in aqueous media and its effects and

dependence on NOM adsorption. A summary of the known effects of different matrix

components on the adsorption of NOM by TiO2 is provided in Table A.1 in Appendix A.

2.5.3 Nanoparticle Agglomeration

Nanoparticles agglomerate quickly when added to aqueous media, though the rate and extent of

agglomeration is impacted by ionic strength and pH (French et al., 2009) as well as the presence

of NOM and the size and shape of the nanomaterials themselves (Hotze et al., 2010). The

majority of the studies that have been conducted on TiO2 nanoparticle agglomeration in natural

or simulated natural water matrices have been motivated by the need to characterize the fate and

transport of these materials within the aquatic environment. Researchers have been particularly

keen to elucidate the mechanisms underlying nanoparticle sedimentation in order to predict the

likelihood that nanoparticles released into the environment will travel far enough to negatively

impact sensitive ecosystems.

Researchers including Liu et al., (2013) and Zhou et al. (2013) have attempted to explain their

agglomeration results in terms of Derjaguin-Landau-Verwey-Overbeak theory (DVLO theory),

which posits that particles in suspension interact via two main forces: charge interactions

33

between the electrical double layers (EDLs) of particles (repulsive) and van der Waals forces

(attractive). The distance between the individual particles governs which of these forces will

dominate and thus the likelihood that they will be attracted to or repelled by one another

(Crittenden et al., 2012). Compression of the EDL by ions in the aqueous medium reduces the

repulsive forces between individual nanoparticles, allowing van der Waal’s attractive forces to

dominate. This results in agglomeration and decreased available surface area, which can lead to

sedimentation or reductions in adsorption capacity and photocatalytic activity.

DVLO theory has gradually been extended to account for other forces that can affect particle

interactions (XDVLO theory):

Magnetic attraction

Hydrophobic interactions

Osmotic repulsion

Elastic-steric repulsion

Bridging attraction

As described by Hotze et al. (2013), DVLO and XDVLO assume that the bulk sizes of the

particles are much greater than those of their surfaces thus that the various interactions occur

between two flat surfaces. This assumption does not always hold for nanoparticles. The small

size of nanoparticles also has impacts on their surface charge and surface interactions because

more of their electrons exist on the surface rather that within the bulk. The concepts underlying

DVLO and XDVLO nonetheless provide a useful framework for evaluating the behavior of

nanoparticles.

As mentioned previously, most researchers who have studied TiO2 nanoparticle agglomeration

have done so in the interest of characterizing sedimentation behavior of various TiO2

nanomaterials under different water quality conditions. The rate at which a particle settles in

aqueous media is determined by three forces: buoyancy, drag, and gravity. All three forces are

affected by the diameter of the particle or agglomerate and the gravitational force is also

impacted by the density of the particle or agglomerate. The magnitudes of all three forces are

also impacted by the shape of the particle or agglomerate as this will impact its projected area

and volume (Crittenden et al., 2012). The effective density of a nanomaterial agglomerate can be

well below the density of that of the actual material (e.g. 4.26 g/cm3 for TiO2) because the

34

agglomerate contains entrapped media (Deloid et al., 2014). It can be inferred based on the sizes

of the agglomerate and the original nanomaterial along with the density of the original

nanoparticle and that of the suspending media using the Sterling equations (see Appendix H) or

determined empirically using a method proposed by Deloid et al. (2014). The shape and structure

of a nanomaterial can affect the shape and effective density of its agglomerates, and therefore its

sedimentation efficacy, as well as the concentration of ions required to induce agglomeration

(Hotze et al., 2010). In a treatment context, the shape and size of the agglomerates also affects

the likelihood that the material will be removed via other drinking water clarification processes

such as membrane filtration (Zhang et al., 2009).

Nanoparticle agglomeration and its effect on surface area can also contribute to the changes in

adsorption efficiency and photocatalytic degradation observed at different pHs and in the

presence of ions and NOM. As a rule, agglomeration and subsequent sedimentation are most

likely to occur when the pH is near the isoelectric point/point of zero charge of the material in

question because at this pH repulsive forces between individual particles are at a minimum (Liu

et al., 2013). Liu et al. (2013) reported that three types of TiO2 nanomaterials were more likely to

agglomerate under high ionic strength conditions than at low ionic strength conditions. This

finding is corroborated by those of Erhayem and Sohn (2014). The type of ions present in

solution may also have an effect – Liu et al. (2013) observed much greater increases in

agglomerate size when Ca2+ was added to the water rather than Na+. They hypothesized that this

was due to the greater ability of Ca2+ to compress the electrical double layer surrounding the

nanomaterials relative to Na+. Greater compression of the electrical double layer results in less

repulsion between individual nanoparticles and thus, greater agglomeration. Numerous

researchers have observed that the presence of natural organic matter increases the stability of

nanomaterials in solution, though this effect is less pronounced in the presence of ions such as

calcium (Zhang et al. 2009; Thio et al., 2011; Liu et al., 2013) and at high NOM concentrations

(Erhayem and Sohn, 2014). According to Zhang et al. (2009), NOM inhibits agglomeration by

increasing the overall negative charge of the particles and thus increasing the repulsive forces

that keep them dispersed in solution. Other researchers have come to different conclusion. For

example, Domingos et al. (2009) attributed NOM’s ability to stabilize nanomaterial suspensions

to increased steric repulsion. Thio et al. (2011) suggested that stabilization was due to both steric

repulsion and changes in electrostatic interactions and that in matrices that contain both NOM

35

and calcium, calcium ions can provide bridging of NOM-coated nanoparticles. The latter

hypothesis was also put forward by Liu et al. (2013).

References

Ali, S., Granbohm, H., Ge, Y., and Singh, V.K. 2016. Crystal structure and photocatalytic

properties of titanate nanotubes prepared by chemical processing and subsequent annealing,

Journal of Materials Science, 51, 7322-7335

Alrousan, D.M.A., Dunlop, P.S.M., McMurray, T.A., and Byrne, J.A. (2009) Photocatalytic

inactivation of E. coli in surface water using immobilized nanoparticle TiO2 films, Water

Research, 43, pp. 47-54

Alrousan, D.M.A., Polo-López, M.I., Dunlop, P.S.M., Fernández-Ibánez, Byrne, J.A. (2012)

Solar photocatalytic disinfection of water with immobilized titanium dioxide in re-circulating

flow CPC reactors, Applied Catalysis B, 128, 126-134

American Public Health Association (2005) Standard Methods for the Examination of Water and

Wastewater, 21st ed., Washington D.C., APHA

Autin, O., Hart, J., Jarvis, P., MacAdam, J., Parsons, S.A., Jefferson, B. (2013) The impact of

background organic matter and alkalinity on the degradation of the pesticide metaldehyde by two

advanced oxidation processes: UV/H2O2 and UV/TiO2, Water Research, 47, 2041-2049

Bavykin, D.V. and Walsh, F.C. (2009) Titanate and Titania Nanotubes: Synthesis, Preparation,

and Application, RSC Publishing

Bellar, T.A., Lichtenberg, J.J., and Kroner, R.C. (1974) The occurrence of organohalides in

chlorinated drinking waters, Journal of the American Water Works Association, 66 (12), 703-

706

Benotti, M.J., Stanford, B.D., Wert, E.C., Snyder, S.A. (2009) Evaluation of a photocatalytic

reactor membrane pilot system for the removal of pharmaceuticals and endocrine disrupting

compounds from water, Water Research, 43, 1513-1522

36

Bickley, R.I., Gonzalez-Carreno, T., Lees, J.S., Palmisano, L. and Tilley, R.J.D. (1991) A

structural investigation of titanium dioxide photocatalysts, Journal of Solid State Chemistry, 92,

p. 178

Bolton, J.R. and Cotton, C.A. (2008) Ultraviolet Disinfection Handbook, American Water Works

Association, Colorado, USA

Bond, T., Goslan, E.H., Jefferson, B., Roddick, F., Fan, L., Parsons, S.A. (2009) Chemical and

biological oxidation of NOM surrogates and effect on HAA formation, Water Research, 43, pp.

2615-2622

Burns, R.A., Crittenden, J.C., Hand, D.W., Selzer, V.H., Sutter, L.L., Salman, S.R. (1999) Effect

of inorganic ions in heterogeneous photocatalysis of TCE, Journal of Environmental

Engineering, 125 (1), 77-85

Butler, E.C. and Davis, A.P. (1993) Photocatalytic oxidation in aqueous titanium dioxide

suspensions: The influence of dissolved transition metals, Journal of Photochemistry and

Photobiology: Chemistry, 70, 273-283

Çeçen and Aktaş (2011) Activated Carbon for Water and Wastewater Treatment, Weinheim,

Germany, Wiley-VCH Verlag and Co.

Centers for Disease Control and Prevention, Department of Health and Human Services,

National Institute for Occupational Safety and Health (2006) Occupational Exposure to

Refractory Ceramic Fibers, Criteria for a Recommended Standard, DHHS (NIOSH) Publication

No. 2006-123

Chen, B., Lee, W., Westerhoff, P., Krasner, S., and Herckes, P. (2010) Solar photocatalysis

kinetics of disinfection byproducts, Water Research, 44 (11), pp. 3401-3409

Chen, D. and Ray, A.K. (2001) Removal of toxic metal ions from wastewater by semiconductor

photocatalysis, Chemical Engineering Science, 56, 1561-1570

Chen, H.Y., Zahraa, O., Bouchy, M. (1997) Inhibition of the adsorption and photocatalytic

degradation of an organic contaminant in an aqueous suspension of TiO2 by inorganic ions,

Journal of Photochemistry and Photobiology A: Chemistry, 108, 37-44

37

Chen, V., Mansouri, J. and Charlton, T. (2010) Biofouling in Membrane Systems, in:

Membranes for Water Treatment, Eds. Peinemann, K-V and Nunes, S.P., Wiley-VCH

Chen, W., Qian, C., Liu, X-Y, and Yu, H-Q (2014) Two-dimensional correlation spectroscopic

analysis on the interaction between humic acids and TiO2 nanoparticles, Environmental Science

and Technology, 48, pp 11118-11126

Cho, M., Chung, H., Choi, W., and Yoon, J. (2004) Linear correlation between inactivation of

E.coli and OH radical concentration in TiO2 photocatalytic disinfection, Water Research, 38, pp.

1069-1077

Chong, M.N., Jin, B., Chow, C.W.K., and Saint, C. (2010) Recent developments in

photocatalytic water treatment technology: A review, Water Research, 44 pp. 2997-3027

Chow, A., Leech, D., Boyer, T., and Singer, P. (2008) Impact of simulated solar irradiation on

disinfection byproduct precursors, Environmental Science and Technology, 42 (15), pp. 5586-

5593

Chowdhury, Z.K., Summers, R.S., Westerhoff, G.P., Leto, B.J., Nowack, K.O., and Corwin, C.J.

(2013) Activated Carbon: Solutions for Improving Water Quality, Passantino, L.B. (Ed.),

Denver, USA, American Water Works Association

Crittenden, J., Trussell, R., Hand, D., Howe, K., and Tchobanoglous (2012) MWH’s Water

Treatment: Principles and Design, 3rd ed., John Wiley and Sons, Hoboken, NJ

Daneshvear, N., Salari, D., Niaei, A, Rasoulifard, M.H., (2005) Immobilization of TiO2

nanopowder on glass beads for the photocatalytic decolourization of an azo dye C.I. Direct Red

23, Journal of Environmental Science and Health, 40, 1605-1617

De la Rubia, Á, Rodríguez, M., Léon, V.M., Prats, D. (2008) Removal of natural organic matter

and THM formation potential by ultra- and nanofiltration of surface water, Water Research, 42,

714-722, doi:10.1016/j.watres.2007.07.049

Deloid, G., Cohen, J.M., Darrah, T., Derk, R., Rojanasakul, L., Pyrgiotakis, G., Wohlleben, W.,

Demokritou, P. (2014) Estimating the effective density of engineered nanomaterials for in vitro

dosimetry, Nature Communications, 5, 3514 doi: 10.1038/ncomms4514

38

Deng, D., Martin, S.T., and Ramanathan, S. (2011) Synthesis of hollow porous nanospheres of

hydroxyl titanium oxalate and their topotactic conversion to anatase titania, Journal of Materials

Research, 26 (12), pp. 1545-1551

Denny, F., Scott, J., Pareek, V., Peng, G.D., Amal, R. (2009) CFD modelling for a TiO2-coated

glass-bead photoreactor irradiated by optical fibres: Photocatalytic degradation of oxalic acid,

Chemical Engineering Science, 64, 1695-1706

Domingos R.F., Tufenkji, N., and Wilkinson, K.J. (2009) Aggreation of titanium dioxide

nanoparticles: Role of fulvic acid, Environmental Science and Technology, 43, 5, 1282-1286,

doi: 10.1021/es8023594

DOW (2012) AdsorbsiaTM As600 Titanium Based Media, Accessed May 5, 2017:

http://msdssearch.dow.com/PublishedLiteratureDOWCOM/dh_08d7/0901b803808d75bc.pdf?fil

epath=liquidseps/pdfs/noreg/177-02512.pdf&fromPage=GetDoc

Edzwald, J.K., Becker, W.C., and Wattier, K.L. (1985) Surrogate parameters for monitoring

organic matter and THM precursors, Journal AWWA, 77 (4), pp. 122-132

Edzwald, J.K. and Tobiason (1999) Enhanced coagulation: US requirements and a broader view,

Water Science and Technology, 40 (9), 63-70

Erhayem, M. (2013) Effect of naturally occurring organic matter (NOOM) type and source on

NOOM adsorption onto titanium dioxide nanoparticles under varying environmental conditions,

Thesis, Florida Institute of Technology, USA

Erhayem, M. and Sohn, M. (2014) Effect of humic acid source on humic acid adsorption onto

titanium dioxide nanoparticles, Science of the Total Environment, 470-471, pp.92-98




Field, R. (2010) Fundamentals of Fouling, in: Membranes for Water Treatment, Eds. Peinemann,

K-V and Nunes, S.P., Wiley-VCH

http://msdssearch.dow.com/PublishedLiteratureDOWCOM/dh_08d7/0901b803808d75bc.pdf?filepath=liquidseps/pdfs/noreg/177-02512.pdf&fromPage=GetDoc

http://msdssearch.dow.com/PublishedLiteratureDOWCOM/dh_08d7/0901b803808d75bc.pdf?filepath=liquidseps/pdfs/noreg/177-02512.pdf&fromPage=GetDoc

39

Fogler, H.S. (2002) Elements of Chemical Reaction Engineering, New Jersey, USA, Prentice

Hall

French, R.A., Jacobson, A.R., Kim, B., Isley, S.L., Penn, R.L., Baveye, P.C. (2009) Influence of

ionic strength, pH, and cation valence on aggregation kinetics of titanium dioxide nanoparticles,

Environmental Science and Technology, 43, 1354-1359



Research, 43, pp. 1597-1610

Gottfried, A., Shepard, A.D., Hardiman, K., Walsh, M.E. (2008) Impact of recycling filter

backwash water on organic removal in coagulation-sedimentation processes, Water Research,

42, 4683-4691

Graver Technologies (2015) MetSorb from Graver Technologies, Accessed May 5, 2017:

http://www.gravertech.com/PDF/MetSorb/lit/metsorb.pdf

Han, S-K, Nam, S-N, Kang, J-W (2002) OH radical monitoring technologies for AOP advanced

oxidation process, Water Science and Technology, 46 (11-12), 7-12,

Hashimoto, K., Irie, H., and Fujishima, A. (2005) TiO2 photocatalysis: A historical overview and

future prospects, Japanese Journal of Applied Physics, 44 (12), pp. 8269-8285

Henderson, M. (2011) A surface science perspective on TiO2 photocatalysis, Surface Science

Reports, 66 (6-7), pp. 185-297

Herrmann, J-M. (2010) Photocatalysis fundamentals revisited to avoid several misconceptions,

Applied Catalysis B: Environmental, 99, pp. 461-468

Hotze, E.M., Phenrat, T., and Lowry, G.V. (2010) Nanoparticle aggregation: Challenges to

understanding transport and reactivity in the environment, Journal of Environmental Quality, 39,

1909-1924, doi:10.2134/jeq2009.0462

Hrudey, S. (2009) Chlorination disinfection by-product, public health risk tradeoffs, and me,

Water Research, 43, 2057-2092

40

Hu, A., Liang, R., Zhang, X., Kurdi, S., Luong, D., Huang, H., Peng, P, Marzbanrad, E., Oakes,

K., Zhou, Y., and Servos, M. (2013) Enhanced photocatalystic degradation of dyes by TiO2

nanobelts with hierarchical structures, Journal of Photochemistry and Photobiology A:

Chemistry, 256, pp. 7-15

Hua, G. and Reckhow, D.A. (2007) Characterization of disinfection byproduct precursors based

on hydrophobicity and molecular size, Environmental Science and Technology, 41 (9), pp. 3309-

3315



pp. 1142-1150

Hyung, H. and Kim, J-H (2008) Natural organic matter (NOM) adsorption to multi-walled

carbon nanotubes: Effect of NOM characteristics and water quality parameters, Environmental

Science and Technology, 42, 4416-4421

Itoh, M., Kunikane, S., Magara, Y. (2001) Evaluation of nanofiltration for disinfection by-

products control in drinking water treatment, Water Science and Technology: Water Supply, 1

(5), 233-243

Jacangelo, J.G., DeMarco, J., Owen, D.M., and Randtke, S.J. (1995) Selected processes for

removing NOM: An overview, Journal of the American Water Works Association, 87 (1), 64-77

Jenks, W.S. (2013) Photocatalytic reaction pathways: Effects of molecular structure, catalyst,

and wavelength, In: Photocatalysis and Water Purification: From Fundamentals to Recent

Applications, Ed. Pichat, P., Wiley-VCH Verlag GmbH and Co.

Karanfil, T. and Kitis, M. (1999) Role of granular activated carbon surface chemistry on the

adsorption of organic compounds. 2. Natural organic matter, Environmental Science and

Technology, 33, 3225-3233



41

Kim, S-C and Lee, D-K (2005) Preparation of TiO2-coated hollow glass beads and their

application to the control of algal growth in eutrophic water, Microchemical Journal, 80, pp.

227-232

Kim, S-H and Shon, H.K. (2007) Adsorption characterization for multi-component organic

matters by titanium oxide (TiO2) in wastewater, Separation Science and Technology, 42, pp.

1775-1792

Kiser, M.A., Westerhoff, P., Benn, T., Wang, Y., Pérez-Rivera, Hristovski, K. (2009) Titanium

nanomaterial removal and release from wastewater treatment plants, Environmental Science and

Technology, 43, 6757-6763 doi: 10.1371/journal.pone.0081239

Koeneman, B.A., Zhang, Y., Westerhoff, P., Chen, Y., Crittenden, J.C., and Capco, D.G. (2014)

Toxicity and cellular responses of intestinal cells exposed to titanium dioxide, Cell Biology and

Toxicology, 26, 225-238

Krasner, S.W. (2009) The formation and control of emerging disinfection by-products of health

concern, Philosophical Transactions of the Royal Society A, 367, 4077-4095

Kuempel, E.D., Sweeney, L.M., Morris, J.B., Jarabek, A.M. (2015) Advances in inhalation

dosimetry models and methods for occupational risk assessment and exposure limit derivation,

Journal of Occupational and Environmental Hygiene, 12, S18-S40

Lamsal..R., Walsh, M.E., Gagnon, G.A. (2011) Comparison of advanced oxidation processes for

the removal of natural organic matter, Water Research, 45, 3263-3269

Lamsal, R., Montreuil, K.R., Kent, F.C., Walsh, M.E., Gagnon, G. (2012) Characterization and

removal of natural organic matter by an integrated membrane system, Desalination, 303, 12-16,

doi: 10.1016/j.desal.2012.06.025

Lee, K. and Choo, K. (2013) Hybridization of TiO2 photocatalysis with coagulation and

flocculation for 1,4-dioxane removal in drinking water treatment, Chemical Engineering

Journal, 231, 227-235

42

Li, F., Yuasa, A., Ebie, K., Azuma, Y., Hagishita, T., Matsui, Y. (2002) Factors affective the

adsorption capacity of dissolved organic matter onto activated carbon: modified isotherm

analysis, Water Research, 36, 4592-4604

Liao, C-H, Kang, S-F, Wu, F-A (2001) Hydroxyl radical scavenging role of chloride and

bicarbonate ions in the H2O2/UV process, Chemosphere, 44, 1193-1200

Liang, R., Hu, A., Hatat-Fraile, M., Zhou, Y.N. (2014) Development of TiO2 nanowires for

membrane filtration applications, in: Nanotechnology for Water Treatment and Purification, Eds.

Hu, A. and Apblett, A., Springer International, Switzerland




Liu, S., Lim, M., Fabris, R., Chow, C., Chiang, K., Drikas, M., and Amal, R. (2008B) Removal

of humic acid using TiO2 photocatalytic process – Fractionation and molecular weight

characterisation studies, Chemosphere, 72, pp. 263-271




Liu, S., Lim, M., and Amal, R. (2014) TiO2-coated natural zeolite: Rapid humic acid adsorption

and effective photocatalytic regeneration, Chemical Engineering Science, 105 pp. 46-52





Loosli, F., Vitorazi, L., Berret, J-F, and Stoll, S. (2015) Towards a better understanding on

agglomeration mechanisms and thermodynamic properties of TiO2 nanoparticles interacting with

natural organic matter, Water Research, 80, pp. 139-148

43

Love, S.A., Maurer-Jones, M.A., Thompson, J.W., Lin, Y-S, Haynes, C.L. (2012) Assessing

nanoparticle toxicity, Annual Review of Analytical Chemistry, 5, 181-205, doi: 10.1146/annurev-

anchem-062011-143134

Lozier, J.C., Jones, G., Bellamy, W. (1997) Integrated membrane treatment in Alaska, Journal of

the American Water Works Association, 89 (10), 50-64

Luttrell, T., Halpegamage, S., Tao, J., Kramer, A., Sutter, E., and Batzill, M. (2014) Why is

anatase a better photocatalyst than rutile? Model studies on epitaxial TiO2 films, Scientific

Reports (www.nature.com), 4, p. 4031

Luck, A. (2007) UV/TiO2 for drinking water treatment: Concurrent degradation of 1,4-dioxane

and removal of iron and manganese, MASc Thesis, University of Toronto

Marhaba, T.F. and Pipada, N.S. (2000) Coagulation: Effectiveness in removing dissolved

organic matter fractions, Environmental Engineering Science, 17 (2), 107-115




McKie, M.J., Taylor-Edmonds, L., Andrews, S.A., Andrews, R.C. (2015) Engineered

biofiltration for the removal of disinfection by-product precursors and genotoxicity, Water

Research, 81, 196-207

Mills, A. and McFarlane, M. (2007) Current and possible future methods of assessing the

activities of photocatalyst films, Catalysis Today, 129, 22-28, doi:10.1016/j.cattod.2007.06.046

Mills, A. (2012), An overview of the methylene blue ISO test for assessing the activities of

photocatalytic films, Applied Catalysis B: Environmental, 128, pp. 144-149

Munla, L., Peldszus, S., and Huck, P.M. (2012) Reversible and irreversible fouling of ceramic

membranes by model solutions, Journal of the American Water Works Association, 104, pp.

E540-E554

44

Murakami, Y., Endo, K., Ohta, I., Nosaka, A., and Nosaka, Y. (2007) Can OH radicals diffuse

from the UV-irradiated photocatalytic surfaces? Laser-induced fluorescence study, Journal of

Physical Chemistry, 111, 11339-11346




Nakata, K. and Fujishima, A. (2012) TiO2 photocatalysis: Design and applications, Journal of

Photochemistry and Photobiology C: Photochemistry Reviews, 13 (3), pp. 169-189

Nosaka, Y. and Nosaka, A.Y. (2013) Identification and roles of the active species generated on

various catalysts, in: Photocatalysis and Water Purification: From Fundamentals to Recent


Ng, M., Kho, E.T., Liu, S., Lim, M., and Amal, R. (2014) Highly adsorptive and regenerative


Journal, 246, pp. 196-203

Ohtani, B., Prieto-Mahaney, O.O., Li, D., and Abe, R. (2010) What is Degussa (Evonik) P25?

Crystalline composition analysis, reconstruction from isolated pure particles and photocatalytic

activity test, Journal of Photochemistry and Photobiology A: Chemistry, 216, pp. 179-182

Ochiai, T. and Fujishima, A. (2013) Design and Optimization of Photocatalytic Water

Purification Reactors, in: Photocatalysis and Water Purification: From Fundamentals to Recent


Ollis, D. (2013) Photocatalytic treatment of water: Irradiance influences, in: Photocatalysis and

Water Purification: From Fundamentals to Recent Applications, Ed. Pichat, P., Wiley-VCH

Verlag GmbH and Co.

Philippe, K.K., Hans, C., MacAdam, J., Jefferson, B., Hart, J., and Parsons, S.A. (2010A)

Photocatalytic oxidation, GAC, and biotreatment combinations: an alternative to the coagulation

of hydrophilic rich waters?, Environmental Technology, 31 (13), pp. 1423-1434

45

Philippe, K.K., Hans, C., MacAdam, J., Jefferson, B., Hart, J., Parsons, S.A. (2010B)



Pifer, A.D. and Fairey, J.L. (2014) Suitability of organic matter surrogates to predict

trihalomethane formation in drinking water sources, Environmental Engineering Science, 31 (3),

pp. 117-126

Purifics (n.d.) Case History: Photo-Cat Chromium (Cr6) Removal to < 1 ppb, Accessed May 5,

2017: http://www.purifics.com/lwdcms/doc-

view.php?module=documents&module_id=468&doc_name=doc

Purifics (n.d.) Chemical Free Case History: 1,4-Dioxane Groundwater Purification for Lockheed

Martin, Accessed May 5, 2017: http://www.purifics.com/lwdcms/doc-

view.php?module=documents&module_id=379&doc_name=doc

Qamar, M., Yoon, C.R., Oh, H.J., Lee, N.H., Park, K., Kim, D.H., Lee, K., Lee, W.J., and Kim,

S.J. 2008. Preparation and photocatalytic activity of nanotubes obtained from titanium dioxide,


Ramsden, C.S., Smith, T.J., Shaw, B.J., and Handy, R.D. (2009) Dietary exposure to titanium

dioxide nanoparticles in rainbow trout (Oncorhyndchus mykiss): No effect on growth but subtle

biochemical disturbances in the brain, Ecotoxicology, 18, 939-951

Richardson, S.D., Plewa, M.J., Wagner, E.D., Schoeny, R., and DeMarini, D.M. (2007)

Occurrence, genotoxicity, and carcinogenicity of regulated and emerging disinfection by-

products in drinking water: A review and roadmap for research, Mutation Research, 636, 178-

242

Robert, D., Keller, V., Keller, N. (2013) Immobilization of a semiconductor photocatalyst on

solid supports: Methods, materials, and applications, in Photocatalysis and Water Purification,

Ed., Pichat, P., Wiley-VCH Verlag GmbH &Co. KGaA

Rook, J.J. (1974) Formation of haloforms during chlorination of natural waters, Journal of Water

Treatment Examination, 23, 234-243

http://www.purifics.com/lwdcms/doc-view.php?module=documents&module_id=468&doc_name=doc




46

Seitz, F., Rosenfeldt, R.R., Müller, M., Lüderwald, S., Schulz, R., Bundschuh, M. (2016)

Quantity and quality of natural organic matter influence the ecotoxicity of titanium dioxide

nanoparticels, Nanotoxicology, 10 (10), 1415-1421, DOI:10.1080/17435390.2016.1222458

Seo, M-H, Yuasa, M., Kida, T., Huh, J-S, Shimanoe, K., Yamazoe, N. (2009) Gas sensing

characteristics and porosity control of nanostructured films composed of TiO2 nanotubes,

Sensors and Actuators B: Chemical, 137, 513-520

Shahbeig, H., Bagheri, N., Ghorbanian, S., Hallajisani, A., and Poorkarimi, S. (2013) A new

adsorption isotherm model of aqueous solutions on granular activated carbon, World Journal of

Modelling and Simulation, 9 (4), pp. 243-254

Shon, H.K., Vigneswaran, S., Ngo, H.H., and Kim, J-H (2005) Chemical coupling of

photocatalysis with flocculation and adsorption in the removal of natural organic matter, Water

Research, 39 (12), pp. 2549-2558

Sohn, J., Amy, G., and Yoon, Y. (2007) Process-train profiles of NOM through a drinking water

treatment plant, Journal of the American Water Works Association, 99 (6), pp. 145-153

Sterling, M.C., Bonner, J.S., Ernest, A.N.S., Page, C.A., Autenrieth, R.L. (2005) Application of

fractal flocculation and vertical transport model to aquatic sol-sediment systems, Water

Research, 39, 1818-1830

Summers, R. and Roberts, P. (1988) Activated Carbon Adsorption of Humic Substances:

Heterodisperse Mixtures and Desorption, Journal of Colloid and Interface Science, 122 (2), pp.

367-381

Summers, R.S., Hooper, S.M., Shukairy, H.M., Solarik, G., and Owen, D. (1996) Assessing DBP

yield: Uniform formation conditions, Journal of the American Water Works Association, 88 (6),

pp. 80-93

Sun, D.D. and Lee, P.F. (2012) TiO2 microsphere for the removal of humic acid from water:

Complex adsorption mechanisms, Separation and Purification Technology, 91, pp. 30-37

47

Sun, L. and Bolton, J. (1996) Determination of the quantum yield for the photochemical

generation of hydroxyl radicals in TiO2 suspensions, Journal of Physical Chemistry, 100, pp.

4127-4134

Thio, B.J.R., Zhou, D., Keller, A. (2011) Influence of natural organic matter on the aggregation

and deposition of titanium dioxide nanoparticles, Journal of Hazardous Materials, 189, 556-563

Tobiason, J.E., Edzwald, J.K., Levesque, B.R., Kaminski, G.K., Dunn, H.J., Galant, P.B. (2003)

Full-scale assessment of waste filter backwash recycle, Journal of the American Water Works

Association, 95 (7), 80-93

Valencia, S., Marin, J., Velasquez, J., Restropo, G., Frimmel, F.H. (2012) Study of pH effects on

the evolution of properties of brown-water natural organic matter as revealed by size-exclusion

chromatography during photocatalytic degradation, Water Research, 46, 1198-1206

Toor, R. and Mohseni, M. (2007) UV-H2O2 based AOP and its integration with biological

activated carbon treatment for DBP reduction in drinking water, Chemosphere, 66, pp. 2087-

2095

Uyguner, C., Bekbolet, M., and Selcuk, H. (2007) A comparative approach to the application of

a physico-chemical and advanced oxidation combined system to natural water samples,

Separation Science and Technology, 42 (7), pp. 1405-1419

Wang, M-Q, Yan, J., Cui, H-P, Du, S-G (2013) Low temperature preparation and

characterization of TiO2 nanoparticles coated glass beads by heterogeneous nucleation method,

Materials Characterization, 76, 39-47

Wassink, J.D., Andrews, R.C., Peiris, R.H., and Legge, R.L. (2011) Evaluation of fluorescence

excitation-emission and LC-OCD as methods of detecting removal of NOM and DBP precursors

by enhanced coagulation, Water Science and Technology: Water Supply, 11 (5), p. 621

Watanabe, T., Nakajima, A., Wang, R., Minabe, M., Koizumi, S., Fujishima, A., Hashimoto, K.

(1999) Photocatalytic activity and photoinduced hydrophilicity of titanium dioxide coated glass,

Thin Solid Films, 351, 260-263

48

White, M.C., Thompson, J.D., Harrington, G.W., Singer, P.C. (1997) Evaluating criteria for

enhanced coagulation compliance, Journal of the American Water Works Association, 89 (5), 64-

77

Wisznioski et al. (2002) Photocatalytic decomposition of humic acids on TiO2 part 1: Discussion

of adsorption and mechanism, Journal of Photochemistry and Photobiology A: Chemistry, 152,

pp. 267-273

Wong, C.L., Tan, Y.N., and Mohamed, A.R. (2011) A review on the formation of titania

nanotube photocatalysts by hydrothermal treatment, Journal of Environmental Management, 92,

1669-1680

World Health Organization (1999) Hazard prevention and control in the work environment,

WHO/SDE/OEH/99.14, Geneva

Wray, H.E., Andrews, R.C., and Bérubé, P.R. (2013) Surface shear stress and membrane fouling

when considering natural water matrices, Desalination, 330, pp. 22-27

Wu, J., Liu, W., Xue, C., Zhou, S., Lan, F., Bi, L., Xu, H., Yang, X., and Zeng, F-D (2009)

Toxicity and penetration of TiO2 nanoparticles in hairless mice and porcine skin after subchronic

dermal exposure, Toxicology Letters, 191, 1-8, doi:10.1016/j.toxlet.2009.05.020

Xing, Z., Zhou, W., Du, F., Qu, Y., Tian, G., Pan, K., Tian, C., and Fu, H. (2014) A floating

macro/mesoporous crystalline anatase TiO2 ceramic with enhanced photocatalytic performance

for recalcitrant wastewater degradation, Dalton Transactions, 43, p. 790

Yang, Y. and Westerhoff, P. (2014) Presence in and release of nanomaterials from consumer

products in: Nanomaterial, Advances in Experimental Medicine and Biology, 811, eds. Capco,

D.G. and Chen, Y., Springer Science + Business Media, Dordrecht, DOI 10.1007/978-94-017-

8739-0_1



Zeng, H.C. (2011) Preparation and integration of nanostructured titanium dioxide, Current

Opinion in Chemical Engineering, 1, pp. 11-17

49

Zhang, H., Du, G., Lu, W., Cheng, L., Zhu, X., and Jiao, Z. (2012) Porous TiO2 hollow

nanospheres: Synthesis, characterization, and enhanced photocatalytic properties,

CrystEngComm, 14, pp. 3793-3801

Zhang, Y., Chen, Y., Westerhoff, P., and Crittenden, J. (2009) Impact of natural organic matter

and divalent cations on the stability of aqueous nanoparticles, Water Research, 43, pp. 4249-

4257

Zhang, X., Pan, J.H., Du, A.J., Fu, W., Sun, D.D., and Leckie, J.O. (2009). Combination of one-

dimensional TiO2 nanowire photocatalytic oxidation with microfiltration for water treatment,


Zheng, Z., Liu, H., Ye, J., Zhao, J., Waclawik, E.R., and Zhu, H. (2010) Structure and

contribution to photocatalytic activity of the interfaces in nanofibers with mixed anatase and

TiO2(B) phases, Journal of Molecular Catalysis A: Chemical, 316, 75-82

Zheng, D., Andrews, R.C., Andrews, S.A., Taylor-Edmonds, L. (2015) Effects of coagulation on

the removal of natural organic matter, genotoxicity, and precursors to halogenated furanones,


Zhou, D., Ji, Z., Jiang, X., Dunphy, D.R., Brinker, J., Keller, A.A. (2013) Influence of material

properties on TiO2 nanoparticle agglomeration, PLOS One, 8, 11, e81239, doi:

10.1371/journal.pone.0081239

50

Materials and Methods

Synthesis and Characterization of Linear Engineered Nanomaterials

3.1.1 Alkaline Hydrothermal Synthesis Procedure

The linear engineered nanomaterials (LENs) were synthesized according to the simple alkaline

hydrothermal method initially introduced by Kasuga et al. (1999) and subsequently modified by

Yuan and Su (2004), Qamar et al. (2008), Zheng et al. (2010), and many other researchers (see

Appendix A). The following equipment was used during the synthesis process:

Mass balance (OHAUS Analytical Plus)

Sonicator (Fritsch Ultrasonic Cleaner Laborette 17)

Acid digester (Parr Instrument Company)

Drying oven (VWR Model 1305 U)

Muffle furnace (Thermolyne Sybron Furnatrol I)

50 mL centrifuge tubes (8)

Centrifuge (Sorvall RC 5C Plus)

Funnel

Erlenmyer flask

Filtering paper (Whatman 1, 185 mm)

Mortar and pestle

A schematic depicting the basic steps of this process is provided in Figure 2.3 in Chapter 2 of

this document. Briefly, 2 g of Aeroxide P25 TiO2 nanoparticles (Evonik Degussa) was measured

out using a mass balance located in a fume hood, added to 60 mL of 10 M NaOH, and stirred

vigorously with a glass stir stick to fully disperse the nanoparticles throughout the NaOH

solution. The mixture was placed in a Teflon-lined container and secured inside an acid digester.

The digester was placed in the muffle furnace set to the desired hydrothermal temperature

setpoint. After 24 hours, the muffle furnace was turned off and allowed to cool to room

temperature. The acid digester was then removed from the furnace and the leftover NaOH was

decanted into a waste container. The remaining material was transferred to a series of eight 50

51

mL centrifuge tubes, which were then topped up with MilliQ water. The tubes were shaken

vigorously and sonicated for two minutes to encourage greater dispersion and contact between

the water and the TiO2 surface. The sonicated centrifuge tubes were placed in the centrifuge and

centrifuged for 30 minutes at 3,500 rpm. After centrifugation, the tubes were removed, the water

was decanted, and then the tubes were refilled with MilliQ water, resonicated, and recentrifuged.

This process was repeated a total of four times. After the final centrifugation step, the material

was immersed in 400 mL of 0.1 N HCl, mixed, and placed in the sonicator for 1 hour. At this

point, the acidic suspension was redistributed into the centrifuge tubes and the

sonication/centrifuge/decant process was repeated twice as described previously. A glass funnel

was lined with a paper filter and placed atop an Erlenmeyer flask to create a simple filtration

apparatus. The final suspension was carefully poured into the funnel and rinsed with an

additional 1 L of water. The filter and filtered material were dried overnight at approximately

70oC in a drying oven. The dried material was then crushed using a mortar and pestle and

calcined at the desired calcination temperature for 4 hours.

As described in Chapter 4, many iterations of the LENs were developed over the course of this

project as impacts of the different steps of the method became better understood. In all three

“generations” of materials were developed as shown in Table 3.1. Precursor materials and

temperature setpoints were originally chosen based on the findings of Yuan and Su (2004) and

later adjusted to encourage the formation of larger, more reactive materials.

An additional rinsing step was added to the basic synthesis procedure for the third generation

LENs to improve the uniformity and settleability of the final products. Essentially, the calcined

materials were immersed in MilliQ water to form a 5 g/L solution, sonicated for five minutes,

and allowed to settle for 24 hours. At that point, all but approximately 100 mL of the water was

removed from the container and the remaining material was resuspended and allowed to settle

for an additional four hours. All but 100 mL of the water was removed from the container and

the materials were removed from the remaining water via filtration with a 0.45 m PES filter

attached to a standard laboratory filtration apparatus and dried overnight at approximately 70oC

in the drying oven.

52

Table 3.1 Precursor materials and temperature setpoints employed during the alkaline

hydrothermal synthesis of LENs in this project

Material Precursors Hydrothermal

Temperature (TH)

Calcination Temperature

(TC)

First Generation

Nanobelts P25 / NaOH 190oC 700oC

Nanowires P25 / KOH 190oC 550oC

Nanotubes P25 / NaOH 130oC 550oC

Second Generation

NB 130/550 P25 / NaOH 130oC 550oC

NB 130/700 P25 / NaOH 130oC 700oC

NB 240/550 P25 / NaOH 240oC 550oC

NB 240/700 P25 / NaOH 240oC 700oC

Third Generation

NB 550 P25 / NaOH 240oC 550oC

NB 700 P25 / NaOH 240oC 700oC

3.1.2 Characterization of LENs

The LENs synthesized in this study were characterized in terms of size and surface

characteristics (SEM and HRTEM), crystal phase structure (XRD or HRTEM/SAED), specific

surface area (BET), photocatalytic activity (methylene blue degradation), and isoelectric point.

The equipment and conditions used for these tests are summarized in the subsections that follow.

3.1.2.1 Scanning Electron Microscopy

A JEOL 6610LV Scanning Electron Microscope operated by the Department of Earth Sciences

at the University of Toronto was used to observe the shapes and sizes of the first generation

LENs.

53

3.1.2.2 X-Ray Diffraction

A Philips X-Ray Diffraction (XRD) system was used to identify the crystal phase structures

present in the first generation LENs. The XRD instrument is owned by the Department of Earth

Sciences at the University of Toronto.

3.1.2.3 HRTEM / SAED

The crystal phase structures present in the second and third generation LENs were identified

using high resolution transmission electron microscopy (HRTEM) and selected area electron

diffraction (SAED) The measurements were conducted using a JEOL 2010F TEM/STEM at the

Canadian Centre for Electron Microscopy (Hamilton, Ontario, Canada). TEM samples were

prepared by drop casting the dispersions onto holey carbon grids. The images were processed

using Gatan Microscopy Suite: Digial MicrographTM and SAED and FFT images were indexed

using CrysTBox – diffractGUI according to Klinger and Jäger (2015).

3.1.2.4 Surface Area Determination

The specific surface area of each of the second and third generation LENs were determined using

the Brunaeur-Emmett-Teller (BET) method for surface area analysis. N2 adsorption isotherms

were measured with a Quantachrome AUTOSORB-1. The samples were outgassed at 200oC

under vacuum for 12 h before the measurement. Surface area was determined by BET method in

a relative pressure range of 0.05 to 0.25.

3.1.2.5 Photocatalytic Activity

The photocatalytic activity of the LENs was assessed using a modified version of ISO method

10678:2010, Determination of photocatalytic activity of surfaces in an aqueous medium by

degradation of methylene blue, as described by Mills (2012). The ISO method is specific to

immobilized TiO2 films, and as such it was necessary to modify the test such that the

photocatalytic activity of the materials could be compared to one another in suspension. Tests

were conducted with a starting methylene blue concentration of 10 mg/L, a TiO2 dose of 0.1 g/L

(100 mg/L), and 30 minutes of irradiation with UVA light. Under these conditions, all of the

second and third generation LENs and P25 nanoparticles achieved between 30% and 90%

54

decolourization of the methylene blue solution, indicating that these experimental conditions

provided measurable changes in analyte concentration and so were appropriate for the evaluation

and comparison of these materials to one another.

3.1.2.6 Isoelectric Point

The IEP of the LENs was determined by measuring their zeta potentials at pH values ranging

from 3 to 9. Zeta potential was measured using a Horiba Zeta Analyzer and all samples were

prepared by additing 0.1 g/L of TiO2 in MilliQ water buffered with 10 mM NaCl before being

adjusted to various pH values using 0.1 M NaOH or HCl. Two aliquots were analyzed from each

sample and the machine measured each aliquot four times.

Water Matrices

Three surface water matrices were used over the course of this project. All water samples were

gathered from the inlet of the WTPs ahead of chlorination and quickly shipped to the DWRG.

Upon arrival, the water samples were stored in a fridge. The DOC, UV254, SUVA, pH, and

alkalinity of the raw water samples were measured periodically throughout the project. The

results of these measurements are summarized in Table 3.2. Preliminary experiments also made

use of a synthetic river water matrix made according to a recipe provided by Linden et al. (2004)

that included Suwannee River NOM isolate obtained from the International Humic Substances

Society (see Chapter 4 for details).

Table 3.2 Characteristics of four water sources (in lab measurements, variable n, error

values represent standard deviation from the mean)

Parameter Units Otonabee River Ottawa River Lake Ontario

DOC mg/L 3.8 – 5.2 4.8 - 6.8 1.6 – 2.0

UV254 1/cm 0.09 – 0.15 0.16 – 0.24 0.02 – 0.03

SUVA L/mg.m 2.0 – 3.3 2.9 – 4.0 0.8 – 1.5

pH 7.8 – 8.2 6.9 – 7.5 7.8 – 8.0

Alkalinity mg as CaCO3/L 83 – 86 26 – 28 89 – 92

55

The water quality data provided in Table 3.2 indicate that the Ottawa River (OTW) water

samples contained more NOM and more aromatic NOM than those obtained from the other two

water matrices. It was also much lower in alkalinity and had a slightly lower pH than the other

water sources. The Otonabee River (OTB) water contained more NOM than the two lake water

samples but was similar to them in terms of alkalinity and pH.

Early in this project, liquid chromatography with organic carbon detection (LC-OCD) was used

to elucidate the effects of adsorption and photocatalytic degradation on different types of NOM

present in the raw water. Additional LC-OCD data has been gathered from published papers,

theses produced by DWRG alumni, and unpublished datasets from this project and others in

order to compare the four water matrices used in this study in terms of LC-OCD fractions. This

data is summarized in Table 3.3 and confirms that the Ottawa River contained a higher

proportion of humic, aromatic substances than did the other water matrices used at different

points in this project

Table 3.3 Percentages of different LC-OCD fractions present in raw water matrices

used in this study (n = 2)

Fraction Otonabee River Ottawa River Lake Ontario

Biopolymers 6% 3 - 9% 12% - 16%

Humic Substances 56% - 58% 62% - 73% 47%

Building Blocks 18% - 20% 12% - 16% 18% - 24%

Low Molecular Weight

Acids

5% 4% - 9% 5% - 12%

Low Molecular Weight

Neutrals

13% 3% - 10% 6% - 13%

Source(s) Gora and Andrews

(2017)

Zheng (2015);

Unpublished data1

Diemert (2012),

Nemani et al. (2016)

1Taylor-Edmonds, L., personal communication, April 25, 2017

The Ontario Ministry of Environment and Climate Change monitors the quality of the influent

and effluent water at all municipal drinking water treatment plants across the province as part of

the Drinking Water Surveillance Program (DWSP). The frequency of sampling varies from plant

to plant, but all systems are sampled at least once a year. The results of the DWSP are publicly

56

available on the ministry’s website. Raw/inlet water quality data from 2010 to 2012 (most recent

data available as of May 2017) was obtained for the three water sources used in this study and is

summarized in Table 3.4. The DOC, pH, and alkalinity data obtained from the DWSP was a

close match to that measured in the DWRG laboratory throughout this project. The two river

water samples contained more NOM (DOC) than the lake water sample while the lake water

sample had higher levels of all parameters associated with ionic content including alkalinity

(carbonate/bicarbonate), hardness (divalent ions), and conductivity and total dissolved solids,

two semi-quantitative measures of ionic strength. It also had higher levels of specific ions such

as sulphate. The Ottawa River water matrix had the lowest pH, alkalinity, hardness, and calcium

content of the three water sources but also had higher levels of turbidity and metals such as

copper, iron, and manganese.

Table 3.4 Additional water quality data for three natural water matrices used in this

study (DWSP 2010-2012)

Parameter Units Otonabee

River Ottawa River Lake Ontario Lake Simcoe

pH 7.9 - 8.5 7.4 – 7.9 8.1 – 8.3 7.9 – 8.5

DOC mg/L 4.7 – 6.0 6.3 – 7.4 1.8 – 2.1 3.8 – 4.5

Turbidity NTU 0.4 – 1.1 2.2 – 4.9 0.2 - 0.4 0.2 – 1.5

Alkalinity mg as CaCO3/L 77 – 100 20 – 41 90 – 94 110 – 120

Hardness mg as CaCO3/L 88 – 103 23 – 38 115 – 119 137 – 140

Chloride mg/L 10 – 13 2 – 5 23 – 26 42 – 45

Calcium mg/L 30 – 35 6 – 10 32 – 33 42 – 43

Magnesium mg/L 3 2 – 3 9 8

Sodium mg/L 6 – 7 2 – 5 13 – 14 24 – 25

Phosphate mg/L 0.001 – 0.005 0.001 – 0.012 0.001 – 0.005 0.001 – 0.008

Sulphate mg/L 5 – 7 5 – 7 25 – 28 18 – 19

Conductivity S/cm 188 – 244 61 -112 300 – 315 378 – 408

Dissolved Solids mg/L n.d. 33 – 52 161 -168 211 – 212

Copper ug/L 0 – 7 107 – 260 1 – 28 2 – 10

Iron ug/L 0 – 30 140 – 290 0 – 20 n.d.

Manganese ug/L 0 – 20 8 – 24 0 – 4 1 – 4

57

Experimental Apparatus

3.3.1 Light Sources

Preliminary experiments were conducted under simulated solar light (Photo Emission Tech,

SS150AA) and a high intensity UVA lamp (Blak-Ray, B100-AP). Detailed descriptions of these

light sources are provided in Chapter 4 (Section 4.1.2). The majority of the experiments

presented in this thesis were conducted using a custom-made UVA LED reactor constructed in-

house according to instructions provided by Robert Liang at the University of Waterloo. The

system consisted of four UVA lamps secured to a stand above a multiple location stir plate that

was able to accommodate four beakers at once. The UVA LED bulbs (LZ1 UV 365 nm Gen2

Emitter, LED Engin Inc.) had a maximum irradiance at 365 nm. The average irradiance across

the surface of the sample was calculated using a spreadsheet developed by Bolton and Linden

(2003) and was determined to be 4.9 mW/cm2. The irradiance of each lamp was confirmed

before each test using a radiometer (International Light, ILT1400) equipped with a sensor

optimized to measure light at 365 nm (International Light, XRL140B).

3.3.2 Additional Apparatus

All adsorption experiments were conducted using an end over end box mixer constructed in-

house at the Department of Civil and Mineral Engineering at the University of Toronto. Amber

bottles were used as batch reactors used for the adsorption experiments to minimize the

likelihood of the sample being exposed to light. Additional information about other apparatus

used for sample preparation and analysis is provided in later sections of this chapter and in the

relevant materials and methods sections in other chapters.

Sample Preparation and Experimental Design

3.4.1 Photocatalysis Tests

All of the dye and NOM degradation experiments included triplicate or quadruplicate samples

prepared by dosing 50 mL samples of raw OTB or OTW with P25 nanoparticles or one of the

LENs and exposing them to 0, 5, 15, 30, 45, and 60 minutes of UVA irradiation. All samples

58

were allowed to mix in the dark for 1 minute before irradiation. Mixing was provided by

magnetic stir bars and UVA irradiation was provided by the UVA LED reactor described in

Section 3.3.1. Dye degradation experiments were conducted with TiO2 doses of 0.1 g/L or 0.25

g/L while the NOM degradation experiments were conducted with a TiO2 dose of 0.25 g/L to

ensure sufficient NOM degradation for subsequent modeling. Samples were analyzed for DOC

and UV light absorbance at 254 nm. The results were evaluated against a pseudo-first-order

model for photocatalytic degradation and also normalized to nanomaterial surface area. Specific

details for each experiment are provided as required in later chapters of this document.

3.4.2 Adsorption Tests

The time required to reach adsorption equilibrium and the effect of increasing TiO2 dose on the

removal of Acid Orange 24 dye, DOC, UV254, THM precursors, and HAA precursors by P25

and the two optimized (a.k.a. third generation) LENs were investigated in a series of adsorption

experiments. In all cases, mixing was provided by an end over end box mixer.

3.4.3 Nanomaterial Regeneration

Regeneration experiments were conducted to determine whether the third generation could be

reused multiple times to remove AO24 and DBP precursor surrogates. Duplicate vials containing

25 mL of 10 mg/L dye solution or raw surface water were dosed with 0.5 g/L of NB 550 or NB

700 and mixed end over end in a box mixer. After 30 minutes the TiO2 was removed from the

samples via filtration and resuspended in 25 mL of millQ purified water. The new suspensions

were mixed with a stir plate and stir bar and regenerated via exposure to UVA light (365 nm)

with an average irradiance of 4.9 mW/cm2 for one hour. The regenerated TiO2 was removed

from the purified water via filtration and then resuspended in a fresh water or dye sample and

mixed for 30 minutes in the box mixer. This process was repeated five times for each LEN.

3.4.4 Settling Tests

3.4.4.1 Settling Tests for Low TiO2 Doses

Low dose settling tests were inspired by the work of Liu et al. (2013) and Erhayem and Sohn

(2014). Both groups studied the effects of water matrix conditions on TiO2 sedimentation. Their

59

tests were conducted using a UV-Vis spectrophotometer. The DWRG’s UV-Vis

spectrophotometer was unable to accurately measure suspensions of some of the LENs at

concentrations above 0.1 g/L, however, so the settling tests for this project were took place in a

Hach 2100 N turbidimeter operating in NTU mode with ratio on, which allowed the turbidimeter

to measure in the 0 to 4000 NTU range and accurately characterize LEN suspensions containing

up to 0.3 g/L of material.

Aliquots of a 10 g/L TiO2 stock solution were dispensed into an appropriate amount of MilliQ

water to create triplicate samples for the settling tests (100 mg/L, 200 mg/L, or 250 mg/L) and

duplicate calibration standards (10 mg/L, 50 mg/L, 100 mg/L, 200 mg/L, and 300 mg/L). Each

calibration standard was sonicated for 5 minutes and then analyzed for turbidity. The turbidity

results were graphed against concentration to develop a calibration curve for each material (see

Appendix C). The relationship between turbidity and concentration was linear within the range

studied for all five materials. For the settling tests, samples were dispensed into the turbidimeter

cuvette, sonicated for 5 minutes, and placed in the turbidimeter for a total of 2 or 3 hours. The

turbidity at the midpoint of the cuvette was recorded at the beginning of the test and at ten

minute intervals thereafter.

3.4.4.2 Settling Tests for High TiO2 Doses

TiO2 doses above 0.3 g/L (300 mg/L) resulted in turbidity and UV-Vis signals well above the

operating ranges of the instruments available in the DWRG laboratory, therefore it was necessary

to adopt an alternative methodology in order to accurately assess the settling behaviour of the 1

g/L TiO2 suspensions used in the final adsorption experiments (Chapter 8). 600 mL of water was

dosed with 1 g/L of TiO2, sonicated for five minutes, and then mixed in the box mixer. After one

hour, the water was distributed into two sets of five tall 60 mL vials and the remaining volume

was reserved as a control. 30 mL aliquots were removed after 5, 10, 15, 30, and 60 minutes and

diluted to one tenth their original concentration. The diluted samples were analyzed using a

HACH turbidimeter.

3.4.5 Filtration Tests

The Time to Filter test in Standard Methods (2710 H) was used to evaluate the filterability of

P25 and the third generation LENs. Essentially, the time to filter a set volume of sample

60

containing the nanomaterials was normalized to the amount of time required to filter pure MilliQ

water through the filter under identical filtration conditions to yield a unitless value, the filtration

index. A high filtration index implies that the suspension is resistant to filtration while a value

close to 1 indicates that the suspension has the same resistance to filtration as pure water.

Sample Analysis

3.5.1 Dyes

The concentration of methylene blue and Acid Orange 24 in the raw and treated water was

determined by measuring the absorbance of the solution at 665 nm for methylene blue and 430

nm for AO24 using an Agilent 8453 UV-Vis spectrophotometer. Sample calibration curves for

the two dyes are provided in Appendix C.

3.5.2 Disinfection Byproduct Surrogates

Raw and treated water samples were filtered through a 0.45 m polyethersulfone (PES)

laboratory filter before analysis. Natural organic matter was quantified as dissolved organic

carbon (DOC) or based on UV absorbance at 254 nm (UV254). DOC was measured on an O/I

Analytical Aurora 1030 TOC analyzer and UV254 was measured using an Agilent 8453 UV-Vis

spectrophotometer. Size exclusion liquid chromatography with organic carbon detection (LC-

OCD) as described by Huber et al. (2011) was conducted at the University of Waterloo on

selection of samples from this project. The results of the analyses were processed using

proprietary software (ChromCalc, DOC-LABOR, Karlsruhe, Germany). Fluorescence

measurements were made on a Varian (now Agilent) Cary Eclipse fluorescence

spectrophotometer.

3.5.3 Disinfection Byproduct Formation and Analysis

The uniform formation conditions (UFC) method as described by Summers et al. (1996) was

used to assess the chlorine demand and DBPfp of the raw water and the water that had been

treated with TiO2. Samples were buffered with a borate solution and adjusted to pH 8 with 1 N

HCl or 1 N NaOH, dosed with chlorine, and stored in the dark at 20oC for 24 hours, after which

61

the free chlorine residual was measured on a Hach DR 2700 according to the DPD Colorimetric

Method (Standard Method 4500-CI G). (APHA 2005). The trihalomethanes and haloacetic acids

formed during the UFC tests were extracted according to Standard Method 6232 B and Standard

Method 6251 B (APHA, 2005) and analyzed on a Agilent 7890B GC-ECD. Standard Method

6232 B and Standard Method 6251 B (APHA, 2005).

3.5.4 Other Analyses

pH was measured throughout this project using a Thermo Scientific Orion Star A111 equipped

with an Orion 9157BNMB pH/ATC probe. Alkalinity was determined using Standard Method

2320 (APHA, 2005).

Quality Control

Each batch of second or third generation LENs was tested before use to ensure that it was

consistent with previous batches using the methylene blue degradation test described in Section

3.1.2.5. Figures D.1 and D.2 in Appendix D show the percent decolourization of methylene blue

dye achieved by each batch of second and third generation LENs, respectively. Quality control

charts for TOC, five THMs, and nine HAAs are provided in Appendix D. TOC standards were

prepared at 3 mg/L and analyzed after each ten samples along with blank samples at the

beginning and end of the run. DBP blanks and 20 g/L check standards were analyzed after

every ten samples.

All water samples were gathered from the inlet of the WTPs ahead of chlorination and quickly

shipped to the DWRG. Upon arrival, the water samples were stored in a fridge. Raw and treated

samples (DOC, UV254, DBPs, etc.) were analyzed as quickly as possible and/or stored in a

fridge between preparation and analysis.

62

References

American Public Health Association (2005) Standard Methods for the Examination of Water and


Bolton, J.R. and Linden, K.G. (2003) Standardization of methods for fluence (UV dose)

determination in bench-scale UV experiments, Journal of Environmental Engineering, 129, 209-

215

Ontario Ministry of Environment and Climate Change (2013) Drinking Water Surveillance

Program, Accessed May 24, 2016: https://www.ontario.ca/data/drinking-water-surveillance-

program

Diemert, S. and Andrews, R.C. (2012) The impact of alum coagulation on pharmaceutically

active compounds, endocrine disrupting compounds, and natural organic matter, Water Science

and Technology: Water Supply, 13 (5), 1348-1357









Klinger, M. and Jäger, A. (2015) Crystallographic Tool Box (CrysTBox): automated tools for

transmission electron microscopists and crystallographers. Journal of Applied Crystallography

48 doi:10.1107/S1600576715017252.

Linden, K.G., Sharpless, C.M., Andrews, S.A., Atasi, K.Z., Korategere, V., Stefan, M., Mel

Suffet, I.H. (2004) Innovative UV Technologies to Oxidize Organic and Organoleptic

Chemicals. Awwa Research Foundation, Denver, CO, USA.

https://www.ontario.ca/data/drinking-water-surveillance-program

https://www.ontario.ca/data/drinking-water-surveillance-program

63







Nemani, V., Taylor-Edmonds, L., Peleato, N.M., Andrews, R.C. (2016) Impact of operational

parameters on biofiltration performance: Organic carbon removal and effluent turbidity, Water

Science and Technology: Water Supply, 16 (6), 1683-1692, DOI: 10.2166/ws.2016.093




Summers, R.S., Hooper, S.M., Shukairy, H.M., Solarik, G., and Owen, D. (1996) Assessing DBP


pp. 80-93






Zheng, D. (2015) Effects of coagulation on the removal of natural organic matter, genotoxicity,

and precursors to halogenated furanones, MASc Thesis, University of Toronto

64

Preliminary Experimental Findings and Concept Development

The project that was initially proposed in early 2013 was part of an NSERC Strategic Project

Grant built around a photocatalytic engineered TiO2 membrane. I was to investigate the

disinfection and fouling reduction abilities of this membrane, which was to be prepared by

researchers at the University of Waterloo. Unfortunately, their work was delayed numerous times

and they were unable to present us with useable membranes until late 2014. In the interim, I

began to develop a new concept and a new project based on the linear engineered nanomaterials

(LENs) that served as precursor materials for the membranes: LENs for the removal of natural

organic matter and disinfection byproduct precursors. The initial results of the experiments

conducted with the three original LENs (first generation LENs) were so promising that a

decision was made to abandon the membrane-based project in favour of the new project.

Preliminary experiments were conducted throughout 2014 to determine the effectiveness of TiO2

adsorption and photocatalysis for NOM removal and to refine the LEN synthesis method adapted

from Kasuga et al. (1999) and Yuan and Su (2004). The work presented in this chapter was, for

the most part, exploratory and not deemed fit for publication for a number of reasons, including

lack of novelty and insufficient replication. It did, however, inform the design and execution of

later experiments, and as such is still relevant to the project. Some of the avenues that were

explored early in the project may also be worth revisiting and eventually lead to interesting

research projects in their own right. Many of the results presented here were presented at various

conferences and symposia in 2014 and 2015.

65

Methods and Materials

4.1.1 Experimental Design

Experimental parameters were chosen after a review of the methods and findings of other

researchers. The light sources, TiO2 doses, irradiation times, and water matrices employed in six

studies focused on the removal of NOM and DBP precursors from surface water or synthetic

water matrices containing commercially available NOM isolates such as Suwannee River NOM

(SRNOM) via TiO2 photocatalysis are presented in Table 4.1.

Table 4.1 Experimental conditions used in previous studies

Study Light Source TiO2 Dose(s) Irradiation Time(s) Water Matrices

Liu et al. (2008) UVA 0.1 g/L 0 min, 30 min, 60 min,

90 min, 120 min, 150

min, 180 min, 210 min,

240 min

Two Australian

surface water

sources

Liu et al. (2010) UVA 0.1 g/L 0 min, 30 min, 60 min,

90 min, 120 min, 150

min

Two Australian

surface water

sources

Gerrity et al. (2009) UV

(wavelengths

not specified)

0.1 g/L, 0.4

g/L, 1 g/L

Researchers measured

energy use by pilot scale

UV/TiO2 system instead

of time

Two Arizona

surface water

sources

Philippe et al.

(2010)

MP UV 1 g/L 0 min, 1 min, 5 min, 10

min

NOM surrogate

solutions

Huang et al. (2008) LP UV 0.1 g/L, 0.3

g/L, 0.5 g/L,

1.0 g/L

0 min, 20 min, 40 min,

60 min, 80 min, 100 min,

120 min

SRNOM

SRNOM + ions

Valencia et al.

(2013)

Simulated

Solar

0.6 g/L 0 min, 30 min, 60 min,

90 min, 120 min, 150

min, 180 min, 210 min,

240 min

Humic acid and

fulvic acid

isolates

66

Based on this literature review, the following experimental conditions were adopted for the

preliminary experiments:

Light sources: Simulated solar and high intensity UVA

TiO2 doses: 0.005 g/L to 1 g/L

Irradiation times: 0 to 60 minutes in increments of 5 or 15 minutes

Water sources: Synthetic river water (SRW, Linden et al., 2014), Lake Ontario water

(LO), and Otonabee River water (OTB)

In all cases, samples were independent of one another, that is, individual samples were prepared

and irradiated for different amounts of time instead of removing aliquots from one large sample

over the course of the experiment. The preparation of independent samples resulted in a more

statistically rigorous dataset and ensured that the height of water in all of the samples was equal

so that they were all exposed to the same dose of light.

4.1.2 Materials

All of the preliminary experiments were conducted in duplicate or triplicate in small batch

reactors mixed using stir bars and stir plates. The mixing rate was controlled to ensure that the

samples were fully mixed without any whirlpool effects, thus ensuring a flat water surface. The

batch reactors had an internal diameter of 6.5 cm and a height of 5 cm. All of the experiments

described in this chapter used 50 mL of sample, thus the height of water in the batch reactors was

1.5 cm. The external walls of the batch reactors were covered to prevent any loss of light.

P25 TiO2 nanoparticles from Evonik Degussa were used for all preliminary method development

experiments, as a precursor material for the LENs, and as a standard point of comparison in the

LENs experiments. Crystalline methylene blue dye and Acid Orange 24 (AO24) were purchased

from Sigma Aldrich. 1,000 mg/L stock solutions of each dye were prepared and stored until

required, at which point they were used to make 10 mg/L working solutions for the experiments.

Three water matrices were used for the preliminary experiments: SRW, LO, and OTB. The

characteristics of the SRW matrix are provided in Table 4.2. All parameters were measured in

67

the laboratory unless indicated. Additional water quality data for LO water and OTB water is

provided in Chapter 3.

Table 4.2 Summary of synthetic water quality

Parameter Units Synthetic

River Water

pH 8.21

Alkalinity mg as CaCO3/L 1171

Hardness mg as CaCO3/L 1141

DOC mg/L 2.9

UV254 0.043

SUVA 1.48

Chloride mg/L 40.02

Calcium mg/L 29.12

Magnesium mg/L 10.02

Sodium mg/L 36.22

Nitrate/Nitrite mg/L 0.672

Sulphate mg/L 55.02

1From: Sokolowski, 2014

2Calculated based on synthetic water recipe from Linden et al., 2014

4.1.3 Light Sources

The majority of the preliminary experiments presented in this chapter were conducted using

simulated solar light. High intensity UVA (“black”) light was used for a small subset of

experiments. The specifications of the two light sources are provided in Table 4.3.

The irradiance of the high intensity lamp was measured using a radiometer from International

Light (ILT1400) equipped with a UVA sensor with maximum sensitivity between 360 and 370

nm (XRL140B). The total irradiance of the portion of light in the 300 to 400 nm range provided

by the solar simulator was provided by a representative from the manufacturer (Sokolowski,

2014). The dose of light provided to each sample was calculated by multiplying the irradiance at

the surface of the sample by the time of exposure.

68

Table 4.3 Light sources used for preliminary photocatalysis experiments

Parameters Solar Simulator Blak-Ray High UVA Lamp

Manufacturer Photo Emission Tech UVP

Product ID SS150AAA B100-AP

Range 300 – 1000 nm n/a

Wavelength of Maximum

Intensity

n/a 365 nm

Average Irradiance 108 mW/cm2 overall

8.27 mW/cm2 UVA (300 – 399

nm)

21.7 mW/cm2 at 2 inches

8.9 mW/cm2 at 10 inches

12 mW/cm2 in experiment

Wattage 1,000 W 100 W

Voltage 220 V 115 V

4.1.4 LEN Synthesis

The three LENs used in the preliminary experiments were synthesized in the laboratory using a

hydrothermal method followed by calcination as described in Section 2.3.2 and Section 3.1. The

basic method was modified using information from a paper published by Yuan and Su in 2004 to

make nanotubes (NTs) and nanowires (NWs) in addition to the basic nanobelts (NBs). A

summary of the conditions used to synthesize the first generation of LENs used in this project is

provided in Table 4.4.

Table 4.4 Summary of synthesis parameters for first generation LENs

Parameter P25 NB NW NT

Precursors -- P25/NaOH P25/KOH P25/NaOH

Hydrothermal Temperature -- 190oC 190oC 130oC

Calcination Temperature -- 700oC 550oC 550oC

Notes Standard material

69

Results and Discussion

4.2.1 Effect of Time, TiO2 Dose, Water Type, and Light Source on NOM Adsorption and Degradation

Preliminary experiments were conducted to evaluate the effects of time, TiO2 dose, water type,

and light source on the adsorption and photocatalytic degradation of NOM, measured as DOC

and UV254, by P25 nanoparticles. The results were generally comparable to those obtained by

other researchers.

4.2.1.1 Adsorption Time

Synthetic river water containing SRNOM was dosed with 0.005 g/L, 0.05 g/L, and 0.5 g/L of

P25 nanoparticles and mixed in the dark for times ranging from one to 30 minutes. The amount

of DOC remaining in the water after adsorption by different doses of TiO2 is presented in Figure

4.1.

Figure 4.1 DOC of synthetic water matrix after adsorption by different doses of P25

nanoparticles

0

0.5

1

1.5

2

2.5

3

3.5

0 5 10 15 20 25 30

DO

C (

mg/L

)

Adsorption Time (min)

0.005 g/L 0.05 g/L 0.5 g/L

70

The results indicate that the adsorption of DOC to the nanoparticles occurred quickly, likely in

less than one minute. Some researchers have also observed fast adsorption kinetics for NOM and

TiO2 (Ng et al., 2014) but others have reported small increases in overall adsorption over time

periods ranging from hours to days (Mwaanga et al., 2014; Erhayem and Sohn, 2014). The very

fast adsorption observed in this experiment was challenged by the findings of later experiments

(see Chapter 5 and Chapter 8), however, it was repeated for Lake Ontario water and Otonabee

River water in this set of experiments (data not shown) so later experiments proceeded with a

one minute adsorption period.

4.2.1.2 TiO2 Dose

The dose of TiO2 added to the water affected not only adsorption but also the extent and rate of

photocatalytic degradation. Figure 4.2 and Figure 4.3 show the change in DOC and UV254 as a

function of time in synthetic water dosed with P25 nanoparticles and irradiated with simulated

solar light.

Figure 4.2 Change in DOC content of synthetic water treated different doses of P25

TiO2 nanoparticles and irradiated by simulated solar light

As was observed in the initial adsorption experiment, adsorption increased as a function of TiO2

dose, ranging from approximately 20% at 0.005 g/L of TiO2 to nearly 80% at 0.5 g/L of TiO2.

-100%

-80%

-60%

-40%

-20%

0%

0 10 20 30 40 50 60

Ch

an

ge

in D

OC

Irradiation Time (min)

0.005 g/L 0.05 g/L 0.1 g/L 0.2 g/L 0.5 g/L

71

Note that adsorption of NOM to TiO2 occurs both before irradiation (dark adsorption) and

dynamically during photocatalysis. In this project, the term adsorption almost always refers to

dark adsorption. One of the most interesting findings of this set of experiments was the repeated

observation of an initial increase in overall DOC upon irradiation of the sample. This slight

increase has been observed by others (Huang et al., 2008; Gerrity et al., 2009), who attributed to

the desorption of intermediate compounds following the degradation of the molecules originally

adsorbed to the TiO2 surface. Another possible explanation is the light induced desorption of

adsorbed NOM molecules upon irradiation of the samples. The photoactivation of TiO2 has

profound effects on its surface characteristics, including its hydrophobicity. Watanabe et al.

(1999) demonstrated that both anatase and rutile forms of TiO2 become more hydrophilic upon

irradiation with UVA light. The change in hydrophilicity began immediately upon irradiation but

proceeded gradually over the course of an hour, but even this initial change may have induced

some highly hydrophobic NOM molecules to desorb from the TiO2 surface.

After 60 minutes, overall DOC removal ranged from approximately 10% at 0.005 g/L TiO2, 40%

at TiO2 concentrations ranging from 0.05 g/L to 0.2 g/L, and 60% at 0.5 g/L TiO2. This is

comparable to results obtained by Huang et al. (2008) with SRNOM and by Liu in their 2008

study with an unidentified water source but below what they achieved in their later studies (Liu

et al., 2010a; Liu et al., 2010b). It should be noted that any decrease in DOC observed after

adsorption represents full mineralization of NOM.

A similar pattern was apparent when UV254 was used as the response parameter. The apparent

increase in UV254 may be related to TiO2 passage through the filter, which would result in the

presence of nanoparticulates that might interfere with absorption measurements. This is more

likely to occur in water matrices that promote nanoparticle stability/disaggregation as described

in Table B.1 in Appendix B.

72

Figure 4.3 Change in the UV254 of synthetic river water treated with different doses of

P25 TiO2 nanoparticles irradiated by simulated solar light

UV254 was removed more quickly and more completely than DOC, particularly at higher

concentrations of TiO2. Liu et al. (2010a) and others have shown that P25 nanoparticles have an

affinity for large aromatic NOM molecules and are more likely to degrade them than smaller,

more hydrophilic NOM compounds. Additionally, UV254 is “removed” when the aromatic

structures in the molecules are broken but before full mineralization (which may include many

steps) is complete, so it is not surprising that it decreased more quickly than DOC. As with DOC,

there was a slight increase in UV254 at short irradiation times followed by a gradual decrease.

The initial increase may be related to the release of aromatic intermediates after photocatalytic

oxidation of the original adsorbed molecules or due to the desorption of these adsorbed

molecules due to changes in the hydrophilicity of the TiO2 surface upon irradiation.

When the effects of initial dark adsorption and light induced desorption were excluded from the

analysis, the DOC and UV254 degradation results of the experiments conducted with TiO2 doses

above 0.05 g/L were a good fit to a pseudo-first order degradation model. The model parameters

as well as the inflection point at which degradation became the dominant phenomenon are

summarized in Table 4.5.

-100%

-80%

-60%

-40%

-20%

0%

20%

0 10 20 30 40 50 60

Ch

an

ge

in U

V2

54


0.005 g/L 0.05 g/L 0.1 g/L 0.2 g/L 0.5 g/L

73

Table 4.5 Pseudo-first order reaction rate constants and fits for DOC and UV254

removal from synthetic river water by different doses of TiO2 P25

nanoparticles irradiated by simulated solar light

TiO2 Dose DOC UV254

k (min-1) R2 Inflection (min) n k (min-1) R2 Inflection (min) n

0.005 g/L n/a n/a n/a n/a n/a n/a n/a n/a

0.05 g/L -0.0082 0.95 15 6 -0.0263 0.98 10 8

0.1 g/L -0.0073 0.97 15 6 -0.0319 0.99 0.5 16

0.2 g/L -0.0079 0.97 15 6 -0.0337 0.97 0.5 16

0.5 g/L -0.0107 0.78 10 8 -0.0339 0.97 5 8

The DOC degradation rate constants were comparable to those reported by Huang et al. (2008),

who used a low pressure UV lamp for irradiation, and Valencia et al. (2013) who used simulated

solar light. Increasing the TiO2 dose from 0.05 g/L to 0.2 g/L had no appreciable effect on k but

increasing it further to 0.5 g/L resulted in a higher rate of DOC removal, though in this case the

data was a poorer fit to the pseudo-first order model than at lower TiO2 doses. In contrast, the

UV254 degradation rate constant increased with TiO2 dose until 0.2 g/L, at which point it

became steady. The UV254 rate constants obtained in this study were higher than those achieved

by Valencia et al., who used HA and FA isolates rather than SRNOM for their experiments. It

may be that SRNOM is more amenable to degradation than the HA and FA isolates prepared by

Valencia et al. or, alternatively, that some other property of the synthetic water matrix promoted

more effective degradation of aromatic NOM.

Overall, the results of this experiment suggest that increasing the dose of TiO2 has a strong effect

on adsorption but less of an effect on degradation rate, particularly above 0.2 g/L. These findings

influenced the choice of TiO2 doses used in the subsequent experiments as well as those used in

the experiments detailed in chapters 5 to 8.

4.2.1.3 Water Type

Real water matrices are impacted by numerous natural and anthropogenic inputs and subject to

seasonal variations, and as such, are more complex than the synthetic water matrices that are

74

used in many laboratory experiments. This can result in confusion and disappointment when a

promising technology is less effective in real water than was originally suggested by experiments

conducted with NOM isolates or surrogate compounds. There are, of course, good reasons to use

synthetic water matrices, including greater control over raw water composition and consistency

and the ability to vary the levels of different matrix components to isolate their effects on the

treatment.

In the case of TiO2 photocatalysis, many researchers including those that have been cited in

earlier sections have explored its effects on NOM isolates (e.g. SRNOM) and NOM surrogates

such as amino acids, carbohydrates, and phenolic compounds. Studies by Liu et al. (2008, 2010a,

2010b) and other members of the Amal research group at the University of New South Wales are

some of the few to employ real water matrices, and their findings suggest that water matrix

components can have strong and conflicting effects on NOM adsorption and degradation by

TiO2. Two water matrices, Lake Ontario water and Otonabee River water, were employed in the

preliminary stages of this project to explore the effectiveness of TiO2 adsorption and

photocatalysis on NOM in real water.

Figure 4.4 shows the effects of adsorption and degradation on the DOC content of Lake Ontario

water. As was observed with the synthetic water, higher TiO2 doses removed more DOC from

the water via adsorption than lower TiO2 doses. Irradiation had little effect on overall DOC

content but did remove 52% of the UV254 in the water at a TiO2 dose of 0.05 g/L and 66% at a

TiO2 dose of 0.5 g/L after 60 minutes of irradiation. It seems likely that photocatalytic

degradation of NOM simply occurred more slowly in the Lake Ontario water than it did in the

synthetic water. The two water matrices contained similar levels of bicarbonate (alkalinity) and

chloride, which are known ROS scavengers (Liao et al., 2001), so overall scavenging potential is

unlikely to explain the discrepancy. Instead, it seems that photocatalytic degradation of NOM

occurred more slowly in the Lake Ontario water than in the synthetic water due to the character

of the NOM present in each matrix. The UV254 absorbance of the synthetic water was nearly

twice that of the Lake Ontario water and SRNOM is known to be highly aromatic. Liu et al.

(2008) showed that aromatic NOM is preferentially degraded by TiO2 photocatalysis.

75

Figure 4.4 Removal of DOC from Lake Ontario water by different doses of P25 TiO2


Figure 4.5 Removal of UV254 from Lake Ontario water by different doses of P25 TiO2


-100%

-80%

-60%

-40%

-20%

0%

20%

0 10 20 30 40 50 60

Ch

an

ge

in D

OC


0.005 g/L 0.05 g/L 0.5 g/L

-100%

-80%

-60%

-40%

-20%

0%

20%

0 10 20 30 40 50 60

Ch

an

ge

in D

OC


0.005 g/L 0.05 g/L 0.5 g/L

76

The Lake Ontario DOC data did not fit a pseudo-first order decay model because of the strong

influence of adsorption before irradiation and desorption (likely of intermediate products) upon

irradiation. The UV254 data (Figure 4.5) was a reasonable fit to the pseudo-first order decay

model at higher TiO2 doses (R2 = 0.83-0.86). The reaction rate constant at 0.05 g/L was 0.0051

min-1, which is only slightly lower than that at 0.5 g/L (0.0056 min-1), suggesting that increased

TiO2 dose did not have a strong effect on reaction rate within this dose range and in this water

matrix.

NOM degradation tests were also conducted using water from the Otonabee River, which

supplies the drinking water treatment plant in Peterborough, Ontario. Graphs depicting the

effects of irradiation on DOC and UV254 are provided in Figure 4.6 and Figure 4.7 below.

Figure 4.6 Removal of DOC from Otonabee River water by different doses of P25 TiO2


As was observed in the tests conducted with synthetic water and Lake Ontario water, an increase

in the TiO2 dose resulted in increased adsorption of DOC from the Otonabee River water. The

extent of adsorption observed at each dose was lower than in the synthetic water experiments.

This may be due to the intrinsic properties of the NOM compounds present in each water matrix

or due to interference by ions or other components of the Otonabee River water matrix. For

-100%

-80%

-60%

-40%

-20%

0%

20%

0 10 20 30 40 50 60

DO

C C

han

ge


0.01 g/L 0.05 g/L 0.1 g/L 0.5 g/L

77

example, as presented in Table B.1 in Appendix B, inorganic ions and metals can compete with

NOM for adsorption sites on the TiO2 surface. Some of these compounds were not present in the

synthetic water matrix but were present in the two real water matrices.

The DOC content of the samples exposed to simulated solar light indicates that NOM was

gradually mineralized over time when the TiO2 dose was at and above 0.05 g/L. At a dose of

0.01 g/L, DOC actually increased upon irradiation, suggesting that smaller, more easily detected

organic compounds were being formed during the initial stages of treatment. This supports the

hypothesis that intermediate products, rather than unreacted parent compounds, desorbed from

the TiO2 surface after irradiation. The rate and extent of DOC degradation was once again less

than that observed with the synthetic water, likely due to the presence of interferents and the

characteristics of the NOM in each water source.

Figure 4.7 Removal of UV254 from Otonabee River water by different doses of P25

TiO2 nanoparticles irradiated by simulated solar light

The effect of TiO2 dose was more straightforward when UV254 was used as the response

parameter. As shown in Figure 4.6, at 0.5 g/L, the P25 nanoparticles were able to remove

approximately 40% of the UV254 in the raw water through adsorption alone and nearly 80%

-100%

-80%

-60%

-40%

-20%

0%

20%

0 10 20 30 40 50 60

Ch

an

ge

in U

V254


0.01 g/L 0.05 g/L 0.1 g/L 0.5 g/L

78

after 60 minutes of simulated solar irradiation. Little or no UV254 removal occurred via

adsorption at lower TiO2 doses but removal did occur upon irradiation.

As shown in Table 4.6, the rates of both DOC and UV254 removal increased steadily with

increasing TiO2 dose. The degradation rate constants obtained in the Otonabee River water

experiments were well below those from the synthetic water experiments but compared

favourably with those reported by Valencia et al. (2013) for fulvic and humic acid isolates. The

Otonabee River UV254 degradation rate constants were within the same range as those from the

Lake Ontario experiments.


removal from Otonabee River water by different doses of TiO2 P25


TiO2 Dose DOC UV254

k (min-1) R2 n k (min-1) R2 n

0.01 g/L 0.0006 0.84 10 -0.0025 0.96 10

0.05 g/L -0.0006 0.57 10 -0.0042 0.97 10

0.1 g/L -0.0010 0.80 10 -0.0059 0.99 10

0.5 g/L -0.0020 0.93 8 -0.0072 0.99 8

The findings of these experiments clearly indicate that the removal of NOM by P25

nanoparticles via both adsorption and photocatalysis was strongly impacted by the composition

of the water matrix. Any new TiO2-based treatment system must be able to operate in real water

matrices, and the results presented here suggest that experiments conducted solely with synthetic

water and/or SRNOM would not adequately predict the effectiveness of either adsorption or

photocatalysis in a real drinking water treatment system.

79

4.2.1.4 Light Source

TiO2 can be photoactivated by any light source that provides irradiation at wavelengths at or

below 385 nm, but in practice, the majority of the studies exploring the use of TiO2

photocatalysis for drinking water treatment have relied on high intensity UVA or germicidal UV

lamps to provide the irradiation required to activate the photocatalyst. Throughout the initial

stages of this project, experiments were conducted using both simulated solar light and UVA

light. The original idea was to eventually compare the effectiveness of the two light sources in

terms of their ability to activate pure TiO2 materials as well as modified TiO2 materials prepared

by our partners at the University of Waterloo, which were designed to take advantage of

wavelengths above 385 nm. Eventually, the project evolved such that a UVA LED light source

was the preferred option, but the results of these early experiments are nonetheless interesting

and may point the way towards areas of future study. Figure 4.8 shows the degradation of DOC

and UV254 as a function of time and light source while Figure 4.9 shows the same data plotted

as a function of UVA light dose. Pseudo-first order reaction rate constants and model fits based

on time and light dose are provided in Table 4.6. All experiments were conducted with Otonabee

River water dosed with 0.15 g/L of P25 TiO2 nanoparticles.

When the data was plotted as a function of time NOM degradation was more effective under

UVA light than simulated solar light. Under the experimental conditions employed in these

experiments, the UVA lamp provided approximately 12 mW/cm2 of UVA light with a maximum

irradiance at 365 nm whereas the solar simulator provided 8.27 mW/cm2 of light between 300

nm and 400 nm. This discrepancy almost certainly explains why more NOM was removed in the

UVA/TiO2 experiments than in the solar/TiO2 experiments.

80


nanoparticles irradiated by simulated solar light or high intensity UVA light

as a function of irradiation time

When the data was instead plotted as a function of UVA dose, the two systems performed

similarly in terms of UV254 removal but UVA/TiO2 was more effective than solar/TiO2 for

DOC degradation. It may be that the simulated solar light, which included wavelengths ranging

from 300 nm to 1,100 nm, was able to photodegrade some portion of the aromatic structures in

the NOM compounds on its own (Winter et al. 2007) or through photo-assisted chemical

reactions with matrix components such as iron (Wu et al., 2005), leading to a reduction of

UV254 but not of DOC. Alternatively or additionally, the 8.27 mW/cm2 irradiance estimate cited

by Sokolowski (2014) includes all wavelengths between 300 nm and 400 nm, but TiO2 is only

activated by wavelengths below 385 nm, so the actual useable irradiance provided by the solar

simulator may have been below 8.27 mW/cm2.

0%

20%

40%

60%

80%

100%

0 20 40 60 80 100 120

NO

M R

emo

va

l


DOC - UVA DOC - Solar UV254 - UVA UV254 - Solar

81


nanoparticles irradiated by simulated solar light and high intensity UVA

light as a function of UVA dose

The pseudo-first order rate constants shown in Table 4.7 confirm that DOC reduction proceeded

more quickly under high intensity UVA light than under simulated solar light. The degradation

rate constants for UV254 removal were essentially equal under both light sources when

degradation was evaluated as a function of time but solar/TiO2 came out ahead when degradation

was evaluated as a function of UVA dose.


removal from synthetic water by 0.15 g/L of TiO2 P25 nanoparticles

irradiated by simulated solar light or high intensity UVA light

DOC UV254

Time k (min-1) R2 n k (min-1) R2 n

Solar -0.0010 0.93 8 -0.0059 0.99 8

UVA -0.0030 0.93 10 -0.0062 0.96 10

UVA Dose k (cm2/mJ) R2 n k (cm2/mJ) R2

Solar -0.0020 0.93 8 -0.0119 0.99 8

UVA -0.0041 0.93 10 -0.0086 0.96 10

0%

20%

40%

60%

80%

100%

0 20 40 60 80 100

NO

M R

emo

va

l

UVA Dose (mJ/cm2)

DOC - UVA DOC - Solar UV254 - UVA UV254 - Solar

82

4.2.1.5 Summary of Findings

The main findings of the preliminary NOM adsorption and photocatalytic degradation

experiments can be summarized as follows:

TiO2 dose affected both NOM adsorption and NOM degradation

Adsorption was highly effective for NOM removal from the synthetic river water but less

so in the real water matrices

NOM was degraded more slowly in the two real water matrices than in the synthetic

water matrix

The rate of UV254 removal was higher in OTB water than LO water

High intensity UVA light was more effective for DOC degradation than simulated solar

light even after the irradiance of each lamp was taken into account

Simulated solar irradiation was more effective than high intensity UVA light for UV254

reduction

4.2.2 Solar Photocatalysis with LENs for NOM Removal

The decision to explore the use of LENs for NOM removal came about gradually. At first, it was

simply an effort to characterize the precursor materials of the TiO2 membranes being prepared at

the University of Waterloo. The initial success of the materials in terms of NOM and dye

removal was promising, but it was an observation made in the summer of 2014 that finally

convinced me to commit to pursuing the LENs as a water treatment option in their own right.

The summer students at that time were responsible for preparing the treated samples for analysis

by filtering them through a 0.45 micron PES filter. They complained to me that the samples

containing P25 took a much longer time to filter than those containing nanobelts (NB), which

were the first LENs I learned to synthesize. Later, we observed that the NBs also settled out of

the water more readily than P25. These early observations suggested that the LENs might be a

good alternative to P25 because they were easy to remove from the water – a perennial challenge

for photocatalysis researchers.

83

4.2.2.1 Synthesis and Characterization of LENs

Three LENs were synthesized, characterized, and compared to standard P25 nanoparticles in

terms of their ability to degrade NOM through photocatalysis. Of the four materials prepared and

tested, only the NBs were easy to photograph using the SEM in Earth Sciences. These

photographs, shown in Figure 4.10 suggest that the NBs are less than 250 nm (0.25 m) in

diameter and 2000 to 5000 nm (2 to 5 m) in length.

Figure 4.10 SEM images of nanobelts (NBs)

Yuan and Su used high resolution TEM to obtain images of their materials. They determined that

the NTs and NWs were approximately 5 to 10 nm (0.005 to 0.010 m) in diameter and over a

micron in length. Unfortunately, the SEM that we had access to did not have high enough

resolution to accurately characterize such small materials, but the pictures I did get weren’t a

total failure, as demonstrated in Figure 4.11. Both pictures show evidence of some longer

nanostructures and the surfaces of the NW and NT agglomerates are wavy and fibrous-seeming,

suggesting that they are made up of many long, skinny linear units.

84

Figure 4.11 SEM images of nanowires (NW) and nanotubes (NT)

X-ray diffraction (XRD) was used to determine the crystal structure of the TiO2 in each of the

first generation LENs (Figure 4.12). It is well-established that anatase is the most

photocatalytically active TiO2 crystal structure (Luttrell et al., 2014) and that P25 NPs are made

up of approximately 75% anatase and 25% rutile (Bickley et al., 1991; Ohtani et al., 2010).

The NBs and NTs were predominantly made up of the anatase fraction, though the wider peaks

found in the NT spectra suggest that these also contained other TiO2 structures. Findings

presented in Chapter 6, which were obtained using a more complex crystal analysis method

(HRTEM and SAED), suggest that the NTs likely contained both anatase and TiO2(B), a less

photocatalytically active form of TiO2. The chromatogram for the NWs was similar to that of the

NTs but also contains peaks not found in the other two chromatograms. These peaks were not

indexed by the XRD instrument available in Earth Sciences, however, the peak just before 2θ =

30 corresponds to peaks that have been alternately identified as rutile, TiO2(B), K2Ti18O17, or

K2Ti6O13 (Yuan and Su, 2004; Zheng et al., 2010). Thus, although the NWs did contain anatase

and, quite likely TiO2(B), they also contained other as of yet unidentified TiO2 crystal structures.

85

Figure 4.12 XRD results for (a) nanobelts (NB), (b) nanowires (NW), and (c) nanotubes

(NT)

4.2.2.2 NOM Removal via Adsorption and Photocatalytic Degradation

The three LENs were compared to P25 in terms of their ability to remove DOC and UV254. The

differences in NOM removal ability between the different nanomaterials were often difficult to

explain with the small amount of information available at this stage of the project, so many of the

hypotheses presented here are speculative and informed by my later work, which is presented in

chapters 5 to 8 of this document.

Table 4.8 presents the percent removal of DOC from Otonabee River water achieved by 0.1 g/L

of each nanomaterial via adsorption and after 60 minutes of irradiation with simulated solar light.

P25 removed approximately 11% of the total DOC in the water via adsorption and approximately

22% after 60 minutes of irradiation. This is in line with the findings presented earlier in this

chapter (e.g. Figure 4.6) and in later chapters. The NBs removed very little DOC via adsorption

and approximately 11% after irradiation. The NB dataset was a particularly good fit to the

pseudo-first order degradation model, suggesting that photocatalytic degradation was the primary

mechanism of NOM removal for this material. This fits with the finding that the NBs contained

primarily anatase, the most photoactive form of TiO2. The NBs were also much larger than any

86

of the other nanomaterials, which means that they have less overall surface area available for

adsorption. This likely explains why they adsorbed so little DOC compared to the other three

materials.

Table 4.8 Summary of percent removal and kinetic parameters - DOC

Nanomaterial Adsorption Only Irradiation (60 min) k (min-1) R2

P25 11 ± 4 % 22 ± 3 % 0.0011 0.83

NB 3 ± 0 % 11 ± 1 % 0.0005 0.94

NW 16 ± 1 % 25 ± 2 % 0.0010 0.87

NT 16 ± 2 % 19 ± 2 % 0.0002 0.41

The NTs removed 16% of the overall DOC via adsorption and only slightly more than that after

60 minutes of irradiation. The low k value obtained for NTs as well as the poor fit of the pseudo-

first order model to the NT DOC removal dataset relative to those of the other materials suggest

that adsorption, rather than irradiation, was the main force driving NOM removal by this

material. The assumed small size of the NTs relative to larger materials such as the NBs may

have resulted in a higher overall surface area in the former case and thus more available

adsorption sites. The poor degradation observed for the NTs may be a function of their crystal

structure – TiO2(B) is known to be less photocatalytically active than anatase and the NTs likely

contained more TiO2(B) than the P25 nanoparticles and NBs. Other materials such as P25 and

NW were effective for both adsorption and degradation.

The UV254 show a similar pattern in terms of adsorption: The NW and NT, both of which likely

had relatively high available surface areas, were the most effective for NOM adsorption and the

NBs were the least effective. In fact, the NBs actually increased the UV254 signal of the sample

after dark adsorption, although this is quite likely due to passage of the material through the 0.45

m filter during sample preparation. This is somewhat counter-intuitive – if the NBs were in fact

the largest of the nanomaterials, they should also have been the least likely to pass through the

filter. In fact, however, most nanomaterials exist as agglomerates in solution, and the size of the

agglomerates depends on the geometry of the nanomaterials (Zhou et al., 2013) as well as their

charge characteristics and the properties of the water matrix (Hotze et al., 2010). These

phenomena, which are discussed in greater detail later in later chapters, may have driven the P25,

NT, and NW materials to form larger agglomerates than those formed by the NBs under similar

conditions. It should also be noted that later iterations of the NB material (see Chapter 7 and

87

Chapter 8) were modified such that they were much less likely to pass through the filters used for

sample preparation.

P25 and NT were the most effective for UV254 removal via photocatalytic degradation as

demonstrated by their higher percent removal after 60 minutes as well as their higher k values

relative to the other two materials (Table 4.9). All four datasets conformed very well to the

pseudo-first order degradation model. It is worth keeping in mind that a reduction in UV254

represents the oxidation of aromatic structures within the NOM molecules rather than their full

mineralization.

Table 4.9 Summary of percent removal and kinetic parameters – UV254

Material Adsorption Only Irradiation (60 min) k (min-1) R2

P25 11 ± 5 % 61 ± 2 % 0.0059 0.99

NB -10 ± 1 %1 43 ± 1 % 0.0045 0.97

NW 14 ± 0 % 48 ± 1 % 0.0036 0.99

NT 13 ± 1 % 63 ± 1 % 0.0062 0.99

1Negative percent removal due to TiO2 passage and subsequent interference in UV-Vis measurement

Some materials, due to either morphology, photoactivity, or other surface characteristics, appear

to have had a particular preference for degrading aromatic NOM (as measured by UV254) over

other types of NOM. The NTs were the most obvious example of this: They had the lowest

reaction rate constant for DOC degradation but the highest one for UV254 reduction. Others,

such as the NWs, showed less preference for aromatic NOM. The reasons for these phenomena

were not explored in detail when the experiments were being conducted or later in the project,

but might make for an interesting side project or MASc project in the future.

The effects of TiO2 adsorption and photocatalysis on NOM quantity and character were further

explored using LC-OCD. The data presented in Figure 4.13 and Figure 4.14 below was obtained

during experiments conducted with Otonabee River water dosed with 0.1 g/L of P25, NBs, NWs,

or NTs and irradiated with simulated solar light.

88

Figure 4.13 Distribution of NOM fractions in Otonabee River water samples dosed with

0.1 g/L of P25, NB, NW, or NT and mixed in the dark for 1 minute

At first glance, it appears that all four nanomaterials were capable of removing at least some

small part of the total concentration of NOM in the raw water through adsorption. P25 and NWs

appear to have been the most effective overall. The biopolymer and humic substances fractions

are the only ones that appear to have been adsorbed to any great degree.

0

1

2

3

4

5

6

Control P25 NB NW NT

DO

C (

mg

/L)

LMW Neutrals

LMW Acids

Building Blocks

Humics

Biopolymers

89

Figure 4.14 Effect of 60 minutes of photocatalysis with four LENs irradiated with

simulated solar light on the distribution of NOM fractions in the sample

The P25 nanoparticles removed NOM through both adsorption and photocatalytic degradation.

Adsorption predominantly affected the biopolymers and humic substances fractions. Degradation

appears to have targeted these fractions along with the building blocks fraction. The only fraction

to increase after irradiation was the LMW acid fraction, suggesting that some of the NOM

compounds from the other fractions were degraded to form LMW acids. Some were also,

presumably, mineralized, resulting in an overall decrease in the concentration of DOC in the

water.

Reductions in total DOC and individual fractions were more modest when NBs were used as the

photocatalyst. Very little NOM appears to have been removed through adsorption alone.

Photocatalytic degradation did reduce the biopolymers and humic substances fractions and

increased the size of the building blocks fraction. There was some overall DOC reduction after

60 minutes, suggesting that at least some NOM was mineralized.

Unlike the NBs, the NWs were quite adept at adsorbing NOM, though this was mostly limited to

the biopolymers and humic substances fractions. Interestingly, upon irradiation there was no

change in the concentration of these two fractions, but the building blocks fraction was

0

1

2

3

4

5

6

Control P25 NB NW NT

DO

C (

mg

/L)

LMW Neutrals

LMW Acids

Building Blocks

Humics

Biopolymers

90

decreased, suggesting that the NOM compounds in this fraction were targeted by photocatalytic

degradation by NWs.

Like the other nanomaterials, the NTs predominantly adsorbed compounds from the biopolymers

and humic substances fractions. After 60 minutes of irradiation the humic substances fraction

decreased while the building blocks, LMW acids, and LWM neutrals fractions all increased. This

suggests that, unlike the other nanomaterials, the NTs create a wide spectrum of intermediate

compounds.

The overall DOC reductions achieved by the various nanomaterials were not impressive. It

should be noted that the DOC values obtained through the LC-OCD analysis were not exactly

the same as those obtained using the DWRG TOC analyzer because the former represent the sum

of the five individual fractions while the latter included all organic matter present in the sample.

No matter what, however, the results indicate that P25 and NTs were the most effective for the

overall reduction of DOC.

Different nanomaterials appear to have had different adsorption affinities for biopolymers and

humic substances. For example, P25 adsorbed approximately 50% of the biopolymers and 20%

of the humic substances in the water whereas the NBs adsorbed only 15% of the biopolymers

and 5% of the humic substances. The NWs were particularly adept at adsorbing biopolymers

(~65%) but less so at adsorbing humic substances (~18%). The NTs adsorbed approximately

45% of the biopolymers and 15% of the humic substances. These differences may be explained

by differences in available surface area but may also be related to differences in charge.

The nanomaterials also varied in their ability to degrade different NOM fractions. The best

reduction of biopolymers after 60 minutes of solar irradiation (85%) was achieved by P25. The

next best materials for biopolymer reduction was the NBs (surprising given the lack of

adsorption) at approximately 70%. The NTs reduced the biopolymer fraction by a respectable

60% after 60 minutes of irradiation. It is difficult to tell how well the nanowires worked because

of the large discrepancy between the two replicate samples.

The results seem to suggest that some LENs worked mainly through adsorption (NWs) while

others worked mainly through photocatalysis (NBs). This was fraction dependent. The results

also suggest that different nanomaterials produce different intermediates during degradation or,

91

alternatively, the different nanomaterials simply work at different rates and with different

contributions from adsorption and photocatalytic degradation.

4.2.2.3 Summary of Findings

The results of the experiments conducted with the first generation LENs can be summarized as

follows:

LENs synthesized at 190oC were larger than those synthesized at 130oC

LENs calcined at 550oC adsorbed NOM more effectively than NBs calcined at 700oC

The DOC, UV254, and LC-OCD results indicate that all of the materials showed a

marked preference for humic substances and other aromatic NOM compounds

4.2.3 LENs for Dye Removal

Indicator dyes, in particular methylene blue dye, are widely used to quickly evaluate the

photocatalytic properties of novel nanomaterials. In these experiments, centrifugation (10,000

rpm for 30 minutes) was used to remove the nanomaterials from the treated solution instead of

filtration because the PES lab filters were found to adsorb some 10% of the overall methylene

blue dye present in solution. Centrifugation was not as effective at removing the materials

because of the propensity of the materials to become resuspended as the dye solution was

decanted out of the centrifugation tubes, which may account for the higher standard deviations

observed among the replicates.

The results of methylene blue degradation tests conducted with P25 nanoparticles and the three

LENs are provided in Figure 4.15. Methylene blue has a pKa of 3.8 and the dye solution had a

pH 5.6 ± 0.2, so the molecules should have had a slightly negative charge. At this pH, the P25

nanoparticles, which have an IEP between 6 and 6.5 would be expected to be electrostatically

neutral or slightly positively charged. It is surprising then that there was no appreciable

adsorption of methylene blue to the surface of the P25 nanoparticles. In this case it seems likely

that factors other than charge interactions predominated in the adsorption (or rather, lack of

adsorption) of methylene blue to P25. The NBs and NTs were also unable to adsorb any

methylene blue dye, but the NWs adsorbed nearly 20%. Whether this was related to charge

effects, surface area, or crystal structure was unclear because these parameters were never

92

definitively elucidated for this material. Other researchers (Xiong et al., 2010) have also

managed to synthesize LENs that adsorb methylene blue dye using a hydrothermal method

similar, but not identical, to that used to synthesize the NWs. Their LENs were multiwalled tubes

with diameters of approximately 10 nm. Based on the XRD pattern of their LENs, which was

similar but not identical to that obtained in this study for the NWs, they hypothesized that they

contained predominantly titanate, a TiO2 structure with limited photocatalytic ability.

P25 and the NBs were the most effective of the four nanomaterials at decolourizing methylene

blue dye. P25’s small size results in a high available surface area, which may explain its superior

degradation behaviour. Despite its lower overall available surface area, NB was also highly

effective, likely because it contained only anatase, which is particularly photoactive. The other

two materials, which contained unidentified non-anatase/non-rutile TiO2 structures, were less

effective for methylene blue decolourization.

Figure 4.15 Adsorption and decolourization of methylene blue dye by 0.1 g/L of P25

nanoparticles or one of three LENs after 0, 15, 30, 45, and 60 minutes of

irradiation with simulated solar light

Acid orange dyes, a subset of azo dyes, contain the azo structure (R-N≡N-R’) and have pKa

values ranging from 9 to 11 (Pérez-Urquiza and Beltrán, 2001). This results in charge

characteristics very different from those of methylene blue dye. AO24 is a sulfonated diazo dye

that is used in the textile industry and is highly recalcitrant to treatment (Chacón et al., 2005). It

0%

20%

40%

60%

80%

100%

120%

P25 NB NW NT

Met

hyle

ne

Blu

e D

ecolo

uri

zati

on

0 min 15 min

93

has previously been used to evaluate the effectiveness of solar photocatalytic disinfection

systems (Bandala et al., 2011). In this project, AO24 was selected mainly because it had been

used in previous photocatalytic experiments at the DWRG. The results of adsorption and

photocatalytic degradation tests conducted with P25 nanoparticles and the three LENs and AO24

dye are presented in Figure 4.16.

Figure 4.16 Adsorption and decolourization of AO24 by 0.1 g/L of P25 nanoparticles or

one of three LENs after 0, 15, 30, 45, and 60 minutes of irradiation with

simulated solar light

One notable finding was that P25 nanoparticles adsorbed more AO24 than they did methylene

blue. Once again, however, this was not easily explained by charge interactions. The AO24 test

solution had an average pH of 6.0 ± 0.3, and at this pH the nanoparticles were neither positively

or negatively charged. None of the LENs adsorbed AO24 in these experiments, though later

generations of LENs did adsorb it effectively (see Chapter 8).

AO24 was more amenable to photocatalytic degradation by all four materials than methylene

blue was. After only 15 minutes of irradiation P25 was able to break down nearly 100% of the

AO24 in solution. The LENs lagged behind P25 but only slightly. Light only control samples

irradiated for 60 minutes did not show evidence of photodecolourization.

0%

20%

40%

60%

80%

100%

120%

P25 NB NW NT

AO

24

Dec

olo

uriz

ati

on

0 min 15 min 30 min 45 min 60 min

94

Summary, Conclusions, and Implications for Future Experiments

4.3.1 Light Source

A decision was made in early 2015 to switch from using simulated solar irradiation to UVA

irradiation. The reasoning underlying this decision was based in part on the preliminary results

presented in Section 4.2.1.4 and in part on the literature review and analysis that took place

during my comprehensive exam. A summary of the characteristics, advantages, and

disadvantages of various UVA light options is presented in Table 4.10.

Table 4.10 Summary of characteristics, advantages, and disadvantages of UVA light

sources

Parameter Solar UVA Lamp Germicidal UV UVA LEDs

Light Source Sun UVA lamp LP or MP UV lamps UVA LEDs

Light Intensity (UVA) Max 30 W/m2

(3 mW/cm2)

Variable Variable Variable

Wavelengths 300 – 1100 nm 365 nm LP: 254 nm

MP: Polychromatic

365 nm

Light Intensity Control Passive Active Active Active

Location Outdoor Indoor Indoor Indoor

Energy Requirements Low High High Low

Lamp Cost None Moderate High Low

Reactor Configuration High surface

area / flow

High surface

area / flow

Tubular Flexible

Concurrent

disinfection

Some (SODIS) No Yes No

By definition, solar/TiO2 processes are designed to operate outdoors or, at the very least in view

of the sun. Although this means that they are at the mercy of global tilt and weather conditions, it

also means that small-scale solar/TiO2 systems can be installed in areas where the construction of

a full-scale building is not feasible. The dependence of solar/TiO2 systems on direct access to

solar light can complicate system operation and design in other ways, however. Most

importantly, even in sunny locations full sun conditions occur for only part of the day. The use of

CPC light collectors can extend this to some degree (Malato et al., 2009) but nightfall is

inevitable and will bring system operation to a halt. This limited operating window means that

95

the daily water demand of the plant must be satisfied over the course of only a few hours. This

translates to high flows through the plant and correspondingly large associated unit processes

able to provide adequate treatment for such high flows. UV/TiO2 processes, by contrast, are

operated indoors. This greatly increases the operational flexibility and allows for the design of

smaller equipment because it can be operated non-stop and in all weather conditions. This, in

theory, makes UV/TiO2 easier to incorporate into existing water treatment trains designed to

provide the daily water demand over an 18 or 24 hour operating cycle.

The vast majority of TiO2 photocatalysis studies make use of high intensity UVA lamps, usually

with a maximum irradiance at 365 nm. These lamps are widely available, emit light in the

required range and come in a variety of configurations, making it easy to adapt them to bench-

scale apparatus. Unlike solar light, they do not provide substantial concurrent disinfection.

Although UVA light is the most common option for TiO2 photocatalysis, the photocatalyst can

be activated by all wavelengths below 385 nm. This means that the low pressure (LP) and

medium pressure (MP) UV lamps commonly used for drinking water disinfection are able to

activate the photocatalyst. These lamps require substantial energy input and can only be used in a

limited number of configurations, however, they can potentially provide concurrent disinfection.

UVA light emitting diodes (LEDs) are an attractive alternative to the other light sources

presented here. They are inexpensive, long lasting, and are less energy intensive than the other

UV lamps. They are also small and easy to integrate into different reactor configurations. Robert

Liang from the University of Waterloo developed a UVA LED batch testing apparatus in 2015

for experiments being conducted at his university and he was willing to help us build one of our

own. This was completed in the summer of 2015 and used for all subsequent experiments. The

irradiance at the centre of the beam of each of the four LED lamps has remained constant (6.25

mW/cm2) for nearly two years. Note that this value corresponds to the irradiance measured at the

centre of the sample at the surface of the water. Despite the installation of collimating cylinders,

the irradiance reaching the samples was not constant but rather dropped off as a function of

distance from the centre of the light beam. A spreadsheet developed by Bolton and Linden

(2003) was used to calculate the average irradiance across the surface of the sample, which

turned out to be approximately 4.9 mW/cm2.

96

4.3.2 Selection of Optimal LENs

A decision was made to carry two of the first generation LENs, NB and NT, forward for future

experiments. The NBs were chosen because they were the largest material, made of nearly pure

anatase, and appeared relatively uniformly sized in the SEM images. The NTs were chosen

because their synthesis procedure was nearly identical to that of the NBs with the exception of

the hydrothermal and calcination temperatures. Based on the results of other researchers (Yuan

and Su, 2004) and the preliminary results in Section 4.2.2.1, it was hypothesized that the

hydrothermal temperature predicted the size of the LENs and the calcination temperature

predicted their crystalline composition (anatase vs. TiO2(B) vs. rutile). Two additional materials

were added to the suite as shown in Table 4.11 in order to help elucidate these effects. A decision

was also made to change from four days of hydrothermal reaction at 190oC to one day of

hydrothermal reaction at 240oC. This change had no apparent effect on the size or reactivity of

the LENs.

Table 4.11 Second generation LENs synthesis conditions

Material Basic Solution Hydrothermal

Temperature (TH)

Calcination

Temperature (TC)

NB 130/550 (NT) NaOH 130oC 550oC

NB 130/700 NaOH 130oC 700oC

NB 240/550 NaOH 240oC 550oC

NB 240/700 (NB) NaOH 240oC 700oC

4.3.3 Natural vs. Synthetic Water Matrices

Finally, a decision was made to use real water matrices rather than synthetic water matrices for

the majority of future experiments. The adsorption and degradation of SRNOM and other NOM

isolates by TiO2 has been explored by numerous researchers including Mwaanga et al. (2014),

Erhayem and Sohn (2014), Huang et al. (2008), and Valencia et al. (2013). Studies by Sanly Liu

(Liu et al., 2008, Liu et al., 2010a; Liu et al. 2010b) and others at the University of New South

Wales in Australia such as Ng et al. (2014) have demonstrated that real water matrices can be

more difficult to treat than synthetic ones but also that TiO2 photocatalysis is nonetheless a

97

competitive option for NOM removal from real water. The decision to use real water matrices for

experiments, most often raw water from the Otonabee River and the Ottawa River, was inspired

by the findings of the researchers cited above as well as a desire to test any resulting new

technology under worst case conditions.

References

Bandala, E.R., González, L., de la Hoz, F., Pelaez, M.A., Dionysiou, D., Dunlop, P.S.M., Byrne,

J.A., and Sanchez, J.L. (2011) Application of azo dyes as dosimetric indicators for enhanced

photocatalytic solar disinfection, Journal of Photochemistry and Photobiology A: Chemistry,

218, 185-191



215

Chacón, J.M., Leal, M.T., Sánchez, M., and Bandala, E.R. (2006) Solar photocatalytic

degradation of azo-dyes by photo-Fenton process, Dyes and Pigments, 69, 144-150

Erhayem, M. and Sohn, M. (2014) Effect of humic acid source on humic acid adsorption onto

titanium dioxide nanoparticles, Science of the Total Environment, 470-471, pp.92-98

Gerrity, D., Mayer, B., Ryu, H., Crittenden, J., Abbaszadegan, M., 2009. A comparison of pilot-

scale photocatalysis and enhanced coagulation for disinfection byproduct mitigation. Water

Research, 43, 1597–1610



pp. 1142-1150

Hotze, E.M., Phenrat, T., and Lowry, G.V. 2010, Nanoparticle aggregation: Challenges to

understanding transport and reactivity, Journal of Environmental Quality, 39, 1909-1924

98



Linden, K.G., Sharpless, C.M., Andrews, S.A., Atasi, K.Z., Korategere, V., Stefan, M., Mel

Suffet, I.H., 2004. Innovative UV Technologies to Oxidize Organic and Organoleptic Chemicals.

Awwa Research Foundation, Denver, CO, USA.

Liao, C-H, Kang, S-F, and Wu, F-A 2001. Hydroxyl radical scavenging role of chloride and


Liu, S., Lim, M., Fabris, R., Chow, C., Chiang, K., Drikas, M., and Amal, R. (2008) Removal of

humic acid using TiO2 photocatalytic process – Fractionation and molecular weight










M. Ng, E.T. Kho, S. Liu, M. Lim, R. Amal, Highly adsorptive and regenerative magnetic TiO2

for natural organic matter (NOM) removal in water, Chemical Engineering Journal, 246 (2014)

196-203

Pérez-Urquiza, M. and Beltrán, J.L. (2001) Determination of the dissociation constants of

sulfonated azo dyes by capillary zone electrophoresis and spectrophotometry methods, Journal

of Chromatography A, 917, 331-336

Philippe, K.K., Hans, C., MacAdam, J., Jefferson, B., Hart, J., and Parsons, S.A. (2010)



99

Sokolowski, A. (2014) Effects of Nanostructured TiO2 Photocatalysis on Disinfection By-

product Formation, Thesis, University of Toronto, Toronto, Canada

Valencia, S., Marin, J.M., Restrepo, G., and Frimmel, F.H. (2013) Evaluations of the

TiO2/simulated solar UV degradations of XAD fractions of the natural organic matter from a bog

lake using size-exclusion chromatography, Water Research, 47, 5130-5138

Wassink, J.D., Andrews, R.C., Peiris, R.H., and Legge, R.L. (2011) Evaluation of fluorescence


by enhanced coagulation, Water Science and Technology: Water Supply, 11 (5), p. 621

Watanabe, T., Nakajima, A., Wang, R., Minabe, M., Koizumi, S., Fujishima, A., and Hashimoto,

K. (1999) Photocatalytic activity and photoinduced hydrophilicity of titanium dioxide coated

glass, Thin Solid Films, 351, 260-263

Winter, A.R., Fish, T.A.E., Playle, R.C., Smith, D.S., and Curtis, P.J. (2007) Photodegradation of

natural organic matter from diverse freshwater sources, Aquatic Toxicology, 84, 215-222

Wu, F.C, Mills, R.B., Cai, Y.R., Evans, R.D., and Dillon, P.J. (2005) Photodegradation-induced

changes in dissolved organic matter in acidic waters, Canadian Journal of Fisheries and Aquatic

Sciences, 62 (5), 1019

Xiong, L., Yang, Y., Mai, J., Sun, W., Zhang, C., Wei, D., Chen, Q., and Ni, J. (2010)

Adsorption behaviour of methylene blue onto titanate nanotubes, Chemical Engineering Journal,

156, 313-320







properties on TiO2 nanoparticle agglomeration, PLOS One 8 e81239

100

Adsorption of Natural Organic Matter and Disinfection Byproduct Precursors from Surface Water onto TiO2

Nanoparticles: pH Effects, Isotherm Modeling, and Implications for the Use of TiO2 for Drinking Water Treatment

The contents of this chapter were published as:




doi: 10.1016/j.chemosphere.2017.01.125

Permission has been obtained from Elsevier to reprint this material in this thesis (license number

416391009421).

The supplementary material associated with this paper is provided in Section 5.6.

Abstract

Titanium dioxide is a photocatalyst that can remove organic contaminants of interest to the

drinking water treatment industry, including natural organic matter (NOM) and disinfection

byproduct precursors. The photocatalytic reaction occurs in two steps: Adsorption of the

contaminant followed by degradation of the adsorbed contaminant upon irradiation with UV

light. The second part of this process can lead to the formation of reactive intermediates and

negative impacts on treated water quality such as increased DBP formation potential (DBPfp).

Adsorption alone does not result in the formation of reactive intermediates and so may prove to

be a safe way to incorporate TiO2 into drinking water treatment processes. The goal of this study

was to expand on the current understanding of NOM adsorption to TiO2 and examine it in a

drinking water context by observing NOM adsorption from real water sources and evaluating the

effects of the resulting reductions on the DBPfp of the treated water. Bottle point isotherm tests

were conducted with raw water from two Canadian water treatment plants adjusted to pH 4, pH

6, and pH 8 and dosed with TiO2 nanoparticles. The DOC results were a good fit to a modified

Freundlich isotherm. DBP precursors and LC-OCD NOM fractions associated with DBP

formation were removed to some extent at all pHs but most effectively at pH 4.

101

Introduction

Nanomaterials, defined as materials with any dimension in the nanoscale or having internal or

surface structure in the nanoscale (ISO, 2010), are increasingly being used in the fields of

electronics, computing, and medicine. Some nanomaterials, including TiO2 nanoparticles, may

also prove to be useful in environmental applications, including the treatment of drinking water

and wastewater.

5.1.1 Titanium Dioxide for Drinking Water Treatment

The use of TiO2 for water purification has been explored by many researchers and at least two

companies have developed small-scale systems based on TiO2 photocatalysis, but it has yet to be

widely applied for municipal drinking water treatment. Photocatalytic degradation of aqueous

contaminants by TiO2 is generally thought to occur in two steps: Adsorption and degradation.

The first step, adsorption, can occur in the absence of light, but degradation only occurs when

TiO2 is irradiated. Upon irradiation, reactive oxygen species (ROS), including the hydroxyl

radical, and oxidative and reducing centres on the surface of the nanoparticle degrade

contaminants that have been adsorbed on the surface of the photocatalyst. This two step reaction

is often described using the Langmuir-Hinshelwood mechanism (Malato et al., 2009). The two

stage nature of photocatalytic degradation differentiates it from other advanced oxidation

processes such as UV/H2O2 and O3/H2O2. Like other AOPs, however, many of the ROS formed

when TiO2 is irradiated with UVA light are non-specific oxidants, so TiO2 photocatalysis has the

potential to provide concurrent disinfection and degradation of organic drinking water

contaminants, including taste and odour compounds and cyanotoxins (Fotiou et al., 2015),

various pharmaceuticals (Avisar et al., 2013; Kanakaraju et al., 2014), and disinfection

byproduct precursors such as natural organic matter (NOM) (Liu et al., 2008, Huang et al., 2007,

Philippe et al., 2010). The photocatalytic degradation process preferentially targets large

aromatic NOM compounds, breaking them down into smaller ones (Huang et al., 2007; Philippe

et al., 2010). This can result in decreased membrane fouling (Huang et al., 2007) and changes in

disinfection byproduct formation potential (DBPfp). The latter is of particular concern because

although some of the degradation products of photocatalysis may be more likely to form DBPs

than the original compounds, others may have an equal or greater DBPfp, particularly at shorter

102

treatment times (Liu et al., 2008). As a result, an adsorption-based process may prove to be a

safer and more effective option for NOM removal from drinking water using TiO2.

5.1.2 Natural Organic Matter

The removal of NOM, a heterogenous mixture of organic compounds found in most surface

water sources, is an important goal in drinking water treatment due to its aesthetic, operational,

and health effects. The latter are primarily linked to the role of NOM as a precursor to the

formation of both regulated and unregulated disinfection byproducts (DBPs). Operational

concerns related to NOM include membrane fouling, competitive adsorption to adsorbent

materials intended for the removal of taste and odour compounds, interference with UV

disinfection, and consumption of chlorine during disinfection.

NOM is commonly quantified as total organic carbon (TOC) or dissolved organic carbon (DOC),

bulk parameters that measure the total mass of organic carbon compounds present in a water

sample without differentiating them from one another. UV-Vis absorbance at 254 nm (UV254)

and fluorescence are also used to characterize NOM, though these methods are specific to NOM

compounds containing aromatic chromophores or fluorophores. More complex methods such as

liquid chromatography with organic carbon detection (LC-OCD) have been developed that allow

researchers to separate NOM into different fractions based on size or chemical characteristics.

The resulting fractions are then quantified using DOC or UV254. LC-OCD separates NOM into

five fractions based on size and/or chemical characteristics as follows: Biopolymers, associated

with membrane fouling (Wray et al., 2013); humic substances, which have been linked to DBP

formation (Wassink et al., 2011); building blocks; low molecular weight acids (LMWA); and

low molecular weight neutrals (LMWN) (Huber et al., 2011). A given NOM sample’s potential

to form regulated and unregulated DBPs can also be assessed more directly by measuring its

DBPfp by chlorinating water samples and measuring the DBPs formed under standardized

conditions.

5.1.3 Adsorption of NOM to TiO2

Studies by Mwaanga et al. (2014), Erhayem and Sohn (2014), and Kim and Shon (2007) have

established that pH and ionic strength have effects on the adsorption of NOM to TiO2 and have

103

noted that larger, more aromatic NOM compounds are adsorbed preferentially. The presence of

bicarbonate, phosphate, and nitrate have all been shown to decrease NOM adsorption to TiO2

while the presence of magnesium and calcium increase the likelihood of NOM adsorption

(Erhayem and Sohn, 2014; Sun and Lee, 2012).

Other studies have described the effects of individual ions on the agglomeration of TiO2

nanomaterials. Agglomeration decreases the overall surface area available for adsorption and as

such is likely to have an impact on the ability of TiO2 nanomaterials to adsorb NOM. Liu et al.

(2013) reported that three types of TiO2 nanomaterials were more likely to agglomerate under

high ionic strength conditions than at low ionic strength conditions. This finding is corroborated

by those of Erhayem and Sohn (2014). The type of ions present in solution may also have an

effect – Liu et al. (2013) observed greater increases in agglomerate size when calcium was added

to the water rather than sodium. They hypothesized that this was due to the greater ability of Ca2+

to compress the electrical double layer surrounding the nanomaterials relative to Na+. Greater

compression of the electrical double layer results in less repulsion between individual

nanoparticles and thus, greater agglomeration.

It has also been observed that the presence of natural organic matter increases the stability of

nanomaterials in solution, though this effect is less pronounced in the presence of ions such as

calcium (Liu et al., 2013; Zhang et al. 2009) and at high NOM concentrations (Erhayem and

Sohn, 2014). According to Zhang et al. (2009), NOM inhibits agglomeration by increasing the

overall negative charge of the particles and thus increasing the repulsive forces that keep them

dispersed in solution.

5.1.4 Adsorption Models

Adsorption processes are usually evaluated in the laboratory using adsorption isotherm models.

The Freundlich isotherm often fits well to empirical data and can be used to model

heterogeneous systems such as the adsorption of organic molecules to activated carbon

(Summers et al., 1988). It can be used to describe multilayer adsorption, reversible adsorption,

and adsorbents with non-uniform adsorption sites (Shahbeig et al., 2013).

Summers and Roberts (1988) found that a modified version of the Freundlich isotherm could be

used to describe the adsorption of NOM to activated carbon when experiments were conducted

104

with a constant initial concentration of NOM and changing doses (D) of activated carbon. They

developed the following equation (referred to in this paper as the SR model) to express this

relationship:

𝑞𝑒 = 𝐾𝑆𝑅 (𝐶𝑒

𝐷)

1

𝑛𝑆𝑅 (5.1)

Where D has units of mg L-1 or g L-1. The SR model has been used and extended upon by

numerous researchers including Karanfil et al. (1999), Li et al. (2002), Hyung and Kim (2008),

and Qi et al. (2008) to characterize the adsorption of NOM to activated carbon as well as carbon

nanotubes. Erhayem and Sohn (2014) modeled the adsorption of Suwannee River humic acids,

fulvic acids, and NOM to P25 TiO2 nanoparticles using the SR isotherm model. They noted that

the SR adsorption constant (and thus the extent of adsorption) increased at lower pH and at

higher ionic strength.

5.1.5 Potential Risks and Opportunities Associated with the Use of TiO2 Nanoparticles

for Water Treatment

Conventional water treatment technologies such as coagulation and activated carbon are

effective for NOM removal but, like all treatment technologies, have limitations. In North

America, coagulation is widely used to remove NOM and turbidity from drinking water but it

does not remove some recalcitrant organics. It also creates a substantial amount of waste, often

referred to as coagulation residuals. Activated carbon readily removes NOM and other organic

compounds but eventually becomes exhausted, and must undergo an expensive and energy

intensive regeneration process if it is to be reused. Although TiO2 is more well known for its

photocatalytic properties, it also adsorbs NOM as part of that process, and is potentially

regenerable onsite (Liu et al., 2014). As such, it might prove to be a useful alternative to existing

treatment options.

TiO2 nanoparticles are not without health and environmental concerns. Inhalation is the route of

exposure of greatest concern for human health (Shi et al., 2013), but the transport of

nanoparticles through the environment is coming under increasing scrutiny (Yang and

Westerhoff, 2014). The adsorption of various water components, including natural organic

matter (NOM), to TiO2 can facilitate the latter’s transport through water systems. Most of the

105

studies that have been conducted to-date on the interactions between NOM and TiO2 have aimed

to elucidate these effects in an effort to predict and minimize the impact of nanomaterials on the

natural environment. Examples include papers by Mwaanga et al. (2014), Erhayem and Sohn

(2014), Kim and Shon (2007), and Liu et al. (2013).

This study aims to expand upon existing research into the adsorption of NOM by TiO2, most of

which has been conducted in a contaminant transport context. It explores the potential

application of TiO2 as an adsorbent in drinking water treatment by studying its behaviour in

natural surface water sources and measuring its ability to remove DOC, UV254, and disinfection

byproduct precursors.


5.2.1 Materials

Evonik Aeroxide P25 TiO2 nanoparticles were purchased from Sigma Aldrich (Canada) and used

without further modification. THM and HAA standards were also purchased from Sigma

Aldrich. DOC standards were made by dissolving potassium hydrogen phthalate into MilliQ

water to create a 1,000 mg/L stock solution, which was then diluted as required. Raw water was

obtained from the inlets of two water treatment plants (WTPs) in Ontario, Canada, both of which

use surface water supplies. The Peterborough WTP is supplied by the Otonabee River while the

R.C. Harris WTP in Toronto draws its water from Lake Ontario, one of the largest lakes in North

America, which provides water to over 9 million people in Canada and the United States. All

water samples were obtained ahead of any pre-chlorination point at the WTP and characterized

upon return to the laboratory. Measured ranges for five relevant water parameters are provided in

Table 5.S.1 in the supplemental file. The water sources varied primarily in terms of their NOM

content and aromaticity. The DOC of the Otonabee River water was approximately 3 to 4 times

higher than that of the Lake Ontario water while its UV254 was approximately 5 times higher.

The SUVA of the Otonabee River water ranged from 2.0 to 2.4 L/mg.m while that of the Lake

Ontario water ranged from 0.8 to 1.0 L/mg.m, indicating that the NOM in the former is more

aromatic than that in the latter.

106

5.2.2 Analytical Methods

Dissolved organic carbon (DOC) was determined using an O.I. Analytical Aurora 1030W TOC

analyzer operating in persulfate oxidation mode and UV absorption at 254 nm (UV254) was

analyzed on an Agilent 8453 UV-Vis analyzer. SUVA was calculated by normalizing UV254 by

DOC. Alkalinity was measured using Standard Method 2320 (APHA, 2005).

Duplicate raw and treated water samples were analyzed using size exclusion liquid

chromatography with organic carbon detection (LC-OCD) as described by Huber et al. (2011).

The results of the analyses were processed using proprietary software (ChromCalc, DOC-

LABOR, Karlsruhe, Germany).

The isoelectric point of the nanoparticles was determined by measuring the zeta potential of the

nanoparticles at different pH values. A series of samples containing 0.1 g/L TiO2 in 10 mM NaCl

were adjusted to pHs ranging from 2 to 9 using 0.1 N HCl or 0.1 N NaOH as per the method

outlined by the Nanotechnology Characterization Laboratory (2009). The zeta potential of the

samples was measured using a Horiba Scientific Nanopartica SZ-100 Nanoparticle Analyzer.

The size of the nanoparticle agglomerates formed at different pHs was determined using a

Malvern MasterSizer 3000.

The uniform formation conditions (UFC) method as described by Summers et al. (1996) was

used to assess the chlorine demand and DBPfp of the raw water and the water that had been

treated with TiO2. The UFC test was designed to mimic the conditions commonly found in

distribution systems in North America. The chlorine demand and DBPfp tests were conducted on

samples buffered with a borate solution and adjusted to pH 8 with 1 N HCl or 1 N NaOH.

Samples were stored in the dark at 20oC for 24 hours, after which the free chlorine residual was

measured using Standard Method 4,500-G (APHA 2005). The DBPfp samples were dosed with

sufficient sodium hypochlorite to ensure that they would have a chlorine residual of 1 ± 0.4 mg/L

after the 24 hour holding time. After 24 hours the trihalomethanes and haloacetic acids formed

during the UFC tests were extracted according to Standard Method 6232 B and Standard Method

6251 B (APHA, 2005) and analyzed on a Agilent 7890B GC-ECD. Standard Method 6232 B and

Standard Method 6251 B (APHA, 2005). Blanks and 20 g/L check standards were analyzed

after every 10 samples.

107

All statistical analyses, including Tukey’s method for multiple comparisons to establish a point

of practical equilibrium and linear regression of the adsorption data to determine KF and KSR,

were conducted at the 95% confidence level. Reported error values represent half of the

calculated confidence interval unless otherwise specified.

5.2.3 Sample Preparation

NOM degradation studies were conducted in a high intensity UV reactor equipped with UV LED

lamps emitting UVA light at 365 cm with an average irradiance of 4.9 mW/cm2 at the surface of

the sample. Unchlorinated raw water from the Peterborough Water Treatment Plant (Otonabee

River water) was dispensed into three 50 mL batch reactors, dosed with 0.25 g/L of P25

nanoparticles, allowed to mix in the dark for one minute, and then exposed to the LED light for

times ranging from 0 to 60 minutes. The treated samples were chlorinated according to the UFC

method and analyzed for THMfp and HAAfp as described in Section 2.2.

The time required to reach a stable adsorption equilibrium between NOM and TiO2 nanoparticles

was determined by adding 75 mL of raw unchlorinated water to duplicate 125 mL amber bottles

and, when necessary, adjusting the pH to 4, 6, or 8 with 1 N HCl or 1N NaOH. The bottles were

dosed with 0.5 g/L of Evonik P25 TiO2 nanoparticles and mixed end-over-end in a box mixer for

times ranging from one to eight hours. This time range was chosen because previous experiments

(results not shown) had indicated that most NOM adsorption occurred within 5 minutes and that

equilibrium likely occurred between one and four hours.

For the bottle point isotherm tests, eight 250 mL amber bottles were filled with 150 mL of raw

water; adjusted to pH 4, 6, or 8; dosed with 0, 0.01, 0.025, 0.05, 0.1, 0.25, 0.5, or 1 g/L of Evonik

P25 TiO2 nanoparticles; and then mixed continuously in the dark for four hours in an end-over-

end box mixer. All experiments were run in triplicate. All replicate samples were analyzed for

DOC, UV254, and SUVA. The samples from one replicate experiment were used to establish

chlorine demand and the remaining samples were analyzed for THMfp and HAAfp. The DOC,

THMfp, and HAAfp results of the bottle point isotherm tests were evaluated for fit against the

linearized Freundlich and SR models.

108

All raw and treated samples were filtered through a 0.45 m polyethersulfone (Pall) laboratory

filter in a standard vacuum filtration apparatus to remove particulate matter and TiO2 ahead of

DOC, UV254, DBPfp, and LC-OCD analysis.


5.3.1 Disinfection Byproduct Formation During Photocatalysis

During the irradiation tests the THMfp and HAAfp of the raw Otonabee River water were

modestly reduced by adsorption alone (15% and 10%) but both increased upon irradiation

(Figure 5.1). The impact on THMfp was particularly dramatic: After only 5 minutes of

irradiation THMfp increased by 61% relative to the control. After 30 minutes, THMfp began to

decrease and after 60 minutes the THMfp of the treated water matched that of the control. The

increase in HAAfp upon irradiation was smaller than that of THMfp and was reversed after 30

minutes of irradiation. After 60 minutes of irradiation HAAfp was reduced by 35% relative to the

control.

Figure 5.1 THMfp and HAAfp of Otonabee River water treated with 0.25 g/L and

irradiated by high intensity UVA LED light

0

50

100

150

200

250

300

Control 0 5 15 30 45 60

DB

Pfp

(

g/L

)


THMfp

HAAfp

109

These results, which are similar to those obtained by Liu et al. (2008) and Philippe et al. (2009),

illustrate the risk associated with the use of photocatalysis for NOM and DBPfp reduction.

Although both THMfp and HAAfp were eventually reduced by the treatment, both increased at

treatment times between 0 and 15 minutes. Shorter treatment times are desirable as they

minimize the amount of space and energy required at full-scale. The modest reductions of

THMfp and HAAfp via adsorption, however, suggest an alternative treatment option – could

higher concentrations of TiO2 adsorb a sufficient amount of DBP precursors to provide a viable

reduction in overall DBPfp and do certain water quality conditions (e.g. pH) favour the

adsorption of DBP precursors to TiO2?

5.3.2 NOM Removal via Adsorption – Time Series Experiments

The practical adsorption equilibrium, defined as the point at which the 95% confidence of

neighbouring means began to overlap and the slope of the line of mean concentration vs. time

could no longer be distinguished from zero, was determined based on the results of the DOC and

UV254 time series experiments conducted in each water source. Irrespective of the water type

used or the parameter observed, the results indicated that the majority of NOM adsorption to the

P25 TiO2 particles occurred within minutes and that a practical adsorption equilibrium was

reached within one hour (see Figure 5.S.1, Figure 5.S.2, and Figure 5.S.3 in the supplementary

material at the end of the chapter). This is similar to results obtained by some researchers

working with TiO2 materials, including P25 nanoparticles (Kim and Shon, 2007; Ng et al.,

2014), though others have suggested that a longer period of time might be required to reach full

equilibrium (Mwaanga et al., 2014; Erhayem, 2013).

Based on the results of the time series experiments conducted in this study, all subsequent

equilibrium experiments were conducted with a four hour adsorption period to ensure that all

data was gathered at a point well beyond the practical point of equilibrium. The results of the

time series experiments also indicated that adsorption was most effective at pH 4 and that

UV254 and SUVA were more strongly affected by the treatment than DOC, hinting that

aromatic NOM may have been preferentially adsorbed by the nanoparticles. The latter effect was

more apparent in the Otonabee River water, likely because it had an initial raw water SUVA that

was two times that of the Lake Ontario water.

110

5.3.3 Effects of pH and TiO2 Dose on Adsorption

Bottle point isotherm tests were conducted to further characterize the adsorption behaviour of the

nanoparticles at the three pH conditions as the dose of TiO2 was varied from 0.01 g/L to 1 g/L.

As shown in Figure 5.2 and Figure 5.S.4 in the supplemental material at the end of the chapter, at

equilibrium, the DOC and UV254 removals observed in the Otonabee River and Lake Ontario

samples were found to be pH dependent and in all cases, more NOM was removed at pH 4 than

at pH 6 and pH 8. Irrespective of pH, increasing the dose of TiO2 added to the water resulted in a

decrease in the amount of DOC remaining in the treated water. In both the Otonabee River

(Figure 5.2A) and Lake Ontario water trials (Figure 5.2B), DOC removal increased quickly as

the TiO2 dose was increased from 0.01 g/L up to 0.25 g/L, but slowed thereafter, though no

definitive plateau was reached at any pH, suggesting that further increases in TiO2 dose beyond

the maximum applied in this study (1 g L-1) may have improved DOC removal even further. At

pH 4 and 1 g/L of TiO2 the DOC of the Otonabee River water was reduced from 4.69 ± 0.12

mg/L to 1.10 ± 0.12 mg/L whereas that of the Lake Ontario water was reduced from 1.64 ± 0.05

mg/Lto 0.69 ± 0.04 mg/L. DOC removal from both water sources was statistically significantly

lower at pH 6 and pH 8 compared to at pH 4.

111

Figure 5.2 Adsorption of DOC from raw unchlorinated water from Otonabee River

water (A) and Lake Ontario water (B) adjusted to pH 4, pH 6, and pH 8 and

mixed with 0.5 g/L of P25 TiO2 nanoparticles for four hours

0

1

2

3

4

5

6

0.0 0.2 0.4 0.6 0.8 1.0 1.2

DO

C (

mg /

L)

TiO2 Dose (g/L)

pH 4

pH 6

pH 8

A

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0

0 0.2 0.4 0.6 0.8 1

DO

C (

mg/L

)

TiO2 Dose (g/L)

pH 4

pH 6

pH 8

B

112

The UV254 results followed similar trends as DOC and are shown in Figure 5.4. UV254 was

removed more effectively than DOC, particularly from the Otonabee River water, where UV254

was reduced from 0.112 ± 0.002 cm-1 to 0.013 ± 0.002 cm-1, approximately 88%, by a 1 g/L dose

of P25 nanoparticles at pH 4. UV254 is a measure of the aromaticity of the NOM present in

water, so the results presented here suggest that aromatic NOM was preferentially removed over

non-aromatic NOM during the adsorption process.

As has been suggested by other researchers (Mwaanga et al., 2014) this pH dependence may be

partially explained through charge interactions. Zeta potential measurements conducted on the

P25 nanoparticles indicated that they had an isoelectic point (IEP) between pH 6 and pH 6.5,

consistent with the literature (Kosmulski, 2009). Thus, at pH 4 they were positively charged, at

pH 8 they were negatively charged, and at pH 6 they were electrostatically neutral. At pH 4 most

NOM compounds would have been neutral or slightly negatively charged and both charge and

hydrophobic interactions likely contributed to adsorption. At pH 6 hydrophobic interactions

between NOM compounds were likely the main contributors to adsorption. At pH 8, both the

nanoparticles and the NOM were negatively charged and thus charge interactions would result in

repulsion, rather than attraction, and any adsorption that took place would have been attributable

to hydrophobic interactions.

Nanoparticle agglomeration and its effect on surface area may also have contributed to the

changes in adsorption efficiency observed at different pHs. Agglomeration is most likely to

occur when the pH is near the isoelectric point/point of zero charge of the material in question

because at this pH repulsive forces between individual particles are at a minimum (Liu et al.,

2013). P25 has an IEP of approximately 6.5, which means that the nanoparticles were more

likely to agglomerate at pH 6 than at pH 4 or pH 8. Indeed, the particle size distributions

presented in Figure 5.3 indicate that the agglomerates formed by the nanoparticles were larger in

Otonabee River water adjusted to pH 6 than in Otonabee River water adjusted to pH 4 or pH 8.

In this study, better adsorption was observed at pH 4 than at pH 6. Agglomeration and the

subsequent reduction in available surface area at pH 6 versus at pH 4 may have contributed to

the poorer adsorption observed at pH 6 whereas charge repulsion between NOM and the TiO2

nanoparticles was more of a driver at pH 8.

113

Figure 5.3 Size distribution of agglomerates of P25 nanoparticles in Otonabee River

water adjusted to pH 4, pH 6, and pH 8

5.3.4 Modeling of Adsorption Isotherms

The DOC results of the bottle point isotherm tests were modeled using the linearized forms of

the Freundlich and modified Freundlich models, as shown in Figure 5.4 (modified Freundlich

model) and Figure 5.S.5 of the supplement (Freundlich model). The isotherm parameters are

summarized in Table 5.1 and Table 5.S.2 in the supplemental file.

At all pHs and in both water sources, the modified Freundlich model was a better fit to the data,

defined as a higher R2 value, than the Freundlich model. The modified Freundlich model is

generally thought to provide a more accurate fit for data from highly heterodisperse systems and

when the isotherms are developed using variable doses of adsorbent, so this result was not

surprising.

0

1

2

3

4

5

6

7

0 5 10 15 20

Per

cen

t

Diameter (m)

pH 4

pH 6

pH 8

114

Table 5.1 Summary of isotherm parameters for the adsorption of NOM from Otonabee

River water onto P25 TiO2 nanoparticles at pH 4, pH 6, and pH 8. Error

values represent the 95% confidence interval on the mean.

Parameter Otonabee River Lake Ontario

pH 4 pH 6 pH 8 pH 4 pH 6 pH 8

DOC

1/nFM 0.4 ± 0.0 0.5 ± 0.0 0.5 ± 0.1 0.6 ± 0.1 0.6 ± 0.1 0.6 ± 0.1

KFM (mg DOC/g TiO2)1-1/n 3.7 ± 0.4 1.3 ± 0.2 1.0 ± 0.2 1.5 ± 0.2 0.8 ± 0.1 0.5 ± 0.1

R2 0.98 0.97 0.94 0.98 0.95 0.90

THMfp

1/nFM 0.4 ± 0.1 0.4 ± 0.1 0.4 ± 0.2 -a - -

KFM (g THMfp/g TiO2)1-1/n 27 ± 17 13 ± 9 8 ± 14 - - -

R2 0.91 0.90 0.71 - - -

HAAfp

1/nFM 0.4 ± 0.2 0.5 ± 0.2 0.7 ± 0.3 - - -

KFM (g HAAfp/g TiO2)1-1/n 8 ± 7 4 ± 6 1 ± 2 - - -

R2 0.81 0.76 0.76 - - -

aPreliminary experiments indicated that TCM and BDCM removal from the LO water at 1 g/L of TiO2

and pH 4 was below the minimum detection limits (TCM = 5.5 g/L, BDCM = 2.9 g/L) and the

formation of DCAA and TCAA in the raw LO water when it was chlorinated according to UFC conditions

was near the minimum detection limit (DCAA = 1.4 g/L, TCAA =1.7 g/L) and unaffected by TiO2

adsorption treatment. As a result, DBPfp isotherms were not developed for the LO water.

Other researchers have observed that aromatic NOM and humic acids are preferentially adsorbed

to P25 nanoparticles over other types of NOM (Erhayem and Sohn, 2014). Given that the two

water sources differ mainly in terms of their NOM concentration and aromaticity, it is not

surprising that the KFM values obtained from the Lake Ontario tests were lower than those

obtained from the Otonabee River tests. The higher 1/nFM values in the Lake Ontario tests also

indicate that adsorption was less favourable in this water than in the Otonabee River water.

115

Figure 5.4 DOC data from Otonabee River water tests (A) and Lake Ontario water tests

(B) fitted to the modified Freundlich model

Within each water source, 1/nFM for DOC was nearly always constant irrespective of pH and KFM

was larger at pH 4 than at pH 6 and pH 8, which further confirms that the adsorption of NOM to

P25 TiO2 nanoparticles was more effective at pH 4 than at pH 6 or pH 8 in both water sources.

The pH 6 and pH 8 confidence intervals for KFM overlapped with one another in the OTB water

source, perhaps suggesting that pH became less of a driver of adsorption capacity when the pH

of the water was equal to or higher than the IEP.

1

10

100

1 10 100 1000

qe

(mg

DO

C/g

TiO

2)

Ce/D (mg DOC/g TiO2)

A

pH 4

pH 6

pH 8

0

1

10

100

0 1 10 100

qe

(mg D

OC

/gT

iO2)

Ce/D (mg DOC/g TiO2)

B

pH 4

pH 6

pH 8

116

These findings are in agreement with those of other researchers (Mwaanga et al., 2014; Erhayem

and Sohn, 2014; Sun and Lee, 2012), which have demonstrated that NOM adsorption to P25

TiO2 nanomaterials occurs more readily at low pH than at high pH. They are also in agreement

with studies that have demonstrated that NOM adsorption to TiO2 can be modeled using the

Freundlich (Wiszniowski et al., 2002) and modified Freundich models (Erhayem and Sohn,

2014). The KFM values reported in the latter study were higher but within the same order of

magnitude as those observed in the current study and the reported 1/nFM values were higher and

more strongly impacted by pH than those observed in the current study. It should be noted that

the Erhayem and Sohn made use of standardized NOM or humic acid isolates (e.g. IHSS) in

synthetic water matrices rather than natural water sources.

The KFM and 1/nFM results of this study and those of other TiO2 researchers are lower than those

achieved by other groups working with activated carbon and nanoscale carbon adsorbents, but

not dramatically so. In their original study, which was conducted with four NOM isolates and

GAC doses similar to the TiO2 doses used in this study, Summers and Roberts (1988) observed

KFM values ranging from 4.22 to 11.4 (mg C/g GAC)1-1/n and 1/nFM values ranging from 0.254 to

0.347. Karanfil et al. (1999) evaluated the adsorption of commercially available NOM isolates

and NOM from natural water onto a series of commercially available and modified activated

carbons. They observed KSR values ranging from 1.754 to 10.695 (mg C/g GAC)1-1/n in the

natural water matrices. Hyung and Kim calculated KFM values ranging from 5.471 to 13.088 (mg

C/g MWNT)1-1/n and 1/nFM values ranging from 0.212 to 0.384 when they evaluated the

adsorption of commercially available NOM isolates onto multi-walled carbon nanotubes.

5.3.5 Adsorption of DBP Precursors

Only a limited amount of THMs and HAAs were formed in the raw Lake Ontario water when it

was chlorinated according to the UFC method (< 50 g/L and < 40 g/L, respectively), so the

adsorption of DBP precursors to the P25 nanoparticles was not explored for this water source.

The raw water and treated samples prepared during the Otonabee River adsorption tests were

analyzed for THMfp and HAAfp and the results are shown in Figure 5.5.

As observed with DOC and UV254 removal, THMfp reduction via adsorption was pH

dependent. Maximum THMfp reduction, 147 ± 16 g/L to 38 ± 16 g/L (74% reduction), was

117

achieved at pH 4 and a TiO2 dose of 1 g/L. Less removal was achieved at this dose at pH 6 (153

± 11 g/L to 75 ± 11 g/L, 51% reduction) and pH 8 (154 ± 15 g/L to 104 ± 15 g/L, 34%

reduction).

Figure 5.5 THMfp (A) and HAAfp (B) of Otonabee River water treated with increasing

concentrations of P25 TiO2 nanoparticles at pH 4, pH 6, and pH 8

HAA precursors were also removed through adsorption and this removal was pH dependent,

though less so than for THM precursors. Figure 5.5B shows that at the highest concentration of

TiO2 (1 g/L) HAAfp was reduced from 39 ± 7 g/L to 19 ± 5 g/L (50% reduction) at pH 4 and

from 36 ± 2 g/L to 23 ± 3 g/L (40% reduction) at pH 6. HAAfp reduction at pH 8 was not

0

25

50

75

100

125

150

175

200

0 0.25 0.5 0.75 1

TH

Mfp

(

g/L

)

TiO2 Dose (g/L)

A

pH 4

pH 6

pH 8

0

10

20

30

40

50

0 0.25 0.5 0.75 1

HA

Afp

(

g/L

)

Dose TiO2 (g/L)

pH 4

pH 6

pH 8

B

118

statistically significant at the 95% confidence level. Some of the variability in the HAA results

can be explained by the fact that the UFC test, which was used to evaluate the THMfp and

HAAfp of the raw and TiO2-treated samples in this study, is conducted at pH 8, which does not

favour the formation of HAAs. As a result, all of the raw and TiO2-treated samples had low

HAAfp, making it difficult to isolate the effects of TiO2 adsorption on HAAfp removal,

particularly at pH 8.

The agreement between the THMfp and HAAfp datasets and the modified Freundlich model are

illustrated in Figure 5.S.6 and the isotherm parameters are summarized in Table 5.1. Although

the R2 values of the THMfp and HAAfp isotherms were lower than those of the DOC isotherms,

the general trends indicate that, with the exception of HAAfp at pH 8, TiO2 was able to remove

significant amounts of THM and HAA precursors from Otonabee River water via adsorption and

that this adsorption could be modeled using the modified Freundlich isotherm model.

Nonetheless, as a whole the isotherm parameters for the two classes of DBPs should be

approached with caution because they were developed using a small dataset that contained

substantial variation at low TiO2 doses. Additional experiments at higher TiO2 doses, using

water sources with higher concentrations of DBP precursors, and/or employing chlorination

regimes more likely to result in THM and HAA formation may help to clarify the how well the

SR model is able to predict the removal of DBP precursors from drinking water by TiO2 as well

as the suitability of the model at different pHs.

5.3.6 Effect of pH on Adsorption of LC-OCD Fractions

A selection of raw and TiO2-treated water samples from the Otonabee River experiment was

analyzed using LC-OCD to determine whether any specific fractions were being removed during

adsorption and whether pH impacted the fractions adsorbed. As shown in Figure 5.6, the

biopolymers and humic substances fractions were targeted for adsorption at all pHs but most

effectively removed at pH 4. The building blocks fraction was also removed to some degree at

pH 4 but was essentially unaffected at pH 6 and pH 8. The low molecular weight acid (LMWA)

and low molecular weight neutral fractions (LMWN) were not adsorbed at any pH. These results

indicate that, consistent with the findings of other researchers (Erhayem and Sohn, 2014), large

and aromatic NOM compounds were preferentially adsorbed by TiO2 nanoparticles and help to

119

explain why U254, which is associated with the humic substances fraction, was sometimes

removed more effectively than overall DOC.

Figure 5.6 LC-OCD fractions present in raw unchlorinated Otonabee River water and

water adjusted to pH 4, pH 6, and pH 8 and mixed with 0.5 g/L of P25 TiO2

nanoparticles for four hours

Conclusions

The results of this study show that during adsorption aromatic NOM (as measured by UV254) is

preferentially removed over non-aromatic NOM and that the efficiency of NOM adsorption to

TiO2 can vary by water source. They also demonstrate that TiO2 nanoparticles preferentially

adsorb larger NOM molecules including the biopolymers and humic substances fractions. pH

was shown to have a strong impact on the removal of NOM, including DBP precursors, from

surface water by TiO2 nanoparticles. Specifically, more adsorption occurred at low pH than at

higher pH. The poorer adsorption observed at pH 6 and pH 8 may be related to both

agglomeration and charge repulsion at higher pH, with the former dominating at pH 6 and the

latter at pH 8.

0

1

2

3

4

5

Raw Water pH 4 pH 6 pH 8

DO

C (

mg

/L) LMWN

LMWA

Building Blocks

Humic Substances

Biopolymers

120

A modified version of the Freundlich isotherm model was found to provide an excellent fit to the

DOC data gathered in this study. The resulting isotherm parameters were within but at the low

end of the range usually observed during NOM adsorption to GAC and carbon nanomaterials,

indicating that, particularly at neutral pH, the TiO2 nanoparticles were less effective than the

adsorbents currently used in drinking water plants. Unlike TiO2, however, GAC is generally

expensive and energy intensive to regenerate and the regeneration must usually be conducted

offsite, whereas TiO2 is potentially regenerable and reusable in place. The THMfp and HAAfp

datasets were also fitted to the modified Freundlich model, with generally positive results. The

results presented in this paper show that TiO2 adsorption is a viable way to remove NOM and

DBP precursors from drinking water and that this removal can be modeled using simple isotherm

models. The results also suggest that researchers hoping to design adsorption-based TiO2

processes should keep in mind that pH adjustment might be required to optimize performance.

Acknowledgements

The authors would like to acknowledge the training provided by Jim Wang, the laboratory

assistance provided by Yijun (Jessie) Gai and Michelli Park, and the support of the Drinking

Water Research Group at the University of Toronto. The authors are also grateful to Dr. Monica

Tudorancea and Dr. Sigrid Peldzsus (University of Waterloo) for performing LC-OCD analyses.

Funding was provided by the National Science and Engineering Research Council of Canada and

the Ontario Ministry of Training, Colleges, and Universities.

121

References

American Public Health Association, 2005. Standard Methods for the Examination of Water and


Avisar, D., Horovitz, I., Lozzi, L., Ruggieri, F., Baker, M., Abel, M-L, Mamane, H., 2013.

Impact of water quality on removal of carbamazepine in natural waters by N-doped TiO2 photo-

catalytic thin film surfaces, Journal of Hazardous Materials, 244-245, 463-471

Chowdhury, Z.K., Summers, R.S., Westerhoff, G.P., Leto, B.J., Nowack, K.O., Corwin, C.J.,

2013. Activated Carbon: Solutions for Improving Water Quality, Passantino, L.B. (Ed.), Denver,

USA, American Water Works Association

Erhayem, M., 2013. Effect of naturally occurring organic matter (NOOM) type and source on



Erhayem, M. and Sohn, M., 2014. Stability studies for titanium dioxide nanoparticles upon



Fotiou, T., Triantis, T.M., Kaloudis, T., Hiskia, A., 2015. Evaluation of the photocatalytic

activity of TiO2 based catalysts for the degradation and mineralization of cyanobacterial toxins

and water off-odor compounds under UV-A, solar, and visible light, Chemical Engineering

Journal, 261, 17-26

Huang, X., Leal, M., Li, Q., 2008. Degradation of natural organic matter by TiO2 photocatalytic

oxidation and its effect on fouling of low-pressure membranes, Water Research, 42, 1142-1150

Huber, S.A., Balz, A., Abert, M., Pronk, W., 2011. Characterisation of aquatic humic and non-

humic matter with size-exclusion chromatography-organic carbon detection and organic nitrogen

detection (LC-OCD-OND), Water Research, 45, 879-888

122

Hyung, H., Kim, J-H, 2008. Natural organic matter (NOM) adsorption to multi-walled carbon

nanotubes: Effect of NOM characteristics and water quality parameters, Environmental Science

and Technology, 42, 4416-4421

International Organization for Standardization, 2010. Nanotechnologies – Methodology for the

classification and categorization of nanomaterials, ISO/TR 11360:2010(E)

Kanakaraju, D., Glass, B.D., Oelgemoller, M., 2014. Titanium dioxide photocatalysis for

pharmaceutical wastewater treatment, Environmental Chemistry Letters, 12, 27-47

Karanfil, T., Kitis, M., Kilduff, J.E., Wigton, A., 1999. Role of granular activated carbon surface

chemistry on the adsorption of organic compounds 2, Environmental Science and Technology,

33, 3225-3233

Li, F., Yuasa, A., Ebie, K., Azuma, Y., Hagishita, T., Matsui, Y., 2002. Factors affecting the

adsorption capacity of dissolved organic matter onto activated carbon: Modified isotherm


Liu, S., Lim, M., Fabris, R., Chow, C., Drikas, M., Amal, R., 2008. TiO2 photocatalysis of



Liu, S., Lim, M., and Amal, R. 2014. TiO2-coated natural zeolite: Rapid humic acid adsorption

and effective photocatalytic regeneration, Chemical Engineering Science, 105, 46-52

Liu, W., Sun, W., Borthwick, A., and Ni, J., 2013. Comparison on aggregation and



Engineering Aspects, 434, 319-328

Loosli, F., Vitorazi, L., Berret, J-F, and Stoll, S., 2015. Towards a better understanding on


natural organic matter, Water Research, 80, 139-148

123

Kim, S-H and Shon, H.K., 2007. Adsorption characterization for multi-component organic

matters by titanium oxide (TiO2) in wastewater, Separation Science and Technology, 42, 1775-

1792

Kosmulski, M., 2009. Compilation of PZC and IEP of sparingly soluble metal oxides and

hydroxides from literature, Advances in Colloid and Interface Science, 152, 14-25

Malato, S., Fernandez-Ibanez, P., Maldonado, M.I., Blanco, J., Gernjak, W., 2009.



Mwaanga, P., Carraway, E.R., Schlautman, M.A., 2014. Preferential sorption of some natural

organic matter fractions to titanium dioxide nanoparticles: influence of pH and ionic strength,

Environmental Monitoring and Assessment, 186, 8833-8844

Nanotechnology Characterization Laboratory, 2009. NCL Method PCC-2: Measuring Zeta

Potential of Nanomaterials, National Cancer Institute, U.S. National Institutes of Health

Ng, M., Kho, E.T., Liu, S., Lim, M., Amal, R., 2014. Highly adsorptive and regenerative


Journal, 246, 196-203

Philippe, K.K. Hans, C. MacAdam, J., Jefferson, B., Hart, J., and Parsons, S.A., 2010.

Photocatalytic oxidation, GAC, and biotreatment combinations: An alternative for the

coagulation of hydrophilic rich waters?, Environmental Technology, 31, 1423-1434

Qi, S. Schideman, L.C., 2008. An overall isotherm for activated carbon adsorption of dissolved

organic matter in water, Water Research, 42, 3353-3360

Shahbeig, H., Bagheri, N., Ghorbanian, S., Hallajisani, A., Poorkarimi, S., 2013. A new


Modelling and Simulation, 9, 243-254

Shi, H., Magaye, R., Castranova, V., and Zhao, J., 2013. Titanium dioxide nanoparticles: A

review of current toxicological data, Particle and Fibre Toxicology, 10:15

124

Summers, R. and Roberts, P., 1988. Activated Carbon Adsorption of Humic Substances:

Heterodisperse Mixtures and Desorption, Journal of Colloid and Interface Science, 122, 367-381

Summers, R.S., Hooper, S.M., Shukairy, H.M., Solarik, G., Owen, D., 1996. Assessing DBP

yield: Uniform formation conditions, Journal of the American Water Works Association, 88, 80-

93

Sun, D.D. and Lee, P.F., 2012. TiO2 microsphere for the removal of humic acid from water:

Complex adsorption mechanisms, Separation and Purification Technology, 91, 30-37

Wassink, J.D., Andrews, R.C., Peiris, R.H., Legge, R.L., 2011. Evaluation of fluorescence


by enhanced coagulation, Water Science and Technology: Water Supply, 11, 621

Wiszniowski, J., Robert, D., Surmacz-Gorska, J., Miksch, K., and Weber, J-V, 2002.

Photocatalytic decomposition of humic acids on TiO2, Part I: Discussion of adsorption and

mechanism, Journal of Photochemistry and Photobiology A: Chemistry, 153, 267-273

Wray, H.E., Andrews, R.C., Bérubé, P.R., 2013. Surface shear stress and membrane fouling

when considering natural water matrices, Desalination, 330, 22-27

Yang, Y. and Westerhoff, P., 2014. Presence in, and Release of, Nanomaterials from Consumer

Products, Nanomaterials, Advances in Experimental Medicine and Biology, 811 (Capco, D.G.

and Chen, Y., editors), Springer Science + Business Media, Berlin

Zhang, Y., Chen, Y., Westerhoff, P., and Crittenden, J., 2009. Impact of natural organic matter

and divalent cations on the stability of aqueous nanoparticles, Water Research, 43, 4249-4257

125

Supplementary Material for Chapter 5

Table 5.S.1 Raw water quality

Parameter Units Otonabee River Lake Ontario

DOC mg L-1 3.8 – 4.9 1.6 – 2.0

UV254 cm-1 0.09 – 0.14 0.02 – 0.03

SUVA L mg-1 m-1 2.0 – 2.4 0.8 – 1.0

pH 7.8 – 8.4 7.8 – 8.0

Alkalinity mg CaCO3 L-1 85 – 91 90 – 93

Table 5.S.2 Freundlich isotherm parameters for DOC

Parameter Otonabee River Lake Ontario

pH 4 pH 6 pH 8 pH 4 pH 6 pH 8

1/nF 1.3 ± 0.2 3.5 ± 0.8 5.3 ± 0.9 3.3 ± 0.5 5.1 ± 1.1 7.1 ± 2.4

KF (mg DOC/g TiO2)1/n 3.1 ± 0.5 0.1 ± 0.1 0.1 ± 0.8 5.2 ± 0.8 1.1 ± 0.3 0.3 ± 0.2

R2 0.96 0.85 0.91 0.91 0.85 0.72

126

Time Series Experiments

Figure 5.S.1 DOC of Otonabee River water (A) and Lake Ontario water (B) dosed with

0.5 g/L of P25 TiO2 nanoparticles and allowed to mix in the dark for between

0 and 480 minutes

0

1

2

3

4

5

0 100 200 300 400 500

DO

C

(mg

/L)

Time (min)

pH 4

pH 6

pH 8

A

0

0.5

1

1.5

2

2.5

0 100 200 300 400 500

DO

C (

mg/L

)

Time (min)

pH 4

pH 6

pH 8

B

A

127

Figure 5.S.2 UV254 of Otonabee River water (A) and Lake Ontario water (B) dosed with


0 and 480 minutes

0

0.02

0.04

0.06

0.08

0.1

0.12

0 100 200 300 400 500

UV

25

4 (

cm-1

)

Time (min)

pH 4

pH 6

pH 8

0

0.01

0.02

0.03

0.04

0 100 200 300 400 500

UV

254 (

cm-1

)

Time (min)

pH 4

pH 6

pH 8

B

A

128

Figure 5.S.3 SUVA of Otonabee River water (A) and Lake Ontario water (B) dosed with


0 and 480 minutes

0

0.5

1

1.5

2

2.5

3

3.5

4

0 100 200 300 400 500

SU

VA

(L

/mg

. m

)

Time (min)

pH 4

pH 6

pH 8

0

1

2

3

0 100 200 300 400 500

SU

VA

(L

/mg

.m

)

Time (min)

pH 4

pH 6

pH 8

A

B

129

UV254 Removal

Figure S.5.4 Adsorption of UV254 from raw unchlorinated water from Otonabee River

water (A) and Lake Ontario water (B) adjusted to pH 4, pH 6, and pH 8 and

mixed with 0.5 g/L of P25 TiO2 nanoparticles for four hours

0.00

0.02

0.04

0.06

0.08

0.10

0.12

0.14

0.0 0.2 0.4 0.6 0.8 1.0 1.2

UV

25

4 (

cm-1

)

TiO2 Dose (g/L)

pH 4

pH 6

pH 8

0.000

0.005

0.010

0.015

0.020

0.025

0 0.2 0.4 0.6 0.8 1

UV

254 (

cm-1

)

TiO2 Dose (g/L)

pH 4

pH 6

pH 8

B

A

130

Freundlich Isotherm Graphs

Fig. 5.S.5 DOC data from Otonabee River water tests (A) and Lake Ontario water tests

(B) fitted to the Freundlich isotherm model

1

10

100

1 10

qe

(mg

DO

C/g

TiO

2)

Ce (mg DOC/L)

A

pH 4

pH 6

pH 8

0.2

2

20

0.2 2

qe

(mg D

OC

/g T

iO2)

Ce (mg DOC/L)

B

pH 4

pH 6

pH 8

131

Linearized Modified Freundlich Model Isotherms for THMfp and HAAfp

Figure S.5.6 THMfp (A) and HAAfp (B) data from Otonabee River water tests fitted to

the linearized modified Freundlich model

10

100

1000

10000

10 100 1000 10000 100000

qe

(g

TH

Mfp

/g T

iO2)

Ce/D (g THMfp/g TiO2)

A

pH 4

Model - pH 4

pH 6

Model - pH 6

pH 8

Model - pH 8

1

10

100

1000

1 10 100 1000 10000

qe

(g H

AA

fp/g

TiO

2)

Ce/D (g HAAfp/g TiO2)

B

pH 4

Model - pH 4

pH 6

Model - pH 6

pH 8

Model - pH 8

132

Development of Settleable Engineered Titanium Dioxide Nanomaterials for the Safe Removal of Disinfection Byproduct Precursors from Drinking Water

Abstract

Four types of linear engineered TiO2 nanomaterials (LENs) were synthesized using a simple

alkaline hydrothermal method and evaluated in terms of their propensity to settle out of water

and their ability to remove natural organic matter (NOM) from river water as measured by two

common disinfection byproduct (DBP) precursor surrogates, DOC and UV254, via adsorption in

the absence of irradiation (dark adsorption) and photocatalytic degradation under UVA LED

irradiation. The size, surface characteristics, and crystallinity of the LENs were manipulated by

varying the temperatures used during the hydrothermal and calcination steps of the synthesis

procedure. All four types of LENs settled out of purified water more effectively than standard

Degussa Evonik Aeroxide P25 nanoparticles but their settling behaviour in the river water

sample was impacted by surface charge effects and interactions with NOM and ionic species

present in the bulk water matrix. The total observed reduction of DBP precursor surrogates by

the LENs ranged from 20 to 50% removal of DOC and from 65 to 90% reduction of UV254 after

60 minutes of irradiation. The electrical energy per order required to remove DOC and UV254

from the water was calculated and found to range from 8 to 36 times higher than that required for

UV/H2O2 treatment but comparable to results reported by other researchers using UV/TiO2 for

NOM removal. The results of this study suggest that a subset of the nanomaterials evaluated in

this study may prove to be a viable alternative to standard TiO2 nanoparticles for the removal of

DBP precursors from drinking water, but also that the characteristics of the water matrix have

important effects on settling efficiency and will require site-specific evaluation.

Introduction

Oxidation processes such as ozonation have become a mainstay of modern drinking water

treatment because they can degrade contaminants that are hazardous and/or recalcitrant to

removal via more traditional methods. In recent decades various advanced oxidation processes

(AOPs) that combine two or more existing treatment technologies to enhance the removal of

133

these recalcitrant compounds have become more popular. In the laboratory, TiO2 photocatalysis

has been used to destroy numerous water contaminants including various types of bacteria

(Rincon and Pulgarin, 2004), taste and odour compounds (Fotiou, 2015), natural organic matter

(Liu et al., 2008; Philippe et al., 2010), and various anthropogenic contaminants (Avisar et al.,

2013, Kanakaraju, 2014).

TiO2 photocatalysis combines a nanoscale semiconductor photocatalyst with UVA light and is an

emerging AOP that may one day prove to be a useful addition to the existing suite of oxidation

processes. When irradiated with UV light at or below 385 nm TiO2 catalyzes the formation of

reactive oxygen species (ROS). The ROS are highly oxidative, and will degrade organic

contaminants adsorbed to the surface of the TiO2 nanoparticle. Adsorbed organic contaminants

can also be oxidized by the electron hole formed when the photocatalyst is activated (Nosaka and

Nosaka, 2013).

Despite the promise that TiO2 holds for drinking water and wastewater treatment, it has yet to be

widely adopted for these purposes, mostly because it has proven difficult to design a reactor that

is simultaneously capable of ensuring adequate treatment efficiency while also working within

the practical confines of a water treatment plant. The need to remove the TiO2 from the water

after treatment is also a major concern because TiO2 nanomaterials are themselves potentially

hazardous to human health (Shi et al., 2013) and the environment (Yang and Westerhoff, 2014).

Some researchers have attempted to address this by immobilizing TiO2 on solid supports.

Examples include nanoparticles attached to or integrated into magnetized particles Leshuk et al.,

2013; Ng et al., 2014), glass beads (Kim and Lee, 2005; Daneshvar et al., 2005), and zeolites

(Liu et al., 2014). Alternatively, other researchers have engineered multidimensional TiO2

materials for water remediation. For example, Xing et al. (2014) engineered a floating

macro/mesoporous TiO2 material and demonstrated its ability to degrade two simple organic

compounds while Hu et al. (2011) developed a freestanding enmeshed TiO2 nanowire membrane

for the removal of pharmaceutical compounds. In this study, we have investigated the use of

settleable LENs synthesized from the most commonly available form of laboratory grade TiO2,

P25 nanoparticles from Evonik Degussa, for the treatment of drinking water.

The alkaline hydrothermal method used to synthesize these materials was first described by

Kasuga et al. (1999) and is by now so widely known that it can be found in the public domain. A

134

precursor compound (usually P25 nanoparticles or anatase nanoparticles) is suspended in

alkaline solution and heated above 100oC for a period of time ranging from 20 hours to 4 days.

The resulting material, which consists of sodium disodium trititanate (Na2Ti3O7), is then washed

with acid and water to remove the Na+ ions and then calcined at temperatures ranging from

300oC to 900oC to yield a final product consisting of H2Ti3O7, TiO2 (anatase or rutile), or other

crystalline structures depending on the temperature used. The linear materials formed at the end

of this process are tubular or belt like with diameters in the nanoscale range and lengths in the

nanoscale or microscale range.

Although many researchers have employed some version of this process to yield linear

nanomaterials, their results are difficult to compare to one another because the studies have

generally employed different synthesis regimes and there remains some debate as to the

individual impacts of the many steps of the procedure on the final product (Wong et al., 2011).

Yuan and Su (2004) synthesized linear nanomaterials with varying precursor materials,

hydrothermal temperatures, and different types and concentration of alkaline solution and found

that all three factors impacted the size, shape, and reactivity of the end products. For example,

higher hydrothermal temperatures resulted in the formation of large, flat materials (nanoribbons

or nanobelts) while lower temperatures yielded smaller nanotubes. Qamar et al. (2008) focused

more specifically on the washing and calcination procedures following hydrothermal synthesis

and found that both steps impacted the chemical structure, shape, morphology, and

photocatalytic activity of the final products. In the present study, we employed a set of synthesis

regimes informed by the findings of Yuan and Su (2004) and Qamar et al. (2008) that we

predicted would enhance the photoactivity and settleability of the nanomaterials.

Natural organic matter (NOM) is a blanket term that encompasses an array of organic carbon

compounds that are formed through the degradation of organisms and their detritus. It is

commonly quantified as total organic carbon (TOC), dissolved organic carbon (DOC), or based

on its ability to absorb UV light at various wavelengths, including 254 nm (UV254). NOM has a

number of undesirable aesthetic, operational, and health effects on drinking water, and its

removal is one of the primary goals in many water treatment plants. Most importantly, the

interaction of NOM with chemical treatment processes can result in the formation of undesirable

reaction products, including disinfection byproducts (DBPs). Certain classes of DBPs, usually

trihalomethanes (THMs) and haloacetic acids (HAAs), are regulated in most jurisdictions in

135

North America. These two classes of DBPs were originally thought to be carcinogenic but more

recent research has revealed that negative health effects are unlikely to occur in humans exposed

to the concentrations of THMs and HAAs commonly found in drinking water (Hrudey, 2009).

Nonetheless, they continue to be regulated, partly because they represent only a small subset of

the DBPs formed when NOM reacts with chemical disinfectants such as chlorine and

monochloramine. Almost all of these are currently unregulated but there is evidence that some

may be more toxic than the DBPs that are currently regulated (Krasner, 2009). For example, the

formation of halogenated furanones such as Mutagen X (MX), a highly genotoxic DBP, has been

shown to be correlated to the formation of HAAs (Zheng et al., 2015). DOC and UV254 are

widely used as surrogate parameters for DBP precursors because they are simple to measure and

well correlated with THM and HAA (van Leeuwen et al., 2005; Pifer and Fairey, 2014]. They

have also been shown to be correlated to MX formation in some water sources (Zheng et al.,

2015).

NOM, including the precursors of regulated and unregulated DBPs, can be removed using

existing water treatment processes such as coagulation but there is demand for alternative NOM

removal processes that are less chemically intensive and produce less waste. The aim of this

research was to evaluate the impact of different steps of the hydrothermal synthesis method on

the settleability of potentially reuseable photocatalytic TiO2 nanomaterials and their ability to

adsorb and photocatalytically degrade NOM and DBP precursors as quantified using DOC and

UV254.

Dyes, which are inexpensive, widely available, and simple to quantify, are often used to track the

progress of oxidation processes, including TiO2 photocatalysis. Oxidation of methylene blue dye

during photocatalysis breaks the compound’s conjugated bonds, resulting in a colour change

from blue to colourless, and as such, methylene blue degradation is widely used to evaluate new

photocatalytic materials. In this study, methylene blue dye was used to assess the overall

photocatalytic activity of four LENs and provided our group with a simple way to compare our

materials to those developed by other researchers.

136

Experimental

6.2.1 Materials

Aeroxide P25 TiO2 nanoparticles were obtained from Sigma Aldrich and used as the precursor

material for the four LENs. They were also used in unmodified form as a reference material

during the degradation experiments and the settling tests. Raw water was obtained from the inlet

of the Peterborough water treatment plant in Ontario, Canada, which treats water from the

Otonabee River. The Otonabee River is typical of many smaller surface water sources in

southern Ontario in that it has moderate levels of NOM and alkalinity, low turbidity, and pH

above neutral. Table 6.1 contains a summary of relevant water quality parameters in the raw

water.

Table 6.1 Summary of raw water quality

Parameter Units Value

DOC mg/L 4.6 ± 0.31

UV254 1/cm 0.12 ± 0.011

SUVA m/mg.L 2.7 ± 0.21

pH 8.2 ± 0.22

Turbidity NTU 0.6 ± 0.22

Alkalinity mg/L as CaCO3 87 ± 72

Hardness mg/L as CaCO3 95 ± 112

Calcium mg/L 32.8 ± 3.72

Magnesium mg/L 3.2 ± 0.32

Sodium mg/L 6.5 ± 0.82

Chloride mg/L 11.5 ± 1.32

Conductivity S/cm 214 ± 192

1Average and standard deviation of samples analyzed in DWRG laboratory

2Average and standard deviation of values obtained from Ontario Drinking Water Surveillance Program

2010-2012

137

6.2.2 Synthesis of Nanostructured Materials

Four LENs were synthesized via a basic hydrothermal method as described by numerous

researchers (Kasuga et al., 1999; Qamar et al., 2008; Yuan and Su, 2004; Liu et al., 2013] to

create tubular or belt-like LENs. Two grams of P25 nanoparticles from Evonik were added to a

Teflon-lined reactor along with 60 mL of a 10 N NaOH and mixed vigorously using a glass rod.

The Teflon-lined reactor was placed in a muffle furnace and heated to 130oC or 240oC. These

temperatures represent the lower and upper limits used by other researchers in the studies that

were reviewed early in this project (e.g. Liu et al., 2013; Seo et al., 2009). Additional

considerations included the fact that Yuan and Su (2004) reported low yields at temperatures

below 150oC, so it seemed prudent to operate at temperatures above this, and the oven originally

used for the hydrothermal step of the synthesis process had an upper limit of 250oC. The

temperature was maintained at this set point for approximately 24 hours and then the muffle

furnace was turned off and the furnace and autoclave were allowed to cool for an additional 24

hours. The contents of the autoclave were washed with 1.2 L of MilliQ water and then placed

into a sonicated acid bath (0.1 N HCl) for one hour. After acidification, the materials were

washed with MilliQ water until they reached the natural pH of the MilliQ water (5.5 to 6). The

washed nanomaterials were dried in an oven at 70oC for 12 hours and then calcined in a muffle

furnace for four hours at 550oC or 700oC. These two temperatures were chosen because studies

conducted by others suggested that they would result in the formation of dissimilar TiO2

polymorphs, specifically TiO2(B) at 550oC and anatase at 700oC (Zheng et al., 2009). The

synthesis conditions of the four nanomaterials used in this study are summarized in Table 6.2.

The temperatures used in the hydrothermal synthesis step and the calcination step were chosen

based on the work of Yuan and Su (2004), who observed that at a set reaction time and NaOH

concentration (10 M) the hydrothermal temperature determined the size and shape of the

products while the calcination temperature determined the crystal structure of the material. They

observed that hydrothermal temperatures ranging from 100oC to 180oC resulted in the formation

of cylindrical nanotubes while hydrothermal temperatures ranging from 180oC to 250oC yielded

flatter nanoribbons/nanobelts. They also observed that materials calcined at 540oC consisted

primarily of TiO2(B), a metastable form of TiO2 while those calcined at temperatures above

700oC for a short time consisted mainly of anatase.

138

Multiple batches of each nanomaterial were synthesized throughout this study. Before being used

for experiments, each batch was evaluated for consistency based on its ability to degrade

methylene blue dye. Quadruplicate samples containing 50 mL of 0.03 mM methylene blue

solution dosed with 0.1 g/L of TiO2 were stirred and exposed to UVA light (365 nm) with an

average irradiance of 4.9 mW/cm2 for 30 minutes (average UV dose of 11.25 J/cm2). The

average results of these simple quality control tests are also presented in Table 6.2.

Table 6.2 Summary of nanomaterial synthesis conditions and percent degradation of

methylene blue dye during quality control tests

Nanomaterial Hydrothermal

Temperature (TH)

Calcination

Temperature (TC)

Average Degradation

of Methylene Blue

NB 130/550 130 oC 550 oC 37 ± 6%

NB 130/700 130 oC 700 oC 54 ± 1 %

NB 240/550 240 oC 550 oC 41 ± 5%

NB 240/700 240 oC 700 oC 83 ± 1%

6.2.3 Characterization of Nanomaterials

The lab synthesized TiO2 nanomaterials were characterized using transmission electron

microscopy (TEM), selected area electron diffraction (SAED), available surface area (Brunauer–

Emmett–Teller adsorption method), isoelectric point (IEP), and agglomerate size in MilliQ

purified water and pre-filtered Otonabee River water.

TEM and SAED observation was conducted using a JEOL 2010F TEM/STEM at the Canadian

Centre for Electron Microscopy (Hamilton, Ontario, Canada). TEM samples were prepared by

drop casting the dispersions onto holey carbon grids. The images were processed using Gatan

Microscopy Suite: Digial MicrographTM and SAED and FFT images were indexed using

CrysTBox – diffractGUI (Klinger and Jäger, 2015). N2 adsorption isotherms were measured with

a Quantachrome AUTOSORB-1. The samples were outgassed at 200oC under vacuum for 12 h

before the measurement. Surface area was determined by BET method in a relative pressure

range of 0.05 to 0.25.

139

The IEP of each LEN was determined by measuring its zeta potential at pH values ranging from

3 to 9. Zeta potential was measured using a Horiba Zeta Analyzer and all zeta potential

experiments were conducted with 0.1 g/L of TiO2 in MilliQ water buffered with 10 mM NaCl

adjusted to various pH values using 0.1 M NaOH or HCl. Two aliquots were analyzed from each

sample.

A Malvern MasterSizer 3000 was used to evaluate the size distribution of the nanomaterial

particles when prepared in MilliQ purified water and Otonabee River water. The latter was

filtered through a 0.45 m filter ahead of TiO2 addition and measurement to avoid interference

by natural particulate matter in the raw water. The background of each water matrix was

evaluated before nanomaterial addition. Sufficient TiO2 was then added to the water to achieve a

15% obscuration value. For all but one material, this obscuration value occurred at a TiO2

concentration of approximately 0.03 to 0.06 g/L. NB 130/700 could not be analyzed under the

same conditions as the other materials because its optical properties made it impossible to

achieve the required obscuration at a concentration comparable to those used for the other

materials. As a result, the size distribution data collected for NB 130/700 has not been included

in this paper. Each sample was measured 10 times and the results were averaged and graphed as

a volume distribution.

6.2.4 Settling Tests

Calibration curves were prepared for each material to relate turbidity to the concentration of TiO2

in the water. For each material, aliquots of a 10 g/L TiO2 stock solution were dispensed into an

appropriate amount of MilliQ water to create duplicate calibration standards (0.01 g/L, 0.05 g/L,

0.1 g/L, 0.2 g/L, and 0.3 g/L). Each calibration standard was sonicated for 5 minutes and then

analyzed for turbidity on a Hach 2100 N turbidimeter operating in NTU mode with ratio on,

which allowed the turbidimeter to measure in the 0 to 4000 NTU range. The resulting turbidity

results were graphed against concentration to develop a calibration curve for each material. The

relationship between turbidity and concentration was linear within the range studied for all five

materials. For the settling tests, samples with a starting concentration of 0.25 g/L of TiO2 were

dispensed into the turbidimeter cuvette, sonicated for 5 minutes, and placed in the turbidimeter

for a total of 3 hours. The turbidity at midpoint of the cuvette was recorded at the beginning of

the test and at ten minute intervals thereafter.

140

6.2.5 Formation of ·OH Radicals

The formation of hydroxyl radicals by each of the nanomaterials was confirmed and quantified

using the same experimental apparatus described in Section 2.6 and a method adapted from

Arlos et al. (2016). Distilled water was spiked with terephthalic acid (TPA), a probe compound

that yields a fluorescent product (2-hydroxyterephthalic acid, HTPA) upon reaction with

hydroxyl radicals. 50 mL samples containing 0.5 mM TPA dissolved in 6 mM NaOH were dosed

with 0.02 g/L of TiO2 and irradiated with UVA LED light for times ranging from 30 seconds to

15 minutes. The fluorescence of the samples was measured at an excitation wavelength of 315

nm and an emission wavelength of 425 nm. Dark controls (no irradiation) and light only (no

TiO2) controls were also prepared. In all cases no hydroxyl radical formation was observed in the

control samples.

6.2.6 Characterization of NOM

Raw and treated water samples were filtered through a 0.45 m polyethersulfone (PES)

laboratory filter before analysis. Natural organic matter was quantified as dissolved organic

carbon (DOC) or based on UV absorbance at 254 nm (UV254). DOC was measured on an O/I

Analytical Aurora 1030 TOC analyzer and UV254 was measured using an Agilent 8453 UV-Vis

spectrophotometer.

6.2.7 Adsorption and Photocatalytic Degradation Under UVA Light

All of the degradation experiments were conducted in quadruplicate at 0, 5, 15, 30, 45, and 60

minutes of irradiation. Samples were prepared in 75 mL batch reactors filled with 50 mL of 0.03

mM methylene blue dye or unchlorinated raw river water obtained from a water treatment plant

in Peterborough, Ontario, Canada. The batch reactors were mixed using magnetic stir bars and

irradiated with UV LEDs (LZ1 UV 365 nm Gen2 Emitter, LED Engin Inc.) with a maximum

irradiance at 365 nm and an average irradiance of 4.9 mW/cm2 at the surface of the sample. The

irradiance of each lamp was confirmed before each test using a radiometer (International Light)

optimized to measure light at 365 nm. The methylene blue degradation experiments were

conducted with a TiO2 dose of 0.1 g/L while the NOM degradation experiments were conducted

with a TiO2 dose of 0.25 g/L to ensure sufficient NOM degradation for subsequent modeling.

Samples were analyzed for DOC and UV light absorbance at 254 nm. The results were evaluated

141

against a pseudo-first-order model for photocatalytic degradation and also normalized to

nanomaterial surface area.

6.2.8 Electrical Energy per Order Calculations

The electrical energy per order (EEO) concept is currently listed as a “figure of merit” for the

evaluation of advanced oxidation processes by IUPAC. Collins and Bolton (2016) define EEO

as:

“…the electrical energy in kilowatt hours (kWh) required to bring about the degradation

of a contaminant C by one order of magnitude in 1 m3 of contaminated water or air.”

The EEO of a given process can be calculated using Equation 6.1, where P is the power

dissipated by the treatment process (kW), V is the volume of water treated in the experiment (L),

Ci is the original concentration of the contaminant, Cf is the final concentration of the

contaminant, and t is the time required to achieve Cf (min).

𝐸𝐸𝑂 =1000 𝑃 𝑡

𝑉 log (𝐶𝑖𝐶𝑓)

(6.1)

For batch experiments, the EEO should be calculated from the electrical energy dose (EED),

which is the electrical energy consumed per unit volume and can be calculated as follows:

𝐸𝐸𝐷 =1000𝑃𝑡

60𝑉 (6.2)

𝐸𝐸𝑂 =𝐸𝐸𝐷

log (𝐶𝑖𝐶𝑓)

(6.3)

In this study, the EEO values calculated using Equation 6.1 were equal to that calculated using

equations 6.2 and 6.3.

142


6.3.1 Nanomaterial Characterization

A summary of the physical and surface characteristics of the five nanomaterials employed in this

study is provided in Table 6.3. As described in the subsequent subsections, the nanomaterials

synthesized at 240oC were larger and had fewer surface defects than those synthesized at 130oC

while the linear nanomaterials calcined at 550oC contained both anatase and had higher IEPs and

surface area than those calcined at 700oC. The nanomaterials calcined at 700oC were more

photocatalytically active and produced more hydroxyl radicals than those calcined at 550oC.

6.3.1.1 TEM

TEM images of the four LENs revealed differences in the shape and size of the nanomaterials

formed under the different synthesis conditions (Figure 6.1). As has been observed by others

(Yuan and Su., 2004), the nanomaterials formed at the lower hydrothermal temperature (130oC)

were smaller in both length and width than those formed when the hydrothermal temperature was

set at 240oC. The materials calcined at 550oC had rough, irregular surfaces while those calcined

at 700oC, irrespective of their geometry, appeared to have smooth but segmented surfaces. The

former were similar to materials prepared by Zheng et al. (2010), which were synthesized using a

similar, though not identical method, and calcined at temperatures ranging from 550oC to 650oC.

The authors attributed the irregular surfaces of their materials to the presence of TiO2(B).

Although the three dimensional shapes of the four nanomaterials were difficult to determine

from the images, NB 240/550 appeared to be flat or belt-like, matching the description of

materials synthesized by Yuan and Su (2004) under similar conditions. Individual particles of

NB 240/700 also appeared flat, but were irregularly shaped and occasionally segmented, perhaps

indicating that particles that were originally rectangular in shape were in the process of being

broken down during the high temperature calcination process.

143

Table 6.3 Characteristics of P25 and four linear engineered nanomaterials

Nanomaterial Crystal Phase(s)

BET Surface

Area

(m2/g)

IEP

·OH Production Rate

Constant

(M/min)

Normalized ·OH Production

Rate Constant

(M/min/m2)

R2

P25 Anatase, Rutile 57 6.0 - 6.1 0.620 ± 0.029 0.870 ± 0.040 0.99

NB 130/550 Anatase, TiO2(B) 99 6 .0 - 6.1 0.127 ± 0.011 0.102 ± 0.009 0.97

NB 130/700 Anatase 30 4.2 0.312 ± 0.154 0.832 ± 0.412 0.97

NB 240/550 Anatase, TiO2(B) 55 6.5 0.104 ± 0.009 0.151 ± 0.013 0.97

NB 240/700 Anatase 19 5.0 0.739 ± 0.058 3.110 ± 0.242 0.98

144

Figure 6.1 TEM images of A: NB 130/550, B: NB 130/700, C: NB 240/550, and

D: NB 240/700

The images of NB 130/550 and NB 130/700 revealed that both samples included a large amount

of smaller nanoparticulate matter in addition to larger linear structures similar to those found in

the samples synthesized at 240oC. As will be described later, the characteristics and behaviour of

these two materials differed in many notable ways from one another and from the P25 precursor

material, suggesting that they may represent nanoparticles that have been modified to some

extent by the synthesis process. Alternatively, they may also or instead be the remnants of larger

tubular or belt-like structures that were broken down into smaller nanoparticles during the

calcination step of the process. A similar phenomenon was observed by Qamar et al. (2008) with

nanotubes synthesized at 150oC.

6.3.1.2 SAED and HRTEM

SAED and HRTEM images are shown in Figure 6.2. In all cases, crystal phase differed based on

the temperature used in the calcination step. The crystalline structure of NB 130/550 belonged to

the TiO2(B) monoclinic system as indicated by the indexed SAED pattern (Figure 6.2a1) with

lattice parameters a = 1.216 nm, b = 0.374 nm, c = 0.651 nm, and 𝛽 = 107.29o in the space group

C2/m. The HRTEM image of NB 130/550 (Figure 6.2a2) contained irregular nanocrystalline

grains with d-spacings of 0.35 nm and 0.38 nm, corresponding to the (110) and (003) planes of

TiO2(B). When the calcination temperature was increased to 700oC as in NB 130/700, the crystal

A B C D

145

phase changed from TiO2(B) to an anatase tetragonal system with lattice parameters a = 0.379

nm and c = 0.951 nm) in Figure 6.2b1. The HRTEM image of NB 130/700 (Figure 6.2b2) depicts

more continuous grains with higher crystallinity compared to NB 130/550 with d-spacings of

0.37 nm and 0.49 nm, corresponding to the (101) and (002) planes of anatase.

At the higher hydrothermal synthesis temperature of 240oC, the SAED and HRTEM images of

NB 240/ 550 (Figure 6.2c) and NB 240/700 (Figure 6.2d) exhibited more continuous grains with

fewer visible surface defects when compared with NB 130/550 and NB 130/700. In Figure 6.2c1,

the SAED image of NB 240/550 was indexed as predominantly TiO2(B) The HRTEM images

(Figure 6.2c2) indicate that TiO2(B) crystalline grains are present with d-spacing of 0.66 nm and

0.35 nm corresponding to the (200) and (011) planes, respectively. There were also anatase

grains present with d-spacings of 0.34 nm and 0.46 nm, corresponding to the (101) and (002)

planes. As with the samples hydrothermally synthesized at 130oC, the samples hydrothermally

synthesized at 240oC exhibited the conversion of TiO2(B) to anatase when increasing the

calcination temperature from 550oC to 700oC as observed in Figure 2d. The d-spacings of NB

240/700 (Figure 6.2d2) were 0.37 nm and 0.47 nm, which match the (101) and (002) planes of

anatase.

146

Figure 6.2 TEM images with SAED indexed regions (yellow) and HRTEM images with

corresponding FT image of LEN samples (figure created by Robert Liang at

the University of Waterloo using results obtained at McMaster University)

147

6.3.1.3 Surface Area

The results of surface area testing (Table 6.3) show that hydrothermal temperature and

calcination temperature both had effects on the BET surface area and pore volumes of the LENs.

Surface area can impact adsorption efficiency and photocatalytic activity, though the latter is not

a simple linear relationship (Qamar et al., 2008). In this study, lower temperatures during the

hydrothermal and calcination steps were associated with higher BET surface area. Thus, NB

130/550 was found to have the highest surface area, the only one above that of P25 and at least

double that of the other LENs, and NB 240/700 had the lowest. The surface area results for NB

130/550 and NB 130/700 presented in Table 6.3 compare favourably with results obtained by

Qamar et al. (2008) and Ali et al. (2016) for materials prepared under similar conditions.

6.3.1.4 Isoelectric Point

The isoelectric point of a substance can be determined by identifying the pH at which the zeta

potential of a particle or colloid is zero. The zeta potential of the four LENs and that of P25

nanoparticles are shown in Figure 6.3. Each point represents the average of four measurements

made on a single aliquot by the zeta potential analyzer. The error bars represent the standard

deviation of these four measurements. The estimated isoelectric point of each nanomaterial

based on the results presented in Figure 6.3 are summarized in Table 6.3.The results indicate that

calcination at 550oC had no or only a small effect on the IEP of the LENs relative to their

precursor material (P25) but calcination at 700oC decreased the IEP substantially, particularly for

NB 130/700. These changes may have had an impact on the interactions between the LENs and

the target contaminants in the raw water, particularly in terms of adsorption behaviour. The IEPs

of the different materials would also be expected to have an effect on their agglomeration and

settling behaviour in different water matrices because particles are generally more likely to

agglomerate and settle when the pH of the water matrix is close to their IEP.

148

Figure 6.3 Determination of isoelectric point of NB 130/550 (A), NB 130/700 (B), NB

240/550 (C), NB 240/700 (D), and P25 nanoparticles (E) using zeta potential

at various pH conditions

-50

-30

-10

10

30

50

2 3 4 5 6 7 8 9 10

Zet

a P

ote

nti

al (m

V)

pH

-50

-30

-10

10

30

50

2 3 4 5 6 7 8 9 10

Zet

a P

ote

nti

al (m

V)

pH -60

-40

-20

0

20

40

60

2 3 4 5 6 7 8 9 10

Zet

a P

ote

nia

l (m

V)

pH

A

-50

-30

-10

10

30

50

2 3 4 5 6 7 8 9 10

Zet

a P

ote

nti

al (m

V)

pH

B

-50

-30

-10

10

30

50

2 3 4 5 6 7 8 9 10

Zet

a P

ote

nti

al (m

V)

pH

C D

E

149

6.3.2 Hydroxyl Radical Formation

Figure 6.S.1 in the supplementary material shows the formation of HTPA via the reaction of

TPA with hydroxyl radicals over time by P25 nanoparticles and the four LENs. Although it

cannot be assumed that there was a one to one relationship between ·OH and HTPA formation –

other researchers have assumed that only 80% of the ·OH formed during photocatalysis interact

with TPA to form HTPA (Ishibashi et al., 2010) – it can be assumed that the number of moles of

·OH formed was at least equal to the number of moles of HTPA formed. The materials calcined

at 700oC, which consisted primarily of anatase, produced far more HTPA, and thus ·OH radicals,

than those calcined at 550oC, which contained both anatase and TiO2(B). As shown in Table 6.3,

the rate of HTPA formation was an excellent fit (R2 = 0.97 to 0.99) to a zero order reaction

model. The rate of HTPA formation ranged from 0.104 ± 0.009 M/min for NB 240/550 to 0.739

± 0.058 M/min for NB 240/700. When the reaction rate constants were normalized to the

available surface area it was even more apparent that NB 240/700, which had a normalized

reaction rate constant of 3.110 ± 0.242 M/min/m2, was far superior to the other materials, which

had normalized reaction rate constants ranging from 0.102 ± 0.009 M/min/m2 for NB 130/550 to

0.870 ± 0.040 M/min/m2 for P25, in terms of ·OH radical formation.

6.3.3 Settling Experiments and Modeling

6.3.3.1 Results of Settling Tests

Conventional settling tanks in full scale water treatment plants are usually rectangular in shape,

operate in a continuous flow through manner, and have detention times ranging from 1.5 to 4

hours (Crittenden et al. 2012). The standard bench-scale tests used to evaluate settling in water

treatment applications, which require a relatively large volume of water, were not feasible for

this study because of materials availability limitations. The simplified settling tests that were

conducted instead clearly showed that the four engineered nanomaterials invariably settled more

quickly than P25 (Figure 6.4) in distilled water. NB 240/700 settled most quickly, followed by

NB 130/550 and NB 240/550. The differences between these three materials were most apparent

between 20 and 100 minutes. NB 130/700 was the slowest to settle and showed more variation

between replicates than the other three engineered nanomaterials. An additional settling test was

150

conducted to provide a qualitative visual counterpart to the data presented in Figure 4.

Photographs were taken of the materials at time zero and after 60 minutes and 24 hours of

settling (Figure 6.5). The photographs clearly illustrate the superior settleability of the

engineered nanomaterials compared to P25 nanoparticles in distilled water.

Nanomaterial synthesis conditions had a strong effect on settling efficiency. In general, the

materials that appeared larger in the TEM pictures (NB 240/550 and NB 240/700) settled more

effectively than the smaller materials. The particle size distributions of all of the materials except

NB 130/700 are shown in Figure 6.S.2 and Figure 6.S.3 in the supplementary material. The

majority of the “particles” measured by the particle sizer were almost certainly agglomerates

because although discrete P25 nanoparticles are known to have a diameter between 20 and 30

nm, the particles detected by the particle sizer were between 10 and 100 times larger than this.

The particle size distribution of NB 240/550 indicates that the solutions made with this material

in both distilled water matrix and the river water matrix contained more large particles than the

suspensions made with the other engineered nanomaterials. The P25 particle size distribution

skewed toward much smaller particle sizes than those of any of the engineered nanomaterials.

All of the materials tested also varied in terms of their uniformity as indicated by the shape of

their particle size distribution. The NB 240/550 particles were the least uniform while those

formed by P25 were the most uniform.

151

Figure 6.4 Settling of P25 nanoparticles and four engineered nanomaterials in purified

water and raw Otonabee River water (n = 3, error bars represent the

standard deviation from the mean)

-100%

-80%

-60%

-40%

-20%

0%

0 20 40 60 80 100 120

Rem

oval

of

Tu

rbid

ity

Time (min)

P25 NB 130/550 NB 130/700 NB 240/550 NB 240/700

-100%

-80%

-60%

-40%

-20%

0%

0 20 40 60 80 100 120

Rem

oval

of

Tu

rbid

ity

Time (min)

P25 NB 130/550 NB 130/700 NB 240/550 NB 240/700

A

B

152

Figure 6.5 Photographs of P25 (A), NB 130/550 (B), NB 130/700 (C), NB 240/550 (D),

and NB 240/700 (E) settling in purified water

0 min

60 min

24 h

A B C D E

153

At first glance, the results of the particle size tests did not support the hypothesis that

agglomerate size alone drove settling behaviour. The largest agglomerates were formed by NB

240/550, but it settled more slowly than NB 130/550 or NB 240/700, both of which had size

distributions that skewed toward smaller particle sizes. This result may have been related to the

conditions of the particle sizing test, which was only able to measure dilute solutions of the

materials (approximately 0.05 g/L) and the behaviour of the materials at this concentration may

not accurately predict their behaviour at higher concentrations. In our study, the larger materials

did, for the most part, settle more quickly than the smaller materials. Other researchers have

noted that the density of nanoparticle agglomerates is often lower than that of the constituent

nanoparticles (Deloid et al., 2014). For example, Liu et al. (2013) determined that the effective

density of the agglomerates formed by their linear engineered TiO2 nanomaterials was 1.2 g/cm3.

Numerous factors can impact the effective density of the agglomerates formed by nanomaterials

in solution including individual particle size, shape, and surface area (Liu et al., 2013; Hotze et

al., 2010). For example, for linear TiO2 nanomaterials, higher surface area has been linked to

greater stabilization of nanomaterial suspensions. NB 240/550, which had a surface area of 55

m2/g, did indeed settle more slowly than NB 240/700, which had a surface area of 19 m2/g. The

surface charge on the particles may also have impacted their settling efficiency because particles

are more likely to agglomerate when the pH of the matrix is close to their IEP. The distilled

water used in this study had a pH between 5.5 and 6. At this pH NB 130/550, NB 240/550, and

NB 240/700 were all neutrally charged but NB 130/700 was negatively charged, making it more

likely to remain dispersed in water. These phenomena alone do not explain the settling behaviour

of NB 130/550, however, indicating that other forces may also have affected the agglomeration

and settling of this nanomaterial in distilled water.

The settling tests were repeated in the Otonabee River water as shown in Figure 4B. In this case,

P25 nanoparticles actually settled more quickly than any of the engineered nanomaterials. This

surprising finding may indicate the agglomerates formed by the P25 nanoparticles in this water

matrix were larger or denser than those formed in the distilled water matrix (see analysis in

supplemental material). In contrast, the settling of the larger LENs (NB 240/550 and NB

240/700) appears to have been hindered in the river water matrix. This effect was less

pronounced for the smaller LENs (NB 130/550 and NB 240/700). The settling behaviour

observed in the river water was likely influenced by various components of the water matrix,

154

particularly pH, NOM, and ions such as calcium, as well as the characteristics of the

nanomaterials themselves. As described in the previous paragraph, pH can influence the surface

charge of the nanomaterial and thus its propensity to agglomerate. The pH of the river water was

approximately 8, which is above the IEPs of all five nanomaterials used in this study. At this pH,

the materials should all be less likely to agglomerate than at pH values closer to their IEPS (pH 4

to 6.5). Increasing levels of ions can minimize the repulsive electrostatic forces that keep

particles from coming together, allowing van der Waals forces to dominate and encouraging

greater agglomeration (Hotze et al., 2010). Additionally, the presence of calcium ions in the

water matrix has been shown to increase the apparent IEP of TiO2 nanomaterials as well as the

size of their agglomerates and sedimentation efficiency (Liu et al., 2013; Hotze et al., 2010),

while humic acid (a major component of NOM) has been shown to have the opposite effects (Liu

et al., 2013; Thio et al., 2011). The size and shape of the nanomaterials can also impact their

sedimentation efficiency because these material characteristics can influence the size and shape

of the resulting agglomerates and the interactions of the materials with water matrix components

(Hotze et al., 2010). For example, Liu et al. (2013) showed that humic acid had a strong

stabilizing effect on suspensions of linear TiO2 nanomaterials but that the effect of ionic strength

was dependent on the constituent ions and their concentration – in some cases, the addition of

calcium actually stabilized the suspensions. Some or all of these phenomena were likely at play

during the settling experiments presented in this study.

6.3.3.2 Modeling of Settling Results

Stokes’ Law is commonly used to model the settling of discrete particles through a liquid

medium. For a hard spherical particle, Stokes’ Law can be simplified to:

𝑣𝑠 =𝑔(𝜌𝑝−𝜌𝑤)𝑑𝑝

2

18𝜇 (6.4)

Where vs is the terminal settling velocity of the particle (m/s), g is the acceleration due to gravity

(9.81 m/s2), ρp is the density of the particle (kg/m3), ρw is the density of the water (kg/m3), dp is

the diameter of the particle (m), and is the viscosity of the water (kg/m.s).

The settling velocity predicted by Stokes’ Law for P25 nanoparticles settling independently in

solution was approximately six orders of magnitude slower than typical settling velocities for

155

small and medium sized alum flocs as listed by Crittenden et al. (2012) (see supplementary

material). The settling velocities predicted for the LENs, which were calculated under the

unrealistic assumption that the LENs or their agglomerates would behave as hard spheres, ranged

from one to two orders of magnitude slower than typical values for alum flocs. If instead it was

assumed that the various nanomaterials formed spherical agglomerates with hydrodynamic

diameters equal to the D50 values obtained during the particle sizing tests, that the density of

these agglomerates was equal to TiO2’s material density (4.26 g/cm3), and that the agglomerates

settled independently of one another, the settling velocities predicted for the four linear

engineered nanomaterials ranged from 5.8 x 10-5 m/s for NB 130/550 in the river water matrix to

9.2 x 10-4 m/s for NB 240/550 in the distilled water matrix. The latter is within the range of

typical settling velocities for small alum flocs provided by Crittenden et al. (2012). Finally, if the

effective density of the agglomerates was assumed to be equal to that reported by Liu et al.

(2013) for their LENs (1.2 g/cm3), the settling velocities of the nanomaterials decreased by

approximately one order of magnitude.

When it was assumed that individual (non-agglomerated) particles settled independently of one

another and that these particles had a density equal to the material density of TiO2 (4.26 g/cm3),

some of the settling behaviour observed in distilled water was a reasonable match to that

predicted by Stokes’ Law. For example, Stokes’ Law predicted that it would take 56 minutes for

50% of the original mass of NB 130/550 to settle out of water, and 50% removal was indeed

achieved in both water matrices within this time interval. In contrast, Stokes’ Law predicted that

50% of the NB 240/550 would settle out of the distilled water matrix within 9 minutes, but it

took approximately an hour to achieve 50% removal of this nanomaterial in the distilled water

matrix and 50% removal was not achieved in the river water matrix within the two hour

timeframe of the test. These assumptions are, however, somewhat unrealistic, as the

nanoparticles and LENs undoubtedly formed agglomerates under most, if not all conditions. The

more likely explanation is that the nanomaterials formed agglomerates with effective densities

below the material density of TiO2 and that this, in combination with the overall size of the

agglomerates, drove their settling behaviour.

As shown in the supplemental file, a sensitivity analysis was conducted to explore the effect of

effective density on nanomaterial settling according to Stokes’ Law. Based on this analysis, it

was determined that agglomerates of NB 130/550, NB 240/550, and NB 240/700 would have to

156

have effective densities of 1.37 g/cm3, 1.02 g/cm3, and 1.21 g/cm3, respectively, to conform to

Stokes’ Law in the distilled water and 1.44 g/cm3, < 1.06 g/cm3, and 1.16 g/cm3 in the river

water matrix. When these values were inputted into the Stokes’ Law equation, the resulting

predicted settling velocities ranged from 3.25 x 10-6 m/s for NB 240/550 in river water to 1.33 x

10-5 m/s for NB 240/700 in distilled water, which are approximately two orders of magnitude

lower than typical values for alum flocs (Crittenden et al. 2012). The same analysis suggested

that P25 nanoparticles formed agglomerates in the river water matrix that were smaller in size

but had a higher greater effective density (3.50 g/cm3) than those formed by the LENs. This

likely explains why the nanoparticles settled more quickly than the LENs in this water matrix.

Overall, this analysis suggests that the settling behaviour of the agglomerates of the P25

nanoparticles and the LENs could, to some extent, be modeled using Stokes’ Law but also that

the LENs settled more slowly than the particles formed during other common water treatment

processes such as coagulation. The practical implication of this is that more and/or larger

sedimentation tanks would be required for a system based around the LENs compared to one

employing coagulation.

The calculations and results presented above are predicated on numerous assumptions, some of

them better supported than others. In order to fully characterize the settling behaviour of the

nanomaterials used in this study it would be necessary to know both the size and the effective

density of the agglomerates formed by each material in the two water matrices. This was beyond

the scope of this proof of concept study, however, it would be a necessary exercise if

sedimentation were selected as the separation mechanism of choice in a TiO2-based water

treatment system.

6.3.4 Photocatalytic Degradation of Methylene Blue Dye Over Time

Methylene blue degradation has been used by many researchers as a measure of the effectiveness

of photocatalytic systems (Mills, 2012). In this study, P25 and NB 240/700 were nearly equal in

terms of their ability to degrade methylene blue (Figure 6.6), achieving 86 ± 4 % and 80 ± 2 %

removal, respectively, after 30 minutes, suggesting that they were equally photocatalytically

active under the experimental conditions despite the fact that NB 240/700 had a substantially

smaller available surface area than the P25 nanoparticles (19 m2/g vs. 57 m2/g), likely because

157

NB 240/700 contained predominantly anatase whereas P25 contains both the rutile and the

anatase phases of TiO2 . NB 130/700, which also consisted predominantly of anatase, was

slightly less effective for methylene blue degradation than NB 240/700, achieving 58 ± 5%

removal after 30 minutes, despite its higher surface area, possibly owing to the larger number of

surface defects on this material relative to NB 240/700. The two LENs that were calcined at

550oC were less photocatalytically active than P25 or the materials calcined at 700oC, achieving

33 ± 4% (NB 130/550) and 40 ± 5% (NB 240/550).

Figure 6.6 Photocatalytic degradation of methylene blue dye by P25 nanoparticles and

four LENs (error bars represent the standard deviation from the mean)

As shown in Table 6.4, In most cases, methylene blue degradation by the LENs was a good fit to

a first order degradation model (R2 > 0.9). This is in line with the findings of other researchers

(Ali et al., 2016; Mills and McFarlane, 2007). The only material where this relationship did not

have a R2 above 0.90 was NB 240/700, which was a result of the very fast degradation observed

between 0 and 5 minutes. The reaction rate constants in Table 6.4 confirm that the LENs

calcined at 700oC, which consisted primarily of anatase, were more effective than those calcined

at 550oC, which consisted primarily of TiO2(B). A small amount of decolourization was

observed in the absence of TiO2 at longer irradiation times. Although methylene blue only

0%

20%

40%

60%

80%

100%

0 10 20 30 40 50 60

Met

hyle

ne

Blu

e D

egra

dati

on


Light Only P25 NB 130/550 NB 130/700 NB 240/550 NB 240/700

158

absorbs a small amount of light at 365 nm, it can nonetheless cause photobleaching due to a

combination of photoreductive and photooxidative reactions (Mills, 2012).

Table 6.4 Removal, reaction rate constants, and R2 values for the pseudo-first-order

degradation of methylene blue dye by UV light, P25 nanoparticles, and four

LENs (error values represent the 95% confidence interval)

Material

Fit of First Order

Degradation Model

(R2)

Reaction Rate

Constant, k (1/min)

Normalized Rate

Constant, knorm

(1/(min*m2))

Light Only 0.52 -0.001 ± 0.000 n/a

P25 0.94 -0.021 ± 0.002 -0.075 ± 0.009

NB 130/550 0.90 -0.010 ± 0.001 -0.020 ± 0.003

NB 130/700 0.95 -0.012 ± 0.001 -0.078 ± 0.008

NB 240/550 0.96 -0.007 ± 0.001 -0.024 ± 0.002

NB 240/700 0.78 -0.012 ± 0.003 -0.130 ± 0.03

The relationship between crystallinity and reactivity was further elucidated by normalizing the

reaction rate constants to the available surface area (TiO2 dose multiplied by the BET surface

area and the volume of the sample). The differences between the unmodified reaction rate

constants and the normalized ones indicate that the overall effectiveness of each material is a

function of both surface area and crystallinity. For example, the normalized rate constant for NB

240/700 was over twice that of P25 and NB 130/700, indicating that it was by far the most

photocatalytically active of the materials tested in this study. This is likely because it had fewer

defects and higher crystallinity as indicated by the more resolved lattice spacings in the HRTEM

images. In contrast, NB 130/550 was the least photocatalytically active of the materials tested.

The HRTEM analysis for this material indicated that it was predominantly TiO2(B), a less

photocatalytically active form of TiO2, and that it contained a large number of surface defects,

which likely further impaired its reactivity. It nonetheless had an unmodified reaction rate

constant nearly equal to that of NB 130/700, which was primarily made up of anatase. However,

the large surface area of NB 130/550 (99 m2/g) relative to the other materials allowed it to

remove a comparable amount of dye despite its relatively low reactivity.

159

6.3.5 Removal of DBP Precursor Surrogates via Adsorption and Photocatalysis

Under the experimental conditions used in this study, P25 nanoparticles removed 20% of the

DOC (Figure 6.7) and 31% of the UV254 (Figure 6.8) present in the raw water through

adsorption alone (i.e. at 0 minutes of irradiation). This is a substantial amount of removal given

the relatively low concentration of TiO2 employed in this study. The four LENs were less

effective that P25 for DOC adsorption but all four nonetheless adsorbed at least a small amount.

NB 130/550 and NB 240/550 were more adsorptive than the two materials calcined at 700oC,

likely owing to their greater surface area. Electrostatic attraction and repulsion effects may also

have contributed to adsorption to some extent. Most NOM compounds are negatively charged

above pH 4, thus in Otonabee River water, which has a pH of approximately 8, NOM would be

negatively charged and the TiO2 nanomaterials would either be electrostatically neutral or

negatively charged (NB 130/700 and NB 240/700). Electrostatic repulsion between negatively

charged materials and negatively charged NOM may have prevented some adsorption that would

otherwise take place due to other forces. This would explain why the materials calcined at

700oC, which have lower IEPs and are negatively charged at ambient pH were less likely to

adsorb NOM than P25 or the two materials calcined at 550oC.

All four LENs were able to degrade DOC and UV254 to some extent as shown in Figure 6.7 and

Figure 6.8. No removal of either parameter was observed in the light only controls (results not

shown). P25 and NB 130/550 were the most effective for both DOC and UV254 removal,

followed closely by NB 240/700. NB 240/550 and NB 130/700 were less effective but

nonetheless removed over 25% of the DOC and approximately 70% of the UV254 in the raw

water upon irradiation with UVA light.

.

160

Figure 6.7 Photocatalytic degradation of DOC by P25 nanoparticles and four LENs

Figure 6.8 Removal of UV254 by photocatalysis with P25 nanoparticles and four LENs

0%

25%

50%

75%

100%

0 15 30 45 60

DO

C R

emo

va

l


P25 NB 130/500 NB 130/700 NB 240/550 NB 240/700

0%

25%

50%

75%

100%

0 15 30 45 60

UV

254 R

emoval


P25 NB 130/500 NB 130/700 NB 240/550 NB 240/700

161

For the most part, the degradation of DOC by P25 and the LENs fit a pseudo-first-order

degradation model. The reaction rate constants (Table 6.5) followed the same trend as the

removal values shown in Figure 6.7 and Figure 6.8, however, when the rate constants were

normalized to the available surface area, a different pattern emerged. The materials calcined at

700oC consistently had higher (1.5-3 times higher) normalized reaction rate constants than P25

and the two materials calcined at 550oC, likely because the former consisted mainly of anatase

and thus were more photocatalytically active and also because they produced more hydroxyl

radicals upon irradiation. P25 and NB 130/550 nonetheless achieved good DOC and UV254

removal owing to their high surface area.

Table 6.5 Reaction rate constants and R2 values for the pseudo-first-order degradation

of DOC by UV light, P25 nanoparticles, and four LENs (error values

represent the 95% confidence interval of the rate constant)

Material DOC UV254

R2 Rate constant, k

(1/min)

Normalized rate

constant, knorm

(1/(min*m2))

R2 Rate constant,

k (1/min)

Normalized rate

constant, knorm

(1/(min*m2))

P25 0.93 -0.003 ± 0.000 -0.005 ± 0.001 0.96 -0.015 ± 0.001 -0.021 ± 0.002

NB 130/550 0.97 -0.004 ± 0.000 -0.003 ± 0.000 0.98 -0.019 ± 0.001 -0.015 ± 0.001

NB 130/700 0.93 -0.002 ± 0.000 -0.005 ± 0.001 0.99 -0.014 ± 0.001 -0.038 ± 0.002

NB 240/550 0.80 -0.001 ± 0.000 -0.001 ± 0.000 0.95 -0.008 ± 0.001 -0.011 ± 0.001

NB 240/700 0.97 -0.003 ± 0.000 -0.013 ± 0.001 0.96 -0.016 ± 0.002 -0.065 ± 0.006

Our previous work (Gora and Andrews, 2017) and that of other researchers (Liu et al.,2008),

though limited to P25 nanoparticles, has demonstrated that UV/TiO2 photocatalysis degrades

larger, more aromatic NOM compounds into smaller, less aromatic ones. This process has been

linked to the formation of intermediate compounds that are more reactive with chlorine (and thus

more likely to form DBPs when chlorine is added for disinfection) after short UV/TiO2 treatment

times followed by the gradual degradation of these intermediates at longer treatment times, with

a resulting decrease in the overall DBPfp of the water (Gora and Andrews, 2017; Liu et al.,

2010). The higher DOC and UV254 degradation rates observed for some of the LENs, especially

after normalization, suggests that these materials may be even more effective than P25 for

162

DBPfp reduction. Specifically, NB 240/700 may prove to react more quickly with DBP

precursors and thus require less time to reach a point where overall DBPfp is decreasing, rather

than increasing, and is the subject of ongoing research.

6.3.6 Electrical Energy per Order

The EEO concept is useful for comparing different types of light-driven systems and processes.

For example, Collins and Bolton (2016) compared EEOs for methylene blue degradation to show

that UV/H2O2 was far more efficient than UV/TiO2 for dye decolourization (EEOUVH2O2 = 0.63

kWh/order/m3 vs. EEOUVTiO2 = 16.4 kWh/order/m3). The EEO for UV/TiO2 reported by Collins

et al. is lower than those calculated for P25 and the various LENs in the current study (Table

6.6). It should be noted, however, that the authors used a much lower starting concentration of

methylene blue (0.32 mg/L), did not report the experimental conditions (UV source, UV

irradiance, H2O2 or TiO2 dose, etc.).

Table 6.6 EEO values provided by Collins and Bolton (2016) for methylene blue

degradation by UV/H2O2 and UV/TiO2 and EEO values for the degradation

of methylene blue by P25 and second generation LENs irradiated by UVA

LEDs

Process Dose Lamp Type Lamp Power Average Irradiance EEO

g/L W mW/cm2 kWh/order/m3

UV/H2O21 --2 UV3 --2 --2 0.63

UV/TiO21 --2 UV3 --2 --2 16.4

Light Only -- UVA LED 2.7 4.9 1,121

P25 0.1 UVA LED 2.7 4.9 42

NB 130/550 0.1 UVA LED 2.7 4.9 95

NB 130//700 0.1 UVA LED 2.7 4.9 81

NB 240/550 0.1 UVA LED 2.7 4.9 133

NB 240/700 0.1 UVA LED 2.7 4.9 69

1From Collins and Bolton (2016)

2Not reported

3Power (W) not specified

163

The EEO values for methylene blue decolourization using P25 and lab synthesized LENs shown

in Table 6.6 follow a similar trend to the reaction rate constants for methylene blue

decolourization in Table 6.5. That is, P25 had the lowest EEO value, implying that it was the

most efficient material. It was followed by NB 240/700 and NB 130/700, the two materials that

were primarily composed of anatase. Finally, the two materials that contained both anatase and

TiO2 had the highest EEOs, implying that a system employing one these nanomaterials would be

less efficient than one employing P25 or one of the nanomaterials calcined at 700oC.

A study by Yen and Yen (2015) explored the use of UV/H2O2 for DOC and THMfp removal

from a synthetic water matrix made with commercial humic acids. Their experiments were

conducted using a 9 W low pressure UV lamp (maximum irradiance at 254 nm) and three doses

of H2O2. EEO values for the removal of DOC and THMfp by P25 and the second generation

LENs are compared to those reported by Yen and Yen (2015) for UV/H2O2 treatment in Figure

6.9.

Figure 6.9 EEO values for DOC removal from synthetic water via UV/H2O2 treatment

with a low pressure UV lamp (Yen and Yen, 2015) and DOC removal from

raw surface water via UV/TiO2 treatment with P25 and four lab synthesized

LENs irradiated with UVA LEDs

0

100

200

300

400

500

600

EE

O (

kW

h/o

rder

/m3)

UV/TiO2UV/H2O2

164

Yen and Yen achieved much lower EEO values for the degradation of DOC using UV/H2O2 than

were obtained in the current study of UV/TiO2 treatment with P25 and lab synthesized LENs

irradiated with UVA LEDs. This may be related to the type of NOM employed in each study

(commercial humic acids vs. real surface water NOM), the absence of ROS scavengers in their

synthetic water matrix, or due to other experimental factors. It may also be that UV/H2O2 is

simply a more effective treatment for NOM removal than UV/TiO2, but this cannot be claimed

with confidence without evidence from parallel experiments run on the same water matrix under

UV/H2O2 and UV/TiO2 experimental conditions that are comparable in terms of energy

utilization and/or materials cost. It is also unclear whether this trend will be hold true for DBP

precursor removal.

6.3.7 Comparison of Reaction Rate Constants and Implications for Degradation Pathways

The degradation rate constants and normalized degradation rate constants for DOC and UV254

were graphed against the formation rate constants for HTPA, a measure of ·OH radical formation

(see Table 6.S.7 in the supplementary material). There was no obvious correlation between the

DOC or UV254 degradation rates and ·OH radical production rates, suggesting that not all of

nanomaterials reduced these parameters via ·OH radical mediated reactions alone. There was a

moderate correlation between the normalized DOC and UV254 degradation rates and ·OH

radical production as predicted by HTPA formation. The correlation was much stronger if P25

removed from the analysis (see Figure 6.S.4). This indicates that the normalized DOC and

UV254 degradation rates were good proxies for ·OH radical-related photocatalytic NOM

degradation by the LENs but not for its photocatalytic degradation by P25. The fact that the

correlation only existed when the NOM degradation rates were normalized to surface area

suggests any additional NOM degradation observed for the LENs was related to surface

phenomena such as oxidation by photo-generated electron holes. NOM degradation by P25

appears to have been more complex, possibly due to the formation of other ROS (e.g.

superoxide) upon irradiation.

The most interesting ramification of these correlations is that different materials

photocatalytically degraded NOM through different oxidative pathways and that for the LENs,

these pathways were, to some extent, determined by the calcination temperature used during

165

synthesis. For example, NB 240/700 and NB 130/550 achieved 86 ± 1% 92 ± 1% reduction in

UV254 after 60 minutes of irradiation. As indicated by the results in Figure S.1 and Table 3,

however, NB 240/700 produced over 5 times as much HTPA, and thus ·OH radicals, as NB

130/550. The fact that NB 130/550 was nonetheless equally capable of reducing the UV254 of

the water indicates that other phenomena contributed to NOM removal by this material and the

discrepancy between the normalized and non-normalized degradation rate constants suggest that

these phenomena were likely surface related (e.g. NOM oxidation via photo-generated holes).

The practical implications of these differences on DBP precursor removal are as of yet unclear,

however, one of them might be that some LENs will be more likely than others to be negatively

impacted by water matrix components such as ROS scavengers, which consume ·OH radicals

and thus slow the overall rate of removal of target contaminants (Liao et al., 2010), or species

that compete with NOM for adsorption sites on the TiO2 surface, which may decrease the

effectiveness of LENs that operate primarily via surface mediated degradation reactions.

Research to further characterize the mechanisms underlying NOM degradation by the different

LENs is ongoing.

Conclusions

Four LENs were successfully synthesized using a simple hydrothermal method. The LENs

differed from one another and from industry standard nanoparticles in terms of size, BET surface

area, and other physical and chemical characteristics. The materials were all able to degrade

substantial amounts of natural organic matter after less than an hour of irradiation with high

intensity UVA LED light at 365 nm.

The materials varied substantially in terms of their ability to degrade two disinfection byproduct

precursor surrogates, DOC concentration and UV254 absorbance, but greater removals were

consistently observed for materials calcined at the higher temperature of 700oC, particularly

when the results were normalized to surface area. The variation was related to surface area,

charge, propensity to agglomerate, crystal phase, and the presence of defects within the crystal

structure. The reaction rates were particularly influenced by the surface area and crystallinity of

the materials. A simple fluorescence-based test was used to compare the nanomaterials in terms

of their propensity to generate hydroxyl radicals when illuminated with UVA LED light. The

166

results suggest that the predominantly anatase materials interacted with NOM primarily via

hydroxyl radical mediated degradation reactions whereas the mixed phased nanomaterials

removed NOM through a combination of adsorption and degradation reactions with photo-

generated holes or ROS other than the ·OH radical.

All four engineered nanomaterials settled out from distilled water more quickly than standard

P25 nanoparticles, likely due to their size and effective density of their agglomerates relative to

those of P25 in this matrix. The results were reversed in a real river water sample – P25 settled

out quickly, possibly due to the presence of agglomeration-inducing ions such as calcium in this

water matrix, but the settling of the larger LENs was slightly hindered, likely due to the presence

of NOM in the river water. Based on their ability to remove NOM their propensity to settle out

of water, a subset of these engineered nanomaterials may be a viable alternative to P25 for

drinking water treatment, though their effectiveness may be limited in water matrices containing

elevated levels of ROS scavengers.

Acknowledgements

The authors would like to acknowledge the contributions of Leonardo Furtado, Jim Wang, Tassia

Brito, Adrielle Costa, Wan-Ying (Jenny) Yue, Nathan Moore, Alireza Mahdavi, and Jeffrey

Siegel to this project. Funding for this study was provided through Canada’s Natural Sciences

and Engineering Research Council’s Strategic Project Grant program [STPGP 430654-12] and

Canada Graduate Scholarship program as well as through the Ontario Graduate Scholarship

program.

167

References

Ali, S., Granbohm, H., Ge, Y., Singh, V.K. (2016) Crystal structure and photocatalytic properties

of titanate nanotubes prepared by chemical processing and subsequent annealing, Journal of

Materials Science, 51, 7322-7335

Arlos, M., Liang, R., Hatat-Fraile, M., Bragg, L., Zhou, Y-N, Servos, M., Andrews, S. (2016)

Photocatalytic decomposition of selected estrogens and their estrogenic activity by UV-LED

irradiated TiO2 immobilized on porous titanium sheets via thermal-chemical oxidation, Journal

of Hazardous Materials, 318, 541-550

Avisar, D., Horovitz, I., Lozzi, L., Ruggieri, F., Baker, M., Abel, M-L, Mamane, H. (2013)

Impact of water quality on removal of carbamazepine in natural waters by N-doped TiO2 photo-

catalytic thin film surfaces, Journal of Hazardous Materials, 244-245, 463-471



215

Collins, J. and Bolton, J.R. (2016) The Advanced Oxidation Handbook, American Water Works

Association, Denver, CO

Crittenden, J.C., Trussell, R.R., Hand, D.W., Howe, K.J., Tchobanoglous, G. (2012) MWH’s

Water Treatment: Principles and Design, 3rd Edition, John Wiley and Sons

Daneshvar, N., Salari, D., Niaei, A., Rasoulifard, M.H., Khataee, A.R. (2005) Immobilization of

TiO2 nanopowder on glass beads for the photocatalytic decolorization of an azo dye C.I. Direct

Red 23, Journal of Environmental Science and Health 40, 1605-1617

Fotiou, T., Triantis, T.M., Kaloudis, T., Hiskia, A. (2015) Evaluation of the photocatalytic



Journal, 261, 17-26

168




Hrudey,S. (2009) Chlorination disinfection by-products, public health risk tradeoffs, and me,

Water Research 43 (2009) 2057-2092

Hu, A., Zhang, X., Oakes, K.D., Peng, P.N., Zhou, Y., Servos, M. (2011) Hydrothermal growth

of free standing TiO2 nanowire membranes for the photocatalytic degradation of

pharmaceuticals, Journal of Hazardous Materials, 189, 278-285

Ishibashi, K, Fujishima, A., Watanabe, T., Hashimoto, K. (2000) Quantum yields of active

oxidative species formed on TiO2 photocatalyst, Journal of Photochemistry and Photobiology A:

Chemistry, 134, 139-142

Kanakaraju, D., Glass, B.D., Oelgemoller, M. (2014) Titanium dioxide photocatalysis for

pharmaceutical wastewater treatment, Environmental Chemistry Letters, 12, 27-47

Kasuga, T., Hiramatsu, M., Hoson, A. Sekino, T., Niihara, K. (1999) Titania nanotubes prepared

by chemical processing, Advanced Materials, 11, 1307-1311

Kim, S-C and Lee, D-K (2005) Preparation of TiO2-coated hollow glass beads and their

application to the control of algal growth in eutrophic water, Microchemical Journal 80, 227-232

Klinger, M. and Jäger, A (2015) Crystallographic Tool Box (CrysTBox): automated tools for

transmission electron microscopists and crystallographers. Journal of Applied Crystallography

48 (2015) doi:10.1107/S1600576715017252.

Krasner, S. (2009) The formation and control of emerging disinfection by-products of health

concern, Philosophical Transactions of the Royal Society, 367, 4077-4095

Leshuk, T., Everett, P., Krishnakumar, Wong, H.K., Linley, S., Gu, F., (2013) Mesoporous

magnetically recyclable photocatalysts for water treatment, Journal of Nanoscience and

Nanotechnology, 13, 3127-3132

169

Liao, C-H, Kang, S-F, Wu, F-A (2010) Hydroxyl radical scavenging role of chloride and


Liu, S., Lim, M., Fabris, R., Chow, C., Chiang, K., Drikas, M., Amal, R. (2008) Removal of


characterisation studies, Chemosphere, 72, 263-271

Liu, S, Lim, M., Amal, R. (2014) TiO2-coated natural zeolite: Rapid humic acid adsorption and

effective photocatalytic regeneration, Chemical Engineering Science 105 (2014) 46-52

Liu, S., Lim, M., Fabris, R., Chow, C.W.K., Drikas, M., Korshin, G., Amal, R. (2010) Multi-


organic matter, Water Research 44, 2525-2532

Liu, W., Sun, W, Borthwick, A.G.L., Ni, J. (2013) Comparison on aggregation and

sedimentation of titanium dioxide, titanate nanotubes, and titanate nanotubes-TiO2: Influence of

pH, ionic strength, and natural organic matter, Colloids and Surfaces A: Physicochem Eng.

Aspects 434, 319-328

Mills, A. and McFarlane, M. (2007) Current and possible future methods of assessing the

activities of photocatalyst films, Catalysis Today, 129, 22-28

Mills, A. (2012) An overview of the methylene blue ISO test for assessing the activities of

photocatalytic films, Applied Catalysis B: Environmental, 128, 144-149

Ng, M., Kho, E.T., Liu, S., Lim, M., Amal, R. (2014) Highly adsorptive and regenerative


Journal, 246, 196-203

Nosaka, Y. and Nosaka, A.Y. (2013) Identification and roles of the active species generated on

various catalysts, in: Photocatalysis and Water Purification: From Fundamentals to Recent

Applications, Ed. Pichat, P., Wiley-VCH Verlag GmbH and Co., pp. 1-24

Philippe, K.K., Hans, C., MacAdam, J., Jefferson, B., Hart, J., Parsons, S.A. (2010)

Photocatalytic oxidation, GAC, and biotreatment combinations: An alternative for the

coagulation of hydrophilic rich waters?, Environmental Technology, 31, 1423-1434

170


trihalomethane formation in drinking water sources, Environmental Engineering Science, 31,

117-126

Qamar, M., Yoon, C.R., Oh, H.J., Lee, N.H., Park, K., Kim, D.H., Lee, K., Lee, W.J., Kim, S.J.,

(2008) Preparation and photocatalytic activity of nanotubes obtained from titanium dioxide,


Rincón, A-G and Pulgarin, C. (2004) Bactericidal action of illuminated TiO2 on pure Escherichia

coli and natural bacterial consortia: post irradiation events in the dark and assessment of the

effective disinfection time, Applied Catalysis B: Environmental, 49, 99-112

Seo, M-H, Yuasa, M., Kida, T., Huh, J-S, Shimanoe, K., Yamazoe, N. (2009) Gas sensing

characteristics and porosity control of nanostructured films composed of TiO2 nanotubes,

Sensors and Actuators B: Chemical, 137, 513-520

Shi, H., Magaye, R., Castranova, V., Zhao, J. (2013) Titanium dioxide nanoparticles: A review

of current toxicological data, Particle and Fibre Toxicology 10:15

Turolla, A, Piazzoli, A., Budarz, J., Wiesner, M., Antonellia, M. (2015) Experimental

measurement and modelling of reactive species generation in TiO2 nanoparticle photocatalysis,

Chemical Engineering Journal, 271, 260-268

van Leeuwen, J., Daly, R., Holmes, R. (2005) Modeling the treatment of drinking water to

maximize dissolved organic matter removal and minimize disinfection by-product formation,

Desalination, 176, 81-89

Wong, C.L., Tan, Y.N., Mohamed, A.R. (2011) A review on the formation of titania nanotube

photocatalysts by hydrothermal treatment, Journal of Environmental Management, 92, 1669-

1680

Xing, Z., Zhou, W., Du, F., Qu, Y., Tian, G., Pan, K., Tian, C., Fu, H. (2014) A floating

macro/mesoporous crystalline anatase TiO2 ceramic with enhanced photocatalytic performance

for recalcitrant wastewater degradation, Dalton Transactions, 43, 790

171

Yang, Y. and Westerhoff, P. (2014) Presence in, and Release of, Nanomaterials from Consumer

Products, Nanomaterials, Advances in Experimental Medicine and Biology, 811 (Capco, D.G.

and Chen, Y., editors), Springer Science + Business Media, Berlin

Yen, H.Y. and Yen, L. S. (2015) Reducing THMfp by H2O2/UV oxidation for humic acid of

small molecular weight, Environmental Technology, 36 (4), 417-423

Yuan, Z-Y and Su, B-L (2004) Titanium oxide nanotubes, nanofibres, and nanowires, Colloids

and Surfaces A: Physicochemical Engineering Aspects, 241, 173-183

Zheng, D., Andrews, R.C., Andrews, S.A., Taylor-Edmonds, L. (2015) Effects of coagulation on




properties on TiO2 nanoparticle agglomeration, PLOS One, 8, e81239

172

Supplementary Material

Figure 6.S.1 HTPA / ·OH radical formation by P25 and four linear engineered

nanomaterials irradiated with UVA LED light (n = 3, error bars represent

standard deviation from the mean)

0

2

4

6

8

10

0 2 4 6 8 10

HT

PA

(u

M)


NB 130/550 NB 130/700 NB 240/550 NB 240/700 P25

173

Figure 6.S.2 Particle size distribution for P25 nanoparticles, NB 130/550, NB 240/550, and

NB 240/700 suspended in distilled water

Figure 6.S.3 Particle size distribution for P25 nanoparticles, NB 130/550, NB 240/550, and

NB 240/700 suspended in river water

0

1

2

3

4

5

6

7

8

0 50 100 150 200 250 300 350 400 450 500

Per

cen

t

Diameter (m)

P25 NB 130/550 NB 240/550 NB 240/700

0

1

2

3

4

5

6

7

8

0 50 100 150 200 250 300 350 400 450 500

Per

cen

t

Diameter (m)

P25 NB 130/550 NB 240/550 NB 240/700

0

2

4

6

8

0 10 20 30 40 50

Per

cen

t

Diameter (m)

A

0

2

4

6

8

0 10 20 30 40 50

Per

cen

t

Diameter (m)

B

174

Settling Calculations and Analysis

Terminal Settling Velocities

For a hard spherical particle, Stokes’ Law can be simplified to:


2

18𝜇 (6.S.1)




Crittenden et al. (2012) list the following terminal settling velocities for sand particles and alum

flocs. Note that alum flocs usually settle according to Type II settling rather than Type I settling,

however, the comparison remains illuminating.

Table 6.S.1 Terminal settling velocities for sand particles and alum flocs (Crittenden et

al., 2012)

Particle Type Approximate Diameter (mm) Terminal Settling Velocity (m/s)

Sand (ρ = 2.56 g/cm3) 1 (1,000,000 nm) 1.38 x 10-1

Sand (ρ = 2.56 g/cm3) 0.2 (200,000 nm) 2.20 x 10-2

Sand (ρ = 2.56 g/cm3) 0.06 (60,000 nm) 2.50 x 10-3

Small Alum Floc n/a 5.56 x 10-4 to 1.25 x 10-3

Medium Alum Floc n/a 8.33 x 10-4 to 1.39 x 10-3

Large Alum Floc n/a 1.11 x 10-3 to 1.53 x 10-3

175

The terminal settling velocities of the various nanomaterials used in this study were predicted

using Stokes’ law under three conditions:

1. Density equal to material density, hydraulic diameter equal to particle diameter or length.

2. Agglomerate density equal to material density, hydraulic diameter of agglomerate equal

to D50 from particle size distribution.

3. Agglomerate density equal to effective density of linear engineered nanomaterial

effective density provided by Liu et al. (2013), hydraulic diameter of agglomerate equal

to D50 from particle size distribution.

Table 6.S.2 Predicted terminal settling velocities for nanomaterials in distilled water

Nanomaterial Predicted Terminal Settling Velocity (m/s)

ρ = 4.26 g/cm3

dh = dparticle

ρ = 4.26 g/cm3

dh = d50

ρ = 1.20 g/cm3

dh = d50

P25 7.10 x 10-10 1.25 x 10-5 7.59 x 10-7

NB 130/550 7.10 x 10-6 7.75 x 10-5 4.80 x 10-6

NB 130/700 7.10 x 10-6 -- --

NB 240/550 4.44 x 10-5 9.16 x 10-4 5.67 x 10-5

NB 240/700 2.84 x 10-5 2.05 x 10-4 1.27 x 10-5

176

Table 6.S.3 Predicted terminal settling velocities for nanomaterials in a river water

Nanomaterial Predicted Terminal Settling Velocity (m/s)

ρ = 4.26 g/cm3

dh = dparticle

ρ = 4.26 g/cm3

dh = d50

ρ = 1.20 g/cm3

dh = d50

P25 7.10 x 10-10 1.75 x 10-5 1.08 x 10-6

NB 130/550 7.10 x 10-6 5.80 x 10-5 3.59 x 10-6

NB 130/700 7.10 x 10-6 -- --

NB 240/550 4.44 x 10-5 1.87 x 10-4 1.16 x 10-5

NB 240/700 2.84 x 10-5 7.82 x 10-5 4.84 x 10-6

Settling Time

The time required for a particle to settle a given distance can be calculated using the following

equation:

𝑃𝑟𝑒𝑑𝑖𝑐𝑡𝑒𝑑 𝑆𝑒𝑡𝑡𝑙𝑖𝑛𝑔 𝑇𝑖𝑚𝑒 (𝑚𝑖𝑛) =𝑆𝑒𝑡𝑡𝑙𝑖𝑛𝑔 𝑉𝑒𝑙𝑜𝑐𝑖𝑡𝑦 (

𝑚

𝑠) × 1,000(

𝑚𝑚

𝑚)× 60(

𝑠

𝑚𝑖𝑛)

𝑆𝑒𝑡𝑡𝑙𝑖𝑛𝑔 𝐷𝑖𝑠𝑡𝑎𝑛𝑐𝑒 (𝑚𝑚) (6.S.2)

The predicted settling time required to achieve 50% removal of the various nanomaterials via

settling was calculated for suspensions in distilled water and river water and compared to the

observed amount of time required to achieve 50% removal of turbidity the various suspensions.

177


distilled water

Nanomaterial Time to 50%

(min)

Predicted Time to 50%

(min)

ρ = 4.26 g/cm3

dh = dparticle

ρ = 4.26 g/cm3

dh = d50

ρ = 1.20 g/cm3

dh = d50

P25 n/a > 500,000 40 648

NB 130/550 45 56 5 83

NB 130/700 n/a 56

NB 240/550 60 9.0 0.43 7.0

NB 240/700 30 14 0.48 7.8


river water

Nanomaterial Time to 50%

(min)

Predicted Time to 50%

(min)

ρ = 4.26 g/cm3

dh = dparticle

ρ = 4.26 g/cm3

dh = d50

ρ = 1.20 g/cm3

dh = d50

P25 30 > 500,000 23 366

NB 130/550 50 56 6.8 110

NB 130/700 > 120 56

NB 240/550 > 120 9 2.1 34

NB 240/700 105 14 1.3 82

The effective density of the agglomerates of the nanomaterials formed in the two water matrices

was predicted using an iterative method.

178

Table 6.S.6 Predicted effective density of agglomerates formed by four nanomaterials in

distilled water and river water

Nanomaterial Predicted Effective Density of Agglomerates (g/cm3)

Distilled Water River Water

P25 n/a 3.50

NB 130/550 1.37 1.44

NB 240/550 1.022 < 1.055

NB 240/700 1.21 1.155

Correlations Between Reaction Rate Constants

Figure 6.S.4 Linear correlation between normalized DOC and UV254 degradation rate

constants and HTPA/hydroxyl radical formation rate constants for four

linear engineered nanomaterials

-1.0E-01

-8.0E-02

-6.0E-02

-4.0E-02

-2.0E-02

0.0E+00

0.0E+00 2.0E-01 4.0E-01 6.0E-01 8.0E-01

Norm

ali

zed

Deg

rad

ati

on

Rate

Con

stan

t

(1/m

in)

Hydroxyl Radical Formation Rate Constant (M/min)

DOC UV254

179

Table 6.S.7 Fit of linear correlation between hydroxyl radical formation rate constants

and NOM degradation rate constants

Parameter Normalized to Surface Area? P25 Included? Fit of Linear Regression

kDOC No Yes 0.20

No No 0.10

Yes Yes 0.70

Yes No 0.98

kUV254 No Yes 0.07

No No 0.07

Yes Yes 0.54

Yes No 0.97

180

Photocatalysis with Engineered TiO2 Nanomaterials to Prevent the Formation of Disinfection Byproducts in Drinking Water

Abstract

Two photocatalytic linear engineered TiO2 nanomaterials (LENs) were synthesized and

evaluated against commercial standard P25 TiO2 nanoparticles in terms of their effects on

common parameters used to measure and characterize natural organic matter (NOM) and

disinfection byproduct (DBP) precursors in drinking water. DBPs, some of which have been

linked to cancer and other negative human health outcomes, are regulated in most jurisdictions in

North America and Europe as well as parts of Asia, Africa, and South America, and the removal

of DBP precursors is a central goal of conventional drinking water treatment plants. All three of

the nanomaterials evaluated in this study were capable of degrading NOM, including DBP

precursors, when irradiated with UVA LED light. The materials differed in terms of crystal

phase structure, surface morphology, and available surface area. These differences impacted the

predominant NOM removal processes occurring in each case and consequently the overall

treatment efficacy of the two materials in the two different water sources. One of the LENs

evaluated in this study, designated NB 700, reduced the DBP formation potential of one of the

water sources by 90%. This nanomaterial was more effective for NOM degradation than

commercial nanoparticles (P25) and was also readily removed from the water via filtration. As

such, it may be a good candidate for future integration in a UVA/TiO2 photocatalytic water

treatment system. Irrespective of the nanomaterial employed, DBP precursor degradation was

faster in the water source with higher NOM and lower alkalinity and hardness. The electrical

energy per order (EEO) required to degrade DOC and THM precursors in one of the water

sources was comparable to reported values for the degradation of these compounds with another

advanced oxidation process (AOP), UV/H2O2. The results of this study underscore the need for

site specific evaluation of novel photocatalytic oxidation processes for drinking water treatment.

181

Introduction

The removal of natural organic matter (NOM) removal is one of the primary goals of modern

drinking water treatment plants because it can interfere with many standard drinking water

treatment processes and reacts with chlorine and other chemical disinfectants to form an array of

regulated and unregulated disinfection byproducts (DBPs). Recent research suggests that

although commonly regulated classes of DBPs such as trihalomethanes (THMs) and haloacetic

acids (HAAs) are less of a concern for human health than unregulated DBPs (Krasner, 2009),

their presence after chlorination has nonetheless been shown to be a good predictor of DBP

formation in general (Zheng et al, 2015). Preventing the formation of DBPs by removing their

precursor compounds (i.e. NOM) ahead of chlorination has become a common practice in

drinking water treatment plants. Coagulation with metal salts has traditionally been the primary

treatment method used for NOM removal in conventional water treatment plants. More recently,

adsorption, high pressure membrane filtration, and various oxidative strategies such as ozonation

and advanced oxidation processes, have also been employed for the removal of DBP precursors.

In this study, we evaluated photocatalysis with linear engineered titanium dioxide (TiO2)

nanomaterials illuminated by UVA LEDs as an alternative to these existing processes.

TiO2 is a photocatalyst that has occasionally been employed for water and wastewater treatment

but has yet to be widely adopted for these purposes. It has been established that TiO2

photocatalysis degrades NOM and that it preferentially targets large and aromatic NOM

compounds (Liu et al., 2008; Gora and Andrews, 2017). The effect of TiO2 photocatalysis on

DBP formation is less clear cut – a few studies have noted decreases in DBP formation potential

(DBPfp), but others have observed increased DBPfp as larger molecules are broken down into

smaller, more reactive ones (Liu et al., 2008; Kent et al., 2011; Huang et al., 2008; Liu et al.,

2010; Philippe et al., 2010). Increases in DBPfp appear to be related to experimental design, in

particular, irradiation time or UV dose (fluence). Studies focusing on short irradiation times have

often noted increases in DBPfp (Philippe et al., 2010) whereas those employing longer

irradiation have usually reported decreases (Liu et al., 2008).

The characteristics of the background water matrix are known to affect the degradation rates of

target compounds. NOM itself is frequently cited as the most important inhibitor of degradation

in studies focused on the photocatalytic removal of more common organic indicators and

182

pollutants (Autin et al., 2013). Many inorganic components of natural water also have effects,

both positive and negative, on the adsorption and degradation of target compounds by TiO2.

Turbidity, which disperses light, and some organic compounds that absorb UVA wavelengths

can reduce the amount of useable light that reaches the photocatalyst. Chloride and bicarbonate

are known scavengers of hydroxyl radicals (Liao et al., 2001), one of the main reactive oxygen

species (ROS) formed during photocatalysis, while other ions such as phosphate and sulphate

can bind to adsorption sites on the surface of the photocatalyst (Abdullah et al., 1990; Chen et

al., 1997). Higher ion levels can lead to a reduction in the electrostatic repulsive forces between

individual nanoparticles, resulting in agglomeration (Hotze et al. 2010), and a decrease in

available surface area, which is likely to affect the extent of adsorption and photocatalysis. The

effects of some inorganic parameters are more complex. For example, although the presence of

calcium ions increases NOM adsorption to TiO2 (Erhayem and Sohn, 2014; Liu et al., 2013), it

also encourages TiO2 nanomaterials to agglomerate (Zhang et al., 2009), and as such may

indirectly slow degradation by decreasing the total available surface area. Iron readily adsorbs to

TiO2 (Chen and Ray, 2001) and both copper and iron can promote faster reactions between TiO2

and organic contaminants (Butler and Davis, 1993; Franch et al., 2005). Many of the effects

described above are pH dependent and some parameters can interact with one another as well as

with NOM and TiO2, making it difficult to predict the overall effect of a given matrix on the rate

of photocatalytic degradation.

Common issues preventing the use of suspended TiO2 in an aqueous medium include the

provision of adequate mixing, the distribution of light within the medium, and the removal of the

photocatalyst after treatment. Much research has been conducted to engineer TiO2 materials that

are easier to remove from water, usually by immobilizing standard anatase or P25 nanoparticles

on solid supports. This has proven challenging, though a few researchers have had success with

magnetic TiO2 nanomaterials (Ng et al., 2014) and TiO2-covered zeolites (Liu et al., 2014).

TiO2-based linear engineered nanomaterials (LENs) including nanotubes, nanowires, and

nanobelts have been synthesized and characterized by research groups around the world in recent

years. These materials are mainly used in sensors and solar cells (Bavykin and Walsh, 2010), but

may also prove to be useful in drinking water applications. Specifically, other researchers have

demonstrated that TiO2 -based LENs can adsorb and degrade NOM (Liu et al., 2013; Zhang et

183

al., 2009), and their large size relative to standard nanoparticles may make them easier to remove

via common drinking water clarification processes such as filtration or sedimentation.

LENs can be synthesized via alkaline hydrothermal, anodic, or template-guided sol-gel methods.

The alkaline hydrothermal method is well established, does not require highly specialized

laboratory equipment or expensive reagents, and is easily manipulated to yield nanosize

materials with different morphological and chemical characteristics (Bavykin and Walsh, 2010).

Survey studies have established that the precursor materials, hydrothermal synthesis temperature,

extent and method of post synthesis cleaning and ion exchange, and calcination temperature have

important effects on the final products of the synthesis process (Yuan and Su, 2004; Wong et al.,

2011; Qamar et al., 2008; Zheng et al., 2010; Ali et al., 2016). The choice of precursor materials

and hydrothermal temperature affects the overall size and aspect ratio of the LENs, with higher

temperatures generally resulting in larger materials (Yuan and Su, 2004). The washing, ion

exchange, and calcination steps affect the surface and crystalline structures of the materials, and

thus their photocatalytic properties (Qamar et al., 2008; Zheng et al., 2010; Ali et al., 2016).

The photocatalytic bleaching of methylene blue under UVA light has become the de facto

standard method to evaluate the photocatalytic properties of novel TiO2 nanomaterials because

methylene blue is easy to monitor, relatively stable and non-toxic, and the bleaching of

methylene blue is a good indicator of a material’s ability to photocatalytically oxidize other

organic pollutants (Mills, 2012). Previous work by our research group has demonstrated that the

ability of standard TiO2 nanoparticles and lab synthesized LENs to degrade NOM (as measured

by DOC or UV254) when illuminated by UVA light can be predicted based on its ability to

photocatalytically bleach methylene blue dye under similar experimental conditions (Gora and

Andrews, 2015).

This study evaluated the photocatalytic degradation of NOM in two Canadian surface waters by

three TiO2 nanomaterials: P25 nanoparticles and two LENs synthesized in our laboratory. The

calcination temperature used in the final step of the LEN synthesis process was varied in order to

manipulate the crystal phase structure and surface properties of the LENs. The main objective of

the study was to evaluate the effects of these LEN properties on the eventual formation of DBPs

after photocatalytic treatment and subsequent chlorination. To the authors’ knowledge, no other

184

research groups have conducted work linking the synthesis conditions of LENs to the

degradation of DBP precursors in drinking water.


7.2.1 Materials

Evonik Degussa P25 TiO2 nanoparticles were used as the reference material for all experiments

and as the precursor material for the two LENs. All remaining reagents were obtained from

Sigma Aldrich. Raw water was obtained from two Canadian water treatment plants (WTPs)

supplied by river water sources. The Otonabee River (OTB), located in Southern Ontario,

supplies the City of Peterborough while the Ottawa River (OTW) supplies the Britannia WTP,

one of the two major WTPs that serve the City of Ottawa. Raw water samples were gathered at

the inlet of each water treatment plant ahead of prechlorination and used without further

modification. A water quality summary is provided in Table 7.1. Both water sources had

historical pH values near 8 and dissolved organic carbon (DOC) levels ranging from

approximately 4 to 6 mg/L. The UV absorbance at 254 nm (UV254), an indicator of the amount

of aromatic carbon present in a water sample, was nearly twice as high in the OTW water as in

the OTB water. The OTW water’s specific UV absorbance (SUVA) value, an indicator of the

overall aromaticity of the NOM present in the water, was 3.7 ± 0.3 m/mg.L compared to 2.6 ±

0.4 m/mg.L in the OTB water, indicating that the NOM in the OTW water was more aromatic in

character than that in the OTB water. The water sources also differed in terms of turbidity,

alkalinity, calcium content, and conductivity (an indicator of ionic strength), many of which can

affect degradation rates (Liao et al., 2001; Abdullah et al. 1990; Chen et al., 1997). NOM

adsorption to TiO2 (Mwaanga et al. 2014; Erhayem and Sohn, 2014; Liu et al., 2013), and/or the

stability of nanomaterial suspensions. (Liu et al., 2013; French et al., 2009; Loosli et al, 2015; Li

et al., 2016). The OTB water had higher alkalinity, a higher overall concentration of ions, and

approximately four times as much calcium as the OTW water while the OTW water contained

higher concentrations of iron and copper relative to the OTB water.

185


Parameter Units Otonabee River (OTB) Ottawa River (OTW)

DOC1 mg/L 4.7 ± 0.2 6.2 ± 0.5

UV2541 1/cm 0.120 ± 0.015 0.234 ± 0.030

SUVA1 m/mg.L 2.6 ± 0.4 3.7 ± 0.3

pH2 8.2 ± 0.2 7.7 ± 0.2

Turbidity2 NTU 0.6 ± 0.2 3.3 ± 1.0

Alkalinity2 mg/L as CaCO3 87 ± 7 28 ± 6

Hardness2 mg/L as CaCO3 95 ± 11 30 ± 6

Calcium2 mg/L 32.8 ± 3.7 8.3 ± 1.5

Magnesium2 mg/L 3.2 ± 0.3 2.2 ± 0.4

Sodium2 mg/L 6.5 ± 0.8 3.4 ± 0.8

Chloride2 mg/L 11.5 ± 1.3 3.3 ± 0.9

Conductivity2 S/cm 214 ± 19 81 ± 13

Aluminum g/L 3.9 ± 1.8 165 ± 47

Copper g/L 0.7 ± 0.1 27 ± 10

Iron2 g/L 19 ± 9 217 ± 42

Manganese2 g/L 10 ± 6 11 ± 4


2Average and standard deviation of values obtained from Ontario Drinking Water Surveillance Program

2010-2012

7.2.2 Apparatus

The UVA LED apparatus used in this study consisted of four UVA lamps secured to a stand

above a multiple location stir plate that was able to accommodate four beakers at once. The UVA

LEDs (LZ1 UV 365 nm Gen2 Emitter, LED Engin Inc.) had a maximum irradiance at 365 nm.

The average irradiance across the surface of the sample was calculated using a spreadsheet

developed by Bolton and Linden (2003) and was determined to be 4.9 mW/cm2. The irradiance

of each lamp was confirmed before each test using a radiometer (International Light, ILT1400)

equipped with a sensor optimized to measure light at 365 nm (International Light, XRL140B).

186

7.2.3 Synthesis and Characterization of Engineered TiO2 Nanomaterials

Two linear TiO2 nanomaterials were synthesized from P25 nanoparticles according to a simple

hydrothermal method first used by Kasuga et al. (1999) and later modified by others including

Yuan and Su (2004). Both materials were synthesized at 240oC and then calcined at 550oC (NB

550) or 700oC (NB 700). They were then rinsed twice with distilled water to remove unreacted

material and/or smaller linear particles, thus insuring a more consistent final product.

Each batch of LENs was evaluated using a quality control test to assess batch to batch

consistency. Triplicate samples containing 50 mL of 0.03 M methylene blue solutions dosed with

0.1 g/L of TiO2 were irradiated with UVA light (365 nm) with an average irradiance of 4.9

mW/cm2 for 30 minutes. After irradiation, the TiO2 was removed from the samples via

centrifugation and the absorbance of the remaining solution at 665 nm was analyzed and used to

calculate the concentration of methylene blue remaining in solution. On average, NB 550

achieved 53% methylene blue removal and NB 700 achieved 89% methylene blue removal.

Batches that came within 5% of the average methylene blue removal were kept and used for

experiments.

The LENs were characterized using transmission electron microscopy (TEM) to observe shape

and surface characteristics, selected area electron diffraction (SAED) to determine crystal phase,

zeta potential at different pH values to identify the isoelectric point, and N2 adsorption isotherms

to obtain surface area. TEM and SAED observation was conducted using a JEOL 2010F

TEM/STEM at the Canadian Centre for Electron Microscopy (Hamilton, Ontario, Canada). TEM

samples were prepared by drop casting the dispersions onto holey carbon grids. The images were

processed using Gatan Microscopy Suite: Digial MicrographTM and SAED and FFT images were

indexed using CrysTBox – diffractGUI (Klinger and Jäger, 2015). N2 adsorption isotherms were

measured with a Quantachrome AUTOSORB-1. The samples were outgassed at 200oC under

vacuum for 12 h before the measurement. Surface area was determined by applying Brunauer–

Emmett–Teller (BET) adsorption method on N2 adsorption isotherms in a relative pressure range

of 0.05-0.25.

The isolectric point (IEP) of the LENs was determined by measuring the zeta potential of 0.1 g/L

TiO2 solutions prepared in 10 mM NaCl and adjusted to pHs ranging from 3 to 9. The pH at

187

which the zeta potential reached 0 was designated the IEP of the material. All samples were

prepared in triplicate and the zeta analyzer made four measurements of each sample.

The formation of hydroxyl radicals by P25 and the two LENs was investigated using a simple

fluorescence-based method as described by Arlos et al. (2016). Briefly, hydroxyl radicals convert

TPA to hydroxyterephthalic acid (HTPA), which fluoresces at approximately 425 nm when

excited by wavelengths between 310 and 320 nm. Thus, the presence of HTPA after

photocatalytic treatment implies that hydroxyl radicals were formed during photocatalysis.

Triplicate 50 mL samples of 0.5 mM TPA in 6 mM of NaOH were dosed with 0.02 g/L of TiO2

and exposed to UVA LED light for 0, 1, 2, 5, and 10 minutes, corresponding to UV doses

(fluence) of 0, 0.4, 0.6, 1.5, 2.9, and 4.4 J/cm2. The treated samples were filtered through a 0.45

micron polyethersulfone (PES) filter to remove the TiO2 from solution and analyzed for the

presence of HTPA. Previous research has suggested that approximately 80% of the total

hydroxyl radicals present in solution will react with TPA to form HTPA (Ishibashi, 2000), and

thus the concentration of HTPA in the treated solution can be assumed to be a conservative low

estimate of the total number of hydroxyl radicals formed during the photocatalytic treatment.

7.2.4 Settling and Filtration

A standard bench top filtration apparatus equipped with a PES lab filter with a pore size of 0.8

m was used to test the filterability of the three nanomaterials using a modified version of the

time to filter test (Standard Methods 2710-H). The apparatus was connected to a vacuum pump

set to provide 34 kPa (4.9 psi) of pressure on the permeate side of the filter. Each PES filter was

flushed with 100 mL of purified water before being used to filter a 50 mL sample. The flux of

purified water under these conditions was 6.8 m/h, which is within the 5 to 15 m/h range

expected for granular media filters in drinking water treatment plants (MWH, 2012). Filtration

samples were prepared with 0.25 g/L of TiO2 and mixed in the dark for 1 minute ahead of

filtration. The time required to filter each sample was recorded and normalized to the amount of

time required to filter 50 mL of purified (MilliQ) water through the apparatus to yield a filtration

index value. The flux of water through the filter was also calculated based on the volume of

water filtered and the exposed surface area of the membrane (7.1 x 10-4 m2). All filtration

experiments were conducted in triplicate.

188

The settling characteristics of P25 and the two LENs were evaluated in MilliQ water and the two

raw river water samples. A 10 g/L stock solution of each material was prepared and sonicated for

five minutes. Aliquots of the stock solution were added to 40 mL of the chosen water matrix and

placed in a Hach turbidimeter. The initial turbidity of the sample was recorded at the beginning

of the text and every ten minutes for two hours. The turbidity results were converted to TiO2

concentration using material-specific calibration curves prepared with standards ranging from 10

mg/L to 3,000 mg/L of TiO2.

The results of these simplified separation and filtration tests cannot be used to predict the long

term behavior of full-scale sedimentation basins or membrane filters, however, they do provide

some indication of the relative settleability and filterability of the LENs used in this study.

7.2.5 NOM and Dye Degradation Experiments

All degradation experiments were conducted in 75 mL continuously mixed batch reactors filled

with 50 mL of 10 mg/L methylene blue dye or raw OTB or OTW water. The reactors were dosed

with 0.25 g/L of TiO2 and exposed to UVA LEDs with an average irradiance of 6.25 mW/cm2

for 0, 5, 15, 30, 45, or 60 minutes, corresponding to approximate UV doses (fluence) of 0, 1.5,

4.4, 8.8, 13.2, and 17.6 J/cm2. The treated river water samples were filtered through a 0.45 m

PES filter to remove the TiO2 nanomaterials and analyzed for UV254, DOC, chlorine demand,

THMfp, and HAAfp. Chlorine demand, THM formation potential (THMfp), and haloacetic acid

formation potential (HAAfp) were assessed at uniform formation conditions (UFC) as described

by Summers et al. (1996). The THMs and HAAs formed during this process were extracted

according to Standard Method 6232 B and Standard Method 6251 B (APHA, 2005) and analyzed

using an Agilent 7890B GC-ECD. The treated methylene blue samples were centrifuged to

remove the TiO2 nanomaterials and then analyzed for absorbance at 665 nm to determine the

concentration of methylene blue remaining in solution. All NOM removal experiments were

conducted in quadruplicate with one replicate being used for chlorine demand and three being

used for DBPfp determination.

7.2.6 Calculations

The average settling rate was calculated using the following equation:

189

Settling Rate (mg

min) =

Co-C

t (7.1)

In an effort to elucidate the effects of surface area and the crystallinity of the nanomaterials the

reaction rate constants were normalized to the available surface area using the following

equation:

𝑘𝑛𝑜𝑟𝑚 =𝑘

𝑆𝐴 × 𝐷𝑇𝑖𝑂2 × 𝑉𝑠𝑎𝑚𝑝𝑙𝑒 (7.2)

where SA is the BET surface area of the nanomaterial (m2/g), DTiO2 is the dose of TiO2 added

(g/L), and Vsample is the volume of the sample (L).

Electrical energy per order (EEO) is currently listed as a “figure of merit” for the evaluation of

advanced oxidation processes by IUPAC (Collins and Bolton, 2016). The EEO of a given

process can be calculated using Equation 7.3, where P is the power dissipated by the treatment

process (kW), V is the volume of water treated in the experiment (L), Ci is the original

concentration of the contaminant, Cf is the final concentration of the contaminant, and t is the

time required to achieve Cf (min).

𝐸𝐸𝑂 =1000 𝑃 𝑡


(7.3)

All statistical analyses were conducted at the 95% confidence level.

Results

7.3.1 Characterization of Engineered TiO2 Nanomaterials

The LENs in this study were characterized in terms of size (TEM), crystal phase structure

(SAED), isoelectric point (zeta potential), surface area (BET isotherm testing), and hydroxyl

radical production. The results are summarized in Table 7.2. The temperature setpoints used

during the LEN synthesis process had important effects on many of these parameters. These have

been discussed in in the context of previous research on TiO2-based LENs.

190

Table 7.2 Shape, size, and surface characteristics of LENs

Material P25 NB 550 NB 700

Shape Spherical Linear Linear

Diameter 21 nm1 -- --

Length -- 0.5 – 2 m 0.5 – 2 m

Width -- 20 – 200 nm 20 – 200 nm

Surface Characteristics -- Speckled Smooth

Predominant Crystal Phase 75% Anatase, 25% Rutile2 Anatase and TiO2 (B) Anatase

BET Surface Area 57 m2/g 30 m2/g 18 m2/g

Isoelectric Point (IEP) pH 6 to 6.5 pH 4 to 4.5 pH 4 to 4.5

k·OH 0.62 ± 0.03 mM/min 0.10 ± 0.01 mM/min 0.74 ± 0.06 mM/min

k·OH (normalized) 10.9 ± 0.5 mM/min/m2 3.5 ± 0.3 mM/min/m2 41.4 ± 3.2 mM/min/m2

1Sigma Aldrich

2Ohtani et al., 2010

Survey studies by Yuan and Su (2004), Wong et al. (2011), Qamar et al. (2008), and others have

established that the precursor materials, hydrothermal synthesis temperature, extent and method

of post synthesis cleaning and ion exchange, and calcination temperature have important effects

on the final products. The choice of precursor materials and hydrothermal temperature affects the

overall size and aspect ratio of the linear nanomaterials, with higher temperatures generally

resulting in larger materials (Yuan and Su, 2004). The washing, ion exchange, and calcination

steps affect the surface and crystalline structures of the materials, and thus their photocatalytic

properties (Qamar et al., 2008; Ali et al., 2016).

The TEM images of the LENs in Figure 7.1 show that both nanomaterials were roughly

rectangular or belt-like in shape. Individual belts ranged from 20 nm to 200 nm in width and

from 500 nm to multiple microns in length. The LENs differed in terms of their surface

topography. The NB 700 particles were smooth and the edges appeared rounded while the NB

550 particles appeared speckled with raised bumps and had sharply defined edges. Zheng et al.

(2010) reported similar surface characteristics for LENs calcined at similar temperatures and

attributed them to the presence of different phases of TiO2, namely anatase and TiO2(B).

191

Figure 7.1 Characterization of NB 550 (A) and NB 700 (B) via TEM and SAED. Figure

created by Robert Liang from the University of Waterloo using results

obtained at McMaster University

Anatase is widely held to be the most photoactive form of TiO2 but some researchers have,

however, reported that mixed phase anatase/TiO2(B) LENs can be even more effective than pure

anatase materials (Zheng et al., 2010). The predominant crystal phases present in the two LENs

in the current study were determined using SAED and TEM analysis. The SAED images of NB

550 (Figure 7.1A1) reveal that NB 550 contained predominantly anatase, however, the TEM

image (Figure 7.1A2) indicated that TiO2(B) crystalline grains were also present with d-spacing

of 0.36 nm and 0.57 nm corresponding to the (011) and (10-1) planes, respectively. The anatase

grains indicate d-spacings for 0.34 nm and 0.38 nm, corresponding to the (101) and (003) planes

of anatase. The interface between anatase and TiO2(B) phases had similar lattice parameters. The

A1: NB 550

B1: NB 700

A2: NB 550

B2: NB 700

192

(101) planes in anatase and (011) planes in TiO2(B) also matched closely. When the calcination

temperature increased to 700oC, the TiO2(B) was converted to anatase as shown in Figure 7.1B.

The d-spacings of NB 700 (Figure 7.1B2) were 0.35 nm and 0.45 nm, which match the (101) and

(002) planes of anatase.

BET isotherm analysis showed that both LENs had less surface area than P25 nanoparticles (30

m2/g and 18 m2/g vs. 57 m2/g). The higher surface area observed for the LENs calcined at 550oC

compared to those calcined at 700oC is consistent with previous studies (Qamar et al., 2008;

Zheng et al., 2010; Ali et al., 2016). In theory, nanomaterials with higher surface area should be

more effective for contaminant adsorption.

The isoelectric points (IEPs) of the LENs occurred at pH values between 4 and 4.5, which is

below that of P25 (6 to 6.5). Previous work has established that the pH of the water has a strong

effect on NOM adsorption to P25 and that adsorption is favoured when pH of the water is below

the IEP the nanomaterial (Mwaanga et al., 2014; Gora and Andrews, 2017). The implication of

this finding is that the LENs will be negatively charged within the pH range commonly found in

natural surface water sources (6.5 to 8.5) and thus may repel negatively charged water

constituents, thus slowing or preventing their degradation. The IEPs of the nanomaterials may

also have had an impact on the degree to which they agglomerated in each water source.

Nanoparticle agglomeration and its effect on surface area can also contribute to the changes in

adsorption efficiency and photocatalytic degradation observed at different pHs and in the

presence of ions and NOM. Nanomaterial agglomeration and subsequent decrease in the overall

available surface area is most likely to occur when the pH is near the isoelectric point/point of

zero charge because at this pH repulsive forces between individual particles are at a minimum

(Liu et al., 2013).

P25 and the two LENs were also evaluated in terms of hydroxyl radical production. As shown in

Figure 7.2, P25 and NB 700 produced more hydroxyl radicals and photogenerated holes than NB

550. This was also reflected in the k·OH values for each material, which are summarized in Table

7.2. The k·OH for NB 700 was 0.74 ± 0.06 mM/min, which was slightly but significantly higher

than that of P25, which was 0.62 ± 0.03 mM/min, at the 95% confidence level. NB 550 lagged

behind the other two materials with a k·OH of only 0.10 ± 0.01 mM/min. NB 700’s superior

hydroxyl radical formation ability relative to P25and NB 550 was further confirmed by the

193

normalized reaction rate constants for the three materials, which are also shown in Table 7.2. The

normalized k·OH for NB 700 was 41.4 ± 3.2 mM/min/m2 while that of P25 was 10.9 ± 0.5

mM/min/m2, indicating that NB 700 produced nearly four times as many moles of HTPA per

unit area as P25. This suggests that NB 700 was more effective at harnessing the available light

energy to drive the formation of hydroxyl radicals.

Figure 7.2 Hydroxyl radical (·OH) radical production by P25, NB 550, and NB 700.

Error bars represent the standard deviation from the mean (n = 3).

7.3.2 Filtration and Settling

Filtration is commonly used to separate TiO2 from water in bench-scale experiments and is a

promising separation option for full-scale water treatment with TiO2 nanomaterials. Filtration

tests were performed with P25 and the new LENs as described in Section 7.2.4. Figure 7.3

shows the filtration index of each raw water and water containing 0.25 g/L of the three

nanomaterials. The filtration indexes of the two raw river water samples were nearly equal to 1,

indicating that they did not present a significant barrier to filtration relative to MilliQ water. The

OTW water had a slightly higher filtration index (1.22 ± 0.07) than the OTB water (1.11 ± 0.04),

likely owing to its higher turbidity and organic content. The raw water flux values for OTB and

OTW water were 6.1 m/h and 5.5 m/h respectively, which is within the accepted range for

granular media filtration but well above that achieved by microfiltration membranes (MWH,

2012).

0

2

4

6

8

10

12

14

P25 NB 550 NB 700

·OH

(u

M)

0 min

1 min

2 min

5 min

10 min

15 min

194

Figure 7.3 Filtration indexes of three TiO2 nanomaterials suspended in purified (MQ)

water, Otonabee River (OTB) water, and Ottawa River (OTW) water

Irrespective of the water matrix used, water flowed through the lab filter more quickly when the

water contained NB 550 or NB 700 rather than P25 nanoparticles. In fact, in the tests conducted

with the two natural water samples there was no statistical difference at the 95% confidence level

between the filtration indexes of the raw water samples and those of the water samples

containing NB 550 or NB 700. The LENs were larger than the membrane pores in at least one

dimension, and as such, may have been more likely to be retained on the surface of the

membrane during filtration than the much smaller P25 nanoparticles, which may have been more

likely to enter and clog the pores of the membrane. Indeed, the samples containing P25 had

filtration indexes three to six times greater than those containing the LENs. This is in agreement

with the findings of Zhang et al. (2009), who used membrane filtration to separate P25

nanoparticles and two LENs from water. Based on the results of membrane fouling tests and

SEM imaging, they hypothesized that the P25 nanoparticles were becoming lodged in the pores

of the membrane during filtration, constricting them and increasing the resistance to flow while

the LENs formed a looser, more porous cake on the membrane surface that had less of an effect

on flow. The experimental results of the current study suggest that similar phenomena were at

play here and clearly demonstrate the superior filterability of the LENs relative to standard P25

nanoparticles in purified water and both surface water matrices. In addition, there was also likely

some degree of nanoparticle agglomeration, a complex phenomenon that is influenced by the pH,

0

1

2

3

4

5

6

7

8

9

MQ OTB OTW

Fil

tra

tio

n I

nd

ex

Raw Water

P25

NB 550

NB 700

195

ionic strength, and organic content of the matrix as well as by the chemical and physical

properties of the nanomaterial in question (French et al., 2009; Liu et al., 2013; Loosli et al.,

2015; Li et al., 2016) and which was not explored in detail in this study (see Appendix H).

Nonetheless, the experimental results clearly demonstrate the superior filterability of the LENs

relative to P25 nanoparticles.

Sedimentation is potentially a more economical alternative to filtration for solids removal, so the

settleability of the LENs was evaluated using simple settling tests. The tests were performed over

a two hour period but in all cases the majority of the observed settling occurred within the first

thirty minutes of the test. The average rate of settling over the first thirty minutes of each test are

presented in Figure 7.4. The rate of settling was influenced by both the characteristics of the

nanomaterials and the water matrices. P25 settled more slowly than the two LENs in MilliQ

water and OTW water, but its settling rate was statistically indistinguishable from theirs in OTB

water. This surprising finding may indicate the agglomerates formed by the P25 nanoparticles in

this water matrix were larger or denser than those formed in the other water samples.

Figure 7.4 Average settling rates of TiO2 nanomaterials suspended in MilliQ (MQ)

water, Otonabee River (OTB) water, and Ottawa River water (OTW)

All three materials settled fastest in the OTB water, which contained higher levels of Ca2+ and

more ions overall than the MilliQ or OTW water whereas the soft, NOM laden OTW hindered

the settling of all three materials. As observed in the tests conducted in MilliQ water, in the

0

1

2

3

4

MQ OTB OTW

Set

tlin

g R

ate

(m

g T

iO2/m

in)

P25

NB 550

NB 700

196

absence of NOM and divalent ions, the materials settled out roughly based on nanoparticle size.

The conductivity of the OTB water was nearly three times that of the OTW water, indicating that

it contained a higher concentration of ions. DVLO theory predicts that increasing levels of ions

can minimize the repulsive electrostatic forces that keep particles from coming together,

allowing van der Waals forces to dominate and encouraging greater agglomeration (Hotze et al.,

2010). Additionally, the presence of calcium ions in the water matrix has been shown to increase

the apparent IEP of TiO2 nanomaterials (Liu et al., 2013) as well as the size of their agglomerates

(French et al., 2009; Zhang et al., 2009), while humic acid (a major component of NOM) has

been shown to have the opposite effects (Thio et al., 2011; Liu et al., 2013; Li et al., 2016). Some

or all of these phenomena were likely at play during the experiments presented in this study.

7.3.3 Degradation of Methylene Blue Dye

The P25 nanoparticles and the two LENs readily degraded methylene blue dye when exposed to

UVA light. P25 and NB 700 achieved over 95% dye decolorization within 30 minutes whereas

NB 550 achieved an average of only 83% degradation after a full hour of irradiation (Figure 7.5).

Figure 7.5 Degradation of methylene blue dye by P25 nanoparticles and two LENs

The degradation data fit well to a pseudo-first-order degradation reaction model for all three

materials as shown in Table 7.3. As others have pointed out (Mills, 2012), the underlying

reaction mechanisms governing the photocatalytic degradation of methylene blue and other large

-3.5

-3

-2.5

-2

-1.5

-1

-0.5

0

0 10 20 30 40 50 60

log (

C/C

o)


P25 NB 550 NB 700

197

organic molecules are likely to be complex despite their apparent fit to a first order degradation

model or the Langmuir-Hinshelwood model. Given the excellent fits obtained under the

conditions used in this study, the rate constants calculated from the apparent first order model

provided a convenient point of comparison between the materials. NB 700 had the fastest

degradation rate followed by P25 nanoparticles and finally by NB 550. This is reflected in the

first order reaction rate constants for the different materials, which ranged from 0.013 min-1 for

NB 550 to 0.060 min-1 for NB 700.

Table 7.3 Reaction parameters for first order degradation of methylene blue dye by

P25 nanoparticles and LENs

k knorm R2

min-1 min-1m-2

P251 -0.047 ± 0.004 -0.066 ± 0.006 0.98

NB 550 -0.013 ± 0.002 -0.033 ± 0.004 0.94

NB 7001 -0.060 ± 0.008 -0.267 ± 0.017 0.95

1Does not include t = 60 min

The normalized reaction rate constants show even more definitively that the superior dye

removal ability of NB 700 relative to P25 is due to its photocatalytic activity, which is a function

of its crystallinity. NB 700 is 100% anatase, the most photocatalytically active form of TiO2,

whereas P25 is only 70 to 85% anatase (Ohtani et al., 2010)

7.3.4 Degradation of Natural Organic Matter (Dissolved Organic Carbon and UV254)

Natural organic matter (measured as DOC and UV254) was adsorbed and degraded by all three

TiO2 nanomaterials under similar conditions as for MB. Results are shown in Figure 7.6. During

the photocatalytic portion of the treatment UV254 decreased more quickly than DOC regardless

of the TiO2 material used. This discrepancy may indicate some preference for aromatic NOM but

is also a function of the parameters themselves: DOC captures all of the original NOM

compounds along with the intermediate organic products of their degradation whereas UV254

measures only that portion of NOM that contains aromatic structures. These aromatic structures,

along with other unsaturated bonds, are easier for ROS to disrupt and as such are the first to be

broken, leading to an overall decrease in UV254.

198

Figure 7.6 Degradation of DOC and UV254 from (A) Otonabee River water and (B)

Ottawa River water (B) by P25 nanoparticles and two LENs

-2

-1.8

-1.6

-1.4

-1.2

-1

-0.8

-0.6

-0.4

-0.2

0

0 10 20 30 40 50 60lo

g (

C/C

o)

Irradiation Time (min) A

P25 - DOC 550 - DOC 700 - DOC

P25 - UV254 550 - UV254 700 - UV254

-2

-1.8

-1.6

-1.4

-1.2

-1

-0.8

-0.6

-0.4

-0.2

0

0 10 20 30 40 50 60

log (

C/C

o)

Irradiation Time (min) B

P25 - DOC 550 - DOC 700 - DOC

P25 - UV254 550 - UV254 700 - UV254

199

NB 550 was less effective for DOC and UV254 removal than P25 nanoparticles or NB 700. The

NB 700 was particularly effective and achieved 100% removal of both UV254 and DOC from

the OTW water within 60 minutes of irradiation, indicating that all of the oxidizable NOM

present in the sample had been mineralized. Decreases in both DOC and UV254 occurred more

slowly in the OTB water than in the OTW water, likely because the former contained higher

levels of known ROS scavengers including chloride (11.5 ± 1.3 mg/L vs. 3.3 ± 0.9 mg/L) and

bicarbonate (87 ± 7 mg/L as CaCO3 vs. 28 ± 6 mg/L as CaCO3). Alternatively or additionally, the

presence of higher concentrations of Ca2+ and other ions in the OTB water compared to the OTW

water (e.g. conductivity of OTB water = 214 ± 19 S/cm, conductivity of OTW water = 81 ± 13

S/cm) may have led to an increased degree of aggregation accompanied by an overall decrease

in available surface area in this water source. Finally, as suggested by its higher SUVA value

(3.7 ± 0.3 L/m.mg vs. 2.6 ± 0.4 L/m.mg), the OTW NOM was more aromatic in nature than the

OTB NOM and as such may have been more vulnerable to oxidation, as has been observed by

other researchers (Liu et al., 2008; Liu et al., 2010).

Adsorption played a minor but notable role in NOM removal in this study. P25 removed 14 ± 5%

of the DOC and 30 ± 2% of the UV254 from the OTB water and 17 ± 6% of the DOC and 20 ±

3% of the UV254 from the OTW water. The LENs adsorbed less NOM; both removed

approximately 6% of DOC and 10% of UV254 from both water sources. The superior adsorptive

ability of the P25 nanoparticles is partly explained by available surface area: The BET surface

area of the P25 nanoparticles was 57 m2/g while those of NB 550 and NB 700 were 30 m2/g and

18 m2/g, respectively. Indeed, when DOC removal was normalized to surface area both P25 and

NB 700 removed 0.08 mg DOC/m2 from OTW water via adsorption (see Table 7.S.1 in the

supplementary material). Water matrix effects also appear to have played a role in adsorption,

particularly with respect to P25 nanoparticles, which removed UV254 from the OTB water more

effectively than from the OTW water. Researchers such as Mwaanga et al. (2014), Erhayem and

Sohn (2014), Sun and Lee (2012), and Liu et al. (2013) have explored the effects of various ions

on the adsorption of NOM by TiO2 nanomaterials and have found that the presence of divalent

ions, in particular calcium, can encourage NOM adsorption to TiO2. OTB water contains more

calcium than OTW water, which may explain why the P25 nanoparticles were able to remove

more UV254 from the former. A more detailed assessment of adsorption effects was beyond the

scope of this study but it is worth further investigation.

200

As was observed with methylene blue degradation, the degradation of both DOC and UV254

was a good fit to a simple pseudo-first order degradation model. The reaction rate constants

shown in Table 7.4 confirm that DOC degradation by P25 and NB 700 proceeded more quickly

for NB 550, and occurred more slowly in the OTB water than in the OTW water. The normalized

reaction rate constants, which are also shown in Table 7.4, show even more definitively that the

superior NOM degradation ability of NB 700 relative to P25 was due to its ability to harness

light to generate ROS and photogenerated holes, which is a function of its crystallinity, rather

than its surface area.

Table 7.4 First order reaction rate constants and normalized reaction rate constants

for DOC and UV254 removal

DOC UV254

k k norm R2 k k norm R2

min-1 min-1m-2 min-1 min-1m-2

OTB

P25 -0.004 ± 0.000 -0.005 ± 0.001 0.94 -0.018 ± 0.001 -0.026 ± 0.002 0.96

NB 550 -0.002 ± 0.000 -0.004 ± 0.001 0.88 -0.005 ± 0.000 -0.013 ± 0.001 0.96

NB 700 -0.005 ± 0.001 -0.024 ± 0.002 0.97 -0.018 ± 0.002 -0.080 ± 0.008 0.97

OTW

P25 -0.008 ± 0.001 -0.011 ± 0.001 0.93 -0.022 ± 0.001 -0.031 ± 0.002 0.98

NB 550 -0.002 ± 0.000 -0.004 ± 0.001 0.88 -0.012 ± 0.001 -0.032 ± 0.004 0.94

NB 700 -0.017 ± 0.003 -0.074 ± 0.015 0.93 -0.027 ± 0.002 -0.120 ± 0.008 0.97

In this study, the main energy input to the experimental apparatus was the UVA LEDs used to

illuminate the samples. Each LED had a rated power demand of 2.7 W. This, along with the

sample volume (50 mL), the time of irradiation (60 minutes), and the concentration of DOC

measured in the raw and treated samples, was inputted into Equation 7.3 (see Section 7.2.6) to

determine the EEO of each material. EEO values for DOC removal by the three nanomaterials

ranged from 37 kWh/order/m3 for NB 700 in OTW water to over 500 kWh/order/m3 for NB 550

in both water matrices. EEO values were higher in OTB, which contained a higher concentration

of ROS scavengers, than OTW water. Yen and Yen (2015) reported an EEO of 30 kWh/order/m3

for DOC degradation by a UV/H2O2 system employing a low pressure UV lamp (maximum

irradiance at 254 nm) and a 10 mg/L dose of H2O2. This is close to the EEO value calculated for

201

NB 700 in the OTW water, suggesting that a UV/TiO2 system employing NB 700 and UVA

LED light might prove to be competitive with UV/H2O2 under certain experimental conditions

and in some water matrices.

7.3.5 Removal and Degradation of Disinfection Byproduct Precursors

In this study, the total THMfp of both the OTB water and the OTW water initially increased

when the samples were exposed to the UVA LED lights irrespective of the TiO2 nanomaterial

used before eventually decreasing at longer irradiation times (Figure 7.7). The rate and extent of

this increase was not constant: THMfp peaked between 5 and 15 minutes of irradiation

depending on the nanomaterial and water matrix used. However, after 60 minutes of irradiation,

the THMfp of the treated water was well below that observed at shorter treatment times. Both

P25 nanoparticles and NB 700 reduced the THMfp of the raw water by over 80% within sixty

minutes of irradiation.

All of the nanomaterials tested exhibited the initial peak followed by eventual THMfp reduction,

but the rates and end points of the THMfp increases and decreases were different, suggesting that

the three materials may have interacted with different components or to different extents with the

raw water NOM. The initial increase in THMfp observed upon irradiation was likely related to

the formation of reactive intermediates during the photocatalytic degradation process (Gora and

Andrews, 2017; Liu et al., 2010). As irradiation time was increased, these reactive intermediates

would themselves have been broken down by the photocatalytic degradation process, resulting in

decreased THMfp. Liu et al. (2010) noted a similar trend in one of the two Australian surface

waters they treated with commercial P25 nanoparticles and UV light. They hypothesized that at

short irradiation times larger NOM molecules were partially broken down such that more

reactive sites became available for chlorine attack, resulting in increased THM formation upon

chlorination. This phenomenon may also have contributed to increased THMfp at short

irradiation times in the current study.

202

Figure 7.7 Reduction in the formation of trihalomethanes in two water matrices after

treatment by (A) P25 in OTB water, (B) NB 550 in OTB water, (C) NB 700 in OTB water,

(D) P25 in OTW water, (E) NB 550 in OTW water, (F) NB 700 in OTW water. Error bars

represent the 95% confidence interval of the mean.

0

50

100

150

200

250

300

350

C 0 5 15 30 45 60

TH

Mfp

(

g/L

)


0

50

100

150

200

250

300

C 0 5 15 30 45 60

TH

Mfp

(

g/L

)


0

50

100

150

200

250

300

C 0 5 15 30 45 60

TH

Mfp

(

g/L

)


0

100

200

300

400

500

600

C 0 5 15 30 45 60

TH

Mfp

(

g/L

)


0

100

200

300

400

500

600

C 0 5 15 30 45 60

TH

Mfp

(

g/L

)


0

100

200

300

400

500

600

C 0 5 15 30 45 60

TH

Mfp

(

g/L

)


A

B

C

D

E

F

Otonabee River Ottawa River

Total THMfp TCMfp BDCMfp

203

In this work, THMfp refers to the sum of four trihalomethane species commonly formed when

NOM from surface water interacts with chlorine. Trichloromethane (TCM) and

bromodichloromethane (BDCM) were the predominant THMs formed upon chlorination of the

raw and treated water samples. Figure 7.7 shows how the relative concentrations of TCM and

BDCM changed over time in the treated samples. In general, the concentration of TCM

decreased continuously with increasing irradiation time but at shorter irradiation times this

reduction was accompanied by a gradual increase in BDCM. As the treatment proceeded, the

BDCM precursors were eventually degraded, contributing to the reduction of overall THMfp.

This is similar to results presented by Gerrity et al. (2009), who explored the degradation of

THM precursors in real surface water sources using a pilot scale UV/TiO2 system. Bromine is

more likely to interact with smaller and more hydrophilic NOM compounds (Kitis and Karanfil,

2002; Liang and Singer, 2003), and previous studies have found that TiO2 photocatalysis can

degrade large hydrophobic NOM compounds into smaller, more hydrophilic ones (Gora and

Andrews, 2017; Liu et al., 2010), which may have been more likely to interact with both chlorine

and bromine upon chlorination to form BDCM.

The composition of the water matrix also had important effects on the degradation of THM

precursors. All three TiO2 nanomaterials reduced the THMfp of OTW water more effectively

than that of the OTB water. As was noted for DOC and UV254, factors that may explain this

discrepancy include the presence of ROS scavengers in the OTB water; reduction in available

surface area due to higher levels of agglomeration driven by a decrease in electrostatic repulsion

in the ion-rich OTB water; increased NOM oxidation in the OTW water related to the presence

of higher concentrations of iron and copper in this matrix (see Table 7.1); and the characteristics

of the NOM, including THM precursors, in each water source. Specifically, the effectiveness of

NB 700 was strongly impacted by the water matrix: it was far more effective for THM precursor

removal in the OTW water matrix than in the OTB water matrix. The former had much lower

alkalinity, a measure of bicarbonate, which is a hydroxyl radical scavenger. As demonstrated in

the hydroxyl radical experiments described in Section 7.3.1 and Figure 7.2, NB 700 was

particularly effective for hydroxyl radical production. As such, its oxidative efficacy may have

been more likely to be impacted by the presence of hydroxyl radical scavengers than that of the

other nanomaterials, which removed NOM predominantly via other oxidative or, in some cases,

adsorptive pathways.

204

Adsorption was less effective for THM precursor removal than it was for DOC and UV254

removal. Adsorption with P25 nanoparticles reduced the THMfp of the OTW water by 28 ± 9%

while adsorption with NB 550 reduced the THMfp of the OTB water by 9 ± 6%. None of the

other material / water matrix combinations resulted in statistically significant removal of THM

precursors via adsorption.

The EEO concept assumes first order degradation kinetics and in most cases in this study,

THMfp removal did not follow a first order reaction model. The exceptions were P25 and NB

700 in the OTW water, and this only after the peak that occurred between 5 and 15 minutes of

irradiation. The EEO value for THMfp reduction by UVA/TiO2 treatment with NB 700 in the

OTW water (51 kWh/order/m3) was comparable to those UV/H2O2 treatment with 10 mg/L of

H2O2 (44 kWh/order/m3) as described by Yen and Yen (2015), indicating that under some

conditions and with some TiO2 materials UV/TiO2 may prove to be competitive with UV/H2O2

for THMfp reduction. This should be confirmed by comparing the two processes under a variety

of experimental conditions and in the same water matrices.

For the most part, the HAAfp results, shown in Figure 7.8, followed the same trends as the

THMfp results. Like THMfp, the term HAAfp refers to the likelihood that a water sample will

form a suite of haloacetic acids. In this study, only dichloroacetic acid (DCAA) and

trichloroacetic acid (TCAA) were formed at levels above 5 g/L upon chlorination. For all three

materials, the overall HAAfp of both water sources decreased slightly during the dark adsorption

step, increased at short irradiation times, and eventually decreased as irradiation time increased.

There were, however, important differences in HAA precursor removal between the two water

matrices and the three materials.

In the OTB experiments (Figure 7.8 A-C), P25 reduced the overall HAAfp of the water from

74.9 ± 9.0 g/L to 45.5 ± 6.4 g/L after 60 minutes of irradiation. When NB 550 and NB 700

were used, the overall HAAfp of the water after 60 minutes of irradiation was equal to that of the

untreated raw water. This surprising finding does not adequately account for the effects of these

nanomaterials on the overall HAAfp of the water at shorter treatment times. In both cases, the

overall HAAfp of the water increased between 0 and 15 minutes of irradiation but decreased

thereafter. The increases in overall HAAfp can be attributed to the initial increase in both

205

DCAAfp and TCAAfp between 0 and 15 minutes. This was followed by a gradual reduction in

TCAAfp over time as TCAA precursors were degraded by further treatment.

Photocatalytic degradation of HAA precursors occurred more readily in the OTW water matrix

(Figure 7.8 D-F), particularly when NB 700 was employed as the photocatalyst. After 60 minutes

of irradiation the HAAfp of the water treated with NB 700 was reduced from 112.9 ± 7.1 g/L to

12.1 ± 7.1 g/L. P25 was nearly as effective as NB 700 in this water matrix, reducing the HAAfp

of the water from 110.4 ± 12.5 g/L to 30.1 ± 12.5 g/L after 60 minutes of irradiation. In both

cases, the DCAAfp of the water increased at shorter irradiation times before decreasing at longer

irradiation times. TCAAfp reduction occurred more quickly than DCAAfp reduction, and for

both P25 and NB 700 the TCAAfp of the water was reduced to below the detection limit after 60

minutes of irradiation, indicating that the photocatalytic treatment was particularly effective for

the removal of TCAA precursors. TCAA precursors are more hydrophobic than DCAA

precursors (Liang and Singer, 2003), and, as shown in earlier work (Gora and Andrews, 2017;

Liu et al., 2010), hydrophobic and aromatic NOM is preferentially degraded by TiO2

photocatalysis. Other researchers have also observed preferential removal of TCAA precursors

over DCAA precursors in AOP systems. These include Toor and Mohseni (2008); who linked

increased DCAAfp in water treated with UV/H2O2 to the formation of aldehydes, known DCAA

precursors, as a result of the partial degradation of NOM; and Bond et al. (2009), who attributed

the increase in DCAAfp to the transformation of hydrophilic NOM compounds (amino acids)

into DCAA precursors via oxidation.

206

Figure 7.8 Reduction in the formation of haloacetic acids in two water matrices after



OTW water. Error bars represent the 95% confidence interval of the mean

Total HAAfp DCAAfp TCAAfp

0

20

40

60

80

100

C 0 5 15 30 45 60

HA

Afp

(

g/L

)


0

20

40

60

80

100

C 0 5 15 30 45 60

HA

Afp

(

g/L

)


0

20

40

60

80

100

C 0 5 15 30 45 60

HA

Afp

(

g/L

)


0

50

100

150

200

250

C 0 5 15 30 45 60

HA

Afp

(

g/L

)


0

50

100

150

200

250

C 0 5 15 30 45 60

HA

Afp

(

g/L

)


0

50

100

150

200

250

C 0 5 15 30 45 60

HA

Afp

(

g/L

)


A

B

C

D

E

F

Otonabee River Ottawa River

207

The effect of the treatment on the formation of individual HAA species was not only

nanomaterial specific but also matrix specific. In the experiments conducted with OTW water,

all three nanomaterials initially increased the amount of DCAA and TCAA precursors in the

water upon irradiation but eventually began to degrade them, resulting in nearly concurrent

reduction of DCAAfp and TCAAfp at longer irradiation times. When P25 and NB 700 were

added to the OTB water and irradiated DCAAfp initially increased or remained constant but this

was not followed by the eventual decrease observed in the OTW experiments. These results

suggest that the two water matrices contain different types of DCAA precursors or that the

degradation of DCAA precursors was in some way inhibited in the OTB water matrix. As was

observed previously for THMfp, NB 700 was more strongly impacted by the presence of

hydroxyl radical scavengers such as bicarbonate (alkalinity) than the other two nanomaterials.

This may be because NOM oxidation by UV/TiO2 treatment with this material proceeded

primarily via hydroxyl radical mediated reactions, which were more likely to be inhibited in the

higher alkalinity water matrix (OTB) than in the lower alkalinity water matrix (OTW).

With the exception of P25 in OTB water, the HAAfp of the adsorption only samples were

statistically indistinguishable from the controls at the 95% confidence level, an indication that

none of the nanomaterials used in this study were reliably able to remove HAA precursors via

adsorption alone at this dose of TiO2. This is in line with previous studies conducted with P25 in

this water source (Gora and Andrews, 2017).

7.3.6 Alternative Measures of System Efficiency: Applied UV Dose and Power per Volume

Another way to track the progress of photocatalytic treatment processes is based on the UV dose,

or fluence, applied to the sample. Ideally, the UV dose would be calculated based on the incident

light throughout the sample, however, as described in Appendix G of this thesis, in the current

study it is unlikely that the UVA LED light applied to the samples penetrated deeply into them.

As a result, it was assumed that the UV dose could be calculated based on the average irradiance

at the surface of the sample. This was calculated to be 4.9 mW/cm2 using a spreadsheet prepared

by Bolton and Linden (2003) as described in Appendix G. This value was multiplied by the

elapsed time (s) to determine the UV dose or fluence (mJ/cm2) at the surface of the sample as

shown in Equation 7.4.

208

UV Dose (mJ/cm2) = Irradiance (mW 𝑐𝑚2⁄ ) × Time (min) × 60 (s 𝑚𝑖𝑛⁄ ) (7.4)

UV dose is less specific to experimental set-up than time and as a result can more easily be

compared to the results of other researchers. It can also be a useful parameter when comparing

different light-based water treatment processes. For example, Autin et al. (2013) demonstrated

that the UV dose (254 nm) required to achieve metaldehyde degradation was equal for UV/TiO2

and UV/H2O2 in the absence of alkalinity and organics. The addition of CaCO3 and NOM

surrogates increased the UV dose required to achieve metaldehyde removal via UV/TiO2 but not

that required for UV/H2O2 treatment. This demonstrated that UV/TiO2 was more likely to be

negatively impacted by the presence of ROS scavengers than UV/H2O2.

Another UV/H2O2 study showed that approximately 3,000 mJ/cm2 of UV light was required to

reduce the THMfp of a Canadian surface water matrix from 238 g/L to 54 g/L (77%) at an

H2O2 dose of 23 mg/L (Toor and Mohseni, 2007). This is well below the UV dose that was

required to achieve a comparable reduction in THMfp from OTW water using NB 700 (~13,000

mJ/cm2) in the current study, indicating that even in a best-case scenario, the UV dose required

to reduce the DBPfp of surface water via UV/TiO2 is unlikely to be comparable to that required

to reduce it via UV/H2O2. It should, however, be noted that Toor and Mohseni did not observe

any significant removal of THM precursors at a fluence of 3,000 mJ/cm2 at a lower H2O2 dose (4

mg/L). Also, their experiments made use of a low pressure UV lamp (max irradiance at 254 nm).

TiO2 can be activated by lower energy wavelengths of up to approximately 380-385 nm and as

such has the potential to be more energy efficient even while requiring a larger UV dose.

An alternative way to compare the efficiency of different treatment systems is to calculate the

power required to remove a given amount of a contaminant.

𝑃𝑜𝑤𝑒𝑟

𝑉𝑜𝑙𝑢𝑚𝑒(𝑘𝑊ℎ 𝑚3⁄ ) =

𝑆𝑦𝑠𝑡𝑒𝑚 𝑃𝑜𝑤𝑒𝑟 𝑅𝑎𝑡𝑖𝑛𝑔 (𝑘𝑊) × 𝑇𝑖𝑚𝑒 (ℎ)

𝑉𝑜𝑙𝑢𝑚𝑒 𝑇𝑟𝑒𝑎𝑡𝑒𝑑 (𝑚3) (7.5)

In this study, the use of UV dose as a parameter hides the main advantage of using UVA LEDs --

the fact that they are far more energy efficient than standard UV germicidal lamps or high

intensity UVA lamps. For example, the study by Autin et al. (2013) took place in a bench-scale

UVC collimated beam apparatus containing four 30 W lamps. This was used to treat a 250 mL

sample and the irradiance at the surface of the sample was 2.23 mW/cm2, thus a UV dose of

3,000 mJ/cm2 corresponded to 22.3 minutes of irradiation and a power per volume of 480

209

kWh/m3. A dose of 3,000 mJ/cm2 in the UVA LED reactor used in the current study corresponds

to an irradiation time of 10.2 minutes and 54 kWh/m3. The UVA LEDs used in the current study

cannot be used for UV/H2O2 because they only emit light at 365 nm, which is not energetic

enough to drive the formation of OH radicals from H2O2.

Figure 7.9 shows the degradation of THM precursors in OTB and OTW water by NB 700 as a

function of time, UV dose, and power per volume.

Figure 7.9 Reduction of THMfp in OTB and OTW water via photocatalysis by 0.25 g/L

of NB 700 irradiated with UVA LEDs (365 nm) as a function of irradiation

time (min), UV dose (J/cm2), and power per treated volume (kWh/m3)

7.3.7 Correlation Between Methylene Blue Degradation, NOM Degradation, and DBPfp

Methylene blue is quickly degraded by TiO2 photocatalysis and as such is often used as a

surrogate parameter for other organic compounds that are more difficult to analyze. Our previous

work (Gora and Andrews, 2015) explored the relationship between methylene blue degradation

and the reduction of DOC and UV254, the two most common DBPfp surrogates used in the

drinking water industry, for P25 and four LENs and found a strong, positive, and significant (p <

0.05) correlation between methylene blue degradation and UV254 and a more modest but still

significant one between methylene blue degradation and DOC. This was also observed in the

0

100

200

300

400

500

0 20 40 60

TH

Mfp

(

g/L

)


OTB

OTW

11.8 18.60

18 360 54

UV Dose (J/cm2)

Power (kWh/m3)

5.8

210

current study. As shown in Table 7.5, the Pearson correlation coefficients between methylene

blue degradation and DOC degradation ranged from 0.74 for NB 700 in OTW water to 0.97 for

NB 550 in OTB water. The correlation between methylene blue degradation and UV254 removal

was much stronger than that with DOC for all three materials, with Pearson correlation

coefficients ranging from 0.91 for NB 700 in OTW water to 0.99 for NB 550 in OTB water.

Table 7.5 Pearson correlation coefficients comparing the degradation of DOC, UV254,

and DBP precursors by three TiO2 nanomaterials in two source waters. Bold

values are significant at the 95% confidence level.

OTB OTW

P25 NB 550 NB 700 P25 NB 550 NB 700

DOC 0.77 0.95 0.74 0.79 0.82 0.77

UV254 0.97 0.97 0.92 0.94 0.97 0.90

THMfp -0.07 -0.97 -0.64 0.32 -0.45 0.50

HAAfp 0.03 -0.16 -0.33 0.60 -0.09 0.52

DOC and UV254 removal are widely used to predict reductions in THMfp because they are

simple, inexpensive surrogate parameters that have been correlated to DBPfp in conventional

water treatment systems (Edzwald et al., 1985). Based on the results obtained in our previous

work (Gora and Andrews, 2015) and in this study, it was hypothesized that it would also be

possible to predict THMfp and HAAfp based on methylene blue degradation. The methylene

blue and DBPfp data presented in the earlier sections of this paper (Section 7.3.3 and Section

7.3.5) clearly demonstrate that methylene blue degradation was not a good predictor of the

effects of the different nanomaterials on DBPfp in either water source because both THMfp and

HAAfp invariably increased relative to the raw water at short irradiation times for all three

materials. This was further confirmed by the general lack of significant correlation between

methylene blue degradation and DBP degradation as shown in Table 7.5. This finding has

important implications for researchers evaluating the safety and effectiveness of potential new

photocatalytic materials for drinking water. Specifically, these results suggest that dyes and other

surrogate compounds are not adequate indicators of a material’s ability to improve the overall

safety of drinking water.

211

DOC removal was significantly correlated to THMfp and HAAfp reduction in the tests

conducted with P25 and NB 700 in OTW water, indicating that in this source water DOC was an

acceptable surrogate for THMfp and HAAfp changes related to the photocatalytic degradation of

NOM by these highly photoactive TiO2 nanomaterials (see Section 7.6 – supplementary

material). DOC was not significantly correlated to THMfp or HAAfp reduction in any of the

other experiments. UV254 was significantly correlated to HAAfp reduction by P25 and NB 700

in the OTW water and to THMfp reduction by NB 700 in OTW water. The lack of a consistent

relationship between the reduction of THMfp and HAAfp and the removal of common NOM

surrogates underscores the fact that that the removal of simple indicator parameters cannot

always be used to predict the removal or reduction of complex parameters such as DBPfp.

Summary and Conclusions

Two TiO2 LENs were characterized based on size, surface characteristics, and crystal structure

and compared to standard commercial P25 TiO2 nanomaterials in terms of their ability to

degrade disinfection byproduct precursors in natural water. The filterability of the three materials

were also evaluated. Both LENs were more quickly removed from purified water and natural

water via filtration than commercial P25 nanoparticles, likely because their larger size prevented

them from becoming stuck in the filter pores.

Although all three materials reduced DOC and UV254 even at short irradiation times, the

THMfp and HAAfp of the treated water initially increased upon irradiation with UVA LED light

irrespective of the material or water source used. The increase in THMfp usually peaked after 5

to 15 minutes of irradiation (UV dose of 1.5 to 2.9 J/cm2, power per volume of 4.5 to 13.5

kWh/m3) and decreased as irradiation time was increased beyond this point. After 60 minutes of

irradiation (corresponding to 17.6 J/cm2 or 54 kWh/m3), one of the LENs, NB 700, removed

more than 90% of the THMfp and HAAfp from one of the water sources. The types of DBP

precursors present in the treated water changed over time as the original NOM compounds were

photocatalytically degraded from the larger, more aromatic precursors of DBPs such as TCM and

TCAA, to smaller, less aromatic precursors of DBPs such as BDCM and DCAA.

212

The EEO required to degrade DOC and THM precursors varied by material and water source and

in some cases was comparable to EEO values for UV/H2O2 reported by others. The EEO results

suggest that a system incorporating NB 700 and UVA LED irradiation may be as energy

efficient as UV/H2O2 in some water matrices, though this should be confirmed in parallel

experiments under a wider range of experimental conditions.

DOC and UV254 removals were not always well correlated to THMfp and HAAfp removal,

possibly due to complications arising from the breakdown of parent NOM compounds into

intermediate compounds that were themselves DBP precursors, particularly at shorter irradiation

times. The results of this study should serve as a caution to researchers looking to quickly

evaluate the photocatalytic properties of novel TiO2 materials to determine whether they can be

incorporated into drinking water treatment processes.

Throughout this study, the characteristics of the water matrix had important effects on the

removal of DBP precursors via degradation and adsorption by the TiO2 nanomaterials. This

finding highlights the need for comprehensive and site-specific evaluation of new engineered

nanomaterials and other advanced oxidation processes ahead of their implementation for

drinking water treatment.

Acknowledgements

The authors would like to acknowledge the assistance of Kennedy Santos, Jim Wang, and

Chuqiao (Kaya) Yuan in the laboratory.

213

References

Abdullah, M., Low, G.K.C., and Matthews, R.W. 1990. Effects of common inorganic anions on

rates of photocatalytic oxidation of organic carbon over illuminated titanium dioxide, Journal of








Bavykin, D.V. and Walsh, F.C. 2010. Titanate and Titania Nanotubes: Synthesis, RSC

Publishing



215

Bond, T., Goslan, E.H., Jefferson, B., Roddick, F., Fan, L., Parsons, S.A. 2009. Chemical and

biological oxidation of NOM surrogates and effect on HAA formation, Water Research, 43,

2615-2622

Butler, E.C. and Davis, A.P. 1993. Photocatalytic oxidation in aqueous titanium dioxide

suspensions: the influence of dissolved transition metals, Journal of Photochemistry and

Photobiology A: Chemistry, 70, 273-283

Chen, H.Y., Zahraa, O., and Bouchy, M. 1997. Inhibition of the adsorption and photocatalytic



Chen, D. and Ray, A. K. 2001. Removal of toxic metal ions from wastewater by semiconductor


214

Gora, S. and Andrews, S. 2017. Adsorption of natural organic matter and disinfection byproduct



Edzwald, J.K., Becker, W.C., and Wattier, K.L. 1985. Surrogate parameters for monitoring

organic matter and THM precursors, Journal AWWA, 77 (4), 122-132

Erhayem, M. and Sohn, M. 2014. Stability studies for titanium dioxide nanoparticles upon


Total Environment, 468-469, 249-257

Franch, M.I., Ayllón, J.A., Peral, J., and Domènech, X. 2005. Enhanced photocatalytic

degradation of maleic acid by Fe(III) adsorption onto the TiO2 surface, Catalysis Today, 101,

245-252

French, R. A., Jacobson, A.R., Kim, B., Isley, S.L., Penn, R.L., and Baveye, P.C. 2009. Influence

of ionic strength, pH, and cation valence on aggregation kinetics of titanium dioxide

nanoparticles, Environmental Science and Technology, 43, 1354-1359

Gora, S.L. and Andrews, S.A. 2015, Adsorption and Photocatalytic Degradation of Methylene

Blue Dye and Natural Organic Matter by Engineered Titanium Dioxide Nanomaterials, 6th

International Water Association (IWA) Specialist Conference on Natural Organic Matter in

Water, Malmö, Sweden



Kasuga, T., Hiramatsu, M., Hoson, A., Sekino, T., and Niihara, K. 1999. Titania nanotubes

prepared by chemical processing, Advanced Materials, 11 (15), 1307-1311

Kent, F., Montreuil, K., Brookman, R., Sanderson, R., Dahn, J., and Gagnon, G. 2011.

Photocatalytic oxidation of DBP precursors using UV with suspended and fixed TiO2, Water

Research, 45, 6173-6180

215

Kitis, M., Karanfil, T., Wigton, A., Kilduff, J.E. 2002. Probing reactivity of dissolved organic

matter for disinfection by-product formation using XAD-8 resin adsorption and ultrafiltration

fractionation, Water Research, 36, 3834-3848

Klinger, M. and Jäger, A.. Crystallographic Tool Box (CrysTBox): automated tools for

transmission electron microscopists and crystallographers. Journal of Applied Crystallography,

48 (6), 2015. doi:10.1107/S1600576715017252.

Krasner, S. 2009. The formation and control of emerging disinfection by-products of health

concern, Philosophical Transactions of the Royal Society 367, 4077-4095

Li, L., Sillanpää, M., and Risto, M. 2016. Influences of water properties on the aggregation and

deposition of engineered titanium dioxide nanoparticles in natural waters, Environmental

Pollution, 291, 132-138

Liang, L. and Singer, P.C. 2003. Factors influencing the formation and relative distribution of

haloacetic acids and trihalomethanes in drinking water, Environmental Science and Technology,

37, 2920-2928

Liao, C-H, Kang, S-F, and Wu, F-A 2001. Hydroxyl radical scavenging role of chloride and





Liu, S., Lim, M., Fabris, R., Chow, C.W.K., Drikas, M., Korshin, G., and Amal, R. 2010. Multi-


organic matter, Water Research, 44, 2525-2532

Liu S. et al. TiO2-coated natural zeolite: Rapid humic acid adsorption and effective

photocatalytic regeneration, Chemical Engineering Science, 105, 46-52 (2014)

Liu, W., Sun, W., Borthwick, A., and Ni, J. 2013. Comparison on aggregation and sedimentation

of titanium dioxide titanate nanotubes and titanate nanotubes-TiO2: Influence of pH, ionic

216

strength, and natural organic matter, Colloids and Surfaces A: Physicochemical Engineering

Aspects, 434, 319-328

Loosli, F., Vitorazi, L., Berret, J-F, and Stoll, S. 2015. Towards a better understanding on



Mills, A. 2012, An overview of the methylene blue ISO test for assessing the activities of

photocatalytic films, Applied Catalysis B: Environmental, 128, 144-149

Mwaanga, P., Carraway, E., and Schlautman, M. 2014. Preferential sorption of some natural

organic matter fractions to titanium dioxide nanoparticles: Influence of pH and ionic strength,


Ng, M. et al. Highly adsorptive and regenerative magnetic TiO2 for natural organic matter

(NOM) removal in water, Chemical Engineering Journal, 246, 196-203 (2014)

Ohtani, B., Prieto-Mahaney, O.O., Li, D., and Abe, R. 2010. What is Degussa (Evonik) P25?

Crystalline composition analysis, reconstruction from isolated pure particles and photocatalytic

activity test, Journal of Photochemistry and Photobiology A: Chemistry, 216, 179-182

Philippe, K., Hans, C., MacAdam, J., Jefferson, B., Hart, J., and Parsons, S. 2010. Photocatalytic

oxidation, GAC, and biotreatment combinations: An alternative to the coagulation of hydrophilic

rich waters?, Environmental Technology, 31 (13), 1423-1434






80-93

Sun, D.D. and Lee, P.F. 2012. TiO2 microsphere for the removal of humic acid from water:


217

Thio, B.J.T., Zhou, D., and Keller, A. 2011. Influence of natural organic matter on the

aggregation and deposition of titanium dioxide nanoparticles, Journal of Hazardous Materials,

189, 556-563

Toor, R. and Mohseni, M.. 2007. UV-H2O2 based AOP and its integration with biological

activated carbon treatment for DBP reduction in drinking water, Chemosphere, 66, 2087-2095

Wong, C.L., Tan, Y.N., and Mohamed, A.R. 2011. A review on the formation of titania nanotube

photocatalysts by hydrothermal treatment, Journal of Environmental Management, 92, 1669-

1680

Yuan, Z-Y and Su B-L 2004. Titanium oxide nanotubes, nanofibres, and nanowires, Colloids


Zhang, Y., Chen, Y., Westerhoff, P., and Crittenden, J. 2009. Impact of natural organic matter





Zheng, D., Andrews, R.C., Andrews, S.A., Taylor-Edmonds, L. 2015. Effects of coagulation on


Water Research 70, 118-129

218


Table 7.S.1 NOM adsorption normalized to available surface area

Parameter Units OTB OTW

P25 NB 550 NB 700 P25 NB 550 NB 700

MB mg MB/m2 0.01 0.03 0.00

DOC mg DOC/m2 0.05 0.04 0.07 0.08 0.06 0.08

UV254

0.0027 0.0013 0.0021 0.0032 0.0025 0.0043

THMfp g THMfp/m2 1.22 1.81 1.41 6.74 2.93 3.58

HAAfp g HAAfp/m2 0.86 0.55 0.00 0.32 2.30 2.50

219

Removal of NOM and Disinfection Byproducts from Drinking Water Using Regenerable Nanoscale Engineered TiO2 Adsorbents

Abstract

Two linear engineered TiO2 nanomaterials (LENs) were synthesized via a simple hydrothermal

method and evaluated as potential regenerable adsorbents for the removal of natural organic

matter (NOM), including disinfection byproduct (DBP) precursors from raw surface water

obtained from two Canadian drinking water treatment plants. The temperature employed in the

final heating step of the synthesis procedure was varied to produce two linear nanomaterials, NB

550 and NB 700. The nanomaterials had similar dimensions but differed in terms of surface

characteristics, surface area, and crystal structure. Unlike the commercial TiO2 nanoparticles,

both LENs were easily removed from the treated water via settling or filtration. The LENs

removed similar amounts of an indicator dye (25% to 30%) but differed in terms of their ability

to remove DBP precursors. NB 550 reduced the trihalomethane (THM) formation potential of

both water sources by up to 40% while NB 700 reduced it by 25% in one water source and 40%

in the other. The adsorption of DOC, UV254, THM precursors, and HAA precursors by

commercial nanoparticles and the LENs fit a modified Freundlich adsorption isotherm model.

When the two new nanomaterials were regenerated via exposure to UVA light, subsequent dye

removal experiments showed no significant reduction in the amount of dye adsorbed over five

regeneration cycles. Regeneration was also successful when the nanomaterials were used to

remove DBP precursors. In most cases, the loss in NOM absorption efficacy was less than 35%

after five regeneration cycles. This loss in adsorption efficacy occurred more quickly for NB

550, the less photoactive of the two materials, and was strongly affected by water source,

suggesting that differences in the amount and type of NOM adsorbed to the photocatalyst and/or

some components of the matrix (e.g. iron, turbidity, alkalinity) may have interfered with

regeneration.

220

Introduction

Adsorption is a well established water treatment process used to remove natural organic matter

(NOM), including disinfection byproduct (DBP) precursors, from drinking water. DBPs are

formed when DBP precursor compounds in the drinking water matrix are exposed to oxidants

such as chlorine during the disinfection step of the overall drinking water treatment process.

Some DBPs are suspected carcinogens and the removal of their precursors ahead of the

disinfection step is an important goal of modern water treatment processes. Existing adsorbents

such as powdered activated carbon (PAC) and granular activated carbon (GAC) are effective for

DBP precursor removal and widely adopted but difficult to regenerate, resulting in the eventual

need for media replacement. Other treatment strategies such as coagulation and flocculation rely

on single use aluminum and ferric salts to destabilize and remove NOM and DBP precursors via

adsorption and precipitation. In large scale water treatment systems, the spent coagulation

chemicals, usually referred to as residuals, are dewatered onsite and then disposed of in landfills.

In smaller systems, the residuals might be released directly to the environment or directed to the

wastewater collection system.

Although most research on titanium dioxide (TiO2) has focused on its use in advanced oxidation

processes (AOPs), it has both adsorptive and oxidative abilities. As with other AOPs, the

degradation of complex organic molecules via TiO2 photocatalysis is a multistep process and full

mineralization may not be achieved within an acceptable treatment time frame. Not only is this

undesirable from a treatment standpoint, but under certain conditions it may result in the

formation of dangerous intermediate products, including highly reactive DBP precursors (Liu et

al., 2010; Gora and Andrews, 2017). This risk can be avoided by instead removing contaminants

via adsorption to TiO2, separating the used TiO2 from the treated water, regenerating the

materials via photocatalysis, then recycling the regenerated TiO2 back into the main treatment

tank. This two-step process would rely exclusively on adsorption for contaminant removal from

the water, thus avoiding the formation of undesirable byproducts during the treatment step.

Numerous researchers have explored the adsorption of NOM to TiO2 nanoparticles, but most of

these studies were conducted in a contaminant transport context. Liu et al. (2014), Ng et al.

(2014), and our own research group (Gora and Andrews, 2017) have demonstrated that NOM

adsorption may be appropriate for drinking water treatment. It is well established that large and

221

aromatic NOM compounds are preferentially removed (Hyung and Kim, 2008; Erhayem and

Sohn, 2014; Gora and Andrews, 2017) and that the effects of the water matrix on NOM

adsorption to TiO2 are complex. Water matrix properties such as pH, ionic strength, the presence

of divalent ions such as calcium, NOM concentration and type, and the presence of other

interferents can increase or decrease the amount of adsorption achieved by modifying the surface

properties of TiO2, competing for adsorption sites, or by encouraging agglomeration, thus

reducing the overall surface area available for adsorption.

The pH of the water matrix has a strong effect on the surface charge of the nanoparticles and the

resulting adsorption of NOM. Adsorption is increased at pH values below the isoelectric point

(IEP) of the TiO2 nanomaterial (pH 6.5 for P25 nanoparticles) and above the pKa of the

contaminant (approximately pH 2 to pH 4 for NOM) due to charge interactions (Mwaanga et al.,

2014; Erhayem and Sohn, 2014; Gora and Andrews, 2017). pH also has an effect on

agglomeration and thus on available surface area because charge effects between nanoparticles

are minimized when the pH of the matrix is near the IEP, resulting in greater agglomeration

(Hotze et al., 2010). Many researchers have observed that the presence of ions, in particular

divalent ions such as calcium, also encourages greater agglomeration of TiO2 nanomaterials (Liu

et al., 2013), likely because the presence of ions diminishes the overall repulsive forces between

individual nanoparticles, allowing attractive van der Waals forces to dominate (Hotze et al.,

2010). However, calcium ions can also promote NOM adsorption via bridging (Liu et al., 2013;

Sun and Lee., 2010). NOM itself has also been shown to have a strong impact on nanomaterial

agglomeration and suspension stability (Zhang et al., 2009; Hotze et al., 2010; Zhou et al., 2013;

Erhayem et al., 2014; Loosli et al., 2014; Liu et al., 2013), but the effects appear to be specific to

NOM concentration, NOM type, and the properties of the nanomaterial in question. For example,

Zhou et al. (2013) observed that 10 mg/L of Suwannee River NOM enhanced the stability of

some nanomaterial suspensions but reduced that of others while Erhayem and Sohn (2014)

observed that NOM concentrations ranging from 10 to 20 mg/L destabilized P25 nanoparticle

suspensions but higher concentrations of NOM restabilized them. Finally, compounds such as

phosphate, nitrate, and carbonate are known to interfere with NOM adsorption to TiO2 and may

compete with DBP precursors for adsorption sites (Chen et al., 1997; Erhayem and Sohn, 2014).

Based on detailed adsorption studies some researchers have hypothesized that the impacts of

ionic strength and pH on NOM adsorption by TiO2 are related to effects of these parameters on

222

NOM shape, conformation, and subsequent behaviour rather than direct effects on the adsorption

process (Hyung and Kim, 2008; Sun and Lee, 2012). NOM compounds can become more tightly

coiled under low pH or high ionic strength conditions, resulting in smaller individual molecule

size and less overall demand for the available surface area of the TiO2 nanomaterial (Hyung and

Kim, 2008). Sun and Lee (2012) suggested that calcium ions encouraged charge neutralization

and flocculation of NOM ahead of adsorption, leading to better overall removal due to reduced

charge repulsion between NOM and TiO2.

The key to making this process work is the development of a regenerable TiO2 material with a

high adsorption capacity that can be removed from the treated water quickly. Degussa / Evonik

P25 Aeroxide nanoparticles, the standard commercial TiO2 nanoparticles used in this and many

other studies, are spherical, have a diameter of approximately 21 nm, and consist of

approximately 80% anatase and 15 to 20% rutile TiO2, with the remainder made up of

amorphous TiO2 (Ohtani et al., 2010). The small size of these particles makes them difficult to

remove using the standard clarification methods used in drinking water treatment plants (e.g.

sedimentation, media filtration, membrane filtration). Many larger TiO2 nanomaterials, some

with complex geometries, have been synthesized by materials scientists in the past three decades.

The simplest of these are the tubular linear nanomaterials first described by Kasuga et al. (1999).

Other researchers, including Yuan and Su (2004), have since modified Kasuga et al.’s original

method to produce linear nanomaterials with different length to width ratios, crystalline

structure, and surface characteristics, including nanobelts, nanofibers, and nanowires. In this

study, two types of belt-like linear nanomaterials were synthesized using Kasuga’s method

modified based on some of the findings of Yuan and Su. The resulting materials were evaluated

for NOM adsorption, regenerability, and ease of removal from water via filtration and settling.

Methods and Materials

8.2.1 Materials

8.2.1.1 Chemicals

Degussa P25 Aeroxide TiO2 nanoparticles and Acid Orange 24 dye (AO24) was obtained from

Sigma Aldrich. AO24 is a sulfonated double azo dye with a molecular weight of 448 g/mol and a

223

peak absorbance at 430 nm that is readily decolourized via photocatalysis. The pKa of AO24 has

not been established but other azo dyes have pKa ranging from pH 10 to pH 11 (Perez-Urquiza

and Beltran, 2001). It has been used in other TiO2 photocatalysis studies as a dosimeter for

microorganisms (Bandala et al., 2011). Preliminary unpublished data from our laboratory

suggested that P25 nanoparticles and LENs adsorbed AO24 at similar rates as they adsorbed

NOM, thus making AO24 a simple and useful NOM surrogate for the initial evaluation of novel

regenerable TiO2-based adsorbents.

8.2.2 Raw Water Quality

Unchlorinated raw water was collected from the influents of two Canadian water treatment

plants (WTP). The Peterborough WTP is supplied by the Otonabee River (OTB) while the

Britannia WTP in Ottawa is supplied by the Ottawa River (OTW). Average values for a number

of parameters specific to DBP formation, adsorption, nanomaterial agglomeration, and

regeneration are summarized in Table 8.1.

The two water sources differed primarily in terms of NOM character and ionic composition. The

OTW water had higher average DOC, UV254, and SUVA values than the OTB water, indicating

that former contained more aromatic NOM and, most likely, a higher concentration of DBP

precursors (Pifer and Fairey, 2014; Zheng et al., 2015). NOM has been shown to stabilize

nanoparticle suspensions by preventing agglomeration and settling (Liu et al., 2013) suggesting

that the nanomaterial suspensions made in OTW water might be less likely to agglomerate and

thus more conducive to adsorption and less likely to settle out quickly. The OTW water also

contains higher concentrations of iron and copper than the OTB water. Iron is readily adsorbed to

the surface of TiO2 (Chen and Ray, 2001) and might be expected to compete with NOM and

DBP precursors for adsorption sites. Iron and copper have been shown to promote faster

degradation of organic contaminants in TiO2 photocatalytic systems at certain pH values (Butler

and Davis, 1993), which may have implications for regeneration.

224


Parameter Units Otonabee River Ottawa River

DOC1 mg/L 4.3 ± 0.3 5.6 ± 0.4

UV2541 1/cm 0.097 ± 0.005 0.183 ± 0.017

SUVA1 m/mg.L 2.3 ± 0.2 3.3 ± 0.3

pH1 8.0 ± 0.2 7.1 ± 0.2

Turbidity2 NTU 0.6 ± 0.2 3.3 ± 1.0

Alkalinity1 mg/L as CaCO3 85 ± 1 27 ± 1

Hardness2 mg/L as CaCO3 95 ± 11 30 ± 6

Calcium2 mg/L 32.8 ± 3.7 8.3 ± 1.5

Magnesium2 mg/L 3.2 ± 0.3 2.2 ± 0.4

Sodium2 mg/L 6.5 ± 0.8 3.4 ± 0.8

Chloride2 mg/L 11.5 ± 1.3 3.3 ± 0.9

Conductivity2 S/cm 214 ± 19 81 ± 13

Copper g/L 0.7 ± 0.1 27 ± 10

Iron2 g/L 19 ± 9 217 ± 42


2Average and standard deviation of values obtained from Ontario Drinking Water Surveillance Program 2010-2012

In contrast, the OTB water had higher alkalinity (a measure of carbonate ion content), hardness

(a measure of divalent cation content), and conductivity (a measure of overall ionic content) than

the OTW water. Higher ionic content is associated with a decrease in the repulsive forces that

maintain the stability of colloid or nanoparticle suspensions (Hotze et al., 2010). When the

repulsive forces are depressed, the nanoparticles begin to agglomerate due to van der Waals

forces. Agglomeration reduces the surface area available for adsorption and photocatalytic

reaction and also encourages settling. Calcium ions contribute to the overall ionic strength of the

water, but also encourages greater NOM adsorption to TiO2, possibly through a bridging

mechanism (Sun and Lee, 2012; Erhayem and Sohn, 2014, Liu et al., 2013). In this experiment,

it was expected that the high ionic content of the OTB water would encourage settling and

interfere with adsorption by inducing the individual particles to agglomerate but conversely that

its high level of calcium might encourage more NOM adsorption.

225

8.2.3 Synthesis and Characterization of Engineered Nanomaterials

Two LENs were synthesized using a simple hydrothermal method pioneered by Kasuga et al.

(1999) and since adapted by many others including Yuan and Su (2004), Qamar et al. (2008),

Zheng et al. (2010), Liu et al. (2013) and Ali et al. (2015). Briefly, P25 nanoparticles were mixed

with 10 M NaOH, placed in a Teflon lined reactor, and heated to 240oC for 24 hours. The

resulting material was repeatedly rinsed with MilliQ water and then immersed in 0.1 M HCl for

one hour before being rinsed with another 1.2 L of MilliQ water. The material was dried

overnight, crushed into a fine powder, then heated to 550oC or 700oC for 4 hours. The finished

product was then immersed in MilliQ water, sonicated for five minutes, and then allowed to

settle out for 24 hours. The MilliQ water and unsettled TiO2 was discarded and the process was

repeated with a settling time of 3 hours. The final product was dried and stored at 20oC.

NB 550 and NB 700 were characterized using transmission electron microscopy (TEM) selected

area electron diffraction (SAED), zeta potential at different pH values to identify the isoelectric

point, and N2 adsorption isotherms. TEM and SAED observation was conducted at the Canadian

Centre for Electron Microscopy (Hamilton, Ontario, Canada) on a JEOL 2010F TEM/STEM.

TEM images were processed using Gatan Microscopy Suite: Digial MicrographTM and SAED

and FFT images were indexed using CrysTBox – diffractGUI (Klinger and Jäger, 2015). A

Quantachrome AUTOSORB-1 was used to determine the N2 adsorption isotherms of the two

materials and their surface areas were determined by applying Brunauer–Emmett–Teller (BET)

adsorption method on N2 adsorption isotherms in a relative pressure range of 0.05 to 0.25. The

isoelectric points (IEP) of the LENs were determined by measuring their zeta potential at

different pH values using a Horiba Scientific Nanopartica SZ-100 Nanoparticle Analyzer and

designating the pH at which the zeta potential was zero as the IEP.

8.2.4 Adsorption Experiments

The time required to reach adsorption equilibrium and the effect of increasing TiO2 dose on the

removal of AO24 dye, DOC, UV254, THMfp, and HAAfp by P25 and the two LENs were

investigated in a series of adsorption experiments. In all cases, mixing was provided by an end

over end box mixer.

226

In the AO24 adsorption time to equilibrium experiments 25 mL of a 10 mg/L solution of AO24

was added to a 40 mL amber vial and dosed with 0.5 g/L of TiO2 and mixed in the dark for 5, 10,

15, 30, 45, or 60 minutes. After mixing, the samples were filtered through a 0.45 m PES filter

to remove the TiO2. The AO24 adsorption isotherm tests were conducted at AO24 concentrations

ranging from 5 mg/L to 500 mg/L and a TiO2 dose of 1 g/L. The adsorption isotherm samples

were mixed for 30 minutes in the dark and separation was achieved via centrifugation to avoid

potential interactions between the PES filters and the different concentrations of AO24 in the

samples. All AO24 adsorption experiments were conducted in duplicate.

The time required to reach equilibrium between the TiO2 nanomaterials and NOM was

determined by dosing duplicate 75 mL of OTB or OTW water with 0.5 g/L of TiO2 and mixing

the samples for times ranging from 2.5 minutes to 4 hours. The treated samples were filtered

through a 0.45 m PES filter and analyzed for DOC and UV254. Quadruplicate 150 mL

adsorption isotherm samples were prepared such that one replicate could be used for chlorine

demand while the remainder were analyzed for DOC, UV254, THMfp, and HAAfp. The samples

were dosed with TiO2 at concentrations ranging from 0.1 g/L to 1.5 g/L and mixed for 3 hours in

the dark after which the TiO2 was removed via filtration.

8.2.5 Regeneration Experiments

Regeneration experiments were conducted to determine whether the two LENs could be reused

multiple times to remove AO24 and DBP precursor surrogates. In the AO24 tests, duplicate vials

containing 25 mL of 10 mg/L dye solution was dosed with 0.5 g/L of NB 550 or NB 700 and

mixed end over end in a box mixer. After 30 minutes the TiO2 was removed from the samples

via filtration and resuspended in 25 mL of millQ purified water. The new suspensions were

mixed with a stir plate and stir bar and exposed to UVA light (365 nm) with an average

irradiance of 4.9 mW/cm2 for one hour. The regenerated LENs were removed from the purified

water via filtration and then resuspended in a fresh volume of AO24 solution and mixed for 30

minutes in the box mixer. This process was repeated five times for each LEN. The same

regeneration process was used to evaluate whether the LENs could be reused to remove NOM as

quantified by UV254, which is a well established surrogate parameter for DBP precursors (Chen

and Westerhoff, 2010, Pifer and Fairey, 2014, Zheng et al., 2015).

227

8.2.6 Filtration and Settling Tests

The filterability of solutions containing the three nanomaterials was determined using the time to

filter test described in Standard Method 2710 H (APHA, 2012). In this method, a defined volume

of water or sludge is filtered through a standard laboratory filtration apparatus equipped with a

0.45 m laboratory membrane filter at a vacuum pressure of 15 mmHg (2 kPa). The time

required for the water in the sample to pass through the filter is recorded and normalized to the

time required to filter an equal volume of purified water under the same conditions. The resulting

parameter is referred to as the filtration index. A high filtration index implies that the sample is

resistant to filtration whereas a filtration index close to 1 indicates that the solution does not

present a major barrier to filtration. Although the time to filter test is not able to accurately

predict the behavior of full-scale filtration processes (e.g. media filtration, membrane filtration)

but it does provide a simple quantitative point of comparison between the different materials and

water sources. In this study, 1 g/L suspensions of TiO2 were prepared in the two river water

samples. The suspensions were sonicated for five minutes and then mixed in the box mixer for

one hour before testing. The time required to filter a 100 mL of each suspension through a 0.45

m PES filter was determined and normalized to the time required to filter 100 mL of MilliQ

water. Raw river water controls were also filtered and all filtration experiments were conducted

in triplicate.

Settling tests were conducted in the two surface water matrices at a TiO2 dose of 1 g/L. Other

researchers, including Thio et al. (2011), Liu et al. (2013), and Erhayem and Sohn (2014), have

explored the sedimentation behavior of TiO2 nanomaterials, though these investigations have

generally been conducted in a contaminant transport context. Usually, a water sample containing

the nanomaterial is placed in a turbidimeter or UV-Vis spectrophotometer and readings are taken

periodically to determine the rate of settling over time. The dose of TiO2 used in this study

resulted in turbidity and UV-Vis signals well above the operating ranges of the instruments

available in our laboratory, As a result, it was necessary to adopt an alternative methodology.

400 mL of water was dosed with 1 g/L of TiO2, sonicated for five minutes, and then mixed in the

box mixer. After one hour, the water was distributed into six tall 60 mL vials and the remaining

volume was reserved as a control. 30 mL aliquots were removed after 5, 10, 15, 30, 45, and 60

minutes and diluted to one tenth their original concentration. The diluted samples were analyzed

using a HACH turbidimeter. The settling tests were conducted in duplicate.

228

8.2.7 Analysis of AO24, DOC, UV Absorbance, and DBP Formation Potential

The concentration of AO24 was quantified by measuring the absorbance of the sample at 430 nm

using an Agilent UV-Vis spectrophotometer. The dissolved organic carbon (DOC) content of the

raw and treated river water samples was determined using an O.I. Analytical Aurora 1030W

TOC analyzer operating in persulfate oxidation model and their UV absorption at 254 nm

(UV254) was analyzed on an Agilent 8453 UV-Vis analyzer.

The chlorine demand and DBP formation potential (DBPfp) of the raw and treated river water

samples were assessed at uniform formation conditions (UFC) as described by Summers et al.

(1996). The THMs and HAAs formed through chlorination were extracted according to Standard

Method 6232 B and Standard Method 6251 B (APHA, 2005) and analyzed using an Agilent

7890B GC-ECD. The treated methylene blue samples were centrifuged to remove the TiO2

nanomaterials and then analyzed for absorbance at 665 nm to determine the concentration of

methylene blue remaining in solution. All NOM removal experiments were conducted in

quadruplicate with one replicate being used for chlorine demand and three being used for DBPfp

determination.

8.2.8 Isotherm Modeling and Other Statistical Analyses

The adsorption datasets were evaluated using one-way ANOVA to elucidate the effects of time

and TiO2 concentration on the extent of AO24 and NOM adsorption by the three TiO2

nanomaterials. Tukey’s pairwise comparison method was used to calculate confidence intervals

on the means at each time and/or concentration point. ANOVA and Dunnett’s method were also

used to evaluate the effects of regeneration on the removal of AO24 and NOM by the two LENs

as well as the results of the filtration tests.

The AO24 adsorption data obtained in MilliQ water was fitted to the linearized Freundlich

isotherm model while the DOC, UV254, THMfp, and HAAfp datasets from the experiments

conducted with the real water matrices were fit to a linearized modified version of the Freundlich

isotherm that is commonly used for NOM adsorption studies (Summers and Roberts, 1988).

𝑞 = 𝐾𝐹𝑀(𝐶𝐷⁄ )

1𝑛⁄

(8.1)

229

This model, which explicitly accounts for adsorbent dose, is commonly used in cases where it is

difficult to vary the initial concentration of the adsorbate and thus adsorbent dose is the changing

variable. Previous work by our research group has demonstrated that it is also appropriate for the

modeling of NOM removal by TiO2 nanoparticles (Gora and Andrews, 2017). Linear regression

was used to determine KF or KFM and 1/n as well as the fit of the linearized isotherm models and

the associated confidence intervals (95%) have been reported all adsorption parameters. The fit

of the resulting two-parameter non-linear adsorption model to the data obtained in the study was

evaluated using Marquardt’s percent standard deviation (MPSD), which is calculated as shown

in Equation 8.2 (Kumar et al., 2008):

𝑀𝑃𝑆𝐷 = 100 √1

𝑛−𝑝∑ (

𝑞𝑒,𝑚𝑒𝑎𝑠−𝑞𝑒,𝑐𝑎𝑙𝑐

𝑞𝑒,𝑚𝑒𝑎𝑠)2

𝑛𝑖=1 (8.2)

All statistical tests were conducted at the 95% confidence level.


8.3.1 Characterization of Engineered Nanomaterials

TEM imaging revealed that both NB 550 and NB 700 were rectangular and belt-like with widths

ranging from 25 to 200 nm and lengths ranging from 100 nm to 10 m (Figure .81). The two

materials had distinctive surface morphologies. NB 550 appeared to be covered in pores or

protuberances whereas NB 700 was smooth but appeared segmented. The LENs also differed

from one another and from P25 nanoparticles in terms of their specific surface areas,

crystallinity, and isoelectric points (IEP). At 29.9 m2/g, NB 550 had a greater surface area than

NB 700, which had a specific surface area of 18.3 m2/g. Both LENs had smaller specific surface

areas than P25 (57.4 m2/g). Commercial P25 nanoparticles contain approximately 75% anatase

and 25% rutile, a combination that is widely held to be conducive to high photocatalytic activity.

In contrast, the results of SAED and TEM analysis indicated that NB 550 contained both anatase

and TiO2(B), a relatively inert form of TiO2 whereas NB 700 consisted entirely of anatase.

Previous studies by our group and other have determined that P25 has an IEP of approximately

6.5 (Gora and Andrews, 2017; Mwaanga et al., 2014), so it is likely that it was electrostatically

230

neutral in all four matrices employed in this study. The IEPs of both LENs fell between pH 4 and

pH 5 as shown in Figure 8.S.1 of the supplemental material section (Section 8.6). Other

researchers have observed that the point of zero charge, which is analogous to the IEP, of TiO2

nanomaterials decreases with increasing particle size (Zhou et al., 2013), so this result was not

unexpected. It implies that both materials were electrostatically neutral in the MilliQ and 10

mg/L AO24 matrices, both of which had pH ranging from 5 to 6, but negatively charged at the

ambient pH of the river water samples, which ranged from 7.1 ± 0.2 for the OTW water to 8.0 ±

0.2 for the OTB water.

Figure 8.1 TEM images of (A) NB 550 and (B) NB 700

8.3.2 Acid Orange 24 Adsorption to TiO2

8.3.2.1 Time to Equilibrium

The results of the experiments conducted to determine the time required to reach equilibrium

between AO24 and the three TiO2 nanomaterials are shown in Figure 8.S.2 in Section 8.6

(supplementary material). Most adsorption studies assume that the true equilibrium between

adsorbent and adsorbate is not reached for many hours or even days, but in this study adsorption

occurred quickly in all cases and there was no statistical difference in removal after 5 minutes of

adsorption. This was declared the point of effective equilibrium and all subsequent AO24

experiments were conducted with a 30 minute adsorption period to ensure that the treatment had

proceeded well beyond the point of effective equilibrium. In this experiment and in all

A B

231

subsequent AO24 experiments the P25 nanoparticles had a higher adsorption capacity for AO24

than NB 550 and NB 700, removing 43 ± 3% of the AO24 from the control vs. 32 ± 2% for NB

550 and 34 ± 3% for NB 700.

8.3.2.2 AO24 Adsorption Isotherms

The adsorption isotherm results were fitted to the linearized form of the Freundlich isotherm.

The results of the analysis are presented in Table 8.2 and in Figure 8.2. The Freundlich isotherm

model was developed empirically and is often able to model systems that do not conform to the

assumptions of the Langmuir isotherm model, namely monolayer adsorption, a homogeneous

adsorbent surface, and no interaction between adsorbed molecules.

Table 8.2 Isotherm parameters for the adsorption of AO24 by P25 and two LENs

Parameters Material

P25 NB 550 NB 700

KF1 2.67 ± 0.82 0.62 ± 0.19 0.53 ± 0.36

KF/SA2 0.047 ± 0.014 0.021 ± 0.006 0.028 ± 0.019

1/n 0.63 ± 0.09 0.83 ± 0.08 0.78 ± 0.16

R2 95% 97% 87%

MPSD 25% 22% 47%

1(mg/g)/(mg/L)1-n

2(mg/m2)/(mg/L)1-n

The good fit of the lines in Figure 8.2 and the high R2 values in Table 8.2 indicate that the

Freundlich model was a good fit for the P25 and NB 550 datasets and adequately described the

adsorption of AO24 to NB 700. The slope parameter, 1/n, was similar, but not equal, for all three

materials, suggesting that their KF values can be compared to one another, but cautiously. The KF

values of NB 550 and NB 700 were statistically equal to one another, indicating that they had

similar AO24 adsorption capacities while the KF of P25 was over four times larger than those of

the LENs, reflecting its higher adsorption capacity for AO24. When the KF values were

normalized to the surface areas of the different materials the P25 was only twice as effective as

the LENs, indicating that the improved removal of the former over the latter was to some extent

a function of differences in surface area (P25 = 57 m2/g; NB 550 30 m2/g; NB 700 18 m2/g).

232

Figure 8.2 AO24 adsorption data fitted to the Freundlich isotherm model

8.3.2.3 Nanomaterial Regeneration After AO24 Adsorption

The results of the regeneration experiments indicate that there was no statistical decrease (95%

confidence level) in the amount of AO24 adsorbed by the LENs over five regeneration cycles

(Figure 8.3). The only exception was a small but statistically significant drop in the amount of

dye adsorbed by NB 700 after the fourth regeneration cycle, which, given the improved dye

removal observed after the fifth regeneration cycle, was likely related to experimental error

rather than an actual reduction in adsorption capacity. As well, the results in Figure 8.3 suggest

that despite the statistical findings, there may in fact have been some reduction in NB 550’s

ability to adsorb AO24 after the third regeneration cycle. The statistical tests were rerun at the

90% confidence level to minimize the chance that the null hypothesis of no change was being

accepted erroneously (Type II error – see Appendix I), but the results remained the same.

Although five regeneration cycles represents only a fraction of the number regeneration cycles

that would likely be required should this technology be adopted in real water and wastewater

treatment installations, it is in line with other proof of concept studies (Liu et al., 2014; Ng et al.,

2015) and is a promising indication that the nanomaterials developed in this study are potentially

applicable as regenerable adsorbents.

1

10

100

1 10 100

log q

e

log Ce

P25 NB 550 NB 700

233

Figure 8.3 AO24 adsorption by virgin and regenerated NB 550 and NB 700. All samples

were prepared in duplicate and error bars represent the 95% confidence

interval on the mean. Legend numbers correspond to the number of

regeneration cycles.

8.3.3 NOM Adsorption to TiO2 Nanomaterials

8.3.3.1 Time to Equilibrium

The time required to reach an effective equilibrium between TiO2 and NOM (measured as DOC

and UV254) was determined for all three TiO2 nanomaterials in both water matrices (OTB and

OTW). The results of these experiments are shown in Figure 8.S.3 in Section 8.6 –

supplementary material. Tukey’s Method for multiple comparisons was used to establish the

time at which the consecutive samples were no longer significantly different from one another at

the 95% confidence level. In almost all cases, there was no significant change in the DOC or

UV254 of the treated samples after 15 minutes, indicating that NOM adsorption to the TiO2

surface occurred quickly. The removal of DOC from the OTW water did not stabilize until

between 90 and 120 minutes. The adsorption of aromatic NOM (measured by UV254) from this

0

2

4

6

8

10

NB 550 NB 700

Ma

ss o

f A

O2

4 A

dso

rbed

to

LE

Ns

(mg

AO

24

/gT

iO2)

Regeneration Cycle

0

1

2

3

4

5

234

water stabilized after ten minutes for all three materials, suggesting that the variable DOC results

were related to slower interactions between TiO2 and non-aromatic NOM. Based on these results

the NOM adsorption isotherm experiments were performed using a three hour adsorption period.

8.3.3.2 Adsorption of DOC and UV254

As illustrated in Figure 8.4, P25 adsorbed more DOC from both water sources than NB 550,

which itself adsorbed more than NB 700. The difference in the mass of DOC adsorbed by the

two LENs was more pronounced than that observed with AO24, particularly in the OTW water.

For example, at a TiO2 dose of 1.5 g/L, NB 550 removed 30 ± 1% of the DOC from the OTW

water but NB 700 only removed 15 ± 1%. At 30 m2/g, the specific surface area of NB 550 is

nearly twice that of NB 700 (18 m2/g). Given that the two materials have similar charge

characteristics, it seems likely that surface area was the most important factor influencing the

amount of DOC adsorbed to the TiO2 surface. In addition, under these same conditions P25,

which has a specific surface area of 57 m2/g, removed 37 ± 1% of the DOC in the raw OTW

water, further supporting the hypothesis that surface area played an important role in the

adsorption of DOC from this water source. This relationship also held true in the experiments

conducted with OTB water, the results of which are also shown in Figure 8.4.

235

Figure 8.4 Adsorption of DOC from Otonabee River water (OTB) and Ottawa River

water (OTW) by P25 nanoparticles and two LENs. Error bars represent the

95% confidence interval on the mean.

The removal of UV254 by the three materials (Figure 8.5) followed similar trends as DOC

removal. At a dose of 1.5 g/L P25 removed 54 ±3 % of the UV254 from the OTB water and 62 ±

5% of the UV254 from the OTW water and NB550 removed 38 ± 5 % of the UV254 from the

OTB water and 49 ± 4% from the OTW water. The OTW water contained more NOM than the

OTB water (5.6 ± 0.4 mg/L vs. 4.3 ± 0.3 mg/L) and this NOM was more aromatic in character

(SUVA of 3.3 ± 0.3 L/mg.m vs. 2.3 ± 0.2 L/mg.m). Research by others has demonstrated that

NOM adsorption increases at higher NOM concentrations (Mwaanga et al., 2014; Erhayem and

Sohn, 2014; Kim and Shon, 2007) and that larger, more aromatic NOM is preferentially

adsorbed by P25 nanoparticles (Erhayem and Sohn, 2014), so this result was not surprising. The

removal of UV254 from the OTB water by NB 700 was similar to its removal from OTW by this

nanomaterial – it removed 30 ± 1% from the former and 25 ± 2 %both of them, suggesting that

NOM adsorption by NB 700 was not as strongly influenced by the amount or type of NOM

present in the water.

2

2.5

3

3.5

4

4.5

5

5.5

6

0 0.25 0.5 0.75 1 1.25 1.5

DO

C (

mg

/L)

TiO2 Dose (g/L)

P25 - OTB NB 550 - OTB NB 700 - OTB

P25 - OTW NB 550 - OTW NB 700 - OTW

236

Figure 8.5 Adsorption of UV254 from Otonabee River water (OTB) and Ottawa River

water (OTW) by P25 nanoparticles and two LENs nanomaterials. Error bars

represent the 95% confidence interval on the mean.

8.3.3.3 Reduction of THMfp and HAAfp via Adsorption of DBP Precursors

The main reason why NOM is removed from drinking water ahead of chlorination is to prevent

the formation of DBPs during the final disinfection step of the overall drinking water treatment

system. Not all NOM compounds are DBP precursors, so although the DOC and UV254 removal

results presented in Section 8.3.3.2 are promising, it was important to prove that the LENs were

specifically capable of adsorbing DBP precursors. As shown in Figure 8.6, all three of the TiO2

nanomaterials adsorbed significant amounts of the precursors of THMs and HAAs, the most

widely regulated DBPs in North America and around the world. At the highest TiO2 dose (1.5

g/L), P25 reduced the THMfp of the OTB water by 48 ± 2 % and that of the OTW water by 61 ±

6%. NB 700 also worked more effectively in the OTW water than in the OTB water, reducing

the THMfp of the former by 38 ± 3% and that of the latter by 27 ± 5%. NB 550 was equally

effective in both water matrices, reducing the THMfp of the OTB water by 37 ± 4% and that of

the OTW water by 43 ± 7%.

0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

0.18

0.2

0 0.25 0.5 0.75 1 1.25 1.5

UV

25

4 (

1/c

m)

TiO2 Dose (g/L)



237

Figure 8.6 Adsorption of THM precursors from Otonabee River water (OTB) and

Ottawa River water (OTW) by P25 nanoparticles and two LENs. Error bars


The three nanomaterials were also able to adsorb significant amounts of HAA precursors from

the two water matrices as shown in Figure 8.7. The HAAfp results were, as a rule, more variable

than the THMfp results, reflecting the increased complexity of the HAA extraction and

derivatization methods. This made it more difficult to identify and quantify trends based on

material type or water matrix. Nonetheless, it was apparent that NB 550 consistently had a

greater affinity for HAA precursors than NB 700 did, removing 51 ± 13% from the OTB water

and 47 ± 13% from the OTW water. NB 700 removed only 24 ± 2% from the OTB water and 31

± 2% from the OTW water.

0

50

100

150

200

250

300

0 0.25 0.5 0.75 1 1.25 1.5

TH

Mfp

(

g/L

)

TiO2 Dose (g/L)



238

Figure 8.7 Adsorption of HAA precursors from Otonabee River water (OTB) and

Ottawa River water (OTW) by P25 nanoparticles and two LENs. Error bars


8.3.3.4 Isotherm Modeling

The DOC, UV254, THMfp, and HAAfp datasets discussed in Section 8.3.3.2 and 8.3.3.3 were

fitted to a modified version of the Freundlich isotherm first introduced by Summers and Roberts

(1988) and since employed by numerous other researchers exploring NOM adsorption to

activated carbon (e.g. Karanfil and Kitis, 1999, Li et al., 2002) and TiO2 (Erhayem and Sohn,

2014). The adsorption isotherms are provided in Figure 8.8 (DOC), Figure 8.9 (THMfp), and

Figure 8.S.4 (UV254) and Figure 8.S.5 (HAAfp) in Section 8.6 (supplementary material) and the

isotherm parameters are presented in Table 8.3.

0

20

40

60

80

100

120

140

0 0.5 1 1.5

HA

Afp

(

g/L

)

TiO2 Dose (g/L)



239

Figure 8.8 DOC adsorption data sets from experiments conducted in (A) Otonabee

River (OTB) water and (B) Ottawa River (OTW) water fitted to a modified

Freundlich isotherm model

0.1

1

10

1 10 100

q (

mg

DO

C/g

TiO

2)

C/D (mg DOC/g TiO2)


0.1

1

10

1 10 100

q (

mg D

OC

/g T

iO2)

C/D (mg DOC/g TiO2)


A

B

240

Table 8.3 Modified Freundlich model isotherm parameters for the removal of DOC,

UV254, THM precursors, and HAA precursors from Otonabee River (OTB)

and Ottawa River (OTW) water by P25 nanoparticles and two LENs

Parameter Water Material KFM1 KFM/SA2 1/n R2 MPSD

DOC OTB P25 0.67 ± 0.15 0.012 ± 0.003 0.55 ± 0.11 88% 25%

NB 550 0.61 ± 0.13 0.020 ± 0.004 0.41 ± 0.09 84% 20%

NB 700 0.44 ± 0.04 0.024 ± 0.002 0.49 ± 0.04 97% 9%

OTW P25 0.90 ± 0.08 0.016 ± 0.001 0.55 ± 0.04 98% 9%

NB 550 0.59 ± 0.20 0.020 ± 0.007 0.38 ± 0.15 70% 22%

NB 700 0.34 ± 0.09 0.019 ± 0.005 0.40 ± 0.11 79% 17%

UV254 OTB P25 0.20 ± 0.04 0.004 ± 0.001 0.47 ± 0.08 89% 23%

NB 550 0.15 ± 0.06 0.005 ± 0.002 0.60 ± 0.16 83% 27%

NB 700 0.06 ± 0.01 0.003 ± 0.000 0.37 ± 0.05 94% 9%

OTW P25 0.25 ± 0.04 0.004 ± 0.001 0.38 ± 0.08 89% 15%

NB 550 0.17 ± 0.03 0.006 ± 0.001 0.40 ± 0.09 83% 23%

NB 700 0.09 ± 0.02 0.005 ± 0.001 0.41 ± 0.15 70% 31%

THMfp OTB P25 2.88 ± 1.98 0.051 ± 0.035 0.68 ± 0.13 93% 20%

NB 550 2.76 ± 2.26 0.092 ± 0.075 0.60 ± 0.12 91% 26%

NB 700 2.20 ± 4.21 0.122 ± 0.234 0.58 ± 0.26 74% 40%

OTW P25 22.5 ± 13.21 0.394 ± 0.232 0.34 ± 0.09 87% 58%

NB 550 11.9 ± 25.8 0.399 ± 0.861 0.36 ± 0.25 53% 44%

NB 700 4.45 ± 4.30 0.247 ± 0.239 0.56 ± 0.14 89% 45%

HAAfp OTB P25 0.53 ± 0.42 0.009 ± 0.007 0.87 ± 0.17 95% 26%

NB 550 1.55 ± 1.07 0.052 ± 0.036 0.73 ± 0.15 92% 26%

NB 700 0.72 ± 0.82 0.040 ± 0.045 0.72 ± 0.25 87% 23%

OTW P25 8.43 ± 9.23 0.148 ± 0.162 0.33 ± 0.20 63% 35%

NB 550 12.91 ± 9.68 0.430 ± 0.323 0.34 ± 0.13 76% 25%

NB 700 1.53 ± 0.87 0.085 ± 0.049 0.66 ± 0.11 96% 15%

1(mg DOC/g TiO2)1-n, (UV254/g TiO2)1-n, (ug THMfp/g TiO2)1-n, (g HAAfp/g TiO2)1-n

2(mg DOC/m2 TiO2)1-n, (UV254/ m2 TiO2)1-n, (ug THMfp/ m2 TiO2)1-n, (g HAAfp/ m2 TiO2)1-n

241

As shown in Table 8.3, the DOC slope parameters (1/n) were statistically equal at the 95%

confidence level for all three materials in both water matrices, indicating that their adsorption

capacities (KFM values) can be compared to one another. The adsorption capacities of the

nanomaterials as estimated from their DOC isotherms indicate that P25 and NB 550 had

statistically equal adsorption capacities for DOC in the OTB water matrix and that P25 had a

greater capacity for DOC adsorption than the two LENs in the OTW water matrix, though this

was less apparent once the KFM values were normalized to surface area. P25’s KFM value was

larger in the OTW water than in the OTB water (KF,OTW = 0.90, KF,OTB = 0.67), indicating that

either the NOM in the OTW water was more amenable to adsorption, likely due to its

hydrophobic nature, or that the OTW matrix was less likely to inhibit adsorption by P25 for other

reasons. For example, the high ion content of the OTB matrix may have induced the P25

nanoparticles to agglomerate more than they did in the OTW water (Liu et al., 2013), thus

reducing the total TiO2 surface area available for adsorption in the OTB experiments compared

to the OTW experiments. NB 550’s KFM values were higher than those of NB 700 in both water

matrices. It should be noted, however, that the adsorption of DOC to NB 550 was not a good fit

to the linearized model, particularly in the OTW water matrix (R2 = 70%), and the full model

developed from the linearized isotherm was a relatively poor fit to the data (MPSD = 22%), so

the KFM values reported for NB 550 are unlikely to be as reliable as those reported for P25 and

NB 700.

The UV254 trends (Figure S.8.4) were, for the most part, similar to the DOC trends. The 1/n

values obtained in the experiments conducted with OTW were statistically equal to one another

(95% confidence level) and as such their KFM values can be compared to one another. P25

showed a higher adsorption capacity for aromatic NOM than the other two materials in the OTW

water matrix. This preference was less apparent when the KFM values were normalized to surface

area, indicating that P25’s higher adsorption capacity was likely a function of its higher surface

area. The relative UV254 adsorption capacities of the nanomaterials from the OTB experiments

were more difficult to compare because 1/n was not constant, however, the results do suggest

that P25 and NB 550 had a statistically equal capacity for aromatic NOM in this water matrix

and that both were superior to NB 700.

242

Figure 8.9 THMfp adsorption data sets from experiments conducted in (A) Otonabee

River (OTB) water and (B) Ottawa River (OTW) water fitted to a modified

Freundlich isotherm model

The THMfp and HAAfp datasets did not fit the modified Freundlich model as readily as the

DOC and UV254 datasets. Considering the higher variability in these parameters owing to the

more complex preparation methods required for the samples, however, the adsorption of DBP

10

100

1000

10 100 1000 10000

q (

g T

HM

fp/g

TiO

2)

C/D (g THMfp/g TiO2)


10

100

1000

10 100 1000 10000

q (

g T

HM

fp/g

TiO

2)

C/D (g THMfp/g TiO2)


B

A

243

precursors to the three TiO2 nanomaterials did provide a reasonable fit to the modified

Freundlich model in many cases, as shown in Table 8.3. For example, the R2 of the linearized

modified Freundlich model was above 90% for both THMfp and HAAfp reduction by P25 and

NB 550 in the OTB water. The 1/n values for P25 and NB 550 in OTB water were statistically

indistinguishable, and the KFM and KFM/SA values for these datasets indicate that the DBP

precursor adsorption capacity of NB 550 was equal to that of P25 in OTB water. A similar trend

was apparent when the TiO2 nanomaterials were used to adsorb HAA precursors from the OTB

water.

The terms THMfp and HAAfp usually refer to the propensity of a water sample to form a variety

of THMs (4) and HAAs (9) when chlorinated under standard conditions. In this study, only two

THMs, trichloromethane (TCM) and bromodichloromethane (BDCM), and two HAAs,

dichloroacetic acid (DCAA) and trichloroacetic acid (TCAA), were formed when the raw and

treated water samples were chlorinated. As shown in Figure 8.S.8 in the supplementary material

(Section 8.6), the ratio of TCM to BDCM formed upon chlorination was unchanged by the

adsorption treatment irrespective of the dose or type of TiO2 nanomaterial added to the water.

This indicates that none of the materials preferentially adsorbed TCM or BDCM precursors. In

contrast, the ratio of DCAA to TCAA formed upon chlorination increased when P25 was used as

an adsorbent (Figure 8.S.9), suggesting that the commercial nanoparticles were more likely to

adsorb TCAA precursors than DCAA precursors. The former are more hydrophobic than the

latter, so this finding further supports the hypothesis that P25 nanoparticles preferentially adsorb

hydrophobic NOM, as suggested by the DOC and UV254 results presented earlier (Figure 8.4,

Figure 8.5, Table 8.3).

Overall, the results of the adsorption isotherm analysis indicate that the modified Freundlich

model was appropriate for most of the matrix/material combinations tested in this study, but

there was no clear trend suggesting that one material or matrix provided an inherently better fit

than any of the others. Should the LENs be adopted for NOM removal from drinking water,

more detailed adsorption studies would prove helpful in elucidating the mechanisms driving the

adsorption of DOC, aromatic NOM (measured as UV254 absorption), and DBP precursors in

different matrices as well as the influences of different water matrix characteristics on the fit of

the model.

244

8.3.4 Regeneration of Engineered Nanomaterials After NOM Adsorption

The regenerability of the LENs is a key factor that will determine whether they can realistically

be employed for drinking water treatment. The amount of aromatic NOM (i.e. NOM measured

by UV254) adsorbed to the virgin and regenerated LENs in the two water matrices are shown in

Figure 8.10 and the UV254 of the water treated by the virgin and regenerated LENs is shown in

Figure 8.S.7 in the supplementary material. There was no statistical difference at the 95%

confidence level in the amount of NOM adsorbed by NB 700 for the OTB water matrix after five

regenerations. However, there was statistically significant loss in the amount of NOM adsorbed

by NB 550 from 0.048 ± 0.015 cm-1/gTiO2 to 0.025 ± 0.015 cm-1/gTiO2 after four regeneration

cycles when it was used to treat OTB water and an earlier and more substantial loss the amount

of NOM adsorbed by both LENs in the tests conducted with OTW water. For example, in the

OTW experiments the amount of NOM adsorbed by NB 550 decreased from 0.098 ± 0.009 cm-

1/gTiO2 TiO2 to 0.081 ± 0.009 cm-1/gTiO2 after the first regeneration and eventually to 0.046 ±

0.009 cm-1/gTiO2 after five regeneration cycles, which represents a 53% overall reduction in

adsorption. The poorer regenerability of NB 550, was likely due to its lower photoactivity

compared to NB 700 as described in Section 8.3.1.

Organic loading rates likely also impacted the regeneration efficacy and reuseability of the LENs

under the chosen regeneration conditions. Although both LENs removed a similar percentage of

DOC and UV254 from the water, the loading of NOM on the nanomaterials (NOM/g TiO2) was

higher when they were used to treat OTW water than when they were used to treat the OTB

water. For example, the fresh NB 550 was loaded with 0.048 ± 0.008 1/cm per gram of TiO2

after being suspended in the OTB water but it adsorbed 0.098 ± 0.004 1/cm per gram of TiO2 in

the OTW water. It is therefore possible that a longer period of irradiation would result in better

regeneration of the LENs after use in OTW water. The mechanisms underlying that adsorption

may also have been different. Thio et al. (2011) and Liu et al. (2013) hypothesized that calcium

ions could act as ionic bridges between NOM and the TiO2 surface, bypassing some of the

repulsive forces that would otherwise keep the two apart. In this study, calcium ions were more

plentiful in the OTB water than in the OTW water, and it may be that the initial adsorption

occurred via different mechanisms, some of which might be reversible in the regeneration matrix

(MilliQ water) or more likely to lead to NOM degradation upon irradiation.

245

Figure 8.10 Adsorption of aromatic NOM (UV254 absorbing NOM) by virgin and

regenerated NB 550 and NB 700. Error bars represent the 95% confidence

interval on the mean and legend numbers correspond to the number of

regeneration cycles.

Other factors that may have impacted regeneration include catalyst fouling by different

components of the natural water matrices or the formation and subsequent adsorption of

recalcitrant intermediates as a result the photocatalytic oxidation of adsorbed NOM during the

regeneration process. The information gathered in this study was insufficient to confirm or deny

either of these possibilities.

Despite the gradual decline in the amount of NOM adsorbed to the LENs that was observed in

some cases, these results are promising and confirm that the two nanomaterials can be

regenerated and reused multiple times for NOM removal from natural water matrices. The

regenerability of the LENs should, however, be confirmed at different TiO2 doses and in other

water matrices before the materials are employed in a real drinking water treatment process.

Adjustments to the regeneration procedure, including longer regeneration time, changes in the

pH or other characteristics of the water used for regeneration, and using different concentrations

of TiO2 during regeneration may further improve the results.

0

0.025

0.05

0.075

0.1

0.125

0.15

NB 550 - OTB NB 700 - OTB NB 550 - OTW NB 700 - OTW

NO

M A

dso

rbed

(1

/cm

/g T

iO2)

0

1

2

3

4

5

246

8.3.5 Removal of Nanomaterials from Treated Water

8.3.5.1 Filtration

The filtration indexes of the different TiO2 suspensions (Figure 8.11) illustrate the superior

filterability of LENs when compared to commercial P25 nanoparticles. A two-way ANOVA

with interactions was conducted on the dataset and it was determined that both material type and

water source had significant impacts on the filterability of the TiO2 suspensions. In the OTW

experiments, the filtration index of the P25 nanoparticles was approximately 6 to 8 times higher

than those of the LENs and in the OTB experiments it was 4 to 5 times higher. The improved

filterability of the LENs compared to the P25 nanoparticles is similar to what was observed by

Zhang et al. (2009) in a study that looked at the removal of LENs and nanoparticles from water

using a bench-scale membrane unit. They noted slower fouling rates with LENs than with

nanoparticles and hypothesized that this was likely due to differences in the deposition of the

materials on the surface of the membrane and within the membrane pores. In both cases, they

observed the formation of cake-like fouling on the surface of the membrane, which gradually

reduced the rate of water passage through the filter. The nanoparticles, however, were also

deposited within the membrane pores, which greatly increased the overall resistance to filtration.

As shown in Figure 8.11, P25 was more readily removed from the OTB water than from the

OTW water. The presence of ions, especially calcium, has been linked to increased

agglomeration and settling of P25 nanoparticles by numerous researchers (Hotze et al., 2010; Liu

et al., 2013; Thio et al., 2011), and the OTB water contained more calcium (32.8 ± 3.7 mg/L vs.

8.3 ± 1.5 mg/L) and other ions (conductivity = 214 ± 19 s/cm vs. 81 ± 13 S/cm) than the OTW

water. Increased agglomeration may have reduced the extent of P25 nanoparticle deposition in

the membrane pores and thus improved the overall filterability of the suspension.

247

Figure 8.11 Filtration indexes of raw water and three TiO2 nanomaterials suspended in

MilliQ water at pH 6 and pH 8 and two raw surface water samples

The filterability of the LENs was not impacted by the water matrix, indicating that they were

removed via size exclusion irrespective of the extent of agglomeration that may or may not have

taken place due to the pH of the water or the presence of ions or NOM. The water matrix did

have significant impacts on the filterability of the LENs relative to the raw water. For example,

the filterability indexes of the NB 550 and NB 700 suspensions made in OTB water matrices

were 3.41 ± 0.03 and 2.67 ± 0.02, respectively, which are both significantly higher than that of

the raw OTB water, which was essentially equal to that of the MilliQ water at 1.03 ± 0.01. In

contrast, the filtration index of the raw OTW water was 6.38 ± 0.52, well above that of the raw

OTB sample and significantly higher than the suspensions of NB 550 and NB 700 made in OTW

water (3.32 ± 0.13 and 2.55 ± 0.19). The OTW water contained more turbidity (3.3 ± 1.0 NTU

vs. 0.6 ± 0.2 NTU) and aromatic NOM (SUVA = 3.3 ± 0.3 L/mg.m vs. 2.3 ± 0.2 L/mg.m) than

the MilliQ and OTB water samples, which may explain its increased resistance to filtration even

in the absence of TiO2. Turbidity and aromatic NOM can foul membranes via cake formation but

also through the deposition of particles within the membrane pores and the adsorption of NOM

to the membrane surface and within the pores, leading to pore constriction as described

previously (Zhang et al., 2009). When TiO2 was added to the OTW water, the resistance to

0

5

10

15

20

25

Raw Water P25 NB 550 NB 700

Fil

tra

tio

n I

nd

ex

OTB

OTW

248

filtration decreased, possibly because the particulate matter and NOM in the raw water became

adsorbed to the TiO2 surface or enmeshed in the nanomaterial agglomerates, reducing the extent

of pore constriction.

In a full-scale membrane filtration process the filters would be periodically backwashed to

remove foulants, and the findings of this study suggest that a full-scale membrane filtration unit

would require more frequent backwashing if TiO2 was added to the water. The amount of

backwashing might be less with the LENs than with commercial P25 nanoparticles. Full-scale

membrane filters preceded by coagulation or powdered activated carbon also require more

frequent backwashing than those filtering unmodified raw water. Although these results suggest

that the LENs would be much easier to remove from treated water via filtration than P25

nanoparticles, they cannot be used to accurately predict the extent of fouling and flow reduction

that would occur in a full-scale media or membrane filter, however, and additional studies should

be conducted with flow through bench-scale membrane filters to track fouling and flow

reduction over time.

8.3.5.2 Sedimentation

Although filtration is the simplest and most intuitive way to remove nanomaterials from water,

other common separation processes used in water treatment plant, including sedimentation

(settling), might also be appropriate under certain conditions. The results of sedimentation

experiments conducted for this study, shown in Figure 8.12, hint at the complex interactions

occurring between the different nanomaterials and the components of the water matrices in

which they were originally suspended.

249

Figure 8.12 Percent removal of turbidity via sedimentation for three TiO2 nanomaterials

suspended in two raw surface water samples

Contrary to what was expected there was no obvious relationship between nanomaterial particle

size and the extent of settling, and in both cases, P25 settled more quickly than the two LENs.

This counterintuitive result, which is nonetheless similar to observations made by Liu et al.

(2013) in their study on the aggregation mechanisms of nanoparticles and linear nanomaterials, is

likely a function of numerous interrelated phenomena including the destabilizing effects of

calcium and other ions, particularly in the OTB water matrix, the stabilizing effect of NOM in

both water matrices, the surface charge of the different materials at the ambient pH of the two

water matrices, and the propensity of each material to form agglomerates of different sizes and

densities based on their shapes and sizes.

As described by Deloid et al. (2014) and discussed in detail in Appendix H, the effective density

of nanomaterial agglomerates is often well below the material density of TiO2 (4.26 g/cm3) and it

may be that the agglomerates formed by the P25 nanoparticles were smaller but denser than

those formed by the LENs. The measurement of effective nanoparticle agglomerate density was

beyond the scope of this study (see Appendix H), but it seems possible that the spherical P25

nanoparticles would be able to form tighter, denser agglomerates than the linear and irregularly

sized LENs. The Sterling equations (Appendix H) were used to estimate the effective densities of

0%

20%

40%

60%

80%

100%

P25 NB 550 NB 700

Tu

rbid

ity

Rem

ov

al

(30

min

)

OTB

OTW

250

the three materials in the different water matrices based on the agglomerate size distributions in

Figure H.2 (Appendix H). When these values were used to predict settling time with Stokes’

Law the resulting estimates were closer to the settling rates observed in the experiments (Table

H.2, Appendix H), indicating that the assumptions made during the calculations were incorrect

and/or that the nanomaterials in this study did not exhibit Type I settling behavior and their

settling velocity could not be described by Stokes’ Law.

The two natural water matrices employed in his study differed mainly in terms of ionic strength,

calcium content, and NOM content and character. As shown in Table 8.1, the conductivity of the

OTB water was nearly three times that of the OTW water. The rate of settling of the P25

nanoparticles was greater in OTB water, indicating that ion content likely had a strong impact on

the stability of P25 in solution at this concentration of TiO2. The presence of ions in the water

destabilizes particle suspensions by reducing the size of the electrical double layer that surrounds

the individual particles. The zeta potential of the materials, which is an indicator of the size of

the electrical double layer, was measured in both water sources, and as shown in Figure 8.13,

was more negative in the OTW water matrix than in the OTB water matrix for all three materials.

A more negative zeta potential implies a more stable suspension, so this result confirms that the

repulsive forces that usually prevent agglomeration were weakened by the contents of the OTB

matrix.

Figure 8.13 Zeta potential of P25 nanoparticles and two LENs in two natural water

matrices

-50

-40

-30

-20

-10

0


Zet

a P

ote

nti

al (m

V)

OTB

OTW

251

It was less clear exactly what role NOM may have played in the stabilization or destabilization of

the nanomaterials in the two natural water matrices, but it seems likely that it contributed to the

increased stability of NB 550 and NB 700 suspensions in OTW water. Liu et al. (2013) explored

the effects of commercially available Suwannee River NOM (SRNOM) on the settling of P25

nanoparticles and LENs similar, though not identical, to those used in this project. They found

that increasing the concentration of SRNOM from 0 to 10 mg/L decreased the size of the LEN

agglomerates and slowed their settling considerably. This effect was less apparent for TiO2

nanoparticles. Based on their analysis of the DVLO profiles of their nanomaterials under

different water quality conditions they attributed the decreases in LEN agglomeration and

settling in the presence of NOM to the formation of an energy barrier due to steric hindrance, as

has also been observed by other researchers (Domingos et al., 2009; Thio et al., 2011). This

energy barrier was decreased in the presence of calcium ions and was not observed for spherical

nanoparticles. Although the LENs employed by Liu et al. differed from those in this study in

terms of crystal phase structure and surface area, the trends that they observed match those

observed in this project and their hypotheses with respect to the effects of NOM and calcium on

the stability of nanomaterials in surface water may also explain the findings of this study.

pH can also have a strong effect on agglomeration and subsequent settling but the ambient pH of

the two natural water matrices was similar, so it is unlikely to explain the observations of this

study. Both of the natural water matrices had ambient pH values between 7 and 8, which is well

above the IEPs of all three nanomaterials, the nanomaterials would hold an overall negative

charge and the resulting repulsive forces between individual particles may have helped to

maintain the stability of the suspensions. The effects of the water matrices on the stability of the

different nanomaterials in this study were undoubtedly complex, and that very complexity will

likely discourage the use of settling as a separation mechanism for TiO2 from drinking water.

252

Conclusions

This study successfully demonstrated the application of two regenerable nanoscale linear

engineered TiO2 adsorbents (LENs) for the removal of an indicator dye and disinfection

byproduct precursors from two natural drinking water matrices. Both LENs were more easily

removed from the water via filtration than commercial P25 TiO2 nanoparticles. An alternative

separation mechanism, sedimentation (settling) was found to be strongly influenced by the water

matrix used. The influences of nanomaterial properties such as surface area, charge, and

photoactivity were elucidated and it was determined that higher surface area was correlated to

better adsorption of disinfection byproduct precursors and precursor surrogates while higher

photoactivity promoted more effective regeneration of the materials under UVA light. The two

LENs were less effective for the adsorption of DOC, UV254, and DBP precursors than standard

P25 nanoparticles, likely due to both surface area effects and charge effects as well as

interactions with the two water matrices used in this study. The removal of DOC and UV254 was

a good fit to a modified Freundlich adsorption isotherm model but this model was less

appropriate for the adsorption of DBP precursors. The adsorption of DBP precursors by the two

LENs was significant but modest (27% to 51%). Further modification of the materials

themselves or of the bulk matrix may be required if this technology is to be implemented

exclusively for DBP control in a drinking water treatment plant. The underlying concept of a

two-step treatment system built on a photocatalytically regenerable adsorbent that is easy to

remove from the water after treatment remains valid, however, and these materials may prove to

be highly effective for the removal of other contaminants of concern from drinking water and

other aqueous matrices upon further testing and development.

Acknowledgements

The authors would like to thank Kaya Yuan, Kennedy Santos, and Katie Dritsas for their

assistance in the laboratory, Jim Wang for training on the various analytical instruments

employed throughout the study, Alireza Mahdavi for the use of the Mastersizer 3000 particle

sizer, and Robert Liang for his assistance developing the bench-scale regeneration apparatus and

arranging some of the nanomaterial characterization tests.

253

References

Abdullah, M., Low, G.K.C., and Matthews, R.W. 1990. Effects of common inorganic anions on

rates of photocatalytic oxidation of organic carbon over illuminated titanium dioxide, Journal of





American Public Health Association, 2005. Standard Methods for the Examination of Water and


Bavykin, D.V. and Walsh, F.C. 2010. Titanate and Titania Nanotubes: Synthesis, RSC

Publishing

Bandala, E., Gonzalez, L., de la Hoz, F., Pelaez, M.A., Dionysiou, D.D., Dunlop, P.S.M., Byrne,

J.A., Sanchez, J.L., 2011. Application of azo dyes and dosimetric indicators for enhanced

photocatalytic solar disinfection (ENPHOSODIS), Journal of Photochemistry and Photobiology

A: Chemistry, 218, 185-191

Bennett, S.W., Zhou, D., Mielke, R., and Keller, A.A. 2012. Photoinduced disaggregation of

TiO2 nanoparticles enables transdermal penetration, PLOS ONE, 7 (11), e48719

Butler, E.C. and Davis, A.P. 1993. Photocatalytic oxidation in aqueous titanium dioxide

suspensions: the influence of dissolved transition metals, Journal of Photochemistry and

Photobiology A: Chemistry, 70, 273-283

Chen, H.Y., Zahraa, O., and Bouchy, M. 1997. Inhibition of the adsorption and photocatalytic



Chen, D. and Ray, A. K. 2001. Removal of toxic metal ions from wastewater by semiconductor


254

Daneshvar, N., Salari, D., Niaei, A., Rasoulifard, M.H., Khataee, A.R. 2005. Immobilization of

TiO2 nanopowder on glass beads for the photocatalytic decolorization of an azo dye C.I. Direct

Red 23, Journal of Environmental Science and Health, 40, 1605-1617

Erhayem, M., 2013. Effect of naturally occurring organic matter (NOOM) type and source on



Erhayem, M. and Sohn, M. 2014. Stability studies for titanium dioxide nanoparticles upon


Total Environment, 468-469, 249-257

Franch, M.I., Ayllón, J.A., Peral, J., and Domènech, X. 2005. Enhanced photocatalytic

degradation of maleic acid by Fe(III) adsorption onto the TiO2 surface, Catalysis Today, 101,

245-252

French, R. A., Jacobson, A.R., Kim, B., Isley, S.L., Penn, R.L., and Baveye, P.C. 2009. Influence

of ionic strength, pH, and cation valence on aggregation kinetics of titanium dioxide

nanoparticles, Environmental Science and Technology, 43, 1354-1359



Huang, X., Leal, M., Li, Q., 2008. Degradation of natural organic matter by TiO2 photocatalytic

oxidation and its effect on fouling of low-pressure membranes, Water Research, 42, 1142-1150

Hyung, H., Kim, J-H, 2008. Natural organic matter (NOM) adsorption to multi-walled carbon

nanotubes: Effect of NOM characteristics and water quality parameters, Environmental Science

and Technology, 42, 4416-4421

Karanfil, T., Kitis, M., Kilduff, J.E., Wigton, A., 1999. Role of granular activated carbon surface

chemistry on the adsorption of organic compounds 2, Environmental Science and Technology,

33, 3225-3233

Kasuga, T., Hiramatsu, M., Hoson, A., Sekino, T., and Niihara, K. 1999. Titania nanotubes

prepared by chemical processing, Advanced Materials, 11 (15), 1307-1311

255

Kim, S-H and Shon, H.K., 2007. Adsorption characterization for multi-component organic

matters by titanium oxide (TiO2) in wastewater, Separation Science and Technology, 42, 1775-

1792

Klinger, M. and Jäger, A. 2015. Crystallographic Tool Box (CrysTBox): automated tools for

transmission electron microscopists and crystallographers. Journal of Applied Crystallography,

48 (6), doi:10.1107/S1600576715017252.

Kumar, K.V., Porkodi, K., and Rocha, F. 2008. Isotherms and thermodynamics by linear and

non-linear regression analysis for the sorption of methylene blue onto activated carbon:

Comparison of various error functions, Journal of Hazardous Materials, 151, 794-804

Li, F., Yuasa, A., Ebie, K., Azuma, Y., Hagishita, T., Matsui, Y., 2002. Factors affecting the

adsorption capacity of dissolved organic matter onto activated carbon: Modified isotherm


Li, L., Sillanpää, M., and Risto, M. 2016. Influences of water properties on the aggregation and

deposition of engineered titanium dioxide nanoparticles in natural waters, Environmental

Pollution, 291, 132-138

Liang, L. and Singer, P.C. 2003. Factors influencing the formation and relative distribution of

haloacetic acids and trihalomethanes in drinking water, Environmental Science and Technology,

37, 2920-2928




Liu, S., Lim, M., Fabris, R., Chow, C.W.K., Drikas, M., Korshin, G., and Amal, R. 2010. Multi-


organic matter, Water Research, 44, 2525-2532

Liu, W., Sun, W., Borthwick, A., and Ni, J. 2013. Comparison on aggregation and sedimentation

of titanium dioxide titanate nanotubes and titanate nanotubes-TiO2: Influence of pH, ionic

256

strength, and natural organic matter, Colloids and Surfaces A: Physicochemical Engineering

Aspects, 434, 319-328

S. Liu, M. Lim, R. Amal, 2014. TiO2-coated natural zeolite: Rapid humic acid adsorption and

effective photocatalytic regeneration, Chemical Engineering Science 105, 46-52

Loosli, F., Vitorazi, L., Berret, J-F, and Stoll, S. 2015. Towards a better understanding on



Mwaanga, P., Carraway, E., and Schlautman, M. 2014. Preferential sorption of some natural

organic matter fractions to titanium dioxide nanoparticles: Influence of pH and ionic strength,


Ng, M., Kho, E.T., Liu, S., Lim, M., Amal, R., 2014. Highly adsorptive and regenerative


Journal, 246, 196-203

Perez Urquiza, M. and Beltran, J.L. 2001. Determination of the dissociation constants of

sulfonated azo dyes by capillary zone electrophoresis and spectrophotometry methods, Journal

of Chromatography A, 917, 331-336

Pifer, A.D. and Fairey,J.L. 2014, Suitability of organic matter surrogates to predict

trihalomethane formation in drinking water sources, Environmental Engineering Science, 31,

117-126




Qi, S. Schideman, L.C., 2008. An overall isotherm for activated carbon adsorption of dissolved

organic matter in water, Water Research, 42, 3353-3360

Shahbeig, H., Bagheri, N., Ghorbanian, S., Hallajisani, A., Poorkarimi, S., 2013. A new


Modelling and Simulation, 9, 243-254

257

Shi, H., Magaye, R., Castranova, V., and Zhao, J., 2013. Titanium dioxide nanoparticles: A

review of current toxicological data, Particle and Fibre Toxicology, 10:15

Summers, R. and Roberts, P., 1988. Activated Carbon Adsorption of Humic Substances:

Heterodisperse Mixtures and Desorption, Journal of Colloid and Interface Science, 122, 367-381



80-93

Sun, D.D. and Lee, P.F. 2012. TiO2 microsphere for the removal of humic acid from water:


Thio, B.J.T., Zhou, D., and Keller, A. 2011. Influence of natural organic matter on the

aggregation and deposition of titanium dioxide nanoparticles, Journal of Hazardous Materials,

189, 556-563

Yuan, Z-Y and Su B-L 2004. Titanium oxide nanotubes, nanofibres, and nanowires, Colloids


Zhang, Y., Chen, Y., Westerhoff, P., and Crittenden, J. 2009. Impact of natural organic matter





Zheng, D., Andrews, R.C., Andrews, S.A., Taylor-Edmonds, L. 2015. Effects of coagulation on


Water Research 70, 118-129

Zheng, Z., Liu, H., Ye, J., Zhao, J., Waclawik, E.R., and Zhu, H., 2010. Structure and



258

Zhou, C., Bashirzadeh, Y., Bernadowski, T.A., Zhang, X. 2016. UV light-induced aggregation of

titania submicron particles, Micromachines, 7, 203, doi: 10.3390/mi7110203

Zhou, D., Ji, Z., Jiang, X., Dunphy, D.R., Brinker, J., Keller, A.A. 2013. Influence of material

properties on TiO2 nanoparticle agglomeration, PLOS ONE, 8 (11), e81239

259


Figure 8.S.1 Zeta potential as a function of pH for two TiO2 LENs

-60

-50

-40

-30

-20

-10

0

10

20

30

2 3 4 5 6 7 8 9 10

Zet

a P

ote

nti

al (m

V)

pH

550-P 700-P

260

Figure 8.S.2 Time series data for AO24 removal by P25 nanoparticles and two LENs

Figure 8.S.3 Time series data for DOC removal by P25 nanoparticles and two LENs from

Otonabee River (OTB) and Ottawa River water (OTW)

2

3

4

5

6

7

0 60 120 180 240

DO

C (

mg/L

)

Adsorption Time (min)

P25-OTB 550-OTB 700-OTB

P25-OTW 550-OTW 700-OTW

0

2

4

6

8

10

0 20 40 60 80 100 120

AO

24

(m

g/L

)

Time (min)

P25 NB 550 NB 700

261

Figure 8.S.4 UV254 isotherms in (A) Otonabee River (OTB) water and (B) Ottawa River

(OTW) water

0.01

0.1

1

0.01 0.1 1 10

log

qA

(cm

-1/g

TiO

2)

log A/D (cm-1/g TiO2)


0.01

0.1

1

0.01 0.1 1 10

log q

A (

cm-1

/g T

iO2)

log A/D (cm-1/g TiO2)


B

A

262

Figure 8.S.5 HAAfp isotherms in Otonabee River (OTB) water and Ottawa River (OTW)

water

1

10

100

1000

10 100 1000

q (

ug

HA

Afp

/g T

iO2)

C/D (ug HAAfp/g TiO2)


1

10

100

1000

10 100 1000 10000

q (

ug H

AA

fp/g

TiO

2)

C/D (ug HAAfp/g TiO2)


B

A

263

Figure 8.S.6 Concentration of AO24 in water treated with virgin and regenerated LENs

Figure 8.S.7 UV254 of OTB and OTW water treated with virgin and regenerated LENs

0.00

2.00

4.00

6.00

8.00

10.00

NB 550 NB 700

Co

nce

ntr

ati

on

of

Aci

d O

ra

ng

e

(mg

/L)

Regeneration Cycle

0

1

2

3

4

5

0.00

0.05

0.10

0.15

0.20

NB 550 - OTB NB 700 - OTB NB 550 - OTW NB 700 - OTW

UV

254 (

1/c

m)

0

1

2

3

4

5

264

Figure 8.S.8 Ratio of TCM to BDCM in surface water treated with increasing doses of

TiO2

Figure 8.S.9 Ratio of DCAA to TCAA in surface water treated with increasing doses of

TiO2

0

2

4

6

8

10

12

0 0.25 0.5 0.75 1 1.25 1.5

TC

M:B

DC

M

TiO2 Dose (g/L)

P25 - OTB NB 550 - OTB NB 700 - OTB P25 - OTW NB 550 - OTW NB 700 - OTW

0

0.5

1

1.5

2

2.5

0 0.25 0.5 0.75 1 1.25 1.5

DC

AA

:TC

AA

TiO2 Dose (g/L)

P25 - OTB NB 550 - OTB NB 700 - OTB P25 - OTW NB 550 - OTW NB 700 - OTW

265

Summary, Conclusions, Engineering Significance, and Implications of Research

The goal of this research was to lay the groundwork for the eventual development of a TiO2-

based drinking water treatment system. The three main challenges that prevent the use of TiO2

for drinking water treatment are:

1. How do we provide light of the appropriate wavelength (<385 nm) and intensity in a

reliable and energy efficient manner?

2. How do we avoid the formation of potentially dangerous intermediate compounds?

3. How do we remove the photocatalyst from the water after treatment?

The following sections summarize the findings of the studies undertaken to address these

challenges, the main conclusions that can be draw from them, their engineering significance, and

implications for future research.

Summary of Findings

The first of the three challenges identified above (i.e. provide light of the appropriate irradiance

and wavelength) was addressed early in the project when a decision was made to use UVA LEDs

as the irradiation source. This decision was based on initial experimental findings, a thorough

literature review, and the increased availability and cost competitiveness of UVA LEDs on the

market. Possible solutions to the two remaining challenges were developed experimentally based

on the specific objectives identified in Section 1.2:

1. Explore the use of standard TiO2 nanoparticles for NOM and DBP precursor removal via

adsorption and photocatalytic degradation.

Preliminary experiments indicated that commercial P25 nanoparticles were able to remove DBP

precursor surrogates such as DOC and UV254 from both synthetic and real water matrices

(Chapter 4). The extent of adsorption and rate of photocatalytic degradation of NOM were both

found to be lower in real water matrices than in a synthetic river water matrix containing

Suwannee River NOM isolate obtained from the International Humic Substances Society. The

266

preliminary experiments also established appropriate ranges for TiO2 dosing, irradiation time,

and other experimental parameters for NOM removal via adsorption and photocatalysis.

A more detailed study was developed based on the preliminary findings and is described in

Chapter 5 of this document. Photocatalytic treatment of Otonabee River water with 0.25 g/L of

P25 nanoparticles increased the THMfp of the water by 88% after 15 minutes of irradiation. This

was followed by a gradual decrease in THMfp with increasing irradiation time but after 60

minutes of irradiation the THMfp of the water was only 10% below that of the untreated water.

During adsorption, aromatic NOM (as measured by UV254), was preferentially removed over

non-aromatic NOM and the efficiency of NOM adsorption to TiO2 varied by water source. TiO2

nanoparticles preferentially adsorbed larger NOM molecules including the biopolymers and

humic substances LC-OCD fractions. pH was shown to have a strong impact on the removal of

NOM, including DBP precursors, from surface water by TiO2 nanoparticles. Specifically, more

adsorption occurred at low pH than at higher pH. This is similar to results presented by

researchers studying NOM adsorption to TiO2 in non-treatment contexts (Mwaanga et al., 2014).

The poorer adsorption observed at pH 6 and pH 8 may be related to both agglomeration and

charge repulsion at higher pH, with the former dominating at pH 6 and the latter at pH 8.

A modified version of the Freundlich isotherm model provided an excellent fit to the DOC data

gathered in this study. The resulting isotherm parameters were within but at the low end of the

range usually observed during NOM adsorption to GAC and carbon nanomaterials, indicating

that, particularly at neutral pH, the TiO2 nanoparticles were less effective than the adsorbents

currently used in drinking water plants. The THMfp and HAAfp datasets were also fitted to the

modified Freundlich model, with generally positive results. The results presented in Chapter 5

show that TiO2 adsorption is a viable way to remove NOM and DBP precursors from drinking

water and that this removal can be modeled using simple isotherm models.

267

2. Develop engineered nanomaterials that are easy to remove from the water via

conventional water treatment clarification processes but retain the adsorptive and

photocatalytic properties of standard TiO2 nanoparticles.

Three sets, or generations, of linear engineered TiO2 nanomaterials (LENs) were synthesized

over the course of this project using variations on a simple hydrothermal synthesis method

originally developed by Kasuga et al. (1999) and later modified by Yuan and Su (2004) and

others (see Appendix A). The first generation of LENs, which is described in Chapter 4, was

based directly on the findings of Yuan and Su (2004). This generation of LENs included

nanotubes, nanobelts, and nanowires created by varying the alkaline precursor solution (NaOH

vs. KOH), the hydrothermal temperature (TH), and the final calcination temperature (TC) of the

synthesis procedure. SEM imaging and XRD analysis suggested that these materials differed

from one another mainly in terms of size and length to width ratio, though they also had different

degrees of reactivity towards indicator dyes (methylene blue and Acid Orange 24) and NOM.

The second generation LENs were variations of the nanobelt and nanotube materials and differed

from one another and from industry standard nanoparticles in terms of size, BET surface area,

and other physical and chemical characteristics. The second generation LENs varied

substantially in terms of their ability to degrade methylene blue dye and as well as their ability to

adsorb and degrade DBP precursor surrogates such as DOC and UV254. The variation was

related to surface area, charge, propensity to agglomerate, crystal phase, and the presence of

defects within the crystal structure. The adsorption and degradation rates were particularly

influenced by the surface area and crystallinity of the nanomaterials. The second generation

LENs settled out of MilliQ water at natural pH (5 to 6) much more quickly than standard P25

nanoparticles.

A subset of the second generation of LENs was selected for further characterization and

evaluation. This final generation of LENs, NB 550 and NB 700, were subjected to an additional

rinsing step and manufactured in sufficient quantity to conduct larger scale DBP removal

experiments. The third generation LENs were both approximately 50 to 100 nm in diameter and

1 to 10 m long but differed in terms of crystallinity, surface area, and surface appearance. They

both settled out of MilliQ water at natural pH (5 to 6) quickly and presented less of a barrier to

filtration relative to commercial P25 nanoparticles.

268

3. Evaluate the use of the linear engineered nanomaterials for DBP precursor removal from

real water matrices via photocatalytic degradation.

The third generation LENs were compared to standard commercial P25 TiO2 nanomaterials in

terms of their ability to degrade disinfection byproduct precursors in two natural surface water

matrices obtained from water treatment plants in Ontario. The filterability and settleability of the

three materials in these water matrices were also evaluated. The results of this study are provided

in Chapter 7 of this thesis. Although all three materials reduced DOC and UV254 even at short

irradiation times, the THMfp and HAAfp of the treated water initially increased upon irradiation

with UVA LED light irrespective of the material or water source used. This is similar to results

published by Liu et al. (2008), Huang et al. (2008), and Gerrity et al. (2009), though all of these

researchers restricted their research to commercial P25 nanoparticles. In this study, the increase

in THMfp usually peaked with 15 minutes of irradiation and decreased as irradiation time was

increased beyond this time. After 60 minutes one of the LENs, NB 700, removed more than 90%

of the THMfp and HAAfp from the Ottawa River water. DBPfp reduction was more modest in

the Otonabee River water matrix, likely because this water source contained ionic species

capable of scavenging reactive oxygen species and/or inducing the nanomaterials to agglomerate,

reducing the surface area available for reaction. DOC and UV254 removal by the LENs was

reasonably well correlated to methylene blue degradation but THMfp and HAAfp removals were

not.

4. Evaluate the use of the linear engineered nanomaterials for DBP precursor removal from

real water matrices via adsorption.

The third generation LENs were also evaluated as two regenerable nanoscale adsorbents. Their

ability to remove an indicator dye and disinfection byproduct precursors from two natural

drinking water sources was assessed using adsorption isotherm experiments. The two LENs were

less effective for the adsorption of DOC, UV254, and DBP precursors than standard P25

nanoparticles, likely due to both surface area effects and charge effects as well as interactions

with the two water matrices used in this study. The removal of these parameters by P25

nanoparticles and by one of the LENs (NB 700) was a good fit to a modified Freundlich

269

adsorption isotherm model but this model did not adequately describe the adsorption of DOC by

the other LEN, NB 550. It was hypothesized that this was due to the distinct surface properties of

NB 550, however, more work is required to confirm that this is the case and to establish which

isotherm model, if any, can be used to describe the adsorption of NOM to this LEN. The

influences of nanomaterial properties such as surface area, charge, and photoactivity were

elucidated and it was determined that higher surface area was correlated to better adsorption

while higher photoactivity promoted more effective regeneration of the materials under UVA

light.

Overall Conclusions

9.2.1 TiO2 Removes Disinfection Byproduct Precursors via Adsorption and Degradation

The findings of this project confirm that DBP precursors can be removed by TiO2 via adsorption

and broken down by TiO2 via photocatalysis. Both commercial P25 nanoparticles and lab

synthesized LENs were able to remove significant amounts of NOM via photocatalysis at TiO2

doses ranging from 0.1 g/L to 0.5 g/L and via adsorption at doses ranging from 0.1 g/L to 1.5

g/L. The adsorption capacity of each nanomaterial was impacted by its available surface area –

materials with higher specific surface areas such as P25 nanoparticles or NB 550, one of the third

generation LENs, adsorbed more NOM than those with lower specific surface areas. It was also

affected by the surface charges of the nanomaterial and the NOM. The more photoactive

materials (e.g. NB 700) were most effective for NOM and DBP precursor degradation.

Throughout this study, the characteristics of the water matrix had important effects on the

removal of DBP precursors via degradation and adsorption by the various the TiO2

nanomaterials. Degradation proceeded more slowly in the water matrix that contained higher

levels of ions, in particular bicarbonate (alkalinity) and calcium, possibly because some of these

ions can act as ROS scavengers but also perhaps because higher concentrations of ions can

compress the electrical double layer that surrounds the nanoparticles, encouraging agglomeration

and an overall reduction in the surface area available for reaction (Liu et al., 2013). This finding

indicates that a photocatalysis-based single step TiO2 treatment process will not be appropriate

270

for all communities and highlights the need for comprehensive and site-specific evaluation of

new engineered nanomaterials and other advanced oxidation processes ahead of their

implementation for drinking water treatment.

9.2.2 Material Synthesis Conditions Determine the NOM Adsorption and Degradation Behaviour of LENs

A total of nine LENs were synthesized over the course of this project. Two of these, NB 550 and

NB 700, were eventually chosen as the best candidates for possible integration into a novel water

treatment process, however, the process of developing the materials also yielded many

interesting findings. The temperature applied during the hydrothermal step (TH) governed the

size of the resulting LENs and indirectly impacted their surface area. The temperature used for

calcination (TC) determined the crystallinity of the LENs and also indirectly impacted their

surface area. Crystallinity (the types and proportion of anatase, rutile, and other TiO2 crystal

phases present in the material) was the main determinant of photocatalytic activity while the

amount of available surface area was found to govern adsorptive behavior. These findings, as

well as those of others (see Appendix A) provide a road map towards the development of new

adsorptive, photocatalytically active, and easily removable LENs for drinking water treatment.

9.2.3 Filtration is the Most Practical Option for Nanomaterial Removal

LENs were more easily removed from purified water and natural water via filtration than

commercial P25 nanoparticles under all of the conditions studied, likely because their larger size

prevented them from becoming deposited or enmeshed in the filter pores. They were also, for the

most part, easier to remove via settling, though this was more apparent in purified water and one

of the two natural surface water matrices. P25’s resistance to filtration was water matrix specific,

possibly because the different water matrices encouraged the formation of different size P25

agglomerates, some of which were small enough to cause pore constriction and others that were

large enough to be fully excluded from the pores. The LENs presented the same low resistance to

filtration in all of the water matrices examined, suggesting that the LENs were large enough to

be removed irrespective of the degree of agglomeration.

In contrast, sedimentation rates were matrix specific for all three nanomaterials. It was

hypothesized that the size and shape of the different materials and their interactions with various

271

components of the water matrices affected the size and effective density of the agglomerates

formed by each of the nanomaterials. This in turn determined how quickly the nanomaterials

would settle out of the water. For example, the presence of calcium has been linked to increased

agglomeration due to electrical double layer (EDL) compression (Liu et al., 2013) and indeed, all

three nanomaterials settled quickly in this water matrix. EDL compression was confirmed using

zeta potential measurements. The materials, particularly the LENs, settled poorly in the OTW

water matrix, which contained high levels of NOM and low levels of ions. Other researchers

have hypothesized that steric repulsion can prevent the agglomeration of NOM coated

nanomaterials (Thio et al., 2011), and this may also have occurred in this study. Attempts were

made to model settling behaviour using Stokes’ Law and the Sterling equations for the

calculation of effective agglomerate density together with particle size distribution data obtained

at a lower TiO2 dose (see Appendix H). Overall, the sedimentation results indicate that settling

would not be an ideal removal mechanism for LENs in high throughput applications such as

drinking water treatment because it is slow, matrix dependent, and difficult to model or optimize

without access to information such as agglomerate size and effective density.

Engineering Significance of Findings

The results presented in this thesis essentially serve as preliminary proof of concept for two

TiO2- based water treatment processes (see Appendix E for conceptual schematics). Of these, the

two-step adsorption and regeneration process with membrane separation is the most likely

candidate for further development, though more work is required to confirm that the processes

will be able to remove other contaminants besides dyes and DBP precursors and that they will

function in a bench-scale flow through configuration and, eventually, at pilot and full-scale.

One of the important practical findings of this study was that further modification of the

materials themselves or of the bulk matrix may be required if this technology is to be

implemented as the primary DBP control strategy in a drinking water treatment system. The

underlying concept of a two-step treatment process built on a photocatalytically regenerable

adsorbent that is easy to remove from the water after treatment remains valid, however, and these

materials may prove to be highly effective for the removal of other contaminants of concern

from drinking water.

272

Given the potential negative impacts of TiO2 nanomaterials on human and environmental health,

it is imperative that any novel process employing such materials for water or wastewater

treatment ensure the complete removal of the materials from the treated water before it is sent to

the drinking water distribution system or returned to the environment. The LENs synthesized in

this study fall within the size range of dangerous “fibrous dusts” as defined by the World Health

Organization (diameter < 3 m, aspect ratio > 3:1) and, in some cases, the size range of

“respirable fibres” as defined by the United States Centres for Disease Control (length > 5 m,

diameter ≤ 1.3 m), and as such may present a human health hazard when they are not

suspended in water (WHO, 1999; CDC, 2006). As a result, if a LEN-based treatment system is

implemented, personal safety equipment will need to be provided to prevent operators from

being exposed to potentially toxic levels of LENs via inhalation in the treatment plant.

Implications for Future Research

The main findings of this study are promising and suggest that a LEN-based two-step adsorption

and photocatalytic regeneration process (Figure E.2 in Appendix E) may one day evolve into a

feasible option for drinking water treatment. More research is required to optimize and scale-up

the LEN synthesis procedure, to explore additional niches in the water and wastewater industry

where such a system might be appropriate, to design and test the adsorption and regeneration

reactors, and to confirm that the LENs can in fact be integrated with membrane separation in a

safe and cost effective manner. A framework for the research required to develop a prototype of

the proposed treatment process is provided in Figure 9.1.

The cost analysis presented in Appendix F suggests that the cost of the materials is the most

important factor driving the overall cost of the proposed treatment process. This is, to some

extent, simply a function of the quality of the costing information available at this time. The

pricing used to estimate material costs in this project was obtained from a company, Novarials,

that specializes in lab grade nanomaterials for research institutions. Their prices may not be

representative of the entire current or future market for these types of materials. Nonetheless,

based on the cost estimates, the optimization and scaling up of the nanomaterial synthesis

procedure should be a special priority in any future research project as it will determine the

273

overall affordability of the resulting system. This will likely involve additional optimization of

the synthesis process to reduce energy requirements for the process and of the nanomaterials

themselves to increase their surface area and photoactivity and thus improve their adsorptivity

and regnerability.

Figure 9.1 Framework for the development of a prototype of a two-step adsorption and

photocatalytic process for drinking water treatment

The proposed treatment system may be appropriate for other applications, including other

drinking water treatment applications. For example, Fotiou et al. (2015) showed that TiO2 can

remove MC-LR, a cyanotoxin, via photocatalysis and, to a lesser extent, adsorption. Their

experiments were conducted at relatively low doses of TiO2, and it might be possible to adsorb

greater amounts of MC-LR at higher doses. In theory, the proposed treatment process may also

work well as a polishing step for the removal of DBP precursors that are recalcitrant to removal

via conventional coagulation-based treatment. These alternative applications may pivot the focus

274

of the project in another direction or simply expand the suite of applications that can be

addressed by the proposed treatment process.

The experiments in this project were all conducted in batch mode, but water treatment systems

work in a dynamic flow through mode. Additional work will need to be done to determine the

proper design parameters for the adsorption and regeneration reactors. For example, the current

concept calls for a closed, serpentine reactor studded with UVA LEDs for the regeneration step.

The size of the channels and the flow rate through the reactor will determine the degree of

hydrodynamic mixing as well as the amount of contact between the UVA light and the LENs.

Additional chemical inputs may also prove to be helpful to encourage better adsorption (e.g. pH

depression) or to aid in the regeneration process.

Finally, based on the results of this project, membrane filtration appears to be the most feasible

option for material separation after adsorption and regeneration. The most appropriate type of

membrane and the fouling effects of the nanomaterials on that chosen membrane will need to be

determined to ensure that the process can be operated over a long period of time.

References

Centers for Disease Control and Prevention, Department of Health and Human Services,

National Institute for Occupational Safety and Health (2006) Occupational Exposure to

Refractory Ceramic Fibers, Criteria for a Recommended Standard, DHHS (NIOSH) Publication

No. 2006-123







Journal, 261, 17-26

275



Research, 43, pp. 1597-1610



pp. 1142-1150


















World Health Organization (1999) Hazard prevention and control in the work environment,

WHO/SDE/OEH/99.14, Geneva



276

Appendix A: Effects of Synthesis Conditions on LEN Characteristics

Table A.1 Summary of LEN synthesis studies

Reference Precursors

Synthesis

Temp (oC) and

Time

Calcination

Temp (oC)

and Time

Length (nm) Width

(nm)

Surface

Area

(m2/g)

Crystal Phase (s) Band Gap

Energy (eV) Other

Kasuga et al.

(1999)

n.s., 10 mM NaOH 110 / 20 h -- 100 8 257

246

“Four coordinate Ti-

O”

Anatase

n.s.

n.s.

Sample C1

Sample D1

Yuan and Su

(2004)

Anatase, 10 M

NaOH

P25, 10 M NaOH

Anatase and P25, 4

– 25 M KOH

100 / 1-2 days

150 / 1-2 days

200 / 1-2 days

220 / 1-2 days

220 / 1-2 days

220 / 1-2 days

90 / 1-2 days

140 / 1-2 days

140 / 1-2 days

200 / 1-2 days

130 – 240 / 1-2

days

--

--

--

--

540 / 2h

700 / 2h

--

--

540 / 2 h

--

--

400 / 2.5 h

600 / 2.5 h

700 / 2.5 h

1000 / 2.5 h

10 – 100+

10 – 100+

10 – 100+

100 – 1000+

100 – 1000+

100 – 1000+

10 – 100+

10 – 100+

50

10 – 100+

1000 – 10000

1000 – 10000

1000 – 10000

1000 – 10000

1000 – 10000

8 – 10

8 – 10

8 – 10

50 – 300

50 - 300

50 - 300

8 – 10

8 – 10

8 – 10

8 – 10

5 – 10

5 – 10

5 – 10

5 – 10

5 – 10

130

210

100

n.s.

n.s.

n.s.

50

325

n.s.

150

250 – 320

250 – 320

250 – 320

250 – 320

250 – 320

Anatase / trititanate



H2Ti5O11H2O

TiO2(B)

Anatase



n.s.


n.s.

K2Ti8O17

K2Ti8O17 / anatase

K2Ti6O13 / anatase

K2Ti6O13 / rutile

n.s.

n.s.

n.s.

n.s.

n.s.

n.s.

n.s.

n.s.

n.s.

n.s

n.s.

n.s.

n.s.

n.s.

n.s.

30% yield

85% yield

Nanoribbon

Nanoribbon

Nanoribbon

Nanotube

Nanotube

Nanorod

Nanotube

Nanowire

Nanowire

Nanowire

Nanowire

Nanowire

Qamar et al.

(2008)

TiOCl2, 10 M NaOH 150 / 48 h --

300 / 2 h

500 / 2 h

700 / 2 h

900 / 2 h

100+

100+

100+

--

100 – 500

6 – 10

6 – 10

6 – 10

--

50 – 200

329

252

109

33

4.2

Trititanate

Trititanate / anatase

Anatase

Anatase / rutile

Rutile / hexatitanate

n.s.

n.s.

n.s.

n.s.

n.s.

Small grains

277

1Samples C and D refer to samples that have been subjected to different washing regimes after the ion exchange step (C = less washing, D = more

washing)

n.s. = not specified

Reference Precursors

Synthesis

Temp (oC) and

Time

Calcination

Temp (oC)

and Time

Length (nm) Width

(nm)

Surface

Area

(m2/g)

Crystal Phase (s) Band Gap

Energy (eV) Other

Bavykin et al.

(2010)

Anatase NaOH and

KOH

100 / 48 h --

400 / 24 h

n.s n.s. 250

180

Titanate

TiO2(B)

n.s.

n.s.

Ali et al. (2016) Antase 10 M NaOH 110 / 21 h --

300 / 3 h

400 / 3 h

500 / 3 h

600 / 3 h

700 / 3 h

60 – 100+

60 – 100+

60 – 100+

60 – 100+

60 – 100+

60 – 100+

8 – 10

8 – 10

8 – 10

8 – 10

8 – 10

8 – 10

157

117

74

53

38

21





Hexatitanate / anatase

Hexatitanate / anatase

3.13

3.10

3.09

3.08

3.07

3.06

Liu et al. (2013) P25 10 M NaOH 130 / 72 h -- 100+ 8 262 Trititanate n.s. IEP = 3.5

Zheng et al.

(2010)

Anatase 10 M

NaOH

180 / 48 h 300 / 4 h

400 / 4 h

500 / 4 h

550 / 4 h

600 / 4 h

650 / 4 h

700 / 4 h

1000+

1000+

1000+

1000+

1000+

1000+

1000+

100 – 200

100 – 200

100 – 200

100 – 200

100 – 200

100 – 200

100 - 200

26

25

25

23

20

18

16

TiO2 (B)

TiO2 (B)

TiO2 (B)

TiO2 (B) / anatase

TiO2 (B) / anatase

TiO2 (B) / anatase

Anatase

3.041

3.046

3.054

3.091

3.183

3.185

3.186

Optimal

Seo et al. (2009) P25 10 M NaOH 160 / 24 h

200 / 24 h

230 / 24 h

600 / 1 h

600 / 1 h

600 / 1 h

n.s.

1000+

1000+

n.s.

50

50

76

23

23

Anatase

Anatase / H2Ti3O7

Anatase / H2Ti3O7

n.s.

n.s.

n.s.

Optimal

Zhang et al.

(2009)

P25 10 M KOH

P25 10 M NaOH

180 / 48 h

180 / 48 h

600 / n.s.

600 / n.s.

100+

1000+

10

20-100

77

37

Anatase

Anatase

n.s.

n.s.

278

Table A.2 Summary of the effects of LEN synthesis conditions on LEN characteristics

Parameter Alkaline Solution Hydrothermal Temperature (TH) Calcination Temperature (TC)

Size/Shape NaOH: Length and width determined

by TH

KOH: Increased length, decreased

width relative to NaOH materials

Lower TH results in smaller nanotube

or nanorod structures

Higher TH results in larger nanobelts

or nanoribbon structures

Minimal independent effect

High TC can result in

sintering/breakdown of some LENs

formed at lower TH

Surface Area NaOH: Lower, affected by TH and TC

KOH: High relative to NaOH

materials

Highest surface area obtained

between TH of 120oC and 180oC

Lower surface area at higher and

lower TH values

Increasing TC results in decreased

surface area

Crystal Phase NaOH: Na-based titanates at some TC

values

KOH: K-based titanates at some TC

values

Minimal effect before calcination

TH can affect the crystal structure

formed after calcination in some

cases

TC is the main driver of crystal structure

TC = 300 oC – 500 oC = titanates

TC = 500 oC – 700 oC = mixed phase

(anatase + TiO2(B) or titanates)

TC = 700 oC – 900 oC = anatase

TC > 900 oC = rutile

Reactivity Unknown TH can affect the crystal structure

formed after calcination and thus the

overall reactivity

Anatase and mixed phase

anatase/TiO2(B), formed at TCs ranging

from 500oC to 900oC, are the most

reactive forms of TiO2

279

Parameter Alkaline Solution Hydrothermal Temperature (TH) Calcination Temperature (TC)

Band Gap Unknown Unknown Effect is dependent on the structure of

the materials formed during

hydrothermal synthesis step

For nanobelts: Increased TC results in

increased band gap energy

For nanotubes/nanorods: Increased TC

results in lower band gap energy

Other Hydrothermal time can affect the

extent of reaction

Calcination time can affect conversion

Extent of washing after ion exchange

can impact crystal phases, especially at

TCs ranging from 300oC to 700oC

280

Appendix B: Matrix Impacts on Adsorption and Photocatalytic Degradation of NOM by TiO2

Table B.1 Matrix effects on NOM adsorption onto TiO2 surface

Parameter Known Effects on Adsorption Source

pH Increased adsorption of NOM onto P25 at low pH Mwaanga et al.

(2014); Erhayem and

Sohn (2014); Valencia

et al. (2012); Patsios

et al. (2012)

Alkalinity More bicarbonate results in less adsorption

Adsorbs to TiO2 and may compete with NOM for

adsorption sites

Erhayem and Sohn

(2014)

Chen et al. (1997)1

Ionic strength Higher ionic strength (ionic strength =

concentration of NaNO3 added to solution) leads to

greater adsorption of NOM onto P25

Higher ionic strength results in more adsorption

Higher ionic strength encourages agglomeration

and subsequent loss of available surface area

Mwaanga et al. (2014)

Erhayem and Sohn,

(2014)

Liu et al. (2013),

Hotze et al. (2010),

Thio et al. (2011)

Concentration of NOM Higher initial NOM concentration results in greater

NOM adsorption

Mwaanga et al.

(2014), Erhayem and

Sohn, (2014), Kim

and Shon (2007)

Type of NOM Humic substances / aromatic compounds are more

readily adsorbed than other NOM

Erhayem and Sohn

(2014)

Calcium Calcium increases the extent of NOM adsorption

onto TiO2

The presence of calcium ions reduces the

electrostatic repulsion between TiO2 and NOM

Calcium makes NOM more hydrophobic, thus

increasing adsorption.

Erhayem and Sohn

(2014)

Sun et al. (2012), Liu

et al. (2013)

Sun et al. (2012)

281

Parameter Known Effects of Adsorption Sources

Magnesium Increased magnesium results in increased NOM

adsorption

Erhayem and Sohn

(2014)

Sodium Minimal change observed Erhayem and Sohn

(2014)

Potassium Minimal change observed Erhayem and Sohn

(2014)

Phosphate More phosphate results in less NOM adsorption


adsorption sites

Erhayem and Sohn

(2014)

Chen et al. (1997)1

Nitrate More nitrate results in less NOM adsorption


adsorption sites

NaNO3 was added to water to increase ionic

strength and seems to have increased the amount of

NOM adsorbed to P25

Erhayem and Sohn

(2014)

Chen et al. (1997)1

Mwaanga et al. (2014)

Sulphate Adsorbs to TiO2 and may compete with NOM for

adsorption sites

Chen et al. (1997)1

Chloride Minimal change observed


adsorption sites

Erhayem and Sohn

(2014)

Chen et al. (1997)1

Iron Iron (Fe(III)) is adsorbed by TiO2 and may compete

with NOM for adsorption sites

Chen and Ray (2001),

Luck (2007)

Manganese Manganese is adsorbed by TiO2 and may compete

with NOM for adsorption sites

Luck (2007)

1Authors note that results may be pH-specific

282

Table B.2 Matrix effects on the photocatalytic degradation of NOM by TiO2

Parameter Known Effects on Degradation Source

pH Greater degradation at higher pH

Degradation is better at pH 5.5 than pH 3.5 or

pH 7

Liu et al. (2008B)

Patsios et al. (2012)

Alkalinity Decreases degradation by scavenging OH

radicals

Carbonate/bicarbonate ion reduced overall

degradation rate of model pollutant

Decreases degradation by encouraging

agglomeration/reducing available surface area

Liao et al. (2001)

Chen et al. (1997)1

Autin et al. (2013)

Ionic Strength Higher ionic strength leads to greater

agglomeration, which reduces overall available

surface area

Liu et al. (2013), Thio et

al. (2011), Hotze et al.

(2010)

Concentration of NOM Faster removal at lower NOM concentrations,

possibly due to greater contribution of

adsorption to overall removal at lower

concentrations

Better removal at higher NOM concentrations

Huang et al. (2008)

Patsios et al. (2012)

Type of NOM Large and aromatic NOM compounds are

targeted for degradation

Liu et al., (2008A), Liu et

al. (2010), Huang et al.

(2008)

Phosphate Phosphate reduced the photocatalytic

degradation rate of a model pollutant

Phosphate adsorbed to the TiO2 surface

preventing the adsorption and degradation of

NOM

Chen et al. (1997)1, Burns

et al. (1999)

Abdullah et al., 1990

283

Parameter Known Effects on Degradation Source

Nitrate Nitrate had a modest inhibitory effect on the

photocatalytic degradation rate of a model

pollutant


et al. (1999)

Chloride Chloride is an OH radical scavenger and as

such may reduce the extent of NOM

degradation during photocatalysis

Chloride ion reduced overall degradation rate

of model pollutant

Liao et al. (2001)


et al. (1999)

Copper Increased degradation of model compound

observed in the presence of Cu(II) at low

concentrations at pH 3

Butler and Davis (1993)

Iron Increased rate of degradation for model

compound observed in the presence of low

concentrations of Fe(III) at pH 3

Decreased rate of degradation for model

compound observed in the presence of high

concentrations of Fe(III) due to competitive

surface reactions and/or increased water

opacity

Inhibition of degradation of model compound

due to catalyst fouling by iron



Burns et al. (1999)

Manganese Inhibition of degradation of model compound

due to catalyst fouling by manganese

Burns et al. (1999)

1Authors note that results may be pH-specific

284

Appendix C: Calibration Curves

This appendix contains a representative set of calibration curves for:

Methylene blue dye (concentration vs. absorbance)

Acid Orange 24 dye (concentration vs. absorbance)

Total organic carbon (concentration vs. area count)

Four THMs (response ratio vs. area count)

Nine HAAs (response ratio vs. area count)

TiO2 dose vs. turbidity

TiO2 dose vs. UV-Vis absorbance at 375 nm

Figure C.1 Representative calibration curve for methylene blue dye

y = 0.1871x + 0.0136

R² = 0.9982

0

0.5

1

1.5

2

0 1 2 3 4 5 6 7 8 9 10

Ab

sorb

an

ce a

t 66

5 n

m (

1/c

m)

Concentration of Methylene Blue (mg/L)

285

Figure C.2 Representative calibration curve for Acid Orange 24

Figure C.3 Representative calibration curve for TOC

y = 0.021x + 0.0009

R² = 0.9974

0

0.05

0.1

0.15

0.2

0.25

0 1 2 3 4 5 6 7 8 9 10

Ab

sorb

an

ce a

t 4

40

nm

(1

/cm

)

Concentration of Acid Orange 24 (mg/L)

y = 4169.8x + 706.51

R² = 0.9991

0

10000

20000

30000

40000

50000

0 1 2 3 4 5 6 7 8 9 10

Are

a C

ou

nt

DOC (mg/L)

286

Figure C.4 Calibration curves for four THMs (Summer 2016)

Table C.1 Parameters and fits of calibration curves (THMs)

THM Species Slope Intercept R2

TCM 0.596 -0.075 0.988

BDCM 2.830 0.157 0.996

CDBM 3.202 -0.273 0.995

TBM 1.376 -0.244 0.989

-5

0

5

10

15

20

25

30

0 1 2 3 4 5 6 7 8

Res

po

nse

Ra

tio

Concentration : IS

TCM

BDCM

CDBM

TBM

287

Figure C.5 Calibration for nine HAA species (Summer 2016)

Table C.2 Parameters and fits of calibration curves (HAAs)

HAA Species Slope Intercept R2

MCAA 130.9 -0.045 0.999

MBAA 942.1 -0.720 0.992

DCAA 1051.9 0.048 0.999

TCAA 4305.1 -1.821 0.999

BCAA 2815.5 -0.284 0.998

DBAA 3462.0 0.932 0.990

BDCAA 4341.5 -5.392 0.993

CDBAA 2949.6 -3.289 0.994

TBAA 2031.2 -2.374 0.992

-20

0

20

40

60

80

100

120

140

0 0.005 0.01 0.015 0.02 0.025 0.03

Res

po

nse

Ra

tio

Concentration : IS

MCAA

MBAA

DCAA

TCAA

BCAA

DBAA

BDCAA

CDBAA

TBAA

288

Figure C.6 Calibration curves for P25 and LENs vs. turbidity

Table C.3 Parameters and fits of calibration curves (TiO2 vs. turbidity)

Nanomaterial Slope Intercept R2

P25 4.84 15.2 0.996

130/550 4.85 -41.5 0.976

130/700 3.91 -75.7 0.949

240/550 4.84 -22.5 0.994

240/700 11.9 37.9 0.998

0

500

1000

1500

2000

2500

3000

3500

4000

0 50 100 150 200 250 300

Tu

rbid

ity

(N

TU

)

TiO2 (mg/L)

P25

130/550

130/700

240/550

240/700

289

Figure C.7 Calibration curves for P25 and LENs vs. UV absorbance at 375 nm

Table C.4 Parameters and fits of calibration curves (TiO2 vs. UV375)

Nanomaterial Slope Intercept R2

P25 0.0241 0.0156 0.999

130/550 0.0106 0.0002 0.997

130/700 0.0037 0.0088 0.987

240/550 0.0172 0.0631 0.987

240/700 0.0143 0.0050 0.964

0

0.5

1

1.5

2

2.5

3

3.5

4

0 50 100 150 200 250

UV

Ab

sorb

an

ce a

t 3

75

nm

TiO2 Dose (mg/L)

P25

130/550

130-700

240/550

240/700

290

Figure C.8 Calibration curves for HTPA vs. fluorescence (Ex: 315 nm, Em: 425 nm)

y = 119.37x + 5.0928

R² = 0.9995

0

200

400

600

800

0 1 2 3 4 5 6 7

Res

pon

se

HTPA (uM)

291

Appendix D: Quality Control

Figure D.1 Quality control results for batches of second generation LENs

Figure D.2 Quality control results for batches of third generation LENs

0%

20%

40%

60%

80%

100%

1 2 3 4

Dec

olo

uriz

ati

on

of

Met

hy

len

e B

lue

Batch

130/550 130/700 240/550 240/700

0%

20%

40%

60%

80%

100%

0 2 4 6 8 10 12 14 16 18

Dec

olo

uri

zati

on

of

Met

hyle

ne

Blu

e D

ye

Batch

NB 550 NB 700

292

Figure D.3 Quality control chart for TOC/DOC

2

2.25

2.5

2.75

3

3.25

3.5

3.75

4

0 10 20 30 40 50 60 70

DO

C (

mg/L

)

Running Standard #

293

Figure D.4 QC chart for TCM Figure D.5 QC chart for BDCM

Figure D.6 QC chart for DCAA Figure D.7 QC chart for TCAA

0

5

10

15

20

25

30

35

40

0 5 10 15 20

TC

M (

ug

/L)

Standard #

0

5

10

15

20

25

30

35

40

0 5 10 15 20

BD

CM

(u

g/L

)

Standard #

0

5

10

15

20

25

30

35

40

0 5 10 15 20

TC

AA

(u

g/L

)

Standard #

0

5

10

15

20

25

30

35

40

0 5 10 15 20

DC

AA

(u

g/L

)

Standard #

294

Appendix E: Proposed TiO2-based Treatment Systems

Figure E.1 Single step photocatalytic system with membrane filtration for separation

295

Figure E.2 Two-step adsorption and regeneration system with membrane filtration for separation

296

Figure E.3 Two-step adsorption and regeneration system with sedimentation for separation

297

Appendix F: Cost Comparison of Proposed TiO2-based Treatment Systems to Existing Water Treatment Processes

Preliminary cost analyses were conducted for the two conceptual TiO2-based treatment processes

that were developed through this research. The potential energy and materials costs associated

with the single step photocatalytic treatment process discussed in Chapter 7 and shown in Figure

E.1 are described first while those associated with the two-step adsorption and regeneration

treatment process discussed in Chapter 8 and shown in Figure E.2 are explored later in this

appendix. The two proposed treatment concepts are compared to existing processes that are

currently used for water treatment in Canada and the implications of the costing analysis are

discussed briefly at the end.

Single Step Photocatalytic Treatment Process

Description

The single step photocatalytic treatment process, which was developed through the experiments

described in Chapter 7 of this document, is shown in Figure E.1. It resembles existing

commercial treatment options (e.g. Photo-CAT by Purifics, London, ON) but, unlike these

options, it makes use of highly photoactive and easily removable TiO2 LENs instead of

commercial nanoparticles. The system is conceptually simple and has the potential to be

relatively low energy and sustainable, which may make it particularly attractive for small and

remote drinking water systems, including those serving many Indigenous communities in

Canada. Health Canada defines small water systems as those serving between 501 and 5,000

users (Health Canada, 2013). Small communities often possess lower economic and operational

capacity than larger ones and the water treatment systems in these communities sometimes

employ minimal or non-traditional treatment processes (e.g. chlorination only), which can

increase the risk of DBP formation (CBCL Limited, 2011; Guilherme and Rodriguez, 2014)

and/or microbiological outbreaks (Murphy et al., 2016). Small systems are also often located in

rural and remote areas, which can increase the cost of shipping of equipment and chemicals to

the plant (Health Canada, 2013). The single step treatment system proposed in this project could

298

potentially provide concurrent disinfection/pathogen removal through photocatalytic degradation

and size exclusion, characteristics that are especially attractive for small and remote water

treatment systems. The results of this project (see Chapter 7) suggest that the single step

treatment option will only be appropriate for water sources with low alkalinity, however, more

work is required to determine the exact relationship between alkalinity and treatability within a

short time frame (< 60 min).

Costing

Capital Costs

At this stage of development, it is difficult to develop accurate capital cost estimates because

many important components of the proposed system have yet to be developed. The single step

treatment process is more likely to be applied at small scale and its capital costs will need to be

competitive with existing small system options for NOM and pathogen removal. Order of

magnitude costing for recent small water treatment projects in Canada are provided in Table F.1.

Population estimates were prepared by dividing the rated flow rate by 483 L/day/capita

(Statistics Canada, 2011) and rounding to the nearest 10.

Table F.1 Recent capital costs for small water treatment systems in Canada

Application Location Flow

Rate

Estimated

Population Cost Cost/m3

Reverse Osmosis Northern Ontario 0.4 MLD 750 $200,000 $667

GAC / Pressure Filtration Northern Ontario 0.4 MLD 750 $125,000 $417

Coagulation / Flocculation Northern Ontario 0.4 MLD 750 $500,000 $1,667

Dissolved Air Flotation Nova Scotia 0.9 MLD 2,000 $500,000 $584

Dissolved Air Flotation Nova Scotia 0.9 MLD 2,000 $925,000 $271

Media Filter / Nanofiltration Nova Scotia 5.8 MLD 12,000 $1,500,000 $974

Media Filter / Nanofiltration Nova Scotia 5.8 MLD 12,000 $1,250,000 $218

Sources: 1Azzeh, J., personal communication, May 2, 2017; 2Chaulk, M., personal communication, May

16, 2017

299

Materials Costs

Since the beginning of this project, research and industry grade TiO2 LENs have started to

become commercially available. These products are sold in amorphous or powdered form and

may require further processing (e.g. calcination) to tailor them to specific applications. Unit costs

for a selection of commercially available products are provided in Table F.2.

Table F.2 Cost of commercially available TiO2 LENs

Vendor Grade Unit Cost (CAD/kg) Description

Novarials Research $427,000 Length = 10 m

Width = 10 nm

Novarials Industrial $40,000 Length = 5 m

Width = 100 nm

Sigma Research $600,000 Width = 25 nm

Assuming that the industrial grade LENs from Novarials are capable of degrading DBP

precursors as effectively as NB 700 (i.e. 90% removal in 60 minutes) at a dose of 0.25 g/L, the

cost of materials would be $9.90 / L. This price is exhorbitant, however, if the single step

treatment option were to be developed further, the LEN synthesis process would need to be

scaled up to manufacture enough LENs to build commercially viable systems. Scale-up should

eventually reduce the unit cost of the material. It should be kept in mind that, unlike other water

treatment chemicals, the LENs are potentially reuseable.

Energy Costs

Assumptions in Table F.3 were used to develop the cost curve shown in Figure F.1.

Table F.3 Assumptions for energy cost analysis – single step treatment process

Parameter Value Units Source

LED Power Rating 2.4 W UVA LED specifications (LED ENGIN)

TiO2 Dose 0.25 g/L Chapter 7

Membrane Energy Costs 96.5 $/MLD Statistics Canada (2011)

Energy Cost 0.154 $/kWh Ontario Energy Board (2017)

300

The costs presented here may overestimate the amount of energy required to run the system

because they were developed based on the bench-scale batch experimental set-up used for the

experiments. As described in Appendix G, with this configuration the light penetration into the

samples was essentially limited to the top layer of water. The flow through reactor depicted in

Figure E.1 is serpentine and studded with LEDs. Depending on the width of the channel, the

amount of light penetration into the sample might be higher than that achieved in this project.

This could substantially reduce the amount of power required for the irradiation step because

each individual nanoparticle would be exposed to a greater amount off incident light and thus the

time required to achieve the dose of light required to drive the photocatalytic oxidation of DBP

precursors would be shorter.

Figure F.1 Estimated annual energy cost for the single step treatment process option as

a function of plant capacity

Other O&M Costs

Other O&M costs such as labour, maintenance, safety equipment, replacement parts, and waste

treatment and disposal are beyond the scope of this project. The proposed single tank system

may, however, have some advantages over existing treatment options in these categories. For

example, although it operates in a different way than coagulation/flocculation, the proposed

$1

$10

$100

$1,000

$10,000

$100,000

$1,000,000

$10,000,000

$100,000,000

100 1,000 10,000 100,000 1,000,000 10,000,000

An

nu

al

En

ergy C

ost

s ($

)

Plant Capacity (L/day)

Total Cost

Irradiation

Separation

301

single tank system fulfills the same DBP precursor role as this technology. Unlike

coagulation/flocculation, however, the proposed system is expected to create few, if any,

residuals that will require further processing and/or disposal. It may also prove to be safer than

some existing oxidation-based systems such as ozonation.

Total O&M Cost

The total O&M cost was calculated as the sum of the estimated material and energy costs.

𝐶𝑜𝑠𝑡𝑇𝑜𝑡𝑎𝑙 = 𝐶𝑜𝑠𝑡𝑀𝑎𝑡 + 𝐶𝑜𝑠𝑡𝐸𝑛𝑒𝑟𝑔𝑦 (𝐼𝑟𝑟𝑎𝑑𝑖𝑎𝑡𝑖𝑜𝑛) + 𝐶𝑜𝑠𝑡𝐸𝑛𝑒𝑟𝑔𝑦 (𝑆𝑒𝑝𝑎𝑟𝑡𝑖𝑜𝑛)

The total O&M cost of the single tank system would be approximately $10/L under the

assumptions used in this analysis. This is much higher than the cost of existing water treatment

systems for NOM removal, however, unlike these systems, the material costs for the single tank

option can be minimized by reusing the LENs many times. Figure F.2 shows how many times

the LENs would need to be reused for the single step option to be competitive with existing

water treatment options as a standalone system in terms of O&M costs. The O&M estimates for

existing treatment systems were developed using costing curves in Cost Estimating Manual for

Water Treatment Facilities (McGivney and Kawamura, 2008) and adjusted for inflation (2008-

2017). Note that the technology is likely to become more competitive at smaller scales and in

remote communities, where energy and shipping costs make up a larger proportion of the total

cost of operating a water treatment system. Additionally, if the system were to be used as a

polishing step as opposed to a standalone system, it may be possible to lower the dose of TiO2

and/or the irradiation time. Finally, as mentioned earlier (Materials Cost), were this option to be

pursued, the LEN synthesis process would need to be scaled up to industrial levels, and this

would likely substantially reduce the cost of the material.

302

Figure F.2 Number of reuses required for the single tank system to be competitive with

existing water treatment systems as a standalone option

Markets

In general, the proposed single step process is only appropriate for low alkalinity waters with

high NOM levels where the reduction of DBPfp is a priority. Many regions of Canada, including

parts of northern and eastern Ontario, the Maritime provinces, Newfoundland and Labrador,

coastal British Columbia, eastern Manitoba, and various parts of Quebec use surface water

sources that meet these criteria (Statistics Canada, 2011). One market where the single step

treatment process might be competitive is at the very small scale. These systems, which produce

between 100 and 1,000 L of water each day, provide water for consumptive uses for small

populations or full water services for residences or institutional facilities in areas where

municipal water is not available for consumptive or non-consumptive uses (e.g. nursing stations).

For example, for nearly a decade the Government of Newfoundland and Labrador has been

sponsoring the design and construction of 1,000 L/day potable water units in many rural and

remote communities in the province. The preliminary energy costs for the single step treatment

process option presented in Figure E.1 are comparable to the initial O&M cost estimates for

these units (Miller et al., 2009; Chaulk and Picco, 2010). The simplicity, minimal chemical

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

1.E-03 1.E-02 1.E-01 1.E+00 1.E+01 1.E+02 1.E+03

# R

euse

s

Flow Rate (MLD)

Conventional Treatment Conventional + Ozone + GAC DAF UF / NF RO

303

requirements, and overall safety of the proposed system as well as the potential provision of

concurrent NOM removal and disinfection would be particularly attractive in these applications.

Alternative Markets

Alternative markets for a small scale, closed loop system capable of mineralizing organic

contaminants might include oil and gas, food production, military, mining, and groundwater

remediation.

Two-step Adsorption and Regeneration Process

Description

The two-step adsorption and regeneration process developed in Chapter 8 of this thesis and

depicted in Figure E.2 in Appendix E represents a potential way to safely incorporate TiO2 into a

drinking water treatment process. As conceived in this project, in this process the adsorbent

LENs would be added to the raw water and mixed in a serpentine flow through reactor to allow

organic contaminants such as DBP precursors to adsorb on the TiO2 surface. The LENs would

then be removed from the treated water via membrane filtration, resuspended in clean water, and

sent through a second serpentine flow through reactor. The second flow through reactor would

be equipped with UVA LEDs that would provide sufficient irradiation to degrade the organics

adsorbed to the LENs. The LENs would then be recaptured in a second membrane separation

step and recycled to the beginning of the process.

Comparison to Existing Water Treatment Processes

Strictly in terms of NOM reduction levels, adsorption and photocatalytic degradation by TiO2

appear to have similar effects on NOM removal as enhanced coagulation. That is, in the absence

of particulate matter, TiO2 adsorption and photocatalytic degradation are able to remove a

sizeable proportion of the dissolved NOM present in the water. Also, enhanced coagulation

targets the aromatic, UV254-absorbing NOM that is commonly associated with trihalomethane

304

(THM) formation (Edzwald et al., 1985; Pifer and Fairey, 2014) but has only limited effects on

other NOM compounds, some of which can also react with disinfectants to yield organic

disinfection byproducts (Hua and Reckhow, 2007). This is also true of TiO2 adsorption and

photocatalysis, both of which target large and aromatic NOM compounds (Liu et al., 2008;

Huang et al., 2008).

That said, conceptually, the TiO2 adsorption and regeneration process (a.k.a. two-step process)

developed over the course of this project and depicted in Figure E.2 more strongly resembles

powdered activated carbon (PAC). Like in the proposed two-step treatment process, PAC is

applied to water as a powder to form a suspension. Organic contaminants become adsorbed to

the surface of the PAC and the loaded adsorbent is removed from the water via conventional or

membrane filtration (Pirbazari et al., 1992; Crittenden et al., 2012). PAC is most commonly used

for the removal of taste and odour compounds and other trace organics and in many cases is only

applied periodically as required. It is single use and must be disposed of after use.

Costing

Capital Costs

Capital costing, which includes design, construction, equipment costs, are beyond the scope of

this project, however, as with the single step treatment system, the cost of the two-step

adsorption and regeneration process will need to be cost competitive with existing treatment

NOM/pathogen removal technologies if it is to be adopted as a standalone process (see Table

F.1). Should the system instead be employed as a post-coagulation polishing step, it will need to

be cost competitive with existing polishing options such as activated carbon. For example, recent

estimates made as part of a water treatment plan upgrade project in the Southern Ontario treating

900 MLD pegged the cost of a new PAC delivery system (including materials) at $2 million, or

$2.20/m3 (Shen, personal communication, May 2, 2017).

Materials Costs

If the two-step treatment process is to be used in standalone mode for NOM removal it must be

able to be competitive with existing options. Adham et al. (1991) prepared formal isotherms

305

measuring the removal of NOM as a target contaminant from natural water using PAC (rather

than NOM as an interferent in the adsorption of other species). They used a variety of PACs to

remove NOM from groundwater with an initial TOC of 2.8 mg/L, which is lower than that used

in this project. No information about NOM character (e.g. UV254, SUVA, etc.) was provided.

Nonetheless, the Freundlich isotherm parameters from the Adham study were used to estimate

the dose of TiO2 required to remove 30% of the DOC from the OTB and OTW matrices and in

both cases, the required dose was approximately 0.05 g/L (50 mg/L).

Table F.4 Assumptions for material cost analysis – two-step treatment process


Adsorption Dose 1.5 g/L Chapter 8

Adsorption Time 30 min Assumed

Regen Dose 1.5 g/L Assumed

Regen Time 60 min Chapter 8

Regen Cycles Unlimited

Percent Removal of DOC 30 % Chapter 8 (NB 550 in OTW)

KF PAC 20.8 Adham et al., 1991

1/n PAC 0.8 Adham et al., 1991

PAC Cost 0.55 USD/lb Wiesner et al., 1994

PAC Cost Adjusted for Inflation 0.90 USD/lb Calculated

USD/CAD Conversion 1.37 CAD/USD May 12, 2017

Even with the generous and unrealistic assumption that the LENs are infinitely regenerable for at

least one year, the materials cost of the LENs was found to be five orders of magnitude greater

than that of PAC for the same degree of NOM removal. As shown in figures F.3 and F.4 below,

this high cost is a function of both the concentration of TiO2 required as well as the unit cost of

the materials, which was assumed to be $40,000/kg based on the price of industrial grade TiO2

LENs from Novarials (Table F.2). The latter parameter had a stronger effect on final materials

cost than the former. The assumed unit cost is very high, and it is unlikely that any LEN-based

treatment process will be feasible until it drops below $40/kg or the nanomaterials/the

regeneration procedure is optimized such that the LENs are indeed reuseable over a long period

of time.

306

Figure F.3 Effects of TiO2 LEN dose and plant capacity on estimated energy costs

Figure F.4 Effects of TiO2 LEN unit cost and plant capacity on estimated annual

materials cost

$100

$1,000

$10,000

$100,000

$1,000,000

$10,000,000

$100,000,000

100 1,000 10,000 100,000 1,000,000 10,000,000

An

nu

al

Ma

teri

als

Co

st (

$)

Capacity (L/day)

0.5 g/L 1 g/L 1.5 g/L

$1

$10

$100

$1,000

$10,000

$100,000

$1,000,000

$10,000,000

$100,000,000

100 1,000 10,000 100,000 1,000,000 10,000,000

An

nu

al

Mate

rials

Cost

($)

Capacity (L/day)

$40/kg $400/kg $4,000/kg $40,000/kg

307

Energy

The energy required to regenerate the LENs after use and separate them after treatment was

estimated using the following assumptions:

Table F.5 Assumptions for energy cost analysis – two-step process


LED Power Rating 2.7 W UVA LED specifications (LED ENGIN)

Regeneration Frequency 12 per day Assumed

Membrane Energy Costs 96.5 $/MLD Statistics Canada (2011)

Energy Cost 0.154 $/kWh Ontario Energy Board (2017)

With these assumptions, the energy cost associated with irradiation during regeneration and the

two membrane separation steps is approximately $0.02/L. Energy use associate with PAC is

expected to be minimal.

Other O&M Costs

Other O&M costs such as labour, maintenance, safety equipment, replacement parts, and waste

treatment and disposal are beyond the scope of this project. The proposed single tank system

may, however, have some advantages over existing treatment options in these categories. For

example, it would create less waste than comparable NOM removal technologies such as

coagulation/flocculation or PAC. At the small and very small scale, the fact that it may

potentially provide concurrent reduction of other contaminants of interest (e.g. pathogens,

metals) may minimize the additional process steps required to achieve clean and safe drinking

water along with their accompanying O&M costs.

Total O&M Cost

The total O&M cost was calculated as the sum of the estimated material and energy costs.

𝐶𝑜𝑠𝑡𝑇𝑜𝑡𝑎𝑙 = 𝐶𝑜𝑠𝑡𝑀𝑎𝑡 + 𝐶𝑜𝑠𝑡𝐸𝑛𝑒𝑟𝑔𝑦 (𝐼𝑟𝑟𝑎𝑑𝑖𝑎𝑡𝑖𝑜𝑛) + 𝐶𝑜𝑠𝑡𝐸𝑛𝑒𝑟𝑔𝑦 (𝑆𝑒𝑝𝑎𝑟𝑡𝑖𝑜𝑛)

The total O&M cost of the proposed two-step treatment process was determined to be

approximately $3.73/L, well above that of PAC, which is fractions of a cent per litre. The high

cost of the LENs also resulted in a high overall O&M cost for the proposed two-step treatment

308

process relative to that of existing treatment processes such as conventional treatment (with and

without ozone), DAF, and various membrane filtration options that are used for NOM removal at

either the large (conventional etc.), small (DAF, UF-NF), or very small scale (RO). The

estimated O&M costs for these existing systems are presented in Figure F.6 as a function of

system capacity (calculated according to equations in McGivney and Kawamura, 2008 and

adjusted for inflation between 2008 and 2017). The proposed two-step system is only

competitive with these processes at the very small scale. Should the cost of the materials come

down thanks to scale-up and/or automation of the synthesis process, the per litre cost of the

proposed two-step system will come down and the system may be competitive with existing

systems or as an addition to existing systems.

Figure F.5 O&M costs as a function of plant capacity for existing water treatment

processes used for NOM removal at large or small scale

Markets

Assuming that the materials and/or regeneration procedure can be optimized such that the

materials become more reuseable and that the overall cost of the materials will come down as a

result of scale-up, the most likely drinking water markets for the proposed two-step treatment

process are as an add on to municipal treatment systems to remove DBP precursors that are

$0.01

$0.10

$1.00

$10.00

$100.00

$1,000.00

$10,000.00

1.E+02 1.E+03 1.E+04 1.E+05 1.E+06 1.E+07 1.E+08 1.E+09

$/m

3

Capacity (L/day)

Conventional Conventional + Ozone DAF UF + NF RO

309

recalcitrant to coagulation or as a treatment process for the removal of taste and odour

compounds and cyanotoxins, which have been shown to be removable by TiO2 (Fotiou et al.

2015). Alternatively, the proposed system might work as a standalone option for small and

remote communities where shipping costs are prohibitive and a simple, self-contained system

that can potentially remove pathogens and other contaminants along with DBP precursors would

be an attractive option.

Summary

The analysis presented here suggests that the treatment concepts that have been developed

through this project are not currently cost competitive with existing treatment options with the

possible exception of very small systems in remote areas. Alternatively, one or both systems may

be appropriate as polishing processes following conventional treatment for the removal of

recalcitrant DBP precursors or other organic contaminants of concern (e.g. T&O compounds,

cyanotoxins).

The two-step treatment process is more novel than the single step treatment process and thus

more likely to be patentable and/or less likely to overlap with the intellectual property of existing

equipment manufacturers or other researchers. Should research continue into the two-step

treatment process, the results of this preliminary cost analysis suggest that research resources are

best allocated to the following:

1. Finding a niche

a. Removal of taste and odour compounds and cyanotoxins

b. Removal of DBP precursors not removed by coagulation or other conventional

processes

c. Removal of target contaminants in alternative markets (e.g. groundwater

remediation, oil and gas, etc.)

2. Improving performance

a. Matrix adjustments to improve adsorption (e.g. pH depression)

b. Optimization of regeneration procedure

310

c. Reconsider use of second generation LENs with higher surface area (e.g. NB

130/550) to improve adsorption

3. Scale-up

a. Scale-up and/or automation of nanomaterial synthesis procedure

b. Development of bench-scale flow through prototype

References

Adham, S.S., Snoeyink, V.L., Clark, M.M., Bersillon, J-L (1991) Predicting and verifying

organics removal by PAC in an ultrafiltration system, Journal of the American Water Works

Association, 83, 12, 81-91

CBCL Limited (2011) Study on characteristics and removal of natural organic matter in drinking

water systems in Newfoundland and Labrador, prepared for the Government of Newfoundland

and Labrador, Department of Environment and Conservation, Water Management Division,

Chaulk, M. and Picco, B. (2010) Drinking Water Safety Initiative, Clean and Safe Drinking

Water Workshop 2010, Gander, NL

Chowdhury, Z.K., Summers, R.S., Westerhoff, G.P., Leto, B.J., Nowack, K.O., and Corwin, C.J.

(2013) Activated Carbon: Solutions for Improving Water Quality, Passantino, L.B. (Ed.),

Denver, USA, American Water Works Association


Treatment: Principles and Design, 3rd ed., John Wiley and Sons, Hoboken, NJ, USA

Edzwald, J.K., Becker, W.C., and Wattier, K.L. (1985) Surrogate parameters for monitoring

organic matter and THM precursors, Journal AWWA, 77 (4), pp. 122-132




Journal, 261, 17-26

311

Guilherme, S. and Rodriguez, M.J. (2014) Occurrence of regulated and non-regulated

disinfection byproducts in small drinking water systems, Chemosphere, 117, 425-532

Health Canada (2013) Guidance for Providing Safe Drinking Water in Areas of Federal

Jurisdiction – Version 2, Pub. Number 130373

Hua, G. and Reckhow, D.A. (2007) Characterization of disinfection byproduct precursors based

on hydrophobicity and molecular size, Environmental Science and Technology, 41 (9), pp. 3309-

3315



pp. 1142-1150

Liu, S., Lim, M., Fabris, R., Chow, C., Chiang, K., Drikas, M., and Amal, R. (2008) Removal of



McGivney, W.T. and Kawamura, S. (2008) Cost Estimating Manual for Water Treatment

Facilities, John Wiley and Sons, Hoboken, NJ, USA

Miller, P., Keefe, F., Kendall, A., Stead, T., Caines, B. (2009) Operation and maintenance of

potable water dispensing units (PWDUs): Small town perspectives from NL, Clean and Safe

Drinking Water Workshop 2009, Gander, NL

Murphy, H.M, Thomas, M.K., Medeiros, D.T., McFayden, S., Pintar, K.D. (2016) Estimating the

number of cases of acute gastrointestinal illness (AGI) associated with Canadian municipal

drinking water systems, Epidemiology and Infection, 144 (7), 1355-1370

Ontario Energy Board (2017) Electricity Rates, Accessed May 12, 2017:

https://www.oeb.ca/rates-and-your-bill/electricity-rates


trihalomethane formation in drinking water sources, Environmental Engineering Science, 31 (3),

pp. 117-126

312

Pirbazari, M., Badriyha, B., Ravindran, V. (1992) MF-PAC for treating waters contaminated

with natural and synthetic organics, Journal of the American Water Works Association, 84, 12,

95-103

Statistics Canada (2011) Survey of Drinking Water Plants, Catalogue no. 16-403-X, Accessed

May 12, 2017: http://www.statcan.gc.ca/pub/16-403-x/16-403-x2013001-eng.pdf

Wiesner, M.R., Hackney, J., Sethi, S., Jacangelo, J.G., and Laine, J-M (1994) Cost estimates for

membrane filtration and conventional treatment, Journal of the American Water Works

Association, 33-41

313

Appendix G: Irradiance Considerations

A spreadsheet developed by Bolton and Linden (2003) was used to calculate the average

irradiance at the surface of the sample assuming that the reactor had a diameter of 6.5 cm, the

volume of the sample was 50 mL, and the distance from the light to the surface of the sample

was 13.7 cm. The spreadsheet, which was originally developed to calculate the time required to

achieve a defined germicidal UV dose, is also able to calculate the average UV dose delivered

throughout the volume of the sample, however, this function does not translate directly to

photocatalytic systems. For one thing, even relatively low doses of TiO2 obscure the passage of

light through the sample due to a combination of absorbance by the material (which can lead to

photoactivation) and light scattering. At the TiO2 doses used in this project (0.1 g/L, 0.25 g/L,

0.5 g/L, and 1 g/L) almost all of the light that enters the sample is absorbed or scattered by the

nanomaterials, resulting in a low average irradiance through the volume of the sample (Figure

G.1).

Figure G.2 shows the average irradiance through different volumes of sample as calculated using

the Bolton spreadsheet at different P25 TiO2 nanoparticle doses in MilliQ distilled water. The

depth of the sample refers to the distance from the top of the sample to the designated value.

Note that these values do not represent the irradiance in different “slices” of the sample because

the existing spreadsheet is not set up to do this. Rather, this graph shows the average irradiance

in the top 1 cm of the sample vs. the average irradiance in different volumes of sample in the

designated reactor. The spreadsheet, which was designed for disinfection applications rather than

photocatalytic reactor design, also fails to distinguish between absorption and light scattering.

Nonetheless, the semi-quantitative results presented in Figure G.2 do indicate that UVA light is

unlikely to have penetrated deeply into the sample, particularly at the higher TiO2 doses used in

this project, and that the majority of photoactivation and photocatalysis occurred within the top

layer(s) of the sample. The samples were constantly and completely mixed throughout the

experiments to minimize the likelihood of stratification through their depth, so in theory,

individual nanoparticles would have been cycled through this top layer regularly..

314

Figure G.1 Absorbance at 365 nm and average irradiance through the volume of the

sample for 50 mL samples of distilled water dosed with varying

concentrations of P25 TiO2 nanoparticles

Figure G.2 Average irradiance through different volumes of sample at different doses of

P25 TiO2 nanoparticles in MilliQ distilled water

0

1

2

3

4

5

0

1

2

3

4

0 0.2 0.4 0.6 0.8 1

Av

era

ge

Irra

dia

nce

Th

rou

gh

Sa

mp

le

(mW

/cm

2)

Ab

sorb

an

ce a

t 3

65

nm

(cm

-1)

TiO2 Dose (g/L)

Absorbance at 365 nm Average Irradiance

0

1

2

3

4

5

0 0.5 1 1.5

Av

era

ge

Irra

dia

nce

Th

rou

gh

Sam

ple

(mW

/cm

2)

Depth (cm)

0 g/L

0.005 g/L

0.05 g/L

0.1 g/L

0.25 g/L

0.5 g/L

1 g/L

315

The Bolton spreadsheet calculates a value called the Petri Factor (PF) to account for variation in

the irradiance reaching the surface of the sample. In ideal collimated beam systems, the PF

would equal 1, indicating that the same amount of light is hitting the sample at all points across

its surface. In reality, the irradiance is likely to be highest at the centre of the collimated beam

and to drop off at the outer edges of the beam. The irradiance of the UVA LEDs used in this

study was measured at various points within the collimated beam as shown in tables G.1 through

G.3. These values were averaged and inputted into the Bolton spreadsheet and used to calculate

the average irradiance at the surface of the sample and throughout its volume.

Table G.1 Irradiance of LED 1

LED 1

x y Irradiance (mW/cm2) x y Irradiance (mW/cm2)

0 -3.0 3.8 -3.0 0 4.1

0 -2.0 4.4 -2.0 0 4.7

0 -1.0 5.8 -1.0 0 5.1

0 0.0 6.5 0.0 0 6.5

0 1.0 5.0 1.0 0 5.1

0 2.0 4.4 2.0 0 4.2

0 3.0 3.7 3.0 0 3.5


LED 2


0 -3.0 4.5 -3.0 0 4.4

0 -2.0 5.2 -2.0 0 4.8

0 -1.0 6.1 -1.0 0 6.3

0 0.0 6.5 0.0 0 6.5

0 1.0 7.2 1.0 0 6.5

0 2.0 5.4 2.0 0 5.3

0 3.0 4.9 3.0 0 4.7

316


LED 3


0 -3.0 4.1 -3.0 0 3.4

0 -2.0 5.2 -2.0 0 4.4

0 -1.0 7.3 -1.0 0 6.1

0 0.0 6.6 0.0 0 6.6

0 1.0 5.9 1.0 0 6.2

0 2.0 5.3 2.0 0 4.7

0 3.0 5 3.0 0 4.5

References



215

317

Appendix H: Sedimentation Analysis

Please note that the discussion in this appendix was originally limited to the settling results from

Chapter 8 of this document. The equations and concepts provided here have since been applied

to the settling results presented in Chapter 6 and Chapter 7.

Stokes’ Law

Stokes’ Law is commonly used to model the settling of discrete particles through a liquid

medium. For a hard spherical particle, Stokes’ Law can be simplified to:


2

18𝜇 (H.1)




Commercial P25 nanoparticles are spherical in shape, but the LENs synthesized for this study

are, by definition, not. Liu et al. (2013) observed that the agglomerates formed by their LENs

were roughly spherical in shape. If it is assumed that the LENs in this project behaved similarly,

it is possible to compare the behavior of all three nanomaterials using Stokes’ Law. Assuming

that the hydraulic radius of P25 is 10 nm, that the hydraulic radius of the LENs is 1,000 nm, that

all of the nanomaterials were of the same density (ρTiO2 = 4.26 g/cm3), that all of the tests were

conducted at 20oC, and that all of the particles settled independently (i.e. Type I settling) the

Stokes’ equation yields the following values for settling velocity and time required to settle 70

mm (as per the experimental set-up for the high TiO2 dose settling experiments described in

Section 3.4.4.2 and Section 8.2.6).

318

Table H.1 Settling time required for P25 and two LENs in MilliQ water

Parameter P25 NB 550 NB 700

Hydrodynamic diameter (nm) 10 1000 1000

Settling velocity (m/s) 7.1E-10 7.10E-06 7.10E-06

Settling time (h) 27,400 2.74 2.74

The results indicate that under the stated assumptions, the average P25 nanoparticle would

require over 25,000 hours to settle out of solution and an average LEN particle would require

2.74 hours to settle. This does not match the behavior of the nanomaterials in this project,

indicating that at least one of our assumptions was incorrect.

Nanoparticle Agglomeration

TiO2 nanoparticle agglomeration in aqueous media has been studied in depth by many

researchers including Liu et al. (2013), Thio et al. (2011), Zhou et al. (2013), and based on their

studies and the results of the preliminary settling analysis, it seems inevitable that the

nanoparticles and LENs used in this study agglomerated to some extent. Figure H.1 shows the

effect of particle/agglomerate size on the time required to settle 70 mm calculated using the same

assumptions listed above.

Figure H.1 Effect of particle/agglomerate size on time required to settle

0.0001

0.001

0.01

0.1

1

10

100

1000

10000

100000

1 10 100 1000 10000 100000

Tim

e to

Set

tle

(h)

Agglomerate Diameter (nm)

319

The majority of the nanoparticles and LENs in this study settled the required distance within one

hour, so based on the results presented in Figure H.1 it’s likely that in most cases, the

particles/agglomerates were between 1,000 nm and 10,000 nm (1 to 10 m) in diameter. In some

cases (e.g. OTB water experiments) the majority of settling occurred within five minutes,

indicating that under these conditions the particles/agglomerates were either larger or had greater

effective densities. Particle size characterization was conducted using equipment in Professor

Siegel’s laboratory in the Department of Civil Engineering, however, these experiments were

conducted at concentrations well below those used in the adsorption and degradation

experiments (0.03 to 0.05 g/L TiO2) because the student in charge of the instrument had

calibrated it to measure his own, more dilute samples and was unwilling to change the settings to

better suit my samples. The results, shown in Figure H.2 below, are nonetheless illuminating.

0

1

2

3

4

5

6

7

8

0 20 40 60 80 100

Per

cen

tage o

f P

art

icle

s

Diameter (um)

MQ

OTB

OTW

0

1

2

3

4

5

6

7

8

0 20 40 60 80 100

Per

cen

tag

e o

f P

art

icle

s

Diameter (um)

MQ

OTB

OTW

B

A

320

Figure H.2 Particle size distributions for P25 nanoparticles (A), NB 550 (B), and NB 700

(C) in MilliQ water (natural pH) and two natural water matrices.

The results of the particle sizing tests cannot be used directly to predict the behavior of the P25

nanoparticles or the LENs at the concentrations used in the experiments in this project (0.1 g/L,

0.25 g/L, 1 g/L). Also, the DLS method itself is underlain by numerous assumptions and only

measures particles within the 0.1 m to 1,000 m range. Even so, the results in Figure H.2

suggest the following:

1. Agglomeration did indeed occur to some extent in all cases.

2. The particle size distributions of the LENs are wider than that of the P25 nanoparticles,

indicating that the assumption of uniformly sized particles/agglomerates is incorrect.

3. Water matrix had a more modest effect on the particle size distribution than nanomaterial

type did.

4. The agglomerates formed at approximately 0.05 g/L TiO2 had diameters between 1 and

100 m, but the majority were between 1 and 10 m as predicted in the analysis

presented earlier (Figure H.1).

5. The agglomerates formed by P25 were always smaller than those formed by the LENs.

6. NB 550 formed the largest agglomerates on average and the most variably sized

agglomerates overall.

7. The agglomerates formed in the OTB water were, in some cases, smaller than those

formed in the MilliQ water (natural pH) or the OTW water.

Taken together, these trends suggest that particle size alone is not an adequate predictor of

sedimentation efficacy. Other researchers (Deloid, et al., 2014) have noted that the effective

0

1

2

3

4

5

6

7

8

0 20 40 60 80 100

Per

cen

tag

e o

f P

art

icle

s

Diameter (um)

MQ

OTB

OTW

C

321

density of nanoparticulate agglomerates can differ substantially from the density of the material

because the agglomerate contains entrapped media (in our case water and naturally occurring

particulate matter) as well as nanoparticles. Liu et al. (2013) reported the density of their TiO2

LENs as 1.2 g/cm3, which is well below TiO2’s material density, which is 4.26 g/cm3. They did

not describe how they determined this value, however, given that it differs substantially from the

material density, it seems likely that this value represents the effective density of the

agglomerates rather than that of the individual LEN particles. The time required to settle 70 mm

as a function of both agglomerate size and density (1.1 g/cm3 to 4.26 g/cm3) is shown in Figure

H.3.

Figure H.3 Time required to settle as a function of particle/agglomerate size and

particle/agglomerate density

The effective agglomerate density can be predicted using the Sterling equations if the

hydrodynamic diameters of the nanoparticle and the agglomerate are known.

𝜀𝑎 = 1 − (𝑑𝑎𝑔𝑔

𝑑𝑚𝑎𝑡)𝐷𝐹−3

(H.2)

𝜌𝑒 = (1 − 𝜀𝑎)𝜌𝑝 − 𝜀𝑎𝜌𝑤 (H.3)

Where εa is the porosity of the agglomerate, which is a function of the diameter of the

agglomerate (dagg), that of the material (dmat) and the fractal dimension constant (DF), which is a

0.0001

0.001

0.01

0.1

1

10

100

1000

10000

100000

1000000

10000000

1 1.5 2 2.5 3 3.5 4 4.5

Hou

rs R

equ

ired

to S

ettl

e

Agglomerate Density (g/cm3)

10 nm

100 nm

1000 nm

10,000 nm

100,000 nm

322

indicator of the shape and the total volume of the agglomerate. Spherical agglomerates have a

DF of 3 while less uniform agglomerates have DF values ranging from 1 to 3 (Sterling et al.,

2005). ρe is the effective density of the agglomerate, ρp is the density of the individual

nanoparticle, and ρw is the density of the water.

These equations were applied to the particle distribution data (d10, d50, and d90 values under

each condition tested) to predict the amount of time required for 10%, 50%, and 90% of each of

the materials to settle out of the different water matrices assuming spherical particles and

agglomerates, LEN nanoparticle diameters of 2 m, and a DF of 2.3 as was used by Deloid et al.

(2014). The results are presented alongside the actual times required to achieve this degree of

removal in Table H.2.

Table H.2 Predicted and actual time required to remove 10%, 50%, and 90% of TiO2

from various water matrices

Test Predicted (min) Actual (min)

10% 50% 90% 10% 50% 90%

P25-MQ 908 2891 9733 -- -- --

P25-OTB 685 2107 7814 0.5 2 30

P25-OTW 769 3831 34710 5 15 90

550-MQ 5 48 590 0.5 1 60

550-OTB 5 48 442 0.5 6 60

550-OTW 4 50 645 4 -- --

700-MQ 12 69 411 0.5 3.5 60

700-OTB 14 76 464 2.5 8 60

700-OTW 15 95 607 5 -- --

The predicted settling time was sensitive to the DF value applied, and this was likely one of the

main reasons for the discrepancy between the predicted and actual results. The three materials

used in this project had differing configurations (spherical vs. linear) and two LENs were

heterogeneous in terms of both particle size and agglomerate size as shown in the TEMs

presented in Section 8.3.1 and the particle size distributions in Figure H.2. As a result, it is quite

likely that they had different agglomerate shapes and as such should have been assigned different

323

DF values, however, in the absence of agglomerate imaging it was difficult to determine the

shape of the agglomerates.

Sedimentation was not the main focus of this project and the results shown here are not

sufficiently detailed to fully model the settling of P25 nanoparticles or the two lab synthesized

LENs in the MilliQ, OTB, and OTW water matrices. Accurate hydrodynamic diameter

measurements at different TiO2 doses would greatly improve the accuracy of the results and aid

in the eventual design of a sedimentation reactor, but more detailed particle characterization may

be required to more fully describe the shape of the agglomerates formed by the three materials

under different water quality conditions to properly characterize the porosity of the agglomerates

and their resulting effective density. Deloid et al. (2014) proposed a (relatively) simpler method

for the determination of effective agglomerate density, but it requires access to an analytical

ultracentrifuge, a specialized piece of equipment commonly used for biochemical applications

that is not available in DWRG laboratory. Should this equipment become available, it may be

possible to determine the effective density of the materials without particle characterization.

Even with this information, however, it may prove challenging to accurately model the settling

behavior of the nanoparticles and LENs as they may not, in fact, follow the assumptions of Type

1 settling.

Settling Trends

Although settling trends were not explored in depth in this study, the settling trends over time in

the real and purified water matrices do suggest that settling behaviour was complex and the

dominant type of settling may have changed over time (Figure H.4 and H.5). In the OTW water,

the LENs settled in a slow but steady manner, suggesting Type I settling (discrete particle

settling). In contrast, the settling behaviour of all three materials in the OTB water matrix was

characterized by large initial drop in turbidity followed by slower, more gradual settling after

approximately 15 minutes. A similar trend was apparent for P25 in OTW water. This more

complex settling pattern suggests that the particles, or more correctly, agglomerates, were not of

uniform size in these suspensions.

324

Figure H.4 Percent removal of turbidity over time via settling in real water matrices

Effect of pH on Filtration and Settling of Nanomaterials in Purified (MilliQ)

Water

Control filterability and sedimentation experiments were run with 1 g/L suspensions of P25 and

the two third generation LENs in MilliQ water at its natural pH (5.5-6) and with the pH adjusted

to 8. The results of the tests run at the natural pH of MilliQ water matched those conducted under

similar conditions with the second generation LENs (Chapter 6) but those from the experiments

run at pH 8 were unexpected.

As shown in Figure H.5, in MilliQ water at natural pH (5.5 to 6), the filtration index of P25 (61.7

± 1.4) was over 60 times greater than that of MilliQ water alone (1) and approximately 25 and 30

times greater than that of suspensions of NB 550 (2.5 ± 0.0) and NB 700 (2.1 ± 0.0), respectively.

-100%

-90%

-80%

-70%

-60%

-50%

-40%

-30%

-20%

-10%

0%

0 10 20 30 40 50 60R

emo

va

l o

f T

urb

idit

y

Settling Time (min)

P25 - OTB

NB 550 - OTB

NB 700 - OTB

P25 - OTW

NB 550 - OTW

NB 700 - OTW

325

Figure H.5 Filtration indexes of raw water and three TiO2 nanomaterials suspended in

MilliQ water at pH 6 and pH 8 and two raw surface water samples

When the pH of the water was adjusted to pH 8 using HCl and/or NaOH, the filterability of the

P25 suspension was reduced to 18.0 ± 0.6, indicating greater filterability, but those of the two

LENs were unaffected. A short experiment was run to determine whether this was caused by

experimental error, specifically error related to pH changes over the course of the test. The

results are shown in Table H.3.

Table H.3 Effects of nanomaterial addition, time, and pH adjustment on the pH of

MilliQ water

Condition P25 NB 550 NB 700

Initial pH 5.94 5.94 5.94

After TiO2 Addition 5.1 6.13 6.1

Adjusted pH 7.9 8.06 8.38

Final pH 6.7 6.63 6.65

The results in Table H.3 confirm that the improved removability of the nanoparticles in MilliQ

water adjusted to pH 8 relative to MilliQ water at its natural pH was likely a function of

experimental conditions. For example, the addition of P25 nanoparticles to MilliQ water, which

lacks buffering capacity, depressed the pH of the water from approximately 6 to 5.1, which is

0

10

20

30

40

50

60

70


Fil

tra

tio

n I

nd

ex

MQ - pH 6

MQ - pH 8

326

well below the material’s IEP. This may have led to decreased agglomeration due to repulsive

charges between the individual nanoparticles and thus increased resistance to filtration. The pH 8

samples also experienced pH depression but in this case due interactions between the unbuffered

MilliQ water and the atmosphere. As a result, the pH “8” tests actually took place at pH 6.7,

which is very close to P25’s IEP (6.5). This is the point where the nanoparticles would be most

likely to agglomerate and thus least likely to clog the pores of the membrane filter. The

experiments were conducted in unbuffered MilliQ water because increased ionic strength has

been linked to increased agglomeration (Hotze et al., 2010) and sedimentation (Erhayem and

Sohn, 2014).

The pH of the water also had strong and sometimes unexpected effects on the sedimentation

efficiency of the nanomaterials. The P25 suspension made in MilliQ water at natural pH (5.5 to

6) remained stable throughout the hour-long test (0% turbidity removal) but the NB 550 and NB

700 suspensions made in MilliQ water settled out quickly (Figure H.6). The turbidity of the LEN

suspensions in MilliQ water decreased by approximately 70% after only five minutes and by

approximately 88% after 60 minutes. This is similar to results for the second generation LENs as

presented in Chapter 6 of this document. P25 settled much more effectively at pH “8” than at pH

“6”, because these matrices in fact had pHs of 6.7 and 5.1, respectively, and P25 agglomeration

is most likely to occur near its IEP (6.5). This does not explain the effects of pH on settling by

NB 550 and NB 700 in MilliQ water. The two materials have similar IEPs (see Figure 8.S.1) and

size and shape characteristics but the settling of NB 550 was strongly impacted by the pH of the

water while that of NB 700 was not. The actual pH of the water was similar in the two NB 550

tests (6.13 vs. 6.63), indicating that this was unlikely to be the cause of the hindered settling

during the pH “8” tests.

327

Figure H.6 Percent removal of turbidity from suspensions made with TiO2

nanomaterials in MilliQ water at pH 6 and pH 8

The individual LEN particles were much larger than the individual P25 nanoparticles (Section

8.3.1), but this size difference does not fully explain the improved removal of the former relative

to the latter. There is some evidence that LENs with higher surface areas are more resistant to

agglomeration than those with smaller surface areas (Zhou et al., 2013), but the reported effects

are small and it is more likely that experimental factors impacted the settling of NB 550.

References


Treatment: Principles and Design, 3rd ed., John Wiley and Sons, Hoboken, NJ

Deloid, G., Cohen, J.M., Darrah, T., Derk, R., Rojanasakul, L., Pyrgiotakis, G., Wohlleben, W.,

Demokritou, P. (2014) Estimating the effective density of engineered nanomaterials for in vitro

dosimetry, Nature Communications, 5, 3514 doi: 10.1038/ncomms4514

0%

20%

40%

60%

80%

100%

P25 NB 550 NB 700

Tu

rbid

ity

Rem

ov

al

(30

min

)

MQ - pH 6

MQ - pH 8

0%

328

Hotze, E.M., Phenrat, T., and Lowry, G.V. (2010) Nanoparticle aggregation: Challenges to

understanding transport and reactivity in the environment, Journal of Environmental Quality, 39,

1909-1924, doi:10.2134/jeq2009.0462





Sterling, M.C., Bonner, J.S., Ernest, A.N.S., Page, C.A., Autenrieth, R.L. (2005) Application of

fractal flocculation and vertical transport model to aquatic sol-sediment systems, Water

Research, 39, 1818-1830




properties on TiO2 nanoparticle agglomeration, PLOS One, 8, 11, e81239, doi:

10.1371/journal.pone.0081239

329

Appendix I: Statistical Analysis of Regeneration Results

The regeneration results were originally analyzed with a one-way ANOVA and Tukey’s and

Dunnett’s methods for comparison of means at the 95% confidence level, which is widely used

as a default confidence level in many fields of science. The statistical tests were then repeated at

the 90% confidence level to minimize the size of the confidence interval and lessen chance of

making a Type II error (i.e. accepting the null hypothesis of no difference between means when a

difference does exist). The results of all of these tests are summarized in Table I.1 and Table I.2.

Table I.1 Statistical analysis of regeneration data – AO24 experiments

Material Regen Mean Confidence

Interval

Tukey's

Grouping

Dunnett's

Grouping

95% 90% 95% 90% 95% 90%

NB 550 0 8.96 1.76 1.40 A A A A

1 8.82 1.76 1.40 A A A A

2 8.95 1.76 1.40 A A A A

3 8.05 1.76 1.40 A A A A

4 8.10 1.76 1.40 A A A A

5 8.07 1.76 1.40 A A A A

NB 700 0 7.67 0.72 0.57 ABC AB A A

1 8.02 0.71 0.57 A A A A

2 7.94 0.71 0.57 AB A A A

3 7.57 0.71 0.57 ABC ABC A A

4 6.90 0.71 0.57 BC BC B B

5 7.19 0.72 0.57 C C A A

Tukey’s Method compares the various means to one another to determine whether differences

exist between consecutive treatment levels whereas Dunnett’s Method compares the means at

different treatment levels to that of a control. At both confidence levels both methods suggested

that there was no significant change in NB 550’s ability to remove AO24 dye after multiple

regenerations. The NB 700 results were more variable at both confidence levels, but it should be

noted that in all cases, the CIs for NB 700 were tighter than those for NB 550 because the

individual measurements were closer to one another for the former than for the latter.

Nonetheless, at both confidence levels the confidence interval of the mean removal of AO24

after five regeneration cycles overlapped with that of the mean removal of AO24 by the virgin

330

material, indicating that the regeneration was effective and that the materials were resuseable at

least six times for this application.

Table I.2 Statistical analysis of regeneration data – NOM (UV254) experiments

Water

Type Material Regen Mean

Confidence

Interval

Tukey's

Grouping

Dunnett's

Grouping

95% 90% 95% 90% 95% 90%

OTB NB 550` 0 0.048 0.015 0.006 A A A A

1 0.036 0.022 0.009 AB AB A A

2 0.041 0.015 0.006 AB ABC A A

3 0.033 0.015 0.006 AB ABC A B

4 0.025 0.015 0.006 B BC B B

5 0.032 0.015 0.006 AB C B B

OTB NB 700 0 0.046 0.013 0.005 A A A A

1 0.042 0.013 0.005 A A A A

2 0.046 0.013 0.005 A A A A

3 0.038 0.013 0.005 A A A A

4 0.039 0.013 0.005 A A A A

5 0.040 0.013 0.005 A A A A

OTW NB 550 0 0.098 0.009 0.003 A A A A

1 0.081 0.009 0.003 B B B B

2 0.082 0.009 0.003 B B B B

3 0.065 0.009 0.003 C C B B

4 0.065 0.009 0.003 C C B B

5 0.046 0.009 0.003 D D B B

OTW NB 700 0 0.103 0.022 0.009 AB A A A

1 0.101 0.022 0.009 AB AB A A

2 0.109 0.022 0.009 A ABC A A

3 0.085 0.022 0.009 ABC BCD A A

4 0.073 0.022 0.009 C CD B B

5 0.079 0.022 0.009 BC D B B

The overall trends, namely that regeneration was more effective in the OTB experiments than in

the OTW experiments (hinting at the existence of some inhibitory component in the OTW water

matrix) and that NB 700 was more readily regenerated than NB 550, were the same irrespective

of the confidence level used for the analysis.

331

Appendix J: Evaluating and Modeling System Performance

Comparison of LEN Performance

Reaction Time and Rate

In chapters 6 and 7 of this thesis, the various lab synthesized LENs were compared based on

their ability to degrade methylene blue, DBP precursor surrogates (DOC, UV254), and DBP

precursors (THMfp, HAAfp). An example of this is provided in Figure J.1, which shows the

removal of total THMfp by NB 700, one of the third generation LENs, from Otonabee River

(OTB) water and Ottawa River (OTW) water.

Figure J.1 Reduction of THMfp in OTB and OTW water matrices via photocatalysis by

NB 700

In many, though not all, cases, the photocatalytic degradation of substrate followed an apparent

first order reaction rate model. As a result, the actual point of comparison between the materials

in most cases was the first order reaction rate constant (k), which was determined by plotting log

(C/Co) vs. log t and taking solving for k.

𝑟 = −𝑘𝐶 (J.1)

log(𝐶) = −𝑘𝑡 + log (𝐶𝑜) (J.2)

0

100

200

300

400

500

0 20 40 60

TH

Mfp

(

g/L

)


OTB

OTW

332

In general, it was not possible to calculate a first order reaction rate constant for THMfp and

HAAfp because these parameters increased at short irradiation times (< 15 min) and, usually,

decreased thereafter. In two cases, however, it was possible to calculate k: P25 in OTW water

and NB 700 in OTW water.

Rate constants are not the only, or necessarily the best, way to compare photocatalytic materials

and systems to one another or to other treatment processes. Many other parameters have been

developed to better characterize and compare the effectiveness and efficiency of these systems.

A selection of these is presented in the sections that follow.

Electrical Energy per Order

The electrical energy per order (EEO) concept is currently listed as a “figure of merit” for the

evaluation of advanced oxidation processes by IUPAC. Collins et al. (2016) define EEO as:

“…the electrical energy in kilowatt hours (kWh) required to bring about the degradation

of a contaminant C by one order of magnitude in 1 m3 of contaminated water or air.”

The EEO of a given process can be calculated using Equation J.3, where P is the power

dissipated by the treatment process (kW), V is the volume of water treated in the experiment (L),

Ci is the original concentration of the contaminant, Cf is the final concentration of the

contaminant, and t is the time required to achieve Cf (min).

𝐸𝐸𝑂 =1000 𝑃 𝑡


(J.3)

For batch experiments, the EEO should be calculated from the electrical energy dose (EED),

which is the electrical energy consumed per unit volume and can be calculated as follows:

𝐸𝐸𝐷 =1000𝑃𝑡

60𝑉 (J.4)

𝐸𝐸𝑂 =𝐸𝐸𝐷

log (𝐶𝑖𝐶𝑓)

(J.5)

333

In this study, the EEO values calculated using Equation J.2 were equal to that calculated using

equations J.4 and J.5.

The EEO concept is useful for comparing different types of light-driven systems and processes.

For example, Collins and Bolton (2016) compared EEOs for methylene blue degradation to show

that UV/H2O2 was far more efficient than UV/TiO2 for dye decolourization (EEOUVH2O2 = 0.63

kWh/order/m3 vs. EEOUVTiO2 = 16.4 kWh/order/m3). The EEO for UV/TiO2 reported by Collins

et al. is lower than those calculated for P25 and the various LENs in the current study (see Table

J.1). It should be noted, however, that the authors used a much lower starting concentration of

methylene blue (0.32 mg/L), did not report the experimental conditions (UV source, UV

irradiance, H2O2 or TiO2 dose, etc.), and that the papers they drew the data from are not

accessible through the University of Toronto library system.

Table J.1 EEO values provided by Collins and Bolton (2016) for methylene blue

degradation by UV/H2O2 and UV/TiO2 and EEO values for the degradation

of methylene blue by P25 and second and third generation LENs irradiated

by UVA LEDs

Process Dose Lamp Type Lamp Power Average Irradiance EEO

g/L W mW/cm2 kWh/order/m3

UV/H2O21 --2 UV3 --2 --2 0.63

UV/TiO21 --2 UV3 --2 --2 16.4

Light Only -- UVA LED 2.7 4.9 1,121

P25 0.1 UVA LED 2.7 4.9 42

NB 130/550 0.1 UVA LED 2.7 4.9 95

NB 130//700 0.1 UVA LED 2.7 4.9 81

NB 240/550 0.1 UVA LED 2.7 4.9 133

NB 240/700 0.1 UVA LED 2.7 4.9 69

P25 0.25 UVA LED 2.7 4.9 36

NB 550 0.25 UVA LED 2.7 4.9 71

NB 700 0.25 UVA LED 2.7 4.9 21

1From Collins and Bolton (2016)

2Not reported

3Power (W) not specified

334

A more detailed (and accessible) study by Yen and Yen (2015) explored the use of UV/H2O2 for

DOC and THMfp removal from a synthetic water matrix made with commercial humic acids.

Their experiments were conducted using a 9 W low pressure UV lamp (maximum irradiance at

254 nm) and three doses of H2O2. EEO values for the removal of DOC and THMfp by P25 and

the third generation LENs are compared to those reported by Yen and Yen (2015) for UV/H2O2

treatment in Figure J.2. Note that in the current study it was only possible to calculate EEO

values for THMfp removal under two conditions, OTW water with P25 and OTW water with NB

700, because these were the only conditions under which first order degradation kinetics were

observed.

Figure J.2 Comparison of EEOs for DOC and THM precursor degradation by

UV/H2O2 and UV/TiO2 with P25 and third generation LENs

From the graph, it is apparent that under these experimental conditions, UV/H2O2 was more

efficient for DOC and THMfp removal in almost all cases. The EEO values for DOC and

THMfp removal by NB 700 in OTW water were, however, comparable to those UV/H2O2

treatment with 10 mg/L of H2O2, indicating that under some conditions, UV/TiO2 may prove to

be competitive with UV/H2O2. This should be confirmed by comparing the two processes in the

same water matrices.

Some of the limitations of the EEO concept include the fact that it’s based on the total energy

required to run the treatment system rather than the dose of light added to the system, making it

0

100

200

300

400

500

600

DOC THMfp

EE

O (

kW

h/o

rder

/m3)

UV/H2O2 - 10 mg/L

UV/H2O2 - 25 mg/L

UV/H2O2 - 50 mg/L

UV/TiO2 - P25 - OTB

UV/TiO2 - P25 - OTW

UV/TiO2 - NB 550 - OTB

UV/TiO2 - NB 550 - OTW

UV/TiO2 - NB 700 - OTB

UV/TiO2 - NB 700 - OTW

335

system/experimental setup specific, and that it assumes first order degradation kinetics. It also

ignores the effect of adsorption, which, as has been shown in chapters 5 and 8 of this study, can

account for a substantial proportion of overall removal of DBP precursors and precursor

surrogates in TiO2-based treatment systems.

Power per Volume

An alternative way to compare the efficiency of different treatment systems is to calculate the

power required to remove a given amount of a contaminant.

𝑃𝑜𝑤𝑒𝑟

𝑉𝑜𝑙𝑢𝑚𝑒(𝑘𝑊ℎ 𝑚3⁄ ) =

𝑆𝑦𝑠𝑡𝑒𝑚 𝑃𝑜𝑤𝑒𝑟 𝑅𝑎𝑡𝑖𝑛𝑔 (𝑘𝑊) × 𝑇𝑖𝑚𝑒 (ℎ)

𝑉𝑜𝑙𝑢𝑚𝑒 𝑇𝑟𝑒𝑎𝑡𝑒𝑑 (𝑚3) (J.6)

Like the EEO value, power per volume is based on total system energy demand rather than

incident light and is therefore system specific. In this study, the system power rating was simply

the power required to run one UVA LED and the volume treated was 50 mL. Power per volume

can be used to calculate the amount of power required to meet a set goal or it can be used as an

alternative to time on the x-axis as shown in Figure J.3.

Figure J.3 Reduction of THMfp in OTB and OTW water matrices via photocatalysis

with NB 700

0

100

200

300

400

500

0 20 40 60

TH

Mfp

(

g/L

)

Power (kWh/m3)

OTB

OTW

336

This method of comparison doesn’t require first order kinetics – though these were assumed for

the analysis presented here in order to determine the amount of power required to remove 90% of

each of the parameters of interest as well as that required to bring the THMfp and HAAfp of the

water to the guideline values recommended in the Guidelines for Canadian Drinking Water

Quality (Health Canada, 2017).

Gerrity et al. (2009) evaluated the use of the Photocat UV/TiO2-based treatment system by

Purifics (London, ON) for DOC and THMfp removal from two Arizona water sources, the Salt

River and the Central Arizona Project Canal (CAP) and presented their results as a function of

power required per cubic meter of water treated (kWh/m3). The two water sources had pH values

of approximately 8 and alkalinity between 100 as 150 mg/L as CaCO3. The Salt River contained

more DOC than the CAP (6.7 to 7.4 mg/L vs. 4.8 to 5.7 mg/L) and the SUVA values of the two

water matrices (1.5 to 1.7 L/mg.m vs. 0.8 to 1 L/mg.m) indicate that the NOM in the Salt River

was more aromatic than that in the CAP. The THMfp of the two water sources was evaluated at

10oC and 28oC, however, only the latter conditions were used for the second stage of the

experiments, which is the stage of interest for the comparison shown below. The THMfp of the

raw Salt River water ranged from 146 to 165 g/L while that of the CAP ranged from 83 to 95

g/L. Therefore, both of the tested water matrices were more similar to the OTB water matrix

than the OTW water matrix, though they both contained less aromatic NOM and had lower

THMfps than both water matrices used in the current study. The power required to achieve 90%

removal of DOC and THMfp as well as that required to achieve the GCDWQ guideline value of

100 g/L of THMs using the Photocat unit and in the experiments conducted for this thesis using

the third generation LENs (see Chapter 7) are compared in Figure J.4 Note that THMfp removal

power requirements were only determined for P25 and NB 700 in OTW water because these are

the only experiments where first order degradation kinetics were observed.

337

Figure J.4 Power required to remove 90% of DOC and THMfp from different water

matrices using UV/TiO2-based treatment processes

The DOC results indicate that, with the exception of the experiments conducted with NB 550,

the bench-scale batch UV/TiO2 treatment setup used in the current study was more energy

efficient than the Photocat system, though this conclusion comes with numerous caveats. For

one, the two studies employed different water matrices. The findings presented in Chapter 7 of

this thesis and earlier results presented by other researchers (Liu et al., 2008) clearly show that

the characteristics of the water matrix, specifically alkalinity, calcium, and NOM type and

concentration, have a strong effect on the photocatalytic degradation of NOM by UV/TiO2. Also,

the number of items included in the overall power demand value for each system was different –

the Photocat system is a flow through pilot-scale installation that includes irradiation (75 W

lamps), pumping, an air compressor, and a control system whereas the power demand value used

to calculate the power requirements for the bench-scale batch experiments included only the

power required to run the UVA-LED, which was 2.7 W/LED. On the other hand, there are

reasons why the bench-scale batch experiments run in the DWRG lab were in fact more energy

efficient. Most notably, NB 700 has been shown to be more photoactive than P25, thus, it

requires less irradiation to accomplish the same amount of degradation. Also, the bench-scale

apparatus uses energy efficient LEDs rather than standard UVA lamps.

The THMfp removal trends were similar in that less power was required to remove THM

precursors from the OTW water using P25 and NB 700 in the batch bench-scale set-up used in

0

100

200

300

400

500

Po

wer

(k

Wh

/m3)

DOC (90%)

THMfp (90%)

THMfp (GCDWQ)

338

this study than was required to remove THM precursors from the Salt River and CAP water

matrices using the flow-through Photocat system. As with DOC, the apparent superior efficiency

of the bench-scale bench system was likely related to numerous experimental and practical

factors, however, the particularly low power requirements associated with NB 700 suggests that

this nanomaterial may prove to be a more energy efficient option than P25 in a single stage

photocatalytic UV/TiO2 treatment system.

Interestingly, Gerrity et al. (2009) observed similar changes in THM species present in water as a

function of time as were observed in this study (see Section 7.3.5 in Chapter 7). That is, there

was an initial increase in both TCMfp and BDCMfp at short treatment times that was followed

by a gradual decrease in the TCMfp accompanied by an increase in BDCMfp and finally an

overall decrease in the formation of all THM species at extended treatment times. The authors of

the study did not provide any hypotheses as to why this may have occurred, but based on the

findings of other researchers cited in the current project (Liu et al., 2008; Toor and Mohseni,

2007; etc.) it seems likely that the photocatalytic treatment degraded larger compounds into

smaller ones that were more reactive with chlorine and that further treatment degraded these

compounds into even smaller ones that reacted more readily with the larger bromine atom.

Notes:

1. The power required to reduce DOC and THMfp was integrated into the cost analysis for

the proposed single step treatment system presented in Appendix E.

2. The power required to remove 90% of a contaminant from the water should be equal to

its EEO value. The discrepancies between the EEO values and the power required to

remove 90% of any given parameter in this analysis are related to the fact that the Cf for

the EEO value was the actual measured concentration of the parameter observed in

samples treated for 60 minutes whereas the C90 used to calculate the power usage rate

was predicted based on the first order reaction rate constant calculated for each parameter

under different experimental conditions.

339

Cost per Volume

In real world applications and projects, the energy efficiency of different water treatment systems

is often expressed in terms of cost per volume:

𝐶𝑜𝑠𝑡 𝑝𝑒𝑟 𝑉𝑜𝑙𝑢𝑚𝑒 ($/𝑚3) = 𝑆𝑦𝑠𝑡𝑒𝑚 𝑃𝑜𝑤𝑒𝑟 𝑅𝑒𝑞𝑢𝑖𝑟𝑒𝑚𝑒𝑛𝑡𝑠 (𝑘𝑊ℎ 𝑚3) × 𝐶𝑜𝑠𝑡 𝑜𝑓 𝑃𝑜𝑤𝑒𝑟 ($ 𝑘𝑊ℎ)⁄⁄ (J.7)

This approach is usually based on the energy requirements of the full treatment system or

process rather than simply the amount of incident light. This is helpful in some ways because it

allows system designers to compare very different treatment options to one another in a holistic

way and also to create site specific cost estimates that can be used by system owners and

operators to choose between different full-scale treatment options. These very advantages,

however, can be drawbacks when the goal is to compare research results from different countries

or jurisdictions that have different currency or energy costs or when comparing bench, pilot, and

full-scale systems, which have differing levels of complexity. For example, in this analysis the

cost of power was assumed to be $0.157/kWh, which was the on-peak cost of energy in Ontario

in May 2017 when this thesis was being written, however, the price has since dropped to

$0.132/kWh (July 17, 2017). Power costs and payment schemes in other Canadian jurisdictions

vary in terms of magnitude and complexity, further complicating the analysis.

Table J.2 shows the approximate cost to reduce the THMfp and HAAfp of the OTW water

matrix by 90% or to GCDWQ guideline levels using P25 and NB 700. The estimates are in

Canadian dollars.

Table J.2 Cost to reduce the THMfp and HAAfp of OTW water via photocatalysis with

P25 and NB 700

Nanomaterial THMfp HAAfp

90% Guideline (100 g/L) 90% Guideline (80 g/L)

P25 $15.70 / m3 $6.27 / m3 $12.85 / m3 $1.57 / m3

NB 700 $7.07 / m3 $3.67 / m3 $8.83 / m3 $ 0.93 / m3

The data in Table J.2 clearly shows that a system employing NB 700 would cost less in terms of

energy than one employing P25. Although not quantified here, NB 700 is also easier to remove

from the water than P25 via filtration (see Section 8.3.5.1 in Chapter 8), which may mean that

the cost of removing the materials via membrane filtration may also be lower. These results can

340

also be used to compare the cost of THMfp reduction using different treatment processes. For

example, Yen and Yen (2015) estimated that removing 90% of the THMfp from their synthetic

water matrix using 10 mg/L H2O2 and a 9 W lamp would cost $1.65/m3 (USD), which translates

to $2.22 CAD/m3 (USD/CAD conversion from June 9, 2017). This is less than the cost of using

P25 or NB 700 to remove THM precursors from OTW water but does not account for the

reuseability of the TiO2 nanomaterials or the fact that the OTW water matrix is more complex

than the synthetic water matrix.

Appendix H includes a detailed cost analysis of the proposed single and two-step treatment

processes that have been proposed based on the results of this project. The proposed systems are

compared to existing water treatment options for NOM removal in terms of energy and materials

costs.

UV Dose / Fluence

Another way to track the progress of photocatalytic treatment processes is based on the UV dose,

or fluence, applied to the sample. Ideally, the UV dose would be calculated based on the incident

light throughout the sample, however, as described in Appendix G of this thesis, in the current

study it is unlikely that the UVA LED light applied to the samples penetrated deeply into them.

As a result, it was assumed that the UV dose could be calculated based on the average irradiance

at the surface of the sample. This was calculated to be 4.9 mW/cm2 using a spreadsheet prepared

by Bolton and Linden (2003) as described in Appendix G. This value was multiplied by the

elapsed time (s) to determine the UV dose or fluence (mJ/cm2) at the surface of the sample as

shown in Equation J.8.

UV Dose (mJ/cm2) = Irradiance (mW 𝑐𝑚2⁄ ) × Time (min) × 60 (s 𝑚𝑖𝑛⁄ ) (J.8)

Figure J.5 shows the effect of photocatalysis with NB 700 on the THMfp of OTB and OTW

water as a function of UV dose.

341

Figure J.5 Reduction of THMfp in OTB and OTW water via photocatalysis with NB

700 as a function of UV dose (fluence)

UV dose is less specific to experimental set-up than time or power/volume and as a result can

more easily be compared to the results of other researchers. It can also be a useful parameter

when comparing different light-based water treatment processes. For example, Autin et al.

(2013) demonstrated that the UV dose (254 nm) required to achieve metaldehyde degradation

was equal for UV/TiO2 and UV/H2O2 in the absence of alkalinity and organics. The addition of

CaCO3 and NOM surrogates increased the UV dose required to achieve metaldehyde removal

via UV/TiO2 but not that required for UV/H2O2 treatment. This demonstrated that UV/TiO2 was

more likely to be negatively impacted by the presence of ROS scavengers than UV/H2O2.

Another UV/H2O2 study showed that approximately 3,000 mJ/cm2 of UV light was required to

reduce the THMfp of a Canadian surface water matrix from 238 g/L to 54 g/L (77%) at an

H2O2 dose of 23 mg/L (Toor and Mohseni, 2007). This is well below the UV dose that was

required to achieve a comparable reduction in THMfp from OTW water using NB 700 (~13,000

mJ/cm2) in the current study, indicating that even in a best-case scenario, the UV dose required

to reduce the DBPfp of surface water via UV/TiO2 is unlikely to be comparable to that required

to reduce it via UV/H2O2. It should, however, be noted that Toor and Mohseni did not observe

any significant removal of THM precursors at a fluence of 3,000 mJ/cm2 at a lower H2O2 dose (4

0

100

200

300

400

500

0 5,000 10,000 15,000 20,000

TH

Mfp

(

g/L

)

UV Dose (mJ/cm2)

OTB

OTW

342

mg/L). Also, their experiments made use of a low pressure UV lamp (max irradiance at 254 nm)

but TiO2 can be activated by lower energy wavelengths of up to approximately 380-385 nm.

UV dose only accounts for irradiation-based portion of treatment and therefore may not

accurately reflect the overall cost of different processes. More importantly, however, UV dose

can be a less desirable parameter when comparing photolytic systems that require or employ

different wavelengths, and thus lamps with different energy ratings. In this study, the use of UV

dose as a parameter hides the main advantage of using UVA LEDs -- the fact that they are far

more energy efficient than standard UV germicidal lamps or high intensity UVA lamps. For

example, the study by Autin et al. (2013) took place in a bench-scale UVC collimated beam

apparatus containing four 30 W lamps. This was used to treat a 250 mL sample and the

irradiance at the surface of the sample was 2.23 mW/cm2, thus a fluence of 3,000 mJ/cm2

corresponded to 22.3 minutes of irradiation and a power per volume of 480 kWh/m3. A dose of

3,000 mJ/cm2 in the UVA-LED reactor used in the current study corresponds to an irradiation

time of 10.2 minutes and 54 kWh/m3. The UVA-LEDs used in the reactor cannot be used for

UV/H2O2 process because they only emit light at 365 nm, which is not energetic enough to drive

the formation of OH radicals from H2O2.

Quantum Yield

Quantum yield, also referred to in some publications as quantum efficiency, is the ratio of

chemical products formed (i.e. reaction events) to the number of photons absorbed (Ollis et al.,

2013) and in photocatalytic systems it gives a measure of the efficiency of electron positive hole

utilization (Ohtani, 2010). The quantum yield () is, theoretically, the rate of disappearance of

the substrate (dC/dt) divided by the number of photons absorbed (Ia) as shown in Equation J.9.

Φ =𝑑𝐶

𝑑𝑡⁄

𝐼𝑎 (J.9)

This apparently simple parameter has numerous underlying assumptions, most notably that each

absorbed photon results in a measurable reaction event and that the reaction occurs in a single

electron exchange step. As described by Ohtani (2010), these assumptions are not appropriate for

photocatalytic systems. The adsorption of a photon by a TiO2 particle liberates an electron and a

343

positive electron hole. In most cases, the two recombine without reacting with water, oxygen, or

any other chemical species. When recombination doesn’t occur, the two can react in numerous

ways, some of which may result in the formation of oxidative species capable of participating in

a measurable reaction event. Also, many desirable oxidation reactions (e.g. the degradation of

DBP precursors) are multi-electron, multi-step reactions.

The determination of Ia can also present challenges because TiO2 nanoparticles both absorb and

scatter light. The number of photons absorbed, rather than scattered, can be determined by

measuring the diffuse reflectance, which describes the proportion of the total incident photons

that are scattered vs. those absorbed, using a UV Vis spectrophotometer equipped with an

integrating sphere. Even when the ratio of scattered to absorbed photons is known, however, it is

only possible to determine apparent quantum efficiencies (Kisch and Bahnemann, 2015) and the

apparent quantum efficiency (also known as the photonic efficiency) are more commonly used to

characterize light utilization in photocatalytic systems.

Finally, many target contaminants, including DBP precursors and indicator dyes such as

methylene blue, also absorb light. As shown in Figure J.6, the methylene blue and OTB and

OTW water matrices had transmittances of nearly 100% at 365 nm, the wavelength emitted by

the UVA LEDs used in the current study, so absorption and scattering of incident light by the

target contaminants in this study was likely negligible, however, in general the fact that target

contaminants can themselves absorb and scatter light further complicates the calculation and

validity of quantum efficiency in many systems.

344

Figure J.6 Transmittance of light through 10 mg/L methylene blue solution and the two

raw water matrices used in this project

Photonic Efficiency

The complexity of quantifying light absorption by photocatalysts has led to many researchers

choosing to characterize their systems based on photonic efficiency () rather than quantum

efficiency. Photonic efficiency is essentially the same as quantum efficiency, however, it is

calculated based on the amount of light hitting the photocatalyst rather than the amount of light

absorbed.

𝜉 =𝑑𝐶

𝑑𝑡⁄

𝐼𝑖 (J.10)

Where dC/dt is the rate of degradation of the target substrate and Ii is the incident light reaching

each nanoparticle. Ii can be difficult to determine because TiO2 nanomaterials can both absorb

and scatter light and different nanomaterials may behave differently in this regard. The photonic

efficiency equation implicitly assumes that all samples absorb the same proportion of the total

incident photons (Kisch and Bahnemann, 2015), an unlikely situation in experiments using

multiple types and doses of TiO2 nanomaterials (see Figure J.7).

0

20

40

60

80

100

200 250 300 350 400 450 500

Tra

nsm

itta

nce

(%

)

Wavelength (nm)

Methylene Blue (10 mg/L) OTB OTW

345

Figure J.7 Absorbance of UV and visible light by 0.05 g/L and 0.1 g/L of P25

nanoparticles and 0.1 g/L of NB 550 and NB 700 suspended in MilliQ water

As described in Appendix G of this thesis, at the TiO2 doses used in this study light penetration

into the sample was likely minimal. It should therefore be possible to assume that all, or at least

the majority, of the photocatalytic activity in the experiment is taking place at the surface of the

sample and may approximate a thin film.

Mills (2012) described the use of photonic efficiency calculated from methylene blue

degradation to compare the performance of photocatalytic films using Equation J.11.

𝜉𝑀𝐵 =𝑑𝐶𝑀𝐵

𝑑𝑡⁄

𝐼𝑖 (J.11)

Where dCMB/dt has units of molecules/cm2/s and Ii has units of photons/cm2/s. Equation J.11 was

used to calculate the photonic efficiencies of P25 (6.0%), NB 550 (1.5%), and NB 700 (6.3%).

This trend is in line with other measures of photoactivity used in the current study, which have

consistently indicated that NB 700 is more photoactive than P25 and NB 550. The implication of

this is that a system employing NB 700 would likely be more energy efficient than one

employing P25 or NB 550. These photonic efficiency values predicted in the current study are

higher than by Mills (< 0.1%) but this might be a function of the experimental setup in the

current study, which employs a stronger lamp and a higher initial concentration of methylene

blue. Also, unlike the systems described by Mills, where the TiO2 nanoparticles are affixed to a

0

0.5

1

1.5

2

2.5

3

200 250 300 350 400 450 500

Ab

sorb

an

ce (

1/c

m)

Wavelength (nm)

P25 (0.05 g/L) P25 (0.1 g/L) NB 550 NB 700

346

solid support, the batch reactors used in the current study contained completely mixed

suspensions of TiO2. As a result, the actual amount of surface area of TiO2 available for reaction

was likely greater than would be available on an immobilized film. Finally, the constant mixing

of the samples in the current study likely resulted in a regular cycling of the nanomaterials from

the surface of the sample to deeper within its volume, and consequently, a constant

replenishment of nanomaterials at the surface of the sample.

Indeed, a recent study employing batch reactors and a different dye (Maxilon Blue) reported

photonic efficiencies for anatase TiO2 nanoparticles were within the same order of magnitude as

those obtained in the current study (Alrobayi et al., 2017). They employed the following

equations to determine a “relative” photonic efficiency based on the irradiance at the surface of

their completely mixed batch samples:

𝜉 =𝑅 × 𝑉

𝐼𝑜 × 𝐴 (J.12)

Where R is the rate of the first order degradation reaction (mg/L/min), V is the volume of the

batch sample (L), A is the irradiated area (cm2), and Io can be calculated from Equation J.13:

𝐼𝑜 =𝐼 × 𝜆

𝑁𝑎× ℎ × 𝑐 (J.13)

In Equation J.13, I is the irradiance at the surface of the sample (mW/cm2), is the wavelength

of the incident light (m), Na is Avogadro’s number, h is the Planck constant, and c is the speed of

light in space (m/s). The relative photonic efficiencies for P25 (5.6%), NB 550 (1.5%), and NB

700 (6.0%) calculated using equations J.12 and J.13 were within 0.4% of the photonic

efficiencies calculated using Equation J.11, indicating that both mathematical approaches yield

similar results and trends.

Although the relative photonic efficiencies reported for anatase nanoparticles by Alrobayi et al.

(2017) were generally between 2 and 6%, most of their experiments employed higher doses of

TiO2 and irradiances than the current study. The relative photonic efficiency reported by

Alrobayi et al. at a TiO2 dose of 0.25 g/L, an irradiance of 5 mW/cm2, and pH of 6.55 (~ 1%)

was lower than those reported for the predominantly anatase NB 700 LEN in the current study.

This is likely due to differences in the experimental set-up, which include different batch

volumes, mixing rates, light type and intensity, and indicator dye, among others.

347

Reaction Pathways

The methods of comparison described above are relatively simple and widely used in the

scientific literature, however, they do not distinguish between the various oxidative processes

occurring in photocatalytic systems and are usually specific to the experimental apparatus and

conditions used by the researchers. The rate at which a given contaminant is oxidized in a

photocatalytic system is a function of:

Direct photolysis

Photocatalytic degradation by electron holes on the surface of the particle

Photocatalytic degradation by OH radicals adsorbed to the surface of the particle

Photocatalytic degradation by other ROS adsorbed to the surface of the particle

The amount of the contaminant adsorbed to the particle (coverage) will also affect the rate of

degradation.

In theory, it is possible to isolate and quantify each of these pathways to better understand the

behavior of the photocatalytic system as described in Section 2.2.2 of Chapter 2 of this thesis. In

the future, it would be interesting to apply some of these methods to determine whether different

photocatalytic materials are more likely to produce different degradation pathways. For example,

in this study, NB 700 had the lowest surface area but the highest overall reactivity, but, beyond

the fact that it contains a higher proportion of anatase than the other nanomaterials used in this

study, the specific reasons for its excellent performance remained unclear and/or speculative

until preliminary ·OH radical formation tests were conducted in June and July of 2017. The

results of this testing showed that the pure anatase LENs (NB 130/700 and NB 240/700)

produced far more ·OH radicals than the mixed phase LENs (NB 130/550 and NB 240/550) did.

As discussed in Chapter 6 of this document, there was a clear linear relationship between the

rates of ·OH radical production and NOM degradation rates, especially when the latter were

normalized to the available surface area. This suggests that these normalized degradation rates

were a good predictor of the amount of NOM degradation occurring as a result of ·OH radical-

mediated reactions. Any additional degradation must have taken place as a result of reactions

mediated by other ROS (e.g. superoxide radical) or photogenerated holes on the surface of the

photocatalyst. Additional testing is required to confirm which of these processes dominated for

each of the nanomaterials used in this study.

348

Modeling the Photocatalytic Degradation of Organic Contaminants

The focus of this project was the development, characterization, and proof-of-concept

application of novel linear engineered nanomaterials for the removal of disinfection byproduct

precursors from real surface water matrices. The results were analyzed to determine adsorption

time and isotherms as well as apparent reaction kinetics and efforts were made to explain the

findings based on the characteristics of the nanomaterials and water matrices employed. In depth

modeling of the adsorption and degradation of DBP precursors by the nanomaterials was outside

the scope of the current project, but would be a worthwhile direction to pursue in the future.

Research groups from numerous fields have attempted to develop models that can account for

the many different processes taking place in aqueous photocatalytic systems. Malato et al. (2009)

proposed a simple method based on three main processes:

Photoactivation: 𝑇𝑖𝑂2 + ℎ𝑣𝑘𝑓→ 𝑒− + ℎ+ (J.14)

Recombination: 𝑒− + ℎ+𝑘𝑟→ 𝑒𝑛𝑒𝑟𝑔𝑦 (J.15)

Oxidation of reactant: ℎ+(𝑜𝑟 ∙ 𝑂𝐻) + 𝑅 𝑘𝑜→ 𝑅1 (J.16)

The researchers extended this to include the effect of irradiation intensity and dissolved oxygen

concentration. The model was later applied by Loeb (2013) in an effort to predict the steady state

formation rate of hydroxyl radicals.

The model proposed by Malato et al. does not, however, account for adsorption effects or the

effects of inhibitory species such as ROS scavengers and competitive adsorbates. A recent paper

by Brame et al. (2015) proposed a complex model for the degradation of organic contaminants

by UV/TiO2 processes. The model, which assumes that adsorption occurs according to the

Langmuir Hinshelwood model, that adsorption capacity (i.e. KA) is independent of irradiation,

that the concentration of ROS in the bulk solution and at the surface is at steady state, and that

degradation reactions take place according to a bi-molecular second order reaction between ROS

and the target contaminant. Brame et al. (2015) assumed that degradation could occur at the

surface or within the bulk water matrix:

349

𝑑𝐶𝐴

𝑑𝑡= −𝑘𝐴𝐶𝑅𝑂𝑆,𝐵𝐶𝐴 −

𝑘𝐴𝐶𝑅𝑂𝑆,𝑆𝐾𝐴𝐶𝐴

1+𝐾𝐴𝐶𝐴 (J.17)

Where dCA/dt is the overall rate of degradation of contaminant A, kA is the degradation rate

constant, KA is the adsorption constant, CROS,B is the concentration of ROS in the bulk water

matrix and CROS,S is the concentration of ROS at the surface.

Through many steps and various assumptions, many of them well supported, Brame et al.

developed a model that accounts for:

ROS mediated degradation in the bulk solution

Adsorption interactions

Degradation of contaminants at the TiO2 surface by adsorbed ROS and photo-generated

holes (combined for simplicity)

Decrease in degradation due to ROS scavengers

Decrease in degradation due to competitive adsorption

Decrease in degradation due to absorption of light by water matrix components (light

attenuation)

The final model proposed by Brame et al. (2015) reads as follows:

𝑑𝐶𝑎

𝑑𝑡=

−𝑃𝑅𝑂𝑆,0

1+𝑘𝑁𝐶𝑁(𝐹+𝐾𝑁𝑆)

𝑘𝐴𝐶𝐴(𝐹+𝐾𝐴𝑆)

10−𝜇ℓ𝐶𝑁 (J.18)

Where PROS is the rate of ROS production, PROS,O is the rate of ROS production in the absence of

any other light absorbing species (light attenuation), 10-lCN is a multiplier to account for light

attenuation, and CN, kN, and KN refer to the concentration, reaction rate constant, and adsorption

constant for inhibitory compound N. The terms F and S are defined as follows (D is the diffusion

coefficient of ROS in the bulk medium):

𝐹 = 1

1+𝑘𝐴𝐶𝐴+𝑘𝑁𝐶𝑁

𝐷

(J.19)

𝑆 =1

1+𝐾𝐴𝐶𝐴+𝐾𝑁𝐶𝑁 (J.20)

The full derivation of the Brame et al. model has been omitted from this document in the interest

of simplicity. The thoroughness of the model is impressive, however, its application to a specific

350

system is no small feat. An extensive array of experiments must be conducted to accurately

calculate the reaction rate and adsorption constants for the target contaminant and the various

inhibitory compounds that may be present in the water matrix. These experiments were well

outside the scope of the current project, however, it may be a useful starting point for future

projects.

A Model to Predict Energy Requirements

As was alluded to earlier in this appendix, it is difficult to compare the results of the current

project to those reported by other AOP researchers, particularly in terms of energy usage,

because the reaction rates observed in each study are a function of the experimental set-up used.

The only way to avoid this would be to directly link the overall degradation rate of the target

contaminant to the number of photons that hit the nanoparticle and result in the formation of

useful oxidative species. As described previously, this is not a simple proposition because of the

numerous oxidative species formed in photocatalytic systems, the importance of adsorption to

the overall degradation process, and the various ROS scavengers, adsorption competitors, and

other interfering constituents that occur in natural water matrices. A simplified model showing

some of the important factors that need to be taken into account is provided in Figure J.8. Please

note that the overall observed degradation rate as defined in this model is the sum of the

following:

A: Apparent degradation rate via hydroxyl-radical-mediated reactions

B: Apparent degradation rate via direct electron-hole-mediated reactions

C: Apparent degradation rate via superoxide-mediated reactions

The simplified model in Figure J.8 does not account for the following:

Nanoparticle agglomeration and subsequent effects on available surface area

The complex and interrelated nature of the reactions between electrons, electron holes,

oxygen, and water molecules that give rise to the various ROS

ROS other than hydroxyl and superoxide radicals

Formation of secondary radicals within the bulk water matrix

351

Figure J.8 Simplified model describing the degradation of an organic contaminant via TiO2 photocatalysis

352

References

Alrobayi, E.M., Algubili, A.M., Aljeboree, A.M., Alkaim, A.F., Hussein, F.H. (2017)

Investigation of photocatalytic removal and photonic efficiency of maxilon blue dye GRL in the

presence of TiO2 nanoparticles, Particulate Science and Technology, 35 (1), 14-20






215

Bolton, J.R. and Cotton, C.A. (2008) Ultraviolet Disinfection Handbook, American Water Works

Association, Colorado, USA

Brame, J., Long, M., Li, Q., Alvarez, P. (2015) Inhibitory effect of natural organic matter or

other background constituents on photocatalytic advanced oxidation processes: Mechanistic

model development and validation, Water Research, 84, 362-371

Collins, J. and Bolton, J.R. (2016) The Advanced Oxidation Handbook, American Water Works

Association, Denver, CO



Research, 43, pp. 1597-1610

Health Canada (2017) Guidelines for Canadian Drinking Water Quality – Summary Table,

Water and Air Quality Bureau, Healthy Environments and Consumer Safety Branch, Health

Canada, Ottawa, Ontario

Kisch, H. and Bahnemann, D. (2015) Best practice in photocatalysis: Comparing rates or

apparent quantum yields? The Journal of Physical Chemistry Letters, 6, 1907-1910

353




Loeb, S. (2013) Nanostructured Photocatalysis for Water Purification, MASc Thesis, University

of Toronto






Ohtani, B. (2010) Photocatalysis A to Z, Journal of Photochemistry and Photobiology C:

Photochemistry Reviews, 11, 157-178

Ollis, D. (2013) Photocatalytic treatment of water: Irradiance influences, in: Photocatalysis and

Water Purification: From Fundamentals to Recent Applications, Ed. Pichat, P., Wiley-VCH

Verlag GmbH and Co.

Toor, R. and Mohseni, M. (2007) UV-H2O2 based AOP and its integration with biological

activated carbon treatment for DBP reduction in drinking water, Chemosphere, 66, pp. 2087-

2095

Yen, H.Y. and Yen, L. S. (2015) Reducing THMfp by H2O2/UV oxidation for humic acid of

small molecular weight, Environmental Technology, 36 (4), 417-423

development and evaluation of photocatalytic linear ... · stephanie gora doctor of philosophy...

Documents