Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Studies of siloxane decomposition in biomethane combustion
(USC Thesis Other)
Studies of siloxane decomposition in biomethane combustion
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
STUDIES OF SILOXANE DECOMPOSITION IN BIOMETHANE COMBUSTION by Mir Aydin Jalali A Dissertation Presented to the FACULTY OF THE USC GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (MECHANICAL ENGINEERING) December 2013 Copyright 2013 Mir Aydin Jalali ii Dedication I would like to dedicate my dissertation work to my loving parents, baba Tajeddin and maman Narcis, for their unending love, encouragement, and support. iii Acknowledgements First and foremost, I would like to thank Professor Fokion N. Egolfopoulos and Professor Theodore T. Tsotsis for giving me the opportunity to conduct research in their prestigious and stimulating research group. I highly appreciated the great amount of freedom I was granted during my work and the dedicated support for my projects. This dissertation is a result of their valuable help, great support, and significant patience. I also wish to thank Professor Paul Ronney, Professor Satwindar Sadhal, and Professor Charles Campbell for serving on my qualifying and dissertation committees. This thesis could not be completed without the generous help of Dr. Mirmohammadyousef Motamedhashemi, who assisted me in every way possible of my research like an older brother. Those long brain-storming discussions deeply helped me to discover new chapters of science as well as life rules. Also, the great help of Dr. Qiyao Feng for instructing me to use our simulation tools is gratefully appreciated. I would also like to thank Dr. Adam Fincham for his magnificent help to build the Laser Extinction setup and also for all scientific and technical assistance and enlightening discussions. His thoughtful comments helped me tremendously. My special thanks go to Dr. Khiza Mazwi for kindly training me and helping me to take AFM images. My wonderful friends in and outside the lab helped me get through the rough times. Okjoo Park, Hugo Burbano, Jagan Jayachandran, Roe Burrell, Runhua Zhao, Vyaas Gururajan, Sahar Soltani, Hamed Barghi, Abtin Ansari thank you so much for all informative suggestions and support. You made my working atmosphere more vibrant, fun and joyful. iv This research was made possible by financial support from the Southern California Gas Company. I would also like to thank my younger brother, Arsalan. He brought laughter into my monotonous PhD life. He has always helped me through hard times and I feel very lucky for having him as my brother. Last, but for certain, not least, I would like to express my deepest appreciation to my dear parents, baba Tajeddin and maman Narcis. They have been my role models, my best friends, and my greatest motivators. I thank them for their endless love and constant support; no matter where they are, they are close to me in my heart. I love you! v Table of Contents Dedication……………………………………….………………………………………..ii Acknowledgements………………………………………………………………….…..iii List of Tables……………………………………………………………………...…….vii List of Figures……………………………………………………………………….....viii Abstract……………………………………………………………………………...…xiii Chapter 1 – Introduction……………………………………………..…………………1 1.1. Overview and Significance……………………………………………………….1 1.2. Project Objectives……………………………………………………..………….4 1.3. Organization of Dissertation……………………………………………………...6 Chapter 2 – Fate of Siloxane Impurities during the Combustion of Renewable Natural Gas…………………………………………………………………………….....8 2.1. Introduction…………………………………………………………………….....8 2.1.1. Biogas History and Background……………………………………………8 2.1.2. What are the Siloxanes?.................................................................................9 2.1.3. Siloxanes Removal Techniques…………………………………………...14 2.1.4. Solid Particles Formation in Flame………………………………………..16 2.2. Experimental Set-up………………………………………………………..……23 2.3. Temperature Measurements………………………………………………..……30 2.3.1. Flat Flame Temperature Profile…………………………………………...30 2.3.2. Bunsen Flame Temperature Profile………………………….……………33 2.4. Reaction Kinetics Studies on Individual Siloxanes………………………..……34 2.4.1. Concentration Profiles…………………………………………………….34 2.4.2. Conversion Profiles…………………………………………………….….37 2.4.3. Rate Constant Parameter Fitting…………………………………………..40 2.5. Solid Particle Volume Fraction Measurements Using Laser Extinction Technique...........................................................................................................................42 2.5.1. Theory and Background…………………………………………...………42 2.5.2. Solid Particle Volume Fraction Measurements for Single Compounds…..43 2.5.2.1. Solid Particle Volume Fraction Profile……………………………...43 2.5.2.2. Maximum Solid Particle Volume Fraction…………………...……..44 2.6. Surface Analyses Results………………………………………………………..45 2.6.1. Introduction……………………………………………………….……….45 2.6.2. Sample Preparation………………………………………………………..46 vi 2.6.3. Results and Discussion……………………………………………...…….47 2.7. Decomposition of Siloxanes’ Mixtures…………………………………………51 2.7.1. Introduction……………………………………………………..…………51 2.7.2. Concentration Profiles…………………………………………………….52 2.7.3. Solid Particle Volume Fraction Measurements……………………….…..58 2.7.4. SEM/EDAX Analyses Results: Particle Deposition Using a Simulated RNG Containing 16ppm v of D 4 +L 4 ……………………………………………….……60 2.7.5. Conclusions………………………………………………………..………63 2.8. Solid State Studies on Siloxanes’ Decomposition Products……………...……..63 2.9. Concluding Remarks…………………………………………………….………73 Chapter 3 – A Study on Silane Decomposition during the Combustion of Renewable Natural Gas………………………………………………………………………….…..75 3.1. Introduction………………………………………………………………..…….75 3.2. Experimental Approach……………………………………………………..…..81 3.3. Modeling Approach……………………………………………………………..87 3.4. Results and Discussion………………………………………………………….88 3.4.1. Flame Temperature Profile………………………………………..………88 3.4.2. Experimental Silane Concentration Profiles and Simulations…………….89 3.4.3. Comparison of the Two Silane Decomposition Models…………………..90 3.4.4. Solid Particle Volume Fraction Measurements…………………………...99 3.4.5. Surface and Particle Analysis Results………………………………...….102 3.5. Concluding Remarks…………..…………………………………………….…106 Chapter 4 – Conclusions and Future Work………………………….......………….108 4.1. Concluding Remarks…………………………………………...………………108 4.2. Future Work………………………………………………………....…………111 References………………………………………………………………….......………113 vii List of Tables Table 2.1. Siloxane Compunds in LFG and some of their properties…………….…12 Table 2.2. Previous studies on the combustion synthesis of silica particles from siloxanes………………………………………………………….………20 Table 2.3. Pre-exponential factors (A) and activation energies (E) for different siloxanes (Jalali et al., 2013)……………………………………….…….41 Table 2.4. Chemical composition (wt.%) of the Ni/Cr alloy………………………..46 Table 2.5. Elemental analysis of Ni/Cr surface for various exposure times to the siloxane/RNG flame…………………………………………….......……49 Table 2.6. Thermochemical analysis of reactions in the silicon-oxygen system……65 Table 3.1. Chemical composition (wt.%) of the Ni/Cr alloy………………….…….87 viii List of Figures Figure 2.1. Siloxane group nomenclatures……………………………………….…..10 Figure 2.2. Examples of Siloxanes’ chemical structure. a) Linear structure of Octamethyltrisiloxane (MDM or L 3 ), b) Cyclical structure of Octamethylcyclotetrasiloxane (D 4 )………………………...………….....10 Figure 2.3. Schematic of the overall experimental configuration…………………....24 Figure 2.4. Schematic of the counter-flow configuration (Adapted from Feng, 2011)………………………………………………………………..........26 Figure 2.5. Renewable Natural Gas flame. a) Opposed-jet premixed flat flame, b) Bunsen flame……………………………………………………….……27 Figure 2.6. A device for temperature measurements…………………………………28 Figure 2.7. Arrangements of mirrors and position of photomultiplier with respect to the burners’ assembly………………………………………........………30 Figure 2.8. Experimental vs. computed flame temperatures in the opposed-jet flame......................................................................................................... 32 Figure 2.9. Experimental temperature profile on the vertical axis of a Bunsen flame..........................................................................................................33 Figure 2.10. Experimental and computed L 2 concentration profiles for various L 2 feed concentrations……………………………………………………...…….34 Figure 2.11. Experimental and computed L 3 concentration profiles for various L 3 feed concentrations……………………………………………………………35 Figure 2.12. Experimental and computed L 4 concentration profiles for various L 4 feed concentrations………………………………………………………...….36 Figure 2.13. Experimental and computed D 4 concentration profiles for various D 4 feed concentrations……………………………………………………………36 Figure 2.14. Experimental and computed L 2 conversion profiles for various L 2 feed concentrations…………………………………………………………....38 ix Figure 2.15. Experimental and computed L 3 conversion profiles for various L 3 feed concentrations………………………………………………..…………..38 Figure 2.16. Experimental and computed L 3 conversion profiles for various L 3 feed concentrations………………………………………………...………….39 Figure 2.17. Experimental and computed D 4 conversion profiles for various D 4 feed concentrations……………………………………………………………39 Figure 2.18. D 4 conversion and solid particle volume fraction profiles………...……..44 Figure 2.19. Maximum solid particle volume fractions for different siloxanes……….45 Figure 2.20. Renewable natural gas Bunsen flame (a) with (16 ppm v of D 4 ) and (b) without siloxanes………………………………………………………...46 Figure 2.21. A silica particle deposited on the Ni/Cr plate after 15s of exposure time and its EDAX spectrum………………………………………………….48 Figure 2.22. SEM image of the surface after 20min of exposure……………………..50 Figure 2.23. AFM images of the metal surface (a) before deposition, average roughness 3.66nm, and (b) after deposition, average roughness 10.64nm……….…51 Figure 2.24. Simulation and experimental concentration profile results for D 4 +L 4 mixture in the pre-flame, flame, and post-flame regions………………...53 Figure 2.25. Comparison between simulation and experimental concentration profile results for D 4 while decomposing individually or in the mixture with L 4 ................................................................................................................54 Figure 2.26. Comparison between simulation and experimental concentration profile results for L 4 while decomposing individually or in the mixture with D 4 ………………………………………………………………….……..54 Figure 2.27. Simulation and experimental concentration profile results for L 2 +L 3 +L 4 mixture in the pre-flame, flame, and post-flame regions…………...……55 Figure 2.28. Comparison between simulation and experimental concentration profile results for L 2 while decomposing individually or in the mixture with L 3 and L 4 …………………………………………………………………….56 x Figure 2.29. Comparison between simulation and experimental concentration profile results for L 3 while decomposing individually or in the mixture with L 2 and L 4 ……………………………………………………………….……57 Figure 2.30. Comparison between simulation and experimental concentration profile results for L 4 while decomposing individually or in the mixture with L 2 and L 3 …………………………………………………………………….57 Figure 2.31. Maximum solid particle volume fractions for mixture of L 2 and L 3 …….58 Figure 2.32. Maximum solid particle volume fractions for mixture of L 2 and L 4 …….59 Figure 2.33. Maximum solid particle volume fractions for mixture of L 2 and D 4 …….59 Figure 2.34. An SEM image of the sample's surface deposited by a RNG Bunsen flame containing 16ppm v of D 4 +L 4 …………………………………………..…61 Figure 2.35. Compositional analysis of the SEM image shown in Figure 2.34…….…62 Figure 2.36. An SEM image of sample's surface using D 4 +L 4 ……………………......62 Figure 2.37. Temperature dependence of the vapor pressure of Si(g) over Si(s,l)……… ■: Solid (Schmude, 1994), □: Liquid (Schmude, 1994), ●: (Drowart et al., 1958), ——: Solid, Calculation (Barin, 1995), ------: Liquid, Calculation (Barin, 1995), ○: (Souchiere and Binh, 1986), ▲: (Gulbransen et al., 1966), …..: Calculation (Honig, 1954; Honig, 1957), ∆: (Batdorf and Smits, 1959)……………………………………………………………...65 Figure 2.38. Temperature dependence of vapor pressure of Si(g) over Si(l). ■: (Tomooka et al., 1999), : (Schmude, 1994), ..: (Drowart et al., 1958), …..: (Barin, 1995)………………………….……67 Figure 2.39. Comparison of available experimental data for SiO (am) evaporation…….69 Figure 2.40. A comparison of the current experimental data for SiO (am) evaporation with thermodynamic assessment of SiO vapor pressure………………....70 Figure 2.41. Vapor pressure of silica at 1.0atm obtained by decomposition in the air...............................................................................................................71 Figure 2.42. Logarithmic plot of partial vapor pressure of SiO(g) over extended temperature range resulting from solid silica dissociation………………72 Figure 3.1. Schematic of the experimental configuration……………………………82 xi Figure 3.2. Schematic of the counter-flow configuration (Adapted from Feng, 2011)……………………………………………………………………..84 Figure 3.3. Computed and experimental silane concentration profiles in the pre-flame, flame, and post-flame regions for various silane feed concentrations…...90 Figure 3.4. Computed and experimental silane concentration profiles in the pre-flame, flame, and post-flame regions of simulated lean ( φ=0.8) RNG flames for the 7 ppm v feed concentrations using the Miller et al. (2004) and Britten et al. (1991) silane decomposition models…………………………...……..91 Figure 3.5. Reaction path analysis resulting in SiO 2 formation in the simulated lean ( φ=0.8) RNG flames for a 7 ppm v feed concentration of silane using the Britten et al. (1991) model. Each number indicates the fractional (%) consumption of the radical from which the arrow begins from………….92 Figure 3.6. Reaction path analysis resulting into SiO 2 formation in the simulated lean ( φ=0.8) RNG flames for a 7 ppm v feed concentration of silane using the Miller et al. (2004) model. Each number indicates the fractional (%) consumption of the radical from which the arrow begins with……….....93 Figure 3.7. Sensitivity of the maximum HSiO(OH) mole fraction on the kinetics of silane decomposition in simulated lean ( φ=0.8) RNG flames for 7 ppm v feed concentrations of silane, computed using the model of Britten et al. (1991)…………………………………………………………………….98 Figure 3.8. Sensitivity of the maximum HSiO(OH) mole fraction on the kinetics of silane decomposition in simulated lean ( φ=0.8) RNG flames for 7 ppm v feed concentrations of silane, computed using the Miller et al. (2004) model………………………………………………………………..……99 Figure 3.9. Simulated and experimental silane conversion and solid particle volume fraction profiles…………………………………………………………100 Figure 3.10. Maximum solid particle volume fraction for silane in comparison with the siloxanes…………………………………………………………...……102 Figure 3.11. An SEM image of deposited particles using 16 ppm v silane for 2 min……………………………………………………………………...104 Figure 3.12. Surface analysis results of the area shown in Figure 3.11………..…….105 xii Figure 3.13. EDAX analysis result of a particle formed using 16ppm v silane for 2 min……………………………………………………………………...106 xiii Abstract Biogas, which is produced from sludge biodegradation in wastewater treatment plants (WWTP), and landfill gas (LFG), which is generated from solid waste in landfills, are potentially important renewable fuels. Gas turbines and conventional internal combustion engines can combust LFG (or biogas) to generate electricity. Aside from their main components, such as methane and carbon dioxide, biogas and LFG may also contain undesirable contaminants. A particularly bothersome such trace constituent is a class of compounds known as siloxanes. The corrosion and damage that can be caused by these impurities may reduce the operating life of power and electricity generation equipment. In this research, the decomposition of siloxanes present in simulated renewable natural gas (RNG), which is LFG (or biogas) after its methane content has been upgraded to meet natural gas (NG) pipeline standards, is experimentally investigated in order to provide a better knowledge of their fate during RNG combustion. In the study, individual siloxanes and their mixtures in trace amounts were introduced into RNG flat flames. The counter- flow experimental technique was utilized which allows one to accurately probe and analyze the mechanisms leading to the formation of silica micro-particles resulting from the decomposition of the siloxanes. In order to probe the nature of deposits formed on surfaces in proximity to the RNG flames, flat Ni/Cr metal strips were placed downstream of a Bunsen flame. The chemical composition, morphology, and structure of the solid particles formed were investigated by scanning electron microscopy (SEM), energy dispersive analysis by X-ray (EDAX), and atomic force microscopy (AFM). It was found that microcrystalline silica particles xiv were generated during RNG combustion, leading to a rapid coverage of the surface of Ni/Cr strips placed in the flame environment, and forming eventually a light white layer of solid particles. The size of these particles was estimated using SEM/EDAX. The effect of siloxane concentration on its conversion along the flame was studied via the use of the GC-MS technique for a number of different feed concentrations. Volume fractions of particles within the flame were measured via the laser extinction method indicating a linear relationship between the concentration in the fuel and the corresponding volume of particles in the flue-gas. The temperature profile was measured experimentally to help identify the kinetics of the burning of the siloxane compounds. This fundamental insight is important in terms of being able to accurately determine the maximum allowable siloxane content for biogas that is safe to use without leading to deposits and micro-particle formation. 1 Chapter 1 – Introduction 1.1. Overview and Significance Biogas from waste water treatment plants (WWTP) and landfill gas (LFG) are considered to be potential important renewable fuels as they contain large amounts of methane (typically >50% -- other major components being CO 2 , N 2 , O 2 , and H 2 O). Their use for energy production (e.g., when used in boilers) and electric power generation is attracting current interest since it addresses the environmental pollution issues resulting from fugitive emissions or the flaring of the gas, and may also offer help towards reducing our country’s reliance on fossil fuels (e.g., Shin et al., 2002). Gas turbines and conventional Internal Combustion (IC) engines can easily combust LFG (and biogas) to generate electricity. Unfortunately, LFG also contains a myriad of other trace non-methane organic compounds (collectively known as NMOC – more than 140 different NMOC have, so far, been identified (e.g., Schweigkofler and Niessner, 2001)), several of which also contain hetero-atoms such as halogens, sulfur, and silicon. The halogen- and sulfur-containing NMOC during combustion produce acids like H 2 SO 4 , HCl, and HF which corrode the energy and electric producing equipment (e.g., the boilers and IC engines) and reduce their operating life (e.g., Ajhar et al., 2010; Popat and Deshusses, 2008). They also contribute towards the 2 formation of acid rain, and as a result may also represent a hazard towards the entire planet. A particularly bothersome trace component in LFG (and biogas) is a class of compounds known as siloxanes. The word siloxane is derived from the words silicon, oxygen, and alkane. Siloxanes are frequently found in many commercial and consumer products, such as detergents, shampoos, deodorants, and cosmetics. Because of their increased use in a variety of products, siloxanes have emerged in recent years as one of the most difficult contaminants to control in LFG. Unfortunately, today, increasing quantities of siloxanes end-up in sewage and landfills where consumers discard these products, eventually becoming trace impurities in the LFG that these landfills generate. During combustion of biogas containing these volatile silicon-based compounds, silicon is known to be released and can be combined with oxygen as well as the various other compounds found in the combustion environment. Silica-containing deposits are found to form on various surfaces in contact with the combustion gases, which sometimes build to a thickness of millimeters, and are difficult to remove by chemical and/or mechanical means. This causes damage to the IC engines, turbines, and boiler tubes, as well as to the air pollution control devices and sensors, and necessitates increased maintenance (e.g., Pacey et al., 1994). Another key problem with these fine silica micro-particles that form, is that unless adequate precautions are taken, they will most likely escape into the atmosphere, where they may pose a risk to both human health and the environment, and may cause serious air pollution problems. This issue has really become of recent keen concern, because of the prospect of biomethane or renewable natural gas (RNG), which is LFG or biogas whose methane 3 content has been enriched to meet natural gas (NG) pipeline standards, that may contain siloxanes, being introduced into the NG supply (pipeline) system and being utilized in home appliances (e.g., stoves, water heaters, etc.). Traditionally, trace constituents in biogas are removed by adsorption or absorption techniques. These techniques are expensive to apply, but remain in common use because there are no other commercially available processes to replace them. Unfortunately, these techniques are not at all effective for the removal of siloxanes, because none of the common adsorption/absorption media are uniquely selective towards the siloxanes in the presence of the other numerous NMOC found in LFG. Siloxanes, as a result, represent a key challenge that the use of renewable fuels like LFG currently faces. There is a clear need, therefore, today to understand how the silica micro-particulates form during combustion of RNG under conditions akin to those encountered in common energy and electricity producing equipment (e.g., IC engines, turbines and boilers), as well as in common household devices (e.g., ovens, water heaters, furnaces, etc.), and how the formed micro-particles deposit on various surfaces in contact with the combustion mixtures. As noted above, formation of such silica-containing films on boiler pipes and turbine blades is, of course, common, but there are also anecdotal reports of silica deposits appearing in residential end-use equipment and devices, likely resulting from the combustion of gas containing siloxanes. If the use of renewable gaseous fuels makes the expected future inroads into residential power and energy generation, siloxanes may represent a serious potential problem for the natural gas distributors. 4 1.2. Project Objectives Prior to initiation of our study there were a limited number of research studies about the decomposition of certain types of siloxanes. However, none of these studies provide a comprehensive understanding of the oxidation kinetics of these impurities. In particular, prior to our own study there was no fundamental investigation focusing on the determination of the decomposition kinetics of these compounds in the RNG flame environment. As a result, the objective of this study was to investigate in the laboratory the fate of siloxanes during combustion of bio-methane or RNG derived from biogas, in order to understand how the solid silica micro-particulates form in the gas phase and to determine the decomposition kinetics as well as the bulk and surface characteristics of these deposits. In order to accomplish these goals, the primary task involved collecting experimental data under well-controlled laboratory conditions with the aim of identifying the chemical and physical mechanisms involved in silica- containing micro-particulate formation and deposition during bio-methane combustion. In order to allow for the generation of fundamental understanding, the opposed-jet counter-flow stagnation flame experimental technique, which allows one to accurately probe and analyze the mechanisms leading to the formation of the silica micro- particles, was utilized. The laser extinction technique was also applied for the in-situ monitoring of particle formation (number density), and for calculating the maximum solid particles volume fraction; mass spectrometric and gas chromatographic techniques were utilized for measuring the kinetics of siloxane decomposition. A simulated RNG composition consisting of 98% CH 4 and 2% CO 2 , spiked with various concentrations of siloxanes, was utilized. During the experiments, different mixtures 5 of siloxanes were introduced into the simulated RNG, in order to investigate the possible effect of siloxanes on each other’s decomposition kinetics in a mixture. A key focus of the study was studying the relationship between the siloxane’s content in the gas and the amount of SiO 2 found in the post-flame region, as well as investigating the nature of deposits formed on surfaces placed downstream of the flames. Such fundamental insight is important in terms of being able to accurately determine the maximum allowable siloxane content for biogas that is safe to use without leading to deposits and to micro-particles formation. This is important in terms of deciding on the type of clean-up processes to be used to prevent SiO 2 deposition during the use of RNG. Specifically, the effect of exposure time to siloxanes, and of the initial temperature of the sample’s surface on the size, chemical composition, and shape of solid silicon-based particles were investigated by the SEM/EDAX technique. Moreover, the change in the morphology and roughness of the surfaces due to particle deposition was analyzed by the AFM technique. In this study, in addition to siloxanes, silane was used as a model silicon- containing precursor and the same experimental techniques, which were used for siloxanes, were applied for silane as well, and the major proposed oxidation mechanisms were appraised experimentally and numerically. By that, the major channels leading to solid silica particles were identified. Comparing the experimental data for these two types of silica forming compounds has provided a comprehensive insight about how these micro-particles form during RNG combustion, and may suggest solutions about control and prevention of air pollution and the sustainable use of these important renewable fuels. 6 In summary, the primary objective of this study was to complete the collection of fundamental experimental flame data in order to identify the chemical and physical mechanisms involved in silica particle formation and deposition during bio-methane combustion. This is in order to evaluate the impact of siloxane decomposition and particle deposition on energy generating and household appliance performance, and to determine the maximum allowable siloxane concentration in the RNG. 1.3. Organization of Dissertation This Thesis is structured as follows: In Chapter 2, the fate of siloxane impurities during the combustion of renewable natural gas is discussed. Different experimental methodologies are applied in order to trace the siloxanes in RNG flames and to see what kind of solid particles are formed as a result of siloxane decomposition. First order global oxidation reactions are assumed and kinetic constants of decomposition reactions are calculated using the experimental concentration profiles. Solid particle volume fractions of different siloxanes are calculated using the laser extinction technique. In order to investigate the identity of the solid particles formed and deposited under different conditions, SEM/EDAX techniques are applied. Also, the effect of solid particles’ impact on the surface of the substrate is discussed in the Chapter. Lastly, the decomposition of siloxane mixtures is investigated theoretically and experimentally and the results are compared to those with individual siloxanes. In Chapter 3, the decomposition of silane as a model silicon-containing precursor is studied. The major channels leading to solid silica production are identified by simulation tools, and the different available oxidation mechanisms are evaluated by 7 the experimental concentration data. Also, the solid particle volume fractions are measured and the results are compared with the ones obtained from siloxanes. The complete surface analyses results are reported and the effect of each parameter, e.g., initial temperature, exposure time, etc. are investigated. Also, the effect of particle deposition on the surface of the substrate is probed and results are reported. Finally, the findings and contributions of this dissertation are summarized and recommendations for future work are presented in Chapter 4. 8 Chapter 2 – Fate of Siloxane Impurities during the Combustion of Renewable Natural Gas 2.1. Introduction 2.1.1. Biogas History and Background Biogas, which is produced from sludge biodegradation in wastewater treatment plants (WWTP), and landfill gas (LFG), which is generated from solid waste in landfills, are potentially important renewable fuels. They typically contain large amounts of methane (40-70vol.%) and also 30-60vol.% carbon dioxide (El-Fadel et al., 1997; Ohannessian et al., 2008; McBean, 2008; Abatzoglou and Boivin, 2009; Schweigkofler and Niessner, 2001; Shin et al., 2002). A comprehensive overview on the biogas and leachate formation mechanisms in landfills is presented in by El-Fadel et al. (1997). Technical, social, economic, and environmental studies of biogas production are all encouraging the use of biogas to generate energy (Boulinguiez and Le Cloirec, 2010; Murphy and McKeogh, 2006). Biogas is generally used for heating, electrical generation and cogeneration (Ohannessian et al., 2008; McBean, 2008). The predictably high percentage of methane in biogas has provided significant opportunities such that there are now hundreds of facilities at which the biogas is being productively used. Additional value arises from the use of biogas because of 9 reduction of its Global Warming Potential (GWP). Concerns about global warming are likely to encourage further capture and utilization, since fugitive emissions from WWTP and landfills are among the largest man-made sources of CO 2 emissions (McBean, 2008). Aside from the main components in biogas (Methane and Carbon Dioxide), it may also contain undesirable contaminants including sulfides (e.g., H 2 S, COS, CS 2 ), ammonia (NH 3 ), volatile organo-silicon compounds (siloxanes), aromatic, aliphatic, mercaptans, halogenated compounds, and other trace substances like oxygen, nitrogen, argon (Ohannessian et al., 2008; Abatzoglou and Boivin, 2009; Shin et al., 2002; Boulinguiez and Le Cloirec, 2010; Eklund et al., 1998). More than 140 other components have been identified, so far, and they reach a total concentration of up to 2000 mg/m 3 or 0.15 vol.% (Schweigkofler and Niessner, 2001). The halogen- and sulfur-containing contaminants during combustion produce acids like H 2 SO 4 , HCl, and HF, which corrode the energy and electric producing equipments (e.g., the boilers and Internal Combustion (IC) engines), thus necessitating more frequent maintenance and reducing operating life. They also contribute towards the formation of acid rain, and as a result may also represent a hazard for plant, human health, and animal life (Schweigkofler and Niessner, 2001). 2.1.2. What are the Siloxanes? A particularly bothersome trace constituent of landfill gas (LFG) is a class of compounds known as siloxanes (see Table 2.1 for a list of frequently reported Si- containing compounds in biogas and LFG) or volatile methyl siloxanes (VMS). The name siloxane is derived from their elements: Silicon + Oxygen + Alkane (Ricaurte 10 Ortega and Subrenat, 2009). Siloxanes are organic compounds containing in their backbone alternating silicon and oxygen atoms and attached methyl groups. Volatile siloxanes, as those typically found in LFG, are often referred to in a shorthand notation as containing M or D components, which are shown in Figure 2.1. M Group D Group Figure 2.1. Siloxane group nomenclatures. Following this nomenclature, the linear octamethyltrisiloxane is represented as MDM (or L 3 ), while octamethylcyclotetrasiloxane, a cyclic siloxane, is represented as D 4 . The molecular structures of these compounds are presented in Figure 2.2a and b. a b Figure 2.2. Examples of siloxanes’ chemical structure. a) Linear structure of octamethyltrisiloxane (MDM or L 3 ), b) Cyclical structure of octamethylcyclotetrasiloxane (D 4 ). 11 Detailed explanations about the chemical structure, physical properties, source, and applications of common siloxanes found in the LFG are provided by Dewil et al. (2006), Ohannessian et al. (2008), and Oshita et al. (2010). Siloxanes are extensively applied in a variety of consumer products, such as cosmetics, detergents, paper coatings, textiles, and shampoos. Also, in the industrial processes of silicon containing chemicals siloxanes are released as a residue (Dewil et al., 2006). Unfortunately, today, increasing quantities of siloxanes end-up in WWTP and in landfills, where consumers discard these products, eventually becoming trace impurities in the LFG that these landfills generate. VMS are identified as the most adverse amongst other contaminants for the utilization of LFG (Ohannessian et al., 2008). Because of the challenges that siloxanes present to the beneficial use of biogas and LFG, they have attracted attention, particularly in recent years. Badjagbo et al. (2010) and Crest et al. (2010), for example, reported on techniques to measure the siloxane content of LFG; this remains a challenge for the renewable energy industry, as field instruments are still lacking in sensitivity and reliability. 12 Name CAS Number Formula MW [g/mol] Boiling Point [ o C] Vapor Pressure [mmHg] Density [g/cc] Hexamethylcyclotrisiloxane (D 3 ) 541-05-9 C 6 H 18 O 3 Si 3 222.46 134 10 1.02 Octamethylcyclotetrasiloxane (D 4 ) 556-67-2 C 8 H 24 O 4 Si 4 296.62 175 1.3 0.96 Decamethylcyclopentasiloxane (D 5 ) 541-02-6 C 10 H 30 O 5 Si 5 370.77 210 0.4 0.96 Dodecamethylcyclohexasiloxane(D 6 ) 540-97-6 C 12 H 36 O 6 Si 6 444.92 245 0.02 0.97 Hexamethyldisiloxane (L 2 ) 107-46-0 C 6 H 18 OSi 2 162.38 101 31 0.76 Octamethyltrisiloxane (L 3 ) 107-51-7 C 8 H 24 O 2 Si 3 236.53 153 3.9 0.82 Decamethyltetrasiloxane (L 4 ) 141-62-8 C 10 H 30 O 3 Si 4 310.69 194 0.55 0.85 Dodecamethylpentasiloxane (L 5 ) 141-63-9 C 12 H 36 O 4 Si 5 384.84 230 0.07 0.88 Trimethylsilanol 1066-40-6 C 3 H 10 OSi 90.20 99 73.9 0.81 Table 2.1. Siloxane compounds in LFG and some of their properties. 13 During biogas combustion, siloxanes are decomposed to silanols (Si-OH) and various carbonyl compounds, which eventually are oxidized to carbon dioxide. Also silanols are reported to be oxidized, at higher temperatures in combustive environments, to silicates (SiO 3 and SiO 4 ) and micro-crystalline silicon dioxide (SiO 2 ) (Dewil et al., 2006; Nair et al., 2012; Schweigkofler and Niessner, 2001), which build a surface film with a thickness of several millimeters, very difficult to remove by chemical or mechanical treatments. These abrasive deposits on the inner walls of engines, heat exchangers, furnaces (Nair et al., 2013), gas turbine blades, boilers, etc., may lead to serious damage, increasing the cost of engine maintenance (Ajhar et al., 2010; Popat and Deshusses, 2008; Ohannessian et al., 2008). In boilers, the silica layers act as thermal insulators interfering with heat exchange (McBean, 2008). In engines, they are known to clog narrow passages (Oshita et al., 2010), increasing the potential for accidental explosions, and they are known to interfere with catalytic treatment systems decreasing their efficiency (Badjagbo et al., 2010; Nair et al., 2012; Ohannessian et al., 2008). Another concern with these fine silica particles is that unless they are removed from the exhaust gas, they will escape into the atmosphere, where they may pose a risk to both human health and the environment. This is an issue of recent concern in California, and the main motivation for this study, because of the prospect of wider use of renewable natural gas (RNG). RNG results from biogas or LFG after their non-methane organic compound (NMOC) impurities are removed, and their methane content is upgraded to meet natural gas (NG) pipeline standards. The concern here is with malfunction of siloxane removal equipment (see below), which may accidentally release these impurities into the NG supply (pipeline) system; this then may result in their being combusted in home appliances (e.g., stoves, 14 water heaters, etc.), thus causing harmful environmental emissions and deleterious effects on human health (Abatzoglou and Boivin, 2009; El-Fadel et al., 1997). 2.1.3. Siloxanes Removal Techniques Given the importance of ridding RNG of siloxanes, it is not surprising that there are numerous reports on different removal techniques (e.g., Ajhar et al., 2010; Dewil et al., 2006). Adsorption is the most common method (e.g., Matsui and Imamura, 2010; Montanari et al., 2010; Oshita et al., 2010; Ricaurte Ortega and Subrenat, 2009), but a key problem is that the media (activated carbon, zeolites, activated alumina, silica gel, etc.) used are not selective toward siloxanes, and they adsorb (often more preferentially) most other NMOC; this, then, reduces the bed's capacity for siloxane adsorption, necessitating frequent regeneration and using multiple beds. Siloxanes are difficult to remove during regeneration (Wheless and Gary, 2002), which results in adsorbents of progressively lower capacity, until replacement becomes necessary, at a great cost. A key challenge with adsorption (and all other physical methods), furthermore, is that it does not change the molecular state of siloxanes. Regeneration involves burning the off-gas, which releases the SiO 2 particles into the atmosphere, and wastes RNG for the incinerator operation. Absorption in solvents (e.g., Selexol and methanol) is another approach (e.g., Schweigkofler and Niessner, 2001), but the problem is the high capital, operating, and maintenance costs. Solvent regeneration is the key to success of reducing solvent disposal and operating costs, and of ensuring long-term operation. Refrigeration also has been tried, but does not appear to be effective on its own (e.g., Wheless and Gary, 2002). Hybrid processes combining refrigeration with adsorption/absorption show 15 more promise (Schweigkofler and Niessner, 2001). However, the high energy consumed for cooling the wet gas is a hindrance to commercialization. Biological treatment has been studied also (e.g., Accettola et al., 2008; Ohannessian et al., 2008; Popat and Deshusses, 2008) because of promise for siloxane "mineralization", but results, so far, are disappointing with conversion harboring ~10%. Removal via membranes has also been proposed (e.g., Ajhar and Melin, 2006), but no experimental data are available currently, and the practicality is unclear, as membranes are, in general, not well suited for removing trace impurities. Reactive approaches have been utilized also. Appels et al. (2008), for example, used peroxidation to reduce the sludge siloxane content prior to biodegradation in a digester. A reduction of 50–85% was observed, which is not competitive with other technologies. Finocchio et al. (2008) studied the decomposition of D 3 on both basic (CaO, MgO) and acidic oxides (Al 2 O 3 , SiO 2 ), and showed that reactive adsorption occurs accompanied by surface silication and release of methane. A model was proposed by Chandramouli and Kamens (2001) for the photochemical oxidation of two siloxanes (L 4 and D 5 ), which these authors studied experimentally. A reaction/solid-partitioning model was implemented in order to explain the formation of the solid products. They concluded that most of the reaction products would partition into the particle phase due to their high affinity for the solid particles. Regeneration of the spent adsorbents is not possible, however, and the basic oxides lose their reactivity in contact with CO 2 due to carbonation. UV photodecomposition of L 2 has been investigated also in bench-top experiments with promising results (Prosser, 2010), but the approach still needs to be field-tested and its economics be investigated further. 16 In summary, conventional methods (adsorption, absorption, refrigeration) face significant hurdles when applied to siloxane removal, but remain in use because there are no other commercial processes to replace them. Novel techniques (e.g., bio- filtration, membranes, reactive approaches) face hurdles of their own and/or have yet to be field-tested. Thus, as one moves forward with the use of RNG, the potential for siloxanes finding their way into RNG equipment and common household appliances remains a challenge, and motivates the need for this study aimed to understand how silica particulates form during combustion of RNG, and how they deposit on various surfaces in contact with the combustion mixtures. 2.1.4. Solid Particles Formation in Flame Particulate formation in flame environments is of current scientific interest, in general, as it is used commercially to prepare powders of a broad range of materials. Flame aerosol technology is widely used in industrial applications to make various commodities, such as ceramics (pigmentary titania and fumed silica) and carbon blacks as well as other specialty chemicals such as alumina and zinc oxide powders (Pratsinis, 1998). Gas-phase combustion synthesis offers significant advantages over other material synthesis processes. This technology is based on its apparent simplicity of a one-step process and no-moving parts machinery (Kammler et al., 2001). Moreover, the high temperature flame environment is self-purifying and can be designed for a wide range of operating conditions, e.g., temperature, reactant concentration, etc. (Wooldridge, 1998). There is a very broad range of various applications of these powders. For example, silica nano-particles have applications in thickening control, thixotropy, and material reinforcement, as well as a stabilizer for 17 suspensions, or as a chemical mechanical polishing (CMP) agent (Kammler et al., 2001). Other applications are in high-temperature superconductor, electronic substrate, and catalysts applications (Wooldridge, 1998). Some of the earliest work on the characterization of particle formation of refractories in flames using silicon tetrachloride (SiCl 4 ) to produce silica was conducted by Ulrich (1971), Ulrich and Subramanian (1977), and Ulrich and Riehl (1982). Their work focused on the agglomeration processes following the formation of silica particles. There are good advances and accomplishments in flame aerosol technology in which many academic groups have conducted valuable research regarding the synthesis of nano-particles (especially silicon dioxide) and the effect of various parameters which may influence particle characteristics. A comprehensive review was published by Wooldridge (1998) where the author provides a summary of experimental and theoretical investigations of the fundamental processes governing gas-phase combustion synthesis of particles. Different experimental techniques and results are evaluated and parameters influencing particle morphology and composition are discussed. In her review she concluded that the particle growth mechanism (i.e., nucleation, deposition, coalescence, and agglomeration) is a strong function of the system conditions. Different system parameters which are identified as effective tools to control the particles morphology are listed in the review. Precursor concentration, and the combined temperature field and residence time of the particles in the synthesis environment are claimed to be the most significant factors (Pratsinis, 1998; Ulrich, 1971; Zhu and Pratsinis, 1997; Formenti et al. 1972). In a earlier study, Chung et al. (1991) concluded that at low SiH 4 concentrations and/or when a high-temperature flame is used, SiO 2 is the species that nucleates. At high SiH 4 concentrations and/or 18 when a low-temperature flame is used, nucleation of SiO x (x = 0 and/or 1) in addition to SiO 2 is believed to occur in the flame. Other system parameters, such as supporting flame stoichiometry, diluent concentration, and gas flow rates, can be used to control morphology to the extent that they affect the system temperature field and the particle residence time. Pratsinis (1998) and Kammler et al. (2001) presented a broad review on the state of knowledge of flame aerosol synthesis of ceramic powders. The process of particle synthesis is described as well as various advancements in this technology and different models (Koch and Friedlander, 1990) describing particle growth and morphology are presented. The particle formation involves three steps: (1) formation of condensing species through chemical reactions of precursor compounds; (2) nucleation of the condensing species to produce initial particles; and (3) growth of the initial particles to form final product particles (Chung et al. 1991). Also the synthesis processes of various nano-powders (e.g., SiO 2 , TiO 2 , and Al 2 O 3 ) from their gaseous precursors are discussed and the effect of different factors such as flame temperature, particle residence time, oxidant type, burner geometry, etc., on the formation and control of particle characteristics in flame aerosol reactors is reviewed and investigated. Another extensive review regarding flame aerosol synthesis technology and particle characterization was published by Gurav et al. (1993) who reviewed almost 400 papers in this field. Relatively less attention has been focused on siloxanes as precursors to silica particulates (for key studies see Table 2.2). For example, Whelan et al. (2004) studied the fate of volatile methyl siloxanes when released into the atmosphere and concluded that they degrade by reacting with OH radicals to form silanols (Sommerlade et al., 19 1993; Whelan et al., 2004). Hardesty and Weinberg (1973) were the first to synthesize silica particles via combustion of L 2 in an electric field–induced, laminar, premixed, flat methane/air flame. They analyzed by transmission electron microscopy (TEM) the silica particles deposited directly onto TEM carbon grids, and concluded that their size depends on the residence time in the flame. The production of high-purity fused silica in a flame from polydimethylsiloxanes was claimed in a patent by Dobbins and McLay (1991). 20 Authors Precursor Fuel Flame Configuration In-Situ Diagnostics Ex-Situ Diagnostics Precursor Concentration Range Particle Gathering Method Remarks Hardesty and Weinberg (1973) L 2 CH 4 /Air Flat Flame Burner, Premixed N/A TEM N/A Directly Deposited onto Carbon TEM Grids The size of silica particles depends on the residence time in the flame. Hurd and Flower (1988) L 2 CH 4 /Air Flat Flame Burner, Premixed, Lean, Laminar Static and Dynamic Light Scattering (514.5nm) TEM Unknown Directly Deposited onto Carbon TEM Grids The growth and aggregation of silica particles studied as a function of position in flame. The growth and aggregation of particles was found as a function of height in the flame. Zachariah and Semerjian (1990) SiH 4 , L 2 , TMS H 2 /O 2 /Ar Counter-flow Diffusion Burner Dynamic Light Scattering, Laser Scattering Dissymmetry (514.5 nm) N/A N/A N/A The effect of precursor on particle growth was investigated. The organo-silicon compounds produce more aggregated structure deposits than the silane. Chagger et al. (1996) L 2 CH 4 /Air/ N 2 Counter-flow Diffusion Burner FTIR, Emission Spectroscopy Micro-probe Sampling, Analyzed by GC/FID, BET Max. 1.3 mol% N/A Si-H and Si-O were shown to be present. The initial particle size was found to be ~10 nm. L 2 decomposes rapidly to SiO and consequently to silica. Briesen et al. (1998) L 2 , D 4 , SiCl 4 CH 4 /O 2 /N 2 Concentric Burners, Premixed and Diffusion Flame N/A BET 59-61% of Saturation for L 2 N/A The effect of precursor type and concentration and oxidant composition was investigated. Particles from L 2 have smaller surface areas than those from SiCl 4 . Ehrman et al. (1998) L 2 CH 4 /O 2 /N 2 Flat Flame Burner, Premixed N/A TEM, EDS EELS, STEM 4.4 × 10 -06 mol/L Deposited onto Copper TEM Grids, using N 2 Dilution The choice of precursor has no observable effect on the particle size. The primary silica particle size was 10±1 nm. Yeh et al. (2001a) L 2 C 3 H 8 /Air H 2 /Air Bunsen, Premixed N/A SEM, EDS, XPS, XRD 0.2-1.2 mol % Deposited onto an Al Plate High purity silica particles were synthesized in both fuels at low concentrations (0.2%). The effect of precursor concentration, flame temperature, water vapor, and O 2 mole fraction on silica synthesis was studied. Butler et al. (2002) L 2 H 2 /O 2 /N 2 Bunsen (Premixed), Concentric Burner (Diffusion Flame) N/A TEM, NMR, IR 0.05-1 mol% Thermophoretically Deposited onto a Copper Grid For flame temperatures >2000K, only SiO 2 condenses out to form particles, but if the flame is cooler than 2000K, SiO can also condense. Ma et al. (2003) L 2 C 3 H 8 /Air Bunsen, Premixed, Lean N/A EDS, XPS, XRD, FTIR, BET/TEM 0.2-1.2 mol% Deposited onto an Al Plate The synthesized products mainly consist of SiO 2 and a small amount of SiO. The nano- sized SiO 2 particles ranged from 2.5-25 nm. Table 2.2. Previous studies on the combustion synthesis of silica particles from siloxanes. 21 Flame temperature and particle residence time have been shown to be the most important parameters determining the characteristics of the product powder (Pratsinis, 1998; Ulrich, 1984). The results of another research indicate that temperature plays a significant role in the silica formation process, by controlling the chemical kinetic rates and through physical changes in particle morphology (Zachariah et al., 1989). Increasing flame temperatures tend to enhance chemical kinetics and homogenous nucleation, leading to smaller particles and more spherical in higher numbers. Hung and Katz (1992) found that increasing flame temperatures resulted in high concentrations of fine particles during synthesis of SiO 2 . By operating a diffusion flame reactor with oxygen instead of air (Zhu and Pratsinis, 1997) the size of silica product particles increased by about 5 times. Using oxygen leads to faster fuel consumption, and therefore to higher flame temperatures and to shorter flames, thus, the precursor oxidation as well as growth and sintering is enhanced. In a similar set- up, Briesen et al. (1998) found a similar increase of the primary product particle size investigating different silica precursors using SiCl 4 , L 2 , and D 4 when they replaced air by oxygen. In another study, Wooldridge (1998) reaches the same conclusion regarding the flame temperature effect on particle size; however, it is mentioned that the temperature effect is more variable than the other factors and it significantly depends on the particle residence time. A formation mechanism for SiO 2 particles was outlined as a function of residence time and temperature by Hung and Katz (1992). It shows that SiO 2 particles form at temperatures around their melting or softening temperature and as they aggregate they form short chain-like structures. Surface growth and aggregation cause continued growth of the particles until the particles melt or soften. 22 Zachariah and Semerjian (1990) and Ehrman et al. (1998) investigated different silica precursors and did not observe significant effects on SiO 2 particles. However, the choice of precursor becomes important when a considerable amount of heat is provided by the combustion of the precursor itself (Briesen et al., 1998). This becomes more pronounced for higher precursor concentrations. Silica particles made from organo-silicon compounds had significantly lower specific surface area (SSA) than those produced from SiCl 4 for all investigated oxidant flow rates in diffusion and premixed aerosol flame reactors. In premixed flames Briesen et al. (1998) correlated the specific surface area of the product with the adiabatic flame temperature, which is determined by the choice of precursor. Zachariah and Semerjian (1990) studied the effect of L 2 on the growth mechanism and morphology of the product silica particles. It was shown that aggregates with larger primary particles are made when organo- silicon precursors rather than SiH 4 are used in synthesis of SiO 2 in diffusion flames. Organo-silicon compounds release more heat upon combustion resulting in higher temperatures than SiH 4 oxidation and in effectively faster sintering rates. The Burner geometry also affects the particle size. The smaller burner means higher exit velocity, implying shorter residence time; therefore, the size of synthesized particles is smaller (Wegner and Pratsinis, 2005). In summary, biogas and LFG are poised to make inroads in energy and electricity generation via their planned addition (as RNG) into the NG pipeline system. Concerns exist with siloxanes, one of their contaminants, finding their way inadvertently into the NG pipeline system and into people's homes and in NG equipment and appliances. Based on the aforementioned considerations, the focus of the present study is the characterizations of the fate of siloxanes during combustion of RNG under well- 23 controlled laboratory conditions. Though some research has already been reported on siloxanes as precursors to silica particles, the conditions in these experiments (including the siloxane concentration) were far removed from those likely to be encountered in RNG combustion. The behavior among the various siloxanes is compared, based on results obtained for linear siloxanes of variable chain length and for cyclic and linear siloxanes of the same chain length. The emphasis is on understanding how silica deposits form on metal surfaces, and determining their bulk and surface characteristics as well as siloxane decomposition kinetics. 2.2. Experimental Set-up The siloxane decomposition was studied in the counter-flow flame configuration, shown schematically in Figure 2.3. Samples were withdrawn from the flame region and analyzed via GC/MSD to determine the siloxane concentration profiles. Temperature profiles were measured using thermocouples, while the laser extinction technique was applied for the in-situ measurement of the solid particles’ volume fraction, f v . In order to investigate the nature of deposits formed on surfaces in proximity to the RNG flames, flat Ni/Cr metal strips were placed downstream of a Bunsen flame. The chemical composition, morphology, and structure of the solid particles were investigated by scanning electron microscopy (SEM), energy dispersive analysis by X-ray (EDAX), and atomic force microscopy (AFM). 24 Figure 2.3. Schematic of the overall experimental configuration. The simulated RNG utilized in this study consists of 98% CH 4 and 2% CO 2 , both 99.97% pure, from the Gilmore Liquid Air Company (South El Monte, CA, USA). Four siloxanes, all commonly encountered in LFG and biogas (Table 2.1), were investigated, namely L 2 (99.5%, CAS: 107-46-0), L 3 (98%, CAS: 107-51-7), L 4 (97%, CAS: 141-62-8), and D 4 (98%, CAS: 556-67-2); all were purchased from Sigma- Aldrich (St. Louis, MO, USA). To generate ppm-level siloxane concentrations, a high-precision syringe pump (Harvard Apparatus HA-5T, PHD2000 using a Hamilton, 1025TLL model #82520 25.0 ml syringe) coupled to a quartz nebulizer with a flush capillary-lapped nozzle (Q-HEN-170-A0.1, 0.1 ml/min @ 170 psi, from Mienhard Glass Products, Golden, CO, USA) was utilized to generate a fine siloxane spray into a heated air stream, followed by further dilution to reach a concentration of ~1000 ppm v . A mass flow controller (MFC; Teledyne Hastings HFC-202) was then utilized 25 to deliver ~1% of this mixture to a premixed RNG/air stream to generate the desired siloxane concentrations in the range of (2 – 30 ppm v ). The ability to generate reliably the required concentrations was verified via a GC/MSD system (Agilent Technologies, 7890A GC System/5975C Inert XL MSD), calibrated using standard liquid samples of siloxanes in ethanol. The concentrations are reported here in terms of ppm v of siloxane in the premixed RNG/air stream. The concentration of siloxanes in biogas varies widely. Dewil et al. (2006), for example, surveyed the siloxane content in biogas from a number of sewage treatment plants and landfill sites, and they report it to be in the range of 4.8 mg/m 3 to 400 mg/m 3 . However, field studies have been reported with a much higher content of siloxanes (Ajhar et al., 2010). For example, Urban et al. (2009) field-tested a catalytic technology with LFG spiked with 200 ppm v (1449 mg/m 3 ) of L 2 and 100 ppm v (1333 mg/m 3 ) of D 4 . By comparison, the ranges of concentrations in the simulated RNG utilized in this study are L 2 (169 – 2368 mg/m 3 ), L 3 (246 – 3072 mg/m 3 ), L 4 (322 – 2580 mg/m 3 ), and D 4 (308 – 2464 mg/m 3 ). The final RNG/air/siloxane gas stream was introduced into the bottom burner. Experiments are carried out under ambient pressure in the counter-flow configuration, developed by Law et al. (1988), Wu and Law (1985), and Egolfopoulos et al. (1989) as schematically shown in Figure 2.4. Each burner was made out of a straight tube as well as a nitrogen co-flow channel surrounding each nozzle. The shape of the co-flow channel assures the isolation of the main jet from the ambient environment. This experimental system allows one to carry out experiments in the stagnation-flow configuration in order to study siloxane decomposition and soot emissions of both premixed and non-premixed (diffusion) flames. Flat flames were stabilized in ambient 26 temperature by counterflowing against this fuel/air stream a N 2 stream from the top burner. The nozzle diameter for both burners was 14 mm, and so was the separation distance between the nozzles. All tubes from the vaporizer to the burner were heated by wrapping them with heating tapes (HTS/Amptek, AWH-051-080DL) to minimize potential condensation and/or adsorption. Two K-type thermocouples (Omega SMP- RT-K-125 G-6) along with two temperature controllers (Omega CN9110A) were used to control the delivery line temperature to maintain the gas-phase siloxane partial pressure below its vapor pressure. For the experiments reported here, the air/fuel equivalence ratio, φ, of the stream exiting the lower burner was kept at 0.8, and the global strain rate K (defined as twice the nozzle exit velocity divided by the burner separation distance) was kept constant at 131.6 s -1 . Figure 2.4. Schematic of the counter-flow configuration (Adapted from Feng, 2011). 27 One such flame generated in the system in Figure 2.4 is shown in Figure 2.5a. For some of the experiments, specifically those involving the interaction of surfaces with the flame environment, there was no nitrogen flow from the top burner, and therefore a Bunsen flame was generated, see Figure 2.5b. a b Figure 2.5. Renewable Natural Gas flame. a) Opposed-jet premixed flat flame, b) Bunsen flame. Thermocouples were used to measure the flame temperature profiles, as noted above. Thermocouples are known to perturb the flow fields, and to interfere with heat transfer mechanisms such as radiation, thus their presence needs to be accounted for (Heitor and Moreira, 1993). In this study, an R-type thermocouple (Omega P13 R- 003) was utilized coated with a Y/BeO ceramic film to minimize its surface reactivity (Kent, 1970; Shandross et al., 1991). To avoid bending of the thin thermocouple wire when placed in the hot flame, which makes it difficult to determine the measurement position, a special device (Figure 2.6) was built similar to that in Cundy et al. (1986) and Ahuja and Miller (1993). The thermocouple assembly was mounted onto a multistage translational stand capable of moving in µm increments. The cold junction 28 temperature was adjusted by calibrating it at room temperature. The radiation corrections were made based on the approach of Shaddix (1999) and McEnally et al. (1997) using an emissivity value of 0.3 (Peterson and Laurendeau, 1985). Figure 2.6. A device for temperature measurements. For measuring the siloxane concentrations in the flame, as previously noted, gas- phase sampling was used with a quartz microprobe with an end opening of ~150 µm positioned on a stand having 3-D (XYZ) movement (Thorlabs, PT3/M, 25 mm XYZ Translation Stage). To begin the sampling, the tip of the probe was positioned at the top of the lower burner (Z=0), and then moved vertically in steps to collect gas samples in the pre-flame, flame, and post-flame regions. Gas phase samples (500 µl) were taken in each position using a 1.0 ml syringe (Hamilton, 1001SL 1.0 ml SYR, model no. 81356) and analyzed by a GC/MSD, as noted above. The positioning system, which includes a cathetometer, was capable of locating accurately the initial position of the probe (as well as of the burners themselves) within 25 µm, thus minimizing experimental uncertainty. 29 To measure f v , a laser-light extinction system, similar to that by Wang et al. (1996), was utilized (Figure 2.7). The system utilized a continuous Ar-ion laser beam (Lexel Laser, model 95) with a wavelength of 488 nm and energy of 10 mW, directed via a series of mirrors and focused with a lens (focal length of 150 mm) before passing through the flame region. The intensity of the light beam was measured with a photomultiplier (Thorlabs, DET100A, large area Si detector, 400 – 1100 nm, Φ=9.8 mm), whose output was processed by a digital oscilloscope (Tektronix TDS 2004B) interfaced to a computer. Two light filters (NE10A-A, NG4-A Coated, mounted absorptive ND Filter, and NE20A-A, NG9-A Coated, mounted absorptive ND Filter) were installed on the photomultiplier to prevent absorption of interfering lights and for protecting the photomultiplier. A data acquisition device (NI USB-6221) using LabView 8.6 was utilized to process the data. The lens, two of the mirrors, and the photomultiplier were mounted in a frame connected to two stepper-motors (Sherline Products, 67130 stepper motor) controlled by an automated motor controller (Anaheim Automation, model no. DPF73353). This allowed the laser beam to move both vertically and horizontally by moving each of the two mirrors separately. After beam alignment via positioning of the lens, mirrors, and of the photomultiplier, the intensity profiles (I/I 0 ) of the beam exiting the flame in the absence and/or presence of solid particles were generated by moving the laser beam both horizontally (y-axis) and vertically (z-axis). Under the fuel-lean conditions in the present study, with no siloxanes in the feed, there was little change in the intensity of the beam when passing through the flame (I/I 0 ~1), so it was assumed that intensity changes in the flames are only due to the presence of solid silica particles. Based on experimental data repeatability, the accuracy of the measured volume fractions was estimated to be 30 ±17% in a fixed alignment system; though the uncertainty is rather large, the experimental results are qualitatively very reproducible. Further details of how f v is measured can be found elsewhere (e.g., Feng et al., 2012). Figure 2.7. Arrangements of mirrors and position of photomultiplier with respect to the burners’ assembly. 2.3. Temperature Measurements 2.3.1. Flat Flame Temperature Profile Flame temperature profiles were measured. Not surprisingly, given the very low (ppm v -level) siloxane concentrations, temperature profiles in the absence and presence of siloxanes were shown to be identical, to the extent that it could be discerned via the intrusive temperature measurements. The experimental temperature profile for the opposed-jet φ=0.8 premixed flame used in this study is shown in Figure 31 2.8, along with the theoretical flame temperatures computed by using a Chemkin- based (Kee et al., 1983, 1989a) opposed-jet flame code (Egolfopoulos and Campbell, 1996; Kee et al., 1989b), coupled with the USC Mech II (Wang et al., 2007) as a chemical kinetics model for RNG. The radiation corrections were made on the experimental flame temperature data based on what was described in detail in (Shaddix, 1999, McEnally et al., 1997). The energy balance equation was used in order to correct the radiative heat loss of the thermocouple wire: ) ( 2 2 2 0 , 4 j g j j g j j T T d Nu k T − = σ ε (E2.1) where j T is the thermocouple junction temperature, g T is the gas temperature, j ε is the emissivity of the wire, σ is the Stephen-Boltzmann constant, g g g T k k = 0 , , g k is the thermal conductivity of the gas, j Nu is the Nusselt number of a cylinder, and j d is the thermocouple wire diameter. The uncertainty for the measured temperature was estimated by the uncertainty in the thermocouple emissivity of the ceramic coating, which varies from 0.3 to 0.6 (Peterson and Laurendeau, 1985). In the temperature corrections of this study the emissivity is assumed ε = 0.3. The cylindrical geometry was used instead of a spherical one because for fine wire thermocouples the heat transfer from the thermocouple wires dominates the junction energy balance (Shaddix, 1999). The expression used for the Nusselt number was obtained from (Andrews et al., 1972): 32 45 . 0 Re 65 . 0 34 . 0 + = j Nu (E2.2) where Re is the Reynolds number. Therefore, the corrected flame temperatures were calculated based on the gas properties, which were obtained by simulation. 0 200 400 600 800 1000 1200 1400 1600 1800 2000 01 2345 678 9 10 11 12 13 14 Distance from bottom burner [mm] Flame Temperature [K] Simulation Experiment Figure 2.8. Experimental vs. computed flame temperatures in the opposed-jet flame. Close agreement is found between the experimental and simulated temperatures in the luminous region. On the other hand, in the pre-flame and post-flame regions the experimental profiles are broader. Similar observations were previously reported by other investigators (e.g., Feng et al., 2010, and references therein), the differences being attributed to the experimental difficulties with the intrusive temperature measurement technique, which are well known and discussed in the literature (e.g., Cundy et al., 1986). For the kinetic data analysis the theoretical temperature profiles were, therefore, utilized, and further pertinent discussion will follow. 33 2.3.2. Bunsen Flame Temperature Profile The temperature profile on the vertical axis of the Bunsen flame was measured as well, the results, which are shown in Figure 2.9, being in agreement with those reported by Mizomoto et al. (1985). When siloxanes were added into the Bunsen flame and the temperature measurements were repeated, again no discernible changes were observed, as expected. The temperature of gas streams inside of the flame cone is higher than what is expected (room temperature) which may be justified by thermal diffusion. Lewis number for the lean methane/air flame is less than unity; as a result, the mass diffusion exceeds thermal diffusion, which implies that burning will be weakened at the tip, with T f <T ad (Mizomoto et al., 1985) As indicated in the experimental measurements, this suppression of flame temperature is observed and the maximum flame temperature occurs a little above the flame tip. 0 500 1000 1500 2000 0 10 20 30 40506070 8090 100 Distance from bottom burner [mm] Flame Temperature [K] Maximum Flame Temperature Flame Tip Figure 2.9. Experimental temperature profile on the vertical axis of a Bunsen flame. 34 2.4. Reaction Kinetics Studies of Individual Siloxanes 2.4.1. Concentration Profiles The concentration profiles of the three different linear siloxanes L 2 , L 3 , and L 4 (studied in order to assess the effect of chain length on reactivity) and the cyclic siloxane D 4 (studied in order to compare its behavior with that of the corresponding linear siloxane L 4 ) are shown in Figures 2.10-13. The concentration measurement results confirm that siloxane decomposition begins before the luminous flame region and completes, for the most part, inside the luminous region. 0.0E+00 5.0E-06 1.0E-05 1.5E-05 0.30.4 0.50.6 0.7 Distance from bottom burner [cm] L2 ppmv Simul. 14ppmv Simul. 7ppmv Simul. 2ppmv Exp. 14ppmv Exp. 7ppmv Exp. 2ppmv Luminous Region Figure 2.10. Experimental and computed L 2 concentration profiles for various L 2 feed concentrations. The concentration of siloxanes vanishes within the luminous flame region, and no siloxane survives and escapes the flame. The decomposition reaction of siloxanes is 35 known to be very fast, e.g., based on a graph in Ehrman et al. (1998), the reaction time of L 2 at 1800 K is in ms. The figure for each siloxane contains data for three different feed concentration, and for all the concentrations profiles are qualitatively identical. 0.0E+00 4.0E-06 8.0E-06 1.2E-05 0.3 0.4 0.5 0.6 0.7 Distance from bottom burner [cm] L3 ppmv Simul. 10ppmv Simul. 5ppmv Simul. 2ppmv Exp. 10ppmv Exp. 5ppmv Exp. 2ppmv Luminous Region Figure 2.11. Experimental and computed L 3 concentration profiles for various L 3 feed concentrations. 36 0.0E+00 2.0E-06 4.0E-06 6.0E-06 8.0E-06 0.3 0.4 0.5 0.6 0.7 Distance from bottom burner [cm] L4 ppmv Simul. 8ppmv Simul. 4ppmv Simul. 2ppmv Exp. 8ppmv Exp. 4ppmv Exp. 2ppmv Luminous Region Figure 2.12. Experimental and computed L 4 concentration profiles for various L 4 feed concentrations. 0.0E+00 4.0E-06 8.0E-06 1.2E-05 1.6E-05 0.3 0.4 0.5 0.6 0.7 0.8 Distance from bottom burner [cm] D4 ppmv Simul. 16ppmv Simul. 8ppmv Simul. 2ppmv Exp. 16ppmv Exp. 8ppmv Exp. 2ppmv Luminous Region Figure 2.13. Experimental and computed D 4 concentration profiles for various D 4 feed concentrations. 37 2.4.2. Conversion Profiles For the same siloxanes, the conversion profiles (defined as the fraction of the molar feed of siloxane that has reacted at that particular position) are also plotted along with calculated conversion profiles obtained from simulations and they are shown in Figures 2.14-17. As with the concentration profiles, the conversion profiles are also sharp, with 100% conversion attained for all siloxanes in the front-end of the flame region with no siloxanes escaping in the post-flame region of the flat flame. Also, note that the conversion profiles (both simulated and experimental) appear to be more or less independent of the feed siloxane concentration, suggesting a first order global decomposition reaction, as one may have expected for these dilute experimental concentrations. Comparing the conversion profiles among the various linear siloxanes, one observes very sharp profiles for L 2 but less so for L 3 and L 4 . In fact, the sharpness in the concentration profiles gradually decreases as the chain length increases, suggesting perhaps slower decomposition kinetics. When comparing the conversion profiles between L 4 and D 4 , the one corresponding to the cyclic siloxane is significantly broader. The reaction zone widths for the linear and cyclic siloxanes were measured to be ~2 mm and ~3 mm, respectively. The simulation results show an acceptable agreement with the experimental data. As one may observe in Figure 2.13, the simulation graphs also indicate a broader reaction zone for the D 4 . However, the simulation results for linear siloxanes demonstrate a very sharp drop in the reaction zone. 38 0.00 20.00 40.00 60.00 80.00 100.00 0.30.4 0.50.6 0.7 Distance from bottom burner [cm] Fraction of Unreacted L2 Simul. 14ppmv Simul. 7ppmv Simul. 2ppmv Exp. 14ppmv Exp. 7ppmv Exp. 2ppmv Luminous Figure 2.14. Experimental and computed L 2 conversion profiles for various L 2 feed concentrations. 0.00 20.00 40.00 60.00 80.00 100.00 0.30.4 0.50.6 0.7 Distance from bottom burner [cm] Fraction of Unreacted L3 Simul. 10ppmv Simul. 5ppmv Simul. 2ppmv Exp. 10ppmv Exp. 5ppmv Exp. 2ppmv Luminous Figure 2.15. Experimental and computed L 3 conversion profiles for various L 3 feed concentrations. 39 0.00 20.00 40.00 60.00 80.00 100.00 0.30.4 0.50.6 0.7 Distance from bottom burner [cm] Fraction of Unreacted L4 Simul. 8ppmv Simul. 4ppmv Simul. 2ppmv Exp. 8ppmv Exp. 4ppmv Exp. 2ppmv Luminous Region Figure 2.16. Experimental and computed L 3 conversion profiles for various L 3 feed concentrations. 0.00 20.00 40.00 60.00 80.00 100.00 0.3 0.4 0.5 0.6 0.7 0.8 Distance from bottom burner [cm] Fraction of Unreacted D4 Simul. 16ppmv Simul. 8ppmv Simul. 2ppmv Exp. 16ppmv Exp. 8ppmv Exp. 2ppmv Luminous Figure 2.17. Experimental and computed D 4 conversion profiles for various D 4 feed concentrations. 40 2.4.3. Rate Constant Parameter Fitting The experimental data in Figures 2.10-13 were fitted assuming that the global reactions for each siloxane in the flame are described as follows: C 2n H 6n O n Si n → Products n=4 (Cyclic) (R2.1) C (2n+2) H (6n+6) O (n-1) Si n → Products n=2,3,4 (Linear) (R2.2) To accomplish this task, the concentration profiles were computed using the opposed-jet flame code. The Lennard-Jones parameters of siloxanes were estimated by the method suggested by Wang and Frenklach (1994). The global decomposition reactions R2.1 and R2.2 above, assumed to be first order with respect to the siloxane concentration, were added to the detailed decomposition mechanism of RNG, given that for lean flames and siloxane concentrations that are very low, any dependence of the rate on the oxygen concentration is assumed to be “lumped” into the pseudo-rate constant. In order to calculate the pre-exponential factor A (1/s) and the activation energy E a (kcal/mol), an extensive search was carried out in the parameter space for the set of A and E a for each siloxane that minimizes the sum of squared differences between the simulated and experimental values. The calculated rate parameters for the four different siloxanes are shown in Table 2.3, and the simulation results are shown as lines in Figures 2.10-13. There is qualitative agreement between experiments and simulations, and the trends appear reasonable. However, the uncertainties, e.g., 95% confidence limits, in the values of the calculated parameters are rather large, not surprising given the many simplifying assumptions that have gone into the simulations. As shown in Table 2.3, E a values for the linear siloxanes increase with 41 increasing chain length. E a for the cyclic siloxane (D 4 ) is larger than that for the linear one (L 4 ), suggesting that the cyclic siloxanes are more difficult to decompose than their linear counterparts with the same number of silicon atoms in the chemical structure. Siloxane Pre-exponential Factor [1/s] Activation Energy [kcal/mol] L 2 3.8±0.34 × 10 12 47±0.7 L 3 2.4±0.29 × 10 12 55±1.1 L 4 9.8±1.45 × 10 11 61±1.3 D 4 1.6±0.14 × 10 12 70±1.0 Table 2.3. Pre-exponential factors (A) and activation energies (E) for different siloxanes (Jalali et al., 2013). To the author’s knowledge, there are a few previous studies reporting rate constants for siloxanes. Ehrman et al. (1998) and Zachariah (1997, unpublished data for decomposition of HMDS, using a method following Sanogo and Zachariah, 1997) reported a rate expression for the decomposition for L 2 , and Sanogo and Zachariah (2007) for D 4 . Their experiments were carried in a 1 m long flow reactor in the temperature range of 1073-1373 K for L 2 and 1058-1197 K for D 4 . When calculating the specific reaction rate for L 2 at 1373 K, using the rate expression reported in Ehrman et al. (1998), its value is 3.3×10 3 s -1 . The corresponding value from the present study is 1.1×10 5 s -1 , approximately 33 times higher. When calculating the rate constant for D 4 at 1197 K, using the rate expression reported in Sanogo and Zachariah (2007), the value is 1.0×10 2 s -1 . The corresponding value from the present study is 42 2.4×10 -1 s -1 which is approximately two orders of magnitude lower. Given the differences in the temperature range between the two studies and the experimental uncertainties (e.g., temperature profiles, the plug-flow assumption to describe the flow reactor, the positional uncertainty in sampling a 2 mm flame, etc.), the extent of disagreements between the different studies is not considered to be excessive. 2.5. Solid Particle Volume Fraction Measurements Using the Laser Extinction Technique 2.5.1. Theory and Background The following equation relates the solid particles volume fraction to the various experimental parameters (Wang et al., 1996): 2 2 1 6Im( ) 2 ext v k f m m λ π = − + (E2.3) where k ext is the extinction coefficient, λ is the wavelength of the incident laser beam, and m is the complex refractive index of the solid silica particles. The real part of the complex refractive index of silica for the 488 nm wavelength is 1.4631 (Tan and Arndt, 2000). The imaginary part of this complex number is 1.0×10 -4 based on a graph reported in Touloukian and DeWitt (1972). Therefore, the complex refractive index used for the volume fraction calculations is m= 1.4631+1.0×10 -4 i. The local extinction coefficient, k ext (r) can be obtained from the following equation using the Abel’s inversion formula: 43 0 22 ln[ ( ) / ] 2 ( ) ext y r Iy I k r dr ry ∞ −= − ∫ (E2.4) where I is the intensity of the laser beam which is detected by the photomultiplier and r is the radius from the center of the flat flame. 2.5.2. Solid Particle Volume Fraction Measurements for Single Compounds 2.5.2.1. Solid Particle Volume Fraction Profile In Figure 2.18, the f v profile in the flame for one of the siloxanes, D 4 , is shown for 8 and 16 ppm v feed concentrations. Shown on the same figure is the simulated fraction of siloxane in the feed that has reacted at a particular position. It is noted that, because of the assumption of a first order global reaction, the simulated conversion profiles are independent of the feed concentration, which is validated also by the data shown in Figure 2.18. Furthermore, the f v profile peaks in the flame, which is a result of two competing effects. On one hand, as the siloxane molecules approach the flame zone, their decomposition intensifies, resulting in greater particle production rates, which tends to increase f v . On the other hand, as the temperature increases within the flame, the gas density keeps decreasing due to dilatation, which tends to decrease f v , given that the particle number density follows closely the variations of the gas phase density for small particles that follow, in principle, the gas phase flow (e.g., Egolfopoulos and Campbell, 1999). 44 0 20 40 60 80 100 01 23 4 5 6 7 Distance from bottom burner [mm] Fraction of reacted D4 0.0E+00 1.0E-08 2.0E-08 3.0E-08 4.0E-08 5.0E-08 Solid particle volume fraction Exp. 16 ppmv Exp. 8 ppmv Simul. 16 ppmv Simul. 8 ppmv Vol. Frac. 16 ppmv Vol. Frac. 8 ppmv Fraction of reacted D4 - Simulation results for 16 and 8ppmv Fraction of reacted D4 - Experimental results for 16 and 8ppmv Solid particle vol. frac. for 16 and 8ppmv Figure 2.18. D 4 conversion and solid particle volume fraction profiles. 2.5.2.2. Maximum Solid Particle Volume Fraction Figure 2.19 depicts the maximum f v values measured in the RNG flame for various siloxanes as a function of their concentrations in the RNG/air feed mixture. The results exhibit a fairly linear relationship between f v and the feed siloxane concentration in the fuel/air mixture. This is expected, given that the siloxanes decompose completely in the flame, as was shown in Figures 2.10-13. As expected also, among the linear siloxanes, f v increases with increasing number of silicon atoms in the parent molecule, and linear (L 4 ) and cyclic (D 4 ) siloxanes with the same number of silica atoms in their structure result in almost the same f v . 45 y = 1E-09x + 4E-09 R 2 = 0.9821 y = 3E-09x - 4E-09 R 2 = 0.9892 y = 3E-09x + 5E-10 R 2 = 0.999 y = 3E-09x + 6E-10 R 2 = 0.9794 0.0E+00 1.0E-08 2.0E-08 3.0E-08 4.0E-08 5.0E-08 6.0E-08 7.0E-08 0 5 10 15 20 25 30 ppmv Maximum solid particle volume fraction L2 L3 L4 D4 Linear (L2) Linear (L3) Linear (L4) Linear (D4) Figure 2.19. Maximum solid particle volume fractions for different siloxanes. 2.6. Surface Analyses Results 2.6.1. Introduction As discussed previously, one important concern about the oxidation of Si- containing fuels is the production of the solid particles within the flame, and their deposition on various surfaces in close proximity with the flame. The presence of silica particles in the RNG flame is often visible in the experiments with the naked eye, e.g., in the Bunsen flames shown in Figure 2.20 in the form of an orange halo surrounding the flame resulting from the black-body radiation of SiO 2 particles at high temperatures (Ma et al., 2003; Yeh et al., 2001b). 46 a b Figure 2.20. Renewable natural gas Bunsen flame (a) with (16 ppm v of D 4 ) and (b) without siloxanes. To probe the nature of deposits formed on surfaces in proximity to the RNG flames, flat Ni/Cr metal strips (5×12 mm, 0.5 mm thick, the exact chemical composition, as reported by the manufacturer is shown in Table 2.4) were placed downstream of a Bunsen flame. The chemical composition and structure of the solid particles were investigated by scanning electron microscopy (SEM) and energy dispersive analysis by X-ray (EDAX). Fe Cr Mo Cu Co W Ni Ti C Mn Si Al Other 17.0- 20.0 20.5- 23.0 8.0- 10.0 <0.5 0.5- 2.5 0.2- 1.0 40.6- 50.5 <0.15 0.05- 0.15 <1.0 <1.0 <0.5 <0.08 Table 2.4. Chemical composition (wt.%) of the Ni/Cr alloy. 2.6.2. Sample Preparation As noted above, in order to study the deposits formed on surfaces in proximity to the siloxane containing RNG flames, flat Ni/Cr metal strips were placed downstream of a Bunsen flame. Prior to their use in the experiments, their surfaces were polished 47 by a polisher (Buehler, Grinder/Polisher EcoMet ® 3) using a slurry solution of 0.3 µm Al 2 O 3 powder, followed by washing in an ultrasonic cleaner first with acetone (15 min), methanol (15 min), and then with deionized water (15 min). In the experiments, the metal strips were placed 10 cm above the exit of the bottom burner tube in a Bunsen-type flame of simulated RNG containing 16 ppm v of D 4 (based on the total air/RNG mixture), φ = 0.8, burner exit velocity of 92 cm/s, and a mixture flow rate of 141.8 cc/s (STP). SEM (JEOL JSM-6610 LV, Software: Scanning Electron Microscope Interface v2.00) and EDAX (JSM 6490, Software: EDAX Genesis v6.2) were utilized to probe the nature of the deposits. 2.6.3. Results and Discussion In the preliminary experiments in which metal strips were exposed to the RNG combustion environment to study the nature of the particulates formed, four different types of particles were identified via SEM/EDAX, namely pure carbon (soot), silicon oxide (SiO x ), silicon oxicarbide (SiO x C y ), and silicon carbide (SiC z ). While the formation of silica (silicon oxide)-type particles was expected, the reasons for the presence of the other two types of particles in a lean RNG/air flame were not entirely clear. Additional experiments demonstrated that the number of carbon-containing particles would significantly diminish if the metal surface was exposed to a pure RNG flame for a certain period of time prior to the siloxane impurity being added to the flame. Subsequently, it was found that by placing the metal strip for 15 min at a position in the post-flame region, where the steady state temperature of the plate surface was about 730 o C, eliminated totally the presence of any pure carbon or carbon-containing particles. The temperature of the hot flue gases hitting the substrate 48 surface at 10 cm above of the bottom burner is 750 °C, and temperature of the back surface is nearly 480 – 500 °C. In all further experiments, therefore, the metal surfaces would be subjected to this pretreatment prior to the siloxanes being added to the RNG flame. At short exposure times, changes on the surface are not visible to the naked eye. However, under the electron microscope, one can observe a few white particles, as Figure 2.21 indicates; under this magnification, one could count 3 – 5 particles/mm 2 . Focusing the EDAX beam on one of these particles indicates that it consists of purely silicon and oxygen, while the other elements are due to the penetration of the beam through the particle to reach the metal plate underneath. That the metal strip surface does not appear visibly coated by silica does not mean that it is not present there, however. In fact, EDAX indicates the presence of small amounts of silicon being present on the surface, apart from the occasional large particle, right from the start. Figure 2.21. A silica particle deposited on the Ni/Cr plate after 15 s of exposure time and its EDAX spectrum. In Table 2.5, the elemental composition of the surface as a function of exposure time is shown. The increasing content of the silicon and the decreasing content of nickel is a tell-tale sign of a surface that is being coated progressively with silica 49 deposits. In the beginning, that is not obvious with the naked eye. For example, after 2 min of exposure, the metal surface has not changed much, but under the electron microscope and under the same magnification with that of Figure 2.21, one can observe 200 – 300 particles/mm 2 . Element Wt% Exposure Time, min C O Ni Al Si Mo Cr Fe 2 0.00 8.77 47.42 0.27 1.64 8.63 17.76 15.51 3 0.00 10.07 45.89 0.00 3.33 8.83 17.17 14.71 4 0.00 14.43 41.67 0.00 5.72 7.97 16.42 13.87 7 0.00 18.79 36.05 0.00 11.33 8.06 13.57 12.20 10 0.00 23.73 26.22 0.00 19.16 6.96 12.88 11.06 20 0.00 37.56 8.06 0.00 36.79 5.97 6.32 5.31 Table 2.5. Elemental analysis of Ni/Cr surface for various exposure times to the siloxane/RNG flame. Eventually the surface becomes completely coated with a white layer, and the presence of the silica layer becomes obvious to the naked eye and also when viewed with electron microscopy, as Figure 2.22 indicates clearly. The colors of the silicon dioxide and silicon-monoxide layers are reported to be white and brown, respectively (Yeh et al., 2001a). However, in all experiments carried out in this study, only a white layer was observed on the surface. 50 Figure 2.22. SEM image of the surface after 20 min of exposure. Despite the fact that it is not obvious to the naked eye, exposure to siloxanes has a major impact on the metal surfaces right from the start. Figure 2.23, for example, shows AFM data with a Ni/Cr polished surface, after only 1 s of total exposure time to the RNG/siloxane mixture. Significant changes, specifically an increase in the surface roughness by a factor of 3, are clearly visible; note that this surface was not previously subjected to the 15 min pretreatment by exposing it to the RNG/air-only mixture. 51 a b Figure 2.23. AFM images of the metal surface (a) before deposition, average roughness 3.66 nm, and (b) after deposition, average roughness 10.64 nm. 2.7. Decomposition of Siloxanes’ Mixtures 2.7.1. Introduction 52 As it is mentioned in the introductory section, siloxanes are an important class of ppm-level impurities in LFG. Most of the biogases extracted from landfills contain a mixture of different types of siloxanes and it is very rare to extract biogas having only a single siloxane. The kinetic studies and other experimental investigations, so far, in this research were carried out on a flat or Bunsen flame containing a single type of siloxane. Therefore, it is very crucial to examine the possible effect of other siloxanes on each others’ decomposition kinetics while they are present in RNG as a mixture. In such studies the laser extinction technique was applied in order to measure the solid particle volume fraction of a RNG flame containing different mixtures of various siloxanes. Finally, surface analyses by SEM/EDAX techniques were utilized to probe the nature of the deposited particles and find the possible differences (chemical composition, size, population, shape, etc) with the solid particles generated in the flames using single siloxanes. 2.7.2. Concentration Profiles The concentration profiles of each siloxane in two different mixtures were experimentally determined and the results were compared with numerical calculations using the kinetic parameters obtained for each siloxane. The first one is a mixture containing D 4 and L 4 (8 ppm v each) in order to study the interaction between the siloxanes having four silicon atoms in their main chain; but with completely different chemical structures. As it is depicted in Figure 2.24, the experimental data agree well with the simulation results. Both siloxanes decompose completely in a narrow flame region. Also, D 4 degrades in a wider region than L 4 does suggesting it is harder to decompose, as also noted previously. 53 0.0E+00 2.0E-06 4.0E-06 6.0E-06 8.0E-06 0.3 0.4 0.5 0.6 0.7 0.8 Distance from bottom burner [cm] ppmv D4 Simulation L4 Simulation D4 Experiment L4 Experiment Luminous Region Figure 2.24. Simulation and experimental concentration profile results for D 4 +L 4 mixture in the pre-flame, flame, and post-flame regions. The experimental and calculation results of D 4 and L 4 decomposition while introduced in the RNG flame individually or in a mixture are demonstrated in Figures 2.25 and 2.26. As one may notice in these figures, there is no difference between oxidation of D 4 or L 4 while they decompose in RNG flame individually or in a mixture with the other siloxane. This is expected, and is attributed to the very low concentration of siloxanes in the flame. Each siloxane molecule has a very low chance to interact with other siloxane molecules, and practically, they cannot have an impact on each other’s decomposition kinetics. 54 0.0E+00 2.0E-06 4.0E-06 6.0E-06 8.0E-06 0.3 0.4 0.5 0.6 0.7 0.8 Distance from bottom burner [cm] D4 ppmv Simulation (Individual) Silumation (Mixture) Experiment (Individual) Experiment (Mixture) Luminous Region Figure 2.25. Comparison between simulation and experimental concentration profile results for D 4 while decomposing individually or in the mixture with L 4 . 0.0E+00 2.0E-06 4.0E-06 6.0E-06 8.0E-06 0.3 0.4 0.5 0.6 0.7 Distance from bottom burner [cm] L4 ppmv Simulation (Individual) Simulation (Mixture) Experiment (Individual) Experiment (Mixture) Luminous Region Figure 2.26. Comparison between simulation and experimental concentration profile results for L 4 while decomposing individually or in the mixture with D 4 . 55 The same experiments were carried out on another mixture containing three linear siloxanes (L 2 , L 3 , and L 4 , 7 ppm v each) in order to study the interaction between siloxanes with different number of silicon atoms in their main chain but identical chemical structure. The experimental and simulation results of each siloxane’s oxidation are depicted in Figure 2.27 and a satisfactory agreement is observed between experimental and calculation results for each siloxane. The simulation results are calculated using the global kinetic constant obtained for each siloxane. All of them, also, decompose completely in the narrow flame region. 0.0E+00 2.0E-06 4.0E-06 6.0E-06 8.0E-06 0.4 0.45 0.5 0.55 0.6 Distance from bottom burner [cm] ppmv L2 Simulation L3 Simulation L4 Simulation L2 Experiment L3 Experiment L4 Experiment Luminous Region Figure 2.27. Simulation and experimental concentration profile results for the L 2 +L 3 +L 4 mixture in the pre-flame, flame, and post-flame regions. 56 Also, the oxidation of each siloxane in the mixture was compared with the decomposition of that siloxane while decomposing individually. These results are demonstrated in Figures 2.28-30. As one may notice in these figures, there is no difference between the profiles of siloxanes while they are decomposing individually or in the mixture with other siloxanes. 0.0E+00 2.0E-06 4.0E-06 6.0E-06 8.0E-06 0.4 0.45 0.5 0.55 0.6 Distance from bottom burner [cm] L2 ppmv Simulation (Individual) Silumation (Mixture) Experiment (Individual) Experiment (Mixture) Luminous Region Figure 2.28. Comparison between simulation and experimental concentration profile results for L 2 while decomposing individually or in the mixture with L 3 and L 4 . 57 0.0E+00 2.0E-06 4.0E-06 6.0E-06 8.0E-06 0.4 0.45 0.5 0.55 0.6 Distance from bottom burner [cm] L3 ppmv Simulation (Individual) Silumation (Mixture) Experiment (Individual) Experiment (Mixture) Luminous Region Figure 2.29. Comparison between simulation and experimental concentration profile results for L 3 while decomposing individually or in the mixture with L 2 and L 4 . 0.0E+00 2.0E-06 4.0E-06 6.0E-06 8.0E-06 0.4 0.45 0.5 0.55 0.6 Distance from bottom burner [cm] L4 ppmv Simulation (Individual) Silumation (Mixture) Experiment (Individual) Experiment (Mixture) Luminous Region Figure 2.30. Comparison between simulation and experimental concentration profile results for L 4 while decomposing individually or in the mixture with L 2 and L 3 . 58 2.7.3. Solid Particle Volume Fraction Measurements In order to investigate the possible effect of presence of different siloxanes in a mixture on decomposition processes and the number of produced particles, the laser extinction technique was applied to measure f v . Three different mixtures were made (L 2 +L 3 , L 2 +L 4 , L 2 +D 4 ) and the f v of each mixture was measured separately in a RNG flame with an equivalence ratio of 0.8. In binary mixtures, the concentration of each siloxane is half of the total concentration of the mixture. As one may notice in Figures 2.31-33, the relationship between f v and siloxane concentration in either individual or mixed siloxanes is still linear. As it is demonstrated in these figures, the f v line for the mixtures falls in between the individual siloxanes in the mixture indicating that the number of silicon atoms in the total feed concentration plays a key role in the magnitude of solid particle volume fraction. No specific effect on the decomposition process as a result of mixing was detected by the laser extinction technique. y = 1E-09x + 4E-09 R 2 = 0.9821 y = 2E-09x - 1E-09 R 2 = 0.9902 y = 3E-09x - 4E-09 R 2 = 0.9892 0.0E+00 1.0E-08 2.0E-08 3.0E-08 4.0E-08 5.0E-08 6.0E-08 7.0E-08 0 5 10 15 20 25 30 ppmv Maximum solid particle volume fraction L2 L2+L3 L3 Linear (L2) Linear (L2+L3) Linear (L3) 59 Figure 2.31. Maximum solid particle volume fractions for mixture of L 2 and L 3 . y = 1E-09x + 4E-09 R 2 = 0.9821 y = 2E-09x + 2E-09 R 2 = 0.995 y = 3E-09x + 5E-10 R 2 = 0.999 0.0E+00 1.0E-08 2.0E-08 3.0E-08 4.0E-08 5.0E-08 0 5 10 15 20 25 30 ppmv Maximum solid particle volume fraction L2 L2+L4 L4 Linear (L2) Linear (L2+L4) Linear (L4) Figure 2.32. Maximum solid particle volume fractions for mixture of L 2 and L 4 . y = 1E-09x + 4E-09 R 2 = 0.9821 y = 2E-09x + 2E-09 R 2 = 0.9898 y = 3E-09x + 6E-10 R 2 = 0.9794 0.0E+00 1.0E-08 2.0E-08 3.0E-08 4.0E-08 5.0E-08 0 5 10 15 20 25 30 ppmv Maximum solid particle volume fraction L2 L2+D4 D4 Linear (L2) Linear (L2+D4) Linear (D4) Figure 2.33. Maximum solid particle volume fractions for mixture of L 2 and D 4 . 60 2.7.4. SEM/EDAX Analyses Results: Particle Deposition Using a Simulated RNG Containing 16 ppm v of D 4 +L 4 In this experiment, a RNG Bunsen flame containing 16 ppm v of the L 4 +D 4 mixture was used and the produced solid particles were deposited on the substrate surface for 2 min. L 4 has the same number of silicon atoms in its chemical structure with D 4 , and the major difference between D 4 and L 4 is the number of carbon atoms. As Figure 2.34 indicates, no specific difference was discerned in deposited particles while using 16 ppm v of pure D 4 or using a D 4 and L 4 mixture. The population of the particles was the same (200 – 300 particles/mm 2 ). Also, the size and morphology of particles are almost same. 61 Figure 2.34. An SEM image of the sample's surface deposited by a RNG Bunsen flame containing 16 ppm v of D 4 +L 4 . The weight percentages of various detected elements over a large surface area of the substrate in Figure 2.35 indicates that there is no tangible difference between using pure D 4 or using a mixture of D 4 and L 4 in the RNG flame. 62 Elements C O Ni Al Si Mo Cr Fe Weight % 0.00 8.23 47.08 0.27 1.68 8.74 18.11 15.89 Figure 2.35. Compositional analysis of the SEM image shown in Figure 2.34. Figure 2.36 shows an SEM image of a corner of the sample. This image provides a better insight of how particles are distributed on the sample. The same population and distribution was observed as in the case of using pure D 4 . Figure 2.36. An SEM image of sample's surface using D 4 +L 4 . 63 2.7.5. Conclusions Since mostly siloxanes exist in biogas in mixtures, it is necessary to gain knowledge about oxidation of siloxanes in the mixture with other siloxanes. Therefore, different experimental techniques are utilized in order to probe the potential change in decomposition processes of siloxanes and produced particles while they are present in RNG flames in a mixture with other siloxanes. No discernable change was detected by mixing of a siloxane with others. Since the concentrations of siloxanes are very low in RNG, each different siloxane molecule has a small chance to interact with other siloxanes and to change their decomposition process. 2.8. Solid State Studies of Siloxane Decomposition Products An important component of a study of particulate formation from siloxane decomposition, is a better understanding of the thermo-physical properties of key solid oxide species potentially generated during the organo-silicon compounds decomposition. As part of this research, an extensive literature review was carried out, therefore, about the phase diagrams and vapor pressure correlations of silicon, silicon monoxide (SiO), silicon dioxide (SiO 2 ), silicon carbide (SiC), and SiCO. Some of the key studies are summarized here. In a study by Brewer and Mastick (1951), it is reported that SiO 2 appears to vaporize by decomposition, but elemental silicon is not the main species, instead SiO(g) and O 2 appear to be the main gaseous species of its vaporization. The existence of SiO(g) as a stable gaseous species indicates its 64 volatility under conditions where neither Si(s) nor SiO 2 (s) have any appreciable vapor pressure. In another probe, a research was conducted on the amorphous solid SiO, and it was demonstrated by Fourier Transform analysis of the scattered intensity data from the material described as “amorphous SiO”, that the material is, in fact, a stoichiometric mixture of SiO 2 and Si. It is also confirmed that the commercially available solid SiO is a composite of Si and SiO 2 (Brady, 1959; Friede and Jansen, 1996). Brewer and Edwards (1954) carried out a fundamental study of the thermo- physical properties of Si, SiO, and SiO 2 at various temperatures. They reported that the metastable SiO solid, prepared by the quenching of SiO gas, begins to disproportionate to Si and SiO 2 at an appreciable rate at ~400-700 K. Solid SiO appears to be thermodynamically unstable at all temperatures below 1450 K. Experimental evidence presented in this study indicated that at a higher temperature solid SiO becomes stable, with a melting point higher than 1975 K. Gulbransen et al. (1966) suggested the main four elementary reactions which occur in the silicon- oxygen system, and subsequently measured and reported their corresponding equilibrium constants at different temperatures (See Table 2.6). According to this study, both SiO(g) and SiO 2 (s) can form. SiO 2 (s) does not appreciably dissociate to SiO(g) in an oxygen atmosphere of 9 ×10 -3 Torr at temperatures below the melting point of silicon (m p =1420 ºC). The solid phase reaction of Si(s) with SiO 2 (s) can proceed in an atmosphere of low-pressure SiO(g), and the reverse reaction in a high- pressure atmosphere of SiO(g). Silicon is especially sensitive to reactions with gases containing carbon under conditions where a SiO 2 (s) film is absent. 65 Equilibrium Constant (logKp) Reactions 927ºC 1127ºC 1327ºC 1527ºC Si(s) + 1/2O 2 (g) → SiO(g) 8.897 8.222 7.707 7.201 Si(s) + O 2 (g) → SiO 2 (s) 30.26 24.65 20.45 17.10 1/2Si(s) + 1/2SiO 2 (s) → SiO(g) -6.333 -4.106 -2.520 -1.347 SiO 2 (s) → SiO(g) + 1/2O 2 (g) -21.37 -16.43 -12.75 -9.90 Table 2.6. Thermochemical analysis of reactions in the silicon-oxygen system. Tomooka et al. (1999) extensively reviewed the past scientific investigations on the temperature dependence of the vapor pressure of Si(g) over solid and liquid silicon, and results are summarized in Figure 2.37. Figure 2.37. Temperature dependence of the vapor pressure of Si(g) over Si(s,l). 66 ■: Solid (Schmude, 1994), □: Liquid (Schmude, 1994), ●: (Drowart et al., 1958), — —: Solid, Calculation (Barin, 1995), ------: Liquid, Calculation (Barin, 1995), ○: (Souchiere and Binh, 1986), ▲: (Gulbransen et al., 1966), …..: Calculation (Honig, 1954; Honig, 1957), ∆: (Batdorf and Smits, 1959). Tomooka et al. (1999) noted that the previous data were scattered, and not all that consistent. Therefore, they measured the vapor pressure of Si(g) over Si(l) (with a time of flight mass spectrometer equipped with a boron nitride Kundsen Cell) in order to obtain more reliable data. The partial vapor pressures of Si(g) over Si(l) are shown as a function of temperature in Figure 2.38 (where they are also compared with previous data). 67 Figure 2.38. Temperature dependence of vapor pressure of Si(g) over Si(l). ■: (Tomooka et al., 1999), : (Schmude, 1994), . . : (Drowart et al., 1958), …..: (Barin, 1995). The Si(g) vapor pressure data in Figure 2.38 are described by the following equation: log(P Si /Pa) = (-2.08±0.10) × 10 4 K/T + (10.84±0.53) (E2.5) A comprehensive investigation was conducted on the vapor pressure of SiO(g) (Ferguson and Nuth, 2008), which is produced by either heating amorphous silicon monoxide: 68 SiO(amorphous) → SiO(g) (R2.3) or mixing silicon with silica: 1/2Si(cristobalite) + 1/2SiO 2 (s) → SiO(g) (R2.4) The study reports that the available experimental data for SiO evaporation (shown in Figure 2.39) are described by the following equation: K T Pa p / 550 17740 ) 39 . 0 29 . 13 ( ) / ( log 10 ± − ± = (E2.6) 69 Figure 2.39. Comparison of available experimental data for SiO (am) evaporation. The symbols shown in Figure 2.39 are reported experimental data in circles (Ferguson and Nuth, 2008), triangles (Geld and Kochnev, 1948), squares (Gunther, 1958), and thick dashed lines (Rocabois et al., 1992). A fit to the experimental data by Ferguson and Nuth (2008) is given by the solid line. In the same study in a wider range of temperatures, Figure 2.40 represents a comparison of the current experimental data for SiO evaporation with thermodynamic modeling of the SiO vapor pressure. Experimental data are reported in circles (Ferguson and Nuth, 2008), triangles (Geld and Kochnev, 1948), squares (Gunther, 1958), and thick solid line (Rocabois et al., 1992). Thermodynamic assessments of SiO vapor pressure are shown 70 in solid line and short dashed line (Schick, 1960), dotted line (Kubaschewski and Chart, 1974), and dashed-dotted line (Schnurre et al., 2004). Figure 2.40. A comparison of the current experimental data for SiO (am) evaporation with thermodynamic assessment of SiO vapor pressure. These data are described by the following equation: 42 . 17 ) / ( log 77 . 2 / 17750 ) / ( log 10 10 + − − = K T K T Pa p (E2.7) 71 A comprehensive review on the thermodynamic analysis of the high-temperature vaporization properties of SiO 2 was conducted by Schick (1960). It is reported that SiO 2 may be vaporized both under oxidizing and non-oxidizing conditions (the former conditions are more relevant for the current study of siloxane decomposition in fuel- lean RNG flames). The reported vapor pressure data at 1.0 atm is shown in Figure 2.41. Figure 2.41. Vapor pressure of silica at 1.0 atm obtained by decomposition in air. The following two reactions were considered: SiO 2 (s) → SiO(g) + 1/2O 2 (g) (R2.5) 72 1/2O 2 (g) → O(g) (R2.6) Since the decomposition temperature is very high, the dissociation of oxygen is considered as well. For temperatures below 2000 K, the relation for the vapor pressure, P SiO , must be modified. This modification is shown in Figure 2.42. Figure 2.42. Logarithmic plot of partial vapor pressure of SiO(g) over extended temperature range resulting from solid silica dissociation. As a result, the data depicted in Figure 2.42 are described by the following equation: ) / 3075 1 ( 72 . 19 T SiO e p − = (E2.8) 73 2.9. Concluding Remarks In summary, biogas and LFG are poised to make inroads in energy and electricity generation via their planned addition as RNG into the NG pipeline system, but their use is hampered by the fact that they contain a variety of trace contaminants, including siloxanes, which are a particularly troublesome class of NMOC. Since current techniques to remove the siloxanes are rather inefficient (Ajhar et al., 2010), and since reliable field analytical equipment for detection is lacking, concerns exist with them finding, inadvertently, their way into the NG pipeline system and thus into people’s homes and in NG equipment and appliances. Based on the aforementioned considerations, the focus of the present study was the characterization of the fate of siloxanes during combustion of RNG under well-controlled laboratory conditions. Siloxane concentration profiles were measured in the opposed-jet flame configuration and it was shown that siloxanes decompose for the most part in the pre-flame and the luminous region with little, if any, escaping in the post-flame region. First-order (with respect to siloxane concentration) global decomposition reactions were assumed, and appear to provide a qualitative fit of the experimental data. The global rate constants were calculated and indicated decomposition rates that become slower with increasing siloxane chain length, and for the cyclic siloxanes when compared to their linear counterparts of the same chain length. Solid particle volume fractions were measured using the laser extinction method, which was shown to be a very reliable technique to be used with such particles. As expected, the results indicate a linear relationship between the siloxane concentration in the fuel and the corresponding particle volume in the flue-gas. The particle volume fractions were found to increase also with increasing number of silicon atoms in the siloxane molecule. The particle volume 74 fractions were found, in addition, to be comparable between linear and cyclic siloxanes with the same number of silicon atoms in the molecule, which is expected based on the fact that these siloxanes completely decompose in the flame environment. Analysis of the metal surfaces exposed to the siloxane-containing RNG mixtures indicates a substantial impact. The surface roughness increases right from the start, with the surface being coated eventually by a white layer of SiO 2 . 75 Chapter 3 – A Study on Silane Decomposition during the Combustion of Renewable Natural Gas 3.1. Introduction Biogas is produced from anaerobic biodegradation of sludge in Waste Water Treatment Plants (WWTP) and of solid waste in Landfills (the gas is then known as landfill gas or LFG). It is considered to be a promising renewable energy source as it contains large amounts of methane (typically, 40-70 vol%) the rest being mostly CO 2 , but also gases like O 2 , N 2 , and Ar in smaller amounts (El-Fadel et al., 1997; Abatzoglou and Boivin, 2009; Shin et al., 2002). Biogas is potentially a very promising fuel for energy and power generation (Ohannessian et al., 2008; McBean, 2008; Murphy and McKeogh, 2006); however, biogas and LFG contain more than 140 adverse trace contaminants (Schweigkofler and Niessner, 2001; Boulinguiez and Le Cloirec, 2010; Eklund et al., 1998) including siloxanes, a particularly problematic trace constituent (Oshita et al., 2010; Dewil et al., 2006). Siloxanes are oxidized and produce SiO 2 micro-particulates (e.g., Dewil et al., 2006) during biogas combustion, leading to the need for more frequent maintenance of electricity generators and reducing their operating life (see, e.g., Popat and Deshusses, 2008; Ajhar et al., 2010; Badjagbo et al., 2010). Another key problem with these fine silica micro-particles that 76 form, is that unless adequate precautions are taken, they will most likely escape into the atmosphere, where they may pose a risk to both human health and the environment, and may cause serious air pollution problems. Because of the challenges siloxanes present to the beneficial use of biogas and LFG, they have attracted attention, particularly in recent years. There are numerous literature studies, for example, on the use of different physical, chemical, and biological techniques for removing siloxanes from biogas and LFG (e.g., see Wheless and Gary, 2002; Ricaurte Ortega and Subrenat, 2009; Matsui and Imamura, 2010; Montanari et al., 2010). Physical absorption and adsorption are most commonly utilized, but these techniques are not at all effective in removing siloxanes, because none of the common adsorption/absorption media are uniquely selective towards the siloxanes in the presence of the other numerous contaminants found in biogas. Today, biogas and LFG in the form of renewable natural gas or RNG (which is biogas after its trace contaminants have been removed and its methane content has been upgraded to meet natural gas (NG) pipeline standards) is finding increased use in California. The potential of siloxanes finding inadvertently their way into the NG supply system, makes even more clear the need to fundamentally understand how the silica micro- particulates form during combustion from siloxanes and how they deposit on various surfaces in contact with the combustion mixtures. Very few investigations have been carried out to date on the siloxane decomposition kinetics; and the few studies that are available, so far, focus mostly on the global decomposition reactions and their derivation of kinetic constants (Ehrman et al., 1998; Zachariah, 1997 (unpublished data for decomposition of Hexamethyldisiloxane, following the method of Sanogo and Zachariah, 1997); 77 Sanogo and Zachariah, 2007; Jalali et al., 2013). However, no rigorous siloxane oxidation mechanism has been reported to date, even for the simplest linear siloxane, hexamethyldisiloxane (L 2 ). For L 2 , but also for all the other larger size siloxane molecules found in biogas and LFG (Jalali et al., 2013), there is still complete lack of information regarding the major decomposition channels leading to SiO and eventually SiO 2 formation. Significant more attention has been focused, however, on silane (SiH 4 ) as a model silicon-containing precursor to produce silica. Several investigations, using different experimental techniques, have been carried out in order to qualitatively and quantitatively understand how silane decomposes and what mechanistic channels lead into SiO and SiO 2 production (see further discussion below). Moreover, silane has received significant attention in the nano-powder industry where flame aerosol technology has been shown to be an extremely useful technique for manufacturing commercial quantities of nano-particles from precursor gases (like silane) in flames (Pratsinis, 1998). Gas-phase combustion synthesis of nano-powders is thought to offer significant advantages over other material synthesis processes, because of its apparent simplicity, as a one-step process, and the fact that it uses machinery with no-moving parts (Kammler et al., 2001). A number of silane decomposition models have been proposed based on various experimental methods, including the measurement of fundamental combustion properties such, as for example, ignition delay times. Some of the early work was done by Newman et al. (1979), Neudorfl et al. (1980), and Jachimowski and McLain (1983) in order to develop a chemical kinetics mechanism for the decomposition of silane/hydrogen mixtures. The models were further improved and more advanced 78 mechanisms were proposed by Coltrin et al. (1984 and 1986). A common idea, among all these early studies, was that the first key step in silane decomposition is a hydrogen abstraction from the silane molecule to form SiH 2 . In a subsequent experimental study by Chung et al. (1985), a counter-flow burner was utilized to produce a flat H 2 /O 2 /Ar diffusion flame to which silane was added to generate silica nano-particles. During the experiments, the temperature as well as the partial pressures of silicon monoxide and oxygen were measured. Assuming that chemical equilibrium prevails in the vapor phase, Chung et al. (1985) calculated the species partial pressures and the concentration of silica particles using the phase diagrams of Schick (1960). In another study carried out by Koda and Fujiwara (1988) the combustion of silane in an opposed jet diffusion flame was investigated and light scattering measurements indicated that the nucleation of solid silica particles intensifies when the silane flow rate is increased beyond a critical value. The existence of SiO, OH, and SiH gas-phase species was confirmed in this study via emission spectroscopy. The experimental study of Zachariah et al. (1989) indicated that temperature plays a significant role in the silica formation process, by controlling the chemical kinetic rates and through physical changes in particle morphology. Increasing flame temperature enhanced chemical kinetics and the homogenous nucleation, leading to higher numbers of smaller particles which were more spherical in shape. The kinetic model proposed by Britten et al. (1991) was the first comprehensive silane decomposition model, incorporating 70 elementary reactions and 25 chemical species. The key feature in this mechanism was postulated earlier by Hartman et al. (1987), and involves the competition between thermal stabilization and decomposition of the 79 excited state silyl-peroxy radical, xH 3 SiOO, formed by the highly exothermic reaction of SiH 3 and O 2 (Murakami et al., 1996); however, the nucleation of SiO 2 to form solid particulate was not considered and SiO 2 was assumed to be created by elementary gas-phase reactions. Another kinetic model of silane decomposition reported by Fukutani et al. (1991a,b) the same year was less comprehensive consisting of 45 elementary reactions. This model concludes that at low temperatures silane decomposes first into SiH 3 , while at high temperatures it decomposes into SiH 2 (Babushok et al., 1998). Koda (1992) published a comprehensive review regarding the pyrolysis of SiH 4 at different conditions. The reaction rate constants of silane with atomic hydrogen and atomic oxygen to produce SiH 3 were investigated and measured by Goumri et al. (1993) and Ding and Marshall (1993), respectively, since the accurate estimates of thermochemical data of Si/O/H-containing molecules are very crucial as a first key step on modeling of the combustion of silicon-containing precursors. Such thorough calculations were carried out by Zachariah and Tsang (1995) who computed the thermochemistry and kinetics of high-temperature reactions of silicon oxy-hydride species (Si x H y O z ). A model of silane combustion using updated thermochemical data was proposed by Babushok et al. (1998) designed to capture the upper explosion limit for silane under isothermal and non-isothermal conditions. Simulation studies showed two distinct reaction mechanisms, one occurring at high temperatures and the other and low temperatures. At high temperatures, reactions of the SiH 2 radical dominate, and at low temperatures, the kinetics of SiH 3 is controlling. Considering the importance of silyl (SiH 3 ) reactions, comprehensive investigations were carried out (Murakami et al., 1996; Kondo et al., 1997) in order to study the kinetics of SiH 3 80 reactions and related compounds. In the most recent silane combustion model suggested by Miller et al. (2004), previous knowledge advancements regarding silyl production and consumption at high and low temperatures were considered and a more realistic decomposition mechanism was developed which includes 69 species and 201 reactions. The particle formation process in flames involves three steps, namely the formation of condensing species, followed by nucleation, and growth. Because these processes all occur in the flame, the characteristics of the flame (e.g., species concentrations, flame temperature, residence time, etc.) all have a great effect on these processes. Chung et al. (1991) experimentally investigated silane diffusion flames and concluded that at low SiH 4 concentrations and/or when a high-temperature flame is used, SiO 2 is the species that nucleates. At high SiH 4 concentrations and/or when a low-temperature flame is used, nucleation of SiO x (x = 0 and/or 1) in addition to SiO 2 is believed to also take place in the flame. Other system parameters, such as flame stoichiometry, diluent concentration, and gas flow rates, can all influence the morphology to the extent that they affect the system temperature field and the particle residence time. A comprehensive review paper was published by Wooldridge (1998) which summarizes experimental and theoretical investigations of the fundamental processes governing gas-phase combustion synthesis of particles and discusses the experimental methods utilized and the parameters influencing particle morphology and composition. The review concludes that the particle growth mechanism (i.e., nucleation, deposition, coalescence, and agglomeration) is a strong function of the system conditions. Precursor concentration, the temperature field and residence time of the particles in the synthesis environment are considered to be the most significant 81 factors influencing particles morphology (Pratsinis, 1998; Ulrich, 1971; Zhu and Pratsinis, 1997; Formenti et al., 1972). Considering the challenges one faces with understanding and modeling of particulate formation during siloxane combustion due to the lack of a reliable decomposition mechanism, silane was chosen in this study as a model silicon- containing compound in order to study silica micro-particulate formation during RNG combustion in place of its siloxane trace contaminants. The focus of the study, whose results are presented in this Chapter, is the characterizations of the fate of silane during combustion of RNG under well-controlled laboratory conditions and at concentration levels typical of those at which the siloxanes are encountered in RNG. The performance of the two main silane decomposition mechanistic models currently available (Britten et al., 1991; Miller et al., 2004) has been evaluated for the conditions encountered in RNG combustion applications. In a subsequent study, the mechanism of silane decomposition more appropriate to describe its fate in the RNG combustion environment is used to describe silica micro-particulate formation in RNG flames. In what follows, first the experimental and modeling approaches are described, and results are presented and discussed next. 3.2. Experimental Approach The silane decomposition was studied in the counterflow flame configuration, shown schematically in Figure 3.1. Temperature profiles were measured using thermocouples, while the laser extinction technique was applied for the in-situ measurement of the solid particles’ volume fraction, f v . Samples were withdrawn from 82 the flame region and analyzed via GC/MSD to determine the silane concentration profiles. To probe the nature of deposits formed on surfaces in proximity to the RNG flames, flat Ni/Cr metal strips were placed downstream of a Bunsen flame and the chemical composition and structure of the solid particles were investigated by scanning electron microscopy (SEM) and energy dispersive analysis by X-Ray (EDAX). Figure 3.1. Schematic of the experimental configuration. The simulated RNG utilized in this study consists of 98% CH 4 and 2% CO 2 , both 99.97% pure, from the Gilmore Liquid Air Company (South El Monte, CA, USA). The silane gas (99.999%, CAS: 7803-62-5) used in this study was purchased from Air Liquide America Specialty Gases LLC (Morrisville, PA, USA). To generate ppm- level silane concentrations in the main feed, a mass flow controller (MFC – Teledyne 83 Hastings HFC-202) was utilized to deliver 0.2 SCCS of silane gas into a heated air stream, followed by further dilution to reach a concentration of ~1000 ppm v . Another mass flow controller (MFC – Teledyne Hastings HFC-202) was then utilized to deliver ~1% of this mixture to a premixed RNG/air stream to generate the desired silane concentrations in the range of (2 – 30 ppm v ). The ability to generate reliably the required concentrations was verified via a GC/MSD system (Agilent Technologies, 7890A GC System; 5975C Inert XL MSD), calibrated using standard gas samples of silane in nitrogen. The final RNG/air/silane feed stream is introduced into the bottom burner of the experimental system, see Figure 3.2 for a schematic of the experimental configuration (Law et al., 1988; Wu and Law, 1985; Egolfopoulos et al., 1989). The burner is a straight tube surrounded by a nitrogen co-flow channel in order to assure the isolation of the main jet from the ambient environment. Flat flames are generated and stabilized by counter-flowing a nitrogen stream from the top burner, against a pre- mixed RNG/air mixture exiting from the bottom burner. The nozzle diameter for both burners was 14 mm, and so was the separation distance between the nozzles. For the experiments reported here, the air/fuel equivalence ratio ( φ) of the stream exiting the lower burner was kept at 0.8, and the global strain rate K (defined as twice the nozzle exit velocity divided by the burner separation distance) was kept constant at 131.6 s -1 . 84 Figure 3.2. Schematic of the counter-flow configuration (Adapted from Feng, 2011). Thermocouples were used to measure the flame temperature profiles, as noted above. Thermocouples are known to perturb the flow fields, and to interfere with heat transfer mechanisms, like radiation, thus their presence needs to be accounted for (Heitor and Moreira, 1993). In this study an R-type thermocouple (Omega P13R-003) was utilized coated with a Y/BeO ceramic film to minimize its surface reactivity (Kent, 1970; Shandross et al., 1991). To avoid bending of the thin thermocouple wire when placed in the hot flame, which makes it difficult to determine the measurement position, a special device was built similar to that in Cundy et al. (1986) and Ahuja and Miller (1993). The thermocouple assembly was mounted onto a multistage translational stand capable of moving in µm increments. The cold junction temperature is adjusted by calibrating it at room temperature. The radiation corrections were made based on the approach of Shaddix (1999) and McEnally et al. (1997) using an emissivity value of 0.3 (Peterson and Laurendeau, 1985). 85 As previously noted, gas-phase samples were taken from the flat flame using a quartz microprobe with an end opening of ~150 µm positioned on a stand having 3-D (XYZ) mobility (Thorlabs, PT3/M, 25 mm XYZ Translation Stage) in order to measure the variations of silane concentration in the simulated RNG flame. To begin the sampling, the tip of the probe was positioned at the top of the lower burner (Z=0), and then moved vertically in steps in order to collect gas samples in the pre-flame, flame, and post-flame regions. The positioning system, which includes a Cathetometer, was capable of locating accurately the initial position of the probe (as well as of the burners themselves) within 25 µm, thus minimizing experimental uncertainty. 500 µl gas-phase samples were taken in each position using a 1.0 ml syringe (Hamilton, 1001SL 1.0 ml SYR, model no. 81356) and analyzed by a GC/MSD, as noted above. To measure f v , a laser-light extinction system, similar to that of Wang et al. (1996), was utilized. The system utilizes a continuous Ar-ion laser beam (Lexel Laser, model 95) with a wavelength of 488 nm and power of 10 mW, directed via a series of mirrors and focused with a lens (focal length of 150 mm) before passing through the flame region. The intensity of the light beam was measured with a photomultiplier (Thorlabs, DET100A, large-area Si detector, 400-1100 nm, Φ=9.8 mm), whose output was processed by a digital oscilloscope (Tektronix TDS 2004B) interfaced to a computer. Two light filters (NE10A-A, NG4-A Coated, mounted absorptive ND Filter, and NE20A-A, NG9-A Coated, mounted absorptive ND Filter) were installed on the photomultiplier to prevent absorption of interfering lights and for protecting the photomultiplier. A data acquisition device (NI USB-6221), using LabView 8.6, was utilized to process the data. The lens, two of the mirrors, and the photomultiplier 86 were mounted in a frame connected to two stepper-motors (Sherline Products, 67130 stepper motor) controlled by an automated motor controller (Anaheim Automation, model no. DPF73353). This allowed the laser beam to move both vertically and horizontally by moving each of the two mirrors separately. After beam alignment via positioning of the lens, mirrors, and of the photomultiplier, the intensity profiles (I/I 0 ) of the beam exiting the flame in the absence and/or presence of solid particles were generated, by moving the laser beam both horizontally (y-axis) and vertically (z-axis). Based on experimental data repeatability, the accuracy of the measured volume fractions was estimated to be ±17% in a fixed alignment system; though the uncertainty is rather large, the experimental results are qualitatively very reproducible. Further details of how f v is measured can be found elsewhere (e.g., Feng et al., 2012). As noted above, in order to study the deposits formed on surfaces in proximity to the silane-containing RNG flames, flat Ni/Cr metal strips (5×12 mm, 0.5 mm thick -- the exact chemical composition, as reported by the manufacturer, is shown in Table 3.1) were placed downstream of a Bunsen flame. Prior to their use in the experiments, their surfaces were polished by a polisher (Buehler, Grinder/Polisher EcoMet ® 3) using a slurry solution of 0.3 µm Al 2 O 3 powder, followed by washing in an ultrasonic cleaner first with acetone (15 min), methanol (15 min), and then with deionized water (15 min). In the experiments, the metal strips were placed 10 cm above the exit of the bottom burner tube in a Bunsen-type flame of simulated RNG containing 16 ppm v of silane (based on the total air/RNG mixture), φ = 0.8, burner exit velocity of 92 cm/s, and a mixture flow rate of 141.8 SCCS. SEM (JEOL JSM-6610 LV) and EDAX (JSM 6490) were utilized to probe the nature of the deposits. 87 Fe Cr Mo Cu Co W Ni Ti C Mn Si Al Other 17.0- 20.0 20.5- 23.0 8.0- 10.0 <0.5 0.5- 2.5 0.2- 1.0 40.6- 50.5 <0.15 0.05- 0.15 <1.0 <1.0 <0.5 <0.08 Table 3.1. Chemical composition (wt.%) of the Ni/Cr alloy. 3.3. Modeling Approach The concentration profiles were computed using a Chemkin-based (Kee et al., 1983; 1989a) opposed-jet flame code (Kee et al., 1989b; Egolfopoulos and Campbell, 1996), coupled with the USC Mech II (Wang et al., 2007) kinetic model. The Lennard-Jones parameters of the various species were estimated by the method suggested by Wang and Frenklach (1994). As noted above, two different silane decomposition mechanisms are studied here. One of these is the kinetic model proposed by Britten et al. (1991), which as discussed previously, was the first comprehensive silane combustion mechanism proposed that uses a system of 70 elementary reaction steps and 25 chemical species. The competition between thermal stabilization and decomposition of the excited state silyl-peroxy radical (xH 3 SiOO) formed by the highly exothermic reaction of SiH 3 and O 2 is the key idea behind this mechanism, as noted earlier. The second model of silane decomposition investigated is the one reported by Miller et al. (2004). The model includes 69 species and 201 elementary reactions and incorporates previous ideas regarding silyl production and consumption at high and low temperatures. The key additional idea incorporated by Miller et al. (2004) is the existence of a transitory state between SiH 3 and HSiO(OH), and the conclusion that the channel producing H atoms (SiH 3 + O 2 → c-OSiH 2 O + H) is dominant at high 88 temperatures. Each of the above two models were added separately to the USC Mech- II. 3.4. Results and Discussion 3.4.1. Flame Temperature Profile Flame temperature profiles were measured for both flat and Bunsen flames in the absence and presence of silane in simulated RNG flames. Not surprisingly, given the very low (ppm-level) silane concentrations, temperature profiles were shown to be identical in the flames with and without silane, to the extent that it could be discerned via the intrusive temperature measurements. The experimental and theoretical temperature profiles for the opposed-jet, φ = 0.8, premixed flame used in this study are reported elsewhere (Jalali et al., 2013). Satisfactory agreement is found between the experimental and simulated temperatures in the luminous region. On the other hand, in the pre-flame and post-flame regions the experimental profiles are broader. Similar observations were previously reported by other investigators (e.g., Feng et al., 2010 and references therein), the differences being attributed to the experimental difficulties with the intrusive temperature measurement technique, which are well known and discussed in the literature (e.g., Cundy et al., 1986). For the kinetic data analysis, the theoretical temperature profiles were, therefore, utilized, and further pertinent discussion will follow. 89 3.4.2. Experimental Silane Concentration Profiles and Simulations In Figure 3.3, the experimental concentration profiles for silane in the pre-flame, flame, and post-flame regions for various silane feed concentrations are shown. The concentration measurement results confirm that silane decomposition begins before the luminous flame region and completes, for the most part, inside the luminous region. Therefore, the profiles are sharp with 100% conversion attained for all concentrations in the front-end of the flame region with no silane escaping from the post-flame region of the flat flame. Also, one may observe that the reaction zone width is very thin suggesting that the decomposition process is very fast. By comparing in Figure 3.3 the concentration profiles, calculated using the decomposition model of Miller et al. (2004), with the experimental data, a good qualitative agreement is observed and the trends appear reasonable. Moreover, when comparing silane concentration profiles with the ones attained using heavy siloxanes, e.g., L 4 or D 4 (Jalali et al., 2013), the negative slopes of silane profiles are much steeper than the ones for siloxanes. Therefore, one may conclude that the silane decomposes easier (faster) than siloxanes. 90 0.0E+00 5.0E-06 1.0E-05 1.5E-05 0.3 0.4 0.5 0.6 0.7 Distance from bottom burner [cm] Silane ppmv Simul. 14ppmv Simul. 7ppmv Simul. 2ppmv Exp. 14ppmv Exp. 7ppmv Exp. 2ppmv Luminous Region Figure 3.3. Computed and experimental silane concentration profiles in the pre-flame, flame, and post-flame regions for various silane feed concentrations. 3.4.3. Comparison of the Two Silane Decomposition Models The performance of the two major silane decomposition models proposed in the literature is evaluated based on the solid experimental concentration data that are obtained in this study. Figure 3.4 shows the computed and experimental results for silane concentration profiles in the pre-flame, flame, and post-flame regions of simulated lean ( φ=0.8) RNG flames for a 7 ppm v feed concentrations using the Miller et al. (2004) and Britten et al. (1991) silane decomposition models. The sum of squared differences, σ, between the simulated and experimental values for the two models is also calculated; the obtained concentration profile using the Miller et al. (2004) model has smaller σ value indicating a better fit (based on that measure of “goodness-of-fit”) to the experimental results. Also, the simulated profile using the 91 Miller et al. (2004) model, qualitatively, shows a somewhat better agreement with the experimental data as well. Similarly, better agreement between the experimental data and the calculated concentration profiles was obtained using Miller et al. (2004) model for all other silane concentrations studied. 0.00E+00 2.00E-06 4.00E-06 6.00E-06 8.00E-06 0.4 0.45 0.5 0.55 0.6 Distance from bottom burner [cm] Silane ppmv Miller et al. (2004) Britten et al. (1991) Exp. 7 ppmv Figure 3.4. Computed and experimental silane concentration profiles in the pre-flame, flame, and post-flame regions of simulated lean ( φ=0.8) RNG flames for the 7 ppm v feed concentrations using the Miller et al. (2004) and Britten et al. (1991) silane decomposition models. Reaction path analysis was also performed on the computed structures of RNG simulated lean ( φ=0.8) flames for the 7 ppm v feed concentration of silane using both the Britten et al. (1991) and the Miller et al. (2004) models. Figures 3.5 and 3.6 depict the major pathways leading into SiO 2 formation for the two different models. 92 SIH3 SIH4 HSIO(OH) SIO SIO2 SIH2 H2SIO H3SIOO HSIO SIOOH 99.9 98.9 100 100 99 82.1 17.6 99.4 100 0.8 xH3SIOO 99.9 8.1 91.1 Figure 3.5. Reaction path analysis resulting in SiO 2 formation in the simulated lean ( φ=0.8) RNG flames for a 7 ppm v feed concentration of silane using the Britten et al. (1991) model. Each number indicates the fractional (%) consumption of the radical from which the arrow begins from. 93 SIH3 SIH4 0.2 99.1 91.1 99.1 6.1 H3SIOO c-OSIH2O w-OSIH2O HSIO(OH) HOSIO SI(OH)2 SIO SIO2 SIH2 H2SIO H3SIOOH H3SIO H2SIOH HSIO HSIOH 5.7 92.4 99.7 99.8 93.3 3.1 6.6 99 100 0.7 99.3 0.6 26.8 10.4 0.8 100 100 1.3 62.4 14.6 99.3 5.3 78.7 100 Figure 3.6. Reaction path analysis resulting into SiO 2 formation in the simulated lean ( φ=0.8) RNG flames for a 7 ppm v feed concentration of silane using the Miller et al. (2004) model. Each number indicates the fractional (%) consumption of the radical from which the arrow begins with. Based on the reaction path analysis, there are similarities but also significant differences between the two models. Both models suggest, for example, that the first key step of silane decomposition is an H-radical abstraction to produce SiH 3 although small amount of SiH 2 may also be produced. This observation is consistent with the heats of formation of SiH 3 and SiH 2 which are +47.8 cal/mol and +58.0 cal/mol, 94 respectively, from which one expects the SiH 3 (based on its less endothermic heat of formation) to be the dominant product of the SiH 4 abstraction reaction. In the model of Miller et al. (2004), 99.1% of SiH 4 is converted into SiH 3 through the following two reactions: SiH 4 + H → H 2 + SiH 3 (80.1% contribution) (R3.1) SiH 4 + OH → SiH 3 + H 2 O (19.0% contribution) (R3.2) and only 0.7% of SiH 4 produces SiH 2 . In the model of Britten et al. (1991), however, the above two reactions have a total 79.3% contribution to the SiH 3 production. More importantly, however, a significantly larger fraction, 17.6%, of SiH 4 generates SiH 2 through the following reaction: SiH 4 → SiH 2 + H 2 (R3.3) Both models suggest that HSiO(OH) is the most important intermediate radical by which the reactants’ pool can be converted into solid products (SiO and eventually SiO 2 ), and the radical appears to play a key role in silica formation in both mechanisms. There are differences in the channels leading to HSiO(OH) formation among the two mechanisms, however. In the model of Miller et al. (2004), three major reactions consume SiH 3 : SiH 3 + O 2 → HSiO(OH) + H (92.4% contribution) (R3.4) SiH 3 + O 2 (+M) → H 3 SiOO(+M) (6.1% contribution) (R3.5) 95 SiH 3 + O 2 → H 2 SiO + OH (1.3% contribution) (R3.6) The heat of formation ( ∆H f ° ) of HSiO(OH) is –125 cal/mol, while the heats of formation for H 3 SiOO and H 2 SiO are fairly similar to each other (–24.2 cal/mol and – 25.3 cal/mol, respectively). Considering the very high exothermic ∆H f ° of HSiO(OH), it is likely to be produced directly from SiH 3 . However, in the model of Britten et al. (1991), it is an indirect route involving xH 3 SiOO (requiring thermal stabilization in order to produce H 3 SiOO), H 3 SiOO, and H 2 SiO that leads into HSiO(OH) formation. As noted above, the direct production of HSiO(OH) from SiH 3 is sufficiently exothermic so that additional intermediate step(s) should not be, in principle, required. The same reasoning may also be applied to the model Britten et al. (1991) itself. It is more likely to generate HSiO(OH) directly from xH 3 SiOO instead of having H 2 SiO generated as an intermediate species and then consumed through the following reaction to produce HSiO(OH), as their model suggests: H 2 SiO + H 2 O → HSiO(OH) + H 2 (R3.7) Although in the model of Miller et al. (2004) most of the H 2 SiO (62.4%) eventually converts into HSiO(OH) according to the following reactions H 2 SiO + OH → HSiO(OH) + H (52.9% contribution) (R3.8) H 2 SiO + HO 2 → HSiO(OH) + OH (1.4% contribution) (R3.9) H 2 SiO + H 2 O → HSiO(OH) + H 2 (8.1% contribution) (R3.10) 96 it is not considered as an important intermediate species since only 1.3% of SiH 3 produces H 2 SiO. In addition to the differences in the routes via which an adequate amount of HSiO(OH) is produced in the radical pool, there is also difference among the two models in the way the HSiO(OH) is consumed. In the model of Miller et al. (2004), the majority of HSiO(OH) converts into Si(OH) 2 (via reaction R3.11, below), with only a small fraction converting directly into SiO 2 , via R3.12. (The rest converts into HOSiO via R3.13). HSiO(OH) → Si(OH) 2 (91.1% contribution) (R3.11) HSiO(OH) + O 2 → HO 2 + H + SiO 2 (5.7% contribution) (R3.12) HSiO(OH) + H → HOSiO + H 2 (3.1% contribution) (R3.13) In the model of Britten et al. (1991), HSiO(OH) decomposes into SiOOH via reactions: HSiO(OH) → SiOOH+H (86.1% contribution) (R3.14) HSiO(OH)+O 2 → SiOOH+HO 2 (13.9% contribution) (R3.15) Considering that the heat of formation of Si(OH) 2 is –115 cal/mol and that of SiOOH –109 cal/mol, one expects the formation of Si(OH) 2 from HSiO(OH) to be a bit more favorable than that of SiOOH. In the model of Britten et al. (1991), SiOOH consumption is the main route of silica formation. However, in the literature (e.g., Wooldridge, 1998), SiO is reported to be the main precursor for silica formation, with 97 the reaction of SiO being considered as the main route of silica production in most reports. In the model of Miller et al. (2004), 99.7% of Si(OH) 2 converts into SiO, and then 99.8% of SiO, through the following reaction, reacts to produce SiO 2 : SiO + O 2 → O + SiO 2 (R3.16) Figure 3.7 depicts the sensitivity coefficients of (X HSiO(OH) ) max on the kinetics of silane decomposition in simulated lean ( φ=0.8) RNG flames for a 7 ppm v feed concentrations of silane, computed using the Britten et al. (1991) model. The sensitivity coefficient is defined here as ) / /( ) / ( A Y Y A i i ∂ ∂ , Y i being the maximum mass fraction of species i and A the pre-exponential factor of an elementary reaction. As it is shown in this figure, large positive sensitivity coefficients indicate that R3.1 and R3.2 are the most important reactions to produce HSiO(OH). Also, one may observe the notable negative influence of the R3.3 reaction on HSiO(OH) formation, which is an unfavorable process that may prevent the system to generate HSiO(OH). 98 -1.50E-01 -1.00E-01 -5.00E-02 0.00E+00 5.00E-02 1.00E-01 1.50E-01 2.00E-01 SIH4+OH<=>SIH3+H2O SIH4+H<=>SIH3+H2 SIH4+O<=>SIH3+OH H2SIO+H2O<=>HSIO(OH)+H2 H2SIO+O2<=>HSIO+HO2 SIH4+HO2<=>SIH3+H2O2 H3SIOO<=>H2SIO+OH SIH4<=>SIH3+H HSIO+OH<=>SIO+H2O HSIO+H<=>SIO+H2 H2SIO+O<=>HSIO+OH HSIO+O2<=>SIO+HO2 H2SIO+H<=>HSIO+H2 SIH4<=>SIH2+H2 Sensitivity Coefficients HSIO(OH) Figure 3.7. Sensitivity of the maximum HSiO(OH) mole fraction on the kinetics of silane decomposition in simulated lean ( φ=0.8) RNG flames for 7 ppm v feed concentrations of silane, computed using the model of Britten et al. (1991). Figure 3.8 shows the sensitivity coefficients of (X HSiO(OH) ) max on the kinetics of silane decomposition in the same simulated lean ( φ=0.8) RNG flames for 7 ppm v feed concentrations of silane, computed using the model of Miller et al. (2004). The R3.1 and R3.2 reactions have again significant positive effect on HSiO(OH) formation. Also, as it is shown in the figure, the production of HSiO(OH) directly depends on reaction R3.4. Reactions R3.11 and R3.12 have a remarkable negative effect on the HSiO(OH) mole fraction since these are the major routes of its consumption. 99 -2.00E-01 -1.00E-01 0.00E+00 1.00E-01 2.00E-01 3.00E-01 4.00E-01 5.00E-01 SIH4+H<=>H2+SIH3 SIH4+OH=>SIH3+H2O SIH3+O2<=>HSIO(OH)+H H3SIOO(+M)<=>HSIO(OH)+H(+M) HSIO(OH)<=>HOSIO+H H2SIO+OH<=>HSIO(OH)+H SIH4+HO2<=>SIH3+H2O2 SI(OH)2<=>H+HOSIO OH+HSIO(OH)<=>H2O+HOSIO SIH3+O2<=>H2SIO+OH H+HSIO(OH)<=>HOSIO+H2 SIH3+O2(+M)<=>H3SIOO(+M) O2+HSIO(OH)<=>HO2+H+SIO2 HSIO(OH)<=>SI(OH)2 Sensitivity Coefficients HSIO(OH) Figure 3.8. Sensitivity of the maximum HSiO(OH) mole fraction on the kinetics of silane decomposition in simulated lean ( φ=0.8) RNG flames for 7 ppm v feed concentrations of silane, computed using the Miller et al. (2004) model. 3.4.4. Solid Particle Volume Fraction Measurements The f v profiles in the silane-laced RNG flat flames (for two different silane feed concentrations of 7 and 14 ppm v ) are plotted in Figure 3.9. The experimental and simulated silane conversions (the latter calculated using the decomposition model of Miller et al. (2004)) for these two concentrations are shown on the same figure as well. Good agreement is observed between experimental and simulated results. As one may notice, the conversion profiles are independent of the feed concentration indicative that overall first order global reaction kinetics may prevail, attributed to very low silane concentrations in the feed gas. Similar observations, in terms of the overall 100 reaction order, were previously made in the study of siloxane combustion in flat RNG flames by Jalali et al. (2013), with the widths of the f v profiles being fairly similar. As was the case with siloxanes, the f v profile for silane again peaks in the flame zone (luminous region). As was explained by Jalali et al. (2013), this is a result of two competing effects: On the one hand, as the silane molecules approach the flame zone, their decomposition intensifies resulting in greater particle production rates, which tends to increase f v . On the other hand, as the temperature increases within the flame, the gas density keeps decreasing due to dilatation, which tends to decrease f v given that the particle number density follows closely the variations of the gas phase density for small particles that follow, in principle, the gas phase flow (e.g., Egolfopoulos and Campbell, 1999). 0 20 40 60 80 100 23 4 5 6 7 Distance from bottom burner [mm] Fraction of reacted Silane 0.0E+00 5.0E-09 1.0E-08 1.5E-08 2.0E-08 2.5E-08 Solid particle volume fraction Exp. 14 ppmv Exp. 7 ppmv Simul. 14 ppmv Simul. 7 ppmv Vol. Frac. 14 ppmv Vol. Frac. 7 ppmv Fraction of reacted Silane - Simulation results for 14 and 7ppmv Fraction of reacted Silane - Experimental results for 14 and 7ppmv Solid particle vol. frac. for 14 and 7ppmv Figure 3.9. Simulated and experimental silane conversion and solid particle volume fraction profiles. 101 Figure 3.10 depicts the maximum f v values measured in the RNG flame for various siloxanes (adapted from Jalali et al., 2013) and silane as a function of their concentrations in the RNG/air feed mixture. As a comparison, the volume fractions of solid silica particles measured is one to two orders of magnitude less than the maximum volume fractions of soot particles measured by Feng et al. (2012) during combustion of normal alkanes, e.g., propane and butane. The results exhibit a fairly linear relationship between f v and the feed siloxane/silane concentration in the fuel/air mixture. This is expected, given that the siloxanes and silane decompose completely in the flame to generate these micro-particulates, as it was shown in Figure 3.3. As expected also, since silane gas has one silicon atom less than L 2 in its chemical structure, the f v of L 2 lies above that of silane. As Figure 3.10 indicates, the laser extinction method is a sensitive indicator of the volume of particles formed, which, in turn, strongly correlates to the number of silicon atoms in the chemical structure of the Si-containing molecules. Using silane as a precursor of solid silica particles in the RNG flame generates also a surrounding orange halo around the flame. This is because of the black body radiation of the SiO 2 particles at high temperatures (Yeh et al., 2001b; Ma et al., 2003). 102 y = 1E-09x + 4E-09 R 2 = 0.9821 y = 3E-09x - 4E-09 R 2 = 0.9892 y = 3E-09x + 5E-10 R 2 = 0.999 y = 3E-09x + 6E-10 R 2 = 0.9794 y = 1E-09x + 6E-09 R 2 = 0.982 0.0E+00 1.0E-08 2.0E-08 3.0E-08 4.0E-08 5.0E-08 6.0E-08 7.0E-08 0 5 10 15 20 25 30 ppmv Maximum solid particle volume fraction L2 L3 L4 D4 Silane Linear (L2) Linear (L3) Linear (L4) Linear (D4) Linear (Silane) Figure 3.10. Maximum solid particle volume fraction for silane in comparison with the siloxanes. 3.4.5. Surface and Particle Analysis Results Metal strips which were prepared by the methods described above, were exposed to the RNG post-flame products, in a preliminary series of experiments, to study the nature of the solid particulates formed. Three different types of particles were identified via SEM/EDAX, namely carbon, SiO x , and SiO x C y . While the formation of silica-type particles was expected for a lean RNG/air combustion mixture laced with a Si-containing precursor, production of carbon and carbon-containing components was unexpected. Additional experiments and investigations demonstrated that the number of carbon and SiO x C y particles would significantly diminish if the metal surface were exposed to a pure RNG flame for a certain period of time prior to the silane being added to the flame (similar observations were also previously reported by Jalali et al. 103 (2013) for siloxane-RNG flames). We attribute this to the oxidation of surface carbon on the metal strips, likely to happen via a number of global surface reactions for carbon oxidation (Libby and Blake, 1981; Dixon-Lewis et al., 1991): C + O 2 → CO 2 (R3.17) 2C + O 2 → 2CO (R3.18) C + CO 2 → 2CO (R3.19) C + H 2 O → CO + H 2 (R3.20) The most likely reaction for surface carbon removal is R3.18 as it is more prone to take place at temperatures between 600 – 1200 K. In our studies, for example, it was found that by placing the metal strip for 15 min at a position in the post-flame region where the steady state temperature of the plate surface was about 730 o C, eliminated totally the presence of any pure carbon or carbon-containing particles (in all further experiments, therefore, the metal surfaces would be subjected to this pretreatment prior to the silane being added to the RNG flame). Since the concentration of silicon-containing precursor is very low (16 ppm v in the experiments reported here), at short exposure times changes on the surface are not visible to the naked eye (however, under the electron microscope a few white particles can be detected). That the metal strip surface does not appear visibly coated by silica does not mean that it is not present there, however. In fact, EDAX indicates the presence of small amounts of silicon being present on the surface, (aside from the occasional large particles formed), right from the start. After 2 min of exposure the metal surface still has not changed visually by much; however, under the electron 104 microscope one observes quite a few particles, ~10 – 15 particles/mm 2 (see Figure 3.11). This particle density is smaller than that observed with siloxanes (D4) at a similar time of exposure (Jalali et al., 2013), consistent also the differences in the f v profiles, see Figure 3.10 Figure 3.11. An SEM image of deposited particles using 16 ppm v silane for 2 min. Figure 3.12 indicates the silicon content of the surface exposed to a silane-RNG flame, which is again substantially lower that observed on surfaces exposed to siloxane-RNG flames, consistent with the observation that silica deposition rates are dependent on the number of silicon atoms the precursor molecule contains. 105 Elements C O Ni Al Si Mo Cr Fe Weight % 0.00 8.31 46.37 0.29 1.19 8.89 18.42 16.53 Figure 3.12. Surface analysis results of the area shown in Figure 3.11. Focusing the EDAX beam on one of these particles (see Figure 3.13) indicates that it consists of purely silicon and oxygen, while the other elements are due to the penetration of the beam through the particle to reach the metal plate underneath. Generally, the morphology and chemical composition of the particles formed are very similar to those observed on surfaces upon exposure to siloxane-RNG flames, though the sizes differ as it takes a much longer exposure to silane-RNG flames (than in the case with siloxane-RNG flames) in order to cover the surface with a fine silica layer. The colors of the silicon dioxide and silicon-monoxide layers are reported to be white and brown, respectively (Yeh et al., 2001a). In all experiments carried out in this study only a white layer was observed on the surface (the same was also true for the siloxane-RNG flames). 106 Elements C O Ni Al Si Mo Cr Fe Weight % 0.00 21.3820.83 2.99 16.048.28 11.82 18.66 Figure 3.13. EDAX analysis result of a particle formed using 16ppm v silane for 2 min. 3.5. Concluding Remarks In summary, biogas is a valuable source of energy that contains, however, a variety of contaminants. Volatile organic silicon compounds (siloxanes) are the most adverse impurities found in biogas (Ohannessian et al., 2008), which could cause many unwanted negative effects on combustion equipment. Therefore, it is of importance to progress the state of knowledge about how these impurities decompose during biogas combustion and the mechanism via which they produce solid particles. There is still not much advancement regarding siloxane decomposition modeling. Taking into account the complexities regarding the understanding of siloxanes’ oxidation, silane was selected in this study as a model silicon-containing precursor of silica solid micro-particulates in order to attain an improved understanding of the mechanisms of precursor degradation and silica formation in ppm v -level concentrations. Many intermediate radicals generated in silane combustion are the ones also thought to form in siloxanes decomposition. Thus, any detailed mechanistic information generated during silane decomposition will also potentially facilitate the way to develop a reliable decomposition model for siloxanes during RNG combustion. As a result, this study is focused on silane decomposition in RNG combustion under well-controlled laboratory conditions. Silane concentration profiles were measured in the opposed-jet flame configuration and it was shown that silane 107 decomposes for the most part in the pre-flame and the luminous region with little, if any, silane molecules escaping in the post-flame region. Two major silane decomposition models have been considered, one from Britten et al. (1991) and the other from Miller et al. (2004). Their performance has been evaluated using the experimental concentration profiles obtained under identical conditions to those encountered during RNG combustion for power generation. In contrast to the past ideas regarding the production and consumption of SiH 2 as the main species in high temperature combustion, by performing path and sensitivity analyses on these mechanisms, it was shown that the main channel of silica formation is initiated by SiH 3 formation. Solid particle volume fractions were also measured using the laser extinction method, which was shown to be a very reliable technique to be used for the measurement of such particles. It was confirmed that the relationship between different silane concentrations present in RNG and the particle volume fraction in the flue-gas is linear. Also, it was shown that adding silane to the RNG flame produces less solid particles than when adding siloxanes at the same concentration level. Surface analyses have been performed on the particles deposited on metal surfaces in contact with silane-containg RNG flames order to determine their bulk and surface characteristics. It was found that, using the same (to those of siloxanes) silane concentrations in the RNG flames, it takes a longer deposition time for the silane than for the siloxanes in order to deposit the same number of particles per unit surface area. 108 Chapter 4 – Conclusions and Future Work 4.1. Concluding Remarks Volatile organic silicon compounds (siloxanes) are, arguably, the most adverse impurities in biogas. Prior to the initiation of this research, siloxanes were known to decompose in RNG flames and to produce solid silica micro-particulates which cause serious damages in different energy equipment. Also, the concern existed that these impurities may inadvertently, find their way into the NG pipeline system and thus into people’s homes and in NG equipment and appliances. As noted in the Introduction of this Thesis, current techniques to remove the siloxanes are rather inefficient (Ajhar et al., 2010), and reliable field analytical equipment for their detection is lacking. Therefore, it is of importance to improve the state of the knowledge about how these impurities decompose during biogas combustion in order produce solid particles. The main objective of this dissertation was, therefore, to provide a better insight into the decomposition of siloxanes and to generate additional knowledge regarding their oxidation mechanisms. Siloxane concentration profiles were measured in the opposed-jet flame configuration and it was shown that siloxanes decompose, for the most part, in the pre-flame and the luminous flame region with small amounts, if any, escaping in the post-flame region. First-order (with respect to the siloxane concentration) global 109 decomposition reaction kinetics appear to provide a qualitative fit of the experimental data. The global rate constants were calculated and they indicated decomposition rates that become slower with increasing siloxane chain length, and for the cyclic siloxanes when compared to their linear counterparts of the same chain length. Solid particle volume fractions were measured as well using the laser extinction method, which was shown to be a very reliable technique to be used for the measurement of such particles. As expected, the results indicate a linear relationship between the siloxane concentration in the fuel and the corresponding particle volume in the flue-gas. The particle volume fractions were found to increase also with the number of silicon atoms in the siloxane molecule. The particle volume fractions were found, in addition, to be comparable between linear and cyclic siloxanes with the same number of silicon atoms in the molecule, which is expected based on the fact that these siloxanes completely decompose in the flame environment. Analysis of the metal surfaces exposed to the flame environment of the siloxane-containing RNG mixtures indicates a substantial impact. The surface roughness increases right from the start upon exposure to the flames, with the surface being coated eventually by a white layer of SiO 2 . Since siloxanes exist in biogas in mixtures the impact of the presence of one siloxane on the decomposition kinetics of another siloxane was investigated as well. No discernible effect was detected by mixing of various siloxanes, which is to be expected, since the concentrations of siloxanes in RNG are very low, with each molecule of siloxanes having a small chance to interact with other siloxanes and to interfere with the decomposition process. 110 There is still not much advancement regarding the development of a siloxane decomposition mechanism. Taking into account the immense complexities involved in the understanding of siloxanes’ oxidation, silane was selected in this study as a model silicon containing precursor of silica solid micro-particulates in order to attain an improved understanding towards precursor degradation and silica formation in low ppm concentrations in combustion mixtures. In our studies, the silane decomposition during RNG combustion was studied under well-controlled laboratory conditions. Silane concentration profiles were measured in the opposed-jet flame configuration and it was shown that, similarly to siloxanes, silane decomposes for the most part in the pre-flame and the luminous flame region with small amounts, if any, escaping in the post-flame region. Solid particle volume fractions were measured again using the laser extinction method. It was confirmed that a linear relationship exists between the silane concentrations present in the RNG and the particle volume fraction in the flue- gas. Also, it was shown that silane-laced RNG flames produce fewer solid particles than the siloxane-containing RNG flame. Surface analyses have been performed on the particles deposited on the metal surfaces in order to determine their bulk and surface characteristics. It was found that using the same concentrations it takes a longer deposition time in the silane flames than in the siloxanes flames in order to deposit the same number of particles. For the interpretation of the obtained experimental silane concentration profiles during RNG combustion, two different silane decomposition models have been considered from Britten et al. (1991) and Miller et al. (2004). By performing path and sensitivity analyses using these mechanisms, it is shown that the main pathway of silica formation is initiated by SiH 3 formation rather than SiH 2 , as it was previously 111 thought of. Many intermediate radicals generated in silane combustion are thought to form during siloxane decomposition as well. Thus, studying silane decomposition kinetics provides valuable information which may facilitate potentially the eventual development of a reliable decomposition mechanism for siloxanes under conditions in which they are used in the energy industry. 4.2. Future Work The present study has made progress in understanding siloxane decomposition during the combustion of renewable natural gas and in characterizing the nature of the deposited particles produced by the oxidation of the different siloxanes. The global reaction rate constants of both linear and cyclic siloxanes have been calculated from the experimental data. In order to gain a comprehensive knowledge of the phenomena that take place during siloxane decomposition in RNG flames, detailed kinetic mechanisms for each siloxane need to be developed. Silane oxidation mechanisms may be a good starting point here, since most of the intermediate radicals produced during silane oxidation are likely to be the ones formed in siloxane decomposition as well. Of value in such future efforts will be the key pathways of silica production that have been identified and verified in this Thesis. The present experimental study of siloxane decomposition must be expanded to larger cyclic siloxanes like D 5 and D 6 , as these are also present in biogas as well. This will require improvements in the current experimental set-up, because these siloxanes have a low vapor pressure and may decompose in the vapor phase prior to entering the RNG flame. Additionally, it is important to carry out the decomposition 112 experiments in lower concentrations, since in a number of landfills the concentrations of organo-silicon compounds are in the order of ppb. To achieve this goal, additional modifications need to be done in the experimental set-up in order to prepare simulated RNG feeds with such low siloxane concentrations. Experiments are also needed in order to investigate combustion of these impurities at various pressures (higher than atmospheric pressure) in order to study the pollutant emission characteristics under conditions more relevant to the real power/energy applications. The generation of such experimental data would allow the further development of new and optimized biogas equipment and applications. 113 References Abatzoglou, N., and Boivin, S. 2009. A review of biogas purification processes. Biofuels, Bioprod. Biorefin., 3, 42–71. Accettola, F., Guebitz, G.M., and Schoeftner, R. 2008. Siloxane removal from biogas by biofiltration: Biodegradation studies. Clean Technol. Environ. Policy, 10, 211–218. Ahuja, S., and Miller, D.L. 1993. Design of a constant tension thermocouple rake suitable for flame studies. Rev. Sci. Instrum., 64, 1358–1359. Ajhar, M., Travesset, M., Yuece, S., and Melin, T. 2010. Siloxane removal from landfill and digester gas – a technology overview. Bioresour. Technol., 101, 2913–2923. Ajhar, M., and Melin, T. 2006. Siloxane removal with gas permeation membranes. Desalination, 200, 234–235. Andrews G.E., Bradley D., and Hundy G.F. 1972. Hot wire anemometer calibration for measurements of small gas velocities. Int. J. Heat Mass Transfer, 15, 1765-1786. Appels, L., Baeyens, J., and Dewil, R. 2008. Siloxane removal from biosolids by peroxidation. Energy Convers. Manage., 49, 2859–2864. Babushok, V.I., Tsang, W., Burgess, D.R., and Zachariah, M.R. 1998. Numerical study of low and high temperature silane combustion. Symp. (Int.) Combust., 27, 2431–2439. 114 Badjagbo, K., Heroux, M., Alaee, M., Moore, S., and Sauve, S. 2010. Quantitative analysis of volatile methylsiloxanes in waste-to-energy landfill biogases using direct APCI-MS/MS. Environ. Sci. Technol., 44, 600–605. Barin, I. 1995. In Sora, K., and Gardiner, J. (Eds) Thermochemical data of pure substances, 3 rd ed., VCH Verlagsgesellschaft mbH, Weinheim (Germany) and VCH Publishers, Inc., New York, NY (USA), Vol. 2. Batdorf, R.L., and Smits, F.M. 1959. Diffusion of impurities into evaporating Silicon. J. Appl. Phys., 30, 259–264. Boulinguiez, B., and Le Cloirec, P. 2010. Adsorption on activated carbons of five selected volatile organic compounds present in biogas: Comparison of granular and fiber cloth materials. Energy Fuels, 24, 4756–4765. Brady, G.W. 1959. A study of amorphous SiO. J. Phys. Chem., 63, 1119–1120. Brewer, L., and Mastick, D.F. 1951. The stability of gaseous diatomic oxides. J. Chem. Phys., 19, 834–843. Brewer, L., and Edwards, R.K. 1954. The stability of SiO solid and gas. J. Phys. Chem., 58, 351–358. Briesen, H., Fuhrmann, A., and Pratsinis, S.E. 1998. The effect of precursor in flame synthesis of SiO 2 . Chem. Eng. Sci., 53, 4105–4112. Britten, J.A., Tong, J., and Westbrook, C.K. 1991. A numerical study of Silane combustion. Symp. (Int.) Combust., 23, 195–202. Butler, C.J., Hayhurst, A.N., and Wynn, E.J.W. 2002. The size and shape of Silica particles produced in flames of H 2 /O 2 /N 2 with a silicon-containing additive. Proc. Combust. Inst., 29, 1047–1054. 115 Chagger, H.K., Hainsworth, D., Patterson, P.M., Pourkashanian, M., and Williams, A. 1996. The formation of SiO 2 from Hexamethyldisiloxane combustion in counterflow methane-air flames. Symp. (Int.) Combust., 26, 1859–1865. Chandramouli, B., and Kamens, R.M. 2001. The photochemical formation and gas- particle partitioning of oxidation products of Decamethylcyclopentasiloxane and Decamethyltetrasiloxane in the atmosphere. Atmos. Environ., 35, 87–95. Chung, S.L., Tsai, M.S., and Lin, H.D. 1991. Formation of particles in a H 2 -O 2 counterflow diffusion flame doped with SiH 4 or SiCl 4 . Combust. Flame, 85, 134–142. Chung, S.L., and Katz, J.L. 1985. The counterflow diffusion flame burner – A new tool for the study of the nucleation of refractory compounds. Combust. Flame, 61, 271–284. Coltrin, M.E., Kee, R.J., and Miller, J.A. 1984. A mathematical model of the coupled fluid mechanics and chemical kinetics in a chemical vapor deposition reactor. J. Electrochem. Soc., 131, 425–434. Coltrin, M.E., Kee, R.J., and Miller, J.A. 1986. A mathematical model of silicon chemical vapor deposition – Further refinements and the effects of thermal diffusion. J. Electrochem. Soc., 133, 1206–1213. Crest, M., Chottier, C., Fine, L., Chatain, V., Ducom, G., Chovelon, J.M., and Germain, P. 2010. On the reliability of sampling and analytical methods to quantify volatile organosilicon compounds (VOSiC) contents in landfill gas. Presented at the Third International Symposium on Energy from Biomass and Waste, Venice, Italy, November 8–11. 116 Cundy, V.A., Morse, J.S., and Senser, D.W. 1986. Constant-tension thermocouple rake suitable for use in flame mode combustion studies. Rev. Sci. Instrum., 57, 1209–1210. Dewil, R., Appels, L., and Baeyens, J. 2006. Energy use of biogas hampered by the presence of siloxanes. Energy Convers. Manage., 47, 1711–1722. Ding, L.Y., and Marshall, P. 1993. Experimental and theoretical studies of the reaction of atomic oxygen with silane. J. Chem. Phys., 98, 8545–8550. Dixon-lewis, G., Bradley, D., and Habik, S.E. 1991. Oxidation rates of carbon particles in methane-air flames. Combust. Flame, 86, 12–20. Dobbins, M.S., and McLay, R.E. 1991. Methods of making fused silica by decomposing siloxanes. U.S. Patent 5,043,002. Drowart, J., De Maria, G., and Inghram, M.G. 1958. Thermodynamic study of SiC utilizing a mass spectrometer. J. Chem. Phys., 29, 1015–1021. Egolfopoulos, F.N., Cho, P., and Law, C.K. 1989. Laminar flame speeds of methane- air mixtures under reduced and elevated pressures. Combust. Flame, 76, 375– 391. Egolfopoulos, F.N., and Campbell, C.S. 1996. Unsteady, counterflowing, strained diffusion flames: Frequency response and scaling. J. Fluid Mech., 318, 1–29. Egolfopoulos, F.N., and Campbell, C.S. 1999. Dynamics and structure of dusty reacting flows: Inert particles in strained, laminar, premixed flames. Combust. Flame, 117, 206–226. 117 Ehrman, S.H., Friedlander, S.K., and Zachariah, M.R. 1998. Characteristics of SiO 2 /TiO 2 nanocomposite particles formed in a premixed flat flame. J. Aerosol Sci., 29, 687–706. Eklund, B., Anderson, E.P., Walker, B.L., and Burrows, D.B. 1998. Characterization of landfill gas composition at the fresh kills municipal solid-waste landfill. Environ. Sci. Technol., 32, 2233–2237. El-Fadel, M., Findikakis, A.N., and Leckie, J.O. 1997. Environmental impacts of solid waste landfilling. J. Environ. Manage., 50, 1–25. Feng, Q., Jalali, A., Fincham, A.M., Wang, Y.L., Tsotsis, T.T., and Egolfopoulos, F.N. 2012. Soot formation in flames of model biodiesel fuels. Combust. Flame, 159, 1876–1893. Feng, Q., Wang, Y.L., Egolfopoulos, F.N., and Tsotsis, T.T. 2010. Fundamental study of the oxidation characteristics and pollutant emissions of model biodiesel fuels. Ind. Eng. Chem. Res., 49, 10392–10398. Feng, Q. 2011. An experimental and modeling study of NO x and soot emissions from Biodiesel and its surrogates. Ph.D. Thesis, University of Southern California. Ferguson, F.T., and Nuth III, J.A. 2008. Vapor pressure of Silicon Monoxide. J. Chem. Eng. Data, 53, 2824–2832. Finocchio, E., Garuti, G., Baldi, M., and Busca, G. 2008. Decomposition of Hexamethylcyclotrisiloxane over solid oxides. Chemosphere, 72, 1659–1663. Formenti, M., Juillet, F., Meriaudeau, P., Teichner, S.J., and Vergnon, P. 1972. Preparation in a hydrogen-oxygen flame of ultrafine metal oxide particles – oxidative properties toward hydrocarbons in presence of ultraviolet radiation. J. Colloid Interface Sci., 39, 79–89. 118 Friede, B., and Jansen, M. 1996. Some comments on so-called ‘silicon monoxide’. J. Non-Cryst. Solids, 204, 202–203. Fukutani, S., Uodome, Y., Kunioshi, N., and Jinno, H. 1991a. Combustion reactions in silane-air flames I. Flat premixed flames. Bull. Chem. Soc. Jpn., 64, 2328– 2334. Fukutani, S., Kunioshi, N., Uodome, Y., and Jinno, H. 1991b. Combustion reactions in silane-air flames II. Counterflow diffusion flame. Bull. Chem. Soc. Jpn., 64, 2335–2340. Geld, P.V., and Kochnev, M.I. 1948. Equilibria of systems involving silicon monoxide. Zh. Prikl. Khim, 21, 1249–1260. Goumri, A., Yuan, W.J., Ding, L.Y., Shi, Y.C., and Marshall, P. 1993. Experimental and theoretical studies of the reaction of atomic hydrogen with silane. Chem. Phys., 177, 233–241. Gulbransen, E.A., Andrew, K.F., and Brassart, F.A. 1966. Oxidation of Silicon at high temperatures and low pressure under flow conditions and the vapor pressure of Silicon. J. Electrochem. Soc., 113, 834–837. Gunther, K.G. 1958. On the measurement of the vapor pressure and evaporation rate of glass-forming substances. Glastechn. Ber., 31, 9–15. Gurav, A., Kodas, T., Pluym, T., and Xiong, Y. 1993. Aerosol processing of materials. Aerosol Sci. Technol., 19, 411–452. Hardesty, D.R., and Weinberg, F.J. 1973. Electrical control of particulate pollutants from flames. Symp. (Int.) Combust., 14, 907–918. 119 Hartman, J.R., Famil-Ghiriha, J., Ring, M.A., and O’Neal, H.E. 1987. Stoichiometry and possible mechanism of SiH 4 -O 2 explosions. Combust. Flame, 68, 43–56. Heitor, M.V., and Moreira, A.L.N. 1993. Thermocouples and sample probes for combustion studies. Prog. Energy Combust. Sci., 19, 259–278. Honig, R.E. 1954. Sublimation studies of Silicon in the mass spectrometer. J. Chem. Phys., 22, 1610–1611. Honig, R.E. 1957. Vapor pressure data for the more common elements. RCA Rev., 18, 195–204. Hung, C.H., and Katz, J.L. 1992. Formation of mixed oxides powders in flames: Part I. TiO 2 -SiO 2 . J. Mater. Res., 7, 1861–1869. Hurd, A.J., and Flower, W.L. 1988. In-Situ growth and structure of fractal Silica aggregates in a flame. J. Colloid Interface Sci., 122, 178–192. Jachimowski, C.J., and McLain, A.G. 1983. A chemical kinetic mechanism for the ignition of silane/hydrogen mixtures, NASA TP 2129. Jalali, A., Motamedhashemi, M.M.Y., Egolfopoulos, F.N., and Tsotsis, T.T. 2013. Fate of Siloxane impurities during the combustion of renewable natural gas. Combust. Sci. Technol., 185, 953–974. Kammler, H.K., Madler, L., and Pratsinis, S.E. 2001. Flame synthesis of nanoparticles. Chem. Eng. Technol., 24, 583–596. Kee, R.J., Miller, J.A., Evans, G.H., and Dixon-Lewis, G. 1989b. A computational model of the structure and extinction of strained, opposed flow, premixed methane-air flames. Symp. (Int.) Combust., 22, 1479–1494. 120 Kee, R.J., Warnatz, J., and Miller, J.A. 1983. FORTRAN computer code package for the evaluation of gas-phase viscosities, conductivities, and diffusion coefficients. Sandia Report SAND83-8209, Sandia National Laboratories, Livermore, CA. Kee, R.J., Rupley, F.M., and Miller, J.A. 1989a. CHEMKIN-II: A FORTRAN chemical kinetics package for the analysis of gas-phase chemical kinetics. Sandia Report SAND89-8009, Sandia National Laboratories, Albuquerque, NM, and Livermore, CA. Kent, J.H. 1970. A noncatalytic coating for platinum-rhodium thermocouples. Combust. Flame, 14, 279–281. Koch, W., and Friedlander, S.K. 1990. The effect of particle coalescence on the surface area of a coagulating aerosol. J. Colloid Interface Sci., 140, 419–427. Koda, S., and, Fujiwara, O. 1988. Silane combustion in an opposed jet diffusion flame. Symp. (Int.) Combust., 21, 1861–1867. Koda, S. 1992. Kinetic aspects of oxidation and combustion of silane and related compounds. Prog. Energy Combust. Sci., 18, 513–528. Kondo, S., Tokuhashi, K., Nagai, H., Takahashi, A., Kaise, M., Sugie, M., Aoyagi, M., Mogi, K., and Minamino, S. 1997. Ab Initio energetic calculations of elementary reactions relevant to low temperature silane oxidation by Gaussian-2 theory. J. Phys. Chem. A, 101, 6015–6022. Kubaschewski O., and Chart, T.G. 1974. Silicon monoxide pressures due to the reaction between solid silicon and silica. J. Chem. Thermodyn., 6, 467–476. Law, C.K., Zhu, D.L., and Yu, G. 1988. Propagation and extinction of stretched premixed flames. Symp. (Int.) Combust., 21, 1419–1426. 121 Libby, P.A., and Blake, T.R. 1981. Burning Carbon particles in the presence of water vapor. Combust. Flame, 41, 123–147. Ma, H.K., Zhao, E., Yeh, C.L., and Chung, K.M. 2003. The formation of nano-size SiO 2 thin film on an aluminum plate with Hexamethyldisilazane (HMDSA) and Hexamethyldisiloxane (HMDSO). J. Therm. Sci., 12, 89–96. Matsui, T., and Imamura, S. 2010. Removal of siloxane from digestion gas of sewage sludge. Bioresour. Technol., 101, S29–S32. McBean, E.A. 2008. Siloxanes in biogases from landfills and wastewater digesters. Can. J. Civ. Eng., 35, 431–436. McEnally, C.S., Koylu, U.O., Pfefferle, L.D., and Rosner, D.E. 1997. Soot volume fraction and temperature measurements in laminar nonpremixed flames using thermocouples. Combust. Flame, 109, 701–720. Miller, T.A., Wooldridge, M.S., and Bozzelli, J.W. 2004. Computational modeling of the SiH 3 + O 2 reaction and silane combustion. Combust. Flame, 137, 73–92. Mizomoto, M., Asaka, Y., Ikai, S., and Law, C.K. 1985. Effects of preferential diffusion on the burning intensity of curved flames. Symp. (Int.) Combust., 20, 1933–1939. Montanari, T., Finocchio, E., Bozzano, I., Garuti, G., Giordano, A., Pistarino, C., and Busca, G. 2010. Purification of landfill biogases from siloxanes by adsorption: A study of silica and 13X zeolite adsorbents on Hexamethylcyclotrisiloxane separation. Chem. Eng. J., 165, 859–863. 122 Murakami, Y., Koshi, M., Matsui, H., Kamiya, K., and Umeyama, H. 1996. Kinetics of the SiH 3 + O 2 reaction: A new transition state for SiO production. J. Phys. Chem., 100, 17501–17506. Murphy, J.D., and McKeogh, E. 2006. The benefits of integrated treatment of wastes for the production of energy. Energy, 31, 294–310. Nair, N., Zhang, X., Gutierrez, J., Chen, J., Egolfopoulos, F., and Tsotsis, T. 2012. On the impact of siloxane impurities on the performance of an engine operating on renewable natural gas. Ind. Eng. Chem. Res., 51, 15786–15795. Nair, N., Vas, A., Zhu, T., Sun, W., Gutierrez, J., Chen, J., Egolfopoulos, F., and Tsotsis, T.T. 2013. Effect of Siloxanes contained in Natural Gas on the operation of a residential furnace. Ind. Eng. Chem. Res., 52, 6253–6261. Neudorfl, P., Jodhan, A., and Strausz, O.P. 1980. Mechanism of the Thermal decomposition of monosilane. J. Phys. Chem., 84, 338–339. Newman, C.G., O’Neal, H.E., Ring, M.A., Leska, F., and Shipley, N. 1979. Kinetics and mechanism of the silane decomposition. Int. J. Chem. Kinet., 11, 1167– 1182. Ohannessian, A., Desjardin, V., Chatain, V., and Germain, P. 2008. Volatile organic silicon compounds: The most undesirable contaminants in biogases. Water Sci. Technol., 58, 1775–1781. Oshita, K., Ishihara, Y., Takaoka, M., Takeda, N., Matsumoto, T., Morisawa, S., and Kitayama, A. 2010. Behavior and adsorptive removal of siloxanes in sewage sludge biogas. Water Sci. Technol., 61, 2003–2012. Pacey, J.G., Doorn, M., and Thorneloe, S.A. 1994. Landfill gas energy utilization: technical and non-technical considerations. SWANA 17 th Annual Landfill Gas 123 Symposium Proceedings, GR-LG 0017, Solid Waste Association of North America, Long Beach, CA, 237–249. Peterson, R.C., and Laurendeau, N.M. 1985. The emittance of yttrium-beryllium oxide thermocouple coating. Combust. Flame, 60, 279–284. Popat, S.C., and Deshusses, M.A. 2008. Biological removal of siloxanes from landfill and digester gases: Opportunities and challenges. Environ. Sci. Technol., 42, 8510–8515. Pratsinis, S.E. 1998. Flame aerosol synthesis of ceramic powders. Prog. Energy Combust. Sci., 24, 197–219. Prosser, R. 2010. Ultraviolet photodecomposition of siloxane. Presented at the Global Waste Management Symposium, San Antonio, TX, October 3–6. Ricaurte Ortega, D., and Subrenat, A. 2009. Siloxane treatment by adsorption into porous materials. Environ. Technol., 30, 1073–1083. Rocabois, P., Chatillon, C., and Bernard, C. 1992. Vapor pressure and evaporation coefficient of SiO (amorphous) and SiO 2 (s) + Si(s) mixtures by the multiple Knudsen cell mass spectrometric method. Rev. Int. Hautes Temper. Refract., 28, 37–48. Sanogo, O., and Zachariah, M.R. 1997. Kinetic studies of the reaction of Tetraethoxysilane (TEOS) with oxygen atoms. J. Electrochem. Soc., 144, 2919–2923. Sanogo, O., and Zachariah, M.R. 2007. Fast-flow reactor study of the thermal decomposition of the Octamethylcyclosiloxane (D 4 ). C. R. Chim., 10, 518– 523. 124 Schick, H.L. 1960. A thermodynamic analysis of the high-temperature vaporization properties of Silica. Chem. Rev., 60, 331–362. Schmude Jr., R.W. 1994. Small clusters of Carbon, Silicon, Germanium, and Tin. Ph.D. Thesis, Texas A&M University. Schnurre, S.M., Grobner, J., and Schmid-Fetzer, R. 2004. Thermodynamics and phase stability in the Si-O system, J. Non-Cryst. Solids, 336, 1–25. Schweigkofler, M., and Niessner, R. 2001. Removal of siloxanes in biogases. J. Hazard. Mater., 83, 183–196. Shaddix, C.R. 1999. Correcting thermocouple measurements for radiation loss: A critical review. In Proceedings of the 33rd National Heat Transfer Conference, Paper HTD99-282, Albuquerque, NM, August 15-17. Shandross, R.A., Longwell, J.P., and Howard, J.B. 1991. Noncatalytic thermocouple coating for low-pressure flames. Combust. Flame, 85, 282–284. Shin, H.C., Park, J.W., Park, K., and Song, H.C. 2002. Removal characteristics of trace compounds of landfill gas by activated carbon adsorption. Environ. Pollut., 119, 227–236. Sommerlade, R., Parlar, H., Wrobel, D., and Kochs, P. 1993. Product analysis and kinetics of the gas-phase reactions of selected organosilicon compounds with OH radicals using a smog chamber-mass spectrometer system. Environ. Sci. Technol., 27, 2435–2440. Souchiere, J.L., and Binh, V.T. 1986. On the evaporation rate of silicon. Surf. Sci., 168, 52–58. 125 Tan, C.Z., and Arndt, J. 2000. Temperature dependence of refractive index of glassy SiO 2 in the infrared wavelength range. J. Phys. Chem. Solids, 61, 1315–1320. Tomooka, T., Shoji, Y., and Matsui, T. 1999. High temperature vapor pressure of Si. J. Mass Spectrom. Soc. Jpn., 47, 49–53. Touloukian, Y.S., and DeWitt, D.P., Purdue University, Lafayette, Ind., Thermophysical Properties Research Center (TPRC). 1972. Thermal Radiative Properties: Nonmetallic Solids. In Ho, C.Y. (Ed.), Thermophysical Properties of Matter – The TPRC Data Series, IFI/Plenum, New York – Washington, Vol. 8, pp. 371–426. Ulrich, G.D. 1971. Theory of particle formation and growth in oxide synthesis flames. Combust. Sci. Technol., 4, 47–57. Ulrich, G.D., and Riehl, J.W. 1982. Aggregation and growth of submicron oxide particles in flames. J. Colloid Interface Sci., 87, 257–265. Ulrich, G.D., and Subramanian, N.S. 1977. Particle growth in flames, III. Coalescence as a rate-controlling process. Combust. Sci. Technol., 17, 119–126. Ulrich, G.D. 1984. Flame synthesis of fine particles. Chem. Eng. News, 62, 22–29. Urban, W., Lohmann, H., and Gomez, J.J.S. 2009. Catalytically upgraded landfill gas as a cost-effective alternative for fuel cells. J. Power Sources, 193, 359–366. Wang, H., Du, D.X., Sung, C.J., and Law, C.K. 1996. Experiments and numerical simulation on soot formation in opposed-jet ethylene diffusion flames. Symp. (Int.) Combust., 26, 2359–2368. Wang, H., and Frenklach, M. 1994. Transport properties of polycyclic aromatic hydrocarbons for flame modeling. Combust. Flame, 96, 163–170. 126 Wang, H., You, X., Joshi, A.V., Davis, S.G., Laskin, A., Egolfopoulos, F.N., and Law, C.K. 2007. USC Mech Version II. High-temperature combustion reaction model of H 2 /CO/C 1 -C 4 compounds. Available at: http://ignis.usc.edu/USC_Mech_II.htm. Wegner, K., and Pratsinis, S.E. 2005. Gas-phase synthesis of nanoparticles: Scale-up and design of flame reactors. Powder Technol., 150, 117–122. Whelan, M.J., Estrada, E., and van Egmond, R. 2004. A modeling assessment of the atmospheric fate of volatile methyl Siloxanes and their reaction products. Chemosphere, 57, 1427–1437. Wheless, E., and Gary, D. 2002. Siloxanes in landfill and digester gas. Presented at the Twenty-Fifth Annual SWANA Landfill Gas Symposium, Monterey, CA, March 25–28. Wooldridge, M.S. 1998. Gas-phase combustion synthesis of particles. Prog. Energy Combust. Sci., 24, 63–87. Wu, C.K., and Law, C.K. 1985. On the determination of laminar flame speeds from stretched flames. Symp. (Int.) Combust., 20, 1941–1949. Yeh, C.L., Zhao, E., and Ma, H.K. 2001a. An experimental investigation of combustion synthesis of silicon dioxide (SiO 2 ) particles in premixed flames. Combust. Sci. Technol., 173, 25–46. Yeh, C.L., Zhao, E., and Ma, H.K. 2001b. Combustion synthesis of SiO 2 on the aluminum plate. J. Therm. Sci., 10, 92–96. 127 Zachariah, M.R., and Semerjian, H.G. 1990. Experimental and numerical studies on refractory particle formation in flames – application to silica growth. High Temp. Sci., 28, 113–125. Zachariah, M.R., Chin, D., Semerjian, H.G., and Katz, J.L. 1989. Silica particle synthesis in a counterflow diffusion flame reactor. Combust. Flame, 78, 287– 298. Zachariah, M.R., Tsang, W. 1995. Theoretical calculation of thermochemistry, energetics, and kinetics of high temperature Si x H y O z reactions. J. Phys. Chem., 99, 5308–5318. Zhu, W.H., and Pratsinis, S.E. 1997. Synthesis of SiO 2 and SnO 2 particles in diffusion flame reactors. AIChE J., 43, 2657–2664.
Abstract (if available)
Abstract
Biogas, which is produced from sludge biodegradation in wastewater treatment plants (WWTP), and landfill gas (LFG), which is generated from solid waste in landfills, are potentially important renewable fuels. Gas turbines and conventional internal combustion engines can combust LFG (or biogas) to generate electricity. Aside from their main components, such as methane and carbon dioxide, biogas and LFG may also contain undesirable contaminants. A particularly bothersome such trace constituent is a class of compounds known as siloxanes. The corrosion and damage that can be caused by these impurities may reduce the operating life of power and electricity generation equipment. In this research, the decomposition of siloxanes present in simulated renewable natural gas (RNG), which is LFG (or biogas) after its methane content has been upgraded to meet natural gas (NG) pipeline standards, is experimentally investigated in order to provide a better knowledge of their fate during RNG combustion. In the study, individual siloxanes and their mixtures in trace amounts were introduced into RNG flat flames. The counter-flow experimental technique was utilized which allows one to accurately probe and analyze the mechanisms leading to the formation of silica micro-particles resulting from the decomposition of the siloxanes. ❧ In order to probe the nature of deposits formed on surfaces in proximity to the RNG flames, flat Ni/Cr metal strips were placed downstream of a Bunsen flame. The chemical composition, morphology, and structure of the solid particles formed were investigated by scanning electron microscopy (SEM), energy dispersive analysis by X-ray (EDAX), and atomic force microscopy (AFM). It was found that microcrystalline silica particles were generated during RNG combustion, leading to a rapid coverage of the surface of Ni/Cr strips placed in the flame environment, and forming eventually a light white layer of solid particles. The size of these particles was estimated using SEM/EDAX. ❧ The effect of siloxane concentration on its conversion along the flame was studied via the use of the GC-MS technique for a number of different feed concentrations. Volume fractions of particles within the flame were measured via the laser extinction method indicating a linear relationship between the concentration in the fuel and the corresponding volume of particles in the flue-gas. The temperature profile was measured experimentally to help identify the kinetics of the burning of the siloxane compounds. This fundamental insight is important in terms of being able to accurately determine the maximum allowable siloxane content for biogas that is safe to use without leading to deposits and micro-particle formation.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Mesoscale SOFC-based power generator system: modeling and experiments
PDF
Studies of combustion characteristics of heavy hydrocarbons in simple and complex flows
PDF
Studies of methane counterflow flames at low pressures
PDF
Lab-scale and field-scale study of siloxane contaminants removal from landfill gas
PDF
Experimental and kinetic modeling studies of flames of H₂, CO, and C₁-C₄ hydrocarbons
PDF
Experimental studies of high pressure combustion using spherically expanding flames
PDF
Flame ignition studies of conventional and alternative jet fuels and surrogate components
PDF
Modeling investigations of fundamental combustion phenomena in low-dimensional configurations
PDF
Novel methods for landfill gas and biogas clean-up
PDF
Accuracy and feasibility of combustion studies under engine relevant conditions
PDF
Re-assessing local structures of turbulent flames via vortex-flame interaction
PDF
A theoretical study of normal alkane combustion
PDF
Studies of Swiss-roll combustors for incineration and reforming applications
PDF
Investigations of fuel effects on turbulent premixed jet flames
PDF
Studies on the flame dynamics and kinetics of alcohols and liquid hydrocarbon fuels
PDF
CFD design of jet-stirred chambers for turbulent flame and chemical kinetics experiments
PDF
Development of a novel heterogeneous flow reactor: soot formation and nanoparticle catalysis
PDF
Design, dynamics, and control of miniature catalytic combustion engines and direct propane PEM fuel cells
PDF
Performance prediction, state estimation and production optimization of a landfill
PDF
Catalytic methane ignition over freely-suspended palladium nanoparticles
Asset Metadata
Creator
Jalali, Mir Aydin
(author)
Core Title
Studies of siloxane decomposition in biomethane combustion
School
Viterbi School of Engineering
Degree
Doctor of Philosophy
Degree Program
Mechanical Engineering
Publication Date
11/20/2013
Defense Date
10/24/2013
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
combustion,deposition,OAI-PMH Harvest,oxidation kinetics,silane,siloxanes,solid silica particles
Format
application/pdf
(imt)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Egolfopoulos, Fokion N. (
committee chair
), Ronney, Paul D. (
committee member
), Sadhal, Satwindar S. (
committee member
), Tsotsis, Theodore T. (
committee member
)
Creator Email
mjalali@usc.edu
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c3-348379
Unique identifier
UC11295963
Identifier
etd-JalaliMirA-2167.pdf (filename),usctheses-c3-348379 (legacy record id)
Legacy Identifier
etd-JalaliMirA-2167.pdf
Dmrecord
348379
Document Type
Dissertation
Format
application/pdf (imt)
Rights
Jalali, Mir Aydin
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Tags
combustion
deposition
oxidation kinetics
silane
siloxanes
solid silica particles