USF Libraries
USF Digital Collections

Mechanistic modeling of photocatalytic water disinfection

MISSING IMAGE

Material Information

Title:
Mechanistic modeling of photocatalytic water disinfection
Physical Description:
Book
Language:
English
Creator:
Dalrymple, Omatoyo Kofi
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla
Publication Date:

Subjects

Subjects / Keywords:
Double Layer
Drinking Water
E. Coli
Solar Applications
Titanium Dioxide
Dissertations, Academic -- Environmental Engineering Microbiology -- Doctoral -- USF   ( lcsh )
Genre:
bibliography   ( marcgt )
non-fiction   ( marcgt )

Notes

Summary:
ABSTRACT: The main goal of this research was to develop a mechanism-based model for photocatalytic disinfection of bacteria in water using suspended catalyst particles in batch reactors. The photocatalytic disinfection process occurs as a semiconductor photocatalyst, most commonly titanium dioxide (TiO<sub>2</sub>), is irradiated with light of wavelength less than 380 nm to produce hydroxyl radicals and other highly reactive oxidants which can inactivate microorganisms. Photocatalytic disinfection involves a complex interaction of many fundamental mechanisms such as light absorption and scattering by semiconductor particles, electrochemical surface reactions, and heterogeneous colloidal stability. Current models, based largely on chemical reacting systems, do not adequately account for these fundamental mechanisms. Even the Langmuir model developed for heterogeneous systems cannot describe the interactions of such large colloidal particles. As a result, it is difficult to assess the combined effects of many important factors which go into the design of a photocatalytic disinfection system. A mechanistic modeling approach is desirable because it provides a framework to understand the influence of many important parameters on the disinfection process. It requires a description of the physical properties of the catalyst, the nature of the suspending electrolyte solution, the physical and chemical properties of the cell surface, and the energetic aspects that influence the interaction of the particles. All these aspects are interrelated. While it is customary to envision the adsorption of reactants unto a catalyst surface, for photocatalytic disinfection involving suspended catalyst particles, multiple catalyst particles adhere to the bacterial surface. In this work a mechanistic model has been developed that simulates the effect of light intensity and catalyst concentration on the disinfection process. The simulations show good agreement with the experimental data for stable colloidal suspensions, that is, suspensions in which rapid aggregation of cells and TiO<sub>2</sub> do not occur. Increased disinfection rates and high levels of inactivation can be achieved by maintaining a relatively low catalyst-to-microbe ratio while maximizing the light intensity. The influence of pH and ionic strength on the disinfection process have been included in the model, but these are only expected to be accurately predicted when the solution remains stable.
Thesis:
Disseration (Ph.D.)--University of South Florida, 2011.
Bibliography:
Includes bibliographical references.
System Details:
Mode of access: World Wide Web.
System Details:
System requirements: World Wide Web browser and PDF reader.
Statement of Responsibility:
by Omatoyo Kofi Dalrymple.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 219 pages.
General Note:
Includes vita.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
usfldc doi - E14-SFE0004895
usfldc handle - e14.4895
System ID:
SFS0028150:00001


This item is only available as the following downloads:


Full Text

PAGE 1

Mechanistic Modeling of Photocatalytic Water Disinfection by Omatoyo Kofi Dalrymple A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy Department of Civil & Environmental Engineering College of Engineering University of South Florida Co Major Professor: Yogi Goswami, Ph.D. Co Major Professor: Maya Trotz, Ph.D. Elias Stefanakos, Ph.D. Vicki Luna, Ph.D. Vinay Gupta, Ph.D. Jeffrey Cunningham, Ph.D. Date of Approval: March 8 2011 Keywords: D rinking W ater D ouble L ayer E. coli S olar A pplications T itanium D ioxide Copyright 2011, Omatoyo Kofi Dalrymple

PAGE 2

DEDICATION To my wife, Rhonda, and son, Jaden; my father, Norman Dalrymple, and mother, Agnes Dalrymple; my siblings, Najuma, Olatunde, Kojo, Palesa, Monifa, Ayodele, and Kayode ; my nieces, Palesa, Ticisa, Jenncia, and Takiyah; and nephews, Randy, Akintunde, Rashad, Keyondre, Raydon, and Azriel

PAGE 3

ACKNOWLEDGEMENTS Research, particularly of this magnitude, can never be accomplished by one person. The work contained herein was conducted under the auspices of the Clean Energy Research Center (CERC) at the University of South Florida with financial support from the Florida Energy Systems Consortium (FESC). Special than ks to my research committee and my co major professors Dr Yogi Goswami and Dr. Maya Trotz Special thanks to Dr. Elias Stefanakos, the Director of the CERC. I also want to thank Dr. Vinay Gupta for allowing the use of his dynamic l ight s cattering equipment, and Dr. Norma Alcantar and her graduate student Eva Williams for teaching me to make lipid vesicles in their lab. A special thank you to Mr. Charles Garretson, our exceptional lab manager and a truly brilliant engineer who can build almost anything that can be conceived. I also want to thank Mr. Ed Haller for his patience in teaching me to use the transmission electron microscope. I am indebted to my colleagues, who offered their assis tance when I needed it the most; Dru Latchman, Ray Morris, Matthew Cutter, Russell Ferlita, Mohammad Abutayeh, Ken Thomas, Ricardo Vasquez, Ana Lucia Prieto, Gokman Demirkaya and Yangyang Zhang.

PAGE 4

i TABLE OF CONTENTS LIST OF TABLES ........................................................................................................... vi LIST OF FIGURES ........................................................................................................ vii LIST OF SYMBOLS ...................................................................................................... xii ABSTRACT ..................................................................................................................... xx CHAPTER 1: INTRODUCTION .................................................................................... 1 1.1 The global water crisis ...................................................................................... 1 1.2 Traditional and low cost disinfection options ................................................... 1 1.3 Advanced treatment processes .......................................................................... 2 1.4 The case for photocatalytic disinfection ........................................................... 3 1.5 Problem statement ............................................................................................. 6 1.6 Research objective ............................................................................................ 7 CHAPTER 2: PHOTOCATALYSIS .............................................................................. 8 2.1 Defi nition .......................................................................................................... 8 2.2 Semiconductor band structure ........................................................................... 8 2.3 Electronic excitation and formation of charge carriers ................................... 10 2.4 Titanium dioxide photocatalyst ....................................................................... 11 2.5 Aqueous phase photocatalysis ........................................................................ 13 CHAPTER 3: MICROBIOLOGICAL CONTAMINATION OF WATER .............. 18 3.1 Pathogenic agents of waterborne diseases ...................................................... 18 3.1.1 Bacteria .............................................................................................18

PAGE 5

ii 3.1.2 Viruses ..............................................................................................19 3.1.3 Protozoa ............................................................................................19 3.2 The model organism: E. coli ........................................................................... 20 3.3 E. coli as an indicator of biological contamination ......................................... 20 3.4 Standards for microbial contamination ........................................................... 22 CHAPTER 4: MICROBECATALYST INTERACTIONS ....................................... 24 4.1 Introduction ..................................................................................................... 24 4.2 Catalyst sur face electrochemistry ................................................................... 24 4.3 Bacterial cell surface electrochemistry ........................................................... 29 4.3.1 Structural composition of bacterial surface ......................................30 4.3.2 Surface charges and ionizable functional groups ..............................34 4.3.3 Electric double layer at bacterial surface ..........................................36 4.4 Microbe catalyst electrical double layer interactions ..................................... 39 CHAPTER 5: REVIEW OF WATER DISINFECTION MODELING .................... 43 5.1 Introduction ..................................................................................................... 43 5.2 Empirical models ............................................................................................ 45 5.2.1 ChickWatson model ........................................................................45 5.2.2 Delayed ChickWatson model ..........................................................46 5.2.3 Hom model ........................................................................................47 5.2.4 Kinetic power law models ................................................................48 5.2.5 Probabilistic models ..........................................................................48 5.3 Mechanistic models ........................................................................................ 50 5.3.1 Series event model ............................................................................50 5.3.2 Multi target model ............................................................................52 5.3.3 Haas model ........................................................................................52 5.3.4 Marugn model .................................................................................53 CHAPTER 6: CONCEPTUAL MODEL FOR PHOTOCATALYSIS ...................... 55 6.1 Introduction ..................................................................................................... 55 6.2 Theoretica l model formulation ....................................................................... 56 6.3 Adsorption kinetics of catalysts and cells ....................................................... 58

PAGE 6

iii 6.3.1 Adsorption in the absence of mechanical mixing .............................60 6.3.2 Adsorption in the presence of mechanical mixing ............................61 6.4 Surface coverage of catalyst on bacteria ......................................................... 61 6.4.1 Surface coverage with low catalyst concentration ............................61 6.4.2 Surface coverage with high catalyst concentrations .........................62 6.5 Kinetics of hydroxyl radicals at interface ....................................................... 65 6.5.1 Generation rate ..................................................................................65 6.5.2 Nature of OH radicals at the bacterial membrane .............................66 6.6 Microbial survival ........................................................................................... 67 6.7 Kinetics of byproduct evolution ...................................................................... 68 6.8 Adsorption and inhibition kinetics of inorganic ions ...................................... 71 6.9 Model for overall inactivation kinetics ........................................................... 74 6.9.1 Ma ss balance of live cells .................................................................74 6.9.2 Mass balance of byproducts ..............................................................75 6.9.3 Mass balance of OH radicals ............................................................75 CHAPTER 7: EXPERIMENTAL DESIGN AND PROTOCOLS ............................. 77 7.1 Selection of experimental factors .................................................................... 77 7.2 Method of data analysis .................................................................................. 78 7.2.1 Statistical analysis .............................................................................78 7.2.2 Numerical analysis ............................................................................79 7.3 Microbiological methods ................................................................................ 79 7.3.1 Preparation of E. coli culture ............................................................79 7.3.2 Cell harvesting and enumeration ......................................................80 7.3.3 Preparation and storage of growth media .........................................81 7.3.4 Preparation and storage of agar plates ..............................................82 7.4 Photocatalytic experiments ............................................................................. 82 7.4.1 Reactor design and setup ..................................................................82 7.4.2 Catalyst stock solution preparation and storage ................................83 7.4.3 Light source ......................................................................................84 7.4.4 Light intensity measurements ...........................................................85 7.4.5 Preparation of working reaction solutions ........................................87 7.4.6 Sampling and error analysis ..............................................................88 7.5 Fatty acid modification and analysis ............................................................... 90

PAGE 7

iv 7.6 Preparation and characterization of model cell membranes ........................... 91 7.6.1 Preparation of lipid film ....................................................................91 7.6.2 Lipid film hydration and extrusion ...................................................92 7.6.3 Size distribution measurement ..........................................................93 7.6.4 Transmission electron microscopy ...................................................93 7.7 Measurement and analysis of byproducts ....................................................... 93 7.7.1 MDA assay ........................................................................................93 7.7.2 Derivative spectroscopy analysis ......................................................94 7.7.3 LOOH assay ......................................................................................94 CHAPTER 8: RESULTS AND DISCUSSION ............................................................ 95 8.1 Fatty acid modification and analysis ............................................................... 95 8.2 Factorial analysis: Main effects ...................................................................... 96 8.2.1 Light intensity ...................................................................................98 8.2.2 TiO2 concentration ..........................................................................100 8.2.3 Fatty acid modification ...................................................................103 8.3 Interaction effects: Light intensity and TiO2 concentration .......................... 103 8.4 Model validation ........................................................................................... 114 8.4.1 Inputs and fitting parameters ..........................................................114 8.4.2 Survival curve predictions ..............................................................121 8.4.3 Influence of light intensity and catalyst concentration ...................125 8.5 Particle interaction effects and colloidal stability ......................................... 125 8.5.1 Influence of ionic strength on disinfection .....................................127 8.5.2 Influence of pH ...............................................................................132 8.6 Byproduct evolution and peroxidation kinetics ............................................ 136 8.6.1 Lipid peroxidation as proof of membrane damage .........................136 8.6.2 Lipid vesicle composition and size distribution ..............................137 8.6.3 MDA production during photocatalytic experiments .....................139 8.6.4 Effect of supplemental fatty acid on MDA production in cells ......140 8.6.5 Correlation between peroxidati on and disinfection ........................143 8.6.6 LOOH production during disinfection ............................................144 CHAPTER 9: CONCLUSIONS .................................................................................. 146

PAGE 8

v CHAPTER 10: RECOMMENDATIONS ................................................................... 148 REFERENCES .............................................................................................................. 149 APPENDICES ............................................................................................................... 174 Appendix A: Computer Codes ............................................................................ 175 Appendix B: Fatty Acid Spectra ......................................................................... 188 ABOUT THE AUTHOR .............................................................................................. 195

PAGE 9

vi LIST OF TABLES Table 1 : Ionizable functional groups located on the surface of E. coli and the associated acidity constants ( ) for zero salt effects at 25C. Data compi led from Martinez et al [109] and Jiang et al [111]. 34 Table 2 : Adsorption equilibrium constants for some common anions on the surface of TiO2 72 Table 3 Incident light intensity in reactors according to lamp combinations 88 Table 4 : Composition of working reaction solutions 89 Table 5 : Composition of electrolytes in final solution 89 Table 6 : Steps in FAME analysis 91 Table 7 : Percent distribution of major fatty acids 96 Table 8 : Rate constants and reaction order as predicted by the model for unmodified and C16:1 modified orga nisms 117 Table 9 : Rate constants and reaction order as predicted by the model for C18:1 and C18:3 modified organisms 118

PAGE 10

vii LIST OF FIGURES Figure 1 : (a) View of a solar collector field and (b) catalyst recovery system (Courtesy of Plataforma Solar de Almera, Spain) 5 Figure 2 : Typical layout of photocatalytic plant for the treatment of water 6 Figure 3 : Simplified energy band diagram of semiconductors 9 Figure 4 : Schem atic of ph otocatalytic processes on the surface of TiO 2 A semiconductor with a band gap of 3.1 eV, TiO2 requires photons with wavelength less than 400 nm [50]. 11 Figure 5 : Band positions of several semiconductors in contact with aqueous electrolyte at pH 1. 13 Figure 6 : Interface of semiconductor and aqueous solution showing band bending for an ntype semiconductor 15 Figure 7 : E. coli grown on mF Endo plates i n the lab 21 Figure 8 : TiO 2 surface in water: (a) water layer [80]; (b) hydroxylated surface [80]; and (c) schematic of double layer according to Stern Grahame model [87] 25 Figure 9 : Surface hydroxylated species of TiO 2 a function of pH calculated according to equations ( 8) and ( 9) using = 2.4 and = 8 as determined by Korman et al [86] for Degussa P25 at 25C 26 Figure 10 : Typical bacterial cell structure (not to scale) [70] 30 Figure 11 : Outer layers of bacteria 32 Figure 12 : Schematic of bacteria water interface [113] 38

PAGE 11

viii Figure 13 : Proposed model for the interaction between a bac terium and a catalyst particle of radii a1 and a2 respectively, separated by X between their surfaces 40 Figure 14 : Typical bacterial inactivation curves: (a) lag survival followed by exponential decay; (b) sigmoidal; (c) ex ponential (loglinear); and (d) concave downward 45 Figure 15 : Surface coverage of catalyst part icles on bacterial cell 57 Figure 16 : Plot of integrated absorption fraction F s for TiO 2 concentration 66 Figure 17 : Schematic of lipid peroxidation 70 Figure 18 : E. coli growth curve fitted with a continuous logistic function 80 Figure 19 : Standard plot for the c orrelation of cell density and optical density 81 Figure 20 : Reactor apparatus 83 Figure 21 : Schematic of UVA fluorescent lamp used in experiments 84 Figure 22 : Spectral power distribution of PL S 9W/08 lamp (source: manufacturer) 85 Figure 23 : Lamp locations on reactor 86 Figure 24 : Typical plots used to determine incident light intensity by actinometry for pairwise combination of lamps 1 and 2 [(a) (c)], and 3 and 4 [(d) (f)] 87 Figure 25 : Schematic of serial dilution of sample 90 Figure 26 : Probability distribution of survival data for E. coli 97 Figure 27 : The main effects plots for (a) TiO 2 concentration; (b) fatty acid modification; and (c) light intensity on mean survival data at 20 min 98

PAGE 12

ix Figure 28 : Effect of light intensity on disinfection for co ntrol organisms at 0.01 g L1 Degussa P25 TiO2 99 Fig ure 29 : Relationship between intensity and average survival at 20 min 100 Figure 30 : Log linear relationship between relatively high catalyst concentration (0.100.50 g L 1) and E. coli survival 101 Figure 31 : Instantaneous formation and settling of TiO 2 cell aggregates; stable solution of 0.10 g L1 TiO2 with 1106 CFU mL1 cells (left); highly unstable suspension of 1 g L1 TiO2 with 1109 CFU mL1 cells; and unstable suspension of 1 g L1 with 1106 CFU mL1 cells 102 Figure 32 : Interaction plots for the three independent factors at 20 min: (a) fatty acid modification vs. TiO2 concentration; (b) light intensity vs. TiO2 concentration; and (c) fatty acid modification vs. light intensity. 104 Figure 33 : Relationship between survival and TiO 2 concentration at high light intensity 105 Figure 34 : Particle interaction an d light transmission in TiO 2 suspensions 106 Figure 35 : TEM image of TiO 2 particles (dark spots) attached to E. coli Images courtesy of Integrative Biology Microscopy Core Facility, University of South Florida 107 Figure 36 : Dependence of limiting catalyst concentration and catalyst diameter 108 Figure 37 : Theoretical adsorption kinetics of TiO 2 particles (25 nm dia. ) unto E. coli surface under hydrodynamic conditions (stir speed was 600 rpm in test tube reactor) 109 Figure 38 : Light transmission through reactor at high light intensity 111 Figure 39 : OH radical generation rate in TiO 2 suspension at pH 7 in deionized water; (a) high intensity I0 = 4.37105 E L1 s1, (b) mid intensity I0= 2.40105 E L1 s1 and (c) low intensity I0=1.35105 E L1 s1 113

PAGE 13

x Figure 40 : Dependence of normalized OH radical generation rate on catalyst concentration 114 Figure 41 : Box plot of the disinfection rate constant k dis obtained from the model 119 Figure 42 : Typical sigmoidal survival of E. coli at low intensity illumination 121 Figure 43 : Survival curve for E. coli treated at low light intensity 123 Figure 44 : Effect of concentration loading on residual survival of E. coli at high light intensity 124 Figure 45 : potential of P25 TiO2 particles and E. coli cells as a function of pH in 0.01 M (open and filled circles) and 0.10 M (open and filled squares) ionic strength respectively. Data modified Liu et al [236], Fernandez Ibanez et al [149], and Suttiponparnit et al [237]. 127 Figure 46 : Influence of salt content on the disinfection process at pH 7 (light intensity = 2.4105 E L1 s1, TiO2 = 0.50 g L1) 128 Figure 47 : Model simulation of the effect of salt content with previously determined rate constants ( kdis = 3.32x105 pM1.5s1; n = 1.5; kOH = 1 L1.5 CFU1 s1 pM0.5) 129 Figure 48 : Simulated results for effect of salt concentration on disinfection (light intensity = 2.4105 E L1 s1, TiO2 = 0.50 g L1) 130 Figure 49 : Settling of TiO 2 cell colloids (0.5 g L 1 and 110 6 CFU L 1 respectively) in 0.01 M (left), 0.10 M (center), and 0.20 M (right) ionic solutions at pH 7 and 25C. 131 Figure 50 : Total interaction energy ( V T ) as a function of separation distance between E. coli (1000 nm dia.) and P25 TiO2 at pH 7 and 25C: a(1) 0.01 M TiO2 1000 nm dia.; a(2) 0.01 M TiO2 25 nm dia.; b(1) 0.10 M TiO2 1000 nm dia.; b(2) 0.10 M TiO2 25 nm dia. 133 Figure 51 : The effect of pH simulated by the model (unmodified cells treated at mid light intensity with 0.01 g L1 TiO2) 134

PAGE 14

xi Figure 52 : Long term (24 hrs) settling of TiO 2 cell colloid s in solutions of pH 3 (left), pH 7 (center), and pH 11 (right) 135 Figure 53 : Total interaction energy ( V T ) as a function of separation distance between E. coli (1000 nm dia.) and P25 TiO2 (1000 nm dia.) at different pH values 136 Figure 54 : Size distribution by volume based on photon correlation spectroscopy of the lipids vesicles in 1 PBS solution (molar ratio 1:1 PE to PG) 138 Figure 55 : TEM images of PE/PG lipid vesicles. Images courtesy of Integrative Biology Microscopy Core Facility, University of South Florida 138 Figure 56 : MDA production during photocatalytic ex periments with P25 TiO2: I0 = 4.85105 E L1s1, N0 11 C FU L1 : (a) unmodified cells; (b) E. coli PE/PG vesicles; (c) cells supplemented with oleic acid; (d) cell s supplemented with linolenic acid. The data are fitted with a fourth order polynomial 142 Figure 57 : Typical curve for the simulation of byproducts from the model 143 Figure 58 : Time characteristics of lipid hydroperoxide detection during photocatalytic treatment: ( E. coli cells; ( with E. coli phospholipids 145

PAGE 15

xii LIST OF SYMBOLS Hammaker constant J radius of bacterial cell m radius of catalyst particle m inactivation rate around the inflection point a ratio of catalyst particle radius to bacterial cell radius / Weibull probability distribution scale parameter Weibull probability distribution shape parameter disinfectant concentration mg L C concentration and the of specifically adsorbing ions mol L C c oncentration of anion mol L 1 concentration of the specific anion species in solution mol L 1 c mass concentration of catalyst c oncentration of byproducts mol L 1 g s concentration of electrolytes in the bulk solution mol L

PAGE 16

xiii concentration of electrolytes in the ion penetrable layer mol L coefficient of dilution D particle diffusivity tensor D diffusion coefficient in the bulk m 2 s D diffusion coefficient of the bacteria m 2 s D catalyst particle diffusion coefficient m 2 s D relative diffusion coefficient m 2 s electron charge C conduction band electron exposure to disinfectant characteristic lethal exposure dose for the particular organism Faraday constant C mol F integrated absorption fraction G rate of OH radical generation mol L 1 s G normalized rate of OH radical generation valence band hole H dimensionless parameter that defines the effective interaction range I absorbed photon flux E L 1 s

PAGE 17

xiv I incident photon flux E L 1 s I transmitted light flux E L 1 s j vector function describing the flows (flux) of n p m s normalized flux m s normalized stationary adsorption flux m s generic disinfection rate constant product of disinfection rate constant and disinfection concentration ( ) Boltzmann constant J K reaction rate constant for byproduct oxidation mol L 1 s diffusion controlled rate constant rate constant for dissociation or organism radical complex and radical quenching and repair observed rate constant for disinfection mol L 1 s rate constant of inactivation OH radical consumption rate constant L n M n 1 CFU 1 s 1 generic acid dissociation constant surface acidity constant acidity constant in bulk solution Langmuir adsorption rate constant for anion on surface of metal oxide

PAGE 18

xv Langmuir adsorption rate constant for disinfection byproducts on surface of metal oxide [ ( ) ] equilibrium constant for the formation of the cell radical complex empirical constant of the Haas model Langmuir parameter for Marug n model L dimensionless double layer thickness given by total concentration of acidic sites on bacterial surface mol m 2 total concentration of basic sites on bacterial surface mol m 2 reaction order with respect to concentration of OH radicals n number concentration of catalyst particles in bulk solution L n number concentration of catalyst particles on surface of bacteria L n number of particles collected on an element of area bacteria concentration CFU L initial concentration of bacteria CFU L Avogadros number number of adsorption sites per unit area m 2

PAGE 19

xvi concentration of damaged cells in the Marug n model CFU L concentration of undamaged cells in the Marug n model CFU L total surface charge rate of byproduct oxidation mol L 1 g s overall disinfection reaction rate for bimolecular reaction CFU L 1 s observed rate of disinfection CFU L s rate of inhibition due to quenching of radicals by anions mol L 1 s s sink terms in continuity equation microbial survival time s absolute temperature K time lag parameter for initial lag phase in the disinfection process s t relaxation time s U particle translation velocity vector ( ) potential energy of double layer interaction between the two spheres at distance J ( ) potential energy of the electrostatic interactions per unit area between two plates at separation J

PAGE 20

xvii ( ) total potential energy of interaction between two spherical particles separated by a distance J ( ) potential energy for van der Waal interaction between the two particles at distance J ( x ) fluid velocity component directly perpendicular to the interface m s x thickness of the organisms diffusive boundary layer m reaction order with respect to cell concentration valence of electrolyte ion Greek symbols coefficient of attenuation per mass concentration of catalyst L g 1 cm 1 empirical constant of the Haas model Gibbs energy of adsorption per molecule J Gibbs energy of specific interaction J fraction of the population with a critical exposure of dielectric constant within the cell membrane layer F m e lectric constant F m relative permittivity relative dielectric constants of the solution dimensionless interaction energy

PAGE 21

xviii zeta potential V dynamic viscosity Pa s dimensionless surface coverage surface coverage of anion species on metal oxide surface coverage of byproducts total surface coverage surface concentration of particles adsorbed during the transient conditions m aximum theoretical surface coverage for an RSA model Debye Huckel parameter m c oncentration time product constant mg s L electrophoretic mobility m 2 s V s urface charge density C m 3 charge density contribution of the ions in the ion penetrable layer C m 3 fixed surface charge C m 2 specifically adsorbed charge per unit area C m 2 charge density in the double layer C m 3 ratio of time to relaxation time t / t quantum yield of radical generation

PAGE 22

xix e lectrostatic potential of the double layer V e lectrostatic surface potential V electrostatic potential at bacterial cell surface V electrostatic potential at catalyst surface V potential difference across the diffuse part of the double layer V Donnan potential V

PAGE 23

xx ABSTRACT The main goal of this research was to develop a mechanism based model for photocatalytic disinfection of bacteria in water using suspended catalyst particles in batch reactors The photocatalytic disinfection process occurs as a semiconductor photocatalyst, most commonly titanium dioxide (TiO2) is irradiated with light of wavelength less than 380 nm to produce hydroxyl radical s and other highly reactive oxidants which can inactivate microorganisms. Photocatalytic disinfection involves a complex interaction of many fundamental mechanisms such as light absorption and scattering by semiconductor particles, electrochemical surface reactions, and heterogeneous colloid al stability C urrent models based largely on chemical reacting systems, do not adequately account for the se fundamental mechanisms Even the Langmuir model developed for heterogeneous systems cannot describe the interactions of such large colloidal particles As a result, it is difficult to assess the combined effects of many important factors which go into the design of a photocatalytic disinfection system. A mechanistic modeling approach is desirable because it provides a framework to understand the influence of many important parameters on the disinfection process. It requires a description of the physical properties of the catalyst, the nature of the suspending electrolyte solution, the physical and chemical properties of the cell surface, and the energetic aspects that influence the in teraction of the particles. All these aspects

PAGE 24

xxi are interrelated. While it is customary to envision the adsorption of reactants unto a catalyst surface, for photocatalytic disinfection involving suspended catalyst particles multiple catalyst particles adher e to the bacterial surface. I n this work a mechanistic model has been developed that simulates the effect of light intensity and catalyst concentration on the disinfection process. The simulations show good agreement wi th the experimental data for stable c olloidal suspensions that is, suspensions in which rapid aggregation of cells and TiO2 do not occur I ncreased disinfection rates and high levels of inactivation can be achieved by maintaining a relatively low catalyst to microbe ratio while maximizing th e light intensity. The influence of pH and ionic strength on the disinfection process have been included in the model, but these are only expected to be accurately predicted when the solution remains stable.

PAGE 25

1 CHAPTER 1: INTRODUCTION 1.1 The global water crisis Waterborne pathogens including viruses, bacteria and protozoa are responsible for 3.5 billion cases of diarrhea each year and 1.8 million deaths as a result of contaminated drinking water The majority of those affected are children under the age of 5 years [1]. Even though there have been outbreaks in developed nations, waterborne diseases are much more prevalent in developing countries particularly among the poor. In general, access to clean water and basic sanitation is a major problem in many poor communities. Acc ording to the United Nations, as much as 50% of the developing world is affected by the main diseases or infections associated with in adequate water supply and sanitation These include diarrhea intestinal helminth infections, dracunculiasis, schistosomiasis, and trachoma [2]. 1.2 Traditional and low cost disinfection options In addition to being chemically nontoxic, water must also be biologically safe to consume ; that means the potential to cause infection must be removed. In many poor communities, boiling water before consumption is the only effective option available for disinfection. However, boiling can be energy intensive especially to meet the needs of large families. So lar disinfection is a low cost alternative in wh ich water in transparent plastic or glass bottles is exposed to direct sunlight. The dual action of solar infrared heating and ultraviolet irradiation inactivates a range of microorganisms [3 5]

PAGE 26

2 Nevertheless, sola r disinfection is limited to small volumes of clear water which must be consumed soon after treatment because of the potential for re growth of pathogens. By far the most common method to disinfect drinking water for the last 100 years is chlorination In the United States, ab out 98% of municipal water treatment facilities use chlorine, and about 200 million residents receive chlorinated drinking water at home [6]. Chlorine is a power ful oxidant and does not only kill pathogens, it also reacts with dissolved natural organic compounds to form man y chlorinated byproducts (DBP s ) Studies show that some classes of DBPs such as trihalomethanes (THMs) and haloacetic acids (HAAs) are potentially mutagenic and carcinogenic [7, 8] The control of DBPs has become important in water treatment adding another level of difficulty to the process Recent Environment al Protection Agency (EPA) regulations have further limited THMs, HAAs and other DBPs (including chlorite and bromate) in drinking water [9]. As a result, many water systems now limit the use of chlorine to high quality groundwater or reduce total organic carbon prior to disinfection. Another concern of chlorine disinfection is that some organisms tend to develop resistance to chlorine or require higher than normal doses for complete inactivation [10, 11] Relatively high residual chlorine concentration can make drinking water taste and smell unpleasant. Nonetheless, chlorination remains an important disinfection method. 1.3 Advanced treatment processes M any advanced a lternative disinfection processes are now available. These include the use of ozone gas, chlorine dioxide, advanced membrane processes, and

PAGE 27

3 germicidal ultraviolet (UV) irradiation Most of these advanced methods are very effective against a wide range of harmful pathogens. However, the cost may be prohibitive since expensive chemicals and costly equipment are required to generate the disinfectant onsite. T hey are often associated with increased process complexity and safety requirements as well Moreover, ozonation produces harmful byproducts including bromate and other brominated DBPs formed in water s with elevated bromide [12, 13] Chlorine dioxide produce s less harmful disinfection byproducts than chlorine, but the formation of chlorite an d chlorate may be a problem for dialysis patients. Also, chlorine dioxide is less effective against rotaviruses and E. coli bacteria UV disinfection makes use of DNAdamaging shortwave radiation (less than 280 nm) which requires the set up of expensive lighting equipment and is associated with increased energy utilization. 1.4 The case for photocatalytic disinfection In general, these advanced techniques are out of reach and often not suited for the local circum stances of developing countries where contaminated water is a real issue. H owever, heterogeneous photocatalysis may be a suitable alternative because i t is capable of utilizing sunlight directly so it can be used in remote areas, and titanium dioxide (TiO2) is widely available. The reactor setup can also be simple either as a suspendedcatalyst application or t he catalyst may be affixed to the re actor walls The actual disinfection of the pathogens occurs as a result of t he highly reactive hydroxyl radical generated during the process placing the technique among advanced oxidation

PAGE 28

4 processes (AOP). Hydroxyl radicals are among the strongest oxidants and are capable of degrading a wide variety of organic and inorganic pollutants [14 18] The first reported killing of microorganisms, including L. acidophilus, S. cerevisiae and E. coli was by Matsunaga et al [19] Many other researchers have since reported on the use of photocatalysis for water disinfection with much attention given to E. coli largely because it is an indicator of fecal contamination in water systems; see for example [14, 2029] Even the more chemically resistant organisms such as Cryptosporidium and Giardia, have been effectively inactivated by photocatalysis [3, 5, 3032] Heterogeneous photocatalysis is particularly adaptable for applications in developing countries, especially in remote and rural areas where energy supply may be prohibitive [33] In addition, TiO2 is abundant in most countries and relatively cheap and photocatalysis is not known to produce the potentially harmful byproducts associated with other disinfection processes. The potential for solar application was previous ly explored for oxidation of chemicals, but Block e t al [34] were among the first researchers to explore the use of solar illumination to drive the disinfection process. In addition, t he engineering and economic feasibility of these systems were explored in detail by Goswami [35] and Goswami et al. [36] Although they are not currently in widespread use, solar photocatalytic systems have been used with much success in pilot facilities [17, 37, 38] Figure 1 and Figure 2 show a solar photocatalytic system operated in Spain [18] and a simplified system layout for flat plate solar reactors.

PAGE 29

5 Nonetheless as with most treatment options, photocatalytic treatment has its challenges. Firstly, TiO2 has shown the most promise and has become the most widely used photocatalyst, but it is only sensitized by near UV radiation or photons with greater energy. This means that only a very small fraction of sunlight ( < 5%) can be used for solar applications However, the modification of TiO2 t hrough doping with me tals and nonmetals t o enhance its visible light capability has shown tremendous promise [39 45] Secondly, slurry reactors are usually more effective than thin films, but they require an additional post treatment step to separ ate the catalyst ( Figure 1b) adding a level of complexity and increased cost. Thirdly, the rate of disinfection is relatively slow compared to other processes, and like UV and ozone, there is no residual protection in a drinking water distribution system Figure 1: (a) View of a solar collector field and (b) catalyst recovery system (Courtesy of Plataforma Solar de Almera, Spain)

PAGE 30

6 Figure 2: Typical layout of photocatalytic plant for the treatment of water 1.5 Probl em statement The design of a disinfection system relies substantially on the knowledge of the inactivation rate of a target or indicator organism(s) by the disinfectant. For photocatalysis the synergistic effect of catalyst concentration and light intensity on the rate of the process determines the most efficient combination of contact time and dose to emp loy. Currently, most of this information is obtained from benchscale studies and extrapolated with a series of empirical models which do not adequately describe photocatalytic disinfection. The most common application is the Chick Watson model used primarily to fit inactivation data with first order decay or modified for data with an initial lag However, frequent deviations from such models have been reported in the literature [21, 46] These models do not allow designers to explicitly determine the overall influence of

PAGE 31

7 important parameters such as catalyst concent ration, light intensity, ionic strength, and pH on the disinfection process. It is difficult to account for many of the complex interactions which occur during photocatalytic inactivation without over-fitting data with numerous empirical parameters. No stu dy to date has proposed a comprehensive mechanistic model to describe the photocatalytic disinfection which can be used to optimize the design of such systems. A major benefit of a mechanistic model is the significant cost reduction associated with performing fewer preliminary experiments to determine the effectiveness of various combinations of catalyst concentration and light intensity for a given organism. 1.6 Research objective The objective of this research was to develop and apply a mechanistic model ing approach to descri be the kinetics of photocatalytic inactivation for batch reactor systems utilizing suspended TiO2 particle s The overall goal was to build a model which could account for the influence of catalyst concentration, light intensity, ionic strength, and cell membrane fatty acid distribution o n the disinfection process The aim is that the model will serve as a predictive tool to design disinfecti on systems, so that water can be disinfected quickly, efficiently and inexpensively.

PAGE 32

8 CHAPTER 2 : PHOTOCATALYSIS 2.1 Definition H eterogeneous p hotocatalysis is the chemical transformation of a substrate at the interfacial boundary of a solid light absorbing catalyst (photocatalyst) and a water or gas phase In this form of photocatalysis, the role of light is to produce active sites on the surface of the photocatalyst so that subsequent chemical reactions may occur [47] As in catalysis, the catalyst remains unchanged at the end of the cycle [47, 48] + ( 1 ) + + ( 2) 2.2 Semiconductor band structure The energy band structure of s emiconductors allows the absorption of light and generation of charge carriers (electron and hole) which participate in photocatalysis. Semiconductor photocatalysts include TiO2, tungsten oxide ( WO3) tungsten sulfide ( WS2) cadmium sulfide ( CdS ) zinc oxide ( ZnO ) and zinc sulfide ( ZnS ) among others. The electrons in the atoms of a semiconductor crystal occupy different energy levels which tend to overlap with those of electrons confined to neighboring atoms. A ccording to the Pauli Exclusion Principle electron energy levels cannot be the same the electronic structure becomes characterized by a set of closely spaced energy levels, forming an energy band. When the band structure is analyzed, a series of allowed and forbidden

PAGE 33

9 energies are obtained resulting in energy bands separated by energy band gaps [49] Although the energy band diagrams of semiconductors are rather complex, they can be simp lified since only the electrons in the highest almost filled band and the lowest almost empty band dominate the behavior of the semiconductor ( Figure 3) Figure 3: Simplified energy band diagram of semiconductors [49] T he almost empty conduction band is identified by a set of horizontal lines the bottom edge of which is labeled Ec. Similarly, the top of the valence band is indicated by a horizontal line labeled Ev. The energy bandgap, Eg, is located between the two bands. T he energy of a fre e electron outside the crystal is called the vacuum level labeled Evacuum [49]

PAGE 34

10 2.3 Electronic excitation and formation of charge carriers At absolute zero temperature, the valence band is completely filled with electrons while the conduction band is empty. A t room temperature, increased thermal energy reduces the band gap slightly as the atomic vibrations increase. This thermal excitation causes some adjustment to t he energy distribution of the electrons, such that a few have enough energy to cross the energy band gap into the conduction band [50]. Another process through which electrons ca n gain energy to cross the band gap is through photoexcitation. In this case, electrons in the valence band absorb the energy from a photon. This is the initiating step in photocatalysis [48, 50] The photon must provide energy gr eater than or equal to the band gap for the electron to cross the barrier ( Figure 4) The electrons which break free from bonds between neighboring atoms in the solid and enter the conduction band are free to move around, and hence can conduct charge or participate in chemical reactions The bonds from which these excited electrons originated are left with electron vacancies or holes The holes are considered positive charge carriers which appear to move around freely as neighboring electron s move in and out of the vacancy [49] The free electron may migrate to a surface site on the semiconductor and participate in a reduction reaction. Similarly, a suitable electron donor at the surface of the material can be oxidized by the valance band hole ( Figure 4) If the conduction band electron returns to the valence band and fills the vacancy, the process is called recombination and is accompanied by a release of heat and or fluorescence.

PAGE 35

11 Figure 4: Schem atic of photocatalytic processes on the surface of TiO2. A semiconductor with a band gap of 3.1 eV, TiO2 requires photons with wavelength less than 400 nm [50] 2.4 Titanium dioxide photocatalyst TiO2 is a model photocatalyst because it is non toxic, stable (does not self oxidize), and highly active [51] The conduction and valence bands lie in energetically favorable positions to both reduce and oxidize adsorbed species ( Figure 5 ). A compound is oxidized on the catalyst surface when its oxidation potential is above the valence band position of the catalyst (dark gray rectangle). Similarly, reduction takes place when the redox potential of the acceptor is below the conduction band position (light gray rectangle). According to Figure 5, TiO2 not only has the oxidation potential to degrade pollutants, but also the reduction potential necessary for splitting water m olecules to create hydrogen gas [52]

PAGE 36

12 There are three crystalline phases of TiO2; anatase, rutile, and brookite. T he anatase and brookite phases are known to be thermodynamically less stable than the rutile phase and are generally converted to rutile at high temperature [53 56] Band structure calculations revealed that rutile and anatase TiO2 have direct and indirect band gaps respectively [53] In a direct band gap semico nductor the conduction band minimum is directly above the valence band maximum that is, they occur at the same wavenumber [49] This makes rutile much more efficient at absorbing light than anatase, but charge carriers generated in the anatase phase have longer lifetimes making it more photocatalytically active than rutile. However, anatase is co mmonly mixed with rutile to help reduce the rate of recombination [48, 51] The band gap energy of anatase is 3.2 eV and hence absorbs photons of 380 nm or less Ru tile has a slightly lower band gap at 3.1 eV and absorbs into the visible range 418 nm [53, 57] There is a wide range of photoreact ivity within mixtures containing variable contents of anatase and rutile. However, Degussa P25 TiO2 has set the standard for photoreactivity in environmental applications [50, 58] It is a nonporous 70% to 30% anatase to rutile mixture [51, 58] P25 is available as high surface area (50 15 m2g1) nanoparticles with an average individual particle siz e of 2030 nm, even though particle agglomeration in solution can reach 300500 nm [51, 59] The small size of the nanoparticles provides high efficiency of surface trapping of the photogenerated electron and hole thus increasing the probability of a photocatalytic process on the surface of the catalyst.

PAGE 37

13 Figure 5: Band positions of several semiconductors in contact with aqueous electrolyte at pH 1. Adapted by permission from Macmillan Publishers Ltd [60] 2.5 Aqueous p hase p hotocatalysis When a semiconductor i s in contact with an aqueous solution, bond formations with water molecules and other ions occur instantaneously There is a movement of charge between the semiconductor and the solution to create the condition s of equilibrium at the interface of the two phases. This is achieved w hen the electrochemical potential s of the two phases are equal [49] The electrochemical potential of the solution is determined by its redox potential, while in semiconductors the electrochemical potential of the electrons is determined by the Fermi level. The Fermi level is th e energy

PAGE 38

14 level occupied by electrons at absolute zero temperature or the level at which the probability of occupation of an electron is 50% [49] On an energy band diagram, the Fermi level would be located at the mid point of the band gap for intrinsic semiconductors and just below the conduction band for ntype semiconductors such as TiO2. The redistribution of charges at the interface produces the space charge region which extends at a considerable distance (100 10,000 Angstroms) below the surface of the semiconductor [49] Likewise, solute and solvent ions with counter charges are distributed from the surface towards the bulk solution. Th e exchange of charge s also induces changes to t he bulk energy levels in the localized area resulting in a curvature to the energy band near the junction. For an ntype semiconductor, the Fermi level is typically higher than the redox potential of the aqueous solution, and hence electrons are transferred from the semiconductor into the solution. Therefore, there is a positive charge associated with the space charge region, and this is reflected in an upward bending of the band edges ( Figure 6). Since most of the charge carriers have been removed from the space charge region, electron transfer reactions occur slowly, if at all. However, if the semiconductor is exposed to radiation of sufficient energy, electrons can now be promoted to the conduction band. Electron hole pairs generated in the region of the electric field, i.e., the spacecharge region, are separated efficiently rather than undergoing immediate recombination. This forces the photogenerated electron towards the bulk of the

PAGE 39

15 semiconductor, where it can be transferred through a surface site to a point where an electron acceptor can be reduced. The photogenerated hole, under the influence of the electric field, migrates to wards the interface to a site where it can oxidize a suitable electron donor in the solution [50] Figure 6: Interface of semi conductor and aqueous solution showing band bending for an n type semiconductor [49] The abso rption of energy and the subsequent generat ion of the electron hole pair are the initiating step s in the photocatalytic process which may be represented as follows [61, 62] : Ti O + h ( Ti O ) + ( Ti O ) ( 3) where is the conduction band electron and is the valence band hole. The interaction of the hole with a water molecule or hydroxide ion produces the very reactive hydroxyl radical ( OH ). These radicals are bound or diffuse from the surface of the semiconductor and act as the primary oxidants in the photocatalytic system [61, 63] The formation of the radicals is illustrated below :

PAGE 40

16 H O + OH + H ( 4 ) O H + OH ( 5) A typical reaction of the bound radical with an organic compound such as glucose may be illustrated as in E quation ( 6) B acterial cells are predominantly water and the major cellular constituents, such as polysaccharides, lipids, proteins and nucleic acids are mostly organic. They react with the hydroxyl radical in a similar way and this subsequently leads to cell death. C H O + H O + OH CO + H O ( 6) Oxidation of compounds may also occur directly via the valence band hole before it is trapped either within the particle or at the particles surface. Nevertheless, the presence of hydroxyl radicals in aqueous solutions of illuminated TiO2 ha s been confirmed by researchers and many intermediates are consistent with those found when organic compounds react with a known source of hydroxyl radicals [64 67] The chemical properties pollutant and the reaction conditions largely determine which mechanism will dominate. However, the presence of hydroxyl radicals is very important for the complete phot ocatalytic destruction of many organic compounds and the inactivation of pathogens. Cho et al. [68] found a linear correlation between hydroxyl radicals and the i nactivation of E. coli in water disinfe ction studies.

PAGE 41

17 The photogenerated conduc tion band electrons are trapped at the surface by TiIV sites and result in TiIII sites. Oxygen adsorbed at TiIII sites may result in the superoxide radical from a charge transfer reaction as shown below : ( Ti ) + O ( Ti ) + O ( 7) The superoxide radical is also relatively reactive and capable of oxidizing cellular constituents. Since all these processes occur simultaneously, photocatalysis may proceed via different pathways depending on the reaction conditions and oxidizable substrates. However, for oxidation of a compound to occur, the presence of oxygen or another suitable electron acceptor (such as hydrogen peroxide) is necessary.

PAGE 42

18 CHAPTER 3: MICROBIOLOG ICAL CONTAMINATION OF WATER 3.1 P athogen ic agents of waterborne diseases P athogens are a class of microorganism s, including bacteria, viruses, and protozoa, able to cause disease in humans (also plants and animals). The majority of w aterborne diseases and infections are caused by bacteria, viruses, and protozoa Pathogens have genetic, biochemical or structural features which allow them to overcome the defense mechanism of the host, and invade and colonize tissues, or produce toxins They are transmitted through the dire ct consumption of contaminated water. In some cases, the consumption of food prepared with contaminated water results in the same infections and diseases [69] In general, microorganisms are ubiquitous, but pathogens tend to enter water sources particularly through contact with human and animal fecal matter. 3.1.1 Bacteria Bacterial pathogens include members of the genus Salmonella and Shigella, cholera causing Vibrio cholera, and some strains of E coli They are mostly rodshaped organisms which infect the gastrointestinal tract and are excreted in t he feces of infected humans and other animals [70] However, there are also some waterborne bacterial pathogens, such as Legionella, Burkholderia pseudomallei and atypical mycobacteria, which can grow in water and soi l [69] Escherichia Salmonella, and Shigella ar e genetically closely related [70] However, while many strains of Escherichia are

PAGE 43

19 harmless, members of the Salmonella and Shigella genus are usually pathogenic. Escherichia are almost universal inhabitants of the intestinal tract of humans and warm blooded animals and m any species play a nutritional role by synthesizing vitamins, particularly vitamin K [70]. 3.1.2 Viruses Viruses are microorganisms that lack many of the attributes of cells, the most important of which is they can only reproduce within a living host cell [70] They are much smaller than bacteria (can range from 10 100 nm) but unlike bacteria, they do not have metabolic abilities of their own They are also known to infect microbial cells. Water borne vir al pathogens include the hepatitis A virus polio virus adenovirus, and rotavirus among others [1, 69, 71] Many are excreted in the feces of infected individuals and may contaminate water intended for drinking. Waterborne viral infections often affect the gastrointestinal tract, and among other symptoms result in severe diarrhea, nausea, and abdominal pain. 3.1.3 Protozoa Protozoa are eukaryotic cells which are generally larger and structurally more complex than bacteria and viruses Cryptosporidium parvum and Giardia duodenalis (previously known as Giardia lamblia) belong to this group of pathogens. They live in the intes tines of humans and large mammals and pose significant threat to public health [72] These two organisms are difficult to disinfect because they are transmitted through

PAGE 44

20 wa ter in dormant, resistant forms, known as cysts and oocysts [11, 73] However, t hey ma y be removed through filtration and other advanced treatment techniques [11, 74, 75] 3.2 The model organism: E. coli E. coli is the name given to a group of rodshaped Gram negative bacteria which usuall y inhabit the intestines of humans and warm blooded animals Gram negative bacteria are cells whose membrane thickness and composition do not allow them to retain the gram stain. On the other hand, Gra m positive bacteria easily retain the gram stain. On average, an E. coli bacterium measures about 0.5 microns in diameter and 1 micron in length. It is a facultative anaerobe, which can switch from aerobic respiration to fermentation to meet its energy nee ds. E. coli is the most studied microorganism in the world. It has found extensive use as a model organism in molecular genetics and molecular biology. However, it is also an excellent model for bacterial pathogens for three important reasons. Firstly, th ere is a wealth of biological data available for E. coli Secondly, other important pathogens such as Salmonella and Shigella are genetically very similar to E. coli Salmonella shares about 50% of its genome with E. coli while Shigella shares about 70% [70] Thirdly, it is easy to culture in the lab and there are many non pathogenic strains to work with 3.3 E. coli as an indicator of biological contamination E. coli along with a number of other similar enteric b acterial species, constitutes the total coliform group. A specific subgroup of this collection is the fecal coliform bacteria, the most common member being E. coli These organisms may be distinguished

PAGE 45

21 from the others by their ability to ferment lactose at 440.5C in the fecal coliform test In addition, when cultured on a specific plate (e.g. mF Endo) a positive result for E. coli is metallic green colonies on a dark purple media ( Figure 7 ) The presence of fecal coliform bacteria in water is usually an indication that fecal matter from hu man s or other animals is present It also suggests that other microorganisms associated wi th fecal matter, and of more significant virulence, possibly exist In this way, E. coli is used as an indicator organism for the biological contamination of water. Figure 7: E. coli grown on mF Endo plates in the lab

PAGE 46

22 3.4 Standa rds for microbial contamination In the United States (US) the EPA sets the rules and establishes the guidelines for drinking water quality. The Safe Drinking Water Act is the primary federal law that governs the provision of potable water to the public [9]. Under the Act, the EPA has the power to set water quality standards. The Agency uses t he Total Coliform Rule, published in 1989, to establish microbiological standards for public water systems [76] The rule set s both nonenforceable maximum contaminant level goals (MCLG) and legal maximum contaminant limits (MCL) for the presence of total coliform in drinkin g water. The MCLG for total coliform which includes E. coli, is set at zer o. The MCL is based on the presence/absence of total coliforms in samples rather than actual counts of bacteria. For water systems which take less than 40 routine samples per month, 39 must be negative for total coliform. For water systems taking more than 40 samples per month, 95% must be negative for total coliform. The number of routine samples per month is determined by the number of consumers that the water system serves. Currently, t he EPA is proposing the elimination of the MCLG and MCL provisions for total co liforms and fecal coliforms, and the inclu sion of an MCLG and MCL for E. coli and a treatment technique for total coliform s [77] The World Health Organization (WHO) provides guidelines to assist countries in verifying drinking wa ter quality. The guidelines are very similar to the US EPA requirements in that E. coli is the indicator organism of choice and the overall goal is to have no indicator organisms present in drinking water. However, WHO suggests that t hermo tolerant coliforms may be used as an alternative to the test for E. coli in many

PAGE 47

23 circumstances. The WHO guideline value for microbial quality is the absence of an indicator organism in 100ml samples [71]

PAGE 48

24 CHAPTER 4: MICROBECATALYST INTERACTIONS 4.1 Introduction For photocatalytic disinfection to occur, microbes must be in close proximity or make contact with the surface of the semiconductor to allow for the exchange of electrons and subsequent chemical reactions. Although TiO2 has been studied extensively to disinfect microorganisms, most of wh at is known about microbe catalyst interactions in aqueous suspensions is qualitative. No study has quantitatively assessed the significance of these interactions on the disinfection process. The important concepts which are related to microbe catalyst int eractions are discussed in this section. Since E. coli is the subject of the investigation, the scope of the discussion has been limited to bacteria. 4.2 Catalyst surface electrochemistry The surface of a metal oxide particle in an electrolyte solution is almost always electrically charged. Upon exposure to water, there is a spontaneous formation of an adsorbed water layer of oriented water dipoles [78, 79] The terminal oxygen atoms at the surface react with water to produce hydroxylated sites ( Figure 8), which are involved in proton exchange reactions imparting a pH dependent surface charge [80 82] In the case of TiO2, the hydroxyl groups on the surface are known to undergo the following acid-base reactions [50] :

PAGE 49

25 TiO H TiOH + H ( 8) TiOH Ti O + H ( 9) where and are the surface acidity constants, which are related to the acidity constant in the bulk solution as [50, 83] : = ( 10) = ( 11) where is the surface potential e is electron charge is the Boltzmann constant and is absolute temperature The pH dependence of the dominant surface species for TiO2 is shown in Figure 9. The surface is known to have a net surface charge of zero close to pH 6 when the neutral TiOH species covers most of the surface sites [ 8486] Figure 8: TiO2 surface in water: (a) water layer [80] ; (b) hydroxylated surface [80] ; and (c) schematic of double layer according to Stern Gra ha m e model [87]

PAGE 50

26 Figure 9: Surface hydroxylated species of TiO2 a function of pH calculated according to equations ( 8) and ( 9) using = 2.4 and = 8 as determined by Korman et al [86] for Degussa P25 at 25C The adsorption of organic molecules or surfa ceactive ions may also occur at the surface. The distribution of the electrolyte ions at the interface and the electric potential play a key role in the stability of catalyst suspensions during photocatalysis [88 91] as well as their post treatment recovery [92, 93] Figure 8 schematically shows the electric double layer at the TiO2 surface in contact with a solutio n according to the SternGrahame model [94, 95] Species are attracted to localized surface sites via electrostatic or hydrophobic effects and displace the primary adsorbed water layer, becoming specifically adsorbed on the oxide surface [96 98] This type of short range interaction is generally called specific adsorption and the ions lose a portion of their hydration shell to become part of the monolayer at the surface. This is particularly the case for anions, since the hydration ene rgies are generally higher for cations [99] The specific adsorption of TiOH TiOH2 +TiO-0% 25% 50% 75% 100% 0 2 4 6 8 10 12% active sitepH

PAGE 51

27 chloride, sulfate, and phosphate ions has been observed on the surface of TiO2 [100, 101] The plane of mean charge of the specifically adsorbed ions defines the inner Helmholtz layer (IHL). The amount of specifically adsorbed charge per unit area can be expressed using a modified Langmuir isotherm [50] : = exp 1 + exp ( 12) in which is the Gibbs energy of adsorption per molecule according to, = + ( 13) where represents the electrostatic interaction energy and is the Gibbs energy of specific interaction. is the number of adsorption sites per unit area. C and z are the bulk concentration and the valence of specifically adsorbing ions, respe ctively. Some ions are adsorbed to the surface through longrange coulombic interactions. They tend to retain their hydration layer and are therefore restricted in their approach to the surface. The mean geometric location of their charge centers defines the outer Helmholtz layer (OHL). The IHL and OHL together constitute the Stern layer. Beyond this region lies the so called diffuse layer in which ions are fully mobile, and whose spacing from one another is a function of the total ionic concentration in bulk solution. The concentration of ions in this layer is governed by the need to maintain overall charge neutrality, including those species adsorbed at the surface of the metal oxide. The concentration of ions in the diffuse layer is described by the Bolt zmann distribution,

PAGE 52

28 ( ) = exp ( ) ( 14) where is molar concentration (mol L1) of the ion in the double and is the concentration in the bulk solution. The electrostatic potential of the double layer is given by the Poisson distribution: = ( 15) where is the charge density given as, = ( 16) and is the dielectric permittivity of the solution. Us ing equations ( 15) and ( 16) the PoissonBoltzmann equation for the electric potential profile is derived as, = exp ( 17) Equation ( 17 ) is restricted to low electrolyte solutions because the ions are treated as point charges. Using the Debye Huckel approximation for low potential, i.e., the electric p otential profile is given as, = ( 18) where is the Debye Huckel parameter and is given by, = ( 19) The solution for equation ( 18) for a double layer around a spherical particle of radius a is given as [102]

PAGE 53

29 = exp ( ) ( 20) in which the potential is the potential difference across the diffuse part of the double layer, which is related to t he charge density in the double layer through, = 1 + ( 21) The total surface charge is given as, = 4 ( 1 + ) ( 22 ) The total surface charge and electrostatic potential of the surface are determining factors for behavior of the colloids in suspension. Particles of similar charge tend to be stabilized as they repel each other. If particles have no charge, there is usually no force to prevent their agglomeration. 4.3 Bacterial cell surface electrochemistry The surface of a bacterium is much more complex than the surface of impenetrable solid colloids. A bacterial surface is a heterogeneous threedimensional arrangement of various biomolecules. The surface propertie s may vary at specific locations as a result of the presence of certain structures. Some cells also have structures that protrude from the surface such as fimbriae, pili, and flagella. Fimbriae and pili are thought to be involved in cell attachment to envi ronmental surfaces, while flagella are special structures used for cell locomotion [70] To understand cell electrochemistry, a brief description of the cell surface is necessary.

PAGE 54

30 4.3.1 Structural composition of bacte rial surface The outer surface of a bacterial cell is made up of a cell wall and cytoplasmic membrane which encircles the fluid cytoplasm ( Figure 10). The cytoplasm i s a complex mixture of substances and structures including deoxyribonucleic acid ( DNA) ribonucleic acid ( RNA), ribosomes and other dissolved and suspended materials. T he cell wall and cell membrane act as barriers to prevent unwanted materials from entering the cell, while also holding the internal contents together. Only water and a few other small, uncharged molecules like oxygen and carbon dioxide diffuse freely across the me mbrane. All other substances enter through active transport or diffuse through trans membrane proteins, whose channels open and close according to the needs of the cell. Figure 10: Typical bacterial cell structure (not to scale) [70]

PAGE 55

31 These outer layers are the p rimary means through which an organism interacts with the environment. Most species of bacteria can be divided into two broad groups based on their cell wall by the Gram staining method simply as Gram positive and Gram negative [70] Figure 11 shows the structure of bacterial cell surfaces. The cell wall of both groups is composed of pe ptidoglycan, a peptide cross linked polysaccharide matrix layer. The peptidoglycan layer is made up sheet s formed from individual strands of peptidoglycan lying adjacent to one another. It account s for as much as 90% of the Gram positive cell wall with several (up to 25) sheets stacked upon each othe r to height of 1580 nm. In G ram negative bacteria, it makes up only about 10% of the cell wall (1 2 nm) and is located between the two phospholipid layers; the outer membrane and the cytoplasmic membrane. Peptidoglycan confers rigidity to maintain shape and internal pressure. I n both Gram negative and G ram positive bacteria, peptidoglycan is very porous and allows particles of approximately 2 nm to pass through [103] Approximately 45% of the surface of Gram negative bacteria may be covered with lipopolysaccharide (LPS), which are anchored in the lipids of the outer membrane. It is ma de up of three distinct regions covalently linked together; a hydrophobic lipid component (lipid A), a core polysaccharide, and O antigen. Some bacterial strains may not possess the O antigen side chain. The LPS core polysaccharide consists of five to ten negatively charged sugar units, which often carry phosphate and carboxylic acid groups. The O antigen consists of 20 to 70 repeating units of three to five sugars, which protrude up to 30 nm or more from the cell surface. It is very like ly that the O antig en plays a

PAGE 56

32 major role in polymer interactions with surfaces reported for Gram negative bacteria [104] Figure 11: Outer layers of bacteria Adapted by permission from Pearson Education, Inc. [70]

PAGE 57

33 Similar to the LPS in Gram negative bacteria, the cell wall of Gram positive bacteria may contain teichoic acids which are attached, directly or indirectly by way of phosphodiester bonds, to carbon 6 of N acetylmuramic residues of the peptidoglycan, or anchored in the underlying lipid bilayer. In the latter case, these are called lipoteichoic acids and are covalently bound to the lipid bilayer via a glyceride. In general, teichoic acids include all wall, me mbrane, or capsular polymers of either ribitol phosphate or glycerophosphate residues. They are connected via phosphodiester bonds and usually have other sugars and Dalanine attached. Both Gram negative and Gram positive bacteria have a cytoplasmic membrane composed almost entirely of lipids and proteins. In Gram negative bacteria, a second phospholipid bilayer is present in the outer cell membrane. P hospholipid bilayers are composed of conventional glycerol phospholipids, mainly phosphatidylethanolamine ( PE) phosphatidylglycerol (PG), and cardiolipin [105107] Phospholipids have a hydrophobic head and two hydrophobic tails and are arranged in a twolayer sheet with the tails pointing towards the center of the layer. The head of the lipid is generally made up of a negatively charged phosphate group and glycerol. The tail is usually a long chain of fatty acid hydrocarbons. Finally, the cell wall and cytoplasmic membrane are populated with proteins which are either firm ly embedded (integral proteins) or associate firmly with one of the membrane structures (peripheral proteins). Some proteins bind substrates or process large

PAGE 58

34 molecules for transport into the cell, while lipoproteins are involved in energy metabolism and ot her important cellular functions. 4.3.2 Surface charges and ionizable functional groups Much of the charge on a bacterial cell surface is derived from functional groups associated with the surface structures. Bioassay studies suggest that the charge on the cell wall results predominantly from proton exchange reactions involving carboxylic, phosphate, and amino moieties [108111] The reactions for the dominant functional groups in E. coli and the range of their associated acidity constants ( ) are shown in Table 1. Table 1: Ionizable functional groups located on the surface of E. coli and the associated acidity constants ( ) for zero salt effects at 25C. Data compiled from Martinez et al [109] and Jiang et al [111] Reaction Location R COOH R C O O + H Proteins, sugars and LPS 2.0 6.0 R N H 3 + R N H 2 + H Proteins and phospholipids 9.0 11.0 R HP O 4 R P O 4 + H Phospholipids 3.2 3.5 R H P O R H P O + H LPS 3.2 3.5 R HP O R P O + H LPS 5.6 7.2

PAGE 59

35 Considering that the site density of carboxyl and phosphate groups is generally greater than amines, the cell surface of E. coli like most bacterial cells is negatively charged at neutral pH [109, 112] In the absence of other ions, the surface charge density resulting from the ionizable functional groups at the bacterial surface may be derived by considering the generic proton exchange reactions, LH L + H ( 23) L H L + H ( 24) where L is the protonbinding site on the cell surface for acidic and basic moieties respectively. The apparent equilibrium constants () for equations ( 23) and ( 24) are defined as, = [ H ] [ L ] [ LH ] ( 25) = [ H ] [ L ] [ L H ] ( 26) The fixed surface charge associated with the various sites is given by, = [ H ] + [ H ] , + [ H ] ( 27) where and are the total concentrations of basic and acidic sites, respectively. The acidity constants associated with each site must be adjusted according to Equations ( 10) and ( 11) to account for the electrostatic influence of the surface.

PAGE 60

36 4.3.3 Electric d ouble layer at bacterial surface Since a bacterial surface has a threedimensional configuration into which ions and solvent molecules are able to penetrate, the bacteriawater interface may best be described as an ion penetrable layer with volume spread electric charge [113117] Figure 12 schematically shows the distribution of ions at the bacterial surface according t o the ionpenetrable model. The charges associated with the ionizable functional groups attract counter ions, but there is no definite boundary at the molecular level. Polymers and surface appendages may also change conformation depending on the ionic char acter of the microscopic local environment [112, 118] Unlike a hard colloidal particle, the bacterial surface has a finite thickness which restricts the charges within the ion penetrable layer. Surface charge density may be deduced from proton titration experiments [110] However, since it is difficult to determine the spatial dis tribution of the charge through the cell membrane, it is usually assumed to be uniformed. The electric potential of the ion penetrable layer is made up of the fixed charges associated with functional groups, as well as the charge density of the ions which have diffused into the layer [117] To derive the electric potential within the layer, Equation ( 15) may be adjusted appropriately as follows = ( ) + ( ) ( 28) where is the charge density contribution of the ions in the ion penetrable layer and is the dielectric constant within the membrane layer. The ions in the membrane have an energy which is equal to and follow the Boltzman distribution. Therefore, the

PAGE 61

37 concentration of ions in the ion penetrable membrane is given by Equation ( 14) The semi permeable cytoplasmic membrane maintains an unequal distribution of ions on either side of the membrane. At equilibrium, the electrostatic potential across the m embrane is called the Donnan potential, Equation ( 14 ) may therefore be rewritten as, = exp ( 29) To satisfy conditions of charge neutrality in the membrane, the following is true + exp = 0 ( 30)

PAGE 62

38 Figure 12: Schematic of bacteria water interface [113] A particular solution for Equation ( 30) gives the Donnan potential for a membrane in contact with a 11 electrolytic solution as [116] = arcsinh 2 ( 31) where F is the Faraday constant.

PAGE 63

39 Various approaches have been taken to derive the electric potential across the cell membrane. A useful approach is to assume infinite thickness of the membrane, even though the solution indicates that the electric field only exists within a finite thicknes s of the membrane [119] However, the origin ( = 0 ) is located at a hy pothetical boundary between the membrane and the electrolyte solution such that < 0 represents the membrane, and > 0 is the electrolyte solution. The Poisson Boltzmann equation for this model is given as, = 1 + exp for < 0 ( 32) = 1 exp for > 0 ( 33) where m and s are the relative dielectric constants of the membrane and the solution respectively. Equations ( 32) and ( 33 ) can be solved numerically after applying the appropriate boundary conditions [117, 119, 120] to yield the electric potential profile across a cell membrane. 4.4 Microbe catalyst e lectrical double layer interactions Since contact between the catalyst and the microbe is a prerequisite for photocatalysis, interactions which enhance contact without destabilizing the suspension should result in more effective disinfection The interaction between the two colloids, as described by classical DLVO theory [121] is governed by the balance of repulsive and attractive forces, usually summed up in electrostatic and van der Waals forces. Electrostatic forces can be both repulsive and attractive depending on the overall charge

PAGE 64

40 of the colloids, while van der Waals interactions are usually attractive. Bacterial surface polymers may also play a major role during the interaction [112, 113, 122124] For simplicity, it may be assumed that both c atalyst and microbes are spherical particles (even though E. coli is rod shaped). It is likely that given the relative size of a bacterium to an individual TiO2 particle, that the system may best be described as a hard spherical particle interacting with a n ion penetrable plate. However, for generality, both particles will be considered spheres ( Figure 13). Taguchi et al [125] calculated the potential energy for the interaction between a sphere covered with an ionpenetrable membrane and a solid spherical particle. Many other cases can be found in the literature which describes specific interactions [126129] particularly the interaction of a spherical particle covered with an ion penetrable layer and a flat solid plate [129] The latter may be applicable to thin film photocatalysis systems. Figure 13: Proposed model for the interaction between a bacterium and a catalyst particle of radii a1 and a2 respectively, separated by X between thei r surfaces

PAGE 65

41 The total potential energy of two spherical particles is given by the sum of their van der Waals and electrostatic interaction energies [125, 127] ( ) = ( ) + ( ) ( 34) Consider two spheres of radii a1 and a2 separated at a distance X ( Figure 13). The potential energy for van der Waal interaction between the two particles is given as, ( ) = + 6 ( 35) where A is the Hammaker constant. The pot ential energy of double layer interaction between the two spheres is ( ) = 2 + ( ) ( 36) where Vpl( x ) is the potential energy of the electrostatic interactions per unit area between two plates at separat ion x During the interaction of the double layers, two cases are introduced for the solid particle [125, 127, 129] ; (1) constan t surface potential; and (2) constant surface charge. The potential inside the organism may be assumed to remain constant at the Donnan potential. Terui et al [126] derive d Vpl for solid particles under assumptions (1) and (2) above interacting with an ionpenetrable particle, respectively, ( ) = 2 1 2 ( ) ( 37) ( ) = 2 + 1 2 ( ) ( 38)

PAGE 66

42 By substituting equations ( 37) and ( 38) into ( 36 ) the potential energy of double layer interaction for a bacterium with TiO2 particle under the constant potential assumption is, ( ) = 4 + 1 4 ( ) ( 39) and under the constant surface charge assumption is, ( ) = 4 + + 1 4 ( ) ( 40) These reactions are important as they define the potential energy of interaction between the suspended colloids. The net interaction energy gives an indication of the colloidal suspension. If the interaction is dominated by van der Waals, then overwhelming attractive forces can lead to irreversible coagulation. If the electrostatic forces dominate, then the particles should be stabilized.

PAGE 67

43 CHAPTER 5: REVIEW OF WA TER DISINFECTION MODELING 5.1 Introduction There have been few attempts to define specific models for photocatalytic disinfection, with most of the current applications based primarily on chemical disinfection. The modeling of water disinfection is importa nt to establish the process kinetics of a specific disinfectant with particular microorganisms. In general, w ater disinfection modeling began as a purely empirical science based on the principles expressed in Chick s law [130] Chick observed that under certain conditions, the inactivation kinetics of microorganisms closely mirrored chemical reactions. Therefore, the fundamental l aws governing chemical reaction kinetics were appl ied to reactions involving microorganisms and a chemical disinfectant. F or a constant disinfection concentration, Chick concluded that the rate of disinfection is proportional to the concentra tion of microorganisms, thus: = ( 41 ) w here is the rate of disinfection given as the number of microbes per volume per unit time, N is the concentration of organisms (cells per unit volume), while k is a rate constant which var ies with the nature and concentration of the disinfectant. In a simple batch reactor, the solution of Equation ( 41) is an exponential decay curve, where N0 i s the initial count of bacteria

PAGE 68

44 = exp ( ) ( 42) Chicks model is a very simplistic formulation, but it has found extensive application where chemical disinfectants such as chlorine, ozone, hydrogen peroxide and chloramines are used [131] In ad dition to its simplicity, it is restricted to first order kinetics, which is just one, and very often, a seldom case in practical disinfection [132] Even though it is based on homogeneous reactions some researchers have applied Chicks formulation to calculate disinfection rate constants for photocatalytic inactivation of viruses and coliform bacteria [27] D isinfection models can be classified into two broad groups, empirical and mechanistic. Empirical models are mathematical expressions aimed at replicating the obse rved behavior of inactivation curves. Such curves can take a variety of shapes as shown in Figure 14. The combination of a number of factors may be responsible for producing each curve, but empirical models are not concerned with the underlying mechanisms. They are applied in areas where the kinetics of a disinfectant is well established. On the other ha nd, in the mechanistic approach a specific inactivation mechanism is first defined and then the model is develop ed The se models tend to be more robust than empirical models, which often cannot be extended beyond the data with which they are calibrated. Mechanistic models can be more flexible and allow the incorporation of many variables. However, microbial inactivation is extremely complex and depends on a wide range of defined and undefined variables [133] This means that even mechanistic models are simplifications and often require empirical approaches to complete them.

PAGE 69

45 Figure 14: Typical bacterial inactivation curves: (a) lag survival followed by exponential decay; (b) sigmoidal; (c) exponential (log linear ); and (d) concave downward 5.2 Empirical models 5.2.1 Chick Watson m odel Watson [134] found that under first order kinetics the relationship between the concentration of a chemical disinfectant and the time of exposure was a constant that produced a specific level of inactivation. Thus, = = ( 43)

PAGE 70

46 In the above equation, d is the coefficient of dilution and c is the disinfectant concentration. This led to the development of CT values, which allow practitioners to calculate how much disinfectant is required to adequately disinfect water, given certain microorganisms and under specified conditions [135] Assuming no disinfectant demand (i.e., c and n are constants), then t he ChickWatson model for a batch system is given by: = exp ( ) ( 44) Rincn et al [136] demonstrated that the model can sometimes fit observed data for photocatalytic inactivation. The inherent assumption is that the disinfectant concentration during photocatalysis is constant and inactivation is first order. In this case Equation ( 44 ) is reduced to, = exp ( ) ( 45) However, it is difficult to make farreaching conclusions and compare different studies based on the ChickWatson model especially when the studi es are conducted under dissimilar conditions 5.2.2 Delayed Chick Watson m odel The delayed ChickWatson model [74] is a modification in which a time lag parameter () is introduced to approximate an initial lag phase in the disinfection

PAGE 71

47 process ( Figure 14a ) For < the pseudofirst order loss of viability is replicated. The model ma y be represented as shown below, = 1 for < exp ( for > ( 46) The delayed Chick Watson model has been used by researchers to estimate CT values for the photocatalytic inactivation of E. coli [68] and Cryptosporidium [30] The hydroxyl radical was assumed to be the dominant disinfectant in these r eactions. 5.2.3 Hom m odel The Hom model [137] present ed a generalized differential equation for the time concentration relationships for the ef fect of a disinfectant on microbes. The expression is given as = ( 47) In the case where the reaction is zero order with respect to time and disinfectant concentration, it reduces to the first order relationship of Chicks l aw. Under condition where m = 0 and d Watson model. However, in the case where m 0 and n and = then the following express ion may be derived, = exp ( 48)

PAGE 72

48 The Hom model is useful for fitting disinfection curves with either an initial lag (when m > 1) or trailing curve (when m < 1). It cannot replicate both conditions simultaneously. 5.2.4 Kinetic p ower l aw m odels K inetic power law models do not make assumptions about the reaction rate order with respect to microbial concentration. The general form is = ( 49 ) The integration of E quation ( 49) gives the following for the survival ratio of organisms: = 1 1 1 + ( + 1 ) ( 50) Similar to the Hom model, E quation ( 50) can fit observed data displaying shoulders ( y < 1) or tailing off behavior ( y > 1). Chang et al [138] used a kinetic power law model and reported a reaction order of x = 1.06 for the inactivation of E. coli with TiO2. They also found that the disinfection rate was proportional to the square root of TiO2 concentration and proportional to incide nt light intensity within a range of 1801660 E s1m2. 5.2.5 Probabilistic models A n alternative modeling approach to disinfection used extensively in food microbiology includes the use of probability functions to determine the distribution of inactivati on times for a population of organisms exposed to a disinfectant [139] The

PAGE 73

49 approach is to consider each cell with a specific sensitivity to a certain level of disinfectant exposure. The survival of an organism during a certain exposure can be described as either alive ( = 1 ) or dead ( = 0 ) This may be written as [140] < = 1 ( 51) = 0 ( 52) where is the characteristic lethal exposure dose for the particular organism. The survival of this organism is essentially a step function and can be approximated by a sigmoid decay function [140143] = 1 1 + exp / ( 53) where is the inactivation rate around the inflection point. For the total population of organisms, the survival curve is given by, ( ) = ( ) ( 54) where is the fraction of the popu lation with a critical exposure of , such that = 1 Like empirical models, p robability based models are not directly concerned with specific reaction kinetics. Instead, it is only important to define the probability distribution of the population s sensitivity to certain levels of exposure. Peleg and Shetty [140] and van Boekel [144] used the Weibull distribution function to describe microbial populat ion sensitivity because it is a flexible function able to account for symmetric and asymmetric distributions. The Weibull probability density function is given as

PAGE 74

50 d d = exp ( 55) where () is the f raction of organisms having a critical exposure of Equation ( 55 ) can be algebraically transformed into the explicit function ( ) = 1 [ ln ( 1 ) ] / ( 56) The survival curve ( ) of the entire population, is obtained by integrating the curves of all the individual organisms, that is, ( ) = 1 { 1 + exp [ ( ( ) ) / ] } d ( 57) 5.3 Mechanistic models 5.3.1 Series event m odel The series event model can be represented by Equation ( 58) The inactivation process is modeled as a progression of discrete damage levels. The organism is assumed to be inactivated at a threshold level of damage [145, 146] ( 58) Each step is characterized by first order kinetics with respect to a constant concentration of chemical disinfectant Each damage level Di has a kinetic constant ki and n is the threshold level of damage The concentration of the disinfectant is assumed constant, so that ki is really a pseudokinetic const ant which can be represented by The disappearance of organisms at damage level D0 is given as

PAGE 75

51 = ( 59) and for level D1 the expression is = ( 60) where and are the concentrations of the organisms at the two damage levels respectively. The total number of surviving organisms is therefore the summation of all organisms below the threshold damage level, i.e., up to Dn 1. The main limitation s of this model are: (1) it requires a large number of damage levels to a ccurately describe inactivation and (2) it is not flexible for analyzing disinfection data since it can only be used to analyze concave curves. In addition, it is unlikely that the underlying chemical r eactions which lead to inactivation would proceed in the very same manner or would have the same effect in every cell. However, by assuming that the kinetic constant is the same at each level, the following generalized expression can be derived for the ser ies event model = + 1 + ( ) ( 61)

PAGE 76

52 5.3.2 Multi target m odel The multi target model is similar to the series event model but instead of damage levels, it assumes each organism contains a finite number of discrete critical targets ( nc) each of which must be attacked for full inactivation of the organism. When derived for batch reactor conditions, the multi target model takes the following form = 1 ( 1 ) ( 62) All the targets are assumed to be equivalent and the damage is randomly distributed among the targets. The probability of inactivating a specific target is given as ( 1 ) As a target is destroyed, the probability of hitting the remaining targets is reduced. 5.3.3 Haas model Haa s [147] developed a model which was applied for the inactivation of viruses by chlorine. However, the model has general applicability. The model was formulated on chemical reaction principles and assumes the existence of an intermediary organism disinfectant complex. + ( 63) With a constant disinfection concentration and first order assumption with regard to cell concentration, the survival of organisms has a Monodtype expression given by,

PAGE 77

53 = + + exp [ ( + ) ] 1 ( + ) ( 64) where and are empirical constants. 5.3.4 Marug n model A mechanistic model was presented by Marug n et al [21] to describe photocatalytic disinfection based on Langmuir type interactions between the microbes a nd catalyst particles. In this model, organisms are assumed to be undamaged, damaged and inactivated. The model takes the form of two different equations which are solved numerically for the adsorption interaction, inactivation, and reaction order constants. These are given as = 1 + + ( 65) = 1 + + ( 66) where and are the concentrations of undamaged and damaged cells, the sum of which gives the total cells surviving the disinfection process; and are the pseudo Langmuir parameters for adsorption and reaction rate respectively The main challenge of this model is that the application of Langmuir type interactions may not be appropriate to describe colloids, especially those as large as TiO2 part icles and microbes [148]

PAGE 78

54 With the exception of the Marugn model, no other mode l has been developed around the mechanisms of photocatalytic disinfection. While a straightforward approach to modeling may be desirable, simplistic formulations tend to neglect many important factors that influence the process. For example, it is impossib le to deduce the influence of catalyst concentration and light intensity from the foregoing models. Therefore, a comprehensive model is needed and should consider the most important mechanisms of the process. It would appear that microbe catalyst particle interactions should be an integral part of such a model, as well as light absorption and scattering, OH radical generation, and inhibition processes.

PAGE 79

55 CHAPTER 6: CONCEPTUAL MODEL FOR PHOTOCATALYSIS 6.1 Introduction In this chapter, a theoretical model for photocatalytic disinfection is presented taking into account the factors influencing bacterial and catalyst particle interaction. The main goal is to derive the reaction rate parameters and show how they can be measured fr om experiment. A quantitative analysis of colloidal adsorption and the subsequent chemical reactions of photocatalytic disinfection is important to the overall process kinetics. P revious attempts to apply models developed for molecular adsorption phenomena and reaction kinetics have proven to be inadequate, because colloidal adsorption proceeds via more complex pathways. In addition, reactions confined to the interface are influenced by the properties of the micro environment of the double layer. As indicat ed in Chapter 4, double layer interactions have considerable influence on the absorption process. The adsorption of colloids is a kinetic process that involves diffusion across the double layer, charge readjustment, and ion exchange processes, each with a characteristic time constant. Due to the fundamental differences between these processes and molecular dynamics, colloidal interaction cannot always be treated with classical statistical mechanic theories [148]

PAGE 80

56 6.2 Theoretical model formulation Consider a reaction suspension containing catalyst particles and bacterial cells. The cat alyst is assumed to be Degussa P25 TiO2 with an average particle diameter of 25 nm. On the other hand, the bacterial cells are much larger having a length of 1000 nm and diameter of 500 nm. Due to the relative size relationship, it is expected that multiple catalyst particles will adsorb to a cell. The electrostatic surface potential of the catalyst is defined by E quation ( 20 ) In like manner, the surface potential profile of the cells is defined by E quations ( 32) and ( 33) Under the pH conditions of interest ( 6 to 8, i.e., mostly neutral) and low electrolyte concentration the TiO2 surface is dominated by noncharged surface hydroxylated species, while the cell surface is mostly negative. The potential energy of interaction between the particles can be described mathematically according to E quations ( 34) to ( 40 ) However, it is easy to see that under the given conditions TiO2 particles would not experience significant repulsion from the bacterial surface because the particles are close to the point of zero charge. Therefore, adsorption of TiO2 to the bacterial cells will mostly be governed by short range van der Waal hydrophilic and hydrophobic forces. For simplicity, it is assumed that the bacterial cell can be repre sented as a sphere of diameter 1000 nm Therefore, imagine a situation where the small spherical catalyst particles surround the much larger bacterial cell as shown in Figure 15. Howeve r, it should be noted that catalyst particles in suspension can agglomerate to sizes comparable with cells [149] Since the repulsive forces are low, t he catalyst particles are able to approach the cell at very close separation distances ( possibly on the order of angstroms).

PAGE 81

57 In some cases, specific bonding may occur with bacterial surface appendages and polymers Figure 15: Surface coverage of catalyst particles on bacterial cell

PAGE 82

58 Hence, with time, TiO2 particles are immobilized at the cell surface. Under illuminated conditions, free radicals, mostly hydroxyl radical s are formed on the surface of the TiO2 and begin to react with bacterial surface sites. The reaction produces byproducts which diffuse away from the interface towards the bulk, but in the process they also react with radicals within the interface. With sufficient time, the cell would have expe rienced significant radical attack which eventually results in the inactivation of the bacterium. 6.3 Adsorption kinetics of catalysts and cells It is important to analyze the amount of TiO2 particles reaching the bacterial surface, since only these partic les are really involved in the photocatalytic process. The analysis would also provide insight into the expected dependence of the process on catalyst concentration. The transport of catalyst particles from the bulk solution to the bacterial surface can be described by the general continuity equation, n t + j = s ( 67) where np is the number concentration of catalyst particles and, t is time, j is a vector function describing the flows (flux) of np, and s is the sink function desc ribing, for example, bulk aggregation of the particles. The flux function involves particle diffusion and convection functions and may be defined as j = D n + U n ( 68) where D is the particle diffusivity tensor and U is the particle translation velocity vector. The terms described in E quations ( 67) and ( 68 ) can be determined by considering the specific particle particle interactions as presented in the C hapter 4.

PAGE 83

59 However, if the system shown previously in Figure 13 is considered in which catalyst particles are approaching the bacterial surface in a dilute colloidal suspension (i.e., n 1012 mL1) then the initial adsorption flux can be considered indep endent of the concentration of particles at the interface [148] The particle con centration varies only along coordinate axis indicated by X i.e., perpendicular to the bacterial surface. Assuming that there is no bulk aggregation of particles, E quation ( 67) may then be adopted in a one dimensional form as n t D n x + ( x ) n x = 0 ( 69) where D is the diffusion coefficient in the bulk and (x ) is the fluid velocity component directly perpendicular to the interface. If it is assumed that there is a primary minimum distance x at the interface where particles approach and are irreversibly adsorbed [102] then the boundary condition at the bacterial interface is given as n = 0 x = x ( 70) and away from the surface n n ( 71) where n is the concentration of particles away from the surface (i.e., in the bulk solution). After applying the boundary conditions, the uniform flux of particles towards the bacterial surface can be obtained as [148] j = D n x = D x n ( 72)

PAGE 84

60 where x is the thickness of the organisms diffusive boundary layer which for small organisms is of a similar magnitude with the characteristic length, a1 in t his case, the organisms radius 6.3.1 Adsorption in the absence of mechanical mixing It is not uncommon during experiments to have a standing suspension of catalyst and bacteria in which the colloids are neutrally buoyant The one dimensional transport equation for the condition in which ( x ) = 0 is given as [148] : n t D 1 r r r n r = 0 ( 73) where r = + + (see Figure 13) and D= D+ D is the relative diffusion coefficient ( D is the diffusion coefficient of the bacteria and D is the catalyst particle diffusion coefficient ; w hen the bacterial diffusion can be neglected). After applying the same boundary conditions a s before, t he uniform adsorption flux of particles to wards the bacterial surface under these conditions is given by [148] ; j ( t ) = D n 1 + 1 1 + a ( 74) where a= / and the dimensionless parameter = t /t. Here t= /D and is time required for the catalyst particle to get across the organisms diffusive boundary layer Therefore, the first term in the parentheses describe s the transient adsorption flux which bec omes negligible when 1 (that is, when t t). It is then clear to see that a constant flux is achieved for times exceeding the relaxation time, hence

PAGE 85

61 j = D n 1 1 + a ( 75) The relaxation time for a catalyst particle with D = 1012 m2 s1 diffusing across a layer of 500 nm thickness would be 0.25 sec which is a negligible time compared to the exposure time required for disinfection (on the order of minutes). 6.3.2 Adsorption in the presence of mechanical mixing Mechanical mix ing of the suspension introduces hydrodynamic shearing forces which maintain s suspension uniformity, but reduce s mass transfer for colloids. The quantitative analysis for the effects of hydrodynamic forces can be complicated, but approximations are ava ilable for simplified scenarios, including colloids in uniform flow in the absence of electrostatic forces The flux of spherical particles towards a spherical surface can be approximated by [148] j = 0 89 D / / n ( 76) where is the velocity of the fluid flow in the bulk phase. 6.4 Surface coverage of catalyst on bacteria 6.4.1 Surface coverage with low catalyst concentration The dimensionless surface coverage is denoted by and is the ratio of the area covered by particles to the total surface area of the collector (in this case the bacter ial surface). Mathematically, this may be expressed as, = n S ( 77)

PAGE 86

62 where ns is the number of particles with diameter collected on an element of area S If the elemental area is defined by vector rs, then the rate of change of surface coverage with time is [148] d dt = n ( r t ) ( 78) where (r, t ) is the normalized flux given by j (r, t )/n. By integratin g equation ( 78 ) the expression for (t ) is obtained as ( t ) = + n ( r ) t ( 79 ) where is the surface concentration of particles adsorbed during the transient conditions and is the normalized stationary adsorption flux previously defined. Equation ( 79 ) is only valid when the initial surface concentration is low so that already adsorbed particles do not have a significant influence (blocking) on the adsorption of new particles. This condition is true when 1 and can be determined from [148] = n ( 80) For a suspension of spheres not subject to mechanical agitation, can be approximated as, = n ( 81) Similarly, for spherical particles in a uniform flow = 0 55 / n / D / ( 82) 6.4.2 Surface coverage with high catalyst concentrations The kinetics of adsorption differs for systems with high colloid concentrations [148, 150] Catalyst particle s already adsorbed at the surface of the bacteria essentially preclude or block other particles from ads orbing within an exclusion zone. Therefore, the

PAGE 87

63 time evolution of the surface coverage is affected by existing coverage. A number of other models can be employed to model these systems (see, for example, reference [148] ). One of the simplest, but powerful approach is t he random se quential adsorption (RSA) model [151] In an RSA simulation, particles are randomly placed at the surface at a constant rate. Once the particle is placed, it is permanently affixed to the surface. Particles are not allowed to overlap, so a surface saturation is eventually reached when there are no more available spaces to fit particles. With this model, the surface is never completely covered. Even though spaces remain, they are not large enough to allow the positio ning of other particle s Hence, the saturation level is commonly referred to as the jamming limit and has a value of 54.7% for monodispersed spheres when only steric effects are considered [151] RSA models have been developed to incorporate short ran ge interactions between particles [150] Even though these assumptions are straightforward, the RSA configuration for high surface concentration, especially in three dimensions can usually only be predicted by numerical simulation [150, 152] However, t he kinetic curves describing the dependence of surface coverage on the adsorption time have been extensively calculated for hard and soft spheres by other authors under many different scenarios including no mixing conditions, electrostatic interactions, and hydrodynamic flows [148, 152, 153] Adamczyk et al [148] provide approximations which can be used in place of complex numerical simulations. T he RSA derived expression for the time evolution of surface coverage can be approximated by, ( ) = ( 1 + H ) 1 0 432 1 j n t ( 83)

PAGE 88

64 where H is a dimensionless parameter that defines the effective interaction range and depends on the energy of interaction and the double layer thickness as indicated by 1. H* may be approximated fro m H = L ln L ln 1 + 1 2 L ln ( 84) where L is the dimensionless double layer thickness give by and is the dimensionless interaction energy [148] For colloidal particles affected by hydrodynamic shear forces, the surface coverage can be approximated by, ( t ) = 1 1 exp ( 85) where is given as = 1 [ 4 ( 1 + H ) + C G ] ( 86) and G = G D 1 ( 87) where G is the shear rate at a given point on the interface and Ch is a dimensionless

PAGE 89

65 6.5 Kinetics of hydroxyl radicals at interface 6.5.1 Generation rate During the illumination of TiO2 particles hydroxyl radicals are produced at the catalyst water interface according to equations ( 3) and ( 5) The generation of the radicals is central to th e overall photocatalytic process At steady state conditions, it is the difference between the rate of light absorption and the recombination rate. As can be imagined, the latter process would be nearly impractical to measure in a real system. T he rate of light absorption is more amenable to experime ntation, but intense light scattering effects still makes this a difficult task However, the incident photon flux I in a solution can be determined by use of actinometry [154157] and the absorbed flux I can be estimated for a sample by determining its integrated absorption fra ction F from spectrophotometric methods Hence, I = I F ( 88) Fs has been previously determined for a range of TiO2 concentrations [158] The chart in Figure 16 has been re constructed based on interpolation and extrapolation of the literature data. Once the rate of adsorbed photon flux is determined, the rate of OH radical gener ation can be estimated by [159] G = I ( 89 ) where is the quantum yield of radical generation. The rate of generation of OH radicals and quantum yields for TiO2 in chemical photocatalytic reaction systems w ere determined by Sun and Bolton [158] according to the method described above. The radical generation rate is a function of catalyst concentration, the physical and chemical

PAGE 90

66 properties of the catalyst, light intensity, and dissolved oxygen concentration. Also important to note is that the addition of hydrogen peroxide has a positive effect on the generation rate [63, 158, 160162] Figure 16: Plot of integrated absorption fraction Fs for TiO2 concentration 6.5.2 Nature of OH radicals at the bacterial membrane In general, there are two theories concerning the nature of radicals at the catalyst surface; (1) radicals remain surface bound to the catalyst during reaction with adsorbed species [19, 48, 64] ; and (2) radicals diffuse away from the surface to react with compounds in solution or on the catalyst surface [64, 163165] It would be very difficult to distinguish between these two possibilities in the overall kinetics of the process. Howe ver, i n the latter case, it is recognized that hydroxyl radicals, in particular are diffusion limited owing to their high reactivity. Depending on the concentration of oxidizable species, hydroxyl radicals have been found to diffuse up to a distance of 10 0.000 0.005 0.010 0.015 0.020 0.025 0.0 0.1 0.2 0.3 0.4 0.5Fs TiO 2 concentration (g/L)

PAGE 91

67 nm away from the site of generation [166, 167] Therefore, it is possible that radicals can diffuse into a bacterial membrane during very close approach with a catalyst surface. The diffusion coefficient of hydroxyl radicals in water has been estimated to be on the order of 109 m2s1 at 25C [168 170] If the nearest substrate is 10 100 nm away from the site of generation, it would take a radical much less than a fraction of a second to move across this range of distance. However, a number of factors may hinder diffusion near the vicinity of the cell membrane, including electrolyte ions hydrophobic zones and the solvation shell around the radical [168, 171, 172] 6.6 M icrobial survival The model presented by Haas [147] may be adopted for the reaction of hydroxyl radical s with cells in a simple bimolecular reaction [ OH ] + [ cell ] [ cell ] + OH ( 90) where the subscripts l and d denote live and dead cells respectively and is the observed rate constant for disinfection. The overall disinfection reaction rate for this bimolecular reaction is given as = [ OH ] [ cell ] ( 91 ) where and are the reaction orders related to radicals and cells respectively. The concentration units for hydroxyl radicals are moles per liter, but for the cell they are given as cell number density (cells per liter). The observed disinfection rate has contributions from (1) the diffusioncontrolled rate constant at which the cell radical complex [cell OH ] is formed,

PAGE 92

68 [ OH ] + [ cell ] [ cell ( OH ) ] ( 92) (2) the rate constant for dissociation (or radical quenching and repair) and (3) the rate constant at which the cell is eventually inactivated after being exposed to the radical. [ cell ( OH ) ] [ cell ] + OH ( 93) It can be shown that the observed disinfection rate has the form = + ( 94) If the inactivation rate constant is much faster than the repair/radical quenching, that is , then as radicals encounter the cell it is rapidly inactivated without time for repair or quenching. In this case, the observed rate is equal to the diffusion rate constant ( = ) and the reaction depends on how fast radicals c an encounter the cells. However, if the inactivation rate is much slower than the repair and quenching mechanisms then the observed rate is given by = = [ ( ) ] ( 95) where [ ( )] is the equilibrium constant for the formation of the cell radical complex. 6.7 Kinetics of b yproduct evolution The effect of free radicals on cellular molecules has long been reported (s ee for example [173, 174] ) In particular, t he hydroxyl radical is very reactive and is capable of injuring virtually all biological macromolecules. Free radicals associated with the

PAGE 93

69 photocatalytic process can react with macromolecules on the bacterial surface, including proteins, polysaccharides, and lipids. Of these, lipids are kno wn to be the most prone to oxidative damage, particularly lipids with unsaturated fatty acids. Proteins are also very susceptible to radical oxidation. The extent of the damage to particular targets depends on a number of factors, including the concentrati on of the target, the reaction rate constants, the relative locations of the target and oxidant, the occurrence of secondary damaging events, occurrence of transfer reactions, and repair and scavenging reactions [175177] In addition, the oxidation of intracellular constituents can occur through the generation of secondary oxidants, such as lipid radicals, hydrogen peroxide and superoxide [178180] Superoxide and hydrogen peroxide can also produce hydroxyl radicals in the intracellular environment through the Fenton reaction involving free iron [181, 182] For E. coli most of the outer membrane is made up of phospholipids. In addition to their abundance, their ease of oxidizability makes this group of biomolecules prime targets for hydroxyl radical attack. Lipid peroxidation has been identified as a leading reac tion mechanism during photocatalysis [20, 178, 183185] The peroxidation of lipids involves three distinct steps: initiation, propagation and termination. Figure 17 illustrates these processes schematically. The initiation reaction occurs when O H abstract s an H atom from the unsaturated fatty acid forming a carboncentered radical. In the propagation reactions th e carbon centered radical reacts with oxygen and yields a peroxyl radical The peroxyl radical then abstracts an H atom from a second fatty acid forming a lipid hydroperoxide (LOOH) and leaving another carbon cent e red free radical [173, 174] The lipid hydr operoxide eventually degrades into malondialdehyde (MDA)

PAGE 94

70 and other unsaturated aldehydes. Termination occurs when two radicals react together forming neutral products ( Figure 17 ) The peroxidation of lipids can often result in damage to biomolecules at sites considerably distant from where the initial free radical reaction occurred [186] Lipid peroxidation can be monitored by assessing the rate of oxygen uptake or the production of byproducts including MDA and lipid hydroperoxides [187, 188] Figure 17: Schematic of lipid peroxidation Most of the byproducts are formed within the interface where the hydroxyl radicals react with the cell surface. Since byproducts can be considered molecular fragments of disinfection, they diffuse throughout the solution and absorb to the catalyst surface. For simplicity, it is assumed that adsorption kinetics can be described by the Langmuir model Hence, in the ab sence of other adsorbing molecules the rate of byproduct oxidation is given as

PAGE 95

71 = 1 + ( 96) where is the reaction rate constant, is the Langmuir adsorption rate constant and the concentration of all byproducts The OH radical is known to react very efficiently with biomolecules at a diffusioncontrolled rate with a reaction rate constant on the order of 109 M1 s1 in homogeneous solutions [189191] 6.8 Adsorption and inhibition kinetics of inorganic ions Inorganic electrolyte ions, particularly anions such as chloride (Cl), sulfate (SO ), phosphate (HPO ) bicarbonate (HCO ), and nitrate (NO ), are known to adsorb to the surface of TiO2 [100, 101] and inhibit the photocatalytic process [192196] However, there has never been any model to quantify the effect of these ions on photocatalytic disinfection efficiency. To include these effects in the current model, the formation of surface comple xes is analyzed. The adsorption of inorganic ions to the surface of TiO2 can be described in terms of ligand exchange reactions with surface hydroxyl groups. This process is similar to complex formation in homogeneous solution, but the apparent equilibrium constants are adjusted to account for the electrostatic effects of the double layer [197] The adsorption kinetics is governed by the properties of the adsorbing ion and the properties of the surface. The primary parameters for a quantitative description of ion adsorption are the acidity constants ( ) of the ionic species and the surface hydroxyl groups, and the constants for the formation of the complexes ( ) With these constants the surface specia tion can be computed as a function of pH and concentration of ionic species

PAGE 96

72 However, for a given pH and low surface coverage, anion adsorption on metal oxide surfaces can be described by the Langmuir equation [101, 198] = [ Ti An ] [ Ti OH ] [ C An ] H + ( 97) where [TiAn ] is the concentration of an adsorbed anion, TiOH is the activity of all protonated surface moieties that can be displaced by the anion, and [ C] is the concentration of the anion in solution. Constants for the formation of complexes by common anions on the surface of TiO2 have been reported in the literature [50, 197, 199] and are given in Table 2 In the absence of other absorbing molecules, Equation ( 97 ) can be rearranged to give the Langmuir equation. = , [ H ] 1 + , [ H ] ( 98) where is the surface coverage of the i th anion species, is the concentration of the specific anion species in solution Table 2: Adsorption e quilibrium constants for some common anions on the surface of TiO2 Anion Equilibrium constant M 1 Cl 110 5 [197] CO 6 10 4 [200] SO 2 10 8 [201] H PO 810 6 [202]

PAGE 97

73 In homogenous solutions inorganic ions react with hydroxyl radical s at diffusion controlled rates. The rate constants and mechanisms for these interactions have been reported [203, 204] However, since the rate of generation (and by extension, the concentration ) of hydroxyl radicals in TiO2 suspensions i s significantly lower than the homogeneous diffusioncontrolled rates, the overall reaction between the ions and the radicals is likely to be limited by the generation rate of radicals T he concentration of radicals during photocatalysis ( 1 108M) is usua lly much lower than the electrolyte concentration [205] If it assumed that the generation of radicals is uniformed across the entire catalyst surface, then t he rate of the inhibition reactions is directly proportional to the extent of coverage The latter may be determined from the specific adsorption isotherms of the various ions in solution [194, 195] Therefore, it is only important to determine the surface coverage of ions to understand the extent of inhibition on the disinfection process. Guillard et al [195] found that electrolyte ions form a salt layer at the surface of TiO2 which prevented the adsorption of organic substrate. In the same way, inorganic ions, due to their molecular size, can approach the catalyst surface and specifically adsorb in a much more efficient way tha n large micron s ized bacterial cells. However, at low salt concentration there is low screening of the cells and there are enough available hydroxyl sites to generate radicals. Under these conditions, the efficiency of disinfection is optimal. Conversely, a t higher conc en trations the opposite is true that is, most of the radicals are consumed by inorganic ions and the cells are screened to a larger extent. Therefore, it can be argued that disinfection must occur as a result of the residual

PAGE 98

74 hydroxyl radicals which are abl e to escape the catalyst surface or interact directly through surface to surface contact. The residual hydroxyl radical generation is the differen ce between the photogeneration rate of radicals and the rate of inhibition. As before in the absence of other absorbing molecules, t he rate of inhibition or radical quenching can b e expressed as a factor of the OH generation rate G as = G = , [ H ] 1 + , [ H ] G ( 99) As 1 all active sites for hydroxyl radical generation are blocked, then the rate of disinfection is at its lowest. is the total surface coverage found by summing the individual coverage of all ionic species. 6.9 M odel for overall inactivation kine tics Now that the important mechanisms for the photocatalytic disinfection process have been defined, the kinetics for the overall process may be determined by performing mass balances for specific variables. This analysis 6.9.1 Mass balance of live cells The survival of cells is given by E quation ( 91 ) The differential form of the equation can be written as d [ cell ] dt = [ OH ] [ cell ] ( 100) The disinfection reaction is peculiar in that it involves the reaction of molecules (usually given in mol L1) and cells (given in CFU L1). Therefore, it is important to recognize that Equation ( 91 ) can be expressed in t w o ways with respect to the reactants; (1) the rate of

PAGE 99

75 disinfection (CFU L1 s1) as given in Equation ( 100 ) where t he units of the disinfection rate constant are Mn s1, and (2) the rate of consumption of hydroxyl radicals given in concentration per time (M s1) To reconcile this irregularity, Equation ( 100) can also be expressed in terms of radical consumption, d [ OH ] dt = OH [ OH ] [ cell ] ( 101) where OH is the reaction rate constant given in units of Ln Mn 1 CFU1 s1. 6.9.2 Mass balance of byproducts In order to account for the accumulation of byproduct, E quation ( 92 ) is rewritten as [ OH ] + [ cell ] [ cell ] + [ BP ] ( 102 ) One of the inherent difficulties of E quation ( 102 ) is that one radical can set off a chain of reactions resulting in numerous byproducts being formed. However, if it is assumed that most of the byproducts result from oxidation of lipids, then the reaction kinetics in the membrane would be very similar to OH radicals reacting with lipids in solution (i.e., outside of a bilayer formation) [206 209] Therefore, if = t he accumulation of by products is given by d dt = OH [ OH ] [ cell ] 1 + + , [ H ] ( 103) 6.9.3 Mass balance of OH radicals The mass balance for OH radicals in the interface between a catalyst and the cell surface is given as

PAGE 100

76 d [ O H ] dt = G ( 1 ) OH [ OH ] [ cell ] ( 104) w here and are the surface coverage of anions and byproducts, respectively. It is customary for researchers to assume that the concentration of OH radicals is constant during the reaction. However, that assumption is not applied here T ogether, Equations ( 100) through ( 104) represent the overall kinetics of the photocatalytic disinfection process.

PAGE 101

77 CHAPTER 7: EXPERIMENTAL DESIGN AND PROTOCOLS 7.1 Selection of experimental factors The rate of inactivation of a ta rget organism is the most important design variable for disinfection systems. For photocatalysis, the rate of disinfection depends on the synergistic effect of catalyst concentration and light intensity, which directly influences the rate of generation of OH radicals in the reactor. If the concentration of radicals can be significantly increased, it is clear to see that the disinfection rate would also increase. This effect has previously been observed when hydrogen peroxide was added to disinfection experi ments with E. coli [16 0] and also in chemical photocatalysis studies [63, 161] Likewise, sink terms such as byproducts or compounds that exert a demand on the OH radicals reduce the overall rate of the reaction [160] The concentration of solution electrolytes (ionic strength) has also been studied and the effects can be explain ed based on the principles laid out in Chapters 4 and 6. Based on the analysis in the previous chapter s the most important operational variables to be tested were catalyst concentration and light intensity. The synergistic effect of these two factors will determine the most efficient combination of contact time and dose to employ. Light intensity was tested at 3 levels ; high, medium, and low. Each level corresponded to a spe cific light intensity value measured in Einstein per volume per time (E L1 s1) Catalyst concentration was tested across 4 levels ; 0.01, 0.10, 0.25, and 0.50 g L1.

PAGE 102

78 In an attempt to account for biological variation in the model one additional factor, f atty acid composition, was also selected. As previously discussed, unsaturated fatty acids have been identified as a major target during photocatalytic disinfection. By modifying the content of specific unsaturated fatty acids in the organisms membrane the effect of this variable could be investigated A factorial experimental design was employed to study the effect of the three independent variables on the disinfection rate and dose responsive behavior of the organism during photocatalysis. Experiments were conducted in triplicates. In addition to the response of microbial survival to various treatments the evolution of byproducts was also monitored for a subset of experiments. MDA is a common biomarker for lipid peroxidation and was used in this study to evaluate the kinetics of byproduct formation [187, 210, 211] Lipid hydroperoxides were also tested in some experiments. Both of these compounds are well known byproducts from membrane peroxidation resulting from a reaction with hydroxyl radicals [211214] The choice to focus on membrane fatty acids and the kinetics of byproduct evolution was validated by using model cellular membranes (liposomes) made from representative fractions of natural E. coli fatty acid s. 7.2 Method of data analysis 7.2.1 Statistical analysis To test for difference s among groups, a one way ANOVA was performed on survival data from 144 experiments, which included all factors at all levels. The null

PAGE 103

79 hypothesis for this test was that there were no differences among the groups. The Tukey test was used to compare groups and examine interaction effects of the factors. Different levels of interaction were examined including main effects, and 2 way and 3 way interactions. 7.2.2 Numerical analysis Equations ( 100) ( 103) and ( 104) were solved numerically by a Runge Kutta method in a MatLab algorithm and constrained to fit the survival data from the 144 experiments using a nonlinear least squares method (See Appendix A). From this procedure values for the disinfection rate constant and reaction order were obtained. 7.3 Microbiological methods 7.3.1 Preparation of E. coli culture Pure cultures of E. coli ( ATCC 25922 ) w ere grown aerobically in 100 mL of Luria broth at 37C on an incubator shaker (250 rpm) The growth kinetics of the organism was determined from experiment by monitoring the turbidity of the suspension with time (Figure 18) The turbidity was measured at 550 nm with a DR/2000 s pectrophotometer (HACH Company).

PAGE 104

80 Figure 18: E. coli growth curve fitted with a continuous logistic function 7.3.2 Cell h arvesting and enumeration Bacteria are known to modify their fatty acid content depending on the growth phase [105, 215217] Therefore, it was important to select a standard time during growth to harvest the organisms since fatty acid content was an independent variable in the experiments. E. coli cells were always harvested after 6 hours of growth from an actively growing media broth by centrifugation at 1380 g for 10 min in a 15mL tube. The cell pellet was washed and resuspended in sterile deionized water (resistivity >16 Mohm cm) This process was performed twice to ensure that most of the broth solution was removed. The turbidity of the suspension was determined as described before using visible light s pectrophotomet r y. A standard curve was developed to correlate turbidit y readings with cell density (CFU m L1) by performing serial dilutions to obtain between 30 and 300 CFU in 100 L on TSA plates The plates were incubated at 37 C and t he 0.0 0.2 0.4 0.6 0.8 1.0 0 2 4 6optical density ( )time (hr) lag phase exponential phase stationary phase

PAGE 105

81 viable cells which appeared after 24 hours were manually counted. The cell suspensio n was diluted to the required final concentration for all experiments. 7.3.3 Preparation and storage of growth media Luria broth was obtained from US Biological ( Swampscott MA) as a dry powder and prepared according to the manufacturers instructions. Th e powder (7.75 g) was added in 450 mL of deionized water while being heated and gently stirred until it was completely dissolved. The pH of the media was adjusted to 7.0 with 1 N NaOH, brought to 500 mL, and finally a utoclaved for 15 minutes at 121 C and 15 psi. The solution was cooled to room temperature before use and the remainder stored at 4C in the refrigerator. Liquid media was used within 14 days. Figure 19: Standard plot for the correlation of cell density and optical density R = 0.9976 0 0.2 0.4 0.6 0 1 2 3 4 5optical density ( )cell count 1011(CFU L1)

PAGE 106

82 7.3.4 Preparation and storage of agar plates Tryptic soy agar was obtained from MP Biomedicals (Solon, OH) as a dry powder and prepared according to the manufacturers instructions. The media was prepared with 40 g of dry powder to 1 L of deionized water. The solution was stirred and heated until it boiled, and then sterilized for 15 minutes at 121C and 15 psi in an autoclave. Sterile media was chilled to 55C before pouring into 100mm 15 mm sterile polystyrene Petri plates Agar plates not used immediately were stored at 4C in a refrigerator and used within 7 days. 7.4 Photocatalytic experiments 7.4.1 Reactor design and setup Experiments were conducted in 30mL borosilicate test tubes (15.35 mm diameter) which were placed in the center of a reactor holder surrounded by lamps ( Figure 20 ). The coefficient of transparency for a 10 mm thick borosilicate glass is within the range of 0.950.99 for wavelengths f rom 360500 nm [218] This provided a s uitable economic alternative to the commonly used, but expensive quartz vessels. The reactor holder was fabricated with different slots for the lamps. This was done so that the light intensity could be varied by adjusting the distance of the lamps to the r eactor vessel. Two lamps were always used and kept equidistant to the reactor Three positions were used to achieve the range of high to low intensity. Lamps were turned on at least 15 minutes prior to experiments to allow them to warm up for stable output The entire unit shown in Figure 20 was placed centered on a magnetic stir plate to continuously stir the solution with a ministir bar (12.7 cm long 0.64 cm dia.)

PAGE 107

83 7.4.2 Catalyst stock solution preparation and storage Degussa P25 TiO2 was used as the catalyst for all photocatalytic experiments. The formulation of this catalyst has been published extensively as containing 75% anatase and 25% rutile with an average s urface area of 50 m2 g1 (see for example [51, 219221] ). A stock solution of 10 g L1 was prepared by vigorously mixing the white powdered catalyst with deionized water, autoclaving and storing the suspension at room temperature in a sealed container. Figure 20: R eactor apparatus

PAGE 108

84 7.4.3 Light source Light for the photocatalytic experiments was provided by 9W UVA lamps ( model PL9W/08 ) from the Phil l ips Lighting company ( Figure 21) They have overall dimensions of 167 mm 2 8 mm. The lamps have a spectral maximum of 365 nm ( Figure 22) and the UVA radiation output is 1.7 W. Figure 21: Schematic of UVA fluorescent lamp used in experiments

PAGE 109

85 Figure 22: Spectral power distribution of PL S 9W/08 lamp (source: manufacturer) 7.4.4 Light intensity measurements Since there were two reactors, light intensity measurements were done for two pairs of lamps at three different positions on the reactor holder. The lamps were numbered 14 and the positions were number ed 13 from the closest to the farthest ( Figure 23 ). The incident light intensity in the reactor solution was determined by azoxybenzene a ctinometry [157] The quantum yield for azoxybenzene is about 0.02 across the UV region 200380 nm and is unreactive in the visible range. Azoxybenzene has a sharp absorption cut off near 380 nm and this, combined with the low quantum yield, means that solutions of azoxybenzene are convenien tly handled under ambient light.

PAGE 110

86 Figure 23: Lamp locations on reactor The procedure included irradiating 20 mL of 4.89 mM azoxybenzene solution in ethanol in the same borosilicate reactor vessel used for photocatalytic disinfecti on experiments. During irradiation, 2mL aliquots were sampled at oneminute intervals for 5 minutes. Two drops of potassium hydroxide solution in ethanol (0.10 N) w ere ad ded to convert the photoproduct ( hydroxyazobenzene ) to its anion form. The samples were photon dose and concentration is given as ln 1 = I t ( 105)

PAGE 111

87 w here is the initial concentration of azox ybenzene (mol L1), P is the concentration of the photoproduct (mol L1) at time t and I0 is the incident light intensity (E L1s1). The slope of the plot ln 1 versus t wa s used to determine I0 ( Figure 24 ) Figure 24: Typical plots used to determine incident light intensity by actinometry for pair wise combination of lamps 1 and 2 [(a) (c)], and 3 and 4 [(d) ( f )] The measurements were conducted periodically over the course of the study. The incident intensity for the pairs of lamps is given in Table 3. 7.4.5 Preparation of working reaction solutions All liquids and vessels, including PBS solution, deionized water, reaction tube, and stir bar were autoclaved for 15 minutes at 121C and 15 psi prior to being used in experiments. The reaction pH was monitored over a series of preliminary experiments and found to be stable during the course of the experiments. Therefore, only initial pH

PAGE 112

88 was recorded for final experiments. Three sets of control experiments were conducted; (1) solution of catalyst and microbes with no light (dark e xperiments); (2) irradiated solutions of microbes with no catalyst present (irradiated blank); and (3) non irradiated solution of microbes only (organism control). The final composition of the reaction solution was made up by adding the appropriate volumes of stock solutions together and then pouring the mixture into the reaction vessel ( Table 4 ). The final composition of electrolytes is shown in T able 5. Table 3: Incident light intensity in reactors according to lamp combinations Lamp s Position Incident intensity, I 0 (E L 1 s 1 ) 1 2 1 4.37 10 5 5.1910 6 1 2 2 2.4010 5 5.1910 6 1 2 3 1.3510 5 2.3010 6 3 4 1 4.8510 5 1.1810 6 3 -4 2 2.5910 5 2.0010 6 3 -4 3 1.5110 5 8.5310 9 7.4.6 Sampling and error analysis During the course of a typical photocatalytic experiment, samples were taken at specified time intervals using a pipette. The sample was serially diluted (Figure 25 ) and incubated as described in section 7.2.2 to determine the mic robial survival. The appropriate dilution level for each time interval was plated in triplicate.

PAGE 113

89 Table 4: Composition of working reaction solutions Working solution# 1 2 3 4 5 Deionized water (mL) 1 8 .60 1 8 .58 18.40 1 8 .10 17.60 10 g L -1 of P25 (mL) 0 .00 0.02 0.20 0.50 1.0 0 510 10 CFU L -1 cells (mL) 0.4 0 0.4 0 0.4 0 0.4 0 0.4 0 1PBS (mL) 1.0 0 1.0 0 1.0 0 1.0 0 1.0 0 pH measurement pH~ 7 3 pH~ 7 3 pH~ 7 3 pH~ 7 3 pH~ 7 3 Solution filled to 20 mL with deionized water and poured into reaction vessel Total volume (mL) 20 20 20 20 20 Final P25 conc. ( g L -1 ) 0 .00 0.01 0.10 0.25 0.5 0 Final cell count (CFU L -1 ) 1 10 9 110 9 110 9 110 9 110 9 Ionic strength (M ) 0.01 0.0 1 0.0 1 0.0 1 0.0 1 Table 5: Composition of electrolytes in final solution Constituent Concentration (mM) NaCl 6.85 KCl 0.14 Na 2 HPO 4 0.50 KH 2 PO 4 0.09 The standard deviation and standard error of viable counts for each sample were determined based on the Poisson distribution. The mean of triplicate experiments was calculated in the usual way and relative error was determined according E quation ( 106 ) = 1 + ( 106)

PAGE 114

90 where is the overall relative error and is the error contribution from technical sources including volume measurements and dilutions. For the Poisson distribution, the mean is equal to the variance Therefore, the standard deviation is given as the square root of the mean. Figure 25: Schematic of serial dilution of sample 7.5 Fatty acid modification and analysis Lipid modification of the E. coli cells was achieved by supplementing the Luria broth growth media with 32 M of palmitoleic (C16:1 n7) oleic (C18:1 n9) and linolenic (C18:3 n3) acids. The se are long chain unsaturated fatty acids with the number of carbons, double bonds, and double bond location indicated in parenthesis respectively.

PAGE 115

91 At least 20 mg of cells w as harvested from an actively gr owing culture at 6 hours and twice pelletized by centrifugation at 1380 g for 15 min in a 15 mL tube The cell pellets were wash ed and suspended in sterile deionized water between centrifugation. The samples were sent frozen to Microbial ID (Newark, DE ) to determine the fatty acid composition by fatty acid methyl ester (FAME) analysis. The general steps in a FAME analysis ( Table 6) include extraction of the fatty acid s by a procedure which consists of saponification in dilute sodium hydroxide/methanol solution, followed by derivatization with dilute hydrochloric acid/methanol solution to give respective methyl esters. The methyl esters are then extracted from the aqueo us phase by the use of an organic solvent and the resulting extract was analyzed by gas chromatography (GC). Table 6: Steps in FAME analysis Step Purpose Harvesting Removal of cells from culture media Saponification Lysis of cells to liberate fatty acids from cellular lipids Methylation Formation of fatty acid methyl esters (FAME) Extraction Transfer of FAMES from the aqueous phase to the organic phase Base wash Aqueous wash of the organic extract prior to GC analysis 7.6 Preparation and characterization of model cell membranes 7.6.1 Preparation of l ipid film The dominant phospholipids in the membrane of E. coli are phosphatidylethanolamine ( PE) and phosphatidylglycerol ( PG ). These natural lipids were obtained from Avanti Polar Lipids (Alabaster, AL) dissolved in chloroform at a

PAGE 116

92 concentration of 5 g L1 each. The lipids were mixed in equal molar proportions Higher ratios of PE to PG were initially used, but the stability of the liposomes was not consistent. The PE/PG solution was transferred to a clean and dry 100 mL round bottom flask and continuously rotated by hand in a water bath at 60C until the solve nt evaporated and a uniform thin lipid film was formed on the surface of the flask. A gentle stream of N2 gas was passed over the film to remove solvent vapor. The flask was left overnight in a chemical hood to allow complete evaporation of all the chlorof orm. 7.6.2 Lipid film hydration and extrusion The lipid films were hydrated with 5 mL of 1 PBS solution by continuously rotating the flask in the water bath maintained at 60oC until all the film was completely dissolved (smooth milky white appearance). At this stage of the process, the lipids are present as sheets of hydrated lamellar films. In order to transform the films to the characteristic cell membrane structure, the solution was forced through 0.8 m polycarbonate membrane. Th is size reduction step was performed using a mini extruder obtained from Avanti Polar Lipids (Alabaster, AL). It consisted of two 1 mL syringes inserted on opposite ends of a filter support assembly. The solution is passed from one syringe to the other across the filter. The ent ire assembly sits on a custom fit heating block. The extruder wa s maintained at 60oC and the solution was passed 12 times across the membrane

PAGE 117

93 7.6.3 Size distribution measurement Th e size distribution of the liposomes was determined by photon c orrelation s pectroscop y using a Malvern Zetasizer Nano series device. The liposome suspension was diluted with 1 PBS prior to measurement. 7.6.4 T ransmission electron microscop y A drop of the PE/PG solution was placed on a Formvar carbon film with 150 square mesh co pper grids and visually examined with a FEI Morgagni 268 TEM after staining with 0.50% uranyl acetate in water. The TEM was operated at 60kV and an Olympus Soft Imaging MegaView III camera was used to collect images. 7.7 Measurement and analysis of byproducts 7.7.1 MDA assay A thiobarbituric acid reactive species ( TBARS ) assay kit was obtained from Northwest Life Science Specialties ( Vancouver, WA ) and used to measure MDA in the samples. Aliquots of 250 L sample solution were added to a microcentrifuge vi al containing 10 L of butylated hydroxytoluene (BHT ) an antioxidant T he acid reagent (250 L) was added and the mixture was centrifuged at 11,000 g for 35 min and then for an additional 20 min to remove solids The supernatant was transferred to new vials and 250 L of thiobarbituric acid (TBA ) reagent was added. The mixture was vigorously shaken on a vortex for 5 counts and then incubated in a water bath at 60C for 1 hour. After incubation, the solution was centrifuged at 10 ,000 g for 3 minutes and absorbance of the supernatant was recorded from 400700 nm on an Ocean Optic USB2000

PAGE 118

94 spectrometer using the OOIBase 32 software and DH 2000BAL UVVIS light source. The spectrometer was calibrated with a standard mercury emission lamp according the manuf acturers instructions prior to measurement. 7.7.2 Derivative spectroscopy analysis Derivative spectroscopy analysis was performed on the absorbance spectra to negate the effects of nonlinear baselines and enhance the spectral signals. A smoothing function was first applied to the spectra according to the method by Savitzky and Golay [222] The second derivative was then selected and the absorbance evaluated at 511 nm The technique was pr ogrammed into a computer code to ensure that the same treatment was performed on all the spectra. 7.7.3 LOOH assay A l ipid hydroperoxide analysis kit was obtained from Northwest Life Science Specialties (Vancouver, WA). The met hod is based on the fact tha t a hydroperoxide present in solution oxidizes ferrous iron ( Fe2+) to ferric iron ( Fe3+) under acidic conditions [188, 223] The resulting ferric iron was detected using xylenol orange, which forms a Fe3+xylenol orange complex. The complex was measured on a spectrophotomet er at 560 nm. The manufacturers assay protocol was followed precisely, except for an additional final centrifugation step to remove solids in the samples.

PAGE 119

95 CHAPTER 8: RESULTS AND DISCUSSION 8.1 Fatty acid modification and analysis The fatty acid profile of the unmodified E. coli was a close match to published profiles of the organisms; see for example [105, 215, 224] The distribution of the main fatty acids is shown in Table 7 (see Appendix B f or full list). The predominant fatty acid was the saturated 16 carbon (palmitic acid). Palmitoleic (C16:1 n 7) and cisvaccenic (C18:1 n7) acids were present in equal proportions and accounted for most of the monounsaturated content. The total polyunsatur ated fatty acid content was below 0.5 0%. Organisms supplemented with oleic acid (C18:1 n9) had an enrichment of this fatty acid in their membrane, even though it was not detected in the control population. The enrichment of oleic acid was accompanied by a reduction in its positional isomer, cisvaccenic acid. The addition of linolenic acid (C18:1 n 3) had a pronounced effect on the fatty acid distribution. The presence of linolenic was not detected in the samples indicating that t he supplemental fatty acid was converted by the organisms to other less unsaturated fatty acids. There were significant changes particularly in the C18:1 group of fatty acids and the appearance of a small fraction of C18:2 in the organism.

PAGE 120

96 Table 7: Percent distribution of major fatty acids Fatty acids Unmodified cells1 Fatty acid supplement Lipid Vesicles2 C16:1 n 7 C18:1 n 9 C18:3 n 3 Saturated C14 8.5 7.0 7.6 7.4 1.8 C15 1.9 1.5 1.6 1.5 8.5 C16 34.8 35.0 31.9 32.6 28.8 C17 2.1 2.0 1.6 2.1 10.9 C18 0.6 0.4 0.3 1.0 0.0 Monounsaturated C16:1 n 7 12. 5 19.9 5 2 9.5 7.1 C18:1 n 7 12. 6 8.6 6.7 17.3 17.1 C18:1 n 9 0.0 0.0 22.2 2.8 4.5 Polyunsaturated C18: 2 n 6 0.4 0.4 0.0 2.7 0.0 Cyclopropane C17 11.1 11.4 6.3 9.0 14.5 C19 1.3 0.5 1.9 1.2 4.0 Total saturated 73.2 70.3 63.6 66.6 68.5 Total unsaturated 26.2 29.2 35.9 32.8 28.7 Uns aturated/ S aturated 0.4 0 .4 0.6 0.5 0.4 1E. coli cells grown in Luria broth and harvested at 6 hours Only major fatty acids are shown. Total fatty acids include all fatty acids detected in analysis. See supplemental information 2 Fatty acid spectra obtained from manufacturer 8.2 Factorial analysis : Main effects In order to make fair comparisons across all groups, the log survival at 20 minutes was selected as the response variable to perform the factorial analysis. This corresponded with the shortest experimental time. Survival data are usually distributed log normally and this was confirmed by conducting a probabi l it y plot as shown in Figure 26.

PAGE 121

97 Figure 26: Probability distribution of survival data for E. coli The main effects are illustrated in Figure 27. The mean of the log of survival is plotted on the vertical axis against the levels of each factor.

PAGE 122

98 Figure 27: The m ain effects pl ot s for (a) TiO2 concentration; (b) fatty acid modification; and (c) light intensity on mean survival data at 20 min 8.2.1 Light i ntensity In the study, l ight intensity was confirmed as the most significant effect on the disinfection process ( Figure 27c ) Figure 28 shows the variation in survival for the three different light intensity levels at the lowest TiO2 concentration. The trend is typical for other concentrations, except that the variation is greatest at concentration value shown. M any workers have found tha t the disinfection rate is usually proportional to the square root of light intensity at relatively high photon fluxes and linear at low flux [22, 23, 158, 160, 225, 226] The latter is true for this study. Compared to most literature values, the intensity levels used in this research would be classified as low fluxes. The results

PAGE 123

99 indicate that disinfection response is linearly proportion al to light intensity as illustrated in Figure 29 Figure 28: Effect of light intensity on disinfection for control organisms at 0.01 g L1 Degussa P25 TiO2

PAGE 124

100 Figure 29: Relation ship between intensity and average survival at 20 min This behavior is directly related to the generation of hydroxyl radicals that occurs as a result of the interaction of the catalyst and light energy. A t high light intensity the recombination of the electronhole pair is enhanced while at low fluxes OH radical formation can compete with recombination [227229] Further, the rate becomes independent of light intensity at higher fluxes and the expected ratelimiting factor becomes the mass transfer [230] 8.2.2 TiO2 concentration The average across all factors (light intensity and fatty acid distribution) show s that disinfection levels at 20 minutes had a log linear relationship with catalyst concentration from 0.10 0.50 g L1 of TiO2 ( Figure 30) D isinfection is m uch lower on average for 0.01 g L1. However, it must be kept in mind that these are main effects ; the 10010-110-210-310-410-5R = 0.99960 1 2 3 4 5log N/N0Light intensity 105(E L1s1)

PAGE 125

101 results are average s for the various combinations Specific interacti ons are discussed in the next section. The interaction between light intensity and catalyst concentration produce d completely different results. Figure 30: Log linear relationship between relatively high catalyst concentration (0 .100.50 g L1) and E. coli survival Without reference to the specific interactions, the general trend for increased disinfection is to reduce catalyst concentration. Block et al [34] made this observation for a similar range of catalyst concentration s This behavior is a direct result of colloidal absorption phenomena and light distribution in the reactor The surface coverage of catalyst particles on the cells is expected t o be relatively lower at low concentrations of TiO2. Very high catalyst concentrations (>0.5 g L1) actually result in destabilization of the colloidal suspension. As the catalyst concentration is increased without a change in 10-110-210-3R = 0.99940 0.1 0.2 0.3 0.4 0.5 0.6log N/NoTiO2concentration (g L1)

PAGE 126

102 pH, the condition for heteroc oagulation is met as the total interaction energy of the colloidal system approaches zero according to Equation ( 34 ) [125] The result is that t he catalyst and microbes particles co flocculate and rapidly settle out of solution ( Figure 31) S ince the process is synergistic, that is, it depends on the interaction of light and TiO2, the level of disinfection is significantly reduced due to the increase shading and scattering of light in high TiO2 suspensions. It indicates that the effectiveness of the process is determined by some optimum surface coverage and a maximum penetration of light. B eyond these values, i ncrease d catalyst concentration r etards the disinfection process. Figure 31: Instantaneous formation and settling of TiO2cell aggregates; stable solution of 0.10 g L1 TiO2 with 16 CFU mL1 cells (left); highly unstable suspension of 1 g L1 TiO2 with 19 CFU mL1 cells; and unstable suspension of 1 g L1 with 106 CFU mL1 cells

PAGE 127

103 8.2.3 Fatty acid modification The effect of fatty acid modification on disinfection did not show a statistically significant difference at the 95% confidence level (p = 0.071 ). Even when the fatty acid modification is analyzed at a specific intensity and catalyst concentration there are no significant differences. O ther researchers [231] working with lipids found that monounsaturated fatty acids tend to retard the progression of peroxidation by acting similar to antioxidants [231] It is believed that they may react with radicals, but somehow slow their progression and block radical chain reactions. However, the specific explanation for this result is still not yet very clear. E ven if the same effect is true in E. coli the effect is not significant in the overa ll disinfection of the organism for the variation of fatty acids in the study. It indicates that while peroxidation of the cell membrane is a key process during disinfection, small changes in the fatty acid content are not sufficient to cause major changes in the disinfection kinetics. 8.3 Interaction effects : Light intensity and Ti O2 concentration L ight intensity and catalyst concentration are evidently the two most important factors to be considered for photocatalysis. The interaction of these two factors is significant at all levels (p = 0.000) By analyzing the main effects, it c an be seen that disinfection efficiency increases as light intensity increases and catalyst concentration decreases. Even though there is some minor sensitivity to high light intensity (result not shown) disinfection was always greater in the presence of the catalyst. A t low and mid light intensity there is much less variation in effectiveness for concentrations from 0.100.50 g L1 TiO2 ( Figure 32) Also, the effecti veness at the same light intensity for 0.01 g

PAGE 128

104 L1 is much less at the chosen time interval when compared to all other concentration values. Figure 32: Interaction plots for the three independent factors at 20 min : (a) fatty acid modification vs. TiO2 concentration; (b) light intensity vs. TiO2 concentration; and (c) fatty acid modification vs. light intensity A t high light intensity the interaction effects change dramatically. The lowest concentration of TiO2 becomes the most eff ective and the effectiveness decreases with catalyst concentration across 2 orders of magnitude ( Figure 33) By doubling the light intensity from the mid to high posi tion, an increase of 5 log units of disinfection was achieved within the same 20 minutes. Whereas, the same increase in light intensity for other concentrations produce d much less disinfection ( Figure 32)

PAGE 129

105 Figure 33: Relationship between survival and TiO2 concentration at high light intensity The interaction between light intensity and catalyst concentration is the most important interaction because the main oxidants in the disinfection process are produced as result of the absorption of light by the catalyst. However, with increasing catalyst concentrations, t he reaction solution becomes saturated and only a portion of the particles receive i rradiation. Although more surface area may be available for reaction the additional catalyst particles do not participate in the reaction and the reaction rate does not increase with g rowing catalyst load beyond the optimum level [232] Three main factors are responsible for these observations; colloidal ad sorption and interaction light transmission through the sol ution and OH generation The interaction of these phenomena is illustrated in the simple model of Figure 34. 10-310-410-510-6R = 0.96290 0.1 0.2 0.3 0.4 0.5 0.6log N/NoTiO2concentration (g L1)

PAGE 130

106 Figure 34: Particle interaction and light transmission in TiO2 suspensions Firs tly, the effects of absorption of TiO2 unto a bacterial surface can be theoretically illustrated based on colloidal absorption theory. From TEM analysis it appears that there is very strong specific adsorption between the TiO2 particles and microbial cells at neutral pH. According to Figure 35, the catalyst particles (dark spots) are bound to the cells (rodshaped features). They also form secondary layers or clusters with each other in some areas. It is interesting to note that the TiO2 particles are not found in isolated areas with themselves, but predominantly occur with the cells. Further, when the theoretical adsorption kinetics of TiO2 to the cell surface is analyzed, it reveals that there is a transition from linear to nonlinear a dsorption for the range of TiO2 concentration used in the research Linear adsorption occurs when the existing adsorption of particles at the bacterial surface does not significantly prevent

PAGE 131

107 other particles from adsorbing [148] This occurs mostly at low particle concentrations (<1012 mL1). Figure 35: TEM im age of TiO2 particles (dark spots) attached to E. coli Images courtesy of Integrative Biology Microscopy Core Facility, University of South Florida However, at higher particle concentrations, the existing coverage blocks other particles and prevent s access to the surface. Under these circumstances, if TiO2 particles could be viewed as carriers of hydroxyl radicals, then it is easy to see that the access of radicals to the surface is also reduced under high concentration. However, for concentrations r anging from 0.10 to 0.50 g L1, this effect does not vary significantly. At

PAGE 132

108 concentrat ion values lower than 0.01 g L1 it appears that this phenomenon is important. Based on Equations ( 80 ) and ( 81) the concentrations that produce linear and nonlinear adsorption under flow and noflow conditions can be estimated as shown in Figure 36. The lower domain of the curves shows the region where linear adsorption occurs According to the figure the given TiO2 concentrations all lie within the non linear adsorption phase for flow conditions Figure 36: Dependence of limiting catalyst concentration and catalyst diameter The TiO2 particle number density was estimated according to published data available for the P25 catalyst such as specific surface area and particle size. 0.01 g/L 0.10 g/L 0.25 g/L 0.50 g/L 1.00 g/L10110010-110-210-310-410-50 50 100 150 200 250 300log TiO2conc (g L1) TiO2 diameter (nm) no flow flow

PAGE 133

109 Even though the RSA model does not account for hydrodynamic forces very well, it was still used to estimate the adsorption kinetics for the given concentrations ( Figure 37) It can be shown that the equilibrium coverage is reached within seconds, even though the kinetics is relatively slower for 0.01 g L1. However, given the time scale of the experiments it can be assumed that the surface coverage is similar for all TiO2 concentrations and ranges between 20 25%. Figure 37: Theoretical adsorption kinetics of TiO2 particles (25 nm dia.) unto E. coli surface under hydrodynamic conditions (stir speed was 600 rpm in test tube reactor ) According to colloidal adsorption processes, hydrodynamic shear forces tend to reduce surface coverage. For a catalyst particle of radius of 2 5 nm, cell radius of 1000 nm and 0% 5% 10% 15% 20% 25% 30% 0 5 10 15 20 25surface covergaetime (s) 0.50 0.25 0.10 0.01 g L 1 g L 1 g L -1 g L-1g L-1

PAGE 134

110 TiO2 concentration of 4x1012 mL1 (0.10 g L1) there is an estimated theoretical initial surface coverage of 52% which is reduced to 25% at equilibrium under flow conditions The second important factor, light transmission, varies with TiO2 concentration as well There is usually an uneven distribution of light in a TiO2 reactor because the light is attenuated as it is transmitted through the solution. As much as 50 percent of the incident light can be absorbed within the first 30 mm from the reactor wall in a suspension of 0.50 g L1. A light transmission test was conducted to develop the profile of light through the reactor used in this study. TiO2 suspension corresponding to the various concentrations was gradually added to a borosilicate Petri plate of similar thickness to the reactor test tube. The light intensity passing through the solution was measured with a UV radiometer placed directly below the plate. The transmission of light is plotted in Figure 38. More than 95% of the incident light passes directly through a suspension of 0.01 g L1, while just about 1% passes through 0.5 g L1. Based on the analysis, an exponential decay of light intensity inside the reactor co uld be established according to: I = I exp ( c ) ( 107 ) where I is the transmitted light intensity, is the coefficient of attenuati on per mass concentration of catalyst (L g1 cm1) c is the mass concentration of catalyst (g L1) and is the distance from the reactor wall (cm) The coefficient of attenuat ion was found to be 0.720.10 L g1 cm1.

PAGE 135

111 Figure 38: Light transmission through reactor at high light intensity According to Figure 38, active photocatalytic activity occurs within a gradually reducing zone close to the reactor surface as the concentration is increased. This means that microorganism s in high concentration suspensions are not exposed to the light as frequently as organisms in lower concentration suspensions [84, 154] Since the bacterial surface coverage of catalyst particles is comparable for all concentrations used in the study, it is clear to see why light intensity has the largest overall effect on the process under the given conditions. This information was used to update the code for the model by writing an algorithm which accounts for the radial variation of intensity in the reactor. Eq uations ( 100 ) and ( 104 ) were adjusted accordin gly 0% 20% 40% 60% 80% 100% 0 1 2 3 4 5 6 7% transmittancedistance from reactor wall (cm) 0.00 0.01 0.10 0.25 0.50 1.00 g L-1g L-1g L-1g L-1 g L-1 g L-1

PAGE 136

112 [ ] = { [ OH ] [ ] } r dr ( 108) d [ O H ] dt = { G ( 1 ) [ OH ] [ cell ] } r dr ( 109) where is the radius of the reactor. Finally, the generation of OH radicals accounts for the significant influence of light intensity and catalyst concentration. The generation of OH radicals as given by Equation ( 89 ) depends on the integrated absorption fraction F of the catalyst suspension (see Equation ( 88) ) Values of F were determined by Sun and Bolton [158] The absorption fraction increases to a maximum with TiO2 concentration. Beyond about 0.10 g L1, there is no sign ificant increase. Values for F were interpolated and extrapolated to construct Figure 39, which shows the expected OH radical generation rate as a function of catal yst concentration. There is no significant increase in the generation rate G for catalyst suspensions exceeding 0.1 g L1. However, the influence of the generation rate on the interaction effect is made much more apparent when the generation rate per mass or per particle is considered (Figure 40 ). There is an exponential drop in the generation rate per mass of catalyst beyond 0.1 g L1. This is a clear indication tha t the additional catalyst particles reduce the efficiency of the process.

PAGE 137

113 Figure 39: OH radical generation rate in TiO2 suspension at pH 7 in deionized water ; (a) high intensity I0 = 4.37105 E L1 s1, (b) mid intensity I0 = 2.405 E L1 s1 and (c) low intensity I0 = 1.35105 E L1 s1 In conclusion, the significance of the interaction of TiO2 and light intensity on disinfection favors lower catalyst concentration and higher light intensity within an optimum range. At low TiO2 concentrations, the colloidal suspension is more stable, the distribution of light is fairly uniform, and there is a higher radical generation rate per mass of catalyst. (a) (b) (c)10-710-810-90.0 0.2 0.4 0.6 0.8 1.0 1.2log G (mol L1s1)TiO2concentration (g L1)

PAGE 138

114 Figure 40: Dependence of normalized OH radical generation rate on catalyst concentration 8.4 Model validation 8.4.1 Inputs and fitting parameters The model developed in the study was very complex, but potentially useful for estimating the effect of a number of parameters such as catalyst concentration, light intensity, salt concentration and pH Inputs to the model included published data on adsorption constants for anions ( Table 2 ) electrolyte con centration ( Table 5 ), the integrated adsorption fraction for specific catalyst concentrations ( Figure 16) the incident light intensity ( Three sets of control experiments were conducted; (1) solution of catalyst 0% 20% 40% 60% 80% 100% 120% 0.0 0.2 0.4 0.6 0.8 1.0normalized generation rate (G')TiO2concentration (g L1)

PAGE 139

115 and microbes with no light (dark experiments); (2) irradiated solutions of microbes with no catalyst present (irradiated blank); and (3) non irradiated solution of microbes only (organism control). The final composition of the reaction solution was made up by adding the appropriate volumes of stock solutions together and then pouring the mixture into the reaction vessel ( Table 4). The final composition of electrolytes is shown in Table 5. Table 3) the reactor radius, and the quantum yield of OH radical generation according to Sun and Bolton [158] The entire model was so lved numerically using a fifth order Runge Kutta method in MATLAB coupled with a least square solver to obtain three unknown parameters ; the se included the disinfection rate constant the reaction order with respect to OH radicals and the OH radical consumption rate constant This is a particular strength of the model. It is able to utilize predetermined values without the need to over fit the model with too many unknown independent parameters. The expectation of the fitting procedure wa s that the rate constants and reaction order should not vary significantly, particularly for a given organism. Previous studies have reported dependence of the rate constant on TiO2 concentration [21] but this study considered tha t to be at odds with reaction kinetic theory. Table 8 and Table 9 show the fi tting parameters and the coefficient of determination of the regression (R2). However, much confidence cannot be placed in the R2 value because the data was fitted across many orders of magnitude. This means that the least square procedure is biased towards the largest numbers which occur at the beginning of the survival curve. A more reliable test was to observe the overall survival curve shape and make actual comparisons

PAGE 140

116 between the final predicted disinfection values and experimental results. Furth er, to improve the accuracy of the fit, the least square fit was performed between a unit matrix and the reciprocal of the model data multiplied by the experimental values.

PAGE 141

117 Table 8: Rate constants and reaction order as predicted by the model for unmodified and C16:1 modified organisms U nmodified C16:1 k dis n k OH R 2 k dis n k OH R 2 TiO2 (pM -n s -1 ) ( ) (L n CFU -1 s -1 pM n-1 ) (pM -n s -1 ) ( ) (L n CFU -1 s -1 pM n-1 ) HIGH 0.01 1.50E+05 1.3 1.00 0.988 1.16E+05 1.2 3.36 0.984 0.10 9.30E+04 1.5 3.49 0.963 1.33E+04 1.3 1.00 0.998 0.25 3.43E+05 1.6 1.08 0.985 1.59E+04 1.3 1.00 0.971 0.50 1.49E+05 1.4 1.72 0.968 5.78E+04 1.4 1.00 0.947 MID 0.01 1.15E+05 1.3 4.25 0.969 1.47E+05 1.4 2.78 0.933 0.10 1.05E+04 1.3 1.00 0.995 1.32E+04 1.3 1.00 0.992 0.25 1.46E+04 1.2 1.00 0.984 1.71E+04 1.3 1.00 0.991 0.50 3.32E+05 1.5 1.00 0.979 3.84E+04 1.3 1.00 0.988 LOW 0.01 7.50E+05 1.6 1.00 0.968 2.17E+05 1.5 1.00 0.982 0.10 1.69E+04 1.3 1.00 0.981 3.35E+04 1.4 1.00 0.987 0.25 4.94E+04 1.4 1.00 0.941 9.16E+05 1.8 1.00 0.960 0.50 1.50E+05 1.6 2.34 0.992 9.31E+04 1.5 1.00 0.948

PAGE 142

118 Table 9: Rate constants and reaction order as predicted by the model for C18:1 and C18:3 modified organisms C18:1 C18:3 k dis n k OH R 2 k dis n k OH R 2 TiO2 (pM -n s -1 ) ( ) (L n CFU -1 s -1 pM n-1 ) (pM -n s -1 ) ( ) (L n CFU -1 s -1 pM n-1 ) HIGH 0.01 2.53E+05 1.4 1.00 0.956 1 .59E+05 1.3 1.00 0.970 0.10 6.48E+04 1.5 1.00 0.966 5.69E+04 1.5 3.56 0.987 0.25 1.78E+04 1.3 1.00 0.983 1.01E+05 1.4 3.29 0.995 0.50 2.30E+04 1.3 1.00 0.993 2.02E+04 1.2 1.00 0.999 MID 0.01 1.35E+05 1.3 3.40 0.971 5.93E+05 1.5 1.32 0.922 0.10 9.82E+03 1.2 1.00 0.966 3.45E+04 1.3 3.49 0.943 0.25 4.37E+04 1.4 1.00 0.996 3.06E+04 1.3 1.00 0.986 0.50 7.11E+04 1.4 1.00 0.977 4.52E+04 1.4 1.00 0.981 LOW 0.01 1.62E+05 1.3 8.44 0.857 4.74E+05 1.6 1.00 0.974 0.10 1.26E+04 1.3 1.00 0.991 1.12E+04 1.3 1.00 0.966 0.25 1.75E+04 1.3 1.00 0.996 9.67E+03 1.2 1.00 0.975 0.50 6.30E+04 1.2 1.00 0.952 1.00E+06 1.7 1.00 0.951

PAGE 143

119 Since the concentration of OH radicals measured in TiO2 suspensions is usually very low [68, 233] the use of pico moles appears to be appropriate to describe the disinfection rate constants. It was observed that the values varied within two order s of magnitude across all experiments This is in keeping with the expectation that the ra te constant should not vary significantly for the same organism. However, when the rate constant is examined as a function of concentration and light intensity, the variation is much great er at higher concentrations. At TiO2 concentrations of 0.01 and 0.10 g L1 the variation is within an order of magnitude. It is believed that these variations are related predominantly to colloidal interactions and the ratio of TiO2 particles to cell numbers, both of which are explained in later sections. Figure 41: Box plot of the disinfection rate constant kdis obtained from the model

PAGE 144

120 The reaction order with respect to OH concentration a lso exhibited a s mall range of variability (1.40.1) In many chemical disinfection studies, the reaction order is usually assumed to be unity. However, the fact that the reaction order had to be greater than one to fit the data is not all that surprising. This is largely because there are numerous reactions of OH with biomolecules which eventually lead to cell inactivation. W hile OH may be the main oxidant, it does not preclude other radicals and oxidizing agents such as O and HO from participating in disinfection r eaction s Hydroxyl radicals are short lived even in pure buffered water because they undergo a recombination reaction to form hydrogen peroxide according to Equation ( 110 ) The second order hydroxyl radical recombination competes with slower first order reactions especially at higher doses when higher concentrations of hydroxyl radicals are produced. 2 OH H O ( 110 ) The formation of hydrogen peroxide also leads to the generation of other radicals, either through the reaction with OH or homolytic scission [50, 61] The hydroxyl radical reacts with H2O2 at a relatively slow er rate (2.7 107 mol1 dm3) and consumes only a small amount of the formed H2O2 [234] OH + H O H O / O + H O ( 111) Even though the concentration of the s uperoxide radical is usually lower than the hydroxyl radical in solution, it has been shown that the former can contribute about 20% of the radical concentration [234]

PAGE 145

121 8.4.2 Survival curve predictions Under the given conditions, t he model produces a sigmoidal survival curve when plotted on the linear axes in accordance with the typical inactivation behavior of E. coli ( Figure 42) The initial lag and the onset of the loglinear phase for most of the disinfection dat a are well defined by the model ( Figure 43) H owever, the greatest challenge seems to be replicating the latter end of the disinfection curves close to the limit of detection. There are a number of factors responsible for this deviation Figure 42: Typical s igmoidal survival of E. coli at low intensity illumination N0 = 16 CFU L1. Firstly, an implicit assumption in the development of the model is that the disinfection process is deterministic. This assumption works well for molecules because their numbers are so incredibly high. However, it can be argued that disinfection begins 0 0.2 0.4 0.6 0.8 1 0 10 20 30 40 50 60 70N/N0time (min) model 0.01 g/L exp 0.10 g/L exp 0.25 g/L exp 0.50 g/L exp

PAGE 146

122 as a deterministic process when the number of microbes in solution is high (109 L1) This means that each microbe has about the same chance of being inactivated. However, as the microbia l numbers drop significantly, it transitions to a stochastic process where the probability of inactivation varies from one organism to the next The stochastic approach to model this behavior would be to define a function which accounts for the changing su rvival probabilities of individual cells [139] The challenge, however, is that stochastic models are mostly empirical and cannot be obtained from deterministic formulations. Even though there may be mechanistic contributions to the probability function, such as uneven distribution of light, particularly in high concentration suspensions of TiO2, it is still very difficult to formulate such a function and determine the influence of many other parameters as in the current model A second challenge, which occurs towards the end of the survival cur ve is the determination of cell numbers close to the limit of detection. At very low concentrations, t here is an inheren t restriction on the number of cells which can appear on agar plates with sufficient accuracy to allow a resolution of the true cell count. In this study the lowest count that could be determined was 1 CFU per 100 L ( that is, 10 CFU mL1) This corresponds to 1 CFU on an agar plate with an associated relative error of 100%. The results indicate that there are significant fluctuations when determining cells at low concentration. The challenge for the model is that close to limit of detection, it predicts a uniform rate of disinfection. It is unlikely that this level of disinfection can be realized in a real population of cells or replicated in the lab.

PAGE 147

123 Figure 43: Survival curve for E. coli treated at low light intensity Lastly, the existence of a finite residual survival, particularly for high catalyst concentration was observed. The residual survival is characterized by a sudden tailing off of the disinfection curve following the exponential decay ( Figure 44) This was determined to be a real phenomenon because the cell count was usually to the right of the limit of detection. As previously explained, it is believed that the uneven di stribution of

PAGE 148

124 light in the high concentration suspensions reduces the exposure of cells in the irradiated fraction of the reactor This is accompanied by a sharp reduction in the disinfection rate. Recall that for suspensions with less catalyst loading the irradiation zone is much wider, that is, the light distribution is more uniformed. As the cells are disinfected, the probability of entering the irradiated zone also drops, but not as much as in the case of high catalyst concentration. The consequence of this phenomenon is that disinfection is more complete in the case of lower concentrations, even if the overall process is slower ( Figure 33 ) Figure 44: Effect of concentration loading on residual survival of E. coli at high light intensity

PAGE 149

125 8.4.3 Influence of light intensity and catalyst concentration The model captures the effect of light intensity and catalyst concentration on the disinfection very well. Without m uch change in the rate constants, it shows that the main effects are dominated by the interaction of these variables as determined previously. The processes involved in this interaction include ligh t tran smission, OH radical generation, and the absorption effects between colloids. Most of the variations from one survival curve to another are related to changes in light intensity and catalyst concentration, since other parameters were held constant. 8.5 Particle interaction effects and colloidal stability Apart from the light absorption processes and chemical reactions involved in photocatalytic disinfection, the results indicate that the interaction of particles in the colloidal suspension has a very significant impact on the disinfection efficiency. These interactions are controlled by such factors as pH, ionic strength, particle size, and particle concentration. The interaction of TiO2 particles and bacterial cells can be explained by DLVO and the s oftparticle theory [125, 129, 235] These theories provide the basis for explaining the effec t of the above parameters on colloidal stability, and hence, on the photocatalytic disinfection process. Colloidal stability is here defined as the ability of the colloids to resist rapid aggregation and settling. The total interaction energy of the particles in solution is given as the sum of the van der Waals and electrostatic interactions given in Equations ( 34) through ( 40) As can be observed from these relationships, the interaction depends on the surface potential of

PAGE 150

126 the particles and the separation distance The former is most commonly derived from e lectrophoret ic studies in which t he electrophoretic mobility e is related to the zeta potential ( potential) at the particle surface through the Smoluchowski equation, = ( 112) The potential is a theoretical approximation of the potential of the inner portion of the diffuse layer which is often used to characterize the stability of colloidal systems It defines the electric potential close to the plane of shear ( hydrodynamic slip plane) where the solvent molecules are not bound t o the particle surface. The ions located in the region from this point toward the particle are assumed to move as a unit with the particle. Using the electrophoretic data of Liu et al [236] Fernandez Ibanez et al [149] and Suttiponpa rnit et al [237] it was possible to construct a graph ( Figure 45) of the potential of TiO2 as a function of pH and ionic strength. Similarly, the electrophoretic data for E. coli reported by Sonohara et al [115] based on the soft particle theory [117] were used to calculate the potential at 0.01 M and 0.10 M ionic concentration as a function of pH

PAGE 151

127 Figure 45: potential of P25 TiO2 particles and E. coli cells as a function of pH in 0.01 M (open and filled circles) and 0.10 M (open and filled squares) ionic strength respectively Data modified Liu et al [236] Fernandez Ibanez et al [149] and Suttiponparnit et al [237] 8.5.1 Influence of ionic strength on disinfection An analysis was performed to test the model for its response ionic strength Two levels of ionic strength were i nvestigated, 0.01 M and 0.2 M The salt composition is reported in Table 5. Based on the experimental data, the overall disinfection p rocess is significantly slowed b y at least two orders of magnitude at the higher ionic strength ( Figure 46 ) T he constants obtained under the 0.01 M ionic strength analysis were used as inputs to the model to assess whether the increase in salt content alone could account for 80 60 40 20 0 20 40 60 2 4 6 8 10 12zeta potential (mV)pH

PAGE 152

128 the difference observed in the experiments. According to the model simulation there was no significant difference in disinfection at these two ionic strength levels. A further theoretical simulation with an ionic strength of 20 M revealed that disinfection would be significantly reduced at this very high electrolyte concentration ( Figure 47 ). Figure 46: Influence of salt content on the disinfection process at pH 7 (light intensity = 2.45 E L1 s1, TiO2 = 0.50 g L1)

PAGE 153

129 As a second approach to modeling the effect of salt concentration on disinfection, the model w as run without using the previously obtained kinetic constants for 0.01 M ionic strength. The newly obtained kinetic constants for the 0.20 M ionic strength were determined to be kdis = 1104 pM1.2s1 and n = 1.2. The results are shown in Figure 48. The differences between the disinfection rate constant for the two ionic strength solutions are attributed to colloidal stability effects Figure 47: Model simulation of the effect of salt content with previously determined rate constants ( kdis = 3.32x105 pM1.5s1; n = 1.5; kOH = 1 L1.5 CFU1 s1 pM0.5)

PAGE 154

130 Figure 48: Simulated results for effect of sal t concentration on disinfection (light intensity = 2.45 E L1 s1, TiO2 = 0.50 g L1) T he effect of salt content in the model was built around a reduction in the generation rate through the blocking of OH sites as more electrolytes adsorb to the catalyst surface. However, the data suggests that colloidal stability is more sensitive to ionic strength than the blocking of OH sites. This is particularly true at neutral pH, where the adsorption of salts to the catalyst surface is not expected to be a significant factor since the TiO2 surface has a very low charge.

PAGE 155

131 However, visual observation of the TiO2microbe suspension shows that the colloids are unstable at salt concentrations exceeding 0.10 M; also confirmed experimentally by other researchers [84, 90, 92, 238] The TiO2 suspended in 0.10 M and 0.20 M ionic solutions flocculated and settled very rapidly, while the colloids remained dispersed in the 0.01 M ionic solution ( Figure 49). This phenomenon is not currently captured in the model The mechanisms through which colloidal destabilization reduces disinfection efficiency has not yet been studied. However, it is suspected that the increase in particle size reduces the rate at which OH radicals are generated. Figure 49: Settling of TiO2cell colloids (0.5 g L1 and 106 CFU L1 respectively) in 0.01 M (left) 0.10 M (center), and 0.20 M (right) ionic solutions at pH 7 and 25C. The destabilizing effect of ionic strength on the TiO2cell suspension can be explained by considering the total interaction energy between the colloids at 0.01 and

PAGE 156

132 0.10 M ionic strength ( Figure 50) The interaction energy was calculated as a function of separation distance, according to Equation ( 34 ) for TiO2 particle s of 25 and 1000 nm and bacterial cells of 1000 nm diameter at pH 7. At the given pH, the surface charge of the catalyst is mostly neutral with less than 10% negative species. However, the cell surface is mostly negative at neutral pH. At a separation distance between 40 nm the catalyst particles of 25 nm diameter begin to experience repulsion from the bacterial surface i n the 0.01 M ionic solution. The strength of the repulsion rapidly increases as the catalyst particle s get closer to the cell surface; as a result, the colloidal suspension is more stable. However, for larger TiO2 particle, there is a primary minimum potential energy around 50 nm from the bacterial surface. At closer separation distances the large particles begin to experience repulsion. In the 0.10 M ionic solution, the interaction energies are much lower. A 25 nm TiO2 particle has virtually no energy barrier preventing it from adsorbing to the cell surface. A larger particle experiences a greater attraction with a primary minimum close to 10 nm from the surface. The low interaction energy and net attractive force creates the conditions for destabilizing the suspension and forces coagulation. 8.5.2 Influence of pH The influence of pH on t he disinfection process was simulated by the model and is illustrated in Figure 51. The simulation confirms the finding of other authors who studied the effect of pH on E. coli disinfection [27, 55, 160, 239] It shows that in the pH range of 68 the disinfection rate is very similar. Simulations of lower and higher pH

PAGE 157

133 were also conducted, but those resul ts cannot be interpreted, since the influence of pH on natural survival of E. coli is not included in the model. Figure 50: Total interaction energy (VT) as a function of separation distance between E. coli (1000 nm dia.) and P25 TiO2 at pH 7 and 25C : a(1) 0.01 M TiO2 1000 nm dia.; a(2) 0.01 M TiO2 25 nm dia.; b(1) 0.10 M TiO2 1000 nm dia.; b(2) 0.10 M TiO2 25 nm dia. It is common knowledge that E. coli survives best within the pH range of 58, and is affected by low and high pH values. The model only accounts for changes in the catalyst surface chemistry and the effect of pH on absorption of anions. Apart from the ability of E. coli to thrive in neutral solutions, the isoeletric point of TiO2 also occurs wi thin this pH range. b(2) a(1) a(2) b(1) 10 8 6 4 2 0 2 4 6 8 10 0 10 20 30 40 50 60total energy, VT(kT)separation distance (nm)

PAGE 158

134 Figure 51: The effect of pH simulated by the model (unmodified cells treated at mid light inten sity with 0.01 g L1 TiO2) However, if the natural survival of the organism was ignored below pH 5, the process would be significantly retarded due to increase in the adsorption of anionic species. At higher pH values this effect would become negligible. The model does not include the effects of cations, which may influence the process at higher pH values. From a co lloidal stability perspective, it was found that pH has less of a destabilizing effect and slower coagulation kinetics than ionic strength. Settling

PAGE 159

135 experiments were conducted at pH 3, 7 and 11 for TiO2cell suspensions. T here instantaneous formation of la rge visible aggregates observed in solutions of ionic strength greater than 0.10 M was not observed for all pH values. However, after several hours (1824 hrs) the colloids in solutions of pH 3 and 11 settled out completely almost complete, whereas the colloids suspended at neutral pH were still stable. Figure 52: Long term (24 hrs) settling of TiO2cell colloids in solutions of pH 3 (left), pH 7 (center), and pH 11 (right) The interaction energy simulated from DLVO could not a ccount for the destabilization at high pH, largely because the cells do not survive in such basic solutions. In the strictest theoretical sense, the solution should be stable at high pH because both the catalyst and cells are negatively charge. However, wh en the cells die, the ability to maintain osmotic balance with the solution is lost and the charges at the cell surface may induce conformations that allow the colloids to destabilize.

PAGE 160

136 Figure 53: Total interaction energy (VT) as a function of separation distance between E. coli (1000 nm dia.) and P25 TiO2 (1000 nm dia.) at different pH values 8.6 Byproduct evolution and peroxidation kinetics 8.6.1 Lipid peroxidation as proof of membrane damage In previous studies of photocata lytic disinfection, lipid peroxidation was used as an index to confirm the effects of OH radicals on cellular membranes during photocatalysis [20, 184] I n these studies the oxidation of PE in homogenous solution was compared to the disinfection of cells However, w hile this approach yielded useful information about byproduct formation, they did not offer muc h information on the kinetics of cell membrane oxidation Another consideration is that pure PE solutions or mixtures enriched in PE are notable for being unstable and adopt a hexagonal phase [240, pH 5 pH 7 pH 6 pH 8 20 15 10 5 0 5 10 15 20 0 10 20 30 40 50total energy, VT(kT)separation distance (nm)

PAGE 161

137 241] They do not spontaneously form lamellar phases in aqueous media as do other phospholipids. They often require a stabilizing agent to main tain a bilayer structure similar to biological membranes. In order to justify the use of peroxidation kinetics and rate constants in the model, lipid vesicles were used as model E. coli membranes. Lipid vesicles of PE were prepared with the addition of PG, which served as a stabilizing agent, but also represented a more realistic and natural E. coli membrane. The vesicles were also sized to be comparable to real cells. C ells and vesicles were then exposed to illumination with TiO2 and the evolution of MDA and LOOH was measured during the experiments to assess membrane peroxidation. 8.6.2 Lipid vesicle composition and size distribution The average diameter of the lipid vesicles was approximately 0.5 m ( Figure 54). Even though the vesicles are not rodshaped like E. coli the results correspond well to the published data on the size of E. coli cells, which measure on average 0.5 m by 1 m [70] The size and shape of the vesicles were confirmed with TEM images as shown in Figure 55. The size distribution of the vesicles was important to establish the precise kinetic behavior of the system. The interaction of the particles (photocatalyst and cell) is based on particle size. The very distinct darkened outline on the features in Figure 55 indicates that these were most likely multilamellar vesicles. Due to the nature of the TEM sample preparation, many of the vesicles seen in the figure were the very large vesicles which

PAGE 162

138 settled out unto th e TEM grid. The fatty acid composition of the vesicles was estimated from the manufacturers data and is shown in Table 7. The predominant unsaturated fatty acid was c is vaccenic acid (C18:1 n 7) in PE and oleic acid (C18:1 n9) in PG. Figure 54: Size distribution by volume based on photon correlation spectroscopy of the lipids vesicles in 1 PBS solution (molar ratio 1:1 PE to PG) Figure 55: TEM images of PE/PG lipid vesicles. Images courtesy of Integrative Biology Microscopy Core Facility, University of South Florida

PAGE 163

139 8.6.3 MDA production during photocatalytic experiments MDA was detected in photocatalytic experiment s containing 1011 CFU L1 in order to increase the levels of MDA detection. Even though the MDA test has some limitations, the evolution of MDA in all the samples was very similar and consistent between experiments. The monotonic accumulation of MDA was ob served during the first 20 30 minutes of the photocatalytic experiments for both unmodified E. coli cells and lipid vesicles. Thereafter, a steady decrease in concentration was recorded ( Figure 56). There was a prolonged increase in MDA for the cells modified with linolenic acid. The overall trend for MDA release during photocatalysis was first observed by Maness et al [20] for the disinfection of E. coli cells under similar conditions. The trend appears to be consistent with the peroxidation of membrane lipids followed by the degradation of MDA (either naturally or photocatalytically). More MDA was produced in the vesicles because they were composed only of fatty acids, whereas cells have their fatty acids distributed in the membrane with other biological structures such as proteins. A common criticism of the TBA assay is that M DA is produced by artifact ual means during the harsh processing conditions of the test [187, 210, 213, 242, 243] However, the use of BHT antioxidant in the test serves to eliminate or reduce the production of MDA during the processing of the sample [242] In addition, the conditions of these tests were much milder compared to the more traditional TBA tests which utilize boiling temperatures to facilitate the reaction with MDA. The most convincing evidence of all is the fact that no measureable MDA concentrations were detected in any of the

PAGE 164

140 control experiments (data not shown), leading to the conclusion that the observed trend resulted from treating the cells and vesicles photocatalytically. The TBA test is the most fr equently used method to detect lipid peroxidation, but it has also been criticized for its non specificity, particularly in complex biological systems. However, it has proven useful in well defined systems such as the oxidation of lipid vesicles [209, 244] Hen ce, when the time characteristic for MDA evolution during oxidation of the model membranes is compared to real cells, there is strong evidence that the trend observed in cells resulted from membrane peroxidation. In addition, the byproduct evolution simula ted by the model is a close match to the observed data ( Figure 57) However, this simulation could possibly include other byproducts apart from MDA. 8.6.4 Effect of su pplemental fatty acid on MDA production in cells For the cells modified with linolenic acid, it was found that MDA accumulation rate was relatively slow compared to the other cells and vesicles ( Figure 56 ). There was a gradual increase which peaked around 45 minutes. Control cells and cells supplemented with oleic acid (C18:1 n 9) did not produce this extended MDA evolution curve, which leads to the belief that the kinetics is affected by the fatty acid composition. However, it is difficult to make a definitive conclusion about the impact of the fatty acid supplementation on MDA production because of the complexity of the system and the undefined sink processes for MDA.

PAGE 165

141 Nonetheless, it is generally expected that increases in polyunsaturated fatty acid content would render the cell more sensitive to oxidation and an increase in the MDA production could be possible. Other studies have shown that t he oxidizability of cells can be altered by supplementation with external fatty acids [208, 231, 245] The results in this case seem to suggest that the enriching of the membrane with monounsaturated fatty acids retards the rate of MDA production, particularly when supplemented with linolenic acid. The actual mechanism by which these monosaturated fatty acids are able to reduce the peroxidation rate is still not clear. However, a poss ible explanation for this observation is the oxidation of monosaturated fatty acids does not produce bioactive byproducts responsible for enhancing membrane peroxidation [231, 246] This effect, described by Lee et al [231] is similar to an antioxidant in which the monosaturated fatty acids serve as a temporary sink for the capture of free radicals, and retard propagation due to their reduced reactivity.

PAGE 166

142 Figure 56: MDA production during photocatalytic ex periments with P25 TiO2: I0 = 4.855 E L1s1, N0 11 C FU L1: (a) unmodified cells; (b) E. coli PE/PG vesicles; (c) cells supplemented with oleic acid; (d) cells supplemented with linolenic acid. The data are fitted with a fourth order polynomial 0 10 20 30 40 50 60 0 0.5 1 1.5 2 Time (min) MDA ( mol/g) 0 10 20 30 40 50 60 0 0.5 1 1.5 2 2.5 3 3.5 Time (min) MDA ( mol/g) 0 10 20 30 40 50 60 0 0.5 1 1.5 2 2.5 Time (min) MDA ( mol/g) 0 10 20 30 40 50 60 0 0.5 1 1.5 2 2.5 Time (min) MDA ( mol/g) (b) (a) (c) (d)

PAGE 167

143 Figure 57: Typical curve for the simulation of byproducts from the model 8.6.5 Correlation between peroxidation and disinfection From the analysis of main effects of fatty acid modification, it was found that there were no significant differences between the organisms. This suggests that while peroxidation is an important process f or disinfection, it is not the sole process. It is very likely that oxidation of proteins and other biomolecules are just as important in the process [247] Polyunsaturated fatty acids are usually very sensitive to oxidati on. However, they were not present in significant proportions in E. coli The MDA produced in these studies could result from both fatty acids and other cellular constituents.

PAGE 168

144 8.6.6 LOOH production during disinfection The i llumination of TiO2 in the prese nce of E. coli cells and lipid vesicles yielded measurable concentrations of hydroperoxides ( Figure 58) The nature of the LOOH test ensures that only peroxide generated from the cells is measured. The kit uses a number of internal controls, which correct for endogenous iron content and possible hydrogen peroxide. A significant increase in LOOH concentration was observed during the early stages of the experiments. There was an apparent decrease in the hydroperoxide content at longer illumination times. As in the case of MDA, this trend indicates that the resulting kinetics is a consequence of both photocatalyticallyinduced formation and decomposition of hydroperoxides. This is consistent with the concomitant generation of MDA during the experiments.

PAGE 169

145 Figure 58: Time characteristics of lipid hydroperoxide detection during photocatalytic treatment: ( E. coli cells; ( E. coli phospholipids 0 10 20 30 40 50 60 0 2 4 6 8 10 12 14 Time (min)LOOH ( mol/g)

PAGE 170

146 CHAPTER 9: CONCLUSIONS The p hotocatalytic disinfection of E. coli with suspended catalyst particles is a complex process that involves the interplay of many phenomena. These include light absorption and scattering, semiconductor photo excitation and charge carrier generation, electrochemical surface reactions (including electron transfer reactions, adsorption, and acid base reactions) and heterogeneous colloidal interactions All these processes play a significant role in the overall inactivation efficiency F or a given solution composition, light intensity and catalyst concentration are the most significant operational factors in the entire process. The combination of light intensity and catalyst concentration determine the light absorption and scattering effects and the OH radical generation rate. Low catalyst concentration and high light intensity favor higher log inactivation. At low TiO2 concentrations, the colloidal suspension is more stable, the distribution of light is fairly uniform, and there is a higher radical generation rate per ma ss of catalyst. The mechanistic model developed in the study is very comprehensive. Apart from light intensity and catalyst concentration, it has the potential to predict the effect of pH and ionic strength on the disinfection process. However, these predi ctions are confined to stable suspensions. The disinfection efficiency is significantly reduced in destabilized suspensions which occur at high ionic strength, excessive particle concentration concentrations, and extreme pH.

PAGE 171

147 It was found that t he evolution of byproducts is consistent with the oxidation of cell membranes. MDA and LOOH w ere detected when E. coli cells and model cell membranes were exposed to photocatalytic action. Not only were the byproducts similar, but the time evolution showed very simil ar trends. However, no statistically significant effect could be observed by modifying the fatty acid profile of the cells. This is attributed to the fact that other biomolecules such as proteins are more abundant than polyunsaturated fatty acids and also react at high rates with the OH radical. Therefore, it can be concluded that even though peroxidation of the membrane is an important process in disinfection of E. coli the fatty acid distribution was not sufficiently altered to observe any changes to the overall disinfection kinetics. Finally, the model is flexible and has good validity for predicting the disinfection behavior of E. coli The reaction rate parameters are within reasonable range and exhibit only small variability especially at low catalyst concentrations The reaction rate order with respect to the OH radical was found to be greater than unity. However, the re is an inherent challenge to replicate residual survival, especially at low cell concentration because of the deterministic na ture of the model T he model predicts uniform inactivation close to an d beyond the limit of detection, which is not always the case. The high fluctuations of bacteria at low concentrations make this challenge very difficult to solve. One technique would be to utilize stochastic models which can define the probability of disinfecting an individual organism based on the reaction composition.

PAGE 172

148 CHAPTER 10: RECOMMENDATIONS The proposed model can be exploited for modeling bacterial survival notwithsta nding the challenges. However this is the first formulation of such a comprehensive model for photocatalytic disinfection. Naturally, many experimental research problems still exist and require attention. The most important would appear to be: experimenta l determination of the adsorption phenomena of TiO2 catalyst particles to bacteria under varying conditions of pH and catalyst concentration developing a stochastic model with a mechanistic basis for disinfection, particularly for treatment of solutions co ntaining a low concentration of cells testing the model for disinfection under flow conditions, particularly under solar conditions including the effects of salt concentration on double layer effects

PAGE 173

149 REFERENCES 1. WHO, Combat ing waterborne disease at the household level 2007, World Health Organization: Geneva. 2. United Nations, Water for People, Water for Life: United Nations World Water Development Report UNESCO and Bergham Books, Barcelona, 2003. 3. J. Lonnen, S. Kilvingt on, S.C. Kehoe, F. Al Touati, and K.G. McGuigan, Solar and photocatalytic disinfection of protozoan, fungal and bacterial microbes in drinking water, Water Res. 39 (2005) 877883. 4. E.F. Duffy, F. Al Touati, S.C. Kehoe, O.A. McLoughlin, L.W. Gill, W. Gernjak, et al., A novel TiO2assisted solar photocatalytic batch process disinfection reactor for the treatment of biological and chemical contaminants in domestic drinking water in developing countries, Solar Energy 77 (2004) 649655. 5. F. Mndez Hermi da, E. Ares Mazs, K.G. McGuigan, M. Boyle, C. Sichel, and P. Fernndez Ibez, Disinfection of drinking water contaminated with Cryptosporidium parvum oocysts under natural sunlight and using the photocatalyst TiO2, J. Photochem. Photobiol. B. 88 (2007) 105 111. 6. G.C. White, The Handbook of Chlorination. 2nd ed., Von Nostrand Reinhold, New York, 1986. 7. M.J. Nieuwenhuijsen, M.B. Toledano, N.E. Eaton, J. Fawell and P. Elliott, Chlorination disinfection byproducts in water and their association with adverse reproductive outcomes: a review, Occup. Environ. Med. 57 (2000) 7385. 8. E.A. Bryant, G.P. Fulton, and G.C. Budd, Disinfection Alternatives for Safe Drinkin g Water Environmental Engineering Series. van Nostrand Reinhold, New York, 1992. 9. Safe Drinking Water Act Amendments of 1996, in P.L. 104182. 1996. p. 16201621.

PAGE 174

150 10. E.W. Rice, J.C. Hoff, and F.W. Schaefer, III, Inactivation of Giardia cysts by chlori ne, Appl. Environ. Microbiol. 43 (1982) 250251. 11. W.Q. Betancourt and J.B. Rose, Drinking water treatment processes for removal of Cryptosporidium and Giardia, Veterinary Parasitology 126 (2004) 219234. 12. S.D. Richardson, A.D. Thruston, T.V. Caughra n, P.H. Chen, T.W. Collette, T.L. Floyd, et al., Identification of new ozone disinfection byproducts in drinking water, Environ. Sci. Technol. 33 (1999) 33683377. 13. S.D. Richardson, A.D. Thruston, T.V. Caughran, P.H. Chen, T.W. Collette, T.L. Floyd, et al., Identification of new drinking water disinfection byproducts formed in the presence of bromide, Environ. Sci. Technol. 33 (1999) 33783383. 14. C. Srinivasan and N. Somasundaram, Bactericidal and detoxification effects of irradiated semiconductor cata lyst, TiO2, Curr. Sci. 85 (2003) 14311438. 15. C. McCullagh, J.M.C. Robertson, D.W. Bahnemann, and P.K.J. Robertson, The application of TiO2 photocatalysis for disinfection of water contaminated with pathogenic microorganisms: a review, Research on Chemi cal Intermediates 33 (2007) 359375. 16. D. Bahnemann, Photocatalytic water treatment: solar energy applications, Solar Energy 77 (2004) 445459. 17. D.Y. Goswami, A review of engineering developments of aqueous phase solar photocatalytic detoxification and disinfection processes, J. Sol. Energy Eng. 119 (1997) 101107. 18. S. Malato, P. Fernndez Ibez, M.I. Maldonado, J. Blanco, and W. Gernjak, Decontamination and disinfection of water by solar photocatalysis: Recent overview and trends, Catal. Today 147 (2009) 159. 19. T. Matsunaga, R. Tomoda, T. Nakajima, and H. Wake, Photoelectrochemical sterilization of microbial cells by semiconductor powders, FEMS Microbiol. Lett. 29 (1985) 211214.

PAGE 175

151 20. P. Maness, S. Smolinski, D.M. Blake, Z. Huang, E.J. Wolfrum and W.A. Jacoby, Bactericidal activity of photocatalytic TiO2 reaction: toward an understanding of its killing mechanism, Appl. Environ. Microbiol. 65 (1999) 40944098. 21. J. Marugn, R. van Grieken, C. Sordo, and C. Cruz, Kinetics of the photocatalytic disinfection of Escherichia coli suspensions, Appl. Catal. B 82 (2008) 2736. 22. O.A. McLoughlin, P.F. Ibanez, W. Gernjak, S.M. Rodri'guez, and L.W. Gill, Photocatalytic disinfection of water using low cost compound parabolic collectors, Solar Energy 7 7 (2004) 625633. 23. A.G. Rincon and C. Pulgarin, Photocatalytical inactivation of E. coli : effect of (continuous intermittent) light intensity and of (suspended fixed) TiO2 concentration, Appl. Catal. B 44 (2003) 263284. 24. A. G. Rincon and C. Pulgari n, 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, Appl. Catal. B 49 (2004) 99112. 25. C.P. Rodrigues, R.L. Ziolli, a nd J.R. Guimares, Inactivation of Escherichia coli in water by TiO2assisted disinfection using solar light, J. Braz. Chem. Soc. 18 (2007) 126134. 26. K. Sunada, Y. Kikuchi, K. Hashimoto, and A. Fujishima, Bactericidal and detoxification effects of TiO2 thin film photocatalysts, Environ. Sci. Technol. 32 (1998) 726728. 27. R.J. Watts, S. Kong, M.P. Orr, G.C. Miller, and B.E. Henry, Photocatalytic inactivation of coliform bacteria and viruses in secondary wastewater effluent, Water Res. 29 (1995) 95100. 28. C. Wei, W.Y. Lin, Z. Zainal, N.E. Williams, K. Zhu, A.P. Kruzic, et al., Bactericidal activity of TiO2 photocatalyst in aqueous media: toward a solar assisted water disinfection system, Environ. Sci. Technol. 28 (1994) 934938. 29. Y. S. Choi and B.W. Kim, Photocatalytic disinfection of E. coli in a UV/TiO2immobilised opticalfibre reactor, J. Chem. Technol. Biotechnol. 75 (2000) 11451150.

PAGE 176

152 30. M. Cho and J. Yoon, Measurement of OH radical CT for inactivating Cryptosporidium parvum using photo/ferrioxalate and photo/TiO2 systems, Appl. Microbiol. 104 (2008) 759766. 31. H. Ryu, D. Gerrity, J.C. Crittenden, and M. Abbaszadegan, Photocatalytic inactivation of Cryptosporidium parvum with TiO2 and low pressure ultraviolet irradiation, Water Res. 42 (2008) 15231530. 32. S. Navalon, M. Alvaro, H. Garcia, D. Escrig, and V. Costa, Photocatalytic water disinfection of Cryptosporidium parvum and Giardia lamblia using a fibrous ceramic TiO2 photocatalyst, Water Sci. Technol. 59 (2009) 639645. 33. A. Cooper, D.Y. Goswami, and S.S. Block, Solar photochemical detoxification and disinfection for water treatment in tropical developing countries, J. Adv. Oxidation Technol. 3 (1998) 151154. 34. S.S. Block, V.P. Seng, and D.W. Goswami, Chemically enhanced sunlight for killing bacteria, J. Sol. Energy Eng. 119 (1997) 8591. 35. D.Y. Goswami, Engineering of Solar Photocatalytic Detoxification and Disinfection in: K.W. Ber (Ed.), Advances in Solar Energy, ASES, 1995, pp. 165210. 36. D.Y. Goswami, S.K. Shar ma, G.D. Mathur, and C.K. Jotshi, Technoeconomic analysis of solar detoxification systems, J. Sol. Energy Eng. 119 (1997) 108113. 37. S. Malato, J. Blanco, A. Vidal, and C. Richter, Photocatalysis with solar energy at a pilotplant scale: an overview, Ap pl. Catal. B 37 (2002) 115. 38. S. Malato, J. Blanco, A. Vidal, P. Fernandez, J. Caceres, P. Trincado, et al., New large solar photocatalytic plant: set up and preliminary results, Chemosphere. 47 (2002) 235240. 39. Q. Li, R. Xie, Y.W. Li, E.A. Mintz, a nd J.K. Shang, Enhanced visible light induced photocatalytic disinfection of E. coli by carbonsensitized nitrogendoped titanium oxide, Environ. Sci. Technol. 41 (2007) 50505056.

PAGE 177

153 40. M.S. Wong, W.C. Chu, D.S. Sun, H.S. Huang, J.H. Chen, P.J. Tsai, et al., Visible light induced bactericidal activity of a nitrogen doped titanium photocatalyst against human pathogens, Appl. Environ. Microbiol. 72 (2006) 61116116. 41. J.A. RengifoHerrera, E. Mielczarski, J. Mielczarski, N.C. Castillo, J. Kiwi, and C. Pulgar in, Escherichia coli inactivation by N, S codoped commercial TiO2 powders under UV and visible light, Appl. Catal. B 84 (2008) 448456. 42. L. Caballero, K.A. Whitehead, N.S. Allen, and J. Verran, Inactivation of Escherichia coli on immobilized TiO2 using fluorescent light, J. Photochem. Photobiol. A 202 (2009) 9298. 43. P. Wu, R. Xie, J.A. Imlay, and J.K. Shang, Visible light induced photocatalytic inactivation of bacteria by composite photocatalysts of palladium oxide and nitrogen doped titanium oxide Appl. Catal. B 88 (2009) 576581. 44. J.C. Yu, W. Ho, J. Yu, H. Yip, P.K. Wong, and J. Zhao, Efficient visible light induced photocatalytic disinfection on sulfur doped nanocrystalline titania, Environmental Science & Technology 39 (2005) 11751179. 45. C. Karunakaran, G. Abiramasundari, P. Gomathisankar, G. Manikandan, and V. Anandi, Cudoped TiO2 nanoparticles for photocatalytic disinfection of bacteria under visible light, J. Colloid Interface Sci. 352 (2010) 68 74. 46. O.K. Dalrymple, E. Stefanakos, M.A. Trotz, and D.Y. Goswami, A review of the mechanisms and modeling of photocatalytic disinfection, Appl. Catal. B 98 (2010) 2738. 47. N. Serpone and A.V. Emeline, Suggested terms and definitions in photocatalysis and radiocatalysis, International Jou rnal of Photoenergy 4 (2002) 91131. 48. A. Mills and S. Le Hunte, An overview of semiconductor photocatalysis, J. Photochem. Photobiol. A 108 (1997) 135. 49. R. Memming, Semiconductor Electrochemistry. Weinheim: Wiley VCH, 2001. 50. M.R. Hoffmann, S.T. Martin, W. Choi, and D.W. Bahnemann, Environmental applications of semiconductor photocatalysis, Chem. Rev. 95 (1995) 6996.

PAGE 178

154 51. A. Wold, Photocatalytic properties of titanium dioxide (TiO2), Chem. Mater. 5 (1993) 280283. 52. M. Ni, M.K.H. Leung, D.Y.C. Leung, and K. Sumathy, A review and recent developments in photocatalytic water splitting using TiO2 for hydrogen production, Renewable and Sustainable Energy Reviews 11 (2007) 401425. 53. S. D. Mo and W.Y. Ching, Electronic and optical properties of thr ee phases of titanium dioxide: Rutile, anatase, and brookite, Physical Review B 51 (1995) 13023. 54. D. Reyes Coronado, G. Rodriguez Gattorno, M.E. Espinosa Pesqueira, C. Cab, R.d. Coss, and G. Oskam, Phase pure TiO2 nanoparticles: anatase, brookite and r utile, Nanotechnology (2008) 145605. 55. D. Gumy, C. Morais, P. Bowen, C. Pulgarin, S. Giraldo, R. Hajdu, et al., Catalytic activity of commercial of TiO2 powders for the abatement of the bacteria ( E. coli ) under solar simulated light: Influence of the is oelectric point, Appl. Catal. B 63 (2006) 7684. 56. C. Di Valentin, G. Pacchioni, and A. Selloni, Electronic structure of defect states in hydroxylated and reduced rutile TiO2(110) surfaces, Phys. Rev. Lett. 97 (2006) 1668034. 57. N. Daude, C. Gout, and C. Jouanin, Electronic band structure of titanium dioxide, Physical Review B 15 (1977) 3229. 58. D.S. Bhatkhande, V.G. Pangarkar, and A.A.C.M. Beenackers, Photocatalytic degradation for environmental applications a review, J. Chem. Technol. Biotechnol. 77 (2002) 102116. 59. S.Y. Chae, M.K. Park, S.K. Lee, T.Y. Kim, S.K. Kim, and W.I. Lee, Preparation of Size Controlled TiO2 Nanoparticles and Derivation of Optically Transparent Photocatalytic Films, Chem. Mater. 15 (2003) 33263331. 60. M. Gratzel, Phot oelectrochemical cells, Nature 414 (2001) 338344.

PAGE 179

155 61. J. M. Herrmann, Heterogeneous photocatalysis: fundamentals and applications to the removal of various types of aqueous pollutants, Catal. Today 53 (1999) 115129. 62. N. Serpone and E. Pelizzetti, Ph otocatalysis: Fundamentals and Applications John Wiley & Sons, Inc., New York, 1989. 63. C.C. Wong and W. Chu, The hydrogen peroxide assisted photocatalytic degradation of alachlor in TiO2 suspensions, Environ. Sci. Technol. 37 (2003) 23102316. 64. C.S. Turchi and D.F. Ollis, Photocatalytic degradation of organic water contaminants: mechanisms involving hydroxyl radical attack, J. Catal. 122 (1990) 178. 65. M. Klare, J. Scheen, K. Vogelsang, H. Jacobs, and J.A. Broekaert, Degradation of short chain alkyl and alkanolamines by TiO2and Pt/TiO2 assisted photocatalysis, Chemosphere 41 (2000) 353362. 66. P.K. Dutta, S.O. Pehkonen, V.K. Sharma, and A.K. Ray, Photocatalytic oxidation of arsenic(III): evidence of hydroxyl radicals, Environ. Sci. Technol. 39 (2005) 18271834. 67. C. Chen, P. Lei, H. Ji, W. Ma, J. Zhao, H. Hidaka, et al., Photocatalysis by titanium dioxide and polyoxometalate/TiO2 cocatalysts. Intermediates and mechanistic study, Environ. Sci. Technol. 38 (2004) 32937. 68. M. Cho, H. Chung, W. Choi, and J. Yoon, Linear correlation between inactivation of E. coli and OH radical concentration in TiO2 photocatalytic disinfection, Water Res. 38 (2004) 10691077. 69. WHO, Emerging issues in water and infectious disease 2003, World Health Organi zation: Geneva. 70. M.T. Magdigan and J.M. Martinko, Brock Biology of Microorganisms 11 ed., Pearson Education, Inc, Upper Saddle River, NJ, 2006.

PAGE 180

156 71. WHO, Guidelines for drinking water quality 3rd ed., World Health Organization, Genva, 2008. 72. R. Matu szewska, Protozoan pathogens of genus Cryptosporidium and Giardia. Part I. Occurrence in water environment and health risk, Rocz. Panstw. Zakl. Hig. 58 (2007) 569577. 73. K. Pond, Water Recreation and Disease. Plausibility of Associated Infections: Acute Effects, Sequelae and Mortality IWA Publishing/World Health Organization, London, 2005. 74. J.L. Rennecker, B.J. Marias, J.H. Owens, and E.W. Rice, Inactivation of Cryptosporidium parvum oocysts with ozone, Water Res. 33 (1999) 24812488. 75. M.W. LeChev allier, W.D. Norton, and R.G. Lee, Giardia and Cryptosporidium spp. in filtered drinking water supplies, Appl. Environ. Microbiol. 57 (1991) 26172621. 76. EPA, Drinking Water; National Primary Drinking Water Regulations; Total Coliforms (Including Fecal C oliforms and E. coli); Final Rule 1989, United States Environmental Protection Agency. 77. EPA, National Primary Drinking Water Regulations: Revisions to the Total Coliform Rule 2010, US Environmental Protection Agency. 78. H.A. Al Abadleh and V.H. Grass ian, Oxide surfaces as environmental interfaces, Surface Science Reports 52 (2003) 63161. 79. G.E. Brown Jr, Surface science: How minerals react with water, Science. 294 (2001) 6769. 80. H.P. Boehm, Acidic and basic properties of hydroxylated metal oxid e surfaces, Discuss. Faraday Soc. 52 (1971) 264275. 81. E.V. Stefanovich and T.N. Truong, Ab initio study of water adsorption on TiO2(110): molecular adsorption versus dissociative chemisorption, Chem. Phys. Lett. 299 (1999) 623629.

PAGE 181

157 82. H.C. William and A.C. Marcus, Brnsted reactions on oxide mineral surfaces and the temperaturedependence of their dissolution rates, Aquatic Sciences. 55 (1993) 304313. 83. A. Davis James and O. Leckie James, Speciation of adsorbed ions at the oxide/water interface Chem ical Modeling in Aqueous Systems, American Chemical Society, 1979, pp. 299317. 84. S. Yurdakal, V. Loddo, B. Bayarri Ferrer, G. Palmisano, V. Augugliaro, J. Gimenez Farreras, et al., Optical properties of TiO2 suspensions: Influence of pH and powder conce ntration on mean particle size, Ind. Eng. Chem. Res. 46 (2007) 76207626. 85. M. Predota, Z. Zhang, P. Fenter, D.J. Wesolowski, and P.T. Cummings, Electric double layer at the rutile (110) surface. 2. adsorption of ions from molecular dynamics and X ray experiments, J. Phys. Chem. B 108 (2004) 1206112072. 86. C. Kormann, D.W. Bahnemann, and M.R. Hoffmann, Photolysis of chloroform and other organic molecules in aqueous titanium dioxide suspensions, Environ. Sci. Technol. 25 (1991) 494500. 87. F. Gonzalez Caballero and V.N. Shilov, Electrical double layer at a colloid particle, Encyclopedia of Surface and Colloid Science: Second Edition, (2006) 1932 1936. 88. T. Imae, K. Muto, and S. Ikeda, The pH dependence of dispersion of TiO2 particles in aqueous surfactant solutions, Colloid Polym. Sci. 269 (1991) 4348. 89. Z.E. Allouni, M.R. Cimpan, P.J. Hl, T. Skodvin, and N.R. Gjerdet, Agglomeration and sedimentation of TiO2 nanoparticles in cell culture medium, Colloi ds Surf. B 68 (2009) 8387. 90. V. Cabuil, P. Leviz, C. Treiner, R. Tsekov, E. Evstatieva, K.W. Stckelhuber, et al., Stability of TiO2 suspensions in reactors for degradation of toxic pollutants in: V. Cabuil, P. Leviz, and C. Treiner (Eds.), Trends in Colloid and Interface Science XVII, Springer Berlin / Heidelberg, 2004, pp. 117120.

PAGE 182

158 91. R.A. French, A.R. Jacobson, B. Kim, S.L. Isley, R.L. Penn, and P.C. Baveye, Influence of ionic strength, pH, and cation valence on aggregation kinetics of titanium dio xide nanoparticles, Environ. Sci. Technol. 43 (2009) 13541359. 92. C. Sentein, B. Guizard, S. Giraud, C. Y, and F. Tngal, Dispersion and stability of TiO2 nanoparticles synthesized by laser pyrolysis in aqueous suspensions, J. Phys.: Conf. Ser. 170 (2009) 012013. 93. R.J. Watts, S. Kong, and W. Lee, Sedimentation and reuse of titanium dioxide: application to suspended photocatalyst reactors, J. Environ. Eng. 121 (1995) 730735. 94. D.C. Grahame, The electrical double layer and the theory of electrocapil larity, Chem. Rev. 41 (1947) 441501. 95. O. Stern, The theory of the electric double layer, Z. Elektrochem 30 (1924) 508. 96. R. Rodriguez, M.A. Blesa, and A.E. Regazzoni, Surface complexation at the TiO2(anatase)/aqueous solution interface: chemisorptio n of catechol. 177 (1996) 122131. 97. J. Lausmaa, P. Lfgren, and B. Kasemo, Adsorption and coadsorption of water and glycine on TiO2, J. Biomed. Mater. Res. 44 (1999) 227 242. 98. Y. X. Weng, L. Li, Y. Liu, L. Wang, and G.Z. Yang, Surface binding forms of carboxylic groups on nanoparticulate TiO2 surface studied by the interfacesensitive transient tripletstate molecular probe, J. Phys. Chem. B 107 (2003) 43564363. 99. P. Olivera, M. Patrito, and H. Sellers, Electronic Structure Calculations of Polyatomic Oxyanions Adsorbed on Metal Surfaces Interfacial electrochemistry: theory, experiment, and applications, Marcel Dekker, Inc. New York, 1999. 100. V.E. Kazarinov, V.N. Andreev, and A.P. Mayorov, Investigation of the adsorption properties of the TiO2 electrode by the radioactive tracer method, Journal of Electroanalytical Chemistry and Interfacial Electrochemistry. 130 (1981) 277285.

PAGE 183

159 101. G. Hornyi, Investigation of the specific adsorption of sulfa te ions on powdered TiO2, J. Colloid Interface Sci. 261 (2003) 580583. 102. M. Bostrm, V. Deniz, G.V. Franks, and B.W. Ninham, Extended DLVO theory: electrostatic and nonelectrostatic forces in oxide suspensions, Adv. Colloid Interface Sci. 123126 (2006) 515. 103. P. Demchick and A.L. Koch, The permeability of the wall fabric of Escherichia coli and Bacillus subtilis, J. Bacteriol. 178 (1996) 768773. 104. M.C.M. van Loosdrecht, J. Lyklema, W. Norde, and A.J.B. Zehnder, Bacterial adhesion: A physicoche mical approach, Microb. Ecol. 17 (1989) 115. 105. W.M. Oleary, Fatty acids of bacteria, Bacteriol. Rev. 26 (1962) 421&. 106. J.E. Cronan, Jr. and E.P. Gelmann, An estimate of the minimum amount of unsaturated fatty acid required for growth of Escherichia coli J. Biol. Chem. 248 (1973) 1188 1195. 107. K. Magnuson, S. Jackowski, C.O. Rock, and J.E. Cronan, Jr., Regulation of fatty acid biosynthesis in Escherichia coli Microbiol. Mol. Biol. Rev. 57 (1993) 522542. 108. J.S. Cox, D.S. Smith, L.A. Warren, and F.G. Ferris, Characterizing heterogeneous bacterial surface functional groups using discrete affinity spectra for proton binding, Environ. Sci. Technol. 33 (1999) 45144521. 109. R.E. Martinez, D.S. Smith, E. Kulczycki, and F.G. Ferris, Determination of intrinsic bacterial surface acidity constants using a Donnan shell model and a continuous pKa distribution method, J. Colloid Interface Sci. 253 (2002) 130139. 110. A. van der Wal, W. Norde, A.J.B. Zehnder, and J. Lyklema, Determination of the total charg e in the cell walls of Gram positive bacteria, Colloids Surf. B 9 (1997) 81100. 111. W. Jiang, A. Saxena, B. Song, B.B. Ward, T.J. Beveridge, and S.C.B. Myneni, Elucidation of functional groups on Gram positive and Gram negative bacterial surfaces using infrared spectroscopy, Langmuir 20 (2004) 1143311442.

PAGE 184

160 112. T.R. Neu, Significance of bacterial surface active compounds in interaction of bacteria with interfaces, Microbiol. Rev. 60 (1996) 151166. 113. A.T. Poortinga, R. Bos, W. Norde, and H.J. Busscher, Electric double layer interactions in bacterial adhesion to surfaces, Surface Science Reports. 47 (2002) 132. 114. A.T. Poortinga, Electric double layer interactions in bacterial adhesion and detachment Institute of Biomedical Materials Science and Applications, University of Groningen, PhD, 2001. 115. R. Sonohara, N. Muramatsu, H. Ohshima, and T. Kondo, Difference in surface properties between Escherichia coli and Staphylococcus aureus as revealed by electrophoretic mobility measurements, Biophys. Chem. 55 (1995) 273277. 116. H. Ohshima and T. Kondo, Relationship among the surface potential, Donnan potential and charge density of ionpenetrable membranes, Biophys. Chem. 38 (1990) 117122. 117. H. Ohshima, Electrophoresis of soft particles, Adv. Colloid Interface Sci. 62 (1995) 189235. 118. H.J. Busscher and A.H. Weerkamp, Specific and nonspecific interactions in bacterial adhesion to solid substrata, FEMS Microbiol. Lett. 46 (1987) 165173. 119. E. Wasserman and A.R. Felmy, Computation of the electrical double layer properties of semipermeable membranes in multicomponent electrolytes, Appl. Environ. Microbiol. 64 (1998) 2295 2300. 120. K. Makino, H. Ohshima, and T. Kondo, Surface potential of an ionpenetrable charged membrane, J. Theor. Biol. 125 (1987) 367368. 121. C.J. Van Oss, Interfacial forces in aqueous media M. Dekker, New York, 1994. 122. A. Luttge, L. Zhang, and K.H. Nealson, Mineral surfaces and their implications for microbial attachment: Results from Monte Carlo simulations and direct surface observations, Am. J. Sci. 305 (2005) 766 790.

PAGE 185

161 123. B.A. Jucker, H. Harms, S.J. Hug, and A.J.B. Zehnder, Adsorption of bacterial surface polysaccharides on mineral oxides is mediated by hydrogen bonds, Colloids Surf. B 9 (1997) 331343. 124. P Landini and A.J.B. Zehnder, The global regulatory hns gene negatively affects adhesion to solid surfaces by anaerobically grown Escherichia coli by modulating expression of flagellar genes and lipopolysaccharide production, J. Bacteriol. 184 (2002) 1522 1529. 125. T. Taguchi, H. Terui, H. Ohshima, and T. Kondo, Interaction between an ion penetrable particle and a solid particle. II. Criteria for heterocoagulation, Colloid & Polymer Science 268 (1990) 8387. 126. H. Terui, T. Taguchi, H. Ohshima, and T. Kondo, Interaction between an ion penetrable particle and a solid particle. 1. Electrical double layer interaction, Colloid & Polymer Science 268 (1990) 7682. 127. H. Ohshima and T. Kondo, Electrostatic interaction of an ionpenetrable sphere with a hard plate: Contribution of image interaction, J. Colloid Interface Sci. 157 (1993) 504508. 128. J. P. Hsu and B.T. Liu, Electrical interaction between two spherical particles covered by an ionpenetrable charged membrane, Chem. Phys. 236 (1998) 6376. 129. Y. C. Kuo, M.Y. Hsieh, and J. P. Hsu, Interactions between a particle covered by an ionpenetrable charged membrane and a charged surface: A modified GouyChapman theory, Langmuir 18 (2002) 27892794. 130. H. Chick, An investigation of the laws of disinfection, J. Hyg. (Lond). 8 (1908) 92158. 131. J.N. McClellan, Modeling chlorine decay and chlorination by product formation in water treatment and distribution, University of Massachusetts Amherst, 2000. 132. C.N. Haas, J. Joffe, U. Anmangandla, and J.C. Hornberger, Development and Validation of Rational Design Methods of Disinfection AWWA Research Foundation and American Water Works Association, Denver, CO, 1995.

PAGE 186

162 133. L.K. Weavers and G.B. Wickramanayake, Kinetics of inactivation of microorganisms in: S.S. Block (Ed.), Disinfection, sterilization, and preservation, Lippincott Williams & Wilkins, 2001, pp. 6578. 134. H.E. Watson, A note on the variation of the rate of disinfection with change in the concentration of the disinfectant, J. Hyg. (Lond) 8 (1908) 536542. 135. EPA, Alternative disinfections and oxidants guidance manual 1999, Office of Water. 136. A.G. Rincon, C. Pulgarin, N. Adler, and P. Peringer, Interaction between E. coli inactivation and DBP precursors -dihydroxybenzene isomers in the photocatalytic process of drinkingwater disinfection with TiO2, J. Photochem. Photobiol. A 139 (2001) 233241. 137. L.W. Hom, Kinetics of chlorine disinfection in an ecosystem, J. Sanitary Eng. Div. 98 (1972) 183194. 138. C. Wei, W.Y. Lin, Z. Z ainal, N.E. Williams, K. Zhu, A.P. Kruzic, et al., Bactericidal activity of TiO2 photocatalyst in aqueous media: Toward a solar assisted water disinfection system, Environ. Sci. Technol. 28 (2002) 934938. 139. M.G. Corradini M.D. Normand, and M. Peleg, Stochastic and deterministic model of microbial heat inactivation, J. Food Sci. 75 (2010) R59R70. 140. M. Peleg and K. Shetty, Modeling microbial populations with the original and modified versions of the continuous and discr ete logistic equations, Crit. Rev. Food Sci. Nutr. 37 (1997) 471 490. 141. M. Peleg, Modeling and simulation of microbial survival during treatments with a dissipating lethal chemical agent, Food Research International 35 (2002) 327 336. 142. M. Peleg and M.B. Cole, Reinterpretation of microbial survival curves, Crit. Rev. Food Sci. Nutr. 38 (1998) 353 380. 143. M. Peleg, M.D. Normand, and E. Damrau, Mathematical interpretation of dose response curves, Bull. Math. Biol. 59 (1997) 747761.

PAGE 187

163 144. M.A.J.S. van Boekel, On the use of the Weibull model to describe thermal inactivation of microbial vegetative cells, Int. J. Food Microbiol. 74 (2002) 139159. 145. B.F. Severin, M.T. Suidan, and R.S. Engelbrecht, Kinetic modeling of U.V. disinfection of water, Wa ter Res. 17 (1983) 16691678. 146. B.F. Severin, M.T. Suidan, and R.S. Engelbrecht, Series event kinetic model for chemical disinfection, J. Environ. Eng. 110 (1984) 430439. 147. C.N. Haas, A mechanistic kinetic model for chlorine disinfection, Environmental Science & Technology 14 (1980) 339340. 148. Z. Adamczyk, B. Siwek, M. Zembala, and P. Belouschek, Kinetics of localized adsorption of colloid particles, Adv. Colloid Interface Sci. 48 (1994) 151280. 149. P. Fernandez Ibanez, J. Blanco, S. Malato, and F.J. de las Nieves, Application of the colloidal stability of TiO2 particles for recovery and reuse in solar photocatalysis, Water Res. 37 (2003) 31803188. 150. M.R. Oberholzer, J.M. Stankovich, S.L. Carnie, D.Y.C. Chan, and A.M. Lenhoff, 2D and 3D in teractions in random sequential adsorption of charged particles, J. Colloid Interface Sci. 194 (1997) 138153. 151. B. Widom, Random sequential addition of hard spheres to a volume, J. Chem. Phys. 44 (1966) 38883894. 152. Z. Adamczyk, M. Zembala, B. Siwek, and P. Warszynski, Structure and ordering in localized adsorption of particles, J. Colloid Interface Sci. 140 (1990) 123137. 153. Z. Adamczyk, Kinetics of diffusioncontrolled adsorption of colloid particles and proteins, J. Colloid Interface Sci. 229 ( 2000) 477489. 154. Q. Yang, P. Ling Ang, M.B. Ray, and S.O. Pehkonen, Light distribution field in catalyst suspensions within an annular photoreactor, Chem. Eng. Sci. 60 (2005) 52555268.

PAGE 188

164 155. C.G. Hatchard and C.A. Parker, A new sensitive chemical actino meter. II. Potassium ferrioxalate as a standard chemical actinometer, Proceedings of the Royal Society of London. Series A, Mathematical and Physical Sciences 235 (1956) 518536. 156. H.J. Kuhn, S.E. Braslavsky, and R. Schmidt, Chemical actinometry, Pure Appl. Chem. 76 (2004) 21052146. 157. N.J. Bunce, J. LaMarre, and S.P. Vaish, Photorearrangement of azoxybenzene to 2hydroxyazobenzene: a convenient chemical actinometer, Photochem. Photobiol. 39 (1984) 531533. 158. L. Sun and J.R. Bolton, Determination of the quantum yield for the photochemical generation of hydroxyl radicals in TiO2 suspensions, J. Phys. Chem. 100 (1996) 41274134. 159. H. Zhang, G. Chen, and D.W. Bahnemann, Environmental Photo(electro)catalysis: Fundamental Principles and Applied Catal ysts, in: C. Comninellis and G. Chen (Eds.), Electrochemistry for the Environment, Springer. New York, 2010. 160. A. G. Rincon and C. Pulgarin, Effect of pH, inorganic ions, organic matter and H2O2 on E. coli K12 photocatalytic inactivation by TiO2: Implic ations in solar water disinfection, Appl. Catal. B 51 (2004) 283302. 161. W. Chu and C.C. Wong, The photocatalytic degradation of dicamba in TiO2 suspensions with the help of hydrogen peroxide by different near UV irradiations, Water Res. 38 (2004) 103743. 162. Y. Wang and C.s. Hong, Effect of hydrogen peroxide, periodate and persulfate on photocatalysis of 2chlorobiphenyl in aqueous TiO2 suspensions, Water Res. 33 (1999) 20312036. 163. J. Thiebaud, F. Thevent, and C. Fittschen, OH radicals and H2O2 m olecules in the gas phase near to TiO2 surfaces, J. Phys. Chem. C 114 (2010) 30823088. 164. M.C. Lee and W. Choi, Solid phase photocatalytic reaction on the soot/TiO2 interface: The role of migrating OH radicals, J. Phys. Chem. B 106 (2002) 11818 11822.

PAGE 189

165 165. Y. Murakami, E. Kenji, A.Y. Nosaka, and Y. Nosaka, Direct detection of OH radicals diffused to the gas phase from the UV irradiated photocatalytic TiO2 surfaces by means of laser induced fluorescence spectroscopy, J. Phys. Chem. B 110 (2006) 1680816811. 166. R. Roots and S. Okada, Estimation of life times and diffusion distances of radicals involved in xray induced DNA strand breaks or killing of mammalian cells, Radiat. Res. 64 (1975) 306320. 167. Y. Murakami, K. Endo, I. Ohta, A.Y. Nosaka and Y. Nosaka, Can OH radicals diffuse from the UV irradiated photocatalytic TiO2 surfaces? Laser induced fluorescence study, J. Phys. Chem. C 111 (2007) 1133911346. 168. I.M. Svishchev and A.Y. Plugatyr Hydroxyl radical in aqueous solution: Computer simulation, J. Phys. Chem. B 109 (2005) 41234128. 169. M.G. Campo and J.R. Grigera, Classical molecular dynamics simulation of the hydroxyl radical in water, J. Chem. Phys. 123 (2005. 170. A. Kupperman, Di ffusion model of the radiation chemistry of aqueous solutions in: G. Silini (Ed.), Radiation Research: Proceedings of the Third International Congress of Radiation Research, Wiley. New York, 1967, pp. 212234. 171. P. Vassilev, M.J. Louwerse, and E.J. Baerends, Ab initio molecular dynamics simulation of the OH radical in liquid water, Chem. Phys. Lett. 398 (2004) 212216. 172. M.G. Campo and J.R. Grigera, Classical molecular dynamics simulation of the hydroxyl radical in water, J. Chem. Phys. 123 (2005) 0845076. 173. B. Halliwell and J. Gutteridge, Free radicals in biology and medicine 2nd ed., Clarendon Press, Oxford, 1989. 174. W.A. Pryor (Ed.), Free radicals in biology Academic Press, Inc, New York, 1976. 175. R.T. Dean, S. Fu, R. Stocker, and M.J. Da vies, Biochemistry and pathology of radical mediated protein oxidation, Biochem. J. 324 (1997) 118.

PAGE 190

166 176. C.L. Hawkins and M.J. Davies, Generation and propagation of radical reactions on proteins, Biochimica et Biophysica Acta (BBA) Bioenergetics. 1504 ( 2001) 196219. 177. J.M.C. Gutteridge, Lipid peroxidation initiated by superoxide dependent hydroxyl radicals using complexed iron and hydrogen peroxide, FEBS Lett. 172 (1984) 245249. 178. Y.W. Cheng, R.C.Y. Chan, and P.K. Wong, Disinfection of Legionella pneumophila by photocatalytic oxidation, Water Res. 41 (2007) 842852. 179. Z. X. Lu, L. Zhou, Z.L. Zhang, W.L. Shi, Z. X. Xie, H. Y. Xie, et al., Cell damage induced by photocatalysis of TiO2 thin films, Langmuir 19 (2003) 87658768. 180. N. p. Huang, X. Min hua, C.w. Yuan, and Y. Rui rong, The study of the photokilling effect and mechanism of ultrafine TiO 2 particles on U937 cells, J. Photochem. Photobiol. A 108 (1997) 229233. 181. I.L. Stefan and F. Irwin, Superoxide and iron: partners in crime, I UBMB Life 48 (1999) 157161. 182. A. Carlioz and D. Touati, Isolation of superoxide dismutase mutants in Escherichia coli Is superoxide dismutase necessary for aerobic life, EMBO J. 5 (1986) 623630. 183. J. Kiwi and V. Nadtochenko, Evidence for the mec hanism of photocatalytic degradation of the bacterial wall membrane at the TiO2 interface by ATR FTIR and laser kinetic spectroscopy, Langmuir 21 (2005) 46314641. 184. J. Kiwi and V. Nadtochenko, New evidence for TiO2 photocatalysis during bilayer lipid peroxidation, J. Phys. Chem. B 108 (2004) 1767517684. 185. V.A. Nadtochenko, A.G. Rincon, S.E. Stanca, and J. Kiwi, Dynamics of E. coli membrane cell peroxidation during TiO2 photocatalysis studied by ATR FTIR spectroscopy and AFM microscopy, J. Photoche m. Photobiol. A 169 (2005) 131137.

PAGE 191

167 186. N. Porter, S. Caldwell, and K. Mills, Mechanisms of free radical oxidation of unsaturated lipids, Lipids 30 (1995) 277290. 187. D.R. Janero, Malondialdehyde and thiobarbituric acidreactivity as diagnostic indice s of lipid peroxidation and peroxidative tissue injury, Free Radic. Biol. Med. 9 (1990) 515540. 188. Z. Y. Jiang, A. Woollard, and S. Wolff, Lipid hydroperoxide measurement by oxidation of Fe2 + in the presence of xylenol orange. Comparison with the TBA assay and an iodometric method, Lipids 26 (1991) 853856. 189. D.J.W. Barber and J.K. Thomas, Reactions of radicals with lecithin bilayers, Radiat. Res. 74 (1978) 5165. 190. J.M. Joseph and C.T. Aravindakumar, Determination of rate constants for the reacti on of hydroxyl radicals with some purines and pyrimidines using sunlight, J. Biochem. Biophys. Methods 42 (2000) 115124. 191. G.V. Buxton, C.L. Greenstock, W.P. Helman, and A.B. Ross, Critical review of rate constants for reactions of hydrated electron, hydrogen atom and hydroxyl radicals (OH/O J. Phys. Chem. Ref. Data. 17 (1988) 513 817. 192. C. Minero, G. Mariella, V. Maurino, D. Vione, and E. Pelizzetti, Photocatalytic transformation of organic compounds in the presence of inorga nic ions. 2. Competitive reactions of phenol and alcohols on a titanium dioxide fluoride system, Langmuir 16 (2000) 89648972. 193. P. Calza and E. Pelizzetti, Photocatalytic transformation of organic compounds in the presence of inorganic ions, Pure Appl. Chem. 73 (2001) 18391848. 194. H.Y. Chen, O. Zahraa, and M. Bouchy, Inhibition of the adsorption and photocatalytic degradation of an organic contaminant in an aqueous suspension of TiO2 by inorganic ions, J. Photochem. Photobiol. A 108 (1997) 3744. 195. C. Guillard, E. Puzenat, H. Lachheb, A. Houas, and J. M. Herrmann, Why inorganic salts decrease the TiO2 photocatalytic efficiency, International Journal of Photoenergy 7 (2005) 19.

PAGE 192

168 196. M. Abdullah, G.K.C. Low, and R.W. Matthews, Effects of comm on inorganic anions on rates of photocatalytic oxidation of organic carbon over illuminated titanium dioxide, J. Phys. Chem. 94 (1990) 68206825. 197. C. Kormann, D.W. Bahnemann, and M.R. Hoffmann, Photolysis of chloroform and other organic molecules in aqueous TiO2 suspensions, Environ. Sci. Technol. 25 (1991) 494500. 198. M.B. McBride, Environmental chemistry of soils Oxford University Press, 1994. 199. D.A. Dzombak and F. Morel, Surface complexation modeling: hydrous ferric oxide Wiley, 1990. 200. Y. Ku, W.H. Lee, and W.Y. Wang, Photocatalytic reduction of carbonate in aqueous solution by UV/TiO2 process, J. Mol. Catal. A: Chem 212 (2004) 191196. 201. S.J. Hug and B. Sulzberger, In situ Fourier transform infrared spectroscopic evidence for the form ation of several different surface complexes of oxalate on TiO2 in the aqueous phase, Langmuir 10 (1994) 35873597. 202. P.A. Connor and A.J. McQuillan, Phosphate adsorption onto TiO2 from aqueous solutions: an in situ internal reflection infrared spectro scopic study, Langmuir 15 (1999) 2916 2921. 203. J. Kochany and E. Lipczynska Kochany, Application of the EPR spintrapping technique for the investigation of the reactions of carbonate, bicarbonate, and phosphate anions with hydroxyl radicals generated by the photolysis of H2O2, Chemosphere 25 (1992) 17691782. 204. R.W. Matthews, H.A. Mahlman, and T.J. Sworski, Elementary processes in the radiolysis of aqueous sulfuric acid solutions. Determinations of both GOH and GSO4, The Journal of Physical Chemistr y 76 (1972) 12651272. 205. C.Y. Chang, K.S. Yao, J.H. Lee, and C.H. Chen, Formation and calculation of hydroxyl radical in the optimal photocatalytic process using the Taguchi method, Environmental Informatics 5 (2007) 655663.

PAGE 193

169 206. I. Tejero, A. Gonzal ezLafont, J.M. Lluch, and L.A. Eriksson, Theoretical modeling of hydroxyl radical induced lipid peroxidation reactions, J. Phys. Chem. B 111 (2007) 56845693. 207. M.A. Cubillos, E.A. Lissi, and E.B. Abuin, Kinetics of lipid peroxidation in compartmental ized systems initiated by a water soluble free radical source, Chem. Phys. Lipids 104 (2000) 4956. 208. B.A. Wagner, G.R. Buettner, and C.P. Burns, Free radical mediated lipid peroxidation in cells: oxidizability is a function of cell lipid bis allylic h ydrogen content, Biochemistry. 33 (2002) 44494453. 209. Q. T. Li, M.H. Yeo, and B.K. Tan, Lipid peroxidation in small and large phospholipid unilamellar vesicles induced by water soluble free radical sources, Biochem. Biophys. Res. Commun. 273 (2000) 7276. 210. J.M.C. Gutteridge, The use of standards for malonyldialdehyde, Anal. Biochem. 69 (1975) 518526. 211. J.M.C. Gutteridge and B. Halliwell, The measurement and mechanism of lipid peroxidation in biological systems, Trends Biochem. Sci. 15 (1990) 129135. 212. S.K. Jain, The accumulation of malonyldialdehyde, a product of fatty acid peroxidation, can disturb aminophospholipid organization in the membrane bilayer of human erythrocytes, J. Biol. Chem. 259 (1984) 33913394. 213. K. Moore and L.J. Roberts, Measurement of lipid peroxidation, Free Radic. Res. 28 (1998) 659671. 214. T. Iwaoka, F. Tabata, and T. Takahashi, Lipid peroxidation and lipid peroxide detected by chemiluminescence, Free Radic. Biol. Med. 3 (1987) 329333. 215. A.G. Marr and J.L. Ingra ham, Effect of temperature on the composition of fatty acids in Escherichia coli J. Bacteriol. 84 (1962) 12601267. 216. C.O. Gill, Effect of growth temperature on the lipids of Pseudomonas fluorescens, J Gen Microbiol 89 (1975) 293298.

PAGE 194

170 217. J.E. Cronan, Jr., Phospholipid alterations during growth of Escherichia coli J. Bacteriol. 95 (1968) 20542061. 218. R.C. Weast, Handbook of Chemistry and Physics. 53 ed., Chemical Rubber Cleveland, 1972. 219. M.C. Hidalgo, G. Coln, and J.A. Navo, Modification of the physicochemical properties of commercial TiO2 samples by soft mechanical activation, J. Photochem. Photobiol. A 148 (2002) 341348. 220. and E. Hooper, Study of sorption properties of various titanium dioxide materials, Czech J. Phys. 49 (1999) 789795. 221. D.V. Bavykin and F.C. Walsh, Titanate and titania nanotubes: synthesis, properties and applications Royal Society of Chemistry, Cam bridge, 2010. 222. A. Savitzky and M.J.E. Golay, Smoothing and differentiation of data by simplified least squares procedures, Anal. Chem. 36 (1964) 16271639. 223. Z. Y. Jiang, J.V. Hunt, and S.P. Wolff, Ferrous ion oxidation in the presence of xylenol or ange for detection of lipid hydroperoxide in low density lipoprotein, Anal. Biochem. 202 (1992) 384389. 224. J.E. Cronan, Jr. and C.O. Rock, The presence of linoleic acid in Escherichia coli cannot be confirmed, J. Bacteriol. 176 (1994) 30693071. 225. A. Pal, S.O. Pehkonen, L.E. Yu, and M.B. Ray, Photocatalytic inactivation of Gram positive and Gram negative bacteria using fluorescent light, J. Photochem. Photobiol. A 186 (2007) 335341. 226. Y. Horie, D.A. David, M. Taya, and S. Tone, Effects of light i ntensity and titanium dioxide concentration on photocatalytic sterilization rates of microbial cells, Ind. Eng. Chem. Res. 35 (1996) 39203926. 227. M. Kaneko, M. Kaneko, and I. Okura (Eds.), Photocatalysis: science and technology Biological and Medical P hysics Series, Springer, New York, 2002.

PAGE 195

171 228. C.S. Turchi. Effect of light intensity on photocatalytic reaction in Potential applications of concentrated solar energy: proceedings of a workshop. 1991. Washington, D.C.: Commission on Engineering and Techni cal Systems (CETS). 229. T.A. Egerton and C.J. King, Influence of light intensity on photoactivity in titanium dioxide pigmented systems, J. Oil Col. Chem. Assoc. 62 (1979) 386391. 230. H.M. Coleman, M.I. Abdullah, B.R. Eggins, and F.L. Palmer, Photocatal ytic degradation of 17[beta] oestradiol, oestriol and 17[alpha] ethynyloestradiol in water monitored using fluorescence spectroscopy, Appl. Catal. B 55 (2005) 2330. 231. C. Lee, J. Barnett, and P.D. Reaven, Liposomes enriched in oleic acid are less susceptible to oxidation and have less proinflammatory activity when exposed to oxidizing conditions, J. Lipid Res. 39 (1998) 12391247. 232. S. Parra, J. Olivero, and C. Pulgarin, Relationships between physicochemical properties and photoreactivity of four bio recalcitrant phenylurea herbicides in aqueous TiO2 suspension, Appl. Catal. B 36 (2002) 7584. 233. C. Y. Chang, Y.H. Hsieh, L. L. Hsieh, K. S. Yao, and T.C. Cheng, Establishment of activity indicator of TiO2 photocatalytic reaction --Hydroxyl radical trapping method, J. Hazard. Mater. 166 (2009) 897 903. 234. C. Watson, I. Janik, T. Zhuang, O. Charva Pulsed electron beam water radiolysis for submicrosecond hydroxyl radical protein footprinting, Anal. Chem. 81 (2009) 24962505. 235. J. Lyklema, H.P. van Leeuwen, and M. Minor, DLVO theory, a dynamic re interpretation, Adv. Colloid Interface Sci. 83 (1999) 3369. 236. G.J. Liu, X.R. Zhang, L. McWilliams, J.W. Talley, and C.R. Neal, Influence of ionic strength, electrolyte type, and NOM on As(V) adsorption onto TiO2, Journal of Environmental Science and Health, Part A: Toxic/Hazardous Substances and Environmental Engineering. 43 (2008) 430 436.

PAGE 196

172 237. K. Suttiponparnit, J. Jiang, M. Sahu, S. Suvachittanont, T. Charinpanitkul, and P. Biswas, Role of surface area, primary particle size, and crystal phase on titanium dioxide nanoparticle dispersion properties, Nanoscale Research Letters. 6 (2010) 18. 238. A. Zita and M. Hermansson, Effects of ionic strength on bac terial adhesion and stability of flocs in a wastewater activated sludge system, Appl. Environ. Microbiol. 60 (1994) 30413048. 239. T. Saito, T. Iwase, J. Horie, and T. Morioka, Mode of photocatalytic bactericidal action of powdered semiconductor TiO2 on m utans streptococci, J. Photochem. Photobiol. B. 14 (1992) 369379. 240. P.R. Cullis and B. De Kruijff, The polymorphic phase behaviour of phosphatidylethanolamines of natural and synthetic origin. A 31P NMR study, Biochimica et Biophysica Acta (BBA) Biom embranes. 513 (1978) 3142. 241. P.L. Yeagle and A. Sen, Hydration and the lamellar to hexagonal(II) phase transition of phosphatidylethanolamine, Biochemistry. 25 (1986) 75187522. 242. A.M. Jentzsch, H. Bachmann, P. Frst, and H.K. Biesalski, Improved an alysis of malondialdehyde in human body fluids, Free Radic. Biol. Med. 20 (1996) 251256. 243. K.J. Dennis and T. Shibamoto, Gas chromatographic determination of malonaldehyde formed by lipid peroxidation, Free Radic. Biol. Med. 7 (1989) 187192. 244. A.F. Vikbjerg, T.L. Andresen, K. Jrgensen, H. Mu, and X. Xu, Oxidative stability of liposomes composed of docosahexaenoic acidcontaining phospholipids, J. Am. Oil Chem. Soc. 84 (2007) 631637. 245. C.M. Hart, J.K. Tolson, and E.R. Block, Supplemental fatty a cids alter lipid peroxidation and oxidant injury in endothelial cells, Am J Physiol Lung Cell Mol Physiol 260 (1991) L481488. 246. J. Cosgrove, D. Church, and W. Pryor, The kinetics of the autoxidation of polyunsaturated fatty acids, Lipids 22 (1987) 299304.

PAGE 197

173 247. M.J. Davies, The oxidative environment and protein damage, Biochimica et Biophysica Acta (BBA) Proteins & Proteomics. 1703 (2005) 93109.

PAGE 198

174 APPENDICES

PAGE 199

175 Appendix A: Computer Codes function [yprime]=rat eeq(t,y,k,r,Ia,mcat) %A function that contains the rate equations for photocatalytic disinfection % m[*OH]+[cell]l = [cell]d + x[BP] -------------(1) % [*OH]+[BP] = [BP] ----------------------------(2) % Since the states are passed in as a single vector, let % y(1) = [*OH], i.e., concentration of OH radicals % y(2) = [cell]l, i.e., concentration of live cells % y(3) = [BP], i.e., byproduct concentration % G = generation rate of *OH (uM/s) % k(1) = rxn rate constant for radicals with cell,% (L/uM/s) % k2 = rxn rate constant with byproducts (L/uM/s) % k(2)= order of rxn wrt [*OH] % k(3) = rxn constant wrt [*OH] % KQ1 = constant of quenching [HCO3] % KQ2 = constant of quenching [CL] % KQ3 = constant of quenching [HPO4] G = Ia*exp( 7*mcat*r);

PAGE 200

176 Appendix A: (Continued) KBP = 1e4; %L/mol KQ1=6e4; %L/mol KQ2=1e5; %L/mol KQ3=8e6; %L/mol HCO3=0.085;%mol/L CL=20*6.9*10^3;%mol/L HPO4=20*0.59*10^3;%mol/L % k(1)=1e5; % k(2)=1.5; % k(3)=1; R=1.535/2; pH = 7; H_conc = 10^ (pH); An_sum = H_conc*(KQ1*HCO3+KQ2*CL+KQ3 *HPO4); theta_BP = KBP*y(3)*1e3/(1+KBP*y(3)*1e 3+An_sum); theta_An = H_conc*(KQ1*HCO3+KQ2*CL+KQ3*HPO4)/(1+KBP*y(3)*1e 3+An_sum); yprime(1)= (G*(1 (theta_An) (theta_BP)) k(3)*((y(1)^k(2))*y(2)))*pi*R; yprime(2)= ( k(1)*((y(1)^k(2)))*y(2))*pi*R; yprime( 3) = (k(3)*((y(1)^k(2)))*y(2) G*theta_BP*5e2)*pi*R;

PAGE 201

177 Appendix A: (Continued) yprime = yprime(:); % This ensures that the vector returned is a column vector

PAGE 202

178 Appendix A: (Continued) function [yprime] = myfun2 (t,y,k,Ia,mcat) %A function that numerically integrates the rate equations with respect to reactor radius yprime=quadv(@(r)rateeq(t,y,k,r,Ia,mcat),0,0.7675,1e 6); yprime=yprime(:); end

PAGE 203

179 Appendix A: (Continued) function [C2]=myodefun(k,inv,Ia,mcat) %A numerical analysis fu nction that solves the set of ODE using a 5th order Runge Kutta method tdata=inv(:,2); inverse=inv(:,1); up=max(tdata); tt=0:0.1:up; % Start time x0 = [0 1 0] ;% Initial conditions options = []; [t,s] = ode45(@myfun2,tt,x0,options,k,Ia,mcat); OH = s( :,1); CELL = s(:,2); BP = s(:,3); ss=max(size(tdata)); Cmod = interp1(tt,CELL,tdata)/max(CELL); dlmwrite(Results \simulation\optional\BYPRODUCTS.txt',BP) dlmwrite(Results \simulation\optional\CELL.txt', CELL) dlmwrite(Results \simulation\optional\OH .txt', OH ) for i=1:ss Cexp(i)=1/inverse(i); C2(i)=Cmod(i)*inverse(i); end C2=C2(:); end

PAGE 204

180 Appendix A : (Continued) function myfitmodel() %Fits data to the mechanistic model developed by Dalrymple et al (2011) %t2 is the time %Cexp is the experimental data %E is the error associated with the data close(gcf) ext = '.xlsx'; files = {'CTRL'};% 'C161' 'C181' 'C183'}; filestr = strcat(files,ext); sheetstr ={'MID'};% 'MID' 'LOW'}; %Reads data directly from MS Excel files cellrange1 = {'M112:R118' 'M6:R12' 'M42:R48' 'M77:R83'}; cellrange2 = {'M177:R183' 'M83:R91'};% 'M6:R14' 'M45:R53' 'M83:R91'}; cellrange3 = {'M130:R141' 'M6:R14' 'M48:R56' 'M89:R97'}; colorset1 = {'ko' 'bs' 'rd' 'g^' 'mv'}; colorset2 = {'k 'b 'r 'g 'm'}; TiO2 = {' 0.01' 0.10' '0.25' '0.50'}; Intensity =[ 1 2 3 ]; TiO2Num=[ 0.01 0.10 0.25 0.50]; QY=0.03; %quantum yield for OH generation

PAGE 205

181 Appendix A: (Continued) Fs = [0.0019 0.0172 0.0193 0.02]; %fraction of adsorbed light filenum = max(size(filestr)); sheetnum = max(size(sheetstr)); op=opti mset('MaxFunEvals',20000,'MaxIter',10000); dlmwrite( Results \simulation\optional \ALL_DATA_coef.csv','') %creates blank file for f = 1:filenum filename=char(filestr(f)); for iii=1:sheetnum sheet=char(sheetstr(iii)); work={'Currently working on filename ' sheet '...'}; update = char(strcat(work)); disp(update); if strcmp(sheet,'HIGH')==1 cellrange = cellrange1; Io = 4.61e 5; else if strcmp(sheet,'MID')==1 cellrange = cellrange2; Io = 2.495e 5; else

PAGE 206

182 Appendix A: (Continued) cellrange = cellrange3; Io = 1.43e 5; end end file = char(files(f)); rangesize = max(size(cellrange)); for r=1:rangesize inverse=[]; yexp=[]; range = char(cellrange(r)); data=xlsread(filename,sheet,range); t2=data(:,1); Cexp = data(:,5); E = data(:,6); K=[0.01 1.5 4]; %initial guesses up = max(t2); ss = max(size(Cexp)); for zz=1:ss inverse(zz)=1/Cexp(zz); yexp(zz) = 1;

PAGE 207

183 Appendix A: (Continued) end inverse=inverse(:); inverse(:,1)=inverse; yexp=yexp(:); inverse(:,2)=t2; Ia=QY*Io*Fs(r)*1e3 mcat=TiO2Num(r) [x, resnorm] = lsqcurvefit(@myodefun,K,inverse,yexp,[1e6 1 1],[1e6 5 1e3],op,Ia,mcat); x resnorm [C]=myodefun2(x,inverse,Ia,mcat); t3=0:0.1:up; t4=t3(:); [D]=interp1(t4,C(:,1),t2); for i=1:ss Y1(i) = log10(Cexp(i)); Y2(i)= log10(D(i)); SSY1= Y1(i)^2; resY1(i)=(Y1(i)Y2(i))^2; SS(i)=(Cexp(i))^2;

PAGE 208

184 Appendix A: (Continued) res(i)=(Cexp(i) D(i))^2; end %Sum of squares in Cexp SSy = sum(SS) ((sum(Cexp))^2/ss); SSres = sum(res); SSyY1=sum(SSY1) ((sum(Y1))^2/ss); SSresY1=sum(resY1); %Goodness of fit parameters Rsq1 = 1 SSres/SSy; Rsq2 = 1 SSresY1/SSyY1; rms1 = sqr t(SSres/ss); rms2 = sqrt(SSresY1/ss); Rsq(1)=Rsq1; Rsq(2)=Rsq2; model_data = [t4 C(:,1)]; conc = char(TiO2(r)); csvfilename1 = char(strcat(file,'_',sheet,'_',conc,'model.txt')); dlmwrite(' \ALL_DATA_coef.csv',[Intensity(iii) TiO2Num(r) x Rsq1],' append')

PAGE 209

185 Appendix A: (Continued) savelocation1= char(strcat( \Results \simulation\optional \',file,' \',csvfilename1)); dlmwrite( savelocation1,model_data,'delimiter',' \t') %adds time variable to file %plots datacolor = char(colorset1(r)); fitcolor = char(colorset2(r)); hold on subplot (2,2,1), plot(t2,Cexp,datacolor,t4,C(:,1),fitcolor) axis([0 max(t2) min(Cexp) max(Cexp)]) xlabel('Time (min)') ylabel('C/C_{o}') hold on errorbar(t2,Cexp,E,datacolor) legend boxoff subplot(2,2,2) semilogy(t2,Cexp,datacolor,t4,C(:,1),fitcolor) hold on errorbar(t2,Cexp,E,datacolor) errorbarlogy;

PAGE 210

186 Appendix A: (Continued) xlabel('Time (min)') ylabel('log C/C_{o}') axis([0 max(t2) min(Cexp) max(Cexp)]) legend boxoff subplot(2,2,3) hold on plot(t4,C(:,2),fitcolor) xlabel('Time (min)') ylabel('Byproduct conc') legend boxoff subplot(2,2,4) hold on plot(t4,C(:,3),fitcolor) xlabel('Time (min)') ylabel('[OH] mol/ L') legend boxoff end imagename = char(strcat( Results \simulation\optional \',file,' \',sheet,'2')); saveas(gcf,imagename,'fig') saveas(gcf,imagename,'png')

PAGE 211

187 Appendix A: (Continued) close(gcf); %close figure window end end

PAGE 212

188 Appendix B: Fatty Acid Spectra Microbial ID, Inc. Volume: DATA File: E103226.46B Samp Ctr: 3 ID Number: 6051 Type: Samp Bottle: 71 Method: RCLIN6 Created: 3/22/2010 3:57:26 PM Created By: sstrauss (Sue Strauss) Sample ID: C USF10 03 (01Luria Broth Direct DL RT Response Ar/Ht RFact ECL Peak Name Perc Comment1 Comment2 0.6998 1.165E+9 0.014 ---6.6894 SOLVENT ---< min rt 0.7726 1710 0.012 ---7.1994 ---< min rt 1.5783 17592 0.008 1.070 12.0018 12:0 3.30 ECL deviates 0.002 Reference 0.001 1.8364 3599 0.009 1.028 12.9998 13:0 0.65 ECL deviates 0.000 Reference 0.003 1.9757 660 0.013 1.011 13.4882 12:0 3OH 0.12 ECL deviates 0.005 2.0733 1554 0.009 ---13.8307 ---2.1079 2758 0.007 ---13.9519 unknown 13.951 ---ECL deviates 0.001 2.1206 49471 0.009 0.995 13.9964 14:0 8.63 ECL deviates 0.004 Reference 0.001 2.2753 6966 0.012 ---14.5094 ---2.3662 825 0.008 0.972 14.8109 15:1 w8c 0.14 ECL deviates 0.003 2.4224 24657 0.009 ---14.9973 15:0 ---ECL deviates 0.003 2.4768 1465 0.009 ---15.1722 ---2.5818 50518 0.009 0.955 15.5101 Sum In Feature 2 8.46 ECL deviates 0.005 14:0 3OH/16:1 iso I 2.6825 74705 0.009 0.948 15.8344 Sum In Feature 3 12.4 ECL deviates 0.006 16:1 w7c/16:1 w6c 2.7097 1071 0.010 0.946 15.9221 16:1 w5c 0.18 ECL deviates 0.006 2.7339 217554 0.009 0.945 15.9999 16:0 36.0 ECL deviates 0.000 Reference 0.003 2.7633 2731 0.013 ---16.0942 ---2.8992 953 0.010 0.935 16.5297 15:0 3OH 0.16 ECL deviates 0.003 2.9881 1929 0.011 0.930 16.8147 17:1 w8c 0.31 ECL deviates 0.000 3.0186 81103 0.009 0.928 16.9126 17:0 cyclo 13.2 ECL deviates 0.002 3.0462 12116 0.009 0.927 17.0011 17:0 1.97 ECL deviates 0.001 Reference 0.002 3.2168 357 0.010 0.919 17.5524 16:0 3OH 0.06 ECL deviates 0.004 3.2796 2075 0.010 0.916 17.7552 Sum In Feature 5 0.33 ECL deviates 0.001 18:2 w6,9c/18:0 3.3079 73697 0.009 0.914 17.8467 Sum In Feature 8 11.8 ECL deviates 0.001 18:1 w7c 3.3256 740 0.010 0.914 17.9041 Sum In Feature 8 0.12 ECL deviates 0.002 18:1 w6c 3.3567 2772 0.010 0.912 18.0045 18:0 0.44 ECL deviates 0.005 Reference 0.000 3.4593 315 0.009 ---18.3434 ---3.4668 353 0.009 ---18.3683 ---3.4991 3233 0.012 ---18.4748 ---3.5665 1769 0.011 ---18.6977 ---3.6392 8911 0.010 0.901 18.9379 19:0 cyclo w8c 1.41 ECL deviates 0.006 3.6568 1223 0.010 0.900 18.9964 19:0 0.19 ECL deviates 0.004 Reference 0.014 3.7066 896 0.014 ---19.1642 ------50518 --------Summed Feature 8.46 12:0 aldehyde ? unknown 10.9525 ------------------16:1 iso I/14:0 3OH 14:0 3OH/16:1 iso I ---74705 --------Summed Feature 12.4 16:1 w7c/16:1 w6c 16:1 w6c/16:1 w7c ---2075 --------Summed Feature 0.33 18:0 ante/18:2 w6,9c 18:2 w6,9c/18:0 ---74437 --------Summed Feature 11.9 18:1 w7c 18:1 w6c ECL Deviation: 0.003 Reference ECL Shift: 0.006 Number Reference Peaks: 7 Total Response: 621153 Total Named: 601869

PAGE 213

189 Appendix B: (Continued) Percent Named: 96.90% Total Amount: 570076 min 0.5 1 1.5 2 2.5 3 3.5 4 pA 10 20 30 40 50 60 70 FID2 B, (E10322.646\B0036051.D) 0.700 0.773 1.578 1.836 1.976 2.073 2.108 2.121 2.275 2.366 2.422 2.477 2.582 2.683 2.710 2.734 2.763 2.899 2.988 3.019 3.046 3.217 3.280 3.308 3.326 3.357 3.459 3.467 3.499 3.566 3.639 3.657 3.707

PAGE 214

190 Appendix B: (Continued) Microbial ID, Inc. Volume: DATA File: E103226.46A Samp Ctr: 5 ID Number: 6052 Type: Samp Bottle: 72 Method: RCLIN6 Created: 3/22/2010 4:06:23 PM Created By: sstrauss (Sue Strauss) Sample ID: C USF1003 (02Oleic Direct DL RT Respons Ar/H RFact ECL Peak Name Perce Comment1 Comment2 0.7415 1.123E+ 0.02 ---6.6887 SOLVENT PEAK ---< min rt 1.6673 15276 0.01 1.069 12.001 12:0 3.53 ECL deviates 0.001 Reference 0.013 1.9354 2399 0.01 1.032 13.000 13:0 0.53 ECL deviates 0.000 Reference 0.012 2.0777 448 0.01 1.016 13.483 12:0 3OH 0.10 ECL deviates 0.000 2.1680 362 0.00 ---13.789 ---2.1802 630 0.00 ---13.831 ---2.2290 39485 0.01 1.001 13.996 14:0 8.53 ECL deviates 0.004 Reference 0.008 2.3884 5210 0.01 ---14.511 unknown ---ECL deviates 0.005 2.4820 364 0.00 0.978 14.813 15:1 w8c 0.08 ECL deviates 0.000 2.5394 17261 0.00 ---14.998 15:0 ---ECL deviates 0.001 2.5951 1097 0.00 ---15.173 ---2.7040 39696 0.00 0.962 15.514 Sum In Feature 8.25 ECL deviates 0.000 14:0 3OH/16:1 2.7934 6772 0.00 0.955 15.795 16:1 w9c 1.40 ECL deviates 0.005 2.8062 34168 0.00 0.955 15.835 Sum In Feature 7.05 ECL deviates 0.005 16:1 w7c/16:1 2.8347 399 0.01 0.953 15.924 16:1 w5c 0.08 ECL deviates 0.003 2.8586 165258 0.00 0.951 15.999 16:0 33.96 ECL deviates 0.000 Reference 0.009 2.8883 1192 0.01 ---16.093 ---2.9563 1102 0.01 ---16.306 ---3.0279 854 0.01 0.941 16.531 15:0 3OH 0.17 ECL deviates 0.002 3.1189 835 0.01 0.936 16.817 17:1 w8c 0.17 ECL deviates 0.002 3.1503 36377 0.01 0.934 16.916 17:0 cyclo 7.34 ECL deviates 0.001 3.1773 7301 0.00 0.932 17.001 17:0 1.47 ECL deviates 0.001 Reference 0.007 3.4285 90120 0.01 0.919 17.797 18:1 w9c 17.90 ECL deviates 0.003 3.4447 31781 0.00 0.919 17.848 Sum In Feature 6.31 ECL deviates 0.001 18:1 w7c 3.4938 1425 0.01 0.916 18.004 18:0 0.28 ECL deviates 0.004 Reference 0.007 3.6391 1518 0.01 ---18.475 ---3.7086 718 0.01 ---18.701 ---3.7718 8920 0.01 0.904 18.906 Sum In Feature 1.74 ECL deviates 0.019 19:0 cyclo 3.7804 5168 0.00 0.904 18.934 19:0 cyclo w8c 1.01 ECL deviates 0.002 3.7996 519 0.00 0.903 18.996 19:0 0.10 ECL deviates 0.003 Reference 0.005 ---39696 --------Summed 8.25 12:0 aldehyde ? unknown ------------------16:1 iso I/14:0 3OH 14:0 3OH/16:1 ---34168 --------Summed 7.05 16:1 w7c/16:1 w6c 16:1 w6c/16:1 ---8920 --------Summed 1.74 19:1w7c/19:1 w6c 19:1 ------------------19:0 cyclo ---31781 --------Summed 6.31 18:1 w7c 18:1 w6c ECL Deviation: 0.005 Reference ECL Shift: 0.009 Number Reference Peaks: 7 Total Response: 494184 Total Named: 487566 Percent Named: 98.66% Total Amount: 462899

PAGE 215

191 Appendix B: (Continued) min 0.5 1 1.5 2 2.5 3 3.5 4 pA 20 40 60 80 100 120 140 FID1 A, (E10322.646\A0056052.D) 0.742 1.667 1.935 2.078 2.168 2.180 2.229 2.388 2.482 2.539 2.595 2.704 2.793 2.806 2.835 2.859 2.888 2.956 3.028 3.119 3.150 3.177 3.429 3.445 3.494 3.639 3.709 3.772 3.780 3.800

PAGE 216

192 Appendix B: (Continued) Microbial ID, Inc. Volume: DATA File: E103226.46B Samp Ctr: 4 ID Number: 6053 Type: Samp Bottle: 73 Method: RCLIN6 Created: 3/22/2010 4:06:23 PM Created By: sstrauss (Sue Strauss) Sample ID: C USF1003 (03Palmitoleic Direct DL RT Respons Ar/H RFact ECL Peak Name Percen Comment1 Comment2 0.6998 1.163E+ 0.01 ---6.6888 SOLVENT PEAK ---< min rt 0.7730 2173 0.01 ---7.2011 ---< min rt 1.5381 657 0.00 ---11.820 unknown 11.825 ---ECL deviates 0.004 1.5786 17130 0.00 1.070 12.001 12:0 3.60 ECL deviates 0.001 Reference 1.8369 2366 0.00 1.028 13.000 13:0 0.48 ECL deviates 0.000 Reference 1.9758 536 0.01 1.011 13.487 12:0 3OH 0.11 ECL deviates 0.005 2.0733 2490 0.00 ---13.829 ---2.1077 2552 0.00 ---13.950 unknown 13.951 ---ECL deviates 0.001 2.1209 36016 0.00 0.995 13.996 14:0 7.04 ECL deviates 0.003 Reference 2.2760 6150 0.01 ---14.510 unknown 14.502 ---ECL deviates 0.005 2.3672 331 0.00 0.972 14.813 15:1 w8c 0.06 ECL deviates 0.001 2.4225 18266 0.00 ---14.996 15:0 ---ECL deviates 0.003 2.4772 1317 0.00 ---15.172 ---2.5820 45545 0.00 0.955 15.510 Sum In Feature 2 8.55 ECL deviates 0.005 14:0 3OH/16:1 2.6828 94182 0.00 0.948 15.834 Sum In Feature 3 17.56 ECL deviates 0.006 16:1 w7c/16:1 2.7105 575 0.00 0.946 15.923 16:1 w5c 0.11 ECL deviates 0.004 2.7342 198761 0.00 0.945 16.000 16:0 36.93 ECL deviates 0.000 Reference 2.7628 2821 0.01 ---16.091 ---2.8992 876 0.01 0.935 16.528 15:0 3OH 0.16 ECL deviates 0.004 2.9256 668 0.01 ---16.613 ---2.9959 1450 0.01 0.930 16.838 17:1 w7c 0.27 ECL deviates 0.003 3.0187 75624 0.00 0.928 16.912 17:0 cyclo 13.80 ECL deviates 0.003 3.0462 10821 0.00 0.927 17.000 17:0 1.97 ECL deviates 0.000 Reference 3.2797 2126 0.01 0.916 17.754 Sum In Feature 5 0.38 ECL deviates 0.002 18:2 3.3080 43366 0.00 0.914 17.845 Sum In Feature 8 7.80 ECL deviates 0.002 18:1 w7c 3.3570 2020 0.01 0.912 18.004 18:0 0.36 ECL deviates 0.004 Reference 3.4659 360 0.00 ---18.364 ---3.4986 3375 0.01 ---18.472 ---3.5667 1727 0.01 ---18.697 ---3.6390 3304 0.01 0.901 18.935 19:0 cyclo w8c 0.59 ECL deviates 0.004 3.6574 1300 0.01 0.900 18.996 19:0 0.23 ECL deviates 0.003 Reference 3.7065 958 0.01 ---19.162 ------45545 --------Summed Feature 8.55 12:0 aldehyde ? unknown ------------------16:1 iso I/14:0 3OH 14:0 3OH/16:1 ---94182 --------Summed Feature 17.56 16:1 w7c/16:1 w6c 16:1 w6c/16:1 ---2126 --------Summed Feature 0.38 18:0 ante/18:2 18:2 ---43366 --------Summed Feature 7.80 18:1 w7c 18:1 w6c ECL Deviation: 0.003 Reference ECL Shift: 0.005 Number Reference Peaks: 7 Total Response: 550047 Total Named: 536329 Percent Named: 97.51% Total Amount: 508649

PAGE 217

193 Appendix B: (Continued) min 0.5 1 1.5 2 2.5 3 3.5 4 pA 10 20 30 40 50 60 FID2 B, (E10322.646\B0046053.D) 0.700 0.773 1.538 1.579 1.837 1.976 2.073 2.108 2.121 2.276 2.367 2.423 2.477 2.582 2.683 2.710 2.734 2.763 2.899 2.926 2.996 3.019 3.046 3.280 3.308 3.357 3.466 3.499 3.567 3.639 3.657 3.707

PAGE 218

194 Appendix B: (Continued) Microbial ID, Inc. Volume: DATA File: E103226.46A Samp Ctr: 6 ID Number: 6054 Type: Samp Bottle: 74 Method: RCLIN6 Created: 3/22/2010 4:15:23 PM Created By: sstrauss (Sue Strauss) Sample ID: C USF1003 (04Linolenic Direct DL RT Respons Ar/H RFact ECL Peak Name Perce Comment1 Comment2 0.7323 167022 0.00 ---6.6278 ---< min rt 0.7411 1.13E+9 0.01 ---6.6861 SOLVENT PEAK ---< min rt 1.6677 13210 0.01 1.069 12.001 12:0 2.99 ECL deviates 0.002 Reference 0.014 1.9357 2489 0.00 1.032 12.999 13:0 0.54 ECL deviates 0.000 Reference 0.013 2.0783 486 0.01 1.016 13.484 12:0 3OH 0.10 ECL deviates 0.001 2.1807 1205 0.00 ---13.831 ---2.2158 2077 0.00 ---13.950 unknown 13.951 ---ECL deviates 0.001 2.2294 37707 0.00 1.001 13.996 14:0 7.98 ECL deviates 0.004 Reference 0.010 2.3890 5513 0.01 ---14.511 unknown 14.502 ---ECL deviates 0.004 2.4828 653 0.00 0.978 14.814 15:1 w8c 0.14 ECL deviates 0.001 2.5401 18075 0.00 ---14.999 15:0 ---ECL deviates 0.000 2.5956 1118 0.00 ---15.173 ---2.7046 37923 0.00 0.962 15.515 Sum In Feature 2 7.72 ECL deviates 0.000 14:0 3OH/16:1 iso 2.8069 54115 0.00 0.955 15.836 Sum In Feature 3 10.93 ECL deviates 0.004 16:1 w7c/16:1 2.8351 774 0.01 0.953 15.924 16:1 w5c 0.16 ECL deviates 0.004 2.8590 169755 0.00 0.951 15.999 16:0 34.18 ECL deviates 0.000 Reference 0.010 2.8894 2400 0.01 ---16.095 ---2.9600 1236 0.02 ---16.316 ---3.0280 826 0.01 0.941 16.530 15:0 3OH 0.16 ECL deviates 0.003 3.1196 1549 0.01 0.936 16.817 17:1 w8c 0.31 ECL deviates 0.003 3.1506 60288 0.01 0.934 16.915 17:0 cyclo 11.92 ECL deviates 0.000 3.1783 8822 0.00 0.932 17.002 17:0 1.74 ECL deviates 0.002 Reference 0.011 3.4169 8936 0.00 0.920 17.758 Sum In Feature 5 1.74 ECL deviates 0.002 18:2 w6,9c/18:0 3.4296 8316 0.00 0.919 17.798 18:1 w9c 1.62 ECL deviates 0.004 3.4447 77610 0.01 0.919 17.846 Sum In Feature 8 15.09 ECL deviates 0.001 18:1 w7c 3.4644 669 0.01 0.918 17.908 Sum In Feature 8 0.13 ECL deviates 0.007 18:1 w6c 3.4941 4036 0.01 0.916 18.003 18:0 0.78 ECL deviates 0.003 Reference 0.008 3.5508 733 0.01 0.914 18.187 17:0 iso 3OH 0.14 ECL deviates 0.006 3.5703 1892 0.01 ---18.250 ---3.6080 1071 0.01 ---18.372 ---3.6395 3246 0.01 ---18.474 ---3.7091 1682 0.01 ---18.700 ---3.7823 7355 0.01 0.904 18.938 19:0 cyclo w8c 1.41 ECL deviates 0.006 3.8005 1157 0.01 0.903 18.997 19:0 0.22 ECL deviates 0.003 Reference 0.002 3.8495 1008 0.01 ---19.159 ------37923 --------Summed Feature 7.72 12:0 aldehyde ? unknown 10.9525 ------------------16:1 iso I/14:0 3OH 14:0 3OH/16:1 iso ---54115 --------Summed Feature 10.93 16:1 w7c/16:1 w6c 16:1 w6c/16:1 ---8936 --------Summed Feature 1.74 18:0 ante/18:2 18:2 w6,9c/18:0 ---78278 --------Summed Feature 15.22 18:1 w7c 18:1 w6c ECL Deviation: 0.003 Reference ECL Shift: 0.010 Number Reference Peaks: 7 Total Response: 512267 Total Named: 497409 Percent Named: 97.10% Total Amount: 472484

PAGE 219

195 ABOUT THE AUTHOR Omatoyo Kofi Dalrymple attended the Stewartville Primary School and then went on to high school education at St. Stanislaus College from 19911996. He began studies at the University of Guyana in Civil Engineering the same year he graduated high school and completed an engineering degree with distinction in 2001. In 2002, he was awarded a CIDA scholarship to re ad for an MSc in Natural Resources Management with a specialization in Climate Change at the University of the West Indies Cave Hill, Barbados He then lived and worked in Barbados for a short while. Kofi moved to Tampa, Florida in 2005 to pursue a Doctor ate in Civil & Environmental Engineering at the University of South Florida. He is very interested in forming alliances to work in areas of physical planning and infrastructural development, environment and natural resources management, water and sanitatio n, and youth development in the Caribbean.


xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam 22 Ka 4500
controlfield tag 007 cr-bnu---uuuuu
008 s2011 flu ob 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0004895
035
(OCoLC)
040
FHM
c FHM
049
FHMM
090
XX9999 (Online)
1 100
Dalrymple, Omatoyo Kofi.
0 245
Mechanistic modeling of photocatalytic water disinfection
h [electronic resource] /
by Omatoyo Kofi Dalrymple.
260
[Tampa, Fla] :
b University of South Florida,
2011.
500
Title from PDF of title page.
Document formatted into pages; contains 219 pages.
Includes vita.
502
Disseration
(Ph.D.)--University of South Florida, 2011.
504
Includes bibliographical references.
516
Text (Electronic dissertation) in PDF format.
520
ABSTRACT: The main goal of this research was to develop a mechanism-based model for photocatalytic disinfection of bacteria in water using suspended catalyst particles in batch reactors. The photocatalytic disinfection process occurs as a semiconductor photocatalyst, most commonly titanium dioxide (TiO2), is irradiated with light of wavelength less than 380 nm to produce hydroxyl radicals and other highly reactive oxidants which can inactivate microorganisms. Photocatalytic disinfection involves a complex interaction of many fundamental mechanisms such as light absorption and scattering by semiconductor particles, electrochemical surface reactions, and heterogeneous colloidal stability. Current models, based largely on chemical reacting systems, do not adequately account for these fundamental mechanisms. Even the Langmuir model developed for heterogeneous systems cannot describe the interactions of such large colloidal particles. As a result, it is difficult to assess the combined effects of many important factors which go into the design of a photocatalytic disinfection system. A mechanistic modeling approach is desirable because it provides a framework to understand the influence of many important parameters on the disinfection process. It requires a description of the physical properties of the catalyst, the nature of the suspending electrolyte solution, the physical and chemical properties of the cell surface, and the energetic aspects that influence the interaction of the particles. All these aspects are interrelated. While it is customary to envision the adsorption of reactants unto a catalyst surface, for photocatalytic disinfection involving suspended catalyst particles, multiple catalyst particles adhere to the bacterial surface. In this work a mechanistic model has been developed that simulates the effect of light intensity and catalyst concentration on the disinfection process. The simulations show good agreement with the experimental data for stable colloidal suspensions, that is, suspensions in which rapid aggregation of cells and TiO2 do not occur. Increased disinfection rates and high levels of inactivation can be achieved by maintaining a relatively low catalyst-to-microbe ratio while maximizing the light intensity. The influence of pH and ionic strength on the disinfection process have been included in the model, but these are only expected to be accurately predicted when the solution remains stable.
538
Mode of access: World Wide Web.
System requirements: World Wide Web browser and PDF reader.
590
Advisor:
Goswami Trotz, Dharendra Maya Y A.
Advisor:
Goswami Trotz, Dharendra Maya Y A.
653
Double Layer
Drinking Water
E. Coli
Solar Applications
Titanium Dioxide
690
Dissertations, Academic
z USF
x Environmental Engineering Microbiology
Doctoral.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.4895