1 COUNTER FLOW SILICA TITANIA PHOTOCATALYTIC REACTOR FOR THE SIMULTANEOUS TREATMENT OF AIR AND WATER CONTAMINATED WITH VOLATILE ORGANIC COMPOUNDS By CHRISTINA AKLY A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVE RSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2009
2 2009 Christina Akly
3 To my parents Bethsy and Salomon
4 ACKNOWLEDGMENTS I would like to thank Dr. Paul A. Chadik, my committee chair and advisor. If I had to name one person from whom I have learned the most about engineering and scientific research, he would be that person His teaching has always been challenging, h is guidanc e and support has always gone beyond expectations and his advice has always encouraged me to go the extra mile. I would like to thank both Dr. Chadik and Dr. David Mazyck for encouraging me to pursue a doctorate degree. Without their advice and support, I probably would have not followed this path, which has been an incredibly rewarding learning experience. I a m also thankful to my other advisory committee members, Dr. Chang Yu Wu and Dr. Spyros Svoronos for their guidance and suggestions during the complet ion of this work. I attribute my success to the many teachers I have learned from throughout my years at the University of Florida I would like to thank some of my fellow researchers at the E nvironmental Engineering Department in UF for providing advice, lab support and in general good research discussions that have significantly contributed to the progress of my work : Mauricio Arias, Felipe Behrens, Gordon Brown, Heather Byrne, Tim English, Heather Fitzpatrick, Beau Kosted, Jennifer Stokke, and Mike Witwe r. Also, I would like to give special thanks to Dr. Hwidong Kim at UF and Rick Loftis at Mazyck Technology Solutions for providing analytical support. They have always gone out of their way to teach me and help me out with the GC, so I a m gratef ul for thei r time and patience. Randy Switt and Dr. Ben Koopman provided important a dvice related to my model, so I a m grateful for their help. Very importantly I would like to thank my family for their unconditional love and support, especially my father who is has always encourage d me to follow my dreams, my friends for their continuous support, even from many miles away; my fianc, Gordon, for always being there for me. Thank you all for listening and being there
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS .................................................................................................................... 4 LIST OF TABLES ................................................................................................................................ 8 LIST OF FIGURES ............................................................................................................................ 10 ABSTRACT ........................................................................................................................................ 15 CHAPTER 1 INTRODUCTION ....................................................................................................................... 17 2 LITERATURE REVIEW ........................................................................................................... 23 H eterogeneous Photocatalysis .................................................................................................... 23 Titanium Dioxide as Catalyst ..................................................................................................... 26 Catalyst Support: Silica Titania Composites ............................................................................. 28 Photocatalysis of Organic Compounds with TiO2 .................................................................... 33 Photocatalytic Oxidation (PCO) of Toluene with TiO2: Mechanisms and Intermediate Formation ................................................................................................................................. 36 Gaseous and Aqueous Phase Toluene Kinetics ......................................................................... 42 Effect of Initial Concentration ............................................................................................ 47 Effect of Light Intensity ...................................................................................................... 48 Effect of Superficial Velocity ............................................................................................. 49 Effect of Temperature .......................................................................................................... 49 Effect of Relative Humidity ................................................................................................ 50 Mass Transfer .............................................................................................................................. 51 Equilibrium Partitioning of Volatile Organ ic Compounds between Air and Water ....... 51 Mass Transfer Principles ..................................................................................................... 52 Mass Transfer Operations: Packed Tower Aerator (PTA) ....................................................... 54 3 MATERIALS AND METHODS ............................................................................................... 63 Pilot -Scale Reactor: Two -Phase Photcatalytic Oxidation Tower (TPOT) .............................. 63 Bench -Scale Reactor for Gas Phase Studies .............................................................................. 65 UV Irradiance Distribution in the Reactor ................................................................................. 67 Synthesis of the Silica Titania Composites (STCs) .................................................................. 71 Characterization of the STCs ...................................................................................................... 72 Analytical Methods for Toluene and Oxidation Byproduc ts ................................................... 74 4 CHARACTERIZATION OF SILICA TITANIA COMPOSITES .......................................... 88 5 GAS PHASE TOLUENE DEGRADATION STUDIES .......................................................... 93
6 Bench -Scale Studies .................................................................................................................... 93 Gas Phase Toluene Studies Using Dry STCs ..................................................................... 93 Adsorption of toluene in a packed bed reactor ........................................................... 93 PCO of toluene: Effect of water vapor ....................................................................... 94 PCO of toluene: Effect of space time ( ) .................................................................... 95 Kinetic analysis ............................................................................................................ 96 Gas Phase Toluene Studies Using Wetted STCs ............................................................. 100 Toluene adsorpt ion in a batch reactor ....................................................................... 100 Toluene adsorption in a continuous flow packed-bed reactor ................................. 102 Pilot -Scale Studies .................................................................................................................... 103 Toluene Adsorption Using TPOT ..................................................................................... 104 Adsorption of toluene using dry STCs ...................................................................... 104 Ads orption of toluene using wetted STCs ................................................................ 105 Toluene PCO Using TPOT ............................................................................................... 106 PCO of toluene using dry STCs ................................................................................ 106 PCO of toluene using wetted STCs ........................................................................... 109 6 AQUEOUS PHASE TOLUENE DEGRADATION STUDIES ............................................ 121 Bench -Scale Studies .................................................................................................................. 121 Pilot -Scale Studies .................................................................................................................... 123 Adsorption of Toluene Using TPOT ................................................................................ 123 PCO of Toluene Using TPOT ........................................................................................... 124 TPOT Hydrodynamics ...................................................................................................... 124 Liquid -Solid Mass Transfer Coefficient ( KLSaC) ............................................................. 130 7 SIMULTANEOUS GAS AND AQUEOUS PHASE TOLUENE DEGRADATION USING THE PILOT SCALE REACTOR .............................................................................. 143 GasLiquid Mass Transfer Coefficient ( KGLaw) ...................................................................... 143 Simultaneous Two-Phase Degradation Studies ....................................................................... 147 8 MATHEMATICAL MODELING AND SIMULATION OF THE TPOT ........................... 165 Model Development .................................................................................................................. 165 Model Simulation ...................................................................................................................... 169 Effects of Operating Parameters on the TPOT Performance ................................................. 171 9 CONCLUSIONS ....................................................................................................................... 184 APPENDIX A PRELIMINARY STUDIES ..................................................................................................... 189 Studies on Systems Integrity ................................................................................................... 189 System Losses .................................................................................................................... 189 Losses Due to Photolysis ................................................................................................... 189
7 Stripping ..................................................................................................................................... 190 Absorption ................................................................................................................................. 192 B ASSESSMENT OF LEACHING OF NANOMATERIALS ................................................. 197 LIST OF REFERENCES ................................................................................................................. 200 BIOGRAPHICAL SKETCH ........................................................................................................... 208
8 LIST OF TABLES Table page 2 1 Oxidation power of various species commonly used in treatment applications ................ 59 2 2 Properties of Degussa P25 ..................................................................................................... 60 3 1 Calculated average UV intensity through the commingled packing at different annuli distances from the UV lamps. ............................................................................................... 77 3 2 Chemicals used in the STC formulation to produce one batch of 140 STC with 30% by mass titania loading. ................................................................................................. 77 3 3 Conditions of the GC/MS and Purge & Trap used for the analysis of gas and aqueous phase toluene. ......................................................................................................................... 78 4 1 Properties of STCs ................................................................................................................. 90 4 2 Phosphorus content from ICP AES results and accessible TiO2 surface area................. 90 5 1 Summary of experimental conditions for the bench-scale gas phase PCO experiments using dry STCs ..................................................................................................................... 110 5 2 Mears ( CM) and Weisz Prater ( CWP) criteria for the determination of mass trans fer influences. ............................................................................................................................. 110 5 3 Experimental conditions for the adsorption experiments using dry STCs in the TPOT 111 6 1 Langmuir and F reundlich isotherms fitting parameters for the aqueous phase toluene adsorption data obtained in batch experiments. ................................................................. 134 6 2 Mean residence time and dispersion coefficients for the high flow ri ngs only packing and commingled packing obtained from the tracer tests. .................................................. 134 6 3 Rate constants and Mears ( CM) criterion for the determination of external mass transfer influences in the aqueou s phase.. .......................................................................... 135 6 4 Experimental conditions for the aqueous phase PCO experiments in the TPOT. ........... 135 7 1 Individual and averaged properties for the two different packing materials used to pack the TPOT. ..................................................................................................................... 153 7 2 Summary of operational conditions and toluene removals for the simultaneous two phase experiments. ............................................................................................................... 153 8 1 Summary of correlations used to determine the parameters involved in the solution to the differential equations developed to model the TPOT. ................................................. 174
9 8 2 Comparison of the actual net toluene removal to the net removal predicted by the two -phase model using different calibration methods. ...................................................... 175 8 3 Comparison of the mass transfer coe fficients calculated using the correlations developed for the aqueous phase and those found by calibrating the model. .................. 175 8 4 Operating conditions selected for the simulation of the systems p erformance. .............. 175 8 5 Reactor characteristics for the end polishing sections of the treatment system. .............. 175
10 LIST OF FIGURES Figure page 1 1 Conceptual design of the photocatalytic system .................................................................. 22 2 1 Energy band diagram of a spherical titania particle showing the basic reaction mechan ism of TiO2 photocatalysis. ....................................................................................... 60 2 2 Structures and properties of rutile and anatase titanium dioxide ........................................ 61 2 3 Linkages of SiO2 tetr ahedras ................................................................................................. 61 2 4 Silanol groups on the silica surface ....................................................................................... 62 2 5 Diagram describing the equilibrium partitioning of a contaminant betw een the air and water phases using the two film theory. ........................................................................ 62 3 1 CAD drawings of the reactor. ................................................................................................ 78 3 2 Dimensions of the TPOT reacto r sh own in horizontal configuration ................................. 79 3 3 Picture of the two different packing materials that make up the commingled packing in the TPOT.. .......................................................................................................................... 79 3 4 Photograph of the reactor (TPOT) during a typical experimental run. ............................... 80 3 5 Emission spectrum of UV C lamps: A) ordinary and B) non -ozone producing UV C lamps. ...................................................................................................................................... 81 3 6 Reactor setup for a typical experimental run ........................................................................ 81 3 7 Gas phase experimental setup and reactor. ........................................................................... 82 3 8 UV irradiance as a function of lamp length measured in ambient air with the UV radiometer sensor placed at 0.95cm and 6.25 cm from the lamp. ....................................... 82 3 9 Top view of th e box setup used to measure the UV intensity as a function of different packing materials at different X distances from the lamp. .............................................. 83 3 10 UV irradiance as a function of distance to the lamp through ambient air using the set up of Figure 3 9 for different extents of reflective surfaces. ............................................... 83 3 11 UV intensity as a function of the distance from the UV lamp through different packing m aterials. ................................................................................................................... 84 3 12 Enlarged version of Figure 311 ............................................................................................ 84
11 3 13 Actual placement of the five UV lamps inside the TPOT measured after p acking the reactor with the commingled packing. .................................................................................. 85 3 14 Fit of the data obtained for UV intensity through the commingled packing. ..................... 85 3 15 UV irradiance distribution inside the reactor.. ..................................................................... 86 3 16 Drying schedule for cylindrical STCs of 9 mm in diameter by 8 mm in height used to pack the TPOT ........................................................................................................................ 87 4 1 Typical nitrogen adsorption/desorption isotherm for the STCs used to pack the pilot scale reactor. ........................................................................................................................... 91 4 2 Typical pore size distributions of the STCs used to pack the pilot -scale reactor. .............. 91 4 3 Comparison of the accessible titanium dioxide surface area for different STC shapes. ... 92 5 1 Gas phase adsorption of toluene followed by PCO using dry STCs and low relative humidity conditions in the bench scale reactor .................................................................. 111 5 2 Normalized effluent toluene concentration during simultaneous adsorption and PCO experiments at different water vapor concentrat ions as a function of run time ............... 112 5 3 Normalized effluent toluene concentrations as a function of run time at v arious space times ...................................................................................................................................... 112 5 4 Linear regression of the Langmuir Hinshelwood model ................................................... 113 5 5 Plan view of the experimental setup of the batch experiments to assess adsorption under different STCs wetting conditions. ........................................................................... 113 5 6 Gas phase adsorption of water vapor and toluene to STCs under different wetting conditions. ............................................................................................................................. 114 5 7 Adsorption of gas phase toluene to STCs under different wetting conditions in a continuous flow reactor. ...................................................................................................... 114 5 8 Adsorption breakt hrough curves for toluene on the dry STCs in the commingled packing of the pilot -scale reactor. ....................................................................................... 1 15 5 9 Adsorption breakthrough curves for water vapor on the dry STCs in the commingled packing of the pilot -scale reactor. ....................................................................................... 115 5 10 Comparison of gas phase toluene adsorption using the TPOT with dry versus pre wetted STCs. ......................................................................................................................... 116 5 11 Adsorption of gas phase toluene followed by PCO in the TPOT using dry STCs and ambient relative humidity. ................................................................................................... 117
12 5 12 Conversion of toluene by PCO in TPOT for 3 studies performed using the same initial conditions: .................................................................................................................. 118 5 13 Conversion of gas phase toluene by PCO in the TPOT after catalyst regeneration. ....... 118 5 14 W ater vapor concentration profile in the TPOT at various run times for the experiment corresponding to the 2nd regeneration study.. ................................................. 119 5 15 Relative output of the UV C lamps used in this resea rch as a function of temperature. 120 5 16 Effect of wetting of the packing on gas phase toluene removal by PCO in the TPOT. .. 120 6 1 Removal of toluene due to adsorption in batch experiments as a function of STCs mass for different initial aqueous phase toluene concentrations. ...................................... 135 6 2 Adsorption isotherm for the aqueou s phase toluene STCs system. ............................... 136 6 3 Aqueous phase toluene adsorption for different liquid loading rates. .............................. 136 6 4 Toluene remo val as a function of inlet toluene concentration in the TPOT at a flowrate of 3.8 L/min. .......................................................................................................... 137 6 5 Aqueous phase toluene removal in the TPOT as a function of flowrate. ......................... 137 6 6 Cumulative RTD ( F(t) ) and RTD (E(t) ) functions obtained from the tracer test analysis for the TPOT at different flowrates and packed with diffe rent packing styles .. 138 6 7 Dispersion coefficients for the TPOT with commingled packing obtained from the tracer tests (data) and determined by the empirical correlation fitted to the data. ........... 139 6 8 Mean residence time as a function of liquid loading rate for the TPOT packed with the commingled packing. ..................................................................................................... 139 6 9 Aqueous phase toluene concentrations predicted using the PFD model as a function of the depth of the packing for different liquid loading rates ( Lm). ................................... 140 6 10 Aqueous phase toluene concentrations predicted using the PFD model as a function of the depth of the packed tower for different initial toluene concentrations (Cin). ......... 140 6 11 Liquid -solid mass transfer coefficient as a function of the Reynolds number. ................ 141 6 12 Liquid -solid mass transfer coefficient as a function of the inlet toluene concentration. 141 6 13 Actual versus liquid -solid mass transfer coefficients calculated fro m the correlation in Equation 6 19 as a function of Re and CIN. .................................................................... 142 7 1 Comparison of the measured overall gas liquid mass transfer coefficients to the coefficients predicted by the Onda corre lation and by the Modified Onda correlation for different liquid flowrates and packing styles as a function of air to water ratio ........ 154
13 7 2 Comparison of the predicted wetted surface area of the packing calculated using the Onda correlation and the Modified Onda correlation as a function of liquid flowrate for the tower packed with the commingled packing. ......................................................... 155 7 3 Comparison of overal l gas liquid mass transfer coefficients for the tower packed with high flow rings only and the commingled packing as a function of air to water ratio for a liquid flowrate of 4.0 L/min. .............................................................................. 155 7 4 Overall gas liquid mass transfer coefficients as a function of air to water ratio for different liquid flowrates for the tower packed with commingled packing. .................... 156 7 5 Effect of the liquid flowrate on KGLaw for different gas flowrates for the tower packed with the commingled packing. ................................................................................ 156 7 6 Comparison of actual saturation conditions in the reactor for the different experimental r uns and the equilibrium concentrations predicted by the Henrys law. .... 157 7 7 Results for two phase experiments using saturated conditions : Experiment (a) ............ 158 7 8 Results for two phase experiments using saturated conditions : Experiment (b). ........... 159 7 9 Results for two phase experiments using saturat ed conditions : Experim ent (c) ............ 160 7 10 Results for two phase experiments using saturated conditions : Experiment (d). ........... 161 7 11 Results for two phase experiments using saturated conditions : Experiment (e) ............ 162 7 12 Results for two phase experiments using undersaturat ed conditions : Experiment (f) .. 163 7 13 Comparison of the removal of toluene in the two-phase experiments to the expected removal in the aqueo us phase only ..................................................................................... 164 8 1 Schematics of the reactor and different ial volume used to determine toluenes mass balance equations. ................................................................................................................ 176 8 2 Gas, liquid and solid phases resistances and concentration profiles in the TPOT. .......... 176 8 3 Comparison of concentration profiles obtained from the model using the correlation derived for the aqueous phase (1) and the one fitted to the two -phase data (2).. ............. 177 8 3 Continued .............................................................................................................................. 178 8 4 Effects of dispersion coefficient on the aqueous and gas phases profiles obtained from the models using the Ks calculated from the correlations. ...................................... 179 8 5 Actual net toluene removal versus the net removal predicted by the two-phase model using different mass transfer coefficients. .......................................................................... 179
14 8 6 Concentration profile for toluene in the TPOTpresent in the gas and aqueous phases in the presence of PCO (UV lamps on) and due to mass transfer only (UV lamps off). 180 8 7 Net toluene removal in the TPOT as a function of packed bed depth .............................. 180 8 8 Complete treatment system showing the simulation results. ............................................. 181 8 9 Effect of different operating parameters on the aqueous phase profile concentrations in the TPOT. ......................................................................................................................... 182 8 10 Effect of different operating parameters on the gas phase profile concentrations in the TPOT ............................................................................................................................. 182 8 11 Effect of different operating parameters on the net toluene conversion in the TPOT as a function of packed bed depth. ...................................................................................... 183 A 1 System losses in the gas phase due to potential leaks ........................................................ 193 A 2 System losses in the aqueous phase due to volatilization and/or leaks. ........................... 193 A 3 Photolysis of toluene in the gas phase ............................................................................... 194 A 4 Photolysis of toluene in the aqueous phase. ....................................................................... 194 A 5 Toluen e stripping in the TPOT packed with high flow rings only using the UV lamps off and on and having the air flow free of contaminant. .................................................... 195 A 6 Toluene absorption in the TPOT packed with plastic h igh flow rings only and the water flow free of contamiants. ........................................................................................... 196 B1 Effects of the UV radiation on the leaching of silica and titania from the TPOT packing. ................................................................................................................................. 198 B2 Effect of the flowrate on the leaching of silica and titania from the packing on the TPOT. .................................................................................................................................... 199 B3 Leaching of silica and titania from the packing in the T POT for several experiments performed at various dates. .................................................................................................. 199
15 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy COUNTER FLOW SILICA TITANIA PHOTOCATALYTIC REACTOR FOR THE SIMULTANEOUS TREATMENT OF AIR AND WATER CONTAMINATED WITH VOLATILE ORGANIC COMPOUNDS By Christina Akly August 2009 Chair: Paul A. Chadik Cochair: David W. Mazyck Major: E nvironmental Engineering Sciences The photocatalytic oxidation (PCO) of VOCs was investigated using a novel countercurrent flow reactor designed to enable the treatment of toluene present in the gas and the aqueous phases simultaneously. The reactor was packed with silica titania composites (STCs) commingled with plastic high flow rings Using this mixed packing style was advantageous as it resulted in a higher UV penetration throughout the reactor. The average UV intensity in the reactor was 220 uW/g irra diated titania U nder dry conditions, the STCs had a high adsorption capacity for toluene; however, this adsorption was completely inhibited by the wetting of the STCs when the two phases were flowing simultaneously. The lack of adsorption hindered the PC O of toluene in the gas phase as it was found to be negligible during the two-phase operation. Likewise, the adsorption of toluene in the aqueous phase was negligible likely due to the short mean residence time in the pilot reactor (< 60 s). However, t he P CO of toluene in the aqueous phase l inearly increase d with concentration due to the larger driving force and decrease d as function of liquid loading rate
16 In the presence of both phases, toluene destruction was only observed under conditions in which the s olution was initially supersaturated with respect to the gas phase. Under these conditions, the net destruction of toluene in the system reached a maximum of about 68%. For the most part, the destruction occurred in the aqueous phase, and it was observed t hat high gas phase flowrates could be detrimental to the toluene destruction in the two -phase reactor. The reactor was modeled using a one dimensional plug flow with dispersion model for the aqueous phase and a plug flow model for the gas phase. The disp ersion coefficient and gas -liquid and liquid -solid mass transfer coefficients were determined from correlations developed in this study. The model was calibrated using the two -phase experiments data by fitting the mass transfer coefficients The mathematical model i s a useful tool to simulate the reactors performance for a given set of operating conditions and investigate the effects of the different variables on the net toluene conversion.
17 CHAPTER 1 INTRODUCTION Many organic compounds are currently reg ulated by the Environmental Protection Agency (EPA) under the National Primary Drinking Water Regulations (NDWRs) due to the threat they pose to human health. Among these compounds are volatile organic compounds (VOCs), which cover a significant portion of these regulated organic chemicals. Many VOCs can be toxic to humans in both the air and the water phase and cause environmental problems. Some of the more alarming chronic effects from VOC exposure are the increased risk of cancer and problems with the ne rvous system, blood, liver, and kidney, among others. Some VOCs cause cancer in animals and some are suspected or known to cause cancer in humans such as benzene, trichloroethylene, vinyl chloride and carbon tetrachloride. S hort term exposure to sufficient ly high concentrations of these contaminants may cause fatigue, nausea, weakness, and confus ion ( USEPA, 2007a ). Environmentally, VOCs can be involved in the production of photochemical oxidants such as ozone which can be toxic to humans when present in the troposphere, the y can be implicated in acid rain formation and even the depletion of stratospheric ozone (Japar et al, 1991 ; Cortese, 1990). Man y VOCs are used as solvents for adhesives, and in the production of rubber, drugs, and paints. The majority of them are used in the production of other chemicals, and fewer are used as flocculants, gasoline additives, or insecticides. Their release to the environment is mostly through emissions and wastewater discharges by production and manufacturing facilities an d spills. Since these compounds are very volatile, they will quickly partition to the gas phase, becoming an air pollution problem. Consequently, many of these compounds are also regulated under the Clean Air Act (CCA) and the Occupational Safety and Healt h Administration (OSHA) as air pollutants. When VOCs do not completely volatilize, they can reach groundwater by leaching through the soil where they may be anaerobically degraded; however, biodegradation is
18 usually slow. Toluene, one of the regulated VOCs is the target compound for this research. Like other VOCs, it can potentially cause serious health effects from acute exposures at levels above the regulated maximum contaminant levels (MCL); in drinking wa ter the MCL is 1 mg/L while the air standards r ecommend a concentration below 300 g/m3 to prevent risk of adverse effects by inhalation ( USEPA, 2007b ; OEHHA, 2002). VOCs regulations are becoming more stringent because of the increased awareness and e vidence of the poten tial adverse effects from exposure to these contaminants. The Clean Air Act of 1990 for example, set the goal of reducing the emissions of 189 toxic c hemicals over a period of 8 years, 70% of these chemicals being VOCs ( Armor, 1992). In 1998, EPA issued a rule limiting VOC emissions from consumer products that requires many United States manufacturers, importers, and distributors to limit the VOC content of their products. In addition, another rule was issued limiting emissions from exterior and interior ho use paints, wood and roof coatings ( USEPA, 2007c ). As a result, these more rigorous regulations need to be addressed by improvements in treatment techniques. Conventional process for air and water treatment include phase tra nsfer, chemical treatment using strong oxidizing agents such as chlorine, potassium permanganate, ozone, hydrogen peroxide and ultraviolet light, thermal and catalytic oxidation, and biological treatment. Such VOCs treatment technologies are currently impl emented in potable water treatment plants, groundwater remediation systems, industrial facilities for both gas and aqueous phase emissions, spills, and indoor air pollution, to mention a few. All these traditional treatment processes have limitations of th eir own. For example, phase transfer methods such as stripping, absorption and adsorption, remove undesired contaminants from one phase but transfer them to another without elimination of the problem. Biological treatments that transform the VOCs can only be applied under very specific conditions because of the sensitivity
19 of microorganisms to undesired toxic compounds. The oxidation methods use either strong oxidants that present serious hazards during practical applications or high energy intensive UV lig ht. Some of the newest technologies investigated for the removal of VOCs are advanced oxidation processes (AOP) involving a solid phase catalyst. These processes are usually preferred over conventional VOCs treatment technologies because they completely o xidized the contaminants to carbon dioxide and water. One of the AOPs that have shown promising results is heterogeneous photocatalysis with titanium dioxide (TiO2). There is a vast research literature database on the removal of VOCs using TiO2. The result s have shown this technology to be effective at completely oxidizing the target contaminants from the gaseous phase and at a slower rate from the aqueous phase. Despite its potential, the feasibility for implementing this technology in large scale systems, particularly those that treat both air and water phases, has not been successfully demonstrated. The research efforts have aimed at two major issues concerning photocatalysis with TiO2: (1) the identification of reaction intermediates, detailed reaction m echanisms, and reaction kinetics; and (2) the optimization of reaction conditions to enhance the photoreaction rate and yield. These issues, which are key factors to large scale implementation, have been broadly investigated; however, they have not been ye t resolved. In addition to the lack of evidence supporting the intrinsic mechanisms and kinetics of VOCs reactions on the catalyst surface, a full size practical reactor configuration and catalyst support for such reactor to implement this technology have not been developed. The reactor needs to be both efficient at achieving the required removal and cost effective compared to currently available technologies. Several factors have impeded the design of effective photocatalytic reactors. Since heterogeneous photocatalysis requires light irradiation to initiate the reactions, sufficient illumination distribution inside the reactor that will activate most of the
20 catalyst is of utmost importance. Another barrier to the development of the photocatalytic reactor is the incorporation of the catalyst inside the reactor. Several configurations have been used in lab -scale experiments, but most of them are usually inconvenient and expensive when considered for large scale systems. Finally, the reactor has to be able to handle fluctuations in flowrates and concentrations that are typical in full -scale applications ( Ray, 1999). In this work, silica titania composites capable of simultaneous adsorption and photocatalytic destruction of organ ic compounds were developed and used in a specially designed counter -flow packed bed reactor capable o f treating both the gas and aqueous phase of the desired contaminant. Toluene was the target volatile organic compound commonly found as a water and air contaminant. The proposed technology can handle the air emissions and contaminated water within one reactor. The overall proposed system, shown in Figure 11, consists of three treatment stages. Given tha t using a counter current flow configuration with both streams being contaminated could cause the undesired transfer of contaminants at both ends of the tower where treated streams meet the contaminated ones, two additional packed sections were included in the final system. The concentrations exiting both e nds of the tower were expected to be very low. Therefore, treating these trace level contaminants should required smaller size packed beds. Additionally, si nce these polishing sections would be treating only one phas e of the contaminants, they would be packed only with the silica titania composites (STC). This part of the system was not be a priority in this research because the treatment of single phase organic contaminants in air or water using STCs has already been proven feasible by other researchers (Holmes et al., 2004; Londeree, 2002; Ludwing, 2004; Pintioniak et al., 2003; Stokke et al., 2006). Therefore, most of the efforts in this work focused on investigatin g the mechanisms o n the section where the two contaminant phases we re treated together.
21 The objectives of this research were the following: 1 D esign and optimize the performance of a counter current flow reactor packed with engineered silica titania composites to enable the simultaneous treatment of ga s phase and aqu eous phase VOCs, using toluene as the target compound. 2 Determine the main factors that affect the toluene removal efficiency (eg. residence time, water vapor concentration, mass transfer, influent contaminant concentration, etc.) in the gas and aqueous phases separately. 3 Determine the main operating factors that affect the toluene removal efficiency (eg. residence time, air to water ratio, inlet gas and aqueous phase concentrations, etc.) in the gas and aqueous phases simultaneously. 4 Investigate the effe ct of the simultaneous two -phase flow in the individual adsorption and photocatalytic efficiencies of the two fluid phases. 5 Determine the effect of the commingled packing o n the overall UV distribution in the reactor 6 Investigate the presence of potentia l byproducts in the effluent flows and adsorbed to the catalyst surface. 7 Determine the overall mass transfer coefficient gas liquid and liquid-solid, for the commingled packing in the reactor and compare it to know n mass transfer correlations. 8 Model the reactor mathematically by using operating parameters a s the main inputs and calibrate the model using collected data for toluene The following hypotheses were investigated in this research : 1 The removal of aqueous phase VOCs will be enhanced in the presence of the air stream in the proposed counter current flow reactor configuration. 2 The UV light distribution will be improved by the use of the commingled packing. 3 The expected presence of a liquid film around the catalyst pellets during the removal of gas -phase VOCs will hinder the adsorption of toluene on the STC However, decreased adsorption will not result in decreased photocatalysis of toluene in the gas phase, since the reaction can still occur in the bulk air 4 The photocatalytic oxidation mechanism for the destruction of VOCs involves the hydroxyl radical attack using the abundant O2 as the scavenger specie s ince water is pre adsorbed to the catalyst and conti nuously supplied in the system together with air
22 5 D espite the short retention time, the con taminants will be completely mineralize to CO2 and water due to excess hydroxyl radical so no byproducts other than CO2 will be present in the effluent streams 6 Deactivation of the catalyst can be overcome using the two phase system configuration. QG = Gas Flowrate Yo, Contaminated Influent Gas SiO2-TiO2 Air Polisher SiO2-TiO2 Water Polisher Clean Water Pall Rings and SiO2-TiO2 Packing Ce Treated Water Ye Treated Air QL = Water Flowrate Co, Contaminated Influent Water UV Lamps Clean Air Figu re 1 1 Conceptual design of the photocatalytic system
23 CHAPTER 2 LITERATURE REVIEW Heterogeneous Photocatalysis In heterogeneous photocatalysis, photoinduced reactions take place on the surface of the catalyst while the catalyst remains intact. Interest in heterogeneous photocatalysis started with the discovery of the photocatalytic splitting of water by Fujishima and Honda in 1972. Semiconductors have been widely studied because of their application in photocatalysis. They can act as sensitizers due to th eir electronic structure. Different from metals, semiconductors posses a band gap, which is a void energy region where no energy levels are available to promote recombination of the electron and hole produced by photoactivation in the solid. The band gap e xtends from the filled valence band (VB) to the vacant conduction band (CB ). Semi conductor photocatalysis can be more appealing than conventional oxidation methods because semiconductors are usually inexpensive, nontoxic, and capable of extended use without substantial loss of photocatalytic activity. One of the most commonly used semiconductors in environmental applications with sufficient band-gap energy to catalyze different chemical reactions is titanium dioxide (TiO2). The band gap energy for TiO2 is 3 .2eV compared to the normal hydrogen electrode (NH E) ( Hoffmann et al., 1995). The initial process in heterogeneous photocatalysis, presented in Equation 2 1 is the generation of electron -hole pairs in the semiconductor part icles by the excitation of an electron with photons of energy, h greater than the bandgap energy ( h > 3.2 eV or < 388 nm for TiO2). In this process, an electron from the CB ( eCB -) is promoted to the VB, creating a hole in the VB (hVB +). The electron in the CB can act as a reductant while the hole in the VB acts as oxidant (Linsebigler et al., 1995). This photoexcitati on process is shown in Figure 1 1. TiO2 + h hVB + + eCB (2 1 )
24 Upon excitation, the electron and hole can follow different pathways ( Linsebigler et al., 1995, Hoffmann et al., 1995, Herrman n et al., 2005): Bulk Recombination: Some of the electron hole pairs generated may recombine within the volume or on the surface of the semiconductor dissipating the input energy as heat. hVB + + eCB TiO2 + heat (2 2 ) Surface Recombination: There is a high probability that the electron -hole pairs which migrate to the surface may recombine with species such as hydroxyl ions ( OH-) and hydroxyl radical s (OH), thus limiting the efficiency of photocatalysis. hVB + + OHH2O (2 3 ) eCB + OH + H+ H2O (2 4 ) Surface Trapping: The electron hole pairs, which migrate to the surface of the catalyst, are t rapped and form primary hydroxyl radicals. Additionally, the electron can get trapped irreversibly to form Ti3+ defects. hVB + + OHOH (2 5 ) hVB + + H2O OH + H+ (2 6 ) eCB + >TiIV TiIII (2 7 ) Electron Scavenging: Oxygen ( O2) scavenges the conduction band electrons through the following consecutive reactions t o form multiple peroxide radicals. TiIII + O2 O2 (2 8) eCB + O2 O2 (2 9 ) H+ + O2 HO2 (2 10) 2H+ + O2 + eCB H2O2 (2 11)
25 Interfacial Charge Transfer and Back Reactions: Some of the radicals as generated above may recombine at the surface of titanium dioxide through a multi -step charge transfer reaction. These back reactions are simply an electron transfer between the surface radicals and the titanium dioxide surface. The back reactions become important when these reactive radicals are trapped near the catalyst surface. TiO2 + 2(OH) TiO2 + 1/2 O2 + H2O (2 1 2 ) TiO2 + 2(H O2) TiO2 + 3 /2 O2 + H2O (2 1 3 ) TiO2 + H2O2 TiO2 + 1 /2 O2 + H2O (2 1 4 ) Degradation: The highly reactive radicals that escape the back reactions lead to complete degradation of the organic (or inorganic) compound, either by initiating an oxidation or reduction pathway. For example, C7H8 + OH Intermediates CO2 + H2O (2 15) The efficiency of photocatalysis can be measured by the numbe r of events occurring per photon absorbed, which is known as the quantum yield. Since measuring the light absorbed in a photocatalytic process is very difficult, it is assumed that all light is absorbed, and the resulting value is defined as the apparent q uantum yield. The quantum yield ( ) can be described by the relationship given in Equation 2 16: kCT / ( kCT + kR) (2 16) where kCT is the rate of the charge transfer process and kR is the rate of the hole recombination (Linsebigler et al., 1995). For an ideal sys tem, the quantum yield would be equal to 1 (rate of hole recombination is either zero or negligible). However, the efficiency of the photocatalytic processes decreases by the recombination of the electron and hole. This recombination process can be retarded by promoting electron and hole traps in the system, either as electron acceptor or
26 donor s pecies or as catalyst surface defects. During catalyst preparation, ideal semiconductors crystal lattices are no t always produced resulting in natural surface and bulk irregularities. These defects are associated with surface electron states which differ in their energy from the bands present in the bulk semiconductor and serve as charge carrier traps that help supp ress electron hole recombination. Hindering such recombination increases the probability for oxidation and reduction processes to take place, hence increasing the efficiency of photocatalysis. ( Fox and Dulay, 1993). Titaniu m Dioxide as Catalyst By far, TiO2 is the most used photocatalyst for air and water treatment applications. This semiconductor is preferred among others because it is chemically and biologically inert; and thus, it does not undergo photo or chemical corros ion. It is also inexpensive when compared to other available catalysts. T he photogenerated holes are highly oxidizing while the photogenerated electrons are sufficiently reducing to produce superoxide from dioxygen. After reaction with water, the holes ca n produce hydroxyl radic als. The hydroxyl radical oxidation potential is more positive than that for ozone (See Table 2 1) The red uction potential for conduction band electrons is in principle negative enough to evolve hydrog en from water, though the electrons can become trapped and lose some of their reducing power ( Fox and Dulay, 1993). Two different crystal structures of TiO2, rutile and anatase, are commonly used in photocatalysis. Of these two forms, anatase shows the h igher photocatalytic activity. The structures of rutile and anatase TiO2 are shown in Figure 2 2 Both structures consist of titanium (Ti4+) atoms surrounded by six oxygen (O2-) atoms forming an octahedron. The differences between the two crystal structure s are the distortion of each octahedron and the assembly pattern of the octahedra chains. The Ti Ti distances in anatase are greater whereas the Ti O distances are shorter than in rutile. Additionally, the rutile octahedron is connected to 10 neighboring o xygen
27 atoms while the anatase structure is in contact with only eight. These structural differences are responsible for the differences in their band electronic structures and mass densities ( Linsebigler et al., 1995). Of th e commercially available titanium dioxides, Degussas P25 is the most researched for its photocataly tic ability. Table 2 2 shows s ome of the physico-chemical characteristics of Degussa P25. Commercial P25 is produced from high-temperature flame hydrolysis of TiCl4 in the presence of oxygen and hydrogen followed by a steam treatment to remove HCl. The resulting product is a catalyst consisting of 70% anatase and 30% rutile phases of T iO2 (Degussa Technical Bulletin, 1990). The anatase phase of TiO2 is considered the more photoreactive of the two. The difference in reactivity of the phases is attributed to: (1) the more positive conduction band of the rutile phase that can prevent molecular oxygen from acting as an electron acce ptor, and (2) the difference in surface properties between the two phase s ( Tanaka et al., 1991 ). Tanaka et al. (1991) showed that the photocatalytic ability of a TiO2 catalyst is foremost affected by the anatase content and secondarily affected by the size of the crystal. Under equal percentages of anatase phase, a TiO2 catalyst containing larger crystals is more efficient than one with smaller crystals. This could be due to the larger migration distance of the holes and elec trons to the surface of the catalyst, thereby decreasing the possibility of recombination ( Tanaka et al., 1991 ). Furthermore, TiO2 in the form of pure anatase might be deactivated to a lesser extent than the combined phase T iO2 for some organic compounds. Marci et al. (2003) for example, showed that when using a titanium dioxide catalyst made of 100% anatase phase no deactivation of the catalyst was encountered. However, when Degussa P25 was u sed under the same conditions, the catalyst was completely deactivated after 20 h of operation. The types of byproducts found in the
28 photooxidation of organic chemicals might be also influence by the type of TiO2 phase used for the reactions (Marci et al., 2003 ). Catalyst S upport : S ilica -Titania C omposites Nanosized titania particles come in the form of a very fine white powder, so developing practical catalyst supports is a challenge for the application in TiO2 photocatal ytic systems. Common methods of implementing this catalyst in gas phase reactors treating VOCs include depositing the titania on substrate particles for packed beds ( Kobayakawa et. al., 1998 ) or on re actor tube walls as a th in film (Maira et. al., 2001 ). Both of these opti ons can be disadvantageous to the photocatalytic process because in the first case, there is only an effective thin layer exposed to the light, and in the second case, the imm obilization of titania particles on tube walls limits the mass transfer. Furthermore, tube wall films do not always provide homogenous coating, and there is a high chance of detachment from the surface. One means to overcome immobilization is to use a flui dized bed. However, this bed will eventually fail as a result of attrition. For the treatment of aqueous phase pollutants, titania slurry is the most commonly a pplied method ( Crittenden et. al., 1997). However, due to the s mall particle sizes, the removal of the semico nductor from solutions is difficul t and energy intensive Additionally, this effective water treatment solution would be difficult to implement for the treatment of air in a reactor system that attempts to trea t both phases. Since catalytic reactions occur at the fluid -solid interface, a large interfacial area can be advantageous in attaining a significant reaction rate. Large surface areas are usually provided by porous structures. Mesoporous materials have at tracted considerable attention after the discovery of M41S family of silicates, which were determined to possess large surface areas, narrow pore distributions and controllable pore sizes ( Kresge et al, 1992). Immobilization of titanium dioxide by embedding it in a silica matrix (Si O2) having a large surface area has shown to be an effective
29 catalyst support for treatment of gas and aqueous phase VOCs ( Belhekar et al.,2002; Boonamnuayvitaya et al, 2006; Holmes, 2003; Kang et al.,2004; Londeree 2002; Ludwig et al 2008; Stokke, 2008). Arai et al. (2006) showed that loading silica to TiO2 using a sol -gel method resulted in 1.5 to 2 times increase in reaction rate co nstants for the photocatalytic oxidation of phenol, nitrobenzene, propionic acid and benzyltrimethylammonium chloride as compared to pure TiO2 suspensions. Similar improvements have been observed in the PCO of toluene in the gaseous phase Zhang (2005) fo r example, determined that the overall removal of toluene increased by about 30% when silica embedded titania was us ed as the catalyst instead of simply pure TiO2. Likewise, Zou et al (2006) s howed that the conversion of toluene reached a steady state at 51% when using Degussa P25 and it increased by 14% when high surface area SiO2TiO2 pellets were used under the same operational conditions. Neither of these studies accounted for adsorption an d photo -oxidation removal separately, so the enhanced conversion might be partially attributed to an increased adsorption due to the greater surface area of the SiO2 TiO2 catalyst. Silica exists as a SiO4 tetrahedron where Si atoms are connected to four oxygen atoms one of which is bounded to another Si atom from another tetrahedron, as shown in Figure 2 3 Silica forms a clear network, transparent to UV radiation, so it does not hinder the absorption of photons by the TiO2 semiconductor. Furthermore, sili ca has a high adsorptive capacity for many organic contaminants since it is commonly the fixed phase in the columns used for gas chromatograph y (Miyabe and Guiochon, 2004). Therefore, incorporating TiO2 to the silica matrix can improve the photocatalytic process by adsorbing the contaminants to the silica surface which will provide longer time for interaction wi th the catalyst. As a result, the probability of the compound to be degraded by the active species formed in the sur face is increased.
30 There have been many research studies investigating methods to prepare SiO2TiO2 composites with photocatalytic properties. The synthesis conditions of these materials have an important impact on the final properties of such composites. Silica titania composites can be formed by two types of interaction between Si and Ti atoms: (1) physical interactions, such as van der Waals attractions and (2) chemical interactions, such as the creation of Ti O -Si bonds. There are few methods widely use d for the preparation of mixed oxides (chemically bonded composites) such as sol -gel hydrolysis, coprecipitation, impregnation and chemical vapor deposition. Each of these methods has different advantages and disadvantages which are related to the difficul ties in attaining homogeneous coatings, controlling changes in titania crystal phases between rutile and anatase, and limiting the TiO2 content in the mixture. In all these methods, titania is incorporated into the silica matrix by replacing a Si atom from the silica network by a Ti atom. Among all the available methods, a sol -gel synthesis was selected since it has shown to overcome most of the common problems encountered in the preparation of SiO2TiO2 composites and has proven to produce materials that c an effectively remove organic compounds from gas and aqueous phases under the influence of UV radiation while enabling the casting of the composites to almost any desired sha pe ( Holmes et al., 2004; Londeree, 2002; Ludwing, 2 004; Pitoniak et al., 2003; Stokke et al., 2006). During the sol gel synthesis, a sol gel, which is a stable dispersion of colloidal particles in a continuous liquid phase, is formed. These silic a -gels a re derived from silicon alkoxides such as tetraethox y orthosilicate (TEOS) and tetramethoxy orthosilicate (TMOS). TEOS is desired over TMOS because, even though the latter hydrolyzes faster, it produces methanol as a byproduct of the process. TEOS, on the other hand, has longer hydrolysis rates but it produ ces ethanol instead. If workers are exposed to the emissions during the manufacturing process, ethanol emissions are
31 less toxic than methanol. The steps involved in the processing of silica gels generally include hydrolysis, condensation, gelation, aging, drying and curing. Initially, the alkoxide used as precursor is mixed with water to undergo hydrolysis. Ethanol is added together with the alkoxide for cosolvency purposes because the alkoxide has low water solubility. During hydrolysis the alkoxide group is replaced by a hydrogen ion by a nucleophilic attack of water on the silicon ato m (Eq uation 217). Hydrolysis can be catalyzed by either acid or bases. In general, acid catalysis increases hydrolysis rate while base catalysis favors condensation. Dependi ng on the concen trations and conditions, re este rification, i.e. the reverse reaction for hydrolysis and condensation, can oc cur ( Hench and West, 1990). 2O H + ROH (2 17) Following hydrolysis condensation reactions occur with siloxane bonding There are two types of condensation, water condensation and alcohol condensation. In the first case, condensation occurs betwe en two silanol groups ( Equation 2 18) while the latter involves a silanol and alkoxide group ( Equation 2 19). W ater and alcohol are the respective byproducts for each reaction Fluoride is added to the process since it is an important catalyst of silica re actions. Under acidic conditions, fluoride has the ability of catalyzing both hydrolysis and condensation reactions. It is believed that fluoride makes the silicic acid m onomer more susceptible to nucleophilic attack by expanding its coordination sphere (Powers, 1998) H + H O Si H2O (Water condensation) (2 18) H O Si (Alcoholic condensation) (2 19) Additional polycondensation occurs to form the SiO2 network, which still contains the alcohol and water within the pores. At this point, during cross linking, the low viscosity liquid like sol is cast into a special mold that prevents adhesion of the sol. After hydrolysis, gelation
32 occurs as colloidal particles grow by polymerization or aggregation. During this step, the viscosity increases sharply and a solid object results in the shape of the mold. Aging or syneresis is the next step in the process. Additional network linkages are formed resulting in the shrinkage of the gel which forces some of the liquid out of the pores. The aged gel has to be strong enough to resist possible cracking during drying. The drying step has an important influence on the final pore size of the sol gel This stage involves the removal of the liquid from the pores of the network. Drying has to be closely controlled to avoid cracking. For xerogel formation, the selected type of sol -gel for this research, the removal of the liquid from the pores is achieve d by evaporation at ambient temperatures and pressures below the critical temperature of the liquid (Hench and West, 1990) The last step to form a mechanically strong gel is curing. This final heat treatment influences the surface chemistry of the gels. Water retained on the surface of the silica can be present in two forms: (1) physisorbed, free water within the ultraporous gel structure, or (2) chemisorbed, as hydroxyl groups associated with the gel surface. Heat treatments using temperatures in the 180oC 200oC range can remove physisorbed species while dehydroxylation, i.e. removal of chemisorbed species requires temperatures above 200oC. Rehydroxylation of the silica surface in the presence of water occurs reversible up to 600oC, such that the surface characteristics are not significantly changed. However, above this temperature, rehydroxylation is less likely and decreases in surface area due to sintering are observed (Londeree, 2002). T he surface functional groups are the main features of a hydrated silica surface and are responsible for the hydrophilic nature of most silica gels. Two functional groups are observed on the silica su rface: silanol groups ( -Si O H) and siloxane groups ( Si -O Si -). Silanol groups,
33 shown in Figure 2 4 act as active adsorption sites while siloxane groups are non -reactive. The average silanol concentration of fully hydroxylated surfaces is about 4.9 silanols/nm2 (Legrand, 1998). Photocatalysis of Organic Compounds with TiO2 Although it is well established that the initial step for the photooxidation process on TiO2 is the excitation of the surface to produce and electron-hole pair, the subsequent chemical events at the fluid -solid interface remain an ambiguous and controversial issue. Different authors have postulated a wide variety of mechanistic pathways to explain the photodegradation of organic compounds and have attempted to determine the rate limiting reactions for these proc esses. Once the semiconductor has been photoexcited, the organic compound can be either oxidized directly by the trapped holes or indirectly by species formed on the surface of the catalyst such as hydroxyl radicals which are strong oxidizing agents ( Hoffmann et al., 1995). The photochemical reaction on TiO2 catalyst in the aqueous phase is generally believed to involve radical species. Most studies have proposed hydroxyl radicals derived from hole trapping by surface hydroxy l groups as the primary oxidizing agent and oxygen as a scavenger for photogenerated electrons. These surface trapping and electron scavenging reactions were previously presented in Equations 2 5 to 2 10. Turchi and Ollis (1990) proposed four cases in which a hydroxyl radical can oxidize an organic molecule in the aqueous phase. These cases are represented by Equations 2 20 to 2 23. Case 1 : The reaction occurs while both species, hydroxyl radical ( TiIVOH ) and organic molecule (X1,ads), are adsorbed resulting in adsorbed products (X2,ads). TiIVOH + X1,ads TiIV + X2 ,ads (2 20) Case 2: The reaction occurs between a nonbound radical (OH ) and an adsorbed organic mol ecule (X1,ads).
34 OH + X1,ads X2,ads (2 21) Case 3: The reaction occurs between an adsorbed radical and a free organic molecule (X1) arriving at the surface of the catalyst and producing a molecule in the bulk phase (X2) TiIVOH + X1 TiIV + X2 (2 22) Case 4 : The reaction occurs be tween two free species in the fluid phase OH + X1 X2 (2 23) The extensive research on photodecomposition of water and photooxidation of organics has provided some evidence of hydroxylated intermediates which in principle, can be formed either by ho molytic attack by a hydroxyl radical or by hydration of a singly oxidized intermediate ( Fox and Dulay, 1993; Linsenbigler et al., 1995). For the case of gas -solid photocatalysis, however, the evidence for OH attack is not as definite. Either radical formation for organic oxidation or direct hole oxidation has been postulated as plausible mechanisms. Because gas p hase molecules can move more freely, hole -electron recombination and hole organic molecule oxidation can be kineticall y competitive. Similarly, the abundance and availability of water molecules in the gas phase is severely diminished as compared to the aqueous phase resulting in a decreased probability of hydroxyl radical formatio n ( Fox and Dulay, 1993). Nonetheless, there is currently no agreement regarding which species are more extensively active in the degradation of organic molecules. Furthermore, it has been difficult to establish an unambiguous mechanism to expl ain the photocatalytic oxidation (PCO) of organics because the data for direct hole oxidation as well as hydroxyl radical attack can result in similar reaction intermediates (Fox and Dulay, 1993; Hoffmann et al.,1995). N ever theless, some of the mechanisms found in the literature that are widely accepted as possible photocatalyitc pathways are presented below (adapted from Serpone and Emelie, 2002):
35 Direct Pathways The Langmuir -Hinshelwood Proce ss: Some researchers have suggested that because photogenerated electron-hole recombination is so fast, interfacial electron transfer is kinetically competitive only when the electron donor or acceptor is preadsorbed on the catalyst surface. Preadsorption has been also suggested as a prerequisite for the reaction to occur at significant high rates (Fox and Dulay, 1993). Consequently, the Langmuir Hinshelwood (LH) mechanism has been proposed as a possible pathway for the PCO of organic compounds using TiO2. The LH mechanism described by Equations 2 24 to 2 28, proposes as the initial step the adsorption and desorption of the organic molecule of interest (X) on the catalyst surface (S) based on Langmuir equilibrium. During the n ext step, the VB hole (hVB +) is trapped by the adsorbed organic molecule (Xads) forming a rea ctive radical state (Xads +). This species can either decay when electron recombination occurs (Eq uation 226) or undergo chemical reactions to regenerate the origi nal surface state and yield products ( Equation 2 27) TiO2 + h hVB + + eCB (2 1 ) X + S (TiO2 surface) Xads (2 24) Xads + hVB + Xads + (2 25) Xads + + eCB Xads (2 26) Xads + Products + S (2 27) The Eley -Rideal Process : The Eley Rideal process has been described by Serpone and Emeline (2002). After the hole -electron pair generatio n ( Equation 2 1 ), the hole is trapped by surface defects (S) producing surface active centers (S+) which can e ither decay by recombination (Equation 229) or act as adsorption sites where the organic compounds can
36 chemisorb to form reactive species (S -X)+ that can further decompose yielding products (Equations 2 30 and 231, respectively). S + hVB + S+ (2 28) S+ + eC B S (2 29) S+ + X (S -X) + (chemisorption) (2 30) (S -X)+ S + Products (2 31) Indirect Pathways The photocatalytic destruction of organic compounds by reactions with unstable species formed as the result of the surface activation is the indirect pathway most commonly c onsidered (Equation 2 32). Researches that have investigated indi rect pathways include Turchi and Ollis (1990), Fan and Yates (1996), and Hennenzel et al. (1998), to name a few. The oxidants formed by water and oxygen adsorption and photodecomposition found in most of the literature are OH, O2 -, HO2 H2O2 and intermediates such as Oamong others ( Hoffmann et al., 1995;Yue et al., 2002). Some of the proposed reactions for their formation were presented in Equations 2 6 and 2 8 to 2 10. OH (or other rad ical) + X Intermediates CO2 + H2O (2 32) Photocatalytic O xidation (PCO) of Toluene with TiO2: Mechanisms and I ntermediate F ormation There seems to be an agreement in the literature that the initial PCO of toluene with TiO2 is very fast. However, the oxidation rate decreases dramatically within the first 30 to 60 minutes. During toluene photocatalysis, most research studies have shown that there is a high potential for catalyst deactivation depending on the experimental conditions (A lberci and Jardim, 1997; Boonamnuayvitaya et al., 2006; Hennezel et al., 1998; Maira et al., 2001a ). This phenomenon is not true for all VOCs, and it seems to be most evident for the case of aromatic compounds such
37 as toluen e and benzene ( Alberci and Jardim, 1997; Lebrawski and Ollis, 2003a .). The deactivation of TiO2 during toluene photocatalysis is usually visually supported by a change in color of the catalyst from white to yellow or brown (Alberci and Jardim, 1997; Eianga et al., 2002). Deactivation does not seem to be strongly related to the concentration of toluene since most studies on gas -phase toluene have reported some degree of deactivation. For example Alberci and Jardim (1997) observed decreases in toluene conversion yields from about 80% to 21% after 150 minutes of continuous operation, attributed to catalyst deactivation, for initial concentrations of 17 ppmv as well as 500 ppmv. Deactivation was le ss evident for lower concentrations, but it was still observed. Similar results were found by Sauer et al. (1995) and Luo and Ollis (1996) under conditions of lower concentrations and shorter contact times. Many other studie s have encountered severe deactivation after 3 h ours of catalyst us e ( Blount et al., 2001; Hennezel et al., 1998; Maira et al., 2001). Literature indicates that the deactivation of the catalyst decreases with increasing humi dification of the air. The actual effect of water and mechanisms for this phenomenon have not been well documented. Similarly, after catalyst poisoning, the catalytic surface can be regenerated by illumination of the photocatalyst in the presence of pure a ir ( Peral and Ollis, 1992). For long term deactivations, however, stronger oxidants such as H2O2 might be required ( Alberci and Jardim, 1997). Catalyst deactivation has been attributed to the accumulation of recalcitrant intermediate species strongly bound to the catalyst surface limiting the oxidation of toluene. The oxidation rate of these intermediates is much slower, and therefore, the overall PCO of toluene takes longer than for other orga nic compounds ( Ameen and Raupp,1999; Augugliario et al., 1990; Blount and Falconer, 2002; Hennenzel et al., 1998; Larson and Falconer, 1997). Ch aracterization of the potential intermediates causing catalyst deactivation is v ery important because it is related to the
38 photocatalyst lifetime which is a factor of the process economics of photocat alysis ( Sauer and Ollis, 1996). Earlier studies by Larson and Falconer (1 997) on the PCO of toluene in the gas phase showed that this compound reacted quickly to form strongly bound intermediates more difficult to degrade and their rate for degradation to CO2 and H2O was much slower. Based on their results, they proposed that the first product from the oxidation of toluene is benzyl alcohol, which further reacts to form benzaldehyde and CO2. They were not able to identify the strongly bound intermediates because the ones they found, benzyl alcohol and benzaldehyde were not cons idered as plausible limiting reactions due to their high rate of oxidation to form CO2 when tested separately. These rates were 20 and 10 times faster than the oxidation rate of toluene. On the other hand, they suggested that a possible intermediate was be nzoic acid since it has been reported by others as a byproduct but they were not able to test for this compound based on the nature of their experiments (Larson and Falconer, 1997). Another group of researchers, Mendez Roman and Cardona Martinez (1998) further supported benzoic acid as the strongly bound intermediate in the degradation of gaseous toluene. In their study, they identified benzaldehyde and benzoic acid as intermediates using the Fourier Transform Infrared (FTIR) spectrum technique, and to a lesser extent, benzyl alcohol using methanol extraction. They proposed that toluene was oxidized to benzaldehyde, which was further oxidized to benzoic acid. They attributed the surface deacti vation of the catalyst to benzoic acid, and they found that this deactivation was retarded when the system was humidified. Their explanation for this phenomenon was that the formation of benzoic acid, which accumulates on the surface and degrades at very s low rates, was slowed down by the water. The water had two possible roles: competition with species that were forming on the
39 surface, or partial regeneration of some of the active sites on the catalyst ( Mendez Roman and Cardo na -Martinez, 1998). Refuting the arguments that suggested benzoic acid as the strongly bound species, is the study of Augugliaro et al. (1990) who found that benzoic acid seems to be an active species on the catalyst surfa ce. Significant amounts of benzene and CO2 were produced by PCO when benzoic acid wa s reacted separately under the same conditions as toluene. The intermediates found in this work were benzaldehyde, benzene, benzyl alcohol, and trace amounts of benzoic aci d. The proposed path for toluene degradation was the hydroxyl radical attack to the toluene molecule forming benzaldehyde, which further reacted with OH to form benzoic acid, and the latter decomposed to benzene and CO2. Benzene was found as a transient product present only at the beginning of the PCO. Among the strongly bound intermediates suggested to decrease the activity of the catalyst were benzyl alcohol, phenol and benzoate -like species. These species would hinder the formation of benzoic acid, and consequently, benzene and CO2 production would decline in the first few hours. Similar to the other authors, they also found that the catalyst deactiv ation was retarded upon humidification, with the catalyst activity lasting about 20 hours for the case of unhumidified systems versus 70 hrs for humidified conditions. Additionally, it was suggested that ben zaldehyde does not oxidize any further in the abs ence of OH, deactivating the catalyst (Augugliario et al., 1990) Blount and Falconer (2002) on the other hand, found that benzaldehyde oxidation was possible even in the absence of water; therefore, they excluded this species as the responsible for catalytic deactivation. Their findings did not contribute any other possible explanation for the deactivation of the catalyst other than the already mentioned strongly bound species in the surface of TiO2. However, based on their results from Temperature -Programmed Hydrogenation
40 (TPH) used to characterized byproducts, they suggested that the intermediates deactivating the catalyst do not appear to be graphitic structures. Matra et al. (1999) identified benzaldehyde as the main photo-oxidation path of toluene. Small amounts of benzene, benzyl alcohol and traces of benzoic acid and phenol were also detected. In their studies they determined that in the presence of water no decrease of photoreactivity of toluene was observed at steady -state conditions. By removing water vapor from the feed, the conversion of toluene to benzaldehyde was almost completely inhibited, and an irreversible deactivation of the catalyst occurred Similar to the Blount and Falconer results, their investigations using FTIR techniques indicated that benzaldehyde is produced on the TiO2 surface even in the absence of water vapor. However, exposure of the catalyst to the UV light in a dry atmosphere r esults in an irreversible consumption of surface hydroxyl groups. This dehydroxilation should be the reason of deactivation of the catalyst under dry conditions ( Matra et al. ,1999). Most of the articles presented in the lite rature do not seem to state a clear understanding and explanation for the intermediates strongly bound to the surface that hinder the activation of the photocatalyst. However, some mechanistic pathways have been postulated to describe the formation of thos e other intermediates that have been mostly characterized in the different studies. Some of these pathways, as described by Hennezel et al. (1998) are presented below: a The primary proposed toluene PCO pathway involves the i nitial formation of the benzyl radical from the abstraction of the hydrogen from the methyl group in toluene. There are 2 pathways that have been proposed for this radical formation: 1 Direct hole transfer to toluene: electron transfer from toluene to the VB hole to form the benzyl radical.
41 2 Hydrogen abstraction by a OH: electron transfer from the methyl group in the toluene molecule to the OH formed by water adsorbed to the catalyst surface. b Once the benzyl radical is formed, the reaction can follow two other pathways: 1 The benzyl radical can react with O2 to form a benzylperoxy radical. 2 The benzyl radical can react with another toluene molecule to initiate a polymerization reaction. c The benzylperoxy radicals can form the commonly found intermediates in toluene PCO by two different pathways: 1 It can couple and form tetroxide, which further decomposes to benzaldehyde, benzyl alcohol and molecular oxygen 2 It can react with a hydroperoxy radical forming monalkyltetroxide, which further decomposes to benzaldehyde, water and, molecular oxygen. d A plausible pathw ay for degrading the most common intermediates is a direct attack on the ring by hydroxyl radicals, VB holes or hydroxide ion. This attack would form mono hydroxylated isomers (hydroxybenzaldehyde, hydroxybenzoic acid, and hydroxybenzyl alcohol). These hyd roxylated compounds are not found in significant amounts, so this pathway is considered to be of secondary importance.
42 Conditions that influence intermediate formation include the contact time between the catalyst and toluene, the percent relative hum idity in t he influent stream, and some studies have also in dicated light intensity as a possible factor ( Alberci and Jardim, 1997; Peral and Ollis, 1992). Lebrawski and Ollis (2003a and b) dev eloped a two-site kinetic model to simulate and predict the deactivation of TiO2 during photocatalysis of aromatic compounds. The model is based on the assumption that there are two types of catalyst sites in the TiO2 surface: (1) sites that can be occupie d by the aromatic contaminants, water molecules or some reaction intermediates and (2) sites that can be only occupied by reaction intermediates or water due to the more hydrophilic nature of these sites and the fact that most of the intermediates are more polar than the original reactant aromatic compound. The model calculates different surface coverages which are used in rate expressions to finally determine effluent concentrations of the aromatic compounds. It also accommodates for catalyst regeneration and reuse when applicable. Although the model does not provide information about the intermediates or possible reaction mechanisms, the re sults show reasonable agreements with experimental data, so the model can be used as a rough estimate of the potential catalyst deactivation in the presence of toluene. (Lebrawski and Ollis, 2003a and b) G aseous and Aqueous P hase Toluene K inetics Literature suggests major differences between the kinetics of heterogeneous photocatalytic rea ctions of organic compounds present in water and in the gas phase. The liquid-solid interface might appear more complex for kinetic studies since more variables need to be controlled, as
43 compared to gas -solid systems. Such variables include the surface com position, surface area and preparation procedures of the catalyst, solution characteristics, such as pH and its effects on the solid surface structure oxygen partial pressure, and diffusion rates in solution and near the catalysts surface among others ( Anpo et a1., 1991). Desp ite all the possible parameters involved in liquid -solid photocatalysis, most of the variables determining the rate of gas phase photocatalysis are the sa me as in liquid -phase photocatalysis. The main difference between liquid -solid and gas -solid photocatalysis of toluene is that the reaction rates for the lat t er can be orders of magnitude larger ( Lichtin and Avudaithai, 1996 ). However, deactivation of the catalyst during gas phase photocatalysis presents a disadvant age compare to aqueous phase photocatalysis. Except for few papers, research studies have not focused on directly comparing gas phase and aqueous phase reactions on TiO2. Most differences have been observed by c ollecting results from individual phase studies, which are harder to compare due to difference s in operational conditions. With the objective of constructing more impartial comparisons, Lichtin and Avudaithai (1996) studied the photocatalysis of some chemi cals in aqueous and gas phases under similar conditions. The chemicals included in the study were acetonitrile, methanol, trichloroethylene and methylene chloride. Their results show that the photoefficiencies for the removal of such organics, defined as t he number of molecules of organic reactant removed per number of photons incident on catalyst, was from 2 to 300 times larger for gas -solid systems as compared to liq uid-solid systems. Not only has the kinetic comparisons between gas -solid and liquid -soli d photocatalysis been neglected, but studies on the simultaneous PCO of contaminants in both phases are virtually nonexistent. The kinetics of reactions in both phases is of great importance in the design of processes, reactors, and the respective scaleup of photocatalytic systems. Accordingly, a short
44 review of the main factors affecting the oxidation rate of toluene in the aqueous and gaseous phases based on the currently available literature is presented in this section. The rate equations for heterogen eous photocatalysis are traditionally dependent on the intrinsic catalyst characteristics as well as mass transfer considerations in the system. The common steps considered in heterogeneous catalyzed reactions are the following ( Satterfield, 1969): 1 ) Mass transfer (diffusion) of the reactants from the bulk fluid to the external surface of the catalyst pellet. 2 ) Diffusion of the reactant from the pore mouth through the catalyst pores to the immediate vicinity of the internal catal ytic surface. 3 ) Adsorption of reactant A onto the catalyst surface 4 ) Reaction on the surface of the catalyst (A B) 5 ) Desorption of the products from the surface 6 ) Diffusion of the products from the interior of the pellet to the pore mouth at the external surface. 7 ) Mass transfer of the products from the external pellet surface to the bulk fluid. The overall rate of the reaction is equal to the rate of the slowest step in the mechanism, given that such step is much slower than the other steps. Typically the rate ca n be dictated by the extent of reaction or the mass transfer. If the reaction rate is controlling, the diffusion steps (1, 2, 6, 7) are very fast compared with the reaction steps (3 5), so the concentrations in the immediate vicinity of the active sites ar e indistinguishable from those in the bulk fluid. For mass transfer limited reactions, the reaction steps are very fast compared to the diffusion steps, so mass transport affects the reaction rate. In mass transfer limited reactions such as systems affected by diffusion from the bulk gas or liquid to the catalyst surface or to the mouth of the catalyst pores, changes in the flow conditions past the catalyst should change the overall reaction rate. For systems where the diffusion within the pores of the cata lyst limits the reaction rate, changes in the external flow conditions will not have any effect on the overall reaction ra te
45 As many heterogeneous catalytic reactions such as the PCO of VOCs on TiO2 are apparently zero order at high reactant concentration and first order at lower concentrations, the Langmuir -Hinshelwoold (LH) rate form, give n by Equation 2 33, is the most commonly assumed for fitting kinetics data.  1LHads LH adskKC dc rk dt KC (2 33) The terms in the equation are the rate of the reaction, r the concentration of the contaminant in solution, C, time, t the reaction rate constant, kLH, the fraction of the surface covered by the reactant, and the adsorption coefficient of the reactant Kads. When low concentrations are considered (KadsC<<1), this equation becomes pseudo first order ( r = kLHKC ), and when high concentrations are used (KC>>1), the reaction simplifies to a zero order reaction (r = kLH). Although most of the observed kinetic data fit this rate form, it is not a sufficient condition to assume the Langmuir Hinshelwood mechanism, which assumes the adsorption of the contaminant to the catalyst surface, as the actual pathway of the reactio n. Recent studies have de monstrated that other mechanisms, including different modifications of the Eley Rideal mechanism can result in the same kinetics of photodegradation as represented by Equation 2 33 (Turchi an d Ollis, 1990, Emeline et al. 1998). It has been shown that the contaminants interaction with photoactivated oxygen via the most common rate limiting steps described by the four cases presented in previous sections and postulated by Turchi and Ollis (1990) can result in overall reaction rate equations that have the same dependence on reactant concentration as the one modeled by the LH form. Accordingly, the LH equation does not provide any insight of the mechanisms between contaminant and excited oxygen (Turchi and Ollis, 1990). Ollis (2005) recently proposed that the apparent simple LH rate form has several origins which are hard to
46 study because of the various kinetic disguises that exist in photocatalysis. He showed tha t out of the two most common assumptions for kinetic mechanism, pseudo-steady state and slow step assumptions, the former is the only one that explains the data available on light intensity dependence. His conclusions illustrate that the studied photocatal yzed reactions do not reach adsorption/desorption equilibrium because the significant reactivity of the active adsorbed species, such as hydroxyl radical, continuously shift from equilibrium (Ollis, 2005) Although the best kinetic model has not been decided upon, common factors influencing the kinetics of toluene in the gas and aqueous phases have been indicated to be, by many researchers, the initial toluene concentration, the UV intensity, the flowrate and to a lesser exte nt the temperature. For gas -phase PCO, the relative humidity also plays a key role in the kinetics of the reaction, whereas for aqueous phase PCO, pH is usually considered. However, a common feature of photocatalytic reactions on TiO2 powders su spended in aqueous solution is the weak dependence of the reaction rate on solution pH (Fox and Dulay, 1993). The isolectric point for TiO2 in water is about pH = 6, meaning that the surface charge is positive at pH values lower than 6 and negative at higher pH. Despite the evidence showing the importance of pH on the particle size, surface charge, and band edge positions of TiO2, photocatalysis of toluene is found to occur at both low and high pH at comparable reaction rates ( Fox and Dulay, 1993). The major trends and effects of the mentioned factors for gas phase and aqueous phase toluen e kinetics are described below. D irect comparisons among rate constants however, even in the same phase are not alwa ys possible due to the differences in operation al conditions among the studies and calculations parameters arbitrarily selected. For example, studies that use the LH rate form to describe the kinetics of toluene PCO may choose to include or exclude the for mation of intermediates. If intermediates are included in the calculations, their KadsC terms must be present
47 in the L H rate equation, even when their adsorption on the surface of TiO2 might be rather weak. However, for simplicity, many studies employ the initial toluene rate for the application on the equation by assuming that intermediates formation at the beginning of the process is negligible ( Xu and Langford l, 2000). As a resu lt, discrepancies among the LH rate constant s determined by different studies can easily occur because the selected time intervals might be different. Effect of I nitial C oncentration Most studies have found that the decomposition and mineralization of toluene decreases as the initial concentratio n o f the contaminant increases and the rates of decomposition increase accordingly There is no consensus on the reaction order as a function of concentration since several t rend s on the increase of reaction rate has been observed for concentration ranges bet ween 0.5 to 2400 ppmv (Buozaza et al, 2006; Maira et al., 2001b; Obee and Brown, 1995; Pengyi et al., 2003). Fewer studies determined that increasing the initial concentration of toluene had a negative effect on the pseudo f irst order reaction constant (Buozaza and Laplanche, 2002; Wang and Ray 2000). These studies also included a wide range of concentrations, between 20 to 800 ppmv. In general, increasing the concentration of toluene increase s the number of molecules that can be adsorb on the surface of the catalyst and thus, oxidized. However, the surface active area of the catalyst is limited, so overloading the system is likely to deactivate the catalyst more rapidly resulting in adverse ef fects on the oxidation rate despite the positive order kinetics involved Furthermore, the divergent results with respect to the initial toluene concentration might be attributed to the differences in catalyst characteristics among the studies as well as o perational conditions of the systems.
48 Effect of Light I ntensity Higher UV wavelengths usually result in higher intensities. The wavelength of the U V lamp s does not seem to change the kinetics of the PCO of toluene in the aqueous phase at low concentrations For example, Holmes et al. (2004) tested the PCO of toluene in the a queous phase in concentrations below 200 g/L by using UV sources that produced 365 nm and 254 nm wavelengths and no changes in the rate constants were observed Thus, it was concluded that as l ong as the light source provided radiant energy greater than 5.12x1019 J (i.e. > 3.2eV or < 388 nm) which is the minimum photon energy required to excite the electrons in the TiO2 surface, i ncreasing the radiant energy did not further contribute to kinetic improvements. Studies comparing the effects of UV intensity on higher concentrations of toluene were not found in the literature for the aqueous phase. On the other hand, the irradiance usually referred to as light intensity, significantly influences the oxidation rate of toluene in gas phase In general, it has been observed that for system s that are not mass transfer limited, the reaction rate ( r ) increases linearly with UV intensity ( I) if the intensity is below 1020 mW/cm2. For higher intensities, greater than 1020 mW/cm2 the oxidation rate is directly proportional to the square root of the intensity meaning that the recombination of photoinduced charges predominates, and accordingly the quantum efficiency decreases (DOliveira et al 1990). These results have been validated for systems treating gas phase toluene. Hager and Bauer (1999) and Wang and Ray (2000) used intensities between 0.3 and 20 mW/cm2 and found that the reaction rate of gas phase toluene increased linearly with UV intensity. Furthermore, it has been shown t hat at high enough intensities the reaction rate does not depend on I anymore and it is instead mass transfer limited. The quantum efficiency (), i.e. the number of molecules transformed per number of photons adsorbed is also
49 a function of intensity, but it follows the opposite trends of the reaction rate because the absorption of photons follows a first order rate in intensity. Hence, at low intensities, is constant; at higher intensities (> 10 20 mW/cm2), is inversely proportional to the square root of the intensity, and at values of I that are high enough not to influence the rate, varies as the inverse of I (Ollis et al. 1991). Effect of Superficial Velocity The superficial velocity affects the mass transfer and transport of pollutants to the catalyst surface. The mass transfer of the reactants to the catalyst can be the rate limiting step in photocatalytic processes. During immobilization of the catalyst, one of the disadvantages is the decreased contact area between catalyst and reactants that can result in lower mass transfer rates. Whe n the reaction is mass transfer limited, the reaction rate increases with increasing velocity For the case of toluene, many studies in the gas phase have shown that toluene is mass transfer limited for low superficial velocites usually below 10 cm/min ( Alberci and Jardim, 1997; Marci et al., 2003) At higher velocities the oxidation rate does not increase anymore and remains constant, so the reactio n becomes kinetically controlled instead of mass transfer limited. Higher superficial velocites are usually desired to overcome mass transfer limitations; however, increasing the velocities decreases the residence time in the reactor resulting in a decreas e in toluene conversion. Accordingly, an optimum combination of velocity and residence time is expected to result in a maximum pollutant removal Effect of Temperature It is well accepted that increasing the temperatu re increases the reaction rate as descr ibed by the Arrhenius equation. This tendency has been widely observed in photocatalytic reactions. Hager and Bauer (1999) for e xample, found that toluene conversion increased whe n the
50 temperature was raised from 7oC to 27oC; however, the conversion significantly decreased when the temperature was raised further up to 75oC. They concluded that at lower temperatures, desorption of the toluene molecules from the catalyst was the limiting step, whereas at higher temperatures, adsorption of these molecules became the rate limiting step. Similarly, the adsorption of water molecules might be also hindered at higher temperatures. Wang and Ray (2000) fo und only small increases in reaction rates for the three temperatures they investigated, 25oC, 35oC and 45oC. They calculated the activation energy of toluene by using a k versus 1/T plot and determined this value to be 31.42 kJ/mol. This value was the largest activation energy found compared to the other compounds studied, which included be nzene and 1,2 dichloroethylene. Effect of R elative H umidity Relative humidity (RH) plays a key role in the oxidation of toluene in the gas -phase. Some authors consider this parameter a requirement to achieve complete ox idation of toluene (Augugliaro et al., 1999). Most literature agrees that increasing the relative humidity of the influent air increases the oxidation rate (Augugliaro et al., 1999; Hennenzel e t al., 1998; Luo and Ollis, 1996; Pengyi et al., 2003; Wang and Ray, 2000). In general, it is well documented that for low concentrations of toluene (< 3 ppm), the rate increases with relative humidity up to about 30% RH, with an optimum between 20% and 30% RH. However, higher RH has a detrimental effect in the oxidation rate, so above 30% RH, the rate rapidly declines. A possible explanation for this phenomenon is that at such low concentrations of contaminant, competition effects predominate and the water molecules can reach the catalyst easier than the toluene molecules adversely affecting the oxidation of toluene (Obee and Brown, 1995). For higher toluene concentrations, the optimum relative humidity is closer to 40% ( Hennenzel et al., 1998; Luo and Ollis, 1996; Obee and Brown, 1995; Pengyi et al., 2003; Wang and Ray, 2000). Above this
51 value, the rate does not increase any further, but it remains constant. Achieving optimum relative humidity enhances the complete mineralization of toluene while reducing the accumulation of intermediates. Increasing relative humidity increases the overall initial conversion of toluene as well as the conversion obtained at steady state ( Augugliaro et al., 1999; Maira et al., 2001). Although the importance of the presence of water for the oxidation of toluene is unquestionable, the actual role of water on the reaction pathway is still not well understood. Mass Transfer Understandi ng the equilibrium partitioning of chemicals between air and water and the mass transfer across the air -water interface is of utmost importance for the purpose of this research. Equilibrium Partitioning of V olatile O rganic C ompounds between A ir and Water S ystems are constantly moving towards the final state of equilibrium. Once equilibrium is attained, the transfer sto ps. ( Hand et al., 1999). Henrys law is used to describe the equilibrium of the contaminants between air and water phases. The law can be assumed to follow a linear trend with respect to the solutes concentration for low concentrations. If the concentration is too high, possible deviations from linearity shoul d be accounted for. In general, the expression is given by Equation 2 34: Ys = H Cs (2 34) where Ys is the gas phase concentration at the interface [M ass/L ength3], H is Henrys constant [dimensionless], and Cs is the aqueous phase concentration at the interface [M ass/L ength3]. The equilibrium partitioning between air and water, H can be influenced by temperature, pressure, ionic strength, surfactants and the solution pH for the case of ionizable species. Given that most treatment operations are carried out at atmospheric pressu re, the impact of pressure on the Henrys constant can be considered negligible. Temperature, however, influences the
52 aqueous solubility of compounds and their vapor pressure. The changes of H with temperature can be predicted by the vant Hoffs equation (Equation 2 3 5 ), assuming that the standard state enthalpy change ( Ho) is constant over the temperature range of interest. 21 2111 expoH HH RTT (2 3 5 ) Ho is the standard state enthalpy change in water due to the dissolution of a compound in water and R is the universal gas constant. H and T ar e the Henrys constant and temperature, respectively with the subscripts representing different conditions. The other factors influencing H do not have a pronounced effect unless extreme conditions are present. Thus, their influence will not be assessed in detail in this review. Hand et al. in the book by AWWA (1999) describes most effects of these parameters and provides a large set of references on the topic. Mass Transfer Principles The rate at which a component is transferred from one phase to another depends upon the degree of departure of the systems from equilibrium. Whether mass transfer occurs from air to water or vice versa, the mechanisms and assumptions are basically the same. The two film theory, developed by Lewis (1916) and Whitman (1923), i s the basis for most mass transfer correlations. According to this theory, turbulence in the two phases dies out near the interface, so equilibrium occurs at the gas/liquid interface, as illustrated in Figure 2 5 Consider for example that a contaminant pr esent in the aqueous phase is in contact with the air phase, as depicted b y Figure 2 5 The tendenc y of the system to reach equilibrium is sufficient for the molecules in the bulk aqueous phase ( Cb) to diffuse in the air phase ( Yb). First, the contaminant in the bulk water phase will move towards the air -water interface due to the concentration difference in this film (Cb > Cs). Similarly, the concentration in the air phase at the air -water interface ( Ys) is larger than
53 the bulk air phase concentration, providing the driving force for the contaminant to move to the bulk air phase ( AWWA,1999) The entir e resistance to transfer is considered to be contained at the two films near the interface where the transfer occurs by purely m olecular diffusion. Local equilibrium is assumed at the interface and concentration gradients at this interface are established much faster than common steady-state diffusion assum es ( Skelland, 1985). Th e overall resistance to mass transfer (RT) is considered to be the sum of the liquid phase ( RL) and the gas phase ( RG) resistance s (Equation 2 3 6 ). Th ese resistances can be defined as the r eciprocal of the liquid phase and gas phase rate constant, respectively ( 2 37) (McCabe et al., 2005): RT = RL + RG (2 36) 111LwlwgwKakaHka (2 37) The rate of transfer of a VOC from the aqueous phase to the gas phase can be described by the linear simplification of F icks Law (Equation 2 3 8 ): J = KLaw [C* Cs] (2 38) where J = mass transfer rate [Mass /t ime /L ength2] KL = liquid side mass transfer coefficient [L ength/time ] aw = wetted interfacial area per volume of tower [L ength2/L ength3] C* = equili brium concentration in aqueous phase [M ass/L ength3] According to E quation 2 3 8 the factors influencing the rate of transfer of a contaminant from the aqueous to the air phase are: the driving force for mass transfer, i.e. the concentration gradient the wetted interfacial area the mass transfer coefficient
54 The driving force of the system can be increased by shifting the systems from equilibrium. In general, increasing the temperature causes shifts from equilibrium. When considering the product of the wette d area and the mass transfer coefficient, ( KLaw), the term is known as the overall mass transfer coefficient. This term will depend on the system and chemical characteristics. All these factors are used and optimized in the design of mass transfer operatio ns, such as strippers and absorbers. Packed towers aerators (PTAs) are mass transfer units commonly used for the stripping of volatile compounds from the aqueous phase and absorption of gases from the gas phase. Because their principles of operation have been implemented on the design of the reactor used for this research, some of their design considerations are included below. Mass Transfer O perations: Packed Tower Aerator (PTA) PTAs are devices that consist of cylindrical columns that treat air and water flowing in a countercurrent mode. These towers are packed with inert solid shapes called tower packing that provide large interfacial areas for the transfer of chemicals between the phases. PTAs can have different types of packing which include random pac king, stacked packing and structured and ordered packing. Random packings are those that are dumped randomly in the tower and their sizes go usually up to 3 in. Stacked packings, which can have sizes between 2 and 8 in., have to be placed in the tower by hand. Similarly, structured packings consist of entire ordered units that are placed inside the tower. The most common type of packing in PTAs is random packing. They tend to be more inexpensive, and they have high interpacking as well as intrapacking poros ity. Higher porosity corres ponds to lower pressure drops through the tower, thus decreasing the overall cost of the PTA ( McCabe et al., 2005 ). The equations used to design PTAs, E quations 2 3 9 to 2 4 2 we re derived under the assumptions that the influent gas stream is free of the contaminant of concern and the flow in the
55 tower is plug flow for both gas and liquid streams so that there is a flat velocity profile across the tower and similar residence time. The plug flow assum ption might not be valid for towers experiencing a lot of channeling, which could be a consequence of nonuniformity or maldistribution of the packing. Similarly, the assumption does not hold true everywhere in the tower since there are different gas flow resistances at the walls or relatively dry packing as compared to the internal body of the tower. Possible axial dispersion can cause the molecules to move at different speeds as well as the back mixin g caused by other flow flowing in opposite direction, r esulting in deviations from the assumptions ( Treybal, 1980). It is well established that the height of packed towers determines the mass transfer efficiency while the towers cross section determines the capacity of the towe r, i.e. the amount of gas that can be treated by the system (Sherwood et al., 1938). Accordingly, PTAs design involves the calculation of two main terms, the height of transfer units ( HTU) and the number of transfer units ( N TU ). NTU is dependent upon the driving force, i.e. the influent ( Co) and effluent (Ce) concentrations of the contaminant of concern in the aqueous phase, as well as the stripping factor ( R ). NTU represents the number of hypothetical stages required for tre atment, and corresponds to the difficulty in removing the solute from the liquid phase. The HTU value, dependent on the liquid flowrate (QL), the cross sectional area of the tower ( A ) and KLaw, characterizes the mass transfer efficiency from the liquid to the gas phase. Z = HTU NTU (2 39) HTU = QL / (A KL aw) (2 40) R = ( QG H) / QL (2 41) 0(1)1 ln 1eC CR R NTU RR (2 42)
56 When the inf luent air stream is contaminated with the contaminant of concern, the design equations cannot be simplified so the complete expression (Equation 2 4 3 ) to determine the height of the packing ( Z ) in the tower has to be used. 0() ()me LlmLCC Z KaDF (2 43) The term Lm is the liquid velocity and DFlm refers to the log mean of the driving force of the reaction which is the difference in concentration at some time t and at equilibrium ( Ct Cs). This value is constantly changing with depth because the bulk liquid concentration at time t ( Ct) is changing. The log mean of the driving force is then given by Eq uation 2 44 where the subscripts o s and e represent the influent, equilibrium (as calculated from the Henrys Law) and eff luent concentrations of the contaminant of concern in the aqueous phase. 00 00()() () ln ln ()eSeS lm S e eSDFDFCCCC DF DF CC DF CC (2 44) Mass transfer c oefficients used for the design of PTAs are obtained from pilot studies when available ( Kavanau gh and Trussell, 1980). However correlations can be helpful to extrapolate real data to other conditions that are different from the experimental conditions. KLaw depends on fluid properties, flow rates and the type of packing. Therefore, experimental pil ot results can be applicable to full -scale results only under the same packing, chemicals and flowrate conditions ( Treybal, 1980). When KLaw is obtained experimentally, errors are more likely to occur for low stripping facto rs. The sensitivity of KLaw measurements decreases as the stripping factor increases. This effect is speculated to occur due to the uncertainty on the values of H which has a lower effect in the stripping factor equation ( R = (QG H) / QL) as the air to w ater ratio increases ( Lamarche and Droste, 1989). Furthermore, the tower height calculated for
57 a full -scale system based on the KLaw obtained from a pilot -scale tends to be a conservative value, meaning that the actual tower usually operates at higher efficiencies. A full -scale system will usually have a larger tower diameter, so the ratio of tower diameter to nominal packing diameter will be greater. An optimum value for this ratio is 12 or greater. Increasing this ratio dec reases the chann eling or short circuiting of the water down the tower which results in higher removal ( Hand et al., 1999). There are some factors that can produce errors in the determination of mass transfer coefficients. One of those factors is the end e ffect of the tower. End effects refer to the removal of volatile chemicals above and below the packing as a result of the air contacting the contaminated water but without the assistance of the packing. Since KLaw represents the mass transfer due to the pa cking section of the tower, significant end effects can cause errors in experimental estimates of the mass transfer coefficients. Conventionally, mass transfer inside and outside the tower is assumed to be additive. Therefore, end effects are usually calcu lated by plotting NTU versus packing height ( Z ). A positive yintercept of the plot is considered as the number of transfer units equivalent to the end effects. This value can be then subtracted from the measured NTU to calculate the actual KLaw (Hand et a l., 1999). Roberts et al., however, showed that the mass transfer due to the packing and the one attributed to the e nd effects occurred under different flow condition; therefore, the NTU values for both cases should not be additive. They explained that a m ore reliable way to corre ct the value of KLaw due to end effects was to measure the NTU in a tower without packing ( Roberts et al.1985). When pilot studies are not available, there are correlations in the literature that al low calculations of mass transfer coefficients. There are 3 well known models for estimating the mass transfer coefficients in air stripping operations: the Sherwood and Holloway model which
58 assumes that the gas phase resistance is negligible, the Shulman et al. and O nda et al. models which account for both, the liquid and gas phase resistances ( Lamarche and Droste, 1989 ). The Onda et al. (1968) correlation is the most widely used for PTA desig n. Additionally, this correlation has been validated by many pilot and full -scale studies ( Roberts el al., 1985 and Hand et al.,1986). According to Lamarche and Droste, the Onda et al. correlation gives a better estimate of the mass transfer coefficients as compared to the Sherwood and Hollow ay and the Shulman et al. correlatio ns. In their study, they found that the Sherwood and Holloway correlation seemed to overpredict the experimental KLaw values probably because it neglec ts the gas phase resistance while the Shulman et al. model, in general, overpredicted high KLaw values while it underpredicted low KLaw values. Th e Onda correlation is given in E quations 2 45 through 2 47. This correlation estimates the wetted area avail able for mass transfer and uses that value to determine the local mass transfer coefficients for the gas and liquid phase s 0.1 0.2 0.05 0.75 22 21exp1.45cmmt m wt tLL LtLLaL aa aga (2 4 5 ) 2 1 0.5 3 3 0.40.0051m LL L tp wLLL LL k ad aDg (2 46) 1 0.7 3 25.23g m g tg tp tgggG kaD ad aD (2 47) Based on this correlation, the wetted area is dependent on the total area of the packing ( at) and the liquid loading rate ( Lm) while the local mass transfer coefficients incorporate the nominal packing diameter ( dp) and the air loading rate ( Gm). The values of Lm and Gm depend on the operating conditions of the system, i.e. the air and water flowrates and the chemical being
59 stripped. Values of dp an d at are characteristics of the packing material. Therefore, an important part of the optimization design for PTAs involves the selection of flowrates as well as the packing material ( Nirmalakhandan et al.,1987). The Onda et al. correlation is valid for Lm values between 0.8 and 43 kg/m2/s (1.1 to 63 gpm/ft2) and Gm val ues between 0.014 and 1.7 kg/m2/s (2.206 and 267.9 cfm/ft2). Equations 2 46 and 2 4 7 were correlated for Raschig rings, Berl saddles, spheres and rods made of ceramic, glass, and polyvinylchloride with nominal packing diameters between 3/8 and 2 in. The co efficient 5.23 shown in the gas side mass transfer coefficient (Equation 2 -4 7 ) is valid for nominal packing diameters greater than 15 mm; packing with smaller diameters were best fit using a coefficient of 2.0 (Onda et al., 1968) Deviations from the experimental range of values are not advisable when using the Onda correlation, though Lamarche and Droste found good agreement between experimental and calculated values for other packing mater ials. Table 2 1. Oxidation power of various species commonly used in treatment applications (Adapted from Three Bond, 2004). Reactive Species Oxidat ion Potential (V) Fluorine 3.03 Hydroxyl radical 2.80 Atomic oxygen (singlet) 2.42 Ozone 2.07 Hydrogen peroxide 1.77 Perhydroxyl radical 1.70 Permanganate 1.69 Hypobromous acid 1.60 Chlorine dioxide 1.56 Hypochlorous acid 1.49 Hypoiodous acid 1. 46 Chlorine 1.36 Bromine 1.09 Iodine 0.73
60 Table 2 2 Properties of Degussa P25 (Degussa Technical Bulletin, 1990) Degussa P25 Properties Value BET surface a rea (m 2 /g) 50 Average primary particle s ize (nm) 21 Band g ap: anatase, rutile (eV) 3.29, 3.05 Point of z ero c harge pH = 6.0 CONDUCTION BAND eh+ VALENCE BAND UV radiation < 388nm Adsorption Adsorption (H2O) Reduction 10-3s Oxidation 10-8s Adsorption (Pollutant) POLLUTANT D E G R A D A T I O N E Electron Energy Eg = 3.2eV Recombination 10-9 s Generation 10-15 s Figure 2 1. Energy band diagram of a spherical titania particle showing the basic reaction mechanism of TiO2 photocatal ysis (Adapted from Herrmann, 2005).
61 Figure 2 2. Structures and properties of rutile and anatase titanium dioxide (Obtained from Linsebigler et al., 1995 ). Figure 2 3. Linkages of SiO2 tetrahedras (Obtained fro m Hench and West, 1990 )
62 Figure 2 4. Silanol groups on the silica surface (Obtained from Legrand, 1998). Air-Water Interface Air Film Water Film Bulk Water Phase Bulk Air Phase ybysys *Cs *CsCb Figure 2 5 Diagram describing the equilibrium partitioning of a contaminant betw een the air and water phases using the two film theory.
63 CHAPTER 3 MATERIALS AND METHOD S Pilot -S cale Reactor: Two P hase Photcatalytic Oxidation Tower (TPOT) The two -phase photocatalytic tower (TPOT) was designed to simultaneously treat both air and water c ontaminated with VOCs, namely toluene. In some cases, however, the TPOT was used to test the removal of toluene in the gas or aqueous phases individually. The countercurrent flow reactor, built by Analytical Research Services (ARS), consisted of a clear PV C cylindrical column of 20.3 cm in diameter and 1.8 m in leng th. The tower was equipped with a liquid inlet and distributor in the form of a spray nozzle at the top; a gas inlet and distributor in the form of a perforated plate at the bottom; liquid and ga s outlets at the bottom and top, respectively; three gas sampling ports vertically aligned 30.5 cm apart on the wall of the column; five 75W UV lamps emitting UV -C radiation (Phillips, TUV 64T5 4P SE UNP) covered by quartz envelopes; and packing o ccupying 1.2 m of the total height of the tower. The packing material wa s supported by the bottom perforated plate. CAD drawings of the reactor as well as its dimensions are shown in Figures 3 1 and 3 2 respectively. The colored fluid connections illustrated in Fi gure 3 1 represent the inlets and outlets for the liquid phase (in blue) and gas phase (in red). The reactor design was based on criteria for the design of typical packed towers used as air strippers or absorbers. It is well established that the height of packed towers determine s the mass transfer efficiency while the towers cross section determines the capacity of the tower. Since the reactor was designed to improve and maximize the mass transfer between air and water phases as well as the diffusion from the bulk to the external surfa ce and pores of the catalyst, the desired height had to be long to provide enough contact time and significant changes over reasonable sampling periods. The main factor controlling the diameter of the reactor was the diameter of the packing material. In order to minimize channeling or short circuiting of the water down the wall
64 of the tower, the ratio of tower diameter to nominal packing diameter should be greater than 12 (Treybal, 1980). Based on this criterion, a small packi ng material wa s desirable to minimize the diameter of the tower. The smallest commercially available tower packing found was plastic high flow rings with a nominal packing diameter of 15 mm (Rauschert Industries Inc.) which was used to compute the preferr ed minimum tower diameter. The reactor packing consisted of silica titania composites (STCs) commingled with the 15 mm nominal diameter plastic high flow rings shown in Figure 3 3. The composites were made of titanium dioxide embedded in a silica matrix w hich served as the TiO2 catalyst support which could be casted in any desired cylindrical pellets size. The commingled packing was chosen over the only STCs packing in order to decrease the pressure drop across the reactor, increase the interfacial area av ailable for mass transfer and i ncrease the penetration of UV irradiance The packed section consisted of about 16% by bulk volume of STCs. The remaining volume was filled wit h plastic high flow rings which had a void fraction of 91% and a specific surface area of 313 m2/m3. The porosity of the tower, determined by dividing the volume of water required to completely fill the packed section of the t ower by the empty volume of the same section was 0.7 6 The mass percentages were 55% and 45% of the total mass in the reactor for STCs and high flow rings respectively. A picture of the actual reactor packed with the palls rings and STCs during a typical experimental run is presented in Figure 3 4. The lamps selected for the reactor emit short -wave UV radiation wi th a peak at 253.7 nm. Unlike most UV lamps emitting UV C radiation, these lamps do not generate ozone because the lamp glass filters out the 185 nm ozone forming line (See Figure 3 5 ). As shown by Equations 3 1 and 3 2, when an oxygen molecule absorbs li ght with energy at 185 nm, two oxygen atoms are produced. These atoms react with another oxygen molecule to
65 yield ozone, which is a strong oxidizing species. For the purpose of understanding the mechanisms involved in the degradation of toluene using TiO2 photocatalysis, other extraneous factors that might contribute to toluene oxidation needed to be controlled, as is the case of ozone forming potential. O2 2O (3 1) O2 + O + M O3 + M (3 2) The reactor setup is shown in Figure 3 6. During a typical experimental run, aqueous phase toluene (Fisher, Certified ACS grade, >99.5% purity) at concentrations ranging from 150 ppb to 2500 ppb entered the reactor at the top inlet. The water was pumped from a 200 L feed tank to the top of the reactor by a centrifugal pump (AMPCO, Model KC2). Gas phase toluene entered the reactor at the bottom inlet of the tower. Concentrated toluene (1001000 ppmv) from a compressed cylinder (Airgas) was diluted to approximately 1 to 10 ppmv by mixing it with air from an air compressor (Coleparmer, Model 07047) before entering the reactor. Samples were taken from the inlets ( Yo, Co) and outlets ( Ye, Ce) for air and water as well as the 3 gas sampling port s on the tower wall. Preliminary studies performed using the reactor packed with high flow rings only and single phase contaminant, showed that there were no significant losses of toluene in either the gas or aqueous phase due to piping, volatilization or photolysis. Additionally, the reactor show ed high efficiency as an air stripper and absorber. The experimental results for these preliminary studies are shown in Appendix A Bench S cale Reactor for Gas Phase Studies Several studies on the degradation of gas phase toluene were carried out at the bench-scale. The reactor and setup used for these experiments, shown in Figure 3 7, consisted of a Pyrex annular reactor of 25 mm annulus with a 25 mm quartz envelope placed a t the center of the
66 reactor packed with kn own masses of STCs to produce heights of approximately 4 to 8 cm. A porous glass frit at the bottom of the cylindrical reactor served as bed and envelope support. The STCs were packed on top of glass beads to improve gas distribution before reaching the c atalyst. The porosity of the packed bed, determined by dividing the volume of the voids in the packed section by the empty volume of the same section was about 0.40. The same type of lamp used in the pilot reactor (Phillips, TUV 64T5 4P SE UNP) was used fo r the bench-scale studies. The intensity was measured using a digital UVX radiometer (UVP) connected to a 254 nm sensor (UVP, Model UVX25). The intensity measured at the outside wall of the envelope was 15 mW/cm2 while the intensities measured through the packing were 91 W/cm2 at 12.5 mm and 1.7 W/cm2 at 25 mm away from the outside wall of the envelope. The gas stream flowed continuously through the reactor in a single pass configuration. During a typical run, toluene from a compressed cylinder ( Airgas 1 00 ppmv) was diluted by mixing with breathing air from another compressed cylinder to a concentration of about 8 ppmv (30 mg/m3). The desired relative humidity (RH) was achieved by bubbling the air through a flask containing deionized water (DI) before mix ing with toluene. The investigated range of RH at room temperatures of 23 2oC varied from 13%, obtained by bypassing the water bubbler, to 90%. The inlet and outlet RH and temperatures were measured using a Hygrothermometer (EXTECH Instruments, 45320) a nd the temperatures inside the reactor and at the lamp envelope were monitored with thermocouples. Adsorption only experiments were performed at room temperature with the UV lamp turned off. For PCO experiments, no adsorption was allowed to take place bef ore turning the lamp on, so the results show the combined effects of adsorption and PCO. For these experiments, the lamp and reactor were allowed to warm up for at least 1 h before starting the gas feed until they reached steady state temperatures correspo nding to 50
67 3oC and 83 2oC for the reactor and lamp envelope, respectively. The gas volumetric flowrates (QG) were varied from 12.5 to 26.7 cm3/s which produced superficial gas velocities ( UG), defined as the volumetric flowrate divided by the cross -se ctional area of the reactor normal to the flow (A ), across the annulus of 0.32 to 0.68 cm/s. The experiments were run until steady state conditions were achieved. Influent and effluent samples were collected at different time intervals using 1 L tedlar bag s (SKC Inc.) and analyzed the same day. Control experiments were performed in the absence of the catalyst to ensure that toluene did not undergo photolysis or other losses due to adsorption to the systems appurtenances. UV Irradiance Distribution in the Reactor Providing a uniform ultraviolet (UV) radiation throughout the reactor is of great importance because the energy emitted by the photons activates the catalyst, which is the first step required for the degradation of toluene The silica titania com posites (STCs) were commingled with the plastic high flow rings to allow further UV light penetration in the reactor and reach as many catalyst sites as possible. To demonstrate the advantage of the commingled packing compared to only STCs and to determine the UV distribution in the reactor, the UV -C i rradiance ( = 254 nm) was measured at different distances from the UV lamp and through di fferent types of packing. The irradiance was measured using a digital UVX radiometer (UVP, Part No. 97 00150) implemented with a 254 nm sensor (UVP, Model UVX 25). Initially, the irradiance was measured as a function of lamp length to ensure that the output of the lamp was constant throughout its length. Figure 3 8 shows the UV irradiance as a function of lamp length averaged for measurements taken using two different lamps. The lamps were allowed to warm up for 15 minutes before the measurements were taken to ensure that the temperature would not influence the intensity measurements The UV irradiance was measured at 2 differ ent distances
68 from the lamp ( X ), 0.95 cm and 6.25 cm The results show that the irradiance is constant throughout the length of the lamp, except for the first 10 cm close to each end of the lamp, which showed lower irradiance The average U V intensities measured were 13.6 mW/cm2 and 3.9 mW/cm2 for the distanc es of 0 .95 and 6.25cm respectively. The intensities were averaged for lamp lengths between 20 cm to 130 cm, so that the lower end intensities were not included. The irradiance greatly decrease d when the detector is placed further from the lamp. After veri fying that the irradiance d id not significantly change throughout the lamp length, the UV intensity was measured using different packing materials. The test setup is shown in Figure 3 9 The packing materials that were tested included: (a) air (as the blan k), (b) high flow rings (c) high flow rings comm ingled with STCs, and (d) STCs. The UV irradiance through air was determined using a cardboard box in several configurations: (1) the cardboard box, (2) the same box with the walls covered by aluminum foil but uncovered at the top, and (3) the aluminum foil covered box also covered at the top. The results for the irradiance as a function of distance to the lamp through air are shown in Figure 3 10. As observed in Figure 3 10, the UV intensity decreases expone ntially with the increase of the distance from the lamp when there is no packing between the UV C sensor and the lamp. Additionally, the irradiance is higher when the box is covered by aluminum foil due to increase in reflection of the light in the presenc e of the foil. Generally, the intensity should decrease as a function of the inverse square of the distance. This is not the case for the results shown above for air. This can be explained by the errors produced by the radiometer readings. The number that the radiometer reads out is a function of the spectral response of the radiometer, coupled with the spectral output of the emitter and taking into account the radiometer spatial response. The spectral response and output were calibrated for the 254 nm lam p used, but the spatial response becomes important when
69 measuring the irradiance at a surface near an extended emitter, such as the long 147 cm lamp used in the tower and these experiments. The sensor gives accurate measurements for point sources that are small in comparison with distance, and they produce lower measurements when used to measure extended light sources at relative small distances (UVP Inc.). Experiments through the other packing styles were performed by using the cardboard box without cover ing the walls with foil to better model the surface of the actual reactor which is PVC and thus nonreflective. The results for these experiments are shown in Figure 3 12. The results presented in this figure include data for small STCs which refers to S TCs of 3 mm in diameter by 5 mm in length that have been used in other studies. The results for these STCs were obtained from the work of Stokke (2008). Placing any type of packing between the UV lamp and the sensor decreased the UV i rradiance dramatically as shown in Figure 3 11 by the large differences between the irradiances obtained through air and through differe nt packing ma terials. The high flow rings allowed the highest UV penetration through the packing while the small STCs allowed the lowest pene tration. UV penetration of the commingled packing was lower than for the packing of high flow rings only ; however, the commingled packing allowed the UV to penetrate up to a distance of approximately 6 cm which is significantly greater than the 2.5 cm perm itted by either type of STCs. Not only was the UV penetration greater for the commingled packing, but also the intensity measured at a given distance was orders of magnitude larger up to about 3 cm from the lamp These results show that packing the reactor with the commingled high flow rings and STCs provide s an advantage since the UV lamps can be further spaced as compared to only STCs, and a better UV radiation distribution can be achieved. These results are very important since the number of lamps in the reactor relates directly to energy costs and temperature increase
70 in the reactor, two factors that are of high importance in large scale applications. The actual placement of the UV lamps in the TPOT is shown in Figure 3 13. The UV irradiance distribution in the reactor was calculated based on the results shown in Figure 3 12. The trend observed for the UV irradiance as a function of distance to the lamp for the commingled packing was fit using two models for different sections of the curve, as shown in Fi gure 3 14. The UV irradiance for distances between 1.3 and 2.5 cm was fit using a linear regression between those two points, given by Equation 3 3 The rest of the graph, i.e. distances to the lamp between 2.5 and 6.3 cm was fit with a third degree polyno mial presented in Equation 3 4 where X is the distance to the UV lamp in cm and the UV irradiance ( I ) is given in W/cm2. I = 1490.03X + 4092.33 (3 3 ) I = -4.847X3 + 96.534X2 634.020X + 1374.7 (3 4 ) By integrating the regression equations and dividing the integral by the difference between the integral bounds (Equation 3 5 ), the average UV intensity was determined for annuli of about 1.3 cm from the lamp increasing up to 6.3 cm as shown in Tabl e 3 1. ()b afxdx Average ba (3 5 ) Any packing placed at a distance greater than 6.3 cm. was considered to receive no UV irradiation. The UV intensity from 0 and 1.3 cm. from the lamp was considered to be the intensity at 1.3 c m since no other smaller distances were measured. The intensity at distances closer than 1.3 cm to the lamps is expected to be greater than the assumed intensity, but this value was used to obtain a conservative estimate of the irradiance. Using the result s in Table 2, the graph in Figure 3 15 was developed showing the UV intensity distribution in the tower for any cross -section of the reactor. This figure shows that 70% of the cross -sectional area of the
71 tower receives some UV radiation while the rest of t he area does not seem to receive any irradiation. Using the intensities calculated in Table 3 1 and the areas determined for each 1.3 cm annulus around the UV lamp, the total intensity in each annulus was calculated. For the sections where annuli overlappe d, the sum of the intensities of the overlapping annuli was used. The average intensity for the section of the surface that is irradiated was then multiplied by the length of the lamp and this value was divided by the total mass of irradiated STCs inside t he reactor 2. Synthesis of the Silica Titania Composites (STCs) The silica titania composites (STC s) were prepared by a sol -gel method initially develo ped by Powers (1998), which used nitric acid and hydrofluoric acid as catalysts to decrease the gelation time by increasing the hydrolysis and condensation rates. The actual formula used to prepare STCs was obtained from Londeree (2002), who adapted Powers (1998) formula to use tetraethylorthsilicate (TEOS) as the silica precursor instead of tetramethylorthosilicate (TMOS) and incorporated the TiO2 catalyst in the silica matrix. For this research the fo rmulation was further modified to decrease the volume of ethanol by 50%. By this formulation, nanopure water, ethanol (Fisher, 200 proof), TEOS, 1M nitric acid (diluted from Fisher, Certified ACS) and 3%v hydrofluoric acid (diluted from 49% Fisher Certifie d ACS) were mixed in volume ratios of 12:6.25:9.25:1:1, respectively. While mixing, 115 mg of TiO2 (Degussa P25) were added to the mixture per mL of TEOS to obtain STCs containing about 30%wt TiO2. The volumes of the chemicals used in the formulation to pr epare one batch of STC are shown in Table 3 3 ; the components were added in the same order as presented in the t able. The titanium dioxide used for all synthesis and experiments had a measured surface area of 49 m2/g, as measured by a Quantachrome Autosorb 1C -MS gas sorption analyzer (NOVA 2200e)
72 The content of each batch was stirred in a Nalgene container and mixed for 20 minutes to allow gelation to occur. The resulting solution was pipetted into Costar polystyrene 24 well assay plates. Since the pellets in the reactor are bigger than those originally prepared by Londeree (2002), the drying schedule was also modified. The complete drying schedule is presented in Figure 3 16. The assay plates were covered with tightly sealed plastic bags to prevent premature evaporation. The aging process included 3 stages: sealed STC drying at room temperature for 3 days followed by 3 days in the oven at 73oC, and continued by another 3 days at 73oC but with the trays uncovered to allow some evaporation. Higher aging tempe ratures than 73oC would decrease the aging time, but since the casting molds are made of polystyrene, the maximum recommended temperature was 80oC. The maximum temperature (80oC) was not used b ecause it was found that using such temperature quickly damaged the assay plates. After aging, the pellets were placed in high temperature resistant trays and exposed to a series of heat treatments using a programmable oven. Initially the temperature was ramped from 25oC to 103oC at a rate of 2oC/min and held constant for 18 hrs. Next, the temperature was increased to 180oC and held for 6 hrs and slowly decreased to room temperature at 2oC/min The last heat treatment in the process was curing at 450oC for about 2 hrs. C haracterization of the STCs Several representat ive samples (about 20) of multiple batch es prepared to make STCs were obtained and analyzed to characterize the composites for their surface area, average pore size and pore volume. Nitrogen adsorption desorption analyses were conducted using a Quantachrom e Autosorb 1C -MS gas sorption analyzer (NOVA 2200e). During the analyses, mortar and pestel crushed STC samples were outgassed for 24 hours at 180 C. The Brunauer, Emmett, and Teller (BET) model was used to determine the specific surface area of the STCs for the nitrogen adsorption/desorption data obtained from the isotherms (P/Po = 0.1 to 0.3) The total pore
73 volume was determined by nitrogen adsorption at a relative pressure of P/Po = 0.99. The average pore size of the composites was calculated using Equation 3 6 d = 4*Vp/S (3 6 ) where d is average pore diameter, Vp is total pore volume, and S is specific surface area. The pore size distribution (PSD) was computed using the method proposed by Barret, Joyner and Halenda (BJH), wh ich assumes cylindrical pore geometry (Barret, 1951). For PS D calculations, desorption isotherms were used since they are generally accepted to be more appropr iate than adsorption isotherms for PSD analysis. A major concern about using silica embedded comp osites and casting them in a pellet shape wa s the loss of activity related to the loss of accessible catalyst surface area due to the potential entrapment of the TiO2 molecules when supported by silica. To determine if the activity of the STCs would remain unchanged in the mixed oxide and the pellet form, the TiO2 surface area accessible in the composite was analyzed using the functionalization procedure proposed by Marugan et al. (2007). This method measures the amount of phenylphosph onic acid ( C6H7O3P PP A) that is reacted with titania to determine the surface area by assuming that the phosphorus (P) atoms incorporated in the composite during reaction are proportional to the TiO2 accessible surface area. PPA is used in the procedure because it can selectiv ely react with titania in the presence of both silica and titania when the PPA solution is aqueous (Mutin et al., 2004). The method used consisted of mixing 200 mg of the STCs (or Degussa P25) with 200 mL of a 4 mM PPA solution prepared with a 4:1 CH3OH:H2O (v/v) mixture. The PPA solution was allowed to mix with the composite material for at least 24 h under vigorous stirring to ensure enough time for the PPA molecules to bond with all the accessible TiO2 likely through Ti O P bonds as proposed by Mutin et al. (2004). The solids were then separated by filtering with a 0.45 m
74 nylon filter and rinsed with at least two times the volume of the sample to ensure complete removal of PPA molecules that were not attached to the surface. The solids were finally d ried for 24 h at 110oC. The catalyst was then dissolved by hydrofluoric acid attack. The catalyst digestion procedure was carried out by mixing about 42 mg of the dried sample in a digestion vessel per liter of a solution containing aqua regia (3:1 solutio n of 12 M HCl and 15 M HNO3) and 29 M HF in volume ratios of 1:5. After 24 hrs of digestion, 8.3 mL and 7.1 mL of H3BO3 and nanopure water, respectively, were added per mL of digested solution. The resulting solution was analyzed for P using a Perkin Elme r Plasma 3200 Inductively Coupled Plasma Atomic Emission Spectroscopy (ICP -A ES) system at a wavelength of 214.914 nm. The P content was correlated to the accessible TiO2 surface area of the composite using the following two formulas. P atoms/nm2 TiO2 = [(P concentration) (mg P/L)] [(TiO2 loading)1 (L/mg TiO2)] [(BET SATiO2)1 (gTiO2/m2)] [(1018) (m2/nm2)] [(30,973.76)1 (mol P/mg P) [(6.022 x 1023) (atoms P/mol P)] m2 TiO2/ g STC = [(P concentration) (mg P/L)] [(STC loading)1 (L/mg STC)] [(P surface coverage)1 (nm2 TiO2/ atoms P)] [(6.022 x 1023) (atoms P/mol P)] [(30,973.76)1 (mol P/mg P)] [(1018) (m2/nm2)] The first formula determines the surface concentration of P atoms in the surface of pure TiO2 Degussa P25 (no silica pr esent) from the P concentration obtained by the ICP and the measured BET surface area of Degussa P25 (49 m2/g). The resulting value, in atoms of P per nm2, was used in the second equation to find the accessible TiO2 surface area of the STCs. Analytical Met hods for Toluene and Oxidation Byproducts Since two phases of toluene, gas and aqueous phases, were investigated in this research, the analytical methods were developed such that minimum changes would have to be incorporated when analyzing either phase of the compound. Both phases of toluene were
75 analyzed using USEPA Method 524.2: Measurement of Purgeable Organic Compounds in Water by Capillary Column Gas Chromatography/ Mass Spectrometry. A ThermoQuest Trace GC/MS with a Te kmar 3100 Purge & Trap extraction system was used in this research. This GC/MS was implemented with a Rtx VMS capillary column (Restek, cat # 19919). The Purge & Trap apparatus used a Vocarb 3000 trap to capture VOCs Aqueous phase toluene samples were collected in 40 mL vials with no headspace and stored at 4oC for no longer than 2 weeks to maintain the samples integrity, based on the guidelines provided by USEPA Method 524.2. When analyzing these samples, 5 mL of the sample we re introduced in the Purg e & Trap where the sample wa s purged and the contaminants captured by the trap using the conditions presented in Table 3 3 to be finally desorbed to the GC/MS for separation and detection. Since minimum changes were desired in the analytical system config uration, the extraction of gas samples was very similar to the extraction of aqueous samples. The gas phase samples were collected in tedlar bags and analyzed in a timely manner since s tudies have shown that the recovery of samples stored in tedlar bags decreases significantly with time ( Andino and Butler, 1995; McGarvey and Shorten, 2000). For this research, it was determined that the samples should be analyz ed within four days from collection because keeping the samples for any longer caused significant losses of the samples due to possible adsorption to the bags. Additionally, cleaning the bag immediately after use was also of utmost importance to keep the bags free of contamination. The cleaning procedure developed that worked best consist ed of flushing the bags 10 times with air followed by 10 times of nitrogen flushing. The typical process for analyzing gas phase samples consisted of initially injecting 5 mL of DI water in the P&T appar atus. The purpose of the water wa s to provide the same
76 humidity con ditions for all samples. Then, 5 mL of the gas sample stored in the tedlar bag were extracted with a gas tight syringe (Hamilton SampleLock, Series 1705). At this point, the purge schedule in the P&T wa s initiated. While purging, the injection valve in the P&T wa s opened and the gas sample in the syringe was injected in the apparatus slowly to avoid overloading the trap since the optimal trap efficiency of VOCs was about 60 mL/min, and the purging gas (helium) was constantly flowing at 35 mL/min. Hence, an injection rate of about 5 mL/min was found to provide satisfactory results. After the gas injection, the valve is closed and the sample continued purging. After purging, the captured sample was desorbed to the GC/MS for analysis following the same oven tem perature schedule as for aqueous phase samples shown in Table 3 3 For all samples analyzed, the response factors (RF) were determined using an internal standard calibration, with fluorobenzene as the internal standard. The reliability of the GC/MS analysi s was verified using the percent relative standard deviation ( RSD ), which is based on the RF of a chemical. The RF and RSD wer e calculated using Equations 3 7 and 3 8 respectively. RF = (AX CIS) / (AIS CX) (3 7 ) RSD = 100 (SRF / MRF) (3 8 ) w here AX is the GC peak area of the target analyte, AIS is the area of the internal standard, CX and CIS are the concentration of the target analyte and internal standard, respectively SRF is the standard deviation of the response factors and MRF is the average response f actor. The USEPA method required RSD below 20%. For all analys e s, the RSD was always below 10%. The minimum detection limit ( MDL ) for each phase of toluene was also determined by analyzing 7 samples at the same low concentration and using Equation 3 9 MDL = SD ta=99%, n1 (3 9 )
77 w here S D is the standard deviation of the replicate analyses, t is the students t value for 99% confidence level with n -1 degrees of freedom (t = 3.143) and n is the number of replicates (n = 7). The MDL for the gas phase was 0. 5 g/L or 0.13 ppmv while for the aqueous phase was about 1 g/ L. The oxidation byproducts monitored during some of the experimental runs were benzene, benzaldehyde, benzyl alcohol, benzoic acid and phenol. The method used to detect toluene, USEPA Method 524.2, also detected benzene. The other potential oxidation byproducts were expected to be found mostly adsorbed to the catalyst surface, so an extraction procedure was used in the cases where catalyst deactivation was encountered. The extraction consisted of placing the crushed STCs in acetonitrile for 2 h. The sus pension was separated using 0.45 m cellulose acetate filters and the so lution was analyzed using a High Performance Liquid Chromatography ( Hitachi, D 7000). Table 3 1. Calculated average UV intensity through the commingled packing at different annuli dis tances from the UV lamps. X (cm) Avg. I ( W/cm2) 0 1.3 2200 1.3 2. 5 1254 2.5 3.8 187 3.8 5.1 43 5.1 6.3 1.4 Table 3 2 Chemicals used in the STC formulation to produce one batch of 140 STC with 30% by mass titania loading. Chemical Amount Nanopure water (DI) (mL) 96 .0 Ethanol (mL) 50 .0 TEOS (mL) 74 .0 1N Nitric Acid (mL) 8 .0 3% v Hydrofluoric Acid (mL) 8 .0 Titanium dioxide (g) 8. 5
78 Table 3 3 Conditions of the GC/MS and Purge & Trap used for the analysis of gas and aqu eous phase toluene. Tekmar 3100 Purge and Trap Conditions Aqueous Phase Gas Phase MCS Line temperature 40 o C 40 o C Purge time 10 min 5 min Dry purge time 2 min 0 min Trap type Carbopack B, Carboxen 1000 and Carboxen 1001 Desorption preheat temperature 220 o C 220 o C Desorption time 6 min 6 min Bake temperature 270 o C 270 o C Bake time 10 min 10 min ThermoQuest Trace GC/MS Condit ions Column Restek, Rtx VMS (fused silica) 30 m x 0.32 mm x 1.8 m Program 40 o C (4 min) to 180 o C at 10 o C/min (2 min) to 200 at 20oC/min Carrier Helium Detector Ion trap, MS, m/z = 45 260 (0.6sec/scan) Figure 3 1. CAD drawings of the reactor A) F igure depicting the final reactor design B) Illustration of the UV lamp connections at the bottom section of the reactor (ARS, 2005) A) B) A)
79 Figure 3 2 Dimensions of the TPOT reactor shown in horizontal configuration. (ARS, 2005). Figure 3 3. Picture of the two different packing materials that make up the commingled packing in the TPOT A) STCs. B ) High flow rings C ) Commingled packing: STCs and high flow rings A) B) C)
80 Figure 3 4 Photograph of the reactor (TPOT) du ring a typical experimental run.
81 Figure 3 5. Emission spectrum of UV -C lamps: A) ordinary and B) non -ozone producing UV -C lamps. Effluent water sample Influent gas sample Water Rotameter Pump Control valve Needle valve Ball valve Influent water sample Toluene cylinder Air Blower Air Rotameter Needle valve Ball valve Gas Rotameter Ball valve Ball valve Ball valve Treated Effluent water Contaminated Influent water Control valve Ball valve Effluent gas sample To fume hood Control valve Centrifugal Pump To sink Figure 3 6. Reactor setup for a typical experimental run A) B)
82 Inlet Sampling Tedlar bag Outlet Sampling To Fume Hood Reactor Rotameter 100ppmvToluene Cylinder Air Cylinder Water Bubbler 30.5 cm 0.64cm gas inlet Pyrex reactor (75mm ID) Porous glass frit 12cmGlass beads Quartz envelope UV lamp 4-8 cm STCs packed bed Gas outlet Removable cap (a) (b) Figure 3 7. Gas phase experimental setup and reactor. (a ) Schematic of the experimental setup. (b) Photocatalytic packed bed reactor 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 0 25 50 75 100 125 150 UV Irradiance (mW/cm 2 ) Lamp Length (cm) X=0.95 cm X=6.25 cm Figure 3 8 UV irradiance as a function of lamp length measured in ambient air with the UV radiometer sensor placed at 0.95cm and 6.25 cm from the lamp.
83 10cm X 20 cmUV Lamp UV sensor Packing 55 cm 70 cm UV meter Figure 3 9 Top view of the box setup used to measure the UV intensity as a function of different packing materials at different X distances from the lamp. 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 0 5 10 15 20 25 UV Irradiance ( W/cm 2 ) Distance to the lamp, X (cm) Opaque box Foil covered box Foil covered/reflected box Figure 3 10. UV irradiance as a function of distance to the lamp through ambient air using the set up of Figure 3 9 for different extents of reflective surfaces.
84 1000 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 0 1 2 3 4 5 6 7 8 UV Irradiance ( W/cm2) Distance to UV Lamp, X (cm) Air (opaque box) Pall rings Pall rings & STCs STCs (9x8) Small STCs(3x5) Figure 3 11. UV intensity as a function of the distance from the UV lamp through different packing materials. 100 200 500 800 1100 1400 1700 2000 2300 2600 0 1 2 3 4 5 6 7 8 UV Irradiance ( W/cm2) Distance to UV Lamp, X (cm) Pall rings Pall rings & STCs STCs (9x8) Small STCs(3x5) Figure 3 12. Enlarged version of Figure 3 11
85 Figure 3 13. Actual placement of the five UV lamp s inside the TPOT measured after packing the reactor with the commingled packing. y = 1,490.03x + 4,092.33 R = 1.00 y = 4.85x 3 + 96.53x 2 634.02x + 1,374.70 R = 1.00 0 500 1000 1500 2000 2500 0 1 2 3 4 5 6 7 8 UV Irradiance ( W/cm 2 ) Distance from UV lamp through commingled packing, X (cm) Data Linear section Curvilinear section Figure 3 14. Fit of the data obtained for UV intensity through the commingled packing
86 1 2 3 4 5(1) 0 0.5 in = 2200 W/cm2(2) 0.5 1.0 in = 1253 .8 W/cm2 (3) 1.0 1.5 in = 186 .5 W/cm2(4) 1.5 2.0 in = 42.5 W/cm2(5) 2.0 2.5 in = 1.4 W/cm2 Figure 3 15. UV irradiance distribution inside the reactor. Inner most (darkest) circles represent the UV lamps. The concentric circles around the lamps show the UV intensity for every 1.3 cm annuli, corresponding to the intensities in Table 3 1. Note intensities from each lamp are not added in this figure.
87 73 103 180 3 days 3 days 3 days 2oC/min 18hrs 6hrs 2oC/min 2oC/min 25 10 days Temperature (oC)Time Covered trays Uncovered trays Figure 3 16. Drying sch edule for cylindrical STCs of 9 mm in diameter by 8 mm in height used to pack the TPOT (Figure not drawn to scale).
88 CHAPTER 4 CHARACTERIZATION OF SILICA TITANIA COMPOSITES The properties of the STCs used for packing the pilot -scale reactor are shown in Ta ble 4 1 The STCs have an average pore diameter of about 1 5 2 a BET surface area of 2 50 m2/g and a pore volume of 0.94 cc/g. A typical nitrogen adsorption -desorption isotherm is shown in Figure 4 1. The STCs exhibit ed a Type IV isotherm with a H1 hystere sis loop (as defined by the IUPAC (Sing et al., 1985)). These isotherms are characteristic of mesoporous materials with uniform spheroid particles that tend to be nonporous or possess a narrow pore size range. The narrow pore size distribution (PSD) was co nfirmed by the PSD analysis shown in Figure 4 2 several for different batches of the same STCs formulation plotted as the differential pore volume as a function of pore diameter The PSD is an important characteristic of the composite catalyst material si nce it can affect the diffusion of toluene in the STCs during adsorption and surface reaction and also the desorption of byproducts after oxidation. The STCs showed a unimodal distribution of pore diameters with more than 90% of the pore volume attributed to pore diameters in the range of 70 and 200 T he peak s of the PSD s shown in Figure 4 2 do not correspond to the hydraulic pore diameter determined for the batches presented in Figure 4 2 as determined by equation 3 6 The BJH method for determining the PSDs of the STCs assumes the ideality of perfectly cylindrical pores, and this ideality did not apply to the STCs. In addition there is also the possibility of the presence of networks within the pellets that are not accounted for by the BJH method. Acc ordingly, the hydraulic pore radius was used as the characteristic pore diameter. Another important characteristic of the STCs that was analyzed was the accessible area of TiO2 in the composite material. As discussed in Chapter 3 the method used for this analysis uses PPA to determine the available area. Since the size of PPA molecules is greater than toluene
89 molecules ( ca. 1116 versus 48 ), the area measured by using this method should also be accessible to toluene molecules Additionally, the PSD analysis showed that most of the pores in the STCs have a diameter greater than 50 which is well above the size of PPA molecules, so potential resistance s d ue to intraparticle diffusion were expected to be insignificant during the procedure. The TiO2 acce ssible surface area analysis was performed using the composites in the form of powder (pellets crushed with mortar and pestle), cylindrical pellets as used in the pilot reactor (9mm x 8mm), and as pellets without any titania loading (silica only). The P co ntent in each of the materials analyzed and the TiO2 available surface area, presented in Table 4 2, were calculated using the formula shown in Chapter 3 The P surface coverage found for Degussa P25 wa s 2.97 atoms P/nm2. This value is comparable to the results found by other researches that have used the same procedure (Marugan et al. 2007; Mutin et al., 2004) Based on the measured BET surface area of Degussa P25 TiO2 (49 m2/g) used in the synthesis of the STCs, the theoretical TiO2 surface area expected in each STC is about 30% of that value or 14.7 m2/g STC ( since STC is about 30%w t TiO2). The results for the accessible surface area of the STCs analyzed p er gram of STC are presented in Table 4 2 and the comparison of this area per gram of TiO2 is shown in Figure 4 3 The results indicate that there was about 20% loss of active TiO2 surface area acces sible for reaction by using silica as the titania support. The decrease in surface area can be attributed to possible clumping or coating of the titania par ticles by silica during the synthesis of the STCs. There w ere no considerable differences, however, in the available area between the composites in powder and pellet form. This shows that the surface area of titania available for reaction is not affected b y the casting of the STCs as the cylindrical pellets. The low available surface area of TiO2 found for the silica
90 only composites confirms that the method measures the functionality of titania only so that the contribution of silica to the total active sur face area can be neglected. The available surface area measured for the powdered STDs is in agreement with the areas found by Byrne et al., where approximately 70% of the titania surface area was found to be accessible for reaction in STCs prepared using a similar formulation as the one used in this research. Table 4 1 Properties of STCs Property Units Value STD Diameter mm 9.1 0.4 Length mm 8.1 1.1 TiO 2 loading %wt 29.6 1.4 BET surface area m 2 /g 250 .0 43 .0 Pore diameter 152 .0 23 .0 Total p ore volume cc/g 0.94 0.06 *STD Standard deviation for at least 20 samples. Table 4 2. Phosphorus content from ICP AES results and accessible TiO2 surface area P Content P Surface Coverage Available T iO 2 SA I naccessible TiO 2 (mg P/g STC) (atoms P/ nm 2 ) ( m 2 /g STC ) (%) Degussa P25 7.49 ( 2 ) 2.97 14.7 Powder.B1 1.82 (3) 11.9 19.1 Powder.B2 1.97 (1) 12.9 12.3 Pellet.B1 1.74 ( 2 ) 11.4 22.4 Silica Only 0.23 (2 ) 1.5 NA *NA Not applicable. (#) # of samples analyzed.
91 0 100 200 300 400 500 600 700 0.0 0.2 0.4 0.6 0.8 1.0 Volume (cc/g) Relative Pressure, P/Po Adsorption Desorption Figure 4 1 Typical nitrogen adsorption/desorption isotherm for the STCs used to pack the pilot scale reactor. 0.000 0.003 0.006 0.009 0.012 0.015 0.018 0 50 100 150 200 250 300 350 400 dV(d), cc/ /g Pore Diameter ( ) B1 (153A) B2 (147A) B3 (127A) B4 (147A) B5 (160A) Figure 4 2. Typical p ore size distr ibutions of the STCs used to pack the pilot -scale reactor. The legend indicates the b atch number (B1 to B5) and the number in parenthesis refers to the measured hydraulic radius in A ngstrom s for the specific batch.
92 49.0 39.7 43.0 38.8 4.9 0.0 10.0 20.0 30.0 40.0 50.0 60.0 TiO 2 Available Surface Area (m 2 /g TiO 2 ) Figure 4 3 Comparison of the accessible titanium dioxide surface area for different STC shapes B1 and B2 refer to two different batches of STC pellets
93 CHAPTER 5 GAS PHASE TOLUENE DE GRADATION STUDIES Bench S cale Studies The objective of the work presented in this chapter was to investigate the oxidat ion of toluene using the STCs in the gas phase only. Several conditions considered to affect the oxidation of toluene, especially when present simultaneously with the aqueous phase were investigated Some of these conditions include the effects of adsorption in PCO, the effect of relative humidity, the degree of wetting of the STCs, and the residence time of the gas phase in the reactor. Many of the results presented in this section were also used to predict the performance of the gas phase end polishing section for the overall proposed treatment system. Gas Phase Toluene Studies U sing Dry STCs Adsorption of toluene in a p acked b ed r eactor The adsorption capacity of the dry STCs for toluene was st udied using the bench-scale annular reactor with a 25 mm annulus. The adsorption experiment was conducted by running toluene through the reactor in the dark until the STCs were completely saturated Based on the large surface area of the composites; adsorption was expected to play an important role during photocatalysis. During this run, no water vapor was added to the system; so the measured RH was about 13%. The toluene inlet concentration was kept relatively constant at 90 10 mg/m3. Once the STCs reached exhaustion, the UV lamps were turn ed on to allow photocatalysis to take place. T he results presented in Figure 5 1 show that th e STCs have a large adsorption capacity for toluene. For the given conditions, it took about 45 h of continuous flow for the STCs to reach exhaustion. The adsorption capacity of the STCs for toluene calculated by integration of the adsorption breakthrough curve, was about 2.5 mg toluene/g STC. Following adsorption, there was significant desorption of toluene when the lamps were turned on. Some desorption was
94 expected to occur due the increase in the reactor temperature from approximately 23oC to 50oC and t he high loading of toluene already in the STCs. After about 14 h of desorption, significant conversion of toluene was observed, but this conversion became insignificant after about 35 h of PCO. The loss of activity of the catalyst was attributed to deactiv ation of the STCs under dry conditions since a yellowish coloration of the STCs was observed at the end of the run. Many other studies of toluene PCO using Degussa P25 TiO2 have correlated this yellow coloration with catalyst deactivation ( Alberci and Jard im, 1997; Einaga et al., 2001). Note that no change in STC color was observed for adsorpt ion only, so deactivation wa s attributed to the reactions occurring during photocatalysis. PCO of toluene: Effect of w ater v apor Due to the large des orption effect s hown in Figure 5 1 the rest of the PCO experiments did not include an adsorption step prior to PCO, so the combined effects of adsorption and PCO are reported. The effect of water vapor on the removal of toluene by PCO was investigated by increasing the R H from 13% to 90%, which corresponds to water vapor concentrations of ca 3,600 ppmv (2,600 mg/m3) and 24,500 ppmv (18,500 mg/m3), respectively. The inlet toluene concentration ( Yo), flowrate (QG), and packed depth ( Z ) were kept constant during both experi ments. The results are shown in Figure 5 2 Initially, the conversion of toluene is large under both dry and high RH conditions, likely due to the combined effect of adsorption and PCO. However, under dry conditions, the outlet effluent concentration conti nuously increased to achieve a pseudo steady state corresponding to about 10% removal after 50 h of experimental run. For high RH the increase in outlet concentrations was significantly smaller. At 90% RH, steady state conditions were achieved after only 20 h and the toluene removal obtained for the following 55 h of the run remained constant at approximately 67%. These results confirm ed the importance of water vapor in the system to prevent the loss of catalyst activity. It is important to
95 note however, t hat the yellowish coloration of the STCs after the runs was still observed for both low and high RH suggesting that there is some deactivation effect still taking place. PCO of toluene: Effect of s pace time ( ) The effect of space time was determined by increasing the flowrate of the packed bed or changing the packed bed depth ( Z ). For all runs, the toluene inlet concentration and RH were kept constant at 30 mg/m3 and 90%, respectively. By increasing the flowrate, the superficial velocity (UG) was corres pondingly increased. Similarly, the toluene and water vapor loadings were also modified as a result in the change in flowrate. The experimental conditions used for the runs showing the effect of space time are shown in Table 5 1 T he effects of space time on the oxidation of toluene are shown in Figure 5 -3 Th e increase in space time resulted in an increase in toluene removal. Initially, the removal wa s high for all space times but it decreased to a lower steady state value with time The decrease d in rem oval is faster for the shorter residence time likely due to the larger t oluene and water vapor loadings. For the longest space time of 25.0 seconds, pseudo steady state was reached after about 60 h of operation The steady state removals achieved were abou t 31% 67 % and 93% for the spac e times of 5.9 s, 12.5 s and 25.0 s respectively. These removals are higher than many of the results reported in the literature for similar residence times For example Maira et al. ( 2001a ) and Belver et al. (2003) who used an annular reactor with TiO2 coated in the walls, found initial removals of 55% and steady state removals of about 5% after 150 min for space times of approximately 5 s. Similarly, Boulamanti et al. (2008) reported about 55% removal for 25 s residence tim e, and they were able to achieve removals greater than 95% only after about 75 s of residence time. By using high water vapor concentrations and longer residence times deactivation was expected to be minimized. However, the yellowish coloration of the STC s was observed in all runs, suggesting that catalyst poisoning was still taking place No byproducts were detected in
96 the effluent flow However, by analyzing the catalyst surface, benzaldehyde and benzoic acid were the species found to be adsorbed to the catalyst. The r esults show ed that despite the catalyst deactivation observed, high toluene removals (>90%) can be achieved and maintained for prolonged periods of time using the STCs Kinetic a nalysis The kinetic analysis for PCO of toluene using dry STCs was performed using the pseudo steady state results from the experiments desc ribed in Table 5 1 Analysis of k inetic data free of mass transfer influences is preferred over such an analysis complicated by mass transfer limitations. Resistance to mass tran sfer can be either external, from the bulk phase to the catalyst pellet, or internal, which refers to the diffusion through the pores. External mass transfer influences are usually assessed by increasing the flowrate in the reactor since it is well establi sh ed that in the presence of external mass transfer influences, the reaction rate increases with fluid velocity. If the system is free of external mass transfer influences the oxidation rate will be the same despite the increase in superficial velocity. Th e average rate of reaction ( r ) used to determine the influence of external mass transfer on toluene oxidation was calculated by Equation 5 1: r = d Y /dt = ( Yo Ye)/ (5 1) where Yo is the influent toluene concentration and Ye is the efflu ent toluene concentration determined at pseudo steady state. These reaction rates are shown in Table 5 2 For the two cases where the superficial velocity was increased by increasing the flowrate and keeping the same packed volume, the reaction rates were very similar, 1.72 x 105 and 1.69 x 105 mol/m3/s. Based on these values, it is expected that external mass transfer influences are not significant. However, when the reactor used to make this comparison is not a differential reactor, such is the case in
97 this study, the intrinsic rate of reaction might also be affected by the increase in flowrate due to the larger differences between inlet and outlet concentrations, making it difficult to differentiate the two effects, unless the studies are performed using two different catalyst beds with the same space time and different velocities (Satterfield, 1970) This case was not investigated in this work. Therefore, to further determine if the external mass transfer resistance was significant, the Mears criterion (CM) was used ( Fogler, 1999) This criterion uses the measured rate of reaction (r ) to determine if the external mass transfer can be neglected. The Mears criterion (CM) states that when the inequality shown in E quation 5 2 is satisfied, external mass tr ansfer is negligible. The variables in CM include R which i s the characteristic pellet radius (radius/2 for cylindrical pellets), n is the reaction order (assumed to be 1 for toluene), kc is the mass transfer coefficient, which was determined using the T hoenes Kramerss correlation shown in Equation 5 3, and Yb, which is the bulk concentration (Folger,1999). Since Yb was unknown and could not be assumed to be the same as Yo, both the influent ( Yo) and effluent ( Ye) concentrations were tested to ensure t he results would apply to both extreme cases of concentration present in the system. CM = ( r R n)/ (kc Yb) < 0.15 for n egligible external mass transfer (5 2 ) 1 1 3 2 1 1/3 2'1.0(Re')() 1 1 (1)cp GpG G G G GGSh Sc kd Ud DD (5 3) The variables used i n the Thoenes Kramers correlation include Re = Re / [(1 ) ] Sh = Sh / [(1 ) ] dp i s the equivalent diameter of a sphere of the same volume, is the void fraction of the packed bed, is the shape factor (external SA/ dp 2), G is the gas dynamic viscosity G is the gas density, and DG is the gas phase diffusivity (DG = 7.42 x 106 m2/s, used for toluene diffusivity in air).
98 The calculation results for the Mears criterion, shown in Table 5 2 indicate d the criterion is satisfied ( CM < 0.15 for no external mass transfer resistance ) for all flow con ditions investigated, thus it was possible to assume that there we re no concentration gradients between the bulk and the external surface of the catalyst pellets. Similarly, the influences of internal mass transfer in the kinetics of the reaction were assessed by using the Weisz Prater Criterion (CWP) When the inequality for the Weisz Prater criterion described by Equation 54 holds, it means that pore diffusion is not important in the sys tem and can be neglected. This criterion is also a function of the measured rate of reaction and the characteristic pellet radius ( R ) in addition to the contaminant surface concentration (in this case assumed to be the same as the bulk concentration, Yo, d ue to the lack of external mass transfer limitations found above), and the effective diffusivity ( De) (Fogler, 1999). CWP = ( r R2)/ ( Yo De) < 1.0 for negligible internal mass transfer (5 4 ) The effective diffusivity, which i s defined by Equation 5 5 is a function of the diffusivity (DG) of the contaminant in the gas phase the catalyst grain porosity (), determined to be 0.7 for the STCs used in this study, and the tortuosity factor for the STCs (c), which was assumed to b e 3 as it usually is for many mesoporous materials ( Satterfield, 1970) De = (D )/ c (5 5 ) The values obtained from the Weisz -Prater expression for the different flow conditions used in the experimental runs are shown in Table 5 2 All these values are less than 1, meaning that there are no diffusion limitations inside the pellets and consequently no co ncentration gradients within the pellets. Based on the Mears and Weisz Prater criteria, the measured rates of reaction at steady state can be considered free of mass transfer influences, so they are expected to be the result of the
99 intrinsic photochemical kinetics only. Consequently, the steady state removals at various space times shown in Figure 5 3 were used to determine the reaction r ate constant for toluene oxidation using the STCs. The Langmuir Hinshelwood (L H) model, described by Equation 56 has been successfully used by many researchers to describe and model the degradation rates of toluene PCO using Degussa P25 TiO2 (Obee and B rown, 1995; Bouzaza et al., 2006; Boulamanti et al., 2008): r = dC/dt = (kK Y )/ (1 + K Y ) (5 6 ) where k is the rate constant, K is the adsorption equilibrium constant, and Y is the bulk contaminant concentration. For low inlet contaminant concentr ations ( KY << 1), the L H kinetic equation can be reduced to a pseudo-first order rate equation, with an overall rate constant k equal to kK shown by Equation 5 7 : Ln( Ye / Yo) = k (5 7 ) The experimental results and the L H model are pres ented in Figure 5 4 The L -H model resulted in a good fit of the data (R2 > 0.99), indicating that the first order reaction simplification is valid The rate constant calculated from the slope of the linear regression equation fit in Figure 5 4 was equal t o 0.12 s1. Bouzaza and Laplanche (2002 ), who used an annular reactor configuration for the PCO of toluene with Degussa P25 TiO2, followed a similar approach for their kinetics analysis and their ca lculated rate constant was 0.066 s1. Other pseudo first o rder kinetic rate constants derived from the L -H model reported in the literature are even lower, in the 7.3x104 s1 range. Compared to the values reported in the literat ure for TiO2 only, the rate constant for toluene using the STCs is greater than many of those cases, suggesting the STCs enhanced performance over using TiO2 alone.
100 Stokke and Mazyck ( 2008), who used the same STCs formulation for the removal of methanol, found in their kinetics analysis a rate constant of 0.40 s1 for methanol. This k is larger than the rate constant found for toluene in this study. Similarly, other researchers investigating the PCO of various VOCs have also found rate constants for alcohols that are higher than those of aromatics under the same operating conditions ( Kim et al., 2002; Bouzaza et al., 2006). This can be expected given the structure of the molecules. Aromatic compounds have a benzene ring which tends to be more difficult to break down during oxidation. Alcohols, on the other hand, tend to be smaller molecule s with an OH terminal group that is usually very reactive. Gas Phase Toluene Studies U sing Wetted STCs Toluene a dsorption in a b atch r eactor The adsorption of toluene on STCs was shown to be significant in the presence of both low and high water vapor conc entrations when the STCs we re dry. During the simultaneous treatment of gas an d aqueous phases, the pellets wer e expected to be wetted at all times. Accordingly, the assessment of gas phase toluene adsorption and PCO under wet STCs conditions was of intere st. Adsorption of toluene was initially compared using dry and pre -wet ted STCs in batch reactors. The batch experiments were performed by suspending about 3 g of STCs ( 10 STC s ) in a bottle containing 50 mL of pure liquid toluene. The pellets were not in di rect contact with the liquid toluene. They were placed in an aluminum container suspende d above the liquid, so that they were only in contact with gas phase toluene resulting from volatilization (vapor pressure 30.6 mmHg at T = 26.5oC). The gas phase conce ntration of toluene inside the bottles was not measured, but it was determined by the vapor pressure of toluene to be approximately 139 mg/L. The experimental setup is shown in Figure 5 5 for the f our different wetting scenarios investigated The first rea ctor contained water vapor as the gas phase and dry STCs as the
101 sorbent ; the other three reactors contain ed toluene vapor as the gas phase and dry STCs ( for reactor 2), STCs pre -wette d with DI water ( for reactor 3) or STCs pre -wetted in a 1 ppm toluene sol ution ( for reactor 4) The STCs were pre -wetted by soaking them in the respective solutions for at least 24 hrs before the experiment. During the experiments, the weight of the STCs was monitored over time. The samples were weigh t ed at the beginning of the experiment and at different time intervals. The uptake was observed for several hours (up to 50 hours) and i t was assumed that the increase of the weight of the STCs was due to the adsorption of the compound present in the gas -phase The STCs used in thes e experiments had an average pore diameter of 147 a BET SA of 248 m2/g, a pore volume of 0.91 cc/g and a TiO2 loading of 29%wt. The results showing the mass of either water vapor or toluene gas adsorbed per gram of STC as a function of time are shown in Figure 5 6. The results in Figure 5 6 show that when the STCs were dry, they had a large adsorption capacity for water vapor and toluene. The adsorption capacity of the STCs for water vapor increased linearly at a rate of approximately 0.3 mg/min and equ ilibrium was not reached even after 50 h of adsorption. T he amount of water vapor found to be adsorbed at the end of the run was about 0.3 g of water vapor per g of STC but it appeared from t he adsorption trend that there wa s still higher capacity for wat er vapor adsorption under dry conditions The adsorption of toluene to the dry STCs reached equilibrium after about 24 h with an adsorption capacity of a pproximately 0.8 g of toluene/ g STC which corresponds to about 21 toluene molecules adsorbed per nm2. This capacity is greater than the one found for the continuous flow experiments. These results suggest that multilayer adsorption might be occurring under batch conditions, which is not the case for continuous flow conditions.
102 Pre -wetting the STCs result ed in decreased adsorption. U nlike dry STCs, the STCs that were pre -wetted in either DI water or 1mg/L toluene solution did not show significant adsorption during the 50 h of experimental run. These results show ed that t he wetting of the STCs c ompletely in hibited adsorption of gas -phase toluene. The inhibition of adsorption was not completely unexpected in the batch experiments because water or an aqueous solution was pre adsorbed to the STCs, which are known to have a high affinity for water, thus ensuring that all available adsorption sites became occu pied by the wetting solution Water molecules were adsorbed to the STCs likely by hydrogen bonding, and these strong bonds could not be outcompeted by the toluene molecules present in the gas phase. Toluene a dsorption in a c ontinuous fl ow p acked -b ed r eactor Although toluene adsorption was inhibited during batch experiments due to the pre -wetting of the packing, the performance of the wetted STCs was also assessed using a continuous flow packed bed reactor. F or these experiments, a small vertical cylindrical reactor with an outside diameter of 4.2 cm packed with the STCs was used. The reactor was similar to the previously described annular reactor but the lamp envelope at the center of the reactor was removed. T he flow rate used in the small reactor was about 1,3 00 mL/min, which translated to an empty bed contact time (EBCT) of approximately 7 s. The toluene concentration used for all runs was 14 3 ppmv. For each run, the reactor was packed with glass beads nea r the gas inlet and 100 pellets (about 34 g STCs) on top of the beads. The beads were used to provide a more uniform distribution of the gas flow before reaching the STCs The same experiment was performed using three different wetting conditions During t he first run, dry STCs were used. For the second and third run, pellets pre -soaked in DI water or 1 mg/L aqueous toluene solution were used. Similar to the batch experiments, t he STCs were wetted by pre -soaking them in the desired
103 solutions. The STCs used in these experiments had an average pore size of 168 a BET SA of 213 m2/g, a pore volume of 0.90 cc/g, and TiO2 loading of 29.0%wt. The re sults for all wetting conditions of the continuous flow experiments are presented in Figure 5 7 Similar to the res ults obtained in the batch reactor the adsorption of gas phase toluene was drastically affected by the presoaking of the STCs with either DI water or 1 ppm toluen e solution. When dry pellets were used, the removal of toluene by adsorption was initially 8 5 %, decreas ing to about 70% after 6 h of adsorption Conversely, when the pellets were presoaked in solution, no significant adsorption of toluene wa s observed. This means that water remained adsorbed to the STCs despite the reactor configuration, prevent in g significant toluene adsorption to the pellets. Pilot -S cale Studies On the one hand, when trying to determine the photocatalytic oxidation of toluene using the STCs, excluding adsorption effects is preferred so that the removal can be only attributed to P CO; thus inhibiting the adsorption of toluene by pre -wetting of the STCs was a potential a lternative for those purposes. However, many of the mechanisms proposed for the oxidation of toluene using TiO2 involve an adsorption step before reaction, so that the reaction could occur between the adsorbed contaminant and the OH radical or t he hole in the titania surface There is no consensus in the literature as to whether the reactio n occurs right at the surface, in the bulk gas phase or a combination of both. For the case of toluene, it was shown earlier that the presence of water vapor plays an important role for the complete conversion of toluene to CO2 and water and at decreasing catalyst deactivation. However, the effect of completely wetting the catalyst d uring gas phase toluene PCO has not been previously investigated. Consequently, it was important to determine the extent of toluene photocatalysis when the STCs were wetted to help elucidate potential reaction mechanisms. PCO experiments could not be perfo rmed at the
104 bench scale due to the large increase in temperature inside the reactor (from 25oC to over 50oC), which caused the adsorbed pre -wetting solution to dry within the first 10 min of the experimental run. Therefore, the effects of STC wetting was o nly assessed at the pilot -scale. Toluene Adsorption U sing TPOT The effect of wetting on gas phase toluene adsorption was investigated in the TPOT before assessing its effect on PCO to ensure that the same effects were observed at the pilot -scale as those f ound at the bench-scale Adsorption of toluene using d ry STCs Initially, the extent of toluene gas phase adsorption on the dry STCs in the pilot -scale reactor was determined. During these experiments, the commingled packing, high flow rings and STCs were dry and the only flowing phase through the reactor was the gas phase contaminated with toluene. The effects of mass loading and superficial velocity on adsorption were investigated by varying the influent toluene concentration ( Yo) and the gas volumetric f lowrate (QG), respectively. The RH was that of the ambient air, since that was the air mixed with the toluene to produce the dilute gas influent. The humidity remained constant at about 45% at 22.3oC. The experimental conditions for the adsorption experime nts performed are given in Table 5 3. The results showing the adsorption breakthrough curves for the dry STCs are presented in Figure 5 8. Different from the results found at the bench -scale, the adsorption of toluene in the TPOT using dry STCs did not show the large adsorption capacity expected given the lar ge number of STCs used to pack the reactor ( ca. 7,500 STCs). Ini tially, the removal of toluene wa s high, but it became negligible after only 4 h of continuous flow for all conditions tested The total a mounts of toluene adsor bed were 156, 93 and 177 g/g of STC for runs 1, 2 and 3, respectively. These
105 values were lower than the amount of toluene adsorbed to the STCs in the bench-scale packed be d reactor (2.5 mg/g STC). Increasing the mass loading rate f rom 309 to 586 g/min did not significantly affect the removal of toluene by adsorption as shown by the breakthrough curves for runs 1 and 2. However, decreasing the superficial velocity resulted in higher toluene removal (runs 2 and 3), likely due to the increase in residence time which allowed more time for adsorption The difference in adsorption trends between the bench -scale and pilot -scale experiments indicate d a possible effect of the commingled packing on the flow patterns of the pilot -scale reacto r. Another possible reason to the lower amounts of toluene adsorbed in the TPOT compared to the bench-scale annular reactor was competition with water vapor which was not as significant at the bench -scale. During the adsorption experiment at the bench -sca le the water vapor concentration was about 2.6 g/m3. This concentration was about 3.5 times higher at the pilot scale. The effluent water vapor concentration normalized by the influent water vapor is shown in Figure 5 9 as a funct ion of run time. These re sults we re very similar to the toluene adsorption breakthrough curves in Figure 5 8 As the superficial velocit y was decreased, higher amounts of water vapor were adsorbed as it occurred for toluene but less water vapor was applied at the lower flowrates. Despite the lower amounts of toluene adsorbed using the TPOT, adsorption of gas phase toluene was found to be significant up to about 4 h of continuous operation for the dry STCs. Adsorption of toluene using w etted STCs The effect of wetting of the STCs on gas phase toluene adsorption was assessed in the TPOT. The wetting of the STCs in the TPOT was performed by completely filling and draining the reactor with tap water, followed by a continuous flow of water at about 6 L/min for 15 to 20 minutes to ensure complete wetting of the STCs. The same experiment al conditions as the one
106 shown in Table 5 3 for run 2 w ere used to perform the test using the wetted STCs. The results showing the adsorption of toluene in the TPOT under dry versus wet STCs conditions are shown in Figure 5 10. Similar to the bench-scale findings, wetting the STCs completely prevent ed adsorption from taking place even after 3 h of experimental run, suggesting that no significant drying of the pre -wetted packing occurred Toluene PCO U sing TP OT PCO of toluene using d ry STCs The photocatalytic oxidation of toluene under dry conditions was investigated in the TPOT. Similar to the results found at the bench-scale, it was shown that pre -adsorbing toluene before PCO resulted in large desorption of toluene after the lamps were turned on due to the high increase in the temperature in the reactor caused by the lamp operation (Figure 5 11) and no significant removal due to PCO could be observed during the length of the experimental run. Consequently, a ll the subsequent experiments performed using dry STCs, showed the combined effects of adsorption and photocatalysis. The results for the combined adsorption/PCO gas phase experim ents using dry STCs are shown by the conversion (XA) or removal of toluene de fined by Equation 5 -8 as a function of the run time. XA = (Yo Ye)/Yo (5 8 ) Since the tower had been previously used with water, the STCs were dried by running ambient air through the reactor for prolonged periods of time and turning the UV l amp s on and off periodically for about 1 h intervals to allow the system to dry faster. The packing of the tower was considered to be dried by visual inspection, measurements of influent and effluent RH and also by inspecting a few samples of the STCs obta ined from inside the reactor.
107 Initially, toluene PCO in the TPOT was tested by running 3 experiments under the same initial conditions to determine the performance of the tower for consecutive runs. The initial conditions were: Yo = 65 10 mg/m3, QG = 14 2 L/min, influent RH = 44% and i nfluent gas t emperature = 23oC. The results for these experiments are shown in Figure 5 12. Based on the bench -scale experiments, the removal of toluene by PCO using dry STCs was expected to be high in the TPOT since th e spa ce time for the experiments was about 15 s, wh i c h corresponded to steady state removals higher than 70% at the bench-scale. However, the removal of toluene observed at the pilot -scale was initially high, as high as 50% conversion, but it rapidly decreased to zero after about 4 h of continuous operation. Furthermore, the TPOT performance at removing toluene decreased after every run. This behavior suggested deactivation of the STCs likely due to poisoning of the catalyst by s trongly adsorbed intermediates. The lower conversion might be also attributed to the lower adsorption of gas phase toluene found earlier, which was likely decreased even further by the large temperature increase in the reactor as a result of the UV lamps. To verify if the catalyst was lo sing activity due to poisoning, the packing was regenerated by flushing ambient air through the reactor for several hours, followed by 1h of UV while flushing air, and repeating the sequence at least 3 times. The conversion of toluene in the TPOT improved after every regeneration as shown in Figure 5 13. The initial removal s increased to about 57% and 70% for the first and second regenerations, respectively. These removals also decreased with time but they did not go to zero. After the first regeneration, the steady state conversion was about 12% while this value improved to about 23% for the second regeneration. Regenerating the tower by flushing air and turning the UV on in the absence of the contaminant likely allowed the adsorbed int ermediates in the ca talyst to be destroyed due to the excess
108 number of photons and electron scavenger (air). Additionally, the water vapor in the clean air used to regenerate the tower, likely provided more water molecules to regenerate the hydroxyl radicals at the surface of the TiO2. As shown by the bench -scale results, the presence of water vapor in the influent gas is an important parameter to decrease catalyst deactivation by increasing conversion of toluene to CO2 and water instead of more persistent intermediates, and also to maintain the catalyst activity. During the PCO experiments, the initial RH was not varied but, but it was measured throughout the length of the tower at the 3 sampling ports on the wall of the tower placed 1 ft apart. The water vapor concentrations as a function of tower height at various run times for the experiment corresponding to the 2nd regeneration study are shown in Figure 5 14. The water vapor entering the reactor decreased a s a function of both reactor length and run time. These results sug gest that water vapor is being consumed during the PCO of toluene. After about 90 min of continuous operation, the water vapor in the reactor is around 3 g/m3, which is similar to the lower water vapor concentration used in the bench -scale experiment that resulted in steady state removals of only 10%. The decrease in toluene removal as a function of time (Figure 5 13), follows the same trend as the decrease in water vapor concentration as a function of time ; indicating that water vapor is a requirement for toluene PCO to take place In addition to the water vapor concentration another important factor that can influence the photocatalytic process is the lamps temperature. The expected output of the UV lamps can be greatly influenced by the temperature; co nsequently, different types of lamps have an optimum temperature range that produces the highest irradiation output. Figure 5 15 shows the relative output of the lamps as a function of temperature for the lamps used in this research. From the figure, it ca n be observed that the maximum output occurs at a temperature of 42.2 oC, and it
109 decreases for higher or lower temperatures. The temperature of the lamps, as measured at the lamp envelope, during the PCO experiments, increased to about 1 25oC as shown in Fi gure 5 13. This temperature is well above the optimum operating temperature of the lamps; so it is expected that the lamps performance is not at its maximum. PCO of toluene using w etted STCs Investigating the wetting of the packing on gas phase toluene PC O was very important due to its implications in the treatment of toluene present in the gas and aqueous phases simultaneously. Accordingly, gas phase toluene PCO was studied under different wetting conditions using the TPOT. Two approaches were used for te sting the effect of packing wetting. First, toluene PCO was tested by pre -wetting the packing with tap water. The pre -wetting procedure consisted of flowing water through the tower at a rate of 7.5 L/min for about 10 minutes before the run The other wetti ng approach was to keep the tower continuously wetted by flowing tap water free of toluene simultaneously with the contaminated gas phase toluene. For the latter approach, some absorption of toluene from the gas phase to the aqueous phase was anticipated, so conversion was calculated by performing a mass balance on toluene in the reactor including both phases. For this experiment, toluene conversion was calculated using Equation 5 9 XA = 1 (Ye QG + Ce QL) / (Yo QG) (5 9 ) The results showin g the conversion of toluene for the different wetting conditions of the packing are shown in Figure 5 16. When the packing was either pre -wetted or continuously wetted, the net removal of toluene was zero, meaning that wetting the packing completely hinder ed photocatalysis. The results for the pre -wetted packing show ed an increase in toluene removal from zer o to almost 30% after about 1 h of the run time. The increased removal after some time can be explained by the drying of the packing due to the heat produced by the UV
110 lamps and evaporation of water as a result of air flow through the packed bed. Having the UV lamps t urned on at all times increased the temperature inside the reactor; thus, drying the commingled packing that was wetted before the experimen t was started. The drying of the packing was confirmed by the water condensation observed at the gas effluent. By keeping t he packing wetted at all times no net r emoval was found during the 3 h run. These results were surprising because although adsorption was shown to be inhibited by wetting of the packing, the photocatalysis of toluene was not anticipated to become negligible in the absence of adsorption given the requirement of water vapor for large toluene removals to occur Creating a water film aroun d the STCs prevented direct contact of the titania embedded in the pellets and the to luene present in the gas phase. These results suggest that the mechanism involved in toluene PCO using STCs requires an adsorption step of the contaminant for the reaction to occu r. Despite having the hydroxyl radical forming potential (i.e. water) pre adsorbed, it is evident that the reaction does not occur in the bulk gas phase since no removal was observed under any wetting conditions. Table 5 1 Summary of e xperimenta l c onditions for the bench-scale gas phase PCO experiments using dry STCs Z Q G U G s cm cm 3 /s cm/s 5.9 4 26.7 0.68 12.5 4 12.5 0.32 25.0 8 12.5 0.32 Table 5 2. Mears ( CM) and Weisz -Prater (CWP) criteria for the determination of mass transfer influe nces. CM <0.15 indicates no significant external mass transfer resistance. CWP<1 indicates no significant internal mass transfer resistance. (s) r 10 5 (mol/m3/s) r (g / L /s) k c 10 2 (m/s) C M < 0.15 Yb = Yo C M < 0.15 Yb = Ye C WP < 1 5.9 1.72 1.59 3.1 1 0.0038 0.005 6 0.15 9 12.5 1.69 1.55 2.1 1 0.005 7 0.017 1 0.15 9 25.0 1.18 1.09 2.2 1 0.003 8 0.051 1 0.11 3
111 Table 5 3. Experimental conditions for the adsorption experiments using dry STCs in the TPOT Run Co Q G U G Mass Loading Influent Water Vapor # mg/m 3 L/min s cm/s g/min g/m 3 1 74 4 142 15.4 7.9 585.8 8.69 0.12 2 39 3 142 15.4 7.9 308.7 8.74 0.08 3 76 5 85 25.7 4.7 360.1 9.10 0.14 Figure 5 1 Gas phase adsorption of toluene followed by PCO using dry STCs and low relative humidity conditions in the bench scale reactor : RH = 13%, Yo = 90 10 mg/m3. QG = 3.3 cm3/s Bed Volume = 42 cm3. Y is the effluent concentration of toluene in the gas phase.
112 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0 10 20 30 40 50 60 70 80 Y/Yo Run Time (h) RH = 13% RH = 90% Figure 5 2 Normalized effluent toluene concentration durin g simultaneous adsorption and PCO experiments at different water vapor concentrations as a function of run time. Yo = 30 mg/m3. QG = 12.5 cm3/s. Bed Volume = 157 cm3. 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0 10 20 30 40 50 60 70 80 90 100 Y/Yo Run Time (h) V/Q = 5.9 s V/Q = 12.5 s V/Q = 25.0 s Figure 5 3 Normalized effluent toluene concentrations as a function of run time at var ious space times. Yo = 30 mg/m3, RH = 90%.
113 Ln(C/Co) = 0.117* 0.331 R = 0.99 0.0 0.5 1.0 1.5 2.0 2.5 3.0 0 3 6 9 12 15 18 21 24 27 30 Ln (Y/Y o ) Space Time, (s) Figure 5 4 Linear regression of the Langmuir Hinshelwood model Gas phase Water vapor Toluene gas Toluene gas 1 4 3 2 STCs degree of wetting Dry Dry Pre-wetted with DI Pre-wetted with 1ppm toluene soln Toluene gas Figure 5 5 Plan view of the e xperimental setup of the batch experiments to assess adsorption under different STCs wetting conditions.
114 0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0 10 20 30 40 50 60 Amount of Gas Adsorbed (g) / g STC Adsorption Time (h) (1) Dry STC/ Water Vapor (2) Dry STC/ Toluene Gas (3) DI Pre wetted STC/ Toluene Gas (4) 1ppm Tol soln pre wetted STCs/ Toluene Gas Fi gure 5 6 Gas phase adsorption of water vapor and toluene to STCs under different wetting conditions 0.0 0.2 0.4 0.6 0.8 1.0 1.2 0 1 2 3 4 5 6 Run Time (h) Y/Yo (1) Dry STCs (2) STC pre-wetted in DI water (3) STCs prewetted in 1ppm tol Figure 5 7 Adsorption of gas phase toluene to STCs under different wetting conditions in a continuous flow reactor
115 0.0 0.2 0.4 0.6 0.8 1.0 0 30 60 90 120 150 180 210 240 270 Ye/Yo Run Time (min) (1) 142L/min 74ug/L (2) 142L/min 39ug/L (3) 85L/min 76ug/L Figure 5 8. Adsorption breakthrough curves for toluene on the dry STCs in the commingled packing of the pilot -scale reactor. 0.0 0.2 0.4 0.6 0.8 1.0 0 30 60 90 120 150 180 210 240 270 RH(Effluent)/RH(Influent) Run Time (min) Run 1 Run 2 Run 3 Figure 5 9. Adsorption breakthrough curves for water vapor on the dry STCs in the commingled packing of the pilot -scale reactor
116 0 0.2 0.4 0.6 0.8 1 0 50 100 150 200 250 300 Ye/Yo Run Time (min) Pre wetted STCs Dry STCs (Run 2) Figure 5 10. Comparison of gas phase toluene adsorption using the TPOT with dry versus pre wetted STCs.
117 Run Time (h) 0 1 2 3 4 5 6 7 8 Y/Yo 0.0 0.2 0.4 0.6 0.8 1.0 20.0 40.0 60.0 80.0 Lamp Temperature ( o C) 0 5 10 15 20 25 30 35 60 90 120 150 Adsorption Photocatalysis Lamp Temp. UV Lamp ON Figure 5 11. Adsorption of gas phase toluene followed by PCO in the TPOT using dry STCs and ambient relative humidity Yo = 85 mg/m3, QG = 142 L/min.
118 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0 60 120 180 240 300 X A Run Time (min) 1st run 2nd run 3rd Run Figure 5 12. Conversion of toluene by PCO in TPOT for 3 studies performed using the same initial conditions: Yo = 65 10 mg/m3, QG = 142 L/min, RHavg = 44%, Tavg= 23oC 0 20 40 60 80 100 120 140 0.0 0.2 0.4 0.6 0.8 1.0 0 50 100 150 200 250 X A Run Time (min) 1st Regenration 2nd Regeneration Lamp Temp Figure 5 13. Conversion of gas phase toluene by PCO in the TPOT after catalyst regeneration. Yo = 60 10 mg/m3, QG = 142 L/min, RHIN,avg = 44%, TIN ,avg = 23oC.
119 0 1,000 2,000 3,000 4,000 5,000 6,000 7,000 8,000 9,000 10,000 30 60 90 120 150 180 240 Water Vapor Concentration (mg/m 3 ) Run Time (min) Inlet 1 FT 2 FT 3 FT Figure 5 14. Water vapor concentration profile in the TPOT at various run times for the experiment corresponding to the 2nd regeneration study. Samples were collected from the influent and at packing depths of 1, 2, and 3 feet.
120 Figure 5 15. Relative output of the UV -C lamps used in this research as a function of temperature ( Obtained from Philips, 2004). 0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0 30 60 90 120 150 180 210 240 X A Run Time (min) Dry Packing (2nd Reg) Pre wetted Packing Continuously Wetted Packing Figure 5 16. Effect of wetting of the packing on gas phase toluene removal by PCO in the TPO T. Yo = 60 10 mg/m3, QG = 142 L/min, QL = 3.8 L/min RHIN,avg = 44%, TIN ,avg = 23oC.
121 CHAPTER 6 AQUEOUS PHASE TOLUENE DEGRADATION STUDIE S Bench S cale Studies Adsorption of Toluene in a Batch Reactor : The extent of aqueous phase toluene adsorption to the S TCs was determined by performing batch adsorption equilibrium experiments. During these experiments, either the initial concentration of toluene or the mass of STCs was kept con stant while the other parameter was changed. The systems were expected to be at equilibrium since adsorption times of at least 24 h were used. When the STCs are placed in any solution, some outgassing of the air present in the pore matrix is observed due to the large internal pore volume of the STCs. T o prevent losses of toluene by volatilization when placing the STCs in the aqueous toluene solution, the STCs were soaked in DI water for about 24 hrs before placing them in the batch reactors. The removal of toluene as a function of STC mass is shown in Figure 6 1 for initial toluene co ncentrations of about 190 g/L and 1000 g/L Increasing the mass of STCs increased the removal of toluene because of the larger available area present for adsorption. However, the initial concentration did not seem to have a noticeable i mpact on the remov al of toluene since both removal curves followed the same trend as a function of STC mass. The slope of the removal curve decreased with increasing STC mass. The maximum concentration removal achieved for the mass of STCs tested was about 75% which corres ponded to approximately 6 g STCs (about 16 STCs pellets) The adsorption results were compared to the two most commonly used single component adsorption isotherms to describe the adsorption of organic contaminants: Langmuir and Freundlich isotherms. The mo dels describing the adsorption density ( q ) as a function of contaminant concentration ( [A] ) are given by equations 6 1 and 6 2 for Langmuir and Freundlich isotherms, respectively. A linearization method was used for each isotherm to find the two fitting
122 pa rameters for the models that best matched the data. The linear forms of the models are shown in Equations 6 3 to 6 5 for the Langmuir type 1 (L1), Langmuir type 2 (L2) and Freundlich isotherms respectively. The fitting parameters, shown in Table 6 1, were calculated from the slope and y intercept found by the linear regression of the plots: 1 /[A] vs. 1/ q (for L1), [A] vs. [A] / q (for L2) and Log [A] vs. Log q (for Freundlich). qLangmuir = Kads qmax [A] / ( 1 + Kads [A]) (6 1) qFreundlich = Kf [A]n (6 2) 1/ qL1 = 1 / ( Kads qmax) 1/ [A] + 1 /qmax (6 3) [A] / qL 2 = 1 / qma x 1/ [A] + 1 /(Kads qmax) (6 4 ) Log q = n Log[A] + LogKf (6 5) The results showing the adsorption density in g of toluene/g of STC as a function of toluene concentration are presented in Figure 6 2 together with the fits for the isotherms evaluated. The adsorption isotherm seems to follow a fairly linear trend for the concentration range investigated (48 to 888 g/L). The adsorpti on density wa s not very large for this aqueous phase toluene STC system since only about 6 g of toluene wer e adsorbed per gram of STC for the highest concentration included (or about 446 molecules of toluene/ m2 based on a 250 m2/g surface area of the STCs). The best fit, determined by comparing the sum of squared differences between the data and the model (SSqD), was obtained by the Freundlich isotherm (SSqD = 0.36), followed by L2 (SSqD = 1.4) and L1 (SSqD = 5.3). The exponent found f or the Freundlich isotherm was close to 1 ( n = 0.87 ), which agrees with the linearity of the adsorption isotherm shown by the data. However, it is important to notice that the isotherm fit does not necessarily imply that the underlying assumptions used to derive the isoth erms (eg. monolayer vs. multilayer
123 adsorption) are valid for the aqueous tol uene STC system. These isotherms provide a mathematical tool to describe the trends shown by the data. Pilot -S cale Studies Adsorption of Toluene U sing TPOT Given the large in ter nal surface area of the STCs and the result s observed at the benchscale, some adsorption of toluene was expected to occur in the aqueous phase Therefore, a queous phase adsorption was tested in the TPOT at different liquid loadi ng rates. Higher loading ra tes we re expected to increase the surface area available for adsorption since they result in larger wetted packing areas. As shown in Figure 6 3 despite the increase of liquid loading rate, no significant removal by adsorption was observed. Previously pre sented batch scale experiments showed significant adsorption of toluene which was not surprising given the adsorption time used of 24 h. Similarly, H olmes et al. (2004) found significant toluene adsorption using the same material of a smaller pellet size in a continuous flow reactor In their reactor system they achieved equilibrium after 5 h. The lack of adsorption in the TPOT is probably due to the short average residence time of the water in the reactor of less than 60 seconds. Although the STCs have a large surface area for adsorption, it seems that they require long contact times for significant adsorption to be observed The time for adsorption available in the TPOT (less than 60 s) wa s not enough to show significant differences in influent and effluent concentrations for the mass loading rates (1.5 3.6 mg/min) used during the experiments. Adsorption is expected to occur until equilibrium is achieved in the STCs in the TPOT, but it does not affect the concentration of the effluent because the remov al for a single pass of fluid is small due to the small residence time
124 PCO of Toluene U sing TPOT The photocatalytic oxidation of toluene in the aqueous phase was tested as a function of two important parameters expected to have a significant effect on PCO: the initial concentration of toluene and the liquid loading rate. The removal of toluene as a function of initial concentration for a flowrate of 3.8 L/min (loading of 3.1 gpm/ft2), shown in Figure 6 4, indicates that increasing the toluene inlet concentr ation resul ted in a higher removal, likely due to the increase in driving force for the transfer of toluene across the liquid-solid film which result ed in higher concentrations at the catalyst surface and thus higher removal. The effect of liquid loading rate in the conversion of toluene can be observed in Figure 6 5 All the results shown in F igure 6 5 were obtained at an initial toluene concentration of approximately 495 37 g/L. The conversion of toluene decreased as the liquid loading rate was increa sed Increasing the liquid loading rate affects three important factors for aqueous PCO : the wetted surface area and the liquid film thickness both increase w hile the mean residence time decrease s. Increasing the area of the catalyst that is wetted with th e contaminated solution should result in higher removals since more of the toluene in the aqueous phase would have the opportunity to be in contact with the catalyst surface ; however, by increasing the film thickness, the liquid -solid resistance is increas ed, so the driving force is adversely affected. Similarly, the lower residence time in the reactor at higher flowrates provide s less reaction time and reduces the removal of toluene by PCO TPOT H ydrodynamics A basic understanding of the hydrodynamic behavior and mass transfer in the tricked bed reactor used in this research is essential for design, scale up, scale -down, and performance prediction purposes. It was of interest to analyze the hydrodynamic behavior of the reactor to
125 obtain estimates of the dis persion coefficients since these are needed to determine the liquid solid mass transfer coefficients used for modeling the reactor. T he hydrodynamic behavi or of trickle packed bed reactors is usually characterized by the axial dispersion and/or the liquid holdup. The axial dispersion is generally specified by an axial dispersion coefficient ( DAX) which provides inform ation about the extent of the deviation of the flow from the ideal plug (or piston) flow ( de Andrade Lima, 2006 ). Tracer studies have long bee n used to characterize the hydraulic behavior of real reactors. TPOT is not a reactor that is completely saturated with water; on the contrary, the water is introduced by a spray nozzle so that the molecules trickle over and through the packing. A tracer test in a trickle bed provides an estimate of the average time it takes the water molecule s to make its way th rough the packing of the tower. The type of tracer test used in the TPOT was the step input tracer test. T his method entails the introduction of t racer at a constant dosage or concentration until the effluent concentration reaches a steady -state value. While continuously feeding the reactor with the tracer, the effluent concentrations are measured and the resulting data provide the cumulative reside nce time distribution function, F(t) Once the cumulative residence time distribution (RTD) function is constructed the residence time density function (E(t) ) for the data is easily obtained as the first derivative of the distribution functio n with respect to time ( Equation 6 6 ) E(t) = dF(t)/dt (6 6 ) Generation of the residence time density function allows calculation of the mean residence time, tm The mean value of the residence time of the water molecules in the reactor is given by the fir st moment of the centroid of the RTD curve, shown in Equation 6 7
126 0 0 0 0() () ()1mtEtdt t tEtdttEt Etdt (6 7 ) Since analytical methods of integration are not adequate to use in discrete data sets such as the collected data from tr acer tests, numerical integration is required in order to calculate tm. The method used for all numerical integrations in this analysis was the trapezoid rule, described by Equation 6 8 : 111 ()*() 2iiiiAreafftt (6 8 ) where f repre sents the function to be integrated over time, and the final result is the sum of all the areas calculated between the points. The axial dispersion coefficient can be calculated by fitting the RTD obtained from the tracer test t o an RTD model that incorpor ates the coefficient as one of the fitting parameters. Several models have been proposed from which an axial coefficient can be identified. Most of these methods are modifications of the most ubiquitous model used in the literature, the one dimensional pis ton dispersion (PD) model (or ideal plug flow with dispersion, or advection dispersion model) described by Equation 6 9 for a non -reactive tracer (Piche et al., 2002). 2 2 AX LdCdCdc DU dzdzdt (6 9 ) The dispersion component in this model is simply a hypothetical term that has the form of Ficks second law added to the plug flow model. The model has no particular mechanistic significance because the use of this model requires empirical evaluation of the dispersion term, i.e. the measured dispersion characteristics need to be correlated with other system properties. Nevertheless, it provides some measure of the back -mixing in the reactor. The closed form
127 mathematical solution for a step input of tracer, which is the metho d used for the tracer tests in this work, can be obtained by treating the stimulus as an injection of the tracer to an open tube which extends from the point of injection to infinity in both direction, so that the initial and boundary conditions are (Weber and DiGiano, 1996) : C = 0 for x > 0 at t = 0 C = CIN for x < 0 at t = 0 C = 0 at x = for t C = CIN at x = for t where CIN is the step concentration of the tracer. For this particular ins tance, the length of the dispersion is unbounded, and the solution for the concentration as a function of position and time is given by the error function. The solution for a specific bed length is shown by Equation 6 1 0 This equation shows that the F(t) curve will broaden as the dispersion coefficient increases. The dispersion number Nd, shown in Equation 6 1 0 is defined as the dispersive mass transport over the advective mass transport or DAX /(U Z ) which is also the inverse of the reactor Peclet number ( Pe ) The relationship indicates that as the dispersion number increases, the transport by dispersion increases relative to the transport by advection. 0.5 0.51 ()0.510.5m mt t out d t in tC Fterf C (6 1 0 ) The tracer of choice for the experiment s was sodium chloride (N aCl) because it can be easily measured using a conductivity meter and it does not adsorb to the plastic packing in the reactor or to the STCs However, w hen STCs are washed with water, the solutions conductivity tends to increase due to the residual fluor ide present in the pellets, thus influencing the conductivity measurements attributed to NaCl Although the STCs were expected to leach fluoride during continuous flow operation and confounding results were anticipated by using NaCl as the tracer, it was d etermined that after 1 0 min of wetting of the packing with tap water,
128 an steady state conductivity was achieved. Additionally, high concentrations of tracer were used to decrease the effect of the potential leaching of fluoride that could increase the cond uctivity measurements Tracer tests were performed using high flow rings only packing as well as the commingled packing. The cumulative RTD s measured during the tracer test and the RTDs calculated from the F(t) data are presented in Figure 6 6 for both pac king configurations Based on the tracer test results, the mean residence time was calculated by numerically deriving the cumulative RTD for each type of packing and flowrate to obtain E(t) and then using Equations 6 -7 and 6 8 T he dispersion coefficient w as determined by minimizing the squared sum of the difference between the modeled values for F(t) obtained from Equation 6 1 0 and the measured results. All these calculated values for the mean residence time and dispersion coefficients for the different li quid loading rates investigated are shown in Table 6 2 The results illustrated in Figure 6 6 for the high flow rings only packing show ed that the residence time distribution s of the water molecules in the TPOT are ver y similar despite the flowrate. For th e case of the commingled packing (Figure 6 6B) the RTDs show ed higher dispersion or wider RTDs for the lower flowrates indicating that the system is approaching plug flow as the flow rate is increased. The dispersion coefficients in the reactor for the co mmingled packing compared to the high flow rings only packing seemed to be slightly higher This is not unexpected because the dispersion in a reactor tends to increase when the heterogeneity of the packing increases. The dispersion coefficients obtained f rom the analysis presented in Table 6 -2, ranged from about 4 to 9 cm2/s. These values followed a linearly increasing trend with increasing flowrate as shown in Figure 6 7 For the lowest flow used, 1.9 L/min, the dispersion coefficient was about
129 4.0 cm2/s This value increased to 6.3 cm2/s when the flow was doubled and to 8.8 cm2/s by increasing the flowrate from 3.8 to 5.7 L/min. The dispersion coefficients calculated from the tracer test analysis were similar t o the values obtained from correlations found in the literature for trickle bed reactors. For example, well known correlations such as those proposed by Otake and Kunugita (1958), Sater and Levenspiel (1966) and Michell and Furzer (1972) and many others summarized by de Andrade Lima (2006) resulted in dispersion coefficients in the range of 1 to 12 cm2/s, which are comparable to the values found for this system. Many of the se correlations available in the literature are functions of the Reynolds (Re) and Galileo (Ga) numbers. The calculation of thes e numbers requires the particle diameter of the packing as one of the known parameters. For the case of the TPOT, the commingled packing makes it difficult to determine one value for this parameter since there are two types of packing both with different d iameters (9 mm for STCs, 15 mm for PR). The value used in the correlations was the average diameter obtained as a weighted average in terms of the bulk volume occupied by each type of packing, which resulted in an average particle diameter of 14.8 mm. A n empirical correlation shown in Figure 6 7, was determined from the data to predict the dispersion coefficients in the TPOT packed with the commingled packing for purposes of modeling This correlation is given in Equation 6 1 1 The two parameters used for the correlation were Re (= dpULw/ L) and Ga (= dp 3w 2g/ L 2) as most of the other correlations in the literature (de Andrade Lima, 2006) The exponent fitted for Ga was 0.10 which is similar to the values found in other correlations. The coefficient for Re, however, was 0.84, which is slightly higher than the values used in the literature which are in the range of 0.5 to 0.7. DAX = 1.98 Re0.84 Ga0.10 (6 1 1 )
130 The mean residence time decreased with flowrate for both types of packing but it was higher for the co mmingled packing compared to the high flow rings only. Mixing the STCs with the pall ring packing decreased the porosity of the tower from about 0.9 for the high flow rings only to 0 .76 for the commingled packing. L ower porosity is an indication of more pa cking present in the tower and increased tortuosity in the system This m eans that there are more tortuous paths for the water to flow through which will tend to increase the holdup in the reactor resulting in higher residence times of the water molecules This will also tend to increase the tailing effect. The mean residence time for the reactor packed with the mixed high flow rings and STCs followed a linear trend with respect to the liquid loading rate as shown in Figure 6 8 for the loadings investigate d in this work Liquid Solid Mass Transfer Coefficient (KLSaC) By knowing the dispersion coefficients in the TPOT as a function of liquid loading rate, the liquid -solid mass transfer coefficients ( KLSaC) can be more easily investigated as shown in this s ection. Mass transfer coefficients from the liquid to the solid catalyst were determined by performing a mass balance on the reactor for the liquid phase and numerically solving the resulting second order differential equation using the finite difference m ethod By this method, finite differences are substituted by the derivatives of the original equation and also the boundary conditions The reactor was modeled using the plug flow with dispersion model (PFD) given in Equation 6 1 2 and solved at steady stat e The corresponding boundary conditions used to solve the model are shown in Equations 6 1 3 and 6 1 4 For the analysis, the central differences shown in Equations 6 1 5 and 6 1 6 were used for the second and first order derivatives, respectively. By perform ing such substitution, a linear differential equation is transformed into a set of simultaneous algebraic equations that can be solved using traditional methods which use some type of matrix multiplication, such as the LU decomposition method Before using the model for
131 the conditions evaluated in this research, it was tested using parameters from examples from Chapra and Canale (2003) to determine if the solutions were correct and ensure that the model was working properly. 2 20AX L LSc SdCdC dC DUKaCC dzdz dt (6 1 2 ) 0 0 AX LINdC DUCC dz a t z = 0 (6 1 3 ) 0ZdC dz a t z = Z (6 1 4 ) 2 11 222iiiCCC dC dzz (6 1 5 ) 112iiCC dC dzz (6 1 6 ) The boundary condition at the reactors inlet (Equation 6 1 3 ) indicates that at t < 0, the reactor is devoid of contaminant and at t = 0, the chemical is injected in the reactor at a constant level of CIN (the flux is constant) The second condition (Equation 6 1 4 ) assumes that the dispersion in the reacto r does not affect the exit rate; so the contaminant leaves the reactor purely as a function of the flow through the outlet. Continuity of flux requires that: r = KLSaC*(C-CS) = k CS n. The surface concentration, CS, shown in Equation 6 1 3 then can be solved for if the reac tion kinetics at the surface are known. However, the intrinsic kinetics for the system are not known in this case, so CS w as assumed to be equal to zero, implying that the contaminant does not accumu late at th e surface of the catalyst. This also implies that the system is mass transfer limited as compared to surface reaction limited because by assuming that the external resistance controls ( KLSaC << k ), so (C -CS) becomes a large value suggesting that CS << C. This assumptio n is not unrealistic if the reaction
132 occurring at the catalyst surface is considered to be i nstantaneous. For the case of photocatalytic reactions, it is expected that the contaminant will react as soon as it reaches the surface of the catalyst given that there is sufficient hydroxyl radicals and UV intensity To determine if the system was mass transfer limited the rates were determined as a function of the loading rates (or superficial velocities) using Equation 6 1 7 since it is well established that in the absence of external mass transfer limitations these rates will remain unchanged. r = dC/dt = (Co Ce)/tm (6 1 7 ) The results for the reaction rates, presented in Table 6 3, indicate that the system appears to be limited by external mass transfer at the lower velocities, but the rates remain ed similar for velocities greater than 0.22 cm/s. However, as explained for the gas phase results of the bench scale kinetic analysis, this is true only for a differential reactor, which is not the case of the TPOT. Therefore, similar to the analysis performed in Chapter 5, the Mears criterion was determined to assess the importance of external mass transfer in the system. The same criterion used for the gas phase (Equation 5 2) was used for the aqueous phase, but for this analysis all the parameters in Equations 5 2 and 53, such as the density, viscosity, and diffusivity refer red to the aqueous phase. The aqueous phase diffusivity (DL ~ 8x1010 m2/s) was calculated using the correlation shown in Equation 6 1 8 (Hand et al., 1999). DL = 13.26105/ ( w 1.14 Vb 0.589) (6 1 8 ) where w is the viscosity of water in centipoises and Vb is the molar volume for toluene (118.3 cm3/mol). The results obtained for the Mears criterion are shown in Ta ble 6 -3 for both the inlet and outlet concentrations since the bulk phase concentration was un known. All the results found were greater than 0.15, indicating that the system is severely limited by mass transfer. Showing
133 that the system is mass transfer lim ited supports the validity of the assumption made about CS. By eliminating CS from Equation 6 13, t he calculation of KLSaC is greatly simplified since it becomes the only unknown in the differential equation. The dispersion coefficient in the model was de termined using the correlation obtained in the previous section shown in Equation 6 1 1 Data on the removal of toluene were available as a function of inlet concentration and liquid loading rate The initial experimental conditions and results for these d ata are shown in Table 6 4 By using the known effluent concentrations (at Z = 122 cm) from the experiments corresponding to a set of given initial operating conditions KLSaC was determined by minimizing the sum of squared differences ( SSqD ) of the actual effluent concentrations and those found with the model by using S olver in Microsoft Excel. The concentration profiles predicted by the model using the fitted KLSaC values are shown in Figures 6 9 and 6 10 for the experiments where either the liquid loadin g rate or the initial concentrations were changed, respectively. Figures 6 9 and 6 10 also show the measured effluent concentrations for each experiment used, which are given in Table 6 4. The results for the fitted KLSaC values, were individually correlat ed as a function of Reynolds (Re) number (Figure 6 1 1 ) and concentration (Figure 6 1 2 ) using a power law model. The coefficients obtained for each individual model were then used as first guesses to fit the liquid -solid mass transfer coefficients in terms of both parameters, Re and CIN. The resulting correlatio n is presented in Equation 6 19. KLSaC (s1) = 1. 2 03Re0. 50 CIN 0.25 (6 19) The exponent for each term is very similar in the single and multiple parameters correlations. The Re number in t he correlation is dimensionless while CIN is given in units of g/L. The KLSaC calculated from this correlation is in units of s1. The comparison of the mass
134 transfer coefficients fitted to the aqueous phase data (data) and those obtained from the correl ation developed as a function of Reynolds number and concentration shown in Figure 6 13, indicates that there is little deviation between the data and the values obtained by the correlation so the correlation given in Equation 6 19 can be used to determi ned the liquid -solid mass transfer coefficient for the range of conditions used in the experiments The correlations for the dispersion coefficient and the liquid -solid mass transfer coefficient were developed in this chapter with the purpose of using the m in the modeling of the reactor when the two phases are flowing simultaneously. Most authors that have investigated the dispersion in trickled bed reactors agree that the dispersion coefficients are not significantly affected by the gas flowrate ( Piche et al. 2002). Therefore, the coefficients developed for the aqueous phase only flow, should still be valid for the two phase flow. Table 6 1. Langmuir and Freundlich isotherms fitting parameters for the aqueous phase toluene adsorption data obtained in bat ch experiments. Isotherm K ads (10 4 ) q max K f n Model L/g toluene g toluene/g STC ( g/g STC)(L/ g) n L1 7.11 39.1 L2 5.25 51.5 Freundlich 0.045 0.87 Table 6 2. Mean residence time and dispersion coefficient s for the high flow rings only pa cking and commingled packing obtained from the tracer tests. High flow rings Only STCs & High flow rings Flowrate t m D AX t m D AX L/min s c m 2 / s s c m 2 / s 1.89 50.27 3.97 3.78 34.50 6.25 40.42 6.29 5.68 31.70 8.82 7.57 29.40 7.21
135 Table 6 3 Rate constants and Mears ( CM) criterion for the determination of external mass transfer influences in the aqueous phase CM <0.15 indicates no significant external mass transfer resistance. U L (cm/s) Liquid Loading (gpm/ft2) r (g/L/s) k c 10 5 (m/s) C M < 0.15 C b = C o C M < 0.15 C b = C e 0.12 1.7 7.3 1.53 2.06 6.74 0.22 3.3 4.2 2.11 1.02 1.64 0.32 4.7 3.9 2.53 0.72 0.97 0.38 5.6 4.4 2.76 0.71 0.92 Table 6 4 Experimental conditions for the aqueous phase PCO experiments in the TPOT. Exp. U L Re C IN C out Actu al K LS a C Predicted K LS a C # cm/s g/L g/L s 1 0 3 s 1 0 3 1 0.12 16.7 532 163 1.39 1.41 2 0.22 31.8 445 275 0.94 0.98 3 0.32 46.1 491 367 0.80 0.83 4 0.38 54.3 514 398 0.82 0.78 5 0.22 31.8 160 108 0.77 0.76 6 0.22 31.8 250 159 0.89 0.85 7 0.22 31.8 908 512 1.14 1.17 8 0.22 31.8 1404 716 1.37 1.30 0 0.2 0.4 0.6 0.8 1 0 1 2 3 4 5 6 7 Toluene Removal (1 C/Co) Mass of STCs (g) Co=1000ppb Co=190ppb Figure 6 1. Removal of toluene due to adsorption in batch experiments as a function of STCs mass for different initial aqueous phase toluene concentrations. Data measured at equilibrium (adsorption ti me of 24 h)
136 0 2 4 6 8 10 12 14 16 18 0 200 400 600 800 1000 Adsorption Density ( g Toluene/g STC) Aqueous Phase Toluene Concentration ( g/L) Data L1 L2 Freundlich Figure 6 2. Adsorption isotherm for the aqueous phase toluene STCs system. Comparison between the data and isotherm models. 0.0 0.2 0.4 0.6 0.8 1.0 1.2 0 50 100 150 200 250 C/Co Run Time (min) 3.8L/min, 127L/min/m^2 7.6L/min, 253L/min/m^2 Figure 6 3. Aqueous phase toluene adsorption for different liquid loading rates. Average Co = 388 g/L for 127 L /min/m2 and 481 g/L for 253 L/min/m2.
137 y = 1.21E 04x + 3.21E 01 R = 0.98 0.0 0.2 0.4 0.6 0.8 1.0 0 250 500 750 1000 1250 1500 Toluene Removal (1 Ce/Co) Aqueous Toluene Concentration ( g/L) Figure 6 4. Toluene removal as a function of inlet toluene concentration in the TPOT at a flowrate of 3.8 L/min. y = 1.4254x 0.972 R = 0.99 0.00 0.20 0.40 0.60 0.80 1.00 0.0 2.0 4.0 6.0 8.0 10.0 1 C/Co Liquid Volumetric Flowrate (L/min) Figure 6 5. Aqueous phase toluene removal in the TPOT as a function of flowrate. Co = 495 37 g/L
138 0.0 0.2 0.4 0.6 0.8 1.0 1.2 0 50 100 150 200 250 300 350 400 F(t) Time (s) PR 1gpm PR 2gpm 0.0E+00 2.5E 03 5.0E 03 7.5E 03 1.0E 02 1.3E 02 1.5E 02 1.8E 02 2.0E 02 0 30 60 90 120 150 180 E(t) Time (s) PR 1gpm PR 2gpm 0.0E+00 5.0E 03 1.0E 02 1.5E 02 2.0E 02 2.5E 02 3.0E 02 3.5E 02 4.0E 02 4.5E 02 5.0E 02 0 30 60 90 120 150 180 E(t) Time (s) PR/STC 0.5gpm PR/STC 1gpm PR/STC 1.5gpm 0.0 0.2 0.4 0.6 0.8 1.0 1.2 0 50 100 150 200 250 300 350 400 F(t) Time (s) PR/STC 0.5gpm PR/STC 1gpm PR/STC 1.5gpm Figur e 6 6 Cumulative RTD ( F(t) ) and RTD ( E(t) ) functions obtained from the tracer test analysis for the TPOT at different flowrates and packed with different packing styles: A) high flow rings only (PR) and B) commingled packing (PR/STC ). A) B )
139 0 2 4 6 0.0 2.0 4.0 6.0 8.0 10.0 12.0 0 0.1 0.2 0.3 0.4 0.5 Liquid Loading Rate (gpm/ft 2 ) D AX (cm 2 /s) Liquid velocity, U L (cm/s) Data Correlation DAX= 1.98*Re0.84Ga0.10 Figu re 6 7. Dispersion coefficients for the TPOT with commingled packing obtained from the tracer tests (data) and determined by the empirical correlation fitted to the data. t m = 5.1059xQ L + 60.844 R > 0.99 0.0 1.0 2.0 3.0 4.0 5.0 0 10 20 30 40 50 60 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 Liquid Loading Rate (gpm/ft 2 ) Mean Residence time, t m (s) Volumetric Liquid Flowrate (L/min) Figure 6 8 M ean residence time as a function of liquid loading rate for the TPOT packed with the commingled packing
140 0 100 200 300 400 500 600 0 20 40 60 80 100 120 140 Toluene concentration ( g/L) Packed Bed Depth, Z (cm) Lm=1.7gpm/ft^2 Lm=3.3gpm/ft^2 Lm=4.7gpm/ft^2 Lm=5.6gpm/ft^2 Data Figure 6 9 A queous phase toluene concentrations predicted using the PFD model as a function of the depth of the packing for different liquid loading rates ( Lm) Co = 495 37 g/L for all loading rates. 0 200 400 600 800 1000 1200 1400 0 20 40 60 80 100 120 140 Toluene concentration ( g/L) Packed Bed Depth, Z (cm) Cin = 160 Cin = 250 Cin = 908 Cin = 1404 Data Figure 6 10. Aqueous phase toluene concentrations predicted using the PFD model as a function of the depth of the packed tower for different initial toluene concentrations ( Cin). QL = 3.8 L/min for all concentrations
141 y = 0.0052x 0.48 R = 0.9532 0.0E+00 2.0E 04 4.0E 04 6.0E 04 8.0E 04 1.0E 03 1.2E 03 1.4E 03 1.6E 03 1.8E 03 0 20 40 60 80 K LS a C (s 1 ) Reynolds Number (Re) Data Power (Data) Figure 6 1 1 Liquid -solid mass transfer coefficie nt as a function of the Reynolds number. y = 0.0002x 0.2464 R = 0.9836 0.0E+00 2.0E 04 4.0E 04 6.0E 04 8.0E 04 1.0E 03 1.2E 03 1.4E 03 1.6E 03 1.8E 03 0 250 500 750 1000 1250 1500 K LS a C (s 1 ) Inlet Toluene Concentration ( g/L) Data Power (Data) Figure 6 1 2 Liquid -solid mass transfer coefficient as a function of the inlet toluene concentration.
142 0.000 0.200 0.400 0.600 0.800 1.000 0 250 500 750 1000 1250 1500 K LS a C x 10 3 (s 1 ) Inlet Toluene Concentration, C IN ( g/L) Correlation Data KLSaC(s-1103) = 0.6493 Re-0.476CIN 0.234 Figure 6 1 3 Actual versus liquid -solid mass transfer coefficients calculated from the correlation in Equati on 6 19 as a function of Re and CIN.
143 CHAPTER 7 S IMULTANEOUS GAS AND AQUEOUS PHASE TOLUENE DEGRADATION USING THE PILOT SCALE REACTOR Gas -Liquid Mass Transfer Coefficient ( KGLaw) The gas liquid mass transfer coefficient is an important parameter in the m odeling of the two -phase reactor. Consequently, the overall gas liquid mass transfer coefficients were determined for both packing configurations used in the large scale reactor: high flow rings only and palls rings commingled with STCs. These coefficients were obtained for different air to water ratios by holding the water flowrate constant and changing the air flowrate in a setup similar to a stripping experiment where the aqueous phase is contaminated and the gas phase is free of contaminants The air f lowrate was kept constant only long enough to reach steady state (at least 5 min). Once steady state was achieved, the air flowrate was increased to reach the next steady state, at which point influent and effluent water phase concentrations were measured. During both experiments, the water flowrate was kept constant. The effect of liquid loading rate was determined by performing the experiment at different water flowrates. Similar to other experiments performed in the TPOT, the packing was wetted in the wa y described in the experimental section. Based on the single phase experiments, adsorption was not a concern for these experiments since it was shown to be insignificant under wetted packing conditions. The overall gas -liquid mass transfer coefficients were calculated from the measured influent and effluent aqueous phase concentrations using Equations 2 39 through 2 4 4 which are repeated here for convenience The results were compared to KG Law values predicted by the Onda correlation, presented in Chapter 2 The Onda correlation given by Equations 2 45 through 2 47 and repeated here for convenience uses certain packing characteristics usually obtained from the packing manufacturer such as the nominal packing diameter ( dp), the critical surface tension (c) and the specific surface area ( at) These characteristics were determined for the STCs in the
144 laboratory. Therefore, the properties used for the calculations when using the commingled packing were based on weighted average of the individual properties available or calculated for both types of packing. The properties were weighted by the bulk volume that each type of packing occupies in the reactor because this weighting method resulted in better fits than using a surface area weighting The properties for high flow rings were obtained from the manufacturer; however, for the STCs, they had to be either calculated, such as the specific surface area (area of cylinder/ volume of cylinder) or found in the literature (e.g. critical surface tension). The packi ng characteristics used for KLSaw calculations are shown in Table 7 1 Z = HTU NTU (2 39) HTU = QL / (A KL aw) (2 40) R = (QG H) / QL (2 41) 0(1)1 ln 1eC CR R NTU RR (2 42) 0() ()me LlmLCC Z KaDF (2 43) 00 00()() () ln ln ()eSeS lm S e eSDFDFCCCC DF DF CC DF CC (2 44) 0.1 0.2 0.05 0.75 22 21exp1.45cmmt m wt tLL LtLLaL aa aga (2 45) 2 1 0.5 3 3 0.40.0051m LL L tp wLLL LL k ad aDg (2 46) 1 0.7 3 25.23g m g tg tp tgggG kaD ad aD (2 47)
145 The results, presented in Figure 7 8 show that the Onda correlation underestimates the actual mass transfer coefficient for all liquid loading rates used and for both types of packing Although the Onda correlation has shown good agreement for many packing styles used in different towers operating at a wide variety of conditions, it h as not been validated for the high flow rings used in this work. For the case of the commingled packing, the disagreement with the cor relation is expected given that the mixed packing is not a conventional packin g, and therefore many of the parameters were arbitrarily estimated. Interestingly, however, it was found that by increasing the wetted surface area proposed by Onda et al. (1968), the mass transfer coefficients could be closely predicted using Equations 2 4 6 and 2 47. The equation predicting the we tted surface area (Equation 2 45) for the Onda correlation, resulted in low wetted areas for the TPOT under the operating co nditions s tudied. Therefore, an approach to improve the predictions using the Onda correlation was to increase the wetted surface area prediction for the reactor, since visual assessment of this parameter during the towers operation seemed to indicate that a greate r percent of the total area was actually being wetted. Accordingly, the equation to predict the wetted area was modified by changing the exponential coefficient used for the ratio of the critical surface tension to the liquid surface tension (c / ) from 0.75 to 0.075, thereby minimizing the effects of the surface tension in the wetting process This change in coefficient resulted in the best fit of the p redicted KGLaw by the Onda correlation to the measured values. The modified equation used to predict the wetted area is shown in Equation 7 1 0.1 0.2 0.05 0.075 22 21exp1.45cmmt m wt tLL LtLLaL aa aga (7 1 )
146 The effect of the modified wetted area equation in the predicted wetted areas for the case of the commingled packing compared to the values obtained by the Onda equation are shown as a function of the liquid flowrate in Figure 7 2 The comparison between the wetted area calculated from the Onda versus the modified Onda equation show that the increase in wetted area is large between 10 and 20 percent for the range of flowrates plotted. Using the modified wetted area, new values for KGLaw were obtained (called modified Onda), and these can be observed in Figure 7 1 in relation to the data and the Onda predictions. There is an excellent agreement between t he data and the Modified Onda predictions for all liquid loading rates tested and for both types of packing used in the tower. This good agreement suggests that the actual wetted area in the reactor is likely higher than the area predicted by the Onda corr elation. The mass transfer coefficients measured for the commingled packing were higher than those measured for the high flow rings only ( Figure 7 3 ). The increase in KGLaw by using the commingled packing can be explained by the larger overall surface area for mass transfer provided by the mixed packing as well as the smaller effective diameter of the packing. As shown in Figure s 7 1 and 7 4 t he overall mass transfer coefficients for both types of packing increased as the air to water ratio increased. As observed in Equation 2 4 7 increasing the gas flowrate results in an increase of the local gas side mass transfer coefficient (kg). Consequently, the gas phase resistance decreases meaning that the overall mass transfer coefficient is increased by lower ing the mass transfer resistance through the film (Equation 2 37) The effect of the liquid loading rate on KGLaw was also investigated by increasing the liquid flowrate during the experiments. The results, presented in Figure 7 5 show that the coefficie nts
147 increase linearly as a function of liquid flowrate The main factor affecting the increase in mass transfer coefficient is the increase in wetted area as a result of the higher liquid loading rate, which increase d from about 97 m2/m3 for 2 L/min to abo ut 140 m2/m3 for 5.7 L/min ( Figure 7 2 ). Simultaneous Two -Phase Degradation Studies These experiments tested the main hypothesis of this research: can a counter flow reactor packed with STCs efficiently enable the simultaneous treatment of gas and aqueous phase toluene ? During these experiments, both streams were contaminated with toluene, and they were allowed to flow simultaneously through the packed bed. Inlet and outlet concentrations in both phases were recorded to determine the overall removal effic iency achieved by the photocatalytic tower. The water flowrate during these experiments was kept constant at 3.8 L/min (liquid loading of 3.1 gpm/ft2) while the air flowrate varied from run to run. When considering the two -phase countercurrent flow of a compound in a packed tower, the direction of transfer of the compound from the gas phase to the aqueous phase or vice versa will be determined by concentrations of toluene in the aqueous and gas phases relative to equilibrium concentrations, as determin ed by Henrys law, in the tower These concepts were explained and reviewed in Chapter 2 Two saturation conditions of toluene can exist in the reactor: The aqueous phase is supersaturated with respect to the gas phase, meaning that the toluene present in t he aqueous phase will tend to transfer to the gas phase. This condition is referred to as stripping. The aqueous phase is undersaturated with respect to the gas phase, meaning that the toluene in the gas phase will tend to transfer to the aqueous phase. Th is condition is referred to as absorption. In the countercurrent flow reactor, the saturation conditions were determined for the top and bottom sections of the reactor. T he conditions at the top of the reactor are represent ed by the influent aqueous phase (Co) and the effluent gas phase ( Ye) concentrations while the bottom of
148 the reactor is represented by the effluent aqueous phase ( Ce) and influent gas phase concentrations ( Yo). These concentrations were plotted in a graph with respect to the equilibrium conditions predicted by the Henrys law in order to determine the tendency of mass transfer in the reactor for the different experiments performed. All the following references to supersaturated and undersaturated are used with respect to the aqueous phase The actual saturation conditions given at the top and bottom of the tower during each of the six two -phase experimental runs (labeled from a to f) and the predicted equilibrium conditions are shown in Figure 7 6 Based on the points representing the co nditions occurring at the top of the tower ( illustrated by squares), one of the experiments was performed with the solution unsaturated with respect to the gas phase at the water inlet at the top of the reactor (i.e. above the equilibrium line) while the r est of the experiments were performed under supersaturated solution conditions (i.e. points are below the equilibrium line). The experiments performed under supersaturated conditions were analyzed in terms of the driving force and the air to water ratio. I nterestingly, the conditions at the bottom of the reactor (represented by the triangles) are very close to the equilibrium line predicted by the Henrys law. These results show that, similar to a stripping/absorption tower, the system tends to move towards equilibrium The letter shown in the legend of Figure 7 6 refers to the name given to the experiment; top and bot refer to the conditions at the top and bottom of the reactor, respectively and the number that follows is the gas flowrate. The results f or each one of the two phase experimental runs are shown in Figure s 7 7 through Figure 7 12. The format for each figure is the same for the different runs. Each figure shows 4 different graphs that can be described as follows: Graphs (A) and (B) show the c hange of the aqueous phase and gas phase concentrations, respectively, as a function of run time (t = 0 when UV lamps were turned on). These plots also include the expected concentrations that would be obtained in both, the aqueous and gas phases, if no toluene oxidation occurred, and there was only mass transfer in the
149 reactor ( Ce if no PCO). These expected concentrations were determined using KG Law values calculated with the modified Onda c orrelation previously determined. Graph (C) show s the toluene mass balance, i.e. the actual overall mass flow into and out of the reactor as calculated using Equation 7 2 Graph (D) show s the summary of the removal of toluene (decrease in toluene concentration represented by positive removal or increase in concentration represented by a negative removal) for the gas phase (1 Ye/Yo) and the aqueous phase (1 Ce/Co). The overall mass flow removal (1 Me/Mo) is also shown. QGYo + QLCo= QG Ye + QL Ce (7 2 ) All the experimental condi tions for runs (a) through (f) are summarized Table 7 2 where QG refers to the volumetric gas flowrate and is given in units of L/min The volumetric liquid flowrate, QL, used for all runs shown in Table 7 2 was 3.8 L/min. The toluene removal in the aqueo us phase was calculated as (1 Ce/Co). The gas phase removal was determined by (1 Ye/Yo). The actual percent toluene removal refers to the net removal in the tower due to mineralization of toluene, and it was calculated by determining the mass flowrates at top and bottom of the reactor Figure 7 13 illustrates the results for the two -phase experiments plotted as the overall removal of toluene in the two phases as a function of the aqueous phase concentration. Additionally, the expected removal of toluene in the aqueous phase as predicted by the correlation shown in Figure 64 for the single phase experiments is also plotted as a function of concentration in the same figure. Notice that the linear correlation for the aqueous phase included concentrations u p to about 1400 g/L Since higher concentrations were used in the two -phase experiments, the line obtained by the correlation was extrapolated to the highest co ncentration observed in the twophase experiments. T he results shown in Figure 7 13 and Table 7 2 indicate that for all the experiments considered as supersaturated at the top of the reactor, i.e. experiments (a) to (e) (See Figure 7 13), the tendency was to strip toluene from the
150 aqueous to the gas phase All the supersaturated experiments showed a large removal of toluene from the aqueous phase (>90%). The gas phase, on the other hand, showed an increase in concentration, and thus negative percent removals. This incre ase was the result of stripping, i.e. the transfer of toluene from the aqueous pha se to the gas phase. Although the gas phase concentration increased significantly, toluene was accumulated in the gas phase at a lower rate than if photocatalysis did not occur as shown by graph B in Figures 7 14 to 718, meaning that to luene was oxidized in the tower For the experiment considered undersaturated, absorption of toluene from the gas phas e to the aqueous phase occurred and no significant removal was achieved in either phase. Based on Figure 7 13, it is evident that the removal of toluene occ urs in the aqueous phase and that for the most part, higher removals can be achieved when the aqueous phase is the only flowing phase. With the exception of experiments b and c, which achieved removals higher than predicted by aqueous phase photocatalysis only, having the two phases flowing simultaneously results in a lower net toluene removal. The main factors that influence the overall re moval of toluene during the two phase PCO include those parameters that will tend to affect the aqueous phase concentra tion in the reactor since the removal occurs mostly in the aqueous phase. Factors that decrease the driving force in the aqueous phase, and thus the overall toluene removal, can be compared from the results presented in Table 7 2 : Air to water ratio : The s upersaturated experiments showed a net removal of toluene between 37 % and 68%, except for experiment (d), which did not show a substantial mass removal (only 6.2 %). The major operational difference associated with experiment (d) was the higher gas flowrate In general, it was observed that by increasing the gas flowrate, the net
151 toluene removal significantly decreased. By increasing the gas flowrate (or air to water ratio), the stripping of toluene to the gas phase is expected to be greater. Based on the si ngle phase experiments; however, toluene is not photocatalyzed in the gas phase, but in the aqueous phase. Therefore, increasing the stripping of toluene decreases the aqueous phase concentration which is the driving force for photocatalysis, and hence the overall removal is adversely affected. Aqueous phase concentration : The influent aqueous phase concentration for experiment (b) is about twice that of experiment (a). The net toluene removal at the higher aqueous concentration was 62.5% compared to 36.6% for experiment (a). The results confirm that the aqueous phase concentration is one of the main driving forces for phototcatalysis in the reactor. Gas phase c oncentration : Lower influent gas phase concentrations can adversely influence the removal of tolu ene in the system because it increases the driving force for toluene transfer to the gas phase by stripping, but the toluene present in this phase will not be oxidized. This tendency can be observed by comparing experiments (b) and (e), which have the same gas flowrate, comparable initial aqueous phase concentrations and water temperatures, but the influent gas phase concentration for experiment (b) is about 6 times larger than for experiment (e). The gas phase concentration difference results in an increa sed water phase removal and an overall decreas e in toluene mass removal of 15% for experiment (e) as compared to (b). Temperature : This parameter was not intentionally varied in this study to determine its effect in the process. However as a result of heat discharged by the UV lamps, small temperature differences were noted. It is well established that, on one hand, higher temperatures increase the volatility of toluene and thus its stripping tendency, so increasing the temperature in the system can have ad verse effects on the net removal of toluene. On the other hand, higher temperatures usually increase oxidation rates, thus favoring toluene removal in the system. The dominant
152 effect cannot be inferred from the experimental runs since the temperature diffe rences were so small. For the case of undersaturated conditions at the water inlet the toluene concentration in the aqueous phase increased while it decreased in the gas phase, i.e. there was toluene absorption in the tower. These results are consistent w ith the undersaturated nature of the system leading to toluene absorption from the gas to the aqueous phase However, there was no net destruction of toluene in the system. Again, photocatalysis in the gas phase seems to be inhibited when the packing was w etted. For undersaturated conditions, most of the toluene was present in the gas phase, so these toluene molecules did not undergo oxidation. The fewer molecules that were transferred to the aqueous phase could potentially undergo photocatalysis. However, the rate of transfer between the phases, the time the molecules need to be in contact with the catalyst once toluene is transferred to the aqueous phase, and the rate of toluene oxidation in the aqueous phase would need to be examined in detail in order to predict the conditions under which toluene removal would occur for undersaturated conditions. By adjusting the conditions to favor the aqueous phase concentration to increase, the overall removal of toluene can be improved in the two phase system as compa red to the single phase. Although most of the removal in the two -phase photocatalytic tower seems to be occurring in the aqueous phase, the complete two -phase system proposed in this study, consisting of the two phase photocatalytic oxidation tower and the two end polishing sections for each phase, has many advantages compared to other systems used to re move VOCs. The main advantage of the TPOT is its versatility d ue to the stripping/ absorption potential of the in the reactor
153 Table 7 1 Individual and averaged properties for the two different packing materials used to pack the TPOT. Property STC High flow rings Commingled Nominal diameter d p (mm) 9.1 15.9 14.8 Specific surface area a t (m 2 /m 3 ) 452.1 313.0 334.5 Critical surface tension c (N/m) 0.027* 0.033 0.0321 Bulk Volume V (L) 6.0 32.8 Value obtained from Gould and Irene (1988). Table 7 2 Summary of operational conditions and toluene removals for the simultaneous two phase experiments. Exp. QG (L/min) Co (g/L) Yo (g/L) Avg. Twater (oC) A queous Removal (%) Gas Removal (%) Net % Toluene Removal Predicted % Aqueous Rem oval Supersaturated Conditions a 142 1288 23 23.5 95.3 52.1 36.6 47.8 b 142 2419 11 24.4 90.2 95.5 62.5 61.5 c 116 2506 6 25.7 95.1 3 40.4 68.0 62.5 d 202 1258 8 25.1 90.7 246.6 6.2 47.4 e 142 2642 2 24.6 94.7 1953.2 47.2 64.2 Undersaturated Conditions f 142 424 161 24.4 99.2 0.12 5.0 37.3
154 0.00 0.10 0.20 0.30 0.40 0 20 40 60 80 100 120 K GL a w (min 1 ) Air to Water Ratio (Q G /Q L ) Data Onda Modified Onda (C) PR & STCs QL= 5.7 L/min 0.00 0.10 0.20 0.30 0.40 0 20 40 60 80 100 120 K GL a w (min 1 ) Air to Water Ratio (Q G /Q L ) Data Onda Modified Onda (A) PR & STCs QL= 2.1 L/min 0.00 0.10 0.20 0.30 0.40 0 20 40 60 80 100 120 K GL a w (min 1 ) Air to Water Ratio (Q G /Q L ) Data Onda Modified Onda (B) PR & STCs QL= 4.0 L/min 0.00 0.10 0.20 0.30 0.40 0 20 40 60 80 100 120 K GL a w (min 1 ) Air to Water Ratio (Q G /Q L ) Data Onda Modified Onda (D)PR Only QL= 4.0 L/min Figure 7 1 Comparison of the measured overall gas -liquid mass transfer coefficients to the coefficients predicted by the Onda correlation and by the Modified Onda correlation for different liquid flowrates and packing styles as a function of air to water ratio. A) to C) show results for the commingled packing while D) refers to the high flow rings only.
155 0 20 40 60 80 100 0 50 100 150 200 250 300 0 10 20 30 40 % Wetted Area Wetted Area (m 2 /m 3 ) Volumetric Liquid Flowrate (L/min) Onda Modified Onda Figure 7 2 Comparison of the predicted wetted surface area of the packing calculated using the Onda correlation and the Modified Onda correlation as a function of liquid flowrate for the tower packed with the commingled pac king. 0.000 0.100 0.200 0.300 0.400 0 20 40 60 80 K GL a w (min 1 ) Air to Water Ratio (Q G /Q L ) PR Only PR & STCs Modified Onda (PR) Modified Onda (STCs/PR) Figure 7 3 Comparison of overall gas liquid mass transfer coefficients for the tower packed with high flow rings only and the commingled packing as a function of air to water ratio for a liquid flowrate of 4.0 L/min.
156 0.000 0.050 0.100 0.150 0.200 0.250 0.300 0.350 0.400 0 20 40 60 80 100 120 K GL a w (min 1 ) Air to Water Ratio (Q G /Q L ) 2.1L/min (Data) 4.0L/min (Data) 5.7L/min (Data) 2.1L/min (Mod. Onda) 4.0L/min (Mod. Onda) 5.7L/min (Mod. Onda) Figure 7 4 Overall gas liq uid mass transfer coefficients as a function of air to water ratio for different liquid flowrates for the tower packed with commingled packing. 0.00 0.10 0.20 0.30 0.40 0 1 2 3 4 5 6 7 8 K GL a w (min 1 ) Volumetric Liquid Flowrate, Q L (L/min) QG = 3cfm QG = 5cfm QG = 7cfm Figure 7 5 Effect of the liquid flowrate on KGLaw for different gas flowrates for the tower packed with the c ommingled packing.
157 0 20 40 60 80 100 120 140 160 180 200 0 500 1000 1500 2000 2500 3000 Gas Phase Toluene Concentration ( g/L) Aqueous Phase Toluene Concentration ( g/L) a,top 5scfm a,bot 5scfm b,top 5scfm b,bot 5scfm c,top 4scfm c,bot 4scfm d,top 7scfm d,bot 7scfm e,top 5scfm e,bot 5scfm f, top 5scfm f, bot 5scfm Equilibrium H Figure 7 6 Comparison of actual saturation conditions in the reactor for the different experimental runs and the equilibrium concentrations predicted by the Henrys law
158 0 200 400 600 800 1000 1200 1400 1600 20 40 60 90 120 Aqueous Phase Concentration (ug/L) Run Time (min) (A) Water Influent (Co) Water Effluent (Ce) Ce if No PCO 0 10 20 30 40 50 60 20 40 60 90 120 Gas Phase Concentration (ug/L) Run Time (min) (B) Gas Influent (Yo) Gas Effluent (Ye) Ye if No PCO 0 2000 4000 6000 8000 10000 20 40 60 90 120 Toluene Mass Flowrate (ug/min) Run Time (min) (C) Mass In Mass Out 2.0 3.0 8.0 13.0 18.0 23.0 28.0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 50 100 150 Toluene Removal in the Gas Phase (1 Ye/Yo) Toluene Removal in Liquid Phase and Overall (1 Ce/Co) or (1 Me/Mo) RunTime (min) (D) Aqueous Phase Total Mass Gas Phase Figure 7 7 Results for two -phase experiments using satu rated cond i tions : Experiment (a) conditions : QL= 3.8L/min QG=142L/min, Twater = 23.5oC. (A) Aqueous phase (B) Gas phase (C) Toluene mass balance (D) Summary of removal s
159 0 500 1000 1500 2000 2500 10 20 30 40 50 Aqueous Phase Concentration (ug/L) Run Time (min) (A) Water Influent (Co) Water Effluent (Ce) Ce if No PCO 0 10 20 30 40 50 60 70 80 10 20 30 40 50 Gas Phase Concentration ( g/L) Run Time (min) (B) Gas Influent (Yo) Gas Effluent (Ye) Ye if No PCO 0 2000 4000 6000 8000 10000 12000 10 20 30 40 50 Toluene Mass Flowrate ( g/min) Run Time (min) (C) Mass In Mass Out 2.0 3.0 8.0 13.0 18.0 23.0 28.0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 20 40 60 Toluene Removal in the Gas Phase 1 Ye/Yo Toluene Removal 1 Ce/Co or 1 Me/Mo RunTime (min) (D) Aqueous Phase Total Mass Gas Phase Figure 7 8 Results for two -phase experiments using saturated cond i tions : Experim ent ( b ) conditions : QL = 3.8 L/min QG = 142 L/min, Twater = 24.4oC. (A) Aqueous phase (B) Gas phase (C) Toluene mass balance (D) Summary of removals
160 0 500 1000 1500 2000 2500 10 20 30 40 50 Aqueous Phase Concentration (ug/L) Run Time (min) (A) Water Influent (Co) Water Effluent (Ce) Ce if No PCO 0 10 20 30 40 50 60 70 80 90 10 20 30 40 50 Gas Phase Concentration ( g/L) Run Time (min) (B) Gas Influent (Yo) Gas Effluent (Ye) Ye if No PCO 0 2000 4000 6000 8000 10000 10 20 30 40 50 Toluene Mass Flowrate ( g/min) Run Time (min) (C) Mass In Mass Out 6.0 1.0 4.0 9.0 14.0 19.0 24.0 29.0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 20 40 60 Toluene Removal in the Gas Phase 1 Ye/Yo Toluene Removal 1 Ce/Co 1 Me/Mo RunTime (min) (D) Aqueous Phase Total Mass Gas Phase Figure 7 9 Results for two -phase experiments using saturated cond i tions : Experiment ( c) conditions: QL = 3.8 L/min QG = 116 L/min, Twater = 25.7oC A) Aqueous phase (B) Gas phase (C) Toluene mass balance (D) Summary of removals.
161 0 200 400 600 800 1000 1200 1400 10 20 30 40 50 Aqueous Phase Concentration (ug/L) Run Time (min) (A) Water Influent (Co) Water Effluent (Ce) Ce if No PCO 0 5 10 15 20 25 30 35 10 20 30 40 50 Gas Phase Concentration (ug/L) Run Time (min) (B) Gas Influent (Yo) Gas Effluent (Ye) Ye if No PCO 0 1000 2000 3000 4000 5000 6000 7000 8000 10 20 30 40 50 Toluene Mass Flowrate (ug/min) Run Time (min) (C) Mass In Mass Out 4.0 0.0 4.0 8.0 12.0 16.0 20.0 0.2 0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 20 40 60 Toluene Removal in the Gas Phase 1 Ye/Yo Toluene Removal 1 Ce/Co or 1 Me/Mo RunTime (min) (D) Aqueous Phase Total Mass Gas Phase Figure 7 10. Results for two -phase experiments using saturated cond i tions : Experiment ( d ) conditions : QL = 3.8 L/min QG = 202 L/min, Twater = 25.1oC (A) Aqueous phase (B) Gas phase (C) Toluene mass balance (D) Summary of removals.
162 0 500 1000 1500 2000 2500 3000 30 60 90 120 Aqueous Phase Concentration (ug/L) Run Time (min) (A) Water Influent (Co) Water Effluent (Ce) Ce if No PCO 0 10 20 30 40 50 60 70 80 30 60 90 120 Gas Phase Concentration (ug/L) Run Time (min) (B) Gas Influent (Yo) Gas Effluent (Ye) Ye if No PCO 0 3000 6000 9000 12000 30 60 90 120 Toluene Mass Flowrate (ug/min) Run Time (min) (C) Mass In Mass Out 30.0 20.0 10.0 0.0 10.0 20.0 30.0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 50 100 150 Toluene Removal in the Gas Phase 1 Ye/Yo Toluene Removal 1 Ce/Co, or 1 Me/Mo RunTime (min) (D) Aqueous Phase Total Mass Gas Phase Figure 7 11. Results for two -phase experiments using saturated cond i tions : Experiment ( e ) conditions : QL = 3.8 L/min QG = 142 L/min, Twater = 24.6oC (A) Aqueous phase (B) Gas phase (C) Toluene mass balance (D) Summary of removals.
163 0 200 400 600 800 1000 1200 30 60 90 120 150 Aqueous Phase Concentration (ug/L) Run Time (min) (A) Water Influent (Co) Water Effluent (Ce) 0 20 40 60 80 100 120 140 160 180 200 30 60 90 120 150 Gas Phase Concentration (ug/L) Run Time (min) (B) Gas Influent (Yo) Gas Effluent (Ye) 0 5000 10000 15000 20000 25000 30000 1 2 3 4 5 Toluene Mass Flowrate (ug/min) Run Time (min) (C) Mass In Mass Out 2 1.6 1.2 0.8 0.4 0 0.4 0.8 0 50 100 150 200 Toluene Removal 1 Ye/Yo, 1 Ce/Co or 1 Me/Mo RunTime (min) (D) Aqueous Phase Gas Phase Total Mass Figure 7 12. Results for two -phase experiments using under saturated cond i tions : Experiment ( f ) conditions: QL = 3.8 L/min QG = 142 L/min, Twater = 24.4oC (A) Aqu eous phase (B) Gas phase (C) Toluene mass balance (D) Summary of removals
164 e c b a d f 0 10 20 30 40 50 60 70 80 90 100 0 500 1000 1500 2000 2500 3000 Overall Toluene Removal (%) Aqueous Phase Initial Toluene Concentration, C o ( g/L) Actual Removal in Two Phase Experiments Removal Predicted for Aqueous Phase Only Figure 7 13. Comparison of the removal of toluene in the two -phase experiments to the expected removal in the aqueous phase only (i.e. in the absence of the gas flow)
165 CHAPTE R 8 MATHEMATICAL MODELIN G AND SIMULATION OF THE TPOT Model Development The two -phase oxidation tower used in this research was mathematically modeled with the purpose of predicting its performance for a ny specified set of conditions and tracking the concen tration profile throughout the length of the reactor The temperature differences in the reactor when both phase s were simultaneously flowing were considered to be small (less than 3oC difference) so isothermal models were appropriate for the simulation of the system investigated. Given that the ratio of the tower diameter to the packing diameter was between 12 and 22, the conditions in the radial direction were considered to be uniform i.e. that the axial velocity over the cross section of the bed is constant (Sater and Levenspiel, 1966). Accordingly, the conversion of toluene was predicted by using one -dimensional models. Given that the gas phase velocities were high, a plug flow with dispersion model was used for toluene in the gas phase as it is usuall y for many stripping towers (Sater and Levenspiel, 1966). For aqueous phase toluene, however, a plug flow with dispersion model was considered. To describe the flow through the tower mathematically and solve the model s some simplifying assumptions were ma de: The axial dispersion is negligible for the gas phase Adsorption of toluene for either phase is negligible Gas and liquid phase concentration profiles are a function of the packing bed length only The area for mass transfer between gas and liquid is the wetted area determined using the total surface area of comingled packing that is available in the reactor The area for the mass transfer between liquid and solid is only the wetted area determined using the catalyst surface area available in the react or. Schematics of the reactor showing the inflows and outflows for the contaminant as well as the differential reactor volume used to write the mass balances on toluene are shown in Figure 8 -
166 1 Additionally, Figure 8 2 shows the mass transfer resistances for the gas, liquid and solid phases included in the modeling of the reactor. Applying the selected models, the mass conservation equations written for each phase, with their respective boundary conditions are shown in Equations 8 1 to 8 3 for the aqueous phase and 8 4 to 85 for the gas phase. The solid phase mass balance is also shown in Equation 8 6. Toluene in the aqueous phase: 2 20AX L GLw LSc SdCdC Y DUKaCKaCC dzdz H (8 1) Boundary conditions At z = 0 AX LINdC DUCC dz (8 2) At z = Z 0 dC dz (8 3) Toluene in the gas phase: 0G GLwdY Y UKaC dz H (8 4) Boundary conditions At z = 0 0 dC dz (8 5) The equality of mass transfer and re action rate is given by: netLSc SrKaCC (8 6)
167 Similar to the aqueous phase solution, the above equations were solved numerically to obtain the axial profiles for the concentrations of toluene in the gas and aqueous phases Both models were solved using the finite difference approximation methods. Spatial derivatives as well as boundary conditions were expressed by central differences approximations shown by Equations 6 15 and 616. The resulting set of algebraic equations was solved using typical methods to solve linear sets of algebraic equations such as the LU decomposition method. The dispersion coefficient and mass transfer coefficients were estimated using the correlations developed in Chapter 7 for the aqueous phase The dispersion coefficient found for the aqueous phase was used in the two -phase system since many other researchers have found that this coefficient is only marginally affected by the gas flowrate (Piche et al. 2002). The surface concentration, Cs, was as sumed to be zero by using the same assumptions as in the aqueous phase model. A summary of all these correlations is given in Table 8 1. The model was initially solve d for the concentration of toluene as a fun c tion of the packed bed depth u sing the correla tions given in Table 8 1 The initial conditions entered in the model were those used for the two -phase experiments (a through e) performed in the TPOT, which are summarized in Table s 7 3 and 8 2 The concentration profiles using these correlations are pr esented in Figure 8 3 and can be identified by the number 1 in the legend. The results in Figure 8 3 show that the effluent concentration s obtained from the model predictions were higher than the actual effluent concentrations measured for the aqueous and the gas phase s during the actual experiments (shown by the single markers in Figure 83) Accordingly, the net toluene conversions predicted by the model were much lower than the conversions obtained during the experimental runs as shown in Table 8 2
168 Th e model includes three main coefficients estimated in this research: the dispersion coefficient the gas liquid mass transfer and the liquid -solid mass transfer coefficient s In order to improve the model predictions, the effects of these coefficients were investigated The dispersion coefficients were not modified because they were found to mostly influence the initial removal of touene compared to the removal at the end of the packed section. As shown in Figure 8 4 increasing the dispersion coefficient r esulted in most of the liquid ph ase removal of toluene to occur in the first few centimeters of the packed section, which did not appear to be very realistic. The mass transfer coefficients, however, had a larger effect on the removal of toluene throughou t the entire length of the packing. Varying the gas liqud mass transfer coefficients resulted in better fits to the gas phase data. Similarly, the liquid -solid mass transfer coefficients influenced mostly the aqueous phase concentrations. Trying to calibra te the model with either coefficient individually did not result in good predictions for both phases. Consequently the model was calibrated by fitting both, the gas -liquid and the liquid-solid mass transfer coefficients to each set of data individually Th e data available for the two -phase experiments only includes the concentrations measured at the inlet and the outlet of the reactor, so the fitting was performed to match the effluent concentration for both phases. Since the reactor flows in a countercurre nt mode, the effluent gas phase concentration is given at z = 0 while the aqueous phase concentration at the outlet occurs at z = Z as shown in Figure 8 3. The calibration was performed by changing both mass transfer coefficients to minimize the SSqD at t he effluent concentrations. The concentration profiles obtained after calibrating the model with the mass transfer coefficients are shown in Figure 8 3, and these profiles can be identified by the number 2 in the
169 legend. Using the fitted liquid -solid mass transfer coefficients resulted in a better fit as compared to the results obtained from the correlation suggesting that there are other phenomena occurring in the reactor when both phases are present that are not being accounted for by the model or cannot be lumped in the KLSac estimates for just the aqueous phase Fitting the KLSac and KGLaw resulted in good agreement for both the gas and aqueous phase concentrations; so the net conversions were very closely approximated with the new values of the mass tr ansfer coefficients ( Figure 8 5 and Table 8 2.) As presented in Table 8 3 results for the liquid -solid mass transfer coefficients used for the calibrated model were larger than the ones predicted from the correlations whereas the gas liquid mass transfe r coefficients did not follow a specific overall tendency compared to the values predicted by the Modified Onda correlation. The mass transfer coefficients predicted by the correlations given in Table 8 1 were very similar for all the experimental conditio ns investigated. This is because all the experiments were performed at the same liquid loading rate, and this is the parameter that seems to have the most effect on the determination of both coefficients using these correlations However, the calibrated co efficients suggest that these coefficients might be also significantly dependent on the gas phase loading rate because larger differences for both mass transfer coefficients were observed for the cases where the gas loading rate was changed. The gas loading rate appears to have a large positive impact on the values of KGLaw and the opposite impact on KLSac. Although the influence of the gas loading rate seems to be important for the determination of both coefficients, a specific trend was not observed for e ither case as a function of gas loading rate. Model Simulation The main objective of modeling the reactor is to be able to make predi ctions about its performance given a set of operating conditions. Therefore, in this section, the performance of
170 the comple te proposed system, TPOT a nd the end polishing sections is illustrated. The initial set of operating conditions is given in Table 8 3.These conditions are similar to the ones used for the two -phase experiments. Also, the reactor used for the simulation is the TPOT, meaning that its dimensions and packing characteristics have not changed. Although the dimensions of the system have been already set for the middle two -phase section, the model can be solved for the length of this section if the desired outlet a queous phase concentration is specified. The final target effluent concentrations, i.e. the concentrations desired at the end of the polishing sections were chosen to be the inhalation chronic reference exposure level for toluene and the MCL for benzene in drinking water, which are equal to 0.30 g/ L and 5 g/L respectively These values were chosen for this section of the study, although other values could be chosen to meet other required air and water concentrations. The MCL for benzene was chosen inste ad of toluene because toluenes MCL is 1 mg/L, which is rather high for the inlet concentrations used in the simulation, so a polishing section for the aqueous phase would not be required to meet th at MCL and for the purposes of this simulation, it was de sired to have large enough aqueous concentrations at the effluent of the TPOT relative to the MCL, so that a liquid phase polishing section would be required to meet the target effluent contaminant level. The two -phase model used the correlations specified in Table 8 1, but the liquid -solid mass transfer coefficient was multiplied by a factor of 3.5 since it was found that these values were likely underestimated by the correlation when used in the model of the two-phase system for most conditions studied. T he concentration profile for the operating conditions specified is shown in Figure 8 6 Furthermore, the concentration profile for the same system operated in the absence of photocataly sis (i.e. with the UV lamps off, so that the only process taking place is mass transfer ) were also plotted in the same graph. The net removal of toluene from the tower,
171 s hown in Figure 8 7 indicates that about 42% of the contaminants mass is removed in this reactor for the operating conditions specified Based on the simul ation for the middle section of the system, the concentrations leaving the TPOT and thus entering the polishing sections were 31.2 g/L and 111.8 g/L for the gas and aqueous phases, respectively. The gas phase reactor was sized based on the pseudo first o rder reaction rate constant found during the small scale gas phase experiments in Chapter 5, which was 0.12 s1. Since the effluent gas from the TPOT is expected to be highly humidified due to its contact with the liquid phase, no significant deactivation of the catalyst is expected to occur in the gas phase polishing section T he liquid phase polishing section was designed using the pseudo first order rate constant found by Holmes et al. (2004), who used a bench -scale reactor packed with the small STCs (3 mm x 5 mm) The concentration range in their experiments was 100 to 200 g/L, which is similar to the concentrations expected to be leaving the two -phase reactor. Different from the middle section, the end polishing sections a re saturated systems, meaning the water is completely filling the reactor instead of just flowing in a trickled mode The gas phase end polishing section w as design ed to ma i nt a in a minimum superficial velocity requirement of 1 cm/s to diminish the influence of mass transfer effects, as determined by the bench-scale experiments. The polishing sections characteristics are shown in Table 8 5 The diameters ( D ) for both sections are larger than those used during the bench -scale experiments; so multiple lamps will need to be included in the reactors to achieve the necessary irradiation and comparable reaction rates The complete treatment system showing the conversion of toluene at the different stages is illustrated in Figure 8 8 Effects of Operating Parameters on the TPOT Performance The effect s of the several operating parameters that can be adjusted in the TPOT during a typical run were in vestigated. An individual parameters effect on toluene conversion was
172 determined by doubling the value of such parameters compared to a base scenario selected. The base scenario had the same initial operating conditions as the one used for the system simulation (QL = 3.785 L/min, QG = 113.5 L/min, Cin = 1500 mg /L, Yin = 10 mg/L, Tw = 24oC). The concentration profiles and removals were determined by solving the two -phase model developed earlier using the correlations given in Table 81 for all required coefficients since based on previous results, these coefficients seem to be a conservative approximation of the actual mass transfer coefficients in the t wo -phase reactor The parameters investigated included the gas and liquid flowrates, the gas and liquid inlet concentrations and the temperature. The aqueous and gas phase concentration profiles are shown in Figures 8 9 and 8 10, respectively. The net conversion of toluene as a function of packed bed depth is presented in Figure 8 1 1 for the different parameters investigated. For the aqueous phase, doubling the initial gas phase concentration, gas flowrate and decreasing the temperature did not have a signi ficant effect on the concentration profile compared to the base scenario. Doubling the liquid loading rate, howe ver, adversely affected the rem o val of toluene in the aqueous phase. T he lower residence time and the thicker liquid film formed as a result of the higher flow rate negatively affects the conversion of toluene to a greater extent than the increase in wetted surface area (~ 20% increase) which improves the conversion of toluene in the aqueous phase. The other factor that influenced the concentratio n profile in the TPOT was increase in initial liquid phase concentration. Even though the concentration was doubled, the concentration sharply decreased with packed bed depth achieving almost the same removal in the aqueous phase as the base scenario, sugg esting that the concentration is an important driving force for the conversion of toluene in the liquid phase.
173 Starting at a concentration of 10 g/L (at Z = 122 cm), the gas phase concentration increased for all conditions studied as it moved through the length of the bed. Factors improving the removal in the gas phase were the water temperature and gas flowrate. Doubling the other parameters resulted in higher gas phase concentration; the liquid flowrate had the highest adverse impact followed by the aque ous inlet concentration and the gas inlet concentration Increasing either the gas or liquid phase concentrations result ed in higher mass of toluene in the reactor, thus it was expected that these concentrations would be higher than the base scenario. Simi larly, doubling the liquid flowrate increases the mass loading, thus compared to the base scenario, the concentrations should be higher if similar removals are expected. However, the results for the doubled gas flowrate were unexpected since increasing the gas flow should result in higher stripping and thus, higher gas phase concentrations as occurred during the experimental runs Decreasing the temperature results in lower H values that decrease the stripping potential, but this parameter does not take in to account its effect on reaction rates. The trends in the profiles shown for the individual phases might not be the same as those showing the net conversion of toluene due to the large difference in flowrates for the gas and aqueous phases. Surprisingly, the parameter that resulted in the highest removal compared to the base scenario was the decreased temperature. By lowering the temperature from 25oC to 10oC, the gas liquid mass transfer coefficient is decreased by a factor of about 2 compared to the bas e scenario, so stripping is significantly affected. However, the effects of temperature on reaction kinetics were not included in the model, so its overall effect cannot be impartially assessed. The increased inlet concentration also resulted in greater re movals and the reasons for this have been previously stated. Doubling either the gas flow or concentration resulted in lower net conversions since they adversely impact the liquid phase removal. Finally, doubling the liquid flowrate
174 resulted in the lowest removal, and the adverse effects of this parameter on rem oval have been explained above. All the effects of the investigated parameters on removal based on the model solution agree with the observed effects obtained from the two -phase experiments. Table 8 1. Summary of correlations used to determine the parameters involved in the solution to the differential equations developed to model the TPOT. Axial Dispersion Coefficient, DAX DAX = 1.98 Re0.84 Ga0.10 (6 11) Overa ll gas liquid mass transfer coefficient, K GL a w 111LwlwgwKakaHka (2 37) 0.1 0.2 0.05 0.075 22 21exp1.45cmmt m wt tLL LtLLaL aa aga (7 1 ) 2 1 0.5 3 3 0.40.0051m LL L tp wLLL LL k ad aDg (2 46) 1 0.7 3 25.23g m g tg tp tgggG kaD ad aD (2 47) Overall liquid solid mass transfer coefficient, K LS a c KLSaC (s1) = 1.2103Re0. 50 CIN 0.25 (6 19)
175 Table 8 2. Comparison of the actual net toluene removal to the net removal predicted by the two phase model using different calibration methods. (1)Ks calculated from correlations in Table 8 1, (2) Both KLSac & KGLaw were modified to cal ibrate the model. Net Toluene Removal using different Ks E xp QG (L/min) Co (g/L) Yo (g/L) Actual (1) K s from correlations (2 ) KLSac & KGLaw fitted a 142 1288 23 0.38 0.19 0.37 b 142 2419 11 0.63 0.29 0.63 c 116 2506 6 0.69 0.30 0.68 d 202 1258 8 0.08 0.23 0.06 e 142 2642 2 0.49 0.32 0.47 Table 8 3 Comparison of the mass transfer coefficients calculated using the correlations developed for the aqueous phase and those found by calibrating the model. (1) Ks from correlations (2 ) K LS a c & K GL a w fitted Exp K GL a w (s 1 x10 3 ) K LS a c (s 1 x10 3 ) K GL a w (s 1 x10 3 ) K LS a c (s 1 x10 3 ) a 4.19 1.30 4.09 4.68 b 4.29 1.52 1.44 4.87 c 4.37 1.54 2.21 6.89 d 4.42 1.29 9.41 1.74 e 4.30 1.56 4.79 5.35 Table 8 4 Operating conditions selected for the simulati on of the systems performance. Parameter Units Value Q G /Q L 30.0 Q L L/min 3.8 Q G L/min 113.5 L m gpm/ft 2 3.1 T water o C 25.0 C in ug/L 1500 .0 Y in ug/L 10.0 Table 8 5 Reactor characteristics for the end polishing sectio ns of the treatment system. Q Z D U Inlet Outlet min L/min M Cm cm/s g/L g/L Gas Phase 0.71 113.5 0.43 49.08 1.0 31.2 0.3 Liquid Phase 18.28 3.8 1.10 28.34 0.10 111.8 5.0
176 Z z z = Z z = 0 Z UL, CoUL, CeUG, YoUG, Ye r V = A*Z Z UL, C(z+z) UL, C(z) UG, Y(z) UG, Y(z+z) Air Water -RA Catalyst Surface NAIaV Figure 8 1. Schematics of the reactor and differential volume used to determine toluenes mass balance equations. Bulk Gas Phase Gas Film Gas-Liquid Interface Liquid Film Liquid-Solid Interface Bulk Liquid Phase Solid Catalyst Liquid Film Figure 8 2. Gas, liquid and solid phases resistances and concentration profiles in the TPOT.
177 0 5 10 15 20 25 30 35 40 45 50 0 400 800 1200 1600 2000 2400 0 20 40 60 80 100 120 140 Gas phase toluene concentration. Y ( g/L) Aqueous phase toluene concetration, C ( g/L) Packed depth, Z (cm) C_1 C_2 Ce_Dat a Y_1 Y_2 Ye_data Exp (b) 0 10 20 30 40 50 60 0 400 800 1200 1600 2000 2400 0 20 40 60 80 100 120 140 Gas phase toluene concentration. Y ( g/L) Aqueous phase toluene concetration, C ( g/L) Packed depth, Z (cm) C_1 C_2 Ce_Dat a Y_1 Y_2 Ye_data Exp (c) 0 5 10 15 20 25 30 35 40 45 0 200 400 600 800 1000 1200 0 20 40 60 80 100 120 140 Gas phase toluene concentration. Y ( g/L) Aqueous phase toluene concetration, C ( g/L) Packed depth, Z (cm) C_1 C_2 Ce_Data Y_1 Y_2 Ye_data Exp (a) Figure 8 3. Comparison of concentration profiles obtained from the model using the correlation derived f or the aqueous phase (1) and the one fitted to the two -phase data (2) Except for the ones identified in the legend as individual points, the markers on the curves do not represent measured data, but were included only to facilitate the identification of t he curves. .
178 0 5 10 15 20 25 30 35 40 0 500 1000 1500 2000 2500 0 20 40 60 80 100 120 140 Gas phase toluene concentration. Y ( g/L) Aqueous phase toluene concetration, C ( g/L) Packed depth, Z (cm) C_1 C_3 C_Data Y_1 Y_3 Y_data Exp (e) 0 5 10 15 20 25 30 0 200 400 600 800 1000 1200 0 20 40 60 80 100 120 140 Gas phase toluene concentration. Y ( g/L) Aqueous phase toluene concetration, C ( g/L) Packed depth, Z (cm) C_1 C_2 Ce_Dat a Y_1 Y_2 Ye_data Exp (d) Figure 8 3. Continued
179 0 5 10 15 20 25 30 35 40 45 50 0 500 1000 1500 2000 0 20 40 60 80 100 120 140 Gas phase toluene concentration, Y ( g/L) Aqueous phase toluene concetration, C ( g/L) Packed depth, Z (cm) C_Dax C_2Dax C_10Dax Y_Dax Y_2Dax Y_10Dax Figure 8 4. Effects of dispersion coefficient on the aqueous and gas phases profiles obtained from the models using the Ks calculated from the correlations. The initial conditions are those for Experiment (b): QL = 3.8 L/min, QG = 142 L/min, Twater = 24.4oC. The markers on the graph do not represent measured data; they are only used to facilitate identifying the respective curves. a b c d e 0.10 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00 0 500 1000 1500 2000 2500 3000 Net Toluene Removal Aqueous phase toluene concentration, C o ( g/L) Data (1) K's from correlation (2) Kls and Kgl fitted Figure 8 5. Actual net toluene removal versus the net removal predicted by the two -phase model using different mass transfer coefficients.
180 0 10 20 30 40 50 60 0 200 400 600 800 1000 1200 1400 0 20 40 60 80 100 120 140 Gas phase toluene concentration. Y (mg/L) Aqueous phase toluene concetration, C ( g/L) Packed depth, Z (cm) C C if no PCO Y Y if not PCO Figure 8 6 Concentration profile for toluene in the TPOTpresent in the gas and aqueous phases in the presence of PCO (UV lamps on) and due to mass transfer only (UV lamps off). 0 0.2 0.4 0.6 0.8 1 0 20 40 60 80 100 120 140 % Toluene Conversion Packed Depth, Z (cm) Figure 8 7 Net tol uene removal in the TPOT as a function of packed bed depth
181 Figure 8 8 Complete treatment system showing the simulation results.
182 0 200 400 600 800 1000 1200 1400 1600 1800 2000 0 20 40 60 80 100 120 140 Aqueous phase toluene concetration, C ( g/L) Packed depth, Z (cm) Base 2QL 2Qg 2Cin 2Yin Tw=10oC Figure 8 9 Effect of different operating parameters on the aqueous phase profile concentrati ons in the TPOT. Base profile: QL=3.785L/min, QG=113.5L/min, Cin=1500mg/L, Yin=10mg/L, Tw = 24oC. 0 10 20 30 40 50 60 70 0 20 40 60 80 100 120 140 Gas phase toluene concentration, Y ( g/L) Packed depth, Z (cm) Base 2QL 2QG 2Cin 2Yin Tw=10oC Figure 8 10. Effect of different operating parameters on the gas phase profile concentrations in the TPOT. Base profile: QL=3.785L/min, QG=113.5L/min, Cin=1 500mg/L, Yin=10mg/L, Tw = 24oC.
183 0 0.2 0.4 0.6 0.8 1 0 20 40 60 80 100 120 140 Net Conversion of toluene Packed depth, Z (cm) Base 2QL 2QG 2Cin 2Yin Tw=10oC Figure 8 1 1 Effect of different operating parameters on the net toluene conversion in the TPOT as a function of packed bed depth. Base profile: QL=3.785L/min, QG=113.5L/min, Cin=1500mg/L, Yin=10mg/L, Tw = 24oC.
184 CHAPTE R 9 CONCLUSIONS The photocatalytic oxidation of toluene was investigated using a novel countercurrent flow reactor designed to enable the treatment of toluene present in the gas phase and the aqueous phase simultaneously. The main conclusions formulated fr om the experimental results were the following: 1 The accessible surface area of titanium dioxide in the STCs developed for the photocatalytic reactor compared to pure Degussa P25 TiO2 decr eased by about 20%. However, no significant differences were found be tween the area available in the STCs when used as a powder or as a large pellet (9mm x 8mm) of cylindrical shape. 2 Gas phase adsorption of toluene was large during both bench and pilot -scale experiments in the absence of a water phase coating the pellets. The adsorption capacity decreased remarkably as temperature increased 3 Gas phase toluene conversions greater than 90% w ere achieved in bench -scale experiments using the STCs only when they were dry and the re was sufficient water vapor concentration in the g as Under the operating conditions investigated, the reaction kinetics were not limited by external or internal resistances to mass transfer. The removal of toluene in the gas phase using dry STCs exhibited pseudo -first order reaction kine tics with rate co nstant of 0.12 s1. 4 Deactivation of the catalyst was observed during gas phase PCO under all conditions investigated at the bench-scale. This deactivation was more pronounced, however, when the relative humidity was low. Accompanying deactivation was the p resence of a yellowish coloration of the STCs. 5 During pilot -scale studies, the extent of PCO of toluene in the gas phase was found to be lower than during bench-scale studies. The decreased activity was attributed to the insufficient relative humidity du ring pilot -scale operations, the significant deactivation encounter ed in the pilot reactor, and the large increase in the lamps temperature. Deactivation was confirmed because the reactor was able to achieve greater toluene conversions after catalyst regeneration. The increase in lamp temperature can affect gas phase PCO in two ways: it increases reactor temperature thus negatively affecting the adsorption of toluene to the catalyst, and it affects the UVC output of the lamps which is temperature sensitive 6 PCO of toluene in the gas phase wa s completely inhibited by the wetting of the STCs, for batch, continuous flow and trickled flow operations.
185 7 Since w etting of the packing resulted in negligible PCO in the gas phase, likely due to the lack of adsorption of toluene to the catalyst, the reaction mechanism for PCO in the gas phase involves an adsorption step so the reactions do not occur at the bulk phase 8 Adsorption of toluene in the aqueous phase was determined at the bench-scale and found to be significant during batch experiments. However, it was negligible during pilot -scale operations at different liquid loading rates most likely due to the much shorter residence times (< 60 s) encountered in the pilot reactor. Adsorption occurs until equilibrium is achi eved, but it does not affect the concentration of the effluent because the removal for a single pass of fluid is small due to the small residence time. 9 PCO of toluene in the aqueous phase was found to linearly increase as a function of inlet concentration and decrease as function of liquid loading rate. Increasing the concentration increases the driving force for transfer between liquid and solid, thus improving toluene conversion. Increasing the loading rate results in opposing effects for PCO. On the one hand it increases the wetted surface area of the packing which should result in improved PCO; on the other hand it increases the thickness of the liquid film in the reactor which increases the resistance to m ass transfer between the liquid-solid interface and it also decreases the residence time of toluene in the reactor, both of which produce an adverse effect on PCO. 10. The mean residence time of toluene in the aqueous phase was determined using a step input tracer test. The mean residence time was greater for the TPOT packed with the commingled packing compared to the only high flow rings packing due to the lower porosity which indicates increased tortuosity and holdup in the system. For the commingled packing, the mean residence time decreased linearly wit h flowrate. 11. The dispersion coefficients obtained from the tracer test increased with liquid superfi ci al velocity, and they were correlated to the Reynolds and Galileo numbers 12. The reactor was modeled fo r the liquid phase only using a one dimensional plug fl ow with dispersion equation, which was numerically solved by the finite differences method. Using this model, overall liquid -solid mass transfer coefficients were obtained by minimizing the sum of squared differences between the model and the experimental results. KLSac was found to be a function of flowrate and inlet concentration, thus it was correlated to the Reynolds number and the influent toluene concentration. 13. The UV distribution in the reactor was improved by using the commingled packing compared to just STCs, as confirmed by the increase in UV irradiance penetration from 2.5 cm to about 6 cm for the STCs only packing and the commingled packing, respectively. 14. It was determined that about 30% of the reactor did not receive any UV irradiance. The aver age UV intensity in the reactor was found to be about 220 W/g of irradiated TiO2. Optimization of the lamp and packing placement in the reactor will provide more efficient use of the UV radiation generated and lower energy requirements.
186 15. Overall gas li quid mass transfer coefficients, which includes the specific surface area were found to be greater for the reactor packed with the commingled packing as compared to just STCs. Improved mass transfer coefficients were the result of the greater surface area avai lable for transfer provided by the commingled packing 16. A modified version of the Onda correlation was found to produce excellent predictions for the overall gas liquid mass transfer coefficients. The modified Onda correlation accoun ted by a greater wetted a rea tha n the one proposed by Onda et al., and the modification was performed by adjusting one of the coefficients of the equation for the wetted area provided by Onda et al. The modification decreased the effect of surface tension in the wetting process. 17. T he gas liquid mass transfer coefficients showed a moderate non-linear increase as a function of the air to -water ratio, and they increased linearly with liquid flowrate. However, KGLaw is expected to achieve a maximum as a function of liquid loading rate, since most of the improvement is due to the increase in wetted area, which will achieve a maximum as a function of the liquid flowrate 18. The two -phase experimental results indicated that compare d to the conversions of toluene found for the aqueous phase onl y, the two -phase system usually resulted in lower conversions for the same aqueous phase concentrations, suggesting that most of the removal in the two -phase system occurs at the liquid solid inter face 19. The removal of toluene in the two -phase countercurren t tower can be adversely affected by: a Decrease in the aqueous phase concentration decreases the driving force for aqueous phase PCO. b Increase in air to water ratio Increases stripping to the gas phase, thus decreasing the aqueous phase concentration. c D ecrease in gas phase concentration increases the gas phase capacity for toluene, which increases the stripping potential, thus decreasing the aqueous phase concentration. 20. The two -phase reactor was modeled using a plug flow with dispersion model for the a queous phase and a plug flow model for the gas phase. The modeling effort with adjust of mass transfer coefficients was successful. 21. The liquid -solid mass transfer coefficients found by the two -phase model were between 1.3 to 6 times greater than the coeffi cients found for the single aqueous phase model. The gas phase flowrate and concentration, which were not part of the correlations, seem to affect these coefficients, although a specific trend was not found. 22. The mathematical model is a useful tool to simul ate the reactors performance for a given set of operating conditions. It also allows investigating the effects of the different variables on the net toluene conversion.
187 Due to the stripping capacity of the TPOT, this system can achieve aqueous phase remov als that single phase photocatalytic systems would not be able to achieve given the residence time in the system. Other authors have ac hieved removals greater than 80 % but using reaction times greater than 40 min (Holmes et al., 2004; Cao et al., 2000). Fo r the average residence time of less than 60 seconds provided in the TPOT, the aqueous phase removal achieved is remarkable. The removals obtained were as high as or higher than those achieved by a stripping tower; however, due to the photocatalytic capaci ty of the system the gas phase concentration exiting the reactor is lower than for a stripping tower relying only on mass transfer. Therefore, if this reactor was used to treat VOCs in the aqueous phase, the treatment of the effluent gas in this system wou ld be cheaper than for a stripping tower using granular activated carbon (GAC) to treat the effluent gas due to the lower toluene gas loading that the GAC system would need to treat. Although the initial capital cost of the photocatalytic system might be g reater than the one for a conventional treatment system, such as a stripping tower treating the off gas using granular activated carbon (GAC), because of the UV lamps and catalyst that need to be incorporated in the stripping tower and in the end polishing sections, the long term cost is expected to be lower because of two reasons: (1) the catalyst is self regenerated which is not the case for activated carbon which needs to be replaced several times during the lifetime of the system, and (2) the TPOT sys tem completely destroys the contaminant, meaning that there are no disposal costs related to this systems, which is not the case for GAC treatment that requires the thermal oxidation of the carbon to completely destroy the contaminants Some recommendation s for future work include: Model validation by testing the tower with different VOCs. Potential coating of the conventional tower packing to make better usage of the catalyst mass in the reactor and improve even further the UV penetration.
188 The competitive effects of VOCs in a multi -component system employing the TPOT should be modeled since these systems will be more indicative of real world applications. Bench -scale studies to determine the reaction rate of toluene PCO in the aqueous phase for modeling of the polishing section Energy requirement comparisons with conventional treatment systems for VOCs. In depth analysis of the oxidation byproducts in the gas phase to better determine the reasons for catalyst deactivation.
189 APPENDIX A PRELIMINARY STUDI ES Studies on Systems Integrity System Losses The potential losses in the system due to piping, leaks or/and volatilization were assessed individually for the gas and aqueous phases using the TPOT reactor packed only with plastic high flow rings i.e. in the absence of the catalyst. Either the gas or aqueous influent was spiked with toluene and passed through the packed tower at the desired flowrate. The UV lamps were turned off during all these experiments while the inlet and outlet concentrations were mo nitored. The gas and aqueous phase results presented in Figures A 1 and A 2 respectively, show the average of at least two replicates of the same experiment and their corresponding standard deviations shown by the error bars. The results obtained from th e gas phase experiments (Figure A 1 ) show that it takes about 20 min for the gas phase concentrations to reach steady state under the given conditions. After steady state is achieved, there are not significant losses in the system since inlet and outlet co ncentrations are very similar. Similar to the gas phase, steady state is reached after about 20 min for the aqueous phase Most of the initial lower effluent concentrations might be attributed to the volatilization of toluene in the reactor. Once the towe r volume is saturated with toluene that transferred to the gas phase, no other losses seem to be present since all the influent toluene is accounted for at the effluent. Losses D ue to P hotolysis Possible losses of toluene due to photolysis were investigat ed. Because the UV lamps used in the reactor are non -ozone prod ucing, no oxidation of toluene wa s expected during exposure to UV radiation in the absence of the catalyst. However, photolysis experiments were still
190 performed to assure that the lamp glass fi ltered the correct wavelength and to confirm the absence of other extraneous factors that could be an issue in the presence of UV radiation. Each contaminant phase was analyzed individually, and during these experiments the UV lamps were kept on at all tim es. For the gas phase experiments, toluene was allowed to flow for 20 min in the dark to reach steady state. After that period elapsed, the UV lamps were turned on. One run for each phase was performed and the results are presented in Figures A 3 and A 4 The results obtained for the gas phase and aqueous phase toluene photolysis experiments are similar to the results previously shown for the other losses in the system. The normalized concentrations are very close to 1, showing no significant differences be tween the inlet and outlet concentrations for either phase. As expected, there is no evidence to show that toluene undergoes photolysis under the given experimental conditions. Stripping The stripping capacity of the tower was evaluated to compare its act ual performance to typical strippers. During these experiments, air and water were simultaneously flowing through the reactor, but only the aqueous phase was contaminated with toluene. The objective of this setup was to determine the removal of toluene due to mass transfer from the aqueous phase to the gas phase as a function of different air to water ratios. Furthermore, the mass transfer coefficients were determined and compared to predictions by the Onda. The influence of temperature changes on the trans fer of toluene when the UV lamps were on or off was also investigated. Three different runs of stripping experiments were performed. Runs 1 and 2 were performed using different air to water ratios to determine their influence in the mass transfer coefficie nts. Run 2 and 3, used the same air to water ratio, but the UV lamps were turned on during the third run to investigate the influence of temperature changes due to heat produced by the high energy lamps. The results for the 3 runs are shown in Figure A 5
191 The results for runs 1 and 2 indicate that changing the air to wat er ratio ( V/Q ) from 18.7 to 45 affects the toluene transfer from the aqueous to the gas phase. V/Q was altered by changing both, the water and the gas flowrate. Furthermore, the water temper ature was 1oC higher for run 1 than for run 2. As explained in the literature review, factors influencing the transfer of contaminants from the aqueous phase to the gas phase include the concentration gradient and the overall mass transfer coefficient. In turn, the mass transfer coefficient is influenced by the temperature and liquid loading rate ( Lm). Comparing runs 1 and 2, the main factor favoring mass transfer in the second run is the large increase in the air to water ratio. The factors opposing mass t ransfer are the decrease in temp erature and liquid loading rate, which result in a decreased mass transfer coefficient. The combination of all these factors resulted in a higher mass transfer in the second run as compared to the first one (89.6% versus 82. 5 % toluene transfer to the gas phase). The overall mass transfer coefficients were calculated to be 0.407 min1 for run 1 and 0.248 min1 for run 2. The coefficients pred icted the Onda correlation underestimated the actual coefficients. By using a modified Onda coefficient that predicts a larger wetted area, as described in Chapter 7, the predicted coefficients were closer to the measured KLaw. As expected, KLaw was smaller for run 2 because of the lower liquid loading rate and temperature. Temperature effects due to the UV lamps were assessed by comparing runs 2 and 3. T he water temperature for run 3 was about 5oC higher than for run 2. Increasing the temperature increases the removal efficiency of toluene from 89.6% (run 2) to 94.9% (run 3). Similarly, the overall mass transfer coefficient was increased to 0.327 min1 for run 3, and the predicted coefficient using the Onda correlatio n was also smaller for this run. Although determining effects of the heat produced by turning on the UV lamps under the same experimental conditions was the m ain goal of run 3, the increase in removal cannot be solely attributed to temperature because
1 92 the initial toluene concentration was higher in run 3 than in run 2, which will increase the driving force for the stripping proc ess; thus also contributing to the higher removal. Absorption Similar to the stripping experiments, absorption experiments were conducted in the presence of both streams flowing through the reactor to determine the tendency of gas phase toluene to transfe r to the aqueous phase if only the gas phase was contaminated with toluene. The transfer to the aqueous phase was monitored by sampling inf luent and effluent streams and t he se resu lts are presented in Figure A 6 The inlet and outlet concentrations of tolu ene in the gas phase in Figure A 5 seem ed very similar, so no significant absorption can be observed from the gas phase results. However, the effluent aqueous concentrations show that some toluene is actually transferring to the aqueous phase despite the h igh volatility of toluene. Although the aqueous phase concentrations might seem large compared to the gas phase concentrations, the gas phase flowrate is greater than the aqueous pha se flowrate, 37 times larger Compared to the stripping capacity of this c ompound, its absorption potential is very small, though it cannot be neglected.
193 0 0 5 1 1.5 2 2.5 3 3.5 0 10 20 30 40 50 Normalized Effluent Gas Phase Toluene Concentration (Ye/Yo) Run Time (min) Figure A 1 System losses in the gas phase due to potential leaks. Packing: plastic high flow rings only. Average influent gas phase concentration Yo = 2.3 ppmv for run 1 a nd 10.4 ppmv for run 2. Gas flowrate, QG = 5 cfm for all runs. 0 0.2 0 4 0.6 0.8 1 1.2 0 10 20 30 40 50 60 Normalized Effluent Aqueous Phase Toluene Concentration (Ce/Co) Run Time (min) Figure A 2 System losses in the aqueous phase due to volatilization and/or leaks. Packing: plastic high flow rings Co = 1.6 mg/L (run 1), 1.7 mg/ L (run 2), and 1.9 mg/L (run 3). QL = 3.8 L /min for all runs.
194 0 0 2 0 4 0.6 0 8 1 1 2 0 10 20 30 40 50 60 Normalized Effluent Gas Phase Toluene Concentration (Ye/Yo) Run Time (min) Figure A 3. Photolysis of toluene in the gas phase. Yo = 10.8 ppmv. QG = 5 cfm. 0 0 2 0.4 0.6 0 8 1 0 10 20 30 40 50 60 Normalized Effluent Aqueous Phase Toluene Concentration (Ce/Co) Run Time (min) Figure A 4 Photolysis of toluene in the aqueous phase. Co = 1. 6 mg/L (run 1), QL = 3.8 L/min.
195 0 200 400 600 800 1000 1200 1400 1600 1800 2000 5 10 20 Aqueous Phase Toluene Concentration ( g/L) Run Time (min) Inlet (Co) Outlet (Ce) V = 5 cfm Q = 2gpm V/Q = 18.7 T = 23oC KLa (min1) = 0.407 (measured) 0.335 (Onda) 0.398 (Onda Modified)(A) 0 200 400 600 800 1000 1200 1400 1600 1800 5 10 20 30 40 50 Aqueous Phase Toluene Concentration ( g/L) Run Time (min) Inlet (Co) Outlet (Ce) V = 6 scfm Q = 1gpm V/Q = 45 T = 22oC KLa (min1) = 0.248 (measured) 0.204 (Onda) 0.242 (Onda Modified)(B) 0 500 1000 1500 2000 2500 3000 3500 10 20 30 40 50 Aqueous Phase Toluene Concentration ( g/L) Run Time (min) Inlet (Co) Outlet (Ce) V = 6 scfm Q = 1gpm V/Q = 45 KLa (min1) = 0.327 (measured) 0.223 (Onda) 0.265 (Onda Modified)(C) Figure A 5. T oluene stripping in the TPOT packed with hi gh flow rings only using the UV lamps off and on and having the air flow free of contaminant (A) Run 1. (B) Run 2. (C) Run 3. UV lamps off for (A) and (B). UV lamps on for (C)
196 0 5 10 15 20 25 5 15 20 30 40 50 Time (min) Gas Phase concentration (ug/L) 0 10 20 30 40 50 60 Aqueous phase conc. (ug/L) Gas Inlet Gas Outlet Water outlet Figure A 6. Toluene absorption in the TPOT packed with plastic high flow rin gs only and the water flow free of contamiants QL = 3.8 L/min QG = 5 cfm. QG/QL = 37.4. UV lamps off
197 APPENDIX B ASSESSMENT OF LEACHI NG OF NANOMATERIALS A fairly new concern about using heterogeneous photocatalysis with titanium dioxide for large scal e applications, specially for applications that will result on the discharge of the effluents to aquatic environments or for drinking water purposes is the potential leaching of the nanomaterials during the systems operation. This is a concern since recen t research studies have shown that even low concentrations of these materials can cause adverse effects on the health of living organisms. For example, Lovern and Kapler (2006) investigated the effects of nanosize TiO2 (30 nm average diameter) on Daphnia m agna, a common filter feeder. They found that the lowest TiO2 concentration causing an observable effect was 2.0 mg/L while a concentration of 5.5 mg/L caused 50% mortality. The mortality rate continued to increase with increasing titanium dioxide concentr ation reaching 100% mortality at 10 mg/L. So far, a concentration of titanium dioxide that causes mortality or severe irreversible adverse health effects has not been determined for either humans or the ecosystem, but many studies suggest that concentratio ns above 2 mg/L might start producing some observable adverse health effects. Accordingly, the effluents from the TPOT were measured for TiO2 and SiO2 concentrations. The effects of the UV radiation on the leaching of silica and titania were investigated by measuring effluents from adsorption and PCO experiment (Figure 9 1). The effects of flowrate on the leaching of materials was also investigated and presented in Figure 9 2. Finally the effects observed with time after multiple runs were also assessed by sampling effluents from many runs performed on different dates (Figure 9 3). In general, all the results showed no titania present at the influent stream whereas some titania was measured at the effluent. However, all the concentrations of titania detecte d were in the ppb range. No trends were observed for the effects of UV radiation and flowrate on the leaching of titania. For the case
198 of silica, the concentrations measured were much higher, in the ppm range, but these values were just barely higher for t he effluent samples compared to the influent samples, meaning that most of the silica measured was already present in the tap water used for the experiments. Furthermore, the leaching of both materials was relatively steady as a function of time, so it was not affected by the number of runs performed in the reactor. These results indicate that although some titania was observed at the effluent of the TPOT, the concentrations were well below those expected to produce adverse health effects on humans or the e nvironment. 1.6 1.8 0.0 1.1 2.9 3.5 3.6 3.7 4.0 4.4 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 30 150 210 Ti ( g/L), Si (mg/L) Run Time (min) Adsorption (2.14.09, 1gpm) Ti In Ti OUT Si IN Si OUT 0.0 0.1 0.0 1.5 3.6 4.0 3.8 3.7 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 30 240 Ti ( g/L), Si (mg/L) Run Time (min) PCO (3.19.09, 1.5gpm) Ti IN Ti OUT Si IN Si OUT Figure B 1. Effects of the UV radiation on the leaching of silica and titania from the TPOT packing.
199 Figure B 2. Effect of the flowrate on the leaching of silica and titania from the packing on the TPOT. 0 0.1 0 0 0 2.9 1.5 3.7 5.5 2.0 3.7 4.0 3.4 3.2 3.2 4.4 3.7 4.1 3.9 3.7 0.0 1.0 2.0 3.0 4.0 5.0 6.0 Ti ( g/L), Si (mg/L) Date Ti IN Ti OUT Si IN Si OUT Figu re B 3. Leaching of silica and titania from the packing in the TPOT for several experiments performed at various dates.
200 LIST OF REFERENCES Alberci, R M ., Jardim, W F. 1997. Photocatalytic destruction of VOCs in the gas phase using titanium dioxide. Appl. Catal B: Environ 14, 55 68. Ameen, M M ., R aupp, G B. 1999. Reversible catalyst deactivation in the photocatalytic oxidation of dilute o -xylene in air. J. Catal. 184, 112 122. Andino, J M ., Butler, J W ., 1991. A Study of the Stability of Methanol -Fueled Vehicle Emissions in Tedlar Bags .Environ Sci Technol. 25, 16441646. Anpo, M., Chiba, K ., Tomonari, M ., Coluccia, S., Che, M ., Fox, M A. 1991. Photocatalysis on native and platinum loaded TiO2 and ZnO catalysts: Origin of different reactivities on wet and dry metal o xides. Bull. Chem. Soc. Jpn. 64, 543551. Arai, J., Tanaka, K ., and Khlaifat, A L. 2006. Photocatalysis of SiO2loaded TiO2. J. Mol Catal A: Chem 243, 85 88. Armor, J N. 1992. Environmental Catalysis. Appl Catal B: Environ 1, 221256. Augugliario, V., Coluccia, S ., Loddo, V ., Marchese, L ., Martra, G ., Palmisano, L., Schiavello, M., 1990. Photocatalytic oxidation of gaseous toluene on anatase TiO2 catalyst: mechanistic aspect s and FT IR investigation. Appl. Catal B: Environ. 20 15 2 7. Barrett, E. P., Joyner, L. G., Halenda, P. P., 1951. The determination of pore volume and area d istributio ns in porous s ub stances. I. Computations from nitrogen i sotherms. J. Amer Chem Soc 73, 373380. Belhekar, A A ., Awate, S V ., Anand, R. 2002. P hotocatalytic activity of titania modified mesoporous silica for pollution control. Ca tal. Commun. 3, 453458. Blount, M C., Kim, D H ., Falconer, J L ., 2001. Transparent thin -film TiO2 photocatalysts with high acti vity. Environ. Sci. Technol. 35, 29882994. Blount, M C., Falconer, J L. 2002. Steady -state surface species during toluene photocatalysis. Appl Catal B: Enviro n. 39, 39 50. Boonamnuayvitaya, V ., Tayamanon, C ., Sae ung, S ., Tanthapanichakoon, W. 2006. Synthesis and characterization of porous me dia produced by a sol -gel method. Chem Eng. Sci. 6 1, 16861691. Boulamanti, A. K., Korologos, C. A., Philippopoulos, C. J., 2008. The rate of photocatalytic oxidation of aromatic volatile organic compounds in the gas phase. Atmos. Environ. 42, 78447850. Bouzaza, A ., Laplanche, A. 2002. Photocatalytic degradation of toluene in the gas phase: comparative study of some TiO2 supports. J. Photochem Photobiol. A: Chem. 150, 207212.
201 Bouzaza, A., Vallet, C ., Laplanche, A., 2006. Photocatalytic degradation of s ome VOCs in the gas phase using an annular flow reactor. Determination of the contribution of mass transfer and chemical reaction steps in the photodegradation process. J. Photochem. Photobiol A: Chem 177, 212217. Byrne, H E Kostedt IV, W L Stokke, J M ., Mazyck, D W. 2009. Characterization of HFcatalyzed silica gels doped with Degussa P25 Titanium Dioxide. J. Non-Cryst Solids. In Press Cao, L., Gao, Z., Suib, S .L., Obee, T .N., Hay, S O ., Freihaut, J D ., 2000. Photocatalytic o xidation of t oluene on n anoscale TiO2 c atalysts: s tudies of d eactivation and regeneration. J. Catal 196, 253 261. Chapra, S.C., Canale, R.P., 2002. Numerical methods for engineers. 4th Ed. McGraw -Hill, New York, NY. Cortese, A D. 1990. Clearing the air. Enviro n. Sci Technol 24, 442448. Crittenden J C., Liu, J ., Hand, D W ., Perram, D L ., 1997. Photocatalytic oxidation of chlorinated hydrocarbons in water. Wat Res .. 31, 429438. de Andrade Lima, L. R. P., 2006. Liquid axial dispersion and holdup in column leaching. Minerals Eng. 19, 37 47. DOliveira, J C., Ai -Sayyed, G ., Pichat, P. 1990. Photodegradation of 2 and 3 -chlorophenol in TiO2 aqueous suspensions. Envir on. Sci & Technol 24, 990996. Degussa Technical Bulletin No. 56, 1990. Highly dispersed metallic oxides produc ed by the AEROSIL process. Einaga, H ., Futamura, S ., Ibusuki,T. 2001 Heterogeneous photocatalytic oxidation of benzene, toluene, cyclohexene and cyclohexane in humidified air: comparison of decomposition behavior on photoirradiated TiO2 catalyst. Appl. Catal B: Environ 38, 215-225. Emeline, A V ., Rudakova, A V ., Ryabchuk, V K ., Serpone, N 1998. Photostimulated reactions at the surface of wide band-gap metal oxides (ZrO2 and TiO2): Interdependence of rates of reactions on pressure -concentration and on light intensity. J. Phys Chem B. 102, 109061 09116. Fan, J ., Yates, J T ., 1996. Mechanism of photooxidation of trichloroethylene on TiO2: detection of intermediates by infrared spec troscopy. J. Am. Chem. Soc. 118, 46864692. Fogler, H. S. 1999. Eleme nts of chemical reaction engineering; Prentice Hall PRT: Upper Saddle River, New Jersey. Fox, M A ., Dulay, M T. 1993. Heterogeneous p hotocatalysis. Chem Rev 93 341357.
202 Gould, G ., Irene, E A. ,1988. An in situ s tudy of a queous HF t reatment of s ilicon by c ontact a ngle m easurement and e llipsometry J. Electrochem. Soc. 135 15351542. Hager, S ., Bauer, R. 1999. Heterogeneous photocatalytic oxidation of organics for air purification by near UV irradiated ti tanium dioxide. Chemosphere. 38, 1549 1559. Hand, D W Crittenden, J C. Gehin, J L ., Lykins, B W ., 1986. Design and e valuation of an a ir s tripping t ower for r emoving VOCs from groundwater. J. AWWA. 78, 8797 Hand D W Hokanson, D R., Crittenden, J C. 1999. Air s tripping and aeration, in: Letterman, R.D. (Ed), Water Quality and Treatment : A Handbook of Community Water Supplies. AWWA and McGrawHill, New York pp. 5.1 5.68. Hench, L L ., West, J K. 1990. The sol gel process. Chem Rev. 90, 3372. Hennenzel, O., Pichat, P ., Ollis, D F. 1998. Benzene and toluene gas -phase photocatalytic degradation over H2O and HCl pretreated TiO2: by -products and mechanisms. J. Photochem Photobiol. A: Chem 118, 197204. Herrmann, J M ., 2005. Heterogeneous photocatalysis: State of the art and present applications. To pics Catal 34 49 65. Hoffmann, M R., Martin, S T ., Choi, W ., Bahnemann, D W. 1995. Environmental applications of semiconductor photocatalysis. Chem Rev. 95, 6996. Holmes, F R. 2003. The performance of a reactor using photocatalysis to degrade a mixture of organic contaminants in aqueous solution. Masters Thesis, University of Florida. Holmes, F R ., Chadik, P A ., Mazyck, D W ., Wu, C Y ., Garton, M J ., Powers, K. W ., Londeree, D J. 2004. Photocatalytic Oxidation of Selected Organic Contaminants in a Cont inuous Flow Reactor Packed With Titania Doped Silica. International Conference on Environmental Systems. Colorado Springs, CO, USA. Hopper J R. Saleh J M Pike R. 2001. Design of multi -phase and catalytic chemical reactors: a s imulation tool for pollution prevention. Clean Prod Proc 3 92 103 Japar, S M ; Wallington, T J ., Rudy, S J ., C hang, T Y. 1991. Ozone -forming p otential of a series of oxygenated organic c ompounds. Environ Sci. Tech 25 415420. Kang, M ., Hong, W J ., Park, M S ., 2004. Synthesis of high concentration titanium incorporated nanoporous silicates (Ti NPS) and their photocatalytic performance for toluene oxidation. Appl Catal B: Enviro 53, 195 205. Kavanaugh, M C., Trussell, R R. 1980. Design of a eration t owers to s trip v olatile o rganic c ontaminants from d rinking water. J. AWWA. 72, 12:684 Kim, S. B., Hong, S. C., 2002. Kinetic study for photocatalytic degradation of volatile organic compounds in air using thin film TiO2 photocatalyst. Appl Catal. B: Environ 35, 305315.
203 Kobayakawa, K ., Sato, C., Sato Y., Fujishima, A., 1998. Continuous -flow photoreactor packed with titanium dioxide immobilized on large silica gel beads to decompose oxalic acid in excess water. J. Photochem Photobiol A: Chem. 118, 65 69. Kresge, C T ., Leonowicz, M E ., Roth, W J ., Vartuli J C., Beck J S. 1992. Ordered mesoporous molecular sieves synthesized by a liquid -crystal template mechanism. Nature 359 (1992) 710712. Lamarche, P ., Droste, R L ., 1989. Air -s tripping m ass t ransfer c orrelations for v olatile o rganics. J. AWWA. January, 7889 Larson, S.A., Falconer, J.L. 1997. Initial reaction steps in photocatalytic oxidation of aromatics. Catal Letters. 44, 57 65. Legrand, P., 1998. O n the silica edge. In: Th e Surface Properties of Silicas. John Wil ey & Sons, Inc., New York, 130200. Lewandowski, M., Ollis, D.F., 2003a. A Two-Site kinetic model simulating apparent deactivation during photocatalytic oxidation of aromatics on titanium dioxide (TiO2). Appl. Catal. B: Environ 43, 309327. Lewandowski, M., Ollis, D.F., 2003b. Extension of a TwoSite transient kinetic model of TiO2 deactivation during photocatalytic oxidation of aromatics: concentration variations and catalyst regeneration studies. Appl. Catal. B: Environ 45, 223 238. Lichtin, N N ., Avudaithai, M. 1996. TiO2-Photocatalyzed oxidative degradation of CH3CN, CH3OH, C 3 HCl, and CH2Cl2 supplied as vapors and in aqueous solution under similar conditions. Environ. l Sci Techno l. 30, 20142020. Linsebigler, A L Lu, G., Yates, J T ., 1995. Photocat alysis on TiO2 surfaces: Principles, mechanisms and selected results. Chem Rev. 95, 735 758. Londeree, DJ. 2002. Silica titania composites for water treatment. Master thesis. University of Florida. Lovern, S B., Klaper, R. 2006. Daphnia magna mortality when exposed to titanium dioxide and fullerene (C60) nanoparticles. Environ. Toxicol. Chem. 25, 11321137. Lou, Y ., Ollis, D F. 1996. Heterogeneous photocatalytic oxidation of trichloroethylene and toluene mixtures in air: Kinetic promotion and inhibition, time -dependent catalyst activity. J. Catal 163, 1 11. Ludwing, C Y. 2004. The performance of silica -titania composites in a packed -bed reactor for the photocatalytic destruction of gray water. Master thesis. University of Florida.. Maira, A J ., Yeung, K .L., Soria, J ., Coronado, J M ., Belver, C ., Lee, C Y ., Augugliaro, V ., 2001a Gas -phase photo -oxidation of toluene using nanometer -size TiO2 catalysts. Appl Catal B: Environ 29, 327336.
204 Maira, A J ., Coronado,J M ., Augugliaro,V ., Yeung, K L ., Conesa, J C., Soria, J. 2001b. F ourier transform infrared study of the performance of nanostructured TiO2 particles for the photocatalytic oxidation of gaseous toluene. J. of Catal 202, 413 420. Marc i G ., Addamo, M ., Augugliaro, V ., Coluccia, S ., -Lpez, E ., Loddo, V ., Martra, G ., Palmisano, L ., Schiavello, M. 2003. Photocatalytic oxidation of toluene on irradiated TiO2: comparison of degradation performance in humidified air, in water and in water containing a zwitterionic surfactant. J. Photochem and Ph otobiol. A: Chem 160, 105 114. Marugn, J. Lpez -Muoz, M. J. Aguado, J. van Grieken, R., 2007. On the comparison of photocatalysts activity: A novel procedure for the measurement of titania surface in TiO2/SiO2 materials. Catal Today. 124, 103109. M artra, G ., Coluccia, S ., Marchese, L ., Augugliaro, V ., Loddo, V ., Palmisano, L ., Schiavello, M. 1999. The role of H2O in the photocatalytic oxidation of toluene in vapour phase on anatase TiO2 catalyst: A FTIR study. Catal. Today 53, 695 702. McCabe, L M, Smith, J C. Harriott, P. 2005. Unit o perations of c hemical e ngineering. 7th Ed. Chem. Eng. Series. New York: McGraw Hill. McGarvey, L J ., Shorten, C V. 2000. The effects of adsorption on the reusability of tedlar air sampling bags. AIHAJ 61, 375 380. Mendez -Roman, R ., Cardona Martinez, N. 1998. Relationship between the formation of surface species and catalyst deactivation during the gas -phase photocatalytic oxidation of toluene. Catal Today. 40, 353365. Michell R.W., Furzer, I.A. 1972. Mixing in trickle flow through packed beds, Chem. Eng. J. 4 53 63 Miyabe, K ., Guiochon, G. 2004. Comparison of the characteristics of adsorption equilibrium and surface diffusion in liquid -solid and gas -solid adsorption on C18 -silica gels. J. Phys Chem B. 108, 29872997. Mo, J. Zhang, Y. Xu, Q. Yang, R., 2009. Effect of TiO2/adsorbent hybrid photocatalysts for toluene decomposition in gas phase. J. Hazardous Mat. In Press, Corrected Proof. Mutin, P. H., Lafond, V., Popa, A. F., Granier, M., Markey, L., Dere ux, A., 2004. Selective surface modification of SiO2TiO2 supports with phosphonic acids. Chem. Materials. 16, 56705675. Nirmalakhandan, N Lee, Y H ., Speece, R E. 1987. Designing a c ost -e fficient a ir -s tripping process. J. AWWA. 79, 5662. Obee, T N ., Brown, R T. 1995. TiO2 Photocatalysis for indoor air applications: Effects of humidity and trace contaminant levels on the oxidation rates of formaldehyde, toluene, and 1 ,3 -butadiene. Environ Sci Technol 29, 12231231.
205 Office of Environmental Health Ha zard Assessment ( OEHHA ) 2002. Air t oxics h ot s pots p rogram r isk a ssessment g uidelines. Part II. Technical Support Document for Describing A vailable Cancer Potency Factors. Ollis, D F ., Pelizzetti, E ., Serpone, N. 1991. Destruction of water contaminants. Environ Sci Technol. 25, 15231529. Ollis, D F. 2005a. Kinetic disguises in heterogeneous photocatalysis Topics Catal. 35, 217223. Ollis, D F. 2005b. Kinetics of liquid phase photocatalyzed reactions: An illuminating approach. J. Phys Chem B. 109, 24392444. Onda, K Takeuchi, H Okumoto, Y. 1968. Mass t ransfer c oefficients b etween g as and l iquid p hases in p acked columns. J. Chem. Eng. J pn. 1, 56 62. Otake, T., Kunugita, E. 1958. Mixing characteris tics of irrigated packed towers. Kagaku Kogaku 22, 144 150. Pengyi, Z ., Fuyan, L ., Gang, Y ., Qing, C ., Wanpeng, Z. 2003. A comparative study on decomposition of gaseous toluene by O3/UV, TiO2/UV and O3/TiO2/UV. J. Photochem Photobiol A: Chem 156, 189194. Peral, J ., and Ollis, D F. 1992. Heterogen eous photocatalytic oxidation of gas -phase organics for air purification acetone, 1 -butanol, butyraldehyde, formaldehyde, and meta -xylene oxidation. J. Catal 136, 554565. Philips Lighting, 2004. UV Disinfection Application Information. Document No. 3222 635 43401. Printed in The Netherlands. Pich, S. Larachi, F., Iliuta, I. Grandjean, B.P.A. 2002. Improving predictions of liquid back mixing in trickle -bed reactors using a neural network approach, J Chem Technol Biotechnol. 77, 989998. Pintar, A. Bercic, G. Levec, J., 1997. Catalytic liquid -phase oxidation of aqueous phenol solutions in a trickle -bed reactor. Chem Eng Sci 52, 41434153. Pitoniak, E., Wu, C.Y., Londeree, D.J., Mazyck, D.W., Bonzongo, J. D., Powers, K.W. Sigmund, W., 2003. N anostructured s ilica -g el d oped with TiO2 for Hg v apor control. J. Nanoparticle Res. 5, 282 292. Powers, K., 1998. The development and characterization of sol gel substrates for chemical and optical applications, Ph.D. Dissertation, University of Florida Ray, A K. 1999. Design, modeling and experimentation of a new large -scale photocatalytic reactor for water treatment. Chem Eng S ci. 54, 31133125.
206 Roberts, P V ., Hopkins, G D ., Munz, C and Riojas, A H. 1985. Evaluating t wo resistance m odels for a ir s tripping of v olatile o rganic compounds in a c ountercurrent, p acked c olumn. Environ Sci Technol. 19 164173. Sater V.E, Levenspiel, O. 1966. Two-phase flow in packed beds. Evaluation of axial dispersion and holdup by moment analysis. I&EC Fundam ent als 5 86 92 Satterfield, C N. 1969. Mass transfer in heterogeneous photocatalysis. M.I.T. Press. Cambridge, MA Sauer, M L Hale M A Ollis, D F. 1995. Heterogeneous photocatalytic oxidation of dilute toluene -chlorocarbon mixtures in air. J. Photoche m and Photobiol A: Chem. 88, 169178. Sauer, M L ., Ollis, D F. 1996. Catalyst deactivation in gas -solid photocatalysis. J. Catal 163, 215217. Serpone, S ., Emelie, V. 2002. Suggested terms and definitions in photocatalysis and radiocatalysis. Int. J. Photoenergy. 4, 91131. Sherwood, T K Shipley, G H ., Holloway, F A L. 1938. Flooding v elocities in p acked c olumns. Ind. Eng. Chem. 30, 765769. Sing, K.S.W. Everett, D.H. Haul, R.A.W. Moscou, L. Pierotti, R.A. Rouquerol, J. Siemieniewska, T. 1985. Reporting physisorption data for gas/solid systems with special reference to the determination of surface area and porosity. Pure Appl Chem 57, 603619. Skelland, A. H. P., 2006. Diffusional mass transfer. Mala bar, Fla: R.E. Krieger Pub. Co. Stokke, J M ., Mazyck, D W ., Wu, C Y ., Sheahan, R. 2006. Photocatalytic oxidation of methanol using silica titania composites in a packed bed reactor. Environ. Prog 25, 312318. Stokke, J M 2008. Photocatalytic oxidation of hazardous air pollutants using silica titania com posites in a packed bed reactor. Ph.D. Dissertation, University of Florida. Stokke, J. M., Mazyck, D. W., 2008. Photocatalytic d egradation of m ethanol u sing s ilica -t itania c omposite p ellets: e ffect of p ore s ize on m ass t ransfer and r eaction k in etics. Environ Sci Technol 42, 38083813. Tanaka, K ., Capule, M ., Hisanaga, T. 1991. Effect of crystallinity of TiO2 on its photocatalytic action. Chem. Phys. Lett. 187, 7376. Three Bond Co.,Ltd., 2004. Titanium Oxide Photocatalyst Thre e Bond Technic al News No. 62 Tokyo, Japan. Treybal, R.E., 1980. Mass Transfer Operations, 3rd Ed., Chem. Eng. Series. New York: McGraw -Hill
207 Turchi, C S ., Ollis, D F. 1990. Photocatalytic degradation of organic water contaminants: m echanisms involving hydroxyl radical attack. J. Catal. 122 178192. United States Environmental Protection Agency (USEPA), 1995. Method 524.2 Measurement of purgeable organic compounds in water by capillary column gas chromatography/ mass sp ectrometry. Edited by JW Munch. United States En vironmental Protection Agency (USEPA), 2007a. An introduction to indoor air quality. Retrieved February 10, 2007, from http://www.epa.gov/iaq/voc.html#Health%20Effects United States Environ mental Protection Agency (USEPA), 2007b Drinking water contaminants: National prim ary drinking water regulations. Retrieved February 12, 2007, from http://www.epa.gov/safewater/co ntaminants/index.html#organic United States Environmental Protection Agency (USEPA), 2007c. Clean Air Act Retrieved February 20, 2007, from http://www.epa.gov/air/caa/ Ultra Violet Products (UVP) Inc. Gui de to Optical Radiation Measurements. Application Note AN 1167. Wang, J H ., Ray, M B. 2000. Application of ultraviolet photooxidation to remove organic pollutants in the gas phase. Separ Pur. Technol. 19, 1120. Web er, W. J. Jr., DiGiano F.A., 1996. Process d ynamics in e nvironmental s ystems. W iley Interscience, New York, NY. Xu, Y ., Langford, C H. 2000. Variation of Langmuir adsorption constant determined for TiO2photocatalyzed degradation of acetophenone under different light intensity. J. Photochem. and Photobiol A: Chem 133, 67 71 Yue, B. Zhou, Y. Xu, J. Wu, Z. Zhang, X. Zou, Y. Jin, S., 2002. Photocatalytic degradation of aqueous 4-chlorophenol by silic a immobilized polyoxometalates. Environ Sci. Technol 36, 132513. Zhang, R B. 2005. Photodegradation of toluene using silica -embedded titania. J. Non Crys. Solids. 351, 21292132. Zou, L ., Luo, Y ., Hooper, M ., Hu, E. 2006. Removal of VOCs by photocatal ysis process using adsorption enhanced TiO2-SiO2 catalyst. Chem. Eng Proc. 45, 959964.
208 BIOGRAPHICAL SKETCH Christina Akly was born in Santa Cruz, Bolivia to Bethsy Flores and Salomon Akly. She gr ew up in her country of origin and came to the United Sta tes in 2001 to pursue her undergraduate degree in environmental engineering s ciences at the University of Florida, where she graduated sum cum laude in May 2005. After completing her degree, she was offered a fellowship to continue her educa tion in UF, and she received a non -thesis m aster s in 2007 and a doc torate in August 2009, both in e nvironmental e ngineering with a focus in potable water treatment using advanced oxidation technologies. Her doctorate degree was performed under the guidance of her adviso r Dr. Paul A. Chadik.