A Thesis in Materials Science and Engineering. Nur Farhana Diyana MOHD YUNOS

Transcription

1 The University of New South Wales Faculty of Science School of Materials Science and Engineering High Temperature Phenomena Occurring during Reactions of Agricultural Wastes in Electric Arc Furnace Steelmaking: Interactions with Gas and Slag Phases A Thesis in Materials Science and Engineering by Nur Farhana Diyana MOHD YUNOS Submitted in Partial Fulfillment of the requirements for the Degree of DOCTOR OF PHILOSOPHY March 2012

2 PLEASE TYPE THE UNIVERSITY OF NEW SOUTH WALES Thesis/Dissertation Sheet Surname or Family name: BINTI MOHD YUNOS First name: NUR FARHANA DIYANA Other name/s: Abbreviation for degree as given in the University calendar: Ph.D School: MATERIALS SCIENCE AND ENGINEERING Faculty: SCIENCE Title: HIGH TEMPERATURE PHENOMENA OCCURING DURING REACTIONS OF AGRICULTURAL WASTES IN ELECTRIC ARC FURNACE STEELMAKING: INTERACTIONS WITH GAS AND SLAG PHASES Abstract 350 words maximum: (PLEASE TYPE) Iron and steel making is an energy intensive industrial sector using mainly coal as the heat source and reduction agent. The industry gives rise to about 10 % of the anthropogenic CO 2 emissions in the world. Due to the challenge for CO 2 mitigation, interest for agricultural waste (palm and coconut shells) use as a renewable energy and carbon source as heating agent and reducing agent contributes to energy conservation and emission reduction, and can partially replace coal and coke. In the present study, the conventional material investigated was metallurgical coke which was blended with different proportions of palm and coconut shells as well as agricultural waste chars in order to reduce the waste in the landfill. Metallurgical coke, palm shell/coke blends and coconut shell/coke blends were combusted in a drop tube furnace (DTF) at 1200 C under 20% O 2 and 80% N 2 gas mixture while palm char was devolatilized at 450 C under N 2 atmosphere. Subsequently, the residual materials were put in contact with an EAF iron oxide rich slags and their interfacial reactions and phenomena have been studied at 1550 C in a horizontal tube furnace under inert atmosphere (1 L/min Ar) with off gases measured using an IR analyser. The initial devolatilization and the subsequent step of combustion of these samples are conducted in a Drop Tube Furnace (DTF) and in a Thermogravimetric Analyser (TGA), respectively, while the sessile drop approach was used to investigate the interfacial reactions taking place in the slag/carbon region. A Thermogravimetric Analyser coupled with Mass Spectrometer (TGA-MS) was also used to study the behavior of coke and agricultural wastes at high temperatures in order to understand the thermal behavior and gas products that evolved at high temperatures. The weight loss profiles, gas formation and products distribution were significantly different between the coke and agricultural waste samples. It was found that more gases were released from agricultural waste than from coke that participated in the subsequent carbon/slag reactions. In the gas phase reaction studies, the blends containing agricultural waste materials indicated higher combustion efficiencies compared to coke alone with an improved surface area resulted from volatile matter removal. The role of chemical structure and properties, as well as inorganic matter in agricultural waste blends also influenced the combustion performance. The rate of devolatilization appears to improve the coke/palm shell blends burnout as well as its foaming behavior when put in contact with an iron oxide rich slag. For carbon/slag interactions, experiments were conducted using the sessile drop technique (1550 C) with off gases (CO, CO 2) measured using an IR analyzer; the wetting behaviour was determined from contact angle measurements and estimation of slag foam volumes were calculated using specialized software. Off gas analyses following the carbon/slag interfacial reactions have been measured for all the carbonaceous materials and significantly different gas concentrations have been observed. The rates of total gas generation (CO+CO 2) from palm char was comparable to those seen in coke; however the gases released from palm chars were extent over a longer period of time and allowed their entrapment in the slag matrix, enhancing the volume of the slag. A slower rate of FeO reduction is seen when coke reacted with the Electric Arc Furnace (EAF) slag, while the palm shell blends showed a faster reduction. Independent of the carbon material used as a substrate, the final stage of reaction reveals comparable contact angles due to similar extents of reduction and Fe deposition at the interface. The steady gas generation seen in palm char compared to coke allows the formation of a highly porous particle, promoting gasification and allowing more gases to be trapped in the slag phase. These results indicate that partial replacement of coke with palm shells is not only viable, but efficient leading to improved/sustained interactions with EAF slag. Optimization between the two phenomena, reduction and foaming is required for improved EAF process performance. Declaration relating to disposition of project thesis/dissertation I hereby grant to the University of New South Wales or its agents the right to archive and to make available my thesis or dissertation in whole or in part in the University libraries in all forms of media, now or here after known, subject to the provisions of the Copyright Act I retain all property rights, such as patent rights. I also retain the right to use in future works (such as articles or books) all or part of this thesis or dissertation. I also authorise University Microfilms to use the 350 word abstract of my thesis in Dissertation Abstracts International (this is applicable to doctoral theses only). Signature.. Witness..... Date The University recognises that there may be exceptional circumstances requiring restrictions on copying or conditions on use. Requests for restriction for a period of up to 2 years must be made in writing. Requests for a longer period of restriction may be considered in exceptional circumstances and require the approval of the Dean of Graduate Research. FOR OFFICE USE ONLY Date of completion of requirements for Award: THIS SHEET IS TO BE GLUED TO THE INSIDE FRONT COVER OF THE THESIS

3 ORIGINALITY STATEMENT I hereby declare that this submission is my own work and to the best of my knowledge it contains no materials previously published or written by another person, nor material which to a substantial extent has been accepted for the award of any other degree or diploma at UNSW or any other educational institution, except where due acknowledgement is made in the thesis. Any contribution made to the research by others, with whom I have worked at UNSW or elsewhere, is explicitly acknowledged in the thesis. I also declare that the intellectual content of this thesis is the product of my own work, except to the extent that assistance from others in the project s design and conception or in style, presentation and linguistic expression is acknowledged. Signed Nur Farhana Diyana Mohd Yunos ii

4 COPYRIGHT STATEMENT I hereby grant the University of New South Wales or its agents the right to archive and to make available my thesis or dissertation in whole or part in the University libraries in all forms of media, now or here after known, subject to the provisions of the Copyright Act I retain all proprietary rights, such as patent rights. I also retain the right to use in future works (such as articles or books) all or part of this thesis or dissertation. I also authorize University Microfilms to use the 350 word abstract of my thesis in Dissertation Abstract International (this is applicable to doctoral theses only). I have either used no substantial portions of copyright material in my thesis or I have obtained permission to use copyright material; where permission has not been granted I have applied/will apply for a partial restriction of the digital copy of my thesis or dissertation.' Signed... Date... AUTHENTICITY STATEMENT I certify that the Library deposit digital copy is a direct equivalent of the final officially approved version of my thesis. No emendation of content has occurred and if there are any minor variations in formatting, they are the result of the conversion to digital format. Signed... Date... iii

5 ACKNOWLEDGEMENTS During my doctoral research I have been accompanied and supported by many people. Now, I have the opportunity to express my gratitude to all of them. I would like to express my deepest sense of gratitude and appreciation to my supervisors Professor Veena Sahajwalla, Assoc Prof Rita Khanna and Dr Magdalena Zaharia for their unrelenting support and guidance throughout my research. My sincerest gratitude goes to them for their excellent guidance and for providing valuable advice, comments and suggestions at various points during the course of this research. Their expertise was also invaluable in interpreting the results and in providing suggestions on preparing papers and research documents. I also wish to acknowledge the financial support provided by the Ministry of Higher Education Malaysia and University of Malaysia Perlis. I am grateful to Mr Narendra Saha Chaudhury for his valuable assistance in setting up and maintaining the experimental apparatus. In addition, special thanks to staff and colleagues at the Centre for Sustainable Materials Research & Technology (SMaRT@UNSW), School of Materials Science and Engineering and Chemical Engineering for their help, comments and advices as well as for making my time enjoyable throughout the course of this project. Finally, I express my deepest gratitude to my husband, Muhammad Asri Idris and family for motivation, patience, sacrifices and forever love through the period. I also thank my lovely daughter Asfa Haziqa, who is my beautiful gift from God, for bearing with me during the course of my Ph.D. iv

6 ABSTRACT Iron and steel making is an energy intensive industrial sector using mainly coal as the heat source and reduction agent. The industry gives rise to about 5 % of the anthropogenic CO 2 emissions in the world. Due to the challenge for CO 2 mitigation, interest for agricultural waste (palm and coconut shells) use as a renewable energy and carbon source as heating agent and reducing agent contributes to energy conservation and emission reduction, and can partially replace coal and coke. In the present study, the conventional material investigated was metallurgical coke which was blended with different proportions of palm and coconut shells as well as agricultural waste chars in order to reduce the waste in the landfill. Metallurgical coke, palm shell/coke blends and coconut shell/coke blends were combusted in a drop tube furnace (DTF) at 1200 C under 20% O 2 and 80% N 2 gas mixture while palm char was devolatilized at 450 C under N 2 atmosphere. Subsequently, the residual materials were put in contact with an EAF iron oxide rich slags and their interfacial reactions and phenomena have been studied at 1550 C in a horizontal tube furnace under inert atmosphere (l L/min Ar) with off gases measured using an IR analyser. The initial devolatilization and the subsequent step of combustion of these samples are conducted in a Drop Tube Furnace (DTF) and in a Thermogravimetric Analyser (TGA), respectively, while the sessile drop approach was used to investigate the interfacial reactions taking place in the slag/carbon region. A Thermogravimetric Analyser coupled with Mass Spectrometer (TGA-MS) was also used to study the behavior of coke and agricultural wastes at high temperatures in order to understand the thermal behavior and gas products that evolved at high temperatures. The weight loss profiles, gas formation and products distribution were significantly different between the coke and agricultural waste samples. It was found that more gases were released from agricultural waste than from coke that participated in the subsequent carbon/slag reactions. In the gas phase reaction studies, the blends containing agricultural waste materials indicated higher combustion efficiencies compared to coke alone with an improved v

7 surface area resulted from volatile matter removal. The role of chemical structure and properties, as well as inorganic matter in agricultural waste blends also influenced the combustion performance. The rate of devolatilization appears to improve the coke/palm shell blends burnout as well as its foaming behavior when put in contact with an iron oxide rich slag. For carbon/slag interactions, experiments were conducted using the sessile drop technique (1550 C) with off gases (CO, CO 2 ) measured using an IR analyzer; the wetting behaviour was determined from contact angle measurements and estimation of slag foam volumes were calculated using specialized software. Off gas analyses following the carbon/slag interfacial reactions have been measured for all the carbonaceous materials and significantly different gas concentrations have been observed. The rates of total gas generation (CO+CO 2 ) from palm char was comparable to those seen in coke; however the gases released from palm char was extent over a longer period of time and allowed their entrapment in the slag matrix, enhancing the volume of the slag. A slower rate of FeO reduction is seen when coke reacted with the Electric Arc Furnace (EAF) slag, while the palm shell blends showed a faster reduction. Independent of the carbon material used as a substrate, the final stage of reaction reveals comparable contact angles due to similar extents of reduction and Fe deposition at the interface. The steady gas generation seen in palm char compared to coke allows the formation of a highly porous particle, promoting gasification and allowing more gases to be trapped in the slag phase. These results indicate that partial replacement of coke with palm shells is not only viable, but efficient leading to improved/sustained interactions with EAF slag. Optimization between the two phenomena, reduction and foaming is required for improved EAF process performance. vi

8 Table of Contents ORIGINALITY STATEMENT... ii COPYRIGHT STATEMENT... iii AUTHENTICITY STATEMENT... iii ACKNOWLEDGEMENTS... iv ABSTRACT... v Table of Contents... vii List of Figures... xiv List of Tables... xxiv List of Publications... xxviii CHAPTER Introduction Background Research Objectives... 5 CHAPTER Literature Review Agricultural Waste as a Source of Renewable Materials Environmental Impact Recycling of Agricultural Wastes in Steelmaking Properties and Composition of Agricultural Waste and Metallurgical Coke Cellulose vii

9 2.4.2 Hemicellulose Lignin Inorganic Minerals Metallurgical Coke Activated Carbon Gas Phase Reactions Pyrolysis Heterogeneous Char Combustion Kinetic Reactions Regimes Inorganic Effects in Combustion Effects Due to the Inorganic Ash Compounds Determination of the Combustion Efficiency of Coke and its Blends with Agricultural Waste Factors Affecting Gas Phase Reactions High Temperature Carbon/Slag Interactions Experimental Techniques for Carbon/Slag Interactions Previous Studies on Carbon/Slag Interaction Factors Affecting Carbon/Slag Interactions FeO Reduction in EAF Steelmaking Slags EAF Operating Conditions with Slags Previous Studies on FeO Reduction and Factors Affecting FeO Reduction by Solid Carbon viii

10 2.7 Summary CHAPTER Experimental Characterization of Experimental Materials Gas Phase Reactions Studies Preparation of Carbonaceous Materials Chemical Characterization of Specimens Thermogravimetric Analysis with Mass Spectroscopy (TGA-MS) X-Ray Diffraction Analysis (XRD) Fourier Transform Infrared Spectroscopy (FTIR) Nuclear Magnetic Resonance Spectroscopy ( 13 C NMR) Physical Characterization of Specimens Surface Area Measurements Scanning Electron Microscopy (SEM) Carbon/slag Interaction Studies Sample Preparation Interfacial Phenomena-Optical Microscopy and SEM Interfacial Phenomena Wetting Behaviour Experimental Apparatus Drop Tube Furnace (DTF) Gas Phase Reaction Studies LECO Carbon Analyser ix

11 3.5.3 Muffle Furnace Ash Measurement Thermogravimetric Analyzer (TGA) Ash Tracer Method Combustion Efficiency Inorganic Tracer Method The mass burnout of coke/agricultural waste blends Minimization of Error in Combustion Efficiency Determination Horizontal Tube Furnace Carbon/slag Interactions Infra-Red Analyzer (IR) Off-Gas Generation X-Ray Fluoroscopy (XRF) Slag Characterization Reproducibility of Carbon/slag Interaction Experiments Minimization of Error in Contact Angle Determination Estimation of Error in Slag Foaming Behavior (V t /V 0 ) Measurements.. 93 CHAPTER Combustion & Structural Transformations of Coke/ Palm Shell Blends: Results & Discussions Gas Phase Reactions of Metallurgical Coke and its Blends with Palm Shell Effect of High Temperature on the Behaviour of the Carbonaceous Material Effect of Blending on Combustion Physical Properties and Structural Transformations Surface Area Measurements BET Surface Area x

12 Structural Transformations - Scanning Electron Microscopy (SEM) The Role of Chemical Properties and Carbon Structures Chemical Structures Chemical Properties X-ray Diffractions Chemical Bonding Fourier Transform Infrared Spectroscopy (FTIR) Carbon Structures Nuclear Magnetic Resonance ( 13 C NMR Spectroscopy) Inorganic Minerals Summary CHAPTER Combustion & Structural Transformations of Coke/ Coconut Shell Blends: Results & Discussion Gas Phase Reactions of Metallurgical Coke and its Blends with Coconut Shell Effect of High Temperature on the Behaviour of the Carbonaceous Material Effect on Blending on Combustion Physical Properties and Structural Transformations Surface Area Measurement BET Surface Area Structural Transformations - Scanning Electron Microscopy (SEM) xi

13 5.1.4 The Role of Chemical Properties and Carbon Structures Chemical Structures Chemical Properties X-ray Diffractions Chemical Bonding Fourier Transform Infrared Spectroscopy (FTIR) Carbon Structures Nuclear Magnetic Resonance ( 13 C NMR Spectroscopy) Inorganic Minerals Comparison of Coke/Palm Shells & Coke/Coconut Shells Blends in Gas Phase Reactions Summary CHAPTER Conclusions Gas Phase Reactions Studies CHAPTER High Temperature Reactions: Carbon/Slag Interactions Results and Discussion Influence of Carbonaceous Material on Carbon/Slag Interactions Off-gas (CO, CO 2 ) Generations Contact Angle Measurements High Temperature, In-situ Observations Slag Foaming Interfacial Phenomena - Optical and SEM Studies xii

14 7.2 FeO Reduction Estimation of the reaction rate constant Discussion on Slag Foaming of Carbonaceous Material Influence of Gas Generation on Slag Foaming Influence of Volatiles from the Agricultural Wastes/Polymers on Slag Foaming Influence of Mineral Matter from the Agricultural Wastes on Slag Foaming Influence of Carbon Structures from the Agricultural Wastes on Slag Foaming Influence of Entrapped Gas Bubbles in the Bulk Slag on Slag Foaming CHAPTER Conclusions Carbon/Slag Interactions Studies CHAPTER Summary and Conclusions CHAPTER References APPENDICES APPENDIX A APPENDIX B xiii

15 List of Figures Figure 2-1 Molecular structure of common fatty acid in palm oil plant Figure 2-2 Polymers of cellulose (Graboski, M. and Bain, R., 1981) Figure 2-3 Monomers of xylan (Graboski, M. and Bain, R., 1981) Figure 2-4 A hypothetical structure of lignin (Blazej, A. and Kosik, M., 1993) Figure 2-5 Process during coal/agricultural wastes pyrolysis, gasification and combustion Figure 2-6 Reaction mechanism of pyrolysis and steam gasification (dotted line represents the reaction with steam): (a) cellulose; (b) lignin Figure 2-7 Proposed catalytic mechanism for the reaction of iron atoms in the char structure (Backreedy, R. I., Jones, J. M., Pourkashanian, M. et al., 2003) Figure 2-8 Proposed catalytic mechanism for the reaction of alkali metals (Huang, H. Y. and Yang, R. T., 1999) Figure 2-9 Representative example of the video screen during data processing. The coordinate for point's p1 and p2 are displayed in the data line just above the image, along with the computed values of volume and area of contact (arbitrary units) (Khanna, R., Spink, J. and Sahajwalla, V., 2007, Rahman, M., 2006) Figure 2-10 Schematic diagram of the novel processing software for volume and contact area measurements (Khanna, R., Mahfuzur, R., Richard, L. et al., 2007) Figure 2-11 The influence of S and P 2 O 5 on the surface tension of lime-silica slags containing 30% FeO at 1673K (Skupien, D. and Gaskell, D., 2000) Figure 2-12 The relationship between the foaming index and viscosity for CaO-SiO 2 - FeO system (Kozakevitch, P. and Olette, M., 1971) xiv

16 Figure 2-13 Foam index as a function of slag basicity (Ito, K. and Fruehan, R., 1989) Figure 3-1 Bar chart of the materials investigated in this study illustrating the relative concentrations of palm shells and MC content in the blends Figure 3-2 Bar chart of the materials investigated in this study illustrating the relative concentrations of coconut shells and MC content in the blends Figure 3-3 Bubble diameter measuring using Adobe Photoshop Figure 3-4 Illustration of the contact angle measurement using ANGLE software Figure 3-5 Schematic diagram of DTF (DTF is an abbreviation of Drop Tube Furnace) Figure 3-6 Front view of the Thermogravimetric Analyzer (TGA) with key features highlighted Figure 3-7 A schematic diagram of the change in mass of representative coke or agricultural waste particle during combustion Figure 3-8 Schematic Diagram of the Horizontal Tube Resistance Furnace used for Gas Entrapment Tests Figure 4-1 (a) (f) The mass loss of metallurgical coke with on-line MS analysis of the gas products during pyrolysis Figure 4-2 (a) (f) The mass loss of palm shells with on-line MS analysis of the gas products during pyrolysis Figure 4-3 (a) (f) The mass loss of palm char with on-line MS analysis of the gas products during pyrolysis Figure 4-4 Weight loss curves of MC and its blends with palm shells at temperature, 1200 ºC in N 2 atmosphere xv

17 Figure 4-5 Comparison of total weight loss curve of MC and its blends with palm shells at temperature, 1200 ºC in N 2 atmosphere Figure 4-6 Effect of blending MC with varying palm shell content in the blends on the combustion at 1200 C in the presence of 80% N 2 ; 20% O Figure 4-7 Effect of volatile matter on (a) blend composition and on (b) the combustion performance of palm shell/mc blends Figure 4-8 Changes in micropore surface area for palm shell before and after combustion in DTF as a function of the palm shell content in the blends Figure 4-9 The changes in micropore surface area for palm shell wastes content in the blend Figure 4-10 Images of (a) raw MC and (b) MC after combustion in gas phase at 1200 ºC (80% N 2 ; 20% O 2 ) Figure 4-11 Images of (a) raw palm shells and (b) palm shell after combustion in gas phase reactions at 1200 ºC (80% N 2 ; 20% O 2 ) Figure 4-12 SEM micrograph of (a) raw MC, (b) MC after combustion at 1200 C in DTF, (c) cross-sectional image of raw MC and (d) cross-section of MC char. 116 Figure 4-13 SEM micrograph of (a) raw palm shell, (b) palm shell after combustion at 1200 C in DTF, (c) cross-sectional image of raw palm shell and (d) crosssection of palm shell char Figure 4-14 Cellulose, hemicellulose and lignin contents of palm shell (Wan Daud, W. M. A. and Wan Ali, W. S., 2004) Figure 4-15 XRD patterns of metallurgical coke before and after combustions (T=1200 ºC with 20% O 2 ; 80% N 2 ) Figure 4-16 XRD patterns of palm shells before and after combustions (T=1200 ºC: 20% O 2 ; 80% N 2 ) xvi

18 Figure 4-17 X-ray diffraction analysis of coke/palm shell blends; P1 before and after combustion temperature 1200 ºC in 20% O 2 and 80% N 2 atmosphere Figure 4-18 X-ray diffraction analysis of coke/palm shell blends; P2 before and after combustion temperature 1200 ºC in 20% O 2 and 80% N 2 atmosphere Figure 4-19 X-ray diffraction analysis of coke/palm shell blends; P3 before and after combustion temperature 1200 ºC in 20% O 2 and 80% N 2 atmosphere Figure 4-20 FTIR spectra of metallurgical coke before and after combustion in DTF at temperature 1200 ºC in 20% O 2 and 80% N 2 atmosphere Figure 4-21 FTIR spectra of palm shells before and after combustion in DTF DTF at temperature 1200 ºC in 20% O 2 and 80% N 2 atmosphere Figure 4-22 CP/MAS 13 C NMR spectra of (a) raw MC and (b) MC at 1200ºC in a drop tube furnace [80% N 2, 20% O 2 ] (* spinning sideband) Figure 4-23 Spectrum of CP/MAS 13 C NMR of (a) raw palm shells and (b) palm shells char at 1200ºC in a drop tube furnace [80% N 2, 20% O 2 ] (* spinning sideband) Figure 4-24 Summarize of mechanism governing the breakdown of the structures before and after combustion in DTF at 1200 C (80% N 2 ; 20% O 2 ) for palm shells and MC Figure 4-25 Fe 2 O 3 component present in MC and MC /palm shell blends ash Figure 4-26 K 2 O and Na 2 O components present in the ash from MC and MC / palm shell blends Figure 5-1 (a) (f) The mass loss of coconut shells with on-line MS analysis of the gas products during pyrolysis Figure 5-2 (a) (f) The mass loss of coconut char with on-line MS analysis of the gas products during pyrolysis xvii

19 Figure 5-3 Weight loss curves of MC and its blends with coconut shells at temperature, 1200 ºC; N 2 atmosphere Figure 5-4 Comparison of total weight loss curve (50 seconds) of MC and its blends with coconut shells at temperature, 1200 ºC; N 2 atmosphere Figure 5-5 Effect of blending MC with varying coconut shell content in the blends on the combustion at 1200 C in the presence of 80% N 2 ; 20 % O Figure 5-6 Effect of volatile matter on (a) blend composition and on (b) the combustion performance of coconut shell/mc blends Figure 5-7 Changes in the micropore surface area before and after combustion in DTF as a function of the coconut shell content in the blends Figure 5-8 The changes in micropore surface area for coconut shell wastes content in the blend Figure 5-9 Images of (a) raw coconut shells and (b) coconut shell after combustion in gas phase reactions at 1200 ºC (80% N 2 ; 20% O 2 ) Figure 5-10 SEM micrographs of transverse section of (a) raw coconut shell, (b) coconut shell after combustion at 1200 C in DTF and cross-sectional images of (c) raw coconut shell, (d) coconut shell char (e) and (f) edge section of coconut shell char Figure 5-11 (a) and (b) SEM micrograph of coconut shell after combustion in DTF showing fragmented particles Figure 5-12 Cellulose, hemicellulose and lignin contents of coconut shell (Wan Daud, W. M. A. and Wan Ali, W. S., 2004) Figure 5-13 XRD patterns of coconut shells before and after combustions (T=1200 ºC: 20% O 2 ; 80% N 2 ) Figure 5-14 X-ray diffraction analysis of coke/coconut shell blends; C1 before and after combustion temperature 1200 ºC in 20% O 2 and 80% N 2 atmosphere xviii

20 Figure 5-15 X-ray diffraction analysis of coke/coconut shell blends; C2 before and after combustion temperature 1200 ºC in 20% O 2 and 80% N 2 atmosphere Figure 5-16 X-ray diffraction analysis of coke/coconut shell blends; C3 before and after combustion temperature 1200 ºC in 20% O 2 and 80% N 2 atmosphere Figure 5-17 FTIR spectra of coconut shell before and after combustion in DTF Figure 5-18 Spectrum of CP/MAS 13 C NMR of (a) raw coconut shell and (b) coconut shell char at 1200ºC in a drop tube furnace [80% N 2, 20% O 2 ] (* spinning sideband) Figure 5-19 Fe 2 O 3 component present in the ash from MC and MC /coconut shell blends Figure 5-20 K 2 O and Na 2 O components present in the ash from MC and MC - coconut shell blends Figure 5-21 (a) Comparison of residual mass (%) in TGA with time for 100% MC and its mixtures with palm shells (b) variation of rate constant of the palm shell blends with volatile matter at 1200ºC in N 2 atmosphere Figure 5-22 (a) Comparison of residual mass (%) in TGA with time for 100% MC and its mixtures with coconut shells (b) variation of rate constant of the coconut shell blends with volatile matter at 1200ºC in N 2 atmosphere Figure 5-23 (a) and (c) Comparison of residual mass (%) in TGA with time for 100% MC and its mixtures with palm and coconut shells (b) and (d) variation of rate constant of the agricultural blends with volatile matter at 1200ºC in N 2 atmosphere Figure 5-24 Combustion performances of palm shell-coke and coconut shell-coke blends (Blend 3* = Blend P3 and Blend C3) Figure 5-25 The changes in micropore surface area for agricultural wastes content in the blend xix

21 Figure 5-26 SEM micrographs of (a) MC, (b) palm shells and (c) coconut shells collected after reaction in the DTF at 1200 C in atmosphere of 20% O 2 and 80% N 2, polished section of the residual particle x Figure 7-1 Schematic diagram illustrating the effect of carbonaceous materials rate of devolatilization on slag foaming Figure 7-2 (I) Metallurgical coke/slag and (II) Palm char/slag with (a) generated gas concentrations (ppm) in terms of CO and CO 2 gases and (b) the cumulative volume of gases (mol) of CO and CO Figure 7-3 Total number of moles (a) carbon and (b) oxygen removed from the metallurgical coke (MC) and palm char (PC) substrate in contact with a slag at temperature 1550 ºC Figure 7-4 Generated gas concentrations, CO+CO 2 (ppm) as a function of time and carbon based materials (a) metallurgical coke, (b) P1 blends, (c) P2 blends and (d) P3 blends Figure 7-5 Total cumulative number of moles of gas generated (CO+CO 2 ) as a function of time and carbonaceous material used Figure 7-6 Variation of contact angle with time for carbonaceous substrate (a) palm char (PC), (b) P1 blends, (c) P2 blends and (d) P3 blends in contact with slag at 1550ºC Figure 7-7 High temperature photographs of slag droplets in contact with (a) 100% MC and (b) 100% Palm char at 1550 ºC as a function of time Figure 7-8 High temperature photographs of slag droplets in contact with (a) 100% MC and (c) P1 blends at 1550 ºC as a function of time Figure 7-9 High temperature photographs of slag droplets in contact with (a) 100% MC and (d) P2 blends at 1550 ºC as a function of time xx

22 Figure 7-10 High temperature photographs of slag droplets in contact with (a) 100% MC and (e) P3 blends at 1550 ºC as a function of time Figure 7-11 (a) Instantaneous gases (CO+CO 2 ), ppm, generated and (b) Volume ratio of 100% Metallurgical coke (MC) and 100% Palm char (PC) with slag with respect to reaction time Figure 7-12 (a) Instantaneous gases (CO+CO 2 ), ppm, generated and (b) Volume ratio of 100% Metallurgical coke (MC) and P1 blends with slag with respect to reaction time Figure 7-13 (a) Instantaneous gases (CO+CO 2 ), ppm, generated and (b) Volume ratio of 100% Metallurgical coke (MC) and P2 blends with slag with respect to reaction time Figure 7-14 (a) Instantaneous gases (CO+CO 2 ), ppm, generated and (b) Volume ratio of 100% Metallurgical coke (MC) and P3 blends with slag with respect to reaction time Figure 7-15 Instantaneous gases (CO+CO 2 ), ppm generated from 100% metallurgical coke (MC), 100% Palm char (PC), P1 blends, P2 blends and P3 blends reacting with slag as a function of time Figure 7-16 Optical microscopy images of (a) Slag/100% metallurgical coke (MC) and (b) Slag/Palm char (PC) as a function of time Figure 7-17 Optical microscopy images of (a) Slag/100% metallurgical coke (MC) and (c) Slag/P1 blends as a function of time Figure 7-18 Optical microscopy images of (a) Slag/100% metallurgical coke (MC) and (d) Slag/P2 blends as a function of time Figure 7-19 Optical microscopy images of (a) Slag/100% metallurgical coke (MC) and (e) Slag/P3 blends as a function of time xxi

23 Figure 7-20 (a) Optical microscopy of MC/slag, (b) SEM, mapping on the inner region of quenched MC/slag and (c) EDS spectra of quenched metallurgical coke/slag assembly at 1550 C after 15 min of contact Figure 7-21 (a) Optical microscopy of Palm char/slag, (b) SEM, mapping on the inner region of quenched Palm char/slag and (c) EDS spectra of quenched Palm char/slag assembly at 1550 C after 15 min of contact Figure 7-22 (a) Optical microscopy of P1 blend/slag, (b) SEM, mapping on the inner region of quenched P1 blend/slag and (c) EDS spectra of quenched P1 blend/slag assembly at 1550 C after 15 min of contact Figure 7-23 Total removed oxygen, moles as a function of time and carbonaceous material used Figure 7-24 FeO concentration in the slag with proceeding reaction Figure 7-25 Reaction rate constant as a function of carbon material used Figure 7-26 (a) Volume ratio of 100% Metallurgical coke (MC) and 100% Palm char (PC) and P3 blends with slag and (b) Gases (CO+CO 2 ), ppm, generated with respect to reaction time Figure 7-27 High temperature images of slag droplet in contact with (a) metallurgical coke (MC), (b) 100% palm char (PC) and P3 blends at 1550 ºC as a function of time Figure 7-28 High temperature images of slag droplet in contact with (a) rubber/coke blend (R3) (Zaharia, M., 2010) and (b) palm shell/coke blend (P3) at 1550 ºC as a function of time Figure 7-29 Reaction rate constant as a function of carbon material used Figure 7-30 Optical microscopy images of (a) Slag/HDPE blend (HDPE 3), (b) Slag/rubber blend (R3), (c) Slag/PET blend (PET 3) (Kongkarat, S., 2011, xxii

24 Rahman, M. M., 2010, Zaharia, M., 2010), (d) Slag/palm shell (P3) blend as a function of time and (e) Slag/100% palm char (PC) Figure 10-1 A schematic diagram of the change in mass of representative coke or agricultural waste particle during combustion xxiii

25 List of Tables Table 2-1 Cellulose, holocellulose and lignin contents of palm shell and coconut shell (Wan Daud, W. M. A. and Wan Ali, W. S., 2004) Table 2-2 Inorganic minerals in agricultural waste by XRF analysis literature review Table 2-3 Ultimate analysis of palm shell wastes material literature review Table 2-4 Proximate analysis of palm shell wastes material literature review Table 2-5 Ultimate analysis of coconut shell wastes material literature review Table 2-6 Proximate analysis of coconut shell wastes material literature review Table 2-7 Data for the SSM Trial Heats: Comparison between Metallurgical Coke and Rubber Tyre/Coke Mixture (Sahajwalla, Zaharia et al. 2009) Table 3-1 Summary of chemical analysis of palm shells, palm shells char (PC), metallurgical coke (MC) and palm shell blends prior to high temperature reactions Table 3-2 Summary of chemical analysis of coconut shell, coconut shells char (CC), metallurgical coke (MC) and coconut shell blends prior to high temperature reactions Table 3-3 Ash analysis for metallurgical coke, agricultural wastes and chars Table 3-4 Ash analysis for coke/agricultural waste blends Table 3-5 Experimental conditions for X-Ray diffraction studies Table 3-6 Experimental conditions for 13 C NMR spectroscopy Table 3-7 Summary of chemical analysis of metallurgical coke and agricultural wastes char after high temperature reactions xxiv

26 Table 3-8 Composition of slag Table 3-9 Operating conditions of DTF for combustion studies Table 3-10 Operating conditions of carbon and sulphur measurements Table 3-11 Experimental conditions for ashing studies Table 3-12 Operating condition of TGA for combustion and devolatilization studies under continuous conditions Table 3-13 The amounts of major inorganic elements (based on their oxides) present in the ash (XRF) of the coke and agricultural waste materials at 1200 C Table 3-14 The experimental conditions for the carbon/slag interactions Table 3-15 The experimental conditions for reduction studies Table 4-1 XRF analysis showing the amounts of major inorganic elements (based on their oxides) present in the ash of the coke, palm shell blends and palm shells following treatment at 1200 C under 20% oxygen and 80% nitrogen atsmosphere Table 4-2 XRD peaks characteristic of raw MC and MC after combustion in DTF 1200 ºC with atmosphere (80% N 2 ; 20% O 2 ) Table 4-3 XRD peaks characteristic of raw palm shells and palm shells after combustion in DTF 1200 ºC with atmosphere (80% N 2 ; 20% O 2 ) Table 4-4 XRD peaks characteristic for all samples P1, P2 and P3 blends before and after combustion in DTF 1200 ºC with atmosphere (80% N 2 ; 20% O 2 ) Table 4-5 Characterization of FTIR peak profiles of raw MC and MC after combustion at 1200 ºC at atmosphere 80% N 2 ; 20% O Table 4-6 Characterization of FTIR peak profiles of raw palm shells and palm shells after combustion at 1200 ºC at atmosphere 80% N 2 ; 20% O xxv

27 Table 4-7 Signal assignments for CP/MAS 13 C NMR of MC (Erdenetsogt, B.-O., Lee, I., Lee, S. K. et al., 2010) Table 4-8 Resonance assignments for CP/MAS 13 C NMR spectra of agricultural wastes (Bardet, M., Hediger, S., Gerbaud, G. et al., 2007, Link, S., Arvelakis, S., Spliethoff, H. et al., 2008) Table 5-1 XRF analysis showing the amounts of major inorganic elements (based on their oxides) present in the ash of the coke, coconut shell blends and coconut shells following treatment at 1200 C under 20% oxygen and 80% nitrogen atsmosphere Table 5-2 XRD peaks characteristic of raw coconut shells and coconut shells after combustion in DTF 1200 ºC with atmosphere (80% N 2 ; 20% O 2 ) Table 5-3 XRD peaks characteristic for all samples C1, C2 and C3 blends before and after combustion in DTF 1200 ºC with atmosphere (80% N 2 ; 20% O 2 ) Table 5-4 Characterization of FTIR peak profiles of raw coconut shells and coconut shells after combustion at 1200 ºC at atmosphere (80% N 2 ; 20% O 2 ) Table 5-5 Type of molecular bonds and bond energies required to break the fuels considered in this study (Blazej, A. and Kosik, M., 1993, Smith, L. H., and, S. L. D. and Fletcher, 1994) Table 7-1 Minimum bubble diameter in slag droplets in contact with metallurgical coke, palm char and palm shell blends Table 7-2 Maximum bubble diameter in slag droplets in contact with metallurgical coke, palm char and palm shell blends Table 7-3 Reaction rate constant (moles/cm 2 s) for 100% metallurgical coke, 100% palm char and palm shell/coke blends Table 7-4 Chemical composition of coke, palm char, rubber and PET xxvi

28 Table 7-5 Comparison of the maximum rate of FeO reduction by different carbonaceous materials obtained from literature Table 7-6 Type of crystallization of polymers and molecular bonds in the carbonaceous materials (Antal, M. J., Jr. and Varhegyi, G., 1995, Orfão, J. J. M., Antunes, F. J. A. and Figueiredo, J. L., 1999, Sharma, R. K., Wooten, J. B., Baliga, V. L. et al., 2004) Table 7-7 The range of diameters of small gas bubbles entrapped in the slag droplet for coke and its blends with polymer between 2 10 minutes of reaction Table 10-1 Ash analysis of metallurgical coke xxvii

29 List of Publications Publications: 1.) M. Yunos, N. F, Zaharia M, Idris M. A., Nath D., Khanna R., Sahajwalla V. (2012), Recycling Agricultural Waste from Palm Shells during Electric Arc Furnace Steelmaking, Energy and Fuels, Vol. 26, (1), pp ) M. Yunos, N. F., Zaharia M., Ahmad K. R., Nath D., Iwase M., Sahajwalla V. (2011), "Structural Transformation of Agricultural Waste/Coke Blends and Their Implications during High Temperature Processes", ISIJ International, Vol. 51, (7), pp ) M. Yunos, N. F., Ahmad K. R., Zaharia M., Sahajwalla V. (2011), "Combustion of Agricultural Waste and Coke Blends during High Temperature Processes: The Effect of Physical, Chemical and Surface Properties", Journal of the Japanese Society for Experimental Mechanics, Vol. 11, (Special Issue), pp Conference: 1.) M. Yunos, N. F, Ahmad K. R, Zaharia M, Sahajwalla V (2011), "Agricultural Wastes-A Resource for EAF Steelmaking", AISTech The Iron & Steel Technology Conference and Exposition, Indianapolis, US, pp xxviii

30 Chapter 1: Introduction CHAPTER 1 1 Introduction 1.1 Background Nowadays the environmental benefits of wood or other forms of biomass/agricultural wastes associated with the reduction of CO 2 in the atmosphere is attracting wide attention due to its CO 2 -neutrality, its contribution to the preservation of natural resources and its flexibility in the production of solid, liquid and gaseous fuels. Biomass/agricultural wastes can be used in the production of chemicals and liquid fuels, charcoal for use in metallurgy, carbon adsorbents from wood and biomass/agricultural wastes and for co-firing with coal/coke in steelmaking. Steelmaking is an energy intensive industrial sector and contributes significantly to the anthropogenic CO 2 emissions in the world. With an estimated value of 20 35% of the annual industrial energy consumption and is one of the largest CO 2 emitter accounting for approximately 5% of the total CO 2 emission (Ooi, T. C., Aries, E., Ewan, B. C. R. et al., 2008, Yanjia, W. and Chandler, W., 2010). Since many of the unit processes in the chain from ore to steel have already evolved to a mature state, and further optimisation of their operation to reduce the emissions is difficult, global concern about increasing greenhouse gas concentrations in the atmosphere drives researchers to look for other solutions, such as capturing and storing the arising CO 2 emissions,(ariyama, T. and Sato, M., 2006, Danloy, G., Berthelemot, A. and Grant, M., 2009, Huijgen, W. J. J. and Comans, R. N. J., 2005) or replacing the fossil hydrocarbons (coal, oil and natural gas) by agricultural waste materials. Two types of agricultural waste materials used for this study are palm shell and coconut shell. 1

31 Chapter 1: Introduction The combustion of carbon based materials releases chemical energy. However, their effective usage as energy depends on the kinetics of their reaction with oxygen and the extent to which they are converted to CO gas (the reaction that liberates energy). Malaysia is the second largest palm oil (Elaeis guineensis) producing country in the world, with 30 million tonnes, out of which 8.2 million tonnes are discarded as palm waste, consisting of empty fruit bunches, fibres and shells. Agricultural waste materials derived from palm shells are among the main renewable waste source in Malaysia. Approximately 2.01 million tonnes of palm shells were generated in 2010 alone, and a steady 5% increase has been seen in the last few years (Azri, S. M., 2008, Chor, L. Y., 2010, Mae, K., Hasegawa, I., Sakai, N. et al., 2000). Coconut shell (Cocos nucifera) consists of a hard and thick bony endocarp material, which presents serious disposal problems in the local environments (Chung, J. K., 1997, Guo, S., Peng, J., Li, W. et al., 2009). Recycling this unused resource would add to its economic value, helping reduce the cost of waste disposal and most importantly, providing an inexpensive alternative to conventional coke (Heschel, W. and Klose, E., 1995, Su, W., Zhou, L. and Zhou, Y., 2006). Agricultural wastes having high volatile matter content, may find their possible utilization in combustion with low volatile coke. As compared to coke, agricultural wastes also contain high oxygen and easy release of volatile matter in a combustor. All these characteristics of agricultural wastes have been found to have large influence on the burn out time of blends of coke and agricultural wastes. Moreover, agricultural waste chars were found to have porous and highly disordered carbon structure and belong to the class of most reactive carbon materials. The porosity within the chars causes more accessibility of the reactive gas to active sites resulting in the very good combustion reactivity. Agricultural waste materials are a renewable source of carbon and a possible replacement for coal/coke used in high temperature processes. The aims of this high temperature study involving blending palm shell/coconut shell wastes with coke with application in high temperature processes include developing an understanding of the following high temperature phenomena: 2

32 Chapter 1: Introduction 1. Gas phase reactions: pyrolysis and its subsequent step combustion; of high importance and novelty is the influence of palm shell and coconut shell wastes blend with coke on the kinetics of devolatilization, gas formations, crystallinity and associated structural transformations as a result of high temperature gas phase reactions. The intra and intermolecular bonding present in the structure of the raw materials are expected to play a significant role in controlling gas phase reactions which can influence the structure of the resulting carbonaceous particle and also subsequent carbon/ slag reactions. 2. Carbon/slag interactions: the generated carbonaceous residue left behind after the combustion reaction can replace some of the coke in promoting FeO reduction reactions and foaming slag. The influence of gas generation, volatile matter and mineral matter from agricultural wastes are expected for controlling in carbon/slag interactions. Slag foaming is a feature of most conventional Electric Arc Furnace steelmaking (EAF) processes. In the modern EAF, maintaining a stable foamed slag to submerge the arc is critical for high productivity operation. In the slag foaming process, carbon is injected into the slag, reacts with FeO in the slag producing CO, which foams the slag. Slag foams have been investigated with smaller bubbles than those used in the previous studies (Hara, S. and Ogino, K., 1992, Zhang, Y., 1995). The bubbles were generated by argon gas injection with the nozzle of multiple small orifices and by the slag/metal interfacial reaction of FeO in the slag with carbon in the liquid iron. Slag foaming is influenced by two main factors: 1) rate of gas evolution by the reduction reactions and 2) the stability of the foamy layer in the melt. The foaming stability depends on the high viscosity, low surface tension and increased suspension of second phase particle (Hara, S., 1990, Nexhip, C., Shouyi, S. and Jahanshahi, S., 2004, Pathak, D. C., 1997). Besides the formation of gases in the slag layer, a very important reaction, when carbon is put in contact with the slag, is the reduction of iron oxide (Nagasaka, T., Hino, M. and Ban-Ya, S., 2000). 3

33 Chapter 1: Introduction The type of carbonaceous material and its properties play a critical role in governing its reaction with gas and slag phases. Previous studies have focused attention on slag properties and their influence on foaming, e.g. the foam index incorporates slag properties (Ito, K. and Fruehan, R., 1989). Our groups investigated the influence of carbonaceous materials on high temperature reactions in EAF steelmaking (Rahman, M., 2006, Rahman, M., Khanna, R., Sahajwalla, V. et al., 2009, Sahajwalla, 2009, Zaharia, M., Sahajwalla, V., Khanna, R. et al., 2009), while Corbari and Fruehan (Corbari, R., Matsuura, H., Halder, S. et al., 2009) measured the rate of carbon/slag reaction using five different types of carbonaceous materials. However, the significant influences of agricultural wastes are not yet fully understood, such as: volatile matter, mineral matter, carbon structures and gas formations. This study investigates the partial replacement of coke with waste materials such as palm shells and coconut shells and their influence upon gas phase reaction and slag/carbon interactions, including the influence of volatiles, gas formations, carbon structures, and mineral matter. The fundamentals of these reactions are key to understanding combustion efficiency and carbon/slag interactions. 4

34 Chapter 1: Introduction 1.2 Research Objectives The present study investigates the influence of metallurgical coke and its blends with agricultural waste materials (palm and coconut shells) on gas phase reactions, carbon/slag interactions and associated gas generation. The objectives of this research are listed below: To study the gas phase reactions of different blends of agricultural wastes/coke and establish the influence of volatiles, gas formations, carbon structure, crystallnity, mineral matter and their release on structural transformations and gas phase reactions. To establish the properties of the residual char and develop fundamental understanding of the interdependence between properties and reactions with gas and slag phases. To study carbon/slag interactions in terms of reduction reactions and subsequent gas entrapment at the carbon/slag interface. Specifically, the aims are: Study the gas generation from FeO reduction as a result of carbon/slag interaction with the subsequent gas entrapment which ultimately influences relative slag volumes obtained. Study gas entrapment in the EAF slag using quantitative technique involving slag volume measurements with respect to time and also through qualitative understanding of gas bubbles trapped within the slag. Study the influence of gas generation, volatiles matter, mineral matter and entrapped gas bubbles in the bulk slag on slag foaming from agricultural wastes. 5

35 Chapter 2: Literature Review CHAPTER 2 2 Literature Review 2.1 Agricultural Waste as a Source of Renewable Materials Agricultural waste is an important contributor to the world economy. Today, various forms of agricultural wastes energy are consumed all over the world. Agricultural waste provides a clean, renewable material source that could dramatically improve the environment, economy and energy security. In developing countries (e.g. Malaysia), the use of agricultural waste is of high interest, since these countries have economy largely based on agriculture and forestry. The use of these materials will depend on the state of the art of safe economic technologies which are able to transform them into manageable products (Sensöz, S., Demiral, I. and Ferdi Gerçel, H., 2006). The use of agricultural waste as carbon/energy source is of interest due to the following benefits: (i) (ii) Agricultural waste is a renewable, potentially sustainable and relatively environmentally benign source of energy. A huge array of diverse materials, frequently stereo chemically defined are available from the agricultural waste giving the user many new structural features to exploit (Bozell, J. J., 2008, Demirbas, A., 2006). (iii) Increased use of agricultural waste would extend the lifetime of diminishing crude oil supplies. (iv) Agricultural waste fuels have negligible sulfur content and therefore, do not contribute to sulfur dioxide emissions that cause acid rain. 6

36 Chapter 2: Literature Review (v) The combustion of agricultural waste produces less ash than coal combustion and the ash produced can be used as a soil additive in fields, etc. (vi) The combustion of agricultural waste is an effective use of waste products that reduces the significant problem of waste disposal, particularly in municipal areas. (vii) Agricultural waste provides a clean, renewable energy source that could improve our environment, economy and energy securities (Annamalai, K., Priyadarsan, S., Arumugam, S. et al., 2007, Bozell, J. J., 2008) and (viii) Use of agricultural waste could be a way to prevent more carbon dioxide production in the atmosphere as it does not increase the atmospheric carbon dioxide level. Agricultural wastes are known to grow in a sustained way through the fixation and release of CO 2, mitigating global warming problems. In fact, the amount of CO 2 produced during the combustion of the fuels is the same amount absorbed during grown of the plants. This is particularly the case of energy crops and agricultural residues (Munir, S., Daood, S. S., Nimmo, W. et al., 2009). At present the palm oil industry generates the most agricultural waste from the oil extraction process such as the mesocarp, fibre, shell, empty fruit bunch (EFB) and palm oil mill effluent (POME). About 2.01 million tons of palm oil wastes are generated every year in Malaysian alone, and this keep increasing at 5% annually (Yang et al., 2006). Coconut shells have little or no economic value and their disposal is costly, also cause environmental problems. It is used as a reference for shell type materials in assessing the viability of other materials. Therefore, by subjecting palm shell and coconut shell to identical experimental conditions, the structure of the carbonaceous material can be compared. Agricultural wastes may vary in its physical and chemical properties due to its diverse origin and species. However, agricultural waste is structurally composed of cellulose, hemicellulose, lignin, extractives and inorganics. From the chemistry point of view, 7

37 Chapter 2: Literature Review agricultural waste is composed of series of long chain hydrocarbons with functional groups such as hydroxyls and carboxyl. Furthermore, it can be defined as a hydrocarbon, which consists mainly of carbon, hydrogen, oxygen and nitrogen. Some types of agricultural waste present significant proportions of inorganic species. The concentration of ashes generated for this inorganic goes 1% in softwoods until 15% in herbaceous biomass and agricultural wastes (Yaman, S., 2004). Agricultural waste is the fourth largest energy source in the world after coal, petroleum and natural gas, providing about 14% of the world s primary energy consumption (Annamalai, K., Priyadarsan, S., Arumugam, S. et al., 2007). Agricultural waste is used to meet a variety of energy needs, including generating electricity, fuelling vehicles and providing process heat for industries (Bridgwater, A. V., 1999, Bridgwater, A. V., Meier, D. and Radlein, D., 1999). It is the only renewable source of carbon that can be converted into convenient solid, liquid and gaseous fuels through different conversion processes (Özbay, N., Pütün, A. E., Uzun, B. B. et al., 2001). 8

38 Chapter 2: Literature Review 2.2 Environmental Impact Atmospheric gases such as carbon dioxide, nitrous oxide and methane can regulate temperature of the earth. These greenhouse gases particularly CO 2 allow energy from the sun to penetrate to the earth, but trap the heat radiated from the earth s surface. Researchers, scientists and others are concerned about those gases being emitted to atmosphere by human activities which will increase the global warming at a rate extraordinary in human history. The CO 2 emission from the usage of fossil fuels that provide about 85% of the total world demand for primary energy, cause an increase of the CO 2 concentration in the atmosphere (Zanzi, R., Sjöström, K. and Björnbom, E., 1996). Emission of mainly sulphur dioxide, nitrous oxide and hydrochloric gases to atmosphere can cause acid rain. Sulphur oxides and nitrogen oxides can be transformed in the atmosphere to H 2 SO 4 and HNO 3. Sulphur oxides are produced in combustion of sulphur bearing fuels such as petroleum and coal. Sulphur oxides emission from the utilization of biomass/agricultural waste is negligible because they contain minimal sulphur. Another acidic gaseous pollutant is hydrochloric acid (HCl) gas, produced from chlorine and mainly associated with combustion of municipal wastes. HCl also plays an important role for dioxin formation during combustion. Special attention is being paid to the nitrogen oxides emission from combustion of nitrogen such as biomass/agricultural waste, coal, peat or municipal waste. The nitrogen oxides emission from combustion of nitrogen comes from two sources, thermal nitrogen oxides and fuel nitrogen oxides. The formed from the nitrogen in the combustion air and its formation is more or less dependent on the temperature and pressure in the combustor. The latter comes from the oxidation of nitrogen in the fuel and is not particularly temperature sensitive. All the oxides of nitrogen also enhance the greenhouse effect. 9

39 Chapter 2: Literature Review During gasification, the nitrogen forms ammonia (NH 3 ). Some hydrogen cyanide (HCN) and nitrogen monoxide (NO) may also be formed. During combustion of the gases, ammonia and cyanides undergo oxidation to nitrogen oxides the pyrolysis in the initial step in both gasification and combustion (Zanzi, R., Sjöström, K. and Björnbom, E., 2002). 2.3 Recycling of Agricultural Wastes in Steelmaking Palm shell is a waste product of the palm oil industry. Currently, the material has no specific technical uses and it creates a huge disposal problem (Hussain, A., Ani, F. N. and Darus, A. N., 2006). It is an added advantage to the oil palm industry if the excess shell can be turned into useful and valuable products. It was estimated that about 1120 kg of palm shells are produced per hectare of oil palm planted area (Azri, S. M., 2008). The proximate analysis data of fixed carbon and ash content in palm shell shows that it is a suitable raw material for the steelmaking (R.C. Bansal, J.-B. D., F. Stoeckli, 1988). Coconut shell is chosen for comparison, since it is commercially used as a precursor for metallurgical industries. It is used as a reference for shell type materials in assessing the viability of other materials. Therefore, by subjecting palm shell and coconut shell to identical experimental conditions, the structure of the carbonaceous material in term of structured can be compared. In previous study, charcoal from hardwood species in small blast furnaces is being employed for iron production in Brazilian steel industry (Emmerich, F. G. and Luengo, C. A., 1994). The use of wood char in ironmaking has been extensively reviewed by Gupta, Burgess and Dell Amico et al. (Burgess, J., 2004, Gupta, R. C., 2003, M. Dell Amico, P. Fung, R. L. and O Connor, J. M. a. M., 2004). High resistivity of charcoal resulted in more efficient operation with respect to energy and electrode consumption. Low ash content charcoal (0.4 %) is derived from eucalyptus type wood, while timber waste (such as saw dust, chips and cuttings) is used to 10

40 Chapter 2: Literature Review improved gas permeability of the burden and prevent charge crusting and decreases electrical conductivity (Spratt, D. M. and Brosnan, J. G., 1990). Babich et al. (Babich, A., Senk, D. and Fernandez, M., 2010) found that the combustion efficiency of all the tested charcoals is better or comparable with conventional coals. The use of palm shell charcoal for the production of good quality steel was considered by Emmerich et al. (Emmerich, F. G. and Luengo, C. A., 1996). Recently, carbon iron ore composite consisting of biomass char coated with submicron iron oxide powder and iron ore fines were proposed to improve the reduction rate in the blast furnace (Watanabe, K., Ueda, S., Inoue, R. et al., 2010). 2.4 Properties and Composition of Agricultural Waste and Metallurgical Coke Agricultural waste contains as main elements of C, H and O. The molecular structure of palm shell based on one C atom can be written as CH x O y (CH 1.61 O 0.51 ) (Lee, D. H., Yang, H., Yan, R. et al., 2007). Palm shell is a waste derived from palm oil which contains as main components saturated and unsaturated aliphatic carboxylic acids with carbon chain length in the range of C 6 up to C 24 such as the palmitic acid, (CH 3 (CH 2 ) 14 COOH) (Figure 2-1). Figure 2-1 Molecular structure of common fatty acid in palm oil plant The chemical composition of agricultural waste is very different from that of coal, coke, etc. The presence of large amounts of oxygen in plant carbohydrate polymers means the pyrolytic chemistry differs sharply from these other fossil fuels (coal). Plant agricultural wastes are essentially a composite material constructed from oxygen-containing organic polymers. The major structural chemical components with high molar masses are 11

41 Chapter 2: Literature Review carbohydrate polymers and oligomers and lignin. The organic and inorganic minerals are also present in agricultural waste. The major constituents consist of cellulose (a polymer glucosan), hemicelluloses (xylan/polyose), lignin, organic extractives, and inorganic minerals. 12

42 Chapter 2: Literature Review Cellulose Cellulose, which is shown in Figure 2-2, approximately constitutes 40% (w/w) of the mass of agricultural waste materials. Its structure is composed of d-glucose (6 carbon sugar) units (C 6 H 10 O 5 ) bound together by ether-type linkages (glycosidic bonds). The weakest bond in the chain is the C-O glycosidic bond. The main functional group present in this compound is the hydroxyl group. Figure 2-2 Polymers of cellulose (Graboski, M. and Bain, R., 1981) Hemicellulose Xylan, the most abundant form of hemicellulose, consists of various forms of monomers of d-xylose (5 carbon sugar units: C 5 H 8 O 4 ). These monomers are linked together by ether-type linkages similar to those of cellulose (Antal, M. J., Jr. and Varhegyi, G., 1995, Graboski, M. and Bain, R., 1981, Shafizadeh, F., 1982). Xylan contains mainly hydroxyl (-OH), carboxyl (-COOH) and methoxyl (CH 3 O-) functional groups. Figure 2-3 depicts some important monomers of xylan. 13

43 Chapter 2: Literature Review Figure 2-3 Monomers of xylan (Graboski, M. and Bain, R., 1981) Lignin Lignin, the non-carbohydrate component of the cell wall, is composed of several monomer units, all of which possess phenylpropane-based structure. Lignin exists as an amorphous form around the cellulose fibres, and cements these fibres together. Lignin is connected to cellulose directly by ether bonds. A hypothetical structure of lignin is illustrated in Figure

44 Chapter 2: Literature Review Figure 2-4 A hypothetical structure of lignin (Blazej, A. and Kosik, M., 1993). From a structural point of view, lignin occupies the strongest structure due to the strength of methylene units (-C 2 H 2 -) and π bonds that constitute the aromatic ring (Antal, M. J., Jr. and Varhegyi, G., 1995, Graboski, M. and Bain, R., 1981, Shafizadeh, F., 1982). When agricultural waste becomes degraded under high temperature, lignin is considered to be most resistant to biochemical degradation and presumed to be largely intact. Lignin, hemicellulose and lignin are deemed to constitute the main chemical structure of agricultural waste materials. Therefore, the macromolecular structure of agricultural waste can be perceived to consist mainly of polymers of aliphatic (from cellulose and hemicellulose) and aromatic molecular (from lignin) incorporated with oxygencontaining functional groups. 15

45 Chapter 2: Literature Review Table 2-1 shows the cellulose and lignin content in the agricultural samples employed in this study. The term of holocellulose represents the total cellulose, which is composed of cellulose and hemicelluloses. Table 2-1 Cellulose, holocellulose and lignin contents of palm shell and coconut shell (Wan Daud, W. M. A. and Wan Ali, W. S., 2004) Compositions (%) Palm Shell Coconut shell Cellulose Holocellulose Lignin Inorganic Minerals Tables 2-2 to 2-6 summarize the inorganic matter and chemical composition present in agricultural waste materials that used for this study as well as the results from previous literature. 16

46 Chapter 2: Literature Review Table 2-2 Inorganic minerals in agricultural waste by XRF analysis literature review Author (Afrane, G. and Achaw, O.-W., 2008) Ash analysis by XRF (%) (Ghani, W. M. A. W., Firdaus, M. S. and Loung, C. J., 2008) Present study Components Coconut Shell Palm Shell Coconut Shell Palm Shell SiO Fe 2 O Al 2 O TiO P 2 O Mn 3 O CaO MgO Na 2 O K 2 O SO

47 Chapter 2: Literature Review Table 2-3 Ultimate analysis of palm shell wastes material literature review Carbon Hydrogen Nitrogen Sulphur Oxygen Author (wt. %) (wt. %) (wt. %) (wt. %) (wt. %) Present study (Hussain, A., 2006) (Mahlia, T. M. I., Abdulmuin, M. Z., Alamsyah, T. M. I. et al., 2001) (Emmerich, F. G. and Luengo, C. A., 1996) (Harimi, M., Megat Ahmad, M. M. H., Sapuan, S. M. et al., ) (Sumathi, S., Bhatia, S., Lee, K. T. et al., 2009) Table 2-4 Proximate analysis of palm shell wastes material literature review Author Volatile matter (wt. %) Fixed Carbon (wt. %) Moisture (wt. %) Ash (wt. %) Present study (Hussain, A., 2006) (Guo, J., Luo, Y., Lua, A. C. et al., 2007) (Lua, A. C., Lau, F. Y. and Guo, J., 2006) (Sumathi, S., Bhatia, S., Lee, K. T. et al., 2009) (Wan Daud, W. M. A. and Wan Ali, W. S., 2004)

48 Chapter 2: Literature Review Table 2-5 Ultimate analysis of coconut shell wastes material literature review Carbon Hydrogen Nitrogen Sulphur Oxygen Author (wt. %) (wt. %) (wt. %) (wt. %) (wt. %) Present study (Bandyopadhyay, S., Chowdhury, R. and Biswas, G. K., 1999) (Wan Daud, W. M. A. and Wan Ali, W. S., 2004) (Tsamba, A. J., Yang, W. and Blasiak, W., 2006) (Mohd Din, A. T., Hameed, B. H. and Ahmad, A. L., 2009) Table 2-6 Proximate analysis of coconut shell wastes material literature review Author Volatile matter (wt. %) Fixed Carbon (wt. %) Moisture (wt. %) Ash (wt. %) Present study (Bandyopadhyay, S., Chowdhury, R. and Biswas, G. K., 1999) (Wan Daud, W. M. A. and Wan Ali, W. S., 2004) (Tsamba, A. J., Yang, W. and Blasiak, W., 2006) (Mohd Din, A. T., Hameed, B. H. and Ahmad, A. L., 2009)

49 Chapter 2: Literature Review Metallurgical Coke Metallurgical coke is a macro porous carbon material of high strength produced by carbonization of coals of specific rank or of coal blends at temperatures up to 1400 K (Díez, M. A., Alvarez, R. and Barriocanal, C., 2002). The metallurgical coke consist a good strength and a fixed carbon content of wt%, with volatiles ranging between 1-5 wt%, and wt% sulphur content. Moisture content is a direct consequence of the coke-quenching process with some dependence on size. Coke is primarily used as a fuel and reductant in electric arc furnaces (Zaharia, M., 2010); however, it also functions as a support for other raw materials in the blast furnace. This material has been widely used in the applications for electric furnaces in production of steel, ferro alloys, graphite and refractories. Coke is also added to steel ladles to re-carburize the melt after discharge from the steelmaking furnace Activated Carbon Activated carbon can be prepared from a large number of materials. These materials are usually high in carbon and volatile contents but low in inorganic contents. Some of the most common precursors for activated carbons are coal, palm shell, coconut shell, lignite and wood. For materials like coal, studies have shown that increasing the temperature initially increased the surface area and porosity and it reached its maximum between C (Wan Daud, W. M. A. and Wan Ali, W. S., 2004). Similar trends was found in lignin, where the effect of high temperature on char porosity and surface area. For palm shells and coconut shells, the surface area and porosity showed an increase with increase temperature. (Lua, A. C. et al., 2006). 20

50 Chapter 2: Literature Review 2.5 Gas Phase Reactions Introduction The processes during heating and combustion of coal/coke are illustrated in Figure 2-5, and they are similar for biomass/agricultural waste except for high volatile matter. The process of release of gases from solid fuels in the absence of oxygen is called pyrolysis, while the combined process of pyrolysis and partial oxidation of fuel in the presence of oxygen is known as gasification. If all combustible gases and solid carbon are oxidized to CO 2 and H 2 O, the process is known as combustion. Homoge nous reactions Volatile Matter Off gas Moisture Vapo ur Coal/coke Biomass/ agricultura l waste Char & Ash (Pyrolysis, coal/coke 400 C, Biomass/ Agricultural waste 200 C) Heterogeneous char reactions (Char combustion) Ash Figure 2-5 Process during coal/agricultural wastes pyrolysis, gasification and combustion 21

51 Chapter 2: Literature Review Pyrolysis Pyrolysis offers a simple way of processing all of the complex polymeric structure found in agricultural waste lignocellulosic residue materials. It can offer high yields of liquid product that retains the elements needed for chemical analysis and products. Pyrolysis is the thermal decomposition process under inert atmosphere through which a biomass/agricultural waste decomposes to various products like solid char, liquid tar and gases, some of which have high calorific values. The operating conditions and the agricultural waste pyrolysis units can maximize the yields of tars and condensable products or char (Antal, M. J., Mochidzuki, K. and Paredes, L. S., 2003, Bridgwater, A. V., 1999, Bridgwater, A. V., 2003). In the former case, the conversion process is indicated as fast pyrolysis (Bridgwater, A. V., 2003) and, after cooling and condensation, bio-oil is obtained. Fast pyrolysis requires high heating and heat transfer rates in the reaction zone, a primary conversion temperature of about 800 K, and short residence time of products in the vapour phase (below 2 s at about 700 K) with a rapid cooling of the vapour-phase products to limit the extent of secondary reactions. Conversely, the conventional or slow pyrolysis, process produces comparable yields of char, gas, and tar species. It is important to describe the evolution of gas and tar volatile products as a function of the process conditions. Yields of CO and H 2 are the desired products of the gasification process and are favoured by these secondary pyrolysis reactions of volatile species. Most of the pyrolysis characteristics reported in literature are for woody materials. A few attempts have been made at correlating the pyrolysis characteristics of biomass with those of its constituents, viz. cellulose, hemicellulose, lignin and extractives (Antal, M. J., Jr. and Varhegyi, G., 1995). Pyrolysis of palm shell waste resulted in biomass chemicals such as methyl derivatives, acetic acid and most certainly phenols derivatives (Islam, M. N., Zailani, R. and Ani, F. N., 1999). Moreover, Yaman (Yaman, S., 2004) presents about two hundreds articles in biomass research studies and none of them is a coconut shells or palm shells. It can conclude that there are a limited number of researchers that have dedicated their investigation to these agricultural wastes. 22

52 Chapter 2: Literature Review Raveendran et al. (Raveendran, K., Ganesh, A. and Khilar, K. C., 1995) studied about 13 different types of biomass samples in which coconut shell is included. However, not too much attention was give particularly to this material. Thus, in this study, palm shell and coconut shell pyrolysis are investigated using thermogravimetry with the endeavor of characterizing their thermal decomposition process, particularly, pyrolysis profiles and gas revolutions. Figure 2-6 present summaries of reaction mechanisms of pyrolysis/steam gasification of cellulose and lignin, respectively by TGA-MS in the literature. Lignin was decomposed in the temperature range of K, yielding 60 wt % of nascent char. Evolution of CO 2 peaked at 673 K. A significant increase in H 2 evolution was observed above 773 K; it reached a peak at 873 K. However, no pronounced evolution of CO and CO 2 was observed in this temperature range. These results imply that aromatization and carbonization of the lignin-nascent char proceed to yield H 2 and char. Evolution of CO 2 and CO exhibited a weak peak at 973 K in accordance with cellulose pyrolysis. Above 823 K, H 2 evolution increased drastically. Subsequently, a steep rise in CO 2 and CO evolution was observed. These suggest that the decomposition of lignocellulosic structure from agricultural waste/biomass will released more gases at high temperature to participate for subsequent carbon/slag interactions in this present study. 23

53 Chapter 2: Literature Review Figure 2-6 Reaction mechanism of pyrolysis and steam gasification (dotted line represents the reaction with steam): (a) cellulose; (b) lignin 24

54 Chapter 2: Literature Review Heterogeneous Char Combustion At high temperatures, solid fuel particles evolve the volatiles that burn in diffusion envelope flames. For most solid fuels, a heterogeneous char combustion phase follows. Heterogeneous char reaction is one of the main controlling steps in the combustion process of carbonaceous materials which affects the heat released in the combustion system. Laurendeau (Laurendeau, N. M., 1978) and later on Smith (Smith, I. W., 1982) and Smith and Smooth (Smith, L. H., and, S. L. D. and Fletcher, 1994) have focused on the kinetic measurements of various carbonaceous materials. However, due to the complexity of these processes, none of the previously developed methods are completely reliable without further experimental measurements. 1. Two stages are encountered in the combustion of a carbon based material (Figure 2-5): Sample pyrolysis during the volatile matter is removed, forming a solid residue with certain physical and chemical structure, i.e. char Carbon based material Char + Volatiles 2. A heterogeneous reaction of the residual char or char combustion C + O 2 = CO 2 Eq. 2-1 C + ½ O 2 = CO Eq. 2-2 The pyrolysis reaction is much faster relative to the heterogeneous char combustion, which spreads over a longer period of time and is not the rate limiting step for the whole combustion process. Nevertheless, it determines the amount and the structure of char generated. 25

55 Chapter 2: Literature Review Kinetic Reactions Regimes Combustion occurs, as a heterogeneous gas solid reaction generally does, in three different regimes controlled by the interplay of transport processes and chemical reactions (Walker, P. L. J., Shelef, M. and Anderson, R. A., 1968). Three different situations may occur that are commonly referred to as regime I (or zone I), regime II, and regime III. In regime I, the overall burning rate is controlled by the chemical heterogeneous reaction between O 2 and the carbon of the char particle. At low temperature zone, O 2 will fully penetrate the porous char so that reactivity increases with total internal surface area. The chemical kinetics is extremely low, therefore the O 2 diffuses into the interior of the porous char faster than it is consumed, providing high diffusion rate. The char burns across the whole particle; the particle size (d p ) remains constant with burn-off, while the particle density (σ p ) decreases. The reaction rate of the particle, ρ m (the reaction rate per unit remaining carbon mass), can be expresses as: m = m ki Ai PO 2 ρ Eq. 2-3 Where: k i, and A i are reactivity constant and specific total surface area of the residue, respectively, while order. m PO 2 represent the oxygen partial pressure as a function of m reaction In regime II (intermediate temperature), the rate is determined by the chemical reaction as well as by internal diffusion of O 2 in the char pores. The reactivity of the char will depend mainly on internal burning and the rate gaseous diffusion through the char wall thickness, total porosity and pore size distribution of the char. During oxidation, large pores increase in size and the material between the large pores tends to shrink. Although the small pores within the granules are associated with a larger surface area for reaction, O 2 penetration into these pores is far less extensive than into the large pores. This limiting condition is referred to as regime II. Smith and Hurt (Smith, L. H., and, S. L. D. and Fletcher, 1994) derived the reaction rate and burnout time for a char particle in zone II, such that: 26

56 Chapter 2: Literature Review m+ 1 m+ 1 [ 2 k A D P ( 1 x) /( m 1) ] 0. 5 ρ = 2 + Eq a σ p i i e O Where: D e is effective pore diffusivity and x is the actual reaction rate to the maximum possible reaction rate. As the temperature is further increased, reactivity may be strongly controlled by diffusion of O 2 through the boundary layer to the particle surface. The rate of diffusion of the reactant gas toward a particle is determined by its external dimensions, so reactivity may be influenced by particle size of the carbonaceous material. Finally, in regime III, external diffusion of O 2 from the bulk phase to the particle surface controls the burning rate. The chemical reactivity of the char will not influence the reaction rate. Nevertheless, for the I st and II nd regime, where the char heterogeneous reaction is controlled by either the chemical reaction of the combination of char chemical reaction and pore diffusion, the char chemical structure and physical structure may have important implications for char reactivity. Many combustion studies indicate that, by using a reactor that entrains the carbon particles in the gas stream at relatively high temperatures, leads to a reaction zone (kinetic regime II) or in the transition region between zone I and II (Hurt, R. H., 1998, Shim, H.-S., Hurt, R. H. and Yang, N. Y. C., 2000, Smith, L. H., and, S. L. D. and Fletcher, 1994) Inorganic Effects in Combustion The combustion or co-combustion of coal/coke and biomass/agricultural waste involves devolatilisation of the solid fuel particle and the gas phase combustion of the volatiles, followed by the burning of the resultant char in an atmosphere, the temperature and oxygen concentration of which are large controlled by the combustion of the volatiles. The reactivities of the coal char vary from one char to another because of differences in the char surface area which determines both the available surface area and the number of active sites, and differences in the content and nature of the inorganics that can act as 27

57 Chapter 2: Literature Review a catalyst or inhibitor. In the case of agricultural waste, the carbon structures can vary particularly due to the incorporation of O-atoms which has a disruptive effect on the char structure. As the original char particle burns the char structure changes and ultimately becomes annealed. This reduces the reactivity of the remaining char that can result in a greater tendency to form unburned carbon. In addition, the ash content increases and this can have an inhibiting effect on the char reactivity. As the temperature reduces towards the end of the combustion chamber the reaction regime changes from zone II to I. At these reduced reaction temperatures catalytic effects due to metals in the ash can become more influential, and the presence or indeed absence of catalytic reactions can play a significant role. The effects of ash inorganic species are still uncertain. Species such as iron and calcium are naturally present in coal often with no effect on the combustion reaction and yet metals are sometimes added as a combustion improver, and the controlling factors are not well understood. Agricultural wastes chars are generally more disordered compared to coal chars and have significant oxygen content. These, and the easy interaction with metal catalysts change the reactivity of the biomass char very significantly (Backreedy, R. I., Jones, J. M., Pourkashanian, M. et al., 2003) Effects Due to the Inorganic Ash Compounds There are two major features that are important in coke and agricultural waste char combustion and these are the formation of ash and catalyst effects. Ash formation Ash is formed during combustion and the fate of this ash is controlled by the temperature in the furnace and the ash composition, i.e. whether it is molten or not. Several approaches have been taken to model this effect and this can be considered to 28

58 Chapter 2: Literature Review have one of two roles. In the first, the ash is considered to form a protective layer (Hampartsoumian, E., Murdoch, P. L., Pourkashanian, M. et al., 1993, Zolin, A., Jensen, A., Jensen, P. A. et al., 2001) around the burning char particle and as reaction proceeds this inhibits combustion. The second approach is to consider that the ash is thrown off the char particle leaving only ash within the surface layer but which still has a blocking effect, that is, only a fraction of the surface is exposed. Application is therefore still difficult because of the uncertainties in applying these effects which can cause a disproportionate amount of inhibition for small char particles. Effects of catalysts Catalysts may influence both devolatilisation and char burnout, but the former is only of significance in the devolatilisation of biomass ( Chen, Y., Charpenay, S., Jensen, A. et al., 1998). It is well known that metals will catalyse the oxidation of carbon at temperatures of about 500 ºC. However, there is some difficulty in determining the influence of catalysts at furnace conditions. Metals may occur in coal combustion either distributed within the char matrix or as separate clusters of metal oxides or in the case of volatile metals as condensed vapour. Three classes of catalysis may be identified as follows: Class 1. Distributed metals: Only a small part of the mineral matter is present as organometallic-type compounds and can result in a distributed form of metals along the reaction interface. Few studies have been made of such but the evidence from studies ( Jones, J. M., Agnew, J., Kennedy, J. et al., 1997) involving Fe and V show that the effect is to not to change the activation energy of char oxidation but to change the pre-exponential factor, that is there is an acceleration of the reaction. Backreedy et al. (Backreedy, R. I., Jones, J. M., Pourkashanian, M. et al., 2002) have put forward a mechanism for this process which is based on the formation of a C metal bond as shown in Figure 2-7 that outlines the 29

59 Chapter 2: Literature Review oxidation mechanism. The effect is twofold, a weakening effect on the adjacent C C bond and a role as an oxygen donor. In the case of alkali metals a different mechanism is likely to hold. In this mechanism (Jones, J. M., Darvell, L. I., Pourkashanian, M. et al., 2005) a C O M bond is formed (where M can be Na, K or Ca) as shown in Figure 2-8 and this could be particularly important in the case of oxygenated biomass compounds. The effect here is the weakening of the adjacent C C bond and hence catalysing the reaction. Class 2. Clusters of metal atoms formed by condensation: This class of catalysts effectively operates by being released into the gas phase during devolatilisation, and typical metals would be potassium and sodium. The metal can deposit in two ways. The first is in the form of clusters and the second is via attachment through a C O bond that can also act as the nucleus for a cluster. There is evidence that in the case of clusters the activation energy for reaction is lower ( Baker, R. T. K., Dumesic, J. A. and Chludzinski, J. J., 1986) and this mechanism is the preferential reaction at low temperatures. Class 3. Ash clusters: The presence of ash clusters does not seem to have much influence on the reaction rate because the degree of contact with carbon at the interface is small. The effects of these catalysts are subsumed in the values of alpha here but it would be desirable to separate the influence of the terms into pure char combustion and the catalytic effect. 30

60 Chapter 2: Literature Review Figure 2-7 Proposed catalytic mechanism for the reaction of iron atoms in the char structure (Backreedy, R. I., Jones, J. M., Pourkashanian, M. et al., 2003) Figure 2-8 Proposed catalytic mechanism for the reaction of alkali metals (Huang, H. Y. and Yang, R. T., 1999) 31

61 Chapter 2: Literature Review Determination of the Combustion Efficiency of Coke and its Blends with Agricultural Waste Factors Affecting Gas Phase Reactions The processes involved in gas phase reactions, such as pyrolysis and combustion are very complex. Depending on the furnace type, carbonaceous material composition, particle size, the relative rates of heating, decomposition and oxygen transfer, the gas phase reactions may occur in separate stages or simultaneously. Furnace type During high temperature gas phase reactions, physical and chemical changes appear. Research on combustion performance of different carbon based material has been reported widely in the literature. Drop tube furnace (DTF), which is an entrained flow reactor, thermogravimetric analyser (TGA) and fixed bed reactor (FB), are the type of furnaces, used in gas phase reaction studies. At low temperature, where the reaction occurs in the kinetic regime I, TGA and FB are the most commonly used reactors. At high temperature, where the reaction occurs in the kinetic controlled regime II and III, the heterogeneous char reactions can finish in a very short time. Neither TGA nor FB has the ability of withdrawing the particle so fast, thus the flow particle technique, in which a stream of carbonaceous particle is entrained into a gas atmosphere is more suitable for this purpose. The drop tube furnace has been widely considered by many authors (El-Samed, A. K. A., Hampartsoumian, E., Farag, T. M. et al., 1990, Kim, B. C. and Gupta, S., 2007, Liddy, J. P., Newey, D. C. and Wilson, T., 1987, Wells, W. F., Kramer, S. K., Smoot, L. D. et al., 1985). 32

62 Chapter 2: Literature Review Heating rate Whilst the combustion process occurs very rapidly, of great importance is the heating rate of the carbonaceous particle. Such a parameter determines the nature of the devolatilization products, including the resulting char. Higher heating rates result in greater amount of volatiles released and thus more changes in the porous structure and the sizes of the remaining char particles ( Kimber, G. M. and Gray, M. D., 1967). Residence Time and Temperature Residence time is a very important parameter in char preparation, since char structure is very sensitive to the residence time, especially in the earlier stages of pyrolysis. When the flow particle technique is employed, such as DTF, the residence time can be adjusted by changing the position of the collecting probe. Preliminary measurements have been performed previously ( Solomon, P. R., Serio, M. A., Carangelo, R. M. et al., 1986) by recording the signals from transistors installed at both ends of the working tube. Fletcher et al. (Fletcher, T. H., Kerstein, A. R., Pugmire, R. J. et al., 1990) and Pugmire et al. (Van Niekerk, D., Pugmire, R. J., S olum, M. S. et al., 2008) have investigated the structural evolution of chars of two American coals, considering their residence time. It was discovered that a higher residence time leads to a more ordered graphitic like carbonaceous structure. During combustion, chars obtained at higher temperatures give carbonaceous matrices of higher surface area. This can be related to the fact that chars obtained at higher temperature have almost totally expelled their total volatile matter, thus forming matrice with larger surface areas which are hence more reactive. Van Niekerk et al. (Van Niekerk, D., Pugmire, R. J., S olum, M. S. et al., 2008) considered a few American coals and applied different temperatures and residence times. They discovered using NMR technique, an increase in the number of carbons per aromatic cluster with increasing temperature and residence times. Arrenillas et al. 33

63 Chapter 2: Literature Review (Arenillas, A., Rubiera, F., Arias, B. et al., 2004) studied the relationship between structure and reactivity of different carbonaceous materials as a function of temperature. During the first stages of combustion, as a result of higher temperature, the crystallites or the so-called turbo static structure developed gradually. An increase in the heat treatment temperature causes a substantial decrease in BET surface area accompanied by a loss in active sites. This study agreed well with the work done by Liming Lu et al. (Lu, L., Sahajwalla, V. and Harris, D., 2000), which observed an increase in the size of the graphitic planes with increasing treatment temperature. Hecker & McDonald et al. (McDonald, K. M., Hyde, W. D. and Hecker, W. C., 1992) prepared in a reactive methane flame at a maximum particle temperature of 1700 K, a series of chars derived from Diets sub-bituminous coal. An increase in CO 2 and N 2 surface area with increasing residence time was seen. Effect of the carbonaceous material composition Every carbonaceous material is characterized and ranked on the basis of its chemical composition, such as volatile matter, ash and carbon content. The carbon content determines the calorific value of the fuel as well as the temperature of the flame. Generally speaking, the gross properties increase with increasing carbon content. The volatile matter content determines the flame stability. It was established that with decreasing volatile matter, the ignition of the carbonaceous materials delayed and the combustion is slow. As the residence time in DTF is of the order of 1 to 2 s, it may be expected that the carbon based material with poor combustion performance will pass through the furnace in a partly unburned state. Volatile present in the composition of the carbonaceous material leads to the formation of pores in the residue when subjected to a high temperature process. Depending on the total amount of volatiles in the carbonaceous matrix a certain number of pores of different dimensions are formed. Depending on their sizes the pores within a carbonaceous particle may be classified into three categories; namely, micropore (<2 34

64 Chapter 2: Literature Review nm), mesopore (2-50 nm) and macropore (>50 nm) (Everett, D. H. and Haynes, J. M., 1972). A pore tree model was initially proposed by Simons and Finson (Simons, G. A. and Finson, M. L., 1979) which tries to explain the change in pore distribution by considering phenomena such as pore combination and pore growing due to chemical reaction. With increasing number and distribution of pores, its surface area increases, allowing the oxygen to penetrate the carbonaceous matrix and deplete it. The surface area developed by micro and mesopores is the one that accounts for increased combustion performance. However, a strong evolution of volatile has an endothermic effect and might deplete the dense phase of oxygen and cause hot spots (Ogada, T. and Werther, J., 1996) and possible particle fragmentation. 35

65 Chapter 2: Literature Review 2.6 High Temperature Carbon/Slag Interactions Introduction The interfacial phenomena between the slag and carbon are expected to play a major role in slag foaming during EAF steelmaking because they could dictate the kinetics of reduction reactions. EAF steelmaking slags mainly contain iron oxide, silica, alumina, lime, manganese and magnesia. In the temperature range of interest, 1500 C to 1700 C, iron oxide, manganese and silica are the main reducible oxides present in the slag. In addition, the ash impurities present in the char also contain iron oxide and silica to varying degrees and could participate in reduction reactions. Slag foaming is achieved by injecting oxygen and carbonaceous materials (carbon/slag interactions) into the molten metal bath and the slag respectively. Some oxygen is required for oxidizing impurities and to reduce the carbon content while the simultaneous injection of carbonaceous materials like inject coke, petroleum coke, natural graphite, synthetic graphite, char etc (slag foaming injectant different carbonaceous materials) and oxygen into the slag layer produces the most effective foamy slag. By understand the carbon/slag interactions is of great importance due to their extensive application in a number of metallurgical processes, such as steelmaking, blast furnace ironmaking and direct reduction ironmaking processes (Hara, S. and Ogino, K., 1992, Jiang, R. and Fruehan, R., 1991, Katayama, H., 1992, Ogawa, Y., 1992). It is widely accepted by the industries involved in high temperature materials processing, that physical property data are extremely valuable in designing new processes and improving process control and product quality. A key carbon/slag interaction is slag foaming which involves the entrapment of gas bubbles in the slag layer when carbon based materials interact with O 2 in the steel melt. This gas foam the slag from its normal thickness to a higher volume depending on different parameters such as: carbon material employed, amount of oxygen injected, 36

66 Chapter 2: Literature Review FeO content in the slag, temperature, etc. CO gas bubbles are generated according to the reaction between solid carbon and the FeO present in the slag: FeO + C = Fe + CO Eq. 2-5 This reaction is endothermic and can lower the temperature of the slag; meanwhile, the carbon entering the bath will react with oxygen as follows: C + O = CO Eq. 2-6 The CO produced will foam the slag and react in part with the available FeO, resulting in the reaction: CO + FeO = Fe + CO 2 Eq. 2-7 Benefits of the slag foaming practice have been widely reported in the literature (Bisio, G., Rubatto, G. and Martini, R., 2000, Wang J and B, Z., 2007). Mostly, adequate slag foaming occurs at the beginning of refining but decreases towards the end of the heat. From a metallurgical point of view, slag foaming is a quite complicated phenomena and it can be difficult to establish as well as control. Operators commonly estimate the foaming behavior during operation by visually inspecting the slag and by listening to the sound of the EAF process. In practice, the steelmaker' aim is to control the foaming behavior such that, the right amount of foam is created at the right time and is maintained for the desired amount of time (Pretorius, E. B. and Carlisle, R. C., 1999) Experimental Techniques for Carbon/Slag Interactions Previously, a number of different experimental techniques have been used to study the entrapment of gases resulting from carbon/slag interactions. Ito and Fruehan (Ito, K. and Fruehan, R., 1989) used a tall alumina crucible where a foamy slag was placed and the surface position of the slag was detected with a stainless steel electric probe while Kitamura and Okohira (Kitamura, S. and Okohira, K., 1992) measured the slag 37

67 Chapter 2: Literature Review height by the slag adhesion length on a steel rod inserted in the crucible at 60 sec intervals. Kapilashrami (Kapilashrami, A. and Görnerup, M., 2006) used X-ray radiographic images and measured the foam height visually. Khanna et al. (Khanna, R., Spink, J. and Sahajwalla, V., 2007, Rahman, M., 2006) used the sessile drop arrangement and a novel video processing software to measure the volume changes in the foamy slag. Dynamic changes in volume and contact area of the foamy slag were determined from the captured images with the help of especially designed software detailed subsequently. A schematic diagram illustrates every step involved in the novel technique, developed with the aim to quantify the slag volume changes during carbon/slag interactions (see Figure 2-9). Detailed image analysis techniques are available in the literature for the accurate determination of contact angles and surface tension from the shape of sessile drops; Maze and Burnet (Maze, C. and Burnet, G., 1971) have developed a numerical algorithm using a nonlinear regression analysis for computing droplet profile from a number of arbitrarily selected coordinates on the experimentally measured sessile drop. However, in a reactive system, a detailed mathematical drop-shape analysis will be extremely difficult due to continuous/sporadic evolution of gases and dynamic changes in the droplet shape and size. As the droplet is not able to reach an equilibrium configuration, fitting a mathematically computed profile to the observed droplet through detailed image analysis is not likely to yield valuable/ reliable information. As the primary focus of this work was on a quantitative estimation of rapidly changing droplet volume and slag foaming, the slag droplet on the substrate was assumed to have a truncated spherical shape. A very large number (>1000) of slag droplets were analyzed to determine slag foaming a a function of time. Minor deviations from the spherical outline were not explicitly taken into account and were neglected (estimated error - <5 pct). Using the specially developed software detailed above, a circular marquee was fitted manually to provide a best fit to the slag droplet, and the centre "0" of the spherical droplet was identified (Figure 2-10). By manually clicking on point " p1 " and " p2," 38

68 Chapter 2: Literature Review the coordinates of three data point (p1: the left inter section point between the slag and the substrate, p2: the right intersection point between the slag and the substrate, and 0 : the centre of the slag droplet) were recorded automatically. With a click on "next," the computer program then brings the next frame into view and the process is repeated. 39

69 Chapter 2: Literature Review Figure 2-9 Representative example of the video screen during data processing. The coordinate for point's p1 and p2 are displayed in the data line just above the image, along with the computed values of volume and area of contact (arbitrary units) (Khanna, R., Spink, J. and Sahajwalla, V., 2007, Rahman, M., 2006) 40

70 Chapter 2: Literature Review START DVD file.vro DVD file.vob > 30min with.mpeg2 video < 30min File_1.VOB. File_n.VOB.mpeg2 data File_1.VOB Virtualdub program Individual bit map frames with.bmp code (24bits) IrfanView program map frames with.gif code (8bits) LINUX based Computation of slag droplet volume in a chos en frame. and END Figure 2-10 Schematic diagram of the novel processing software for volume and contact area measurements (Khanna, R., Mahfuzur, R., Richard, L. et al., 2007) 41

71 Chapter 2: Literature Review The overall process is semiautomatic, with the user playing an important role in identifying the optimum spherical shape for the slag droplet at a given instant of time and contact points p1 and p2 across the slag/substrate interfacial region. The volume of the slag droplet and area of contact between the slag and substrate were computed in terms of the radius, r of the droplet and the truncated height, h (Figure 2-10). The data from all images were stored in a data file for further processing. While it is rather difficult to measure the absolute volume of the droplet in the absence of a reference marker, this technique generates the foaming ratio V/V 0 and area ratio A/A 0 as a function of time, where V 0 and A 0 are, respectively, the initial volume and contact area for the slag droplet. Sessile drop method The Sessile Drop Method is a method used for the characterization of solid surface energies, and in some cases, aspects of liquid surface energies. The main premise of the method is that by placing a droplet of liquid with a known surface energy, the shape of the drop, specifically the contact angle, and the known surface energy of the liquid are the parameters which can be used to calculate the surface energy of the solid sample. The liquid used for such experiments is referred to as the probe liquid, and the use of several different probe liquids is required. The theoretical background of this method is based on the work of Bashforth and Adams (Bashforth, F. and Adams, J. C., 1983) who solved Laplace equation for a sessile drop and the solution has been used to determine surface tension from the profile of the liquid drop. 42

72 Chapter 2: Literature Review Previous Studies on Carbon/Slag Interaction Slag foaming is a common feature in steelmaking processes. It is found in most converters and is prevalent in the blast furnace, during pretreatment of hot metal, and in bath smelting processes. Foaming of slags is also common in the electric arc furnace (EAF). The foam has both desirable and undesirable effects on the process performance. Because of its relevance in metallurgical processes, a number of studies (Fruehan, R., 1977, Ito, K. and Fruehan, R., 1989, Jiang, R. and Fruehan, R., 1991, Kim, H. S., 2001) were carried out by different investigators to characterize the foaminess of slag. The stability of slag foams has mostly been measured using the inert gas injection technique, originally developed by Bikerman (Bikerman, J. J., Perri, J. and Booth, R., 1953). He defined a constant of proportionality termed the foaming index Σ, based on the observation that the volume of foam formed at a steady state was proportional to the gas flowrate, Q. The constant of proportionality was named foaming index. V = Σ.Q Eq. 2-8 where: V is the foam volume at steady state and Q is the gas flowrate blown through a column of liquid slag. Ito and Fruehan (Ito, K. and Fruehan, R., 1989) and Jiang and Fruehan (Jiang, R. and Fruehan, R., 1991) developed a foaming index in terms of the average travelling time of gas through the foam and concluded that the bubble size affect the slag foaming stability. h = U Σ Eq. 2-9 where: h is the height of the foam at steady state when gas with superficial velocity U is passed through it. The fluctuations behavior changes due to slag composition had been investigated by Kapilashrami et.al.(kapilashrami, A. and Görnerup, M., 2006). Morales et al. (Morales, R. D. and Rodríguez-Hernández, H., 2003) quantified the foaming 43

73 Chapter 2: Literature Review behavior by introducing a foaming index which is correlated to the changes in the slag volume and the rate of generated gas. Foaming index, Σ can be obtained from the slope of a plot of foam height versus superficial gas velocity. Σ represents the residence time (or average travelling time) of the gas in the foam layer. This procedure was used by Cooper and Kitchener (Cooper, C. and Kitchener, J., 1959) Swisher and McCabe (Swisher, J. and McCabe, 1964), Kozakevitch (Kozakevitch, P. and Olette, M., 1971), and Hara and Ogino (Hara, S. and Ogino, K., 1992) for CaO SiO 2 and CaO SiO 2 FeO slags. The foaming index depends primarily on the slag composition and process temperature. The physical properties of the slags, i.e. density (ρ), viscosity (ɳ) and surface tension (σ) are related to the composition and temperature of the slag and therefore will affect foaming. Zhang and Fruehan (Zhang, Y. and Fruehan, R., 1995) incorporated the effect of bubble size (d b ) and carbonaceous particles on slag foaming. Mukai et.al (Mukai, K., Nakamura, T. and Terashima, H., 1992, Mukai, K., Toguri, J. M., Kodama, I. et al., 1986) focused on the size of the bubbles and where the bubble size from the inert gas blowing technique did not correspond to the bubble size caused by CO gas generated at the slag and metal interface. 44

74 Chapter 2: Literature Review Factors Affecting Carbon/Slag Interactions Carbon/slag interactions have been a subject of great interest due to their extensive applications in a large number of metallurgical processes. The study is particularly relevant to the ironmaking processes, including the blast furnace ironmaking process. Since the prediction and control of carbon/slag interactions are required for current and future ironmaking and steelmaking processes, an understanding of the factors affecting these interactions are very important. Cooper and Kitchener et al., (Cooper, C. and Kitchener, J., 1959 ) investigated the significance of surface tension during slag foaming process in their work. It is believed that the presence of a surface-active substance, to decrease in surface tension of the slag, is an essential requirement for achieving better slag foaming. The adsorption of the substance lowers the surface tension of the bubble films and provides protective layers for stopping bubble coalescence. Cooper and Kitchener et al., (Cooper, C. and Kitchener, J., 1959) also proved that there exists a direct relationship between the foam stability and the amount of surface-active species. Since then, Kitamura & Okohira (Kitamura, S. and Okohira, K., 1992) reported that P 2 O 5 content was responsible for the decrease of surface tension in the CaO-SiO 2 -Al 2 O 3 -P 2 O 5 -TiO 2 -MgO slag system. Swisher and McCabe (Swisher, J. H. and McCabe, C. L., 1964) measured the foam life of CaO-SiO 2 slags and they have found that, the foam life of CaO-SiO 2 slag increases with decreasing basicity and temperature. Hara and Ogino (Hara, S. and Ogino, K., 1986) studied the effect of slag composition, surface active additives and gas composition on the foaming behavior of FeO-SiO 2 -CaO slags and concluded that foam height increases sharply when the ratio of O/Si gets decreased below 3.5. They showed that the melt viscosity does not have much role in the foaming behavior of the slags, but surface tension plays an important role on foam life. Differences between the dynamic and static surface tensions of the film might also lead to a strong elastic surface. Swisher and McCabe (Swisher, J. H. and McCabe, C. L., 1964) studied the effect of Cr 2 O 3 on the foaming of CaO- SiO 2 slags in the temperature range of C when the CaO/ SiO 2 ratio was equal to 0.8. They reported that 45

75 Chapter 2: Literature Review Cr 2 O 3 increased the foam life of liquid slags. In order to explain this phenomenon they applied Gibbs' elasticity theory and estimated changes in the thickness of the bubble lamellae. The surface tension was seen to increase but only at very low extents; therefore the Gibbs ' theory was ruled out and they concluded that the Marangoni elasticity effect and slag viscosity are possible contributors. The foaming phenomenon from the surface chemistry point of view was discussed in detail by Kozakevitch (Kozakevitch, P., 1969). He considered the film surface as a two dimensional "surface phase" Kozakevitch conducted extensive studies on other factors influencing carbon/slag interactions such as surface tensions of slags. He considered the CaO- SiO 2 -FeO systems and found that the surface tension depends on the surface activity of SiO 2 which in turn is dependent on the CaO and FeO content of the melts. CaO decreases the activity of SiO 2 more than an equivalent amount of FeO does and hence the dissolution of CaO removes silica from the surface layer. Gaskel and Skupien (Skupien, D. and Gaskell, D., 2000) measured the surface tension of melts from CaO-FeO-SiO 2 system, where the iron oxide content was maintained constant at 30 wt pct. The data were used to correlate slag foaming indexes with surface tension and viscosity. Foam life was determined to increase with increasing viscosity and decreasing surface tension. The influence of P 2 O 5 and sulphur were studied and the authors observed that the addition of sulphur increased surface tension while more phosphorus oxide caused a decrease in surface tension, thus P 2 O 5 was considered to be surface active (Figure 2-11). Figure 2-11 shows the influence of S and P 2 O 5 addition on the surface tension of 30% FeO containing slags. Similar behavior has been observed in previous work Kozakevitch, Elliot, Bhattacharyya and Gaskel. (Bhattacharyya, P. and Gaskell, D., 1996, Elliott, J. F., 1988, Kozakevitch, P. and Olette, M., 1971). 46

76 Chapter 2: Literature Review Figure 2-11 The influence of S and P 2 O 5 on the surface tension of lime-silica slags containing 30% FeO at 1673K (Skupien, D. and Gaskell, D., 2000) Ito and Fruehan (Ito, K. and Fruehan, R., 1989) quantified the slag foaminess for CaO-SiO 2 -FeO slags in the temperature range of C. The effect of P 2 O 5, S, MgO and CaF 2 on the foaming was also studied. As expected, slag foaming increased with increasing viscosity and decreasing surface tension. Also CaO and MgO were regarded as suspended second phase particle that stabilized the foam and had a larger effect on slag foamability bringing changes in viscosity and surface tension of the studied slags. Kozakevitch (Kozakevitch, P. and Olette, M., 1971) considered that a high viscosity is perhaps the most obvious factor in the stabilization of foams, as a high viscosity has the role of retarding the rate of drainage in the film. Figure 2-12 shows the relation between the foaming index and the slag viscosity for the CaO-SiO 2 -FeO system. The foaming 47

77 Chapter 2: Literature Review index increases with increasing viscosity and high foaming prevents the drainage of the liquid slag from the film between the bubbles. Figure 2-12 The relationship between the foaming index and viscosity for CaO-SiO 2 - FeO system (Kozakevitch, P. and Olette, M., 1971) Considering the CaO- SiO 2 -FeO system with 30% FeO, Ito and Fruehan (Ito, K., 1989, Ito, K. and Fruehan, R., 1989) performed in-depth studies on the effect of basicity on the foaming index. The liquid slag contained 5% AlO 3 dissolved from the crucible. As shown in Figure 2-13 at 1400 C the foam index decreased with increasing basicity up to 1.22, the increase was attributed to the presence of the second phase particles (CaO or 2CaO.SiO 2 ) ~ which was previously proved and reported to increase the foam index. 48

78 Chapter 2: Literature Review Figure 2-13 Foam index as a function of slag basicity (Ito, K. and Fruehan, R., 1989) FeO Reduction in EAF Steelmaking Slags Iron oxide reduction is the predominant reaction in EAF steelmaking process. This reduction might be influenced by the nature carbon injected, contact area of reaction, injection rate, gas generation and reaction rate. In the reduction process, molten iron oxide contained slag is reduced by solid carbon or saturated liquid iron. And the reduction of FeO in the slag, CO is generated in large amounts, which is mainly responsible for the slag foaming in EAF steelmaking process. The details knowledge of reduction kinetics and mechanism for the EAF steelmaking is of particular significant. Reduction of FeO from slag is one of the fundamental reactions in iron and steel making. Normally, coke and char coal are commonly used as a reactant in the metallurgical process to remove oxygen and hence reduce metal oxides in slag. It has been suggested that, three different mechanisms are responsible depending on the level of FeO content present in the slag (Paramguru, RK, 1997). Generally, the steps of reactions are (Eq to 2-15): 49

79 Chapter 2: Literature Review 1) Reduction of molten FeO (pure or in slag) with solid carbon (Sarma, B., Cramb, A. and Fruehan, R., 1996). 2) Reduction of molten FeO with carbon in carbon saturated iron melt 3) Reduction of molten FeO with CO gas (Bafghi, M., 1992). All the chemical reactions proceed through the gas film at a steady state. From the optical microscopic examinations of the present study, Fe is the reduction product at the carbon/slag interface. Iron could be the result of gas reaction at the gas/slag interface, where the reduced iron was carried out to the slag surface by the evolved gas bubbles passing through molten slag (Min, D. and Fruehan, R., 1992). Hydrogen also plays an important role as reductant. Basically, coals and agricultural waste materials contain significant amount of hydrogen as volatile matter, the mechanism of hydrogen reaction should be understood to control the processes. The reactions that are believed to occur at steelmaking temperatures ( C) are divided into the following four categories: 1. Conversion of lignocellulosic compound in agricultural waste materials into hydrocarbons (C n H m ): Lignocellulosic C n H m Eq Thermochemical decomposition of hydrocarbons into carbon and hydrogen: m C nh m( g ) = nc( s) + H 2( g ) Eq The hydrocarbon could also act as a sink for CO 2, producing CO and H2 (Eq. 2-12): m C nh m( g ) + nco2( g ) = 2nCO( g ) + H 2( g ) Eq Reduction of iron oxide by hydrogen, carbon and carbon monoxide: FeO + H = Fe + H O Eq ( g ) ( l) 2 ( g ) 50

80 Chapter 2: Literature Review FeO + C = Fe + CO Eq ( s) ( l) ( g ) FeO + CO = Fe + CO Eq ( g ) ( l) 2 ( g ) 4. Auxiliary reactions: (i) Water gas shift reaction: H O + CO = H + CO Eq ( v) ( g ) 2( g ) 2( g ) (ii) Reaction of water vapour with C produced from the cracking of C n H m : H O + C = H + CO Eq ( g ) ( s) 2( g ) ( g ) (iii) Boudouard reaction: CO + C = CO Eq ( g ) ( s) 2 ( g ) (iv) Carbon dissolution into molten metal: C( s ) = C Eq From the available literature, it was found that the rate of hydrogen reduction is much faster than those with other reducing agents. Katayama et.al (Katayama, H., 1992) first studied the reduction rate of iron oxide with a blast furnace slag using a mixture H 2 - H 2 O. They reported that the kinetics of hydrogen reduction is pure iron oxide is very fast; the reaction rate was estimated with a mixed control model. Considering the previous work, Hayashi and Iguchi (Hayashi, S. and Iguchi, Y., 1994) measured the rate of reduction of pure liquid iron oxide, FeO-CaO and FeO-SiO 2 binary melts with a gas conveying system at 1773 K. They agreed on a faster rate of reduction when hydrogen was present in the system. Xie and Belton (Xie, D. and Belton, G., 2003) observed reduction rates of ferric iron in slags by H 2 -H 2 O were a factor of 2 or 3 times than those by CO-CO 2, while Hayes 51

81 Chapter 2: Literature Review (Hayes, P., 1979) develop maximum chemical reaction rate constants for iron oxide reduction. It was demonstrated that for H 2, the reduction rate is one order magnitude greater than that for CO at 1300 ºC and a factor of 5 times greater at 1600 ºC. Most of the studies performed on reduction of FeO containing metallurgical slags with hydrogen were found to be in good agreement and the general trend consisted in increased reduction with usage of hydrogen as reducing agent EAF Operating Conditions with Slags From previous study, the industrial (OneSteel) used waste rubber tyres as partial replacement of coke with slags containing ~26-27 % FeO. A comparison of the trial data showed improvements when coke rubber blend was injected in the furnace. A detailed paper regarding rubber blend trials as well as a comparison with the conventional coke was published by Sahajwalla and Zaharia et al. (Sahajwalla, Zaharia et al. 2009). As the EAF operating condition can vary over time, it was ensured that the coke comparison heats had similar operating conditions to the rubber heats. The partial replacement of coke as feedstock has been successfully demonstrated in commercial-scale trials using an industrial furnace operated by OneSteel. Table 2-7 Data for the SSM Trial Heats: Comparison between Metallurgical Coke and Rubber Tyre/Coke Mixture (Sahajwalla, Zaharia et al. 2009) Specific EE (kwh/t) Melting rate (Tonnes per min, POT) Carbon (kg/heat) FeO (%) Coke Blend R (115 kgs rubber 52

82 Chapter 2: Literature Review Previous Studies on FeO Reduction and Factors Affecting FeO Reduction by Solid Carbon The reduction of iron oxide in molten metallurgical slags by solid carbon is a very complex reaction and could be dictated by a large number of factors such as: composition of the slag, carbonaceous material used, experimental conditions (temperature and pressure) and gasses atmosphere. Previous studies have focused mainly on the slag characteristics and temperature started with 1300 ºC and reached about 1600 ºC (Fruehan, R. J., 1997, Hara, S., 1990, Hara, S. and Ogino, K., 1992, Ito, K. and Fruehan, R., 1989). However, the nature of the carbonaceous material and its influence upon the reduction reaction is not fully understood and additional information is required to fully understand its implication on the reduction process. Beer et. al (Beer, H. P. and Engell, H. L., 1970) measured the rate of the reaction between CaO-SiO 2 -Al 2 O 3 with 70% FeO with graphite and coke crucible in temperature range of 1200 C to 1500 C by thermogravimetric method. The reduction was found to be 1st order with reaction 2.39, the slag/gas reaction, being the rate limiting reaction. They observed that no significant difference for the rates measured with graphite and coke crucible. This suggested that the type of carbon used does not influence the overall reaction kinetics. The rate for the given slag with graphite crucible at 1400 C was measured to be x 10-5 mol.cm -2.s -1. The contact angle between slag and coke was observed to be greater than 90, implying that the slag was non-wetting with the coke. Fruehan (Fruehan, R., 1977) have investigated the rate of reduction of Fe 2 O 3 and FeO by charcoal, coal and coke, in an inert atmosphere within the temperature range of 900 C to 1200 C. They reported that the reaction takes place by means of the gaseous intermediates CO and CO 2. The rates of reduction of FeO and Fe 2 O 3 by CO are relatively fast and the CO 2 /CO ratio for the oxidation of carbon was determined by their equilibrium. Haiping et. al (Sun, H. and Easman, W., 2007 ) stated that according to the thermodynamic estimation of the reaction (2) FeO from the slag will eventually be removed to a negligible low level if carbon is in excess in the system and enough 53

83 Chapter 2: Literature Review reaction time is applied at temperatures ranging from C. FeO content in the slag can be considered as the driving force for the reaction rate with temperature is considered because of the increase in rates of mass transfer of reactant or the interfacial reaction at higher temperatures. Ozawa et. al (Ozawa, M. and Kitagawa, S., 1986) investigated the reduction of FeO in molten slags by solid carbon in electric arc furnace steelmaking and pointed out that factor like quality of solid carbon injected, boundary area of reaction, wherein solid carbon reacts with FeO and nature of slag could influence the reduction rate of FeO in slag. They reported that, the volatile matter in solid carbon has a greater influence on the reducing reactions. In the coke with higher volatile matter, the reduction was controlled by the chemical reactions, while in the coke with lower volatile matter; the reduction was controlled by the transport of FeO in slag. The reduction rate depends on the basicity of molten slag, reduction rate increases as the basicity of molten slag increase. They estimated carbon content of molten steel by using of the fundamental data on waste gas. Recent studies performed by Corbari and Fruehan (Corbari, R., Matsuura, H., Halder, S. et al., 2009, Matsuura, H. and Fruehan, R. J., 2009) measured the rate of reaction of FeO in slags with five different types of carbonaceous materials: graphite, bituminous coal, bituminous char, anthracite coal and anthracite char. The presence of coals brought additional release of gas in form of volatiles. The release of the volatiles fragments of the coal creating additional surface area and increasing the CO generation rate. The controlling mechanism for the reaction appeared to be quite complex and controlled by more than one single process; whereas the precise rate controlling mechanism was not determine. Jouhari et. al (Jouhari, A. K., Galgali, R. K., Datta, P. et al., 2000) conducted experiments on reduction of FeO in molten slag, using a graphite crucible in a 50 KW capacity plasma reactor. The crucible itself served as reductant which implies that the reduction of FeO takes place by solid carbon. Influence of amount of FeO in the slag, crucible surface area, initial slag height and the effect of CaF 2 addition on reduction rate 54

84 Chapter 2: Literature Review was investigated. They found that the reaction kinetics follows the 1st order reaction which implies that the FeO diffusion in the slag phase controls the reaction rate. They confirmed it by linear variation of rate constant with respect to FeO. Surface active elements in liquid metals, such as oxygen and sulfur, have shown to retard the rate of interfacial reactions (Richardson, F. D., 1974), P 2 O 5 and SiO 2, which are surface active in slags have shown to reduce the rate of reduction of iron oxide in slags. Most literatures found that the rate increased with increasing of % FeO in the slag and (Galgali, R. K., Datta, P., Ray, A. K. et al., 2001, Min, D., 1999, Paramguru, RK, 1997, Sarma, B., Cramb, A. and Fruehan, R., 1996) they agreed that one of the most significant parameter in determination of the rate of the reduction in the slag chemistry. The rate was found increased almost three times when the basicity (CaO/SiO 2 ) increased from 0.5 to 1.5, however at basicity at 2.0, the rate was lower than at 0.5. The viscosity of the slag and diffusivity of FeO depend on the slag chemistry, thereby the slag composition affects the mass transfer coefficient and hence the reaction rate. The reaction rate decreases when a strong surface active agent such as P 2 O 5 is present in the slag. Surface active agents block the reaction sites at the slag/gas interface and thereby reduced the rate. Bafghi et. al (Bafghi, M., 1992) found that the addition of 1 % P 2 O 5 reduced the reduction rate by 50 % when the reaction was primarily chemically controlled. From the above literature survey, it can be understood that, the nature of iron oxide reduction in molten slag by solid carbon is complex and could be influenced by several factors including slag composition, FeO content in slag, different carbonaceous materials as well as the experimental conditions such as temperature and very few investigations have been carried out to study the reduction reaction between EAF steelmaking slag and solid carbon. Most previous studies have focused on the slag characteristics and experimental temperatures. Despite using different experimental techniques, the reduction reaction kinetics with solid carbon is still unclear. Moreover, most of the studies were conducted below 1550 C, which is lower than the appropriate steel making temperature. Limited work has been reported on reduction rates higher 55

85 Chapter 2: Literature Review than 1550 C and also the reaction mechanism that limits the reduction reaction is yet to be established. 56

86 Chapter 2: Literature Review 2.7 Summary Combustion is defined as the burning of a fuel and oxidant to produce heat and energy release. However, the effective usage as an energy resource depends on the kinetics of their reaction with oxygen and the extent to convert CO gas. Compared with metallurgical coke, agricultural waste materials are characterized by higher contents of volatile matter, carbon and oxygen with low ash content. These materials contain high amount of energy that might be supplied when subjected to high temperature reactions. The gas phase reactions at high temperature (> 1000 ºC) of the agricultural wastes have not been investigated before. This present study is focusing on understanding the influence of palm and coconut shells blend with coke on the kinetics devolatilization and associated structural transformations as a result of high temperature gas phase reactions. From the literatures, the intra and intermolecular bonding that present in agricultural wastes might play a significant role in controlling gas phase reactions which can influence the structure of the resulting carbonaceous particle and also subsequent carbon/slag reactions. Slag foaming is an important phenomenon in many metallurgical processes and EAF steelmaking. Slag foamability is as critical a parameter as CO gas generation in order to obtain foam. Foamability is controlled by the slag phase physical properties viscosity, surface tension and density and it s hard to achieve it due to an unstable phenomenon. The literature reviewed is pointed that helps in describing the reactions and mechanism taking place in such a complex system. The slag foaming behavior from agricultural waste materials with application in EAF steelmaking has not been done. The present study is focused on understanding the presence of palm shell char and carbonaceous residue left behind after combustion reaction which might used to replace coke in promoting a stable foaming and continuously as well as FeO reduction reactions. The sessile drop technique had been approached at temperature 1550 ºC (steelmaking temperature) to investigate the reactivity and interfacial phenomena of EAF slag/carbon interactions with an aim to 57

87 Chapter 2: Literature Review determine the role played by the carbonaceous material characteristics towards optimum slag foaming. The carbonaceous materials and their associated properties could influence carbon/slag interactions, i.e.: wettability, FeO reductions and gas entrapment (high temperature) for better foaming behavior. 58

88 Chapter 3: Experimental CHAPTER 3 3 Experimental Characterization of Experimental Materials Key objectives of this study are to correlate the characteristics of solid residue from agricultural wastes in gas phase reactions and carbon/slag interactions. Metallurgical coke, palm and coconut shells waste in different proportions were mixed and characterized in terms of ash, volatile matter, moisture content and elemental characteristics. The combustion efficiency was calculated after subjecting materials to combustion in a drop tube furnace (DTF) and was determined on the basis of raw material and the residual mass collected after reactions in the DTF. In the present study interactions of EAF iron oxide rich slag with the residual char collected after gas phase reactions in the DTF and associate interfacial phenomena was investigated and detailed results are presented in chapter. Carbon/slag interaction investigations were carried out using horizontal tube furnace; these included the formation of gases, their entrapment and subsequent generation in molten slag. A detailed description of techniques used for these investigations, and experimental approach has been presented in the following sections. 59

89 Chapter 3: Experimental Gas Phase Reactions Studies The carbonaceous materials were subjected into drop tube furnace heat treatment and thermogravimetric analyser for gas phase reactions studies. These materials were characterised using chemical analysis, gas products, carbon structure analysis and physical analysis. Details of experimental approach are presented below in a number of sections. 3.1 Preparation of Carbonaceous Materials Carbonaceous materials used for this study were metallurgical coke (MC), agricultural wastes and mixtures of metallurgical coke and agricultural waste over a wide range of concentrations. Metallurgical coke, typically used as carbon injecting material in the EAF steelmaking at OneSteel Sydney Mill, Australia represents the parent material to be partially replaced by the agricultural waste products (palm shells and coconut shells). Figure 3-1 and Figure 3-2 represent relative concentrations of materials used in this study. 60

90 Chapter 3: Experimental MC P1 P 2 P3 PC Figure 3-1 Bar chart of the materials investigated in this study illustrating the relative concentrations of palm shells and MC content in the blends MC C1 C 2 C 3 CC Figure 3-2 Bar chart of the materials investigated in this study illustrating the relative concentrations of coconut shells and MC content in the blends 61

91 Chapter 3: Experimental MC was ground and sieved to a particle size in the range of mm to minimize the effect of particle size on experimental results. Palm shells (PS) and coconut shells (CS) were crushed to smaller sizes by using a cutting mill Pulverisette 15 (Fritsch GmbH, Idar-Oberstein, German) and sieved to a particle size mm. The chemical analysis including proximate (air dry base, %) and ultimate (dry ash free, %) analysis are conducted on all materials as well on the combusted residues. All these analysis have been performed by Amdel Laboratories based on Australian Standards. Tables 3-1 and 3-2 show the chemical analysis of samples prior to gas phase reaction investigations. 62

92 Chapter 3: Experimental Table 3-1 Summary of chemical analysis of palm shells, palm shells char (PC), metallurgical coke (MC) and palm shell blends prior to high temperature reactions Proximate Analysis (%) Ultimate Analysis (%) Volatile Fixed Matter Carbon Ash Moisture C H O S N MC Palm Shell PC P P P Table 3-2 Summary of chemical analysis of coconut shell, coconut shells char (CC), metallurgical coke (MC) and coconut shell blends prior to high temperature reactions Proximate Analysis (%) Ultimate Analysis (%) Volatile Fixed Matter Carbon Ash Moisture C H O S N MC Coconut Shell CC C C C

93 Chapter 3: Experimental The ash analysis by XRF of raw materials is presented in Table 3-3 and agricultural waste blends in Table 3-4. Table 3-3 Ash analysis for metallurgical coke, agricultural wastes and chars Ash analysis by XRF (wt. %) Components MC Palm Palm shells Coconut Shells char Shells SiO Coconut shells char 13.7 Fe 2 O Al 2 O TiO P 2 O Mn 3 O CaO MgO Na 2 O K 2 O SO

94 Chapter 3: Experimental Table 3-4 Ash analysis for coke/agricultural waste blends Ash analysis by XRF (%) Components P1 P2 P3 C1 C2 C3 SiO Fe 2 O Al 2 O TiO P 2 O Mn 3 O CaO MgO Na 2 O K 2 O SO

95 Chapter 3: Experimental 3.2 Chemical Characterization of Specimens Thermogravimetric Analysis with Mass Spectroscopy (TGA-MS) The powdered samples of the individual agricultural wastes and coke were investigated using thermogravimetric analysis (TGA) under non-isothermal conditions from 0 ºC to 1450 ºC to continuously monitor the evolution of gases that are released over a wide range of temperatures. The TGA experiments were performed simultaneously using a thermogravimeter (STA 449 F1 Jupiter, Netzsch Instruments, Inc.) and a quadrupole mass spectrometer (MS; QMS 403 Aëolos) connected to a thermobalance where three coaxial tubes were used for the connection of the two instruments. The inner tube is a capillary transfer line connected to the MS detector ion source. The outer tube is connected to a vacuum source, in order to produce stable laminar flow conditions. Sampling period and duration are adjusted by an electronically controlled solenoid valve. The inert gas helium was used with a flow rate of 0.1 L/min in the thermobalance. The samples, approximately 6 mg were heated in the thermobalance from 0 ºC to 1450 C with a rate of 10 ⁰C min 1. The intensity of the following selected ions at m/z 2 (H + 2 ), m/z 15 (CH + 3 ), m/z 16 (CH + 4 ), m/z 18 (H 2 O), m/z 27 (HCN), m/z 28 (CO), m/z 28 (N + 2 ), m/z 30 (C 2 H + 6 ), m/z 40 (C 3 H + 4 ), m/z 40 (NO) and m/z 44 (CO 2 ) were continuously detected with thermogravimetric parameters (temperature, mass) at different times X-Ray Diffraction Analysis (XRD) XRD is a well established technique, with good reproducibility, that characterizes the atomic level crystal structure of carbonaceous materials. It uses a small amount of sample and collects most of the scattered intensities. A thin and smooth layer of fine raw material and residue collected after gas-phase reactions was set on a glass sample holder. Acetone was used as an adhesive to allow sample to attach to the folder. Table 3-5 summarizes the experimental conditions for this x-ray diffraction study. 66

96 Chapter 3: Experimental Table 3-5 Experimental conditions for X-Ray diffraction studies Operating parameters Values/condition X-Ray Diffraction Unit Model Siemens D5000 Water Flow Rate > 3.4 l/min Water Temperature < 30 ºC Voltage 30 kv Current 30 ma Scanning Range 10º - 55º Step Size 0.02º Speed 0.5 º/min Fourier Transform Infrared Spectroscopy (FTIR) FTIR spectroscopy was used to understand the nature of chemical bonding structure. A small amount of each sample (approximately 0.5 mg) was mixed with 200 mg of KBr to produce a gleam pellet. The Infrared signal was determine in the range cm - 1. Bands were identified by comparison to published assignments (Andreas, G., 2003, Cloke, M., Lester, E. and Thompson, A. W., 2002). 67

97 Chapter 3: Experimental Nuclear Magnetic Resonance Spectroscopy ( 13 C NMR) 13 C NMR spectroscopy was used to identify various carbon groups in MC and agricultural wastes and also in combusted char samples. This method is used to investigate the influence of high temperatures on the char samples, and to compare the carbon groups left behind after combustion in the char samples compared to the raw MC and agricultural waste samples. The 13 C CPMAS spectra were obtained from the agricultural waste samples relatively a short time (maximum number of scans was 208). For the char samples, the needed number of scans was in the range of , because chars were less crystalline compared to agricultural waste samples. The single-pulse NMR method resulted in similar spectra, with 3 times shorter scanning time. Table 3-6 summarizes the experimental conditions for 13 C NMR Spectroscopy. Table 3-6 Experimental conditions for 13 C NMR spectroscopy Operating parameters Values/condition 13 C NMR Spectroscopy Bruker Avance III 300 Solid State spectrometer Frequency (sodium) Spinning Speed ZrO 2 Rotors Signal Enhancement Software Signal Identification 75 MHz 4 khz 4 mm cross-polarization (CP) and 1 H gated decoupling TOPSPIN software 68

98 Chapter 3: Experimental 3.3 Physical Characterization of Specimens Surface Area Measurements Physical chracterization of the samples was performed using quantitative Brunauer Emmett Teller (BET) to measure the surface area. Pore surface area can be determine by both nitrogen and carbon dioxide adsorption onto the surface of the samples. In the present study, the surface area was measured by using nitrogen as adsorbent. A TriStar 3000 V6.02 A surface area analyser was used for these measuremnet. Prior to analysis, samples were outgassed at 200 C under inert nitrogen atmosphere, at least 24 h prior to the measurement Scanning Electron Microscopy (SEM) To understand the structural transformation of the samples study after gas phase reaction studies, the surface morphology of char particles is examined using a high resolution Scanning Electron Micoscopy (SEM) for observation the pores and cells structure of the materials. Original and combusted samples were examined by Hitachi 3400X scanning electron microscope (SEM) operating at an accelerating voltage of 20 kv. In order to do so, char particles are dispersed initially in a prepared resin solution. The char resin mixture is kept under vacuum for 1 hour to remove the entrained gas. After being subjected to metallographic studies, the cross section of the mounted char particles are gold coated and imagined under scanning microscopy. 69

99 Chapter 3: Experimental Carbon/slag Interaction Studies Sessile drop investigation was focused on the reactions occurring between different carbonaceous materials and EAF slags. The sessile drop technique was employed as the technique that allows the study of reaction products as a function of time. In these experiments, furnace tube was purge with Ar with a flow rate of 1.0 L/min. The experimental assembly was monitored continuously by CCD camera, and the reaction time was measured using a counter which was started when the slag sample melted. Once the sample was pushed in, and the counter was started after the melting of the slag droplet on the substrate, the time was carefully monitored and the droplet and substrate quenched (by withdrawing the tray into the cold zone of the furnace) after a fixed period of time. The continuous off-gas evolved during the reactions was measured by using Infrared analyzer (IR). Samples were produced after carbon/slag reactions were quenched for the following times: 1, 2, 10 and 15 minutes to measure the gas bubbles diameter and observing the iron oxide reductions. Once these droplets were quenched, the droplet was removed from the substrate for further interfacial investigations using an optical microscopy and Scanning Electron Microscope (SEM, see section 3.5). This type of analysis allowed the time dependant growth of the interfacial material to be determined. 3.4 Sample Preparation The carbonaceous materials used in this study were initially combusted in a DTF in an atmosphere of 20% O 2 and 80% N 2 at a temperature of 1200 C. The collected residues were ground to fine powders (< 60 µm) using a Rocklab rotation grinder fed with compressed air. Metallurgical coke supplied by OneSteel Sydney mill represented the conventional and parent material used in the present study. A range of proportions of palm shell and coconut shell were mixed with metallurgical coke and the proportions are given in Figure 3-1 and 3-2 while the proximate analyses are presented in Tables 3-1 and 3-2. The analysis was performed at Amdel laboratories following Australian Standards. 70

100 Chapter 3: Experimental Metallurgical coke was used without prior oxidative and thermal treatment, while the mixtures due to high volatile content were combusted in the DTF to partially remove the volatile matter. The agricultural wastes were devolatilized at 450 ºC under N 2 atmosphere for an hour to remove all the moisture. The ultimate and proximate analysis of the carbon based materials used in this study is presented in Table 3-7 while the ash analyses of the parent materials are presented in Table 3-3. Table 3-7 Summary of chemical analysis of metallurgical coke and agricultural wastes char after high temperature reactions Proximate (air dry base) (%) and ultimate analysis (dry ash free) (%) Met. coke Palm char Coconut char Moisture Ash Volatile matter Fixed Carbon Sulfur Total Carbon Hydrogen

101 Chapter 3: Experimental Substrates were compacted in a specially designed steel die using a hydraulic press while applying a force of 4 MPa. The amount of carbonaceous material placed in the steel die was kept constant for all the experimental studies. The compacted, cylindrical substrate had a diameter of 15 mm and thickness of 3 4 mm and was then removed from the die and placed on the alumina stage of the sessile-drop experimental assembly. The industrial slag was provided by OneSteel Sydney mill, from Rooty Hill, Australia for carbon/ slag interactions and the composition is presented in Table 3-8. Table 3-8 Composition of slag EAF slag composition wt. (%) FeO 34.9 CaO 30.4 Al 2 O MgO 10.9 SiO MnO

102 Chapter 3: Experimental Interfacial Phenomena-Optical Microscopy and SEM After the completion of the carbon/slag interaction experiments; e.g. coke-palm shell blends/slag assembly was quenched. The samples were mounted in a mixture of epoxy resin-hardener solution in a ratio of 4 to 1. These were then sectioned vertically so that the carbon/slag interface as well as the slag droplet comes in one plane after polishing. Silicon carbide papers of sizes starting with 250 and reaching 4000 µm were used along with water as a lubricant solution. For the final polishing, diamond paste was spread on the polishing papers. Following the metallographic studies, the quenched assembly was put in a furnace at 60 C for removing the moisture and ensuring a moisture free surface. An optical microscope, Nikon Epiphot 200 at magnification of 500 µm used to capture the slag cross section. A representative example of the interfacial region between metallurgical coke/slag assemblies after 2 minutes of contact is shown in Figure The measured bubble diameters from the optical images for various samples at different times have been reported in Chapter 6. The quantitative estimation of these diameters was carried out using the magnification scale marked on optical microscopic images by Adobe Photoshop 7.0 software. The measuring method is shown on Figure

103 Chapter 3: Experimental 3 rd 1 st 500 µm Figure 3-3 Bubble diameter measuring using Adobe Photoshop 7.0 For SEM and EDS analysis, the samples were subjected to carbon coating as this enables a semi quantitative chemical analysis of the sample. The instrument used in the present study was a HITACHI S3400I present in the Electron microscope unit of UNSW, while the energy dispersive analyser was an Oxford ISIS model Interfacial Phenomena Wetting Behaviour The carbon/slag interaction experiments were carried out and the images of the wetting of the slag on the substrate were captured from the DVD of the complete process using computer software. 74

104 Chapter 3: Experimental Figure 3-4 Illustration of the contact angle measurement using ANGLE software The captured images were analysed using ANGLE software that uses a non-linear regression calculation procedure to determine the left and right contact angles. The average of the contact angle values were used for the present studies. The details for the software are described elsewhere (Sahajwalla, V. 2006). Figure 3-4 illustrates an example of the contact angle measurement using ANGLE software. 75

105 Chapter 3: Experimental 3.5 Experimental Apparatus Drop Tube Furnace (DTF) Gas Phase Reaction Studies A Drop Tube Furnace (DTF) was used for the gas-phase studies in this study (Zaharia, M., Sahajwalla, V., Kim, B.-C. et al., 2009). This furnace is capable of operating up to temperatures of 1650 C, this furnace is specially designed to simulate conditions experienced by the combustion of carbonaceous materials in the EAF steelmaking. A schematic diagram of the DTF used in this investigation is presented in Figure 3-4, while the experimental conditions of DTF are given in Table 3-9. The main elements of DTF consist of a feeding system (Schenck Accurate, Whitewater, Wisconsin), a sampling probe, an electrically heated furnace and a gas distribution system. The furnace is equipped with two type-b thermocouples; the external one (thermocouple 2) is used for furnace ramping and the internal one (thermocouple 1) for an improved response during the measurements. Two gas inlets are present, N 2 represents the main gas and is used to carry the solid reactants into the reaction zone and the secondary gas flow which contains the gaseous reactant, O 2, is designed to meet the primary gas flow at the tip of the water cooled injector. Three mass flow controllers are used to adjust flow rate and composition of both flows. 76

106 Chapter 3: Experimental Primary gas and Carbonaceous particle Thermocouple 1 to Controller Water cooled injector Secondary gas Working tube Flow Straightener Thermocouple 2 to Controller Reaction zone Heating elements Water cooled collector Residual char and off gas out Figure 3-5 Schematic diagram of DTF (DTF is an abbreviation of Drop Tube Furnace) 77

107 Chapter 3: Experimental Table 3-9 Operating conditions of DTF for combustion studies Operating parameters Values/condition Furnace Drop Tube Furnace Model HT VTF (Radatherm Pty. Ltd.) Temperature 1200 C Carbonaceous Materials Metallurgical coke (OneSteel Ltd., Sydney, Australia) Type of Organic Wastes Palm Shells and Coconut Shells (UniMAP, Malaysia) Particle Size mm/ mm Material Injection Rate 0.05 g/sec Gas Composition 20% O 2 ; 80% N 2 Gas Flow Rate 1.00 L/min Residence Time 1-2 sec Heating Rate 104 K/sec LECO Carbon Analyser In order to measure the carbon content of each coke/agricultural wastes blend and its corresponding carbonaceous residue left after combustion in DTF, A LECO analyser was used. The carbon and sulphur analyser (model SC-444DR , LECO Corporation, Michigan, USA) is a non-dispersive, infrared, digitally-controlled instrument designated to measure the carbon and sulphur content in a wide variety of organic materials such as coal, cokes and oil, as well as some inorganic materials such as soil, cement and limestone. The experimental parameters for carbon and sulphur measurement are presented in Table

108 Chapter 3: Experimental Table 3-10 Operating conditions of carbon and sulphur measurements Operating parameters Values/condition Furnace LECO-model SC-444DR Temperature 1250 C Sample Mass g Gas Composition O 2, 99.9% purity Muffle Furnace Ash Measurement Ashing of coke/organic waste mixtures was carried out in an electrically heated muffle furnace suitable for temperature up to 1200 C. Each sample and their corresponding carbonaceous residues left behind after combustion in the DTF, were placed in clean alumina crucibles and were weighed using an analytical balance (model 1702, Sartorius AG, Gottingen, Germany). After weighing, the crucibles were placed in the centre of a muffle furnace and heated up. The samples were allowed to cool down room temperature inside the furnace. The total mass of the crucible and the remaining ash was weighed and recorded. Table 3-11 summarizes the experimental conditions for this ashing study. Table 3-11 Experimental conditions for ashing studies Operating parameters Values/condition Furnace Box Furnace Model (Lindberg/Blue M) Temperature 1200 C Sample Mass 1.5 g Duration 3 hours 79

109 Chapter 3: Experimental Thermogravimetric Analyzer (TGA) A custom made TGA furnace was used to record the loss in weight of the carbonaceous samples on gas phase reactions. It has been extensively used as a means of determining devolatilization characteristics and kinetic parameters of carbonaceous materials. The TGA consists of a mechanically movable vertical alumina tube furnace, a Percisa (1212 MSCS) analytical balance having 1.2 kg of capacity (1 mg accuracy) provided with bottom loading facility, data logging computer and a high temperature furnace. The furnace temperature was controlled by using a thermocouple external to the reaction tube, while the sample temperature was monitored through a thermocouple located under the sample holder inside the reaction tube. A schematic diagram of TGA analyzer is presented in Figure 3-5, while Table 3-12 summarizes the experimental conditions for the thermogravimetric studies employed in this study. 80

110 Chapter 3: Experimental Balance Chamber Purge Line Cooling water Suspe nsion Wire Expandable Bellows Sample Holder Gas Inlet Alumina Tube Heating Elements Isothermal Zone (5mm) Thermocouple Gas Out let Purging Gas Figure 3-6 Front view of the Thermogravimetric Analyzer (TGA) with key features highlighted 81

111 Chapter 3: Experimental Table 3-12 Operating condition of TGA for combustion and devolatilization studies under continuous conditions Operating parameters Values/condition Furnace TGA Temperature 1250 C Carbonaceous Materials Metallurgical coke (OneSteel Ltd., Sydney, Australia) Type of Organic Wastes Palm Shells and Coconut Shells (UniMAP, Malaysia) Particle Size mm/ mm Material Injection Rate 0.05 g/sec Gas Composition N 2, 20% O % N 2 Gas Flow Rate 1.00 L/min Residence Time 1-2 sec The sample was weighed and placed in a high temperature resistant glass sample holder, which was suspended in the sealed cold part of the analyzer. Weight loss was continuously recorded by a data logger connected to a balance and computer when the sample was heated from at 1250 C. 82

112 Chapter 3: Experimental Ash Tracer Method Combustion Efficiency At high temperatures, combustion can occur very rapidly; for example the combustion of a pulverized char sample which is a matter of 1 4 seconds (Smith, L. H., and, S. L. D. and Fletcher, 1994). Considering the short time needed for such a reaction, conventional techniques, such as TGA and Fixed bed reactor are not suitable. Therefore an entrained flow reactor, such as DTF was selected for this purpose. A metered stream of mixture of N 2 and O 2 gases was provided allowing the carbon particle to combust. Partly burned char particles were collected at the bottom of the DTF in a steel collecting probe. Further on, the samples were subjected to chemical analysis to measure the residual carbon content with the aid of LECO analyser, while the ash contents were determined from high temperature interactions in a muffle furnace. Since it is quite difficult to recover the reacted carbon samples, the conventional gravimetric analysis is not suitable to calculate the combustion efficiency. Therefore, the combustion efficiency is calculated by the ash tracer method (El-Samed, A. K. A., Hampartsoumian, E., Farag, T. M. et al., 1990, Hampartsoumian, E., Murdoch, P. L., Pourkashanian, M. et al., 1993) based on the chemical analysis as follows: ɳ = (1 - (A 0. C i )/A i.c 0 )) x 100% Eq. 3-1 where: A 0 and A i are ash content (%) before and after combustion, while C 0 and C i represent carbon content (%) before and after combustion Inorganic Tracer Method Results presented in Table 3-13 are described partly in terms of the extent of mass burnout, which is crucial for the determination of the extent of combustion reactivity. For this reason, the methods utilized in the analysis of the magnitude of the mass burnout are discussed in this section. 83

113 Chapter 3: Experimental Table 3-13 The amounts of major inorganic elements (based on their oxides) present in the ash (XRF) of the coke and agricultural waste materials at 1200 C Material Amount of SiO 2 (%w/w of raw fuel) Amount of Al 2 O 3 (%w/w of raw fuel) Amount of TiO 2 (%w/w of raw fuel) MC Palm shells Coconut shells Agricultural waste materials have a high alkaline content which can easily vaporises at combustion temperatures (Bryers, R. W., 1996, Korbee, R. E., S.; Heere, P.G.T.; Kiel, J.H.A., 1998, Zevenhoven-Onderwater, M., Backman, R., Skrifvars, B. J. et al., 2001). Potentially, this can lead to a large error if the conventional ash tracer method is employed for the burnout measurement. To avoid such error, the extent of mass conversion during the devolatilization and combustion of the agricultural waste was determined using an inorganic tracer method. Principles of the inorganic tracer method and the derivation of relevant mathematical equations are described below: 84

114 Chapter 3: Experimental A raw coke or agricultural wastes particle m 0 x a, 0 x i, 0 A raw coke or agricultural wastes char particle m 1 x a, 1 x i, 1 A raw coke or agricultural wastes ash particle m a x i, a Initial Stage (Stage 0) Initial Stage (Stage I) Initial Stage (Stage II) Figure 3-7 A schematic diagram of the change in mass of representative coke or agricultural waste particle during combustion This method (Meesri, C., 2003) is based on the assumption that the mass of the high melting temperature inorganic materials, especially silica (Si), alumina (Al), and titanium (Ti) is conserved during combustion of coke. Using Si, Al or Ti as x i in the following equations represents the mass percentage of oxides, it can be demonstrated that: (m 0 )(x i, 0 ) = (m 1 )( x i, 1 ) = (m a ) (x i, a ) Eq. 3-2 where, according to Figure 3-6, m 0 is the mass of raw coke/agricultural wastes particle, x i, 0 is the mass percentage of oxides (Si, Al or Ti), m 1 is the mass of coke/agricultural wastes char particle, x i, 1 is the mass percentages of oxides (Si, Al or Ti), m a is the mass of coke/agricultural wastes ash particle, x i, a is the average value of mass percentages of oxides (Si, Al or Ti). Therefore: (m 1 ) = (m 0 )( x i, 0 )/ (x i, 1 ) Eq. 3-3 (m a ) = (m 0 )( x i, 0 )/ (x i, a ) Eq. 3-4 The extent of mass burnout of solid materials can be expressed as (U; % w/w, daf): 85

115 Chapter 3: Experimental U = (m 0 - m 1 )/( m 0 - m a ) Eq. 3-5 Hence, Eq. 3-6 may represent based on the extent of oxides (SiO 2, Al 2 O 3 or TiO 2 ): xi, 0 1 ( m0 m1 ) xi, 1 ( xi, 1 xi, 0 ) xi, a U = = = Eq. 3-6 ( m0 m ) x, a i 0 ( xi, a xi, 0 ) xi, 1 1 x, i a The value of U based on the mass percentage of SiO 2, Al 2 O 3 or TiO 2 can be obtained by repeating the procedure above (Mitchell, R. E. and Akanetuk, A. E. J., 1996 ). Accuracy of the results obtained from Eq. 3-6 is estimated to be in the range of ± 0.9 % at the 95 % confidence level (based on the precision data of the XRF analysis of Si, Al and Ti; details see Table 3-3 and 3-4) The mass burnout of coke/agricultural waste blends For the mass burnout calculation of the agricultural wastes/coke blended materials, the inorganic tracer method was used by using the same equation. For this purpose, the following assumptions were made: (1) All interactions between vaporised inorganic species (eg. Alkali and alkali earth based gaseous species) of agricultural waste and coke mineral matter (eg. Si, Ti, and Al-based compounds) during combustion of their blends that can possibly take place are assumed to occur via a physisorption (gas-solid interaction) process (Dayton, D. C., Belle-Oudry, D. and Nordin, A., 1999) (2) No physical interaction occurs between the blended material particles during combustion. (3) The liberations of incorporated silica (Si), alumina (Al) and titanium (Ti) elements of the blended materials due to convective transport (Baxter, L. L., Miles, T. R., Jenkins, B. M. et al., 1998) and vaporisation are ignored. 86

116 Chapter 3: Experimental Minimization of Error in Combustion Efficiency Determination Calibration of the carbon-sulphur analyzer using a certified reference coal sample was carried out to increase the accuracy of the carbon output values and to minimise measurement errors. However, the analyzed data were still found to have high variability. For this, measurements were repeated several times for all samples until two values were found to be within 2% of each other. Although random errors may still persist, this method of measurement can be adopted to minimise errors and to produce reliable results. This is evident from the good agreement between the measured carbon content in coke and that obtained from chemical analysis. A similar method of measurement were carried out for the ashing of coke, so that repeated measurements were conducted for all samples until two values were found to be within 1% of each other. An average of the two was then calculated for combustion efficiency estimation. The calculation of the samples burnout was repeated at least five times for consistency and reproducibility. Errors in quantitative determination of combustion efficiency were minimized by reputability tests and were generally less than 0.3% Horizontal Tube Furnace Carbon/slag Interactions The sessile drop approach was used to investigate carbon/slag interactions with the aid of an electrically heated laboratory-scale horizontal tube furnace (HF). The furnace tube has an inside diameter of 50 mm and was fabricated from double-walled, vacuuminsulated stainless steel tubes fitted with a cooling fan to dissipate heat. A schematic diagram of the experimental setup is shown in Figure 3-7 and Table

117 Chapter 3: Experimental Thermocouple Alumina Tube Slag Sample Coke/agricultural wastes substrate Quartz Window Date/Time Generator PC DVD Stainless Steel Rod Gas Inlet Alumina Rod Alumina Tray Gas Outlet CCD Camera Figure 3-8 Schematic Diagram of the Horizontal Tube Resistance Furnace used for Gas Entrapment Tests Table 3-14 The experimental conditions for the carbon/slag interactions Operating parameters Values/condition Furnace Horizontal Tube Resistance Furnace Model HTTF 60/20 Temperature 1550 C Substrate Mass 1.6 g Slag EAF slag (OneSteel Ltd.) Slag Mass g Substrate Surface Area 1 cm 2 Argon Flow rate 1.00 L/min Duration 30 minutes 88

118 Chapter 3: Experimental The carbonaceous residue collected after combustion in the DTF was ground for 1 minute in a ring mill pulveriser. The residue was weighed and compacted into a substrate disc of 1 cm 2 area using a hydraulic dry presser for a few minutes, due to the powdery nature of the material. The industrial EAF slag (OneSteel Ltd., Sydney, Australia) was placed on top of the formed substrate. Alumina powder was filled and compacted into the cavity of a graphite specimen holder. Initially, the carbon based substrate and slag assembly was held on the specimen holder, which could be pushed to the centre of the hot zone in the furnace with the help of an alumina rod. The cold zone of the furnace hosted the carbon/slag assembly until the desired temperature of reaction, 1550 ºC was attained. Sample assembly was then inserted in the hot zone; this eliminated any reaction that could occur at lower temperatures and possibly influence the phenomena to be studied at the temperature of interest. The gas flow through the furnace tube was controlled by a digital mass flow meter throughout the experiment, while a high-resolution coloured charge-coupled device (CCD) camera provided with IRIS TV zoom lens was used to capture the live in-situ phenomena in the furnace. The melting of slag marked the beginning of contact time. The output from the camera was channeled to a TV monitor and a video cassette recorder (VCR) to record the entire process as a function of time. This feature allows specific images, displaying the contact between the slag and carbonaceous material, to be captured as a function of time from the DVD into the computer. A time-date generator was used in the system to display the duration of the experimental process. All the experimental runs were recorded up to 30 minutes. Our group at UNSW has developed a novel video processing software (Khanna, R., Mahfuzur, R., Richard, L. et al., 2007) which has significantly enhanced the capabilities of the sessile drop technique for quantitative estimations of slag foaming. Through continuous monitoring of the droplet volume and area of contact, a rapid and quantitative estimation of slag foaming can be carried out determining the extent and stability of the foam as a function of time. 89

119 Chapter 3: Experimental Infra-Red Analyzer (IR) Off-Gas Generation Analysis gaseous emission evolved during carbon/slag interactions were performed using a continuous gas analyser. Coupled with visual examination of the reacting carbon/slag system at high temperatures and off-gas (CO, CO 2 ) data, these studies could be used to investigate the role of chemical reactions, gas generation, carbonaceous materials, and slag composition on carbon/slag interactions (Table 3-15). Table 3-15 The experimental conditions for reduction studies Operating parameters Values/condition Furnace Horizontal Tube Resistance Furnace Model HTTF 60/20 Advance optima continuous gas analyser model Infra-red analyser ABB^ AO2020 Temperature 1550 C Substrate Mass 1.6 g Slag EAF slag (OneSteel Ltd.) Slag Mass g Substrate Area 1 cm 2 Argon Flow rate 1.0 L/min Span gas for calibration 9% CO + 9% CO 2 Duration 30 minutes 90

120 Chapter 3: Experimental Connected to an electrically heated furnace, similar to the one presented for the sessile drop technique, and a data logging computer, CO and CO 2 gases were monitored. The analyser is capable of measuring gasses to approximately ppm and 9.00 vol% levels. The furnace had been purged continuously with argon gas (99.99% purity) to ensure an inert atmosphere and to remove any possible contaminating gases. To ensure that the carbonaceous pellet is the only carbon source in the off gas, the graphite holder was replaced by an alumina tray attached to a steel rod. The sample assembly is kept in the cold zone for 30 minutes. After the furnace attains the hot zone the desired temperature, the sample is inserted and gases are being monitored for 30 minutes. No appreciable change in the gas composition was observed at times beyond 30 minutes. The standard calibration gas to calibrate the ppm scales of the instrument were used by following the standard calibration procedure built in the analyser. Ar of 99.98% purity was used as the zero gas, while mixture of CO and CO 2 was used as span gas X-Ray Fluoroscopy (XRF) Slag Characterization The EAF slag samples were analysed by Oxford Instruments Lab-X3500, which is located at Onesteel, Sydney, a dedicated, robust and reliable bench top XRF slag analyser. The Lab-X3500 performs a quantitative analysis of calcia, alumina, silica, manganese and iron in the slag sample and provides an accurate, precise and reliable measurement using Energy Dispersive (EDXRF) and Wavelength Dispersive (WDXRF) X-ray Fluorescence. X-ray Fluorescence (XRF) instruments work by exposing a sample to be measured to a beam of X-rays. The atoms of the sample absorb energy from the X- rays, become temporarily excited and then emit secondary X-rays. Each chemical element emits x-rays at a unique energy. By measuring the intensity and characteristic energy of the emitted X-rays, an XRF analyser can provide qualitative and quantitative analysis regarding the composition of the material being tested. 91

121 Chapter 3: Experimental 3.6 Reproducibility of Carbon/slag Interaction Experiments Minimization of Error in Contact Angle Determination There could be several potential sources of error in contact angle measurements in the experiments performed in the present study. The contact angle could be influenced by the following: The sample in the furnace of the CCD camera is not perfectly horizontal, and/or The camera and the carbon/slag interface are not in one line The software used for contact angle determination relies in human judgement to fix the edge of the droplet and the substrate. The experimental setup used does allow overcoming of these few errors. For example, slight tilt of the sample or the camera with respect to the horizontal line can be corrected during contact angle determination using the de-skewing feature of the software. The capturing system allows bright and sharply defined images to be processed and stored. This allows the edge of the droplet and the substrate to be defined more accurately. In addition to these features built into the experimental setup, was taken to minimize the errors in contact angle measurements. Several measures have been taken towards minimizing the errors, such that: (1) The camera is ensured to be parallel to the two horizontal axes and not tilted. Thus, the level of the sample can be adjusted while it is inserted into the furnace hot zone by checking the image projected on the television screen. (2) The software used for contact angle measurements uses a magnification ratio as a user defined input, which relates the computer pixels to horizontal and vertical dimensions of the object being focused. This eliminates any errors due to the distortion of the image due to the aspect ratio of the computer screen. This magnification ratio is also a function of distance of the object from the camera. The magnification ratio is determined by first 92

122 Chapter 3: Experimental focusing the camera on to the several grids of known dimensions, at fixed distance from the camera. The magnification ratio and the distance of the camera from the sample, in the furnace are then maintained constant, as determined during the calibration process. The software is then checked for accuracy by using them on systems whose contact angles are well reported. (3) Finally, repeated measures are taken and an average of the contact angle is reported. The variation in contact angels during repeated measurements is observed to be generally less than ±5%. Bending of the sample assembly during the high temperature experiments was observed for some experiments. It was calculated and the maximum angle of bending was around 2 from the original axis. The initial bending was observed from the captured images. However, the slag droplet, contact angle was measured on a static image which is unlikely to be affected by the bending of the sample Estimation of Error in Slag Foaming Behavior (V t /V 0 ) Measurements All the slag foaming experiments with coke and different agricultural blends were repeated at least three times. The experimental procedures were similar for all experiments and not much difference was observed on the measured parameters. In general, the experimental results are reproducible to a reasonably good consistency. For all slag foaming experiments the CCD camera was kept at the same distance and height from the sample assembly and the different measurement showed minor differences. Concerning the quantity of slag and carbon material for the substrate, a great care was taken to ensure precise measurements. However, a small difference was observed for different gas measurements, which suggested that the slag amount was not accurate and the error bar was set at ± The software used for volume measurements relayed on human judgement to fix the edge of the droplet and the substrate. During measurement, extreme care was taken to minimize the human error. 93

123 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends CHAPTER 4 4 Combustion & Structural Transformations of Coke/ Palm Shell Blends: Results & Discussions 4.1 Gas Phase Reactions of Metallurgical Coke and its Blends with Palm Shell For gas phase studies, samples were prepared by blending MC with palm shell particles in a range of proportions (see Figure 3-2), while the chemical compositions of the samples used in the present is shown in Table 3-1. Blends P1, P2 and P3 were chosen for detailed examination. The proximate (air dry base, %) and ultimate analyses (dry ash free, %) of the samples were carried out at Amdel Laboratories and Technical Services; and UNSW based on Australian standards. The agricultural wastes contain a low amount of ash and have relatively low sulfur and high volatile contents as compared to the coke used in the present study. Agricultural waste has high volatile matter and oxygen and thereby has high capacity for easy release of volatile matter in a combustor. All these characteristics of agricultural waste have been found to have significant influence on the burn out time of blends of coal and agricultural waste (Biagini, E., Barontini, F. and Tognotti, L., 2006, Wornat, M. J., Hurt, R. H., Yang, N. Y. C. et al., 1995). 94

124 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends Effect of High Temperature on the Behaviour of the Carbonaceous Material From the chemical point of view, agricultural waste can be regarded as a mixture of three components (cellulose, hemicellulose, lignin) and trace amounts of mineral matter (McKendry, P., 2002), and the pyrolytic behavior of agricultural waste can be considered as a sum of the three components. So far, numerous studies based on the main components have been carried out; most of them were focused on developing kinetics models for predicting the behavior of agricultural waste/biomass pyrolysis at low temperature (< 900 ºC) (Raveendran, K., Ganesh, A. and Khilar, K. C., 1996, Yan, R., Yang, H., Chin, T. et al., 2005, Yang, H., Yan, R., Chen, H., Lee, D. H. et al., 2006). However, there are limited studies reported on the pyrolysis behavior of agricultural waste/biomass at high temperature. The present study will focus on the behavior of agricultural waste materials through a Thermogravimetric Analysis combined with Mass Spectroscopy (TGA-MS) to determine the thermal degradation and gas formation at high temperatures (1450 ºC). The gas evolution products at high temperature are investigated starting at room temperature up to 1450 ºC to be further used for carbon/slag interaction in Chapter 7. Observed thermal behavior and gas evolution of coke during pyrolysis is shown in Figure 4-1 (a) (f). During pyrolysis, it was seen that coke shows a very limited weight loss, which starts around 200 ºC and then decreases slowly (Figure 4-1 (a)). Such a behavior was expected based on its composition, i.e. low volatile matter and prior history of heat treatment. Various gases species detected were CO, CO 2, CH + 3, CH 4, H 2 and HCN and N 2. H 2 started to evolve at 700 C (Figure 4-1 (d)) while water had been removed completely due to coke pretreatment (Figure 4-1 (e)). The CO peaks were observed at temperatures 450 ºC and 900 ºC, while CO 2 started to evolve around 600 ºC (Figure 4-1 (b)). Light gases such as CH 3 and CH 4 were released in the temperature range of ºC (Figure 4-1 (c)). Figure 4-1 (f), shows the HCN evolved around temperature ºC and N 2 around ºC. It is well known that HCN is one of the main NO precursors. This shows that metallurgical coke produced higher NO emissions during pyrolysis compared to palm shells, whereas TGA-MS did not detect 95

125 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends the nitrogen gas from palm shells (Chambrion, P., Orikasa, H., Suzuki, T. et al., 1997). 96

126 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends Metallurgical Coke Mass (%) (a) MC Ion Current, x (A) (d) Hydrogen H Temp. ( C) Temp (ºC) 3.0 (b) CO/ CO (e) Water Ion Current, x (A) CO Ion Current, x (A) H2O 0.5 CO Temp (ºC) Temp (ºC) Ion Current, x (A) 1.2 (c) Hydrocarbons CH CH3 Ion Current, x (A) 2.0 (f) HCN/N Temp (ºC) Temp (ºC) N2 HCN Figure 4-1 (a) (f) The mass loss of metallurgical coke with on-line MS analysis of the gas products during pyrolysis 97

127 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends Figure 4-2 (a) - (f) shows the changes in weight loss during pyrolysis of palm shells along with the gas formation products; (e.g. CO, CO 2, CH + 4, CH + 3, C 2 H 6, C 3 H 4, H 2, and H 2 O). The TGA curve of palm shells showed two degradation steps (Figure 4-2 (a)). The first weight loss is attributed to moisture release taking place at around 110 ºC. The light gas products from palm sample, including H 2, CH 4, C 2 H 6, C 3 H 4, CO, CO 2 and H 2 O are produced in the temperature range ( ºC) resulting in a significant weight loss (Figure 4-2 (a) to (f)). This is in good agreement with previous literatures that reported palm shell devolatilization occurs up to 600 ºC (Din, A. T. M., Hameed, B. H. and Ahmad, a. A. L., 2005). However, the present study goes beyond that and after 950 ºC, it is seen that the palm shell continues to lose weight indicating that gases are still being released. The proportion of H 2 O, CH 4 and higher hydrocarbons decreased considerably whereas the amount of CO and CO 2 increased to a high extent. Hydrogen showed an increase, however to a lower extent. According to Yang et. al, the evolving CO and CO 2 at low temperatures is mainly caused by breaking of double bonding (C=O) in biomass at ºC. The breaking and reforming of aromatic rings would release H 2 at higher temperature (> 400 ºC) (Haiping, Y., Yan, R., Chen, H. et al., 2007, Yang, H., Yan, R., Chin, T. et al., 2004). After 950 ºC, methane is expected to form CO and H 2 and this might be one of the reasons behind the increase seen in CO gas (Figure 4-2 (b)). Further, the reaction of water vapor with C produced from cracking the hydrocarbons (Cagigas, A., Escudero, J. B., Low, M. J. D. et al., 1987, Di Blasi, C., Signorelli, G., Di Russo, C. et al., 1999, Serio, M. A., Hamblen, D. G., Markham, J. R. et al., 1987, Worasuwannarak, N., Sonobe, T. and Tanthapanichakoon, W., 2007) present in palm shell releases CO and H 2 (Figure 4-2 (b) and (d)). The interaction of cellulose and lignin during pyrolysis was found to release H 2 and other hydrocarbons at 900 C (Hasegawa and Yang et. al. (Haiping, Y., Yan, R., Chen, H. et al., 2007, Hasegawa, I., Tabata, K., Okuma, O. et al., 2004, Yang, H., Yan, R., Chen, H., Lee, D. H. et al., 2006)). Palm shell contains high lignin (53.4 %) and cellulose (29.7 %) (Wan Daud W M A, Wan Ali W S and Sulaiman M Z 2003, Yunos, N. F. M., Zaharia, M., Ahmad, K. R. et al., 2011) and their heat treatment at high temperature releases a considerable amount of 98

128 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends gases. Moreover, the inorganic matter could also influence the formation of gas, tar and char from wood/ biomass pyrolysis at temperature 800 C (Hosoya, Takashi, Kawamoto, Haruo and Saka, Shiro, 2007). 99

129 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends Palm Shells Mass (%) (a) Palm Shells Ion Current, x (A) (d) Hydrogen H Temp. ( C) Temp (ºC) Ion Current, x (A) (b) CO CO/ CO2 Ion Current, x (A) (e) Water H2O 0.5 CO Temp (ºC) Temp (ºC) Ion Current, x (A) 1.2 (c) Hydrocarbons CH4 Ion Current, x (A) (f) Hydrocarbons C3H4 2.0 C2H6 CH Temp (ºC) Temp (ºC) Figure 4-2 (a) (f) The mass loss of palm shells with on-line MS analysis of the gas products during pyrolysis 100

130 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends The pyrolysis characteristics for palm char is shown in Figure 4-3 (a) (f). This sample is prepared for carbon/slag interactions to partially replace the coke in steelmaking (Chapter 7). The decomposition occurred slowly under the whole temperature range from ambient to 1450 ºC, a low mass loss (~40 wt. %) was observed. Two peaks of CO releasing (350 and 750 ºC) were seen at lower temperature (<600 ºC) while one big jump was found at high temperatures (>600 ºC) (Figure 4-3 (b)). The release of CO was attributed to carbonyl (C O C) and carboxyl (C=O) at low temperature and the second was attributed to thermal cracking of tar residue in the solid sample (Yang, H., Yan, R., Chen, H., Zheng, C. et al., 2006). CO 2 was released out at 350 ºC, and achieved the maximum value at 400 ºC. CO 2 release was mainly caused by the primary pyrolysis (temperature < 400 C), while secondary pyrolysis (temperature >600 C) was the main reason for release of CO and CH 4. H 2 gas was released out at a higher temperature (500 ºC) and the release of H 2 increased greatly with rising temperature (Figure 4-3 (d)). On the other hand, the release of C 3 H 4 and C 2 H 6 was generally very low and water had been removed completely (Figures 4-3 (c), (e) and (f)). The cracking of the hydrocarbons will increase the hydrogen content which is favored by a higher temperature of carbon/slag interactions (1550 ºC). The palm chars are expected to improve the carbon/ slag reactions due to the availability of gases released such as H 2, CO and CO 2 and other hydrocarbons above 400 ºC that will participate in the later stages of reaction (carbon/slag interactions). 101

131 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends Palm Char Mass (%) (a) Ion Current, x (A) (d) Hydrogen H Temp. ( C) Temp (ºC) Ion Current, x (A) (b) CO CO/ CO2 CO2 Ion Current, x (A) (e) Water H2O Temp (ºC) Temp (ºC) 2.5 (c) Hydrocarbons 2.0 (f) Hydrocarbons Ion Current, x (A) CH3 CH4 Ion Current, x 10-8 (A) C3H4 C2H Temp (ºC) Temp (ºC) Figure 4-3 (a) (f) The mass loss of palm char with on-line MS analysis of the gas products during pyrolysis 102

132 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends Effect of Blending on Combustion Combustion of carbonaceous materials involves 2 stages: an initial thermal decomposition stage, where volatiles are released (see Figure 4-4) accompanied by physical and chemical changes, followed by the subsequent combustion of gases and the formation of solid residue. The pyrolysis and combustion behavior of the palm shell, coke and the corresponding blends was investigated by using a Thermogravimetric Analysis under isothermal conditions at a fixed temperature of 1200 ºC. The weight loss curve is plotted as a function of reaction time in Figure 4-4 and the comparison of the total weight loss are showed in Figure 4-5. The total weight loss in the final stage of pyrolysis was found to be 4.9% for MC alone, reaching 11.6% for an initial addition of palm shell as in P1, 15.7% for P2 blends and increasing to 21.3% total weight loss when more palm shell was added in the blend as in P3. These results are in good agreement with the volatile matter content and previously published literature (Meesri, C. and Moghtaderi, B., 2002, Vuthaluru, H. B., 2004). The difference in weight loss may be attributed to the difference in the strength of molecular structure of the fuels. The polymers of cellulose, hemicellulose and lignin, which constitute the macromolecular structure of the biomass and agricultural waste materials, are linked together with relatively weak ether bonds (R O R, bond energy of kj mol -1 ). These bonds are less resistant to heat at low temperatures ( ºC) (Blazej, A. and Kosik, M., 1993). In contrast, the immobile phase present in the coke structure, which mostly comprises dense polycyclic aromatic hydrocarbons linked together by C=C (aromatic ring) bonds with bond energy of 1000 kj mol -1, are more resistant to the heat (Smith, L. H., and, S. L. D. and Fletcher, 1994). The coke used for the present study is expected to have a high aromatic content due to its structure. As a consequence, a small amount of fragmented polycyclic aromatic compounds are expected to result from coke. Hence, the mass loss is lower compared to samples from the palm shell in the blends. The carbonaceous materials used in this study were consumed by thermal decomposition (devolatilization), thus palm shells are expected release volatiles during 103

133 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends combustion process (see Figure 4-2) where more gases especially hydrocarbons were released. This high level volatile yield occurs over a relatively short time and is believed to influence the time for complete combustion when compared to coke. Our results are similar to the values expected on the basis of high volatile yield of blended materials, which was previously been reported by a number of authors (Hussain, A., 2006, Jones, J. M., Kubacki, M., Kubica, K. et al., 2005). Focusing attention on the initial stage of reaction (Figure 4-4), volatiles were released in the first 10 s. At 1200 ºC, the temperature of the test, the degradation time of blends decreased with increasing palm shell content in the blend possibly due to the decomposition of the lignocellulosic materials. Rates of pyrolysis are studied in the next chapter (Chapter 5), where the comparison between palm shells/ coke blend and coconut shells/ coke blend are clearly explained to demonstrate the differences in their combustion performance and structural transformations. 104

134 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends Residual Mass, % MC P1 P2 P Time, sec Figure 4-4 Weight loss curves of MC and its blends with palm shells at temperature, 1200 ºC in N 2 atmosphere 25.0 Total Weight Loss, % P3 P2 P1 MC Palm shells content in the blend Figure 4-5 Comparison of total weight loss curve of MC and its blends with palm shells at temperature, 1200 ºC in N 2 atmosphere 105

135 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends The ash tracer method was used to indicate the combustion performance of the present carbonaceous materials. On the basis of measured ash and carbon content of the samples, ash and carbon content of the combusted residue, the burnout of coke and its blends with palm shell has been estimated using Equation 3-1. The calculation of the samples burnout was repeated at least five times for consistency and reproducibility. Errors in quantitative determination of combustion efficiency were minimized by repeatability tests and were generally less than 0.3% (Appendix A). Figure 4-6 compares the combustion efficiencies of MC and its blends with palm shell in different proportions (from P1 up to P3 blends) at a temperature of 1200 ºC using a mix gas composed of 80% N 2 and 20% O 2. The burnout of coke/palm shell blends are higher than the burnout of MC, increasing with each stepwise addition of agricultural wastes in the blend. Coke burnout was estimated to be 10% while 100% palm shells attained 14.3% (Figure 4-6). By replacing coke with palm shell content in the blends, an increase in the burnout is seen. The burnout of palm shell blends was seen to increase gradually, such an increase is seen for the blends with palm shell of up to P3 blends (P1 = 16.5 %; P2 = 19.3 %; and P3 = 23.5 %) respectively. The combustion performance of the parent materials is lower than the combustion performance of their corresponding blends indicating that the volatile matter and carbon content require optimum concentrations to improve the combustion performance. 106

136 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends Burnout, % MC P1 P2 P3 Raw palm shells Palm shells content in the blend Figure 4-6 Effect of blending MC with varying palm shell content in the blends on the combustion at 1200 C in the presence of 80% N 2 ; 20% O 2 Variation in the combustion performance can be attributed to some extent to differences in chemical properties of coke and its corresponding blends, such as volatile matter, mineral matter and carbon structure. Proximate analysis of the samples indicated an increase in VM with increasing palm shell content in the blend which is expected to increase the combustion performance (Figure 4-7). The presence of palm shell in the blends may had an impact upon the structure during the release and combustion of their high VM and hence increase in char burnout. The modification of structure and high VM from previous study (polypropylene and rubber) also showed high combustion performance (Gupta, S., Sahajwalla, V. and Wood, J., 2006, Zaharia, M., Sahajwalla, V., Kim, B.-C. et al., 2009). 107

137 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends Volatile Matter, % (a) 71.3% 25.0% 19.3% 6.1% 12.5% MC P1 P2 P3 Palm Shell Burnout, % (b) P3 P2 P1 MC Palm shells Volatile Matter, % Figure 4-7 Effect of volatile matter on (a) blend composition and on (b) the combustion performance of palm shell/mc blends 108

138 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends Previously, a study performed by Hussain et al., Moghtaderi et al. and Vamvuka et al. on the combustion of palm shell waste blends with coal by means of thermogravimetric analysis, in the temperature range ºC, revealed similar findings where the combustion performance increases when the proportion of palm shell blends increases (Hussain, A., 2006, Moghtaderi, B., Meesri, C. and Wall, T. F., 2004, Vamvuka, D., Kakaras, E., Kastanaki, E. et al., 2003). By blending the two fuels in proportion of agricultural wastes and coal, the combustion characteristics were attributed to the coupling effect of agricultural wastes and coal. Present observations and studies reported in the literature agree on the influence on the combustion performance brought by the high volatile content and mineral matter has an effect on the dominant combustion of palm shell wastes (Agblevor, F. A. and Besler, S., 1996, Baxter, L. L., Miles, T. R., Jenkins, B. M. et al., 1998, Edye, L. A., 1992). Devolatilization starts at low temperatures and this indicates the process is expected to take place as soon as the material is exposed to high temperature environment. The volatiles consist mainly of combustibles and significant amount of energy is released during gas phase reactions. Agricultural waste materials have a high alkaline content which can easily vaporize at combustion temperatures (Bryers, R. W., 1996, Korbee, R. E., S.; Heere, P.G.T.; Kiel, J.H.A., 1998, Zevenhoven-Onderwater, M., Backman, R., Skrifvars, B. J. et al., 2001). Potentially, this can lead to a large error if the conventional ash tracer method is employed for the burnout measurement. To avoid such error, the extent of mass conversion during devolatilization and combustion of agricultural waste materials was determined using an inorganic tracer method (Meesri, C., 2003). Table 4-1 indicates that the amounts of the inorganic species present in the fullycombusted MC/agricultural waste blends obtained from the measurement. Table 4-1 shows that the amount of oxides still exist after combustion. Approximately 2 % of oxides still exist in palm shells. This confirms the conserved-mass principle of the hardto-vaporise inorganic elements of coke and agricultural waste during combustion of their blends in the range of blending ratios investigated. More importantly, it highlights the suitability of the modified inorganic tracer method in quantifying the accurate values of mass burn-off during the combustion of MC/agricultural waste blends. 109

139 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends Table 4-1 XRF analysis showing the amounts of major inorganic elements (based on their oxides) present in the ash of the coke, palm shell blends and palm shells following treatment at 1200 C under 20% oxygen and 80% nitrogen atsmosphere Material Amount of SiO 2 (%w/w of raw fuel) Amount of Al 2 O 3 (%w/w of raw fuel) Amount of TiO 2 (%w/w of raw fuel) Metallurgical Coke Palm shell P P P

140 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends Physical Properties and Structural Transformations The burnout improvement of the blends could be attributed to the modification of the physical structure of the resulting char. Physical chracterization of the samples were performed using quantitative (BET surface area) and qualitative analyse (SEM) Surface Area Measurements BET Surface Area Figure 4-8 shows the graph of micropore surface area as a function of carbonaceous material and reaction stage. The results demonstrated minor changes in the BET surface area for coke, from 31.6 to 34.9 m 2 / g. The surface area of the MC did not change significantly. The palm shell sample has a microporous structure with small a value of surface area of m 2 / g. On the other hand, the palm shell blends (Figure 4-8) showed significant differences before and after reaction samples values fluctuating from 9.4 to 24.2 m 2 / g for P1 blends; from 2.7 to 21.9 m 2 / g for P2 blends and from 1.5 to 12.0 m 2 / g for P3 blends. The differences in surface area before and after reaction in the DTF are attributed to pores opening up with volatile matter evolution. The change in BET surface area of the blend char samples was higher compared to the change in surface area of MC. 111

141 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends BET Surface Area (m 2 /g) Before DTF After DTF 0.0 MC P1 P2 P3 Palm shells content in the blend (%) Palm shells Figure 4-8 Changes in micropore surface area for palm shell before and after combustion in DTF as a function of the palm shell content in the blends The changes in surface area before and after combustion have been estimated for coke and its blends in different proportions with palm shell and an index of surface area change has been defined as ΔSA. Figure 4-9 shows that the change is more significant in the case of palm shell blends (P1 blends = 61.2 %; P2 blends = 87.6 %; P3 blends = 87.5% and raw palm shells = %), while coke (coke = 9.5 %) shows the least change. This supports the fact that for palm shell blends, the combustion shows an increase compared in coke (see Figure 4-6). 112

142 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends P2 P3 Palm shells Δ Surface Area (%) P1 MC 0 Palm shells content in the blend (%) Figure 4-9 The changes in micropore surface area for palm shell wastes content in the blend The images of raw sample are presented in Figures 4-10 and 4-11 and the structural transformations after combustion are studied by using a scanning electron microscope (SEM). Metallurgical coke (MC), is produced by the carbonization of coals or coal blends at temperatures up to 1127 C to produce a macroporous carbon material of high strength and relatively large lump size. From Figure 4-10 (a), metallurgical cokes are grey, hard, and porous and may appear glassy in some samples. After reactions in the gas phase (Figure 4-10 (b)), the physical structure of MC still has similar features as the raw coke due to prior treatment of MC. Carbon texture and microstructure are two main factors that affect reactivity. 113

143 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends (a) (b) 0.05 mm 0.05 mm Raw MC MC after combustion in DTF Figure 4-10 Images of (a) raw MC and (b) MC after combustion in gas phase at 1200 ºC (80% N 2 ; 20% O 2 ) Meanwhile, in Figures 4-11 (a) and (b) show the palm shell samples used for gas phase reactions. Palm shells, an agricultural waste by-product from palm oil mills share the same characteristics as its other sibling, coconut shells. They both possess a highly complex pore structure and fibre matrix (Panagiota P., Dimitrios K. and D., E., 2008). In Figure 4-11 (a), the outer layer of the palm shells seen to be fibrous with dark brown color. After gas phase reactions, palm shell char turned black in color with high microporosity structure (Figure 4-11 (b)). 114

144 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends (a) (b) 0.05 mm 0.05 mm Raw Palm Shells Palm Shells after combustion in DTF Figure 4-11 Images of (a) raw palm shells and (b) palm shell after combustion in gas phase reactions at 1200 ºC (80% N 2 ; 20% O 2 ) Structural Transformations - Scanning Electron Microscopy (SEM) Further investigations using SEM have been conducted to describe the morphological characteristics of MC and palm shell wastes. A comparison between the residual chars and their corresponding raw materials is also presented to draw conclusions on the structural changes after the combustion stage. Figure 4-12 shows the morphology of 100% metallurgical coke with residual ash appearing as bright white spots (Figure 4-12 (a) and (b)). MC has high ash content in its composition (see Table 3-1), which might accumulate at the receding surfaces of the burning char impeding pore enlargement. Consequently, oxygen diffusion is delayed, and a lower combustion performance is expected. The cross-sections of coke particles before and after combustion in DTF are taken to support the above statement and are illustrated in Figure 4-12 (c) and (d), respectively. The morphology of the raw coke shows a very porous matrix characterized by bright well defined edges (Figure 4-12 (c)). Following combustion, no significant changes are seen (Figure 4-12 (d)) in the coke particle, which is also to be expected based on the BET surface area measurements presented in Figure 4-8 and the amount of volatiles available in the raw sample. 115

145 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends Figure 4-12 SEM micrograph of (a) raw MC, (b) MC after combustion at 1200 C in DTF, (c) cross-sectional image of raw MC and (d) cross-section of MC char Typical SEM micrographs of palm shell and its corresponding char obtained at 1200ºC in O 2 and N 2 mixed atmosphere are illustrated in Figure 4-13 (a) (d). Ordered compact cells seem to define the raw palm shell structure with small pores dispersed throughout (Figure 4-13 (a)). The cross-section of the shell shows isodiametric polygonal flattened cells with well-defined lumen arranged in a honeycomb like pattern (Figure 4-13 (a) and (c)). The combustion stage is expected to cause noticeable changes in the shells' structure. Microphotographs of the external surfaces of the char obtained at 1200 C (Figure 4-13 (d)) indicate progressive destruction of the cell lumen and wall rupture during combustion. Removal of cell contents and consequently, the opening-up of cellular structures give rise to distinctive features with porous networks, attributed to the evaporation of volatile material (Figure 4-13 (c)). Cetin, et al. (Cetin, E., Moghtaderi, B., Gupta, R. et al., 2004), while studying about the effect of pyrolysis in 116

146 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends the agricultural char structure, found that the cell structure practically did not exist after devolatilization. They observed that the lack of cell structure in high heating rate chars could be assigned to melting of the cell structure and the occurrence of plastic transformation. Figure 4-13 SEM micrograph of (a) raw palm shell, (b) palm shell after combustion at 1200 C in DTF, (c) cross-sectional image of raw palm shell and (d) cross-section of palm shell char 117

147 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends The Role of Chemical Properties and Carbon Structures Chemical Structures The chemical structure and the major organic components in agricultural wastes play an important role in gas phase reactions. The basic components present in agricultural waste can be classified as cellulose, hemicelluloses and lignin (Figure 4-14). Cellulose is a polysaccharide with a general formula ([C 6 (H 2 O) 5 ] n ) and an average molecular weight in the range of 300, ,000 (g/mol). The cellulose molecule has a crystalline structure, strong, and resistant to hydrolysis. In contrast, hemicellulose ([C 5 (H 2 O) 4 ] n ) has a random, amorphous structure with little strength, while lignin ([C 10 H 12 O 3 ] n ) consists of a highly three dimensional cross-linked structure. In terms of stability to thermal degradation, the most stable is lignin, followed by cellulose while the least stable structure is found in hemicelluloses (Demirbas, A., 2001, Fengel, D., 1983). The cellulose and hemicellulose content in the palm shell materials (Figure 4-14) may enhance the combustion characteristics. The decomposition of cellulose compounds that have structures of branched chains of polysaccharides with no aromatic compounds can be easily volatilized. Cellulose is the most abundant volatile species, and this is due to the to the relatively low degradation temperature compared to hemicellulose and lignin. In the present study, palm shell contains 29.7% cellulose, 16.9% hemicellulose and 53.4% lignin. The cellulose content in the palm shell is expected to diminish after combustion where the XRD, NMR and FTIR spectra showed the results in next sections. While coke sample contains mainly aromatic carbon, with strong bonds that resistance at high temperatures, and therefore higher thermal stability. Two successive steps are occurring during combustion of palm shells. First, the cellulose components are volatilized (low thermal degradation), thus the developed char porosity increases and that oxygen easily diffuses into the particle pores and channels. Next, the lignin components can also react with the diffused oxygen although the reactivity of lignin itself is low (Asri, G. and Ichiro, N., 2007). In other words, the char 118

148 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends morphology and the properties of various components will be some of the most important factors controlling palm shells reactivity during combustion. MC Allotropes of Carbon Palm Shell Cellulose 29.7 Hemicellulose 16.9 Lignin 53.4 Figure 4-14 Cellulose, hemicellulose and lignin contents of palm shell (Wan Daud, W. M. A. and Wan Ali, W. S., 2004) 119

149 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends Chemical Properties X-ray Diffractions The structural modifications induced by blending coke with palm shells have been characterized by X-ray diffraction using the untreated blends and their corresponding heat treated chars. Coke is known to contain short range graphite like structure, i.e. crystalline carbon (2θ = 26 ), of the order of nanometers (~2nm). The crystalline carbon has an intermediate structure between graphitic and amorphous, so called turbostatic structure or random layer structure, as suggested by Biscoe and Warren (Biscoe, J. an d Warren, B. E., 1942). In addition to the graphite like structure, coke displays disordered material, giving rise to high background intensity. The XRD profiles of raw MC and MC after combustion in DTF are displayed in Figure 4-15 and the peak characterization is presented in Table 4-2. Figure 4-15 shows the XRD patterns of raw MC. The carbon reflection peaks were observed in at 2θ =26.5º and 42.4º, corresponding to (002) and (102) bands respectively. The (002) band of carbon is attributed to stacking of aromatic layers, while the (102) band arise from the inplane structure of the aromatics. Minerals identified in metallurgical cokes by previous studies are quartz (SiO 2 ), iron oxides, mullite (Al 6 Si 2 O 13 ), fluorapatite (Ca 5 (PO 4 ) 3 F), pyrrhotite (Fe1-xS), brookite (TiO 2 ), anatase (TiO 2 ), cristobalite (SiO 2 ), alkali feldspars ((K, Na)(AlSi 3 O 8 )) and aluminosilicates (Grigore, M., Sakurovs, R., French, D. et al., 2008). The main minerals can be found in raw MC are quartz (SiO 2 ), mullite [Al 6 Si 2 O 13 ; 2θ =33.1º (022)], calcite [CaCO 3 ; 2θ =29.3º (002)] and iron oxides [Hercynite; 2θ =21.6º (200)] (see Table 4-2). Slightly changes can be seen after combustion and detecting that calcite peak at 29.3 º and iron oxides at 21.6 º are visible. 120

150 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends Raw MC SiO 2 FeAl O 2 4 C Ca 2 CO 3 Al 6 Si 2 O 13 SiO 2 C, SiO 2 SiO 2 C MC after DTF C C C FeAl 2 O 4,SiO 2 SiO 2 Intensity (counts) SiO 2 Al 6 Si 2 O 13 SiO 2 C, SiO 2 SiO 2 C FeAl 2 O 4,SiO 2 SiO Scattering angle (2θ) Figure 4-15 XRD patterns of metallurgical coke before and after combustions (T=1200 ºC with 20% O 2 ; 80% N 2 ) 121

151 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends Table 4-2 XRD peaks characteristic of raw MC and MC after combustion in DTF 1200 ºC with atmosphere (80% N 2 ; 20% O 2 ) Sample Raw MC MC after DTF Peak Position (2θ) d- spacing (Å) Crystal Size h k l Possible Compound (chemical formula) Herynite (FeAl 2 O 4 ) Graphite (C) Calcite (CaCO 3 ) Mullite (Al 6 Si 2 O 13 ) Graphite (C) Quartz (SiO 2 ) Graphite (C) Mullite (Al 6 Si 2 O 13 ) Graphite (C) Quartz (SiO 2 ) Palm shells are characterized as an amorphous material. In Figure 4-16 and Table 4-3 showed the raw palm shell has a peak designated as μ band, observed at 2θ = 15.8º, which may be attributed to the packing of the structure such as aliphatic side chains or condensed saturated rings, or the adjacent chains of linear polymer (Ko, T.-H., Kuo, W.-S. and Chang, Y.-H., 2001, Lu, L., Kong, C., Sahajwalla, V. et al., 2002). The peaks detected the basic compounds in agricultural wastes which is cellulose (2θ = 15.8 and 29.4º). As expected, due to long branch of carbon ring which is hemicellulose and lignin, the peak does not appear since they are amorphous structure. The other oxides elements such as sodium, aluminum and silicate [2θ = 34.7º (132)] also detected. The aliphatic side rings are the volatiles contained in the carbonaceous material matrix which will be transported out of the carbon particle during the combustion processes. At high reaction temperatures, the amorphous carbonaceous constituents and the reactive sites, diminish via oxidation and graphitization of the carbon matrix (Lu, L., Sahajwalla, V. and Harris, D., 2000). It can be observed that (Figure 4-16) the amorphous carbonaceous constituents in the raw palm shells undergo conversion during 122

152 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends combustion. The amorphous diffused peak is decreased as combustion proceeds and this attributed to the decrease in the amount of carbon active sites and breakdown the cellulosic structure (cellulose). This is due to the consumption of volatiles and structural ordering of their carbon matrices. From a structural point of view, lignin occupies the strongest structure due to the methylene units (-C 2 H 2 -) and π bonds that constitute the aromatic ring (Antal, M. J., Jr. and Varhegyi, G., 1995, Graboski, M. and Bain, R., 1981, Shafizadeh, F., 1982). When agricultural materials degraded at high temperature, lignin is modified to aromatic structure. 123

153 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends Raw Palm Shells Cellulose (Cn(H 2 O)n) C Cellulose Na 6, Al 4 Si 4 O 17 SiO 2 C (Cn(H O)n) 2 SiO 2 Intensity (counts) Palm Shells after DTF SiO 2 (Cn(H O)n) 2 C (Cn(H 2 O)n) Na Al Si O SiO 6, SiO 2 C (Cn(H O)n) 2 SiO Scattering angle (2θ) Figure 4-16 XRD patterns of palm shells before and after combustions (T=1200 ºC: 20% O 2 ; 80% N 2 ) 124

154 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends Table 4-3 XRD peaks characteristic of raw palm shells and palm shells after combustion in DTF 1200 ºC with atmosphere (80% N 2 ; 20% O 2 ) Sample Raw palm shells Palm shells after DTF Peak Position (2θ) d- spacing (Å) Crystal Size Possible Compound (chemical formula) h k l Cellulose [C 6 (H 2 O) 5 )]n Graphite (C) Cellulose [C 6 (H 2 O) 5 )]n Anorthite (Ca, Na)(Si, Al)O Carbon (C n (H 2 O) n ) Quartz (SiO 2 ) Quartz (SiO 2 ) Graphite (C) Anorthite (Ca, Na)(Si, Al)O Carbon (C n (H 2 O) n ) Quartz (SiO 2 ) Qualitatively, Figures 4-17 to 4-19 shows that, with increasing the combustion temperature, the (002) peak position shifts from 24.4º to 26.5º, closer to that of graphite (26.6º) and all the palm shell/coke blends peaks become more sharper and intense. In general, diffused and broad bands in XRD patterns represent the existence of short range order in the carbon structure, while sharp and narrow peaks correspond to highly crystalline with high degree of long-range order. In the palm shell blends, the cellulose peak had been detected and diminishes by high temperature reactions compared to raw coke (Figure 4-15). 125

155 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends Raw_P1 blends C Cellulose SiO 2 C P1 blends after DTF C CaCO 3 C SiO 2 C ; SiO SiO 2 2 C FeAl 2 O 4,SiO 2 SiO 2 Intensity (counts) SiO 2 FeAl O 2 4 C CaCO Al Si O C C SiO 2 C ; SiO SiO 2 2 C FeAl 2 O 4,SiO 2 SiO Scattering angle (2θ) Figure 4-17 X-ray diffraction analysis of coke/palm shell blends; P1 before and after combustion temperature 1200 ºC in 20% O 2 and 80% N 2 atmosphere 126

156 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends Raw_P2 blends C Cellulose SiO 2 FeAl O 2 4 C CaCO 3 C SiO 2 C ; SiO SiO 2 2 C FeAl 2 O 4,SiO 2 SiO 2 Intensity (counts) P2 blends after DTF C SiO 2 FeAl O 2 4 C CaCO 3 Al 6 Si 2 O 13 C C SiO 2 C ; SiO SiO 2 2 C FeAl 2 O 4,SiO 2 SiO Scattering angle (2θ) Figure 4-18 X-ray diffraction analysis of coke/palm shell blends; P2 before and after combustion temperature 1200 ºC in 20% O 2 and 80% N 2 atmosphere 127

157 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends Raw_P3 blends C Intensity (counts) Cellulose SiO 2 FeAl O 2 4 C CaCO 3 Al 6 Si 2 O 13 C SiO 2 C ; SiO SiO 2 2 C FeAl 2 O 4,SiO 2 SiO 2 P3 blends after DTF C SiO 2 FeAl O 2 4 C CaCO 3 Al 6 Si 2 O 13 C C SiO 2 C ; SiO SiO 2 2 C FeAl 2 O 4,SiO 2 SiO Scattering angle (2θ) Figure 4-19 X-ray diffraction analysis of coke/palm shell blends; P3 before and after combustion temperature 1200 ºC in 20% O 2 and 80% N 2 atmosphere 128

158 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends When the samples were combusted in DTF, changes in coke and its palm shell blends structures become evident (Figures 4-17 to 4-19). This result has significant implications regarding the differences in the combustion performance of the derived solid residues which is discussed in the early section of this chapter. Volatile consumption in the high temperature process brought changes in the structure of the residual particle, increasing the burnout of the blends. A previous study (Gupta, S., Sahajwalla, V., Chaubal, P. et al., 2005, Lu, L., Sahajwalla, V., Kong, C. et al., 2001, Sahajwalla, 2006, Zaharia, M., Sahajwalla, V., Kim, B.-C. et al., 2009) on combustion of plastics/rubber/coke blends implied the possible contribution of the polymeric materials modifying the coke/char structure due to the heat released by the combustion of polymeric volatiles and increasing the burnout. A summary of Figures 4-17 to 4-19 are show in Table 4-4. The peak emerging from palm shell/coke blends (P1, P2 and P3 blends sample) at 2θ = 21.9 º (100), 43.1º (103) and 50.1º (113) strongly suggest the existence of the crystalline compounds of SiO 2 (quartz), the most abundant oxide as the ash analysis showed. The peaks situated at 2θ = 12.8º and correspond to carbon ring of cellulose [C 6 (H 2 O) 5 )]n compound in the palm shell blends compared to raw coke. 129

159 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends Table 4-4 XRD peaks characteristic for all samples P1, P2 and P3 blends before and after combustion in DTF 1200 ºC with atmosphere (80% N 2 ; 20% O 2 ) Sample Raw P1, P2, P3 blends P1, P2, P3 blends after DTF Peak Position (2θ) d- spacing (Å) Crystal Size h k l Possible Compound (chemical formula) Cellulose [C 6 (H 2 O) 5 )]n Quartz (SiO 2 ) Graphite (C) Calcite (CaCO 3 ) Carbon (C n (H 2 O) n ) Quartz (SiO 2 ) Graphite (C) Calcite (CaCO 3 ) Mullite (Al 6 Si 2 O 13 ) Carbon (C n (H 2 O) n ) Quartz (SiO 2 ) 130

160 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends Chemical Bonding Fourier Transform Infrared Spectroscopy (FTIR) During the combustion process, the high tempearture and the presence of oxygen leads to possible disruption in the chain molecules. The intermolecular strenght may be broken if the bonds are weaker and the structure is simpler (Sahajwalla, V., Zaharia, M., Kongkarat, S. et al., 2009). The organic structure of coke can be regarded as consisting of heterogenous aromatic structures, while sulphur, oxygen and nitrogen are also presented in functional groups. Fourier transform infrared (FTIR) spectroscopy is a widely used analytical technique for determining the different functional groups of a carbonaceous structure. This method, is able to reveal carbohydrogenated structures (aromatic and aliphatic) and heteroatomic functions (mainly oxygenated). Moreover it can detect the presence of minerals and is currently considered to be the most powerful techniques for coke and biomass characterization (Cloke, M., Lester, E. and Thompson, A. W., 2002, Guo, Y. and Bustin, R. M., 1998). FTIR spectras of raw and heat treated coke are presented in Figure 4-20 while the characterization of the peaks are shown in Table 4-5. MC shows intense bands at 3,402 3,416 cm 1 and this attributed to OH stretching from H 2 O or phenol groups. In the aliphatic stretching region (3,000 2,800 cm 1 ), distinct peaks are observed at 2,851 cm 1 and 2,918 2,926 cm 1, attributed to symmetric and asymmetric CH 2 stretching, respectively. A great abundance of C=O and C O R structures are noted, as revealed by the intensity of the peaks in the 1,800 1,000 cm 1 region. This zone of oxygen-containing functional groups is characterized by a very intense peak at 1,618 1,622 cm 1, which is attributed either to C=O or C=C aromatic ring stretching. The aromatic ring (C=C) stretching was seen in the after combustion sample (Table 4-5). However, the stretches due to C=O in chars showed a continues decrease in intensity with temperature, indicating a loss of these functionalities at high temperature. These bands changed into single bonds (C-H and C-C). The raw MC peaks were seen at 1,032 1,047 cm 1 may also result from silicate minerals (Si O bonds) as the amount of 131

161 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends SiO 2 in the ash sample is significant (Table 3-3). A very prominent band at 893 cm 1 could be related to aromatic out-of-plane C C deformations % Transmittance Moisture CH 2 stretching C=O C=C C-CH 3 bending C-C alkene C-CH bending 3 C C bond SiO 2 Raw MC Moisture CH 2 stretching O-H stretching C=C C-CH bending 3 C-C stretching C C bond 85 MC after DTF Wavenumbers (cm -1 ) Figure 4-20 FTIR spectra of metallurgical coke before and after combustion in DTF at temperature 1200 ºC in 20% O 2 and 80% N 2 atmosphere 132

162 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends Table 4-5 Characterization of FTIR peak profiles of raw MC and MC after combustion Peak location range (cm -1 ) at 1200 ºC at atmosphere 80% N 2 ; 20% O 2 Possible Functional Groups Raw MC MC after combustion O-H stretching C-H asymmetric and symmetric stretching in methyl and methylene group C=O stretching in acetyl and uronic ester groups or in carboxylic group of ferulic and coumaric acids C=C, symmetric aromatic O-Hydroxyl diaryl ketones N-H bending in primary amine Aromatic rings (lignin) CH 3 (aliphatic) C-H rocking in alkenes or C-H stretching in methyl and phenolic alcohol Si-CH 2 stretching in alkenes or C-C plus C-O plus C=O stretching C-C alkenes C-O deformation in secondary alcohol and aliphatic ether or aromatic; C-H in plan deformation 1060 plus C-O deformation in primary alcohol C C alkynes

163 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends The presence of water impurities and other O-H bonding are present in the raw palm shell spectra (Figure 4-21 and Table 4-6). The absorbance peaks stretching corresponded to O-H observed at 3600 and 3200 cm -1. Of particular interest are the bands corresponding to the OH group as well as their vibrations relative to other bands such as aromatic CH 3 stretches. Sharp peaks of medium-to-strong intensity are usually found in the region cm -1 corresponding to aromatic CH 3 functional groups while the overlapping peaks between 1000 and 1300 cm -1 suggested to be due to the presence of alcohols and phenols. The absorption peaks of raw palm shells in the region 1780 and 1640 cm -1 represent C=O stretching vibration indicating the presence of ketones and aldehydes. The presence of both O-H vibration and C=O stretching vibration may also reveal the presence of carboxyclic acids. The possible presence of alkenes (C=C) is indicated by the absorbance peaks situated between 1680 and 1580 cm -1. The absorption peak between 900 and 650 cm -1 might correspond to single polycyclic and substituted aromatic groups. These functional groups and the composition have been identified in the pyrolytic derived oil from palm shells when using a fixed bed reactor (Ani, F. N., 1997, Islam, M. N., Zailani, R. and Ani, F. N., 1999, Williams, P. T. and Horne, P. A., 1995). In Table 4-6, the bond stretching of C=O became visible and the stretch of aliphatic rings (1459 cm -1 ) was lost after combustion. This also indicated a loss of the OH group and an enhancement of the aromatic nature with increase in temperature. The band due to OH stretch (1372 cm -1 ) was almost missing in the combusted palm, indicating that the char had completely lost OH groups. The increased aromatization results in a polyaromatic structure of the palm char and a decrease in the intensity of the aromatic CH stretch as the concentration of the carbon atoms at the surface decreases. The result from XRD spectra from palm char also showed consistency where the decrease of intensity for cellulose peak and more carbon ring (aromatic) peaks increases were observed (Figure 4-16). It showed that all aliphatic rings are consumed after combustion. A similar observation was made by Boon et al. (Boon, J. J., Pastorova, I., Botto, R. E. et al., 1994) where the aromatic character was observed. 134

164 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends % Transmittance Moisture CH 2 stretching C=O O-H stretching N-H stretching C-CH 3 bending Aromatic (lignin) C-C stretching C-C alkenes C-CH bending 3 C-1 ring Raw Palm Shells Moisture CH 2 stretching O-H stretching N-H stretching C-CH 3 bending Aromatic (lignin) C-O C-CH bending Wavenumbers (cm -1 ) Palm shells after DTF Figure 4-21 FTIR spectra of palm shells before and after combustion in DTF DTF at temperature 1200 ºC in 20% O 2 and 80% N 2 atmosphere 135

165 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends Table 4-6 Characterization of FTIR peak profiles of raw palm shells and palm shells after combustion at 1200 ºC at atmosphere 80% N 2 ; 20% O 2 Peak location Raw Palm Palm Shells range (cm -1 Possible Functional Groups ) Shells after DTF O-H stretching C-H asymmetric and symmetric stretching in methyl and methylene group C=O stretching in acetyl and uronic ester groups or in carboxylic group 1736 of ferulic and coumaric acids C=C, symmetric aromatic O-Hydroxyl diaryl ketones N-H bending in primary amine Aromatic rings (lignin) CH 3 (aliphatic) C-H rocking in alkanes or C-H stretching in methyl and phenolic 1372 alcohol Si-CH 2 stretching in alkane or C-C plus C-O plus C=O stretching C-C alkenes 1166 C-O deformation in secondary alcohol and aliphatic ether or aromatic; C-H in plan deformation plus C-O deformation in primary alcohol C-1 groups frequency

166 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends Carbon Structures Nuclear Magnetic Resonance ( 13 C NMR Spectroscopy) 13 C NMR spectroscopy was used to identify the various carbon groups present in coke, agricultural waste materials and the corresponding combusted char samples. The present study investigates the influence of high temperature and oxidising atmosphere on the residual mass of coke and agricultural wastes by comparing the carbon groups before and after combustion. The 13 C NMR spectra acquired for coke before and after combustion are shown in Figure 4-22 (a) and (b) and signal assignments data are showed in Table 4-7. The comparison between the spectra shows only the slight modifications. The spectra can generally be divided into two main chemical regions: aliphatic carbons (0 90 ppm) and aromatic carbons ( ppm), including carbonyl/carboxyl and phenolic groups. Figure 4-22 (a) shows aromatic C group, which are aromatic C H groups (124 ppm) and oxygenated structures C O (151 ppm). 137

167 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends (a) Protonated aromatic carbon Raw MC -O C O Aliphatic C aromatic [ppm] (b) Protonated aromatic carbon MC at 1200ºC * aromatic * [ppm] Figure 4-22 CP/MAS 13 C NMR spectra of (a) raw MC and (b) MC at 1200ºC in a drop tube furnace [80% N 2, 20% O 2 ] (* spinning sideband) 138

168 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends Table 4-7 Signal assignments for CP/MAS 13 C NMR of MC (Erdenetsogt, B.-O., Lee, I., Lee, S. K. et al., 2010) Chemical shift (ppm) Assignment Ketone, quinine, aldehyde C=O, HC=O Carboxyl, ester, quinone COO, COOH O-substituted aromatic C-O, C-OH Aromatic CH, C Aromatic CH Sacchride, alcohol, ether CHOH, CH 2 OH Methoxy, methyne, quaternary CH 3 O-, CH-NH, CH, C Methylene CH Methyl CH 3 The spectrum shows a broad peak at 32 ppm with two small humps centered at around 25 ppm, which can be assigned to CH 2 attached to the aromatic rings and complex aliphatic carbon (Pan, V. H. and Maciel, G. E., 1993). However, after high temperature gas phase reaction, coke (Figure 4-22 (a)) showed the destruction of aliphatic groups and the formation of alcohols, ethers, and carbonyl groups with more aromatic C and CH at 128 ppm. The agricultural waste materials show spectra scan within relatively a short time (maximum number of scans was 208). While for the char sample, the number of scans was in the range of , due to the fact that chars are amorphous. Agricultural waste samples show a variety of carbon resonances, while those of the char samples reveal only a single resonance. Figure 4-23 and Table 4-8 summarizes the signal assignments obtained from the NMR spectra of the palm shell samples. Figure 4-23 (a) shows raw palm shell shows a peak at 22 ppm assigned to the acetyl methyl groups of hemicelluloses. At 56 ppm, the methoxy group of lignin is present. The peak at 65 ppm is assigned to the aliphatic C-6 carbons of crystalline cellulose. In 139

169 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends the region of 72 and 75 ppm, the palm shells have the highest peak, corresponding to the C-2, C-3, and C-5 carbons of cellulose. The raw palm shells show C-4 amorphous and crystalline cellulose at 84 and 89 ppm, respectively, while at 105 ppm a sharp peak appears indicating the presence of anomeric carbon of sugars. The signals in the region of 116 ppm represent the lignin quaiacyl C-3 and C-5 only detected in palm shell. Lignin syringyl C-1, syringyl C-5, and quaiacyl C- 2 are also detected in raw palm shell samples exhibiting a broad line in the region between 134 and 137 ppm. The signals detected 153 ppm are assigned to syringyl C-3 and syringyl C-5. In the case of palm shells, the signal at 162 ppm corresponds to 4- hydorxyphenyl C-4. A peak of carbonyl groups (COOH) of hemicelluloses is seen at 173 ppm. These groups are well known to have a high reactivity (Scholze, B., Hanser, C. and Meier, D., 2001). Figure 4-23 (b) shows the NMR spectra of the residual chars derived from the combustion of palm shell materials. The palm shells prepared at 1200 C (80% N 2, 20% O 2 ) show one peak at ppm assigned to aromatic lignin (124 ppm-issuing from polyaromatic hydrocarbons). The aromaticity of the chars increased due to the loss of aliphatic groups. The broad resonances in Figure 4-23 (b) demonstrate the amorphous nature of the chars due to the presence of free radicals and the complex structure of this highly dipolar coupled materials. Different authors have reported the changes in composition of woody materials following thermal treatment (Inari, G. N., Mounguengui, S., Dumarçay, S. et al., 2007). New carbon groups were seen to evolve. However, at the end temperatures of pyrolysis (600 C), only aromatic groups were left behind (Bardet, M., Hediger, S., Gerbaud, G. et al., 2007, Freitas, J. C. C., Emmerich, F. G. and Bonagamba, T. J., 2000, Ramesh K.Sharma, 2000). Sharma et al. (Sharma, R. K., Wooten, J. B., Baliga, V. L. et al., 2004) showed that the amount of aromatic carbons in a lignin char is less than in the original sample material. Less changes were seen in the residual coke samples while the palm shell waste materials showed significantly different spectra following the high temperature reaction 140

170 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends in the DTF. The loss of lignocellulosic structure present in palm shells (Figure 4-23) might lead to higher combustion peformance because the molecular structures are easier to break compared to coke carbon structure which is more aromatic (Figure 4-22) and the bonds are stronger. In Figure 4-24, the comparison between palm shells and MC had shown a similar results by using different analytical tools such as XRD, FTIR and NMR. Palm shells showed that all the aliphatic groups (stuctures) had been removed completely after combustion while coke with aromatic structures are still left behind. The mechanisms governing during the breakdown of the structures had completely discussed before. These results are consistency to show that palm shell blends had better combustion performance compared to coke alone due to its chemical structures. 141

171 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends (a) Cellu lose C-2, C-3, C-5 Raw palm shells Acetyl carbonyl groups of hemicellulose Ketones O R-C-R O -C-O O -C-OCH3 Oxygen substituted (-OCH and OH) aromatic carbons Lignin Lignin aromatic Aromatic C-1 carbon of sugars (cellulose) O -C-O C-4 carbon CH3 cr am C-6 carbon crystalline cellu lose Methoxy group Lignin:-OCH3 Acetyl methyl groups of hemicellu lose CH3 CH2 in saturated aliphatic chain Aliphatic (b) Aromatic -C [ppm] Palm shells at 1200ºC * aromatic [ppm] Figure 4-23 Spectrum of CP/MAS 13 C NMR of (a) raw palm shells and (b) palm shells char at 1200ºC in a drop tube furnace [80% N 2, 20% O 2 ] (* spinning sideband) 142

172 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends Table 4-8 Resonance assignments for CP/MAS 13 C NMR spectra of agricultural wastes (Bardet, M., Hediger, S., Gerbaud, G. et al., 2007, Link, S., Arvelakis, S., Spliethoff, H. et al., 2008) Chemical shift (ppm) Assignment COOH, principally acetyl groups Phenolic acid free COOH Hydroxyphenyl C Ether-linked syringyl, C-3/C Ether-linked guaiacyl, C Guaiacyl and syringyl, not ether-linked, C-3/C Phenylcoumaran Syringyl C Guaiacyl C-1, syringyl C Guaiacyl C Guaiacyl C-5, 4-hydroxyphenyl C-3/C Arabinofuranosyl C Cellulose C-1, xylan C-1, syringyl C-2/C Hemicellulose C Crystalline cellulose C Amorphous cellulose C-4, xylan C Cellulose and xylan, C-2/C-3/C Crystalline cellulose C-6, xylan C Amorphous cellulose C OCH 3 Lignin methoxyl 21.4 Acetyl CH 3 C=O hemicellulose 3.3 TKS (internal standard) 143

173 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends Palm shells Met. Coke (MC) Analytical Tools Che mical Structure : Cellulose ([C6(H2O)5]n) Hemicellulose ([C5(H2O)4]n) Lignin ([C10H12O3]n) Cellulose peak detected at 2θ=12.8,29.4 Che mical Structure : Allotrops of Carbon (C=C) Graphitic (002) peak detected at 2θ=26.5 XRD analysis FTIR analysis Cellulose diminished After combustion ( T= 1200 C, 80%N2 + 20% O2) peak Aliphatic rings (CH3) detected at 1459 cm -1 Graphitic (002) peak still existed Aromatic rings (C=C) detected at 1556 cm -1 After combustion ( T= 1200 C, 80%N2 + 20% O2) All aliphatic rings (CH3) are consumed Aromatic rings (C=C) still existed NMR analysis Cellulose, hemicellulose and lignin detected Aliphatic and aromatic rings detected After combustion ( T= 1200 C, 80%N2 + 20% O2) Only aromatic groups were left Aliphatic groups diminished and aromatic groups were left behind Figure 4-24 Summarize of mechanism governing the breakdown of the structures before and after combustion in DTF at 1200 C (80% N 2 ; 20% O 2 ) for palm shells and MC 144

174 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends Inorganic Minerals The composition of ash is also an important indicator of the reactivity of materials (Vogt, D. and Depoux, M., 1990). Iron (Lindert, M. and Timmer, R. M. C., 1991), sodium and potassium oxides are known to act as catalysts in gas phase reactions. Hematite (Fe 2 O 3 ) and sodium and potassium oxide (Na 2 O, K 2 O) was found to catalyze the reaction (Velden, B. v. d., Trouw, J., C haigneau, R. et al., 1999). The alkali metals in palm shell materials are expected to vaporise faster compared to coke due to the weak bonding structure present available in the agricultural waste materials. The composition of the ash present in the studied samples (Figure 4-25 and 4-26) show an increase in the proportion of alkali metal in palm shells compared to coke Fe 2 O 3 in Ash (%) Palm shells 10 MC P1 P2 P3 0 Palm shell content in the blends (%) Figure 4-25 Fe 2 O 3 component present in MC and MC /palm shell blends ash 145

175 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends Iron oxide content showed significant variation in the 2 samples studied (see Table 3-3) and Figure The ash of the palm shell has Fe 2 O 3 = 20.5 wt.% while the Fe 2 O 3 present in the coke ash (5.7 wt.%), this could influence the combustion performance of the blends to a certain extent. Increasing the proportion of palm shells in the blend, the iron oxide content also will increase. Previous studied also found that the influence of iron oxide will enhance the combustion performance (Backreedy, R. I., Jones, J. M., Pourkashanian, M. et al., 2003, Yamashita, H. and Tomita, A., 1993) K2O: MC/palm shell blends Na2O: MC/palm shell blends K 2 O in Ash (%) 6 4 Na 2 O in Ash (%) 2 0 MC P1 P2 P3 Palm shells Palm shell content in the blends (%) Figure 4-26 K 2 O and Na 2 O components present in the ash from MC and MC / palm shell blends 146

176 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends Table 3-3 showed the presence of (Na 2 O, K 2 O) in coke ash is less than 1 wt.%. On the other hand, palm shell ash contains Na 2 O = 1.1 wt.% and K 2 O = 4.0 wt.%. By blending coke with palm shells, the presence of alkali (Na 2 O, K 2 O) in the palm shells ash (Figure 4-25) might catalytically enhance the combustion performance of the blends (Carlos, A. C., Hooshang, P. and Christian, R., 2001). Sodium ions are very small and can penetrate into the palm shells texture and break the intermolecular hydrogen bridges with increasing temperature. These ions can react with cellulose (lignocellulosic structure from biomass) through active alcohol groups of cellulose (Vamvuka, D., Troulinos, S. and Kastanaki, E., 2006, Yang, H., Yan, R., Chen, H., Zheng, C. et al., 2006). Further explanation of coconut shells and its influence on the combustion performance is discussed in Chapter

177 Chapter 4: Combustion & Structural Transformations of Coke/Palm Shell Blends 4.2 Summary The combustion efficiency of metallurgical coke and palm shell blends in different proportions have been investigated. The chemical and physical properties and the chars associated with reactions in gas phases are summarize below: 1.) At temperatures above 1000 ºC, the gas products evolving from palm shells pyrolysis measured by TGA-MS showed CO, CO 2 and H 2 as the main gases. The lignocellulosic structure of the palm shells break down and releasing gases. 2.) The combustion performance was seen to increase with increasing palm shells concentration in the blend. The role of volatiles, structural transformations, carbon structures and inorganic minerals are the factors governing the reaction specifically. 3.) The changes in surface area are more significant when the proportion of agricultural waste increases. SEM micrograph supported the BET measurement showing enlarged pores developed in the palm shell wastes. Palm shell s cell structures were seen to open up to a significant extent and structural changes were observed accompanied by higher surface area which results in improvement in combustion efficiency at high temperatures. 4.) The difference in the molecular structure of the carbonaceous materials are showed different behavior where the palm shell blends are better in combustion performance. The analytical tools such as XRD, FTIR and NMR analysis showed the mechanism structures changes for both materials. Palm shells lost all the aliphatic rings which contribute to increase the performance while coke structures still remain the aromatic rings at high temperature reactions. 148

178 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends CHAPTER 5 5 Combustion & Structural Transformations of Coke/ Coconut Shell Blends: Results & Discussion 5.1 Gas Phase Reactions of Metallurgical Coke and its Blends with Coconut Shell For gas phase studies, samples were prepared by blending MC with coconut shell particles in a range of proportions (see Figure 3-2), while the chemical compositions of the samples used in the present is shown in Table 3-2. Blends C1, C2 and C3 were chosen for detailed examination. The proximate (air dry base, %) and ultimate analyses (dry ash free, %) of the samples were carried out at Amdel Laboratories and Technical Services; and UNSW based on Australian standards. Coconut shell is suitable for preparing microporous carbon due to its excellent natural structure and low ash content (Heschel, W. and Klose, E., 1995, Kirubakaran, C. J., Krishnaiah, K. and Seshadri, S. K., 1991). Moreover, coconut shell is also considered superior to those obtained from other carbon sources mainly because of small macrospores structure which renders it more effective for the adsorption of gas/vapor compounds (Prabhakar, K., Maheshwari, R. C. and Vimal, O. P., 1986). Conversion of this coconut shells into new carbon material is recommended. Pyrolysis process is of key importance as thermal degradation of solid fuels occurs during both combustion and gasification. It has a key influence over the quality of the char that is either gasified or burned. As it is known, agricultural wastes thermoconversion technologies are strongly influenced by the reactivity of the char produced in the pyrolysis phase. Lignocellulosic agricultural waste basic components are hemicellulose, cellulose and lignin. Researchers have already confirmed these structures cellulose ( ºC.) and hemicellulose ( ºC) will decompose faster than 149

179 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends aromatic structures (lignin = ºC). This is the interval in which the main decomposition takes place and accounts for the greatest decomposition in the biomass pyrolysis process consisting of degradation reactions (Fisher, T., Hajaligol, M., Waymack, B. et al., 2002, Raveendran, K., Ganesh, A. and Khilar, K. C., 1996). The characteristics of individual coke and coconut shell particles in the blend are expected to jointly influence the combustion behavior of the blends. The influence of coconut shells composition and ash content on pyrolysis and combustion performance characteristics is discussed. 150

180 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends Effect of High Temperature on the Behaviour of the Carbonaceous Material Thermogravimetric analysis, besides providing a means for the preliminary assessment of fuel values in agricultural waste materials (Norton, G. A., 1993), allows for a prior knowledge of initial and final temperatures for their combustion as well as other relevant data such as maximum reactivity temperature or total combustion time. On the other hand, the use of TGA-MS techniques is well established for the characterization of gaseous products arising from thermal degradation of various materials (Conesa, J. A., Marcilla, A., Moral, R. et al., 1998), allowing to relate weight loss signals and gaseous emission (Otero, M., Díez, C., Calvo, L. F. et al., 2002). This present study investigated the thermal behavior and gas products of coconut shells released in high temperatures processes. The weight loss and gas formation of palm shells and coke has been discussed in Chapter 4. Figure 5-1 (a) to (f) shows the changes in weight and gas formation rates during the pyrolysis of coconut shells. The integrated product distributions (e.g. CO, CO 2, CH + 4, CH + 3, C 2 H 6, C 3 H 4, H 2, and H 2 O) are also shown in the Figure 5-1. The first weight loss was attributed to the moisture released at 110 ºC. The weight of coconut shells showed a decrease at temperature 150 ºC and continued losing weight around ºC for all the samples. After that, the weight decreased gradually when increasing the pyrolysis temperature further from 1000 ºC to 1450 ºC for coconut shell samples. The formation behavior of H 2, CH 4, H 2 O, CO, CO 2 and hydrocarbons accompanied the sharp decomposition seen in the first weight loss ( C). However, the CO, CO 2 and H 2 formation accompanied the second weight loss. H 2 was released at a higher temperature (>1000 ºC) attributed to the hemicellulose and lignin structure present in coconuts shells. H 2 and CH 4 might be attributed to the higher content of aromatic ring and O CH 3 functional groups in the origin lignin sample, as the H 2 from organics pyrolysis mainly came from the cracking and deformation of C=C and C H while CH 4 was mainly brought by the cracking of methoxyl. Cellulose resulted in the highest CO release, due to the higher carbonyl content in it while hemicellulose produced high CO 2 because of 151

181 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends the higher carboxyl content (Haiping, Y., Yan, R., Chen, H. et al., 2007). The release of C 3 H 4 and C 2 H 6 was generally very low (Figure 5-1 (c) and (f)). 152

182 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends Coconut Shells Mass (%) (a) Ion Current, x (A) (d) Hydrogen H Temp. ( C) Temp (ºC) Ion Current, x (A) (b) CO CO/ CO2 Ion Current, x (A) (e) Water H2O 0.5 CO Temp (ºC) Temp (ºC) Ion Current, x (A) 1.2 (c) Hydrocarbons CH4 Ion Current, x (A) (f) Hydrocarbons C3H4 2.0 C2H6 CH Temp (ºC) Temp (ºC) Figure 5-1 (a) (f) The mass loss of coconut shells with on-line MS analysis of the gas products during pyrolysis 153

183 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends The pyrolysis characteristics for coconut char is shown in Figure 5-2 (a) (f). The decomposition took place slowly under the whole temperature range from ambient to 1450 ºC. Two peaks of CO release (450 and 600 ºC) were shown at lower temperature (<600 ºC) while one peak was found at high temperatures (>600 ºC) (Figure 5-2 (b)). The release of CO was attributed of carbonyl (C O C) and carboxyl (C=O) at low temperature and the second was attributed to the thermal cracking of tar (Yang, H., Yan, R., Chen, H., Zheng, C. et al., 2006). CO 2 was released at 200 ºC and 1000 ºC. CO 2 release was mainly caused by the primary pyrolysis, while secondary pyrolysis was the main source for release of CO and CH 4. H 2 gas was released at a higher temperature (600 ºC) (Figure 5-2 (d)). 154

184 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends Coconut Char Mass (%) (a) Ion Current, x (A) (d) Hydrogen H Temp. ( C) Temp (ºC) Ion Current, x (A) (b) CO CO/ CO2 CO2 Ion Current, x (A) (e) Water H2O Ion Current, x (A) Temp (ºC) 6.0 (c) Hydrocarbons 5.0 CH CH3 1.0 C2H Temp (ºC) Ion Current, x 10-8 (A) Temp (ºC) 2.0 (f) Hydrocarbons C3H Temp (ºC) Figure 5-2 (a) (f) The mass loss of coconut char with on-line MS analysis of the gas products during pyrolysis 155

185 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends Effect on Blending on Combustion The thermal behavior of coke and its blends with coconut shell blends during pyrolysis in N 2 atmosphere, under isothermal conditions was studied and is shown in Figure 5-3. The pyrolysis reaction is much faster relative to heterogeneous char combustion, which is spread over a longer period of time. The total weight loss of MC and its blends with coconut shell during pyrolysis occurring in less than 50 seconds. Under thermal effect, the maximum weight loss observed in the TGA test was in order as proximate analysis, (being 2.1 % MC, 8.4 % for C1 blends, 10.5 % total weight loss for C2 blends and 11.1 % weight loss for C3 blends as can be seen in Figure 5-3. Residual mass decreased with increasing coconut shell content in the blend. This trend is due to the high volatile content of the coconut shells compared to coke alone. The difference is attributed to the differences in the strength of the molecular structure of the material. The lignocellulosic bond strength from coconut shells are relatively weaker and less heat resistance compared to coke structure (Sadhukhan, A. K., Gupta, P. and Saha, R. K., 2009, Smith, L. H., and, S. L. D. and Fletcher, 1994, Ulloa, C. A., Gordon, A. L. and García, X. A., 2009, Vuthaluru, H. B., 2004). Hence, the mass loss is lower (higher residual mass) compared to the pyrolysis of blends containing coconut shell proportions. Nitrogen atmosphere and high temperatures leads to weight loss derived mainly from volatile release, whereas combustion conditions leads to oxidation of volatile matter and residual carbon as shown in Figures 5-3 and 5-4. Different behaviors are exhibited by coke and their corresponding coconut shell blends, with larger fractions of coconut shell released as volatiles during the combustion process. This high volatile yield was seen to occur over a relatively short time and is believed to influence the time required for complete combustion when compared to coke alone (Yoshizawa, N., Maruyama, K., Yamada, Y. et al., 2000). 156

186 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends MC Residual Mass, % C1 C2 C Time, sec Figure 5-3 Weight loss curves of MC and its blends with coconut shells at temperature, 1200 ºC; N 2 atmosphere 25.0 Total Weight Loss, % C1 C2 C3 0.0 MC Coconut shells content in the blend Figure 5-4 Comparison of total weight loss curve (50 seconds) of MC and its blends with coconut shells at temperature, 1200 ºC; N 2 atmosphere 157

187 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends In order to quantify the combustion performances, the burnout of the samples has been calculated on the basis of ash conversion. Figure 5-5 exhibits the extent of burnout in the drop tube furnace of the coke and its blends in different proportions with coconut shells at a fixed temperature of 1200 ºC and 80% N 2 ; 20% O 2 concentration. Coke burnout values were 10 % while 100 % coconut shells showed burnout values of 15.1 %. It can be seen that coconut shell blends (C1 to C3 blends) showed higher burnout values compared to the parent coke. A small increase is observed with increasing coconut shell content in the blend as expected on the basis of their volatile matter content (C1 = 20.7 %; C2 = 21.1 % and C3 = 22.2 %) Burnout, % C1 C2 C3 Raw coconut shells 10.0 MC Coconut shells content in the blend Figure 5-5 Effect of blending MC with varying coconut shell content in the blends on the combustion at 1200 C in the presence of 80% N 2 ; 20 % O 2 158

188 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends The amount of volatile matter present in the carbonaceous material blends are known to influence the burnout values to a certain extent (Sahajwalla, V., Zaharia, M., Kongkarat, S. et al., 2009). Proximate analyses of the coconut shell/coke blends indicate an increase in VM with increasing coconut shell content in the blend (Figure 5-6 (a)). An almost linear increase in burn-out is observed with increasing VM for the three blends (Figure 5-6 (b)). Higher combustion efficiency of coconut shell/coke blends is expected to be significantly influenced by the quantity and nature of volatiles released during gas phase reactions. However, for 100% coconut shells with higher VM showed a low value of burnout. This is due to low of fixed carbon content in coconut shells. Thus, by blending with MC which is high fixed carbon content, the blending is expected to improve the burnout. 159

189 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends (a) Volatile Matter, % % 12.2% 77.8% 27.9% 19.4% MC C1 C2 C3 Coconut Shell Burnout, % (b) C1 MC C2 C3 Coconut shells Volatile Matter, % Figure 5-6 Effect of volatile matter on (a) blend composition and on (b) the combustion performance of coconut shell/mc blends 160

190 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends The changes in the original structure of the matrix of the coconut shells material is the result of the cross-linking of the reactive points brought about by the disruption of the original polymeric structure of the precursor material (i.e. the cellulosic and lignin units of the raw material) and the subsequent reconstitution of a new matrix structure during gas phase reactions that improve the combustion performance. These will be discussed in next section Moreover, agricultural wastes (coconut shells) differ from coke in many important ways, including the organic, inorganic, energy content, and physical properties. Relative to coke, agricultural waste materials generally has less carbon, more oxygen, more silica, sodium, phosphorus and potassium, lower heating value, higher moisture content, and lower density (Demirbas, A., 2006). The influence of inorganic matter will be discussed in sub-section Table 5-1 showed the amounts of the inorganic species present in the fully-combusted MC/agricultural waste blends obtained from the measurement. Table 5-1 shows that the amount of oxides still exist in coconut shells after combustion with approximately 2 %. This method is used in order to quantify the accurate values of mass burn-off (ash tracer method) during the combustion of MC/agricultural waste blends. 161

191 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends Table 5-1 XRF analysis showing the amounts of major inorganic elements (based on their oxides) present in the ash of the coke, coconut shell blends and coconut shells following treatment at 1200 C under 20% oxygen and 80% nitrogen atsmosphere Material Amount of SiO 2 (%w/w of raw fuel) Amount of Al 2 O 3 (%w/w of raw fuel) Amount of TiO 2 (%w/w of raw fuel) Metallurgical Coke Coconut shell C C C

192 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends Physical Properties and Structural Transformations It has been established that at elevated temperatures and low residence times, as in the case of coal combustion, the reaction is likely to be controlled by pore diffusion and chemical reactions on pore surfaces. In such case, the physical parameters of the reacted materials play a very important role. Therefore, in the present work, the surface accessibility of coke and its blends with coconut shell was investigated in detail and was assessed quantitatively by BET surface area measurement and qualitatively by SEM photomicrographs of the char particles collected after oxidation reaction in the drop tube furnace Surface Area Measurement BET Surface Area The BET surface area was measured for MC and its mixtures in different proportions with coconut shells; N 2 was pushed in the micropore channels. Figure 5-7 illustrates the BET surface area of the raw agricultural waste materials containing coconut shell samples which showed an increase in the surface area after combustion in DTF. The BET surface area results demonstrated minor changes in the BET surface area for coke, from 31.6 to 34.9 m 2 /g, while a high surface area had developed for coconut shell blends after the reaction in DTF i.e., from 26.3 to 36.7 m 2 /g for C1 blend, from 13.9 to 25.7 m 2 /g for C2 blend, and from 7.4 to 17.2 m 2 /g for C3 blend in Figure 5-7. In the case of coconut shell blends, the initial increase in combustion efficiency is supported by the increase in surface area (high surface area = 36.7 m 2 in Figure 5-7), followed by a small increase in combustion efficiency (Figure 5-5). 163

193 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends BET Surface Area (m 2 /g) Before DTF After DTF 0.0 MC C1 C2 C3 Coconut shells Coconut shell content in the blends (%) Figure 5-7 Changes in the micropore surface area before and after combustion in DTF as a function of the coconut shell content in the blends The changes in surface area before and after combustion have been estimated for coke and its blends in different proportions with coconut shell and an index of surface area change has been defined as ΔSA (%) in Figure 5-8. In the case of coconut shell blends, the initial increase in the combustion efficiency is supported by the first ΔSA which is rises from C1 blends = 28.4 % to C2 blends = 45.7 %, however was seen to decrease afterwards for C3 in the blends. The drop in ΔSA might be due to particle fragmentation (coconut shell sample = 23.6 %). Further SEM studies were undertaken to investigate this aspect (Figure 5-11). 164

194 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends Δ Surface Area (%) C2 C C1 Coconut shells MC Coconut shells content in the blend (%) Figure 5-8 The changes in micropore surface area for coconut shell wastes content in the blend Surface area modification during combustion is known to influence crack formation (Mitchell, R. E. and Akanetuk, A. E. J., 1996). It showed that a sharp change in the micropore surface area promotes a severe cracking and further SEM imaging is considered to demonstrate the behavior of coconut shell/ coke blends. From Figure 5-9 (a), the raw shells (brown color which is the fiber from the shell of the coconuts. After combustion in gas phase reactions, the physical structure turned into dark black color with porous structures. The ash particles (light grey particle) can be seen in the coconut shell chars (Figure 5-9 (b)). 165

195 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends (a) (b) 0.05 mm 0.05 mm Raw Coconut Shells Coconut Shells after combustion in DTF Figure 5-9 Images of (a) raw coconut shells and (b) coconut shell after combustion in gas phase reactions at 1200 ºC (80% N 2 ; 20% O 2 ) Structural Transformations - Scanning Electron Microscopy (SEM) The physical properties and structural transformations of raw coke and coke after combustion by SEM have been presented in Chapter 4. SEM micrographs of coconut shell and its residual mass after combustion are presented in Figure 5-10 (a) (f). As it can be seen, the raw coconut shell and its cross-section clearly show pore structures which are made of cylinder-like tubes (Figure 5-10 (a)) composed of layers of several flat sheets. The cross section of the raw coconut shell gives more detail on the morphology of the cell like tubes which are seen to lie almost distinct from each other. After combustion the individual sheets of the tube walls have fused together and are no longer distinguishable from each other (Figure 5-10 (d) and (f)). Moreover, the walls of the individual tubes have also undergone a similar transformation. These have fused together at their point of contact with the result that a single solid matrix has been formed with interspersed pores. A similar observation on the pore structure and transformation following combustion was made by Achaw et.al. (Achaw, O. W. and Afrane, G., 2008) implying that the cell structure was lost after devolatilization. 166

196 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends Figure 5-10 SEM micrographs of transverse section of (a) raw coconut shell, (b) coconut shell after combustion at 1200 C in DTF and cross-sectional images of (c) raw coconut shell, (d) coconut shell char (e) and (f) edge section of coconut shell char 167

197 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends The micrographs of 100% coconut shell following combustion are taken to support the surface area measurements (Figure 5-11). As it can be seen from Figure 5-11 (a) and (b) particle fragmentation is clearly present and a small fragment has detached from the particle (Figure 5-11 (b)). A narrow fault can be seen on the particle surface (Figure 5-11 (a)) indicating that it is likely to form another fragment during further reactions. It is suggested that the fault in the shell of the particle results from the connection of neighbouring macropores on the particle surface that connect the external environment to the internal void. Moreover, the irregular thickness of the coconut shell particles provides some weak parts, which may be consumed rapidly, contributing to the formation of the surface pores and particle fragmentation when compared to palm shell. The association of these physical changes to reaction would influence the combustion performance as well as other factors; e.g. VM, ash/catalytic effects and chemical structures. 168

198 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends Figure 5-11 (a) and (b) SEM micrograph of coconut shell after combustion in DTF showing fragmented particles The char particles resulting from the coconut shells seem to contain a greater number of large pores when compared to those in coke-char particles, which may be attributed to the evolution of volatile species from the interior of the particle during the combustion process. 169

199 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends The Role of Chemical Properties and Carbon Structures Chemical Structures Lignocellulosic of agricultural waste basic components are hemicellulose, cellulose and lignin. Previous study (Haiping, Y., Yan, R., Chen, H. et al., 2007), confirmed that lignin starts decomposing at low temperatures ( C) and continues to decompose at a low rate until approx. 900 C. Hemicellulose is the second component to start decomposing, followed by cellulose, in a narrow temperature interval from about 200 to 400 C. This is the interval in which the main decomposition takes place and accounts for the greatest decomposition in the biomass pyrolysis process consisting of degradation reactions (Fisher, T., Hajaligol, M., Waymack, B. et al., 2002, Hosoya, T., Kawamoto, H. and Saka, S., 2007). Figure 5-12, shows the content of cellulose was 19.8 %, hemicellulose equally 50.1 % and lignin was 30.1 %. The high content of cellulose and hemicellulose will enhance the combustions performance compared with coke which contains aromatic carbon. MC Allotropes of Carbon Coconut shell Cellulose 19.8 Hemicellulose 50.1 Lignin 30.1 Figure 5-12 Cellulose, hemicellulose and lignin contents of coconut shell (Wan Daud, W. M. A. and Wan Ali, W. S., 2004). 170

200 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends Chemical Properties X-ray Diffractions Structural modifications of coconut shell and its blends with MC have been characterized by X-ray diffraction experiments in Figure 5-13 and Table 5-2 illustrates the resulting x-ray spectra. Based on visual inspections of Figure 4-15 (Chapter 4), it could be suggested that coke peaks correspond to a type of intermediate structure between graphitic and amorphous carbon with a wide range of carbon lattice spacing. The raw coconut shells show characteristic peaks of amorphous carbonaceous constituents in their structural matrices. A peak designated as μ band, was observed at 2θ = 15.8º, which may be attributed to the packing of the structure such as aliphatic side chains or condensed saturated rings which was cellulose. Moreover, the peaks detected the basic compounds in coconut shells was oxides elements such as sodium, aluminum and silicate. It can be observed from Figure 5-13 that the amorphous carbonaceous constituents in the raw coconut shells undergo massive conversion during devolatilisation as indicated by the disappearance of the hump and the presence of a plateau section in the diffraction spectrum between The long carbon ring which is expected to be lignin was assigned at 2θ ~ 44.6). The overall feature of the diffraction spectrum of the coconut shells after combustion resembles by previous reported by (Wornat, M. J., Hurt, R. H., Yang, N. Y. C. et al., 1995) at similar combustion temperatures and mass burnout. 171

201 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends Raw Coconut Shells Cellulose (Cn(H 2 O)n) Na Al Si O SiO 6, C (Cn(H O)n) 2 Intensity (counts) Coconut Shells after DTF SiO 2 (Cn(H O)n) 2 C (Cn(H 2 O)n) Na 6, Al 4 Si 4 O 17 SiO 2 SiO 2 C (Cn(H O)n) 2 SiO Scattering angle (2θ) Figure 5-13 XRD patterns of coconut shells before and after combustions (T=1200 ºC: 20% O 2 ; 80% N 2 ) 172

202 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends Table 5-2 XRD peaks characteristic of raw coconut shells and coconut shells after combustion in DTF 1200 ºC with atmosphere (80% N 2 ; 20% O 2 ) Sample Peak Position (2θ) d- spacing (Å) Crystal Size h k l Possible Compound (chemical formula) Raw coconut shells Coconut shells after DTF Cellulose [C 6 (H 2 O) 5 )]n Graphite (C) Anorthite (Ca, Na)(Si, Al)O Carbon (C n (H 2 O) n ) Quartz (SiO 2 ) Quartz (SiO 2 ) Graphite (C) Anorthite (Ca, Na)(Si, Al)O Quartz (SiO 2 ) Carbon (C n (H 2 O) n ) Quartz (SiO 2 ) 173

203 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends Figures 5-14 to 5-16 show that, with increasing combustion temperature, the (002) peak position shifts from 24.4º to 26.5º, closer to that of graphite (26.6º) and all the coconut shells/coke blends peaks become more sharper. In general, diffuse and broad bands in XRD patterns represent the existence of short range order in the carbon structure, while sharp and narrow peaks correspond to highly crystalline region with high degree of long-range order. A summary of peak characteristics are presented in Table 5-3. The only noticeable peak emerging from coconut shell/coke blends (C1, C2 and C3 blends sample), its diffraction spectrum situated at 2θ = 21.9º (100), 43.1º (103) and 50.1º (112) strongly suggest the existence of the compounds of SiO 2 (quartz), the most abundant oxide in the ash. Highest carbon peak was found at 2θ = º 26.5 (002). At peak θ 2= 29º and 44.6º corresponds to carbon ring of compound in the coconut shell blends. After combustion peak characteristics of cellulose (2θ = 29.4º) was seen to diminish with high temperature of reactions. 174

204 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends Raw_C1 blends SiO 2 FeAl O 2 4 C1 blends after DTF C C Intensity (counts) Cellulose C CaCO 3 C SiO 2 C ; SiO 2 C C H N O SiO SiO 2 FeAl 2 O 4 C CaCO 3 Al Si O C C SiO 2 C ; SiO SiO 2 2 C FeAl 2 O 4,SiO 2 SiO 2 SiO Scattering angle (2θ) Figure 5-14 X-ray diffraction analysis of coke/coconut shell blends; C1 before and after combustion temperature 1200 ºC in 20% O 2 and 80% N 2 atmosphere 175

205 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends Raw_C2 blends C Cellulose SiO 2 FeAl 2 O 4 C CaCO 3 C C2 blends after DTF C SiO 2 C ; SiO SiO 2 2 C FeAl 2 O 4,SiO 2 SiO 2 Intensity (counts) SiO 2 FeAl 2 O 4 C CaCO 3 Al Si O C C SiO 2 C ; SiO SiO 2 2 C FeAl 2 O 4,SiO 2 SiO 2 SiO Scattering angle (2θ) Figure 5-15 X-ray diffraction analysis of coke/coconut shell blends; C2 before and after combustion temperature 1200 ºC in 20% O 2 and 80% N 2 atmosphere 176

206 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends Raw_C3 blends C Intensity (counts) Cellulose SiO 2 FeAl O 2 4 C CaCO 3 C SiO 2 C ; SiO SiO 2 2 C C3 blends after DTF FeAl 2 O 4,SiO 2 SiO C SiO 2 FeAl 2 O 4 C CaCO 3 Al 6 Si 2 O 13 C C SiO 2 C ; SiO SiO 2 2 C FeAl 2 O 4,SiO 2 SiO 2 SiO Scattering angle (2θ) Figure 5-16 X-ray diffraction analysis of coke/coconut shell blends; C3 before and after combustion temperature 1200 ºC in 20% O 2 and 80% N 2 atmosphere 177

207 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends Table 5-3 XRD peaks characteristic for all samples C1, C2 and C3 blends before and after combustion in DTF 1200 ºC with atmosphere (80% N 2 ; 20% O 2 ) Sample Raw C1, C2, C3 blends C1, C2, C3 blends after DTF Peak Position (2θ) d- spacing (Å) Crystal Size h k l Possible Compound (chemical formula) Cellulose [C 6 (H 2 O) 5 )]n Quartz (SiO 2 ) Calcite (CaCO 3 ) Carbon (C n (H 2 O) n ) Quartz (SiO 2 ) Graphite (C), Quartz (SiO 2 ) Calcite (CaCO 3 ) Mullite (Al 6 Si 2 O 13 ) Carbon (C n (H 2 O) n ) Quartz (SiO 2 ) 178

208 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends Chemical Bonding Fourier Transform Infrared Spectroscopy (FTIR) The assignment of peaks in the FTIR spectra was carried out by comparison with literature data (Biagini, E., Barontini, F. and Tognotti, L., 2006, Shao, J., Yan, R., Chen, H. et al., 2007). Table 5-4 shows the regions of the spectra for the present work and FTIR spectra of the coconut shell before and after reaction in DTF are given in Figure Bands corresponding to aromatic CH, OH and CO can be distinguished. The presence of the sharp peaks between 1640 and 1700 cm -1 could be attributed to C O (carbonyl) stretching vibration indicative of ketones, phenols, carboxylic acids or aldehydes for raw sample, and represent C=C stretching vibrations indicative of alkenes and aromatics. The samples after combusted showed that the moisture was not completely removed due to a few seconds of reaction in DTF. 179

209 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends Raw Coconut Shells 90 % Transmittance Moisture CH 2 stretching C=O N-H stretching aromatic rings (lignin) C-CH bending 3 C-H C-C alkene C-O C-CH bending 3 C-1 ring Coconut shells after DTF Moisture CH 2 stretching O-H stretching N-H stretching C-CH 3 bending C-O C-CH bending Aromatic (lignin) Wavenumbers (cm -1 ) Figure 5-17 FTIR spectra of coconut shell before and after combustion in DTF 180

210 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends The spectrum of the combusted coconut shell was seen to decrease after burning. The spectrum showed the absence of most bands and mainly an aromatic polymer of carbon atoms while the aliphatic rings were not strongly detected. Mok et al. (Mok, W. S. L., Antal, M. J., Szabo, P. et al., 2002) made similar observations from the FTIR analysis of cellulose chars prepared at temperatures of up to 450 C. The major chemical changes were reported to be due to dehydration, carbonyl group formation and elimination, decomposition of aliphatic and formation of aromatic char units has been revealed. This study indicated that at least some steps (e.g. dehydrations, decomposition of aliphatic) in the chemical transformations occurring during devolatilization may be similar for different biomasses. The peak at around 1513 cm -1 could correspond to C-C stretching vibrations related to alkenes and aromatic components. Below 1500 cm -1, all bands are complex and have their origin in a variety of vibrational modes. C H stretching and bending vibrations between 1380 and 1465 cm -1 indicate the presence of alkane groups in pyrolysis oils derived from biomass (Tsai, W. T., Lee, M. K. and Chang, Y. M., 2006). Absorption bands possibly due to C O vibrations from carbonyl components (i.e., alcohols, esters, carboxylic acids or ethers) seem to occur between 1300 and 900 cm

211 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends Table 5-4 Characterization of FTIR peak profiles of raw coconut shells and coconut Peak location range (cm -1 ) shells after combustion at 1200 ºC at atmosphere (80% N 2 ; 20% O 2 ) Possible Functional Groups Raw Coconut Shells Coconut Shells after DTF O-H stretching C-H asymmetric and symmetric stretching in methyl and methylene group C=O stretching in acetyl and uronic ester groups or in carboxylic group of ferulic and coumaric acids C=C, symmetric aromatic O-Hydroxyl diaryl ketones N-H bending in primary amine Aromatic rings (lignin) CH 3 (aliphatic) C-H rocking in alkanes or C-H stretching in methyl and phenolic alcohol Si-CH 2 stretching in alkane or C-C plus C-O plus C=O stretching C-C alkenes C-O deformation in secondary alcohol and aliphatic ether or aromatic; C-H in plan deformation plus C-O deformation in primary alcohol C-1 groups frequency

212 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends Carbon Structures Nuclear Magnetic Resonance ( 13 C NMR Spectroscopy) A large number of structural studies based on solid-state 13 C NMR spectroscopy of crystalline and amorphous cellulose, cellulose polymorphs, and model compounds have been reported in literature (Alesiani, M., Proietti, F., Capuani, S. et al., 2005, Sekiguchi, Y., Frye, J. S. and Shafizadeh, F., 1983, Snape, C. E., Axelson, D. E., Botto, R. E. et al., 1989). Solid-state NMR proved to be especially useful to characterize the process of degradation of wood/ biomass/ agricultural waste, revealing significant differences in the chemical structure. Solid-state CP/MAS 13 C NMR spectroscopy is a powerful tool for investigation of cellulose structure. If it is used in conjunction with spectral fitting, it is possible to study and monitor changes in both cell wall structure and composition determined by chemical, mechanical, and thermal factors. The carbon structure of coke had been discussed in Chapter 4 (Figure 4-22). Raw cokes mainly consist of aliphatic and aromatic carbon structures. However, coconut shells showed variety of carbon resonances. The raw coconut shells shows a peak at 22 ppm assigned to the acetyl methyl groups of hemicelluloses in Figure 5-18 (a) and resonance assignment of peak was presented in Table 4-8. A peak at 56 ppm is attributed to the methoxy groups of lignin. The peak at 65 ppm is aliphatic C-6 carbons of crystalline cellulose (Link, S., Arvelakis, S., Spliethoff, H. et al., 2008). Moreover, at peak at 72 and 75 ppm, the coconut shells have the highest peak, corresponding to the C-2, C-3, and C-5 carbons of cellulose. C-4 amorphous and crystalline cellulose at 84 and 89 ppm present in the raw coconut shells, respectively, while at 105 ppm a sharp peak indicating the presence of anomeric carbon of sugars (aldehyde and ketones). Lignin syringyl C-1, syringyl C-5, and quaiacyl C-2 are also detected in the raw coconut samples exhibiting a broad line in the region between 134 and 137 ppm. The signals at 153 ppm correspond to syringyl C-3 and syringyl C-5. In the case of coconut shell, the signal at 162 ppm corresponds to 4-hydorxyphenyl C-4. A peak of carbonyl groups (COOH) of hemicelluloses seen at 173 ppm which are well known to have a high reactivity (Scholze, B., Hanser, C. and Meier, D., 2001). 183

213 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends The NMR spectra of the residual chars derived from combustion of coconut shell wastes are presented in Figure 5-18 (b). As can be seen in this figure, only one peak is detected at ppm which is assigned to aromatic lignin (124 ppm-issuing from polyaromatic hydrocarbons). In can be concluded that aromaticity of the chars increased due to the loss of the less stable aliphatic groups. The broad resonances in Figure 5-18 (b) demonstrate the amorphous nature of the chars due to the presence of free radicals and the complex structure of this material. The aromatic component of lignin is very resistant to thermal degradation, and the resulting char is highly refractory. The lignocellulosic structure was lost and transformed to polycyclic material with a preponderance of aromatic structures as the temperature of treatment increases. All cellulose and lingo-cellulosic materials under thermal treatment with final temperatures between 800 and 1000 C were seen to undergo structural transformation, resulting in a more ordered structure (Bardet, M., Hediger, S., Gerbaud, G. et al., 2007). As in the present study, residues reaches end temperature of 1200ºC, such a transformation is expected for the employed agricultural waste materials. 184

214 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends (a) Cellu lose C-2, C-3, C-5 Raw coconut shells Acetyl carbonyl groups of hemicellulose Ketones O R-C-R O -C-O O -C-OCH3 Oxygen substituted (-OCH and OH) aromatic carbons Lignin Lignin aromatic Aromatic C-1 carbon of sugars (cellulose) O -C-O C-4 carbon CH3 cr am C-6 carbon crystalline cellu lose Methoxy group Lignin:-OCH3 Acetyl methyl groups of hemicellu lose CH3 CH2 in saturated aliphatic chain Aliphatic (b) [ppm] Coconut shells at 1200ºC Aromatic -C * aromatic * [ppm] Figure 5-18 Spectrum of CP/MAS 13 C NMR of (a) raw coconut shell and (b) coconut shell char at 1200ºC in a drop tube furnace [80% N 2, 20% O 2 ] (* spinning sideband) 185

215 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends Inorganic Minerals Minerals, such as iron and alkali metals in agricultural wastes contribute to the mineral matter in the blends and influence combustion. Mineral phases in coke reactivity were investigated Grigore et al. (Grigore, M., Sakurovs, R., French, D. et al., 2006) identified Australian coke containing more than 7 wt.% of iron oxide in the ash having catalytic effects. From Figure 5-19, coke had lower content of Fe 2 O 3 in the ash (5.7 wt. %) than coconut shells which was 46.8 wt. % of Fe 2 O 3 in the ash. This values are higher than the Fe 2 O 3 content in coke, thus could influence the combustion performance of the blends. Increasing the proportion of coconut shells in the blend, the iron oxide content also will increase. Backreedy et.al. (Backreedy, R. I., Jones, J. M., Pourkashanian, M. et al., 2002) studied the mechanism on the formation of a C - metal bond during devolatilisation. The alkali metals, C O M bond (where M can be Fe, Na, K or Ca) was seen to be formed which effected the C C bond and hence catalysed the gas phase reaction (R Backreedy, J. M. J., 2002). Yamashita et. al (Yamashita, H. and Tomita, A., 1993) found that during devolatilization, the chemical form of iron species changed stepwise with high temperature. Thus, a rapid increase of the burnout is expected. 186

216 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends 50 Coconut shells 40 Fe 2 O 3 in Ash (%) C2 C3 MC C1 0 Coconut shell content in the blends (%) Figure 5-19 Fe 2 O 3 component present in the ash from MC and MC /coconut shell blends Figure 5-20 showed the presence of (Na 2 O, K 2 O) in coke ash is less than 1 wt.%. However, coconut shell has higher content of Na 2 O = 7.9 wt.% and K 2 O = 16.9 wt.%. By blending coke with agricultural wastes, the presence of alkali (Na 2 O, K 2 O) in the agricultural wastes ash (Figure 5-20) might catalytically enhance the combustion performance of the blends (Carlos, A. C., Hooshang, P. and Christian, R., 2001). From previous study, alkali metals such as potassium and sodium in coke are associated with aluminosilicate in an unexchangeable ion form and they are believed to be catalytically inactive (Lang, R. J. and Neavel, R. C., 1982). 187

217 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends K 2 O in Ash (%) K2O: MC/coconut shell blends Na2O: MC/coconut shell blends Na 2 O in Ash (%) 0 MC C1 C2 C3 Coconut shells Coconut shell content in the blends (%) Figure 5-20 K 2 O and Na 2 O components present in the ash from MC and MC - coconut shell blends The inorganic elements are located at the edge sites of the carbon layer and at the vacancies between the carbon layers in the graphitic-like carbon building blocks of coke/agricultural blends also known as carbon active sites (Laurendeau, N. M., 1978, Mims, C. A., Chludzinski, J. J., P abst, J. K. et al., 1984, Walker Jr, P. L., Taylor, R. L. and Ranish, J. M., 1991). Their location allows them to catalytically enhance oxygen reactivity of neighbouring carbon atoms in the carbon matrix (MacPhee, J. A., Charland, J. P. and Giroux, L., 2006, Walker Jr, P. L., Taylor, R. L. and Ranish, J. M., 1991). Rate of pyrolysis are reported in the next section 5.2, while a thorough comparison between palm shell/coke and coconuts shell/coke blends is clearly explained along with the differences in their combustion and structural transformations. 188

218 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends 5.2 Comparison of Coke/Palm Shells & Coke/Coconut Shells Blends in Gas Phase Reactions This section reports comparison of coke/palm shell and coke/coconut shell blends in gas phase reactions. As the combustion reaction involves an initial thermal decomposition, followed by pyrolysed char combustion, the weight loss of the carbon based materials was studied in 100 % N 2 and in a gas mixture consisting of 20% O 2 and 80% N 2, respectively. The weight loss curve is plotted as a function of reaction time in Figure 5-21 (a) and Figure 5-22 (a). The total weight loss of MC and it blends with palm shell and coconut shell occurred in less than 100 seconds. Such trend supports a lack of synergy between the two fuels, which is in good accordance with the previously published literatures (Meesri, C. and Moghtaderi, B., 2002, Vuthaluru, H. B., 2004). The extent of volatile release can be quantitatively compared in terms of devolatilization rates (Kim, B.-c., Gupta, S., Lee, S.-h. et al., 2008), which is calculated from the weight loss data by considering the first-order kinetics (Eq. 5.1): - dw / dt = K (W 0 W ) Eq. 5-1 Where: W,W, and dw / dt are the initial mass of the sample, the final mass of the solid residue, and the rate of the mass change in a time range of 20s, and K is the rate constant, in s -1, respectively. Correlating the rate constant of the tested samples with the actual amount of volatiles present in the initial blends Figure 5-21 (b) and Figure 5-22 (b), an increase in devolatilization rates is seen with an increase in agricultural content in the blends. Faster rates of devolatilization were seen for P3 blends compared to coke/coconut blends, while pure coke, as expected, shows the slowest rate. Figure 5-21 (b) showed an increased in rates of devolatilization from C1 to C2 blends which then decreased for the C3 sample. The difference in lignocellulosic structure of both samples agricultural waste materials is expected to influence the rates. 189

219 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends On the other hand, palm shell blends showed a continuous increase with increasing palm shell proportion in the blend. Previous studies on the pyrolysis of agricultural waste (Antal, M. J., Jr. and Varhegyi, G., 1995), report that lignin gives a higher char yield than cellulose or hemicellulose. A relatively greater weight loss of palm shell (and lower residual mass) could be attributed to its higher cellulose content (palm shells cellulose = 29.7 %; coconut shells cellulose = 19.8%), which is less stable compared to lignin (Figure 4-14). These results suggest that cellulose content in the agricultural wastes may enhance the combustion characteristics and decomposition of lignin since the cellulose compounds have a structure of branching chain of polysaccharides and no aromatic compounds, which are easily volatilized. Consequently, the agricultural wastes will burn at the flowing steps; first, the cellulose components in the agricultural wastes are volatilized, so that the porosity in the char particles of agricultural wastes increases and that oxygen easily diffuses into the char particles. Next, the lignin components in the agricultural wastes can also react with oxygen diffused even if the reactivity of lignin itself is low. In other words, this discussion suggests that the char morphology will be one of the important indices to evaluate the agricultural wastes reactivity during combustion. 190

220 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends Residual Mass, % MC P1 P2 P (a) Time, sec 10 (b) Rate constant (K, s -1 x 10-3 ) P1 P2 P3 MC Volatile matter (%) Figure 5-21 (a) Comparison of residual mass (%) in TGA with time for 100% MC and its mixtures with palm shells (b) variation of rate constant of the palm shell blends with volatile matter at 1200ºC in N 2 atmosphere 191

221 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends MC Residual Mass, % C1 C2 C (a) Time, sec 10 (b) Rate constant (K, s -1 x 10-3 ) C1 C2 C3 MC Volatile matter (%) Figure 5-22 (a) Comparison of residual mass (%) in TGA with time for 100% MC and its mixtures with coconut shells (b) variation of rate constant of the coconut shell blends with volatile matter at 1200ºC in N 2 atmosphere 192

222 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends Thermogravimetric analysis has been performed for MC and its corresponding palm shell and coconut shell blends as compared with previous study (rubber blends) (Zaharia, M., 2010). The weight loss curve is plotted as a function of reaction time in Figure 5-23 (a) and (c). The curve for each agricultural blend was found along with the curves of the reference material (MC). Such trend supports a lack of synergy between the two fuels, which is in good accordance with the previously published literatures (Zaharia, M., 2010). The devolatilization of rubber/coke was chosen for comparison from a previous study (Sahajwalla, V., Zaharia, M., Kongkarat, S. et al., 2009, Zaharia, M., 2010) due to the volatile matter content in the polymeric material. Rubber was chosen since it s containing high volatile that similar to agricultural wastes (~ %). Correlating the rate constant of the tested samples with the actual amount of volatiles present in the initial blends (Figure 5-23 (b) and (d)), an increase in devolatilization rates with an increasing agricultural content in the blends is seen. The rate constant of palm shell blends (P3 blends) is similar to the rate constant of rubber blend (R3) which is same in proportion. However, lower rates of devolatilization were seen for coconut shell blends compared to pure coke, which is expected, shows the slowest rate. Coconuts shells blend showed a lower rate due to its structure transformation (particle fragmentation) during devolatilization compared to palm shells blend. This phenomenon occurred very fast at high temperature where the devolatilization rates would decreased. Different behaviors are shown by coke and its blends with agricultural wastes, with larger fractions from the blends being released as volatiles during the combustion process. This high amount of volatile was seen to occur over a relatively short time and is believed to influence the time required for complete combustion compared to coke. The estimated weight loss rate for P3 blend is the fastest, while 100% coke showed the lowest decomposition rate. On the other hand, palm shell blends show a continuous increase with increasing palm shell proportion in the blend. The greater weight loss of palm shell (and lower residual mass) could be attributed to its higher cellulose content, which is less stable compared 193

223 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends to lignin. Therefore, cellulose might have a more dominant influence on devolatilization compared to lignin. Previous studies on the pyrolysis of agricultural waste (Antal, M. J., Jr. and Varhegyi, G., 1995), agrees that lignin gives a higher char yield than cellulose or hemicellulose. However, in the case of rubber, the long hydrocarbon structure with higher volatile matter would influence the rate of devolatilization compared to agricultural waste materials. Moreover, such a behavior is expected considering the high moisture content in the agricultural wastes retarding the VM released. However, the presence of potassium, iron oxide and sodium present in agricultural waste, is known to be a strong catalyst in gas phase reactions (Lang, R. J. and Neavel, R. C., 1982, Miura, K., Hashimoto, K. and Silveston, P. L., 1989, Yunos, N. F. M., Zaharia, M., Ahmad, K. R. et al., 2011), promoting a high devolatilization rate compared to coke alone. 194

224 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends Residual Mass, % MC P1 P2 P3 R3 Residual Mass, % MC C1 C2 C3 R Rate constant (K, s -1 x 10-3 ) (a) Time, sec (b) P1 MC Volatile matter (%) P2 R3 P3 Rate constant (K, s -1 x 10-3 ) (c) Time, sec (d) R3 4 C2 C3 2 C1 MC Volatile matter (%) Figure 5-23 (a) and (c) Comparison of residual mass (%) in TGA with time for 100% MC and its mixtures with palm and coconut shells (b) and (d) variation of rate constant of the agricultural blends with volatile matter at 1200ºC in N 2 atmosphere Where: [MC =100% coke; P3 = palm shell blends; C3 = coconut shell blends; R3 = rubber blends]. 195

225 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends Furthermore, the difference is also attributed to the difference in the strength of the molecular structure of the fuels. The polymers cellulose, hemicellulose and lignin are linked together with relatively weak ether bonds are less resistant to the heat at low temperatures (Blazej, A. and Kosik, M., 1993). In contrast, the coke structure, which mostly comprises dense polycyclic aromatic hydrocarbons, are more resistant to the heat (Blazej, A. and Kosik, M., 1993, Smith, L. H., and, S. L. D. and Fletcher, 1994). The coke considered for the present study was expected to have a high aromatic content. As a consequence, a small amount of fragmented polycyclic aromatic compounds would be expected to result from the immobile phase. Hence the mass loss was lower (high residual mass) than from pyrolysis of blends involving agricultural waste samples. The lower amount of volatiles, the slowest rate of devolatilization and the highly aromatic structure present in coke leads to lower mass loss and little change in structure. Table 5-5 summarizes the intramolecular bonds present in the samples used in the present study along with the corresponding energies required to break these bonds. Table 5-5 Type of molecular bonds and bond energies required to break the fuels considered in this study (Blazej, A. and Kosik, M., 1993, Smith, L. H., and, S. L. D. and Fletcher, 1994) Samples Bonds Bond Energies (kj/mol) Metallurgical coke, MC C=C >1000 Cellulose, hemicellulose and lignin (agricultural waste/woody/ biomass) R-O-R ~

226 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends Quantitative determination of the combustion performance of coke and its corresponding agricultural waste blends indicated a direct correlation showing an increase in burnout with increasing agricultural waste proportions in the blend. Higher burnout values recorded by the palm shells blend over the coconut shells blend were observed. (Figure 5-24) Burnout, % Palm shells Coconut shells 0.0 MC Blend 1 Blend 2 Blend 3 Palm & Coconut shells content in the blend Figure 5-24 Combustion performances of palm shell-coke and coconut shell-coke blends (Blend 3* = Blend P3 and Blend C3) Looking at the decomposition pattern of palm shells and coconut shells, it may be assumed that the slightly higher burnout of the palm shells over the coconut shells may be due to the direct conversion of lignocellulosic structure, changes in surface area and inorganic material that is present in the agricultural waste materials. 197

227 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends The development of porosity in the residual chars seems to depend on the amount of volatile matter removed and the subsequent structural changes of the residual carbon. Agricultural waste materials as a raw have high levels of volatiles. When used in conjunction with coke matrix, structural and physical changes might occur because of the volatiles released following combustion. Figure 5-25 shows that the change is more significant in the case of palm blends compared to coconut blends, while coke develops the least change. This will support the observations that for palm shell the combustion show an increase (continuous) (see thermogravimetric analysis in Figure 5-22). In the case of coconut shell, the initial increase in combustion efficiency is supported by the first ΔSA which is appreciable and rises. However the combustion was decreased afterward for C3 blends due to particle fragmentation compared to palm shell samples discussed earlier. The changes in the original structure of the matrix of the raw material may be attributed to the cross-linking of the reactive points of the cylinders brought about by the disruption of the original polymeric structure of the precursor material (i.e. the cellulosic and lignin units of the raw material) and the subsequent reconstitution of a new matrix structure during the devolatilization as observed by Byrne and Marsh (Marsh, H. and Reinoso, F. R., 2006). 198

228 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends 100 Palm shells 80 P2 P3 Δ Surface Area (%) P1 C2 C3 20 C1 Coconut shells 0 MC Agricultural wastes content in the blend (%) Figure 5-25 The changes in micropore surface area for agricultural wastes content in the blend Scanning electron microscopy was used for a quantitative determination of the residual char morphology. Figures 5-26 (a) (c) show the structure of 3 samples blended with coke develop a high porous structure. For palm shells sample, the pore structures are opening up while coconut shells showed that cylinders were still present, even though the interiors are clearly being consumed. Thus, after the initial consumption of the interiors, the cylinders are not providing any further opportunities for increase, as it is difficult to find too many edges, and the surface area itself keeps decreasing, compared to palm shells (Figure 5-26 (b)) where we clearly saw the entire structure uniformly opening up and creating significant changes in surface area (Figure 4-13 in Chapter 4). 199

229 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends (a) (b) (c) Met. coke Palm shells Coconut shells Figure 5-26 SEM micrographs of (a) MC, (b) palm shells and (c) coconut shells collected after reaction in the DTF at 1200 C in atmosphere of 20% O 2 and 80% N 2, polished section of the residual particle x1000 A good agreement between the surface area (porosity) and morphological changes of the agricultural/coke blends with combustion performance and devolatilization is established in this study. Similar correlations between physical parameters of the carbonaceous materials and their derived burnout were observed in the literature (Sahajwalla, V., Zaharia, M., Kongkarat, S. et al., 2009, Zaharia, M., Sahajwalla, V., Kim, B.-C. et al., 2009). These researchers found that the changes in morphology would enhance the combustion performance at similar conditions. 200

230 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends 5.3 Summary Investigation of comparative combustion characteristics and performances of coke and its blends in different proportions with palm and coconut shells have been investigated. The effect of addition of agricultural waste on the combustion behavior of its blends with metallurgical coke are summarize below: 1) Fundamental pyrolysis and combustion behaviors for two types of agricultural wastes were tested by a TGA-MS. At temperatures above 1000 ºC, the gas products evolving from coconut and palm shells pyrolysis measured by MS showed CO, CO 2 and H 2 were released. 2) Increased combustion efficiency is observed with increasing agricultural wastes proportion in the blend. 3) The effect on the structure of the particles, such as surface accessibility developed following gas phase reactions in the DTF also influences the combustion efficiency of coke/palm shell and coke/coconut shell blends. 4) Good correlations were found between the extent of mass burnout of the blended residues and that of their pore, therefore supporting the dominant influence of the carbon surface accessibility (porosity) and blend components. The SEM photomicrographs illustrated a good agreement with the trends observed in measured surface area in the carbonaceous matrix. 5) Minerals, such as iron and alkali metals in agricultural waste contribute to the mineral matter in the blends and influence combustion performances. 6) Combustion efficiency depends on the amount of mineral matter coupled with the rate of release of VM products and also with the effect on the structure of the particle formed. The findings clearly indicate that blending coke with recycled agricultural waste materials would enhance overall combustion efficiency. 201

231 Chapter 5: Combustion & Structural Transformations of Coke/Coconut Shell Blends 7) Palm shell blends developed higher devolatilization rates when compared to coconut shell blends where the lower devolatilization rates appear to affect the structure. The difference in lignocellulosic structure of the materials is strongly connected to their kinetic behavior. 202

232 Chapter 6: Conclusions Gas Phase Reactions CHAPTER 6 6 Conclusions Gas Phase Reactions Studies The combustion characteristics and performances of coke and agricultural waste blends in different proportion (palm and coconut shells) have been investigated. The aim of this study is to understand the fundamentals of gas phase reactions and structural transformations. TGA (N 2 atmosphere) and DTF (20% O 2 and 80% N 2 gas mixture) are used for gas phase reaction purposes. The thermal decomposition (devolatilisation) and char oxidation of MC and agricultural waste materials were studied. Residual mass was found to decrease with increasing the agricultural wastes content in the blend. This behavior might be due to the high volatile content of agricultural waste materials when compared to coke alone. The intensity of volatile release was quantitatively compared in terms of devolatilization rates. Coconut shell blends show a steady reaction rate with increasing concentration of coconut in the blend, while palm shell blends show an increase as the proportion of the palm shell in the blend rises. At temperatures above 1000 ºC, the gas products evolving from palm and coconut shells pyrolysis measured by TGA-MS showed CO, CO 2 and H 2 as the main gases. These were attributed to the lignocellulosic structure present in the in agricultural waste allowing a continuous gas release for participation in the subsequent carbon/slag interactions. The combustion performance of coke and its blends with agricultural wastes was quantified. A higher burnout was estimated for the palm/coconut blends when compared to coke alone. 203

233 Chapter 6: Conclusions Gas Phase Reactions Surface area measurements and qualitative morphological characterization of the agricultural/coke blends through SEM were conducted. The changes in surface area, before and after gas phase reactions, were estimated and palm shell blends showed a larger increase in surface area compared to coconut shell blends, while the changes occurring in the coke particles are marginal. Particle fragmentation is believed to hinder pore development in coconut shell while an agreement with the trends observed in the rate of devolatilization is consistent and support with an increase in combustion performance of palm shell blends. A highly amorphous structure characterized the palm and coconut shell wastes with broad diffuse spectra and a significant amount of highly disordered material. The changes in surface area are more significant when the proportion of agricultural waste materials increases. SEM micrograph supported the BET measurement showing enlarged pores develop in the agricultural waste materials. Palm shell s cell structures were seen to open up to a significant extent and structural changes were observed accompanied by higher surface area. In contrast, the structural transformations that occur in the coconut shell blends retain the cylindrical cell structures, which show a lower surface area compared to palm shell blends. Minerals, such as iron and alkali metals in agricultural waste also contribute to the mineral matter in the blends and influence combustion. The combined modification of pore and cell structure of agricultural waste materials (palm and coconut shells) are shown to contribute to their structural transformations, which results in improvement in combustion efficiency in high temperature processes. 204

234 Chapter 7: Discussions on Carbon/Slag Interactions CHAPTER 7 7 High Temperature Reactions: Carbon/Slag Interactions Results and Discussion The sessile drop method was used to investigate carbon/slag interactions at 1550 ºC in a horizontal tube furnace under inert atmosphere (1 L/min Ar, 99.99% purity). These studies include the off-gas generations, interfacial phenomena and slag foaming behavior of EAF slags. The carbonaceous materials used were metallurgical coke, palm and coconut shells char and the blends with different proportions together with an EAF rich iron oxides slag. The agricultural waste/coke blends were initially subjected to rapid gas reaction in a drop tube furnace (DTF) in an atmosphere of (20% O 2 and 80% N 2 ) gas mixture at temperature of 1200 ºC while agricultural wastes were devolatilized at 450 ºC, as explained in Chapter 3. Coke was used without prior treatment due to low amount of volatile present in its matrix. Later on, these carbonaceous materials were subjected to carbon/slag reactions. These carbon/slag interactions were monitored continuously throughout the experimental run by video recording of reaction assembly as well as gas generation. When a carbonaceous material is injected in the EAF furnace it experiences an initial devolatilization, followed by combustion and later on the residues interact with the available slag. The iron oxide rich EAF slag reacts with the available carbon as a result metallic iron is produced. The reaction between carbon and FeO in the slag leads to the formation of CO and CO 2 gases which are entrapped in the slag phase and subsequently released. The characteristics of the carbonaceous materials in the residual chars obtained after devolatilization and subsequent combustion reaction will influence carbon/slag 205

235 Chapter 7: Discussions on Carbon/Slag Interactions interactions. Thermogravimetric analysis (TGA) has been performed on the raw materials and the devolatilization rates of coke and coke/agricultural waste blends were estimated. The rates of devolatilization of the carbonaceous materials studied in this present work will affect the gases released, influencing the carbon/slag interactions. Medium rates of volatile matter evolution will lead to steady gas releases which might be available during subsequent contact with the EAF slag allowing the entrapment of gases over a longer period of time, while higher rates of gas generations might lead to poor gas entrapment allowing a fast passage through the slag (Figure 7-1). high Fast gas release Poor foaming Carbonaceous material Rate of devolatilization medium Steady gas release Stable foaming low Low gas release Poor foaming Figure 7-1 Schematic diagram illustrating the effect of carbonaceous materials rate of devolatilization on slag foaming The initial rapid gas phase reaction of coke/palm shell blends were achieved during passage through DTF. This resulted in volatiles still being present in the blends as unburned hydrocarbons. The structure of the resulting carbon also plays a very important role in the subsequent carbon/slag interactions. An ordered graphitic structure has a low degree of reactivity; while a more disordered material contains active sites \ and volatiles (hydrocarbons) are very reactive. The reactions occurring at the slag/carbon interface (1550 ºC) are expected to be affected by the presence of an increased level of hydrocarbons. These hydrocarbons 206

236 Chapter 7: Discussions on Carbon/Slag Interactions (C n H m ) are expected to further decompose into carbon and hydrogen in Eq (Dankwah, J. R., Koshy, P., Saha-Chaudhury, N. M. et al., 2011). The hydrocarbon could also act as a sink for CO 2 gas to produce CO and H 2 in Eq When put in contact with an EAF rich iron oxide slag, the presence of FeO leads to reduction reaction depending on the reducing agents including C, CO and H 2 released at high temperature. The reduction of FeO by hydrocarbons leads to the formation of CO, CO 2 and H 2 O. These gases can later be found at slag/metal interface allowing the entire bath to become foamy. A fluctuation in volume is dependent on the rate of gas evolution, the size of gas bubble as well as physical and chemical properties of the slag. Moreover, CO 2 can be produced from reaction with H 2 O based on auxiliary reaction and H 2 O with C to produce CO from cracking the hydrocarbon (Dankwah, J. R., Koshy, P., Saha-Chaudhury, N. M. et al., 2011, Demirbas, A., 2002). The reactivity of carbon is also related to the porosity of structure and more porous structure will result in a larger exposed area per unit mass and this will further enhance gasification of carbon. The gases released from the carbonaceous material will assist with carbon/slag reactions including FeO reduction. The interactions between carbons based materials with EAF rich slags at high temperature leading gas generation. Part of the gases generated remains trapped in the slag for a certain period of time depending on the process parameters and slag properties. Interfacial reactions between solid carbon and the slag phase determine their degree of interaction. 207

237 Chapter 7: Discussions on Carbon/Slag Interactions 7.1 Influence of Carbonaceous Material on Carbon/Slag Interactions For a better understanding of carbon/slag interactions, off-gas measurement enable determination of the amounts of CO and CO 2 formed during carbon/slag interactions and FeO reduction as a function of time. In the present study, palm shell sample is chosen for comparison with conventional material (coke) due to its capability to maintain the slag volume and easy to quantify the sizes. While, coconut shell reacted very fast with slag that makes hard to quantify the slag volume ratio and contact angle measurement because it kept sinking into substrate. The gas entrapment phenomena has been quantified using sessile drop arrangement and the novel processing software (Khanna, R., Mahfuzur, R., Richard, L. et al., 2007) for estimation of slag volume ratios as a function of time and quantitative measurement of the reduced metal formation and gas behavior by using optical microscopy and SEM. These carbonaceous materials will react with an EAF rich iron oxide slag, containing 34.9 % iron oxide. The complete slag composition is presented in detailed in Table 3-8. When the samples assembly attained the desired temperature, changes in the slag volume were observed. Fluctuations were clearly visible and were attributed to subsequent formation and gas release. Depending on the chemical nature and composition of the carbonaceous material different volume ratio trends have been observed through in-situ high temperature determination of slag volumes as a function of time are presented in the following sub-sections. 208

238 Chapter 7: Discussions on Carbon/Slag Interactions Off-gas (CO, CO 2 ) Generations Off-gas generation (CO and CO 2 ) during the high temperature reactions at 1550 ºC between the carbonaceous substrate and the slag (predominantly reduction of iron oxide by carbon) was determined by using IR analyser results as shown in Figure 7-2. Figures 7-2 (I and II (a)) respectively provide volume concentrations (ppm) in terms of CO and CO 2 gases resulting from the interactions of the EAF slag with coke and palm char samples. Metallurgical coke showed significant level of carbon/slag interaction with the CO gas concentration emitted reaching ppm after 100 s of contact (Figure 7-2 (I, a)). The CO gas concentration from palm shells char/slag showed a sharp rise, attaining a value of ppm in less than 100 seconds, increased slowly and the reaction slowing down after 300 s (Figure 7-2 (II, a)). The CO concentration decreased gradually which attributed to the complex structure found in the palm shells breaking down at a slower pace. The CO 2 volume was seen to be significantly lower (~200 ppm) compared to coke (13000 ppm) and this might be due to the hydrogen content present in the agricultural wastes char (~ 6 %) which acts as a CO 2 sink (Dankwah, J. R., Koshy, P., Saha-Chaudhury, N. M. et al., 2011). Figure 7-2 (I and II (b)) shows the cumulative volumes of gases emitted (CO, CO 2 ) from a metallurgical coke/slag as well as from the palm char/slag. In general, there is a time delay observed of about seconds involved before the Infrared analyser starts recording the CO and CO 2 data. The product gases take this time to travel from the reaction site to the Infrared analyser. This was initially ascertained by passing Argon gas through the furnace tube and the observing time needed for the IR to respond. The off-gas results for metallurgical coke were significantly different from palm char for both CO and CO 2 levels. After 100 s reaction at 1550 C, the cumulative volume of CO gas emitted from the metallurgical coke was seen to increase up to 0.5 x 10-4 mol with CO 2 levels = 0.25 x 10-4 mol. On the other hand, CO gas emitted from palm char was nearly 3 times higher (1.15 x 10-4 mol) than the corresponding result from coke. The total amount of CO 2 evolved gases from palm char was lower than coke indicated that agricultural waste has potential to reduce the CO 2 emissions. These results were in 209

239 Chapter 7: Discussions on Carbon/Slag Interactions good agreement with TGA-MS analysis where more CO and CO 2 were released at high temperatures (Figure 4-3). A higher content of oxygen present in the agricultural waste was also an important factor affecting the high temperatures reactions of agricultural wastes char with EAF steelmaking slags. Palm char contain high oxygen level where the gas will release as CO and CO 2 during the interaction with slags. (I) Metallurgical coke/ slag Gas generated (CO/CO2), ppm (a) CO CO Cumulative volume of gases, x 10-4 (mol) (b) CO CO Time, s Time, s (II) Palm char/ slag Gas generated (CO/CO2), ppm (a) CO 5000 CO Cumulative volume of gases, x 10-4 (mol) 1.4 (b) 1.2 CO CO Time, s Time, s Figure 7-2 (I) Metallurgical coke/slag and (II) Palm char/slag with (a) generated gas concentrations (ppm) in terms of CO and CO 2 gases and (b) the cumulative volume of gases (mol) of CO and CO 2 210

240 Chapter 7: Discussions on Carbon/Slag Interactions A number of reduction reactions involving oxides in EAF slags, ash impurities, and oxidation of carbon are expected to take place in the interfacial reactions. In Figure 7-3, the number of moles of carbon and oxygen removed during metallurgical coke/slag and palm char/slag interactions have been plotted. Extensive gas generation was observed from palm char in contact with slag which is predominantly CO. Metallurgical coke showed slightly lower levels of gas generation. Moles of oxygen and carbon removed increased as a function of temperature indicating enhanced reactivity and iron oxide reduction. 211

241 Chapter 7: Discussions on Carbon/Slag Interactions Carbon removed x10-5,moles (a) MC PC Time, s Oxygen removed x10-5,moles (b) MC PC Time, s Figure 7-3 Total number of moles (a) carbon and (b) oxygen removed from the metallurgical coke (MC) and palm char (PC) substrate in contact with a slag at temperature 1550 ºC 212

242 Chapter 7: Discussions on Carbon/Slag Interactions Figures 7-4 (a) (d) respectively shows the gas (CO+CO 2 ) generation for palm char, metallurgical coke and its blends with palm shells during interactions with slag at 1550 C. From the total gas (CO+CO 2 ) generated graph, it can be observed that, with increasing palm shells concentration in the blends, the total gas generation was decreased. The might be due to the lower of CO 2 generated from palm shell blends where the CO 2 generated showed a lower level than coke. Metallurgical coke showed a significant level of carbon/slag interaction with the CO gas emitted reaching ppm after 100 s of contact. CO gas from palm shells/coke blends was higher than that from metallurgical coke substrates. In the initial stages of interaction, the highest amounts of total cumulative of moles generated (CO+CO 2 ) gases were emitted by P1 blends (1.2 x 10-5 moles) in Figure 7-5 (b). The CO 2 gas was seen decreasing with increasing the palm shell concentration in the blends indicating less of CO 2 emissions (Figure 7-5 (b) (d). 213

243 Chapter 7: Discussions on Carbon/Slag Interactions Gas generated (CO/CO2), ppm (a) MC CO CO Time, s Gas generated (CO/CO2), ppm (b) P1 blends CO CO Time, s Gas generated (CO/CO2), ppm (c) P2 blends CO CO Time, s Gas generated (CO/CO2), ppm (d) P3 blends CO CO Time, s Figure 7-4 Generated gas concentrations, CO+CO 2 (ppm) as a function of time and carbon based materials (a) metallurgical coke, (b) P1 blends, (c) P2 blends and (d) P3 blends 214

244 Chapter 7: Discussions on Carbon/Slag Interactions Figure 7-5 (a) (d) shows the total amount of removed gasses produced from metallurgical coke and palm shell blends when put in contact with an EAF slag. In general, there is a time delay observed about 40 to 60 seconds involved before the Infrared analyser starts recording the CO and CO 2 gases. The product gases take this time to travel from the reaction site to the Infrared analyser. This was initially ascertained by passing argon through the furnace tube and observing the time needed for the IR to respond. As a result, all the graphs show the reduction starting after a delay of 40 to 60 seconds. No significant differences were observed in the time of initiation of the evolution of gases with the variation of carbonaceous material or slag employed in the present study. The variations in amounts of removed oxygen, i.e. the amount of CO and CO 2 show a decrease with increasing palm shell content in the blends. By direct observations of the graphs in Figure 7-5, it can be observed that the lowest amount gas emissions are obtained from the raw coke, increasing when palm shells were partially replaced the carbonaceous substrate as in P1 blends (Figure 7-5 (b)). 215

245 Chapter 7: Discussions on Carbon/Slag Interactions Total (CO+CO2) removed x 10-5, moles (a) 100% MC 100% PC Total (CO+CO2) removed x 10-5, moles (b) 100% MC P1 blends Time, s Time, s Total (CO+CO2) removed x 10-5, moles (c) 100% MC P2 blends Total (CO+CO2) removed x 10-5, moles (d) 100% MC P3 blends Time, s Time, s Figure 7-5 Total cumulative number of moles of gas generated (CO+CO 2 ) as a function of time and carbonaceous material used 216

246 Chapter 7: Discussions on Carbon/Slag Interactions Contact Angle Measurements The slag/carbon interactions of metallurgical coke, palm char and palm shells/coke blends were measured at 1550 C with EAF slags are presented in Figures 7-6 (a) to (d) in terms of contact angles. For metallurgical coke, the initial contact angle was 107 and then it increased up to 124 at 150 s. Then the contact angle decreased to 102 at 1000 s, indicating a minor improvement in the wetting behavior with time (Figure 7-6 (a)). For palm shell/coke blends, the wettability was seen to improve slightly with decreasing levels of concentrations of palm shells. The slag droplet showed high fluctuating of contact angles during the experimental run. For Blend P1 showed dynamic non-wetting behavior with the contact angle was higher than the corresponding values for metallurgical coke (Figure 7-6 (b)). The slags also showed a poor wettability on Blend P2 where high contact angles rapidly increased to 122 in 10 min (Figure 7-6 (c)). While the contact angles for Blend P3 was 118 for 100 s and decreased to 100 at 10 min (Figure 7-6 (d)). These fluctuations are believed to have been caused by gases generated during carbon/slag interactions, their trapping and subsequent release from the slag droplet. Based on thermodynamic considerations, the wetting line in a wetting carbon/slag system will move only if there is a net decrease in the free energy at the carbon/slag interface. This decrease in the free energy during wetting is a result of interfacial reaction and lowering of interfacial tension. According to Pask (Pask, J., 1987 ), the solid carbon substrate is an active participant in the reaction and the free energy changes during the reactions contribute to reduction in the interfacial energy. However, the initial contact angle being high, generally >100º, the decrease in the contact angle is to be expected where the volume expansion will control the measurements indicating high gas entrapments. Concluding from the measured values of the contact angles, there were observed a positive influence of FeO reduction reaction on improving wettability in palm char with higher amount of iron oxide in the ash, which is influenced by composition of carbon content, slag composition (FeO content) and temperature. 217

247 Chapter 7: Discussions on Carbon/Slag Interactions (a) (b) Contact angle ( ) Contact angle ( ) MC PC Time, s 90 MC P Time, s (c) (d) Contact angle ( ) Contact angle ( ) MC P Time, s 90 MC P Time, s Figure 7-6 Variation of contact angle with time for carbonaceous substrate (a) palm char (PC), (b) P1 blends, (c) P2 blends and (d) P3 blends in contact with slag at 1550ºC 218

248 Chapter 7: Discussions on Carbon/Slag Interactions High Temperature, In-situ Observations The sessile drop method was used to investigate the gas entrapment in metallurgical slags as a result of carbon/slag interactions. The changes during carbon/slag interactions were monitored continuously throughout the experimental run by video recording and monitoring gas data. In order to qualitatively demonstrate the gas entrapment phenomenon a few representative dynamic wetting images of the slag droplet in contact with palm char, metallurgical coke and its blends with palm shell are shown in Figures 7-7 to 7-7 (a) (e). Gas bubbles are predominantly visible in the video recording and are generated throughout the interface just after melting. In fact, the generated gas spreads into the slag droplet increasing its volume. The volume decreases when the gases are released. The entire sequence of events is presented as a function of time and t=0 represents the initial melting stage. The size of the slag droplet in contact with metallurgical coke (Figure 7-7 (a)) did not show a significant variation with time during the initial stages. After 10 minutes of reaction, the slag droplet fluctuated to a small extent in volume and maintained a reasonably spherical shape. The fluctuations in droplet size are associated with the generation and subsequent release of gases. In the case of palm shell char, the volume increase and associated were observed fluctuations throughout the experiment, indicated a sustained improvement in slag volume (Figure 7-8 (b)). The EAF slag reacting with P1 blends increased in volume and attained a spherical shape. Fluctuations appeared throughout the experiment, showing an improvement in volume when compared to coke alone (Figure 7-9 (c)). When the slag reacted with P2 blends (Figure 7-10 (d)) an initial droplet of high volume was seen, followed by a fast released of gases and immediate decrease in droplet. The fluctuations continue up to 15 min of contact. A different behavior was seen when the EAF slag reacted with the carbonaceous substrate, P3 blends. An increase in volume was clearly visible due to a continuous/sporadic evolution of CO gas bubbles within the slag droplet. Fluctuations appeared throughout the experiment, showed an improvement in volume compared to 219

249 Chapter 7: Discussions on Carbon/Slag Interactions coke alone. (Figure 7-10 (e)). Images were captured for every 5 s during the initial stages of contact and every 30 s during later periods. 220

250 Chapter 7: Discussions on Carbon/Slag Interactions (a) Metallurgical Coke (MC) t = 60 sec t = 120 sec t = 600 sec t = 900 sec Palm char (PC) (b) t = 60 sec t = 120 sec t = 600 sec t = 900 sec Figure 7-7 High temperature photographs of slag droplets in contact with (a) 100% MC and (b) 100% Palm char at 1550 ºC as a function of time 221

251 Chapter 7: Discussions on Carbon/Slag Interactions (a) Metallurgical Coke (MC) t = 60 sec t = 120 sec t = 600 sec t = 900 sec P1 blends (c) t = 60 sec t = 120 sec t = 600 sec t = 900 sec Figure 7-8 High temperature photographs of slag droplets in contact with (a) 100% MC and (c) P1 blends at 1550 ºC as a function of time 222

252 Chapter 7: Discussions on Carbon/Slag Interactions (a) Metallurgical Coke (MC) t = 60 sec t = 120 sec t = 600 sec t = 900 sec P2 blends (d) t = 60 sec t = 120 sec t = 600 sec t = 900 sec Figure 7-9 High temperature photographs of slag droplets in contact with (a) 100% MC and (d) P2 blends at 1550 ºC as a function of time 223

253 Chapter 7: Discussions on Carbon/Slag Interactions (a) Metallurgical Coke (MC) t = 60 sec t = 120 sec t = 600 sec t = 900 sec P3 blends (e) t = 60 sec t = 120 sec t = 600 sec t = 900 sec Figure 7-10 High temperature photographs of slag droplets in contact with (a) 100% MC and (e) P3 blends at 1550 ºC as a function of time 224

254 Chapter 7: Discussions on Carbon/Slag Interactions Slag Foaming Carbon/slag interactions were performed in high temperature furnaces with monitoring facilities to observe and quantify during the reactions. All experimental runs were recorded up to 30 minutes, but due to insignificant changes in the slag volume after 1000 seconds no more data points are presented. Gas hold-up in the slag droplet was measured in terms of V t /V 0, where V t is the volume of slag droplet at time, t and V 0 is the initial slag volume (Figure 7-11). At the initial slag melting stage is represented by t = 0, where the volume ratio is 1. Changes are observed corresponding to CO and CO 2 generation and entrapment, as an increased droplet volume. After a certain period of time, gas release has reduced considerably and V t /V 0 stabilizes. Figure 7-11 (a) presents the volumes of CO and CO 2 generated at any time, t for 100% palm char and 100% coke, monitored by the IR gas analyzer. The instantaneous gases (CO+CO 2 ) generated peaked around ppm for coke and the evolution of CO and CO 2 gases released from palm char slowly decreased with time (Figure 7-11 (a)). These results indicate that the high levels of gas generation from coke/slag released than palm char/slag resulting in faster gas escaping after 300 seconds of reactions. From Figure 7-11 (b), the palm shells char has a significant variation in volume ratios, with the volume ratio fluctuating between 1.0 and 1.3 up to 100 s and between 1.0 and 1.2 after 200 s. This indicates that the decrease in slag volume took longer when palm shell char was used, suggesting significant levels of gas entrapment and subsequent release during carbon/slag interactions. Metallurgical coke has the volume ratio in the range of 0.8 to 1.0. After 600 s the volume ratio decreased to 0.7 indicating a lower extent of gas entrapment by the slag. The results suggest the potential of palm shells char injection, to give immediate and observable differences to slag foaming, increasing foam volume. 225

255 Chapter 7: Discussions on Carbon/Slag Interactions (CO+CO 2 ), ppm (a) 100% MC 100% PC Time, s Volume ratio, V t /V (b) 100% MC 100% PC Time, s Figure 7-11 (a) Instantaneous gases (CO+CO 2 ), ppm, generated and (b) Volume ratio of 100% Metallurgical coke (MC) and 100% Palm char (PC) with slag with respect to reaction time 226

256 Chapter 7: Discussions on Carbon/Slag Interactions Figure 7-12 (b) reveals the volume ratio V t /V 0 of the slag reacting with metallurgical coke and P1 blends. The slag volume ratios for P1 blends showed significantly different trends with much higher level of gases in the slag volume in the first 100 seconds of reaction, sustained for 400 seconds and continue fluctuated after 600 seconds (1.8) where the gas generated from P1 blends showed high gas released, ppm (Figure 7-12 (a)). This indicates that the decrease in slag volume took much longer than coke suggesting significant levels of gas entrapped which was subsequently released. P2 blends showed a volume ratio comparable to P1 blends (Figure 7-13 (b)). The volume ratio showed 1.4 for first 100 seconds due to the gas trapped within the slag, dropping down to a value of 0.8 after 300 seconds. After 700 seconds, the fluctuation resulted in high volume ratio 1.7 and dropping down afterwards with the release gases and consumption of carbon. It was seen that the gas generated from P2 blends are lower than P1 blends (Figure 7-13 (a); ppm). P3 blends showed significantly lower levels of gas entrapment compared to other blends. The volume ratio (V t /V 0 ) showed large fluctuations, with the drop growing up to 1.9 due to the gas trapped within the slag droplet and then dropping down to a much smaller size with the release of the gas. This trend continued several times during the first 200 s of contact; the volume ratio eventually settled down to 1.1 after 1000 s (Figure 7-14 (b)). For palm shell blends, slag foaming behavior was found to be similar for all cases where after 600 seconds of reaction, the volume sizes increased (Figures 7-12 and 7-14). This might be due to the presence of small bubbles that hold up all through the period of time reactions. P1 blends showed the best gas entrapment and foaming behavior compare to coke and other blends. When more proportion of palm shell in the blends (P2 and P3 blends), the volatile matter was increase compared to P1 blends. Volatile matter was expected to loss during experimental where the substrate needs to stay in the cool zone before purge into hot zone. Even in the cool zone, the temperature is reached 1450 C. At this temperature, palm shells blend was burn before the exactly time to interact with slag at 227

257 Chapter 7: Discussions on Carbon/Slag Interactions 1550 C. Thus, the gas released from the carbon/slag interactions for P2 and P3 blends was expected to be low than P1 blends. A stable slag volume is strongly related to an optimum rate of gas generation where 100% palm char shows a steady behavior compared to the three blends. Higher gas generation leads to higher flow rate of gas with increase velocities. When the rate of gas generation is too high, the slag is unable to hold the gases and the foaming phenomenon is affected. 228

258 Chapter 7: Discussions on Carbon/Slag Interactions (CO+CO 2 ), ppm (a) 100% MC P1 blends Time, s (b) 100% MC P1 blends Volume ratio, V t /V Time, s Figure 7-12 (a) Instantaneous gases (CO+CO 2 ), ppm, generated and (b) Volume ratio of 100% Metallurgical coke (MC) and P1 blends with slag with respect to reaction time 229

259 Chapter 7: Discussions on Carbon/Slag Interactions (CO+CO 2 ), ppm (a) 100% MC P2 blends Time, s (b) 100% MC P2 blends Volume ratio, V t /V Time, s Figure 7-13 (a) Instantaneous gases (CO+CO 2 ), ppm, generated and (b) Volume ratio of 100% Metallurgical coke (MC) and P2 blends with slag with respect to reaction time 230

260 Chapter 7: Discussions on Carbon/Slag Interactions (CO+CO 2 ), ppm (a) 100% MC P3 blends Time, s (b) 100% MC P3 blends Volume ratio, V t /V Time, s Figure 7-14 (a) Instantaneous gases (CO+CO 2 ), ppm, generated and (b) Volume ratio of 100% Metallurgical coke (MC) and P3 blends with slag with respect to reaction time 231

261 Chapter 7: Discussions on Carbon/Slag Interactions Figure 7-15 has been plotted for a quick visual comparison of the instantaneous gases (CO+CO 2 ), ppm generated when EAF slag put in contact with coke, palm char, P1 blends, P2 blends and P3 blends. The trend observed in the gas generation data for all samples are expected based on the amount of residual gases available due to palm shells addition and increased surface area following initial gas phase reactions. While the amount of gases generated for coke peaked at a value around ppm, P1 blends attained ppm followed by P2 (59000 ppm) and P3 blends (52000 ppm). The lower gas generated was seen with increasing the palm shell in the blends due to lower CO 2 gas generations. This might be due to the slower reactivity of CO 2 with coke where higher levels of CO 2 are left behind which does not help with gas generation of CO. Whereas for 100% palm char, the CO 2 reactivity was greater as FeO reduction continues further generated CO 2 and converted to CO. The gas was observed over an extended period time (Figure 4-3). Thus, it can conclude that 100% palm char released the gas steady and continuously; this expected due to slowly breaking of the lignocellulosic structure for participation in the subsequent carbon/slag interactions. 232

262 Chapter 7: Discussions on Carbon/Slag Interactions (CO+CO 2 ), ppm % MC 100% PC P1 blends P2 blends P3 blends Time, s Figure 7-15 Instantaneous gases (CO+CO 2 ), ppm generated from 100% metallurgical coke (MC), 100% Palm char (PC), P1 blends, P2 blends and P3 blends reacting with slag as a function of time 233

263 Chapter 7: Discussions on Carbon/Slag Interactions Interfacial Phenomena - Optical and SEM Studies The sample was pushed in the hot zone and the counter was started after the slag droplet showed the first signs of melting. Throughout the experiment the reaction time was closely monitored. The slag droplet resting on the carbonaceous substrate was quenched by withdrawing the supporting tray into the cold zone of the furnace after fixed periods of time, being: 60, 120, 600 and 900 seconds. The optical images exhibited different characteristics depending on the carbon based material used as substrate. Optical micrographs of the quenched samples during reactions at different times were investigated to develop an understanding of the role of the gas entrapment on slag foaming (Figure 7-16 (a) and (b)). Extensive reduction of iron oxide (Figure 7-16 (a)) was observed after 1 min of contact when the EAF slag was put in contact with coke. A large number of molten iron droplets (shiny round particles) were also observed dispersed throughout the slag matrix sitting either at the carbon/slag or gas/slag interface. Gas bubbles trapped within the slag droplet were generally quite small in size. After 10 min of contact, the reduced iron in slag droplet was seen to be precipitated on the substrate. A close inspection of the cross section of the slag/palm char sample revealed a different behavior. From Figure 7-16 (b), we observe a large numbers of small diameter gas bubbles were dispersed throughout the entire slag volume. Thus, overall slag volume is dictated by large number of small bubbles. With increasing time while the large gas bubbles were still present, their numbers were much reduced. The evolution of gases was seen clearly and was due to the reaction of carbon with the FeO in the slag droplet, resulting in the release of CO and CO 2 gases. The subsequent entrapment and released these gases contribute to volume fluctuations of the slag droplets; Fe was seen deposited at the slag/carbon interface. At the latter stages of the reaction, as the reduction was approaching completion where the molten metal resting at the interface was appeared as a bright round shiny droplet. Significantly different behavior (Figure 7-17 (c)) was observed when the same EAF slag was put in contact with the residue resulting from P1 blends. From the figure, we 234

264 Chapter 7: Discussions on Carbon/Slag Interactions observe large bubble and small of gas bubbles dispersed throughout the entire slag volume. Metallic iron is observed at the slag/gas interface and the overall slag volume is dictated by the large number of small bubbles. With elapsing time, small gas bubbles are still present, although not as numerous as the initial stages of reaction. However, these small bubbles contribute to forming a large bubble as seen after 15 minutes of reaction. Metallographic examination of the slag/p2 blends assembly shows in the first minute of reaction small and large gas bubbles in the cross sections (Figure 7-18 (d)). Small shiny particle are dispersed throughout the slag area representing the reduced Fe. Metallic Fe is seen also in different locations such as the slag/gas and the carbon/slag interface. After 15 minutes of reaction, the optical images illustrate the reduced iron is deposited at the carbon/slag interface. Examination of slag/p3 blends assembly, indicated after 60 seconds of reaction, a large gaseous region trapped between the substrate and the slag droplet along with several much smaller gas bubbles (Figure 7-19 (e)). With increasing reaction time, gases still present in the droplet. With the reduction of iron oxide in slag nearing completion, the molten iron droplets tended to increase in size and precipitate on the substrate. It can concluded that 100% palm char showed better foaming where the presence of smaller bubbles all through the time period would sustain the foaming when compared to palm shell blends (Figure 7-11 (b.)). The iron reduction rate was expected to be lower for palm char due to a steady released of gases (Figure 7-24). 235

265 Chapter 7: Discussions on Carbon/Slag Interactions (a) Metallurgical coke (MC) (b) Gas Fe 500 µm µm µm t=60 sec t=120 sec Slag Fe Gas 500 µm 500 µm 500 µm t=600 sec t=900 sec (b) Palm char (PC) 500 µm t=60 sec 500 µm t=120 sec Gas Slag 500 µm t=600 sec 500 µm Fe t=900 sec Figure 7-16 Optical microscopy images of (a) Slag/100% metallurgical coke (MC) and (b) Slag/Palm char (PC) as a function of time 236

266 Chapter 7: Discussions on Carbon/Slag Interactions (a) Metallurgical coke (MC) (b) Gas Fe 500 µm µm µm t=60 sec t=120 sec Slag Fe Gas 500 µm 500 µm 500 µm t=600 sec t=900 sec (c) P1 blends 500 µm 500 µm t=60 sec t=120 sec Slag Fe Gas 500 µm t=600 sec 500 µm t=900 sec Figure 7-17 Optical microscopy images of (a) Slag/100% metallurgical coke (MC) and (c) Slag/P1 blends as a function of time 237

267 Chapter 7: Discussions on Carbon/Slag Interactions (a) Metallurgical coke (MC) (b) Gas Fe 500 µm µm µm t=60 sec t=120 sec Slag Fe Gas 500 µm 500 µm 500 µm t=600 sec t=900 sec (d) P2 blends 500 µm 500 µm t=60 sec t=120 sec Fe Gas 500 µm 500 µm Slag t=600 sec t=900 sec Figure 7-18 Optical microscopy images of (a) Slag/100% metallurgical coke (MC) and (d) Slag/P2 blends as a function of time 238

268 Chapter 7: Discussions on Carbon/Slag Interactions (a) Metallurgical coke (MC) (b) Gas Fe 500 µm µm µm t=60 sec t=120 sec Slag Fe Gas 500 µm 500 µm 500 µm t=600 sec t=900 sec (e) P3 blends 500 µm t=60 sec 500 µm t=120 sec Slag Gas Fe 500 µm 500 µm t=600 sec t=900 sec Figure 7-19 Optical microscopy images of (a) Slag/100% metallurgical coke (MC) and (e) Slag/P3 blends as a function of time 239

269 Chapter 7: Discussions on Carbon/Slag Interactions In Tables 7-1 and 7-2, the measured bubble diameters from the optical images for various samples at different times are reported. The quantitative estimation of these diameters was carried out using the magnification scale marked on optical microscopy images by Adobe Photoshop 7.0 software. The minimum bubble diameters observed for palm char were generally smaller than the corresponding values measured for metallurgical coke. Due to their better retention, these smaller bubbles are expected to lead to improved slag foaming. Maximum gas bubble diameters with P3 blends were generally higher than the corresponding values for metallurgical coke. Palm shell/coke blends did not show any well defined trends as a function of palm shells concentration. On the other hand, minimum bubble diameters observed for palm shell/coke blends were generally smaller than the corresponding values measured for metallurgical coke. These smaller bubbles in the blends could contribute to improved slag foaming observed for the blends, due to their better retention, which leads to larger slag volumes. Table 7-1 Minimum bubble diameter in slag droplets in contact with metallurgical coke, palm char and palm shell blends Minimum Bubble Diameter (µm) Sample 1 min 2 mins 10 mins 15 mins Met. Coke (MC) Palm char (PC) P1 blends P2 blends P3 blends

270 Chapter 7: Discussions on Carbon/Slag Interactions Table 7-2 Maximum bubble diameter in slag droplets in contact with metallurgical coke, palm char and palm shell blends Maximum Bubble Diameter (µm) Sample 1 min 2 mins 10 mins 15 mins Met. Coke (MC) Palm char (PC) P1 blends P2 blends P3 blends These results are consistent with previous research where the minimum size gas bubble diameters in rubber-coke blends/slag leads to better foaming (Rahman, M. M., 2010, Zaharia, M., 2010). Teasdale and Hayes (Teasdale, S. L. and Hayes, P. C., 2005) proposed a gas ferrying mechanism, wherein CO gas produced in the initial reaction between carbon and slag gets transported through the gas phase and reacts with FeO in slag to produce CO 2 and metallic iron. This CO 2 is ferried back to carbon, where gasification of CO 2 to CO takes place via the Boudouard reaction. Gas entrapment in slags as a result of carbon/slag interactions is a dynamic phenomenon depending on a number of physical and chemical factors. In addition, ash oxides present in the substrate can also participate in reduction reaction and associated gas generation. These could also account for excess CO and CO 2 generated (Rahman, M., Khanna, R., Sahajwalla, V. et al., 2009). A SEM/EDS with mapping technique, on the slag droplet after 15 min of reaction with carbonaceous material is shown in Figures 7-20 to Predominant components in this region were CaO, MgO, Al 2 O 3 and SiO 2 with levels of molten iron and sulphur being observed (Figure 7-20 (c)). EDS spectra for metallurgical coke/slag (Figure 7-20 (b)) showed local regions of molten iron superimposed on other slag oxides. Red regions was found in round shapes which represent the reduced metal oxides (FeO). The 241

271 Chapter 7: Discussions on Carbon/Slag Interactions level of sulphur was higher than palm char/slag (Figure 7-21 (b)). Sulphur plays an important role in suppressing slag foaming which tends to increase the size of CO bubbles and affect stability (Kapilashrami, A. and Görnerup, M., 2006, Morales, R. D. and Rodríguez-Hernández, H., 2003). It was in good agreement where the size of bubbles increases from MC (Table 7-1) due to the influence of high sulphur. (a) O Fe S 500 µm (b) (c) Ca MC after 15 min reactions 10 µm Mg Figure 7-20 (a) Optical microscopy of MC/slag, (b) SEM, mapping on the inner region of quenched MC/slag and (c) EDS spectra of quenched metallurgical coke/slag assembly at 1550 C after 15 min of contact 242

272 Chapter 7: Discussions on Carbon/Slag Interactions Figure 7-21 shows the SEM micrograph with mapping of palm char at 1550 ºC after reaction with slag. EDS analysis is conducted in three specific points to determine the variations of composition. The presence of oxygen was seen while the level of sulphur is not detected. Palm char showed a better slag foaming compared to coke and the influence could be due to lower sulphur and high content of phosphorus (P 2 O 5 ) and iron oxide (Fe 2 O 3 ) that present in the ash (Figure 7-21 (b) and Table 3-1). This analysis is consistent with the results from the size of bubbles where palm char shows smaller bubble sizes compared to coke (Table 7-1). Gaskell and Skupien et. al had summarized the influence of sulphur and phosphorus on surface tension, and P 2 O 5 lowered the surface tension thus allowing more stability of the gas bubble generations. (Skupien, D. and Gaskell, D., 2000). (a) O Fe S 500 µm (b) (c) Ca Palm char after 15 min reactions 30 µm Mg Figure 7-21 (a) Optical microscopy of Palm char/slag, (b) SEM, mapping on the inner region of quenched Palm char/slag and (c) EDS spectra of quenched Palm char/slag assembly at 1550 C after 15 min of contact 243

273 Chapter 7: Discussions on Carbon/Slag Interactions Further, the blends of palm shell and coke in contact with slag is presented in Figure The SEM/EDS with mapping technique shows P1 blends in contact with slag after 15 minutes of reaction. The micrograph indicated that iron is deposited as a result of carbon/slag interactions releasing gas bubbles which lead to fluctuations in the liquid slag droplet. It is assumed that the produced metallic iron has significant levels as seen in EDS spectra. The other selected images show the slag contains CaO, SiO 2, MgO and Al 2 O 3. With increasing palm shells content in the carbonaceous blend the gas generation increase which is important for foaming. This can be explained on the basis of increased volatile matter in the palm shells and rates of FeO reduction which leads to greater levels of gas release inside the slag thus will lead to increase the bubble size. The present studies show that an improvement in slag foaming is governed by gas generation, its entrapment and subsequent release. These phenomena are controlled by both properties of carbonaceous materials and slag properties. Carbonaceous materials have a significant influence on the kinetics of reactions that generate gases and also contribute to changes in slag compositions during/slag carbon interactions. 244

274 Chapter 7: Discussions on Carbon/Slag Interactions (a) O Fe S 500 µm (b) (c) Ca P1 after 15 min reactions 10 µm Mg Figure 7-22 (a) Optical microscopy of P1 blend/slag, (b) SEM, mapping on the inner region of quenched P1 blend/slag and (c) EDS spectra of quenched P1 blend/slag assembly at 1550 C after 15 min of contact 245

275 Chapter 7: Discussions on Carbon/Slag Interactions 7.2 FeO Reduction In the direct reduction, carbon is presumed to react directly with iron oxide (Fe 2 O 3 ), which is very rapid then converted to FeO, followed by further reduction of FeO to Fe, producing carbon monoxide with carbon dioxide. The reduction of FeO to Fe will takes place at a slower step (Rao, Y., 1971). The direct reduction may be visualized as beginning at the points of contact between iron oxide and carbon particles; and oxygen is removed from the solid in the form of CO and CO 2. The assumption can be made into 2 steps; where a slag containing FeO and Fe 2 O 3 can be reduced by carbon (Equation 2-11 and Equation 2-12). As it assumed that the reduction of Fe 2 O 3 to FeO occurs quite rapidly it has not been considered for calculation purposes. Based on previous studies (Fruehan, R., 1977) which indicated that in the first stage of reduction, Fe 2 O 3 to FeO, the reaction product was essentially all CO 2, the possibility of carbon gasification reaction in controlling the reduction rate is expected. The off-gas analysis resulting from iron oxide rich slag in contact with various carbonaceous materials in the present study allowed the quantification of the reduction reaction. Carbon based materials such as coke, coal chars and graphite are commonly used to remove oxygen from the metallic oxide component. The gas generated from above mentioned reaction will form bubbles at the carbon/slag interface leading to the formation of a gas film. The maximum amount of oxygen that can be removed from both the slag and the carbonaceous substrate is calculated in terms of number of moles of oxygen. The oxygen content in the evolved gas by the reduction reaction can be accounted for the reduction of FeO in the slag. According to the mass balance for oxygen, the following relationships can be derived: FeO + xc = Fe + ( 2x 1) CO + (1 x) CO 2 n = n +2. n Eq. 7-1 FeO CO CO d( nfeo) dt = A n FeO Eq

276 Chapter 7: Discussions on Carbon/Slag Interactions where: n and A are respectively the number of moles (mole -1 ), and A is the reaction area (cm 2 ). The data obtained from IR analyzer on the reduction reactions during carbon/slag interactions are presented in Figure The off-gas analyzer monitored the amount of CO and CO 2 gases evolved during the carbon/slag interactions, ppm values. The volume flow rates of CO and CO 2 gases were calculated considering that 1 liter/min of Argon was purging the furnace environment. Based on standard temperature and pressure (STP) conditions one mole of a standard gas occupies 22.4 liters of volume. From Equation 7-2 the time dependent reduction involves the removal of one mole of oxygen during the reduction of one mole of FeO. Thus, each of moles of CO detected corresponds to a mole of oxygen removed from the FeO present in the slag, while each mole of CO 2 corresponds to two moles of oxygen. A further conversion into equivalent number of moles of oxygen removed have been considered and plotted as a function of time (Figure 7-23). From Figure 7-23 (a), palm char samples showed a higher number of moles of oxygen were removed from the slag when compared to the amounts of moles of oxygen from the interaction with metallurgical coke however with a slower rate. When palm shells were added with coke (P1 blends), the oxygen content estimated from the evolved gases increased further when compared to coke/slag assembly (Figure 7-23 (b)). However, when palm shells accounted for higher proportions in the carbonaceous blends, as P2 and P3 blends; the amounts of oxygen removed were seen decreased. This might be attributed due to lower of CO 2 released from agricultural waste materials. P1 blends/slag led to the highest amount of removed oxygen accounting for the highest amount of oxygen removed when compared to all carbonaceous substrates. It was expected because after combustion in DTF, the residual char derived from the palm shell blends conserved unburned volatiles as Table 3-7 shows. Moreover, the role of solid coke was expected to gasify the CO 2 and generated CO, thus would release more gases compared to other blends where more palm shells rather than coke. Modifications in the carbonaceous structure of palm shell/coke blends appeared during the initial gas phase reactions (Yunos, N. F. M., Zaharia, M., Ahmad, K. R. et al., 247

277 Chapter 7: Discussions on Carbon/Slag Interactions 2011) which improved the generation of reducing gas and thus increased the quantity of removed oxygen. The rate of gas generation is enhanced when palm shell/coke blends are compared to coke alone. Another factor that could influence the total of moles removed could be the gases generated from breaking of lignocellulosic structure from palm shells. In Chapter 4 showed that the breaking of lignocellulosic structure at high temperature will release a considerable amount of CO and CO 2 gases that will participate in carbon/slag interactions. 248

278 Chapter 7: Discussions on Carbon/Slag Interactions Total removed O, moles (a) 100% MC 100% PC Total removed O, moles (b) 100% MC P1 blends Time, sec Time, sec Total removed O, moles (c) 100% MC P2 blends Total removed O, moles (d) 100% MC P3 blends Time, sec Time, sec Figure 7-23 Total removed oxygen, moles as a function of time and carbonaceous material used The change in FeO content in the slag was calculated using the amount of oxygen evolved of the blends and the initial FeO content in the slag (see Appendix B). The FeO concentration in the slag with respect with to time is estimated and further represented in Figure Three distinct stages are observed: I. An incubation period attributes to a delay due to the time required for the transport of the gas and the response from the IR analyzer. The remaining time period before the initiation of reaction is possibly due to the nucleation and growth of the generated gas bubbles. 249

279 Chapter 7: Discussions on Carbon/Slag Interactions II. A steady state period. III. A degradation period characterized by local deficiency of FeO at the slag/gas interface. Because alterations in experimental conditions change the reaction rates of region (I) and (III), the reaction-rate constant is estimated from the slope region (II), where the concentration of FeO changes linearly with time. FeO concentration, % Temp: 1550 ºC Stage I Stage II 100% MC 100% PC P1 blends P2 blends P3 blends Stage III Time, sec Figure 7-24 FeO concentration in the slag with proceeding reaction 250

280 Chapter 7: Discussions on Carbon/Slag Interactions An initial slow reduction section followed by a linear rapid reduction and finally a slow degradation region characterised by local deficiency of FeO. The pure palm char showed the largest duration of the initial slow region followed by the pure coke; the length of the initial slow region appeared to decrease with a decreased in the level of palm shells blended with the coke (Figure 7-24). The reduction of the FeO present in the slag took place in about 300 seconds, and FeO in the slag was reduced completely for coke sample. When P1 blends was the carbonacous substrate, the FeO concentration in the slag decreased faster being consumed in about 160 seconds. Increasing the palm shell in the blends as in P3 blends, a lower reduction of the available FeO in the slag was observed. However, all the carbonaceous blends showed faster FeO reduction when compared to coke alone. On the other hand, 100% palm char showed a lower rate of FeO reduction and taking a longer period of time to be completely reduced in about 360 seconds. From Figure 7-24 the time required for complete reduction of the FeO from the slag decreased with decreasing amount of palm shell blended in coke. The role of solid coke was expected to gasify the CO 2 and generated CO, thus would release more gases compared to other blends where more palm shells rather than coke. The relatively lower times of reduction recorded for the blends (50 to 60 seconds) compared to pure coke may be attributed to the presence of a hydrogen environment provided by the agricultural waste materials. Higher hydrogen contents promote fluidity at high temperatures and gas formation (Ono-Nakazato, H., Yonezawa, T. and Usui, T., 2003, Sohn, I. and Fruehan, R., 2005). Hydrogen gas, apart from being a faster reducing agent than both carbon monoxide and solid carbon also enhances the rate of reduction of iron oxides by carbon monoxide (Ono-Nakazato, H., Yonezawa, T. and Usui, T., 2003) when it is added to a reduction system containing the latter. Optical and SEM studies are cnsidered to support the calculations based on gas generation data and are presented in section

281 Chapter 7: Discussions on Carbon/Slag Interactions Estimation of the reaction rate constant The rate of reduction is equivalent to the rate of CO and CO 2 gases produced from the system. Iron oxide is the main oxide getting reduced as a result of the interaction of slag with various carbonaceous materials under investigations. Based on these gases, the variation of the percentage of FeO remianing in the system is plotted as a function of time. The slope of the graph is calculated to determine the reaction rate (R) by considering linear portion. A first order kinetic reduction is asummed and a similar aprroach has been reported in the literature (Bafghi, M., 1993, Mehta, A. S. and Sahajwalla, V., 2000, Sarma, B., Cramb, A. and Fruehan, R., 1996, Story, S., Sarma, B., Fruehan, R. et al., 1998). R ( S /100) * ρ / MW FeO = Eq. 7-3 FeO s FeO where: R FeO is the reaction rate of FeO (moles/cm 3 s), S FeO is the slope of the graph, % FeO/s, ρ s is the density of the slag (g/cm 3 ) and MW FeO is the molecular weight of FeO (g/mole). On the basis of the reaction rate, R, calculated using Eq. 7-4, the apparent reaction rate constant is calculated as follows: R = K ( A/ V ) Eq. 7-4 FeO 0 slag Combining Eq. 7-7 and 7-8 the reaction rate constant is given by Eq. 7-5: S ρ FeO slag K 0.. MWFeO 100 A V = Eq. 7-5 slag where: K is the reaction rate constant in moles/cm 2 s, A is the interfacial area of contact in cm 2 and V slag is the slag volume in cm 3. The reaction rate, expressed in mol.cm -2.s -1 is the slope of the points devided by the interfacial area. The rates are deduced from CO and CO 2 gas volumes assuming the oxygen is removed as CO and CO 2. From Figure 7-25, it was noted that previously in 252

282 Chapter 7: Discussions on Carbon/Slag Interactions section 7.1.1, that the CO gas generation (moles) decreases proportionally to palm shell content in the blends, recording the highest value when the blend contained the lowest proportion as in P1 blends. Using Eq. 7-4 to 7-5 and the amount of gas generated from reaction of slag/coke and slag/coke-palm shell blends, the rates of reduction were calculated and are presented in Figure 7-25 as a function of carbon materials used as a substrate. 7.0 Rate constant, moles/cm 2 s (x10-5 ) Met. coke Palm char P1 blends P2 blends P3 blends Figure 7-25 Reaction rate constant as a function of carbon material used The rate of reaction of FeO (Table 7-3) decreased with the proportion of palm shells blended with the coke, with the blends recording much higher values than for pure coke or pure palm char. For 100% metallurgical coke, a value of 2.16 x 10-5 moles/cm 2 s was calculated as the maximum rate, while a lower value of x 10-5 moles/cm 2 s was calculated for the 100% palm char. The corresponding rates for the blends were 6.23 x 10-5 moles/cm 2 s, 4.37 x 10-5 moles/cm 2 s and 3.10 x 10-5 moles/cm 2 s for blends P1, P2 and P3, respectively. 253

283 Chapter 7: Discussions on Carbon/Slag Interactions Table 7-3 Reaction rate constant (moles/cm 2 s) for 100% metallurgical coke, 100% palm char and palm shell/coke blends Substrate Reaction rate, K (moles/cm 2 s) Met. coke x 10-5 Palm char x 10-5 P1 blends x 10-5 P2 blends x 10-5 P3 blends x 10-5 Slower rates of gas generation from 100% palm char made it easier for slag to trap gases. A relatively slower rate of gas generation could have a lower impact on the modification of slag composition and would allow for gas bubbles to be trapped in the slag sample instead of rapidly escaping from the slag (Yunos, N. F. M., Zaharia, M., Ahmad, K. R. et al., 2011). These resulted in the better slag foaming in the case of the palm char compared to coke. The reason for such a behavior might be the oxygen and oxides in the ash of palm char which reacted as surface active. Oxygen present will decrease the surface tension in iron melt and the interfacial tension at the slag/metal boundary. Surface active elements in liquid have shown to retard the rate of interfacial reactions (Richardson, F. D., 1974). Oxides such as P 2 O 5 and SiO 2, which are surface active in slags, have shown to reduce the rate of reduction of iron oxide in slags. Palm char sample contains high P 2 O 5 and SiO 2 in the ash might be one of the factors to lower the rate of FeO reduction. The other reason of lower rate of reduction might be due to the low of fixed carbon content (55.6 wt. %) where more carbon is needed to react with slag in producing more gases. In a study performed by Fruehan (Fruehan, R., 1977) on the rate of reduction of iron oxide by different carbon based materials such as coconut charcoal, coal char and metallurgical coke in an inert atmosphere (T = 900 ºC and T = 1200 ºC), a strong dependence of the reduction rate on the type and the amount of carbon used was seen. 254

284 Chapter 7: Discussions on Carbon/Slag Interactions Thus, palm shell blends showed better reduction compared to coke and palm char due to the carbon in coke which could gasify CO 2 to CO gas during the interactions. Such findings support an overall rate of reduction controlled by the rate of oxidation of carbon. At high temperatures, oxidation of carbon and partially mass transfer of FeO was reported to control the overall reaction. 255

285 Chapter 7: Discussions on Carbon/Slag Interactions 7.3 Discussion on Slag Foaming of Carbonaceous Material Interfacial phenomena occurring between polymer/coke blends with steelmaking slags were investigated at 1550 ºC (Rahman, M., 2006, Zaharia, M., Sahajwalla, V., Kim, B.-C. et al., 2009). This section presents in-depth discussion on the effect of palm shell blends and char, on the slag/carbon interactions, along with the effect of rubber and polyethylene (PET), which are polymers containing high levels of volatiles. The substitution of waste polymers by blending with metallurgical coke was found to modify blends characteristics; significant differences have been observed in the carbon/slag interactions of polymer/coke blends in previous studies (Zaharia, M and Kongkarat, S. 2011). Metallurgical coke, rubber/coke blends (R3), polymers, namely Polyethylene Terephthalate (PET)/coke blends, palm shell/coke blends (P3) and 100% palm char have been chosen for comparison in the slag foaming behavior due to distinct in chemical structure. Rubber contains high carbon, sulfur and hydrogen while PET contains carbon, hydrogen and high oxygen. The discussions on volatiles produced from rubber and PET are presented to compare them with palm shells which contains volatiles and oxygen. Table 7-4 shows the chemical composition of rubber, PET, palm char and metallurgical coke. Table 7-4 Chemical composition of coke, palm char, rubber and PET Materials VM C H O N S Ash Met. coke Palm char P3 blend *Rubber, R **PET *(Zaharia, M., 2010) **(Kongkarat, S., 2011) 256

286 Chapter 7: Discussions on Carbon/Slag Interactions Influence of Gas Generation on Slag Foaming The rate of gas generation has been reported to have a significant effect on the slag foaming behavior and this influence was found to depend on the basic characteristics of carbonaceous materials. Rahman et. al. (Rahman, M. M., 2010, Sahajwalla, V., Rahman, M., Khanna, R. et al., 2009) investigated interactions between an EAF slag containing high levels of FeO with metallurgical coke and natural graphite at 1550 ºC using the sessile drop technique. These authors found that the gas generation of both CO and CO 2 from the slag/coke assembly was significantly higher than that of slag/graphite, also indicating an extensive/rapid FeO reduction by metallurgical coke. Even though the rate of gas generation for coke was very high, poor gas entrapment was observed within molten slag. Rapid gas generation can lead to the escape of gas bubbles from the slag sample. It was mentioned that high levels of gas generation resulted in a strong likelihood of convective transport of reactants and products across the slag/coke interface with oxides in coke ash partially dissolving in molten slag and modifying slag composition. The carbonaceous materials (coke, 100% palm char and P3 blends) under investigation showed significant differences in gas generation and entrapment and their foaming behavior (Figure 7-26). Metallurgical coke showed lowest level of gas generation as well as lowest slag foaming which was not sustained over extended period. High levels of off-gases (CO, CO 2 ) were emitted from the slag/metallurgical coke substrate; CO 2 levels were relatively much higher than 100% palm char. 100% palm char showed a steady gas released and continuously fluctuating for a longer time. On the other hand, P3 blends showed increased gas generation compared to coke and fluctuating behavior with droplets increasing in size and then releasing gas. A high volume ratio greater than 1 was maintained for an extended period. 257

287 Chapter 7: Discussions on Carbon/Slag Interactions (a) 100% MC 100% PC P3 blends Volume ratio, V t /V Time, s (CO+CO 2 ), ppm (b) 100% MC 100% PC P3 blends Time, s Figure 7-26 (a) Volume ratio of 100% Metallurgical coke (MC) and 100% Palm char (PC) and P3 blends with slag and (b) Gases (CO+CO 2 ), ppm, generated with respect to reaction time 258

288 Chapter 7: Discussions on Carbon/Slag Interactions Just from visual observation these carbonaceous materials (palm shells) influenced the foaming behavior. High temperature images captured during specific times of reaction showed an increase in volume when carbonaceous material consisted palm shells, while coke show similar slag droplet (Figure 7-27 (a) (c)). Volume ratio measurements and optical imaging are also considered to determine the gas entrapment and the size of bubbles present in the slag after interactions with coke, 100% palm char and P3 blends. (c) P3 blends (b) 100% Palm char (a) 100% MC t = 60 sec t = 120 sec t = 600 sec t = 900 sec Figure 7-27 High temperature images of slag droplet in contact with (a) metallurgical coke (MC), (b) 100% palm char (PC) and P3 blends at 1550 ºC as a function of time 259

289 Chapter 7: Discussions on Carbon/Slag Interactions Slag composition can be affected in two different ways: reduction reactions and possible transfer of ash oxides from coke into molten slag. The extent of FeO reduction as a function of time can in turn lead to the differences in physical properties of the molten slag, such as viscosity, surface tension and density, and this would affect slag foaming (Paramguru, RK 1997). In the present study, palm char showed a relatively slow rate of CO and CO 2 generation compared to the case of coke; and a greater level of gas entrapment was observed in the slag sample. A higher rate of gas generation in the case of coke may lead to local turbulence at the interface and transfer ash oxides into the molten slag and modify slag composition. The modification of slag composition may lead to the changes in the physical properties of slag, and thus impact the ability of slag to hold the gas (Rahman et al. 2009). Conversely, a relatively slower rate of gas generation in the case of palm char could cause the gas bubbles to be trapped in the slag sample instead of rapidly escaping from the slag. In the other case, palm shell blends showed higher rate of FeO reduction compared to coke (Table 7-3). While in the case of rubber (Zaharia, M., 2010) their interaction with slag showed a high rate of gas generation which resulted a higher FeO reduction rate compared to coke. These blends (rubber/palm shells blend) showed similar behaviour due to volatile matter content, where palm shell/coke has 25 wt. % and rubber/coke blend has 20.9 wt. % (Table 7-3). Optimal rates of gas generations are beneficial for both improved foaming and FeO reduction. The higher rates of gas generation for blends compared to coke are important for optimisation. For 100% palm char, the FeO reduction rate is about half that of coke which indicates that the blends lead to optimum outcomes not the 100% palm char. These showed palm char is better in slag foaming compared to coke. 260

290 Chapter 7: Discussions on Carbon/Slag Interactions Influence of Volatiles from the Agricultural Wastes/Polymers on Slag Foaming The foaming of slag is predominantly caused by the retention of CO and CO 2 generated from the reactions between slag and solid or solute carbon. The palm chars are expected to improve the carbon/slag interactions due to the availability of gases such as CO and CO 2 and other hydrocarbons above 900 ºC. When palm shells reacted with an iron oxide rich EAF slag, the presence of FeO leads to a reduction reaction depending on the reducing agents including C, CO and H 2 released at high temperature (Chapter 4). This result is attributed to the presence of hydrogen that promotes gas formation. H 2 peaks were also detected in the last stages of thermal decomposition (Figure 4-1 in Chapter 4). H 2 comes from the condensation of aromatic structures (lignin) or the decomposition of heterocyclic compounds (cellulose/ hemicellulose), processes that occur at high temperatures (Haiping, Y., Yan, R., Chen, H. et al., 2007, Hasegawa, I., Tabata, K., Okuma, O. et al., 2004, van Heek, K. H. and Hodek, W., 1994 ). Hydrogen gas, apart from being a faster reducing agent than both carbon monoxide and solid carbon, also enhances the rate of FeO reduction (Fruehan, R., 1977). It can be seen that from equations: FeO + H 2 (g) = Fe + H 2 O Eq. 7-1 FeO + CO (g) = Fe + CO 2 (g) Eq. 7-2 The reduction of FeO by H 2 and CO will produce CO 2 and H 2 O. Moreover, CO 2 produced by Eq. (7-11) will react with any available carbon at high temperatures, thus the key advantage palm shell comes from its increased hydrogen content, in comparison to metallurgical coke. Interactions between FeO containing slag with plastics/coke and rubber/coke blends at 1550 ºC using the sessile drop technique were respectively investigated by (Kongkarat, S., 2011, Rahman, M. M., 2010, Zaharia, M., 2010). Using a similar proportion of agricultural waste/coke blend used in the present study, (Rahman, M. M., 2010) reported that the maximum FeO reduction rate for metallurgical coke was 1.52 x

291 Chapter 7: Discussions on Carbon/Slag Interactions mole.cm -2.s -1, while the reaction rate was observed to increase when a HDPE/Coke blend was used with a maximum value of 1.91 x 10-5 mole.cm -2.s -1. Similarly, (Zaharia, M., 2010) reported that the maximum rate of FeO reduction for metallurgical coke was 1.89 x 10-5 mole.cm -2.s -1, while the reaction rate was also found to increase when a rubber tyre/coke blend was used with a maximum value of 2.5 x 10-5 mole.cm -2.s -1. Recently, the FeO reaction using PET/Coke was also conducted at 1550 ºC by (Kongkarat, S., 2011), where the reaction rate for PET/coke blend was 1.5 x 10-5 mole.cm -2.s -1 and coke was 1.74 x 10-5 mole.cm -2.s -1. The reaction rate for coke, palm shell/coke and 100% palm char in the present study were compared with other values reported in literature and is shown in Table 7-5. It was found that rubber showed a higher rate of FeO reduction than that of coke. However, the results observed in the present study indicate an opposite trend for 100% palm char, where the rate of FeO reduction was found to be lower than that for coke. The lower rate of palm char might be due to low fixed carbon; while in the case of rubber and palm shell blends, they showed similar high rate of FeO reduction with high fixed carbon content. Table 7-5 Comparison of the maximum rate of FeO reduction by different carbonaceous materials obtained from literature Materials Reaction Rate Researcher Coke 1.89 x 10-5 mole.cm -2.s -1 (Zaharia, M., 2010) Rubber/Coke 2.5 x 10-5 mole.cm -2.s -1 (Zaharia, M., 2010) Coke 2.2 x 10-5 mole.cm -2.s -1 Present study Palm char 0.94 x 10-5 mole.cm -2.s -1 Present study Palm shell/coke 3.1 x 10-5 mole.cm -2.s -1 Present study 262

292 Chapter 7: Discussions on Carbon/Slag Interactions The differences between the values of reaction rates in the case of metallurgical coke reported in the present study and from the previous studies (Kongkarat, S., 2011, Zaharia, M., 2010), are possibly due to the difference in the properties of slags and cokes used. In the case of polymer/coke blends and palm shell/coke blends, the polymer and lignocellulosic characteristics were found to have a significant effect on their interactions with slag. As shown in Table 7-4, e.g. rubber tyre which contains high levels of carbon and hydrogen where higher carbon and hydrogen is in the form of volatile matter (CH4 and H2), can aid and supplement gas generation and FeO reduction when reacted with molten slag. In the case of 100% palm char, the oxygen content that present could also oxidize Fe product to generate FeO. That is why 100% palm char shows an extended time period which the reaction appears to take place. The influence of oxygen which may reoxidize some of the reduced Fe in the slag, then reform FeO into the bulk slag (Kongkarat, S., 2011). These results show that FeO in the slag was not only reduced by solid carbon, but also by volatiles in the 100% palm char when compared to rubber/coke blends where rubber does not contain any oxygen. These additional reactions can contribute to the improvement in slag foaming behaviour in the case of palm shell/coke blends compared to coke. Ozawa et. al (Ozawa, M. and Kitagawa, S., 1986) investigated the reduction of FeO in molten slags by solid carbon in electric arc furnace steelmaking and pointed out that factors like quality of solid carbon injected, boundary area of reaction, wherein solid carbon reacts with FeO and nature of slag could influence the reduction rate of FeO in slag. They reported that, the volatile matter in solid carbon has a greater influence on the reducing reactions. The reduction of carbon materials with high volatile was controlled by the chemical reactions, while in the carbon with lower volatile matter; the reduction was controlled by the transport of FeO in slag. 263

293 Chapter 7: Discussions on Carbon/Slag Interactions Influence of Mineral Matter from the Agricultural Wastes on Slag Foaming Due to high levels of gas generation, there is therefore a strong likelihood of gas escaping quickly due to greater velocity. Due to extensive reduction of iron oxide, its levels in slag will be lowered. According to slag phase diagrams (Slag Atlas, 1995), a lower iron oxide content will increase the melting point of slag; which will in turn reduce its liquid fraction and enhance the viscosity of the slag. Surface active elements, such as oxygen and sulphur can affect the surface tension (γ) of the liquid metal (Jimbo, I. et al., 1993). A previous study on surface tension and interfacial tensions shows that the surface tension of liquid iron decreases with increasing oxygen and sulphur content. Oxygen is expected to come from ash of palm char and mineral matter (oxides) in the coke. The influence of oxygen which may reoxidize some of the reduced Fe in the slag, then reform FeO into the bulk slag. In the present study, the FeO level should decreased ever time, thus the rate will be lower as we see in the case for 100% palm char. Other interactions such as mineral matter (SiO 2 ) from metallurgical coke affect the surface tension and making it easier for gases to escape (Hara, S. and Ogino, K., 1986, Rahman, M. M., 2010). Silica was a mineral matter (57.4 wt. %) present in coke ash (17.2 wt. %) (Table 3-3 and Table 7-4), and its diffusion from the substrate into slag will lead to a lowering of surface tension of the slag. Alumina was another significant mineral matter (26.5 wt. %) present in coke ash. According to atomistic simulations of (Khanna, R. and Sahajwalla, V., 2005), poor wetting between alumina and molten iron results in a strong tendency between alumina and liquid iron to be mutually exclusive from their immediate neighborhood. A similar finding was observed from (Mehta, A. S. and Sahajwalla, V., 2000). In the present study, the interactions of palm char (5.6 % ash) and metallurgical coke (17.2 % ash) with EAF slag showed major differences in gas generation, carbon/slag interactions and rate of iron oxide reduction (Table 7-3). Metallurgical coke showed rapid iron oxide reduction, very high rates of gas generation but poor foaming behavior. Palm char on the other hand showed excellent slag volumes but slower reduction of iron oxide. Slower rates of gas generation made it easier for slag to trap gases. The reason 264

294 Chapter 7: Discussions on Carbon/Slag Interactions for such a behavior might be due to the components in the mineral matter (ash) of palm char which are surface active (P 2 O 5 and SiO 2 ). Surface active components in liquid have shown to retard the rate of interfacial reactions (Richardson, F. D., 1974). Components such as P 2 O 5 and SiO 2, which are surface active in slags, have shown to reduce the rate of reduction of iron oxide in slags. Palm char sample contains P 2 O 5 and SiO 2 in the ash which might be one of the factors for lower rate of FeO reduction (Table 3-3). The influence of P 2 O 5 and sulphur were studied and the previous authors observed that the addition of sulphur increased surface tension while more phosphorus oxide caused a decrease in surface tension, thus P 2 O 5 was considered to be surface active (Figure 2-11). Figure 7-28 shows the influence of S and P 2 O 5 addition on the surface tension of 30% FeO containing slags. Similar behavior has been observed in previous work Kozakevitch, Elliot, Bhattacharyya and Gaskell. (Bhattacharyya, P. and Gaskell, D., 1996, Elliott, J. F., 1988, Kozakevitch, P. and Olette, M., 1971) Influence of Carbon Structures from the Agricultural Wastes on Slag Foaming The other factor that influence a better foaming from agricultural waste is the interactions of lignocellulosic structure (cellulose, hemicellulose and lignin) in palm shells to break at high temperatures could also influence for low FeO reduction rate (Table 7-6). Hydrogen bonding, intermolecular and intramolecular, is recognized as one important linkage between cellulose and hemicellulose (Henriksson, Å. and Gatenholm, P., 2001, Schmidt, M., Gierlinger, N., Schade, U. et al., 2006) while hydrogen bonding in lignin was found to exist chemically linked to polysaccharides (Miyoshi, K., Uezu, K., Sakurai, K. et al., 2006). It was found that the hydrogen bonding in lignin is more stable at high temperatures (Zhang, X., Yang, W. an d Blasiak, W., 2011) which is attributed to the high degree of branching and formation of highly condensed aromatic. Thus, the gas generated from palm shell char was released at slower rate due to the complex structure of lignin. 265

295 Chapter 7: Discussions on Carbon/Slag Interactions Table 7-6 Type of crystallization of polymers and molecular bonds in the carbonaceous materials (Antal, M. J., Jr. and Varhegyi, G., 1995, Orfão, J. J. M., Antunes, F. J. A. and Figueiredo, J. L., 1999, Sharma, R. K., Wooten, J. B., Baliga, V. L. et al., 2004) Samples Crystallization Structure Bond Met. coke amorphous (turbostatic) C=C Lignin amorphous (-C 2 H 2 -), π bonds aromatic Cellulose crystalline [(C 6 H 10 O 5 ) n ] Hemicellulose amorphous [(C 5 H 8 O 4 ) n ] Rubber (SBR) amorphous C-C within aromatic ring, C-S Rubber is characterized by a three dimensional framework following the vulcanizing process when sulfur was introduced and C-S and S-S bonds were formed (Kameda, T. and Asakura, T., 2003). When rubber content was introduced in the carbonaceous blend, the gas generation increases compared to coke which is important for foaming. This can be explained on the basis of increased volatile matter in the rubber and higher rates of FeO reduction which leads to greater levels of gas release inside the slag. This can be understood on the basis of influence of sulfur on the surface tension of slags. Gaskel and Skupiel (Skupien, D. and Gaskell, D., 2000) measured the surface tensions in CaO-FeO-SiO 2 system with a view to provide a better understanding of the phenomenon of slag foaming and considered the influence of S on steelmaking slags. Each slag contained 30 wt. pct FeO with the balance varying from CaO/SiO 2 (wt. pct) = 0.43 to CaO/SiO 2 (wt. pct) = 1.5. Sulfur was introduced as CaS and the surface tension of melts in the range of 0 to 3 wt pct were measured at 1400 ºC. Increasing the surface tension (effects of sulfur) leads to an increase in the interfacial energy and thus will lead to increasing the bubble size (Table 7-7). Therefore, as sulfur influences size of bubbles, eventually an increase in minimum bubble size decreases the 266

296 Chapter 7: Discussions on Carbon/Slag Interactions stability of foam. This suggested that when there is greater proportion of rubber in the system the stability of foam will decrease. Palm shell/coke blends sample with low sulfur content, was expected to improve the foam stability compared to rubber/coke blend samples. Figure 7-28 showed the comparison of high temperature images captured during specific times of reaction showed an increased in volume of palm shell/coke blends (P3) and rubber/coke blends (R3). (a) P3 blend R3 blend (b) t = 60 sec t = 120 sec t = 240 sec Figure 7-28 High temperature images of slag droplet in contact with (a) rubber/coke blend (R3) (Zaharia, M., 2010) and (b) palm shell/coke blend (P3) at 1550 ºC as a function of time A series of cyclic reaction (Donskoi, E. and McElwain, D., 2003, Shi, J. Y., Donskoi, E., McElwain, D. L. S. et al., 2008) is set up as the carbon monoxide and hydrogen (products of the carbon gasification) in turn partially react with the iron oxide and further reactions continue. These cyclic reactions persist until all the iron oxide has been reduced to metallic iron, provided there is enough carbon in the system (Shi, J. Y., Donskoi, E., McElwain, D. L. S. et al., 2008) as is the case in the present study. It is 267

297 Chapter 7: Discussions on Carbon/Slag Interactions thus clear that the presence of hydrogen and some fixed carbon in a carbonaceous material are both essential to initiate and maintain these cyclic reactions to obtain a high rate of reduction. This clearly explains why the blends generally perform better than both pure coke and the pure agricultural waste. The rates of reduction were presented to show the blend is better than the parent materials (palm char and coke) in Figure 7-29 as a function of carbon materials used as a substrate; where 100% metallurgical coke, a value of 2.16 x 10-5 moles/cm 2 s, 100% palm char showed value of x 10-5 moles/cm 2 s and P3 blend showed 3.10 x 10-5 moles/cm 2 s. Rate constant, moles/cm 2 s (x10-5 ) % Met. coke 100% Palm char P3 blends Carbonaceous material Figure 7-29 Reaction rate constant as a function of carbon material used 268

298 Chapter 7: Discussions on Carbon/Slag Interactions Influence of Entrapped Gas Bubbles in the Bulk Slag on Slag Foaming The foaming of slag is a result of the gas bubbles being retrained in the bulk liquid slag, and is influenced by amount and size of the entrapped gas bubbles. ((Zhang, Y. and Fruehan, R., 1995)) studied the effects of gas bubble size generated by argon gas injected into liquid slag and by the slag/metal interfacial reactions between the solute carbon in liquid metal and FeO in the slag on slag foaming. The authors concluded that foams with very fine bubbles have spherical bubble cells and are very stable, while foams with larger bubbles are less stable. (Rahman, M. M., 2010) measured the diameters of entrapped gas bubbles in the quenched droplets after reaction with coke and HDPE/Coke blends. This author reported that the minimum gas bubble diameter of the quenched slag droplets after 2, 4 and 8 minutes of reaction with coke was 90, 92 and 92 µm, respectively, while it was 29, 36, 39 µm in the case of HDPE/Coke (Figure 7-31 (a)). (Kongkarat, S., 2011) reported PET/Coke showed the smallest gas bubbles (36-62 µm) (Figure 7-30 (c)). The small gas bubbles are generated from the reduction reactions of FeO and these bubbles could contribute to both sustaining and improving the slag foaming behavior where a similar trend had been reported by (Rahman, M. M., 2010) showing that the gas bubble sizes for polymer/coke blends are generally smaller than that for coke alone, which is expected to lead to a better slag foaming behavior. A close inspection of the cross section of the slag/carbonaceous sample revealed the small gas bubbles in Figure

299 Chapter 7: Discussions on Carbon/Slag Interactions 500 µm 500 µm 500 µm 500 µm (e) 100% Palm char (c) PET/coke (PET 3) (b) rubber/ coke (R3) (d) Palm shell/ coke (P3) (a) HDPE/coke (HDPE 3) 500 µm 500 µm 500 µm 500 µm t = 60 sec t = 120 sec t = 600 sec t = 900 sec Figure 7-30 Optical microscopy images of (a) Slag/HDPE blend (HDPE 3), (b) Slag/rubber blend (R3), (c) Slag/PET blend (PET 3) (Kongkarat, S., 2011, Rahman, M. M., 2010, Zaharia, M., 2010), (d) Slag/palm shell (P3) blend as a function of time and (e) Slag/100% palm char (PC) In the present study, the bubble diameters of palm char (38-56 µm) were smaller than coke and rubber/coke (37-85 µm)(zaharia, M., 2010) in Figure 7-30 (b) and (d). The 270