Investigation of Sintering Kinetics of Magnetite Pellets ... - DiVA portal

20 downloads 0 Views 13MB Size Report
Dec 24, 2014 - particles of magnetite from taconite ore2. Also, the raw material for the current study is magnetite ore from the LKAB mines in Sweden; therefore ...
L ICE N T IAT E T H E S I S

ISSN 1402-1757 ISBN 978-91-7583-405-4 (print) ISBN 978-91-7583-406-1 (pdf)

Investigation of Sintering Kinetics of Magnetite Pellets During Induration

Luleå University of Technology 2015

Kamesh Sandeep

Department of Civil, Environmental and Natural Resources Engineering Division of Minerals and Metallurgical Engineering

Investigation of Sintering Kinetics of Magnetite Pellets During Induration

T. Kamesh Sandeep Kumar

Investigation of Sintering Kinetics of Magnetite Pellets During Induration

Kamesh Sandeep

Luleå University of Technology Department of Civil, Environmental and Natural Resources Engineering Division of Minerals and Metallurgical Engineering

Printed by Luleå University of Technology, Graphic Production 2015 ISSN 1402-1757 ISBN 978-91-7583-405-4(print) ISBN 978-91-7583-406-1 (pdf) Luleå 2015 www.ltu.se

Abstract One of the measures of development and economy of a nation is its per capita consumption of steel, and the demand is fulfilled by iron ore. In the context of increasing environmental constraints and ores becoming leaner and leaner, recycling and utilization of ore fines becomes necessary. Pelletization, being one of the major agglomeration techniques is increasingly practiced across the world to produce agglomerates that can be fed into the metallurgical furnaces (blast furnaces) for subsequent processing. In Europe, Sweden has the richest iron ore deposits, and mining and metals production contributes majorly to its net export. Luossavaara – Kiirunavaara AB (LKAB) operates with magnetite ore-bodies in the northern Sweden to produce iron ore (magnetite) pellets (26 MTPA), and exports about 70 % of its product to European steel producers. The quality of the pellets produced is of utmost importance to meet the demands of the stakeholders. Therefore, constant efforts are necessary to maintain and improve the quality of magnetite pellets, and it is necessary to enhance the understanding underlying physico-chemical mechanisms during pellet production. Magnetite pellets prepared from the fines are indurated to attain the quality standards in terms of strength and other metallurgical properties. The quality of the pellet is primarily determined by the physico-chemical changes the pellet undergoes as it makes an excursion through the gaseous and thermal environment in the induration furnace. Among these physico-chemical processes, the oxidation of magnetite phase and the sintering of oxidized magnetite (hematite) and magnetite (non-oxidized) phases are vital. Rates of these processes not only depend on the thermal and gaseous environment to which the pellets are exposed in the induration reactor, these processes are also interdependent. Therefore, a doctorate project is undertaken to systematically understand these processes in isolation to the extent possible and quantify the process kinetics seeking the physics. Further, with the help of modeling methodologies overall induration process can be simulated using integration of the individual quantified process kinetics. With this motivation, the current study is focused on investigating the sintering phenomena involved during induration of magnetite pellets. Experiments with single pellets were designed to understand and quantify the sintering behavior of oxidized magnetite (hematite) and magnetite independently. The kinetics of sintering can be described using power law (‫ ݐܭ‬௡ ) and Arrhenius (݈݊൫ܶ‫( ܭ‬ଵ/௡) ൯ = ݈݊ ‫ܭ‬Ԣ െ

ொ ோ்

) equations. In the

experiments, single pellets were exposed to different thermal profiles in a controlled atmosphere, and their in-situ shrinkage was captured continuously by a novel technique using Optical Dilatometer. It was found that the sintering behavior captured by shrinkage of the pellet can be quantified using three isothermal kinetic parameters; namely – activation energy (ܳ), preexponential factor (‫ ܭ‬ᇱ ) and time exponent (݊). The values of activation energy and time exponent derived suggest that sintering of oxidized magnetite (hematite) is dominated by a single diffusion mechanism, whereas sintering of magnetite showed two distinct mechanisms; one operating at lower temperatures and the other at higher temperatures. The isothermal sintering kinetic equation is also extended to predict the nonisothermal sintering for both oxidized magnetite and magnetite, and validated with the laboratory experiments. In future, this will further be useful in predicting the sintering state of pellets during induration in plant scale operations.

2

Acknowledgements I am very thankful to my supervisor’s Prof. Nurni N. Viswanathan (LTU, IITB) for providing me the opportunity, continuous guidance and channelizing on technical front as well as many other areas of life; Dr. Charlotte Andersson (LKAB) and Dr. Hesham Ahmed (LTU) for many crucial discussions as and when required and efforts in managing things so that I can work smoothly, and Prof. Bo Björkman (LTU) for valuable visionary inputs throughout the course of project. I extend my sincere thanks to Prof. Åke Sandström (LTU), who accepted to take up the responsibility of supervisor in a short notice, and proactively involved in the progress of project and thesis writing. Financial support from Hjalmar Lundbohm Research Centre (HLRC) is gratefully acknowledged. I would also like to thank LKAB for providing the raw materials for the project, and allowed to utilize their laboratory services. I am thankful to quite a few personals from LKAB for their technical support and valuable feedback; Ola Eriksson, Daniel Marjavaara, Gustaf Magnusson, Anders Dahlin, Axel Stahlstrom, Klauss Weigel and Kjell-Ove Mickelsson. I also thank Prof. N. B. Ballal, Prof. M. P. Gururajan (IITB) and Prof. S. Seetharaman (KTH) for valuable discussions in critical stages of the project. I am thankful to the colleagues and friends from Process Metallurgy, MiMeR and the Department of Civil, Environmental and Natural Resources Engineering (SBN) at LTU. Finally, I am grateful to my parents who have supported and prayed for me all the way through even many miles away, little sister Sandhya, and my extremely special and lovely Aparna Lohiya for being with me forever through all the aspects of life, emotionally supportive, sometimes technically as well; and truly a source of inspiration and motivation. Thank You, T. Kamesh Sandeep Kumar Luleå, 2015

3

List of papers International Journals (peer reviewed) Paper I Estimation of Sintering Kinetics of Oxidized Magnetite Pellet Using Optical Dilatometer -

T. K. Sandeep Kumar, N. N. Viswanathan, H. Ahmed, C. Andersson and B. Björkman

(Published in Metallurgical and Materials Transaction B, Volume 46B, Number 2, 2015, pp. 635-643) Paper II Estimation of Sintering Kinetics of Magnetite Pellet using Optical Dilatometer -

T. K. Sandeep Kumar, N. N. Viswanathan, H. Ahmed, C. Andersson and B. Björkman (Submitted to Metallurgical and Materials Transaction B, 2015)

International Conferences contributions, not included in the thesis: Studying the Sintering Behavior of Oxidized Magnetite Pellet during Induration - T. K. Sandeep Kumar, N. N. Viswanathan, H. Ahmed, C. Andersson and B. Björkman (7th International Congress on the Science and Technology of Ironmaking (ICSTI): AISTech, May, 2015, Cleveland, United States of America) Effect of Heating Rates on the Sintering of Oxidized Magnetite Pellets during Induration - T. K. Sandeep Kumar, H. Ahmed, N. N. Viswanathan, C. Andersson and G. Magnusson (2nd European Steel Technology and Application Days (ESTAD): METEC, June, 2015, Dusseldorf, Germany)

Author’s contribution to the appended papers: Most of the planning, experimental work, evaluation and writing were done by the author. Other co-authors have contributed in a supervisory capacity, timely discussions and reading-reviewing the manuscripts.

4

Table of Contents Acknowledgements .................................................................................................3 List of papers ..........................................................................................................4 1.

Introduction .....................................................................................................7

2.

Background .................................................................................................... 12 2.1

Agglomeration Techniques ............................................................................................ 13

2.1.1 Briquetting .................................................................................................................... 13 2.1.2 Travelling Grate Sintering ..............................................................................................14 2.1.3 Pelletization ................................................................................................................... 15 2.2

Physico – Chemical phenomena during Magnetite pellets Induration ............................. 18

2.3

Sintering Mechanisms .................................................................................................... 21

2.3.1 Evaporation – Condensation (E-C) ................................................................................ 24 2.3.2 Surface diffusion (SD) .................................................................................................... 24 2.3.3 Grain Boundary Diffusion (GB) ..................................................................................... 25 2.3.4 Volume Diffusion (VD) ................................................................................................. 25 2.3.5 Plastic Flow (PF) ............................................................................................................ 26 2.4

Sintering Kinetics........................................................................................................... 26

3.

Scope ............................................................................................................. 30

4.

Approach and Methodology ............................................................................. 31

5.

Experimental Methods and Analysis .................................................................. 34 5.1

Raw Materials ............................................................................................................... 34

5.2

Green Pellet production ................................................................................................. 35

5.3

Oxidation ...................................................................................................................... 36

5.4

Sintering ........................................................................................................................ 37

5.5

Characterization............................................................................................................. 40

5.5.1 Density and Porosity measurements ............................................................................... 40 5.5.2 X – Ray Diffraction (XRD) .......................................................................................... 43 5.5.3 Microscopic studies ........................................................................................................ 44 6.

Results and Discussion..................................................................................... 45 6.1

Preliminary Microstructural Evaluation .......................................................................... 45 5

6.2

Shrinkage during Sintering ............................................................................................. 49

6.3

Degree of Sintering ........................................................................................................ 50

6.4

Sintering Rate ............................................................................................................... 53

6.5

Sintering Ratio .............................................................................................................. 56

6.6

Estimation of Sintering Kinetic Parameters ..................................................................... 58

6.7

Sintering Prediction using Kinetic Parameters ................................................................ 65

7.

Concluding Remarks ....................................................................................... 68

8.

Conclusions .................................................................................................... 71

9.

Future Plan..................................................................................................... 72

10. Bibliography ................................................................................................... 73

6

1.

Introduction

In the current era of globalization, manufacturing industries play a vital role in the growth and development of a nation. Iron and Steel industries are particularly crucial, as steel finds its usage in virtually every application in the manufacturing, transportation and construction to surgical to automobiles, and therefore, per capita consumption of steel is considered one of the critical measures of development. Steel is the most sustainable and resource efficient material with maximum recyclability without loss of quality1. There is a continuous drive towards growth and progress, and hence, an increasing demand for steel, and in-turn, hot metal (iron), which is met by the availability and supply of iron ores. Iron ores are mostly available in earth’s crust as hematite and magnetite. From the early 1900s, the iron ores of metallurgical grade were reduced in smelting furnaces to molten iron and subsequently processed in oxidizing furnaces to steel. Over time, the iron ore grades are getting leaner and leaner due to continuous excavation and with the increasing environmental awareness of particulate emissions to the atmosphere necessitated the need for recycling and reutilization of the iron ores. The fines, mostly generated during mining of the ores, dust emissions from the furnaces and the steel plant solid wastes, must be beneficiated and agglomerated, so that they can be reutilized in the metallurgical furnaces for subsequent processing. Agglomeration is the size enlargement process whereby fine solid particles adhere to each other, and pelletization is one of the widely practiced agglomeration techniques2. Unlike other agglomeration techniques, pelletization can utilize solid particles at ultra-fine scales; and going to finer scales enables liberation, which in turns helps in better beneficiation. Pelletization uses much finer sized particles ( 65 % is considered to be of metallurgical grade for efficient reduction in iron making furnaces. During mining of the ores, large amount of fines are generated which cannot be used directly in the furnaces are stockpiled. With the continuous excavation over the years, ores grades are getting leaner and leaner, both in terms of the availability of lumps and lower Fe content. The Fe content in these ores can be enhanced through mineral beneficiation processes at fine scales. The ores beneficiated at fine scales calls for agglomeration techniques for their usage in iron making processes. Further, particulate emissions from the furnaces containing iron and carbon constituents are affecting the environment and contribute significantly to the greenhouse gases. This raises concerns over the environment, and government bodies have laid stringent constraints on the dust emissions from plant as well as on carbon footprints50. Therefore, to make the process more sustainable and environment friendly, there emerges the need to recycle and re-utilize these fines. Agglomeration techniques play a crucial role in these efforts. Agglomeration is the size enlargement process whereby fine particles adhere to each other, and heat hardened to form agglomerates which can be fed into the furnace as iron bearing materials. Agglomerates intended for use in metallurgical furnaces should possess optimum quality parameters, as per the standards designed by the International Organization for Standardization (ISO) such as chemical composition (basicity), crushing strength (ISO 4700), tumbler and abrasive indices (ISO 3271), free swelling index (ISO 4698) reducibility (ISO 4695), reduction degradation under load (ISO 7992). There are three agglomeration techniques in practice; namely, briquetting, sintering (travelling grate) and pelletizing. Images of their products are shown in Figure 2.

12

(a)

(b)

(c) Figure 2: Iron ore agglomerates – a) Briquettes, b) Sinters, and c) Pellets

2.1

Agglomeration Techniques

2.1.1 Briquetting In briquetting, the fines in the size range of 0.15 mm to 3 mm are pressed against each other with binder or without binders at certain pressures into blocks of suitable shape and size, and hardened2,51,52. The appropriate proportions of iron ore fines, additives such as flux and coke breeze/ coal dust are mixed together, and sprinkled with moisture and binder to form a homogenous green mix. The green mix is then passed into the briquetting roll press, where optimum compressive force is applied to form green briquettes or composites. The rolls of the 13

press contain the impressions of the desired shape and size of the finished product. It is a matrixdistributed compaction process; therefore, the raw materials need to be thoroughly mixed to have uniform distribution of binder and moisture. The moisture required for briquetting is relatively high, 10 to 15 % by weight of the raw material, approximately. Depending on the compressive strength requirements of the destined furnace, the green briquettes are subjected to hot briquetting or cold briquetting process. In hot briquetting, the composites are heat hardened in the shaft furnace to attain high strength (220 – 250 kg/briquette), whereas in cold briquetting53, the composites are either sun-dried in sun or in muffle furnace at 378 K (105oC), for 24 hours to achieve a strength of approximately 50 – 80 kg/briquette54. A wide range of binders55-57, both organic and inorganic, such as bentonite, cement, molasses, tar, starch, lignosulfonate, limestone, etc. have been used over the years. The choice and amount of the binder depends on their characteristics considering the subsequent process. Briquetting requires less capital investment and was used to produce briquettes for blast furnaces started during World War II (1940s), but slowly decreased and ceased by about 1960 because of the increased demand for agglomerates to meet the needs of operating larger blast furnaces efficiently. However, in recent times there has been a revived interest in briquetting with the development of alternate iron making processes and other metallurgical operations. This is because of the efforts to utilize the fines, especially coal dust and other steel plant wastes, producing briquettes which possess relatively lower strength and quality for use in blast furnace but sufficient enough for use in alternate iron making furnaces, LD furnaces or in Electric Arc Furnaces (EAF)54,58-60.

2.1.2 Travelling Grate Sintering Travelling Grate Sintering is also known as the Dwight Lloyd Sintering process, named after the scientists who discovered it. In this process the green mix of raw materials including coke breeze is laid as a packed bed onto the travelling strand with permeable grate. This is passed through an ignition hood, subjected to heating at high temperatures (1523 K (1250oC) – 1553 K (1280oC)) while moving continuously under the downdraft suction to produce the sinter cake2,52,61-64. The coke breeze provides a combustion front which moves downwards from the top of the sinter bed. The sinter cake is then crushed, and the desired size fraction is dispatched for further processing in blast furnace, while the smaller size is reused as return fines. The raw materials for sintering – iron 14

ore fines (< 6.3 mm), fluxes, coke breeze, return fines and other metallurgical wastes (< 3.15 mm) having proper granulometry are mixed together with the optimum moisture to form the green mix. Furthermore, sintering can recycle various steel plant solid wastes such as sludge, flue dust from blast furnace and basic oxygen furnace, mill scales, etc. Because of its flexibility in the usage of raw materials fines, the travelling grate sintering process has been used extensively throughout the world by iron and steel manufacturers since the from early 20th century. Sinter is the major iron-bearing burden which replaced lumps to about 60 – 85 % in blast furnaces. The limitation of travelling grate sintering process is that ultrafine particles (< 0.15 mm) cannot be used in the green mix, as this would affect the permeability of the bed, resulting in inferior quality sinters.

2.1.3 Pelletization In the early 1970s2,52,63-65, the limitation of travelling grate sintering technique and increasing environmental awareness necessitated the use of pelletization broadly. As opposed to the other agglomeration processes, which use particles in millimeter size, pelletization uses fines in the size range of microns. Pelletization is an agglomeration process whereby fine particles (< 150 microns or 0.15 mm) adhere to each other with the help of moisture and binding media, balled into spherical compacts and indurated to produce pellets with desired physical, chemical and morphological properties. Pelletization was invented for the purpose of utilizing the ultrafine particles of magnetite from taconite ore2. Also, the raw material for the current study is magnetite ore from the LKAB mines in Sweden; therefore, the focus of this thesis will henceforth be on the pelletization of magnetite ore fines. The pelletization process for magnetite ores can be broadly categorized into three sub-processes – raw material preparation, balling and induration, a schematic of process flow is shown in Figure 3. Firstly, the magnetite ore is beneficiated and ground in ball mills by wet grinding circuit. The output slurry of concentrate from ball mills having desired particle size distribution is sent to vacuum filters for the removal of excess moisture, producing the filter cake. In terms of size distribution, for good quality pelletization, 65 % of material should be below 45 microns (Mesh #350)15. The additives – fluxes and binder are added to the filter cake, and mixed in a counter rotating mixer while water is sprinkled to attain a homogenous green mix. The green mix containing 8 – 10 % moisture is rolled either in balling drums or discs to produce green pellets. After screening, the green pellets of desired size range (9 – 15

16 mm in diameter) with adequate properties are chosen for further processing, whereas under and over-sizes are recycled to the mixture. The green pellets are not strong enough to be used directly in the iron making furnace; therefore, they need to be strengthened by heat hardening process in an induration furnace. During induration, the packed bed of green pellets undergoes drying, preheating, firing and cooling. These are achieved by allowing hot gases to flow upward and downward in straight, circular or rotary kilns containing pellets for efficient heat transfer. This imparts strength and metallurgical properties to the pellets. The indurated magnetite pellets are collected and dispatched for further processing in iron making furnaces or shipped over long distances to various iron and steel manufacturers throughout the world.

Figure 3: Schematic of typical process flow sheet for magnetite pelletization The green pellets are exposed to a typical thermal profile in the induration furnace, as shown in Figure 4, and the hot gases (air or oxygen enriched-air) are allowed to flow upward as well as downward through the packed bed of pellets for efficient heat transfer2,52,65. The bed of green pellets is passed through the updraft drying zone (UDD) mostly to evaporate moisture, followed by downdraft drying (DDD) to gently release the chemically bonded hydroxides, raising the 16

temperature of the gas to approximately 573 K (300oC). After drying, the pellet bed reaches the preheating zone (PHZ) under downdraft suction, and increases the temperature of the gas from 573 to 1553/1573 K (300 to 1280/1300oC), approximately, where a major part of the oxidation takes place followed by initiation of sintering. The supposedly oxidized pellets after preheating are exposed to firing zone (FZ), operating isothermally at 1553/1573 K (1280 / 1300oC), where particles inside the pellets gets sintered (incipiently fused) in solid state to each other by forming strong bonds15. Firing of pellets can be done either in the straight grate, where the hot gases are sucked through the bottom of the bed or in the rotary kiln, where the pellets are exposed to a counter current flame of hot gas in the rotating drum. This is followed by final oxidation of the sintered pellets in the after firing zone (AFZ) at lower temperatures (~ 1273 K (1000oC)) and annular cooling zones (CZ1 and CZ2) to room temperature. Thereafter, the magnetite indurated pellets are collected, stored and transported to nearby iron and steel making plants or shipped over the long distances to different countries.

Figure 4: Thermal profile in different zones of pellet Induration Furnace

17

2.2

Physico – Chemical phenomena during Magnetite pellets Induration

The major phenomena during induration of magnetite pellet comprise drying, oxidation and sintering. Drying is a crucial phenomenon in the beginning of induration, and needs to be done carefully, as it affects the treatment in successive zones of the furnace, accordingly. The poorly dried pellets can form internal cracks because of the thermal shocks, crumble into fines and thereby affecting the permeability of bed, which could be detrimental to the quality of pellets2,65. There have been numerous studies66-71 to understand and model the drying phenomenon so as to design the optimum thermal profile for efficiently drying the pellets. The other two phenomena – oxidation and sintering are primarily responsible for imparting strength to the pellets during induration6,7; hence, they require thorough investigation. The oxidation proceeds radially inward from the surface to the center of the pellet following shrinking-core model, as seen in Figure 1. On the other hand, sintering occur both in the oxidized magnetite shell as well as in the magnetite core. The degree of sintering depends also on the thermal profile inside the pellet.

Wynnyckyj and McCurdy7 found that the extent of oxidation and sintering reactions during induration of magnetite pellets is responsible for the evolution of pellet morphology (microstructures) accompanied by the structural changes, which in turn determines the quality of pellet. Therefore, it is necessary to understand the mechanism of these phenomena and the effects of various raw material and processing parameters on pellet quality and performance. Papanastassiou and Bitsianes6 made an effort to understand the kinetics of the oxidation mechanisms of magnetite pellets, whereas, Wynnyckyj and McCurdy7 focus on identifying the reason responsible for the large shrinkage gradient in the pellet during induration leading to the formation of duplex structure (layered inhomogeneity), which is detrimental to the pellet quality. Forsmo10,15 recently studied the oxidation and sintering mechanisms of the magnetite pellet, separately by performing Thermo-Gravimetric (mass change) and Dilatometric (shrinkage) experiments, respectively. They found that the oxidation is a stepwise phenomenon where firstly, magnetite grains oxidizes to metastable Ȗ – hematite (maghemite) at temperatures below 773 K (500oC), and then further oxidize to Į – hematite with an increase in temperature beyond 773 K (500oC). The magnetite particles oxidize at a rapid rate initially, but as the particles are covered by the layer of hematite, the oxidation rate decreases. This is because of the limited diffusion of 18

oxygen through the hematite layer in to the particles. Oxidation gives rise to strong bonding of the grains within the pellet at the points of grain contact. The heat energy released during the highly exothermic oxidation increases the temperature inside the pellet significantly, which can be used further for the sintering reactions, as shown in Equation (1). 4 Fe3O4 + O2 = 6 Fe2O3;

ǻH = – 490 kJ/mol72

(1)

The increase in temperature beyond 1373 K (1100oC) facilitates sintering in the pellet, causing the phase transformations in order to attain the necessary metallurgical strength. The irregular shaped pores in the pellet merge during sintering and rounded off, resulting in the increase in pore size with decrease in total porosity, this is associated with the volumetric shrinkage of the pellet. Ideally, it is desirable to have complete oxidation of the magnetite pellet before sintering begins, so as to attain a homogenous structure (phase distribution) inside the pellet, but quite often this may not be the situation in industrial operations. The limited diffusion of oxygen into the pellet due to the hematite layer may cause sintering to begin before it is completely oxidized. Therefore, a nonoxidized magnetite core might sinter simultaneously with the hematite shell, which results in gradient shrinkage in the pellet. The gap in the gradient of shrinkage increases when the temperature inside pellet is above 1473 K (1200oC), at which point hematite begins to shrink at a rapid rate along with that of magnetite7. This results in the formation of a so-called duplex structure with a heavily sintered magnetite core shrinking more than the less sintered outer shell of hematite. The fraction of the duplex structure depends on the extent of oxidation and sintering of the magnetite pellet. This causes structural stress in the pellet, which results in concentric cracks, and can significantly affect the pellet quality7,15. On the other hand, additives and raw material properties also have a significant effect on the evolution of the pellet quality, and are explored quite extensively. Forsmo15 studied the influence of green pellet properties such as fineness of magnetite ore on the induration mechanism during pelletization. It was found that the finer the magnetite concentrate the faster is the rate of sintering for both the hematite and magnetite phases, leading to the formation of larger fraction of unwanted duplex structure. Based on the microstructural observations, the optimum fineness for good quality magnetite pelletization was 65 % of particles should pass through 45 μm sieve. Cho13 extended the study of different particle size on magnetite concentrate with the focus on oxidation 19

kinetics for the purpose of modeling, and found similar observations. Tang et al.14 explored the use of oxygen-enriched air during oxidation of magnetite pellets and found that it enhances the degree of oxidation at lower temperatures before sintering begins, and hence, attempts to lower the fraction of duplex structure. In addition to this, dolomite – fluxed magnetite pellets had been micro-structurally investigated to understand the process of oxidation, sintering, and subsequent reduction8,73,74. They found that the presence of bentonite and low-melting dolomite flux adjacent to the magnetite particles promotes sintering by the formation of bridges preferentially along the grain boundaries, and imparts strength to the pellet. Firth and Garden75 later on investigated the calcination of limestone and dolomite in magnetite pellets and their interaction with oxidation mechanisms during induration. Based on the findings, Firth76 attempted to model the calcination and oxidation reactions separately, and collaborated to predict the extent of these mechanisms and their effect on the microstructural evolution of the pellets. Granse77 studied the swelling behavior of slag forming constituents in magnetite pellets during reduction, and found that vitreous slag formed is useful to avoid the abnormal swelling. Further in the current laboratory, Semberg12 studied the effect of additives used for magnetite pelletization on the subsequent mechanism of reducibility in the blast furnace. The mechanisms of bond formation by sintering during reduction of magnetite fluxed pellets were reflected in their microstructures, and in turn, their quality. Various flux additives such as calcite, quartzite and olivine were explored in the study, and found by microstructural investigation that magnesium from the olivine flux can diffuse farthest into the magnetite matrix at high temperatures in reducing atmospheres forming magnesioferrite, which imparts efficient high temperature strength and improves reducibility in the blast furnace. It is always beneficial to have forehand understanding of pellet properties and performance in relation to the various raw materials and operating parameters. Several researchers17,18,31,78,79 have done bulk scale (kg) experiments in the laboratories, pilot plant, etc., with different raw materials (Fe content, fineness, density, porosity, etc.), furnace atmosphere (‫݌‬ைమ ) and operating parameters (temperature, time, heating rate, etc.) to establish trends with the indurated pellet quality parameters (FeO content, crushing and tumbling strength, reducibility, reduction degradation strength, etc.). The quality parameters of the pellets obtained by the experimentations are empirically correlated to develop reactor scale models for the whole induration process, to predict the pellet qualities and the furnace performance. However, they have often been limited in their ability to predict the quality of pellets and its distribution. Interestingly, most of the aforementioned studies are primarily qualitative in nature, and an indirect way of prediction using 20

empirical models. A quantitative approach characterizing the induration phenomena in a single pellet can help in understanding the spread in pellet quality at the reactor scale, optimizing operating parameters for new raw material mix as well as optimizing raw materials mix for pellet quality. In order to quantify the induration process in a single pellet, a detailed laboratory investigation for the individual process mechanisms and the corresponding kinetic parameters for quantification are necessary. Further, these kinetic parameters for different processes can be used to predict overall induration process through integration of these processes through mathematical modeling methodology. As mentioned earlier sintering of magnetite and oxidized magnetite, oxidation of magnetite and heat transfer through a pellet are important processes during induration of a single pellet. The present thesis focuses on sintering process of magnetite and oxidized magnetite phases. Powder metallurgists have extensively studied sintering mechanisms for sintering metallic and ceramic powder green compacts. A brief summary of these mechanisms is depicted in the following section as it forms the basis for the sintering of magnetite and oxidized magnetite phases.

2.3

Sintering Mechanisms

Sintering is a thermally activated phenomenon, where particles adhere to each other by diffusional mass transport of atoms, which is studied for decades32,33,80-97 and comprehensively discussed in the section. The driving force for the sintering is the decrease in excess free surface energy of the irregularly shaped particles. The temperature of the sintering depends on the material and particle size. Mostly, the materials exhibit sintering temperatures equivalent to 0.5 – 0.8 times their melting temperatures. There are three broadly types of sintering, depending on the principal adhering phenomena – solid-state sintering, liquid-state sintering and pressure-assisted sintering. Although, liquid and pressure-assisted sintering is also prevalent, the focus of this study is on solidstate sintering whereby solid particles of magnetite in the pellet sinter to produce strengthened agglomerates. Solid-state sintering has been studied for decades to understand the progress of sintering and the mechanisms responsible. Important stages during solid-state sintering can be explained by the two-

21

sphere model, as shown in Figure 584,85,88,98,99. Firstly, the two spherical particles of same size (D) come in contact with each other at random orientations, and adhere at the contacts because of the weak van der Waals forces. With the increase in temperature, the initial stage sintering proceeds by the growth of neck rapidly at the particle interface, without any significant densification or shrinkage. As the neck grows, the compact become stronger. In the intermediate stage, as the temperature approaches the sintering temperature, the neck size increases with the evolution of grain boundaries, giving larger average particle size with fewer particles. Due to this, the interparticle distance decreases resulting in shrinkage, and the pore structure becomes smooth and spherical. Finally, the particles merge at the last stage of sintering, forming agglomerated particle 1.26 times the initial individual particle size (1.26 D), when they are exposed at high temperature for a very long (infinite) time. Although, solid-state sintering in the pelletization of mineral ore fines (magnetite ore fines) extends up to the intermediate stage, where, the objective is only to agglomerate the particles.

Figure 5: Schematic of the different stages of sintering as it progresses

The mechanisms during solid state sintering of oxides can be determined from their path of mass transport depending on the driving force. They can be broadly classified in two categories – surface transport and bulk transport, as shown in Figure 632,95. The mass transport is mainly associated with the diffusion of atoms across different particle sites. The mode of dominant mechanism may vary depending on the temperature, particle size and extent of sintering. When the particles come close to each other, surface transport mechanisms such as evaporation condensation and surface diffusion, produce neck formation and grow without densification (shrinkage) due to the mass flow originating and terminating at the particle surface. In bulk transport, mechanisms such as volume diffusion, grain boundary diffusion, plastic flow and viscous 22

flow, the neck growth promotes densification (shrinkage) because the movement of mass is from the particle interior and deposited at neck. Surface transport dominates at low temperatures, whereas bulk transport is most active at higher temperatures. At still higher temperatures, the elementary stage of liquid sintering may sometimes become dominant and fastens rate of sintering. This could be due to the formation of low melting temperature slag, especially along the grain boundaries. The sinter bond between the contacting particles is the critical region because atoms are deposited there to reduce the surface energy. Generally, all the key sintering measures are related to the mass transport rates and subsequent neck growth and pore changes.

E-C

SD

VD

PF

GB B VD

Figure 6: Mass transport diffusion mechanisms during sintering

23

The aforementioned diffusion mechanisms, which are responsible for the mass transport during solid-state sintering, are briefly discussed. 2.3.1 Evaporation – Condensation (E-C) Evaporation – Condensation mechanism is associated with transport of vapor across the particle surface leading to repositioning of the atoms32,84,100. Evaporation occurs at the surface and transport across the pore space and condenses on a nearby surface. This results in the reduction of the total surface area, as bonds grow between particles in contact without any change in the inter particle distance. This is often ignored, as it is very slow at typical sintering temperatures. 2.3.2 Surface diffusion (SD) Surface diffusion involves the movement of atoms between the defect sites at surface such as ledges, kinks, vacancies and adatoms, when thermally activated without any shrinkage or densification, as depicted schematically in terrace-ledge-kink model in Figure 7101,102. A typical surface diffusion event involves three steps; firstly, the atom is detached by breaking the existing bonds, typically at a kink, moves randomly across the surface, and finally reattaches at an available surface site, possibly again at a kink. The activation energy required for the motion of these atoms is usually lower than the other diffusion mechanisms and is influenced by the sintering temperature. The rate of surface diffusion gradually slows down with the progress of sintering, as the available surface area is lost to sintering bonds32-34,80,94,103.

Figure 7: Terrace model for surface diffusion showing motion of atoms from defect sites such as kinks, adatom, and ledge. 24

2.3.3 Grain Boundary Diffusion (GB) In grain boundary diffusion the mass flows along the grain boundaries which serve as vacancy sinks and redistributes at sinter bonds between the particles, resulting in the shrinkage and densification32,85,90,94,98. As sintering progresses, new grain boundaries emerge at the sinter bonds with the decrease in free surface energy, and therefore, also decrease the sintering rate. The activation energy required for the net flow mass in grain boundary diffusion is usually in between the surface diffusion and volume diffusion32. The grain boundary becomes inefficient to serve as vacancy sink at later stages of sintering, and the solid structure becomes so strong that it resists further densification, and then the mass transport takes place between pores via the grain boundary, leading to pore coarsening. Surface diffusion proceeds together with the grain boundary to give cooperative mass flow, where densification is controlled by transport along the grain boundary and mass redistribution, while surface smoothing occur by surface diffusion33,83,87. 2.3.4 Volume Diffusion (VD) Volume diffusion, also known as bulk or lattice diffusion, involves the motion of vacancies through a crystalline structure either by interstitial or substitutional mechanisms. Volume diffusion can happen by any of the three diffusion paths, as depicted by two-particle geometry in Figure 832-34,88,90,93,104. In the first path, the vacancies can move from a surface source (neck) to a surface site (particle) through the particle interior without any densification, and is termed as adhesion. The second path, termed volume diffusion densification, involves vacancy flow to the interparticle grain boundary from the neck surface, associated with shrinkage and densification. This is because, effectively, a layer of atoms moves in the direction opposite to the contact between particles, reducing the inter-particle distance as the sinter bond grows. The final path is where the vacancies can be annihilated by dislocations via a process termed dislocation climb. It involves cooperative action by both dislocations and vacancies and results in densification, since the vacancy source is not at a surface.

25

Figure 8: Volume Diffusion paths 2.3.5 Plastic Flow (PF) Plastic flow is the motion of dislocations under stress99. The mass began to flow plastically without diffusion by dislocation climb due to vacancy absorption and dislocation glide due to surface stresses, exceeding the flow at sintering temperatures32,100. 2.4

Sintering Kinetics

Sintering kinetics infers about the rate and the extent of reaction, for the given raw material and processing parameters, on the basis of minimization of surface energy. The information on kinetics can help in designing the optimum thermal treatment for the materials intended to be sintered. Therefore, it is beneficial to explore the possible ways to determine the kinetics of sintering. Numerous efforts33,34,36,43,44,46,80,95,104-106 have been made from as early as the 1940s to estimate the kinetics of the sintering phenomenon and understand the governing mechanism. Kuczynski34 developed an early approach to estimate kinetics while studying the sintering of alumina compacts by quantifying their microstructural observations; intermittently at predefined temperature intervals following the whole thermal profile. The extent of sintering was estimated from the microstructures of the alumina compacts at different temperatures with the help of a term called neck size ratio. It is the ratio of neck size (ܺ) to diameter (‫ )ܦ‬of the particle (Figure 5), which indicates the progress of sintering and is also related to the green and sintered fractional density of the pellet, according to the Equation (2), 26

௑ ஽

ீ௥௘௘௡ ி௥௔௖௧௜௢௡௔௟ ஽௘௡௦௜௧௬

ଵ/ଷ ଵ/ଶ

= 4 ቈ1 െ ቀௌ௜௡௧௘௥௘ௗ ி௥௔௖௧௜௢௡௔௟ ஽௘௡௦௜௧௬ቁ



(2)

German32 developed the sinter model for very fine iron powder (20 μm) on the basis of neck size ratio as a measure for the extent of sintering, which helps to determine the diffusion mechanism dominant at any sintering temperature (homologous temperature = sintering temperature/melting temperature), as shown in Figure 9.

It can be inferred that at low degree of

sintering (ܺൗ‫~ ܦ‬0.01), particles attach together by adhesion at lower sintering temperatures, and as the temperature increases, neck size increases with the surface diffusion being the dominant mechanism, whereas at even higher temperatures, the dominant mechanism for mass transport is grain boundary diffusion agglomerating two particles by forming strong bonds, and finally, a fully dense (ܺൗ‫~ ܦ‬1) particle. The other transport mechanisms may still contribute to neck growth, even if they are not dominant. The effect of simultaneous transport mechanism is to increase the overall rate of neck growth, but possibly with a decreased rate of shrinkage.

Figure 9: Sintering Diagram representing different sintering mechanism as a function of neck-size ratio and the temperature for fine (20 μm) iron powder 27

There were few sintering diagrams developed by the similar approach107,108. But the method of measuring the neck size and diameter of a particle at microscopic level was done manually under an optical microscope for several grains. It was, therefore a time consuming method and also prone to errors due to manual intervention. Thereafter, there have been various efforts by powder metallurgists to capture the sintering mechanisms. Surface area reduction, change in stress and energy, neck size ratio, coordination number or the contact points of grains were the approaches based on the microstructural quantification34,82,104,109-112. Since sintering is associated with the shrinkage of the material composite, researchers have explored this as a basis to overcome the limitations of microstructural approaches and developed the methods to determine sintering kinetics by capturing the macro-structural changes. Although there have been exhaustive studies to understand and estimate the sintering kinetics in the field of pharmaceutical, refractory, ceramics, glass, and even nuclear materials, there are very few on the iron ores. In 1974, Wynnyckyj and Fahidy33 made an early attempt to study the sintering kinetics of iron ores. They studied commercial magnetite and synthetic hematite using the approach of shadow photograph. Interestingly, there has not subsequently been much of work on the iron ores. Dilatometers have been used to measure the dimensional changes of the compact because of shrinkage during the sintering process. Traditionally, a push–rod dilatometer based on the principle of Linear Variable Differential Transducer (LVDT) was used10,36,38-42,44,45,47-49,105,113-116, where an alumina rod rests on the composite sample with a small amount of load (~ 5 to 20 grams) acting on it to sensitively capture the linear change in dimensions during isothermal heating, and then adapted various approaches to derive kinetics. It assumes that the variation in dimensions are isotropic i.e., change equally in all directions, during the sintering process, and hence estimate the change in relative volume and density of the composites. The ratio of relative change in these parameters is related to the initial reaction mechanism by power law and Arrhenius equation, to estimate the kinetic parameters – reaction rate and activation energy, as discussed in later sections. Although this method is used extensively, there are a few limitations, which were reported by Karamanov et al.117. They studied the sintering kinetics of alumina composite by the recently developed optical dilatometer, and found that the load exerting at the contact of push-rod with 28

the sample might interfere with sintering kinetics. The change in dimensions of composite is anisotropic in reality, and assuming isotropic changes lowers the efficiency of the analysis. Also, push-rod dilatometer is primarily designed for cylindrical composite, where the width to height ratio is 2:3, so that the base of the alumina rod can rest properly on the surface of the cylindrical sample, and may not be appropriate for the spherical samples, where only a single contact point exists at any random orientation. Therefore, an optical dilatometer, which is based on the principle of light optics, has been developed to capture the shrinkage of the pellet during sintering without any contact, and hence does not interfere in the process. This might be advantageous in the near future and it may be used extensively for estimating the sintering kinetics for compacts of various materials.

29

3.

Scope

The objective of the entire project is to model the induration behavior of magnetite pellets by studying the sintering, oxidation and heat transfer phenomena, in isolation and their mutual interference. This can be done by quantifying these phenomena to deduce their kinetic parameters, and later on, integrating them together for simulating the overall induration of magnetite pellets. In the scope of work, the sintering phenomenon is being studied independently for both oxidized and non-oxidized magnetite pellets. Sintering of single pellet is captured by means of optical dilatometer under isothermal as well as non-isothermal profiles. Optical dilatometer does not have the provision for inserting the sample directly in the isothermal sintering temperature. The isothermal sintering kinetics have been estimated, and extended to the non-isothermal segment to be in close proximation to industrial conditions. The overall sintering profiles for the pellets exposed to different heating profiles have been mathematically predicted and validated.

30

4.

Approach and Methodology

As mentioned earlier, the current study is focused on investigating the sintering phenomenon during induration of the magnetite pellets in isolation, as shown in Figure 1. The sintering of magnetite is a combination of sintering of two phases – oxidized magnetite (hematite) and nonoxidized magnetite (magnetite). To understand the overall sintering behavior of magnetite pellets, the two phases have further been investigated independently for quantification, and henceforth estimation of kinetics. Wynnyckyj and Fahidy33 quantified the iron ore sintering process, and they suggested that the volumetric shrinkage (Equation (3)) of the composite during sintering can be used to quantify the process. ܸ‫݁݃ܽ݇݊݅ݎ݄ܵ ܿ݅ݎݐ݁݉ݑ݈݋‬, ߚ௏ =

௏ି௏బ

(3)

௏బ

Where, ܸ௢ is the initial volume of composite, and ܸ is the volume of composite during induration They derived the term sintering ratio (ߛ), which is defined as the ratio of fraction of pellet sintered (ܺ) to the fraction yet to be sintered(1 െ ܺ), as mentioned in Equation (4).

ܵ݅݊‫݋݅ݐܴܽ ݃݊݅ݎ݁ݐ‬, ߛ =

௙௥௔௖௧௜௢௡ ௢௙ ௣௘௟௟௘௧ ௦௜௡௧௘௥௘ௗ ௙௥௔௖௧௜௢௡ ௢௙ ௣௘௟௟௘௧ ௬௘௧ ௧௢ ௕௘ ௦௜௡௧௘௥௘ௗ

=

௑ ଵି௑

(4)

Isothermal sintering for the two spherical particles is modeled as sintering ratio, (ߛ) following power law relation with reaction time, as mentioned in Equation (5), ߛ = ‫ ݐܭ‬௡

(5)

Where, ‫ ݐ‬is the isothermal time, ݊ is time exponent and ‫ ܭ‬is rate constant. Please note that that at a constant temperature the rate of sintering progressively decreases with time, as the driving force available (surface area of grains) diminishes with the increasing extent of sintering. Hence, the variation of sintering ratio with time is not expected to be a linear relation; a 31

simple power law is assumed which was further corroborated with experimental observations. As sintering is a thermally activated process, the rate constant, ‫ ܭ‬is expected to follow Arrhenius law is expressed as Equation (6) ln൫ܶ‫( ܭ‬ଵ/௡) ൯ = ln ‫ܭ‬Ԣ െ



(6)

ோ்

Here, ܳ is the activation energy at sintering temperature ܶ, ‫ܭ‬Ԣ is the pre-exponential factor, and ܴ is gas constant. In the current investigation an optical dilatometer has been used for the first time to capture the macroscopic shrinkage of the spherical pellet. This is because it does not have any contact load /force on the pellet sample, and also it captures the area change (two dimensional) during sintering, which adds an extra dimension to otherwise linear change (one dimensional). This helps in quantifying the sintering process more efficiently to estimate the kinetic parameters. Thereafter, sintering phenomena characterized through shrinkage measurement can be used to check the validity of the proposed sintering kinetics, and further, these kinetic parameters can be estimated to quantify the sintering phenomena. Such quantification can be used to predict the induration behavior of pellets, as it makes an excursion through thermal and gaseous environment in the reactor. Experiments are designed on the single pellet in the range of the temperatures for isothermal sintering, independently, for oxidized and non-oxidized magnetite pellets. The thermal profile used for the study is a combination of non-isothermal heating followed by the isothermal holding and cooling section. This has been done in order to be more similar to the real situations, and hence, the pellet is not introduced directly into the isothermal zone of the furnace. Therefore, proper measures have been taken to account for using the combined thermal profile while deriving sintering kinetics. The isothermal holding time is chosen so as to have adequate residence time for sintering to occur. The plan for the experiment is mentioned in Table I. The experiments are executed separately with a constant heating rate of 30 K/min to different peak sintering temperatures at intervals of 50 K for both the materials over the entire range of temperatures. The exception being at the lower sintering temperatures for the magnetite pellets, where experiments are also conducted at intervals of 25 K in the effort to capture the small dimensional changes. 32

Further, for the purpose of validating the model developed for prediction of the sintering behavior of pellets, experiments are also designed at three different heating rates at the same isothermal sintering temperature.

Table I: Experimental design to study sintering kinetics of oxidized and non-oxidized magnetite pellets Oxidized Magnetite Pellets Pellet

Sintering

Heating

Temperature

Rate

K (oC)

K/min o

( C/min)

Non-oxidized Magnetite Pellets

Isothermal Hold Time

1423 (1150)

30

40

HP2

1473 (1200)

30

20

HP3

1523 (1250)

30

20

HP5

1573 (1300)

1623 (1350)

30

30

20

20

Heating

Temperature

Rate

K (oC)

min

HP1

HP4

Pellet

Sintering

K/min o

( C/min)

Isothermal Hold Time min

MP1

1173 (900)

30

90

MP2

1223 (950)

30

60

MP3

1248 (975)

30

60

MP4

1273 (1000)

30

60

MP5

1298 (1025)

30

60

MP6

1323 (1050)

30

60

MP7

1373 (1100)

30

40

MP8

1423 (1150)

30

40

MP9

1473 (1200)

30

20

MP10 1523 (1250)

30

20

MP11 1573 (1300)

30

20

MP12 1623 (1350)

30

20

HP6

1573 (1300)

15

20

MP13 1573 (1300)

15

20

HP7

1573 (1300)

45

20

MP14 1573 (1300)

45

20

33

5.

Experimental Methods and Analysis

5.1

Raw Materials

The raw materials used in the current study are magnetite fines concentrate and bentonite. Magnetite ores from the mine of LKAB in Malmberget (Sweden) are ground, beneficiated and wet screened to produce concentrate to be used for pelletizing. In order to focus on investigating the behavior of magnetite pellets, additives such as flux and other solid wastes have not been considered. The only additive used is bentonite which acts as a binder for the fine particles in the pellet. About 15 kg of concentrate has been collected at once, stored in containers and covered with paraffin sheets to avoid any absorption of extra moisture. This is to ensure that the raw material chemistry and properties remain constant for the experiments during the entire course of project. The concentrate (MPC) is analyzed for its properties, such as chemical composition, moisture, fineness, true density and specific surface area, as mentioned in Table II. The representative concentrate sample for analysis is taken by means of coning and quartering. Moisture content is evaluated using an Infrared Moisture Analyzer MA150 (Sartorius AG, Germany), as per ISO 3087. The known amount (100 gm.) of concentrate is heated at 378 K (105oC) for 8 hours, and the weight loss due to evaporation is measured to determine the moisture content of the concentrate. The sample is heated in an oven furnace at 378 K (105oC) for 8 hours used for further analysis and to ensure that it is free from moisture. The concentrate is analyzed for particle size distribution or fineness and specific surface area using a Laser Diffraction Particle Size Analyzer and BET (Brunauer–Emmett–Teller (BET) Surface Area Analyzer, respectively; whereas, true density is measured using a helium gas pycnometer (Accupyc II 1340, Micromeritics, USA). Further, the chemical analysis of concentrate and bentonite is done with X-ray fluorescence (XRF). The dried concentrate is crushed in a pulverize mill to below 37 μm (Mesh No. 400). The concentrate is bombarded with high energy X-rays for elemental identification by the emission of characteristic "secondary" (or fluorescent) X-ray peaks, whereas area under the peaks determines the respective quantification of the elements. The elemental composition obtained is then converted to the respective stable oxides and normalized. Further, FeO analysis of the magnetite concentrate is done by titration. The analysis obtained shows that magnetite concentrate contains 34

Fe3O4 > 95 % with minor fractions of Al2O3 and SiO2 as gangue making up to 0.6 %, and 7.2 % moisture by weight. Table II: Properties of Raw Materials Chemical Composition

Physical Properties of Magnetite Conc.

Magnetite Conc.

Bentonite

(wt. %)

(wt. %)

Fe

71.06

SiO2

0.39

CaO

0.12

True Density (gm./cm3)

5.13

3.63

Moisture (wt. %)

7.20

51.64

Fineness (% of -45 μm)

6.43

MgO

0.26

0.68

Ȉ Na2O + K2O

0.078

3.30

5.2

65 2

3

Specific Surface Area (cm /cm ) 2

Blaine Number (cm /gm.)

9900 1930

Green Pellet production

Green pellet production or balling is the method of making moist spherical pellets from the raw materials mixed proportionately, known as green pellets2,63. Green pellets are prepared in the balling drums in two stages – nuclei or (seed) formation and their growth. The balling drum is 0.8 m in diameter, with the thickness of 0.15 m. The binder bentonite (0.7 wt. %) is added to concentrate, and mixed in a laboratory mixer (Eirich R02, Germany) to form the green mix. Green mix (batch of 7 kg) is gradually scattered in the balling drum, rotating at 37 revolutions per minute (rpm) to the prepare seeds (3.5 – 5 mm diameter). Moisture (0.5%) is added by spraying in short intervals over the scattered green mix, until the optimum moisture requirement is achieved. The nuclei are fed into the balling drum, rotating at 47 rpm, and the green mix is added in parts over the nuclei to produce green pellets. The green pellets, thus produced are screened to achieve 75-80 % in the size range of 9 – 10 mm diameter. The green pellets are then dried at 378 K (105oC) for 8 hours, and stored in a desiccator to avoid moisture absorption.

35

5.3

Oxidation

To study the sintering kinetics of oxidized magnetite, the dried magnetite pellet is fully oxidized to hematite prior to sintering. Oxidation of magnetite to hematite is a highly exothermic reaction and associated with the weight gain owing to the diffusion of oxygen atoms into the magnetite lattice. The aim is to oxidize the magnetite pellet to hematite to the maximum possible extent to make sure that the oxidation phenomenon does not interfere with the sintering phenomena of oxidized magnetite. The degree of oxidation (݂) is defined as the extent of oxidation the pellet has achieved. It is evaluated by the ratio of percentage of weight gain for magnetite pellet exposed to thermal profile during experiment to the maximum possible (theoretical) weight gain ((Equation (8)). It ranges from 0 to 1. The theoretical weight gain during oxidation of Fe3O4 to Fe2O3 is calculated to be 3.36 %.

‫= )݂( ݊݋݅ݐܽ݀݅ݔܱ ݂݋ ݁݁ݎ݃݁ܦ‬

ௐ௘௜௚௛௧ ௚௔௜௡ (%) ௗ௨௥௜௡௚ ௧௛௘ ௘௫௣௘௥௜௠௘௡௧ ்௛௘௢௥௜௧௜௖௔௟ ௪௘௜௚௛௧ ௚௔௜௡ (%)

(8)

Since the magnetite concentrate (MPC) used here has minor fractions of gangue and volatile minerals, their contribution to weight loss during oxidation is obtained by performing heating experiments in TGA from 298 K (25oC) to 1473 K (1200oC) under inert nitrogen atmosphere. The average weight loss for three experimental runs is found to be 0.22 %, which is taken into consideration while calculating the degree of oxidation. Thereafter, some preliminary oxidation experiments have been done at different temperatures in air, held isothermally for variable duration to design the optimum thermal profile for nigh magnetite pellet oxidation. The dried magnetite pellets are placed in a Carbolite Chamber Furnace and exposed to temperatures between 923 K (650oC) to 1073 K (800oC) at an interval of 50 K, and held at that temperature for variable duration from 2 to 24 hours. It is found that the degree of oxidation at 923 K (650oC) is approximately 75 %, which is not enough to continue with the sintering studies, whereas at 1073 K (800oC), although the degree of oxidation is above 90 %, there might have been sintering initiated at a very early stage. The magnetite pellets exposed to 1023 K (750oC) and held for 24 36

hours achieve a degree of oxidation above 90 % within 4 hours with no signs of sintering, and thereafter increase at a very slow rate without much significant oxidation, even at the end of 24 hours. Therefore, the optimum thermal profile designed for oxidizing the magnetite pellets is heating from room temperature to 1023 K ( 750oC), and holding isothermally for 4 hours, so as to achieve degree of oxidation 90% or above, which is in satisfaction with the sintering studies to follow15. The oxidized magnetite pellets are furnace cooled back to room temperature, and transferred for the sintering studies.

5.4

Sintering

Sintering is the principle phenomena during induration of magnetite pellets and is responsible for physico-chemical changes in the pellet. Quantification of the sintering phenomena for both oxidized magnetite (hematite) and non-oxidized magnetite pellets, in isolation, is the prime focus of this study. Since sintering causes volumetric shrinkage of the composite on macro-structural scale; it is used as a measure to quantify sintering for estimating their kinetics. A recently developed optical dilatometer by Misura HSM-ODHT is used, which works on the principle of light optics and continuously captures the in-situ shadow images of the sample exposed to the thermal profile, as depicted in Figure 10. It measures the change in area (shrinkage) of shadow images (two dimensional) with respect to temperature and time during the process; hence, anisotropy is taken into consideration by adding an extra dimension as opposed to linear changes by push-rod dilatometer. Since it does not involve any physical contact with the sample; it is suitable for spherical pellet composites without interfering with the sintering process. Therefore, the optical dilatometer has been used for estimating the sintering kinetics in the current investigation.

37

Figure 10: Schematic of Optical Dilatometer

It comprises three principal units mounted on an optical bench: a continuously illuminated halogen light source, a horizontal tube furnace (100 mm in length and 20 mm in diameter) and an image capturing microscope with recording facility. It is equipped with a double-beam optical measuring system having two lenses attached to the image capturing device – one compatible for small samples (2 to 5 mm diameter), and the other for bigger samples (6 to 12 mm diameter). Therefore, the size of the pellet chosen is 9 – 10 mm, rather than industrially preferred 10 – 12.5 mm diameter because of the instrumental limitations in the optical dilatometer, which cannot capture the images within the frame size beyond 12 mm diameter in size. The pellet sample is placed on a small alumina plate (15 x 15 mm2) resting on a thermocouple inside the tube furnace, adjusted for focus, and the microscope transfers the image of spherical pellet sample from the furnace at 5X magnification through a quartz window and onto the recording camera. The typical thermal profile should consist of non-isothermal heating, followed by holding isothermally at the desired temperature for a certain time and finally allowed to furnace cool back to room temperature. The entire set-up is assembled to a computer system facilitated with specific Misura Thermal Analysis software to acquire and store the images of pellets subjected to a thermal profile at predetermined time or temperature intervals.

38

Before proceeding with the experiments assuming uniform temperature across the pellet, one has to consider the difference in the temperature of pellets at the surface and at the core, when they are exposed to heat. The surface reaches the exposed temperature, whereas core remains at the initial temperature, and hence, there exist a thermal gradient across the pellet. As the reaction proceeds with time, the heat front propagates radially inwards increasing the temperature inside the pellet, and decreases the thermal gradient slowly with time and temperature of the whole pellet becomes homogenous when heat front reaches to core of the pellet. The time required to attain the homogenous temperature for the pellet is known as the characteristic time(߬), and needs to be estimated prior to assuming the uniform pellet temperature for further analysis. Some preliminary experiments have been done to measure the time required for the pellet temperature to become homogenous. This was done by embedding thermocouples (K-type) at the center and at the surface of the pellet to measure the difference between surface and center temperatures, respectively, during heating of the pellet in a furnace (at 1473 K (1200oC). It did not show any significant difference at the center and the surface temperatures after 30 seconds. This showed that the characteristic time for the center temperature to reach the surface temperature is very small as compared to that of sintering process. Therefore, it is reasonable to assume that the pellet has a uniform temperature throughout. These findings were also supported through analytical heat transfer calculations118, where transient conduction through the sphere is used to estimate the characteristic time for magnetite pellet. To study the sintering kinetics of oxidized magnetite (hematite) pellets during induration, the pellets are exposed to the thermal profile comprises of non-isothermal heating from room temperature to 800oC at a rate of 50 K/min (as pellet was already oxidized at 1023 K (750oC)), and then to the desired sintering temperature from 1423 – 1623 K (1150 – 1350oC) at a rate of 30 K/min, held isothermally at that temperature for 20 – 40 minutes, and finally furnace cooled to room temperature. In order to avoid stagnancy and probable deficiency of oxygen during sintering, the air (20% oxygen and 80% nitrogen) is flown continuously at the optimum rate of 0.3 liters per minute across the pellet in the furnace. The shadow images of the pellet in process are preset to capture at an interval of 15 seconds for the whole thermal profile. Similarly, for sintering studies of non-oxidized magnetite pellets, independently, the thermal profile followed is heating from room temperature to the desired sintering temperature at a rate of 39

30 K/min non-isothermally, held isothermally for the adequate time for sintering to happen from 20 – 90 minutes, and then furnace cooled to room temperature. As sintering of magnetite begins at relatively lower temperature as compared to hematite, the experiments are designed for isothermal sintering temperatures from 1173 – 1623 K (900 – 1350oC), and at intervals of 50 K at higher temperatures and 25 K at lower temperatures with an effort to capture the small dimensional changes. Since magnetite is very prone to oxidation even at lower temperatures (around 673 K (400oC)), all the experiments are done under inert atmosphere. This is done by passing argon (99.995 %) gas continuously at the maximum allowable flow rate of 0.5 liters per minute, so as to keep positive pressure inside the furnace to avoid any oxygen infringement from the surroundings. All the experiments have been repeated at least twice for reproducibility. In order to validate the predicted sintering behavior from the estimated kinetics, the experiments are designed for both oxidized magnetite and non-oxidized magnetite pellets. The thermal profile consists of non-isothermal heating at three different heating rates of 15, 30 and 45 K/min to the isothermal sintering temperature of 1573 K (1300oC), held for 20 minutes and cooled in the furnace, while continuously capturing the shadow images and further analyzed.

5.5

Characterization

Characterization is an essential part of research to ensure that the experiments, analysis and interpretation are aligned to derive appropriate findings. Characterization studies for the current work include density and porosity measurements, X – Ray Diffraction and Light Optical Microscopy. The green magnetite pellets used to study the sintering behavior are characterized at each stage – dried, oxidized and sintered for all the experiments; for both oxidized magnetite and magnetite pellets. 5.5.1 Density and Porosity measurements The two types of densities used for characterization of pellets are true density and bulk density. True or skeletal density is the ratio of mass to the volume of the sample excluding pores i.e., true volume. The true density of pellets is measured by means of a Helium Pycnometer (Accupyc II 40

1340, Micromeritics, USA), which works on the principle of gas displacement, and is much more accurate and reproducible than the traditional Archimedes water displacement method. Inert gases, such as helium or nitrogen, can be used as the displacement medium. Helium gas is used as the displacement medium because it is lighter and can penetrate deeper into the open pores. As experiments are designed on a single pellet, the same pellet is used for subsequent experimentation; therefore, the true density of the pellet needs to be measured in a nondestructive manner keeping the whole pellet intact. Hence, the sample cell of volume 100 cm3 with an opening of 12 mm diameter is chosen for the same purpose and used for all the pellets under consideration at each stage. The true density of the pellet measured using helium pycnometer has a standard deviation of 1 %. On the other hand, measurement of bulk density without destroying the surface of the pellet is quite tricky. The bulk density or envelope density is measured usually by mercury porosimeter or fine sand powder pycnometer (Geopyc 1360 by M/s Micromeritics, USA), both of which hamper the surface of the pellet and hence cannot be used for further experimentation. This necessitates development of a novel method of measuring bulk density in a non-destructive manner. This is achieved by using the setup of Light Table Imaging (LTI) developed by LKAB and MBV Systems AB. The set-up utilizes the concept of shadow imaging and comprises of a brightly illuminated table, a camera to capture images and a computer to store and analyze the images, as shown in Figure 11. The single pellet sample is kept on the illumination table, and its image is captured by the camera (Canon EOS with Sigma 1: 2.8 lens) equipped with and transferred to computer system for further analysis. The bright illumination and the dull reflectance of the pellet result in an image with a sharp boundary between the pellet and background.

41

Figure 11: Light Table Imaging setup

The shadow image of the pellet was analyzed for area fraction under the circular perimeter using the pixel count method with the help of image analysis software “Image J”. The pixel count is calibrated to a known distance in millimeters (mm) which is then used as a reference for all measurements of area fraction. Later, the area fraction of the image is extrapolated to volume of the pellet, assuming that the pellet is perfectly spherical, and divided by its weight to achieve the bulk density. A total of 20 images were taken for each pellet by orienting the pellet in different directions to capture the pellet surface from several angles to arrive at a statistically acceptable mean value. Since this non-destructive method of measuring bulk density is tailor made; it is standardized by comparing with that of a sand powder pycnometer (Geopyc 1360). This is done by measuring the bulk density values for standard sized Nitrile Buna Rubber (NBR) spheres from sand powder pycnometer (Geopyc 1360) and derived the Correlating Factor (CF). For making the comparison, NBR spheres of four different sets of calibrated size (6, 10, 11, 12 mm diameter) are analyzed by both the methods for 10 spheres in each set of size. The correlating factor is found out to be 1.2, and is being multiplied to the results of volume fraction obtained from analyzing the images from LTI. Once bulk density and true density are measured, porosity can be determined by Equation (9). Porosity evolved from the progress of the reaction inside the pellet is an indication of the 42

sintering. The density and porosity values obtained for oxidized magnetite and dried magnetite pellets are given in Table III.

ܲ‫ݕݐ݅ݏ݋ݎ݋‬, % =

்௥௨௘ ஽௘௡௦௜௧௬ି஻௨௟௞ ஽௘௡௦௜௧௬ ்௥௨௘ ஽௘௡௦௜௧௬

ܺ 100

(9)

Table III: Density and Porosity values Oxidized Magnetite

Dried Magnetite

Pellets

Pellets

3

5.30 – 5.45

5.15 – 5.40

3

3.40 – 3.71

2.18 – 2.43

31.68 – 37.63

52.43 – 59.58

True Density (gm./cm ) Bulk Density (gm./cm ) Porosity (%)

5.5.2 X – Ray Diffraction (XRD) The pellets of oxidized magnetite and non-oxidized magnetite after sintering at different temperatures are analyzed for its mineral constituents by powder XRD using Bregg-Brentano geometry with copper (Cu) target. The pellet is cut in two halves by diamond saw, with one half grinded in mortar for XRD analysis, whereas the other one is used for microscopic analysis. The instrumental setup used for XRD is provided by PANalytical. Since only half of the pellet is used in powdered form, the amount is too small for it to be loaded in a regular sample case. Therefore, silicon sample case which is designed for a low amount of sample is used, where a single layer of powder is carefully poured and placed in the sample holder. Accordingly, the diaphragm, incident and anti-scatter slit are adjusted. The sample holder along with the samples is then placed inside the XRD chamber and scanned in the 2ș range of 10 – 90o. The resulting diffractogram is analyzed by High Score Plus software for mineral identification by matching the peak positions using the database developed by International Centre for Diffraction Data (JCPDS-ICDD). 43

5.5.3 Microscopic studies Light Optical Microscopy (LOM) is used for studying the microstructures of the sintered pellets. The other half of the pellet sample cut by diamond saw is cold mounted in epoxy and polished to a fineness of 1 μm with diamond pastes. The optical images of sintered pellets exposed to different temperatures are captured at three locations – center, middle and surface on the pellet cross section, and at different magnifications. The optical microscope supplied by Zeiss (Imager.M2m) is integrated with image analyzing software provided by AxioVision. Further, the overview microstructure of the whole pellet is generated, where the several images are progressively captured within a frame of 5 X 5 μm2 at 200X across the pellet surface, and are stitched together. These stitched images are intended to be automatically analyzed for quantification of different phases present in the whole pellet while considering various image analyzing parameters.

44

6.

Results and Discussion

The results from light optical microscopy and x-ray diffraction are presented at first as preliminary validation for experiments carried out, followed by the experimental results obtained from the optical dilatometer and their subsequent analysis for both oxidized and non-oxidized magnetite pellets.

6.1

Preliminary Microstructural Evaluation

The purpose of microstructural evaluations presented here serves mainly to ensure that the aforementioned approach and the analysis adapted are aligned in a proper direction, especially for non-oxidized magnetite pellets. The detailed study on microstructural characterization and their correlation with the derived sintering kinetics is planned to be done in near future. First of all, the overall view of sintered non-oxidized magnetite pellets is generated by stitching several microstructures at the magnification of 200X, with the focus on rim of pellet cross section, and one of them is shown in Figure 12. This was done to ensure that there is no significant oxidation when the magnetite pellet is exposed to the designed thermal and gaseous atmosphere while studying its sintering kinetics. It is observed that there are traces of oxidation at the rim of the magnetite pellet. The maximum uniform thickness of the spread of these traces is approximately 50 – 70 μm. Assuming that the oxidation layer is concentric, the degree of oxidation calculated is found to be less than 4.2 %.

45

Figure 12: Microstructure of a magnetite pellet (10 mm in diameter) (right) exposed to a sintering temperature of 1473 K (1200oC) with focus (left) on rim of the pellet.

Figure 13: XRD peaks of raw magnetite concentrate and the pellet exposed to highest sintering temperature 1623 K (1350oC) 46

The microstructural observations are coupled with the XRD studies for the same purpose. The XRD patterns for raw material (magnetite concentrate) and non-oxidized magnetite pellet exposed to the highest sintering temperature (1623 K (1350oC)) are shown in Figure 13, where major magnetite peaks were found (without those of hematite). The hematite reference peak positions are also indicated. Similar diffractograms with magnetite peaks have been observed for all pellets with sintering temperatures varying from 1173 to 1623 K (900 to 1300oC). Since the detection limit for mineral constituents in XRD is approximately 2 % for fairly homogenous material, this substantiates the microstructural observation that oxidation at the rim of pellet is less than 3 %. Since this is such a small percentage and that too at the surface of pellet and not at the grains level, it is not expected to interfere or have a significant effect on sintering of magnetite pellets. Microstructures have also been observed to follow the progress of sintering when they are exposed to different thermal profiles. The different stages of sintering for oxidized and non-oxidized magnetite pellets at a magnification of 200X can be seen in Figure 14, where initially grains come closer at lower temperatures, while neck formation begins at intermediate temperatures and grains adhere to each other by forming a continuous network and rounding the pores at higher temperatures.

47

(a)

(b) Figure 14: Microstructures of (a) oxidized magnetite and (b) non-oxidized magnetite pellets following sintering stages with increase in temperature at a magnification of 200X

48

6.2

Shrinkage during Sintering

The shrinkage during sintering, in terms of change in area fraction is measured by an optical dilatometer. The typical plot of the percentage area change (Equation (10)) with respect to time obtained from the optical dilatometer following a desired thermal profile is shown in Figure 15. The area change plots for a couple of experimental runs for same thermal profile show that the results from optical dilatometer are quite reproducible, and found to be consistent. ߜ஺,௢௩௘௥௔௟௟ =

஺ି஺బ ஺బ

× 100

(10)

where, Ԣ‫ܣ‬Ԣ refers to projected area of the pellet at any instant of time and Ԣ‫ܣ‬଴ Ԣ refers to that at the start of the experiment.

Figure 15: Typical plot for area change with time and temperature obtained from an optical dilatometer during induration of pellets 49

As mentioned, the typical thermal profile for the induration of pellets consists of three segments – non-isothermal heating, isothermal holding, and furnace cooling. The pellet expands initially, and after reaching a maximum it begins to shrink under the influence of designed thermal profile. The expansion of the pellet initially is by virtue of material’s thermal expansion property, and once the temperature reaches around 1173 – 1273 K (900 – 1000oC), the sintering phenomena begins to dominate, resulting in the overall shrinkage of the pellet. Shrinkage in the isothermal regime is due to sintering alone and thereafter, during cooling, the pellet continues to shrink along with the influence of thermal contraction. Since the experiments are quite reproducible, the result of one of the experiments is shown from here onward for the sake of clarification, except few instances, where both the experimental results have been plotted.

6.3

Degree of Sintering

Since the overall change in size of the pellet is due to the combination of the thermal expansion/contraction and the sintering phenomena. It is therefore necessary to isolate the thermal expansion/contraction phenomena from the overall percentage change in area to obtain the shrinkage due to sintering alone. This is done by plotting the percentage of area change as a function of temperature for both oxidized magnetite and non-oxidized magnetite pellets, as shown in Figure 16. This infers that the pellet expands linearly at low temperatures where sintering is insignificant, then shrinks non-linearly as temperatures go beyond 1173 – 1273 K (900 – 1000oC), and subsequently shrinks linearly during the cooling segment once the temperature is sufficiently low where the sintering becomes insignificant. Thermal expansion and thermal contraction are the material properties, and theoretically their slopes should be equal, which is also evident from the similar slopes obtained for area change during the heating and cooling segments. Therefore, sintering can be extracted from the overall shrinkage of the pellet, with the help of known slope of thermal expansion, which accounts for thermal contraction in calculations.

50

(a)

(b) Figure 16: Area change with respect to temperatures for (a) oxidized magnetite and (b) nonoxidized magnetite pellets exposed to different isothermal sintering temperatures during induration 51

The volumetric thermal coefficient of expansion (ߙ) by definition is given by Equation (11), ଵ ௗ௏

ߙ = ௏ ௗ்

(11)

By integration, ο௏

ቀ ௏ ቁ = exp൫ߙ(ܶ െ ܶ଴ )൯ ؆ 1 + ߙ(ܶ െ ܶ଴ ) బ

(Reasonable approximation for Į 10-6 and temperature < 1573 K (1300oC), vacancy mechanism is predominant126, which is plausible to operate in the range of temperatures being studied in the current investigation. However, the negative activation energy corresponding to this mechanism has also not been observed. Interestingly, Xuebin et al.121 observed a trend in activation energy for the sintering kinetics of hydroxyapatite similar to that estimated in the current study for magnetite pellets. They correlated the kinetic parameters ݊ and ܳ to the morphology, which infers the probable mechanisms responsible for sintering. They speculate that the high activation energy at lower temperatures is due to the initial neck formation through reaction and surface diffusion. Contemplating in the similar context for the findings of sintering of magnetite pellet, it might be possible that the bentonite/gangue minerals initiate neck formation through reaction and solid-state diffusion at the initial stages. Later, at higher temperatures, there might be formation of liquid phase bridges, decreasing the activation energy for sintering73. Quantifying the extent of sintering by microstructural evaluation, and correlating them to the currently observed sintering kinetics may impart more insights into the sintering mechanisms at different temperatures. The quantification of morphologies using the automated image analysis on microstructures of pellets is planned in the near future.

70

8.

Conclusions

The sintering kinetics of single magnetite pellets during induration has been successfully estimated with the approach of capturing in situ shrinkage, and quantifying them. An optical dilatometer is successfully used to capture the shrinkage, and hence sintering of the pellet, instead of traditionally used push-rod dilatometer. Sintering phenomena during induration of magnetite pellets are categorized into that of oxidized magnetite and non-oxidized magnetite, and studied independently. Sintering is quantified using isothermal shrinkage data by estimating three kinetic parameters – activation energy (ܳ), pre-exponential factor (‫ܭ‬Ԣ) and time exponent (݊). The sintering of oxidized magnetite pellets estimated the value of ݊ to be 0.22, while activation energy is 509 kJ/mole over the range of sintering temperatures (1373 – 1623 K (1100 – 1350oC)). This suggests that there exists a single dominant mechanism during sintering of oxidized magnetite pellets. On the other hand, the sintering of non-oxidized magnetite pellets shows that the value of ݊ varies from 0.45 to 0.15 with the increase in temperature. The two activation energy values estimated for the sintering of magnetite pellets are 477 and 148 kJ/mole at lower temperatures (1173 – 1373 K (900 – 1100oC)) and higher temperatures (1373 – 1623 K (1100 – 1350oC)), respectively, suggesting the possibility of two distinct mechanisms. The extents of sintering have been predicted under non-isothermal conditions as well by incorporating the variations in the above mentioned kinetic parameters(݊, ‫ܭ‬Ԣ and ܳ), with temperatures, for both oxidized and nonoxidized magnetite pellets, and are validated using experimental data. Quantitative description of sintering in iron ore pellets is useful in predicting the state of the pellet during industrial induration process, and also helps in optimizing the appropriate raw material mixes for making pellets, as well as pellet quality and performance in subsequent iron making process. This can also provide additional advantage when designing the furnace for efficient operation, with relatively lower investment on experimentations, considering the variability in the raw materials in future.

71

9.

Future Plan

With the objective of modeling the evolution of magnetite pellet during induration process; following are the steps ahead for future in succession to the findings of sintering phenomena obtained in current study: i.

The next phase in the project is to experimentally study the other major phenomenon during induration, i.e. oxidation of magnetite pellets, independently, and estimate their kinetics.

ii.

The independent behavior of magnetite pellet under oxidation and sintering phenomena during induration needs to be characterized, and correlated to their estimated kinetic parameters. Therefore, the detailed characterization of the experimental pellets will be done in terms of density, porosity, pore size distribution and most importantly microstructural evaluation. It is also intended to quantify the morphology of the pellets by automated image analysis to derive the degree of sintering, and substantiate the extent of processes found experimentally.

iii.

Thereafter, in the last phase of project, the kinetic parameters estimated for oxidation (of magnetite) and sintering (of oxidized magnetite and non-oxidized magnetite) phenomena in isolation, will be integrated together to develop a mathematical model simulating the overall induration process of magnetite pellet. The model developed will be validated by single pellet experiments following the complete induration profile for the pellets exposed to different temperatures, and at different heating rates.

The model for predicting induration behavior of magnetite pellet will provide an effective aid to the furnace operator, and helps in designing the optimum operating profile to achieve the desired pellet quality, improving the furnace productivity and performance. It can also be very useful to design the operating profile considering the variability in the raw material mix in the future.

72

10.

Bibliography

1. "The Three R's of Sustainable Steel", World Steel Association, 2009. 2. Ball D.F., Darntell J., Grieve A. and Wild R.: "Agglomeration of Iron Ores", 1973, Heinemann Educational Books Limited, London. 3. Aaro,L.: "LKAB – The Way Forward", 2nd METEC - InSteelCon, Dusseldorf, Germany, 2011, pp. 1-5. 4. LKAB: "Annual and Sustainability Report", Lulea, 2014, . 5. Cooke S.R. and Stowasser Jr W.F.: "The Effect of Heat Treatment and Certain Additives on the Strength of Fired Magnetite Pellets", Trans. AIME, 1952, 193, (19521), pp. 1223-1230. 6. Papanastassiou D. and Bitsianes G.: "Mechanisms and Kinetics Underlying the Oxidation of Magnetite in the Induration of Iron Ore Pellets", Metallurgical Transactions, 1973, 4, (2), pp. 487496. 7. Wynnyckyj J.R. and McCurdy W.A.: "Causes of the Shell and Core Structure and Layered Inhomogeneity in Iron Ore Pellets.” Metallurgical Transactions, 1974, 5, (10), pp. 2207-2215. 8. Niiniskorpi V.: "Phases and Microstructures in LKAB's Olivine and Dolomite Fluxed Pellets", 60th Ironmaking Conference, 2001, pp. 767-780. 9. Kawatra S.K. and Ripke S.J.: "Sintering of Magnetite Pellets under Oxidizing, Neutral and Reducing Atmospheres", Miner Metall Process, 2003, 20, (4), pp. 165. 10. Forsmo S.P.E., Forsmo S., Samskog P. and Björkman B.M.T.: "Mechanisms in Oxidation And Sintering Of Magnetite Iron Ore Green Pellets", Powder Technol, 2008, 183, (2), pp. 247259. 11. Semberg P., Rutqvist A., Andersson C. and Björkman B.: "Interaction between Iron Oxides and Olivine in Magnetite Based Pellets during Reduction at Temperatures below 1000°C", Ironmaking and Steelmaking, 2011, 38, (5), pp. 321-328. 73

12. Semberg P., Andersson C. and Björkman B.: "Interaction between Iron Oxides and Olivine in Magnetite Pellets During Reduction to Wustite at Temperatures of 1000-1300°C", ISIJ Int, 2013, 53, (3), pp. 391-398. 13. Cho H.J., Tang M. and Pistorius P.C.: "Magnetite Particle Size Distribution and Pellet Oxidation", Metallurgical and Materials Transactions B, 2014, 45, (4), pp. 1213-1220. 14. Tang M., Cho H.J. and Pistorius P.C.: "Early Gaseous Oxygen Enrichment To Enhance Magnetite Pellet Oxidation", Metallurgical and Materials Transactions B, 2014, 45 B, (August), pp. 1304-1314. 15. Forsmo S.: "Influence of Green Pellet Properties on Pelletizing Of Magnetite Iron Ore", [PhD Thesis], Luleå Tekniska Universitet, 2007. 16. Libsch K.D. and Tarby S.K.: "Magnetite Oxidation in a Traveling Grate Pellet Plant: A Computer Model", Metallurgical Transactions, 1973, 4, (5), pp. 1347-1353. 17. Thurlby J., Batterham R. and Turner R.: "Development and Validation of a Mathematical Model for the Moving Grate Induration of Iron Ore Pellets", Int J Miner Process, 1979, 6, (1), pp. 43-64. 18. Thurlby J.:"A Dynamic Mathematical Model of the Complete Grate/Kiln Iron–Ore Pellet Induration Process", Metallurgical Transactions B, 1988, 19, (1), pp. 103-112. 19. Küçükada K., Thibault J., Hodouin D., Paquet G. and Caron S.: "Modeling of a Pilot Scale Iron Ore Pellet Induration Furnace", Can Metall Q, 1994, 33, (1), pp. 1-12. 20. Chejne Janna F., Ruiz Sierra J. and Badie J.: "Mathematical Model For Energy Saving In Induration of Iron Ore Pellets", Ironmaking Steelmaking, 1996, 23, (5), pp. 406-410. 21. Cross M. and Blot P.: "Optimizing the Operation Of Straight-Grate Iron-Ore Pellet Induration Systems Using Process Models", Metallurgical and Materials Transactions B, 1999, 30, (4), pp. 803-813.

74

22. Hamidi A. and Payab H.: "Mathematical Model for Energy Saving in Induration of Iron Ore Pellets containing Solid Fuel", International Journal of Engineering Transactions B, 2003, 16, pp. 265278. 23. Sadrnezhaad S., Ferdowsi A. and Payab H.: "Mathematical Model for a Straight Grate Iron Ore Pellet Induration Process of Industrial Scale", Computational Materials Science, 2008, 44, (2), pp. 296-302. 24. Barati M.: "Dynamic Simulation of Pellet Induration Process in Straight-Grate System", Int J Miner Process, 2008, 89, (1), pp. 30-39. 25. Nordgren S., Lindblom B., Dahl J. and Wang C.: "Process Integration in an Iron Ore Upgrading Process System E Analysis of Mass and Energy Flows within a Straight Grate Induration Furnace", 18th International Congress of Chemical and Process Engineering, Prague, Czech Republic, 2008. 26. Granstrom B.R., Lundstrom T.S., Marjavaara B.D. And Toyra S.: "CFD Modeling of the Flow through a Grate-Kiln", 7th International Conference on CFD in the Minerals and Process Industries - CSIRO, Melbourne, Australia, 2009, pp. 1-6. 27. Gustafsson G.: "Mechanical Characterization And Modeling Of Iron Ore Pellets", 2012, . 28. Athayde M., Cotta L., Possani L., Trindade L. and Contini A.: "Pelletizing Pilot Scale Simulation through Mathematical Modelling: A Goethite- Martite Iron Ore Study", METEC and 2nd ESTAD, Dusseldorf, Germany, 2015, pp. 1-6. 29. Reidetschlaeger J., Hiebl B. and Fehringer E.: "Numerical Simulation of the Sintering and the Pelletizing Processes", METEC - 2nd ESTAD, Dusseldorf, Germany, 2015, pp. 1-9. 30. Batterham R.: "Modeling the Development of Strength in Pellets", Metallurgical Transactions B, 1986, 17, (3), pp. 479-485. 31. Davis R.: "Mathematical Model of Magnetite Oxidation in a Rotary Kiln Furnace", The Canadian Journal of Chemical Engineering, 1996, 74, (6), pp. 1004-1009.

75

32. German R.M.: "Sintering Theory and Practice", 1996, 1, Wiley Interscience, New York. 33. Wynnyckyj J.R. and Fahidy T.Z.: "Solid State Sintering in the Induration of Iron Ore Pellets", Metall Trans, 1974, 5, (5), pp. 991-1000. 34. Kuczynski G.C.: "Self-diffusion In Sintering of Metallic Particles", AIME TRANS, 1949, 185, pp. 169-178. 35. Semberg P., Rutqvist A., Andersson C. and Björkman B.: "Interactions between Iron Oxides and the Additives Quartzite, Calcite and Olivine in Magnetite Based Pellets", ISIJ Int, 2011, 51, (2), pp. 173-180. 36. Hansen J.D., Rusin R.P., Teng M. and Johnson D.L. :"Combined-Stage Sintering Model", J Am Ceram Soc, 1992, 75, (5), pp. 1129-1135. 37. Kosanoviü D., Filipoviü S., Obradoviü N., Pavloviü V. and Ristiü M.: "Microstructure Evolution and Sintering Kinetics of ZnO", Journal of Applied Engineering Science, 2011, 9, (2), pp. 317-322. 38. Holková Z., Pach L., Kovár V. and Svetík S.: "Kinetic study of Al2O3 Sintering by Dilatometry", Ceramics-Silikáty, 2003, 47, (1), pp. 13-19. 39. Ewsuk K.G., Ellerby D.T. and Diantonio C.B.: "Analysis of Nanocrystalline and Microcrystalline ZnO Sintering using Master Sintering Curves", J Am Ceram Soc, 2006, 89, (6), pp. 2003-2009. 40. Mazaheri M., Golestani-Fard F. and Simchi A.: "An Investigation on Densification of Nanocrystalline 3Y-TZP using Master Sintering Curve", 10th International Conference and Exhibition of the European Ceramic Society, pp.1-5. 41. Nikoliü M.V., Pavloviü V.P., Pavloviü V.B., Labus N. and Stojanoviü B.: "Application of the Master Sintering Curve Theory to Non-Isothermal Sintering of BaTiO3 Ceramics", Materials science forum, 2005, Trans Tech Publ, pp. 417-422.

76

42. Gasik M., Keski-Honkola A. and Bilotsky Y.: "Sintering Kinetics of Titanium and Titanium Alloy Powders", 17th Plansee Seminar, 2009, pp. 70/1-70/10. 43. Coble R.L.: "Initial Sintering of Alumina and Hematite", J Am Ceram Soc, 1958, 41, (2), pp. 55-62. 44. Johnson D.L. and Cutler I.B.: "Diffusion Sintering: II Initial Sintering Kinetics of Alumina", J Am Ceram Soc, 1963, 46, (11), pp. 545-550. 45. Kutty T., Khan K., Hegde P., Sengupta A., Majumdar S. and Kamath H.: "Determination of Activation Energy of Sintering of ThO2 - U3O8 Pellets using the Master Sintering Curve Approach", Science of Sintering, 2003, 35, (3), pp. 125-132. 46. Zeng W., Gao L., Gui L. and Guo J.: "Sintering Kinetics of Į- Al2O3 Powder", Ceram Int, 1999, 25, (8), pp. 723-726. 47. Lahiri D., Rao S., Rao G. and Srivastava R.: "Study on Sintering Kinetics and Activation Energy of UO2 Pellets using Three Different Methods", J Nucl Mater, 2006, 357, (1), pp. 88-96. 48. Dehaudt P., Bourgeois L. and Chevrel H.: "Activation Energy of UO2 and UO2x Sintering", J Nucl Mater, 2001, 299, (3), pp. 250-259. 49. Su H. and Johnson D.L.: "Master Sintering Curve: A Practical Approach to Sintering", J Am Ceram Soc, 1996, 79, (12), pp. 3211-3217. 50. Joshi S., Krishnan R. and Lave L.: "Estimating the Hidden Costs of Environmental Regulation", The Accounting Review, 2001, 76, (2), pp. 171-198. 51. Kuskov V.B., Kuskov J.V., Kornev A.V. and Trushko V.L.: "Preparation of Iron Ore For Metallurgical Processing", CIS Iron and Steel Review, 2011, pp. 4-5. 52. Fruehan R.J.: "The Making, Shaping and Treating of Steel: Ironmaking Volume". 1999, 2, AISE Steel Foundation.

77

53. Svensson K.J.V.: "Process for Cold-Hardening Of Shaped Bodies", US Patent (US3490895 A). 1970. 54. Dwarapudi S., Tathavadkar V., Rao B.C., Sandeep Kumar T.K., Ghosh T.K. and Denys M.: "Development of Cold Bonded Chromite Pellets for Ferrochrome Production in Submerged Arc Furnace", ISIJ Int, 2013, 53, (1), pp. 9-17. 55. Eisele T. and Kawatra S.: "A Review of Binders in Iron Ore Pelletization", Miner Process Extr Metal Rev, 2003, 24, (1), pp. 1-90. 56. Singh M. and Björkman B.: "Effect of Processing Parameters on the Swelling Behavior of Cement-Bonded Briquettes", ISIJ Int, 2004, 44, (1), pp. 59-68. 57. Dutta D.K., Bordoloi D. and Borthakur P.C.: "Investigation on Reduction of Cement Binder in Cold Bonded Pelletization of Iron Ore Fines", Int J Miner Process, 1997, 49, (1), pp. 97-105. 58. Su F., Lampinen H. and Robinson R.: "Recycling of Sludge And Dust To The BOF Converter By Cold Bonded Pelletizing", ISIJ Int, 2004, 44, (4), pp. 770-776. 59. Robinson R.: "High Temperature Properties of By-Product Cold Bonded Pellets Containing Blast Furnace Flue Dust", Thermochimica acta, 2005, 432, (1), pp. 112-123. 60. Ahmed H., Persson A., Sundqvist L. and Bjorkman B.: "Energy Efficient Recycling of InPlant Fines", Energy, World Academy of Science, Engineering and Technology, 1:9689. 61. Patisson F., Bellot J., Ablitzer D., Marliere E., Dulcy C. and Steiler J.: "MathematicalModeling Of Iron-Ore Sintering Process", Ironmaking Steelmaking, 1991, 18, (2), pp. 89-95. 62. Mitterlehner J., Löffler G., Winter F., et al.: "Modeling and Simulation of Heat Front Propagation in the Iron Ore Sintering Process", ISIJ Int, 2004, 44, (1), pp. 11-20. 63. Biswas A.K.: "Principles of Blast Furnace Ironmaking", Cootha Publishing House, 1981. 64. Ghosh A. and Chatterjee A.: "Iron making and Steelmaking: Theory and Practice". 2008, PHI Learning Pvt. Ltd. 78

65. Mayer K.: "Pelletization of Iron Ores", Springer, 1980. 66. Feng J., Zhang Y., Zheng H., Xie X. and Zhang C.: "Drying and Preheating Processes of Iron Ore Pellets in a Traveling Grate", International Journal of Minerals, Metallurgy, and Materials, 2010, 17, (5), pp. 535-540. 67. Ljung A., Lundström T.S., Marjavaara B.D. and Tano K.: "Convective Drying of an Individual Iron Ore Pellet–Analysis with CFD", Int J Heat Mass Transfer, 2011, 54, (17), pp. 38823890. 68. Ljung A., Frishfelds V., Lundström T.S. and Marjavaara B.D.: "Discrete and Continuous Modeling of Heat and Mass Transport in Drying of a Bed of Iron Ore Pellets", Drying Technol, 2012, 30, (7), pp. 760-773. 69. Majumder S., Natekar P.V. and Runkana V.: "Virtual Indurator: A Tool for Simulation of Induration of Wet Iron Ore Pellets On A Moving Grate", Comput Chem Eng, 2009, 33, (6), pp. 1141-1152. 70. Tsukerman T., Duchesne C. and Hodouin D.: "On the Drying Rates of Individual Iron Oxide Pellets", Int J Miner Process, 2007, 83, (3), pp. 99-115. 71. Morrissey J.P.: "Discrete Element Modeling Of Iron Ore Pellets to include the Effects of Moisture and Fines", [PhD Thesis], The University of Edinburgh, 2013. 72. Knache O., Kubaschewski O. and Hesselmann K.: "Thermochemical Properties of Inorganic Substances", Springer-Verlag, Berlin, 1991. 73. Bentell L.: "Means of Improving High-Temperature Reducibility of Pellets for Blast Furnace Use", Scand J Metall, 1981, 10, (5), pp. 205-209. 74. Friel J.J. and Erickson E.S.: "Chemistry, Microstructure, and Reduction Characteristics of Dolomite-Fluxed Magnetite Pellets", Metallurgical Transactions B, 1980, 11, (2), pp. 233-243.

79

75. Firth A.R. and Garden J.F.: "Interactions between Magnetite Oxidation and Flux Calcination during Iron Ore Pellet Induration", Metall Mat Trans B Process Metall Mat Process Sci, 2008, 39, (4), pp. 524-533. 76. Firth A.R., Douglas J.D. and Roy D.: "Understanding Reactions In Iron Ore Pellets", 2011, pp. 413-423. 77. Granse L.: "Influence of Slag - Forming Additions on the Swelling of Pellets from Very Rich Magnetite Concentrates", Proc Internat Conf Sci Technol Iron Steel, Tokyo, 1971, pp. 45-51. 78. Dwarapudi S., Banerjee P., Chaudhary P., et al.: "Effect of Fluxing Agents on the Swelling Behavior of Hematite Pellets", Int J Miner Process, 2014, 126, pp. 76-89. 79. Dwarapudi S., Ghosh T.K., Shankar A., Tathavadkar V., Bhattacharjee D. and Venugopal R.: "Effect of Pyroxenite Flux on the Quality and Microstructure of Hematite Pellets", Int J Miner Process, 2010, 96, (1), pp. 45-53. 80. Kuczynski G.C.: "Study of the Sintering of Glass", Journal Of Applied Physics, 1949, 20, (12), pp. 1160-1163. 81. Hayes P.: "Chapter 4.7 - factors influencing reaction area", Treatise on Process Metallurgy, Seetharaman Seshadri, ed., 2014, Elsevier, Boston. 82. German R. and Munir Z.: "Surface Area Reduction during Isothermal Sintering", J Am Ceram Soc, 1976, 59, (9-10), pp. 379-383. 83. German R. and Munir Z.: "A Kinetic Model for the Reduction in Surface Area during Initial Stage Sintering", Sintering and Catalysis, Springer, 1975. 84. Wakai F. and Aldinger F.: "Sintering through Surface Motion by the Difference in Mean Curvature", Acta materialia, 2003, 51, (14), pp. 4013-4024. 85. Xu X., Lu P. and German R.M.: "Densification and Strength Evolution in Solid-State Sintering Part II Strength Model", J Mater Sci, 2002, 37, (1), pp. 117-126.

80

86. German R.M.: "Powder metallurgy and particulate materials processing: The processes, materials, products, properties, and applications", 2005, Metal Powder Industries Federation, Princeton, NJ. 87. Mehrer H.: "Diffusion in Solids: Fundamentals, Methods, Materials, Diffusion-Controlled Processes". 2007, 155, Springer Science & Business Media. 88. Ristiü M.M. and Miloseviü S.: "Frenkel's Theory of Sintering", Science of Sintering, 2006, 38, (1), pp. 7-11. 89. Shaler A. and Wulff J.: "Mechanism of Sintering", Industrial & Engineering Chemistry, 1948, 40, (5), pp. 838-842. 90. Mackenzie J. and Shuttleworth R.: "A Phenomenological Theory of Sintering", Proceedings of the Physical Society, Section B, 1949, 62, (12), pp. 833. 91. Hausner H.: "Discussion on the Definition of the Term “Sintering”", Sintering–New developments: Proceedings of the 4th International Round Table Conference on Sintering, 1977, pp. 5-10. 92. Kingery W.: "Regelation, Surface Diffusion and Ice Sintering", J Appl Phys, 1960, 31, (5), pp. 833-838. 93. Frenkel J.: "On the surface Motion of Particles in Crystals and the Natural Roughness of Crystalline Faces", J.Phys.USSR, 1945, 9, pp. 392-398. 94. Fahidy T. and Wynnyckyj J.: "Simulation of a Sintering Pellet", Simulation, 1973, 20, (3), pp. 73-79. 95. Johnson D.L.: "Comment on Temperature-Gradient-Driven Diffusion in Rapid-Rate Sintering", J Am Ceram Soc, 1990, 73, (8), pp. 2576-2578. 96. Somiya S. and Moriyoshi Y.: "Sintering key papers", 2012, Springer Science & Business Media. 97. Shi J.: "Solid State Sintering Of Ceramics: Pore Microstructure Models, Densification Equations and Applications", J Mater Sci, 1999, 34, (15), Pp. 3801-3812. 81

98. Burke J.: "Role of Grain Boundaries in Sintering", J Am Ceram Soc, 1957, 40, (3), pp. 80-85. 99. Clark P. and White J.: "Some Aspects of Sintering", Trans. Brit. Ceram. Soc, 1950, 49, (7), pp. 305-333. 100. Kingery W.D. and Berg M.: "Study of the Initial Stages Of Sintering Solids By Viscous Flow, Evaporation-Condensation, and Self-Diffusion", J Appl Phys, 1955, 26, (10), pp. 1205-1212. 101. Kossel W.: "Extenoling the Law of Bravais", Nach.Ges.Wiss.Gottingen, 1927, 143. 102. Stranski I.: "Zur Theorie des Kristallwachstums", Z Phys Chem, 1928, 136, (259), pp. 259278. 103. German R. and Lathrop J.: "Simulation of Spherical Powder Sintering By Surface Diffusion", J Mater Sci, 1978, 13, (5), pp. 921-929. 104. German R. and Munir Z.: "Morphology Relations during Bulk-Transport Sintering", Metallurgical and Materials Transactions A, 1975, 6, (12), pp. 2229-2234. 105. Young W.S. And Cutler I.B.: "Initial Sintering with Constant Rates of Heating", J Am Ceram Soc, 1970, 53, (12), pp. 659-663. 106. Coble R.L.: "Effects of Particle-Size Distribution in Initial-Stage Sintering", J Am Ceram Soc, 1973, 56, (9), pp. 461-466. 107. Ashby M.: "A First Report on Sintering Diagrams", Acta Metallurgica, 1974, 22, (3), pp. 275289. 108. Swinkels F. and Ashby M.: "A Second Report on Sintering Diagrams", Acta Metallurgica, 1981, 29, (2), pp. 259-281. 109. German R. and Munir Z.: "The Geometry of Sintering Wires", J Mater Sci, 1975, 10, (10), pp. 1719-1724.

82

110. Nichols F.: "Theory of Sintering of Wires by Surface Diffusion", Acta Metallurgica, 1968, 16, (1), pp. 103-113. 111. Beere W.: "The Sintering and Morphology of Interconnected Porosity in UO2 Powder Compacts", Journal of Materials Science, 1973, 8, (12), pp. 1717-1724. 112. Herring C.: "Effect of Change of Scale on Sintering Phenomena", J Appl Phys, 1950, 21, (4), pp. 301-303. 113. Liu Z.Y., Loh N.H., Khor K.A. and Tor S.B.: "Sintering Activation Energy Of Powder Injection Molded 316L Stainless Steel", Scr Mater, 2001, 44, (7), pp. 1131-1137. 114. Fang T., Shiue J. and Shiau F.: "On the Evaluation of the Activation Energy of Sintering", Mater Chem Phys, 2003, 80, (1), pp. 108-113. 115. Blaine D., Gurosik J., Park S., Heaney D. and German R.: "Master Sintering Curve Concepts as Applied to the Sintering of Molybdenum", Metallurgical and Materials Transactions A, 2006, 37A, (3), pp. 715-720. 116. Shao W., Chen S., Li D., Cao H. and Zhang S.: "Construction of the Master Sintering Curve for Submicron Size Į-Al2O3 based on Non-Isothermal Sintering Containing Lower Heating Rates Only", Materials Science (0137-1339), 2009, 27, (2). 117. Karamanov A., Dzhantov B., Paganelli M. and Sighinolfi D.: "Glass Transition Temperature and Activation Energy of Sintering by Optical Dilatometry", Thermochimica Acta, 2013, 553, (32), pp. 1-7. 118. Nurni V.N. and Ballal B.N.: "Chapter 4.1 - rate phenomena in process metallurgy", Treatise on Process Metallurgy, Seetharaman Seshadri, ed., 2014, Elsevier, Boston. 119. Skinner B.J.: "Section 6: Thermal expansion", Geological Society of America Memoirs, 1966, 97, pp. 75-96. 120. German R.M., Messing G.L. and Cornwall R.G. : "Sintering technology", 1996, Marcel Dekker. 83

121. Xuebin Z., Yunfei D., Songlin W., Jie X. and Yi F.: "Sintering Behavior and Kinetic Evaluation of Hydroxyapatite Bio-Ceramics from Bovine Bone", Ceramics – Silikáty, 2010, 54, (3), pp. 248-252. 122. Gaskell D.R.: "Introduction to the Thermodynamics of Materials". 5th ed., 2008, 2, CRC Press, New York, USA. 123. Spencer P. and Kubaschewski O.: "A Thermodynamic Assessment of the Iron-Oxygen System", Calphad, 1978, 2, (2), pp. 147-167. 124. Vallet P. and Carel C.: "The Fe-O (iron-oxygen) Phase Diagram in the Range of the Nonstoichiometric Monoxide and Magnetite at the Fe-rich Limit: Reduction Diagrams", Journal of Phase Equilibria, 1989, 10, (3), pp. 209-218. 125. Becker K.D., Wurmb V. V., and Litterst F. J.: "Mössbauer Study of High-Temperature Diffusion in Magnetite", Hyperfine Interactions, 1990, 56, (1-4), pp. 1431-1435. 126. Dieckmann R. and Schmalzried H.: "Defects and Cation Diffusion in Magnetite (I)", Berichte der Bunsengesellschaft für Physikalische Chemie, 1977, 81, (3), pp. 344-347. 127. Dieckmann R. and Schmalzried H.: "Defects and Cation Diffusion in Magnetite (II)", Berichte der Bunsengesellschaft für Physikalische Chemie, 1977, 81, (4), pp. 414-419. 128. Lewis G., Catlow C.R.A. and Cormack A.: "Defect Structure and Migration in Fe3O4", Journal of Physics and Chemistry of Solids, 1985, 46, (11), pp. 1227-1233. 129. Hallström S., Höglund L. and Ågren J.: "Modeling of Iron Diffusion in the Iron Oxides Magnetite and Hematite with Variable Stoichiometry", Acta Materialia, 2011, 59, (1), pp. 53-60. 130. Schmalzried H.: "Diffusion in Oxides", Reactivity of Solids, 1988, 5, (4), pp. 269-278. 131. Peterson N.L., Chen W. and Wolf D.: "Correlation and Isotope Effects for Cation Diffusion in Magnetite", Journal of Physics and Chemistry of Solids, 1980, 41, (7), pp. 709-719.

84

132. Van Orman J.A. and Crispin K.L.: "Diffusion in Oxides", Reviews in Mineralogy and Geochemistry, 2010, 72, (1), pp. 757-825. 133. Levy D., Giustetto R. and Hoser A.: "Structure of Magnetite (Fe3O4) above the Curie temperature: A Cation Ordering Study", Physics and Chemistry of Minerals, 2012, 39, (2), pp. 169176. 134. Becker K. D., Wurmb V.V. and Litterst F. J.: "Disorder and Diffusional Motion of Cations in Magnetite, Fe3-įO4, Studied by Mössbauer Spectroscopy at High Temperatures", Journal of Physics and Chemistry of Solids, 1993, 54, (8), pp. 923-935. 135. Litterst F.: "Examples for Electronic and Structural Dynamics from Mössbauer Spectroscopy", Hyperfine Interactions, 1986, 27, (1-4), pp. 123-134.

85

Paper I

Estimation of Sintering Kinetics of Oxidized Magnetite Pellet Using Optical Dilatometer T.K. SANDEEP KUMAR, NEELAKANTAN NURNI VISWANATHAN, HESHAM M. AHMED, CHARLOTTE ANDERSSON, and BO BJO¨RKMAN The quality of magnetite pellet is primarily determined by the physico-chemical changes the pellet undergoes as it makes excursion through the gaseous and thermal environment in the induration furnace. Among these physico-chemical processes, the oxidation of magnetite phase and the sintering of oxidized magnetite (hematite) and magnetite (non-oxidized) phases are vital. Rates of these processes not only depend on the thermal and gaseous environment the pellet gets exposed in the induration reactor but also interdependent on each other. Therefore, a systematic study should involve understanding these processes in isolation to the extent possible and quantify them seeking the physics. With this motivation, the present paper focusses on investigating the sintering kinetics of oxidized magnetite pellet. For the current investigation, sintering experiments were carried out on pellets containing more than 95 pct magnetite concentrate from LKAB’s mine, dried and oxidized to completion at sufficiently low temperature to avoid sintering. The sintering behavior of this oxidized pellet is quantified through shrinkage captured by Optical Dilatometer. The extent of sintering characterized by sintering ratio found to follow a power law with time i.e., Ktn. The rate constant K for sintering was determined for different temperatures from isothermal experiments. The rate constant, K, varies  Q ; and the activation energy (Q) and reaction rate with temperature as ln TKð1=nÞ ¼ ln K0  RT constant (K¢) are estimated. Further, the sintering kinetic equation was also extended to a nonisothermal environment and validated using laboratory experiments. DOI: 10.1007/s11663-014-0273-y  The Minerals, Metals & Materials Society and ASM International 2014

I.

INTRODUCTION

IRON ores becoming leaner necessitate beneficiation at finer scales. Globally, this has led to growth in pelletization as an agglomeration process. In addition, pelletization of steel plant solid wastes enables them to be used in various iron and steel processes. Pelletization has been widely practiced for magnetite ores, as it offers an added advantage in terms of energy generated from the oxidation of magnetite to hematite. In Sweden, pelletization is the most important agglomeration process used industrially. Swedish steel industries have T.K. SANDEEP KUMAR, Doctorate Student, and BO BJO¨RKMAN, Chaired Professor, are with the Department of Civil, Environmental and Natural Resources (SBN), Lulea University of Technology (LTU), 97187 Lulea˚, Sweden. Contact e-mail: kamesh. [email protected] NEELAKANTAN NURNI VISWANATHAN, Visiting Professor is with the Department of Civil, Environmental and Natural Resources (SBN), Lulea University of Technology (LTU), and also Professor with the Department of Metallurgical Engineering & Materials Science, Indian Institute of Technology Bombay (IITB), Mumbai 400076, India. HESHAM M. AHMED, Assistant Professor, is with the Department of Civil, Environmental and Natural Resources (SBN), Lulea University of Technology (LTU), and also with the Central Metallurgical Research & Development Institute (CMRDI), Cairo, Egypt. CHARLOTTE ANDERSSON, Specialist/Senior Researcher, is with the Oxidation Metallurgy, Research & Development, Luossavaara-Kiirunavaara AB (LKAB), Malmberget, 98381 Ga¨llivare, Sweden. Manuscript submitted August 18, 2014. Article published online December 24, 2014. METALLURGICAL AND MATERIALS TRANSACTIONS B

stopped using the other widely used agglomeration process—sintering altogether and have pioneered in operating blast furnaces with cent percent pellets. Magnetite ore grounded and beneficiated to an appropriate grade and size distribution is mixed with binder and agglomerated into spherical balls in a disk or drum pelletizer. The green pellets are strengthened through heat hardening process known as induration for subsequent use in iron making units such as blast furnace and directreduced iron processes. Induration is carried out in a straight grate furnace or in straight grate furnace followed by a rotary Kiln furnace. In straight grate, green pellets form a packed bed that is continuously moved through a long furnace. The hot gases are allowed to flow upward as well as downward through the packed bed for efficient heat transfer. In case of rotary Kiln process, partially processed pellets from a straight grate furnace are processed further in a rotating drum furnace. Pellet quality primarily measured in terms of strength and reducibility depends on the excursion the pellet makes through the thermal and gaseous environment in the induration furnace. During induration, magnetite pellets undergo primarily drying, oxidation and sintering. Among these processes, drying almost gets completed at about 373 K (100 C). Some of the chemically bonded water may not get released until the pellet temperature reaches 523 K (250 C).[1] At higher temperatures, primarily oxidation of magnetite and strengthening (sintering) occur. Ideally for a good pellet quality, the oxidation of the pellet should be followed by sintering of the oxidized VOLUME 46B, APRIL 2015—635

magnetite grains.[1,2] The oxidation determined by the intrinsic oxidation kinetics and diffusion of gases through the pores of the pellet proceeds from the pellet surface to the interior. This oxidation process is an exothermic process that generates heat within the pellet and can result in non-uniform temperature within the pellet. Simultaneously, the sintering of the oxidized magnetite (hematite) grains as well as the non-oxidized magnetite grains is initiated depending on the temperature and the sintering kinetics. It has been found that sintering of magnetite grains starts at much lower temperatures than that of hematite grains.[1,3,4] Sintering, leading to a reduction of pore volume, adversely affects the oxidation phenomena. The interaction between oxidation, sintering, and the associated heat transfer can result in so-called duplex structure with a non-oxidized magnetite core with an oxidized shell.[1,4–6] This interrelationship between various phenomena depends on the rate of these individual processes which in turn is determined by the processing conditions and the initial raw material. In order to predict and quantify these physico-chemical phenomena during induration, a project has been initiated jointly by Lulea University of Technology (LTU) and Luossavaara-Kiirunavaara AB (LKAB). At present, models at the reactor scale exist for predicting the overall performance of the induration furnace. However, their ability to predict the quality of the pellets and its distribution is very limited.[7–12] For realizing these objectives, it is important that the physico-chemical phenomena are quantitatively described at the pellet scale and subsequently can to be used as an input to the models at the reactor scale. The methodology adopted is to study each of the aforementioned phenomena in isolation to the extent possible, quantify them and further integrate them through modeling techniques to understand the interaction among these phenomena. Such model can provide (1) building blocks for reactor scale models to predict pellet quality as a function of process parameters and (2) correlate the raw material quality in terms of content, additions, size distribution, etc., to the induration process and pellet quality. These can help in designing process parameters for new grades of raw materials as well as improving existing processes. Sintering is one of the important phenomena during induration processes. The objective of this paper is to quantify the sintering of oxidized magnetite (hematite) pellet using optical dilatometer by deducing the kinetic parameters. Subsequently, the deduced kinetic parameters are validated with further experiments. Sintering kinetics of non-oxidized magnetite pellet will be presented in a future communication. Powder metallurgists have been investigating sintering kinetics of the powder compacts of metallic and ceramic powders for many years.[13–21] The extent of sintering of the powder compact is monitored through shrinkage measured using contact dilatometer. Numerous experiments have been reported on sintering studies on ceramic materials such as alumina, barium titanate, zirconia, uranium oxides, thorium oxides, etc., as well as metallic materials such as Ti, Fe, Zr, stainless steel, etc.[1,14,22–31] Traditionally, dilatometric measurements have been performed on briquettes having an alumina 636—VOLUME 46B, APRIL 2015

push-rod to quantify shrinkage. The push-rod measures the linear displacement and cannot account for 2-D or 3-D spatial changes. A recent development by Expert System Solutions (ESS), Italy—Optical Dilatometer has facilitated to measure the dimensional variations solely on their optical images without having any contact with the sample. It works on principle of monitoring the periphery of the sample captured by two digital cameras equipped with high magnification and long working distance optical systems. Karamanov et al.[32] compared the sintering kinetics estimated with the help of traditional push-rod dilatometer to that using the optical dilatometer. They found significant difference between these two measurements. They attributed the difference to the additional force exerted by the alumina push-rod over the sample contributing to its shrinkage. In this paper, a method to capture shrinkage of spherical pellets of approximately 10-mm diameter using Optical Dilatometer is established with an in situ measurement of area variation during sintering. This helps in assessing the shrinkage in 2-D. Any inhomogeneity in the green pellet can potentially get reflected in the shrinkage measurements. Sintering is a complex process involving transport of species through various paths, such as bulk diffusion, surface diffusion, transport through the gaseous phase, etc., driven primarily to reduce the total surface energy.[33] Most of the studies in ceramic and metallic powder processing were focused in finding the predominant mechanisms operating at different extents of sintering and sintering temperatures.[1,6,16,21,25,30,31,34,35] Sintering studies on iron ore to deduce their kinetics have been relatively less. It can be noted that unlike conventional powder processing where the objective is to achieve close to 100 percent densification, in iron ore sintering, the objective is only to achieve sintering enough to provide sufficient strength during handling and subsequent operation in blast furnace and other reactors. At the same stage, sufficient porosity within in the pellet is necessary to achieve reduction in least time. Thus, sintering mechanisms responsible during the initial stages are only important for iron ore sintering. Wynnyckyj and Fahidy[36] made an early in-road by studying the sintering kinetics of pure hematite reagent powder and commercial magnetite concentrate in the form of briquettes under isothermal conditions. They proposed a power law relation with time for isothermal sintering based on similar studies on sintering of metallic minerals and the same has been used in the present study. This model is described in detail at a later section. Interestingly, authors could not find any further quantitative studies either on magnetite or hematite sintering, though dilatometer measurements have been used to study sintering behavior.[1]

II.

EXPERIMENTAL DETAILS

A. Raw Material The raw material chosen for the study was a concentrate from LKAB’s Malmberget mine. It was preserved METALLURGICAL AND MATERIALS TRANSACTIONS B

Table I.

Chemical Composition of Raw Materials Concentrate (wt pct)

Bentonite (wt pct)

71.06 0.3994 0.1198 0.2596 0.078

3.63 51.64 6.43 0.68 3.30

Fe SiO2 CaO MgO RNa2O + K2O

pellet at different angles and orientations were taken and analyzed using software ImageJ. The measurement of the bulk volume using this technique was calibrated using standard-sized dull luster spheres. The mean porosity of the pellet can be obtained from their bulk and true densities. D. Oxidation

carefully to avoid atmospheric moisture absorption. The concentrate contains Fe3O4 >95 pct with Al2O3 and SiO2 95pct) magnetite, are used for the sintering studies. Optical Dilatometer is used to capture the sintering behavior of the magnetite pellet and determine their isothermal kinetics by deducing the three parameters, namely – activation energy (Q), pre-exponential factor (K’) and time exponent (n) with the help of power law and Arrhenius equation. It is interesting to find that the time exponent (n) is decreasing with the increase in sintering temperature. It is also interesting to note that the activation energy for sintering of magnetite pellet shows no single value. From the present investigation, two activation energies – 477 kJ/mole (1173 – 1373 K) and 148 kJ/mole (1373 – 1623 K) were deduced for sintering of magnetite, suggesting two different mechanisms operating at lower and other at higher temperatures. The estimated kinetic parameters were used to predict the non-isothermal sintering behavior of magnetite using the sintering kinetic model. Predicted results were validated using experimental data.

Keywords: Induration, Sintering Kinetics, Magnetite pellet, Optical Dilatometer, Activation Energy, Shrinkage, Diffusion mechanism, Non-stoichiometry

1

Introduction In the prospects of ores getting leaner and finer, pelletization is increasing globally as the most widely practiced agglomeration technique for the ore fines. The major ores being the iron bearing minerals – hematite and magnetite whose fines are being pelletized at large scale across the world. Magnetite ore fines, in particular have the advantage because of its exothermic oxidation reaction making it more suitable and efficient for induration subsequent to pelletization. In Sweden, where major source of iron ore is magnetite has developed the expertise in pelletization over the years, as it is the largely practiced agglomeration technique. The blast furnaces in Sweden operate primarily with pellets except for a small proportion of briquettes made out of steel plant solid waste materials. In pelletization process, magnetite ore is ground and mixed with bentonite as binder then balled in pelletizer to produce green pellets. These green pellets are then heat hardened during induration. Thus, induration is a vital process in pelletization to attain desired pellet quality parameters of strength and reducibility. Induration is carried out either in a straight grate furnace or in a grate kiln furnace. The hot gases are allowed to flow upward as well as downward through the packed bed and in counter-current direction across the furnace for efficient heat transfer. During induration, magnetite pellet undergoes drying, oxidation and sintering1. During induration of magnetite pellets, both oxidation and sintering occur. Ideally, it is desirable to have sintering after the complete oxidation of magnetite pellet1,2. In reality, sintering and oxidation may proceed simultaneously. Magnetite sintering will start at a lower temperature than hematite sintering, and remaining magnetite in the core can therefore shrink away from the hematite shell, forming a duplex structure with non-oxidized magnetite core and oxidized shell1,37 . This is further complicated by the heat generated within the pellet from the exothermic oxidation of magnetite, causing sintering of magnetite to occur earlier than otherwise. Therefore, a systematic approach would be to investigate sintering, oxidation and the heat transfer phenomena independently; and the mutual interference of these phenomena can be understood using an appropriate mathematical model at the pellet scale. Further, such a model at the pellet scale can be integrated with models at the reactor scale8-13 to predict pellet quality. A collaborative project by Lulea University of Technology (LTU) and Luossavaara – Kiirunavaara AB (LKAB) has been taken up for this purpose. The project has been formulated in parts to study for each of these processes in isolation. The sintering kinetics of oxidized magnetite (hematite) has already been determined in isolation14. The characteristics of magnetite are expected to differ from that of hematite, and hence the focus of the current paper is to independently determine the sintering kinetics of magnetite pellets. The outcome of magnetite sintering along with that of oxidized magnetite (hematite) and their microstructural analysis could be a step towards understanding the overall induration process at pellet scale. This will provide an aid to consider the raw material variability in future (backward integration) and also on the process control parameters to achieve the desired pellet quality (forward integration), and in turn improve the 2

production efficiency. Therefore, it becomes essential to quantify magnetite sintering kinetics in continuation to that of oxidized magnetite sintering kinetics. Kinetics of process can be evaluated by quantifying them with the three parameters namely – activation energy (Q), pre-exponential factor (K’) and time exponent (n). Sintering kinetics has been studied from the early 1940’s quite extensively in the field of powder metallurgy15-25 whereas those of iron ore26,27 have been relatively less. It can be noted that unlike conventional powder processing where the objective is to achieve close to complete densification, in iron ore sintering, the objective is to achieve sintering enough to provide sufficient strength during handling and subsequent operation in iron making furnaces. At the same time, sufficient porosity within the pellet is necessary to achieve reduction in least time. In 1974, Wynnyckyj and Fahidy made an early attempt by studying the sintering kinetics of pure hematite reagent powder and commercial magnetite concentrate in the form of briquettes under isothermal conditions28. They proposed a power law which relates the extent of sintering with time under isothermal conditions and can be quantified by measuring its shrinkage. Similar approach has been adapted by Kumar et al14 recently, wherein shrinkage data of oxidized magnetite, is successfully quantified to estimate the above mentioned three kinetic parameters. Kumar et al14 used Optical Dilatometer as a novel method to measure shrinkage of pellet instead of traditionally used push-rod dilatometer. This new contactless method is also better suited for spherical pellets29. It works on the principle of monitoring the shadow of the pellet sample captured by digital camera equipped with high magnification and long working distance optical systems. It measures the dimensional variations solely on their optical images without having any contact with the sample. The same approach is now extended to estimate the kinetic parameters by quantifying the sintering phenomena of magnetite. Dimensional changes of a single magnetite pellet (approximately 10 mm diameter) kept in a furnace under argon atmosphere are measured using optical dilatometer to study the sintering behavior of the pellet.

Experimental Details The raw material used was the same concentrate as used in the previously studied hematite sintering14. The concentrate was collected from LKAB’s mine in Malmberget, which upon passed by open grinding circuit in ball mills make it suitable for pelletization. The concentrate was targeted to have ILQHQHVVRISDVVLQJWKURXJKȝPVFUHHQVDQGspecific surface area ~9900 cm2/cm3 (Blaine No. = 1930 cm2/gm.) measured by BET Surface Area analyzer. The concentrate having Fe3O4 with Al2O3 and SiO2 < 0.6 %, and 7 % moisture by weight was mixed, with 0.5 % dosage of bentonite as binder, in a laboratory mixer (Eirich R02). The chemical analysis of MPC and bentonite is mentioned in Table I.

3

Table I: Chemical Composition of Raw Materials

Fe SiO2 CaO MgO ɇEĂ2O + K2O

Concentrate (wt. %) 71.06 0.39 0.12 0.26 0.078

Bentonite (wt. %) 3.63 51.64 6.43 0.68 3.30

The mix (7 kg) was then fed to the drum pelletizer (0.8 m diameter) for balling, producing nucleation seeds (3.5 – 5 mm diameter) initially, and then green pellets of desired size fraction (9 – 10 mm diameter) was screened and collected. The desired size here, is smaller than the widely practiced industrial average size (10 – 12 mm) of pellets because of the limitations imposed by optical dilatometer on sample size which have been discussed elsewhere14. The green pellets were dried in an oven at 378 K (105oC) overnight, and stored in a desiccator during the course of the whole project. The pellets were carefully characterized for moisture content, true or skeletal density and bulk density before and after sintering experiments. Moisture content was evaluated by Infrared Moisture Analyzer MA150 (Sartorius AG, Germany) as per the ISO standards. Since the study is focused on single pellet experiments, the surface of the pellet needs to be protected from any contamination to make it suitable for further processing. Therefore, skeletal (true) and envelope (bulk) density of single pellet as a whole were measured by AccuPyc II 1340 (Micromeritics Inc., USA) and a tailor made Light Table Imaging (LTI) method, respectively without disrupting the pellet surface. The mean porosity of the pellet was then obtained from their skeletal and envelope densities. Figure 1 shows a schematic of the optical dilatometer used for the study. More details on characterization methodology and experimental set can be found elsewhere14.

Figure 1: Schematic of Optical Dilatometer (Misura@ HSM – ODHT)

4

Magnetite green pellets in all the experiments were exposed to the thermal profile from room temperature to desired sintering temperature at a heating rate of 30 K/min, held isothermally for 20 to 90 minutes and then allowed to be furnace cooled. The isothermal holding time was chosen so as to have adequate residence time for sintering to occur. The experiments designed for magnetite sintering studies in isolation are mentioned in Table II. The experiments were designed with temperature intervals of 50 K at higher temperatures (above 1323 K (1050oC)) and 25 K at lower temperatures with an effort to capture the small dimensional changes. Table II: Experimental design to study sintering kinetics of magnetite pellets Pellet Sintering Temperature, K Isothermal Time, min

P1

P2

P3

P4

P5

P6

P7

P8

P9

P10

P11

P12

1173

1223

1248

1273

1298

1323

1373

1423

1473

1523

1573

1623

90

60

60

60

60

60

40

40

20

20

20

20

A single magnetite pellet was kept under inert atmosphere to minimize its oxidation. A constant flow of argon (99.995 %) gas at the maximum allowable flow rate of 8.33 x 10-6 m3/sec (0.5 liters per minute) was used to have inert atmosphere across the pellet in the furnace. Higher flow rate of argon is maintained to keep positive pressure inside the furnace all the time to avoid any air infringement into the furnace which otherwise may lead to substantial oxidation. In order to confirm this, sintered magnetite pellets were ground and analyzed with X–Ray Diffraction (XRD) with a copper target and they were also investigated with Light Optical Microscopy (LOM). The shadow images of the pellet were captured continuously by optical dilatometer at an interval of 15 seconds to observe the shrinkage during sintering.

Results XRD and Microstructures The XRD patterns for raw material (magnetite concentrate) and pellet exposed to the highest sintering temperature (1623 K (1350oC)) are shown in Figure 2, where major magnetite peaks (96-900-5814) were found (without those of hematite). The hematite reference (96-901-4881) peak positions are also indicated in the figure. Similar diffractograms with magnetite peaks have been observed for all pellets with sintering temperatures varying from 1173 to 1573 K (900 to 1300oC). The microstructure of a 10 mm diameter fired magnetite pellet is shown in Figure 3 with focus on its rim. It is observed that there are traces of oxidation at the rim of the magnetite pellet. The maximum thickness of the spread of these traces is approximately 50 – 70 μm. Assuming that the oxidation layer is concentric, the degree of oxidation calculated is found to be less than 4.2 %. Since, this is such a small percentage and that too at the surface, it is not expected to interfere or have significant effect on sintering of magnetite pellet. 5

Figure 2: XRD peaks of magnetite concentrate and pellet exposed to highest sintering temperature (1623 K)

Figure 3: Microstructure of a 10 mm diameter exposed to sintering temperature of 1423 K (a) with focus on rim and (b) the overview of whole pellet.

6

Sintering Degree and its Rate The typical plot resulting from the sintering of magnetite from the optical dilatometer is shown in Figure 4, where the percentage change in area (ߜ஺,௢௩௘௥௔௟௟ ) of a pellet is plotted against time for the given thermal profile. ߜ஺,௢௩௘௥௔௟௟ =

஺ି஺బ ஺బ

× 100

(1)

where, ‘A’ refers to projected area of the pellet at any instant of time and ‘A0’ refers to that at the start of the experiment. Experiments have been repeated with same thermal profile to ensure the reproducibility of the results which is quite good as depicted in Figure 4, and only one of them is plotted further for brevity. However, experiments were repeated for all thermal profiles. The reproducibility of these experiments are good considering the fact that pellet to pellet variation in porosity and bulk density is expected. The temperature within the pellet is assumed to be uniform as it is observed from the preliminary confirmatory experiments, and well supported by analytical heat transfer calculations30.

Figure 4: Typical plot obtained from Optical Dilatometer for percentage area change with time

Figure 4 shows that initially the pellet expands thermally, and shrinks later due to sintering at higher temperatures (> 1223 K) and thermal contraction during cooling14,31. Thus, the overall 7

change in size of pellet is due to the combination of thermal expansion/contraction and the sintering phenomena. Isolating the linear thermal expansion/contraction from the overall percentage change in area by plotting it as the function of temperature (see Figure 5), and equating their similar slopes yields the shrinkage of pellet due to sintering alone. The noise in area change measurement at the start of experiment can be attributed to image analysis by software. Since, the percentage change in area is measured from optical dilatometer, volumetric thermal coefficient of expansion (Į) is related to its area thermal coefficient of expansion (ȕ), and is expressed in Eq. [2]14,32. The values of thermal expansion estimated is comparable with that obtained in the literature33,34. ଵ ௗ஺

ߚ = ஺ ௗ் =

ଶ ଷ

ߙ

(2)

Subsequently, the area change due to sintering alone is used for further analysis.

Figure 5: Area change for magnetite pellets during sintering at different temperatures

Sintering is a temperature dependent process, and the degree of sintering increases with increase in temperature for magnetite pellets. This is also evident from the micrographs of fired magnetite pellets shown in Figure 6, where grains come closer to each other at lower temperature (1273 K), 8

form the neck while sintering progresses with increase in temperature (1473 K), and agglomerated at higher temperatures (1573 K).

Figure 6: Optical Microstructures of fired magnetite pellet at 1273 K, 1473 K and 1573 K, at magnification of 200 X

Sintering rate describes the progress of sintering, and is quantified as the degree of sintering (percentage of area change) achieved per unit time during the course of induration. The relation between the sintering rate of magnetite pellet and sintering temperature is shown in Figure 7. It shows that the sintering rate increases in the non-isothermal segment whereas it decreases with time in the isothermal segment. This is because the driving force for sintering, the surface area of grains in the pellet (pore surface area), decreases with the extent of sintering.

9

Figure 7: Sintering Rate of a magnetite pellet for a given thermal profile

Sintering Ratio (Ȗ) Sintering ratio for the pellet is defined as the ratio of the sintering accomplished to the sintering yet to be accomplished as expressed in Eq. [3]. In 1974, Wynnyckyj and Fahidy28 have suggested, that this sintering ratio can be a measure to quantify sintering of pellets by capturing their shrinkage. In the preceding study, Kumar et al14 has successfully quantified the sintering of oxidized magnetite (hematite) pellets with a similar approach. Thus, in this study the same approach has now been extended to quantify the sintering of magnetite pellets. ߛ=

ௌ௜௡௧௘௥௜௡௚ ௔௖௖௢௠௣௟௜௦௛௘ௗ ௌ௜௡௧௘௥௜௡௚ ௬௘௧ ௧௢ ௕௘ ௔௖௖௢௠௣௟௜௦௛௘ௗ

=

௏బ ି௏ ௏ି ௏೟ೝೠ೐

(3)

where, Vo is the initial volume of the pellet, V is the volume of the pellet at any instant during sintering and Vtrue is the volume if the pellet would have undergone complete sintering with no pores remaining. Since only the change in area of pellet during sintering is measured it needs to 10

be related to the sintering ratio of pellets. Bulk density and true density of pellets measured by LTI and Helium pycnometer respectively, are used and rearranged to obtain an expression for sintering ratio in terms of area change due to sintering. ߛ=

ିఋಲ,ೞ೔೙೟೐ೝ೔೙೒ ఋಲ,೟ೝೠ೐ ା ఋಲ,ೞ೔೙೟೐ೝ೔೙೒

(4)

where, įA,sintering is area change at any instant during sintering and į(A,true) is area change if the pellet would have sintered with zero porosity. This sintering ratio (extent of sintering) is shown in Figure 8. It is plotted against time for pellets exposed to different isothermal temperatures and further used for determination of kinetics.

Figure 8: Sintering ratio for magnetite pellets during sintering at different temperatures

Sintering Kinetics Sintering of magnetite pellets are quantified by determining the aforementioned three basic kinetic parameters, namely – time exponent, activation energy and rate constant. The isothermal sintering time (t) of the reaction (assuming two-sphere particle model) is related to the extent of sintering (Ȗ) with a time exponent (n), as expressed in Eq. [5]14,28. The time exponent of the reaction determines the reaction mechanisms which are dominant during sintering of pellet at various stages as it progress. 11

ߛ = ‫ ݐܭ‬௡

(5)

where, K is the rate constant. Further, this reaction constant term can be related to temperature with an Arrhenius form of expression (Eq. [6]). ݈݊൫ܶ‫( ܭ‬ଵ/௡) ൯ = ݈݊ ‫ܭ‬Ԣ െ

ொ ோ்

(6)

where, Q is the activation energy and ‫ܭ‬Ԣ is pre-exponential factor. The isothermal segment of sintering ratio curves (Figure 8) is used to estimate the sintering kinetic parameters. The sintering ratios of the pellet at the beginning and at any instant of isothermal sintering have been compared according to Eq. [5] giving the following expression, ఊ

݈݊ ቀఊ‫כ‬ቁ = ݊ ݈݊

(௧ ‫ כ‬ା ௧೘ ) ௧‫כ‬

(7)

where, Ȗ*as the sintering ratio at the start of the isothermal section, t* corresponds to a time if the pellet had attained a sintering ratio of Ȗ* from the start under isothermal condition, and tm is the measured time from the beginning to any instant of isothermal section corresponding to sintering ratio Ȗ, as shown in Figure 9 . The parameters n and t* were estimated using least square fit for experiments at different sintering temperatures for magnetite pellets, and has been plotted as shown in Figure 10.

Figure 9: Depiction of parameters of power law from sintering ratio plots with respect to time

12

Figure 10: Time exponent ‘n’ estimated from the slope of curves, for magnetite pellets sintered at different temperatures

Please note that linearity of points (R2 > 0.975) at different temperatures in Figure 10 confirms the validity of Eq. [5] in describing the sintering kinetics. However, it can be observed that at lower temperatures, there are significant fluctuations about the straight line whereas at higher temperatures the fluctuations are much less. This might corresponds to the fact that at lower temperatures the shape of the pellet, as well as variation in shape and packing of grains in the pellet, may be influencing the initial sintering phenomena. These fluctuations might also be attributed to the fact that the shrinkage values are so small at low temperatures that there might be uncertainties in the measurements. Ideally, all the points in Figure 10 would have fallen on one line, for a single value of n, as suggested by Eq. [6]. However, the present results show a variation in the slope, ‘n’, for different sintering temperatures, as tabulated in Table III. In order to explore further, the exponent ‘n’ was plotted as a function of temperature as shown in Figure 11. The value of ‘n’ varies from 0.45 to 0.15 in the range of temperatures from at 1173K

13

to 1673K. A linear fit for predicting n as a function of temperature (with two data points considered as outliers) is given by ݊ = െ4 ܺ 10ିସ ܶ + 0.82

(8)

Figure 11: Variation of time exponent ‘n’ for pellets exposed to different sintering temperatures Further, from the variation of sintering ratio, Ȗ, with time, t, in the isothermal section the values of sintering rate constant, K, in Eq. [5] for different temperatures is estimated knowing respective Ȗ*, t* and n. In order to estimate the activation energy, Q and the pre-exponential factor, K’, ln (KT1/n) is plotted as a function 1/T as shown in Figure 12. Two separate lines can be inferred from the plot, as illustrated in Figure 12. Corresponding to these lines, the ranges of temperatures, activation energies Q1 and Q2, and the pre-exponential factors K1’ and K2’ are shown in Table III.

14

Table III: Kinetic parameters for sintering of magnetite pellets Sintering Temp. o



C

Activation Energy

Pre-exponential Factor

kJ/mole

K/sec

1173

0.35

1223

0.31

1248

0.20

ܳଶ = 477 kJ/mole

‫ܭ‬Ԣଶ = 3.66 X 1014

1273

0.44

(1173 – 1373 K)

(1173 – 1373 K)

1298

0.25

1323

0.35

1373

0.33

1423

0.30

1473

0.25

ܳଵ = 148 kJ/mole

‫ܭ‬Ԣଵ = 3.51 X 102

1523

0.22

(1373 – 1623 K)

(1373 – 1623 K)

1573

0.20

1623

0.15

15

Figure 12: Activation Energies and Pre-exponential factors for the sintering of magnetite pellets

The activation energies for magnetite pellet sintering at lower temperatures is higher than at higher temperatures suggesting different sintering mechanisms at lower and higher temperatures.

Sintering Prediction In order to simulate the actual industrial process where sintering occurs under non-isothermal conditions, the estimated kinetic parameters determined for isothermal magnetite sintering by Eq. [6 – 8] are further extended to non-isothermal sintering by similar approach which had been adapted for the sintering of oxidized magnetite14. The care has been taken to incorporate the variation of ‘n’ with temperature and also the two activation energies while marching in time for VXIILFLHQWO\ VPDOO VWHS RI ¨t (discretized time – temperature plot) to generate the profile for sintering ratios as expressed in Eq. [9].

16

ߛ௧ାο௧ = ‫ܶ(ܭ‬௧ା௱௧ ) ቆ൬

ఊ೟

௄೅೟శ೩೟

ଵ/௡





+ ο‫ ݐ‬ቇ

(9)

where, ߛ௧ and ߛ௧ା௱௧ denote the sintering ratios at time t and W¨W, respectively and corresponding ఊ

temperatures be denoted by ܶ௧ and ܶ௧ା௱௧ ., and ቀ௄(் ೟

ଵ/௡



೟శ೩೟ )

is the time that would have taken to

achieve sintering ratio of ߛ௧ isothermally at temperatureܶ௧ା௱௧ . For the purpose of validating the sintering profiles predicted by Eq. [9], experiments have also been performed with pellets exposed to three different heating rates up to a temperature of 1573 K in optical dilatometer, and are compared in Figure 13 (a). Sintering profiles have also been predicted for pellets exposed to different temperatures at a heating rate of 30 K/min, and are compared with those obtained experimentally. Although all of them have similar profiles, for the sake of clarity, a few of them at low, intermediate and higher temperatures are shown in Figure 13 (b).

Figure 13: Predicted and estimated sintering ratios for pellets exposed to (a) different heating rates up to 1573 K, and (b) different temperatures at same heating rate of 30 K/min

Figure 13 depicts the sintering state of a pellet under variable heating rates that it experiences during a complete induration cycle. The predicted sintering states for the pellets exposed to different heating rates are in very good agreement with the experimental ones, whereas those exposed to different temperatures were found to deviate somewhat at lower temperatures but at higher temperatures they appears to be in good agreement. This may be attributed to the 17

aforementioned reasons for the fluctuations in the Figure 10 at lower temperatures. This establishes that by using the sintering kinetic parameters, namely, n, K’ and Q, it can be possible to predict the extent of sintering for any non-isothermal profile using Eq. [9].

Discussion The kinetic parameters of sintering, namely, time exponent, n and the activation energy, Q can be correlated to various sintering mechanisms. In the present study, the time exponent for magnetite sintering is found to be decreasing with increasing temperature, and two activation energies were deduced over the entire temperature range of study; suggesting two distinct mechanisms. Numerous investigations on solid state sintering15,33-42 have proposed n increasing with increase in temperature, and attributed the variation of n to various sintering mechanisms – surface diffusion (n =1/7), grain boundary diffusion (n =1/6), volume diffusion (n =1/5) and viscous flow (n =1/2). However, the current observations are not in correspondence with these propositions. Interestingly, the decreasing trend of n with increasing temperature for magnetite sintering was also observed by Wynnyckyj and Fahidy28. As far as activation energy is concerned, Xuebin et al42 also found two activation energies for the sintering of hydroxyapatite over the temperature ranges, and are consistent with the current observations. In this context, further insights in to the current observations are being sought. Magnetite is a non-stoichiometric ionic solid as shown in the FeO – Fe2O3 binary phase diagram in Figure 14 and has significant excess oxygen ions at higher temperature leading to defects which aid diffusion in these solids. It should be noted that for a fixed partial pressure of oxygen, the amount of oxygen dissolved in magnetite decreases with increase in temperature. Sintering mechanisms are primary diffusion based. Thus, a closer look at the diffusion in magnetite may throw some light in to their sintering kinetics estimated in the present study.

18

Figure 14: The phase diagram for the system FeO – Fe2O3 showing the positions of the oxygen (atm) isobars32 Many researchers have studied the cation tracer diffusion and jump frequency (Mossbauer spectroscopy) of magnetite in solid state43-54 with partial pressure of oxygen (‫݌‬ைమ ) and temperature. They found that, diffusion of iron in magnetite lattice occurs by two different mechanisms – interstitial mechanism (at lower ‫݌‬ைమ and higher temperature) and vacancy mechanism (at higher ‫݌‬ைమ and lower temperature). It is also interesting to note that the activation energy for interstitial mechanism is positive whereas that for vacancy mechanism is negative. Thus, for a given ‫݌‬ைమ as temperature increases, the diffusion mechanism shifts from the vacancy to the interstitial mechanism. Such shift in mechanisms suggests that the activation energy increases when temperature increases from lower to higher, which does not imply with the current findings. Furthermore, for ‫݌‬ைమ > 10-6 and temperature < 1573 K, vacancy mechanism is predominant which is plausible to operate in the range of temperatures being studied in the current investigation. However, the negative activation energy corresponding to this mechanism has also not been observed.

19

Interestingly, Xuebin et al42, observed a trend in activation energy for the sintering kinetics of hydroxyapatite similar to that estimated in the current study for magnetite pellets. They correlated the kinetic parameters, n and Q to the morphology, which infers about the probable mechanisms responsible for sintering. They speculate that the high activation energy at lower temperatures is because of the initial neck formation through reaction and surface diffusion. Contemplating in the similar context for the findings of sintering of magnetite pellet, it might be a possibility that the bentonite/gangue minerals initiate neck formation through reaction and solid state diffusion at the initial stages. Later at higher temperatures, there might be formation of liquid phase bridges55 decreasing the activation energy for sintering. Morphological studies using the automated image analysis of the microstructures of pellets can impart more insights into the sintering mechanisms at different temperatures. These studies are in progress and will be presented in future communication.

Conclusions The sintering kinetics of magnetite pellet under inert atmosphere has been successfully estimated, in succession to oxidized magnetite pellet (hematite). Optical dilatometer is used to capture the shrinkage, and hence sintering of the pellet. Sintering is quantified using isothermal shrinkage data by estimating three kinetic parameters, namely, activation energy (Q) and pre-exponential factor (K’) and a time exponent (n). The sintering of magnetite pellet shows that the value of ‘n’ varies from 0.45 to 0.15 with the increase in temperature. The two activation energy values estimated for the sintering of magnetite pellets are 477 and 148 kJ/mole, respectively, at lower temperatures (1173 – 1373 K) and higher temperatures (1373 – 1623 K) suggesting the possibility of two distinct mechanisms. The extent of sintering have been predicted under nonisothermal conditions as well by incorporating the variations in above mentioned kinetic parameters (n, K’ and Q) with temperatures, and are validated using experimental data.

Acknowledgement Authors thank the Hjalmar Lundbohm Research Centre (HLRC) for their financial support. We also thank Ola Eriksson, Daniel Marjavaara, Gustaf Magnusson, Magnus Stafstedt, and Anders Dahlin of LKAB for their technical support. We also thank Prof. S. Seetharaman, an Emeritus professor of Royal Institute of Technology (KTH), Stockholm for valuable discussions.

20

List of Symbols:

ߜ஺,௢௩௘௥௔௟௟ ߜ஺,௦௜௡௧௘௥௜௡௚ ߜ஺ ߜ஺,௧௥௨௘ ߙ ȕ V0 V ܸ௧௥௨௘ T0 T ߛ Ȗ* ߩ

ߩ௧௥௨௘ ߩ଴ ‫ݐ‬ t* tm n K1 ’ K2 ’ Q1 Q2 R Ȗt ȖW¨W Tt TW¨W K(TW¨W )

Overall percentage area change at any instant during induration Percentage area change due to sintering at any instant during induration Percentage area change at any instant Percentage area change when pellet has no pores Volumetric thermal coefficient of expansion Area thermal coefficient of expansion Initial volume of a material Volume of material at any temperature Volume of the pellet if it would have undergone complete sintering with no pores Initial temperature (t = 0) Temperature at any instant Sintering Ratio of the pellet at any instant in the isothermal section Sintering ratio at the start of the isothermal section Bulk density of the pellet at any instant True density of the pellet Initial bulk density of pellet Time for sintering reaction Time corresponds if the pellet had attained a sintering ratio of Ȗ* from the start under isothermal condition Measured time in isothermal section Time exponent Pre exponential factor at high temperatures Pre exponential factor at low temperatures Activation energy at high temperatures Activation energy at low temperatures Universal gas constant Sintering ratio at time t Sintering ratio at time W¨W Temperature at time t Temperature at time W¨W Rate constant at W¨W

21

References [1] S. P. E. Forsmo, S. -. Forsmo, P. -. Samskog, and B. M. T. Björkman: Powder Technol, 2008, vol. 183, pp. 247-59. [2] S. R. Cooke and W. F. Stowasser Jr: Trans. AIME, 1952, vol. 193, pp. 1223-30. [3] V. Niiniskorpi: Ironmaking Conference Proceedings, 2001, pp. 767-80. [4] M. Tang, H. J. Cho, and P. C. Pistorius: Metall. Mater. Trans. B, 2014, vol. 45 B, pp. 130414. [5] J. R. Wynnyckyj and W. A. McCurdy: Metall Trans., 1974, vol. 5, pp. 2207-15. [6] P. Semberg, A. Rutqvist, C. Andersson, and B. Björkman: Ironmaking and Steelmaking, 2011, vol. 38, pp. 321-28. [7] P. Semberg, C. Andersson, and B. Björkman: ISIJ Int., 2013, vol. 53, pp. 391-98. [8] J. Thurlby, R. Batterham, and R. Turner: Int. J. Miner. Process., 1979, vol. 6, pp. 43-64. [9] J. Thurlby: Metall. Trans. B, 1988, vol. 19B, pp. 103-12. [10] R. A. Dave: Can. J. Chem. Eng., 1996, vol. 74, pp. 1004-09. [11] M. Cross and P. Blot: Metall. and Mater. Trans. B, 1999, vol. 30B, pp. 803-13. [12] S. Sadrnezhaad, A. Ferdowsi, and H. Payab: Comput. Mater. Sci., 2008, vol. 44, pp. 296-02. [13] M. Barati: Int. J. Miner. Process., 2008, vol. 89, pp. 30-39. [14] T. K. Sandeep Kumar, N. N. Viswanathan, H. M. Ahmed, C. Andersson, and B. Björkman: Metall. and Mat. Trans. B, 2015, vol. 46B, pp. 635-43. [15] G. C. Kuczynski: Journal of Applied Physics, 1949, vol. 20, pp. 1160-63.

22

[16] D. L. Johnson and I. B. Cutler: J. Am. Ceram. Soc., 1963, vol. 46, pp. 545-50. [17] W. S. Young and I. B. Cutler: J. Am. Ceram. Soc., 1970, vol. 53, pp. 659-63. [18] J. D. Hansen, R. P. Rusin, M. Teng, and D. L. Johnson: J. Am. Ceram. Soc., 1992, vol. 75, pp. 1129-35. >@ 0 9 1LNROLü 9 3 3DYORYLü 9 % 3DYORYLü 1 /DEXV DQG % 6WRMDQRYLü 0DWHU 6FL Forum, 2005, vol. 494, pp. 417-22. [20] P. Dehaudt, L. Bourgeois, and H. Chevrel: J. Nucl. Mater., 2001, vol. 299, pp. 250-59. [21] W. Zeng, L. Gao, L. Gui, and J. Guo: Ceram. Int., 1999, vol. 25, pp. 723-26. [22] Z. Holková, L. Pach, V. Kovár, and S. Svetík: Ceramics-Silik., 2003, vol. 47, pp. 13-19. [23] T. Kutty, K. Khan, P. Hegde, A. Sengupta, S. Majumdar, and H. Kamath: Sci. Sinter., 2003, vol. 35, pp. 125-32. [24] D. Lahiri, S. Rao, G. Rao, and R. Srivastava: J. Nucl. Mater., 2006, vol. 357, pp. 88-96. [25] M. Mazaheri, F. Golestani-Fard, and A. Simchi: 10th International Conference and Exhibition of the European Ceramic Society, Berlin, 2007. [26] Z. Y. Liu, N. H. Loh, K. A. Khor, and S. B. Tor: Scr. Mater., 2001, vol. 44, pp. 1131-37. [27] R. L. Coble: J. Am. Ceram. Soc., 1958, vol. 41, pp. 55-62. [28] J. R. Wynnyckyj and T. Z. Fahidy: Metall. Trans., 1974, vol. 5, pp. 991-1000. [29] A. Karamanov, B. Dzhantov, M. Paganelli, and D. Sighinolfi: Thermochim. Acta, 2013, vol. 553, pp. 1-7. [30] V. N. Nurni and B. N. Ballal: Treatise in Process Metallurgy, vol. 1, Elsevier, Oxford, UK, 2014. 23

[31] S. R. B. Cooke and T. E. Ban: Transactions AIME _ Minning Engineering, 1952, vol. 6, pp. 1053-58. [32] D. R. Gaskell: Introduction to the Thermodynamics of Materials, 2nd ed., CRC Press, New York, USA, 2008, pp. 6. [33] B. J. Skinner: Thermal Expansion - Geological Society of America Memoirs, 1966, vol. 97, pp. 75-96. [34] D. Taylor: Transactions and journal of the British Ceramic Society, 1985, vol. 84, pp. 121127. [35] G. C. Kuczynski: AIME Trans., 1949, vol. 185, pp. 169-78. [36] R. German and Z. Munir: Metallurgical and Materials Transactions A, 1975, vol. 6, pp. 2229-34. [37] R. M. German: Sintering Theory and Practice, Wiley Interscience, New York, 1996. [38] R. German and Z. Munir: J. Am. Ceram. Soc., 1976, vol. 59, pp. 379-83. [39] R. German and Z. Munir: J. Mater. Sci., 1975, vol. 10, pp. 1719-24. [40] F. Nichols: Acta Metall., 1968, vol. 16, pp. 103-13. [41] W. Beere: Journal of Materials Science, 1973, vol. 8, pp. 1717-24. [42] W. Kingery: J. Appl. Phys., 1960, vol. 31, pp. 833-38. [43] J. Dedrick and A. Gerds: J. Appl. Phys., 1949, vol. 20, pp. 1042-44. [44] Z. Xuebin, D. Yunfei, W. Songlin, X. Jie, F. Yi: Ceramics – Silik., 2010, vol. 54, pp. 24852.

24

[45] R. Dieckmann and H. Schmalzried: Berichte Der Bunsengesellschaft Für Physikalische Chemie, 1977, vol. 81, pp. 344-47. [46] R. Dieckmann and H. Schmalzried: Berichte Der Bunsengesellschaft Für Physikalische Chemie, 1977, vol. 81, pp. 414-19. [47] G. Lewis, C. R. A. Catlow, and A. Cormack: Journal of Physics and Chemistry of Solids, 1985, vol. 46, pp. 1227-33. [48] S. Hallström, L. Höglund, and J. Ågren: Acta Materialia, 2011, vol. 59, pp. 53-60. [49] H. Schmalzried: Reactivity of Solids, 1988, vol. 5, pp. 269-78. [50] N. L. Peterson, W. Chen, and D. Wolf: Journal of Physics and Chemistry of Solids, 1980, vol. 41, pp. 709-19. [51] J. A. Van Orman and K. L. Crispin: Reviews in Mineralogy and Geochemistry, 2010, vol. 72, pp. 757-825. [52] D. Levy, R. Giustetto, and A. Hoser: Physics and Chemistry of Minerals, 2012, vol. 39, pp. 169-76. [53] K. D. Becker, V. V. Wurmb, and F. J. Litterst: Journal of Physics and Chemistry of Solids, 1993, vol. 54, pp. 923-35. [54] K. D. Becker, V. V. Wurmb, and F. J. Litterst: Hyperfine Interactions, 1990, vol. 56, pp. 1431-35. [55] K. Becker: Solid State Ionics, 1990, vol. 39, pp. 27-35. [56] F. Litterst: Hyperfine Interactions, 1986, vol. 27, pp. 123-34. [57] L. Bentell: Scandinivian Journal of Metallurgy, 1981, vol. 10, pp. 205-09. 25