Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Determination of the influence of interdendritic segregation during the solidification of freckle-prone… Auburtin, Philippe Bernard Lucien 1995

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_1995-0503.pdf [ 10.92MB ]
Metadata
JSON: 831-1.0078520.json
JSON-LD: 831-1.0078520-ld.json
RDF/XML (Pretty): 831-1.0078520-rdf.xml
RDF/JSON: 831-1.0078520-rdf.json
Turtle: 831-1.0078520-turtle.txt
N-Triples: 831-1.0078520-rdf-ntriples.txt
Original Record: 831-1.0078520-source.json
Full Text
831-1.0078520-fulltext.txt
Citation
831-1.0078520.ris

Full Text

DETERMINATION OF THE INFLUENCE OF INTERDENDRITIC SEGREGATION DURING THE SOLIDIFICATION OF FRECKLE-PRONE ALLOYS by PHILIPPE BERNARD LUCIEN AUBURTIN B.A.Sc, Ecole Centrale Paris (FRANCE), 1992 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF T H E REQUIREMENTS FOR T H E D E G R E E OF M A S T E R OF APPLIED SCIENCE in T H E F A C U L T Y OF G R A D U A T E STUDIES (Department of Metals and Materials Engineering) We accept this thesis as conforming to the required standard ^ T H E UNIVERSITY OF BRITISH COLUMBIA August 1995 © Philippe Bernard Lucien Auburtin, 1995 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of The University of British Columbia Vancouver, Canada Date SejA-,- ^  JWC DE-6 (2/88) Abstract Freckles are presently one of the major defects encountered in advanced casting technology. The current state-of-the-art knowledge about freckles is discussed in an extensive critical literature review, including the main physical characteristics of freckles, the current theory of density inversion leading to channel segregation and the various experiments and mathematical models developed so far. This review emphasizes the fact that no quantitative research on freckles has been carried out on actual industrial alloys yet. The goal of this thesis is to start this quantitative research by measuring interdendritic segregation profiles and evaluating the extent of density inversion in the mushy zone and its possible effects on freckle formation. Chemical compositions and segregation were measured by electron microprobe on directionally solidified and quenched samples of the following alloys : IN718, MAR-M002, MAR-M247, C-276 and T l . Freckled samples were also chemically analyzed. Densities, function of temperature and composition, were estimated by a numerical model. It was found that freckles initiated from density inversion of the order of 0.01 (g/cm 3 ) /°C under typical superalloy casting conditions. In superalloys, it was estimated that freckles tend to initiate relatively close to the top of the mushy zone (20°C below the liquidus temperature) where fraction liquid is still high (about 40-60%). In high carbon alloys, especially tool steels, precipitation of heavy alloying elements into carbides may act as a trigger for freckling. Finally, the potential importance in the analysis of freckles of the segregation of minor alloying light elements such as carbon, silicon, zirconium, manganese, phosphorus, etc. was outlined. ii Table of Contents Abstract ii Table of Contents iii List of Tables v List of Figures vi List of Symbols viii Acknowledgments x 1. I N T R O D U C T I O N 1 2. C R I T I C A L L I T E R A T U R E R E V I E W 3 2.1. F R E C K L E S IN T H E INDUSTRY 3 2.1.1. Morphology 3 2.1.2. Chemical composition 6 2.1.3. Influencing factors 6 2.1.3.1. Shape and geometry 6 2.1.3.2. Solidification parameters and operating conditions 7 2.1.3.3. Alloy composition 7 2.1.4. Physical properties of freckles 8 2.1.5. Cost of freckles 8 2.1.6. Center-segregation and freckles 9 2.2. M E C H A N I S M S O F F R E C K L E F O R M A T I O N 10 2.2.1. Freckling theory 10 2.2.1.1. Density inversion theory 10 2.2.1.2. Driving force 12 2.2.2. Freckle initiation and growth 14 2.2.3. Fluid flow velocity in freckles 16 2.3. A N A L O G S Y S T E M S A N D SPECIAL EXPERIMENTS 19 2.3.1. Lead-based systems 20 2.3.2. Aqueous ammonium chloride 22 2.3.3. Other analog systems 26 2.3.4. Experiments with superalloys 26 2.3.5. Special experiments 29 2.3.5.1. Low gravity experiments 29 2.3.5.2. Mold rotation 29 2.4. F R E C K L E M O D E L S A N D CRITERIA 31 2.4.1. Models basis 31 2.4.2. Models results 32 2.4.2.1. Bennon & Incropera 32 2.4.2.2. Felicelli et al. 33 2.4.3. Freckling criteria 36 2.4.3.1. Initial perturbation and Rayleigh number 36 2.4.3.2. Effect of growth rate and thermal gradient on freckling 38 2.4.3.3. Flemings' criterion 3 9 2.4.3.4. Other criteria 40 2.5. L I T E R A T U R E R E V I E W CONCLUSION 41 iii 3. O B J E C T I V E S AND E X P E R I M E N T A L P R O C E D U R E 42 3.1. O B J E C T I V E S 42 3.2. C H O I C E O F T H E A L L O Y S 43 3.3. S A M P L E PREPARATION : T H E DSQ F U R N A C E . 45 3.4. S A M P L E ANALYSIS : S E M - E D X M E A S U R E M E N T S 49 3.5. D E N S I T Y CONVERSION 55 3.5.1. Mathematical model 5 5 3.5.2. Model's limitations 57 4. R E S U L T S 59 4.1. DSQ S A M P L E S 59 4.2. INTERDENDRITIC S E G R E G A T I O N A N D LIQUID DENSITY 63 4.3. F R E C K L E D S A M P L E S 71 4.4. F R A C T I O N LIQUID A N D PRIMARY DENDRITE A R M SPACING 76 5. DISCUSSION 77 5.1. V A L I D A T I O N O F T H E M E A S U R E M E N T S 77 5.1.1. Primary dendrite arm spacing 77 5.1.2. Freckle composition measurements 78 5.1.3. Segregation measurements 78 5.1.4. Fraction liquid 81 5.2. E X T E N T O F T H E DENSITY INVERSIONS 82 5.2.1. Theoretical provisions 82 5.2.2. Freckles density 83 5.2.3. Segregation, related density, and freckle formation 86 5.2.4. Possible influence of non-measured elements on freckle formation 88 5.2.4.1. Alloy IN718 88 5.2.4.2. Tool steels T l and M2 91 5.2.4.3. Other alloys 93 6. C O N C L U S I O N 95 6.1. S U M M A R Y 95 6.2. R E C O M M E N D A T I O N S F O R F U T U R E W O R K 96 References 98 Appendix A : Expressions of the permeability factor 105 Appendix B : The Scheil and Lever rules 106 Appendix C : DSQ samples E D X data. 107 Appendix D : Graphic determination of the freckle initiation position. 112 iv List of Tables Table 1: Table of the compositions (wt%) of a freckle and surrounding matrix area in IN718 [5].6 Table 2 : Summary of observations in bottom chilled rotating cylindrical mold [26]. 30 Table 3 : Standard compositions (in wt%) and melting range of chosen alloys [44]. 44 Table 4 : Compositions of the products of a DSQ run with MAR-M247. 62 Table 5 : Freckles and surrounding matrix compositions. 71 Table 6 : Primary dendrite arm spacing in DSQ samples. 76 Table 7 : Partition coefficients of various elements in industrial alloys. 79 Table 8 : Measured and calculated (Scheil and Lever rules) eutectic compositions in DSQ alloys.79 Table 9 : Measured and back-calculated fraction liquid at various temperatures in the mushy zone for 3 DSQ alloys. 81 Table 10 : Freckle and surrounding matrix calculated densities (after "METALS"). 84 Table 11: Published freckle/matrix compositions for various alloys and corresponding calculated liquid densities. 85 Table 12 : Published average and eutectic compositions for various alloys and corresponding calculated liquid densities (after "METALS"). 87 Table 13 : Calculated effect of carbon segregation and carbide precipitation on the density of the interdendritic liquid in IN718. 89 Table CI : Numerical data measured by EDX microprobe on DSQ IN718. 107 Table C2 ; Numerical data measured by EDX microprobe on DSQ MAR-M002. 108 Table C3 : Numerical data measured by EDX microprobe on DSQ MAR-M247. 109 Table C4 : Numerical data measured by EDX microprobe on DSQ C-276. 110 Table C5 : Numerical data measured by EDX microprobe o n D S Q T l . I l l v List of Figures Figure 1 : Hollowed Udimet 700 freckled casting (4 in. OD by 3 in. high) [6]. 5 Figure 2 : Mar-M200 (4 in. diam.) ingot showing freckling distribution [6]. 5 Figure 3 : Longitudinal section of a freckle in Pb-15 wt%Sn [11]. 5 Figure 4 : Close-up of one of the freckle lines from figure 2 [6]. 5 Figure 5 : Phase diagram and corresponding casting geometry in vertical upward directional solidification of a binary alloy [9]. 12 Figure 6 ; Schematic concentration (a) and density (b) profiles above and below the dendritic front [24]. 12 Figure 7 : (a) Schematic phase diagram (b) Corresponding freckle potential. [8] 13 Figure 8 : Schematic representation of liquid layer perturbation leading to plume and freckle formation [9]. 14 Figure 9 : Schematic of the velocity profile and boundary conditions for flow of a cylindrical plume of liquid through a quiescent liquid of higher viscosity [24]. 17 Figure 10 : Phase diagrams : (a) Pb-Sn, (b) Pb-Sb [28]. 20 Figure 11 : Cross section of a Pb-2wt%Sb cylindrical ingot showing freckle dots [28]. 21 Figure 12 : Composition change across a freckle in Pb-17wt%Sn [11]. 21 Figure 13 : Phase diagram of NH 4 C1-H 2 0 [27]. 22 Figure 14 : NH 4 C1-H 2 0 ingot casting, showing colored solution pouring out from behind the dendritic front through "A" segregate channels [7]. 24 Figure 15 : Example of apparatus used for studying unidirectional solidification of NH 4 C1-H 2 0 [8]. 24 Figure 17 : General view of plumes in NH 4Cl-70wt%H 2O [9]. 24 Figure 18 : Freckle lines along the uppermost portion of the cylindrical surface of a Mar-M200 ingot, directionally solidified with its axis tipped 35° from horizontal [8]. 27 Figure 19 : Schematic of commercial furnace used for directional solidification of cluster molds [3]. 28 Figure 20 : Macrosegregation after complete solidification of a NH 4Cl-70wt%H 2O slab (vertical left side chill). Bennon & Incropera's model [39]. 33 Figure 21: Freckle formation in Pb-10wt%Sn (bottom chill), (a) Fraction liquid (b) Streamlines (Size of the figure : 2.5mmx4.5mm) Felicelli et al. model [20]. 34 Figure 22 : "No freckle" operating domain criterion [8]. 39 Figure 23: Schematic diagram of the DSQ furnace. 46 Figure 24 : Various recorded temperature profiles in the DSQ furnace. 48 Figure 25 : Schematic cross-section window of a DSQ sample. 49 Figure 26 : Schematic representation of the analysis of a DSQ sample. 50 Figure 27 : Etched cross-sections at various depths in DSQ MAR-M247 (Mag. X51). 52 Figure 27 : (continued) 53 Figure 28 : Comparison of calculated and measured densities at the liquidus temperature for Fe-6wt%W-C melts, versus carbon content. 58 Figure 29 : Etched longitudinal section of a whole DSQ MAR-M002 (Mag. X6.5). 60 Figure 30 : Etched longitudinal section of the top of the mushy zone in DSQ MAR-M002 (Mag. X22). 61 Figure 31 : Interdendritic liquid segregation and fraction liquid profiles along the mushy zone in DSQIN718. 65 vi Figure 32 : Interdendritic liquid segregation and fraction liquid profiles along the mushy zone in DSQ MAR-M002. 66 Figure 33 : Interdendritic liquid segregation and fraction liquid profiles along the mushy zone in DSQ MAR-M247. 67 Figure 34 : Interdendritic liquid segregation profiles along the mushy zone in DSQ C-276. 68 Figure 35 : Interdendritic liquid segregation profiles along the mushy zone in DSQ T l . 69 Figure 36 : Interdendritic liquid density profiles computed by "METALS" for 5 DSQ alloys. 70 Figure 37 : Cross-section of a IN718 ingot exhibiting freckles (Mag. X0.87). 72 Figure 38 : Cross-section of a forged T l billet exhibiting a ring of freckles at mid-radius (Mag. X0.6). 72 Figure 39 : DS MAR-M002 test slab (dark freckle in lower right corner) (Mag. X0.9). 73 Figure 40 : Cross-section of a C-276 slab. Freckles are light etching. They appear flattened after the slab was rolled down. (Mag. X0.55). 73 Figure 41 : SEM picture of a freckle in IN718 exhibiting large Niobium carbides (Mag. X60). 74 Figure 42 : Porosity associated with a freckle in MAR-M002 (polished sample) (SEM picture) (Mag. X300). 74 Figure 43 : SEM close-up of the carbides in a freckle in T l (Mag. XI200). 75 Figure 44 : SEM close-up of the carbides in the bulk matrix in T l (Mag. X1200). 75 Figure 45 : Primary dendrite arm spacing versus cooling rate [54]. 77 Figure 46 : Segregation profiles in tool steel M2 [65,66]. 92 Figure 47 : Calculated density profiles in tool steel M2, assuming 3 different carbon segregation patterns. 94 vii List of Symbols Symbols SI Unit Meaning a mol Number of moles A m 2 Area C wt% Solute concentration D m2/s Diffusivity fL % Fraction liquid g m/s2 Gravitational acceleration (=9.81 m/s2) G °C/m Thermal gradient h m Characteristic linear dimension H J/m 3 Enthalpy k - Partition coefficient K m Permeability KT J/ms°C Thermal conductivity Le - Lewis number MV m3/mol Molar volume P Torr Pressure Pe - Peclet number Pr - Prandtl number r m Radius R m/s Solidification rate Ra - Rayleigh number Re - Reynolds number t s Time T °C Temperature V m/s Fluid flow velocity W kg Weight x m Horizontal coordinate z m Vertical coordinate Greek Symbols SI Unit Meaning a 1/°C Thermal expansion coefficient P l/wt% Solutal expansion coefficient y - Constant 8 (kg/m3)/m Density inversion magnitude E °C/s Rate of temperature change r| kg/m.s Dynamic viscosity X m Dendrite arm spacing v m2/s Kinematic viscosity p kg/m3 Density x - Tortuosity factor co rpm Rotation rate v|/ J Potential energy viii Superscripts Meaning /' Chemical element / j Alloy; K Permeability dependent * Critical Subscripts Meaning 1 Primary 2 Secondary E Eutectic L Liquid Liq Liquidus max Maximum min Minimum mp Melting point o Reference, Initial S Solid Sol Solidus T Thermal T/S Thermosolutal X x-axis component z z-axis component Abbreviations Meaning C E T Columnar to Equiaxed Transition E B M Electron Beam Melting D T A Differential Thermal Analysis DS Directionally Solidified DSQ Directional Solidification and Quench ESR Electro-Slag Remelting E D X Energy Dispersion Spectrometry HIP Hot Isostatic Pressing LST Local Solidification Time N/R Not Recorded SX Single Crystal SEM Scanning Electron Microscope V A R Vacuum Arc Remelting ix Acknowledgments The author would like to thank particularly his supervisor, Dr. A. Mitchell, for his invaluable guidance throughout this M.A.Sc, as well as A. Schmalz for all his help during the design, building and operating of the experimental equipment. All the support staff in the department of Metals & Materials Engineering at UBC was also most helpful. A special thank should also go to Dr. P. Quested and Dr. K. Mills at National Physical Laboratories (UK) for providing the much used "METALS" model. The help from the following companies concerning the alloys studied in this thesis was most appreciated : Aubert & Duval, Inco Alloys International, Rolls-Royce, Sandia, Special Melted Products, Special Metals Corporation, Teledyne Allvac/Vasco. Finally, the author is very thankful to Ms. K. Roe for her much needed moral support. x L_ I N T R O D U C T I O N In today's industrial world, virtually all metals undergo a melting and casting operation. Indeed, many of the present metallurgical extraction processes yield liquid metal. Alloying additions are also much easier to make with liquid metal. Moreover, metal refining (finer structure, inclusion removal, etc.) usually involves melting. Liquid metal is then cast into ingots or directly into near-net-shape parts. Although casting can be a relatively basic and elementary operation as in the case of ingots meant to be remelted, casting can also be a very advanced technology, providing large added value to the material, as in the case of single crystal blades for aerospace turbine engines. There are a wide variety of casting techniques available, depending on the requirements of the product and on the metal behavior (melting point, reactivity, etc.) : mold casting, continuous casting, skull casting, investment casting, etc. Melting and casting can also be closely coupled, as in the case of vacuum arc remelting (VAR), electro-slag remelting (ESR) or electron beam melting (EBM). The shape of the final product usually dictates the type of casting to be employed : a billet to be rolled, a disc to be forged and a near-net-shape blade require different types of casting. However, most of all, the determining factor is really the specifications of the physical properties of the final part. Industrial applications increasingly require metal alloy parts that can withstand higher and higher mechanical stresses and temperatures, parts that are more and more resistant to corrosion, fatigue failure, creep, etc. Thus casting techniques have to conform to greater and greater quality standards, especially in the case of superalloys. Apart from porosity and shrinkage, one of the main quality problems associated with alloy castings is segregation. Although microsegregation (segregation at the dendrite scale), by its very nature in alloy systems, cannot be avoided during casting, it can usually be eliminated by 1 a proper heat treatment. However, castings can also be subject to macrosegregation (segregation at the casting scale), as a result of physical movement of liquid or solid phases [1]. Unlike microsegregation, macrosegregation can be very difficult, and in many cases impossible to eliminate through thermomechanical treatment because of the very low rate of solid state diffusion of some solute species [2]. Moreover, macrosegregation can lead to the formation of isolated misoriented grains, which is highly undesirable in the case of directionally solidified (DS) or single crystal (SX) parts [3,4]. The term macrosegregation is usually employed to describe any kind of non-uniform chemical composition. For example, "white spots" in V A R superalloy ingots are the remains of pieces of crown shelf, cut by the electric arc, that sank in the liquid pool without remelting. The chemical composition of the crown being somewhat different from the rest of the ingot, these white spots appear as macrosegregates [2,5]. Macrosegregation can also result from the continuous fluid flow and solute partitioning occurring during the entire duration of the casting, such as center segregation. More specifically, this thesis will focus on the occurrence of freckles in various alloys. Freckles are a special kind of macrosegregation related directly to the solidification process. These macrosegregates are the result of the combination of interdendritic microsegregation and fluid flow of the liquid alloy. This thesis will begin with a critical state-of-the-art literature review on freckles. The objectives of this research project as well as the corresponding experimental procedure will follow. The bulk of the experimental results will then be presented in the next part of this thesis. These results will be detailed, analyzed and discussed in the following chapter. Finally, the main significant findings as well as the recommendations for future work will be reported in the conclusion section of this thesis. 2 2^ C R I T I C A L L I T E R A T U R E R E V I E W This chapter is a critical review of the scientific knowledge about freckles. It gathers the principal results and observations from research carried over the past 25 years. It also outlines the main elements lacking for a full understanding (and eliminating) of the freckling phenomenon. This chapter will first describe the main features of freckles as they are found in the industry. It will then explain the now widely accepted theory of the mechanism responsible for freckling. The various experiments as well as mathematical models that have been developed to date and that led to the understanding of the freckle mechanism will also be reported. 2*1^  F R E C K L E S IN T H E INDUSTRY Freckles, also known as "channel segregates", are a special kind of macrosegregation. They can be found in a wide variety of castings such as V A R and ESR superalloys billets [2,5], DS and SX castings of superalloys [6], and large killed steel ingots [7]. 2.1.1. Morphology Freckles appear as dark etching spots in transverse sections of arc-melted superalloy ingots [8], hence their name. Freckles are actually cylindrical pencil-shaped segregates, of the order of 1mm in diameter, running parallel to the direction of gravity. In the case of arc-melted ingots, freckles are usually located in the center to mid-radius of the billet. In the case of unidirectional solidification of superalloys, freckle lines are normally located on the exterior surface of the casting (Figures 1 and 2). Freckles are usually evenly spaced (periodic spacing around the circumference of the castings in Figures 1 and 2 for example) [6]. In the case of DS or SX castings, freckles have been visually observed to run along most of the length of the casting. However, in the case of larger ingots exhibiting internal freckles, it 3 has not been determined yet whether freckles are continuous or not along the whole ingot. In small DS or SX castings, freckles do not normally originate right at the bottom of the casting, but rather at some given distance from the chill. Two adjacent freckles can sometimes merge into a single bigger one (see Figures 2 and 3), but divergence or split is never observed. Some freckle lines can also simply terminate [6]. A close look at a freckle line (Figure 4) shows that a freckle is a linear trail of equiaxed grains [6,9]. Some porosity and feeding shrinkage may also be noted in and adjacent to the freckle line. The misoriented grains associated with freckles do not normally penetrate deeply into the casting because they cannot effectively compete with the rapidly growing [001] preferred orientation [6]. In the case of a killed steel ingot [7], freckles are usually referred to as "A" segregates or "V" segregates due to their inclined orientation on a longitudinal ingot section. "A" segregates are arranged over the surface of cones, the wide part being at the base of the ingot. They extend from the columnar into the equiaxed region at the edge of the ingot. "V" segregates have a less well defined shape and are distributed over inverted cones. They occur at the center of the ingot. Although rather different in appearance, "A" or "V" segregates and freckles exhibit the same chemical characteristics, and their occurrence has been shown to arise from the same fluid flow phenomena. Therefore, since most of the research published so far deals exclusively with "freckles" in directionally solidified alloys, all further discussions in this literature review will focus mainly on freckles. An extensive survey about "A" and "V" segregates has already been published elsewhere [10] 4 Figure 1 : Hollowed Udimet 700 freckled casting (4 in. OD by 3 in. high) [6]. Figure 3 : Longitudinal section of a freckle in Pb-15wt%Sn[ll]. Figure 2 : Mar-M200 (4 in. diam.) ingot showing freckling distribution [6]. Figure 4 : Close-up of one of the freckle lines from Figure 1 [6]. 5 2.1.2. Chemical composition Freckles have been observed to be enriched in those elements which segregate normally (i.e. rejected in the liquid during dendrite growth) and depleted in those elements that segregate inversely [6]. Thus, freckles are shifted toward the eutectic composition. In the case of nickel-based superalloys, the y solid phase usually rejects Al , Hf, Nb, Ta, Ti [12,13]. Table 1 is an example of composition differences between freckle and matrix in superalloy IN718. In this case, the freckles are strongly enriched in Nb and Ti and contain an excessive amount of Laves phase [2,5]. It is interesting to mention that there is no significant chemical composition difference inside a freckle between center and outer radius, or between top and bottom of the casting [6]. Inside Outside Element Freckle Freckle A l 0.43 0.67 Si 0.16 0.12 Ti 1.33 0.97 Cr 17.36 18.58 Fe 15.23 17.62 N i 52.55 53.19 Nb 9.43 5.46 Mo 3.51 3.38 Table 1: Compositions (in wt%) of a freckle and surrounding matrix area in IN718 [5]. 2,1.3, Influencing factors The probability of occurrence of freckles has been observed to depend on several factors. 2.1.3.1 .Shape and geometry The number of freckle trails has been reported to be proportional to the ingot cross-section area [2,6]. Moreover, it would seem that there exists a mirumum freckling area below which freckles are normally not present, such as in small diameter rods [6]. 6 It has also been observed that freckles are least likely to form in regions where the cross-section area of a part with a complex shape is increasing rapidly [6]. In the case of arc-melted superalloys ingots, ESR ingots have been observed to be more prone to freckling than V A R ingots. This observation has been related to the V-shaped pool and mushy zone profile in ESR, as opposed to a U-shaped profile in V A R [5]. 2.1.3.2.Solidification parameters and operating conditions Many publications also suggest that freckling can be significantly reduced and even avoided by allowing steeper thermal gradients and faster solidification rates. This result has been linked to the observation that freckles are reduced by a shorter mushy zone and a shallower liquid pool [2,5,6,14]. On the other hand, high power input for higher melting rates in V A R or ESR produces a strong electromagnetic force field, inducing liquid convection which could result in a higher degree of channel segregation [2,5,15]. 2.1.3.3 .Alloy composition Freckling is highly dependent on alloy composition. Experiments involving binary alloys showed that, below a certain alloying limit, no freckles develop. Above this limit, the number of freckles seems to increase with the solute concentration. For example, in Ni-Al castings, no freckles were found at compositions of lwt% and 5wt%Al, a few freckle grains were noted at 8wt%Al, and there were numerous freckles at 10wt%Al [6]. The kind of alloying element is also important. Superalloys with high aluminum (segregating normally) or tungsten (segregating inversely) levels are reported to be more freckle prone [6]. 7 2.1.4. Physical properties of freckles Due to their relatively large size (diameter over 1mm) and the negligible solid state diffusion rate of some alloying elements (such as Nb in IN718), freckles are virtually non removable by any amount of thermomechanical processing [2,5,12]. Moreover, freckles even tend to remelt during hot isostatic pressing (HIP), homogenizing or solutioning heat treatments, when the temperature is raised close to the eutectic temperature in order to dissolve the Laves precipitates in the casting. In any case, equiaxed grains cannot be reoriented to recover DS or SX castings. The exact loss of properties due to freckles has yet to be quantified. To date, no extensive data on the mechanical properties of freckled castings has been published. It has been reported however that freckled specimens showed poor ductility and a reduction of about 30% in yield strength [5]. Moreover, without necessarily be deleterious, trails of random equiaxed grains should not improve fatigue life, creep properties or corrosion resistance of DS and SX castings. 2.1.5. Cost of freckles Freckles in superalloy castings have been considered a totally unacceptable defect since the 1960's, when they were first linked to the failure of some military engines. In view of the physical properties described above, freckled castings have to be scrapped. When comparing the cost of an acceptable DS or SX blade (over $2000 as-cast) to its raw material value (about $30), and knowing the rejection rates (about 30% for small blades, up to 70% for bigger castings) due to various defects (freckles, spurious grains [3,16], columnar to equiaxed transition (GET) [17,18], dimensional misfit, etc.), the importance of understanding and tackling the problem of freckles becomes obvious. Moreover, any manufacturer who would accidentally produce freckles in a single one arc-melted ingot would automatically be disqualified as a supplier. 8 2.1,6. Center-segregation and freckles It is important to mention that it is not uncommon for cases of center-segregation to be mistakenly identified as freckles, especially in the case of forged ingots. Indeed, center macrosegregates, usually initially in the form of round pockets after casting, can look like fine pencil-shaped freckles after forging, as they etch differently from the surrounding matrix. Although center-segregation and freckling both arise from interdendritic fluid flow (the case of freckles will be discussed more in detail in the following sections of this thesis), they differ on one fundamental aspect - the density of the interdendritic liquid. While freckles result from the upward flow of a lighter liquid (see chapter 2.2), center-segregation is usually the result of a slow downward motion and accumulation of a heavier interdendritic liquid toward the bottom of a U-shaped mushy zone. Thus, these are two very different types of macrosegregation. Their elimination from industrial castings will therefore require different actions. Mistaking one defect for the other might lead to rather unexpected research interpretations. 9 2.2. M E C H A N I S M S O F F R E C K L E F O R M A T I O N Although freckles can develop in several different types of casting, all the research published to date has dealt almost exclusively with one type of casting, the vertical upward directional solidification of binary alloys. The parameters related to this type of solidification (thermal gradient, growth rate, solute concentration, etc.) are only one-dimensional, making the interpretation of experimental results and the explanation of the freckling phenomenon more straightforward and simple. This is the first step toward the extrapolation to other more complicated systems such as vertical downward or horizontal directional solidification, or ingots, of complex alloys. The following discussion in this review will therefore be concerned primarily with the vertical upward directional solidification of binary alloys, unless otherwise stated. 2.2.1. Freckling theory 2.2.1.1 .Density inversion theory The vertical upward directional solidification of a binary alloy is presented schematically in Figure 5. The heat flow is vertical downward (for example, chill plate at the bottom and hot furnace at the top), creating a vertical thermal gradient along the casting. The casting can be separated into three distinct regions : (1) The bottom region is fully solid. There is no bulk movement and solid-state diffusion is usually considered negligible [19]. (2) The top region is fully liquid. It is usually considered uniform in solute concentration because of liquid convection. (3) The middle region is the dendritic mushy zone. At the height of the dendrite tips, the solute concentration in the liquid is CL, slightly greater than C0 (average composition of 10 the casting), due to solute rejection. Deeper in the mushy zone (lower height), the solute concentration in the interdendritic liquid increases up to the eutectic concentration CE at the mushy/solid interface. Thus, in addition to a thermal gradient, there also exits a variable solute concentration gradient in the liquid between the bottom of the mushy zone and the top of the casting. Moreover, the density, p, of a liquid alloy can be considered to be dependent on its temperature and its solute concentration. In this case, p is usually expressed in the following form [8,9,20,21,22,23]: p = p 0 x[l-a.(r-7; 0 )-p.(C-C 0 )] [Eq.l] In the case of most metallic alloys and analog systems, a is always positive (i.e. the density decreases when the temperature increases). However, p can be either positive or negative, usually depending on the relative densities of the solute and the solvent in binary alloys, and on the segregation sign. In the cases where the rejected solute is lighter than the solvent (for example, Ni-10wt%Al), p is usually positive. The combined influence of the temperature and concentration profiles in the liquid and mushy zone can lead to a density profile similar to that in Figure 6. It can be seen that the solute gradient in the mushy zone has a driving effect on the density inversion whereas temperature as a stabilizing effect. It is interesting to note that the same kind of density profile can be obtained when the solute is preferentially solidified in the dendrites (liquid depleted in solute) and is heavier than the solvent (like tungsten or rhenium for example [3]). However, the following discussions will usually consider the former more documented case of a lighter normally segregating solute. 11 a> 1111111 1111111 (a) Concentration -<-) Thermal V / S z (*) Solutal Density concentrat ion Figure 5 : Phase diagram and corresponding Figure 6 ; Schematic concentration (a) and casting geometry in vertical upward directional density (b) profiles above and below the solidification of a binary alloy [9]. dendritic front [24]. Given such a density profile, it can be seen that the interdendritic liquid lower in the mushy zone (enriched in solute) is less dense than the liquid at the dendrites tips. This is a case of density inversion at the growth front. This system is inherently unstable. This instability can lead to some form of fluid convection in order to reduce the potential energy of the system [9]. This phenomenon is known as "thermosolutal" or "double diffusive" convection, since it arises from the influence of both thermal and solute concentration gradients. It is now widely agreed that thermosolutal convection is the cause of freckling [9,19,22,25]. It is interesting to notice that this phenomenon is not unique to metallurgy and is believed to be responsible for convective flows in oceanographic ("salt fingers"), geological and even stellar contexts [14]. 2.2.1.2.Driving force It is possible to quantify the extend of the density inversion phenomenon. The calculation in a simple case has been done by Giamei [8]. In the case of vertical directional solidification, the casting can be considered as a stack of thin layers of thickness dz, temperature T(z), and composition C(z). By knowing the temperature 12 and concentration profiles and using equation [Eq.l], the density p(z), and thus the mass of each layer, is determined. By integrating over the whole liquid domain, the absolute gravitational potential energy of the system is calculated. Then, by mathematically redistributing the same liquid layers so that the heaviest layers are now located at the bottom of the stack, and integrating again, the minimal gravitational potential energy is found. The difference A T of these two energies is a measure of the maximum driving force for instability and convection in the system, called "freckle potential" [8]. In the case of a linear density profile similar to that in Figure 6(b), a schematic profile of the freckle potential as a function of concentration is given in Figure 7. fa) T. Figure 7 : (a) Schematic phase diagram (b) Corresponding freckle potential [8]. <b) c t C e, c f f COMPOSITION It was found that A T is proportional to \IG . Thus, an increase in the thermal gradient decreases the freckle potential. This has been confirmed experimentally and industrially. Figure 7(b) also shows that the freckle potential is maximum around the solid solubility limit C*, and is zero at C=0 (pure element) and C=Qj (no mushy zone). Thus, if a critical freckle potential A T * must be exceeded before freckling will occur, then freckling should be confined to a definite 13 composition range that includes the maximum solid solubility limit [8].This is consistent with experimental observations onNi-Al alloys [6]. Giamei has refined the freckle potential calculation in order to take into account the variations of liquid fraction with height in the mushy zone due to the dendrite array. The result of the calculation in this more complex case is that the freckle potential still varies as the inverse square of the thermal gradient and is maximum at the solid solubility limit [8]. However, no calculation has been done for alloys containing more than two elements. 2,2,2, Freckle initiation and growth The following discussion about freckle initiation and growth is mostly based on the works of Hellawell, Sample and Sarazin [9,14,24,26,27,28]. Figure 8 illustrates freckle initiation and growth, based on a simplification of the density inversion theory. In Figure 8(a), a less dense solute enriched "boundary layer" lies at the top of the mushy zone below a heavier bulk liquid. Figure 8 : Schematic representation of liquid layer perturbation leading to plume and freckle formation [9]. 14 This system is metastable. Under a small perturbation, this system will tend to loose its excess gravitational potential energy. The perturbation itself depicted in Figure 8(b) is unstable. Indeed, for metals or analog systems, the rate of solute diffusion is negligible compared to thermal diffusivity. Thus, the protuberance can still be considered solute enriched, but also close to thermal equilibrium with the hotter bulk liquid. Thus the density difference is further increased, enhancing the growth of the perturbation (Figure 8(c)). It is to be noted that, as long as the perturbation is fed by the liquid of the boundary layer at the dendrite tips, the dendrites located directly underneath the perturbation tend to grow a little faster and to produce a small hillock (Figure 8(c)). However, as the perturbation keeps growing, it begins to be also fed by interdendritic liquid from deeper in the mushy zone. This interdendritic liquid is closer to the eutectic composition and is quickly heated as it rises. This liquid is no longer in thermodynamic equilibrium with the upper parts of the dendrites and tends to melt a channel upward. The very motion of the upward fluid flow could also be strong enough to erode the dendrite arms. Thus, the dendrites located underneath the initial perturbation now tend to disappear by melting and/or erosion (see Figure 8(d)). This decreases the resistance to fluid flow, causing even further erosion and remelting [29]. A liquid channel is thus created in the mushy zone and thermosolutal convection is established. This entire nucleation event has been observed to occur within seconds in the case of NH 4C1-H 20 castings [14]. The overall convection pattern associated with a freckle is shown in Figure 8(e). The light interdendritic liquid flowing through the mushy zone to feed the freckle plume is slowly replaced by the heavier fluid from the bulk liquid. This results in a quasi steady state flow pattern which can persist until the dendritic interface reaches the top of the casting [14]. 15 It is also to be noted that the dendritic crystal fragments eroded by the freckle flow can either be ejected upward (and remelt) or sink to the bottom of the channel depending on local fluid velocities. It would seem that the equiaxed grain structure of the freckles in the solidified casting results from these polycrystalline debris collecting at the bottom of the channels [9]. Another explanation for the equiaxed structure of freckle trails might also be the following : freckles, with their higher fraction of eutectic, would tend to partially remelt during solutioning heat treatment, yielding local recrystallization upon cooling. However, comparison of a freckled casting before and after heat treatment would be required to validate this second theory. The mechanisms described above were confirmed by experimental observations of NH 4C1-H 20 castings, as will be explained in the next part about analog systems. A similar step-by-step explanation of the initiation and growth and "A" and "V" segregates has been described elsewhere by Beech and co-workers from observations on Al-21wt%Cu alloys [29]. The main difference between Beech's and Hellawell's theory is the following: in the case of freckles, fluid flow and channels originate together and are closely linked, whereas in the case of "A" and "V" segregates, convective fluid flow is assumed to already exist and enlarge preferential channels resulting from irregularities in the growth front. Both theories have been experimentally validated, but certainly apply to different types of castings (SX or DS blades versus large killed steel ingots). 2.2.3. Fluid flow velocity in freckles As a first approach, one might consider plume flow to be analogous to flow along a tube. Neglecting end effects, this gives a parabolic velocity distribution in the plume described by the Poiseuille equation [24]. However, instead of a no-slip condition at the plume walls, the plume is 16 assumed to drag some of the quiescent bulk liquid, causing the radial velocity to decay exponentially around the plume, resulting in a global velocity distribution shown schematically in Figure 9. Figure 9 : Schematic of the velocity profile and boundary conditions for flow of a cylindrical plume of liquid through a quiescent liquid of higher viscosity [24]. o Plume Bulk A 7?, V1 = VFW d V / d r = 0 - - L - ^ i — V, = 0 The maximum velocity at the center of the plume is given by a modified Poiseuille equation [9,24] : 4r( - ( l n ^ ) r2 r [Eq.2] Where Ap is of the form Ap = - p 0 x (a. AT+ p. AC) (see [Eq.l]) rj = plume/freckle radius (typically 0.5mm) r2 = plume drag radius Taking r2 equal to half the inter-plume spacing (typically r 2 « 5 m m ) yields velocities which are in very good agreement with experimental observations ( F m a x « 7 m m / s in NH4CI-H2O and F m a x « l m m / s in organic succinonitrile-ethanol systems). Extrapolation to the Pb-Sn or Pb-Sb metallic systems yields values in the range of lOOmm/s [9], which cannot be verified optically. It is to be noted that this calculation took no account of the slow but necessary downward flow which must occur in order to balance the overall convection. 17 Taking into account these velocities V, as well as the observed plume radius r for various systems, calculation of the Reynolds number Re=Vr/\ yields values well below 103 (turbulence limit). This means that freckle plumes are streamlined flow [9]. It is also possible to calculate the thermal and solutal Peclet numbers : Vr ^Thermal = ~ [Eq3] •^Thermal Vr Pesoiutai = — [Eq-4] L'Soltital For the three metallic, aqueous and organic systems, PeTherma^5-25 while PeSolutal-102-5.104 [9]. These numbers indicate that observed flow rates are a little too rapid to allow thermal equilibrium between plume and bulk fluids but are much too rapid for any significant solute diffusion. The same conclusion can be drawn from a large Lewis number : Le=DThermalIDSoiutal. It is usually observed in experimental castings that the heights to come to thermal equilibrium are small by comparison with actual plume heights, but those estimated for solutal equilibrium are actually larger. In any case, when the freckle plumes reach the top of the casting, they have been observed to break down, leaving the bulk liquid uniform in temperature and composition [9]. 18 2 A . A N A L O G S Y S T E M S A N D SPECIAL E X P E R I M E N T S Although channel segregates have been observed to be a major technical problem mainly in superalloys castings or steel ingots, very little research has been done with superalloys or steel. References [3], [6] and [8] were the only papers found in the literature dealing directly with freckle segregates in superalloys. Indeed, most of the research published so far has been based on experiments involving so-called "analog systems". Two analog systems have been widely used : (1) Lead-based binary alloys : Pb-Sn, Pb-Sb. (2) Binary aqueous ammonium chloride : NH 4C1-H 20 There are several advantages in using analog systems : (1) During solidification, analog systems behave very similarly to superalloys or steel: they are subject to segregation, and exhibit dendrites in a mushy zone. (2) All analog systems are binary alloys : they contain only one solute. Therefore, considerations about the phase diagram, or the effects of concentration are easier to analyze, especially since they are not subject to phase precipitation (such as carbides or Laves phase) like superalloys. (3) The melting range of analog systems is at much lower temperature than that of superalloys or steel. Thus, experiments and measurements are easier and cheaper to set up, especially on large scale samples. (4) Aqueous ammonium chloride is a transparent system. This enables direct visual observation of the mushy zone, growth front, convection patterns and freckle formation during solidification. This chapter will present the various analog systems and their utilization in research about freckles, along with the significant results of some experiments. 19 2.3.1. Lead-based systems Lead-based systems have been widely used to simulate interdendritic fluid flow and freckles [9,11,14,24,27,28,30,31]. These systems were chosen to simulate freckling because tin and antimony segregate normally and are substantially less dense than lead. Relevant phase diagrams are shown in Figure 10. Figure 10 : Phase diagrams: (a) Pb-Sn, (b) Pb-Sb [28]. - 300 200k 2 100 5* \ J l l l l l i f t . * . v-»y " Cl>> +<P-Sn> 80 40 WEIGHT % Sn (a) U 30O Research focused on the behavior of lead-based alloys in the case of vertical directional solidification of cylindrical castings. Directional solidification was accomplished by water cooling the base of the casting and slowly decreasing the furnace temperature. Like for industrial alloys, the actual analysis of the lead-based casting could only be done when the metal had cooled to room temperature. Solidified ingots were sectioned longitudinally (as in Figure 3) or transversally (as in Figure 11), polished, and etched to reveal the final structure of the casting. 20 Figure 11 : Cross section of a Pb-2wt%Sb cylindrical ingot showing freckle dots [28]. It was then possible to measure values for channel diameter, channel length, channel spacing, dendrite spacing, distance between channel initiation and bottom chill, number of freckles, etc. Chemical compositions were also measured, yielding curves such as in Figure 12. Figure 12 : Composition change across a freckle in Pb-17wt%Sn[ll]. Experiments on lead-based alloys have also shown that there exists a range of concentrations where freckling is possible (shown numerically in section 2.2.1.2). The lower limit, below which no freckle is observed, lies between 2.5 and 5wt%Sn in Pb-Sn and between 1 and 2wt%Sb in Pb-Sb. Although no sharp upper limit can be defined, it has been observed that, over 15wt%Sn in Pb-Sn or 3wt%Sb in Pb-Sb, channels become rather diffuse, of irregular section, and confined to the center of specimens [28]. This is similar to the behavior of Ni-Al alloys reported earlier, where freckles start to appear for an aluminum content greater than 5 to 8wt%. This strongly suggests that freckles depend a lot on the proportion of eutectic. 30 , 20 I0h 1 I 0 0.5 1.0 1.5 cm DISTANCE FROM MOLD WALL 21 In any case, since it is not possible to observe visually lead-based systems anymore than superalloys or steel directly during solidification, the experiments involving Pb-Sn or Pb-Sb are mainly a confirmation of the proposed theories. They do not really provide a better insight to freckling in industrial alloys. 2.3.2f Aqueous ammonium chloride In order to provide the visual insight missing with lead-based systems and superalloys or steel, most experiments were carried out using the aqueous ammonium chloride analog system [7,8,9,14,22,24,26,27,32,33]. The phase diagram of NH 4C1-H 20 is given in Figure 13. 100 • i i 1 — i 1— 8 0 J / 6 0 7 " it! 4 0 1 Figure 13 : Phase diagram of 5 20 L / ! NH.CI-L NH 4 C1-H 2 0 [27]. i 0 ! 1 - 2 0 • 1 1 1 1 O 10 20 30 4 0 50 60 WEIGHT «7„ NH 4 CI The practical range of compositions lies around 60 to 70wt% of "solute" H 2 0 (dotted line on the diagram). Water is the species that is rejected by NH4C1 freezing dendrites. It is interesting to note that the NH 4C1-H 20 system shows no solid solubility, resulting in a greater liquid fraction in the mushy zone, and therefore a greater permeability of the dendritic array [9]. This case is particularly favorable to fluid flow and freckle development. 22 In addition to a low melting point, the aqueous ammonium chloride system has a significant advantage : the solidified NH4C1 is opaque, the mushy zone is translucent and the liquid is clear, enabling direct visual observation when the casting is contained in a transparent mold (acrylic, Pyrex, or Plexiglas). Castings are usually made by pouring NH 4C1-H 20 at 80-90°C (enough superheat to avoid equiaxed solidification) into the mold. Mold cooling is controlled by a flow of liquid nitrogen (-196°C) through copper attachments. Slabs (with side chill) to observe "A" segregates [7,33] (Figure 14) as well as cylinders to observe freckles [8,9] (Figure 15) have been cast. In addition to enabling direct observation of the growing dendritic front in real time (such as the small hillocks around the channel openings in Figure 16), the NH 4 Cl-H z O system also provides the possibility of observing the fluid flow patterns in the casting : (1) The refractive index of the liquid depends on the composition and temperature. Thus freckle plumes, which have a different temperature and composition from the bulk liquid, are readily visible (Figure 17). (2) Fluid flow can also be traced by additions of a colored marker, such as red potassium permanganate crystals K M n 0 4 (Figure 14). As seen in Figure 17, the upper part of the freckle plumes has been observed to be helical and to rotate about its axis. A parallel may be drawn with plumes of smoke rising into still air. Helices are supposed to develop to allow a more efficient downward displacement of the surrounding liquid [9]. There is presently no way of knowing if such helical flow occurs in metals since it does not influence channel propagation. In such transparent castings, the entrainment of broken dendrite tips in the plume flows was also observed [9]. 23 Figure 14 : NH 4 C1-H 2 0 ingot casting, Figure 15 : Example of apparatus used for studying showing colored solution pouring out unidirectional solidification of NH 4C1-H 20 [8]. from behind the dendritic front through "A" segregate channels [7]. Figure 16 : Side view of a channel at the mold wall, showing erosion [8]. lOmrn Figure 17 : General view of plumes in NH 4Cl-70wt%H 2O [9]. 24 Several variations of the experiments involving transparent analog systems have also been investigated and are briefly reported below. (All these experiments provided the base for the theory of freckle formation described earlier.) (1) When the mold was tipped about the vertical, plumes were observed to keep rising antiparallel to the direction of gravity, independently of the mold angle [8]. (2) By locally thermally insulating part of the bottom of the mold, the dendritic front was no longer horizontal flat but convex or concave, depending on the insulation design. Freckles were observed to develop only at the outer surface of the ingot when the mushy zone is concave, whereas freckles occurred only in the interior of the ingot with a convex mushy zone [8]. This suggest that the less dense liquid flows interdendritically to the highest region in the mushy zone, producing jets preferentially in this area [8]. (3) Increasing the thermal gradient and growth rate during solidification has also been shown to eliminate existing freckles [8]. (4) It has also been attempted to create artificial freckle channels by carefully drilling a 1mm diameter hole in the mushy zone with a fine sharpened silica tube. However, the channels created in this manner failed to propagate and were soon grown over [14]. The same silica tube was used to draw up a small amount of liquid just ahead of the dendrite tips. This created an artificial perturbation that always led to the formation of a channel at the exact position where the plume had been started [14]. (5) Other experiments, such as mold spinning, or reduced gravity, described later in this review, have also been conducted to study the possibility of freckle elimination. 25 2.3.3. Other analog systems Channel segregation has also been studied using other kinds of analog systems. Most of the calculations by Flemings and co-workers [1,34,35,36,37] were validated by castings of Al-4.5wt%Cu alloys. Al-21wt%Cu and Al-25wt%Ge have also been used to study the formation of "V" segregates [29]. In addition to metallic and aqueous systems, Hellawell et al. [9] used the organic succinonitrile-ethanol system (NCCH 2 CH 2 CN-C 2 H 5 OH), also referred to as SCN-EtOH. Most of their organic experiments were conducted with SCN-15wt%EtOH. However, results were not considered as satisfying as with NH 4C1-H 20, due to lower refractive indices, ethanol volatility, monotectic in the phase diagram, etc. A very interesting observation has nevertheless been made by Hellawell et al. [9]: by comparing the three analog systems Pb-Sn, NH 4C1-H 20 and SCN-EtOH, it has been found that, although all three materials exhibit rather different physical properties (thermal and solutal diffusivity, viscosity, density, etc.), the freckle radius and channel spacing were all identical (about 0.5mm and 5-10mm respectively). Freckles in superalloys and steel also exhibit the same dimensional characteristics. 2.3.4. Experiments with superalloys Although most of the research has been related to analog systems, a few experiments have actually been carried out with superalloys [3,6,8,38]. A cylindrical ingot of MAR-M200 (1.25in. diameter) was directionally solidified with its axis tipped at 35° about the vertical. The result is presented in Figure 18, showing vertical freckle lines joining together to form a single freckle line along the uppermost portion of the cylinder. 26 The underside of the ingot was freckle free. This confirms that freckles in superalloys are due to vertical upward fluid flow [8]. Figure 18 : Freckle lines along the uppermost portion of the cylindrical surface of a Mar-M200 ingot, directionally solidified with its axis tipped 35° from horizontal [8]. Two ingots were also solidified under the same conditions [8] : Ni-20at%Al shows profuse freckling, while Ni-13at%Ta shows no freckling at all. Since aluminum and tantalum segregate similarly in nickel, this experiment emphasizes the importance of the density of the 3 3 rejected solute (Al is much lighter (p=2.7g/cm ) and Ta is much heavier (p=16.6g/cm ) than Ni (p=8.9g/cm3)). Comparisons between superalloys have also been made : it was found, in investment casting of single-crystals, that Ni-Al alloys, because of their higher thermal conductivities, allowed higher cooling rates and lower temperature gradients than Rene N5, resulting in a lower tendency to form freckles [38]. The formation of freckles, as well as the nucleation of misoriented grains, in directional solidification of nickel-base single crystals was also investigated over a wide range of chemical compositions, imposed thermal gradients and growth rates [3]. The alloys contained Al , Cr, Co, 27 Hf, Re, Ta, W and Ni and were solidified in furnaces such as depicted in Figure 19 at rates varying from 4.23x10"4 to 1.13xl0"2 cm/s under temperature gradients at the solidification front covering the range 0.3-140°C/cm. SaptaHay Figure 19 : Schematic of commercial furnace used for directional solidification of cluster molds [3]. LI I v DiimiiMi It was found that high level of Re or W (heavy elements depleted in interdendritic fluid) or low levels of Ta (heavy element rejected in interdendritic fluid) increased the potential for density inversion, and thus the freckle probability [3]. It was also found that no freckle occurred in the castings showing a primary dendrite arm spacing lower than 320p.m (corresponding to a cooling rate GxR of 0.1°C/s) for a given alloy [3]. This can be related to the permeability of the mushy zone, which must be high enough to allow fluid flow, for a given density inversion driving force. It is important to notice that in all these experiments, the interpretation relative to the effects of various alloying elements such as A l , W or Ta is purely qualitative. Moreover, precipitation of various phases (carbides, Laves) common to most superalloys, was never considered. 28 2.3.5. Special experiments The following special experiments can bring a new insight on possible original ways to prevent freckling. 2.3.5.1 .Low gravity experiments Since freckling, due to thermosolutal convection, is driven by a density inversion phenomenon, it is strongly influenced by the action of gravity. N A S A founded some research to examine the effect of "low gravity" on freckle formation [32]. Directional solidification of NH4Cl-72wt%H20 was carried out aboard an aircraft flying along parabolas, yielding up to 30s of approximately 10"2g alternating with up to 2g during the 90s pullout. Solidification was carried continuously during the alternating low-gravity and high-gravity cycles. It was observed that the low-g runs did not plume, and in fact, showed convective fields to be indiscernible, whereas Ig runs, and especially pullout periods (2g), showed rather strong convection [32]. This can be linked directly to the Rayleigh number (proportional to g) which would be lower than its critical value in the low-g runs, indicating the stop of any convection. In any case, except in space, long time low-g conditions are not possible. Thus, this is not a practical solution to eliminate freckling at an industrial scale. 2.3.5.2.Mold rotation Mold rotation as a way of preventing freckle formation has been researched both experimentally [26] and numerically [22]. Sample and Hellawell [26] studied the effect of slow rotation rates (oo<10rpm)(as opposed to centrifugation rates ((D«100rpm)) on the directional solidification of slabs and cylinders of aqueous ammonium chloride. Observations were recorded 29 for stationary vertical, stationary inclined, rotated about the vertical and rotated about an inclined axis systems (for various rotation rates and axis angles). In the case of slab mold, channel segregates appeared in every casting. It has been suggested that rotational movement is not ideal for a narrow slab shape, and that a slow rocking motion about an axis normal to the broad faces would probably be more appropriate then rotation [26]. In the case of a cylindrical mold, results are summarized in Table 2. 1. Stationary, vertical Freckles and vertical plumes after 15 to 20 min. 2. Stationary, inclined Freckles and vertical plumes on upper side. Rotation, vertical axis 3. 5 rpm and 10 rpm Freckles and spiral plumes after about 15 min. Rotation. 30 deg. inclination 4. 1 rpm Few central freckles, central plumes. 5. 2.5 rpm trace of central unevenness. 6. 5 rpm No freckles or plumes. 7. 10 rpm Concave front - small freckles at mold walls after 45 min. Rotation, 5 rpm 8. 20 deg. Inclination No freckles or obvious plumes. 9. 10 deg. Inclination Very fine freckles after 45 min. Table 2 : Observations in bottom chilled rotating cylindrical NH 4 C1-H 2 0 casting [26]. In a further similar experiment, freckles were initially allowed to develop in a stationary vertical mold. The mold was then tilted at 30° to the vertical and rotated at 5rpm. The freckle plumes disappeared almost instantly and the channels were quickly grown over. Later, the mold was returned to its stationary vertical position, and freckles initiated again (although at different positions than before). Thus, it can be seen from Table 2 and this experiment, that a rotation rate of 5rpm on an axis at 30° about the vertical seems to be a very efficient way to prevent freckling [26]. Initially, this result was thought to be due to the ever changing direction of gravity relative 30 to the casting. However, it would seem that freckle elimination is actually due to the translation of bulk liquid across the liquidus front, smoothing any perturbation before it could grow into a "salt finger" and then a freckle [22]. Neilson & Incropera [22] built a numerical model simulating freckling in a cylindrical directional solidification mold rotating about a vertical axis. They found that for steady rotation, channels were predicted to develop in much the same manner as for stationary solidification. Moreover, rotation about a vertical axis even seemed to have a plume stabilizing effect. Intermittent rotation (spin-up and spin-down cycles) was also investigated and found to substantially reduce freckling. However, freckles are still expected to appear in this case, at the centerline and at the outer radius of the casting [22]. These results are in agreement with those presented in Table 2. 2A_ F R E C K L E M O D E L S A N D CRITERIA 2.4.1. Models basis Several models related to freckle formation have been developed. The two major computer models are those of Bennon & Incropera [39,40,41,42] and Felicelli, Heinrich & Poirier [19,20,21]. A few other models have also been presented elsewhere [23,25,33]. All these models are two-dimensional and are solved by meshing the casting and using finite differences or finite elements methods. They all consider the directional solidification of a binary alloy. They model thermosolutal convection, and the effects of the fluid flow in the liquid and mushy zone on the solidification front (only the effects of remelting arid not those of dendrite erosion are considered). 31 In order to remain as simple as possible, all these models were written under the same following assumptions [20,41] : (1) the fluid flow is always laminar, (2) the fluid is of constant viscosity, (3) liquid and solid have equal densities (i.e. fluid motion induced by solidification shrinkage is considered negligible compared to buoyancy driven convection), (4) density is constant, except in the buoyancy terms of the momentum equations, (5) no solute diffusion and no bulk movement in the solid, (6) the thermal properties are constant and equal in both the liquid and the solid, (7) local thermodynamic equilibrium exists (i.e. concentration in the interdendritic liquid and in the solidifying dendrites are linked by the phase diagram). Al l the models are based on the same continuum equations governing the conservation of mass, momentum (x and z directions), energy and solute. The main variables are the density, the velocity, the enthalpy and the solute mass fraction. For further details, the reader should refer directly to the corresponding publications. 2.4.2. Models results 2.4.2.1 .Bennon & Incropera Bennon & Incropera's model [39,40,41,42] considers the horizontal solidification of aqueous ammonium chloride (NH4Cl-70wt%H2O) against a vertical chilled wall. This calculation confirms the theories previously presented. As water-rich interdendritic fluids, adjacent to the solidus front, rise under the action of positive buoyancy forces, they are 32 forced, by the constraint of overall mass conservation, to deflect into warmer regions of the mushy zone near the liquidus line. As the temperature of these solute-rich fluids increases, localized melting of dendrites occurs, increasing the liquid volume fraction and permeability, thus enhancing further flow, erosion and melting [39]. This leads to "A" segregates in the final solidified ingot (Figure 20). The localized solute enrichment of these segregation bands comes at the expense of solute depleted regions near the bottom and at several small regions near the center of the mushy region [39]. Further details about the evolution of the fluid flow velocity, liquid composition and profile of the growth front from start to finish have also been reported. Figure 20 ; Macrosegregation after complete solidification of a NH 4Cl-70wt%H 2O slab (vertical left side chill). Bennon & Incropera's model [39]. f H l O I 1 0.47-0.56 0.56-0.64 ?m 0.64-0.72 0.72-0.80 2.4.2.2.Felicelli et al. Felicelli et al. developed a model [19,20,21] simulating the formation of freckles in vertical upward directional solidification of Pb-10wt%Sn. In Figure 21, the observed channel is of the size of one to several primary dendrite spacing. This channel was induced by introducing a small initial perturbation in the .33 concentration of the melt along the vertical centerline (similar to the experiments when liquid was drawn up in a fine silica tube [9,14,24]) and letting the system evolve thereafter [20]. In the streamline contours (Figure 21(b)), one can clearly distinguish the rising liquid plume associated with the channel. If the system is left to evolve undisturbed (i.e. no initial perturbation), channel formation occurs preferentially on the sides of the casting, as experimentally observed with superalloys [6]. The reason for the growth of channels along the side walls is believed to be the condition of zero horizontal velocity in the fluid at the walls [20]. This has been confirmed by artificially imposing V=0 at the center of the casting, resulting in the growth of channels at the centerline as well as at the side walls. This model was also able to predict the experimentally observed patterns in the case of a convex or concave mushy zone, or of an ingot tilted about the direction of gravity. Figure 21 : Freckle formation in Pb-10wt%Sn (bottom chill), (a) Fraction liquid (b) Streamlines (Size of the figure : 2.5mmx4.5mm) Felicelli et al. model [20]. CONTOUR FROM 0.05 TO 0.95 CONTOUR FROM -3.6 TO 3.6 BY 0.05 BY 0.4 (a) (b) Felicelli et al. drew the following conclusions about freckle formation from observations made with their model [20]: 34 (1) The isotherms are practically flat because of a low Prandtl number for Pb-Sn (Pr=v/Dj). Thus, the convection is almost entirely driven by the solute concentration field. (2) The convection in the bulk liquid penetrates deeply into the mushy zone, although the velocities decrease by two or three orders of magnitude below the upper 20% of the mushy zone. (3) Because of the slow solute diffusion, the columns of solute rich liquid, rising from the bottom of the mushy zone in the casting by convection, tend to melt solidified dendrites in order to maintain their thermodynamic equilibrium. It is thus possible to observe the formation of a channel, developing backwards from the tip of the dendrites toward the bottom of the mushy zone (or, in a different referential, the mushy zone grows around the solute-rich liquid columns). (4) Although they may originate at a different location, the channels tend to grow along regions of solute accumulation. For example, the solute accumulates along the side walls because of the limitation of lateral flow. Such solute accumulation, enhancing channel propagation, could also occur in regions of high anisotropy in the mushy zone (permeability KX«KZ), as might be the case in grain boundaries, dendrite misalignments, dendrite debris deposits, etc. In any case, Felicelli et al. admit themselves that their model is mainly qualitative, and that quantitative data are yet to be evaluated. Particularly, a critical aspect of the model lies in the choice of the permeability function, which must be predicted properly, especially in the neighborhood of the dendrite tips [19]. Unfortunately, it is in this region, where the volume fraction liquid is above 0.6, that the value of the permeability is least known [31]. Various mathematical formulations of the permeability are presented in appendix A. 35 2.4.3. Freckling criteria One of the main goals of the research on freckles is to develop a reliable criterion for freckling. This will enable to know whether or not a given alloy is going to exhibit freckles for given casting geometry and conditions (thermal gradient, casting speed, etc.). Several of these criteria have been developed that can be readily used in conjunction with various "maps" (temperature, fraction liquid, fluid velocity, etc.) of complex-shaped castings computed by most of the current casting models such as Magma or Procast. 2.4.3.1 .Initial perturbation and Rayleigh number As mentioned earlier, the mechanism of instability leading to freckles due to density inversion can be briefly described as follow : if a small parcel of fluid is displaced upwards, its temperature equilibrates quickly but its solute content does not, since the diffusivity of heat is some two orders of magnitude greater than that of solute. The fluid parcel is thus solute richer than its surroundings, becoming increasingly buoyant as it continues to accelerate upwards. This situation can be described numerically by a dimensionless thermosolutal Rayleigh number Ra-p/g [14]: The numerator and denominator in equation [5] both have equivalent units of pressure per unit area (N/m4). The numerator can be physically thought of as the buoyant pressure exerted upward due to the density inversion. The denominator, on the contrary, can be thought of as the opposing or restraining pressure due to the fluid viscosity and diffusion of heat in the system [14]. (Although it was not clearly explained in the literature, it would seem that Dj^^, rather 36 than DSoiulah was chosen to characterize the evolution of the system, because of the much faster diffusion rate of heat over that of solute.) Equation [Eq.5] can be used as a criterion for freckling : the system is considered to perturb when Rajys exceeds some critical value. Sarazin and Hellawell [9,14] found that, with h being given the value of the primary dendrite arm spacing in the mushy zone (h=X{), the critical value of the thermosolutal Rayleigh number was RaT/g*>l. It is interesting to note that this value of 1 corresponds to the buoyant pressure (numerator) overcoming the restraining pressure (denominator). Since the dendrite spacing can be linked to operating parameters such as the growth rate or the temperature gradient [14], the thermosolutal Rayleigh number can be calculated for a given alloy and given operation conditions and used as a criterion for the probability of freckling. It is important to mention that this criterion can be linked even more closely to the physical characteristics of a mushy zone. Indeed, for a network of permeability K, in a unidirectional density gradient, the thermosolutal Rayleigh number can actually be written : dp R^TIS = r~T [Eq-6] TIS T\DT/Kk\ 1 4 J This modification takes into account the fact that a dense network of dendrites (of low permeability K) will dampen fluid flow. In this case, freckle formation would require a higher density inversion as driving force. (Permeability expressions have been calculated in the past (appendix A) and usually depend on dendrite spacing and fraction liquid.) Fowler [23] developed a similar criterion, in which RaT/s could be compared to two (instead of one) critical Rayleigh numbers Rx and R2. Fowler stated that thermosolutal convection would occur when Ram > Rx and that freckle formation would only be expected for RaT/g>R2>R\. 37 It is important to remember that these criteria based on the local thermoslutal Rayleigh number are not extensive criteria. For example, it was reported in section 2.1.3.1 that freckling does not occur in small castings having a cross-section area smaller than the minimum freckle area. However, this effect does not appear in the Rayleigh number criteria. 2.4.3.2.Effect of growth rate and thermal gradient on freckling The following criterion was developed in the case of steady state directional solidification (constant thermal gradient G and growth rate R) [8]. As mentioned earlier, freckles formation is not instantaneous. It can be considered that there exists, for a given alloy, a critical time Ar* necessary for the density inversion induced flow to appear and erode and/or melt a channel in the mushy zone. Thus, if the local solidification time Ar is greater than At*, a freckle line can develop. T — T AT The local solidification time (LST) can be written : Ar = —— — = [Eq.7] RG RG * AT To avoid freckles, Ar must be lower than Ar : R>—;— [Eq.8] Ar G Moreover, it has also been shown that for G>G (critical thermal gradient), freckling is suppressed because there is insufficient gravitational potential energy available to drive the convective plumes (see chapter 2.2.1.2) [8]. Finally, to avoid equiaxed solidification, steady state unidirectional solidification is KG restricted to growth rates such as: R < ^ s [Eq.9] (where KT= Thermal conductivity (in J/ C.m.s) G$ = Thermal gradient in the solid at the solidus (in °C/m) AH = Latent heat for: Liquid at TUq Solid at Ts0i (in J/m3)) By considering all these equations, it is possible to define a thermal gradient/growth rate domain in which freckles should not appear, as shown in Figure 22. 38 Figure 22 : "No freckle1 operating domain criterion [8]. THERMAL GRADIENT 2.4.3.3.Flemings' criterion Flemings et al. [34,35,36] developed a set of equations describing macrosegregation in castings and ingots, resulting from interdendritic fluid flow during solidification due to thermal contraction and solidification shrinkage. In another article [37], the action of gravity on a fluid of variable density was added to the fluid flow causes. The main basis to these equations is the application of D'Arcy's law to a small volume element in the mushy zone. The end result of these calculations was a local criterion for the onset of freckles. Channel type segregates (e.g. freckles, "A" segregates) are considered to result from a flow instability which occurs at the following critical condition (vectorial equation) [37] : Although complicated in appearance, this critical condition simply states that the formation of a channel segregate or a freckle is possible when the interdendritic fluid flow velocity is greater than the isotherms velocity. v . v r < - l [Eq.10] 8 39 2.4.3 AOther criteria Various criteria yielding maps of freckling probability have also been developed for use in conjunction with casting softwares [43] : GR -The gradient acceleration parameter: (GAP) = [11] LST Gv -The Xue porosity function : (XUE) = — [ 1 2 ] LST (Where LST- local solidification time) Although these two criteria (especially the Xue function) can give a relatively good estimate of the probability of occurrence of defects such as freckles, they are completely empirical and none of them is entirely satisfactory. Additional research is necessary in order to formulate a comprehensive prediction criteria for freckling which would be linked to a physical theory [43]. 40 2.5. L I T E R A T U R E R E V I E W CONCLUSION This chapter is a state-of-the-art literature review focusing on the formation of freckles and channel segregates. These macrosegregates have been encountered and studied in a wide variety of systems, although superalloy casting is the main industrial problem concerned. They appear as solute-rich trails of misoriented grains. They are specific to certain alloy compositions. They result from a density inversion phenomenon in the bulk liquid and mushy zone during casting. This density inversion, when it becomes unstable, creates thermosolutal convection. The associated fluid flow has been found responsible for the melting and erosion of preferential channels in the dendritic array. The characteristic features of channel segregates are very similar in several different materials and the relevant experiments using various analog systems (lead-based alloys, aqueous ammonium chloride, superalloys) have been described. Related numerical results such as the gravitational driving force, the thermosolutal Rayleigh number, the effect of dendrite spacing and permeability, have also been presented, as well as comprehensive computer models. It has been found that small primary dendrite arm spacing (corresponding to a low permeability of the mushy zone), high thermal gradient, high growth rate, high fluid viscosity, reduced gravity and mold rotation about an inclined axis all have beneficial effects on the prevention of freckling. 41 3^ O B J E C T I V E S A N D E X P E R I M E N T A L P R O C E D U R E S ^ L . O B J E C T I V E S In view of the literature review in chapter 2, several observations can be made : (1) Freckling is a common recurring problem for some specific alloys, especially superalloys. (2) Although the superalloy industry is the first concerned by freckles, the vast majority of the research to date has been focusing on analog systems. The little research done on superalloys was merely observational and qualitative. (3) The foremost result of the experimental and modeling research is now widely accepted : freckles, in any system, result from the thermosolutal convection and upward fluid flow generated by density inversion. (4) In the case of superalloys, the large cost of freckles could possibly be substantially reduced and even avoided if a suitable freckling criterion was determined. (5) Such a criterion will only be developed for superalloys through the quantitative study in superalloys of segregation in the mushy zone and its effect on density inversion. The main objective of this thesis is to start this quantitative research with superalloys about freckles. It is to be noted that this thesis is based on the assumption that the current theory linking freckles to density inversion in the liquid is indeed true. The first step of this quantitative research is the following : samples of a few alloys will be chosen for their tendency to form freckles. They will be directionally solidified under vacuum in an induction furnace and quenched. The composition of the quenched interdendritic liquid along the mushy zone in these samples will then be measured by conventional energy dispersion spectrometry (EDX). These compositions will then be transposed into estimated densities 42 through a mathematical model. These results will enable the quantitative evaluation of the density inversion magnitude encountered in superalloys. This procedure will also underline some of the main differences between superalloys and binary systems. Another objective of this thesis is also to estimate directly the differences in composition and density between freckles and bulk matrix on various freckled samples from the industry. The comparison of these freckle compositions with the segregation profiles obtained from the quenched samples should also provide a good estimate of the depth in the mushy zone at which freckles initiate. It is important to remember that all these experimental objectives are directed toward a more ultimate goal: to develop a freckling criterion (similar to the Rayleigh number criterion for example), as well as the corresponding fitting parameters for various superalloys. The experimental setup and procedures used to fulfill these objectives are described in the following sections of this chapter. 3.2. C H O I C E O F T H E A L L O Y S In order to investigate the relationship between density inversion and freckling, five alloy compositions have been chosen in the scope of this thesis : IN718, MAR-M002, MAR-M247, C-276 and T l . The standard compositions and melting range of these alloys are shown in Table 3. IN718 is usually V A R or ESR ingot melted and then forged into turbine disks. It is well known to be freckle prone. IN718, being one of the most common of all superalloys, was therefore a logical choice in this study. 43 MAR-M002 and MAR-M247 are usually used for DS blade castings. MAR-M002 is known to exhibit freckling whereas freckles have never been reported for MAR-M247, although both alloys are relatively similar in composition. The comparison of the segregation patterns of these two alloys should therefore provide a better insight on the amount of density inversion necessary to form freckles. Moreover, measurements on MAR-M247 will be compared to published data in order to validate the experimental procedure of this thesis. C-276, despite its Ni-based composition, does not exhibit the usual y/y' structure characterizing superalloys. Its main application is as pitting and crevice corrosion resistant material in aqueous environments [44]. It is known to be subject to freckling. T l is a tool steel. It is now rarely used because of its high cost, due to its high tungsten content. Freckles were reported in T l billets. Although T l is not a superalloy, its choice in this thesis was driven by the large density difference between W and Fe, and the possible effects of tungsten segregation on freckling. Solidus Liquidus Al C Co Cr Fe Hf Mn Mo Nb Ni Re Ta Ti V W IN 718 1260 °C 1336 °C 0.5 0.03 0.4 19.0 BAL. 3.0 5.5 52.5 1.0 MAR-M002 1249 °C 1365°C 5.5 0.15 10.0 9.0 1.3 BAL. 2.5 1.5 10.0 MAR-M247 1280 °C 1360 °C 5.5 0.15 10.0 8.4 1.4 0.6 BAL. 3.0 1.0 10.0 C-276 1325 °C 1370°C 0.01 15.5 6.0 0.4 16.0 BAL. 4.0 T1 «1320°C* *1440°C * 0.75 4.0 BAL. 0.3 1.1 18.0 Table 3 : Standard compositions (in wt%) and melting range of chosen alloys [44]. (* : estimated from reference [45,46]) 44 3.3. S A M P L E PREPARATION : T H E D S Q F U R N A C E . A schematic diagram of the DSQ (directional solidification and quench) furnace is presented in Figure 23. It is a vacuum induction furnace. The water cooled copper coil is connected to a 5kW, 450kHz power supply. The melting chamber of the furnace is evacuated by a mechanical pump and a diffusion pump. Alloys to be melted were machined into small rods (5.5mm diameter, 60-70mm length). These samples were lowered in the furnace at the bottom of long crucibles. These crucibles were made with a fine alumina tube (approximately 6mm ID, 7mm OD and 285mm length) plugged with alumina paste. The plugs (20-30mm length) at the bottom of the tubes were allowed to dry for over a day at 100°C and cured at 900°C for several hours. Various plug designs were attempted (flat top, conical top, pierced top) in order to enhance DS or SX growth. However, no significant difference was observed in the results of the various plugs. Due to the fairly small size of the samples (required for an efficient quench), thermal data could not be collected during the runs with the actual alloys of interest. However, several temperature profiles of the furnace in steady-state conditions have been recorded with (W-3%Re/W-25%Re) thermocouples during trial runs. These profiles are presented in Figure 24. There is a relatively high degree of uncertainty (up to 150-200°C) concerning the absolute temperature at a given height in the furnace. This effect is thought to be related to the possible "aging" and degradation of the Mo foil susceptor after several runs. In any case, the important parameter relevant to the directional solidification of the samples is the temperature gradient in the solidification zone at the exit of the "hot zone" (i.e. for 100< z < 140mm). This gradient, calculated by linear regression, was very consistent in every run. Its average value is 9.4±0.5°C/mm. 45 Vacuum Seal Quartz Tube Induction Coil Mo foil Susceptor Alumina Thermal Shield Water-cooled Copper Quench PSQ FURNACE Driving Rod To Vacuum Pump Crucible Metal Sample Alumina Plug To Vacuum Pump Figure 23: Schematic diagram of the DSQ furnace. 46 The vacuum gauge readings were always very consistent for every run during the solidification period of the sample : P « 3.10"5 - 5.10"5 Torr. However, it is not possible to know precisely what the exact vacuum pressure was in the immediate proximity of the sample. Vacuum was considered sufficiently good when cast samples would come out of the furnace without any scale or oxide "skin". A typical run would usually follow the sequence described below : (1) The sample, inside the crucible tube, is positioned above the hot zone (z « 0 mm). (2) The cold furnace is evacuated (P » 10"4 Torr). (3) The power in the induction coil is switched on. (4) After one hour, pressure and temperature steady-state is reached. (5) The sample is lowered through the hot zone at low speed (typically 0.025mm/s, hence about 1.5 hour for a total travel distance of 120-150 mm). (6) When the sample is halfway solidified, the crucible is quickly (in about Is) lowered into the center of the water-cooled copper quench. After the furnace had cooled down, the samples were removed, longitudinally polished, and etched to check the efficiency of the directional solidification and quench. From the machining of the sample to this final observation, each run requires about 8 to 10 hours per sample, under favorable conditions. 47 1900 1700 1500 DSQ Temperature Profiles 1300 o o •*•> CO §.1100 E o 900 700 500 40 60 80 100 120 140 160° 180 V e r t i c a l P o s i t i o n z ( m m ) Figure 24 : Various recorded temperature profiles in the DSQ furnace. 48 S A M P L E ANALYSIS : S E M - E D X M E A S U R E M E N T S In order to estimate density inversion magnitudes, the main goal of the experimental part of this thesis is to analyze the chemical composition of the interdendritic liquid along the mushy zone and to derive the corresponding segregation profiles for various alloys. In order to locate interdendritic liquid with precision at various depths in the mushy zone, it was imperative to work on cross-sections of the D S Q samples. This enabled the unambiguous localization of the dendrite cores, and hence, the localization of the interdendritic liquid (see Figure 25). Typical EDX Measurement Areas Dendrite Centers Quenched Interdendritic Liquid Schematic cross-section window of a DSQ sample. Directionally solidified and quenched samples were first cut at both ends to eliminate most of the quenched liquid zone and any "non-directional" area in the solid zone. The top cross-sections of the samples were then repeatedly polished off. This technique enables the study of as many cross-sections as desired at virtually any depth in the mushy zone (cutting the DSQ samples into slices was not feasible due to the relatively short mushy zone lengths (typically 10mm)). (See Figure 26) 49 50 Each cross-section was referenced by its distance z to the bottom of the sample. This distance was measured with a Vernier caliper with a precision of ±25 um (i.e. ±0.25% on a 10mm mushy zone). Special care was taken to determine as accurately as possible the position zLiq of the cross-section showing the first sign of the tip of the dendrites. This position corresponds to the liquidus temperature in the sample at the moment of the quench. Knowing the linear gradient G in the furnace and the actual melting range of the alloy (liquidus and solidus temperatures), it is then possible to calculate the position zSol of the bottom of the mushy zone at temperature TSoi: zSoi = Zuq-(TLiq-TSol)/G [Eq.13] Knowing the position z of each cross-section, it is also possible to back calculate their respective temperature at the moment of the quench : T(z) = TLiq-G.(zLiq-z) [Eq.14] In the rest of this thesis, depth in the mushy zone will usually be referred to by its temperature. For each DSQ sample, each cross-section was carefully polished down to 5u.m diamond and lightly etched with Kalling's II (2g CuC12 + 40ml HC1 + 40-80ml ethanol) or Marble's (10gCuSO4 + 50ml HC1 + 50ml water) etch for visual observation. An example of various cross-sections is given in Figure 27. The number N of primary dendrites was also counted in these pictures. Knowing the actual area A represented on these pictures (depending on the magnification), the primary dendrite arm spacing was then calculated [47]: Xx=(AIN)m [Eq.15] By outlining the actual dendrites, it was also possible to evaluate by image analysis the fraction liquid at various depths (or temperatures) in the mushy zone for some alloys. 51 (a) Quenched liquid (7>1360°C) (b) Dendrite tips (7>1360°C) Figure 27 : Etched cross-sections at various depths in DSQ MAR-M247 (Mag. X51). 52 (c) Middle of the mushy zone (7>1335°C) (d) Middle of the mushy zone (7>1321°C) Figure 27: (continued) 53 In some cases, at the top of the mushy zone where the tip of the dendrites is rather small (as fine as the quenched liquid structure), relevant areas (such as dendrite centers and interdendritic liquid) were marked on the etched samples with microhardness diamond marks (at 500g or lOOOg weight). Before being analyzed, the samples were then slightly polished again on the 5um diamond polishing wheel, enough to eliminate the etched material, but without changing the height z of the cross-section in any significant way. The microhardness marks were still highly visible. This final polishing is mandatory in order to avoid the influence of the etch on the chemical analysis. It is to be noted that once the dendrites were much coarser than the quenched liquid structure, they could easily be distinguished by backscatter on the scanning electron microscope (SEM) screen without any etch. The interdendritic liquid, usually enriched in light elements such as A l or Ti , appears darker than the dendrites. Etching and marking of the areas of interest was no longer necessary. For each sample, each cross-section was observed in the SEM. Once an area of interest was focused upon, it was sampled by EDX microprobe, and a normalized quantitative analysis of the spectrum was calculated. The size of the measurement window was always chosen as large as possible, without overlapping the limits of the area of interest (see Figure 25). This provided representative average measurements, smoothing out any local variation (fine structure in the quenched liquid, carbides, Laves phase, etc.). All the values reported in this thesis are average values, over at least 4 measurement areas. In every case, chemical analysis recorded all the major and minor metallic elements relevant to the standard composition of each alloy. Elements below 0.2wt% in the standard composition were not taken into account. However^ carbon, although a significant alloying 54 element, especially in tool steel T l , could not be incorporated in the quantitative analysis, due to the current limitations of the E D X technology. It is not the topic of this thesis to describe the principles of E D X analysis. For further details, the reader should refer to any relevant text book. However, it is important to mention that under the best possible conditions, the EDX microprobe precision is not expected to achieve better than ±5% relative for the major elements (>5-10wt%). For minor elements (<5wt%), this precision is expected to be much worse, due to the statistical nature of the E D X technology and to poorer precision in the spectrum's peaks analysis. 3 J L D E N S I T Y CONVERSION 3.5.1. Mathematical model Freckling arises from a density inversion in the liquid. Thus, it is necessary to translate the measured chemical compositions into densities at a given temperature. Since no physical measurement of the density was possible in the scope of this thesis's experiments, a mathematical model was used. This model, named "METALS", provided courtesy of National Physical Laboratories NPL (UK), is based on a weighted average of the molar volumes of each pure element forming the alloy (along the same principle, this model is also capable of calculating alloy enthalpies, viscosities, thermal conductivities and diffusivities). This approximation is now a widely accepted approach [48,49,50,51]. A replica of the density calculation of this model was rewritten for this thesis, in order to include a few chemical elements which were not taken into account in "METALS", such as Hf, Re and Sb. Considering 55 that the thermal behavior of pure elements is known (and linear), the basic equations in these models are presented below : At the temperature T (in °C), lower than the liquidus temperature TLiq of the alloy, the molar volume in the solid phase MVS of each pure element / is given by : MV^T) = AtV&S'C) x (1 + a'sx(T-25)) [Eq.15] At the temperature T (in °C), the molar volume in the liquid phase MVL of each pure element i (of melting point Tmp) is given by a similar equation : MV^T) = MV1^) x (1 + a'Lx(T-TUq)) [Eq.16] with MVL(TLiq) = MVL(fmp) x (1 + a!Lx(TLiq-fmp)) [Eq.17] For a given total weight W of an alloy of known composition, the number of mole a' of each element is also known. Thus the density of the alloy in the solid and liquid state, at any given temperature T, can be calculated as follow : p 5 (7)= Wl V&xMV'jT))] [Eq.18] and pL(T)=W/[I&xMriL(Tm [Eq.19] This model is accurate to about 5% according to NPL. For this thesis, it was tested and showed very good agreement with various liquid densities reported in the literature [46,48]. In the case of superalloys, NPL suggested to incorporate the following correction coefficient: PactuarPcaicuiated* (100-1.16xw/%4Q/100 [Eq.20] This correction coefficient has been derived from the comparison of the actual and calculated densities of various alloys at room temperature. Although no extensive verification 56 has been done yet, it is believed that this correction is still valid for melts, and should bring the model's accuracy to about 1 or 2%. However, due to the absence of confirmation, this correction coefficient will not be used in the density calculations. Moreover, as can be seen in the segregation profiles, aluminum does not segregate much. Thus the only effect of this correction would be to slightly shift all the densities toward lower values by a fixed amount for a given alloy. Thus it would not bring a better insight about density inversion. All the densities reported in the rest of this thesis are derived from this model. Note : In the present case of directional solidification, interdendritic liquid at any depth is assumed to be in thermodynamic equilibrium with the solidifying outer core of the dendrites. Thus, at any depth, the interdendritic liquid is at its liquidus temperature, which is also the actual temperature in the furnace at this given depth. Hence, in all the density calculation equations, for any alloy composition, it was considered that T=TLiq. 3,5.2, Model's limitations This model is based on a simple weighted average of the molar volumes - e.g. the more of a light element, the lighter the alloy. Unfortunately, this is not a good approximation in the case of interstitial elements, such as carbon. Addition of carbon may increase the total weight without necessarily increasing the volume. Thus, as the carbon content increases, the density of the alloy could increase, instead of decreasing as predicted by "METALS". This can be fully appreciated on the graph shown in Figure 28, which compares the actual measured density at the liquidus temperature of a Fe-6wt%W-C melt [52], and the calculated corresponding values, for various carbon contents. It can be seen in this graph that the differences between calculated and measured 57 densities may be as high as 0.3g/cm3. The problem lies in the fact that freckles could arise from density differences as low as O.lg/cm . This is lower than the expected precision of "METALS" when interstitial elements are involved. Thus, "METALS" calculations involving elements such as carbon or silicon should probably be regarded as qualitative approximations. M e a s u r e d a n d c a l c u l a t e d d e n s i t y v a r i a t i o n s v e r s u s c a r b o n c o n t e n t , a t t h e l i q u i d u s t e m p e r a t u r e , i n F e - 6 w t % W 0 0.5 1 1.5 2 2.5 Carbon Content (in wt%) Figure 28 : Comparison of calculated and measured densities at the liquidus temperature for Fe-6wt%W-C melts, versus carbon content. (Measured densities reported in [52]; Calculated densities computed by "METALS") 58 4^ R E S U L T S 4 ,L_ D S Q S A M P L E S A typical cast sample is shown on Figure 29. Depending on the alloy and the experimental conditions, cast samples can be single crystals or exhibit 2 or 3 elongated grains. Although some samples started solidifying directionally from the very bottom of the crucible, others exhibited a initial region of competition between several grains. It can be noted in all the samples that the preferred orientation of the dendrites in the cast samples is not exactly vertical (about 5° to 10° off the vertical). This is probably due to the slight initial misorientation of the leading grain(s) at the bottom of the sample. It might be possible to correct this misorientation with the design of a proper grain selector at the bottom of the crucible. However, no special design, other than that of the top of the plug (with no significant result) was attempted. In any case, the slight angle of the dendrites should have no effect at all on the segregation measurements. This angle would however affect the permeability in the vertical direction and should be taken into account in the case of the study of the growth of an actual freckle. The top of the mushy zone (and the tip of the dendrites) was always clearly visible on the lightly etched longitudinal section of the DSQ samples (see Figure 30). The quenched liquid zone exhibited a much finer structure (with ternary and possibly quaternary dendrite arms) than the rather coarse dendritic array of the mushy zone. It can be seen that the top of the mushy zone was always flat horizontal (perpendicular to the sample length), indicating that isotherms in the DSQ furnace are flat and horizontal as expected. 59 Quenched Liquid Dendrite Tips Figure 29 : Etched longitudinal section of a whole DSQ MAR-M002 sample (Mag. X6.5). DS Section Growth Direction i Top of the Crucible Plug Quenched Liquid Dendrite Tips Figure 30 : Etched longitudinal section of the top of the mushy zone in DSQ MAR-M002 (Mag. X22). (same sample as in Figure 29) DS Section Growth Direction I In most experiments, the inside walls of the crucible above the sample were covered with a thin metallic film and small metal beads (about 0.5mm diameter). The corresponding chemical compositions, as well as those of the bulk sample before and after melting, in the case of DSQ MAR-M247, are presented in Table 4. It can be seen that some of the chromium evaporated from the molten sample and deposited in a thin film on the colder walls of the crucible above the hot zone. This is due to the combination of a relatively high temperature in the hot zone, a high vacuum in the furnace and a high partial pressure for Cr. The beads are most likely the result of projections issued from the melt due to flow in the liquid pool or degassing. Light variations in the bulk sample composition between before and after melting, as observed here, are not unusual in superalloys and have been reported elsewhere [6,53]. In any case, the bulk sample composition 61 is not affected significantly, and values are still within the expected precision limits of E D X measurements. Element Thin film Beads Bulk sample (before melting) Bulk sample (after melting) Al 0.15 6.58 6.87 6.58 Ti 0.00 1.01 0.99 1.10 Cr 64.58 7.63 8.47 8.38 Co 9.23 10.14 10.09 10.00 Ni 24.72 57.36 57.03 58.24 Hf 0.45 2.15 2.86 2.10 Ta 0.88 4.54 3.36 3.58 W 0.00 10.59 10.33 10.03 Table 4 : Compositions of the products of a DSQ run with MAR-M247. Various average chemical analysis were also taken along the MAR-M247 and C-276 samples, in the solid, mushy and liquid zones. No significant composition differences were observed between solid and liquid, or between various heights in the solid. This means that there was little or no solute rejected ahead of the growth front. Thus, samples were not subject to any zone refining effect. This is coherent with the usual behavior of superalloys castings. Moreover, composition of the "liquid" zone was also very uniform, confirming either that no macrosegregation took place along the sample, or that the liquid was well mixed. No freckles were observed in the DSQ samples. This was expected, due to the relatively small cross-section area of the samples (about 24mm ), which is lower than the minimum freckling area (see section 4.3). 62 INTERDENDRITIC S E G R E G A T I O N A N D LIQUID DENSITY The segregation of the significant alloying elements along the mushy zone is presented in Figures 31, 32, 33, 34 and 35 for IN718, MAR-M002, MAR-M247, C-276 and T l respectively. The main element Ni (or Fe in the case of Tl) , constituting the balance of the normalized compositions was not plotted in these graphs. In the case of IN718, MAR-M002 and MAR-M247, the fraction l iquid^ is also presented on the same graphs. For all 5 alloys, the densities corresponding to the compositions and temperatures at various depths in the mushy zone were computed by "METALS". The resulting density profiles are shown in picture 36. The solid curves in Figures 31 to 36 are third degree polynomial regressions relative to the plotted data. All the numerical values corresponding to these graphs, as well as other E D X analysis results (such as eutectic precipitate and surrounding matrix, carbides, dendrite centers) for the various alloys studied, are gathered in Tables CI to C5 in appendix C. Al l these values are averages over 4 or more measurements. It is to be noted that the standard deviations associated with the measurements were always of the order of ±5% relative (or less) for the major elements and up to ±15% relative for the minor elements, which is within the expected precision capabilities of E D X microprobe. For each alloy in Figure 36, the effect of the E D X imprecision on the density calculations was estimated in the following manner : - Lowest possible density : (a) All light alloying elements (namely A l , Ti , Cr, Fe) were increased by a full E D X imprecision (+5% relative for major elements (>5wt%) and +15% relative for minor elements (<5wt%)). (b) Al l heavy alloying elements (namely Hf, Ta, W, Mo) were decreased by a full E D X imprecision, (c) Nickel (or iron for Tl ) adjusted to the reminder to 100wt%. 63 - Highest possible density : same procedure with an increase of the heavy elements and a decrease of the light elements. The resulting sensitivity analysis is represented with error bars in Figure 36. It is to be noted that in reality, the actual standard deviation associated with the density profiles is much less than that depicted by the error bars since it is very unlikely that every heavy elements and every light elements be shifted of the full imprecision amount in opposite directions, all on the same average composition (average over at least 4 E D X measurements). Note : In Figure 32, in the case of MAR-M002, 3 sets of values are reported (at 1343°C, 1337°C and 1322°C) which are not included in the composition curves. It is believed that these measurements are erroneous and do not represent the actual interdendritic segregation pattern, as is explained below. From the top of the mushy zone, the interdendritic liquid exhibited a relatively large amount of carbides. From this observation, interdendritic liquid was (wrongly) associated with carbides. These 3 sets of measures thus focused on areas exhibiting carbides. However, as shown by the increasing W content and decreasing Hf content, it became obvious that these carbides were actually embedded in the outer layers of dendrites. The next measurement areas were then chosen accordingly, focusing on the presence of eutectic precipitates (andpossibly carbides) rather than just carbides. The same mistake was not repeated with MAR-M247. In any case, this leads to an interesting observation : carbides seem to start precipitating around or above 1350°C in MAR-M002 (an accurate temperature value would require a DTA), and it would appear that they are then grown over rather than pushed away by the dendritic array. However, confirmation of this observation is not the topic of this thesis and would require further work. 64 I n t e r d e n d r i t i c l i q u i d s e g r e g a t i o n a n d f r a c t i o n l i q u i d p r o f i l e s a l o n g t h e m u s h y z o n e i n D S Q IN718 18 ., _ 100 17 16 15 4-' 14 13 4^  12 11 10 9 8 7 4-6 5 l 4 3 4-2 1 Tsoi ~ 1260°C T"|jq = 1336°C 1260 1280 1300 1320 Temperature in the mushy zone (in °C) 80 4eo g, '5 • c .2 40 1 I 20 0 1340 o Al A Ti O Cr • Fe e Co Cu Nb Mo fL figure 31: Interdendritic liquid segregation and fraction liquid profiles along the mushy zone DSQIN718 65 I n t e r d e n d r i t i c l i q u i d s e g r e g a t i o n a n d f r a c t i o n l i q u i d p r o f i l e s a l o n g t h e m u s h y z o n e i n D S Q M A R - M 0 0 2 13 . 100 Tsoi ~ 1249°C T|_iq = 1365°C 80 60 12 11 10 9 8 7 6 5 4 3 2 4-1 0 -I 1 1 1 1 1 1 1_ 0 1240 1260 1280 1300 1320 1340 1360 1380 Temperature in the mushy zone (in °C) 40 20 3 o Al o Ti • Cr • Co A Hf Ta E W m b - fL Figure 32 ; Interdendritic liquid segregation and fraction liquid profiles along the mushy zone DSQ MAR-M002 66 I n t e r d e n d r i t i c l i q u i d s e g r e g a t i o n a n d f r a c t i o n l i q u i d p r o f i l e s a l o n g t h e m u s h y z o n e i n D S Q M A R - M 2 4 7 12 100 1260 1280 1300 1320 1340 Temperature in the mushy zone (in °C) 1360 o Al A Ti O Cr • Co e Mo & Hf 0 Ta m W - - - fL Figure 33 : Interdendritic liquid segregation and fraction liquid profiles along the mushy zone DSQ MAR-M247 67 I n t e r d e n d r i t i c l i q u i d s e g r e g a t i o n p r o f i l e s a l o n g t h e m u s h y z o n e i n D S Q C-276 27 26 25 24 23 22 21 20 19 18 -17 -16 -15 14 13 12 11 4-10 9 8 7 6 -5 -4 3 2 1 0 Tsoi ~ 1325X T~Liq = 1370°C 1320 1330 1340 1350 1360 Temperature in the mushy zone (in °C) 1370 o Cr Fe o Mo • W Figure 34 : Interdendritic liquid segregation profiles along the mushy zone in DSQ C-276 68 I n t e r d e n d r i t i c l i q u i d s e g r e g a t i o n p r o f i l e s a l o n g t h e m u s h y z o n e i n D S Q T1 24 4-23 22 21 -20 -19 18 17 4-16 15 14 -13 -12 11 10 9 8 7 6 5 4 3 2 4-1 0 a-"Tsoi ~ 1320°C 1440°C + 1300 1350 1400 Temperature in the mushy zone (in °C) 1450 o V A Cr O Mn • W Figure 35 ; Interdendritic liquid segregation profiles along the mushy zone in DSQ T l 69 I n t e r d e n d r i t i c l i q u i d d e n s i t y v a r i a t i o n s a l o n g t h e m u s h y z o n e o f v a r i o u s a l l o y s Figure 36 : Interdendritic liquid density profiles computed by "METALS" for 5 DSQ alloys. 70 43i_ F R E C K L E D S A M P L E S Various industrial castings with freckles were also studied for this work. Pictures of these freckled samples are shown in Figures 37, 38, 39, 40; Freckles in IN718 could be easily distinguished on a polished sample by their relatively high concentration of big niobium carbides (see Figure 41). Only in MAR-M002 did the freckles show significant porosity (Figure 42). It was actually the only way to locate them since other features (carbides, grains) appeared no different from the surrounding matrix. Freckles in T l exhibited bigger more interconnected tungsten carbides than the surrounding matrix (Figures 43 and 44). Freckles in C-276 did not exhibit any significant features and could only be located by strong etching. The average compositions of freckles and surrounding matrix were measured by E D X microprobe and are given in Table 5 for IN718, MAR-M002, C-276 and T l . IN718 M a t r i x F r e c k l e Al 0.33 0.14 Ti 0.83 1.20 Cr 18.48 16.93 Fe 19.37 16.8C Co 0.00 0.00 Ni 51.68 50.16 Cu 0.35 0.50 Nb 5.55 10.11 Mo 3.41 4.15 Total 100.0C 100.0C M M 0 0 2 Mat r i x F reck le Al 6.82 7.54 Ti 1.26 1.82 Cr 9.18 7.98 Co 10.18 8.93 Ni 58.07 60.73 Hf 1.83 4.55 Ta 2.55 2.65 W 10.12 5.81 Total 100.0C 100.0C C-276 M a t r i x F r e c k l e Cr 16.05 1666 Fe 6.56 6.49 Ni 55.60 53.33 Mo 17.28 19.81 W 4.51 3.71 Total 100.0C 100.0C II M a t r i x F r e c k l e Iv 1.24 1.61 Cr 4.58 5.65 Fe 73.86 68.71 w 20.33 24.04 Total 100.0CS 100.0C Table 5 : Freckles and surrounding matrix compositions (in wt%). 71 Figure 37 : Cross-section of a IN718 ingot exhibiting freckles (Mag. X0.87). Figure 38 : Cross-section of a forged T l billet exhibiting a ring of freckles at mid-radius (Mag. X0.6). 72 Figure 39 : DS MAR-M002 test slab (dark freckle in lower right corner) (Mag. X0.9). Figure 40 : Cross-section of a C-276 slab. Freckles are light etching. They appear flattened after the slab was rolled down. (Mag. X0.55). Figure 41 : SEM picture of a freckle in IN718 exhibiting large Niobium carbides.(Mag. X60). Figure 42 : Porosity associated with a freckle in MAR-M002 (polished sample) (SEM picture) (Mag. X300). 031208 20KV X300 100u 74 Figure 43 : SEM close-up of the carbides in a freckle in T l . (Mag. X 1 2 0 0 ) . 031339 £0KV X1£0 250umC Figure 44 : SEM close-up of the carbides in the bulk matrix in T l . (Mag. X I 2 0 0 ) . 75 The average minimum area per freckle "cell", A m i n , was also evaluated on two samples : 2 2 (a) in IN718, a count of 64 freckles was found on a 1600mm area: A m i n « 2 5 m m . • 2 2 (b) in T l , a count of 16 freckles was found on a 504mm area : Aminw31.5mm . These values, of the order of the DSQ samples cross-section area, confirm that freckles were not expected to develop in the experimental samples. These values are slightly lower than Amin=25-100mm , yielded by freckle spacing of the order of 5-10mm, reported for various analog systems [9]. However, these values are well below a reported minimum freckle area of about 300mm for Udimet 700 [6]. This suggests that minimum freckling area may depend on alloy chemistry. 4.4. F R A C T I O N LIQUID A N D PRIMARY DENDRITE A R M SPACING Fraction liquid was measured by image analysis of pictures of cross-sections of the DSQ samples for three alloys : IN718, MAR-M002 and MAR-M247. Results were presented with the interdendritic liquid compositions in Figures 31, 32 and 33 respectively. In the case of C-276 and T l , although the dendrite centers could be easily distinguished, there was too much uncertainty concerning the actual outline of the dendrites and measurements of the fraction liquid were not carried out. The primary dendrite arm spacing for 4 DSQ alloys was also evaluated from various cross-sections pictures. Average values are presented in Table 6. Alloy IN718 MAR-M002 MAR-M247 T l A-! (in u.m) 380 300 290 450 Table 6 : Primarv dend rite arm spacing in DSQ samples. Note : Fraction liquid and primary dendrite spacing are important parameters in the determination of a freckling criterion (such as the Rayleigh number criterion). They enable the calculation of the permeability of the mushy zone (see appendix A). Permeability is a key factor, since it controls fluid flow, and thus freckle formation. 76 5_ D I S C U S S I O N 5.1. V A L I D A T I O N O F T H E M E A S U R E M E N T S 5.1.1. Primary dendrite arm spacing In the case of superalloys, primary dendrite arm spacing has previously been linked to gradient G and solidification rate R at the growth front [54]. (Figure 45) With a measured X,!«250-400u,m for the D S Q samples, it can be estimated from this chart that G.ite0.2 °C/s. The recorded lowering speeds for IN718, MAR-M002 and MAR-M247 in the D S Q furnace were 2.1xl0"5 m/s, 1.8x10'5 m/s, 1.7xl0"5 m/s respectively. Combined with a cooling rate of 0.2 °C/s, these lowering speed values yield an average temperature gradient of about 10°C/mm, confirming the validity of the actual temperature measurements reported in Figure 24. 1000 100 DENDRITE SPACING 10^ 0.01 10 I I I | 100 I I I 1000 I ' M 10000 I s U PRIMARY ARMS J 0 IN-100 O IN-713C & IN-718 V IN-738LC O IN-792 O MAR-M-246 Cs NI-9AL-4CR 0 NASAIR-100 J I I I I j i i L I I I I L J l_L o.i 1 10 COOUNG RATE (GxR),°C/SEC 100 Figure 45 : Primary dendrite arm spacing versus cooling rate [54]. 77 5.1.2. Freckle composition measurements Relatively few freckle and surrounding matrix compositions have been published in the literature. Comparison between literature values and this thesis' measurements was only possible in the case of IN718. Tables 1 and 5 have already presented both sets of data. They are in very good agreement, confirming the validity of the experimental procedure used in this thesis to measure freckle and surrounding matrix compositions. 5.1.3. Segregation measurements The aim of this section is to validate the measured interdendritic compositions. For each element /' of each alloy j, the initial partition coefficient ka was calculated : Concentrations in the dendrites centers and in the top of liquid are reported for each alloy in appendix C. It has been assumed, as a first approximation, that this initial partition coefficient is valid throughout the mushy zone (although it is known that partition coefficients usually vary with the fraction liquid [53]). Average and/or initial partition coefficients of various elements in various alloys found in the literature as well as this work are reported in Table 7. The Scheil and Lever equations were then applied for each element to calculate a theoretical eutectic concentration. (Superalloys usually exhibit a relatively small fraction of eutectic (fE<\0%) [57]. For calculation purposes, the fraction solid at the eutectic was approximated to/^0.95 for the Scheil equation, and7^ =1 for the Lever rule.) (For further details on the Scheil and Lever rules, see appendix B or refer directly to [55].) These values given by the Scheil and Lever rules should provide a good estimation of the limits for the real eutectic composition. These limits were then compared to the actual measured DendriteCenter [Eq.21] TopLiquid 78 eutectic compositions in the DSQ samples. It is to be noted that, given the experimental conditions, the DSQ samples segregation profiles would be expected to follow more closely the Scheil equation than the Lever rule (see appendix B). Results are presented in Table 8. Alloy Reference Al C Co Cr Fe Hi Mn Mo Nb Ni Re Si Ta Ti V W Binary Ni-X Av. k (a), (b)* 0.80 0.28* 1.00 0.93 0.10 0.67 1.00 0.82 0.70 0.85 1.40 MAR-M200 Av. k (c) 0.90 0.92 0.60 1.20 MAR-M200+Hf Av. k (c) 1.07 0.23 1.00 SX1 Av. k (d) 0.78 1.09 1.11 0.92 2.40 0.66 1.75 IN100 Av. k (a) 0.99 1.07 0.99 1.00 0.63 1.00 MAR-M247 Av. k (a) 0.97 1.07 1.00 0.13 1.00 0.79 0.76 1,18 MAR-M247 Av. k (e) 0.94 1.07 1.00 0.12 1.00 0.77 0.73 1.23 MAR-M247 I nit. k0 (e) 0.88 1.11 0.10 0.65 0.60 1.31 MAR-M002 I nit. k0 CO 0.91 1.10 0.88 0.06 0.29 0.53 1.29 IN718 I nit. k0 (g) 1.02 1.06 0.87 0.48 1.05 0.48 0.63 A^verage Partition Coefficients 0.92 1.07 0.98 1.06 0.12 0.91 0.48 0.99 2.40 0.48 0.66 0.69 0.93 1.34 IN718 Init. k0 Present work 1.08 1.05 1.06 1.10 0.99 0.44 1.02 0.59 MAR-M002 I nit. k0 Present work 0.91 1.09 0.96 0.48 1.00 0.77 0.56 1.25 MAR-M247 Init. k0 Present work 0.80 1.10 0.91 0.52 0.89 0.99 0.88 0.54 1.36 C-276 Init. k0 Present work 0.99 1.05 0.86 1.04 0.99 |TI Init. k0 Present work 0.87 1.06 1.38 0.61 0.78 Table 7 : Partition coefficients of various elements in industrial alloys. (References : (a)=[56], (b)=[57], (c)=[53], (d)=[3], (e)=[58], (r>[59], (g)=[61]) IN718 Scheil (fs=0.95) Lever Measured M A R - M 2 4 7 Scheil (fs=0.95) Lever Measured (in wt%) (in wt%) Al 0.48 0.57 0.18 Al 11.57 7.96 6.28 Ti 3.39 1.68 1.74 Ti 4.28 1.99 1.50 Cr 12.42 14.09 11.92 Cr 11.12 9.34 9.18 Fe 12.99 16.07 13.36 Co 7.34 9.09 8.59 Co 0.40 0.44 0.42 Ni 59.88 58.95 53.45 Ni 51.58 53.79 51.90 Mo 0.94 0.76 1.41 Cu 0.49 0.47 0.51 Hf 8.90 4.04 11.47 Nb 32.45 13.82 15.62 Ta 4.69 3.71 3.11 Mo 3.82 3.74 4.35 W 3.22 6.97 5.00| C-276 Scheil Lever Measured (in wt%) (/s=0.95) (fs=1) Cr 15.09 14,69 14.56 Fe 5.72 6.28 5.58 Ni 49.28 53.90 49.33 Mo 27.83 21.29 26.52 I W 4.48 4.38 4.01 1 1 Scheil Lever Measured (in wt%) (fs=0.95) (f s=D V 3.43 1.76 1.84 Cr 7.61 5.88 5.50 Mn 0.15 0.34 0.43 Fe 62.93 71.05 66.47 1 W 35.76 23.74 25.76 M A R - M 0 0 2 Scheil Lever Measured I (in wt%) f/s=0.95) (fs=1) Al 8.15 6.90 4.98 Ti 5.51 2.64 2.03 Cr 9.82 9.14 8.91 Co 7.79 9.30 8.76 Ni 57.51 57.62 56.69 Hf 11.42 5.00 12.11 Ta 6.30 4.09 1.73 W 4.74 8.07 4.79 Table 8 : Measured and calculated (Scheil and Lever rules) eutectic compositions in DSQ alloys. 79 It can be seen in Table 8 that, in the case of IN718, C-276 and T l , final eutectic compositions are in very good agreement with the calculated Scheil-Lever range. In the case of IN718, only A l and Mo show a slight deviation : the A l content, being lower than lwt% is close to the detection limits of E D X analysis. Thus, values show poor precision in any case. Molybdenum's initial partition coefficient in IN718, measured at 0.99, may actually be closer to 0.90, and then yield a eutectic concentration of 4.35wt%. Apart from A l and Mo, measured initial partition coefficients in IN718 compare very well with the literature (see Table 7). In the case of MAR-M002 and MAR-M247, only Cr, Co and Ni are in good agreement with the Scheil-Lever range. The A l eutectic concentration is lower than expected from the calculation. This is probably due to the fact that the partition coefficient of A l usually increases significantly with the fraction solid, from 0.8 up to 0.95 or even 1.2 [53]. Although the measured initial coefficient for Ti , Ta and W is in very good agreement with the published data (Table 7), the actual eutectic concentrations of Ti and Ta (and W to a lesser extent) are much lower than expected from the calculation. This is because much of Ta, Ti, Hf and W is removed from the interdendritic liquid by the large amount of carbide precipitation taking place in MAR-M002 and MAR-M247 (both alloys have a relatively high average carbon content : 0.15wt%). The metal content in these carbides has been measured to be about 54Ta + 15W + 15Hf + 12Ti + 4(Ni+Mo+Al+Co).(See appendix C). In the case of Hf in MAR-M002 and MAR-M247, literature values reported in Table 7 indicate a average partition coefficient of about 0.13. This value, combined with an average concentration of about 2.2wt% in the top liquid, yields by Lever-Scheil calculation eutectic compositions in the range 17-30wt%. In this case, the actual measured concentration of Hf (about 12wt%) is confirmed (12wt% is still lower than the calculated range, but this was expected, like 80 for Ta, Ti and W, because of carbide precipitation). The unrealistic value of 0.5 measured for the initial partition coefficient of Hf probably comes from the following. Center dendrite compositions were measured on cross-sections toward the bottom of the mushy zone and not at the dendrite tips. Since solid-state diffusion is considered negligible, this should not be a problem. However, in this case, for strongly segregating elements such as Hf, too big and/or slightly off-center a measurement window will yield too high a dendrite center concentration. Nevertheless, the effect on the measure of the partition coefficient of other less strongly segregating elements should be minimal, as is confirmed by values in good agreement with the literature. Finally, it can be noticed in Table 8 that, contrarily to the predictions in appendix B, some elements tend to be closer to the Lever eutectic value than the Scheil eutectic value. This might be linked to the actual deviation of the partition coefficient from its initial value k0. 5.1.4. Fraction liquid Knowing the concentration profiles of the alloying elements and their respective partition coefficients, it was possible to back calculate the fraction liquid along the mushy zone for each alloy. The results are presented in Table 9. IN718 Temperature (in °C) 1336.0 1328.5 1321.0 Scheil fL 1.00 0.71 0.44 Measured fL 0.99 0.74 0.48J MAR-M002 Temperature (in °C) 1365.0 1359.4 1354.7 Scheil f L 1.00 0.80 0.63 Measured fL 0.93 0.81 0.71 MAR-M247 Temperature (in °C) 1360.0 1334.6 1321.0 Scheil f L 1.00 0.53 0.25 Measured f L 0.94 0.50 0.18 Table 9 : Measured and back-calculated fraction liquid at various temperatures in the mushy zone for 3 DSQ alloys. 81 Only the elements which behaved according to the Scheil and Lever equations, as seen in the previous section, were considered in Table 9, namely Nb in IN718 (ko=0A4), W in MAR-M002 (k=l.25) and W in MAR-M247 (&0=1.36). It can be seen that the fraction liquid measured by image analysis show relatively good agreement with the back calculated values. 52^ E X T E N T O F T H E DENSITY INVERSIONS 5,2,1, Theoretical provisions The following calculations are based on the Rayleigh number, which seems to be the freckling criterion the most closely linked to the density inversion theory. The thermosolutal Rayleigh number RaT/s has been written [14]: According to [14], freckling occurs when Ram » 1. The density inversion order of magnitude 8 would then be : 5 = ^ = ^ Pq.23] dz gk\ (Note : In the following expressions, numerical values for x\, DT, g and Xx are in SI units.) In the case of Pb-10wt%Sn [14]: 6 * 2 5 x 1 0 ' 11*10 = Q 3 5 ( g ^ y ^ ^ • ^ x l O - 4 ) In the case of Pb-2wt%Sb [14]: 6 * 3 x 1 0 1 0 . = 0.38 (g/cm3)/mm 9.8-(3x10^) In the case of superalloys : 8 « ^ X ^ ^ . = 0.3 (g/cm3)/mm (for superalloys, numerical data for pure liquid Nickel at 1500°C, after "METALS") 82 For superalloys, with a typical primary dendrite arm spacing of 350u.m, freckling would therefore be expected to occur following density inversions of 0.3(g/cm3)/mm, or 0 .03(g /cm 3 ) / °C. , in a thermal gradient of 10°C/mm. Similarly, it is possible to include the permeability K of the dendritic array in these calculations (see equation [Eq.6]). Among all the permeability expressions reported in appendix A, Poirier's [31] seems to be the best adapted to the present case. Poirier reported the following permeability equation for flow parallel to the dendrites in analog systems : K = 3.75x10" 4./^, 2 (with 0.2 <fL< 0.6) [Eq.24] Further calculation of the density inversion order of magnitude is however not possible because no critical value of the permeability dependent Ra m has been reported yet. If freckles are expected to develop in the upper part of the mushy zone (fL « 0.5) (see section 5.2.3), then the critical value of Ra T/s can be estimated to be : Ra m » Ra m x K/Xx « 1.10 .In any case, more work involving actual castings is required to determine the exact Ra m . 5-2.2. Freckles density The densities of the freckles and the surrounding matrix at various temperatures have been calculated with "METALS" from the measured compositions (Table 5). The results are presented in Table 10. Moreover, freckle/matrix compositions have been published for various alloys. These compositions, as well as the corresponding densities calculated by "METALS" at various temperatures are presented in Table 11. It can be seen that, although all the freckle compositions (from this thesis (Table 10) and from the literature (Table 11)) are indeed shifted toward the eutectic, as expected, freckles do not seem to necessarily have a lower density than 83 the surrounding matrix. In IN718 and T l , the calculated densities show that freckles seem actually heavier than the surrounding matrix. In C-276, there is no significant density difference. In MAR-M002, MAR-M200 and Udimet 700, freckles are between 0.5% and 3% lighter. Only in SX1 and lead-based alloys, specifically designed to exhibit freckles, does the density difference reach 6 or 7%. Thus, it would seem that freckles result from very small density differences, of the order of 0. lg/cm . Note : The fact that the freckles in some systems appear heavier than the surrounding matrix does not invalidate the density inversion theory. These results might arise from the limits on EDX measurements and density calculations. Table 10 : Freckle and Al loy Area of interest L iquid Density (in g/cm 3 ) IN718 Matrix (TU (f1336 0C) Freckle (7U qFl336°C) Freckle (7"Sor1260°C) 7.49 7.57 7.64 surrounding matrix calculated densities (after "METALS"). MAR-M002 Matrix (77,^=1365°C) Freckle (7"a,=1365°C) Freckle (7" S or1249°C) 7.04 6.81 6.92 C-276 Matrix ( 7 ^ 1370°C) Freckle (7U (j=1370°C) Freckle (TS or1325°C) 8.12 8.12 8.16 T1 Matrix (T L i q=1440°C) j 8.09 Freckle (T L i q=1440°C) I 8.29 Freckle (TSor1320°C)j 8.41 84 I S X i (a) Al Co Cr Hf Ni Re Ta W Total Matrix(wt%) 6.00 12.50 4.50 0.16 57.74 6.30 7.00 5.80 100.0 |Freckle(wt%) 8.20 11.00 3.60 0.18 61.82 2.30 10.00 2.90 100.0 Liquid Densities P(1300°C)= P(1300°C)= M a r - M 2 0 0 (b) Al Co Cr Nb Ni Ti W Total P(1300°C)= P(1300°C)= Matrix(wt%) 4.90 10.90 9.00 1.00 58.70 2.30 13.20 100.0 Freckle(wt%) 5.19 10.57 9.18 1.20 59.20 2.60 12.01 100.0 7.56 g/cm3 7.02 g/cm3 7.37 g/cm3 7.25 g/cm3 I N - 7 1 8 ( c ) Al Cr Fe Mo Nb Ni Ti Total Matrix (wt%) 0.67 18.58 17.62 3.38 5.46 53.19 0.97 99.9 Freckle(wt%) 0.45 17.36 15.23 3.51 9.43 52.55 1.33 99.8 P(1336°C)= P(1260°C)= 7.43 g/cm3 7.57 g/cm3 U d i m e t 700 (b) Al Co Cr Mo Ni Ti Total Matrix(wt%) 4.19 14.75 14.00 3.80 59.66 3.60 100.0 Freckle(wt%) 4.27 14.16 14.00 3.90 59.90 3.82 100.0 P(1300°C)= P(1300°C)= 6.93 g/cm3 6.90 g/cm3 P b - S n (d) Pb Sn Total Matrix(wt%) 83.00 17.00 100.0 Freckle(wt%) 71.00 29.00 100.0 P(290°C)= P(270°C)= 9.82 g/cm3 9.29 g/cm3 P b - S b (d) Pb Sb Total Matrix(wt%) 98.00 2.00 100.0 P(315°C)= Freckle(wt%) 96.00 4.00 100.0 P(300°C)= 10.57 g/cm3 10.47 g/cm3 Table 11 : Published freckle/matrix compositions for various alloys and corresponding calculated liquid densities. (References : (a)=[3], (b)=[6], (c)=[2], (d)=[ll]) 85 5,2,3, Segregation, related density, and freckle formation Density profiles for the five DSQ alloys have been presented in Figure 36. It can be seen that in IN718, C-276 and T l , calculated densities of the interdendritic liquid increases with depth in the mushy zone. Although this does not correspond to the density inversion theory, it could have been expected since the calculated freckle densities for these alloys were also greater than that of the surrounding matrix. However, in the case of MAR-M002 and MAR-M247, there is a drop in the interdendritic liquid density slightly below the top of the mushy zone. This density inversion, between the tip of the dendrites and the lowest density point is of the order of 0.03(g/cm ) / ° C for MAR-M002 and 0.005(g/cm ) / ° C for MAR-M247 ("slopes" in Figure 36). According to previous calculations with the Rayleigh number (section 5.2.1), 0.03(g/cm3)/°C could be enough to produce freckling, whereas 0.005(g/cm ) / ° C should be too low. This is consistent with the fact that MAR-M002 exhibits freckles and MAR-M247 does not. It is also interesting to notice that in MAR-M002, the composition of the interdendritic liquid associated with the point of lowest density (at 7 « 1 3 4 0 O C ) corresponds exactly to the measured freckle composition (comparison between Table 5 and Figure 32). Similar comparison between freckle composition and segregation profiles for IN718, C-276 and T l , would locate freckle initiation at 7 « 1 3 2 0 O C , JH«1363°C, and 7 « 1 3 5 0 O C respectively. A graphic illustration of the comparison between Table 5 and Figures 31 to 35 is proposed in appendix D. Thus, it can be seen that in the case of IN718, MAR-M002 and C-276, freckles would seem to initiate relatively close (about 20°C below TLiq) to the top of the mushy zone where ^ « 4 0 - 6 0 % . This is coherent with the fact that freckles will develop more easily in regions of high permeability. The case of T l will be discussed in the next section. 86 Moreover, average and eutectic compositions published in the literature are presented in Table 12. The corresponding densities calculated by "METALS", as well as the estimated total density inversion 5 over the whole mushy zone are also presented in Table 12. Liquid Densities and Mar-M247 (a) Al Co Cr Hf Ni Ta Ti W Total Density Inversion 5 Initial Composition 5.50 10.00 8.25 1.50 58.75 5.00 1.00 10.00 100.0 P(1360°C)= 7.36 g/cm3 P(1280°C)= 7.67 g/cm3 Eutectic Composition 6.50 8.50 8.25 20.50 38.25 10.00 2.00 6.00 100.0 Mar-M200+Hf (b) Al Co Cr Hf Nb Ni Ti W Total Initial Composition 5.40 10.00 8.60 1.90 0.00 59.70 2.10 12.30 100.0 Eutectic Composition 4.80 4.40 7.60 9.50 1.50 66.30 2.90 3.00 100.0 Mar-M200 (b) Al Co Cr Nb Ni Ti W Total Initial Composition 4.50 10.50 9.40 0.00 61.70 2.10 11.80 100.0 Eutectic Composition 6.00 8.70 11.60 1.60 61.80 4.00 6.30 100.0 M2 (c) Initial Composition Eutectic Composition Cr 4.00 6.50 Fe 82.70 74.20 Mo 5.50 9.80 1.80 2.00 W 6.00 7.50 Total 100.0 100.0 IPb-Sn (d) {Initial Composition jEutectic Composition Pb 90.00 38.10 Sn 10.00 61.90 Total 100.0 100.0 Pb-Sb (d) Initial Composition Eutectic Composition Pb 98.00 88.80 Sb 2.00 11.20 Total 100.0 100.0 8= -0.004(g/cm3)/oC P(1370°C)= P(1315°C)= P(1370°C)= P(1315°C)= 8= P(1420°C)= P(1231°C)= P(320°C)= P(183°C)= P(315°C)= P(251°C)= 8= 7.23 g/cm3 7.16 g/cm3 0.001(g/cm3)/°C 7.32 g/cm3 6.80 g/cm3 0.009[g/cm3)/oC 7.46 g/cm3 7.79 g/cm3 0.002(g/cm3)/oC 10.14 g/cm: 8.12 g/cm: 8= 0.015(g/cm3)/oC 10.57 g/cm3 10.12 g/cm3 0.007(g/cm3)/oC Table 12 : Published average and eutectic compositions for various alloys and corresponding calculated liquid densities (after "METALS"). Where b=(9Liq-pE)l{TLiq-TE) (References : (a)=[56], (b)=[53], (c)=[65], (d)=[l 1]) 87 In all cases, it can be seen that the density inversion 8 over the whole mushy zone is lower than the necessary 0.02-0.03(g/cm )/°C required for freckling. This result, coupled with a very low permeability close to the eutectic, confirms that freckles do not initiate at the bottom of the mushy zone, but rather close to the top. This is further confirmed in binary systems [11] : In Pb-2wt%Sb ( 7 / ^ 3 1 5 ° C ) , freckle composition averaged 4wt%Sb (at 7/«300°C=7^ < r 15°C according to the phase diagram in Figure 10), and was indeed only slightly shifted toward the eutectic composition (0=11.2wt%Sb). In Pb-17wt%Sn (TLiq*290°C), freckle composition averaged 29wt%Sn (at 7>270oC=7,£,1?-20oC according to the phase diagram in Figure 10), and was also only partially shifted toward the eutectic composition (C£=61.9wt%Sn). 5.2.4. Possible influence of non-measured elements on freckle formation The density of the interdendritic liquid in IN718, C-276 and T l was observed to increase with depth in the mushy zone (Figure 36). It may thus be possible that the observed macrosegregates in these alloys are actually due to center-segregation rather than real freckling (see section 2.1.6). However, it will be assumed in this section 5.2.4 that the segregation mechanism is indeed freckling. The following calculations by "METALS" may provide a better insight on the role on freckle formation of some elements that were not included in the E D X composition measurements, such as carbon or silicon. 5.2.4.I. Allov IN718 In IN718, the carbon is expected to segregate toward the liquid from 0.03wt% (average composition) up to about 0.25wt%. Carbon is also known to precipitate into niobium carbides. 88 From a reported Nb:C weight ratio of 93:12 [60], and a metal composition (in wt%) in the carbides of (1.5Fe + 4.7Ni + 2.5Cr + 80.4Nb + 2.5Mo + 8.4Ti)[61], the calculated composition (in wt%) of these carbides is (9.4C + 2.27Cr + 1.36Fe + 2.27Mo + 72.85Nb + 4.26Ni + 7.61Ti). Carbides precipitate over a wide range of temperatures, the start temperature having been reported to be around 1250°C-1280°C [62,63]. However, for the sake of the argument, it will be assumed that the start temperature corresponds to the apparent freckle initiation location : r=1321°C (15°C below the liquidus) (see section 5.2.3). The density of the interdendritic liquid at 1321°C was then computed, assuming that 0, 20, 40, 60, 80 or 100% of the carbides precipitated at this temperature. (Precipitated carbides are considered not to participate to the overall density of the liquid.) The results are presented in Table 13. Temperature (°C) Tuq 1321 1321 1321 1321 1321 1321 1 Carbide precipitation 0% 0% 20% 40% 60% 80% 100% J Liquid density (g/cm3) 7.48 7.51 7.51 7.52 7.53 7.54 7.55 Table 13 ; Calculated effect of carbon segregation and carbide precipitation on the density of the interdendritic liquid in IN718. It can be seen that 0.25wt%C at 1321°C would not be enough to create density inversion. Moreover, carbide precipitation tends to increase the density of the interdendritic liquid, and is therefore not favoring freckle formation. Segregation up to lwt%C, and no carbide precipitation would be required in order to achieve the minimum necessary density inversion of 0.01 (g/cm )/°C. Such conditions are rather unlikely and may not even be sufficient to produce freckles. 89 However, it is also possible to consider the potential effect of silicon on the liquid density. Results are reported below : Liquidus composition with 0.03wt%C and 0.3wt%Si : p^( l336°C)=7 .44g /cm 3 1321°C composition with 0.25wt%C and 0.3wt%Si : p^(1321°C)=7.46g/cm 3 1321°C composition with 0.25wt%C and 0.6wt%Si : p^(1321°C)=7.42g/cm 3 1321°C composition with 0.25wt%C and 1 .Owt%Si : pil(?(13210C)=7.36g/cm3 It can be seen that Si would have to segregate up to 0.6-1.0wt% to produce a density inversion susceptible to create freckles. These values are realistic according to the Scheil equation (given a partition coefficient of 0.4, an initial concentration of 0.3wt%Si and a fraction liquid of 40% at 1321°C). However, further work would be required to investigate the influence of Si on freckling in IN718. It is interesting to note that IN718 ingots were initially categorized as freckle-prone in the 1970's when they contained about 0.3wt%Si. Since then, in an effort to reduce the amount of Laves phase (stabilized by Si), the silicon content in IN718 has been reduced to about 0.01wt%. Much fewer cases of freckles have been reported in IN718 since then. It is also possible that these more recent cases were actually center segregation (resulting from a heavier interdendritic liquid) and were mistaken for freckles due to previous history. This tends to confirm the important role of the silicon content on freckling in IN718. In any case, the observation of numerous large NbC carbides in freckles in IN718 (see Figure 41) is a further confirmation of the expected high carbon content in the freckles. Zirconium, segregating heavily toward the liquid, may also play a role in enhancing density inversion. Further work involving C, Si, Zr content measurements to study their segregation pattern and influence on density is required. 90 5.2.4.2.Tool steels T l and M2 In the case of T l , like for IN718, freckle density was found to be higher than that of the surrounding matrix (see Table 10). However, the calculation did not take into account the carbon content. Respective carbon content in the freckle and the matrix was estimated by image analysis of pictures 43 and 44. Inside a freckle (picture 43), carbides represented about 40% of the area, against only 15% in the matrix (Figure 44). Considering an average carbon content of 0.8wt% in the matrix, the carbon content in the freckle was therefore expected to be 40/15=2.7 times higher, i.e. about 2wt%C. Liquid density calculations then yielded the following results : Matrix with 0.8wt%C at 1440°C : pi/<7=7.94 g/cm3 Freckle with 2.0wt%C at 1350°C : p L i q =7.98 g/cm3 Freckle with 2.5wt%C at 1350°C : p L i q =7.89 g/cm3 Freckle with 3.0wt%C at 1350°C : p L i q =7.80 g/cm3 (1350°C is the estimated freckle initiation position.) It can be seen that carbon segregation yields some degree of density inversion and may possibly be responsible for freckle formation. Another case of freckle forming tool steel is M2 [64]. Interestingly, the interdendritic liquid segregation profile down the mushy zone of a directionally solidified sample has been measured in the past [65] and is presented in Figure 46. 91 Figure 46 : Segregation profiles in tool steel M2 [65,66]. SUrf '•act el rctic ion End of PVfitOCt'C ' •action 16,0 I L 16,5 !%i distance in mm 30,4 1309 1243 1240 temporaluw in ° C 31.S 1231 The corresponding density profiles have been calculated with "METALS" in 3 cases, assuming no segregation of carbon (wt%C = constant = 0.9 = average composition), segregation from 0.9 up to 2wt% toward the liquid (most likely case) and segregation from 0.9 up to 3wt% respectively. These profiles are presented in Figure 47. It can be seen that density always increases with depth in the mushy zone until the carbide precipitation temperature (around 1240°C). Then, close to the bottom of the mushy zone, there is a sharp density inversion because the liquid is depleted in carbide forming heavy elements (Mo and W). Thus, it would seem that for tool steels, the trigger for freckle initiation is the precipitation of heavy elements into carbides. But in order for the freckle to develop through the mushy zone, a high carbon content of the order of 2 to 3wt% (due to segregation) is probably necessary. 92 However, it is important to notice that carbide precipitation, in tool steel or any other carbide forming alloy, actually plays a double role : on one hand, it favors freckle formation by removing heavy elements from the interdendritic liquid ; on the other hand, carbide precipitates between dendrite arms tend to lower the permeability of the mushy zone and impede fluid flow and freckle formation [67]. In any case, further work (such as the measurement of the actual carbon segregation profile and carbon content in freckles) is necessary. Other elements, such as Mn, P, Si may also play a role on freckle formation. 5.2.4.3.Other alloys Carbon segregation and carbide precipitation in MAR-M002 and MAR-M247 also plays a role in favoring freckle formation since it lowers the overall content of the interdendritic liquid in heavy elements such as W, Ta and Hf. It is interesting to note that the argument suggesting a high Ta content, to increase the interdendritic liquid density and to prevent freckling [3], will only be valid in non carbide-forming alloys (such as SX castings). In carbon containing alloys, increasing the Ta content might on the contrary favor density inversion at the carbide precipitation temperature. In the case of C-276, density is found to increase slightly with depth in the mushy zone (see Figure 36) and freckles were calculated to be slightly heavier than the matrix. However, although no density inversion is observed, all the densities are in a very narrow range (less than 0.1 g/cm difference). Therefore, even though C-276 contains only relatively small amounts of Al , C, B, Zr, Si, segregation of these light elements might create a density inversion sufficient to 93 produce freckles in a manner similar to that in IN718. In any case, because of the small concentrations involved, E D X analysis may not be suitable and further work is required. Estimated interdendritic liquid density along the mushy zone in M2 7.25 H 1 1 1 1 1 1200 1250 1300 1350 1400 1450 Temperature (°C) Figure 47 : Calculated density profiles in tool steel M2, assuming 3 different carbon segregation patterns. 94 £ _ C O N C L U S I O N 6.1. S U M M A R Y In the literature review part of this thesis, the characteristic features of freckles were described, the current theory on freckle formation, based on density inversion, was explained and the main experiments and mathematical models were reported. Based on this extensive literature review, it became obvious that virtually no quantitative research had been carried out on specific industrial alloys. The objective of this thesis was to start this quantitative research on a few selected industrial alloys and to estimate the amount of density inversion associated with freckling. Along the mushy zone of directionally solidified and quenched alloys (IN718, MAR-M002, MAR-M247, C-276 and Tl) , segregation profiles, as well as fraction liquid profiles in some cases, were measured and showed very good agreement with literature data. Several freckle compositions, as well as matrix and eutectic compositions, were also measured and/or reported from the literature. Associated densities were computed by a mathematical model. The following conclusions could be drawn : (1) It was found that density inversion as low as 0.01 (g/cm 3 ) /°C could possibly initiate freckles, in castings with the range of dendrite spacing found in superalloys (typically 100-400um). (2) In the case of superalloys, freckle compositions indicated that freckles initiated at about 20°C below the liquidus temperature, at the top of the mushy zone where the fraction liquid is in the range 40 to 60% and permeability was high enough to allow fluid flow. 95 (3) It was also found in some carbon-containing alloys, as is the case for MAR-M002, T l and M2, that freckle initiation was probably triggered by carbide precipitation, removing heavy elements (such as Ta, W, Hf, Mo) from the interdendritic liquid, creating the necessary density inversion for freckling. (4) The influence of the segregation of minor light elements such as C, Si, Zr, A l , Mn, P was also discussed as a possible significant factor in freckle initiation. However, low atomic number and/or low overall concentration levels made the effect of these elements difficult to quantify by simple EDX measurements. More work would however be necessary in order to further validate some of these results. (LL. R E C O M M E N D A T I O N S F O R F U T U R E W O R K The main basis of this work relied upon the measurement of interdendritic liquid composition profiles by EDX microprobe. However, localization of the interdendritic liquid on cross-sections of a DSQ sample was not always obvious. Thus, the segregation profiles reported in this work should be validated by an alternate method. Such method could be to draw a minute amount of interdendritic liquid from a known position in the mushy zone through a fine alumina tube and compare its composition (measured afterwards by E D X microprobe) to that given by the segregation profile. Such method, physically isolating the metal of interest, could also possibly allow the chemical analysis of some light elements (such C), which is not currently possible by EDX. This thesis also relied heavily on the density calculations by "METALS". Although in very good agreement with published data, this model cannot handle accurately enough some 96 elements such as carbon and thus, cannot provide results with the precision of 0.1 g/cm required for freckle study. Density of various alloy compositions at various temperatures should actually be measured physically rather than computed. Several methods of density measurements have been reported in the literature [48]. A precise differential thermal analysis (DTA) of the various chosen alloys would also be of great interest to determine accurately the melting range as well as the carbide precipitation temperature. The trigger effect of carbide precipitation on freckling could thus be validated. The freckle theory, although widely accepted, could also be unambiguously validated in the case of many industrial alloys by directionally solidifying a superalloy casting big enough to produce freckles and then quenching it. The quench should exhibit a plume above the mushy zone, if indeed freckles in industrial alloys are similar to those in analog systems. This kind of test would also eliminate any possible confusion between freckles and center-segregation. Finally, in order to ultimately develop a criterion for freckling, more work is also required to determine permeability expressions as well as critical values (such as Ra*T/s) for various industrial alloys. 97 References [I] M.C. Flemings : "Segregation in castings and ingots" 1990 Elliott Symposium Proceedings / pp.253-272 [2] K.O. Yu, J.A. Domingue : "Control of solidification structure in VAR andESR processed alloy 718 ingots" Superalloy 718, Min., Met. and Mat. Soc, E.A. Loria Ed., 1989 / pp.33-48 [3] T .M. Pollock, W.H. Murphy, E H . Goldman, D.L. Uram, J.S. Tu : "Grain defect formation during directional solidification of nickel base single crystals" Superalloys 1992, Min., Met. and Mat. Soc, S.D. Antolovitch et al. Ed., 1992 / pp. 125-134 [4] T .M. Pollock, W.H. Murphy : "The breakdown of single crystal solidification in high refractory nickel-base alloys" To be published (Met. Trans.) [5] K.O. Yu, J.A. Domingue, G.E. Maurer, H.D. Flanders : "Macrosegregation in ESR and VAR processes" Journal of Metals, Jan. 1986 / pp.46-50 [6] A.F. Giamei, B.H. Kear : "On the nature of freckles in nickel base superalloys" Met. Trans., Vol. 1, August 1970 / pp.2185-2192 [7] R.J. McDonald, ID . Hunt: "Fluid motion through the partially solid regions of a casting and its importance in understanding A type segregation" Trans. Metall. Soc. AIME, Vol. 245, Sept. 1969/pp. 1993-1997 [8] S.M. Copley, A.F. Giamei, S.M. Johnson, M.F. Hornbecker : "The origin of freckles in unidirectionally solidified castings" Met. Trans., Vol. 1, August 1970 / pp.2193-2204 [9] J.R. Sarazin, A. Hellawell, R.S. Steube : "Channel convection in partly solidified systems" Phil. Trans. R. Soc. Lond. A (1993) 345 / pp.507-544 [10] J.J. Moore, N.A. Shah : "Mechanisms offormation of A- and V-segregation in cast steel" Int. Metals Reviews, Vol. 28, No. 6, 1983 / pp.338-355 [II] L . Wang, V. Laxmanan, J.F. Wallace : "Gravitational macrosegregation in unidirectionally solidified lead-tin alloy" Met. Trans. A, Vol. 19A, Nov. 1988 / pp.2687-2694 98 [12] R.G. Carlson, J.F. Radavich : "Microstructural characterization of cast 718" Superalloy 718, Min., Met. and Mat. Soc., E.A. Loria Ed., 1989 / pp.79-95 [13] J.F. Radavich : "The physical metallurgy of cast and wrought alloy 718" Superalloy 718, Min., Met. and Mat. Soc, E.A. Loria Ed., 1989 / pp.229-240 [14] J.R. Sarazin, A. Hellawell: "Channel flow in partly solidified alloy systems" Advances in Phase Transition, Oct. 1987 / pp. 101-115 [15] S. Morimoto, A. Yoshinari: "High speed single crystal casting technique " Superalloys 1988, The Metallurgical Society, S. Reichman et al. Ed., 1988 / pp.325-334 [16] A. Mitchell: "Actual and potential manufacturing problems in the production of superalloy components for industrial gas turbine application Advanced Materials and Processing Conference, Vol. 1, K.S. Shin et al Ed, June 1995 / pp.133 [17] R.B. Mahapatra, F. Weinberg : "The columnar to equiaxed transition in tin-lead alloys" Met. Trans. B, Vol. 18B, June 1987 / pp.425-432 [18] I. Ziv, F. Weinberg "The columnar-to-equiaxed transition in Al 3 pet Cu " Met. Trans. B, Vol. 20B, October 1989 / pp.731-734 [19] J.C. Heinrich, S. Felicelli, D R . Poirier : "Vertical solidification of dendritic binary alloys" Computer Methods in Applied Mech. and Eng. 89 (1991) / pp.435-461 [20] J.C. Heinrich, S. Felicelli, D R . Poirier : "Simulation offreckles during vertical solidification of binary alloys" Met. Trans. B, Vol. 22B, Dec. 1991 / pp.847-859 [21] J.C. Heinrich, S. Felicelli, D R . Poirier : "Numerical model for dendritic solidification of binary alloys" Num. Heat Transfer, Part B, Vol. 23 (1993) / pp.461-481 [22] D.G. Neilson, F.P. Incropera : "Effect of rotation on fluid motion and channel formation during unidirectional solidification of a binary alloy" Int. J. Heat Mass Transfer, Vol. 36, No. 2, 1993 / pp.489-505 99 [23] A.C. Fowler : "The formation of freckles in binary alloys" IMA Journal of Applied Mathematics (1985), 35 / pp. 159-174 [24] J.R. Sarazin, A. Hellawell: "Segregation arising from natural convection during solidification" EPD Congress '91, Min., Met. and Mat. Soc, D R . Gaskel Ed., 1991 / pp.511-522 [25] G.B. McFadden, R.G. Rehm, S.R. Corriel, W. Chuck, K.A. Morrish : "Thermosolutal convection during directional solidification" Met. Trans. A, Vol. 15A, Dec 1984 I pp.2125-2137 [26] A. Sample, A. Hellawell: "The effect of mold precession on channel and macrosegregation in ammonium chloride-water analog castings" Met. Trans. B, Vol.l3B, Sept. 1982/pp.495-501 [27] A .K. Sample, A. Hellawell: "The mechanisms of formation and prevention of channel segregation during alloy solidification" Met. Trans. A, Vol. 15A, Dec 1984 / pp.2163-2173 [28] J.R. Sarazin, A. Hellawell: "Channel formation in Pb-Sn, Pb-Sb, and Pb-Sn-Sb alloy ingots and comparison with the system NH4CI-H20" Met. Trans. A, Vol. 19A, July 1988 / pp.1861-1871 [29] M R . Bridge, M.P. Stephenson, J. Beech : "Direct observations on channel segregate formation in aluminum alloys" Metals technology, Nov. 1982, Vol. 9 / pp.429-433 [30] N. Streat, F. Weinberg : "Interdendriticfluidflow in a lead-tin alloy" Met. Trans. B, Vol.7B, Sept. 1976 / pp.417-423 [31] D R . Poirier: "Permeability for flow of interdendritic liquid in columnar-dendritic alloys" Met. Trans. B, Vol. 18B, March 1987 / pp.245-255 [32] M.H. McCay, T.D. McCay : "Experimental measurement of solutal layers in unidirectional solidification" J. Thermophysics, Vol. 2, No. 3, July 1988 / pp. 197-202 [33] S.Asai, I. Muchi: "Theoretical analysis and model experiments on the formation mechanism of channel type segregation" Trans. ISIJ, Vol. 18, 1978 / pp.90-98 100 [34] M.C. Flemings, G.E. Nereo : "Macrosegregation: Part I" Trans. Metall. Soc. ATME, Vol. 239, Sept. 1967 / pp. 1449-1460 [35] M.C. Flemings, R. Mehrabian, G.E. Nereo : "Macrosegregation: Part II" Trans. Metall. Soc. ATME, Vol. 242, Jan. 1968 / pp.41-49 [36] M.C. Flemings, G.E. Nereo : "Macrosegregation: Part III" Trans. Metall. Soc. ATME, Vol. 242, Jan. 1968 /pp.50-55 [37] R. Mehrabian, M . Keane, M.C. Flemings : "Interdendritic fluid flow and macrosegregation; influence of gravity" Met. Trans., Vol. 1, May 1970 / pp. 1209-1220 [38] K.O. Yu, J.A. Oti, W.S. Walston : "Investment casting of NiAl single-crystal alloys" JOM, May 1993 /pp.49-51 [39] W D . Bennon, F.P. Incropera : "The evolution of macrosegregation in statically cast binary ingots" Met. Trans. B, Vol. 18B, Sept. 1987/pp.611-616 [40] W.D. Bennon, F.P. Incropera : "A continuum model for momentum, heat and species transport in binary solid-liquid phase change systems : I. Model formulation" Int. J. Heat Mass Transfer, Vol. 30, No. 10, 1987 / pp.2161-2170 [41] W.D. Bennon, F.P. Incropera : "A continuum model for momentum, heat and species transport in binary solid-liquid phase change systems : II. Application to solidification in a rectangular cavity" Int. J. Heat Mass Transfer, Vol. 30, No. 10, 1987 / pp.2171-2187 [42] W D . Bennon, F.P. Incropera : "Numerical analysis of binary solid-liquid phase change using a continuum model" Num. Heat Transfer, Vol. 13, 1988 / pp.277-296 [43] A L . Purvis, C R . Hanslits, R.S. Diehm : "Modeling characteristics for solidification in single-crystal, investment cast superalloys" JOM, Jan. 1994/pp.38-41 [44] A S M International Metals Handbook : "Properties and selection : irons, steels and high performance alloys" A S M International, Vol. 1, 10th edition, J.R. Davis et al. Ed., 1990 / pp.950-1006 101 [45] A. Grellier : "Toolsteel 18-4-1 solidification data" Aubert & Duval, private communication (28/07/95). [46] P. Bates : "Toolsteel Tl data" Special Melted Products, private communication (18/07/95). [47] N. Streat: "Interdendritic fluid flow" P h D Thesis, Univ. Of British Columbia, Dec. 1973 [48] A.F. Crawley : "Densities of liquid metals and alloys" Int. Metallurgical Reviews, Vol. 19, 1974/pp.32-47 [49] A. Sharan, T. Nagasaka, A.W. Cramb : "Densities in liquidFe-Ni andFe-Cr alloys" Met. Trans. B, Vol. 25B, December 1994 / pp.939-942 [50] K.C. Mills, P.N. Quested : "Measurements of the physical properties of liquid metals" Liquid Metal Processing and Casting Conf. Proa, Am. Vacuum Soc, A. Mitchell et al. Ed., Sept. 1994/pp. 226-240 [51] T. Iida, R.I.L. Guthrie : "The physical properties of liquid metals" Clarendon Press, 1988 [52] S.Yu. Denisov, O.I. Ostrovskii, N.V. Gritsenko, S.Yu. Nefedov, G.F. Stasyuk : "Viscosity and density ofFe-W-C melts and their change during solidification" Steel in the USSR, Vol. 16, May 1986 / pp.213-214 [53] R.Sellamuthu, A.F. Giamei: "Measurement of segregation and distribution coefficients in MAR-M200 and Hafnium-modified MAR-M200 superalloys" Met. Trans. A, Vol. 17A, March 1986 / pp.419-428 [54] G.K. Bouse, J.R. Mihalisin : "Metallurgy of investment cast superalloy components" Superalloys, Supercomposites and Superceramics, J.K. Tien et al. Ed. 1989 / pp.99-148 [55] W. Kurz, D.J. Fisher : "Fundamentals of solidification" Trans Tech Publications, 1992 / pp. 123-125 102 [56] Y. Haruna, A. Mitchell, A.J. Schmalz : "Some observations on the recycling of superalloys by EBCHM" Liquid Metal Processing and Casting Conf. Proa, Am. Vacuum Soc, A. Mitchell et al. Ed., Sept. 1994/pp. 181-204 [57] G.A. Knorovsky, M.J. Cieslak, T.J. Headley, A D . Romig,Jr., W.F. Hammetter : "Inconel 718 : a solidification diagram " Met. Trans. A, Vol. 20A, October 1989 / pp.2149-2158 [58] K . L . Zeisler-Mashl, B.J: Pletka : "Segregation during solidification in the MAR-M247 system " Superalloys 1992, Minerals, Met. and Mat. Soc, S.D. Antolovitch et al. Ed. 1992 / pp. 175-184 [59] D. Ford : "MAR-M002 initial partition coefficients" Rolls-Royce, private communication (07/02/95). [60] A. Mitchell: "The present status of melting technology for alloy 718 " Superalloy 718, Minerals, Metals and Materials Society, E.A. Loria Ed., 1989 / pp. 1-15 [61] M.J. Cieslak, G.A. Knorovsky, T.J. Headley, A D . Romig,Jr. : "The solidification metallurgy of alloy 718 and other Nb-containing superalloys" Superalloy 718, Minerals, Metals and Materials society, E.A. Loria Ed., 1989 / pp.59-68 [62] B. Radhkrishnan, R.G. Thompson : "Solidification of the nickel-base superalloy 718 : a phase diagram approach " Met. Trans. A, Vol. 20A, December 1989 / pp.2866-2868 [63] A. Mitchell, S.L. Cockcroft, C.E. Schvezov, A.J. Schmalz, J-N. Loquet, J. Fernihough : "Primary carbide and nitride precipitation in superalloys containing niobium " To be published (J. of High Temp. Mat.) [64] A.S. Ballantyne, A. Mitchell: "Prediction of structure in industrial VAR, ESR and PAR ingots using computed local solidification times" Sheffield Metall. and Eng. Ass., Conf. Proc, July 1976 / pp.363-370 [65] H. Brandis, E . Haberling, H.H. Weigand : "Metallurgical aspects of carbides in high speed steels" ATME conference proceedings, February 1980 / pp. 1-18 [66] H. Fredriksson, M . Nica : "The influence of vanadium, silicon and carbon on the eutectic reaction in M2 high speed steels" Scandinavian J. of metallurgy, 8, 1979 / pp.243-253 103 [67] P. Auburtin, S. Cockcroft, J. Feraihough, A. Mitchell, A. Schmalz : "Metallurgical aspects of the casting of single-crystal airfoils for use in aero-engines and industrial gas turbines" Materials Engineering in Turbines and Compressors Conf. Proc, R.D. Conroy et al. Ed., April 1995/pp.547-556 [68] K.S. Oh, Y.K. Shin, Y.W. Chang : "The role of combination stirring and final stirring pool thickness on center defects of continuously cast high carbon steel blooms" Iron & Steelmaker, Apr. 1994 / pp.43-55 [69] R.C. Weast: "Handbook of chemistry and physics" CRC Press, 58th edition, 1977-78 [70] AsbJ0rn Mo : "An internal variable description of solidification suitable for macrosegregation modeling" Met. Trans. B, Vol. 25B, Aug. 1994 / pp.597-605 104 Appendix A : Expressions of the permeability factor Permeability is a key factor in the determination of a freckling criterion such as the thermosolutal Rayleigh number. Numerous permeability factors have been reported and argued upon in the literature. Various common expressions to calculate the permeability K are presented below. In Al-Cu ingot castings, Flemings and co-workers [37] assumed an isotropic permeability K, linked to the volume fraction liquid/^ by the following relationship : K = yxf? [Eq.Al] (Where y depends on the primary dendrite arm spacing.) Streat and Weinberg [30] derived from the Hagen-Poiseuille equation in lead-tin ingots : * = ^ T - A 2 [Eq.A2] (Where T is a tortuosity factor estimated equal to 4.6) The permeability has also been found depending on the cooling rate 8. The following experimental equation was determined for steel ingots [68]: In K = In 0.10s0 8 1 + 19.6e0 3 3 lnfL [Eq.A3] Poirier [31] also published various experimental equations about the permeability in Pb-20wt%Sn and borneol-paraffin systems. The permeability was usually found to be fairly well described by equations of the type : K = y.X\X\fl [Eq.A4] (Where x, y, z are experimentally fitted exponents (usually between 0.5 and 4)) Other expressions include the following : (Where K0 is a constant depending on the mushy zone structure) X] - K depending on dendrite arm spacing [23]: K = — - [Eq.A6] Xr\ 2 3 (Where X is a numerical factor of the order of 10 or 10 ) Note : In most cases, these mathematical calculations of the permeability are valid only for a limited range of liquid fractions (typically 0.2<fL<0.6). Liquid fractions can be determined by the lever rule or the Scheil equation, as described in Appendix B. Kozeny-Carman equation [33,39] : K = K0 \2 J [Eq.A5] 105 Appendix B : The Scheil and Lever rules The Scheil and Lever rules are two extreme simplification equations derived from a more complex (and numerically non integrable) diffusion equation. They are very useful to relate the average composition of the casting C0, the composition of the solidifying solid C*s, the composition of the liquid Q , the partition coefficient k and the fraction liquid fL at any given stage during solidification. The practical equations, in their integrated form, are presented below [55]: C 1 Scheil equation: — = —rr- [Eq.Bl] Co fi L C 1 Lever rule : — = [Eq.B2] C 0 \-(\-k)(l-fL) In both cases, C s is then given by : C s = k.CL [Eq.B3] In both cases, the liquid is always assumed to be uniform in solute concentration (i.e. diffusion of the solute throughout the liquid is much faster than the actual solidification event). This assumption can be mathematically written : h (Where : DLSolute is the solute diffiisivity in the liquid h is the system size) In the case of DSQ samples : rrSolute« 5.10"9 m2/s [8] h « 2.10"4 m (half the primary dendrite spacing) LST » 4 0 0 s This yield a numerical value for the left hand side of equation [Eq.B4] of the order of 50, verifying the assumption. In the case of the Scheil equation, a further assumption is that the solid state solute diffusion is negligible. On the contrary, the Lever rule assumes total homogeneity of the solid concentration (high rate of diffusion of the solute). Reality lies in-between these two rules. In the case of the DSQ samples, it is found that: Km"!*? x l Q - 3 < < x [ E q B 5 ] h (Where : rfSolute = solute diffiisivity in the solid : I?Solute «10" 1 3 m2/s [69]) This confirms the usual assumption of negligible solid state diffusion in industrial castings. Therefore, actual segregation in the DSQ samples should in theory follow the Scheil equation rather than the Lever rule. Note : If the partition coefficient k is not constant during solidification (which is true as a general rule), the above equations cannot be used in their integrated form and require numerical integration by computer. Computer efficient codes have been reported in the literature [70]. 106 Appendix C : DSQ samples E D X data. IN718 Tuq-Tsoi'-G-1336 °C 1260 °C 9.4 °C/mm I n t e r - D e n d r i t i c " L i q u i d " C o m p o s i t i o n Average values (wt%) (zso/= 7.81 mm) z (mm) 15.90 15.65 15.10 14.30 13.50 12.40 10.80 9.50 7.95 7-(°C) 1336.0 1333.7 1328.5 1321.0 1313.4 1303.1 1288.1 1275.8 1261.3 Al 0.62 0.51 0.51 0.45 0.50 0.42 0.48 0.31 0.18 Ti 0.99 1.11 1.15 1.40 1.48 1.60 1.58 1.60 1.74 Cr 14.98 14.78 14.68 13.83 13.34 13.05 12.36 12.64 11.92 Fe 17.74 17.28 16.98 15.82 15:17 14.56 13.95 14.33 13.36 Co 0.46 0.49 0.49 0.49 0.42 0.41 0.38 0.47 0.42 Ni 54.92 54.63 54.41 54.10 52.99 52.62 52.22 52.85 51.90 C u 0.46 0.47 0.41 0.51 0.46 0.19 0.33 0.35 0.51 Nb 6.14 7.08 7.41 9.68 11.52 12.88 14.42 13.32 15.62 Mo 3.71 3.66 3.94 3.73 4.12 4.28 4.27 4.13 4.35 Total 100.00 100.00 100.00 100.00 100.00 100.01 100.00 100.00 100.00 P(g/crrf) 7.49 7.51 7.52 7.55 7.57 7.60 7.62 7.65 7.711 D e n d r i t e - C e n t e r C o m p o s i t i o n (wt%) Average values Al 0.54 Ti 0.59 Cr 15.91 Fe 19.58 Co 0.48 Ni 56.06 Cu 0.45 Zr N/R Nb 2.73 Mo 3.67 Total 100.00 Mat r i x a r o u n d E u t e c t i c P r e c i p i t a t e s (in w t%) Average values Al 0.19 Ti 1.41 Cr 14.22 Fe 16.10 Co 0.39 Ni 55.06 Cu 0.36 Zr N/R Nb 8.15 Mo 4.12 Total 99.99 E u t e c t i c P r e c i p i t a t e s (in w t%) Average values Al 0.00 Ti 2.02 Cr 6.65 Fe 7.98 Co 0.11 Ni 48.18 Cu 0.12 Zr 1.86 Nb 28.93 Mo 4.18 Total 100.00 (N/R = Not Recorded) Table C I : Numerical data measured by EDX microprobe on DSQ IN718. 107 MAR-M002 Tuq-Tsol  = G = 1 3 6 5 ° C 1249 °C 9.4 ° C / m m I n t e r - D e n d r i t i c " L i q u i d " C o m p o s i t i o n Average values (wt%) ( z S o / = 7.66 m m ) z (mm) 20.00 19.40 18.90 17.70 17.00 15.40 12.95 11.30 10.50 8.60 7 (°C) 1365.0 1359.4 1354.7 1343.4 1336.8 1321.8 1298.7 1283.2 1275.7 1257.8 Al 6.31 7.36 7.67 6.89 6.54 6.03 5.59 5.46 4.96 4.98 Ti 1.49 1.56 1.70 2.03 1.89 1.88 2.10 2.05 2.00 2.03 Cr 8.81 8.77 8.91 8.92 8.75 8.48 9.15 9.04 9.21 8.91 Co 10.10 9.88 9.63 9.51 9.70 9.75 8.71 8.95 8.93 8.76 Ni 57.67 57.59 57.47 56.50 56.32 55.54 57.72 57.92 56.52 56.69 Hf 2.38 2.15 2.42 2.76 1.97 1.57 9.18 8.56 11.02 12.11 T a 3.14 3.15 3.21 4.19 4.56 5.27 2.59 2.26 2.26 1.73 W 10.11 9.56 9.00 9.21 10.28 11.46 4.96 5.76 5.10 4.79 Total 100.00 100.00 100.00 100.00 100.00 99.99 100.00 100.00 100.00 99.99 p(g/cm 3) 7.15 6.95 6.89 7.05 7.16 7.33 7.23 7.27 7.40 7.41 D e n d r i t e - C e n t e r Ma t r i x a r o u n d E u t e c t i c C a r b i d e s C o m p o s i t i o n E u t e c t i c P r e c i p i t a t e s P r e c i p i t a t e s ( C h i n e s e S c r i p t ) (wt%) (in w t%) (in w t%) (in w t % ) Average values Average values Average values Average values Al 5.77 Al 7.34 Al 1.72 Al 0.72 Ti 0.84 Ti 2.68 Ti 1.00 Ti 13.22 Cr 8.49 Cr 5.97 Cr 7.47 Cr 0.42 Co 10.97 Co 7.90 Co 6.88 Co 0.18 Ni 57.72 Ni 62.68 Ni 44.67 Ni 2.13 Zr N/R Zr N/R Zr 2.84 Zr 0.81 Hf 1.13 Hf 6.58 Hf 33.03 Hf 15.30 T a 2.41 Ta 2.42 Ta 0.70 T a 51.91 W 12.66 W 4.44 W 2.25 W 15.52J Total 100.00 Total 100.00 Total 100.57 Total 100.20J (N/R = Not Recorded) Table C2 : Numerical data measured by EDX microprobe on DSQ MAR-M002. 108 MAR-M247 Tuq-Tsoi = G = 1360 °C 1280 °C 9.4 °C /mm I n t e r - D e n d r i t i c " L i q u i d " C o m p o s i t i o n Average values (wt%) ( z S o / = 8.79 m m ) z (mm) 17.30 16.20 14.60 13.15 11.15 9.85 8.80 7.50 7 T Q 1360.0 1349.7 1334.6 1321.0 1302.2 1290.0 1280.1 1267.9 Al 6.38 5.95 6.95 6.71 6.02 6.08 5.87 6.28 Ti 1.07 1.14 1.52 1.78 1.75 1.64 1.64 1.50 Cr 8.50 8.52 8.77 8.58 8.67 8.45 8.95 9.18 Co 10.05 10.24 9.34 8.57 8.66 8.77 8.67 8.59 Ni 58.49 58.13 57.47 56.60 56.28 56.67 55.28 53.45 Mo 0.68 0.85 0.91 0.84 1.04 0.84 1.11 1.41 Hf 2.09 1.91 3.13 6.97 9.14 9.44 11.06 11.47 Ta 3.26 3.57 4.40 4.21 3.19 3.17 2.98 3.11 W 9.48 9.70 7.52 5.74 5.27 4.95 4.44 5.00 Total 99.99 100.00 100.01 100.00 100.01 100.00 99.99 100.00 p (g/cm3) 7.15 7.25 7.04 7.09 7.22 7.22 7.27 7.26 D e n d r i t e - C e n t e r C o m p o s i t i o n (wt%) Average values Al 5.11 Ti 0.58 Cr 7.74 Co 11.10 Ni 58.03 Mo 0.61 Hf 1.08 T a j 2.87 W 12.89 Total 100.00 M a t r i x a r o u n d E u t e c t i c P r e c i p i t a t e s (in w t%) Average values Al 6.23 Ti 1.55 Cr 11.39 Co 10.11 Ni 55.65 Mo 1.09 Hf 5.40 T a 2.37 W 6.21 Total 100.00 Eutectic P r e c i p i t a t e s (in w t%) Average values Al 1.30 Ti 0.76 Cr 2.65 Co 5.79 Ni 46.42 Mo 0.44 Hf 37.98 Ta 2.46 W 2.19 Total 99.99 C a r b i d e s ( C h i n e s e S c r i p t ) (in w t % ) Average va lues Al 0.71 Ti 10.06 Cr 0.58 Co 0.00 Ni 2 46 Mo 1.45 Hf 15.38 T a 54.88 W 14.49 Total 100.00I Table C3 ; Numerical data measured by E D X microprobe on DSQ MAR-M247. 109 C-276 Tuq-Tsol  = G = 1370 °C 1325°C 9.4 °C /mm I n t e r - D e n d r i t i c " L i q u i d " C o m p o s i t i o n Average values (wt%) (Zsoi= 12.51 mm) z (mm) 17.30 16.00 15.00 13.95 12.65 7~fC) 1370.0 1357.8 1348.4 1338.5 1326.3 Cr 14.49 14.87 14.98 14.82 14.56 Fe 6.58 6.08 5.92 6.09 5.58 Ni 56.29 52.93 50.25 51.80 49.33 Mo 18.31 21.68 25.28 23.09 26.52 W 4.33 4.44 3.56 4.21 4.01 Total 100.00 99.99 99.99 100.00 100.00 P (g/cm3) 8.15 8.23 8.26 8.26 8.33 D e n d r i t e -Cen te r Ma t r i x a r o u n d E u t e c t i c C o m p o s i t i o n E u t e c t i c P r e c i p i t a t e s P r e c i p i t a t e s (wt%) (in wt%) (in w t%) Average values Average values Average values Cr 14.29 Cr 14.92 Cr 11.29 Fe 6.88 Fe 5.61 Fe 3.53 Ni 58.79 Ni 50.51 Ni 27.47 Zr N/R Zr 0.25 Zr 2.32 Mo 15.76 Mo 25.66 Mo 50.48 W 4.28 W 3.04 W 4.92 Total 100.00 Total 99.99 Total 100.00 (N/R = Not Recorded) Table C4 : Numerical data measured by E D X microprobe on DSQ C-276. 110 T1 Tuq-Tsoi  = G = 1440 °C 1320 °C 9.4 ° C / m m I n t e r - D e n d r i t i c " L i q u i d " C o m p o s i t i o n Average values (wt%) ( z S o / = 14.23 m m ) z (mm) 27.00 26.50 23.90 22.10 20.40 17.90 15.10 13.20 T ( ° C ) 1440.0 1435.3 1410.9 1393.9 1378.0 1354.5 1328.1 1310.3 V 1.08 1.22 1.37 1.43 1.57 1.89 1.78 1.84 Cr 5.09 4.92 4.99 5.03 5.13 5.59 5.43 5.50 M n N/R 0.46 0.47 0.40 0.47 0.46 0.43 0.43 Fe 75.31 73.73 71.74 71.04 69.37 66.59 66.99 66.47 W 18.53 19.68 21.44 22.11 23.48 25.47 25.37 25.76 Total 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 P(g/crrf) 7.99 8.04 8.16 8.22 8.31 8.45 8.47 8.51 D e n d r i t e - C e n t e r C o m p o s i t i o n (wt%) Average values V 0.89 Cr 3.90 Mn 0.37 Fe 76.80 W 18.14 Total 100.09 Ma t r i x a r o u n d C a r b i d e s (in w t%) Average values V 1.42 Cr 6.14 Mn 0.40 Fe 82.75 W 9.29 Total 100.00 Carbide P r e c i p i t a t e s (in w t%) Average values V 3.25 Cr 4.56 Mn 0.00 Fe 27.69 W 64.51 Total 100.00 Not Recorded) T a b l e C 5 : Numerical data measured by EDX microprobe on DSQ T l . I l l Appendix D : Graphic determination of the freckle initiation position Freckle composition for several alloying elements (found in Table 6) is plotted as horizontal dashed lines on the segregation profiles graph for IN718 (Figure 31). The intersection points of these horizontal lines with the segregation profiles all fall within a relatively narrow range of temperatures (about 1320°C). This is the temperature of freckle initiation. A similar construction can be done for MAR-M002, C-276 and T l . 18 17 16 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 S e g r e g a t i o n a n d f r a c t i o n l i q u i d p r o f i l e s a l o n g t h e m u s h y z o n e i n D S Q IN718 100 Freckle composition in Fe Freckle composition in Nb Tsoi ~ 1260°C T|_iq = 1336°C Freckle composition in Ti \ i i £ c 0) .2 in U-\ 1 1260 1280 1300 1320 Temperature in the mushy zone (in °C) 80 60 # I 40 '5. cr • c o U 2 20 0 1340 o Al A Ti O Cr • Fe • Co • Cu Nb a Mo - f l 112 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0078520/manifest

Comment

Related Items