<%BANNER%>

Effect of a supersolvus heat treatment on the microstructure and mechanical properties of a powder metallurgy processed ...

University of Florida Institutional Repository

PAGE 1

EFFECT OF A SUPERSOLVUS HEAT TR EATMENT ON THE MICROSTRUCTURE AND MECHANICAL PROPERTIES OF A POWDER METALLURGY PROCESSED NICKEL-BASE SUPERALLOY By DARRYL SLADE STOLZ A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2004

PAGE 2

Copyright 2004 by Darryl Slade Stolz

PAGE 3

This dissertation is dedicated to the loving memo ry of Earl A. Stolz, Jr. and Bradford F. Gifford, Jr.

PAGE 4

iv ACKNOWLEDGMENTS I would like to thank Dr. Gerhard E. Fuch s for all of his support and guidance over the years. His knowledge of ma terials science in general, an d superalloys specifically has helped me grow tremendously as a researcher in the pursuit of my Ph.D. degree. I would also like to thank all of my committee members for their guidance during this project: Dr. Robert T. DeHoff, Dr. Mi chael J. Kaufman, Dr. John J. Mecholsky, Jr., Dr. Luisa A. Dempere, and Dr. Nagaraj K. Arakere. I w ould like to especially thank Sean Conway and Crucible Compaction Metals (Oakdale, PA) for all of the materials provided for this study. In addition, I would like to acknowledge the help of Gerald R. Bourne and the Major Analytical Instrumentation Center (M AIC) at the University of Florida for assistance with my research. I thank all of the members of the High Temperature Alloys Laboratory for their technical assistance, a nd perhaps more importantly for all of the good memories that I will keep with me after I leave Gainesville. Fi nally, I would like to thank my family for their love, support, a nd patience throughout my educational career. Without all of this help, I truly c ould not have completed this project.

PAGE 5

v TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iv LIST OF TABLES...........................................................................................................viii LIST OF FIGURES.............................................................................................................x ABSTRACT.....................................................................................................................xv i CHAPTER 1 BACKGROUND.............................................................................................................1 Composition and Microstructure..................................................................................2 Chemistry..............................................................................................................3 Gamma Matrix.......................................................................................................4 Gamma Prime Phase..............................................................................................4 Carbides and Borides.............................................................................................6 Strengthening Mechanisms...........................................................................................8 Powder Metallurgy Superalloys.................................................................................10 Powder Processing...............................................................................................13 Powder production.......................................................................................13 Powder consolidation...................................................................................16 Challenges...........................................................................................................18 Alloy 720.............................................................................................................27 Chemistry and processing............................................................................27 Microstructure..............................................................................................31 Objectives...................................................................................................................35 2 EXPERIMENTAL PROCEDURES..............................................................................37 Materials.....................................................................................................................3 7 Development of Heat Treatment................................................................................37 Heat Treatment Trials..........................................................................................39 Heat Treatment....................................................................................................41 Characterization..........................................................................................................43 Determination of ’ Solvus..................................................................................43 Differential thermal analysis........................................................................44 Metallographic examination.........................................................................44

PAGE 6

vi Microstructural Analysis.....................................................................................46 Mechanical Testing.....................................................................................................47 Low Cycle Fatigue Testing.................................................................................48 High Cycle Fatigue Testing.................................................................................54 Tensile Testing....................................................................................................55 Creep Testing.......................................................................................................56 Fracture Analysis........................................................................................................59 3 RESULTS..................................................................................................................... .60 Microstructure.............................................................................................................60 Development of Heat Treatment.........................................................................60 Heat treatment trials.....................................................................................61 Heat treatment..............................................................................................67 Characterization...................................................................................................68 Determination of ’ Solvus...........................................................................68 Microstructural characterization..................................................................72 Mechanical Testing and Fracture Analysis.................................................................75 Low Cycle Fatigue..............................................................................................75 High Cycle Fatigue..............................................................................................81 HCF results...................................................................................................82 HCF fractography.........................................................................................84 Tensile Testing....................................................................................................99 Tensile results...............................................................................................99 Tensile fractography...................................................................................101 Creep Testing.....................................................................................................105 Creep results...............................................................................................105 Creep fractography.....................................................................................107 4 DISCUSSION..............................................................................................................114 Microstructure...........................................................................................................114 Heat Treatment..................................................................................................114 Characterization.................................................................................................116 Temperature of ’ solvus............................................................................117 Microstructure............................................................................................118 Mechanical Testing...................................................................................................124 Hardness Testing...............................................................................................125 Fatigue Testing..................................................................................................126 Tensile Testing..................................................................................................139 Creep Testing.....................................................................................................143 Conclusions...............................................................................................................146 Future Work..............................................................................................................148

PAGE 7

vii APPENDIX A DIFFERENTIAL THERMAL ANALYSIS GRAPHS...............................................150 B STATISTICAL DATA...............................................................................................153 Hardness Testing......................................................................................................153 The Volume Fraction of the Gamma Prime Phase...................................................155 Gamma Prime Size Distributions.............................................................................158 Grain Size.................................................................................................................162 Stress Intensity Factors.............................................................................................163 C TENSILE FRACTURE MECHANICS......................................................................164 LIST OF REFERENCES.................................................................................................166 BIOGRAPHICAL SKETCH...........................................................................................173

PAGE 8

viii LIST OF TABLES Table page 1-1: Engine systems using forged P/M superalloys..........................................................12 1-2: Engines and airframe system s using as-HIP P/M superalloys..................................12 1-3: Nominal chemistry of Alloy 720 and Alloy 720LI...................................................27 2-1: Sieve analysis of the master pow der blends (MPBs) in weight percent....................38 2-2: Chemistry analysis of the mast er powder blends in weight percent.........................38 2-3: Heat treatme nt trial matrix.........................................................................................39 2-4: The LCF test matrix...................................................................................................54 2-5: The HCF test matrix..................................................................................................55 2-6: The Tensile test matrix..............................................................................................56 2-7: The Creep test matrix................................................................................................58 3-1: The DTA results from M&P Laboratories................................................................69 3-2: The DTA results fro m Dirats Laboratories...............................................................70 3-3: The total ’ volume fraction, and ASTM grain size..................................................74 3-4: Results from Rockwell C hard ness tests for both heat treatments............................74 3-5: The HCF test results and crack initiation types.........................................................87 3-6: Area of crack growth regions on the HCF fracture surfaces.....................................94 3-7: Tensile test results..................................................................................................... 99 3-8: The creep results......................................................................................................10 6 4-1: Stress intensity factors and plastic zone sizes for the initial crack in HCF tests.....132 4-2: Stress intensity factors and plastic z one sizes for the final crack in HCF tests.......134

PAGE 9

ix B-1: The hardness test ing statistical data........................................................................153 B-2: The t-test data for the hardness testing...................................................................154 B-3: The primary ’ volume fraction data for the standard heat treatment.....................156 B-4: The secondary ’ volume fraction data for the standard heat treatment.................156 B-5: The tertiary ’ volume fraction data for th e standard heat treatment......................157 B-6: The primary ’ volume fraction data for the alternate heat treatment....................157 B-7: The secondary ’ volume fraction data for the alternate heat treatment.................158 B-8: The primary ’ size data for the standard heat treatment........................................159 B-9: The secondary ’ size data for the standard heat treatment....................................159 B-10: The tertiary ’ size data for the standard heat treatment.......................................160 B-11: The primary ’ size data for the alternate heat treatment......................................161 B-12: The secondary ’ size data for the alte rnate heat treatment..................................161 B-13: The grain size data for the standard heat treatment..............................................163 B-14: The grain size data for the alternate heat treatment..............................................163 B-15: The t-test data for the initi al crack stress intensity factors...................................163 C-1: Area of crack in fractured tensile specimens..........................................................164 C-2: Stress intensity fa ctor for tensile tests.....................................................................165

PAGE 10

x LIST OF FIGURES Figure page 1-1. Nickel-aluminum phase diagram................................................................................5 1-2: The Ni3Al solid solution field at approximat ely 1150C for various ternary alloys..6 1-3: Flow stress as a f unction of temperature for Ni3Al....................................................7 1-4: Effect of valency difference on hardeni ng of nickel alloys. Nv is electron vacancy number of the solute...................................................................................................9 1-5: Vertical gas atomizer................................................................................................14 1-6: Formation of spherical metal powder by gas atomization........................................15 1-7: Material and fabrication saving s with P/M processing of superalloys.....................18 1-8: Plastic strain suffered by smaller a nd larger particles duri ng HIPing of a bimodal particle size distribution of powders........................................................................20 1-9: PPB initiation site in doped Ren 95........................................................................21 1-10: Type 1 ceramic inclus ion fatigue initiation site......................................................22 1-11: Comparison of average LCF lives of HIP vs. HIP + Forge and Extrude + Forge Ren 95.....................................................................................................................24 1-12: Fatigue initiation at a crystallographic defect.........................................................25 1-13: Average low cycle fatigue life for a P/M superalloy during 1980 through 1996....25 1-14: Effect of thermal exposure on microstr uctures of subsolvus h eat treated Alloy 720 and Alloy 720LI.......................................................................................................29 1-15: Experimental TTT diagram for the form ation of 0.5 and 1.0 wt% of sigma in Alloy 720LI........................................................................................................................30 1-16: Mean diameter of the cooling ’ as a function of the inte rrupt temperature in Alloy 720LI........................................................................................................................32 1-17: P/M Alloy 720 microstructure.................................................................................34

PAGE 11

xi 1-18: Optical micrographs of Astroloy.............................................................................35 2-1: JEOL JSM 6400 SEM.............................................................................................42 2-2: Carbolite box furnace...............................................................................................45 2-3: Tensile sample design...............................................................................................49 2-4: Creep sample design.................................................................................................50 2-5: Low cycle fatigue sample design.............................................................................51 2-6: Instron-Satec se rvo-hydraulic test frame..................................................................52 2-7: Instron-Satec Creep frame........................................................................................57 3-1: Heat treatment trial sample s with various soak temperatures..................................62 3-2: Heat treatment trial samp les with different cooling rates.........................................63 3-3: Heat treatment trial sa mples with different soak times and final temperature before fan-air-cool...............................................................................................................64 3-4: Heat treatment schedules..........................................................................................65 3-5: The standard and alternat e heat treatment microstructures......................................66 3-6: MPB 96SW865 after solution heat treating at 1175C (2147F).............................66 3-7: MPB 96SW865 after solution heat treating at 1165C (2129F).............................67 3-8: The ’ solvus trials quenched in iced br ine after an hour at the solutioning temperature...............................................................................................................71 3-9: The solvus trials that were fu lly solutioned and then control-cooled.......................72 3-10: The / ’ microstructure in heat treated Alloy 720LI...............................................73 3-11: Grain boundary microstructure of heat treated Alloy 720LI...................................74 3-12: Low cycle fatigue sample that wa s misaligned and tested in compression.............76 3-13: Push/pull rods.......................................................................................................... 76 3-14: Low cycle fatigue sample with notches machined just after th e threaded region and before the shoulder of the sample............................................................................77 3-15: Ceramic inclusion type defect in sta ndard heat treatment sample tested at 538C (1000F) and a strain range of 1.1%........................................................................79

PAGE 12

xii 3-16: Standard heat treatment sample test ed at 538C (1000F) and a strain range of 1.0%..........................................................................................................................7 9 3-17: Alternate heat treatment sample test ed at 538F (1000C) and a strain range of 1.0%..........................................................................................................................8 0 3-18: Relatively flat featureless region just after crack initiation in low cycle fatigue....81 3-19: Representative micrographs of the fast fracture region of LCF samples................82 3-20: High cycle fatigue S-N curves for specimens tested at 538C (1000F)................83 3-21: High cycle fatigue S-N curves for specimens tested at 649C (1000F)...............84 3-22: High cycle fatigue S-N curves for test s of the standard heat treatment at both 538C (1000F) and 649C (1200F).......................................................................85 3-23: High cycle fatigue S-N curves for test s of the alternate heat treatment at 538C (1000F) and 649C (1200F)..................................................................................86 3-24: Optical micrographs of HCF fracture surfaces of samples tested at 649C and 1086 MPa..........................................................................................................................86 3-25: Ceramic agglomerate as in itiation point in standard heat treatment sample tested at 538C (1000F) and a stress range of 1034 MPa (150 Ksi).....................................87 3-26: EDS spectra of crack in itiation for standard heat trea tment sample tested at 538C and 1034 MPa...........................................................................................................88 3-27: Ceramic agglomerate as initiation point for standard heat treatment sample tested at 538C (1000F) and a stress range of 1086 MPa (157.5 Ksi)..............................88 3-28: Slip band cracking as in itiation point an alternate heat treatment sample tested at 538C (1000F) and 993 MPa (144 Ksi)..................................................................89 3-29: Primary carbide as the in itiation type for the standard heat treatment sample tested at 649C (1200F) and 1086 MPa (157.5 Ksi).........................................................89 3-30: EDS spectra of Ti-rich particle as initi ation point of standard heat treatment sample tested at 649C and 1086 MPa.................................................................................90 3-31: Unknown initiation type for alternat e heat treatment sample tested at 649C (1200F) and 1034 MPa (150 Ksi)...........................................................................91 3-32: Crack origin for alternate heat trea tment sample tested at 538C (1000F) and 1086 MPa (157.5 Ksi).......................................................................................................91 3-33: Initial crack regi on of fatigue samples....................................................................92

PAGE 13

xiii 3-34: Fracutre surface near initiation in stan dard heat treatment sample tested at 538C (1000F) and a stress range of 1034 MPa (150 Ksi)................................................93 3-35: Fracture surface of standard heat tr eatment tested at 538C (1000F) and a stress range of 1086 MPa (157.5 Ksi) outside of discolored region.................................93 3-36: Fracture surfaces far from origin of samples tested at 538C (1000F) and 993 MPa (144 Ksi)..........................................................................................................94 3-37: Initial crack area versus stress ra nge for the high cycle fatigue specimens............95 3-38: Final crack area vers us stress range for the hi gh cycle fatigue specimens..............96 3-39: Secondary cracks at primary and secondary ’ precipitates in samples tested at 538C and 993 MPa.................................................................................................97 3-40: Secondary cracks in fatigue samples tested at 1034 MPa.......................................97 3-41: Fatigue samples tested at 538C and 1086 MPA showing secondary cracks.........98 3-42: Yield strength vs. temper ature for both heat treatments........................................100 3-43: Elongation at failure vs. temp erature for both heat treatments.............................101 3-44: Tensile stress-strain curve for standa rd heat treatment samp le tested at 1400F..102 3-45: Tensile stress-strain cu rve for alternate h eat treatment sample tested at 1400F..102 3-46: Crack initiation re gion in standard heat treatment tensile samples.......................103 3-47: Crack initiation region in alternate heat treatment tensile samples treatment.......104 3-48: Larson-Miller plot for creep rupture in Alloy 720LI.............................................107 3-49: Fast fracture propagation path for sta ndard heat treatment sample tested at 677C (1250F) and 1034 MPa (150 Ksi).........................................................................108 3-50: Fast fracture propagation path for a lternate heat treatment at 704C (1300F) and 689MPa (100Ksi)...................................................................................................109 3-51: Standard heat treatment sample te sted at 760C (1400F) and 483 MPa (70 Ksi) showing fast fracture propagation path..................................................................109 3-52: Standard heat treatment sample test ed at 677C and 1034 MPa with crack initiation at a grain boundary triple point..............................................................................110 3-53: Transgranular cracking during fatigue testing in samples tested at 677C and 1034 MPa........................................................................................................................111

PAGE 14

xiv 3-54: Large secondary crack in alternate heat treatment sa mple tested at 677C and 1034 MPa........................................................................................................................111 3-55: Secondary cracks in high temperature creep tests.................................................111 3-56: Grain boundary crack initiation in al ternate heat treatment creep samples...........112 3-57: Secondary cracks in creep samples tested at 677C (1250F) and 793 MPa (115 Ksi).........................................................................................................................11 3 4-1: The tertiary and “quaternary” ’ precipitates in the standard heat treatment.........119 4-2: Serrated grain boundary structure present in the alternate heat treatment microstructure.........................................................................................................120 4-3: Histogram of ’ volume fraction for the two heat treatments..................................121 4-4: Secondary and tertiary ’ precipitates....................................................................122 4-5: Comparison of fast fracture feat ures during HCF fatigue and grain size...............131 4-6: Secondary cracking in standard heat treatment fatigue samples............................135 4-7: Secondary cracks propagating transgranul arly in alternate heat treatment fatigue samples...................................................................................................................136 4-8: Large pore at a grain bounda ry triple point in alternate heat treatment sample tested at 538C and 1086 MPa.........................................................................................136 4-9: Long secondary cracks in alternate he at treatment fatigue sample tested at 538C and 1086 MPa.........................................................................................................137 4-10: Yield strength vs test temperature for both heat treatments.................................140 4-11: Optical pictures showing the fr acture surfaces of tensile samples........................141 4-12: Alternate heat treatment creep sa mple tested at 732C and 483 MPa showing secondary cracks initiating at and propagating along grain boundaries.................144 4-13: Alternate heat treatment creep samp le tested at 677C and 1034 MPa showing void nucleation at a grain boundary triple point and at the interface between secondary ’ precipitates and the matrix.....................................................................................145 A-1: The DTA Graph of sample UMBCa......................................................................150 A-2: The DTA graph of sample UMBCb......................................................................150 A-3: The DTA graph of sample UMBEa.......................................................................151

PAGE 15

xv A-4: The DTA graph of sample UMBEb......................................................................151 A-5: The DTA graph of sample UMBCc......................................................................152 A-6: The DTA graph of sample UMBEc.......................................................................150

PAGE 16

xvi Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy EFFECT OF A SUPERSOLVUS HEAT TR EATMENT ON THE MICROSTRUCTURE AND MECHANICAL PROPERTIES OF A POWDER METALLURGY PROCESSED NICKEL-BASE SUPERALLOY By Darryl Slade Stolz August 2004 Chair: Gerhard E Fuchs Major Department: Materials Science and Engineering Powder Metallurgy (P/M) processed nickel -base superalloys are used as turbine disk materials in jet engines. The P/M processing resu lts in a homogenous microstructure. Large amounts of strengtheni ng elements can be incorporated into the chemistry of these P/M alloys. In addition, the ability to pr oduce near net-shaped parts with powder consolidation may of fer the potential for large cost savings. However, the fatigue properties of P/M superalloys in th e as-consolidated form have suffered because of the defect sensitivity of the as-cons olidated microstructure. Expensive, thermomechanical steps are necessary to break down defects, so that the P/M parts can be considered defect-tolerant. As a result, th e true potential cost savings for using P/M superalloys in turbines have never been realized. This program was undertaken to examine the potential for utilizing an alternate heat treatment with P/M Alloy 720LI to generate a potentially defect-tolerant

PAGE 17

xvii microstructure. This heat treatment had a soak above the ’ solvus temperature followed by a controlled cool through th e solvus. This produced grains with a regular array of large dendritic-shaped secondary ’ within the grains. Mechanical testing was carried out to fully evaluate the effect of this alternate heat treatment on the mechanical properties of Alloy 720LI. The standard heat treatment had longer lifetimes at the lower stress range conditions during high cycle fatigue; however, the alternate heat treatment was supe rior at the highest stress range. Fracture analysis suggests that this is due to the grain size differen ce. During tensile testing, the standard heat treatment had higher yield and ultimate strengths but lower ductility than the alternate heat treatment. This is thought to be due to th e larger amounts of tertiary ’ present in the microstructure produced by the standard heat treatment. Finally, the standard heat treatment had longer creep lifetim es at the lowest test temperature. The alternate heat treatment performed better at the higher test temperatures. While the microstructure did not improve the fatigue properties across the board, the improved understanding of the microstructural evolut ion during heat treatment will help in developing new heat treatments that may provide the defect -tolerance that is necessary.

PAGE 18

1 CHAPTER 1 BACKGROUND Superalloys have been used extensively as high temperature materials in turbine engines since the 1950s [1]. In addition to jet engines and industrial gas turbines, they have seen service in space vehicles, rocket engines, nuclear reactors, submarines, steam power plants, petrochemical equipment, he ating elements, and furnace parts [2, 3]. Superalloys are generally defined by their unique ability to retain high strength at temperatures approaching 90% of their me lting point and for times up to 100,000 hours at slightly lower temperatures [4]. This unique feature for a structural material coupled with the high oxidation resistance, makes supera lloys the best choice for very demanding environments like jet engines a nd industrial gas turbines. There are three main classes of supera lloys: nickel-base, cobalt-base, and nickel-iron-base. However, Ni-base alloys generally are considered the most important, and are the most widely used of these cl asses, owing to their excellent blend of mechanical and chemical properties. As turb ine engine materials, they are required to have high strength as well as good creep, fatigue and corrosion resistance [3]. The first jet engine developed by Sir Frank Whittle opera ted with a thrust-to-weight ratio of 1.5:1 [1]. Modern commercial engines operate at a ratio of 6-8:1 w ith advanced engines approaching 10:1. The increases in propulsion output in jet engines have been made possible by improvements in materials as well as engine design. The demand for more efficient and powerful turbines has steadily increased the operating temperature and mechanical stress demands for superalloys. Ni-base superalloys have met these demands

PAGE 19

2 through improvements in alloy processing and chemistry. Turbine blades have the highest operational temperatures of the supe ralloys used in jet engines, making creep resistance vital. Single crystal alloys with excellent creep resistance have been developed as blade alloys in current jet engines. On the other hand, turbine di sks operate at higher loads but somewhat lower temperatures. They require good fracture toughness, low crack growth rate, and ease of inspectability [5]. Turbine disk designers have examined powder processing as a means of producing disks with higher strength and better fatigue resistance. These alloys have an excellent blend of properties; how ever, their high costs have limited their usage in turbine engines. Composition and Microstructure The alloy chemistry of Ni-base superall oys has evolved tremendously over the years. The initial alloys used in ga s turbine engines were modifications of oxidation-resistant stainless steels [6]. The addition of aluminum and titanium to the Nimonic (80% Ni, 20% Cr) alloy series led to the precipitation of the intermetallic phase, Ni3Al. These were the first pr ecipitation-hardenable supera lloys. These alloys have higher strength than alloys strengthened so lely by solid-solution hardening, and than those strengthened by oxide dispersion stre ngthening [3]. Precipitation-hardenable superalloys are preferred for the most dema nding environmental conditions. The first superalloys had relatively simple compositi ons; however current generations can be a confusing blend of varied elements that are added to improve processing, mechanical properties, oxidation resistance, and even density. Even though there has been a tremendous amount of research into the effect of chemistry change s on the properties of these alloys, much still remains to be lear ned about the complex interactions between

PAGE 20

3 various elements and the effects that slight changes in composition will have on the phase stability and mechanical prope rties of the various alloys. Chemistry Precipitation hardenable Ni-base superall oy phase systems consist of a nickel matrix, gamma ( ), which is strengthened both by so lid-solution alloying elements and by the intermetallic precipitate, gamma prime ( ’). Aluminum and titanium are the primary elements added to increase the amount of ’ in the alloy [4]. Niobium and tantalum can also be added as substitutional elements for Al and Ti. These ’ forming elements are added in amounts up to 8%. The most common solid-solution strengthening elements for the matrix are cobalt, iron, chromium, molybde num, tungsten, titanium, and aluminum. While titanium and aluminum are added mostly as precipitation hardeners, they do help strengthen the matrix by solid-solution substituti on. Ni-base superalloys generally contain 10-20% chromium, as it forms a pr otective oxide scale in addition to the solid-solution strengtheni ng benefits. Aluminum also forms a protective Al2O3 scale for oxidation and hot corrosion resistance. Between 5 and 15% Co is added because of its effect on the precipitation behavior of ’. The addition of Co e ffectively alters the slope of the ’ solvus curve, which changes the ’ solvus temperatur e, depending on the ’ volume fraction in the alloy. Add itionally, it increases the amount of ’ that will precipitate out of solution during heat treatment. Small amounts of carbon, boron, zirconium, and hafnium are added to impr ove grain boundary streng th. Finally, it is important that the amounts of so called “tramp” elements (like silicon, phosphorus, sulfur, oxygen, and nitrogen) as well as other minor elements are ca refully controlled to very small levels.

PAGE 21

4 Gamma Matrix The continuous matrix phase, is a face centered cubic (FCC) Ni-base phase that is austenitic in structure [4]. Pure nick el has neither a high elastic modulus nor a low diffusivity, which are necessary to promote creep rupture resistan ce. Its ability to be used at elevated temperatures for long times is attributed to the following three reasons: Nickel has a nearly filled third electron shell, which allows for high alloying content. It has the tendency to form the protective Cr2O3 oxide scale with additions of chromium. It has an added inclination to form the oxidation-resistant Al2O3 scale. The matrix is strengthened primarily by a lloying with solid-solution elements. Gamma Prime Phase The precipitate phase, ’, is an ordered L12 intermetallic phase that is stable over a fairly narrow compositional window (Figure 11) [7]. Its nominal composition is Ni3Al; however, both the nickel and the aluminum can be substituted for, by various alloying elements [4]. Figure 1-2 [8] shows how various elements substitute in NiAl. As can be seen, cobalt and copper will substitute for ni ckel. However, titanium, silicon, vanadium, and manganese all partition to the aluminum s ites. Molybdenum, chromium, and iron are equally likely to substitute for nickel as th ey are for aluminum. Nickel atoms occupy the face-centered sites and aluminum atoms the corn er sites [8]. More recent studies have found that niobium and hafnium also part ition to the aluminum sites [4]. Ni3Al exhibits long-range order up to its melting point, and pr ecipitates coherently with the austenitic matrix [3, 7]. In early superalloys, this pr ecipitate was spherical; however as alloying

PAGE 22

5 Figure 1-1. Nickel-aluminum phase diagram. Reprinted with permission from N. S. Stoloff, “Wrought and P/M Superalloys,” ASM Handbook 1 ASMInternational, 2003, Figure 1, p. 952. content was increased, cuboidal precipitates we re observed [3]. It was determined that the change in ’ morphology was related to the lattice mismatch between and ’. Spherical precipitates were noticed when the mismatch was between 0.0 and 0.2%. As the mismatch increased to between 0.5 and 1.0%, cuboidal precipitat es were observed. Finally, plate-like ’ were present when the mismatch reached 1.25% and above. Perhaps

PAGE 23

6 Figure 1-2: The Ni3Al solid solution field at approxi mately 1150C for various ternary alloys. Reprinted with permission fr om R. W. Guard and J. H. Westbrook, “Alloying Behavior of Ni3Al ( ’ phase),” Transactions of th e Metallurgical Society of AIME 215 1959, AIME, Figure 4, p. 810. the most important property of ’ is that its yield strength increases with increasing temperature, making it ideal fo r high-temperature applica tions (Figure 1-3) [9]. Additionally, ’ is generally not considered a frac ture-initiation site, because of its inherent ductility [4]. Carbides and Borides Carbides and borides precipitate in Ni-bas e superalloys in various forms including MC (metal carbide), M23C6, M6C, and M3B2. The role that they play in the microstructure and mechanical properties of superalloys is fairly complex. In some cases, their presence can be detrimental; however, other forms are beneficial to the mechanical properties. While they can lo wer the ductility, their presence can also

PAGE 24

7 Figure 1-3: Flow stress as a function of temperature for Ni3Al. Reprinted with permission from R. G. Davies and N. S. Stoloff, “On the Yield Stress of Aged Ni-Al Alloys,” Transactions of the Metallu rgical Society of AIME 233 1965, AIME, Figure 4, p. 717. increase the creep rupture life and the chemical stability of the matrix [4]. Davis et al. [3] and Ross and Sims [4] describe different t ypes of carbides. MC carbides generally precipitate as discrete random cubic or script particles, along grain boundaries and within the grains. They are FCC and have little or no orientation relationshi p to the matrix. In superalloys, their preferred order of stability is HfC, TaC, NbC, and TiC (with HfC being the most stable). The M23C6 carbides form primarily as discontinuous, blocky particles with a complex cubic structure, during h eat treatment or service between 760 and 980C from the degeneration of MC carbides and carbon left in the matrix. They are generally found along grain boundaries, but can also form on twin bands, stacking faults, and twin ends. The formation reaction is MC + M23C6 + ’ These carbides form mostly in alloys w ith at least moderately high chromium contents. With the presence of tungste n and molybdenum, they form with the

PAGE 25

8 composition Cr21(Mo, W)2C6. Finally, M6C carbides precipitate in blocky form along grain boundaries, or sometimes as Widmansttte n plates within the grains [3]. These carbides also have a complex cubic structur e. They form between 815 and 980C when the molybdenum and tungsten contents are highe r than 6-8 atomic percent. They are stable at high temperatures, and thei r composition can range widely. The M3B2 borides precipitate as hard particles, with a wide variety of shapes from blocky to half-moon. They have a tetragonal unit cell, and act to improve grain boundary strength. Strengthening Mechanisms As stated earlier, the two main modes of strengthening in Ni-base superalloys are solid-solution strengthening and precipitation hardening. Many mechanisms contribute to strengthening from the substitution of so lid-solution elements: size misfit, modulus misfit, stacking-fault energy, and short-range or der. The size of the substitutional atom was shown to be an important factor in so lid-solution strengtheni ng [10]. The solute atoms have small elastic stress fields that can impede dislocation motion [11]. However, the model by Mott and Nabarro [12] to account for size misfit overestimates the strengthening effect of solute atoms. The vale nce state of the solute atom also affects the amount of solid-solution strengthening for a pa rticular solute element [13, 14]. Pelloux and Grant [15] confirmed these findings for superalloys (Figure 1-4). These valency effects have been explained as being due to the modulus differences among different alloys [16]. Local variations in the modul us can attract and pin dislocations [11]. Stacking-fault interactions are an additiona l solid-solution strengthening mechanism that was first discovered by Suzuki [10, 11]. Solu te atoms can preferentially segregate to stacking faults. In turn, the stacking-fault en ergy is lowered, and the separation between

PAGE 26

9 Figure 1-4: Effect of valenc y difference on hardening of nickel alloys. Nv is electron vacancy number of the solute. Reprinted with permission from R. M. Pelloux and N. J. Grant, “Solid Solution and Second Phase Strengthening of Nickel Alloys at High and Low Temperatures,” Transactions of th e Metallurgical Society of AIME 218 1960, AIME, Figure 5, p. 234. partial dislocations is increased. The motion of these partials is much more difficult, as a result. Finally, short-range ordering can occu r in some nickel solid-solutions [12]. A dislocation moving through a region with shor t-range order will require more work for propagation, as it locally di srupts the energetically favor able ordered state [11]. Strengthening from precipitates can be c onsidered as additive to solid-solution strengthening. Precipitation of ’ in superalloys causes stre ngthening due to coherency strains between the matrix and precipitate, and the presence of order in the particles [12]. The strain field locally around each precipitate provides a repe lling force for dislocation motion. The mechanisms of order strengthenin g are controlled by the type of interaction between the dislocations and the precipitat es. Below a critical precipitate size, dislocations will shear the ’ particles. The ordered prec ipitate is left out of phase, creating an antiphase boundary (APB) in the precipitate. This extra force needed to create APB’s in the alloy results in order strengthening. The size and spacing of the ’

PAGE 27

10 precipitates has a large impact on the amount of strengthen ing from ordered particles [11]. However, if the preci pitates are incoherent and la rger than a critical size, dislocations will bypass them. For superalloys this generally happens according to the Orowan bowing model. As a dislocation appr oaches the large precip itates, it begins to bow around them. The dislocation eventual ly bypasses the par ticles, leaving a dislocation loop that exerts a stress on approaching dislocati ons. This stress field from the dislocation loops results in an in crease in the strength of the alloy. Powder Metallurgy Superalloys Powder metallurgy (P/M) superalloys were first examined as a novel way of cooling turbine blades through transpiration cooling [17]. A mixt ure of elemental and master-alloy powders were cold-compacted a nd sintered. However, it was only possible to achieve approximately 90% of the theoreti cal density. As a result, these alloys had inadequate strength for use as turbine blad es. In the early 1960s, pre-alloyed powders were developed through the use of water at omization [18]. However, these alloys suffered from poor fatigue strengths in comp arison to the conventi onal wrought alloys. Eventually, with the use of vacuum melting and improved powder cleanliness in the mid 1960s, it was possible to produce P/M superalloys with sufficient mechanical properties, that were a potential alternative to c onventional cast and wr ought alloys [17]. As turbine manufacturers sought more effi cient jet engines in the early 1970s, the property requirements for turbine disk materi als increased beyond the capabilities of the existing alloys [19]. The new alloys were required to have a high strength-to-weight ratio, good hot workability, and higher operationa l temperatures. To meet these needs, alloy designers began developi ng cast and wrought alloys wi th higher concentrations of solid-solution strengthening elements, as well as ’ forming elements (aluminum and

PAGE 28

11 titanium) [20, 21]. However, these all oys experienced major difficulties with segregation, hot workability, and ductility defi ciencies as the alloying contents and ingot size necessary for larger turb ine disks increased [19, 22]. These newer cast and wrought alloys were practically unforgeable [21]. With pre-alloyed powder processing, more rapid solidification rates were easily possible. These faster solidification rates resulted in much smaller dendrite-arm spacing within the po wder particles, as opposed to those seen in conventionally cast ingots [23]. The increase in hom ogeneity for these alloys by powder processing makes them easy to fo rge isothermally through superplastic deformation [19, 22]. Powder-processed disk alloys provide many advantages over disk alloys processed by conventional ingot metallurgy [17, 20, 24]. Reduced segregation in structure that allows for more heavily alloyed compositions. Improved hot workability after compaction. Near net-shape production of parts that has potential for cost-savings and conservation of rare and expensive elements. Finer grain size. Reduced carbon segregation. Ability to produce unique structures like oxide dispersion-strengthened (ODS) alloys. With all of these advantages, P/M superalloys were first put into operational service in 1974 in the Pratt & Whitney F100 engine on the F-15 Eagle fighter jet [22]. A survey of engines in use in 1996 shows that powder superalloys have become widely used in both commercial and military jets since their inception. Engines that use P/M alloys

PAGE 29

12 after extrusion and isothermal forging are shown in Table 1-1; while Table 1-2 shows engines that incorporate as-hot isos tatically pressed (as-HIP) alloys. Table 1-1: Engine systems using forged P/M superalloys.* Engine Number produced through 1996 GE T-700 9674 GE F-404/414 3077 GE-F110 2259 GE 90 38 PW F100 6496 PW 2000 951 PW 4000 1819 *Reprinted with permission from J. H. Moll and B. McTiernan, “Powder Metallurgy Superalloys,” ASM Handbook, 7 ASM-International, 2003, Table 1. Table 1-2: Engines and airframe sy stems using as-HIP P/M superalloys. GE Aircraft Engines CFM International AlliedSignal Turbine Airframe Turbine Airframe Turbine Airframe CF680C2/E1 B747-400, 767 AWACS CFM563B1/B2 B 737-300 GTC-1313[A] B2 F108-CF100 KC-135-R CFM563B2/3C B 737-400 GTC-1319[D] MD-90 F110-GE100 F-16 C/D/G CFM56-3B4 B 737-500 GTC-1319[B] B737-X F110-GE129 F-16 D/G CFM56-5A1 A-320-100/200 GTC-1319[A] Airbus A319, A320, A321 F110-GE400 F-14B CFM565A/5B A-319 GTCP-331200 B757/B767 B747-200B F118-GE100 B2 CFM565B1/B2 A-321-100 GTCP-331250 A300, A310, C17A F404-GE400/402 F/A-18 C/D, F-117A CFM565C2/C4 A-340200/300 GTCP-331350 Airbus A330, A340 T-700-GE401 Bell AH 1W, Sikorsky SH60B/F CFM56-7 B737-700 GTCP-331500 B777 T-700-GE700 Sikorsky UH-60A/L CFM56-8 B 737600/800 RE-220 Gulfstream V Global Express CRJ700 *Reprinted with permission from J. H. Moll and B. McTiernan, “Powder Metallurgy Superalloys,” ASM Handbook 7 ASM-International, 2003, Table 2. The initial P/M processed superalloys we re derived through slight modifications to the chemistries of existing alloys like IN-100, Astroloy, an d Ren 95 [22]. Eventually,

PAGE 30

13 alloys like AF115 and AF2-1DA were designed specifically for P/M processing, to take advantage of the high level of alloying that can be accommodated by this route. However, these alloys had low resistance to crack growth from defects in the microstructure. With defect tolerance in mind, current research has centered on more defect-tolerant alloys like N18, Ren 88DT, and Udimet 720. Powder Processing A wide range of powder processing met hods is available to a turbine disk manufacturer. These include powder pr oduction, consolidation, post-consolidation thermo-mechanical processing (TMP), and heat treatment. All of these different processing routes give the manufacturer the flexibility to alter the microstructure and, therefore, the mechanical properties to fit the requirements for almost any particular application. Powder production Many different powder fabrication technique s are used for the production of metal powders. These include mechanical fabricat ion like machining or milling, electrolytic fabrication, chemical fabrication, and atomizatio n [25]. Of all these different techniques, gas atomization to produce pre-alloyed powder is the most important for the production of P/M superalloys [17]. Within the cat egory of atomization, there are a few subcategories of powder fabrication proce sses including inert-gas atomization, soluble gas atomization (vacuum atomization) centrifugal atomization including rotating-electrode pro cess (REP) and electron-beam rota ting process (EBRP), and rapid solidification techni ques [17, 20, 22]. The REP process has received some attenti on from turbine disk manufacturers [26]; however, the most research into powder fabric ation of superalloys has examined inert-gas

PAGE 31

14 atomization. Typical cooling rates for iner t-gas atomization of superalloys are in the range of 104 to 106 C/sec [27]. A typical gas atom ization rig is shown in Figure 1-5 [25]. In this process, vacuum induction melting (VIM) is used to melt high-purity raw Figure 1-5: Vertical gas atomizer. Reprinted with permission from Powder Metallurgy Science 1994, Metal Powder Industries Federation, 105 College Road East, Princeton, New Jersey, USA. Figure 3.14, p. 101. materials in a crucible at the top of the atomization rig [22]. The melted alloy passes through the nozzle in a thin st ream. High-pressure argon gas is then impinged the stream, to break up the melt stock into fine spherical droplets. Alternatively, nitrogen gas can be used; however, it is necessary to alter the alloy composition to account for any increase in nitrogen and the effects that will have on th e final alloy. These droplets solidify as they descend through the cooling towe r to the bottom of the atom ization rig, where they are collected for consolidation. The fine particles are collected in a side unit, using a cyclone separator.

PAGE 32

15 The particle size and size distribution can be affected by many variables during the atomization process. Generally, the mo re energy that is imparted onto the molten metal by the gas, the finer the powder yiel d produced [25]. Depending on the nozzle design, the gas can impinge the metal stream either at an angle or tangentially. In addition to the geometric parameters of the nozzle, other factors influencing the powder characteristics are gas pressure type of gas, diameter (d ’) of the melt stream, flow characteristics of the molten material, and the ratio of mass transport of the melt and atomizing gas [17]. An expanded view of th e metal stream during atomization illustrates the development of spherical particles from the initial melt stream (Figure 1-6) [25]. Figure 1-6: Formation of spherical metal powder by gas atomization. Reprinted with permission from Powder Metallurgy Science 1994, Metal Powder Industries Federation, 105 College Road East, Pr inceton, New Jersey, USA. Figure 3.16, p. 103. The gas around the molten stream expands, causing a depressurization of the liquid metal [25]. The stream, in turn, expands into a hollow cone. The high surface-area-to-volume ratio of this cone lead s to instability in the geometry that, in conjunction with the inert gas stream, leads to breakup of th e cone into long ligaments.

PAGE 33

16 Eventually, these ligaments break up into ellipsoids, and finally into small spherical droplets. The amount of superheat in the me tal before atomization plays a key role in determining the final shape and size of the powder. The level of superheat must be sufficient so that the particles do not solidify be fore becoming spheres. It is also crucial to eliminate any agglomerations of particles or satellite pa rticles by proper design of the atomization process. Either of these can pot entially result in poor packing of the powder before consolidation. Powder consolidation Conventional powder-compaction techniques, like cold pressing and sintering, are insufficient in achieving a fully dense P/M superalloy billet, because of the incompressibility of superalloy powder, and th e susceptibility to forming oxide scales at sintering temperature [17]. Th erefore, compaction techniques that involve high pressure at elevated temperatures are necessary. Early work on P/M superalloy compaction techniques examined forge compaction [28] and extrusion [28, 29] as means of achieving fully dense P/M billets. Another novel co mpaction process that has received some attention is consolidation by atmospheric pressure (CAPR) [30]. However, the advancements in HIP processing of powders along with potential cost savings make this the preferred route for consolidation of superalloy powders into billet form. The basic principle behind HIPing is the application of high pressure gas to consolidate the powder within an autoclave at elevated temperatures. The heater is generally located inside of the pressure vessel. Atomized powder, loaded into a container, is placed in the autoclave for comp action. These containers can be fabricated out of sheet metal, glass, or ceramic. Wh ile the glass and ceramic molds provide for better flexibility in both size and final part complexity, the metal cans are not as brittle

PAGE 34

17 [17, 31]. They can handle higher autoclave load ing. It is ideal for the powder to be spherical and have a wide size distribution, to facilitate higher packing densities before compaction [32]. The containers are loaded und er vibration, to achie ve packing densities approaching 80% of the theoretical density before consolidation. The cans are then evacuated and sealed before being placed in the HIP chamber. During HIPing, densification is though t to take place in a similar manner to sintering [32]. The mechanisms describing the initial bonding of powder particles during a HIP cycle [32, 33] are as follows: Initial elastic deformation. Plastic deformation of the particles by dislocation. Power-law creep or dislocation creep. Nabarro-Herring creep or volume diffusional creep. Coble creep or grain bou ndary diffusional creep. The final stages of densification occur th rough the same mechanisms that govern neck growth during sintering [32]. It is possible to produce a wide range of si zes and shapes of final parts in the asHIP condition with proper container selection and de sign. Through production of net-shape and near net-shape HIPed parts, there is a great potential for reduction in starting material for P/M supe ralloy parts (Figure 1-7) [ 20]. This production of netshaped parts, in addition to the intrinsic homogeneity of atomized powders, translates directly into a cost savings for turbine disk manufacturers, as they could possibly eliminate forging operations, excess material, and costs associated with machining the final part [34, 35]. In 1976, the development of as-HIP turbine disks was viewed as “one of the most exciting and technologically impor tant metallurgical developments in recent

PAGE 35

18 years” and one that could “re volutionize the ai rcraft engine rotati ng parts industry” [34, p.502]. Figure 1-7: Material and fabrication savings with P/M processing of superalloys. Reprinted with permission from Eds. J. R. Davis and Davis & Associates, “Powder Metallurgy Superalloys,” Heat Resistant Materials ASMInternational, 1997, Figure 1, p. 272. Challenges Unfortunately, these potentia l cost savings have never been fully realized by the turbine disk manufacturers. Powder-processed superalloys in the as-H IPed + heat treated condition do not contain enough i nherent damage-tolerance for use as disk alloys in jet engines. “A damage tolerant design assumes that a component has a flaw (i.e. crack) of a size just below the non-destruc tive inspection detect able limit. The times to inspection and/or retirement are based on the crack grow th rate of the flaw.” [36, p. 388] This

PAGE 36

19 approach enables disk manufacturers to design around potential failures through knowledge of, and confidence in, tensile strengt h and creep and fatigue lifetimes. While as-HIPed P/M superalloys ha d sufficient tensile and creep capabilities through heat treatment alone [34, 37], it was soon evident that the low cy cle fatigue (LCF) lifetimes contained large amounts of scatter and were highly dependant on the cleanliness of the powder and the consolidated billets freedom fr om defects [35]. The three main defect types found in P/M superalloys are prior pa rticle boundaries, ceramic inclusions, and thermally induced porosity [38]. Prior particle boundaries (PPBs) have been the most extensively studied of these defect types for P/M superalloys. A HIPed billet that suffers from PPB defects will exhibit this network through the whole microstr ucture causing this defect type to be the most detrimental to the fatigue lifetime of the alloy. Powd er particles that are left undeformed from the HIP processing can beco me decorated with detrimental secondary phases during the HIP cycle. La rger particles are strained to a much smaller degree than small particles during a typical HIP cycle (Fi gure 1-8) [38]. PPB networks tend to form on these larger undeformed powde r particles [39]. Through careful analysis, it has been determined that the PPB defects consist of a semi-continuous network of carbides, oxides, oxy-carbides, and perhaps oxy-carbonit rides [23, 40, and 41]. It is thought that during the initial stages of densification during HIPing that titanium and carbon within the powder migrate to the particle surfaces as a result of oxides on the outside of the particles [40]. This results in a film of stable titanium oxy-carbides on the prior particle boundaries. These carbides are mostly MC in natu re. However, there is some debate as

PAGE 37

20 Figure 1-8: Plastic strain suffered by sma ller and larger partic les during HIPing of a bimodal particle size distribution of pow ders. Reprinted with permission from R. D. Kissinger, S. V. Nair, and J. K. Tien, “The Infuluence of Powder Particle Size Distribution and Pressu re on the Kinetics of Hot Isostatic Pressing (HIP) Consolidation of P/M Superalloy Ren 95,” Superalloys 1984 Eds. M. Gell, C. S. Kortovich, R. H. Bricknell, W. B. Kent, and J. F. Radavich, The Metallurgical So ciety of AIME, 1984, Figure 6, p. 291. to the mechanisms for precipitation of these deleterious carbide and oxide precipitates. Some theorize that oxides on the powder partic le surfaces serve as nucleation sites for carbides and borides [41]. However, others cl aim that these surfaces are not preferential sites for carbide nucleation [42]. Only extremely fine carbides were found on the surface of these powder particles. In stead, the metal-metal interfaces between powder particles that have stuck together as the result of co llisions during atomization are the preferential sites for carbide precipitation. A detailed study by D. R. Chang et al. [39] examined the effects of various processing defects on mechanical propert ies of P/M Ren 95. PPBs were discovered to have an average size of approximately 9,700 m2 with a max size up

PAGE 38

21 to 161,300 m2 when samples were made from -150 (< 106 m) mesh powder. In this study, -150 mesh Ren 95 was seeded with va rious dopants to induce fatigue failures at PPBs and ceramic inclusions. The dopants chos en to lead to PPB networks were Buna N (an organic carbide former) and mill scale (a n inorganic oxide former). The average lifetime of as-HIPed Ren 95 was reduced fr om 37,000 cycles in the baseline (no dopant) condition to 8,675 cycles in the seeded conditi ons. Additionally, the frequency of failure at PPB sites was increased from 5% to 95%. A typical PPB fracture initiation site from this study is shown in Figure 1-9 [39]. Figure 1-9: PPB initiation site in doped Ren 95. Reprinted with permission from D. R. Chang, D. D. Krueger, and R. A. Sprague, “Superalloy Powder Processing, Properties and Turbine Disk A pplications,” Superalloys 1984 Eds. M. Gell, C. S. Kortovich, R. H. Bricknell, W. B. Kent, and J. F. Radavich, The Metallurgical Society of AIME, 1984, Figure 8A, p. 258. Ceramic inclusions have also been de trimental to the fatigue properties of as-HIPed powder superalloys. A typical failur e origin associated with ceramic defects is displayed in Figure 1-10 [39]. These inclusions originate from the melting crucible, the pouring tundish, and the atomizing nozzle wi th aluminum, zirconium, magnesium, and calcium being the major metallic elements [39]. The composition of the inclusions depends entirely on the ceramic materials used during production of th e powder. In -150

PAGE 39

22 Figure 1-10: Type 1 ceramic inclusion fatigue initiation site. Repr inted with permission from D. R. Chang, D. D. Krueger, and R. A. Sprague, “Superalloy Powder Processing, Properties and Turbine Disk Applications,” Superalloys 1984 Eds. M. Gell, C. S. Kortovich, R. H. Bricknell, W. B. Kent, and J. F. Radavich, The Metallurgical Soci ety of AIME, 1984, Figure 3A, p. 250. mesh Ren 95, these inclusions were found to have an average size of approximately 6500 m2 and on some occasions up to 64,500 m2. A study by W. H. Chang et al. [43] seeded -150 mesh Ren 95 with Al2O3 to facilitate fatigue initiation at ceramic inclusions. LCF specimens were tested at 1000F and 0.66% strain range. The average lifetime was decreased from 36,840 cycles for unseeded mate rial to 3,411 cycles for Ren 95 seeded with Al2O3. The third type of defect associated w ith premature failure of P/M superalloy specimens tested in fatigue is therma lly induced porosity (TIP). During argon atomization, some of the larger particles can form voids that contain entrapped argon. The entrapped gas can then expand at high temperatures during HIP cycles or heat treatment, forming much larger voids that have a deleterious effect on the fatigue lifetimes of P/M alloys. These voids are generally less than 1300 m2. Due to their Initiation site

PAGE 40

23 smaller size, TIP defects are not as detrimen tal to fatigue life as PPB’s and ceramic inclusions. To overcome the detriment in properties th at these defects have caused in as-HIP P/M superalloys, turbine disk manufacturers ha ve turned to thermomechanical processing (TMP) of the alloys. Ren 95 was HIPed and isothermally forged to an 80% reduction [43]. This hot working resulted in a large in crease in LCF lifetime. The extrusion step broke up and reduced the size of the PPB networks; however only had a small effect on the ceramic inclusions. A second study examined the effect of cons olidation by extrusion followed by isothermal forging [39]. The extrusion was carried out with a 6.5 to 1 reduction ratio. The forging resulted in a 60% re duction of the consolid ated material. As a result, the tensile strength and the LCF lifetime were improved dramatically (Figure 1-11) [39]. The extrusion and isothermal forging steps served to break up both the ceramic inclusions and PPB networks. As a result, the maximum size of these defects in the TMP material is much smaller, reducing th eir tendency to be crack initiation sites. More samples failed at voids and crystallographic origins than in the non-TMP material. Figure 1-12 [39] is a fractogra ph of a typical crystallogra phic crack initiation site. Longer lifetimes are generally seen from samples that fail at these defects. Unfortunately, the TMP steps needed to pr oduce a defect-tolerant microstructure are costly and limit the usage of as-HIP pow der superalloys in the turbine industry. Research into process improvements in recent years has improved the fatigue lifetime of as-HIP material to the point that the original hope for cost reduction for turbine disks is becoming more achievable. The average LCF lifetime of as-HIP alloys has increased dramatically as can be seen in Figure 1-13 [22]. Im provements in the gas

PAGE 41

24 Figure 1-11: Comparison of average LCF liv es of HIP vs. HIP + Forge and Extrude + Forge Ren 95. Reprinted with permissi on from D. R. Chang, D. D. Krueger, and R. A. Sprague, “Superalloy Powd er Processing, Properties and Turbine Disk Applications,” Superalloys 1984 Eds. M. Gell, C. S. Kortovich, R. H. Bricknell, W. B. Kent, and J. F. Ra davich, The Metallurgical Society of AIME, 1984, Figure 11, p. 262.

PAGE 42

25 Figure 1-12: Fatigue initiati on at a crystallographic defect Reprinted with permission from D. R. Chang, D. D. Krueger, and R. A. Sprague, “Superalloy Powder Processing, Properties and Turbine Disk Applications,” Superalloys 1984 Eds. M. Gell, C. S. Kortovich, R. H. Bricknell, W. B. Kent, and J. F. Radavich, The Metallurgical Soci ety of AIME, 1984, Figure 10B, p. 260. Figure 1-13: Average low cycle fatigue life for a P/M supera lloy during 1980 through 1996. Reprinted with permission from J. H. Moll and B. McTiernan, “Powder Metallurgy Superalloys,” ASM Handbook 7 ASM-International, 2003, Figure 30, p. 900. atomization process have re sulted in fewer ceramic in clusions. Switching from commercial purity argon to high purity ar gon limited the number of particulates contained in the argon gas itself [38]. Studies on the melting practices have shown that

PAGE 43

26 magnesia-based crucibles and electroslag re melting can reduce the oxide concentrations compared to other crucibles or melting methods [17]. Additionally, the size of potential inclusions has been reduced simply by siev ing the atomized powder to finer particle sizes. The average particle size has been reduced from approximately 250 m to smaller than 105 m [22]. This has been made possible in part by the ability to produce finer powder yields from atomization. Reduci ng the levels of boron and carbon in the chemistry of P/M superalloys also helps minimize the PPBs. More care has also been taken when handling the powder before consolidation, to limit the amount of contaminants. Improvements in modeling the pa rticle deformation and resulting densities of HIP cycles have allowed powder manufact urers to better understand HIP processing and to optimize the HIP cycle parameters to minimize the effect of defects on the LCF lifetimes. Original studies on deformati on during HIPing concentrated on monosized particles [33]. To more closely approximate real world multimodal particle distributions, models were developed to study the densific ation of a bimodal distribution through HIP [44, 45]. Using these models, HIP cycles that provide more uniform deformation to all particles can be developed, resu lting in an improved as-HIP material. Some have even suggested performing the HIP cycles at higher su b-solidus temperatures to facilitate grain growth past the PPB’s. Finally, improve ments in non-destructive evaluation (NDE) techniques have been used to help insure th at defects are of sufficiently small size before any parts are put into service. It is with these improvements in superalloy powder processing in mind that a recent study examined the mechanical properties of as-HIP Alloy 720 [27]. With cleaner production methods and a supersolvus heat treatment, the properties compared favorably with ex truded + isothermally forged Alloy 720.

PAGE 44

27 Alloy 720 Alloy 720 LI (low inclusion chemistry) is a powder processed nickel-base superalloy that has begun to be widely used in both the aircraft and land-based gas turbine industries over the last ten years [46]. It is a derivation from Alloy 720, which was originally developed as a cast and wrought turbine blade alloy for use in land-based gas turbines [47, 48]. To lim it segregation from casting, Alloy 720 LI can be processed through powder metallurgy as an alternativ e to the traditiona l cast and wrought processing route. Chemistry and processing Alloy 720 was initially designed to be ch emically resistant to environmental attack through both oxidation and sulfidizat ion. The composition of Alloy 720 and Alloy 720LI are given in Table 1-3 [46]. Titanium and aluminum were added as pr ecipitation Table 1-3: Nominal chemistr y of Alloy 720 and Alloy 720LI. Ni Cr Co Mo W Ti Al C B Zr Alloy 720 Bal 18.0 14.8 3.0 1.25 5.0 2.5 0.035 0.033 0.03 Alloy 720LI Bal 16.0 15.0 3.0 1.25 5.0 2.5 0.0100.025 0.0100.020 0.03 strengtheners because of their tendency to form ’. Molybdenum, tungsten, chromium, and cobalt are all used for solid -solution streng thening of the matrix. Alloy 720 contains a high amount of chromium for oxi dation and sulfidization resistance. The formation of TiC, Mo2B3, and Cr23(C,B)6 along the grain boundaries effectively limits the diffusion of oxygen and sulfur along the grain bo undaries [47, 49]. This helps to retard the crack growth rate at elevated temperat ures compared to other cast and wrought blade materials.

PAGE 45

28 As a blade material, Alloy 720 was engineer ed to be used for up to 10,000 hours at 900C. A coarse grained microstructure was optimal for turbine blad e applications where creep rupture lifetime was vitally important. However, because of Alloy 720’s combination of exceptional strength and hot workability, turbine manufacturers became interested in using the alloy in a fine grained cond ition as a potential turbine disk material [47]. It was soon discovered that the fine grain Alloy 720 is fairly susceptible to the formation of the deleterious sigma ( ) phase [50]. The forma tion of the topologically closed-packed (TCP) phase led to a rapid deteriorati on in the tensile ductility, creep resistance, and toughness of the alloy. The most common nucleation sites for phase in superalloys are at large grain boundary ’ and at carbides [51]. The phase forms along the grain boundaries, as globular precipitates similar to chro mium carbides. Alternatively, it can form as platelet or needle-like precipitates. phase has an approximate composition of (Cr0.5, Mo0.1), (Ni0.2, Co0.2) in Alloy 720 [52]. Not only can these precipitates act as crack initiation sites, but th ey also can deplete th e matrix of chromium [52]. The loss of chromium will lower the strength of the matrix, thereby causing a deficit in the mechanical propertie s of the alloy as a whole. The chemistry was then refined to th e current Alloy 720 LI composition to improve phase stability and reduce casting defe cts such as carbide and boride stringers [46]. The amount of chromium was reduced by two weight percen t to decrease the stability of Additionally, the carbon and boron levels were reduced to decrease the number of nucleation sites av ailable for the formation of The change in chemistry

PAGE 46

29 has a noticeable impact on the microstructure of exposed samples as can clearly be seen in Figure 1-14 [50]. These prec ipitates were found to be M23(C,B)6 and phase. Reed et Figure 1-14: Effect of thermal exposure on microstructures of subsolvus heat treated Alloy 720 and Alloy 720LI. A) Alloy 720 as-heat treated. B) Alloy 720 after 750C for 1000 hours. C) Alloy 720LI as -heat treated. D) Alloy 720LI after 750C for 1000 hours. Reprinted with permission from P. W. Keefe, S. O. Mancuso, and G. E. Maurer, “Effects of Heat Treatment and Chemistry on the Long-Term Stability of a High Stre ngth Nickel-Based Superalloy,” Superalloys 1992 Eds. S. D. Antolovich, R. W. Stusrud, R. A. MacKay, D. L. Anton, T. Khan, R. D. Kissinger, D. L. Klarston, TMS, 1992, Figure 4, p. 493. al. [52] constructed an expe rimental time-temperature-tran sition (TTT) diagram for Alloy 720 and Alloy 720LI that shows that the chemis try change effectively delayed the onset of formation for Alloy 720LI. (Figur e 1-15) [52]. The time for 1.0% formation has

PAGE 47

30 Figure 1-15: Experimental TTT diagram for the formation of 0.5 and 1.0 wt% of sigma in Alloy 720LI. Reprinted with permi ssion from R. C. Reed, M. P. Jackson, and Y. S. Na, “Characterization and Modeling of the Precipitation of the Sigma Phase in UDIMET 720 and UDIMET 720LI,” Metallurgical and Materials Transactions A 30A 1999, Figure 9, p. 529. been increased from 10 hours for Alloy 720 to over 1000 hours for Alloy 720LI. Additionally, the nose of the TTT diagram has been decreased to about 750C for the new composition, as opposed to 800C for the orig inal Alloy 720. It is important to note that Alloy 720 LI exhibits excellent micr ostructural stability below 700C, which is above the temperatures that a high pressure turbine disk rim will see in service. While Alloy 720LI is able to be produced by traditional cast and wrought processing, it is difficult to pr oduce a billet of the alloy that is homogenous in grain size and chemistry. It is being processed in up to 250 mm diameter billets for Rolls-Royce engines [54]. However, even with the most current melting and ingot conversion techniques, Alloy 720LI suffers from ’ banding and from inhomogeneous grain sizes. Unrecrystallized grains up to ASTM 0 grain si ze can be found near the center of the billet

PAGE 48

31 and can be difficult to fully eliminate. As a result, many turbine manufacturers have preferred the powder meta llurgy processing route. Microstructure Alloy 720LI has been reported in literatu re to contain approximately 55 volume percent ’ [46]. Both its microstructure and th e resultant mechanical properties can be easily manipulated through heat treatmen t and forging [55]. Like many Ni-base superalloys, Alloy 720LI has a multi-modal distribution of ’ precipitates. It is important to note that the nomenclature for the different ’ sizes in P/M superall oys is different than that used for single crystal superall oys. In P/M superalloys, primary ’ consists of large blocky precipitates that nucl eate discretely along the grain boundaries. These particles decorate and pin the grain boundaries, impeding grain growth during solution heat treatment. The sec ondary and tertiary ’ precipitates are formed by nucleation and growth during cooling from a supersolvus heat treatment. It has been proposed that the secondary ’ particles nucleate durin g quenching from the heat treatment temperature [56]. As the temperature conti nues to decrease the secondary ’ precipitates grow in size, rejecting solute into the matrix. This continues until the diffusion rate of the solute is too slow for the precipitates to grow any fu rther. The alloy eventually cools to a point where the driving force for nucleation of the tertiary ’ is large enough to occur. These particles grow in size up to 90 nm during the aging heat treatment. It is thought that the nucleation and growth pr ocess for the secondary ’ is in competition with that of the tertiary ’ [57]. At slower cool ing rates, the secondary ’ can grow larger, leaving less supersaturation in the matrix for nucleation of the tertiary ’. However, at faster cooling rates, the supersaturation in the matrix is greater when the nece ssary undercooling for

PAGE 49

32 nucleation of the tertiary ’ is reached, resulting in a larg er volume fraction of ultrafine ’. It is readily possible to ta ilor the size and morphology of the secondary (cooling) ’ through heat treatment alone. With a c onstant cooling rate of 55C/min, the temperature at which an Alloy 720LI samp le was removed and quenched had a great impact on the size of the secondary ’ (Figure 1-16) [57]. The size of the secondary ’ Figure 1-16: Mean diameter of the cooling ’ as a function of the interrupt temperature in Alloy 720LI. Reprinted with permissi on from J. Mao and K. Chang, “Growth Kinetics of ’ precipitates in P/M Superalloys ,” Materials Design Approaches and Experiences Eds. J.-C. Zhao, M. Fahrmann, and T. M. Pollock, TMS, 2001, Figure 5, p. 314. was increased from 0.15mm to 0.25mm by changi ng the test interrupt temperature from 1121C to 650C. Additionally, the cooling rate used during a supersolvus solution heat treatment greatly effects on the size and morphology of the cooling ’ [58]. A wide range of varying morphologies were produced by c ontrolling the cooling rates with different quenching media and by wrapping the samples in insulation to mimic the cooling rates seen by large turbine disks. As the cooling rate fell below 1.0C/sec, the ’ size rose

PAGE 50

33 dramatically from approximately 0.2 m to 1.0 m. Using all of this data, a TTT curve was constructed for the pr ecipitation of different ’ morphologies in supersolvus heat treated Alloy 720LI [58]. The shape of secondary ’ progresses from spherical to dendritic to fan-type pr ecipitates as the cooli ng rate is slowed down. Although the carbon and boron levels in Alloy 720LI are low, carbi des and borides are still present and impact the mechanical prope rties. The P/M material contains small uniform carbide precipitates, while the cas t and wrought material has large primary carbides [59]. The most common type of carbi de found in this alloy is the MC carbide. The M23C6 carbides and M3B2 borides are also present in smaller amounts. As the cooling rate is decreased, the amount of th ese two phases increased to the point where they can exceed the amount of primary carbides at the slowest cooling rates. It was also discovered that aging time increased the amount of M23C6 and M3B2 present in the alloy. These carbides and borides precipitate mostly along the grain boundaries. They appear as the fine white precipitates in the micrograph below (Figure 1-17) [60] For this alloy, the MC carbide composition is predicted to be (Cr0.80Mo0.19Ni0.01)23C6 [61]. The borides on the other hand are expected to be of the composition (Mo0.76Cr0.24)3B2. After electrolytic extraction and x-ray diffraction analysis, it was discovered that th e weight fraction of M23C6, and M3B2 for subsolvus heat treated cast and wrought Alloy 720LI were all below 0.5% even after 3000 hour exposures at 800C [61]. The MC-type carbides were found in micrographs of Alloy 720LI but not in All oy 720 in this study. However, the amount in Alloy 720LI was not able to be measured by x-ray due to the ove rlap of their peaks with the peaks of M23C6 carbides and M3B2 borides, in addition to the low amount of MC carbides present in this study. These grain boundary carbides have been shown to reduce

PAGE 51

34 grain boundary sliding at elevat ed temperatures which improve s the creep resistance of the alloy [17]. The knowledge and ability to control these two additional strengthening mechanisms in P/M superalloys is important in improving the mechanical properties of these alloys. Figure 1-17: P/M Alloy 720 microstructure. Reprinted with permission from K. A. Green, J. A. Lemsky, and R. M. Gasior, “Development of Isothermally Forged P/M Udimet 720 for Turbine Disk Applications,” Superalloys 1996 Eds. R. D. Kissinger, D. J. Deye, D. L. Anton, A. D. Cetel, M. V. Nathal, T. M. Pollock, and D. A. Woodford, TMS, 1996, Figure 5, p. 701. In addition, the grain boundary microstructu re istself plays an important role in the mechanical properties of Alloy 720LI. Through control of heat treatment, it is possible to form serrations in the grain boundari es. They form during controlled cooling through the ’ solvus [62]. Their amplitude and period are determined by the homogenization temperature, th e cooling rate, and the final temperature for the controlled cooling. Their presence has been noted in Alloy 720 LI by Furer and Fecht [63]. It is generally thought that these serrations de velop as a result of coarsening of the ’ precipitates close to the grai n boundaries. These grain boundar y serrations (Figure 1-18)

PAGE 52

35 [64] have been demonstrated to have a posit ive effect on creep fatigue crack growth rate by impeding grain boundary sliding [64, 65]. Figure 1-18: Optical micrographs of Astr oloy. A) “Smooth” gr ain boundaries. B) Serrated grain boundaries. Reprinted with permission from H. L. Danflou, M. Marty, and A. Walder, “Formation of Serrated Grain Boundaries and Their Effect on the Mechanical Properties in a P/M Nickel Base Superalloy,” Superalloys 1992 Eds. S. D. Antolovich, R. W. Stusrud, R. A. MacKay, D. L. Anton, T. Khan, R. D. Kissinger, and D. L. Klarstrom, TMS, 1992, Figure 1, p. 65. Objectives The increased homogeneity of powder pro cessed Ni-base superalloys has led to dramatic improvements in the mechanical pr operties compared to traditional cast and wrought processing. However, the as-consolid ated and heat treated P/M superalloys donot contain enough inherent defect tolerance to warrant usage in applications like jet engines, where unexpected failure can lead to tragic conseq uences. These alloys need expensive thermomechanical processing steps to reduce the size of de fects so that they are below the critical crack initiation length for fatigue failure. Elimination of these processing steps would result in a tremendous re duction in the cost of finished and flightready turbine disks.

PAGE 53

36 The purpose of this study was to try to im prove and limit the scatter in mechanical properties of as-HIPed Alloy 720LI through heat treatment alone. In unpublished research by Pratt & Whitney, a supersolvus he at treatment improved the fatigue lifetime and fracture toughness of single crystal superalloys. A m odification of this heat treatment was given to Alloy 720LI to dete rmine the effect of a supersolvus heat treatment with a slow, controlled cooli ng rate on the microstructure, mechanical properties, and fracture mechanics of P/M Alloy 720LI. The creep lifetime, tensile strength, and fatigue lifetimes of specimens with a supersolvus heat treatment and a standard subsolvus heat treat ment were compared. Detailed fracture analysis of the tested specimens was also performed to dete rmine the mechanisms of failure for the two heat treatments.

PAGE 54

37 CHAPTER 2 EXPERIMENTAL PROCEDURES Materials All materials used in this study were pr ocessed by Crucible Compaction Co. in Oakdale, PA. The alloys were gas atomi zed in the 2268 Kg (5000 lb) production gas atomization unit, using high-purity argon as the atomizing gas. The powder was collected and then sieved to -230 mesh (63 m) size, with the ma jority of the powder below 400 mesh (38 m). All powder particles coarse r than the 230 mesh size were removed from the powder lot and returned as revert to be used in further atomization runs. The powder was then loaded into stai nless or mild steel HIP cans and compacted. The HIP cycle used for all material in this study was below the gamma prime ( ’) solvus (1129C/100.0 MPa/4 h or 2065F/14.5 Ksi/4 h). The test material came from eleven different master powder blends atomized at Crucible. Ther e were only small differences in particle size and overall composition from one master powder blend to another. The sieve analysis for the various master powder blends is listed in Table 2-1. The compositions of all of the master powder bl ends, along with the nominal composition of the Alloy 720 LI, are listed in Table 2-2. All samples were provided in the as-HIPed condition. Development of Heat Treatment Previous research by Pratt & Whitney ha s indicated that a supersolvus heat treatment followed by a slow cool through the ’ solvus temperature would produce the desired microstructure and mech anical properties in their sing le crystal alloys. In this

PAGE 55

38 Table 2-1: Sieve analysis of the master powder blends (MPBs) in weight percent MPB +230 -230/+270 -270/+325 -325/+400 -400 95SW860 0 0.7 8.4 13.2 77.7 95SW861 0 0.6 8.1 13.7 77.6 96SW863 0 0.5 8.2 14.5 76.8 96SW864 0 0.5 6.4 12.7 80.4 96SW865 0 0.4 5.6 13.1 80.9 96SW887 0 0.9 8.9 14.2 76.0 96SW888 0 0.9 6.4 14.9 77.8 97SW984 0 0.4 7.5 11.1 81.0 97SW985 0 0.2 6.2 10.8 82.8 98SW035 0 0.5 5.7 13.4 80.4 98SW036 0 0.2 6.0 11.4 82.4 Table 2-2: Chemistry analysis of the master powder blends in weight percent MPB Ni Cr Co Mo W Ti Al C B Zr Nominal Bal 16.57 14.713.00 1.28 5.02 2.49 0.010 0.012 0.038 95SW860 Bal 16.64 14.783.00 1.27 4.95 2.48 0.008 0.013 0.037 95SW861 Bal 16.67 14.883.01 1.28 4.91 2.55 0.010 0.013 0.040 96SW863 Bal 16.85 14.872.81 1.26 5.00 2.52 0.011 0.012 0.038 96SW864 Bal 16.53 14.703.00 1.25 4.98 2.58 0.011 0.013 0.038 96SW865 Bal 16.60 14.772.99 1.26 5.00 2.52 0.009 0.012 0.040 96SW887 Bal 16.64 14.743.00 1.27 4.98 2.47 0.012 0.014 0.037 96SW888 Bal 16.57 14.713.00 1.28 5.02 2.49 0.010 0.012 0.038 97SW984 Bal 16.73 14.723.00 1.26 4.92 2.43 0.007 0.014 0.037 97SW985 Bal 16.73 14.723.01 1.26 4.91 2.45 0.006 0.011 0.037 98SW035 Bal 16.6 14.633.01 1.29 5.03 2.47 0.011 0.011 0.040 98SW036 Bal 16.7 14.652.99 1.27 4.93 2.49 0.010 0.012 0.040 microstructure, there is a regular array of dendritic-shaped secondary ’ in the matrix. However, specific details of the Pratt & Whitney heat treatment were not available. To determine the alternate supersolvus heat tr eatment schedule need ed to produce this

PAGE 56

39 microstructure in Alloy 720, various heat tr eatment parameters were studied including soak temperature, soak time, cooling rate, a nd final temperature before fan air cooling. The final heat treatment was verifi ed through microstructural analysis. Heat Treatment Trials Two solution heat treatments of Alloy 720LI were compared in this study. The control samples were given a standard subsolvus heat treatment from the literature. This heat treatment schedule is 1100C (2012F)/2 h. After two hours, the samples were fan-air-cooled (FAC). The alternate h eat treatment was developed to match the microstructure that Pratt & Whitney’s study obtained after a supers olvus heat treatment for their single crystal superalloys. Si nce these P/M alloys are very different compositionally than the single crystal alloys utilized by Pratt & Whitney, the exact heat treatment temperature and the c ooling rate necessary to obtain the desi red microstructure in the Alloy 720LI were not known. As a result it was necessary to perform several heat treatment trials to determine the temperature and cooling rate needed to produce the alternate microstructure. The test matrix for determining the alternate heat treatment schedule is shown in Table 2-3. The sample s were heat treated in a Lindberg model Table 2-3: Heat treatment trial matrix Sample ID Soak Temp.(C) Hold Time(h) Cooling Rate(C/min) Final Temp. A2 1175 2 3.3 1020 A3 1200 2 7.4 1020 A4 1225 2 7.2 1020 C1 1175 2 2.75 1075 C2 1175 2 0.82 1020 D2 1175 2 2.03 1020 A1 1175 4 2.75 1020 UMB1 1175 2 3.0 1020 D1 1200 2 2.75 1020

PAGE 57

40 54233-V tube furnace with a Eurotherm model 818S temperature controller with a 50.8 mm (2 in) diameter mullite tube. The cont rol thermocouple was a type R thermocouple with alumina sheathing positioned radially to the hot zone, but outside the mullite tube. A type K thermocouple, with 304 stainless st eel sheathing, was insert ed through the front end of the tube furnace into the hot zone to monitor the actual temperature of the samples throughout the heat treatment cycle. All samp les were heated from room temperature to their hold temperature, held for two hours, and then cooled slowly in the furnace at various cooling rates to a final temperatur e of 1020C (1868F). This final temperature was selected in order to insure that the sa mples were cooled at least 100C (180F) below the solvus temperature so that all precipit ation from cooling was accomplished before the more rapid cooling in air. Upon completion of the heat treatments the samples were removed from the furnace and allowed to fan-air-cool to room te mperature. They were metallographically prepared by first mounting in di allyl phthalate usi ng a Leco model PR-10 mounting press. The metallographic samples were then co arsely ground on a Leco model BG 20 belt grinder with a 60 grit SiC be lt to remove the oxidation laye r and to polish far enough into the sample to offset any temperature loss that may have occurred during transfer from the furnace to the quench bucket. The metallographic samples were then polished by hand using successive SiC papers of 120, 240, 320, 400, 600, and 800 grit. At each grit size, the samples were polished in the same direction until all of the scratches from the previous step had been removed. The samp le was then rotated 90 and polished on the next highest grit paper. Once they were pol ished to 800 grit paper, fine polishing was performed using a Buehler Billiard cloth on a Leco model VP-50 po lishing wheel, with

PAGE 58

41 alumina suspensions of 15 m (5.9X10-4 in) and 5 m (2.0X10-4 in). The final polishing was performed on the same metallographic po lishing wheel, with alum ina suspensions of 1.0 m (3.9X10-5 in) and 0.3 m (1.2X10-5 in) on a Buehler Microcloth. Great care was taken between polishing steps to clean the sa mples in water using a Leco ultrasonic bath and to clean the polishing wheels, so there would be no contamination of larger sized alumina particles during the fine polishing. The surfaces were examined for scratches with an Olympus light optical microscope be fore moving on to the next polishing step. After polishing, the specimen surfaces were cleaned with methanol Pratt & Whitney etchant #17 (25 mL HCl, 25 mL HNO3, 25 mL H20, and 0.75 g molybdic acid) was then swabbed on the samples with cotton swabs for 5-20 s to etch the microstructure. This etchant selectively attacks the ’ precipitates to the matrix. These samples were then examined using a JEOL JSM 6400 (Figure 2-1) scanning electron microscope (SEM) at 15 KV accelerating voltage and a 15 mm (0.59 in) working distance. Heat Treatment After the alternate heat treatment sche dule was determined through these heat treatment trials and microstructural analysis, the heat treatment was carried out on the provided Alloy 720 LI mechanical testing bars at the University of Florida in an Applied Test Systems box furnace. When the samples were removed from the furnace they were given a fan-air-cool to room temperature. A type K thermocouple was used as the control thermocouple, with a second type K thermocouple with Nextel sheathing positioned directly on the sample bars for accurate monito ring of the temperature of the bars. This second thermocouple was placed under the test bars so that the bead of the thermocouple was in contact with the test bars at all times Half of the bars received the standard

PAGE 59

42 Figure 2-1: JEOL JSM 6400 SEM solution heat treatment. The alternate heat treatment was 1175C (2147F) for 2 h with a 3.0C/min (5.4F/min) cooling rate in th e furnace to 1020C (1868F). The samples were then removed and given a fan-air-cool to room temperature. While this heat treatment worked for the majority of the ma ster powder blends, the soak temperature had to be reduced to 1165C (2129F) to produce th e desired microstructu re for three of the MPBs as will be discussed in the results section. The same agi ng heat treatment was given to all sample bars regardless of the solution heat treatment. The aging heat treatment used in this study was the sta ndard Alloy 720 aging h eat treatment found in literature. The schedule was 760C (1400F) fo r 24 h with a fan air cool and then 650C (1200F) for 16 h followed by a fan air cool. This heat treatment was performed in the same box furnace with the same thermocouple set-up as during the solution heat treatments. Thirty-six bars were provided in the as-H IP condition for creep or tensile testing. These bars were approximately 19.1 mm ( 0.75 in) on a side and 101.6 mm (4.0 in) long.

PAGE 60

43 Eighteen creep/tensile bars were heat treated using the standa rd heat treatment, while the other eighteen were solution heat treated ac cording to the alternate heat treatment schedule. In addition, thirty-five bars we re provided in the as-HIP condition for low cycle fatigue testing. These bars were approximately 19.1mm (0.75 in) by 25.4 mm (1.0 in) on edge and 152.4 mm (6.0 in) in length. Se venteen of the LCF bars were given the standard heat treatment, while the other eight een test bars were gi ven the alternate heat treatment. To verify the heat treatment, small microstructural analysis samples of each master powder blend were include d along with the mechanical te sting bars in each of the heat treatment runs. These samples were cubes with dimensions of approximately 19.1 mm (0.75 in) on edge. Characterization After the heat treatments were complete, both the ’ solvus temperature and the microstructures of the two heat treatments we re fully characterized. It was important to determine the solvus temperature in order to understand the microstr uctural evolution of Alloy 720LI during the heat treatments. The microstructural charac terization of the two heat treatments served not only to help unde rstand the precipitation behavior during the heat treatments, but also to help determine the effect of the two microstructures on the mechanical properties. Determination of ’ Solvus The ’ solvus temperature was determined in two complimentary ways. The two methods were differential thermal analysis and metallographic examination of the microstructure at various temperatures. These results were then compared to accurately

PAGE 61

44 determine the ’ solvus temperature, the solidus temperature, and the liquidus temperature of Alloy 720LI. Differential thermal analysis Differential thermal analysis (DTA) was performed at two independent laboratories. Samples for DTA were sect ioned using a Leco diamond saw to produce specimens approximately 6.4 mm (0.25 in) by 9.5 mm (0.375 in) by 15.9 mm (0.625 in) in size. All DTA samples were in the as-H IPed condition with no heat treatment. Two samples from each of two different master pow der blends were sent to M&P Laboratories in Schenectady, NY. These samples were heated at a constant rate of 10C/min (18F/min) to 1600C (2912F). The temperat ure difference between the sample and a reference sample of pure nickel was measur ed. These results were plotted versus temperature. One more sample of each of th e two MPBs was sent to Dirats Laboratories in Westfield, MA. These samples were heat ed at 20C/min (36F/min). Once again, the temperature difference between the sample a nd a reference sample was measured. Two curves were plotted for these results. The fi rst is of the temperature difference versus temperature. The second curve is of the derivative temperature difference versus temperature. All reaction temperatures were then identified by the inflection points present on the plots. Metallographic examination In addition to DTA, solvus temperatures were determined by metallographic examination of quenched microstructures. For this portion of th e study, a bar of Alloy 720 LI in the as-HIP condition was sectioned into smaller samples using a LECO cut-off saw. The metallographic examination samp les were cut to approximately 6.4 mm (0.25 in) by 19.1 mm (0.75 in) by 19.1 mm (0.75 in). A Nextel-sheathed t ype K thermocouple

PAGE 62

45 was attached to each sample with 80Ni-20C r wire to measure the temperature of the sample accurately while in the furnace. Thes e samples were then heated in a Carbolite model CWF 13/23 box furnace with PM 2000 el ements (Figure 2-2) to various temperatures above and below the expected so lvus temperature. One set of samples was Figure 2-2: Carbolite box furnace heated to temperatures at 5C (9F) intervals between 1100C (2012F) and 1200C (2192F). These samples were allowed to soak at temperature for one hour before being removed with tongs and quickly dropped into an iced-brine quen ch to “freeze in” the microstructure at the elevated temperature and prevent the formation of ’ during cooling. This quenching medium was prepared by filling a stainless steel bucket with tap water. Salt was then added to the water and stirred until the water was supersaturated with salt. Then ice was added and the water stirred unt il the ice no longer me lted immediately. A second set of samples were solution heat tr eated at 1175C (2147F) for 1 h. They were

PAGE 63

46 then slow cooled at 3C/min (5.4F/min) dow n to temperatures at 5C (9F) intervals between 1120C (2048F) and 1140C (2084F). Th ese samples were allowed to soak at that temperature for an hour to allow precip itation to occur. All of these samples were then removed with tongs at their final temperature and quickly dropped into the iced-brine quench. Once the samples were cooled, they were also metallographically prepared by first mounting in diallyl phthala te and then polishing as previously described. They were etched with Pratt & Whitney etchant #17 to reveal the / ’ microstructure. The SEM was used to examine the micrsostructures, so that the temperatures at which all phases go into solution for this alloy could be determined. These results were then compared with the DTA data to better understand the micros tructural evolution of Alloy 720 during supersolvus heat treatment. Microstructural Analysis When the heat treatment of all of th e sample bars was completed, the small microstructural analysis samples were analyz ed to fully characterize the microstructures of both the alternate and sta ndard heat treatments. Each metallographic sample was mounted and polished to 0.3 m (1.2X10-5 in) following the procedure listed above. In order to reveal the / ’ microstructure, these samples were swabbed with etchant #17. They were then examined on the SEM so the ’ morphology could be fully characterized after both heat treatments. An as-HIP sample and samples that had been heat treated but not aged were also examined with the SE M and with secondary electron imaging using the field emission gun (FEG) of an FEI Strata DB235 focused ion beam (FIB).

PAGE 64

47 In addition, an analysis of the ’ volume fraction was performed according to the American Society for Testing and Materials (A STM) standard E562-95 [66]. At least ten fields of view from the SEM were used for each sample for this analysis. A circular grid with 24 intersection points was then overl aid on the printed micrographs. Every intersection between a gr id intersection and a ’ particle was counte d. The total number of intersections was divided by the total number of grid points for that sample, giving an approximate ’ volume fraction. The microstructural analysis samples for grain size determination were polished down to 0.3 m (1.2X10-5 in) to remove all etching effects. The samples were then etched with waterless Kallings etchant (5 g CuCl2, 100 mL HCl, and 100 mL CH3CH2OH) reveal the grain boundar ies. The average ASTM grain size was measured according to ASTM standard E112-96 [67]. For this analysis, five fields of view were analyzed for each heat treatment. The Abrams three-circle procedure was used to count the number of times the circles inte rcept grains in the micrograph. The final characterization technique that was utilized was Rockwell C hardness. The hardness of samples in the as-compact ed condition, the as-solutioned condition for each heat treatment, and in the as-aged condi tion for each heat treatment were measured using a Rockwell hardness indenter with th e Rockwell C tip. Twelve hardness values were determined for each sample to determ ine an average Rockwell C hardness value for each condition. Mechanical Testing Upon completion of the heat treatment, al l of the mechanical test bars were shipped to Crucible Compaction to be machined into test specimens. Crucible

PAGE 65

48 outsourced the machine jobs to three di fferent machine shops. Twelve of the creep/tensile bars (six from each heat treatm ent) were sent to Westmoreland Mechanical Testing & Research Inc. in Youngstown, PA to be machined into tensile samples according to the schematic in Figure 2-3. Th e gauge section for the tensile samples was 6.4 mm (0.25 in) in diameter and 31.8 mm (1.25 in) in leng th. The other twenty-four samples (twelve from each heat treatment) were sent to Joliet Metallurgical Laboratories, Inc. in Joliet, IL to be machined into cr eep samples according to Figure 2-4. The gauge section on the creep samples was 4.5 mm (0.177 in) in diameter and 26.0 mm (1.025 in) in length. Pin holes were machined into the shoulder of the creep specimens for attaching the extensometer frame with screws. The LCF test bars were then machined at Metcut Research Associates, Inc. in Cincinnati, OH according to Figure 2-5. The LCF test specimens had a gauge section that was 6.4 mm (0.25 in) in diameter and 19.1 mm (0.75 in) in length. These samples were later se nt to Low Stress Grind in Cincinnati, OH to machine a 0.25 mm (0.01 in) notch was m achined onto both shoulders of all LCF samples for attaching the knife edges of the ex tensometer frame. All gage surfaces were low stress ground to reduce or eliminate m achining effects. Furthermore, the LCF samples were polished by hand to elimin ate possible crack in itiation sites. Low Cycle Fatigue Testing The low cycle fatigue testing (LCF) was perf ormed at the University of Florida in Gainesville, Florida in the High Temperat ure Alloys Laboratory using an Instron servo-hydraulic frame (Model TC 25). (Figure 26). This frame has an actuator with six inches of travel and is capable of loads up to 9500 Kg (21000 lbs). The LCF testing was computer controlled through the Instron Fast Track 8800 computer. The Fast Track

PAGE 66

49 Figure 2-3: Tensile sa mple design. All dimensions are in inches.

PAGE 67

50 Figure 2-4: Creep sample design. All dimensions are in inches.

PAGE 68

51 Figure 2-5: Low cycle fati gue sample design. All dimensions are in inches.

PAGE 69

52 Figure 2-6: Instron-Satec servo-hydraulic test frame computer was connected to a traditional PC on which the individual tests were setup and controlled using the LCF module of Instron’s Fast Track Console software package. Loads/strains were applied to the test sp ecimens through pull/push-rods and threaded grips. The pull/push-rods were machined by Low Stress Grind in Cincinnati, OH from material (Udimet 720) cast and wrought by P CC Airfoils in Minerva, OH. The grips were obtained from Satec and made from Ma r-M 247 alloy. The specimens were heated to the test temperature by a Satec (M odel SF-12 2230) clam-shell Power Positioning Furnace with Kanthal elements. Temperatur e was controlled by a Eurotherm model 2416 controller using a type K thermocouple. A s econd thermocouple was used to verify the test temperature. The two t ype K thermocouples, with Nextel fiber sheathing, were tied on to the gauge section of each sample using 26 AWG 80Ni-20Cr bare thermocouple

PAGE 70

53 wire. A high temperature Instron extensomet er frame was attached to the sample using replaceable round knife edge inserts with a 12.7 mm (0.5 in) diameter to allow the extensometer to measure strain outside of th e hot zone of the furnace. The extensometer frame and the inserts were made out of Haynes 214. An Instron model 2620-826 dynamic extensometer with 12.7 mm (0.5 in) gauge length and 2.54 mm (0.1 in) travel length was attached to the extensometer fram e using straight knife edge attachments and size 014 neoprene O-rings. Once loaded in the test frame, the specimens were heated to the test temperature by the furnace and allowed to soak at the test temperature for thirty minutes before beginning the tests. The low cycle fatigue test matrix is lis ted below in Table 2-4. For all test specimens in this study, the third letter of the sample ID refers to the heat treatment. “A” is the standard heat treatment, while “B” refers to the alternate heat treatment. All tests were run with an R ( min/ max) ratio of 0.0. Therefore, the minimum strain value for every cycle of every fatigue test was 0.0%. The maximum strain value for each cycle is equivalent to the value for th e entire strain rang e. Two samples of each heat treatment were tested at three different temperatures and three different stra in ranges. The LCF testing was performed in strain control through the first 25000 cycles at 0.333 Hz. If the specimens did not fracture before 25000 cycles the test was switched to load control at the maximum and minimum loads during the stra in controlled portion of the test. The load control portion of the test was pe rformed at 10 Hz until fracture or until 1,000,000 cycles when the test was stopped and the sample was considered a run-out.

PAGE 71

54 Table 2-4: The LCF test matrix Heat Treatment MPB Serial Number Sample ID Temp.(C) Strain Range 96SW865 137789 ULA7 538 1.0% 96SW864 135211T ULA9b 538 1.0% 95SW861 133978B ULA1 538 1.1% 95SW860 133870T ULA2 538 1.1% 96SW863 135276B ULA13 538 1.2% 96SW864 135195B ULA14 538 1.2% 95SW861 133468T ULA3 649 1.0% 95SW860 133870B ULA5 649 1.0% 96SW865 134340B ULA10 649 1.1% 96SW864 135203B ULA11 649 1.1% 96SW865 134277B ULA15 649 1.2% Standard 96SW865 134367T ULA16 649 1.2% 96SW863 135294B ULB2b 538 1.0% 96SW865 134367B ULB11 538 1.0% 95SW860 133825B ULB3 538 1.1% 96SW864 135227B ULB4 538 1.1% 96SW864 135211B ULB9 538 1.2% 96SW863 135285B ULB10 538 1.2% 95SW861 133468B ULB5 649 1.0% 96SW864 135227T ULB6 649 1.0% 95SW861 133978 ULB7 649 1.1% 96SW863 135276T ULB8 649 1.1% 96SW864 135195T ULB13 649 1.2% Alernate 96SW864 135203T ULB15 649 1.2% High Cycle Fatigue Testing All high cycle fatigue testing (HCF) wa s performed in the High Temperature Alloys laboratory at the University of Florid a using the same servo-hydraulic frame as for the LCF testing. The HCF tests were cont rolled using Instorn’s Fast Track Console software. The samples were loaded using th e same push/pull rods and grips as in the LCF testing. The clam-shell furnace was en closed around the test specimen and heated to the test temperature. Tw o type K thermocouples were atta ched to the ga uge section of the specimens for temperature control. After a fifteen minute soak at test temperature, the HCF tests were started. These tests were run in load control at 10 Hz with an R ( min/ max) ratio of 0.1. A test run with a stre ss range of 993 MPa (144 Ksi) was cycled

PAGE 72

55 between a minimum load of 110.3 MPa (16 Ks i) and a maximum load of 1103 MPa (160 Ksi). Samples were tested until failure or until run-out, which was designated as 1,000,000 cycles. The test matrix for the HCF samples is shown in Table 2-5. Table 2-5: The HCF test matrix Heat Treatment MPB Serial Number Sample ID Temp.(C) Stress Range (MPa) 95SW860 133870B ULA5 538 993 96SW864 135203B ULA11 538 1034 96SW865 134367T ULA16 538 1086 95SW861 133468T ULA3 649 993 96SW863 135276B ULA13 649 1034 Standard 96SW865 134340B ULA10 649 1086 95SW861 133978 ULB7 538 993 96SW863 135285B ULB10 538 1034 96SW865 134358B ULB17 538 1086 96SW864 135227T ULB6 649 993 96SW864 135195T ULB13 649 1034 Alternate 96SW864 135211B ULB9 649 1086 Tensile Testing The tensile testing was also carried out on the Instron servo-h ydraulic test frame in the High Temperature Materials laboratory of the University of Florida. The Merlin module of the Fast Track software was used fo r set up and control of the tensile tests. The pull-rods and grips for the tensile testi ng were machined out of IN-713 superalloy by Satec. Two type K thermocouples, with Ne xtel sheathing, were a ttached to the gauge section of each elevated temperature tensil e test as in the LCF testing. The same clam-shell furnace heated the samples to their test temperature. The samples were given at least a fifteen minute hold at temperature to allow the temperature inside the furnace to equilibrate before testing was commenced. The same extensometer frame from the LCF testing was used with 6.4 mm (0.25 in) diameter knife edge inserts for the elevated temperature tensile tests. The room temp erature tensile tests did not require an extensometer frame, as the extensometer coul d be attached directly to the gauge section

PAGE 73

56 of the specimens. An Instron model 2630-110 static extensometer with a 25.4 mm (1.0 in) gauge length and 25.4 mm (1.0 in) of travel was used for measuring the strain for the tensile tests. This extensometer was attach ed to the sample with 6.4 mm (0.25 in) steel clips. The test matrix for the tensile testing is shown below in Table 2-6. All tensile tests were conducted at a strain rate of 2.5 mm/ min (0.1 in/min). The Merlin software recorded the data and then was used to produce stress/strain curves for each of the tensile tests. The software also calculated the yiel d strength, ultimate tens ile strength, ultimate strain, and fracture strength. Table 2-6: The Tensile test matrix Heat Treatment MPB Serial Number Sample ID Temp.(C) 96SW888 137789 UCA1 25 98SW036 143569 UCA4 25 97SW984 142274 UCA5 649 96SW888 137789 UCA7 649 96SW888 137816 UCA8 760 Standard 97SW984 142184 UCA10 760 96SW887 138142 UCB1 25 96SW888 137762 UCB4 25 96SW887 138142 UCB2 649 98SW036 143596 UCB5 649 96SW887 138160 UCB3 760 Alterante 96SW888 137816 UCB8 760 Creep Testing The creep specimens in this study were te sted on a Satec model M3 creep frame as pictured in Figure 2-7. The High Temperature Alloys Laboratory at the University of Florida has four of these creep frames for high temperature creep testing. These frames have a sixteen-to-one ratio on the lever arm between the load pan and the sample. Two of the frames are outfitted with a Satec (model SF-16 2230) Power Positioning Furnace. These furnaces have three z ones (top, middle, and bottom) that can be controlled independently to achieve the correct temperat ure along the entire gauge length of the test

PAGE 74

57 Figure 2-7: Instron-Satec Creep frame specimen. These furnaces have Kanthal elem ents. The other two creep frames have a Satec (model KSF 2-8-18) Power Positioning Furnace. These furnaces have a single hot zone with MoSi2 elements that are capable of test ing up to temperatures of up to 1500C (2732F). All of these furnaces are contro lled by a standard PC using the NuVision Mentor software supplied by Satec. The pu ll-rods and couplings used for the creep testing are the same as those described in th e tensile testing secti on above. Three type K thermocouples were attached directly to the gauge section of each sample with 80Ni-20Cr wire. These three thermocouples allowed the computer to independently control each furnace zone on each frame so that the power output of each zone needed to maintain the correct temperature along the whole sample could be obtained. A high-temperature Instron extensometer frame wa s attached to each sample with screws into the pin holes on the samp le shoulder. This frame and the screws were machined

PAGE 75

58 from Haynes 214. A Satec model 9234K linear variable differentia l transformer (LVDT) displacement transducer was attached to the extensometer frame outside of the hot zone of the furnace for measuring strain on the sample during testing. In total, twenty creep tests were performed according to the test matrix in Table 2-7. Two samples from each heat treatmen t were tested at every condition. Each specimen was given a one hour soak at the test temperature before beginning the test. The weight was step loaded in order to m easure an approximate value for the elastic modulus for the samples. Time to 0.1% 0.2%, 0.5%, 1.0%, 2.0%, 5.0% creep and time to creep rupture were recorded by the Mentor software. In addition, the minimum creep rate of each creep test was measured. Table 2-7: The Creep test matrix Heat Treatment MPB Serial Number Sample ID Temp.(C) Stress (MPa) 96SW888 137789 UCA12 677 1034 97SW984 142274 UCA16 677 1034 96SW888 137762 UCA9 677 793 97SW984 142184 UCA11 677 793 96SW888 137825 UCA13 704 689 96SW887 138160 UCA18 704 689 95SW860 133834T UCA20 732 483 95SW861 133807T UCA21 732 483 95SW860 133816B UCA19 760 483 Standard 95SW861 133807B UCA22 760 483 96SW888 137825 UCB11 677 1034 97SW985 139888 UCB18 677 1034 96SW888 137762 UCB7 677 793 98SW036 143569 UCB10 677 793 98SW036 143596 UCB6 704 689 96SW888 137825 UCB12 704 689 95SW861 133861T UCB19 732 483 95SW861 133798B UCB20 732 483 96SW888 137816 UCB14 760 483 Alternate 95SW861 133789T UCB23 760 483

PAGE 76

59 Fracture Analysis After failure, samples from each condition from the fatigue, creep, and tensile testing were evaluated to examine the fracture features and determine crack initiation site, type, and propagation path. Before analysis, all samples were cleaned in a Leco ultrasonic bath in acetone followed by metha nol to clean any dirt or dust from the fracture surface. The samples we re first viewed with a simple light optical microscope to get an overall view of the pr opagation path. Next the samples were examined on the SEM to reveal further details of the fracture su rface. If the crack in itiation site could be located for the sample, EDS was performed to determine the type of initiation site, e.g. inclusion, pore, machined crack, etc. Additi onally, the microstructure of the crack path was investigated along the length of the fr acture surface to determine what paths of failure were active during the va rious stages of fracture, i.e. intergranular, transgranular, ductile, or brittle. When SEM and optical examination of the fracture surface was complete, about 6.4 mm (0.25 in) of the specimen tip at fracture was sectioned off using a Leco (model VC-50) diamond saw. These samp les were mounted in diallyl phthalate and polished and etched to reveal any interactions between s econdary cracks and the alloy’s microstructure. The samples were polished down to 0.3 m (1.2X10-5 in) as in the previous section on metallographic sample prep aration. These samples were then etched with Pratt & Whitney etchant #17 to reveal th e microstructure and examined on the SEM.

PAGE 77

60 CHAPTER 3 RESULTS Microstructure Development of Heat Treatment The first step of this study was developing the heat treatment that would result in the desired ’ microstructure. This microstructure was to have a regular array of large dendritic-shaped ’, surrounded by tertiary ’. Additionally, the grain size was to be slightly larger than both the as-HIPed and the standard heat treated microstructures. The only known variables for producing this mi crostructure were that the solution temperature was to be above the ’ solvus temperature and that there was to be a controlled-cool through this solvus temperature. The ’ solvus temperature is reported in literature to be approximately 1150-1155C (2102-2111F) [46]. With this approximate value for the ’ solvus temperature, a set of heat treatment trials were run to develop the alternate, super-solvus, heat treatment for use in this study. Once the proper heat treatment was developed, the heat treatments were carried out on machined test bars. As a control variable, the standard heat treatment consisted of a two hour soak below the ’ solvus temperature at 1100C (2012F) followed by a fan-air-cool. Sm all, microstructural analysis samples were included for each Master Powder Blend (MPB) with all of the heat treatments. These samples were then examined to check the heat treatments and to fully characterize the microstructure.

PAGE 78

61 Heat treatment trials Small, as-HIPed specimens were given vari ous heat treatments according to the test matrix in Table 2-3. During these trials, the soak temperature, soak time, cooling rate, and final temperature before the fan-air-cool were varied to determine the proper heat treatment specifications. After the alternate heat treatment was developed, samples from each master powder blend were given both the alternate heat treatment as well as the standard heat treatment. The starting point for the alternate heat treatment was a soak at 1175C (2147F). This temperature was chosen si nce it is above the reported ’ solvus temperature. A soak time of two hours was origin ally selected to allow enough time for all of the ’ to go fully into solution. The baseline sample for this study was A2 and had an 1175C (2147F) soak with a 3.3C/minute (5.9F/min) cooli ng rate to 1020C (1868F). The first heat treatment parameter that was examined was th e solution temperature. Soak temperatures of 1200C (2192F) and 1225C (2237F) we re examined, in addition to 1175C (2147F). Micrographs of these heat treatme nts can be found in Figure 3-1. Samples A3 and D1 were solution heat treated at 1200 C (2192F), with A3 cooled at a rate of 7.4C/min (13.3F/min) and D1 at a rate of 2.8C/minute (5.0F/min). Finally, sample A4 was soaked at 1225C (2237F) and cooled at a rate of 7.2C/min (13.0F/min). All four of these samples were held at their soak temperature for two hours and cooled to 1020C (1868F) before they were removed from the furnace for the fan air cool. The higher heat treat temperatures did not result in the desired microstructure regardless of the cooling rate. From this set of trial samples it was determined that 1175C (2147F) would be used as the solution heat treatment temperature for the alternate heat treatment.

PAGE 79

62 A) B) C) D) Figure 3-1: Heat treatment trial samples with various soak te mperatures. A) A2 (1175C). B) A3 (1200C). C) D1 (1200C). D) A4 (1225C). The next heat treatment parameter that wa s studied was the cooling rate from the solution temperature through the ’ solvus. When examining the previous samples, it was determined that a cooling rate of approxima tely 7C/min (12.6F/min) was too fast. Sample D1 was closer to the desired microstructure than A3 Slower cooling rates were examined to determine the optimal cooling rate. These ranged from 0.82C/min (1.5F/min) to 3.3C/min (5.9F/min). These samples are shown in Figure 3-2. Sample C2 was given a slow cooling rate of 0.82C/min (1.5F/min) down to 1020C (1868F) and then fan air cooled to room temperatur e. Sample D2 was cooled at a rate of 2.0C/min (3.7F/min). Finally, A2, as di scussed above, was cooled at 3.3C/min (5.9F/min). All of these samples were soaked at 1175C (2147F) for two

PAGE 80

63 A) B) C) Figure 3-2: Heat treatment trial samples with different cooling rates. A) C2 (0.82C/min). B) D2 (2.0C/ min). C) A2 (3.3C/min). hours before the controlled cooling. They we re also all removed from the furnace when their temperature reached 1020C (1868F). As can be seen from the micrographs, both D2 and A2 produced the desired microstructu re. The cooling rate for the final heat treatment was chosen as 3.0C/min (5.4F/mi n) as a compromise between the cooling rates of these two samples. The final two parameters examined were the soak time and the temperature at which the sample was removed from the furnace for the fan air cool. A longer hold at the solution temperature was examined as well as a higher final temper ature. Micrographs for these heat treatment trials can be found be low in Figure 3-3. A2 was given a 2 h hold and a 3.3C/min (5.9F/min) cooling rate to 1020C (1868F). A1, on the other hand,

PAGE 81

64 A) B) C) Figure 3-3: Heat treatment trial samples with different soak times and final temperature before fan-air-cool. A) A2 (2 h so ak, 3.3C/min, and 1020C final temp.). B) A1 (4 h soak, 2.8C/min, and 1020C). C) C1 (2 h soak, 2.8C/min, and 1075C). was held at temperature for 4 h and cooled at a rate of 2.8C/min 5.0F/min) until 1020C (1868F). Sample C1 was soaked for 2 h a nd slow cooled at 2.8C/min (5.0F/min) until a final temperature of 1075C (1967F). The lo nger hold time had an adverse affect on the microstructure, prod ucing irregular shaped ’ precipitates. A dditionally, the higher final temperature showed signs of dendritic shaped ’, but ultimately did not allow enough time for the larger precipitates to grow From these results it was decided that the final heat treatment would have a 2 h hold and a final temperature of 1020C (1868F).

PAGE 82

65 The final heat treatment schedules for both the standard a nd alternate heat treatments are shown below in Figure 3-4. Once the final alternate heat treatment was A) B) Figure 3-4: Heat treatment sche dules. A) Standard heat tr eatment. B) Alternate heat treatment. determined, samples of each Master Powder Blend (MPB) for the mechanical testing were given both the standard and alternate h eat treatment as a check that these heat treatments work for all of them. Representa tive micrographs of the two heat treatments for these samples are contained in Figure 35. The standard heat treatment produced a similar microstructure to Figure 3-5a and Fi gure 3-5b in all of the MPBs. However, the alternate heat treatment pr oduced the desired, dendritic ’ microstructure in only eight of the eleven MPBs. The other thr ee (96SW865, 97SW984, and 98SW035) had microstructures similar to those found in Figure 3-6. The secondary ’ was irregular in shape. To counter this problem, the solu tion heat treatment schedule was lowered to 1165C (2129F). This lower solution temperat ure worked well and produced the desired microstructure in these three MPB’s (Figure 3-7).

PAGE 83

66 A) B) C) D) Figure 3-5: The standard and al ternate heat treatment microstr uctures. A) Standard heat treatment at 4300X magnifi cation. B) Standard heat treatment at 570X. C) Alternate heat treatment at 4300X. D) Alternate heat treatment at 570 X. A) B) Figure 3-6: MPB 96SW865 after solution he at treating at 1175C (2147F). A) 4300X magnification. B) 570X magnification.

PAGE 84

67 A) B) Figure 3-7: MPB 96SW865 after solution he at treating at 1165C (2129F). A) 4300X magnification. B) 570X magnification. Heat treatment The standard, subsolvus heat treatment was carried out in two separate batches. These samples were heated to 1100C (2012F) at 10C/min. They were left at this temperature for 2 h. Upon completion of the h eat treatment, they we re removed from the furnace and given a fan-air-cooling. The a lternate, supersolvus heat treatment was carried out in three batches. The first tw o batches were ramped to 1175C (2147F) at 10C/min. A two hour soak at this temp erature was followed by a controlled cool (3C/min) through the ’ solvus until the final temp erature of 1020C (1868F) was reached. At this temperature, the test bars were removed and give n a fan-air-cooling. The third batch consisted of the followi ng MPB’s: 96SW865, 97S W984, and 98SW035. Due to the differences noted between these three MPB’s and the remaining samples during the heat treatment trials, the so lution temperature was lowered to 1165C (2129F). All of the samples were then given the same aging heat tr eatment in a total of four batches. The first step of the age involved heating the samples to 760C (1400F); holding them for 8 h, then fan air cooling them back to room temperature. The final step was to heat the test bars to 650C (1202F) and hold them for 24 h before fan-air-cooling them to room temperature. Only one problem was encountered during these heat

PAGE 85

68 treatments. A few of the creep and tensile bars developed a significant oxide scale that eliminated these bars as possible test specimens It was determined that these particular bars had been sectioned so that they were either too close to the stainless steel HIP can or actually contained part of the can. New samp les were shipped from Crucible Compaction to the University of Florida for completion of the heat treatments. These samples were tested for HIP can remnants by passing a magnet over them. Any sample that still contained part of the HIP can would be ma gnetic, while the ones that only contained Alloy 720LI would not be magnetic. These sa mples were heat treat ed according to the schedules listed above. When all the heat treatments were completed, the test bars were sent back to Crucible to be machined into test specimens. Characterization After the heat treatments were complete a full characteriza tion of Alloy 720LI’s response to the heat treatments was carried out. First, the ’solvus temperature was determined as a comparison to the reporte d value from literature. Two methods of determining this temperature were used. The first was through differential thermal analysis (DTA) and the second was by metallo graphic inspection of quenched samples. Additionally, the ’ microstructure was fully eval uated through SEM, TEM, grain size measurements, volume fraction measurements, and hardness indents. Determination of ’ Solvus During the preliminary heat treatments, th ree of the eight master powder blends (MPBs) of Alloy 720LI had a different micros tructural response to the heat treatment than the other MPBs. This differenc e was thought to be related to the ’ solvus temperatures of the different master powder blends as there was no noticeable trend with

PAGE 86

69 chemistry. To verify this phenomenon and to determine the ’ solvus temperatures of the MPBs, samples were sent off for differential thermal analysis. MPB # 96SW863 and MPB # 96SW865 were chos en as representative samples of the MPBs thought to have higher ’ solvus and lower ’ solvus temperatures respectively. Two samples from each MPB were sectioned and sent for DTA analysis at two independent laboratories. The purpose of using two laboratories was to ensure the accuracy of the results. Th e results obtained from the DTA analysis performed at the M&P Laboratories are in Table 3-1. The “m aximum thermal effect” is the temperature Table 3-1: The DTA resu lts from M&P Laboratories 96SW863a 96SW863b 96SW865a 96SW865b Max Thermal Effect C (F) 1162 (2124) 1160 (2120) 1158 (2116) 1150 (2102) ’ soluvs C (F) 1192 (2178) 1192 (2178) 1200 (2192) 1193 (2179) Solidus C (F) 1279 (2334) 1279 (2334) 1279 (2334) 1280 (2336) Liquidus C (F) 1352 (2466) 1349 (2460) 1348 (2458) 1349 (2460) at which the rate of volume change in the ’ precipitates is at a maximum during heating [69]. This was measured at a local minimu m that occurred in the DTA curve, while the ’ solvus temperature was determined from a local maximum immediately following the “maximum thermal effect”. The solidus wa s measured at the change in slope that occurred immediately before the minimum in the graph. Finally, the liquidus was measured at this minimum. The results of the samples sent to Dirats Laboratories for DTA analysis are contained in Table 3-2. The ’ solvus temperature was measured at a local maximum in the derivative of temperatur e curve. On the other hand, the solidus and liquidus were measured in the same method as M&P Laboratories. Dirats

PAGE 87

70 Table 3-2: The DTA results from Dirats Laboratories 96SW863c 96SW865c ’ solvus C (F) 1182 (2160) 1185 (2165) Solidus C (F) 1251 (2284) 1235 (2255) Liquidus C (F) 1334 (1169) 1333 (2431) Laboratories did not report a value for the ma ximum thermal effect of the two samples; however, from the graphs the value can be estimated as approximately 1165C for sample 96SW863c and 1160C for sample 96SW865c. As can be seen from these two tables, M&P measured a lower temperature for th e maximum thermal effect, but a higher temperature for the ’ solvus temperature. All of the DTA graphs can be found in Appendix A. To more accurately determine the ’ solvus temperature for Alloy 720LI, the DTA data above was complimented with a meta llographic study of samples quenched from various temperatures. The first set of these samples was heated to temperatures at 5C (9F) intervals between 1100C (2012F) and 1200C (2192F). They were allowed to soak at the temperature for 1 h and then que nched in an iced-brine solution. These samples were then polished, etched, and examined on the SEM. The important temperature range for the solutioning of ’ was determined to be between 1120C (2048F) and 1140C (2084F) during this st udy. Figure 3-8 shows micrographs for samples heated to these temperatur es. At 1140C (2084F), all of the ’ was in solution. The only contrast present after etching was due to preferential attack of the grain boundaries and some small, disc rete carbides. The sample s heated to 1130C (2066F) and 1125C (2057F) exhibited coarse grain boundary ’ in addition to very small amounts of the secondary, intragranular ’ that had not completely gone into solution

PAGE 88

71 after 1 h. Finally, the sample heated to 1120C (2048F) had no secondary or primary ’ in solution. A) B) C) D) Figure 3-8: The ’ solvus trials quenched in iced brine after an hour at the solutioning temperature. A) 1140C (2084F) B) 1130C (2066F). C) 1125C (2057F). D) 1120C (2048F). Additionally, a number of samples were he ated to 1175C (2147F) and allowed to soak for 1 h to allow all second phase partic les to go into solution. These samples were then control cooled in the furnace to temp eratures at 5C (41F) intervals between 1120C (2048F) and 1140C (2084F). They were allowed to soak at this temperature for 1 h to allow all precipitation of secondary phases to occur. After 1 h, the samples were quenched in an iced-brine solution. Figure 3-9 contains micrographs of these samples. Once again, the sample quenched after soaking at 1140C (2084F) only had contrast due to preferential etching of the grain boundaries and discrete carbides. No ’

PAGE 89

72 A) B) C) D) Figure 3-9: The solvus trials that were fully solutioned a nd then control-cooled. A) 1140C (2084F). B) 1130C (2066F). C) 1125C (2057F). D) 1120C (2048F). had precipitated out at this temperature. After cooling to 1130C (2066F), the primary ’ had precipitated along the grain bounda ries. The 1125C (2057F) sample had no additional precipitation. However, the sa mple that was cooled to 1120C (2048F) before quenching exhibited secondary, intragranular ’ in addition to the primary ’. Microstructural characterization Small blocks of each master powder blend were heat treated along with the sample bars to ensure that the proper microstructure was achieved for every test specimen and to fully characterize the microstructures that we re produced. Representa tive micrographs of the two heat treatments afte r etching with Pratt & Whitney etchant #17 are included in Figure 3-10. As can be seen, the standard heat treatment produced smaller, more

PAGE 90

73 A) B) Figure 3-10: The / ’ microstructure in heat treated A lloy 720LI. A) The standard heat treatment. B) The alternate heat treatment. cuboidal, secondary ’ precipitates compared to the larger, dendritic ’ produced by the alternate heat treatment. Both samples contained large primary ’ precipitates along the grain boundaries as seen in the far left of Figure 3-10A and th e far right of Figure 3-10B. The ultrafine, tertiary ’ is readily evident in the standard heat treatment samples. However, a FEG SEM was needed to view these tertiary precipitates in the alternate heat treatment because of their smaller size and the better resolution capabilities of this instrument. The ’ volume fraction and the average ASTM grain size were measured as detailed in Chapter 2. The ’ volume fraction was calculated on specimens etched with Pratt & Whitney etchant #17. This etchant did not reveal the grain boundary morphology as well as it revealed the / ’ microstructure. As a result, wate rless Kallings solution was used to etch the grain boundaries for the grain size measurements. Figure 3-11 shows micrographs of both the standard and alte rnate heat treatments after etching with waterless Kallings reagent. The standard he at treatment has a smaller and more uniform grain size distribution, while the alternate heat treatment has la rger and more varied grain

PAGE 91

74 A) B) Figure 3-11: Grain boundary microstructure of heat treated Alloy 720LI. A) The standard heat treatment. B) The alternate heat treatment. size distribution. The data and calculations for the ’ volume fraction measurements, mean ’ size, and the ASTM average grain size can be found in Appendix B. Table 3-3 shows the results of all of these calculations. While both heat treatments result in the same ’ volume fraction, the alternate heat trea tment has a larger average ASTM grain size than the standard heat treatment. Add itionally, the porosity of the as-HIP material was measured on a polished and unetched sample to be 0.3 0.3% by volume. Table 3-3: The total ’ volume fraction, and ASTM grain size Heat Treatment ’ volume fraction ASTM Grain Size Standard 50.7 11. 8 Alternate 49.6 8.2 Hardness tests were conducted on heat tr eated samples after the solution heat treatment and after the aging heat treatment. The results of the Rockwell C hardness tests are contained in Table 3-4 below. While bot h heat treatments produced an increase in Table 3-4: Results from Rockwell C ha rdness tests for both heat treatments Heat Treatment as-HIP Solution H.T. Aging H.T. Standard 42.5 0.3 45.8 0.3 Alternate 37.8 0.6 39.9 0.2 42.9 0.2 the hardness of the alloys, the standard h eat treatment has a greater hardness than the alternate heat treatment both before and afte r the aging heat treatment. The statistical

PAGE 92

75 analysis for the hardness testing can be found in Appendix B. All of these data sets are statistically unique from each ot her. The increase in hardne ss after aging indicates that both heat treatments successfully produced the ultrafine tertiary ’. The data and statistical calculations for the hardness te sting are also contained in Appendix B. Mechanical Testing and Fracture Analysis Low Cycle Fatigue Numerous problems were encountered when attempting to test samples in LCF. The first difficulty was with the push/pull rods initially used for the testing. A test was setup with the same push/pull rods that had b een used during the creep and tensile testing. These were 27.3 cm (10.75 in) long and 1.91 cm (0.75 in) in diameter. The first sample was loaded, the test conditions were input into the computer, and the furnace was heated to the test temperature. However, when the test was started, the sample broke on the first cycle. After examination of the deformed sample, it was obvious that the test had been run without proper sample alignment (Figure 3-12). One of the push/pull rods and the extensometer frame were bent beyond repair. This was probably the result of the sample being off-axis during the compression cycle. It was necessary to design new push/pull rods specifically for fatigue testing that had a smaller length-to-diameter ratio to eliminate any potential bending during compre ssion. The new design also contained a lip that would be flush against the actuator so that any potential deflection due to slack in the threads would be minimized. These new ba rs were cast by Special Metals (Figure 3-13a). They were then shipped to Low Stress Grind to be machined into the new push/pull rods. Figure 3-13b shows the new push/pull rods used for fatigue testing.

PAGE 93

76 Figure 3-12: Low cycle fatigue sample that was misaligned and tested in compression A) B) Figure 3-13: Push/pull rods. A) Ascast B) Machined into final form. After a lengthy setback due to the produc tion of these new push/pull rods, testing was resumed. The new push/pull rods worked better, but some additional issues still needed to be resolved. Anytime a sample was loaded, the test w ould begin properly, but it was obvious after a few cycles that the ma ximum and minimum load s were not at the expected values. Either the extensometer frame was slipping on the sample itself or the extensometer was slipping on the extensometer frame. In this setup, round knife-edge inserts attached directly to the polished specimen surface. This worked well for the

PAGE 94

77 initial cycle; however, the knife -edges would slip with each cy cle. To counteract this, two fatigue samples were sent to Low Stress Grind to machine a notch into the shoulder for the inserts to “hold onto” during the fa tigue testing. Additionally, notches were machined onto the extensometer frame where the extensometer attaches to ensure that the extensometer itself would not slip. When th e new notched specimens were received, they were loaded into the frame and tested. The system appeared to control the strain well on these tests, however they both broke at the no tch, invalidating the data accumulated from these tests. Two more samples were then sent to Low Stress Grind to machine a notch just after the threaded region of the sample Figure 3-14 shows one of these samples. Figure 3-14: Low cycle fatigue sample with notches machined just after the threaded region and before the shoulder of the sample Larger diameter (12.7 mm) insert s were also ordered from Satec for these new samples. This new set-up appeared to work well on th ese two samples. At this point, the remaining samples were shipped to Low Stre ss Grind to have similar notches machined onto their shoulders. Notch

PAGE 95

78 A number of tests were run to failure a nd some to run-out at 1,000,000 cycles. The computer appeared to be controlling stra in well; however, the maximum and minimum stresses and the amount of plastic strain re ported during the tests were different from what was expected. At this point, the total le ngth of the tested samples was measured to determine the actual deformation that they ha d endured. These values were much lower than those reported by the fatigue software While the notches had minimized the slippage, they still had not completely eliminated it. It was decided to test the remaining samples in high cycle fatigue (HCF) rather th an waste any more fatigue specimens. The LCF fracture surfaces were examined as they could provide some valuable insight into the effect of the different heat treatments and their resulting micr ostructure on the crack initiation type and propagation mode. Three samples failed for each heat treatment, all at 538C (1000F). The strain ranges were supposed to be either 1.0% or 1.1% for all of these tests. However, due to the complications listed above, these strain ranges were not accurate for the strain actually seen by the samples. All three stan dard heat treatment samples that fractured failed from processing defects. Two of th e samples failed from ceramic inclusion initiation types as seen in Figure 3-15. The third standard heat treatment sample contained two different types of defects that initiated cracks. These can be seen in Figure 3-16. The first defect was in the middle of the sample. This initiation point had too much mechanical damage to determine the type of defect. The sec ond initiation (ceramic inclusion) was located at the bot tom of the sample as seen in the overview micrograph. It was inside the sample, but very near to the surface. This initia tion was responsible for the eventual fast fracture and failure of this sample. It is important to note that faceted

PAGE 96

79 Figure 3-15: Ceramic inclusion type defect in standard heat treatment sample tested at 538C (1000F) and a strain range of 1.1% A) B) C) Figure 3-16: Standard heat treatment sample tested at 538C (1000F) and a strain range of 1.0%. A) The overview of the fracture surface. B) Initiation from unknown type of defect. C) Initiation from a ceramic agglomerate. grains were found near the vari ous defects for the standard h eat treatment. These grains are examples of slip band cracking; however, they did not function as the crack instigator.

PAGE 97

80 In contrast to the standard heat treatmen t samples, the alternate heat treatment LCF fracture surfaces all had faceted grains as the initiation. Thes e cracks all initiated inside the sample, but near the surface. A represen tative micrograph of one of these initiation sites is in Figure 3-17. The arrows in th ese micrographs indicate the direction of crack propagation. The grain that served as the initiation point also ha s a twin boundary that A) B) C) D) Figure 3-17: Alternate heat treatment sample tested at 538F (1000C) and a strain range of 1.0%. A) 570X magnification. B) 285X. C) 115X. d) 57.5X. resulted from the deformation to the grain. The lower magnificati on fractographs show secondary cracks propagating radial ly outward from this initia tion point. After initiation, the cracks propagated in a similar manner for both the standard and the alternate heat treatments. Immediately after the initiation re gion, there was a flat relatively featureless region of crack growth as can be seen in Fi gure 3-18. This region of the crack surface

PAGE 98

81 A) B) Figure 3-18: Relatively flat featureless region just after crack initiation in low cycle fatigue. A) Standard heat treatment B) Alternate heat treatment. extended approximately one millimeter from th e initiation in all directions. It was marked by discoloration on the fracture surf ace due to longer exposure to the oxidizing atmosphere during testing. At the edge of this discolored region, the surface transitions from a featureless region to what can best be described as an ar ea of undulating peaks and valleys. These features persists until either the specimen surface or shear lips. As the crack moves further from the discolored regi on, these “hills and va lleys” become more and more pronounced. There appears to be a si ze difference between the features for the two heat treatments in this area. Figure 319 shows representative micrographs of this area of the propagation paths. High Cycle Fatigue After the initial complications with th e LCF testing, high cycle fatigue (HCF) tests were conducted to determin e the effect of the supersol vus heat treatment on fatigue lifetime and crack initiation type compared to the standard subsolvus heat treatment. A test matrix was setup to determine the role of microstructure in the fatigue failure of the remaining samples. Tests were performed with an R ratio ( min/ max) of 0.1 at 538C (1000F) and 649C (1200F) and at three diffe rent stress ranges: 993MPa (144Ksi),

PAGE 99

82 A) B) Figure 3-19: Representative micrographs of the fast fracture region of LCF samples. A) Standard heat treatment. B) Alternate heat treatment. 1034MPa (150Ksi), and 1086MPa (157.5Ksi). A ll tests were run at a frequency of 10 Hz. The fracture surfaces were then examined to evaluate the fracture features and to determine the crack initiation type and the mode of crack propagation. HCF results The HCF testing was completed without any of the problems encountered during LCF testing. All of the tests failed be fore 1,000,000 cycles. There were three extra fatigue specimens (two standard heat trea tment and one alternate heat treatment) remaining after all test cond itions had been examined. Af ter examining the HCF test results, these samples were test ed at conditions to determine the amount of data scatter in fatigue properties and when test results in dicated a longer or s horter than expected lifetime. The stress range is pl otted versus the number of cy cles to failure on a log scale for both heat treatments at 538C (1000F) in Figure 3-20. At the low and middle stress ranges of 993 MPa (144 Ksi) and 1034 MPa (150 Ksi), the standard heat treatment had superior fatigue lifetimes. However, at the highest stress range of 1086 MPa (157.5 Ksi), the alternate heat treatment ha d the longer fatigue lifetime. Similar results were found for the testing performed at 649C (1200F). (Figur e 3-21) Once again, the standard heat

PAGE 100

83 980 1000 1020 1040 1060 1080 1100 10,000100,0001,000,000 NfStress Range (MPa) Alternate H.T. Standard H.T. Figure 3-20: High cycle fatigue S-N curves for specimens tested at 538C (1000F) treatment exhibited the longer lifetimes at th e low and middle stress ra nges, but not at the highest stress range. This data would suggest that there is a change in crack initiation or propagation behavior at higher stresses that increases the fatigue lifetime of the alternate heat treatment compared to the standard heat treatment. Analysis of the fracture surfaces would evaluate this phenomenon. It is also interesting to plot the S-N curv es for the two different test temperatures for each heat treatment. Figure 3-22 contains a pl ot of the standard heat treatment at both 538C (1000F) and 649C (1200F). The lowe r test temperature exhibited longer fatigue lifetimes at both the high and the middl e stress ranges. However, as the stress range went below about 1000 MPa, the higher test temperature had the superior lifetime.

PAGE 101

84 The same type of relationship between test te mperature, stress range, and cycles to failure was seen in the alternate heat tr eatment as well (Figure 3-23). 980 1000 1020 1040 1060 1080 1100 100001000001000000 NfStress Range (MPa) Alternate H.T. Standard H.T. Figure 3-21: High cycle fatigue S-N curv es for specimens tested at 649C (1000F) HCF fractography After failure, the fractured surfaces we re examined using both an optical microscope and a SEM. The fracture surfaces of both heat treatments looked similar when first examined visually. Optical pict ures of fracture surfaces from standard and alternate heat treatments are contained in Fi gure 3-24. The crack initiation points on the fracture surfaces of the HCF sa mples could readily be determined. Circular regions of discoloration were present at th e point of origin for all samp les, due to being exposed to the oxidizing environment for longer times than the fast fracture area. The samples in Figure 3-24 are situated so that the crack or igin is at the bottom of the picture.

PAGE 102

85 980 1000 1020 1040 1060 1080 1100 100001000001000000 NfStress Range (MPa) 538C 649C Figure 3-22: High cycle fatigue S-N curves for tests of the standard heat treatment at both 538C (1000F) and 649C (1200F) Additionally, the region of slow crack growth was visible and ranged in shape fromsemi-elliptical to circular. This can be seen at the bottom of both of the optical pictures in Figure 3-24. The fatigue sample s had small shear lips on the sides of the sample leading away from the initiation site. After the fracture surfaces were examin ed optically, they were viewed on the SEM to determine the crack initiation poi nts and the surface features at higher magnifications. The crack initiation points vari ed in appearance from sample to sample, but there were in general three different type s of initiations: ceramic inclusions, primary carbides, and slip band cracking. None of the samples failed from cracks that initiated at PPB particles. Table 3-5 lists the fatigue lif etime and crack initiati on type for each of the HCF specimens.

PAGE 103

86 980 1000 1020 1040 1060 1080 1100 10,000100,0001,000,000 NfStress Range (MPa) 538C 649C Figure 3-23: High cycle fatigue S-N curves for tests of the alternate heat treatment at 538C (1000F) and 649C (1200F) A) B) Figure 3-24: Optical micrographs of HCF fr acture surfaces of samples tested at 649C and 1086 MPa. A) Standard heat treatm ent. B) Alternate heat treatment. Ceramic inclusions can be rather dele terious to the fatigue lives of Ni-base superalloys. They can be present in the form of either a large blocky inclusion or an agglomerate of small ceramic particles. Figure 3-25 shows a typi cal fracture initiation site at a ceramic inclusion. When viewed w ith backscattered electron imaging (BSE) in

PAGE 104

87 Table 3-5: The HCF test resu lts and crack initiation types Sample Temperature C (F) a MPa (Ksi) Nf Initiation Type Standard H.T. ULA5 538 (1000) 993 (144) 330,536 Ceramic(Al2O3) ULA15 538 (1000) 993 (144) 432,715 Al2O3 ULA11 538 (1000) 1034 (150) 331,895 Al2O3 ULA14 538 (1000) 1034 (150) 371,796 Al2O3 ULA16 538 (1000) 1086 (157.5) 22,828 Al2O3 ULA3 649 (1200) 993 (144) 661,396 Al2O3 ULA13 649 (1200) 1034 (150) 96,859 Al2O3 ULA10 649 (1200) 1086 (157.5) 14,244 TiC? Alternate H.T. ULB7 538 (1000) 993 (144) 85,646 Crystallographic ULB10 538 (1000) 1034 (150) 64,431 Crystallographic ULB17 538 (1000) 1086 (157.5) 45,868 Unknown ULB6 649 (1200) 993 (144) 181,975 Crystallographic ULB13 649 (1200) 1034 (150) 13,994 Al2O3 ULB15 649 (1200) 1034 (150) 19,629 grain ULB9 649 (1200) 1086 (157.5) 17,324 Al2O3 A) B) Figure 3-25: Ceramic agglomerate as initiati on point in standard heat treatment sample tested at 538C (1000F) and a stre ss range of 1034 MPa (150 Ksi). A) Secondary electron imaging (SE). B) B ackscattered electron imaging (BSE). EDS showed these to be Al2O3 inclusions. the SEM, these defects showed contrast due to atomic number difference between the particles and the matrix. This type of init iation was found mostly in the standard heat treatment at both test temperatures. Only two of the alternate h eat treatment samples contained this type of initia tion. These two specimens were tested at the middle and high stress ranges respectively at 649C (1200F). Energy dispersive spectrometry (EDS) was

PAGE 105

88 performed on these agglomerates to reveal the approximate compositi on of these particles (Figure 3-26). The ceramic agglom erates were determined to be Al2O3. Some samples Figure 3-26: EDS spectra of crack initiation fo r standard heat treatment sample tested at 538C and 1034 MPa contained regions of the fract ure initiation sites that were very round and were first thought to be PPB defects. However, upon cl oser inspection using backscattered electron imaging and EDS, it was revealed that these were also Al2O3 inclusions. (Figure 3-27). Once again, the phase identification was verified using BSE imaging and EDS. A) B) Figure 3-27: Ceramic agglomerate as initiati on point for standard heat treatment sample tested at 538C (1000F) and a stress range of 1086 MPa (157.5 Ksi). A) SE imaging. B) BSE imaging. The ceramic agglomerate was identified as Al2O3. The second main type of initiation site observed on the HCF fracture surfaces was associated with slip band cracking. It is ch aracterized by faceted feat ures that appear as brittle fracture on first analys is. Figure 3-28 shows fractographs typical of this type of failure. While slip band cracking served as th e source of crack initiation in the alternate

PAGE 106

89 Figure 3-28: Slip band cracking as initiation point an alternate heat treatment sample tested at 538C (1000F) and 993 MPa (144 Ksi) heat treatment and not in the standard heat treatment, they were present surrounding most of the standard heat treatment crack initia tion points. The slip band cracking features were larger in the alternate he at treatment due to the larger grain size. Secondary cracks were evident leading radially away from these initiation sites. The final type of initiation s ite encountered in this study was at a Ti-rich particle. The only sample to show this type of crack initiation was the standard heat treatment sample tested at 649C (1200F) and 1086 MPa (157.5 Ksi). This failu re started at the specimen surface as seen in Figure 3-29. A crack was observed running through the A) B) Figure 3-29: Primary carbide as the initiation type for the st andard heat treatment sample tested at 649C (1200F) and 1086 MPa ( 157.5 Ksi). A) SE imaging. B) BSE imaging.

PAGE 107

90 length of the particle, as well as many seconda ry cracks leading radi ally away from the origin. As can be seen, this precipitate a ppears darker than the matrix phase during BSE imaging, indicating a lower atomic number. ED S revealed that this particulate had large amounts of titanium. The EDS spectrum for this particle is in Figure 3-30. This particle maybe a ceramic inclusion or even a TiO partic le; however, it is unclear at this point the exact nature of this particle. Figure 3-30: EDS spectra of Ti-ri ch particle as initiation poin t of standard heat treatment sample tested at 649C and 1086 MPa There were a few samples that it was not po ssible to determine the type of fracture initiation. Figure 3-31 shows a clear initiation feature for the alternate heat treatment sample tested at 649C (1200F) and 1034 MPa (150 Ksi). The BSE image did not reveal any noticeable atomic number difference nor did th e EDS spectra conclusively determine what type of particle this was. In addition, the initiation type for the alternate heat treatment sample tested at 538C ( 1000F) and 1086 MPa (157.5 Ksi) could not be determined (Figure 3-32). One final interesting note on the crack initiation points is that the samples which had internal crack initiation always contained slip band cr acking in the area immediately surrounding the crack origin, whereas those that failed at the surface did not have any slip

PAGE 108

91 A) B) Figure 3-31: Unknown initiation ty pe for alternate heat treatm ent sample tested at 649C (1200F) and 1034 MPa (150 Ksi). A) SE imaging. B) BSE imaging. Figure 3-32: Crack origin for alternate heat treatment sample tested at 538C (1000F) and 1086 MPa (157.5 Ksi) band cracking around the defect. Th is held true for both heat treatments and can be seen in the micrographs in Figure 3-33. In addition to examination of the fractur e initiation types, the fracture surface features along the propagation path were examined in the SEM to determine the crack propagation path. Micrographs were taken at three different locations along the fracture path. Inside the region of discoloration for each sample, the fracture surface appeared flat and relatively featureless as can be seen by representative microgra phs in Figure 3-34. This portion of the fracture surface has the same appearance for all test conditions for both heat treatments. In some areas, fatigue st riations were present. Unfortunately, these

PAGE 109

92 A) B) C) D) Figure 3-33: Initial crack regi on of fatigue samples. A) Standard heat treatment that failed internally. B) Standa rd heat treatment that failed at the surface. C) Alternate heat treatment that failed intern ally. D) Alternate heat treatment that failed at the surface. areas are not distinct enough to measure the striation spaci ng and eventually approximate the fracture toughness of the two heat treatments. The region immediately after th e discolored area displaye d a noticeable change in microstructure. The crack path became much more tortuous at this point. The surfaces transitioned from being flat and featureless to being undulat ing and more indicative of ductile fracture. Figure 3-35 is a represen tative fractograph from this area of the specimen surface. Both heat treatments resu lted in similar features indicative of interparticle or intergranular failure. These interparticle/intergranular features were present throughout the remainder of the fracture surfaces up to the point of fast fracture.

PAGE 110

93 A) B) Figure 3-34: Fracutre surface near initiation in standard heat treatment sample tested at 538C (1000F) and a stress range of 1034 MPa (150 Ksi). A) 570X. B) 2850X. Figure 3-35: Fracture surface of standard he at treatment tested at 538C (1000F) and a stress range of 1086 MPa (157.5 Ksi) outside of discolored region However, there appears to be a slight size difference for the features between the two heat treatments. The alternate heat treatment surfaces appear to ha ve larger features that are more varied in size than the standard heat treatment surfaces. These differences can be seen in Figure 3-36. The indi vidual test conditions, however do not appear to have an effect on the surface features of the propa gation path within each heat treatment. In order to complete a detailed study of the fracture mechanics during the fatigue testing of the two heat treatments, the fract ure surface features were measured from SEM micrographs of all of the surfaces. The initial area of crack growth (Stage I) before the

PAGE 111

94 A) B) Figure 3-36: Fracture surfaces far from origin of samples tested at 538C (1000F) and 993 MPa (144 Ksi). A) Standard heat treat ment. B) Alternate heat treatment. transition to transgranular cr ack growth was measured in addition to the area of total crack growth before fast fracture. These m easurements can be found in Table 3-6. In Table 3-6: Area of crack growth regions on the HCF fracture surfaces Sample Temp (C) (MPa) max (MPa) Nf Ai ( m2) Af (mm2) Standard Heat Treatment ULA5 538 993 1103 330536 22000 1.3 ULA15 538 993 1103 432715 26000 1.5 ULA14 538 1034 1149 371796 14000 1.4 ULA11 538 1034 1149 331896 15000 1.4 ULA16 538 1086 1207 22829 5000 1.3 ULA3 649 993 1109 661397 22000 2.5 ULA13 649 1034 1149 96859 15000 2.2 ULA10 649 1086 1207 14245 7000 1.7 Alternate Heat Treatment ULB7 538 993 1103 85647 41000 1.5 ULB10 538 1034 1149 64432 16000 1.4 ULB6 649 993 1103 181979 56000 2.0 ULB15 649 1034 1149 19629 13000 1.4 ULB13 649 1034 1149 13994 16000 1.4 ULB9 649 1086 1207 17324 11000 1.4 this table, Ai refers to the area of the initial crack before transgranular crack growth, and Af refers to the final area of sl ow crack growth just prior to overload and fast fracture. These two areas are plotted versus stress range in Figures 3-37 and 3-38. The alternate

PAGE 112

95 980 1000 1020 1040 1060 1080 1100 00.000000010.000000020.000000030.000000040.000000050.00000006 Initial Crack Area (m2)Stress Range (Mpa) Standard 538C Standard 649C Alternate 538C Alternate 649C Standard 538C Standard 649C Alternate 538C Alternate 649C Figure 3-37: Initial crack ar ea versus stress range for the high cycle fatigue specimens heat treatment has a larger area of slow crack growth before the transition to transgranular crack growth for both test temp eratures and all stress ranges. The final crack size increases with decreasing stress ra nge for both heat treatments at both test temperatures. If the specimen fractures at the max load of the fatigue cycle, then it would follow that all of the specimens fractured at the same critical stress intensity value (KC) for each test temperature. A smaller stress range would require a larger crack before the stress intensity factor reaches KC. Additionally, the two heat treatments resulted in similar areas for the slow crack growth re gion at 538C. However, at 649C, the standard heat treatment had a significantly larger final area of slow crack growth. After careful analysis of the fracture surf aces was complete, the fractured ends of the specimens were sectioned, mounted longitudi nally, polished, and etch ed to reveal the

PAGE 113

96 980 1000 1020 1040 1060 1080 1100 00.00000050.0000010.00000150.0000020.00000250.000003 Final Crack Area (m2)Stress Range (Mpa) Standard 538C Standard 649C Alternate 538C Alternate 649C Standard 538C Standard 649C Alternate 538C Alternate 649C Figure 3-38: Final crack area versus stress range for the high cycle fatigue specimens path of the secondary cracks through the mi crostructure. The samples from both heat treatments tested at both test temperatur es and a stress ranges of 993 MPa (144 Ksi) exhibited secondary cracks at the bounda ry between the primary and secondary ’ and the matrix. This can be seen in Figure 3-39. The arrows in these pictures indicate the tensile direction. The alternate heat treatme nt samples exhibited secondary cracks that formed in lines through the grains, while th e standard heat treatment sample’s cracks appeared to be more randomly placed. At both test temperatures at the stress range of 1034 MPa (150 Ksi), samples from the two heat treatments had different seconda ry crack morphologies. These can be seen in Figure 3-40. The standard heat treatment samples still exhibited secondary cracks at the primary and secondary ’ precipitates; however, now these cracks appear to line up in

PAGE 114

97 A) B) Figure 3-39: Secondary crack s at primary and secondary ’ precipitates in samples tested at 538C and 993 MPa. A) Standard h eat treatment. B) Alternate heat treatment. The arrows show the tensile direction. A) B) C) D) Figure 3-40: Secondary cracks in fatigue samples tested at 1034 MPa. A) Standard heat treatment at 538C. B) Standard heat treatment at 538C again. C) Alternate heat treatment at 538C. D) A lternate heat treatment at 649C. a curved manner through the grains. In cont rast, the alternate heat treatment samples exhibited the same behavior as in the low stress range. The seconda ry cracks occurred at Cracks

PAGE 115

98 both primary and secondary ’ precipitates, but they did were not aligned in rounded shapes. Instead, they appeared to form in straight lines through the grains. Finally, at the highest stra in range of 1086 MPa (157.5 Ksi) samples from both heat treatments exhibited seconda ry cracks at TiC and Al2O3 particles. As with the other stress ranges, there were also cracks at the primary and secondary ’ precipitates. There were not many long cracks present. Some of these propagated al ong the grain boundaries and some formed rounded shapes within the gr ains. Both heat treatments exhibited both of these types of secondary cracks. All of these features can be seen below in Figure 3-41. A) B) C) D) Figure 3-41: Fatigue samples tested at 538C and 1086 MPA showing secondary cracks. A) At a TiC in standard heat trea tment. B) Along grain boundaries in standard heat treatment. C) At a Ti C in alternate heat treatment. D) Transgranularly in alternate heat treatment. Crack Voids

PAGE 116

99 Tensile Testing The tensile testing was performed at room temperature, 538C (1000F), 649C (1200F), and 760C (1400F) on an Instron servohydraulic test frame. Alloys designed as turbine disk materials need excellent tensile strength at elevated temperatures. An improvement in one mechanical property can have an adverse impact on the other mechanical properties. Therefore, it was n ecessary to determine if the alternate heat treatment would not result in too la rge of a debit in tensile strength. Tensile results Table 3-7 shows the results of the tensile testing perf ormed in this study. The extensometer slipped on some of the samples after yielding. As a result, the data for percent strain at UTS was not available. In addition, the data for samples UCA1 and Table 3-7: Tens ile test results Sample Temp. (C) Modulus (GPa) Yield Strength (MPa) UTS (MPa) Strength at failure (MPa) Total Elongation (%) Standard Heat Treatment UCA4 RT 228 1225 1710 1675 23.3 UCA23 RT 234 1241 1696 1669 21.1 UCA24 538 172 1078 1510 1464 23.3 UCA5 649 181 1087 1471 1324 23.0 UCA7 649 156 1103 1476 1351 22.4 UCA8 760 174 1018 1180 1103 11.1 UCA10 760 171 1069 1145 1076 10.7 Alternate Heat Treatment UCB4 RT 224 990 1524 1517 27.6 UCB21 RT 225 1066 1568 1558 25.3 UCB2 649 212 880 1367 1296 29.4 UCB5 649 185 887 1395 1317 26.0 UCB3 760 112 827 1123 1034 17.1 UCB8 760 165 780 1103 993 17.4 UCB1 that appear in the original matrix in Table 2-6 was lost due to a software error during testing. There were a few extra samp les that were machined with the creep geometry. To make up for the lost tensile samp les, these two extra samples were tested at

PAGE 117

100 room temperature. Also, sample UCA24 was tested at 538C. The standard heat treatment had greater yield stress, ultimate te nsile strengths, and fa ilure strengths at all conditions compared to the al ternate heat treatment. However, the standard heat treatment tensile samples displayed a corres ponding deficit in strain at maximum load and elongation at failure. All of the elevated temperature tensile te sts for both conditions had a slight decrease in strength before failure, denoting that th e samples experienced some amount of necking prior to failure. The room temperature tests had a smaller decrease in strength. The alternate heat treatment samples basically showed no evidence of necking. The standard heat treatment specimens showed very minimal amounts of necking. Figure 3-42 shows the decrease in yiel d strength with temperature for both heat treatments. As can be seen, the yield strength at 649C (1200F) is less than that at room 1233 1028 1095 883.5 1043.5 803.5 0 200 400 600 800 1000 1200 1400 Yield Strength (MPa) 25649760 Temperature (C) Standard H.T. Alternate H.T. Figure 3-42: Yield strength vs. te mperature for both heat treatments

PAGE 118

101 temperature. As the temperature is in creased to 760C (1400F), the yield strength decreased even further. The elongation stay ed constant up to 649C (1200F) (Figure 343). However, there was a precipitous drop in elongation at 760C (1400F). Representative stress-strain curves are shown for each heat treatment for a sample tested at 760C (1400C) in Figure 3-44 and 3-45. 22.2 26.5 22.7 27.7 10.9 17.3 0 5 10 15 20 25 30 Elongation (%) 25649760 Temperature (C) Standard H.T. Alternate H.T. Figure 3-43: Elongation at failure vs. temperature for both heat treatments Tensile fractography Fractured tensile samples from each test condition were analyzed on the SEM to determine crack initiation type, if possible, and crack propagati on paths. Figure 3-46 shows the crack initiation region for one sample at each test temperature for the standard heat treatment. The samples tested at r oom temperature and 649C (1200F) both show a

PAGE 119

102 0 20 40 60 80 100 120 140 160 180 0246810121416 Tensile strain (%)Tensile stress (Ksi) Figure 3-44: Tensile stress-strain curve for standard heat treatment sample tested at 1400F 0 20 40 60 80 100 120 140 160 180 0510152025 Tensile strain (%)Tensile stress (Ksi) Figure 3-45: Tensile stress-strain curve for alternate heat treatment sample tested at 1400F

PAGE 120

103 A) B) C) Figure 3-46: Crack initiation region in standard heat treatment tensile samples. A) Tested at room temperature. B) Test ed at 649C (1200F). C) Tested at 760C (1400F). distinct crack initiation point. EDS analysis of this initiation di d not have enough signal to determine the composition of the initiation since the defect was well below the fracture surface. However, the fracture surfaces of these samples were similar to those of the HCF fracture surfaces. Therefore, it is t hought that this initiation is due to an agglomerate of Al2O3. On the other hand, the sample tested at 760C (1400F) did not show a unique point of origin. Crack propagati on features pointed ba ck to a region near the surface; however, it was not possible to determine the type of crack origin on this sample. While it was easy to locate the crack origin on the alternate heat treatment tensile samples, it was not possible to determine the type of fracture origin for any of these

PAGE 121

104 samples. Micrographs of the origin area for each of the tensile test temperatures are shown in Figure 3-47. The samples tested at room temperature and 760C (1400F) A) B) C) Figure 3-47: Crack initiation re gion in alternate heat treatment tensile samples treatment. A) Room temperature. B) 649C (1200F). C) 760C (1400F). failed from the propagation of a crack that be gan near the surface. One of the samples that were tested at 649C (1200F) (Figure 3-47b ) actually had at least two distinct origin points. The surface had a fan-shape appear ance indicating that multiple cracks were propagating through the sample at failure. In Figure 3-47b, one clear initiation point can be seen in the lower left corner of the speci men, with another directly in the middle of the sample. Secondary cracks can be seen propa gating radially from bot h of these initiation points.

PAGE 122

105 Creep Testing Creep testing was carried out at several different temperature and load conditions on both the standard and the alte rnate heat treatments. Whil e fatigue properties are more critical for turbine disk materi als, it is still important to maintain good creep strength especially in the rim (outer edge connected to the turbine blades). However, generally an increase in tensile or fati gue properties for a material is accompanied by a corresponding decrease in creep properties. This creep te st matrix was designed to insure that any improvements in tensile or fatigue propertie s through the alternate heat treatment would not result in a deficit in creep rupture lifetime. Creep results The results of the creep testing performed in this study are contained in Table 3-8. For the standard heat treatment, the conditi on that produced the longest lifetime was at 677C (1250F) and 793 MPa (115 Ksi). The shorte st creep rupture lifetime for this heat treatment was found at 677C (1250F) and 1034 MP a (150 Ksi). In contrast, the test conditions that produced the longest creep lifet imes for the alternate heat treatment were 732C (1350F) and 483 MPa (70 Ksi) and the shortest lifetimes were 677C (1250F) and 1034 MPa (150 Ksi). It is also evident th at the alternate heat treatment produced a much more ductile material than the sta ndard heat treatment, as the elongation was greater at all test condi tions for the alternate h eat treatment. It should also be noted that generally longer creep lifetimes corresponded to lower elongation values for both heat treatments. The creep frames additionally recorded time to 0.1%, 0.2%, 0.5%, 1.0%, 2.0%, and 5.0% creep for each of the tests. This data was compiled to produce Larson-Miller (L-M) plots for each of these times as well as fo r time to creep rupture. L-M curves show

PAGE 123

106 Table 3-8: The creep results Sample Temp. (C) Stress (MPa) Tf (Hours) Elongation % Standard Heat Treatment UCA12 677 1034 11.1 6.3 UCA16 677 1034 17.1 7.0 UCA9 677 793 396.8 5.9 UCA11 677 793 362.7 7.2 UCA13 704 689 143.9 8.5 UCA18 704 689 128.9 10.8 UCA20 732 483 134.5 4.9 UCA21 732 483 157.9 7.1 UCA19 760 483 25.7 3.6 UCA22 760 483 34.4 7.6 Alternate Heat Treatment UCB11 677 1034 5.8 13.0 UCB18 677 1034 4.8 14.7 UCB7 677 793 288.4 10.9 UCB10 677 793 233.4 16.0 UCB6 704 689 148.0 17.5 UCB12 704 689 220.9 14.0 UCB19 732 483 452.1 7.4 UCB20 732 483 559.4 9.1 UCB14 760 483 112.0 9.7 UCB23 760 483 126.2 11.0 stress on a log stress scale versus the L-M parameter, P, on a traditional scale. This parameter is calculated using the equation: P = Ta [Log (tr + 20)] where Ta is the absolute test temperature and tr is the time to the percent creep for that particular plot. In such a plot, the higher te mperature and longer lifetime tests will be at greater values of the P parameter. This provides a succinct way of comparing creep lifetimes for different microstructures of the sa me alloy or for different alloys altogether. Data points with larger P values represent longer lifetimes and gr eater temperatures. Therefore, given the same stre ss condition and test temperature, points that lie to the right on the plot have better creep properties than those that lie to th e left of the graph.

PAGE 124

107 The Larson-Miller plot for rupture is s hown below in Figure 3-48. The alternate heat treatment has the best lifetime for creep at 704C (1300F) and 689 MPa (100 Ksi), 732C (1350F) and 483 MPa (70 Ksi), and 76 0C (1400F) and 483 MPa (70 Ksi) for all strain values. These are the points that lie on the far right of the graph. However, at the lower temperature and higher stress c onditions the standard heat treatment outperforms the altern ate heat treatment. 100 1000 10000 1919.52020.52121.52222.52323.5P=Ta[log(tR)+20]Stress (MPa) Standard Heat Treatment Alternate Heat Treatment Figure 3-48: Larson-Miller plot for creep rupture in Alloy 720LI Creep fractography To fully understand the effect of the two mi crostructures on the creep properties of this alloy, it was necessary to examine the fracture surfaces of creep samples from each test condition for each heat treatment. Th erefore, the surfaces were examined on the SEM. If possible, failure origins were noted and analyzed. Additionally, the

PAGE 125

108 microstructure of the propagating crack path was inspected to determine the propagation mode. In general, the creep fracture surfaces were not as straight forward as the fatigue or tensile sample surfaces. Discoloration due to oxidation was again used to locate the origin region for most of these samples; how ever, it was not as easy to determine the type of origin for the creep samples as it was with the fatigue samples. Due to the difficulty in determining the fracture origin type for the creep samples, more attention was paid to the fracture surf ace features during propagation. There was a distinct change in the types of features present for some of the standard heat treatment specimens from near the origin to points furt her from the origin (Figure 3-49). Near the origin, the scale of the fracture features is on the order of 5 m. However, further down the fracture path, these features even tually reach a scale of about 20-30 m in diameter signaling a change in crack propagation path. Th e change in feature size is not nearly as dramatic for the alternate h eat treatment samples tested at 704C (1300F) and 689 MPa (100 Ksi) (Figure 3-50). There still does appe ar to be a slight cha nge at these conditions, but the transition is not as obvious as at the higher stress conditions. Finally, as the test A) B) Figure 3-49: Fast fracture propagation path for standard heat treatment sample tested at 677C (1250F) and 1034 MPa (150 Ksi). A) Near the initiation. B) In the fast propagation region.

PAGE 126

109 A) B) Figure 3-50: Fast fracture propagation path for alternate heat treatment at 704C (1300F) and 689MPa (100Ksi). A) Near the initiation. B) In the fast propagation region. temperature rose to 760C (1400F) and the lo ad was reduced to 483 MPa (70 Ksi), there was no change at all in feature size between these two regions. Figure 3-51 shows one of the samples tested at these conditions. Once examination of the fracture surfaces wa s completed, the fractured ends of the specimens were sectioned approximately 1 cm (0.39 in) from the fracture. The sectioned ends were mounted longitudina lly, polished, and etched to examine the path of secondary A) B) Figure 3-51: Standard heat treatment samp le tested at 760C (1400F) and 483 MPa (70 Ksi) showing fast fracture propagation path.. A) Near the initiation B) During the fast propagation region. cracks through the microstructure. This pr oved to be a useful way to view both the initiation and propagation behavior of cracks in the creep tests. The cross-sectioned

PAGE 127

110 samples not only served to catalogue differences between the two heat treatments but also between test conditions within the same heat treatment. One sample was examined for each heat treatment at each test condition. For both heat treatments, the lowest temp erature, 677C (1250F), and the greatest stress state, 1034 MPa (150 Ksi), resulted in few secondary cracks propagating through the microstructure near the fracture surface. The standard heat treatment sample exhibited some voids that formed at grain boundary triple points (Figure 3-52). Samples from both heat treatments exhibited cracks th at propagated transgranularly (Figure 3-53). Although the large crack in the standard heat treatment micrograph appears to follow the grain boundary on the left side of the picture, it clearly pr opagates through the grain on the right side of the micrograph. The alternat e heat treatment sample contained at least one crack that propagated from the fractur e surface down through the sample along the grain boundaries (Figure 3-54). This mi crograph also shows the propensity for microvoids to form within the grains at this test condition. Figure 3-52: Standard heat treatment samp le tested at 677C and 1034 MPa with crack initiation at a grain boundary triple point In contrast, all the samples tested at 704C (1300F) and higher exhibited many secondary cracks in the microstructure (Fig ure 3-55). These cracks propagate mostly Crack

PAGE 128

111 A) B) Figure 3-53: Transgranular cr acking during fatigue testing in samples tested at 677C and 1034 MPa. A) Standard heat treatment. B) Alternate heat treatment. Figure 3-54: Large secondary crack in alternate heat trea tment sample tested at 677C and 1034 MPa A) B) Figure 3-55: Secondary cracks in high temp erature creep tests. A) Standard heat treatment sample tested at 760C and 483 MPa. B) Alternate heat treatment sample tested at 732C and 483 MPa. Crack Crack Crack

PAGE 129

112 along the grain boundaries. They ap pear to nucleate at primary ’ precipitates, with adjacent cracks linking up during testing. Figu re 3-56a shows cracks initiating at primary ’ precipitates, while Figure 3-56b shows sma ll voids forming along the grain boundaries. Crack initiation within the grains is not observed at th ese higher temperature and lower stress test conditions fo r either heat treatment. A) B) Figure 3-56: Grain boundary crack initiation in alte rnate heat treatment creep samples. A) Tested at 704C and 689 MPa. B) Tested at 760C and 483 MPa. The only test condition that exhibited differe nt microstructural behavior for the two heat treatments was 677C (1250F) and 793 MPa (115 Ksi). The standard heat treatment sample exhibited many long cracks similar to the higher temperature tests, while the alternate heat treatment sample had a limited number of secondary cracks similar the highest stress condi tion. These differences can be seen in Figure 3-57. The cracks in the standard heat treatment sa mple propagate entirely along the grain boundaries, with very few void in itiation sites found within the grains. In contrast, cracks in the alternate heat treatment sample show ed equal preference for initiating at grain boundary triple points as well as with in the grains at dendritic-shaped ’ precipitates. However, the grain boundary voids were much larger than those within the grains. Crack Crack

PAGE 130

113 A) B) C) D) Figure 3-57: Secondary cracks in creep sa mples tested at 677C (1250F) and 793 MPa (115 Ksi). A) Standard heat treatment at 570X. B) Standard heat treatment at 2850X. C) Alternate heat treatment at 570X. D) Alternate heat treatment at 2850X.

PAGE 131

114 CHAPTER 4 DISCUSSION Microstructure Heat Treatment This study investigated the impact of an alternate, supersolvus solution heat treatment on the microstructure and mechan ical properties of powder processed Alloy 720LI. This alternate heat treatment was compared to a standard, subsolvus heat treatment. The alternate h eat treatment was designed to produce a microstructure with large, dendritic-shaped ’ precipitates in larger grains than in the standard heat treatment. In a study by Pratt & Whitney, a si milar microstructure dramatically improved the fracture toughness and fatigue properties of a single crystal alloy. The investigation examined the role of various heat trea tment parameters on the precipitation of ’ in this alloy. The first parameter examined was the soluti on heat treatment temperature. It was necessary for the solution temperature to be above the ’ solvus to completely dissolve all of the ’ in the alloy into the matrix. However, if the so lution temperature was too high, the resulting microstructure after slow cooli ng was more irregular in shape. Additionally, the greater heat treatment temp erature would require a longer to tal time for the entire heat treatment. This would result in an unnecessary increase in cost to heat treat actual parts. A solution temperature of 1175C (2147F) was c hosen for the alternate heat treatment in this study.

PAGE 132

115 The cooling rate through the ’ solvus temperature had the single biggest impact on the resulting size and shape of the secondary ’ precipitates. Too fa st of a cooling rate resulted in smaller, more cuboidal ’ precipitates. (Figure 3-1B ). Conversely, when the cooling rate was too slow, the ’ precipitates formed large, but irregular shaped dendrites. (Figure 3-2a) An intermediate cooling ra te of approximately 3.0C/min (5.4F/min) produced a regular array of dendritic-shaped ’ precipitates as desired. The final temperature before fan air cooli ng also affected the size of the secondary ’ precipitates. Both final temperat ures examined, 1020C (1868F) and 1075C (1967F), produced dendritic-shaped ’ when given the proper cooling rate. However, as can be seen in Figure 3-3, the lower temper ature produced larger, more fully developed dendrites than the higher temperature. Therefore, 1020C (1868F) was chosen as the final temperature for the heat treatment. The microstructural variations with heat tr eatment parameters agree with those seen in literature [70]. At fast cooling rates, more undercooling is achieved and the precipitate nucleation density is high. Th e resulting precipitates are cons trained in their growth and cannot develop into dendrites, but remain more cuboidal. However, at lower cooling rates, the number of nucleation sites decreases. The precipita tes begin as spheroids. The spheres can grow into cubes and eventual ly dendrites through diffusion controlled growth. In a near zero misfit alloy like Alloy 720LI, the precipitates eventually lose coherency with the matrix during the dendritic grow th. The growth direction during this process is along the <111> directions of the original ’ cubes. While this chosen heat treatment work ed well for most of the master powder blend’s (MPBs) used in this study, three MPBs required a slightly lo wer soak temperature

PAGE 133

116 in order to develop the alternate microstructure There do not appear to be any consistent differences in chemistry from one MPB to the next that would explain this difference. Differential thermal analysis (DTA) curves were also examined to attempt to determine if there was a change in the ’ solvus temperature that could account for this difference. From Tables 3-1 and 3-2, ther e is a slight difference in ’ solvus and the maximum thermal effect between MPB #96SW863 to #96S W865. The former MPB has a slightly greater temperature for the maximum thermal effect, but a slightly lesser ’ solvus temperature than the latter MPB. The maximu m thermal effect corresponds to the point where the ’ phase has the greatest rate of dissoluti on into the matrix. It appears that the soak temperature can not be too far above this temperature in order to achieve the desired microstructure through this heat treatment. Th is agreed with the heat treatment trials, where greater soak temperatures produced irregular shaped ’ precipitates. A decrease in the soak temperature by 10C produced the alte rnate microstructure for the three MPBs in question. Characterization Once the heat treatments were complete, the microstructure of both heat treatments was fully characterized. Th is included accurately determining the ’ solvus temperature and determining the volume fraction and size range of all modes of ’ particles found in the microstruc ture of both heat treatments This data helps understand the microstructural evolution during heat tr eatment and the strengthening and fracture mechanisms active durin g mechanical testing.

PAGE 134

117 Temperature of ’ solvus Due to the role of the ’ solvus temperature in both the standard and the alternate heat treatments, it was important to accurately determine the ’ solvus temperature for Alloy 720LI to fully understand the microstruc tural evolution during heat treatment. DTA analysis was performed at two inde pendent laboratories to determine the ’ solvus, the solidus, and the liquidus of Alloy 720LI. Since the DTA analysis is performed onheating, it can lead to results that diffe r from the equilibrium values of these temperatures. This is due to superheati ng in the alloy before dissolution or melting reactions takes place. Therefore, microstr uctural analysis samples were examined to compliment the DTA data and to determine the solvus temperatur es after quenching from a soak temperature both on-heating and on-cooling. The DTA analysis measured the ’ solvus temperature to be approximately 11851195C (2165-2183F). This is well above the va lue typically reported in literature of 1150-1155C (2102-2111F). The temperature repor ted as the maximum thermal effect is much closer to this reported ’ solvus temperature. More over, the metallographic study of quenched samples determined that there were at least two separate ’ solvus temperatures. These are approximately 1130C (2066F) and 1120C (2048F), and are lesser than the reported solvus temperature. These two temperatures correspond to the solvus temperatures for the primary and the secondary ’ precipitates respectively. It would also follow that there is another solvus temperature for the nucleation of the fine tertiary ’ precipitates at an even lower temperature. DTA data can be misleading since it is not an equilibrium method of determining various reaction temperatures and can only give a range of temperatures over which the

PAGE 135

118 ’ phase goes into solution. The ’ solvus temperature is determined as the temperature at which the last ’ phase goes into solution during heati ng [69]. As a result, DTA analysis will tend to overestimate the value of the solvus temperatures versus the equilibrium solvus temperature. Additionally, it is diffi cult to resolve two peaks on a DTA curve that occur within 10C (18F) of each other. The DTA data suggests that there is only one ’ solvus; however, the metallographic study disput ed this by revealing two distinct solvus temperatures for the primary and secondary ’ precipitates. While DTA is a useful means to approximate reaction temperatures, it is beneficial to augment its data with a metallographic study of quenchedin microstructures for accura te determination of solvus temperatures. Microstructure The microstructure of both the alternate and the standard heat treatments were characterized to help understand the role th e microstructure plays in the mechanical properties of these two heat treatments. The calculations and statistics for the microstructural characterizati on are found in Appendix B. The standard heat treatment consisted of small grains (ASTM 11.8) that are approximately 5.3 0.8 m (2.1 0.3 X 10-4in) in diameter and decora ted with large, primary ’ precipitates. These particles are blocky in shape with an average size of 1.7 0.4 m (6.7 1.6 X 10-5in) on edge. Within the grains, the secondary and tertiary ’ precipitated as regular arrays of small cubes approximately 0.6 0.2 m (2.5 0.7 X 10-5in) and 230 50 nm (9.1 1.9 X 106in) on edge respectively. Additionally, even smaller ’ precipitates were seen when using a FEG SEM. These “quaternary” ’ precipitates can be seen in Figure 4-1. They are approximately 40nm (1.6X10-6in) on edge. The total ’ volume fraction was found to

PAGE 136

119 be 51 3% for this heat treatment. The primary ’ had a volume fraction of 17 3%, while the secondary ’ had a volume fraction of 9 2% There was almost as much tertiary ’ as there was combined primary and secondary ’ in this heat treatment. The tertiary ’ had a volume fraction of 25 4%. It wa s not possible to resolve the quaternary ’ well enough to measure an accurate volume fr action. However, it is assumed to be less than 1.0% by volume. Figure 4-1: The tertia ry and “quaternary” ’ precipitates in the standard heat treatment In contrast, the alternate heat treatment resulted in a larger grain size (ASTM 8.2) that is approximately 18.4 1.6 m (7.2 0.6 X 10-4 in) in diameter. The primary ’ precipitates pin grain boundary motion during the standard subsolvus heat treatment. However, during the soak for the alternate heat treatment, these primary ’ particles are completely in solution, giving the grain boundaries the freedom to grow in a manner that Quatenary ’ p reci p itates Tertiary ’ p reci p itates

PAGE 137

120 minimizes the total surface area of the grains The grain boundaries developed a serrated structure in the alternate heat tr eatment due to the dendritic-shaped ’ precipitates impinging on the grain boundaries during growt h. This can be seen in Figure 4-2. During cooling from the maximum heat treatment temperature, the primary ’ particles Figure 4-2: Serrated grain bounda ry structure present in th e alternate heat treatment microstructure precipitate along the grain boundaries. As the temperat ure is decreased further, the secondary ’ precipitates begin to nuc leate intragranularly. Finally, at even lower temperatures, the fine tertiary ’ particles precipitate out of the matrix. In this heat treatment, the secondary ’ forms a regular array of dendritic-shaped ’ within the grains. In this heat treatment, the primary ’ occupy 6 2 % by volume and their average size is 1.4 0.6 m (5.5 2.4 X 10-5 in). The secondary ’ occupy a much larger volume in this heat treatment: 44 4 %. These precipita tes have a mean lineal intercept of 1.0 0.2 m (3.9 0.6 X 10-5 in); however they can be as large as 5 m (1.97X10-4in) from tip to tip. However, this is a misleading size due to grain orientation and the <111> growth direction of these precipitates. It is difficult to obtain a true radius for these particles. Due to the large size of these secondary ’ precipitates, not much ’ remains in solution after the initial solution heat treatment. As a result, the tertiary ’ can not grow as large

PAGE 138

121 1.7 m 1.4 m 0.63 m 1.0 m 230 nm 40 nm 40 nm 0 5 10 15 20 25 30 35 40 45 Volume Fraction PrimarySecondaryTertiaryQuatenary morphology Standard H. T. Alternate H. T.as in the standard heat treatment. Their volu me fraction is assumed to be less than 1.0%. They are approximately 40 nm (1.6X10-6 in) on edge in this heat treatment. No quaternary ’ was evident in this heat treatment. The total ’ volume fraction for this heat treatment was measured to be 50 3 %. Both heat treatments have the same volume fraction of ’. Figure 4-3 is a histogram of the ’ volume fractions for the different modes of precipitate in each heat treatment. Figure 4-3: Histogram of ’ volume fraction for the two heat treatments Samples of both heat treatments without th e aging heat treatment, along with an asHIPed sample, were polished and etched to study the precipitation and growth of the tertiary ’. Micrographs of these three samples are displayed in Figure 4-4. The large precipitates in these micrographs are primary ’ in the as-HIPed micrograph and

PAGE 139

122 A) B) C) Figure 4-4: Secondary and tertiary ’ precipitates. A) As-HIPed microstructure. B) Standard heat treatment w ithout aging heat treatmen ts. C) Alternate heat treatment without agi ng heat treatments. secondary ’ particles in the standard and altern ate heat treatment micrographs. Fine tertiary ’ precipitates can be seen in the as -HIPed and standard heat treatment micrographs; however, this fine, tertiary ’ is not visible in the alternate heat treatment. The tertiary ’ in the as-HIP and standard he at treatment microstructures are approximately 50nm on edge. However, there is a much higher vol ume fraction visible after the standard heat treatment. It is possible that the tertiary ’ has already precipitated from solution after the alternat e heat treatment but are not visible because of limitations

PAGE 140

123 in polishing, etching, and microsco pe resolution. The other possi bility is that the tertiary ’ precipitates may actually nucleate duri ng subsequent aging heat treatments. From this characterization data, the micr ostructural evolution during consolidation and heat treatment can be accurately desc ribed. The HIP cycle was carried out at 1129C. According to literature, this would have been below the ’ solvus temperature for this alloy. From the data collected in this project though, it is thought that this is below the primary ’ solvus temperature, but, in fact, is above the secondary ’ solvus temperature. The primary ’, which is only present at th e grain boundaries will restrict grain boundary motion and prevent the grains from growing during consolidation; however, the seconda ry and tertiary ’ will go into solution. During cooling from the consolidation temperature, ’ precipitates intergranularly in at least two modes (secondary and tertiary). It also appears that at the ri ght conditions, a fourth mode (quaternary) of ’ will precipitate during c ooling. The result of this HIP cycle is a fine grain sized material with primary ’ along the grain boundaries and irregular shaped secondary and tertiary ’ within the grains. The standard heat treatment is applied at a subsolvus temperature. This heat treatment is a partial solution heat treatment designed to refine the size and shape of the secondary ’ precipitates. It is not clear from the results collected during this study whether this heat treatment is a solution heat treatment or if it is only a high temperature aging heat treatment. It would be interesti ng to examine the solvus temperature of the tertiary ’ with a metallographic study below 1120 C. A comparison of the secondary ’ size difference before and after the solution h eat treatment would help in determining if this heat treatment is a true solution heat treatment.

PAGE 141

124 During the standard heat treatment, the secondary ’ transforms from slightly irregular shaped particles to more or less cuboidal shaped particles through diffusion. During this process, the size and shape of th e secondary precipitates is refined until they are cubes that are 0.63 0.17 m (2.5 0.7 X 10-5in) on edge. In addition, the tertiary ’, which previously nucleated during the HIP cy cle, begins to grow in size and volume fraction through diffusion. It is on ly after the final aging heat treatments that the tertiary and quaternary ’ reach their final size and volume fr actions. These final aging heat treatments only affect the fine ’ and not the primary or secondary ’ precipitates. In contrast, the alternate heat treatment is carried out at a supers olvus temperature. The ’ phase that precipitates during the consolid ation is completely in solution at this temperature. Without any primary ’ precipitates present, the gr ains are free to grow past their original size within the powder particles. During the slow cooling through the solvus temperature, ’ precipitates nucleate both along the grain boundaries as primary ’ and within the grains as sec ondary and possibly as tertiary ’ precipitates. It is not possible to tell when the tertiary ’ precipitates during the alternate heat treatment with the current data. During the slow cooling, the secondary ’ precipitates grow along the <111> directions to form dendritic-shaped part icles. Finally, the aging heat treatments possibly nucleate and increase the size and volume fraction of the tertiary ’ precipitates. Mechanical Testing After heat treating, the test bars were mach ined into mechanical test specimens. Hardness, tensile, creep and fatigue testing were performed on samples from both heat treatments to understand the effects of the new, alternate microstructure on the

PAGE 142

125 mechanical properties of Alloy 720LI. In addition, fractography was performed on fractured test specimens to understand the mech anisms of failure in both heat treatments. Hardness Testing There was a significant difference in the Rockwell C hardness for the two heat treatments, both before and after the aging he at treatments (Table 3-4). The differences in hardness can be related dire ctly to the microstructures of the two heat treatments. The size and larger volume fr action of the tertiary ’ precipitates in the standard heat treatment appear to result in a greater st rength due to more effective blocking of dislocation motion. The secondary and tertiary ’ in the standard heat treatment are coherent and have a small mean free path. As a result, the dislocations shear these precipitates causing antiphase boundaries to de velop [12]. In contra st, the alternate heat treatment has a very small volume fraction of tertiary ’ precipitates. In addition, the larger ’ precipitates are incoherent and more wi dely spaced. It is not clear if the dislocations are shearing these precipitates or bowing around them. It follows that the different ’ morphologies of the two heat treatments leads to the difference in hardness. The standard heat treatment has a moderate amount of small, finely spaced, coherent secondary ’ precipitates compared to the large amount of widely spaced, incoherent secondary ’ in the alternate heat treatment. The more finely spaced, coherent precipitates are expected to be more eff ective at blocking dislocation motion due to ’ shearing and the creation of an tiphase boundaries [12]. In addition, the standard heat treatment has a large amount of fine, closely-spaced tertiary ’. The alternate heat treatment, in contrast, has very little of this tertiary ’ precipitate. This ultrafine phase of g’ is thought to be very effective at bloc king dislocation motion because of its small

PAGE 143

126 mean lineal intercept. The higher hardness and larger volume fraction of tertiary ’ in the standard heat treatment s uggests that this mode of ’ has a large impact on the hardness and, in turn, strength of the alloy. In generally, the hardne ss of metals is proportional to the tensile strength [11]. A second reason for th ese differences in hardness is the smaller grain size of the standard heat treatment. The grain boundaries block dislocation motion from one grain to the next. As a result, fine grained materials exhibit greater strength and greater hardness. Finally, both heat treatments experienced an increase in hardness after the aging heat treatment. During these ag ing heat treatments, only the tertiary ’ precipitates grow. The other modes of ’ precipitates remain ba sically unchanged in morphology and volume fraction. This increase in hardness after aging is due to the coarsening of the tertiary ’ precipitates. Fatigue Testing The goal of this study was to develop an alternate heat treatment that produced a microstructure with better fatigue properties. Unfortunately, the standard heat treatment had longer high cycle fatigue (HCF) lifetimes th an the alternate heat treatment at stress ranges of 1034 MPa (150 Ksi) and 993 MPa (1 44 Ksi) at both test temperatures. However, the alternate heat treatment outperf ormed the standard heat treatment at the greatest stress range of 1086 MPa (157.5 Ksi) at both 538C (1000F) and 649C (1200F). A detailed study of the fracture su rfaces was performed to help identify the fracture mechanisms that led to these results. The initial step of fatigue failure is pl astic deformation before the nucleation of microcracks around a processing defect or persis tent slip band [71]. A sufficient amount of deformation will cause microcracks to init iate in the matrix surrounding a defect or

PAGE 144

127 cracked slip band. These microcracks propagate until they link up to form one or more large crack. The next phase of failure is fa tigue crack propagation before the onset of overload and fast fracture when these cracks re ach a threshold size. Fatigue failure in metals is generally broken up into three st ages [72]. During Stage I, fatigue crack initiation occurs along with some crack growt h. This most often occurs due to cracking of slip bands because of the repetitive reve rsals of deformation along these slip bands during fatigue testing. The crack follows crystallographic planes and the fracture surfaces are left with faceted features. During Stage II, the fatigue crack propagates transgranularly, leaving behind fatigue striations on a relativ ely featureless surface. Finally, in Stage III crack gr owth, the specimen undergoes over load and the crack growth is unstable until final fracture. The initiation point for each specimen was located by SEM and identified using energy dispersive spectrometry (EDS). Every standard heat treatment specimen failed at a processing defect. The majority of these were at ceramic agglomerates introduced to the alloy possibly from the melting crucible the pouring tundish, or the nozzle during atomization. Screening through fine mesh sieves removes many of these ceramic inclusions; however, some of them are sma ller than the sieve sizes used. It is not economically feasible to sieve to a finer mesh size, so it is ne cessary to engineer the alloy to have some degree of tolerance to these incl usions. The smaller grains in the standard heat treatment have a smaller displacement along the slip bands during fatigue testing [17]. This slows down the formation of micr ocracks around the slip bands. As a result, a small grained material is not as likely to fail by slip band cracking. Instead, it is more probable that a small grained material will fail from cracks emanating from processing

PAGE 145

128 defects. While the majority of the fatigue specimens failed at ceramic inclusions, one of the samples, ULA10, fractured at a Ti-rich pa rticle left on the surface of the machined test specimen (Figure 3-29). This carbide was cracked either before or during testing and as a result initiated microc racks in the matrix surroundi ng it. These microcracks eventually linked up and led to failure. The standard heat treatment samples te sted at the low and middle stress ranges failed at defects that were near the surface but not actually on the su rface. These defects were surrounded by a region of faceted fracture f eatures (Figure 3-33A). This area is an example of Stage I fatigue crack growth [72]. This region eventually transitioned to a flat, featureless region after approximately 100 m (3.94X10-3 in). This flat region is indicative of Stage II transgranular fatigue crack growth [72]. In contrast, the two tests at the highest stress range actually failed at processing defects on the surface of the machined samples. These defects were onl y surrounded by a flat, re latively featureless region (Figure 3-33B). There were no faceted f eatures near the failure initiation point for the highest stress range condition. In these samples, Stage I crack growth was completely bypassed and the cracks propagated entirely in a Stage II crack growth mode until fast fracture. At larger values of strain during fatigue testing, cracks will initiate at most defects in the material. The stress intensit y at defects on the surf ace have a greater stress intensity than internal defects as can be seen in Equations 4-1 and 4-2 below. Cracks at the surface will propagate faster under these conditions because they have greater stress intensity than cracks at internal defects and because of the deleteri ous effects of oxidation [17]. In contrast, at small strains, cracks will only initiate at acute inclusions, and internal initiation is favored.

PAGE 146

129 The alternate heat treatmen t test samples had a little more variability in their initiation sites (Table 3-4). Three of the seve n samples exhibited initiations at persistent slip bands. Ceramic inclusions led to failu re in two of the samples. The fracture initiation of the final two samples is unclear In one specimen, ULB17, no clear initiation point was discovered after examining the fract ure surface (Figure 3-32). Another sample, ULB15, clearly failed from a crack that be gan at a processing defect on the surface. However, after examining it with both back scattered electron imag ing (BSE) and EDS, it is still not clear what type of defect is present. It seems to be a large grain containing no ’ precipitates. (Figure 3-31). It is not cl ear how such a defect might have formed and remained throughout the heat treatment. This defect has cracks that extend from its interior into the surrounding matrix. Excluding sample ULB17 (no initiation poi nt found), the only samples to fail internally for the alte rnate heat treatment were the two tested at 993MPa (144Ksi). The rest failed at surface defects. The two intern al failures were at large faceted persistent slip bands (Figure 3-33C). These were su rrounded immediately by a region of faceted features as with the standard heat treatm ent samples. This region again extended for approximately 100 m (3.94X10-3in) before entering a region of Stage II crack propagation. However, the samples that failed at surface defects were again only surrounded by a relatively featureless region w ith no faceted features present (Figure 333D). The faceted features are ex amples of typical Stage I fatigue fracture by slip band cracking [72]. The dislocations move along the active slip planes, (111) in this case, until they reach the grain boundaries. After enough dislocations pile up at the grain

PAGE 147

130 boundaries, microcracks begin forming around th e grain. These microcracks link up and lead to the initial crack front that eventually pr opagates to failure. When this initial crack is formed, the grain fractures along the slip pl ane. This leaves a faceted feature on both halves of the fracture surface. These features resemble brittle fracture visually, but are the result of dislocation motion. As the cr acks grew, the fracture surfaces exhibited a transition to typical Stage II frac ture that is defined by a relati vely flat region. The cracks propagate transgranularly through this region of slow crack growth. Fatigue striations are sometimes present on the specimens in this re gion of the surface. These striations are caused by the repetitive opening and closing of the crack tip during propagation. Eventually, the crack reaches a point where th e specimen is overloaded and fast fracture begins. This is marked by the transition from the transgranular regi on to an area with a more tortuous surface features as seen in Fi gure 3-35. At this point, the crack transitions to either an intergranular or interparticle prop agation mode here. It is not clear which of these two modes of failure is active for this all oy. It is possible that they are both active. The feature sizes appear to match the grain size for the heat treatments; however, the features are much more rounded than those usua lly seen from intergra nular fracture. This could be due to cracking at the particle boundaries (Figure 4-5). The depth and width of the initial cracks were measured for all of the fatigue samples. The area used for this initial crack measurement coincided with the onset of the Stage II transgranular fracture. It was th en possible to measure the maximum stress intensity as well as a change in stress intens ity for the stress ranges tested. The equations used for the fracture mechanics calculations were taken from Anderson [73]. The flaws

PAGE 148

131 A) B) C) D) Figure 4-5: Comparison of fast fracture features during HCF fatigue and grain size. A) Standard heat treatment grai n size. B) Standard heat treatment fatigue fracture surface features. C) Alternate heat treat ment grain size. D) Alternate heat treatment fatigue fracture surface features. were approximated based on the stress intensit y factors for elliptical and semi-elliptical cracks. The equation for an embedded elliptical crack is a K max (4-1) The equation for a surface, semi-elliptical crack is a K 12 1max (4-2) where a is the crack depth, Kmax is the stress intensity factor at the maximum stress, and is a geometrical factor based on the crack shape and size. can be estimated by the following equation:

PAGE 149

132 2 28 8 3 c a (4-3) Once the stress intensity factor for these initi al cracks was determined, the plastic zone size was estimated by the plane stress and plane strain approximations: 2 22 1ys pK r (plane stress) (4-3) 2 26 1ys pK r (plane strain) (4-4) This equation was applied to both Kmax and K All of this data is contained in Table 4-1. Table 4-1: Stress intensity fact ors and plastic zone sizes for the initial crack in HCF tests Sample Kmax (MPa*m0.5) rp(Kmax) ( m) plane stress rp(Kmax) ( m) plane strain K (MPa*m0.5) rp( K) ( m) plane stress rp( K) ( m) plane strain Standard Heat Treatment ULA5 12 18 6 10 15 5 ULA15 12 20 7 11 16 5 ULA14 11 16 5 10 13 4 ULA11 11 16 5 10 13 4 ULA16 11 17 6 10 14 5 ULA3 12 18 6 10 15 5 ULA13 11 16 5 10 13 4 ULA10 13 21 7 11 17 6 Alternate Heat Treatment ULB7 15 45 15 13 36 12 ULB10 14 44 15 13 35 12 ULB6 15 44 15 13 35 12 ULB15 14 4 14 13 34 11 ULB13 14 42 14 13 34 11 ULB9 14 40 13 13 32 11 The calculations and statistics for these stre ss intensity factors can be found in Appendix B. The yield strength of the alternate heat treatment at 538C was estimated to be approximately 870 MPa. This estimate was ba sed on the yield streng th of the alternate heat treatment at 649C in conjunction with th e lower yield strength in the standard heat

PAGE 150

133 treatment tensile sample at 538C. The altern ate heat treatment fatigue samples transition to transgranular Stage II crack growth at a larger stress intensity and plastic zone size. It has been suggested previously that this tran sition occurs when the plastic zone size is approximately equal to the grain size [74]. This is the case in this st udy if the plane strain approximation is used. As these test bars were not designed to be either plane strain or plane stress, the actual situation will lie in between these two extreme cases. From this data, it is definitely possible that the tran sition to Stage II crack growth occurs when the calculated plastic zone size is approximately equal to the grain size for the two different heat treatments. However, since the cr ack area for the two heat treatments is approximately equal, it can not be ruled out that this transition simply occurs at a threshold crack size when the microcracks li nk up into one continuous crack front. One way to determine the mechanism for this tran sition in crack propagation mode would be to test samples at smaller stress ranges. If there is a threshold crack size needed for microcrack link-up, then the initial crack si ze would not change with stress range. However, if this transition is controlled by the size of the plastic zone compared to the grain size, then the initial crack size should increase with decreasing stress range. In this case, the plastic zone size and stress intensity factor for the initial crack would remain constant with changes in the stress range. In the same manner, the crack dimensions for the final slow crack growth region were measured from SEM micrographs. The st ress intensity factor and plastic zone size were also measured for both the max stress for each cycle. It is assumed that Kmax approaches K as the crack approaches overload. Th e results are listed in Table 4-2. Once again, the yield strength for the alternat e heat treatment at 538C was taken to be

PAGE 151

134 Table 4-2: Stress intensity f actors and plastic zone sizes for the final crack in HCF tests Sample ys (MPa) max (MPa) Kmax (MPa*m0.5) rp(Kmax) ( m) Standard Heat Treatment ULA5 1078 1103 37 190 ULA15 1078 1103 39 211 ULA14 1078 1149 40 218 ULA11 1078 1149 41 227 ULA16 1078 1207 42 238 ULA3 1095 1103 45 274 ULA13 1095 1149 45 272 ULA10 1095 1207 45 263 Alternate Heat Treatment. ULB7 ~870 1103 40 334 ULB10 ~870 1149 39 322 ULB6 883.5 1103 42 360 ULB15 883.5 1149 41 342 ULB13 883.5 1149 41 345 ULB9 883.5 1207 43 384 870 MPa. These values are lower than the KC values reported in literature of up to 70 MPa(m)0.5 [75]. While the value of Kmax at the final crack size is not equivalent to the KIC fracture toughness, it can be used to compare the stress intensity (KC) to cause overload for the two heat treatments in this st udy. As can be seen by this table, the KC of the two heat treatments is approximately eq ual. Additionally, Alloy 720LI has a higher stress intensity factor at 649C than at 538 C. These samples appear to fracture when Kmax reaches a threshold value irrespective of the plastic zone size. One way to determine whether this is in fact the case w ould be to test da/dn samples from both heat treatments to accurately determine KC. Any interactions between s econdary cracks and the microstructure were examined by polishing and etching transverse slices of the gauge section. For the standard heat treatment samples tested at both temperatur es and a stress range of 993 MPa (144 Ksi), the secondary cracks developed at primary ’ precipitates and at grain boundary triple

PAGE 152

135 points. There were very few secondary cracks observed in samples te sted at this stress range. The samples tested at stress range s of 1034 MPa (150 Ksi) and 1086 MPa (157.5 Ksi) exhibited more secondary cracks that appeared to nucleate al ong sites that formed rounded shapes through the grains There were also some cr acks that nucleated along the grain boundaries as well. Examples of bot h are found in Figure 4-6. The rounded appearance of these secondary cracks through the grains su ggests that they are along prior particle boundaries. At the highest st ress condition, there is also evidence of a secondary crack initiatin g at a cracked TiC particle. (Figure 3-41A). A) B) Figure 4-6: Secondary cracking in standard heat treatment fatigue samples. A) Running through the grains in sample tested at 538C and 1034 MPa. B) Propagating intergranularly in sample tested at 538C and 1086 MPa. In the alternate heat treatment samples te sted at both temperatures, the secondary cracks appear to have a slightly different nature than in the standard heat treatment. At stress ranges of 993 MPa (144 Ksi) and 1034 MPa (150 Ksi), the cracks initiate at both primary and secondary ’ precipitates. These cracks appear to occur in straight lines at an angle to the tensile direction as seen in Figure 4-7. This suggests that these cracks might be forming along the slip bands at the / ’interface. The samples tested at a stress range of 1086 MPa (157.5 Ksi) had a few seconda ry cracks that began at TiC’s and Al2O3

PAGE 153

136 A) B) Figure 4-7: Secondary cracks propagating transgranularly in alternate heat treatment fatigue samples. A) Tested at 538C and 993 MPa. B) Tested at 538C and 1034MPa. ceramic inclusions as seen in Figure 3-41C. They also had voids initiating at the interface between the matrix a nd the primary and secondary ’ precipitates as seen in Figure 3-40c. Finally, there were some instances of large pores forming at grain boundary triple points. (Figure 4-8). These ar e thought to form as a result of thermally Figure 4-8: Large pore at a gr ain boundary triple point in al ternate heat treatment sample tested at 538C and 1086 MPa induced porosity (TIP) due to entrapped argon gas from atomization. For the most part, there were not many large cracks at this condi tion. However, the one s that were present appeared to propagate transgranularly as seen below in Figure 4-9. There were no

PAGE 154

137 secondary cracks that appeared to follow prior particle boundaries as in the standard heat treatment specimens. Figure 4-9: Long secondary cracks in alternat e heat treatment fatigue sample tested at 538C and 1086 MPa While the fatigue test data shows that th e standard heat trea tment outperforms the alternate heat treatment at the low and mi ddle stress ranges, but not at the high stress range; it is necessary to exam ine the tests on a case by case basis in order to understand the mechanisms behind these differences. At the highest stress ra nge (1086 MPa), all samples failed from cracks that began at pr ocessing defects on the surface. These cracks propagated entirely in Stage II crack growth. For this condition, the alternate heat treatment lasted more cycles, but had a smaller crack size before fast fracture. At this high of a stress range, it is reasonable to assume that the number of cycles to initiate a continuous crack front is small compared to the total lifetime. The number of cycles for Stage II crack propagation domina ted the lifetime of the fatigue specimens. With this in mind, it can be concluded that the alternate heat treatment had a slower crack growth rate. It took more cycles to propagate a shorter di stance. This differen ce in propagation rate would be applicable to all of the fatigue tests for the Stage II crack growth region.

PAGE 155

138 At the middle stress range (1034 MPa), the standard heat treatment samples failed at internal defects, while the alternate heat treatment samples failed from either surface defects or persistent slip bands at the surf ace. This difference in initiation location is probably partly caused by the difference in yi eld strengths between the two materials. The standard heat treatment had a greater yiel d strength than the alternate heat treatment (Table 3-7). For the standard heat treatment, the fatigue tests are run only slightly into the plastic regime. In contrast, the alternat e heat treatment samples received significant plastic deformation on each cycle. The stre ss intensity is high enough to favor crack initiation at the surface. Microcracks at def ects on the surface initiate rather quickly. They link up to form a continuous crack front in a relatively small number of cycles, and once again the number of cycles for Stage II propagation dominates the lifetime. However, it takes a significant number of cycles for internal crack initiation and Stage I propagation for the standard heat treatment. While the Stage II propagation rate may be quicker, the lifetime is dominated by crack init iation. The slower time to crack initiation explains why the standard heat treatment sa mples outperform the alternate heat treatment samples at the middle stress range. At the lowest stress range (993 MPa), all of the samples failed at internal initiation sites. The standard heat treatment samples failed from cracks originating at processing defects, while the alternate heat treatment samples failed at cracks propagating from the slip bands. At this stress range, the lifetimes of the test specimens are controlled by the number of cycles to crack initiation. While the standard heat treatment produces persistent slip bands, the displacement of th e slip bands at the gr ain boundaries is small due to the small grain size. As a result, these samples tend to fail at small processing

PAGE 156

139 defects that were not eliminated during the pow der screening process in stead of at the slip bands. These defects are slightly larger in size than the grains. In contrast, the large grains of the alternate heat treatment pr oduce very large displacements at the grain boundaries through slip mechanisms during the fatigue testin g. These large displacements lead to a faster crack initiati on than the defects found in the standard heat treatment because the initial cracks are larger. This difference in cycles to crack initiation lead to longer lifetimes for the standard h eat treatment at the lowest stress range. In general, higher ductility materials have longer fatigue lifetimes at higher strain (stress) values [17]. Conversely, higher st rength materials have l onger fatigue lifetimes at lower strain (stress) values. At low stre ss ranges, it takes a gr eater amount of cycles for the microcracks surrounding a processing de fect or slip band to finally link up and form the initial crack. As a result, the fatigue lifetime is dominated by the number of cycles to crack initiation. The large size of the grains in the alternate heat treatment results in a faster crack initiation compared to the small processing defects at which failure initiates in the standa rd heat treatment because the stress intensity around the large grains is greater than that around the small grains. The smaller grain sized material (standard heat treatment) has longer fatigue lifetimes at the small stress ranges. From this data, it is expected that at ev en larger stress ranges, the alternate heat treatment will outperform the standard heat tr eatment. However, at lower stress ranges the standard heat treatment will perform bette r. The microstructure does not seem to have as much of an impact on the fatigue properties as grain size does. Tensile Testing At all test temperatures, the standard h eat treatment had greater values for yield strength, ultimate tensile strength (UTS), a nd strength at failure than the alternate heat

PAGE 157

140 treatment. However, the standard heat trea tment had a corresponding deficit in ductility. While both heat treatments maintain thei r strength and elongation up to 649C (1200F), they suffer a deficit in these properties at 760C (1400F). This can be seen in Figure 4-10. Additionally, the macroscopic features of both heat treatments were similar for 700 800 900 1000 1100 1200 1300 0100200300400500600700800 Temperature (C)Yield Strength (MPa) Standard H.T. Alternate H.T. Figure 4-10: Yield strength vs. test temperature for both heat treatments each test temperature. The room temperatur e tests had a flat surface with small shear lips. The tests at 649C (1200F) had a ve ry distinct radial crack pattern with pronounced ridges leading away fr om the initiation source and small shear lips. Finally, the 760C (1400F) tests had surfaces with a fl at arrowhead-shaped section leading away from the origin and larger shear lips on either side of the arrowhead. Optical pictures of these surface features ca n be found in Figure 4-11.

PAGE 158

141 A) B) C) Figure 4-11: Optical pictures showing the fract ure surfaces of tensile samples. A) Tested at room temperature. B) Tested at 649C (1200F). C) Tested at 760C (1400F). The fracture surfaces of the tensile tests showed no difference in crack nucleation or propagation between the two heat treatments. Neither h eat treatment had a preference for crack origin type. A few crack s began at ceramic particles (Al2O3); however, the majority of the fractures began at undetermi ned origin types. Neither heat treatment displayed a tendency to exhibit failure initiation at ceramic inclusions. Additionally, the microstructural features along the crack path were similar for both heat treatments. A study of the fracture mechanics of the tens ile samples is contained in Appendix C. Samples were polished to view secondary cr acks interacting with the microstructure. However, the tensile specimens had almost no s econdary cracks for either heat treatment.

PAGE 159

142 It is thought that the differe nce observed in tensile prope rties between the two heat treatments is due to the ’ morphology and, to a lesser extent, the grain size differences between the two heat treatments. As discu ssed in the section on hardness testing, the ’ morphology and volume fraction of the standard heat treatment provide an effective hindrance to dislocation motion. Dislocations shear the secondary, tertiary, and possibly quaternary ’ precipitates creating antiphase boundari es within the precipitates. It is necessary for a second dislocati on to shear the precipitates in order to restore the ordered structure. The large volume fraction of tertiary ’ in the standard heat treatment plays a crucial role in the strength of this microstructu re. In addition, the standard heat treatment has a finer grain size. It has a much higher grain boundary density. These grain boundaries are also an effective barrier to di slocation motion through the test specimens. In contrast, the alternate heat treatment ha s a microstructure that does not provide as effective of a barrier to dislocation motion. The small amount of tertiary ’ does not provide much additional strengthening in th e samples given this heat treatment. The dendritic-shaped secondary ’ precipitates are larger and have a greater mean free path between them than in the standard heat treatmen t. It is possible that the dislocations are able to bypass these ’ precipitates instead of shearing them. This bypass mechanism produces dislocation loops that provide so me strengthening but not as much as the shearing mechanism produces. Finally, the la rger grain size does not block dislocation motion as much as the small grains present in samples given the sta ndard heat treatment. Since the microstructure of the standard heat treatment has more hindrances to dislocation motion than the alternate heat tr eatment, it takes a gr eater stress to produce dislocation motion and, therefor e, plastic strain during tensile loading. As a result, the

PAGE 160

143 yield stress, UTS, and stress at failure are all greater in the standard heat treatment samples. Likewise, the microstructure of th e alternate heat treatment specimens allows for much easier dislocation motion over longer distances. This heat treatment produces a more ductile microstructure. Creep Testing During creep testing, the alternate heat treatment outperformed the standard heat treatment at the tests run at 704C (1300F) and above. However, the opposite was the case for the tests at 677C (1250F). This is evident from the results in Table 3-8. However, the alternate heat treatment sample s experienced greater elongation at all test conditions. The fracture surfaces of select samples, al ong with polished, longitudinal slices of the gauge section of these samples, were examined to determine the mechanisms involved during creep testing. It was not possible to determin e the type of crack origin for any of these samples by examining the fracture surface. The origin was located using two complimentary methods: discoloration due to longer oxidation times at the origin and radial cracks leading from the origin. Howe ver, none of the origins found were of a distinct defect type. They appeared to be due to crystallographic failure rather than to processing defects. The path of secondary cracks within the microstructure was examined by polishing into a longitudinal slic e of the gauge section of the creep specimens. The tests at 704C (1300F) and higher had a high density of secondary cracks propagating roughly perpendicular to the tensile axis. When these specimens were treated with a grain boundary etchant (waterless Kallings reagent), it was obvious that these secondary cracks ran along the grain boundaries fo r both heat treatments. Mo reover, higher magnification

PAGE 161

144 SEM micrographs revealed that the grain bound aries served as nucle ation sites for voids or cavities to form during creep testing. Th is can be seen in Figure 4-12. These cracks Figure 4-12: Alternate heat treatment cr eep sample tested at 732C and 483 MPa showing secondary cracks initiating at and propaga ting along grain boundaries appear to initiate at cavitations formed between the grain boundary and the primary ’ precipitates as shown in Figur e 3-56. These cracks are an indication that the failure mechanism at these test conditions is govern ed by grain boundary sliding [76]. This mechanism explains why the alternate heat tr eatment test specimens have longer lifetimes at these test conditions. The alternate heat treatment micros tructure contains a smaller grain boundary density due to its larger grain size. Therefor e, there is a much smaller area for cracks to nucleate. Additionally, th e alternate heat treatment has successfully produced a serrated grain boundary structure th at provides additional resistance to grain boundary sliding. As a result, this heat tr eatment is more resistant to grain boundary sliding than the standard heat treatment. In contrast, the specimens tested at 677C (1250F) and 1034MPa (150Ksi) contained very few secondary cracks in the tr ansverse slices. These cracks nucleated at voids created either at grain boundary tr iple points or betw een the secondary ’ precipitates and the matrix. There does not appear to be a preference for these cracks to

PAGE 162

145 nucleate at either the grain boundaries or within the grains for this test condition. Voids nucleating at both sites can be s een in Figure 4-13. This type of cracking indicates that Figure 4-13: Alternate heat treatment cr eep sample tested at 677C and 1034 MPa showing void nucleation at a grain bounda ry triple point a nd at the interface between secondary ’ precipitates and the matrix grain boundary sliding is not the primary mechan ism at work at this condition. Instead, dislocation motion is the main mechanism that leads to creep failure. As was stated before, the standard heat treatment micros tructure is more effective at blocking dislocation motion. The standa rd heat treatment has a larger volume fraction of tertiary ’, as well as small grains, that are very e ffective at blocking disl ocation motion. In contrast, the alternate heat treat ment has very little tertiary ’, large grains, and large, incoherent secondary g’ that are not as effective at blocki ng dislocation motion. Hence, the microstructure in samples gi ven the standard heat treatment is more resistant to creep at lower temperatures and higher stresses. The only test condition where the primary mechanism for failure seems to differ is 677C (1250F) and 793 MPa (115 Ksi). Th e secondary cracks for the standard heat treatment sample nucleate both intergranularly and intragranluarly. However, the cracks along the grain boundaries appear to be more fr equent and longer. These are similar to the secondary cracks present in the high temper ature creep tests. Grain boundary sliding

PAGE 163

146 is the main mechanism leading to failure in the standard heat treatment sample. However, the alternate heat treatment sample at this test condition has relatively few secondary cracks that appear to nucleate without bias to the grain boundaries or within the grains in a similar fashion to the tests at 1034 MPa (150 Ksi). It is thought that the mechanism that led to failure for this sample is dislocation motion along the slip planes. The fracture surface features along the crack path also provide clues as to the dominant mechanism for creep failure. The samp les tested at the highe st stress condition, 1034MPa (150Ksi), have small dimples near the cr ack initiation site; wh ile those tested at higher temperatures have larger rounded feat ures (Figure 3-49, 350, and 3-51). These larger features correspond roughl y to the grain size of the material. This is an indication that the fracture mechanism near the initiation site is along the grain boundaries. This is evidence of grain boundary sliding as the prim ary mechanism for the samples tested at higher temperatures and dislocation motion as the mechanism for the samples at the highest stress condition. As the cracks propagate radia lly from the initiation site, all samples regardless of test c ondition eventually have larger features representative of grain size. Therefore, once the cracks reach ed fast fracture, they propagated along the grain boundaries. Conclusions The following conclusions can be made base d on this study on the effect of a slow cooling rate from above the ’ solvus temperature on the mi crostructure and mechanical properties of Alloy 720LI: 1. The morphology of the secondary ’ precipitates can be read ily tailored by varying the cooling rate afte r a supersolvus soak.

PAGE 164

147 2. At a cooling rate of approximately 3C/min (5F/min), the secondary ’ precipitates formed with a dendritic-shaped morphology. These precipitates have a mean lineal intercept of 1.0 0.2 m (3.9 0.6 X 10-5 in), but can grow up to 5 m (1.97X104in) from tip to tip. 3. Tertiary ’ precipitates were cuboidal and approximately 40nm (1.6X10-6in) on edge after aging heat treatments for the alternate heat treatment material. 4. The standard (subsolvus) heat treatmen t resulted in cuboidal, secondary ’ that was approximately 0.6 0.2 m (2.5 0.7 X 10-5in ). Aging heat treatments produced cuboidal tertiary ’ that was about 230 48 nm (9.1 1.9 X 10-6in) on edge and “quaternary” ’ that was approximately 40nm (1.6X10-6in) on edge. 5. The alternate heat treatment produced a larger grain size (ASTM 8.2) than the standard heat treatment (ASTM 11.8) due to solutioning of the primary ’ precipitates at the soak temperature. Add itionally, the grains in the alternate heat treatment material formed coarse serrati ons during the heat treatment from the growth of the large secondary ’ precipitates. 6. The ’ volume fraction for both heat treatments was approximately 50%. For the alternate heat treatment, the primary and secondary ’ phase accounted for almost all of the ’ phase volume. However, in the st andard heat treatment, the primary and secondary ’ phase accounted for about half ’ phase present and the tertiary ’ phase accounted for the other half. 7. The standard heat treatment had a greater hardness after both the solution and the aging heat treatments than the alternate heat treatment. 8. The standard heat treatment had longer fa tigue lifetimes at the low and mid-range stress conditions, but shorter lifetim es at the greatest stress range. 9. The fatigue life was controlled by crac k initiation at the low and middle stress ranges, but by crack propagation at the greatest stress range. 10. Grain size had the biggest effect on fatigue lifetimes. 11. The standard heat treatment had larger yield strengths, te nsile strengths, and strengths at failure for all test temperatures. The larger amount of tertiary ’ plays an important role in this greater streng th. Additionally, the smaller grains are a more effective barrier to dislocation motion. 12. The alternate heat treatment ha d more ductility at all test temperatures. This is also a result of the ’ morphology and the larger grains not being as effective a barrier to dislocation motion. 13. The standard heat treatment had longer creep lifetimes for both stress conditions at 677C (1250F). However, the alternate heat treatment had longer lifetimes for all

PAGE 165

148 tests at 704C (1300F) and a bove. In general, the alte rnate heat treatment creep samples had greater values of total elongation at failure. 14. At the lower temperatures, dislocatio n motion was the primary mechanism for creep failure. The standard h eat treatment had longer lifetimes. 15. At the higher temperatures, grain boundary sliding was the failure mechanism. The larger grain size and corr espondingly lower grain boundary density of the alternate heat treatment was not affected by grain boundary sliding as much as the smaller grains in the standa rd heat treatment. Future Work This research investigated the effect of a supersolvus heat tr eatment with a slow cooling rate on the microstructure and mechanical properties of a P/M Ni-base superalloy, Alloy 720LI. The supersolvus he at treatment did not perform as well in tensile and fatigue testing as hoped. Howeve r, this heat treatment was not completely refined. Further studies into the pr ecipitation behavior of the secondary ’ particles in P/M Alloy 720LI would help understand the variety of ’ morphologies that can be achieved through variation in the heat treatment parameters. It would also be interesting to examine the tensile properties for a range of microstructures. Th is study suggests that most of the strengthening for the standard he at treatment is from the development of the tertiary ’ and not the secondary ’. With a fast enough cooling rate from a supersolvus heat treatment temperature and subsequent aging at the proper temperatures, it may be possible to develop a fairly fine grained material with mostly tertiary ’ precipitates. Another interesting micr ostructure that could be develope d is one similar to the alternate heat treatment in this study, but with a smaller grain size. This might be able to be accomplished if the soak time is reduced from 2 h. The alternate heat treatmen t material did have a slow er crack propagation rate during the fatigue testing as a result of its larger grain size. Since the alternate heat

PAGE 166

149 treatment microstructure is much more ductile it would be benefici al to compare the LCF lifetimes of the two heat treatments. It would also be important to perform HCF testing at lower and higher stress ranges as those in this study. This study was not able to establish the repeatability of the fatigue results for the alternate heat treatment. It is possible that this heat treatment’s tendency to fracture at slip bands instead of processing defects means that its fatigue lifetimes are more repeatable than the standard heat treatment. Even with lower lifetimes, th is repeatability would be a step towards developing the desired defect-t olerant microstructure. Fa tigue crack growth testing would also help characterize the crack propaga tion rate and the toughness of the two heat treatments.

PAGE 167

150 APPENDIX A DIFFERENTIAL THERMAL ANALYSIS GRAPHS Figure A-1: The DTA Gr aph of sample UMBCa Figure A-2: The DTA graph of sample UMBCb

PAGE 168

151 Figure A-3: The DTA graph of sample UMBEa Figure A-4: The DTA graph of sample UMBEb

PAGE 169

152 Figure A-5: The DTA gr aph of sample UMBCc Figure A-6: The DTA graph of sample UMBEc

PAGE 170

153 APPENDIX B STATISTICAL DATA Hardness Testing Rockwell C hardness testing was performed on an as-HIP sample as well as samples given the standard and alternate heat treatments, both before and after the aging heat treatment. In total, there were five different sets of hardne ss data. The individual measurements and the statistics for these measurements are given below in Table B-1. Table B-1: The hardness testing statistical data AsHIP Std. H.T. w/out Age Std. H.T. w/ Age Alt H.T. w/out Age Alt H.T. w/ Age 1 37.4 42.3 45.4 40.4 42.9 2 37.2 42 46.1 39.7 43.2 3 37.1 42.8 45.4 39.7 42.7 4 38.4 42.4 45.7 39.9 43.3 5 38 42.5 46.5 39.7 42.5 6 38.5 42.4 46 39.8 42.7 7 42.5 46.1 40.3 43.3 8 43.3 45.4 40 42.8 9 42.7 45.6 40 43 10 42.3 45.4 39.8 43 Mean 37.8 42.5 45.8 39.9 42.9 Std. Dev. 0.615 0.352 0.392 0.250 0.272 t 2.571 2.262 2.262 2.262 2.262 95% CI 0.6 0.3 0.3 0.2 0.2 The standard deviation is given by the following equation: 2 / 1 1 2) ( 1 1 n ix i x n s (B-1) The 95% confidence interval is calculated by this equation: n s t CI % 95 (B-2)

PAGE 171

154 where the t parameter is estimated for the 95% confidence level acco rding to the number of degrees of freedom for each data set. In addition, t-tests were performed to determine whether these data sets are unique from one another. To accomplish this, th e standard deviations of the two tests in question were weighted to determ ine the pooled sample estimator, sp 2. The equation for this is as follows: 2 1 12 1 2 2 2 2 1 1 2 n n s n s n sp (B-3) Then, the difference in the means is compared with a 95% confidence interval by Equation B-4: 2 1 2 2 / 2 11 1 n n s t x xp (B-4) If this interval contains zero, then the two da ta sets are not distinctly statistically. The ttest values for the hardness tests are contained in Table B-2. All of these data sets were statistically distinct from each other. Table B-2: The t-test da ta for the hardness testing sp 2 t Mean Difference 95% CI As-HIP vs. Std. 0.215 2.145 4.75 0.51 As-HIP vs. Std. + Age 0.234 2.145 7.99 0.54 As-Hip vs. Alt. 0.175 2.145 2.16 0.46 As-HIP vs. Alt. +Age 0.183 2.145 5.17 0.47 Std. vs. Std. + Age 0.139 2.101 3.24 0.35 Std. vs. Alt. 0.093 2.101 2.59 0.29 Std. vs. Alt. + Age 0.099 2.101 0.42 0.30 Std. + Age vs. Alt. 0.108 2.101 5.83 0.31 Std. + Age vs. Alt. + Age 0.114 2.101 2.82 0.32 Alt. vs. Alt. + Age 0.068 2.101 3.01 0.25

PAGE 172

155 The Volume Fraction of the Gamma Prime Phase The volume fraction of the ’ phase was calculated for each mode of precipitate present in each of the two heat treatmen ts using ASTM Standard E562-99. Volume fractions were determined for the primary, secondary, and tertiary ’ in the standard heat treatment and for the primary and secondary ’ in the alternate heat treatment. The volume fraction of the tertiary ’ in the alternate heat treatment and the quaternary ’ in the standard heat treatment were assumed to be less than 1.0%. A circular grid of 24 points was superimposed onto each micrograph. The number of test points that fell within the respective ’ mode was counted and then divi ded by the total number of grid points. These values were then averaged for all ten fields of view. The standard deviation, the 95% confidence interval, a nd the percent relative accuracy were all calculated. The number of counts for each fiel d of view (out of a total of 24 possible points), the volume fraction for each field of view (PP(i)), the average volume fraction for all of the fields of view ( PP ), the standard deviation (s ), t parameter, and the 95% confidence interval are all listed in this table. The corresponding equations for these calculations are: n i P Pi P n P11 (B-5) 2 / 1 1 21 1 n i P PP i P n s (B-6) n s t CI % 95 (B-7)

PAGE 173

156 where n is the total number of fi elds of view and t is the sa me parameter described in the hardness testing section. The calculations for each of the different modes of ’ are contained in Tables B-3 through B-7. Table B-3: The primary ’ volume fraction data for the standard heat treatment Point Counts PP(i) PP Std. Dev. t 95%CI 4 16.7 5 20.8 3 12.5 4 16.7 5 20.8 5 20.8 4 16.7 2.5 10.4 4 16.7 5 20.8 17.3 3.68 2.262 2.63 Table B-4: The secondary ’ volume fraction data for the standard heat treatment Point Counts PP(i) PP Std. Dev. t 95% CI 2 8.33 3 12.5 3 12.5 2 8.33 1.5 6.25 2 8.33 1.5 6.25 2.5 10.4 1.5 6.25 2 8.33 8.75 2.37 2.262 1.69

PAGE 174

157 Table B-5: The tertiary ’ volume fraction data for the standard heat treatment Point Counts PP(i) PP Std. Dev. t 95% CI 6 25.0 7 29.2 6.5 27.1 4 16.7 8 33.3 5.5 22.9 5 20.8 7 29.2 6 25.0 4 16.7 24.6 5.45 2.262 3.90 Table B-6: The primary ’ volume fraction data for the alternate heat treatment Point Counts PP(i) PP Std. Dev. t 95% CI 0.0 0.00 0.0 0.00 1.0 4.17 1.5 6.25 1.0 4.17 2.0 8.33 0.0 0.00 1.0 4.17 0.0 0.00 4.0 16.7 2.5 10.4 1.5 6.25 2.0 8.33 2.0 8.33 2.5 10.4 0.0 0.00 1.0 4.17 2.0 8.33 0.5 2.08 3.0 12.5 5.73 4.73 2.093 2.21

PAGE 175

158 Table B-7: The secondary ’ volume fraction data for th e alternate heat treatment Point Counts PP(i) PP Std. Dev. t 95% CI 14 58.3 11.5 47.9 8 33.3 9 37.5 9.5 39.6 14.5 60.4 14.5 60.4 12 50.0 12 50.0 8.5 35.4 8 33.3 11.5 47.9 9.5 39.6 9.5 39.6 8.5 35.4 13.5 56.3 10 41.7 9 37.5 10 41.7 7.5 31.3 43.9 9.43 2.093 4.41 Gamma Prime Size Distributions In addition to their volume fracti ons, the average size of each mode of ’ precipitates in the two heat treatments wa s calculated. Once again, the tertiary ’ in the alternate heat treatment and the quaternary ’ in the standard heat treatment were not analyzed because of their small volum e fraction. A line of known length was superimposed on the micrographs and the numbe r of intercepts betw een the line and the ’ precipitates were counted. The mean (< x>), standard devia tion, and confidence interval for the number of intercepts were calculated. The averag e number of counts per m () was calculated by dividing by the total line length. The surface area (SV) of the precipitates was taken as twice this va lue. Finally, the mean lineal intercept was calculated according to the following equation:

PAGE 176

159 V VS V 4 (B-8) where VV is the volume fraction of that mode of ’. This mean lineal intercept is equivalent to the average precipitate size. Th e data for these calculations is in Tables B-8 through B-12. Table B-8: The primary ’ size data for the stan dard heat treatment Intercepts Std. Dev. Line Length ( m) (1/ m) SV (1/ m) < > ( m) 95% CI 3 9 4 8 5 8 5 5 11 6.4 2.7 31.9 0.20 0.40 1.71 0.39 Table B-9: The secondary ’ size data for the standard heat treatment Intercepts Std. Dev. Line Length ( m) (1/ m) SV (1/ m) < > ( m) 95% CI 12 12 15 5 5 9 12 7 3 8.9 4.1 31.9 0.28 0.56 0.63 0.17

PAGE 177

160 Table B-10: The tertiary ’ size data for the stan dard heat treatment Intercepts Std. Dev. Line Length ( m) (1/ m) SV (1/ m) < > ( m) 95% CI 6 2 3 6 3 7 15 6 13 3 13 3 14 5 3 6 0 10 4 7 9 9 12 11 11 8 4 3 3 4 6.8 4.0 3.19 2.12 4.25 .232 .048

PAGE 178

161 Table B-11: The primary ’ size data for the alte rnate heat treatment Intercepts Std. Dev. Line Length ( m) (1/ m) SV (1/ m) < > ( m) 95% CI 0 1 1 2 2 0 4 0 0 0 2 4 0 8 4 2 2 0 2 0 2 0 2 2 7 0 2 8 0 0 1.9 2.3 24.0 0.079 0.158 1.45 0.62 Table B-12: The secondary ’ size data for the alte rnate heat treatment Intercepts Std. Dev. Line Length ( m) (1/ m) SV (1/ m) < > ( m) 95% CI 21 25 20 22 20 18 24 20 19 19 20.8 2.25 24.0 0.87 1.73 1.01 0.15

PAGE 179

162 Grain Size The average ASTM grain size was calcul ated using the Abrams Three-Circle Procedure found in ASTM standard E112. Three concentric circles with a total circumference length of 500 mm were supe rimposed on a micrograph. The number of intersections between a grai n boundary and the circles (Pi) were counted. Five fields of view were analyzed for each heat treatment. For each field of view, the number of grain boundary intersections per unit line () were calculated by: M L x / PL (B-9) where L is the total test line length (500mm) a nd M is the magnification. With this value, the mean lineal intercept value for each field, < >, by: LP 1 (B-10) The average grain size was then calculated by averaging the numb er of grain boundary intersections per unit line over the five fields of view. The grain size was calculated using the following equation: 3 3 log 64 610 LP G (B-11) After determining the average grain size, th e standard deviation, the 95% confidence interval, and the percen t relative accuracy were all determined in the same manner as the volume fraction described previously. The data for the grain size calculations are contained in Table B-13 and Table B-14.

PAGE 180

163 Table B-13: The grain size data for the standard heat treatment Intercepts Std. Dev. Line Length ( m) (1/ m) < > ( m) 95% CI ASTM Grain Size 53 59 65 48 54 55.8 6.46 299 0.19 5.35 0.77 11.8 Table B-14: The grain size data for the alternate heat treatment Intercepts Std. Dev. Line Length ( m) (1/ m) < > ( m) 95% CI ASTM Grain Size 50 46 48 48 55 49.4 3.44 909 0.054 18.4 1.6 8.2 Stress Intensity Factors A statistical analysis was performed on the initial crack stress intensity factors to determine whether the data for the standard he at treatment was distinct from the data for the alternate heat treatment. As with the hardness testing data, a t-test was performed to determine this. The eight data points for Kmax of the standard h eat treatment were compared against the six data points for Kmax of the alternate heat treatment. Table B-15 contains the pertinent data for this study. When the confidence interval is added or subtracted to the difference in means, the re sulting range does not contain zero. The two data sets have distinctly different means. Table B-15: The t-test data for the initial crack stress intensity factors (MPa*m0.5) Std. Dev. sp 2 t Mean Difference 95% CI Standard H.T. 11.4 0.72 Alternate H.T. 14.4 0.05 0.32 2.179 3.0 0.67

PAGE 181

164 APPENDIX C TENSILE FRACTURE MECHANICS As with the fatigue fracture surfaces, th e crack dimensions for the flaws that eventually led to tensile failure was measur ed when possible. Due to difficulty in locating the crack initiati on point on samples UCA24, UCB4, and UCB2, it was not possible to measure the crack depth of these samples. The crack depth measured for the rest of the samples is contained in Table C-1. It should be noted that measurements of the crack dimensions for the tensile fracture surfaces was much more difficult than with the fatigue fracture surfaces. The crack area was not as distinct as with the fatigue samples. This is in part because of the difference in fracture and crack growth between tensile testing and fatigue testing. In fatigue testing, there is a period of slow deformation and crack growth followed by fast fracture at overload. Tensile testing, on the other hand, contains large amounts of plastic de formation during a fast fracture mode at overload. Table C-1: Area of crack in fractured tensile specimens Sample Temp. (C) YS (MPa) UTS (MPa) Crack Depth Standard Heat Treatment UCA4 25 1225 1710 54 m UCA23 25 1241 1696 84 m UCA5 649 1087 1471 0.38 mm UCA7 649 1103 1476 0.40 mm UCA8 760 1018 1180 130 m UCA10 760 1069 1145 35 m Alternate Heat Treatment UCB21 25 1066 1568 65 m UCB5 649 887 1395 0.41 mm UCB3 760 827 1123 77 m UCB8 760 780 1103 63 m

PAGE 182

165 From these data, an approximate stress inte nsity factor at failure was calculated based on the estimated flaw size and the ultimat e tensile strength. Results are shown in Table C-2. While it was more difficult to m easure the crack size in the fractured tensile Table C-2: Stress intensity factor for tensile tests Sample Temp. (C) UTS (MPa) Elongation (%) Kmax (MPa*m0.5) Standard Heat Treatment UCA4 25 1710 23.3 18 UCA23 25 1696 21.1 19 UCA5 649 1471 23.0 40 UCA7 649 1476 22.4 37 UCA8 760 1180 11.1 8 UCA10 760 1145 10.7 17 Alternate Heat Treatment UCB21 25 1568 25.3 16 UCB5 649 1395 26.0 35 UCB3 760 1123 17.1 12 UCB8 760 1103 17.4 11 samples compared to the fatigue samples, th e stress intensity factors calculated for the tensile samples at 649C and the fatigue samples at 649C are similar. Additionally, it follows that the stress intensity factor at failure would drop dramatically at 760C as a result of the large drop in ultimate tensile strength and total elongation. However, the results of a similar analysis for the ro om temperature tests do not exhibit a good correlation with the expected stress intensity factor at overload. The stress intensity factor at failure in these test s should be closer to the values at 649C, but are much lower than expected. This is probably due to the difficulty in measuring tensile fracture features on a sample that has undergone heavy plastic deformation and the very different types of deformation during te nsile and fatigue testing.

PAGE 183

166 LIST OF REFERENCES 1. R. Schafrik and R. Sprague, “Saga of Gas Turbine Materials: Part I,” Advanced Materials & Processes 162, No. 3, 2004, pp. 33-36. 2. C. T. Sims, “Superalloys: Gene sis and Character,” Superalloys II eds. C. T. Sims, N. S. Stoloff, and W. C. Hagel, W iley, New York, New York 1987, pp. 3-26. 3. Eds. J. R. Davis and Davis & Associates “Metallurgy, Processing, and Properties of Superalloys,” Heat-Resistant Materials ASM-International, Materials Park, Ohio, 1997, pp. 221-254. 4. E. W. Ross and C. T. Sims, “Nic kel-Base Alloys,” Superalloys II eds. C. T. Sims, N. S. Stoloff, and W. C. Hagel, W iley, New York, New York, 1987, pp. 97-133. 5. H. E. Miller and W. L. Chambers, “G as Turbine Design and Superalloys,” Superalloys II eds. C. T. Sims, N. S. Stoloff, and W. C. Hagel, Wiley, New York, New York, 1987, pp. 27-57. 6. R. Schafrik and R. Sprague, “Saga of Gas Turbine Materials: Part II,” Advanced Materials & Processes 162, No. 4, 2004, pp. 27-30. 7. N. S. Stoloff, “Wrought and P/M Superalloys,” ASM Handbook 1, ASMInternational, Materials Park, Ohio, 1993. 8. R. W. Guard and J. H. West brook, “Alloying Behavior of Ni3Al ( ’ phase),” Transactions of the Metallurgical Society of AIME 215, 1959, pp. 807-814. 9. R. G. Davies and N. S. Stoloff, “On the Yield Stress of Aged Ni-Al Alloys,” Transactions of the Metallurgical Society of AIME 233, 1965, pp. 714-719. 10. R. W. K. Honeycombe, The Pl astic Deformation of Metals 2nd ed., Edward Arnold, London, England, 1984. 11. G. E. Dieter, Mechanical Metallurgy 3rd ed., McGraw-Hill, Boston, Massachusetts, 1986. 12. N. S. Stoloff, “Fundamentals of Strengthening,” Superalloys II Eds. C. T. Sims, N. S. Stoloff, and W. C. Hagel, Wile y, New York, New York, 1987, pp. 61-96.

PAGE 184

167 167 13. J. E. Dorn, P. Pietrokowsky, and T. E. Ti etz, “The Effect of Alloying Elements on the Plastic Properties of Aluminum Alloys,” Transactions of AIME 188, 1950, pp. 933-943. 14. W. R. Hibbard, Jr., “Solid Solution Hardening,” Transactions of AIME, 212, 1958, pp. 1-5. 15. R. M. Pelloux and N. J. Grant, “Solid Solution and Second Phase Strengthening of Nickel Alloys at High and Low Temperatures,” Transactions of the Metallurgical Society of AIME 218, 1960, pp. 232-237. 16. R. L. Fleischer, “Substitu tional Solution Hardening,” Acta Metallurgica 11, 1963, pp. 203-209. 17. G. H. Gessinger, Powder Metallurgy of Superalloys Butterworths, London, England, 1984. 18. V. A. Tracey, G. T. Poyner, and J. F. Watkinson, Journal of Metals 13, 1961, pp. 363-369. 19. B. Ewing, F. Rizzo, and C. zurLippe, “Powder Metallurgy Products for Advanced Gas Turbine Applications,” Superalloys 1972 The Metallurgical Society of AIME, 1972, pp. BB1–BB20. 20. Eds. J. R. Davis and Davis & Associates “Powder Metallurgy Superalloys,” HeatResistant Materials ASM-International, Material s Park, Ohio, 1997, pp. 272-289. 21. R. D. Eng and D. J. Evans, “High Strength HIP Consolidated Merl 76 Disks,” Superalloys 1980 Eds. J. K. Tien, S. T. Wlodek, H. Morrow III, M. Gell, and G. E. Maurer, The Metallurgical Soci ety of AIME, 1980, pp. 491-500. 22. J. H. Moll and B. McTiernan, “Powde r Metallurgy Superalloys,” ASM Handbook 7, ASM-International, Materials Park, Ohio, 1993. 23. L. N. Moskowitz, R. M. Pelloux, and N. J. Grant, “Properties of IN-100 Processed by Powder Metallurgy,” Superalloys 1972 The Metallurgical Society of AIME, 1972, pp. Z1–Z25. 24. S. Reichman and D. S. Chang, “Powder Metallurgy,” Superalloys II Eds. C.T. Sims, N. S. Stoloff, and W. C. Hagel, Wiley, New York, New York, 1987, pp. 459493. 25. R. M. German, Powder Metallurgy Science 2nd ed., Metal Powder Industries Federation, Princeton, New Jersey, 1994. 26. G. Raisson and Y. Honnorat, “P.M Superalloy for High Temperature Components,” Superalloys 1976 Eds. B. H. Kear, D. R. Muzyka, J. K. Tien, and S. T. Wlodek, The Metallurgical So ciety of AIME, 1976, pp. 473-482.

PAGE 185

168 168 27. J. H. Moll and J. J. Conway, “Characteris tics and Properties of As-HIP P/M Alloy 720,” Superalloys 2000 Eds. T. M. Pollock, R. D. Kissinger, R. R. Bowman, K. A. Green, M. McLean, S. L. Olson, and J. J. Schirra TMS, 2000, pp. 135-142. 28. C. A. Morris and J. W. Smythe, “Compa rison of Astroloy Powder Consolidation Processes, Forging, and Testing of As troloy Powder Billet ,” Superalloys 1972 The Metallurgical Society of AIME, 1972, pp. Y1–Y26. 29. J. F. Barker and E. H. VanDerMolen, “E ffect of Processing Variables on PowderMetallurgy Ren 95,” Superalloys 1972 The Metallurgical Society of AIME, 1972, pp. AA1–AA23. 30. J. D. Buzzanell and L. W. Lherbier, “P rocessing Effects on the Properties of P/M Ren 95 Near Net Shapes,” Superalloys 1980 Eds. J. K. Tien, S. T. Wlodek, H. Morrow III, M. Gell, and G. E. Maurer, The Metallurgical Society of AIME, 1980, pp. 149-158. 31. J. N. Fleck, V. K. Chandhok, and L. P. Clark, “HIP of Near-Net Turbine Disks,” Superalloys 1976 Eds. B. H. Kear, D. R. Muzyka, J. K. Tien, and S. T. Wlodek, The Metallurgical Society of AIME, 1976, pp. 509-521. 32. J. J. Conway and F. J. Rizzo, “Hot Isostatic Pressing of Metal Powders,” ASM Handbook 7, ASM-International, Materials Park, Ohio, 1993. 33. E. Arzt, M. F. Ashby, and K. E. Easterling, “Practical Applicati ons of Hot-Isostatic Pressing Diagrams: Four Case Studies,” Metallurgical Transactions A 14A, 1983, pp. 211-221. 34. J. L. Bartos and P. S. Mathur, “Developm ent of Hot Isostatically Pressed (As-HIP) Powder Metallurgy Ren 95 Turbine Hardware,” Superalloys 1976 Eds. B. H. Kear, D. R. Muzyka, J. K. Tien, and S. T. Wlodek, The Metallurgical Society of AIME, 1976, pp. 495-508. 35. W. Betz, B. Borchert, and W. Track, “Com parative Assessment of PM Material For Turbine Disks,” Superalloys 1980 Eds. J. K. Tien, S. T. Wlodek, H. Morrow III, M. Gell, and G. E. Maurer, The Metallu rgical Society of AIME, 1980, pp. 643-650. 36. K. R. Bain and R. M. Pelloux, “Effect of Oxygen on Creep Crack Growth in PM/HIP Nickel-Base Supe ralloys,” Superalloys 1984 Eds. M. Gell, C. S. Kortovich, R. H. Bricknell, W. B. Kent, and J. F. Radavich, The Metallurgical Society of AIME, 1984, pp. 387-396. 37. D. M. Carlson, “P/M AF115 Dual Property Disk Process Development,” Superalloys 1980 Eds. J. K. Tien, S. T. Wlodek, H. Morrow III, M. Gell, and G. E. Maurer, The Metallurgical Soci ety of AIME, 1980, pp. 501-511.

PAGE 186

169 169 38. R. D. Kissinger, S. V. Nair, and J. K. Tien, “Influence of Po wder Particle Size Distribution and Pressure on the Kine tics of Hot Isostatic Pressing (HIP) Consolidation of P/M Superall oy Ren 95,” Superalloys 1984 Eds. M. Gell, C. S. Kortovich, R. H. Bricknell, W. B. Kent, and J. F. Radavich, The Metallurgical Society of AIME, 1984, pp. 285-294. 39. D. R. Chang, D. D. Krueger, and R. A. Sprague, “Superalloy Powder Processing, Properties and Turbine Disk A pplications,” Superalloys 1984 Eds. M. Gell, C. S. Kortovich, R. H. Bricknell, W. B. Kent, and J. F. Radavich, The Metallurgical Society of AIME, 1984, pp. 245-273. 40. C. Aubin, J. H. Davidson, and J. P. Tro ttier, “The Influence of Powder Particle Surface Composition on the Properties of a Nickel-Based Superalloy Produced by Hot Isostatic Pressing,” Superalloys 1980 Eds. J. K. Tien, S. T. Wlodek, H. Morrow III, W. B. Kent, and J. F. Rada vich, The Metallurgical Society of AIME, 1980, pp. 345-354. 41. G. E. Maurer, W. Castledine, F. A. Sc hweizer, S. Mancuso, “Development of HIP Consolidated P/M Superalloys for Conven tional Forging to Gas Turbine Engine Components,” Superalloys 1996 eds. R. D. Kissinger, D. J. Deye, D. L. Anton, A. D. Cetel, M. V. Nathal, T. M. Pollock and D. A. Woodford, TMS, 1996, pp. 645652. 42. M. Dahln and H. Fischmeister, “Car bide Precipitation in Superalloys,” Superalloys 1980 Eds. J. K. Tien, S. T. Wlodek, H. Morrow III, M. Gell, and G. E. Maurer, The Metallurgical Soci ety of AIME, 1980, pp. 449-454. 43. W. H. Chang, H. M. Green, and R. A. Sprague, “Defect Analyses of P/M Superalloys,” Rapid Solidification Pro cessing Principles and Technologies III Ed. R. Mehrabian, Bureau of Standards, 1982, pp. 500-509. 44. S. V. Nair and J. K. Tien, “Densification Mechanism Maps for Hot Isostatic Pressing (HIP) of Uneq ual Sized Particles,” Metallurgical Transactions A 18A, 1987, pp. 97-107. 45. J. C. Borofka, R. D. Kissinger, and J. K. Tien, “HIP Modeling of Superalloy Powders,” Superalloys 1988 Eds. S. Reichman, D. N. Duhl, G. Maurer, S. Antolovich, and C. Lund, TMS, 1988, pp. 111-120. 46. D. U. Furrer, “A Review of U720LI A lloy and Process Development,” Materials Design Approaches and Experiences Eds. J.-C. Zhao, M. Fahrmann, and T. M. Pollock, TMS, Warrendale, Pennsylvania, 2001, pp. 281-296. 47. F. E. Sczerzenie and G. E. Maurer “Development of Udimet 720 For High Strength Disk Applications,” Superalloys 1984 Eds. J. K. Tien, S. T. Wlodek, H. Morrow III, M. Gell, and G. E. Maurer, The Metallurgical Society of AIME, 1984, pp. 573-580.

PAGE 187

170 170 48. K. R. Bain, M. L. Gambone, J. M. Hyzak, and M. C. Thomas, “Development of Damage Tolerant Microstructure s in Udimet 720,” Superalloys 1988 Eds. S. Reichman, D. N. Duhl, G. Maurer, S. An tolovich, and C. Lund, TMS, 1988, pp. 1322. 49. G. A. Whitlow, C. G. Beck, R. Viswanathan, and E. A. Crombie, “The Effects of a Liquid Sulfate/Chloride E nvironment on Superalloy Stre ss Rupture Properties at 1300 F (704 C),” Metallurgical Transactions A 15A, 1984, pp. 23-28. 50. P. W. Keefe, S. O. Mancuso, and G. E. Maurer, “Effects of Heat Treatment and Chemistry on the Long-Term Phase Stabil ity of a High Strength Nickel-Based Superalloy,” Superalloys 1992 Eds. S. D. Antolovich, R. W. Stusrud, R. A. MacKay, D. L. Anton, T. Khan, R. D. Ki ssinger, D. L. Klarst rom, TMS, 1992, pp. 487-496. 51. J. R. Mihalisin, C. G. Bieber, and R. T. Grant, “Sigma-Its Occurrence, Effect, and Control in Nickel-Base Superalloys,” Transactions o the Metallurgical Society of AIME 242, 1968, pp. 2399-2414. 52. R. C. Reed, M. P. Jackson, and Y. S. Na, “Characterization and Modeling of the Precipitation of the Sigma Phase in UDIMET 720 and UDIMET 720LI,” Metallurgical and Materials Transactions A 30A, 1999, pp. 521-533. 53. N. Saunders, “Phase Diagram Calculations for Ni-Based Superalloys,” Superalloys 1996 Eds. R. D. Kissinger, D. J. Deye, D. L. Anton, A. D. Cetel, M. V. Nathal, T. M. Pollock, and D. A. Woodford, TMS, 1996, pp. 101-110. 54. D. J. Bryant and D. G. McIntosh, “The Manufacture and Evaluation of a Large Turbine Disc in Cast and Wrought Alloy 720LI,” Superalloys 1996 Eds. R. D. Kissinger, D. J. Deye, D. L. Anton, A. D. Cetel, M. V. Nathal, T. M. Pollock, and D. A. Woodford, TMS, 1996, pp. 713-722. 55. H. Hattori, M. Takekawa, D. Furrer, and R. J. Noel, “Evaluation of P/M U720 for Gas Turbine Engine Disk Application,” Superalloys 1996 Eds. R. D. Kissinger, D. J. Deye, D. L. Anton, A. D. Cetel, M. V. Nathal, T. M. Pollock, and D. A. Woodford, TMS, 1996, pp. 705-711. 56. M. P. Jackson and R. C. Reed, “Heat Tr eatment of UDIMET 720Li: The Effect of Microstructure on Properties”, Materials Science & Engineering A,” A259, 1999, pp. 85-97. 57. J. Mao and K. Chang, “Growth Kinetics of ’ Precipitates in P/M Superalloys,” Materials Design Approaches and Experiences Eds. J.-C. Zhao, M. Fahrmann, and T. M. Pollock, TMS, Warrendale, Pennsylvania, 2001, pp. 309-319. 58. D. Furrer and H. Fecht, “ ’ Formation in Superalloy U720LI,” Scripta Materialia 40, 1999, pp. 1215-1220.

PAGE 188

171 171 59. D. Furrer and H. Fecht, “Superalloys for Turbine Disc Applications,” JOM 51, 1999, pp. 14-17. 60. K. A. Green, J. A. Lemsky, and R. M. Gasior, “Development of Isothermally Forged P/M Udimet 720 for Turbine Di sk Applications,” Superalloys 1996 Eds. R. D. Kissinger, D. J. Deye, D. L. Anton, A. D. Cetel, M. V. Nathal, T. M. Pollock, and D. A. Woodford, TMS, 1996, pp. 697-703. 61. S. E. Kim, M. P. Jackson, R. C. Reed, C. Small, A. James, and N. K. Park, “Quanitification of the Minor Precip itates in UDIMET™ Alloy 720(LI) Using Electrolytic Extraction and X-ray Diffraction,” Materails Science and Engineering A A245, 1998, pp. 225-232. 62. M. F. Henry, Y. S. Yoo, D. Y. Yoon, and J. Choi, “The Dendritic Growth of ’ Precipitates and Grain Boundary Serrati on in a Model Nickel-Base Superalloy,” Metallurgical Transactions A 24A, 1993, pp. 1733-1743. 63. D. Furrer and H. Fecht, “Microstructure and Mechanical Property Development in Superalloy U720LI,” Superalloys 2000 Eds. T. M. Pollock, R. D. Kissinger, R. R. Bowman, K. A. Green, M. McLean, S. L. Olson, and J. J. Schirra, TMS, 2000, pp. 415-424. 64. H. L. Danflou, M. Marty, and A. Wald er, “Formation of Serrated Grain Boundaries and Their Effect on the Mechanical Propert ies in a P/M Nickel Base Superalloy,” Superalloys 1992 Eds. S. D. Antolovich, R. W. Stusrud, R. A. MacKay, D. L. Anton, T. Khan, R. D. Kissinger, and D. L. Klarstrom, TMS, 1992, pp. 63-72. 65. H. L. Danflou, M. Macia, T. H. Sander s, and T. Khan, “Mechanisms of Formation of Serrated Grain Boundaries in Nickel Base Superalloys ,” Superalloys 1996 Eds. R. D. Kissinger, D. J. Deye, D. L. Anton, A. D. Cetel, M. V. Nathal, T. M. Pollock, and D. A. Woodford, TMS, 1996, pp. 119-127. 66. ASTM Standard E 562-99, “Standard Test Method for Determining Volume Fraction by Systematic Manual Point Count,” 2001 Annual Book of ASTM Standards 3.01, ASTM, West Conshohocken, Pe nnsylvania, 2001, pp. 538-543. 67. ASTM Standard E 112-96, “Standard Test Methods for Determining Average Grain Size,” 2001 Annual Book of ASTM Standards 3.01, ASTM, West Conshohocken, Pennsylvania, 2001, pp. 237-259. 68. ASTM Standard E 930-99, “Standard Test Methods for Estimating the Largest Grain Observed in a Metallographic Section (ALA Grain Size),” 2001 Annual Book of ASTM Standards 3.01, ASTM, West Conshohocken, Pennsylvania, 2001, pp. 712-717. 69. D. L. Sponseller, “Differential Thermal Analysis of Nickel-Base Superalloys,” Superalloys 1996 Eds. R. D. Kissinger, D. J. Deye, D. L. Anton, A. D. Cetel, M. V. Nathal, T. M. Pollock, and D. A. Woodford, TMS, 1996, pp. 259-270.

PAGE 189

172 172 70. R. A. Ricks, A. J. Porter, and R. C. Ecob, “The Growth of ’ Precipitates in NickelBase Superalloys,” Acta Metallurgica, 31, 1983, pp. 43-53. 71. “Fatigue and Fracture Mechanics,” ASM Handbook 8, ASM-International, Materials Park, Ohio, 2003. 72. V. Kerlins, “Modes of Fracture,” ASM Handbook 12, ASM-International, Materials Park, Ohio, 2003. 73. T. L. Anderson, Fracture Mechanics CRC Press, Boca Raton, Florida, 1991. 74. J. E. King, “Effects of Grain Size and Microstructure on Threshold Values and Near Threshold Crack Growth in Powder-formed Ni-base Superalloy,” Metal Science 16, 1982, pp. 345-355. 75. A. Banik and K. Green, “The Mechani cal Property Response of Turbine Disks Produced Using Advanced PM Proce ssing Techniques,” Superalloys 2000 Eds. T. M. Pollock, R. D. Kissinger, R. R. Bowman, K. A. Green, M. McLean, S. L. Olson, and J. J. Schirra, TMS, 1996, pp. 69-74. 76. R. W. Herzberg, Deformation and Fractur e Mechanics of Engineering Materials 5th ed., Wiley, New York, New York, 1996.

PAGE 190

173 BIOGRAPHICAL SKETCH Darryl Slade Stolz was born in New Orleans, LA, in September of 1976. He graduated from Jesuit High School in New Orleans, LA, in May 1994. He then matriculated at the University of Notre Dame in South Bend, IN. While a student there, Slade was a member of the Men’s Swimming T eam, participating in the sprint freestyle and butterfly events. He graduated Cum Laude with a Bachelor of Science degree in chemical engineering, in May 1998. In August of 1998, he enrolled at the University of Florida in the Department of Materials Scien ce and Engineering. He received a Master of Science degree in May 2002. During the summer of 2002, Slade completed an internship with Pratt & Whitney Aircraft Engine s, in East Hartford, CT. He is currently scheduled to graduate with a Doctor of Philosophy degr ee in materials science and engineering, in August 2004.


Permanent Link: http://ufdc.ufl.edu/UFE0006480/00001

Material Information

Title: Effect of a supersolvus heat treatment on the microstructure and mechanical properties of a powder metallurgy processed nickel-base superalloy
Physical Description: Book
Language: English
Creator: Stolz, Darryl Slade ( Dissertant )
Fuchs, Gerhard E. ( Thesis advisor )
DeHoff, Robert T. ( Reviewer )
Kaufman, Michael J. ( Reviewer )
Mecholsky, John J. ( Reviewer )
Dempere, Luisa A. ( Reviewer )
Arakere, Nagaraj K. ( Reviewer )
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2004
Copyright Date: 2004

Subjects

Subjects / Keywords: Materials Science and Engineering thesis, Ph.D   ( local )
Dissertations, Academic -- UF -- Materials Science and Engineering   ( local )

Notes

Abstract: Powder Metallurgy (P/M) processed nickel-base superalloys are used as turbine disk materials in jet engines. The P/M processing results in a homogenous microstructure. Large amounts of strengthening elements can be incorporated into the chemistry of these P/M alloys. In addition, the ability to produce near net-shaped parts with powder consolidation may offer the potential for large cost savings. However, the fatigue properties of P/M superalloys in the as-consolidated form have suffered because of the defect sensitivity of the as-consolidated microstructure. Expensive, thermomechanical steps are necessary to break down defects, so that the P/M parts can be considered defect-tolerant. As a result, the true potential cost savings for using P/M superalloys in turbines have never been realized. This program was undertaken to examine the potential for utilizing an alternate heat treatment with P/M Alloy 720LI to generate a potentially defect-tolerant microstructure. This heat treatment had a soak above the ϐʹ solvus temperature followed by a controlled cool through the solvus. This produced gamma grains with a regular array of large dendritic-shaped secondary ϐʹ within the grains. Mechanical testing was carried out to fully evaluate the effect of this alternate heat treatment on the mechanical properties of Alloy 720LI. The standard heat treatment had longer lifetimes at the lower stress range conditions during high cycle fatigue; however, the alternate heat treatment was superior at the highest stress range. Fracture analysis suggests that this is due to the grain size difference. During tensile testing, the standard heat treatment had higher yield and ultimate strengths but lower ductility than the alternate heat treatment. This is thought to be due to the larger amounts of tertiary ϐʹ present in the microstructure produced by the standard heat treatment. Finally, the standard heat treatment had longer creep lifetimes at the lowest test temperature. The alternate heat treatment performed better at the higher test temperatures. While the microstructure did not improve the fatigue properties across the board, the improved understanding of the microstructural evolution during heat treatment will help in developing new heat treatments that may provide the defect-tolerance that is necessary.
Subject: Nickel, powder, superalloy, u720
General Note: Title from title page of source document.
General Note: Document formatted into pages; contains 190 pages.
General Note: Includes vita.
Thesis: Thesis (Ph.D.)--University of Florida, 2004.
Bibliography: Includes bibliographical references.
General Note: Text (Electronic thesis) in PDF format.

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: UFE0006480:00001

Permanent Link: http://ufdc.ufl.edu/UFE0006480/00001

Material Information

Title: Effect of a supersolvus heat treatment on the microstructure and mechanical properties of a powder metallurgy processed nickel-base superalloy
Physical Description: Book
Language: English
Creator: Stolz, Darryl Slade ( Dissertant )
Fuchs, Gerhard E. ( Thesis advisor )
DeHoff, Robert T. ( Reviewer )
Kaufman, Michael J. ( Reviewer )
Mecholsky, John J. ( Reviewer )
Dempere, Luisa A. ( Reviewer )
Arakere, Nagaraj K. ( Reviewer )
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2004
Copyright Date: 2004

Subjects

Subjects / Keywords: Materials Science and Engineering thesis, Ph.D   ( local )
Dissertations, Academic -- UF -- Materials Science and Engineering   ( local )

Notes

Abstract: Powder Metallurgy (P/M) processed nickel-base superalloys are used as turbine disk materials in jet engines. The P/M processing results in a homogenous microstructure. Large amounts of strengthening elements can be incorporated into the chemistry of these P/M alloys. In addition, the ability to produce near net-shaped parts with powder consolidation may offer the potential for large cost savings. However, the fatigue properties of P/M superalloys in the as-consolidated form have suffered because of the defect sensitivity of the as-consolidated microstructure. Expensive, thermomechanical steps are necessary to break down defects, so that the P/M parts can be considered defect-tolerant. As a result, the true potential cost savings for using P/M superalloys in turbines have never been realized. This program was undertaken to examine the potential for utilizing an alternate heat treatment with P/M Alloy 720LI to generate a potentially defect-tolerant microstructure. This heat treatment had a soak above the ϐʹ solvus temperature followed by a controlled cool through the solvus. This produced gamma grains with a regular array of large dendritic-shaped secondary ϐʹ within the grains. Mechanical testing was carried out to fully evaluate the effect of this alternate heat treatment on the mechanical properties of Alloy 720LI. The standard heat treatment had longer lifetimes at the lower stress range conditions during high cycle fatigue; however, the alternate heat treatment was superior at the highest stress range. Fracture analysis suggests that this is due to the grain size difference. During tensile testing, the standard heat treatment had higher yield and ultimate strengths but lower ductility than the alternate heat treatment. This is thought to be due to the larger amounts of tertiary ϐʹ present in the microstructure produced by the standard heat treatment. Finally, the standard heat treatment had longer creep lifetimes at the lowest test temperature. The alternate heat treatment performed better at the higher test temperatures. While the microstructure did not improve the fatigue properties across the board, the improved understanding of the microstructural evolution during heat treatment will help in developing new heat treatments that may provide the defect-tolerance that is necessary.
Subject: Nickel, powder, superalloy, u720
General Note: Title from title page of source document.
General Note: Document formatted into pages; contains 190 pages.
General Note: Includes vita.
Thesis: Thesis (Ph.D.)--University of Florida, 2004.
Bibliography: Includes bibliographical references.
General Note: Text (Electronic thesis) in PDF format.

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: UFE0006480:00001


This item has the following downloads:


Full Text












EFFECT OF A SUPERSOLVUS HEAT TREATMENT ON THE MICROSTRUCTURE
AND MECHANICAL PROPERTIES OF A POWDER METALLURGY PROCESSED
NICKEL-BASE SUPERALLOY














By

DARRYL SLADE STOLZ


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


2004

































Copyright 2004

by

Darryl Slade Stolz

































This dissertation is dedicated to the loving memory of Earl A. Stolz, Jr. and Bradford F.
Gifford, Jr.















ACKNOWLEDGMENTS

I would like to thank Dr. Gerhard E. Fuchs for all of his support and guidance over

the years. His knowledge of materials science in general, and superalloys specifically has

helped me grow tremendously as a researcher in the pursuit of my Ph.D. degree. I would

also like to thank all of my committee members for their guidance during this project:

Dr. Robert T. DeHoff, Dr. Michael J. Kaufman, Dr. John J. Mecholsky, Jr., Dr. Luisa A.

Dempere, and Dr. Nagaraj K. Arakere. I would like to especially thank Sean Conway

and Crucible Compaction Metals (Oakdale, PA) for all of the materials provided for this

study. In addition, I would like to acknowledge the help of Gerald R. Bourne and the

Major Analytical Instrumentation Center (MAIC) at the University of Florida for

assistance with my research. I thank all of the members of the High Temperature Alloys

Laboratory for their technical assistance, and perhaps more importantly for all of the

good memories that I will keep with me after I leave Gainesville. Finally, I would like to

thank my family for their love, support, and patience throughout my educational career.

Without all of this help, I truly could not have completed this project.
















TABLE OF CONTENTS



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

LIST OF TABLES .......................................... viii

LIST O F FIG U RE S ............... .......................................... ...x.... .. .... .x

ABSTRACT ................................................... ................. xvi

CHAPTER

1 BACKGROUND .................................... .. ......... ............ .............. .. 1

C om position and M icrostructure ..................................... ..................... ............... 2
C hem istry .............. ...................................................................3...........
G am m a M atrix ................................................................ ...... .. ............... 4
G am m a P rim e P hase ...................................................................... ...............4...
C arbides and B orides .... ............................................................... .............. 6
Strengthening M echanism s....................................... ......................... ............... 8
Pow der M etallurgy Superalloys ......................................................... 10
Pow der Processing... ................................................................................ 13
P ow der production ....................................... ........................ .............. 13
Pow der consolidation ....................................................... 16
C challenges .................................................................................................... 18
A lloy 720 ...................................................................................................... 27
Chem istry and processing ....................... ............................................ 27
M icrostructure ........................................................................................... 3 1
O b j e c tiv e s ................................................................................................................... 3 5

2 EXPERIM EN TAL PROCEDURES......................................................... ................ 37

M a te ria ls ..................... ............................................................ .................................3 7
D evelopm ent of H eat Treatm ent ........................................................... ................ 37
H eat T reatm ent T rials.......................................... ........................ ................ 39
H eat Treatm ent ............................................................................................... 41
C characterization ........... .. .................. ................ ............ .......... .... ........ ...........43
D eterm nation of y' Solvus............................................................. ................ 43
D ifferential therm al analysis .................................................. 44
M etallographic exam ination.................................................... ................ 44


v









M icrostructural A naly sis ....................................... ...................... ................ 46
M mechanical Testing ........................ .. ........... .......................................47
Low Cycle Fatigue Testing ......................................................... 48
H igh C ycle F atigue T testing ............................................................ ................ 54
T ensile T testing ............................................................................................... 55
C reep T testing .................................................................................................. 56
F ra ctu re A n aly sis ........................................................................................................5 9

3 R E S U L T S ............................................................................................ ..................... 6 0

M icrostructure...........................................................................60
Developm ent of Heat Treatm ent ......................................................................60
H eat treatm ent trials ..................................................................................61
H eat treatm ent ........................................................................................... 67
C haracterization ................................................................................................ 68
D eterm nation of y' Solvus........................................................................68
M icrostructural characterization ...............................................................72
M mechanical Testing and Fracture Analysis ............... ............................ ............... 75
L ow C ycle F atigue ...........................................................................................75
H igh C y cle F atigu e ............... ..................................... ............... ................ 8 1
H C F resu lts ........................................................ ............... .... ......... .. 82
H C F fractography .............................................. ...................... ... ......... .. 84
Tensile Testing ......... ..... .... .. .................................99
T ensile results............................................... ............... . ............ 99
Tensile fractography ..................................... ............... ...... 101
C reep T e stin g ......................................................................................... 10 5
Creep results ....................................................................... 105
C reep fractography ............................................ .................... ................ 107

4 D ISC U SSIO N ................................................................................... ............... 114

M ic ro stru ctu re ...........................................................................................................1 14
H eat Treatm ent ....... ................................................................................. 114
C characterization .................................................. ............... . ........... 116
Temperature of y' solvus.....................................117
M icrostructure ............. .... ......................................................... 118
M echanical Testing .............................................................................................124
H ardness Testing .............................. ............................................. 125
Fatigue Testing ............... ........ .. ....... ......................... 126
T en sile T estin g ...................................................................................... 13 9
Creep Testing ........................ .. ........... ................ ............... 143
C onclu sion s ... ....................................... ............... ..... .. ...... ........ 146
F u tu re W o rk ...... ..................................... ............... ........... .......... .. 14 8









APPENDIX

A DIFFERENTIAL THERMAL ANALYSIS GRAPHS.................... ...................150

B ST A T IST IC A L D A T A ............................................................................................... 153

Hardness Testing .................. ......... ............ ............... 153
The Volume Fraction of the Gamma Prime Phase ....................... ...................155
Gam m a Prim e Size Distributions ...... ........... .......... ...................... 158
G ra in S iz e ................................................................................................................ 1 6 2
Stress Intensity F actors .... ............................................................... .............. 163

C TENSILE FRACTURE MECHANICS .......... ..........................164

LIST O F R EFEREN CE S ... ................................................................... ............... 166

BIOGRAPH ICAL SKETCH .................. .............................................................. 173
















LIST OF TABLES

Table page

1-1: Engine system s using forged P/M superalloys........................................ ............... 12

1-2: Engines and airframe systems using as-HIP P/M superalloys. ................12

1-3: Nominal chemistry of Alloy 720 and Alloy 720LI. .............................................27

2-1: Sieve analysis of the master powder blends (MPBs) in weight percent.................38

2-2: Chemistry analysis of the master powder blends in weight percent. ......................38

2-3: H eat treatm ent trial m atrix ......................................... ........................ ................ 39

2-4: The L C F test m atrix..... .................................................................... .. ............. 54

2-5: The H CF test m atrix ............... ................ .............................................. 55

2-6 : T he T ensile test m atrix ........................................................................... ................ 56

2-7: The C reep test m atrix. ............................ ............................................ 58

3-1: The DTA results from M &P Laboratories ........................................... ................ 69

3-2: The DTA results from Dirats Laboratories .......................................... ................ 70

3-3: The total y' volume fraction, and ASTM grain size............................................... 74

3-4: Results from Rockwell C hardness tests for both heat treatments. ......................... 74

3-5: The HCF test results and crack initiation types.................................... ................ 87

3-6: Area of crack growth regions on the HCF fracture surfaces..............................94

3-7: Tensile test results. .................................. ....... ..................... 99

3-8: The creep results ............................ ............ ............................. 106

4-1: Stress intensity factors and plastic zone sizes for the initial crack in HCF tests..... 132

4-2: Stress intensity factors and plastic zone sizes for the final crack in HCF tests....... 134









B-1: The hardness testing statistical data...... ........ ...................... 153

B-2: The t-test data for the hardness testing. .......... ... ......................................... 154

B-3: The primary y' volume fraction data for the standard heat treatment...................156

B-4: The secondary y' volume fraction data for the standard heat treatment ...............156

B-5: The tertiary y' volume fraction data for the standard heat treatment....................157

B-6: The primary y' volume fraction data for the alternate heat treatment. ..................157

B-7: The secondary y' volume fraction data for the alternate heat treatment...............158

B-8: The primary y' size data for the standard heat treatment.................................. 159

B-9: The secondary y' size data for the standard heat treatment. ..............................159

B-10: The tertiary y' size data for the standard heat treatment................................. 160

B-11: The primary y' size data for the alternate heat treatment................................ 161

B-12: The secondary y' size data for the alternate heat treatment...............................161

B- 13: The grain size data for the standard heat treatment. .................. ...................163

B-14: The grain size data for the alternate heat treatment......................................163

B-15: The t-test data for the initial crack stress intensity factors. ..............................163

C-1: Area of crack in fractured tensile specimens. ...... ... ..................................... 164

C-2: Stress intensity factor for tensile tests...... .... ........................ 165















LIST OF FIGURES


Figure page

1-1. N ickel-alum inum phase diagram ........................................................... ...............5...

1-2: The Ni3Al solid solution field at approximately 11500C for various ternary alloys..6

1-3: Flow stress as a function of temperature for Ni3Al. .............................................7...

1-4: Effect of valency difference on hardening of nickel alloys. Nv is electron vacancy
num ber of the solute ..... .. ................................ .......................................... 9

1-5: V ertical gas atom izer .............................................................................. ............... 14

1-6: Formation of spherical metal powder by gas atomization...................................15

1-7: Material and fabrication savings with P/M processing of superalloys..................... 18

1-8: Plastic strain suffered by smaller and larger particles during HIPing of a bimodal
particle size distribution of pow ders ................................................... ................ 20

1-9: PPB initiation site in doped Rene 95 .......................................................... 21

1-10: Type 1 ceramic inclusion fatigue initiation site. ................................ ................ 22

1-11: Comparison of average LCF lives of HIP vs. HIP + Forge and Extrude + Forge
R e n e 9 5 .................................................................................................................. ... 2 4

1-12: Fatigue initiation at a crystallographic defect. ................................... ................ 25

1-13: Average low cycle fatigue life for a P/M superalloy during 1980 through 1996. ...25

1-14: Effect of thermal exposure on microstructures of subsolvus heat treated Alloy 720
an d A lloy 7 2 0L I. ...................................................................................................... 2 9

1-15: Experimental TTT diagram for the formation of 0.5 and 1.0 wt% of sigma in Alloy
7 2 0 L I. ..................................................................................................... .......... 3 0

1-16: Mean diameter of the cooling y' as a function of the interrupt temperature in Alloy
7 2 0 L I. ..................................................................................................... ........ .. 3 2

1-17: P/M A lloy 720 m icrostructure............................................................... ............... 34









1-18: O ptical m icrographs of A stroloy ....................................................... ................ 35

2-1: JE O L JSM 6400 SE M ............................................................................. ............... 42

2-2: C arbolite box furnace. ............................................... ............. ................ 45

2-3: Tensile sample design ........................................................................ 49

2-4: C reep sam ple design .. .................................................................... .............. 50

2-5: L ow cycle fatigue sam ple design ........................................................ ................ 51

2-6: Instron-Satec servo-hydraulic test fram e............................................. ................ 52

2-7: Instron-Satec C reep fram e ....................................... ........................ ................ 57

3-1: Heat treatment trial samples with various soak temperatures ............................. 62

3-2: Heat treatment trial samples with different cooling rates....................................63

3-3: Heat treatment trial samples with different soak times and final temperature before
fa n -a ir-c o o l .............................................................................................................. 6 4

3-4 : H eat treatm ent schedules ........................................ ......................... ................ 65

3-5: The standard and alternate heat treatment microstructures .................................66

3-6: MPB 96SW865 after solution heat treating at 11750C (21470F).......................... 66

3-7: MPB 96SW865 after solution heat treating at 11650C (21290F).......................... 67

3-8: The y' solvus trials quenched in iced brine after an hour at the solutioning
tem p eratu re .............................................................................................................. 7 1

3-9: The solvus trials that were fully solutioned and then control-cooled....................72

3-10: The y/y' microstructure in heat treated Alloy 720LI. ........................................73

3-11: Grain boundary microstructure of heat treated Alloy 720LI...............................74

3-12: Low cycle fatigue sample that was misaligned and tested in compression ..........76

3 -13 : P u sh /p u ll ro d s .......................................................................................................... 7 6

3-14: Low cycle fatigue sample with notches machined just after the threaded region and
before the shoulder of the sam ple. ...................... ............................................ 77

3-15: Ceramic inclusion type defect in standard heat treatment sample tested at 538C
(1000F) and a strain range of 1.1% ................... .............. ......... ................ 79









3-16: Standard heat treatment sample tested at 5380C (1000F) and a strain range of
1 .0 % .......................................................................................................................... 7 9

3-17: Alternate heat treatment sample tested at 538F (10000C) and a strain range of
1 .0 % .......................................................................................................................... 8 0

3-18: Relatively flat featureless region just after crack initiation in low cycle fatigue ....81

3-19: Representative micrographs of the fast fracture region of LCF samples .............82

3-20: High cycle fatigue S-N curves for specimens tested at 5380C (1000F). .............83

3-21: High cycle fatigue S-N curves for specimens tested at 6490C (10000F). ..............84

3-22: High cycle fatigue S-N curves for tests of the standard heat treatment at both
5380C (1000F) and 6490C (12000F)................ ............... 85

3-23: High cycle fatigue S-N curves for tests of the alternate heat treatment at 5380C
(10000F) and 6490C (12000F ) ............................................................. ................ 86

3-24: Optical micrographs of HCF fracture surfaces of samples tested at 6490C and 1086
M P a ....................................................................................................... ........ .. 8 6

3-25: Ceramic agglomerate as initiation point in standard heat treatment sample tested at
5380C (10000F) and a stress range of 1034 MPa (150 Ksi)...............................87

3-26: EDS spectra of crack initiation for standard heat treatment sample tested at 5380C
and 1034 M Pa ........................................................................... ........ ............... 88

3-27: Ceramic agglomerate as initiation point for standard heat treatment sample tested
at 5380C (10000F) and a stress range of 1086 MPa (157.5 Ksi) ...........................88

3-28: Slip band cracking as initiation point an alternate heat treatment sample tested at
5380C (10000F) and 993 M Pa (144 K si)............................................. ................ 89

3-29: Primary carbide as the initiation type for the standard heat treatment sample tested
at 6490C (12000F) and 1086 M Pa (157.5 K si).................................... ................ 89

3-30: EDS spectra of Ti-rich particle as initiation point of standard heat treatment sample
tested at 6490C and 1086 M Pa .................................... ..................... ................ 90

3-31: Unknown initiation type for alternate heat treatment sample tested at 649C
(12000F) and 1034 M Pa (150 K si)...................................................... ................ 91

3-32: Crack origin for alternate heat treatment sample tested at 5380C (10000F) and 1086
M P a (1 5 7 .5 K si) ....................................................................................................... 9 1

3-33: Initial crack region of fatigue sam ples ............................................... ................ 92









3-34: Fracutre surface near initiation in standard heat treatment sample tested at 538C
(1000F) and a stress range of 1034 M Pa (150 Ksi) .......................... ................ 93

3-35: Fracture surface of standard heat treatment tested at 5380C (10000F) and a stress
range of 1086 MPa (157.5 Ksi) outside of discolored region ...............................93

3-36: Fracture surfaces far from origin of samples tested at 5380C (10000F) and 993
M P a (14 4 K si) .......................................................................................................... 9 4

3-37: Initial crack area versus stress range for the high cycle fatigue specimens. ...........95

3-38: Final crack area versus stress range for the high cycle fatigue specimens..............96

3-39: Secondary cracks at primary and secondary y' precipitates in samples tested at
5380C and 993 M Pa. ............................ ........................................... 97

3-40: Secondary cracks in fatigue samples tested at 1034 MPa..................................97

3-41: Fatigue samples tested at 5380C and 1086 MPA showing secondary cracks .........98

3-42: Yield strength vs. temperature for both heat treatments...................................100

3-43: Elongation at failure vs. temperature for both heat treatments. ...........................101

3-44: Tensile stress-strain curve for standard heat treatment sample tested at 14000F. .102

3-45: Tensile stress-strain curve for alternate heat treatment sample tested at 14000F. .102

3-46: Crack initiation region in standard heat treatment tensile samples .....................103

3-47: Crack initiation region in alternate heat treatment tensile samples treatment....... 104

3-48: Larson-Miller plot for creep rupture in Alloy 720LI.................. ...................107

3-49: Fast fracture propagation path for standard heat treatment sample tested at 677C
(12500F) and 1034 M Pa (150 Ksi) ....... ....... ...................... 108

3-50: Fast fracture propagation path for alternate heat treatment at 7040C (13000F) and
689MPa (100Ksi) ........................... ......... ....................... 109

3-51: Standard heat treatment sample tested at 7600C (14000F) and 483 MPa (70 Ksi)
show ing fast fracture propagation path.. .............................................................. 109

3-52: Standard heat treatment sample tested at 6770C and 1034 MPa with crack initiation
at a grain boundary triple point. ...... ........... .......... .....................1... 10

3-53: Transgranular cracking during fatigue testing in samples tested at 6770C and 1034
M P a ........................................................... ........................................... ............. 1 1 1









3-54: Large secondary crack in alternate heat treatment sample tested at 6770C and 1034
M P a ........................................................... ............................................ ............. 1 1 1

3-55: Secondary cracks in high temperature creep tests...............................................111

3-56: Grain boundary crack initiation in alternate heat treatment creep samples........ 112

3-57: Secondary cracks in creep samples tested at 6770C (12500F) and 793 MPa (115
K si) ...................................................................................................... ....... .. 1 13

4-1: The tertiary and "quaternary" y' precipitates in the standard heat treatment ........ 119

4-2: Serrated grain boundary structure present in the alternate heat treatment
m icro stru ctu re ......................................................................................................... 12 0

4-3: Histogram ofy' volume fraction for the two heat treatments................................121

4-4: Secondary and tertiary y' precipitates. ...... ... ......................... 122

4-5: Comparison of fast fracture features during HCF fatigue and grain size.............131

4-6: Secondary cracking in standard heat treatment fatigue samples ..........................135

4-7: Secondary cracks propagating transgranularly in alternate heat treatment fatigue
sa m p le s ................................................................................................................ ... 1 3 6

4-8: Large pore at a grain boundary triple point in alternate heat treatment sample tested
at 538C and 1086 M Pa. ................ ............................................................ 136

4-9: Long secondary cracks in alternate heat treatment fatigue sample tested at 5380C
an d 10 8 6 M P a ......................................................................................................... 13 7

4-10: Yield strength vs. test temperature for both heat treatments...............................140

4-11: Optical pictures showing the fracture surfaces of tensile samples..................... 141

4-12: Alternate heat treatment creep sample tested at 7320C and 483 MPa showing
secondary cracks initiating at and propagating along grain boundaries...............144

4-13: Alternate heat treatment creep sample tested at 6770C and 1034 MPa showing void
nucleation at a grain boundary triple point and at the interface between secondary y'
precipitates and the m atrix. ................ ........................................................ 145

A-l: The DTA Graph of sample UMBCa......................................... 150

A-2: The DTA graph of sample UMBCb. ......... ......................... 150

A-3: The DTA graph of sample UMBEa........................................... 151









A-4: The DTA graph of sample UMBEb. ...... ... .......................... 151

A-5: The DTA graph of sample UMBCc. ...... ... .......................... 152

A-6: The DTA graph of sample UMBEc........................................... 150















Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

EFFECT OF A SUPERSOLVUS HEAT TREATMENT ON THE MICROSTRUCTURE
AND MECHANICAL PROPERTIES OF A POWDER METALLURGY PROCESSED
NICKEL-BASE SUPERALLOY

By

Darryl Slade Stolz

August 2004

Chair: Gerhard E Fuchs
Major Department: Materials Science and Engineering

Powder Metallurgy (P/M) processed nickel-base superalloys are used as turbine

disk materials in jet engines. The P/M processing results in a homogenous

microstructure. Large amounts of strengthening elements can be incorporated into the

chemistry of these P/M alloys. In addition, the ability to produce near net-shaped parts

with powder consolidation may offer the potential for large cost savings. However, the

fatigue properties of P/M superalloys in the as-consolidated form have suffered because

of the defect sensitivity of the as-consolidated microstructure. Expensive,

thermomechanical steps are necessary to break down defects, so that the P/M parts can be

considered defect-tolerant. As a result, the true potential cost savings for using P/M

superalloys in turbines have never been realized.

This program was undertaken to examine the potential for utilizing an alternate

heat treatment with P/M Alloy 720LI to generate a potentially defect-tolerant









microstructure. This heat treatment had a soak above the y' solvus temperature followed

by a controlled cool through the solvus. This produced y grains with a regular array of

large dendritic-shaped secondary y' within the grains.

Mechanical testing was carried out to fully evaluate the effect of this alternate

heat treatment on the mechanical properties of Alloy 720LI. The standard heat treatment

had longer lifetimes at the lower stress range conditions during high cycle fatigue;

however, the alternate heat treatment was superior at the highest stress range. Fracture

analysis suggests that this is due to the grain size difference. During tensile testing, the

standard heat treatment had higher yield and ultimate strengths but lower ductility than

the alternate heat treatment. This is thought to be due to the larger amounts of tertiary y'

present in the microstructure produced by the standard heat treatment. Finally, the

standard heat treatment had longer creep lifetimes at the lowest test temperature. The

alternate heat treatment performed better at the higher test temperatures. While the

microstructure did not improve the fatigue properties across the board, the improved

understanding of the microstructural evolution during heat treatment will help in

developing new heat treatments that may provide the defect-tolerance that is necessary.














CHAPTER 1
BACKGROUND

Superalloys have been used extensively as high temperature materials in turbine

engines since the 1950s [1]. In addition to jet engines and industrial gas turbines, they

have seen service in space vehicles, rocket engines, nuclear reactors, submarines, steam

power plants, petrochemical equipment, heating elements, and furnace parts [2, 3].

Superalloys are generally defined by their unique ability to retain high strength at

temperatures approaching 90% of their melting point and for times up to 100,000 hours at

slightly lower temperatures [4]. This unique feature for a structural material coupled with

the high oxidation resistance, makes superalloys the best choice for very demanding

environments like jet engines and industrial gas turbines.

There are three main classes of superalloys: nickel-base, cobalt-base, and

nickel-iron-base. However, Ni-base alloys generally are considered the most important,

and are the most widely used of these classes, owing to their excellent blend of

mechanical and chemical properties. As turbine engine materials, they are required to

have high strength as well as good creep, fatigue, and corrosion resistance [3]. The first

jet engine developed by Sir Frank Whittle operated with a thrust-to-weight ratio of 1.5:1

[1]. Modern commercial engines operate at a ratio of 6-8:1 with advanced engines

approaching 10:1. The increases in propulsion output in jet engines have been made

possible by improvements in materials as well as engine design. The demand for more

efficient and powerful turbines has steadily increased the operating temperature and

mechanical stress demands for superalloys. Ni-base superalloys have met these demands









through improvements in alloy processing and chemistry. Turbine blades have the

highest operational temperatures of the superalloys used in jet engines, making creep

resistance vital. Single crystal alloys with excellent creep resistance have been developed

as blade alloys in current jet engines. On the other hand, turbine disks operate at higher

loads but somewhat lower temperatures. They require good fracture toughness, low

crack growth rate, and ease of inspectability [5]. Turbine disk designers have examined

powder processing as a means of producing disks with higher strength and better fatigue

resistance. These alloys have an excellent blend of properties; however, their high costs

have limited their usage in turbine engines.

Composition and Microstructure

The alloy chemistry of Ni-base superalloys has evolved tremendously over the

years. The initial alloys used in gas turbine engines were modifications of

oxidation-resistant stainless steels [6]. The addition of aluminum and titanium to the

Nimonic (80% Ni, 20% Cr) alloy series led to the precipitation of the intermetallic phase,

Ni3Al. These were the first precipitation-hardenable superalloys. These alloys have

higher strength than alloys strengthened solely by solid-solution hardening, and than

those strengthened by oxide dispersion strengthening [3]. Precipitation-hardenable

superalloys are preferred for the most demanding environmental conditions. The first

superalloys had relatively simple compositions; however current generations can be a

confusing blend of varied elements that are added to improve processing, mechanical

properties, oxidation resistance, and even density. Even though there has been a

tremendous amount of research into the effect of chemistry changes on the properties of

these alloys, much still remains to be learned about the complex interactions between









various elements and the effects that slight changes in composition will have on the phase

stability and mechanical properties of the various alloys.

Chemistry

Precipitation hardenable Ni-base superalloy phase systems consist of a nickel

matrix, gamma (y), which is strengthened both by solid-solution alloying elements and by

the intermetallic precipitate, gamma prime (y'). Aluminum and titanium are the primary

elements added to increase the amount ofy' in the alloy [4]. Niobium and tantalum can

also be added as substitutional elements for Al and Ti. These y' forming elements are

added in amounts up to 8%. The most common solid-solution strengthening elements for

the y matrix are cobalt, iron, chromium, molybdenum, tungsten, titanium, and aluminum.

While titanium and aluminum are added mostly as precipitation hardeners, they do help

strengthen the y matrix by solid-solution substitution. Ni-base superalloys generally

contain 10-20% chromium, as it forms a protective oxide scale in addition to the

solid-solution strengthening benefits. Aluminum also forms a protective A1203 scale for

oxidation and hot corrosion resistance. Between 5 and 15% Co is added because of its

effect on the precipitation behavior of y'. The addition of Co effectively alters the slope

of the y' solvus curve, which changes the y' solvus temperature, depending on the y'

volume fraction in the alloy. Additionally, it increases the amount ofy' that will

precipitate out of solution during heat treatment. Small amounts of carbon, boron,

zirconium, and hafnium are added to improve grain boundary strength. Finally, it is

important that the amounts of so called "tramp" elements (like silicon, phosphorus,

sulfur, oxygen, and nitrogen) as well as other minor elements are carefully controlled to

very small levels.









Gamma Matrix

The continuous matrix phase, y, is a face centered cubic (FCC) Ni-base phase that

is austenitic in structure [4]. Pure nickel has neither a high elastic modulus nor a low

diffusivity, which are necessary to promote creep rupture resistance. Its ability to be used

at elevated temperatures for long times is attributed to the following three reasons:

* Nickel has a nearly filled third electron shell, which allows for high alloying
content.

* It has the tendency to form the protective Cr203 oxide scale with additions of
chromium.

* It has an added inclination to form the oxidation-resistant A1203 scale.

The y matrix is strengthened primarily by alloying with solid-solution elements.

Gamma Prime Phase

The precipitate phase, y', is an ordered L12 intermetallic phase that is stable over a

fairly narrow compositional window (Figure 1-1) [7]. Its nominal composition is Ni3Al;

however, both the nickel and the aluminum can be substituted for, by various alloying

elements [4]. Figure 1-2 [8] shows how various elements substitute in Ni3Al. As can be

seen, cobalt and copper will substitute for nickel. However, titanium, silicon, vanadium,

and manganese all partition to the aluminum sites. Molybdenum, chromium, and iron are

equally likely to substitute for nickel as they are for aluminum. Nickel atoms occupy the

face-centered sites and aluminum atoms the corner sites [8]. More recent studies have

found that niobium and hafnium also partition to the aluminum sites [4]. Ni3Al exhibits

long-range order up to its melting point, and precipitates coherently with the austenitic

matrix [3, 7]. In early superalloys, this precipitate was spherical; however as alloying













Nickel, at.%
10 20 30


40 50 60


70 BI 90


U
'1


E
H


1700 .-. I....-i i -1
1640 TC, 68.5%

1600 ---:


1500 --

8. 1455 C
1400 -- -- -- -.8 7%
1397 *C, 83% 1367 T
088%
1300 -


1200 ------
1135'C

1100 IG 1 -59


1000- __


900 -D --- -

28 42 56

660.37'

700 i------ -- ---- -- -
-../6. 640 *C




--A I


400 __ 1 I II I


I I I I /|59C

I I curie l
temperature
nn I I I I


2010






1650


1470


1290


1110


930


750


Al 10 20 3D 40 50 60 70 80 00 Ni
Nickel, wt%


Figure 1-1. Nickel-aluminum phase diagram. Reprinted with permission from N. S.

Stoloff, "Wrought and P/M Superalloys," ASM Handbook, 1, ASM-

International, 2003, Figure 1, p. 952.


content was increased, cuboidal precipitates were observed [3]. It was determined that


the change in y' morphology was related to the lattice mismatch between y and y'.


Spherical precipitates were noticed when the mismatch was between 0.0 and 0.2%. As


the mismatch increased to between 0.5 and 1.0%, cuboidal precipitates were observed.


Finally, plate-like y' were present when the mismatch reached 1.25% and above. Perhaps


2190

































Figure 1-2: The Ni3Al solid solution field at approximately 11500C for various ternary
alloys. Reprinted with permission from R. W. Guard and J. H. Westbrook,
"Alloying Behavior of Ni3Al (y' phase)," Transactions of the Metallurgical
Society ofAIME, 215, 1959, AIME, Figure 4, p. 810.

the most important property of y' is that its yield strength increases with increasing

temperature, making it ideal for high-temperature applications (Figure 1-3) [9].

Additionally, y' is generally not considered a fracture-initiation site, because of its

inherent ductility [4].

Carbides and Borides

Carbides and borides precipitate in Ni-base superalloys in various forms including

MC (metal carbide), M23C6, M6C, and M3B2. The role that they play in the

microstructure and mechanical properties of superalloys is fairly complex. In some

cases, their presence can be detrimental; however, other forms are beneficial to the

mechanical properties. While they can lower the ductility, their presence can also











NIAI
80 -

70 0 Present Results
2 AFjInnRef. 6
0 0

W 50 -
n-
I 40-
0
-J
IL
S30 -

20 -




-200 -100 0 100 200 300 400 500 600 700 800 900 1000
TEMPERATURE,*C


Figure 1-3: Flow stress as a function of temperature for Ni3Al. Reprinted with
permission from R. G. Davies and N. S. Stoloff, "On the Yield Stress of Aged
Ni-Al Alloys," Transactions of the Metallurgical Society ofAIME, 233, 1965,
AIME, Figure 4, p. 717.

increase the creep rupture life and the chemical stability of the matrix [4]. Davis et al. [3]

and Ross and Sims [4] describe different types of carbides. MC carbides generally

precipitate as discrete random cubic or script particles, along grain boundaries and within

the grains. They are FCC and have little or no orientation relationship to the matrix. In

superalloys, their preferred order of stability is HfC, TaC, NbC, and TiC (with HfC being

the most stable). The M23C6 carbides form primarily as discontinuous, blocky particles

with a complex cubic structure, during heat treatment or service between 760 and 980C

from the degeneration of MC carbides and carbon left in the matrix. They are generally

found along grain boundaries, but can also form on twin bands, stacking faults, and twin

ends. The formation reaction is


MC + y M23C6 + y'


These carbides form mostly in alloys with at least moderately high chromium

contents. With the presence of tungsten and molybdenum, they form with the









composition Cr2i(Mo, W)2C6. Finally, M6C carbides precipitate in blocky form along

grain boundaries, or sometimes as Widmanstatten plates within the grains [3]. These

carbides also have a complex cubic structure. They form between 815 and 9800C when

the molybdenum and tungsten contents are higher than 6-8 atomic percent. They are

stable at high temperatures, and their composition can range widely. The M3B2 borides

precipitate as hard particles, with a wide variety of shapes from blocky to half-moon.

They have a tetragonal unit cell, and act to improve grain boundary strength.

Strengthening Mechanisms

As stated earlier, the two main modes of strengthening in Ni-base superalloys are

solid-solution strengthening and precipitation hardening. Many mechanisms contribute

to strengthening from the substitution of solid-solution elements: size misfit, modulus

misfit, stacking-fault energy, and short-range order. The size of the substitutional atom

was shown to be an important factor in solid-solution strengthening [10]. The solute

atoms have small elastic stress fields that can impede dislocation motion [11]. However,

the model by Mott and Nabarro [12] to account for size misfit overestimates the

strengthening effect of solute atoms. The valence state of the solute atom also affects the

amount of solid-solution strengthening for a particular solute element [13, 14]. Pelloux

and Grant [15] confirmed these findings for superalloys (Figure 1-4). These valency

effects have been explained as being due to the modulus differences among different

alloys [16]. Local variations in the modulus can attract and pin dislocations [11].

Stacking-fault interactions are an additional solid-solution strengthening mechanism that

was first discovered by Suzuki [10, 11]. Solute atoms can preferentially segregate to

stacking faults. In turn, the stacking-fault energy is lowered, and the separation between










g 7.0 Ti

S5.6

-E 4.2 Cr o
0 Mo
O 2W 0Fe
>-2.8-
W Co
1.4 -

0 I I I I I
Ti V Cr Mn Fe Co Ni Solute
Mo Element
W
6.66 5.66 4.66 3.66 2.66 1.71 0.66-N,

Figure 1-4: Effect of valency difference on hardening of nickel alloys. Nv is electron
vacancy number of the solute. Reprinted with permission from R. M. Pelloux
and N. J. Grant, "Solid Solution and Second Phase Strengthening of Nickel
Alloys at High and Low Temperatures," Transactions of the Metallurgical
Society ofAIME, 218, 1960, AIME, Figure 5, p. 234.

partial dislocations is increased. The motion of these partial is much more difficult, as a

result. Finally, short-range ordering can occur in some nickel solid-solutions [12]. A

dislocation moving through a region with short-range order will require more work for

propagation, as it locally disrupts the energetically favorable ordered state [11].

Strengthening from precipitates can be considered as additive to solid-solution

strengthening. Precipitation of y' in superalloys causes strengthening due to coherency

strains between the matrix and precipitate, and the presence of order in the particles [12].

The strain field locally around each precipitate provides a repelling force for dislocation

motion. The mechanisms of order strengthening are controlled by the type of interaction

between the dislocations and the precipitates. Below a critical precipitate size,

dislocations will shear the y' particles. The ordered precipitate is left out of phase,

creating an antiphase boundary (APB) in the precipitate. This extra force needed to

create APB's in the alloy results in order strengthening. The size and spacing of the y'









precipitates has a large impact on the amount of strengthening from ordered particles

[11]. However, if the precipitates are incoherent and larger than a critical size,

dislocations will bypass them. For superalloys, this generally happens according to the

Orowan bowing model. As a dislocation approaches the large precipitates, it begins to

bow around them. The dislocation eventually bypasses the particles, leaving a

dislocation loop that exerts a stress on approaching dislocations. This stress field from

the dislocation loops results in an increase in the strength of the alloy.

Powder Metallurgy Superalloys

Powder metallurgy (P/M) superalloys were first examined as a novel way of

cooling turbine blades through transpiration cooling [17]. A mixture of elemental and

master-alloy powders were cold-compacted and sintered. However, it was only possible

to achieve approximately 90% of the theoretical density. As a result, these alloys had

inadequate strength for use as turbine blades. In the early 1960s, pre-alloyed powders

were developed through the use of water atomization [18]. However, these alloys

suffered from poor fatigue strengths in comparison to the conventional wrought alloys.

Eventually, with the use of vacuum melting and improved powder cleanliness in the mid

1960s, it was possible to produce P/M superalloys with sufficient mechanical properties,

that were a potential alternative to conventional cast and wrought alloys [17].

As turbine manufacturers sought more efficient jet engines in the early 1970s, the

property requirements for turbine disk materials increased beyond the capabilities of the

existing alloys [19]. The new alloys were required to have a high strength-to-weight

ratio, good hot workability, and higher operational temperatures. To meet these needs,

alloy designers began developing cast and wrought alloys with higher concentrations of

solid-solution strengthening elements, as well as y' forming elements (aluminum and









titanium) [20, 21]. However, these alloys experienced major difficulties with

segregation, hot workability, and ductility deficiencies as the alloying contents and ingot

size necessary for larger turbine disks increased [19, 22]. These newer cast and wrought

alloys were practically unforgeable [21]. With pre-alloyed powder processing, more

rapid solidification rates were easily possible. These faster solidification rates resulted in

much smaller dendrite-arm spacing within the powder particles, as opposed to those seen

in conventionally cast ingots [23]. The increase in homogeneity for these alloys by

powder processing makes them easy to forge isothermally through superplastic

deformation [19, 22].

Powder-processed disk alloys provide many advantages over disk alloys processed

by conventional ingot metallurgy [17, 20, 24].

* Reduced segregation in structure that allows for more heavily alloyed
compositions.

* Improved hot workability after compaction.

* Near net-shape production of parts that has potential for cost-savings and
conservation of rare and expensive elements.

* Finer grain size.

* Reduced carbon segregation.

* Ability to produce unique structures like oxide dispersion-strengthened (ODS)
alloys.

With all of these advantages, P/M superalloys were first put into operational service

in 1974 in the Pratt & Whitney F100 engine on the F-15 Eagle fighter jet [22]. A survey

of engines in use in 1996 shows that powder superalloys have become widely used in

both commercial and military jets since their inception. Engines that use P/M alloys









after extrusion and isothermal forging are shown in Table 1-1; while Table 1-2 shows

engines that incorporate as-hot isostatically pressed (as-HIP) alloys.


Table 1-1: Engine systems using forged P/M superalloys.*


Engine


Number produced
through 1996


GE T-700 9674
GE F-404/414 3077
GE-F110 2259
GE 90 38
PW F100 6496
PW 2000 951
PW 4000 1819
*Reprinted with permission from J. H. Moll and B. McTiernan, "Powder Metallurgy
Superalloys," ASM Handbook, 7, ASM-International, 2003, Table 1.

Table 1-2: Engines and airframe systems using as-HIP P/M superalloys. *
GE Aircraft Engines CFM International AlliedSignal
Turbine Airframe Turbine Airframe Turbine Airframe
CF6- B747-400, CFM56- GTC-131-
B 737-300 B2
80C2/E1 767 AWACS 3B1/B2 3[A]
F108-CF- CFM56- GTC-131-
-- KC-135-R CFM56 B 737-400 GT MD-90
100 3B2/3C 9[D]
FlO-GE- F-16 C/D/G CFM56-3B4 B 737-500 GT131- B737-X
100 9[B]
Fl10-GE- F-16 D/G FM56-5A A-320-100/- GTC-131- Airbus A319,
129 200 9[A] A320,A321
F 110-GE- CFM56- GTCP-331- B757/B767
400 5A/5B 200 B747-200B
F118-GE- CFM56- A-321-100 GTCP-331- A300, A310,
100 5B1/B2 250 C17A
F404-GE- F/A-18 C/D, CFM56- A-340- GTCP-331- Airbus A330,
400/402 F-117A 5C2/C4 200/300 350 A340
T-700-GE- Bell AH 1W, GTCP-331-
Sikorsky SH- CFM56-7 B737-700 B777
401 60B/F 500
Gulfstream V
T-700-GE- Sikorsky CFM56-8 B 737- RE-220 Global
700 UH-60A/L 600/800 Express CRJ-
700
*Reprinted with permission from J. H. Moll and B. McTiernan, "Powder Metallurgy
Superalloys," ASM Handbook, 7, ASM-International, 2003, Table 2.

The initial P/M processed superalloys were derived through slight modifications

to the chemistries of existing alloys like IN-100, Astroloy, and Rene 95 [22]. Eventually,









alloys like AF115 and AF2-1DA were designed specifically for P/M processing, to take

advantage of the high level of alloying that can be accommodated by this route.

However, these alloys had low resistance to crack growth from defects in the

microstructure. With defect tolerance in mind, current research has centered on more

defect-tolerant alloys like N18, Rene 88DT, and Udimet 720.

Powder Processing

A wide range of powder processing methods is available to a turbine disk

manufacturer. These include powder production, consolidation, post-consolidation

thermo-mechanical processing (TMP), and heat treatment. All of these different

processing routes give the manufacturer the flexibility to alter the microstructure and,

therefore, the mechanical properties to fit the requirements for almost any particular

application.

Powder production

Many different powder fabrication techniques are used for the production of metal

powders. These include mechanical fabrication like machining or milling, electrolytic

fabrication, chemical fabrication, and atomization [25]. Of all these different techniques,

gas atomization to produce pre-alloyed powder is the most important for the production

of P/M superalloys [17]. Within the category of atomization, there are a few

subcategories of powder fabrication processes including inert-gas atomization, soluble

gas atomization (vacuum atomization), centrifugal atomization including

rotating-electrode process (REP) and electron-beam rotating process (EBRP), and rapid

solidification techniques [17, 20, 22].

The REP process has received some attention from turbine disk manufacturers [26];

however, the most research into powder fabrication of superalloys has examined inert-gas









atomization. Typical cooling rates for inert-gas atomization of superalloys are in the

range of 104 to 106 C/sec [27]. A typical gas atomization rig is shown in Figure 1-5

[25]. In this process, vacuum induction melting (VIM) is used to melt high-purity raw



melt vacuum
induction
gas melter
source





i j i ,,,, -chamber




fine
powder


collection
c chamber


Figure 1-5: Vertical gas atomizer. Reprinted with permission from Powder Metallurgy
Science, 1994, Metal Powder Industries Federation, 105 College Road East,
Princeton, New Jersey, USA. Figure 3.14, p. 101.

materials in a crucible at the top of the atomization rig [22]. The melted alloy passes

through the nozzle in a thin stream. High-pressure argon gas is then impinged the stream,

to break up the melt stock into fine spherical droplets. Alternatively, nitrogen gas can be

used; however, it is necessary to alter the alloy composition to account for any increase in

nitrogen and the effects that will have on the final alloy. These droplets solidify as they

descend through the cooling tower to the bottom of the atomization rig, where they are

collected for consolidation. The fine particles are collected in a side unit, using a cyclone

separator.









The particle size and size distribution can be affected by many variables during

the atomization process. Generally, the more energy that is imparted onto the molten

metal by the gas, the finer the powder yield produced [25]. Depending on the nozzle

design, the gas can impinge the metal stream either at an angle or tangentially. In

addition to the geometric parameters of the nozzle, other factors influencing the powder

characteristics are gas pressure, type of gas, diameter (d') of the melt stream, flow

characteristics of the molten material, and the ratio of mass transport of the melt and

atomizing gas [17]. An expanded view of the metal stream during atomization illustrates

the development of spherical particles from the initial melt stream (Figure 1-6) [25].

melt stream
nozzle-Y \
ggas
gas "
expansion -'
zone sheet

0 R ligament

o 8ellipsoid
0 D
0 0 0o 0 sphere
0O 00


Figure 1-6: Formation of spherical metal powder by gas atomization. Reprinted with
permission from Powder Metallurgy Science, 1994, Metal Powder Industries
Federation, 105 College Road East, Princeton, New Jersey, USA. Figure
3.16, p. 103.

The gas around the molten stream expands, causing a depressurization of the liquid

metal [25]. The stream, in turn, expands into a hollow cone. The high

surface-area-to-volume ratio of this cone leads to instability in the geometry that, in

conjunction with the inert gas stream, leads to breakup of the cone into long ligaments.









Eventually, these ligaments break up into ellipsoids, and finally into small spherical

droplets. The amount of superheat in the metal before atomization plays a key role in

determining the final shape and size of the powder. The level of superheat must be

sufficient so that the particles do not solidify before becoming spheres. It is also crucial

to eliminate any agglomerations of particles or satellite particles by proper design of the

atomization process. Either of these can potentially result in poor packing of the powder

before consolidation.

Powder consolidation

Conventional powder-compaction techniques, like cold pressing and sintering, are

insufficient in achieving a fully dense P/M superalloy billet, because of the

incompressibility of superalloy powder, and the susceptibility to forming oxide scales at

sintering temperature [17]. Therefore, compaction techniques that involve high pressure

at elevated temperatures are necessary. Early work on P/M superalloy compaction

techniques examined forge compaction [28] and extrusion [28, 29] as means of achieving

fully dense P/M billets. Another novel compaction process that has received some

attention is consolidation by atmospheric pressure (CAPR) [30]. However, the

advancements in HIP processing of powders along with potential cost savings make this

the preferred route for consolidation of superalloy powders into billet form.

The basic principle behind HIPing is the application of high pressure gas to

consolidate the powder within an autoclave at elevated temperatures. The heater is

generally located inside of the pressure vessel. Atomized powder, loaded into a

container, is placed in the autoclave for compaction. These containers can be fabricated

out of sheet metal, glass, or ceramic. While the glass and ceramic molds provide for

better flexibility in both size and final part complexity, the metal cans are not as brittle









[17, 31]. They can handle higher autoclave loading. It is ideal for the powder to be

spherical and have a wide size distribution, to facilitate higher packing densities before

compaction [32]. The containers are loaded under vibration, to achieve packing densities

approaching 80% of the theoretical density before consolidation. The cans are then

evacuated and sealed before being placed in the HIP chamber. During HIPing,

densification is thought to take place in a similar manner to sintering [32]. The

mechanisms describing the initial bonding of powder particles during a HIP cycle [32,

33] are as follows:

* Initial elastic deformation.
* Plastic deformation of the particles by dislocation.
* Power-law creep or dislocation creep.
* Nabarro-Herring creep or volume diffusional creep.
* Coble creep or grain boundary diffusional creep.

The final stages of densification occur through the same mechanisms that govern neck

growth during sintering [32].

It is possible to produce a wide range of sizes and shapes of final parts in the as-

HIP condition with proper container selection and design. Through production of

net-shape and near net-shape HIPed parts, there is a great potential for reduction in

starting material for P/M superalloy parts (Figure 1-7) [20]. This production of net-

shaped parts, in addition to the intrinsic homogeneity of atomized powders, translates

directly into a cost savings for turbine disk manufacturers, as they could possibly

eliminate forging operations, excess material, and costs associated with machining the

final part [34, 35]. In 1976, the development of as-HIP turbine disks was viewed as "one

of the most exciting and technologically important metallurgical developments in recent











years" and one that could "revolutionize the aircraft engine rotating parts industry" [34,

p.502].

Conventional processing
vacuum melting mnd Hot isostatic pressing Direct hot
cold forging plus hot forging isostatic pressing

Ingot ( c
.. P/M reform Near-net shape part
95 kg (210 Ib) 33kg (72 Ib) 18kg (401b)

Pancake
forge /
93 kg (205 Ib)

Blocker
forge -
91 kg (200 Ib)

Finish 33 kg (72 Ib)
forge
77 kg (170 Ib)

Machined
for heat
treating 75 kg (165 Ib)
20 kg (45 Ib) 18 kg (40 Ib)
_. __ Sonic bC
Machined 32 kg (165 Ib)
sonic Jj-- [ 1
envelope



5 kg(11 b)
Finished part

Figure 1-7: Material and fabrication savings with P/M processing of superalloys.
Reprinted with permission from Eds. J. R. Davis and Davis & Associates,
"Powder Metallurgy Superalloys," Heat Resistant Materials, ASM-
International, 1997, Figure 1, p. 272.

Challenges

Unfortunately, these potential cost savings have never been fully realized by the

turbine disk manufacturers. Powder-processed superalloys in the as-HIPed + heat treated

condition do not contain enough inherent damage-tolerance for use as disk alloys in jet

engines. "A damage tolerant design assumes that a component has a flaw (i.e. crack) of a

size just below the non-destructive inspection detectable limit. The times to inspection

and/or retirement are based on the crack growth rate of the flaw." [36, p. 388] This









approach enables disk manufacturers to design around potential failures through

knowledge of, and confidence in, tensile strength and creep and fatigue lifetimes. While

as-HIPed P/M superalloys had sufficient tensile and creep capabilities through heat

treatment alone [34, 37], it was soon evident that the low cycle fatigue (LCF) lifetimes

contained large amounts of scatter and were highly dependant on the cleanliness of the

powder and the consolidated billets freedom from defects [35]. The three main defect

types found in P/M superalloys are prior particle boundaries, ceramic inclusions, and

thermally induced porosity [38].

Prior particle boundaries (PPBs) have been the most extensively studied of these

defect types for P/M superalloys. A HIPed billet that suffers from PPB defects will

exhibit this network through the whole microstructure causing this defect type to be the

most detrimental to the fatigue lifetime of the alloy. Powder particles that are left

undeformed from the HIP processing can become decorated with detrimental secondary

phases during the HIP cycle. Larger particles are strained to a much smaller degree than

small particles during a typical HIP cycle (Figure 1-8) [38]. PPB networks tend to form

on these larger undeformed powder particles [39]. Through careful analysis, it has been

determined that the PPB defects consist of a semi-continuous network of carbides,

oxides, oxy-carbides, and perhaps oxy-carbonitrides [23, 40, and 41]. It is thought that

during the initial stages of densification during HIPing that titanium and carbon within

the powder migrate to the particle surfaces as a result of oxides on the outside of the

particles [40]. This results in a film of stable titanium oxy-carbides on the prior particle

boundaries. These carbides are mostly MC in nature. However, there is some debate as




















5
H'- /








0
2
largr particles
1-

0 I I
0.65 0.70 0.75 0.80
Relative Density

Figure 1-8: Plastic strain suffered by smaller and larger particles during HIPing of a
bimodal particle size distribution of powders. Reprinted with permission from
R. D. Kissinger, S. V. Nair, and J. K. Tien, "The Infuluence of Powder
Particle Size Distribution and Pressure on the Kinetics of Hot Isostatic
Pressing (HIP) Consolidation of P/M Superalloy Rene 95," Superalloys 1984,
Eds. M. Gell, C. S. Kortovich, R. H. Bricknell, W. B. Kent, and J. F.
Radavich, The Metallurgical Society of AIME, 1984, Figure 6, p. 291.

to the mechanisms for precipitation of these deleterious carbide and oxide precipitates.

Some theorize that oxides on the powder particle surfaces serve as nucleation sites for

carbides and borides [41]. However, others claim that these surfaces are not preferential

sites for carbide nucleation [42]. Only extremely fine carbides were found on the surface

of these powder particles. Instead, the metal-metal interfaces between powder particles

that have stuck together as the result of collisions during atomization are the preferential

sites for carbide precipitation. A detailed study by D. R. Chang et al. [39] examined the

effects of various processing defects on mechanical properties of P/M Rene 95. PPBs

were discovered to have an average size of approximately 9,700 rtm2 with a max size up









to 161,300 ntm2 when samples were made from -150 (< 106 [tm) mesh powder. In this

study, -150 mesh Rene 95 was seeded with various dopants to induce fatigue failures at

PPBs and ceramic inclusions. The dopants chosen to lead to PPB networks were Buna N

(an organic carbide former) and mill scale (an inorganic oxide former). The average

lifetime of as-HIPed Rene 95 was reduced from 37,000 cycles in the baseline (no dopant)

condition to 8,675 cycles in the seeded conditions. Additionally, the frequency of failure

at PPB sites was increased from 5% to 95%. A typical PPB fracture initiation site from

this study is shown in Figure 1-9 [39].















Figure 1-9: PPB initiation site in doped Rene 95. Reprinted with permission from D. R.
Chang, D. D. Krueger, and R. A. Sprague, "Superalloy Powder Processing,
Properties and Turbine Disk Applications," Superalloys 1984, Eds. M. Gell,
C. S. Kortovich, R. H. Bricknell, W. B. Kent, and J. F. Radavich, The
Metallurgical Society of AIME, 1984, Figure 8A, p. 258.

Ceramic inclusions have also been detrimental to the fatigue properties of

as-HIPed powder superalloys. A typical failure origin associated with ceramic defects is

displayed in Figure 1-10 [39]. These inclusions originate from the melting crucible, the

pouring tundish, and the atomizing nozzle with aluminum, zirconium, magnesium, and

calcium being the major metallic elements [39]. The composition of the inclusions

depends entirely on the ceramic materials used during production of the powder. In -150


























Figure 1-10: Type 1 ceramic inclusion fatigue initiation site. Reprinted with permission
from D. R. Chang, D. D. Krueger, and R. A. Sprague, "Superalloy Powder
Processing, Properties and Turbine Disk Applications," Superalloys 1984,
Eds. M. Gell, C. S. Kortovich, R. H. Bricknell, W. B. Kent, and J. F.
Radavich, The Metallurgical Society of AIME, 1984, Figure 3A, p. 250.

mesh Rene 95, these inclusions were found to have an average size of approximately

6500 tnm2 and on some occasions up to 64,500 trm2. A study by W. H. Chang et al. [43]

seeded -150 mesh Rene 95 with A1203 to facilitate fatigue initiation at ceramic inclusions.

LCF specimens were tested at 10000F and 0.66% strain range. The average lifetime was

decreased from 36,840 cycles for unseeded material to 3,411 cycles for Rene 95 seeded

with A1203.

The third type of defect associated with premature failure of P/M superalloy

specimens tested in fatigue is thermally induced porosity (TIP). During argon

atomization, some of the larger particles can form voids that contain entrapped argon.

The entrapped gas can then expand at high temperatures during HIP cycles or heat

treatment, forming much larger voids that have a deleterious effect on the fatigue

lifetimes of P/M alloys. These voids are generally less than 1300 trm2. Due to their









smaller size, TIP defects are not as detrimental to fatigue life as PPB's and ceramic

inclusions.

To overcome the detriment in properties that these defects have caused in as-HIP

P/M superalloys, turbine disk manufacturers have turned to thermomechanical processing

(TMP) of the alloys. Rene 95 was HIPed and isothermally forged to an 80% reduction

[43]. This hot working resulted in a large increase in LCF lifetime. The extrusion step

broke up and reduced the size of the PPB networks; however only had a small effect on

the ceramic inclusions. A second study examined the effect of consolidation by extrusion

followed by isothermal forging [39]. The extrusion was carried out with a 6.5 to 1

reduction ratio. The forging resulted in a 60% reduction of the consolidated material. As

a result, the tensile strength and the LCF lifetime were improved dramatically (Figure

1-11) [39]. The extrusion and isothermal forging steps served to break up both the

ceramic inclusions and PPB networks. As a result, the maximum size of these defects in

the TMP material is much smaller, reducing their tendency to be crack initiation sites.

More samples failed at voids and crystallographic origins than in the non-TMP material.

Figure 1-12 [39] is a fractograph of a typical crystallographic crack initiation site.

Longer lifetimes are generally seen from samples that fail at these defects.

Unfortunately, the TMP steps needed to produce a defect-tolerant microstructure are

costly and limit the usage of as-HIP powder superalloys in the turbine industry.

Research into process improvements in recent years has improved the fatigue

lifetime of as-HIP material to the point that the original hope for cost reduction for

turbine disks is becoming more achievable. The average LCF lifetime of as-HIP alloys

has increased dramatically as can be seen in Figure 1-13 [22]. Improvements in the gas












EX + F 750F LCF






HIP VE.VE.






8 4 B
10 10 10 to10
CYCLES TO FAILURE


EX + F 100OF LCF



VEXF AVE.



HIP AVE.






1 4 B
10 10 10 10
CYCLES TO FAILURE


H + F 1000F LCF







HIP AVE.






10 10 to10 10
CYCLES TO FAILURE


Figure 1-11: Comparison of average LCF lives of HIP vs. HIP + Forge and Extrude +
Forge Rene 95. Reprinted with permission from D. R. Chang, D. D. Krueger,
and R. A. Sprague, "Superalloy Powder Processing, Properties and Turbine
Disk Applications," Superalloys 1984, Eds. M. Gell, C. S. Kortovich, R. H.
Bricknell, W. B. Kent, and J. F. Radavich, The Metallurgical Society of
AIME, 1984, Figure 11, p. 262.



























Figure 1-12: Fatigue initiation at a crystallographic defect. Reprinted with permission
from D. R. Chang, D. D. Krueger, and R. A. Sprague, "Superalloy Powder
Processing, Properties and Turbine Disk Applications," Superalloys 1984,
Eds. M. Gell, C. S. Kortovich, R. H. Bricknell, W. B. Kent, and J. F.
Radavich, The Metallurgical Society of AIME, 1984, Figure 10B, p. 260.

250
Average life in LCF strain-controlled tests at 1000 OF (540 "C)
";" 200


150


S 100


50



1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996
Year

Figure 1-13: Average low cycle fatigue life for a P/M superalloy during 1980 through
1996. Reprinted with permission from J. H. Moll and B. McTiernan, "Powder
Metallurgy Superalloys," ASM Handbook, 7, ASM-International, 2003,
Figure 30, p. 900.

atomization process have resulted in fewer ceramic inclusions. Switching from

commercial purity argon to high purity argon limited the number of particulates

contained in the argon gas itself [38]. Studies on the melting practices have shown that









magnesia-based crucibles and electroslag remelting can reduce the oxide concentrations

compared to other crucibles or melting methods [17]. Additionally, the size of potential

inclusions has been reduced simply by sieving the atomized powder to finer particle

sizes. The average particle size has been reduced from approximately 250 jtm to smaller

than 105 |tm [22]. This has been made possible in part by the ability to produce finer

powder yields from atomization. Reducing the levels of boron and carbon in the

chemistry of P/M superalloys also helps minimize the PPBs. More care has also been

taken when handling the powder before consolidation, to limit the amount of

contaminants. Improvements in modeling the particle deformation and resulting densities

of HIP cycles have allowed powder manufacturers to better understand HIP processing

and to optimize the HIP cycle parameters to minimize the effect of defects on the LCF

lifetimes. Original studies on deformation during HIPing concentrated on monosized

particles [33]. To more closely approximate real world multimodal particle distributions,

models were developed to study the densification of a bimodal distribution through HIP

[44, 45]. Using these models, HIP cycles that provide more uniform deformation to all

particles can be developed, resulting in an improved as-HIP material. Some have even

suggested performing the HIP cycles at higher sub-solidus temperatures to facilitate grain

growth past the PPB's. Finally, improvements in non-destructive evaluation (NDE)

techniques have been used to help insure that defects are of sufficiently small size before

any parts are put into service. It is with these improvements in superalloy powder

processing in mind that a recent study examined the mechanical properties of as-HIP

Alloy 720 [27]. With cleaner production methods and a supersolvus heat treatment, the

properties compared favorably with extruded + isothermally forged Alloy 720.









Alloy 720

Alloy 720 LI (low inclusion chemistry) is a powder processed nickel-base

superalloy that has begun to be widely used in both the aircraft and land-based gas

turbine industries over the last ten years [46]. It is a derivation from Alloy 720, which

was originally developed as a cast and wrought turbine blade alloy for use in land-based

gas turbines [47, 48]. To limit segregation from casting, Alloy 720 LI can be processed

through powder metallurgy as an alternative to the traditional cast and wrought

processing route.

Chemistry and processing

Alloy 720 was initially designed to be chemically resistant to environmental

attack through both oxidation and sulfidization. The composition of Alloy 720 and Alloy

720LI are given in Table 1-3 [46]. Titanium and aluminum were added as precipitation

Table 1-3: Nominal chemistry of Alloy 720 and Alloy 720LI.
Ni Cr Co Mo W Ti Al C B Zr
Alloy 720 Bal 18.0 14.8 3.0 1.25 5.0 2.5 0.035 0.033 0.03
Alloy Bal 16.0 15.0 3.0 1.25 5.0 2.5 0.010- 0.010- 0.03
720LI 0.025 0.020

strengtheners because of their tendency to form y'. Molybdenum, tungsten, chromium,

and cobalt are all used for solid-solution strengthening of the y matrix. Alloy 720

contains a high amount of chromium for oxidation and sulfidization resistance. The

formation of TiC, Mo2B3, and Cr23(C,B)6 along the grain boundaries effectively limits the

diffusion of oxygen and sulfur along the grain boundaries [47, 49]. This helps to retard

the crack growth rate at elevated temperatures compared to other cast and wrought blade

materials.









As a blade material, Alloy 720 was engineered to be used for up to 10,000 hours at

9000C. A coarse grained microstructure was optimal for turbine blade applications where

creep rupture lifetime was vitally important. However, because of Alloy 720's

combination of exceptional strength and hot workability, turbine manufacturers became

interested in using the alloy in a fine grained condition as a potential turbine disk material

[47].

It was soon discovered that the fine grain Alloy 720 is fairly susceptible to the

formation of the deleterious sigma (G) phase [50]. The formation of the topologically

closed-packed (TCP) c phase led to a rapid deterioration in the tensile ductility, creep

resistance, and toughness of the alloy. The most common nucleation sites for CY phase in

superalloys are at large grain boundary y' and at carbides [51]. The CY phase forms along

the grain boundaries, as globular precipitates similar to chromium carbides. Alternatively,

it can form as platelet or needle-like precipitates. cy phase has an approximate

composition of (Cro.5, Moo.1), (Nio.2, Co00.2) in Alloy 720 [52]. Not only can these

precipitates act as crack initiation sites, but they also can deplete the matrix of chromium

[52]. The loss of chromium will lower the strength of the matrix, thereby causing a

deficit in the mechanical properties of the alloy as a whole.

The chemistry was then refined to the current Alloy 720 LI composition to

improve phase stability and reduce casting defects such as carbide and boride stringers

[46]. The amount of chromium was reduced by two weight percent to decrease the

stability of c [53]. Additionally, the carbon and boron levels were reduced to decrease

the number of nucleation sites available for the formation of G. The change in chemistry









has a noticeable impact on the microstructure of exposed samples as can clearly be seen

in Figure 1-14 [50]. These precipitates were found to be M23(C,B)6 and CY phase. Reed et






























Figure 1-14: Effect of thermal exposure on microstructures of subsolvus heat treated
Alloy 720 and Alloy 720LI. A) Alloy 720 as-heat treated. B) Alloy 720 after
7500C for 1000 hours. C) Alloy 720LI as-heat treated. D) Alloy 720LI after
7500C for 1000 hours. Reprinted with permission from P. W. Keefe, S. 0.
Mancuso, and G. E. Maurer, "Effects of Heat Treatment and Chemistry on the
Long-Term Stability of a High Strength Nickel-Based Superalloy,"
Superalloys 1992, Eds. S. D. Antolovich, R. W. Stusrud, R. A. MacKay, D. L.
Anton, T. Khan, R. D. Kissinger, D. L. Klarston, TMS, 1992, Figure 4, p. 493.

al. [52] constructed an experimental time-temperature-transition (TTT) diagram for Alloy

720 and Alloy 720LI that shows that the chemistry change effectively delayed the onset

of y formation for Alloy 720LI. (Figure 1-15) [52]. The time for 1.0% cy formation has









850


S800 -
-A-




700
6 750 -- i






650 ... .
100 1000 10000
Time (hours)

Figure 1-15: Experimental TTT diagram for the formation of 0.5 and 1.0 wt% of sigma
in Alloy 720LI. Reprinted with permission from R. C. Reed, M. P. Jackson,
and Y. S. Na, "Characterization and Modeling of the Precipitation of the
Sigma Phase in UDIMET 720 and UDIMET 720LI," Metallurgical and
Materials Transactions A, 30A, 1999, Figure 9, p. 529.

been increased from 10 hours for Alloy 720 to over 1000 hours for Alloy 720LI.

Additionally, the nose of the TTT diagram has been decreased to about 7500C for the

new composition, as opposed to 8000C for the original Alloy 720. It is important to note

that Alloy 720 LI exhibits excellent microstructural stability below 7000C, which is

above the temperatures that a high pressure turbine disk rim will see in service.

While Alloy 720LI is able to be produced by traditional cast and wrought

processing, it is difficult to produce a billet of the alloy that is homogenous in grain size

and chemistry. It is being processed in up to 250 mm diameter billets for Rolls-Royce

engines [54]. However, even with the most current melting and ingot conversion

techniques, Alloy 720LI suffers from y' banding and from inhomogeneous grain sizes.

Unrecrystallized grains up to ASTM 0 grain size can be found near the center of the billet









and can be difficult to fully eliminate. As a result, many turbine manufacturers have

preferred the powder metallurgy processing route.

Microstructure

Alloy 720LI has been reported in literature to contain approximately 55 volume

percent y' [46]. Both its microstructure and the resultant mechanical properties can be

easily manipulated through heat treatment and forging [55]. Like many Ni-base

superalloys, Alloy 720LI has a multi-modal distribution of y' precipitates. It is important

to note that the nomenclature for the different y' sizes in P/M superalloys is different than

that used for single crystal superalloys. In P/M superalloys, primary y' consists of large

blocky precipitates that nucleate discretely along the y grain boundaries. These particles

decorate and pin the y grain boundaries, impeding grain growth during solution heat

treatment. The secondary and tertiary y' precipitates are formed by nucleation and

growth during cooling from a supersolvus heat treatment. It has been proposed that the

secondary y' particles nucleate during quenching from the heat treatment temperature

[56]. As the temperature continues to decrease, the secondary y' precipitates grow in

size, rejecting solute into the matrix. This continues until the diffusion rate of the solute

is too slow for the precipitates to grow any further. The alloy eventually cools to a point

where the driving force for nucleation of the tertiary y' is large enough to occur. These

particles grow in size up to 90 nm during the aging heat treatment. It is thought that the

nucleation and growth process for the secondary y' is in competition with that of the

tertiary y' [57]. At slower cooling rates, the secondary y' can grow larger, leaving less

supersaturation in the matrix for nucleation of the tertiary y'. However, at faster cooling

rates, the supersaturation in the matrix is greater when the necessary undercooling for









nucleation of the tertiary y' is reached, resulting in a larger volume fraction of ultrafine

Y.

It is readily possible to tailor the size and morphology of the secondary (cooling)

y' through heat treatment alone. With a constant cooling rate of 550C/min, the

temperature at which an Alloy 720LI sample was removed and quenched had a great

impact on the size of the secondary y' (Figure 1-16) [57]. The size of the secondary y'

0.4

0.35 -U720LI- --


0 T
z 0.3 .-- --- -



M .




0.05 y--0--4X -+- -- 3.97E-- -
rr 0.1524 -- ---

^ 0.1 --_1-..----.


R2 = 9.61E-01
0 I I _. __-
1200 1100 1000 900 800 700 600
INTERRUPT TEMPERATURE, aC

Figure 1-16: Mean diameter of the cooling y' as a function of the interrupt temperature in
Alloy 720LI. Reprinted with permission from J. Mao and K. Chang, "Growth
Kinetics of y' precipitates in P/M Superalloys," Materials Design Approaches
and Experiences, Eds. J.-C. Zhao, M. Fahrmann, and T. M. Pollock, TMS,
2001, Figure 5, p. 314.

was increased from 0.15mm to 0.25mm by changing the test interrupt temperature from

11210C to 6500C. Additionally, the cooling rate used during a supersolvus solution heat

treatment greatly effects on the size and morphology of the cooling y' [58]. A wide range

of varying morphologies were produced by controlling the cooling rates with different

quenching media and by wrapping the samples in insulation to mimic the cooling rates

seen by large turbine disks. As the cooling rate fell below 1.00C/sec, the y' size rose









dramatically from approximately 0.2 |tm to 1.0 |tm. Using all of this data, a TTT curve

was constructed for the precipitation of different y' morphologies in supersolvus heat

treated Alloy 720LI [58]. The shape of secondary y' progresses from spherical to

dendritic to fan-type precipitates as the cooling rate is slowed down.

Although the carbon and boron levels in Alloy 720LI are low, carbides and borides are

still present and impact the mechanical properties. The P/M material contains small

uniform carbide precipitates, while the cast and wrought material has large primary

carbides [59]. The most common type of carbide found in this alloy is the MC carbide.

The M23C6 carbides and M3B2 borides are also present in smaller amounts. As the

cooling rate is decreased, the amount of these two phases increased to the point where

they can exceed the amount of primary carbides at the slowest cooling rates. It was also

discovered that aging time increased the amount of M23C6 and M3B2 present in the alloy.

These carbides and borides precipitate mostly along the grain boundaries. They appear as

the fine white precipitates in the micrograph below (Figure 1-17) [60]. For this alloy, the

MC carbide composition is predicted to be (Cro.8oMoo.19Nio.o01)23C6 [61]. The borides on

the other hand are expected to be of the composition (Moo.76Cro.24)3B2. After electrolytic

extraction and x-ray diffraction analysis, it was discovered that the weight fraction of Y,

M23C6, and M3B2 for subsolvus heat treated cast and wrought Alloy 720LI were all below

0.5% even after 3000 hour exposures at 8000C [61]. The MC-type carbides were found

in micrographs of Alloy 720LI but not in Alloy 720 in this study. However, the amount

in Alloy 720LI was not able to be measured by x-ray due to the overlap of their peaks

with the peaks of M23C6 carbides and M3B2 borides, in addition to the low amount of MC

carbides present in this study. These grain boundary carbides have been shown to reduce









grain boundary sliding at elevated temperatures which improves the creep resistance of

the alloy [17]. The knowledge and ability to control these two additional strengthening

mechanisms in P/M superalloys is important in improving the mechanical properties of

these alloys.

















Figure 1-17: P/M Alloy 720 microstructure. Reprinted with permission from K. A.
Green, J. A. Lemsky, and R. M. Gasior, "Development of Isothermally Forged
P/M Udimet 720 for Turbine Disk Applications," Superalloys 1996, Eds. R.
D. Kissinger, D. J. Deye, D. L. Anton, A. D. Cetel, M. V. Nathal, T. M.
Pollock, and D. A. Woodford, TMS, 1996, Figure 5, p. 701.

In addition, the grain boundary microstructure itself plays an important role in

the mechanical properties of Alloy 720LI. Through control of heat treatment, it is

possible to form serrations in the grain boundaries. They form during controlled cooling

through the y' solvus [62]. Their amplitude and period are determined by the

homogenization temperature, the cooling rate, and the final temperature for the controlled

cooling. Their presence has been noted in Alloy 720 LI by Furer and Fecht [63]. It is

generally thought that these serrations develop as a result of coarsening of the y'

precipitates close to the grain boundaries. These grain boundary serrations (Figure 1-18)









[64] have been demonstrated to have a positive effect on creep fatigue crack growth rate

by impeding grain boundary sliding [64, 65].

(a) (b)














Figure 1-18: Optical micrographs of Astroloy. A) "Smooth" grain boundaries. B)
Serrated grain boundaries. Reprinted with permission from H. L. Danflou, M.
Marty, and A. Walder, "Formation of Serrated Grain Boundaries and Their
Effect on the Mechanical Properties in a P/M Nickel Base Superalloy,"
Superalloys 1992, Eds. S. D. Antolovich, R. W. Stusrud, R. A. MacKay, D. L.
Anton, T. Khan, R. D. Kissinger, and D. L. Klarstrom, TMS, 1992, Figure 1,
p. 65.

Objectives

The increased homogeneity of powder processed Ni-base superalloys has led to

dramatic improvements in the mechanical properties compared to traditional cast and

wrought processing. However, the as-consolidated and heat treated P/M superalloys

donot contain enough inherent defect tolerance to warrant usage in applications like jet

engines, where unexpected failure can lead to tragic consequences. These alloys need

expensive thermomechanical processing steps to reduce the size of defects so that they

are below the critical crack initiation length for fatigue failure. Elimination of these

processing steps would result in a tremendous reduction in the cost of finished and flight-

ready turbine disks.









The purpose of this study was to try to improve and limit the scatter in mechanical

properties of as-HIPed Alloy 720LI through heat treatment alone. In unpublished

research by Pratt & Whitney, a supersolvus heat treatment improved the fatigue lifetime

and fracture toughness of single crystal superalloys. A modification of this heat

treatment was given to Alloy 720LI to determine the effect of a supersolvus heat

treatment with a slow, controlled cooling rate on the microstructure, mechanical

properties, and fracture mechanics of P/M Alloy 720LI. The creep lifetime, tensile

strength, and fatigue lifetimes of specimens with a supersolvus heat treatment and a

standard subsolvus heat treatment were compared. Detailed fracture analysis of the

tested specimens was also performed to determine the mechanisms of failure for the two

heat treatments.














CHAPTER 2
EXPERIMENTAL PROCEDURES

Materials

All materials used in this study were processed by Crucible Compaction Co. in

Oakdale, PA. The alloys were gas atomized in the 2268 Kg (5000 lb) production gas

atomization unit, using high-purity argon as the atomizing gas. The powder was

collected and then sieved to -230 mesh (63 [tm) size, with the majority of the powder

below 400 mesh (38 [tm). All powder particles coarser than the 230 mesh size were

removed from the powder lot and returned as revert to be used in further atomization

runs. The powder was then loaded into stainless or mild steel HIP cans and compacted.

The HIP cycle used for all material in this study was below the gamma prime (y') solvus

(11290C/100.0 MIPa/4 h or 20650F/14.5 Ksi/4 h). The test material came from eleven

different master powder blends atomized at Crucible. There were only small differences

in particle size and overall composition from one master powder blend to another. The

sieve analysis for the various master powder blends is listed in Table 2-1. The

compositions of all of the master powder blends, along with the nominal composition of

the Alloy 720 LI, are listed in Table 2-2. All samples were provided in the as-HIPed

condition.

Development of Heat Treatment

Previous research by Pratt & Whitney has indicated that a supersolvus heat

treatment followed by a slow cool through the y' solvus temperature would produce the

desired microstructure and mechanical properties in their single crystal alloys. In this










Table 2-1: Sieve analysis of the master powder blends (MPBs) in weight percent


95SW860 0 0.7 8.4 13.2 77.7
95SW861 0 0.6 8.1 13.7 77.6
96SW863 0 0.5 8.2 14.5 76.8
96SW864 0 0.5 6.4 12.7 80.4
96SW865 0 0.4 5.6 13.1 80.9
96SW887 0 0.9 8.9 14.2 76.0
96SW888 0 0.9 6.4 14.9 77.8
97SW984 0 0.4 7.5 11.1 81.0
97SW985 0 0.2 6.2 10.8 82.8
98SW035 0 0.5 5.7 13.4 80.4
98SW036 0 0.2 6.0 11.4 82.4



Table 2-2: Chemistry analysis of the master powder blends in weight percent
MPB Ni Cr Co Mo W Ti Al C B Zr

Nominal Bal 16.57 14.71 3.00 1.28 5.02 2.49 0.010 0.012 0.038
95SW860 Bal 16.64 14.78 3.00 1.27 4.95 2.48 0.008 0.013 0.037
95SW861 Bal 16.67 14.88 3.01 1.28 4.91 2.55 0.010 0.013 0.040

96SW863 Bal 16.85 14.87 2.81 1.26 5.00 2.52 0.011 0.012 0.038
96SW864 Bal 16.53 14.70 3.00 1.25 4.98 2.58 0.011 0.013 0.038
96SW865 Bal 16.60 14.77 2.99 1.26 5.00 2.52 0.009 0.012 0.040

96SW887 Bal 16.64 14.74 3.00 1.27 4.98 2.47 0.012 0.014 0.037
96SW888 Bal 16.57 14.71 3.00 1.28 5.02 2.49 0.010 0.012 0.038
97SW984 Bal 16.73 14.72 3.00 1.26 4.92 2.43 0.007 0.014 0.037

97SW985 Bal 16.73 14.72 3.01 1.26 4.91 2.45 0.006 0.011 0.037
98SW035 Bal 16.6 14.63 3.01 1.29 5.03 2.47 0.011 0.011 0.040
98SW036 Bal 16.7 14.65 2.99 1.27 4.93 2.49 0.010 0.012 0.040



microstructure, there is a regular array of dendritic-shaped secondary y' in the y matrix.

However, specific details of the Pratt & Whitney heat treatment were not available. To

determine the alternate supersolvus heat treatment schedule needed to produce this


MPB


+230


-230/+270


-270/+325


-325/+400


-400









microstructure in Alloy 720, various heat treatment parameters were studied including

soak temperature, soak time, cooling rate, and final temperature before fan air cooling.

The final heat treatment was verified through microstructural analysis.

Heat Treatment Trials

Two solution heat treatments of Alloy 720LI were compared in this study. The

control samples were given a standard subsolvus heat treatment from the literature. This

heat treatment schedule is 11000C (20120F)/2 h. After two hours, the samples were

fan-air-cooled (FAC). The alternate heat treatment was developed to match the

microstructure that Pratt & Whitney's study obtained after a supersolvus heat treatment

for their single crystal superalloys. Since these P/M alloys are very different

compositionally than the single crystal alloys utilized by Pratt & Whitney, the exact heat

treatment temperature and the cooling rate necessary to obtain the desired microstructure

in the Alloy 720LI were not known. As a result, it was necessary to perform several heat

treatment trials to determine the temperature and cooling rate needed to produce the

alternate microstructure. The test matrix for determining the alternate heat treatment

schedule is shown in Table 2-3. The samples were heat treated in a Lindberg model

Table 2-3: Heat treatment trial matrix
pID Soak Hold Cooling Final
Temp.(C) Time(h) Rate(C/min) Temp.
A2 1175 2 3.3 1020
A3 1200 2 7.4 1020
A4 1225 2 7.2 1020
C1 1175 2 2.75 1075
C2 1175 2 0.82 1020
D2 1175 2 2.03 1020
Al 1175 4 2.75 1020
UMB1 1175 2 3.0 1020
D1 1200 2 2.75 1020









54233-V tube furnace with a Eurotherm model 818S temperature controller with a 50.8

mm (2 in) diameter mullite tube. The control thermocouple was a type R thermocouple

with alumina sheathing positioned radially to the hot zone, but outside the mullite tube.

A type K thermocouple, with 304 stainless steel sheathing, was inserted through the front

end of the tube furnace into the hot zone to monitor the actual temperature of the samples

throughout the heat treatment cycle. All samples were heated from room temperature to

their hold temperature, held for two hours, and then cooled slowly in the furnace at

various cooling rates to a final temperature of 10200C (18680F). This final temperature

was selected in order to insure that the samples were cooled at least 1000C (1800F) below

the solvus temperature so that all precipitation from cooling was accomplished before the

more rapid cooling in air.

Upon completion of the heat treatments, the samples were removed from the

furnace and allowed to fan-air-cool to room temperature. They were metallographically

prepared by first mounting in diallyl phthalate using a Leco model PR-10 mounting press.

The metallographic samples were then coarsely ground on a Leco model BG 20 belt

grinder with a 60 grit SiC belt to remove the oxidation layer and to polish far enough into

the sample to offset any temperature loss that may have occurred during transfer from the

furnace to the quench bucket. The metallographic samples were then polished by hand

using successive SiC papers of 120, 240, 320, 400, 600, and 800 grit. At each grit size,

the samples were polished in the same direction until all of the scratches from the

previous step had been removed. The sample was then rotated 900 and polished on the

next highest grit paper. Once they were polished to 800 grit paper, fine polishing was

performed using a Buehler Billiard cloth on a Leco model VP-50 polishing wheel, with









alumina suspensions of 15 rtm (5.9X10-4 in) and 5 rtm (2.0X10-4 in). The final polishing

was performed on the same metallographic polishing wheel, with alumina suspensions of

1.0 |tm (3.9X10- in) and 0.3 |tm (1.2X105 in) on a Buehler Microcloth. Great care was

taken between polishing steps to clean the samples in water using a Leco ultrasonic bath

and to clean the polishing wheels, so there would be no contamination of larger sized

alumina particles during the fine polishing. The surfaces were examined for scratches

with an Olympus light optical microscope before moving on to the next polishing step.

After polishing, the specimen surfaces were cleaned with methanol. Pratt & Whitney

etchant #17 (25 mL HC1, 25 mL HNO3, 25 mL H20, and 0.75 g molybdic acid) was then

swabbed on the samples with cotton swabs for 5-20 s to etch the microstructure. This

etchant selectively attacks the y' precipitates to the y matrix. These samples were then

examined using a JEOL JSM 6400 (Figure 2-1) scanning electron microscope (SEM) at

15 KV accelerating voltage and a 15 mm (0.59 in) working distance.

Heat Treatment

After the alternate heat treatment schedule was determined through these heat

treatment trials and microstructural analysis, the heat treatment was carried out on the

provided Alloy 720 LI mechanical testing bars at the University of Florida in an Applied

Test Systems box furnace. When the samples were removed from the furnace they were

given a fan-air-cool to room temperature. A type K thermocouple was used as the control

thermocouple, with a second type K thermocouple with Nextel sheathing positioned

directly on the sample bars for accurate monitoring of the temperature of the bars. This

second thermocouple was placed under the test bars so that the bead of the thermocouple

was in contact with the test bars at all times. Half of the bars received the standard




























Figure 2-1: JEOL JSM 6400 SEM

solution heat treatment. The alternate heat treatment was 11750C (21470F) for 2 h with a

3.00C/min (5.40F/min) cooling rate in the furnace to 10200C (18680F). The samples

were then removed and given a fan-air-cool to room temperature. While this heat

treatment worked for the majority of the master powder blends, the soak temperature had

to be reduced to 11650C (21290F) to produce the desired microstructure for three of the

MPBs as will be discussed in the results section. The same aging heat treatment was

given to all sample bars regardless of the solution heat treatment. The aging heat

treatment used in this study was the standard Alloy 720 aging heat treatment found in

literature. The schedule was 7600C (14000F) for 24 h with a fan air cool and then 650C

(12000F) for 16 h followed by a fan air cool. This heat treatment was performed in the

same box furnace with the same thermocouple set-up as during the solution heat

treatments.

Thirty-six bars were provided in the as-HIP condition for creep or tensile testing.

These bars were approximately 19.1 mm (0.75 in) on a side and 101.6 mm (4.0 in) long.









Eighteen creep/tensile bars were heat treated using the standard heat treatment, while the

other eighteen were solution heat treated according to the alternate heat treatment

schedule. In addition, thirty-five bars were provided in the as-HIP condition for low

cycle fatigue testing. These bars were approximately 19.1mm (0.75 in) by 25.4 mm (1.0

in) on edge and 152.4 mm (6.0 in) in length. Seventeen of the LCF bars were given the

standard heat treatment, while the other eighteen test bars were given the alternate heat

treatment. To verify the heat treatment, small microstructural analysis samples of each

master powder blend were included along with the mechanical testing bars in each of the

heat treatment runs. These samples were cubes with dimensions of approximately 19.1

mm (0.75 in) on edge.

Characterization

After the heat treatments were complete, both the y' solvus temperature and the

microstructures of the two heat treatments were fully characterized. It was important to

determine the solvus temperature in order to understand the microstructural evolution of

Alloy 720LI during the heat treatments. The microstructural characterization of the two

heat treatments served not only to help understand the precipitation behavior during the

heat treatments, but also to help determine the effect of the two microstructures on the

mechanical properties.

Determination of y' Solvus

The y' solvus temperature was determined in two complimentary ways. The two

methods were differential thermal analysis and metallographic examination of the

microstructure at various temperatures. These results were then compared to accurately









determine the y' solvus temperature, the solidus temperature, and the liquidus

temperature of Alloy 720LI.

Differential thermal analysis

Differential thermal analysis (DTA) was performed at two independent

laboratories. Samples for DTA were sectioned using a Leco diamond saw to produce

specimens approximately 6.4 mm (0.25 in) by 9.5 mm (0.375 in) by 15.9 mm (0.625 in)

in size. All DTA samples were in the as-HIPed condition with no heat treatment. Two

samples from each of two different master powder blends were sent to M&P Laboratories

in Schenectady, NY. These samples were heated at a constant rate of 100C/min

(180F/min) to 16000C (29120F). The temperature difference between the sample and a

reference sample of pure nickel was measured. These results were plotted versus

temperature. One more sample of each of the two MPBs was sent to Dirats Laboratories

in Westfield, MA. These samples were heated at 200C/min (360F/min). Once again, the

temperature difference between the sample and a reference sample was measured. Two

curves were plotted for these results. The first is of the temperature difference versus

temperature. The second curve is of the derivative temperature difference versus

temperature. All reaction temperatures were then identified by the inflection points

present on the plots.

Metallographic examination

In addition to DTA, solvus temperatures were determined by metallographic

examination of quenched microstructures. For this portion of the study, a bar of Alloy

720 LI in the as-HIP condition was sectioned into smaller samples using a LECO cut-off

saw. The metallographic examination samples were cut to approximately 6.4 mm (0.25

in) by 19.1 mm (0.75 in) by 19.1 mm (0.75 in). A Nextel-sheathed type K thermocouple









was attached to each sample with 80Ni-20Cr wire to measure the temperature of the

sample accurately while in the furnace. These samples were then heated in a Carbolite

model CWF 13/23 box furnace with PM 2000 elements (Figure 2-2) to various

temperatures above and below the expected solvus temperature. One set of samples was























Figure 2-2: Carbolite box furnace

heated to temperatures at 5C (9F) intervals between 1100C (2012F) and 1200C

(21920F). These samples were allowed to soak at temperature for one hour before being

removed with tongs and quickly dropped into an iced-brine quench to "freeze in" the

microstructure at the elevated temperature and prevent the formation ofy' during cooling.

This quenching medium was prepared by filling a stainless steel bucket with tap water.

Salt was then added to the water and stirred until the water was supersaturated with salt.

Then ice was added and the water stirred until the ice no longer melted immediately. A

second set of samples were solution heat treated at 11750C (2147F) for 1 h. They were









then slow cooled at 30C/min (5.40F/min) down to temperatures at 50C (90F) intervals

between 11200C (20480F) and 11400C (20840F). These samples were allowed to soak at

that temperature for an hour to allow precipitation to occur. All of these samples were

then removed with tongs at their final temperature and quickly dropped into the

iced-brine quench.

Once the samples were cooled, they were also metallographically prepared by

first mounting in diallyl phthalate and then polishing as previously described. They were

etched with Pratt & Whitney etchant #17 to reveal the y/y' microstructure. The SEM was

used to examine the micrsostructures, so that the temperatures at which all phases go into

solution for this alloy could be determined. These results were then compared with the

DTA data to better understand the microstructural evolution of Alloy 720 during

supersolvus heat treatment.

Microstructural Analysis

When the heat treatment of all of the sample bars was completed, the small

microstructural analysis samples were analyzed to fully characterize the microstructures

of both the alternate and standard heat treatments. Each metallographic sample was

mounted and polished to 0.3 |tm (1.2X105 in) following the procedure listed above. In

order to reveal the y/y' microstructure, these samples were swabbed with etchant #17.

They were then examined on the SEM so the y' morphology could be fully characterized

after both heat treatments. An as-HIP sample and samples that had been heat treated but

not aged were also examined with the SEM and with secondary electron imaging using

the field emission gun (FEG) of an FEI Strata DB235 focused ion beam (FIB).









In addition, an analysis of the y' volume fraction was performed according to the

American Society for Testing and Materials (ASTM) standard E562-95 [66]. At least ten

fields of view from the SEM were used for each sample for this analysis. A circular grid

with 24 intersection points was then overlaid on the printed micrographs. Every

intersection between a grid intersection and a y' particle was counted. The total number

of intersections was divided by the total number of grid points for that sample, giving an

approximate y' volume fraction.

The microstructural analysis samples for grain size determination were polished

down to 0.3 |tm (1.2X105 in) to remove all etching effects. The samples were then

etched with waterless Kallings etchant (5 g CuCl2, 100 mL HC1, and 100 mL

CH3CH2OH) reveal the grain boundaries. The average ASTM grain size was measured

according to ASTM standard El 12-96 [67]. For this analysis, five fields of view were

analyzed for each heat treatment. The Abrams three-circle procedure was used to count

the number of times the circles intercept grains in the micrograph.

The final characterization technique that was utilized was Rockwell C hardness.

The hardness of samples in the as-compacted condition, the as-solutioned condition for

each heat treatment, and in the as-aged condition for each heat treatment were measured

using a Rockwell hardness indenter with the Rockwell C tip. Twelve hardness values

were determined for each sample to determine an average Rockwell C hardness value for

each condition.

Mechanical Testing

Upon completion of the heat treatment, all of the mechanical test bars were

shipped to Crucible Compaction to be machined into test specimens. Crucible









outsourced the machine jobs to three different machine shops. Twelve of the

creep/tensile bars (six from each heat treatment) were sent to Westmoreland Mechanical

Testing & Research Inc. in Youngstown, PA to be machined into tensile samples

according to the schematic in Figure 2-3. The gauge section for the tensile samples was

6.4 mm (0.25 in) in diameter and 31.8 mm (1.25 in) in length. The other twenty-four

samples (twelve from each heat treatment) were sent to Joliet Metallurgical Laboratories,

Inc. in Joliet, IL to be machined into creep samples according to Figure 2-4. The gauge

section on the creep samples was 4.5 mm (0.177 in) in diameter and 26.0 mm (1.025 in)

in length. Pin holes were machined into the shoulder of the creep specimens for attaching

the extensometer frame with screws. The LCF test bars were then machined at Metcut

Research Associates, Inc. in Cincinnati, OH according to Figure 2-5. The LCF test

specimens had a gauge section that was 6.4 mm (0.25 in) in diameter and 19.1 mm (0.75

in) in length. These samples were later sent to Low Stress Grind in Cincinnati, OH to

machine a 0.25 mm (0.01 in) notch was machined onto both shoulders of all LCF

samples for attaching the knife edges of the extensometer frame. All gage surfaces were

low stress ground to reduce or eliminate machining effects. Furthermore, the LCF

samples were polished by hand to eliminate possible crack initiation sites.

Low Cycle Fatigue Testing

The low cycle fatigue testing (LCF) was performed at the University of Florida in

Gainesville, Florida in the High Temperature Alloys Laboratory using an Instron

servo-hydraulic frame (Model TC 25). (Figure 2-6). This frame has an actuator with six

inches of travel and is capable of loads up to 9500 Kg (21000 lbs). The LCF testing was

computer controlled through the Instron Fast Track 8800 computer. The Fast Track


















WESTMORELAND MECHANICAL TESTING & RESEARCH INC.
Youngstown, Pa. 724-537-3131
Inspection Sheet


Sample ID D-1 L G TL/2 R. Cone
UCAI | 137789 0 2506 3.2535 1.2337 0.6949_-- 0.2500 0o0001
UCA4 1143569 0.2502 3.2460 1.2464 0.6938 .- 2500 00002
UCAS 142274 0.2500 3.2455 1.2374 0.6921 0.2500 0.0001
UCA7 137789 0.2500 3.2525 1L2396 0.6992 0.2500 0.0002
UCA8J137816 0.2508 3.2510 1.2424 -0.6916 02500 0.0001
UCA10 114218 02507 3.2515 1.2427 0.6959 0.2500 0.0001
UCB I 138142 0.2494 32455 1.2536 0.6974 0.2500 0.0002
UCB4 1137762 0.2506 .2505 1.2309 0.6992 02500 00001
UCB2 138142 0.2505 3.2410 1.2446 0.7000 0.2500 0.0001
UCB5| 143596 .2501 3.2450 1.2512 0.6935 02500 0.0001
UCB31 138160 0.2507 3.2420 1.2391 0.6940 02500 0.0001
UCB8J 137816 0.2503 3.2470 L- 2383 0.6926 0.2500- 0.0001


Attn: Mr. Sean Conway
Company: Crucible Compaction Metals
Total Samples: 12
WMT&R #: 0-14345

Comment: Threads and finish are acceptable.


Inspector DCW 11 -28-2000 09:47:09


- -- Measuring Device Columns
KNOWINGLYORWILLFULLY FALSIFYNGOR.CONCEALING G T 2 R
I MMTLiN.FU oNR TsFoNN.OOMMAIGALSE. Optical ComparatorX D- G TL/2 R
SnciTousoR FRAUDULENTSITIEMENTSOIR Optical Comparator Y Cone
REPRESETIATIONS FREIN COULD CON UTE 0-6 Inch Caliper L
A FELONY PUNISHABLE UNDER FEDERAL STATRMTES nc aliper L


Figure 2-3: Tensile sample design. All dimensions are in inches.


TeslLog Lot No.
533992 KT55
533993 KT56
533994 KT57
533995 KT58
533996 KT59
533997 KT61
533998 KT90
533999 KT91
534000 KT92_
534001 KT93
534002 KT94
534003 KT95














































CONTROLLED TO A MINIMUM OF .001
AND A MAXIMUM OF .00. LONGITUDINAL
FINAL FINISHING SHOULD BE OBTAINED
WITH AUTOMATIC LAPPING OPERATION IN
IAGE SECTION THE AXIAL DIRECTION (APPLIES TO ITEM
PIN B OIN C POLISH NO. .B.7 AND 8 OMLV).
o0 0 RADIAL S. CENTERS PERMISSIBLE ON ENDS.
1.010 3.060
1.00 3.00 A0AL6. 1001 WRITTEN DIMENSIONAL
1.010 1.360 INSPECT ON HEOUIREO.
* oo0 3.060 LONGITUDINAL Ms uni mrS mEv nai II nitn'isLI
1.010 ,.360 LONGITUDINAL Imicmi D&Iwr I.I Ih *.u a..

.5G 2.900
.510 2.860 RADIAL
.O 2.60 LONGITUGINAL $$10CWG FEN ...it aND .....0AE.VA1.12
.io0 2.560 A______



.540 2.900 IDINAL
.510 2.860 LONGITUDINAL
iiN si NLuSS O itss THIEflISE SPtCIFItD ALL I f' CREEPIRUPTURE
TOLERANCES ANO REOVIREMEITS i
ON 296(o00 (CAG 044 11l
APPl T Td TlHIS 0A diAWIN 's i .. in" -h






Figure 2-4: Creep sample design. All dimensions are in inches.


















- CRIND: SEE r NISH
CAGE OPERATION

010 .001 Z
-.251/.249 0
LCNG.


.500/.498 0 q


1'
Notch


rOi0 .001 Z

1/2-20 THREADS 3A
/ .4675/.4643 PITCH 0
S.008/.010 ROOT RADIUS
.8?O LONG

-

#2 CD L CS
\ .125/.135 0
.L50 MAX, DEPTH


I 45- CHAMFER


Figure 2-5: Low cycle fatigue sample design. All dimensions are in inches.


ME"h



































Figure 2-6: Instron-Satec servo-hydraulic test frame

computer was connected to a traditional PC on which the individual tests were setup and

controlled using the LCF module of Instron's Fast Track Console software package.

Loads/strains were applied to the test specimens through pull/push-rods and threaded

grips. The pull/push-rods were machined by Low Stress Grind in Cincinnati, OH from

material (Udimet 720) cast and wrought by PCC Airfoils in Minerva, OH. The grips

were obtained from Satec and made from Mar-M 247 alloy. The specimens were heated

to the test temperature by a Satec (Model SF-12 2230) clam-shell Power Positioning

Furnace with Kanthal elements. Temperature was controlled by a Eurotherm model 2416

controller using a type K thermocouple. A second thermocouple was used to verify the

test temperature. The two type K thermocouples, with Nextel fiber sheathing, were tied

on to the gauge section of each sample using 26 AWG 80Ni-20Cr bare thermocouple









wire. A high temperature Instron extensometer frame was attached to the sample using

replaceable round knife edge inserts with a 12.7 mm (0.5 in) diameter to allow the

extensometer to measure strain outside of the hot zone of the furnace. The extensometer

frame and the inserts were made out of Haynes 214. An Instron model 2620-826

dynamic extensometer with 12.7 mm (0.5 in) gauge length and +2.54 mm (0.1 in) travel

length was attached to the extensometer frame using straight knife edge attachments and

size 014 neoprene O-rings. Once loaded in the test frame, the specimens were heated to

the test temperature by the furnace and allowed to soak at the test temperature for thirty

minutes before beginning the tests.

The low cycle fatigue test matrix is listed below in Table 2-4. For all test

specimens in this study, the third letter of the sample ID refers to the heat treatment. "A"

is the standard heat treatment, while "B" refers to the alternate heat treatment. All tests

were run with an R (smin/Smax) ratio of 0.0. Therefore, the minimum strain value for every

cycle of every fatigue test was 0.0%. The maximum strain value for each cycle is

equivalent to the value for the entire strain range. Two samples of each heat treatment

were tested at three different temperatures and three different strain ranges. The LCF

testing was performed in strain control through the first 25000 cycles at 0.333 Hz. If the

specimens did not fracture before 25000 cycles, the test was switched to load control at

the maximum and minimum loads during the strain controlled portion of the test. The

load control portion of the test was performed at 10 Hz until fracture or until 1,000,000

cycles when the test was stopped and the sample was considered a run-out.









Table 2-4: The LCF test matrix
Heat Serial Strain
Heat MPB a Sample ID Temp.(C) S
Treatment MPB Number Sample ID Temp.(Range
96SW865 137789 ULA7 538 1.0%
96SW864 135211T ULA9b 538 1.0%
95SW861 133978B ULA1 538 1.1%
95SW860 133870T ULA2 538 1.1%
96SW863 135276B ULA13 538 1.2%
96SW864 135195B ULA14 538 1.2%
95SW861 133468T ULA3 649 1.0%
95SW860 133870B ULA5 649 1.0%
96SW865 134340B ULA10 649 1.1%
96SW864 135203B ULA11 649 1.1%
96SW865 134277B ULA15 649 1.2%
96SW865 134367T ULA16 649 1.2%
96SW863 135294B ULB2b 538 1.0%
96SW865 134367B ULB11 538 1.0%
95SW860 133825B ULB3 538 1.1%
96SW864 135227B ULB4 538 1.1%
96SW864 135211B ULB9 538 1.2%
96SW863 135285B ULB10 538 1.2%
95SW861 133468B ULB5 649 1.0%
96SW864 135227T ULB6 649 1.0%
95SW861 133978 ULB7 649 1.1%
96SW863 135276T ULB8 649 1.1%
96SW864 135195T ULB13 649 1.2%
_96SW864 135203T ULB15 649 1.2%

High Cycle Fatigue Testing

All high cycle fatigue testing (HCF) was performed in the High Temperature

Alloys laboratory at the University of Florida using the same servo-hydraulic frame as for

the LCF testing. The HCF tests were controlled using Instorn's Fast Track Console

software. The samples were loaded using the same push/pull rods and grips as in the

LCF testing. The clam-shell furnace was enclosed around the test specimen and heated

to the test temperature. Two type K thermocouples were attached to the gauge section of

the specimens for temperature control. After a fifteen minute soak at test temperature,

the HCF tests were started. These tests were run in load control at 10 Hz with an R

(Gmin/Gmax) ratio of 0.1. A test run with a stress range of 993 MPa (144 Ksi) was cycled









between a minimum load of 110.3 MPa (16 Ksi) and a maximum load of 1103 MPa (160

Ksi). Samples were tested until failure or until run-out, which was designated as

1,000,000 cycles. The test matrix for the HCF samples is shown in Table 2-5.

Table 2-5: The HCF test matrix
Heat MPB Serial Stress Range
Treatment Number Sample ID Temp.(C) (MPa)
95SW860 133870B ULA5 538 993
96SW864 135203B ULA11 538 1034
96SW865 134367T ULA16 538 1086
95SW861 133468T ULA3 649 993
96SW863 135276B ULA13 649 1034
96SW865 134340B ULA10 649 1086
95SW861 133978 ULB7 538 993
96SW863 135285B ULB10 538 1034
96SW865 134358B ULB17 538 1086
96SW864 135227T ULB6 649 993
96SW864 135195T ULB13 649 1034
_96SW864 135211B ULB9 649 1086

Tensile Testing

The tensile testing was also carried out on the Instron servo-hydraulic test frame

in the High Temperature Materials laboratory of the University of Florida. The Merlin

module of the Fast Track software was used for set up and control of the tensile tests.

The pull-rods and grips for the tensile testing were machined out of IN-713 superalloy by

Satec. Two type K thermocouples, with Nextel sheathing, were attached to the gauge

section of each elevated temperature tensile test as in the LCF testing. The same

clam-shell furnace heated the samples to their test temperature. The samples were given

at least a fifteen minute hold at temperature to allow the temperature inside the furnace to

equilibrate before testing was commenced. The same extensometer frame from the LCF

testing was used with 6.4 mm (0.25 in) diameter knife edge inserts for the elevated

temperature tensile tests. The room temperature tensile tests did not require an

extensometer frame, as the extensometer could be attached directly to the gauge section









of the specimens. An Instron model 2630-110 static extensometer with a 25.4 mm (1.0

in) gauge length and 25.4 mm (1.0 in) of travel was used for measuring the strain for the

tensile tests. This extensometer was attached to the sample with 6.4 mm (0.25 in) steel

clips. The test matrix for the tensile testing is shown below in Table 2-6. All tensile tests

were conducted at a strain rate of 2.5 mm/min (0.1 in/min). The Merlin software

recorded the data and then was used to produce stress/strain curves for each of the tensile

tests. The software also calculated the yield strength, ultimate tensile strength, ultimate

strain, and fracture strength.

Table 2-6: The Tensile test matrix
Heat Serial Sample Temp.(
MPB Temp.(C)
Treatment Number ID
96SW888 137789 UCA1 25
98SW036 143569 UCA4 25
97SW984 142274 UCA5 649
96SW888 137789 UCA7 649
96SW888 137816 UCA8 760
97SW984 142184 UCA10 760
96SW887 138142 UCB1 25
96SW888 137762 UCB4 25
96SW887 138142 UCB2 649
98SW036 143596 UCB5 649
96SW887 138160 UCB3 760
96SW888 137816 UCB8 760

Creep Testing

The creep specimens in this study were tested on a Satec model M3 creep frame as

pictured in Figure 2-7. The High Temperature Alloys Laboratory at the University of

Florida has four of these creep frames for high temperature creep testing. These frames

have a sixteen-to-one ratio on the lever arm between the load pan and the sample. Two

of the frames are outfitted with a Satec (model SF-16 2230) Power Positioning Furnace.

These furnaces have three zones (top, middle, and bottom) that can be controlled

independently to achieve the correct temperature along the entire gauge length of the test

































Figure 2-7: Instron-Satec Creep frame

specimen. These furnaces have Kanthal elements. The other two creep frames have a

Satec (model KSF 2-8-18) Power Positioning Furnace. These furnaces have a single hot

zone with MoSi2 elements that are capable of testing up to temperatures of up to 1500C

(27320F). All of these furnaces are controlled by a standard PC using the NuVision

Mentor software supplied by Satec. The pull-rods and couplings used for the creep

testing are the same as those described in the tensile testing section above. Three type K

thermocouples were attached directly to the gauge section of each sample with

80Ni-20Cr wire. These three thermocouples allowed the computer to independently

control each furnace zone on each frame so that the power output of each zone needed to

maintain the correct temperature along the whole sample could be obtained. A

high-temperature Instron extensometer frame was attached to each sample with screws

into the pin holes on the sample shoulder. This frame and the screws were machined









from Haynes 214. A Satec model 9234K linear variable differential transformer (LVDT)

displacement transducer was attached to the extensometer frame outside of the hot zone

of the furnace for measuring strain on the sample during testing.

In total, twenty creep tests were performed according to the test matrix in Table

2-7. Two samples from each heat treatment were tested at every condition. Each

specimen was given a one hour soak at the test temperature before beginning the test.

The weight was step loaded in order to measure an approximate value for the elastic

modulus for the samples. Time to 0.1%, 0.2%, 0.5%, 1.0%, 2.0%, 5.0% creep and time

to creep rupture were recorded by the Mentor software. In addition, the minimum creep

rate of each creep test was measured.

Table 2-7: The Creep test matrix
Heat MPB Serial Sample Temp.(C) Stress
Treatment Number ID (MPa)
96SW888 137789 UCA12 677 1034
97SW984 142274 UCA16 677 1034
96SW888 137762 UCA9 677 793
97SW984 142184 UCA11 677 793
96SW888 137825 UCA13 704 689
96SW887 138160 UCA18 704 689
95SW860 133834T UCA20 732 483
95SW861 133807T UCA21 732 483
95SW860 133816B UCA19 760 483
95SW861 133807B UCA22 760 483
96SW888 137825 UCB11 677 1034
97SW985 139888 UCB18 677 1034
96SW888 137762 UCB7 677 793
98SW036 143569 UCB10 677 793
teate 98SW036 143596 UCB6 704 689
96SW888 137825 UCB12 704 689
95SW861 133861T UCB19 732 483
95SW861 133798B UCB20 732 483
96SW888 137816 UCB14 760 483
95SW861 133789T UCB23 760 483









Fracture Analysis

After failure, samples from each condition from the fatigue, creep, and tensile

testing were evaluated to examine the fracture features and determine crack initiation site,

type, and propagation path. Before analysis, all samples were cleaned in a Leco

ultrasonic bath in acetone followed by methanol to clean any dirt or dust from the

fracture surface. The samples were first viewed with a simple light optical microscope to

get an overall view of the propagation path. Next the samples were examined on the

SEM to reveal further details of the fracture surface. If the crack initiation site could be

located for the sample, EDS was performed to determine the type of initiation site, e.g.

inclusion, pore, machined crack, etc. Additionally, the microstructure of the crack path

was investigated along the length of the fracture surface to determine what paths of

failure were active during the various stages of fracture, i.e. intergranular, transgranular,

ductile, or brittle. When SEM and optical examination of the fracture surface was

complete, about 6.4 mm (0.25 in) of the specimen tip at fracture was sectioned off using a

Leco (model VC-50) diamond saw. These samples were mounted in diallyl phthalate and

polished and etched to reveal any interactions between secondary cracks and the alloy's

microstructure. The samples were polished down to 0.3 |tm (1.2X105 in) as in the

previous section on metallographic sample preparation. These samples were then etched

with Pratt & Whitney etchant #17 to reveal the microstructure and examined on the SEM.














CHAPTER 3
RESULTS

Microstructure

Development of Heat Treatment

The first step of this study was developing the heat treatment that would result in

the desired y-y' microstructure. This microstructure was to have a regular array of large

dendritic-shaped y', surrounded by tertiary y'. Additionally, the grain size was to be

slightly larger than both the as-HIPed and the standard heat treated microstructures. The

only known variables for producing this microstructure were that the solution

temperature was to be above the y' solvus temperature and that there was to be a

controlled-cool through this solvus temperature.

The y' solvus temperature is reported in literature to be approximately

1150-11550C (2102-2111F) [46]. With this approximate value for the y' solvus

temperature, a set of heat treatment trials were run to develop the alternate, super-solvus,

heat treatment for use in this study. Once the proper heat treatment was developed, the

heat treatments were carried out on machined test bars. As a control variable, the

standard heat treatment consisted of a two hour soak below the y' solvus temperature at

11000C (20120F) followed by a fan-air-cool. Small, microstructural analysis samples

were included for each Master Powder Blend (MPB) with all of the heat treatments.

These samples were then examined to check the heat treatments and to fully characterize

the microstructure.









Heat treatment trials

Small, as-HIPed specimens were given various heat treatments according to the test

matrix in Table 2-3. During these trials, the soak temperature, soak time, cooling rate,

and final temperature before the fan-air-cool were varied to determine the proper heat

treatment specifications. After the alternate heat treatment was developed, samples from

each master powder blend were given both the alternate heat treatment as well as the

standard heat treatment.

The starting point for the alternate heat treatment was a soak at 11750C (21470F).

This temperature was chosen since it is above the reported y' solvus temperature. A soak

time of two hours was originally selected to allow enough time for all of the y' to go fully

into solution. The baseline sample for this study was A2 and had an 11750C (21470F)

soak with a 3.30C/minute (5.90F/min) cooling rate to 10200C (18680F). The first heat

treatment parameter that was examined was the solution temperature. Soak temperatures

of 12000C (21920F) and 12250C (22370F) were examined, in addition to 1175C

(21470F). Micrographs of these heat treatments can be found in Figure 3-1. Samples A3

and Dl were solution heat treated at 1200C (2192F), with A3 cooled at a rate of

7.40C/min (13.30F/min) and Dl at a rate of 2.80C/minute (5.0F/min). Finally, sample

A4 was soaked at 12250C (22370F) and cooled at a rate of 7.20C/min (13.00F/min). All

four of these samples were held at their soak temperature for two hours and cooled to

10200C (18680F) before they were removed from the furnace for the fan air cool. The

higher heat treat temperatures did not result in the desired microstructure regardless of

the cooling rate. From this set of trial samples it was determined that 11750C (21470F)

would be used as the solution heat treatment temperature for the alternate heat treatment.



















A) B)













Figure 3-1: Heat treatment trial samples with various soak temperatures. A) A2
(11750C). B) A3 (12000C). C) Dl (12000C). D) A4 (12250C).

The next heat treatment parameter that was studied was the cooling rate from the

solution temperature through the y' solvus. When examining the previous samples, it was

determined that a cooling rate of approximately 70C/min (12.60F/min) was too fast.

Sample Dl was closer to the desired microstructure than A3. Slower cooling rates were

examined to determine the optimal cooling rate. These ranged from 0.820C/min

(1.50F/min) to 3.30C/min (5.90F/min). These samples are shown in Figure 3-2. Sample

C2 was given a slow cooling rate of 0.820C/min (1.50F/min) down to 10200C (18680F)

and then fan air cooled to room temperature. Sample D2 was cooled at a rate of

2.00C/min (3.70F/min). Finally, A2, as discussed above, was cooled at 3.30C/min

(5.90F/min). All of these samples were soaked at 11750C (21470F) for two



















A) B)













Figure 3-2: Heat treatment trial samples with different cooling rates. A) C2
(0.820C/min). B) D2 (2.00C/min). C) A2 (3.30C/min).

hours before the controlled cooling. They were also all removed from the furnace when

their temperature reached 10200C (18680F). As can be seen from the micrographs, both

D2 and A2 produced the desired microstructure. The cooling rate for the final heat

treatment was chosen as 3.00C/min (5.40F/min) as a compromise between the cooling

rates of these two samples.

The final two parameters examined were the soak time and the temperature at

which the sample was removed from the furnace for the fan air cool. A longer hold at the

solution temperature was examined as well as a higher final temperature. Micrographs

for these heat treatment trials can be found below in Figure 3-3. A2 was given a 2 h hold

and a 3.30C/min (5.90F/min) cooling rate to 10200C (18680F). Al, on the other hand,


























C)







Figure 3-3: Heat treatment trial samples with different soak times and final temperature
before fan-air-cool. A) A2 (2 h soak, 3.30C/min, and 10200C final temp.). B)
Al (4 h soak, 2.80C/min, and 10200C). C) Cl (2 h soak, 2.80C/min, and
10750C).

was held at temperature for 4 h and cooled at a rate of 2.80C/min 5.00F/min) until 1020C

(1868F). Sample Cl was soaked for 2 h and slow cooled at 2.8C/min (5.0F/min) until

a final temperature of 10750C (19670F). The longer hold time had an adverse affect on

the microstructure, producing irregular shaped y' precipitates. Additionally, the higher

final temperature showed signs of dendritic shaped y', but ultimately did not allow

enough time for the larger precipitates to grow. From these results, it was decided that

the final heat treatment would have a 2 h hold and a final temperature of 1020C

(18680F).










The final heat treatment schedules for both the standard and alternate heat

treatments are shown below in Figure 3-4. Once the final alternate heat treatment was

1100*C for 2 hours


10C/min \ FAC



A)RT ,RT

1175C for2 hours
13C/ 0in


10Cn/miin FAC


B) RT RT


Figure 3-4: Heat treatment schedules. A) Standard heat treatment. B) Alternate heat
treatment.

determined, samples of each Master Powder Blend (MPB) for the mechanical testing

were given both the standard and alternate heat treatment as a check that these heat

treatments work for all of them. Representative micrographs of the two heat treatments

for these samples are contained in Figure 3-5. The standard heat treatment produced a

similar microstructure to Figure 3-5a and Figure 3-5b in all of the MPBs. However, the

alternate heat treatment produced the desired, dendritic y' microstructure in only eight of

the eleven MPBs. The other three (96SW865, 97SW984, and 98SW035) had

microstructures similar to those found in Figure 3-6. The secondary y' was irregular in

shape. To counter this problem, the solution heat treatment schedule was lowered to

11650C (21290F). This lower solution temperature worked well and produced the desired

microstructure in these three MPB's (Figure 3-7).


































Figure 3-5: The standard and alternate heat treatment microstructures. A) Standard heat
treatment at 4300X magnification. B) Standard heat treatment at 570X. C)
Alternate heat treatment at 4300X. D) Alternate heat treatment at 570 X.











A)- B)

Figure 3-6: MPB 96SW865 after solution heat treating at 11750C (21470F). A) 4300X
magnification. B) 570X magnification.



















A) B)
Figure 3-7: MPB 96SW865 after solution heat treating at 11650C (21290F). A) 4300X
magnification. B) 570X magnification.

Heat treatment

The standard, subsolvus heat treatment was carried out in two separate batches.

These samples were heated to 11000C (20120F) at 100C/min. They were left at this

temperature for 2 h. Upon completion of the heat treatment, they were removed from the

furnace and given a fan-air-cooling. The alternate, supersolvus heat treatment was

carried out in three batches. The first two batches were ramped to 11750C (21470F) at

100C/min. A two hour soak at this temperature was followed by a controlled cool

(30C/min) through the y' solvus until the final temperature of 10200C (18680F) was

reached. At this temperature, the test bars were removed and given a fan-air-cooling.

The third batch consisted of the following MPB's: 96SW865, 97SW984, and 98SW035.

Due to the differences noted between these three MPB's and the remaining samples

during the heat treatment trials, the solution temperature was lowered to 1165C

(21290F). All of the samples were then given the same aging heat treatment in a total of

four batches. The first step of the age involved heating the samples to 7600C (14000F);

holding them for 8 h, then fan air cooling them back to room temperature. The final step

was to heat the test bars to 6500C (12020F) and hold them for 24 h before fan-air-cooling

them to room temperature. Only one problem was encountered during these heat









treatments. A few of the creep and tensile bars developed a significant oxide scale that

eliminated these bars as possible test specimens. It was determined that these particular

bars had been sectioned so that they were either too close to the stainless steel HIP can or

actually contained part of the can. New samples were shipped from Crucible Compaction

to the University of Florida for completion of the heat treatments. These samples were

tested for HIP can remnants by passing a magnet over them. Any sample that still

contained part of the HIP can would be magnetic, while the ones that only contained

Alloy 720LI would not be magnetic. These samples were heat treated according to the

schedules listed above. When all the heat treatments were completed, the test bars were

sent back to Crucible to be machined into test specimens.

Characterization

After the heat treatments were complete, a full characterization of Alloy 720LI's

response to the heat treatments was carried out. First, the y'solvus temperature was

determined as a comparison to the reported value from literature. Two methods of

determining this temperature were used. The first was through differential thermal

analysis (DTA) and the second was by metallographic inspection of quenched samples.

Additionally, the y-y' microstructure was fully evaluated through SEM, TEM, grain size

measurements, volume fraction measurements, and hardness indents.

Determination of y' Solvus

During the preliminary heat treatments, three of the eight master powder blends

(MPBs) of Alloy 720LI had a different microstructural response to the heat treatment

than the other MPBs. This difference was thought to be related to the y' solvus

temperatures of the different master powder blends as there was no noticeable trend with









chemistry. To verify this phenomenon and to determine the y' solvus temperatures of the

MPBs, samples were sent off for differential thermal analysis.

MPB # 96SW863 and MPB # 96SW865 were chosen as representative samples of

the MPBs thought to have higher y' solvus and lower y' solvus temperatures respectively.

Two samples from each MPB were sectioned and sent for DTA analysis at two

independent laboratories. The purpose of using two laboratories was to ensure the

accuracy of the results. The results obtained from the DTA analysis performed at the

M&P Laboratories are in Table 3-1. The "maximum thermal effect" is the temperature

Table 3-1: The DTA results from M&P Laboratories
96SW863a 96SW863b 96SW865a 96SW865b
Max Thermal
Mxfetl0 1162(2124) 1160(2120) 1158(2116) 1150(2102)
Effect oC (OF)
y' soluvs C (OF) 1192(2178) 1192(2178) 1200(2192) 1193 (2179)
Solidus oC (OF) 1279 (2334) 1279 (2334) 1279 (2334) 1280 (2336)
Liquidus oC (OF) 1352 (2466) 1349 (2460) 1348 (2458) 1349 (2460)

at which the rate of volume change in the y' precipitates is at a maximum during heating

[69]. This was measured at a local minimum that occurred in the DTA curve, while the

y' solvus temperature was determined from a local maximum immediately following the

"maximum thermal effect". The solidus was measured at the change in slope that

occurred immediately before the minimum in the graph. Finally, the liquidus was

measured at this minimum. The results of the samples sent to Dirats Laboratories for

DTA analysis are contained in Table 3-2. The y' solvus temperature was measured at a

local maximum in the derivative of temperature curve. On the other hand, the solidus

and liquidus were measured in the same method as M&P Laboratories. Dirats









Table 3-2: The DTA results from Dirats Laboratories
96SW863c 96SW865c
y' solvus C (F) 1182 (2160) 1185 (2165)
Solidus C (F) 1251 (2284) 1235 (2255)
Liquidus C (F) 1334 (1169) 1333 (2431)

Laboratories did not report a value for the maximum thermal effect of the two samples;

however, from the graphs the value can be estimated as approximately 11650C for sample

96SW863c and 11600C for sample 96SW865c. As can be seen from these two tables,

M&P measured a lower temperature for the maximum thermal effect, but a higher

temperature for the y' solvus temperature. All of the DTA graphs can be found in

Appendix A.

To more accurately determine the y' solvus temperature for Alloy 720LI, the DTA

data above was complimented with a metallographic study of samples quenched from

various temperatures. The first set of these samples was heated to temperatures at 50C

(90F) intervals between 11000C (20120F) and 12000C (21920F). They were allowed to

soak at the temperature for 1 h and then quenched in an iced-brine solution. These

samples were then polished, etched, and examined on the SEM. The important

temperature range for the solutioning of y' was determined to be between 11200C

(20480F) and 11400C (20840F) during this study. Figure 3-8 shows micrographs for

samples heated to these temperatures. At 11400C (20840F), all of the y' was in solution.

The only contrast present after etching was due to preferential attack of the grain

boundaries and some small, discrete carbides. The samples heated to 11300C (20660F)

and 11250C (20570F) exhibited coarse grain boundary y' in addition to very small

amounts of the secondary, intragranular y' that had not completely gone into solution









after 1 h. Finally, the sample heated to 11200C (20480F) had no secondary or primary y'

in solution.











A) B)













Figure 3-8: The y' solvus trials quenched in iced brine after an hour at the solutioning
temperature. A) 11400C (20840F). B) 11300C (20660F). C) 1125C
(20570F). D) 11200C (20480F).

Additionally, a number of samples were heated to 11750C (21470F) and allowed to

soak for 1 h to allow all second phase particles to go into solution. These samples were

then control cooled in the furnace to temperatures at 50C (410F) intervals between

11200C (20480F) and 11400C (20840F). They were allowed to soak at this temperature

for 1 h to allow all precipitation of secondary phases to occur. After 1 h, the samples

were quenched in an iced-brine solution. Figure 3-9 contains micrographs of these

samples. Once again, the sample quenched after soaking at 11400C (20840F) only had

contrast due to preferential etching of the grain boundaries and discrete carbides. No y'



















A) B)













Figure 3-9: The solvus trials that were fully solutioned and then control-cooled. A)
11400C (20840F). B) 11300C (20660F). C) 11250C (20570F). D) 1120C
(20480F).

had precipitated out at this temperature. After cooling to 11300C (20660F), the primary

y' had precipitated along the grain boundaries. The 11250C (20570F) sample had no

additional precipitation. However, the sample that was cooled to 11200C (20480F)

before quenching exhibited secondary, intragranular y' in addition to the primary y'.

Microstructural characterization

Small blocks of each master powder blend were heat treated along with the sample

bars to ensure that the proper microstructure was achieved for every test specimen and to

fully characterize the microstructures that were produced. Representative micrographs of

the two heat treatments after etching with Pratt & Whitney etchant #17 are included in

Figure 3-10. As can be seen, the standard heat treatment produced smaller, more






















Figure 3-10: The y/y' microstructure in heat treated Alloy 720LI. A) The standard heat
treatment. B) The alternate heat treatment.

cuboidal, secondary y' precipitates compared to the larger, dendritic y' produced by the

alternate heat treatment. Both samples contained large primary y' precipitates along the

grain boundaries as seen in the far left of Figure 3-10A and the far right of Figure 3-10B.

The ultrafine, tertiary y' is readily evident in the standard heat treatment samples.

However, a FEG SEM was needed to view these tertiary precipitates in the alternate heat

treatment because of their smaller size and the better resolution capabilities of this

instrument.

The y' volume fraction and the average ASTM grain size were measured as detailed

in Chapter 2. The y' volume fraction was calculated on specimens etched with Pratt &

Whitney etchant #17. This etchant did not reveal the grain boundary morphology as well

as it revealed the y/y' microstructure. As a result, waterless Kallings solution was used to

etch the grain boundaries for the grain size measurements. Figure 3-11 shows

micrographs of both the standard and alternate heat treatments after etching with

waterless Kallings reagent. The standard heat treatment has a smaller and more uniform

grain size distribution, while the alternate heat treatment has larger and more varied grain




















A) B)

Figure 3-11: Grain boundary microstructure of heat treated Alloy 720LI. A) The
standard heat treatment. B) The alternate heat treatment.

size distribution. The data and calculations for the y' volume fraction measurements,

mean y' size, and the ASTM average grain size can be found in Appendix B. Table 3-3

shows the results of all of these calculations. While both heat treatments result in the

same y' volume fraction, the alternate heat treatment has a larger average ASTM grain

size than the standard heat treatment. Additionally, the porosity of the as-HIP material

was measured on a polished and unetched sample to be 0.3 + 0.3% by volume.

Table 3-3: The total y' volume fraction, and ASTM grain size
Heat Treatment volume ASTM Grain
Heat Treatment .
fraction Size
Standard 50.7 11.8
Alternate 49.6 8.2

Hardness tests were conducted on heat treated samples after the solution heat

treatment and after the aging heat treatment. The results of the Rockwell C hardness tests

are contained in Table 3-4 below. While both heat treatments produced an increase in

Table 3-4: Results from Rockwell C hardness tests for both heat treatments
Heat Treatment as-HIP Solution H.T. Aging H.T.
Standard 42.5 0.3 45.8 0.3
Alternate 39.9 0.2 42.9 + 0.2

the hardness of the alloys, the standard heat treatment has a greater hardness than the

alternate heat treatment both before and after the aging heat treatment. The statistical









analysis for the hardness testing can be found in Appendix B. All of these data sets are

statistically unique from each other. The increase in hardness after aging indicates that

both heat treatments successfully produced the ultrafine tertiary y'. The data and

statistical calculations for the hardness testing are also contained in Appendix B.

Mechanical Testing and Fracture Analysis

Low Cycle Fatigue

Numerous problems were encountered when attempting to test samples in LCF.

The first difficulty was with the push/pull rods initially used for the testing. A test was

setup with the same push/pull rods that had been used during the creep and tensile testing.

These were 27.3 cm (10.75 in) long and 1.91 cm (0.75 in) in diameter. The first sample

was loaded, the test conditions were input into the computer, and the furnace was heated

to the test temperature. However, when the test was started, the sample broke on the first

cycle. After examination of the deformed sample, it was obvious that the test had been

run without proper sample alignment (Figure 3-12). One of the push/pull rods and the

extensometer frame were bent beyond repair. This was probably the result of the sample

being off-axis during the compression cycle. It was necessary to design new push/pull

rods specifically for fatigue testing that had a smaller length-to-diameter ratio to

eliminate any potential bending during compression. The new design also contained a lip

that would be flush against the actuator so that any potential deflection due to slack in the

threads would be minimized. These new bars were cast by Special Metals (Figure

3-13a). They were then shipped to Low Stress Grind to be machined into the new

push/pull rods. Figure 3-13b shows the new push/pull rods used for fatigue testing.





























Figure 3-12: Low cycle fatigue sample that was misaligned and tested in compression


............ FIT.


Figure 3-13: Push/pull rods. A) As-cast B) Machined into final form.

After a lengthy setback due to the production of these new push/pull rods, testing

was resumed. The new push/pull rods worked better, but some additional issues still

needed to be resolved. Anytime a sample was loaded, the test would begin properly, but

it was obvious after a few cycles that the maximum and minimum loads were not at the

expected values. Either the extensometer frame was slipping on the sample itself or the

extensometer was slipping on the extensometer frame. In this setup, round knife-edge

inserts attached directly to the polished specimen surface. This worked well for the









initial cycle; however, the knife-edges would slip with each cycle. To counteract this,

two fatigue samples were sent to Low Stress Grind to machine a notch into the shoulder

for the inserts to "hold onto" during the fatigue testing. Additionally, notches were

machined onto the extensometer frame where the extensometer attaches to ensure that the

extensometer itself would not slip. When the new notched specimens were received, they

were loaded into the frame and tested. The system appeared to control the strain well on

these tests, however they both broke at the notch, invalidating the data accumulated from

these tests. Two more samples were then sent to Low Stress Grind to machine a notch

just after the threaded region of the sample. Figure 3-14 shows one of these samples.



I
^ Notch
i-















Figure 3-14: Low cycle fatigue sample with notches machined just after the threaded
region and before the shoulder of the sample

Larger diameter (12.7 mm) inserts were also ordered from Satec for these new samples.

This new set-up appeared to work well on these two samples. At this point, the

remaining samples were shipped to Low Stress Grind to have similar notches machined

onto their shoulders.









A number of tests were run to failure and some to run-out at 1,000,000 cycles. The

computer appeared to be controlling strain well; however, the maximum and minimum

stresses and the amount of plastic strain reported during the tests were different from

what was expected. At this point, the total length of the tested samples was measured to

determine the actual deformation that they had endured. These values were much lower

than those reported by the fatigue software. While the notches had minimized the

slippage, they still had not completely eliminated it. It was decided to test the remaining

samples in high cycle fatigue (HCF) rather than waste any more fatigue specimens. The

LCF fracture surfaces were examined as they could provide some valuable insight into

the effect of the different heat treatments and their resulting microstructure on the crack

initiation type and propagation mode.

Three samples failed for each heat treatment, all at 5380C (10000F). The strain

ranges were supposed to be either 1.0% or 1.1% for all of these tests. However, due to

the complications listed above, these strain ranges were not accurate for the strain

actually seen by the samples. All three standard heat treatment samples that fractured

failed from processing defects. Two of the samples failed from ceramic inclusion

initiation types as seen in Figure 3-15. The third standard heat treatment sample

contained two different types of defects that initiated cracks. These can be seen in Figure

3-16. The first defect was in the middle of the sample. This initiation point had too

much mechanical damage to determine the type of defect. The second initiation (ceramic

inclusion) was located at the bottom of the sample as seen in the overview micrograph. It

was inside the sample, but very near to the surface. This initiation was responsible for

the eventual fast fracture and failure of this sample. It is important to note that faceted




















Figure 3-15: Ceramic inclusion type defect in standard heat treatment sample tested at
5380C (10000F) and a strain range of 1.1%











A)












Figure 3-16: Standard heat treatment sample tested at 5380C (10000F) and a strain range
of 1.0%. A) The overview of the fracture surface. B) Initiation from
unknown type of defect. C) Initiation from a ceramic agglomerate.

grains were found near the various defects for the standard heat treatment. These grains

are examples of slip band cracking; however, they did not function as the crack instigator.









In contrast to the standard heat treatment samples, the alternate heat treatment LCF

fracture surfaces all had faceted grains as the initiation. These cracks all initiated inside

the sample, but near the surface. A representative micrograph of one of these initiation

sites is in Figure 3-17. The arrows in these micrographs indicate the direction of crack

propagation. The grain that served as the initiation point also has a twin boundary that
























C) D)

Figure 3-17: Alternate heat treatment sample tested at 5380F (10000C) and a strain range
of 1.0%. A) 570X magnification. B) 285X. C) 115X. d) 57.5X.

resulted from the deformation to the grain. The lower magnification fractographs show

secondary cracks propagating radially outward from this initiation point. After initiation,

the cracks propagated in a similar manner for both the standard and the alternate heat

treatments. Immediately after the initiation region, there was a flat relatively featureless

region of crack growth as can be seen in Figure 3-18. This region of the crack surface



















A) B)

Figure 3-18: Relatively flat featureless region just after crack initiation in low cycle
fatigue. A) Standard heat treatment. B) Alternate heat treatment.

extended approximately one millimeter from the initiation in all directions. It was

marked by discoloration on the fracture surface due to longer exposure to the oxidizing

atmosphere during testing. At the edge of this discolored region, the surface transitions

from a featureless region to what can best be described as an area of undulating peaks and

valleys. These features persists until either the specimen surface or shear lips. As the

crack moves further from the discolored region, these "hills and valleys" become more

and more pronounced. There appears to be a size difference between the features for the

two heat treatments in this area. Figure 3-19 shows representative micrographs of this

area of the propagation paths.

High Cycle Fatigue

After the initial complications with the LCF testing, high cycle fatigue (HCF)

tests were conducted to determine the effect of the supersolvus heat treatment on fatigue

lifetime and crack initiation type compared to the standard subsolvus heat treatment. A

test matrix was setup to determine the role of microstructure in the fatigue failure of the

remaining samples. Tests were performed with an R ratio (amin/amax) of 0.1 at 538C

(10000F) and 6490C (12000F) and at three different stress ranges: 993MPa (144Ksi),



















A) B)

Figure 3-19: Representative micrographs of the fast fracture region of LCF samples. A)
Standard heat treatment. B) Alternate heat treatment.

1034MPa (150Ksi), and 1086MPa (157.5Ksi). All tests were run at a frequency of 10

Hz. The fracture surfaces were then examined to evaluate the fracture features and to

determine the crack initiation type and the mode of crack propagation.

HCF results

The HCF testing was completed without any of the problems encountered during

LCF testing. All of the tests failed before 1,000,000 cycles. There were three extra

fatigue specimens (two standard heat treatment and one alternate heat treatment)

remaining after all test conditions had been examined. After examining the HCF test

results, these samples were tested at conditions to determine the amount of data scatter in

fatigue properties and when test results indicated a longer or shorter than expected

lifetime. The stress range is plotted versus the number of cycles to failure on a log scale

for both heat treatments at 5380C (10000F) in Figure 3-20. At the low and middle stress

ranges of 993 MPa (144 Ksi) and 1034 MPa (150 Ksi), the standard heat treatment had

superior fatigue lifetimes. However, at the highest stress range of 1086 MPa (157.5 Ksi),

the alternate heat treatment had the longer fatigue lifetime. Similar results were found for

the testing performed at 6490C (12000F). (Figure 3-21) Once again, the standard heat











1100



1080


*Alternate H T
1060
Standard H T


1040



1020



1000



980
10,000 100,000 1,000,000
Nf


Figure 3-20: High cycle fatigue S-N curves for specimens tested at 5380C (10000F)

treatment exhibited the longer lifetimes at the low and middle stress ranges, but not at the

highest stress range. This data would suggest that there is a change in crack initiation or


propagation behavior at higher stresses that increases the fatigue lifetime of the alternate

heat treatment compared to the standard heat treatment. Analysis of the fracture surfaces

would evaluate this phenomenon.

It is also interesting to plot the S-N curves for the two different test temperatures

for each heat treatment. Figure 3-22 contains a plot of the standard heat treatment at both

5380C (10000F) and 6490C (12000F). The lower test temperature exhibited longer

fatigue lifetimes at both the high and the middle stress ranges. However, as the stress

range went below about 1000 MPa, the higher test temperature had the superior lifetime.