"Applied Science, Faculty of"@en . "Materials Engineering, Department of"@en . "DSpace"@en . "UBCV"@en . "Gandhi, Chetak"@en . "2009-06-17T19:40:58Z"@en . "1999"@en . "Master of Applied Science - MASc"@en . "University of British Columbia"@en . "As customer demands become more stringent for canbody stock, it\r\nbecomes essential to understand the complex interaction between the processing\r\nconditions and resulting product properties. This research focused on investigating the\r\ninfluence of homogenization process parameters (heat-up rate, soak temperature and\r\ntime) on the microstructural evolution of an AA3104 aluminum alloy used for canbody\r\nstock. Experiments were conducted on samples taken from an industrial D.C. cast ingot\r\nand homogenized in a programmable temperature controlled laboratory furnace under\r\nvarious thermal profiles (i.e. homogenization temperatures 550\u00C2\u00B0C, 580\u00C2\u00B0C and 610\u00C2\u00B0C at\r\nvarious heating rates and homogenization soak times of up to ten hours). The samples\r\nwere then characterized in terms of their microstructure (retained manganese in solid\r\nsolution, percentage a-phase, and size distribution and density of dispersoids).\r\nThe homogenization process parameters were found to affect the evolving\r\nmicrostructure profoundly with:\r\n\u00E2\u0080\u00A2 An increase in heat-up rate favoring a reduction in the number of evolving dispersoids.\r\n\u00E2\u0080\u00A2 An increase in soak temperature increasing the Mn in solid solution, and decreasing the\r\nnumber of dispersoids that form.\r\n\u00E2\u0080\u00A2 An increase in soak time up to 3 hrs increasing the volume percent of \u00CE\u00B1-\r\nAl\u00E2\u0082\u0081\u00E2\u0082\u0082(Fe,Mn)\u00E2\u0082\u0083Si.\r\nBased on this work, a homogenization profile for optimum microstructure and\r\ntexture development would include a fast heat-up rate to a high soak temperature\r\n(610\u00C2\u00B0C) with moderate soak times (up to 3 hrs)."@en . "https://circle.library.ubc.ca/rest/handle/2429/9389?expand=metadata"@en . "14001091 bytes"@en . "application/pdf"@en . "EFFECT OF HOMOGENIZATION ON THE MICROSTRUCTURAL DEVELOPMENT IN A D.C. CAST AA3104 ALUMINUM ALLOY USED FOR CANBODY STOCK by C H E T A K G A N D H I B . Tech., Indian Institute of Technology, Kharagpur, India, 1994 A T H E S I S S U B M I T T E D I N P A R T I A L F U L F I L L M E N T O F T H E R E Q U I R E M E N T S F O R T H E D E G R E E O F M A S T E R O F A P P L I E D S C I E N C E in T H E F A C U L T Y O F G R A D U A T E S T U D I E S (Department o f Metals and Materials Engineering) ' W e accept the thesis as comforming to the required standard T H E U N I V E R S I T Y O F B R I T I S H C O L U M B I A A p r i l , 1999 \u00C2\u00A9 Chetak Gandhi, 1999 in presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of The University of British Columbia Vancouver, Canada DE-6 (2/88) Abstract A s customer demands become more stringent for canbody stock, it becomes essential to understand the complex interaction between the processing conditions and resulting product properties. This research focused on investigating the influence o f homogenization process parameters (heat-up rate, soak temperature and time) on the microstructural evolution o f an A A 3 1 0 4 aluminum alloy used for canbody stock. Experiments were conducted on samples taken from an industrial D . C . cast ingot and homogenized i n a programmable temperature controlled laboratory furnace under various thermal profiles (i.e. homogenization temperatures 550\u00C2\u00B0C, 580\u00C2\u00B0C and 610\u00C2\u00B0C at various heating rates and homogenization soak times of up to ten hours). The samples were then characterized i n terms o f their microstructure (retained manganese i n solid solution, percentage a-phase, and size distribution and density o f dispersoids). The homogenization process parameters were found to affect the evolving microstructure profoundly with: \u00E2\u0080\u00A2 A n increase i n heat-up rate favoring a reduction in the number o f evolving dispersoids. \u00E2\u0080\u00A2 A n increase in soak temperature increasing the M n i n solid solution, and decreasing the number o f dispersoids that form. \u00E2\u0080\u00A2 A n increase i n soak time up to 3 hrs increasing the volume percent o f a -A l i 2 ( F e , M n ) 3 S i . Based on this work, a homogenization profile for optimum microstructure and texture development would include a fast heat-up rate to a high soak temperature (610\u00C2\u00B0C) with moderate soak times (up to 3 hrs). i i Table of Contents Abstract ii Table of Contents iii List of Figures vi List of Tables ix Acknowledgements x 1 Introduction 1 2 Literature Review 5 2.1 Industrial Can Mak ing Operation 5 2.2 Solidification - As-cast Microstructure 6 2.3 Microstructural Changes During Homogenization 10 2.3.1 Equilibration o f Al loy ing Elements and Removal o f Microsegregation 11 2.3.2 Precipitation o f Dispersoids 14 2.3.3 Modification o f As-Cast Constituent Phases 15 2.3.4 M n in Sol id Solution 17 2.4 Effect o f Composition on the As-Cast and Homogenized Microstructure 19 2.4.1 Si l icon 19 2.4.2 Iron 20 2.4.3 Copper 20 2.4.4 Magnesium 21 2.4.5 Manganese 22 i i i 2.5 Effect of Homogenization Process Parameters on Microstructure 23 2.5.1 Heat-up Rate 24 2.5.2 Soak Temperature 24 2.5.3 Soak time 29 2.5.4 Second Stage Soak 30 2.6 Models for Microstructural Changes during Homogenization 32 2.7 Summary 34 3. Scope and Objectives 35 3.1 Scope 35 3.2 Objectives 36 4. Experimental 37 4.1 Start Material 37 4.2 Homogenization Experiments 38 4.2.1 Laboratory furnace 38 4.2.2 Homogenization profiles 39 4.3 Manganese Retained i n Sol id Solution 42 4.4 Microstructure Characterization / Metallography 43 4.4.1 Grinding and Polishing Procedure 43 4.4.2 Etching 44 4.4.3 Electron Probe M i c r o Analysis ( E P M A ) Line Scans 47 4.4.4 Image Analysis 47 4.5 As-Cast Characterization... 48 iv 5. Results and Discussion 50 5.1 Microstructure Evolution during Homogenization 50 5.1.1 Effect o f Heat-up Rate 50 5.1.1.1 Effect on M n in Sol id Solution 51 5.1.1.2Constituent Particles and percent a - A l i 2 ( F e , M n ) 3 S i 53 5.1.1.3Evolution o f Dispersoids 54 5.1.1.4Removal o f Microsegregation 58 5.1.2 Effect o f Soak Temperature and Soak Time 59 5.1.2.1 Effect on M n in Sol id Solution 59 5.1.2.2Constituent Particles and Percent a - A l i 2 ( F e , M n ) 3 S i 62 5.1.2.3 Evolution o f Dispersoids 64 5.1.2.4Cool D o w n to Rol l ing Temperature 66 5.2 Influence o f Homogenization Microstructure on Texture/Microstructure after Hot-Rol l ing 70 6. Summary and Conclusions 73 6.1 Summary 73 6.2 Conclusions 75 6.3 Future Work 77 Bibliography 78 Appendix A .....83 Appendix B 86 Appendix C 90 V List of Figures Figure 1.1- Aluminum beverage can demand [1 ]. 1 Figure 1.2- Typical processing sequence for the production o f final gauge sheet [3]. 3 Figure 2.1 - Photomicrographs showing the size and shape o f the constituents in as-cast ingot at: (a) edge and (b) center locations [12]. 7 Figure 2.2 - Diffusion coeffients for various elements in aluminum at different temperatures [18]. 12 Figure 2.3 - Effect o f 0.06% Fe and 0.02% C u on the loss o f supersaturation in Al -1 M n Al loys at 550\u00C2\u00B0C [30]. 21 Figure 2.4 - M n precipitation vs. time and temperature for an as-cast 3104 al loy [34]. 25 Figure 2.5 - M n content i n a-phase constituents as a function o f homogenization temperature [26]. 27 Figure 2.6 - Examples o f secondary precipitates in T E M thin foil specimens o f A A 3 0 0 4 alloy at (a) 600\u00C2\u00B0C and (b) 560\u00C2\u00B0C [27]. 28 Figure 2.7 - T E M photomicrographs showing the a-dispersoids i n ingot homogenized at 577\u00C2\u00B0C for (a) 14 hrs and (b) 26 hrs, followed by water quench [12]. 30 Figure 4.1 - Laboratory furnace set-up: a ) overview and b ) cross-sectional view. 38 Figure 4.2 - Experimental homogenization profiles with varying heat-up rates (dots mdicating where samples were taken for microstructure characterization). 40 Figure 4.3 - Experimental homogenization profiles with varying soak temperatures (550\u00C2\u00B0C, 580\u00C2\u00B0C and 610\u00C2\u00B0C). 40 Figure 4.4 - Experimental thermal profiles used for samples homogenized at different soak temperatures. 41 Figure 4.6 - Resistivity o f aluminum alloy 3004 as a function o f the percentage o f M n in solid solution [43]. 43 Figure 4.7 - As-cast microstructure (1000X) showing the various second phase particles. 45 Figure 4.8 - E D X / S E M results confirming the etching difference between a -A l i 2 ( F e , M n ) 3 S i and Al6(Fe,Mn) particle. 46 v i Figure 4.9 - Variation of the cumulative average per field with the measured quantity per field for volume fraction second phase. 48 Figure 5.1 - Effect of heat-up rate on % M n in solid solution. 51 Figure 5.2 - Effect o f heat-up rate on % a-Ali2(Fe,Mn)3Si after a 3 hour hold at 580\u00C2\u00B0C. 54 Figure 5.3 - Microstructures of samples for 580\u00C2\u00B0C soak temperature after 3 hour hold for: a) 279\u00C2\u00B0C/hr and b) 47\u00C2\u00B0C/hr. 55 Figure 5.4 - Microstructures of samples at 450\u00C2\u00B0C during heat-up for: a) 279\u00C2\u00B0C/hr and b) 47\u00C2\u00B0C/hr. 57 Figure 5.5 - M n in solid solution at different soak temperatures at a) 47\u00C2\u00B0C/hr and b) 279\u00C2\u00B0C/hr. 61 Figure 5.6 - Effect o f soak temperature on %a-Ali2(Fe,Mn)3Si after a soak period o f three hours. 62 Figure 5.7 - Effect o f soak time on %a-Ali 2(Fe,Mn)3Si at various soak temperatures at a heat-up rate o f 47\u00C2\u00B0C/hr. 63 Figure 5.8 - Microstructures (1000X) o f samples after three hour hold with a heat-up rate o f 47\u00C2\u00B0C/hr for soak temperatures of: a) 550\u00C2\u00B0C, b) 580\u00C2\u00B0C and c) 610\u00C2\u00B0C. 66 Figure 5.9 - M n i n solid solution for homogenization profiles o f 550\u00C2\u00B0C, 580\u00C2\u00B0C and 610\u00C2\u00B0C for three hour hold followed by cool down to 500\u00C2\u00B0C. 67 Figure 5.10 - Microstructures (1000X) o f samples after a cool down to 500\u00C2\u00B0C for soak temperatures of: a) 550\u00C2\u00B0C, b) 580\u00C2\u00B0C and c) 610\u00C2\u00B0C. 69 Figure A l - Temperature variations between furnace (580\u00C2\u00B0C) and sample o f original furnace at a) two sample positions in the furnace, and b) on a raised platform near the furnace thermocouple. 84 Figure A 2 - Temperature variations between furnace (580\u00C2\u00B0C) and sample temperature in the retrofitted furnace. 85 Figure B l - Microstructures o f samples at 580\u00C2\u00B0C soak temperature after 3 hr hold for a) 37\u00C2\u00B0C/hr, b) 47\u00C2\u00B0C/hr, c) 70\u00C2\u00B0C/hr, d ) l 12\u00C2\u00B0C/hr, and e) 279\u00C2\u00B0C/hr. 89 Figure C l - E P M A line scans ( M g and M n ) for the as-cast microstructure. 91 Figure C 2 - E P M A line scans (Mg and Mn) for 47\u00C2\u00B0C/hr profile at a) 400\u00C2\u00B0C, b) 500\u00C2\u00B0C, 580\u00C2\u00B0C (0 hrs), and d) 580\u00C2\u00B0C (3hrs). 93 v i i Figure C3 - E P M A line scans ( M g and Mn) for 279\u00C2\u00B0C/hr profile at a) 400\u00C2\u00B0C, b) 500\u00C2\u00B0C, 580\u00C2\u00B0C (0 hrs), and d) 580\u00C2\u00B0C (3hrs). v i i i List of Tables Table 1.1 - Mechanical properties of aluminum sheet in H I 9 condition. 2 Table 2 . 1 - Nominal weight percent composition of selected 3xxx alloys. 5 Table 2.2 - Size variation of cells and stringers with various casting methods [13]. 8 Table 2.3 - Relative solid solubility and rate o f diffusion o f major alloying elements in 3xxx series alloys [10]. 12 Table 2.4 - Effects o f homogenization treatment and Si content on oc-phase constituents (i.e. a-Ali2(Fe,Mn) 3Si) [4]. 16 Table 2.5 - Change in a-Ali2(Fe,Mn) 3 Si dispersoid mean size during preheat heat-up and soak [16]. 29 Table 4.1 - Composition o f A A 3 1 0 4 alloy used. 37 Table 4.2 - Range o f homogenization process parameters used i n the study. 39 Table 4.3 - Composition o f A A 3 1 0 4 and effect o f elements on resistivity [18]. 42 Table 4.4 - Grinding and polishing procedure followed. 44 Table 4.5 - As-cast characterization between 5 - 2 5 cm from surface. 49 Table 5.1 - M i n i m u m M n in solid solution and corresponding temperature for different heat-up rates. 52 Table 5.2 - M n and M g segregation results for 47\u00C2\u00B0C/hr and 279\u00C2\u00B0C/hr profile at various temperatures during heat-up. 58 ix Acknowledgements I would like to thank, firstly, my advisor, Dr. Mary A . Wells , for her constant support and encouragement throughout this work. I would also like to thank N S E R C , Canada and Alcan International Ltd. for the material and financial help, and Dr. D . J. L loyd ( K R D C , Alcan Int. Ltd.) for many useful discussions and suggestions. Special thanks to M r . Ross Mcleod and M r . Car l N g for preparing all my samples, to M r . Rudy Cardeno for the image analysis work, to M s . M a r y Mager for her instructions and advice with the S E M , to M r . Serge Mil lare for the help with the furnace and other electrical issues, and to Peter M u s i l for help with the furnace repair and fittings. I would like to convey many thanks to M r . Gary Lockhart for his help with experimental set-up and valuable advice and Rob Stevenson for his help i n the project. Finally, this work would not have been completed without the encouragement and support o f my family and friends. x 1 Introduction Over the last decade there has been considerable growth in the use of aluminum alloys for rigid packaging of food products, with the most significant increase being in the beverage container market. The world market for beverage containers has seen a steady rise over the last ten years [1] (Figure 1.1). The continuing strong demand for aluminum alloys in this very competitive market is forcing aluminum sheet producers to critically examine their production processes in terms of both cost and final material properties. Mideast & Africa Hi Europe M U S . & Canada Latin America & Mexico Pacific Figure 1.1 - Aluminum beverage can demand [1]. 1 Aluminum beverage cans are deep drawn from the final cold rolled sheets o f A A 3 0 0 4 or A A 3 1 0 4 aluminum alloys [2, 3]. These alloys are essentially A l - M n - M g alloys also containing C u , S i and Fe. The current demand is for A A 3 1 0 4 final canbody sheet that is < 0.3 m m gauge and with strength corresponding to the H I 9 condition (Table 1.1) Table 1.1 - Mechanical properties o f aluminum sheet in H I 9 condition. Tensile strength (UTS) 295 M P a Y i e l d Strength 285 M P a Elongation 2% Besides strength, elongation and gauge, another important requirement for the final sheet is its crystallographic texture. Undesirable textures i n the sheet leads to \"earing\" which may cause trouble during can making as well as yield losses as they must be trimmed off. The phenomenon o f \"earing\" refers to undulations on the r im o f cups formed during cup drawing or wa l l ironing, the high points named as \"ears\". During processing, earing is controlled by balancing two types of textures; i.e., by developing a strong cube texture (0-90\u00C2\u00B0 earing) upon recrystallization after hot roll ing followed by the superposition o f a rolling texture (45\u00C2\u00B0-45\u00C2\u00B0 earing) during subsequent heavy cold rolling to final gauge [4, 5 ,6] . The resulting product w i l l not show zero earing, but rather small mixed ears of similar size. 2 Typical ly processing o f Direct C h i l l (D.C.) cast aluminum ingots (generally 300 -760 mm thick) to final gauge sheet (-0.3 mm thick) for the beverage can market involves a number o f processing steps (Figure 1.2). Process time Figure 1.2 - Typical processing sequence for the production o f final gauge sheet [3]. The process steps include: i) homogenization (temperatures above ~550\u00C2\u00B0C and soaked for a controlled period). ii) breakdown rolling (the temperature falls from ~500\u00C2\u00B0C to ~300\u00C2\u00B0C and the ingot is reduced to a slab o f ~25 mm thickness). i i i ) tandem rolling (multiple stand continuous mi l l with 3-4 stands produce coiled metal ~2.5 m m thick that recrystallizes without furnace annealing). 3 iv) cold rolling (-87% reduction in thickness to final sheet with < 0.3 mm thick). v) final heat treatment. One area in sheet production that plays a crucial role in the final sheet gauge properties of canbody stock is homogenization. Homogenization, also known as pre-heating, is typically used to: (i) remove microsegregation of alloying elements in the ingot and (ii) raise the ingot temperature for hot rolling. The most economical way to run this process would be to have a cycle as short as possible and at as low a temperature as possible. Balanced with this is the need to produce an optimum microstructure capable of producing the desired properties in the final gauge sheet. In addition, the microstructure of the homogenized ingot can influence the strength, formability and the texture of the final gauge sheet for the canbody stock. Some of the important microstractural features in the as-homogenized ingot include relatively large constituent particles, fine dispersoids that are precipitated during processing and the solute levels in solid solution [7]. Each of these can affect the recrystallization process and in particular, the strength of the cube texture after hot deformation, which is of paramount importance in controlling anisotropy in the final product [5, 8]. In an effort to understand and quantify the microstructural changes that occur during homogenization, this study examined the effect of the homogenization process on key microstructural features. 4 2 Literature Review 2.1 Industrial Can Making Operation The manufacture o f beverage cans is one o f the most demanding sheet applications for aluminum [9]. The primary alloys used for canbody stock are AA3004 and A A 3 1 0 4 , which are non-heat-treatable aluminum alloys. The nominal specification for these alloys is given in Table 2.1 but typically beverage manufacturers operate on much tighter specifications than those given. AA3004 and A A 3 1 0 4 are essentially the same alloys with the exception of the tighter M n range in A A 3 0 0 4 . Table 2 . 1 - Nominal weight percent composition o f selected 3xxx alloys. A l l o y M g M n Fe S i C u A A 3 0 0 4 0.8-1.3 1.0-1.5 0.7 max. 0.3 max. 0.25 max. A A 3 1 0 4 0.8-1.3 0.8-1.4 0.8 max. 0.4 max. 0.05-0.25 Conventional body stock for beverage cans is produced from 600 to 750 mm diameter ingots by semi-continuous direct-chill (D.C.) casting or electromagnetic casting ( E M C ) processes. Other methods for casting canbody stock that have been tried experimentally but are currently not widely used are: (a) continuous casting o f 13 to 25 5 mm thick slabs by introducing liquid metal between parallel mold surfaces (e.g., the Hazelett water-cooled steel belt casters, Alusuisse Caster II and Lauener block casters which employ large metal chi l l blocks) (b) continuous casting o f 6 to 12 m m thick strips produced by introducing liquid metal between water cooled rolls, followed by cold rolling to final gauge without any preheating or hot rolling (e.g., Pechiney Jumbo 3C caster and the Hunter process). 2.2 Solidification - As-cast Microstructure The as-cast microstructure defines the starting structure for the study o f microstructure evolution during homogenization. The cast structure o f A A 3 x x x consists o f a cellular dendritic aluminum matrix with stringers containing intermetallic phases along the cell boundaries, remnants o f the last highly segregated melt to freeze (Figure 2.1). The main features defining the cast structure are: (a) the cell size and intermetallic particle size, (b) the degree o f solute supersaturation, and (c) the morphology o f the intermetallics. Due to the thickness of the ingot, the freezing rate varies from the edge to the center (Figure 2.1) [10, 11, 12]. The freezing rate can also vary greatly based on the casting practices. A s the freezing rate during casting increases, the dendrite cel l size decreases while the stringers become thinner and envelop the cell boundaries more continuously (Table 2.2). The cell size plays an important role in the equilibration process o f the alloying elements and the spatial arrangement o f the precipitating dispersoids during homogenization. Typically, the finer the cell size the smaller the distance which the elements must travel within the cell and therefore homogenization 6 occurs more quickly and the precipitate stringers tend to be finer yet engulf the cells more continuously [10, 11, 12]. I I a 50 Lim I I b 50 nm Figure 2.1 - Photomicrographs showing the size and shape of the constituents in as-cast ingot at: (a) edge and (b) center locations [12]. 7 Table 2.2 - Size variation of cells and stringers with various casting methods [13]. Thickness Freezing Cell Size Stringer Shape Process Rate- edge thickness (mm) (\u00C2\u00B0C/s) (urn) (um) Ingot D.C* 660 1 60 1.0 Billet D .C* 200 10 45 0.7 Slab Belt/Block 19 20 30 0.5 Stationary Strip Mold 13 100 10 0.2 Strip Twin Mold 6 700 6 0.1 *NOTE: The freezing rate at the center of a D.C cast ingot is ~0.25\u00C2\u00B0C/sec whereas at the edge is ~l\u00C2\u00B0C/sec) [13]. The intermetallic phases that form in AA3104 consist of two major types: orthorhombic Al6(Fe,Mn) and cubic a-Ali2(Fe,Mn)3Si. The relative amounts of these phases are determined by the alloy composition as well as the solidification rate [11, 12, 14,15]. In AA3104 under commercial D.C-cast ingot cooling rates of about l\u00C2\u00B0C/s (Table 2), about 85% of the primary constituent particles in the as-cast structure correspond to the Al6(Fe,Mn) phase and the remainder to oc-Ali2(Fe,Mn)3Si [14]. At these cooling rates, about 25-30% of the Mn precipitates out of solid solution in the form of intermetallic particles during solidification while the rest of the Mn (-70-75%) remains in solid solution, thereby producing an as-cast structure which is in a supersaturated metastable 8 solid solution condition [14]. Moreover, the as-cast structure exhibits microsegregation of M g , S i , and to a lesser extent M n , across the dendritic cellular cast structure [16]. The formation of phases and the temperatures at which they nucleate during the solidification process of A A 3 0 0 4 have been summarized by Backerud et al [17]. Initially as the molten aluminum begins to cool and its temperature reaches ~648-652\u00C2\u00B0C the liquid w i l l begin to solidify and form an aluminum dendritic matrix according to : 1. L iquid (L) -> A l , dendritic network @ 648-652\u00C2\u00B0C The first constituent particles A l 6 (Fe ,Mn) start to nucleate at the periphery o f the growing dendrites at ~643\u00C2\u00B0C according the eutectic phase transformation 2; 2. L iquid (L) -> A l + Al 6 (Fe ,Mn) @ 643\u00C2\u00B0C A t lower temperatures (~638\u00C2\u00B0C), a peritectic reaction occurs between the (Mn,Fe)Al6 particles and the remaining liquid which has been enriched in Si according to 3 a: 3a. Liquid (L) + A l 6 (Fe ,Mn) -> a - A l 1 2 ( F e , M n ) 3 S i @ 638\u00C2\u00B0C A s the liquid continues to be enriched in solute the a-Ali2(Fe,Mn)3Si phase can form directly from it according to 3b: 3b. Liquid (L) -> A l i 2 ( F e , M n ) 3 S i @ 638\u00C2\u00B0C Finally, at slightly lower temperatures, eutectic P-Mg 2Si particles precipitate from the melt given the high S i and M g in the alloy according to 4; 4. L iquid (L) -> A1 + A l 1 2 ( F e , M n ) 3 S i + M g 2 S i @ 630\u00C2\u00B0C Homogenization, which occurs after casting of the ingot, modifies the as-cast microstructure significantly and has a direct influence on both the hot-rolled and cold-9 rolled sheet properties. A s a result it is important to understand the microstructural changes which occur during homogenization, the effect o f homogenization parameters on the microstructure evolution and how this behaviour can be represented mathematically. 2.3 Microstructural Changes During Homogenization During homogenization, the D . C . cast ingot is heated and soaked at a temperature in the range o f about 10-100\u00C2\u00B0C below the solidus for the alloy and 20-100\u00C2\u00B0C above the desired initial hot working temperature. For A A 3 x x x alloys, the solidus ranges from 630-645\u00C2\u00B0C, homogenization temperatures are typically 530-620\u00C2\u00B0C and the initial hot rolling temperatures are 480-520\u00C2\u00B0C [10]. The basic purpose of the homogenization process is to reduce microsegregation that exists in the as-cast ingot as well as preheat the ingot to the desired hot rolling temperature. Concurrent with the constitutional equilibration and reduction i n microsegregation o f some elements in solid solution during homogenization, several other rmcrostructural changes occur in the matrix, such as: i) precipitation o f supersaturated elements as dispersoids, ii) dissolution o f unstable phases or precipitates, i i i ) transformation o f Alg(Fe,Mn) to a-Ali2(Fe,Mn)3Si, and iv) coarsening of the stable intermetallic particles. 10 2.3.1 Equilibration of Alloying Elements and Removal of Microsegregation Microsegregation in the as-cast ingot and equilibration o f alloying elements during homogenization depend upon a number o f factors: their solid solubility in the aluminum matrix, their rate o f diffusion in the aluminum matrix for the element and the cell size o f the cast microstructure. In A A 3 x x x aluminum alloys both M n and Fe have relatively low solubility in the aluminum matrix and hence they tend to be the elements which w i l l be in a supersaturated condition in the as-cast aluminum matrix (Table 2.3). During homogenization, the supersaturation of M n and Fe within the dendritic cells is largely relieved by the precipitation o f fine (Fe,Mn)-rich dispersoids consisting primarily o f A l i2 (Mn,Fe) 3 Si phase [16], or Ale(Mn,Fe) i f the S i content is less than about 0.07 wt% [4]. In the as-cast aluminum structure both M g and S i w i l l exhibit a significant amount o f microsegregation or coring across the dendritic cellular as-cast microstructure. The segregation pattern existing i n the as-cast microstructure w i l l be unstable as it is heated up to and held at the homogenization temperature and equilibration o f these elements w i l l begin to occur during heat-up to the homogenization soak temperature. In fact, by the time the ingot has reached 500\u00C2\u00B0C, the M g and S i levels have typically homogenized completely across the cell [16]. 11 Table 2.3 - Relative solid solubility and rate o f diffusion of major alloying elements in 3xxx series alloys [10]. Moderate solubility, M n , L o w diffusion rate Appreciable solubility, Cu , M g , Si High diffusion rate Extremely low solubility, Fe Very low diffusion rate 1200 1000 800 600 F 1 < 1 i 1 \u00E2\u0080\u00A2 1 600 500 cr 8 i P T \u00E2\u0080\u0094 i \u00E2\u0080\u0094 i c \u00E2\u0080\u0094 i \u00E2\u0080\u0094 I I I | i i i 400 300 C - i 1 1 1\u00E2\u0080\u0094l 1.4 (000/T Figure 2.2 - Diffusion coefflents for various elements in aluminum at different temperatures [18]. 12 The mechanism behind the homogenization process is diffusion and M n and Fe have much lower diffusion coefficients (Figure 2.2). Since most Fe stays out o f solid solution, the rate determining element is M n . Diffusion o f M n \ is influenced by both temperature and time and can be described mathematically by Fick's law: dt dx2 This equation states analytically how the concentration (C) w i l l change with time (t) according to a temperature dependent constant D which is called the diffusion coefficient. D depends on the temperature according to the Arrhenius equation: D = D 0 e x p - Q / R T (2) Where: D 0 = constant Q = activation energy (kJ/mole) R = gas constant (8.314 J/mole\u00C2\u00B0K) T = temperature (\u00C2\u00B0K) It is this exponential dependence on temperature that causes diffusional removal o f concentration gradients to take place more rapidly at higher temperatures. Since this 13 process is governed entirely by diffusion, three inain factors w i l l determine the time needed for this to occur: i) temperature, ii) the actual values o f D 0 and Q in the Arrhenius equation which w i l l differ for different elements, and iii) the cell size o f the cast microstructure since this w i l l determine the distance over which diffusion must take place. 2.3.2 Precipitat ion of Dispersoids Dispersoids are secondary precipitates that form during homogenization as a result of supersaturation of particular elements in the matrix. In A A 3 x x x alloys, these dispersoids are mostly oc-Ali 2(Fe,Mn)3Si, or sometimes Al6(Fe,Mn) i f the S i is lower than 0.08% in the ingot [4, 10, 16]. The formation of these precipitates during the homogenization process is a complex procedure involving the nucleation o f the new phase into the aluminum matrix followed by growth until all of the excess solute has been consumed. In A A 3 x x x aluminum alloys, as the ingot is heated above 300\u00C2\u00B0C, needles o f and more sparsely distributed square platelets of P-Mg2Si begin to precipitate in the A l matrix [16]. A t a slightly higher temperature of around 400\u00C2\u00B0C, precipitates of oc-Ali2(Fe,Mn)3Si begin to nucleate on the pre-existing (3-Mg2Si needles which subsequently dissolve as the temperature is increased. A s the temperature is further increased, a-Ali 2 (Fe,Mn)3Si dispersoids continue to precipitate while the existing precipitates begin to grow through coarsening mechanisms. Both the volume fraction and average size o f dispersoids increases until ~ 480\u00C2\u00B0C when the M n precipitation rates peak 14 [4, 19]. Above that temperature, re-dissolution o f dispersoids take place due to increased solid solubility o f M n in A l at higher temperatures [4, 16, 20]. The re-dissolution of dispersoids causes their number to decrease, however, those dispersoids that remain do undergo growth [16]. This shows that while the dispersoids nucleation period stops at ~480\u00C2\u00B0C, the growth period continues at higher temperatures. Subsequent cooling from the peak preheating temperature to the hot rolling temperature causes renewed precipitation and growth of oc-Ali2(Fe,Mn)3Si dispersoids. The final average size of dispersoids depends on the homogenization profile and can vary from 50nm to about 0.5 urn in diameter [4,16, 19]. 2.3.3 Modification of As-Cast Constituent Phases The iron-rich intermetallic particles which form during solidification (Ai^(Fe,Mn)) are not removed during homogenization since the alloy is never heated into a single phase field and the diffusion coefficient of Fe in aluminum is so low. However, some changes, which are both time and temperature dependent, in the as-cast constituent particles do begin to occur during the homogenization process: i) Ale(Fe,Mn) starts to transform to a-Al(Fe ,Mn)Si , and ii) Spherodization of the particles. Between 480\u00C2\u00B0C and 593\u00C2\u00B0C, the normally dominant Ale(Fe,Mn) constituent phase gradually transforms to Ali2(Fe,Mn)3Si phase. This transformation occurs due to localized enrichment o f the matrix with S i , originally present in the form o f M g 2 S i i n the as-cast microstructure but which dissolves during the early stages o f the homogenization 15 process. The transformation is initiated by nucleation of a-Ali2(Fe,Mn)3Si at the interface between Ai6(Fe,Mn) and the solute enriched matrix [21]. The a-Ali2(Fe,Mn)3Si phase then grows into and consumes the Ale(Fe,Mn) particles. Depending on the composition o f the alloy, this transformation can either go quickly to completion or be stopped before completion because of a lack of S i . In order for this reaction to proceed, M n and S i must diffuse through the matrix, and the rate o f the reaction is usually controlled by M n diffusion. If Si is tied up as M g 2 S i , due to high M g or low S i , the transformation is slowed down due a lack o f S i . In general, this transformation is promoted by a higher Si content and a higher homogenization temperature (Table 2.4) [4, 22, 23]. Table 2.4 - Effects o f homogenization treatment and S i content on a-phase constituents (i.e. a - A l i 2 ( F e , M n ) 3 S i ) [4]. Simulated Homogenization Treatment Percent of a - A l i 2 ( F e , M n ) 3 S i 0.21% S i Ingot 0.30% S i Ingot As-cast 10 20 Heated to 427\u00C2\u00B0C 10 20 Heated to 482\u00C2\u00B0C 25 35 Heated to 538\u00C2\u00B0C 50 75 Heated to 566\u00C2\u00B0C 70 97 Heated to 593\u00C2\u00B0C 85 100 Heated to 616\u00C2\u00B0C 85 100 Heated to 632\u00C2\u00B0C 50 100 Heated to 638\u00C2\u00B0C 0 100 16 The exact significance o f the transformation o f Al6(Fe,Mn) to a -Al (Fe ,Mn)S i is not completely understood although an increased level o f a -Al (Fe ,Mn)S i has been reported to improve the drawing and ironing behaviour in canmaking; the a -Al (Fe ,Mn)S i particles are harder than the Al6(Fe,Mn) particles and help in cleaning the dies thereby prevent galling [4]. The constituent particles can also change their morphology during homogenization. Initially, the particles have a plate-like elongated and irregular shape with a high surface area. During homogenization, spherodization o f particles can start to take place as the elongated particles break up into smaller fragments having a more globular appearance [16, 24, 25]. 2.3.4 M n in Solid Solution The level o f M n in solid solution after the homogenization process for A A 3 x x x alloys can influence the properties of the final gauge sheet. During the homogenization process, the amount o f M n in solid solution w i l l initially drop as M n precipitation rates peak at temperatures between 475\u00C2\u00B0C and 500\u00C2\u00B0C depending on the alloy composition [4, 16, 20]. A s the temperature is further increased, the amount o f M n in solid solution increases as the solubility o f M n i n the matrix increases. A t the end o f homogenization, however, there is considerably less M n i n solid solution than i n the as-cast microstructure. M n depletion in the solid solution takes place in three ways: i) Precipitation of Mn-r ich dispersoids. 17 ii) Transformation o f A l 6 (Fe ,Mn) to a - A l i 2 ( F e , M n ) 3 S i : Total (Fe+Mn) in A l 6 ( F e , M n ) is about 25 wt.% compared with 32 wt.% o f (Fe+Mn) in a - A l i 2 ( F e , M n ) 3 S i phase [4, 26, 27]. Wi th very little Fe available in the matrix, it is mostly achieved through M n enrichment. i i i) Enrichment o f M n in both dispersoids and eutectic constituents [10, 26, 27]. During homogenization, M n is known to diffuse into the particles thereby increasing the % M n in both Al 6 (Fe ,Mn) [10] and a - A l i 2 ( F e , M n ) 3 S i [26, 27] particles. Hence, though the (Fe+Mn)/Al i 2 (Fe ,Mn) 3 Si ratio o f the particles stays fairly constant, the Mn/(Fe+Mn) increases with an increase in homogenization temperature or decrease in cell size o f the ingot. 18 2.4 Effect of Composition on the As-Cast and Homogenized Microstructure The exact composition o f the alloy w i l l have a significant influence on both the structure that forms during solidification of the ingot as wel l as some o f the microstructural changes which occur during the homogenization process 2.4.1 Si l icon Commercial producers o f beverage canbody stock presently have between 0.12% and 0.30% S i i n their alloys. Varying silicon in the ingot affects the developing microstructure i n the following ways i) Effect on constituent particles: an increase in percentage o f Si in the alloy gives a higher percentage o f cc-Ali2(Fe,Mn)3Si in the as-cast structure and also promotes the transformation o f A l 6 (Fe ,Mn) to a-Ali 2(Fe,Mn)3Si during various stages of homogenization treatment [4, 28] (Table 2.4). ii) Effect on M n solubility: S i enhances the precipitation o f M n as we l l as other constituent elements such as Fe and M g . Increasing S i reduces the amount o f M n in solid solution in the as-cast as well as homogenized ingot [4, 28]. S i reduces M n content in solid solution by causing a higher proportion o f a-Ali2(Fe,Mn)3Si to form, thereby removing more M n from solid solution (Table 2.4). Whi le A l 6 ( F e , M n ) contains about 25 wt% (Fe + Mn) , oc-Al i 2 (Fe ,Mn) 3 Si contains about 32 wt% [26]. 19 i i i) Effect on Dispersoids: A s mentioned earlier, the dispersoids formed are mostly a -Ali2(Fe,Mn)3Si, but i f S i is below 0.08%, the dispersoids which form are o f the type Al 6 (Fe ,Mn) . 2.4.2 I ron The effect of iron is very pronounced in an A l - M n - S i alloy. Even a very small addition o f Fe to the A l - M n alloy greatly enhances the number of constituent particles which form facilitating the M n precipitation and preventing the formation o f the ternary metastable phases ( A l x M n y F e z ) usually encountered in the A l - M n - F e alloys [29]. N o ternary precipitation phase appears, but rather Fe atoms substitute for M n atoms in A l e M n . During homogenization, M n enters Al6(Fe,Mn) until the phase has achieved its maximum stability at Al-pi \u00C2\u00BB > 10 10 - A ] 17 u Exposure Time (Hrs) Figure 2.3 - Effect of 0.06% Fe and 0.02% Cu on the loss of supersaturation in Al-1 M n Alloys at 550\u00C2\u00B0C [30]. 2.4.4 Magnesium Magnesium (Mg) contributes to the overall strength of the alloy through solid solution hardening and work hardening [31]. AA3004/AA3104 have often been considered as non-heat-treatable A l - M n alloys to which M g is added to improve its work hardening characteristics. Furthermore, a comparison of similar AA3xxx alloys indicates that the M g containing AA3104 requires substantially less homogenization time than the equivalent AA3003 without M g [32]. This has been explained by the role of Mg^Si matrix precipitation acting as nucleation sites for a-dispersoids during the early stages of ingot heating [4,16, 32]. 21 2.4.5 Manganese M n in canbody stock provides large hard particles effective in die-cleaning during drawing operations. It also imparts some strength to the final gauge sheet through solution hardening and excellent corrosion resistance [33]. Varying the M n content o f the ingot from 0.6wt% to 0.9wt% w i l l give a higher M n in solid solution throughout the homogenization process [26]. However, the resulting microstructure, in terms of volume fraction and size distribution of constituent particles and precipitation and dissolution of dispersoids, is not significantly different. 22 2.5 Effect of Homogenization Process Parameters on Microstructure Although studies have been completed over the past decade [10, 24, 25, 34] in an effort to understand the effect of the homogenization profile on the microstructural evolution, the ability to predict an optimum homogenization profile for a given starting structure and alloy composition has not yet been developed. In fact, currently, industry uses a wide variety of homogenization profiles to process the same alloy even though the starting structure and chemistry may be very similar. Industrial homogenization profiles can vary significantly in terms of the soak temperature and time as well as the overall homogenization profile (i.e, a second stage soak at a lower temperature after the primary soak is complete). The final microstructure of the as-cast material after homogenization is chiefly dependent on the type of thermal profile (temperature and time) it has undergone. The parameters that can be varied in any thermal profile are: \u00E2\u0080\u00A2 heat-up rate, \u00E2\u0080\u00A2 soak temperature, \u00E2\u0080\u00A2 soak time, and \u00E2\u0080\u00A2 cool down rate. Each o f these process parameters can affect the microstructural evolution during the homogenization process.. 23 2.5.1 Heat-up Rate Currently, industrial heat-up rates during homogenization o f ingot material typically range from around 45\u00C2\u00B0C/hr (soaking pit furnace) to around 70\u00C2\u00B0C/hr (walking beam furnace)[35]. To date, almost no research has been done on the effect of heat-up rate on the microstructure evolution or properties of homogenized canbody stock material. Sheppard [36] was the only researcher who studied the effect o f heat-up rate on the microstructure evolution of a 7xxx series alloy. His findings showed that a slower heat-up rate to homogenization temperature had no influence on the constituent particles however that it did affect the dispersoid size and distribution, with the slower heating rate giving more dispersoids with a more homogeneous distribution. His results were explained using classical precipitate nucleation theory; a slower ramp provided a longer period within the temperature zone corresponding to maximum nucleation rates hence more dispersoids could precipitate. Later during the soak period at the homogenization temperature, precipitate growth and precipitate nucleation become competing processes whereby pre-existing nuclei grew at the expense o f further nucleation. 2.5.2 Soak Temperature Currently, the homogenization soak temperature for A A 3 x x x alloys used in the industry lies between 560\u00C2\u00B0C and 610\u00C2\u00B0C [35]. The ingots are heated to the soak temperature and held at this temperature for a period o f time to achieve the desired microstructure. Soak temperature also known as the homogenization temperature, impacts most aspects o f the microstructure such as: 24 i) percentage M n in solid solution, ii) constituent particles, and iii) dispersoids. Goodrich [34] investigated the effect of soak temperature on the final homogenized structure and percent M n left in solid solution. He estimated the M n in . solid solution using resistivity measurements (Figure 2.4). 6.Sr 4\" \" 100 150 200 250 3 0 0 3 3 0 400 4 5 0 SOO S50 600 Tenparatu-* QU) Figure 2.4 - M n precipitation vs. time and temperature for an as-cast 3104 alloy [34]. 25 A s shown, after around 550\u00C2\u00B0C, the difference between resistivity values ( % M n in solid solution) for 1 hr and 200 hrs is not very different indicating that M n equilibration rates are sufficiently faster at those temperatures. It can also be inferred from the graph that equilibrium M n in solid solution rises with an increase in temperature above 480\u00C2\u00B0C. This temperature can be referred to as the peak M n precipitation temperature since at this temperature there is least amount of M n in solid solution. This is due to the higher solubility o f M n in A l at higher temperatures [4, 16, 34]. Constituent particles undergo a variety o f changes as the homogenization temperature is raised over the peak M n precipitation temperature such as: (i) Al6(Fe,Mn) to a-Ali2(Fe,Mn)3Si transformation. (ii) Particle coarsening and spherodizing. (iii) M n enrichment in particles. The effect o f temperature on Al<5(Fe,Mn) to a-Ali2(Fe,Mn)3Si transformation can be estimated from Table 2.4. It is clear that an increase i n temperature increases the %cc-A l i 2 ( F e , M n ) 3 S i up to 593\u00C2\u00B0C for both S i levels. A t higher temperatures o f 632\u00C2\u00B0C and 638\u00C2\u00B0C, where incipient melting takes place, transformation kinetics become greatly enhanced and a reduction in the %a-Ali2(Fe,Mn) 3 Si is seen in the lower S i alloy. The constituent particles have shown a rounding o f their edges (spherodization), coarsening and break up o f the three-dimensional dendritic network when held at higher homogenization temperatures [22, 37,27]. Wi th an increase in homogenization temperatures, the M n levels i n particles is known to stabilize at different levels (Figure 2.5) [26]. Sun [26] suggested that it was 26 possible to determine the homogenization temperature from the composition of the eutectic constituents. 14.0 Sample A Sample B 10.54 1 \u00E2\u0080\u0094 i 1 580. 590 600 610 620 630 Preheating Temperature, C Figure 2.5 - Mn content in cc-phase constituents as a function of homogenization temperature [26]. Quite a few researchers [10,12, 16, 26, 27] have found that dispersoids that form during ingot heating become fewer and coarser as homogenization temperature increase. 27 There are fewer and coarser dispersoids present in a sample homogenized at 600\u00C2\u00B0C than one homogenized at 560\u00C2\u00B0C (Figure 2.6) [27]. 4j Figure 2.6 - Examples o f secondary precipitates in T E M thin foil specimens o f A A 3 0 0 4 alloy at (a) 600\u00C2\u00B0C and (b) 560\u00C2\u00B0C [27]. Bolingbroke [16] has characterized the increase in size o f the dispersoids with increase in temperature during heat-up (Table 2.5). It is clear that dispersoid coarsening is a direct function o f temperature. 28 Table 2.5 - Change in a - A l i 2 ( F e , M n ) 3 S i dispersoid mean size during preheat heat-up and soak [16]. Preheat treatment Average a-Ali2(Fe,Mn)3Si precipitate size (nm) As-cast -Heat-up to 400\u00C2\u00B0C 35 Heat-up to 450\u00C2\u00B0C 47 Heat-up to 500\u00C2\u00B0C 66 Heat-up to 600\u00C2\u00B0C 108 Heat-up to 600\u00C2\u00B0C, cooled to 500\u00C2\u00B0C at 30\u00C2\u00B0C/hr 136 Heat-up to 600\u00C2\u00B0C, soaked for 8 hrs and cooled to 500\u00C2\u00B0C at 30\u00C2\u00B0C/hr 150 2.5.3 Soak time The soak time at the high temperature allows the reaction kinetics to proceed. Both M g and S i have a high diffusion coefficient and w i l l equilibrate even as the temperature reaches the homogenization temperature. M n on the other hand, has a low diffusion coefficient and it is the rate determining step for the transformation of Al6(Fe,Mn) to ot-Ali2(Fe,Mn)3Si and dispersoid precipitation during the homogenization treatment. A s the homogenization temperature increases, the time required for 29 homogenization decreases [12, 38]. This is partly because the M n diffusion rates become faster and also partly because the equilibrium M n in solid solution is higher at higher temperature. Besides M n equilibration, time at homogenization temperature also allows for precipitate coarsening. For example, there is no change in M n level for an ingot homogenized at 577\u00C2\u00B0C for 14 hrs and for 26 hrs but T E M microstructures shows that a -dispersoids are almost twice the size after 26 hrs at 577\u00C2\u00B0C [12](Figure. 2.7) a 0.5 um b 0,Sj,m Figure 2.7 - T E M photomicrographs showing the cc-dispersoids in ingot homogenized at 577\u00C2\u00B0C for (a) 14 hrs and (b) 26 hrs, followed by water quench [12]. 2.5.4 Second Stage Soak Traditionally, industrial homogenization cycles consisted o f heating the ingot to a prescribed temperature and holding it there for a certain length o f time followed by air 30 cool to the hot roll ing temperature. Recently, researchers have investigated the merit of altering the traditional homogenization cycle to include a second soak period at a lower temperature on the microstructural evolution in the ingot [26, 39]. Hence, a typical two stage homogenization profile consists of heat-up, hold at high temperature (560\u00C2\u00B0C -610\u00C2\u00B0C) followed by controlled cooling to a lower temperature (450\u00C2\u00B0C - 500\u00C2\u00B0C) and a soak at this temperature prior to hot rolling. One major difference in microstructure after a second stage soak is the lower M n in solid solution compared to a traditional homogenization profile [39]. During the second stage cool, M n precipitation occurs in a controlled manner such that the wt % M n in solid solution is lower and at the same time the number o f dispersoids is not increased [12]. 31 2.6 Models for Microstructural Changes during Homogenization Very few mathematical models have been developed to predict the microstructural evolution in an aluminum ingot as a function of time and temperature during homogenization. Only two models were found in the literature that had been developed to model microstructural changes during homogenization of A A 3 x x x aluminum alloys. A mathematical model is an attractive way to describe microstructural evolution, because the response o f the microstructural features such as dispersoids and constituent particles to time, temperature and composition can be quantified. To date, two researchers have attempted to model dispersoid formation during homogenization o f A A 3 x x x alloys [34,40]. Goodrich [34] used an Avrami type formulation (equation 3) to model the amount o f M n i n solution for AA3104 . The precipitation or dissolution rate constants i n the model were dependent on temperature and the state o f the preheat cycle and these were determined experimentally from isothermal treatments of an ingot. The fit was apparently quite good with Avrami exponents between 0.2-0.3. X = l-exp(-kt n) (3) where X is the fraction of M n precipitated, k is the rate constant, 32 t is the time in hours, n is related to the type o f reaction. X is expressed as: X = ^ l - (4) c. - c, Where: C(t) is the dissolved M n at any time t, Cj is the initial concentration o f dissolved M n , C e is the \"pract ica l \" M n solubility for this alloy. Goodr ich ' s mode l was never validated using industrial measurements so it is difficult to comment on the accuracy o f the predictions. Suni [40, 41] developed a phys ical ly based model to predict the dispersoid evolution during homogenization o f A A 3 0 0 3 and A A 3 0 0 4 a luminum alloys. H i s mode l combined the processes o f nucleation, growth and coarsening so that particle size, density and vo lume fraction o f dispersoids cou ld be predicted. M o d e l parameters were determined using industrial A l c o a data and the mode l predictions were then validated using industrial data and in general predicted sizes compared we l l to measurements. 33 2.7 Summary Industrial can making is a complex process comprising various steps with many variable process parameters. Homogenization is the starting step in the processing of the ingot to the final sheet. It is important as the microstructure of the homogenized ingot can influence the strength, formability and the texture o f the final gauge sheet for the canbody stock. Although studies have been completed over the past decade [10, 24, 25, 34] in an effort to understand the effect of the homogenization on the microstructural evolution, the ability to predict an optimum homogenization profile for a given starting structure and alloy composition has not yet been developed. Although some studies have been done investigating the effect o f temperature and time of homogenization on the microstructures o f A A 3 x x x aluminum alloys [21-26], there is almost no work studying the effect o f heat-up rate. A s such, there is no comprehensive study on the combined effect o f the various homogenization parameters (heat-up rate, temperature, time, cool-down rate) on the microstructure evolution. 34 3 . Scope and Objectives 3.1 Scope The goal o f homogenization process is not only to preheat the material to the desired hot rolling temperature but also allow segregated areas in the material to equilibrate. Currently in industrial practice, a wide variety of homogenization practices exist i n terms o f soak temperature, soak times and heat-up rate. In addition, newer cycles involving a second stage cool followed by a soak at a lower temperature have started to be investigated. To date, a comprehensive study quantifying the effect o f homogenization practice on the microstructural evolution has not been done. Such a study w i l l be o f benefit to aluminum sheet metal producers in that it w i l l address the balance required i n terms o f developing the most economic practice in conjunction with one that obtains an optimum rmcrostructure for hot rolling. Although some studies have been done investigating the effect o f temperature and time o f homogenization on the microstructures o f A A 3 x x x aluminum alloys [21, 26], there is almost no work studying the effect of heat-up rate. A s such, there is no comprehensive study on the combined effect o f the various homogenization parameters (heat-up rate, temperature, time). A s a result, the purpose o f this study was to examine various homogenization profiles and their effect of on the microstructural development during homogenization o f an AA3104 aluminum alloy. 35 3.2 Objectives The objectives o f the present study are as follows: to simulate typical industrial homogenization profiles in a laboratory furnace, to characterize the effect of homogenization process parameters on microstructural evolution of aluminum alloy AA3104 . 36 4. Experimental 4.1 Start Material The experimental work was undertaken on an as-cast commercial A A 3 1 0 4 aluminum alloy. The material was received from Alcan International and was taken from an ingot that had been previously D . C . cast. The chemical composition o f the alloy as supplied by Alcan and verified through low magnification E D X / S E M is shown in Table 4.1. Table 4.1 - Composition of AA3104 alloy used. A l l o y M g (wt%) M n (wt%) Fe (wt%) S i (wt%) C u (wt%) A A 3 1 0 4 1.20 0.87 0.40 0.20 0.18 Microstructure and composition can vary dramatically across an ingot and the starting microstructure can greatly influence the microstructure evolution during homogenization [10, 11, 12]. To ensure homogeneity in the starting specimens, samples were cut 7-15cm from the ingot edge and were tested to ensure similar initial conductivity values. This ensured that conductivity (a measure o f the amount o f solute i n solid solution) and cell size differences were minimized. 37 4.2 H o m o g e n i z a t i o n E x p e r i m e n t s 4.2.1 Labora to ry furnace The samples were homogenized in a programmable controlled laboratory furnace. To ensure accurate temperature control and homogeneity within the furnace, an aluminum tube was wel l fitted into the furnace with the controlling thermocouple inside the aluminum tube (Figure 4.1). Measurements of thermal profiles inside the modified furnace using instrumented samples indicated that the temperature variations were \u00C2\u00B1 2\u00C2\u00B0C between the samples (Appendix A ) . K-thermocouple Aluminum Tube Furnace \ Samples Programmable temperature controller Aluminum Tube Samples a ) Overview b ) Cross-sectional view Figure 4.1 - Laboratory furnace set-up: a ) overview and b ) cross-sectional view. 38 4.2.2 Homogenization profiles A wide range of homogenization profiles were simulated in the laboratory furnace so that the effect o f heat-up rate, soak temperature, and soak time on the microstructural evolution could be quantified (Table 4.2). Table 4.2 - Range o f homogenization process parameters used in the study. Process Parameter Range Heat-up rate 3 7 - 2 7 9 \u00C2\u00B0 C / h r Soak temperature 5 5 0 - 6 1 0 \u00C2\u00B0C Soak time 0 - 1 0 hrs In addition, the effect o f cool down to the rolling temperature after the homogenization treatment on the microstructure evolution was also studied. Samples which were taken for microstructural characterization were quenched immediately from the homogenization cycle to ensure that the structure was representative o f what is seen during homogenization. Figures 4.2-4.4 show the experimental homogenization cycles used as wel l as where samples were taken for microstructural characterization. 39 0 5 10 15 20 25 30 Time (hrs) Figure 4.2 - Experimental homogenization profiles with varying heat-up rates (dots indicating where samples were taken for microstructure characterization). 700 0 5 10 15 20 25 Time (hrs) Figure 4.3 - Experimental homogenization profiles with varying soak temperatures ( 5 5 0 \u00C2\u00B0 C , 5 8 0 o C a n d 6 1 0 \u00C2\u00B0 C ) . 40 After an ingot is homogenized in industry, it cools down to the temperature at which hot rolling starts, also known as the lay-on temperature. Due to cool down, microstructural changes, especially in terms of M n in solid solution are expected. Therefore, experiments that were conducted at different homogenization temperatures were cooled down to 500\u00C2\u00B0C from the soak temperature in about half an hour (Figure 4.4). In order to cool the samples down to 500\u00C2\u00B0C, the furnace was turned off and the door was partially left open. 700 Time (hrs) Figure 4 . 4 - Experimental thermal profiles used for samples homogenized at different soak temperatures. 4 1 4.3 Manganese Retained in Solid Solution Electrical conductivity measurements were used to determine the amount o f solute in solid solution. The sample conductivity was measured at 20\u00C2\u00B0C as a percentage of International Annealed Copper Standard (%IACS) [18] using a Verimet 4900C eddy current probe. The readings (in % I A C S ) could then be directly used to estimate the % M n in and out of solid solution. This is because M g , a major alloying element, has a much lower effect on conductivity than M n (Table 4.3) [18], almost al l the Fe stays out of solution during the treatment, and S i and C u levels in the alloy are too low to significantly affect the conductivity readings. Resistivity values could be obtained from conductivity measurements by the simple equation [42]: p =172.41/ (%IACS) (5) where p is resistivity i n p.Qcm. Conversion o f resistivity values to wt.% M n i n solid solution can then be done using the graph shown in Figure 4.6 [43]. Table 4.3 - Composition of A A 3 1 0 4 and effect o f elements on resistivity [18]. Element A A 3 1 0 4 composition (wt%) Average increase in resistivity per wt% i n solution (p.Qcm) M n 0.87 2.94 M g 1.20 0.54 Fe 0.4 2.56 S i 0.2 1.02 42 1.0 12 1.4 MoC%) in SS Figure 4.6 - Resistivity of aluminum alloy 3004 as a function of the percentage of M n in solid solution [43]. 4 . 4 Microstructure Characterization / Metallography 4.4.1 Grinding and Polishing Procedure Samples were cold mounted in an epoxy resin. A l l samples were polished using the Buehler Ecomet I V Polisher/Grinder which is a ten mount automatic polisher. The following schedule (Table 4.4) was standardised after trial and error and consultation with Tech-Met Canada: 43 Table 4.4 - Grinding and polishing procedure followed. Abrasive type Abrasive Size Lubricant Pressure (psi) Time (min) Additional comment S i C paper 180 grit Water 20 2 Repeated step t i l l surface was flat S iC paper 320 grit Water 40 2 Changed paper after 1 min i f murkiness not noticed Diamond suspension on 6um Metadi 40 3 Texmet 1000 cloth fluid Diamond suspension on Texmet 1000 cloth l u m Metadi fluid 40 3 Colloidal S i 0 2 suspension 0.05um 30 2 Washed with water for last 10 sees 4.4.2 E tch ing The chemical etchant selected had to adequately distinguish between the two predominant constituent phases namely a-Ali 2(Fe,Mn)3Si and AJ.6(Fe,Mn). From literature, two etchants were found to do this, 10% H 3 P 0 4 at 20\u00C2\u00B0C [16], and H 2 S 0 4 at 80\u00C2\u00B0C [12]. H3PO4 was tried and found to distinguish the two phases adequately for image analysis. Quantification o f dispersoids was also attempted, and H3PO4 was effective in a qualitative viewing o f dispersoids. 0.5 % H F , an etchant used in literature [18] was also tried but was unsuccessful. 44 Samples were etched with 10% H3PO4 for 8 minutes at room temperature to distinguish between the second phases (a-Ali2(Fe,Mn) 3 Si and Al 6 (Fe ,Mn)) . The a-Ali2(Fe,Mn)3Si etched considerably darker while Al6(Fe,Mn) was grey when viewed optically (Figure 4.7). The differentiation was confirmed by analysis using E D X / S E M (Figure 4.8) where only the darker particle showed a Si peak confirming it to be oc-A l i 2 ( F e , M n ) 3 S i . oc-Ali 2 (Fe,Mn) 3 Si M & S i A l 6 (Fe ,Mn) (etched black) (black) (etched grey) Figure 4.7 - As-cast microstructure (1000X) showing the various second phase particles. 45 17-Sep-19S8 14:41:37 Dark P a r t i c l e V e r t . 5000 counts Dlsp= 1 Quantex> fi 01 P r e s e t -E l a p s e d ^ 200 sees 200 sees J S I n rg B n ns-4- 0 .000 Range- 10.230 keV Ta I B T T 10.110 -> I n t e g r a l 8 - 306306 Particle etched black showing A l , S i , M n and Fe peaks. 17-Sep-1998 14:46:41 Greu P a r t i c l e V e r t - 5000 counts D i s p - 1 Quantex> \u00E2\u0080\u00A2 \u00E2\u0080\u00A2 01 P r e s e t -E l apsedc 286 sees 20B sees n re I T 4- 0 .000 Range-nm rs~ 10.H38 keV TT TO V3 10.110 -t I n t e g r a l 0 \u00C2\u00B0 338423 Particle etched grey showing A l , M n and Fe peaks. Figure 4.8 - E D X / S E M results confirming the etching difference between cx-A l i 2 ( F e , M n ) 3 S i and Alo cc-Ali2(Fe,Mn)3Si transformation was measured as percent a-Ali2(Fe,Mn)3Si of total second phase and plotted for a three hour hold at 580\u00C2\u00B0C for the various heat-up rates (Figure 5.2). Although heat-up rates do not have a significant effect on percentage transformation, there is a trend showing a higher a-phase at faster heat-up rates. Due to the high standard error during measurement, 95% confidence interval error bars suggest that no conclusive statement can be made about the increasing trend indicated on the graph. 53 60% 50% CD co cd 40% 30% 20% 37\u00C2\u00B0C/hr 47\u00C2\u00B0C/hr 70\u00C2\u00B0C/hr 112\u00C2\u00B0C/hr 279\u00C2\u00B0C/hr Figure 5.2 - Effect of heat-up rate on % a-Ali2(Fe,Mn)3Si after a 3 hour hold at 580\u00C2\u00B0C. 5.1.1.3 Evolution of Dispersoids Microstructures were taken from samples at a soak temperature of 580\u00C2\u00B0C after a three hour hold for each of the heat-up rates (see Appendix B for a list of pictures for each profile). Dispersoids (seen as the fine dots in the microstructure) formed from a faster heat-up rate were larger and fewer than ones formed from a slower heat-up rate profile (Figure 5.3). 54 20um b) Figure 5.3 - Microstructures of samples for 580\u00C2\u00B0C soak temperature after 3 hour hold for: a) 279\u00C2\u00B0C/hr and b) 47\u00C2\u00B0C/hr. 55 M n precipitates during heat-up in the form o f a-Ali2(Fe,Mn) 3 Si dispersoids with the M n precipitation kinetics bearing a direct relation to the dispersoid evolution [4, 19]. Since there has been no study on heat-up rate for A A 3xxx alloys, no results or theory was found in literature to confirm or explain these findings. Evaluating % M n in solid solution at specific temperatures during heat-up revealed that there was more precipitation in the 47\u00C2\u00B0C/hr heat-up profile than the 279\u00C2\u00B0C/hr profile (Figure 5.4). It was reasoned that due to the small time range during heat-up for the faster heat-up rate, the M n precipitation rates were lower causing less dispersoids to nucleate. This was confirmed by the smaller minima of M n in solid solution, as illustrated in Figure 5.1. The growth o f the dispersoids, however, continued during the soak period and the fewer dispersoids grew larger with the available M n in solid solution. 56 Figure 5.4 - Micro-structures of samples at 450\u00C2\u00B0C during heat-up for: a) 279\u00C2\u00B0C/hr and b) 47\u00C2\u00B0C/hr. 57 5.1.1.4 Removal of Microsegregation Electron Probe Micro Analysis ( E P M A ) scans carried out on samples at various temperatures during heat-up showed that M g and M n microsegregation was removed from the samples at different temperatures (Table 5.2) (for micrographs see Appendix C) . M g microsegregation also referred to as M g coring [33] is high in the as cast structure and is considerably reduced by the time the temperature reaches 500\u00C2\u00B0C for the 47\u00C2\u00B0C/hr profile and by 580\u00C2\u00B0C for the 279\u00C2\u00B0C/hr profile. Table 5.2 - M n and M g segregation results for 47\u00C2\u00B0C/hr and 279\u00C2\u00B0C/hr profile at various temperatures during heat-up. Condition Heat-up profile 47\u00C2\u00B0C/hr Heat-up profile 279\u00C2\u00B0C/hr M g M n M g M n As-cast H i g h Medium High Med ium 4 0 0 \u00C2\u00B0 C H i g h Medium H i g h Med ium 5 0 0 \u00C2\u00B0 C Medium Medium High Med ium 5 8 0 \u00C2\u00B0 C L o w L o w L o w M e d i u m 5 8 0 \u00C2\u00B0 C 3hr L o w L o w L o w L o w 58 5.1.2 Effect of Soak Temperature and Soak Time 5.1.2.1 Effect on M n in Solid Solution A s the soak temperature increases, the amount of M n in solid solution at the soak temperature increases for both a fast (279\u00C2\u00B0C/hr) and slow (47\u00C2\u00B0C/hr) heat-up rate (Figure 5.5). This is in tune with the literature since M n solubility increases with an increase in temperature [4, 26]. The effect o f soak time at 550\u00C2\u00B0C, 580\u00C2\u00B0C and 610\u00C2\u00B0C can also be estimated from Figure 5.5 for both heat-up rates. The results can be analyzed for two distinct regions: i) Between 0 and 0.5 hrs - most profiles show an increase in the M n i n solid solution especially for the higher temperature profiles (580\u00C2\u00B0C and 610\u00C2\u00B0C). This could be attributed to the fact that at 0 hr hold even though the samples have reached the soak temperature the M n re-dissolution kinetics are still catching up to reach the equilibrium amount o f M n in solid solution for that temperature. ii) Between 0.5 hrs and 10 hrs: The results for this range show a slight drop in % M n in solid solution with increasing soak time at higher soak temperatures (~600\u00C2\u00B0C) whereas a steady amount or even a slight rise can be seen at lower soak temperatures (~550\u00C2\u00B0C). The decrease in % M n in solid solution during the soak period at higher homogenization temperatures have also been noted by other researchers [12, 19, 21]. The reason for this could be that at higher soak temperatures other processes competing for M n become more favorable such as: 59 Al6(Fe,Mn) -> a-Ali2(Fe,Mn) 3Si: A higher % a - A l i 2 ( F e , M n ) 3 S i is reported at longer soak periods [4] (Figure 5.7) and there is a higher % M n in a-Ali2(Fe,Mn) 3 Si as compared to Al6(Fe,Mn), and % M n in constituent particles: A higher % M n is seen in both a - A l i 2 ( F e , M n ) 3 S i and Al6(Fe,Mn) at higher soak temperatures [12, 26, 27]. 60 \u00E2\u0080\u00A2 550\u00C2\u00B0C ^580\u00C2\u00B0C H610\u00C2\u00B0C 0.6 0 0.5 3 10 Soak time at temperature (hrs) b) Figure 5.5 - M n in solid solution at different soak temperatures at a) 47\u00C2\u00B0C/hr and b) 279\u00C2\u00B0C/hr. 61 5.1.2.2 Constituent Particles and Percent a-Ali2(Fe,Mn)3Si Particle coarsening occurs at high temperature hold during homogenization [10, 12, 24] with more coarsening at higher temperatures. However, in the present study, a significant coarsening was not observed in the particles. The particles at 610\u00C2\u00B0C, however, were seen to be more rounded than at 550\u00C2\u00B0C (Figure 5.8). With higher soak temperatures, the Al6(Fe,Mn) to a-Ali2(Fe,Mn)3Si transformation is also enhanced such that the percentage of a-Ali2(Fe,Mn) 3Si is higher for higher soak temperatures (Figure 5.6). This trend is seen at both slow and fast heat-up rates with the total %a-Ali 2 (Fe,Mn) 3 Si being higher at the faster heat-up rates. 147\u00C2\u00B0C/hr 3 hr hold \u00E2\u0080\u00A2 279\u00C2\u00B0C/hr 3 hr hold \u00E2\u0080\u00A2 As cast 22 550 580 Temperature (\u00C2\u00B0C) 610 Figure 5.6 - Effect of soak temperature on %a-Ali 2(Fe,Mn) 3Si after a soak period of three hours. 62 A s the soak time is increased the percentage a-phase increases and eventually levels off after a soak time of 3 hours is reached. The leveling was more pronounced for higher soak temperatures (Figure 5.7). This could be attributed to the unavailability of M n due to higher M n in solid solution at the end o f 10 hrs for the higher soak temperature (Figure 5.5). \u00C2\u00B0 5 5 0 \u00C2\u00B0 C ^ 5 8 0 \u00C2\u00B0 C S 6 1 0 \u00C2\u00B0 C 60% 0 3 10 Soak time at temperature (hrs) Figure 5.7 - Effect o f soak time on %oc-Ali2(Fe,Mn)3Si at various soak temperatures at a heat-up rate o f 47\u00C2\u00B0C/hr. 63 5.1.2.3 Evolut ion of Dispersoids The effect o f soak temperature on the evolution of dispersoids can be estimated qualitatively from the microstructure inspection o f samples from 550\u00C2\u00B0C, 580\u00C2\u00B0C and 610\u00C2\u00B0C soak temperature after three hour hold with a heat-up rate o f 47\u00C2\u00B0C/hr (Figure 5.8). The dispersoids at 550\u00C2\u00B0C are smaller and more in number, the number decreasing and the size increasing with an increase i n the soak temperature t i l l at 610\u00C2\u00B0C there are very few dispersoids observed, and the ones present are very coarse. There are two main mechanisms that contribute to the observed results: i) Growth o f dispersoids is a diffusion-based process and as temperature rises from 550\u00C2\u00B0 to 610\u00C2\u00B0C, the rates o f this process increase rapidly. ii) Increased solubility o f M n at higher temperatures lead to a dissolution of the smaller dispersoids into the matrix thereby explaining the low number o f dispersoids seen i n the sample which had been homogenized at 610\u00C2\u00B0C. 64 *. 20um a) 550\u00C2\u00B0C 20u: b ) 5 8 0 \u00C2\u00B0 C 65 / ' ' f 20um c ) 6 1 0 \u00C2\u00B0 C Figure 5.8 - Microstructures (1000X) of samples after three hour hold with a heat-up rate of 47\u00C2\u00B0C/hr for soak temperatures of: a) 550\u00C2\u00B0C, b) 580\u00C2\u00B0C and c) 610\u00C2\u00B0C. 5.1.2.4 Cool Down to Rolling Temperature The effect of soak temperature is seen to translate to a difference in % M n in solid solution even when the samples were cooled down to a similar temperature of 500\u00C2\u00B0C (Figure 5.9), with the samples which had been soaked at a higher temperature exhibiting a higher level of M n in solid solution. 66 e - 550\u00C2\u00B0C (46.5\u00C2\u00B0C/hr) 580\u00C2\u00B0C (46.5\u00C2\u00B0C/hr) - * ^ 6 1 0 \u00C2\u00B0 C (46.5\u00C2\u00B0C/hr) 10 12 14 16 18 Time (hrs) Figure 5.9 - M n in solid solution for homogenization profiles of 550\u00C2\u00B0C, 580\u00C2\u00B0C and 610\u00C2\u00B0C for three hour hold followed by cool down to 500\u00C2\u00B0C. During cool down, very little change takes place in the morphology or size of the constituent particles, the precipitating Mn comes out enlarging the pre-existing dispersoids (Figure 5.10). Hence, the higher number of dispersoids in the 550\u00C2\u00B0C sample (Figure 5.8) are also carried forward at the end of the cool down to 500\u00C2\u00B0C. 67 b) 580\u00C2\u00B0 68 / * \ 1 % < % 20um c) 610\u00C2\u00B0C Figure 5.10 - Microstructures (1000X) of samples after a cool down to 500\u00C2\u00B0C for soak temperatures of: a) 550\u00C2\u00B0C, b) 580\u00C2\u00B0C and c) 610\u00C2\u00B0C. 69 5.2 Influence of Homogenization Microstructure on Texture/Microstructure after Hot-Rolling The size and shape o f the constituent and dispersoid phases formed in the ingot during homogenization influence the size distribution of the second phases in the final gauge sheet [4, 12, 37, 45, 46] as well as the recrystallization kinetics and texture formation after hot rolling. During hot rolling, the eutectic networks o f constituents break down and depending on the rolling temperature, recrystallization o f the grains initiates during/after rolling [11, 12, 48]. The morphology and size distribution o f the constituent particles (a-Ali2(Fe,Mn)3Si and Al6(Fe,Mn)) affects the cube texture evolution due to particle-stimulated nucleation (PSN) during recrystallization after hot-rolling, die-cleaning during ironing i n the bodymaker operation, and tear-off tendency [4, 12, 45, 48, 47]. The dispersoids, on the other hand, affect the texture by retarding recrystallization [45, 46,48]. The morphology o f the constituent particles in terms o f the percent a -Ali2(Fe,Mn) 3Si phase determines the die-cleaning efficiency o f the alloy during drawing. The constituent a-Ali2(Te,Mn)3Si phase is relatively harder (Vickers micro-hardness o f 900-950 Hv) than the A l 6 (Fe ,Mn) phase, which has hardness o f 700-750 H v [4, 16, 22]. Hence a certain fraction (greater than 40%) o f a-Ali2(Fe,Mn)3Si constituent particles is desirable for prevention o f galling (the seizure of metal on the working surfaces o f the dies observed when the constituent particles are not hard enough). The majority o f the 70 manufacturers in the industry control the percent a-Ali2(Fe,Mn)3Si phase fraction between 50 and 60% [23]. To minimize earing during the can drawing operation, adequate amounts o f cube texture (0/90\u00C2\u00B0 ears) are needed after hot rolling. However, large constituent particles can cause deformation zones to form around them during hot rolling, and during subsequent annealing, these deformed zones become nucleation sites for recrystallization. P S N results in random texture, which reduces the effective cube content, and hence decreases the cube texture in the final gauge sheet [45, 5]. Hence, it is not favorable to have a large number of constituent particles. M n retained i n solid solution retards recrystallization during annealing after hot rolling [12]. It is desired to achieve a totally recrystallized microstructure before the final cold rol l step to obtain a sufficient amount o f cube texture [39]. Hence, it is desirable to have as low a % M n in solid solution after hot rolling as possible. The number and size distribution of dispersoids that evolve during homogenization are important parameters in determining the amount o f cube texture formed. These dispersoids exert a retarding force on the low/high angle grain boundary, which can affect the kinetics of recovery, recrystallization and grain growth. This effect is known as Zener drag [45] and can be expressed as: Pz = (3Fv*y)/2r (6) Where F v is the volume fraction o f randomly distributed spherical particles o f radius r, and y is the grain boundary energy reported to be 324 mJ/m 2 [49] for aluminium. From equation (6) it can be deduced that the Zener drag is a function o f volume fraction 71 (Fv) and size r (radius) o f the dispersoids such that a large volume fraction of small dispersoids increases Zener drag. 72 6. Summary and Conclusions During the manufacturing o f aluminum beverage cans from the ingot through the cold rolled sheet, it is important to control the microstructure and texture evolution as they w i l l control the final mechanical properties and performance o f the alloy during the can making operation. Since the homogenization profile the ingot undergoes, w i l l influence the starting microstructure prior to hot rolling, it is critical to characterize and quantify the influence o f various homogenization parameters on the microstructure evolution and determine the \"optimum\" homogenized microstructure. 6.1 Summary The present research was primarily concerned with characterizing and quantifying the effect o f homogenization process parameters on the microstructural evolution o f an AA3104 aluminum alloy. The following observations can be made based on the study: i) Heat-up rate to the soak temperature affected the amount o f M n precipitating out and re-dissolving in solid solution during heat up. However, the total amount o f M n that precipitated out o f solid solution at the end o f three hours at the soak temperature was almost identical for each heat-up rate. The variation in M n during heat-up affected the size and shape o f the evolving dispersoids; the dispersoids became fewer and appeared coarser as the heat-up rate increased possibly due to the lower amount of M n precipitation during heat-up. The shape and size o f the constituent particles was not 73 affected by the heat-up rate, but there was a slight increase in the percent a -Ali 2(Fe,Mn)3Si seen for the faster heat-up rates. This may have occurred because o f the extra M n that was available due to fewer dispersoids. A n increase i n heat-up rate also caused a delay in reduction o f M g and M n microsegregation across the cell . i i) A s soak temperature was increased, the amount of M n in solid solution increased for each heat-up rate. The rise in M n in solid solution was accompanied in parallel by a drop in dispersoids. Higher soak temperatures also resulted in a higher %oc-Ali 2(Fe,Mn)3Si particles. i i i) Increasing soak times also increased the %cx-Ali2(Fe,Mn) 3Si initially but started tapering off for longer soak times o f 3 hrs as the equilibrium amount o f %cx-Al] 2(Fe,Mn)3Si was approached for the soak temperature. Higher soak temperature profiles tapered faster due to the lower amount o f M n in solid solution. M n in solid solution was affected inversely with increasing soak times. A drop was seen in % M h i n solid solution with increasing soak times between 0.5 to 3 hrs. Between 0 and 0.5 hrs the M n level still increased i n all profiles due to lag i n M n dissolution kinetics to reach the equilibrium amount o f M n for that soak temperature. Dispersoids on the other hand, were qualitatively assessed to have coarsened slightly without a noticeable increase in their number. Since the microstructure after the homogenization process w i l l influence the microstructure and sheet properties that develop during and after hot rolling, it is important to understand the influence o f key microstructure features on the both the microstructure evolution during hot rolling as well as the final material properties. A summary of desired microstructure characteristics after homogenization include 74 \u00E2\u0080\u00A2 a fair amount o f %a-Al]2(Fe,Mn)3Si preferably over 50% and definitely not below 40% (to minimize galling), \u00E2\u0080\u00A2 minimum coarsening o f the constituent particles so as to obtain a minimum number o f large constituent particles after hot rolling (particles greater than l - 2um promote P S N during recrystallization [12]), \u00E2\u0080\u00A2 a small volume fraction of dispersoids which are as large as possible so as to minimize the Zener drag effect which can inhibit recrystallization and thereby lower the total amount of cube texture after hot rolling, and \u00E2\u0080\u00A2 low % M n in solid solution after hot rolling so that the temperature at which recrystallization occurs is not too high and the material can self-anneal during coiling. 6.2 Conclusions From the results, the following conclusions can be made for a suitable homogenization profile: \u00E2\u0080\u00A2 A faster heat-up rate is more favorable in terms o f the formation o f fewer dispersoids and a slightly higher volume fraction of a-Ali2(Fe,Mn)3Si. \u00E2\u0080\u00A2 A soak temperature o f 610\u00C2\u00B0C is prefered as it achieves the required volume fraction of a-Ali2(Te,Mn)3Si and leads to a lower volume fraction o f dispersoids with a larger size. The disadvantage o f a higher soak temperature is that there w i l l be a higher level of M n in solid solution during hot rolling which could inhibit recrystallization during 75 annealing after hot rolling. However, this could be lowered before annealing by adjusting the rolling temperatures. \u00E2\u0080\u00A2 Soak times at temperature allowed the % a - A l i 2 ( F e , M n ) 3 S i to increase (-15% from 0 to 3 hours), however, given a high soak temperature, high soak times could increase the size o f the constituent particles promoting P S N . Beyond three hours there did not appear to be much change in the microstructure even for the material heated using a fast heat-up rate Based on this study, it appears that for an AA3104 alloy used for canbody stock the optimum homogenization profile includes: a fast heat-up rate to a high homogenization temperature (580\u00C2\u00B0C - 610\u00C2\u00B0C) with moderate soak times (0.5-3 hrs). Other conclusions that can be drawn from the study are: i) There is a significant difference in the microstructure obtained between the industrial standard heat-up rate range (45\u00C2\u00B0C/hr - 70\u00C2\u00B0C/hr) profile and very fast heat-up rates. Hence, i n order to apply laboratory based homogenization results, it is critical that similar heat-up rates to industrial conditions be applied to the samples (i.e., samples which have been heated using salt baths would give very different microstructures than the ones industrially heated even i f they have been held at 580\u00C2\u00B0C for 3 hrs.) ii) Even though the heat-up rate had a significant impact on the microstructure evolution, within the current industrial heat-up rate range (i.e., soaking pits (~50\u00C2\u00B0C/hr) V s . walking beam furnaces (~70\u00C2\u00B0C/hr)), there is not a significant difference i n the microstructure obtained 76 6.3 Future Work Based on this study, the following are some of the topics that may be considered for future work: i) Quantification o f dispersoids was not possible with optical image analysis or S E M . Hence, T E M analysis of multiple fields within samples in each profile is recommended to complete quantification and numerical analysis of microstructure evolution during homogenization. ii) One o f the interesting observations in this study was the lowering o f M n i n solid solution with time at higher temperatures. To explain the results and validate the theory proposed extensive fundamental research is required studying the thermodynamic relationship between M n in solid solution, Al<5(Fe,Mn) -> a-Ali2(Fe,Mn) 3 Si transformation and M n in Ai6(Fe,Mn) and a-Aln(Fe,Mn)3Si. There is very little literature available covering the thermodynamics involved in the A A 3 1 0 4 alloys. To the best o f the author's knowledge the A l - M g - M n - F e - S i phase diagram is not yet known; it is under investigation in the European C O S T 507 (Co-operation in Science and Technology) action [50, 51, 52]. i i i) Finally, modeling the microstructure evolution would be a useful undertaking especially i f the thermodynamic data is available. The model would predict the microstructure characteristics given homogenization process parameters. The model could be tested against the results obtained in this study. 77 Bibliography 1. A L C O A website; Sources: Alcoa , Canadean, C C I , K A A L , I M E S , \" http://www.alcoaxorri7frarneset.asp ?page=%2Finvestor%2Ffinancial%2Findex%2Eas u\" 2. Sanders E.R. , Lege D J . ; Hartman T .L . , \"Aluminum rigid container sheet for the packaging industry \", Aluminium. 65 (9), 1989, 941-947. 3. Nes E . , Hutchinson W . B . , Proc. 10th Riso Symposium. 1989, Eds. Bilde-Sorensen et al, 233, Roskilde, Denmark, 4. Westerman E.J . , \"Sil icon: a vital alloying element in aluminum beverage can body stock\", Aluminum Al loys for Packaging. Eds. J .G. Morris , H . D . Merchant, E .J . Westerman and P . L . Morris , The Minerals, Metals & Materials Society, Warrendale, P A , U S A , 1993,1-16. 5. Bolingbroke R . K . , Creed E . , Marshall G J . , Ricks R . A . , \"The effect o f Composition and Microstructure on the Recrystallization o f AA3004\" , A luminum Al loys for Packaging, eds. J .G. Morris , H . D . Merchant, E . J . Westerman and P . L . Morris , (Warrendale, P A : T M S , 1993), 215-224. 6. Hirsh, J.R., \"Texture Evolution During Hot Rol l ing in A l Al loys\" , Hot Deformation o f Aluminum A l l o y s . Eds. Langdon T .G . , Merchant H . D . , Morr is J .G. , and Zaidi M . A . , The Minerals Metals & Materials Society, 1991, 379-388. 7. Kyang Y . and Morris J .G. , \"The effect of structure on the mechanical behavior and stretch formability o f constitutionally dynamic 3000 series aluminum alloys\", Materials Science and Engineering. 1977, 59-74. 8. Hutchinson W . B . , Oscarsson A and. Karlsson A , \"Control o f microstructure and earing behaviour i n aluminum alloy A A 3 0 0 4 hot bands\", Materials Science and Technology, vol . 5, 1989, 1118. 9. Regan, P .C . , \"Recent Advances in Aluminum Strip Casting and Continuous Rol l ing Technology\u00E2\u0080\u0094Implications for Aluminum Can Body Sheet Production\", Light Metal Age. February 1992, 58-62. 78 10. Merchant, H . D . ; Kattamis, T .Z. ; Scharf: \"Homogenization o f Aluminum Al loys\" , Homogenization and Annealing o f Aluminum and Copper Al loys , eds. H . D . Merchant, J. Crane, and E . H . Chia, (Warrendale, P A : T M S , 1988), 1-52. 11. D ing S .X . , Ren B . & Morris J .G., \"Influence o f Initial Structure on Hot Rol l ing and Recrystallization Textures in A A 3 0 0 4 Aluminum A l l o y \" , Advances in Hot Deformation Textures and Microstructures. Eds. Jonas J.J. Bieler T .R. and Bowman K . J . , 1994, 281-288. 12. Kamat R . G . , \"AA3104 Can-body stock ingot: Characterization and homogenization\", J . O . M . vol . 48, N o . 6, 1996, 34-38. 13. M i k i I., Kosgue H . & Nagahama K . , \" Supersaturation and Decomposition o f A l - F e Al loys During Solidification\", J. Japan Inst. Light Metals. 1975, vo l 25, 1-12. 14. Chen L . , Morris J .G .& Das S.K., \"The effect o f cooling rate during casting on the structure and mechanical property behavior o f A A 3 0 0 4 aluminum alloy\", Continuous Casting o f Non-Ferrous Metals and Al loys Eds. Mercant H . D . , et al, The Minerals, Metals & Materials Society, Warrendale, P A , U S A , 1989, 269-284. 15. Mondolfo L . F . , Aluminum Al loys . Structure and Properties. Butterworth & Co. Ltd. , London, 1976. 16. Bolingbroke R . K . , Marshall G.J . & Ricks R . A . , \"Microstructural Development During Preheating o f AA3004\" , The 3 r d International Conference on Aluminum Al loys . Trondeim, Norway, 1992, 285-290. 17. Backerud L . , K r o l E . , & Tamminen J., Solidification Characteristics o f Aluminum Volume 1: Wrought Al loys . Skan Aluminum, Sweden, 1986. 18. Hatch J . E . , Aluminum: Properties and Physical Metallurgy. American Society for Metals, Ohio, U S A , 1984, 61-65, 6-9,140 & 204-206. 19. Rouns T . N . , \"Composition and Preheating on the Dispersoid and Insoluble Constituent Particle Evolution in 3xxx Al loys\" , Aluminum alloys for Packaging HI. Eds. Das S.K. , (Warrendale, P A : T M S , 1993), 3-20. 20. Palmer, S .L. ; L i , Z . : \"The Single-Step Preheating o f A A 3 1 0 4 Can-Body Stock Ingots,\" J O M (6) 1996, 30-33. 21. Tromberg E . , Dons A . L . and Arnberg L . , \"Investigation o f the Al6(Fe,Mn) -> cc-Ali2(Fe,Mn)3Si phase during homogenization o f AA3003 and A A 3 0 0 4 aluminum alloys\", 3 r d International Conference on Aluminum alloys. Trondheim, Norway, 1992, 270-275. 79 22. Watanabe H . , Ohori K . & Taheuchi Y . , \"Phase Change in 3004 Base A l loys at Elevated Temperatures\", Aluminum 60.1984, 310-321. 23. Wang, X . ; Kamat, R . G . : \" A Technique to Measure Intermetallic Size Distribution in Aluminum Can Body Stock,\" Aluminum Al loys for Packaging II. eds. J .G. Morris , S .K. Das and H.S . Goodrich, (Warrendale, P A : T M S , 1996), 209-222. 24. Kattamis, T .Z . ; Merchant, H . D . ; Skolianos, S.; Scharf, G . : \"Homogenization and Coarsening in Cast 3004 Aluminum A l l o y , \" Aluminum 65 (4) (1989), 367-376 25. Kattamis, T.Z. ; Merchant, H . D . ; Scharf, G . : \"Influence o f Homogenization on Intermetallic Coarsening in Aluminum A l l o y 3004,\" Homogenization and Annealing of Aluminum and Copper Al loys , ed., H . D . Merchant, J . Crane and E . H . Chia, (Warrendale, P A : T M S , 1988), 117-135. 26. T .C . Sun, \"The effect o f preheating on A A 3 1 0 4 aluminum alloy ingot structure and particulate composition\", Aluminum Al loys for Packaging. Eds. J .G. Morris , H . D . Merchant, E.J . Westerman and P . L . Morris, The Minerals, Metals & Materials Society, Warrendale, P A , U S A , 1993, 31-46. 27. Karlson A , Oscarsson A , Lehtinen B & Hutchinson W . B . , \"Influence o f Homogenization on Structure and Earing in Aluminum alloy 3004\", Homogenization and Annealing o f Aluminum and Copper Al loys , ed., H . D . Merchant, J . Crane and E . H . Chia, (Warrendale, P A : T M S , 1988), 99-116. 28. Anyalebechi, P . N . ; Rouns, T . N . ; Sanders, Jr., R . E . , \"Microstructural Analysis o f Second Phases Developed During Casting and Preheating o f 3xxx Aluminum alloys\", R A S E L M ' 9 1 . Tokyo, Japan, 1991, 923-928. 29. Theler J.J., & Furrer P., \"Influence o f Fe on the Course o f Precipitation in A l - M n Al loys\" , Aluminium. 50, vol 7,1974,467-472. 30. Goel D . B . , Furrer P. & Warlimont H . , \" Recrystallization and Precipitation i n A l - M n Al loys\" , Aluminium 50. vol 8, 1974, 511-521. 31. Usui E . , Takashi I. & Noboru S., \"Influence o f M n and M g Additions on Hot Deformation o f Aluminum and Aluminum alloys\", Z . Metallkde.. 1986, vo l 77, 179-187. 32. Marshall G.J. , \"Microstructural Control During Processing o f Aluminum Canning Al loys\" , Materials Science Forum. 1996, vols. 217-222, 19-30. 80 33. Sanders R . E . , Baufmann S.F. & Stumpf H . C . , \"Wrought Non-Heat-Treatable Aluminum Al loys\" , Aluminum Al loys - Contemporary Research and Applications. 1988, Eds Vasudevan A . K . & Doherty R .D . , 75-83. 34. Goodrich H.S. , \" A Model for the Precipitation/Dissolution o f M n During Commercial Homogenization o f Aluminum A l l o y 3104,\" Aluminum Al loys for Packaging, eds J .G. Morris , H . D . Merchant, E .J . Westerman and P . L . Morris , (Warrendale, P A : T M S , 1993), 47-60. 35. Private communication with Alcan International. 36. Jackson A . & Sheppard T., \"Structural Modifications occurring during the Homogenization of some 7xxx alloys\". E T 96: Profiles of Change. 6l International Aluminum Extrusion Technology Seminar. 1996, 541-550. 37. Kattamis, T .Z . ; Merchant, H . D . ; Scharf, G . : \"Influence o f Homogenization on Intermetallic Coarsening in Aluminum A l l o y 3004,\" Homogenization and Annealing o f Aluminum and Copper Al loys , ed., H . D . Merchant, J. Crane and E . H . Chia, (Warrendale, P A : T M S , 1988), 117-135. 38. E . Tromberg, A . L . Dons and L . Arnberg, \"Investigation of the Al6(Fe,Mn) -> oc-Ali2(Fe,Mn)3Si phase during homogenization o f AA3003 and A A 3 0 0 4 aluminum alloys\", 3 r d International Conference on Aluminum alloys. Trondheim, Norway, 1992, 270-275. 39. L i , Z . ; L i , C . X . ; Morris , J .G. : \"Precipitate Behavior o f A A 3004 Aluminum A l l o y \" , Aluminum Al loys for Packaging, eds. J .G. Morris, H . D . Merchant, E . J . Westerman and P . L . Morris , (Warrendale, P A : T M S , 1993), 61-69. 40. Suni J.P. & Shuey R.T. , \"Modeling Dispersoid and Constituent Particle Evolution in 3xxx Al loys\" , Aluminum alloys for Packaging III. Eds. Das S .K. , (Warrendale, P A : T M S , 1993), 21-37. 41. Suni J.P., Shuey R.T . & Doherty R .D . , \"Dispersoid Model ing i n 3xxx Al loys\" , Aluminum Al loys for Packaging JJ. eds. J .G. Morris , S .K. Das and H.S . Goodrich, (Warrendale, P A : T M S , 1996), 145-159. 42. Anyalebechi P . N . , Rouns T . N , Sanders, Jr. R .E . , \"Effects o f Cool ing Rate and Grain Refining on Constituent Phase Particle Size In As-Cast 3004 A l l o y , \" Light Metals 1991. ed. E . L . R o y (Warrendale, P A : T M S , 1990), 821-850. 43. Ti lak R . V . , . Morris J .G, \"Studies o f the effect o f thermomechanical treatments on the supersaturation content of strip-cast aluminum alloy 3004,\" Materials Science and Engineering. 73 (1985), 139-150. 81 44. Tromberg E . , Dons A . D . , Amber L . , \"Investigation o f the Al6(Fe,Mn) -> cx-Ali2(Fe,Mn)3Si phase transformation during homogenization o f A A 3 0 0 3 and A A 3 0 0 4 Aluminum Al loys\" , The 3 r d International Conference o f Alumin ium Al loys . 1992, 270-275. 45. Humphreys F.J . , Hatherly. M , Recrystallization and Related Annealing Phenomena. Elseview Science Inc., N e w York 1995. 46. Es-Said O.S., Morris J .G. & Merchant H .D . , \"The Influence o f Particles on the Recrystallization of Aluminum Al loys\" , Homogenization and Annealing of Aluminum and Copper Al loys , eds. H . D . Merchant, J. Crane, and E . H . Chia, (Warrendale, P A : T M S , 1988), 183-208. 47. Sun, T .C. : \"Surface and Metallurgy Effects in Wal l Ironing o f A A 3104,\" Aluminum Al loys for Packaging. Eds. J .G. Morris , H . D . Merchant, E . J . Westerman and P . L . Morris , (Warrendale, P A : T M S , 1993), 183-192. 48. L i Z . , L i C . X . , Morris J. G . , \" A Comparison of Recrystallization behavior of D . C . Cast and Continous Strip Cast A A 3 0 0 4 Aluminum Al loy\" , Aluminum Al loys for Packaging, eds. J .G. Morris , H . D . Merchant, E.J . Westerman and P . L . Morris , (Warrendale, P A : T M S , 1993), 299. 49. Murr L . E . , Interfacial Phenomena in metals and Al loys . Reading, P . A . Addison-Wesley, 1975, 131. 50. Sigl i C , Vichery H . & Grange B . , \"Computer Assisted Metallurgy for Packaging Al loys\" , \", Aluminum alloys for Packaging ITI. Eds. Das S .K. , (Warrendale, P A : T M S , 1993), 189-197. 51. K o l b y P., S ig l i C . & Simensen C.J . , \"Solubility Limit of M n and S i in A l - M n - s i at 550\u00C2\u00B0C\", Aluminum alloys: Their Physical and Mechanical Properties ( I C A A 4 ) , Ed . T . N . Sanders & E . A . Starke, 1994, 513-519. 52. Fries, S.G., Jantzen, T., \"Compilation of C A L P H A D formation enthalpy data- Binary Intermetallic Compounds in the C O S T 507 Gibbsian Database\", Thermochimica Acta, vo l . 314, no. 1-2, (1998), 23-32. 82 Appendix A Temperature profile of aluminum samples in original and retrofitted (i.e. with aluminum tube) furnace. 83 b) Figure A l - Temperature variations between furnace (580\u00C2\u00B0C) and sample o f original furnace at a) two sample positions in the furnace, and b) on a raised platform near the furnace thermocouple. 84 Figure Al - Temperature variations between furnace (580\u00C2\u00B0C) and sample temperature i n the retrofitted furnace. 85 Appendix B Microstructures o f samples heated to 580\u00C2\u00B0C using different heat-up rates and held for three hours: 86 87 88 e) 279\u00C2\u00B0C/hr Figure B l - Microstructures o f samples at 580\u00C2\u00B0C soak temperature after 3 hr hold for a) 37\u00C2\u00B0C/hr, b) 47\u00C2\u00B0C/hr, c) 70\u00C2\u00B0C/hr, d)112\u00C2\u00B0C/hr, and e) 279\u00C2\u00B0C/hr. 89 Appendix C E P M A line scans (Mg and Mn) at various temperatures during heat-up to 580\u00C2\u00B0C using two different heat-up rates (47\u00C2\u00B0C/hr and 279\u00C2\u00B0C/hr). 90 Figure C l - E P M A line scans ( M g and Mn) for the as-cast microstructure. 91 c) 580\u00C2\u00B0C 0 hr hold d) 5 8 0 \u00C2\u00B0 C 3 h r h o l d .^^^^^^H^HBMl Figure C2 - E P M A line scans ( M g and Mn) for 47\u00C2\u00B0C/hr profile at a) 400\u00C2\u00B0C, b) 500\u00C2\u00B0C, 580\u00C2\u00B0C (0 hrs), and d) 580\u00C2\u00B0C (3hrs). 93 a) 400\u00C2\u00B0C b)500\u00C2\u00B0C 10 um i 1 Figure C3 - E P M A line scans (Mg and Mn) for 279\u00C2\u00B0C/hr profile at a) 400\u00C2\u00B0C, b) 500\u00C2\u00B0C, 580\u00C2\u00B0C (0 hrs), and d) 580\u00C2\u00B0C (3hrs). 95 "@en . "Thesis/Dissertation"@en . "1999-05"@en . "10.14288/1.0078699"@en . "eng"@en . "Materials Engineering"@en . "Vancouver : University of British Columbia Library"@en . "University of British Columbia"@en . "For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use."@en . "Graduate"@en . "Effect of homogenization on the microstructural development in a d.c. cast aa3104 aluminum alloy used for canbody stock"@en . "Text"@en . "http://hdl.handle.net/2429/9389"@en .