sign in sign up
<<
Home , Core (manufacturing), Metalworking, Riser (casting)

The Effect of Processing Parameters and Composition on the Microstructure Formation and Quality of DC Cast Alloys

Majed M. R. Jaradeh

Doctoral Thesis In Materials Processing Department of Material Science and Engineering School of Industrial Engineering and Management KTH-Royal Institute of Technology SE-10044 Stockholm, Sweden

Department of Engineering, Physics and Mathematics Mid Sweden University SE-851 70 Sundsvall, Sweden

Akademisk avhandling som med tillstånd av Kungliga Tekniska Högskolan (KTH) framlägges till offentlig granskning för avläggande av Teknologie Doktorsexamen, Todsdagen den 14 Dec 2006, kl 10.15, sal F3, Lindstedtsvägen 26, KTH, Stockholm.

i

The Effect of Processing Parameters and Alloy Composition on the Microstructure Formation and Quality of DC Cast Aluminium Alloys

Majed Jaradeh Department of Material Science, Royal Institute of Technology SE-100 44 Stockholm, Sweden Abstract The objective of this research is to increase the understanding of the solidification behaviour of some industrially important wrought aluminium alloys. The investigation methods range from direct investigations of as-cast to laboratory-scale techniques in which is simulated. The methods span from directional solidification at different cooling rates to more fundamental and controlled techniques such as DTA and DSC. The microstructure characteristics of the have been investigated by optical and Scanning Electron microscopy. tests were used to evaluate the mechanical properties. The effects of adding alloying elements to 3XXX and 6XXX aluminium alloys have been studied with special focus on the effects of Zn, Cu, Si and Ti. These elements influence the strength and corrosion properties, which are important for the performance of final components of these alloys. Solidification studies of 0-5wt% Zn additions to 3003 alloys showed that the most important effect on the microstructure was noticed at 2.5 wt% Zn, where the structure was fine, and the hardness had a maximum. Si addition to a level of about 2% gave a finer structure, having a relatively large fraction of eutectic structure, however, it also gave a long solidification interval. The addition of small amounts of Cu, 0.35 and 1.0 wt%, showed a beneficial effect on the hardness. Differences have been observed in the ingot surface microstructures of 6xxx billets with different Mg and Si ratios. Excess Si compositions showed a coarser grain structure and more precipitations with possible negative implications for surface defect formation during DC casting. The comparison of alloys of different Ti content showed that the addition of to a level of about 0.15 wt% gave a coarser grain structure than alloys with a normal Ti content for grain refinement, i.e. < 0.02 wt%, although a better corrosion resistance can be obtained at higher Ti contents. The larger grain size results in crack sensitivity during DC casting. A macroscopic technique was developed, based on a NaOH solution, and used in inclusion assessment along DC cast billets. Good quantitative data with respect to the size and spatial distribution of inclusions were obtained. The results from studied billets reveal a decreasing number of inclusions going from bottom to top, and the presence of a -shaped distribution of a large number of small defects in the beginning of the casting. The present study shows how composition modifications, i.e. additions of certain amounts of alloying elements to the 3xxx and 6xxx Al alloys, significantly change the microstructures of the materials, its castability, and consequently its mechanical properties.

Keywords: Aluminium wrought alloys, AA3xxx and 6xxx, Direct Casting, Unidirectional Solidification, Bridgman technique, Process parameters, Microstructure, Composition modification, Thermal analysis, Inclusions, Metallographical investigations.

ii

Acknowledgments This work has been carried out at Mid Sweden University, at the Department of Engineering, Physics and Mathematics. I would like to express my sincere appreciation to my professors, Torbjörn Carlberg, the main thesis supervisor and Hasse Fredriksson (co-supervisor), for their scientific guidance, encouragement and kindness throughout my graduate studies and the completion of this dissertation. The great inspiration and confidence that they have given me make me believe that I can continue my academic dream. I am also grateful to the staff members of the Sundsvall branch of the Department of Engineering, Physics and Mathematics, at Mid Sweden University and the Department of Materials processing at the Royal Institute of Technology in Stockholm, especially Advenit Makaya. The financial support from EU, Objective 1, has been of great importance for my investigation, and my co- workers and I would also like to thank Kubikenborg Aluminium AB for their support and provision of the sample materials. Last, I would like to thank my family and friends who have motivated and supported me through the completion of this dissertation.

Sweden, December 2006 Majed M. R. Jaradeh

iii

List of

The thesis includes the following supplements: Supplement 1 Analysis of solidification in a Bridgman furnace as a simulation of DC casting of aluminium alloy slabs. Majed Jaradeh and Torbjörn Carlberg. Accepted for publication in Materials Science and Technology, 2006 Supplement 2 Solidification studies of 6xxx alloys with different Mg and Si contents. Hu Jin, Majed Jaradeh and Torbjörn Carlberg. Proceedings, Light . TMS, San Francisco, California, USA, Feb. 2005, pp. 1039-1044. Supplement 3 Directional solidification of type AA3003 alloys with addition of Zn and Si. Majed Jaradeh and Torbjörn Carlberg. Proceedings, Solidification of Aluminum Alloys. TMS, Charlotte, North Carolina, USA. March 2004, pp. 53-62. Supplement 4 DTA and DSC Studies of aluminium 3003 alloys with Zn and Cu additions. Majed Jaradeh and Torbjörn Carlberg. Submitted to Metallurgical and Materials Transactions A. 2006 Supplement 5 Effect of titanium additions on the microstructure of DC-cast aluminium alloys. Majed Jaradeh and Torbjörn Carlberg. Materials Science and Engineering A. Vol. 413-414, 15 December 2005, pp. 277-282. Supplement 6 Method Developed for Quantitative Assessment of Inclusions in Aluminium Billets. Majed Jaradeh and Torbjörn Carlberg. To be published in Proceedings, Light Metals, TMS. Orlando, Florida, USA. Feb. 2007. Supplement 7 Solidification Studies of Automotive Heat Exchanger Materials. Torbjörn Carlberg, Majed Jaradeh and H. Kamgou Kamga. JOM. Vol. 58, no. 11. Nov, 2006.pp 56-61

iv

TABLE OF CONTENTS

Abstract...... ii Acknowledgments ...... iii List of papers...... iv TABLE OF CONTENTS ...... v Chapter 1 ...... 6 1 Introduction and Background of Thesis ...... 6 1.1 Direct Chill Casting of Aluminium Alloys...... 6 1.2 Aluminium Alloys and Effects of Alloying Elements...... 9 1.3 Aluminium Melts Cleanliness and Treatment ...... 11 1.4 The Scope of the Thesis...... 14 1.5 References...... 15 Chapter 2 ...... 19 2 Experimental methods...... 19 2.1 Materials...... 19 2.2 Experimental Techniques...... 19 Chapter 3 ...... 28 3 Results and discussion...... 28 3.1 Results of thermal conditions in the Bridgman furnace...... 28 3.2 Results of different Mg and Si contents on solidification of 6XXX alloys... 32 3.3 Results of additions of Zn and Si on the solidification of 3003 alloys...... 35 3.4 Results of DTA and DSC studies of 3003 alloys containing Zn and Cu...... 37 3.5 Results of additions of excess titanium...... 41 3.6 Results of assessment of inclusions in aluminium billets...... 43 4 General Conclusions of Thesis...... 46

v

Chapter 1

1 Introduction and Background of Thesis Aluminium and its alloys have excellent properties: they are light in weight, corrosion- resistant, good electrical and thermal conductors and relatively easy to produce. They find their main uses in the transportation, , packaging and electrical industry. Aluminium can be produced via two different routes: primary aluminium production from , and recycling from process scrap and used aluminium products. The production of primary aluminium consists of three steps: bauxite , alumina production and electrolysis. The bauxite is refined into alumina (Al2O3) by the Bayer process and in reduction plants primary aluminium is extracted from alumina by the Hall-Heroult electrolytic process [1-3]. For all aluminium alloy components, the casting process plays an important role in controlling the properties of the final products. Semicontinuous direct-chill (DC) casting is currently the most widely used process in the production of aluminium alloys [4,5]. This casting technique is commonly used to produce ingots that undergo subsequent processing. Ingots of rectangular cross-section are commonly rolled to sheet, foil and plate; cylindrical ingots are extruded to form rods, bars and wires [6]. The casting method was developed independently by VAW (Germany) and Alcoa (USA), based on the patent of Roth and W. T. Ennor in the early 1930s and 1940s respectively [7,8]. A brief detail of the process description and DC machine casting is reported below.

1.1 Direct Chill Casting of Aluminium Alloys In the semicontinuous direct chill (DC) casting process, molten aluminium is poured from the furnace through a troughing system into a mould table containing moulds with a cross-sectional shape of the desired ingot, which may be rectangular for , round for billets. A starting head, mounted to a hydraulic ram, forms a sort of false bottom to the mould. When the in the mould reaches a definite height, the bottom block is lowered at a constant rate equivalent to the metal flow. A skin solidifies on the first contact of the melt with the mould wall, a shell to hold the liquid metal. The direct chill is applied just below the mould exit by spraying cooling water onto the ingot to cool the surface and further solidify the metal. The water-cooled mould and the water sprayed on the hot ingot cause a very rapid solidification. Casting speeds normally vary with the ingot size and alloy type and are typically in the 1-3 mm/s range (1. 10-3 to 3. 10-3 ms-1) [4]. The cooling rate varies in relation to the mould walls and is approximately 0.5 oC/s in the ingot centre to approximately 20 oC/s near the ingot surface. The thermal gradient is 0.3-5.0 K/mm, and the pouring temperature of the aluminium melt alloys is near 690-725 oC. The water flow rate is ramped up in the beginning but constant during steady state castings. The DC casting apparatus of rectangular cross-sections used for rolling ingots, shown in Fig.1.1, is composed of a water-cooled rectangular mould, made of a or aluminium alloy, with a or ingot stool resting on a casting table.

Figure 1.1: Schematic of direct-chill (DC) casting machine. Recent developments in the DC casting include the electromagnetic casting, the hot- top system and the air slip process. The air slip mould was developed specifically for production of billets. The process involves the formation of a gas barrier between the molten metal and the graphite-lined mould [5,9]. The benefits gained by industries from using the mentioned new variants of the process over the basic DC cast process are the improvement of the surface microstructure and minimization of surface defects. Although the DC casting technology has been used for a long time, several problems are still unsolved, particularly in relation to the demand for increased productivity and the reduction of defects (e.g. ingot cracking, shrinkage, porosity, inclusions, etc.). From a productivity standpoint, the casthouse would preferably produce the largest possible ingots at the highest possible casting speeds, but reality intervenes, with ingot size and aspect ratio (ingot thickness/ingot width) as well as casting speed restricted by the occurrence of casting defects, such as cracks and hot tears [10-14]. The quality and the commercial applications of the aluminium cast products depend, to an important extent, on their microstructure. The microstructure variables of interest in aluminium alloy castings are Dendrite Arm Spacing (DAS), grain size and type, amount and morphology of intermetallic particles. Each of these variables determines the mechanical properties of the part. The values of the variables in the as-cast part depend on the solidification characteristics during casting, such as cooling and solidification rates, the alloy composition, additives such as modifiers and grain refiners, etc. [15,16]. A characteristic feature of the process of ingot casting of aluminium and its alloys by a semicontinuous method is that, the solidification parameters, which influence the solidified microstructure are not constant across the thickness of the ingot, i.e. each elementary volume of the cast metal is solidified with different rate. Its magnitude depends on the parameters used in the casting process, (e.g. the temperature of the melt, the casting speed, the amount of cooling water), on the position in relation to the mould walls and the axis of the ingot being cast. It means that this phenomenon is critical in the case of the casting of large ingots of rectangular cross-section, which is used as the charge to make rolled products. Wan et al., [17] measured the solidification rates of DC cast ingots by in situ thermocouple tests. The solidification rate during

cooling was observed to be faster on the surface of the ingot and was observed to decrease towards the centre of the ingot. This change in the solidification rate during cooling causes the change in microstructure, which results in a variation of mechanical properties at different locations of the ingot. In direct chill (DC) casting of aluminium and its alloys, it is a well-established practice to add a grain refiner to promote the formation of a fine, equiaxed grain structure, to improve many properties of the as-cast metal and to facilitate subsequent processing [18-20]. Grain refinement allows a more uniform distribution of alloying elements, and it reduces porosity and the risk of defects especially cracks in the cast structures. Grain refinement by inoculation is commonly achieved by the addition of a master alloy to the melt, typically in the form of a rod or a waffle. Various types of master alloys can be used. The most common master alloys are based on the Al-Ti-B system, (e.g. Al-5% Ti-1% B). The amount that needs to be added is calculated on the basis of a concentration of 0.02 wt.% Ti (with respect to the weight of the molten metal to be treated). Typical addition levels of master alloys are between 0.5 and 2 Kg ton-1 of melt in DC ingots [21]. Much work has been directed to understand the mechanisms of grain refinements [21-25]. Of all the mechanisms proposed so far, duplex nucleation theory is the most recent, first proposed by Mohanty and Gruzleski [26] and further developed by Schumacher and Greer [27-29]. This theory explains the mechanism of grain refinement of Al alloys by the addition of AlTiB master alloys, which suggests that TiB2 particles need to be coated with a thin layer of Al3Ti to be effective nucleant particles. A second important effect is the grain growth restriction, i.e. when alloying elements restrict the growth of aluminium dendrites and form an undercooled melt zone [30]. Ti is the most potent growth restriction element among all other alloying element in aluminium alloys. Both the partitioning of solute elements and the added nucleant particles are now thought to affect the grain refinement process. Minimizing casting defects and improving the productivity of the DC-casting processes through a better understanding of the mechanisms involved during the casting and the solidification are the focal points of efforts worldwide. However, the investigation of an industrial scale ingot is expensive, time-consuming and tedious, so it is becoming increasingly important to study solidification process and microstructure formation within the laboratory. A number of small-scale casting techniques have been developed to minimize the need for the casting of full size ingots during research on DC-casting phenomena and the development of new alloys. These experimental techniques [21,31] either mimic solidification conditions within the ingot (e.g. Direct-chill, DC simulator) [32,33] or allow solidification to occur under controlled conditions, so that the fundamentals of solidification can be investigated (e.g. Bridgman technique). These two directional solidification techniques were used in the course of this work to develop our understanding of the microstructure formation, and they are fully described in chapter 2, section 2.2.1. Both techniques have also been used previously to investigate the formation of intermetallic-phase selection [34-37], and fundamental theory associated with grain refinements in wrought aluminium alloys [38,39]. Other types of experiments available to trace changes in the solidification and precipitation of aluminium alloys involve the use of thermal analysis techniques. Two

types of thermal analysis methods have been used in this research: differential thermal analysis (DTA) and differential scanning calorimetry (DSC). These devices are well documented in literature and widely used to calculate material properties such as latent heat and specific heat and to determine the reaction kinetics during phase transformations [40-42]. Although the development of DTA and DSC instruments made it possible to determine values faster, more conveniently and at a lower cost, these techniques can be less accurate to investigate the solidification of DC-casting conditions and to predict commercial DC-cast microstructure. The solidification of a DC-cast ingot is complex, and normally it does not occur under equilibrium conditions. The techniques mentioned were designed to determine equilibrium properties and are limited to small sample volumes. Further details of the experiments and the analysis methods involving the DTA and DSC techniques applied in this research are detailed in chapter 2, section 2.2.2. The aluminium can be tailored to the applications required by modifying the composition of its alloys, i.e. through the addition of different kinds and quantities of alloying elements. Many metals can be added to aluminium, according to the product demand. Manganese (Mn), silicon (Si), magnesium (Mg), copper (Cu), (Fe), (Zn) and titanium (Ti) are among the most commonly used alloying elements. The different alloying elements have different functions because of their different solid solubilities in aluminium. Some elements are used as solid-solution strengtheners, others form desirable intermetallic compounds and therefore induce different effects [43]. In the next section the purposes of adding different alloying elements are reviewed as well as their effects on the microstructures and properties important for their applications.

1.2 Aluminium Alloys and Effects of Alloying Elements In general, pure aluminium has a very low yield strength of about 7 Mpa, i.e. it is too soft and ductile to be used in structural and industrial applications, and therefore it needs to be hardened and strengthened. One common way to strengthen a metal is to add alloying elements that cause both precipitation and solution , which results in a high strength-to-weight ratio. The amount of useful alloying elements, which can be incorporated, is controlled by their solid solubility and the cooling rate achieved during casting. Since there are no allotropic phase transformations in aluminium, much of the control of the microstructure and properties relies on precipitation reactions. The solubility of solute in the matrix is therefore of importance. Up to 70 wt.% of zinc can dissolve in aluminium, followed by 17.4 wt.% magnesium, 5.65 wt.% copper, 1.82wt% manganese and 1.65wt% silicon [44]. Certain properties of aluminium are largely independent of the content of alloying elements and the processing chain. Examples are the modulus of elasticity of common wrought alloys, which is important for many applications, E = 70 GPa, the density, S = 2700 kg/m3, and the thermal expansion coefficient, 24 x 10-6 /K. Most other properties, for example strength and corrosion resistance, are very sensitive to the composition and microstructure of the material [45].

Basically, two types of alloys have been studied for this thesis; the AA3xxx (Mn rich) and AA6xxx (Al-Mg-Si) series. The aluminium alloy 3003 type is a preferred alloy for heat exchanger materials in car radiators or air conditioning units [46, 47]. This type of alloys has the advantage of being non-heat treatable, i.e. they do not require any post- ageing treatment. In this type of alloys, Mn is always present at slightly above 1%. It improves the strength, both by solution and dispersoid hardening, at room temperature and at brazing temperature, and it also improves the corrosion resistance and the castability. Common modifications or enhancements in the mechanical properties of the alloy, i.e. the improvement of strength and enhanced corrosion resistance are made through the additions of certain alloying elements, such as Zn, Si and Cu. Additions of different amounts of zinc to the 3003 alloy have been studied, as this element has a large solubility in aluminium and thus, it can improve the strength, by solution hardening both at room temperature and at brazing temperature. Another interesting aspect of Zn alloying is that it opens up the possibility to give corrosion protection to selected parts of the heat exchangers. Zinc additions lower the corrosion potential, and can thus be used in part that can act as sacrificial anodes, while other parts are being protected [48]. Similarly, copper has an electrochemical effect on the material and alters the corrosion resistance [49]. In addition, Cu contributes to the solid solution strengthening of the material. Silicon additions are also of interest due to the low solubility of this element and the potential for the precipitation of virtually pure silicon and of intermetallic particles of silicon within the matrix, which increases the hardness and abrasion resistance of the metal. Si also has a low density (2.34 g cm- 3), which may be an advantage to keep the overall weight of the component low. In addition, Si also increases the amount of the AlFeSi phases, which has a positive effect on mechanical properties. In brazing applications, the most important use of Si is in clad materials, i.e. as a filling metal during brazing, which has been rolled onto the material. Here, the compositions are from 7 to 11% in order to lower the melting point sufficiently to facilitate the brazing. The AA6xxx series alloys are very popular in the extrusion industry and are generally known as high-speed extrusion alloys. This type of alloys is increasingly attractive to industries such as transportation because of their high specific strength. Alloys in the 6xxx series contain silicon and magnesium approximately in the proportions required to form magnesium silicide (Mg2Si). The formation of (Mg,Si) particles is facilitated when the Mg and Si contents are high. The proper ratio for Mg2Si is Mg/Si= 1.73, but this is impossible to achieve with ordinary operating tolerances; thus, most alloys have either magnesium or silicon excess. Magnesium excess to better corrosion resistance but lower strength and formability [50]; silicon excess produces higher strength without loss of formability and [50,51], but there is some tendency to intergranular corrosion [51]. In the thesis, the influence of alloy compositions is studied, particularly the balance of Mg and Si concentrations on the solidification microstructure and ingot surface defects. Another alloy addition, which is common in many aluminium alloys, is the titanium. Ti may be added as an alloying element for its grain refinement effect or as a grain refining agent of Al-Ti-B master alloys. The maximum content of Ti that can usefully be incorporated in aluminium alloy ingots, cast by normal techniques, is of the order of <0.02 wt.%, depending on the alloy and grain refinement technique. On the other

hand, if Ti is added in excess of 0.1%, it is believed that improved corrosion resistance can be obtained [52].

1.3 Aluminium Melts Cleanliness and Treatment One of the most serious problems in aluminium casting is that cast structures are usually associated with inclusions. The detrimental effects of the presence of inclusions are well known. The presence of inclusions can results in increased gas porosity, poor surface finish, pinholes in thin foils, decreased fluidity, reduced corrosion resistance, loss of mechanical properties, especially fatigue and , poor machinability and excessive wear, etc [53-57]. Although much work has been reported on the study of inclusions, the problem of measuring metal cleanliness through a sensitive quantitative method still exists. The main problem concerning determining the inclusions size distribution is that they are heterogeneously distributed in the melt, and that their concentrations are extremely low. In the following, the main types of inclusions, the removal techniques during the metal casting of aluminium, and the methods of assessment are reviewed. Inclusion is defined as any exogenous solid or liquid phase present in molten aluminium. The types of inclusions found in aluminium are numerous and complex, and their formation depend primarily on the alloy composition, the melt treatment practice, and the casting technology. There is three important indices when defining melt cleanliness: the composition of the inclusions, the particle size of the inclusions, and the size distribution of the inclusions. 1.3.1 Inclusions: Types and Sources

The inclusions commonly found in cast aluminium alloys include Alumina (Al2O3 particles and films), titanium diborides (TiB2 particles), magnesia (MgO particle), and magnesium aluminates (MgAl2O4 particles and films) [58]. Inclusions in aluminium alloys may be from two basic sources [58-60]: the inclusions occurring from outside the melt, called exogenous, and include refractory particles resulting from the degeneration of crucibles or furnace linings. They may appear in several forms with sizes from microns to millimetres. The indigenous inclusions, i.e. those, which arise from within the melt, may be of several types and are usually only microns thick (but may extend to several millimetres). The size at which an inclusion becomes potentially troublesome of course varies with the end-use application. It has been shown in [61] that the production of quality casting necessitates that the inclusion contents should be of the order of several volume parts per billion at the most and that the particle size of the population is less than 50μm. In general, for critical applications, any inclusion larger than 10-20 in diameters is considered to be deleterious, although even smaller inclusions can cause problems if present in sufficient quantity [54, 57,62]. In the metal casting industry, the treatment and cleanliness of the molten metal is of great importance. A number of commercially accepted melt treatment techniques are being used by aluminium to remove and separate inclusions from the molten aluminium alloy prior to casting. They include various methods of fluxing [63], degassing, [64,65] and filtration [66-68] in the furnaces and in the gating system.

1.3.2 The Formation of Inclusion The formation of inclusions during aluminium casting is inevitable, since aluminium is very reactive and readily forms a stable oxide on the liquid surface by reacting with air or moisture. The rate of oxidation increases with temperature and is substantially greater in molten than in solid aluminium. While the oxide coatings or oxide skins that initially form on the surface of molten aluminium are highly protective and self-limiting, any agitation or turbulence in the treatment and handling of the melt increases the risk of oxide entrainment and immediate reformation of additional oxides. Oxide concentration can increase when alloying additions are stirred into the melt, when reactive elements and compounds are immersed, when the metal is transferred from the conventional metal treatment system to the casting stations, when the metal is drawn for pouring, and when the metal is poured into the mould cavity. The problem of removing aluminium oxides depends on the almost identical density of the inclusions and the melt. For this reason, the inclusions do not rise to the surface and tend to remain inside the metal. However, Al2O3 inclusions become buoyant as a result of absorbed gases, and rise to the melt surface [69].

Magnesium oxide (MgO) and magnesium aluminates (MgAl2O4) inclusions occur in most aluminium alloy containing magnesium. Since MgO and MgAl2O4 have lower free formation energy than Al2O3, such inclusions tend to form preferentially, particularly when the Mg content of the alloy is greater than 0.005 wt.% [70,71]. The addition of AlTiB grain refiner material often results in 1-3 micron titanium diboride (TiB2) particles. Such small particles individually do not cause any problem, but they often form 10-50 microns sized agglomerates. These large agglomerates can be a cause of concern in a casting. This is because TiB2 particles are relatively much harder than aluminium. They can be detrimental to critical applications such as thin sheet, micron foils, fine wire and applications requiring special surface finishes [72]. According to [57], TiB2 is considered to be a type of inclusions, whereas TiAl3, which is soluble under normally encountered conditions, is not. 1.3.3 The Assessment of Inclusion The assessment of metal cleanliness is an integral part of any casting operation. The optimum technique would provide the assessment of three inclusion parameters: size, distribution, and composition. Several techniques of melt cleanliness are available to assess the cleanliness of molten aluminum alloys [57,73,74]. Among the commonly used techniques, the following can be mentioned: The LiMCA (Liquid Metal Cleanliness Analyzer) technique [75,76] is an on-line melt cleanliness determination, utilizing electrical resistivity to detect inclusions in the 20- 300 micron range, depending on the orifice size of the sampling tube used. This technique is based on Electrical Sensing Zone (ESZ), as shown in Fig 1.2. A constant electrical current is applied between the and flows through the liquid metal in the small orifice of the tube. The presence of an inclusion in the melt flowing through the orifice increases the electrical resistance measured at the orifice. The signals are amplified and their amplitude is measured digitally. The advantage of this

technique is that the impact of furnace preparation, alloying practice, feedstock mix, settling time and similar parameters on melt cleanliness, is easily determined. The technique characterizes the cleanliness of a melt at time intervals of the order of 1 minute, i.e. it delivers an immediate response. The disadvantage of this technique is that it also detects gas bubbles from degassing as particles, which complicates the analysis, and it cannot provide information on the , shape, or the physical state of the inclusions. The small orifice (~ 300μm) may become plugged or blocked, if the metal contains relatively large inclusions. The cost of the measurement instrument is expensive.

Figure 1.2: Schematic of the LiMCA principle. The PoDFA (Porous Disc Filtration Analysis) [59,77] is an off-line technique, developed by Alcan International Limited, in which a fixed volume of metal is forced through a filter, concentrating the inclusions on the surface of the filter, as is shown in Fig 1.3.

Figure 1.3: Schematic of the PoDFA Analysis 57. Subsequent metallographic evaluation is needed to measure the amount and type of inclusions present. While it is inexpensive to collect samples, this method requires

time and expertise to acquire the measurements. It provides information about the type of inclusions and their relative proportions. However, in the PoDFA technique, since the assessment involves concentration of the inclusions, some information is lost, such as the distribution of the inclusions. Since the concentration also includes some degree of agglomeration, the information about size and shape might also be lost. Furthermore, the inclusion size is limited by the filter-porous disk size, i.e. small particles cannot be detected by the PoDFA method. Ultrasonic test is a non-destructive technique, and it transmits an ultrasonic wave through the metal sample. Inclusions are identified through sound damping [78]. Considerable work has been carried out, using , to detect inclusions in molten aluminium alloys [78,79]. The disadvantage of this technique is that it remains expensive and can only detect inclusions greater than 100μm. Other less common and less widely used methods to measure the inclusion content of metals are neutron activation analysis (used to measure oxygen concentration), arc- emissions spectrometry [80], and selective dissolution techniques etc. [81]. However, these methods are also relatively expensive tedious and time-consuming. The acid dissolution test, that is, extracting out the non-metallic inclusion by the complete dissolution of the aluminium matrix in mixed acids and the collection of the insoluble [82,83]. This test is limited by the difficulty of dissolving anything more than a relatively small sample, and the fact that some of the compounds and contaminants are also soluble in the solutions used. When it comes to the assessments of inclusions in the solidified product, a small number of inclusion assessment methods are known, and there is a lack of literature data. A typical method used is the light microscopy examination. Obtaining results by this method is, however, time-consuming, and at the time that the results are received, the metal produced may already have left the house. Due to the low inclusion content, and the high tendency of inclusion to segregate into a nonuniform distribution, the direct microscopic analysis of the surface of metal samples is not practical, since a very large area would have to be observed for quantitative assessment. In the present thesis a simpler test for inclusion assessment in solidified Al alloy castings has therefore been developed, using a NaOH macroetching [84,85]. One of the reasons for the use of the macroetching technique is related to the random nature of the occurrence of inclusions in Al alloys. Another reason could be attributed to the difficulties with the detection and the quantitative assessment of inclusions.

1.4 The Scope of the Thesis This study has been performed to provide an understating of the solidification behaviour and the microstructure formation generated during the solidification of DC casting of some industrially important wrought aluminium alloys. The intentions are alloys development and to increase our understanding of the effects of the casting and solidification parameters, the amount of certain alloying addition, and the dimensions of casting products on the castability and resulting microstructures of the 3000 and 6000 series alloys in particular. An additional aim of the project was to simulate the conditions (e.g. cooling rate) in the large DC casting process, using directional solidification experiments with a Bridgman-type furnace and a DC simulator.

The project results will provide general guidelines to improve the quality and the performance of the final products, to reduce casting difficulties and defects, to reduce scrap, and to contribute on optimizing DC process parameters, with regard to castability and ingot dimensions. The thesis starts with an analysis of the thermal field and the solidification process in samples during growth in a Bridgman furnace, in order to provide information about the growth conditions, such as the temperature gradient, and the position and velocity of the solid/liquid interface during directional growth [Suppl. 1]. The thesis also covers the study of the solidification behaviour of aluminium wrought alloys and the effect of excess additions of certain alloying elements on the microstructure formation [Suppl. 2-5]. The equilibrium solidification behaviour of the heat exchanger alloys of the AA3003 type containing an excess of Zn and Cu were studied by a series of thermal analyses (DTA, and DSC). The results of these experiments are presented in [Suppl. 4]. The last part of the thesis deals with quality issues, namely the location and content of inclusions in DC-cast billets [Suppl. 6], and an application of the studied materials to automotive production [Suppl. 7].

1.5 References 1. Stobart, Patrick D. Centenary of the Hall & Heroult Processes, 1886-1986. London: International Primary Aluminium Institute, 138 pp, 1986. 2. Altenpohl, Dietrich G. Aluminum: Technology, Applications, and Environment: A profile of a Modern Metal. Book: Aluminium from Within. 6th Edition. Publ: Warrendale PA 15086, USA; Minerals, Metals & Materials Society (TMS); Nov.1997. Ed: J.G.Kaufman, S.K.Das. 3. Glenn Switkes. Foiling the aluminum indstry. A toolkit for communitles, Activists,Consumers, workers. Published by International Rivers Network, August 2005. 4. Grandfield, J.F. and McGlade, P.T. DC casting of aluminium: process behaviour and technology. Materials Forum, Australia. Vol. 20, pp. 29-51. 1996. 5. Grandfield, J.F. DC casting of aluminium: a short review of process development. Fifth Australian Asian Pacific Conference. Aluminium Cast House Technology: Theory & Practice; Queensland; Australia; June 1997. Pp. 231-243. 6. King, F. Aluminium and its Alloys. Ellis Horwood, Market Cross House, Cooper Street, Chichester, West Sussex PO19 1EB, UK, pp. 313.1987 7. Roth, W. Deutsches Reichs Patent Nr. 974203. 1936 8. W. T. Ennor, US patent 2301027, Nov. 3, 1942. 9. Faunce, J P; Wagstaff, F E; Shaw, H. New Casting Method for Improving Billet Quality. Light Metals, Los Angeles, Calif; U.S.A; 27 Feb.-1 Mar. 1984. pp. 1145-1158. 10. Jacoby, J. E. Direct chill casting defects. Fifth Australian Asian Pacific Conference. Aluminium Cast House Technology: Theory & Practice; Queensland; Australia; June 1997. pp. 245-251. 11. Suyitno, Eskin D.G., Savran V.I., and Katgerman. L Effects of Alloy Composition and Casting Speed on Structure Formation and Hot Tearing during Direct-Chill Casting of Al-Cu Alloys. Metallurgical and Materials Transactions A. Vol. 35A, no. 11, pp. 3551-3561A. Nov. 2004 12. Eskin D G, Savran V. I, and Katgerman L. Effects of Melt Temperature and Casting Speed on the Structure and Defect Formation during Direct-Chill Casting of an Al-Cu Alloy. Metallurgical and Materials Transactions A. Vol. 36A, no. 9, pp. 2561A. Sept. 2005. 13. Eskin, D G; Zuidema, J; Savran, V I; Katgerman, L. Structure formation and macrosegregation under different process conditions during DC casting. Materials Science and Engineering A. Vol. 384, no. 1-2, pp. 232-244. Oct. 2004. 14. Eskin D.G., Zuidema (Jr) J, Savran V.I. and Katgerman L. Effect of Process Parameters on Structure Formation during Direct-Chill Casting of an Al-Cu Alloy. Materials Forum. Vol. 28, pp. 487-493. 2004.

15. Buxmann, K and , E. Solidification Conditions and Microstructure in Continuously Cast Aluminum. J. Met. Vol. 34, no. 4, pp. 28-34. Apr. 1982. 16. Anyalebechi, P N. Effects of alloying elements and solidification conditions on secondary dendrite arm spacing in aluminum alloys. EPD Congress 2004 as held at the 2004 TMS Annual Meeting; Charlotte, NC; USA; Mar. 2004. pp 217-233. 17. Wan, J. Lu, H-M, Chang, K-M, and Harris, J. As-Cast Mechanical Properties of High Strength Aluminum Alloy. Light Metals; San Antonio, TX; USA; Feb. 1998. pp. 1065-1070. 18. Bunn, A. M, Schumacher, P, Kearns M. A, Boothroyd, C. B, and Greer, A. L. Grain refinement by Al-Ti-B alloys in aluminium melts: a study of the mechanisms of poisoning by zirconium. Materials Science and Technology (UK). Vol. 15, no. 10, pp. 1115-1123. Oct. 1999 19. Cooper, P and Barber, A. Review of the latest developments and best use of grain refiners. 2nd International Melt Quality Workshop, Prague, Czech Republic, 15-17th October 2003. 20. Kashyap, K T; Chandrashekar, T. Effects and mechanisms of grain refinement in aluminium alloys. Bulletin of Materials Science (). Vol. 24, no. 4, pp. 345-353. Aug. 200. 21. Quested, T.E. (2004): Ph.D. Thesis. “Solidification of Inoculated Aluminium Alloys”. Department of materials science and , University of Cambridge, Cambridge UK. 22. Kearns, M.A, Thistlethwaite S.R. and Cooper P. S. Recent Advances in understanding the mechanism of aluminium Grain refinement by TiBAl master alloys. Light Metals; Anaheim, California; USA; Feb. 1996. pp. 713-720. 23. Pearson, J and Cooper, P.S. A review of the basics of grain refining. Sixth Australian Asian Pacific Conference on Aluminum Cast House Technology; Sydney; Australia; July 1999. pp. 109-118. 24. Easton M, StJohn D. (1999a): Metall Mat Trans A 1999; 30A: 1613. 25. Easton M, StJohn D. (1999b): Metall Mat Trans A 1999; 30A: 1625. 26. Mohanty P.S, Gruzleski, J.E. Mechanism of grain refinement in aluminum. Acta Metall. Mater. Vol. 43, no. 5, pp. 2001-2012. May 1995. 27. Schumacher, P., Greer, A. L. (1994a). Heterogeneously nucleated alpha -Al in amorphous aluminium alloys. Materials Science and Engineering A (Switzerland). Vol. A178, no. 1-2, pp. 309-313. Apr. 1994 28. Schumacher, P., Greer, A. L. (1994b): Mater. Sci. Eng. A, 182, 1335. 29. Schumacher, P., Greer, A. L., Worth, J., Evans, P. V., Kearns, M. A., Fisher, P., Green, A. H. New studies of nucleation mechanisms in aluminium alloys: implications for grain refinement practice. Materials Science and Technology (UK). Vol. 14, no. 5, pp. 394-404. May 1998. 30. Johnsson, M and Backerud, L. The influence of composition on equiaxed crystal growth mechanisms and grain size in aluminum alloys. Zeitschrift fur Metallkunde (Germany). Vol. 87, no. 3, pp. 216-220. Mar. 1996. 31. Meredith, M.W, Worthm J, Brown, J.M. and Hamerton, R. G. Understanding DC-cast microstructures using laboratory-based solidification techniques. Light Metals 2003 as held at the 132nd TMS Annual Meeting; San Diego, CA; USA; Mar. 2003. pp. 1111-1118. 32. Yang X., Pekgüleryüz M.Ö., and Bouchard M. Modelling of Solidification of AL Alloys in a Laboratory Scale DC-Simulator. Light Metals 2000 as held at the 129th TMS Annual Meeting; Nashville, TN; USA; Mar. 2000. pp. 591-596. 33. Heiberg G. Gandin Ch.-A., Goerner H., and Arnberg L. Metallurgical and Materials Transactions A. Vol. 35A, no. 9, pp. 2981-2991A. Sept. 2004 34. Maggs, S. J., Cochrane, R. F., Flood, S. C., Evans, P. V. The effect of trace element of intermetallic phase selection in simulated dc castings. Light Metals 1995, Las Vegas NV, USA; Las Vegas NV; USA; Feb. 1995. pp. 1039-1047. 35. Maggs, S. J. (1996): Ph.D. Thesis, University of Leeds, UK. 36. Meredith, M. W., Greer, A. L., Evans, P. V., Hamerton, R. G. Light Metals; San Diego, CA; USA; 28 Feb.-4 Mar. 1999. pp. 811-817. 37. Meredith, M. W. (1999): Ph.D. Thesis, University of Cambridge, UK. 38. Quested, T.E., and Greer, A. L. Modelling of microstructural evolution in Bridgman specimens. Solidification of Aluminum Alloys as held at the 2004 TMS Annual Meeting; Charlotte, NC; USA; Mar. 2004. pp. 43-52.

39. Greer, A.L., Quested, T.E., and Spalding, J.E. (2002): Modelling of grain refinement in directional solidification. Light Metals 2002 as held at the 131st TMS Annual Meeting; Seattle, WA; USA; Feb. 2002. pp. 687-694. 40. Speyer, R.F. Thermal Analysis of Materials. Marcel Dekker, Inc. (USA), pp. 297, 1994. 41. Tamminen, J. “Thermal Analysis for Investigation of Solidification Mechanisms in Metals and Alloys”, Ph.D. Thesis, Univ. Stockholm, Sweden, 1988. 42. Starink M.J. Analysis of aluminium based alloys by calorimetry: quantitative analysis of reactions and reaction kinetics. International Materials Reviews. Vol. 49, no. 3-4, pp. 191-226. June-Aug. 2004 43. Stefanescu, Doru M; Ruxanda, Roxana. Solidification Structures of Aluminum Alloys. Materials Park, OH: ASM International, 2004. Pp. 107-115. 44. Polmear, I.J., Light Alloys- metallurgy of the light metals. London, 1989: Edward Arnold. 45. Altenpohl, D. Aluminum Viewed from Within. Aluminium-Verlag, 1982. ISBN 3-87017-138-3. 46. Karlsson, A. Braze Clad aluminium materials for automotive heat exchangers some recent developments. INCAL '98: International Conference on Aluminium; New Delhi; India; Feb. 1998. pp. 189-199. 47. Gray, Alan. The growth of aluminium in automotive heat exchangers. Aluminium. Vol. 81, no. 3, pp. 197-201. Mar. 2005 48. Takemoto,T; Okamoto, I. Investigation on the Corrosion of Brazed Aluminium. Kei Kinzoku Yosetsu. J. Light Met. Weld. Constr. Vol. 23, no. 1, pp. 2-13. Jan. 1985. 49. Itagaki, T; Toma, K. Corrosion Resistance of Al--Mn--Cu Alloys. 76th Conference of the Institute of Light Metals; Osaka; Japan; May 1989. pp. 125-126. 50. Davis, J.R. Corrosion of Aluminum and Aluminum Alloys. ASM International, Member/Customer Service Center, Materials Park, OH 44073-0002, USA, 1999. pp 313. 51. Hirth, S.M., Marshall, G J, Court, S A, Lioyd, D J. Effects of Si on the aging behaviour and formability of aluminium alloys based on AA6016. Materials Science and Engineering A (Switzerland). Vol. 319-321, pp. 452-456. Dec. 2001 52. Shoji, M; Yamauchi, S. Influence of Titanium Additive on the Corrosion Resistance of Al--Mn-- Mg--Si--Cu Based Alloys. 76th Conference of the Japan Institute of Light Metals; Osaka; Japan; May 1989. pp. 127-128. 53. Lessiter, M J, and Rasmussen W.M. To pour or not to pour - The dilemma of assessing your aluminum melt's cleanliness. Modern casting, February 1996. 45-48. 54. Lessiter, M J. Understanding Inclusions in Aluminum Castings. Modern Casting (USA). Vol. 83, no. 1, pp. 29-31. Jan. 1993 55. Eklund, J.E. On the effect of impurities on the solidification and mechanical behaviour of primary and secondary commercial purity aluminium and aluminium alloys-PhD thesis. Helsinki University of technology, Espoo, Finland. 1991. 135P. 56. Eckert, C E; Miller, R E; Apelian, D; Mutharasan, R. Molten Aluminum Filtration: Fundamentals and Models. Light Metals 1984; Los Angeles, Calif ; U.S.A ; 27 Feb.-1 Mar. 1984. pp. 1281-1304. 57. Doutre, D, Gariepy B, Martin J. P and Dube, G. Aluminum Cleanliness Monitoring: Methods and Applications in Process Development and Quality Control. Light Metals 1985; New York, New York; USA; Feb. 1985. pp. 1179-1195. 58. Simensen, C.J and Berg, G. A Survey of Inclusions In Aluminum. Aluminium. Vol. 56, no. 5, pp. 335-338. May 1980 59. Apelian, D. How clean is the metal you cast? The issue of assessment; a status report. 3rd International Conference on Molten Aluminum Processing, Florida; USA; Nov. 1992. pp. 1-15. 60. Eckert, C.E. The origin and identification of inclusions in foundry alloys. 3rd International Conference on Molten Aluminum Processing; Orlando, Florida; USA; Nov. 1992. pp. 17-50. 61. Apelian, D and Mutharasan, R. Filtration: a melt Refining Method. J. Met. Vol. 32, no. 9, pp. 14- 18. Sept. 1980 62. Gallo, R. Correlating Inclusion Sizes with Aluminum Casting. Modern Casting. Vol. 94, no. 6, pp. 27-30. June 2004 63. Gallo, R. Development, Evaluation, and Application of Granular and Powder Fluxes in Transfer Ladles, Crucible, and Reverberatory Furnaces. 6th International AFS Conference on Molten Aluminium Processing; Lake Buena Vista, FL; USA; Nov. 2001. 15 pp. 2002.

64. Martins, L. C. B., and Sigworth, G. K. Inclusion Removal by Flotation and Stirring. 2nd International Conference on Molten Aluminum Processing; Florida; Nov. 1989. pp. 16.1-16.28. 65. Groteke, D.E. Influence of SNIF Treatment on Characteristics of Aluminum Foundry Alloys. Transactions of the American Foundrymen's Society, Vol. 93; Pittsburgh, Pennsylvania; USA; 29 Apr.-3 May 1985. pp. 953-960. 66. Schmahl, J.R. Aubrey L. S. and Martins, L.C.B. Recent Advances in Molten Aluminum Filtration Technology. 4th International Conference on Molten Aluminium Processing; Orlando, FL; USA; Nov. 1995. pp. 71-102. 67. Groteke, D.E. The Reduction of Inclusions in Aluminum by Filtration. Mod. Cast. Vol. 73, no. 4, pp. 25-27. Apr. 1983 68. Neff, D. V. Continuous, "re-usable" filtraton systems for aluminum foundries. 4th International Conference on Molten Aluminium Processing; Orlando, FL; USA; Nov. 1995. pp. 121-137. 69. Crepeau, P N. Molten aluminum contamination: gas, inclusions and . Modern Casting (USA). Vol. 87, no. 7, pp. 39-41. July 1997 70. Campbell, J.C. Castings 2nd Edition (Book). Oxford, Butterworth Heinemann, 2003. 335 pp. 71. Fuoco, R., Correa, E.R., Bastos, M., and Escudero, L.S., “Characterization of Some Types of Oxide Inclusions In Aluminum Alloy Castings”, AFS Transactions, v. 107, pp. 287-294 (1999). 72. Whitehead, A. J., Cooper, P. S., and McCarthy, R. W. An Evaluation of Metal Cleanliness and Grain Refinement of 5182 Aluminum Alloy DC Cast Ingot Using Al-3%Ti-0.15%C and Al-3%Ti- 1%B Grain Refiners. Light Metals 1999; San Diego, CA; USA; 28 Feb.-4 Mar. 1999. pp. 763-772. 73. Dasgupta, S., Apelian, D. Interaction of initial melt cleanliness, casting process and product quality: cleanliness requirements fit for a specific use. 5th International AFS Conference on Molten Aluminium Processing; Orlando, FL; USA; Nov. 1998. pp. 234-258. 74. Simard, A., Prouix, J; Paguin, D; Samuel, F.H. Capability study of the Improved PREFIL- Footprinter to measure liquid Aluminium Cleanliness and Comparison with Different Techniques. 6th International AFS Conference on Molten Aluminium Processing; Lake Buena Vista, FL; USA; Nov. 2001. 20 pp. 2002 75. Guthrie, R.I.L. and D.A. Doutre, D.A. On-Line Measurements of Inclusions in Liquid Melts. Refining and Alloying of Liquid Aluminium and Ferro-Alloys. The Norwegian Institute of Technology; Trondheim; Norway; Aug. 1985. pp. 145-163. 76. Ware, TN; Cooksey, M; Couper, M J. Measurement of the performance of in-line processes using LiMCA. Seventh Australian Asian Pacific Conference on Aluminum Cast House Technology; Hobart; Australia; Sept. 2001. pp. 45-54. 77. Utigard, T.A. and Sommerville, I.D. CLEANLINESS OF ALUMINUM AND STEEL: A COMPARISON OF ASSESSMENT METHODS. Light Metals 2005 as held at the 134th TMS Annual Meeting; San Francisco, CA; USA; Feb. 2005. pp. 951-986. 78. Gallo, R; Mountford, H; Sommerville, I. Ultrasound for On-Line Inclusion Detection in Molten Aluminum Alloys: Technology Assessment. AFS International Conference on Structural Aluminum Castings: Recent Advancements for the and Manufacturing of Reliable Structural Aluminum Castings; Orlando, FL; USA; Nov. 2003. pp. 179-194. 79. Mansfield, T L.Ultrasonic Technology for Measuring Molten Aluminum Quality. Light Metals; Dallas; Tex; Feb. 1982. pp. 969-981. 80. Semensen, C J. Gas-Chromatographic Analysis of Carbides in Aluminium and Magnesium. Fresenius Z. anal. Chem. Vol. 292, no. 3, pp. 207-212. 1978. 81. Simensen, C.J. Sampling and Analysis of Impurities in Aluminium. Refining and Alloying of Liquid Aluminium and Ferro-Alloys. The Norwegian Institute of Technology; Trondheim; Norway; Aug. 1985. pp. 165-171. 82. Siemensen, C.J., and Strand, G. Analysis of Inclusions in Aluminium by Dissolution of the Samples in Hydrochloric/Nitric Acid. Fresenius Z. Anal. Chem. Vol. 308, no. 1, pp. 11-16. 83. LaVelle, D. L. Determining Inclusions in Aluminium Alloy Castings. Foundry/ October 1969. pp 154-157. 84. Vander Voort. Metallography: Principles and Practice. ISBN: 0-87170-672-5 85. Petzow G., Metallographic Etching, ASM (American Society for Metals), 1978, P. 39.

Chapter 2

2 Experimental methods

2.1 Materials The materials used in the thesis were aluminium alloys of mainly AA3003 and 6xxx series, provided by Kubikenborg Aluminium AB. The standard chemical compositions of these types of alloys are given in Table 2.1. The material was DC-cast as billets or slabs with different dimensions. Samples for remelting investigations were taken from a region of 10 mm away from the edge of the ingots in order to get representative compositions. Table 2.1: Chemical compositions (wt%) of aluminium alloys, 3xxx and 6xxx series. Element 3xxx 6xxx Mn 0.9-1.5 0-0.8 Fe <0.7 0-0.8 Si <0.6 0.4-1.3 Mg 0.037 0.3-1.2 Cu <0.2 Zn 0.1 Al Balanced Balanced

2.2 Experimental Techniques The following laboratory techniques were used in this thesis to simulate industrial scale DC casting: 2.2.1 Directional Solidification Directional solidification methods are widely used techniques to simulate and study fundamental issues associated with the solidification during industrial DC casting. These methods include the Bridgman technique and the DC simulator. The DC simulator provides a range of solidification conditions over the length of the samples, which replicate the conditions present in full-scale DC casting. Bridgman solidification allows a series of samples to be cast that can cover the complete range of solidification conditions found in full-scale DC casting. Measurements of temperature were made within these techniques by inserting thermocouples at representative positions; the recorded temperature data obtained were then used to determine the thermal field during solidification and to validate modelling of the process.

2.2.1.1 Bridgman technique The Bridgman directional solidification apparatus used for this study is illustrated schematically in Fig. 2.1. It consists of three zones, a hot zone with a potassium heat pipe, an adiabatic zone about 0.2 m long, and a cold zone with water filled heat pipe. The furnace is equipped with transition device and facility.

Figure 2.1: Schematic of the Bridgman experimental equipment. The Bridgman samples were prepared from the DC cast ingots by remelting in alumina crucibles at 800 oC, with subsequent casting in copper moulds of 5 mm diameter. The cast rods of a total length of 115 mm were then transferred into alumina tubes of 5 mm inner diameters. The specimens were melted in the Bridgman furnace in an argon atmosphere at 800oC by the furnace 7 cm downwards with 2mm/min melting speed. The specimens were held at that position for one hour to ensure stabilization of the solidification front. Growth was then driven by moving up the furnace at different constant pulling rates. The furnace motion was controlled by motor and gearbox, and the movement was monitored by means of a pointer, which was attached to the furnace and moves along a vertical scale. Applied pulling speeds were in the range 1 to 50mm/min. After a certain pulling distance of the furnace ~ 30-55 mm, the solidification was interrupt as the sample was quenched by dropping it downward into water, located under the furnace. The measurements of temperature profiles during Bridgman solidification were carried out in some experiments by inserting thermocouples within the samples of the composition 3003 aluminium alloys. A K-type thermocouple wire, 0.5 mm diameter was inserted into a thin alumina protection tube, 0.8 mm inner diameter, and 1.6 mm outer diameter. These were placed within the Bridgman samples in a hole drilled inside the sample close to the wall of the alumina tube surrounding the sample. The position of the thermocouple tip was; either 50 mm, 80 mm or 100 mm from the bottom of the sample as seen in Fig. 2.2. For each thermocouple position, the sample was placed in the Bridgman furnace in a standard position. By further analysis of the recorded temperature data it was possible to calculate the temperature gradient, position and velocity of the solidification front and cooling rate during the growth. The temperature gradient inside the sample was about 4.7 °C/mm at the solidification front. The cooling rate for the 50 mm min–1 pulling rate was in the range of 0.1-1.2 °C/sec, and a corresponding analysis for the lowest pull rate, 1 mm/min, gave cooling rates in the range between 0.05-0.1 °C/sec.

Figure 2.2: Schematic of locations of the thermocouple. In the Bridgman solidification, during pulling at the high rate, 50 mm/min for 55 mm, the growing interface did not follow the furnace pull rate. The solid/liquid front had only moved about 10 mm. The actual growth rates were calculated from the measurements of the thermal field data, and if the average growth rates are estimated as the ratio of solidification distance over solidification time, the resulting average growth rates become similar.

2.2.1.2 DC simulator A method that can simulate industrial DC casting processes is the chilled casting test apparatus, also knows as DC-simulator, in which, the solidification occurs in a direction approximately perpendicular to a chill surface. The cooling rates close to the chill surface are approximately similar to those in the shell zone of a DC-cast ingot.

Figure 2.3: Schematic illustration DC simulator apparatus. The experimental set up used for the DC simulator is illustrated in Fig 2.3. It consists of a copper block (100mm x 100mm x 30mm) isolated from the sides by ceramic insulator plates lining of 20 mm thickness, and a casting cup (120mm x 80mm x 30mm), made of steel that also served as a and was positioned directly on top of the copper mould. The metal was poured into alumina tubes of 12 mm outside

diameter and 8 mm inner diameter, which were positioned in the copper mould to shape the melt and avoid transverse heat transfer to the copper mould. Directional solidification was achieved by cooling the bottom surface of the copper plate. The thermal field was determined during solidification by temperature measurements using K-type thermocouples placed at heights, 5 mm and 25 mm above the chill surface. The cooling rate at 5 mm from the chill surface was 2 oC/s, the solidification rate 1 mm/s, and the temperature gradient 2 oC/mm. The conditions, (e.g. velocity, cooling rate and thermal gradient) were changed a long the length of the sample throughout the solidification process in a way similar to the variation through the thickness of a commercial ingot. For example, the cooling rate measured closer to the bottom of the sample is approximately similar to the cooling conditions present at the surface of a commercial ingot, whereas slightly further into the sample, the cooling rate was measured to be much lower. Since the conditions are changing constantly through the ingot, so do aspects of the microstructure. 2.2.2 Thermal Analysis Techniques Differential thermal analysis (DTA) and differential scanning calorimetry (DSC) experiments are widely utilized techniques to simply and accurately measure the temperatures of phase reactions in materials, as well as to study small as-cast samples with controlled cooling rate. These calorimetric techniques have been proved to be a useful (complementary) tool in microstructure characterization and their results are used to obtain information regarding the solidification behavior and precipitation reactions occurring in alloys on heating and cooling.

2.2.2.1 Differential thermal analysis (DTA) Differential thermal analysis (DTA) is a technique, which measures the difference in temperature between the investigated sample and a reference material (a thermally inert material- graphite), which are subject to the same heating or cooling as a function of time. When a sample undergoes a phase transformation, it will either absorb, endothermic process, or release heat, exothermic process. The integrated temperature difference as a function of time is a measure of the amount of heat that is absorbed or released during the phase transformation. A schematic diagram of a DTA set-up is shown in Fig 2.4.

Figure 2.4: Schematic of DTA set-up.

DTA specimens of a cylindrical shape of about 13 mm length and 8 mm diameter were heated to 800oC under an argon atmosphere. The experimental data were collected during melting and solidification at heating and cooling rates of 10 and 4 oC/min (0.167 and 0.067K/s). DTA curves were obtained using a sample holder and with two thermocouples (S-type). The sample temperature was measured at the centre of the sample and a reference temperature was measured in a graphite sleeve. During solidification some samples were quenched in water at temperatures that have been chosen according to earlier recorded DTA cooling curves to investigate the start of the solidification, their reactions and formed microstructures.

2.2.2.2 Differential Scanning Calorimetry (DSC) DSC is a technique determining the variation in the heat flow exchanged between the sample and the surrounding as a function of temperature or time. At heating or cooling transformations taking place in the material are accompanied by heat exchange. DSC determines the temperatures of these transformations and quantifies their heat exchanges, i.e. enthalpy changes in the alloys. Fig. 2.5 shows a schematic of a Differential Scanning Calorimetry instrument of heat flux type DSC. It consists of a reference pan and a sample pan contained in a constant temperature enclosure.

Figure 2.5: A heat flux type DSC (left) Heat source (right) Sample carrier. The measurements were performed using a NETZSCH model DSC 404 with disc- shaped samples having a weight of approximately 70 mg encapsulated in alumina pan. (62 mg) was used as a reference. During the measurements, the samples were protected with flowing argon. Temperature scans were made in the range 50-700oC with constant heating and cooling rates of 20, 10, 4, and 1 K/min. The amount of energy, which has to be supplied to or withdrawn from the sample to maintain zero temperature differential between the sample and the reference is the experimental parameter displayed as the ordinate of thermal analysis curve. 2.2.3 Testing

2.2.3.1 Mechanical testing Hardness testing has always been important in the metalworking industry. Measuring hardness is a relatively quick, cheap and easy way to assess the strength of a material

without the need to prepare tensile test samples. For example, it may be a convenient way of investigating the progress of . The micro hardness measurements were performed using a Vickers micro hardness test apparatus under a known load. Pressing a point with a tip on the sample surface with a known force, the size of the diagonal is taken as a measure of the hardness. In the thesis the hardness testing was carried out to evaluate the mechanical properties. In the Bridgman samples, the hardness has been tested with a small load (50 N), in the centre of the dendrites. In the DC simulator samples, the overall structure was tested using a load of 300 N. An average of minimum 10 hardness indentations on each sample surface was measured. Compressions strength tests have been also performed at elevated temperatures to assess the strength of a material. Samples prepared for hot compression were cut 15 mm height and machined to completely flat edges. They were taken 5, 20 and 35 mm from the chill surface to measure the effect of different cooling rates. Isothermal compression tests were performed using a MTS load frame. The testing was conducted at temperatures 400, 450 and 500°C and at strain rates of 0.01 and 1/s. A Lenton resistance furnace, regulated with temperature controllers, was used for heating two water-cooled parallel platens in which two thermocouples (k type) were installed to monitor the temperature of the specimen. Oil based graphite was used as the lubricant between the specimen and the platens to prevent frictional effects during the test. When the platens had settled at the desired test temperature, a specimen was clamped between them and held at this temperature for three minutes before compression. Tests were run using constant velocity control profiles. Load and stroke were acquired during the compression test by a data acquisition system and later converted to true stress-true strain curves.

2.2.3.2 Macroscopic Etching Test A macroetching technique has been used in this thesis to study inclusions in solidified aluminium billets. It was employed to satisfy easy handling, easy sampling and quick evaluation of inclusions in a large area at low cost. The principle idea is to employ visual inspection of macroetch surface of a test piece; a sample of the ingot cross section, and count the number of inclusion appearing as etch pits on the surface regardless of its size, kind and type. The technique can be a useful tool in the evaluation of the cleanliness of the melt, and the inclusions content and distribution along aluminium cast billets. A disk, cut in transverse direction of a DC-cast Al billet, is grounded and macroetched by immersing in hot NaOH solution. The conditions are 10-20 g Sodium hydroxide in 100 ml water, at 60 to 70°C, for about 5 to 15 min. After etching, every kind of inclusions such as, Al2O3, MgO, MgAlO4, TiB2, etc, generate holes, i.e. etch pits in the surface that can be seen by the naked eye. The surface condition for parts to be macroetched is not critical. A finely polished surface is not required; a flat surface obtained by a machine is adequate. However, the surface should be free of grease or oil to permit even etching.

To easy characterize the inclusion distribution on billet slices, mapping of the etched surfaces were made by using a COPY machine, and the copies were then scanned to a computer. The number of inclusions that become visible as black dots were then counted using an image analysis program (KAPPA software). Fig. 2.6 shows an example of pictures of the lateral distribution of pits on etched billet cross sections.

2.5cm from billet bottom block 56cm from bottom block 2.5cm from billet top Figure 2.6: Image of the etched billet cross sections. In Fig. 2.7a-b it can be seen by optical micrographs how an inclusion developes, during subsequent etch steps, into a large etch pit, visible by the naked eye. It is the etching of a TiB2 clusters that is followed.

No etch 10sec etch 30sec etch

90sec etch 15min etch Figure 2.7a: Subsequent etch steps of a TiB2 cluster.

No etch 10sec etch 30sec etch

90sec etch 5.5min etch 15min etch

Figure 2.7b: Subsequent etch steps of an oxide+TiB2 clusters. 2.2.4 Metallographic investigations Microstructure characterization of the investigated aluminium alloys was carried out by optical and electron microscopy. Measurements of microstructure features, such as secondary dendrite arm spacings (SDAS), grain size and the chemical compositions of the secondary phases were carried out. The investigations were made both on as-cast samples of direct chill (DC) castings and on specimens solidified in a Bridgman furnace or under controlled solidification conditions. The solidified samples were cut, grinded on 320 SiC-grit papers and polished with suitable diamond suspension agents, 9 μm, on a hard cloth followed by polishing on a soft polishing cloth with 3 µm and 1 µm diamond paste and finally a 3- 5 minutes treatment with an OPS-solution. Then, in order to reveal the structure, etching was carried out with a reagent using, HF solution, (0.5 ml HF in 100 ml water) for 20 seconds. After etching, the dendrite structure was revealed. For easy grain measurements, the samples were anodized with a anode in a medium of 24% Ethanol, 74%H2O, 1%HF, 1%HBF4 and direct current of 50 V for 90 seconds. The directionally solidified samples were prepared for optical microscopy analysis using longitudinal sections. For the DC cast ingots, samples for microstructural examinations of the size 5 x 15mm were cut at appropriate distances from the surface toward the centre of the ingot.

An optical microscope with attached camera, and software programs called Image– Pro, and/or KAPPA Image were used in the metallographic analysis of the microstructure features. The secondary dendrite arm spacings were measured by averaging the distance between adjacent side branches of a primary dendrite. The measurements of grain sizes were performed in light optical microscopy using polarized light, as the number of grains on the circumference of a circle. In addition to the light optical microscope investigations, a scanning electron microscope (SEM) equipped with an energy dispersive x-ray (EDX) detector was used to provide information about the chemical compositions of the precipitated secondary phases and inclusions types in the investigated alloys. In the presented work transmission electron microscopy (TEM) was also used primarily to determine inclusion particles, which were difficult to detect in SEM due to their small size. Because in many instances no well-developed dendritic structure i.e. dendritic crystals growth with a clear side-branching structure, could be revealed; a technique of measuring the distances between secondary phases and liquid films was applied to obtain numerical values of the coarseness of the structures. This was done by placing lines randomly in the structures and measuring the average free distances along the lines.

Chapter 3

3 Results and discussion

3.1 Results of thermal conditions in the Bridgman furnace Temperature measurements within a sample of aluminium AA 3003 alloys in a Bridgman type furnace during growth (and no-growth conditions) were performed in Suppl. 1. The technique was employed to determine the evolution of temperature field and also to provide information about the growth conditions such as, the temperature gradient, position and velocity of the s/l interface during the directional growth. Thus it makes it possible to understand the performance of the used furnace and to adjust the furnace parameters to simulate the range of solidification conditions that may be found in industrial-scale DC casting. Table 3.1: Compositions of grain refined Al 3003 alloys (wt.%). Alloy Mn Fe Si Mg Ti Cu Zn 3003 1.14 0.50 0.135 0.002 0.018 0.126 0.008 The results of temperature measurements at three thermocouple positions 50, 80 and 100 mm from the bottom of a sample during solidification by furnace pulling with different speeds 1, 10 and 50 mm/min are shown in Figure 3.1. The temperature was recorded at regular intervals approximately 5mm apart during the furnace motion. By further analysis of the recorded temperature data and mathematical treatment, it was possible to calculate the temperature gradient, position and velocity of the solidification front and cooling rate during the growth.

850 100mm

750 80mm

650 50mm

1mm/min 550 10mm/min 50mm/min 450 1mm/min 10mm/min Temperature, oC 50mm/min 350 1mm/min 10mm/min 50mm/min 250 0 10203040506070 Furnace traverse distance, mm

Figure 3.1: Temperatures recorded during furnace pulling as a function of pulled distance. In the mathematical treatment, quadratic functions of the type shown in Equation 3.1 were fitted to the temperature data selected at the three thermocouple positions at different time interval and thus, the temperatures could be obtained as a function of time for all positions in a sample.

2 T(t) = C1 X + C2 X + C3 (3.1)

Cooling curves could thus be obtained for differed pulling rates. Fig. 3.2 shows the calculated cooling curves at a pulling rate of 50 mm/min as an example.

850 800 750 700 650

Temperature (oC) 600 550 0 0,2 0,4 0,6 0,8 1 1,2 1,4 Time (min) 45mm 50mm 60mm 70mm 80mm 90mm 100mm 110mm 115mm

Figure 3.2: Calculated cooling curves at 50mm/min pulling rate. By using the temperature data, the positions of an isotherm can be plotted as a function of time. This has been done in Fig 3.3, where both the solidus and the isotherms have been plotted for two different pulling rates.

100 1,8 1mm/min VL=0.99mm/min 1,6 90 VS=0.97mm/min 1,4 80 1,2 1 70 0,8 60 0,6 I L 0,4 Interface Position, mm

I S Front velocity mm/min 50 VL 0,2 VS 40 0 0 102030405060 Furnace pulling time, min

Figure 3.3a: Interface position and velocity versus time.

60 25 58 50mm/min VL=8.14mm/min 20 56 VS=7.82mm/min 54 15 52 50 10 48 5 46 I L I S Interface position, mm

44 Front velocity, mm/min VL 0 42 VS 40 -5 00,511,5 Furnace pulling time, min

Figure 3.3b: Interface position and velocity versus time.

The slopes of the linear fit of s/l location curves give values for average growth rates as listed in the Fig 3.3. The derivative of each points in the location curves give values of the instantaneous growth rates as a function of time for the studied pulling velocities, Fig. 3.3. Table 3.2, shows the operating conditions of some of the directional solidification experiments for the furnace test. It is clear that there is a deviation between growth distances and pulled distances, i.e. the average growth rate was not equal to the furnace-pulling rate at higher pull rates. Table 3.2: Operating conditions for directional solidified experiments. Furnace Grown Pull distance Calculated Furnace Growth distance in average Pulling rate Time at solidified growth rates [mm/min] [min] quenching sample (mm/min) (mm) (mm) 50 55 1.1 8 7 50 70 1.4 11 8 8 40 5 25 5 5 40 8 33 4 1 10 10 7 0.7 1 30 30 31 1.0 Typically the growth rate starts with a transient, which is followed by a period of constant rate, but at the end, the growth rate increase again. The temperature gradient at the s/l front can also be obtained from the temperature data as the derivative of the quadratics function at the liquidus and solidus temperature. The values of the gradients at the melt interface for the three studied furnace speed, are plotted in Fig. 3.4. It can be observed that the gradient decrease with distance and time during the growth. The maximum gradient of approximately 6oC/mm is observed at the beginning of the solidification.

6,5 1mm/min 6 10mm/min 5,5 50mm/min 5 4,5 C/mm o 4 3,5 3 Temperature gradients, 2,5 40 50 60 70 80 90 100 Distance from the bottom of the sample, mm

Figure 3.4: Calculated temperature gradient as a function of distance. A one dimensional heat flow model (Equ. 3.2) could account for first order effect with regard to the found growth rate variations along the samples. The temperature gradient o in the liquid at the interface, GL get the value 4.6 C/cm at the low pull rate. This is close to

4.8, which was obtained in Fig 3.4 in the stable gradient zone, i.e. 65-80mm from the bottom of the sample. The analytical heat flow modelling can therefore successfully be applied to experiments that reach steady state, but for higher growth rates the boundary conditions for equation (3.2) are not met.

Vo T − T V yl G = hot cold . o eαL (3.2) L ⎛ Vo ⎞ yl α L αL 2⎜ e −1⎟ ⎝ ⎠ Microstructural analysis from Bridgman experiments showed clearly how the solidification parameters affect the resultant microstructure. Equiaxed microstructure observed in samples solidified with high growth rate and mixed columnar-equiaxed microstructures were observed at the lowest solidification rate, Fig. 3.5.

Figure 3.5: The shape of the solidification front for high and low growth rates respectively. The directional solidified microstructures were compared with the DC-cast microstructure. It was observed that the majority of solidification in Bridgman samples adopts a similar morphology to that in DC cast, Fig 3.6.

Figure 3.6: Microstructures in DC cast ingots and in a directional solidification samples faraway from solidification front respectively. 3.1.1 Conclusions The objectives of this work were to investigate whether the thermal conditions in a Bridgman furnace, were comparable to the conditions during DC-casting. It was found that the used Bridgman furnace quite well could be adapted to DC-casting simulation of certain parts of the ingots. In Bridgman solidification, steady-state solidification can be achieved over a larger proportion of the length of the sample by using longer samples; in the current set-up, this was limited by the sample manufacturing and by the size of the furnace.

3.2 Results of different Mg and Si contents on solidification of 6XXX alloys In the case of Al 6xxx alloys, it is Mg and Si that form the strengthening precipitates. Alloy with higher contents of Si than Mg gives a good strength and extrudability. However the casthouse experience is that such alloys are more sensitive to surface defects, and thus more difficult to cast than alloys with balance content of Si and Mg. The influence of the balance between Si and Mg concentrations on the microstructures of some important extrusion alloys, namely 6063, 6005A and 6082 were investigated for the aim to increase the knowledge of their solidification behaviour, and to study the effect of addition of excess Si on the microstructure of AA 6063 alloy. The investigations were made for the compositions given in Table 3.3. Table 3.3: Compositions of the investigated alloys (wt.%). Alloy Type Mg Si Fe Mn Ti Cu 6063 0.4 0.43 0.17 0.02 0.012 0.001 6063+Si 0.42 0.62 0.17 0.02 0.010 0.003 6082 0.64 1.01 0.20 0.48 0.024 0.010 6005A 0.55 0.70 0.21 0.13 0.011 0.105 The as-cast microstructures of the three DC cast ingots 6063, 6005A, 6082 were analysed by optical microscope, and the grain sizes were measured along the diameter of billet slices. No significant difference in gain structures was observed between the three alloys as can bee seen in Fig. 3.7.

200

150

6063 100 6005A

Grain size [Microns] 6082 50 0 40 80 120 Radial distance [mm]

Figure 3.7: The grain size as a function of the radius of the three studied billets. The microstructures of as-cast samples of the investigated 6005A and 6082 alloys at the surface zone are shown in Fig 3.8a. The surface structures show significant differences in the two alloys surface region. The 6005A alloy has a wider enriched segregation region with large precipitations in grain boundaries, while the secondary phases at the 6082 surface seem to concentrate strongly to a more narrower segregate layer and have the appearance of plate like precipitations. The investigated alloys were also studied in experiments performed by the Bridgman technique, which simulate the growth in the large commercial ingots. Samples from the three ingots were directionally solidified at pulling velocities between 5–50 mm min–1, (0.83 10-4 - 8.3 10-4 m/s), which give cooling rate in the order of 0.25-1.0oC/s.

The temperature gradients inside the sample were varied, and were 25oC/cm and 35oC/cm for the two-furnace temperatures 800 and 1000oC respectively.

40μm 40μm

6005A 6082 Figure 3.8a: Surface structure of 6005A and 6082 ingots. Results of the Bridgman experiments showed that, increased Si contents result in larger grain size. Fig. 3.9 shows the grain size (GS) in quantitative form as a function of Si content, and Fig. 3.10, shows that the directionally solidified microstructures close to the quenched front were clearly influenced by the Si content.

200 GS, 50mm/min GS, 5mm/min 150

100 Grain Size (GS), [Micron] 50 0,4 0,5 0,6 0,7 0,8 0,9 1 1,1 Si content in Alloy [wt%]

Figure 3.9: Grain size as a function of Si contents.

200μm

0.43% Si 0.62% Si 1.01% Si Figure 3.10: Directionally solidified structures close to the quenched front, pulling rate 50mm/min. Figure 3.11. Shows the structure coarsening from quenched front to start of growth for two directionally solidified samples of 6063 alloy with different Si contents. It can clearly be observed in Figure 3.11 that Si additions promote development of secondary

dendrite arms and result in a coarser structure with a larger grain sizes, but with shorter distances between secondary phases. The final structure at room temperature is much coarser than the structure formed at solidification, implying that the transformations during cooling are very important for the final structure and the materials properties.

200μm

0.43% Si 0.62% Si Figure 3.11: Structure coarsening from quenched front (top) to start of growth (bottom) for two directional solidified samples at 50 mm/min pulling speed. 3.2.1 Conclusion Differences have been noted in the investigated three ingots; especially the ingots surface structures showed significant variations, which probably influences the surface defect formation. The mechanisms for that is not yet clear, but the inverse segregation to the surface also plays an important roll giving significantly higher alloy contents in this region.

3.3 Results of additions of Zn and Si on the solidification of 3003 alloys Zinc is often chosen as an alloying element because; it is highly soluble and can be dissolved in solid state in aluminium to high concentrations. Thus, the strength can be improved by solution hardening. It is expected that application of Zn additions to current aluminium heat exchanger materials, 3003 type alloys can provide a significant improvement in the corrosion performance of heat exchangers. Si is used for strength improvement as it contributes to both particles and solid solution strengthening. The effect of the additions of Zn and Si on the microstructure formation of 3003 Al alloys was investigated by Bridgman directional solidification technique with quenching were applied. The chemical compositions of the base alloys are as tabulated in Table 3.4. Table 3.4: Composition (wt.%) of AA3003 alloys prior to additions. Alloy Dimension Mn Fe Si Mg Ti Cu Zn Zr AA3003 mm x mm A 1440x330 1.23 0.500 0.145 0.037 0.019 0.146 0.021 0.004 B 1440x600 1.160 0.510 0.489 0.009 0.018 0.125 1.514 0.002 Alloy (A) was used in experiments with Zn additions and alloy (B) was used in the experiments with Si additions. The different levels of Zn additions were 1.5, 2.5, and 5.2 wt.% and the Si addition to alloy (B) was 2 wt.%. Significant effects on the structures of the alloys formed during directional solidification experiments were found as a result of varying amount of alloy additions. For the zinc additions the microstructure became finer i.e. smaller SDAS and grain size as the Zn content increased to 2.5%, but at 5.2 wt% a coarser structure was formed. In Fig. 3.12 the quenched solidification fronts in the Bridgman samples are shown. In the 2.5% alloy, a larger number of less developed dendrites appear giving a finer grain size. Also between the dendrite arms more secondary phases were precipi- tated in this alloy, which hinders the coarsening after the solidification resulting in a final structure being finer both with regard to dendrite arm spacing and grain size.

0.1mm

0.02% Zn 1.4% Zn 2.5% Zn 5.2% Zn Figure 3.12: Microstructures of quenched interfaces in Bridgman samples with different Zn level. Additions of Si, which have a significantly lower solubility than Zn, quickly give a finer structure, larger amounts of secondary phase precipitations and a longer solidification interval. Fig. 3.13 shows the refining effect of Si additions.

400μm

0.47% Si 1.9% Si Figure 3.13: Microstructures along the Bridgman solidified samples from quenched front (top) to start of growth (bottom) with two different Si contents. For Zn additions in the range of 0-5 wt.%, the hardness measured by a Vickers micro hardness test apparatus, showed clear effect, Fig. 3.14. In the Bridgman samples, hardness values from the centre of dendrite arms increased with increased Zn content, as Zn goes into solid solution. In the DC simulator samples, also containing 0.35-0.9 wt% Cu, the hardness values from the overall structure, show a clear peak at 2.5 wt.% Zn as a result of both largest fraction of θ-phase and finest structure observed at this composition.

110 Bridgman sample 100 DC simulator-5mm 90 DC simulator-25mm 80 70 60

Hardness (HV) 50 40 30 0123456 wt% Zn

Figure 3.14: Hardness of alloys with different Zn contents. Bridgman samples measured in centre of dendrites, and DC simulator samples measured in overall structure by larger indentations. 3.3.1 Conclusion Solidification studies of Zn additions to Al 3003 alloys showed that the most important effect on microstructure was noticed at 2.5 wt.% Zn where the structure was fine and the hardness had a maximum. Si addition to a level of about 2% gave a finer structure, and a relatively large fraction of eutectic structure, but also a long solidification interval, which limits the use of Si as a strengthen element to about 0.6%.

3.4 Results of DTA and DSC studies of 3003 alloys containing Zn and Cu The purpose of this was to give a better understanding of the influence of different Zn and Cu additions on the solidification behaviour, phase precipitation and resulting microstructure in Al 3003 alloys. The cooling curves were recorded during the solidification in DTA and DSC experiments. The alloys compositions were Al 3003 alloy with Zn additions in range 0-5wt.% and 0.15-1.0wt% Cu, Table 3.5. DTA curves obtained for 3003 alloys containing about 0.15 wt.% Cu and different Zn additions during solidification at a constant cooling rate of 4 K/min (0.067oC/s) are shown in Fig. 3.15. For the three lower Zn concentrations, i.e. 0.02, 1.5 and 2.5 wt%, only one second peak appears in the curves, giving the temperatures for the start of secondary phase precipitation. The values of these temperatures for each alloy are listed in Fig. 3.15. For 5.2% Zn alloy the peak is not clearly visible, and only a slight flattening of the curve, which indicates a small amount of secondary phase formation, can be seen. The liquidus, the solidus and the temperatures of the secondary phase precipitation are decreasing with increasing percentage of Zn contents in the alloys. The values of these characteristics are listed in Table 3.5. Similar effects, i.e. melting point depression were also observed for the alloys with different Cu contents.

16 Tsecondary 14 0.02 Zn 628 precipitation 12 1.5Zn 639 T0.02=645 C o / 10 2.5Zn T =642

r 1.5

8 T2.5=640 -T 5.2 Zn 645 s 6 T5.2=634 4 T=T o Δ 2 To=651 C 0 -2580 590 600 610 620 630 640 650 660 670 o T sample / C

Figure 3.15: DTA cooling curves of 3003 alloys for different Zn additions cooled at a rate of 4oC/min. The open circles on the DTA thermogram indicate where samples were quenched. DSC cooling curves recorded at the rate of 4oC/min are presented in Fig. 3.16, which shows the effect of varying Zn and Cu on the investigated 3003 alloys.

0,5

0 600 610 620 630 640 650 660 670 -0,5 0.02%Zn+0.15Cu -1 2.5Zn+0.15Cu 2.5Zn+0.94Cu DSC /mW/mg -1,5 2.3%Zn+0.35%Cu 5.2%Zn+0.15Cu -2 Temperature /oC

Figure 3.16: DSC cooling curves of 3003 alloys with different Zn and Cu content at a cooling rate of 4 K/min.

In Table 3.5, it can clearly be observed that the increased Zn content and the presence of excess Cu have a significant effect on the melting point depression, on the solidification ranges, and temperatures of the secondary phase precipitation. Especially at the highest alloy contents, i.e. in the samples containing 5.2Zn+0.15Cu and 2.5Zn+1.0Cu, the melting point depressions are obvious. Table 3.5: Data from DTA and DSC experiments at rate of 4 K/min O O O O Alloys TLiquid ( C) Tp-precipitation ( C) TSolid ( C) T-range ( C) name DTA DSC DTA DSC DTA DSC DTA DSC 0.02Zn 651 657 645 640 641 630.5 10 27 1.5Zn 649 642 638 11 2.5Zn 648 654 640 637 635 626 13 28 5.2Zn 645 650 634 635 625 621 20 29 2.5Zn0.35Cu 648 653 640 637 634 625 14 27 2.5Zn1.0Cu 644 650 636 635 624 622 20 28 The results of the microstructure investigations of the as-quenched DTA samples with different Zn are shown in Fig 3.17. It can be seen how the different crystallization morphologies change as the Zn and Cu contents increase. At 2.5% Zn, a globular structure, i.e. less developed dendrites were formed. At 5.2% Zn a coarser dendritic structure with larger solidification intervals was formed. The analysis of the alloys with different Cu contents, Fig. 3.18, shows that an increase of the amount of Cu gives a more developed dendritic growth morphology.

800μm

(a) (b) (c) Figure 3.17: Optical microstructures of the quenched DTA samples (a) 1.5Zn+0.15Cu o o (Tquench=645), (b) 2.5Zn+0.15Cu (Tquench=639 C) and (c) 5.2Zn+0.15Cu (Tquench=628 C).

800μm

(a) (b) (c) Figure 3.18: Optical microstructures of the quenched DTA samples contain 2.5%Zn and different o o Cu. (a) 2.5Zn+0.15Cu (Tquench=639 C) (b) 2.5Zn+0.35Cu (Tquench=642 C) and (c) 2.5Zn+1.0Cu o (Tquench=635 C)

Although only one second peak can be seen on the DTA cooling curves, two secondary phases were observed in the investigated microstructures. Type 1 was identified by EDX-analysis in a SEM to be Al6(MnFe) and type 2 was α– Al15(Fe,Mn)3Si2, which is formed both from the liquid and as a transformation of the first phase. Fig. 3.19a shows the appearance of these phases in the base alloy and Fig. 3.19b shows a partly transformed particle.

(a) (b) Figure 3.19: SEM images of (a) the precipitated phases in the base alloy and (b) shows the transformation where α–Al(Fe,Mn)-Si is brighter than the Al6(Fe,Mn) phase. EDX-analysis also indicated how the Zn and Cu entered into, and stabilized the different phases. Generally only Zn goes into the type 1-phase as Al6(Fe,Mn,Zn) but both Zn and Cu can be found in the type 2-phase as Al15(Fe,Mn,Zn,Cu)3Si2, i.e. at 4- 5% Zn, the Al6(Fe,Mn,Zn) phase is dominating, and Si is mainly found in the Al matrix. EDX analysis of structures formed by DC simulator test at higher cooling rate than the DTA samples indicated a similar effect occurred but even stronger. Figure 3.20 shows clearly how the amount of secondary phases increase with the Zn content until 2.5%, but upon further Zn additions it decreases again. Also the dendrite arm spacing has a minimum at 2.5%, and the SEM pictures reveal a segregation in the 1.5 and 4% Zn structures, but not in the 2.5% alloy.

0.02%Zn 1.5%Zn

2.5%Zn 4%Zn

Figure 3.20: Backscatter SEM microstructures showing the effect of different Zn contents

EDX analysis of the DC simulator structures also reveal that an eutectic phase in the 2.5% Zn and 1.0% Cu alloy, has a composition corresponding to Al2(Cu,Zn,Si) resembling the appearance of the θ-phase, Al2Cu, in the Al-Cu system. In the 4% Zn alloy the dominating phase is Al6(Fe,Mn,Zn) with only small amounts of the light Al2(Cu,Zn,Si) phase. This is in agreement with the DTA results, where Al6(Fe,Mn,Zn) is strongly dominating at higher Zn contents. 3.4.1 Conclusions The DTA and DSC analysis showed clearly the differences caused by the additions of Zn and Cu to the balance 3003 alloys. The results of microstructural investigation showed that at 2.5% Zn, globular structure was formed, whereas the alloy with 5.2% Zn has a coarser structure and relatively large solidification range. In all the DTA and DSC samples, where the cooling rates are low, and the solidification processed close to equilibrium conditions; only two secondary phases were observed in the alloys. Neither Zn nor Cu did form own intermetallic phases but were found in the solid solution and in precipitated phases as Al6(Fe,Mn,Zn) and Al15(Fe,Mn,Zn,Cu)3Si2. However ingot solidification occurs at a higher cooling rate, and therefore at increased Zn contents in the presence of Cu, the Al2(Cu,Zn,Si) start to form at 1.5% and has a maximum at 2.5%, while at higher Zn concentration Al6(Fe,Mn,Zn) is the dominating phase.

3.5 Results of additions of excess titanium Titanium is usually added to many aluminium alloys because it is a potent grain refiner, i.e. gives smaller grain size, and at higher additions can possibly improve the corrosion resistance. However the casthouse experience shows that if Ti is added as an alloying element and not only as a grain refiner to the level of the peritectic point and above 0.15-0.20 wt.%, makes it more difficult to cast the alloy without cracks. The effect of different Ti concentrations was studied by both characterizing the grain structure over the thickness of DC cast rolling ingots, and by studying the solidification microstructures of Bridgman directionally solidified samples. The investigated alloy compositions, as shown in Table 3.6, were Al 3003 alloys produced by DC casting. The grain refiner was Al-5%Ti-1%B. Table 3.6: Compositions of the investigated alloys. Alloy Dimension Mn Fe Si Mg Ti Cu Zn Zr mm x mm 1440X400 0.76 0.198 0.142 0.212 0.148 0.283 0.010 1440x460 0.843 0.216 0.139 0.002 0.155 0.305 0.006 1440x460 0.931 0.209 0.139 0.083 0.163 0.276 0.011 1440x600 1.14 0.5 0.135 0.002 0.018 0.126 0.008 1440x600 1.18 0.53 0.147 0.009 0.029 0.115 0.041 1440x600 1.55 0.258 0.748 0.013 0.013 0.042 1.429 0.143 Figure 3.21, shows the grain sizes for varying Ti levels, measured from the surface to the centre of several DC cast ingots. It can be seen that, the high Ti alloys have a coarser structures than the alloys with the normal Ti concentrations for grain refinement, i.e. <0.02 %Ti. Fig. 3.21 also shows the poisoning effect of Zr, i.e. the grain refining mechanism cease to work.

600 500 +Zr 400 300 200 100

Grain size [Microns] 0 0 100 200 300 Distance from ingot surface [mm] Ti=0.148wt% Ti=0.155wt% Ti=0.163wt% Ti=0.018wt% Ti=0.029wt% Ti=0.013%+Zr

Figure 3.21: The as-cast grain size as a function of distance from the ingot surface for different Ti contents and for an alloy with Zr. Experiments were done by Bridgman directional solidification with high and normal Ti contents. It was found that the high Ti alloys had a finer grain size and stronger ten-

dency to equiaxed growth, Figure 3.22, which is opposite to the results of the DC cast ingots where the high Ti alloy showed a coarser grain structure.

The results were explained by an analysis of the dissolution time of the Al3Ti particles, which showed that the dissolution time increased from about 60 to 200 seconds when the Ti content increased from 0.015 to 0.15%. If the Al3Ti particles do not totally dissolve before the melt starts to solidify, a few large Al3Ti particles will nucleate the aluminum grains instead of the numerous boride particles, which in this case are smaller and not activated as nucleation points.

Figure 3.22: Microstructures of quenched solidification fronts for low Ti and high Ti respectively, solidified at 50mm/min pulling rate. Solidification direction upwards with quenched liquid on top.

This conclusion was supported by an analysis of the bulk ingot structure. Some Al3Ti particles could be revealed in the centre of aluminium dendrites, as shown in Figure 3.23. The undissolved large particles are therefore likely the reason to the coarser structure and the higher cracking sensitivity in the high Ti alloys. The reason why the structures in the Bridgman experiments were finer at the higher Ti content was that in these cases the time for dissolution of Al3Ti particles was long enough and therefore the high Ti level could increase the growth restriction and give a finer structure.

100μm 100μm

0.155% Ti 0.163% Ti

Figure 3.23: Al3Ti at the centers of grains in the microstructure of DC-cast ingots. 3.5.1 Conclusion Comparison of alloys of different Ti contents showed that; addition of titanium to a level > 0.02% gives a coarser grain structure than alloys with a lower Ti content. This is probably a result of that the Al3Ti particles in grain refiner master alloys are not totally dissolved during the casting process implying nucleation on larger but few particle. The larger grains sizes in the high Ti alloys are most likely the results of high crack sensitivity during DC casting of these alloys.

3.6 Results of assessment of inclusions in aluminium billets The knowledge of the amounts, distribution and locations of inclusions along DC cast billets length and cross section is of large industrial importance. A good assessment method could result in reduced scrap, i.e. an improvement of the yield by decreasing the off length at the ingot ends and scalping depth at the ingot surface. Due to the low inclusion content, and the high tendency of inclusion to segregate into a nonuniform distribution, the direct microscopic analysis of small metallographic samples is not practical, since a very large area would have to be observed for quantitative assessment. Macroetching technique based on NaOH acid solution has been developed to study inclusions in solidified aluminium billets. It was employed to satisfy easy handling, easy sampling and quick evaluation of inclusions in a large area at low cost. Four billets of two different types of alloys, AA 6063 and AA 6082 were analysed. Two of the billets namely, A1 and A2 were of the same charge but from different positions of the casting table, while the billets named B, and C were from different charges. The technique is that, a disk of the billets cross-section is immersed in hot NaOH solution. After the deep etching, inclusions generate holes/etch pits on the surface. The results of subsequent etch steps in Fig. 3.24 show how a TiB2 inclusion develops into a visible hole in the surface.

no etch 10sec etch 30sec etch

90sec etch 5.5min etch 15min etch

Figure 3.24: Subsequent deep etching of an oxide+TiB2 cluster inclusion.

An example of a macroscopic view of the inclusion distributions on deep etched cross- sections of billets is presented in Fig 3.25. The photographs show a non-uniform lateral distribution of holes (black dots), which correspond to inclusions. A high inclusion content in the billets surface regions can generally be seen compared to interior. It can also be seen that the first billet disk at position 2.5 cm from bottom block, shows a large number of very small holes, taking a shape of circular zone close to the mould wall.

2.5cm 12cm 60cm 788cm

Figure 3.25: Photographs of the etched billet cross sections, showing the lateral distribution of etch pits along the length of a billet (samples position from the bottom block). The quantity of inclusions on the etched surfaces of different ingots was measured regardless of the pits type and kind by an image analysis program. The holes were classified based on their maximum length. Three size ranges were used and results are presented in Fig. 3.26. It shows the size distribution changes along those billets. Higher densities of etch pits are revealed at both ends, i.e. at the bottom and top parts, compared to the main part of the billet in between. At the bottom samples, i.e. 0-3 cm from the bottom block, the highest densities of inclusions are located. At 5-12 cm the relative amount of large etch pits increased compared to the rest of the billet.

100 A1: 0.20.4 90 A2: 0.20.4 B: 0.20.4 80 70 60 50 40 30

Numbers of holes (length) 20 10 A1 A2 B 0 4.5-5 11-12. 56-57 786-790 788-792 Sample position from the bottom end along the billets (cm)

Figure 3.26: The variations of etch pits counted in transverse sections along three billets of two different charges. Optical microscopic charactarization were performed on selected pieces 20mm x 10 mm, which were cut from the studied macroscopic etched sections. In the slices from 2.5 cm from the bottom block, the majority of discovered imputities were aluminium oxides, (Al2O3) and borides cluster (TiB2). At ~56 cm form the bottom block, magnesium oxides were found in additions to Al2O3, and TiB2. Fig 3.27 shows typical appearnaces of the different inclusion types in these two positions.

(a) (b) Figure 3.27a: Optical micrograph showing inclusions of (a) Oxide particles and (b) oxides films in sample position 2.5 cm from bottom block of the billet A1.

(a) (b) (c)

Figure 3.27b: Optical micrograph showing inclusions (a) spinal MgAl2O4 (b) MgO and (c) TiB2 cluster in sample position at 56 cm from bottom block of the billet A1. 3.6.1 Conclusions The results of the studied billets revealed similar distribution tendency of the etch pits along the billets length. A trend in decreasing number of etches pits from the bottom block toward the top with an increase again at the very top of the billets. Also the presence of a ring shaped distribution of a large number of small defects in the beginning of the casting, i.e. just above the bottom block could be revealed. These higher densities of pits in the beginning of the casting are related to the disruption of the oxides films due to the strong turbulence during pouring the liquid aluminium to the mould. The slight increase in the inclusion frequency at the very top part of the billet can be explained by the effect that at the end of the casting process the melt flow is disturbed as the melt level in the launder system start to decay. The ring shaped small inclusions are identified as TiB2 particles and associated with the grain refinment proceedure. During casting practice excess of grain refiners are often added in the start of the casting. The general high etch pits density closer to the billets surface might be couple to the melt flow conditions and the high solidification in this resgion.

4 General Conclusions of Thesis The solidification process and the resulting microstructure are very important for the mechanical properties of the cast materials. By increasing the basic knowledge of the crystallisation process it will be possible to improve the properties and tailor the material for different applications. The properties of aluminium alloys are influenced by a number of microstructural features, which are controlled by chemistry and processing. The present study was carried out to investigate the role of some major operating parameters applied in aluminum foundries and that of minor alloying element additions on microstructure, properties, and quality in 3003 and 6xxx wrought aluminium alloys, using laboratory casting techniques. From an analysis of the results, the following general conclusions may be drawn. • Addition of alloying elements such as Zn, Si, Cu, and Ti can be used to improve properties and tailor the material for different applications, if the amounts are optimized. • Another way of improving the properties is by employing optimum additions of grain refiners. • Inclusions are a problem, and better techniques to detect them and their distribution, will improve the possibilities to limit the problem.