<%BANNER%>

The Characterization of Freckle Casting Defects in Directionally Solidified Nickel-Base Superalloy Turbine Blades


PAGE 1

THE CHARACTERIZATION OF FR ECKLE CASTING DEFECTS IN DIRECTIONALLY SOLIDIFIED NICKEL-BASE SUPERALLOY TURBINE BLADES By ERIK M. MUELLER A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2003

PAGE 2

Copyright 2003 by Erik M. Mueller

PAGE 3

This document is dedicated to everyone across the world who dreams and hopes for a better world.

PAGE 4

ACKNOWLEDGMENTS I would like first and foremost to thank my parents and God for giving me the opportunity to attend the University of Florida and receive a higher education. Many people across the world are unable to receive such a gift, so I hope I have made the best use of my years at school. I would also like to thank my sister, Melissa, and all my grandparents, aunts, uncles, and cousins for their continual support throughout my life. My good friends, Andy, Chris, and Julio, have also been there to show me what real friendship is like and to support me during this time. Of chief importance is Dr. Michael Kaufman, who has guided me through this project and has also been there as a friend for me when I needed help. Dr. Amelia Dempere was instrumental in providing me with the monetary support to be in graduate school to do this project. Without her and the Major Analytical Instrumentation Center (MAIC), this project would never have been possible. I would also like to Dr. Richard Connell for introducing me to materials engineering and Dr. Gerhard Fuchs for introducing me to superalloys. iv

PAGE 5

TABLE OF CONTENTS Page ACKNOWLEDGMENTS.................................................................................................iv LIST OF TABLES............................................................................................................vii LIST OF FIGURES.........................................................................................................viii ABSTRACT.....................................................................................................................xiii CHAPTER 1. INTRODUCTION........................................................................................................1 2. LITERATURE REVIEW.............................................................................................5 2.1. Nickel-Base Superalloys........................................................................................5 2.1.1. Chemical Composition................................................................................5 2.1.2. Role of Alloying Elements..........................................................................7 2.1.3. Other Phases in Nickel-Base Superalloys...................................................9 2.2. Processing of Superalloys....................................................................................12 2.2.1. Investment Casting....................................................................................12 2.2.2. Directional Solidification (DS)..................................................................14 2.2.3. Microstructural Features............................................................................16 2.3. Freckles................................................................................................................19 2.3.1. Theories of Freckle Formation..................................................................20 2.3.2. Modeling....................................................................................................24 3. METHODS AND MATERIALS...............................................................................28 3.1. Materials..............................................................................................................28 3.2. Optical Microscopy.............................................................................................28 3.3. Preparation of Samples........................................................................................29 3.4. Stereology............................................................................................................30 3.5. SEM and TEM.....................................................................................................32 3.6. Electron Backscattered Diffraction......................................................................33 3.7. Nanoindentation...................................................................................................34 v

PAGE 6

4. RESULTS...................................................................................................................36 4.1. Microstructural Characterization.........................................................................36 4.1.1. Compositions.............................................................................................38 4.1.2. Surface of the Castings..............................................................................41 4.1.3. Eutectic Colonies and Porosity..................................................................45 4.1.4. Eutectic at Grain Boundaries.....................................................................48 4.1.5. Eta Phase Formation..................................................................................51 4.1.6. Morphology and Microstructure of Constituents near Surface.................57 4.2. Stereology Results...............................................................................................64 4.2.1. Stereology on DS-1 Casting......................................................................64 4.2.2. Stereology on DS-2 Casting......................................................................67 4.2.3. Stereology for Individual Freckle Grains..................................................69 4.3. EBSD...................................................................................................................72 4.4. Nanoindenter Results...........................................................................................90 5. ANALYSIS AND DISCUSSION..............................................................................96 5.1. Analysis of Phases...............................................................................................96 5.2. Freckle Formation..............................................................................................100 6. CONCLUSIONS......................................................................................................105 LIST OF REFERENCES.................................................................................................107 BIOGRAPHICAL SKETCH...........................................................................................111 vi

PAGE 7

LIST OF TABLES Table page 1. Common stereological formulas and relations to microstructural features...................32 2. Estimated nominal compositions of DS-1 and DS-2 alloy............................................39 3. EDS weight percents of elements for phases in Figure 36............................................56 4. Misorientations and common zone axes of DS-1 freckle grains...................................90 vii

PAGE 8

LIST OF FIGURES Figure page 1. Binary phase diagram of Ni and Al.................................................................................6 2. Crystal structures of FCC Ni, and Ni 3 Al, .................................................................7 3. The investment casting process.....................................................................................13 4. Schematics of (a) a DS furnace and (b) a single crystal furnace...................................15 5. Isosurfaces of volume fraction and stream traces emerging from a channel.................21 6. Isosurfaces of solute concentration, showing solute enrichment within channels and the emerging plumes......................................................................................................22 7. The effect of growth velocity (R) and thermal gradient (G) on freckle formation........24 8. Illustration showing the castings pinching back on itself after being cut by the wire EDM.........................................................................................................................29 9. Illustration of the directions cut in relation to the primary dendrite arms (not to scale).........................................................................................................................30 10. Manual counting methods used for stereology............................................................31 11. Illustration of indent head mechanism in the Triboindenter .......................................34 12. Photograph of the DS-1 castings and a labeled freckle chain......................................37 13. Photograph of the DS-2 castings with some visible freckle chains labeled................37 14. EDS spectrum of DS-2 alloy at 15 kV.........................................................................38 15. Comparison of freckle chain and matrix compositions in weight percent from DS-1 alloy without Ni........................................................................................................40 16. Comparison of freckle chain and matrix compositions in weight percent from DS-2 alloy without Ni........................................................................................................40 17. SEM (Scanning Electron Micrograph) of etched DS-1 surface showing dendritic structure....................................................................................................................41 viii

PAGE 9

18. SEM of etched surface of DS-2 casting revealing dendritic structureof freckle chain. ......................................................................................................................42 19. SEM of unaltered DS-2 alloy surface (100X)..............................................................43 20. SEM of surface of DS-2 casting showing charging alumina phase (650X)................43 21. BSE (Backscattered Electron) SEM of Figure 20........................................................44 22. X-Ray map of Figure 21 revealing the Al enrichment in the charging phase and the Ta and Ti enrichment in the high Z phase................................................................44 23. Comparison of EDS data from matrix and freckle regions of unaltered surface of DS-2 alloy.......................................................................................................................45 24. SEM of eutectic colony at a freckle boundary in DS-2 alloy (1000X)........................46 25. SEM of a eutectic colony in DS-1. Note the web-like morphology of the (2000X)....................................................................................................................46 26. SEM of different section of eutectic in DS-2 alloy at the casting surface (2000X)....47 27. Optical micrograph of cross section of DS-2 casting..................................................47 28. SEM of porosity near a colony of at a freckle grain boundary (1000X)..................48 29. SEM of DS-2 alloy near surface showing elongated at the boundaries between three freckles.....................................................................................................................49 30. SEM of DS-1 alloy at boundary of freckle grains and matrix.....................................49 31. SEM of DS boundary in DS-1 (5500X).......................................................................50 32. SEM showing intergranular at a freckle grain boundary in DS-2 (900X)................50 33. SEM of intergranular between two eutectic colonies at a freckle grain boundary from DS-2 (750X)....................................................................................................51 34. Optical image of a freckle grain in DS-1 casting showing various phases (50X).......52 35. BSE SEM of freckle boundary in the DS-1 showing phase (500X)........................53 36. BSE SEM showing phase, carbides, and TCP phase at a DS-1 freckle boundary (1900X)....................................................................................................................53 37. X-ray map of Figure 36, showing elements common to phase................................54 38. SEM of Phase (3300X).............................................................................................54 ix

PAGE 10

39. X-ray map of Figure 38, showing elements common to phase................................55 40. SEM of MC carbide in matrix (10000X).....................................................................55 41. X-ray map of MC carbide in Figure 40.......................................................................56 42. BF TEM image of containing stacking faults stacking faults near a carbide (30000X)..................................................................................................................57 43. BF TEM closeup of stacking faults above the carbide from Figure 42 (100000X)....57 44. SEM of a cross section of the DS-2 casting surface at a freckle chain (1400X).........58 45. SEM of a large eutectic colony near the external surface of the DS-2 casting (1000X)....................................................................................................................59 46. SEM of large eutectic colony at surface (450X)..........................................................59 47. SEM of carbide in the center of large eutectic colony near casting surface (1800X)....................................................................................................................60 48. SEM of carbide near surface (2500X).........................................................................61 49. BSE SEM of Figure 42 (2500X)..................................................................................61 50. SEM of casting surface showing the "smearing" of the + morphology (3500X)..62 51. BSE SEM of bare surface of DS-2 near polished regions (3000X).............................63 52. BSE SEM of unpolished DS-2 surface........................................................................63 53. SEM of virgin surface near polished surface...............................................................64 54. Volume percent of DS-1 eutectic at 125X found through stereology.........................65 55. Volume percent of MC carbides in DS-1 at 1000X found through stereology...........66 56. Volume percent of DS-1 phase at 125X found through stereology.........................66 57. Volume percent of DS-1 porosity at 125X found through stereology.........................67 58. Volume percent of MC carbides in DS-2 at 1000X found through stereology...........68 59. Volume percent of DS-2 eutectic at 125X found through stereology.........................68 60. Volume percent of DS-2 porosity at 125X found through stereology.........................69 61. Volume percent of eutectic in DS-1 alloy at 125X......................................................71 x

PAGE 11

62. Volume percent of in DS-1 alloy at 125X................................................................71 63. Volume percent of eutectic in DS-2 alloy at 125X at cuts 2 and 3..............................72 64. EBSD Kikuchi map of + matrix in DS-1 casting...................................................73 65. A [001] Pole Figure of matrix and eutectic regions.....................................................74 66. Schematic of Oxford OPAL system showing directions in the pole figure from Figure 65..................................................................................................................75 67. SEM of adjacent to eutectic colony (950X).............................................................76 68. EBSD Kikuchi map of eutectic from Figure 67...........................................................77 69. EBSD Kikuchi map of from Figure 67.....................................................................78 70. Optical image of DS-1 freckle chain. (a) The unaltered image at 50X. (b) Individual freckle chains are numbered and colored.................................................................79 71. EBSD Kikuchi map of a spot probed in Freckle 1.......................................................80 72. The [001] Pole figure of freckle 1 in relation to the matrix. The freckle grain results are highlighted red....................................................................................................80 73. EBSD Kikuchi map of a spot probed in Freckle 2.......................................................81 74. The [001] Pole figure of freckle 2 in relation to the matrix. The freckle grain results are highlighted red....................................................................................................81 75. EBSD Kikuchi map of a spot probed in Freckle 3.......................................................82 76. The [001] Pole figure of freckle 3 in relation to the matrix. The freckle grain results are highlighted red....................................................................................................82 77. EBSD Kikuchi map of a spot probed in Freckle 4.......................................................83 78. The [001] Pole figure of freckle 4 in relation to the matrix. The freckle grain results are highlighted red....................................................................................................83 79. EBSD Kikuchi map of a spot probed in Freckle 5.......................................................84 80. The [001] Pole figure of freckle 5 in relation to the matrix. The freckle grain results are highlighted red....................................................................................................84 81. EBSD Kikuchi map of a spot probed in Freckle 6.......................................................85 82. The [001] Pole figure of freckle 6 in relation to the matrix. The freckle grain results are highlighted red....................................................................................................85 xi

PAGE 12

83. EBSD Kikuchi map of a spot probed in Freckle 7.......................................................86 84. The [001] Pole figure of freckle 7 in relation to the matrix. The freckle grain results are highlighted red....................................................................................................86 85. EBSD Kikuchi map of a spot probed in Freckle 8.......................................................87 86. The [001] Pole figure of freckle 8 in relation to the matrix. The freckle grain results are highlighted red....................................................................................................87 87. EBSD Kikuchi map of a spot probed in Freckle 9.......................................................88 88. The [001] Pole figure of freckle 9 in relation to the matrix. The freckle grain results are highlighted red....................................................................................................88 89. EBSD Kikuchi map of a spot probed in Freckle 10.....................................................89 90. The [001] Pole figure of freckle 10 in relation to the matrix. The freckle grain results are highlighted red....................................................................................................89 91. Nanohardness comparison of and matrix ...............................................................91 92. Nanohardness comparison of eutectic ', and matrix ............................................92 93. Nanohardness comparison for in the eutectic colonies and in the matrix.................92 94. Nanohardness comparison of the in the eutectic colonies and in the matrix............93 95. Complex modulus map of DS-1 sample; phase is shown on the left.......................94 96. Complex modulus map of DS-1 sample in the matrix regions showing cuboidal '...94 97. Relaxed modulus comparison of and from the matrix regions..............................95 98. Relaxed modulus comparison of matrix and eutectic '............................................95 99. Transverse optical image of freckle in an ingot of Gatorized Waspaloy.....................99 100. Optical image of Figure 99 at higher magnification.................................................99 xii

PAGE 13

Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science THE CHARACTERIZATION OF FRECKLE CASTING DEFECTS IN DIRECTIONALLY SOLIDIFIED NICKEL-BASE SUPERALLOY TURBINE BLADES By Erik M. Mueller August 2003 Chair: Dr. Michael Kaufman Major Department: Materials Science and Engineering The purpose of this project was to characterize equiaxed grains, or freckles, that appear on the surface of directionally solidified superalloy turbine blades. There has been extensive work on modeling the formation of freckle defects on castings using mathematical models and 3-dimensional imaging. Unfortunately, not all the model predictions can explain every microstructure and defect created in a system with upwards of ten components. During a post-casting etch, much of the surface of these castings is lostthis project analyzed two castings with unaltered, unetched surfaces in an attempt to observe the microstructure in these regions. Stereology was also used to quantify the purported increase in various microstructural features near the freckles. Indeed a marked increase in porosity, and + eutectic colonies was seen in freckled regions of the castings. Several morphologies of were observed, including grain boundary , adjacent to grain boundaries, and eutectic . It will be shown that these eutectic colonies xiii

PAGE 14

do not form by eutectic growth, but rather by cellular solidification. Further, there appears to be no preferred orientations of the freckle grains relative to the casting. Although not exclusive to freckle regions, the surface of the casting has an external alumina layer that is likely the result from a reaction between the ceramic mold and the Al in the liquid metal. Further, there are an unusually large number of MC carbides immediately underneath the alumina layer. These reactions, the freckle characteristics, and the microstructural evolution during single crystal growth will be described in some detail. xiv

PAGE 15

CHAPTER 1 INTRODUCTION Todays modern jet aircraft, both military and civilian, demand higher performance, efficiency, and endurance from their engines. For many decades, the aerospace industry has believed that ceramics would eventually fill the role as the material of choice for turbine blades and vanes in jet engines. However, no available ceramic has both the toughness and high temperature strength requirements necessary in these sometimes corrosive environments. Therefore, the material of choice continues to be superalloys, particularly those of alloyed nickel. Materials that provide relatively high strength and environmental resistance at elevated temperatures have been lumped into the category of high-temperature materials. When this subject is broached, materials that come to mind are usually ceramics, steels, refractory metals, superalloys, and certain carbon composites. However, one must take into account as to how one defines the phrase high temperature. Can a specific temperature threshold be assigned, or is it relative to each material? For example, 2024 aluminum begins to melt at 502C, and significant creep occurs above 300C. This could obviously not be used in critical aircraft engines that operate in excess of 1000C. Therefore, the general rule is to base the designation of high-temperature to 30 to 50% of the melting point of a material. Above this 50% melting point, diffusion and other atomic motion mechanisms accumulate and the material loses usefulness. Therefore, it takes careful materials selection to find a suitable material for this 1

PAGE 16

2 application. For the aforementioned turbine blade, one would choose a superalloy material. Other materials would likely be ruled out because Ceramics would not have the fracture toughness to withstand the high stresses in the turbine Refractory metals would oxide too fast Steels would creep and yield under these conditions. Thus, for these applications, Ni-based superalloys are unique in that they not only possess the high strength and toughness at high temperatures, but they also have effective resistance against oxidation and corrosion if properly alloyed. For aircraft turbine engines, the goal is to maximize the efficiency. By increasing the inlet temperature 56C (100F), the power output will increase 4%. 1 Nickel-base superalloys are distinctive in that, unlike most materials, they can operate at temperatures near 85% of their melting temperature. Today, aircraft engine manufacturers are busy trying to increase the engine output, which in turn requires better composition control. Nickel-base superalloys outperform cobalt-base superalloys because of the coherent precipitate phase, Ni 3 Al. This phase increases the strength and rupture life of a turbine blade dramatically, especially as temperature increases. 2 Besides alloy chemistry, there are other techniques to improve temperature capabilitiesdirectional solidification and thermal barrier coatings. Directional solidification (DS) greatly improves the rupture strength and life of a turbine blade by reducing the number of grain boundaries normal to the applied stress. These grain boundaries are the reason for failure at elevated temperatures due to grain boundary fracture. A step above DS superalloys is blades cast as single crystals. Single crystal processing eliminates grain boundaries altogether, and these parts have been shown to

PAGE 17

3 operate at temperatures 25-50C higher than the same alloys directionally solidified. In addition, fatigue life is improved as well. In spite of their superior properties, a major problem with using nickel-base superalloys is that they are extremely expensive. Many of the alloying elements (such as tantalum, rhenium, and ruthenium) cost in excess of $400 a pound. Add to the fact that the DS and single crystal casting methods are difficult to perform, and the price soars astronomically. Therefore, it is crucial to increase the quality control of the casting process and reduce the amount of material that has to be scrapped. A major problem in the turbine blade industry has been the formation of casting defects exclusive to DS and single crystal castings. These include scale and freckles. Superalloy scale is a rough raised layer on the casting surfaces that typically contains oxides and other undesirable phasesthis can usually be removed by sandblasting or low-pressure water blasting. Freckles and other internal defects, though, cannot be removed as their depth is considerably greater. Freckles in single crystal and DS castings appear as long chains of equiaxed grains that run vertically along the sides of the castings. They ruin the enhancements of the DS and single crystal processes by acting as failure sites for creep rupture. To detect freckles, the as-cast turbine blades are etched and visually inspected; if a freckle trail is discovered, the entire casting must be scrapped. Therefore, it is of significant importance to the jet propulsion and gas turbine industries to determine the cause of freckle defects and prevent their formation. Consequently, the goal of this project is to characterize these freckle defects in DS superalloy castings and understand the mechanism of their formation. There has been

PAGE 18

4 extensive work on modeling the formation of freckle defects on castings using mathematical models and 3-dimensional imaging. Unfortunately, not all of the model predictions correlate exactly to the actual castings defects. One of the major drawbacks is the etching process used to remove the surface of the casting to reveal freckles. A good portion of the surface layer is removed during this process, which may yield clues to the origin of these freckle defects. Furthermore, many researchers have commented on the increase in various microstructural features near the freckled areas of the castings. However, no true quantitative measurements have yet been made. In this study stereological methods will be applied to the castings to determine what, if any, differences exist in the volume fraction of microstructural features in and near freckles relative to the rest of the casting. Other characterization techniques will also be employed to investigate these features qualitativelyshape, size, and relative location. The orientation of these features and the freckle grains themselves, with respect to the casting, will be investigated as well.

PAGE 19

CHAPTER 2 LITERATURE REVIEW As stated in Chapter 1, most of the research in characterizing freckle defects in nickel-base superalloys and correlating them with the mathematical modeling results has already been produced. This section will explain, in depth, about nickel base superalloys, their processing, chemistry, and the nature and theory behind their defects including freckles. 2.1. Nickel-Base Superalloys Nickel base superalloys are often used for applications that require high strength and fracture toughness at elevated temperatures. These alloys are especially useful for the hottest areas of aircraft engines and industrial gas turbine engines (IGTs). This is because, in addition to being strong and tough, they are resistant to fatigue, oxidation, creep, and hot corrosion. These properties stem from the microstructure of the and phases of nickel-base superalloys, the addition of various elements, and the casting processes and heat treatments used. 2.1.1. Chemical Composition What makes superalloys unique is the coherent interface between the austenitic FCC structure and the L1 2 structure of Ni 3 Al (see Figure 1). 3 Both have nearly identical lattice parameters creating extremely low mismatch (see Figure 2). 2 The morphology of the is determined by the mismatch: spheres occur from 0 to 0.2% and cubes appear from 0.5 to 1%. 2 The amount of misfit can be controlled, in part, by the concentration of other formers. In fact, up to 60% of the Al in Ni 3 Al can be substituted 5

PAGE 20

6 by Ti, Ta, and Nb. The phase is a superlattice that shows long-range ordering (LRO) to its melting point. Since this is the primary source of Ni superalloy strength, the volume percentage of will greatly affect the strength of the alloy. Current commercial cast alloys have greater than 50% , which results in higher creep strength but dramatically reduces their hot workability. This increase in creep strength with increasing temperature is due to an anomalous increase of flow stress with temperature. 1 Figure 1 Binary phase diagram of Ni and Al

PAGE 21

7 Figure 2 Crystal structures of FCC Ni, and Ni 3 Al, 2.1.2. Role of Alloying Elements Modern superalloy compositions contain 8 or more elements, creating a very complex system to describe thermodynamically. It would be extremely difficult to construct and interpret an 8 to 10-dimensional phase diagram. 4 Furthermore, many of the sought after phases and microstructures of nickel-base superalloys are metastable and must be developed through careful heat treatments. Finally, many of these elements tend to segregate to different areas of the castingeither the dendrite core or the interdendritic regions. One of the base elements for any nickel-base alloy is chromium, and in fact, the original superalloy was Ni-20Cr. Cr is an important element in that its primary function is to protect the alloy from oxidation and other forms of corrosion. Cr will form a protective oxide layer on the surface that will adhere well to the surface with a matching coefficient of thermal expansion. Cr is particularly effective in preventing hot corrosion and sulfidation. Unfortunately, the Cr 2 O 3 layer is only effective until about 900C. Cr also acts as a solid solution strengthener and carbide former. 2

PAGE 22

8 Another major element used to environmentally protect nickel-based superalloys is aluminum. Al, like Cr, forms a protective layer on the surface, but is more stable than Cr 2 O 3 at higher temperatures. However, Al 2 O 3 does not protect well in corrosive environments, such as sulfur intensive hot corrosion, so both Al and Cr are needed in relatively high quantities to protect the part. The primary function of Al, though, is to strengthen the alloy by forming . While being a solid-solution strengthener, Co also alters the solvus temperature and aids in expanding the temperature window for heat treatments. Co also reduces the stacking fault energy (SFE) of the alloy, which improves creep resistance and makes dislocation cross slip around particles more difficult. 1,5 Ti, Ta, Nb, W, Re, and Mo are all used as solid-solution strengtheners. Ta, Ti, and Nb strengthen the precipitates by enhancing the anti-phase boundary (APB) energy, while W, Re, and Mo strengthen the matrix. These elements, along with Cr, also react with carbon to form the various carbides present in superalloys. For various reasons usually stemming from corrosion and oxidation resistance, recent trends in DS and single crystal castings favor using more Ta, W, and Re for strengthening, as opposed to Mo and V, which were used in older alloy designs. Mo and V lower oxidation resistance while Re and W show synergistic effects in strengthening the / interface. Ta strengthens , and has been shown to reduce freckle formation, 6 as will be discussed later. The concentration of these refractory elements must be controlled, however, as they increase the density, cost, and likelihood of forming deleterious topologically close-packed (TCP) phases.

PAGE 23

9 C, B, Zr, and Hf are added to strengthen and stabilize grain boundaries. C reacts with Hf, Zr, and other elements of the casting (Ta and Ti) to form MC carbides, which retard grain boundary sliding and alleviate residual stresses in the casting. Hf and Zr are particularly effective in that they react with deleterious impurities, such as S and P, before all other elements. B, likewise, forms borides that also strengthen the grain boundaries. However, due to directional solidification and single crystal casting technologies, the need for grain boundary strengtheners and stabilizers has diminished. 2.1.3. Other Phases in Nickel-Base Superalloys Besides the and phases, there are many other phases that may be present in superalloys. Some, like carbides, are desired under certain conditions. However, the majority are detrimental phases that initiate fast fracture and reduce service life. The C in superalloys tends to react with one or more alloying elements in solution to form carbides. Carbides serve as excellent grain-boundary strengtheners improving creep rupture life dramatically. They also strengthen the rest of the casting and act as traps for harmful elements such as P and S. Carbides formed in superalloys can be classified as either primary or secondary. Primary carbides are those that form from the melt and are almost always the MC phase (eg. TaC, TiC) with an NaCl structure. These carbides are randomly oriented with the matrix and are located both interand intragranularly. MC carbides tend to be extremely stable such that they typically survive various heat treatment schedules. Their morphology can be either script, cuboidal, or blocky. 2 Secondary carbides are those that form from solid solution, usually during lower-temperature heat treatments or during service. Their constitution and structure depend on

PAGE 24

10 overall alloy composition and include M 23 C 6 M 7 C 3 and M 6 C carbides. These carbides usually appear at grain boundaries and can be created and altered in reactions between MC carbides and the matrix. M 7 C 3 carbides are trigonal, form in low Cr concentrations, and are unstable. They usually decompose into M 23 C 6 carbides, which have complex cubic structures and usually form in alloys with higher Cr concentrations. M 23 C 6 are created either as a product of M 7 C 3 decomposition, or as a reaction of MC and The M 6 C carbides are similar to that of the M 23 C 6 in structure and morphology, except that they tend to form at higher levels of W and Mo. M 6 C appear to be more stable at higher temperatures than M 23 C 6 1 Borides, like carbides, act as excellent grain boundary strengtheners. They are usually hard, blocky particles of tetragonal M 3 B 2 and form at small concentrations of B (<0.03 wt %). However, single crystal alloys usually lack carbon and boron due to the absence of grain boundaries. The philosophy behind most single crystal processing is to increase the amount of for strengthening. However, as will be discussed later, carbon may have other beneficial effects on single crystal (SX) and DS superalloy castings. In alloys with high concentrations of Nb, the phase will form. This is a coherent body-centered tetragonal (bct) phase of Ni 3 Nb. This phase strengthens the alloy primarily because of its high lattice mismatch strains with the matrix. Similar to , will provide high strength at low temperatures, but unlike , it is a metastable phase which becomes unstable above 650C. Furthermore, one must ensure a carefully controlled heat treatment to prevent the formation of the stable -phase of Ni 3 Nb. This is an incoherent orthorhombic phase that can strengthen grain boundaries in small amounts, but in large amounts (more volume percent than ) is detrimental to the alloy.

PAGE 25

11 The hexagonal -phase (D0 24 ) of Ni 3 (Ti,Ta) is often found in superalloys with high (Ti+Ta)/Al ratios, especially after considerable service life. The phase can also form during solidification after the + eutectic 7 and can appear in a pearlite structure of alternate lamellae with or as a ribbon-like Widmansttten pattern. 2 In either case, the phase grows slowly, but can eventually grow larger than the particles. Usually occurring near grain boundaries, this incoherent brittle phase will reduce the ductility and rupture strength of the alloy. There has been limited investigation into formation. This phase has been found to be a component in the scale that forms on castings of PWA 1483. 7 This was found to contain a percentage of Si that dissolved from the mold. It was also found that the precipitation of MC carbides reduced formation due to the reduction in the remaining Ti and Ta contents in the alloy. 7 There are a myriad of TCP phases that can be present in a casting that are detrimental to the performance of the part. These phases are complex in nature, incoherent, and brittle resulting in a lowered rupture life, ductility, and creep resistance. These include Laves, P, and G phase. 1,2,8 The phase is rhombohedral and forms at high temperatures in alloys with high W and Mo. The tetragonal phase, which also appears in stainless steels, forms elongated, irregular globules with a structure similar to M 6 C carbides and appears during service at elevated temperatures. Laves, P, and G phases are all complex, irregularly-shaped globules, which are strongly promoted by silicon and greatly reduce fatigue and creep resistance. 8,9 With all TCP phases, including only a slight change in local composition is is needed to force the composition past the stability threshold.

PAGE 26

12 Besides these TCP phases, a variety of nitrides, oxides, hydrides, and other ceramic impurities can form and thus reduce the desired properties not only by acting as crack initiation sites, but also by removing alloying elements from solid solution. Therefore, it is imperative to control the chemistry of each casting and protect the environment during casting. 2.2. Processing of Superalloys Nickel-base superalloys need to be processed under special conditions. They must be cast at high temperatures due to their high melting temperature (T M ), but must also be cast in a vacuum to prevent excessive oxidation and contamination. Those alloys with high concentrations of solid-solution strengtheners, especially formers, cannot be hot or cold worked and must therefore be cast. The casting or forging of nickel-base superalloy parts is a multistage process in controlled environments. First, the raw materials and reused scrap are vacuum induction melted (VIM). VIM involves melting of the material in a vacuum to prevent the formation of slag and control the reactivity of the elements. This melting step produces ingots with coarse, non-uniform grain sizes that must be refined for forged components. Either vacuum arc remelting (VAR) or electroslag remelting (ESR) are performed to refine the ingot. Both of these processes remelt the original VIM ingot to improve the microstructure and purify the alloy. 2 ESR differs from VAR in that it can be carried out in air. For castings, though, only the VIM process is necessary. 2.2.1. Investment Casting Due to the complex shapes often required for blades and vanes, many superalloy components are best produced by lost-wax investment casting (see Figure 3). 10 Lost wax investment casting involves the creation of a replica made of wax that must be designed

PAGE 27

13 to compensate for shrinkage. Any desired holes or cavities are produced by ceramic cores inserted into the wax pattern. The wax pattern is then assembled with other patterns to form a tree. The tree is then dipped repeatedly into slurry baths while rotating to insure complete coverage and uniform thickness. When a sufficient slurry coating has been applied, the mold is covered with coarse stucco particles. This way, fine and smooth ceramic particles conform to the wax shape inside of the mold, whereas the stucco provides strength and prevents cracking and excess runoff of the wet slurry. This process is repeated several times until the mold is sufficiently thick. The ceramic mold is then dried and placed into an oven for the dewaxing process. Here, the mold is heated to remove the wax and leave the hollow ceramic shell. It is important to control the rate of temperature change to prevent cracking in the mold. After the dewaxing process, the mold is sintered. Figure 3 The investment casting process

PAGE 28

14 To make the actual casting, the ingots of superalloy are melted and poured into a preheated ceramic shell. After cooling, the ceramic shell is removed and usually destroyed, and the metal parts are machined off the tree. However, for DS and SX castings, considerably more care (and cost) must be added to the investment casting process. 2.2.2. Directional Solidification (DS) The process of directional solidification (DS) helped to revolutionize the development of modern nickel-base superalloys. DS alloys have higher creep rupture life in addition to improved fatigue resistance and ductility. This is due to the vast reduction in grain boundary surface area, which acts as an initiation site for creep failure. Fatigue resistance is maximized due to the <100> preferred growth direction of the DS grains (the <100> orientation has the lowest modulus, as opposed to other growth directions). 2,5 The DS process for superalloy components involves an investment procedure but, in this case, a special furnace is used. The ceramic mold must be heated to above the liquidus temperature of the alloy in the furnace. The bottom of the mold is placed onto a chill plate of Cu or some other conductive metal that is water-cooled. After filling the shell with molten alloy, it is slowly withdrawn from the furnace, which is also above the liquidus temperature. This allows the original <100> grains to continuously grow throughout the casting (see Figure 4). 2

PAGE 29

15 Figure 4 Schematics of (a) a DS furnace and (b) a single crystal furnace It is important to control the processing conditions of DS components in order to prevent a myriad of defects. Upon contact with the water-cooled chill, the molten alloy immediately solidifies and a thermal gradient (G) is established. This gradient is controlled and influenced by the temperature of the chill, the size of the casting, the temperature of the upper and lower chambers of the furnace, and other variables. The casting cannot be withdrawn until the solid dendrites have reached the area of the casting at the liquidus temperature. Because of the low thermal conductivity of Ni, the effect of the chill is small and the majority of heat must be dissipated by other means: conduction through the casting, conduction from the ceramic mold, and radiation emanating from the mold. At steady-state conditions, the thermal gradient is controlled by the size of the casting and the baffle. To ensure that the growth of columnar grains does not break down, G and the withdrawal velocity (R) must be carefully controlled. Defects frequently occur in these castings and include grain boundary cracking, bulging, non-metallic inclusions, and recrystallization. Grain boundary cracking is caused by a hoop stress from ceramic cores during cooling and can be prevented by

PAGE 30

16 adding Hf to the alloy. Bulging is caused by creep of the mold shell, causing flat surfaces in the casting to bulge out. 11 This distorts the ceramic mold, and inclusions are formed from parts of the mold that break off when the molten alloy is poured into the mold. Both of these defects can be resolved with better mold design and construction. However, the defects of chief concern are those involving the nucleation of equiaxed grains. Any recrystallization or nucleation of new grains defeats the purpose of the DS process and results in the part being scrapped. Freckles, spurious grains, and columnar-to-equiaxed transitions are a few of the defects that can occur. These will be discussed in Section 2.3. A further advancement to DS casting was the development of SX castings with Ni-based superalloys. The same DS process is used, except that a seed or grain selector is added. Often helical in shape, the grain selector allows the <100> direction to dominate and eventually be the only grain growing into the component. An alternate method involves using a previously cast SX piece as a seed. The seed is allowed to partially melt and then the new solidification follows the growth direction of the seed. The advantage of the seed is that it allows any direction and orientation about that direction to be grown, though this process must be more rigidly controlled. The defects involved with SX castings are the same as those for DS except for grain boundary cracking. The absence of grain boundaries allows the removal of C, B, Hf, and Zr from alloy design. This increases the incipient melting temperature of the alloy and is one of the reasons for enhanced creep rupture life. 2.2.3. Microstructural Features The microstructure of these DS alloys, and most solidified structures, stems from supercooling issues. As the liquid cools, the solidification front, which is initially a flat

PAGE 31

17 surface, releases its latent heat of fusion at the interface. This increases the temperature on the liquid side of the interface, which creates a temperature inversion and results in cellular growth. This thermal supercooling, which exists in other solidification processes, is not as valid for DS as constitutional supercooling. 12 Here, the alloying elements partitionthe solid that forms has a different composition from that of the original liquid. The liquid at the interface contains excess solute, which shifts the equilibrium liquidus temperature. If above the actual liquid temperature, it causes the solidification front to break down forming perturbations that lead to the formation of cells and dendrites. Eventually, more solute is built up in the lateral directions, forming perturbations on the advancing cells. These secondary perturbations lead to the formation of both secondary and tertiary dendrite arms on the already existing primary dendrite arms. 12,13 The spacing of the primary dendrite arms () is related to the temperature gradient (G) and growth velocity (R) in the casting: mnRGK (1) K is a constant specific to the alloy, and m 0.25 and n -0.5 for many nickel-base alloys. 14-16 In nickel-base superalloys, Cr, W, Re, and Mo segregate to the dendrite core while Ta, Ti, and Al remain in the interdendritic liquid. 2 Thus, the dendrite arms are and MC carbides begin to precipitate as cooling proceeds (usually 5-30C below the liquidus 2,17,18 ). The solidification concludes with the + eutectic that occurs over a small, 5C window. 19 The precipitates in the regions spread throughout the dendrites and large amounts of + eutectic remains where the interdendritic fluid solidified last.

PAGE 32

18 The dendritic structure of the castings is not always perfectly aligned with respect to the bottom of the chill. This is due to the development of curvature in the liquidus isotherm that occurs when the conduction of heat through the casting is much less than the conduction and radiation of heat out of the mold walls. The opposite effect occurs in VAR ingots in that there is more heat on the outside than inside, causing a convex upward liquidus isotherm. Since the dendrites tend to grow normal to the solidification front (or antiparallel to the direction of heat flux), the dendrites will bend somewhat when the isotherm curves. This is the cause of splaying of dendrites and can cause the part to be rejected if the deviation from <100> is severe enough (usually around 10 from other subgrains). As described above, alloying elements partition and microsegregate during dendritic solidification. Another process called macrosegregation occurs when solute is transported over large distances. Macrosegregation is a result of convection caused by density differences in the casting. A form of macrosegregation, called inverse segregation, can occur in DS castings from the back flow of solute-enriched interdendritic liquid through the interdendritic channels in the casting as it cools. 20 In Al-Cu alloys, the Cu solute can actually emanate from the chill surface and alongside the casting. 21-22 This backflow of interdendritic fluid is controlled by the viscosity and permeability of the fluid. As the primary dendrite arm spacing (PDAS) increases, the permeability through the dendritic mushy zone will increase. The PDAS is inversely proportional to the thermal gradient and withdrawal rate, so segregation will be favored at low G and R.

PAGE 33

19 2.3. Freckles Freckle casting defects have been discovered in many kinds of various DS alloys besides superalloys. They were found originally in iron-base superalloys and have been seen in steels, solders, and various other alloys near the eutectic composition. 23 They have been observed in both castings and ingots. 23-24 In all these alloys, the freckle defects were found to be long chains of equiaxed grains parallel to the primary dendrite arms whose composition was close to that of the interdendritic regions. They have been found on the surface of castings and near the midradius of ingots. The number of, and even presence of, freckle chains has been found to increase as the diameter of the casting increases. Consistent with this observation, there is a minimum diameter or size at which freckle chains will not formeven if conditions are ideal for freckling. Freckle chains begin to form at a certain length above the chill in DS and SX castings. The chains commonly converge as one moves up the casting, but divergence does not occurs. It was observed by Giamei that the freckle chains are as deep as they are wideapproximately 1-2 mm. 23 Freckle grains are observed to be replete with misoriented grains, eutectic, carbides, and porosity. Unfortunately, there is no quantitative or stereological data to confirm this and most reports are based on qualitative observations only. In superalloys such as CMSX4, Waspaloy, Mar-M 200, and PWA 1480, the freckle grains are purportedly enriched in those elements that partition to the liquid during solidification, such as Al, Ta, Nb, and Ti. There is also a lack of those elements that segregate to the dendrite core in freckles (Co, Cr, and other refractory elements). In fact, freckling often occurs in those alloys with high concentrations of W and Re. However,

PAGE 34

20 additions of Ta and sometimes Nb have been shown to reduce the propensity for freckle formation in these castings. Besides castings, freckles have been found in VAR processed ingots of Waspaloy. 24 The freckles in this case occur at the midradius rather than the surface. This is one of the largest problems with freckles in ingots in that they cannot be detected, even by ultrasound, unless they are cut open. Due to the large size of these ingots, this is an undesirable solution. Characterization of these ingot freckles resulted in large quantities of script carbides and -phase with a Widmansttten morphology. This explains the increased hardness in the freckle areas and the propensity for fatigue and fracture initiation sites. Even though VAR ingots eventually are forged into polycrystalline parts, the presence of these ingot freckles greatly reduces uniformity in the ingot and can be a source of fracture during processing. 2.3.1. Theories of Freckle Formation From the earliest research into freckle formation, it was immediately noticed that freckle chains would always form at a distance away from the chill, where the thermal gradient and growth velocity of the mushy zone had decreased. 23 In addition, the compositional shift towards the eutectic composition in the freckle chains pointed to a clue of the origin of these defects. Copley was the first to see freckle chains actually forming by using a solution of water and NH 4 Cl near eutectic composition. 25 The solution was directionally solidified on a chill cooled with liquid nitrogen. As was soon discovered, interdendritic liquid from inside the mushy zone tended to jet upward from the mush into the liquid layer. The liquid would remain immiscible due partly to its cooler temperature. Along the way, these jets would erode dendrites and break them

PAGE 35

21 apart where they would redissolve into the liquid or act as nucleation sites for equiaxed grains. It was also shown that the jets emanate from the highest part of the mushy zone. In castings, the mushy zone is concave upward due to the surface being cooler than the rest of the interior. Thus, the freckle trails appear on the surface. In VAR ingots where the mushy zone is convex upward, the jet plumes emanate to the casting interior. The jet trails are also always parallel to the direction of gravity. Figure 5 show vectors of channel flow calculated for a solder alloy. 26 Solute from the mushy zone flows up, forcing the liquid above the mush down into it. If a casting is directionally solidified at an angle so that the mushy zone is tipped (since primary dendrites always grow opposite to the direction of heat flux), the jet plumes will still grow parallel to gravity. 25 Figure 5 Isosurfaces of volume fraction and stream traces emerging from a channel The relation of this H 2 O-NH 4 Cl system to superalloy solidification was made in later research. The freckle grains tended to have an increase in both Al and Ti, but lacked

PAGE 36

22 the heavier W and Re, which were found in the dendrite core. 6 Therefore, it was found that in nickel-base superalloys, excess segregation of the heavier refractory elements into the dendrite core and lighter formers into the interdendritic fluid create a density inversion. At a critical G and R, a pressure increase in the interdendritic fluid forces it out at the highest point of the mushy zone as small jet plume. During this process, named thermosolutal convection, the plumes break off dendrite arms as they shoot upward. 6,27 The interdendritic fluid does not readily mix with the rest of the liquid due to its difference in density and temperature between the two liquids. This interdendritic plume eventually solidifies as a chain of small, equiaxed grains where the nuclei are purportedly the dendrite arms removed by the plumes. Figure 6 shows a model of these plumes and how they remain immiscible in the liquid. 26 Figure 6 Isosurfaces of solute concentration, showing solute enrichment within channels and the emerging plumes. Knowing the cause, the question was then figuring out how to prevent the formation of freckles. Copley proposed controlling the R and G. At high R, the amount

PAGE 37

23 of heat released into the mushy zone must be less than what is conducted through the solid to maintain the flow of heat out of the casting. Thus, the growth rate can be defined as HGKRST (2) where K T and G S are the conductivity and temperature gradient of the solid, respectively, and H is the heat of fusion. 25 Under equilibrium, freckle formation is controlled by the difference in temperature between the liquidus and solidus (T), the minimum time necessary to produce freckles (t), and the gradient, G. Therefore, the growth rate can be defined as GtTR1 (3) As Copley discovered, above a critical thermal gradient (G*) the mushy zone becomes too small to accommodate the convective currents that would produce freckles (see Figure 7). 25 Other solutions, though, involved altering the alloy chemistry. The major cause of freckle formation stems from substantial segregation of the heavier elements to the dendrite core. Reducing the concentration of W and/or Re would reduce the propensity to form freckles. However, in doing so the strengthening effects of these elements are sacrificed. In castings with high Ta contents (or Ta to W + Re ratios), few, if any, freckle chains appear. Ta partitions to the interdendritic liquid, which helps to increase its density and hinder thermosolutal convection. 6,28

PAGE 38

24 Figure 7 The effect of growth velocity (R) and thermal gradient (G) on freckle formation. Most SX alloys contain no C since there are no grain boundaries to strengthen. However, small additions of C have been shown to hinder freckle formation by altering the segregation behavior of certain elements. 29 It was shown that MC carbides, while containing no W, reduced it from segregating to the dendrite core. More likely, though, the carbides that formed increased the permeability in the interdendritic fluid by increasing the drag force and, due to their high density, stave off the density inversion that occurs in the mushy zone. 29 2.3.2. Modeling While a change to the chemistry of these alloys helps prevent the onset of freckling, this may not be acceptable for all castings. As previously shown, castings with high Ta concentrations have the tendency to form Therefore, returning to the fine-tuning of processing conditions was investigated. Consequently, a large amount of effort has been

PAGE 39

25 placed into modeling the situation of thermosolutal convection and what impact various conditions have on the process. This has led to the creation of models and criteria that attempt to describe the conditions that promote freckle formation during solidification. The first mathematical constructions of freckle channel formation were originally developed by Sarazin and Hellawell on a variety of solder alloys. 27 These binary Pb-Sn and Pb-Sb systems were chosen due to their simplicity (rather than the multicomponent superalloys). Their research attempted to use permeability and thermodynamic data to calculate various dimensionless fluid mechanics numbers. Using these numbers, one could determine the conditions necessary for thermosolutal convection to occur. It was discovered that there is a critical Rayleigh number below which freckling will not occur. More recently, research has been slated towards three-dimensional modeling of freckle-forming binary alloys. 26,30 The Rayleigh number (Ra) is a dimensionless number used for convection and is officially defined as the driving force to flow over the resistance to flow or TDhTgRa3 (4) where g is gravitational acceleration, the coefficient of thermal expansion, the dynamic viscosity, D T the thermal diffusivity, and h some length or dimension. 31 Above a critical Rayleigh number (Ra*), turbulent flow will commence. There have been differing opinions of how to define and apply the Rayleigh criterion to freckle formationusing mathematical criteria 25,27-28,32 or conservation equations. 33-35 Auburtin et al. believed conservation equations cannot be practically applied to industry and derived their Ra criterion from mathematical standards. 36 Their version,

PAGE 40

26 TDhdzdgRa4 (5) uses the density inversion (d/dz) instead of a temperature difference. One of the difficult parts in applying this criterion to superalloy castings is how to define the characteristic length, h. Here, Auburtin defines it as the primary dendrite arm spacing. Beckermann et al., however, define their Rayleigh criterion through conservation equations: 37 TDhKgRa 0 (6) Beckermann uses the mean permeability ( K ) in his equation and defines the characteristic length as the mushy zone height. It is believed that this approach is more applicable to a large variety of casting conditions. Nevertheless, both authors have determined how to use Rayleigh criteria to predict freckling: Determine the local Rayleigh number (Ra MAX ) from the local G and R either through thermocouple measurements or analyzing primary dendrite arm spacings. Determine the critical Rayleigh number (Ra*) below which freckling will not occur. Ensure that Ra MAX < Ra*. It should be noted that Ra* will only predict at what point freckling will not occur; it cannot predict freckle formation. 36,37 Even if the local Ra is higher than Ra*, freckles may not form due to small casting cross sections, complex shapes, or small vertical heights. Likewise, Ra* will not predict where or how many freckle chains will form. Both authors also discovered that tilting the casting significantly decreases Ra*, increasing the likelihood of freckle formation. 25,36,37

PAGE 41

27 Schneider et al. continued the work of Beckermann et al. by using the conservation equations with several computer simulators previously used by the steel industry. 35 The simulations included tracking the concentrations of various elements in the convective channels in castings with various thermal gradients and growth velocities, which were combined here as cooling rate: RGT (7) Most importantly, a critical PDAS was found above which the system becomes unstable and convection occurs. For CMSX2, the critical primary dendrite arm spacing was 320m. 35 Pollock and Murphy also correlated the formation of convective jets to the columnar-to-equiaxed transition (CET) in DS and SX alloys. This transition is a breakdown of the single crystal or columnar grains caused by a blockage of columnar growth brought on by contact with the melt walls, remelting, and fragmentation of dendrite arms, or some other mechanism. 28 Freckling and CET occur under similar conditions. In both, it was found that at low G, secondary and tertiary dendrite arms remelt, detach, and fragment. These dendrite fragments serve as nucleation sites for new grains. Furthermore, the detached arms also impede the advancing solidification front, reducing R as well. This complicates the situation but allows a total transition from single or columnar to small, equiaxed grains.

PAGE 42

CHAPTER 3 METHODS AND MATERIALS 3.1. Materials The materials used for this procedure were two nickel-base superalloy castings directionally solidified from the Howmet Corporation in Hampton, VA. Due to proprietary concerns, the composition and processing variables were not disclosed. It was therefore a challenge to try to compare compositions of the materials and to determine if any elements had segregated. Energy dispersive spectroscopy (EDS), which will be discussed later, was used to ascertain the chemistries, and compare them to similar industrial compositions from literature. 3.2. Optical Microscopy The purpose of the optical microscopy was to examine the microstructural features, and to measure the PDAS. The main goal was to try to discover if there was more porosity, eutectic, and other features around the freckled region than in the rest of the casting. Furthermore, the amounts microstructural features inside and near the freckle grains were contrasted with the rest of the casting. The PDAS was found using the optical microscope at low magnifications (usually 50X) on specimens cut perpendicular to the growth direction of the primary dendrite. The PDAS was then measured from one dendrite core to the nearest diagonal dendrite core. This process was repeated multiple times to reduce the variance in the results. 28

PAGE 43

29 3.3. Preparation of Samples The samples for optical microscopy were cut using a wire electrical discharge machine (EDM). The samples were ground in sequence with 120, 240, 320, 400, 600, 800, and 1200-grit SiC polishing paper. The fine polishing was performed using a 1.0 m alumina suspension. The surface features were brought out by using an etchant of 80 mL hydrochloric acid, 2 mL nitric acid, 11 mL water, and 16g FeCl. 38 These samples were then viewed through the Leco Neophot 21 optical microscope. The process of cutting the bars was difficult, especially for the DS-2 samples. A wire EDM was chosen over a cut-off wheel because when the cutoff blade was deep into the sample, the sample would move causing the blade to shatter. Similar problems were encountered with the wire EDM. As the wire cut deeper into the sample, the cut metal around the wire would close in on itself, pinching the wire off (see Figure 8). This may have been due to residual stresses in both castings. Therefore, careful, slow, and repetitious EDM cuts had to be made for the each desired specimen. Figure 8 Illustration showing the castings pinching back on itself after being cut by the wire EDM

PAGE 44

30 Another complication was that the only regions of interest in these bulky samples was near the surface containing the freckle chains and extended only a few millimeters into the surface. Therefore, carefully designed cuts were made in the freckled regions from and were less than 10 mm into the sample. In order to get the best picture to characterize the samples, cuts were made in three directions (see Figure 9): 1. Parallel to the growth direction and parallel to the casting surface 2. Parallel to the growth direction and perpendicular to the casting surface 3. Perpendicular to the both growth direction and the casting surface. In reality, however, Cut 1 is not really a cut, but rather the surface polished and etched to reveal the grain structure. Figure 9 Illustration of the directions cut in relation to the primary dendrite arms (not to scale). 3.4. Stereology In order to quantify the microstructural features in the alloys, stereological methods were used. Stereology involves using information from a two dimensional image to find three-dimensional properties. This includes using point, line, and area probes and

PAGE 45

31 running the data through expected value theorems to get geometric properties of the structure. 39 For example, from a count of points, one can infer the volume fraction of a microstructural feature. Some of the probing techniques are shown in Figure 10. These probes are then used to find properties of the microstructure (see Table 1). This project used these techniques to determine The quantitative volume percentages of microstructural features inside the freckles The quantitative volume percentages of microstructural features in freckled regions (so as to include the grain boundary areas) The microstructural constituents of interest are eutectic, porosity, carbides, and -phase. These were contrasted with the measurements found in the rest of the casting. The stereological measurements were taken from optical photographs as well as scanning electron microscope (SEM) micrographs (see below). Figure 10 Manual counting methods used for stereology

PAGE 46

32 Table 1 Common stereological formulas and relations to microstructural features Probe CountRelationProperty FoundPoint Count = VVVolume FractionLine Intercept Count = 0.5 SVSurface Area DensityArea Point Count = 0.5 LVLength DensityFeature Count = MV/2Total CurvatureArea Tangent Count = MV/Total Curvature It is important to ensure that the probe set used is isotropic, uniform, and random. 39 In addition, the correct number of probes must be used to ensure proper data collection and filter out noise. Part of the problem for this project was that the target for sampling was in and near the freckle grains. This reduces the chance of a random set of fields being sampled and can throw off the actual volume percent of the desired constituents. Therefore, for this project, a set of 20 fields, each with 324 point grids were chosen to take volume fraction measurements in random spots inside the freckled area, and in areas in which freckles existed. These were contrasted with a sampling amount for areas in the casting, near the surface, that showed no freckle formations. There was a difference noted between the probes inside the freckle and those outside for those fields that encompassed freckled and non-freckled regions. Furthermore, the results from the position of the freckle chains were compared. If no trend can be associated with the position, then any cross-section will be random and uniform. Otherwise, an extra interpretive factor must be added. 3.5. SEM and TEM Samples were prepared for SEM and transmission electron microscope (TEM) similar to those fashioned for optical microscopythey were cut using the EDM. The samples were also polished with 120, 240, 320, 400, 600, 800, and 1200-grit SiC polishing paper. Fine polishing was done with 1.0 m alumina suspension. An etching

PAGE 47

33 solution from Van der Voort was used contained 80 mL hydrochloric acid, 2 mL nitric acid, 11 mL water, and 16g FeCl. 38 These samples were preheated at 180F in hot water before etching. The SEM used was a JEOL SEM 6400 equipped with Oxford Link ISIS EDS and Oxford OPAL electron backscattered secondary electron detector (EBSD). The samples prepared for TEM were ground and polished in a similar manner to those for SEM and optical microscopy. However, these thin slices were ground to approximately 100 m before polishing mechanically. The thin polished specimens were pressed to form 3 mm disks. These disks were then electropolished in a solution of 50 mL perchloric acid, 100 mL of ethylene glycol, and 350 mL of methanol at -20C, 10 mV, and 1 to 5 A until a small hole appeared. The area around the hole needed to be sufficiently thin to be electron transparent. If not, the specimens were ion milled at a low angle (~12) to thin the regions near the hole. The TEMs used were a JEOL TEM200CX and a Philips TEM420 equipped with an EDAX EDS system. 3.6. Electron Backscattered Diffraction As previously mentioned, the JEOL SEM 6400 is fitted with an Oxford OPAL system. Similar to convergent beam diffraction patterns, the EBSD technique collects data to create a Kikuchi map, which contains crystallographic data along with information on orientation relationships and grain misorientation. This is done when the detector, which is a phosphor-coated screen, collects backscattered electrons and forms a pattern of lines whose configurations are determined by the orientation and structure of the crystal. The samples for EBSD are 10 mm wide and less than 1 mm thick and are placed at 70 to the detector. The accelerating voltage used was 20 kV.

PAGE 48

34 3.7. Nanoindentation A Hysitron Triboindenter nanoindenter was used to determine the hardness and moduli of different microconstituents in the castings. The Triboindenter, which is controlled by a Triboscan transducer, works by affixing a nanotip to an electrostatically held plate between two other plates (see Figure 9). The outer plates are fixed to electrodes where a potential is created by adding an AC current 180 out of phase. As the center plate moves, the potential changes allowing for displacement measurement. Force is measured by a DC current applied to the center plate. Figure 11 Illustration of indent head mechanism in the Triboindenter Using the nanoindenter, nanohardness measurements and modulus mapping could be performed. Nanoindentation involves forcing the tip into the surface with a known load. The depth is monitored and force is measured as the tip is released to give hardness and relaxed modulus measurements. Multiple indents were performed on various

PAGE 49

35 constituents in both castings with a Berkovich indenter at loads ranging from 500 to 5000 N. This range of loads was used to counteract problems associated at both extremes: small loads can produce results with increased scatter due to the greater impact of noise, and large loads can create indents too large to accurately characterize small phases. Modulus mapping was performed by oscillating the tip to collect force and phase measurements while also gathering topographical data.

PAGE 50

CHAPTER 4 RESULTS The results acquired are divided into four parts. The first discusses the SEM and optical pictures showing and describing the microstructure and morphology of the superalloy samples. The next section shows the stereological data of the microstructure. The third section details the grain orientations and results found from the EBSD data. The last section discusses data found from the nanoindenter. 4.1. Microstructural Characterization There were two large samples obtained from the Howmet Corporation. One, which will be herein designated DS-1, is a section off of an airfoil approximately 300 mm tall, 70 mm wide, and up to 25 mm thick (see Figure 12). The second sample set is from a larger 1.2 m long industrial gas turbine blade, 300 mm wide, and 120 mm thick, with two hollow cavities running through the middle (see Figure 13). This specimen was cut from the top of the DS casting. Close examination of Figure 13 reveals freckle chains on the side of the castingssome of these are pointed out in the figure. The primary difference between the two castings is that the DS-1 casting had been etched to reveal freckle grains, whereas the DS-2 castings had notthey still had their original, undisturbed surface. While this was intriguing to study the possible surface effects of the second casting, it also presented a problem when using the wire EDM to cut them. Specifically, the virgin surface was not sufficiently conductive to be spark cut due to the remnant oxide/ceramic that remained after casting. This was overcome by scratching the surface to be cut and having the wire cut at precisely that location. 36

PAGE 51

37 Figure 12 Photograph of the DS-1 castings and a labeled freckle chain. Figure 13 Photograph of the DS-2 castings with some visible freckle chains labeled.

PAGE 52

38 4.1.1. Compositions The overall compositions of the DS-1 and DS-2 castings were not revealed by Howmet due to their proprietary nature. In an effort to determine both global and local compositional information, EDS spectra were obtained from the samples at low magnifications in an area, not point, mode in order to accommodate as much area as possible. These scans included the freckled regions, and were large enough to incorporate all the phases and constituents to prevent biasing the results from the segregation of elements. Multiple scans were run in order to minimize deviations. Figure 14 shows an EDS spectrum from the DS-2 casting. Spectra from the DS-1 casting were similar to the DS-2 and had the same elements present. Figure 14 EDS spectrum of DS-2 alloy at 15 kV. Using the atomic number, absorption, and fluorescence (or ZAF) correction software on the EDS system, a semi-quantitative analysis was performed on the qualitative data in the EDS spectra. This technique uses matrices, summations, and formula integrations to estimate weight and atomic percent of a sample based on peak

PAGE 53

39 height, elements found, etc. 40 The weight percentages for both DS-1 and DS-2 alloys, excluding carbon, are shown in Table 2. The resulting analysis suggested that the DS-2 casting is similar to a Rene N4 alloy minus Nb and Hf, and DS-1 seemed comparable to PWA 1483. 2 Table 2 Estimated nominal compositions of DS-1 and DS-2 alloy. These compositions are compared to that of Rene N4 and PWA 1483. CAlTiCrCoNiMoTaWNbHfDS-1 (Small Airfoil)xxxxx2.96.112.89.359.11.83.24.8PWA 14830.073.6412.8960.81.943.8DS-2 (Large Pieces)xxxxx2.93.59.97.8641.34.26.3Rene N40.064.23.59.87.5621.54.860.50.15 In addition, semi-quantitative EDS analysis was performed at a higher magnification to contrast compositions in the freckled regions with that of the rest of the casting (or matrix). The results, without carbon, are shown in Figure 15 for DS-1 and Figure 16 for DS-2. For both alloys, the freckled regions were higher in Al, Ti, and Ta, and lower in Cr, Co, W, and Mo. It should be noted that ZAF semi-quantitative analysis is susceptible to error during the EDS collection, and from the nature of the ZAF correction itself.

PAGE 54

40 02468101214Weight % AlTiCrCoMoTaWCompositional Differences for DS-1 Freckle Matrix Figure 15 Comparison of freckle chain and matrix compositions in weight percent from DS-1 alloy without Ni. 02468101214Weight % AlTiCrCoMoTaWCompositional Differences in DS-2 Casting Freckle Matrix Figure 16 Comparison of freckle chain and matrix compositions in weight percent from DS-2 alloy without Ni. The freckles in the DS-1 casting about the same Al, Ti, Cr, and Mo amounts as DS-2, but the amounts of Co, Ta, and W and were different. The Co and W differences between the freckle grains of the two castings correspond to the compositional differences between the two alloys as a whole. However, the freckle Ta differences are too great to associate with composition and likely indicates substantial Ta segregation in DS-2 to the interdendritic during solidification.

PAGE 55

41 4.1.2. Surface of the Castings The freckle chains were observed visually on the surface macroscopically, and were studied microscopically as well. The dendritic structure was revealed in Figure 17 for DS-1 and in Figure 18 for DS-2. The primary (and secondary) dendrites are all arranged in a <100> type direction. Here, as shown in Figure 17, the dendritic structure of the freckle grains is more randomly oriented, and no direct relation could be determined by visual inspection alone. The orientations were identified later using EBSD. Figure 17 SEM (Scanning Electron Micrograph) of etched DS-1 surface showing dendritic structure. Freckle grains run diagonal from top left to bottom right (35X).

PAGE 56

42 Figure 18 SEM of etched surface of DS-2 casting revealing dendritic structureof freckle chain. The chain runs left to right in the middle of the picture (50X). As noted above, the surface of the DS-2 castings was initially unaltered after it was broken out of the mold. This virgin surface (see Figure 19) exhibits charging non-metallic phases approximately 100 to 400 m in size. Further inspection of this phase (see Figure 20) in the backscattered mode shows the charging particles are of low atomic weight, and also revealed some smaller, high atomic weight phases (see Figure 21). The X-ray map (see Figure 22), reveals the larger charging phase to be alumina while the smaller, brighter phase is likely (Ta,Ti)C carbide. Traces of ZrO 2 were also found on the surface, most likely left over from the ceramics mold. It was discovered that the overall composition of the surface differed in the freckled regions and in the rest of the casting (here called matrix). From Figure 23, one can see these compositional differences. There may have been a larger percentage of alumina in the freckled regions, or it might just reflect the results from Figure 16.

PAGE 57

43 Figure 19 SEM of unaltered DS-2 alloy surface (100X). Figure 20 SEM of surface of DS-2 casting showing charging alumina phase (650X).

PAGE 58

44 Figure 21 BSE (Backscattered Electron) SEM of Figure 20. The brighter areas have elements with higher atomic numbers (high Z). Figure 22 X-Ray map of Figure 21 revealing the Al enrichment in the charging phase and the Ta and Ti enrichment in the high Z phase.

PAGE 59

45 051015202530COAlTiCrCoNiTaWeight Percent Matrix Freckle Figure 23 Comparison of EDS data from matrix and freckle regions of unaltered surface of DS-2 alloy. 4.1.3. Eutectic Colonies and Porosity Large amounts of eutectic were found in and near the freckle grain boundaries. Qualitatively, the volume percent seemed larger at and around the freckles than the rest of the casting. Quantitative data describing these differences are in Section 4.2. The microstructure of the eutectic colonies in the current castings and in other cast Ni-base superalloys, when cross-sectioned, appears as a fan of small rods or plates in a pool of . This is shown in Figure 24. The structure can also take up other forms, such as a web (see Figure 25). From other planes, such as the surface cut in Figure 26, the eutectic colony may show a different structure. In the current alloys, the eutectic is believed to be the last liquid to freeze, and thus has the highest content of those elements that partition away from the dendrite core (Al, Ta, Ti). Furthermore, in the DS-2 castings, there was porosity near the eutectic colonies in the freckle chain (as well as elsewhere in the casting). This is shown in the optical micrograph in Figure 27. In the DS-1 alloy, though, porosity is also associated with the hexagonal or Ni 3 (TaTi), phase

PAGE 60

46 at the eutectic/matrix boundaries (see Figure 28). The phase will be discussed in Section 4.1.5. Figure 24 SEM of eutectic colony at a freckle boundary in DS-2 alloy (1000X). Figure 25 SEM of a eutectic colony in DS-1. Note the web-like morphology of the (2000X).

PAGE 61

47 Figure 26 SEM of different section of eutectic in DS-2 alloy at the casting surface. The eutectic is in the middle (2000X). Figure 27 Optical micrograph of cross section of DS-2 casting. In the freckle chain on the left, the dark spots are porosity next to the white eutectic colonies (50X).

PAGE 62

48 Figure 28 SEM of porosity near a colony of at a freckle grain boundary (1000X). 4.1.4. Eutectic at Grain Boundaries One of the more interesting solidification structures is shown in Figure 29. Nearing the boundary between the freckle and the matrix, the appears to be elongated. While certain castings have been shown to contain a film of at grain boundaries, this morphology was unexpected and was found in both alloys (see Figure 30). This elongated structure was also seen near and carbides at the freckle-freckle and freckle-matrix boundaries. The DS grain boundaries, while having some carbides and eutectic at the boundaries, never exhibited the elongated structure of , and often there were no other phases present besides and (see Figure 31). Besides the elongated morphology, blocky was found at the grain boundaries involving freckles when there are no other intergranular constituents, such as eutectic, or carbides, present (see Figure 32). The grain boundary is approximately 5 to 10

PAGE 63

49 Figure 29 SEM of DS-2 alloy near surface showing elongated at the boundaries between three freckles. Figure 30 SEM of DS-1 alloy at boundary of freckle grains and matrix. The "elongates" at the freckle boundary.

PAGE 64

50 Figure 31 SEM of DS boundary in DS-1 (5500X). Figure 32 SEM showing intergranular at a freckle grain boundary in DS-2 (900X).

PAGE 65

51 times larger than the intragranular and is rectangular is shape, rather than cuboidal. In Figure 33, one can see more of the elongated and the blocky at a freckle-matrix boundary. Figure 33 SEM of intergranular between two eutectic colonies at a freckle grain boundary from DS-2. Note the elongated inside the freckle grain at the boundary (750X). 4.1.5. Eta Phase Formation Upon investigation in the SEM, the phase was frequently observed in one of the castings. Figure 34 is an optical image showing some of this phase around a freckle grain in the DS-1 alloy. Figure 35 is an SEM image of the same freckle grain. This phase was found primarily at the boundary between the freckle and the rest of the casting (as shown in Figure 35), but was also found inside the freckle grain itself. Furthermore, as stated previously, there were large eutectic regions near these freckled regions as well.

PAGE 66

52 It is important to distinguish between the phase and the carbides, since in backscattered mode, both appeared lighter than the + matrix (see Figure 36). This distinction was made from EDS and X-ray mapping. The phase, or Ni 3 (Ti,Ta), usually has little solubility for Al or Cr. Likewise, the MC carbides have little solubility for Ni. An X-ray dot map of Figure 36 is shown on Figure 37 and shows regions enriched in Ni and Ti, and depleted in Cr. This is shown again in an area with more (Figure 38), and its X-ray map clearly shows the enrichment in Ni and Ti, as well as low Mo, Al, and Cr amounts (see Figure 39). To differentiate the phase from the primary carbides, an SEM micrograph and corresponding X-ray dot maps of a primary MC carbide is shown in Figures 40 and 41, respectively. This phase is high in Ta and Ti, while being low in Ni and Cr, indicating it is an MC carbide. The compositions of the different phases in Figure 36 are given in Table 3. Figure 34 Optical image of a freckle grain in DS-1 casting showing various phases (50X).

PAGE 67

53 Figure 35 BSE SEM of freckle boundary in the DS-1 showing phase (500X). Figure 36 BSE SEM showing phase, carbides, and TCP phase at a DS-1 freckle boundary (1900X).

PAGE 68

54 Figure 37 X-ray map of Figure 36, showing elements common to phase. Figure 38 SEM of Phase (3300X).

PAGE 69

55 Figure 39 X-ray map of Figure 38, showing elements common to phase. Figure 40 SEM of MC carbide in matrix (10000X)

PAGE 70

56 Figure 41 X-ray map of MC carbide in Figure 40. Table 3 EDS weight percents of elements for phases in Figure 36. Constituen ts Al Ti Cr Co Ni Mo Ta W MC Ca rbide 0.2* 30.0 1.2 0.5* 3.9 4.2 49.3 10.7 phas e 8 5 0.1* 2.7 44. 1.3 3.4 30. 1.5 15.5 Phas e 3.5 12.2 3.6 6.3 63.6 0.9 7.4 2.5 Matrix ( + ) 3.3 5.7 20.3 9.9 51.6 2.8 2.3 4.0 No sThe present in the DS-1 casting appeared to be plate-like in morphology. There also seemed to be more of the aforementioned elongated near the plates at the freckle-matrix border. From TEM micrographs, was found to contain stacking faults and to be located near a carbide (see Figures 42 and 43). EDS analysis in the area of the stacking faults showed mostly Ni and Ti, and little of the other elements. TEM also showed stacking faults in that was near ignificant amount present

PAGE 71

57 Figure 42 BF TEM image of containing stacking faults stacking faults near a carbide (30000X). Figure 43 BF TEM closeup of stacking faults above the carbide from Figure 42 (100000X). 4.1.6. Morphology and Microstructure of Constituents near Surface As previously mentioned, the surfaces of the DS-2 castings were not modified prior to investigation. This meant that the surface of these castings had been in contact with the ceramic mold during the DS process. The actual surface was characterized as

PAGE 72

58 described in Section 4.1.1. In this section, a freckle chain was cross-sectioned (Cut 2) and examined to investigate its microstructure near the surface. There appears to be an 8 m thick layer at the surface that does not contain the cuboidal (see Figure 44). Near the surface, large amounts of eutectic were present, as seen in Figure 45. Figure 46 shows another large eutectic colony with the layer above it. There is also a large pore and carbides present. These colonies are commonly seen in cast superalloys, but rarely of this size. Another large eutectic area is seen in Figure 47. In this case, a large MC carbide is located in the middle of the eutectic colony and the plates appear in a starburst pattern emanating from the carbide. The typical matrix + structure is present between the carbide and the surface. It should be noted that EDS was performed on all the constituents discussed above to properly identify each phase. Figure 44 SEM of a cross section of the DS-2 casting surface at a freckle chain (1400X).

PAGE 73

59 Figure 45 SEM of a large eutectic colony near the external surface of the DS-2 casting (1000X). Figure 46 SEM of large eutectic colony at surface (450X).

PAGE 74

60 Figure 47 SEM of carbide in the center of large eutectic colony near casting surface (1800X). Further analysis of the surface was performed and Figure 48 shows a better resolved image of the surface layer beside a carbide near the surface. In this figure, the layer shows a smearing feature, with the smearing angled towards the carbide. To ensure that these waves were not entirely topological, a backscattered image was taken of Figure 48 (see Figure 49). The smearing, or waves, are lower in atomic number than the surface layer, and have the same contrast as . A closer look at these waves can be seen in Figure 50, where the appears to be smeared in one direction and then back in another. Unfortunately, cross-sections of the bare surface not at 90 did not reproduce this feature. A shallow cut towards the sample at approximately 15 was performed to tie together the images of the virgin surface with the cross-sections. Figure 51 shows a

PAGE 75

61 Figure 48 SEM of carbide near surface (2500X). Figure 49 BSE SEM of Figure 42 (2500X).

PAGE 76

62 Figure 50 SEM of casting surface showing the "smearing" of the + morphology (3500X). backscattered image where alumina is covering carbides and the casting. It appears that on the unpolished surface, bare is seen. ZrO 2 was found on top of Al 2 O 3 in a few regions, along with what appeared to be trace particles of Si. There was smooth surface sparsely seen on this sample, possibly covering the bare from Figure 51 (see Figure 52). It is likely that since this microstructure is seen so infrequently, it is a eutectic colony at the surface. Figure 53 shows more of the bare cuboidal on the surface. The flat areas are polished. EDS of the crevices between the cuboidal showed higher levels of O than the surrounding .

PAGE 77

63 Figure 51 BSE SEM of bare surface of DS-2 near polished regions. The dark phase is alumina, the white phases are MC carbides, and the grey is the casting. An unpolished part of the casting (lower right) shows bare (3000X). Figure 52 BSE SEM of unpolished DS-2 surface. The surface layer can be seen covering the bare seen in Figure 51.

PAGE 78

64 Figure 53 SEM of virgin surface near polished surface. Bare cuboidal can be seen, along with MC carbides that appear bright. The flat regions are polished areas (1200X). 4.2. Stereology Results As previously stated, there appeared to be large quantities of eutectic, carbides, and porosity in both castings in and near the freckled regions. To quantify this observation, stereological techniques were used to determine the volume percent of these constituents in both castings and in the various regions in these castings. Stereological measurements were performed at several magnifications until the lowest standard deviation in the measurements was foundpoint counts for carbides were performed at 1000X in the optical microscope and SEM, and point counts for eutectic, porosity, and (if present) were performed at 125X primarily on the optical microscope. 4.2.1. Stereology on DS-1 Casting The constituents observed in the DS-1 alloy were eutectic, porosity, carbides, and phase. Due to the small number of freckle chains on the DS-1 castings, only Cuts 1 and 3

PAGE 79

65 had significant data. Figure 54 shows the volume percent differences of eutectic in the freckled regions of the casting. These areas include the freckled grains, grain boundaries, and immediate surrounding regions of less than 30 m near the chains. Figure 55, 56, and 57 show the stereology results of carbides, and porosity from the DS-1 castings. The regions labeled matrix are regions at least 1 mm from any freckle chain and should represent the rest of the DS casting. V olume Percent Comparison of Eutectic in DS-1 Alloy at 125X0123456713Cut #Volume % Freckle Matrix Figure 54 Volume percent of DS-1 eutectic at 125X found through stereology.

PAGE 80

66 V olume Percent Comparison of Carbides in DS-1 Alloy at 1000X.00.511.522.533.5413Cut #Volume % Freckle Matrix Figure 55 Volume percent of MC carbides in DS-1 at 1000X found through stereology. Volume Percent Comparison of Eta Phase in DS-1 Alloy at 125X0123456713Cut #Volume % Freckle Matrix Figure 56 Volume percent of DS-1 phase at 125X found through stereology

PAGE 81

67 V olume Percent Comparison of Porosity in DS-1 Alloy at 125X00.20.40.60.811.21.41.61.8213Cut #Volume % Freckle Matrix Figure 57 Volume percent of DS-1 porosity at 125X found through stereology. From the above figures, it is clear that there is more eutectic and in and around the freckle grains. There is also more porosity in these regions. Statistically, however, it is not possible to determine if there are more or fewer carbides present in the freckled or matrix regions. In fact, on Cut 3, there appeared to be more carbides present in the matrix than in the freckled regions. 4.2.2. Stereology on DS-2 Casting While there was no in the DS-2 alloy, there were eutectic, porosity, and carbides and it was possible to collect data from all three cuts (the same labeling conditions were used for these stereological results). Figure 58 shows the volume percent of carbides, and Figures 59 and 60 show the results for eutectic and porosity, respectively.

PAGE 82

68 V olume Percent Comparison of Carbides in DS-2 Alloy at 1000X00.511.522.533.544.5123Cut #Volume % Freckle Matrix Figure 58 Volume percent of MC carbides in DS-2 at 1000X found through stereology. V olume Percent Comparison of Eutectic in DS-2 Alloy at 125X024681012141618123Cut #Volume % Freckle Matrix Figure 59 Volume percent of DS-2 eutectic at 125X found through stereology.

PAGE 83

69 V olume Percent Comparison of Porosity in DS-2 Alloy at 125X-101234567123Cut #Volume % Freckle Matrix Figure 60 Volume percent of DS-2 porosity at 125X found through stereology. The results for the microconstituents of the DS-2 castings were similar to the stereological volume percents of those for the DS-1 alloy. The percentage of carbides seemed higher in the freckled regions, but this could not be proven statistically. Furthermore, the percentage of eutectic and porosity in the freckled regions was markedly higher than in the matrix, especially in the cross sections (Cuts 2 and 3). While the amount of eutectic present in DS-2 freckles was up to twice as high as in the DS-1 alloy, the ratio of eutectic in the freckled regions to the matrix was similar (approximately six times). 4.2.3. Stereology for Individual Freckle Grains One of the goals for this project was to compare and contrast the amount of the various phases in the freckled regions with those in the matrix, but also to contrast the volume percent of constituents found inside the freckle grains themselves (excluding the grain boundaries and phases at the grain boundaries). However, this process proved difficult for several reasons. First, while it was relatively simple to distinguish the freckle

PAGE 84

70 chains from the rest of the casting, it was difficult to distinguish the individual grains inside the chains. For example, the freckle grain boundaries are usually partially covered with (if present) and eutectic, but these phases are also present in between dendrite arms. Freckles that can be distinguished with the naked eye do not necessarily contrast the same way at higher magnifications in an optical microscope or in an SEM. This is especially a problem at magnifications greater than 100X. At and near the grain boundaries, it is difficult to determine if a phase is inside or outside a grain, or if it is between two grains. Therefore, individual freckle grains that could be distinguished yielded far less data than was found in the previous battery of tests. There were simply not enough grains to produce a sizeable amount of data. Statistically, conclusions should not be drawn from the results. Nevertheless, the results found are shown in Figure 61 (eutectic in DS-1), Figure 62 ( in DS-1), and Figure 63 (eutectic in DS-2). In all figures, the mean volume percent of the individual freckle grains is essentially the same as that of the freckled regions, but with a higher deviation. While one may hypothesize that the freckled region and individual freckle grains have identical volume fractions of eutectic and it cannot be proven in this study due to the small sampling population. Besides quantifying volume percent of constituents in the two castings, PDAS was measured on the same specimens with Cut 3, since this is normal to the primary growth direction. The PDAS for DS-1 is 528 m ( m) and 493 m ( m) for DS-2. No critical PDAS for freckle formation could be calculated because no castings of the same compositions without freckles were available. However, both were larger than the aforementioned critical PDAS for CMSX-2 m. 35

PAGE 85

71 Volume Percent Comparison of Eutectic in DS-1 Alloy01234567891Cut #Volume % Freckle Region Matrix Freckle Grain Figure 61 Volume percent of eutectic in DS-1 alloy at 125X. This figure compares freckle regions with individual freckle grains. Volume Percent Comparison of Eta Phase in DS-1 Alloy0123456789101Cut #Volume % Freckle Region Matrix Freckle Grain Figure 62 Volume percent of in DS-1 alloy at 125X. This figure compares freckle regions with individual freckle grains.

PAGE 86

72 Volume Percent Comparison of Eutectic in DS-2 Alloy0246810121416182023Cut #Volume % Freckle Region Matrix Freckle Grains Figure 63 Volume percent of eutectic in DS-2 alloy at 125X at cuts 2 and 3. This figure compares freckle regions with individual freckle grains. 4.3. EBSD EBSD patterns were generated from the freckle and matrix grains in the DS-1 casting in order to determine the orientations of several different phases, and to find what orientation relationships, if any, existed between individual freckle grains and the rest of the casting. EBSD was first used to establish the orientation of this part of the casting, and to relate any orientation relationships or misorientations of the various constituents in matrix. This task was first performed on the + eutectic to ensure that it was indeed oriented with the dendrite arms for the rest of the casting. In order to do this, the orientation of the matrix was first found. Figure 64 shows an EBSD map of the matrix region from Cut 1. This Kikuchi map is labeled; the green crosshairs towards the top indicate the direction normal to the sample, and the cyan crosshairs towards the center indicate the direction of the electron beam. The normal direction of this sample seems to

PAGE 87

73 be in a <120> direction, which is close to a <110>. This result tends to agree with the fact that the secondary dendrite arms in this part of the sample are rotated away from perfectly orthogonal. In theory, this angle should be around 26. Figure 64 EBSD Kikuchi map of + matrix in DS-1 casting. Next, EBSD data was compiled from various matrix points, and also various eutectic regions. The eutectic was carefully measured by putting a point probe on each section of a eutectic fan. The results were similar to the matrix EBSD map in Figure 64, and were compiled as a pole figure in Figure 65. This pole figure is viewed from a [001] normal direction where RD is the rolling direction (which is the same as the growth direction for this sample), and TD is the transverse direction (see Figure 66). 41 Besides

PAGE 88

74 showing the orientation of the matrix and the eutectic, Figure 65 shows that the orientation of eutectic is identical to that of the rest of the matrix. In moving the electron beam from one side of a eutectic fan to the other (in other words, across each cell), the EBSD map rotated slightly. However, this rotation was too small to be shown on a pole figure. Figure 65 A [001] Pole Figure of matrix and eutectic regions. EBSD was also used to find the relationship between the phase and the eutectic. If a relationship exists between the and the eutectic, then that relationship would hold true between the and the matrix. Figure 67 shows to the right of a eutectic fan. The EBSD map of the eutectic is shown in Figure 68, and the map of the phase is

PAGE 89

75 Figure 66 Schematic of Oxford OPAL system showing directions in the pole figure from Figure 65. shown in Figure 69. While the orientation of eutectic was relatively fixed throughout the sample, the orientation changed from piece to piece. However, the basal plane of the (seen as the thick line on the bottom of Figure 69) matches up with the close-packed (111) plane of the (similar position in Figure 68). Many directions match up as well, such as <112> and < 0101 >, and the <011> and < 02 11> directions. Figure 70 shows a part of a freckle chain taken in the optical microscope. Figure 70 shows the same freckle chain, with the individual freckle chains color coordinated. EBSD was performed on each of these freckle grains along with a few random spots not in a particular freckle grain, but in the freckle chain region. The Kikuchi maps of each grain, along with the [001] pole figure of each grain are shown as Figures 71 to Figure 90. Table 4 shows the misorientations of each grain, and the common zone axis between the freckle and the rest of the casting.

PAGE 90

76 Figure 67 SEM of adjacent to eutectic colony (950X).

PAGE 91

77 (111) Figure 68 EBSD Kikuchi map of eutectic from Figure 67.

PAGE 92

78 (0001) Figure 69 EBSD Kikuchi map of from Figure 67.

PAGE 93

79 Figure 70 Optical image of DS-1 freckle chain. (a) The unaltered image at 50X. (b) Individual freckle chains are numbered and colored.

PAGE 94

80 Figure 71 EBSD Kikuchi map of a spot probed in Freckle 1. Figure 72 The [001] Pole figure of freckle 1 in relation to the matrix. The freckle grain results are highlighted red

PAGE 95

81 Figure 73 EBSD Kikuchi map of a spot probed in Freckle 2. Figure 74 The [001] Pole figure of freckle 2 in relation to the matrix. The freckle grain results are highlighted red.

PAGE 96

82 Figure 75 EBSD Kikuchi map of a spot probed in Freckle 3. Figure 76 The [001] Pole figure of freckle 3 in relation to the matrix. The freckle grain results are highlighted red.

PAGE 97

83 Figure 77 EBSD Kikuchi map of a spot probed in Freckle 4. Figure 78 The [001] Pole figure of freckle 4 in relation to the matrix. The freckle grain results are highlighted red.

PAGE 98

84 Figure 79 EBSD Kikuchi map of a spot probed in Freckle 5. Figure 80 The [001] Pole figure of freckle 5 in relation to the matrix. The freckle grain results are highlighted red.

PAGE 99

85 Figure 81 EBSD Kikuchi map of a spot probed in Freckle 6. Figure 82 The [001] Pole figure of freckle 6 in relation to the matrix. The freckle grain results are highlighted red.

PAGE 100

86 Figure 83 EBSD Kikuchi map of a spot probed in Freckle 7. Figure 84 The [001] Pole figure of freckle 7 in relation to the matrix. The freckle grain results are highlighted red.

PAGE 101

87 Figure 85 EBSD Kikuchi map of a spot probed in Freckle 8. Figure 86 The [001] Pole figure of freckle 8 in relation to the matrix. The freckle grain results are highlighted red.

PAGE 102

88 Figure 87 EBSD Kikuchi map of a spot probed in Freckle 9. Figure 88 The [001] Pole figure of freckle 9 in relation to the matrix. The freckle grain results are highlighted red.

PAGE 103

89 Figure 89 EBSD Kikuchi map of a spot probed in Freckle 10. Figure 90 The [001] Pole figure of freckle 10 in relation to the matrix. The freckle grain results are highlighted red.

PAGE 104

90 Table 4 Misorientations and common zone axes of DS-1 freckle grains. Freckle Number Angle of Misorientation Common Zone Axis 1 47 [7,-19,19] 2 35 [16,14,17] 3 32 [14,-17,16] 4 38 [2,-2,-1] 5 54 [-15,21,-11] 6 13 [5,-9,26] 7 46 [-23,9,-11] 8 59 [14,16,3] 9 60 [15,16,-17] 10 47 [14,-23,-3] These orientations show that there is little if any relationship between the freckles and the matrix regions. This is further backed up by the fact that the common zone axes between the freckle grains and the matrix have high indices. Spots A, B, and C showed the same orientation as the matrix. 4.4. Nanoindenter Results The Triboindenter was used to identify both the hardness and modulus of the microconstituents. No differentiation was made between the freckles and the matrix since the scale of these grains is too large (for example, one would be comparing in freckles with in the matrix). Rather, the following measurements were performed on

PAGE 105

91 phase Matrix phase Cuboidal phase in the matrix Eutectic phase Plates of in eutectic The hardness values for the phase were higher than in the matrix (see Figure 91). Likewise, the hardness of the eutectic was higher than the matrix (see Figure 92). The may be harder than the eutectic , but this cannot be statistically determined. As shown in Figure 93, the inside the eutectic regions was found to have a higher hardness than the matrix When doing a similar comparison with the , no real difference was seen (see Figure 94). The variance in the matrix was high because the indent was often larger than the small precipitates, especially at higher loads. This may have also happened with the eutectic Nanohardness values of DS-1 (Eta vs matrix)0246810120100020003000400050006000Load (N)Nanohardness (GPa) Matrix Figure 91 Nanohardness comparison of and matrix

PAGE 106

92 Nanohardness values of DS-1 (Eta vs matrix)0246810120100020003000400050006000Load (N)Nanohardness (GPa) Matrix Eutectic Figure 92 Nanohardness comparison of eutectic ', and matrix Nanohardness values of DS-1 ( comparison)012345670100020003000400050006000Load (N)Nanohardness (GPa) Eutectic Matrix Figure 93 Nanohardness comparison for in the eutectic colonies and in the matrix. The indent at 5000 N may have been too large to get an accurate measurement.

PAGE 107

93 Nanohardness (' Comparison)024681012140100020003000400050006000Load (N)Nanohardness (GPa) Eutectic Matrix Figure 94 Nanohardness comparison of the in the eutectic colonies and in the matrix. No precise result could be found for the matrix at 5000 N due to the large indent size. Dynamic testing was also performed to create modulus maps of various regions. Figure 95 shows the complex modulus, which is a composite of the stored and loss modulus, for an area of known phase. Figure 96 shows the complex modulus matrix region. The higher modulus areas indicate higher stiffness, and it is shown that has the highest stiffness, and that is stiffer than . The numbers shown are more relative than absolute. When compared to the relaxed modulus results obtained from the static mode of the nanoindenter, the is stiffer than the (see Figure 97). A similar comparison between eutectic and matrix revealed no real difference between the moduli of either phases (see Figure 98).

PAGE 108

94 Figure 95 Complex modulus map of DS-1 sample; phase is shown on the left. Figure 96 Complex modulus map of DS-1 sample in the matrix regions showing cuboidal '.

PAGE 109

95 Relaxed Modulus Comparison of and 05010015020025030035005001000150020002500Load (N)Relaxed Modulus (GPa) Figure 97 Relaxed modulus comparison of and from the matrix regions. Relaxed Modulus Comparison of Eutetic and 0501001502002503000100020003000400050006000Load (N)Relaxed Modulus (GPa) Figure 98 Relaxed modulus comparison of matrix and eutectic '.

PAGE 110

CHAPTER 5 ANALYSIS AND DISCUSSION It has been clearly demonstrated that freckle formation in castings is detrimental to the properties desired from the DS process. The samples analyzed in this study were actually commercial castings that were scrapped because they displayed freckle chains. In this chapter, the data obtained from this project will be correlated with previous findings to create a better understanding of the formation of the deleterious freckle grains. 5.1. Analysis of Phases The microstructures of both DS-1 and DS-2 alloys are consistent with the expected microstructure of DS castings. Both underwent unidirectional dendritic growth and both contained interdendritic regions of eutectic + . The primary dendrites contained the typical and cuboidal microstructures desired by the industry. Elements that usually segregate out of the dendrite core, i.e., those that have a fractional partition coefficient less than one (e.g., Al, Ti, Ta) were found at higher concentrations in the eutectic regions, while the solid-solution strengtheners with partition coefficients greater than one (e.g., Cr, W, Co, Mo) were enriched in the primary dendrite cores. This segregation and the large quantities of alloying additions lead to constitutional supercooling and the resulting dendritic solidification in these alloys. Many of the phases described in previous literature for freckled castings and superalloys in general were observed and identified in both alloys. In DS-1, was identified by multiple characterization techniques. Backscattered SEM showed this phase to have higher atomic number contrast than the or , but less than the MC 96

PAGE 111

97 carbides while the EDS data indicated a phase with mostly Ni and Ti. TEM analysis confirmed the D0 24 structure of the its proximity to the eutectic and the expected orientation relationship. In addition, hardness and modulus tests with the nanoindenter confirmed that this phase was harder and stiffer than other phases, which again pointed to However, no was found in DS-2 likely because of a lower Ti/Al ratio than that of DS-1. The had an orientation relationship with the eutectic : 0211||101)0001(||}111{ (8) Since the eutectic had an identical orientation with the matrix, this relationship holds true for the matrix and as well, and agrees with previous findings. 42 The TEM images showing the stacking faults would suggest this is prone to faulting where the stacking faults in the are local regions of and vice-versa. The areas where appeared alone (not adjacent to a eutectic colony) are likely from a sectioning effect of the sample preparation. Besides was observed in various morphologies throughout the freckle grains in both alloys. While this included the common cuboidal in a matrix and the + eutectic, elongated and rectangular was also observed at the freckle-freckle and freckle-matrix boundaries. The rectangular grain boundary could be a film or plate with holes surrounding a freckle grain. While there is literature that suggests that a film of grain boundary frequently occurs during service and improves rupture properties, 2 the morphology observed in this project has not been previously documented. Carbides were also found in both alloys. The all of those identified were primary MC carbides, which in DS-1 and DS-2 appeared to be (Ta,Ti)C. Secondary, M 23 C 6

PAGE 112

98 carbides were not found in either casting, likely because no heat treatments had been performed on these castings. Some TCP phases, most likely was found in DS-1. Statistically and quantitatively, no more carbides were found in the freckled regions than in the matrix, as was previously reported. 23,24,43 However, there appeared to be an increase in carbides on the virgin surface of the DS-2 casting although there was no obvious difference between the surface of the freckle regions and that of the general casting. These constituents correspond well with the results found by Genereux with midradius VAR ingot freckles. 24 While freckles from both ingots and castings are characterized by an increase in elements that partition to the interdendritic liquid and result in the increased presence of carbides, and , the morphologies of the phases and overall microstructures are different (see Figures 99 and 100 for images of VAR freckles 23 ). Freckles in castings are chains of equiaxed grains surrounded by and containing a larger percentage of and than the rest of the casting, whereas the freckles in ingots are regions of the ingot with an increased percentage of these phases. The freckles in ingots usually result in cracking in the ingot during forging. For those ingots where freckling does not cause this problem, the freckles are not retained in the final product due to extensive hot working of the material. Freckles in castings, though, initiate creep failure from their grain boundaries. The extensive porosity and segregation near the freckles found in this study may indicate that freckles could also be initiation sites for other failure mechanisms not necessarily related to creep, such as fracture or fatigue cycling.

PAGE 113

99 Figure 99 Transverse optical image of freckle in an ingot of Gatorized Waspaloy. The freckle is the darker region in the middle. Figure 100 Optical image of Figure 99 at higher magnification. The white phase is and the dark thin plates are '.

PAGE 114

100 The hardness and modulus results seem to correlate well to accepted literature. The hardness of and is higher than that of the However, the elastic modulus of the matrix was shown to be lower during the dynamic modulus mapping. Static tests of relaxed modulus showed no statistical difference between the two phases. Previous authors 2,44 reported that the phase should have a higher Youngs modulus than the -Ni 3 Al (207 GPa for annealed Ni and 179 GPa for pure Ni 3 Al). The similar modulus values from static testing likely result from noise (e.g., indents were too large). An indent that was too large in the matrix would interact with the interface between and , resulting in an increase in hardness and modulus values. 5.2. Freckle Formation There is little from the results gathered in this project that would dispute the mechanism of thermosolutal convection as the cause of freckle grains on the surface of DS and SX castings. However, several qualitative hypotheses have been backed up with quantitative data, and more information on the virgin surfaces of these castings has been provided. There appeared to be more freckle chains in the DS-2 castings than in the DS-1. These chains were straighter than the bowed freckles chains in DS-1, and they were more closely spaced. This, along with the higher PDAS in DS-2 and larger casting diameter, corresponds to more convective jets occurring in DS-2 than in DS-1. However, it should be noted that the location of these samples in the original castings is unknown, and the number of freckle chains decreases as one goes form the chill to the top of the casting. 23,25

PAGE 115

101 As was previously observed, the freckle regions were found to be enriched in the elements that segregate to the interdendritic regions (Al, Ti and Ta). 6 Even so, as was also observed, many of the larger freckle chains exhibited a dendritic structure similar in size and spacing to the matrix regions. This would tend to support the notion that freckle chains are caused from the dendrite arms of the mushy zone being ripped off by the low-density plumes arising from the mush. At the same time though, these freckle grains exhibited a compositional shift towards the interdendritic liquid composition. Since the dendrite arms have the least solute (Ta, Ti, Al) and the plume emanating from the mush has the maximum amount of solute, this mechanism of freckle formation seems questionable. The measured PDAS for both alloys was above the reported critical values for freckle formation from previous research. 25,35 Large PDAS indicate that the cooling rate (the product of G and R) was low, and extensive macrosegregation of elements in the casting occurred. Therefore, conditions were suitable for the onset of thermosolutal convection. Furthermore, the dendritic structure of freckle grains was similar to that of the matrix. Most likely, the dendrites torn away from the mushy zone were large enough to nucleate new equiaxed grains, but small enough as to contribute minimally to the freckle composition. This would be especially true if partial remelting of the dendrite arms occurred as would be expected if the arms are convectively transported to higher temperature regions. The random orientation of these grains seen from EBSD, which agrees with previous work, 45 shows that the nucleation of these grains is not dependent on the matrix.

PAGE 116

102 It is clear that the freckled regions contain greater amounts of elements that partition to the liquid during solidification and, in turn, more eutectic and (if present) appear in these areas than in the rest of the casting. The eutectic fans at high-angle grain boundaries display an orientation relationship with the grain adjacent to the web-like + portion rather than the coarser lamellar regions. Furthermore, in DS-1, usually appeared adjacent to the lamellar regions of the eutectic colonies and displayed an orientation relationship (see Equation 8) with the eutectic colony. The structure of the eutectic colony is described best by Figure 24, which appears to be a longitudinal section of the colony, and Figure 26, which appears to be a transverse section. This indicates that in these multicomponent superalloys, there is no exact eutectic point, but rather a trough or other geometric construction that describes the solidification of the + colony. This agrees with previous work that showed eutectic growth in superalloys over a 5C temperature range. 19 It is likely that in the both alloys, the web-like regions solidify first. This region is actually small cells of . At some point, some of the cells grow and reject elements that segregate to the dendrite core, causing small layer to form between the cells. This growth continues until the last solidification occursin DS-1 and alloys with high Ti/Al ratios, the last liquid becomes This may stem from the fact that the theoretical solidus for (1304C) is lower than that for (1385C). 3 In alloys where does not form, the eutectic colony completes the solidification process. An excess amount of porosity was also found between the freckle grains and the matrix, and between the freckles themselves. One possible explanation could be pullout of the large eutectic colonies (and in DS-1) during grinding and polishing of the sample. However, such pullout should have been seen in the eutectic colonies as well as

PAGE 117

103 in the matrix. Furthermore, if porosity were caused by the removal of eutectic colonies, then the volume fraction of eutectic in these freckle regions would have been lower. Therefore, the porosity must be a real consequence of solidification as has been observed in previous studies of freckle chains. 2,23,43 It is unlikely the porosity is caused by gases since all processing occurs in a vacuum, but rather when the liquid feeding to the advancing solid is slowed by the meandering dendrite arms in the mushy zone 2 Solidification shrinkage, or macroshrinkage, occurs at the last regions of solidification 2 and could also be a factor in the formation of porosity. More porosity was observed in DS-2 than DS-1 likely because the remaining liquid could not freeze as but had to continue the growth of established eutectic colonies. The morphology and quantity of found between freckle-freckle and freckle-matrix boundaries were not seen at the DS boundaries in either casting. Therefore, the grain boundary is a freckle-related microstructure and is likely caused by excess solute in the freckle chains and grain boundary diffusion of solute to enable coarsening of this . This mechanism also can cause the elongated in the regions surrounding the grain boundaries in the freckled regions. Excess solute and grain boundary diffusion cause the cuboidal near these boundaries to coarsen more than in other regions. Both the excess solute and grain boundary transport are necessary to create this morphology, as it was not present at the DS boundaries. The microstructure and phases present at the surface of the casting appear to be formed by a reaction between the ceramic mold and the liquid metal. This situation is not limited to freckled regions, but does involve surface mechanisms that may influence freckles. Large amounts of MC carbides and alumina were present on the surface and, in

PAGE 118

104 most cases, the alumina layer covered the other phases present. Some zirconia was seen, but only covering alumina. It is likely that a reaction occurred 46 between silica in the ceramic mold and Al in the molten metal: 2 Al + 3 SiO Al 2 O 3 + 3 Si (9) This would explain the trace Si amounts seen on the surface (although this Si may have been from SiC used while grinding). These and any amounts of Si would be so minute as to only contribute to background in EDS and have little or no effect of the composition and microstructure of the castingfreckle or matrix. A similar mechanism at the surface causing the formation of an alumina layer in CMSX4 and PWA 1483 castings was shown to eventually lead to the formation of scale. 47 Large amounts of carbides were seen on the surface as well, which is a reaction of the Ti and Ta in the liquid metal with C. Whether the C is exclusively from the alloy, or if it is from an outside source, such as residual C left from dewaxing of the ceramic mold, is unknown. It is likely that there was an alumina layer between the casting and the mold along the entire surface of the casting before the ceramic was knocked off. During the removal process, the majority of alumina remained adhered to the casting while some was still stuck to the mold, leaving the casting surface that was observed. Upon removal of the ceramic shell, some of the appears to have been removed as well. Because of the oxide found between the surface cuboidal , it is likely that the oxidized faster than the . This corresponds with a lower Al concentration in the making it less resistant to oxidation than the . It is unlikely that would have been pulled away from unless it had oxidized prior to the removal of the mold.

PAGE 119

CHAPTER 6 CONCLUSIONS The objective of this study was to characterize freckle defects in directionally solidified superalloy castings and to understand their mechanism of formation. The following conclusions can be drawn from this study: The freckle grain regions have higher concentrations of those elements that inversely segregate (Al, Ta, and Ti) similar to the interdendritic composition. This indicates that the freckles are a product of interdendritic fluid emanating from within the mushy zone. Freckled regions have higher volume percentages of eutectic, porosity, and (if present) than the rest of the casting. No orientation relationship occurs between the individual freckle grains and the rest of the casting. Dendrite arms torn from the mushy zone in thermosolutal convection are large enough to nucleate freckle grains in the interdendritic plumes, yet small enough to have a minimal effect on the freckle composition when the arms partially remelt. During the last stages of solidification, web-like + eutectic colonies form from the remaining interdendritic fluid. However, a true eutectic transformation does not occur, but rather small cells of that reject elements that form During this growth, some of the cells grow forming a coarse lamellar structure. If the Ti/Al ratio is high enough, this solidification concludes with the formation of whose basal plane is oriented with the {111} of the eutectic. Porosity that forms in and around the freckle grains is caused by a lack of interdendritic fluid feeding the area around the freckles and is also caused by solidification shrinkage of the casting and freckles. Along with the eutectic colonies, this porosity will greatly reduce the toughness and fatigue of the casting. Elongated and intergranular occurring at the freckle grain boundaries is a product of higher solute concentrations and grain boundary transport. Both these mechanisms allow for unique, coarser morphologies of at the freckle-freckle and freckle-matrix boundaries. 105

PAGE 120

106 Not exclusively present in freckle regions, a thin layer of Al 2 O 3 is formed at the surface of the DS castings from a reaction between the ceramic mold and liquid metal. After precipitation, the alumina grows preferentially in the rather than the . Upon removal of the shell, some alumina is left behind and some is detached leaving bare cuboidal .

PAGE 121

LIST OF REFERENCES 1. C. T. Sims, N. S. Stoloff, W.C. Hagel: Superalloys II, John Wiley & Sons, New York, NY (1987) 2. J.R. Davis, ed: Heat Resistant Materials, ASM International, Materials Park, OH (1997) 3. H. Baker, ed: ASM Handbook Volume 3Alloy Phase Diagrams, ASM International, Materials Park, OH (1992) 4. F. Rhines: Phase Diagrams in Metallurgy: Their Development and Application, McGraw-Hill, York, PA (1956) 5. G.E. Dieter: Mechanical Metallurgy, McGraw-Hill, York, PA (1961) 6. T.M. Pollock, W.H. Murphy, E.H. Goldman, D.L. Uram, J.S. Tu: Superalloys 1992, S.D Antolovich, R.W. Stusrud, R.A MacKay, D.L. Anton, T. Khan, R.D. Kissinger, D.L. Klarstrom, eds., TMS Warrendale, PA, pp. 125-134 (1992) 7. G.K. Bouse: Superalloys 1996, R.D Kissinger, D. Deye, D.L. Anton, A. Cetel, M. Nathal, T.M. Pollock, D. Woodford, eds., TMS, Warrendale, PA, pp. 163-177 (1996) 8. R. Darolia, D.F. Lahrman, R.D. Field: Superalloys 1988, S. Reichman, D.N. Duhl, G. Maurer, S. Antolovich, C. Lund, eds., TMS, Warrendale, PA, pp. 255-264 (1988) 9. S.T. Wlodek: The Stability of Superalloys, 1999 TMS Annual Meeting, San Diego, CA, 1999 10. R.A. Horton: ASM Handbook Volume 15Casting, ASM International, Materials Park, OH, pp. 253-270 (1992) 11. S. Tin, T.M. Pollock, W.T. King: Superalloys 2000, R.D. Kissinger, T.M. Pollock, R.R. Bowman, K.A. Green, M. McLean, S.L. Olson, J.J. Schirra, eds., TMS, Warrendale, PA 201-210 (2000) 12. D. Porter, K. Easterling: Phase Transformations in Metals and Alloys, Chapman and Hall, London (1995) 107

PAGE 122

108 13. W. Kurz, D.J. Fisher: Fundamentals of Solidification, Trans Tech Publications Ltd, Brookfield, VT (1989) 14. V.A Wills, D.G. McCartney: Mat. Sci. Eng. A, Vol. A145, pp. 223-232 (1991) 15. M. Vijayakumar, S.N. Tewari, J.E. Lee, P.A. Curreri: Mat. Sci. Eng. A, Vol. A132, pp. 195-201 (1991) 16. N. DSouza, M.G. Ardakani, M. McLean, B.A. Shollock: Met. Trans. A, Vol. 31A, pp. 2877-2886 (2000) 17. W.R. Sun, J.H. Lee, S.M. Seo, S.J. Choe, Z.Q. Hu: Mat. Sci. Eng. A, pp. 143-149 (1999) 18. R. Hamar, I. Cournes, S. Hamar-Thibault: High Temperature Alloys for Gas Turbines and Other Applications, pt. 1, pp. 767-776 (1986) 19. S.N. Tewari, M. Vijayakumar, J.E. Lee, P.A. Curreri: Mat. Sci. Eng. A, Vol A141, pp. 97-102 (1991) 20. S. Minakawa, I.V. Samarasekera, F. Weinburg: Met. Trans. B, Vol 16B, pp. 595-604 (1985) 21. M.C. Flemings, G.E. Nereo: Trans. AIME, Vol. 239, pp. 1449-1461 (1967) 22. J.S. Kirkaldy, W.V. Youdelis: Trans. AIME, Vol. 212, pp. 833-840 (1958) 23. A.F. Giamei, B.H. Kear: Met. Trans, Vol. 1, pp. 2185-2192 (1970) 24. P.D. Generaux, C.A. Borg: Superalloys 2000, R.D. Kissinger, T.M. Pollock, R.R. Bowman, K.A. Green, M. McLean, S.L. Olson, J.J. Schirra, eds., TMS, Warrendale, PA pp. 19-27 (2000) 25. S.M. Copley, A.F. Giamei, S.M. Johnson, M.F. Hornbecker: Met. Trans., Vol. 1, pp. 2193-2204 (1970) 26. S.D. Fellicelli, J.C. Heinrich, D.R. Poirier: J. Crystal Growth, Vol. 191 pp. 879-888 (1998) 27. J.R. Sarazin, A. Hellawell: Met. Trans. A, Vol. 19A pp. 1861-1871 (1988) 28. T.M. Pollock, W.H. Murphy: Met. Trans. A, Vol. 27A, pp. 1081-1094 (1996) 29. S. Tin, T.M. Pollock, W.H. Murphy: Met. Trans. A, Vol. 32A, pp. 1743-1753 (2001) 30. C. Frueh, D.R. Poirier, S.D. Fellicelli: Met. Sci. Eng. A, Vol. A328 pp. 245-255 (2000)

PAGE 123

109 31. F.P. Incropera, D.P. DeWitt: Fundamentals of Heat and Mass Transfer, John Wiley and Sons, New York (1996) 32. P. Auburtin, A. Mitchell: Liquid Metal Processing and Casting AVS Conf., AVS, Santa Fe, NM, pp. 18-34 (1997) 33. J.C. Heinrich, S. Fellicelli, D.R. Poirier: Num. Heat Transfer, Part B, Vol. 23, pp. 461-481 (1993) 34. H. Combreau, G. Lesoult: Modeling of Casting, Welding, and Advanced Soldification Processes VI, T.S. Piwonka, V. Voller, L. Katgerman, eds., TMS, Warrendale, PA, pp. 201-208 (1993) 35. M.C. Schneider, J.P. Gu, C. Beckermann, W.J. Boettinger, U.R. Kattner: Metall. Mater. Trans. A, Vol. 28A, pp. 1517-1531 (1997) 36. P. Auburtin, T. Wang, S.L. Cockcroft, A. Mitchell: Met. Trans. B, Vol 31B, pp. 801-811 (2000) 37. C. Beckermann, J.P. Gu, W.J. Boettinger: Met. Trans. A, Vol. 31A, pp. 2545-2557 (2000) 38. G. F. Van Der Voort: Metallography: Principles and Practice, McGraw-Hill, New York (1984) 39. R.T. DeHoff: Materials Microstructure, University Copy Center, Gainesville, FL (2001) 40. J.I. Goldstein: Scanning Electron Microscopy and X-Ray Microanalysis, Plenum Press, New York (1994) 41. Link Opal Operators Guide, Oxford Instruments PLC, United Kingdom (1997) 42. R. Zhao, R. Beltran: Met. Trans. A, Vol. 32A, pp. 1271-1282 (2001) 43. P. Auburtin, S.L. Cockcroft, A. Mitchell, T. Wang: Superalloys 2000, R.D. Kissinger, T.M. Pollock, R.R. Bowman, K.A. Green, M. McLean, S.L. Olson, J.J. Schirra, eds., TMS, Warrendale, PA 225-261 (2000) 44. MatWeb Material Property Data, http://www.matweb.com/ accessed May 2003 45. R. Schadt, I. Wagner, J. Pruehs, P.R. Sahm: Superalloys 2000, R.D. Kissinger, T.M. Pollock, R.R. Bowman, K.A. Green, M. McLean, S.L. Olson, J.J. Schirra, eds., TMS, Warrendale, PA 211-218 (2000) 46. Y. Virieux, J. Desmaison, J. Labbe, A. Gabriel: Key Engineering Materials, Vol. 132-136, pp. 1657-1660 (1997)

PAGE 124

110 47. K. Kloske: Investigation of Scale Formation on the Surface of Two Directionally Solidified Nickel-Base Superalloys, Masters Thesis, University of Florida (2001)

PAGE 125

BIOGRAPHICAL SKETCH Erik Michael Matthew Mueller was born on July 7, 1978, to Michael J. Mueller and his wife Sally Leona in Jacksonville, FL. There he attended Sacred Heart Elementary and Bishop Kenny High School, where he started at defensive line for the football team and threw shotput and discus for the track team. After graduating in 1996, Erik moved to Gainesville, FL, where he attended the University of Florida. His undergraduate research focused on fractal analysis of lead zirconate/lead titanate fracture surfaces under the tutelage of Dr. J. J. Mecholsky. In May 2000, he graduated cum laude with a Bachelor of Science in materials science and engineering with a focus in engineering ceramics, and began an internship at American Technical Ceramics. There he worked on particle size reduction for processing multi-layered ceramic capacitors. He then entered graduate school later that year at the University of Florida and redirected his focus to metallurgy and electron microscopy. 111


Permanent Link: http://ufdc.ufl.edu/UFE0000919/00001

Material Information

Title: The Characterization of Freckle Casting Defects in Directionally Solidified Nickel-Base Superalloy Turbine Blades
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0000919:00001

Permanent Link: http://ufdc.ufl.edu/UFE0000919/00001

Material Information

Title: The Characterization of Freckle Casting Defects in Directionally Solidified Nickel-Base Superalloy Turbine Blades
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0000919:00001


This item has the following downloads:


Full Text












THE CHARACTERIZATION OF FRECKLE CASTING DEFECTS
IN DIRECTIONALLY SOLIDIFIED
NICKEL-BASE SUPERALLOY TURBINE BLADES















By

ERIK M. MUELLER


A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE

UNIVERSITY OF FLORIDA


2003

































Copyright 2003

by

Erik M. Mueller

































This document is dedicated to everyone across the world who dreams and hopes for a
better world.















ACKNOWLEDGMENTS

I would like first and foremost to thank my parents and God for giving me the

opportunity to attend the University of Florida and receive a higher education. Many

people across the world are unable to receive such a gift, so I hope I have made the best

use of my years at school. I would also like to thank my sister, Melissa, and all my

grandparents, aunts, uncles, and cousins for their continual support throughout my life.

My good friends, Andy, Chris, and Julio, have also been there to show me what real

friendship is like and to support me during this time. Of chief importance is Dr. Michael

Kaufman, who has guided me through this project and has also been there as a friend for

me when I needed help. Dr. Amelia Dempere was instrumental in providing me with the

monetary support to be in graduate school to do this project. Without her and the Major

Analytical Instrumentation Center (MAIC), this project would never have been possible.

I would also like to Dr. Richard Connell for introducing me to materials engineering and

Dr. Gerhard Fuchs for introducing me to superalloys.
















TABLE OF CONTENTS

Page

A C K N O W L E D G M E N T S ................................................................................................. iv

LIST OF TABLES ...................... ........ .................. ....... ............ .. vii

LIST OF FIGURES ............. .. ..... ...... ........ ....... ............. ............. viii

A B S T R A C T .......................................... ..................................................x iii

CHAPTER

1. IN TRODU CTION ................................................. ...... .................

2. LITER A TU RE R EV IEW .............................................................. ....................... 5

2.1. N ickel-B ase Superalloys....................................... .......................................5
2.1.1. C hem ical C om position ........................................... .......................... 5
2.1.2. Role of Alloying Elements ........... ..................................... ................7
2.1.3. Other Phases in Nickel-Base Superalloys ................................................9
2.2. Processing of Superalloys............................................................... ........ ....12
2.2.1. Investm ent Casting ............................................................. ............ 12
2.2.2. D irectional Solidification (D S)....................................... ............... 14
2.2.3. Microstructural Features................................. 16
2.3. Freckles.......................... ..........................19
2.3.1. Theories of Freckle Formation ............ .............. ..........................20
2.3.2. M odeling................................................... 24

3. M ETHOD S AND M ATERIALS ........................................ ......................... 28

3 .1. M materials ............................................................................................................2 8
3.2. O optical M icroscopy ........................................ .................. ............... 28
3.3. Preparation of Sam ples .......................................................................... ......... ....29
3.4. Stereology ............................................................... ......... 30
3.5. SEM and TEM ......... ........... .. ....... ........ ... ....... ............... ..32
3.6. Electron Backscattered Diffraction............................................. ...............33
3.7. Nanoindentation........ ........ ...... .... .. .. ......... .......... 34





v









4 R E SU L T S ....................................................... 36

4.1. Microstructural Characterization ................ .... .............. 36
4 .1.1. C om positions ................................................... .. ............ .... ..... 38
4.1.2. Surface of the Castings......... ................. ......... ................... .... 41
4.1.3. Eutectic Colonies and Porosity ....................................... ...............45
4.1.4. Eutectic at Grain B oundaries.................................. ....................... 48
4.1.5. Eta Phase Form ation................................................ .. .. ............ ...............51
4.1.6. Morphology and Microstructure of Constituents near Surface .................57
4.2. Stereology Results ............................. ...................... .. ............... 64
4.2.1. Stereology on D S-1 Casting ....................................................... ......... 64
4.2.2. Stereology on DS-2 Casting .........................................................67
4.2.3. Stereology for Individual Freckle Grains ..............................................69
4 .3 E B S D .......................................................................... 7 2
4.4. N anoindenter R results ............................................................ ............... 90

5. ANALYSIS AND DISCU SSION ........................................ ......................... 96

5 .1. A n aly sis of P h cases ...... ... .... ................................................................... 9 6
5.2. Freckle Form action .................. ................................... .. ........ .... 100

6. CONCLU SION S .................................. .. .......... .. .............105

LIST OF REFEREN CES ........................................................... .. ............... 107

BIOGRAPHICAL SKETCH ........................................................... .................111
















LIST OF TABLES

Table pge

1. Common stereological formulas and relations to microstructural features ...................32

2. Estimated nominal compositions of DS-1 and DS-2 alloy.........................................39

3. EDS weight percent of elements for phases in Figure 36. ..........................................56

4. Misorientations and common zone axes of DS-1 freckle grains ................................90
















LIST OF FIGURES


Figure p

1. Binary phase diagram of N i and A l ........................................... .......................... 6

2. Crystal structures of FCC Ni, y, and Ni3A y'....................................... ...............7

3. The invest ent casting process ......................................................... .....................13

4. Schematics of (a) a DS furnace and (b) a single crystal furnace ..................................15

5. Isosurfaces of volume fraction and stream traces emerging from a channel ...............21

6. Isosurfaces of solute concentration, showing solute enrichment within channels and the
em erging plum es. .......................... ...... ................................... .. ..... 22

7. The effect of growth velocity (R) and thermal gradient (G) on freckle formation........24

8. Illustration showing the castings pinching back on itself after being cut by the wire
E D M ...................................... .................................................... 2 9

9. Illustration of the directions cut in relation to the primary dendrite arms (not to
sc ale) ...................................................... .... ................. 3 0

10. Manual counting methods used for stereology .................................................31

11. Illustration of indent head mechanism in the Triboindenter ................. .................34

12. Photograph of the DS-1 castings and a labeled freckle chain................ ..................37

13. Photograph of the DS-2 castings with some visible freckle chains labeled. ...............37

14. ED S spectrum of D S-2 alloy at 15 kV .......... .................................... ............... 38

15. Comparison of freckle chain and matrix compositions in weight percent from DS-1
alloy without Ni .................................... ............................... .........40

16. Comparison of freckle chain and matrix compositions in weight percent from DS-2
alloy without Ni.................................... ............................... .........40

17. SEM (Scanning Electron Micrograph) of etched DS-1 surface showing dendritic
stru ctu re .......................................................................... 4 1









18. SEM of etched surface of DS-2 casting revealing dendritic structureof freckle
ch ain ..................................................................................... 4 2

19. SEM of unaltered DS-2 alloy surface (100X).......... ... ........................43

20. SEM of surface of DS-2 casting showing charging alumina phase (650X). ..............43

21. BSE (Backscattered Electron) SEM of Figure 20...............................44

22. X-Ray map of Figure 21 revealing the Al enrichment in the charging phase and the
Ta and Ti enrichment in the high Z phase..................................... ...... ............... 44

23. Comparison of EDS data from matrix and freckle regions of unaltered surface of DS-
2 alloy ................................................................................ 4 5

24. SEM of eutectic colony at a freckle boundary in DS-2 alloy (1000X)........................ 46

25. SEM of a eutectic colony in DS-1. Note the web-like morphology of the y
(2 0 0 0 X ). .......................................................... ................ 4 6

26. SEM of different section of eutectic in DS-2 alloy at the casting surface (2000X). ...47

27. Optical micrograph of cross section of DS-2 casting.. ................................................47

28. SEM of porosity near a colony of r at a freckle grain boundary (1000X). ................48

29. SEM of DS-2 alloy near surface showing elongated y' at the boundaries between three
freckles. .............................................................................49

30. SEM of DS-1 alloy at boundary of freckle grains and matrix...............................49

31. SEM of DS boundary in DS-1 (5500X).... .................... ................. ........ ....... 50

32. SEM showing intergranular y' at a freckle grain boundary in DS-2 (900X)...............50

33. SEM of intergranular y' between two eutectic colonies at a freckle grain boundary
from DS-2 (750X). .................................... .. ....... .. ............ 51

34. Optical image of a freckle grain in DS-1 casting showing various phases (50X). ......52

35. BSE SEM of freckle boundary in the DS-1 showing r phase (500X). .......................53

36. BSE SEM showing r phase, carbides, and TCP phase at a DS-1 freckle boundary
(19 0 0 X ). .......................................................... ................ 5 3

37. X-ray map of Figure 36, showing elements common to r phase...............................54

38. SEM of Phase (3300X ) .................................................. ........................................... 54









39. X-ray map of Figure 38, showing elements common to r phase...............................55

40. SEM of M C carbide in matrix (10000X)........................ ........................ 55

41. X-ray map of M C carbide in Figure 40. ........................................... ............... 56

42. BF TEM image of r containing stacking faults stacking faults near a carbide
(3 0 0 0 0X ). ......................................................... ................ 57

43. BF TEM closeup of stacking faults above the carbide from Figure 42 (100000X). ...57

44. SEM of a cross section of the DS-2 casting surface at a freckle chain (1400X). .......58

45. SEM of a large eutectic colony near the external surface of the DS-2 casting
(1000X ) ..............................................................................59

46. SEM of large eutectic colony at surface (450X) ..............................59

47. SEM of carbide in the center of large eutectic colony near casting surface
(1800X ) ..............................................................................60

48. SEM of carbide near surface (2500X). ............................... ................................ 61

49. B SE SEM of Figure 42 (2500X ).......................................... ............................ 61

50. SEM of casting surface showing the "smearing" of the y + y' morphology (3500X)..62

51. BSE SEM of bare surface of DS-2 near polished regions (3000X)...........................63

52. BSE SEM of unpolished D S-2 surface............................................. ...............63

53. SEM of virgin surface near polished surface................................... 64

54. Volume percent of DS-1 eutectic at 125X found through stereology .......................65

55. Volume percent of MC carbides in DS-1 at 1000X found through stereology. .........66

56. Volume percent of DS-1 r phase at 125X found through stereology .......................66

57. Volume percent of DS-1 porosity at 125X found through stereology........................67

58. Volume percent of MC carbides in DS-2 at 1000X found through stereology. .........68

59. Volume percent of DS-2 eutectic at 125X found through stereology .......................68

60. Volume percent of DS-2 porosity at 125X found through stereology........................69

61. Volume percent of eutectic in DS-1 alloy at 125X .................................... ... ......71









62. Volume percent of r in DS-1 alloy at 125X........................................................71

63. Volume percent of eutectic in DS-2 alloy at 125X at cuts 2 and 3..............................72

64. EBSD Kikuchi map of y + y' matrix in DS-1 casting ............. .......................73

65. A [001] Pole Figure of matrix and eutectic regions................................ ............... 74

66. Schematic of Oxford OPAL system showing directions in the pole figure from
F igu re 6 5. ......................................................... ................ 7 5

67. SEM of r adjacent to eutectic colony (950X). ...............................76

68. EBSD Kikuchi map of eutectic from Figure 67................................ ............... 77

69. EBSD Kikuchi map of from Figure 67.......................................... ............... 78

70. Optical image of DS-1 freckle chain. (a) The unaltered image at 50X. (b) Individual
freckle chains are numbered and colored. ...................................... ............... 79

71. EBSD Kikuchi map of a spot probed in Freckle 1..................................................... 80

72. The [001] Pole figure of freckle 1 in relation to the matrix. The freckle grain results
are highlighted red............. .................................... .............. ........ 80

73. EBSD Kikuchi map of a spot probed in Freckle 2...................................................81

74. The [001] Pole figure of freckle 2 in relation to the matrix. The freckle grain results
are highlighted red............ ................. .... .. ......... .. ........ .. ........... ....81

75. EBSD Kikuchi map of a spot probed in Freckle 3..................... ................................. 82

76. The [001] Pole figure of freckle 3 in relation to the matrix. The freckle grain results
are highlighted red............ ................. .... .. ......... .. ........ .. .............. .. 82

77. EBSD Kikuchi map of a spot probed in Freckle 4..................... ................................. 83

78. The [001] Pole figure of freckle 4 in relation to the matrix. The freckle grain results
are highlighted red............ ................. .... .. ......... .. ........ .. .............. .. 83

79. EBSD Kikuchi map of a spot probed in Freckle 5..................... ................................. 84

80. The [001] Pole figure of freckle 5 in relation to the matrix. The freckle grain results
are highlighted red............ ................. .... .. ......... .. ........ .. .............. .. 84

81. EBSD Kikuchi map of a spot probed in Freckle 6 .............................. ...............85

82. The [001] Pole figure of freckle 6 in relation to the matrix. The freckle grain results
are highlighted red............ ................. .... .. ......... .. ........ .. .............. .. 85









83. EBSD Kikuchi map of a spot probed in Freckle 7 .................. ............................86

84. The [001] Pole figure of freckle 7 in relation to the matrix. The freckle grain results
are highlighted red............ ................. .... .. ......... .. ........ .. .............. .. 86

85. EBSD Kikuchi map of a spot probed in Freckle 8..................................................... 87

86. The [001] Pole figure of freckle 8 in relation to the matrix. The freckle grain results
are highlighted red............ ................. .... .. ......... .. ........ .. .............. .. 87

87. EBSD Kikuchi map of a spot probed in Freckle 9..................................................... 88

88. The [001] Pole figure of freckle 9 in relation to the matrix. The freckle grain results
are highlighted red............ ................. .... .. ......... .. ........ .. .............. .. 88

89. EBSD Kikuchi map of a spot probed in Freckle 10.......................... .................. 89

90. The [001] Pole figure of freckle 10 in relation to the matrix. The freckle grain results
are highlighted red............ ................. .... .. ......... .. ........ .. .............. .. 89

91. Nanohardness comparison of r and matrix y................... ...............91

92. Nanohardness comparison of r, eutectic y', and matrix y ........................................92

93. Nanohardness comparison for y in the eutectic colonies and in the matrix ...............92

94. Nanohardness comparison of the y' in the eutectic colonies and in the matrix...........93

95. Complex modulus map of DS-1 sample; r phase is shown on the left.....................94

96. Complex modulus map of DS-1 sample in the matrix regions showing cuboidal y'...94

97. Relaxed modulus comparison of y' and y from the matrix regions..............................95

98. Relaxed modulus comparison of matrix y and eutectic y'.............................................95

99. Transverse optical image of freckle in an ingot of Gatorized Waspaloy...................99

100. Optical image of Figure 99 at higher magnification.............................. .............99















Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Science

THE CHARACTERIZATION OF FRECKLE CASTING DEFECTS
IN DIRECTIONALLY SOLIDIFIED
NICKEL-BASE SUPERALLOY TURBINE BLADES


By

Erik M. Mueller

August 2003

Chair: Dr. Michael Kaufman
Major Department: Materials Science and Engineering

The purpose of this project was to characterize equiaxed grains, or freckles, that

appear on the surface of directionally solidified superalloy turbine blades. There has

been extensive work on modeling the formation of freckle defects on castings using

mathematical models and 3-dimensional imaging. Unfortunately, not all the model

predictions can explain every microstructure and defect created in a system with upwards

often components. During a post-casting etch, much of the surface of these castings is

lost-this project analyzed two castings with unaltered, unetched surfaces in an attempt

to observe the microstructure in these regions. Stereology was also used to quantify the

purported increase in various microstructural features near the freckles. Indeed a marked

increase in porosity, r, and y + y' eutectic colonies was seen in freckled regions of the

castings. Several morphologies ofy' were observed, including grain boundary y', y'

adjacent to grain boundaries, and eutectic y'. It will be shown that these eutectic colonies









do not form by eutectic growth, but rather by cellular solidification. Further, there

appears to be no preferred orientations of the freckle grains relative to the casting.

Although not exclusive to freckle regions, the surface of the casting has an external

alumina layer that is likely the result from a reaction between the ceramic mold and the

Al in the liquid metal. Further, there are an unusually large number of MC carbides

immediately underneath the alumina layer. These reactions, the freckle characteristics,

and the microstructural evolution during single crystal growth will be described in some

detail.














CHAPTER 1
INTRODUCTION

Today's modem jet aircraft, both military and civilian, demand higher

performance, efficiency, and endurance from their engines. For many decades, the

aerospace industry has believed that ceramics would eventually fill the role as the

material of choice for turbine blades and vanes in jet engines. However, no available

ceramic has both the toughness and high temperature strength requirements necessary in

these sometimes corrosive environments. Therefore, the material of choice continues to

be superalloys, particularly those of alloyed nickel.

Materials that provide relatively high strength and environmental resistance at

elevated temperatures have been lumped into the category of high-temperature materials.

When this subject is broached, materials that come to mind are usually ceramics, steels,

refractory metals, superalloys, and certain carbon composites. However, one must take

into account as to how one defines the phrase "high temperature." Can a specific

temperature threshold be assigned, or is it relative to each material? For example, 2024

aluminum begins to melt at 5020C, and significant creep occurs above 3000C. This could

obviously not be used in critical aircraft engines that operate in excess of 10000C.

Therefore, the general rule is to base the designation of "high-temperature" to 30 to

50% of the melting point of a material. Above this 50% melting point, diffusion and

other atomic motion mechanisms accumulate and the material loses usefulness.

Therefore, it takes careful materials selection to find a suitable material for this









application. For the aforementioned turbine blade, one would choose a superalloy

material. Other materials would likely be ruled out because

* Ceramics would not have the fracture toughness to withstand the high stresses in
the turbine

* Refractory metals would oxide too fast

* Steels would creep and yield under these conditions.

Thus, for these applications, Ni-based superalloys are unique in that they not only possess

the high strength and toughness at high temperatures, but they also have effective

resistance against oxidation and corrosion if properly alloyed.

For aircraft turbine engines, the goal is to maximize the efficiency. By increasing

the inlet temperature 560C (100F), the power output will increase 4%.1 Nickel-base

superalloys are distinctive in that, unlike most materials, they can operate at temperatures

near 85% of their melting temperature. Today, aircraft engine manufacturers are busy

trying to increase the engine output, which in turn requires better composition control.

Nickel-base superalloys outperform cobalt-base superalloys because of the coherent

precipitate phase, Ni3Al. This phase increases the strength and rupture life of a turbine

blade dramatically, especially as temperature increases.2

Besides alloy chemistry, there are other techniques to improve temperature

capabilities-directional solidification and thermal barrier coatings. Directional

solidification (DS) greatly improves the rupture strength and life of a turbine blade by

reducing the number of grain boundaries normal to the applied stress. These grain

boundaries are the reason for failure at elevated temperatures due to grain boundary

fracture. A step above DS superalloys is blades cast as single crystals. Single crystal

processing eliminates grain boundaries altogether, and these parts have been shown to









operate at temperatures 25-500C higher than the same alloys directionally solidified. In

addition, fatigue life is improved as well.

In spite of their superior properties, a major problem with using nickel-base

superalloys is that they are extremely expensive. Many of the alloying elements (such as

tantalum, rhenium, and ruthenium) cost in excess of $400 a pound. Add to the fact that

the DS and single crystal casting methods are difficult to perform, and the price soars

astronomically. Therefore, it is crucial to increase the quality control of the casting

process and reduce the amount of material that has to be scrapped. A major problem in

the turbine blade industry has been the formation of casting defects exclusive to DS and

single crystal castings. These include scale and freckles. Superalloy scale is a rough

raised layer on the casting surfaces that typically contains oxides and other undesirable

phases-this can usually be removed by sandblasting or low-pressure water blasting.

Freckles and other internal defects, though, cannot be removed as their depth is

considerably greater.

Freckles in single crystal and DS castings appear as long chains of equiaxed grains

that run vertically along the sides of the castings. They ruin the enhancements of the DS

and single crystal processes by acting as failure sites for creep rupture. To detect

freckles, the as-cast turbine blades are etched and visually inspected; if a freckle trail is

discovered, the entire casting must be scrapped. Therefore, it is of significant importance

to the jet propulsion and gas turbine industries to determine the cause of freckle defects

and prevent their formation.

Consequently, the goal of this project is to characterize these freckle defects in DS

superalloy castings and understand the mechanism of their formation. There has been









extensive work on modeling the formation of freckle defects on castings using

mathematical models and 3-dimensional imaging. Unfortunately, not all of the model

predictions correlate exactly to the actual castings defects. One of the major drawbacks

is the etching process used to remove the surface of the casting to reveal freckles. A

good portion of the surface layer is removed during this process, which may yield clues

to the origin of these freckle defects.

Furthermore, many researchers have commented on the increase in various

microstructural features near the freckled areas of the castings. However, no true

quantitative measurements have yet been made. In this study stereological methods will

be applied to the castings to determine what, if any, differences exist in the volume

fraction of microstructural features in and near freckles relative to the rest of the casting.

Other characterization techniques will also be employed to investigate these features

qualitatively-shape, size, and relative location. The orientation of these features and the

freckle grains themselves, with respect to the casting, will be investigated as well.














CHAPTER 2
LITERATURE REVIEW

As stated in Chapter 1, most of the research in characterizing freckle defects in

nickel-base superalloys and correlating them with the mathematical modeling results has

already been produced. This section will explain, in depth, about nickel base superalloys,

their processing, chemistry, and the nature and theory behind their defects including

freckles.

2.1. Nickel-Base Superalloys

Nickel base superalloys are often used for applications that require high strength

and fracture toughness at elevated temperatures. These alloys are especially useful for

the hottest areas of aircraft engines and industrial gas turbine engines (IGTs). This is

because, in addition to being strong and tough, they are resistant to fatigue, oxidation,

creep, and hot corrosion. These properties stem from the microstructure of the y and y'

phases of nickel-base superalloys, the addition of various elements, and the casting

processes and heat treatments used.

2.1.1. Chemical Composition

What makes superalloys unique is the coherent interface between the austenitic

FCC y structure and the L12 y' structure of Ni3Al (see Figure 1).3 Both have nearly

identical lattice parameters creating extremely low mismatch (see Figure 2).2 The

morphology of the y' is determined by the mismatch: spheres occur from 0 to 0.2% and

cubes appear from 0.5 to 1%.2 The amount of misfit can be controlled, in part, by the

concentration of other y' former. In fact, up to 60% of the Al in Ni3Al can be substituted











by Ti, Ta, and Nb. The y' phase is a superlattice that shows long-range ordering (LRO)


to its melting point. Since this is the primary source ofNi superalloy strength, the


volume percentage of y' will greatly affect the strength of the alloy. Current commercial


cast alloys have greater than 50% y', which results in higher creep strength but

dramatically reduces their hot workability. This increase in creep strength with

increasing temperature is due to an anomalous increase of flow stress with temperature. 1

A. l .i . ..... ............ ....
I IIf 0 sO to 7 I U 9}I



laC r/ i. \.










kAl NI








-------- -.......-........
2U IED AD 510 80 0 i


A I % QI NM N








Figure 1 Binary phase diagram of Ni and Al
Figure 1 Binary phase diagram ofNi and Al









L12 FCC
NiYAI Ni


O *
0 0 *
O


O Al
*Ni Ni
ao = 3.561 ao = 3.532 A
3.568 A
Figure 2 Crystal structures ofFCC Ni, y, and Ni3A1, y'

2.1.2. Role of Alloying Elements

Modem superalloy compositions contain 8 or more elements, creating a very

complex system to describe thermodynamically. It would be extremely difficult to

construct and interpret an 8 to 10-dimensional phase diagram.4 Furthermore, many of the

sought after phases and microstructures of nickel-base superalloys are metastable and

must be developed through careful heat treatments. Finally, many of these elements tend

to segregate to different areas of the casting-either the dendrite core or the interdendritic

regions.

One of the base elements for any nickel-base alloy is chromium, and in fact, the

original superalloy was Ni-20Cr. Cr is an important element in that its primary function

is to protect the alloy from oxidation and other forms of corrosion. Cr will form a

protective oxide layer on the surface that will adhere well to the surface with a matching

coefficient of thermal expansion. Cr is particularly effective in preventing hot corrosion

and sulfidation. Unfortunately, the Cr203 layer is only effective until about 9000C. Cr

also acts as a solid solution strengthener and carbide former.2









Another major element used to environmentally protect nickel-based superalloys is

aluminum. Al, like Cr, forms a protective layer on the surface, but is more stable than

Cr203 at higher temperatures. However, A1203 does not protect well in corrosive

environments, such as sulfur intensive hot corrosion, so both Al and Cr are needed in

relatively high quantities to protect the part. The primary function of Al, though, is to

strengthen the alloy by forming y'.

While being a solid-solution strengthener, Co also alters the y' solvus temperature

and aids in expanding the temperature window for heat treatments. Co also reduces the

stacking fault energy (SFE) of the alloy, which improves creep resistance and makes

dislocation cross slip around y' particles more difficult.1'5

Ti, Ta, Nb, W, Re, and Mo are all used as solid-solution strengtheners. Ta, Ti, and

Nb strengthen the y' precipitates by enhancing the anti-phase boundary (APB) energy,

while W, Re, and Mo strengthen the y matrix. These elements, along with Cr, also react

with carbon to form the various carbides present in superalloys. For various reasons

usually stemming from corrosion and oxidation resistance, recent trends in DS and single

crystal castings favor using more Ta, W, and Re for strengthening, as opposed to Mo and

V, which were used in older alloy designs. Mo and V lower oxidation resistance while

Re and W show synergistic effects in strengthening the y/y' interface. Ta strengthens y',

and has been shown to reduce freckle formation,6 as will be discussed later. The

concentration of these refractory elements must be controlled, however, as they increase

the density, cost, and likelihood of forming deleterious topologically close-packed (TCP)

phases.









C, B, Zr, and Hf are added to strengthen and stabilize grain boundaries. C reacts

with Hf, Zr, and other elements of the casting (Ta and Ti) to form MC carbides, which

retard grain boundary sliding and alleviate residual stresses in the casting. Hf and Zr are

particularly effective in that they react with deleterious impurities, such as S and P,

before all other elements. B, likewise, forms borides that also strengthen the grain

boundaries. However, due to directional solidification and single crystal casting

technologies, the need for grain boundary strengtheners and stabilizers has diminished.

2.1.3. Other Phases in Nickel-Base Superalloys

Besides the y and y' phases, there are many other phases that may be present in

superalloys. Some, like carbides, are desired under certain conditions. However, the

majority are detrimental phases that initiate fast fracture and reduce service life.

The C in superalloys tends to react with one or more alloying elements in solution

to form carbides. Carbides serve as excellent grain-boundary strengtheners improving

creep rupture life dramatically. They also strengthen the rest of the casting and act as

traps for harmful elements such as P and S. Carbides formed in superalloys can be

classified as either primary or secondary. Primary carbides are those that form from the

melt and are almost always the MC phase (eg. TaC, TiC) with an NaCl structure. These

carbides are randomly oriented with the matrix and are located both inter- and

intragranularly. MC carbides tend to be extremely stable such that they typically survive

various heat treatment schedules. Their morphology can be either script, cuboidal, or

blocky.2

Secondary carbides are those that form from solid solution, usually during lower-

temperature heat treatments or during service. Their constitution and structure depend on









overall alloy composition and include M23C6, M7C3, and M6C carbides. These carbides

usually appear at grain boundaries and can be created and altered in reactions between

MC carbides and the y matrix. M7C3 carbides are trigonal, form in low Cr concentrations,

and are unstable. They usually decompose into M23C6 carbides, which have complex

cubic structures and usually form in alloys with higher Cr concentrations. M23C6 are

created either as a product of M7C3 decomposition, or as a reaction of MC and y. The

M6C carbides are similar to that of the M23C6 in structure and morphology, except that

they tend to form at higher levels of W and Mo. M6C appear to be more stable at higher

temperatures than M23C6.1

Borides, like carbides, act as excellent grain boundary strengtheners. They are

usually hard, blocky particles of tetragonal M3B2, and form at small concentrations of B

(<0.03 wt %). However, single crystal alloys usually lack carbon and boron due to the

absence of grain boundaries. The philosophy behind most single crystal processing is to

increase the amount ofy' for strengthening. However, as will be discussed later, carbon

may have other beneficial effects on single crystal (SX) and DS superalloy castings.

In alloys with high concentrations of Nb, the y" phase will form. This is a coherent

body-centered tetragonal (bct) phase of Ni3Nb. This phase strengthens the alloy

primarily because of its high lattice mismatch strains with the matrix. Similar to y', y"

will provide high strength at low temperatures, but unlike y', it is a metastable phase

which becomes unstable above 650C. Furthermore, one must ensure a carefully

controlled heat treatment to prevent the formation of the stable 6-phase of Ni3Nb. This is

an incoherent orthorhombic phase that can strengthen grain boundaries in small amounts,

but in large amounts (more volume percent than y") is detrimental to the alloy.









The hexagonal r-phase (D024) of Ni3(Ti,Ta) is often found in superalloys with high

(Ti+Ta)/Al ratios, especially after considerable service life. The r phase can also form

during solidification after the y + y' eutectic7 and can appear in a pearlite structure of

alternate lamellae with y or as a ribbon-like Widmanstatten pattern.2 In either case, the

r phase grows slowly, but can eventually grow larger than the y' particles. Usually

occurring near grain boundaries, this incoherent brittle phase will reduce the ductility and

rupture strength of the alloy. There has been limited investigation into r formation. This

phase has been found to be a component in the scale that forms on castings of PWA

1483.7 This r was found to contain a percentage of Si that dissolved from the mold. It

was also found that the precipitation of MC carbides reduced r formation due to the

reduction in the remaining Ti and Ta contents in the alloy.7

There are a myriad of TCP phases that can be present in a casting that are

detrimental to the performance of the part. These phases are complex in nature,

incoherent, and brittle resulting in a lowered rupture life, ductility, and creep resistance.

These include [t, Laves, c, P, and G phase.1'2'8 The [t phase is rhombohedral and forms at

high temperatures in alloys with high W and Mo. The tetragonal CY phase, which also

appears in stainless steels, forms elongated, irregular globules with a structure similar to

M6C carbides and appears during service at elevated temperatures. Laves, P, and G

phases are all complex, irregularly-shaped globules, which are strongly promoted by

silicon and greatly reduce fatigue and creep resistance.8'9 With all TCP phases, including

r, only a slight change in local composition is is needed to force the composition past the

stability threshold.









Besides these TCP phases, a variety of nitrides, oxides, hydrides, and other ceramic

impurities can form and thus reduce the desired properties not only by acting as crack

initiation sites, but also by removing alloying elements from solid solution. Therefore, it

is imperative to control the chemistry of each casting and protect the environment during

casting.

2.2. Processing of Superalloys

Nickel-base superalloys need to be processed under special conditions. They must

be cast at high temperatures due to their high melting temperature (TM), but must also be

cast in a vacuum to prevent excessive oxidation and contamination. Those alloys with

high concentrations of solid-solution strengtheners, especially y' former, cannot be hot

or cold worked and must therefore be cast.

The casting or forging of nickel-base superalloy parts is a multistage process in

controlled environments. First, the raw materials and reused scrap are vacuum induction

melted (VIM). VIM involves melting of the material in a vacuum to prevent the

formation of slag and control the reactivity of the elements. This melting step produces

ingots with coarse, non-uniform grain sizes that must be refined for forged components.

Either vacuum arc remelting (VAR) or electroslag remelting (ESR) are performed to

refine the ingot. Both of these processes remelt the original VIM ingot to improve the

microstructure and purify the alloy.2 ESR differs from VAR in that it can be carried out

in air. For castings, though, only the VIM process is necessary.

2.2.1. Investment Casting

Due to the complex shapes often required for blades and vanes, many superalloy

components are best produced by lost-wax investment casting (see Figure 3).10 Lost wax

investment casting involves the creation of a replica made of wax that must be designed









to compensate for shrinkage. Any desired holes or cavities are produced by ceramic

cores inserted into the wax pattern. The wax pattern is then assembled with other

patterns to form a 'tree.' The tree is then dipped repeatedly into slurry baths while

rotating to insure complete coverage and uniform thickness. When a sufficient slurry

coating has been applied, the mold is covered with coarse stucco particles. This way, fine

and smooth ceramic particles conform to the wax shape inside of the mold, whereas the

stucco provides strength and prevents cracking and excess runoff of the wet slurry. This

process is repeated several times until the mold is sufficiently thick.

The ceramic mold is then dried and placed into an oven for the dewaxing process.

Here, the mold is heated to remove the wax and leave the hollow ceramic shell. It is

important to control the rate of temperature change to prevent cracking in the mold.

After the dewaxing process, the mold is sintered.











Pattern Assembly Investing Stuccoing Dewaxing


Firing Pouring Knockout Finishing Inspection
.Figure 3 -The investment casting process
Figure 3 The investment casting process









To make the actual casting, the ingots of superalloy are melted and poured into a

preheated ceramic shell. After cooling, the ceramic shell is removed and usually

destroyed, and the metal parts are machined off the 'tree.' However, for DS and SX

castings, considerably more care (and cost) must be added to the investment casting

process.

2.2.2. Directional Solidification (DS)

The process of directional solidification (DS) helped to revolutionize the

development of modem nickel-base superalloys. DS alloys have higher creep rupture life

in addition to improved fatigue resistance and ductility. This is due to the vast reduction

in grain boundary surface area, which acts as an initiation site for creep failure. Fatigue

resistance is maximized due to the <100> preferred growth direction of the DS grains

(the <100> orientation has the lowest modulus, as opposed to other growth directions).2,5

The DS process for superalloy components involves an investment procedure but,

in this case, a special furnace is used. The ceramic mold must be heated to above the

liquidus temperature of the alloy in the furnace. The bottom of the mold is placed onto a

chill plate of Cu or some other conductive metal that is water-cooled. After filling the

shell with molten alloy, it is slowly withdrawn from the furnace, which is also above the

liquidus temperature. This allows the original <100> grains to continuously grow

throughout the casting (see Figure 4).2









Insulation Susceptor Induction
Susoeptor induction i ,/ coil
Scol Radiation
I,--Radiation -- Ieatng
S heating 0



0 0 0 Molten 0l

0 metal Radiation Baffle
Ceramic adiation cooling
heating -i






mold /+ cooling Ceramic cstal selector
mold
Solidification 00 00 Water-cooled
front I chill Water-cooled Columnar-grain


Figure 4 Schematics of (a) a DS furnace and (b) a single crystal furnace

It is important to control the processing conditions ofDS components in order to

prevent a myriad of defects. Upon contact with the water-cooled chill, the molten alloy

immediately solidifies and a thermal gradient (G) is established. This gradient is

controlled and influenced by the temperature of the chill, the size of the casting, the

temperature of the upper and lower chambers of the furnace, and other variables. The

casting cannot be withdrawn until the solid dendrites have reached the area of the casting

at the liquids temperature. Because of the low thermal conductivity ofNi, the effect of

the chill is small and the majority of heat must be dissipated by other means: conduction

through the casting, conduction from the ceramic mold, and radiation emanating from the

mold. At steady-state conditions, the thermal gradient is controlled by the size of the

casting and the baffle. To ensure that the growth of columnar grains does not break

down, G and the withdrawal velocity (R) must be carefully controlled.

Defects frequently occur in these castings and include grain boundary cracking,

bulging, non-metallic inclusions, and recrystallization. Grain boundary cracking is

caused by a hoop stress from ceramic cores during cooling and can be prevented by









adding Hfto the alloy. Bulging is caused by creep of the mold shell, causing flat surfaces

in the casting to bulge out.1 This distorts the ceramic mold, and inclusions are formed

from parts of the mold that break off when the molten alloy is poured into the mold.

Both of these defects can be resolved with better mold design and construction.

However, the defects of chief concern are those involving the nucleation of equiaxed

grains. Any recrystallization or nucleation of new grains defeats the purpose of the DS

process and results in the part being scrapped. Freckles, spurious grains, and columnar-

to-equiaxed transitions are a few of the defects that can occur. These will be discussed in

Section 2.3.

A further advancement to DS casting was the development of SX castings with Ni-

based superalloys. The same DS process is used, except that a seed or grain selector is

added. Often helical in shape, the grain selector allows the <100> direction to dominate

and eventually be the only grain growing into the component. An alternate method

involves using a previously cast SX piece as a seed. The seed is allowed to partially melt

and then the new solidification follows the growth direction of the seed. The advantage

of the seed is that it allows any direction and orientation about that direction to be grown,

though this process must be more rigidly controlled. The defects involved with SX

castings are the same as those for DS except for grain boundary cracking. The absence of

grain boundaries allows the removal of C, B, Hf, and Zr from alloy design. This

increases the incipient melting temperature of the alloy and is one of the reasons for

enhanced creep rupture life.

2.2.3. Microstructural Features

The microstructure of these DS alloys, and most solidified structures, stems from

supercooling issues. As the liquid cools, the solidification front, which is initially a flat









surface, releases its latent heat of fusion at the interface. This increases the temperature

on the liquid side of the interface, which creates a temperature inversion and results in

cellular growth. This thermal supercooling, which exists in other solidification processes,

is not as valid for DS as constitutional supercooling.12 Here, the alloying elements

partition-the solid that forms has a different composition from that of the original liquid.

The liquid at the interface contains excess solute, which shifts the equilibrium liquidus

temperature. If above the actual liquid temperature, it causes the solidification front to

break down forming perturbations that lead to the formation of cells and dendrites.

Eventually, more solute is built up in the lateral directions, forming perturbations on the

advancing cells. These secondary perturbations lead to the formation of both secondary

and tertiary dendrite arms on the already existing primary dendrite arms.12,13 The spacing

of the primary dendrite arms (k) is related to the temperature gradient (G) and growth

velocity (R) in the casting:

A =K G" Rm (1)
K is a constant specific to the alloy, and m z 0.25 and n z -0.5 for many nickel-base

alloys. 14-16

In nickel-base superalloys, Cr, W, Re, and Mo segregate to the dendrite core while

Ta, Ti, and Al remain in the interdendritic liquid.2 Thus, the dendrite arms are y, and MC

carbides begin to precipitate as cooling proceeds (usually 5-300C below the

liquidus2'17'18). The solidification concludes with the y + y' eutectic that occurs over a

small, 50C window.19 The y' precipitates in the y regions spread throughout the y

dendrites and large amounts of y + y' eutectic remains where the interdendritic fluid

solidified last.









The dendritic structure of the castings is not always perfectly aligned with respect

to the bottom of the chill. This is due to the development of curvature in the liquidus

isotherm that occurs when the conduction of heat through the casting is much less than

the conduction and radiation of heat out of the mold walls. The opposite effect occurs in

VAR ingots in that there is more heat on the outside than inside, causing a convex

upward liquidus isotherm. Since the dendrites tend to grow normal to the solidification

front (or antiparallel to the direction of heat flux), the dendrites will bend somewhat when

the isotherm curves. This is the cause of "splaying" of dendrites and can cause the part to

be rejected if the deviation from <100> is severe enough (usually around 100 from other

subgrains).

As described above, alloying elements partition and microsegregate during

dendritic solidification. Another process called macrosegregation occurs when solute is

transported over large distances. Macrosegregation is a result of convection caused by

density differences in the casting. A form of macrosegregation, called inverse

segregation, can occur in DS castings from the back flow of solute-enriched

interdendritic liquid through the interdendritic channels in the casting as it cools.20 In Al-

Cu alloys, the Cu solute can actually emanate from the chill surface and alongside the

casting.2122 This backflow of interdendritic fluid is controlled by the viscosity and

permeability of the fluid. As the primary dendrite arm spacing (PDAS) increases, the

permeability through the dendritic mushy zone will increase. The PDAS is inversely

proportional to the thermal gradient and withdrawal rate, so segregation will be favored at

low G and R.









2.3. Freckles

Freckle casting defects have been discovered in many kinds of various DS alloys

besides superalloys. They were found originally in iron-base superalloys and have been

seen in steels, solders, and various other alloys near the eutectic composition.23 They

have been observed in both castings and ingots.23-24 In all these alloys, the freckle defects

were found to be long chains of equiaxed grains parallel to the primary dendrite arms

whose composition was close to that of the interdendritic regions. They have been found

on the surface of castings and near the midradius of ingots. The number of, and even

presence of, freckle chains has been found to increase as the diameter of the casting

increases. Consistent with this observation, there is a minimum diameter or size at which

freckle chains will not form-even if conditions are ideal for freckling.

Freckle chains begin to form at a certain length above the chill in DS and SX

castings. The chains commonly converge as one moves up the casting, but divergence

does not occurs. It was observed by Giamei that the freckle chains are as deep as they are

wide-approximately 1-2 mm.23 Freckle grains are observed to be replete with

misoriented grains, eutectic, carbides, and porosity. Unfortunately, there is no

quantitative or stereological data to confirm this and most reports are based on qualitative

observations only.

In superalloys such as CMSX4, Waspaloy, Mar-M 200, and PWA 1480, the freckle

grains are purportedly enriched in those elements that partition to the liquid during

solidification, such as Al, Ta, Nb, and Ti. There is also a lack of those elements that

segregate to the dendrite core in freckles (Co, Cr, and other refractory elements). In fact,

freckling often occurs in those alloys with high concentrations of W and Re. However,









additions of Ta and sometimes Nb have been shown to reduce the propensity for freckle

formation in these castings.

Besides castings, freckles have been found in VAR processed ingots of

Waspaloy.24 The freckles in this case occur at the midradius rather than the surface. This

is one of the largest problems with freckles in ingots in that they cannot be detected, even

by ultrasound, unless they are cut open. Due to the large size of these ingots, this is an

undesirable solution. Characterization of these ingot freckles resulted in large quantities

of script carbides and r-phase with a Widmanstatten morphology. This explains the

increased hardness in the freckle areas and the propensity for fatigue and fracture

initiation sites. Even though VAR ingots eventually are forged into polycrystalline parts,

the presence of these ingot freckles greatly reduces uniformity in the ingot and can be a

source of fracture during processing.

2.3.1. Theories of Freckle Formation

From the earliest research into freckle formation, it was immediately noticed that

freckle chains would always form at a distance away from the chill, where the thermal

gradient and growth velocity of the mushy zone had decreased.23 In addition, the

compositional shift towards the eutectic composition in the freckle chains pointed to a

clue of the origin of these defects. Copley was the first to see freckle chains actually

forming by using a solution of water and NH4C1 near eutectic composition.25 The

solution was directionally solidified on a chill cooled with liquid nitrogen. As was soon

discovered, interdendritic liquid from inside the mushy zone tended to "jet" upward from

the mush into the liquid layer. The liquid would remain immiscible due partly to its

cooler temperature. Along the way, these jets would erode dendrites and break them










apart where they would redissolve into the liquid or act as nucleation sites for equiaxed

grains.

It was also shown that the jets emanate from the highest part of the mushy zone. In

castings, the mushy zone is concave upward due to the surface being cooler than the rest

of the interior. Thus, the freckle trails appear on the surface. In VAR ingots where the

mushy zone is convex upward, the jet plumes emanate to the casting interior. The jet

trails are also always parallel to the direction of gravity. Figure 5 show vectors of

channel flow calculated for a solder alloy.26 Solute from the mushy zone flows up,

forcing the liquid above the mush down into it. If a casting is directionally solidified at

an angle so that the mushy zone is tipped (since primary dendrites always grow opposite

to the direction of heat flux), the jet plumes will still grow parallel to gravity.25



Cbar Sn

10.39
10,17
9.94
9.72



Frac Liq

0.94
0,88
0.81
0.75
0.69
0.63:


0 -44


Figure 5 Isosurfaces of volume fraction and stream traces emerging from a channel

The relation of this H20-NH4C1 system to superalloy solidification was made in

later research. The freckle grains tended to have an increase in both Al and Ti, but lacked










the heavier W and Re, which were found in the dendrite core.6 Therefore, it was found

that in nickel-base superalloys, excess segregation of the heavier refractory elements into

the dendrite core and lighter y' former into the interdendritic fluid create a density

inversion. At a critical G and R, a pressure increase in the interdendritic fluid forces it

out at the highest point of the mushy zone as small jet plume. During this process, named

thermosolutal convection, the plumes break off dendrite arms as they shoot upward.6'27

The interdendritic fluid does not readily mix with the rest of the liquid due to its

difference in density and temperature between the two liquids. This interdendritic plume

eventually solidifies as a chain of small, equiaxed grains where the nuclei are purportedly

the dendrite arms removed by the plumes. Figure 6 shows a model of these plumes and

how they remain immiscible in the liquid.26








*044
i U











Figure 6 Isosurfaces of solute concentration, showing solute enrichment within
channels and the emerging plumes.

Knowing the cause, the question was then figuring out how to prevent the

formation of freckles. Copley proposed controlling the R and G. At high R, the amount









of heat released into the mushy zone must be less than what is conducted through the

solid to maintain the flow of heat out of the casting. Thus, the growth rate can be defined

as


R= (2)
AH

where KT and Gs are the conductivity and temperature gradient of the solid, respectively,

and AH is the heat of fusion.25 Under equilibrium, freckle formation is controlled by the

difference in temperature between the liquidus and solidus (AT), the minimum time

necessary to produce freckles (At), and the gradient, G. Therefore, the growth rate can be

defined as

AT 1
R =A (3)
At G

As Copley discovered, above a critical thermal gradient (G*) the mushy zone becomes

too small to accommodate the convective currents that would produce freckles (see

Figure 7).25

Other solutions, though, involved altering the alloy chemistry. The major cause of

freckle formation stems from substantial segregation of the heavier elements to the

dendrite core. Reducing the concentration of W and/or Re would reduce the propensity

to form freckles. However, in doing so the strengthening effects of these elements are

sacrificed. In castings with high Ta contents (or Ta to W + Re ratios), few, if any, freckle

chains appear. Ta partitions to the interdendritic liquid, which helps to increase its

density and hinder thermosolutal convection.6'28





























G"
THERMAL GRADIENT

Figure 7 The effect of growth velocity (R) and thermal gradient (G) on freckle
formation.

Most SX alloys contain no C since there are no grain boundaries to strengthen.

However, small additions of C have been shown to hinder freckle formation by altering

the segregation behavior of certain elements.29 It was shown that MC carbides, while

containing no W, reduced it from segregating to the dendrite core. More likely, though,

the carbides that formed increased the permeability in the interdendritic fluid by

increasing the drag force and, due to their high density, stave off the density inversion

that occurs in the mushy zone.29

2.3.2. Modeling

While a change to the chemistry of these alloys helps prevent the onset of freckling,

this may not be acceptable for all castings. As previously shown, castings with high Ta

concentrations have the tendency to form r. Therefore, returning to the fine-tuning of

processing conditions was investigated. Consequently, a large amount of effort has been









placed into modeling the situation of thermosolutal convection and what impact various

conditions have on the process. This has led to the creation of models and criteria that

attempt to describe the conditions that promote freckle formation during solidification.

The first mathematical constructions of freckle channel formation were originally

developed by Sarazin and Hellawell on a variety of solder alloys.27 These binary Pb-Sn

and Pb-Sb systems were chosen due to their simplicity (rather than the multicomponent

superalloys). Their research attempted to use permeability and thermodynamic data to

calculate various dimensionless fluid mechanics numbers. Using these numbers, one

could determine the conditions necessary for thermosolutal convection to occur. It was

discovered that there is a critical Rayleigh number below which freckling will not occur.

More recently, research has been slated towards three-dimensional modeling of freckle-

forming binary alloys.26'30

The Rayleigh number (Ra) is a dimensionless number used for convection and is

officially defined as the driving force to flow over the resistance to flow or

g'a-AT'h3
Ra = (4)
v-D

where g is gravitational acceleration, ac the coefficient of thermal expansion, v the

dynamic viscosity, DT the thermal diffusivity, and h some length or dimension.31 Above

a critical Rayleigh number (Ra*), turbulent flow will commence. There have been

differing opinions of how to define and apply the Rayleigh criterion to freckle

formation-using mathematical criteria25 27o28'32 or conservation equations.3335 Auburtin

et al. believed conservation equations cannot be practically applied to industry and

derived their Ra criterion from mathematical standards.36 Their version,










g.dp h4
Ra = (5)
v-D

uses the density inversion (dp/dz) instead of a temperature difference. One of the

difficult parts in applying this criterion to superalloy castings is how to define the

characteristic length, h. Here, Auburtin defines it as the primary dendrite arm spacing.

Beckermann et al., however, define their Rayleigh criterion through conservation

equations:37

g.Ap -K.h
0 / P
Ra = (6)
vD,

Beckermann uses the mean permeability (K) in his equation and defines the

characteristic length as the mushy zone height. It is believed that this approach is more

applicable to a large variety of casting conditions. Nevertheless, both authors have

determined how to use Rayleigh criteria to predict freckling:

* Determine the local Rayleigh number (RaMAx) from the local G and R either
through thermocouple measurements or analyzing primary dendrite arm spacings.

* Determine the critical Rayleigh number (Ra*) below which freckling will not
occur.

* Ensure that RaMAX < Ra*.

It should be noted that Ra* will only predict at what point freckling will not occur;

it cannot predict freckle formation.36,37 Even if the local Ra is higher than Ra*, freckles

may not form due to small casting cross sections, complex shapes, or small vertical

heights. Likewise, Ra* will not predict where or how many freckle chains will form.

Both authors also discovered that tilting the casting significantly decreases Ra*,

increasing the likelihood of freckle formation.25'36'37









Schneider et al. continued the work of Beckermann et al. by using the conservation

equations with several computer simulators previously used by the steel industry.35 The

simulations included tracking the concentrations of various elements in the convective

channels in castings with various thermal gradients and growth velocities, which were

combined here as cooling rate:

T= GR (7)

Most importantly, a critical PDAS was found above which the system becomes unstable

and convection occurs. For CMSX2, the critical primary dendrite arm spacing was

320 tm.35

Pollock and Murphy also correlated the formation of convective jets to the

columnar-to-equiaxed transition (CET) in DS and SX alloys. This transition is a

breakdown of the single crystal or columnar grains caused by a blockage of columnar

growth brought on by contact with the melt walls, remelting, and fragmentation of

dendrite arms, or some other mechanism.28 Freckling and CET occur under similar

conditions. In both, it was found that at low G, secondary and tertiary dendrite arms

remelt, detach, and fragment. These dendrite fragments serve as nucleation sites for new

grains. Furthermore, the detached arms also impede the advancing solidification front,

reducing R as well. This complicates the situation but allows a total transition from

single or columnar to small, equiaxed grains.














CHAPTER 3
METHODS AND MATERIALS

3.1. Materials

The materials used for this procedure were two nickel-base superalloy castings

directionally solidified from the Howmet Corporation in Hampton, VA. Due to

proprietary concerns, the composition and processing variables were not disclosed. It

was therefore a challenge to try to compare compositions of the materials and to

determine if any elements had segregated. Energy dispersive spectroscopy (EDS), which

will be discussed later, was used to ascertain the chemistries, and compare them to

similar industrial compositions from literature.

3.2. Optical Microscopy

The purpose of the optical microscopy was to examine the microstructural features,

and to measure the PDAS. The main goal was to try to discover if there was more

porosity, eutectic, and other features around the freckled region than in the rest of the

casting. Furthermore, the amounts microstructural features inside and near the freckle

grains were contrasted with the rest of the casting.

The PDAS was found using the optical microscope at low magnifications (usually

50X) on specimens cut perpendicular to the growth direction of the primary dendrite.

The PDAS was then measured from one dendrite core to the nearest diagonal dendrite

core. This process was repeated multiple times to reduce the variance in the results.









3.3. Preparation of Samples

The samples for optical microscopy were cut using a wire electrical discharge

machine (EDM). The samples were ground in sequence with 120, 240, 320, 400, 600,

800, and 1200-grit SiC polishing paper. The fine polishing was performed using a 1.0

|tm alumina suspension. The surface features were brought out by using an etchant of 80

mL hydrochloric acid, 2 mL nitric acid, 11 mL water, and 16g FeC1.38 These samples

were then viewed through the Leco Neophot 21 optical microscope.

The process of cutting the bars was difficult, especially for the DS-2 samples. A

wire EDM was chosen over a cut-off wheel because when the cutoff blade was deep into

the sample, the sample would move causing the blade to shatter. Similar problems were

encountered with the wire EDM. As the wire cut deeper into the sample, the cut metal

around the wire would close in on itself, pinching the wire off (see Figure 8). This may

have been due to residual stresses in both castings. Therefore, careful, slow, and

repetitious EDM cuts had to be made for the each desired specimen.



















Figure 8 Illustration showing the castings pinching back on itself after being cut by the
wire EDM









Another complication was that the only regions of interest in these bulky samples

was near the surface containing the freckle chains and extended only a few millimeters

into the surface. Therefore, carefully designed cuts were made in the freckled regions

from and were less than 10 mm into the sample. In order to get the best picture to

characterize the samples, cuts were made in three directions (see Figure 9):

1. Parallel to the growth direction and parallel to the casting surface

2. Parallel to the growth direction and perpendicular to the casting surface

3. Perpendicular to the both growth direction and the casting surface.

In reality, however, Cut 1 is not really a cut, but rather the surface polished and etched to

reveal the grain structure.


Cut I :C i '--
I Cut 2
I Cut 3













Figure 9 Illustration of the directions cut in relation to the primary dendrite arms (not to
scale).

3.4. Stereology

In order to quantify the microstructural features in the alloys, stereological methods

were used. Stereology involves using information from a two dimensional image to find

three-dimensional properties. This includes using point, line, and area probes and









running the data through expected value theorems to get geometric properties of the

structure.39 For example, from a count of points, one can infer the volume fraction of a

microstructural feature. Some of the probing techniques are shown in Figure 10. These

probes are then used to find properties of the microstructure (see Table 1). This project

used these techniques to determine

* The quantitative volume percentages of microstructural features inside the freckles

* The quantitative volume percentages of microstructural features in freckled regions
(so as to include the grain boundary areas)

The microstructural constituents of interest are eutectic, porosity, carbides, and r-phase.

These were contrasted with the measurements found in the rest of the casting. The

stereological measurements were taken from optical photographs as well as scanning

electron microscope (SEM) micrographs (see below).


Figure 10 Manual counting methods used for stereology









Table 1 Common stereological formulas and relations to microstructural features
Probe Count Relation Property Found
Point Count = Vv Volume Fraction
Line Intercept Count = 0.5 Sv Surface Area Density
Area Point Count = 0.5 Lv Length Density
Feature Count = Mv/27 Total Curvature
Area Tangent Count = Mv/, Total Curvature

It is important to ensure that the probe set used is isotropic, uniform, and random.39

In addition, the correct number of probes must be used to ensure proper data collection

and filter out noise. Part of the problem for this project was that the target for sampling

was in and near the freckle grains. This reduces the chance of a random set of fields

being sampled and can throw off the actual volume percent of the desired constituents.

Therefore, for this project, a set of 20 fields, each with 324 point grids were chosen

to take volume fraction measurements in random spots inside the freckled area, and in

areas in which freckles existed. These were contrasted with a sampling amount for areas

in the casting, near the surface, that showed no freckle formations. There was a

difference noted between the probes inside the freckle and those outside for those fields

that encompassed freckled and non-freckled regions. Furthermore, the results from the

position of the freckle chains were compared. If no trend can be associated with the

position, then any cross-section will be random and uniform. Otherwise, an extra

interpretive factor must be added.

3.5. SEM and TEM

Samples were prepared for SEM and transmission electron microscope (TEM)

similar to those fashioned for optical microscopy-they were cut using the EDM. The

samples were also polished with 120, 240, 320, 400, 600, 800, and 1200-grit SiC

polishing paper. Fine polishing was done with 1.0 |tm alumina suspension. An etching









solution from Van der Voort was used contained 80 mL hydrochloric acid, 2 mL nitric

acid, 11 mL water, and 16g FeC1.38 These samples were preheated at 1800F in hot water

before etching. The SEM used was a JEOL SEM 6400 equipped with Oxford Link ISIS

EDS and Oxford OPAL electron backscattered secondary electron detector (EBSD).

The samples prepared for TEM were ground and polished in a similar manner to

those for SEM and optical microscopy. However, these thin slices were ground to

approximately 100 |tm before polishing mechanically. The thin polished specimens were

pressed to form 3 mm disks. These disks were then electropolished in a solution of 50

mL perchloric acid, 100 mL of ethylene glycol, and 350 mL of methanol at -200C, 10

mV, and 1 to 5 A until a small hole appeared. The area around the hole needed to be

sufficiently thin to be electron transparent. If not, the specimens were ion milled at a low

angle (-120) to thin the regions near the hole. The TEMs used were a JEOL TEM200CX

and a Philips TEM420 equipped with an EDAX EDS system.

3.6. Electron Backscattered Diffraction

As previously mentioned, the JEOL SEM 6400 is fitted with an Oxford OPAL

system. Similar to convergent beam diffraction patterns, the EBSD technique collects

data to create a Kikuchi map, which contains crystallographic data along with

information on orientation relationships and grain misorientation. This is done when the

detector, which is a phosphor-coated screen, collects backscattered electrons and forms a

pattern of lines whose configurations are determined by the orientation and structure of

the crystal. The samples for EBSD are 10 mm wide and less than 1 mm thick and are

placed at 700 to the detector. The accelerating voltage used was 20 kV.









3.7. Nanoindentation

A Hysitron Triboindenter nanoindenter was used to determine the hardness and

moduli of different microconstituents in the castings. The Triboindenter, which is

controlled by a Triboscan transducer, works by affixing a nanotip to an electrostatically

held plate between two other plates (see Figure 9). The outer plates are fixed to

electrodes where a potential is created by adding an AC current 1800 out of phase. As the

center plate moves, the potential changes allowing for displacement measurement. Force

is measured by a DC current applied to the center plate.



Plates Electrode














Indent Tip


Figure 11 Illustration of indent head mechanism in the Triboindenter.

Using the nanoindenter, nanohardness measurements and modulus mapping could

be performed. Nanoindentation involves forcing the tip into the surface with a known

load. The depth is monitored and force is measured as the tip is released to give hardness

and relaxed modulus measurements. Multiple indents were performed on various






35


constituents in both castings with a Berkovich indenter at loads ranging from 500 to 5000

[[N. This range of loads was used to counteract problems associated at both extremes:

small loads can produce results with increased scatter due to the greater impact of noise,

and large loads can create indents too large to accurately characterize small phases.

Modulus mapping was performed by oscillating the tip to collect force and phase

measurements while also gathering topographical data.














CHAPTER 4
RESULTS

The results acquired are divided into four parts. The first discusses the SEM and

optical pictures showing and describing the microstructure and morphology of the

superalloy samples. The next section shows the stereological data of the microstructure.

The third section details the grain orientations and results found from the EBSD data.

The last section discusses data found from the nanoindenter.

4.1. Microstructural Characterization

There were two large samples obtained from the Howmet Corporation. One, which

will be herein designated DS-1, is a section off of an airfoil approximately 300 mm tall,

70 mm wide, and up to 25 mm thick (see Figure 12). The second sample set is from a

larger 1.2 m long industrial gas turbine blade, 300 mm wide, and 120 mm thick, with two

hollow cavities running through the middle (see Figure 13). This specimen was cut from

the top of the DS casting. Close examination of Figure 13 reveals freckle chains on the

side of the castings-some of these are pointed out in the figure.

The primary difference between the two castings is that the DS-1 casting had been

etched to reveal freckle grains, whereas the DS-2 castings had not-they still had their

original, undisturbed surface. While this was intriguing to study the possible surface

effects of the second casting, it also presented a problem when using the wire EDM to cut

them. Specifically, the "virgin" surface was not sufficiently conductive to be spark cut

due to the remnant oxide/ceramic that remained after casting. This was overcome by

scratching the surface to be cut and having the wire cut at precisely that location.





































Figure 12 Photograph of the DS-1 castings and a labeled freckle chain.


Figure 13 Photograph of the DS-2 castings with some visible freckle chains labeled.
Figure 13 Photograph of the DS-2 castings with some visible freckle chains labeled.










4.1.1. Compositions

The overall compositions of the DS-1 and DS-2 castings were not revealed by

Howmet due to their proprietary nature. In an effort to determine both global and local

compositional information, EDS spectra were obtained from the samples at low

magnifications in an area, not point, mode in order to accommodate as much area as

possible. These scans included the freckled regions, and were large enough to

incorporate all the phases and constituents to prevent biasing the results from the

segregation of elements. Multiple scans were run in order to minimize deviations.

Figure 14 shows an EDS spectrum from the DS-2 casting. Spectra from the DS-1 casting

were similar to the DS-2 and had the same elements present.



Counts





1000-







Crl .
500- A|



0- -UF
5 10
Energy (keV)

Figure 14 EDS spectrum of DS-2 alloy at 15 kV.

Using the atomic number, absorption, and fluorescence (or ZAF) correction

software on the EDS system, a semi-quantitative analysis was performed on the

qualitative data in the EDS spectra. This technique uses matrices, summations, and

formula integration to estimate weight and atomic percent of a sample based on peak









height, elements found, etc.40 The weight percentages for both DS-1 and DS-2 alloys,

excluding carbon, are shown in Table 2. The resulting analysis suggested that the DS-2

casting is similar to a Rene N4 alloy minus Nb and Hf, and DS-1 seemed comparable to

PWA 1483.2

Table 2 Estimated nominal compositions of DS-1 and DS-2 alloy. These compositions
are compared to that of Rene N4 and PWA 1483.
C Al Ti Cr Co Ni Mo Ta W Nb Hf
DS-1 (Small Airfoil)
xxxxx 2.9 6.1 12.8 9.3 59.1 1.8 3.2 4.8
PWA 1483
0.07 3.6 4 12.8 9 60.8 1.9 4 3.8
DS-2 (Large Pieces)
xxxxx 2.9 3.5 9.9 7.8 64 1.3 4.2 6.3
Rene N4
0.06 4.2 3.5 9.8 7.5 62 1.5 4.8 6 0.5 0.15

In addition, semi-quantitative EDS analysis was performed at a higher

magnification to contrast compositions in the freckled regions with that of the rest of the

casting (or matrix). The results, without carbon, are shown in Figure 15 for DS-1 and

Figure 16 for DS-2. For both alloys, the freckled regions were higher in Al, Ti, and Ta,

and lower in Cr, Co, W, and Mo. It should be noted that ZAF semi-quantitative analysis

is susceptible to error during the EDS collection, and from the nature of the ZAF

correction itself.







40



Compositional Differences for DS-1

14
12
1'0'

) Freckle
|) Matrix
4



Al Ti Cr Co Mo Ta W

Figure 15 Comparison of freckle chain and matrix compositions in weight percent from
DS-1 alloy without Ni.


Compositional Differences in DS-2 Casting

14J
1"
10

.c Freckle
,- Matrix




Al Ti Cr Co Mo Ta W

Figure 16 Comparison of freckle chain and matrix compositions in weight percent from
DS-2 alloy without Ni.

The freckles in the DS-1 casting about the same Al, Ti, Cr, and Mo amounts as DS-

2, but the amounts of Co, Ta, and W and were different. The Co and W differences

between the freckle grains of the two castings correspond to the compositional

differences between the two alloys as a whole. However, the freckle Ta differences are

too great to associate with composition and likely indicates substantial Ta segregation in

DS-2 to the interdendritic during solidification.









4.1.2. Surface of the Castings

The freckle chains were observed visually on the surface macroscopically, and

were studied microscopically as well. The dendritic structure was revealed in Figure 17

for DS-1 and in Figure 18 for DS-2. The primary (and secondary) dendrites are all

arranged in a <100> type direction. Here, as shown in Figure 17, the dendritic structure

of the freckle grains is more randomly oriented, and no direct relation could be

determined by visual inspection alone. The orientations were identified later using

EBSD.


Figure 17 SEM (Scanning Electron Micrograph) of etched DS-1 surface showing
dendritic structure. Freckle grains run diagonal from top left to bottom right
(35X).































Figure 18 SEM of etched surface of DS-2 casting revealing dendritic structureof freckle
chain. The chain runs left to right in the middle of the picture (50X).

As noted above, the surface of the DS-2 castings was initially unaltered after it was

broken out of the mold. This "virgin" surface (see Figure 19) exhibits charging non-

metallic phases approximately 100 to 400 |tm in size. Further inspection of this phase

(see Figure 20) in the backscattered mode shows the charging particles are of low atomic

weight, and also revealed some smaller, high atomic weight phases (see Figure 21). The

X-ray map (see Figure 22), reveals the larger charging phase to be alumina while the

smaller, brighter phase is likely (Ta,Ti)C carbide. Traces of ZrO2 were also found on the

surface, most likely left over from the ceramics mold.

It was discovered that the overall composition of the surface differed in the freckled

regions and in the rest of the casting (here called "matrix"). From Figure 23, one can see

these compositional differences. There may have been a larger percentage of alumina in

the freckled regions, or it might just reflect the results from Figure 16.



















Figure 19 SEM of unaltered DS-2 alloy surface (100X).

^4S^


Figure 20 SEM of surface of DS-2 casting showing charging alumina phase (650X).

































Figure 21 BSE (Backscattered Electron) SEM of Figure 20. The brighter areas have
elements with higher atomic numbers (high Z).


IIIMUAfF 9F;F;


igure 22 X-Ray map of Figure 21 revealing the Al enrichment in the charging phase
and the Ta and Ti enrichment in the high Z phase.











30

25

20

C m Matrix
2 Freckle

10
10

5-


C 0 Al T1 Cr Co Ni Ta

Figure 23 Comparison of EDS data from matrix and freckle regions of unaltered surface
of DS-2 alloy.

4.1.3. Eutectic Colonies and Porosity

Large amounts of eutectic were found in and near the freckle grain boundaries.

Qualitatively, the volume percent seemed larger at and around the freckles than the rest

of the casting. Quantitative data describing these differences are in Section 4.2.

The microstructure of the eutectic colonies in the current castings and in other cast

Ni-base superalloys, when cross-sectioned, appears as a fan of small y rods or plates in a

"pool" ofy'. This is shown in Figure 24. The y structure can also take up other forms,

such as a web (see Figure 25). From other planes, such as the surface cut in Figure 26,

the eutectic colony may show a different structure. In the current alloys, the eutectic is

believed to be the last liquid to freeze, and thus has the highest content of those elements

that partition away from the dendrite core (Al, Ta, Ti). Furthermore, in the DS-2

castings, there was porosity near the eutectic colonies in the freckle chain (as well as

elsewhere in the casting). This is shown in the optical micrograph in Figure 27. In the

DS-1 alloy, though, porosity is also associated with the hexagonal r, or Ni3(TaTi), phase









at the eutectic/matrix boundaries (see Figure 28). The r phase will be discussed in

Section 4.1.5.


Figure 24 SEM of eutectic colony at a freckle boundary in DS-2 alloy (1000X).


Figure 25 SEM of a eutectic colony in DS-1. Note the web-like morphology of the y
(2000X).






























- SEM of different section of eutectic in DS-2 alloy at the casting surface. The
eutectic is in the middle (2000X).


Figure 27 Optical micrograph of cross section of DS-2 casting. In the freckle chain on
the left, the dark spots are porosity next to the white eutectic colonies (50X).






























Figure 28 SEM of porosity near a colony of r at a freckle grain boundary (1000X).

4.1.4. Eutectic at Grain Boundaries

One of the more interesting solidification structures is shown in Figure 29. Nearing

the boundary between the freckle and the matrix, the y' appears to be elongated. While

certain castings have been shown to contain a film ofy' at grain boundaries, this

morphology was unexpected and was found in both alloys (see Figure 30). This

elongated structure was also seen near q and carbides at the freckle-freckle and freckle-

matrix boundaries. The DS grain boundaries, while having some carbides and eutectic at

the boundaries, never exhibited the elongated structure of y', and often there were no

other phases present besides y and y' (see Figure 31).

Besides the elongated y' morphology, blocky y' was found at the grain boundaries

involving freckles when there are no other intergranular constituents, such as eutectic, q,

or carbides, present (see Figure 32). The grain boundary y' is approximately 5 to 10































Figure 29 SEM of DS-2 alloy near surface showing elongated y' at the boundaries
between three freckles.


Figure 30 SEM of DS-1 alloy at boundary of freckle grains and matrix. The y'
"elongates" at the freckle boundary.






























Figure 31 SEM of DS boundary in DS-1 (5500X).


Figure 32 SEM showing intergranular y' at a freckle grain boundary in DS-2 (900X).









times larger than the intragranular y' and is rectangular is shape, rather than cuboidal. In

Figure 33, one can see more of the elongated y' and the blocky y' at a freckle-matrix

boundary.























Figure 33 SEM of intergranular y' between two eutectic colonies at a freckle grain
boundary from DS-2. Note the elongated y' inside the freckle grain at the
boundary (750X).

4.1.5. Eta Phase Formation

Upon investigation in the SEM, the q phase was frequently observed in one of the

castings. Figure 34 is an optical image showing some of this q phase around a freckle

grain in the DS-1 alloy. Figure 35 is an SEM image of the same freckle grain. This

phase was found primarily at the boundary between the freckle and the rest of the casting

(as shown in Figure 35), but was also found inside the freckle grain itself. Furthermore,

as stated previously, there were large eutectic regions near these freckled regions as well.









It is important to distinguish between the r phase and the carbides, since in

backscattered mode, both appeared lighter than the y + y' matrix (see Figure 36). This

distinction was made from EDS and X-ray mapping. The r phase, or Ni3(Ti,Ta), usually

has little solubility for Al or Cr. Likewise, the MC carbides have little solubility for Ni.

An X-ray dot map of Figure 36 is shown on Figure 37 and shows regions enriched in Ni

and Ti, and depleted in Cr. This is shown again in an area with more r (Figure 38), and

its X-ray map clearly shows the enrichment in Ni and Ti, as well as low Mo, Al, and Cr

amounts (see Figure 39). To differentiate the r phase from the primary carbides, an SEM

micrograph and corresponding X-ray dot maps of a primary MC carbide is shown in

Figures 40 and 41, respectively. This phase is high in Ta and Ti, while being low in Ni

and Cr, indicating it is an MC carbide. The compositions of the different phases in

Figure 36 are given in Table 3.

,2 j .. : m-3Y .Dc. Wr -- I


Figure 34 Optical image of a freckle grain in DS-1 casting showing various phases
(50X).






























Figure 35 BSE SEM of freckle boundary in the DS-1 showing r phase (500X).


Figure 36 BSE SEM showing r phase, carbides, and TCP phase at a DS-1 freckle
boundary (1900X).





























Figure 37 X-ray map of Figure 36, showing elements common to r phase.


Figure 38 SEM of r Phase (3300X).

























1 f, I


IF"" c


II I -.


I I


Figure 39 X-ray map of Figure 38, showing elements common to qr phase.


Figure 40 SEM of MC carbide in matrix (10000X)


I I- -










-7 I


IFl~-'~ -


I I' -


I I '-'' -


figure 41 x-ray map ot IVILt carDae in figure 4u.

Table 3 EDS weight percent of elements for phases in Figure 36.
Constituents Al Ti Cr Co Ni Mo Ta W
N' Carbide ,2 c 5 3 L 4 2 403 I,
Gphise 1' 2 44 S 13 34 3 5 1 5 15 5
i Phase 35 12 3 3 _3 f 4
Nlatri\x 33 5 2' 3 0 5n1 4_ 2 i,
(" + "

* No significant amount present

The qr present in the DS-1 casting appeared to be plate-like in morphology. There


also seemed to be more of the aforementioned elongated y' near the r plates at the


freckle-matrix border. From TEM micrographs, qr was found to contain stacking faults

and to be located near a carbide (see Figures 42 and 43). EDS analysis in the area of the

stacking faults showed mostly Ni and Ti, and little of the other elements. TEM also

showed stacking faults in y' that was near qr.



























Figure 42 BF TEM image of r containing stacking faults stacking faults near a carbide
(30000X).


Figure 43 BF TEM closeup of stacking faults above the carbide from Figure 42
(100000X).

4.1.6. Morphology and Microstructure of Constituents near Surface

As previously mentioned, the surfaces of the DS-2 castings were not modified prior

to investigation. This meant that the surface of these castings had been in contact with

the ceramic mold during the DS process. The actual surface was characterized as









described in Section 4.1.1. In this section, a freckle chain was cross-sectioned (Cut 2)

and examined to investigate its microstructure near the surface.

There appears to be an 8 |tm thick layer at the surface that does not contain the

cuboidal y' (see Figure 44). Near the surface, large amounts of eutectic were present, as

seen in Figure 45. Figure 46 shows another large eutectic colony with the y layer above

it. There is also a large pore and carbides present. These colonies are commonly seen in

cast superalloys, but rarely of this size. Another large eutectic area is seen in Figure 47.

In this case, a large MC carbide is located in the middle of the eutectic colony and the y'

plates appear in a starburstt" pattern emanating from the carbide. The typical matrix y +

y' structure is present between the carbide and the surface. It should be noted that EDS

was performed on all the constituents discussed above to properly identify each phase.


figure 44 sNhVI ot a cross section ot the US-2 casting surface at a treckle chain
(1400X).












-L z zCo


Figure 45 SEM of a large eutectic colony near the external surface of the DS-2 casting
(1000X).


Figure 46 SEM of large eutectic colony at surface (450X).


-- I

































Figure 47 SEM of carbide in the center of large eutectic colony near casting surface
(1800X).

Further analysis of the surface was performed and Figure 48 shows a better

resolved image of the surface y layer beside a carbide near the surface. In this figure, the

y layer shows a "smearing" feature, with the smearing angled towards the carbide. To

ensure that these waves were not entirely topological, a backscattered image was taken of

Figure 48 (see Figure 49). The smearing, or "waves," are lower in atomic number than

the surface layer, and have the same contrast as y'. A closer look at these waves can be

seen in Figure 50, where the y' appears to be smeared in one direction and then back in

another.

Unfortunately, cross-sections of the bare surface not at 900 did not reproduce this

feature. A shallow cut towards the sample at approximately 150 was performed to tie

together the images of the virgin surface with the cross-sections. Figure 51 shows a






































fIgure 48 SEM ot carbide near surface (ZSUUX).


figure 49 BSh SEM ot Mlgure 42 (23UUX).
































Figure 50 SEM of casting surface showing the "smearing" of the y + y' morphology
(3500X).

backscattered image where alumina is covering carbides and the casting. It appears that

on the unpolished surface, bare y' is seen. ZrO2 was found on top of A1203 in a few

regions, along with what appeared to be trace particles of Si. There was smooth surface

y' sparsely seen on this sample, possibly covering the bare y' from Figure 51 (see Figure

52). It is likely that since this microstructure is seen so infrequently, it is a eutectic

colony at the surface. Figure 53 shows more of the bare cuboidal y' on the surface. The

flat areas are polished. EDS of the crevices between the cuboidal y' showed higher levels

of O than the surrounding y'.


4EW,































Figure 51 BSE SEM of bare surface of DS-2 near polished regions. The dark phase is
alumina, the white phases are MC carbides, and the grey is the casting. An
unpolished part of the casting (lower right) shows bare y' (3000X).


Figure 52 BSE SEM of unpolished DS-2 surface. The surface y'layer can be seen
covering the bare y' seen in Figure 51.































Figure 53 SEM of virgin surface near polished surface. Bare cuboidal y' can be seen,
along with MC carbides that appear bright. The flat regions are polished areas
(1200X).

4.2. Stereology Results

As previously stated, there appeared to be large quantities of eutectic, carbides, and

porosity in both castings in and near the freckled regions. To quantify this observation,

stereological techniques were used to determine the volume percent of these constituents

in both castings and in the various regions in these castings. Stereological measurements

were performed at several magnifications until the lowest standard deviation in the

measurements was found-point counts for carbides were performed at 1000X in the

optical microscope and SEM, and point counts for eutectic, porosity, and r (if present)

were performed at 125X primarily on the optical microscope.

4.2.1. Stereology on DS-1 Casting

The constituents observed in the DS-1 alloy were eutectic, porosity, carbides, and r

phase. Due to the small number of freckle chains on the DS-1 castings, only Cuts 1 and 3










had significant data. Figure 54 shows the volume percent differences of eutectic in the

freckled regions of the casting. These areas include the freckled grains, grain boundaries,

and immediate surrounding regions of less than 30 |tm near the chains. Figure 55, 56,

and 57 show the stereology results of carbides, r, and porosity from the DS-1 castings.

The regions labeled "matrix" are regions at least 1 mm from any freckle chain and should

represent the rest of the DS casting.


Volume Percent Comparison of Eutectic in DS-1
Alloy at 125X

7
6
5
S4 Freckle
3E Matrix
S-3
2-
1
0
1 3
Cut #
Figure 54 Volume percent of DS-1 eutectic at 125X found through stereology.






66



Volume Percent Comparison of Carbides in DS-1
Alloy at 1000X.


4
3.5
3
2.5
2
1.5
1
0.5
0


* Freckle
* Matrix


1 3
Cut #


Figure 55 Volume percent of MC carbides in DS-1 at 1000X found through stereology.



Volume Percent Comparison of Eta Phase in DS-
1 Alloy at 125X


0 -


* Freckle
* Matrix


Cut #


Figure 56 Volume percent of DS-1 r phase at 125X found through stereology










Volume Percent Comparison of Porosity in DS-1
Alloy at 125X

2
1.8 -
1.6
1.4
m 1.2 Freckle
1E 1 T Matrix
o 0.8 -
>0.6
0.4
0.2
0
1 3
Cut #
Figure 57 Volume percent of DS-1 porosity at 125X found through stereology.

From the above figures, it is clear that there is more eutectic and r in and around

the freckle grains. There is also more porosity in these regions. Statistically, however, it

is not possible to determine if there are more or fewer carbides present in the freckled or

matrix regions. In fact, on Cut 3, there appeared to be more carbides present in the

matrix than in the freckled regions.

4.2.2. Stereology on DS-2 Casting

While there was no q in the DS-2 alloy, there were eutectic, porosity, and carbides

and it was possible to collect data from all three cuts (the same labeling conditions were

used for these stereological results). Figure 58 shows the volume percent of carbides, and

Figures 59 and 60 show the results for eutectic and porosity, respectively.






68



Volume Percent Comparison of Carbides in DS-2
Alloy at 1000X


4.5
4
3.5
3
2.5
2
1.5
1
0.5
0


* Freckle
* Matrix


1 2
Cut #


Figure 58 Volume percent of MC carbides in DS-2 at 1000X found through stereology.



Volume Percent Comparison of Eutectic in DS-2
Alloy at 125X


* Freckle
* Matrix


2
Cut #


Figure 59 Volume percent of DS-2 eutectic at 125X found through stereology.






69



Volume Percent Comparison of Porosity in DS-2
Alloy at 125X
7
6-

5-
4 -
3 E Freckle
E 3 -
M Matrix
> 2-


0-
1 2 3
Cut #
Figure 60 Volume percent of DS-2 porosity at 125X found through stereology.

The results for the microconstituents of the DS-2 castings were similar to the

stereological volume percent of those for the DS-1 alloy. The percentage of carbides

seemed higher in the freckled regions, but this could not be proven statistically.

Furthermore, the percentage of eutectic and porosity in the freckled regions was

markedly higher than in the matrix, especially in the cross sections (Cuts 2 and 3). While

the amount of eutectic present in DS-2 freckles was up to twice as high as in the DS-1

alloy, the ratio of eutectic in the freckled regions to the matrix was similar

(approximately six times).

4.2.3. Stereology for Individual Freckle Grains

One of the goals for this project was to compare and contrast the amount of the

various phases in the freckled regions with those in the matrix, but also to contrast the

volume percent of constituents found inside the freckle grains themselves (excluding the

grain boundaries and phases at the grain boundaries). However, this process proved

difficult for several reasons. First, while it was relatively simple to distinguish the freckle









chains from the rest of the casting, it was difficult to distinguish the individual grains

inside the chains. For example, the freckle grain boundaries are usually partially covered

with q (if present) and eutectic, but these phases are also present in between dendrite

arms. Freckles that can be distinguished with the naked eye do not necessarily contrast

the same way at higher magnifications in an optical microscope or in an SEM. This is

especially a problem at magnifications greater than 100X. At and near the grain

boundaries, it is difficult to determine if a phase is inside or outside a grain, or if it is

between two grains.

Therefore, individual freckle grains that could be distinguished yielded far less data

than was found in the previous battery of tests. There were simply not enough grains to

produce a sizeable amount of data. Statistically, conclusions should not be drawn from

the results. Nevertheless, the results found are shown in Figure 61 eutecticc in DS-1),

Figure 62 (r in DS-1), and Figure 63 eutecticc in DS-2). In all figures, the mean volume

percent of the individual freckle grains is essentially the same as that of the freckled

regions, but with a higher deviation. While one may hypothesize that the freckled region

and individual freckle grains have identical volume fractions of eutectic and q, it cannot

be proven in this study due to the small sampling population.

Besides quantifying volume percent of constituents in the two castings, PDAS was

measured on the same specimens with Cut 3, since this is normal to the primary growth

direction. The PDAS for DS-1 is 528 [tm (91 min) and 493 tm (+86 min) for DS-2. No

critical PDAS for freckle formation could be calculated because no castings of the same

compositions without freckles were available. However, both were larger than the

aforementioned critical PDAS for CMSX-2-320 am.35






71





Volume Percent Comparison of Eutectic in DS-1 Alloy

9
8
7
61 Freckle Region
5 r s wh i l f e Matrix
n 4 Freckle Grain
3
2
1
0

Cut #
Figure 61 Volume percent of eutectic in DS-1 alloy at 125X. This figure compares
freckle regions with individual freckle grains.


Volume Percent Comparison of Eta Phase in DS-1 Alloy


* Freckle Region
* Matrix
o Freckle Grain


1
Cut #


Figure 62 Volume percent of q in DS-1 alloy at 125X. This figure compares freckle
regions with individual freckle grains.











Volume Percent Comparison of Eutectic in DS-2 Alloy

20
18
16
14 U Freckle Region
2 12
w12 U Matrix
E 10
210 E Freckle Grains
0 8
6-
4-
2
0
2 3
Cut #
Figure 63 Volume percent of eutectic in DS-2 alloy at 125X at cuts 2 and 3. This figure
compares freckle regions with individual freckle grains.

4.3. EBSD

EBSD patterns were generated from the freckle and matrix grains in the DS-1

casting in order to determine the orientations of several different phases, and to find what

orientation relationships, if any, existed between individual freckle grains and the rest of

the casting. EBSD was first used to establish the orientation of this part of the casting,

and to relate any orientation relationships or misorientations of the various constituents in

matrix.

This task was first performed on the y + y' eutectic to ensure that it was indeed

oriented with the dendrite arms for the rest of the casting. In order to do this, the

orientation of the matrix was first found. Figure 64 shows an EBSD map of the matrix

region from Cut 1. This Kikuchi map is labeled; the green crosshairs towards the top

indicate the direction normal to the sample, and the cyan crosshairs towards the center

indicate the direction of the electron beam. The normal direction of this sample seems to









be in a <120> direction, which is close to a <110>. This result tends to agree with the

fact that the secondary dendrite arms in this part of the sample are rotated away from

perfectly orthogonal. In theory, this angle should be around 26.


Figure 64 EBSD Kikuchi map of y + y' matrix in DS-1 casting.

Next, EBSD data was compiled from various matrix points, and also various

eutectic regions. The eutectic was carefully measured by putting a point probe on each

section of a eutectic "fan." The results were similar to the matrix EBSD map in Figure

64, and were compiled as a pole figure in Figure 65. This pole figure is viewed from a

[001] normal direction where RD is the rolling direction (which is the same as the growth

direction for this sample), and TD is the transverse direction (see Figure 66).41 Besides









showing the orientation of the matrix and the eutectic, Figure 65 shows that the

orientation of eutectic is identical to that of the rest of the matrix. In moving the electron

beam from one side of a eutectic "fan" to the other (in other words, across each y' cell),

the EBSD map rotated slightly. However, this rotation was too small to be shown on a

pole figure.


Figure 65 A [001] Pole Figure of matrix and eutectic regions.

EBSD was also used to find the relationship between the r phase and the eutectic.

If a relationship exists between the r and the eutectic, then that relationship would hold

true between the r and the matrix. Figure 67 shows r to the right of a eutectic "fan."

The EBSD map of the eutectic is shown in Figure 68, and the map of the r phase is


Sample Name : 05-12-2003 Ni
Material: Nickel
Measurements : 11
RD
0.0.1} -- --






























Figure 66 Schematic of Oxford OPAL system showing directions in the pole figure
from Figure 65.

shown in Figure 69. While the orientation of eutectic was relatively fixed throughout the

sample, the r orientation changed from piece to piece. However, the basal plane of the r

(seen as the thick line on the bottom of Figure 69) matches up with the close-packed

(111) plane of the y' (similar position in Figure 68). Many directions match up as well,

such as <112> and <01 10 >, and the <011> and <1120 > directions.

Figure 70 shows a part of a freckle chain taken in the optical microscope. Figure

70 shows the same freckle chain, with the individual freckle chains color coordinated.

EBSD was performed on each of these freckle grains along with a few random spots not

in a particular freckle grain, but in the freckle chain region. The Kikuchi maps of each

grain, along with the [001] pole figure of each grain are shown as Figures 71 to Figure

90. Table 4 shows the misorientations of each grain, and the common zone axis between

the freckle and the rest of the casting.
































Figure 67 SEM of r adjacent to eutectic colony (950X).


























(111)


1 21 a-.:


Figure 68 EBSD Kikuchi map of eutectic from Figure 67.


























(0001)


Figure 69 EBSD Kikuchi map of r from Figure 67.










(b)



Freckle 9

Freckle 8








Freckle 4

Freckle 3
Freckle 2


Freckle 1 -


Figure 70 Optical image of DS-1 freckle chain. (a) The unaltered image at 50X. (b)
Individual freckle chains are numbered and colored.


-Freckle 10


-Spot C



SSpot B


-Freckle 7

-Freckle 6

- Freckle 5

-, Spot A










































Figure 71 EBSD Kikuchi map of a spot probed in Freckle 1.


Sample Name : 5-12-2103 Ni
MIater: Nickl
Ma#iuaenits: 11
HO
FID
(00.11


Figure 72 The [001] Pole figure of freckle 1 in relation to the matrix. The freckle grain
results are highlighted red








81


































Figure 73 EBSD Kikuchi map of a spot probed in Freckle 2.



Macuremieni : 11
Ho
(0,0.1 I -
1 II 2 1 ......

















1 -1 o
1-^-- --r









1 0
RD



















Figure 74 The [001] Pole figure of freckle 2 in relation to the matrix. The freckle grain
results are highlighted red.










































Figure 75 EBSD Kikuchi map of a spot probed in Freckle 3.

Sample HMae ; 05-12-203 Fleck 3
Mateial: Nickel
NMacurenenla : 10
RD
(0,0,11--


Figure 76 The [001] Pole figure of freckle 3 in relation to the matrix. The freckle grain
results are highlighted red.








83










1I




U11 2



















Figure 77 EBSD Kikuchi map of a spot probed in Freckle 4.


Sample Name : Ni 05-08
Material: Nickel
Measurements : 16
RD
(0.0.1}






TD












Figure 78 The [001] Pole figure of freckle 4 in relation to the matrix. The freckle grain
results are highlighted red.









































Figure 79 EBSD Kikuchi map of a spot probed in Freckle 5.

Sampl NH me ; 05-12-203 Fleck 1
Material: Nickel
NMacurenenla : 10
RD
(0,0,11 ~- -


Figure 80 The [001] Pole figure of freckle 5 in relation to the matrix. The freckle grain
results are highlighted red.










































Figure 81 EBSD Kikuchi map ot a spot probed in Freckle 6.


Sample Ma*e : 06-12-2003 Fleck 6
Material: Nickel
MUeawreanIn : 12
RD






1 4:- --


Figure 82 The [001] Pole figure of freckle 6 in relation to the matrix. The freckle grain
results are highlighted red.










































Figure 83 EBSD Kikuchi map of a spot probed in Freckle 7.

Sampl NHame ; 05-12-2;3 Fleck 6
Material: Nickel
NMacuremenla : 10
RD
(0,0,11--


Figure 84 The [001] Pole figure of freckle 7 in relation to the matrix. The freckle grain
results are highlighted red.