Part I. Destruction of environmental pollutants by microwave heating and discharge plasmas. Part II. Generation of hydrogen from water and methane using discharge plasmasкод для вставкиСкачать
UMI MICROFILMED 2003 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. INFORMATION TO USERS This manuscript has been reproduced from the microfilm master. UMI films the text directly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter face, while others may be from any type of computer printer. The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print colored or poor quality illustrations and photographs, print bleedthrough, substandard margins, and improper alignment can adversely affect reproduction. In the unlikely event that the author did not send UMI a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion. Oversize materials (e.g., maps, drawings, charts) are reproduced by sectioning the original, beginning at the upper left-hand comer and continuing from left to right in equal sections with small overlaps. ProQuest Information and Learning 300 North Zeeb Road. Ann Arbor, Ml 48106-1346 USA 800-521-0600 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Part I. DESTRUCTION OF ENVIRONMENTAL POLLUTANTS BY MICROWAVE HEATING AND DISCHARGE PLASMAS Part II. GENERATION OF HYDROGEN FROM WATER AND METHANE USING DISCHARGE PLASMAS Franz-Josef Spiess, Ph.D. University o f Connecticut, 2003 Concerns about environmental pollution have caused new regulations to be implemented on a global level. Freons were shown as the principal cause o f the depletion o f the stratospheric ozone layer, and global action was set forth to reduce their emissions by the 1987 Montreal Protocol. Effective, environmental friendly decomposition o f these pollutants is sought. In this work, the efficient destruction o f two Freon species, Freon 21 and Freon 142B, using silent discharge plasmas was demonstrated. Gas chromatography was used to follow their decomposition in mixtures with oxygen, nitrogen, and water. Optical emission spectroscopy and mass spectrometry were employed to gain insight into the reaction mechanism. Global warming caused by greenhouse gases such as carbon tetrafluoride is a further environmental challenge that mankind faces. The breakdown of this pollutant was achieved by gas discharges but mainly by microwave heating in this study. Activated carbon was used as a catalyst in the presence of water to decompose a mixture o f nitrogen Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Franz-Josef Spiess—University of Connecticut, 2003 and carbon tetrafluoride. The effect of power on the conversion and catalyst deactivation were investigated. A possible reaction scheme was based on mass spectrometric results from decomposition studies. Fuel cells are considered the future of transportation. In order for the fuel cells to become more competitive, the cost of hydrogen production has to become much lower than in processes used today such as steam reforming o f methane. In this work, the hydrogen production from a mixture of water and nitrogen and a mixture o f water and methane was examined using discharge plasmas. The influence o f the reactor type and of other parameters such as input voltage and admixtures of nitrogen or argon was investigated. A reaction mechanism was proposed based on the mass spectrometric data. In a related study, the metal effect of the inner electrode used in the decomposition of methane and the resulting evolution of hydrogen by discharge plasmas was investigated. The effect of the flow rate and the cleaning of the electrode were examined as further objectives o f this study. A mechanism was developed using mass spectrometry and combined gas chromatography and mass spectrometry. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Part I. DESTRUCTION OF ENVIRONMENTAL POLLUTANTS BY MICROWAVE HEATING AND DISCHARGE PLASMAS Part II. GENERATION OF HYDROGEN FROM WATER AND METHANE USING DISCHARGE PLASMAS Franz-Josef Spiess Vordiplom, Universitat Konstanz, Germany, 1994 A Dissertation Submitted in Partial Fulfillment o f the Requirements for the Degree o f Doctor of Philosophy at the University o f Connecticut 2003 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UMI Number 3080929 Copyright 2003 by Spiess, Franz-Josef All rights reserved. UMI* UMI Microform 3080929 Copyright 2003 by ProQuest Information and Learning Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code. ProQuest Information and Learning Company 300 North Zeeb Road P.O. Box 1346 Ann Arbor, Ml 48106-1346 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Copyright by Franz-Josef Spiess 2003 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. APPROVAL PAGE Doctor of Philosophy Dissertation Part I. DESTRUCTION OF ENVIRONMENTAL POLLUTANTS BY MICROWAVE HEATING AND DISCHARGE PLASMAS Part II. GENERATION OF HYDROGEN FROM WATER AND METHANE USING DISCHARGE PLASMAS Presented by Franz-Josef Spiess, Vordiplom ^ Major Advisor jJmjX r Steven L. Suib, Ph.D. Associate Advisor a a a Associate Advisor - ■ ,r __ JohiTTanaka, Ph.D. ^ ' James D. Stuart, Ph.D. University o f Connecticut 2003 u Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Dedicated to my wife Daniela iii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ACKNOWLEDGEMENTS I would like to express my deepest and sincerest gratitude to my major advisor, Dr. Steven L. Suib, for his continued support, patience, guidance, and encouragement throughout my graduate career. I am especially grateful to my associate advisor Dr. John Tanaka for his advice, assistance and guidance throughout the years. Also, I would like to thank my other associate advisor, Dr. James D. Stuart, for his assistance and guidance. I wish to express my gratitude to Dr. Francis S. Galasso for all his encouragement, assistance, support, guidance and advice. A Ph.D. degree is similar to building a house. A lot o f people were involved in helping me build it. I am grateful for having these people and would like to thank them. Especially technical assistance is fundamental. Therefore, I would like to thank Dr. William Willis, Mr. Mark Marrotte, Mr. David Osier, and Mrs. Debra Avery. And I would like to express my gratitude to all the people doing the important little things and thank in this context Mrs. Janice Nuhn, Mrs. Dianne Tillman, Mrs. Nancy Coogan, Mrs. Charlene Fuller, and Mrs. Sally Chappell. I wish to give special thanks to all my co-workers over the years, especially Dr, Stephanie Brock, Dr. Scott Segal, Dr. Guangunag Xia, Dr. Aimin Huang, Dr. Lixin Cao, Dr. Xiao Chen, Dr. Jie Chen, Mr. Young-Chan Son, Mr. Jeffrey Rozak, and Ms. Theresa Hugener. I also thank all the member o f the Suib research group for their help and support over the years, especially Mr. Javier Garces, Mr. Jun Cai, Ms. Maggie Gulbinska, Mr. Daniel Conde, Ms. Laura Espinal, Mr. Vinit Makwana, and Mrs. Beatriz Hincapie. I would further thank Yuji Hayashi and Kanji Irie for all their help, advice, cooperation and support in the plasma reactions. iv Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. I am very thankful and grateful for all the friendships formed over my graduate career at UConn, which made life easier and more entertaining over the years. I am especially grateful to have formed friendships with Dr, Michael Miller, Dr. Scott Segal, Mr. Jeffrey Rozak, Dr. Ralf Mason, Dr. Travis Taylor, and Ms. Claudia Gollmeier. I would also like to thank Mr. Andre Popp for introducing me to my future wife. Finally, I would like to thank in particular my wife, Daniela Dunst, for all her love, patience, and sacrifices. I am indebted to her indefinitely for being here with me the last 3 years after having a long-distance relationship for a year. I am also thankful to my family, especially my three brothers to hold up the bond we have. The financial support I received during my graduate studies at UConn was provided by the following sources. I acknowledge JFCC and PLANET JAPAN for funding the Freon decomposition project, NGK for funding the carbon tetrafluoride decomposition studies, and DAIDO STEEL who funded the hydrogen production work. v Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. TABLE OF CONTENTS Page I. Introduction. 1 A. Overview. 1 B. Background Information. 2 1. Plasma Fundamentals. 2 a. The Plasma State. 2 b. Sheaths. 4 c. Physical Properties of Plasmas. 5 d. Processes Governing the Plasma. 10 e. Process at the Electrode and Other Surfaces and Interaction with Walls. 22 f. Gas Discharges and Breakdown in Gases. 27 g. Glow Discharges. 29 h. Radiofrequency Discharges and Microwave Discharges. 31 i. Silent and Corona Discharges. 33 2. Microwave Heating. 35 a. Microwave Applications. 35 b. Microwave Dielectric Heating. 36 C. Goals in this Dissertation. 40 D. References. 42 vi Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. II. Destruction o f Environmental Pollutants by Microwave Heating and Discharge Plasmas. 44 A. Destruction o f Freon Species by Discharge Plasma. 44 1. Introduction. 44 2. Experimental Section. 49 a. Preparation o f Gas Mixtures. 49 b. Plasma Reactor. 49 c. Experimental Setup and Parameters. SO d. Product Analysis. 51 e. Plasma Diagnostics. 52 3. Results. 52 a. Voltage and Current Characterization o f the Plasma. 52 b. Destruction o f Freons. 54 c. Optical Emission Studies. 62 d. Mass Spectroscopic Studies. 69 e. Power and Efficiency Measurements. 72 f. Long Term Stability. 75 4. Discussion. 75 a. Activation o f the Freon Molecules. 75 b. Model for Decomposition. 77 c. Efficiency Considerations. 83 5. Conclusions. 85 6. References. 87 vii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. B. Destruction o f Carbon Tetrafluoride by Discharge Plasma and Microwave Heating. 90 1. Introduction. 90 2. Experimental Section. 95 a. Preparation of Gas Mixtures. 95 b. Experimental Setup and Microwave Apparatus. 95 c. Plasma Experiment Setup. 97 d. Product Analysis and Catalyst Characterization. 98 e) Safety Concerns. 99 3. Results. 99 a. Discharge Plasma Studies. 99 b. Microwave Studies. 101 c. Extent o f Reaction and Characterization o f Catalyst. 108 4. Discussion. 109 a. Reaction Scheme and Proposed Mechanism. 109 b. Deactivation. 112 c. Glow Discharge and Zeolite Addition. 113 5. Conclusions. 113 6. References. 115 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. III. Generation o f Hydrogen from Water and Methane Using Discharge Plasma. 117 A. Hydrogen Production from Water and Methane. 117 1. Introduction. 117 2. Experimental Section. 121 a. Preparation o f Gas Mixtures. 121 b. Tubular Plasma Reactor. 122 c. Experimental Setup and Circuit Design. 123 d. Product Analysis and Characterization. 123 3. Results. 125 a. Water Splitting Using Various Reactors. 125 b. Water Splitting Using a Porous Cu Reactor in a Nitrogen Atmosphere. 127 c. Water Splitting Using a Porous Cu Reactor in a Methane Atmosphere. 129 d. Effect o f Admixtures to the Methane/Water System. 131 e. Reactions o f the Water/Methane System at Higher Voltages. 133 f. Mass Spectroscopic Studies. 135 g. Analysis of Carbon Deposits on a Cu Electrode. 137 4. Discussion. 138 a. Water Splitting in a Water/Nitrogen System. 138 b. Water Splitting in a Water/Methane System. 140 c. Effects of Admixtures and Effects o f Flow Rate. 143 ix Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5. Conclusions. 144 6. References. 146 B. Metal Effect and Flow Rate Effect in the Hydrogen Production from Methane 149 1. Introduction. 149 2. Experimental Section. 153 a. Preparation o f Gas Mixtures. 153 b. Tubular Reactor Setup. 153 c. Reaction and Circuit Setup. 155 d. Product Analysis and Characterization. 156 3. Results. 157 a. Methane Decomposition Using Different Electrodes. 157 b. Cleaning the Electrodes and Methane Decomposition. 162 c. Effect of Flow Rate on Methane Decomposition and Hydrogen Evolution. 163 d. Effect o f Input Voltage on Methane Decomposition and Hydrogen Evolution. 165 e. Mass Spectroscopic Studies. 166 f. Characterization o f Electrodes and Electrode Deposits. 169 4. Discussion. 171 a. Metal Effect. 171 b. Methane Decomposition. 172 x Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. c. Effect of Flow Rate, Input Voltage andElectrode Cleaning. 174 5. Conclusions. 175 6. References. 177 Appendix I. Future Work. 180 1. Freon Decomposition. 180 2. Decomposition of Carbon Tetrafluoride. 181 3. Hydrogen Production from Water and Methane. 182 4. Hydrogen Production from Methane. 182 Appendix II. Other Research Projects. 184 1. Decomposition o f 1-Butene using TitaniumDioxide Photocatalysis. 184 2. Alloy Preparation. 186 3. MCM-41 Synthesis. 187 xi Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LIST OF TABLES II-1. Efficiency as a function of flow rate, percent conversion, and power at a concentration of Freon 21 o f 0.5% and of oxygen o f 5.0% in helium and different input voltages (Vin). II-2. Efficiency as a function of flow rate, percent conversion, and power at a concentration of Freon 21 of 0.5% in helium and different input voltages (Vin). II-3. Major greenhouse gases, origins, lifetimes and global warming potentials. III-l. World primary energy demand history and projection until 2030. III-2. Trends of various air pollutants over the past 30 years in the United States in million tons/year. III-3. Usage of natural gas in trillion cubic feet. III-4. Methane decomposition for various metal inner electrodes at 2.56 kV input voltage and 20 mL/min CH«. III-5. Hydrogen production for various metal inner electrodes at 2.56 kV input voltage and 20 mL/min CH4 . III-6. Influence of cleaning the Pd inner electrode on methane decomposition at 2.56 kV input voltage and 20 mL/min CH4 . III-7. Influence of cleaning the Pd inner electrode on hydrogen production at 2.56 kV input voltage and 20 mL/min CH4 . Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. HI-8. Compounds found in the GC/MS analyses o f samples taken under plasma conditions. xiii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LIST OF FIGURES 1-1. Colliding spheres. 1-2. Elemental gas-surface interactions observed in the treatment o f silicon by a S ilty l^ plasma. 1-3. Surface processes involving energetic particle bombardment. I-4. Regions o f a glow discharge. II-1. Schematic o f the experimental setup for plasma generation and monitoring the voltage and current conditions. II-2. (a) Current and (b) voltage vs. time wave form for the closed circuit plasma conditions for a 0.5% Freon 21 and 5.0% oxygen in He mixture in an Fe PACT reactor. II-3. Destruction of 0.5% Freon 21 in 5.0% oxygen and He at 19.4 mL/min and Vp-p= 2.560 kV. II-4. Destruction of 0.5% Freon 21 in 5.0% oxygen and He at 20 mL/min and variable input voltage. II-5. Destruction of 0.5% Freon 21 in He at 20 mL/min and variable input voltage. II-6. Destruction of 0.1 % Freon 21 in He at 20 mL/min and variable input voltage. II-7. Destruction of 0.5% Freon 21 in 5.0% oxygen/water and He at 20 mL/min and variable input voltage. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. II-8. Destruction o f 0.5% Freon 21 in 5.0% oxygen/water and He at Vp-p= 2.440 kV mL/min and variable flow rates. 60 11-9. CO2 production from of 0.5% Freon 21 in 5.0% oxygen/water and He at Vp-p= 3.440 kV mL/min and 20 mL/min. 60 11-10. (a) Destruction of 0.5% Freon 142B in 5.0% oxygen/water and He and (b) CO2 production at variable voltage and 20 mL/min. 61 II-l 1. OEM spectra of (a) pure He, (b) 0.25% Freon 142B in He at 20 cc/min flow rate and 5.6 kV voltage. 63 II-l 1. OEM spectra of (c) 5% O2 and 0.25% Freon 142B in He, and (d) 5% O2 and 0.25% Freon 142B in He with H2O at 20 cc/min 64 flow rate and 5.6 kV voltage. 11-12. OEM spectra of (a) 0.5% Freon 21 in He, (b) 5%02 and 0.5% Freon 21 in He, and (c) 5% O2 and 0.5% Freon 21 in He with H2O at 20 cc/min flow rate and 5.6 kV voltage. 68 II-13. Mass spectra of (a) pure He, (b) 1.0% Freon 21 in He before plasma, and (c) 1.0% Freon 21 in He with plasma conditions. 70 11-14. Correlation of average power to input voltage for 0.5% Freon 21 in He and 0.5% Freon 21 in 5.0% oxygen and He. 72 11-15. Schematic o f the greenhouse effect. 90 11-16. Potential impacts of climate change on the environment. 92 II-l 7. Reaction setup for CF4 decomposition. 96 n -1 8 . Photograph o f the reaction apparatus. 96 II-19. Sketch of tubular PACT reactor. 98 xv Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 11-20. Destruction of 2% CF4 in 3% H2O and N2 at Vp-p= 12.2 kV at 11-2 1 . 100 cc/min. 100 Decomposition of CF4 versus time using the high surface area activated carbon at 10 mL/min and 2% C F4,3% H2O and in a balance o f N2. 102 11-22. Plot of CF4 decomposition versus time, top plot is the CF4 plot, bottom plot is the CO2 plot, power was turned on to 480 W at 46 min. 104 11-23. Plot of CF4 conversion versus time, power was turned on to 480 W at 46 min. 105 11-24. Dependence of the conversion on the power and the corresponding trend line. 106 11-25. Hydrogen production with and without CF4 present, and using the activated carbon catalyst at various power settings. 11-26. Photograph showing the extent of reaction at various power settings. 107 108 11-27. Simplified reaction scheme for the presence and absence o f CF4, activated carbon is presented as AC. 110 II-28. Reaction mechanism for the CF4 destruction. Ill III-1. Photograph showing Pd and porous copper electrode. 122 III-2. Schematic Setup for tubular PACT reactor. 123 III-3. Photographs o f porous Cu electrode and Pd electrode setup. 123 III-4. Circuit diagram for the tubular plasma reactor. 124 xvi Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. III-5. Water splitting at various flow rates, Au Fan, Vnra=500 V 3.0%H2O in N 2. 126 III-6. Hydrogen production (vol%) using a feed of 3.1% water in methane at various flow rates. 128 III-7. Hydrogen yield in % using a feed o f 3.1% water in methane at various flow rates. 129 IH-8. Hydrogen production (vol%) using a feed o f 3.1% water in methane at various flow rates. 130 III-9. Decomposition o f methane in relation to flow rate and conversion o f water and methane to hydrogen in relation to flow rate. 131 III-10. Effect of admixture o f argon on the hydrogen production at 20 cc/min. 132 IU-11. Effect of admixture of nitrogen on the hydrogen production at 20 cc/min. 132 III-12. Hydrogen production (vol%) using a feed o f varying methane content in air at 2.2 kV (rms), 20 cc/min flow, porous Cu reactor. 134 III-13. Hydrogen production (vol%) using a feed of varying methane content in air at 2.2 kV (rms), 20 cc/min flow, Pd reactor. 134 III-14. Mass spectrum o f the methane/water/nitrogen system. 136 III-l 5. Mass spectrum of the water/nitrogen system. 136 III-l 6. Mass spectrum of the air/methane system. 137 III-17. Formation of petroleum and natural gas. 149 III-18. Reactor setup for exchangeable electrode in methane reactions. 154 III-19. Exchangeable electrode, example of a Pd electrode. 155 xvii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 111-20. Mass spectrum o f reaction with Ni electrode showing methane destruction and hydrogen evolution. 158 111-21. Mass spectrum of reaction with Pd electrode showing methane destruction and hydrogen evolution. 159 111-22. Hydrogen production (vol%) using a feed methane at various flow rates and 2.56 kV. 164 111-23. Methane decomposition in % using a feed methane at various flow rates and 2.56 kV. 164 111-24. Methane decomposition in % and hydrogen evolution in vol% using a feed o f 20 mL/min methane at various input voltages. 165 111-25. Mass spectrum of methane decomposition reaction with power off. 166 111-26. Mass spectrum of methane decomposition reaction with power on. 167 111-27. X-Ray diffraction pattern of copper coated electrode. 169 111-28. X-Ray diffraction pattern of tin/tin oxide coated electrode. 170 xviii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER I. INTRODUCTION A. OVERVIEW. Environmental pollution is one o f the greatest man-made problems o f our time. At the same time, pollution is one of the great challenges mankind faces this century. Some of the environmental effects o f pollution are still disputed, especially in the field o f global wanning. There is insufficient data to support the hypothesis of global wanning but more and more proof has been found for a lasting wanning trend.1 International agreements in the form o f the Kyoto Protocol were struck to limit the exhaust of greenhouse gases.2 Other pollution problems were addressed as well such as the depletion o f the ozone layer by the Montreal Protocol and the following amendments.3 Additionally, the interest in fuel cell technology and in low polluting transportation with hydrogen as a fuel is rising.4 The work described within this dissertation deals with processes to decompose harmful environmental pollutants and to produce hydrogen for application in fuel cells and in vehicles. The emphasis will be on plasma processes which find more and more applications in today’s industry.5 In this chapter, basic concepts of plasma physics and chemistry will be presented that will be useful for discussions in later chapters o f this thesis. In addition, the concept of microwave heating will be discussed briefly as relevant to this thesis. 1 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 B. BACKGROUND INFORMATION 1. Plasma Fundamentals. a. The Plasma State. The term “plasma” is used in a variety of scientific fields such as biology and physics. The first use o f the term in the context of gas discharges is attributed to Langmuir in 1927.6 He described a plasma as a strictly defined region of uniform glow in a vacuum discharge tube, but nowadays it includes almost all phenomena associated with ionized gases.7 A plasma, also called the fourth state of matter, is an assortment of randomly moving charged particles which is as a whole electrically neutral but is a good conductor of electricity.78 Apart from the charged particles which are positive and negative ions and electrons, plasmas contain two further components: photons (quanta of electromagnetic radiation) and electrically neutral gas molecules. The molecules present consisting of one or more atoms occupy either a ground or an excited energy state in regards to rotation, vibration or electronic nature. Positive ions may carry charges of one or more and be in an excited or ground state, while negative ions are singly charged. These plasmas are usually generated by an electric discharge in a gas.6 The plasma state is described by an equally charged particle density o f electrons and ions and overall and an equilibrium temperature. To promote pure substances into the plasma state, temperatures o f up to 20,000 K for helium are needed if thermal equilibrium is imposed. In most practical discharges, however, thermal equilibrium is almost never achieved in either low-pressure or atmospheric pressure systems. Most discharges discussed in this thesis are characterized as electrically driven and weakly ionized. Consequently, the supplied power heats up the free moving electrons predominantly Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. whereas the heavy ions collide with the neutral background gas species and transfer their energy efficiently. The electron temperature is therefore orders of magnitude higher than the temperature o f the ions. Typical electron temperatures are on the order of 10,000 to 100,000 K. or higher for low-pressure glow discharges and somewhat lower for highpressure arc discharges.8 Plasmas can be divided into two main groups and then into subgroups with the electron temperature being one of the main characteristics. The first distinction is made in the temperature o f the plasma; a division into high- and low-temperature plasmas. Hightemperature plasmas have an ion and electron temperature of larger than 10 million K. with fusion plasmas as an example. Whereas low-temperature plasmas are subdivided into two classes. Thermal low-temperature plasmas have an equal ion and electron temperature o f less than 20,000 K; an arc plasma at atmospheric pressure being an example. Non-thermal low temperature plasmas have an ion temperature of around room temperature and electron temperatures of around 100,000 K; a low-pressure glow discharge is an example for this type of plasma.6 Plasmas are overall neutral but very strong electric fields called micro-fields which are able to form regionally on the microscopic level especially in highly ionized plasmas. Due to the motion of the particles, these fields fluctuate in magnitude and direction.1 Space charges might form on a macroscopic level due to external influences and changes in the movement of charge-carriers in the induced electrical field. These charge-carriers are also subject to Coulombic interactions between each other, yielding to collective interaction at high densities of charge-carriers. Plasmas themselves tend to minimize the external and magnetic fields inside the body o f the plasma contrary to the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. surrounding areas.9 Because o f the ionization o f the gas, the number o f particles and, therefore, the total pressure increases, depend on the degree o f ionization. In highly ionized plasmas, the increase is significant and this has been confirmed in experiments, but Dalton’s law o f partial pressure still holds.6 b. Sheaths. Quasi-neutral plasmas are connected to wall surfaces by thin positively charged layers called sheaths. The reason lies in the difference in the thermal velocity o f the electrons which is 100 times greater than that of the ions. The net charge density is zero as well as the electric potential and electric field; therefore the fast-moving electrons are not restricted and lost very rapidly to the walls. On a very short timescale, electrons lost near the walls lead to a very thin positive ion layer near each wall, a sheath, in which the ion density is much larger than the electron density. The result of the net positive sheaths is a potential profile that is positive within the plasma itself and approaches zero very quickly near both walls leading to a restricting potential valley for electrons and a potential hill for ions. Electric fields in the sheaths are directed from the plasma to the wall; electrons in the sheaths are pulled back into the plasma whereas ions are accelerated into the walls.8 The distinction into bulk plasma and sheaths is common to all plasmas. The bulk plasma is quasi-neutral and strength of instantaneous and time-averaged fields is low. The plasma dynamics are distinguished into high and low-pressure processes. At high pressure diffuse ion loss is dominant, whereas at low pressure free-fall ion loss is important. In the positively charged sheaths, strong fields are present. The dynamics Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5 required in this case encompass a variety o f ion space charge sheath laws such as low voltage sheaths and high voltage sheath models including collisionless and collisional Child laws and modifications. The interface of plasma and sheaths requires certain conditions. The mean ion velocity at the interface has to be equal to the ion-sound velocity uB, also called Bohm velocity, which is given in Equation 1, in which e and M are the ion charge and mass and Te the electron temperature in Volts.8 c. Physical Properties o f Plasmas. The plasmas discussed in this thesis are atmospheric pressure non-equilibrium plasmas, which can also be describe as non-thermal low temperature plasmas. These plasmas are characterized as weakly ionized plasmas and their properties resemble those of transient high-pressure glow discharges.10 Some concepts, which were developed for fully ionized plasmas, also hold for weakly ionized plasmas. The idea o f a “shielding distance D” was developed for fully ionized gases. The uniformly distributed charges of a plasma establish a positively charged local electric field, which attracts electrons, thereby increasing its strength and repels slightly the heavy positive ions decreasing the field at greater distances. The field, which decreases proportional to the inverse square of the distance, is limited to a finite region and is zero outside of it. The shielding distance D is defined as the distance at which the field has the value of zero. In typical discharges, the electron temperature is 10,000 to 100,000 K and the charge densities o f 109 to 1014 cm'3 are observed; this yields according to Equation 2 a shielding distance D o f 10'1to 1O'4 cm (N is the particle density and T is the electron temperature). Consequently, this distance is Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. larger compared to the average distance between charges <iK r> which is on the order o f 10'3 to 10*5 cm. This property of the shielding distance D is also applicable for partially ionized and non-isothermal plasmas. If a plasma is bounded by an insulating wall, a local field arises, the afore-mentioned sheaths, with a thickness on the order of D.6 The shielding distance D is also called the Debye screening length Xq. A plasma state is deemed ideal if in a sphere of the radius A© many charge carriers are present in other words:— » 1 , with nt0 being the electron density at neutrality. The electrostatic energies is much smaller than the thermal energy at the mean distance; if it exceeds the thermal energy the plasma is considered non-ideal.9 In order to maintain quasi-neutrality, the kinetic energy o f the charge carriers has to be much larger than the space charges field energy. The acceptable deviation of the electron density nt Ane = ne0 - nt from neutrality ( = ni0; is the ion density at I k neutrality) is given by-— —< — , where L is a characteristic plasma length (radius of a 1 plasma column),. Inside Xd, as defined above, the plasma might deviate from neutrality; the dynamics of this is controlled by the Langmuir plasma frequency.9 One o f the main difficulties when dealing with non-thermal low-temperature plasmas is that the determination of the distribution functions F for the electron velocity or energy in the plasma for every case under existing terms requires different approaches and solutions. One way is to use an adapted solution Boltzmann equation. This was a step Reproduced with permission o f the copyright owner. Further reproduction prohibited without permission. in the right direction in the last decades of the last century especially by treatment of the complexity of the collision integral and the terms involving time and space variables. The complete solution is still not achieved and therefore approximation of the distribution functions by simpler formulae is used. Equation 3 shows a simplified electron energy distribution f 0(U) , which has proven to be o f value; with U being the electron energy, Ue the distribution parameter (kinetic electron energy), m being a positive integer, T being the Gamma-function whose values are tabulated, and a as defined in Equation 3. For m = 1 a Maxwell distribution is given by Equation 3, whereas m = 2 yields a Druyvesteyn distribution.9 Q /o (^ ) = ^T7Te (i/ m)(3_2")/2" ■£(£) ;m > 0;a = ------------------(3) ' — r(3/2m ) The kinetic temperature of the electrons can be determined by using the mean energy Um according to Equation 4, with e0 being the charge o f the electron and the other terms as defined above. The electron temperature T, for the afore-mentioned Maxwell distribution (m = l) is therefore given by Tt (Kelvin) = 11600C/e(Fb/tt)or approximately as 1 V s 104 K.9 f ^ W. -e ,U m -,Um = ] u ,l2f 0(U)dU (4) 0 An important factor for particle and energy transport is the electron flow density, which is dependent on the electron density and drift velocity. If mathematically expressed, the density includes the electrical field drift and the action o f diffusion and thermodiffusion. The mathematical terms are approximated, mobility, diffusion Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. coefficients and free path terms are introduced, and transport coefficients are yielded. The particle flow is coupled with the flow o f energy.9 Inhomogeneous regions are observed in plasmas near insulating walls or floating metallic surfaces. In these regions, behavior is observed similar to a neutral gas under the influence of external forces at the Boltzmann equilibrium. These regions are not sheaths due to validity o f plasma conditions and quasi-neutrality in these cases. Nevertheless, there are significant differences in plasmas from the Boltzmann equilibrium with neutral gases. Unlike with a neutral gas, the forces originate from space charges in the plasma. The plasma deviates from the Maxwell distribution function, and this has to be considered. Variations in the distribution function in space are observed. One of the conditions for the Boltzmann equilibrium is the loss of particle flow in the flow direction. Deviations from the Maxwell distribution are fairly common in plasmas and primarily due to collisions with heavy particles such as ions; therefore, it is important to consider the spatial variations of the electron distribution function prompting a thorough investigation of the electron energy dependencies. Recent studies show the non-local complex character of balancing power and momentum in space-dependent plasmas originating from the Boltzmann equation. If collisions are ignored, an energetic quasi equilibrium causes the prevalence and stability o f the Maxwell distribution.9 The movement o f ions and electrons inside a plasma is interrelated by electric space charges which cause equal electron and ion drifts if external influences are missing. The steady state drift velocities of electrons and ions come together as the velocity of ambipolar diffusion v^, when these particles drift towards insulating walls. The range of ambipolar diffusions exhibits similarities to the Boltzmann equilibrium.9 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. If a plasma is exactly at equilibrium, it is possible to establish the degree o f ionization by employing the Saha-Eggert equation. For non-equilibrium low-temperature plasmas, the determination o f the degree o f ionization necessitates an energy balance o f electrons yielding as shown in Equation 5 a simple and useful term for this value. n. n 2*^nP IV (5) 3Sloae0Uen‘ In this equation, z, is the mean free time between electron collisions, P/V is the power density provided to the plasma, 6 ^ is the mean fraction of energy that electrons lose in a single collision, n and nt are the total particle and electron density respectively, e0 is the electron charge, Ue is the electron energy. The loss for elastic collisions is on the 2m order o f HT4to 10*s w i t h ^ = Stl = — - . In the case o f inelastic collisions this loss djMl M is dependant on Ue is usually larger by one or two orders of magnitude. The term tots but it shows only slight variations and is viewed constant especially in regards to the P IV 9 range o f the term — — ; this approximation holds well at pressures above 1 kPa. n The mobility of a particle n , which describes the influence of an electric field £ on a particle and its gain o f velocity o f kinetic energy (characterized as the drift velocity u ) therein, is given by equation 6. Herein, the charge o f the particle e , the mass of the particlem , and the time between collisions r are important variables.11 “ =— e t (6) // = — E nt Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 10 The electrical conductivity o f plasmas a is given by the mobility of the electrons ( fit » ft, ) and the electron density (n ( * n ,) and can be approximated to Equation 7 by using the definition of the mobility given above wheree0,n e, m,and xt are the electron charge, density, mass, and mean free time o f flight respectively. ° =el nJ t / me (7) A distinction has to be made in regards to the degree o f ionization. In weakly ionized plasmas, the mean free time of flight r f is described by electron-atom collisions and independent of nt resulting ino ocn,. For fully ionized plasmas, re is defined by Coulomb collisions with t t « 1I nt and the resulting conductivity is constant and can be described by the Spitzer equation.9 Due to the presence o f space charges, the electrons show quasi-elastic coupling to fe h T the ionic surroundings resulting in oscillations with the frequency con = - °— with £0 V£ om e being the vacuum permittivity and all other parameters as defined above. This frequency is detrimental to the propagation of electromagnetic waves in the plasma. Strong damping of waves with a wavelength a is observed forty < (op, whereas at (o = o)p strong reflection at the interface o f the plasma is seen with a refractive index given by the Eccle relation.9 d. Processes Governing the Plasma. The plasmas usually used at atmospheric pressure are weakly ionized. These complex systems contain mostly neutral gas atoms and to a lesser degree electrons and Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 11 positive ions. The main suppliers o f power for plasmas are electric fields acting on particularly electrons and other charged particles. Due to the minute electron mass, energy transfer stemming from elastic collisions between the electron and the heavier gas particles is very poor, and as a result the mean kinetic energy o f the electrons is much larger than that of the other particles leading to a non-thermal plasma condition. Enough electrons have therefore the appropriate energy to participate in inelastic collisions with atoms and molecules and thereby losing energy and causing depopulation o f higher energy states. The electrons are far from an equilibrium state due to the actions o f electric fields, elastic and inelastic collisions and therefore their description is difficult. This can be done by either particle simulation or the formulation of electron kinetic equations and its solutions.9 The kinetic treatment of electrons is aimed at yielding its velocity distribution and other macroscopic properties, which influence the overall behavior of the plasma; electrons are the species providing the activation of the working gas by various collision processes. The velocity distribution function is dependent on the specific plasma state. Furthermore, various collision parameters and processes such as elastic collisions and excitation have to be taken into account as well as electric field parameters. Complexity is added when inelastic processes are considered and rotational and vibrational excitation take place. The solution and the approximations necessary to get it is ultimately dependent on and specific for the specific plasma state, steady-state, time-dependent or space-dependent.9 Collisions in the plasma can be categorized into two main classes: elastic and inelastic collisions. In elastic collisions, the only energy exchanged is the translational Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 12 energy. If spherical particles are assumed for this case, the outcome of the collision can be forecasted based on the laws o f conservation o f energy and momentum. The collision depends only on the impact angle 6 (as shown in Figure 1-1) between the velocity vector of the moving particle and the line joining centers at impact as shown in Figure 1-1. A particle o f mass m, and kinetic energy £, strikes a stationary particle of mass m2transfer the energy A e . The energy transferred is defined by Equation 8, which yields the maximum transferred energy for a central collision. The average energy fraction transferred is ytS^and ifm, « m 2 -> 2m L « 1 . Electrons consequently only m2 lose a very small fraction o f their kinetic energy in each elastic collision.11 4m, m2 2 4m,m2 A* = ex ----y cos 9 ,5 ^ = ------1-^ -2- (8) (m , -t-m2) (m , + m 2) Moving particle Velocity e1 Stationary partide Figure 1-1. Colliding spheres (Adapted from Ref. 11). The average energy loss for an electron is assumed as zero given that the uniformity of the temperature and therefore a constant average energy for all gas particles. The term <5defined above can be used in systems where one group of particles has a higher Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 13 energy than another; 6 characterizes the average fraction o f excess energy transferred in a collision of particles from both groups. The transfer of energy by electrons to stationary particles is a realistic situation found in many discharges, in which they transfer only minute amounts of energy in elastic collisions resulting in a difference in particle temperature. The electron temperature is orders o f magnitude higher than that o f other particles such as ions. Heavy particles, on the other hand, tend to yield no net transfer o f energy in collisions because they generally have similar energies; nevertheless they are able to participate in elastic collisions showing very effective transfer o f energy, in contrast to electrons.11 The situation for momentum transfer is different. In the casern, « rrij, only a very small fraction of the velocity of m]will be passed on tom2, yielding to a situation where the relative velocity equals that of m, at an angle ^ , the angle of scatter. The probability of all values of <fi is equal in the case of/n, which means that isotropic scattering occurs. An electron has therefore a velocity o f exactly zero, on the average, after an elastic collision; all the momentum is transferred from the electron to the heavy particle. These findings are based on the assumption that the cross sections are constant for real particles, which they are not. For Q, the total cross section per unit volume, two definitions can be found: 1. The probability o f a particle making a collision in unit distance 2. The total effective area per unit volume o f gas presented to a single particle; which means in practical term, the larger this area of the target particle is, i.e. an atom, the more probable is an interaction o f the target particle with the single projectile particle, i.e. an electron. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 14 However, the values are close enough to the true values. For momentum transfer this holds as well, the error is almost negligible and the dominant scattering process is forward scattering in contrast to isotropic scattering as predicted by the elastic-sphere model.11 In the classical sense, an inelastic collision is characterized by conservation o f momentum but the total kinetic energy is reduced and the heavier partner absorbs energy in a specific form (i.e. heats up). A collision is deemed inelastic in the microscopic sense when the internal excitation energy o f a particle is altered. In the case o f discharges, the important excitations are generally the ones of electron energy levels. An electron is able to cause excitation or ionization o f atoms if its kinetic energy has the appropriate value to overcome the energy difference between electron levels. An inelastic collision occurs if the kinetic energy of the electron £, is larger than the energy between the mth and nth l e v e l T h e cross section for£„„ is zero for increases to a maximum with in c re a s in g a n d then decreases at larger^,. If the mth state is the ground state, is equal to the ionization potential, the process itself is then called electron impact ionization. Further excitation or ionization of an already excited atom is possible if the atom collides with adequately fast electrons. For this case, the cross section is naturally higher for a given £,, but the threshold value is lower.11 Superelastic collisions, the reverse process o f excitation by electron collision, can occur as well. The electron receives the excitations energy in the form o f kinetic energy. A three-particle collision in which a positive ion collides with two electrons, recombines with one o f the electrons and transfers kinetic energy corresponding to the ionization potential to the other electron is the reversal o f the ionization. The reversibility o f each Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. IS and every collision is valid and the equilibrium o f a gas assures that each process and reverse process occur at the same rate.11 Neutral gas atoms and positive ions can cause impact ionization and excitation, as well. In this case, the cross-sections are to a great extent smaller compared to electron impact, the exception being at high energies (200 eV and above) where the relative particle speed is similar to the one of electrons. Furthermore, not the electron mass alone as in usual collisions, but the reduced mass has to be used when the particles have comparable velocities. This reduced mass has to have the appropriate relative kinetic energy (kinetic energy o f both particles relative to the reduced mass) or exceeding it to induce the transition. As before, the reverse process yields kinetic energy, which equals the energy of the ionization or excitation process.11 Ionization can occur by processes other than electron or particle impact. If the temperature of the gas, the temperature o f the neutral and positive particles, not the electron temperature, is high enough, atoms might acquire enough random energy to induce ionization of other atoms by a single collision. This thermal ionization is an important part of high-pressure arc discharges. In flames chemi-ionization takes place by chemical exchange providing the necessary energy. If there are not particles with enough energy to induce ionization in a single collision, cumulative ionization becomes important. Several consecutive collisions from any kind of ionization (thermal, electron impact, ...) give the collision partner the necessary energy for ionization, through multiple stages of excitation. At high pressure and temperature, cumulative ionization plays a significant role because the frequent collisions yield a high probability that an atom in an excited state is involved in an ionizing collision.11 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. I 16 Other ionization processes o f interest are auto-ionization, surface or contact ionization and photoionization. Auto-ionization occurs when an atom becomes doublyexcited (two electrons in higher valance energy levels) by successive collisions and instead o f going back to its ground state releases one o f the electrons and the other one falls back to the lower energy level. The condition for this to happen is that the total excitation energy is greater than the energy required for ionization, and the released electron receives the balance of internal energy; if this balance is small the step is more likely." Surface ionization occurs when a neutral particle interacts with a metallic surface and loses an electron to this surface leaving as a positive ion. Ionization is favored if the work function of the metal (potential which has to be overcome in order for an electron to leave the surface of a metal; constant for a given metal but lower than the metal’s ionization potential) is larger than the ionization potential of the interacting atom; the ionization is only significant if particles of low ionization potential such as alkali metals are involved. This effect combined with the thermionic emission o f a hot metal surface is used in a quiescent plasma (Q-machine), where heated tungsten plates are employed to sustain a plasma of alkali metal vapor in a magnetic field. The stable plasma shows a thermal equilibrium with the plates, superior to most gas discharges for experimental use." Photoionization is the extreme case of photoexcitation. Photoexcitation is the I reverse process to spontaneous emission, described below. An atom is elevated to a higher energy level by absorption of quanta o f light in the form o f a photon. The frequency o f the absorption is governed by Equation 10, given below. If the photon has Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. I 17 enough energy to ionize the atom, which is in its ground state, the process is called photoionization. The condition for ionization o f an atom with ionization potential Fjby a photon o f the frequency / is given by Equation 9 with h being Planck’s constant. hf> eV , (9) If the photon has a frequency that exceeds the critical frequency given by Equation 9, the photon is still absorbed and the excess energy is given to the liberated electron as kinetic energy. The critical frequency is in the X-ray section of the electromagnetic spectrum for most gases, but for metal vapors it lies in the ultraviolet region.11 Collisions of the second kind are important in charge transfer processes. These collisions are inelastic collisions in which potential energy is lost or transferred. On collision o f a positive ion and an atom, the positive charge is transferred which happens by the ion receiving an electron from the neutral atom. If the two particles are of the same species, they go on with unchanged kinetic energy but appear to have exchanged it because fast atoms appear to be slowed down whereas slow atoms seem speeded up leading to problems for ion mobility studies in gases. The cross sections for the charge transfer are usually greater than for other inelastic collisions involving ions and atoms. Excitation energy can also be transferred between atoms. If the excitation energy is larger than its ionization energy, ionization can occur for the collision partner. The likelihood increases if the Penning effect is used where the excited state is metastable and the excited atom takes advantage of its longer lifetime to participate in collisions.11 The ionization involving free radicals is important. Free radicals are abundant in molecular plasmas, and they are the reactive species in these plasmas making plasmaassisted processing a feasible and innovative technology for industry and research. The Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 18 ionization mechanism is characterized by electron impact which is competing with dissociative ionization. A well researched example is the carbon tetrafluoride chemistry, where the cross sections for these two processes have been determined.9 In most discharges only positive ions, electrons and neutral gas particles are present, but in some gases negative ions form by combination of electrons with neutral particles. In most cases, these gases are electronegative gases. They lack electrons in their outermost valence shell and tend to complete this shell by attracting electrons; this tendency is called electron affinity. Examples are oxygen and halogen gases. The process itself in which a negative ion is formed upon collision o f an electron with a neutral particle is defined as electron attachment. The cross-section for electron attachment is a great deal smaller than for most inelastic collisions and declines with rising electron velocity. An attachment coefficient 6 can be defined as the ratio o f all electron-atom collisions to attachment collisions; the inverse of this ratio mirrors the attachment probability for a single collision. The electron affinity is typically on the order of 1 eV. Upon attachment the electron affinity energy and the energy of the electron are given off in several ways. Radiation can be dissipated, kinetic energy can be transferred to a third body (more probable at higher gas densities), or dissociation can occur if molecules are involved. Electron attachment is important because it provides a way to remove free electrons from ionized gases.11 Similar to electron attachment is random or volume recombination. This process can happens when a positive ion is involved in a collision with an electron or positive ion. The probability of recombination depends on the electrostatic attraction and the type o f recombination; electron-ion recombination is less likely than ion-ion recombination Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 19 due to the high velocity o f electron. The probabilities are similar to those o f electron attachment, decreasing with increasing velocity o f the particles involved, as is the case with the dissipation mechanism. When radiation is given off, the process is called radiative recombination. Third body collisions or dissociation can occur, as well.11 As stated before, the probabilities for ion-ion recombination are velocity dependant but more likely due to the inherently lower velocities compared to electrons. The velocity itself is dependant on the gas density or on the pressure, at constant temperature. The recombination is described by two theories depending on the pressure regiment. At low pressures, Thomson theory forecast the probability to be proportional to pressure. At high pressures, the probability only depends on the particle mobility and consequently is inversely proportional to the particle density according to Langevin theory. The maximum probability at the transition point o f these theories is at 1 atmosphere. A recombination coefficient a can be defined for both, electron-ion and ionion recombination. Important is recombination for de-ionization at high pressures where diffusion is slow compared to low pressures where diffusion is usually a more important process.11 As mentioned before, radiation is one o f the dissipation mechanisms in recombination and electron attachment reactions. An excited atom could also lose the acquired energy by spontaneous de-excitation into a lower energy level, not necessarily the ground level, and emit a quantum of radiation with a frequency /b ased on Equation 10 with h being Planck’s constant. h f = As (10) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 20 This likelihood of this process to happen depends immensely on the lifetime o f the involved excited state, the average time between excitation and spontaneous emission when collisions are absent. The lifetime is usually on the order of 10*8 s, but metastable excited states can exhibit lifetimes o f 10~3 s or longer. Consequently, if the collisions happen with a frequency considerably lower than 10* s, the spontaneous emission occurs with a higher probability before the particle collides. For metastable particles, on the contrary, collisions are the much more conceivable mechanism of energy loss. Induced or stimulated emission is also possible, when there is enough density of radiation already in the surrounding area of the atom at a suitable frequency.11 Radiative recombination is another recombination process and the reverse of photoionization, which was mentioned before. In this process, the ion-ion or electron-ion pair recombines and releases its potential energy and relative kinetic energy in the form o f a photon. Because the particles have a variety of velocities and therefore relative kinetic energies as they participate in recombining collisions, no distinct frequency for the emitted photons are observed, but frequencies which range upwards from a minimum corresponding to the ionization potential. This frequency is the limit o f the observed weak continuum spectrum extending into short wavelengths. Various of these continuum spectra could be observed if recombination occurs into a range of level apart from the ground state. Radiative attachment causes an effect similar to the one observed with radiative recombination. The probability o f radiation absorption by a particle resulting in excitation or ionization is dependent on the gas density and the frequency o f the radiation (showing enhanced probabilities at frequencies corresponding to excitation potentials). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 21 Free electrons can emit radiation as bremsstrahlung when they encounter positive ions, remain unattached but lose kinetic energy.11 A different excitation process, vibrational excitation, is particularly important for molecules where there is an interaction between the molecule and electrons. The cases have to be discerned for vibrational excitation, resonant and non-resonant processes in terms of cross sections. In the resonant case, the electron is in the vicinity of the target molecule for an extended period o f time, longer than the transit time, leading to a more effective distortion o f the molecule and an increase in cross section. If the resonance lifetime t is much smaller than the vibrational period ( r <<10^5), the nuclei do not vibrate during this time period. The colliding electron attaches itself to the target molecule yielding a transition state, which on emissive decay releases the electron leaving the molecule vibrationally excited. If the lifetime is similar to the vibrational period and considerably larger than the transit time, the “boomerang model” can be applied. In this model, an electron causes nuclear waves to propagate by attachment and later by emissive detachment. This results in vibrational excitation o f various vibrational states. If the lifetime increases even more, many nuclear vibrations are possible during r . For reaction purposes, molecules with higher vibrational energies can help make endothermic reactions occur. Dissociative attachment is a process similar to vibrational excitation. In both processes an electron attaches itself to a molecule, but the outcome is different, dissociation of the molecule in one case, and a vibrationally excited state and a free electron in the other case, and depends on various factors such as temperature and favorable energy states.12 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 22 The energy thresholds for vibrational excitation and dissociative attachment are on the order of a few tenth o f an electron volt These low energy electron collision mechanisms are some of the most efficient energy-Ioss pathways for plasma electrons. They thereby reduce the high-energy tail o f the electron density function of the plasma.9 e. Process at the Electrode and Other Surfaces and Interaction with Walb. The electrodes initiate the discharge and provide the charged particles to maintain the plasma. These electrodes emit electrons in several ways from its surface. Electrons are bound to the surface and the work function 4 o f the metal has to be overcome for the electron to leave. One method to supply the necessary energy to these electrons is to raise the electrode temperature; this process is called thermionic emission and was mentioned earlier. The emitted current density j with respect to the absolute temperature T can be calculated using Equation 11, where eis the charge of the electron, kis the Boltzmann’s constant and A is 60 A/cm2. Metals and certain oxides are used in specific cathodes for thermionic emission based on their low 4 since the emission is high in this case; for most metals this effect becomes important at temperatures higher than 1000 K. The emission can be reduced by sputtering, the removal o f particles from the electrode by the impact of gas phase ions." j = A T2e*lkT (11) Another way to supply the necessary energy for electron release is by the action o f a photon with a frequency / with a condition similar to Equation 9, but 4 *s substituted for Vi. From this relation, the minimum frequency, the threshold frequency can be determined which is for most metals in the ultraviolet, but for alkali metals in the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 23 infrared. If the photon has energy beyond the threshold, the emitted electron receives kinetic energy and the surface heats up slightly. The probability o f this happening is often expressed as the photoelectric yield, which is material specific.11 Electrons can be released from the electrode by the impact o f plasma electrons, positive ions or neutral atoms. The mechanism itself is not completely understood, however the impacting electrons have to have high electron energies (100-1000 eV). The yield is about 1.5 electrons per impacting electron for metals and higher for insulators. This process is unimportant in DC discharges, but important in high frequency discharges at low pressure. If a positive ion is to remove an electron from the electrode surface, it has to have a total energy o f twice e<f> because it releases one electron and needs one for neutralization. The probability o f this process happening depends on ion velocity, gas used and electrode material. These ions can cause sputtering on the electrode surface and could be given off as neutral particles without removing an electron. Excited or ground state atoms can release surface electrons upon impact if the total energy is equivalent t o ^ . It is easily observed with metastable atoms, but generally the yield is low for this process.11 An electric field (normal to a surface and drive electrons from it) can use the Schottky effect and thereby increases the thermionic emission due to the decrease in the work function. At high electric fields the current flow is higher than the thermionic value due to the pull o f the field on the electrode electrons. High values of the electric field can be achieved in low voltage discharge by space charges over short distances and do not require high discharge voltages.11 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 24 Particles interact with surfaces including electrodes in many different ways including scattering, sputtering, deposition, implantation, modification and erosion. Several methods are used to model these interactions with collision being the primary one apart from molecular dynamics considerations. Scattering, easily modeled by collision methods, can be described in terms o f scattering angle and energy transfer in elastic or inelastic collisions with surface atoms. Impacting ions not scattered or reflected are either implanted into the solid or deposited onto the surface. The energy loss (depending on nuclear and electronic interactions called stopping power) determines how deep the ions penetrate. Backscattering of ions is large at low impact energies (10-50 eV) but tails off then and can be expressed by a reflection coefficient.9 Physical sputtering can happen by various mechanisms. Primary knock-on and secondary knock-on particles are distinguished if they are directly impacted by the projectile or if they are influenced by a cascade o f impacts and receive energy from other target particles. The direction, inward or outward, in which the impacting particle was moving, can be important, as well. Low impact energies favor outwards primary knock out processes, whereas at higher energies inward secondary knock-on processes dominate. The degree of sputtering can be expressed by the sputtering yield (sputtered atoms per impact ion), which is dependent on the masses and charges o f the particles involved as well as the impact angle. Sputtering can also be caused by cluster and highly charged ions. These ions cause potential sputtering, based on the potential energy o f these ions; this process can remove more than 100 particles per impact. Sputtering involving electrons as the impact partner and as the leaving partner has been discussed above and is usually called electron emission.9 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 25 Chemical sputtering (CS) relies on chemical effects and involves bond breakage of surface atoms. These effects depend strongly on the reactivity of the surface and the agents in the gas phase and also the temperature. Chemical sputtering is considered a multi-step process in which a volatile molecule is formed after an impact reaction of projectile and substrate atom and leaves the surface immediately (direct CS) or after a certain residence time (delayed CS). The formation can be exothermic; therefore no minimum energy transfer is needed, especially if oxygen ions or other reactive species are involved. The kinetic energy distribution is shifted to lower energies, and thermal desorption contributes also. Examples of the usage are dry etching for computer chips or plasma cleaning.9 Surface interactions are important for the plasma-wall interactions; various surface interactions are depicted in Figure 1-2, the example of a Si/SiFtyFk-system. A particle that is not reflected upon impact is adsorbed onto the surface and held by weak electromagnetic bonds. Physisorption is weaker compared to chemisorption and is based on dipole-dipole interactions or dispersion forces. Chemisorption is caused by valence forces o f the exchanging electron orbitals o f substrate atoms and adsorbed particles forming a strong chemical bond to the substrate. Upon approach to the surface, particles can break apart and then become adsorbed, this is called dissociative chemisorption. The adsorption can be characterized by the degree of coverage (adsorbed particles per available site), which depends on surface morphology, vacant sites and other parameters. Adsorbed species migrate along the surface in a surface diffusion process. This diffusion depends on crystal structure whereas diffusion coefficients depend on the structure o f the crystal face, and chemical composition. Bulk diffusion is possible as well. Desorption is Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 26 possible after a certain residence time; desorption mechanism include thermal and collisional desorption. If equilibrium can be established, Langmuir isotherm can be determined based on the balance of adsorption and desorption.9 Absorption of SiH3 (sticking to passivated surface) Diffusion (hopping) H, molecule 0*0 0*0 Q O' Q i » Desorption (thermal) Chemisorption (resulting in dissociation) 0 -0 0-0 Physisorption 0-0 Q 0 Silicon surface Dangling bond (silicon) Hydrogen passivation Figure 1-2. Elemental gas-surface interactions observed in the treatment of silicon by a SiHVHz plasma (Adapted from Ref. 9). Reflection of an impacting particle happens when its energy is too high for participation in adsorption or too low for sputtering or implantation, or no appropriate surface site is available. The impacting particles lose energy yielding de-excitation o f the particles, heating of the surface until thermal equilibrium with the surrounding gas, and collision cascades causing activation of surface reactions (mixing). Apart from physical sputtering (the erosion of solid atoms) and chemical sputtering (an Eley-Rideal mechanism), the surface film reaction following a Langmuir-Hinshelwood model is an important process. In this type of interaction, the gas phase particle is adsorbed onto the surface before the chemical reaction occurs after a specific time period. A recombination Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 27 is a characteristic step in this reaction. A very good summary o f the aforementioned reactions is given in Figure 1-3. Deposition on surface Surface activation Surface diffusion \ •• • surface* bulk Surface film reaction • • • ./// II ' X Damaging impact Physical sputtering Implantation into bulk Collision cascade Figure 1*3. Surface processes involving energetic particle bombardment (Adapted from Ref. 9). f. Gas Discharges and Breakdown in Gases. Gas discharges in the steady state can be classified by the current they carry: 1) Townsend or dark discharge with currents up to 10^ A. 2) Glow discharge with currents between I O'6 A and 10'1A. 3) Arc discharge with currents of 10*1A and higher. The discharges dealt with in this thesis fall into the glow discharge region with a current of a few mA. When these discharges are started, they go through a region called the Townsend region where they behave similarly to Townsend plasmas up until a breakdown voltage is reached, even at atmospheric pressure. Before a discharge is Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 28 started, there is a certain equilibrium between the production and recombination o f charged particles. When the voltage is increased, the current increases up to saturation plateau where ions and electrons reach the electrodes before recombination. With increasing voltage the current stays more or less constant up to the breakdown region where the current increases almost exponentially and the plasma becomes self-sustaining. The breakdown voltage is dependent on gas conditions, electrode spacing, material and shape. At the breakdown point, due to the increase in the electric field, the electrons are accelerated more and more between their collisions up to a point where their energy exceeds the ionization potential before a collision. Therefore, they are able to ionize atoms on their way from cathode to anode producing positive ions and more electrons. These electrons themselves can participate in collisions and ionize atoms yielding a cascading effect. This is similar to starting an avalanche and the discharge becomes self sustained. More electrons reach the anode and additionally positive ions reach the cathode leading to an increase in current. This mechanism was proposed at low pressures but holds fairly well at atmospheric pressure. The breakdown voltage can be determined using Paschen’s law where it depends only on the gas pressure and the electrode spacing. The condition for the breakdown is given by the Townsend criterion as given in Equation 12, where d is the electrode spacing, y is the cathode yield, and a is Townsend’s first ionization coefficient (ionizing collisions made on average by moving 1 cm in the direction o f the electric field).11 ye* =\ (12) Reproduced with permission o f the copyright owner. Further reproduction prohibited without permission. 29 At higher gas pressures, around atmospheric, breakdown not only involves electron avalanches originating from the cathode, but also streamers originating from the anode reaching the cathode. These streamers are caused by space charges at the anode yielding an increased field and therefore causing additional avalanches by a few photoionized electrons in the space charge region; breakdown occurs when these streamers reach the cathode. Corona discharges at these pressures do not really have a breakdown, but they become self-sustaining with burst coronas and streamers until the breakdown into a high-voltage discharge. At high frequencies, AC discharges can be treated similar to DC discharges in respect to breakdown up to a few kHz. Breakdown is characterized in this case by electron diffusion and ionizing collisions in the oscillating field; breakdown occurs if the diffusion loss is smaller than the collision production rate.9 g. Glow Discharges. Glow discharges as electronically driven discharges, usually direct current discharges, have to be discerned from microwave induced discharges and capacitively driven discharges such as RF plasmas. Glow discharges are divided into sub-normal Cow V), normal, and abnormal glow discharges (high V) depending on the applied voltage and the resulting current. In the subnormal region the voltage decreases and the current increases. The normal discharge is characterized by a constant current density at the cathode with the discharge only partially covering the cathode surface and by the normal cathode fall (drastic drop in electrical potential) depending on the gas and cathode material. The abnormal region (close to the glow-to-arc-transition) shows increasing current with increasing voltages, total coverage of the cathode by the discharge. Arc Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 30 discharges, characterized by a drop in current at increasing voltage, are outside the scope of this thesis, but are a very important class o f gas discharges.9 Cathode dark space Aston dark space ( Faraday dark space Anode gk)W Cathode glow Negative glow positive column Anode dark space Cathode fall Figure 1-4. Regions o f a glow discharge (Adapted from Ref. 11). The glow discharge shows distinct regions as seen in Figure 1-4. The cathode dark space, where the electrical potential drops significantly (cathode fall), separates the negative glow, the brightest region of the discharge and where the electric field is close to zero, from the cathode. The negative glow is followed by the Faraday dark space, a diffuse region, and the bright positive column. The positive column, homogenous or streaked, fills almost all the space to the anode, where it is separated by an anode glow and an anode dark space. Almost all important reactions, also sputtering and etching of the electrode, take place in the cathode regions sustaining the discharge. Positive ions from the negative glow are accelerated by the electric field of the cathode fall towards the cathode surface where they release secondary electrons, which in turn are accelerated in the cathode fall to strike heavy particles. The produced positive ions in the cathode fall and negative glow start the cycle all over again. The positive column, where electrons gain energy, is only formed if there is a long, narrow discharge gap with wall losses of Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 31 charged particles. Cathodes heat up as the discharge current increases; therefore measures (resistor in series with reactor) have to taken in order to prevent the transition to an arc. These discharges usually occur at low pressures, but pressures up to atmospheric are possible. A hollow cathode reactor is a specific and common design type in which an increased current is available.9 h. Radiofrequency Discharges and Microwave Discharges. Radiofrequency discharges (RF discharges) are employed in a frequency range of 1-100 MHz, with a typical frequency being 13.56 MHz. At lower frequencies, a DC discharge case, accelerated ions moving towards electrodes producing secondary electrons, with alternating electrodes are observed. At higher frequencies, however, ions and electrons no longer reach the electrode surface during acceleration. RF plasmas are divided into two categories due to the difference in power coupling: capacitively coupled discharges, E-discharges, and inductively coupled discharges, H-discharges.9 Capacitively coupled RF discharges are operated at pressures from 1-103 Pa and have a setup consisting of a generator combined with an impedance matching network (providing peak efficiency and minimum refection) and the reactor with electrodes. The electrodes are surrounded by sheath regions and a bulk plasma in between them. Two forms of plasma are observed; the a mode (avalanching electrons) with lower current and positive voltage-current (U-I) characteristics whereas the y mode (ion impact secondary electrons) is the opposite. These modes show differences in the sheath regions and glow intensity, the y mode being more intense. A characteristic for these RF plasmas is the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. I 32 self-bias, negative DC potential between plasma and powered electrode due to a coupling capacitor and different electrode size.9 Inductively coupled RF discharges are produced by an electric field generated by a transformer from a RF current in a conductor in which the electrons are accelerated. These plasmas produce high election densities at low ion energies. They can operate in a static magnetic field (helicon discharge) and the typical frequency is equal to that of the other RF plasma type. This special type of discharge uses special antennae and special mechanisms are provided (wave dampening explained by collision theory). Neutral loop discharge is an interesting technology variation of inductively coupled plasmas at low pressures.9 Microwave discharges are characterized by a frequency o f 2.45 GHz and a short period of the exciting microwave field. The power absorption is dependent on the electron-neutral collision frequency, therefore on the gas pressure and nature of the gas. These plasmas can operate at pressures up to atmospheric and microwave power absorbed in the skin sheath, maximum penetration o f microwaves, is transferred as energy into the plasma. They consist o f a MW power supply, a circulator (protect power supply from reflected power), an applicator (optimize energy transfer and minimize reflection), and the plasma load. Three basic types o f reactors can be discriminated: discharges in closed, open, or resonance structures. Surface wave plasmas can also be generated by microwaves; the surface wave spreading along the boundary layer of the plasma column and dielectric vessel, the plasma absorbing the wave energy. Generation of low collision plasmas at low pressures and therefore low power absorption is aided by Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 33 the use of magnetic fields. Overall, plasmas can also be generated by particle beams, usually electron beams, and by using pulsed direct current or microwave sources.11 i. Silent and Corona Discharges. The discharges dealt with in this thesis are strictly speaking silent discharges but they are generally referred to as glow or silent glow discharges because of the use o f an alternating current they show a lot of behavior characteristics of a glow discharge, e.g. homogeneity. Silent and corona discharges are special types of RF discharges at elevated (atmospheric) pressures. Silent discharges, also called dielectric barrier discharges, are usually operated at atmospheric pressure and have an insulating layer on one or both electrodes or an insulating substance in the discharge gap. Surface discharges are similar in nature and employ some dielectric in some way or another but differently from silent discharges. Usually, these materials have high dielectric strengths such as quartz, glass or aluminum oxide and act as a plasma stabilizer. In conventional plasmas at atmospheric pressures, sparks cause local heating and non-uniformity, but in silent discharges with the presence o f the dielectric as a capacitor in series with the plasma, microdischarges o f short pulses result which are statistically spread in space and time as the potential across the gas gap reaches the breakdown voltage. The ions generated by the discharge travel towards the dielectric and are stored at its surface; this space charge accumulates and generates a reverse electric field yielding discharge termination. If a sinusoidal voltage potential is used, this process can become continuous because the afore-mentioned process starts over and over again.13 Reproduced with permission o f the copyright owner. Further reproduction prohibited without permission. 34 The microdischarges preferentially use the channels o f previous microdischarges again upon the ignition o f the process by applying reversed voltage if an alternating current o f sinusoid shape is used. A sinusoid waveform, which can be altered in form, amplitude and frequency, represents one way to control this type o f discharge. Other ways involve changing the width o f the gas gap and the dielectric thickness or even the material itself; the nature o f the gas in the gap and the flow rate are important parameters as well. Homogeneous glow discharges using a dielectric have been reported by employing sinusoidal voltage of appropriate frequency. In these discharges, electrons generally activate the background gas, which in turn with or without electrons produce free radicals, which induce the desired chemical reactions.9 Corona discharges are comparatively low power self-sustained discharges at atmospheric pressure. The corona occurs on a sharp edge such as a needle or a thin wire in the presence of strong electric fields caused by the strong curvature of one o f the electrodes yielding ionization of neutral gas molecules.8 Therefore, primary ionization happens in the corona region constituting only a small fraction o f the whole plasma volume. A current flow is established with increasing voltage in a region from the corona onset till the potential o f spark breakdown. A faint visual glow around the electrode (positive corona) or concentrated in certain areas (negative corona) goes along with luminous streamers towards the other electrode. In direct current coronas, frequently burst coronas are observed with very regular current pulses (Trichel pulses) due to the periodic build-up and removal o f space charges thereby altering the magnitude of ionization processes. Direct current corona discharges can be discerned from pulsed corona discharges.9 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 35 Pulsed coronas are used to yield a high concentration o f free radicals by positive streamers caused by fast-rising high-voltage pulses. At high voltages, streamers bridge the whole gap, and positive streamers spread from the anode towards the cathode at very high speeds. High-energy electrons constitute the streamer head and are very efficient at excitation and ionization. In the streamer channel, the neutral gas temperature is very close to ambient. Short pulses (smaller than 1 ps) avoid thermalization and breakdown. Radical formation can be optimized by pulse times close to streamer transit times, therefore requiring advanced high voltage switching technology.9 2. Microwave Heating. a. Microwave Applications. Microwaves cover the region of the electromagnetic spectrum from wavelengths between 1 mm and 30 cm, wavelengths longer than visible light; equivalent frequencies are 1-300 GHz.14 The wavelength region between 1 and 25 cm is broadly used for RADAR transmission and for telecommunication purposes. Due to these requirements, two wavelengths in the microwave section of the spectrum have been designated for domestic and industrial use, 12.2 cm (2.45 GHz), the frequency used for domestic microwave appliances, and 33.3 cm (900 Hz). Other frequencies can only be used when properly shielded against radiation loss. Microwaves are used in a variety of fields such as for RADAR but most everyone is familiar with the household item, the microwave oven. For the development o f RADAR, a magnetron was developed during World War II at the University o f Birmingham and later produced on an industrial scale with help of the US. Microwaves can heat water rapidly. In the 1950s, commercial and domestic Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 36 appliances for heating and cooking food started to emerge. The widespread use o f these ovens started in the 1970s due to efficient Japanese technology transfer and worldwide marketing. 15 b. Microwave Dielectric Heating. Microwaves have been used extensively in industry and research. Microwave heating has to be discerned from the aforementioned microwave plasmas where gas phase reactions take place at sometimes very high temperatures. Microwave heating, also described as microwave dielectric heating, involves the capability o f some solids and liquids to convert electromagnetic energy into heat and thereby drive or facilitate chemical reactions. The properties o f the molecules themselves determine the extent of the in-situ energy conversion of microwaves. Therefore, the properties of the material can be controlled and fine-tuned in order to lead to the desired selectivity and in some cases to the desired products. 15 Solids and liquids behave differently from gases in respect to microwave spectroscopy. Gases show distinct sharp lines due to transitions between the molecule’s quantized rotational states; in liquids and solids these lines are too broad to be observed because molecules in these cannot generally rotate independently. In these phases, microwave dielectric loss heating effects are important. If high frequency electromagnetic waves are applied to these materials, the energy is transferred and these materials heat by the force of an electric field on charged particles. If the particles are able to move freely in a material, only a current is induced. If, on the other hand, the motion of charge carriers is limited by bonds or attachment in any form they will advance Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 37 until a counteracting force balances them resulting in dielectric polarization. The microwave heating effects by conduction and dielectric polarization depend on the applied frequency and power. 15 An electric field is able to polarize charges in a substance, and the polarization of these materials is unable to follow the quick reversals o f the electric field. This is one of the origins of microwave dielectric heating. The total polarization a , o f a molecule is the sum o f the individual polarization: electronic ( a e), atomic ( a a), dipolar ( a d), and interfacial polarization (a , ). The influence o f an oscillating field on the components of the polarization depends on the timescale of the orientation and disorientation events compared to the radiation frequency. The timescales for a eand a tt are much larger than the microwave frequency, whereas a d and maybe a, have timescales similar to the microwave frequencies thereby contributing to dielectric heating while the other two do not. 15 The dipolar polarization is caused by the dipole moment, which is due to electronegativity differences in the specific molecule. At low frequency, the electric field oscillates slower than the response time of the dipoles, therefore the dipoles rotate into alignment with a very small energy transfer and the temperature remains almost constant. If the frequency is high, the dipoles do not rotate resulting in no net energy transfer because the field oscillation is higher than the response time. In the microwave field region, the response time is very similar to the oscillation time of the electric field. Due to the field action the dipoles rotate but there is a time lag between the resulting polarization and the field oscillation; as a result the dipoles absoib electric field energy and heat up. 15 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 38 The dielectric constants', at a maximum at low frequencies (maximum energy storage) and the dielectric heat losss", efficiency o f conversion of electromagnetic radiation into heat, characterize the dielectric properties o f a material. Both values are frequency dependent with e" having a maximum as s ' falls. Two definitions using s ' and s" are listed in Equation 13. The ability of a material to convert electromagnetic energy into heat energy at a given temperature and frequency is defined as tan <5 .1S s is the complex permittivity which describes the dielectric behavior of a material. 16 tan J = — ;s = s'-y 's” (13) s' The amount o f time the dipole of a molecule needs for polarization and depolarization is determined by the relaxation time constant. A specific polarization mechanism is then important when this value is comparable to the inverse excitation frequency. Water shows dielectric loss over a wide frequency range with a maximum around 20 GHz. The normal frequency for domestic and most industrial uses is at 2.45 GHz, lower than the maximum. The reason for this is that at lower frequencies the penetration of the material to be heated is better expressed by the penetration depth. This penetration depth Dp is defined by Equation 14, where ^ i s the corresponding wavelength of the microwave radiation. 16 Dp ac Aq-Js T s " (14) Debye’s equations for dielectrics are valid for solids and liquids. In liquids, dipoles have no preferential direction and are constantly changing due to thermal agitation. Relaxation according to Debye is due to frictional forces in the medium. In solids, a molecule interacts variably with its neighbors, and therefore its dipole has many Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 39 equilibrium positions separated by potential barriers. Many polar organic solids are found to have only small dielectric losses, but in most other cases for organic solids considerable dielectric absorption is observed at room temperature and decreases with increasing temperature. 13 Interfacial polarization is observed in inhomogeneous materials. These materials are suspensions of conducting particles in a non-conducting matrix showing a frequency dependent dielectric constant. The dielectric loss is caused by the increase o f charges between interfaces and identified as the Maxwell-Wagner effect. The importance o f this effect inside the microwave region is somewhat poorly characterized. The basis for this uncertainty is that it is not clear that absorption based around a frequency of 107 Hz tails into the microwave frequency region. Models by the Wagner theory were developed for these phenomena including dielectric constant and conductivity to yield a dielectric loss value. 15 Conduction losses are an extension of interfacial polarization. Here, the amount of the conducing phase dispersed in a non-conducting matrix is increased to the point where the interaction between the individual conducting areas comes into play. The Wagner theory is expanded to fit these interactions. The Maxwell-Wagner effect in this case is described as a combination of differing dielectric constants and conductivities resulting in the definition of the complex permittivity which is similar to the one defined in Equation 13 but includes the conductivity contributions. The real part is covered by the simple Debye theory, whereas the complex part is characterized by a single relaxation time response and an extra conductive part depending on the DC conductivity itself. In some instances, the conductive losses can be bigger than the dipolar relaxation effects Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 40 especially for highly conductive solids and liquids. The temperature increase due to the action o f microwave radiation depend on the specific heat capacity o f the substance and the nns field intensity, but the mathematic determination o f the increases becomes rather complex due to the temperature dependency o f the physical properties of the solid or liquid. 15 In the case of water, its high polarity makes it useful for microwave heating. The dielectric loss depends heavily on the form of water. For example, there is a difference of three to four orders of magnitude between the absorption in liquid (free) water and ice. Aqueous substances show a number of relaxations corresponding to the different states o f water, bound water is usually considered to be intermediate, somewhere between ice and free water. Furthermore, in mixtures the dielectric loss usually drops below the value o f pure water; in saline solutions significant changes have been observed caused by the appearance o f bound water. 16 C. GOALS IN THIS DISSERTATION. This research consists of two major parts which may lead to new fundamental information and industrially useful applications for discharge plasmas and microwave heating. The first part involves the decomposition of two Freon species and carbon tetrafluoride using these methods. Freon 21 (Dichlorofluoromethane) and Freon 142B (1chloro-1,1 -difluoroethane), two volatile organic compounds (VOCs), were tested in a plasma discharge system with regards to how changes in applied input voltage, input concentration, flow rate, and additives such as oxygen and water affect conversions. Optical emission spectroscopy in conjunction with mass spectrometry was used to Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 41 suggest a reaction mechanism in these systems. Carbon tetrafluoride was treated by a discharge plasma and by microwave heating. The conversion was optimized using different additives and catalysts, hi the case o f the microwave heating, a reaction mechanism was suggested based on mass spectrometric data. In the second part, hydrogen production from different sources was studied. Hydrogen was produced from water in a balance of nitrogen (nitrogen is the diluent gas), and from water in a balance o f methane and methane alone. The influence o f these different gas mixtures on hydrogen production was investigated. How the conversion versus hydrogen changed when flow rate and applied voltage were varied and admixtures were added was also studied. In the case of just using methane, the influence o f different coated metal electrodes was investigated, and the carbon species deposited on the electrode during the reactions were identified by mass spectrometry and surface methods such as X-ray diffraction and X-Ray Photoelectron Spectroscopy (XPS). Finally, a small chapter is included in the appendix which describes some of the other research projects that I have worked on during my time at UConn. This includes a study o f the decomposition of 1-butene using titania as a photocatalyst, the preparation of alloy materials for incorporation as anode materials in secondary batteries and the synthesis o f MCM-41 materials as catalysts for the synthesis of o-diphenylamine. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 42 D. REFERENCES. 1) A Beginner’s Guide to the UN Framework Convention on Climate Change, http://unfccc.int/resoiirce/beeinnerJitml. December 2002. 2) United Nations Framework Convention on Climate Change, http://unfccc.int/. December 2002. 3) The UNEP Ozone Secretariat, http://www.unep.ore/ozone/. December 2002. 4) United States Department o f Energy Hydrogen Information Network, http://www.eren.doe.eov/hvdroeen/. December 2002. 5) Coalition fo r Plasma Science: Plasma Applications, http://www.plasmacoalition.ore/applications.htm. December 2002. 6) Von Engel, A. Electric Plasmas: Their Nature and Uses, International Publication Service, New Yoric, 1983. 7) Dresvin, S. V. Physics and Technology o f Low-Temperature Plasmas, The Iowa State University Press, Ames, LA, 1977. 8) Lieberman, M. A.; Lichtenberg, A. J. Principles o f Plasma Discharges and Materials Processing, John Wiley & Sons, New Yoric, 1994. 9) Hippier, R.; Pfau, S.; Schmidt, M.; Schoenbach, K. H.; Low Temperature Plasma Physics, Wiley-VCH Verlag, Berlin, Germany, 2001. 10)Nozaki, T.; Miyasaki, Y.; Unno, Y.; Okazaki, K. 15thInternational Symposium on Plasma Science, Orleans, France, July 2001, Vol.4, 1591-1596. U)Howatson, A. M. An Introduction to Gas Discharges, Pergamon Press, Oxford, England, 1976 (Second Edition). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 43 12)Kunhardt, E. E.; Luessen, L. H.; Electrical Breakdown and Discharges in Plasma: Part A: Fundamental Processes and Breakdown, Plenum Press, New York, 1983. 13)Vercammen, K. L. L.; Berezin, A. A.; Lox, F.; Chang, J.-S. J. Adv. Qxid. Technol., 1997,2(2). 312-329. 14) Imagine the Universe, NASA Laboratory fo r High Energy Astrophysics/ Goddard Space Center, http://imagine.gsfc.nasa.gov/docs/dict ip.html#microwave. 15)Mingos, D. M. P.; Baghurst, D. R. Chem. Soc. Rev., 1991,20. 1-47. 16)Thuery, J. In Microwaves: Industrial, Scientific and Medical Applications: Grant, E.H., Ed.; Artech House, Inc. Norwood, MA. 1992. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER II. DESTRUCTION OF ENVIRONMENTAL POLLUTANTS BY MICROWAVE HEATING AND DISCHARGE PLASMAS A. DESTRUCTION OF FREON SPECIES BY DISCHARGE PLASMA. 1. Introduction. Freons, also known as ChloroFluoroCarbons (CFCs), are a class of halocarbons containing chlorine and fluorine. A specific naming system was developed due to their importance in industry. This naming system assigns each compound a 3-digit number, which is written after the word Freon; sometimes this number is written after the word CFC or HCFC. The last digit is the number of fluorine atoms. The next to last digit is one plus the number o f hydrogen atoms. The first digit is the number of carbon atoms minus one. If this digit is zero then it is not written. The balance of atoms required to make the valency of carbon atoms up to 4 is accounted for by chlorine atoms. The letters are given to compounds with the same sum formula with the most symmetric isomer having no letter; for the next symmetric isomer an “a” is added after the Freon name. 1 Freons are volatile organic compounds (VOCs) that were widely used in industry but their use is decreasing and is better regulated. These compounds have been used as propellants in aerosol sprays, refrigerants, solvents in the electronic industry, as well as foam blowing agents. Freons are safe to use because they are chemically very inert. Atmospheric pollutants do not react with Freons because of the inertness o f the Freons, and therefore, Freons get slowly transported into the stratosphere over many years where these compounds undergo photochemical reactions.2 44 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 45 These photochemical reactions yield species that are involved in the destruction o f the stratospheric ozone layer. The CFC itself is not responsible for the ozone loss in the stratosphere, but its decay products from UV photolysis certainly are. Most of the photochemically produced chlorine radicals end up as hydrogen chloride or chlorine nitrate, the so called “reservoir species” .3,4 These species decompose providing small amounts of atomic chlorine and chlorine monoxide, which act as catalysts for the destruction of ozone. Each chlorine atom introduced into the stratosphere destroys thousands and thousands o f ozone molecules before removal, the effect is even more pronounced for bromine, which is 1 0 -1 0 0 times more destructive than chlorine and does not have a “reservoir”, but is less abundant in the stratosphere. The chemistry o f ozone destruction is very complicated.5 The Freons used in these experiments are actually hydrochorofluorocarbons (HCFCs), Freon 21 (Dichlorofluoromethane) and Freon 142B (l-chloro-1,1- difluoroethane). HCFCs contain a certain amount of hydrogen atoms in the molecule. These molecules replaced CFCs as refrigerants; a good example is HCFC-22 replacing CFC-12. HCFCs are more readily attacked by the hydroxyl radical in the lower atmosphere because of the presence of a hydrogen atom in the molecule. This reaction significantly reduces the amount of HCFCs reaching the stratosphere. The potential of HCFCs to destroy ozone is just 1 to 10% of that of CFC>12, a standard used to measure the ozone depletion potential o f compounds, and is even smaller for hydrofluorocarbons (HFCs) .6 The short and long term effects of these species are disputed.7 Studies done by Hayman et al.8 indicate that these compounds have a very low effect on the ground level Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 46 ozone formation, but their high stability results in a high global warming potential and a non-zero contribution to the stratospheric ozone depletion. The first international conference concerning the ozone depletion was the 1985 Vienna Convention for the Protection o f the Ozone Layer. In 1987, the primary international agreement providing control measures on the production and consumption of ozone depleting species was signed in Montreal, the Montreal Protocol. As o f December 2002 184 countries have ratified the Protocol. The Protocol was amended several times in the following decade, the Beijing Amendment in 1999 being the latest; this amendment is in effect and ratified by 45 countries. In this last amendment, bromochloromethane was included on the list o f the ozone depleting substances, and the production o f HCFCs will be limited by 2004 in industrialized countries 9 The parties meet annually to check its operation. The depletion of the ozone layer and other environmental pollution problems such as acid rain and global warming led to increased research in destroying environmentally harmful compounds such as VOCs and CFCs. Traditionally, thermal methods, such as incineration and thermal plasmas, were used to achieve “cleaner** products. 1011 These methods turned out to be associated with high installation costs and toxic exhaust gases. Catalytic oxidation and adsorption are other removal methods but similar to incineration yield to problems due to deactivation over time and poisoning of the catalyst. 11 Takita et al. 12 found promising results for the decomposition of CCI2F2 and CH2FCF3 with water and oxygen playing a crucial role therein. Furthermore, electrochemical methods proved to be highly efficient in the destruction of CFC-12 and 13.13,14 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 47 In recent times, the focus for the removal o f environmental pollutants such as Freons shilled to the use of non-thermal plasmas. In this case, non-thermal plasmas need to be distinguished between microwave (MW) plasmas and electronic plasmas. MW plasmas are used in the power regime o f a hundred kW to several MW and at variable low pulse rates under low pressure. Gritsinin et al.ls studied the effect of a MW discharge on CFCs. Effective destruction o f CCI2F2 takes place and fluorine rich species such as CF4 and CF3CI were produced. An appropriate mechanism was postulated involving chemical transformations and photodissociation. Aleksandrov et al. 16 demonstrated that a plasma catalytic cycle employing electrons and negative ions in the case o f CCU decomposition in humid air in a MW discharge afterglow reduces energy expense. Jasinski et al. 17 tested a microwave torch plasma reactor and achieved almost complete destruction of Freon 11 at medium power levels (around 400 W), atmospheric pressure and high Freon concentrations in nitrogen and proposed a prototype system. Electronic plasmas can be categorized depending on the frequency used to drive the discharge and the type of discharge itself. RF discharge plasmas have to be distinguished from silent, glow and surface discharge plasmas and other discharge plasma types. Radio frequency (RF) plasmas favor electron impact dissociation over electron attachment dissociation as Stoffels et al. 18 observed for the dissociation of CCI2F2 (up to 90%) at low pressures (0-400 mTorr) and high power input. Wang et al. 19 showed for the same compound that high decomposition rates (up to 94%) and high selectivity to CH4 and C2H2 (up to 80%) are achievable in a low pressure hydrogen-argon RF plasma system. A dielectric barrier (silent) discharge at low frequency was used by Korzekwa et Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 48 al.20 to decompose a mixture o f volatile organic compounds (VOCs) including Freons at atmospheric pressure. Oda et al.21 investigated the surface discharge o f CFC-113 in air at atmospheric pressure and showed high concentrations of hydrochloric acid, carbon dioxide, and nitrogen suboxide at higher decomposition rates, and intermediate halogenated carbons and hydrocarbons at low rates as products. The same authors established that there are no power efficiency differences between a ceramic, a coil type, used in the previous study, and a coaxial reactor (yielding a silent discharge), but there are some minor differences with respect to peak voltage.22 For the same compound, Freon 113, Akhmedzhanov et al.23 found trends similar to Stoffels’ 18 work favoring electron impact and collisions at high Freon content and dissociative attachment at lower content. For laser sparks and slipping surface discharge treatment o f CF2CI2 a reaction scheme similar to the one for a MW plasma by Gritsinin15 was proposed by Akhvlediani et al.24 but unlike Gritsinin’s 15 mechanism, photodissociation was not part of the mechanism. In our case the glow plasma decomposition of HCFC-21 and 142B was investigated and a mechanism for their decompositions was proposed using optical emission spectroscopy and mass spectrometry results. Furthermore, the influences of factors such as input voltage and flow rate was studied, and power consumption and efficiencies are discussed in this part of the thesis. The reactor employed in this case is a coaxial reactor favoring silent or dielectric barrier discharges.22 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 49 2. Experimental Section. a. Preparation o f Gas Mixtures. A 1.0% mixture of Freon 21 in helium was prepared by mixing 1.S psi Freon 21 (Dichlorofluoromethane, 98 % purchased from Aldrich Chemicals) with Helium UHP (Ultra High Purity, obtained from Connecticut Airgas) to achieve a total pressure o f ISO psi. The 0.5% mixture of Freon 142 B in helium was prepared accordingly. A 10% mixture o f oxygen in helium was used as obtained from Connecticut Airgas. b. Plasma Reactor. The reactions involving Freon destruction were carried out in a Plasma And Catalysis integrated Technologies (PACT) tubular type reactor. The reactor consisted o f three parts. The inner electrode was an iron rod with a diameter o f 9.50 nun. A Pyrex glass tube of 9.91mm inner and 11.89 mm outer diameter for earlier experiments and later a quartz tube of the same dimensions, which served as a dielectric, were used to separate the inner from the outer electrode. The quartz tube was used because it is transparent in the UV region, which is crucial in the use of optical emission spectroscopy. The outer electrode was wrapped around the glass tube and was subsequently wrapped with copper wire to achieve a close coverage o f the glass tube in order to achieve uniform plasma. Aluminum and copper foil of 19.7 cm length were used for this purpose. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 50 c. Experimental Setup and Parameters. A schematic o f the experimental setup used is shown in Figure II-1. The voltage to operate the tubular plasma reactor was produced by a Japan-Inter UHV-10 AC high voltage generator with an operating frequency o f between 8.0 and 8.1 kHz. Input voltage and input current values were measured by using a Tektronix 6015A High Voltage Probe connected to a Yokogawa digital oscilloscope DL1520; in some cases a HV 15 HF 15 kV Oscilloscope Gas in High Voltage Probe Ground Tubular PACT Reactor Voltage probe Figure II-l. Schematic of the experimental setup for plasma generation and monitoring the voltage and current conditions. DC and Peak-to-Peak AC probe was used. The measurement o f the input voltage required an additional voltage probe. The voltage measured across a 100 O resistor (Rs) placed in series was used to monitor the input current. The other two resistors Ri (100kD, 225 W), and R2 (lOOkH, 225 W), parallel to the tubular reactor, were used to divide the current from the AC power supply. AC Voltage values between 1.000 and 5.440 kV were applied to the electrodes to produce plasmas; these voltages are peak-to peak voltages. The Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. reactions were carried out at atmospheric pressure and room temperature. The reaction gases were introduced by mixing Freon in helium and oxygen in helium (moisture added if needed) online to achieve a 0.5% concentration of Freon and 5 % of oxygen, respectively while using flow rates o f 20 up to 80 mL/min. In later set-ups two additions were done to prevent damage o f the analysis apparatus. A water bubbler was installed to trap most o f the produced hydrogen fluoride and hydrogen chloride, and a subsequent water trap was added to prevent too much water from reaching the GC thereby damaging the instrument. d. Product Analysis. The reaction products were analyzed using a Hewlett Packard 5890 Series II Gas Chromatograph (GC) employing a Thermal Conductivity Detector (TCD). The columns used were an Alltech Poropak N 80/100 column ( 6 ft x 1/8 in. x 0.085 in. inner diameter stainless steel) for the Freon analysis, and a Supelco 45/60 Carboxen 1000 (5 ft. x 1/8 in. stainless steel) for the carbon monoxide/ carbon dioxide analysis. To measure the disappearance o f the Freon peak the Poropak N column was used. The Carboxen column was used to measure the yield of CO2 and compared with a calibration done previously. The ratio of CO to CO2 is the ratio o f the corresponding areas. The calibration of the columns for CO2 was done with gas mixtures prepared beforehand in the lab. Mass spectroscopic studies were carried out using a MKS-UTI PPT quadrupole residual gas analyzer mass spectrometer with a Faraday cup detector and a variable highpressure sampling manifold. The m/z range achievable with this instrument lies between Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 52 0 and 200 m/z units. The sampling was done on-line, and the monitored spectra were almost in real time. e. Plasma Diagnostics. Diagnosis of the species involved in the plasma reactions was done using a 270 M SPEX optical emission spectrometer with a charge-coupled device (CCD), which was cooled by liquid nitrogen. Emission spectra of the Freon plasmas were recorded in the wavelength range from 200 to 900 nm. A slightly modified setup was used for conducting these experiments. A quartz tube had to be used in these experiments as the reaction tube because it yielded better optical emission data. Three holes were drilled into the aluminum foil, one in the front part, one in the middle, and one at the end o f the plasma zone so as to collect light being emitted under plasma conditions. The emitted light from the plasma was collected and transported by a fiber optic cable to the monochromator of the instrument. 3. Results. a. Voltage and Current Characterization o f the Plasma. Typical voltage data from the oscilloscope are shown in Figure II-2. Data for the closed circuit representing plasma conditions are shown in Figure II-2b. As the voltage was switched on with the reactor in the circuit a certain voltage was necessary to achieve plasma conditions, which could be correlated to conversion. In the case of Freon 21 a voltage of over 1.60 kV was needed at a Freon 21 concentration of 0.5% and o f over 1.020 kV at a Freon 21 concentration of 0.1 %. The voltage reading on the oscilloscope Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 53 showed that during the initiation period the shape o f the voltage curve kept building up until reaching a steady value after about 30 s. The final shape o f the resulting curve for the closed circuit was closely related to the shape o f the input voltage curve. The power 6 4 2 0 -2 -4 -6 0 100 0 100 200 300 400 soo 400 SOO Time(microsec) 1500 1000 500 (b) -500 -1000 •1500 200 300 Tim e(m icrosec) F igure II-2. (a) Current and (b) voltage vs. time wave form for the closed circuit plasma conditions for a 0.5% Freon 21 and 5.0% oxygen in He mixture in an Fe PACT reactor. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 54 supply did not produce a perfect sine wave as observed in the resulting voltage waveform. The current wave, measured as a voltage wave, which was averaged out, was in phase with respect to the voltage wave, see Figure II-2a. The current wave (averaged) went down relatively sharply from the maximum, wiggled around the zero value, and then dropped to a minimum, and continued in this fashion. There was a slight difference o f about 5-10 ns between the current minima and maxima (slightly ahead) and the voltage minima and maxima which might be due to the measurement setup or due to the plasma itself. The non-averaged current curve showed a different behavior. The waveform oscillated and showed maxima and minima where the voltage wave had its maxima and minima. Furthermore, depending on conditions, up to two current maxima were observed between correlated voltage maxima. b. Destruction ofFreons. For the study on the removal ofFreons just one reactor setup was used. This setup consisted of an iron inner electrode and either copper or aluminum as the outer electrode. Effects of input voltage, flow rate, and the addition of water and oxygen were investigated as the main purpose o f this work. The focus o f these studies were the reactions with 0.5% Freon 21 as the probe gas, but further investigations included reactions with 0.1% Freon 21 and 0.25% Freon 142 B. The reaction gas was mixed online in a 1:1 ratio with respect to flow rate with helium, oxygen in helium or oxygen and water in helium depending on the objectives of the reaction study. Before the reaction, an equilibrium time of 1.5-2 h was preferred and Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 55 checked by a constant area count on the gas chromatograph. The reaction itself usually took about 2 h. An equilibrium state was additionally desired at the end o f the reaction. Consequently, the run was sometimes prolonged accordingly to reach equilibrium or aborted if no Freon peak was present. A similar trend was seen for the data gathered for the three different conditions at a constant flow rate. A sharp increase in conversion was observed over the first 20-40 min of the reaction. The conversion then leveled off for the remaining time until reaching equilibrium. Differences in the equilibrium conversion value and furthermore differences in the time in which equilibrium was reached were identified for the three reaction environments. When Freon 21 was mixed with oxygen in helium, the characteristic behavior was observed with respect to conversion as seen in Figure II-3. After the initial equilibration period, the conversion increased to 60 to 70% conversion over the first 40 min and then leveled off at about 95% after about 120-140 min. The conversion of Freon 21 above a certain threshold input voltage was independent of the applied input voltage and followed the same trend from this low voltage up to higher voltages if the conversion data for several different voltages were plotted against time (Figure II-4). The conversion results turned out slightly different when just Freon was used at a concentration o f 0.5% in helium. The conversion followed the same trend over the first phase o f the reaction. The difference was that some dependence o f the final conversion on the applied input voltage was observed. The upper limit for the conversion with 95% Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 56 100-1 80 - 60 - 40 - 20 - 0 -20 20 40 100 80 60 120 Time in min Figure D-3. Destruction o f 0.5% Freon 21 in 5.0% oxygen and He at 19.4 mL/min and Vp-p= 2.560 kV. 100 80 e ▼ O B □ * £ e 0 1 60 -. D ▼ V 40 - V= 2.280 kV V= 2.680 kV V= 1.680 kV V= 3.160 kV V -4.040 kV V= 4.480 kV J 20 H o e -i-O - •20 I I — i— i 20 40 60 80 100 120 140 Time in min Figure II-4. Destruction o f 0.5% Freon 21 in 5.0% oxygen and He at 20 mL/min and variable input voltage. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 57 was the same as in the previous case (Figure II-5). The lower limit was 75% for applied input voltages o f 1.700 kV and higher with the same barrier input voltage o f 1.600 kV. 100 80 * 60 £ o e t o o 40 □ 20 e— •20 o 0 - • V= 5.040 kV ■ V- 4.240 kV A V= 3.360 kV V V= 2.680 kV ♦ V= 2.080 kV O V= 1.680 kV □ V= 1.560 kV - I ------------ 1—- -------1------------1— 20 40 60 80 — i--------------------- 1— 100 120 140 Time in min Figure 0 -5 . Destruction o f 0.5% Freon 21 in He at 20 mL/min and variable input voltage. In addition, a different Freon 21 concentration o f 0.1% in helium was used as well (Figure II-6). The conversion was similar to the one with a 0.5% concentration of Freon 21. The only major difference regarding these data was that conversion of 100% was attainable, and the lower limit of conversion above the barrier voltage was 80%. The barrier input voltage o f 1.020 kV for a lower Freon 21 concentration in this case was clearly lower than that for the higher Freon concentration. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 58 100 n O* Q* 80 - £ 60 > 40 e o • 0 □ A ♦ A O ▼ □ O o V= 1.560 kV V= 1.020 kV V= 1.060 kV V= 2.200 kV V= 2.560 kV V= 3.040 kV V= 3.400 kV V= 3.960 kV □ □ 20 0 o o o - * ▼“i— -cf t i 0 -Id—i— -20 a 20 - i— 40 — i— — i— 60 80 — i— 100 120 140 Time in min Figure II-6. Destruction o f 0.1 % Freon 21 in He at 20 mL/min and variable input voltage. The conversion of Freon 21 seemed unaffected by the addition of water to a mixture of 0.5% Freon and 5 % oxygen (Figure II-7.). The conversion was as high as in the other cases at the same Freon concentration. Furthermore, the barrier input voltage of 1.60 kV had as similar value as in the before-mentioned cases. In this case, the effect of the flow rate on conversion was studied (Figure II-8). The expected dependence o f the conversion on flow rate was observed. At low flow rates, the conversion was high (88% at 15 mL/min), decreased to 78% at 40 mL/min and was even lower at high flow rates (47% at 100 mL/min). An interesting observation was that at 100 mL/min the conversion increased if water was omitted from the feed. The flow rate also affected CO2 production. At a higher voltage (3.440 kV), conversion to CO2 decreased by 30% when the flow rate was changed from 20 to 40 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 59 100 9 * 8 oO 80 OKo * £ 60 - o C 5 □° o □O o ♦ 40 - • ■ ♦ 0 □ A 0 0 o o 20 a1 - V= 5.520 kV V= 5.040 kV V= 4.480 kV V= 4.000 kV V= 3.550 kV V= 3.040 kV V= 2.480 kV V= 2.160 kV O 3C O 0 - -20 20 40 60 80 100 120 Time in min Figure D-7. Destruction o f 0.5% Freon 21 in 5.0% oxygen/water and He at 20 mL/min and variable input voltage. mL/min and by 42% if changed to 60 mL/min. The selectivity to CO2 was a parameter that was more directly influenced by the admixture of oxygen or oxygen and water to the Freon species (Figure II-9). In the case of just the Freon species, merely traces o f oxygen were present in the reaction system and only small amounts of CO2 were formed. If oxygen was introduced into the reaction system, the CO2 production increased by 90% compared to the oxygen deficient case and the admixture of water, additional to oxygen, increased this value by another 25%. Moreover, CO2 was the main carbon oxide formed in these reactions as seen at a CO/CO2 ratio o f 0.05 to 0.07, and by earlier data that indicated that CO2 showed selectivity up to 95% depending on input voltage. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 60 100 0 80 a A i * £ c 0 ▲ v * * 60 g 1 40 : o o • □ A V O ■ r* " T' 20 -20 — i— 40 60 — 20 cc/min 15 cc/min 30 cc/min 40 cc/min 100 cc/min 100 cc/min (2) i— 120 100 80 140 Tim* in min Figure II-8. Destruction o f 0.5% Freon 21 in 5.0% oxygen/water and He at Vp-p= 2.440 kV mL/min and variable flow rates. 1e+6 9e+5 8e+5 ^ 7e+5 £ 6e4-5 a « 5e+5 4e+5 - o" u 39+5 □ • 2e+5 A 1e+5 — i— -20 0 20 just Freon 21 Freon+oxygen Freon+oxygen+water — t— — i— — i— 40 60 80 100 Tim* (min) Figure II-9. CO2 production from o f 0.5% Freon 21 in 5.0% oxygen/water and He at Vp-p= 3.440 kV mL/min and 20 mL/min. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 61 100 80 * e £ (a) 60 - I§ s I a • 20 2 3 4 Input Voltage in kV 6000 -| 5000 i a e (b) 4000 - I i 5 c 8 c 8 3000 - 2000 - CM O O 1000 - 00 1 2 3 4 S 6 Input Voltage in kV Figure 11-10. (a) Destruction o f 0.5% Freon 142B in 5.0% oxygen/water and He and (b) CO2 production at variable voltage and 20 mL/min. A few experiments were conducted with a different Freon, Freon 142B. However, only qualitative conclusions could be drawn from results for this Freon (Figure II-10). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 62 The conversion increased with increasing input voltage, but the increase seemed more linear than in the case where Freon 21 was used, where the conversion was more exponential with oxygen and water admixed. This could also be a result o f the lack of enough data points. The selectivity to CO2 showed the same trend as the conversion and reached almost 100% selectivity. Studies at other conditions, such as Freon by itself or Freon with oxygen admixture, were not carried out. c. Optical Emission Studies. Optical emission studies were carried out to gain insight into the mechanism of different plasma reactions involving Freons. In these studies, several different mixtures of the two Freons, Freon 21 and Freon 142 B, with or without oxygen and with and without water in helium and helium by itself were analyzed using this method. Their respective spectra were taken at total flow rates o f 20 mL/min and identical voltage conditions at the maximum possible voltage o f about 5.5 kV. Three sets o f data were taken for these two Freons at all three observation holes in the plasma zone. Additionally, background spectra for pure helium were collected at all three holes for comparison purposes. The background spectrum included several characteristic lines as shown in Figure II-l la. The strongest line by far was the one at 706.5 nm, representing helium.25 Other strong characteristic helium lines were observed at 587.6 nm, 667.6 nm, and 728.2 nm, as well as a line of He+ at 656.3 nm.25 Lines o f oxygen radicals (O’) and various nitrogen lines were present as well. Strong lines of oxygen radicals were at 777.2 nm and 844.64 nm.25 The nitrogen lines were mainly due to the weaker first positive system o f N2 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 63 6000 ,4000 (a) 2000 200 W avelength (nm CM 2000» (nm) He 1500 - I 1000 • <3 500 He , He 200 400 600 Wavelength (nm) He 800 Figure 11-11. OEM spectra of (a) pure He, (b) 0.25% Freon 142B in He at 20 cc/min flow rate and 5.6 kV voltage. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 64 250 200 - He He 150 - JS c 3 8 100 ■ He He 50 - He He He He 200 O' He' 800 600 400 Wavelength (nm) 300 He 250 - 200 - 100 ■ He (A C 3 O O He 50 - He He He He' 200 400 600 800 Wavelength (nm) Figure 11-11. OEM spectra of (c) 5% O2 and 0.25% Freon 142B in He, and (d) 5% O2 and 0.25% Freon 142B in He with H2 O at 20 cc/min flow rate and 5.6 kV voltage. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 65 (B3n g-A3I u+) at 639.8 nm, and the stronger first negative system o f N 2+ (B2Zu+-X2Sg+) at 391.2 nm, and 427.6 nm.26 Additionally to the weaker signal o f all these species, very weak signals that might be attributed to OH and H20+ were found. The spectra o f the three holes were identical with hole #1 being the most intense and hole #2 being least intense in the background spectrum. Lines of He2 and He2+ were absent in these spectra. The lines o f oxygen and nitrogen species in these spectra were most likely caused by the discharge o f air between the CCD detector tip and the foil surrounding the hole where the spectra are taken. The spectra changed dramatically upon introduction of Freon 142B into the system as shown in Figure II-l lb. The intensity o f the main helium line dropped by about 75%. The intensities of other helium lines also decreased but in less dramatic fashion, down 35% (667.8nm) to 58% (728.2 nm), whereas the intensity of the He+ line at 656.3 nm more than doubled. All the nitrogen peaks and oxygen peaks disappeared in these spectra. The described changes in the helium lines showed that the interaction between the different states of helium, represented by the different wavelengths, and the Freon molecules were of different strength. Due to the low concentration of the Freon molecules (2500 ppm) and the strength of their transitions the additional lines that appeared were relatively weak. Peaks due to fluorine radicals were seen between 730 nm and 760 nm with the major ones being at 740.0 nm and 733.3 nm, and minor ones at 742.7 nm, 755.3 nm, and 757.3 nm.25 A system o f fluorine bands could be observed around 430 nm with the species being F+and F2+ (in more detail below).25,26 Moreover, a weak signal of CH (A2A-X2n ) could be seen at 431.0 nm which maybe coincided with the F2+ at about the same wavelength.26 This signal was only seen in hole #1 and #2, Reproduced with permission o f the copyright owner. Further reproduction prohibited without permission. however the signal was the most intense in hole #1. Quite a few transitions of the hydrogen radical were present as well. These transitions were located at 433.8 nm (weak), 486.1 nm (medium), and 656.2 nm (strong) . 23 The presence of these transitions could be the reason why the peak at this wavelength, which coincided with He*, increased in intensity because one molecule o f Freon 142B could yield three hydrogen radicals. Further analysis of the multi-line system between 425 nm and 435 nm showed lines of C* or Cl at 426.6 nm, F* at 427.8 nm and 430.0 nm, F2+ at 428-430 nm, and C2+ at 432.7 nm .25,26 There were still more peaks which could not be assigned in this signal system. This multi-line system was also observed in hole #2, however its intensity was weaker compared to hole #2. Data from hole #3 only showed two peaks, one at 427.6 which could be F* and an unidentified line at 431.7 nm. Signals of chlorine radicals were found at wavelengths of 774.5 nm, 808.8 nm, and 837.6 nm. The later two were the more intense ones. Nevertheless, no lines of C2, CF, CF2, CI2, HF or F2 were observed in these spectra. More drastic changes in these spectra were caused by the further addition of oxygen (5%) to this Freon/helium system (Figure II-11c). The helium peaks were reduced further in intensity, some by up to 85% (705.6 nm); others were reduced by 3050%. Weak lines due to the addition of oxygen were observed at 777.2 nm and 846.6 nm; these were due to oxygen radicals. The hydrogen lines were still present due to the CH content but they were reduced (as seen in the peak at 656.6 nm which coincided with the one of He*). The observed lines of fluorine and chlorine radicals, and CH were absent in these spectra. The reduction in line intensity agreed well with earlier observations made in our research group that the addition of oxygen into helium systems drastically Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. i 67 decreases the intensity o f the helium lines.27 The addition of water vapor to this system resulted in no significant changes in the spectra (Figure II-1Id). The intensities of the helium peaks were slightly higher and the ones o f the oxygen radical peaks were slightly lower. The hydrogen lines seemed to have the same intensity as before, regardless of the addition of water. The only major difference was that the line at 706.5 nm was no longer the strongest line; the most intense transition was now at 667.6 nm. This was due to different interactions o f the carrier gas and the other gases. The presence o f OH radicals was debatable because the peaks could not be clearly distinguished from the background. These experiments were also performed with a mixture o f 5000 ppm Freon 21 in helium; see Figure II-12a. These spectra showed the same trends as observed in the spectra o f Freon 142B. Peaks o f chlorine, fluorine and hydrogen radicals, and CH as well as the other two fluorine species (F+ and F2 ) and carbon species (C+ and C2+) appeared upon addition of the Freon. The only distinction lay in the intensities. These systems were prominent in the spectra of Freon 142B, but they were more intense in the ones of Freon 21. The lines around the CH line looked more refined. Due to higher intensity o f the CH transitions a second and third CH transition could be observed. The second transition was a very weak C2I+-X2n transition at 314.5 nm. The third transition was part o f the system around 390 nm, located exactly at 387.0 nm, and represented a B21-X2II transition. The hydrogen line at 656.6 nm (coinciding with He'*) seemed to be less intense since one molecule of Freon 21 could yield just one hydrogen radical as observed for hole #1, which seemed consistent. However, at hole #3 (not shown) this line had a relatively high intensity. The addition o f oxygen and then water had the same effect as in the spectra of Freon 142 B (Figure II-12b and c). The observed lines o f the fluorine and chlorine Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 68 1400 (a) 1200 • 1000 ■ 800 • 600 - 400 ■ 200 - He CH. r.F. He He He H He' He F . Cl 200 400 600 800 Wavelength (nm) He He He 200 (b) ■ 1 O He H He' 0 1 600 800 Wavelength (nm) 250 He 200 He He ■ 150- I0 0 - He 50- He He He He Ht , 200 400 He 600 Wavelength (nm) 800 Figure 11-12. OEM spectra of (a) 0.5% Freon 21 in He, (b) 5 % 0 2 and 0.5% Freon 21 in He, and (c) 5% O2 and 0.5% Freon 21 in He with H2 O at 20 cc/min flow rate and 5.6 kV voltage. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 69 radicals and CH disappeared as they did in the spectra of Freon 142 B, the oxygen radical lines were still present, and the hydrogen system was still unchanged after water addition. d. Mass Spectroscopic Studies. These studies were undertaken to determine the products in the case o f a Freon/helium plasma. Three mass spectra that represented three different settings are shown in Figure 11-13. The first two were background spectra for helium and Freon ( 1 .0 %) in helium, respectively; the third was a spectrum o f the species after plasma reaction. The helium background spectrum showed traces of water, nitrogen, oxygen and carbon dioxide as seen in the respective peaks due to residues in the lines from the reactor to the mass spectrometer. The helium peak itself was removed due to its very high intensity. The spectrum of Freon 21 agreed well with the literature spectrum .28 By comparing the last two spectra, Freon 21 was clearly destroyed and new species were formed. Determining the exact nature o f the new species was not easy because this spectrum was a convolution of several spectra o f the products and the reactant, Freon 21, with different weights and the relatively poor resolution of the mass spectrometer. The resulting new peaks could be qualitatively assigned to several different species by comparison with literature spectra.28 The decrease of the molecular ion peak of Freon 21 was evident in Figure II-13 c. Freon 21 seemed to be dissociated by about 87%, which was in good agreement with previously discussed gas chromatography data. The actual conversion should be higher because the products of the decomposition overlap with the molecular peak at this m/z Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 70 2e-8 to 1e-8 (a) t: S. 200 2e-a Freon 21 (b) t 0 K 1m i a. & ■c le-fl S. 100 200 150 m/z Plasma b (c) ,2 S a CHFO 2 (KM.) a. c f 4.c f,c i CCI,F co,f: CCI, CHF, c ,a .F , ■S' S \ 0 50 100 150 200 m/z Figure 11*13. Mass spectra o f (a) pure He, (b) 1.0% Freon 21 in He before plasma, and (c) 1.0% Freon 21 in He with plasma conditions. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 71 ratio. The major new peaks were the ones at m/z ratios o f 51, 85 and 87, 101 and 103, and a series of peaks at and above an m/z o f 117. The destruction o f Freon 21 seemed to favor the evolution o f chlorine rich species which is in contrast to previously mentioned work by Stoffels where fluorine rich species were favored under low pressure conditions (0-400 mTorr) . 18 The pair with m/z ratios o f 101 and 103 represented CCI3F and had a selectivity o f about 12% considering only this molecule contributes to these peaks. The other pair o f peaks at 85 and 87 could be assigned to CCI2F2 with a selectivity o f about 10 % with the same considerations as in the previous case. Besides, the peaks at m/z ratios of 51, 69, and 117 could be assigned to CHF3, CF4, and CF3CI, and CCI4 , respectively. The peak at an m/z o f 51 was likely to be CHF3 because others with a peak at an m/z of 51 did not have the other corresponding peaks in the spectrum o f the plasma. CF4 and CF3CI had to be part o f the peak at an m/z of 69 because the ratio of m/z ratios o f 69 to 67 is too high for being just from Freon 21. The peaks from an m/z of 117 onwards were likely to be C2s and C3s such as 1,1,2,2tetrachloro-1 ,2-difluoroethane. Some of these had to also have a peak at an m/z o f 102 because the abundance of this ion was too high. This peak showed a conversion of 28%, which was not as high as that observed for the molecular ion at an m/z of 67, which is 87%. Some trends regarding the formation of small fluorine- and chlorine-containing molecules could be drawn from the mass spectrometry data. Comparison of the data of masses 19 and 38 at plasma condition and before plasma indicated formation of hydrogen fluoride and no formation of fluorine, as seen in the increasing peak at an m/z of 19 under Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 72 plasma conditions and the decreasing peak at an m/z o f 38. The decrease in the peaks at m/z ratios o f35,37, and 70 indicated hydrogen chloride and chlorine were not formed. e. Power and Efficiency Measurements. The power used in the reactions was calculated by the integration o f voltage versus current curves. The method used for integration was the trapezoid rule. These calculations were carried out using Sigma Plot 3.0. If the power is plotted against the applied input voltage, the best fit is a straight line with an intersect at about 1.60 kV. This 0.6 □ 0.5 - • - □ t § Q. & □ • <1 < T ■ 1 & 0.1 - % • □ A % • ■4 % 0.0 2 3 Freon 0.5% Freon 0.5% Oxygen 5%+Freon 0.5% i ■ i 1 6 4 Input Voltage(kV) Figure 11-14. Correlation o f average power to input voltage for 0.5% Freon 21 in He and 0.5% Freon 21 in 5.0% oxygen and He. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 73 is the barrier voltage for most cases for Freon 21 at a concentration of 0.5% (Figure II14). Sets of data for the case involving just Freon 21 in helium and one for oxygen and Freon 21 in helium were used to calculate the efficiency o f these reactions. Data for these calculations are listed in Tables II-1 and II-2.29 In equation 1, F represents the flow rate in mL/s, Co the Freon concentration in volume % used, x tbe conversion in %, P the power consumed in W, 22,400 mL/mol, the molar volume at 273 K, 0.93 the correction factor for the molar volume from 273 K to 298 K. Efficiency, mol/J = (0.93 x F x Co x x)/(l s x 22,400 x P) (1) Table II-l. Efficiency as a function of flow rate, percent conversion, and power at a concentration of Freon 21 of 0.5% and of oxygen o f 5.0% in helium and different input voltages (Vin). Flow Rate Efficiency (mL/min) (mol/J) 100 2 0 .1 1.08x10"® 0.0484 100 19.9 1.42x1 O' 6 2.160 0.0750 100 2 0 .1 9.20X10*1 2.360 0.141 100 2 0 .0 4.90x10"' 2.480 0.215 100 2 0 .0 3.20x10"' 2.760 0.218 100 2 0 .1 3.17x10"' Vi„(kV) Power (W) Conversion (%) 2.040 0.0640 2.080 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 74 Table II-2. Efficiency as a function o f flow rate, percent conversion, and power at a concentration of Freon 21 o f 0.5% in helium and different input voltages (Vjn). Flow Rate Efficiency (mL/min) (mol/J) 76 2 0 .0 8.29x10"' 0.131 83 2 0 .0 4.38x10-’ 3.400 0.269 95 2 0 .0 2.44x10"7 4.040 0.390 93 2 0 .0 1.65x10"' 4.720 0.509 100 2 0 .0 1.36x10’ Vi„(kV) Power (W) Conversion (%) 1.640 0.0634 2.280 The efficiency of these reactions is fairly low at about 1.0 x 1 0 -6 mol/J. The efficiency itself decreases by an order of magnitude with increasing power while increasing the input voltage from 1.640 to 4.720 kV in the case o f Freon 21 by itself, but this trend is seen in the mixture with oxygen as well. In this mixture, the efficiency is slightly higher than for Freon 21 alone, as seen at about 2.3 kV where the Freon case shows an efficiency of 4.38x1 O' 7 mol/J compared to 4.90x10*7 mol/J for Freon and oxygen. Oxidation with oxygen alone is favored over the oxidation involving water vapor and oxygen as shown in thermodynamic calculations of the total oxidation o f Freon 21. This information is obtained from calculations of AH0 and AG° for both scenarios. The reaction involving oxygen has AH0—383.55 kJ/mol and AG°=-416.93 kJ/mol, compared to AH°=-234.05 kJ/mol and AG°=-286.62 kJ/mol for reaction without oxygen. Furthermore, this means that chlorine is thermodynamically favored over hydrogen chloride as a product in the oxidation of Freon 21. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 75 f. Long Term Stability. The long-term stability o f the electrode and the effect o f the reaction on the quartz glass tube used in this study are important factors to consider. The stability o f the iron electrode depends on the conditions of the electrode and reaction conditions. If water is used in these reactions, the electrode and the glass tube have to be cleaned after about 2 days o f 8 hour/day continual operation. If oxygen is in the reaction mixture, rust is the only major problem, and the electrode lasts at least a week. Freon by itself does not impact on the electrode performance. The effect on the glass tube is that it becomes “blind” due to the formation of silica or reaction with produced hydrogen fluoride. The iron electrode is cleaned by using sandpaper, and the glass tube is rinsed with water first and then with acetone to remove residues. A component that has to be exchanged from time to time is the NUPRO filter (7 pm pore diameter), which is placed in line before the gas chromatograph to prevent damage to the instrument. 4. Discussion. a. Activation o f the Freon Molecules. The glow discharge plasmas used in these experiments were dielectric-barrier or silent discharges. Furthermore, these discharge plasmas were non-equilibrium plasmas at atmospheric pressure. In these plasmas the real gas temperature was close to room temperature whereas the electronic temperature was in the order o f several thousands of Kelvin. This was caused by the nature of activation. The electric field accelerated the electrons to a certain speed and therefore gave them a certain energy. All species in the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 76 plasma were consequently not thermally activated but instead were activated by highenergy electrons directly or indirectly.30 The optical emission data suggested that the Freon molecules similar to others (water and oxygen molecules) were activated indirectly by excited carrier gas species, in this case excited helium atoms or ions (as discussed later); this was in good agreement with the results in a paper published recently by Luo et al. , which states this as the major source o f reactant molecule activation. The optical emission work presented here encompassed reactions of two similar Freon molecules. These Freons molecules readily reacted in the discharge plasma with the excited helium atoms and ions. The helium atoms were excited by inelastic collisions with high-energy electrons into different excited states. The spectra o f just helium as well as others showed several different excited states of helium; the nature o f the oxygen and nitrogen lines was discussed previously. The most important excited state in the activation o f the Freon molecules was the 33S state with a line at 706.5 nm.31 This line showed the biggest intensity drop upon Freon addition exhibiting the best energy transfer efficiency. Other helium lines showed similar behavior but the efficiency was smaller. The decreases in efficiencies were as follows: He 33S (706.5 nm) > He 3'S (728.2 nm) > He 33D (587.6 nm) > He 3'D (667.8 nm )» He+ (656.3 nm). The assignment o f the He+ line posed a problem due to its overlap with a H line. The extent to which each line contributed to the final line is unknown. Similar trends were observed when oxygen was added to the Freon/helium system. The extent to which the peaks were reduced is similar to the addition of Freon resulting in the same order o f the excited states. Upon addition o f water to the system the interactions o f the molecules present were slightly weaker as seen in the stronger helium lines compared to the system without water. This could be caused Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 77 by interaction o f oxygen with water because the oxygen radical lines were slightly weaker. b. Modelfo r Decomposition. The model proposed as shown in equations 2-21 is valid for the decomposition of Freon 21 without oxygen or water; it involves inelastic collisions o f various partners. The following is the model for Freon 21 pieced together from optical emission data and mass spectrometry data with consideration from thermodynamic data (bond strength)25: CHFC12 + He* -> CHFCI2* + He (2) CHFCh* CHFC1 + Cl* (3 a) CHFCh* -» CFCI2 + H* (3 b) (minor) CHFC1 + He* CHFCr* + He (4) CHFC1* -> CHF + Cl* (5 a) CHFC1 * -» CFC1 + H* (5 b) CHF + He* -» CHF * + He ( 6 a) CFC1 + He* CFC1 * + He ( 6 b) (minor) C H F * -» C H + F (7 a a ) CHF * CF + H* (7 ab) CFC1 * -> CF + Cl (7 ba) (minor) CFC1* -» CC1* + F (7 bb) (minor) CF + He* -> CF * + He ( 8 a) CH + He* CH * + He ( 8 b) C F * -» C + F ( 9 a ) Reproduced with permission o f the copyright owner. Further reproduction prohibited without permission. I 78 C H * -» C + H '(9 b ) Recombination: H + F -> HF (10) F + F -» F 2 (11) H +C l -»HC1(12) Cl +C1 ->C12 (13) H +H -»H 2(14) CFC12 + CFC12 -> C2F2Cl4 (15) CHF + F2 -> CHF3 (16) CFC12 + F -> CF2C12 (17) CFC12 +C1 ->CFC13 (18) CF + F + F2 CF4 (19) CFC1 +F 2 -»C F 3C1 (20) CC1+C1 +C12 -»CCL, (21) In reaction 2, a major pathway for the activation o f Freon 21 was the reaction with the 33S state of helium, depicted as He*. The observed species o f C+, C2*, F+ and F2+ could be explained by electron impact ionization o f species formed in the reaction 2 -2 1 ; these reactions are shown in reactions 22-25: C +e' -> C++2 e (22) C++e' -» C2++2 e' (23) P + e*-»F,f+2e'(24) F2 +e‘ ■> F2++2 e‘ (25) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 79 Small molecules containing fluorine and chloride such as hydrogen fluoride were formed by reactions 10-14. Hydrogen fluoride was the major species formed under plasma condition as seen in trends from the mass spectrometry data, whereas other species such as hydrogen chloride and fluorine did not seem to play a major role under these conditions. C2s, C3s and higher hydrocarbons could be formed by radical recombination; the evolution of substituted methane analogs and C2s such as CFCI2, as observed in the mass spectra of Freon 21 under plasma condition, is well documented by reactions 15-21. The formation of aromatic compounds by radical recombination containing fluorine was confirmed by NMR analysis of the residue on the iron electrode after plasma reactions (results are not shown in this thesis). The Freon 142B followed a similar reaction scheme. This scheme is described in reactions 26-57 and based mostly on analogies to the Freon 21 scheme and spectroscopic data: CH3CCIF2+ He* -» CH3CCIF2* + He (26) CH3CCIF2* -» CH3CF2 + C l (27 a) CH3CCIF2* -» CCIF2 + CH3 (27 b) CH3CCIF2* -» CH3CCIF + F (27 c) CH3CF2 + H e* -» CH 3CF2 * + H e CCIF2 + He* (28 a) CCIF2 * + He (28 ba) CH3 + He* -> CH3 * + He (28 bb) CH3CCIF + He* -» CH3CCIF * + He (28 c) CH3CF2 * -» CH3CF + F (29 aa) CH3CF2 * -» CF2 + CH 3 (29 ab) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 80 CC1F2 * -» CF2 + c r (29 baa) CC1F2 * -» CFC1 + F (29 bab) CH3* -> CH2 + H (29 bb) CHjCCIF * CH3CF + C r (29 ca) CHjCCIF * -> CH3CC1 + F (29 cb) CHjCCIF * CFC1 + CH 3 (29 cc) CH3CF + He* -» CH3CF * + He (30 aa) CF2 + He* -» CF2 * + He (30 ab) CFC1 + He* CFC1 * + He (30 bab) CH2 + He* -> CH2 * + He (30 bb) CH3CCI + He* CH3CCI * + He (30 cb) CH3CF * CH3C + F (31 aaa) CH3CF * CF + CH 3 (31 aab) CF2 * ->CF + F (3 1 ab) CFC1 * -> CC1 + F (31 baba) CFC1 * -> CF + C r (31 babb) CH2* -> CH + H -(31 bb) CH3CCI * CH3C + C r (31 cba) CH3CCI * -> CC1 + CH 3 (32 ebb) CH3C + He* -> CH3C * + He (32 aaa) CF + He* CF * + He (32 aab) CC1 + He* -> CC1 * + He (32 baba) CH + He* -» CH * + He (32 bb) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHjC * -» C + CH 3 (33 aaa) CF * -> C + F (33 aab) CC1* -> C +C 1'(33 baba) CH * C + H' (33 bb) Recombination: H +F HF (34) F + F -» F2 (35) H + Cl -> HC1 (36) Cl +C1 ->C12 (37) H + H -»H 2(38) CH3CF2 + CH3CF2 -» C4H6F4 (39) CH3CF2 + CH3CFC1 “> C4H6F3C1 (40) CH3CFC1 +CH 3CFC1 C4H6F2C12 (41) CH3CF2 + CH3 “> C3H6F2 (42) CH3CFC1 + CH3 -» C3H*FC1 (43) CH3CFC1 + CC1F2 -» C3H3F3C12 (44) CH3CF2 + CC1F2 C3H3F4C1 (45) CC1F2 + CC1F2 -» C2F4C12 (46) CC1F2 + CH3 -» C2H3F2C1 (47) CH3 +CH 3 ->C 2H6 (48) CH3CFC1 + F -> CH3CF2C1 (50) CH3CFC1 + Cl -» CH3CFC12 (51) CH3CF2 + F ->CH 3CF3 (52) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 82 CH3CF2 +C1 -» CH3CF2CI (53) CCIF2 + F -> CCIF3 (54) CCIF2' + Cl’ CC12F2 (55) CH3 + F - » C H j F ( 56) CH3 +C1 -»CH 3CI(57) Generally, for Freon 142 B, the abstraction of a hydrogen radical was substituted by the abstraction o f a methyl radical. The only pathway for the formation o f hydrogen and CH radicals was the stepwise breakup o f methyl radicals as seen in reactions 26-33. The higher hydrogen content in Freon 142B explained the more intense hydrogen lines in the spectra of this Freon. The lower intensity for the CH line was due to the stepwise disintegration o f the methyl radicals resulting in a lower probability o f CH generation. The formation o f fluorine compounds seemed to be favored due to the higher abundance of fluorine in Freon 142B compared to that in Freon 21. The same trends for the evolution of higher hydrocarbons should apply for Freon 142B but the likelihood o f this should be greater than in the case of Freon 21 due to the presence of methyl radicals and other carbon containing radicals or fragments such as substituted ethyl radicals. The evolution of unsubstituted methyl, ethyl and higher alkyl species was unlikely due to the fast reactions of the hydrogen radical as in the case of Freon 21. The model for the decomposition o f Freon 21 had to be complemented if oxygen and water were present in the feed gas. The following equations had to be added to account for the change in reaction conditions: O2 + H e * O 2 * + He (58) O2 * 0 + O (59) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 83 CH + 0 -» CO + H (60) CO + 0 -> CO2 (61) H20 + He* -» H20* + He (62) H20 * - » 0 H +H (63) CH + OH -> CO + H2 (64) CO + OH COz + H (65) F + O H - » 0 + H F (6 6 ) Cl + OH -» O + HC1 (67) Upon addition of oxygen and in the second case of using oxygen and water the reaction products changed drastically because these species were in excess in the reaction mixture. The principal model o f reactions 2-9 is still valid under these conditions but has to be modified by reactions 58-67. Since no lines of species in the first reactions (2-9) were observed in the optical emission spectra it was assumed that these are very quickly scavenged by O’, OH , and H' as shown in reactions 60, 64,66, and 67. The formation of CO and COz in this system was explained by reactions 60-67. The excess of oxygen in the feed explained the appearance of O . The formation of hydrocarbons could not be ruled out, especially with the formation of a polymer-like film on the electrode, which could be due to a fluorinated hydrocarbon polymer. The model for Freon 142 B under these conditions is identical. c. Efficiency Considerations. Mass spectrometry data and gas chromatography data for the destruction of both Freons showed very high conversions up to 100%. This “efficient” destruction was Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. caused by the influence o f the plasma. The easy abstraction o f chlorine or hydrogen radicals by electron or ion impact seemed to be the main driving force to yield almost complete destruction o f the Freon molecules. In order to be viable for industrial applications the reaction efficiencies with the respect to energy input have to be examined and to be optimized. Data o f Table II-1 and II-2 suggest that these reactions have low energy efficiencies. As discussed before the efficiency decreases with increasing voltage, which is expected because more molecules have the appropriate energy for reaction but the collision between the molecules were less selective and less efficient than for the case at lower voltage. The efficiency o f the destruction was slightly higher in the case when oxygen is added; this is believed to be an effect o f the presence of abundant oxygen radicals. Generally, plasmas are a good and efficient way to break down hydrocarbons, which was shown in earlier research in our group.32 The difficulty with Freons is that they are very hard to decompose due to their stability. In further studies in this field, CF4 is even harder to decompose (conversions being lower than 20% under the same conditions) than these Freons due to its increased stability.33 To improve the efficiency of this method, parameters such as frequency, input voltage, and surface of the electrode can be altered to achieve increased efficiency. A “cleaning gas” might be used to purge the reactor after usage to increase efficiency. The low efficiency is also influenced by the electrical setup. There were substantial losses at the resistors in the setup as well as at the power supply. In order to minimize these losses more efficient power supplies and maybe also resistors have to be selected and tested in this setup. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 85 5. Conclusions. Hydrochlorofluorocarbons (HCFCs) are a very important substitute for the more dangerous chlorofluorocarbons (CFCs). In the coming years, further restrictions are likely to be implemented to phase out these substitutes. In the Beijing Amendment to the Montreal Protocol, the decision was made to freeze the production o f HCFCs in industrialized countries starting in 2004. Therefore, effective ways to eliminate these pollutants are crucial. High voltage glow plasmas are effective in destroying the pollutant Freons at atmospheric pressure and at room temperature. In these reactions almost complete destruction was observed at low energy efficiencies. Based on the mass spectrometry data and the optical emission data a model for the destruction o f Freon 21 and Freon 142B was proposed and is in line with the experimental data. The main products of the reactions depend on the conditions. If the Freon was used by itself, hydrogen fluoride, chlorinated and fluorinated methane analogs were formed with other hydrocarbons such as C2s, C3s with aromatic compounds evolving at the same time. In the presence o f oxygen, hydrogen fluoride was still one of the main products but carbon dioxide was the other major product with good selectivity, with small amounts o f carbon monoxide forming as well. The presence of other hydrocarbons was not investigated in this case, but it is likely that they formed especially higher hydrocarbons because o f the polymer-like coverage o f the electrode. The data for further addition o f water were very similar to the data with oxygen and Freon. In this case, the conversion seemed almost independent from the applied input voltage. Therefore, good conversions were possible at low voltages with relatively high efficiencies when Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 86 employing Freon and oxygen. This system is very much suited for the very efficient destruction o f Freons under the various conditions employed in this research. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 87 6. References. 1) Freon Naming System, Environmental Sciences Course, Nova Scotia Agricultural College, httn://www.nsac.ns.ca/cnvsci/courses/csl 10-7001/imii6 /frer>ns.html. December 2002. 2) Solomon, S. Nature 1990,347, 347-354. 3) Rowland, F. S.; Molina, M. J. Rev. Geophys. & Space Phys. 1975,13, 1-35. 4) Molina, M. J.; Rowland, F. S. Nature 1974,249, 810-812. 5) Rowland, F. S. Ann. Rev. Phys. Chem. 1991,42, 731-768. 6) Ravishankara, A. R.; Tumipseed, A. A.; Jensen, N. R.; Barone, S.; Mills, M.; Howard, C. J.; Solomon, S. Science 1994,263, 71-75. 7) Solomon, S.; Albritton, D. L. Nature 1992,357, 33-37. 8) Hayman, G. D.; Derwent, R. G. Environ. Sci. Technnol. 1997,31, 327-336. 9) The UNEP Ozone Secretariat, http://www.unep.org/ozone/. December 2002. 10) Vercammen, K. L. L; Berezin, A. A.; Lox, F.; Chang, J. S. J. Adv. Qxid. Technol. 1997,2, 312-329. 11) Ruddy, E. N.; Caroll, L. A. Chemical Engineering Progress 1993,89, 28-35. 12) Takita, Y.; Ishihara, T. Catalysis Surveyfrom Japan 1998,21, 165-173. 13) Sonoyama, N.; Sakata, T. Environ. Sci. Technnol. 1998,32, 375-378. 14) Sonoyama, N.; Sakata, T. Environ. Sci. Technnol. 1998,32, 4005-4009. 15)Gritsinin, S. I.; Kossyi, I. A.; Nisakyan, M. A.; Silakov, V.P. Plasma Phys. Rep. 1997,23, 242-250. 16) Aleksandrov, N. L.; Dobkin, S. V.; Konchakov, A. M.; Novitskii, D. A. Plasma Phys. Rep. 1994,20. 442-448. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 88 17) Jasinski, M.; Mizeraczyk, J.; Zakizewski, Z.; Ohkubo, T.; Chang, J.-S. J. Phys. D 2002,35. 2274-2280. 18) StofFels, W. W.; Stoffels, E.; Haverlag, M.; Kroesen, G. M. W.; de Hoog, F. J. J. Vac. Sci. Technol. 1995,13, 2058-2066. 19) Wang, Y.-F.; Lee, W.-J.; Chen, C.-Y.; Hsieh, L-T. Environ. Sci. Technol. 1999,33, 2234-2240. 20) Korzekwa, R. A.; Rosocha, L. A. J. Adv. Oxid. Technol. 1999,4, 390-399. 21) Oda, T.; Yamashita, R.; Tanaka, K.; Takahashi, T.; Masuda, S. IEEE Trans. Ind. Appl. 1996,32. 1044-1050. 22) Oda, T.; Yamashita, R.; Takahashi, T.; Masuda, S. IEEE Trans. Ind. Appl. 1996,32. 227-232. 23) Akhmedzhanov, R. A.; Vikharev, A. L.; Gorbachev, A. M.; Ivanov, O.A.; Kolysko, A.L. High Temp. 1997,35. 514-527. 24) Akhvlediani, Z. G.; Barkhudarov, E. M.; Gelashvili, G. V.; Kossyi, I. A.; Melitauri, I. T.; Taktakishvili, M. I. Plasma Phys. Rep. 1996,22, 428-435. 25) CRC Handbook o f Chemistry and Physics, 75U| Edition, Lide, D.R., CRC Press; Boca Raton, FL, 1994, Section 10,1-127. 26)Pearse, R. W. B.; Gaydon, A. G., Identification o f Molecular Spectra, Chapman and Hall, New York, NY, 1976. 27) Luo, J.; Suib, S. L.; Hayashi, Y.; Matsumoto, H. J. Phys. Chem. A, 1999,103, 6151-6161. 28) Stein, S. E., NIST Standard Reference Database Number 69- November 1998 Release. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 89 29) Luo, J.; Suib, S. L.; Marquez, M.; Hayashi, Y.; Matsumoto, H. J. Phys. Chem. A, 1998,102, 7954-7963. 30) Dresvin, S. V., Physics and Technology o f Low-Temperature Plasmas, The Iowa State University Press, Ames, LA, 1977. 31) Reader, J.; Corliss, C. H.W.; Wiese, L.; Martin, G. A., Wavelengths and Transition Probabilities fo r Atoms and Atomic Ions; Natl. Stand. Ref. Data Ser., Natl. Bur. Stand. (U.S.) 6 8 (1980). 32) Huang, A.; Xia, G. G.; Wang, J.; Suib, S. L.; Hayashi, Y.; Matsumoto, H. J. Catal. 2000,189,349-359. 33) Spiess, F.-J.; Suib, S. L.; Hayashi, Y.; Matsumoto, H. ACS Symposium Series: Utilization o f Greenhouse Gases, in press. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. B. DESTRUCTION OF CARBON TETRAFLUORIDE BY DISCHARGE PLASMA AND MICROWAVE HEATING. I. Introduction. Carbon tetrafluoride (Freon 14) was found in the air of various European countries and in both hemispheres o f the troposphere (0-14 km) in the 1970s.1,2 Cicerone’s calculations and modeling showed that carbon tetrafluoride is a nearly inert gas in the atmosphere.3 The residence time o f CF4 in the atmosphere is calculated to be on the order o f 1 0 ,0 0 0 years, up to 100 times higher than for simple chlorofluorocarbons. Furthermore, CF4 traps the outgoing planetary infrared radiation in its intense band at about 8 micrometers. These properties make CF4 a very potent greenhouse gas whose atmospheric concentration is believed to be due to mostly industrial sources; natural sources seem to be negligible. Atmosphere Some IR radiation lost to space / / Some IR radiation absorbed and re-emitted by greenhouse gases / resulting in warming of surface / Incoming ' Solar radiation Solar energy absorbed by surface, warms surface Emission of IR radiation as result of warming of surface Figure 11-15. Schematic of the greenhouse effect (Adapted from Ref 4). 90 Reproduced with permission o f the copyright owner. Further reproduction prohibited without permission. The greenhouse effect is schematically shown in Figure 11-15. Incoming solar radiation can be reflected by the atmosphere or clouds or reach the surface o f the earth. The radiation reaching earth warms up its surface and part o f it is irradiated back as infrared (IR) radiation. O f this back-irradiated radiation part is lost to space but part o f it is absorbed by molecules classified as greenhouse gases. These molecules re-emit this radiation and thereby contribute to the warming o f the earth’s surface and the troposphere. This starts a cycle and more heat is emitted and re-emitted. To a certain degree, there is a natural greenhouse effect, because otherwise earth would be an “ice planet" similar to Mars.4 Table II-3. Major greenhouse gases, origins, lifetimes and global warming potentials (Adapted from Ref. 4). Greenhouse Pre-industrial Gas Concentration Carbon Dioxide 278 000 ppbv Concentration Atmospheric In 1994 358 0 0 0 ppbv Sources Global Lifetime Wanning (Years) Potential Variable Fossil 1 Fuel Methane 700 ppbv 1721 ppbv 1 2 .2 Fossil 21 Fuel Nitrous Oxide 275 ppbv 311 ppbv 120 Fertilizer 310 CFC-12 0 0.503 ppbv 102 Liquid 6200- Coolants 7100 Liquid 1300- Coolants MOO HCFC-22 Perfluoromethane 0 0 0.105 ppbv 0.070 ppbv 1 2 .1 50 000 Aluminum 6 500 Production Sulfur Hexafluoride 0 0.032 ppbv 3 200 Dielectric Fluid Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 23 900 92 The group of greenhouse gases consists o f different gases with very different atmospheric lifetimes and concentrations as seen in Table II-3. The present atmospheric concentrations are higher by far than the pre-industrial concentrations. The main contributors are carbon dioxide and methane. Perfluoromethane (another name for carbon tetrafluoride) has a relatively low atmospheric concentration but its very long atmospheric lifetime and its very high global wanning potential and therefore contributes significantly to global warming.4 Temperature Rise Rising Sea Levels More Infectious More Precipitation Change in Change in Lower Crop Water Forest Respiratory Yields Composition, Quality, Illness Supply. Weather-Related Higher Irrigation Geographic Demand Competition Range, Health Mortality Beach Erosion, Loss of Habitat, Diminishing Glaciers, Inundation of Coastal Lands Figure 11-16. Potential impacts of climate change on the environment (Adapted from Ref. 4). The impact of climate change is multi-faceted (Figure 11-16). Sea levels are expected to rise by 40-60 cm by 2100, and the average earth temperature depending on the scenario is expected to rise by 1-5°C by 2100. To address this problem the 1997 Kyoto protocol to the United Nations Framework Convention on Climate Change was Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 93 struck.4 In this Protocol, the industrial countries agreed to cut their emissions by 5.2% over the 2008 to 2012 period; this includes CF* 5 This originally included a 7% reduction of emission for the United States. New negotiations were set up to ultimately define the goals for the individual partners o f the Protocol, hi 2001, an agreement was reached to implement the Kyoto Protocol and its operational details (Marrakech Accords). The individual targets are 8 % for Switzerland, most Central and East European states, and the European Union (the EU has distributed different rates among its member states); and 6 % for Canada, Hungary, Japan, and Poland. Russia, New Zealand, and Ukraine are to stabilize their emissions, while Norway may crease emissions by up to 1%, and Iceland 10%. The United States and Australia opted out of the agreement which goes into effect if the countries responsible for 55% o f the greenhouse gas emissions (as o f 1990) ratify it. As o f December 2002, this tally is up to 43.7%. With Russia’s approval (17.4%) in early 2003 the Protocol will go into effect shortly thereafter.6 CF4 is a by-product in aluminum production and is used in the semiconductor industry.7*10 In the semiconductor industry, carbon tetrafluoride is one o f the gases used in etching (dry etching silicon wafers using radio frequency plasmas) and cleaning (plasma etch cleaning o f plasma enhanced CVD reactors after deposition) processes.7 The emissions from these processes in the US semiconductor industry, which could be reduced by 98% using a capture and recycle system or by using plasma or thermal decomposition (as mentioned below), amass up to 230 tons per year, about 5-10% of global emissions, compared to 6,800 tons per year by the US aluminum smeltering industry.7,8 Carbon tetrafluoride is one of the by-products in the electrolytic smelting of alumina to produce aluminum metal and 1.3 to 3.6 kg CF4 is produced per ton of Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 94 aluminum produced (1985).9 Emissions from newer cells are lower, up to 20 times lower in 1999 than in 1993 as reported in a Canadian study.7,10 The exceptionally long atmospheric lifetime of carbon tetrafluoride necessitates finding solutions to destroy this potent greenhouse gas. So far mostly catalytic and plasma discharge methods have been used to decompose CF* 11' 19 Wei et al. 11 reported destruction efficiencies over 99% using a microwave-generated surface wave plasma under optimized input power, pressure, and oxygen admixture. Grytsinin et al. 12 showed effective decomposition using a slipping surface discharge. Tonnis et al. 13 showed the importance of water addition in a Litmas RF abatement reactor. Xu et al. 14 concluded that water as an additive could cause destruction o f 99.6% of CF4, and adds that in the presence o f nitrogen NO* could be formed. Takita et al. 15 used AlPCVrare earth phosphate catalysts to decompose CF4 at conversions of about 50%. Kanno et al. 16 developed a catalyst, which destroyed CF4 with a conversion of over 99.9%. Burdeniuc17 showed mineralization of chlorofluorocarbons using sodium oxalate and achieved 95 to 100% destruction. Similar results were reported by Lee et al.18; they showed that by using heated alkali halides carbon tetrafluoride can be effectively destroyed to yield alkali metals. Several papers such as Jacobsohn et al. 19 describe the decomposition as a means of depositing fluorinated amorphous-carbon films. Our research group performed research on low-temperature AC discharge plasmas involving CF4 utilizing different reactors at atmospheric pressure compared to the ones mentioned before which were done at lower pressure or in vacuum.20 In this research, we first studied the decomposition using AC plasmas with carbon tetrafluoride and nitrogen as the balance gas, but the conversions were too low, so our focus shifted to Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 95 the use o f microwave heating because preliminary data were very promising.21 The microwave heating was carried out at medium power (0-1200 W) using water as a supplement and activated carbon as a catalyst. The experiments were performed using an ASTEX microwave unit. 2. Experimental Section. a. Preparation o f Gas Mixtures. A 5% mixture of carbon tetrafluoride in nitrogen was purchased from Connecticut Airgas Inc. This mixture was diluted online with nitrogen to achieve the desired carbon tetrafluoride concentrations. Water vapor was introduced into the reaction system by the gas mixtures from a bubbler filled with deionized water, yielding a concentration of 3% at room temperature of 25°C and atmospheric pressure. The content o f the feed gas is 4% CF4, 3% water, and a balance o f nitrogen at a flow rate o f 15 mL/min for the microwave reactions. In earlier microwave reactions, the feed gas was 2% CF4, 3% water, and a balance o f nitrogen at a flow rate o f 10 mL/min. For the plasma reaction, the same mixture was used, but at a flow rate of 100 mL/min. b. Experimental Setup and Microwave Apparatus. A simple setup was used as shown in Figure 11-17. The feed is a mixture of carbon tetrafluoride, water, and nitrogen. The activated carbon (100 mg) was placed in a quartz tube (3/8 in. outer diameter) and secured with quartz wool. This tube was then placed into the ASTEX microwave cavity (Figure II-18). The ASTEX unit consists of an ASTEX microwave power source module GK.139 with a magnetron type GL 130WC, a 3 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 96 stub tuner model AX3041, and an ASTEX applicator model AX7020. The power was emitted in pulses with a frequency o f 120 MHz (every 8.3 ns). All parameters were kept constant. Only the power level was adjusted as necessary. 15 cc/min flow of C F* H2 O and N 2 MW heating 0-1200 W AVI I X I— h Syringe I I MS 00 Figure 11-17. Reaction setup for CF4 decomposition. ‘ '} Gas out 10 cm Figure 11-18. Photograph of the reaction apparatus. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 97 A water scrubber was placed in front o f the mass spectrometer to prevent hydrogen fluoride (HF) from getting into critical parts o f the unit and furthermore trap HF in order to prevent HF from escaping into the laboratory atmosphere. The reaction system was equilibrated at the beginning and then the power was turned on to the desired level and continued until equilibrium was achieved. c. Plasma Experiment Setup. For the discharge plasma experiments, a set-up similar to the one used in the Freon experiments was used (Figure II-19). The discharge reactor consisted of a Pd metal rod (about 1 0 0 pm thick coating on copper rods and produced by electroless plating) o f 8 mm diameter as an inner electrode and aluminum foil wrapped around a quartz tube o f 1 0 mm inner diameter as the outer electrode. The effective plasma zone was 20 cm. The voltage set-up is identical to the one in the Freon experiment (Figure II-1). The exception is that resistor R2 was a variable with SO, 100, and ISO kfl, usually with a 22S W rating. The power was generated by an UHV-10 power supply and the voltage, current and their respective waveforms were monitored by a DL-1S40 Yokogawa oscilloscope using a high- (Tektronix P602S) and a low-voltage probe (Yokogawa 70996). The fan Pact reactor used a similar setup, but a TREK variable frequency power supply was used. This reactor consisted o f an inner rotor o f S.S9 cm diameter with 10 evenly spaced fan blades protruding 0.34 cm from the fan and a stator with an inner diameter of 6.33 cm. This resulted in an electrode gap o f 0.3 mm. The width o f the fan was 1.66 cm, and the resulting plasma was about 0.25 cm3. The inner and outer electrodes were coated with Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 98 gold (about 100 pm thick) by electroless plating. An inserted fan motor in the rotor provided rotation speeds o f3600 revolutions per minute. Inner electrode 8 mmOD b Quartz tube 10 mm ID 20 cm Outer electrode, Al foil I G as in G as out Figure 11-19. Sketch of tubular PACT reactor. d. Product Analysis and Catalyst Characterization. In the early experiments, sampling was done using a syringe and then injected into a Hewlett Packard Series II gas chromatograph (GC) equipped with an Alltech Poropak N 80/100 with a thermal conductivity detector; the plasma experiments used online sampling by automatic injection into the GC. Later on, the reaction mixture was continuously monitored with a MKS-UTI PPT quadrupole residual gas analyzer mass spectrometer (MS) with a Faraday cup detector and a variable high-pressure sampling manifold. The catalysts used were either prepared in our research group or used as obtained from commercial vendors. Surface area and pore size measurements were performed using a Micrometries ASAP 2010 BET system. Ion chromatography (Dionex Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 99 40001) was carried out at the Environmental Research Institute at the University o f Connecticut Fluorine/Fluoride was extracted from an activated carbon sample by water extraction. No acids or bases were used during extraction. The aqueous sample was then analyzed for fluoride; the minimum detection limit by ion chromatography is 1 mg F/L. Infrared Analysis was done with a Nicolet 750 Fourier transform infrared spectrometer with a mercury-cadmium-telluride detector and a KBr beam splitter. Scanning electron microscopy (SEM) was performed using an Amray 1810 SEM instrument with an attached Econ-4 PV9700145 energy dispersive X-ray analyzer. e) Safety Concerns. CF4 is a colorless, odorless, and nonflammable gas which at a level of 895,000 ppm/15 minutes can lead to hypoxia with dizziness, disorientation, incoordination, narcosis, nausea, and vomiting. Therefore, these reactions were carried out in a fume hood. The formed hydrogen fluoride is captured in a scrubber to prevent damage to instruments and for safety concerns and will be disposed o f properly. The microwave unit is safe to use in the described experiments. 3. Results. a. Discharge Plasma Studies. Initial studies using discharge plasmas employed many different electrodes and carbon tetrafluoride concentrations. The experiments presented here were done using a Pd electrode as the inner electrode and using 2% CF4 and 3% water in a balance o f nitrogen at flow rates of 100 mL/min. In these reactions, the influence of the resistor R2 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 100 parallel to the discharge reactor was investigated. It seemed that 100 kfi is the optimal resistance in this case because at higher and lower resistance lower conversions were observed. At a peak-to-peak input voltage of 12.2 kV up to 12% conversion was observed at 100 kH, whereas 7% were seen at 150 kfi and 10% at 50 kQ (Figure 11-20). Different power supplies were used but the conversion remained low due to limitations of the power supply or because no uniform plasma could be established in the fan PACT reactors.21 In the case of the fan reactor, a variable frequency power supply was used at voltages similar to the ones mentioned above. The conversion, however, was even smaller in a Au fan reactor with 1-2% in the presence o f water. •••e , c c & £ 5c 5 -2 - -200 0 200 400 600 800 Time on stream in min Figure 11-20. Destruction o f 2% CF4 in 3% H2O and N2 at Vp^,= 12.2 kV at 100 cc/min. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 101 b. Microwave Studies. Due to the low conversions using discharge reactors, more effective ways to decompose CF4 were sought. MS studies during the discharge plasma experiments suggested that CF4 alone or CF4 and water were the most suitable systems for the decomposition. Studies using microwave heating exhibited promising results under similar conditions and were therefore pursued. The first set of reactions was carried out with the equipment described above (Figure 11-17 and 11-18) at a constant flow rate of 100 mlVmin. This procedure was attempted in order to reproduce the conditions used for the earlier plasma experiments. However, due to limitations o f this setup, because the catalyst would not stay in place in the quartz tube, the flow rate was lowered to 10 mL/min. Several catalysts were examined for these types of experiments. The catalysts tested were several different kinds of activated carbon with various surface areas, a Ni filament type catalyst, V2 O5 supported on AI2 O 3 , and a zeolite SA/activated carbon mixture as well as a Ni filament type catalyst/activated carbon mixture. Of these catalysts, only the activated carbon and the mixed system containing activated carbon exhibited any change in the CF4 concentration in these experiments. No reaction took place without the presence of water vapor. In the case of activated carbon, several different kinds of activated carbon were used. In the earlier studies a high surface area species (3,100 m2/g) was used. The flow rate in these experiments was 10 cc/min with a feed of 2% CF4, 3% H2O and in balance of nitrogen (95% nitrogen). Figure 11-21 shows the concentration decrease o f CF4 versus Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 102 time. Complete destruction o f CF4 was achieved almost initially, but very rapid deactivation also took place. Furthermore, two mixed systems were investigated. A system containing zeolite SA and the high surface area activated carbon was tested as well as a mixture o f the activated carbon and a nickel filament type catalyst These studies were done at the same total flow rate o f 10 mL/min and CF4 concentration as used before. Both systems showed lower conversion than the activated carbon alone. The conversion in the case o f the nickel catalyst reached only about 20%, in the case o f the zeolite about 40 to 50%. However the power needed to get a similar conversion with the zeolite. 120 -r 100 - 2* c c 80 60 - 0 IS a> c 8 40 - 20 0-20 •60 1 r— -40 0 20 ■! i 1 I I 40 60 80 100 120 140 Time on Stream in min Figure 11-21. Decomposition of CF4 versus time using the high surface area activated carbon at 10 mL/min and 2% CF4, 3% H2O and in a balance of N2. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 103 In the more recent studies, the catalyst was changed to an activated carbon with a lower surface area. Furthermore, the flow rate was increased to IS mL/min from 10 mL/min, as was the concentration o f carbon tetrafluoride from 2 to 4% in the feed gas; the water vapor level remained constant at 3%. The analysis method was changed from the cruder method o f syringe sampling and injection into a gas chromatograph to on-line sampling with a mass spectrometer. In this new setup and under the new conditions, a different activated carbon with a surface area of 2,200 m2/g was tested as well, but conversions ranged only from 30-40% destruction of carbon tetrafluoride. Only a few reactions were performed with this catalyst. The results presented from here on are the ones for a commercial brand of activated carbon available from Sigma-Aldrich, part number 64365-11-3, a lower surface area species. Surface area data will be presented later for this species. A typical decomposition versus time plot is shown in Figure 11-22 as obtained with the mass spectrometer. The conversion in the case shown below was 74% (based on the disappearance of the CF4 signal from the mass spectrometer) as was the appearance o f one o f the major side products, carbon dioxide. The activated carbon deactivated rather quickly, as seen in the time versus conversion plot (Figure U-23). The deactivation of this species was slower and less rapid than in the case of the higher surface species (as seen in Figure 11-21). In this case, it took about 100 min for the activated carbon to deactivate, but the catalyst stayed relatively active for over 60 min. In other cases for this species similar deactivation times were observed. Generally, the deactivation time tended to be longer if the conversion (and therefore power) was lower, and shorter if the conversion was higher. In the case o f the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 104 best conversion o f 95% with this species, the deactivation time was about 60 min. The deactivation for the high surface area activated carbon was about 60-80 min, but a great deal of activity was already lost in the first 30 min o f the reaction. Similar observations regarding deactivation times were made using this catalyst type. The deactivation was also generally dependent on power as in the case mentioned above. 2e-7 Power on 2e-7 - C £ 3 CO 1®-7 £ CL 76 ra s - 8 Q. Power on 0 50 100 150 200 Tim e on stream in min Figure 11-22. Plot of CF4 decomposition versus time, top plot is the CF4 plot (dashed line), bottom plot is the CO2 plot (solid line), power was turned on to 480 W at 46 min (arrow). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 105 The dependence o f the conversion on power was also studied. As shown in Figure 11-24, the maximum conversion was achieved at medium power levels of 300-400 W (whole range 0-1200 W). The maximum conversion observed was 95% measured as the disappearance of the CF4 in the MS plot. This was in good agreement with earlier data (data not presented) that showed that the maximum conversion is in the region o f400 to 500 W. The overall conversion in that case was slightly lower than in the data shown in Figure 11-24. The maximum conversion in that set of data was 75%. 80 60 O 40 82 § o o 20 Power on 0 50 100 150 200 Time on stream in min Figure 11-23. Plot of CF4 conversion versus time, power was turned on to 480 W at 46 min (arrow). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 106 A lot o f other species were detected apart from carbon tetrafluoride and carbon dioxide in the mass spectrometric studies (as shown in Figure 11-22). The reaction produced significant amounts of hydrogen, about 10-50,000 ppm depending on reaction conditions and power level. Other gases detected were oxygen, methane, nitrogen (used as the balance gas in these reactions), methane, and water vapor. Carbon monoxide might be present but its mass coincides with nitrogen, so the signal would be buried under the larger signal for nitrogen, the predominant gas in the system with a concentration of 100 90- trendline 8070- 60 I 50- I "■ ° 30- 10- 0 100 200 300 400 500 600 700 Power inW about 94% in the feed. Nitrogen oxides might be present as well as discussed by Xu et al.u, but the mass spectrometer was not tuned to that mass, unfortunately. Figure 11-24. Dependence of the conversion on the power (black) and the corresponding trend line (gray). The hydrogen level and its influence on the reaction were investigated. Hydrogen evolved as soon as the microwave power is turned on. The hydrogen level in the mass spectrometer peaked shortly thereafter and then levels off again. Figure 11-25 shows these Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 107 peak levels at various power settings. The graph clearly shows that in a blank run o f the reaction the hydrogen level was higher than in a run with the carbon tetrafluoride present. The activated carbon catalyst was used in both reactions. 200 300 400 500 600 Pow er in W Figure n-25. Hydrogen production with (dots) and without CF4 present (squares), and using the activated carbon catalyst at various power settings. The oxygen and carbon dioxide levels were also compared in the same study. The CO2 level was slightly higher for the blank run compared to runs under normal reaction conditions. The oxygen leveled normally levels off in the entire course o f the reaction, maybe due to absorption in the water scrubber, but in two cases significant oxygen production was observed under normal reaction conditions. In the blank run, the usual pattern was observed, but a short spike in the oxygen concentration was observed when the power was switched on. The water vapor level showed a slight increase when the power was turned, probably due to water coming off the activated carbon where water Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 108 was adsorbed. Methane was evident in the system as seen in the peaks after the power was turned on at an m/z o f IS. The main methane peak is at an m/z o f 16, but this peak coincides with a secondary peak o f oxygen at an m/z o f 16. The line at an m/z o f IS was the second strongest line in the methane spectrum at about 90%. Methane may be present in the blank run but this is unknown. c. Extent o f Reaction and Characterization o f Catalyst. The extent of the reaction could also be seen, based on how the reaction tube appeared after the reaction. The higher the extent o f the reaction, the more “decomposed” was the reaction tube due to the glow and HF production as is shown in Figure 11-26. Figure 11-26. Photograph showing the extent of reaction at various power settings. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 109 The surface area o f the activated carbon was determined to be 663 m2/g by BET. The activated carbon consists of meso- and micropores. The surface area for below 20 A powder is 349 m2/g, and between 20 and 200 A, it is 280 m2/g. The surface area of one activated carbon used in one o f the reactions was measured to be 623 m2/g (BET). This trend shows a slight reduction in surface area; a probable cause will be discussed later. Ion Chromatography showed that there is a significant amount of fluoride ion in the spent catalyst after the microwave reaction and extraction o f the sample with water. The fluoride concentration was 4.1 mg/L at a minimum detection limit of 1.0 mg/L. IR analysis of spent catalysts samples showed band consistent with C-F bond stretches at about 1250, 1030 and 650 cm. However, EDX analysis did not showed the presence of fluorine on spent catalyst samples. 4. Discussion. a. Reaction Scheme and Proposed Mechanism. The water molecules have a twofold role in this reaction. Water was the primary energy absorber for the microwave radiation by dielectric heating (as discussed beforehand), but also water was the hydrogen and oxygen source needed in this reaction. A schematic o f the simplified reaction scheme o f the major reactions in the microwave heating cavity is shown in Figure 11*27. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 110 NoCE, H ,0 CFa H2» °2 h 2, o 2 H ,0 \^ c o 2 CO N, ch4 Nj hf Figure 11-27. Simplified reaction scheme for the presence and absence o f CF4, activated carbon is presented as AC. Hydrogen, oxygen, and carbon dioxide were formed in both cases. Carbon monoxide and nitrogen oxides might form [reference 12 notes that in a system such as this these gases should be expected]. Methane surely formed in the case o f CF4, but might also form without CF4. The levels maybe different. Hydrogen fluoride is only formed in the presence o f CF4.Water in the feed and on the catalyst (some comes off) was split up in both cases, when the power o f the microwave unit was turned on. The formed hydrogen and oxygen radicals or atoms and the hydroxyl radical were the active species in this reaction. A reaction mechanism involving this process is shown in Figure 11-28. In the absence of carbon tetrafluoride, the reaction to hydrogen and oxygen and the reaction to carbon oxides, after the reaction with activated carbon, were predominant. Nitrogen oxides might form as well because the nitrogen signal decreases slightly when the power was turned on, the formation was also favored under these conditions according to the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. I ll literature.12 Methane may form as well because hydrogen atoms are available to react with carbon o f the catalyst. Other products were not observed in the absence o f carbon tetrafluoride. co, co o O H H- FHF Figure II-28. Reaction mechanism for the CF< destruction. Things are somewhat different in the presence of carbon tetrafluoride. The fluorine radicals compete for the hydrogen radicals. This observation is evident from a decrease in the hydrogen signal in the mass spectrometer. Hydrogen fluoride is formed by the combination o f these two radicals. HF is captured in a water scrubber inline before HF could enter the mass spectrometer. Therefore only in two cases was HF observed as a significant signal. Methane is formed by the combination of carbon from the carbon tetrafluoride or from the catalyst (indicated as AC in Figure U-28). The oxygen signal seems to be a somewhat stronger than in the case of the blank run, favoring the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 112 combination o f oxygen molecules over carbon dioxide. This explains the slightly lower signal (reduced by a few percent) for carbon dioxide, which forms from CF4 or from the catalyst, in the case where CF4 is present Activated carbon plays an important role in the reaction. Water gets adsorbed onto the surface by means o f the surface hydroxyl groups. When the power is turned on, the adsorbed water is released and some additional water is formed by the reaction of hydroxyl groups. Therefore, activated carbon does not have an active site per se; activated carbon is more o f a mediator. However, the possibility that the activated carbon acted as an energy mediator cannot be ruled out. Mingos et al.22 reported heating of a 25 g carbon sample in a 1 kW microwave (2.45 GHz) for 1 min with a 1000 ml vented water load caused the temperature to rise from room temperature to 1283°C. This indicates that carbon is a very good microwave absorber by dielectric heating via conduction loss. Therefore, the water could be thermally broken apart by the high temperature achieved by the activated carbon and causes the generation of the proposed radical species. This would furthermore explain the observed partial deformation of the quartz tube after the reaction. Nevertheless, water still is crucial in these reactions as the reactive medium because no decomposition was observed in the absence of water. b. Deactivation. The reason for the deactivation of the catalyst is not completely understood. Initially, hydrogen fluoride formed in the reaction was thought to poison the catalyst, but infrared and EDX analyses gave conflicting results about whether there was any fluorine Reproduced with permission o f the copyright owner. Further reproduction prohibited without permission. 113 on the catalyst. However, the ion chromatography results show conclusively the presence of fluoride ion in the spent catalyst. Additionally, carbon (coke) formed in the decomposition o f the carbon tetrafluoride and deposits on the catalyst surface and hinders access to the active sites. This could be explained by the decrease in surface area, as seen in the BET surface analysis. This coking theory is supported by the literature.23 Therefore, the deactivation mechanism is a combination of coking yielding to an increase in surface area and the fluorination o f the catalyst surface. c. Glow Discharge and Zeolite Addition. The lower conversions observed in the case o f the zeolite mixture with activated carbon were due to the replacement with the zeolite that can take up a lot more water than did activated carbon. Therefore, more power was needed to release the water adsorbed in the matrix in the case o f the zeolite to achieve comparable conversion to the case without the zeolite. A similar effect was observed for the mixture of activated carbon with a nickel filament type catalyst In the glow discharge reactions, not enough energy was supplied to break up the molecules or not the matching combination o f electrode or reactor was found. As Huang et al.20 observed, it is possible to break up carbon tetrafluoride using glow and arc discharges. 5. Conclusions. The destruction of carbon tetrafluoride by microwave heating represents an efficient and easy means to eliminate this greenhouse gas. Carbon tetrafluoride was Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. almost completely decomposed. The main products were hydrogen fluoride, carbon dioxide, oxygen, and hydrogen. Furthermore, small amounts o f methane were formed and were observed. The production o f small amounts o f carbon monoxide and nitrogen oxides is expected but was not examined. The activated carbon catalyst deactivated after a short period o f time o f about 100 min. The deactivation time depends in general on the power. This deactivation seemed to due to a combination o f coking onto the catalyst and fluorination of the catalyst surface. Furthermore, water was essential in these reactions as the energy mediator caused by dielectric heating and as the source for the reactive species such as hydrogen and oxygen radicals and atoms as well as hydroxyl radicals. A reaction scheme explaining the observed phenomena was proposed and is justified based on our data. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 115 6. References. 1) Gassmann, M. Naturwissenschafien 1974,61, 127. 2) Rasmussen, R.; Penkett, S.; Prosser, N. Nature 1979,277, 549. 3) Cicerone, R. J. Science 1979,206, 59-60. 4) United Nations Framework Convention on Climate Change, Vital Climate Graphics: Climate Change, http://www.grida.no/climate/vital/index.htm. December 2002. 5) Gosselin, A.; Fradet, C. In Light Metals 2000, Proceedings o f the International Symposium on Light Metals, Ottawa, ON, Canada, Aug. 20-23, 2000; Canadian Institute o f Mining, Metallurgy and Petroleum: Montreal, Canada, 2000,353-363. 6) United Nations Framework Convention on Climate Change, Press Release: Kyoto Protocol receives 100th ratification, December 18th, 2002. 7) Marinelli, L.; Worth, W. Global Warming: A White Paper on the Science, Policies and Control Technologies that Impact the U.S. Semiconductor Industry; Technology Transfer # 9311207A-TR, SEMATECH, 1994. 8) Li, Y. E. D.; Paganessi, J. E.; Rufin, D. ACS Symposium Series 2001, 766, 62-75. 9) Weston, R. E. Jr. Atmos. Environ. 1996,30(16), 2901-2910. 10)Mackay, G. I.; Karecki, D. R.; Pisano, J. T.; Schiff, H. I. In Light Metals 2000, Proceedings o f the International Symposium on Light Metals, Ottawa, ON, Canada, Aug. 20-23, 2000; Canadian Institute of Mining, Metallurgy and Petroleum: Montreal, Canada, 2000, 339-352. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 116 11) Wei, T.-C.; Fang, Y. S Abatement o f CF4, CiFt and CHF3 in Microwave Generated Surface Wave Plasmas, 29th IEEE International Conference on Plasma Science, 2002. 12)Grytsinin, S. I.; Korchagina, E. G.; Kossyi, I. A.; Misakyan, M. A.; Silakov, V. P.; Tarasov, N. M.; Temchin, S. M. Plasma Sources Sci. Technol. 2001,10, 125-133. 13)Tonnis, E. J.; Graves, D. B.; Vartanian; Beu, L.; Lii, T.; Jewett, R. J. Vac Sci Technol. A 2000,18,213-231. 14) Xu, X.; Rauf, S.; Kushner, M. J. J. Vac Sci Technol. A 2000,18,393. 15)Takita, Y.; Ninomiya, M.; Miyake, H.; Wakamatsu, H.; Yoshinaga, Y.; Ishihara, T. Phys. Chem. Chem. Phys. 1999,1, 4501-4504. 16)Kanno, S.; Dceda, S.; Yamashita, H.; Azuhata, S.; Irie, K.; Tamata, S. Mater. Res. Soc. Symp. Proc. 1998,497, 59-64. 17) Burdeniuc, J.; Crabtree, R. H. Science 1996,271, 340-341. 18) Lee, M. C.; Choi, W. Environ. Sci. Technol. 2002,36, 1367-1371. 19) Jacobsohn, L. G.; Franceschini, D. F.; Maia da Costa, M.E.H.; Freire Jr. F. L. J. Vac. Sci. Technol. A 2000,18. 2230-2238. 20) Huang A.; Suib, S. L. Res. Chem. Interm. 2001,27, 957-974. 21) Spiess, F.-J.; Suib, S. L.; Hayashi, Y.; Matsumoto, H. unpublished results. 22) Mingos, D. M. P.; Baghurst, D. R. Chem. Soc. Rev., 1991,20, 1-47. 23) Wiersma, A.; van de Sandt, E. J. A. X.; Makee, M.; Moulijn, J. A. Appl. Catal., A 2001,272. 223-238. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER III. GENERATION OF HYDROGEN FROM WATER AND METHANE USING DISCHARGE PLASMA A. HYDROGEN PRODUCTION FROM WATER AND METHANE. 1. Introduction. The demand for energy will further increase this century due to economic and population growth. The increase is furthermore caused by increased industrialization and urbanization in developing countries which are catching up to the industrialized countries. It has been estimated that the global energy demand will rise by 1.7% per year until 2030.Less than the growth o f 2.1% that has occurred over the last 30 years. As seen in Table III-1, fossil fuels already meet most o f the present energy demands and are expected to meet 90% o f the increased demand.1 Fossil fuels are coal, crude oil and natural gas. The reserves and supply of these fossil fuels are not limitless; they are non-renewable. New sources are found on a constant basis; however in some cases it is not economical to exploit them. But with dwindling supplies in the coming decades, it will become necessary to exploit less oil rich deposits more intensely than now and even deposits under oceans as done in the North Sea; this would certainly make these fuels more and more expensive.2 This could lead to the exploitation o f ‘unconventional’ fossil fuels such as tar sands, oil shale and gas hydrates.3 Estimates on how long fossil fuels will last differ greatly. A good assumption under current growth conditions is that gas and crude oil will last for about 60 years, and coal will last well into the next century regardless of environmental and price concerns.2 117 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 118 Table III-l. World primary energy demand history and projection until 2030 in Million tons o f oil equivalent (Mtoe) (Adapted from Ref. 1). Fuel 1971-2000 (Mtoe) 2000-2030 (Mtoe) Oil I 140 2 140 Gas I 190 2 120 Coal 915 1 260 Nuclear 645 20 Hydro 120 140 Other Renewables 145 380 Apart from the increasing price of fossil fuels which we greatly depend on especially crude oil and natural gas, environmental impact is a major concern these days. The burning o f fossil fuels caused and still causes major air pollution problems. Trends for several major air pollutants are shown in Table m-2. The use of sulfur rich fuels causes acid rain and the destruction of forests, buildings and the acidification of lakes leading to the collapse of the aquatic ecosystem. Respiratory problems can also arise from sulfuric acid aerosols. Carbon monoxide (CO) is of concern because CO is an asphyxiating poison. Nitrogen oxides and volatile organic compounds (VOCs) are indirect air pollutants. They are the main components o f the formation process of tropospheric ozone or photochemical smog. This photochemical smog causes respiratory problems by the action of the formed ozone molecules. PMio classifies the pollution group o f particulate matter with a diameter smaller than 10 pm. These are o f concern due to their effect on the respiratory system, especially if they carry toxic or dangerous chemicals; the smaller the particles the deeper the penetration into the lung causing the greatest potential health risk.2 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 119 Table III-2. Trends of various air pollutants over the past 30 years in the United States in million tons/year (Adapted from Ref. 4). Pollutant 1970 1980 1989 1999 SO2 28 22 20 16 CO 117 106 96 88 NO, 18 21 20 21 VOC 28 23 20 16 PM,0 N/A N/A 36 23 Government actions have significantly reduced the amount of air pollutants by legislation such as the Clean Air Act o f 1990 (see trends in Table HI-2). The greatest environmental problem, however, does not stem from these air pollutants but from the end product o f every combustion process, carbon dioxide. Carbon dioxide is one o f the major contributors to global warming (details on global warming were described in the previous chapter). Solutions to decrease the CO2 emissions are important but at the same time a balance of enough energy to drive the economical and technological progress of mankind is needed.3 There are several different ways to achieve the goal of cutting CO2 emissions. First o f all, the efficiency of contemporary processes such as automobiles or photovoltaic cells can be improved but for many cases efficiency limits have been reached such as in turbines. Decarbonization is the reduction o f carbon emission per released energy unit such as in the shift from coal to the less CO2 intensive natural gas.5 Sequestration can be done by planting trees to absorb CO2 or store CO2 in the deep ocean; the long term effects such as sudden release or aquatic implications o f this are unknown.3,5 The use of renewable energies such as photovoltaic, solar thermal, biomass, wind, hydropower, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 120 ocean and geothermal, and tidal technologies has to be expanded and made competitive to fossil fuels to have an impact on CO2 emission reduction. The use of nuclear fusion and fission are alternatives as well but are highly controversial due to nuclear waste issues.5 In the eyes o f many people, hydrogen is viewed as the fuel of the future being a carbon-neutral fuel. In this context, fuel cells come into focus as the energy supplier in a hydrogen based energy technology. Fuel cells are considered to be the future of transportation and energy production. In a fuel cell, hydrogen and oxygen are consumed and produce only water as an exhaust gas.s In order for the fuel cells to become more competitive, the cost of hydrogen production has to be much less. There are several ways hydrogen can be produce today. Electrolysis, water splitting via solar energy or thermochemical water splitting can be used as described in a review by Czuppon et al.6. On an industrial scale, hydrogen is produced by steam reforming natural gas.5 Licht et al.7,8 report the possibility o f over 18% conversion o f water by solar energy to hydrogen fuel based on calculations and experiments involving RuS2 as one of the catalysts. Other materials such as oxide semiconductor photocatalysts, for example doped InTa04 or Ti0 2 (modified or in conjunction with sonication) have been investigated as well.9' 11 Miller et al.12developed multi-junction thin film photoelectrodes containing amorphous silicon or copper-indium-gallium-diselenide with expected solar to hydrogen conversions of over 10%. Mechano-catalytic methods and biological hydrogen from fuel gases and water are promising concepts, as well.13 T-Raissi14 analyzed new hydrogen production technologies such as thermochemical H2S reformation o f methane with and without the use of a solar component and catalyzed micro-reformers for the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 121 decomposition o f ammonia. Algae have been observed to produce hydrogen by photosynthesis and hydrogenase activity in the presence o f background oxygen which inhibits hydrogen production at high levels.15 Venugopalan et al.16 studied electrical discharge o f water vapor in the context of hydrogen peroxide production which is detected on cold surfaces where the exhaust gases from the plasma condensed. Fujii et al.17 used methane oxygen plasmas and found hydrogen peroxide but they also showed that in preliminary experiments no evidence o f this compound in a microwave water discharge plasma. Maerk18 studied the dissociation degree of water vapor at various pressures using a hollow-cathode discharge. MelikAslanova et al.19 used plasma discharges burning in supersonic flows to decompose water with a conversion o f about 20% Our research group previously worked on water splitting with discharge plasmas.20-22 In two cases20,21 argon was the diluent gas; in the other,22 the diluent was helium. In this work, methane was used as the diluent with admixtures of nitrogen or argon, and in some cases nitrogen as the diluent. 2. Experimental Section. a. Preparation o f Gas Mixtures. Gases used were obtained from commercial vendors such as Airgas. Nitrogen and Argon were o f ultra high purity grade. Methane was o f CP grade (99.5% pure). Water vapor was introduced into the reaction system by the gas mixtures from a bubbler filled with deionized water, yielding a concentration o f 3% at room temperature of 25°C and atmospheric pressure. The content of the feed gas is 3% water (due to its vapor pressure Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 122 at room temperature) and in a balance o f nitrogen or methane with admixtures o f nitrogen or argon as needed at flow rates varying from 5-250 mL/min for the plasma reactions. 10cm Pd Electrode 10cm Cu Electrode Figure III-l. Photograph showing Pd and porous Cu electrode. b. Tubular Plasma Reactor. The water splitting reactions were carried out in a Plasma And Catalysis integrated Technologies (PACT) tubular type reactor. The reactor consists o f three parts. The inner electrode was either a set of two porous copper electrodes or a Pd electrode both having diameters of 8 mm and lengths 6 cm (porous part) and 10 cm respectively (as shown in Figure III-l); these two electrode materials were used in the experiments unless otherwise noted. The Pd electrode used was dumbbell shaped and 3 cm each on the wide part and 4 cm on the narrow part (diameter 6 mm). A quartz glass tube o f 10 mm inner and 12 mm outer diameter, which served as a dielectric, was used to separate the inner from the outer electrode. The outer electrode is wrapped around the glass tube and is subsequently wrapped with copper wire to achieve a close coverage o f the glass tube in order to achieve a uniform plasma. Aluminum foil of 13.5 cm length was used for this purpose yielding a plasma zone o f the same length. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 123 c. Experimental Setup and Circuit Design. A simple setup was used as shown in Figure III-2. The setup differed slightly between the porous copper electrode and the Pd electrode setup with the copper tube also functioning as the flow tube (Figure III-3). The gases used were mixed online and water introduced by a bubbler at atmospheric pressure and room temperature. Porous, Hollow Cu electrode Outer Electrode, Al Figure III-2. Schematic Setup for tubular PACT reactor. Pd Reactor Cu Reactor Figure ID-3. Photographs of porous Cu electrode and Pd electrode setup. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 124 A circuit diagram is shown in Figure III-4. This setup was used in the later experiments at higher voltages, the only difference compared to earlier setups being that the resistor before the reactor was removed. The voltage was supplied by an UHV-10 AC high voltage power supply (PS) at frequencies between 8.0 and 8.1 kHz. Input voltage and input current values were measured by a DL-1540 Yokogawa oscilloscope using a high- (Tektronix P6025) (HVP) and a low-voltage probe (Yokogawa 70996) (LVP), respectively. The voltage across a 100 Q resistor (Rs) placed in series with the reactor measured by the LVP was equivalent to the input current. The resistor Rl (lOOkfl, 225 W) was the load resistor parallel to the reactor, preventing too much voltage from reaching the reactor R. AC Voltage values between 1.000 and 2.600 kV were applied to the electrodes to produce plasmas; these voltages were rms (root mean squared) voltages. The system was equilibrated at the beginning o f a run and then the power supply was turned to the desired level and continued until an equilibrium value was achieved. AA/V PS= Power Supply R= Reactor HVP= High Voltage Probe LVP= Low Voltage Probe R,= Standard Resistor Rl= Load Resistor Figure III-4. Circuit diagram for the tubular plasma reactor. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 125 The fan PACT reactor used a similar circuit and experimental setup. The voltage was also supplied by the UV-10 power supply. The dimensions and characteristics o f the fan reactor are the same as mentioned in chapter II o f this thesis. The rotor and stator used in these experiments were coated by a thin layer o f gold (about 100 pm thick). The description of the T-shaped reactor can be found elsewhere.24 d. Product Analysis and Characterization. The reaction mixture was continuously monitored with a MKS-UTI PPT quadrupole residual gas analyzer mass spectrometer with a Faraday cup detector and a variable high-pressure sampling manifold. In earlier experiments, a Hewlett Packard 5890 Series II gas chromatograph (GC) equipped with a Carboxen-100 (45/60 mesh) column and a thermal conductivity detector (TCD) were used. The electrodes used were provided by Dr. Hayashi o f ASE or designed and assembled in our lab. Deposits on the electrodes (as seen in Figure III-3) were analyzed at the Environmental Research Institute at the University o f Connecticut. These samples were characterized using a Perkin Elmer 2400 carbon, hydrogen, nitrogen analyzer and a OI Analytical 700 total organic carbon analyzer. 3. Results. a. Water Splitting Using Various Reactors. In early experiments, a gold fan PACT reactor was used. This study was done at various flow rates. These experiments showed that a high hydrogen yield o f 24% could be achieved at 25 mL/min at an applied voltage o f 500 V (V„m) (Figure III-5). The Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 126 conversions o f hydrogen turned out to be just a few percents when ultra-high purity air was used instead o f nitrogen. 25 -| • • 20 - • • ge c • • • 15- 1 10>t ru * 50 - 25 cc/min 50 cc/min 10 cc/min • I 0 1 20 1 40 "" 1 60 1 80 i 100 120 time in min Figure I I I - 5 . Water splitting at various flow rates, Au Fan, V „n s= 5 0 0 V 3.0%H2O in N 2. In other experiments, a tubular reactor with copper being the active electrode was used in a setup similar to the Pd electrode setup. The initial conversions were very low (around 1%). Heating o f the water bubbler to 65°C yielded very low conversions due to condensation in the colder reactor and the sampling line. Changes in the setup such as cooling of the plasma zone and using the water bubbler at room temperature increased the conversion further, the conversion went up from the initial 1% to a 8-10% hydrogen yield. Changing the flow rate to 45 mL/min from the initial value o f 100 mL/min increased the conversion to 11% compared to 2.5% at 100 mL/min. When the electrode material was changed to platinum, the hydrogen yield was in the same range (8-10%) but it was slightly lower. However, the conversion was lower at Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 127 the same applied voltage (maximum of 8.7%) than that for copper (maximum of 10.5%), but the conversion was highest at high flow rates (100 mL/min), unlike for copper (45 mL/min). A change in the applied voltage further increased the hydrogen yield. A voltage of 2.6 kV, which was used in the comparison with copper, turned out to be too high. The optimum value was 2.0 kV where the hydrogen yield was a steady 14%. In a T-shaped tubular reactor, the conversions were smaller than in the previously mentioned cases. b. Water Splitting Using a Porous Cu Reactor in a Nitrogen Atmosphere. The experimental conditions were similar to the ones used in earlier experiments. The peak-to-peak voltage used was 1.0 to 1.1 kV, with Vnns o f220 to 235 V, and currents around 200 mAmps yielding a power of about 5 Watt. The electrode used was the porous hollow copper electrode in all these experiments. In these experiments, water (3.1%) was used in a balance of nitrogen. The reaction was observed at various flow rates ranging from 5-250 mL/min, and the voltage (peak-to-peak) was kept at a constant 1.022 kV. A calibration was carried out before and after the experiment and yielded the same values using prepared standard mixtures. The hydrogen production, in this case as volume-%, is shown in the Figure III-6. The production decreased in an exponential fashion as the flow rate was increased similar to the results in the water/methane system; the production of hydrogen is however significantly lower. There may be a maximum conversion in the range around 3 mL/min, but the lowest achievable flow rate was 3.5 mL/min; below that it was difficult to achieve flow due to the rotameter used in the setup. The highest hydrogen production achieved was about 1.9% at a water input of 3.1%. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 128 2.0 1.8 1.6 1.4 <M X 1 -2 ^ 1.0 5 0.8 0.6 0.4 0.2 0.0 0 50 100 150 200 Flow rate in mL/min Figure III-6. Hydrogen production (vol%) using a feed of 3.1% water in nitrogen at various flow rates. Because water was the only possible hydrogen source, the hydrogen yield could be calculated based on the assumption that a complete reaction yields 3.1 vol% hydrogen. This yield with respect to the applied flow rate is shown in Figure III-7. The yield followed the same trends as the volume yield of hydrogen. The highest yield achieved was about 60%. Reproduced with permission o f the copyright owner. Further reproduction prohibited without permission. 129 70 60 50 >8 £ 40 2 © ’>* CM 30 X 20 10 I 0 0 50 100 150 200 Flow rate in mL/min Figure III-7. Hydrogen yield in % using a feed of 3.1% water in nitrogen at various flow rates. c. Water Splitting Using a Porous Cu Reactor in a Methane Atmosphere. These reactions were carried out under the same conditions as the water/nitrogen reaction. The same porous copper electrode was used. Instead o f a balance of nitrogen, a balance of methane was used in these reactions. The effect o f different flow rates was investigated in the range from 3-200 mL/min. The voltage was kept at a constant 1.020 kV (peak-to-peak). Calibrations were carried out with standard mixtures o f hydrogen in a balance o f methane. The hydrogen production (here as volume-%) is shown in the Figure III-8. An exponential decline with respect to the applied flow rate was observed. There may be a Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 130 maximum conversion in the range around 5 mlVmin, but the lowest achievable flow rate was S mL/min; below that it was difficult to achieve flow because o f the rotameter used in the setup. The highest hydrogen production achieved was about 15% (total volume percent o f the mixture). This is not the conversion o f water or methane to hydrogen. The input concentration of water is 3.1% (by volume). 16 14 12 C I 4 2 0 « 1U 1U 2U S I S I M 41.6 53.1 63 6U 106 123 136 143 186 166 214 231 flow rat* (mL/min) Figure III-8. Hydrogen production (vol%) using a feed o f 3.1% water in methane at various flow rates. The decomposition o f methane follows the same trend as the hydrogen yield as seen in Figure HI-9 (squares). The methane decomposition is highest at low flow rates at about 35%. The decomposition rate o f methane is lower than the hydrogen yield calculated from methane as seen in Figure III-9 (dots). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 131 40 - e£ C c & 2 30 - 20 - 10 - 1 3 0 50 100 150 200 250 Flow rate in mL/min Figure III-9. Decomposition o f methane in relation to flow rate (squares) and conversion o f water and methane to hydrogen in relation to flow rate (circles). d. Effect o f Admixtures to the Methane/Water System. In this set o f experiments, the flow rate was set at 20 mL/min. This time, argon, was mixed in different ratios with methane. A significant effect was observed on the conversion in this case as seen in Figure 111-10. The conversion dropped and was almost constant in the range from 25 to 75% by volume argon at the set flow rate. The same experiment was carried out, at the exact flow rate. In this case nitrogen was mixed with methane at different ratios (Figure III-l 1). The results are comparable to the ones obtained in the argon case, but the conversion was slightly lower in the nitrogen case. The conversion in pure nitrogen was significantly lower than in the case o f methane alone; in the case of argon it is slightly higher. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 132 20 - * c c 15- & e 5c - 10- 5 □ • 0O 0 20 100 80 60 40 Decomposition Conversion % ch 4 Figure 111-10. Effect of admixture of argon on the hydrogen production at 20 cc/min. 20 ■S c o - 15- § so □ e 0 60 20 60 40 Decomposition Conversion 60 100 % ch 4 Figure 111-11. Effect of admixture of nitrogen on the hydrogen production at 20 cc/min. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 133 e. Reactions o f the Water/Methane System at Higher Voltages. For these experiments, the setup shown in Figure III— 4 was used. This modification resulted in higher input voltages which were about 10 times higher than in the water/methane reactions in the previous section. The input voltage used was about 2.2 kV (Vra) compared to about 230 V in the other cases. This resulted in currents of about 30 mA and a power use o f about 120 W. The yields and the hydrogen production were higher than in previous experiments under these conditions, as expected. Hydrogen yields over 100 % were achieved in the water/nitrogen system. At a flow of 20 mL/min rate, a yield o f 5.6 vol% hydrogen was achieved, and 10.2 vol% hydrogen at a flow rate of 15 mL/min. The theoretical yield was 3.1 vol% hydrogen based on the water input o f 3.1 vol%. A blank run with just nitrogen in the feed yielded 0.58 Vol-% hydrogen at a flow rate of 20 mL/min. In the case o f just air in the feed, a hydrogen volume yield of 0.06% was observed at the same flow rate. In a different set o f experiments, the influence of the methane content in a balance o f air was investigated. No water was added to the feed in this case and the flow rates ranged from 15-35 mL/min, but were mostly at 20 mL/min. The voltage was set at 2.2 kV (Vrms). The two different electrodes, porous copper and the dumbbell shaped Pd electrode, were used resulting in the use o f a slightly different reactor design. Linear trends were observed in the hydrogen yield with respect to the methane content in the feed (Figure III-12 and 111-13). The maximum conversion at applied flow rates were around 30% conversion. The conversions were slightly higher in a porous copper reactor compared to conversions with an electrode coated with palladium. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 134 12 - 10 ■ CM 12 14 16 18 24 22 20 26 Methane Content in % Figure ID-12. Hydrogen production (vol%) using a feed o f varying methane content in air at 2.2 kV (rms), 20 cc/min flow, porous Cu reactor. 30 25- 10 - 10 20 30 40 50 60 70 Methane Content in % Figure HI-13. Hydrogen production (vol%) using a feed o f varying methane content in air at 2.2 kV (rms), 20 cc/min flow, Pd reactor. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 135 f Mass Spectroscopic Studies. All analysis in later experiments was done by on-line mass spectrometry. Usually, the mass spectrometer (MS) was tuned to specific m/z ratios such as 2 for hydrogen and 44 for carbon dioxide, and these masses were monitored over the equilibration and reaction periods. Additionally, the whole range of the MS, m/z ratios between 1 and 200, was scanned. The reaction products depended on the gases used in the reaction. In the typical reaction system of water in methane and nitrogen a variety o f different products were observed apart from the unreacted input gases (Figure QI-14). Nitrogen oxides (NOx), (m/z ratios o f 30 and 46), carbon monoxide (28), carbon dioxide (44), hydrogen (2), oxygen (16 and 32), hydrogen cyanide (26 and 27) and various hydrocarbon species were found in these spectra. The hydrocarbons observed were CnH^i species which were consistent with alkanes (CnH2n+2)- The peaks at m/z ratios of 39,41,43, 57, and 71 could be attributed to alkanes such as butane (B), pentane (P), hexane (Hx), heptane (Hp), their isomers and other higher alkanes. The corresponding alkenes were also part o f the m/z ratios of 39 and 41 and are part o f the systems at m/z ratios of 57 and 71 with m/z of 56 and 70. When just water and nitrogen are used, the mass spectrum is simpler (Figure HI15). There was less overlap from various peaks and less species were detected compared to the previous system. Nitrogen oxides (NOx), carbon monoxide, carbon dioxide, hydrogen, and oxygen were also found as products in this case as well as the input gases nitrogen and water. Only traces o f the mentioned hydrocarbons were found in this spectrum. Hydrogen cyanide did not seem to be present in this case. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 136 1.50608 1.406-08 1.30608 1.20606 Hx 1.10606 | 1.00608 £ 9.00608 i 8.00609 7.00609 j i Hx 6.00609 5.00609 4.00609 Hx 3.00609 Hx 2.00609 1.00609 0.006*00 ...............U r r r , .........\ t<C 1 , ......................................... wt f h ....... 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96 101 m/z Figure III-14. Mass spectrum o f the methane/water/nitrogen system. 2.00600 1.00600 1.00600 ^ 1.40600 € 120600 • 1.00600 & 0.00606 ^ 0.00606 a. 4.00600 2.00606 0.006*00 m/z Figure 111-15. Mass spectrum of the water/nitrogen system. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 137 In the system o f just air and methane, the input gases were the major peaks detected in this mass spectrum (Figure III-16). Significant amounts o f nitrogen oxides (NOx) and carbon monoxide were formed. Furthermore, carbon dioxide, hydrogen, hydrogen cyanide, and CH species o f higher molecular weight such as C4H4 (l-buten-3yne or cyclobutadiene) at a m/z of 52 (this could also be cyanogen (C2N2)), CNH species, and CHO species were observed. Dinitrogentrioxide was attributed to a m/z of 60. The peak at an m/z o f 34 showed that there were small amounts o f hydrogen peroxide formed in this reaction. I NO, CO COz 1 6 0 13 17 21 28 29 33 37 41 46 87 61 66 06 73 77 61 68 80 03 97 Figure 111-16. Mass spectrum o f the air/methane system. g. Analysis o f Carbon Deposits on a Cu Electrode. Carbon deposits form on the copper electrode. These deposits were collected and sent in for analysis. These deposits were difficult to remove from the electrode, especially Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 138 due to the brittle character o f the electrode. Carbon, hydrogen, nitrogen analysis and total organic carbon analysis were performed on these samples. This analysis found that 76.1% carbon and 8.22% hydrogen are present in this sample. This resulted in an empirical formula for carbon and hydrogen o f C3H4. These deposits are also observed on the other electrodes used and are usually washed off if the electrode is to be used again; deposits can be clearly seen on the Pd electrode in Figure III-l. 4. Discussion. a. Water Splitting in a Water/Nitrogen System. In the experiments involving just water and nitrogen, several different electrodes and reactor types were used. A gold fan reactor showed the best conversion with 24% in early experiments whereas in later experiments only the porous copper system achieved a higher conversion with 60% at a very low flow rate o f 3.5 mL/min; this system will be commented upon later. All other system showed conversions o f 14% or lower with a tubular Pt electrode being the best; even a T-shaped tubular reactor showed lower conversions. The difference in conversions was discussed by Luo et al.22,23 for a similar system using helium or argon as the diluent gas and comparisons to nitrogen were made in terms o f energy transfer based on optical emission data. The transfer o f energy in the plasma from the diluent to water is less efficient for nitrogen than it is for argon or helium. Additionally, the dissociation of water into its elements is a highly non-spontaneous and endothermic process (AH0 = +241.8 kJ/mol and AG° = +228.6 KJ/mol at 25°C).25 The Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 139 conversions are therefore expected to be low. The observed trend in increasing conversion with increasing voltage can be explained in terms o f energy transfer. If the voltage is increased more power and therefore more energy is transferred to the plasma. The electrons have higher energy, and the collisions with nitrogen yield more activated nitrogen molecules which are also more energetic. As a result, more energy is transferred to water molecules in an increased number o f collisions. More water molecules dissociate and the probability to form hydrogen is therefore higher. Luo et al.22 also proposed a mechanism (shown in equations 1-12 below) for their similar system explaining the influence of the metal electrode. This mechanism is valid in the case of nitrogen as well. The metal electrode participates by scavenging oxygen radicals (equations 10-12). Noble and stable metals such as platinum show higher activities because no or only a minute oxide layer is formed. The fan reactor achieves activation at higher speeds, and oxygen radicals recombine more readily compared to the results in a tubular reactor due to the mixing and rotation o f the fan. N2+ e - » N 2*/N2*+(1) H20 + N2*/N2*+ -» H20* + N2 (2) H20 * H* + HO* (3) H O + N2*/N2*+ H* + O + N2 (4) 2H «-»H 2(5) 20 Oz (6) 20H* -» HzOz (7) 0 2 + 0 - > 0 3(8) O + 0H» ■> H20 (9) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 140 0* "*■Mnftct (MOJmftte (10) x O + M w k ^ C M O J h id l) 2 O + M -> O2 (1 2 ) When the water/nitrogen system was used, these runs were performed after runs with water and methane. The system was purged for a period o f time with nitrogen or oxygen to eliminate deposits, but the hydrocarbon deposits were not probably removed completely. This could explain the higher conversion when this system was used in the porous copper reactor. The hydrogen yield o f more than 100 % was definitely due to hydrocarbon deposits because in that case no purging of the reactor was performed after the reactions involving methane. b. Water Splitting in a Water/Methane System. Methane requires less energy to be broken up into its elements (AH0 = +74.8 kJ/mol and AG° = +50.7 KJ/mol at 25°C).25 The enthalpies for the reactions of methane and water to hydrogen and carbon monoxide or carbon dioxide are +206.1 KJ/mol and +164.9 KJ/mol, respectively. From an energy standpoint, the reaction to coke seems to be preferred over the reaction with water vapor to the carbon oxides. Mass spectrometric data suggested that the formation of carbon monoxide is preferred over the formation of carbon dioxide. This conclusion was based on the increase in the m/z ratio of 28 which is the peak for nitrogen and carbon monoxide, but because the peak at an m/z of 14 actually decreases, this increase was due to the formation o f carbon monoxide. This increase was bigger than the increase observed for the carbon dioxide peak at an m/z of 44. It was hard to make a judgment if the coke Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 141 formation is preferred over the formation o f the carbon oxides due to the difficulty o f establishing a carbon balance. The deposits on the electrodes also contain hydrogen and other species. Thus, it was hard to get them completely off the electrode for elemental analysis. Coke formation takes place and all these three reaction contribute with different weight to the formation o f hydrogen. A hydrogen yield based upon the input o f water and methane was difficult to define. A mechanism for the reaction o f methane in the presence o f water can be suggested (equations 13-38). A detailed mechanism for the formation o f hydrocarbon will be presented in the second part o f this chapter. The mechanism starts by electron activation of methane and shows the formation of coke, the carbon oxides, hydrogen, and oxygen, as well as ozone and hydrogen peroxide; of these two species only hydrogen peroxide was detected in minute amounts. This mechanism is justified by the observation o f the aforementioned products as well as the observation of various hydrocarbon species such as alkanes and alkenes (which will be discussed in the second part o f this chapter in detail). Furthermore, good evidence for this mechanism is the elemental analysis o f the carbon deposits on the electrode yielding a carbon to hydrogen ratio of 3:4, indicating saturated and aromatic hydrocarbons. CH4 + e CH4 */ CH4*+ (13) CH,*/ CH4*+ -> CH3-/ CH3-++ H- (14) CH3-/ CH3-++ CH4 */ CH4 *+-» CH4 + CH3-*/ CH3-*+ (15) CH3-*/ CH3-*+-» CH2-/ CH2*++ H- (16) CH:«/ CH2«++ CH4 */ CH4 *+-» CH4 + CH2'* / CH2-*+(17) CH2-*/ CH2-*+-> CH-/ CH-++ H- (18) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CH*/ CH*+ + CH4 */ C H S** CH4 + CH**/ CH**+ (19) CH**/ CH**+-> C*/ C*++ H* (20) H20 + CH4 */ CH4*+ * H20* + CH4 (21) H20* H* + HO* (22) HO* + CH4 */ CH4 *+ “> H* + O* + CH« (23) 2H* * H2 (24) 2 0 -> 0 2 (25) 20H* -> H202 (26) 0 2 + O* O3 (27) O + OH* -> H20 (28) CH* + O* CO* + H* (29) CO* + O* -> C 02 (30) CH* + OH*->CO + H2 (31) C 0 + 0H *-> C 0 2 + H* (32) C* + O - » C 0 (33) CCH O* -> COz* (34) C 02*-> C 02 + e (35) O* + Mjujfjce -> (MO)suHace (36) X O* + Mbulk * (M O x)bulk(37) 2 O* + M 0 2 (38) When oxygen or nitrogen (in excess) is added, this mechanism has to be supplemented by reactions accounting for the formation of the nitrogen oxides. The carbon oxides are already accounted for, but the formation of oxygen radicals has to be Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 143 added. These reactions are shown in equations 39-46. The activation of water in this case can also take place by equations 1-9. The reaction to hydrogen cyanide is also accounted for in this addition to the general model. 02 + e'-» 0 2 * (3 9 ) Cb* * 2 O (40) N2+ e '-» N2*/N2*+ (41) N2*/N2*+ -» 2 N*/ N*+ (42) N«/ N»+ + O NO/ NO+ (43) NO/ NO+ + O -> N 02«/ N 02,+ (44) N 02«/ N 02*+ -> N 02/ N02+ + e* (45) N 02 + NO -> N20 3 (46) CH* + N -» HCN (47) c. Effects o f Admixtures and Effects o f Flow Rate. The admixtures of argon and nitrogen to the water/methane system caused decreases in the decomposition o f methane and conversion to hydrogen. The decomposition and conversion dropped due to the dilution effect compared to the case when only methane and water were used. The conversion was however fairly constant between 25 and 75% methane content. This could be due to more effective energy transfer of the diluent to methane and water. In the case of nitrogen, all these levels are lower because nitrogen is less effective in energy transfer than argon is. As seen in the plots o f hydrogen production versus flow rate in the case of water and methane and water and nitrogen, the hydrogen level dropped almost in a negative Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 144 exponential trend. The flow rate F is related to the residence time tn by the following equation 48, with Vri.«^. being the plasma volume, usually on the order of 1 mL in tubular reactors. tfeS= Vpliggi / F (48) Consequently, the higher the flow rate is, the lower is the residence time. The conversion or hydrogen production is directly related to the residence time. At low residence times, the molecules have enough time to dissociate and to recombine and therefore the conversion is high, but at higher residence time, there is not enough time for all these reaction to take place and the conversion is lower. The negative exponential decline can be explained by ltinetic factors. Due to the lower residence time, mass transfer problems arise, and the contact time of the reactants is too low for the kinetics which are slow on this timescale. Drastic changes in the concentration profile around the electrodes at high flow rates also cause the negative exponential decline. 5. Conclusions. The decomposition o f a combination o f methane and water by discharge plasma represents a promising way to produce hydrogen. Hydrogen production of up to 30 vol% at higher voltages is shown. Hydrogen yields in systems only containing water and nitrogen were up to 60% at low rates and high power, and the yields suggested that certain electrodes and reactor types were preferential for the production o f hydrogen. The dependence o f the hydrogen yield and hydrogen production as well as the methane decomposition on the applied flow rate was studied. These parameters demonstrated a negative exponential decline with respect to the flow rate which is expected from kinetic Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 145 models. Furthermore, the effects of admixtures of nitrogen and argon are demonstrated. The main products were hydrogen, coke and carbon monoxide and dioxide; if air or nitrogen is added nitrogen oxides and hydrogen cyanide were also major products. A reaction scheme is suggested and seemed validated by the species found in the mass spectra and on the deposits. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 146 6. References. 1) World Primary Energy Demand in World Energy Outlook 2002, Organization for Economic Cooperation and Development, Paris France, 2002. 2) Spiro, T. G.; Stigliani, W. M. Chemistry o f the Environment 2nd edition, Prentice Hall, Upper Saddle River, NJ, 2003. 3) Dresselhaus, M. S.; Thomas, I. L. Nature 2001,414, 332-337. 4) Air Quality in GEO: Global Environment Outlook 3 (Original Source: US EPA 2001), http://www.grida.no/geo/geo3/enelish/index.htm. January 2003. 5) Hoffert, M. I.; Caldeira, K.; Benford, G.; Criswell, G.; Green, C.; Herzog, H.; Jain, A. K.; Kheshgi, H. S.; Lackner, K. S.; Lewis, J. S.; Lightfoot, H. D.; Manheimer, W.; Mankins, J. C.; Mauel, M. E.; Perkins, L. J.; Schlesinger, M. E.; Volk, T.; Wigley, T. M. L. Science 2002,298, 981- 987. 6) Czuppon, T. A.; Knez, S. A.; Newsome, D. S. in Encyclopedia o f Chemical Technology', Kirk-Ohmer; 4th ed.; Wiley: New York, NY, 1991; Vol. 13. 7) Licbt, S.; Wang, B.; Mukeiji, S.; Soga, T.; Umeno, M.; Tributsch, H. Int. J. Hydrogen Energy 2001,26, 653-659. 8) Licht, S.; Ghosh, S.; Tributsch, H.; Fiechter, S. Sol. Energy Mater. Sol. Cells 2002, 70, 471-480. 9) Zou, Z.; Ye, J.; Sayama, K.; Arakawa, H. Nature 2001,414, 625-627 10) Abe, R.; Sayama, K.; Domen, K.; Arakawa, H. Chem. Phys. Lett. 2001,344, 339344. 11)Harada, H.; Hosoki, C.; Kudo, A. J. Photochem. Photobiol., A 2001, 141, 219224. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 147 12) Miller, E.; Rocheleau, R. Proceedings o f the 2001 U.S. DOE Hydrogen Program Review, Baltimore, MD, United States 2001,359-381. 13)Hitoki, G.; Takata, T.; Ikeda, S.; Hara, M.; Kondo, J. N.; Kakihana, M.; Domen, K. Catal. Today 2000,63, 175-181. 14)T-Raissi, A. Proceedings o f the 2001 U.S. DOE Hydrogen Program Review, Baltimore, MD, United States 2001,879-905. 15) Lee, J.; Greenbaum, E. Proceedings o f the 2001 U.S. DOE Hydrogen Program Review, Baltimore, MD, United States 2001,31-44. 16) Venugopalan, M.; Jones, R. A. Chem. Rev. 1966,6 6 ,133-160. 17) Fujii, T.; Iijima, S.; Iwase, K.; Arulmozhiraja, S. J. Phys. Chem. A 2001, 105, 10089-10092. 18) Maerk, T. D. Acta Phys. Austriaca 1978,49(1), 67-70. 19) Melik-Aslanova, T. A.; Rusanov, V. D.; Fridman, A. A.; Abbasov, A. S.; Potapkin, B. V.; Shil'nikov, V. I. Izv. Akad. Nauk Az. SSR, Ser. Fiz.-Tekh. Mat. Nauk 1985,6(5), 88-96. 20) Chen, X.; Marquez, M.; Rozak, J.; Marun, C.; Luo, J.; Suib, S. L.; Hayashi, Y.; Matsumoto, H. J. Catal. 1998,178, 372-377. 21) Chen, X.; Suib, S. L.; Hayashi, Y.; Matsumoto, H. J. Catal. 2001,201, 198-205. 22) Luo, J.; Suib, S. L.; Hayashi, Y.; Matsumoto, H. J. Res. Chem. Intermed. 2000, 26, 849-874. 23) Luo, J.; Suib, S. L.; Hayashi, Y.; Matsumoto, H. J. Phys. Chem. A. 1999; 103, 6151-6161. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 148 24) Huang, A.; Xia, G.; Wang, J.; Suib, S. L.; Hayashi, Y.; Matsumoto, H J. Catal. 2000,189. 349-359. 25) Masteiton, W. L.; Hurley, C. N. Chemistry: Principles and Reactions 2nd Edition; Saunders College Publishing, Orlando, FL; 1993. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. B. METAL EFFECT AND FLOW RATE EFFECT IN THE HYDROGEN PRODUCTION FROM METHANE. 1. Introduction. Methane is the main component o f natural gas. Typical methane content is about 95%; it can range from 87 - 96%. Other main components are ethane, propane, butanes, carbon dioxide, and nitrogen. Natural gas is formed over millions of years (Figure III-17). Organic matter from plankton, plants and animals sank upon death to the bottom o f lakes and river beds where the organic matter was deposited into sand and clay sediments. These deposits were transported further by the action of currents and rivers themselves into larger bodies of water where they were finally deposited and formed a layer of organic silt producing the so-called source beds (1 in Figure III-17). Over time more and more sediment layers accumulated on top of this layer creating significant heat, sometimes up to 100°C or more (2 in Figure III-17). In this heat, complex chemical reactions occurred over millions of years; temperature between 60 and 120°C resulted in the formation of petroleum (3 in Figure III-17). If the petroleum was subjected to temperatures above 150°C, cracking occurred into lighter hydrocarbons such as methane, ethane, propane and butane yielding natural gas. Depending on the source of organic matter wet or dry natural gas was obtained. Organic matter derived from animal matter yielded wet natural gas high in methane and propane; organic matter derived from plants, which is the predominant process, yielded dry natural gas high in methane and minor amounts o f ethane, propane and butane. Large amounts o f natural gas found these days have not originated from a thermocatalytic origin but were produced by bacterial action. This process yielded methane concentrations of up to 99% in natural gas, which is found 149 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ISO at much higher levels in the earth’s crust compared to thennocatalytically formed natural gas.1 Organic matter dies and forms source bed (1) Petroleum forms under action of heat over millions of years Sediment layers 1 accumulate, formation starts (2) Figure 111-17. Formation o f petroleum and natural gas (Adapted from Ref. 2). Methane (C H 4) and natural gas, the least CO2 intensive of the fossil fuels, are used in a variety of different fields (Table III-3). Natural gas is used in residential applications for space and water heating and for cooking purposes. Natural gas can also be employed for air conditioning and production of electricity by natural gas fuel cells. Commercially, natural gas is also important for heating and cooling, and is utilized in the food industry for cooking and food preparation. Natural gas powered generators and back-up generators are becoming used more and more in commercial and office buildings. Industry alone uses 43% of all natural gas; in addition, natural gas is used for heating in the pulp and paper, metals, chemicals, petroleum refining, stone, clay and glass, plastic, and food processing industries. Useful applications include waste treatment and incineration, metal preheating (particularly for iron and steel), drying and dehumidification, glass melting, and fueling industrial boilers. Natural gas is the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. feedstock for the production o f hydrogen, methanol and related chemicals, and a source for the extraction o f butane, ethane, and propane which find applications as feedstock in the fertilizer and pharmaceutical industry. Natural gas is furthermore applied to a great extent for the generation o f electricity in steam generating units for turbines. Natural gas powered vehicles have become more popular due to environmental concerns.3 Table m -3. Usage o f natural gas in trillion cubic feet (Adapted from Ref 4). Usage 1990 2000 2010 2020 Industrial 5.5 8.0 9.3 10.4 1.4 3.8 7.0 11.4 Residential 4.4 4.8 5.6 6.1 Commercial 2.6 3.0 3.8 4.1 N/A N/A 0.2 0.2 Electricity Generation Natural Gas Vehicles N/A= Not Applicable An alternative source of fossil fuels and methane might be methane hydrates. Methane hydrates are methane and water trapped in cages o f cathrates, molecule-sized pockets of ice-crystals. These hydrates are stable at high pressures or at low temperatures and are found deep under ocean floors and in the Arctic permafrost regions. They were formed by organic decomposition by microbes on the ocean floor or in Arctic regions. Huge amounts are believed to be present around the world; conservative estimates account for over 300,000 trillion cubic feet, twice to ten times as much as fossil fuels. Upon burning, these hydrates release less CO2 and sulfur products than conventional fossil fuels. But the extraction and mining of these deposits cause great problems such as Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 152 working at the great depths they are found at, the expansion o f the compound upon extraction and the release o f the greenhouse gas methane.5 Methane is a very potent greenhouse gas and has a global wanning potential o f 21 times that o f carbon dioxide with an atmospheric concentration o f 1721 ppb. Major sources o f methane are releases from natural gas and petroleum production, from rice paddies, waste dumps, and livestock.6 Therefore ways to reduce CO2 emissions from natural gas are sought. The production of hydrogen from natural gas might be one alternative. In natural gas fuel cells, reformers (steam reformers, partial oxidation reformers, and auto-thermal reformers) produce internally the needed hydrogen for the electrochemical processes.7 Traditionally, hydrogen was and still is produced by steam reforming natural gas. This is done at temperatures between 500 and 800°C employing Ni/MgAfeOv This process is highly endothermic and requires considerable energy input for the vaporization of water and heating the reaction.8 Recently, this process has been used to steam reform biomass such as organic waste using commercial multicomponent catalysts, which generally contain Ni, K, Ca, and Mg on alumina-based supports.9 New partial oxidation catalyst for fuel cell applications were developed by Pino et al.10 using ceria supported platinum catalysts showing hydrogen selectivities o f over 90%. Bromberg et al.11 used thermal plasmas (several thousand K) to produce up to 50% hydrogen in plasma reformers at 2-3 kW power input. Mutaf-Yardimci et al.12 observed 16% conversion in the reforming of CO2 and CH4 in a pulsed corona discharge reactor after preheating the mixture thermally to 900°C. Liu et al.13 observed dielectric barrier discharges during diamond deposition using optical emission spectroscopy and Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 153 reported the excitation temperature and various radical species. Atmospheric pressure glow plasma enhanced CVD system for simultaneous production of hydrogen and carbon nanotubes (CNTs) by direct methane decomposition was used by Nozaki et al. 14 employing methane/hydrogen/helium gas mixtures. Our group performed various reactions involving reformation of CH4 and CO2. Brock et al. 15 used a Pt tubular and fan PACT reactor for this reaction, whereas Huang et al. 16 used a novel T-shaped tubular reactor with various electrode setups. This work studied the effect of using different metal electrodes on the hydrogen production from methane silent glow discharge plasmas. 2. Experimental Section. a. Preparation o f Gas Mixtures. Methane used in these experiments was obtained from commercial vendors such as Airgas. Its purity was of CP grade (99.5% pure). The content o f the feed gas is just a feed of pure methane at room temperature and atmospheric pressure. The flow rates used ranged over 5-250 mL/min for the plasma reactions. Ultra high purity air, also acquired from Airgas, was used to purge the reactors to remove some of the coke deposits that form typically in these reactions. Purges overnight were done in some cases. b. Tubular Reactor Setup. The reactions were carried out with a Plasma And Catalysis integrated Technologies (PACT) tubular reactor with exchangeable electrode. This reactor is depicted in Figure III-18. The reactor consisted o f an exchangeable electrode o f 10 cm Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 154 length and 8 mm outer diameter, an example was a Pd electrode shown in Figure III-19, a quartz tube o f 15 cm length and 10 mm inner and 12 mm outer diameter acting as a dielectric separating outer and inner electrodes, and aluminum foil of 10 cm was wrapped around this outer electrode. The exchangeable electrode consisted o f the electrode itself for which various materials such as gold and rhodium were used, which were electroplated onto a copper rod at a thickness of 100 pm, and a metal rod of 15 cm length and 2 mm outer diameter onto which this electrode was screwed. This metal rod was in a glass tube o f the same diameters as the quartz tube and o f 7 cm length and held in place by Swagelock fittings. The exposed end of this metal rod (outside of the Swagelock fitting) was used as the electric contact for the inner electrode to the high voltage source. grounded Inner electrode High voltage \ 7 C H ,in G as out Inner electrode- Blow up Figure III-18. Reactor setup for exchangeable electrodes in methane reactions. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 155 Pd Electrode Figure III-19. Exchangeable electrode, example o f a Pd electrode. c. Reaction and Circuit Setup. A setup similar to the one shown in Figure III-2 was used. This setup is also illustrated in Figure III-3 as the Pd electrode setup. The reaction gas methane was introduced at room temperature and atmospheric pressure into the tubular reactor and its flow was adjusted by a rotameter in-line with a set flow rate o f 20 mL/min in most experiments. In another study the flow rate was altered between 10 and 200 mL/min to investigate the flow rate dependence of the conversion. The system was equilibrated at the beginning of a run and then the power supply was turned to the desired level and continued until an equilibrium value was achieved. The circuit design in these experiments was identical with the one shown in Figure III-4. The high voltage was generated by an UHV-10 AC high voltage power Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 156 supply at frequencies between 8.0 and 8.1 kHz. Input voltage and input current values were measured by a DL-1540 Yokogawa oscilloscope using a high- (Tektronix P6025) and a low-voltage probe (Yokogawa 70996), respectively. The voltage across a 100 Q resistor in series with the reactor was used to determine the input current (a voltage is measured but the oscilloscope reading can be manipulated in a way that displays the current by taking the resistor into account). The resistor parallel to the reactor was the load resistor (lOOkft, 225W), preventing too much voltage from reaching the reactor. AC Voltage values around 2.600 kV were applied to the electrodes to produce glow discharge plasmas; these voltages were rms (root mean squared) voltages. d. Product Analysis and Characterization. The reaction mixture was continuously monitored with a MKS-UTI PPT quadrupole residual gas analyzer mass spectrometer (MS) with a Faraday cup detector and a variable high-pressure sampling manifold. Samples from the gas stream were taken into a sample loop and analyzed using a Hewlett Packard 5970 series mass selective detector coupled to a Hewlett Packard 5890 Gas Chromatography (GC/MS) equipped with a TCD detector with a HP-1 column (cross linked methyl siloxane) 12.5 m x 0.2 m x 0.33 pm film thickness. The electrodes used were provided by Dr. Hayashi of ASE or designed and assembled in our lab. Some electrodes were treated in our lab; copper and tin and tin oxide particles were deposited onto these electrodes by electroless plating. Deposits on the electrodes were investigated by using a Scintag XDS-2000 diffractometer with Cu Ka X-ray radiation and a Leybold vacuum system equipped with Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 157 a SPECS multichannel energy analyzer XPS spectrometer with Mg and Al anodes as the X-ray source. 3. Results. a. Methane Decomposition Using Different Electrodes. In these experiments, the effect of using different metal inner electrodes on the conversion of methane and the production of hydrogen was investigated. The conditions for this reaction were set at 20 mL/min and 2.56 kV input voltage (tins input voltage); pure methane was the reaction gas. All parameter were not changed during the reactions, the only variable was the material o f the inner electrode. The setup was purged with ultra high purity air and an air plasma was employed to “clean” the electrodes o f residues from cleaning and rinsing the electrode before each set of reactions. The reaction was equilibrated at the beginning and then the power was turned to 2.56 kV to start the decomposition of methane. The power was turned off when the reaction reached an equilibrium level, usually after 30-45 min. This procedure was repeated two times for a total of 3 runs under plasma conditions. After each set of reactions the setup was purged with air and an air plasma was applied, usually overnight. The used electrodes and the quartz tube were then cleaned using water and acetone. The reactions yielded very different results. In general, the destruction o f methane decreased during the 3 runs as did the production o f hydrogen. The difference between the metal electrodes was to which extent the methane decomposition and the hydrogen evolution decreased. Two examples are shown in Figures 111-20 and III-21; the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 158 decomposition o f methane and hydrogen production is shown in the mass spectra for nickel and for palladium, respectively. 1.2e-5 -|-------------------------------------------------------------------------------------------------- Methane y n ^ 8.0e-6- L V a S i g 6.0e-6 Power Poweron 5 6 f 4.0e-€ - * on Power on Power on Power Power on I V VV u Hydrogen 2 .0e-€ 0,U U i ,n, n , 0 100 200 300 Time in min Figure III-20. Mass spectrum o f reaction with Ni electrode showing methane destruction and hydrogen evolution. Nickel deactivated significantly as shown in Figure III-20 above. In the case of nickel, as seen in Tables III-4 and III-5, the methane decomposition dropped from 21.8% in the first run to 7.0% in the third run. The drop in hydrogen production was correlated well with the decrease in methane decomposition and showed a similar drop, from 10.6 vol% (run 1) to 5.6 vol% (run 3). A less drastic decrease was observed for the hydrogen production using a palladium inner electrode (Figure III-21). The hydrogen evolution dropped from 16.5 vol% to 15.2 vol%. As in the case of nickel, the decomposition of methane showed a similar trend; the decrease in decomposition was almost insignificant, from 21.2% for run 1 to 20.3% for run 3. The changes in the baseline for methane could Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 159 be due to slight changes in flow rate of methane and maybe by a baseline drift caused by the mass spectrometer itself. Methane 1.0e-5 ■ I c 8.0e-6 S Power on Power on Power on a. * £ 4.0e-€ a CL 0.0 -I 0 Hydrogen 50 100 150 Time in min Figure III-21. Mass spectrum o f reaction with Pd electrode showing methane destruction and hydrogen evolution. The tables below (Tables III-4 and III-5) summarize the behavior of the different metal inner electrodes used in these experiments. The choice o f the metal for the electrode influenced the conversion of methane and the evolution o f hydrogen. The electrodes seemed to deactivate during the 3 runs, but the degree o f deactivation was different for each individual metal. The decomposition was usually in a range from 2030%. Both nickel electrodes, one of them electroplated commercially and the other one coated with nickel nanoparticles by spraying a sol and then heating, showed almost the same decline in decomposition and evolution, as did the gold electrode, which appeared very corroded after the reactions. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 160 Table ID-4. Methane decomposition for various metal inner electrodes at 2.56 kV input voltage and 20 mL/min CH4. Methane Methane Methane Decomposition Decomposition Decomposition Run 1 (%) Run 2 (%) Run 3 (%) Rh 22.6 23.1 18.8 Ni (coated) 19.9 14.3 10.0 Ni 21.8 12.0 7.0 Pt 22.4 22.9 23.2 Au 20.5 14.6 6.5 Pd 20.5 19.1 22.0 Fe 23.4 22.1 21.2 Cu 18.2 17.4 17.1 Pd (new) 21.2 20.6 20.3 Sn (coated) 18.5 24.6 28.2 Cu (coated) 27.7 21.7 22.4 Metal For most other metals the conversion of methane dropped by 1 to 5% and the evolution o f hydrogen decreased by 1 to 2 vol% in most cases. The deactivation affected the decomposition o f methane more than the production of hydrogen. The exception was the electrode coated with tin and tin oxide by electroless plating. This electrode showed increasing ability to decompose methane in the 3 runs, but also exhibited a slight increase in the production of hydrogen. In the case o f the iron electrode, both parameters were almost constant, but the hydrogen production dropped significantly in the fourth run, not shown in Table III-5, down to 11.5 vol%. The electrode coated with copper nanoparticles by electroless plating showed a similar behavior as the solid copper electrode with respect to hydrogen production, but the coated electrode decomposed methane more Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 161 effectively than the solid electrode did. The two palladium electrodes showed very similar trends in their ability to destruct methane, whereas the “newer” electrode yielded a higher hydrogen production compared to the “older” electrode; these electrodes might be from different sources and might have a different core material onto which the palladium was coated, for the “newer” electrode the core material was copper. Palladium, platinum and the tin/tin oxide coated electrodes showed the best overall performance with a high rate of methane decomposition but at the same time still yielded significant amounts o f hydrogen. These three electrodes also exhibited the least deactivation o f all the tested inner electrodes. Reversing the polarity o f the electrodes, making the inner electrode the ground electrode, caused no change in decomposition or evolution in the case o f gold. Table III-5. Hydrogen production for various metal inner electrodes at 2.S6 kV input voltage and 20 mL/min C H 4. Hydrogen Evolution Hydrogen Evolution Hydrogen Evolution Run 1 (vol%) Run 2 (vol%) Run 3 (vol%) Rh 12.5 11.0 10.5 Ni (coated) 9.2 7.0 5.7 Ni 10.6 7.7 5.6 Pt 12.5 12.1 11.7 Au 11.2 8.8 6.7 Pd 12.5 12.1 11.3 Fe 13.3 14.5 14.3 Cu 14.5 14.0 13.9 Pd (new) 16.5 15.7 15.2 Sn (coated) 17.4 18.1 18.7 Cu (coated) 13.8 13.5 13.1 Metal Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 162 b. Cleaning the Electrodes and Methane Decomposition. Experiments were performed to investigate the influence of cleaning the electrodes (as described above) on the decomposition o f methane and the evolution of hydrogen. The results of these experiments are summarized in Tables III-6 and III-7. The fresh and the cleaned electrode showed methane decomposition of comparable value. The hydrogen evolution, however, was significantly lower when these two cases were compared; the hydrogen production dropped by about 3 vol%. When this electrode was used again, but this time without cleaning, a very significant drop in activity was observed. The decomposition rate dropped by 6-8% depending on the run number, and furthermore, the hydrogen production decreased accordingly. The drop in hydrogen evolution was less pronounced in the first run exhibiting a value closer to the one of the case when the electrode was cleaned. In the second and third run, however, significant differences were observed in degree o f hydrogen production; the values differed by about 3 vol%. Table III-6. Influence of cleaning the Pd inner electrode on methane decomposition at 2.S6 kV input voltage and 20 mL/min C H 4 . Methane Methane Methane Decomposition Decomposition Decomposition Run 1 (%) Run 2 (%) Run 3 (%) Pd, new 21.2 20.6 20.3 Pd, cleaned 22.4 20.4 20.2 Pd, not cleaned 14.5 15.9 14.0 Electrode Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 163 Table III-7. Influence o f cleaning the Pd inner electrode on hydrogen production at 2.S6 kV input voltage and 20 mL/min CH4 . Hydrogen Evolution Hydrogen Evolution Hydrogen Evolution Run 1 (vol%) Run 2 (vol%) Run 3 (vol%) Pd, new 16.5 15.7 15.2 Pd, cleaned 13.6 13.1 11.9 Pd, not cleaned 12.9 10.2 8.5 Electrode c. Effect o f Flow Rate on Methane Decomposition and Hydrogen Evolution. In this set o f experiments, the effect o f altering the input flow rate on the hydrogen production and methane destruction was studied. The flow rate was varied between values o f 10 and 140 mL/min; all other parameters such as the input voltage were left unchanged in these reactions. The values reported here are the average values of 3 runs, as before. In these runs, the decomposition rates and evolution rates were similar to the average rates. A similar trend was observed in the decomposition of methane and in the evolution o f hydrogen, as shown in Figure 111-22 and 111-23. The hydrogen production showed an exponential decline with a maximum at low flow rate (Figure IU22). The maximum hydrogen evolution achieved was 19.5 vol% at a flow rate o f 10 mL/min, and the hydrogen production seemed to level off at around 4 vol% at higher flow rates. The decomposition of methane exhibited the same exponential decline which was observed for the hydrogen production (Figure III-23). The decomposition rate was high (about 34%) at low flow rates (10 mL/min) and decreased to an almost constant value (6-7%) at high flow rates (over 100 mL/min). The highest decomposition rate in this study o f 33-34% was achieved at a low flow rate o f 10 mL/min. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 164 22 20 • 18 • £ 1 c 16 ■ J 12 ■ 14 - 10 - s Z N 8 ■ 6 ■ 4 2 60 80 160 100 Flow rate in mL/min Figure 111-22. Hydrogen production (vol%) using a feed methane at various flow rates and 2.56 kV. 60 80 100 120 140 160 Flow rate in mL/min Figure 111-23. Methane decomposition in % using a feed methane at various flow rates and 2.56 kV. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 165 d. Effect o f Input Voltage on Methane Decomposition and Hydrogen Evolution. The influence of the input voltage on the methane decomposition rate and the hydrogen production was investigated in this study. As before, the values represented here are average values from 3 consecutive runs. The hydrogen evolution and the methane decomposition showed different trends (Figure III-24). The hydrogen evolution showed a nearly linear trend with increasing input voltage. The hydrogen production increased from about 5 vol% at 1.75 kV input voltage to 14 vol% at 2.56 kV. The decomposition of methane, however, exhibited a very dissimilar behavior. The decomposition rate increases from a very low value o f 7% at an input voltage o f 1.75 kV to a plateau of about 18-20% at voltages over 2.0 kV. It seemed that the decomposition above this threshold is independent of an increase in applied input voltage 3? 20 2.0 22 2.4 2.6 2.8 Input voltage In kV Figure ID-24. Methane decomposition in % and hydrogen evolution in vol% using a feed of 20 mL/min methane at various input voltages. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 166 e. Mass Spectroscopic Studies. Investigation of the products o f the methane decomposition was the focus of this study. The products were analyzed by a mass spectrometer and by a GC/MS. The mass spectra with the power off is shown in Figure 111-25 and compared to the mass spectra o f the reaction products as shown in Figure 01-26. In the mass spectra when the power was off, taken after a reaction was run, showed that predominately methane with an m/z ratio o f 16 and associated peaks at m/z ratios o f 14 and IS were present. Other major peaks in this spectrum were assigned to hydrogen radicals and hydrogen at m/z ratios o f 1 and 2, respectively, and to alkane species at m/z ratios o f 27, 28 and 29 (major peak) and the corresponding higher peaks around m/z ratios of 40, 58, and higher m/z ratios in period distances. These peaks are also apparent in the spectrum when the power was off, but here their intensity is lower. These peaks appear in this spectrum because the spectrum was taken after a reaction and residues o f this reaction seemed to be present afterwards. 1.20607 1. 0064)7 s5 0.00608 6.00608 S 4.00608 I.H , CH. 2.00608 0.006*00 1 6 11 16 21 26 31 A .-A 36 41 46 51 96 61 16 71 76 81 H 91 86 101 106 111 116 121 126 Figure III-25. Mass spectrum of methane decomposition reaction with power off. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 167 3.00E-07 I e t i 1.00E-07 He Pe CH. I.U. X L .U * L^ ...U 1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96 -___ 101 106 111 116 121 126 mtz Figure 111-26. Mass spectrum of methane decomposition reaction with power on. The peaks which were weak in the afore-mentioned spectrum are much more intense in the spectrum under plasma conditions (Figure 111-26). The hydrogen peak at an m/z of 2 is more intense than in the other spectrum whereas the methane peak is decreased according to its decomposition. The peaks centered around an m/z of 28 could be attributed to alkane species such as ethane (E), propane (Pr), butane (B), pentane (Pe), hexane (Hx), and heptane (Hp) and their isomers, but also can be part of systems from cycloalkanes and alkenes. The group of peaks with m/z ratios of 37, 39 and 41 was assigned to the same species as above with the exception o f ethane. Peaks at higher m/z ratios are due to higher alkanes and their isomer and the corresponding cycloalkanes and alkenes. The peaks around an m/z of 57 could be caused by the alkanes butane, pentane, hexane, heptane, octane and higher alkanes, as well as isomers of these such as 2,2- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 168 dimethyl pentane, 2-methyl hexane and 3-methyl pentane. Alkenes such as 2-methyl-1pentene have to be accounted for as well when assigning these peaks. In the resulting spectra from the GC/MS analysis of samples taken under plasma conditions, a multitude o f compounds were found as listed in Table III-8. The main products of the reaction, ethane and lower alkanes were not listed in this table as well as hydrogen and methane because it was difficult to distinguish them from the higher hydrocarbons in the spectra, but they are clearly present as shown by the mass spectrometric analysis. The species listed here were the major species observed in the spectra obtained using the GC/MS and gave the best quality agreement when assigning them with the GC/MS software. This list is not complete because numerous minor peaks are present and due to the multitude of species present, it was very difficult to identify * these species. The major species are branched alkanes and alkenes as well straight alkenes and substituted cycloalkanes (Table III-8). Table III-8. Compounds found in the GC/MS analyses of samples taken under plasma conditions. 2-methyl propane 2-methyl-1-propene methyl cyclopentane 2-methyl-1-pentene 3-methyl-Z-2-pentene 2,3,4-trimethyl-E-hexane 2,2,3-trimethyl hexane 2,3-dimethyl pentane 3,3,4-trimethyl hexane 2,3,5-trimethyl hexane 1-propene 2-butene-E 2-pentene 3-methyl pentane 2-methyl pentane cis- 1,2-dimethyl trans-1,2-dimethyl cyclopropane cyclopropane 2-methyl-2-butene Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 169 / Characterization o f Electrodes and Electrode Deposits. X-Ray diffraction patterns of the copper and tin/tin oxide coated electrodes were collected and are shown in Figures 111-27 and ID-28, respectively. In the pattern for the copper coated electrode no copper oxides were observed. The pattern matches those o f copper metal and carbon having a diamond structure, as shown in Figure III-27. In the case o f the tin/tin oxide coated electrode, the same carbon species can be identified again (Figure 111-28). However, the assignment of the tin and tin oxide species is a bit more complicated. The only patterns that match the shown pattern are that of tin metal and that of P-tin oxide (SnO). Compared to the patterns from the database, several strong peaks are missing which could be due to preferential orientation o f the spherical electrode in the diffractometer. 3500 Cu C(diamond) 3000 - 250 0 - >. c 3c 2000 5 1500 JS • - Cu a: 1000 Cu C(diamond) - Q 500 - 20 60 40 80 20 Figure III-27. X-Ray diffraction pattern o f copper coated electrode. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 170 Qdiamond) C(diamond) 20 40 60 80 20 Figure 111-28. X-Ray diffraction pattern of tin/tin oxide coated electrode. The analysis o f the carbon deposits washed off the electrode surface with toluene yielded no useful information because only toluene was detected using the GC/MS. XPS analysis of a palladium electrode coated with carbon deposits yielded only the identity o f the palladium metal by observation of the Pd 3d, 3ds/2 and 3d3£ peaks; before the analysis it was known that the electrode is either palladium or platinum coated. Furthermore, the carbon C Is peak was identified, but no statement could be made about the nature o f the carbon species, whether the carbon species was aromatic or aliphatic carbon deposited on the surface of the palladium electrode. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 171 4. Discussion. a. Metal Effect. Luo et al.17 described the metal effect in the water splitting reaction. Oxygen radicals passivated the surface o f the metal and depending on the reactivity o f the metal with oxygen, surface oxides were formed. Stable metals showed therefore higher activities for water splitting. In the carbon dioxide reforming of methane, a difference in deactivation was observed. Nickel catalysts were effective in these reactions but rapid deactivation from coking limited their efficiency.18,19 Noble metal catalysts such as platinum catalysts showed a reduction in coking and therefore a prolonged period of activity.20'23 Similar trends were observed in the study of methane decomposition. The noble and unreactive metals such as platinum, rhodium and palladium showed high decomposition and hydrogen evolution activities in these reactions and deactivation was relatively low. The decomposition ability of platinum is slightly higher than that o f palladium, but palladium showed higher conversion to hydrogen than platinum. The reason might lie in the different effect o f coking on the hydrogen evolution and in a higher catalytic activity of palladium. Ordonez et al.24 observed more pronounced deactivation of a platinum catalyst compared to a palladium catalyst due to coking. Gold is also a member of the noble metal class but showed significant deactivation in the decomposition reaction which resulted in low hydrogen evolution. Initial activity o f gold was almost as high as that of palladium and platinum however its activity drops off sharply in the next 2 runs. The major deactivation was likely caused by a combination of sintering of the electrode as evident in the corroded appearance of the electrode after the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 172 reaction and coking. Avgouropoulos et al.25 observed a significant deactivation of gold catalysts compared to copper and platinum catalysts in the selective oxidation of carbon monoxide. Other authors attributed this loss in activity to sintering and coking.26,27 Copper as a less reactive metal also yielded high decomposition rates as well as high hydrogen evolution efficiency. The higher decomposition rate for the copper coated electrode might be due to the higher surface area of the copper nanoparticles on the electrode. The deactivation of the nickel electrode seemed to be a combination of coking and the increased reactivity o f the electrode. Similar behavior would be expected for the reactive iron electrode but the reactivity of the iron electrode with the reaction products seemed suppressed in the methane reaction. The high and increasing activity o f the tin/tin oxide electrode might be due to the presence of oxygen in tin oxide. b. Methane Decomposition. Thermodynamically, the reaction to coke and hydrogen seemed to be preferred in the decomposition of methane. Mass spectrometric data suggested that the formation of other compounds such as alkanes, alkanes, and cycloalkanes occurred as well. This conclusion was based on the appearance of peaks other than those o f methane and hydrogen in these mass spectra. The presence of other compounds was confirmed by analyzing samples using the GC/MS. Species found included ethane and ethylene, part of the major system around the m/z ratio o f 28, and analogous hydrocarbons in system with 12 to 13 mass units spacings in between them as is typical for methyl and methylene groups. The system around an m/z o f 41 for example contained peaks for propane and Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 173 \ higher alkanes which showed this m/z value in their individual mass spectra due to defragmentation. A mechanism for the decomposition of methane can be suggested (equations 117). The mechanism is initiated by electron activation o f methane and shows the formation of coke, hydrogen and alkanes, alkenes and cycloalkanes. The formation of aromatic compounds seemed probable. The form o f coke observed on the electrodes might be diamond like as seen in the X-ray diffraction pattern shown before. This mechanism is based on observations in the mass spectra and the results for the samples analyzed with the GC/MS. The collisions for the formation o f higher hydrocarbons such as propane and butane as shown in equations 12 and higher are simplified and might be the result of sequential collisions. CH4 + e - » C H 4 * / C H 4 * + ( l ) CH4 */ CH4*+ -» CH3-/ CH3-++ H- (2) CH3-/ CH3-++ CH4 */ CH4 *+-» CH4 + CH3«*/ CH3-*+ (3) CH3-*/ CH3-*+-» CH2-/ CH2-++ H* (4) CH2-/ CH2-++ CH4 */ CH4 *+“> CH4 + CH2**/ CH2-*+ (5) CH2-*/ CH2«*+-> CH•/ CH»++ H* (6) CH«/ CH«+ + CH4*/ CH4*+“> CH4 + CH**/ CH**+ (7) CH-*/ CH-*+"> C•/ C-++ H- (8) 2H- -> H2 (9) 2 CH3- -> CH3CH3 (10) 2 CH22 CH3- + CH2- CH2CH2 (11) CH3CH2CH3 (12) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 174 CH3* + CH2* + CH* -» CH2CHCH3 (13) 2 CH3* + 2 CH2* CH3CH2CH2CH3 (14) 3 CH2* -> CH2CH2CH2 (cyclic) (15) 3 CH3* + CH* CH3CHCH3CH3 (16) 3 CH3* + CH* + CH2*-> CH3CHCH3CH2CH3 (17) c. Effect o f Flow Rate, Input Voltage and Electrode Cleaning. The plots of hydrogen production versus flow rate and the methane decomposition versus flow rate showed a drop almost in a negative exponential trend similar to the trends seen in the water splitting reactions involving methane. The same reasoning can be used to explain this phenomenon. The residence time decreased as the flow rate increased and therefore the time the reactant was in contact with the plasma and the electrode was lower and there was not enough time for a multitude o f reactions to take place; the average kinetics of the reaction was slower than the contact time. The effect of the cleaning of the electrode showed the influence o f coking on the decomposition of methane and the evolution of hydrogen. The coke produced in the reactions deposited on the electrodes and decreased the conversions. This furthermore showed that the metal acted catalytically in these reactions because when the electrodes are covered with coke, deactivation occurred and the decomposition rate and the hydrogen evolution dropped. The influence o f the input voltage is best described in terms o f energetic electrons. The higher the input voltage was, the higher was the energy input into the plasma. This caused more and more electrons to have the appropriate energy to excite Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 175 methane and induced the ionization and breakup o f methane as proposed in the reaction scheme. This influence was clearly seen in the dependence o f the hydrogen evolution on the applied input voltage, which followed a linear trend in this case. This linear trend can be understood in terms of the electrons inducing more efficient breakup of methane and the increased production o f hydrogen radicals which then formed hydrogen molecules. Methane decomposition, however, showed a different behavior; the increase in input voltage induced higher decomposition rates up to a value o f about 2 kV. Above this value the decomposition rates seemed independent of the applied input voltage. This indicated that a certain energy was necessary to induce efficient breakup o f methane and above this energy an increase in input voltage and therefore electron energy did not influence the methane breakup efficiency anymore. 5. Conclusions. The effect of using different metal electrodes on the methane decomposition and hydrogen evolution was investigated in this research. Noble metals such as palladium and platinum showed the highest decomposition rates and highest hydrogen evolution as well as the lowest deactivation. Significant deactivation was observed for nickel and gold electrodes due to reactivity with reaction products, coking and sintering. Electrodes coated with copper nanoparticles and tin/tin oxide nanoparticles showed high activity. In the case o f the copper coated electrode, the activity was comparable to the activity o f the electroplated copper electrode, but the decomposition rate of methane was slightly higher probably due to the characteristics of the coating. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 176 Several other factors influencing the methane decomposition were also studied. The flow rate has a significant impact on both the methane decomposition and the hydrogen evolution. With increasing flow rate the decomposition rate and the hydrogen production showed a negative exponential decline due to the decreased residence time in the plasma zone. Cleaning the metal electrode had a profound impact on the methane decomposition due to coking and indicated thereby the catalytic role o f the electrode in these reactions. Increasing applied input voltage increased the hydrogen production and the methane decomposition up to the threshold by supplying more energetic electrons. Other major products apart from hydrogen are lower and higher alkanes, alkenes, and cycloalkanes. Branched alkanes were also detected. A reaction scheme accounting for the reaction products was proposed based on the data gathered by using mass spectrometry and analysis of samples using the GC/MS. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 177 6. References. 1) Petroleum and natural gas A-Z, Wintershall AG, http://www.wintershall.de/www/wintershall.nsf/name/a- 2 e-EN: January 2003. 2) Energy Story, Chapter 8: Fossil Fuels, California Energy Commission, http://www.energyquest.ca.gov/storv/chapter08.html. January 2003. 3) Uses of Natural Gas, NaturalGas.org, http://www.naturalgas.org/overview/uses.asp. January 2003. 4) U.S. Natural Gas Markets: Recent Trends and Prospects fo r the Future, Energy Information Administration, U.S. Department o f Energy, Washington, DC, May 2001. 5) Wolfson, W. Frozen Gas under the Sea, Chemweb.com: The Alchemist: News, http://www.chemweb.com/alchem/articles/103S208837900.html. November 2002. 6) United Nations Framework Convention on Climate Change, Graphics: Vital Climate Climate Change, http://www.grida.no/climate/vital/index.htm. January 2003. 7) Fuel Cells 2000's Frequently Ashed Questions, Fuel Cells 2000, http://www.faelcells.org/fcfaas.htm. January 2003. 8) Rostrup-Nielsen, J. R. In Catalytic Steam Reforming Catalysis, Science & Technology; Anderson, J. R.; Boudart, M., Eds.; Springer: Berlin, 1984; vol. S. 9) Wang, D.; Czeraik, S.; Chomet, E. Energy and Fuels, 1998,12, 19-24. 10) Pino, L.; Recupero, V.; Beninati, S.; Shukla, A. K.; Hegde, M. S.; Parthasarathi, B. Appl. Catal, A 2002,225, 63-75. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 178 11) Bromberg, L.; Cohn, D. R.; Rabinovich, A.; Alexeev, N. Proceedings o f the 1998 U.S. DOE Hydrogen Program Review, Baltimore, MD, United States 1998,627-639. 12)Mutaf-Yardimci, O.; Saveliev, A. V.; Fridman A. A.; Kennedy, L. A. Int. J. Hydrogen Energy 1998,23, 1109-1 111. 13) Liu, D.; Xu, Y.; Yang, X.; Yu, S.; Sun, Q.; Zhu, A.; Ma, T. Diamond Relat. Mater. 2002,11, 1491-1495. 14)Nozaki, T.; Kimura, Y.; Okazaki, K. 76/* ESCAMPIG and 5th ICRP Joint Conference, 14-18 July 2002, Grenoble, France, 37-38. 15) Brock, S. L.; Shimojo, T.; Suib, S. L.; Hayashi, Y.; Matsumoto, H. J. Res. Chem. Intermed. 2002,28, 13-24. 16) Huang, A.; Xia, G.; Wang, J.; Suib, S. L.; Hayashi, Y.; Matsumoto, H. J. Catal. 2000,189, 349-359. 17) Luo, J.; Suib, S. L.; Hayashi, Y.; Matsumoto, H. J. Res. Chem. Intermed. 2000, 26, 849-874. 18)Lu, G. Q.; Wand, S. CHEMTECH (January), 1999,37-43. 19)Ruckenstein, E.; Hu, Y. H. J. Catal. 1996,162, 230-238. 20) Ashcroft, A. T.; Cheetham, A. K.; Green, M. L. H.; Vernon, P. D. F. Nature 1991, 352, 225-226. 21)Tsipouriari, V. A.; Efstathiou, A. M.; Verykios, X. E. J. Catal. 1996,161, 31-42. 22) Bradford, M. C. J.; Vannice, M. A. J. Catal. 1998,173, 157-171. 23) Stagg, S. M.; Romedo, E.; Padro, C.; Resasco, D. E. J. Catal. 1998,178, 137-145. 24) Ordonez, S.; Diez, F. V.; Sastre, H. Appl. Catal., B 2001,31, 113-122. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 179 25)Avgouropoulos, G.; Ioannides, T.; Papadopoulou, C.; Natista, J.; Hocevar, S.; Matralis, H. K. Catal. Today 2002, 72. 157-167. 26) Wang, G.; Zhang, W.; Lian, H.; Liu, Q.; Jiang, D.; Wu, T. React. Kinet. Catal. Lett. 2002, 75. 343-351. 27) Kang, Y.-M.; Wan, B.-Z. Catal. Today 1997,35. 379-392. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. APPENDIX I: FUTURE WORK 1. Freon Decomposition. In future experiments it would be interesting to investigate the metal effect in the Freon decomposition reaction as observed in the methane decomposition. The decomposition reaction was studied just using a stainless steel electrode. Using a different metal electrode might optimize the power input and lead to better energy efficiencies in the decomposition reaction. The energy efficiency might also be increased by using different, more effective power supplies and variable frequency power supplies such as the TREK power supply used in carbon tetrafluoride reactions. The power supply used in the Freon reactions showed significant power losses in the voltage generation and some losses in the resistor setup. Freons were used as refrigerants, and there are still significant quantities stockpiled around the world which have to be destroyed. Decomposition reactions at higher Freon concentrations up to the use of pure Freon seem therefore feasible. The use o f higher Freon concentrations necessitates the study o f more effective hydrogen fluoride capture mechanisms than a simple water trap such as mineralization. If refrigerants leak out, low concentrations might be escaping into the air. Studies of Freon decomposition at low levels in the ppm range could be beneficial for treating factory exhausts containing low levels of Freons. The analysis o f the reaction products could be complemented by using a cold trap. In the cold trap the reaction products could be captured as solids. The vessel with the captured products could be slowly heated and the products separated by boiling point. 180 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 181 Alternatively, the effluent gas during heating could be analyzed using a mass spectrometer or a mass spectrometer coupled with a gas chromatograph yielding detailed information on the separated reaction products. 2. Decomposition of Carbon Tetrafluoride. In this decomposition reaction, carbon tetrafluoride was destroyed with the help o f water molecules in the presence of activated carbon. Zeolite SA was used in a mixture with activated carbon but showed lower decomposition rates than activated carbon alone. Zeolites can hold a lot o f water molecules in their cage structure. Therefore, different classes o f zeolites might be tested by themselves and compared to the zeolite mixture already tested. A systematic study of different zeolites in this decomposition reaction might yield higher decomposition rates and less deactivation of the used catalyst depending on their ability to bind and easily release water molecules. Further studies into the deactivation mechanism of the activated carbon used in these reactions would be helpful. A combination of fluorination and coking of the activated carbon surface was suggested to cause the observed deactivation. Investigation of the spent activated carbon could involve several different methods. Thermogravimetric analysis o f the activated carbon could yield information on the coking o f the catalyst if used in a carbon monoxide atmosphere. Electron microscopy could yield information of the fluorination and coking when used in conjunction with elemental mapping. Transmission electron microscopy used together with an energy dispersive X-ray analyzer could yield this valuable data. Scanning electron microscopy might also yield insight into the deactivation mechanism. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 182 3. Hydrogen Production from Water and Methane. The porous copper electrode yielded the highest conversion in the hydrogen production from water and methane. However, the mechanical stability is undesirably low; therefore the stability has to be improved. Just one layer o f copper mesh is wrapped onto a copper tube; this causes the brittle nature o f the electrode. Measures to make these electrodes sturdier should be undertaken. Further investigation o f the carbonaceous deposits on the electrodes would be helpful in order to confirm the observation o f diamond-like carbon in the X-ray diffraction patterns. Transmission electron microscopy could yield important information for the confirmation of this observation. Ways to reduce the formation o f the carbonaceous deposits on the electrodes might be beneficial in prolonging the lifetime o f the electrodes and decreasing the deactivation o f the electrode. In this context the controlled admixture of oxygen might reduce coking by the formation o f carbon oxides and the dependence of coking on the oxygen concentration could be studied, but oxygen probably also reduces the production of hydrogen by forming water with hydrogen radicals. 4. Hydrogen Production from Methane. The metal effect was studied in the decomposition o f methane. Two coated electrodes with copper and tin/tin oxide nanoparticles were investigated. Further studies in this context could include the use o f electrodes coated by nanoparticles o f various metals. Enhancement of catalytic activity could be achieved by depositing various catalysts onto these electrodes. Octahedral molecular sieves (OMS) materials have been Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 183 shown to stick on glass and copper metal by dip coating in a gel precursor solution and annealing at elevated temperatures. Therefore, electrodes could be coated with active manganese oxides materials such as OMS-2 and tested and evaluated in the decomposition reaction o f methane. Other possible materials are titanium dioxide and maybe even zeolites if they stick to metal. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. APPENDIX II: OTHER RESEARCH PROJECTS 1. Decomposition of 1-Butene using Titanium Dioxide Photocatalysis. Several other research projects were also pursued in addition to the work presented in this dissertation. These include the decomposition of 1-butene using the photocatalyst titanium dioxide, the development o f alloy mixtures for novel lithium-ion batteries, and the synthesis of modified MCM-41 materials as catalysts in the production o f o-diphenylamine. Photocatalysis was used to decompose low concentration o f 1-butene in the first work. Photoassisted catalysis involving Ti02 has been carried out as a means of removing harmful organic pollutants from water and ambient air, even at low levels. A thin film reactor was used in this study at high flow rates to test mixed films of titania with zirconia, tantalum oxide, niobium oxide, KTaC)3, mordenite, and zinc oxide, respectively. The probe gas was 1-butene due to its relatively high activity, and since 1-butene is a model compound for volatile organic compounds (VOCs). The thin film reactor was a flat glass-plate photocatalytic reactor in which reactions were usually run at high flow rates of about 3.6 L/min in order to eliminate the influence of mass transfer from the bulk and at ambient temperature and pressure. The composition of the flow gas was that of an artificial atmosphere with a level of 1-butene between 1-10 ppm as a pollutant, and a variable water vapor content was added by a water bubbler. The thin films were prepared via dip coating using suspensions of Degussa P-25 titania with the appropriate amount of dopant such as zirconia or mordenite. Analysis was done using a gas chromatograph and UV lamps were the light source in these reactions. 184 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 185 The data for six different systems including one for the established photocatalyst P-25 were collected. Only two of the tested systems exhibited higher activity for 1butene decomposition than the model catalyst Degussa P-25 in the humidity regime tested. The zirconia/titania mixtures o f 5.3 % and 0.56 % showed the highest improvement compared to pure Degussa P 25 in the humidity range from 0 to 4500 ppm relative humidity. The low concentration and the high concentration films showed almost identical conversions in the range tested. The initial conversion was up to 5 % higher than that for pure titania P 25 between 1000 and 2000 ppm relative humidity, but the difference in conversion between these systems and pure titania decreased with increasing humidity and the trend reversed at about 3000 ppm relative humidity. An increase of initial conversion of about 20 % was observed in the thin film system employing high flow rates. XRD data showed the presence of zirconia in these thin films and changes in the d-spacing of the titania peaks were observed. The mordenite mixed system did not show enhancement in conversion contrary to earlier results. It was observed that the initial conversion is reduced by 2-4 % compared to P-25 in our dynamic high flow rate system. The conversion can be improved slightly by heating the films to 90°C; it is increased 1-2 %. High humidity in the ambient air might have played a role in decreasing the conversion in the later studies. Other mixed oxide films have also been tested. These metal oxide mixed films showed activities similar to the ones using titania P-25 or showed slightly lower activities (about 2-3 %). The addition of metal oxides had no detrimental effect on the activity o f titania, but an enhancement such as in the case of zirconia had not been observed. Zinc oxide was the only system in which a significant drop in activity was observed. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 186 In a different study, the effect o f 1-butene concentration on conversion was investigated. The conversion decreased in an exponential fashion as the concentration of 1-butene was increased. This phenomenon could be explained by saturation of the available active sites on the catalyst surface because the lower the concentration is, the more sites are available for the pollutant. If the concentration increased, the availability of sites decreased and competition took place. A similar explanation can be used to explain the humidity trends. In addition to fewer available sites with increasing water on the surface (increasing humidity) the more competition between the pollutant and water took place. Pure zinc oxide and zirconia films were tested under the same conditions as the mixed systems. Both systems showed no photocatalytic activity with UV excitation and in contrast to results with the titania mixed systems; both pure films exhibited a blue-green luminescent glow upon irradiation with UV light due to impurities in the zirconia and zinc oxide. 2. Alloy Preparation. In this research, alloys were prepared which were intended to be used as anode materials in secondary batteries. Systems were chosen based on cost effectiveness and environmental friendliness. Three alloy systems were initially investigated. These alloys included silver with addition of either silicon, zinc or germanium. These samples were heated in a muffle furnace to temperatures up to 600°C whereas earlier samples were heated in a tube furnace to temperatures as high as 1000°C. After the heating overnight, these samples were then analyzed by X-ray diffraction. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 187 Different mixing ratios of the alloys were investigated. The silver/germanium system and the silver/silicon system did not show any alloy formation but the silver/silicon mixture exhibited some enhancement charging/discharging behavior when incorporated into a test cell. However, the silver/zinc system showed alloy formation with a multitude o f phases present. Therefore the focus shifted to the silver/zinc system. The main product o f these reactions was however zinc oxide even if argon was used as a purge gas due to the porous nature of the muffle furnace. Pressing the samples into pellets and using polyvinyl alcohol as a binder yielded coking o f the pellet In a later stage, these pellets were heated under nitrogen or hydrogen in a CVD reactor yielding distinctive alloy phases with one phase being dominant. Microwave heating synthesis o f silver/zinc alloys from silver and zinc salt precursors was unsuccessful. 3. MCM-41 Synthesis. New catalysts were sought for the synthesis o f o-diphenylamine from aniline and nitrobenzene. Siliceous and aluminosilicate MCM-41 materials of high surface area and channel aperture around 40 A were synthesized as determined from the XRD data. The siliceous species showed higher selectivity to the para product than the ortho product (Ratio para: ortho was 2:1). With changes in the synthesis procedure tubular and spherical MCM-41 materials were synthesized. The acid exchange of a siliceous MCM41 material caused the collapse of the structure as observed in the XRD data. A sulfated zirconia/MCM-41 material was prepared by the incipient wetness technique. The selectivity, however, did not change with the conversion still being low. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.