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
/
Novel methods for landfill gas and biogas clean-up
(USC Thesis Other)
Novel methods for landfill gas and biogas clean-up
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
NOVEL METHODS FOR LANDFILL GAS AND BIOGAS CLEAN-UP by Nitin Narayanan Nair 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 (CHEMICAL ENGINEERING) August 2013 Copyright 2013 Nitin Narayanan Nair i Dedication This Thesis is dedicated to my father Mr. Narayanan Nair and my mother Mrs. Leena Nair and also to the scientific community currently in pursuit of clean sources of energy so that we can protect our only home, this planet Earth. ii Acknowledgments I would like to express my sincere regards to my advisors Professor Theodore T. Tsotsis and Professor Fokion Egolfopoulos for giving me a chance to do my PhD under their valuable supervision. This dissertation is a result of their help, support, and patience. I also wish to thank Prof. Katherine Shing for serving on my qualifying and dissertation committees. This Thesis could not also be finished without the help of Dr. Mirmohammedyousef Motamedhhashemi, who was with me from the beginning of my career at USC and assisted me in many aspects of my research; Dr. Ryan Mourhatch who taught me how to use the GC/MS technique, Dr. Aadesh Harale who taught me how to do membrane permeation experiments, and Ms. Xiaojie Yan who took the TEM images, and who carried out the He pycnometry measurements. I would also like to thank all my colleagues and friends in our research group who helped me in different ways: Mr. Alireza Divaslar, Mr. Arjun Vas, Mr Tongyu Zhu, Mr. Wenjing Sun, Mr. Aydin Jalali, Ms. Sahar Soltani, Ms. Basabdatta Roychaudhuri, Mr. Wangxue Deng, Ms. Malak Khojasteh, and many others. A special thanks to the colleagues from the Southern California Gas Company, Mr. Jack Chen, Mr . Jorge Gutierrez, Ms. May Lew, Dr. Siari Sosa and Mr. George. I would also like to thank my colleagues from GC Environmental Inc. Mr. Richard Prosser, Dr. Jerry Ren and Mr. Kambiz Zortani. iii Many thanks also go to Mrs. Tina Silva, the Department’s Instructional Lab Manager, and Mr. Shokry Bastorous, the Instructional Lab Assistant, for helping me in every way possible throughout these past four years. Special gratitude is extended to the administrative staff of the Mork Family Department of Chemical Engineering and Materials Science, Ms. Karen Woo, Mr. Martin Olekszyk, Ms. Heather Alexander, Ms. Aimee Barnard, and Ms. Angeline Fugelso for all their help and support throughout my graduate studies. Above all, I would like to thank my parents, Narayanan P Nair and Leena Nair, my brother Nived Nair and my entire extended family for their patience, encouragement and support all through during my Ph.D. studies. This research was made possible primarily by financial support from the NSF. Additional technical support came from the Media and Process Technology, Inc., the Southern California Gas Company, and GC Environmental, which are all gratefully acknowledged. iv Table of Contents Dedication……. ................................................................................................................... i Acknowledgments............................................................................................................... ii List of Tables….. ............................................................................................................... vi List of Figures…. ............................................................................................................. viii Nomenclature…. ................................................................................................................ xi Abstract………. ............................................................................................................... xiii Chapter 1 Motivation and Background ..................................................................... 1 1.1 Landfills and Landfill Gas Utilization ................................................................................... 2 1.2 Current Landfill Gas Clean-Up Technologies ....................................................................... 5 1.3 Flow-Through Catalytic Membrane Reactors ..................................................................... 28 1.3.1 Complete Conversion Integral FTCMR ....................................................................... 30 1.3.2 Selective Integral FTCMR ........................................................................................... 38 1.3.3. Selective Differential FTCMR .................................................................................... 52 1.4 The Siloxane Problem ......................................................................................................... 59 1.5 Our Study ............................................................................................................................ 61 1.5.1. Use of FTCMR for Removal of Chlorinated and Sulfided Contaminants in LFG ...... 61 1.5.2. Study the Impact of Siloxanes in LFG on Common Appliances such as Engines and Furnaces ................................................................................................................................ 68 1.5.3. Use of UV Photodecomposition for the Removal of Siloxanes .................................. 69 Chapter 2 A Flow-Through Catalytic Membrane Reactor for Landfill Gas Clean-up 71 2.1. Introduction ........................................................................................................................ 72 2.2. Experimental Section .......................................................................................................... 79 2.3. Experimental Results and Discussion................................................................................. 87 2.6. Conclusions ...................................................................................................................... 112 Chapter 3 The Impact of Siloxane Impurities on Engines Operating on Renewable Natural Gas ........................................................................................... 113 3. 1 Introduction ...................................................................................................................... 114 3.2 Experimental Procedures ................................................................................................... 121 3.3 Results and Discussion ...................................................................................................... 129 3.4. Conclusions ...................................................................................................................... 149 Chapter 4 Effect of Siloxanes Contained in Natural Gas on the Operation of a Residential Furnace ............................................................................... 150 v 4.1. Introduction ...................................................................................................................... 151 4.2. Experimental Set-up ......................................................................................................... 156 4.3. Results and Discussion ..................................................................................................... 162 4.4. Concluding Remarks ........................................................................................................ 176 Chapter 5 A Novel Technique for the Removal of Siloxanes from Landfill Gas using UV Photodecomposition ............................................................. 177 5.1. Introduction ...................................................................................................................... 178 5.2. Experimental Results and Discussion............................................................................... 184 5.3 Conclusions ....................................................................................................................... 193 Chapter 6 Ideas for Future Work ........................................................................... 194 References……. .............................................................................................................. 197 vi List of Tables Table 1.1: Literature review of the catalytic oxidation of chlorinated compounds .......... 17 Table 1.2: Literature review of the catalytic oxidation of sulfided compounds ............... 22 Table 1.3: Literature review of the catalytic oxidation of fluorinated compounds .......... 24 Table 1.4: Catalytic activity of catalyst A towards the removal of model VOC compounds. ...................................................................................................... 26 Table 1.5: Catalytic activity of catalyst B towards removal of model minor compounds.27 Table 1.6: Literature review of complete conversion integral FTCMR ........................... 31 Table 1.7: Literature review of selective integral FTCMR ............................................... 40 Table 1.8: Literature review of selective differential FTCMR ......................................... 53 Table 2.1: Table showing the different membranes prepared and their uses .................... 83 Table 2.2: Table showing the Simulated Landfill Gas (SLFG) components and their composition ...................................................................................................... 84 Table 2.3: Table showing the parameters in the permeance equation for the EM1 (unimpregnated) and CM2 (catalytic) membrane. .......................................... 92 Table 2.4: Table showing the parameters in the permeance equation for the CM6 membrane. ........................................................................................................ 93 Table 2.5: CO Chemisorption results for the CM4 membrane ......................................... 98 Table 3.1: Siloxanes limits in LFG and digester gas recommended by various engine manufacturers. ................................................................................................ 117 Table 3.2: Siloxane compounds in LFG and some of its properties from Jalali et al. [2013] ............................................................................................................. 125 Table 3.3: Oil analysis data for two used oil samples from the siloxane and non-siloxane engines............................................................................................................ 146 Table 5.1: Absorption cross-section of different LFG components ............................... 184 Table 5.2: Conversion of L2 and D4 (5ppm+5ppm) for different energy lamps ........... 186 vii Table 5.3: Conversion of chlorinated and sulfided compounds for different energy lamps ........................................................................................................................ 187 viii List of Figures Figure 1.1: Comparison of the FTCMR concept with the granular fixed-bed and monolithic reactor technologies for VOC destruction. .................................... 65 Figure 1.2: Left, conversion of toluene in a FTCMR as a function of temperature; right conversion of 1,200 ppm MEK and comparison between a FTCMR and a monolithic reactor. ........................................................................................... 67 Figure 2.1: Experimental set-up of the permeation experiments ...................................... 81 Figure 2.2: Experimental set-up of the reaction experiments ........................................... 85 Figure 2.3: (a) Permeance results for He for an unimpregnated membrane. (b) Permeance results for Ar for an unimpregnated membrane ............................................... 88 Figure 2.4: (a) Permeance results for He for catalytic membrane. (b) Permeance results for Ar for catalytic membrane. (c)Knudsen % flow for catalytic membrane .. 90 Figure 2.5: SEM picture of a three-layer membrane (EM1) showing the two top membrane layers (0.01 micron and 0.05 micron pore size) and the support ... 94 Figure 2.6: BSE image for the CM7 membrane showing the Pt catalyst particels deposited in the membrane............................................................................... 95 Figure 2.7: EDX analysis showing the distribution of Pt along the thickness of the CM3 membrane ......................................................................................................... 96 Figure 2.8: TEM image of the Pt/Al 2 O 3 membrane .......................................................... 97 Figure 2.9: FTCMR light-off temperature results for membrane CM1 .......................... 100 Figure 2.10: FTCMR vs. monolith results for CM1 membrane ..................................... 102 Figure 2.11: FTCMR results for various LFG components for different flow-rates for membrane CM2 .............................................................................................. 104 Figure 2.12: FTCMR vs. Monolith comparison for SLFG components for membrane CM2 ............................................................................................................... 106 Figure 2.13: FTCMR deactivation test results for different runs for CM2 ..................... 109 Figure 2.14: Monolith deactivation test for membrane CM2 ......................................... 111 ix Figure 3.1: Top, laboratory engine with the catalyst monolith attached; bottom: catalyst monolith connected to the engine muffler. .................................................... 122 Figure 3.2: Top, schematic of the experimental apparatus for the siloxane engine; bottom, schematic of the experimental apparatus for the non-siloxane engine. ......... 123 Figure 3.3: (a) CO and CH 4 flue-gas concentrations for the non-siloxane engine. NO, (b)NO 2 and NO x flue-gas concentrations for the non-siloxane engine. ......... 130 Figure 3.4: (a) CO and CH 4 flue-gas concentrations for the siloxane engine. (b) NO, NO 2 and NO x flue-gas concentrations for the siloxane engine. ............................. 132 Figure 3.5: CO conversion vs. time on stream for the non-siloxane engine ................... 133 Figure 3.6: CO conversion vs. time on stream and volume of siloxane injected for the siloxane engine. .............................................................................................. 134 Figure 3.7: Silica profile on the catalyst monolith surface along its length. .................. 135 Figure 3.8: BET data of the catalyst monolith for both the siloxane and non-siloxane engines............................................................................................................ 136 Figure 3.9: SEM image of the spark-plug metal surface for the siloxane engine coated by a silica film. .................................................................................................... 138 Figure 3.10: EDX of the metal surface of the non-siloxane engine spark-plug (top) and of the silica coated spark-plug surface for the siloxane engine (bottom). .......... 139 Figure 3.11: Sensor voltage signals. ............................................................................... 141 Figure 3.12: SEM-EDX results of the siloxane engine oxygen sensor........................... 143 Figure 3.13: Photograph of the piston head of the siloxane engine showing the silica film deposition. ...................................................................................................... 144 Figure 3.14: EDX results of the siloxane engine valve. ................................................. 145 Figure 3.15: Particle size distribution in the flue-gas. .................................................... 148 Figure 4.1: (a) L2 siloxane. (b) D4 siloxane. .................................................................. 154 Figure 4.2: Furnace ......................................................................................................... 157 Figure 4.3: Flame sensor ................................................................................................. 158 Figure 4.4: Experimental configuration. ......................................................................... 159 x Figure 4.5: Sensor current vs. time on stream for various siloxane concentrations, ppm v . ........................................................................................................................ 162 Figure 4.6: (a) Sensor current vs. moles of siloxane fed to the furnace for a siloxane concentration of 20 ppm v ; the experiment was run twice each time with a different sensor. (b) Sensor current vs. moles of siloxane fed to the furnace for two different siloxane concentrations of 2 ppm v and 10 ppm v each tested with a different sensor. .............................................................................................. 165 Figure 4.7: SEM image of the tip of the flame sensor showing the silica layer (indicated by the double-arrow on the left). .................................................................... 167 Figure 4.8: EDX analysis of the surface of the tip of the flame sensor showing the silica layer thickness. ............................................................................................... 168 Figure 4.9: Schematic diagram of the furnace, showing the parts, which were analyzed by EDX for the presence of silica deposits. ........................................................ 169 Figure 4.10: (a) Deposits on the condenser coil assembly of the furnace. (b) EDX analysis of the deposits on the condenser assembly. ................................................... 171 Figure 4.11: (a) EDX analysis of the deposits on the top-half of the combustion chamber. (b) EDX analysis of the deposits on the tubing farthest from the combustion chamber. ......................................................................................................... 173 Figure 4.12: EDX analysis showing kilo counts of elements of the dried residue from the furnace water condensate. .............................................................................. 174 Figure 4.13: Particle size distribution in the flue-gas from the furnace operating with and without siloxanes. ........................................................................................... 175 Figure 5.1: Absorption spectrum of some silicon-containing compounds ..................... 183 Figure 5.2: MiniRAE 2000 showing the inlet and outlet ports ....................................... 185 Figure 5.3: (a) UV lmap. (b) UV lamp enclosed in a PVC pipe and modified for reactor operations ....................................................................................................... 188 Figure 5.4: Experimental set-up ...................................................................................... 190 Figure 5.5: Conversion of 10ppm L2 and 10ppm D4 in LFG2 gas for different flow-rates ........................................................................................................................ 191 Figure 5.6: Effect of flow-rate on the conversion of VOC compounds in LFG ............. 192 Figure 6.1: Schematic drawing of a LFG clean-up process with SiO 2 separation .......... 195 xi Nomenclature ɛ Membrane porosity τ Memerane tortuosity μ Gas viscosity (Pa.s) R Ideal gas constant(J gmol -1 K -1 ) T Temperature(K) p Partial pressure(Pa) y mole fraction N flux (mol s -1 m -2 ) D ij Binary molecular diffusion coefficient (m 2 s -1 ) D ik Knudsen diffusion coefficient for i component (m 2 s -1 ) P Total pressure(Pa) d p Pore diameter (m) Rxn Reaction rate for the species i (mol s - 1 m -3 ) Mi Molecular weight of the species i (kg kmol -1 ) B e Membrane geometrical prameter r radial distance (m) z Longitudinal distance (m) xii Subscripts L Main LFG without contaminants c LFG component i Component i in inner out out Superscripts e Effective 0 Inlet conditions t tube-side s shell-side xiii Abstract In this study a novel catalytic oxidation technology appropriate for landfill gas (LFG) clean-up based on the flow-through catalytic membrane reactor (FTCMR) concept has been studied. For the experiments, a model LFG stream has been used with a volatile organic compound (VOC) composition which was shown previously by the authors to simulate well the behavior of real LFG in field-scale investigations. Asymmetric tubular alumina membranes were used in the research and were rendered catalytic by wet impregnation. Their pore-structure characteristics were measured with single-gas permeation tests, as they are important in determining the transport mechanisms of the VOC through the catalytically active membrane layer. When comparing the FTCMR with the more conventional reactors the “yardstick” of success is the ability of the FTCMR to operate under lower temperature for a given level of conversion, and/or attain higher conversion under the same conditions and catalyst loading. For the LFG application, light-off temperature experiments showed promising results when compared to the monolith reactor. Also, no catalyst deactivation was observed during the time-on-stream experiments, proving that the FTCMR is robust towards corrosive by-products (e.g., HCl) produced during the oxidation reactions. Siloxanes are another major class of compounds detected in LFG. This Thesis is also a study on the impact of siloxanes on various types of equipment using LFG which is not treated for siloxanes. Specifically, in this study an internal combustion engine and a residential furnace operating on natural gas (NG) spiked with siloxanes have been studied xiv experimentally with the goal of understanding the impact of siloxane impurities on their performance. These impurities are shown to completely decompose during NG combustion in the engine to form silica microparticulates. These coat the internal metal surfaces in the equipment and severely reduce their efficiency and damage important components, such as furnace flame sensors and engine oxygen sensors. A method to remove these siloxane impurities has also been studied in this thesis based on UV Photodecomposition. Specifically, this Thesis also describes efforts to evaluate the technical feasibility and environmental implications of a novel technology for the treatment of biogas and LFG which involves the in situ conversion of the siloxanes, typically found in the gas, into inert silicon dioxide via a photochemical conversion process. The approach involves using high energy UV light to convert the siloxanes into SiO 2 powder, which can be conveniently removed from the biogas via a downstream filter. The technique is shown to be very effective with high siloxane conversions attained in the laboratory. 1 Chapter 1 Motivation and Background 2 1.1 Landfills and Landfill Gas Utilization The emphasis in this study is on the treatment and further potential beneficiation of landfill gas (LFG). Landfills are among the oldest and most common methods of waste disposal. Even today, landfilling is considered to be the best available worldwide solution for our mounting solid waste disposal problems. In the early years, landfills were considered to have no or minimal impact on the surrounding environment. However, studies conducted by various environmental agencies and other groups in recent years at several landfill sites have shown that landfills, when poorly designed and operated, can in fact have a major impact on the environment. Landfills generate two different effluent streams: a gas stream, known as landfill gas and a liquid stream known as leachate (EPA- 450/3-90-011a, 1991; Baker et al. [1992]). Both these landfill products contain a variety of trace toxic chemical compounds, which are produced by the various physical, biological and chemical waste decomposition processes that occur in the landfills. Leachate is a toxic effluent that must be dealt appropriately before disposal. LFG, on the other hand, is potentially an important renewable fuel, as it typically contains more than 50% methane. The United States Environmental Protection Agency (EPA) estimates that as of June 28 th 2012, there were approximately 594 operational LFG energy projects in the United States and 540 additional landfills that are good candidates for methane extraction. They are also, potentially, the largest source of anthropogenic (man-made) methane emissions in the U.S. According to a survey by EPA, 126.3 teragrams of carbon 3 dioxide equivalent of methane was produced in landfills in the year 2008 [http://www.epa.gov/outreach/lmop/basicinfo/index.html]. Unfortunately, today a large fraction of it is flared in order to control the emissions of methane (a potent greenhouse gas, by some accounts 20 times more powerful than CO 2 ) and to control odor problems arising from sulfur-containing compounds in the landfill gas. The rest is utilized for power generation applications. There are several options for converting LFG into energy. They include: Electricity Generation… The generation of electricity from LFG accounts for up to about two–thirds of the currently operational LFG utilization projects in the United States. Electricity for on–site use or sale to the grid can be generated using a number of different technologies, including internal combustion (IC) engines, turbines, microturbines, and fuel cells. The vast majority of projects use IC (reciprocating) engines or turbines, with microturbine technology being used at smaller landfills, and in niche applications. Technologies such as Stirling and Rankine-cycle engines and fuel cells are still under development [http://www.epa.gov/outreach/lmop/basicinfo/index.html]. Direct Use… Directly using the LFG to offset the use of other fuels (e.g., natural gas, coal, fuel oil) is occurring in about one–third of the current projects. This direct use of LFG can be in a boiler, dryer, kiln, green-house, or in other thermal applications. LFG can also be used directly to evaporate leachate. Reported innovative direct uses include: firing pottery and glass–blowing kilns; powering and heating green-houses and in an ice 4 rink; and heating water for an aquaculture (fish farming) operation. Current industries using LFG for energy production include auto manufacturing, chemical production, food processing, pharmaceuticals, cement and brick manufacturing, wastewater treatment, consumer electronics and products, paper and steel production, and prisons and hospitals, just to name a few. Co-generation…Co-generation (also known as combined heat and power generation or CHP) projects using LFG generate both electricity and thermal energy, usually in the form of steam or hot water. Several co-generation projects have been installed at industrial operations, using both engines and turbines. The efficiency gains of capturing the thermal energy, in addition to electricity generation, can make these projects very attractive economically. [http://www.epa.gov/outreach/lmop/basicinfo/index.html]. Alternate Fuels…Production of alternate fuels from LFG is an emerging technical area. LFG has been successfully delivered (after removal of its impurities and the CO 2 that it contains) to the natural gas pipeline system as a high–BTU fuel. LFG has also been converted to vehicle fuel in the form of compressed natural gas (CNG) and liquefied natural gas (LNG). Projects to convert LFG to methanol are also in the planning stages [http://www.epa.gov/outreach/lmop/basicinfo/index.html]. Before LFG becomes available for use in the aforementioned applications, it must be cleaned-up and appropriately conditioned. Some of the methods currently utilized for LFG clean-up are described below. 5 1.2 Current Landfill Gas Clean-Up Technologies LFG consists primarily of CH 4 and CO 2 , a few per cent of N 2 , O 2 , and H 2 O, and smaller amounts of H 2 , H 2 S, and NH 3 . It also contains many trace organic compounds, other than CH 4 (collectively known as NMOC), several of which contain halogens and sulfur. Among the halogen-containing compounds, one finds chloro/fluoro substituted CH 4 , C 2 H 6 , and C 2 H 4 homologues (including CCl 4 , chloroform, methylene and methyl chloride, dichloro-, thrichloro-, and tetrachloro-ethane, PCE, and vinyl chloride among others), mono- and dichloro-benzene, and other polychlorinated aromatic hydrocarbons [Baker et al., 1992]. The various sulfur compounds, in addition to H 2 S, include carbonyl sulfide (COS), dimethylsulfide (DMS), and various mercaptans. The concentrations of such compounds vary widely among various landfills, from <100 ppb to >several hundred ppm. Even for the same landfill, the LFG composition varies with time, but also among various places in the landfill, reflecting the different range of wastes deposited during its lifetime. A direct way of improving the fuel quality of LFG is by removing the CO 2 that it contains, thus enhancing its methane content. A number of approaches have been utilized for that purpose. For example, Rautenbach and Welsch [1994] used polyamide membranes, and Harasimowicz and Orluk [2007] have used polyimide-type membranes to separate CO 2 from methane in LFG in order to enrich the LFG with CH 4 . Other techniques such as pressure swing adsorption [Cavenati and Grande, 2005] have also been utilized. The use of such conventional approaches dictates that the LFG trace 6 contaminants must be removed prior to bringing it in contact with the membrane and/or adsorbents as both are highly susceptible to such impurities. NMOC removal from LFG, prior to combustion (or clean-up), is a technical challenge. Current state-of-the-art gas treatments (see below) fail to deliver as promised. There are many reasons for that, but process economics is the most critical one, since to date no purification system is able to eliminate all the trace NMOC that are present in LFG, implying the need to implement a number of sequential steps, which target specific substances. Another key consideration is the inability of most of the current purification systems to cope with concentration peaks, which are usually observed in landfill gas for these kinds of pollutants. The following discussion describes briefly some of the common methods currently available for the clean-up of LFG. Adsorption Methods…One of the most common methods of controlling the emissions of trace compounds in LFG is through the use of adsorption. Activated carbon (AC) is the material of choice for such an application. Activated carbons are versatile adsorbents which find frequent use in many industrial gas separation and purification processes because of their large surface area, microporous structure, high adsorption capacity, and high degree of surface activity [Pradhan et al., 1999]. Though adsorption via activated carbon (but also via other adsorbent media) is commonly utilized, it has only proven marginally effective for LFG clean-up. This is because of the broad array of low- concentration NMOC found in LFG, as well as due to the simultaneous presence of other 7 highly adsorbable species (e.g., aromatic hydrocarbons) which by coating the AC surface result in low adsorption capacity for the former compounds, particularly the lower molecular weight (MW) NMOC, (e.g., vinyl chloride, and other chlorofluoromethane compounds). This results in premature adsorber bed saturation. One further problem with adsorption, as well as with other conventional technologies like absorption (see below), is that the NMOC are not destroyed, but simply transferred to and “stored” into the adsorption/absorption media. This presents risks during media regeneration and in the handling and storage of spent media prior to their disposal. In addition, the regenerated media have, often, diminished adsorption capacity, and successive regenerations yield progressively lower efficiencies, until media replacement becomes necessary, which comes at a great cost. Moreover, the operating cost of the regeneration process itself is relatively high and, because of the process complexity, there are multiple opportunities for system failure. It is the costs associated with regeneration that often make adsorption a very expensive method for the clean-up of landfill gas (and biogas, in general). Other materials like silica gel have shown comparable removal capacities (e.g., towards siloxanes, a troublesome LFG impurity) to active carbons with the advantage of being more readily regenerable by thermal treatment [Schweigkofler et al., 2001]. However, due to their hydrophilicity, their adsorption capacities are remarkably reduced in LFG which contains substantial amounts of water. 8 Absorption Methods…Water scrubbing is a reasonably effective and relatively inexpensive method for treating gas streams containing high concentrations of H 2 S, but for lower concentration of H 2 S this technology seems to be less efficient. Though water has a relatively good efficiency for H 2 S removal, still large water volumes are needed, and reboiling is often necessary for regeneration. Water scrubbing is not ideal for complete LFG clean-up since it has very poor removal efficiencies for other common LFG impurities (other than H 2 S), because of their low solubilities in water and high vapor pressures. Absorption methods have been applied for the removal of targeted substances like siloxanes (commonly encountered in LFG) either through the use of concentrated solutions of acids (which have been shown to have moderate removal efficiencies – alkaline solutions are also effective for the removal of siloxanes, however, the presence of CO 2 in LFG makes their use impractical). The application of gas scrubbing with organic solvents has been shown not to be very efficient, particularly for the volatile silicon-containing compounds (e.g., hexamethyldisiloxane (L2) or trimethylsilane) which are not easy to remove from the gas phase [Dewil et al., 2006]. Furthermore, the utilization of organic solvents complicates plant design due to safety concerns, and hence results in high operational costs. Chilling Processes…Cooling systems are usually utilized for LFG treatment in order to remove water, and thus avoid corrosion damage in engine parts. Parallel to water removal other substances, such as heavy aromatics and siloxanes can, at least partially, be removed via chilling. It has been shown that at −25 o C a 25.9% removal of volatile 9 methyl siloxanes (VMS) is possible and that a removal of around 99% can only be achieved by freezing temperatures as low as −70 o C [McBean et al., 2008]. According to Schweigkofler [2001], 89% of the silicon compounds in a dried gas can be removed by a cooling at a temperature of 5 o C. Higher removal efficiencies may be achieved at lower temperatures. Previous studies carried out by Siloxa Engineering AG and the Fraunhofer Institute UMSICHT have shown that freeze-cooling systems for the removal of siloxanes are only cost-effective in combination with active carbon in the scenario of high flow rates with relatively high loads of siloxanes. In general, the energetic requirements for the application of deep chilling systems are so high, as experience for sewage gas treatment indicates, that such a process is generally not economically feasible [Urban et al., 2009]. Catalytic Methods…A process involving the destruction of the NMOC via in situ catalytic reaction with H 2 was developed previously by our group in collaboration with EPRI, and was field-tested at the Anoca landfill in Minnesota [He et al., 1997]. The process utilizes catalytic hydroprocessing with Co-Mo/Al 2 O 3 or Ni-Mo/Al 2 O 3 catalysts in an atmospheric pressure reactor, together with conventional adsorption technology for the removal of the HCl and H 2 S by-products. It has been shown effective in reducing the levels of the Cl-containing and S-containing NMOC, down to their analytical detection limits. Furthermore, no catalyst deactivation was observed during 1000 h of field-testing. However, the amount of H 2 required (~5 vol % of the LFG treated) is undesirable in comparison with the ppm-level of contaminants treated, and is a challenge to generate and handle, particularly for small landfills or in remote locations; furthermore, the H 2 10 safety risks have not been accepted by the Landfill industry. Catalytic oxidation is also a potential alternative technology which has previously been utilized for the treatment in various contaminated streams of many of the same NMOC encountered in LFG. As noted above, many of these NMOC contain halogens and sulfur. Among the halogen-containing compounds, one finds chloro/fluoro-substituted CH 4 , C 2 H 6 , and C 2 H 4 homologues (including chloroform, methylene and methyl chloride, dichloro, thrichloro, and tetrachloroethane, PCE, and vinyl chloride among others), mono- and dichloro-benzene, and other polychlorinated aromatic hydrocarbons [Baker et al., 1992]. The various sulfur compounds include carbonyl sulfide (COS), dimethylsulfide (DMS), and various mercaptans. Below, the various studies to date on the catalytic oxidation of halogen-containg and sulfur-containg VOC are briefly reviewed. Chlorinated Hydrocarbons…There are several papers that deal with the catalytic oxidation of chlorinated hydrocarbons, so the discussion here focuses primarily on compounds that are commonly encountered in landfill gas. Table 1.1 summarizes the key studies, so far, on the catalytic oxidation of chlorinated hydrocarbons. Dai et al. [2013] studied the lower temperature catalytic combustion of chlorinated hydrocarbons (CHCs), including chlorobenzene (CB), 1,2-dichloroethane (DCE) and trichloroethylene (TCE), over RuO 2 supported on Ti-doped CeO 2 catalysts (Ru/Ti–CeO 2 ) and the effects of preparation methods, Ti content, Ru content, inlet CB concentration 11 and space velocity, oxygen concentration and water vapor. The results show that the doping of Ti can improve the catalytic activity and stability of CeO 2 based catalysts. The authors were also able to achieve stable conversions for all of the CHCs for temperatures as low as 200 o C with CeO 2 based catalysts by the loading or doping of Ru and Ti. Ma et al., [2012] studied the catalytic oxidation of 1,2-dichlorobenzene (o-DCB) using CaCO 3 /α-Fe 2 O 3 nanocomposites which are environmentally friendly. The authors found that, the nanocomposite with 9.5 mol% Ca showed the highest catalytic activity, which could be attributed to the promoting effect of CaCO 3 on α-Fe 2 O 3 with smaller crystallite size. Experimental results in the presence of water indicated that, due to the competitive adsorption of water on the active sites, there was a local minimum of catalytic activity at 350 °C. Hyung Ik et al., [2010] studied the catalytic conversion of 1,2-Dichlorobenzene over mesoporous zeolite materials. According to the authors, this is the first time that these recently developed mesoporous materials from zeolites (MMZ) were used as a support for an oxidation catalyst. The catalytic oxidation of 1,2-dichlorobenzene over Pt/MMZ was carried out, and the catalytic activity was compared with that of Pt/ γ-Al 2 O 3 . Pt/MMZ showed the highest activity for the catalyst tested. The authors found that the catalytic activity of the Pt/MMZ was maintained (>40%) at low temperatures (250 o C), while the other catalysts showed extremely low activity at these temperatures. The high catalytic activity of Pt/MMZ was attributed to both the satisfacory acidity and mesoporosity of the MMZ support. 12 Yoon and Lee et al. [2004] have studied the oxidation of TCE on Ru-CrO x /Al 2 O 3 catalysts. The authors have reported that both Brönsted and Lewis acid sites existed on the catalyst surface. These sites provided adsorption sites for both molecular oxygen and TCE. Musialik-Piotrowska et al. [2002] studied the catalytic oxidation of trichloroethylene oxidation in the presence of non-halogenated compounds, like toluene and ethanol in an air stream. Over platinum catalysts, water had a positive effect on the conversion and the selectivity of the reaction towards HCl. In addition to this it was found that over noble metal catalysts, the presence of the TCE decreased the oxidation rate for the two non-halogenated compounds. Kulazynski et al. [2002] also studied the catalytic combustion of TCE. They found that TiO 2 -SiO 2 is a stable catalyst support material for TCE combustion, and it also exhibits its own (small) activity toward TCE combustion as well; catalysts supported on TiO 2 -SiO 2 differ in activity according to the type of active phase. For example, metal-oxide-based catalysts (chromia, vanadia) are very active in the oxidation of TCE, however, they lose their active components, especially at higher reaction temperature; additionally they are inhibited by the water present in the feed (which consisted of 12% O 2 , 88% N 2 ). Noble metal (Pt, Pd)-based catalysts are less active per gr of catalyst, but much more active per mole of active phase than the oxidic ones for TCE oxidation. Furthermore, these catalysts are less sensitive to the water content in the feed, and their activity is inhibited to a lower extent by the chlorine-containing reaction products than the activity of oxidic catalysts. Toledo et al. [2001] have reported that Pt-based catalysts have the longest life (or the lowest deactivation rate) for the decomposition of dichloromethane (DCM) and TCE. The 13 authors reported that chromia- and vanadia-based catalysts did not resist the attack by the nascent Cl, and were quickly lost from the surface. They also reported that the deactivation detected in some Pt based catalyst can be attributed to the catalyst support (Al 2 O 3 ) and not the Pt metal itself. Everaert et al. [2004] studied the destruction of chlorinated VOC’s on vanadium–titanium catalysts coated on a sintered metal–fleece. They measured the destruction efficiency in the range of 260-340 o C. The authors have reported that the oxidation of chlorinated VOC’s proceeds to a higher extent with increase in temperature, with the multiple chlorine substituted compounds having higher reactivity. Yim et al. [2002] studied the destruction of PCDD/F (polychlorinated dibenzodioxins/ Furan) on a vanadium-based catalyst. The authors have reported that water vapor inhibits the reaction of the PCDD/F. This may be due to the fact that water reduces the contact between the PCDD/F molecules and active sites of the catalyst. Studies on the catalytic oxidation of chlorinated aromatic hydrocarbons have been carried out by Krishnamoorthy et al. [2000]. The authors have investigated the catalytic oxidation of 1,2-dichlorobenzene over a series of transition metal oxides (i.e., Cr 2 O 3 , V 2 O 5 , MoO 3 , Fe 2 O 3 , and Co 3 O 4 ) supported on TiO 2 and Al 2 O 3 . The activity of the different catalysts for this reaction depends on the nature of the transition metal oxides utilized, with Cr 2 O 3 -and V 2 O 5 -based catalysts being the most active ones. With the 14 exception of the cobalt oxide catalysts, the TiO 2 -supported systems were more active than the corresponding Al 2 O 3 -supported ones, indicating that the metal oxide–support interactions are significant in this reaction. Experiments conducted in the presence of water indicate an inhibiting effect for the V 2 O 5 - and Cr 2 O 3 -based catalysts and a promoting effect for the Co 3 O 4 /TiO 2 catalyst. The Fe 2 O 3 - and MoO 3 -based catalysts were unaffected by the presence of water. Competitive adsorption between the surface species and water is suspected to be the reason for the inhibition, while the promoting effect can be attributed to the reaction of water with surface Cl − . Van den Brink et al. [1998] conducted a study of the oxidation of chlorobenzene. They found that Pt seems to be responsible for the formation of polychlorinated benzenes, which are formed by chlorination of adsorbed (chloro)benzene-species through Pt oxy- chlorides. The addition of water to the air carrier reduces the formation of the polychlorinated benzenes and improves the conversion. When the γ-Al 2 O 3 support alone is applied for the catalytic oxidation, complete conversion of chlorobenzene is reached only at ~550 °C, but without production of polychlorinated benzenes. 15 Authors Contaminants Catalysts Temp (ºC) Remarks Dai [2013] Chlorobenzene, 1,2- dichloroethane, trichloroethylene Ru/Ti–CeO 2 300 o C Improvement in stability of CeO 2 based catalysts by the introduction of Ru and Ti is general and not related with the type of CHCs. Ma [2012] 1,2-dichlorobenzene CaCO 3 /α-Fe 2 O 3 nanocomposites 350 °C Nanocomposite with 9.5 mol% Ca showed the highest catalytic activity, which could be attributed to the promoting effect of CaCO 3 on α-Fe 2 O 3 with smaller crystallite size Hyung Ik [2010] 1,2-Dichlorobenzene Pt/Zeolite 250 o C The recently developed mesoporous materials from zeolites (MMZ) were used for the first time as a support for an oxidation catalyst. The catalytic oxidation of 1,2- dichlorobenzene over Pt/MMZ was carried out, and the catalytic activity was compared with that of Pt/ γ-Al 2 O 3 . 16 Authors Contaminants Catalysts Temp (ºC) Remarks Everaert [2004] monochlorobenzene, 1,-2 dichlorobenzene and o-chlorophenol V 2 O 5 - WO 3 /TiO 2 260– 340 Oxidation of chlorinated VOC proceeds to a higher extent with increase in temperature with multiple chlorine substitution enhancing the reactivity Yoon and Lee [2004] Trichloroethylene Ru- CrO x /Al 2 O 3 249 Ru promoted the catalytic activity and the resistance to deactivation Musialik- Piotrowska et.al, [2002] Trichloroethylene Pd and Pt on Al 2 O 3 support 240 on Pt and 270 on Pd Water had a positive effect on conversion in Pt catalysts Kulazynski [2002] Trichloroethylene Cr 2 O 3 , V 2 O 5 , Pt, Pd over SiO 2 -TiO 2 support 400C CrO x and VO x are very active, but they are inhibited by water present in the feed. Pt and Pd are active at higher temperatures but less sensitive to water content Yim et al. [2002] PCDDs/PCDFs CrO x /TiO 2 325 for (99%) Water vapor inhibits the rate of reaction of PCDD/F Toledo et al. [2001] TCE. Dichloro- methane, and chlorobenzene Pt, Pd, Ru- based catalysts 200- 450 The Pt-based catalysts were not as active as the chromia and vanadia-based catalysts, but they have a much longer useful life. 17 Authors Contaminants Catalysts Temp (ºC) Remarks Krishnamoorthy, et al. [2000] 1,2- Dichlorobenzene (DCB) Cr 2 O 3 , V 2 O 5 ,Fe2O 3 , MoO 3, Co 3 O 3 Cr 2 O 3 and V 2 O 5 exhibited the highest activity van den Brink et al. [1998] Chlorobenzene Pt/Al 2 O 3 400- 600 Pt/ γ-Al 2 O 3 , although widely used, is not a suitable catalyst for the catalytic combustion of chlorobenzene (and chlorinated aromatics in general) Table 1.1: Literature review of the catalytic oxidation of chlorinated compounds From the above literature studies, we can conclude that Pt-based catalysts are the best ones for chlorinated hydrocarbon oxidation, due to their inertness towards attack by HCl and Cl. Also Pt-based catalysts work best in the presence of water, which is commonly present in LFG. Sulfided Compounds…There are also a number of prior studies on the catalytic oxidation of sulfided hydrocarbons, so discussion in this introduction focuses again primarily on compounds that are commonly encountered in landfill gas. Table 1.2 below summarizes some of the key studies in this area. 18 Pope et al. [1976] in an earlier study tested cobalt oxide as a catalyst in the oxidation of dimethyl sulfide (DMS). The authors report that they were able to achieve complete oxidation at 235 o C for 4 ppm and at 370 o C for 10 ppm of DMS in air. Also they found that when they passed air along with DMS no SO 3 and SO 2 were observed. It appears that the sulfur oxides released from the reaction are incorporated into the Co 3 O 4 catalyst, hence causing much faster deactivation. Pope et al. [1976] also used a platinum honeycomb catalyst and found that dimethyl sulfide was removed completely at 250 o C. At lower temperatures, there was a marked deficit in the amount of carbon dioxide formed, probably due to the formation of partial oxidation products. The formation of sulfur dioxide exhibited a more complex behavior at lower temperature due to the fact that it adsorbs on the catalyst surface. Devulapelli et al. [2008] studied the oxidation of dimethyl sulfide with ozone. High conversions were achieved only at high temperatures (>250 °C). Reaction with oxygen only showed very low conversion of DMS at low temperatures, and produced high concentrations of dimethyl sulfoxide (DMSO). The authors report that the best catalytic activity was obtained over 10 wt.% CuO-10 wt.% MoO 3 supported on γ-Al 2 O 3 , which showed 100% DMS conversion and high selectivity (~96%) towards complete oxidation products such as CO 2 and SO 2 . Also they found a stronger acid strength in the CuO- MoO 3 /γ-Al 2 O 3 than in the CuO/γ-Al 2 O 3 catalyst. The results also revealed that high temperatures and higher than stoichiometric amounts of ozone favor total conversion of DMS into CO 2 and SO 2. 19 Hwang et al. [2011] studied the vapor phase oxidation of DMS with ozone over ion- exchanged zeolites. Ozone was used as an oxidant to assess the oxidation capability of Ag/ZSM-5, Mn/ZSM-5 and Ag-Mn/ZSM-5 catalysts towrds DMS at both room temperature and 130 °C. Ion-exchange with silver ions (Ag + ) strengthened the adsorption of DMS, resulting in an increased oxidation capacity for DMS. Furthermore, the introduction of manganese ions (Mn 2+ ) strengthened the oxidation capability of DMS, thus enhancing the selectivity towards SO 2 obtained from the oxidation. Chu et al. [2001] studied the catalytic incineration of ethyl-mercaptan, typically emitted from the petrochemical industry, over a Pt/Al 2 O 3 fixed-bed catalytic reactor. The results show that the conversion of C 2 H 5 SH increases as the inlet temperature increases and the space velocity decreases. For the temperature range (200-260 °C), the higher the C 2 H 5 SH concentration is, the lower its conversion. The O 2 concentration has a positive effect on the conversion of C 2 H 5 SH. C 2 H 5 SH has a poisoning effect on the Pt/Al 2 O 3 catalyst, especially at lower temperature. However, the sulfur-poisoning effect of C 2 H 5 SH on the catalyst can be reduced by raising the operating temperature to a sufficiently high value. The existence of CH 3 SH has no effects on the conversion of C 2 H 5 SH. Single metal oxides, particularly Co 3 O 4 , MnO 2 and CuO 1 were evaluated by Heyes, et al. [1982] as the catalysts for the destructive oxidation of organic air pollutants, e.g., in malodorous process emissions, with special focus on their resistance to deactivation by 20 sulphur compounds, as compared with the Pt-honeycomb catalysts. The authors found that tests with low odorant concentrations, 100 ppm v in air, demonstrated highly efficient conversion of n-butanal or methyl mercaptan to acceptable products over the best oxides. The ability to destroy the butanal in mixtures with mercaptan, during which the mercaptan was completely removed, decreased in the order: CuO Pt > MnO 2 > V 2 O 5 > CO 3 O 4 . Values of the odor detection threshold measured at the catalyst exit by dynamic- dilution olfactometry were interpreted in terms of unreacted butanal and the selectivity in conversion of the odorants to CO + CO 2 . 21 Authors Contaminants Catalyst Temp Remarks Hwang [2011] (CH 3 ) 2 S Ag/ZSM- 5, Mn/ZSM- 5 and Ag- Mn/ZSM- 5 130 o C Silver ions present in the zeolite (Ag + ) strengthened the adsorption of DMS, whereas manganese ions (Mn 2+ ) strengthened the oxidation capability of DMS Devulapelli et al. [2008] (CH 3 ) 2 S V 2 O 5 /TiO 2 >250 o C Increase in temperature favors oxidation to SO 2 and CO 2 Devulapelli et al. [2008] (CH 3 ) 2 S Cu, Mo, Cr and Mn oxides supported on γ- alumina 100-200 o C Best catalytic activity was obtained over 10- wt.%CuO-10- wt.%MoO 3 /γ- alumina catalyst Chu H et al. [2001] C 2 H 5 SH / CH 3 SH Pt/Al 2 O 3 200-260 o C Poisoning effect can be reduced by raising the operating temp. Heyes et al. [1982] CH 3 SH CuO, Pt, MnO 2 , V 2 O 5 CO 3 O 4 - The ability to destroy the VOC is in the following order CuO = Pt > MnO 2 > V 2 O 5 > CO 3 O 4 . Pope et al. [1978] (CH 3 ) 2 S Pt/Co 3 O 4 267 o C Pt catalysts were found to be more active than Co 3 O 4 22 Authors Contaminants Catalyst Temp Remarks Pope et al. [1976] (CH 3 ) 2 S Co 3 O 4 215-433 o C Sulphur dioxide released from the reaction is incorporated into the Co 3 O 4 catalyst, hence causing much faster deactivation Table 1.2: Literature review of the catalytic oxidation of sulfided compounds From the above literature studies for sulfided hydrocarbons we can conclude that Pt- based catalysts have again the best ability to destroy malodorous sulfided hydrocarbons in the presence of simple hydrocarbons such as methane and ethane. This fact is particularly important for the oxidation of such compounds in LFG, as it contains a substantial fraction of methane. Fluorinated Hydrocarbons...There are also a number of studies on the catalytic oxidation of fluorinated hydrocarbons. Table 1.3 below summarizes some of the key studies of compounds relevant to LFG. In an early study Aida et al. [1990] discussed the use of gold nanocatalysts in the decomposition of CFC12. They chose gold due to its chemical inertness. However, no data supporting the stability of catalyst were reported. The authors reported that Al 2 O 3 can itself decompose fluorinated hydrocarbons (but, on the other hand, it cannot decompose purely chlorinated hydrocarbons). 23 Bickle et al. [1994] compared Pt/Al 2 O 3 and Pt/ZrO 2 based catalysts for the oxidation of CFC113. The authors found that Pt/ZrO 2 was more tolerant to fluorine species and gave a higher conversion of 99% over a longer time frame of 300 h. Nagata et al. [1994] also studied the oxidation of CFC115 in the presence of V 2 O 5 and WO x supported on γ-Al 2 O 3 for the oxidation of CFC115. The authors concluded the following: (1) γ-alumina- supported tungsten( VI) oxide and vanadium( VI) oxide catalysts exhibited the highest activity among all the catalysts investigated; (2) The acid sites of the catalyst were necessary for the oxidative decomposition of CFC-115 in the presence of hydrocarbons such as methane and ethane, which explains why alumina gave the highest activity; (3) the water formed by the combustion of hydrocarbons suppresses the catalyst deactivation in the decomposition of CFC115. Tajima et al. [1996] have reported that γ-Al 2 O 3 resulted in 85% conversion for CFC113 at 500 o C, whereas for SiO 2 -TiO 2 and ZrO 2 they reported conversions of 40% and 0 % respectively. The authors also reported that the oxidation products on γ-Al 2 O 3 were CO and CO 2 , and no other carbon containing products were found. Ning et al. [2011] studied the catalytic decomposition of CFC-12 over solid super-acid Mo 2 O 3 /ZrO 2 catalyst. The authors reported that CO 2 and CClF 3 were the main products and no CO was detected as by-product. The decomposition activity of solid super-acid catalyst largely depended on the content of ZrO 2 and the calcination temperatures of 24 Mo 2 O 3 /ZrO 2 . The highest activity for catalytic decomposition of CFC-12 was at 450°C and the optimum content of ZrO 2 in the Mo 2 O 3 /ZrO 2 catalyst was 20–40% by weight. Authors Contaminants Catalyst Temp Remarks Ning [2011] CFC-12 Mo 2 O 3 /ZrO 2 450°C The decomposition activity of solid super-acid catalyst largely depended on the content of ZrO 2 and the calcination temperatures of Mo 2 O 3 /ZrO 2 . Tajima et al. [1996] CFC113 γ-Al 2 O 3 , SiO 2 -TiO 2 , ZrO 2 pellets 500 o C γ-Al 2 O 3 gave 85% conversion which was significantly higher than other catalysts Bickle et al. [1994] CFC113 Al 2 O 3 ,Pt/ ZrO 2 -(PO 4 ) 500 o C Zirconia was more tolerant to fluorine but Pt can give 99% conversion Nagata et al. [1994] CFC115 γ-Al 2 O 3 , V 2 O 5 , Wox 600 o C γ-Al 2 O 3 supported tungsten oxide gave the highest activity Aida et al. [1990] CCl2F2 Au/Al 2 O 3 200- 400 o C Al 2 O 3 can itself decompose fluorinated hydrocarbons, but in the case of just chlorinated hydrocarbons it cannot do that Table 1.3: Literature review of the catalytic oxidation of fluorinated compounds From the above literature studies it can be concluded that Pt/Al 2 O 3 catalysts provide higher conversion of fluorinated hydrocarbons at lower temperatures than all other catalysts studied. HF is the product of the complete oxidation of fluorinated hydrocarbons which is likely to have an impact on the structural stability of the catalyst support. However, fluorinated hydrocarbons are present in ppm levels in landfill gas, and the novel reactor that is proposed for study here provides for a more effective way for 25 removing the HF from the catalytic pore environment than conventional reactors, so the impact of treating fluorinated VOC may not be as severe. Although the individual VOCs encountered in LFG are destroyed by catalytic methods including oxidation, a complex matrix such as landfill gas makes their removal potentially more challenging. That may explain why NMOC removal in LFG via catalytic oxidation may be an attractive alternative, but has yet to be practiced commercially. In addition, conventional catalytic oxidation technology, based on either granular packed-beds or monolithic catalytic converters, may not be appropriate for the LFG application due to the fact that the oxidation temperature required is undesirably high, ~300 to 650 °C. This feature is not, however, intrinsic to the catalytic process itself, but rather the result of technical barriers associated with existing reactor technology. We know of only one published study on the catalytic oxidation of NMOC in LFG. In this study Urban et al. [2009] used two different types of V 2 O 5 –TiO 2 catalysts for the oxidation of VOC’s found in simulated landfill gas. One of the catalysts is composed of only V 2 O 5 and TiO 2 (catalyst B), while the other catalyst (catalyst A) contained also significant amounts of WO 3 , MnO 2 , CuO, and Fe 2 O 3 , which may act as promoters to enhance the activity of the catalyst. Studies were carried out in a lab-scale packed-bed tubular reactor at 300 o C. Four different model VOC compounds were chosen for the study, namely benzene or toluene as the aromatic VOC, CFC113, H 2 S and TCE. Lab tests 26 with only the individual model compounds in N 2 and in the presence of O 2 or water were carried with catalyst A and the results are shown in Table 1.4. Model compound Concentration [ppmv] Carrier gas N 2 [ml min −1 ] O 2 [ml min −1 ] H 2 O [ml min −1 ] Conversion [%] TCE 213 504 10 97 TCE 208 504 0.8 88 CFC-113 183 504 10 78 CFC-113 185 504 0.8 40 Benzene 202 504 10 99.8 Benzene 200 504 0.8 ~0 H 2 S (with 100 ppmv O 2 ) 45 504 ~100 Table 1.4: Catalytic activity of catalyst A towards the removal of model VOC compounds. The catalyst appears effective towards oxidation of H 2 S, TCE, and benzene, but not so much towards the oxidation of CFC113. The presence of water appeared to have a detrimental effect on the conversion of benzene, which the authors attributed to the hydroxylation of the V 5+ catalytic centers which are thought to be responsible for the oxidation of the aromatic VOC (using a vanadia-based catalyst for the treatment of LFG which is normally saturated with water would then appear to be questionable). Lab studies using the catalyst B were also carried out in the presence of individual model VOC in simulated LFG containing 45% CH 4 , 25% CO 2 , 1% O 2 , 29% N 2 and the results are shown in Table 1.5. 27 Model compound Concentration [ppmv] LFG total [ml min −1 ] O 2 fraction [ml min −1 ] Conversion [%] TCE 219 1000 10 93.2 CFC-113 189 1000 10 1.4 Toluene 351 1000 10 99.8 H 2 S 107 1000 10 ~100 Table 1.5: Catalytic activity of catalyst B towards removal of model minor compounds. The catalyst showed good activity towards TCE, toluene and H 2 S, but rather poor activity towards CFC113. The presence of water, however, had again a serious detrimental effect towards the oxidation of the toluene. In conclusion, though catalytic oxidation technology has been studied for the decomposition of a few select VOC compounds found in LFG, no such commercial technology currently exists for the treatment of LFG. What we propose here is to study a novel catalytic oxidation technology appropriate for LFG clean-up, which potentially overcomes many of the current technical challenges that the conventional catalytic oxidation technology faces. It makes use of the flow-through catalytic membrane reactor (FTCMR) concept -- see discussion below-- endowed with an oxidation nanocatalyst. This is an important goal, as economical, environmental and energy advantages will be realized, with a process cost-effectively removing the NMOC from LFG. 28 1.3 Flow-Through Catalytic Membrane Reactors (Please note that the studies of the prior technical literature on FTCMR discussed here were done collaboratively with my USC Group colleague, Dr. Motamedhashemi and, thus, significant overlap exists between this section and the corresponding section on FTCMR in his Thesis published in December 2012). Membranes are generally thought as structures, which are permeable to one or more components of a surrounding fluid but impermeable to its other components. They can be made from different materials such as metals, ceramics and polymers [Melin et al., 2007]. Process integration through the application of membrane reactors [Sanchez-Marcano & Tsotsis, 2002], which combine reaction and membrane separation in a single unit (thus creating substantial synergies) promises numerous benefits when compared to conventional processes. In a membrane reactor the membrane can fulfill different functions: 1. The membrane selectively removes one or more products from the reaction mixture (‘Extractor’); 2. It controls the addition of reactants to the reaction mixture (‘Distributor’); 3. It intensifies the contact between the reactants and catalysts (‘Contactor’). Catalytic membranes without a separative function have been applied successfully as microstructured reactors [Dixon et al., 2003; Motamedhashemi et al., 2011]. If the reactant mixture is forced to flow through the pores of a membrane which have been impregnated with catalyst, the intensive contact allows for high catalytic activity with negligible mass transport resistances [Zaspalis et al., 1991; Saracco et al., 1994]. Such 29 reactors allow for precise control of the contact time between reactants and catalysts, and represent a new important class of membrane reactors potentially equal in importance with the permselective extractor and distributor membrane reactors. An early review dealing exclusively with the topic of flow-through catalytic membrane reactors (FTCMR) was published by Groschel et al. [2005]. The authors classified FTCMR into gas-phase integral reactors with stoichiometric feed of reactants, and liquid-phase differential reactors consisting of a loop of a membrane module and saturation tank. In the FTCMR concept, an unselective porous catalytic membrane is typically utilized and the reactants are forced to flow through the membrane. The function of the membrane is to provide a reaction space with short controlled residence time and high catalytic activity. In classical fixed-bed reactors the conversion is limited by pore diffusion in the catalyst pellets. In the FTCMR the catalyst is placed inside the membrane pores and the reactants flow convectively through the pores, the ensuing intensive contact between reactants and catalyst resulting in a high catalytic activity. The motivation for applying an FTCMR is to either aim to reach complete conversion in minimum time or space, taking advantage of the high catalytic efficiency, or to reach maximum selectivity for a given reaction due to the narrow contact time distribution. Westermann and Melin [2009] classify FTCMR into three different classes, namely: 1. Complete Conversion Integral FTMCR 2. Selective Integral FTCMR 3. Selective Differential FTCMR 30 1.3.1 Complete Conversion Integral FTCMR For the complete conversion integral FTCMR the premixed reactants flow through the catalytic membrane in a single pass with the aim of reaching complete conversion, taking advantage of the high catalytic efficiency caused by the intensive contact between the reactants and the catalyst. This mode has been applied to several gas-phase reactions most recently in the oxidation of Dimethyl Methyl Phosphonate (DMMP) in air [Motamedhashemi et al, 2011]. Most of the applications are in the area of decomposition of volatile organic compounds, including photocatalytic oxidations. Table 1.6 below list some of the key studies done on complete conversion integral FTCMR. 31 Authors Reaction Membran e Catalyst dpore (nm) tcat (µm) Yamada et al. [1988] 1-Butene isomerization Anodized Al 2 O 3 None ~100 5-100 Saracco and Specchia [1994] VOC Decomposition Porous ceramic γ-Al 2 O 3 , urea method 1000-15000 1500 Pina et al. [1996] VOC Decomposition γ-Al 2 O 3 on MF Pt, wet impregnatio n 3.5 ~1700 Zalamea et al. [1999] VOC Decomposition γ-Al 2 O 3 on MF Pt, wet impregnatio n 2000-10000 1000 Splinter et al. [2002] CO oxidation Porous Silicon Pd wet impregnatio n 4000-8000 70 Maira et al. [2003] TCE photocatalytic oxidation Zeolite or porous steel plate TiO 2 ~1 5 Tsuru et al. [2003] Methanol photocatlytic oxidation -Al 2 O 3 - MF TiO 2 colloidal sol 6.5 n/a Picasso et al. [2003] MEK Alumina/ Stainless Steel Fe 2 O 3 5(Alumina) / 500(stainles s steel) n/a Motamedhashe mi et al. [2011] DMMP oxidation in Air γ-Al 2 O 3 on - Al 2 O 3 Pt 50 n/a Chea et al [ 2012] Guaiacol and 2,6- dimethoxyphen ol degradation -Al 2 O 3 Laccase 200-1400 n/a García-García et al. [2013] Water gas shift reaction Al 2 O 3 hollow fibre 10% CuO/CeO 2 0.1- 0.2 μm n/a Table 1.6: Literature review of complete conversion integral FTCMR 32 Yamada et al. [1988] was the first to report the application of a complete conversion integral FTCMR. They prepared catalytic membranes by means of anodic oxidation of aluminum plates. In the reactor setup the fluid permeates from one side to the other through the catalytic membrane. For the model reaction of the isomerization of 1-butene the membrane reactor showed similar catalytic activity when compared to the fixed-bed packed with powder of the anodized alumina film. For potential applications in flue-gas cleaning, Saracco, et al. [1994] catalytically activated ceramic porous filters by applying an alumina layer. The idea here is for the filter to mechanically remove the particles, and for the catalytically active membrane to decompose the chemical pollutants such as NO x and VOC that are forced to flow through it. The capability of the filters is assessed by performing a model reaction, namely the dehydration of isopropanol, which is catalytically promoted by the γ-alumina itself. Nearly complete conversion can be achieved for superficial velocities of industrial interest. The concept of a FTCMR working in the Knudsen-diffusion range is investigated for complete combustion of VOC by Pina, et al. [1996]. In a commercial MF-membrane with a γ-Al 2 O 3 separation layer and pore diameters of 200 nm an additional γ-Al 2 O 3 phase is deposited by the sol–gel method and a Pt catalyst (0.13 wt.%) is introduced by wet impregnation. A VOC-containing air stream (100–5100 ppm toluene) is forced to 33 permeate through the Pt/Al 2 O 3 catalytic membrane operating in the Knudsen-diffusion regime which provides intimate contact between the VOC molecules and the combustion sites, thus minimizing the diffusion resistances and providing a highly efficient use of the catalyst. The minimum temperatures to achieve complete conversion with this FTCMR are reported to be low. On the other hand, pressure drops are a concern with such a reactor (0.18 – 0.35 bar for gas flows of 0.4 – 1.0 L/min). While the investigations of Pina, et al. [1996] are limited to the Knudsen flow regime (KFR), Zalamea, et al. [1999] utilized catalytic membranes with wider pores operating in the mixed-flow regime for the combustion of VOC in order to minimize pressure drops which were identified as the major drawback when operating in the KFR. The increased pore size of these membranes results in significant laminar flow contributions. On these tubular membranes with a uniform pore structure, Zalamea, et al. [1999] deposited 2–3 wt.% of γ-Al 2 O 3 , as well as 0.15 wt.% of Pt catalyst by wet impregnation. The catalytic membranes had a BET surface of 5.9 m 2 /g, with most of the pore volume occupied by the large pores (2–10 μm). The temperatures required for complete conversion in the mixed- flow regime membranes were higher compared to those of the asymmetric membranes with high Knudsen-flow contribution, but for the latter the pressure drop was higher. Zalamea et al. [1999] state that considering the operating costs associated with larger pressure drops, the mixed-flow regime may be preferable in spite of an increase in combustion temperature by 15–50 o C. 34 For the pre-processing of gas mixtures in gas analyzers, Splinter et al. [2002] designed a FTCMR making use of a micromembrane. The aim of the FTCMR is to convert CO into CO 2 in order to increase the sensor selectivity. The reactor is fabricated by the combination of anisotropic silicon etching to create the membrane and porous silicon technology to perforate the membrane. The advantage of porous silicon as a flow-through system is the adjustable pore diameters (here, 4–8 m) with a high open porosity, a large reactive surface and a ramified structure, which is supposed to enable gas turbulence. The membrane area is smaller than a square millimeter resulting in porous silicon surfaces between 8.6 and 39.4 cm 2 . The applied gas molecule retention times are around 1 ms. The silicon substrate was impregnated with Pd by simmering it in Pd(Acetate) 2 dissolved in toluene. The complete porous silicon surface area (~100 m 2 /cm 3 ) was thus covered with a Pd layer of 700 nm. In comparison to more conventional reactors, the FTCMR showed a higher conversion, a low dead volume and small dimensions. The time of reaction depends on the membrane thickness, which together with the membrane active area can be adjusted to convert nearly 100% of the CO. Maira, et al. [2003] investigated the effect of the flow configuration on the gas-phase photocatalytic oxidation of trichloroethylene, applying catalytic zeolite membranes and a catalytic stainless steel plate both impregnated with TiO 2 photocatalyst. While both the catalytic stainless steel (SS) plate and the hybrid zeolite-TiO 2 membrane catalyst show catalytic activity operating in the parallel flow mode, the conversion can be radically increased by switching either into the flow-through or into a mixed-flow mode. In terms 35 of its selectivity, the catalytic membrane outperformed the catalytic plate, which is explained by a possible retention of the larger pollutant molecules by the membrane. The gas-phase oxidation of methanol as a model volatile organic component in a photocatalytic membrane reactor was studied by Tsuru, et al. [2003]. After fabricating TiO 2 membranes with pore sizes of several nm on cylindrical -alumina MF membranes, they used them to study the decomposition of methanol. In their experiments, an air stream with a methanol load of 1000 ppm is fed into the shell side of the membrane tube, which is simultaneously irradiated with black-light lamps to catalyze the generation of OH radicals on the TiO 2 surface. Two different flow patterns are applied: In the first mode without permeation, the product stream is also taken from the shell side, allowing the reactants to contact the catalytic surface by diffusion only. In the second mode with permeation, the reactants leave the reactor on the tube side of the membrane, forcing them to flow convectively to the surface and through the membrane pores, and thus providing for more efficient contact between the reactants and the catalytically active TiO 2 surface. A higher decomposition rate is observed for the reactor mode with membrane permeation. This is attributed to two factors: the enhanced transport to the surface by forced convection in addition to the transport by diffusion, as well as to the utilization of a larger surface area. Picasso et al. [2003] studied the concept of a Knudsen contactor for total combustion of methyl-ethyl ketone (MEK) in diluted streams using Fe 2 O 3 as catalyst. The membranes 36 were prepared from 90 mm long, 10 mm o.d. asymmetric ceramic tubes (Inocermic) with 5 nm pores in the γ-Al 2 O 3 thin layer and stainless steel tubes of similar dimensions with an effective pore size of 500 nm. The reaction was carried out with MEK (in the range 500–2000 ppm v ) in air at temperatures as low as 255 °C. The authors found that flow- through catalytic membranes based on these membranes and operating in the Knudsen permeation regime have shown a higher efficiency in the complete combustion of MEK than their bulk catalyst counterparts under equivalent experimental conditions. Motamedhashemi et al. [2011] investigated the use of 3 layer γ-Al 2 O 3 on -Al 2 O 3 membrane with Pt catalyst for removing Dimethyl Methyl Phosphonate (DMMP) from air. Preliminary experiments are reported for different DMMP feed concentrations and reactor temperatures, which demonstrate the potential advantage of the FTCMR in the complete catalytic oxidation of this important chemical warfare agent (CWA) simulant. Complete destruction of low and high concentrations of DMMP was achieved at lower temperatures compared to the values obtained in this study for a wall-coated plug-flow (monolith) reactor containing the same amount of catalytic metal. The authors also developed a mathematical model to provide a better understanding of the fundamental transport phenomena underpinning the FTCMR operation. Chea et al. [2012] investigated the use of enzymatic membranes for the degradation of two phenolic compounds, namely 2,6-dimethoxyphenol (DMP) and guiaicol by laccases from Trametes versicolor using a FTCMR concept to improve contact between catalytic 37 sites and substrates. The enzymatic membranes were prepared by grafting laccase onto a gelatin layer previously deposited on α-alumina supports. The steps of the immobilization procedure were optimized in order to improve the catalytic efficiency and stability of the membrane. Their studies showed that the optimized membrane is very effective for removing DMP from aqueous model solutions. However, fouling was observed during long-term experiments due to the formation of polymers resulting from DMP degradation. García-García et al. [2013] studied the use of asymmetric and symmetric Al 2 O 3 hollow fibres as a support with a 10% CuO/CeO 2 catalyst in the development of an asymmetric hollow fibre reactor (AHFR) and a symmetric hollow fibre reactor (SHFR), respectively. The water gas shift (WGS) reaction was chosen as a sample reaction to compare the performances of both AHFR and SHFR with a traditional fixed bed reactor (FBR). The catalytic activity tests in the FBR were carried out using the powder ground from either the asymmetric or symmetric Al 2 O 3 hollow fibre impregnated with 10% CuO/CeO 2 . Two different configurations, “dead-end” and “open-end”, were studied in the AHFR and SHFR. The experimental results show that, despite the differences observed between the AHFR and SHFR, both reactors offer important advantages over conventional FBRs including high catalytic activity along with a better selectivity. The performance of the AHFR working either with “dead-end” or “open-end” configurations at 400 °C was 6 times higher than that obtained in the FBR. The authors contend that the micro-channeled structure improves the mass transfer during the reaction. 38 1.3.2 Selective Integral FTCMR This mode of operation is similar to the one described in the previous section, in the sense that the premixed reactants flow through the catalytic membrane in a single pass. In addition to the high catalytic efficiency of the FTCMR, the concept, however, takes advantage of the narrow residence time distribution in the membrane, which limits the sequential reactions towards undesired byproducts. For such an FTCMR a pin-hole free microstructure is essential to prevent maldistribution effects. Intrinsic maldistribution effects are well known for monolithic reactors Kreutzer [2005], but have not been directly measured for membrane reactors. Selective reactions performed in such reactors often prove indirectly the narrow residence time distribution in such reactors when compared to competing reactor concepts. The selective-integral membrane reactor concept has been applied to several selective reactions, mainly in the gas phase, such as partial oxidation, partial hydrogenation, oligomerization or coupling reactions of hydrocarbons. In a few cases a liquid phase is also present during the reaction, allowing for the use of the term ‘three-phase reactor’. The membranes, reactions and catalysts utilized are summarized in Table 1.7. 39 Authors Reaction Membrane Catalyst dpore (nm) tcat (µm) Zaspalis et al. [1991] MeOH selective dehydrogenation -Al 2 O 3 γ-Al 2 O 3 3-4 7-8 Kobayashi et al. [1996] Propane selective epoxidation Microporous glass Cs-Ag, Re-Ag, Ag 2 O 220- 480 1000 Lange et al. [1998] Hexadiene selective hydrogenation Asymmetric ceramic Pt 5 0.2-0.4 Ma et al. [1998] Methane oxidative coupling Asymmetric -Al 2 O 3 Sm 2 O 3 20- 5000 30 Lambert and Gonzalez [1999] Acetylene/butadi ene partial hydrogenation Asymmetric γ-Al 2 O 3 Pt/ γ- Al 2 O 3 by sol- gel 3.6 11 Alfonso et al. [2001] Butane oxidative and non- oxidative dehydrogenation -Al 2 O 3 ; layer 1 V/ MgO 200/12 000 30/1000 -Al 2 O 3 ; layer 2 Pt-Sn/ γ- Al 2 O 3 200 30 Schuessler et al. [2001] Methanol autothermal reforming Cu- matrix Cu/ZnO 10000 800 Vincent and Gonzalez [2002] Acetylene partial hydrogenation Asymmetric γ-Al 2 O 3 Pd/ γ- Al 2 O 3 3.6 5 Torres et al. [2003] Isobutene dimerization Porous ceramic Zeolite layer 1.4-5.0 3.5 Zhu et al. [2003] Propane selective oxidation -Al 2 O 3 support AgBi V and Mo oxide 2-20 5-10 Fritsch et al. [2004] Isobutene dimerization Solid Acid on porous polymer Solid acid coploym er 410- 610 3.8-150 Groschel et al.[2005] Propene partial hydrogenation Porous polymer Pd particles 3600 1000 Khassin et al. [2005] Fischer-Tropsch synthesis Composite porous material Co-Al co- precipita tion 500- 3000 5000 40 Authors Reaction Membra ne Cataly st Dpore(n m) Tcatalyst (micron) Pellin et al. [2006] Cyclohexane oxidation dehydrogenation Anodized Al 2 O 3 VO, ALD 10-38 70 Mucherie et al. [2007] Propane oxidation dehydrogenation Anodized Al 2 O 3 VO, ALD 40 70 Westermann et al. [2009] Acetylene partial hydrogenation Anodic Al 2 O 3 Pd wet impreg nation 200 60 Minyukova et al. (2012) Hydrogentaion of trigylicerides Carbon composit e membran e CuZn/ CuCr n/a n/a Table 1.7: Literature review of selective integral FTCMR The first application of the FTCMR concept for selective reactions dates back to the early 1990’s, when Zaspalis et al. [1991] used non-separative catalytically active alumina membranes for the dehydrogenation of methanol to formaldehyde. Two different configurations were investigated: In mode A, both reactants enter from the same side but the products are allowed to leave on both sides of the membrane; in mode B all the feed is forced to diffuse through the membrane. In both configurations the membranes exhibit a high catalytic activity for the dehydrogenation of methanol, but the reactant configuration and the mode of operation were shown to be important parameters determining the performance of the catalytic membrane reactor. In mode A, a distribution 41 of the products on opposite sides of the membrane is observed. High purging rates on the permeate side increase the activity of the membrane by preventing back diffusion. The flow-through mode B leads to maximum conversions but minimum selectivities due to the comparatively high residence time of methanol in the catalytically active layer. The observed activity per gr of material is up to ten times higher than in a comparable tubular reactor with a catalytic bed of the same material. This is attributed to the higher effective surface area accessible to the reactants. The heterogeneously-catalyzed selective epoxidation of propylene in an FTCMR is studied by Kobayashi et al. [1996]. Three different catalysts (Cs-Ag, Re-Ag and Ag 2 O) are immobilized in the pores of a microporous glass membrane, and applied to three different reactor systems: a convection-flow reactor, a diffusion-flow reactor, and a packed-bed-flow reactor. Reaction rates are determined for CO 2 and propylene oxide, keeping the total conversion of propylene below 10%. Comparing the three immobilized catalysts, the selectivity towards propylene oxide increases with the amount of intermediate formed on the catalyst, which strongly depends on the types of catalyst and the reactor system utilized. The highest selectivity towards propylene oxide is achieved with the convection-flow reactor and the Re-Ag catalyst, demonstrating hysteresis kinetics depending on the increase or decrease in the propylene concentration. The convective flow through the membrane pores effectively enhances the selectivity towards propylene oxide. 42 Lange et al. [1998] utilized an asymmetric ceramic membrane with an active top layer (thickness 0.2–0.4 µm, pore size 5 nm) for the selective hydrogenation of hexene. Pt was chosen as the catalyst, in spite of its rather poor selectivity for selective hydrogenation reactions, in order to emphasize the influence of the mass transfer resistance. The observed hydrogenation activity is significantly higher than with the comparable conventional catalysts, while the prevention of back-mixing increases the selectivity. The authors noted that the low permeability of the membrane represented the major drawback of the reactor. Oxidative coupling of methane was investigated by Ma et al. [1998] in different membrane reactor configurations including a radial-flow catalytic membrane reactor using tubular -alumina membranes impregnated with samarium nitrate. The premixed reaction mixture of oxygen, methane and helium diluent is fed into the tube-side of the membrane reactor, forcing it to flow through the pores. This increases the linear velocity of reactant gases over the catalyst reducing the external mass transfer limitations. Porous catalytic membranes with different pore diameters (0.02-5 m) were applied in the temperature range of 750–900 o C under various flow rates. The performance of all the FTCMR was reported to be superior to that of a packed-bed reactor made from the crushed catalytic membranes. The membranes with the smallest pores show the best results for both conversion and selectivity, confirming the benefits of the Knudsen regime. The exact location of catalyst inside the membrane was also regarded as an important parameter for the reactor performance. 43 Lambert and Gonzalez [1999] investigated the partial hydrogenation of hydrocarbons (HC) such as acetylene and 1,3-butadiene in a catalytic membrane reactor. The active Pd/γ-Al 2 O 3 membrane layer is prepared by the sol–gel method on a mesoporous γ-Al 2 O 3 tube. The measured pore diameter of 3.6 nm allows for gas separation by Knudsen diffusion. The reactions with 10% HC/Ar mixtures and different HC/H 2 ratios are performed both with separated feed of reactants and in the premixed mode, where H 2 is added to the reactant stream and fed from the tube side. The hydrogenation reactions performed in premixed mode result in the highest selectivity to the partially hydrogenated products while maintaining high conversion without any loss of hydrocarbon species. For the acetylene reaction a high selectivity of ethylene (85%) at reasonably high conversions (76%) is obtained with an excess of H 2 (H 2 /C 2 H 2 = 8) and a temperature of 200 o C at a total flow rate of 18 ml/min. The increased selectivity is explained with the decreased contact time between acetylene and the catalyst. The term ‘short contact time reactor’ is used by Vincent and Gonzalez [2002] for their FTCMR, pointing out the major advantage compared to other reactor concepts. The selective hydrogenation of acetylene is performed by forcing a dilute C 2 H 2 /H 2 /Ar mixture radially through a thin Pd/γ-Al 2 O 3 catalytic membrane (thickness 5μm) into the shell- side. High conversions coupled with a high selectivity are observed, the latter increasing with temperature. The analysis aims at determining whether the membrane can be considered as a thin film, if Knudsen diffusion is contributing, and if a critical membrane 44 thickness exists for ethylene selectivity and conversion. A one-parameter dispersion equation was chosen to model the successive chemical reaction through a thin catalytic layer. A membrane thickness of 2.5 µm is determined to give the maximum ethylene conversion at the given conditions, as a result of tailoring of the residence time. Alfonso et al. [2001] applied an FTCMR making use of an alumina membrane with two distinct catalytic layers in order to carry out the combined oxidative and non-oxidative dehydrogenation of butane. The oxidative dehydrogenation layer is prepared by impregnation with V supported on MgO. The second layer consists of Pt-Sn on γ-Al 2 O 3 prepared by a sol–gel procedure. A premixed (butane/oxygen/inert) feed with an excess of butane enters the membrane and takes part in two consecutive reactions in the two different catalytic layers. The oxidative dehydrogenation of butane takes place in the first layer, which contains the V/MgO catalyst, consuming most if not all the oxygen in the feed. The product stream of the first reaction subsequently enters the second catalytic layer (Pt-Sn/γ-Al 2 O 3 ), where the non-oxidative dehydrogenation takes place. The endothermic non-oxidative dehydrogenation step consumes the heat that is generated in the first reaction (thermal coupling), and both the unreacted butane and the product butene are dehydrogenated. Additionally, the products of the first step, CO 2 and steam, are expected to act effectively as inhibitors of coking in the second catalytic layer. If the catalytic materials in both layers can be regulated so that complete oxygen consumption is obtained within the oxidative layer, the two-layer membrane reactor allows for stable operation with high conversion and selectivity. 45 Schuessler et al. [2001] have developed a reactor design for the autothermal reforming of methanol for mobile applications which involves flowing through a stack of thin porous catalytic discs. Though this setup cannot explicitly be thought of as a membrane reactor it features similar aspects with an FTCMR. Autothermal reactors are characterized by high- temperature gradients in the catalytic zone, which for the reforming reaction leads to the formation of undesired by-products such as CO and CH 4 . To attain nearly isothermal conditions, the heat conduction of the catalytic bed must be improved. Schuessler et al. [2001] achieve these requirements through the use of Cu/ZnO catalyst particles (5–20 m) mixed with Cu powder and pressed into porous disks, which are subsequently sintered at moderate temperatures. The resulting mixed-matrix metal membranes are catalytically active, efficiently transport heat and provide an open structure for mass transport. This allows for an operation at an optimum temperature, resulting in high reaction rates and selectivities. Additionally, the uniform flow distribution is beneficial for the dynamic behavior of the reactor, resulting in negligible fluctuation of stack temperature and product concentration. After attempts to improve reaction yield and selectivity by means of inert membrane reactors in the distributor or extractor modes, Zhu et al. [2003] applied a catalytic membrane reactor in the flow-through mode to the selective oxidation of propane to acrolein. Benefits expected from the application of the FTCMR were because of the high contact surface combined with a short and controlled contact time in the 5–10 m thin 46 catalytic layer which prevents total oxidation. A tubular mesoporous mixed oxide catalytic membrane with a length of 6 cm is prepared by means of a modified sol–gel method. The catalytic reaction is carried out at 400 o C at atmospheric pressures. The FTCMR, operated with reactants flowing from the shell- to the tube-side, is compared with a fixed-bed reactor. The FTCMR produced much higher selectivities towards acrolein (>50%), compared to <10% for the FBR, which produces mainly methanol and acetic acid as the liquid-phase product. A composite zeolite membrane on a porous ceramic tubular support is applied to the catalytic oligomerization of isobutene to isooctene by Torres et al. [2003]. The reactants are forced to permeate through the zeolite membrane film (a few µm thick), which acts as a catalyst for the reaction. The FTCMR attains conversions which are comparable with those in a fixed-bed reactor but no deactivation due to coking and long chain species was observed. This is explained to be due to the short and controlled residence times in the membrane pores that are achieved in the flow-through mode, which prevents further oligomerization. The authors obtained yields of the desired C 8 components close to 60% and theorize that the selectivity towards these intermediate products can be further increased by controlling the residence time. The dimerization of isobutene in an FTCMR was also studied by Fritsch et al. [2004] through the application of polymer-based composite catalytic membranes based on Nafion® and other solid acid catalysts. The membranes are generated by deposition of a 47 porous reactive layer (generated by mixing the catalysts in solution with a polymer binder) on the top of a porous polymeric tubular membrane with a diameter of 20 mm. The influence of different catalysts and different binders on conversion and selectivity were investigated with structural parameters such as porosity, catalyst partition and catalyst accessibility determining the final membrane reactivity. The experiments were performed by adjusting the reactant mass flows through the membrane (the outlet pressure was kept atmospheric). The conversion of isobutene increases with the flow rate (and thus the upstream pressure) up to values of 90% without a significant change of selectivity to the dimer, which reaches around 14%. The authors theorize that the liquid products fill the membrane pores, and therefore a certain flow is required to purge the pores from products and oligomers. For all catalysts the conversion improved by increasing the temperature in a range from 30 to 50 o C, whereas in most cases the selectivity decreased. High selectivities of 86 % were achieved for conversions of 22 %. The authors note the difficulty of properly comparing the performance of membrane reactors with that of conventional reactors in terms of parameters such as space-time yield. Porous polymer membranes based on polyacrylic acid networks containing catalytically active palladium nanoparticles are synthesized by Gröschel et al. [2005]. The effects of porosity, catalyst loading and flow rate on the catalytic behavior of the membrane are examined using the gas-phase partial hydrogenation of propyne to propylene as a model reaction. The resulting membrane porosity is varied between 34 and 72% with a Pd 48 content of 2 mg. The applied flow rates correspond to residence times of 1–4 s, assuming steady state conditions. Larger pores lead to longer residence times but worse catalyst distribution, resulting in an optimum porosity as a compromise between residence time and catalyst distribution. After variation of the catalyst content in the membrane all measured selectivities fall on the same trajectory when plotted against the conversion, indicating that the kinetics are not limited by mass-transfer effects. The activity and selectivity obtained in the membrane reactor are compared with those achieved in a fixed-bed reactor filled with porous or egg-shell catalysts. For equal catalyst load and residence times the porous catalysts show low conversion and poor selectivity, whereas both membrane catalyst and egg-shell catalyst gave equally high conversions and selectivities around 85%, corresponding to an optimized utilization of the Pd. Simulations were performed in order to distinguish between kinetic and mass-transfer control. A simplified first-order rate law was fit to the experimental data. Three-phase Fischer–Tropsch synthesis (CO hydrogenation to liquid hydrocarbons) in catalytic membrane reactors was investigated by Khassin et al. [2005]. For the traditional slurry and circulating fluidized bed reactors the catalyst concentration is too low. In fixed-bed reactors there is a problem in choosing small catalyst particles (to reduce diffusional limitations), as they result in increased hydraulic resistance; using egg-shell type catalysts may offer a solution, but the concentration of the catalyst per unit particle volume is rather low. The authors suggest using heat-conductive ‘flow-through contactor membranes’ as a solution. The membrane is made up of a Co-Al co-precipitated catalyst. 49 The pores with effective radius above 2–3 μm are gas-filled, and are referred to as the transport pores, while the smaller pores are flooded with liquid products and are thus not permeable. As a liquid phase fills the smaller pores and leaves only the larger pores permeable for gas flow, the pressure drop across the membrane is an important parameter for the membrane performance. A broad pore size distribution results in gas flow by-pass which diminishes the selectivity towards higher hydrocarbons. The space-time yield achieved with the membranes (200 kg HC/m 3 h) is much higher than in the traditional reactor designs for Fischer–Tropsch synthesis, and the observed catalytic activity is up to three times higher than in a comparable slurry reactor. A computational study was performed by Albo et al. [2006] in order to predict the transport and kinetic behavior inside these nanostructured catalytic membrane reactors fabricated by anodic oxidation and atomic layer deposition. The selective oxidation of hydrocarbons was chosen as the model reaction in order to assess the prevention of over- oxidation as a result of the short contact times. Detailed atomistic molecular dynamics simulations in pores with diameters between 10 and 150 nm, identified Knudsen diffusion as the dominant mass transfer mechanism, particularly in the smaller pores and at lower pressures, while surface diffusion was only present at temperatures below 700 K. Pellin et al. [2006] synthesized mesoporous, ultra-uniform inorganic catalytic membranes by a combination of anodic oxidation of aluminum and atomic layer deposition. With this approach, pore diameter and wall composition can be controlled along the entire pore 50 length. The pores of the anodic alumina membranes (thickness 70 µm) with initial pore diameters of 40 nm are coated with a thin alumina layer of 1 or 15 nm respectively, resulting in pore diameters of either 38 or 10 nm. The catalytic performance of the membranes is compared to that of an uncoated alumina membrane, after adding one monolayer of vanadium oxide on each. Additionally, the membranes are compared to a conventional high surface area γ-alumina powder catalyst impregnated with vanadium oxide. As a test reaction, the oxidative dehydrogenation of cyclohexane is studied at temperatures of 450 o C, yielding higher selectivities for the partial oxidation product cyclohexene for all three catalytic membranes compared to that of the catalytic powder. The observed reduction of secondary reactions is attributed to a decrease of contact time by a factor of 1000–10,000 compared to the conventional catalyst bed. For higher temperatures, the specificity of the membrane is reported to decrease, but in all cases the membranes outperform the conventional supported alumina catalysts. While the conversions of O 2 and C 6 H 12 is lower for the 10 nm membrane than for those with larger pores, the conversion per mole of vanadium is significantly higher. The same group [Mucherie et al., 2007] using the same approach studied the oxidative dehydrogenation of propane. Catalytic layers composed of vanadium species supported on different metal oxides (Al 2 O 3 , Nb 2 O 5 , TiO 2 ) are compared. Highest reactivity is observed for TiO 2 , which is attributed to the highest dispersion, whereas the selectivity at 3% conversion is highest for the Al 2 O 3 -based catalyst. 51 Westermann et al. [2009] investigated the application of commercially-available anodized alumina membranes (possessing a regular structure of parallel open pore channels with very uniform pore channels with diameters of 200 nm and a length of 60 μm) for selective gas-phase reactions by means of a model reaction namely the selective hydrogenation of acetylene. The membranes were rendered catalytic by impregnation with Pd. The narrow membrane channels intensify the contact between reactants and catalyst resulting in high conversion, while the uniform channel structure leads to a narrow residence time distribution resulting in high selectivity. Westermann et al. [2009] modeled the reactive membrane as a bundle of parallel plug-flow reactors taking into consideration the effects of pore size distribution and axial dispersion. Higher flow- through velocity leads to lower axial dispersion but increases the pressure drop. Deviations from an ideal pore size distribution could, in principle, be counteracted by flowing through several membranes in series. Pressure drop measurements confirmed that the flow through the membrane consisted both of a laminar-flow and a Knudsen-flow contribution, the ratio being varied by altering the pressure. Minyukova et al. [2012] investigated the use of a permeable composite membrane (PCM) for the hydrogenation of fatty acid triglycerides. The authors found that by varying the composition of PCM active component and the conditions of reaction it is possible to transform the triglycerides of fatty acids into valuable chemicals, as fatty esters of higher alcohols. With their high heat conductivity (allowing for isothermal operation of the catalyst bed), their high mechanical strength, their high active component loading per unit 52 volume and easy separation of the reaction products, the authors believe that Cu-based PCMs may have very god potential as a catalysts for transformations of fatty acid triglyserides. 1.3.3. Selective Differential FTCMR Partial hydrogenations of liquid reactants are of industrial interest. The reaction rate is often limited by the low solubility of hydrogen in the liquid phase. Thus, low conversions argue against performing these reactions in a single-pass integral reactor. An alternative that takes advantage of the intensive contact between reactants and catalyst caused by convective flow through the pores, but which also allows for higher total conversions is the use of differential FTCMR in combination with recycling and resaturation. This concept has been applied to the hydrogenation of nitrates in water and to the partial hydrogenation of several unsaturated liquid hydrocarbons. An overview of the membranes and catalysts utilized and reactions studied is given in Table 1.8. 53 Authors Reaction Membrane Catalyst dpore (nm) tcat (µm) Ilinitch et al. [2000] Water Denitrification SiO 2 /Al 2 O 3 ceramics Pd-Cu/γ-Al 2 O 3 1000 4800 Reif et al. [2003] Water Denitrification Symmetric ceramic Pd 3000 1500 Schmidt et al. [2005] Cyclooctadiene ; 1- octyne; phenyl acetylene; geraniol partial hydrogenation Porous PAA Pd 130-380 1000 Fritsch and Bengston [2006] Sunflower oil partial hydrogenation Porous polymer (PES, PAI) Pt/Pd impregnation/casting 3900- 9000 130-230 Purnama et al. [2006] -Methyl- styrene partial hydrogenation -Al 2 O 3 Pd wet impregnation 1900 1000 Schmidt and Schomacker [2006] Sunflower oil partial hydrogenation -Al 2 O 3 Pd/Pt wet impregnation 3000 500 Schmidt et al. [2007] Cyclooctadiene partial hydrogenation -Al 2 O 3 Pd wet impregnation 600- 3000 500 Liguori et al. [2013] 3-hexyn-1-ol partial hydrogenation hybrid zirconia/ polyvinyl alcohol (PVA) Pd nanoparticles (4.6 nm) n/a n/a Table 1.8: Literature review of selective differential FTCMR Ilinitch et al. [2000] studied the reduction of nitrate ions in water by hydrogen in three different types of reactors: (A) a fixed-bed of supported Pd-Cu/Al 2 O 3 catalysts; (B) a differential FTCMR (utilizing a macroporous SiO 2 /Al 2 O 3 ceramic membrane with a 54 uniform structure with a pore diameter around 1 µm which is also impregnated with Pd and Cu) with liquid recycling; and (C) a conventional catalytic membrane reactor operating in the absence of forced flow. In configuration (A) pronounced internal diffusion limitations are observed. Configuration (B) minimizes the diffusional limitations and results in increased catalytic activity when compared to configuration (C). When comparing the specific catalytic activity of the membrane to the one of the conventional powder catalyst, the latter shows a higher value, indicating that the active material inside the membrane may not be fully utilized. Reif et al. [2003] compared the FTCMR concept for aqueous nitrate and nitrite reduction with the catalytic diffuser concept, in which the reactants are delivered from different sides of the membrane. Both setups induce active contact between the reactants and the catalysts, while only the FTCMR concept allows for the elimination of pore diffusion and very short contact times. Due to high pressure drops in the FTCMR mode, conventional ceramic membranes with pore diameters of 100 nm in the active layer are found to be unsuitable. Consequently symmetric supports with pore diameters around 3 m are coated with Pd. Flowing from the outside to the inside reduces plugging of the membrane. For the nitrite hydrogenation in FTCMR mode, a significantly higher activity and a lower formation of ammonium is observed, both depending on the trans-membrane flux. 55 For the partial hydrogenation of several unsaturated hydrocarbons Schmidt et al. [2005] used the differential FTCMR concept, and compared its performance to that of the slurry reactor and of the fixed-bed reactor. The FTCMR is constructed as a loop of a saturation vessel and a porous membrane made from cross-linked polyacrylic acid integrated with catalytic palladium nanoparticles. Four different unsaturated reactants are investigated: Cyclooctadiene, 1-octyne, phenyl-acetylene, and geraniol. The liquid reaction mixture is resaturated with hydrogen up to 100 times. With the applied flow velocity of 2 × 10 −4 m/s the convection of the reactant stream is at least one order of magnitude faster than the diffusion velocity (of the order of 10 −5 m/s). The membranes had a thickness of 1mm, pore sizes between 130 and 380 nm, porosities between 35 and 70 % and a Pd content of 1wt.%, and were placed on a porous support plate in the reactor module. For the hydrogenation of octyne the desired reaction is much faster than the undesired reaction, so that little influence of mass transfer is observed. For phenyl acetylene and cyclooctadiene the desired reaction is still faster than the undesired one, but mass transfer is more significant, which can be seen in the increased selectivity obtained in the membrane reactor and slurry reactor compared to the fixed-bed reactor. For the geraniol reaction, the rates are nearly equal. For this reaction the FTCMR reaches even higher selectivities than the slurry reactor, whereas the activity is lower due to the strong decrease of hydrogen concentration during each pass through the membrane, with the result that the average hydrogen concentration is substantially lower than the saturation concentration. 56 Purnama et al. [2006] performed the partial hydrogenation of -methyl-styrene to cumene in a catalytic membrane reactor using tubular macroporous -Al 2 O 3 membranes impregnated with Pd. The measured productivity, defined as conversion per time and catalyst mass, in the FTCMR exceeds that of other types of membrane reactors as well as other conventional reactors, like slurry, trickle-bed, and bubble column reactor. The authors note that the flow rate must be adjusted to the catalyst mass, so that the hydrogen conversion reaches 100% exactly when the reactants leave the membrane. Fritsch and Bengtson [2006] developed catalytically active porous membranes for the selective hydrogenation of viscous liquids such as sunflower oil in a membrane reactor. The high viscosity and the low hydrogen solubility induce severe mass transfer limitations for such reactions. Industrially, vegetable oil hydrogenation takes place in stirred tank reactors with finely dispersed catalyst at temperatures between 170 and 200 o C, at 2–5 bar of H 2 pressure. In this process, up to 50% of the undesired trans-isomerized fatty acids are generated. By applying the FTCMR concept, the content of trans-isomers can be reduced while at the same time the immobilization of the catalyst inside the membrane pores avoids expensive filtration steps. For their study the authors prepared porous polymer membranes from polyethersulfone and polyamideimide with and without inorganic alumina filler. These membranes show high water fluxes of about 30,000 L/(m 2 h bar) and oil fluxes of 900–2000 L/(m 2 h bar) at 60 o C. Catalysts were introduced either by wet impregnation of the membrane in a catalyst precursor solution followed by calcination and chemical reduction or by addition of ready-made supported catalysts to 57 the membrane casting solutions. The catalytically activated membranes with Pt-contents between 0.1 and 1 g/m 2 were applied to hydrogenate refined sunflower oil. Within 6 h about a half of the linoleic acid is hydrogenated, which is the major compound in sunflower oil triglycerides. Pt shows a similar activity to Pd catalysts and generates less trans-isomers. The membrane activity is stated to be sufficiently high to allow the industrial use of the FTCMR for the hydrogenation of viscous liquids. The same reactor approach is applied to the hydrogenation of cyclooctadiene, reaching the same selectivities as in slurry experiments and outperforming the comparable fixed- bed reactor [Schmidt et al., 2007]. For the same amount of Pd, membranes with smaller pores (0.6 m) are more active than membranes with bigger pores (1.9 or 3.0 µm). The experiments were also transferred from the laboratory to the pilot scale using a 27- capillary catalytic membrane module. The obtained space-time yields were again much higher than those reported for conventional fixed-bed or trickle-bed pellet catalysts or slurry catalysts in bubble columns [2008]. The encouraging scale-up results make this FTCMR setup among the most promising candidates for the first industrial applications of catalytic membrane reactors [2007]. Schmidt and Schomäcker [2007] also used a differential FTCMR for partial hydrogenation of sunflower oil using n-heptane as solvent and compared the reactor performance with that of a slurry reactor with powder catalyst. A porous γ-Al 2 O 3 membrane is impregnated with Pd or Pt as the active catalyst. The membrane reactor 58 reaches significantly lower contents of the fully hydrogenated stearic acid, but also higher contents of undesired trans-isomers than the slurry reactor. Both results are attributed to the lower hydrogen concentration present in the membrane reactor. Liguori et al. [2013] showed that palladium nanoparticles can be easily grown within a hybrid zirconia/PVA membrane matrix, to prepare contactor type catalytic membranes suitable for application to partial hydrogenation reactions in the liquid phase under stable conditions. The authors showed that, the partial hydrogenation reaction of the challenging substrate 3-hexyn-1-ol could be for the first time carried out by a catalytic-type membrane, showing no significant loss of catalytic performance over prolonged reaction period. 59 1.4 The Siloxane Problem As our study further progressed in the use of FTCMR for removal of chlorinated and sulfided compounds from landfill gas, we came across a particularly troubling class of compounds called siloxanes present in landfill gas. They are key components of many commercial and consumer products, such as detergents, shampoos, deodorants, and other cosmetics which unfortunately find their way into various landfills where such products are discarded. Because of their increased use, siloxanes have emerged in recent years as one of the most difficult contaminants to control in landfill gas and biogas. Siloxanes are organic compounds containing alternating silicon and oxygen atoms in their backbone, with various attached groups [Nair et al., 2012; Nair et al., 2013]. They are made primarily through the hydrolysis of chlorosilanes (which are themselves produced via the chlorination of Si by CH 3 Cl), the process resulting in the basic raw materials (monomers) from which a host of products are made. Most siloxanes volatize quickly into the atmosphere, during use, where eventually they degrade into carbon dioxide, silica, and water (Dow Corning, 1997). An increasing number of siloxanes, however, end-up in wastewater or in landfills when consumers rinse or discard products. The type and nature of these siloxane compounds (which will be discussed in more detail in Chapters 3-5) creates an inherent problem for landfill gas users. The oxidation of these compounds creates silica particulates (SiO 2 ). These particulates are highly abrasive and they deposit on catalysts and block the pores of the membrane. Which is why FTCMR cannot be used for the removal of these class of compounds. For these compounds (when present in the 60 LFG) we need a different removal technique; some of our preliminary efforts testing this technique which will be discussed in Chapter 5 of this Thesis. 61 1.5 Our Study Our study has three different focus areas, as follows: 1.5.1. Use of FTCMR for Removal of Chlorinated and Sulfided Contaminants in LFG (This work in this Thesis is supported by the National Science Foundation. Some of the discussion in this section are, thus, adopted from the related proposal to the Funding Agency). As the discussion above indicates, FTCMR are finding a broad range of applications. What we propose here, therefore, is to study their use as part of a novel catalytic oxidation technology appropriate for LFG clean-up, which potentially overcomes many of the current technical challenges that the conventional catalytic technology faces. The FTCMR we propose will be, in addition, endowed with an oxidation nanocatalyst. From the above literature review of catalysts utilized for the oxidation of VOC typical of those encountered in LFG, we choose Pt as the active component and mesoporous asymmetric membranes (see below). Further development of the proposed technology is important, as economical, environmental and energy advantages will be realized, with a process cost-effectively removing the NMOC from LFG. 62 The FTCMR we propose to utilize for LFG clean-up, making use of a porous ceramic membrane with catalyst deposited on the surface of its pores, is shown schematically in Fig. 1.1. The premise of the operation of this FTCMR is that all the NMOC contaminants found in LFG are forced through the pores and are brought, by the gas flow, directly in contact with the catalyst surface. If the flow through the membrane is maintained in the Knudsen flow regime (KFR), collisions with the membrane pore walls (and hence the catalyst sites) prevail over gas-gas collisions. Hence, in the KFR, all of the NMOC contaminants of LFG are guaranteed to come into contact with the catalyst surface and to be reacted, even though the membrane layer thickness is in the μm range. This presents potentially a great advantage over the situation one encounters with a granular packed- bed or a monolithic converter, in which the NMOC must diffuse through the catalyst support pores to the active catalyst sites. In KFR, very rapid and high conversion (“deep” oxidation) of the contaminants can be achieved. By contrast, in a packed-bed or in a monolithic reactor, the collision of the trace contaminant molecules with the catalyst is inefficient; this becomes particularly problematic as the contaminant concentration decreases, and external mass transfer limitations increase. The pressure drop across the FTCMR is thought as a potential challenge. However, the ultra-thin reactor thickness and the high permeability of the membranes significantly minimize the pressure drops needed, with transmembrane pressure differences of 1-2 psi being, typically, sufficient. Since the LFG produced is typically pressurized, no additional compressor costs are, therefore, needed to operate the FTCMR 63 For the FTCMR to operate in the KFR the mean free-path of the molecules must compare with the membrane pore size. The mean free-path of gas molecules at atmospheric pressures and room temperature is in the ~1,000Å (0.1μm) range. If the pore size is less than the mean free-path, molecules will collide with the pore wall more frequently than with each other. The permeance of gas molecule through a straight cylindrical pore under the KFR is as follows: [Permeance a ] = c/(M a RT) 1/2 (1.1) where c is a constant, M a is the molecular weight of gas a, R is the gas constant, and T is the temperature. A simple criterion to assess whether the operation is under the KFR is to compare the measured permeance ratios of the gas components to the ideal ones expected in the KFR (known as Knudsen selectivity) [Permeance a /Permeance b ] = (M b /M a ) 1/2 . For viscous flow of gas in pores, this permeance ratio is 1, indicating complete lack of Knudsen type of flow. The potential advantages of the proposed FTCMR operated in the KFR are summarized below: Less Intraparticle Mass Transfer Resistance…One of the major advantages that FTCMR offers is low intraparticle mass transfer resistance. The main reason that FTCMR achieves higher conversion is because in the FTCMR there is direct contact between the reactant and the catalyst as the reactant gas 64 is transported to and from the catalyst by Knudsen diffusion. The same cannot be said for conventional catalyst. In a conventional catalyst the reactant molecules have to counter-diffuse against the products through the catalyst pellets to reach the active catalyst sites in them. This results in slower transport of the reactant and product gases to and from the catalyst sites. Figure 1.1 shows this fundamental difference between the two reactor concepts. Because mass transfer limitations are eliminated, the intrinsic catalyst reactivity in an FTCMR can be fully exploited. Less by-product formation…In addition to the more effective catalyst- reactant collisions, FTCMR is also superior because it allows less inter molecular collisions. These inter-molecular collisions can result in the formation of by-products. In the FTCMR the molecules collide mostly with the catalyst surface rather than with each other, thus effectively increasing the molecule-active site contact that controls the overall reaction rate, and avoiding gas phase parallel reactions. 65 Less Catalyst Poisoning…The reaction products (e.g., HCl and SO 2 ), can reduce the catalyst activity. Since FTCMR has a low mass transfer resistance the gaseous products can quickly diffuse out, thus minimizing or completely eliminating this particular problem. Pore diameter, ~0.1 µm Contaminant molecule Mean free path, ~0.1 µm Expanded View Simplified straight pore through the porous thin film Under Knudsen flow regime (KFR), the molecules collide with the pore wall more frequently than with each other as shown here. Porous ceramic substrate, 0.5 µm pore dia. and ~1 mm thickness α-Al 2 O 3 porous thin film with 0.05 to 0.2 µm pore size available, and 10 to 50 µm thickness Feed with contaminants VOC Depleted Permeate with reaction products Inside tubular surface coated with nanocatalyst for catalytic oxidation Feed with contaminants Permeate with reaction products FAST REACTION Pore Flow Reactor Concept Bulk flow occurs through the pores. Reactants are (rapidly) convected to the active catalyst site. Reaction rate is very fast in comparison to packed catalyst bed. VOC Laden Feed VOC Depleted Permeate Catalyst Support Pore Granular Catalyst Support VOC Depleted LNG Stream SLOW REACTION RATE Packed Bed/Monolithic Reactor Bulk flow occurs normal to the pores in granular or monolithic reactors; reactants (and products) must (slowly) diffuse to the active catalyst site. The slow diffusion rate reduces the overall reaction rate in comparison with the pore flow reactor given above. Catalyst Dispersed On Support VOC Laden LNG Stream Figure 1.1: Comparison of the FTCMR concept with the granular fixed-bed and monolithic reactor technologies for VOC destruction. 66 Complete Destruction of the VOC even at Trace Concentrations…Since in FTMCR all the gas are forced to flow through the membrane, all the reactants are ultimately brought into contact with the catalyst surface, and are hence reacted. In contrast, for conventional reactors, a larger reactor would be required to achieve complete destruction, because mass transfer dominates the reaction rate at trace levels. Lower Operating Temperature…Because of the increased contact between the catalyst and the reactants the FTCMR also does not need as high temperatures to operate as a conventional reactor. This will greatly assist in preserving CH 4 in the LFG gas. In addition, when using the nanocatalysts, the mild FTCMR operating temperature will bring advantages in minimizing their sintering; this is one of the major limitations in the use of such materials in conventional reactors. As noted above, the FTCMR has been demonstrated to be more effective for the destruction of trace model VOC in contaminated air steams in comparison to the conventional reactors. For example, destruction of MEK and toluene in air were compared in a FTCMR and in a monolithic reactor [Pina et al., 1996]. The FTCMR achieved 100% toluene conversion, independent of the feed concentration. Further, MEK oxidation is complete at ~200 C for an FTCMR, as shown in Fig. 1.2 [Pina et al., 1997] versus <95% conversion at ~300 C in the monolithic reactor. Note that, as Fig. 1.2 67 (right) shows, the conversion in the monolithic reactor did not exceed 95% at any temperature, indicative of mass transfer being rate limiting; this is typical in the conventional reactor design. Similar leveling-off in conversion with increasing temperature is well known in VOC combustion on monolithic catalysts [Mazzarino et al, 1993] (it is important to note that for the experiments in Fig. 1.2, even using a high activity nanocatalyst would not improve the conversion, since it is mass transfer limited). Two additional points, pertaining to the effect of VOC concentration should be noted (Fig. 1.2, left). First, even at the lowest concentration, >99% conversion is achieved. Second, the temperature to achieve >99% conversion decreases with decreasing concentration. These results demonstrate that the FTCMR can achieve high oxidation efficiency even at dilute VOC concentrations. No indication exists of “leveling off” in conversion at any VOC concentration, unlike that observed in a monolithic contactor (Fig. 1.2, right). Figure 1.2: Left, conversion of toluene in a FTCMR as a function of temperature; right conversion of 1,200 ppm MEK and comparison between a FTCMR and a monolithic reactor. 68 In summary, preliminary results have evidenced the advantages of the FTCMR for NMOC oxidation. The unique efficiency results from the fundamental characteristics of the reactor operated under the KFR, where all the contaminant molecules are guaranteed to have multiple contacts with the catalyst surface. Little concrete is known, on the other hand, about the complex reaction and transport processes that occur in the FTCMR. The emphasis in this project, therefore, is on gaining a better fundamental insight in these phenomena, and on understanding and modeling the catalytic combustion of the complex LFG heteroatom compounds. A better fundamental understanding will lead to the technological advances needed for the further development and commercial application of the FTCMR concept. 1.5.2. Study the Impact of Siloxanes in LFG on Common Appliances such as Engines and Furnaces This part of the study was done in collaboration with the Southern California Gas Company. In the investigation, an internal combustion engine and a residential furnace operating on natural gas (NG) spiked with siloxanes have been studied experimentally with the goal of understanding the impact of siloxane impurities on their performance. These impurities are shown to completely decompose during NG combustion in the engine to form silica microparticulates [Nair et al., 2012; Nair et al., 2013]. These coated the internal metal surfaces in the engine (e.g., the piston heads) as well as the engine’s 69 oxygen sensors, spark-plugs, and also collected in the engine oil. They also coated the flame sensor of the furnace the condenser coils, and the tailpipes, and they also accumulated in the water that condenses on the furnace’s flue vent. The coating of the flame sensor presented the key challenge for the furnace operation because after a certain period of exposure to the siloxanes it was no longer able to sense the flame, thus causing the furnace to stop operating. A certain fraction of them, furthermore, are carried out of the engine in the flue-gas and they deposit inside a catalyst monolith bed placed downstream of the engine resulting in its severe deactivation. Similarly, a fraction of the silica particles of submicron size got entrained in and escaped through the flue-gas exiting the furnace which could also be a potential health hazard (see more detailed discussion in Chapter 4). All these results point out the critical importance for engine and furnace performance of adequately removing these siloxane impurities from NG prior to its use. 1.5.3. Use of UV Photodecomposition for the Removal of Siloxanes This part of the study was done in collaboration with GC Environmental, Inc. The main aim of this study was to use a UV Photodecomposition reactor to remove the harmful siloxanes from LFG and augment the use of FTCMR for LFG clean-up. This Thesis (Chapter 5) describes our preliminary efforts to evaluate the technical feasibility and environmental implications of a novel technology for the treatment of biogas, which 70 involves the in situ conversion of siloxanes, typically found in the gas, into inert silicon dioxide via a photochemical process. The approach involves using high energy UV light to convert the siloxanes into SiO 2 powder which can be conveniently removed from the biogas via a downstream filter. In our study, the technique is shown to be very effective with high siloxane conversions attained in the laboratory.. The potential advantages of this technology are: 1. Less operating costs: Since the use of UV photodecomposition involves the use of a UV light, the only operating costs is the amount of power consumed by the UV light. 2. Low maintenance: The only maintenance required for the UV lamp in the periodic cleaning-up of all the silica that coats the glass cover of the UV lamp. 71 Chapter 2 A Flow-Through Catalytic Membrane Reactor for Landfill Gas Clean-up 72 2.1. Introduction Landfill gas (LFG) is potentially a valuable renewable fuel because of the substantial concentration of methane that it contains. But the presence of various organic trace compounds, collectively known as non-methane organic compounds or NMOC, in LFG many of which contain in their structure heteroatoms such as chorine, fluorine, sulfur, etc. [Ohannessian et al., 2008; Abatzoglou et al., 2009; Shin et al., 2002; Boulinguiez et al., 2010; Eklund et al., 1998], presents challenges for its use. Burning the gas for power generation (or even when flaring it in order to dispose it), releases gases such as hydrogen chloride, sulfur dioxide, etc. (if such impurities are not removed), which are toxic and harmful to both humans and the environment. Therefore, there is a strong incentive today to develop effective technologies for removing the toxic compounds from landfill gas prior to its utilization as a fuel (for recent reviews, see [Nair et al., 2012; Nair et al., 2013]). The most common method of controlling the emissions of trace compounds in LFG is through adsorption, for example, by using activated carbon (AC) [Pradhan et al., 1999] or silica gel [Schweigkofler et al., 2001]. This approach has only proven marginally effective for LFG clean-up, however. This is because of the broad array of the low- concentration NMOC found in LFG; a number of these would normally not present a problem (e.g., various aromatic hydrocarbons) when LFG is combusted to produce electricity. However, when they pass through the adsorption beds these highly adsorbable species coat the active surface of the adsorbents and diminish their adsorption capacity 73 towards the more harmful, but significantly less adsorbable, heteroatom containing compounds (e.g., vinyl chloride or VCM). This then necessitates frequent regeneration of the beds, which consumes energy, and eventually requires replacing the adsorbents all together, which is costly. (Water vapor, which is always present in LFG, interferes also with the operation of some of these adsorbents, e.g., silica gel, because they are hydrophilic). Refrigeration has also been utilized for the removal of condensable VOC from LFG and has been shown effective [Schweigkofler et al., 2001; Dewil et al., 2006]. But the energetic requirements for the application of deep chilling systems are very high, as experience for sewage gas treatment indicates, so that these processes are generally perceived not economically feasible [Urban et al., 2009]. One of the problems with the physical methods for LFG clean-up is that the various impurities are not destroyed but they are concentrated instead on a different medium (e.g., the adsorbent) from which they have to be removed and disposed. Reactive processes, on the other hand, offer the advantage that they mineralize the NMOC, so no further disposal is needed. They have also received attention for the treatment of LFG. For example, a process involving the destruction of the NMOC in LFG via catalytic hydroprocessing over Co-Mo/Al 2 O 3 or Ni-Mo/Al 2 O 3 catalysts in an atmospheric pressure reactor, together with conventional adsorption technology for the removal of the HCl and H 2 S reaction by-products has been developed previously by this group (in collaboration with EPRI), and field-tested at the Anoca landfill in Minnesota [He et al., 1997]. The process was shown effective in reducing the levels of the Cl and S containing NMOC, 74 down to their analytical detection limits, and no catalyst deactivation was observed during 1000 h of field-testing. However, the amount of H 2 required (~5 vol % of LFG treated) is undesirable in comparison with the ppm-level of contaminants treated, and is a challenge to generate and handle, particularly for small landfills or in remote locations; furthermore, the H 2 safety risks have not been accepted by the Landfill industry. Catalytic oxidation has also been shown effective for the treatment of halogenated [Krishnamoothy et al., 2000; Toledo et al., 2001; Musialik-Piotrowska et al., 2002] and sulfided compounds [Chu et al., 2001; Devulapelli et al., 2008] and has been investigated for the treatment of LFG [Urban et al., 2009]. Krishnamoorthy et al., 2000 for example studied the catalytic oxidation of 1,2-Dichlorobenzene (a commonly occurring LFG component) on a variety of transition metal oxides such as Cr 2 O 3 , V 2 O 5 , MoO 3 , Fe 2 O 3 , and Co 3 O 4 supported on TiO 2 and Al 2 O 3 and found them very effective. Toledo et al. [2001] have reported that Pt-based catalysts have the longest life (or the lowest deactivation rate) for the decomposition of dichloromethane (DCM) and trichloroethylene (TCE). The authors reported that chromia- and vanadia-based catalysts did not resist the attack by the nascent Cl, and were quickly lost from the surface. They also reported that the deactivation detected in some Pt-based catalysts can be attributed to the catalyst support (Al 2 O 3 ) and not the Pt. Musialik-Piotrowska et al. [2002] studied the catalytic oxidation of trichloroethylene oxidation in the presence of non-halogenated compounds, like toluene and ethanol. Over Pt catalysts water had a positive effect on the conversion and the selectivity of the reaction to HCl (this is an important finding for the use of such catalysts in LFG as well as it also contains water vapor as well). Chu et al. 75 [2001] studied the catalytic incineration of ethyl-mercaptan, typically emitted from the petrochemical industry, over a Pt/Al 2 O 3 fixed-bed catalytic reactor. The results show that the conversion of C 2 H 5 SH increase as the inlet temperature increases and the space velocity decreases. C 2 H 5 SH was reported to have a poisoning effect on the Pt/Al 2 O 3 catalyst at lower temperatures, which can be reduced by raising the operating temperature to a sufficiently high value. Devulapelli et al. [2008] studied the oxidation of dimethyl sulfide (DMS) with ozone. High conversions were achieved only at high temperatures (>250 °C). The authors report that the best catalytic activity was obtained over 10 wt.% CuO-10 wt.% MoO 3 supported on γ-Al 2 O 3 , which showed 100% DMS conversion and high selectivity (~96%) towards complete oxidation products such as CO 2 and SO 2 . One of the latest techniques for removing H 2 S has been the use of iron-chelated solutions which can act as a pseudo-catalyst while having high affinity for sulfides [López et al., 2012]. Urban et al. [2009] have investigated the use of catalysts for removing chlorinated, sulfided and other VOCs from LFG. The authors tested the use of V 2 O 5 –TiO 2 in lab-scale tubular reactor to remove VOCs such as TCE, CFC-113, benzene, toluene and H 2 S from simulated LFG (CH 4 45vol%, CO 2 25vol%, O 2 1vol% and balance N 2 ) and they were able to achieve >90% conversion for almost all the compounds except for the CFC’s. The model substances used for the tests to represent the typical minor compounds, i.e., H 2 S, benzene, toluene, TCE and the chlorofluorocarbon CFC-113, were enriched into the system by conventional evaporation. One of the catalysts is composed of only V 2 O 5 and TiO 2 (catalyst B), while the other catalyst (catalyst A) contained also significant amounts of WO 3 , MnO 2 , CuO, and Fe 2 O 3 , which may act as promoters to enhance the activity of 76 the catalyst. Studies were carried out in a lab-scale packed-bed tubular reactor at 300 o C for dry and humidified simulated LFG. In these experiments Catalyst A showed good conversion for the VOC compounds such as benzene and CFC-113 which are tough to oxidize compared to Catalyst B. Flow-through catalytic membrane reactors (FTCMR), which are the focus of this study, are receiving current attention for a variety of interesting applications (for recent reviews, see Westermann and Melin [2009], Motamedhashemi et al. [2011] and Soltani et al. [2013]). These reactors are a special class of the so-called contactor-type membrane reactors [Soltani et al., 2013], which make use of porous catalytic membrane that bring together the reactants that are forced to flow through them. The function of the membrane is to provide a reaction environment with short controlled residence time and high catalytic activity. In classical fixed-bed reactors the conversion is limited by pore diffusion in the catalyst pellets. In the FTCMR the catalyst is placed inside the membrane pores and the reactants flow convectively through the pores, the ensuing intensive contact between reactants and catalyst resulting in a high catalytic activity. The motivation for applying an FTCMR is to either aim to reach complete conversion in minimum time or space, taking advantage of the high catalytic efficiency, or to reach maximum selectivity for a given reaction due to the narrow contact time distribution. FTCMR have found use in the environmental field. Illinitch et al. [2000] and Illinitch et al. [2003] used FTCMR for the removal of harmful nitrates generated from industrial and 77 agricultural use. Kochkodan et al. [2008] studied the efficiency of photocatalytic membrane reactors for removing harmful chemicals like phenol, chlorophenol, nitrophenol and hydroquinone from water. For flue-gas cleaning, Saracco et al. [1994] catalytically activated ceramic porous filters by applying an alumina layer to decompose chemical pollutants, such as NO x and VOC that were forced to flow through it. An FTCMR operating in the Knudsen-flow regime (KFR) was investigated by Pina et al. [1996] for the destruction of toluene in air. Pina et al. [1997] also studied the catalytic oxidation of toluene and methyl ethyl ketone (MEK) using Pt/Al 2 O 3 membranes. Zalamea et al. [1999] utilized catalytic membranes with wider pores, operating in the mixed-flow regime for the combustion of MEK and n-hexane in order to minimize the pressure drop which was identified as a potential drawback when operating in the KFR. Picasso et al. [2007] evaluated Ce/Mn-Al 2 O 3 and Ce/Zr-Al 2 O 3 catalysts for the combustion of n-hexane using FTCMR. The authors found that the Ce/Mn-based FTCMR is more active than Ce/Zr-based ones for the combustion of n-hexane. Maira et al. [2003] investigated the effect of the flow configuration in a MR on the gas-phase photocatalytic oxidation of trichloroethylene (TCE), applying catalytic zeolite membranes and a catalytic stainless steel plate both impregnated with TiO 2 photocatalyst. They found that the degradation of TCE was higher (48%) in the flow-through mode, than in the monolith (parallel) mode (20%). The gas-phase oxidation of methanol, as a model VOC was also studied in a photocatalytic membrane reactor by Tsuru et al. [2003]. FTCMR studies have also been carried out by our group [Motamedhhashemi et al. 2011]. The FTCMR was used to oxidize a chemical warfare simulant dimethyl methyl 78 phosphonate (or DMMP). The FTCMR was able to completely remove the DMMP from an air stream at temperatures as low as 200 o C. As the discussion above indicates, FTCMR are finding today a broad range of applications in the environmental field. This paper is a study of their use for LFG clean- up, aiming to potentially overcome the current technical challenges that the conventional catalytic technology faces, specifically catalyst deactivation due to the inefficient way of products diffusing out from the catalyst and increased temperatures for higher conversion. To the best of our knowledge, this is the first time that this novel reactor concept is applied to this important and practical problem. In what follows, the experimental set-up and the techniques utilized are first described. The experimental observations are then presented. A model is then presented and is validated by the experimental data. The model is then utilized to compare the behavior of the FTCMR with monolith reactors on an equitable basis. (This is a key aspect of the work presented in the Chapter as such direct comparisons are challenging). Discussion and conclusions are then presented on the potential use of FTCMR for LFG clean-up. 79 2.2. Experimental Section In this study porous tubular ceramic membranes were utilized which were provided for the project by our industrial partner in this project Media and Process Technology, Inc. (MPT) of Pittsburgh, PA. These membranes are asymmetric in nature consisting of a macroporous substrate (known as the membrane support) which is made of α-alumina. On the top of this support one deposits one or more mesoporous layers (α-alumina or γ- alumina) either by slip-casting or sol-gel techniques. For the membranes in this study two such layers are placed on the inside of the support tubes. One of the objectives of the membrane characterization efforts is to gain a better understanding of the pore structure characteristics of such membranes in order to meet the technical requirements of the LFG application. Average pore size and pore size distribution is, of course, important in determining the flow regime that prevails during FTCMR operation. The characteristics of these membranes, including the average pore size, pore size distribution (PSD), surface area and thickness of the individual layers were determined previously by Dr. Motamedhashemi in our Group via flow perporometry, scanning electron microscopy (SEM), and BET analysis based on N 2 gas adsorption, and are tabulated in his Thesis [Motamedhashemi , 2012]. For the modeling results to be presented later in this Chapter some of the membrane characteristics were adapted directly from there. 80 Measuring the transport characteristics of these membranes is, of course, a key way of identifying their ability to function as FTCMR and to determine the potential presence of defects and pinholes that may degrade performance. Such measurements, furthermore, allow one to gauge whether Knudsen flow prevails though the membrane during the FTCMR experiments. For the transport measurements we utilized He and Ar (which are inert gases thought to neither adsorb on the surface of the membranes nor to condense in their structure), and we measured their permeation rates at various temperatures. Most of the membranes used in this study were 9 cm in length, but some longer (25 cm) membranes have been studied as well. The ends of the membranes (~1 cm long on each end) were glazed using a nonporous gloss glaze (Duncan GL612) so as to avoid leaks during the sealing in the reactor, leaving an active length of ~7 cm (23 cm in the case of the 25 cm long membranes) for each membrane for permeation/reaction. For the measurements, the membrane is installed in a stainless steel (SS) module utilizing high-temperature graphite O-rings for sealing (for a schematic of the experimental system, see Fig. 2.1 below). In this system the inert gas is fed continuously through the membrane tube. During the experiments the pressures of the gas at the inlet and outlet of the retentate side and the permeate sides are measured (using Omega PX 180-100GV pressure transducers connected to an Omega DP3300-MV read-out box) together with the pressure drop between the inlet in the feed side and the permeate side (using an Omegadyne model # PX409-030DDUI differential pressure gauge connected to an Omega DP3300-MV read-out box). The flow of the gas transporting through the 81 membrane to the permeate side was measured with the aid of an Alltech 0-100 cc bubble flow-meter. The membrane module was placed in a 6-zone ATS furnace model #3210, and its temperature was controlled with the aid of an Omega CN1507 controller. The pressure of the system was varied by a VWR pressure regulator (Cat No. 55850-260). Module isothermality was verified with the aid of an Omega k-type sliding thermocouple connected to an Omega CN9000a display unit was used. Figure 2.1: Experimental set-up of the permeation experiments 82 The membranes were rendered catalytic using Pt as a catalyst via a wet impregnation technique. For that we used an 8wt% hexachloroplatinic acid solution [H 2 PtCl 6 · 6H 2 O ] [Sigma Aldrich part # 206083]. The membrane was first soaked by placing in a small conical flask and continuously stirred using a magnetic stirrer to ensure homogeneity. Before impregnation the outside surface of the membrane was covered with a Teflon tape. The membrane was kept in the conical flask for a certain period. The wet- impregnated membrane was then dried at 380 K in an oven for 24 hr and calcined at 400 o C in an air stream at 120 cc/min for 2hr. It was then installed inside the reactor reduced by flowing H 2 at 400 o C at 30 cc/min for 2 hr. With this procedure a number of catalytically impregnated membranes were prepared which are shown in Table 2.1 below. (All these membranes, other than membrane CM4 are 3-layer membranes with a top layer having a nominal pore diameter of 100 Å – as reported by MPT, an intermediate layer with a nominal pore diameter of 500 Å and the support, whereas the CM4 membrane consists of the support with the top layer having a pore diameter of 500 Å). 83 Membrane # Pt wt% Type of experiments using the membrane CM1 0.72 FTCMR light-off temperature, Monolith light-off temperature CM2 0.04 Catalytic membrane permeance, FTCMR-Monolith experiment, FTCMR deactivation test CM3 0.21 EDX line analysis CM4 0.18 CO Chemisorption test CM5 1.2 25 cm long, FTCMR-Monolith experiment CM6 0.44 Ar permeance for catalytic and unimpregnated membrane CM7 0.05 BSE imaging CM8 0.05 TEM imaging EM1 N/A He-Ar permeance for unimpregnated membrane permeance EM2 N/A Unimpregnated membrane for SEM analysis Table 2.1: Table showing the different membranes prepared and their uses The Pt content of the membranes was measured by weighing them before impregnation and then weighing them again after drying using a Sartorius mass balance (model # CP225D). To characterize the Pt-impregnated membranes, SEM-EDAX analysis was conducted on the membrane to determine the Pt distribution and also to find the thickness of the membrane layer. CO Chemisorption was used to characterize the state of the metal catalyst on the CM4 membrane using a Micromeritics Chemisorb 2720 apparatus. TEM analysis was also utilized for finding the size of the Pt particles deposited on the CM8 membrane. In the FTCMR experiments a simulated landfill gas (SLFG) was utilized whose composition is shown in Table 2.2 below. 84 Compound Composition 1,3- Dichlorobenzene 5 ppm Trichlorofluoromethane 50 ppm Carbonyl Sulfide 50 ppm Dimethyl Sulfide 50 ppm Vinyl Chloride 50 ppm Oxygen 1% Nitrogen 9% Carbon Dioxide 38% Methane Balance Table 2.2: Table showing the Simulated Landfill Gas (SLFG) components and their composition As noted in the Introduction, in past experiments by our group this SLFG proved to be a good test gas to simulate in the laboratory real field-scale conditions (He et al., 1997). As Table 2.2 indicates, in addition to the common fixed gases it contains five NMOC commonly found in real LFG. The experimental set-up for both the reaction and the permeation studies is given in Fig. 2.2 below. 85 Figure 2.2: Experimental set-up of the reaction experiments A Brooks 5850 CH 4 mass flow controller (MFC) was used to control the flow of the SLFG. To analyze the components a GC-MS (HP6890 GC attached to a HP5973 MS) with an HP-5 MS capillary column was used. (For the analysis of the five SLFG VOC, the GC oven was kept at a temperature of 30 o C for 5 min, and then was ramped up to 200 o C at a rate of 20 o C /min). For the calibration of the GC-MS we used the SLFG itself by diluting it with Ar. The rationale for doing so is that in the FTCMR experiments the main goal is to measure the conversion of the various components, knowing the exact precise concentration in ppm is not terribly important. (Typical calibration results for the 86 various VOC components indicate very good linearity). For comparing experimentally the FTCMR behavior to that of the monolith, experiments were repeated for the same conditions with the same membrane but in the “monolith mode” of operation, i.e., with the permeate side being closed and the reject side being open. For the FTCMR experiments, the reject-side of the membrane was closed and the gases were forced to pass through the membrane. 87 2.3. Experimental Results and Discussion From permeation experiments with a number of three-layer unimpregnated membranes the permeance ratio for He to Ar was on the average 2.7. Since the ratio for He and Ar permeance calculated from the formula S.F = √ ⁄ , where M Ar and M He are the molecular weights of Ar and He respectively is 3.1, one can infer that no significant defect or cracks in the membrane existed. The permeation data for one of these membranes (EM1) prior to being catalytically impregnated are shown in Fig. 2.3a and 2.3b below. 88 (a) (b) Figure 2.3: (a) Permeance results for He for an unimpregnated membrane. (b) Permeance results for Ar for an unimpregnated membrane R² = 0.9974 R² = 0.9706 R² = 0.9548 1.1E-07 1.2E-07 1.3E-07 1.4E-07 1.5E-07 1.6E-07 1.7E-07 1.8E-07 1.9E-07 0.0000002 0 5 10 15 20 25 Permeance (m 3 /(m 2 .Pa.s) Avg Pressure (psig) Helium Permeance Chart 150 degC 200 degC 250 degC R² = 0.9992 R² = 0.9996 R² = 0.8691 1.1E-08 2.1E-08 3.1E-08 4.1E-08 5.1E-08 6.1E-08 7.1E-08 8.1E-08 9.1E-08 1.01E-07 0 5 10 15 20 25 Permeance (m 3 /(m 2 .Pa.s) Avg Pressure (psig) Argon Permeance Chart 150 degC 200 degC 250 degC 89 The permeation results for an impregnated catalytic membrane (this is membrane CM2 in Table 2.1) and its transport characteristics were also investigated with single-gas permeation tests involving He and Ar, and the data are shown in Figs. 2.4a and 2.4b below. 90 (a) (b) (c) Figure 2.4: (a) Permeance results for He for catalytic membrane. (b) Permeance results for Ar for catalytic membrane. (c)Knudsen % flow for catalytic membrane R² = 0.9985 R² = 0.9994 R² = 0.9954 3.4E-08 3.5E-08 3.6E-08 3.7E-08 3.8E-08 3.9E-08 4E-08 4.1E-08 0 1 2 3 4 5 6 Permeance (m 3 /m 2 .pa.s) Avg Pressure (psig) Helium Permeance Chart Helium 150degC Helium 200degC Helium 250degC R² = 0.9787 R² = 0.9818 R² = 0.9829 1.2E-08 1.4E-08 1.6E-08 1.8E-08 2E-08 2.2E-08 0 2 4 6 8 10 12 Permeance (m 3 /m 2 .pa.s) Avg Pressure (psig) Argon Permeance Chart Argon 150degC Argon 200degC Argon 250degC 30 50 70 90 130 150 170 190 210 230 250 270 Knudsen % Temperature ( o C) Knudsen % vs. Temperature Argon Helium 91 The data in Fig. 2.3 and 2.4 are analyzed using the following Equation [Motamedhashemi, 2012]. ( ⁄ ) [ √ ] (2.1) (2.1a) ⁄ (2.1b) where r out (m) is the outer radius of the membrane, r in (m) is the inner radius of the membrane, R (J/mol.K) is the universal gas constant, T is the temperature in K, ε is the porosity, τ is the tortuosity, μ is the gas viscosity, P av is the average pressure (Pa), M is the molecular weight of the gas (g/mol), B 0 (m 2 ) is the viscous flow parameter and K 0 (m) is the Knudsen flow parameter. The estimated transport properties for the above data are shown in Table 2.3 below. There is good consistency for the values for B 0 and K 0 for the catalytically impregnated membrane (CM2) but not so for the unimpregnated membranes for which totally unrealistic values for the pore diameter of the membrane are calculated (not shown in the Table). This is a well-known problem [Motamedhashemi, 2012] when using the Equation above to fit the data of multi-layer membranes, with experimental errors in the measurements often resulting in totally unrealistic values of the parameters being calculated. 92 Membrane B 0 (m 2 ) K 0 (m) dp (m) ε/ τ He(catalytic) (423K) 2.30E-18 4.75E-10 3.88E-08 4.90E-02 Ar (catalytic)(423K) 1.81E-18 5.01E-10 2.89E-08 6.94E-02 He (catalytic) (473K) 1.93E-18 4.49E-10 3.43E-08 5.24E-02 Ar (catalytic)(473K) 1.64E-18 4.74E-10 2.77E-08 6.85E-02 He (catalytic)(523K) 1.76E-18 4.27E-10 3.29E-08 5.19E-02 Ar (catalytic)(523K) 1.76E-18 4.06E-10 3.48E-08 4.67E-02 He (unimpreganted)(423K) 5.12E-14 3.17E-09 Ar (unimpregnated)(423K) 3.01E-14 3.01E-09 He (unimpreganted)(473K) 2.75E-14 1.50E-09 Ar (unimpregnated)(473K) 2.95E-14 2.84E-09 He (unimpreganted)(523K) 2.93E-14 1.42E-09 Ar (unimpregnated) (523K) 2.82E-14 2.25E-09 Table 2.3: Table showing the parameters in the permeance equation for the EM1 (unimpregnated) and CM2 (catalytic) membrane. The permeance experiment was also done for a same unimpregnated and a catalytic membrane (CM6 membrane in Table 2.1) and its transport characteristics were investigated with single-gas permeation tests involving Ar, and the data calculated from the above formulas are shown Table 2.4 below. 93 Membrane B 0 (m 2 ) K 0 (m) d p (m) ε/ τ Ar (373K)(Unimpregnated) 1.07E-09 Ar (423K)(Unimpregnated) 1.51E-18 1E-09 1.2E-08 3.33E-01 Ar (473K)(Unimpregnated) 1.31E-18 9.48E-10 1.11E-08 3.43E-01 Ar (373K)(Catalytic) 8.21E-19 4.27E-10 1.54E-08 1.11E-01 Ar (423K) )(Catalytic) 6.03E-19 5.01E-10 9.63E-09 2.08E-01 Ar (473K) )(Catalytic) 9.83E-19 4.74E-10 1.66E-08 1.14E-01 Table 2.4: Table showing the parameters in the permeance equation for the CM6 membrane. Table 2.4 above shows a fair consistency for the calculated values of the transport parameters B 0 and K 0 (other than of B 0 for 373K for the unimpregnated membrane, the results for which is not shown). On the other hand, this is not the case when calculating the values of and though the values are not unrealistic and, in fact, the value for is close of the nominal value of 100 Å reported for the top layer by MPT. As noted above, similar problems have been encountered by other investigators before [for a review, see Motamedhashemi, 2012] when predicting the properties of multilayer membranes from similar transport data. As Figs. 2.3 and 2.4 show, impregnating the membrane decreases its permeance. In the FTCMR field a key metric of membrane performance is the so-called Knudsen fraction (Knudsen %), which is defined as the ratio of the membrane flux via Knudsen transport divided by the total flux. As Figs. 2.3 and 2.4 show, for these membranes the Knudsen % decreases with P av and also depends on the temperature of the experiment. The Knudsen % as a function of temperature, for P av =0.5 psig, for the CM2 membrane is shown in Fig. 94 2.4c above. Fig. 2.4c also shows that the Knudsen % value increases with increasing temperature which is an expected and positive result from these experiments. Fig. 2.5 below shows the SEM picture of membrane EM2. The membrane layers are clearly visible in this picture. As mentioned in the Experimental Section the SEM results provide us the thickness of the various membrane layers to be used in our future simulations, the results of which are discussed below. Figure 2.5: SEM picture of a three-layer membrane (EM1) showing the two top membrane layers (0.01 micron and 0.05 micron pore size) and the support 0.01 micron pore size membrane 0.05 micron pore size membrane 0.3 -0.4 micron pore size support 95 Fig. 2.6 below shows the back-scattering electron image (BSE) of the Pt catalyst particles as they are distributed on the top layer of the membrane CM7. Figure 2.6: BSE image for the CM7 membrane showing the Pt catalyst particels deposited in the membrane The white dots are the Pt particles, and from this figure it can be seen that they are uniformly distributed. An EDX line scan was also conducted in order to investigate the Pt metal distribution along the thickness of the CM3 membrane, and the results are shown in Fig. 2.7 below. 96 Figure 2.7: EDX analysis showing the distribution of Pt along the thickness of the CM3 membrane As this Figure shows, most of the platinum is deposited in the 100 o A region which is where most of the Knudsen diffusion mentioned above takes place, which is an important observation. Even though, the BSE image above gives a good indication of the Pt distribution, it is not a good indicator of showing Pt sizes. For this TEM studies were done by this group. Figure 2.8 below is a TEM image of the CM8 membrane (these studies were done by Dr. Motamedhashemi). The black spots indicate the Platinum particles. From this figure it was found that the majority of the particles are smaller than 10 nm. 97 Figure 2.8: TEM image of the Pt/Al 2 O 3 membrane 98 CO chemisorption (using a Chemisorb 2720 apparatus) was also utilized with one of the catalytic membranes (CM4 in Table 2.1) and the results are shown in Table 2.5. (The same Table is also shown in Motamedhashemi [2012]). Average Weight of Support+Membrane, (g) 3.62418 Average Weight of Catalyst+Support+Membrane (after regeneration), (g) 3.5446 Weight of Pt for Regenerated Catalyst, (dimensionless) 0.69E-02 Weight of Pt per Gram of Regenerated Catalyst, (dimensionless) 0.19E-02 Average Weight of Sample Used for Chemisorption, (g) 0.1953 Weight of Platinum Deposited on the Sample, (g) 0.38E-03 Number of Pt Moles Deposited on the Sample 1.94E-06 TCD Calibration Signal per 100 microliter of CO 0.21 Observed TCD Signal for the Sample 0.17 Pressure of Experiment, (Pa) 103191 Temperature of Experiment, (K) 298 Number of CO Moles Adsorbed on the Sample, (moles) 7.93E-07 Fraction of Pt Moles available for Reaction, (dimensionless) 0.41 Table 2.5: CO Chemisorption results for the CM4 membrane With the assumption that one molecule of CO attaches to one Pt atom, one can calculate the moles of active Pt that are available on the catalytic membrane. From the chemisortion results it was found that 40.8% of the Pt was active (i.e., exposed and available towards CO 99 chemisorption). Pt has a FCC (face centered cubic) crystalline structure. For this FCC structure a total of 12n 2 +2 atoms are on the particle surface whereas 4n 3 -6n 2 +3n-1 atoms are in the interior. Assuming that the chemisorption occurs only at the surface of Pt, a straightforward calculation of the FCC structure from the above formula indicates that ~40% of active sites should have 9 unit cells of the FCC. Pt crystals have a density of 21.5 g/cm 3 and each edge of the cell has a length of 4 times the radii. From this analysis the particle size comes to be ~ 3.6nm which is somewhat lower than average value obtained from the TEM analysis [Motamedhashemi, 2012]. Before using the membrane for treating the SLFG the reactivity of the module was first studied, and it was found to be non-reactive. Light-off temperature (which is defined as the temperature where 100% oxidation of a component takes place) experiments were carried out initially using the CM1 membrane as a part of a preliminary effort to find out the reactivity of the FTCMR in treating the SLFG. These results are given in Figure 2.9 below. 100 Figure 2.9: FTCMR light-off temperature results for membrane CM1 0 20 40 60 80 100 120 150 170 190 210 230 250 270 Conversion % Temperature ( o C) COS 0 20 40 60 80 100 120 150 170 190 210 230 250 270 Conversion % Temperature ( o C) VC 0 20 40 60 80 100 120 150 200 250 300 350 400 450 Conversion % Temperature( o C) TCFM 0 20 40 60 80 100 120 150 170 190 210 230 250 270 290 Conversion % Temperature ( o C) DMS 0 20 40 60 80 100 120 150 170 190 210 230 250 270 290 310 330 Conversion % Temperature ( o C) DCB 101 The results show that the light-off temperatures for COS, VC, TCFM DMS and DCB are 220 o C, 240 o C, 300 o C, 240 o C and 290 o C respectively. These good light-off temperatures indicated that the CM1 membrane is highly active. The reactor was then turned into the monolith mode and the same light-off temperature experiments were carried out. A comparison of the monolith and FTCMR results are given in Fig.2.10 below for the various components. 102 Figure 2.10: FTCMR vs. monolith results for CM1 membrane 0 20 40 60 80 100 120 50 100 150 200 250 300 350 400 Conversion % Temperature ( o C) Carbonyl Sulfide Knudsen Monolith 0 20 40 60 80 100 120 50 100 150 200 250 300 350 Conversion % Temperature ( o C) Vinyl Chloride Knudsen Monolith 0 20 40 60 80 100 120 50 100 150 200 250 300 350 400 450 Conversion % Temperature ( o C) Trichlorofluoromethane Knudsen Monolith 0 20 40 60 80 100 120 0 50 100 150 200 250 300 350 Conversion % Temperature ( o C) Dimethyl Sulfide Knudsen Monolith DichloroBenzene 0 20 40 60 80 100 120 0 100 200 300 400 500 Temperature (C) Conversion % Monolith Mode Knudsen Mode ! 103 These results indicate that although the components eventually reach their light-off temperatures in the monolith mode, FTCMR gets there first at lower temperatures except for DCB. After these preliminary experiments, a more detailed investigation was undertaken using the CM2 catalytic membrane. The results for the reaction for the 5 components of SLFG at different flow rates and temperatures for the CM2 membranes are shown in Fig. 2.11 below. 104 Figure 2.11: FTCMR results for various LFG components for different flow-rates for membrane CM2 105 As shown clearly in these figures the FTCMR is capable of removing up to 100% of all the NMOC components at 350 o C except for the TCFM. This could be because TCFM is hard to oxidize due to the chlorine and fluorine atoms present in it. Another observation that can be seen from the FTCMR results of CM1 and CM2 membrane is that most of the components (except TCFM) reach 100% conversion at lower temperatures of around 300 o C for CM1 membrane, whereas for CM2 membrane the temperatures are higher at 350 o C. This is due to the low Pt loading of CM2 membrane (0.04%) vs. that of the CM1 membrane (0.72%). The membrane reactor was then run in an open-flow mode from the flow-through mode to simulate a monolith reactor. The results for this experiment for all the 5 components are shown in Fig. 2.12 below. 106 Figure 2.12: FTCMR vs. Monolith comparison for SLFG components for membrane CM2 107 Although the conversion for the NMOC components is higher for monolith at lower temperatures it does not reach complete conversion as the FTCMR does even at 350 o C. As can be clearly seen from these results the FTCMR is thus far superior to monolith in achieving 100% conversion for all compounds at the lowest temperature. Another key advantage of the FTCMR is that no deactivation was observed for this reactor. To verify this, a series of experiments was carried out (using the CM2 membrane after the series of the experiments described above were completed) for which the SLFG flow rate was kept constant at 0.5 cc/s. For this series of experiments, starting at room temperature, the FTCMR temperature was first raised in flowing N 2 to 350 o C at a rate of 1 o C/min When the FTCMR reached this temperature the SLFG gas was switched on and was left at this temperature for 6 hr and the conversions of the various components were measured. Upon completion of the experiment at this temperature, the SLFG was switched off and the FTCMR temperature was then lowered in flowing N 2 to 300 o C at - 0.8 o C/min. Once the FTCMR reached this temperature the SLFG gas was switched on again and the membrane was left there again for 6 hr and the conversion measured at the end of the time period; the procedure was then repeated two additional times for two other FTCMR temperatures (250 o C and 200 o C). Upon completion of the experiment at 200 o C the temperature was raised in flowing N 2 to 250 o C, the SLFG was turned and left there for 6 hr to measure the conversion; then the SLFG was turned off and the temperature was again raised in flowing N 2 to 300 o C, the SLFG turned on and the conversion was noted after 6 hr, and then again the temperature was raised (in flowing 108 nitrogen) to 350 o C, the SLFG was turned on and the conversion measured after for 6 hr. After that the procedure was repeated for three additional temperatures (300 o C, 250 o C, and 200 o C). Fig. 2.13 below shows the results of these three FTCMR experimental runs. 109 Figure 2.13: FTCMR deactivation test results for different runs for CM2 110 As can be seen from this figure no visible deactivation is observed for the FTCMR, which is not the case with the monolith reactor, see discussion below. The reason for the added resistance of the FTCMR to deactivation is not entirely well understood at this point (and further experiments are planned in the future to validate this significant advantage over the more conventional monolith reactor and to try to identify the causes for the observed behavior). One explanation may be that in the FTCMR mode of operation there is less of an accumulation of the potential by-products of the reaction (e.g., HCl, HF, SO 2 , etc.) that may cause deactivation (but again, we have no experimental evidence currently to support such a hypothesis). As noted above, in order to study the behavior of the CM2 membrane when operating as a catalyst monolith, the reactor was then switched into the catalytic monolith mode and two additional series of experiments were carried out. In these experiments the above experimental procedure for the FTCMR was followed for two runs for all four temperatures (for and SLFG flow rate of 0.5 cc/s). The results from these experiments are shown in Fig. 2.14 below. These results indicate the CM2 membrane, when operating in the catalytic monolith mode, suffers a decline in its activity (particularly for the oxidation of COS and TCFM). For the VC, DMS and DCB components deactivation can be observed at higher temperatures ~350 o C, where already the FTCMR had exhibited a superior performance. 111 Figure 2.14: Monolith deactivation test for membrane CM2 112 2.6. Conclusions Landfill gas is potentially a valuable renewable fuel because of the methane that it contains. But the presence of impurities in the landfill gas presents challenges for its effective utilization. Burning the gas for power generation or even when we flare the landfill gas, if these impurities are not removed, it releases toxic gases such as hydrogen chloride, sulfur dioxide, etc. which are harmful to both humans and the environment. There is a strong incentive, therefore, to develop effective technologies for removing the various NMOC compounds from landfill gas prior to its utilization as a fuel. In this study, a novel catalytic oxidation technology appropriate for landfill gas clean-up based on the concept of a “Pore Flow Reactor” impregnated with an oxidizing nanocatalyst has been studied. For our experiments, a simulated landfill gas containing 60% CH 4 30% CO 2 , 9 % N 2 , 1% O 2 , 50 ppm DMS, 50 ppm COS, 50 ppm VC, 50 ppm TCFM and 5 ppm DCB has been utilized. γ-Alumina membranes have been used as the FTCMR along with Pt as a nanocatalyst. The catalyst has been characterized using SEM, EDAX and CO chemisorption methods. The 1% oxygen present in the gas has been found sufficient to oxidize all the NMOC impurities. The effect of operating temperatures and feed flow on reactor conversion and catalyst deactivation are investigated. It has been found that complete destruction (>99% conversion) of all the impurities can be obtained at relatively low temperatures. 113 Chapter 3 The Impact of Siloxane Impurities on Engines Operating on Renewable Natural Gas The research presented in this chapter has already been published [Nair et al., 2012] 114 3. 1 Introduction As mentioned in the introduction chapter biogas from sludge biodegradation in waste- water treatment plants (WWTP), and landfill gas (LFG) generated from the decomposition of solid waste in landfills also contain a particularly problematic class of trace constituents known as siloxanes [Ohannessian et al.,2008]. During combustion of LFG, siloxanes decompose into SiO 2 microparticulates [Schweigkofler et al., 2001; Dewil et al., 2006] which not only pose a threat to equipment (see discussion to follow) but also to human health and the environment, unless adequate measures are taken to remove them from the exhaust. Both are current keen concerns in California, and key drivers for this study, because of the prospect of using bio-methane or renewable natural gas (RNG), which results from either biogas or LFG, after their impurities have been removed and their methane content has been upgraded to meet natural gas (NG) pipeline standards. The concern here is with potential malfunction of equipment for siloxane removal – see discussion below – which may lead into their accidental release into the NG supply (pipeline) system; this then may result in their combustion in home appliances (e.g., stoves, water heaters, etc.), thus causing harmful environmental emissions and deleterious effects on human health [ El-Fadel et al., 1997; Abatzoglou et al. 2009]. Because of the technical challenges siloxanes present to the beneficial use of biogas and LFG, they have attracted the attention of researchers in the renewables area, particularly in recent years. Badjagbo et al. [2010] and Crest et al. [ 2010], for example, reported on 115 the challenge of sampling and analyzing siloxanes in LFG, this still remaining a major challenge since field instruments are still lacking in sensitivity and reliability. Furthermore, there are numerous reports on different techniques for their removal. 11,16 Adsorption is the most common method [Oshita et al., 2010; Matsui et al., 2010; Montanari et al.,2010], but a key problem is that the media utilized, are not particularly selective towards siloxanes, and they adsorb most other non-methane organic compounds (NMOC) in LFG; this reduces the bed’s capacity for siloxane adsorption, necessitating frequent regeneration. Siloxanes are difficult to remove effectively from spent adsorbents during regeneration [Wheless et al., 2002], which results in adsorbents of progressively lower capacity, until media 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, which when released from the beds are still the same as when they entered. Regeneration involves burning the off-gas, which releases SiO 2 particles into the atmosphere, and consumes methane fuel to operate the incinerators. Absorption at relatively high pressures in solvents (e.g., Selexol TM and methanol) is another approach [Schweigkofler et al., 2001]. The problem with absorption is high capital and operating and maintenance costs. Solvent regeneration (and robustness) is also key to success to reduce solvent disposal and operating costs, and to ensure a long- term operation. Refrigeration has been tried also, but is not effective on its own [Wheless et al., 2002]. Hybrid processes combining refrigeration with adsorption/absorption show 116 more promise [Schweigofler et al., 2001]. However, the high energy consumed for cooling large amounts of wet gas is a major drawback for commercialization. Further, as with adsorption and absorption, siloxanes are not converted, and thus require incineration for their destruction. Biological treatment to remove siloxanes has been studied also [Ohannessian et al. 2008; Popat et al., 2008; Accettola et al. 2008], but the results, so far, are disappointing with conversions ~10%. Siloxane removal via membranes has also been proposed [Ajhar et al., 2006], however, no experimental data are currently available, and the practical implementation is unclear, as membranes are, in general, not well-suited for removing trace impurities. Reactive approaches have also been utilized, for example, peroxidation to reduce the siloxane content of sludge in a digester producing biogas [Apples et al., 2008]. A reduction of 50–85% was observed, which is not high compared with other competing technologies, and it will require, in most instances, an additional polishing step to comply with siloxane limits by engine manufacturers. Table 3.1 shows the typical such limits proposed by various engine manufacturers. 117 Engine Manufacturer Siloxane mg/m 3 in LFG Waukesha 25 Jenbacher 10 Caterpillar 10 Deutz 5 Solar Turbines 0.1 IR Microturbines 0.06 Capstone Microturbines 0.03 Table 3.1: Siloxanes limits in LFG and digester gas recommended by various engine manufacturers. Finocchio et al., 2008, have studied the decomposition of D 3 on the surface of basic (CaO, MgO) and acidic oxides (Al 2 O 3 , SiO 2 ), showing that reactive adsorption occurs accompanied by surface silication and release of methane. Adsorbent regeneration is not possible, however, and the basic oxides lose their reactivity when in contact with CO 2 due to surface carbonation. UV photodecomposition of L 2 has been also investigated by this team in bench-top experiments with promising results [Prosser et al., 2010], but the approach needs still to be field-tested, and its economics to be further investigated. In summary, conventional methods (adsorption, absorption, refrigeration) face significant technical and economic hurdles for siloxane removal from LFG and biogas but remain in use because there are currently no other commercial processes to replace them. Newer approaches (e.g., biofiltration, membrane separation, reactive approaches, etc.) face challenges of their own and/or have yet to be field-tested. As one starts using RNG, 118 therefore, the potential for these impurities finding their way into NG equipment and common household appliances remains a key challenge, and points out the need for systematic studies of the fate of siloxanes during combustion in such devices and their impact on performance and associated emissions. As noted above, there are numerous reports of field-scale observations of silica films (often several mm thick) forming on internal surfaces of equipment operating on biogas [Schweigkofler et al., 2001; Dewil et al., 2006], and proving difficult to remove by either chemical or mechanical treatment. These abrasive silica deposits forming on the inner walls of engines, on gas turbine blades, and on heat exchange surfaces in boilers, have the potential to lead to serious damage, thus necessitating more frequent maintenance and increasing the cost of operating these devices [Ajhar et al.,2010; Popat et al., 2008]. In boilers, the silica layer acts as a thermal insulator, interfering with the heat exchange operations [McBean et al., 2008]. In engines, the silica deposits have been reported to clog narrow passages [Oshita et al., 2010], increasing thus the potential for accidental explosions. Martin et al., 1996, have reported that the combustion of siloxane-containing biogas in lean-mix, spark-ignition engines results in substantial wear even after short running-times. Siloxanes have been reported to interfere also with the operation of catalytic treatment systems of the exhaust gas, often decreasing their efficiency [Ohannessian et al., 2008; Badjagbo et al., 2010]. The impact of organosilicon compounds on catalytic oxidation 119 catalysts had previously also attracted attention in the context of the abatement of VOC emissions from printing shops. In these studies, organosilicon compounds contained in the printing ink are reported to decompose inside the catalysts and deposit as silica and to mask the active sites [Libanati et al., 1998]. Organosilicon compounds such as L2 are also known to poison Pt and Pd supported catalysts by virtue of the coating and blocking of the surface of the precious metals by silicon atoms [Cullis et al., 1984; Gentry et al. 1978]. More recent studies of deactivation of Pt/Al 2 O 3 supported catalysts have revealed that the silica deposition on the catalyst surface is promoted by the active catalyst metal [Larsson et al., 2007]. As alluded previously, another potential concern with the fine silica microparticles that form is that they may be carried out in the flue-gas of the NG equipment (e.g., engines, boilers and furnaces), and unless adequate measures are taken to remove them from the exhaust, they are likely to escape into the atmosphere, where they may pose a risk to both human health and the environment [El-Fadel et al., 1997; Abatzoglou et al., 2009]. The same concern exists, of course, as well with home appliances such as stoves and ovens operating on RNG. In summary, attention has been focused in recent years on the fate of siloxane impurities during RNG combustion in NG equipment and common household appliances and the impact on their performance as well as on human health and the environment. Though there are reports of field observations of inorganic deposits on various internal surfaces of 120 combustion equipment, there are no open literature reports of systematic studies, under well-controlled laboratory conditions, investigating the various phenomena involved. The results of fundamental combustion studies of siloxane decomposition in a simulated RNG/air flat flame using the counterflow experimental technique was studied by Jalali et al., 2013. A key conclusion from that study is that siloxanes readily decompose in the RNG combustion environment to form pure silica particulates that coat exposed metal surfaces. The model combustion experimental configuration allowed, furthermore, the determination of the global reaction kinetics of siloxane decomposition in the RNG flame environment. In this study, the focus is on a well-instrumented IC engine operating on real NG laced with trace amounts of two common siloxanes (L2 and D4). Laboratory engine studies, due to their mechanical complexity, cannot provide the fundamental insight counter flow combustion studies provide. They are, however, an important link between fundamental combustion investigations and commercial engines operating under real field conditions. In what follows, the experimental set-up and the techniques utilized are first described. The experimental observations are then presented and discussed in terms of the potential implications of the siloxanes’ presence in NG on the operation of real world engines. 121 3.2 Experimental Procedures In these studies two Honda EU2000i gasoline electric generators were utilized. The generators were modified in order to be able to run on NG instead of gasoline. This was accomplished by converting the original gasoline carburetor into a modified NG/gasoline carburetor and then connecting it to a NG regulator. A photograph of one of the engines and of the catalyst monolith attached is shown in Fig. 3.1 below. 122 Figure 3.1: Top, laboratory engine with the catalyst monolith attached; bottom: catalyst monolith connected to the engine muffler. Note that the catalyst bed attached to the engine (Fig. 3.1, above) was purchased separately (DCL Inc., oxidation catalyst Model #RC4x4x1-24) since the engine did not come equipped with it, because its emissions when running on gasoline were below the regulated threshold. The catalyst bed was of monolith shape and the manufacturer reports that it contains V-Zn-Ba on an alumina wash-coat. Though the original engine did not come equipped with an oxygen sensor, in order to investigate the impact of siloxane impurities on this important component of larger size engines, an oxygen sensor (Bosch, model#0258010078) was installed in both engines. The sensor was installed on the exhaust gas stream before the catalyst bed, and during operation its voltage was monitored via a multimeter. 123 Figure 3.2: Top, schematic of the experimental apparatus for the siloxane engine; bottom, schematic of the experimental apparatus for the non-siloxane engine. 124 Figure 3.2 above shows the schematic of the overall experimental engine set-up. Each engine was connected, during operation, to two GE electric bulbs, 250W and 150W respectively, corresponding to a total electric load of 400W (chosen because, according to the manufacturer, the engine runs most efficiently at that load). Performance was monitored through the r.p.m display unit of the engine. The flow of pipeline-quality NG to the engine was controlled by a small valve in the NG regulator, and was monitored via gas-flow meters (American Meter Company, model DTM 200A). The energy efficiency (calculated as (%) , where W is the engine energy output in Watts, F the gas flow rate in ft 3 /s, E the energy content of NG equal to 1020 BTU/ft 3 , P a conversion factor equal to 1055.05 J/BTU) was also recorded as an indicator of engine performance. The two engines operated side-by-side for the same period of time and under the same NG flow rate and overall conditions (e.g., air to fuel ratio or A/F), other than the fact that the NG going into one of the engines (hereinafter, named the siloxane engine) was “spiked” with an equimolar mixture of two siloxanes, namely L2 and D4, which are the two most common linear (L2) and cyclic (D4) siloxanes in LFG and biogas (see Table 3.2 below). 125 Name CAS Numb er Formula MW [g/mo l] Boilin g Point [ o C] Vapor Pressu re [mmH g] Densit y [g/cc] Hexamethylcyclotrisiloxane (D 3 ) 541- 05-9 C 6 H 18 O 3 S i 3 222.4 6 134 10 1.02 Octamethylcyclotetrasiloxane (D 4 ) 556- 67-2 C 8 H 24 O 4 S i 4 296.6 2 175 1.3 0.96 Decamethylcyclopentasiloxane (D 5 ) 541- 02-6 C 10 H 30 O 5 Si 5 370.7 7 210 0.4 0.96 Dodecamethylcyclohexasiloxa ne(D 6 ) 540- 97-6 C 12 H 36 O 6 Si 6 444.9 2 245 0.02 0.97 Hexamethyldisiloxane (L 2 ) 107- 46-0 C 6 H 18 OSi 2 162.3 8 101 31 0.76 Octamethyltrisiloxane (L 3 ) 107- 51-7 C 8 H 24 O 2 S i 3 236.5 3 153 3.9 0.82 Decamethyltetrasiloxane (L 4 ) 141- 62-8 C 10 H 30 O 3 Si 4 310.6 9 194 0.55 0.85 Dodecamethylpentasiloxane (L 5 ) 141- 63-9 C 12 H 36 O 4 Si 5 384.8 4 230 0.07 0.88 Trimethylsilanol 1066- 40-6 C 3 H 10 OSi 90.20 99 73.9 0.81 Table 3.2: Siloxane compounds in LFG and some of its properties from Jalali et al. [2013] To generate ppm-level concentrations of siloxanes, a high-precision syringe pump (Harvard Apparatus model PHD 2000) coupled to a quartz Nebulizer with a flush capillary-lapped nozzle (Meinhard model TR-20-A0.5) was utilized to generate a fine siloxane spray into a heated NG stream whose flow rate was controlled by a mass flow controller (Cole Parmer model# 55932). The siloxane-containing NG stream was then mixed with the main NG stream in order to generate the NG feed-stream into the engine with the desired siloxane concentration. The ability of the set-up to generate reliably the required siloxane concentrations was verified by periodic injection of gas samples into a 126 GC/MSD system (Agilent Technologies, 7890A GC System/5975C Inert XL MSD). The GC/MSD system was calibrated using standard liquid samples of siloxanes in ethanol. The GC/MSD measurements were also cross-checked with the estimated concentrations based on the amount of liquid siloxane injected and the gas flow rates. The composition of the exhaust gas coming out of the two engines was analyzed in two locations immediately after exiting the engines (and before entering the catalyst bed) as well as after exiting the catalyst bed. A Testo 350XL Gas Emissions Analyzer was used to measure the concentrations of CO, CO 2 , CH 4 , NO, NO 2 and NO x . The Analyzer was calibrated every seven days (according to EPA protocol) using standard gas cylinders (450.5 ppm CO in N 2 ; 215 ppm NO and 100 ppm NO 2 in N 2 ; and 1394 ppm CH 4 in N 2 ). Omega J- type thermocouples were installed at the inlet and outlet of the catalyst bed to measure the bed temperature at these two positions. The catalyst bed was heated by the hot exhaust gas stream and no effort was made to control its temperature (which, however, remained relatively constant during operation). The pressure drop across the bed was also measured by a differential pressure sensor (Dwyer Series 475 digital manometer). Both engines were operated for 500 hr. For the first 10 hrs. the concentration of siloxanes into the siloxane engine was varied. After that initial period, the total concentration of siloxanes (L2+D4) stayed constant at 10 ppm v . This siloxane content (10 ppm v ) is very 127 much in line with what is encountered in the LFG from various landfill sites around the world. For example, Wheless and Gary, 2002, reported the amount of siloxane found in two landfill sites (C-Modern and C-Kiefer) to be 20 ppm V . According to Dewil et al., 2006, the siloxane concentration ranges from 4.8 mg/m 3 up to 400 mg/m 3 , averaging 71.4 mg/m 3 . The conversion factor between these two common ways to express siloxane concentration depends on the M.W. of the given siloxane. For example, for L2 1 mg/m 3 corresponds to 0.138 ppm v , so the aforementioned range by Dewil et al., 2006], corresponds to (0.7 – 55.2 ppm v ), averaging 9.9 ppm v , close to the concentration of 10 ppm v used in this study. The catalyst bed connected to the siloxane engine was removed after 403 hr of operation (by that point it had completely deactivated, see discussion below). For control purposes, the catalyst in the non-siloxane engine was also replaced even though it was still active. The silica profiles along the deactivated bed were analyzed using SEM and EDX. The bed (for both engines) was subsequently replaced with another bed of the same composition, which was left connected to the engine for an additional 97 hr. Upon termination of the study, the structural characteristics of the second deactivated bed were again analyzed via SEM/EDX, BET (for studying the overall surface area and pore volume), and CO Chemisorption analysis to detect any potential decrease in the active sites of the catalyst. 128 After 200 hr of operation the siloxane engine spark-plugs were removed and their surface analyzed for the presence of silica deposits via SEM. After 467.5 hr of operation, an oxygen sensor was installed and its performance was monitored via the voltage signal coming from the sensor. At the end of the 500 hr point of engine operation the sensors from both engines were removed and their performance was tested (at the Advanced Transportation Technology Center, of Long Beach City College) using a Bosch MTS 5200 Engine Analyzer equipped with a Digital Storage Oscilloscope. Subsequently, the sensors’ surface was analyzed using SEM/EDX. The oil for both engines was replaced every 100 hr and analyzed for the presence of metals and other heteroatoms, via a variety of analytical techniques, by Jet Care International, Inc. In order to verify whether silica microparticulates manage to escape the catalyst bed, their presence was analyzed at the end of the exhaust pipe venting into the hood, via the use of a scanning mobility particle sizer (SMPS). Upon completion of the testing, the siloxane engine was completely dismantled, and the internal surfaces and components were examined for the presence of silica deposits. 129 3.3 Results and Discussion The flue-gas composition data for the non-siloxane engine immediately after the engine (and before the catalyst bed) and after the catalyst bed are shown in Fig. 3.3 below (note that these data are with the 2 nd catalyst bed). 130 (a) (b) Figure 3.3: (a) CO and CH 4 flue-gas concentrations for the non-siloxane engine. NO, (b)NO 2 and NO x flue-gas concentrations for the non-siloxane engine. 131 Note that the NO x species (defined as the sum of NO+NO 2 – though some of the NO 2 converts into NO) and CH 4 remain unaffected by this catalyst. However, there is substantial conversion of CO (see also Fig. 3.4, below). Figure 3.4 below shows the corresponding flue-gas composition data for the siloxane engine and the behavior is qualitatively similar to that for the non-siloxane engine. In Fig. 3.4 the data are plotted both in terms of the time of operation but also in terms of the total siloxane amount the engine has been exposed to during the operation of this particular catalyst bed. 132 (a) (b) Figure 3.4: (a) CO and CH 4 flue-gas concentrations for the siloxane engine. (b) NO, NO 2 and NO x flue-gas concentrations for the siloxane engine. 133 Figure 3.5 below shows the CO conversion for the non-siloxane engine, which indicates that the catalytic activity remains constant for the duration of engine operation, with conversion harboring ~90%. Figure 3.5: CO conversion vs. time on stream for the non-siloxane engine 134 Figure 3.6: CO conversion vs. time on stream and volume of siloxane injected for the siloxane engine. Figure 3.6 shows the corresponding CO conversion for the siloxane engine, again plotted both in terms of time on stream as well as with respect to the total volume of siloxanes injected into the engine. Catalytic conversion starts at ~ 90% (as is the case with the non- siloxane engine, see Fig. 3.5) but it declines gradually right from the start reaching values lower than 50% by the time the engine operation was terminated (for the first catalyst bed which stayed much longer on stream, as noted above, the catalyst activity continued to decline throughout the operation of the engine). 135 Figure 3.7: Silica profile on the catalyst monolith surface along its length. Figure 3.7 above shows an EDX scan of the catalyst’s surface along the length of the bed which is 8.7 cm long (this EDX profile was taken with the first catalyst monolith after 403 hr on stream). A silica film covers the whole monolith surface with the heavier coverage observed on the front end of the bed, as expected. 136 Figure 3.8: BET data of the catalyst monolith for both the siloxane and non-siloxane engines. The deposition of silica also impacts greatly the pore structure characteristics, as Fig. 3.8 above indicates which shows the pore size distribution (PSD) measured using BET for samples taken from both the siloxane and non-siloxane engine (the data in Fig. 3.8 were generated by taking a 1 g sample from the midsection of the second catalyst bed without crushing it and analyzing it in a Micromeritics ASAP 2010 BET apparatus). When comparing the BET data between the catalyst monolith connected to the siloxane engine with that of the non-siloxane engine, one observes significant decreases in both the pore 137 volume (from 0.0272 cm 3 /g to 0.0242 cm 3 /g) as well as in the total surface area (from 6.563 m 2 /g to 4.911 m 2 /g). In addition, the average pore size (as determined using the BJH analysis method of the adsorption branch of the data) of the siloxane engine monolith is shifted towards the smallest pore sizes, indicating a progressive plugging of the support pores in the pore structure. CO Chemisorption analysis, using an Autosorb iQ (Model # 373) instrument, was also performed on the same (as with the BET analysis) spent catalyst beds of both engines. The analysis was carried out using a 1.24 g sample (from the inlet section) from each bed at an isothermal temperature of 45 o C, and by varying the CO pressure from 80 torr to 720 torr, and measuring the CO uptake at various pressures, and extrapolating the resulting linear plot to 0 torr pressure to calculate the amount of CO chemisorbed by the catalysts. For the catalyst bed from the engine without siloxanes the amount was 0.761 μmol/g, while the corresponding amount for the catalyst from the siloxane engine was 0.043 μmol/g, indicating a decrease in the active surface area of ~94.4% for the silica-coated catalyst. Silica microparticulates also coat the internal surfaces in the engine, as one may expect. (One expectation, prior to the initiation of the study, was that the silica, by coating the spark-plug surfaces, would significantly interfere with the engine performance; the expectation of engine performance deterioration did not materialize, however, see further discussion below). After 97 hr of operation, for example, the spark-plugs from both engines were removed and visually inspected. The presence of silica deposits on the spark-plug from the siloxane engine was clearly evident. After 200 hr of operation, the 138 spark-plugs from both engines were removed again, and their surfaces examined by SEM/EDX. Figure 3.9: SEM image of the spark-plug metal surface for the siloxane engine coated by a silica film. Figure 3.9 above shows a SEM image of the silica film formed on the spark-plug metal surface from the siloxane engine. The silica deposit structure consists of a compact layer that appears strongly adhering to the metal surface underneath and a top layer with a more porous and less dense nature that easily flakes from the surface. Figure 3.10 below shows the EDX data of the metal surface of the non-siloxane engine spark-plug as well as that of the siloxane engine spark-plug. 139 Figure 3.10: EDX of the metal surface of the non-siloxane engine spark-plug (top) and of the silica coated spark-plug surface for the siloxane engine (bottom). 140 As expected, no Si was detected on the non-siloxane engine spark-plug surface while only Si and O were observed on the siloxane-engine spark-plug. However, the spark-plug metal surfaces (not shown here) where the engine spark is generated (gap) appear equally clear and free of any deposits in both spark-plugs, consistent with the fact that there were no changes in engine operation observed, as noted above. These spark-plugs were then re-installed in their respective engines, and left on the engines for the duration of the experiment (500 hr). In the siloxane engine, the spark-plug continued to function well without any problems till the experiment was terminated. After 500 hr, the spark-plugs for both engines continued to be free of silica deposits on the gap. As noted previously, the Honda engines did not come equipped with an oxygen sensor so, instead, an oxygen sensor was purchased and installed in both engines (these particular sensors were installed at the 460 hr point of engine operation). 141 Figure 3.11: Sensor voltage signals. Oxygen sensors are a critical component of large NG internal combustion engines as they determine whether the engine operates either fuel-rich or fuel-lean and help control the operating conditions and performance, including the NO X and the other emissions. To monitor the performance of the sensor, its voltage was recorded. Figure 3.11 reports the readings for both the siloxane and the non-siloxane engines. The voltage signals increased with time on stream for both engines, and more so for the siloxane engine. On the other hand, the signals were very noisy and they did not appear to indicate any discernible qualitative differences in their performance. (Since these sensors do not control the Honda engine performance, any damage to them will not manifest itself as a problem for engine operation; that will not be the case, however, for engines whose operation depends on the sensor to control performance and the emissions). Therefore, in 142 order to examine the impact, if any, that siloxanes may have on oxygen sensor performance both devices were removed at the 500 hr point of engine operation for testing using a Bosch MTS 5200 Engine Analyzer, as previously noted. The testing involved monitoring the voltage pulse width and the response time of the sensors, as the engine is first forced to run fuel-rich, and is then allowed to adjust to the proper air-to- fuel mixture, i.e., running with maximum fuel efficiency depending on the load. If an oxygen sensor functions properly, the response time is expected to be less than 100 ms. The testing results indicated that the response time of the non-siloxane engine sensor was 96 ms, indicating good performance, whereas that of the siloxane-engine oxygen sensor was 124 ms, indicating failed performance. (If this particular sensor was controlling the engine operation, it would trigger the check engine light to come on). After these sensor tests were completed, the sensor surfaces were studied by SEM and EDX. Figure 3.12 shows the SEM micrograph of the siloxane sensor surface which illustrates the presence of deposits, while EDX indicates the presence of Si and oxygen indicative that these are silica films. The non-siloxane engine sensor was also analyzed by SEM/EDX as well and no silica films or deposits were found on it. 143 Figure 3.12: SEM-EDX results of the siloxane engine oxygen sensor. 144 Figure 3.13: Photograph of the piston head of the siloxane engine showing the silica film deposition. After the completion of the 500 hr run the siloxane engine was dismantled and its internal surfaces were examined. Extensive silica deposition was observed. Figure 3.13 above, for example, shows a photograph of the piston head where a silica film is clearly visible. Visual examination of the engine valve surfaces indicated substantial coatings on the valves from the siloxane engine as compared to the valves from the non-siloxane engine. The EDX results of one of these valve surfaces, shown in Fig. 3.14 below, confirm that the deposits are silica. 145 Figure 3.14: EDX results of the siloxane engine valve. The presence of silica deposits on the internal engine surfaces suggests the presence of these particulates in the engine oil as well. The engine manufacturer recommends an oil change after every 100 hrs. of operation. After two of these oil changes, the used oil from both engines was analyzed for the presence of metals. The results are shown in Table 3.3 below. 146 Elements 1st oil sample 2nd oil sample Non- Siloxane Siloxane Non- Siloxane Siloxane Sulfur (ppm) 2100 2200 2000 2000 Chlorine (ppm) <50 <50 50 <50 Sodium (ppm) <1 <1 <1 <1 Boron (ppm) 1 1 <1 <1 Silicon (ppm) 13 205 11 375 Iron (ppm) 27 24 34 25 Aluminium (ppm) 8 7 7 5 Chromium (ppm) 1 <1 <1 <1 Molybdenum (ppm) 163 149 112 110 Copper (ppm) 6 5 4 6 Lead (ppm) <1 <1 <1 <1 Tin (ppm) <1 <1 <1 <1 Nickel (ppm) <1 <1 1 1 Titanium (ppm) <1 <1 <1 <1 Silver (ppm) <1 3 1 <1 Vanadium (ppm) <1 <1 <1 <1 Manganese (ppm) <1 <1 <1 <1 Phosphorous (%wt) 0.0714 0.0721 0.0725 0.0738 Zinc (%wt) 0.0932 0.0919 0.0852 0.0909 Calcium (%wt) 0.2099 0.2504 0.2044 0.2073 Magnesium (%wt) 0.0009 0.0009 0.0008 0.0008 Table 3.3: Oil analysis data for two used oil samples from the siloxane and non- siloxane engines. There was a significant concentration of silicon element in the used oils from the siloxane engine when compared to the clean oil and the used oil from the non-siloxane engine oil. The presence of a large concentration of silica particulates in the engine oil indicates the potential for accelerated engine wear of moving parts, such as pistons, likely to lead to the need for frequent replacement. Despite the coating of the internal engine surfaces 147 with silica deposits including the spark-plugs and the piston head, there was little discernible evidence of any negative impact on engine performance, however. For example, the r.p.m data indicated no changes in the power output of the engine. As noted previously, one of the concerns with the siloxanes present in RNG is the potential for the silica microparticulates escaping into the environment. This prospect is more of concern, of course, with home appliances (e.g., stoves, ovens, indoor furnaces, etc.) rather than with NG equipment which generally operate outdoors. As previously noted, for the measurement of the particulates in the engine flue-gas a scanning mobility particle sizer was utilized, manufactured by TSI (model 3936). The flue-gas was sampled at the exit of the catalyst bed. 148 Figure 3.15: Particle size distribution in the flue-gas. Figure 3.15 above shows the particle size distribution for both the siloxane and the non- siloxane engines. The SMPS is unable to detect any particulates in the non-siloxane engine, while the same is not true for the siloxane engine for which the SMPS detects particles ranging in size from 10 nm to 180 nm with a mean particle size of ~73 nm, and a mass concentration of 13 μg/m 3 . One should be reminded, furthermore, that these silica particulates are those that have managed to escape the catalyst monolith. 149 3.4. Conclusions In this chapter, an internal combustion engine operating on natural gas spiked with two common siloxanes has been studied experimentally, with the goal of understanding the impact of siloxane impurities on equipment (an IC engine) and environment. These impurities were shown to completely decompose during NG combustion in the engine to form silica microparticulates. These coat the internal metal surfaces in the engine, such as the piston heads, as well as the engine’s oxygen sensors and spark-plug, and they also collect in the engine oil. A certain fraction of them, furthermore, are carried out of the engine in the flue-gas and they deposit inside a catalyst monolith bed placed downstream of the engine resulting in its severe deactivation. In addition, a fraction of these particles of submicron size escape through the catalyst bed. These engine studies are consistent with prior fundamental studies which indicate that siloxane impurities readily decompose in the RNG combustion environment to form silica particulates that coat exposed metal surfaces. They also point out the critical importance of adequately removing these siloxane impurities from RNG prior to its use for engine performance. 150 Chapter 4 Effect of Siloxanes Contained in Natural Gas on the Operation of a Residential Furnace The research presented in this chapter has already been published [Nair et al., 2013] 151 4.1. Introduction There are many facilities worldwide using biomethane (biogas or LFG) for production of power or electricity, and concerns about global warming are likely to encourage their further capture and utilization [Mcbean et al., 2008]. Unfortunately, biogas and LFG contain in addition to methane a number of undesirable trace contaminants, including sulfided (e.g., H 2 S, COS, CS 2 ) and halogenated compounds [Ohannessian et al., 2008; Abatzoglou et al., 2009; Shin et al., 2002; Boulinguiez et al., 2010; Eklund et al., 1998] as well as a particularly problematic class of trace constituents known as siloxanes. During combustion, siloxanes decompose into silica microparticulates (see discussion in Section 3.1) which pose a threat to equipment, if not removed from the gas, but also to the environment unless adequate measures are taken to remove them from the flue-gas. This is the key driver for this and prior studies by the authors [Jalali et al., 2013; Nair et al., 2012] because of the prospect of using renewable natural gas (RNG). RNG is biogas after its impurities have been removed and its methane content has been upgraded to meet natural gas (NG) pipeline-quality standards. A major concern is the potential malfunction of equipment for siloxane removal from RNG – see discussion below – which could lead into their accidental release into the NG supply (pipeline) system. Combustion of natural gas containing siloxanes may result in the formation of silica microparticulates that may form deposits in home appliances (e.g., furnaces, stoves, water heaters, etc.), thus causing 152 operating issues, as well create harmful environmental emissions [El-Fadel et al., 1997; Abatzoglou et al., 2009]. Because of the challenges siloxanes present to the beneficial use of biomethane, they have attracted the attention of researchers in the renewable energy area. As one starts using RNG, the potential for these impurities finding their way into NG equipment and common household appliances remains a key challenge, and highlights the need for systematic studies of the fate of siloxanes during combustion in such devices and their impact on performance and associated emissions. There are reports of field-scale observations of silica films forming in equipment operating on biogas [Dewil et al., 2006; Schweigkofler et al., 2001] and proving difficult to remove. These abrasive silica deposits forming on engine pistons [Nair et al., 2012] on gas turbine blades, and on heat exchange surfaces in boilers, have the potential to cause serious equipment damage, necessitating more frequent maintenance, and increasing the operating cost of these devices [Ajhar et al., 2010; Popat et al., 2008]. Siloxanes also interfere with the operation of catalytic treatment systems of the exhaust gas, often decreasing their efficiency [Ohannessian et al., 2008; Nair et al., 2012; Badjagbo et al., 2010]. Another potential concern with the fine silica microparticles that form, is that they may be carried out in the flue-gas of the NG equipment [Nair et al., 2012], and unless adequate measures are taken to remove them from the exhaust, they are likely to escape into the atmosphere, where they may pose a risk to both human health and the 153 environment [El-Fadel et al., 1997; Abatzoglou et al., 2009]. The same concern exists with unvented home appliances such as stoves and ovens operating on RNG. Attention has been focused in recent years on the fate of siloxane impurities during RNG combustion in NG equipment and common household appliances and the impact on their performance as well as on human health and the environment. Though there are reports of field observations of inorganic deposits on various internal surfaces of combustion equipment, the authors are not aware of any open literature reports of systematic studies, under well-controlled laboratory conditions, other than recent studies by the authors [Jalali et al., 2013; Nair et al., 2012] investigating the various phenomena involved. This paper represents the third in a series of publications by the authors aiming to remedy the situation and to bridge this knowledge gap. The first paper in the series [Jalali et al., 2013] reports the results of fundamental combustion studies of siloxane decomposition in a simulated RNG/air flat flame established in the counterflow configuration. A key conclusion from that study was that siloxanes completely decompose in the RNG combustion environment to form pure silica particulates that coat exposed metal surfaces. The model combustion experimental configuration allowed the determination of the global reaction kinetics of siloxane decomposition in the RNG flame environment. The focus of the second paper [Nair et al., 2012] was on the understanding of the impact of siloxane impurities on the performance of an internal combustion (IC) engine 154 operating on real NG laced with trace amounts of two common siloxanes (L2 and D4, see Figure 4.1). (a) (b) Figure 4.1: (a) L2 siloxane. (b) D4 siloxane. These impurities were shown to decompose completely during NG combustion in the engine to form silica microparticulates, which coated the internal metal surfaces in the engine (e.g., the piston heads) as well as the engine’s oxygen sensors and spark plugs, and they were collected also in the engine oil. A certain fraction of them were carried out of the engine in the flue-gas and deposited inside a catalyst monolith bed placed 155 downstream of the engine resulting in its severe deactivation. These engine studies [Nair et al., 2012] are consistent with the prior fundamental studies by the authors [Jalali et al., 2013] that indicated that siloxane impurities decompose readily in the NG combustion environment to form silica particulates that coat exposed metal surfaces. They also point out the critical importance for engine performance of adequately removing these siloxane impurities from NG prior to its use. The goal of the present study is to investigate the impact of the presence of siloxane impurities in RNG on a common household appliance, namely a residential pulse combustion furnace. Considering the fact that RNG is already being blended into the natural gas pipeline system in California and the possibility of having more of it added in the future, the potential is there for siloxane impurities inadvertently finding their way into people’s homes, and for being combusted in common household appliances. The choice of this particular furnace for the study was because it uses a flame sensor, and prior anecdotal evidence existed that this type of furnace had flame sensing issues when operating with RNG containing siloxanes (after initiating testing of the furnace, the authors were also made aware of anecdotal evidence that boilers and water-heaters that use a flame sensor had ignition issues when operating with RNG containing siloxanes as well). In what follows, the experimental set-up and the techniques utilized in this study are described first. The experimental observations are presented and then discussed in terms of the potential implications of the siloxanes presence in NG on the operation of a real home appliance, namely a residential furnace. 156 4.2. Experimental Set-up A Lennox 80,000 Btuh PULSE Series Direct Vent Central furnace (model no. G14Q3- 80-19) was utilized; this furnace prior to this study was used in a residence for a number of years. A photograph of the furnace is shown in Figure 4.2. 157 Figure 4.2: Furnace 158 This furnace comes with a flame sensor (Part Number: 64K60), illustrated in Figure 4.3, that is used as a safety device for shutting the furnace down when no flame is detected inside the combustion chamber. Figure 4.3: Flame sensor The furnace flame sensor was connected to a HP 3465B multimeter, in-series, which measured the flame sensor’s current output. The main flow of NG to the furnace was controlled by a NG regulator, and was monitored via gas-flow meters (American Meter Company, model DTM 200A). Figure 4.4 shows the schematic of the overall experimental furnace set-up. 159 Figure 4.4: Experimental configuration. To generate various ppm v -level concentrations of siloxanes (ranging from 2 ppm v to 20 ppm v of an equimolar mixture of (L2+D4), a high-precision syringe pump (Harvard Apparatus model PHD 2000) coupled to a quartz nebulizer with a flush capillary-lapped nozzle (Meinhard model TR-20-A0.5) was utilized to generate a fine siloxane spray into a heated (via a heating tape) NG stream. The flow rate was controlled by a mass flow controller (Cole Parmer model# 55932). The siloxane-containing NG stream was then 160 mixed with the main NG stream in order to generate the NG feed-stream into the furnace with the desired siloxane concentration. The ability of the set-up to generate reliably the required total siloxane concentrations was verified by periodic injection of gas samples into a Gas Chromatograph equipped with a mass spectrometric detector (GC/MSD) system (Agilent Technologies, 7890A GC System/5975C Inert XL MSD). The GC/MSD system was calibrated using standard liquid samples of the two siloxanes in ethanol. The GC/MSD measurements were also cross-checked with the estimated concentrations based on the amount of liquid siloxane injected and the gas flow rates. The range of siloxane concentrations (2-20 ppm v ) investigated in this study is very much in line with what is encountered in the biogas from various landfills around the world. For example, Wheless and Gary, 2002, report the amount of siloxane found in two landfill sites (C-Modern and C-Kiefer) to be 20 ppm V . According to Dewil et al., 2006, the siloxane concentration ranges from 4.8 mg/m 3 up to 400 mg/m 3 , averaging 71.4 mg/m 3 . The conversion factor between these two common ways to express siloxane concentration depends on the M.W. of the given siloxane. For example, for L2 1 mg/m 3 corresponds to 0.138 ppm v , so the aforementioned range by Dewil et al., 2006, corresponds to (0.7 – 55.2 ppm v ), averaging 9.9 ppm v , which is the mid-point of the concentration range of (2-20) ppm v investigated in this study. 161 In order to monitor the impact of siloxanes on the overall performance of the furnace, two Moore Industries TRY PC Programmable Temperature Transmitters were installed at the return air opening, and at the supply air opening of the furnace. The data recorded by the temperature transmitters were logged in by a Logic Beach Intellilogger IL-80. Deposits on the sensor were analyzed by Scanning Electron Microscopy (SEM) and Energy Dispersive X-Ray Spectroscopy (EDX). The furnace was operated in the presence of siloxanes for a total period of 218.5 hr. A key focus of the study was the fate of the silica particles that form after combustion. For that purpose, the water condensate coming from the furnace flue-gas was collected, evaporated, and the solid residue analyzed using EDX. To verify whether silica microparticulates manage to escape into the furnace exhaust, the flue-gas was analyzed at the end of the exhaust pipe, via the use of a Scanning Mobility Particle Sizer (SMPS). Upon termination of the study, the furnace was carefully dismantled, and its various parts were visually inspected and photographed, and the chemical nature of the deposits was analyzed by EDX. 162 4.3. Results and Discussion Figure 4.5 summarizes the performance of the furnace sensor in terms of its current measured as a function of time on stream, starting with the time a given sensor was installed on the home furnace. Figure 4.5: Sensor current vs. time on stream for various siloxane concentrations, ppm v . 0 1 2 3 4 5 6 7 0 10 20 30 40 50 60 70 80 Sensor Current ( μA) Time (hr) 20 ppm siloxane 20 ppm siloxane(repeat) 10 ppm siloxane 2 ppm siloxane No siloxane 163 Since exposure to siloxanes completely damages the sensors to the point where they are no longer able to detect the flame – see discussion to follow – this necessitated frequent replacement of the sensors; as a result, each current vs. time profile, corresponding to a given concentration of siloxanes in Figure 4.5, was measured with a different sensor. A sensor would be declared completely damaged from exposure to siloxanes, if the furnace would not start after three manual attempts, each manual attempt consisting of five automatic tries by the ignition controller. Prior to testing the sensor in the presence of siloxanes, the sensor’s performance was tested when operating on pure NG, without any siloxanes added. No current changes were detected, and the sensor performed well, with the furnace igniting without any issues. The first 45 hr of this sensor’s operation, operating on pure NG are shown in Figure 4.5 (showing no change in current output), serving as the basis of comparison for the subsequent tests. The experiments, the data from which are plotted in Figure 4.5, were carried out in order to study the impact of varying the concentration level of siloxanes intermixed with the natural gas on sensor performance. Three different siloxane (equimolar (L2+D4) mixture) concentrations were tested, namely 20 ppm v , 10 ppm v , and 2 ppm v , the latter representing the lowest concentration level that the siloxane delivery system used in this study can generate reliably. Each experiment used a new sensor. Two experiments using a different sensor but the same concentration of siloxanes (20 ppm v ) are also shown in Figure 4.5. The end-point of each current vs. time profile in Figure 4.5 signifies the time when the sensor stopped operating, and the furnace would no longer start. As the data in Figure 4.5 164 indicate, there are some differences between the “failure times” among the sensors operating on various levels of siloxanes. For example, the sensor operating on 2 ppm v of siloxanes failed approximately 28 hours later than the one operating on 20 ppm v , but on the other hand the one operating at 10 ppm v failed approximately 10 hours earlier. No clear trends are obvious, however, and in fact whatever differences exist they may be attributable to differences among the properties of the sensors themselves (see also discussion below about uncertainties with the sensor’s failure test itself). Furthermore, what exactly causes a certain sensor to fail and prevents the furnace from starting is not entirely clear either, since the starting currents for the sensors seem to vary widely, as is the final current at which the sensors fail. The lack of apparent trends with respect to the concentration of siloxanes is evident in Figures 4.6(a) and 4.6(b). 165 (a) (b) Figure 4.6: (a) Sensor current vs. moles of siloxane fed to the furnace for a siloxane concentration of 20 ppm v ; the experiment was run twice each time with a different sensor. (b) Sensor current vs. moles of siloxane fed to the furnace for two different siloxane concentrations of 2 ppm v and 10 ppm v each tested with a different sensor. 0 1 2 3 4 5 6 0 10 20 30 40 50 60 70 80 90 Sensor Current ( μA) Moles of siloxane (mmoles) 20 ppm siloxane 20 ppm siloxane (repeat) 0 1 2 3 4 5 6 7 0 5 10 15 20 25 30 35 Sensor Current ( μA) Moles of siloxane (mmoles) 10 ppm siloxane 2 ppm siloxane 166 In these figures, results are plotted in terms of sensor current vs. the total amount of siloxanes injected into the furnace, starting from the point when this particular sensor was installed to the point when it stopped operating. There is no clear correlation between the total amount of siloxanes introduced into the furnace and the time that it takes for the sensor to fail. This lack of trends may be also a result of the inexact nature of the condition that defines the failure mode for the sensor (three manual attempts, each manual attempt consisting of five automatic tries by the ignition controller to re-start from the first moment of furnace shut-down), in addition to the variability among the various sensors. It was observed, for example, that for some of the failed sensors after waiting overnight, the furnace would re-start the next day and would operate a few additional hours before failing again. However, field experience indicates that a typical homeowner is likely, after attempting a few times to re-start the furnace, to conclude that the furnace has already failed and to call for service rather than attempting to re-start the furnace the next day. All these technical issues aside, a key conclusion from the data in Figures 4.5 and 4.6 is that the presence of siloxanes, even at single-digit ppm v concentration levels, has a very detrimental effect on furnace operation. This conclusion is very much in line with conclusions reached in the study of Nair et al., 2012, where the presence of siloxanes in NG was shown to have a very detrimental effect on the performance of the oxygen sensors for the IC engine investigated. 167 The reasons for sensor failure become clearer when observing Figure 4.7, which shows the SEM image of the tip of the sensor (left side in Figure 4.3). Figure 4.7: SEM image of the tip of the flame sensor showing the silica layer (indicated by the double-arrow on the left). On the surface of the tip of the sensor exposed to the siloxane-containing natural gas one observes the formation of a several hundred micrometers thick layer of deposits (similar deposits also form along all other metal surfaces of the sensor). Figure 4.8 depicts an EDX line-scan indicating the (Si/Cr) ratio as a function of position across the same cross- section of the sensor. Silica layer ~ 345microns thick 168 Figure 4.8: EDX analysis of the surface of the tip of the flame sensor showing the silica layer thickness. Si-containing deposits, several hundred micrometers thick, can be seen forming on the sensor surface exposed to the siloxane-spiked natural gas (the lack of silica deposits in the middle is because the deposits there were removed in order to be able to determine their thickness in Figure 4.7). The top layers of the silica deposits appeared to be flaky and could be removed with ease. The bottom layers however, adhered strongly to the metal surface. Upon the completion of the experiments, the home furnace was carefully dismantled and its various internal parts were examined for the presence of deposits. Figure 4.9 shows a detailed engineering diagram of the home furnace illustrated in Figure 4.2, indicating the 0 1 2 3 4 5 6 7 8 9 0 500 1000 1500 2000 2500 Si/Cr Distance in Microns 169 various internal parts of the furnace that were removed and further examined for the presence of Si-containing deposits. Figure 4.9: Schematic diagram of the furnace, showing the parts, which were analyzed by EDX for the presence of silica deposits. Figure 4.10(a) shows the presence of thick deposits on the internal surface of one of the coils in the condenser coil assembly. EDX analysis of these deposits indicates that they are pure silica, as shown in Figure 4.10(b). Deposits were detected also on the upper part 170 of the combustion chamber as well as on the tubing connected to the combustion chamber. 171 (a) (b) Figure 4.10: (a) Deposits on the condenser coil assembly of the furnace. (b) EDX analysis of the deposits on the condenser assembly. Silica Deposit 172 EDX analyses of deposits from the combustion chamber (Figure 4.11(a)) and from both the tubing furthest away (Figure 4.10(b)) and closest to the combustion chamber (not shown here) indicate that they are silica films. 173 (a) (b) Figure 4.11: (a) EDX analysis of the deposits on the top-half of the combustion chamber. (b) EDX analysis of the deposits on the tubing farthest from the combustion chamber. 174 Silica was detected also in the condensate water collected from the furnace, as the EDX analysis (shown in Figure 4.12) of the dried residual matter from that water indicates. Figure 4.12: EDX analysis showing kilo counts of elements of the dried residue from the furnace water condensate. A certain fraction of the silica particles (beyond those attached on the furnace surfaces and dispersed in the condensate) still get entrained in the flue-gas exiting the home furnace. Figure 4.13 shows the size distribution of the particles found in the flue-gas of the furnace, as detected by SMPS. 175 Figure 4.13: Particle size distribution in the flue-gas from the furnace operating with and without siloxanes. When the furnace operates with pure natural gas under these fuel-lean operating conditions, the SMPS detects no particulate matter in the flue-gas. The same is not true when the furnace is operating on siloxane-spiked natural gas, even at the lowest 2 ppm v concentration level for which SMPS detects the presence of silica microparticulates in the flue-gas, with an average particle diameter of ~ 75 nm. Although these nano-particles may have some health implications [Napierska et al., 2010], the real risk needs to be further evaluated in practical settings. Despite the presence of silica deposits throughout the internal surface of the furnace, including those causing significant blockage of the condenser coil assembly (as shown in Figure 4.10(a)), during operation of this furnace there was no discernible impact of such deposits on its heat transfer characteristics. 0 500 1000 1500 2000 2500 3000 0 100 200 300 400 500 600 700 Particle Count (#/cm 3 ) Diameter of particles(nm) With 2 ppm Siloxane Non-Siloxane 176 4.4. Concluding Remarks In this study, a residential pulse-combustion furnace operating on natural gas spiked with two common siloxanes (L2+D4) has been studied experimentally with the goal of understanding the impact of siloxane impurities on furnace performance. These impurities were shown to decompose completely during NG combustion in the furnace to form silica microparticulates and films. These microparticulates coated internal surfaces of the furnace and its components, such as the flame sensor, the condenser coils and the tailpipes. Silica particulates also accumulated in the condensate water coming from the flue-gas. In addition, a fraction of the silica particles of submicron size escape, via entrainment, through the flue-gas. The coating on the flame sensor proved particularly detrimental for furnace performance, as after a certain period of exposure to siloxanes the sensor stopped detecting the flame, thus causing the furnace to stop operating. These furnace studies, in agreement with prior fundamental combustion studies by the authors 10 , indicate that siloxane impurities readily decompose in the NG combustion environment to form silica particulates that coat some of the exposed metal surfaces and seriously interfere with furnace operation. They also highlight the need for completely removing these impurities from biogas prior to its use in residential appliances. 177 Chapter 5 A Novel Technique for the Removal of Siloxanes from Landfill Gas using UV Photodecomposition 178 5.1. Introduction As mentioned in the previous chapters the use of biogas has its own challenges and none of them more important than removing siloxanes. The silicon-containing NMOC, specifically various volatile siloxanes, which are the focus of the present Chapter, are converted into silica microparticles, and, therefore, constitute a potential source of particulate air pollution. In addition, these particles build-up on various surfaces in the boiler tubes and the combustion chambers, causing reduction in catalytic activity of engines and most importantly necessitating frequent and expensive maintenance as discussed in Chapters 3 and 4. As noted previously in this Thesis, a number of techniques have been investigated in the past for the removal of the halogenated and sulfided NMOC from biogas, but they have generally not proven successful to rid the gas of the volatile siloxanes. The siloxane compounds often encountered in LFG and their properties are shown in Table 3.2, in Chapter 1 [Jalali et al., 2012]. Their concentrations, typically ~15 mg/m 3 (but concentrations as high as 50 mg/m 3 are not uncommon) often range beyond the limits placed by most manufacturers (see Table 3.1 for a list of such limits); as a result, most engine manufacturers today highly recommend gas pretreatment to remove the siloxanes. 179 As with the treatment of the other trace constituents in LFG, adsorption, absorption and refrigeration are the techniques currently utilized, which are expensive to apply, but remain commonly in use because there are no other commercially available processes to replace them, thus the focus in this effort. Adsorbents used to remove gaseous siloxanes, include activated carbon, activated alumina and silica gel. Ricaurte Ortega and Subrenat [2009] compared the application of different porous media including activated carbon, zeolite and silica gel in adsorption process. They observed that the mass transfer within the porous material was more rapid for activated carbon than for the zeolite and silica gel. Also, they found that light siloxanes (L 2 – Hexamethyldisiloxane) with a lower boiling point breakthrough much faster than their heavier counterparts, necessitating frequent regeneration which is relatively expensive, and because of the process complexity carries with it the risk for system failure. Typically, the adsorption media utilized are not particularly selective to siloxanes, and will adsorb most other heavy hydrocarbons, thus the NMOC present in the gas significantly reduce the bed capacity for siloxanes. Normally adsorption capacities (defined by the siloxane adsorbed when breakthrough is detected) for various active carbon media range from 2,000-17,000 mg/lb (0.4 - 3.7 wt.%) with virgin media having the highest capacity [Wheless et al., 2002]. Since the adsorption capacity of regenerated media, is less than that of virgin media, successive onsite regenerations yield a progressively lower capacity, until media replacement is necessary, which comes at a great cost. One key challenge with adsorption is that it does not change the siloxane molecules, which when released from the beds are still the same as when 180 entering the beds. Siloxane disposal, during regeneration, involves burning the off-gas, which releases silica oxide particles to the atmosphere. Another approach for the removal of gaseous siloxanes is physical absorption using solvents like Selexol TM and methanol operating at relatively high pressures in order to be effective. In a study conducted by Schweigkofler et al. (2001), hot concentrated nitric acid was used in the absorption process to remove siloxanes. They found that elevated temperature was necessary in this application and 70 – 75% removal efficiency was achieved. Solvent regeneration is also key to such processes in order to reduce solvents disposal and system operating costs, and to assure long-term operation without significant consumption of the scrubbing solution (solvent recycle impacts the siloxane removal efficiency, however, due to its accumulation in the solvent). The capital/operating costs of the absorption processes are similar to those for adsorption, but performance is inferior and significant process improvement is, therefore, needed. For the refrigeration process, based on the study at the Calabasas landfill microturbine facility, Wheless (2002) reported that more than 88% of the initial concentrations were still present. Some studies focused on the combination of traditional technologies and new adsorption/absorption media were also reported. Schweigkofler et al. [2001] used a refrigeration condenser with the adsorption bed to remove 98% of siloxanes. However, the high energy consumed by the refrigeration equipment for cooling huge amounts of 181 sewage gas is the drawback. They also reported the use of meadow ore for the adsorption process. Low efficiency (31 – 75%) of siloxane removal was observed. Some non-traditional methods for siloxane removal were developed over these years. Popat et al. [2008] used biological treatment to control the siloxanes. A low efficiency (~10%) was observed in their study. Ajhar et al. [2006] used AspenPlus simulation models to identify gas permeation as a potential new method for siloxane removal. However, this technology is in its early stages and no experimental data are available. Appels et al. [2008] conducted a study using a peroxidation method to reduce the siloxane content. A reduction of 50 – 85% was observed, which is not so effective when compared with existing technologies. It should be clear from the above discussion that the commonly utilized technique for siloxane removal from LFG and biogas face significant technical and economic hurdles, and the new techniques available still do not show high destruction efficiencies. There is a need, therefore, to develop novel techniques to rid these gases of these components efficiently. UV photodecomposition is a promising technique in that regard, as it has been previously shown effective in the treatment of a variety of organic compounds at trace amounts in air and other contaminated gas streams. UV photodecomposition has, in fact, been studied by a number of previous studies in order to decompose a number of silicon containing compounds [Dalton et al., 1985; Ouchi et al., 1999; Urbanova et al., 2001; Yingling et al., 2003]. Urbanova et al. [2001], for example, compared IR laser 182 thermolysis and UV laser photolysis for the decomposition of 1,3-diethyldisiloxane ([H 2 (C 2 H 5 )Si] 2 O). Their study revealed that typical hydrocarbon products of UV photolysis are ethane (C 2 H 6 , a major product) accompanied by smaller amounts of ethylene (C 2 H 4 ), methane (CH 4 ), propane (C 3 H 8 ), and butane (C 4 H 10 ). They reported that the cleavage of the Si-O bond does not take place under photolysis. Dalton also showed that Si-O bond cleavage is not a feasible photochemical step. Even though their findings provide great insight into the photodecomposition of siloxane, their work focuses on the reaction pathway of photoreaction and there is no reaction rate information available to systematically evaluate the removal of siloxanes. During the photochemical reaction of siloxane in LFG, a small amount of oxygen will be typically present which is likely to participate in the photoreaction. During the photodecomposition process, the Si-C bonds of the siloxane and O=O bonds of oxygen are broken upon receiving the UV energy [Kockarts et al., 1976; Dalton et al., 1985; Ouchi et al., 1999; Pola et al., 2002] . In the reaction that then follows, the generated Si- containing radical is quickly intercepted by oxygen radicals to form solid deposit material [Ouyang et al., 2000; Brinkmann et al., 2001; Pola et al., 2002]. The energy needed to perform the photodecomposition of siloxanes is wavelength-dependent, each compound characterized by its own absorption spectrum. The absorption spectra of some silicon- containing organics are shown, for example, in Figure 5.1 below. The absorption cross- section of hexamethyldisiloxane (L 2 ) has a maximum value of 3.5x10 -17 cm 2 /molecule at a wavelength of 180 nm. 183 Figure 5.1: Absorption spectrum of some silicon-containing compounds As noted above, for the reaction the existence of reactive oxygen radicals is important, and ozone can serve as one of the sources. Ozone is a naturally occurring allotrope of oxygen that can be produced from oxygen in situ during UV radiation. For example, UV wavelengths shorter than 200 nm are capable of producing ozone from molecular oxygen and can, at the same time, break the Si-CH 3 bond, creating reactive fragments which are susceptible to an ozone attack. Owing to the fact that the absorption spectrum of CH 4 is located in the IR, UV radiation is not likely to cleave CH 4 , the most abundant organic compound in landfill gas. Therefore, in the proposed application of the photochemical clean-up process in biogas, one can expect a high reaction rate for the photodecomposition of siloxane with almost no effect on the fuel quality of the biogas. 184 5.2. Experimental Results and Discussion Before starting the main experiments, some preliminary experiments were carried out with inexpensive, and readily available UV measurement devices. These experiments were carried out with the goal of understanding the effect of the UV energy on the siloxane. Table 5.1 below shows the absorption cross-section of some of the LFG gas compounds for different wavelengths. A point to be noted here is that the larger the absorption cross-section the greater the chance of the compound reacting with the UV light. Compound Absoprtion Cross-section (cm 2 /molecule) 185 nm Wavelength 254 nm Wavelength Carbonyl Sulfide 1.90E-19 1.00E-20 Vinyl Chloride 3.40E-17 0 Trichlorofluoromethane 2.55E-18 1.00E-22 Dimethyl Sulfide 1.00E-18 0 Dichlorobenzene N/A N/A L2 1.50E-17 0 D4 N/A N/A Water 2.50E-18 0 Table 5.1: Absorption cross-section of different LFG components In the preliminary experiments two MiniRAE 2000 gas analyzers, shown in Figure 5.2 below, were used. The first contained a 10.6 eV radio frequency excited lamp and the other a 11.7 eV radio frequency excited lamps, each with a power rating of 0.1 watt. 185 Figure 5.2: MiniRAE 2000 showing the inlet and outlet ports As noted above, these gas analyzers were studied for the sole purpose of using their high- energy UV lamps for reaction. The gases going in and out of the instrument were measured for their composition by a GC/MS (GC-HP 6890 Series, MS Detector-HP 5973, GC Column- HP-624) with the following settings: GC temperature-31 o C rising at 5.0 o C/min, final temperature- 230 o C; GC flow-24.8 psi, 1.6 L/min, 51 cm/sec; MS temperature-200 o C. The experiments were carried out with LFG gas used (herein after referred to as LFG2) in Chapter 2 whose composition is shown in Table 2.2 (containing Gas Outlet Port Gas Inlet Port 186 30% CO 2 , 9%N 2 , 1%O 2 with 50 ppm Carbonyl Sulfide, 50 ppm Vinyl Chloride, 50 ppm Trichlorofluoromethane, 50 ppm Dimethyl Sulfide, 5 ppm DCB, and balance CH 4 ). The concentration of the siloxanes was kept at 5ppm L2 and 5 ppm D4 simulating the typical concentrations observed in landfill sites. The results for the reaction of (5ppm L2 and 5ppm D4) mixture for these two lamps are given in Table 5.2 below. Compound %Conversion 10.6 eV lamp 11.7 eV lamp D4 30 36 L2 22 24 Table 5.2: Conversion of L2 and D4 (5ppm+5ppm) for different energy lamps Apart from siloxanes some conversion was also observed for the chlorinated and sulfided contaminants in LFG. The results for these compounds are shown in Table 5.3. (The conversion rates for DCB are not reported in this Table due to experimental difficulties with the analysis method; efforts are currently underway to resolve this problem). 187 Compound %Conversion 10.6 eV lamp 11.7 eV lamp Carbonyl Sulfide 5 16 Vinyl Chloride 8 8 Trichlorofluoromethane 4 3 Dimethyl Sulfide 7 3 Table 5.3: Conversion of chlorinated and sulfided compounds for different energy lamps The above results show that the higher the energy the higher is the conversion of the siloxanes. The data above also shows that some of the chlorinated and sulfided compounds can also be removed from LFG and this application will be the focus of our future work. Since the above UV lamps were of low power (0.1 watt) and similar higher power lamps are not commercially available we continued with our studies using an Atlantic Ultraviolet Corp. 16 W lamp. The picture of this lamp is given in Figure 5.3 below. 188 (a) (b) Figure 5.3: (a) UV lmap. (b) UV lamp enclosed in a PVC pipe and modified for reactor operations Air + Siloxane PVC pipe Quartz tube UV Lamp Ballast Modified UV lamp to collect the sample Gas Sample Point 189 In the initial set of experiments the UV photo-oxidation reactor shown in Figure 5.3b, which makes use of a UV lamp (G10T51/2VH, 16 W, Atlantic Ultraviolet Corp.) with electronic ballast, siloxane mixture inlet, and high purity fused quartz, was utilized. This outlet of quartz tube has been modified for the sampling of product gas. These UV lamps operate in much the same way as fluorescent lamps. UV radiation is emitted from electron flow through ionized mercury vapor to produce UV energy in most units. The difference between the two lamps is that the fluorescent lamp bulb is coated with phosphorous, which converts the UV radiation to visible light. The UV lamp is not coated, so it transmits the UV radiation generated by the arc. These UV-C lamps emit their maximum energy output at a wavelength of 254nm (4.8 eV) and 185nm (6.7 eV) with high-purity quartz. The lamps typically used in UV consist of a quartz tube filled with an inert gas, such as argon, and a small amount of mercury. Ballasts are transformers that control the power to the UV lamps. The entire experimental set-up for the UV photodecomposition study is shown below in figure 5.4. 190 Figure 5.4: Experimental set-up In our experiments, the siloxane vapor is generated by the nebulizer (Figure 5.4). The siloxane/LFG2 gas mixture was fed into the reaction chamber shown in Figure 5.4. The analysis of siloxane and reaction products were carried out by GC/MS (GC-HP 6890 Series, MS Detector-HP 5973, GC Column- HP-624) with the same GC oven settings as used in the preliminary experiments. The product was sampled by a needle syringe (5 L NS104305, National Scientific Comp.). A 0.2 m pore syringe filter (#F2504-4, National Scientific Comp.) was used for the syringe to remove particulates from the gas sample before being analyzed. The flow rate of the gas stream was varied to find the effect of residence time on the conversion of the siloxanes. The results of this experiment are given in Figure 5.5 below. The results show that as the flow rate decreases (residence time increases) the conversion of the siloxanes increases. 191 Figure 5.5: Conversion of 10ppm L2 and 10ppm D4 in LFG2 gas for different flow- rates As noted in the preliminary experiments earlier some conversion was also observed for the chlorinated and sulfided VOC’s present in the LFG. The results of their conversion vs. residence time are given below in Figure 5.6. 192 Figure 5.6: Effect of flow-rate on the conversion of VOC compounds in LFG 193 5.3 Conclusions Based on the experimental results 80% L 2 and 60% D 4 siloxane removal can be achieved for 1 cc/s flow-rate. The UV reactor has also demonstrated some ability to remove the chlorinated and sulfided NMOC’s present in simulated LFG. Studies are currently under way on the effect of the moisture content present in LFG. The efficiency of these reactors can be further expanded by use of reflective coatings. Studies are currently underway, therefore, by using the GORE ® Diffuse Reflector coatings to see if it can improve the efficiency of this reactor. A high power 65W lamp is also currently being studied in addition to the 16W lamp. This will help in determining the effect of power on the conversion of these impurities. 194 Chapter 6 Ideas for Future Work 195 The results obtained from both the FTCMR and UV photodecomposition experiments show good promise in the use of these techniques for removal of contaminants from LFG. The use of UV light in the removal of NMOC’s, as shown in the preliminary experiments in Chapter 5, needs to be investigated further in order to optimize the whole system. Although both techniques have and are being tested in the laboratory, there are currently no data available under actual field conditions treating real biogas or LFG. Such tests are, therefore, strongly recommended. Another recommendation is to investigate the combined use of both techniques. The diagram of a simple pilot treatment plant set-up is given below in Figure 6.1 Figure 6.1: Schematic drawing of a LFG clean-up process with SiO 2 separation 196 The experimental data obtained from the FTCMR and UV photodecomposition lab-scale studies are still very preliminary in terms of being able to develop cost estimates of the proposed processes. Some preliminary estimates indicate that for the combined process the key challenge currently is with the efficiency of the utilization of the UV lamps, so the focus in our current and future efforts will be with the use of reflective coatings and the optimal placement of the UV lamps. For the FTCMR, the key challenge that lies ahead is improving the permeation flux of the membranes. Preliminary estimates indicate that a factor of 5-10 times increase in the permeation rates is what is needed to make the technology competitive. Currently, efforts are under way at MPT in improving the permeation characteristics of such membranes as part of a separate joint project that focuses on the use of the FTCMR technology for the destruction of chemical warfare agents. 197 References Abatzoglou, N., Boivin, S., “A Review of Biogas Purification Processes.”, Biofuels Bioproducts & Biorefining – Biofpr., 3(2009), 42. Accettola F., Guebitz G.M., Schoeftner R., “Siloxane Removal from Biogas by Biofiltration: Biodegradation Studies.”, Clean Technol. Environ. Policy., 10 (2008), 211. Aida, T., Higuchi, R., & Niiyama, H. (1990). Decomposition of Freon-12 and Methyl Chloride over Supported Gold Catalysts. Chemistry Letters, 19(12), 2247-2250. Ajhar M., Melin T., “Siloxane Removal with Gas Permeation Membranes.”, Desalination., 200(2006), 234. Ajhar M., Travesset M., Yuece S., Melin T., “Siloxane Removal from Landfill and Digester Gas – A Technology Overview.”, Bioresour. Technol., 101(2010), 2913. Albo S.E., Broadbelt L.J., and Snurr R.Q., “Multiscale modeling of transport and residence times in nanostructured membranes”, AIChE Journal, 52 (2006) 3679–3687. Alfonso M., Menendez M., and Santamaria J., “Coupling of consecutive reactions in a two-layer, flow-through catalytic membrane”, Industrial & Engineering Chemistry Research, 40 (2001), 1058–1064. Appels L., Baeyens J., Dewil R., “Siloxane Removal from Biosolids by 198 Peroxidation.”, Energy Convers. Manage., 49(2008), 2859. Augenstein D., Pacey J. 1992, Landfill Gas Energy Utilization: Technology Options and Case Studies, EPA-600-R-92-116, U.S. Environmental Protection Agency, Research Triangle Park. Badjagbo K., Heroux M., Alaee M., Moore S., Sauve S., “Quantitative Analysis of Volatile Methylsiloxanes in Waste-to-Energy Landfill Biogases Using Direct APCI-MS/MS.”, Environ. Sci. Technol., 44(2010), 600. Baker, L.; Capouya, R.; Cenci, C.; Crooks, R.; Hwang, R. The landfill testing program; data analysis and evaluation guidelines. CAPCOA/CARB Report, Sept 1990. Bickle, G. M., Suzuki, T., & Mitarai, Y. (1994). Catalytic destruction of chlorofluorocarbons and toxic chlorinated hydrocarbons. Applied Catalysis B: Environmental, 4(2), 141-153. Boulinguiez, B., Le Cloirec, P., “Adsorption on Activated Carbons of Five Selected Volatile Organic Compounds Present in Biogas: Comparison of Granular and Fiber Cloth Materials.”, Energy Fuels., 24(2010), 4756. Brinkmann, M., Chan, V. Z. H., Thomas, E. L., Lee, V. Y., Miller, R. D., Hadjichristidis, N., & Avgeropoulos, A. (2001). Room-temperature synthesis of a-SiO2 thin films by UV-assisted ozonolysis of a polymer precursor. Chemistry of materials, 13(3), 967-972. Caro J., Caspary K., Hamel C., Hoting B., Kölsch P., Langanke B., Nassauer K., Schiestel T., Schmidt A., Schomäcker R.,, Seidel-Morgenstern A., 199 Tsotsas E., Voigt I., Wang H., WarsitzR., Werth S., and Wolf A., “Catalytic membrane reactors for partial oxidation using perovskite hollow fiber membranes and for partial hydrogenation using a catalytic membrane contactor”, Industrial & Engineering Chemistry Research, 46 (2007), 2286–2294. Cavenati Simone, Grande Carlos A, “Upgrade of methane from landfill gas by pressure swing adsorption”, Energy & Fuels, 19(2005), 2545-2555. Chea, V., Paolucci-Jeanjean, D., Belleville, M. P., & Sanchez, J. (2012). Optimization and characterization of an enzymatic membrane for the degradation of phenolic compounds. Catalysis Today, 193(1), 49-56. Chu, H., Chiou, Y. Y., Horng, K. H., & Tseng, T. K. (2001). Catalytic incineration of C2H5SH and its mixture with CH3SH over a Pt/Al2O3 catalyst. Journal of environmental engineering, 127(5), 438-442. Crest M., Chottier, C., Fine L., Chatain V., Ducom G., Chovelon J.M., Germain P., “On the Reliability of Sampling and Analytical Methods to Quantify Volatile Organo Silicon Compounds (VOSiC) Contents in Landfill Gas.”, Proceedings of the Third International Symposium on Energy from Biomass and Waste, Venice, Italy. (2010). Cullis C.F., Willatt B.M., “The Inhibition of Hydrocarbon Oxidation Over Supported Precious Metal Catalysts.”, Journal of Catalysis., 86(1984), 187. Dalton J.C. 1985, Organic Photochemistry, ed. Padwa A., Dekker M., New York, vol 7, ch3. Dai, Q., Bai, S., Wang, J., Li, M., Wang, X., & Lu, G. (2013). The effect 200 of TiO< sub> 2</sub> doping on catalytic performances of Ru/CeO< sub> 2</sub> catalysts during catalytic combustion of chlorobenzene. Applied Catalysis B: Environmental. DeMers, L.D. and Renner R.C. 1992, Alternative Disinfection Technologies For Small Drinking Water Systems. AWWARF. Devulapelli, V. G., & Sahle-Demessie, E. (2008). Catalytic oxidation of dimethyl sulfide with ozone: Effects of promoter and physico-chemical properties of metal oxide catalysts. Applied Catalysis A: General, 348(1), 86-93. Dewil R., Appels L., Baeyens J., “Energy use of biogas hampered by the presence of siloxanes”, Journal of Energy Conversion and Management, 47(2006) 1711-1722. Dewil, R., Appels, L., Baeyens, J., “Analysis of Volatile Siloxanes in Waste Activated Sludge.”, Talanta., 74(2007), 14. Dow Corning, Basic Silicon Production and Siloxane Polymerization www.dowcorning.com/content/sitech/sitechbasics/siloxane_polymerization.asp Dow Corning. 1997, Environmental information updates. “An overview of volatile Methylsiloxane (VMS) fluids in the environment”. Eklund, B., Anderson, E.P., Walker, B.L., Burrows, D.B., “Characterization of Landfill Gas Composition at the Fresh Kills Municipal Solid- Waste Landfill.”, Environ. Sci. Technol., 32(1998), 2233. 201 El-Fadel, M., Findikakis, A.N., Leckie, J.O., “Environmental Impacts of Solid Waste Landfilling”, J. Environ. Manage., 50(1997), 1. Finocchio E., Garuti G., Baldi M., Busca G., “Decomposition of Hexamethylcyclotrisiloxane over Solid Oxides.”, Chemosphere., 72(2008), 1659. Fritsch D., Randjelovic I., and Keil F., “Application of a forced-flow catalytic membrane reactor for the dimerisation of isobutene”, Catalysis Today, 98 (2004), 295–308. Fritsch D., and Bengtson G., “Catalytic polymer membranes for high temperature hydrogenation of viscous liquids”, Advanced Engineering Materials, 8 (2006a), 386–389. Fritsch D., and Bengtson G., “Development of catalytically reactive porous membranes for the selective hydrogenation of sunflower oil”, Catalysis Today, 118 (2006b), 21–127. García-García, F. R., & Li, K. (2013). New catalytic reactors prepared from symmetric and asymmetric ceramic hollow fibres. Applied Catalysis A: General. Gentry S. J., Jones A., “Poisoning and inhibition of catalytic oxidations: I. The effect of Silicone Vapour on the Gas-Phase Oxidations of Methane, Propene, Carbon Monoxide and Hydrogen over Platinum and Palladium Catalysts.”, J. Chem. Technol. Biotechnol., 28(1978), 727. Griffin, P. Siloxanes in Wastewater and Biogas, Seminar, Severn Trent Water, U.K, (2004). 202 Gröschel L., Haidar R., Beyer A., Colfen H., Frank B., Schomåcker R., “Hydrogenation of propyne in palladium-containing polyacrylic acid membranes and its characterization, Ind.Eng.Chem. Res., 44(2005) 9064-9070. Haidar R., Beyer A., Colfen H., Frank B., Schomåcker R., “Hydrogenation of propyne in palladium-containing polyacrylic acid membranes and its characterization, Ind.Eng.Chem. Res., 44(2005) 9064-9070. Harasimowicz, M., Orluk, P., Zakrzewska-Trznadel, G., Chmielewski, A. G., “Application of polyimide membranes for biogas purification and enrichment”, Journal of hazardous materials, 144(2007), 698-702. He C., Herman D., Minet R., Tsotsis T., “A catalytic/sorption hybrid process for landfill gas cleanup”, Journal of Industrial Engineering and Chemistry Research, 36(1997), 4100-4107. Heyes, C. J., Irwin, J. G., Johnson, H. A., & Moss, R. L. (1982). The catalytic oxidation of organic air pollutants part 1. Single metal oxide catalysts. Journal of Chemical Technology and Biotechnology, 32(7‐12), 1025-1033. Hwang, C. L., & Tai, N. H. (2011). Vapor phase oxidation of dimethyl sulfide with ozone over ion-exchanged zeolites. Applied Catalysis A: General, 393(1), 251-256. Ilinich O.M., Cuperus F.P., Nosova L.V., and Gribov E.N., “Catalytic membrane in reduction of aqueous nitrates: Operational principles and catalytic performance”, Catalysis Today, 56 (2000)-137–145. 203 Ilinich O.M., Gribov E. N., and Simonov P.A., “Water denitrification over catalytic membranes: Hydrogen spillover and catalytic activity of macroporous membranes loaded with Pd and Cu”, Catalysis Today, 82 (2003), 49–56. Jalali, A., Yousef Motamedhashemi, M. M., Egolfopoulos, F., & Tsotsis, T. (2013). Fate of Siloxane Impurities during the Combustion of Renewable Natural Gas. Combustion Science and Technology, (just-accepted). Khassin A.A., “Catalytic membrane reactor for conversion of syngas to liquid hydrocarbons”, Energeia, 16 (2005), 1–3. Khassin A., Sipatrov A., Yurieva T., Chermashentseva G., Rudina N., Parmon V., “Performance of a catalytic membrane reactor for the Fischer-Tropsch synthesis”, Catal.Today., 105(2005), 362-366. Kobayashi M., Togawa J., Kanno T., Horiuchi J.I., Tada K., “Dramatic innovation of propene epoxidation efficiency derived from a forced flow membrane reactor”, J.Chem.Technol. Biotechnol. 78(2003), 303-307. Kochkodan, V. M., Rolya, E. A., & Goncharuk, V. V. (2009). Photocatalytic membrane reactors for water treatment from organic pollutants. Journal of Water Chemistry and Technology, 31(4), 227-237. Kockarts G., “Absorption and Photodissociation in the Schumann-Runge Bands of Molecular Oxygen in the Terrestrial Atmosphere”, Space Sci. 24(1976), 589. 204 Krishnamoorthy, S., Rivas, J. A., & Amiridis, M. D., “Catalytic oxidation of 1, 2-dichlorobenzene over supported transition metal oxides.”, Journal of Catalysis, 193(2000), 264. Kułażyński, M., Van Ommen, J. G., Trawczyński, J., & Walendziewski, J. (2002). Catalytic combustion of trichloroethylene over TiO< sub> 2</sub>-SiO< sub> 2</sub> supported catalysts. Applied Catalysis B: Environmental, 36(3), 239-247. Lambert C., Gonzalez R., “Activity and selectivity of a Pd/gamma-Al 2 O 3 catalytic membrane in the partial hydrogenation reactions of acetylene and 1,3- butadiene”, Catal.Lett., 57(1999), 1-7. Lange C., Storck S., Tesche B., Maier W.F., “Selective hydrogenation reactions with a microporous membrane catalyst, prepared by sol-gel dip coating”, J.Catal., 175(1998), 280-293. Landfill methane outreach program, EPA, http://www.epa.gov/outreach/lmop/basicinfo/index.html Larsson A. C., Rahmani M., Arnby K., Sohrabi M., Skoglundh M., Cruise N., Sanati M., “Pilot-scale Investigation of Pt/alumina Catalysts Deactivation by Organosilicon in the Total Oxidation of Hydrocarbons.”, Topics in Catalysis., 45 (2007), 121. Lee, H. I., Kim, J. M., Lee, J. Y., Park, Y. K., Jeon, J. K., Yim, J. H., ... & Jeong, K. E. (2010). Catalytic conversion of 1, 2-dichlorobenzene over mesoporous materials from zeolite. Journal of nanoscience and nanotechnology, 205 10(5), 3639-3642. Libanati C., Ullenius D.A., Pereira C.J., “Silica Deactivation of bead VOC catalysts.”, Applied Catalysis B: Environmental., 43(1998), 21. Liguori, F., Barbaro, P., Giordano, C., & Sawa, H. (2013). Partial hydrogenation reactions over Pd-containing hybrid inorganic/polymeric catalytic membranes. Applied Catalysis A: General. Lüdtke K., Peinemann K.V., Kasche V., and Behling R.D., “Nitrate removal of drinking water by means of catalytically active membranes”, Journal of Membrane Science, 151 (1998), 3–11. Ma, Y. H., Moser, W. R., Dixon, A. G., Ramachandra, A. M., Lu, Y., Binkerd, C., “Oxidative coupling of methane using inorganic membrane reactors”, (No. DE--AC22-92PC92113--22). Federal Energy Technology Center, Morgantown, WV (US); Federal Energy Technology Center, Pittsburgh, PA (US), 1998. Ma, X., Sun, Q., Feng, X., He, X., Guo, J., Sun, H., & Cao, H. (2012). Catalytic oxidation of 1, 2-dichlorobenzene over CaCO< sub> 3</sub>/α-Fe< sub> 2</sub> O< sub> 3</sub> nanocomposite catalysts. Applied Catalysis A: General. Maira A.J., Lau W.N., Lee C.Y., Yue P.L., Chan C.K., Yeung K.L., “Performance of a membrane-catalyst for photocatalytic oxidation of volatile organic compounds”, Chem.Eng.Sci., 58(2003), 959-962. Martin P., Ellersdorfer E., Zeman A., “Stadtentwaesserungswerke 206 Muenchen. Volatile siloxanes in wastewater and biogas and their effects on combustion engines.”, Korrespondenz Abwasser., 43(1996), 1574. Matsui T., Imamura S., “Removal of Siloxane from Digestion Gas of Sewage Sludge.”, Bioresour. Technol., 101(2010), S29. McBean, E.A., “Siloxanes in Biogases from Landfills and Wastewater Digesters.”, Can. J. Civ. Eng., 35(2008), 431. Minyukova, T. P., Shtertser, N. V., Khassin, A. A., & Yurieva, T. M. (2012). Permeable composite membrane as a catalytically active contactor for hydrogenation reactions. Catalysis Today. Montanari T., Finocchio, E., Bozzano I., Garuti G., Giordano A., Pistarino, C., Busca G., “Purification of Landfill Biogases from Siloxanes by Adsorption: A Study of Silica and 13X Zeolite Adsorbents on Hexamethylcyclotrisiloxane Separation.”, Chem. Eng. J., 165(2010), 859. Motamedhashemi Yousef M.M., Egolfopoulos F., Tsotsis T.T., “Application of a flow-through catalytic membrane reactor (FTCMR) for the destruction of a chemical warfare simulant ”, Journal of Membrane Science, 376(2011), 119-131. Motamedhashemi Yousef M.M., “A flow-through catalytic membrane reactor for destruction of a chemical warfare simulant ”, A Phd Thesis (2012). Mucherie S., Kim H.S., Marshall C.L., Libera J.A., Elam J.W., Stair P.C., “Oxidative dehydrogenation of propane over nanostructured membrane catalyst”, Proceedings of North American Catalysis Society (20 NAM), 2007. 207 Murphy J.D., McKeogh E., “The Benefits of Integrated Treatment of Wastes for the Production of Energy.”, Energy., 31(2006), 294. Musialik-Piotrowska, A., & Syczewska, K. (2002). Catalytic oxidation of trichloroethylene in two-component mixtures with selected volatile organic compounds. Catalysis today, 73(3), 333-342. Nagata, H., Takakura, T., Tashiro, S., Kishida, M., Mizuno, K., Tamori, I., & Wakabayashi, K. (1994). Catalytic oxidative decomposition of chlorofluorocarbons (CFCs) in the presence of hydrocarbons. Applied Catalysis B: Environmental, 5(1), 23-31. Nair N.N., Zhang X., Gutierrez J., Chen J., Egolfopoulos F., Tsotsis T., “On the Impact of Siloxane Impurities on the Performance of an Engine Operating on Renewable Natural Gas.”, Ind. Eng. Chem. Res., 51(2012), 15786. Nair, Nitin, Arjun Vas, Tongyu Zhao, Wenjing Sun, Jorge Gutierrez, Jack Chen, Fokion N. Egolfopoulos, and Theodore T. Tsotsis. "Effect of Siloxane Contained in Natural Gas on the Operation of a Residential Furnace." Industrial & Engineering Chemistry Research (2013). Napierska D., Thomassen L.C., Lison D., Martens J.A., Hoet, P.H., “The nanosilica hazard: another variable entity.”, Part. Fibre. Toxicol., 7(2010), 39. Ning, P., Wang, X., Bart, H. J., Liu, T., Huang, J., Wang, Y., & Gao, H. (2011). Catalytic Decomposition of CFC-12 over Solid Superacid Mo 2 O 3/ZrO 2. Journal of Environmental Engineering, 137(10), 897-902. Ohannessian, A., Desjardin, V., Chatain, V., Germain, P., “Volatile 208 Organic Silicon Compounds: The Most Undesirable Contaminants in Biogases.”, Water Sci. Technol., 58(2008), 1775. Oshita K., Ishihara Y., Takaoka M., Takeda N., Matsumoto T., Morisawa S., Kitayama A., “Behavior and Adsorptive Removal of Siloxanes in Sewage Sludge Biogas.”, Water Sci. Technol., 61(2010), 2003-2012. Ouchi A., Koga A., Bastl Z., Pola J., “UV-laser induced photolysis of trimethyl (vinyloxy) silane for chemical vapour deposition of polysiloxane films” , Appl. Organoment. Chem., 13 (1999), 643. Ouyang, M., Yuan, C., Muisener, R. J., Boulares, A., & Koberstein, J. T. (2000). Conversion of some siloxane polymers to silicon oxide by UV/ozone photochemical processes. Chemistry of materials, 12(6), 1591-1596. Pacey J.G., Doorn M., Thorneloe S. 1994, "Landfill gas energy utilization: technical and non-technical considerations", SWANA 17th Annual Landfill Gas Symposium Proceedings, GR-LG 0017, Solid Waste Association of North America, Long Beach, 237-249. Pellin M., Stair P., Xiong G., Elam J., Birrell J., Curtiss L., George S., Han C., Iton L., Kung H., Kung M., Wang H., “Mesoporous catalytic membranes: Synthetic control of pore size and wall composition, Catal. Lett., 102(2005), 127- 130. Picasso, G., Quintilla, A., Pina, M. P., & Herguido, J. (2003). Total combustion of methyl-ethyl ketone over Fe< sub> 2</sub> O< sub> 3</sub> 209 based catalytic membrane reactors. Applied Catalysis B: Environmental, 46(1), 133-143. Picasso, G., Gutierrez, M., Pina, M. P., & Herguido, J. (2007). Preparation and characterization of Ce-Zr and Ce-Mn based oxides for< i> n</i>-hexane combustion: Application to catalytic membrane reactors. Chemical Engineering Journal, 126(2), 119-130. Pina M., Menendez M., Santamaria J., “The Knudsen-diffusion catalytic membrane reactor: an efficient contactor for the combustion of volatile organic compounds”, Journal of Applied Catalysis, B11 (1996), L19-L27. Pina, M. P., Irusta, S., Menéndez, M., Santamaria, J., Hughes, R., & Boag, N. (1997). Combustion of volatile organic compounds over platinum-based catalytic membranes. Industrial & engineering chemistry research, 36(11), 4557- 4566. Pola J., Galikova A., Galik A., Blechta V., Bastl Z., Subrt J., Ouchi A., “UV Laser Photolysis of Disiloxanes for Chemical Vapor Deposition of Nano- Textured Silicones”, Chem. Mater, 14 (2002), 144. Poling, B. E., Prausnitz, J. M., & O’Connell, J. P. The Properties of Gases and Liquids, 2000. There is no corresponding record for this reference, 656-731. Popat S.C., Deshusses M.A., “Biological Removal of Siloxanes from Landfill and Digester Gases: Opportunities and Challenges.”, Environ. Sci. Technol., 42(2008), 8510. Pope, D., Walker, D. S., & Moss, R. L. (1976). Evaluation of cobalt oxide 210 catalysts for the oxidation of low concentrations of organic compounds in air. Atmospheric Environment (1967), 10(11), 951-956. Pradhan B.K., Sandle, N.K., “Effect of different oxidizing agent treatments on the surface properties of activated carbons”, Carbon, 37 (1999), 1323-1332. Prosser R.W., Ren J.Y., Egolfopoulos F.N., Tsotsis T.T., “UV Photodecomposition of Siloxane.”, CEC EISG Final Report. (Grant # 06-25). (2010). Pina M., Menendez M., Santamaria J., “The Knudsen-diffusion catalytic membrane reactor: an efficient contactor for the combustion of volatile organic compounds”, Journal of Applied Catalysis, B11 (1996), L19-L27. Purnama H., Kurr P., Schmidt A., Schomäcker R., Voigt I., Wolf A., and Warsitz R., [alpha]-methylstyrene hydrogenation in a flow-through membrane reactor”, AIChE Journal, 52 (2006), 2805–2811. Rautenbach, R., & Welsch, K. (1994). Treatment of landfill gas by gas permeation—pilot plant results and comparison to alternatives. Journal of membrane science, 87(1), 107-118. Reif M. and Dittmeyer R., “Porous, catalytically active ceramic membranes for gas-liquid reactions: A comparison between catalytic diffuser and forced through flow concept”, Catalysis Today, 82 (2003), 3–14. Reid, R. C., Prausnitz, J. M., & Poling, B. E. (1987). The properties of liquids and gases. McGraw-Hill, New York. 211 Ricaurte Ortega D., Subrenat A., “Siloxane Treatment by Adsorption into Porous Materials.”, Environ. Technol., 30 (2009), 1073. Sanchez, J. G., & Tsotsis, T. T. (2001). Catalytic membranes and catalytic membrane reactors. J. Wiley. Saracco G., Specchia V., “Catalytic ceramic filters for flue gas cleaning. 2: Catalytic performance and modeling thereof”, Journal of Industrial Engineering and Chemistry Research, 34(1995), 1480-1487. Schmidt A., Haidar R., and Reinhard Schomäcker R., “Selectivity of partial hydrogenation reactions performed in a pore-through-flow catalytic membrane reactor”, Catalysis Today, 104 (2005) 305–312. Schmidt A. and Schomäcker R., “Partial hydrogenation of sunflower oil in a membrane reactor”, Journal of Molecular Catalysis A: Chemical, 271 (2007), 192–199. Schmidt A., Wolf A., Warsitz R., Dittmeyer R., Urbanczyk D., Voigt I., Fischer G., and Schomäcker R., ”A pore-flow-through membrane reactor for partial hydrogenation of 1,5-cyclooctadiene. AIChE Journal, 54 (2008), 258–268. Schuessler M., Lamla O., Stefanovski T., Klein C., Megede zur D., “Autothermal reforming of methanol in an isothermal reactor-concept and realization”, Chem.Eng. Technol. 24(2001), 1141-1145. Schweigkofler M., Niessner R., “Removal of siloxanes in biogases”, Journal of Hazardous Materials, 83 (2001), 183-196. 212 SCS Energy. 2005, Siloxane quantification, removal and impact on landfill gas utilization facilities. 8 th annual LMOP conference and project expo, Maryland. Shin, H.C., Park, J.W., Park, K., Song, H.C., “Removal Characteristics of Trace Compounds of Landfill Gas by Activated Carbon Adsorption.”, Environ. Pollut., 119(2002), 227. Soltani S., Sahimi M., Tsotsis T.T., “Catalytic Membrane Reactors: A Brief Overview 2013”, Encyclopedia of Membrane Science and Technology, Wiley-VCH, 2013. Splinter A., Strumann J., Bartels O., Benecke W., “Micro membrane reactor: a flow-through membrane for gas pre-combustion”, Sens. Actuators, B83(2002), 169-174. Stair P., Marshall C., Xiong G., Feng H., Pellin M., Elam J., Curtiss L., Iton L., Kung H., Kung M., and Wang H.H., “Novel, uniform nanostructured catalytic membranes”, Topics in Catalysis, 39 (2006), 181–186. Tajima, M., Niwa, M., Fujii, Y., Koinuma, Y., Aizawa, R., Kushiyama, S., ... & Ohuˆchi, H. (1996). Decomposition of chlorofluorocarbons in the presence of water over zeolite catalyst. Applied Catalysis B: Environmental, 9(1), 167-177. Thorneloe S.A. 2003, U.S. EPA’s Field test programs to update data on landfill gas emissions, 9 th International Waste Management and Landfill Conference, Sardinia, Italy. 213 Toledo, J. M., Corella, J., & Sanz, A. (2001). Noble metal‐ based catalysts for total oxidation of chlorinated hydrocarbons. Environmental progress, 20(3), 167-174. Torres M., Lopez L., Dominguez J.M., Mantilla A., Ferrat G., Gutierrez M., Maubert M., “Olefins catalytic oligomerization on new composites of beta- zeolite films supported on [alpha]- Al 2 O membranes”, Chem.Eng.J., 92(2003), 1- 6. Tsuru T., Kan-no T., Yoshioka T., Asaeda M., “A photocatalytic membrane reactor for gas-phase rreactions using porous titanium oxide membranes”, Catal. Today, 82(2003), 41-48. Urban W., Lohmann H., Salazar Gomez J.I., “Catalytically upgraded landfill gas as a cost-effective alternative for fuel cells”, Journal of Power Sources, 193(2009), 359-366. Urbanova M., Bastl Z., Subrt J. and Pola J, “IR laser-induced thermolysis and UV laser-induced photolysis of 1,3-diethyldisiloxane: chemical vapor deposition of nanotextured hydridoalkylsilicones.”, Journal of Materials Chemistry, 11 (2001), 1557. van den Brink, R. W., Louw, R., & Mulder, P. (1998). Formation of polychlorinated benzenes during the catalytic combustion of chlorobenzene using a Pt/< i> γ</i>-Al< sub> 2</sub> O< sub> 3</sub> catalyst. Applied Catalysis B: Environmental, 16(3), 219-226. 214 Vincent M.J. and R. D. Gonzalez R.D., “Selective hydrogenation of acetylene through a short contact time reactor”, AIChE Journal, 48 (2002), 1257– 1267. Westermann T., Melin T., “Flow-through catalytic membrane reactors- principles and applications”, Journal of Chemical Engineering and Processing: Process Intensification, 48(2009), 17-28. White, G.C. 1992, Handbook of Chlorination and Alternative Disinfectants. Van Nostrand Reinhold, New York, NY. Wheless, E.; Gary, D. Siloxanes in Landfill and Digester Gas. Twenty- Fifth Annual SWANA Landfill Gas Symposium, Monterey, CA. (2002). Wheless E., Pierce J., 2004, Siloxane in landfill and digester gas update. http://mercury.scseng.com/Papers/Pierce_2004Siloxanes_Update_Paper.pdf Yamada M., Fugii K., Haru H., Itabashi K., “Preparation and catalytic properties of special alumina membrane formed by anodic oxidation of aluminum “, Tech.Rep., The Light Metal Educational Foundation, Inc. Research Group for Functionalizing of Aluminum and its Surface Films, 1988, 175-182. Yingling Y.G., Garrison B.J., “Photochemical ablation of organic solids.” Nuclear Instruments and Methods in Physics Research B, 202 (2003), 188. Yim, S. D., Koh, D. J., & Nam, I. S. (2002). A pilot plant study for catalytic decomposition of PCDDs/PCDFs over supported chromium oxide catalysts. Catalysis today, 75(1), 269-276. 215 Zalamea S., Pina M., Villellas A., Menendez M., Santamaria J., “Combustion of volatile organic compounds over mixed-regime catalytic membranes”, React.Kinet.Catal.Lett., 67(1999), 13-19. Zaspalis V.T., Praag van W., Keizer K., Van Ommen J.G., Ross J.R.H., Burggraaf A.J., “Reaction of methanol over catalytically active alumina membranes”, Appl. Catal., 74(1991), 205-222. Zhu B., Li H., and Yang W., “AgBiVMo oxide catalytic membrane for selective oxidation of propane to acrolein”, Catalysis Today, 82 (2003), 91–98.
Abstract (if available)
Abstract
In this study a novel catalytic oxidation technology appropriate for landfill gas (LFG) clean-up based on the flow-through catalytic membrane reactor (FTCMR) concept has been studied. For the experiments, a model LFG stream has been used with a volatile organic compound (VOC) composition which was shown previously by the authors to simulate well the behavior of real LFG in field-scale investigations. Asymmetric tubular alumina membranes were used in the research and were rendered catalytic by wet impregnation. Their pore-structure characteristics were measured with single-gas permeation tests, as they are important in determining the transport mechanisms of the VOC through the catalytically active membrane layer. When comparing the FTCMR with the more conventional reactors the ""yardstick"" of success is the ability of the FTCMR to operate under lower temperature for a given level of conversion, and/or attain higher conversion under the same conditions and catalyst loading. For the LFG application, light-off temperature experiments showed promising results when compared to the monolith reactor. Also, no catalyst deactivation was observed during the time-on-stream experiments, proving that the FTCMR is robust towards corrosive by-products (e.g., HCl) produced during the oxidation reactions. ❧ Siloxanes are another major class of compounds detected in LFG. This Thesis is also a study on the impact of siloxanes on various types of equipment using LFG which is not treated for siloxanes. Specifically, in this study an internal combustion engine and a residential furnace operating on natural gas (NG) spiked with siloxanes have been studied experimentally with the goal of understanding the impact of siloxane impurities on their performance. These impurities are shown to completely decompose during NG combustion in the engine to form silica microparticulates. These coat the internal metal surfaces in the equipment and severely reduce their efficiency and damage important components, such as furnace flame sensors and engine oxygen sensors. A method to remove these siloxane impurities has also been studied in this thesis based on UV Photodecomposition. Specifically, this Thesis also describes efforts to evaluate the technical feasibility and environmental implications of a novel technology for the treatment of biogas and LFG which involves the in situ conversion of the siloxanes, typically found in the gas, into inert silicon dioxide via a photochemical conversion process. The approach involves using high energy UV light to convert the siloxanes into SiO₂ powder, which can be conveniently removed from the biogas via a downstream filter. The technique is shown to be very effective with high siloxane conversions attained in the laboratory.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Lab-scale and field-scale study of siloxane contaminants removal from landfill gas
PDF
A flow-through membrane reactor for destruction of a chemical warfare simulant
PDF
The use of carbon molecule sieve and Pd membranes for conventional and reactive applications
PDF
Studies of siloxane decomposition in biomethane combustion
PDF
Biogas reforming: conventional and reactive separation processes and the preparation and characterization of related materials
PDF
The purification of contaminated air streams via recative separation techniques
PDF
A high efficiency, ultra-compact process for pre-combustion CO₂ capture
PDF
Performance prediction, state estimation and production optimization of a landfill
PDF
On the use of membrane reactors in biomass utilization
PDF
Methanol synthesis in the membrane reactor
PDF
Mesoscale SOFC-based power generator system: modeling and experiments
PDF
Fabrication of nanoporous silicon carbide membranes for gas separation applications
PDF
Methanol synthesis in a membrane reactor
PDF
Process intensification in hydrogen production via membrane-based reactive separations
PDF
An integrated 'one-box' process for hydrogen production
PDF
Catalytic methane ignition over freely-suspended palladium nanoparticles
PDF
Methanol synthesis in a contactor-type membrane reactor
PDF
A study of the application of membrane-based reactive separation to the carbon dioxide methanation
PDF
The study of mass transfer in gas shales and the optimization of hydraulic stimulation processes via additives
PDF
A hybrid adsorbent-membrane reactor (HAMR) system for hydrogen production
Asset Metadata
Creator
Nair, Nitin Narayanan
(author)
Core Title
Novel methods for landfill gas and biogas clean-up
School
Viterbi School of Engineering
Degree
Doctor of Philosophy
Degree Program
Chemical Engineering
Publication Date
07/09/2013
Defense Date
06/19/2013
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
biogas,landfill gas,membrane reactor,methane,NMOC,OAI-PMH Harvest,siloxanes,VOC
Format
application/pdf
(imt)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Tsotsis, Theodore T. (
committee chair
)
Creator Email
nairnitinn@gmail.com,nnair@usc.edu
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c3-285994
Unique identifier
UC11294452
Identifier
etd-NairNitinN-1732.pdf (filename),usctheses-c3-285994 (legacy record id)
Legacy Identifier
etd-NairNitinN-1732.pdf
Dmrecord
285994
Document Type
Dissertation
Format
application/pdf (imt)
Rights
Nair, Nitin Narayanan
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
biogas
landfill gas
membrane reactor
methane
NMOC
siloxanes
VOC