Development of a Nanoscale Precipitation-Strengthened Creep-Resistant Aluminum Alloy Containing Trialuminide Precipitates

Transcription

1 NORTHWESTERN UNIVERSITY Development of a Nanoscale Precipitation-Strengthened Creep-Resistant Aluminum Alloy Containing Trialuminide Precipitates A DISSERTATION SUBMITTED TO THE GRADUATE SCHOOL IN PARTIAL FULFILLMENT OF THE REQUIREMENTS for the degree DOCTOR OF PHILOSOPHY Field of Materials Science and Engineering By Keith Edward Knipling EVANSTON, IL December 2006

2 2 Copyright by Keith Edward Knipling 2006 All Rights Reserved

3 3 ABSTRACT Development of a Nanoscale Precipitation-Strengthened Creep-Resistant Aluminum Alloy Containing Trialuminide Precipitates Keith Edward Knipling This research is toward developing a castable and heat-treatable precipitation-strengthened aluminum alloy exhibiting coarsening- and creep resistance at temperatures exceeding 400 C. Criteria for selecting alloying elements capable of producing such an alloy are established. Those systems forming Al 3 M trialuminide compounds with a cubic L1 2 crystal structure are favored, and based on a review of the existing literature, these are assessed in terms of solid-solubility and diffusivity in α-al(satisfying the need for slow coarsening kinetics), and castability (which is discussed based on the binary phase diagrams). The first Group 3 element, Sc, and the second Group 4 element, Zr, are shown to be most promising. These expectations are confirmed by an initial study on the Al-Ti system, which demonstrates that conventionally-solidified alloys are not capable of precipitation strengthening. The Al-Zr system, by contrast, exhibits precipitation of nanometer-scale Al 3 Zr (L1 2 ) producing pronounced precipitation hardening when aged at 375, 400, or 425 C. The Al 3 Zr precipitates are coarsening resistant and have the metastable L1 2 structure up to 500 C, a result of very sluggish diffusion of Zr in α-al. Ternary additions of Ti are also investigated, forming Al 3 (Zr 1 x Ti x ) (L1 2 ) precipitates with a reduced lattice parameter mismatch with α-al, potentially improving the coarsening resistance. The composition of Al 3 (Zr 1 x Ti x ) precipitates formed at 375 or 425 C are measured directly using 3-D atom-probe tomography. At these temperatures, the Zr:Ti atomic ratio in the precipitates is about 10 and 5, respectively, indicating that most of the available Ti fails to partition to the Al 3 (Zr 1 x Ti x ) phase. This is consistent with prior studies on Al-Sc alloys, where the slower-diffusing ternary solute species make up a small fraction of the Al 3 Sc-based precipitates. Despite the confirmed presence of Ti, Al 3 (Zr 1 x Ti x ) precipitates exhibit no improvement in terms of coarsening resistance compared to binary Al 3 Zr. Mechanical properties of the Al-Zr and Al-Zr-Ti alloys are investigated utilizing Vickers microhardness and creep. The alloys deformed by creep at C exhibit a dislocation climb-controlled threshold stress, ca MPa. The binary Al-Zr and ternary Al-Zr-Ti alloys behave similarly under ambient- and high temperature loading, consistent with the similar microstructures of the two alloys.

4 4 ACKNOWLEDGMENTS This research would not have been possible without the guidance of my thesis advisors, Profs. D. C. Dunand, D. N. Seidman, and M. E. Fine. It was an honor to be your student. I thank also Prof. M. Asta (University of California, Davis) and Dr. J. L. Murray (Alocoa) for numerous stimulating conversations. I am most appreciative of my colleagues in both the Dunand and Seidman research groups who were there in the trenches with me. Foremost, I acknowledge Dr. D. Isheim for his insatiable enthusiasm for metallurgy and always making himself available for interesting discussions. I especially thank R. A. Karnesky and M. van Dalen, my colleagues working on aluminum, for putting up with me for four years. I thank my parents for their unending love and support throughout my (lengthy) education. Finally, I am indebted to Alison who was always there for me; you bore the burden of this thesis as much as I. This research was supported by the United States Department of Energy, Basic Sciences Division, under contract DE-FG02-02ER Financial support was also provided by a Walter P. Murphy Fellowship at Northwestern University.

5 5 CONTENTS 1 Criteria for Developing High-Temperature Aluminum Alloys Introduction General strengthening requirements Summary Selection Criteria Capability of forming strengthening intermetallic phases Low solid-solubility in α-al Small diffusivity in α-al Ability to be conventionally cast Discussion Alloying additions with transition elements Alloying additions with lanthanide elements Summary Chapter Summary Peritectic Solidification Introduction Nomenclature Mechanisms of α Formation Peritectic reaction Peritectic transformation Direct solidification of supersaturated α Metastable Properitectic Phases Compositionally metastable properitectic phases Structurally metastable properitectic phases Grain refinement

6 CONTENTS Chapter Summary Mechanisms of α formation Possible solidification sequences as a function of cooling rate Solute redistribution during α-al nucleation Nucleation and Precipitation Strengthening in Dilute Al-Ti and Al-Zr Alloys Introduction Experimental Procedures Alloy compositions and preparation Aging treatments and analytical techniques Experimental Results Optical microscopy Age hardening Electron microscopies Electrical conductivity Discussion Diffusion kinetics Classical nucleation theory Limitations on increasing the solute concentration Comparison with previous studies Common industrial uses of Ti and Zr additions to Al Chapter Summary Atom-Probe Tomographic Studies of Precipitation in Al-0.1Zr-0.1Ti (at.%) Alloys Introduction Experimental Procedures Experimental Results Solute segregation and the precipitated microstructure Three-dimensional atom-probe tomographic reconstructions Quantitative chemical composition of the Al 3 (Zr 1 x Ti x ) precipitates Time-of-flight mass spectra Correlation between measured Ti concentration proximity to precipitates. 99

7 CONTENTS Discussion Local magnification of the Al 3 (Zr 1 x Ti x ) precipitates Compositions of the Al 3 (Zr 1 x Ti x ) precipitates Residual Ti in the α-al solid-solution Comparison to 3-D APT studies on Al-Sc-Ti/Zr alloys Chapter Summary Microstructure of Al-Zr and Al-Zr-Ti Alloys during Aging at 375, 400, or 425 C Introduction Experimental Procedures Experimental Results As-cast microstructure Solute segregation and the precipitated microstructure Age hardening Precipitate morphologies Discussion Comparison to Previous Studies on Al-Zr-V/Ti Alloys Chapter Summary Microstructural Coarsening in Al-Zr and Al-Zr-Ti Alloys, T 450 C Introduction Experimental Procedures Experimental Results: Isothermal Aging at 500 C Vickers microhardness Electron microscopies Experimental Results: Isochronal Aging, C Vickers microhardness and electrical conductivity Transmission electron microscopy Discussion Transformation to the equilibrium D0 23 structure Mechanism of microstructural coarsening Mechanism of strengthening

8 CONTENTS Effect of aging temperature Effect of Ti additions Chapter Summary Creep of Al-Zr and Al-Zr-Ti Alloys Introduction Experimental Procedures Experimental Results Discussion Transformation to the equilibrium D0 23 structure Differences in creep behavior between Al-Zr and Al-Zr-Ti alloys Comparison to Al-Sc alloys Chapter Summary Final Remarks and Suggestions for Future Work Introduction Improvements with Judicious Aging Alloys Containing Ta Alloying Additions to Al-Zr Alloys Al-Zr-Mg Alloys Al-Zr-Sc Alloys References 197 A Alloy Preparation and Castability 225 A.1 Advantages of Arc-Melting A.1.1 High melt temperatures A.1.2 Inert atmosphere A.1.3 Accelerated solidification rates B Homogenization and Other Prior Annealing Treatments 228 B.1 Homogenization (or Solution) Anneals B.2 Pre-Aging at 500 C

9 CONTENTS 9 C Length Scales of Analytical Techniques and Potential Biases 232 C.1 Length Scales of Analytical Techniques C.2 Biases Associated with Sample Preparation D Study on an Al-Zr-Ti Alloy Prepared by Parameswaran et al. 236 E Potential Contamination of Fe and/or Si, and its Effects on Precipitation 240

10 10 LIST OF FIGURES 1.1 The Orowan stress, Eq.(1.1), as a function of mean precipitate radius, R, for various volume fractions, f, of dispersed phase Alloying additions to Al that form thermodynamically stable trialuminide (Al 3 M) intermetallic compounds, with the equilibrium structure indicated The (a) L1 2, (b) D0 22, and (c) D0 23 structures Reported binary phase diagrams for dilute (<1 at.%) additions of the Group 3, 4, and 5 transition elements alloyed with Al Measured activation enthalpies (Q) of 3d transition element solutes and of 4sp non-transition metal solutes in α-al Calculated diffusivities at 300, 400, and 660 C (T m of Al) of 3d transition element solutes and of 4sp non-transition metal solutes in α-al Semi logarithmic plot of diffusivity in α-al versus reciprocal temperature for the elements which form L1 2 trialuminide phases with Al Comparison between eutectic and peritectic reactions Undesirable (a) and desirable (b) characteristics of peritectic systems, as they relate to the potential for development of castable precipitation-strengthened alloys Schematic peritectic phase diagram indicating symbols used for phases, compositions, and temperatures Free energy curves for the equilibrium phases at the Al-rich end of the Al-Ti system above the peritectic temperature T p. From Kerr et al. (1974) Equilibrium Al-Ti phase diagram with metastable extrapolations constructed from microprobe analyses. From Kerr et al. (1974) Effect of solidification rate and temperature of the melt on the morphology of primary Al 3 Zr precipitates. As solidification rate increases, the primary phase becomes more equiaxed. From Brodova et al. (2001) Transmission electron micrographs showing the 3-D morphology of metastable cubic L1 2 Al 3 Ti primary (properitectic) precipitates. From Majumdar et al. (1993)

11 LIST OF FIGURES Relation between the grain sizes and Ti concentration in Al-Ti alloys solidified at various cooling rates. From Ohashi and Ichikawa (1973) Relation between the grain sizes and Zr concentration in Al-Zr alloys solidified at various cooling rates. From Ohashi and Ichikawa (1973) Schematic equilibrium (full lines) and non-equilibrium (broken and dotted lines) peritectic phase diagram Effects of cooling rate and Zr concentration on the solidification microstructures of the Al- Zr alloys investigated by Hori et al.(1981) Solidus-liquidus relationships for hypothetical dilute binary alloys. (a) k 0 < 1 (b) k 0 > Consequences of eutectic (k 0 < 1) and peritectic (k 0 > 1) solidification on the resulting solute distribution in cast Al alloys Macrostructure of as-cast alloys Al-0.18Ti, Al-0.22Ti, and Al-0.19Zr Vickers microhardness vs. aging time at 375 C for the Al-Ti and Al-Zr alloys investigated Vickers microhardness vs. aging time at 425 C for the Al-Ti and Al-Zr alloys investigated SEM micrographs of electropolished TEM foils, showing no evidence of precipitation of Al 3 Ti and copious precipitation of Al 3 Zr (L1 2 ) after aging Al-0.22Ti and Al-0.19Zr (at.%) alloys at 425 C Centered superlattice dark-field TEM micrographs of Al-0.19Zr aged at 425 C for 400 h displaying fine, coherent, homogeneously-distributed Al 3 Zr (L1 2 ) precipitates within the highly-supersaturated dendrites Calculated root-mean-square diffusion distances, 6Dt, for Ti and Zr in α-al at 375 and 425 C Equilibrium Al-rich Al-Ti binary phase diagram Equilibrium Al-rich Al-Zr binary phase diagram Critical cooling rate for suppressing primary Al 3 Ti or Al 3 Zr and achieving supersaturated α-al solid-solution Centered superlattice dark-field TEM micrographs of alloy Al-0.1Zr-0.1Ti(a) aged at 375 C for 1,600 h. A representative nm 3 3-D APT analysis volume is indicated for comparative purposes EDS concentration profiles of Zr and Ti across three solute-rich dendrite arms in as-cast Al-0.1Zr-0.1Ti(b) LEAP reconstruction of an Al 3 (Zr 1 x Ti x ) precipitate in alloy Al-0.1Zr-0.1Ti(b) aged for 400 h at 425 C

12 LIST OF FIGURES Proxigram displaying the distribution of Zr and Ti atoms for the precipitate reconstructed in Figure 4.3 (Al-0.1Zr-0.1Ti(b) aged at 425 C for 400 h). The precipitate is Zr-rich, with a small fraction of the Zr lattice sites replaced by Ti atoms. The ratio of Zr:Ti is about 5, corresponding to Al 3 (Zr 0.83 Ti 0.17 ) Proxigrams displaying the distribution of Zr and Ti atoms in Al 3 (Zr 1 x Ti x ) precipitates formed in alloy Al-0.1Zr-0.1Ti(a) aged at 375 C for 20, 400, and 1,600 h. The ratio of Zr:Ti is about 10, regardless of aging time, corresponding to Al 3 (Zr 0.91 Ti 0.09 ) Atom-probe time-of-flight mass spectrum of Al-0.1Zr-0.1Ti(b) aged for 400 h at 425 C (reconstruction shown in Figure 4.3) LEAP reconstruction of Al-0.1Zr-0.1Ti(a) aged for 1,600 h at 375 C. One large (>25 nm radius) Al 3 (Zr 1 x Ti x ) precipitate is detected Proxigram showing the distribution of Zr and Ti atoms in the Al 3 (Zr 1 x Ti x ) precipitate displayed in Figure 4.7, Al-0.1Zr-0.1Ti(a) aged at 375 C for 1,600 h. The precipitate is slightly super-stoichiometric (25 30 at.% total solute) with very little Ti (ca. 1 at.%) Isothermal sections at 375 or 425 C of a calculated ternary phase diagram for the Al-Zr- Ti system in the Al-rich corner, showing the α-al solid-solution in equilibrium with the metastable Al 3 (Zr 1 x Ti x ) (L1 2 ) and Al 3 Ti (D0 22 ) A 3 nm-thick slice, cut parallel to the analysis direction, through the reconstruction in Figure Macrostructure of as-cast alloys Al-0.1Zr(b), Al-0.1Zr-0.1Ti(b), Al-0.2Zr, and Al-0.2Zr-0.2Ti SEM secondary electron images of as-solidified Al-0.2Zr. Several petal-like primary Al 3 Zr precipitates (light contrast) are observed SEM secondary electron images of as-solidified Al-0.2Zr-0.2Ti SEM micrograph of a primary Al 3 (Zr 1 x Ti x ) precipitate, showing also positions of quantitative EDS measurements EDS concentration profiles of Zr and Ti across three solute-rich dendrite arms in as-cast Al-0.1Zr-0.1Ti(b) and Al-0.2Zr-0.2Ti TEM and SEM micrographs of Al-0.1Zr-0.1Ti(a) aged at 375 C for 1,600 h Vickers microhardness vs. aging time for Al-0.1Zr(a) and Al-0.1Zr(b) at 375, 400, or 425 C Vickers microhardness vs. aging time for Al-0.2Zr at 375, 400, or 425 C Vickers microhardness vs. aging time for Al-0.1Zr-0.1Ti(a) and Al-0.1Zr-0.1Ti(b) at 375, 400, or 425 C Vickers microhardness vs. aging time for Al-0.2Zr-0.2Ti at 375, 400, or 425 C Centered superlattice dark-field TEM micrograph of an interdendritic channel between two dendrite arms in alloy Al-0.2Zr-0.2Ti aged at 425 C for 100 h

13 LIST OF FIGURES Centered superlattice dark-field TEM micrographs of heterogeneously-nucleated spheroidal L1 2 interdendritic precipitates Development of growth instabilities with increasing precipitate radius for interdendritic Al 3 (Zr 1 x Ti x ) precipitates in Al-0.1Zr-0.1Ti(b) aged at 425 C for 400 h SEM micrographs of Al-0.2Zr and TEM micrographs of Al-0.2Zr-0.2Ti, aged at 425 C for 400 h. Gradients in precipitate radius vary considerably throughout the microstructure, even across the same interdendritic channel Complementary dark-field and bright-field TEM micrographs showing loss of coherency in the interdendritic precipitate size gradients TEM micrographs of rod-shaped interdendritic Al 3 Zr or Al 3 (Zr 1 x Ti x ) precipitates, oriented in 100 matrix directions SEM secondary electron images of primary Al 3 (Zr 1 x Ti x ) precipitates in alloy Al-0.2Zr- 0.2Ti aged at 425 C for 400 h, illustrating solid-state precipitation of Al 3 (Zr 1 x Ti x ) (L1 2 ) in supersaturated α-al surrounding the primary precipitates Centered superlattice dark-field TEM micrographs demonstrating solid-state precipitation of spheroidal Al 3 (Zr 1 x Ti x ) (L1 2 ) precipitates in supersaturated α-al solid-solution surrounding a petal-like Al 3 (Zr 1 x Ti x ) (L1 2 ) primary precipitate in Al-0.2Zr aged at 425 C for 400 h Centered superlattice dark-field TEM micrographs of Al-0.1Zr(b) and Al-0.1Zr-0.1Ti(b) aged at 425 C for 1600 h. Within the dendrites, Al 3 (Zr 1 x Ti x ) (L1 2 ) precipitates exhibit no improvement in coarsening resistance as compared to Al 3 Zr(L1 2 ) Comparison of L1 2 -structured Al 3 (Zr 1 x M x ) (M = V and/or Ti) average precipitate radius ( R, in nm) with aging time at 425 C (M = V and/or Ti) Montage of centered superlattice dark-field TEM micrographs of Al-0.2Zr aged at 425 C for 400 h. The apparent mean radius, R, of spheroidal Al 3 Zr (L1 2 ) precipitates is strongly dependent on location in the specimen Vickers microhardness vs. exposure at 500 C for peak-aged Al-0.1Zr(b) and Al-0.1Zr-0.1Ti(b) alloys Montage of bright-field TEM micrographs of Al-0.1Zr-0.1Ti(b) aged at 500 C for 100 h (after aging at 375 C for 100 h) Montage of centered superlattice dark-field TEM micrographs of Al-0.1Zr-0.1Ti(b) aged at 500 C for 100 h (after aging at 375 C for 100 h) SEM micrograph of heterogeneously-nucleated Al 3 Zr (D0 23 ) precipitates on dislocations or small-angle grain boundaries in Al-0.1Zr(b) aged at 500 C for 100 h (after aging at 375 C for 100 h)

14 LIST OF FIGURES Schematic of the three possible orientation relationships with α-al for the disc-like equilibrium D0 23 precipitates Complementary bright-field and dark-field TEM micrographs of heterogeneously-nucleated L1 2 - and D0 23 -structured Al 3 (Zr 1 x Ti x ) precipitates on dislocations in Al-0.1Zr-0.1Ti(b) aged at 500 C for 100 h (after aging at 375 C for 100 h). Many of the L1 2 -structured precipitates exhibit attributed to antiphase boundaries (APBs) along 100 directions, indicating transformation to the equilibrium D0 23 structure Bright-field TEM micrograph of disc-like D0 23 precipitates in three possible orientations and corresponding selected area diffraction patterns (SADPs) of the equilibrium D0 23 phase in Al-0.1Zr-0.1Ti(b) aged at 500 C for 100 h (after aging at 375 C for 100 h).the patterns are indexed in Figure Complementary bright-field and dark-field TEM micrographs of heterogeneously-nucleated D0 23 precipitates on dislocations in Al-0.1Zr-0.1Ti(b) aged at 500 C for 100 h (after aging at 375 C for 100 h) Calculated electron diffraction patterns from the equilibrium tetragonal D0 23 (closed circles, T subscript) and metastable cubic L1 2 (crosses, C subscript) Al 3 Zr phases in three possible orientation relationships with α-al (open circles). The spots not indexed (e.g. 002 T, 006 T ) are kinematically-forbidden Complementary dark-field and bright-field TEM micrographs of dendritic precipitates in Al-0.1Zr(b) and Al-0.1Zr-0.1Ti(b) aged at 500 C for 100 h (after aging at 375 C for 100 h) Centered superlattice dark-field TEM micrographs of Al-0.1Zr-0.1Ti(b) aged at 500 C for 100 h (after aging at 375 C for 100 h). Many of the spheroidal L1 2 -structured precipitates exhibit APBs along 100 directions, indicating a partial transformation to the equilibrium D0 23 structure Vickers microhardness and electrical conductivity during isochronal aging of Al-0.1Zr(c) and Al-0.1Zr-0.1Ti(c) Bright-field and centered superlattice dark-field TEM micrographs of Al-0.1Zr-0.1Ti(c) isochronally aged to 450 C Centered superlattice dark-field TEM micrographs of Al-0.1Zr-0.1Ti(c) isochronally aged to 525 C Complementary bright-field and dark-field TEM micrographs of Al-0.1Zr-0.1Ti(c) isochronally aged to 575 C. Most of the metastable L1 2 precipitates are transformed to the disc-like D0 23 structure Complementary dark-field and bright-field TEM micrographs of dendritic precipitates in Al-0.1Zr(c) and Al-0.1Zr-0.1Ti(c) isochronally aged to 575 C

15 LIST OF FIGURES Schematic of the L1 2 to D0 23 transformation in Al 3 Zr precipitates Schematic of four stages of microstructural coarsening occurring on the nanometer-scale of the precipitates and the micrometer-scale of the dendrites Centered superlattice dark-field TEM micrographs displaying the four stages of microstructural coarsening shown schematically in Figure Vickers microhardness yield stress increment vs. mean precipitate radius R for Al-0.1Zr(c) and Al-0.1Zr-0.1Ti(c) isochronally aged at 450, 525, and 575 C Schematic as-solidified structure and the expected precipitation behavior for a dendriticallysolidified α solid-solution in which k 0 > Al-rich equilibrium binary Al-Zr phase diagram and calculated metastable L1 2 Al 3 Zr and Al 3 (Zr 1 x Ti x ) solvus by Murray. Additions of Ti reduce the solid-solubility of Zr Variation of strain, ɛ, with time for alloy Al-0.1Zr(d) crept at 300 C for four levels of stress Double logarithmic plot of minimum creep rate at 300, 350, or 400 C vs. applied stress, for Al-0.1Zr(d) and Al-0.1Zr-0.1Ti(d) ln( ɛ) vs. ln(σ σ th ) for the data in Figure Series of optical micrographs illustrating the macrostructure of alloy Al-0.1Zr(b). The ingot cross-section is composed entirely of columnar grains (etched using Poultan s reagent) Series of optical micrographs illustrating the macrostructure of alloy Al-0.1Zr-0.1Ti(b). The grain size is more equiaxed and refined than that of Al-0.1Zr(b), Figure 7.4 (etched using Poultan s reagent) Dependence of the lattice parameter of the metastable Al 3 (Zr 1 x Ti x ) (L1 2 ) phase on the stoichiometric parameter x Minimum creep rate at 300 C vs. applied stress, comparing data in Figure 7.2 to Al-Sc, Al- Sc-Zr, and Al-Sc-Ti alloys Minimum creep rate at 350 or 400 C vs. applied stress, comparing data in Figure 7.2 to Al- Sc-Ti alloys Vickers microhardness and electrical conductivity during isochronal aging of Al-0.1Zr, Al- 0.1Sc and Al-0.1Zr-0.1Sc (at.%) Vickers microhardness and electrical conductivity during isochronal aging of Al-0.06Zr, Al- 0.06Sc and Al-0.06Zr-0.06Sc (at.%) A.1 Optical micrograph of a conventionally-solidified hyperperitectic Al-Ti alloy. Numerous plate-like Al 3 Ti primary precipitates are observed

16 LIST OF FIGURES 16 B.1 Observed Vickers microhardness of Al-0.2 at.% Zr subjected to various annealing treatments prior to isothermal aging at 375 C for 100 h B.2 SEM micrographs of a conventionally-solidified Al-2.36Zn-0.26Zr (at.%) alloy following homogenization at 640 C for 650 h C.1 Length scales of the analytical techniques used C.2 SEM micrograph displaying precipitate-rich dendrites around the specimen edge of an electropolished TEM foil C.3 Vickers microhardness value as a function of indent diagonal length for a 200 g load C.4 Bright-field and centered superlattice dark-field TEM micrographs showing preferentially electropolishing around the precipitate-rich dendrites D.1 Variation of spheroidal L1 2 precipitate radius vs. time for Al 3 (Zr 1 x Ti x ) and Al 3 (Zr 1 x V x ) precipitates at 425 C D.2 Vickers microhardness vs. aging time for Al-0.19Zr-0.18Ti (at.%) at 425 C D.3 Centered superlattice dark-field TEM micrograph of Al 3 (Zr 1 x Ti x ) (L1 2 ) precipitates observed after aging at 425 C for 400 h (preceded by a 500 C 1 h pre-age). The precipitates have a mean radius R of 19.2±3.6 nm E.1 SEM secondary and backscattered electron images displaying Fe-rich precipitates decorating the grain boundaries in as-solidified Al-0.2Zr-0.2Ti

17 17 LIST OF TABLES 1.1 Reported lattice parameters and the corresponding mismatch with Al for cubic (L1 2 ) and related tetragonal (D0 22 or D0 23 ) Al 3 M trialuminide intermetallic compounds Equilibrium maximum solid-solubility (C max ) and solubility at 400 C (C 400 ) in binary Al-M alloys that form cubic (L1 2 ) and related tetragonal (D0 22 or D0 23 ) Al 3 M trialuminide intermetallic compounds Measured diffusion data for selected transition metal solutes in α-al Measured diffusion data for selected lanthanide solutes in α-al Invariant reactions in binary Al-M alloys which form cubic (L1 2 ) and related tetragonal (D0 22 or D0 23 ) Al 3 M trialuminide intermetallic compounds Designations and compositions (at.%) of the Al-Ti and Al-Zr alloys investigated Calculation of chemical driving force, F ch, effecting nucleation of Al 3 Ti (L1 2 ) and Al 3 Zr (L1 2 ) Compositions and aging conditions of the Al-0.1Zr-0.1Ti alloys investigated by 3-D APT Number of precipitates intercepted by 3-D APT, total number of atoms collected, number of Ti atoms collected and resulting Ti concentration of the α-al solid-solution Calculated evaporation fields (V nm -1 ) for singly- (E 1 ), doubly- (E 2 ), and triply- (E 3 ) charged Al, Zr, and Ti ions Compositions and aging conditions of the Al-Zr and Al-Zr-Ti alloys investigated Quantitative EDS analysis of the properitectic precipitate of Figure 5.4 in as-cast Al-0.2Zr- 0.2Ti Mean precipitate radii, R, observed in the dendrite centers from six different Al-Zr and Al-Zr-Ti (three of each) specimens after extended aging times at 425 C Reported coarsening rates of L1 2 Al 3 Zr-based precipitates in Al-Zr, Al-Zr-Ti, and Al-Zr-Ti-V alloys aged at 425 C Compositions and aging conditions of the Al-Zr and Al-Zr-Ti alloys investigated

18 LIST OF TABLES Mean precipitate radii, R, of precipitates observed in the dendrite centers of Al-0.1Zr(b) and Al-0.1Zr-0.1Ti(b) aged at 500 C for 100 h (after aging at 375 C for 100 h) Mean precipitate radii, R, observed in the dendrite centers in Al-0.1Zr(c) and Al-0.1Zr- 0.1Ti(c) isochronally aged at 450, 525, and 575 C Mean precipitate radii, R, observed in the dendrite centers after different thermal histories Compositions and aging conditions of the Al-Zr and Al-Zr-Ti alloys investigated by creep Vickers microhardness of the aged Al-Zr and Al-Zr-Ti alloys before and after creep Threshold stress, σ th, of the Al-Zr and Al-Zr-Ti alloys crept at 300, 350, or 400 C

19 19 CHAPTER 1 Criteria for Developing High-Temperature Aluminum Alloys Four criteria are described for the selection of alloying elements capable of producing castable, precipitation-strengthened Al alloys with high-temperature stability and strength: these alloying elements must: (i) be capable of forming a suitable strengthening phase; (ii) exhibit low solid-solubility in α-al; (iii) display a small solute diffusivity in α-al; and (iv) retain the ability for the alloy to be conventionally solidified. With regard to criterion (i), only those systems forming Al 3M trialuminide compounds with a cubic L1 2 crystal structure are considered, which are chemically and structurally analogous to Ni 3Al in the Ni-based superalloys. Eight elements, clustered in the same region of the periodic table, fulfill criterion (i): the first Group 3 transition metal (Sc), the three Group 4 transition metals (Ti, Zr, Hf), and the four latest lanthanide elements (Er, Tm, Yb, Lu). Based on a review of the existing literature, these elements are assessed in terms of criteria (ii) and (iii), which satisfy the need for a dispersion in Al with slow coarsening kinetics, and criteria (iv), which is discussed based on the binary phase diagrams. 1.1 Introduction IMPROVED STRENGTH AT ELEVATED TEMPERATURES has been a continuing goal in aluminum alloy development for more than three decades (e.g., [1 3] or a number of papers in [4]). Aluminum-based alloys have several characteristics that make them especially attractive for the development of high-temperature, high-strength alloys. As with the Ni-based superalloys, the unit cell of Al is face-centered cubic (fcc), whose close-packed structure is more creep resistant than more open crystalline structures. Moreover, Al alloys are naturally oxidation resistant due to an extremely stable passivating protective oxide layer. For weight-sensitive applications, the low density of Al alloys allow for materials with high specific strengths. Finally, Al alloys are considerably more economical than existing high-temperature aerospace (e.g., Ni- and Ti-based) alloys. Historically, most efforts to develop high-strength, thermally-stable Al alloys have sought al-

20 CHAPTER 1 CRITERIA FOR DEVELOPING HIGH-TEMPERATURE ALUMINUM ALLOYS 20 loying elements that exhibit both limited solid-solubility and low diffusivity in α-al. This approach was originally promoted by Adam [2] who argued, based on diffusion-controlled coarsening theory, that dispersed phases formed from such alloying additions would be resistant to Ostwald ripening. Borne out of these ideas are the rapidly-solidified alloys based on the eutectic Al-Fe system that, to date, represent the most promising high-temperature Al-based alloys. These include the well-known Al-Fe-V-Si alloys developed by Skinner et al. [5 7], as well as more complex Al-Fe based systems with ternary and often quaternary additions such as Ce, Ni, Co, Zr, Mo, V [2,3,8 10]. These alloys, however, derive their high-temperature strength from a large volume fraction of stable precipitates that form directly from the melt during rapid solidification. A conventional ingot metallurgy approach to alloy development offers several benefits, both from performance and economic standpoints. Aluminum alloys compacted from powders produced, for example, by rapid solidification processing (RSP) and mechanical alloying (MA) are prone to brittleness, in part due to residual porosity and the presence of oxides from the original powders. Also, the powder production and subsequent compaction stages negatively impact the commercial competitiveness of these alloys. More importantly, solid-state precipitation during post-solidification aging offers the potential for generating a much finer dispersion of strengthening phases than those formed in the melt. This review considers the general requirements and challenges for developing such a castable, precipitation-strengthened, thermally-stable Al-based alloy. The current work is distinguished from other reviews, most notably by Starke and colleagues [11 13], devoted to the intelligent design of high-strength aluminum alloys. This work is concerned less with design, but rather focuses on four broad criteria that a suitable alloying addition to Al must meet. Specifically, alloying additions that are (i) capable of forming a trialuminide strengthening phase, exhibit (ii) low solid-solubility and (iii) low diffusivity in α-al, and (iv) retain the ability for the alloy to be conventionally solidified, are sought General strengthening requirements Physical metallurgy of Ni-based and Al-based alloys Before discussing the challenges of developing high-strength, high-temperature Al alloys, it is valuable to consider what are certainly the most complex and successfully engineered hightemperature alloys developed to date the Ni-based superalloys. Modern Ni-based superalloys

21 SECTION 1.1 INTRODUCTION 21 are able to sustain stresses of the order of 150 MPa for thousands of hours, operating at temperatures ca. 0.75T m. This extraordinary creep resistance is achieved primarily by additions of Al to produce ordered Ni 3 Al precipitates (γ ), with the L1 2 structure, which are both isomorphous and coherent with the fcc Ni-rich matrix (γ). The solubility in Ni of γ -producing Al is substantial, thus allowing for very large volume fractions of precipitated γ which, in many commercial alloys, exceeds 70 vol.%. Development of Al alloys by conventional solidification processing is subject to the restriction that appreciable solubility (> 1 at.%) at equilibrium is limited to eight alloying elements Zn, Ag, Mg, Li, Ga, Ge, Cu, Si (in order of decreasing maximum solubility) which are situated near Al in the periodic table [14 17]. This limitation necessarily restricts the equilibrium volume fraction of any precipitated phase (excluding those for the above elements of which only Ag, Mg, Li, and Cu form stable intermetallic compounds with Al) to a value that is less than 1 vol.%, which is one to two orders of magnitude smaller than typically found in Ni-based superalloys. Precipitation strengthening mechanisms To appreciate the ramifications of the generally very limited solid-solubility of alloying elements in α-al (and the concomitantly low volume fraction of the dispersed phase), it is necessary to consider the quantitative effect of precipitate volume fraction on the predicted strengthening increment in precipitation-strengthened alloys. The strengthening produced by the interaction of dislocations with a dispersion of incoherent, inpenetrable particles within a matrix phase was first described by Orowan [18], and has been reviewed thoroughly (e.g., [19 24]). The shear stress required for a dislocation to loop around a precipitate is inversely proportional to the edge-to-edge distance between precipitates, and the increase in yield strength σ or due to this mechanism is given by [25]: σ or = M 0.4 Gb π (1 ν) ( ) ln 2 R b ; (1.1) λ where M is the Taylor factor for the matrix, G and ν are the shear modulus and Poissons ratio of the matrix, b is the magnitude of the Burgers vector, R is the mean planar precipitate radius (not equal to the mean radius, R ), and λ is an effective inter-precipitate distance, which takes into account the finite size of the precipitates. Both R and λ depend on the distribution of precipitate

22 CHAPTER 1 CRITERIA FOR DEVELOPING HIGH-TEMPERATURE ALUMINUM ALLOYS 22 sizes. For a monodispersed assembly, these parameters are given by [20, 22, 23]: R = π R (1.2) 4 and λ = ( ) 2π 3f 2π R ; (1.3) 4 where f is the precipitate volume fraction. Equations (1.2) and (1.3) are also good approximations for polydispersed arrays [23]. A plot of Orowan stress as a function of precipitate radius for different volume fractions, generated using Eq.(1.1), is displayed in Figure 1.1 with parameters pertinent to Al: M = 3.06 [26], b = nm [27], G = 25.4 GPa [27], and ν = [26]. Because of the intrinsically low precipitate volume fraction in Al-based alloys, it is critical that these dispersed phases be small (of the order of 10 nm or less) and remain small (resist coarsening) throughout thermal exposure during operation. This requirement does not apply to Ni-based superalloys, given their much higher solubility for different alloying elements, thus allowing for higher volume fractions of the ordered strengthening phase, γ. Also, the higher shear modulus of Ni (G = 78.9 GPa [27]) increases the disparity in Orowan stresses between Ni-based and Al-based alloys. The strengthening mechanisms are also somewhat different in coarse-grained Ni- and Albased alloys, as might be anticipated on account of the large disparity in precipitate volume fraction, and in neither system is deformation at elevated temperature governed directly by the Orowan mechanism of dislocation looping. During creep deformation of Ni-based superalloys, dislocations are generally confined to the narrow γ channels between the large γ precipitates, where complex dislocation networks form and inhibit further dislocation motion [28, 29]. In Al-based precipitation-strengthened alloys, sufficient thermal energy is usually available under creep conditions to allow glissile dislocations to circumvent coherent precipitates by climbing out of their glide plane. The increase in length of the dislocation during the climb process results in a threshold stress (which is linearly proportional to the Orowan stress [30 32]), below which creep deformation is not measurable. For coherent precipitates, elastic interactions due to precipitate-matrix modulus and lattice parameter mismatches can further increase the creep threshold stress [33]. These elastic interactions generally result in an optimum precipitate size for creep resistance, whereas dislocation climb predicts that alloys with the smallest precipitates

23 SECTION 1.1 INTRODUCTION 23 Fig. 1.1: The Orowan stress, Eq.(1.1), as a function of mean precipitate radius, R, for various volume fractions, f, of dispersed phase. have the greatest threshold stress (since it is proportional to the Orowan stress), provided precipitates are not sheared. Nevertheless, the comparison in Figure 1.1 shows the critical need for very fine, and thus very coarsening-resistant, precipitates in Al alloys to compensate for the intrinsically limited volume fraction obtained during conventional casting of these alloys. Precipitate stability Ostwald ripening (coarsening) occurs during the latest stages of precipitation and involves the growth of larger precipitates at the expense of smaller ones. The kinetics of this process are controlled by volume diffusion, as solute atoms are transferred through the matrix from the shrinking precipitates to the growing ones, if the evaporation-condensation model obtains. In their classic work on the coarsening of a binary alloy, Lifshitz and Slyozov [34] and Wagner [35] (LSW)

24 CHAPTER 1 CRITERIA FOR DEVELOPING HIGH-TEMPERATURE ALUMINUM ALLOYS 24 showed that the average precipitate size R increases with time t according to: R(t) 3 R(t = 0) 3 = kt; (1.4) where R(t) is the average precipitate radius at time t, R(t = 0) is the average initial precipitate radius at the onset of coarsening, and k is the rate constant given by [36]: k ( Dσ ) 2. (1.5) Ce β Ce α Here, D is the diffusivity of the rate-controlling solute, σ is the precipitate-matrix interfacial free energy, and Ce β and Ce α are the equilibrium solubilities (assuming a planar interface) of the solute species in the precipitate and matrix phases, respectively. For any creep-resistant alloy, it is essential that the dispersion of precipitates resist coarsening during prolonged exposure at elevated service temperatures. As indicated in Figure 1.1, this requirement is especially imperative for Al-based alloys because of the generally limited solubility of most solutes in α-al and the concomitant limited volume fractions (f 0.01) of dispersed phases Summary Based on the behavior of modern Ni-based superalloys, which remain mechanically strong at temperatures exceeding 75% of their absolute melting temperature, it is conceivable that Albased alloys could be analogously developed that would be useful to 425 C (0.75 Tm). The creep resistance of Ni-based superalloys is conferred by very large volume fractions of the precipitated Ni 3 Al ordered phase (γ ), that have the L1 2 structure. An effective high-temperature Al alloy should thus exhibit a similar structural constitution, with suitable alloying additions to Al exhibiting the following qualities: Capability of forming strengthening intermetallic phases. As is true for γ in the Ni-based systems, a high-temperature Al alloy should contain a large volume fraction of a suitable dispersed phase, which must be thermodynamically stable and difficult to shear by dislocations at the intended service temperature. These precipitated phases should also exhibit a similar crystal structure to, and a low lattice parameter mismatch with, the α-al solid-

25 SECTION 1.2 SELECTION CRITERIA 25 solution. Low solid-solubility in α-al. A low equilibrium solid-solubility at the intended service temperature is necessary to retard volume diffusion-controlled coarsening (Eq.(1.5)) and prevent dissolution of the precipitated phases. By the lever rule, limited solid-solubility also maximizes the equilibrium volume fraction of the dispersed phase. Small diffusivity in α-al. Limited diffusivity of the solutes in α-al should also stifle volume diffusion-controlled coarsening (Eq.(1.5)), allowing the precipitates to remain effective barriers to dislocation motion at elevated temperatures. Ability to be conventionally cast. This review is concerned with developing Al alloys via standard ingot metallurgy processing routes, and therefore the alloy must be amenable to conventional casting. In the following sections, potential alloying additions to Al with respect to each of the above four criteria are systematically evaluated. 1.2 Selection Criteria for Castable Precipitation-Strengthened Alloys Capability of forming strengthening intermetallic phases The first criterion stated requires that any suitable system must have the capability to form a fine dispersion of a secondary strengthening phase. As suggested by Fine et al. [1, 37, 38], intermetallic compounds formed with Al are the most promising candidates for strengthening phases in ductile, thermally-stable dispersion-strengthened Al-based alloys. While a number of potential Al-rich intermetallics can be used to strengthen Al [39], trialuminide compounds of the type Al 3 M (where M is an element of the transition metals, lanthanide, or actinide series) have particularly attractive characteristics that include low density (they are nominally 75% Al on an atomic basis), high specific strength, good thermal stability (they have generally very high melting points), and excellent oxidation resistance (again, mostly due to the high Al content). Moreover, the trialuminides are directly analogous, in terms of chemistry, to the γ Ni 3 Al (L1 2 ) ordered precipitates in the Ni-based alloys. Extending the Ni-based alloy analogy further, it is desirable that these dispersed trialuminide precipitates have the cubic L1 2 structure. The similarity in crystal structure between the matrix

26 CHAPTER 1 CRITERIA FOR DEVELOPING HIGH-TEMPERATURE ALUMINUM ALLOYS 26 Thermodynamically Stable Trialuminide Intermetallic Compounds Cubic (L1 2 ) Tetragonal (D0 22 or D0 23 ) Hexagonal (D0 19, D0 18, D0 24 ) or Rhombohedral Orthorhombic (D0 11 ), Monoclinic, or Unknown Bold outline indicates trialuminide is the Al-richest intermetallic phase Fig. 1.2: Alloying additions to Al that form thermodynamically stable trialuminide (Al 3 M) intermetallic compounds, with the equilibrium structure indicated. and precipitated phases allows for a coherent interface between the two phases which, in turn, maximizes the strengthening efficacy of the dispersed phase (e.g., by allowing for elastic interactions between dislocations and misfitting precipitates). Furthermore, coherency minimizes the surface energy per unit area of the heterophase interface, conferring stability at high temperatures by reducing the driving force for precipitate coarsening; i.e., the excess free energy associated with the total interfacial area between the precipitate phase and the matrix. A review of the published phase diagrams and crystallographic data [40 42] indicates that a number of alloying additions crystallize to form stable Al 3 M trialuminides, as shown graphically in the periodic table 1 of Figure 1.2. The high-symmetry cubic L1 2 and related tetragonal D Groups of the periodic table are denoted by the recommended new IUPAC (International Union of Pure and Applied Chemistry) convention where columns are numbered with Arabic numerals from 1 18, corresponding to the number of s, p, and d orbital electrons. This system is less ambiguous than the older, but mutually confusing, schemes i.e., the old IUPAC and CAS (Chemical Abstract Service) designations which label columns with Roman numerals followed by either the letter A or B. Group 3 (new IUPAC), Group III-A (old IUPAC), and Group III-B (CAS)

27 SECTION 1.2 SELECTION CRITERIA 27 and D0 23 structures are prevalent among the early transition elements (Groups 3 5), with other lower-symmetry structures obtained with a few of the transition elements of later groups (Fe, Co, Ni, Re, Ir). Trialuminide intermetallic compounds are even more abundant among the lanthanide (rare earth, RE) elements; with the exception of Eu, all RE elements form thermodynamically stable Al 3 RE compounds. This is also the case for several of the early actinide elements (Th, U, Np, Pu). Other metastable Al 3 M trialuminides also exist, e.g. Al 3 Li, which is a potent strengthening phase in aerospace Al-based alloys, but they are not discussed further since their metastability does not fulfill the criterion of high-temperature stability. They may, however, find use as ternary (or higher order) alloying additions in Al-M systems exhibiting stable Al 3 M trialuminides. Trialuminides formed from the transition elements Of the only seven thermodynamically stable L1 2 trialuminides (Figure 1.2), Al 3 Sc has generated by far the most attention in the scientific literature (see Royset and Ryum [43] for a comprehensive review of the role of Al 3 Sc in Al alloys). Besides Al 3 Sc, no other thermodynamically stable L1 2 trialuminides exist among the transition elements. Just below Sc in the periodic table is Y, which forms an Al 3 Y trialuminide with an equilibrium hexagonal D0 19 (Ni 3 Sn-type) structure. In rapidly-solidified hypereutectic alloys, however, Foley et al. [44] reported the existence of a metastable cubic L1 2 Al 3 Y phase that formed during solidification. Also near to Sc are the Group 4 (Ti, Zr, Hf) and Group 5 (V, Nb, Ta) elements, which crystallize with the body-centered tetragonal D0 22 (or D0 23 for Al 3 Zr) structures. In their monolithic form, these trialuminides have received considerable attention as potential high-strength, hightemperature structural materials, most notably Al 3 Ti since it is the least dense of this class (3.36 g cm 3 ) [45 50]. Unfortunately, however, the low-symmetry tetragonal structure makes these phases intrinsically brittle. The D0 22 and D0 23 structures are, however, closely related to the cubic L1 2 structure (Figure 1.3) and much effort has concentrated on alloying these binary intermetallics to transform them to the higher symmetry L1 2 structure, in the hope that the increased number of independent slip systems will improve toughness. For example, Al 3 Ti (D0 22 ) can be transformed to the cubic L1 2 structure by alloying with late fourth-period transition elements refer to the same column in the periodic table.

28 CHAPTER 1 CRITERIA FOR DEVELOPING HIGH-TEMPERATURE ALUMINUM ALLOYS 28 Fig. 1.3: The (a) L1 2, (b) D0 22, and (c) D0 23 structures. Adapted from Yamaguchi [45]. such as Cr, Mn, Fe, Co, Ni, Cu, or Zn [47, 51 61]. Similarly, Li, Cr, Mn, Fe, Ni, and Cu have been added to Al 3 Zr to increase the stability of the cubic L1 2 structure [62 64], and Cu and Zn have also been shown to stabilize the L1 2 structure of Al 3 Hf [65]. Carlsson and Meschter [66] and Xu and Freeman [67,68] have shown by ab initio calculations that the stability of the D0 22 /D0 23 structure relative to the L1 2 increases rapidly as the transition metal d-electron count increases. Therefore, the likelihood of transforming the stable tetragonal structure to a metastable cubic L1 2 phase is greater for the Group 4 (Ti, Zr, Hf) elements than it is for the Group 5 (V, Nb, Ta). Indeed, while Al 3 Ti and Al 3 Zr have been successfully stabilized into the L1 2 structure by alloying additions of late fourth-period transition elements, similar efforts to produce cubic L1 2 Group 5 trialuminides, such as Al 3 Nb and Al 3 Ta, have proven unsuccessful [47, 69]. Since the free energy difference between the equilibrium and metastable structures is so small, one might expect the L1 2 structure of the Group 4 trialuminides (Al 3 Ti, Al 3 Zr, Al 3 Hf) to be readily achievable when precipitated from solid-solution. Indeed, the decomposition sequence during aging of supersaturated Al-Ti [70 75], Al-Zr [14, 76 84], and Al-Hf [85 93] solidsolutions has been reported to occur firstly by the formation of a metastable cubic L1 2 Al 3 M

29 SECTION 1.2 SELECTION CRITERIA 29 phase, with prolonged exposure (hundreds of hours) at high temperatures (> 450 C) required before this phase transforms to the equilibrium tetragonal Al 3 M structure. Furthermore, single phase L1 2 -structured Group 4 transition metal trialuminides (Al 3 Ti, Al 3 Zr, Al 3 Hf) have been produced through mechanical alloying, which do not transform to their respective equilibrium tetragonal structures until after heating at very high temperatures (485 C, 550 C, and 750 C for Al 3 Ti, Al 3 Zr, and Al 3 Hf, respectively) [94]. The Group 4 cubic L1 2 trialuminides, while thermodynamically metastable, readily precipitate from supersaturated solid-solutions and are kinetically stable at temperatures well in excess of 400 C. These transition elements are also extraordinarily slow diffusers in α-al as discussed below, and therefore show considerable promise as thermally stable secondary phases in precipitation-strengthened, high-temperature Al-based alloys. Trialuminides formed from elements of the lanthanide series There is a monotonic decrease in the radius of the rare earth atoms across the lanthanide period, and this variation in atomic radius has been shown to influence strongly the stability, structure, and composition of the intermetallic compounds formed in the Al-RE systems [95 98]. With decreasing RE atomic radius, the structure of the corresponding trialuminide compound exhibits progressively more cubic character [95 97]. For larger metallic radii (Z = 57 64: La, Ce, Pr, Nd, Pm 2, Sm, or Gd), the hexagonal D0 19 structure (Ni 3 Sn-type) is found. Those elements of intermediate size (Z = 65 67: Tb, Dy, Ho) possess rhombohedral and hexagonal (Ba 3 Pb-, Ni 3 Ti-, and Al 3 Ho-type, respectively) structures in which cubic and hexagonal stacking is mixed. For the smallest radius RE atoms (Z = 68 71: Er, Tm, Yb, Lu) the cubic L1 2 (Cu 3 Au-type) structure is observed. The composition of the terminal Al-rich intermetallic phase is also strongly dependent on the atomic radius of the RE addition. On traversing the lanthanide period from Ce (Z = 58) to Lu (Z = 71), there is a transition in the most Al-rich intermetallic phase from Al 11 RE 3 (also referred to as Al 4 RE by some authors), being stable for early elements of the period, to Al 3 RE which is stable for the late lanthanides. According to the published equilibrium phase diagrams [40 42], Al 3 RE is the terminal intermetallic compound for the smaller lanthanide elements beyond Sm (Z = 62), inclusive. The trialuminide phases of the early, larger, lanthanide series (Ce (Z = 58) 2 The Al-Pm system is not well known, but is assumed to be similar to Al-Nd [40].

30 CHAPTER 1 CRITERIA FOR DEVELOPING HIGH-TEMPERATURE ALUMINUM ALLOYS 30 to Pm (Z = 61)), therefore, are not in equilibrium with α-al and hence may be not precipitated during aging. The point at which this transition in composition of the Al-rich intermetallic phase occurs has been disputed, however [98]. In the Al-Gd system, for example, Al 3 Gd is the equilibrium terminal intermetallic but Savage et al. [98, 99] report the stabilization of Al 4 Gd/Al 11 Gd 3 at high cooling rates during solidification. The existence of a metastable Al 4 Y/Al 11 Y 3 phase has similarly been reported in the as-cast structure of rapidly-solidified Al-Y alloys [100,101], even though Al 3 Y is the equilibrium terminal phase. The transition in trialuminide structure from hexagonal/orthorhombic to cubic is also rather sensitive to small perturbations. Figure 1.2, which displays the equilibrium structures, indicates that thermodynamically stable cubic L1 2 structures are obtained for RE additions of Er (Z = 68) and beyond. Cannon and Hall [96], however, showed that applied pressure produced a structural transformation toward the more cubic structure in the lanthanide trialuminides, and obtained cubic L1 2 structures for Al 3 Ho and Al 3 Dy through such a pressure-induced polymorphic change. Conversely, impurities of Si may stabilize a non-equilibrium, rhombohedral (Al 3 Ho-type) structure for Al 3 Er [40, 102]. Trialuminides formed from elements of the actinide series Consider the trialuminides of the actinide series, of which there are four equilibrium Al 3 M phases (Al 3 Th, Al 3 U, Al 3 Np, Al 3 Pu) as indicated in Figure 1.2. Of particular interest are Al 3 U and Al 3 Np, which form thermodynamically stable cubic L1 2 structures. None of the actinide trialuminides, however, are in terminal equilibrium with with their respective α-al solid-solutions [40, 41], and they therefore cannot be precipitated from α-al during aging. While actinide metals could be considered as ternary elements to modify other Al 3 M phases, their radioactive nature prevents practical engineering applications. Lattice parameters of the cubic and tetragonal trialuminide phases A small lattice parameter mismatch between the precipitate and matrix is essential for minimizing the interfacial free energy driving Ostwald ripening (Eq.(1.5)) and for maintaining a coherent and coplanar heterophase interface between the two phases [1,37,38,84]. It is therefore useful to compare the lattice parameters, a, among the cubic L1 2 trialuminides in Figure 1.2, as displayed

31 SECTION 1.2 SELECTION CRITERIA 31 in Table 1.1. For the Group 4 (Ti, Zr, Hf) and Group 5 (V, Nb, Ta) elements that form metastable cubic L1 2 trialuminides, data for the related equilibrium tetragonal D0 22 and D0 23 structures are also provided. The absolute lattice parameter mismatch, δ, for the cubic L1 2 structures is: δ = a (1.6) a 0 where a 0 = Å is lattice parameter of Al. Taking into account the mismatch along both the a- and c- axes, the parameter δ for the tetragonal phases may be written as [37, 39, 94]: δ = where n = 2 for D0 22 and n = 4 for D0 23. [ 2 1 a + a 0 1 c n a 0 The lattice parameters and corresponding mismatches displayed in Table 1.1 are published values for the pure binary Al 3 M trialuminides, measured at room temperature, which may be modified significantly by alloying or thermal expansion at elevated temperature. The cubic and tetragonal phases in Table 1.1 are isostructural, respectively, and so there is generally extensive mutual solubility between them with a concomitant shift in lattice parameter, which is approximately linear with composition assuming Vegard s law obtains. Fine et al., for example, showed that the lattice parameters of both the stable (D0 23 ) and metastable (L1 2 ) Al 3 Zr phases could be reduced by additions of Ti, Hf, or V [84, 110, 111]. The Al-Zr-V system, in particular, was studied extensively [84, ], and the reduced lattice parameter mismatch was observed to decrease the rate of Ostwald ripening for both metastable cubic L1 2 and the equilibrium tetragonal D0 23 phases [37]. Similar improvements have been obtained in Al-Sc based alloys, where partitioning of Zr to Al 3 (Sc 1 x Zr x ) [ ] and Ti to Al 3 (Sc 1 x Ti x ) [118] has been shown to offer improved stability of the precipitated Al 3 Sc-based trialuminides in dilute alloys compared to binary Al 3 Sc [119]. These and other Group 3 (Y), Group 4 (Ti, Zr, Hf), and Group 5 (V, Nb, Ta) elements reduce the lattice parameter of Al 3 Sc (L1 2 ) [120,121], thus reducing the mismatch with Al. As discussed later, these ternary additions to the Al-Zr [84, ] and Al-Sc [ ] alloys are slower diffusers than either Zr or Sc, thus further improving coarsening resistance. ] (1.7)

32 CHAPTER 1 CRITERIA FOR DEVELOPING HIGH-TEMPERATURE ALUMINUM ALLOYS 32 Table 1.1: Reported lattice parameters and the corresponding mismatch with Al for cubic (L1 2 ) and related tetragonal (D0 22 or D0 23 ) Al 3 M trialuminide intermetallic compounds. Phase Structure Lattice parameters Mismatch with Al Absolute mismatch, δ References (Å) Group 3 transition elements Al 3 Sc L1 2 a = % 1.32% [103] Al 3 Y L1 2 a a = % 4.55% [44] Group 4 transition elements Al 3 Ti L1 2 a a = % 2.04% [94] D0 22 a = % 5.36% [39, 104] c = % Al 3 Zr L1 2 a a = % 0.75% [79, 94, 105] D0 23 a = % 2.89% [39, 104] c = % Al 3 Hf L1 2 a a = % 0.04% [94] D0 22 b a = % 5.98% [106] c = % Group 5 transition elements Al 3 V L1 2 a,c a = % 4.44% [68] D0 22 a = % 5.35% [39, 104, 107] c = % Al 3 Nb L1 2 a a = % 1.49% [39] L1 2 a,c a = % 3.20% [68] D0 22 a = % 5.47% [39, 104] c = % Al 3 Ta d D0 22 a = % 5.26% [104] c = % Lanthanide series (rare earths) Al 3 Er L1 2 a = % 4.08% [ ] Al 3 Tm L1 2 a = % 3.79% [107] Al 3 Yb L1 2 a = % 3.71% [107] Al 3 Lu L1 2 a = % 3.39% [107, 109] Actinide series Al 3 U L1 2 a = % 5.24% [107] Al 3 Np L1 2 a = % 5.20% [107] a Metastable. b Al 3 Hf exists in two different crystallographic forms: a stable high temperature D0 23 phase and a stable low temperature D0 22 phase. The D0 22 structure is the one relevant to solid-state precipitation. c Calculated [68]. d No reported metastable cubic L1 2 Al 3 Ta.

33 SECTION 1.2 SELECTION CRITERIA 33 Summary Trialuminide (Al 3 M-type) intermetallic compounds have many beneficial characteristics including low density, high elastic modulus, high melting points, and are often stable with Al. They are therefore ideal dispersed strengthening phases for high-strength thermally-stable Al-based alloys. The cubic L1 2 -structured trialuminides are especially attractive since these ordered fcc structures are commensurate with Al. While 31 elements form trialuminides when alloyed with Al, as shown in Figure 1.2, only six elements Sc, Er, Tm, Yb, Lu, U, and Np form thermodynamically stable cubic L1 2 Al 3 M structures. Several metastable L1 2 structures exist, most notably among the Group 4 and Group 5 elements. The Group 4 elements (Ti, Zr, Hf) are especially attractive since the degree of metastability of the cubic L1 2 trialuminide is very slight Low solid-solubility in α-al Figure 1.4 shows the published phase diagrams of the Group 3 (Sc, Y, La), Group 4 (Ti, Zr, Hf), and Group 5 (V, Nb, Ta) transition elements alloyed with Al, positioned as they are situated in the periodic table. Many of the arguments employed for considering the capacity for precipitation strengthening (this section) and also castability (section 1.2.4) will be made with reference to features in the equilibrium binary phase diagrams. While the particular features are discussed later, a few patterns in Figure 1.4 are worth commenting on now. The Group 3 (Sc, Y, La) elements exhibit terminal eutectics with Al, while the Group 4 (Ti, Zr, Hf) and Group 5 (V, Nb, Ta) are peritectics. Of the eutectic Group 3 elements, Sc is unique in that it exhibits the highest solid-solubility and the lowest liquid solubility at the eutectic temperature. The other eutectic-forming elements, Y and La (as well as most of the lanthanides alloyed with Al), exhibit rather high liquid solubility of solute with very little solid-solubility, as shown in Figures 1.4(b) and 1.4(c) for Al-Y and Al-La, respectively. These features will have a profound influence on the conduciveness of the various systems for development into castable precipitation-strengthened alloys, as discussed in detail later. Requirements for precipitation strengthening In the interest of precipitation-strengthening, a low solid-solubility is desired to maximize the chemical driving force for nucleation and, concomitantly, the equilibrium volume fraction of

34 CHAPTER 1 CRITERIA FOR DEVELOPING HIGH-TEMPERATURE ALUMINUM ALLOYS (Al) L 0.28 at.% L + Al 3 Sc 660 C 0.23 at.% (Al) + Al 3 Sc L (Al) at.% L + Al 3 Ti C 0.79 at.% (Al) + Al 3 Ti L 0.10 at.% (Al) 0.33 at.% L + Al 23 V 4 L + Al 21 V 2 L + Al 3 V L + Al 45 V 7 (Al) + Al 21 V C 688 C 670 C C Atomic Percent Sc Atomic Percent Ti Atomic Percent V (a) (d) (g) (Al) (Al) + L 0.05 at.% L (Al) + Al 3 Y 639 C L at.% at.% (Al) L + Al 3 Zr (Al) + Al 3 Zr C (Al) L at.% at.% L + Al 3 Nb (Al) + Al 3 Nb C Atomic Percent Y Atomic Percent Zr Atomic Percent Nb (b) (e) (h) (Al) + L 0.01 at.% (Al) L (Al) + Al 11 La C L L + Al 3 Hf at.% C 650 C at.% (Al) (Al) + Al 3 Hf (Al) + Al 3 Hf L L + Al 3 Ta at.% at.% (Al) (Al) + Al 3 Ta 662 C Atomic Percent La Atomic Percent Hf Atomic Percent Ta (c) (f) (i) Fig. 1.4: Reported binary phase diagrams for dilute (<1 at.%) additions of the Group 3, 4, and 5 transition elements alloyed with Al. References for the diagrams are: Al-Sc [42,103]; Al-Y [42]; Al-La [42]; Al-Ti [122]; Al-Zr [123]; Al-Hf [42,106]; Al-V [42]; Al-Nb [42,124]; Al-Ta [42].

35 SECTION 1.2 SELECTION CRITERIA 35 the precipitated phase. Moreover, according to volume diffusion-controlled coarsening theory (Eq.(1.5)), limited solid-solubility is also favorable for retarding Ostwald ripening of the precipitate dispersion, which is essential for creep resistance. As noted previously, appreciable solubility in α-alexists only for eight elements (Zn, Ag, Mg, Li, Ga, Ge, Cu, Si) [14 17]. Solid-solubility is therefore considered primarily from the standpoint of maximizing the potential for precipitation strengthening. The basic requirement for precipitation strengthening, as originally formulated in the seminal work of Merica et al. [125] to explain Wilm s serendipitous discovery of age hardening, is decreasing solute solubility with decreasing temperature. This criterion alone is not very useful since virtually all elements, when alloyed with Al, exhibit this behavior. A more discriminating feature by which to compare various systems, therefore, is to consider the potential for obtaining large volume fractions of dispersed phases which, as discussed in detail by Ryum [126], scales with the maximum solubility, C max. This is exactly true only for eutectic systems as the liquid solubility of solute in peritectic systems is often the limiting factor for determining the amount of solute that may be quenched into solid-solution. This point is discussed in more detail when considering castability. A large maximum solubility, C max, is also essential for solutionizing the alloys in the single phase α-al solid-solution prior to precipitation aging, which is necessary for achieving a homogenous distribution of supersaturated solute atoms after quenching, as well as a potential supersaturation of vacancies capable of accelerating precipitation. In addition to a large C max, limited solubility at intermediate aging/service temperatures, say 400 C (C 400 ), is favored for driving nucleation as well as retarding Ostwald ripening of the precipitated phase. Finally, precipitation of an Al 3 M trialuminide from solid-solution requires that Al 3 M is the most Al-rich intermetallic compound in the system (i.e., Al 3 M exists in equilibrium with the terminal α-al solid-solution). Several of the transition metals (V, Co, La, Ir), the early lanthanides (Ce, Pr, Nd, Pm), and the actinides (Th, U, Np, Pu) all form trialuminides that are not the terminal intermetallic compound, and hence Al 3 M may not be precipitated from solid-solution in these systems. Table 1.2 considers the cubic L1 2 Al 3 M-forming systems in Figure 1.2 with respect to the criteria favorable for precipitation strengthening. As indicated, all trialuminides except for Al 3 V, Al 3 U, and Al 3 Np are in equilibrium with α-al. Unfortunately, few of these elements exhibit ap-

36 CHAPTER 1 CRITERIA FOR DEVELOPING HIGH-TEMPERATURE ALUMINUM ALLOYS 36 Table 1.2: Equilibrium maximum solid-solubility (C max) and solubility at 400 C (C 400 ) in binary Al-M alloys that form cubic (L1 2 ) and related tetragonal (D0 22 or D0 23 ) Al 3 M trialuminide intermetallic compounds. Element C max C 400 α-al-al 3 M References (at.%) (at.%) equilibrium Group 3 transition elements Sc Yes [42, 103] Y Yes [40, 42] Group 4 transition elements Ti Yes [122] Zr Yes [123] Hf Yes [42, 106] Group 5 transition elements V 0.33 < 0.15 No [42] Nb Yes [42] Ta Yes [42] Lanthanide series (rare earths) Er 0 0 Yes [40] Tm 0 0 Yes [40] Yb a 0.18 < 0.1 Yes [42] Lu 0 0 Yes [40] Actinide series U No [40, 42] Np b [40] a Based on the published phase diagram in reference [42]. The original references [127,128] for this data, however, did not measure C max. Furthermore, reference [40] claims there is no significant solubility of Yb in α-al. b No phase diagram available, although presumed to be similar to Al-U [40]. preciable solubility in α-al. Moreover, the solid-solubility of the RE elements, several of which form thermodynamically stable L1 2 trialuminides, is particularly quite low. This may be readily explained by the substantial size difference between Al and RE atoms [95] Small diffusivity in α-al Slow diffusion kinetics are an essential requirement for retention of strength for any alloy subjected to long-term exposure at elevated temperatures. This requirement is especially true for

37 SECTION 1.2 SELECTION CRITERIA 37 any Al-based system due to the intrinsically low volume fraction of dispersed phases and the concomitant necessity for fine precipitates with exceptional coarsening resistance (Figure 1.1). From an experimental point-of-view, it is difficult to carry out diffusion experiments in Al. Cited difficulties include the kinetic barrier for diffusion of a radioactive Al isotope through the passiviting oxide layer, the extremely low solid-solubility of solutes in α-al, and a strong tendency to form intermetallic compounds [129]. As a result, some of the early diffusion data for solutes in α-al are unreliable. The advent of readily available radioactive isotopes after 1950, and more recent developments with ion implantation, have mitigated many of these experimental difficulties. Nevertheless, for several elements (e.g., Ti), measurements of impurity diffusion coefficients are still hampered primarily by the unavailability of suitable and inexpensive radioactive isotopes. In light of these advances in experimental techniques, impurity diffusion in α-al (especially that of the transition elements) has received renewed interest in recent years [ ], and the most reliable data, with particular emphasis on the solutes forming cubic L1 2 Al 3 M intermetallics, are reviewed here. Trends in diffusivity among the transition elements It is fortunate that the transition elements are anomalously slow diffusers in α-al, characterized by large activation enthalpies, large pre-exponential factors, and a wide range of variation of diffusivity values as compared with Al self-diffusion. Measured activation enthalpies for tracer diffusion (Q) and pre-exponential factors (D 0 ) for all of the 3d transition elements and other selected 4d- and 5d transition elements are listed in Table 1.3. The majority of these data was obtained from two review articles by Mehrer et al. [129, 130] and Fujikawa [131], which provide the most authoritative theoretical interpretations for transition metal diffusion in α-al. Data for some of the diffusing elements were also obtained from a recent review by Du et al. [132] and a somewhat older article by Grammatikakis [133]. Also, the comprehensive Landolt-Börnstein review [134] was utilized. Data for 3d transition elements and of 4sp non-transition elements (foreign atoms from the same row of the periodic table) are depicted graphically in Figure 1.5 and 1.6. Figure 1.5 shows the activation enthalpies for tracer diffusion versus the position in the periodic table. Figure 1.6 shows the calculated diffusivities near the melting point of Al (660 C) as well as two other tem-

38 CHAPTER 1 CRITERIA FOR DEVELOPING HIGH-TEMPERATURE ALUMINUM ALLOYS 38 Self-diffusion Table 1.3: Measured diffusion data for selected transition metal solutes in α-al. Pre-exponential, D 0 Activation enthalpy, Q D at 400 C References m 2 s 1 kj mol 1 ev atom 1 m 2 s 1 Original reference Cited in Al Dais et al. [135] [129] Fourth period (3d) transition elements Sc Fujikawa [136] [131] Ti Bergner and van Chi [137] [129, ] V Bergner and van Chi [137] [129, ] Cr Rummel et al. [129] [129, 131] Mn Rummel et al. [129] [129, 131] Fe Rummel et al. [129] [129, 131] Co Rummel et al. [129] [129, 131, 132] Ni Erdélyi et al. [138] [129, 131, 133, 134] Cu Fujikawa and Hirano [139] [129, 131, 134] Zn Peterson and Rothman [140] [129, 131, 133, 134] Fifth period (4d) transition elements Zr Marumo et al. [141] [131, 133, 134, 142] Mo van Chi and Bergner [143] [131, 133, 134] Sixth period (5d) transition elements La Murarka and Agarwala [144] [145] Hf Minamino [131] W van Chi and Bergner [143] peratures of interest (300 and 400 C). The 3d sublevel starts to fill with Sc (Z = 21, [Ar]4s 2 3d 1 ) and becomes filled at Zn (Z = 30, [Ar]4s 2 3d 10 ), and it is evident that valence has a strong influence on both the activation enthalpies and the corresponding diffusivities. This dependence is most obvious in Figure 1.6, where the calculated diffusivity increases with increasing number of d electrons from V to Co by nearly six orders of magnitude at 660 C. It is well known that attractive or repulsive interactions among vacancies and substitutionally dissolved solute atoms may lead to higher or lower diffusivities of solute atoms compared with

39 2 ) SECTION 1.2 SELECTION CRITERIA A c tiv a tio n e n th a lp y (k J m o l -1 ) Q fo r A l s e lf-d iffu s io n A c tiv a tio n e n th a lp y (e V a to m s -1 ) D iffu s iv ity (m A l, C A l, C A l, C C C C S c T i V C r M n F e C o N i C u Z n G a G e S c T i V C r M n F e C o N i C u Z n G a G e Fig. 1.5: Measured activation enthalpies (Q) of 3d transition element solutes and of 4sp non-transition metal solutes in α-al. Data for Ga and Ge are from reference [129]. Fig. 1.6: Calculated diffusivities at 300, 400, and 660 C (T m of Al) of 3d transition element solutes and of 4sp nontransition metal solutes in α-al. Data for Ga and Ge are from reference [129]. the self diffusivity of the solvent atom. This relationship between the vacancy-solute binding free energy and the valence difference between solute and solvent atoms was originally discussed by Lazarus [146] and further elaborated on by LeClaire [147]. While the so-called Lazarus-LeClaire model works well for electropositive impurities in noble metals, it is known (e.g., [148, 149]) that this model is unsatisfactory for several solvents, including Al. More accurate theoretical interpretations have been provided by Hoshino et al. [150] and most recently by Sandberg and Holmestad [151] studying vacancy-solute interactions in α-al based on local-density-functional theory. The calculations indicate that while 4sp (and 5sp) solute atoms are attracted to the vacancy, the interaction for 3d (and 4d) transition elements is repulsive and exhibits a maximum near the middle of the 3d and 4d rows, which is reasonably consistent with the measured activation enthalpies displayed in Figure 1.5. Alexander and Slifkin [152] observed that the row of the diffusing species and hence atomic radius also has an influence on diffusivity; the diffusivities of the Group 11 solutes, for example, increases in the sequence Cu, Ag, Au. Nevertheless, Rummel et al. [129] noted that this influence amounts to at most a factor of about 2, and is therefore minor compared to the strong valency effect. Elements of the same group may be assumed to exhibit similar diffusion kinetics in α-al, as evidenced by the data for the homovalent Group 3 (Ti, Zr, Hf) and Group 6 (Cr, Mo, W) solutes in Table 1.3, exhibiting a factor of about 20 difference for the diffusivity at 400 C.

40 CHAPTER 1 CRITERIA FOR DEVELOPING HIGH-TEMPERATURE ALUMINUM ALLOYS 40 Table 1.4: Measured diffusion data for selected lanthanide solutes in α-al. Pre-exponential, D 0 Activation enthalpy, Q D at 400 C References m 2 s 1 kj mol 1 ev atom 1 m 2 s 1 Original Reference Cited in Ce Murarka and Agarwala [144] [145, 154] Pr Murarka and Agarwala [144] [145, 154] Nd Murarka and Agarwala [144] [145, 154] Sm Murarka and Agarwala [144] [145, 154] Trends in diffusivity among the lanthanide elements Reported diffusion data for the lanthanide or actinide elements in α-al appears to be limited to a study by Murarka and Agarwala [144], which reports diffusion data for the early lanthanide elements (La, Ce, Pr, Nd, and Sm). These data are presented in Table 1.4 (Table 1.3 for La). As indicated, the diffusivities of the lanthanides generally lie between those of the Group 3 (Sc) and Group 4 (Ti, Zr, Hf) elements, as might be expected considering the position of the lanthanides in the periodic table. This intermediate diffusivity of the lanthanides in α-al is substantiated by a decomposition study of melt spun Al-Ti, Al-Zr, and Al-Er alloys by Angers et al. [153]. The coarsening rates of the precipitated Al 3 Er (L1 2 ) phases were more rapid than those for Al 3 Ti (L1 2 ) or Al 3 Zr (L1 2 ), most likely due to faster diffusion kinetics. The activation enthalpy for tracer diffusion is very high compared to the other elements shown in Figure 1.7, and this unexpected experimental result demands that this experiment be repeated using modern analytical methods. Even more useful would be measurements of diffusivities for the other lanthanide elements. Summary The transition elements are anomalously slow diffusers in α-al, with diffusivities several orders of magnitude smaller than that for Al self-diffusion. This anomalous behavior is ascribed to repulsive vacancy-solute interactions which, according to experimental evidence (Figure 1.5), reaches a maximum near the Group 5 (V, Nb, Ta) column of the periodic table. The period of the diffusing species has only a moderate effect (compared to valence) and so homovalent solutes of the same group in the periodic table may be assumed to obey similar diffusion kinetics in α- Al. Diffusion of the lanthanide elements in α-al has received comparatively little attention, but available data for the light lanthanides indicate that diffusivities are relatively small, intermedi-

41 SECTION 1.2 SELECTION CRITERIA A l S c s -1 ) G r o u p 3 S e lf-d iffu s io n D iffu s iv ity (m V N d C e S m G r o u p 5 G r o u p 4 L a n th a n id e s P r L a T i Z r H f /T (K -1 ) Fig. 1.7: Semi logarithmic plot of diffusivity in α-al versus reciprocal temperature for the elements which form L1 2 trialuminide phases with Al. The data for Ce, Pr, Nd, and Sm are assumed to be representative of the L1 2 -forming lanthanides (Er, Tm, Yb, and Lu). ate between those of the Group 3 (Sc) and Group 4 (Ti, Zr, Hf) elements. The diffusion behavior of elements capable of forming precipitated cubic L1 2 trialuminides Al 3 M formed with elements of the Group 3, Group 4, Group 5, and lanthanide series and which are in equilibrium with α-al in Table 1.2 is presented in the Arrhenius diagram of Figure 1.7. The diffusion data for the early lanthanides (Ce, Pr, Nd, and Sm) is assumed to be representative of the late lanthanides (Er, Tm, Yb, Lu), which form Al 3 M (L1 2 ) trialuminides Ability to be conventionally cast This review is concerned with developing precipitation-strengthened Al-based alloys produced via conventional ingot metallurgy routes, which requires that any suitable system be amenable

42 CHAPTER 1 CRITERIA FOR DEVELOPING HIGH-TEMPERATURE ALUMINUM ALLOYS 42 L L L + α L + β T p T m o f p u re s o lv e n t P e rite c tic p o in t T e m p e r a tu r e α L + α α+ β L + β T e E u te c tic p o in t T 0 α α+ β T 0 k 0 = C m a x / C e < 1 k 0 = C m a x / C L p > 1 C m a x C e C L p C m a x C o m p o s itio n C o m p o s itio n (a ) (b ) Fig. 1.8: Comparison between eutectic (a) and peritectic (b) reactions. In the context of the present discussion, α refers to α-al and β refers to Al 3 M. to conventional casting. Castability is considered only in general terms, primarily with respect to features in the reported binary phase diagrams that the various elements form with Al. The transition elements, as well as those of the lanthanide and actinide series, form rather complex binary systems when alloyed with Al, in which one or more intermetallic phases occur [15, 17]. In these systems, eutectic phase equilibria generally exists between the liquid, the α-al terminal solid-solution, and the Al-rich intermetallic phase. The Al-Cu, Al-Ni, and Al-Sc systems are familiar examples. However, α-al solid-solutions are formed via the peritectic reaction between the liquid and the Al-rich intermetallic phase with the Group 4 (Ti, Zr, Hf), Group 5 (V, Nb, Ta) and Group 6 (Cr, Mo, W) transition elements. Details of the reactions are provided in Table 1.5 for the systems considered; i.e., those that form cubic (L1 2 ) or related tetragonal (D0 22 or D0 23 ) Al 3 M trialuminide phases. Peritectic alloys To appreciate the complications associated with the existence of a peritectic reaction in the phase diagram, consider Figure 1.8 which compares the essential differences between eutectic and peritectic systems. These distinctions are enumerated below: 1. The reaction itself. Both eutectic and peritectic reactions represent invariant points (three phases in equilibrium). The eutectic reaction involves decomposition of a single phase

43 SECTION 1.2 SELECTION CRITERIA 43 Table 1.5: Invariant reactions in binary Al-M alloys which form cubic (L1 2 ) and related tetragonal (D0 22 or D0 23 ) Al 3 M trialuminide intermetallic compounds. Reaction Type Reaction Temp. Liquid solubility Solid-solubility k 0 References ( C) C e or C Lp (at.%) C max (at.%) Group 3 transition elements Sc Eutectic [42, 103] Y Eutectic [40, 42] Group 4 transition elements Ti Peritectic [122] Zr Peritectic [123] Hf Peritectic [42, 106] Group 5 transition elements V Peritectic [42] Nb Peritectic [42] Ta Peritectic [42] Lanthanide series (rare earths) Er Eutectic [40] Tm Eutectic [40] Yb Eutectic a a [40, 42] Lu Eutectic [40] Actinide series U Eutectic [40, 42] Np b [40] a Based on the published phase diagram in reference [42]. The original references [127,128] for this data, however, did not measure C max. Furthermore, reference [40] claims there is no significant solubility of Yb in α-al. b No phase diagram available, although presumed to be similar to Al-U [40]. liquid into two different solid phases (L α + β), while the peritectic reaction is the formation of a single solid phase by the reaction of a different solid phase with the liquid (L + β α). 2. Solidification sequence. In a dilute eutectic system the first solid to form is the solute-poor α solid-solution, whereas for a peritectic the first solid to form can be the solute-rich β phase. 3. Reaction temperature. The reaction temperature is less than the melting point of pure solvent in eutectic alloys; for peritectic systems the opposite is true.

44 CHAPTER 1 CRITERIA FOR DEVELOPING HIGH-TEMPERATURE ALUMINUM ALLOYS Liquid-solid partition coefficient. The equilibrium partition coefficient, k 0, for solidification of the α solid-solution is less than unity in a eutectic system, while it is greater than unity in a peritectic system. This parameter dictates the solute distribution in cast alloys and therefore influences precipitation of dispersed phases during post-solidification aging. The key difference and limitation with regard to castability of the Al-based peritectic systems compared to the eutectics in Table 1.5 relates to the solidification sequence. In an alloy of peritectic composition (i.e., with composition exceeding C Lp, Figure 1.8(b)) solidified under equilibrium conditions, the first solid to form is the solute-rich primary (or properitectic) Al 3 M phase. Consequently there is a strong tendency to lose a significant amount of solute to this primary phase when such peritectic alloys are conventionally cast. Moreover, as shown in Figure 1.8, there is a necessary liquidus elevation with solute content in peritectic systems. Thus, in order to melt (completely) a peritectic alloy (i.e., form a single-phase liquid) it is necessary to heat (and hold) the melt above this elevated temperature to dissolve completely the primary Al 3 M phase. This increase in melting temperature is significant in the Al-based peritectic systems (e.g., Al-Ti, Al-Zr, Al-Nb) considering the extraordinarily high melting points of the trialuminide phases (1380 C, 1580 C, and 1680 C for Al 3 Ti, Al 3 Zr, and Al 3 Nb, respectively). For developing a creep-resistant alloy, peritectic systems present an additional challenge due to the potent grain refinement associated with primary Al 3 M precipitation in cast alloys. It is well known [124, ] that minor additions of the Group 4 (Ti, Zr, Hf), Group 5 (V, Nb, Ta), and Group 6 (Cr, Mo, W) elements may be used to refine the grain structure in cast Al alloys. Though the actual mechanism is disputed [156, 157, 159, 166, 168, 169], the marked grain refinement is generally attributed to the presence of primary Al 3 M precipitates, which act as heterogeneous nuclei during solidification of the melt. To avoid rapid diffusional creep associated with a refined grain structure it is necessary to suppress nucleation of the properitectic Al 3 M phase which, for a given cooling rate, is achieved by reducing the solute content of the alloy [ ]. Therefore, the already-limited solid-solubility of most alloying elements in α-al is further reduced by the necessity to avoid properitectic precipitation in the peritectic systems. What characteristics of the equilibrium phase diagram are conducive to properitectic suppression? In alloys produced by RSP, the resistance of peritectic alloys to primary Al 3 M formation has been shown [173, 174] to be related to the undercooling necessary to reach the metastable liquidus of the α-al solid-solution; i.e., the temperature difference, at a composition of inter-

45 SECTION 1.2 SELECTION CRITERIA 45 est, between the β liquidus and extrapolated α-al liquidus. While RSP is outside the scope of the present discussion, it is useful to invoke similar arguments in evaluating peritectic alloys in terms of conventional castability. A minimum undercooling for quenching into the metastable α-al phase field is favored in systems that have a shallow β liquidus and a steep α-al liquidus. In the six Al-based peritectic systems of interest, the peritectic reaction temperatures are only a few degrees above the melting point of pure Al (Figure 1.4, Table 1.5), and so the slopes of the α-al liquidus curves in these systems are shallow. Therefore, the slope of the β liquidus is the primary influence determining the ease of properitectic suppression. A second approach to minimizing this undercooling (without changing the slope of the β liquidus) is achieved by a general increase in the solubility of β in the liquid i.e., a shift to the right of the liquidus curve. Castability in peritectic alloys is therefore dictated primarily by the liquid solubility of solute, whereas the propensity for precipitation strengthening, as discussed previously, is determined by the solid-solubility; high liquid solubilities being favored for castability and low solidsolubilities required for precipitation. These criteria are conveniently expressed in terms of the equilibrium solid-liquid partition coefficient, k 0, which is the ratio of solute composition of the solid and liquid phases in local equilibrium during solidification [ ] 3. Making the usual assumption that both the liquidus and solidus boundaries are straight lines, k 0 is then constant at all temperatures and may be expressed in terms of the solid and liquid compositions at the peritectic temperature: k 0 = C max /C Lp. All peritectic alloys exhibit partition coefficients greater than unity (Figure 1.8(b)), and so maximum liquid solubility and minimum solid-solubility is obtained in systems with partition coefficients approaching unity (i.e., minimum disparity between liquid and solid-solubilities). Figure 1.9 shows undesirable and desirable features of hypothetical peritectic systems based on the preceding arguments. In Figure 1.9(a) it is apparent that large undercoolings are necessary to obviate the primary β phase when casting peritectic alloys of this type. Moreover, on account of the large disparity in liquid and solid-solubilities (k 0 1), it is difficult to achieve a significant supersaturation of solute for post-solidification aging. This reduced driving force 3 The partition coefficient k 0 is usually defined where the compositions are in wt.%. Compositions are reported in at.% since, assuming equal molar volumes between Al 3M and Al (reasonable considering the data in Table 1.1), the supersaturation (in at.%) is a direct measure of the equilibrium volume fraction of precipitated phase. Furthermore, the difference between k 0 defined by compositions in wt.% or at.% is negligible for the dilute concentrations considered presently.

46 CHAPTER 1 CRITERIA FOR DEVELOPING HIGH-TEMPERATURE ALUMINUM ALLOYS 46 S h a llo w β liq u id u s L S te e p β liq u id u s L L + β L + β L + α k 0 ~ 1 T p > > T m T e m p e r a tu r e T m k 0 > > 1 L + α α α+ β T p ~ T m T m α α+ β S te e p α s o lv u s S h a llo w α s o lv u s C L p C m a x C L p C m a x C o m p o s itio n C o m p o s itio n (a ) (b ) Fig. 1.9: Undesirable (a) and desirable (b) characteristics of peritectic systems, as they relate to the potential for development of castable precipitation-strengthened alloys. for precipitation is exacerbated by a relatively steep α-al solvus, i.e., the solid-solubility does not decrease appreciably with temperature. The hypothetical phase diagram in Figure 1.9(b) exhibits features favorable for conventional solidification processing. By virtue of the large liquid solubility of the β phase, a considerable amount of solute may be added before precipitating primary β during solidification. Furthermore, there is a relatively low solubility of solute at the peritectic reaction temperature (k 0 1), which diminishes rapidly with decreasing temperature, providing a maximum supersaturation of solute for precipitation strengthening. It is even conceivable, on account of the large liquid solubility coupled with the very low solid-solubility, that nonperitectic compositions (i.e., with composition less than C Lp [171]) may provide appreciable precipitation strengthening in a system resembling Figure 1.9(b). This is most desirable since the possibility of nucleating properitectic β, which is responsible for most of the complications of casting peritectic alloys (loss of solute during solidification and grain refinement), is eliminated. Finally, the reaction temperature T p in the system of Figure 1.9(b) is much greater than that in Figure 1.9(a). Not only does this tend to maximize the slope of the α-al liquidus (desirable for minimizing the undercooling necessary to suppress properitectic β) but, more importantly, such a feature is favorable in any high-temperature alloy since a high reaction temperature extends the temperature range where the reinforced two-phase (α+β) solid is thermodynamically stable.

47 SECTION 1.3 DISCUSSION 47 This could, in principle, extend the service temperature of the alloy beyond the melting point of pure Al. Eutectic alloys Casting eutectic alloys (of hypoeutectic composition) is considerably less complicated than for peritectic alloys since the first solid to form is the solvent-rich α solid-solution (Figure 1.8(a)). Nevertheless, it is worthwhile commenting on a few features in the eutectic phase diagram as they relate to the potential of a given system for developing high-strength high-temperature alloys. As in any alloy undergoing dendritic solidification, microsegregation of solute is minimized for solid-liquid partition coefficients, k 0, approaching unity. In eutectic alloys, k 0 = C max /C e is less than unity since the α-al liquidus has a greater solubility than the α-al solidus, Figure 1.8(a). The disparity in solute solubility is what drives microsegregation, and so systems in which k 0 is near unity are especially amenable to casting, since the degree of microsegregation is small. For high-temperature applications, a eutectic system poses a potential limitation since, by definition, the reaction temperature is less than the melting point of pure Al (Figure 1.8(a)). In an extreme case, the eutectic reaction temperature could conceivably limit the service temperature of the alloy since its melting point is reduced. Deep Al-M binary eutectic temperatures exist for elements such as Ga and Zn, but not for those forming stable or metastable L1 2 trialuminides (Figure 1.4, Table 1.5). 1.3 Discussion Alloying additions with transition elements As indicated in Table 1.1, transition elements that form cubic L1 2 trialuminides are limited to the early elements near Sc in the periodic table (Ti, Zr, Hf, V, Nb, Y), with Al 3 Sc as the only thermodynamically stable L1 2 structure. While a metastable cubic L1 2 Al 3 Y has been reported [44], this investigation was on highly supersaturated melt spun alloys and precipitation of Al 3 Y (L1 2 ) was observed to commence during solidification. The likelihood of precipitating a similar phase in a conventionally-solidified alloy during post-solidification aging seems unlikely. The Group 4 (Ti, Zr, Hf) and Group 5 (V, Nb, Ta) elements form metastable cubic L1 2 trialu-

48 CHAPTER 1 CRITERIA FOR DEVELOPING HIGH-TEMPERATURE ALUMINUM ALLOYS 48 minides, with the degree of metastability increasing from the former to the latter group (section 1.2.1). Therefore, the Group 5 trialuminides such as Al 3 Nb and Al 3 Ta are poor candidates for cubic L1 2 modification. Indeed, as indicated in Table 1.1, the existence of Al 3 Ta (L1 2 ) is unknown to the authors. This is unfortunate, since both the Al-Nb and Al-Ta peritectic systems (Figures 1.4(h) and 1.4(i), respectively) exhibit favorable characteristics (small k 0 and limited solid-solubility, C 400 ) and, considering they are also Group 5 elements, should be very slow diffusers in α-al (Figure 1.7). Furthermore, unlike the Group 4 (Ti, Zr, Hf) elements, reports of precipitation during aging of Al-Nb and Al-Ta alloys are unknown to the authors. Among the transition elements, therefore, Sc and the neighboring Group 4 (Ti, Zr, Hf) elements seem to offer the most potential for producing a precipitated dispersion of L1 2 precipitates formed during solid-state aging. Stable L1 2 Sc trialuminide precipitates Aside from being the only system forming a thermodynamically stable cubic L1 2 trialuminide, Al-Sc exhibits other unique features that make it attractive for developing conventionally-cast precipitation-strengthened alloys. As displayed in Figure 1.4(a) and Table 1.5, the eutectic Al-Sc system features an exceptionally high reaction temperature (660 C, within 1 C of the melting point of pure Al), with a relatively large solid-solubility (0.23 at.% Sc) and comparatively low liquid solubility (0.28 at.% Sc) at the reaction temperature. This similarity in solid and liquid solubilities leads to an equilibrium partition coefficient, k 0, very near unity (k 0 = 0.82, Table 1.5), which minimizes segregation during solidification and makes this system especially amenable to conventional casting. The extremely high eutectic temperature does not limit the service temperature of Al-Sc alloys for high-temperature applications. In fact, the liquid and solid eutectic solubilities are so close and the reaction temperature so high that the first published phase diagrams for Al-Sc indicated a terminal peritectic reaction like the neighboring Al-Ti, Al-Zr, and Al-Hf systems [43]. The maximum solid-solubility of Sc in α-al is relatively large, exceeded for the elements in Table 1.2 only by Ti and V (and similar to Ta), and it diminishes significantly with decreasing temperature (C 400 = 0.01 at.% Sc, Figure 1.4(a) and Table 1.2), thus maximizing the volume fraction of Al 3 Sc achievable during post-solidification aging. Of the systems in Table 1.2, only Al-Ti and Al- Ta exhibit a comparably high decrease in solid-solubility with temperature. This fortuitous con-

49 SECTION 1.3 DISCUSSION 49 fluence of desirable characteristics a stable L1 2 trialuminide, a solvus boundary conducive to precipitation strengthening, eutectic phase equilibria favoring conventional solidification, and a very high eutectic temperature makes the Al-Sc system particularly well-suited for developing castable, precipitation-strengthened alloys. Additions of Sc, forming coherent nanoscale Al 3 Sc (L1 2 ) precipitates, provides the highest increment of strengthening (at room temperature) per atomic percent of any alloying element added to Al [178]. In addition to the marked precipitation hardening response, these precipitates are also very effective at inhibiting recrystallization and maintaining a fine uniform microstructure in wrought Al alloys [43]. The potent strengthening from precipitated Al 3 Sc (L1 2 ), the high eutectic temperature, and the relatively small diffusivity of Sc in α-al (Figure 1.7) indicate that Al-Sc alloys possess significant potential for developing conventionally-solidified creep-resistant Al-based alloys. Indeed, recent studies by Dunand, Seidman, and colleagues have shown conventionally-solidified Al- Sc alloys exhibiting remarkably high coarsening and creep resistance at 300 C [ ], which may be improved with ternary additions of Mg [ ], Zr [ ], and Ti [118]. The latter two elements, together with many other candidates with diffusivitivies below that of Sc [120], segregate to the Al 3 Sc phase without changing its L1 2 structure. These Sc substitutions also reduce the relatively high cost of Sc additions. Metastable L1 2 trialuminides of the Group 4 transition elements The Group 4 transition elements (Ti, Zr, Hf) are extremely slow diffusers in α-al (Figure 1.7), and therefore offer potentially significant improvements in thermal stability compared to Al-Sc alloys. While the L1 2 trialuminides formed from the Group 4 elements are metastable, these are slow to transform to their respective equilibrium tetragonal structures (section 1.2.1) and therefore seem promising as thermally stable dispersed phases in high-temperature Al-based alloys. The Group 4 elements, however, form peritectics with Al which, as discussed in section 1.2.4, introduces significant complications to produce alloys by conventional ingot metallurgy routes. The Al-Ti system (Figure 1.4(d)) seems especially attractive due to the very sluggish diffusion kinetics of Ti in α-al (Figure 1.7). Moreover, of all the systems in Table 1.2, Ti has the highest maximum solid-solubility in α-al (0.79 at.% Ti), suggesting that relatively high volume fractions of precipitated Al 3 Ti may be obtained. The Al-Ti system has, however, a comparatively limited

50 CHAPTER 1 CRITERIA FOR DEVELOPING HIGH-TEMPERATURE ALUMINUM ALLOYS 50 liquid solubility (k 0 = 10, Table 1.5). Hence, for alloy compositions approaching the maximum solubility there will be a strong tendency for primary Al 3 Ti precipitation during solidification and so appreciable Ti concentrations are unlikely to remain in solid-solution after conventional solidification. Moreover, the solid-solubility of Ti is still rather large at appropriate aging temperatures (C 400 = 0.13 at.% Ti), limiting both the chemical driving force for nucleation and the volume fraction of the dispersed phase (if formed). As reviewed in Chapter 3, virtually all precipitation studies for the Al-Ti system have investigated alloys prepared by nonequilibrium means: RSP and MA, techniques that circumvent the difficulties encountered during conventional solidification of Al-Ti alloys. In the Al-Hf system (Figure 1.4(f)) there is a relatively small disparity between the liquid and solid-solubilities at the peritectic temperature (k 0 = 2.4, Table 1.5), implying that alloys with compositions corresponding to the maximum solubility (0.186 at.% Hf, Table 1.2) may be obtained with conventional casting techniques. This solid-solubility, however, diminishes only slightly with decreasing temperature (C 400 = at.% Hf, Figure 1.4(f) and Table 1.2), which limits the potential for precipitation strengthening. Indeed, in all of the precipitation studies on Al-Hf alloys cited previously [85 93], the alloys were generally highly supersaturated and produced by relatively rapid (nonequilibrium) chill casting methods. Ryum [85], for example, investigated the decomposition of supersaturated 0.27 at.% Hf solid-solutions produced by chill casting. Although the cooling rate was not reported, it is reasonable to believe it was about C s -1, which was the solidification rate reported by Hori et al. [86, 87, 91, 92] on investigations of chill cast alloys containing up to 0.79 at.% Hf. It was reported [86 88] that the solid-solubility of Hf in α-al could be extended to approximately 0.5 at.% Hf at this cooling rate, and such high values of supersaturations, well in excess of the maximum solid-solubility, were required to effect continuous precipitation of Al 3 Hf (L1 2 ) [88], with a pronounced precipitation hardening response in alloys aged at 350 to 450 C [86, 87]. Other studies investigated even more supersaturated alloys (up to 1.0 at.% Hf) obtained by rapid solidification at a rate of approximately 10 7 C s -1 [88,90,93]. Unlike Ti and Hf, Zr exhibits negligible solubility in α-al at temperatures of interest for postsolidification aging (C 400 < at.% Zr, Figure 1.4(e) and Table 1.2). Consequently, even very dilute alloying additions may produce an appreciable precipitation hardening response as shown by Hori et al. [185, 186], who reported precipitation hardening in alloys containing 0.07 at.% Zr aged at 350 to 450 C. Ichikawa and Ohashi [187] similarly observed a significant age hardening

51 SECTION 1.3 DISCUSSION 51 response when dilute chill cast alloys, containing as little as 0.12 at.% Zr, were aged at 300 to 500 C. Furthermore, Nes [79] has reported precipitation of Al 3 Zr (L1 2 ) in very dilute (hypoperitectic, 0.05 at.% Zr) alloys. Moreover, the role of dilute additions of Zr to commercial wrought alloys as recrystallization inhibitors is well known [78, ], where fine (20 30 nm) dispersions of coherent Al 3 Zr (L1 2 ) precipitates are used to pin grain and subgrain boundaries during annealing. The Zr concentration required to inhibit recrystallization in conventionally cast commercial alloys is very low, typically at.% Zr [191] Alloying additions with lanthanide elements There would seem to exist significant potential for the use of late rare earth elements (Er, Tm, Yb, Lu) in Al, as they form thermodynamically stable L1 2 trialuminides (Figure 1.2), are likely slower diffusers than Sc (Figure 1.7), and, unlike the Group 4 elements, exhibit the preferred eutectic phase equilibrium (Table 1.5), which is favorable for conventional ingot metallurgy processing. Indeed, a number of Al-RE systems Al-La [100, ], Al-Ce [195, 196], Al- Nd [195,197,198], Al-Gd [98,99,198,199], Al-Sm [99], and Al-Er [99,153,192,198, ], in particular have received considerable attention for development into dispersion-strengthened alloys for elevated temperature applications. All of these studies, however, investigated highlysupersaturated (sometimes hypereutectic) alloys prepared by RSP. The attraction to the RE elements stems from their generally large solubility in the liquid state, very limited solid-solubility, and small solid state diffusivity in α-al. The large liquid solubility is conducive to solid-solubility extension by RSP, while the very limited solid-solubility and small diffusivity retards volume diffusion-controlled coarsening of the dispersed intermetallic phases during thermal exposure. As in the Al-Fe based alloys described previously, the strengthening dispersions in these alloys form during solidification, either as lamellar eutectic constituents, as a primary phase in hypereutectic alloys, or are precipitated in the solid state on cooling from solidification. The microstructures in these rapidly-solidified alloys, as monitored by microhardness, are generally stable beyond 300 C [98,192,193, ,199,201,202], as might be anticipated due to the small diffusivity and solid-solubility of the RE elements in α-al. Since the strengthening dispersions form on solidification, precipitation hardening during post solidification aging is generally not observed in these alloys. This lack of hardening is a

52 CHAPTER 1 CRITERIA FOR DEVELOPING HIGH-TEMPERATURE ALUMINUM ALLOYS 52 consequence of the very limited equilibrium solubility of the RE elements in α-al (Table 1.2), which assures almost complete precipitation of solute on cooling. Even in highly supersaturated Al-La (0.8 at.% La [192, 193]) and Al-Er (0.7 at.% Er [192, 202]) alloys produced by RSP, the amount of La and Er retained in solid-solution was negligible as indicated by lattice parameter measurements using x-ray diffraction (XRD) of the as-solidified α-al solid-solution. Solid-state precipitation of Al 3 Er (L1 2 ) has, however, been reported by Angers et al. [153] in extremely supersaturated (hypereutectic) rapidly-solidified Al-Er alloys containing 6.25 at.% Er. Furthermore, a weak indication of additional precipitation hardening (beyond the strengthening provided by the as-solidifed dispersions) was shown in other studies on Al-Er alloys [201], as well as Al-Gd alloys [98, 199], but the origin of this hardness increase was not reported. More support for Al-Er as a promising system capable of precipitation strengthening may be provided by Nie et al. [ ], who indicated that coherent nanoscale Al 3 Er (L1 2 ) dispersions can be potent recrystallization inhibitors in Al alloys, comparable in effect to the more commonly-used Al 3 Zr and Al 3 Sc phases in commercial alloys. Although their experimental techniques were not explicit about this point, micrographs indicate that coherent Al 3 Er precipitates were not precipitated during aging but rather formed during cooling after solidification, similar to what Foley et al. [44] reported for metastable Al 3 Y (L1 2 ) precipitation described previously. While the Al-RE systems show considerable promise for development by RSP, their conduciveness to conventional solidification is limited. Indeed, the very properties that make the Al-RE systems attractive for RSP, namely a very high liquid solubility and low solid-solubility (in other words k 0 1), limits their potential when these alloys are produced by conventional casting techniques. Because of the very limited solid-solubility, a supersaturation of solute is impossible to achieve under moderate cooling rates since there will be a strong driving force for precipitation during cooling, as shown in studies on Al-Er [203, 204] and Al-Y [44]. A partition coefficient deviating far from unity also predicts that these alloys will be prone to significant solute segregation during solidification, as substantiated by Ruder and Eliezer [193, 202] comparing the microstructures of Al-La and Al-Er solidified conventionally and with RSP. Moreover, due to the very limited maximum solid-solubility of the RE elements in α-al (Table 1.2), post-solidification homogenization is not possible.

53 SECTION 1.4 CHAPTER SUMMARY Summary It is unfortunate that the slowest diffusers in α-al i.e., the Groups 4 6 transition elements are also the only elements in the periodic table forming terminal peritectics with Al (all other alloying additions exhibit eutectic or monotectic phase equilibria) [15]. The existence of a peritectic reaction in the phase diagram reduces significantly the conduciveness of these systems to conventional casting. It is equally unfortunate that all of the late lanthanide elements, with the possible exception of Yb (see notes in Table. 1.2 and 1.5), exhibit negligible solid-solubility in α- Al. This severely limits the potential for precipitation strengthening since virtually all available solute is precipitated out of solid-solution during post-solidification cooling. The Group 3 element, Sc, exhibits a unique eutectic phase equilibrium with Al, which makes it particularly conducive to developing conventionally-solidified precipitation strengthened alloys. Moreover, Al 3 Sc is a thermodynamically stable L1 2 trialuminide, which exists in two-phase equilibrium with Al to the high eutectic temperature of 660 C. Unfortunately, the very high cost of Sc limits its application. Moreover, compared to the other transition elements, Sc is only a moderately slow diffuser in α-al (Figure 1.7). Of the slower-diffusing peritectic-forming elements that form L1 2 trialuminides (Groups 4 and 5), the Group 4 (Ti, Zr, Hf) are strongly favored since the L1 2 structure in these systems is only slightly metastable. Of the Group 4 elements, only Zr exhibits negligible solubility at suitable aging temperatures (Figure 1.4), both maximizing the volume fraction of the precipitated Al 3 Zr phase as well as improving its resistance to coarsening. 1.4 Chapter Summary This review has summarized basic criteria required for a conventionally-solidified precipitationstrengthened Al-based alloy for high-temperature applications: (i) Solid-state precipitation upon aging of coherent trialuminide Al 3 M with the L1 2 crystal structure, to achieve high strengthening and low coarsening; (ii) Shallow α-al solvus curve, to maximize the volume fraction of precipitated Al 3 M (and thus increase strengthening); and the concomitant low solid-solubility at the aging temperature to minimize coarsening;

54 CHAPTER 1 CRITERIA FOR DEVELOPING HIGH-TEMPERATURE ALUMINUM ALLOYS 54 (iii) Small diffusivity of M in α-al, to minimize Al 3 M coarsening and the associated loss of strength; (iv) Solid-liquid partition coefficient (k 0 ) near unity, to minimize segregation and accommodate conventional solidification. For peritectic systems, a shallow Al 3 M liquidus boundary is desirable for minimizing the casting temperature and suppressing Al 3 M primary precipitation during solidification. Criterion (i) narrows the potential candidates to only eight elements, all situated near Sc in the periodic table: the first Group 3 element (Sc), the three Group 4 elements (Ti, Zr, Hf) and the four heaviest rare-earth elements (Er, Tm, Yb, Lu). Among these elements, Sc and Zr stand out for the following reasons: The Al-Sc system exhibits a unique combination of a shallow solvus curve conducive to precipitation strengthening, an eutectic phase equilibrium favoring conventional solidification, and thermodynamically stable Al 3 Sc with the L1 2 structure. Sc is, however, the fastest diffuser in α-al and the priciest of the above eight elements. The Al-Zr system is characterized by one of the smallest diffusion rates and lattice parameter mismatch between Al 3 M and Al, as well as price among the eight candidates. The L1 2 structure of Al 3 Zr, however, is metastable. Furthermore, the Al-Zr system is peritectic which limits the concentration of Zr retained in solid-solution after conventional solidification, ultimately limiting the strength attainable by precipitation strengthening. Unlike the other Group 4 systems (Al-Ti and Al-Hf), the solid-solubility of Zr in α-al is negligible and so precipitation strengthening can be obtained in conventionally cast alloys. The steep α-al solvus, however, limits the possibility of post-solidification homogenization.

55 55 CHAPTER 2 Peritectic Solidification The peritectic reaction is the formation of the peritectic phase (α) by reaction of the primary phase (β) directly with the liquid (L) at the peritectic temperature (T p). Despite the existence of a peritectic reaction in an equilibrium phase diagram the possibility of the reaction taking place is dependent on the average alloy composition and the cooling rate from the melt, with increasing cooling rates and decreasing solute concentration generally limiting the extent of the reaction. Still faster rates of cooling prevent nucleation of the properitectic β phase, and supersaturated α may be formed directly from the melt. Terminal peritectic equilibrium introduces several complications with regard to castability, which were introduced in Chapter 1. This chapter is about understanding these complications, and more importantly how to avoid them. 2.1 Introduction IT WAS SHOWN IN CHAPTER 1 that the slowest diffusers in α-al occupy the Groups 4 6 in the periodic table. It is unfortunate that these same elements Ti, Zr, Hf, V, Nb, Ta, Cr, Mo, W are also the only elements exhibiting peritectic phase equilibria with the terminal α-al solid-solution. As discussed (Section 1.2.4), this peritectic equilibrium limits the feasibility for conventional solidification of these alloys because of the associated: 1. Liquidus elevation. The liquidus of a peritectic alloy rises continuously with increasing solute concentration, Figure 1.8(b). 2. Solute loss from primary precipitation of Al 3 M. In a peritectic alloy (of peritectic composition) the first solid form is the solute-rich properitectic β phase (Al 3 M for the systems discussed). Under equilibrium conditions, the solute concentration of the liquid phase follows the liquidus with the last solid to form having a composition necessarily less than the maximum solid-solubility.

56 CHAPTER 2 PERITECTIC SOLIDIFICATION Grain refinement associated properitectic precipitation. When Al 3 M forms primarily in the melt there is a concomitant refinement of the as-cast grain size of the alloy. This is deleterious for creep resistance since a refined grain structure can promote rapid diffusional creep. While the increase in melting temperature is unavoidable, the extent of properitectic Al 3 M precipitation (and hence the amount of solute lost to this phase, as well as the degree of grain refinement) is highly dependent on the concentration of solute in the alloy and the cooling rate at which the alloy is solidified. It is necessary, therefore, to understand the kinetics are of the peritectic reaction and what non-equilibrium solidification sequences are possible. The seminal paper on this topic, which was on dilute Al-Ti alloys, is that of Kerr et al. [170]. The first line of this paper aptly reads: The peritectic transformation appears to have puzzled students and authors alike. 2.2 Nomenclature Establishing clear definitions will be beneficial before discussing in detail the possible solidification sequences in peritectic systems. Peritectic solidification has called the attention of many researchers, most notably Kerr et al. [170,171] and St. John et al. [172, ], and consequently ambiguities in nomenclature have arisen [171]. We will use the nomenclature introduced by St. John with terminology suggested by Kerr and Kurz [171]. A peritectic reaction as part of a schematic binary phase diagram is shown in Figure 2.1. The peritectic reaction involves decomposition of β, and hence this phase is termed the properitectic phase; similarly, the product phase, α, is called the peritectic phase. The term peritectic composition refers to the composition C αp of the peritectic phase α at temperature T p, and corresponds to the maximum solid-solubility of solute in the α phase. This is an invariant point and is analogous to the eutectic composition in eutectic systems, since only at the composition C αp can the high temperature β and L phases be replaced completely by the lower temperature α phase on cooling. The liquid composition C Lp is termed the peritectic liquid. 1 At temperatures above T p all alloys to the right of the peritectic liquid C Lp will initially form some β phase directly from the liquid, whereas alloys with less than C Lp will not. Since the pres- 1 As discussed by Kerr and Kurz [171], C Lp has also been termed the peritectic point by others [170, 175] or peritectic limit by St. John [172,206,208]. Moreover, peritectic point has also been used to refer to C αp by St. John [206].

57 SECTION 2.3 MECHANISMS OF α FORMATION 57 N o n p e rite c tic H y p o p e rite c tic H y p e rp e rite c tic L L + β T e m p e r a tu r e T p α + L β T m α α + β A l C L p C α p C β p C o m p o s itio n (% M ) Fig. 2.1: Schematic peritectic phase diagram indicating symbols used for phases, compositions, and temperatures. During the peritectic reaction, primary phase (β) of composition C βp reacts with liquid (L) of composition C Lp to form peritectic α phase with composition C αp. ence of β (the properitectic phase) is required for the reaction, these alloys are thus designated as peritectic (C 0 > C Lp ) and nonperitectic (C 0 < C Lp ). Peritectic alloys are further distinguished as hypoperitectic (C 0 < C αp ) and hyperperitectic (C 0 > C αp ), as indicated in Figure Mechanisms of α Formation: The Peritectic Reaction, Peritectic Transformation, and α Nucleation Directly from the Melt Solidification of dilute Al alloys involves formation of α-al solid-solution from the melt (L). For a peritectic alloy there are three possible mechanisms of forming α-al: (i) the peritectic reaction; (ii) the peritectic transformation; and (iii) direct nucleation of supersaturated α-al.

58 CHAPTER 2 PERITECTIC SOLIDIFICATION Peritectic reaction Consider a peritectic alloy (i.e. of initial composition between C Lp and C βp ) cooled slowly from the melt. The equilibrium freezing sequence begins with the nucleation of the primary phase β below the liquidus, with subsequent growth of this phase as the L + β mixture is cooled. Withdrawal of solute from the melt will cause the liquid composition to follow the β liquidus with decreasing temperature until the interdendritic liquid has composition C Lp at the peritectic temperature T p. At this point, liquid of composition C Lp and primary β of composition C βp react together to give α: L + β α. The Gibbs Phase Law requires that, at T p, the three phases remain in local equilibrium until the reaction is complete, resulting in final equilibrium quantities of α, β, and L as dictated by the bulk alloy composition C 0 and the phase diagram. For an alloy precisely of peritectic composition, C αp, all of the α phase is produced by decomposition of properitectic β via the peritectic reaction, with no β remaining thereafter. For hypoperitectic alloys (i.e. those with initial composition between C Lp and C αp ), there is enough liquid remaining at T p to consume all of the primary β which has formed during cooling between the β liquidus and T p. Hyperperitectic alloys (between C αp and C βp ) have insufficient liquid to react with all of the properitectic β so that some primary phase remains in the final solidified alloy [209] Peritectic transformation In reality, the peritectic reaction can continue only until each primary β particle is completely encapsulated with α, at which point the three phase contact required for equilibrium is no longer possible. However, the α phase may continue to form at the expense of β by diffusion through the peritectic envelope enclosing each primary particle. This mechanism is termed the peritectic transformation.the rate of α formation by this mechanism is diffusion-controlled, and St. John and Hogan [206] showed that the peritectic transformation is negligible in the Al-Ti system, with the majority of the peritectic α solid-solution formed directly from the melt. Similar behavior is expected for the other Al-based peritectic systems. Since the peritectic transformation does not occur to any significant extent, properitectic β formed in the liquid is not consumed and hence the solute contained in this phase cannot be recovered back into α.

59 SECTION 2.3 MECHANISMS OF α FORMATION Direct solidification of supersaturated α Consider the schematic free energy curves in Figure 2.2. Under equilibrium conditions a hyperperitectic liquid of composition C 0 will lower its energy by the nucleation of properitectic β with the attendant change in the liquid composition to that of the β liquidus (this is the process of primary β precipitation described in Section 2.3.1). The nucleation of the primary β phase, however, requires a severe fluctuation in composition (25 at.% solute for a trialuminide primary Al 3 M phase compared to C at.% for the dilute Al alloys studied in this thesis). As shown in Figure 2.2, the free energy of the system can be lowered metastably by forming supersaturated α, which may be favored kinetically under accelerated cooling rates. In Figure 2.2, the common tangent to the α and liquid free energy curves touches them at higher solute concentrations than the equilibrium liquid composition, and these compositions will correspond to extrapolations of the equilibrium solidus and liquidus curves since both the equilibrium phase boundaries and the metastable extensions are defined by common tangents to the same curves [170]. This point was demonstrated convincingly by Kerr et al. [170, 210] in their seminal study on Al-Ti alloys. Using microprobe and concurrent thermal analyses, these researchers showed that with relatively high cooling rates a supersaturated α solid-solution nucleated above T p (Figure 2.3), with the measured liquidus and solidus compositions following their respective extrapolated values. Considering the equilibrium β liquidus, which increases rapidly with solute concentration (Figure 2.1), the amount of undercooling required to reach the extrapolated α-liquidus, and hence the cooling rate required to suppress properitectic formation, increases with solute concentration. There is consequently a critical cooling rate above which β phase nucleation is completely suppressed in favor of the α solid-solution, with suppression of the properitectic β phase becoming more difficult with increasing solute concentration. Nucleation of the peritectic phase α therefore occurs independently of the peritectic reaction, and crystallization of α can occur above the peritectic temperature T p. 2 2 This elevation of the reaction temperature is exactly opposite to that observed for eutectic systems, where the temperature of thermal arrest is depressed as the cooling rate of the melt is increased.

60 CHAPTER 2 PERITECTIC SOLIDIFICATION 60 Fig. 2.2: Free energy curves for the equilibrium phases at the Al-rich end of the Al-Ti system above the peritectic temperature T p. The relative driving forces for the equilibrium (nucleation of properitectic β) and nonequilibrium (nucleation of peritectic α) possibilities are illustrated for composition C 0. From Kerr et al. [170]. Fig. 2.3: Equilibrium Al-Ti phase diagram with metastable extrapolations constructed from microprobe analyses. From Kerr et al. [170]. 2.4 Intermediate Cooling Rates: Metastable Properitectic Phases As discussed, for any peritectic alloy the amount of properitectic precipitated in the melt decreases with increasing solidification rate and/or decreasing solute content; for sufficiently rapid cooling, the formation of the properitectic Al 3 M phase may be suppressed completely, resulting in the solidification of the peritectic α phase directly from the melt. At intermediate cooling rates, however, nucleation of the peritectic α phase may be preceded by precipitation of a metastable properitectic phase, β. As discussed below, these metastable properitectic phases (with the L1 2 structure) are the most desirable form of primary Al 3 M for grain refinement of commercial Al alloys, and consequently understanding the morphology, structure, and formation mechanism of these metastable phases as a function of cooling rate has received considerable attention in the literature (e.g., Figures 2.9) Compositionally metastable properitectic phases As discussed above, the direct nucleation of supersaturated α-al is favored at fast cooling rates because of the significant composition fluctuation required to precipitate the equilibrium tri-

61 SECTION 2.4 METASTABLE PROPERITECTIC PHASES 61 aluminide properitectic β (Figure 2.2). For the same reason, metastable sub-stoichiometric (Al-rich) properitectic phases may be favored under intermediate cooling rates since the composition fluctuation required to form these Al x M phases is less than that of the stoichiometric trialuminide (Al 3 M) properitectic. This phenomenon has been studied most extensively in Al- Ti alloys, where the occurrence of Al 3 Ti [71, 160, 211, 212], Al 17 Ti 4, Al 15 Ti 4 [210], Al 9 Ti [213], and Al x Ti [163] metastable properitectic phases have been reported. Recent convincing evidence [214, 215] suggests that the Ti concentration in these metastable primary phases is approximately 20 at.%, corresponding to Al 4 Ti, which is also in reasonable agreement with that reported originally by Kerr et al. [170, 210] Structurally metastable properitectic phases In addition to a possible sub-stoichiometric composition, intermediate cooling rates can also lead to structurally metastable properitectic phases. For the Al-based peritectic systems considered (i.e., transition elements of the Groups 4 6 alloyed with Al), the primary β phase formed under slow cooling of the melt is of equilibrium tetragonal (D0 22 or D0 23 ) crystal structure, exhibiting a coarse plate- or needle-like shape in cross-section, typically µm in length. At smaller solute concentrations and higher freezing rates these properitectic phases become finer and more equiaxed [80, 210, 211, 216], eventually exhibiting a morphology which is generally described as cuboidal or petal-like (Figure 2.4). Attending this change in morphology is a change in crystal structure from the equilibrium tetragonal to a metastable cubic L1 2 structure, which was first identified by Ohashi and Ichikawa investigating Al-Zr and Al-Ti alloys [160, 217]. The internal structure of these metastable L1 2 properitectic phases has been disputed. Several investigations on metastable L1 2 Al 3 Ti [71,160,219], Al 3 Zr [160,216,220], and Al 3 Hf [86,88] described the primary precipitates as two-phase aggregates, consisting of fine discontinuous lamellae of Al 3 M (L1 2 ) in their respective solid-solutions. Majumdar et al. [72, 214], however, showed that the petal-like Al 3 Ti primary precipitates were homogenous single phase. These differences in morphology have been attributed to breakdown of a stable growth front of Al 3 M during solidification [215]. Coarser metastable L1 2 cuboidal primary precipitates exhibit prominent extensions along 111 body diagonals, Figure 2.5, which has been reported for primary Al 3 Ti (L1 2 ) [71, 72, 214], Al 3 Zr (L1 2 ) [216,220,221], and Al 3 Hf (L1 2 ) [88,90]. An explanation for this preferred 111 growth

62 CHAPTER 2 PERITECTIC SOLIDIFICATION 62 Fig. 2.4: Effect of solidification rate and temperature of the melt on the morphology of primary Al 3 Zr precipitates formed in a hyperperitectic Al-Zr alloy (0.45 at.% Zr). As solidification rate increases, the primary phase becomes more equiaxed. (a) Solidification rate of C s -1, cooled from 1100 C. (b) Solidification rate of C s -1, cooled from 1100 C. (c) Solidification rate of C s -1, cooled from 1300 C. The efficacy of the grain refinement from the cuboidal particles is readily apparent. In addition to solidification rate, overheating the melt above the liquidus also affects the morphology of the primary phase, which is related to the degree of primary Al 3 Zr dissolution in the melt. From Brodova et al. [218]. is that {111} planes of any L1 2 Al 3 M phase have Al and M atoms in the ratio 1:1, while the {200} planes have only Al atoms. Thus, for these properitectic L1 2 phases to grow in 100 directions, two kinds of atomic planes must be placed one upon the other. Growth in the 111 direction, however, simply consists of sequential deposition of the same {111} planes consisting of equiatomic fractions of Al and M atoms. It is argued [216, 220, 221] that the latter may be favorable kinetically. As these primary phases coarsen this preferred 111 growth commonly involves the development of secondary branching and an extension into dendritic growth along the cube diagonals. Therefore, the morphology of these cuboidal particles is often described as dendritic, with four main branches generating from a solute-rich center [222, 223] Grain refinement As discussed previously (Section 1.2.4), primary Al 3 M precipitates are efficient heterogeneous nucleants of α-al during solidification, and additions of the Groups 4 6 elements (particularly

63 SECTION 2.4 METASTABLE PROPERITECTIC PHASES 63 Fig. 2.5: Transmission electron micrographs showing the 3-D morphology of metastable cubic L1 2 Al 3 Ti primary (properitectic) precipitates. (a) [ 11 1] β, (b) [1 1 1] β, and (c) [00 1] β orientations. From Majumdar et al. [214]. Ti) are routinely used as grain refiners in cast Al alloys. The grain size decreases as solute content increases, with significant grain refinement occurring when the solubility limit (for a given cooling rate) is exceeded and the Al 3 M phase precipitates primarily. This grain refinement phenomenon is especially well documented in the Group 4 systems (Al-Ti [160, 210, 219, 224], Al- Zr [80, 160, 216, 220, 222], and Al-Hf [86 88, 223]), and is shown in Figures 2.6 and 2.7 for Al-Ti and Al-Zr alloys, respectively, solidified under various cooling rates. The most effective form of the properitectic Al 3 M phase for grain refinement is the metastable L1 2 structure [80, 87, 216, 220, 223] (see Figure 2.4(c), in which each grain contains a petal-like Al 3 Zr primary precipitate), whose cubic structure is commensurate with- and exhibits a small lattice parameter mismatch to fcc α-al. The lattice parameters are Å for Al 3 Ti [72,160,214, 225], Å for Al 3 Zr [217], and Å for Al 3 Hf [223]. The metastable cubic primary precipitates are oriented with the cube-on-cube relationship with their respective surrounding α-al solid-solutions, (100) β (100) α and [001] β [001] α [72, 160, 213, 214, 217, 222, 226], indicating that α-al grows epitaxially from the Al 3 M primary phase. 3 3 This is demonstrated also in Figure 5.18.

64 CHAPTER 2 PERITECTIC SOLIDIFICATION 64 Fig. 2.6: Relation between the grain sizes and Ti concentration in Al-Ti alloys solidified at various cooling rates. From Ohashi and Ichikawa [160]. Fig. 2.7: Relation between the grain sizes and Zr concentration in Al-Zr alloys solidified at various cooling rates. From Ohashi and Ichikawa [160]. Grain refinement in eutectic systems The mechanism of grain refinement just described is not unique to peritectic alloys. Al-Sc alloys, which are eutectic, are also known for their potent grain refinement in cast alloys [43, ]. As with the peritectic alloys, grain refinement relies on the primary precipitation of the Al 3 Sc phase in the melt. Considering the schematic eutectic phase diagram in Figure 1.8(a), however, primary Al 3 Sc precipitation may occur only when the alloy composition is hypereutectic. In a peritectic alloy, Figure 1.8(b), primary precipitation of Al 3 M occurs generally at much lower solute concentrations. This, along with the much higher costs of Sc, explains why peritectic systems are favored industrially as grain refining inoculants to Al alloys. However, unlike in the peritectic alloys, in the Al-Sc system the L1 2 intermetallic compound is an equilibrium phase and grain refinement will consequently occur at any cooling rate.

65 SECTION 2.5 CHAPTER SUMMARY 65 T β T β T α T e m p e r a tu r e T p T p T m A l C o m p o s itio n (% M ) C 0 Fig. 2.8: Schematic equilibrium (full lines) and non-equilibrium (broken and dotted lines) peritectic phase diagram. Adapted from Nes et al. [232]. 2.5 Chapter Summary Mechanisms of α formation During the equilibrium solidification of a peritectic alloy, properitectic β (Al 3 M) is the first solid phase to form. At the peritectic temperature T p this phase and the liquid undergo a peritectic reaction (L + β α) to complete the solidification process. This textbook definition of the peritectic reaction is rarely realized in practice as it is stifled by the formation of an envelope of α solid-solution encapsulating the primary β dendrites. The three-phase contact required for equilibrium is no longer maintained, and further transformation must occur by diffusion through this peritectic α layer. This mechanism, denoted as the peritectic transformation, does not occur to any significant extent in the Al-Ti, Al-Zr, or Al-Hf systems and so virtually all of the α solid-solution is formed by solidification directly from the melt (with or without prior properi-

66 CHAPTER 2 PERITECTIC SOLIDIFICATION 66 Fig. 2.9: Effects of cooling rate and Zr concentration on the solidification microstructures of the Al-Zr alloys investigated by Hori et al. [216, 220]. Open circles indicate α solid-solution (no observed primary phase); half-filled circles indicate angular primary phase (composed of fine dendritic metastable L1 2 Al 3 Zr) + α solid-solution; filled circles indicate needle-like primary phase (equilibrium D0 22 Al 3 Zr) + α solid-solution. From Hori et al. [216, 220]. tectic β precipitation) Possible solidification sequences as a function of cooling rate Consider a hyperperitectic alloy of initial composition C 0, as shown in Figure 2.8. Under slow cooling from the melt the equilibrium primary β phase will be nucleated at temperatures below T β, resulting in coarse plate- or needle-like primary precipitates of equilibrium Al 3 M surrounded by a (solute-depleted) α-al solid-solution. Increased cooling rates can suppress the formation of the equilibrium phase and at temperatures below T β the melt will also be supersaturated with respect to the metastable Al 3 M L1 2 phase. Although this phase has a higher free energy than the equilibrium β, its formation may be favored due to a lower surface energy with the melt. By a further increase in cooling rate the formation of metastable β may also be suppressed, causing the melt eventually, at temperatures below T α, to be supersaturated with respect to α-al

67 SECTION 2.5 CHAPTER SUMMARY 67 L iq u id u s k 0 < 1 k 0 > 1 L iq u id u s T e m p e r a tu r e T S o lid u s T S o lid u s C S C L C L C S C o m p o s itio n C o m p o s itio n (a ) (b ) Fig. 2.10: Solidus-liquidus relationships for hypothetical dilute binary alloys where the solutes either lower (k 0 < 1) or raise k 0 > 1 the melting point of the alloy relative to the pure solvent. solid-solution. This latter scenario results in direct precipitation of α-al, which is most desirable for solid-state precipitation during post-solidification aging, since no solute is lost to primary phases and the solid-solution is metastably supersaturated. In real alloys, the situation is as depicted in Figure 2.9, which shows the interrelation of cooling rate and solute content on the observed properitectic morphology and microstructure of Al-Zr alloys [216,220] 4. To a specific cooling rate corresponds a critical solute concentration below which primary precipitation of properitectic (whether equilibrium β or metastable β ) will not occur Solute redistribution during α-al nucleation Solidification is a first-order phase transformation, characterized by a discontinuous change in equilibrium atom fractions of solute in the liquid and solid phases; the composition of the solid necessarily differs from that of the liquid in equilibrium with it, as conveyed quantitatively by the solidus and liquidus phase boundaries [233, 234]. When direct nucleation of α-al occurs in peritectic alloys, the solidification follows the dotted lines of the metastable extrapolated α liquidus and solidus extensions, Figure 2.8. The difference in composition at the growing solid-liquid interface, assuming that local equi- 4 Similar studies have been performed on Al-Ti alloys and other Al-Zr alloys, as discussed in Figure 3.9

68 CHAPTER 2 PERITECTIC SOLIDIFICATION 68 librium exists 5, can be described by the equilibrium distribution (or partition) coefficient [ ]: k 0 = C S C L which is simply the ratio of the concentrations of the solid and the liquid specified by the solidus and liquidus, respectively (Figure 2.10). As discussed in Section 1.2.4, k 0 may be expressed in terms of the solid and liquid compositions at the eutectic or peritectic reaction temperature, Table 1.5. It is of interest to compare the solute redistribution that occurs during solidification of α-al solid-solution in eutectic (k 0 < 1) and peritectic (k 0 > 1) systems. On paper, the two situations are no different. The starting condition is exactly the same, with the initial solid being of composition k 0 C 0 in both cases. During solidification, solvent (k 0 < 1) or solute (k 0 > 1) is rejected. As long as the partition coefficient k 0 is constant, some liquid will remain until an invariant temperature is reached. For eutectic systems this the eutectic point, where the remainder of the liquid then solidifies at eutectic composition (Figures 2.11(a) and (b)); for peritectic systems this is the melting point of pure Al, resulting in precipitate-free interdendritic regions during postsolidification aging (Figures 2.11(c) and (d)). As discussed in subsequent chapters, this segregation of solutes influences the distribution, morphology, and composition of the precipitates formed during post-solidification aging. It is also important to recognize that the nonuniform distributions of precipitates displayed in Figure 2.11(c) and 2.11(d) are not unprecedented; similar dendritic precipitate distributions have been observed during the decomposition of supersaturated Al-Ti [70, 72, 236], Al-Zr [76, 77, 80], and Al-Hf [86 88, 90, 223] solid-solutions all terminal peritectic systems with α-al. 5 This is generally a valid assumption for solid/liquid interface [ ].

69 SECTION 2.5 CHAPTER SUMMARY 69 Fig. 2.11: Consequences of eutectic (k 0 < 1) and peritectic (k 0 > 1) solidification on the resulting solute distribution in cast Al alloys. Panels (a) and (b) are optical micrographs from ternary Al-Mg-Si eutectic alloys cooled at ca. 5 C s -1, from [235]. Panel (c) is an SEM micrograph showing relief contrast between precipitate-rich and precipitate-free regions in an electropolished Al-Zr-Ti alloy aged at 425 C. Panel (d) is a dark-field TEM micrograph, showing the dendritic distribution of nanonscale Al 3 (Zr, Ti) (L1 2 ) precipitates in an Al-Zr-Ti alloy aged at 375 C.

70 70 CHAPTER 3 Nucleation and Precipitation Strengthening in Dilute Al-Ti and Al-Zr Alloys Two conventionally-solidified Al-Ti alloys (0.18 and 0.22 at.%ti) exhibit no hardening after aging up to 3,200 h at 375 or 425 C. This is due to lack of precipitation of Al 3Ti, as confirmed by electron microscopy and electrical conductivity measurements. This behavior is in contrast to that of an Al-0.19Zr (at.%) alloy, which displays strong age hardening at both temperatures due to precipitation of Al 3Zr (L1 2). These observations are interpreted within the context of the equilibrium phase diagrams. The disparity in solid and liquid solubilities of Ti in α-al is much greater than that of Zr in α-al; the relatively small liquid-solubility of Ti in α-al limits the amount of solute retained in solidsolution during solidification, while the comparatively high solid-solubility reduces the supersaturation effecting precipitation during post-solidification aging. The lattice parameter mismatch of Al 3Ti (L1 2) with α-al is also larger than that of Al 3Zr (L1 2), further hindering nucleation of Al 3Ti. Classical nucleation theory indicates that the minimum solute supersaturation required to overcome the elastic strain energy accompanying homogeneous nucleation cannot be obtained during conventional solidification of Al-Ti alloys (unlike for Al-Zr alloys), thus explaining the absence of Al 3Ti precipitation and presence of Al 3Zr precipitation. 3.1 Introduction AS DISCUSSED IN CHAPTER 1, the Group 4 transition metals (Ti, Zr, Hf) constitute a group of alloying additions to Al that show particular promise for developing creep-resistant, thermally-stable Al-based alloys. In each of these systems, an ordered Al 3 M (where M = Ti, Zr, or Hf) trialuminide may be precipitated from supersaturated solid-solution during postsoldification aging. While the equilibrium structure of these Al 3 M trialuminides is tetragonal (D0 22 for Al 3 Ti and Al 3 Hf 1 and D0 23 for Al 3 Zr), decomposition of supersaturated Al-M solidsolutions occurs initially by the formation of nanometer-scale metastable cubic L1 2 Al 3 M pre- 1 Al 3Hf exists in two different crystallographic structures: a stable high temperature D0 23 phase and a stable low temperature D0 22 phase [85]. The D0 22 structure is the one relevant to solid-state precipitation, while D0 23 is the relevant structure when considering primary precipitation from the liquid.

71 SECTION 3.1 INTRODUCTION 71 cipitates exhibiting small lattice parameter mismatches with α-al, which transform to their respective equilibrium structures after prolonged exposure (ca h) to elevated temperatures (ca. 450 C). Moreover, these transition elements are anomalously slow diffusers in α-al, with very limited equilibrium solid-solubilities, enabling precipitated Al 3 M to be resistant to Ostwald ripening in accordance with volume diffusion-controlled coarsening theory. For high-temperature applications, Ti seems especially promising among the Group 4 elements since it has the smallest atomic weight and is also the slowest diffuser in α-al (Section 1.2.3). Indeed, Al alloys strengthened by Al 3 Ti dispersions have garnered considerable interest as potential lightweight high-temperature structural materials, and the thermal stability and strength at high temperatures of alloys reinforced with Al 3 Ti dispersions is well-documented [73, ]. These studies, however, investigated alloys prepared by rapid solidification processing (RSP) [73, ] or mechanical alloying (MA) [237, 238, ] techniques, whose nonequilibrium processing routes circumvent the difficulties encountered during conventional solidification of Al-Ti alloys. The problems with casting these alloys arise from the fact that dilute additions of Ti or Zr (as well as Hf) to Al exhibit peritectic phase equilibria with the terminal α-al solid-solution. The first solid to form under equilibrium conditions is the properitectic, or primary, Al 3 M (M = Ti, Zr, Hf) ordered phase (for alloys of peritectic composition, i.e., those enriched beyond the minimum liquid-solubility of solute). These solute-rich primary phases readily nucleate and grow into coarse (ca. 100 µm) precipitates during conventional casting, leaving the remaining melt, and ultimately the solidified α-al solid-solution, substantially depleted in solute. This limits the potential for precipitation strengthening since the supersaturation and therefore the chemical driving force for solid-state nucleation, as well as the equilibrium volume fraction of the dispersed phase (if formed) is significantly reduced. The equilibrium structure of the primary precipitates is tetragonal (D0 22 for Al 3 Ti and D0 23 for Al 3 Zr and Al 3 Hf 1 ), but under certain conditions primary Al 3 M may precipitate also with the metastable cubic L1 2 structure [160]. These metastable primary precipitates are isostructural to, and exhibit a small lattice parameter mismatch with, the α-al solid-solution, and therefore act as efficient heterogeneous nucleants during solidification of α-al. The resulting grain refinement in cast alloys is a well known and frequently exploited phenomenon industrially for Ti additions to Al [ ,160], but has also been observed in Al-Zr [80,160,216,220,222] (and Al-Hf [86 88,

72 CHAPTER 3 NUCLEATION AND PRECIPITATION STRENGTHENING IN DILUTE AL-TI AND AL-ZR ALLOYS 72 Table 3.1: Designations and compositions (at.%) of the Al-Ti and Al-Zr alloys investigated. Designation Bodycote ATI Wah Chang Al-0.18Ti Al-0.22Ti Al-0.19Zr ]) alloys whenever primary precipitation of Al 3 M (L1 2 ) occurs. Both consequences of properitectic precipitation solute depletion from primary Al 3 M precipitation or potent grain refinement from metastable L1 2 Al 3 M precipitation are undesirable for high-strength high-temperature applications since the reduced solute concentration retained in solid-solution leads to smaller volume fractions of precipitated strengthening phases formed during aging, and the possibility of grain refinement is also deleterious for creep resistance since a refined grain structure promotes rapid diffusional creep. This study compares the feasibility of developing conventionally-cast and aged, precipitationstrengthened, dilute Al-based alloys alloyed with the Group 4 elements Ti or Zr. Most studies related to conventionally-solidified Al-Ti alloys have focused on the thermal stability of the relatively coarse primary Al 3 Ti phase formed during solidification [212, ]. As noted in references [263,264], the scientific literature pertaining to decomposition or age hardening of Al-Ti solid-solutions produced by conventional ingot metallurgy processing is rather limited. The primary objective, therefore, is to report and discuss the precipitation behavior in Al-Ti alloys, with the data for Al-Zr provided primarily as a comparison. In Chapter 5 the precipitation behavior and stability of Al 3 Zr precipitated from conventionally-solidified Al-Zr alloys is discussed in detail. 3.2 Experimental Procedures Alloy compositions and preparation Two Al-Ti alloys (Al-0.18Ti and Al-0.22Ti) and one Al-Zr alloy (Al-0.19Zr) were investigated (compositions indicated in Table 3.1). Small (ca. 7 g) buttons of these alloys were prepared by arcmelting in a gettered purified argon atmosphere using a non-consumable tungsten electrode

73 SECTION 3.2 EXPERIMENTAL PROCEDURES 73 and a water-cooled copper cathode as the crucible. The charges were melted a minimum of ten times, and inverted between melts to ensure alloy homogeneity. The moderately enhanced cooling rate associated with the chilled crucible was estimated to be of the order of C s -1, comparable to continuous casting. 2 For dilute Al-Ti and Al-Zr alloys, as well as other peritectic systems, to a specific cooling rate corresponds a critical solute concentration below which primary precipitation of Al 3 M (M = Ti or Zr) does not occur [ ]. The compositions in Table 3.1 were expected, based on solidification studies (Figure 3.9, discussed below), to be near the threshold for avoiding primary precipitation of Al 3 M under the moderate cooling rates investigated. Both Al-Ti alloys were prepared by melting nominally at.% Al (Atlantic Equipment Engineers, Bergenfield, NJ) with pure Al 3 Ti master alloy. A first advantage of this approach is that the melting point of Al 3 Ti (1350 C) is lower than that of pure Ti (1688 C). Additionally, this ordered structure is a line compound whose composition and constitution is entirely homogeneous. This is not true of commercial master alloys (containing typically 5 10 wt.% Ti) which exhibit coarse (ca. 100 µm) primary Al 3 Ti precipitates. For the dilute alloys in this investigation and the small mass of the button ingots, such inhomogeneous master alloys introduce unacceptable uncertainties in chemical composition. The monolithic Al 3 Ti master alloy was prepared by arcmelting amounts of pure constituents (99.99% Al, Atlantic Equipment Engineers; % Ti, Alfa Aesar, Ward Hill, MA) corresponding to the stoichiometry of the Al 3 Ti phase; chemical homogeneity and structural uniformity of the button ingot was verified with X-ray diffraction. The Al-0.19Zr alloy was prepared from a dilute master alloy containing 1.9 wt.% Zr, which was dilution cast from a commercial 10 wt.% Zr master alloy (KB Alloys, Reading, PA). The chemical compositions of both Al-Ti alloys were verified independently by Bodycote Materials Testing (Skokie, IL) and ATI Wah Chang (Albany, OR), while the composition of Al-0.19Zr was checked by ATI Wah Chang only Aging treatments and analytical techniques The as-cast alloys were isothermally aged at 375 or 425 C, within the range of temperatures reported to exhibit a strong age hardening response for Al-Ti alloys produced by RSP [73, 74, 2 The solidification rate of the arc-melter is discussed briefly in Appendix A. 3 See Appendix E.

74 CHAPTER 3 NUCLEATION AND PRECIPITATION STRENGTHENING IN DILUTE AL-TI AND AL-ZR ALLOYS , 265, 266] or MA [246] and Al-Zr alloys produced by RSP [14, 264, ] or chill-casting [187, 190, 271]. Precipitation of Al 3 Ti or Al 3 Zr during aging was monitored by a number of techniques, described below. First, Vickers microhardness measurements were performed at room temperature on metallographically polished sections using a load of 200 g and a dwell time of 5 s. Second, precipitation of Al 3 Ti or Al 3 Zr was assessed directly by transmission electron microscopy (TEM). TEM foils were prepared by mechanical grinding sections of aged specimens to a thickness of ca. 100 µm. Discs of 3 mm diameter were punched from these sections and thinned to perforation by twin-jet electropolishing at 20 V d.c. (Struers TenuPol-5) using a 10 vol.% solution of perchloric acid in methanol at -40 C. Electropolished TEM foils were also examined using a LEO 1525 high-resolution field-emission gun scanning electron microscope (SEM) operated at 3 kv with a short working distance (3 mm) using a secondary electron in-lens detector. As discussed below, the alloys are highly segregated on the micrometer scale, which is not easily discerned in the TEM. The advantage of SEM as a complement to TEM is the much larger field-of-view from both the larger range in magnification and also since observation of specimens in the SEM are not subject to the limitation that the foil be electron transparent. Furthermore, the resolution of the SEM, ca. 1 nm, is capable of resolving nanometer-scale Al 3 Ti or Al 3 Zr precipitates. Finally, for the Al-Ti alloys, which did not exhibit detectable age hardening or evidence for precipitation of Al 3 Ti by electron microscopy, electrical conductivity measurements were also performed to monitor the decomposition of the supersaturated solid-solutions during extended aging at 425 C. The conductivity measurements were performed using a SIGMATEST (Foerster Instruments, Pittsburgh, PA) eddy current apparatus at room temperature. Five measurements were recorded, each corresponding to a different frequency (60, 120, 240, 480, 960 khz), on each specimen. For consistency, a single specimen of each alloy (aged for various times) was used for conductivity measurements. 3.3 Experimental Results Optical microscopy Figure 3.1 shows the as-cast macrostructure of the alloys studied. The solidification macrostructure is typical of cast alloys, with coarse columnar grains, originating at the bottom surface of the

75 SECTION 3.3 EXPERIMENTAL RESULTS 75 Fig. 3.1: Macrostructure of as-cast alloys Al-0.18Ti, Al-0.22Ti, and Al-0.19Zr (etched using Poultan s reagent), showing various degrees of grain refinement. ingot (which was in contact with the chilled copper crucible of the arc-melter), growing upward toward a zone of equiaxed grains at the center of the ingot. The relative sizes of the columnar and equiaxed zones is strongly dependent on the solute content of the alloy and, more precisely, the extent of properitectic Al 3 M (M = Ti or Zr) precipitation, since these primary phases are potent grain refiners, as discussed above. Alloy Al-0.18Ti is the most dilute of those studied (Table 3.1), and the extent of the columnar zone, which comprises nearly half of the ingot cross-section in this alloy, is also greatest. Within the equiaxed zone, the grains are relatively coarse, with grain sizes ranging from mm across. The effect of a small increase in Ti concentration in these conventionally-solidified alloys is demonstrated for Al-0.22Ti, which exhibits a dramatic refinement in grain size with a fine columnar zone that adjoins an even finer equiaxed region of small ( µm) grains in the upper half of the ingot. Moderate grain refinement is exhibited also for Al-0.19Zr, which is intermediate between that of alloy Al-0.18Ti and alloy Al-0.22Ti. Nearly the entire ingot cross-section of Al-0.19Zr is equiaxed, with grain sizes ranging from mm in size. The pronounced grain refinement observed in Figure 3.1 for Al-0.22Ti is due to copious precipitation of petal-like primary Al 3 Ti, which were observed in metallographically polished

76 CHAPTER 3 NUCLEATION AND PRECIPITATION STRENGTHENING IN DILUTE AL-TI AND AL-ZR ALLOYS 76 V ic k e r s m ic r o h a r d n e s s (M P a ) A l T i A l T i 8 0 A l Z r d a y w e e k s V ic k e r s m ic r o h a r d n e s s (H V ) V ic k e r s m ic r o h a r d n e s s (M P a ) A l T i A l T i A l Z r d a y w e e k s V ic k e r s m ic r o h a r d n e s s (H V ) A s -c a s t A s -c a s t Fig. 3.2: Vickers microhardness vs. aging time at 375 C for the Al-Ti and Al-Zr alloys investigated. Fig. 3.3: Vickers microhardness vs. aging time at 425 C for the Al-Ti and Al-Zr alloys investigated. specimens by SEM. This petal-like morphology is characteristic of the metastable L1 2 Al 3 M phase [160], whose cubic structure is commensurate with fcc α-al and acts as an effective heterogeneous nucleant of α-al during solidification. The presence of these primary precipitates indicates also that, for the conventional casting conditions used in this study, exceeding ca. 0.2 at.% solute (Ti or Zr) results in primary precipitation of Al 3 M and hence does not increase the amount of solute retained in solid-solution Age hardening Figures 3.2 and 3.3 display the observed microhardness of the Al-Ti and Al-Zr alloys after isothermal aging at 375 and 425 C, respectively. Each data point represents a minimum of 20 measurements, with the standard deviation of these measurements indicated by the error bars. At both temperatures, the Al-Ti alloys exhibit a negligible age hardening response after extended aging times up to 3,200 h (19 weeks). This lack of strengthening contrasts with the significant precipitation hardening response of the Zr-containing alloy, which commences at times as short as 0.1 h (6 min) at 425 C, and achieves peak strength within 25 h at both 375 and 425 C. As described below, the pronounced strengthening is due to precipitation of fine (< 10 nm) coherent Al 3 Zr (L1 2 ) precipitates; no evidence for precipitation of a similar Al 3 Ti phase is observed. The ascast hardness values of Al-0.22Ti and Al-0.19Zr are both 242 MPa, approximately 20 MPa greater than that of Al-0.18Ti. This moderate strength increase is probably attributable to solid-solution

77 SECTION 3.3 EXPERIMENTAL RESULTS 77 Fig. 3.4: SEM micrographs of electropolished TEM foils: (a) as-cast Al-0.22Ti, showing preferential etching around the soluteenriched dendrites; (b) Al-0.22Ti aged at 425 C for 1,600 h, showing no evidence of precipitation of Al 3 Ti within the solute-enriched dendrites; (c) Al-0.19Zr aged at 425 C for 400 h, showing similar dendritic distributions of solute; (d) Magnified view of Al 3 Zr precipitate-rich dendritic regions and precipitate-free interdendritic channels. strengthening from dissolved Ti or Zr in the as-cast alloys [14,272]. While there is a refinement in the as-cast grain size with increasing solute concentration (Figure 3.1), Hall-Petch strengthening is not applicable because the scale of the microhardness indent (ca. 125 µm) relative to the grain sizes in Figure Electron microscopies The origin of the precipitation hardening response indicated in Figures 3.2 and 3.3 was investigated directly for alloys Al-0.22Ti and Al-0.19Zr using SEM, Figure 3.4. The dendritic distribution

78 CHAPTER 3 NUCLEATION AND PRECIPITATION STRENGTHENING IN DILUTE AL-TI AND AL-ZR ALLOYS 78 Fig. 3.5: Centered superlattice dark-field TEM micrographs of Al-0.19Zr aged at 425 C for 400 h. Images obtained under dynamical two-beam conditions, the specimen oriented near the [110] zone axis and imaged with the (1 10) L1 2 superlattice reflection as indicated in the inset diffraction pattern. (a) Precipitate-rich dendritic regions separated by a precipitate-free interdendritic channel similar to those in Figures 3.4(c) and 3.4(d). (b) Fine, coherent, homogeneously-distributed Al 3 Zr (L1 2 ) precipitates within the highly-supersaturated dendrite. The precipitates in this region have a mean radius of R = 6.7 ± 1.7 nm. of solute species formed during solidification in both alloys is revealed by preferential etching of the foil surface during electropolishing. Ryum [85] observed similar resistance to electropolishing in the dendritic cells of as-cast specimens of Al-Hf alloys. Figure 3.4(a) shows the as-cast structure for Al-0.22Ti, where the relief contrast around the dendrite arms is due to variations in solute concentration in solid-solution between the dendritic and interdendritic regions of the alloy. The center of the dendrites are enriched in solute [70, 170, 210, 273], and are apparently more resistant to electropolishing. Figure 3.4(b) shows the same Al-0.22Ti alloy after extended aging (1,600 h) at 425 C. The dendritic cells are still delineated, but no evidence for precipitation of Al 3 Ti within the supersaturated cells is observed, consistent also with the negligible age hardening of the Al-Ti alloys demonstrated in Figures 3.2 and 3.3. Analyses of the same specimen by TEM, both in real space (strain contrast) and reciprocal space (selected area electron diffraction), also provided no evidence for Al 3 Ti precipitates, or any other precipitated phase. Figures 3.4(c) and 3.4(d) show similar SEM micrographs for Al-0.19Zr aged 400 h at 425 C. As with the Ti alloys, dendrites are clearly delineated from the electropolishing process. Unlike the Al-Ti alloys, however, the solute-rich dendrites decompose during aging to form regions with high number density of homogeneously-distributed, nanometer-scale Al 3 Zr precipitates. Figure 3.5 shows superlattice dark-field TEM micrographs of the same specimen, exhibiting again

79 SECTION 3.4 DISCUSSION 79 precipitate-rich and precipitate-free regions associated with the initially dendritic Zr distribution. Within the highly-supersaturated dendrites, Figure 3.5(b), the Al 3 Zr precipitates have the metastable cubic L1 2 structure, are coherent with the α-al solid-solution, and are small ( R = 6.7 ± 1.7 nm). It is those precipitates that are responsible for the marked precipitation hardening response of Al-0.19Zr presented in Figures 3.2 and Electrical conductivity The electrical conductivity of a disordered binary substitutional alloy is a parabolic function of the absolute solute concentration [274], and is therefore a sensitive means of monitoring the decomposition of a supersaturated solid-solution during aging. Electrical conductivity measurements were performed on both Al-Ti alloys (Al-0.18Ti and Al-0.22Ti) during extended aging (up to 1,600 h) at 425 C. No detectable change in conductivity was observed, indicating that solute atoms remained dissolved in supersaturated solid-solution and did not precipitate as Al 3 Ti, corroborating the lack of precipitation observed by hardness and electron microscopies. As shown in Chapter 5, the decomposition of more dilute concentrations than those studied presently are easily detected by electrical conductivity. Moreover, Royset and Ryum [275] recently monitored in detail the decomposition kinetics of dilute Al-0.12Sc (at.%) alloys using a similar eddy current apparatus. 3.4 Discussion The hardness data in Figures 3.2 and 3.3 demonstrate that a conventionally-solidified Al-0.19Zr alloy exhibits pronounced precipitation hardening at 375 and 425 C, unlike Al-Ti alloys with similar solute concentrations. The strong age hardening response of the Zr-containing alloy is due to precipitation of fine (< 10 nm) coherent Al 3 Zr (L1 2 ) precipitates, shown by SEM (Figures 3.4(c) and 3.4(d)) and TEM (Figure 3.5). No evidence for precipitation of a similar Al 3 Ti phase is observed in the Al-Ti alloys by SEM or TEM. Moreover, no appreciable change in electrical conductivity is observed after aging Al-0.18Ti or Al-0.22Ti at 425 C for 1,600 h, indicating that the Ti concentration in α-al solid-solution is unchanged, confirming the lack of precipitation of Al 3 Ti.

80 CHAPTER 3 NUCLEATION AND PRECIPITATION STRENGTHENING IN DILUTE AL-TI AND AL-ZR ALLOYS 80 D iffu s io n d is ta n c e (n m ) C a lc u la te d r o o t-m e a n -s q u a r e d d iffu s io n d is ta n c e, (6 D t) 1 / A g in g tim e (h ) Fig. 3.6: Calculated root-mean-squared diffusion distances, 6Dt, for Ti and Zr in α-al at 375 and 425 C to t = 3,200 h, comparable to the aging conditions in the hardness curves of Figures 3.2 and Diffusion kinetics The measured tracer diffusivity of Ti in α-al is smaller than that of Zr in α-al, and it is conceivable that sluggish diffusion might account for the lack of Al 3 Ti precipitation. The diffusivities, D, of Ti and Zr in α-al are given by an Arrhenius relationship: D = D 0 exp ( Q/R g T ), (3.1) where R g is the ideal gas constant, T is the absolute temperature, Q is the activation enthalpy for solute diffusion, and D 0 is the pre-exponential factor. With values Q = 260 and 242 kj mol -1 and D 0 = and m 2 s -1 for Ti and Zr, respectively (Table 1.3), Figure 3.6 shows the calculated root-mean-squared (RMS) diffusion distances, 6Dt (where t is the aging time), for Ti and Zr in α-al at 375 and 425 C. Therefore, the diffusivity of Ti atoms at 425 C is comparable to that of Zr atoms at 375 C. The hardness curve of Figure 3.2 demonstrates that the Al-Zr binary alloy reaches near-peak hardness by 6.25 h of aging at 375 C, which corresponds to an RMS diffusion distance of 18 nm for Zr atoms. For the slower-diffusing Ti atoms, a similar RMS diffusion distance can be achieved

81 SECTION 3.4 DISCUSSION 81 after 115 h at 375 C or 4 h at 425 C, indicating that Ti atoms diffuse fast enough to support the nucleation and growth of Al 3 Ti precipitates. Atom-probe tomography studies on Al-Sc-Ti alloys aged at 300 C [118] and Al-Zr-Ti alloys aged at 375 and 425 C (Chapter 4) demonstrate that Ti segregates to the precipitated Al 3 (Sc 1 x Ti x ) and Al 3 (Zr 1 x Ti x ) precipitates, further evidence that the limited mobility of Ti is not the limiting factor explaining the unobserved precipitation of Al 3 Ti Classical nucleation theory The inability to nucleate Al 3 Ti is a result of an insufficient supersaturation of Ti in solid-solution following conventional solidification. Figures 3.7 and 3.8 display dilute equilibrium binary phase diagrams for the Al-Ti and Al-Zr systems based on experimental data. From a casting standpoint, a suitable alloying element with Al should produce a low liquidus temperature or, equivalently, exhibit high solubility in the liquid. This liquidus criterion facilitates melting into the single phase liquid and also reduces the tendency for primary phase precipitation during solidification. For precipitation strengthening, limited solid-solubility is required to maximize the chemical driving force for nucleation as well as the volume fraction of the precipitated phase. From Figures 3.7 and 3.8, the disparity in the liquid and solid-solubilities of Ti in α-al is much greater than that of Zr. The equilibrium phase boundaries in Figures 3.7 and 3.8 correspond to the equilibriumstructured trialuminides: D0 22 for Al 3 Ti and D0 23 for Al 3 Zr. The solubility limits of Ti and Zr in α-al, in equilibrium with their respective metastable L1 2 structure Al 3 M trialuminides, have been evaluated in recent ab initio calculations by Liu and Asta [276, 277]: C Al 3Ti α = exp ( /T ) C Al 3Zr α = exp ( /T ) (3.2) where C Al 3Ti α and C Al 3Zr α are the solubilities (expressed in atomic fraction) of Ti and Zr in equilibrium with their respective metastable L1 2 precipitates. These calculated solvus boundaries are indicated by the dotted lines in Figures 3.7 and 3.8. According to classical nucleation theory [ ], the steady-state nucleation current (nu-

82 CHAPTER 3 NUCLEATION AND PRECIPITATION STRENGTHENING IN DILUTE AL-TI AND AL-ZR ALLOYS L L + A l 3 T i L a t.% L + A l 3 Z r (A l) a t.% C o n c e n tra tio n s o f T i s tu d ie d a t.% C M e ta s ta b le L 1 2 S o lv u s (A l) + A l 3 T i (A l) M e ta s ta b le L 1 2 S o lv u s a t.% a t.% (A l) + A l 3 Z r C o n c e n tra tio n o f Z r s tu d ie d C A to m ic P e r c e n t T i A to m ic P e r c e n t Z r Fig. 3.7: Equilibrium Al-rich Al-Ti binary phase diagram (adapted from Murray [122]). Metastable L1 2 Al 3 Ti solvus calculated by Liu and Asta [276]. Fig. 3.8: Equilibrium Al-rich Al-Zr binary phase diagram (adapted from Murray [123]). Metastable L1 2 Al 3 Zr solvus calculated by Liu and Asta [276]. cleation rate per unit volume) may be written: J exp ( F ) ( exp Q ), (3.3) R g T R g T where F is the net reversible work required for the formation of a critical nucleus, given by: F = 16πσ 3 3( F ch F el ) 2. (3.4) The steady-state nucleation current, Eq.(3.3), is thus a function of the chemical driving force, F ch ; the reduction of this driving force by the elastic strain energy, F el ; and the activation enthalpy for diffusion, Q. As discussed below, these three parameters hinder the nucleation of Al 3 Ti compared to Al 3 Zr. Chemical driving force, F ch, for nucleation of Al 3 M The metastable L1 2 solvus boundaries in Figures 3.7 and 3.8 demonstrate that the solid-solubility of Ti in α-al is significantly greater than that of Zr, which dictates the chemical driving force for

83 SECTION 3.4 DISCUSSION 83 homogenous nucleation of metastable L1 2 Al 3 M (M = Ti or Zr). Assuming ideal solution behavior (valid for the dilute concentrations discussed), this driving force is proportional to the natural logarithm of the supersaturation ratio, C 0 /C α, where C 0 is the concentration (atomic fraction) of M dissolved in solid-solution and C α is the solid-solubility of M in α-al given by Eqs.(3.2). Following Doherty [283], the chemical driving force per unit volume driving precipitation of the ordered Al 3 M phase may be estimated as: F ch = R gt V Al3 M ( CAl3 M C α 1 C α ) ln C 0 C α, (3.5) where C Al3 M = 0.25 is the atomic fraction of solute in the ordered trialuminide, and V Al3 M = N a (a Al3 M) 3 /4 (N a is Avogadro s number) is the molar volume of the precipitated phase. Figures 3.7 and 3.8 indicate that, for a given solute concentration (C 0 ), the supersaturation driving precipitation of Al 3 Zr is significantly greater than that for Al 3 Ti, which partially explains the unobserved precipitation of Al 3 Ti because of the reduced supersaturation ratio C 0 /C α entering Eq.(3.5). The solute concentration C 0 is, however, not constant throughout the alloys, because of the segregation visible in the SEM micrographs of Figure 3.4. For both Al-Ti and Al- Zr, the dendrites are supersaturated with respect to the overall bulk alloy composition, C 0, and so in both cases the local supersaturation of solute can be greater than the bulk compositions indicated in Table 3.1 and Figures 3.7 and 3.8. The degree of this segregation was not measured directly (e.g., by electron microprobe analysis, EPMA), but may be estimated from the equilibrium solid-liquid partition coefficient, k 0 = C S /C L, which quantifies the difference in composition of the solid and liquid phases in local equilibrium during solidification [ ]. Making the usual assumption that both the liquidus and solidus boundaries are straight lines, k 0 may be expressed in terms of the solid and liquid compositions at the peritectic temperature. From Figures 3.7 and 3.8, k 0 = 10 for Ti and k 0 = 2.5 for Zr. Therefore, the first solid to form is predicted to be enriched by a factor of 10 and 2.5 for Ti and Zr, respectively. This is in reasonable agreement with results of Kerr et al. [170, 210], who measured the concentration of Ti across the dendritic cells in Al-Ti alloys containing up to 0.28 at.% Ti using EPMA, and observed an enrichment in Ti of 5 6 times in the dendrite centers. Setiukov and Fridlyander [273] similarly observed that the Ti concentration in the central zones of the dendritic cells exceeded its average value by 6 8 times in alloys containing at.% Ti. All present Al-Ti and Al-Zr alloys are therefore locally supersaturated well beyond

84 CHAPTER 3 NUCLEATION AND PRECIPITATION STRENGTHENING IN DILUTE AL-TI AND AL-ZR ALLOYS 84 their respective equilibrium solubilities of solute, and so an insufficient supersaturation of solute, F ch, alone cannot account for the absence of Al 3 Ti precipitates. As indicated in Eq.(3.4), the influence of elastic strain energy, F el, must also be considered since this reduces the net driving force for nucleation. Reduction of the driving force by the elastic strain energy, F el Solid-state reactions involve strain energy since precipitating a second phase, which is fully or partially coherent with the matrix, requires the straining of the lattices in both the matrix and precipitated phases. This strain energy, F el, diminishes the net driving force, Eq.(3.4), and hence the rate of nucleation, Eq.(3.3). Considering only dilatational strains and assuming elastic isotropy, the strain energy per unit volume for a coherent inclusion may be written [279, 280]: F el = 2µ ( ) 1 + ν δ 2, (3.6) 1 ν where µ = 25.4 GPa [27] and ν = [26] are the shear modulus and Poisson s ratio of Al at room temperature. The average lattice parameter mismatch at room temperature, δ, of both the tetragonal and metastable cubic (L1 2 ) phases are significantly greater for Al 3 Ti (δ = 2.04% and 5.36% for L1 2 and D0 22, respectively) than for Al 3 Zr (δ = 0.75% and 2.89% for L1 2 and D0 23, respectively), Table 1.1. Thus, the elastic strain energy impeding nucleation of Al 3 Ti (L1 2 ) is over 7 times greater than that of Al 3 Zr (L1 2 ). Critical solute concentration, C 0, to overcome F el Compared to Al 3 Zr (L1 2 ), nucleation of Al 3 Ti (L1 2 ) is hindered by: (i) a reduced chemical driving force ( F ch ) for a given C 0 ; and (ii) a greater elastic strain energy opposing nucleation ( F el ). To explain why nucleation of Al 3 Zr (L1 2 ) is observed and that of Al 3 Ti (L1 2 ) is not, consider the relative magnitudes of F ch and F el entering Eq.(3.4) in both systems. The chemical driving force, F ch (Eq.(3.5)), is a strong function of C 0 and is calculated using the parameters in Table 3.2. Equating F ch and F el gives a critical solute concentration necessary to overcome the elastic strain energy barrier associated with Al 3 Ti ( F el = 45 MJ m -3 ) and Al 3 Zr ( F el = 6 MJ m -3 ) nucleation. For Al-Ti alloys, the critical value of C 0 is at.% Ti and for Al-Zr it is at.% Zr.

85 SECTION 3.4 DISCUSSION 85 Table 3.2: Calculation of chemical driving force, F ch, effecting nucleation of Al 3 Ti (L1 2 ) and Al 3 Zr (L1 2 ). V Al3 M C α at 425 C (Eq.(3.2)) C Al3 M F ch (Eq.(3.5)) (m 3 mol 1 ) (atomic fraction) (atomic fraction) (MJ m -3 ) Al 3 Ti (L1 2 ) ln (229.4 C 0 ) Al 3 Zr (L1 2 ) ln ( C 0 ) Table 3.1 indicates that the bulk compositions of both of the Al-Ti alloys studied are less than the critical threshold of at.% Ti whereas Al-0.19Zr is supersaturated well beyond at.% Zr, thus explaining the observed precipitation of Al 3 Zr (L1 2 ) and the lack thereof for Al 3 Ti. This explanation is actually not strictly complete because, as discussed, the alloys are segregated and hence locally supersaturated beyond the bulk compositions in Table 3.1. Nevertheless, the small chemical driving force, F ch, coupled with the large elastic strain energy barrier, F el, in the Al- Ti system as compared to the Al-Zr system is an important contribution to the explanation for the difference in precipitation behavior between these two systems. Activation enthalpy for solute diffusion, Q As indicated in Eq.(3.3), the nucleation current is strongly dependent on temperature. The solute supersaturation (and, correspondingly, F ch ) increases with decreasing temperature (Eqs.(3.2)) while the solute diffusivity D diminishes rapidly (Eq.(3.1)). These competing influences result in an intermediate temperature for optimum nucleation kinetics [282]. As discussed in Section 3.4.1, the activation enthalpy for solute diffusion, Q, for Ti in α-al (260 kj mol -1 ) is greater than that of Zr (242 kj mol -1 ), which tends to increase the temperature for maximum nucleation current for Al 3 Ti. As discussed, F ch driving precipitation of Al 3 Ti is significantly less than that for Al 3 Ti for a given C 0 (Figures 3.7 and 3.8). This reduced driving force is exacerbated by the higher Q for Ti diffusion since at the temperatures where Ti is mobile, the supersaturation is not large enough to effect nucleation of Al 3 Ti Limitations on increasing the solute concentration The discussion in Section indicates that an increased concentration of Ti (C 0 ) beyond those studied here is required for precipitation of Al 3 Ti. This, however, is not possible under con-

86 CHAPTER 3 NUCLEATION AND PRECIPITATION STRENGTHENING IN DILUTE AL-TI AND AL-ZR ALLOYS T i Z r (d ) (c ) (a ) (b ) (a ) T i o r Z r c o n c e n tr a tio n (a t.% ) Fig. 3.9: Critical cooling rate for suppressing primary Al 3 Ti (solid lines) or Al 3 Zr (dashed lines) and achieving supersaturated α-al solid-solution, as reported by: (a) Ohashi and Ichikawa [160]; (b) Kerr et al. [170]; (c) Hori et al. [71,219]; and (d) Hori et al. [216,220]. ventional casting conditions. As demonstrated in Figure 3.1, the marked grain refinement observed with increasing Ti concentration is due to primary precipitation of Al 3 Ti in the melt, and increasing the alloy composition does not increase the amount of Ti retained in α-al solidsolution. The interrelation between solidification rate, solute concentration, and solidified microstructure of the Al-Ti and Al-Zr binary systems has been the subject of a number of studies [71,160,170, 216,219, 220], which are summarized in Figure 3.9. These curves show the measured solidification rate, as a function of solute concentration, necessary to suppress primary precipitation of Al 3 Ti or Al 3 Zr, thereby obtaining supersaturated α-al solid-solution. Figure 3.9 indicates that in order to achieve a concentration of at.% Ti supersaturated in α-al solidsolution, required to just offset the elastic strain energy associated with nucleation of Al 3 Ti, accelerated cooling rates of C s -1 (based on the data of Hori et al. [71,219]) are required. To achieve a greater driving force for nucleation higher cooling rates are required still. Nucleation of Al 3 Zr requires a small supersaturation to overcome the elastic strain energy barrier, which is

87 SECTION 3.4 DISCUSSION 87 easily achieved for even the slowest cooling rates in Figure Comparison with previous studies Decomposition behavior of Al-Ti solid-solutions In contrast to the closely-related Al-Zr and Al-Hf systems, the precipitation behavior of Al 3 Ti in Al-Ti alloys has been debated in the scientific literature. As with Al-Zr and Al-Hf alloys, the decomposition sequence of supersaturated Al-Ti solid-solutions has been reported to occur firstly by the precipitation of a metastable Al 3 Ti (L1 2 ) phase, which then transforms to the equilibrium tetragonal (D0 22 ) structure after prolonged aging (Chapter 1). Asboll and Ryum [236], however, observed only precipitation of the equilibrium D0 22 phase in chill cast Al-0.34Ti (at.%) alloys aged C. These researchers also noted that the decomposition kinetics of Al 3 Ti were sluggish compared to that of Al alloyed with other elements of the same subgroup, Zr or Hf. Pandey and Suryanarayana [93] similarly compared the decomposition behavior of rapidlysolidified Al alloyed with the Group 4 elements (Ti, Zr, Hf) and noted that, while a metastable L1 2 Al 3 M ordered phase is observed in Al-Zr and Al-Hf alloys, this is not the case for Al-Ti alloys. This is in accord with other studies [243,262,269] reporting only precipitation of the equilibrium D0 22 phase in Al-Ti alloys aged in the broad temperature range of C. Park and Kim [269], however, reported that the equilibrium D0 22 structure was preceded by another metastable tetragonal Al 3 Ti phase in Al-1.13Ti (at.%) alloys aged at 400 C. Asboll and Ryum [236] also observed tetragonal Al 3 Ti precipitates which were not of the D0 22 structure, in addition to those of the equilibrium D0 22 phase. Other studies report no precipitation during aging of supersaturated Al-Ti solid-solutions. Ohashi et al. [284] found no evidence for precipitation of Al 3 Ti, using microhardness and electrical resistivity measurements, in Al-Ti alloys containing up to 0.45 at.% Ti produced by RSP during 100 h of aging in the range of C. St. John et al. [261] similarly observed no solidstate precipitation of Al 3 Ti when aging an Al-1.1Ti (at.%) alloy for 24 h at 435 C; their alloy did, however, contain coarse primary Al 3 Ti precipitates, indicating the amount of Ti initially in solidsolution was reduced. Nucleation of coherent Al 3 Ti (L1 2 ) precipitates has been reported only for highly supersaturated alloys produced by nonequilibrium means (RSP). Ohashi and Ichikawa [70] observed Al 3 Ti (L1 2 ) precipitates in alloys containing at.% Ti aged at C. Muddle et al. [72 75]

88 CHAPTER 3 NUCLEATION AND PRECIPITATION STRENGTHENING IN DILUTE AL-TI AND AL-ZR ALLOYS 88 similarly showed precipitation of coherent Al 3 Ti (L1 2 ) in melt-spun alloys containing at.% Ti aged at C. Similar supersaturations cannot be achieved by conventional casting, Figures 3.1 and 3.9, since Ti precipitates out in the liquid as properitectic Al 3 Ti at moderate cooling rates. Age hardening in the Al-Ti and Al-Zr systems As with the decomposition behavior from solid-solution, the age hardening response accompanying Al 3 Ti precipitation is similarly disputed. Dobatkin et al. [268] observed that, among several transition metals alloyed with Al, Al-Ti alloys exhibit one of the weakest hardening responses in isochronal (2 h) aging experiments. This lack of hardening, however, was not due to the absence of precipitation since decomposition of the supersaturated solid-solution, as monitored by electrical resistivity, was commenced upon aging at ca. 300 C. In similar isochronal aging experiments, Park and Kim [269] also confirmed precipitation of Al 3 Ti (D0 22 ) by X-ray diffraction and TEM in an Al-1.13Ti (at.%) aged at 400 C, but observed no accompanying increase in microhardness. Precipitation of Al 3 Ti with no apparent increase in strength was shown more conclusively by Asboll and Ryum [236], who aged a chill-cast Al-0.34Ti (at.%) alloy at temperatures between C. No increase in hardness was observed after up to 24 days, but precipitation of irregularly-shaped lath Al 3 Ti (D0 22 ) precipitates was confirmed by TEM to occur within the highly supersaturated dendritic cells of the alloy. The precipitates were relatively coarse ( nm in length) and exhibited several, non-specific, orientation relationships with the surrounding α-al solid-solution; other studies [72, 260] have also reported an ill-defined orientation relationship between Al 3 Ti (D0 22 ) and α-al. As discussed above, both the metastable L1 2 and equilibrium D0 22 structures of Al 3 Ti exhibit a large lattice parameter mismatch with α-al. This precludes the formation of a coherent Al 3 Ti strengthening phase, and may explain the reported negligible age hardening even when precipitation of Al 3 Ti is confirmed. The questionable age hardening behavior in Al-Ti alloys contrasts significantly with the otherwise similar Al-Zr system, in which a strong age hardening effect, even in conventionally cast alloys, is welldocumented [78, , 267].

89 SECTION 3.5 CHAPTER SUMMARY Common industrial uses of Ti and Zr additions to Al Perhaps the most illustrative comparison of Al-Ti and Al-Zr alloys is found by considering their primary industrial roles. As discussed previously, minor additions of Ti are used to refine the grain structure in commercial aluminum alloys, where primary Al 3 Ti precipitates act as heterogeneous nuclei during solidification of the melt [ , 160]. A solid state counterpart to this grain refinement effect is realized when dilute additions of Zr are added to commercial wrought alloys as recrystallization inhibitors [78, 168, ]. The precipitation of fine (20 30 nm) dispersions of coherent Al 3 Zr (L1 2 ) precipitates, formed during post-solidification aging, are effective barriers to grain boundary motion. Titanium is thus added for the purpose of forming primary precipitates in the melt whereas Zr additions form coherent dispersoids precipitated from supersaturated solid-solution. This suggests that: (i) Zr, compared to Ti, is easier to retain in solid-solution during conventional solidification; and (ii) Al 3 Zr compared to Al 3 Ti is more readily precipitated during post-solidification aging. These considerations indicate that the Al- Zr system is a better candidate than the Al-Ti system for developing a castable precipitationstrengthened alloy. 3.5 Chapter Summary This investigation has compared the decomposition of supersaturated, conventionally-cast Al- 0.2Ti and Al-0.2Zr (at.%) alloys, upon aging at 375 or 425 C. The Zr-containing alloy exhibits a strong age hardening response at both temperatures, due to copious precipitation of coherent, nanometer-scale Al 3 Zr (L1 2 ) precipitates within the supersaturated dendritic cells of the alloy. Precipitation of a similar Al 3 Ti phase is unobserved during extended aging times (up to 3,200 h at 425 C). Compared to Al 3 Zr, nucleation of Al 3 Ti (L1 2 ) is hindered by: (i) a reduced chemical driving force ( F ch ) driving nucleation (for a given C 0 ); (ii) a greater elastic strain energy opposing nucleation ( F el ); and (iii) a greater activation enthalpy for solute diffusion (Q). The minimum supersaturation to effect precipitation of Al 3 Ti cannot be achieved by conventional solidification of binary Al-Ti alloys, since increasing the Ti concentration causes precipitation of primary Al 3 Ti, and thus does not contribute to a supersaturation of solute in solid-solution. The Al-Zr system is more promising system for developing a castable alloy capable of precipitation hardening.

90 90 CHAPTER 4 Atom-Probe Tomographic Studies of Precipitation in Al-0.1Zr-0.1Ti (at.%) Alloys Atom-probe tomography was utilized to verify directly the chemical composition of Al 3(Zr 1 xti x) (L1 2 structure) precipitates formed in Al-0.1Zr-0.1Ti (at.%) alloys during extended aging times at 375 or 425 C. At these temperatures, the Zr:Ti atomic ratio in the precipitates is about 10 and 5, respectively, indicating that most of the available Ti fails to partition to the Al 3(Zr 1 xti x) phase. This is consistent with prior studies on Al-Sc-Zr and Al-Sc-Ti alloys, where the slower-diffusing Zr and Ti atoms make up a small fraction of the Al 3Sc-based precipitates. Unlike these alloys, however, the Al-Zr-Ti alloys do not exhibit evidence for interfacial segregation of Ti to the α-al-al 3(Zr 1 xti x) interface, a result that may be affected by the disparity in evaporation fields of the two phases. 4.1 Introduction IT WAS SHOWN IN CHAPTERS 1 AND 3 that the Al-Zr system shows particular promise for developing creep-resistant, thermally-stable Al-based alloys. It is well established that decomposition of supersaturated Al-Zr solid-solutions occurs initially by the formation of nanometerscale Al 3 Zr precipitates with a metastable cubic L1 2 structure, which transform to the equilibrium D0 23 phase after prolonged aging at elevated temperatures (> 450 C). The high stability of the L1 2 metastable phase is attributed to slow diffusion kinetics and a small lattice parameter mismatch with α-al (Chapter 1). Fine and coworkers showed that this lattice parameter mismatch is reduced by additions of Ti, V, or Hf [84,110,111], resulting in improved coarsening resistance of L1 2 Al 3 (Zr 1 x V x ) [84, ] and Al 3 (Zr 1 x Ti x ) [285] precipitates, which they ascribed to a reduced matrix/precipitate interfacial free energy. These findings motivated subsequent studies on Al-Zr-V [270,286], Al-Zr- Ti [264, 271, ], and Al-Zr-Ti-V [286, 294, 295] alloys, in which the precipitated ternary and quaternary trialuminides were reported to offer improved coarsening resistance and thermal

91 SECTION 4.2 EXPERIMENTAL PROCEDURES 91 stability compared to binary Al 3 Zr precipitates [84]. 1 This improved stability was attributed to a reduced lattice parameter mismatch, as suggested by Fine, but in none of these studies (with the exception of reference [271]) was the composition of the precipitates (and hence the lattice parameter mismatch) measured directly. Three-dimensional atom-probe tomography (3-D APT) [ ] is based on the fieldevaporation of surface atoms, as ions in different charge states, that are identified chemically by time-of-flight mass spectrometry and whose positions are deduced from the coordinates of ion impacts on a 2-D position-sensitive detector. Small, nanometer-scale concentration fluctuations are examined on the sub-nanometer-scale in real space, without data deconvolution, and without comparison to standards. In this respect, 3-D APT is uniquely suited to examine directly the effects of microalloying in nanoscale precipitation processes [299, 300]. In this chapter, the compositions of nanometer-scale Al 3 (Zr 1 x Ti x ) precipitates formed upon aging at 375 or 425 C for various aging times are measured directly by 3-D APT. The effect of the precipitate compositions on the microstructures and mechanical properties of these alloys is discussed in Chapters 5 and Experimental Procedures Two ternary Al-0.1Zr-0.1Ti (all compositions are given in at.%) alloys were investigated, designated Al-0.1Zr-0.1Ti(a) and Al-0.1Zr-0.1Ti(b). The alloys were prepared by melting nominally at.% Al (Atlantic Equipment Engineers, Bergenfield, NJ) with pure Al 3 Ti master alloy and a dilute Al-0.57Zr (at.%) master alloy employing non-consumable electrode arc-melting in a gettered purified argon atmosphere. The monolithic Al 3 Ti master alloy was prepared by arcmelting amounts of the pure constituents 99.99% Al (Atlantic Equipment Engineers, Bergenfield, NJ); % Ti (Alfa Aesar, Ward Hill, MA to obtain the stoichiometric Al 3 Ti phase; the chemical homogeneity and structural uniformity of the Al 3 Ti button ingot was verified with x-ray diffraction. The Al-0.57Zr master alloy was dilution cast from a commercial 10 wt.% Zr master alloy provided by KB Alloys (Reading, PA). The chemical composition of alloy Al-0.1Zr- 0.1Ti(b) was confirmed by bulk chemical analysis performed by direct current plasma emission spectroscopy (ATI Wah Chang, Albany, OR). As-cast specimens were isothermally aged at 375 or 1 These studies are discussed in detail in Chapter 5. 2 See Appendix E.

92 CHAPTER 4 ATOM-PROBE TOMOGRAPHIC STUDIES OF PRECIPITATION IN AL-0.1ZR-0.1TI (AT.%) ALLOYS 92 Table 4.1: Compositions and aging conditions of the Al-0.1Zr-0.1Ti alloys investigated by 3-D APT. Nominal comp. (at.%) Verified comp. (at.%) Aging conditions Zr Ti Zr Ti Temperature ( C) Time (h) Al-0.1Zr-0.1Ti(a) , 100, 400, 1600 Al-0.1Zr-0.1Ti(b) C, within the range of temperatures shown to produce a strong precipitation strengthening response for Al-Zr alloys (Chapter 5). The alloy designations, compositions, and aging conditions investigated by 3-D APT are summarized in Table 4.1. Needle-like 3-D APT specimens were prepared by a two-step electropolishing procedure. Specimen blanks, excised from aged specimens and mechanically ground to approximately mm 3, were initially shaped into long needle-like tips with a slight taper angle using a solution of 10 vol.% perchloric acid in acetic acid at ca. 10 V d.c. at room temperature. Final electropolishing involved formation of a neck near the tip apex, controlled by carefully limiting the amount of chemical solution in contact with the area of the neck using a loop-polishing apparatus [301, 302]. For this final tip-preparation procedure, a solution of 2 vol.% perchloric acid in butoxyethanol at 3 8 V d.c. was employed. The resulting specimen is a sharply-pointed tip with an end radius of curvature less than 50 nm. A new class of atom probe tomographs, the local electrode atom-probe (LEAP ) [ ], was used for most 3-D APT analyses. Pulsed field-evaporation was conducted under ultrahigh vacuum (UHV) conditions (<10 10 Torr gauge pressure) at a specimen temperature of 30 K utilizing a pulse fraction (ratio of the pulse voltage to the steady-state d.c. imaging voltage) of 15 20% and a pulse frequency of 200 khz. A fixed flight path of 80 mm was used for all analyses. Some 3-D APT analyses were also performed using an energy-compensated optical positionsensitive atom-probe [ ] (referred to in the text as conventional 3-D APT), with a pulse fraction of 20%, 1.5 khz pulse frequency, and a specimen temperature of 30 K in a vacuum of <10 9 Torr gauge pressure. Post-analysis data visualization and evaluation were performed with IVAS (Imago Scientific Instruments: Madison, WI) and custom software developed at Northwestern University called ADAM or currently APEX [309]. Quantitative precipitate compositions and solute composition profiles were obtained using the proximity histogram method [310, 311], measuring solute con-

93 SECTION 4.3 EXPERIMENTAL RESULTS 93 Fig. 4.1: Centered superlattice dark-field TEM micrographs of alloy Al-0.1Zr-0.1Ti(a) aged at 375 C for 1,600 h. Images obtained under dynamical two-beam conditions, with the specimen oriented near the [221] zone axis and imaged with the (1 10) L1 2 superlattice reflection from the Al 3 (Zr 1 x Ti x) precipitates. (a) Low magnification image showing dendritic distribution of Al 3 (Zr 1 x Ti x) precipitates. (b) Magnified view of fine (3 5 nm radius), coherent, homogeneously-distributed L1 2 Al 3 (Zr 1 x Ti x) precipitates within the highly-supersaturated dendrites. A representative nm 3 3-D APT analysis volume is indicated for comparative purposes. centration profiles in the α-al matrix and Al 3 (Zr 1 x Ti x ) precipitate phases with respect to a 5 at.% Zr isoconcentration surface delineating the two phases. 4.3 Experimental Results Solute segregation and the precipitated microstructure Figure 4.1 shows the precipitated microstructure of the alloys, obtained after aging alloy Al- 0.1Zr-0.1Ti(a) for 1,600 h at 375 C. The nanometer-scale Al 3 (Zr 1 x Ti x ) (L1 2 ) precipitates are nonuniformly distributed throughout the alloy, reflecting the microsegregation of Zr and Ti solute atoms in the as-cast alloys. Both solute species, when alloyed with Al, raise the melting point of the alloy relative to that of pure Al. Consequently, the liquidus and solidus boundaries of the α-al solid-solution have positive slopes and k 0, the equilibrium partition coefficient for solidification, is greater than unity in both systems. The first solid to form during solidification is therefore richer in Zr and Ti compared to the bulk alloy composition, resulting in soluterich dendritic cells surrounded by solute-depleted interdendritic channels. Upon aging, only the enriched dendritic cells contain enough solute in solid-solution to cause precipitation of

94 CHAPTER 4 ATOM-PROBE TOMOGRAPHIC STUDIES OF PRECIPITATION IN AL-0.1ZR-0.1TI (AT.%) ALLOYS Z r C o m p o s itio n (a t.% ) C 0 T i 0.1 C D is ta n c e (µm ) Fig. 4.2: Concentration profiles of Zr and Ti, measured by EDS, across three solute-rich dendrite arms in as-cast Al-0.1Zr-0.1Ti(b). Both Zr and Ti segregate to the dendrite centers (k 0 >1). Al 3 (Zr 1 x Ti x ). The initial distribution of solute was measured in as-cast Al-0.1Zr-0.1Ti(b) by quantitative energy dispersive X-ray spectroscopy (EDS) using a JEOL JSM-7000F scanning electron microscope (SEM) equipped with a ThermoNORAN EDS detector. The composition profile across 35 µm spanning three dendrites, Figure 4.2, demonstrates that the dendrite centers are enriched nearly twofold in Zr and threefold in Ti, compared to their respective bulk compositions (Table 4.1). The interdendritic regions are similarly virtually solute-free. Not only is the spatial distribution of precipitates influenced by the initial inhomogeneous supersaturation of solute atoms after solidification, but there is also a strong effect on precipitate size and morphology, as well as the nucleation mechanism. Consider the precipitates labeled A and B in Figure 4.1(b). Within the center of the dendrites (A), where the Zr and Ti supersaturations are greatest, the precipitates are small (3 5 nm radius), coherent, and are homogeneously distributed. Other precipitates outside the dendrites (B) are larger (about 25 nm radius) and are heterogeneously distributed, indicating that they are most likely heterogeneously nucleated. The nucleation and growth of these precipitates during aging at 375, 400, or 425 C is described

95 SECTION 4.3 EXPERIMENTAL RESULTS 95 in Chapter 5. A typical nm 3 analysis volume for 3-D APT is displayed in Figure 4.1(b). As compared to the relatively coarse (micrometer-scale) dendritic distribution of solute and precipitates, the 3-D volume analyzed by 3-D APT is exceedingly small and therefore the locallymeasured solute composition of the alloy, as well as the likelihood of precipitate interception, varies between 3-D APT analyses. Indeed, it is demonstrated below that the solute concentration measured during an analysis is correlated to the probability of precipitate interception. This also means the local alloy composition from which the Al 3 (Zr 1 x Ti x ) precipitates formed is a priori unknown Three-dimensional atom-probe tomographic reconstructions Figure 4.3 presents a 3-D reconstruction of an Al 3 (Zr 1 x Ti x ) precipitate intercepted during a LEAP tomographic analysis. The reconstructed volume is nm 3, containing identified atoms represented by dots. Figure 4.3(a) displays the Al atoms, with only 25% shown for the sake of clarity. The paucity of atoms in the precipitate interior and the shell of atomic enrichment at the matrix/precipitate interface are reconstruction artifacts, characteristic of a strong local magnification effect arising from disparate evaporation fields between the α-al matrix and Al 3 (Zr 1 x Ti x ) precipitates. This artifact, and its possible influence on the measured precipitate compositions, is discussed below. Figure 4.3(b) shows only the solute (Zr and Ti) atoms, indicating that most of the solute atoms partition to the Al 3 (Zr 1 x Ti x ) phase. While the precipitates are Zr-rich (i.e., primarily Al 3 Zr as discussed below), Figure 4.3(c) shows unambiguously that Ti atoms partition to the precipitate, forming an Al 3 (Zr 1 x Ti x ) intermetallic Quantitative chemical composition of the Al 3 (Zr 1 x Ti x ) precipitates The reconstructions in Figure 4.3 indicate that both Zr and Ti partition to the precipitated Al 3 (Zr 1 x Ti x ) phase. This information is conveyed more quantitatively in Figure 4.4, which exhibits a proximity histogram, or proxigram for short [310, 311], displaying quantitatively the composition of the Al 3 (Zr 1 x Ti x ) precipitate depicted in Figure 4.3. Similar to a linear composition profile, the proxigram is a plot of concentration as a function of distance, indicating the partitioning behavior of solute atoms between the α-al matrix and the Al 3 (Zr 1 x Ti x ) precipitates. Rather than using a single linear direction, however, the proxigram considers each point

96 CHAPTER 4 ATOM-PROBE TOMOGRAPHIC STUDIES OF PRECIPITATION IN AL-0.1ZR-0.1TI (AT.%) ALLOYS 96 Fig. 4.3: LEAP reconstruction of an Al 3 (Zr 1 x Ti x) precipitate in alloy Al-0.1Zr-0.1Ti(b) aged for 400 h at 425 C. (a) Reconstruction showing only Al atoms, with only 25% displayed for the sake of clarity. (b) Reconstruction showing Zr and Ti atoms, indicating that the solute atoms partition to the Al 3 (Zr 1 x Ti x) precipitate. (c) Reconstruction showing only Ti atoms. of 3-D space with respect to a reference surface (a 5 at.% Zr isoconcentration surface), which is physically more representative of the genuine concentration profiles. Hence, a proxigram displays in 2-D information that is characteristic of 3-D. Figure 4.4 indicates that, compared to the nominal value of 25 at.% solute expected for a trialuminide (Al 3 M-type) ordered phase, the precipitate in Figure 4.3 is sub-stoichiometric, containing approximately 20 at.% total solute. Similar sub-stoichiometric compositions have been observed in metastable L1 2 primary (formed in the melt) trialuminides in Al-Ti alloys, with precipitate compositions (determined by EDS) containing approximately 20 at.% Ti, corresponding to Al 4 Ti [170, 210, 214, 215]. 3 This departure from stoichiometry may be attributable to unpositioned solute ions (Zr and Ti) field-evaporated from the precipitate phase. Sha and Cerezo [312] showed that it is necessary to include both correctly-positioned as well as unpositioned ions in order to achieve stoichiometric compositions of Al 3 Zr precipitates formed in a commercial 7050 Al alloy. The precipitates require a much higher electric field for field-evaporation than the α-al solid-solution, discussed below, which generates more multiple ionization events during field-evaporation resulting in higher proportions of these ions not being positioned. Assuming the fraction of unpositioned ions is indiscriminant with respect to solute ion species, the ratio of solutes measured in the Al 3 (Zr 1 x Ti x ) precipitates herein should reflect the correct precipitate composition. Despite the overall substoichiometric precipitate composition, Figure 4.4 demonstrates that Ti is a minor 3 See also Section

97 SECTION 4.3 EXPERIMENTAL RESULTS 97 α α Z r, T i c o m p o s itio n (a t.% ) Z r, T i c o m p o s itio n (a t.% ) D is ta n c e fr o m th e in te r fa c e (n m ) D is ta n c e fr o m th e in te r fa c e (n m ) Fig. 4.4: Proxigram displaying the distribution of Zr and Ti atoms relative to the position of a 5 at.% Zr isoconcentration surface for the precipitate reconstructed in Figure 4.3 (Al-0.1Zr-0.1Ti(b) aged at 425 C for 400 h). The precipitate is Zr-rich, with a small fraction of the Zr lattice sites replaced by Ti atoms. The ratio of Zr:Ti is about 5, corresponding to Al 3 (Zr 0.83 Ti 0.17 ). Fig. 4.5: Proxigrams displaying the distribution of Zr and Ti atoms in Al 3 (Zr 1 x Ti x) precipitates formed in alloy Al-0.1Zr-0.1Ti(a) aged at 375 C for the times indicated. The 20 h aging time results are based on four precipitates and a total of atoms, the results for 400 h are based on 10 precipitates and a total of atoms, and the results for 1,600 h are based on four precipitates and a total of atoms. The ratio of Zr:Ti is about 10, regardless of aging time, corresponding to Al 3 (Zr 0.91 Ti 0.09 ). constituent in the Al 3 (Zr 1 x Ti x ) precipitate, with an average Zr:Ti ratio of 5 (corresponding to x = 0.17) within the precipitate. This is much greater than the overall Zr:Ti ratio of unity in the alloy (Table 4.1). Figure 4.5 shows proxigrams for precipitates formed in Al-0.1Zr-0.1Ti(a) aged for 20, 400, or 1,600 h at 375 C. Compared to Figure 4.4, these concentration profiles are closer to stoichiometry (25 at.% total solute) and, as with Figure 4.4 (425 C aging temperature), Ti is clearly a minor constituent in the Al 3 (Zr 1 x Ti x ) phase precipitated at 375 C. After 20 h at 375 C, the concentration of Ti in the precipitate is 3 4 at.%, comparable to that displayed in Figure 4.4 for Al-0.1Zr- 0.1Ti(b) aged 400 h at 425 CḞigures 4.4 and 4.5 indicate that the composition of the Al 3 (Zr 1 x Ti x ) precipitates varies little with aging temperature (425 C, Figure 4.4), or aging time (20 1,600 h at 375 C, Figure 4.5). At 375 C, the compositions of the Al 3 (Zr 1 x Ti x ) precipitates displayed in Figure 4.5 each have a Zr:Ti ratio of about 10, corresponding to x = 0.09.

98 CHAPTER 4 ATOM-PROBE TOMOGRAPHIC STUDIES OF PRECIPITATION IN AL-0.1ZR-0.1TI (AT.%) ALLOYS Time-of-flight mass spectra Figure 4.6(a) shows the time-of-flight mass spectrum for the entire analysis reconstructed in Figure 4.3. The prominent Al peaks correspond to singly- (Al 1+, 27 amu) and doubly-charged (Al 2+, 13.5 amu) ions produced during pulsed field-evaporation; Al has only one isotope. The peaks for the solute (Zr and Ti) atoms, while less pronounced (a result which is expected, considering the compositions in Table 4.1), are nevertheless clearly identifiable. Zirconium is field-evaporated in both the doubly- (Zr 2+, amu) and triply-charged (Zr 3+, amu) states, whereas Ti atoms are observed only in the doubly-charged (Ti 2+, amu) state. For all species, the stable isotopes are clearly distinguished from one another, demonstrating the good mass resolution of the 3-D APT technique. Compare now the mass spectrum in Figure 4.6(a) to the one displayed in Figure 4.6(b), which represents an analyzed volume from the α-al solid-solution below the precipitate displayed in Figure 4.3. The Zr peaks, which are distinct in Figure 4.6(a), are no longer visible in Figure 4.6(b), indicating that all detectable Zr atoms partition to the Al 3 (Zr 1 x Ti x ) phase. The Ti peaks, by contrast, are still quite prominent in Figure 4.6(b). By counting the number of atoms corresponding to these peaks, the concentration of Ti in α-al solid-solution is determined to be ± at.%, which is somewhat greater than the overall composition of alloy Al-0.1Zr-0.1Ti(b) (Table 4.1). Considering the microsegregation of solute demonstrated in Figure 4.2, this result is not surprising since the as-cast alloys are highly segregated, with some regions significantly enriched in solute and other regions depleted. The compositions in Table 4.1 are bulk chemical compositions whereas those measured by 3-D APT are locally sampled. Furthermore, as shown in the next section, there is a correlation between measured solute concentration in the α-al solid-solution and proximity to precipitates, as anticipated. The mass spectrum in Figure 4.6(c) represents the ions field-evaporated from the core of the Al 3 (Zr 1 x Ti x ) precipitate displayed in Figure 4.3; as anticipated, the Zr and Ti peaks from this solute-rich phase are distinct. There is also a shift in charge states of the field-evaporated ions between the two phases. The spectrum of field-evaporated ions from the α-al solid-solution matrix, Figure 4.6(b), shows that 95% of the Al ions are field-evaporated in the singly-charged state (Al 1+ ), whereas in the Al 3 (Zr 1 x Ti x ) precipitates, Figure 4.6(c), Al is almost entirely doubly-charged (Al 2+ ). These higher charge states are indicative of the larger evaporation fields required to field-evaporate the ions from Al 3 (Zr 1 x Ti x ), leading to a pronounced local magnification effect as discussed

99 SECTION 4.3 EXPERIMENTAL RESULTS (a ) E n tir e a n a ly s is 2 7 A l A l T i T i T i T i T i Z r Z r Z r Z r Z r Z r Z r Z r Z r Z r N u m b e r o f e v e n ts (b ) α-a l s o lid s o lu tio n 2 7 A l T i T i A l (c ) A l 3 (Z r 1 -x,t i x ) p r e c ip ita te 2 7 A l A l Z r T i Z r Z r Z r T i T i Z r Z r Z r T i T i Z r Z r M a s s -to -c h a r g e -s ta te r a tio (a m u ) Fig. 4.6: Atom-probe time-of-flight mass spectrum of Al-0.1Zr-0.1Ti(b) aged for 400 h at 425 C (reconstruction shown in Figure 4.3). (a) Mass spectrum for the entire analysis, representing identified atoms, of which atoms (0.679 ± at.%) are identified as Zr and atoms (0.282 ± at.%) are Ti. (b) Mass spectrum indicating the composition of the matrix (α-al solid-solution). This mass spectrum represents identified atoms, of which 224 atoms (0.107 ± at.%) are Ti; there is no evidence of Zr (<0.010 at.%), indicating that all detectable Zr atoms have partitioned to the Al 3 (Zr 1 x Ti x) precipitate. (c) Mass spectrum indicating the core composition of the Al 3 (Zr 1 x Ti x) precipitate. This spectrum represents atoms, of which 1508 atoms (17.00 ± 0.40 at.%) are Zr and 335 atoms (3.77 ± 0.20 at.%) are Ti. This corresponds to the plateau composition in Figure 4.4. below Correlation between measured Ti concentration proximity to precipitates Considering the initial dendritic distribution of solute species after solidification (Figure 4.2), and the resulting inhomogeneous distribution of Al 3 (Zr 1 x Ti x ) precipitates after aging (Figure 4.1), it is anticipated that 3-D APT analyses be frequently precipitate-free. Indeed, at least half of all performed analyses fail to intercept precipitates but rather sample interdendritic

100 CHAPTER 4 ATOM-PROBE TOMOGRAPHIC STUDIES OF PRECIPITATION IN AL-0.1ZR-0.1TI (AT.%) ALLOYS 100 Table 4.2: Number of precipitates intercepted by 3-D APT, total number of atoms collected, number of Ti atoms collected and resulting Ti concentration of the α-al solid-solution (expressed in terms of the bulk value, C 0 = at.% Ti). Alloy Number of precipitates Total number of atoms ( 10 6 ) Ti atoms C/C 0 Al-0.1Zr-0.1Ti(a), 375 C, 100 h ± 0.07 Al-0.1Zr-0.1Ti(a), 375 C, 100 h ± 0.04 Al-0.1Zr-0.1Ti(a), 375 C, 1,600 h ± 0.03 Al-0.1Zr-0.1Ti(a), 375 C, 1,600 h ± 0.05 Al-0.1Zr-0.1Ti(a), 375 C, 1,600 h ± 0.03 Al-0.1Zr-0.1Ti(b), 425 C, 400 h , ± 0.04 Al-0.1Zr-0.1Ti(a), 375 C, 20 h ± 0.08 Al-0.1Zr-0.1Ti(a), 375 C, 1,600 h ± 0.03 Al-0.1Zr-0.1Ti(a), 375 C, 100 h ± 0.07 Al-0.1Zr-0.1Ti(a), 375 C, 1,600 h ± 0.08 Al-0.1Zr-0.1Ti(a), 375 C, 400 h ± 0.12 Al-0.1Zr-0.1Ti(b), 425 C, 400 h ± 0.07 Al-0.1Zr-0.1Ti(a), 375 C, 20 h ± 0.08 Al-0.1Zr-0.1Ti(a), 375 C, 1,600 h , ± 0.06 Al-0.1Zr-0.1Ti(a), 375 C, 1,600 h , ± 0.08 Al-0.1Zr-0.1Ti(b), 425 C, 400 h , ± 0.10 precipitate-free volumes of α-al solid-solution. These analyses, though precipitate-free, are nevertheless valuable since they provide a direct, atom-by-atom chemical verification of the local solute concentration in the particular volume sampled. As shown in Figure 4.1, the spatial distribution of Al 3 (Zr 1 x Ti x ) precipitates is correlated directly to the initial dendritic distribution of solute atoms formed during solidification. Figure 4.6 indicates that, during aging, nearly all of the Zr atoms partition to the precipitated phase, whereas a substantial fraction of the Ti is observed in the α-al solid-solution around (ca nm 3, a typical volume analyzed by 3-D APT) the precipitates. Since Zr and Ti partition in the same manner (Figure 4.2) there should be a correlation between measured Ti concentration in the α-al solid-solution and proximity to the precipitate-rich dendrites in the microstructure. Table 4.2 compares the results of several LEAP and conventional APT analyses for various aging conditions of the Al-0.1Zr-0.1Ti alloys, arranged in order of increasing measured Ti concentration in the α-al solid-solution. For all but three analyses that intercepted precipitates, Table 4.2, the Ti concentration measured locally in the α-al solid-solution is enriched relative to the overall value of at.% Ti (i.e., C/C 0 > 1).

101 SECTION 4.4 DISCUSSION 101 Similarly, all analyses that do not detect precipitates have smaller Ti concentrations than the nominal value (C/C 0 < 1), with the α-al solid-solutions of some analyses containing undetectable amounts of Ti (C/C 0 0). There appears to be a critical value, between C/C 0 = 0.61 and 0.69 as indicated by the horizontal rule in Table 4.2, which separates the precipitate-free analyses from the precipitate-containing ones. There is one outlier at C/C 0 = 0.39, Table 4.2, which represents an analysis that intercepted a precipitate, but whose surrounding α-al solid-solution is depleted in Ti compared to its overall concentration in the alloy. A portion of the corresponding 3-D reconstruction is shown in Figure 4.7, in which a large (R > 25 nm) Al 3 (Zr 1 x Ti x ) precipitate is intercepted at the end of the analysis. The reconstructed precipitate is comparable in radius to precipitate B in the TEM micrograph in Figure 4.1(b)); the large radius of this precipitate suggests that the analyzed volume is interdendritic, consistent with the depleted measured Ti concentration in the α-al solid-solution (Table 4.2). This hypothesis is supported by the composition, proxigram in Figure 4.8, of the precipitate reconstructed in Figure 4.7. Not only is the matrix depleted in Ti (C/C 0 = 0.39, Table 4.2), but the measured Ti concentration in the Al 3 (Zr 1 x Ti x ) precipitate is also small (ca. 1 at.% Ti) compared to the the values in the proxigrams displayed in Figures 4.4 and 4.5. That there is little Ti in both the α-al solid-solution and precipitate phases suggests that this analysis intercepted an interdendritic precipitate (e.g., B in Figure 4.1(b)). 4.4 Discussion Because of the significant microsegregation of solute species in these alloys (Figure 4.1), each localized region sampled by a given 3-D APT analysis can be thought of as probing a unique (and unknown, a priori) alloy composition. Figure 4.9 shows calculated [313] isothermal sections of the Al-Zr-Ti phase diagram at 375 C (solid lines) or 425 C (dashed lines), which indicates three-phase equilibria among α-al, Al 3 (Zr 1 x Ti x ) (L1 2 ), and Al 3 Ti (D0 22 ) in the Al-rich corner. Superimposed are the data from the EDS linescan in Figure 4.2, showing measured ternary alloy compositions across solute-rich dendritic and solute-depleted interdendritic regions in the alloy. While the local alloy composition varies significantly, Figure 4.9 illustrates that at both temperatures (375 or 425 C), most of the alloy is in the α-al + Al 3 (Zr 1 x Ti x ) (L1 2 ) two-phase

102 CHAPTER 4 ATOM-PROBE TOMOGRAPHIC STUDIES OF PRECIPITATION IN AL-0.1ZR-0.1TI (AT.%) ALLOYS 102 Fig. 4.7: LEAP reconstruction of Al-0.1Zr-0.1Ti(a) aged for 1,600 h at 375 C. The dimensions of the rectangular parallelepiped are nm 3, containing identified atoms. One large (>25 nm radius) Al 3 (Zr 1 x Ti x) precipitate is detected. (a) Reconstruction showing only Al atoms, with only 25% of the atoms displayed for clarity. (b) Reconstruction showing Zr and Ti atoms. (c) Reconstruction showing only Ti atoms. α Z r, T i c o m p o s itio n (a t.% ) D is ta n c e fr o m th e in te r fa c e (n m ) Fig. 4.8: Proxigram showing the distribution of Zr and Ti atoms in the Al 3 (Zr 1 x Ti x) precipitate displayed in Figure 4.7, Al-0.1Zr- 0.1Ti(a) aged at 375 C for 1,600 h. The precipitate is slightly super-stoichiometric (25 30 at.% total solute) with very little Ti (ca. 1 at.%).

103 SECTION 4.4 DISCUSSION α B u lk ra tio o f s o lu te s Z r (a t.% ) α α T i (a t.% ) α Fig. 4.9: Isothermal sections at 375 or 425 C of a calculated ternary phase diagram for the Al-Zr-Ti system in the Al-rich corner, showing the α-al solid-solution in equilibrium with the metastable Al 3 (Zr 1 x Ti x) (L1 2 ) and Al 3 Ti (D0 22 ). The locally-measured alloy compositions (by EDS, Figure 4.2) are indicated by the circles. Phase diagram calculated by Murray [313]. field with a few points (from the interdendritic regions) lying in the single-phase α-al solidsolution. No points are situated in the α-al + Al 3 (Zr 1 x Ti x ) (L1 2 ) + Al 3 Ti (D0 22 ) three-phase triangle. This is consistent with the observed precipitated microstructure (Figure 4.1), which shows only Al 3 (Zr 1 x Ti x ) (L1 2 ) precipitates concentrated in the dendrite centers and precipitate-free α-al solid-solution in the interdendritic channels. The bulk ratio of solute atoms in the alloy is nearly unity (Table 4.1), yet the data in Figures 4.2 and 4.9 show a bias in the measured Zr:Ti ratio of approximately 2:3. This apparent enrichment of Ti in the locally sampled region of the alloy (35 µm linescan, Figure 4.2) could represent a macrosegregation of solutes. It is also possible that this apparent enrichment of Ti is an artifact of the EDS technique. The data presented in Figure 4.2 were not calibrated with respect to known standards, and are therefore strictly semi-quantitative. Nevertheless, the mean alloy composition measured by EDS is comparable to the bulk alloy composition (Figure 4.9), indicating that the EDS results are fairly accurate. The variation in alloy composition, shown in Figures 4.2 and 4.9, makes the discussion of the measured compositions of the Al 3 (Zr 1 x Ti x ) precipitates difficult since the local alloy composition from which these precipitates form is a priori unknown. The situation is complicated further by a significant disparity in evaporation fields between the α-al and Al 3 (Zr 1 x Ti x ) phases,

104 CHAPTER 4 ATOM-PROBE TOMOGRAPHIC STUDIES OF PRECIPITATION IN AL-0.1ZR-0.1TI (AT.%) ALLOYS 104 Fig. 4.10: A 3 nm-thick slice, cut parallel to the analysis direction, through the reconstruction in Figure 4.7. This volume contains identified atoms, with all atoms displayed. resulting in distortions in atomic density in the reconstructed volume, as well as potential biases in the measured precipitate compositions. Nevertheless, consistent with the proxigrams presented in Figures 4.4, 4.5, and 4.8 (as well as others measured), two conclusions are drawn: (i) the precipitated phase is predominantly Zr-rich with a small amount of Ti partitioning to the Al 3 (Zr 1 x Ti x ) precipitates; and (ii) there is no evidence of interfacial segregation of solute at the α-al/al 3 (Zr 1 x Ti x ) heterophase interface Local magnification of the Al 3 (Zr 1 x Ti x ) precipitates Before discussing the measured precipitate compositions, the disparate evaporation fields of the Al 3 (Zr 1 x Ti x ) and α-al phases, and the potential effects on measured solute concentrations in these phases and at their interface, must be discussed. Figures 4.3 and 4.7 exhibit noticeable aberrations in the reconstructed volumes, including: (i) reduced atomic density of the reconstructed precipitates; (ii) a shell of enriched atomic density at the matrix/precipitate interface; and (iii) compression of the precipitate dimension in the analysis direction. These artifacts are attributable to a local magnification effect, which can occur during field-evaporation of twophase systems when the phases have disparate evaporation fields, leading to variations in lo-

105 SECTION 4.4 DISCUSSION 105 cal curvature of the specimen surface, and ultimately aberrations in the trajectories of fieldevaporated ions [296, 298, 314]. These aberrations are not accounted for in the reconstruction procedure, leading to distortions in the reconstructed volumes. The apparent local atomic density in the 3-D APT reconstruction provides an indication of the severity of local focusing or defocusing effects resulting from ion trajectory aberrations [315, 316]. The differences in the reconstructed atomic density of the α-al solid-solution and Al 3 (Zr 1 x Ti x ) precipitates are clear in Figure 4.10, which shows all identified atoms in a thin (3 nm) slice parallel to the analysis direction through the reconstruction in Figure 4.7. The {111}- type atomic planes in the α-al matrix are well resolved and with the correct interplanar spacing, indicative of an accurately reconstructed volume using reconstruction parameters appropriate for the α-al matrix. The same is not true for the reconstructed volume inside the precipitate. The spheroidal precipitates are coherent and coplanar with the α-al matrix (Chapter 5), and therefore the {111} planes should be continuous through the Al 3 (Zr 1 x Ti x ) precipitate. The reconstructed precipitate also exhibits a lower atomic density than the surrounding α-al matrix. At ambient temperature, the lattice parameters of Al and Al 3 Zr are similar, and nm, respectively (Table 1.1), resulting in a 2% difference in molar volume. This small difference would be visually indiscernible in a correctly reconstructed volume; it is obvious in Figure 4.10, however, that the reconstructed local atomic density of the α-al solid-solution is significantly greater than that of the Al 3 (Zr 1 x Ti x ) precipitate, with a measured density ratio of 3.4. The differences in atomic densities between the two phases in the reconstruction of Figure 4.3 is even greater, 5.9. This greater disparity is also consistent with the possibly greater fraction of unpositioned solute species (Zr and Ti) in the Al 3 (Zr 1 x Ti x ) precipitate in explaining the sub-stoichiometric compositions observed in Figure 4.4. The difference in evaporation fields between the α-al matrix and Al 3 (Zr 1 x Ti x ) precipitates is demonstrated quantitatively by considering the mass spectra in Figure 4.6. As discussed, 95% of the detected Al ions are field evaporated as Al 1+ from the α-al solid-solution (Figure 4.6(b)), whereas Al 2+ is predominant from the Al 3 (Zr 1 x Ti x ) precipitates (Figure 4.6(c)), indicative of the higher electric fields required to field-evaporate ions from the Al 3 (Zr 1 x Ti x ) precipitates. This electric field is estimated from the calculated evaporation fields [317] for the constituent species (Al, Zr, and Ti) in the precipitates, Table 4.3. Because the Al ions field-evaporated from the Al 3 (Zr 1 x Ti x ) precipitates are almost all Al 3+, the electric field required to field-evaporate

106 CHAPTER 4 ATOM-PROBE TOMOGRAPHIC STUDIES OF PRECIPITATION IN AL-0.1ZR-0.1TI (AT.%) ALLOYS 106 Table 4.3: Calculated evaporation fields (V nm -1 ) for singly- (E 1 ), doubly- (E 2 ), and triply- (E 3 ) charged Al, Zr, and Ti ions [317]. The expected charge states are underlined. Element E 1 E 2 E 3 Al Zr Ti ions then must exceed 35 V nm -1 (E 3 = 35 V nm -1, Table 4.3). This is also consistent with the observed charge states of Zr ions in the precipitates: 93% of the field-evaporated ions are Zr 3+ (E 3 = 35 V nm -1 ) rather than Zr 2+ (E 2 = 28 V nm -1 ). These results agree with a recent analysis by Sha and Cerezo [312], who used field ion microscopy and 3-D APT to estimate an evaporation field of 36 V nm -1 for Al 3 Zr precipitates in a commercial 7050 Al alloy. The effect of local magnification aberrations on the measured compositions of nanometerscale precipitates by 3-D APT has been addressed both experimentally and with modeling [ , 318]. Miller and Hetherington [318] showed, for a one-dimensional atom-probe, that while trajectory overlap leads to significant spatial distortions in the reconstruction, the bulk compositions of the phases are generally unbiased. Recent simulations by Blavette et al. [314, 316] have similarly concluded that the measured compositions from the inner core of precipitates are unaffected by ion trajectory overlap effects occurring at the reconstructed matrix/precipitate interface. The evaporation field of Al 3 Zr (E Al3 Zr = 36 V nm -1 ) is considerably greater than that of the α- Al solid-solution (E Al = 19 V nm -1 ), as measured by Sha and Cerezo [312] and confirmed herein. The corresponding evaporation field ratio [314], ɛ = E Al3 Zr/E Al = 1.9, represents a greater disparity in evaporation fields than those modeled by Blavette et al. (0.85 ɛ 1.15) [314, 316]. Sha and Cerezo, however, measured directly the extent of ion trajectory overlap for experimental reconstructions and showed that, for Al 3 Zr precipitates ranging from nm in radius, the experimentally-measured ion trajectory overlaps increased from nm with increasing precipitate size. The reconstructed precipitates in Figures 4.3 and 4.7 are larger than this, and so the extent of trajectory overlap (and hence the convolution in measured composition at the interface) may be greater for the precipitates analyzed presently. It is therefore suspected that the precipitate composition profiles presented in Figures 4.4,

107 SECTION 4.4 DISCUSSION , and 4.8 are less accurate than indicated at the matrix/precipitate heterophase interface. Nevertheless, the reported plateau compositions of the proxigrams (and in particular the solute ratios) at the center of the precipitates are thought to be reliable Compositions of the Al 3 (Zr 1 x Ti x ) precipitates The proxigrams in Figures 4.4, 4.5, and 4.8 are representative of the compositions of all the Al 3 (Zr 1 x Ti x ) precipitates observed by both LEAP tomography and conventional APT. The precipitates are invariably Zr-rich, with only a small fraction of Zr replaced by Ti. In all cases, the absolute Ti concentration in the trialuminide is less than 5 at.%, despite the nominally equal concentrations of Zr and Ti initially in the alloys (Table 4.1). The proxigrams shown in Figure 4.5, which indicate the temporal evolution of the Al 3 (Zr 1 x Ti x ) precipitates at 375 C, suggest that the Zr:Ti ratio varies little with time and is approximately 10. The proxigram in Figure 4.4, for a precipitate formed after 400 h of aging at 425 C, demonstrates that the Zr:Ti ratio is approximately 5, roughly twice that at 375 C, corresponding to a precipitate composition of Al 3 (Zr 0.83 Ti 0.17 ). Like the composition of the alloy from which they form (Figure 4.2), however, the concentrations of solutes in the precipitates are probably not uniform throughout the volume. Indeed, as shown in Figure 4.8, the measured Zr:Ti ratio can deviate substantially from those stated above. The small concentration of Ti observed in the precipitates could be reflective of: (i) the substantially higher solid-solubility of Ti in α-al; and (ii) the smaller diffusivity of Ti in α-al compared to that of Zr. Figure 4.9 indicates that the solid-solubility of Al 3 Zr (L1 2 ) in metastable equilibrium with α-al is and at.% Zr at 375 and 425 C, respectively. The solid-solubility of Al 3 Ti (D0 22 ) in α-al is an order of magnitude greater: and at.% Ti at 375 and 425 C, respectively. This greater solubility of Ti would seemingly account for the small partitioning of Ti to the Al 3 (Zr 1 x Ti x ) precipitates and the residual Ti observed in α-al solid-solution in precipitate-containing 3-D APT analyses (Table 4.2). The calculated ternary Al-Zr-Ti phase diagram (Figure 4.9), however, indicates that both solutes are supersaturated with respect to the Al 3 (Zr 1 x Ti x ) solvus, particularly in the solute-rich dendrites. Thus, the small Ti concentration in the Al 3 (Zr 1 x Ti x ) precipitates, as well as the residual Ti observed in α-al solid-solution around them, is not simply a function of the equilibrium solubilities of Zr and Ti. The measured tracer diffusivity of Ti in α-al is smaller than that of Zr in α-al (Table 1.3). Therefore kinetics are probably the dominant influence dictating the observed small concentra-

108 CHAPTER 4 ATOM-PROBE TOMOGRAPHIC STUDIES OF PRECIPITATION IN AL-0.1ZR-0.1TI (AT.%) ALLOYS 108 tion of Ti in the Al 3 (Zr 1 x Ti x ) precipitates. The diffusivities of Zr and Ti in α-al are given by an Arrhenius relationship, D = D 0 exp ( Q/R g T ), where Q = 242 and 260 kj mol -1 and D 0 = and m 2 s -1 for Zr and Ti, respectively (Table 1.3). The root-mean-squared (RMS) diffusion distances, 6Dt, of Zr and Ti after 1,600 h at 375 C are 279 and 65 nm, respectively; after 400 h at 425 C these distances are 698 and 184 nm, respectively. These calculated RMS diffusion distances are much smaller than the dimensions of the initial dendritic distribution of solute (0.5 4 µm in Figure 4.1 and 3 10 µm in Figure 4.2). Therefore, achieving a homogenous distribution of Ti is impossible for reasonable aging treatments. Even at 375 C, however, the calculated RMS diffusion distance of Ti is greater than the mean edge-to-edge inter-precipitate distance (ca nm) in the precipitate-rich dendrite centers (Figure 4.1). The diffusion distance of Ti is also greater than that of a typical 3-D APT reconstruction (ca nm 3 as shown schematically in Figure 4.1). The growth of the precipitates may therefore be interface limited and not diffusion-limited. Additionally, diffusion of a solute in an ordered compound is slower than in the matrix because of correlation effects on the diffusivity. There is only one other study where the composition of Al 3 (Zr 1 x Ti x ) precipitates was measured directly. Using scanning transmission electron microscopy and energy dispersive X-ray spectroscopy, Sato et al. [271] measured the composition of Al 3 (Zr 1 x Ti x ) precipitates formed after aging an Al-0.09Zr-0.17Ti (at.%) for 168 h at 500 C. Despite the enhanced kinetics at 500 C and the greater supersaturation of Ti, the estimated Zr:Ti ratio in the precipitates was approximately 10, consistent with our findings that only a small fraction of the available Ti partitions to the Al 3 (Zr 1 x Ti x ) phase. Our results agree also with other 3-D APT studies on Al-Sc-Ti [118] and Al-Sc-Zr [116, 319, 320] alloys, discussed below, where the slower-diffusing Ti and Zr atoms are generally a minor constituent in the precipitated trialuminide Residual Ti in the α-al solid-solution Table 4.2 demonstrates that the amount of Ti in α-al observed locally by 3-D APT can be correlated to the proximity of the 3-D APT analysis to the precipitate-rich and therefore solute-rich regions in the alloy. This is indeed illustrative of how little Ti partitions to the Al 3 (Zr 1 x Ti x ) precipitate during aging at both 375 and 425 C. That the local Ti concentration in α-al solidsolution is sometimes enriched (C/C 0 > 1) and other times depleted (C/C 0 < 1) relative to the bulk composition is reflective of the initial inhomogeneous distribution of solute atoms shown

109 SECTION 4.4 DISCUSSION 109 quantitatively in Figure 4.2. In the as-cast alloy the concentration of Ti, as measured by EDS, is shown to be enriched nearly threefold relative to the bulk alloy composition within the supersaturated dendrites. This is in reasonable agreement with the over twofold enrichment of Ti in solid-solution near precipitates measured by 3-D APT (Table 4.2). That there is only a twofold enrichment of Ti in α-al in the aged alloys may be explained in part by the fact that Ti atoms partition to the precipitated Al 3 (Zr 1 x Ti x ) phase, as shown in the mass spectra of Figure 4.6 and the proxigrams in Figures 4.4, 4.5, and 4.8, and hence Ti is depleted from the α-al solid-solution. As discussed, the solid-solubility of Al 3 Ti (D0 22 ) in α-al is or at.% Ti at 375 or 425 C, respectively, which correspond to the Ti composition of the point in the α-al + Al 3 (Zr 1 x Ti x ) (L1 2 ) + Al 3 Ti (D0 22 ) tie triangle in Figure 4.9. Since all regions of the alloy are in the α-al solid-solution or the α-al + Al 3 (Zr 1 x Ti x ) (L1 2 ) two-phase field (Figure 4.9), any tie line drawn between these two phases must intercept the α-al solvus at less than at.% Ti at 375 C or at.% Ti at 425 C. Table 4.2 shows that the measured Ti concentration in α-al solid-solution for three precipitate-containing analyses performed on Al-0.1Zr-0.1Ti(a) aged at 375 C contain more than at.% Ti, the equilibrium maximum solubility at this temperature. Similarly, the α-al solid-solution from one precipitate-containing analysis performed on Al-0.1Zr-0.1Ti(b) aged 400 h at 425 C contained ± at.% Ti (C/C 0 = 2.15 ± 0.10), well beyond the maximum solubility of at.% Ti at 425 C. The fact that concentrations beyond the equilibrium solubility are measured in some precipitate-containing 3-D APT analyses is indicative that the alloys have not yet reached global equilibrium after 1,600 h at 375 C or 400 h at 425 C. As described in the Introduction, improving the thermal stability of Al 3 Zr with ternary or quaternary additions of Ti and/or V has generated considerable interest in the literature [84, , 264, 271, ]. Most of these studies, however, assumed that the ratio of solutes in the precipitated trialuminides was the same as that initially in solid-solution (i.e., the bulk alloy composition). This is at first glance a reasonable assumption since the transition elements are generally sparingly soluble in α-al and all solute atoms should therefore partition to the precipitated phase. In the present study the alloys had initially the same atomic fraction of solute species (Table 4.1), yet this Zr:Ti ratio of unity is never observed in the precipitated Al 3 (Zr 1 x Ti x ) phase. Moreover, as evidenced by Table 4.2, the assumption of negligible solubility of solutes in α-al is invalid; diffusion kinetics in α-al, which is anomalously slow and varies widely among the

110 CHAPTER 4 ATOM-PROBE TOMOGRAPHIC STUDIES OF PRECIPITATION IN AL-0.1ZR-0.1TI (AT.%) ALLOYS 110 transition elements (Section 1.2.3), is instead the dominant factor dictating the relative amounts of solute atoms in the precipitated phase. The precipitate compositions cited in previous studies [84, , 264, ] are therefore highly speculative (Sato et al. [271], however, verified the compositions of their Al 3 (Zr 1 x Ti x ) preicpitates by EDS, as noted) Comparison to 3-D APT studies on Al-Sc-Ti/Zr alloys It is valuable to compare our measured precipitate compositions to those for Al-0.06Sc-0.06Ti (at.%) alloys reported in a similar 3-D APT study by van Dalen et al. [118]. Like Al 3 Zr, the ordered Al 3 Sc phase precipitates from a supersaturated solid-solution during aging, and also has the L1 2 structure; unlike Al 3 Zr, however, Al 3 Sc is a thermodynamically stable L1 2 ordered structure. Even after aging at 300 C for 1,536 h (64 days), van Dalen et al. [118] showed that Ti atoms partition weakly to the Al 3 (Sc 1 x Ti x ) phase, forming precipitates with the average composition Al 3 (Sc 0.94 Ti 0.06 ), corresponding to a Sc:Ti ratio of 16, with Ti remaining primarily in α-al solidsolution. The amount of Ti measured in α-al solid-solution by 3-D APT decreased continuously with aging time at 300 C, yet still exceeded the calculated equilibrium value in α-al after 1,536 h, indicating that equilibrium had not yet been reached. The higher coarsening resistance of Al 3 Zr compared to Al 3 Sc, due to the slower diffusion kinetics of Zr (Section 1.2.3), allows us to age at higher temperatures (up to 425 C versus 300 C for the Al-Sc-Ti alloys [118]). This amounts to a significant difference in diffusion kinetics, with the calculated diffusivity of Ti being times greater at 425 C than at 300 C. Accordingly, the proxigram shown in Figure 4.4 indicates that the composition of a precipitate formed after 400 h at 425 C is Al 3 (Zr 0.83 Ti 0.17 ). Figure 4.5, for precipitates formed at 375 C, indicates compositions of Al 3 (Zr 0.91 Ti 0.09 ). These both represent a larger fraction of Ti compared to Al 3 (Sc 0.94 Ti 0.06 ) observed after 1,536 h at 300 C [118]. This may be explained in terms of diffusion kinetics: the calculated RMS diffusion distance for Ti in α-al after 1,536 h at 300 C is 3 nm; and after 20 h at 375 C (corresponding to one of the proxigrams in Figure 4.5) the calculated RMS diffusion distance of Ti is 7 nm. The distribution of Ti in the Al 3 (Sc 1 x Ti x ) precipitates studied by van Dalen et al. [118] is inhomogeneous, with the precipitates consisting of an Al 3 Sc core surrounded by a spherical Tienriched outer shell. Similar cored composite precipitates are well-documented for Al-Sc-Zr alloys, as observed by 3-D APT [116,319,320], analytical high resolution electron microscopy [321],

111 SECTION 4.5 CHAPTER SUMMARY 111 and lattice kinetic Monte Carlo simulations [322]. Both Ti and Zr reduce the lattice parameter of Al 3 Sc (L1 2 ) [120, 121], thus reducing the lattice parameter mismatch with α-al. The enriched outer shell of the mismatch-reducing solutes (Ti or Zr) within the precipitate should therefore minimize the elastic strain energy of the coherent precipitate [116,118,320,321]. It has also been suggested [116, ] that, because of the much faster diffusion rate of Sc compared to Ti or Zr, only Sc atoms are kinetically able to participate in the early stages of nucleation and growth. These pre-existing clusters then act as heterogeneous nucleation sites for the less-mobile Ti or Zr atoms, accounting for the observed cored precipitate compositions. In the Al-Zr-Ti system, Ti is the slowest diffuser (Section 1.2.3) and has also been shown to reduce the lattice parameter mismatch with α-al of the metastable Al 3 Zr (L1 2 ) phase [289]. By analogy with the Al-Sc-Ti or Al-Sc-Zr systems, therefore, one might expect segregation of Ti to the matrix/precipitate interface. Nevertheless, our results provide no evidence for such segregation at the α-al/al 3 (Zr 1 x Ti x ) interface after extended aging at 375 or 425 C. The lattice parameter mismatch of Al 3 Zr (L1 2 ) with α-al at room temperature, δ = +0.75%, is less than that of Al 3 Sc, δ = +1.32% (Table 1.1), which means the elastic strain energy (proportional to the square of the mismatch [323,324]) of the coherent heterophase interface is also less. There is also the possibility that interfacial segregation of Ti may be veiled by the significant local magnification for this system, as discussed. 4.5 Chapter Summary Atom-probe tomography was applied to measure directly the chemical compositions of Al 3 (Zr 1 x Ti x ) (L1 2 structure) precipitates formed in conventionally-solidified Al-0.1Zr-0.1Ti (at.%) alloys during extended aging times at 375 C (to 1,600 h) or 425 C (to 400 h). The following results were obtained and discussed: Dendritic solidification results in non-uniform distributions of Zr and Ti, which segregate to the dendrite centers and are enriched by a factor of 2 and 3, respectively. The interdendritic regions are concurrently solute-depleted. During subsequent aging, Al 3 (Zr 1 x Ti x ) precipitates are similarly non-uniformly distributed in a dendritic manner. The Zr:Ti atomic ratio in the Al 3 (Zr 1 x Ti x ) precipitates, as measured by atom-probe tomography, is 10 (x = 0.09) at 375 C and 5 (x = 0.17) at 425 C, indicating that Ti partitions

112 CHAPTER 4 ATOM-PROBE TOMOGRAPHIC STUDIES OF PRECIPITATION IN AL-0.1ZR-0.1TI (AT.%) ALLOYS 112 weakly to the Al 3 (Zr 1 x Ti x ) phase. The precipitate compositions vary little with aging time. While Ti unambiguously partitions to the Al 3 (Zr 1 x Ti x ) precipitate, residual Ti is observed in analyzed volumes ( nm 3 ) of the α-al solid-solution near to the Al 3 (Zr 1 x Ti x ) precipitates, which is beyond the calculated solubility. The small concentration of Ti in the Al 3 (Zr 1 x Ti x ) precipitates, as well as the enriched Ti concentration beyond the calculated solubility in α-al, indicates that local equilibrium has not been achieved after 1,600 h at 375 C or 400 h at 425 C. This is consistent also with 3-D APT studies on Al-Sc-Ti alloys aged at 300 C [118]. There is a correlation between measured Ti concentration in α-al solid-solution (within the volume of a typical 3-D APT analysis, ca nm 3 ) and proximity of the analysis to the precipitate-rich dendrites, reflective of the initial non-uniform distribution of solute atoms. There is a significant disparity in evaporation fields between the α-al matrix (19 V nm -1 ) and Al 3 (Zr 1 x Ti x ) precipitates (36 V nm -1 ), leading to trajectory aberrations in the fieldevaporated ions. Ion trajectory overlap at the reconstructed matrix/precipitate heterophase interface means the measured composition at this interface is uncertain. The measured composition in the center of the precipitates should be unaffected by trajectory overlap and are therefore reliable.

113 113 CHAPTER 5 Microstructure and Mechanical Properties of Al-Zr and Al-Zr-Ti Alloys during Isothermal Aging at 375, 400, or 425 C Precipitation of Al 3Zr (L1 2) and Al 3(Zr 1 xti x) (L1 2) is investigated in conventionallysolidified Al-0.1Zr, Al-0.2Zr, Al-0.1Zr-0.1Ti, and Al-0.2Zr-0.2Ti (at.%) alloys isothermally aged at 375, 400, or 425 C. Pronounced hardening accompanies precipitation of Al 3Zr or Al 3(Zr 1 xti x) for all alloys investigated. The magnitude of the peak-aged hardness is controlled primarily by the Zr concentration. The alloys overage sluggishly, with no benefit from Ti additions in delaying overaging. After extended aging times (up to 1,600 h) at 425 C, there is no difference in the mean precipitate radii of Al 3Zr (L1 2) or Al 3(Zr 1 xti x) (L1 2) precipitates, indicating that there is no benefit in terms of coarsening resistance from ternary alloying additions of Ti. Previous coarsening studies on Al 3(Zr 1 xv x) (L1 2), Al 3(Zr 1 xti x) (L1 2), and Al 3(Zr 1 x y, V x, Ti y) (L1 2) precipitates at 425 C are reviewed and discussed, and are compared to the present results. 5.1 Introduction AS SHOWN IN PREVIOUS CHAPTERS, THE AL-ZR SYSTEM shows particular promise for developing thermally-stable precipitation-strengthened Al alloys. In this system, an ordered trialuminide, Al 3 Zr, precipitates during aging from supersaturated solid-solution. While the equilibrium structure of Al 3 Zr is tetragonal (D0 23 ), decomposition of supersaturated Al-Zr solidsolutions occurs initially by the formation of nanometer-scale Al 3 Zr precipitates with a metastable cubic L1 2 structure, structurally and chemically analogous to the Ni 3 Al (γ ) phase in the Ni-based superalloys, which are thermally stable at high homologous temperatures (Chapter 1). Fine has long argued [1, 37, 38, 84] that even greater stability can be achieved by improving the lattice parameter mismatch between Al 3 Zr and the α-al solid-solution, and showed that the lattice parameters of both the stable D0 23 and metastable L1 2 Al 3 Zr precipitates are reduced by additions of Ti, Hf, or V [84, 110, 111]. The Al-Zr-V system, in particular, was studied exten-

114 CHAPTER 5 MICROSTRUCTURE OF AL-ZR AND AL-ZR-TI ALLOYS DURING AGING AT 375, 400, OR 425 C 114 sively [84, ]; by substituting V for Zr, Al 3 (Zr 1 x V x ) (L1 2 ) precipitates were shown to exhibit reduced coarsening compared with that of the binary Al 3 Zr precipitates [84, 113, 114]. The improved stability of the Al 3 (Zr 1 x V x ) precipitates proved to be effective barriers to dislocation motion during creep at 425 C [ ]. However, the formation of an additional Al 10 V phase, particularly at grain boundaries, deteriorated the mechanical properties [112, 113]. Titanium also reduces the lattice parameter mismatch of Al 3 Zr (L1 2 ) with α-al [289] and, since Al 3 Ti is the terminal intermetallic compound (Figure 1.4, Table 1.2), the problem of additional embrittling intermetallics is mitigated. Furthermore, an initial investigation by Parameswaran et al. [285] showed that an Al-Zr-Ti alloy containing 1 vol.% Al 3 (Zr 0.75 Ti 0.25 ) exhibited much improved coarsening resistance at 425 C compared to Al-Zr and Al-Zr-V alloys of similar precipitate volume fractions [84]. The present study is an extension of the one by Parameswaran et al. [285]. The Al 3 (Zr 1 x Ti x ) precipitate compositions, measured by 3-D atom-probe tomography, are discussed in Chapter 4. This chapter reports on the effects of Ti additions to dilute Al-Zr alloys by examining their ambient temperature mechanical properties, utilizing Vickers microhardness measurements, and correlating these results to their microstructure during isothermal aging at 375, 400, and 425 C. The effect of Ti additions on the Al 3 (Zr 1 x Ti x ) (L1 2 ) coarsening rates at 425 C is also assessed. In Chapter 6, similar alloys are studied at temperatures in excess of 450 C, where the L1 2 to D0 23 transformation is discussed. Elevated-temperature mechanical properties, in the form of creep properties, are discussed in Chapter Experimental Procedures A series of binary Al-Zr and ternary Al-Zr-Ti alloys were investigated; alloy designations, compositions, and aging conditions are summarized in Table 5.1. The alloys were prepared employing non-consumable electrode arc-melting, as described in Sections 3.2 and 4.2. The verified compositions in Table 5.1 were obtained by bulk chemical analysis performed by direct current plasma emission spectroscopy (ATI Wah Chang, Albany, OR). As-cast specimens were isothermally aged at 375, 400, or 425 C in air, within the range of temperatures shown to produce a strong precipitation strengthening response for Al-Zr alloys produced by RSP [14, 264, ] or chill-casting [187, 190, 271]. The microstructures obtained after aging were investigated

115 SECTION 5.3 EXPERIMENTAL RESULTS 115 Table 5.1: Compositions and aging conditions of the Al-Zr and Al-Zr-Ti alloys investigated. Nominal comp. (at.%) Verified comp. (at.%) Aging temperatures ( C) Zr Ti Zr Ti Al-0.1Zr(a) Al-0.1Zr(b) , 400, 425 Al-0.2Zr , 400, 425 Al-0.1Zr-0.1Ti(a) Al-0.1Zr-0.1Ti(b) , 400, 425 Al-0.2Zr-0.2Ti , 400, 425 in both alloys by transmission electron microscopy (TEM) and scanning electron microscopy (SEM), using similar instruments and techniques described in Section Experimental Results As-cast microstructure Figure 5.1 shows the as-cast macrostructure of the alloys studied. The solidification macrostructure is typical of cast alloys, with coarse columnar grains, originating at the bottom surface of the ingot (which was in contact with the chilled copper crucible of the arc-meler), growing upward toward a zone of equiaxed grains near the center of the ingot. The relative sizes of the columnar and equiaxed zones is strongly dependent on the solute content of the alloy and, more precisely, the extent of properitectic Al 3 M (M = Zr or Zr,Ti) precipitation, since these primary phases are potent grain refiners, as discussed previously. Alloy Al-0.1Zr(b) is the most dilute of those studied (Table 5.1), and the extent of the columnar zone, which comprises the entire ingot cross-section in this alloy, is also greatest. The grains in the columnar zone are large, mm wide, and ca. 5 mm long. The effect of doubling the solute concentration with an addition of 0.1 at.% Ti is demonstrated for Al-0.1Zr-0.1Ti(b). For this alloy, the upper half of the ingot exhibits a zone of coarse ( mm diameter) equiaxed grains. Alloy Al-0.2Zr exhibits a finer structure still, where the grains in the columnar zone are typically 0.2 mm wide and the extent of this zone is also less. Consequently, the majority of the ingot is equiaxed, with finer grain sizes ( mm) than for alloy Al-0.1Zr(b). Alloy Al-0.2Zr-

116 CHAPTER 5 MICROSTRUCTURE OF AL-ZR AND AL-ZR-TI ALLOYS DURING AGING AT 375, 400, OR 425 C 116 Fig. 5.1: Macrostructure of as-cast alloys Al-0.1Zr(b), Al-0.1Zr-0.1Ti(b), Al-0.2Zr, and Al-0.2Zr-0.2Ti (etched using Poultan s reagent), showing various degrees of grain refinement. 0.2Ti, the most concentrated alloy studied (Table 5.1), exhibits the finest grain size; the entire ingot cross-section is equiaxed with very fine grains ca. 100 µm diameter. Figures 5.2 and 5.3 display SEM micrographs of metallographically-polished Al-0.2Zr and Al- 0.2Zr-0.2Ti, respectively. Both alloys exhibit petal-like precipitates of the properitectic phase (Al 3 Zr or Al 3 (Zr 1 x Ti x )), responsible for the grain refinement observed in Figure 5.1. This petallike morphology is characteristic of the metastable L1 2 structure [160], whose cubic structure is commensurate with fcc α-al and acts as an effective heterogeneous nucleant of α-al during solidification. The presence of these primary precipitates indicates also that, for the conventional casting conditions used in this study, exceeding ca. 0.2 at.% solute (Ti or Zr) results in primary precipitation of the properitectic trialuminide and hence decreases the amount of solute retained in solid-solution. No primary precipitates are observed by SEM in the more-dilute Al-0.1Zr and Al-0.1Zr-0.1Ti alloys, consistent with the coarser grain structure in Figure 5.1. Composition of the Al 3 (Zr 1 x Ti x ) primary precipitates The composition of one of the primary Al 3 (Zr 1 x Ti x ) precipitates in alloy Al-0.2Zr-0.2Ti (Figure 5.3) was measured using quantitative energy-dispersive spectrometry (EDS) in SEM, using a

117 SECTION 5.3 EXPERIMENTAL RESULTS 117 Fig. 5.2: SEM secondary electron images of as-solidified Al-0.2Zr. Panel (a) displays the ingot top. Z-contrast delineates Zr-rich dendrite arms (lighter contrast). Panel (b) displays the ingot bottom. Several petal-like primary Al 3 Zr precipitates (light contrast) are observed. Fig. 5.3: SEM secondary electron images of as-solidified Al-0.2Zr-0.2Ti. Numerous Al 3 (Zr 1 x Ti x) primary precipitates are observed. Panel (a) displays the ingot top. Numerous Al 3 (Zr 1 x Ti x) primary precipitates are observed. Panel (b) displays the ingot bottom. Fewer primary precipitates are observed. In Figures 5.2 and 5.3 solute-enriched dendritic cells are also visible. Hitachi S-3500 operated at 20 kv and equipped with an Oxford Instruments EDS system. X-ray spectra were acquired for 10 s at a constant sample working distance of 15 mm. The beam was used at maximum focusing (i.e. with a diameter of a few nm) with a beam current of 0.66 na, calibrated with an internal Faraday cup. The recorded spectra were compared quantitatively to previously-recorded pure reference standards generated under identical beam conditions. Figure 5.4 shows the analyzed properitectic Al 3 (Zr 1 x Ti x ) precipitate with the approximate

118 CHAPTER 5 MICROSTRUCTURE OF AL-ZR AND AL-ZR-TI ALLOYS DURING AGING AT 375, 400, OR 425 C 118 Table 5.2: Quantitative EDS analysis of the properitectic precipitate of Figure 5.4 in as-cast Al-0.2Zr-0.2Ti. Position EDS compositions (at.%) Ti:Zr Zr Ti Total Fig. 5.4: SEM micrograph of a primary Al 3 (Zr 1 x Ti x) precipitate seen in Figure 5.3(a) for Al-0.2Zr-0.2Ti. Quantitative EDS measurements, reported in Table 5.2, were performed at the positions indicated. positions of the four EDS measurements. The measured compositions are displayed in Table 5.2. While the data are quantitatively accurate, the compositions are not reflective of the primary phase since the detected X-rays by EDS originate from an interaction volume (ca. 1 µm 3 ) that includes the underlying α-al solid-solution, explaining why the total solute content is less than the 25 at.% expected for the primary trialuminide. Nevertheless, the measured concentrations in Table 5.2 show unambiguously that both Ti and Zr segregate to the primary phase, and that these precipitates are slightly Ti-rich with a Ti:Zr ratio of approximately 1.5. The halo surrounding the primary phase in Figure 5.4 may be a sub-surface projection of the properitectic precipitate. This could also represent solute-enriched α-al solid-solution enveloping the primary phase, since the first solid to form is nucleated heterogeneously onto the Al 3 (Zr 1 x Ti x ) precipitate. Quantitative EDS measurements, position 4 in Table 5.2, indicate that this halo is enriched in both Zr and Ti, supporting both hypotheses. It is shown in Figures 5.17 and 5.18 that solid-state spheroidal Al 3 (Zr 1 x Ti x ) (L1 2 ) precipitates nucleate in regions surrounding the petal-like primary phases during post-solidification aging, suggesting that the halo in Figure 5.4 indicates an enrichment in solute concentration in α-al solid-solution Solute segregation and the precipitated microstructure In addition to the primary precipitates described above, Figures 5.2 and 5.3 reveal also the soluteenriched dendrites in the as-solidified alloys. The dendritic regions exhibit lighter contrast, in-

119 SECTION 5.3 EXPERIMENTAL RESULTS Zr 0.8 Zr Composition (at.%) C 0 Ti Composition (at.%) C 0 Ti 0.1 C C Distance (µm) Distance (µm) (a) (b) Fig. 5.5: Concentration profiles of Zr and Ti, measured by EDS, across three solute-rich dendrite arms in as-cast ternary Al-Zr-Ti alloys: (a) Al-0.1Zr-0.1Ti(b); (b): Al-0.2Zr-0.2Ti. Both Zr and Ti segregate to the dendrite centers (k 0 >1). dicative of the higher solute concentration in α-al solid-solution (Z-contrast from backscattered electrons). This concentration variation between the dendritic and interdendritic regions was quantified by EDS using a JEOL JSM-7000F SEM equipped with a ThermoNORAN EDS detector (procedure discussed previously, Section 4.2). Measured composition profiles spanning three dendrite branches are displayed in Figure 5.5 for alloys Al-0.1Zr-0.1Ti(b) and Al-0.2Zr-0.2Ti. The two composition profiles in Figure 5.5 are similar in that the dendrite centers are enriched approximately twofold in Zr and threefold in Ti compared to their respective bulk compositions (Table 5.1). The interdendritic regions are concurrently solute-depleted. The mean concentrations of Zr and Ti in Figure 5.5(a) are 0.10 at.% Zr and 0.14 at.% Ti; in Figure 5.5(b) these values are 0.28 at.% Zr and 0.40 at.% Ti. It is noteworthy that both composition profiles exhibit a bias in the measured Ti:Zr ratio of approximately 1.4 (as noted also in Figure 4.9). This apparent enrichment of Ti in the locally sampled region of the alloy (ca. 10 µm) could represent a macrosegregation of solutes. The fact that both composition profiles exhibit a similar bias, however, suggests that the apparent enrichment of Ti is an artifact of the EDS technique. Unlike the data in Table 5.2, the data in Figure 5.5 were not calibrated with respect to known standards, and are therefore strictly semi-quantitative. Nevertheless, the mean alloy composition measured by EDS is comparable to the bulk alloy composition, C 0 (especially true

120 CHAPTER 5 MICROSTRUCTURE OF AL-ZR AND AL-ZR-TI ALLOYS DURING AGING AT 375, 400, OR 425 C 120 Fig. 5.6: TEM and SEM micrographs of Al-0.1Zr-0.1Ti(a) aged at 375 C for 1,600 h. Panels (a) and (b) are centered dark-field TEM images showing the dendritic distribution of nanometer-scale Al 3 (Zr, Ti) (L1 2 ) precipitates. Panels (c) through (f) are SEM secondary electron images, which illustrate the inhomogenous nature of the dendritic distribution.

121 SECTION 5.3 EXPERIMENTAL RESULTS 121 for Zr), indicating that the EDS results are reasonably accurate. During post-solidification aging, nanometer-scale Al 3 Zr (L1 2 ) or Al 3 (Zr 1 x Ti x ) (L1 2 ) precipitates form, non-uniformly distributed throughout the alloy, reflecting the dendritic microsegregation of Zr and Ti solute atoms in the as-cast alloys. Figure 5.6 shows the precipitated microstructure of the alloys, obtained after aging alloy Al-0.1Zr-0.1Ti(a) at 375 C for 1,600 h Age hardening Figures display the observed microhardness of the Al-Zr and Al-Zr-Ti alloys after isothermal aging at 375, 400, or 425 C. Each data point represents a minimum of 20 measurements, with the standard deviation of these measurements indicated by the error bars. The position of the measurement within the ingot (i.e., ingot top, center, bottom) was also recorded since the cooling rate of the melt and therefore the amount of solute retained in solid-solution probably varies with proximity to the chilled copper crucible. A positional-dependence on hardness was observed only in Al-0.2Zr-0.2Ti, where the ingot bottom is consistently 70 MPa harder than the rest of the alloy. This increased hardness is probably a result of suppressed properitectic precipitation of Al 3 (Zr 1 x Ti x ) near the ingot bottom, as shown in Figure 5.3. To minimize the effect of solute loss to this primary phase, therefore, only hardness data measured near the bottom surface of alloy Al-0.2Zr-0.2Ti are reported in Figure There is generally no significant difference in the observed hardness between the Al-Zr and Al-Zr-Ti alloys for a given Zr concentration (0.1 or 0.2 at.%) and aging temperature. For all alloys investigated, the peak hardness decreases with increasing aging temperature, attributable to the natural reduction in volume fraction of the dispersed phase due to the temperature-dependence of solute solid-solubility. There is a particularly pronounced drop in plateau hardness between aging at 400 and 425 C in the alloys containing 0.1 at.% Zr (Figures 5.7 and 5.9), suggesting a significant decrease in volume fraction of the Al 3 Zr (L1 2 ) or Al 3 (Zr 1 x Ti x ) (L1 2 ) precipitates. The reduction in hardness at 425 C is, however, less for Al-0.1Zr-0.1Ti(b), suggesting that Ti additions increase the volume fraction of L1 2 precipitates, which is particularly noticeable at higher temperatures. 1 Differences in age hardening behavior are even less distinct between Al-0.2Zr and Al-0.2Zr- 1 In Chapter 4 it is shown by 3-D APT that Ti partitions to the precipitates, forming Al 3(Zr 1 xti x), and hence an increase in volume fraction is not unexpected. However, the effect is small since the measured Zr:Ti atomic ratio is 5 10, indicating that Ti partitions weakly to the precipitated phase.

122 CHAPTER 5 MICROSTRUCTURE OF AL-ZR AND AL-ZR-TI ALLOYS DURING AGING AT 375, 400, OR 425 C A l-0.1 Z r A l-0.1 Z r -0.1 T i 8 0 V ic k e r s m ic r o h a r d n e s s (M P a ) V ic k e r s m ic r o h a r d n e s s (H V ) V ic k e r s m ic r o h a r d n e s s (M P a ) V ic k e r s m ic r o h a r d n e s s (H V ) A g in g tim e (h ) A g in g tim e (h ) Fig. 5.7: Vickers microhardness vs. aging time for Al- 0.1Zr(a) and Al-0.1Zr(b) at 375, 400, or 425 C. Fig. 5.9: Vickers microhardness vs. aging time for Al- 0.1Zr-0.1Ti(a) and Al-0.1Zr-0.1Ti(b) at 375, 400, or 425 C A l-0.2 Z r A l-0.2 Z r -0.2 T i 8 0 V ic k e r s m ic r o h a r d n e s s (M P a ) V ic k e r s m ic r o h a r d n e s s (H V ) V ic k e r s m ic r o h a r d n e s s (M P a ) V ic k e r s m ic r o h a r d n e s s (H V ) A g in g tim e (h ) A g in g tim e (h ) Fig. 5.8: Vickers microhardness vs. aging time for Al-0.2Zr at 375, 400, or 425 C. Fig. 5.10: Vickers microhardness vs. aging time for Al- 0.2Zr-0.2Ti at 375, 400, or 425 C. 0.2Ti. These alloys, however, contain primary (properitectic) Al 3 Zr or Al 3 (Zr 1 x Ti x ) precipitates, Figures 5.2 and 5.3, which are especially numerous in alloy Al-0.2Zr-0.2Ti. The solute retained in α-al solid-solution is therefore less than the bulk concentrations indicated in Table 5.1, and the overall volume fraction of dispersed phases precipitated during post-solidification aging is similarly unknown. Table 5.2 demonstrates that both Ti and Zr segregate to the primary phase in alloy Al-0.2Zr-0.2Ti. Moreover, these precipitates are more numerous in alloy Al-0.2Zr-0.2Ti than in Al-0.2Zr, as demonstrated by the pronounced grain refinement in Figure 5.1 and also shown in Figure 5.3(a). The supersaturation of Zr retained in α-al solid-solution in alloy Al-

123 SECTION 5.3 EXPERIMENTAL RESULTS Zr-0.2Ti may indeed be less than that in Al-0.2Zr, and the similar hardness values for these alloys, Figures 5.8 and 5.10, may indicate a similar benefit from Ti additions as that shown for Al-0.1Zr-0.1Ti(b), Figure 5.7. In addition to the noticeable increase in peak strength at 425 C, Ti additions appear to also accelerate the onset of precipitation hardening in the more-dilute alloys containing 0.1 at.% Zr (Figures 5.7 and 5.9). According to classical nucleation theory, the incubation time for nucleation [ , 325] is inversely proportional to the square of the chemical driving force, which is dependent on the solute supersaturation (Eq.(3.5)). The increased hardness at higher temperatures and the accelerated incubation time for nucleation in Al-0.1Zr-0.1Ti(b) compared to Al-0.1Zr(b) are indicative of a larger supersaturation of solute. The incubation time varies also with aging temperature. For Al-0.2Zr and Al-0.2Zr-0.2Ti, Figures 5.8 and 5.10, the incubation time decreases with increasing aging temperature. The exact opposite effect is observed for Al-0.1Zr-0.1Ti(b), Figure 5.9, where the onset of hardening is noticeably accelerated compared to aging at 400 or 425 C. Finally, for alloy Al-0.1Zr(b), Figure 5.7, there is little effect of aging temperature on the incubation time. The hardness values for short aging times (t < 6.5 h) in Figures were secured with rather small specimens from various positions in the ingot cross section. The inconsistent variations in incubation time may therefore reflect slight variations in solute concentration due to different solidification rates with respect to the location in the ingot. A more detailed study focussing on the early stages of the decomposition of these alloys during isothermal aging, accompanied also with electrical conductivity measurements, would be beneficial Precipitate morphologies The SEM and TEM micrographs in Figure 5.6 suggest that the precipitated microstructure is binary in nature, with fine, spheroidal, coherent Al 3 Zr (L1 2 ) or Al 3 (Zr 1 x Ti x ) (L1 2 ) precipitates confined to the dendritic cells, which are surrounded by precipitate-free interdendritic channels. In reality, the situation is not so simple due to complicated solute concentration gradients around the dendrites. Figure 5.11 reveals the precipitated microstructure in an interdendritic channel between two precipitate-rich dendrite arms, produced in alloy Al-0.2Zr-0.2Ti after aging at 400 C for 100 h (peak hardness, Figure 5.10). In any precipitation reaction, the precipitate size decreases with increasing solute content (for a given aging temperature) since the chemical

124 CHAPTER 5 MICROSTRUCTURE OF AL-ZR AND AL-ZR-TI ALLOYS DURING AGING AT 375, 400, OR 425 C 124 Fig. 5.11: Centered superlattice dark-field TEM micrograph of an interdendritic channel between two dendrite arms in alloy Al- 0.2Zr-0.2Ti aged at 425 C for 100 h. The foil edge is on the right-hand side of the micrograph, with increasing specimen thickness towards the left-hand side. Within the dendrites themselves, the Al 3 (Zr 1 x Ti x) precipitates are small and homogeneously distributed. Because the foil is thicker in these regions (due to differential thinning), the small (<10 nm) precipitates are clearly resolved only near the foil edge. There is a gradient in the size and number density of the larger interdendritic precipitates, becoming smaller and more numerous with proximity to the dendrite stalk (towards the left-hand side of the figure). Fig. 5.12: Centered superlattice dark-field TEM micrographs of heterogeneously-nucleated spheroidal L1 2 interdendritic precipitates. Panel (a) displays Al-0.2Zr-0.2Ti aged at 425 C for 100 h (detail of interdendritic region in Figure 5.11). Panel (b) displays Al-0.2Zr aged at 400 C for 100 h. The precipitate-rich dendritic regions are visible in the upper right and lower left.

125 SECTION 5.3 EXPERIMENTAL RESULTS 125 Fig. 5.13: Development of growth instabilities with increasing precipitate radius for interdendritic Al 3 (Zr 1 x Ti x) precipitates in Al-0.1Zr-0.1Ti(b) aged at 425 C for 400 h. driving force for precipitation increases with supersaturation. This leads to a smaller critical radius for nucleation and hence smaller precipitates. Within the center of the dendrites, where the solute supersaturation is largest (Figure 5.5), precipitates are small and homogeneouslydistributed at high number densities. The supersaturation decays with lateral position from the center of the dendritic cells and correspondingly, the precipitates become progressively larger. The interdendritic regions contain insufficient solute to effect homogeneous nucleation so these regions are largely precipitate-free. Precipitates that form here do so heterogeneously, Figure 5.12, as evidenced by their arrangement in linear arrays. These precipitates are likely formed on dislocations since the strain energy barrier for precipitation via this mechanism is significantly reduced [326, 327]. Cauliflower-shaped interdendritic precipitates In the interdendritic regions, where the spheroidal L1 2 precipitates are largest and their volume fraction is smallest, cauliflower-like morphologies are frequently observed, Figure These arise from instabilities during the growth of the Al 3 Zr or Al 3 (Zr 1 x Ti x ) precipitates, and similar morphologies have been observed for Al 3 Sc (L1 2 ) [117, 119, 328] and Al 3 Li (L1 2 ) [329] precipitates when nucleated under small supersaturations. Because of their small number density, the interdendritic precipitates in the present alloys are small enough for them to grow in a super-

126 CHAPTER 5 MICROSTRUCTURE OF AL-ZR AND AL-ZR-TI ALLOYS DURING AGING AT 375, 400, OR 425 C 126 saturated matrix before their diffusion fields commence overlapping. The morphology is then determined by the growth conditions rather than the equilibrium conditions. The gradients in precipitate radius that develop between the dendritic and interdendritic regions are not uniform. Figure 5.14(a c) are SEM micrographs of alloy Al-0.2Zr aged at 425 C for 400 h, and Figure 5.14(d f) are TEM micrographs of alloy Al-0.2Zr-0.2Ti aged at 425 C for 400 h. The TEM micrograph in Figure 5.14(d) demonstrates that the size gradients occurring on either side of the same interdendritic channel may vary considerably. On the left-hand side, Figure 5.14(f), the change in precipitate radius is gradual, similar to that shown in Figure On the right-hand side of the precipitate-free channel, Figure 5.14(d), the gradient in precipitate size is abrupt. These gradients, occurring over the micrometer-scale, are best visualized by SEM, Figure 5.14(a), which shows similar coexistence of sharp and broad size gradients across a precipitate-free interdendritic channel. The various gradients in precipitate radius probably reflect the initial distribution of solute species, and are thus a function of the critical radius for nucleation as discussed above. Loss of coherency in Al 3 Zr and Al 3 (Zr 1 x Ti x ) precipitates The gradual precipitate size gradients occurring in the interdendritic channels provides an opportunity to assess the change in coherency of the Al 3 Zr and Al 3 (Zr 1 x Ti x ) precipitates. Figure 5.15 displays three pairs of complementary dark-field and bright-field TEM micrographs recorded under dynamical two-beam conditions. Under such imaging conditions, coherent precipitates observed in bright-field exhibit characteristic Ashby-Brown [330] strain contrast, as shown for the precipitates near the center of Figure 5.15(b) exhibiting distinct lines of no contrast normal to the diffraction vector, g. For slightly larger interdendritic precipitates, coherency strain contrast is lost and striped lines of contrast normal to g appear in the images of many of the precipitates. These are interpreted as contrast due to interfacial dislocations, thus signaling the transition to semicoherency of the Al 3 Zr or Al 3 (Zr 1 x Ti x ) precipitates. Figures 5.15(d) and 5.15(f) illustrate other examples of this phenomenon in different alloys aged at different temperatures. Iwamura and Miura [331] show similar transitions in strain contrast in loss of coherency for Al 3 Sc (L1 2 ) precipitates. Coherency across a precipitate-matrix heterophase interface is generally lost when the pre-

127 SECTION 5.3 EXPERIMENTAL RESULTS 127 Fig. 5.14: SEM micrographs of Al-0.2Zr, Panels (a) through (c), and TEM micrographs of alloy Al-0.2Zr-0.2Ti, Panels (d) through (f), aged at 425 C for 400 h. Gradients in precipitate radius vary considerably throughout the microstructure, even across the same interdendritic channel. Panels (c) and (f) illustrate the development of growth instabilities, and demonstrate that the microstructures of the alloys are indistinguishable.

128 CHAPTER 5 MICROSTRUCTURE OF AL-ZR AND AL-ZR-TI ALLOYS DURING AGING AT 375, 400, OR 425 C 128 Fig. 5.15: Complementary dark-field and bright-field TEM micrographs showing loss of coherency in the interdendritic precipitate size gradients. Panels (a) and (b) display Al-0.1Zr(b) aged at 425 C for 1,600 h. Loss of coherency occurs at R = 35 nm. Panels (c) and (d) display Al-0.2Zr aged at 425 C for 400 h. Loss of coherency occurs at R = 31 nm. Panels (e) and (f) display Al-0.1Zr-0.1Ti(a) aged at 375 C for 1,600 h. Loss of coherency occurs at R = 25 nm.

129 SECTION 5.3 EXPERIMENTAL RESULTS 129 cipitate size exceeds a critical value, which typically corresponds to n = 1 δ, (5.1) where n is the number of atomic planes in the matrix and δ is the lattice parameter mismatch between the two phases. The radius of the semicoherent precipitates indicated by the arrows in Figures 5.15(b), 5.15(d), and 5.15(f) are 35 nm (for Al-0.1Zr(b) aged at 425 C for 1,600 h), 31 nm (for Al-0.2Zr aged at 425 C for 400 h), and 25 nm (for Al-0.1Zr-0.1Ti(a) aged at 375 C for 1,600 h), respectively. These precipitate radii are slightly larger than in Al 3 Sc precipitates, which retain coherency to approximately 20 nm radius [43, 119, 331, 332]. This disparity may be understood considering the reported lattice parameters of the L1 2 Al 3 Zr and Al 3 Sc phases, which are 4.08 Å [79, 94, 105] and Å [103], respectively. A coherent Al 3 Zr (L1 2 ) precipitate has a smaller mismatch (δ = +0.75%) than that of Al 3 Sc (δ = +1.33%) and, by Eq.(5.1), should therefore retain coherency to larger radii. Marquis and Seidman [119] argued that, for an Al 3 Sc-α-Al heterophase interface Eq.(5.1) predicts n = 80 planes which, considering a 0.2 nm spacing between {200} planes, corresponds to a distance of 16 nm, which is in good agreement with the observed coherency loss for Al 3 Sc precipitates of 20 nm radii. Using similar arguments, a metastable L1 2 Al 3 Zr precipitate should be expected to remain coherent to n = 130 planes, corresponding to radii of 26 nm. This is in reasonable agreement with the observed coherency loss in Figure As discussed by Royset and Ryum [43, 332] for Al 3 Sc (L1 2 ) precipitates, the critical radius for coherency loss is actually temperature dependent. Since the thermal expansion is different for the matrix and precipitate phases, coherency is expected to be maintained for larger precipitate radii at higher temperatures. It is interesting that for the specimens aged at 425 C, Figures 5.15(b) and 5.15(d), the observed loss of coherency occurs at larger radii (R = 35 nm and 31 nm, respectively) than in the one aged at 375 C, Figure 5.15(f) (R =25 nm). Rod-like and plate-like interdendritic L1 2 precipitates A second precipitate morphology observed in the interdendritic channels is that of orthogonal rods and plates oriented in 100 matrix directions and having the metastable L1 2 structure, Figure Both rod-like and plate-like morphologies are observed, as confirmed by tilting experiments. A characteristic feature of these precipitates is the appearance of closely-spaced pairs of

130 CHAPTER 5 MICROSTRUCTURE OF AL-ZR AND AL-ZR-TI ALLOYS DURING AGING AT 375, 400, OR 425 C 130 Fig. 5.16: TEM micrographs of rod-shaped interdendritic Al 3 Zr or Al 3 (Zr 1 x Ti x) precipitates, oriented in 100 matrix directions. Panel (a) is a centered dark-field image, [100] zone axis, of precipitate-rich dendrites in Al-0.1Zr-0.1Ti(a) aged at 375 C for 1,600 h. Panel (b) is a detailed view of closely-spaced pair of rod-like precipitates shown in Panel (a). Panel (c) is a bright-field image, [111] zone axis, of Al-0.1Zr(a) aged at 375 C for 1,600 h. Arrays of spheroidal precipitates are also visible. Panel (d) displays a centered superlattice dark-field image of several rod-like and plate-like Al 3 (Zr 1 x Ti x) (L1 2 ) precipitates. Panel (e) is a bright-field image, [110] zone axis, of Al-0.1Zr(b) aged at 425 C for 1,600 h. Panel (f) is a centered superlattice dark-field image of rod-like Al 3 Zr (L1 2 ) precipitates and arrays of spheroidal Al 3 Zr (L1 2 ) precipitates.

131 SECTION 5.3 EXPERIMENTAL RESULTS 131 rods, Figure 5.16(b). These are identical to those described by Nes [79] in a hypoperitectic Al-Zr alloy (0.05 at.% Zr) aged at 460 C. Ryum [85] observed similar orthogonal rods and plates during the decomposition of Al-0.027Hf (at.%) alloys after aging at 450 C for 50 h. In the present alloys, these rod-shaped precipitates are predominantly in the interdendritic regions where the solute concentration is smallest. Heterogeneous nucleation dominates in these regions, as discussed, and Nes [79] observed that precipitation of the rod-like and platelike precipitate morphologies is closely associated with matrix dislocations, proposing that a precipitate growth/dislocation climb interaction is responsible for their appearance. Indeed, a dislocation is seen protruding into the matrix from both ends of the rod-shaped precipitate in the lower right of Figure 5.16(e). Near to the rod- and plate-shaped precipitates are linear arrays of spheroidal interdendritic L1 2 precipitates, Figures 5.16(c), 5.16(d), and 5.16(f). Nes [79] observed similar arrays of Al 3 Zr (L1 2 ) precipitates which were described as rods partly chopped up into rows of spherical particles. As with the rods, the arrays of spheroidal precipitates were aligned along 100 directions. The arrays observed presently seem to be aligned in particular orientations, Figures 5.16(c) and 5.16(f), but the orientation relationship with α-al is not strict. Zedalis and Fine [84] observed similar arrays of Al 3 Zr (L1 2 ) precipitates in Al-0.22Zr (at.%) alloys aged at 450 C, which they attributed to coalescence of spheroidal precipitates. Precipitation in supersaturated α-al enveloping primary (properitectic) precipitates In the more concentrated Al-0.2Zr and Al-0.2Zr-0.2Ti alloys, primary Al 3 Zr or Al 3 (Zr 1 x Ti x ) precipitates (Figures 5.2 and 5.3) act as heterogeneous nucleants during solidification of α-al. Since k 0 > 1, α-al solid-solution enveloping the primary phase is enriched in solute, as shown by EDS in the halo surrounding the primary Al 3 (Zr 1 x Ti x ) precipitate in Figure 5.4. This phenomenon is also observed in aged specimens of alloy Al-0.2Zr-0.2Ti, Figures 5.17 and 5.18, where solidstate precipitation of Al 3 (Zr 1 x Ti x ) (L1 2 ) is observed in α-al surrounding the petal-like primary phase. Figure 5.17 shows SEM micrographs of electropolished TEM foils of alloy Al-0.2Zr-0.2Ti aged at 425 C for 400 h. The origin of the grain refinement observed in Figure 5.1 is apparent in Figures 5.17(a) and 5.17(b), where most grains contain a petal-like primary Al 3 (Zr 1 x Ti x ) precipitate. Closer inspection of these primary phases, Figures 5.17(c) 5.17(f), demonstrate that

132 CHAPTER 5 MICROSTRUCTURE OF AL-ZR AND AL-ZR-TI ALLOYS DURING AGING AT 375, 400, OR 425 C 132 Fig. 5.17: SEM secondary electron images of primary Al 3 (Zr 1 x Ti x) precipitates in alloy Al-0.2Zr-0.2Ti aged at 425 C for 400 h. Panels (a) and (b) display numerous petal-like primary precipitates, similar to those shown in Figure 5.3(a), and the accompanying grain refining effect due to heterogeneous nucleation of α-al during solidification. Panels (c) through (f) reveal the solid-state precipitation of Al 3 (Zr 1 x Ti x) (L1 2 ) in supersaturated α-al surrounding the primary precipitates. Gradients in size and number density vary of the solid-state precipitates varies with initial solute concentration, Panel (f), as discussed previously.

133 SECTION 5.3 EXPERIMENTAL RESULTS 133 Fig. 5.18: Centered superlattice dark-field TEM micrographs demonstrating solid-state precipitation of spheroidal Al3(Zr1 xtix) (L12) in supersaturated α-al solid-solution surrounding a petal-like Al3(Zr1 xtix) (L12) primary precipitate in Al-0.2Zr aged at 425 C for 400 h. [100] zone axis imaged with the (001) L12 superlattice reflection. Panel (a) displays SEM secondary electron images of a similar Al3(Zr1 xtix) primary precipitate visible in an electropolished TEM foil. Panel (b) is a TEM micrograph of dendritic filaments at the periphery of the primary precipitate. Panel (c) displays homogeneously-distributed spheroidal Al3(Zr1 xtix) (L12) precipitates nucleated during post-solidification aging. In Panel (d) precipitates have a mean radius R = 6.3 ± 1.9 nm, Table 5.3, and also exhibit antiphase boundaries. Panel (e) demonstrates that towards the periphery, the interdendritic precipitates develop growth instabilities, shown in detail in Panel (f), and described previously (Figures 5.13 and 5.14).

134 CHAPTER 5 MICROSTRUCTURE OF AL-ZR AND AL-ZR-TI ALLOYS DURING AGING AT 375, 400, OR 425 C 134 the α-al solid-solution enveloping them is initially highly supersaturated, as evidenced by a concentric layer of homogeneously-distributed nanometer-scale spheroidal Al 3 (Zr 1 x Ti x ) (L1 2 ) precipitates surrounding the petal-like precipitate. Here again we observe similar gradients in precipitate size and morphology as described in Figures 5.13 and Consider Figure 5.17(f). Near to the primary precipitate where the first α-al solid-solution was nucleated during solidification, the solute supersaturation is greatest (k 0 >1) and the precipitates are fine and homogeneously distributed in high number densities. The solute supersaturation decreases outward from the primary precipitate periphery and the solid-state spheroidal precipitates are coarser and in smaller number densities. These variations in precipitate morphologies are illustrated also in the series of superlattice dark-field TEM micrographs in Figure As shown in Figure 5.18, the primary petal-like precipitates have the metastable L1 2 structure, as evidenced by their bright contrast when imaged with an L1 2 superlattice reflection. This observation is not unexpected since the petal-like morphology is characteristic of the metastable L1 2 structure [160], as discussed previously and also in Section Figure 5.18 demonstrates also that the primary precipitates and the spheroidal ones precipitated in solid-solution around them share the same orientation relationship with the α-al matrix, indicating that α-al nucleates epitaxially on the metastable L1 2 primary precipitate. The primary precipitates themselves consist of fine dendritic filaments, Figures 5.17(f) and 5.18, reflective of growth instabilities during their growth in the melt. This observed morphology is consistent with what has been observed for Al 3 Ti, Al 3 Zr, and Al 3 Hf metastable L1 2 primary precipitates, as discussed in Section Small dendritic precipitates The most prevalent precipitate morphology in terms of number density, and certainly the one exerting the most influence on the observed mechanical properties, is the large population of small, coherent, homogeneously-distributed spheroidal Al 3 Zr (L1 2 ) or Al 3 (Zr 1 x Ti x ) (L1 2 ) precipitates found in the precipitate-rich dendritic cells (e.g., Figure 5.6). An example of these precipitates is shown in Figure 5.18(d), where the mean radius, R, of Al 3 (Zr 1 x Ti x ) (L1 2 ) precipitates is 6.3 ± 1.9 nm after aging at 425 C for 400 h. This is impressive coarsening resistance indeed, but Al 3 (Zr 1 x Ti x ) (L1 2 ) precipitates have been shown [285] to exhibit dramaticallyreduced precipitate radii compared to binary Al 3 Zr (L1 2 ). How different are the precipitate radii

135 SECTION 5.3 EXPERIMENTAL RESULTS 135 Fig. 5.19: Centered superlattice dark-field TEM micrographs of Al-0.1Zr(b) (Panels (a) and (b)) and Al-0.1Zr-0.1Ti(b) (Panels (c) and (d)) aged at 425 C for 1600 h. Images obtained with a low-index L1 2 superlattice reflection as indicated in the inset diffraction patterns. Panel (a) displays an interdendritic channel separating two precipitate-rich dendritic regions. Panel (b) demonstrates that, within the dendrites, the Al 3 Zr (L1 2 ) precipitates have a mean radius of R = 10.9 ± 1.9 nm. Panel (c) shows precipitate-rich dendritic regions separated by a 4 5 µm wide precipitate-lean interdendritic channel in Al-0.1Zr-0.1Ti(b). L1 2 -structured orthogonal rods or plates are observed in the channel. The dark spots are holes in the foil, remnant of large spheroidal interdendritic precipitates dislodged during sample preparation. Panel (d) displays Al 3 (Zr 1 x Ti x) (L1 2 ) precipitates within the dendrites, R = 11.6 ± 2.3 nm, showing no improvement in coarsening resistance as compared to Al 3 Zr (L1 2 ) (Panel (b)). for Al 3 Zr (L1 2 ) under identical aging conditions? Such a comparison is illustrated in Figure 5.19, which shows the precipitated microstructures formed in alloys Al-0.1Zr(b) and Al-0.1Zr-0.1Ti(b) after aging at 425 C for 1,600 h. 2 Despite the inhomogeneities occurring throughout both alloys, within the dendrites themselves the alloys are indistinguishable: R = 10.9 ± 1.9 nm for Al 3 Zr 2 It is more appropriate to use the dilute Al-0.1Zr and Al-0.1Zr-0.1Ti alloys for a direct comparison since both alloys are coarse-grained, Figure 5.1, and are free of primary precipitates.

136 CHAPTER 5 MICROSTRUCTURE OF AL-ZR AND AL-ZR-TI ALLOYS DURING AGING AT 375, 400, OR 425 C 136 Table 5.3: Mean precipitate radii, R, observed in the dendrite centers from six different Al-Zr and Al-Zr-Ti (three of each) specimens after extended aging times at 425 C. Alloy Aging treatment Mean precipitate radius, R (nm) Number of precipitates counted Al-0.1Zr(b) 1,600 h at 425 C 10.9 ± Al-0.1Zr-0.1Ti(b) 1,600 h at 425 C 11.6 ± Al-0.2Zr 400 h at 425 C 5.1 ± ± Al-0.2Zr-0.2Ti 400 h at 425 C 6.3 ± ± and R = 11.6 ± 2.3 nm for Al 3 (Zr 1 x Ti x ). Table 5.3 summarizes results of these and other precipitate size analyses on more concentrated Al-0.2Zr and Al-0.2Zr-0.2Ti alloys, which indicate also that the coarsening rates of Al 3 Zr and Al 3 (Zr 1 x Ti x ) precipitates in these alloys are identical after extended aging times at 425 C (0.75T m ). The similarity in dendritic precipitate size between the Al-Zr and Al-Zr-Ti alloys after extending aging times is also reflected in the comparable hardness of the Al-Zr and Al-Zr-Ti alloys in Figures At 400 C and below, the alloys exhibit identical precipitation hardening behavior after extended aging times. While the Al-Zr-Ti alloys tend to be harder at shorter aging times or at temperatures greater than 400 C, this effect may be interpreted in terms of solute supersaturation influencing the incubation for nucleation and the volume fraction of precipitate-rich dendritic regions, and is not an effect of precipitate radius or coarsening resistance. The effect of Ti additions is discussed more thoroughly in Chapter 6, Section The data in Table 5.3 illustrate a second point. The two mean precipitate sizes reported for Al- 0.2Zr and Al-0.2Zr-0.2Ti represent different physical specimens from each alloy. In other words, these mean radii represent the precipitate populations in two different dendrites. The vast inhomogeneities in precipitate size, growth, and morphology (e.g., Figures 5.14 and 5.16) are not observed in the dendrites, and the consistency in the observed mean radii suggests that the solute supersaturation (and hence the critical radius for nucleation as well as the local volume fraction) is more-or-less constant from one dendrite to the next. This contention is supported also by the measured EDS linescans, Figure 5.5, which indicate that the solute concentration is similar for all dendritic regions.

137 SECTION 5.4 DISCUSSION 137 Transformation to the equilibrium D0 23 structure Despite extended aging times (up to 1,600 h) at 425 C (0.75T m ), equilibrium D0 23 -structured are not observed by TEM. Indeed, as discussed in Chapter 6, Al 3 Zr (L1 2 ) is kinetically stable at high homologous temperatures, with significant coarsening and transformation to the D0 23 structure commencing at ca. 500 C (0.83T m ), much higher than the temperatures investigated presently. It is important to point out, however, that the precipitates in Figure 5.18(d) exhibit structural faults, characterized by sharp lines of no contrast parallel to {100} planes inside the L1 2 precipitates. The faults are attributed to antiphase boundaries (APBs) generated in the transition to an imperfect tetragonal (D0 23 ) structure by the formation of an APB with a displacement vector a/2 110 on {100} type planes, and hence represent the early stages of the transformation to the D0 23 structure. This L1 2 to D0 23 transformation is discussed in more detail in Chapter Discussion Comparison to Previous Studies on Al-Zr-V/Ti Alloys As described in Section 4.1 and the Introduction to this chapter, improving the coarsening resistance of Al 3 Zr with ternary or quaternary additions of V and/or Ti has generated considerable interest in the literature. These studies are summarized in Table 5.4, which indicates the alloy compositions, volume fractions and purported compositions 3 of the L1 2 precipitates, and their measured coarsening rates at 425 C. The measured precipitate radii of the L1 2 -structured Al 3 (Zr 1 x M x ) (M = V and/or Ti) precipitates as a function of aging time at 425 C are displayed in Figure Anomalous behavior in arc-melted alloys An immediately apparent observation in Figure 5.20 is that the studies on arc-melted alloys [84, 285, 333] (i.e., those containing 1 vol.% precipitates see also Table 5.4) tend to exhibit larger precipitates with faster coarsening rates than the subsequent studies on melt-spun alloys [113, 114, 286, , 334]. Indeed, several researchers [113, 286, 294, 295] noted that the precipitates they studied in melt-spun Al-Zr-V and Al-Zr-Ti-V alloys were consistently smaller than those 3 As pointed out in Section 4.1, with exception to reference [271], the precipitate compositions were not measured directly in the studies in Table 5.4.

138 CHAPTER 5 MICROSTRUCTURE OF AL-ZR AND AL-ZR-TI ALLOYS DURING AGING AT 375, 400, OR 425 C 138 Table 5.4: Reported coarsening rates of L12 Al3Zr-based precipitates in Al-Zr, Al-Zr-Ti, and Al-Zr-Ti-V alloys aged at 425 C Study Alloy comps. (at.%) Method Cubic L12 phase Nom. vol. frac. Pre-age Rate constant, k Refs Zr Ti V (%) (m 3 h -1 ) Studies on Al-Zr alloys Zedalis and Fine (1986) 0.24 Arc-melted Al3Zr 1 vol.% None [84, 333] Studies on Al-Zr-V alloys Zedalis and Fine (1986) a Arc-melted Al3(V0.875Zr0.125) 1 vol.% None [84, 333] Arc-melted Al3(V0.725Zr0.275) 1 vol.% 500 C1 h Chen et al.( ) Melt-spun Al3(V0.75Zr0.25) 5 vol.% 500 C2.5 h [113, 114] Han et al.(1995) Melt-spun Al3(V0.675Zr0.325) 8 vol.% 500 C1 h [286] Studies on Al-Zr-Ti alloys Parameswaran et al.(1989) b Arc-melted Al3(Zr0.75Ti0.25) 1 vol.% 500 C1 h [285] a Tsau and Chen (2002) Melt-spun Al3(Zr0.75Ti0.25) 5 vol.% 500 C1 h [293] Melt-spun Al3(Zr0.25Ti0.75) 5 vol.% 500 C1 h [293] Studies on Al-Zr-Ti-V alloys Lee et al.(1993) Melt-spun Al3(Zr0.50V0.40Ti0.10) 10 vol.% 500 C1 h [294] Melt-spun Al3(Zr0.40V0.40Ti0.20) 10 vol.% 500 C1 h Chung et al.(1995) Melt-spun Al3(Zr0.5V0.4Ti0.1) 8 vol.% 500 C1 h [295] Han et al.(1995) Melt-spun Al3(Zr0.595V0.25Ti0.155) 8 vol.% 500 C1 h [286] a Based on reported data [333], the coarsening rate constant k for the Al 3(V0.725Zr0.275) precipitates is actually , not m 3 h 1, as reported in [84]. b Rate constant not reported in their study. Reported to be m 3 h 1 in [294], presumably obtained by linearly fitting plotted data in [285].

139 SECTION 5.4 DISCUSSION v o l.% A l 3 Z r Z e d a lis a n d F in e ( ) 1 v o l.% A l 3 (V Z r ) Z e d a lis a n d F in e ( ) M e a n p r e c ip ita te r a d iu s (n m ) v o l.% A l 3 (V Z r ) Z e d a lis a n d F in e ( ) 1 v o l.% A l 3 (Z r.7 5 T i.2 5 ) P a ra m e s w a ra n e t a l. ( ) 1 0 v o l.% A l 3 (Z r.4 0 V.4 0 T i.2 0 ) 5 v o l.% A l 3 (Z r.7 5 T i.2 5 ) L e e e t a l. ( ) T s a u a n d C h e n ( ) 1 0 v o l.% A l 3 (Z r.5 0 V.4 0 T i.1 0 ) L e e e t a l. ( ) 5 v o l.% A l 3 (T i.7 5 Z r.2 5 ) 5 v o l.% A l 3 (V.7 5 Z r.2 5 ) T s a u a n d C h e n ( ) C h e n e t a l. ( ) 8 v o l.% A l 3 (Z r.5 V.4 T i.1 ) C h u n g e t a l. ( ) 8 v o l.% A l 3 (Z r V.2 5 T i ) 8 v o l.% A l 3 (V Z r ) H a n e t a l. ( ) H a n e t a l. ( ) Fig. 5.20: Comparison of L1 2 -structured Al 3 (Zr 1 x M x) (M = V and/or Ti) average precipitate radius ( R, in nm) with aging time at 425 C. Open symbols represent studies by Fine and coworkers on Al-Zr [84, 333], Al-Zr-V [84, 113, 114, 333, 334], and Al-Zr-Ti [285] alloys. Shaded symbols are studies by Lee et al.on Al-Zr-V [286] and Al-Zr-Ti-V alloys [286, 294, 295], and half-shaded symbols are from a study by Tsau and Chen on Al-Zr-Ti alloys [293]. Specific alloy compositions and reported coarsening rates indicated in Table 5.4. observed previously by Fine and coworkers on Al-Zr, Al-Zr-V and Al-Zr-Ti alloys produced by arc-melting. The effect of finite volume fraction of precipitates on coarsening rate, well-studied theoretically and experimentally [36, 335, 336], is always increasing coarsening rate with increasing volume fraction. The apparent violation of this rule further supported the belief that ternary additions of V and/or Ti had dramatically improved the coarsening resistance of Al 3 Zr, in spite of the higher volume fractions in the melt-spun studies. Lee and colleagues [286, 294, 295] attributed the discrepancies in coarsening behavior between the initial arc-melted studies of Zedalis and Fine [84, 333] and those that followed to differences in the pre-aging treatment prior to aging. 4 They contended that the pre-aging treat- 4 This is one of the reasons why the specimens were not pre-aged in the present study. Pre-aging was also found empirically to be detrimental to achievable hardness, Figure B.1.

140 CHAPTER 5 MICROSTRUCTURE OF AL-ZR AND AL-ZR-TI ALLOYS DURING AGING AT 375, 400, OR 425 C 140 ment has substantially decreased the amount of supersaturation and slowed down the growth kinetics of the precipitates. A more probable explanation for the discrepancies in Figure 5.20 is that the initial studies on Al-Zr and Al-Zr-V alloys by Zedalis and Fine [84,333] were biased by an inhomogeneous dendritic ditribution of precipitates. Indeed, Zedalis and Fine [84] acknowledged that the density of precipitates is not uniform but varies somewhat from place to place in their alloys, and likened the microstructure to that reported in a study by Izumi and Oelschlägel [76] who observed Al 3 Zr (L1 2 ) precipitates statistically distributed within the dendritic cells of the cast structure. Furthermore, Zedalis and Fine [84] themselves questioned the validity of their measured coarsening rates, stating that their calculated value interfacial free energy, σ, driving coarsening is a hundred times larger or more than expected, and suspected that the nonuniform distribution of precipitates may be playing a role for this anomalous result. The (i) anomalously fast coarsening and (ii) anomalously large precipitates observed in the initial studies by Zedalis and Fine [84, 333] can be easily explained within the context of a dendritic distribution of precipitates. Consider the TEM micrographs in Figure What is the mean radius of the Al 3 Zr precipitates? The most apparent population is the ca. 20 nm spheroidal Al 3 Zr (L1 2 ) precipitates observed in the interdendritic regions of the specimen, where the foil is thinnest. The most numerous population is the ca. 5 nm precipitates in the dendrites, which are not obvious because of their smaller size and the much-reduced contrast from absorption (incoherent scattering) occurring in the thicker dendritic regions. The precipitate radii Zedalis and Fine measured, R = nm (Figure 5.20), suggest that the precipitates they studied were interdendritic. 5 These tend to be larger, as discussed above, because of an initially smaller solute supersaturation leading to a larger critical radius for nucleation. Moreover, as suggested by Robson [337], their smaller number density allows them to grow to larger radii before exhausting their surrounding α-al solid-solutions of solute, further accounting for their increased size (and also enhancing the observed gradients in their radius, Figure 5.13). In other words, the interdendritic precipitate size is evolving by a growth mechanism rather than a coarsening one, which would explain the anomalously large radii and anamolously fast coarsening rates observed in reference [84]. Why are the precipitates in the melt-spun alloys consistently smaller than those observed 5 While this theory may seem far-fetch, consider also the study in Appendix D on an alloy prepared by Parameswaran et al. [285].

141 SECTION 5.4 DISCUSSION 141 Fig. 5.21: Montage of centered superlattice dark-field TEM micrographs of Al-0.2Zr aged at 425 C for 400 h. The apparent mean radius, R, of spheroidal Al 3 Zr (L1 2 ) precipitates is strongly dependent on location in the specimen. in the alloys produced by arc-melting? As discussed in Section 1.1.1, Al has particular limitations in its capacity to form extensive solid-solutions with all but a few neighboring elements, so the extension of solid-solubility by RSP will lead to immediately smaller precipitates because of the larger achievable supersaturation of solutes. A second effect of rapid solidification is the stabilization of plane-front solidification, thus minimizing or eliminating microsegregation [16, 338, 339]. Because alloys produced by RSP exhibit little or no microsegregation of

142 CHAPTER 5 MICROSTRUCTURE OF AL-ZR AND AL-ZR-TI ALLOYS DURING AGING AT 375, 400, OR 425 C 142 solutes, the precipitates that form are not subject to the various inhomogeneities in size and morphology associated with dendritically-distributed precipitates in arc-melted alloys. The apparently improved coarsening resistance, despite the higher volume fraction of precipitates in the melt-spun alloys, is also not surprising. Figure 5.6, however, demonstrates that the apparent coarsening kinetics of precipitates in the densely-populated dendrites are much slower than those in regions of smaller volume fraction, thus seemingly violating volume diffusioncontrolled coarsening theory. Al 3 Zr precipitates, despite their high number densities in the dendrites, exhibit impressive stability at 425 C (Table 5.3). Precipitate compositions Finally, consider the reported precipitate compositions in Table 5.4; in all cases the implicit assumption is that the ratio of solutes in the precipitated trialuminides is the same as that initially in solid-solution (i.e., the bulk alloy composition). This is at first glance a reasonable assumption since the transition elements are generally sparingly soluble in α-al and all solute atoms should therefore partition to the precipitated phase, as argued by Lee and colleagues [294, 295, 340]. It is shown in the present study, Chapter 4, that for Al-Zr-Ti alloys having the same atomic fraction of solute species, this Zr:Ti ratio of unity is never observed in the precipitated Al 3 (Zr 1 x Ti x ) phase. Moreover, as evidenced by Table 4.2, the assumption of negligible solubility of solutes in α-al is invalid; diffusion kinetics in α-al, which is anomalously slow and varies widely among the transition elements (Section 1.2.3), is instead the dominant factor dictating the relative amounts of solute atoms in the precipitated phase. Both Ti and V are slower diffusers than Zr in α-al, Figure 1.7 (especially true of V), and the precipitate compositions in Table 5.4 are therefore highly speculative. Summary Discounting the anomalous results from the studies on arc-melted alloys, the various Al 3 (Zr 1 x V x ) (L1 2 ), Al 3 (Zr 1 x Ti x ) (L1 2 ), and Al 3 (Zr 1 x y, V x, Ti y ) (L1 2 ) precipitates in Figure 5.20 have radii R = 3 7 nm after 400 h at 425 C, which are comparable to the radii of the Al 3 Zr (L1 2 ) and Al 3 (Zr 1 x Ti x ) (L1 2 ) dendritic precipitates in the present Al-Zr and Al-Zr-Ti alloys after 400 and 1,600 h at 425 C which range in size from R = 5 7 nm after 400 h and R = nm after 1,600 h (Table 5.3). Binary Al 3 Zr (L1 2 ) precipitates are more coarsening resistant than the

143 SECTION 5.5 CHAPTER SUMMARY 143 data of Zedalis and Fine [84] indicates, and the apparent improvement by V and/or Ti additions claimed in the studies of Table 5.4 are probably artifacts of the non-uniform precipitate morphologies associated with an inhomogenous dendritic distribution of precipitates in the initial studies on arc-melted alloys. As shown presently, ternary additions of solute have little effect on the coarsening behavior at 425 C (0.75T m ). 5.5 Chapter Summary Precipitation of Al 3 Zr (L1 2 ) and Al 3 (Zr 1 x Ti x ) (L1 2 ) was investigated in conventionally-solidified Al-0.1Zr, Al-0.2Zr, Al-0.1Zr-0.1Ti, and Al-0.2Zr-0.2Ti (at.%) alloys isothermally aged at 375, 400, or 425 C. The following results were obtained and discussed: Dendritic solidification results in non-uniform distributions of Zr and Ti, which segregate to the dendrite centers and are enriched by a factor of 2 and 3, respectively (Figure 5.5). The interdendritic regions are concurrently solute-depleted. During subsequent aging, Al 3 Zr Al 3 (Zr 1 x Ti x ) precipitates are similarly non-uniformly distributed in a dendritic manner (Figure 5.6). Pronounced hardening accompanies precipitation of Al 3 Zr or Al 3 (Zr 1 x Ti x ) at 375, 400, or 425 C for all alloys investigated (Figures ). The magnitude of the peak-aged hardness is controlled primarily by the Zr concentration. The alloys overage sluggishly, even after 3,200 h at 425 C, with no obvious benefit from Ti additions in delaying overaging. For shorter aging times (t < 10 h), Ti accelerates the nucleation kinetics, and for higher temperatures (425 C), the peak hardness is increased. These are effects of solute supersaturation (incubation time and precipitate volume fraction) and is not an indication of coarsening resistance. The precipitated microstructures in the alloys are complicated by the initial segregation of solute species; because of the locally-varying solute concentration, different precipitate morphologies are observed. These include: (i) larger interdendritic precipitates, frequently exhibiting growth instabilities (Figure 5.13); (ii) rods and plates aligned in 100 matrix directions (Figure 5.16); and (iii) small (< 10 nm) spheroidal L1 2 precipitates in the centers

144 CHAPTER 5 MICROSTRUCTURE OF AL-ZR AND AL-ZR-TI ALLOYS DURING AGING AT 375, 400, OR 425 C 144 of the dendritic cells (Figure 5.19). Even after aging at 425 C (0.75T m ) for 1,600 h, the equilibrium D0 23 structure (Al 3 Zr or Al 3 (Zr 1 x Ti x )) is not observed. The original motivation for adding Ti was to form Al 3 (Zr 1 x Ti x ) precipitates, that exhibit a reduced lattice parameter mismatch with α-al, thereby improving the coarsening resistance of the Ti-containing precipitates. Despite the confirmed partitioning of Ti to the Al 3 (Zr 1 x Ti x ) precipitates by 3-D atom-probe tomography (Chapter 4), there is no benefit in terms of coarsening resistance of the small dendritic Al 3 (Zr 1 x Ti x ) (L1 2 ) precipitates compared to Al 3 Zr (L1 2 ) during extended isothermal aging at 425 C (Table 5.3), consistent also with the similar overaging behavior observed by Vickers microhardness (Figures ). The apparent improvement of the coarsening resistance of Al 3 Zr (L1 2 ) by V and/or Ti additions claimed in the studies of Table 5.4 may be artifacts of a non-uniform dendritic distribution of precipitates in the initial studies on arc-melted Al-Zr alloys. Metastable Al 3 Zr (L1 2 ) precipitates are intrinsically coarsening resistant at 425 C (0.75T m ).

145 145 CHAPTER 6 Microstructural Coarsening in Al-Zr and Al-Zr-Ti Alloys, T 450 C The transformation of Al 3Zr (L1 2) and Al 3(Zr 1 xti x) (L1 2) precipitates to their respective equilibrium D0 23 structures is investigated in conventionally-solidified Al-0.1Zr and Al-0.1Zr-0.1Ti (at.%) alloys isothermally aged at 500 C or isochronally aged C. There is no benefit from Ti additions, both in terms of coarsening resistance of the metastable L1 2 precipitates, or in delaying the L1 2 to D0 23 transformation. Both alloys overage at the same rate at or above 500 C, during which spheroidal L1 2 precipitates transform to disc-like D0 23 precipitates ca. 200 nm in diameter and 50 nm in thickness, exhibiting a cube-on-cube orientation relationship with α-al. The transformation occurs heterogeneously on dislocations because of a large lattice parameter mismatch of the equilibrium D0 23 phase with α-al. The transformation is extraordinarily sluggish; even at 575 C, coherent L1 2 precipitates are also observed. Mechanisms of microstructural coarsening and strengthening are discussed with respect to the micrometer-scale dendritic distribution of precipitates. 6.1 Introduction CHAPTER 5 DISCUSSED the microstructure and ambient temperature mechanical properties of Al-Zr and Al-Zr-Ti alloys during isothermal aging at 375, 400, and 425 C. Despite extended aging times (3,200 h) at 425 C (0.75T m of Al), the alloys exhibit no appreciable overaging and the precipitates do not transform to the equilibrium D0 23 structure. This chapter investigates similar alloys during exposure at higher temperatures: (i) isothermally at 500 C (0.83T m ); and isochronally at C (0.94T m ). Microstructural coarsening and strengthening mechanisms, with respect to the micrometer-scale dendritic distribution of precipitates, are developed and the L1 2 to D0 23 transformation is discussed.

146 CHAPTER 6 MICROSTRUCTURAL COARSENING IN AL-ZR AND AL-ZR-TI ALLOYS, T 450 C 146 Table 6.1: Compositions and aging conditions of the Al-Zr and Al-Zr-Ti alloys investigated. Nominal comp. (at.%) Verified comp. (at.%) Aging temperatures ( C) Zr Ti Zr Ti Al-0.1Zr(b) Al-0.1Zr(c) Isochronal aging ( ) Al-0.1Zr-0.1Ti(b) Al-0.1Zr-0.1Ti(c) Isochronal aging ( ) 6.2 Experimental Procedures A series of binary Al-Zr and ternary Al-Zr-Ti alloys were investigated; alloy designations, compositions, and aging conditions are summarized in Table 6.1. The alloys were prepared employing non-consumable electrode arc-melting, as described in Sections 3.2 and 4.2. The verified compositions in Table 6.1 were obtained by bulk chemical analysis performed by direct current plasma emission spectroscopy (ATI Wah Chang, Albany, OR). To investigate the overaging behavior isothermally, peak-aged specimens (aged isothermally at 375 C for 100 h to ca. 450 MPa, Figures 5.7 and 5.9) of Al-0.1Zr(b) and Al-0.1Zr-0.1ZTi(b) were overaged isothermally at 500 C. The microstructures obtained after aging for 100 h at 500 C were investigated in both alloys by transmission electron microscopy (TEM) and scanning electron microscopy (SEM), using similar instruments and techniques described in Section 3.2. A second set of alloys, Al-0.1Zr(c) and Al-0.1Zr-0.1Ti(c), were aged in a series of multi-step 3 h isochronal anneals beginning at 300 C and terminating at 600 C, in 25 C increments. Between each aging step, the specimens were water-quenched and precipitation of Al 3 Zr or Al 3 (Zr 1 x Ti x ) was monitored by Vickers microhardness and electrical conductivity measurements. The conductivity measurements were performed using a SIGMATEST (Foerster Instruments, Pittsburgh, PA) eddy current apparatus at room temperature. Five measurements were recorded, each corresponding to a different frequency (60, 120, 240, 480, 960 khz), on each specimen. For consistency, a single specimen of each alloy was used for conductivity measurements, which was measured between each isochronal aging treatment. The microstructures obtained after isochronal aging to 450, 525, or 575 C were investigated in both alloys by TEM.

147 SECTION 6.3 EXPERIMENTAL RESULTS: ISOTHERMAL AGING AT 500 C V ic k e r s m ic r o h a r d n e s s (M P a ) V ic k e r s m ic r o h a r d n e s s (H V ) Fig. 6.1: Vickers microhardness vs. exposure at 500 C for peak-aged Al-0.1Zr(b) and Al-0.1Zr-0.1Ti(b) alloys. 6.3 Experimental Results: Isothermal Aging at 500 C Vickers microhardness Figure 6.1 displays Vickers microhardness as a function of aging time at 500 C, indicating that both Al-0.1Zr(b) and Al-0.1Zr-0.1Ti(b) alloys overage at the same rate with no improved coarsening resistance of Al 3 (Zr 1 x Ti x ) (L1 2 ) as compared to Al 3 Zr(L1 2 ). These results are consistent with those of Chapter 5, where the precipitate radii of Al 3 Zr(L1 2 ) and Al 3 (Zr 1 x Ti x ) (L1 2 ) precipitates were statistically identical after extended aging times at 425 C (Table 5.3) Electron microscopies Precipitate dissolution and transformation to the equilibrium D0 23 structure The overaging behavior exhibited in Figure 6.1 was investigated directly by TEM and SEM, and is due to dissolution of the small, high number density, dendritic spheroidal Al 3 Zr (L1 2 ) or Al 3 (Zr 1 x Ti x ) (L1 2 ) precipitates which then re-precipitate on grain boundaries or dislocations, ultimately forming extensive arrays of heterogeneously-nucleated equilibrium D0 23 -structured

148 CHAPTER 6 MICROSTRUCTURAL COARSENING IN AL-ZR AND AL-ZR-TI ALLOYS, T 450 C 148 Fig. 6.2: Montage of bright-field TEM micrographs of Al-0.1Zr-0.1Ti(b) aged at 500 C for 100 h (after aging at 375 C for 100 h). Imaged with g = (020).

149 SECTION 6.3 EXPERIMENTAL RESULTS: ISOTHERMAL AGING AT 500 C 149 Fig. 6.3: Montage of centered superlattice dark-field TEM micrographs of Al-0.1Zr-0.1Ti(b) aged at 500 C for 100 h (after aging at 375 C for 100 h). Imaged with the g = (010) superlattice reflection.

150 CHAPTER 6 MICROSTRUCTURAL COARSENING IN AL-ZR AND AL-ZR-TI ALLOYS, T 450 C 150 Fig. 6.4: SEM micrograph of heterogeneously-nucleated Al 3 Zr (D0 23 ) precipitates on dislocations or small-angle grain boundaries in Al-0.1Zr(b) aged at 500 C for 100 h (after aging at 375 C for 100 h). The morphology of the D0 23 phase is disc-like, nm in diameter and nm in thickness, and exhibit one of three possible orientation relationships with the α-al solid-solution. precipitates. Figures 6.2 and 6.3 display complementary bright-field and dark-field micrographs, respectively, of Al-0.1Zr-0.1Ti(b) aged at 500 C for 100 h (after aging at 375 C for 100 h). Several dislocations are shown, on which numerous disc-like, ca nm diameter and 50 nm thick, equilibrium D0 23 -structured precipitates have formed (precipitates A in Figures 6.2 and 6.3). The dislocations and linear arrays of precipitates are most apparent in the bright-field micrographs, Figure 6.2. The arrays of heterogeneously-nucleated D0 23 precipitates extend for hundreds of micrometers and are observable also by SEM, Figure 6.4, which shows similar arrays of disc-shaped

151 SECTION 6.3 EXPERIMENTAL RESULTS: ISOTHERMAL AGING AT 500 C 151 [001] Al [100] Al [010] Al c-axis c-axis c-axis I: c-axis [100] Al II: c-axis [010] Al III: c-axis [001] Al Fig. 6.5: Schematic of the three possible orientation relationships with α-al for the disc-like equilibrium D0 23 precipitates. Adapted from Chen [334]. precipitates along several dislocations extending over 50 µm in length. It initially seems unlikely that such a long dislocation length would randomly exist exactly parallel to the specimen surface in Figure 6.4 (the probably of observing a dislocation by TEM is much larger since a 3-D volume, ca. 100 nm thick, is observable). For a metal at room temperature, the dislocation density is of the order of m 2 [341], or one dislocation per µm 2 area. Thus, the area represented in Figure 6.4(a) can be expected to contain 18 dislocations many more than what is visually apparent intersecting the specimen at various orientations, with one of them parallel to the specimen surface. The disc-like D0 23 precipitates exist in one of three orientations: two orthogonal edge-on views and one face-on when viewed along [100] Al, indicating a cube-on-cube orientation relationship with the α-al solid-solution, shown schematically in Figure 6.5. Only precipitates with certain orientation relationships, those whose c-axis is parallel to [010] Al, are illuminated in darkfield when imaged with g = (010) in Figure 6.3; this indicates that the precipitates have the D0 23 structure, as discussed in more detail below. Transformation mechanism to the equilibrium D0 23 structure The equilibrium D0 23 -structured precipitates are not nucleated directly from supersaturated α- Al solid-solution during aging, but rather form by dissolution and subsequent re-precipitation of solutes from nearby pre-existing L1 2 precipitates formed during the prior isothermal aging treatment at 375 C. The consumption of the pre-existing L1 2 precipitates is unambiguously shown in the bottom-most precipitate-rich dendrite in Figure 6.3, which initially contained a relatively

152 CHAPTER 6 MICROSTRUCTURAL COARSENING IN AL-ZR AND AL-ZR-TI ALLOYS, T 450 C 152 high number density of homogenenously-distributed spheroidal L1 2 Al 3 (Zr 1 x Ti x ) precipitates in the peak-aged condition (100 h at 375 C). During aging at 500 C, the equilibrium D0 23 precipitates form heterogeneously along a dislocation cutting through the dendrite, leaving behind a precipitate-free cylinder approximately 500 nm in radius around the dislocation (precipitates A in Figures 6.2 and 6.3). A similar precipitate-free zone extends 1.5 µm to either side of the heterogeneously-nucleated array of disc-like Al 3 Zr precipitates in Figure 6.4(c). The widths of these precipitate-free zones is comparable to the calculated 1 root-mean-squared diffusion distance, 6Dt, of Zr or Ti after 100 h at 500 C, which is 2.6 or 0.8 µm, respectively. Arrays of spheroidal L1 2 precipitates, labeled B in Figures 6.2 and 6.3, are also nucleated heterogeneously along the dislocation. These precipitates are surrounded by a narrower, nm wide, precipitate-depleted zone, suggesting that the metastable L1 2 structure represents an intermediate stage of microstructural coarsening. Figure 6.6 demonstrates that many of the heterogeneously-nucleated L1 2 precipitates exhibit preferential growth in 100 directions, forming elongated cigar-shaped precipitates. They also contain antiphase boundaries (APBs), characterized by sharp lines of no-contrast parallel to {100} planes inside the L1 2 precipitates, and represent a transition to the equilibrium D0 23 structure as discussed below. The precipitates labeled C, most obvious in Figure 6.2, are other examples of the transformation to the equilibrium D0 23 structure occurring heterogeneously on dislocations cutting through a precipitate-rich dendrite, leaving behind precipitate-depleted zones surrounding the dislocation. Finally, regions marked D in Figures 6.2 and 6.3 indicate a second mechanism of microstructural coarsening. Here, larger spheroidal L1 2 -structured precipitates at the dendrite edge have coarsened at the expense of the smaller precipitates immediately on the interior side of the dendrites, leaving behind a narrow, ca nm, precipitate-free zone on either side of the ca. 1 µm wide interdendritic channel. This reduction in volume fraction of the small dendritic precipitates also contributes to the drop in hardness during exposure at 500 C, Figure 6.1. Structure and morphology of the D0 23 precipitates Figure 6.7 displays a bright-field TEM micrograph of three orthogonal D0 23 -structured precipitates viewed along [100] Al, exhibiting the three orientation relationships depicted schematically in Figure 6.5. Selected area diffraction patterns (SADPs) from two of the precipitates are also 1 Using diffusivities reported in Table 1.3.

153 SECTION 6.3 EXPERIMENTAL RESULTS: ISOTHERMAL AGING AT 500 C 153 Fig. 6.6: Complementary bright-field and dark-field TEM micrographs of heterogeneously-nucleated L1 2 - and D0 23 -structured Al 3 (Zr 1 x Ti x) precipitates on dislocations in Al-0.1Zr-0.1Ti(b) aged at 500 C for 100 h (after aging at 375 C for 100 h). Many of the L1 2 -structured precipitates exhibit attributed to antiphase boundaries (APBs) along 100 directions, indicating transformation to the equilibrium D0 23 structure. shown in Figure 6.7, which agree with kinematically-simulated diffraction patterns in Figure 6.9, indicating that the precipitates in Figure 6.7 have the equilibrium D0 23 structure of Al 3 Zr. The disc-like morphology of the equilibrium Al 3 Zr precipitates reflects the tetragonal symmetry of the D0 23 unit cell. The lattice parameters of Al 3 Zr (D0 23 ) are a = b = and c = Å, resulting in lattice parameter mismatches with α-al (a = Å) of δ a = δ b = -0.88% and δ c = 6.92% (Table 1.1). Since the lattice parameter mismatch is most severe in the c-direction, the energetic preference is for precipitates that are thin along c, explaining the disc-like morphology. The large overall lattice parameter mismatch with α-al, δ = 2.89% (Table 1.1), also explains why the

154 CHAPTER 6 MICROSTRUCTURAL COARSENING IN AL-ZR AND AL-ZR-TI ALLOYS, T 450 C 154 Fig. 6.7: Bright-field TEM micrograph of disc-like D0 23 precipitates in three possible orientations and corresponding selected area diffraction patterns (SADPs) of the equilibrium D0 23 phase in Al-0.1Zr-0.1Ti(b) aged at 500 C for 100 h (after aging at 375 C for 100 h).the patterns are indexed in Figure 6.9. equilibrium phase is only observed heterogeneously-nucleated along dislocations. Orientation relationships and diffraction contrast of the D0 23 precipitates The disc-like D0 23 -strcutured precipitates exhibit a cube-on-cube orientation relationship with α-al, with the three possible orientations, designated I, II, or III as viewed along [100] Al (Figure 6.5). The equilibrium precipitates are observed only on dislocations, and appear to be at least partially coherent with α-al, as evidenced by strong strain contrast when imaged under two-beam conditions, Figure 6.8(a). Because of their large size (ca. 200 nm in diameter), the disc-like precipitates are most likely semicoherent with α-al, which is supported also by the appearance of striped lines of contrast normal to g in the images of many of the precipitates when viewed face-on (i.e., those in orientation I). These are interpreted as contrast due to interfacial dislocations, indicating semicoherency with the α-al matrix. The dark-field micrographs in Figure 6.8 illustrate a second diffraction contrast phenomenon. As observed previously (e.g. in Figures 6.3 and 6.6), and shown in Figure 6.8(b), only the disclike precipitates whose c-axes are parallel to [010] Al (orientation II) are illuminated when imaged with the g = (010) L1 2 superlattice reflection. Similarly, only precipitates whose c-axes are parallel to [100] Al (orientation I) are illuminated when imaged with g = (011), Figure 6.8(d). These diffrac-

155 SECTION 6.3 EXPERIMENTAL RESULTS: ISOTHERMAL AGING AT 500 C 155 Fig. 6.8: Complementary bright-field and dark-field TEM micrographs of heterogeneously-nucleated D0 23 precipitates on dislocations in Al-0.1Zr-0.1Ti(b) aged at 500 C for 100 h (after aging at 375 C for 100 h). The strong contrast and presence of interfacial dislocations in bright-field indicates that the disc-like precipitates are semicoherent with α-al. In dark-field, only certain orientations are illuminated, depending on the g vector that is operating. tion contrast phenomena may be understood considering the calculated diffraction patterns for the D0 23 and L1 2 structures of Al 3 Zr in Figure 6.9. For the D0 23 precipitates in orientation II, the 004 T reflections nearly coincide with the 010 C reflections from the L1 2 precipitates, explaining why only those tetragonal precipitates in orientation II are illuminated in dark-field when imaged with the g = (010) L1 2 superlattice reflection as in Figure 6.3 and 6.8(b). With g = (011) operating, Figure 6.8(d), only the disc-like precipitates in orientation I are illuminated, since the 011 T reflections from these precipitates overlap with the 011 C superlattice reflections from the L1 2 precipitates.

156 CHAPTER 6 MICROSTRUCTURAL COARSENING IN AL-ZR AND AL-ZR-TI ALLOYS, T 450 C Al 022 Al 002 Al 024 T 028 T 022 Al 002 Al 022 Al 020 T 012 C 220 T 020 T 012 C 008 T T C 028 T 017 T 110 T 001 C 011 C 021 C 011 T 013 T 015 T 017 T 019 T 001 C 011 C 021 C 015 T 001 C C 021 C T 024 T 013 T 200 T 004 T 008 T 011 T 020 T C 020 Al C 020 Al C 020 Al I: c-axis [100] Al II: c-axis [010] Al III: c-axis [001] Al Fig. 6.9: Calculated electron diffraction patterns from the equilibrium tetragonal D0 23 (closed circles, T subscript) and metastable cubic L1 2 (crosses, C subscript) Al 3 Zr phases in three possible orientation relationships with α-al (open circles). The spots not indexed (e.g. 002 T, 006 T ) are kinematically-forbidden. Small dendritic precipitates Figure 6.10 shows that the precipitates in both Al-0.1Zr(b) and Al-0.1Zr-0.1Ti(b) are L1 2 -structured, coherent with α-al, and exhibit impressive coarsening resistance at 500 C. The mean radii of the dendritic precipitates in Figures 6.10(a) and 6.10(c) are displayed in Table 6.2; after aging at 500 C for 100 h (preceded by aging at 375 C for 100 h), the mean radii of the Al 3 Zr and Al 3 (Zr 1 x Ti x ) precipitates are statistically identical: 12.3 ± 2.9 and 12.0 ± 4.0 nm, respectively, which is consistent also with the identical drop in hardness exhibited by both alloys during exposure at 500 C, Figure 6.1. Figure 6.11 shows that the larger dendritic precipitates (R 20 nm) exhibit planar faults, indicating a partial transition to the D0 23 structure. Nevertheless, as shown in Figure 6.10, the vast majority of the dendritic precipitates are L1 2 -structured and fully coherent, indicating that the L1 2 to D0 23 transformation does not occur spontaneously even after 100 h at 500 C. The transformation instead requires pre-exisitng defects and other heterogeneous nucleation sites.

157 SECTION 6.3 EXPERIMENTAL RESULTS: ISOTHERMAL AGING AT 500 C 157 Fig. 6.10: Complementary dark-field and bright-field TEM micrographs of dendritic precipitates in Al-0.1Zr(b) (panels a and b) and Al-0.1Zr-0.1Ti(b) (panels c and d) aged at 500 C for 100 h (after aging at 375 C for 100 h). The Al 3 Zr or Al 3 (Zr 1 x Ti x) precipitates have the metastable L1 2 structure and are coherent with the α-al solid-solution, with a mean radius ca. 12 nm. Table 6.2: Mean precipitate radii, R, of precipitates observed in the dendrite centers of Al-0.1Zr(b) and Al-0.1Zr-0.1Ti(b) aged at 500 C for 100 h (after aging at 375 C for 100 h). Alloy Aging treatment Mean precipitate radius, R (nm) Number of precipitates Al-0.1Zr(b) 500 C for 100 h (after 375 C for 100 h) 12.3 ± Al-0.1Zr-0.1Ti(b) 500 C for 100 h (after 375 C for 100 h) 12.0 ±

158 CHAPTER 6 MICROSTRUCTURAL COARSENING IN AL-ZR AND AL-ZR-TI ALLOYS, T 450 C 158 Fig. 6.11: Centered superlattice dark-field TEM micrographs of Al-0.1Zr-0.1Ti(b) aged at 500 C for 100 h (after aging at 375 C for 100 h). Many of the spheroidal L1 2 -structured precipitates exhibit APBs along 100 directions, indicating a partial transformation to the equilibrium D0 23 structure. 6.4 Experimental Results: Isochronal Aging, C Vickers microhardness and electrical conductivity Figure 6.12 indicates the precipitation behavior of Al-0.1Zr(c) and Al-0.1Zr-0.1Ti(c) during 3 h isochronal aging, as monitored by Vickers microhardness and electrical conductivity. Precipitation of Al 3 Zr commences between 350 and 375 C in the binary alloy Al-0.1Zr(c), as evidenced by the increase in strength and the accompanying change in electrical conductivity. For the Ti-containing alloy, Al-0.1Zr-0.Ti(c), precipitation of Al 3 (Zr 1 x Ti x ) begins at temperatures lower by about 25 C, consistent with the accelerated isothermal nucleation kinetics (shorter incubation time) for Al-0.1Zr-0.1Ti(b), Figure 5.9, compared with Al-0.1Zr(b), Figure 5.7. The maximum strength achieved during isochronal aging is also slightly greater (ca. 25 MPa) in the Ticontaining alloy indicating a finer population, or larger volume fraction, of Al 3 (Zr 1 x Ti x ) precipitates. The amplitude of the change in conductivity between as-cast and peak-aged condition is, however, less for Al-0.1Zr-0.1Ti(c) (2.7 MS m -1 ) than it is for Al-0.1Zr(c) (3.7 MS m -1 ), suggesting that less solute is precipitated from solid-solution in the ternary alloy despite having twice the initial total solute concentration. The most obvious effect of Ti additions is the reduced conductivity initially in Al-0.1Zr-0.1Ti(c) compared to that of Al-0.1Zr(c), which might explain the reduced absolute change in conductivity for the ternary alloy. Barghout et al. [342] showed that

159 ) SECTION 6.4 EXPERIMENTAL RESULTS: ISOCHRONAL AGING, C V ic k e r s m ic r o h a r d n e s s (M P a ) A l-0.1 Z r-0.1 T i(c ) A l-0.1 Z r(c ) E le c tr ic a l c o n d u c tiv ity (M S m A l-0.1 Z r(c ) A l-0.1 Z r-0.1 T i(c ) A s -c a s t Fig. 6.12: Vickers microhardness and electrical conductivity evolution during isochronal aging (3 h at each temperature) of Al- 0.1Zr(c) and Al-0.1Zr-0.1Ti(c). the presence of second phase precipitates reduces the electrical conductivity in Al-Fe and Al-Mg alloys. The smaller change in conductivity for Al-0.1Zr-0.1Ti(c) may therefore indicate a greater volume fraction of Al 3 (Zr 1 x Ti x ) precipitates compared to Al 3 Zr precipitates in Al-0.1Zr(c), consistent with the increased hardness of the ternary alloy over the binary. The Vickers microhardness and electrical conductivity data in Figure 6.12 indicate that Ti additions have a slight effect on the hardness of the alloys in the peak-aged condition, and a negligible influence on the total amount of solute precipitated from solid-solution. These observations are consistent with the results in Chapter 4, where only a small amount of Ti is precipitated from α-al solid-solution in Al-Zr-Ti alloys, and also the hardness data in Figures , where the effect of Ti is small compared to that of Zr in dictated the hardness of the alloys during

160 CHAPTER 6 MICROSTRUCTURAL COARSENING IN AL-ZR AND AL-ZR-TI ALLOYS, T 450 C 160 Fig. 6.13: Bright-field and centered superlattice dark-field TEM micrographs of Al-0.1Zr-0.1Ti(c) isochronally aged to 450 C. (a) Precipitate-rich dendrite are clearly defined. Contrast in bright-field is due to thickness contrast. (b) Dark-field micrograph displaying several precipitate-rich dendrites. (c) Arrays of coalescing precipitates near a dendrite edge. (d) Dendritic L1 2 precipitates with a mean radius R = 3.4±0.5 nm. isothermal aging Transmission electron microscopy Microstructural coarsening of the alloys, which occurs for T > 450 C in Figure 6.12, was investigated directly by TEM in both alloys after isochronal aging to 450, 525, and 575 C. Figure 6.13 displays the microstructure in Al-0.1Zr-0.1Ti(c) isochronally aged to 450 C, which corresponds to the onset of coarsening in Figure Distinct, well-defined precipitate-rich dendrites are observed in Figures 6.13(a) and 6.13(b); the contrast observed in bright-field, Fig-

161 SECTION 6.4 EXPERIMENTAL RESULTS: ISOCHRONAL AGING, C 161 Fig. 6.14: Centered superlattice dark-field TEM micrographs of Al-0.1Zr-0.1Ti(c) isochronally aged to 525 C. (a) Precipitate-rich dendrite (outline dotted). D0 23 -structured precipitates are nucleated heterogeneously on a dislocation, leaving behind a ca. 400 nm wide precipitate-free zone. (b) Magnified region from panel (a). (c) Arrays of coalescing precipitates in an earlier stage of coarsening. (d) Dendritic L1 2 precipitates with a mean radius R = 6.6±1.9 nm. ure 6.13(a), is from thickness differences arising from preferential thinning of the interdendritic channels during electropolishing, discussed also in Appendix C. Figure 6.13(c) displays precipitates near a dendrite edge, where precipitate coalescence along ca. 100 nm linear arrays are observed within the precipitate-rich dendrite. It is believed that these precipitates are coalescing on dislocations, ultimately leading to vast arrays of heterogeneously-nucleated D0 23 precipitates during the latter stages of coarsening. Figure 6.13(d) shows that within the dendrites, the precipitates are small and spheroidal, with a mean radius R = 3.4 ± 0.5 nm. Figure 6.14 shows the microstructure in Al-0.1Zr-0.1Ti(c) isochronally aged to 525 C, which

162 CHAPTER 6 MICROSTRUCTURAL COARSENING IN AL-ZR AND AL-ZR-TI ALLOYS, T 450 C 162 Fig. 6.15: Complementary bright-field and dark-field TEM micrographs of Al-0.1Zr-0.1Ti(c) isochronally aged to 575 C. Most of the metastable L1 2 precipitates are transformed to the disc-like D0 23 structure. corresponds to an intermediate stage of coarsening. Figures 6.14(a) and 6.14(b) display disc-like D0 23 precipitates heterogenenously nucleated on a dislocation cutting through a precipitaterich dendrite. A surrounding precipitate-free cylinder, up to 400 nm in radius, is formed as a result of the D0 23 formation. The calculated cumulative root-mean-squared diffusion distance, 6Dt, of Zr or Ti atoms isochronally aged to 525 C is 1.8 or 0.5 µm, respectively, which is comparable to the radius of the precipitate-free cylinder in Figure 6.14(b); this is consistent with the mechanism of microstructural coarsening described previously in Figures 6.3 and 6.4. Other regions of the alloy, Figure 6.14(c), exhibit earlier stages of microstructural coarsening with smaller precipitate-free zones surrounding other arrays of coalescing precipitates. Within the dendrites, most precipitates are spheroidal and have the L1 2 structure, with a mean radius of 6.6 ± 1.9 nm. By 575 C, the majority of the precipitates have transformed to the equilibrium D0 23 structure, demonstrated in Figure The disc-like D0 23 precipitates exhibit a cube-on-cube orientation relationship with α-al, as discussed previously in Figure 6.5. There are, however, other regions of the alloy containing spheroidal L1 2 precipitates, indicated in Figure 6.15(b). These precipitates

163 SECTION 6.4 EXPERIMENTAL RESULTS: ISOCHRONAL AGING, C 163 Fig. 6.16: Complementary dark-field and bright-field TEM micrographs of dendritic precipitates in Al-0.1Zr(c) (panels a and b) and Al-0.1Zr-0.1Ti(c) (panels c and d) isochronally aged to 575 C. The Al 3 Zr or Al 3 (Zr 1 x Ti x) precipitates have the metastable L1 2 structure and are coherent with the α-al solid-solution, with a mean radius ca nm. exhibit impressive coarsening resistance despite exposure at 575 C (0.91T m ), with a mean radius of 16.3 ± 3.4 nm. Figure 6.16 displays complementary dark-field and bright-field TEM micrographs of dendritic L1 2 precipitates in Al-0.1Zr(c) and Al-0.1Zr-0.1Ti(c) isochronally aged to 575 C. The precipitates have the same radius, 16.9 ± 4.4 nm and 16.3 ± 3.4 nm, respectively, demonstrating that there is no difference in coarsening of spheroidal Al 3 Zr(L1 2 ) and Al 3 (Zr 1 x Ti x ) (L1 2 ) precipitates at 575 C. Moreover, the dendritic L1 2 precipitates are coherent with α-al, demonstrated by the Ashby-Brown strain contrast exhibited in the bright-field micrographs in Figure Con-

164 CHAPTER 6 MICROSTRUCTURAL COARSENING IN AL-ZR AND AL-ZR-TI ALLOYS, T 450 C 164 Table 6.3: Mean precipitate radii, R, observed in the dendrite centers in Al-0.1Zr(c) and Al-0.1Zr-0.1Ti(c) isochronally aged at 450, 525, and 575 C. Alloy Last aging treatment ( C) Mean precipitate radius, R (nm) Number of precipitates counted Al-0.1Zr(c) ± ± ± Al-0.1Zr-0.1Ti(c) ± ± ± sidering the size of these precipitates, their coherency is not too surprising since their radii are below the R = nm threshold for coherency loss discussed in Figure Figures 6.13, 6.14, and 6.15 are for Al-0.1Zr-0.1Ti(c) aged isochronally at 450, 525, and 575 C, respectively, but are representative of the binary Al-0.1Zr(c) as well. Indeed, Table 6.3 demonstrates that measured precipitate radii after isochronal aging at 450, 525, and 575 C are statistically identical for Al-0.1Zr(c) and Al-0.1Zr-0.1Ti(c), consistent with their similar overaging behavior demonstrated in Figure As found in Chapter 5, the addition of Ti has little effect on the coarsening of Al 3 Zr (L1 2 ) precipitates. 6.5 Discussion Transformation to the equilibrium D0 23 structure The L1 2 to D0 23 transformation is summarized schematically in Figure Spheroidal L1 2 precipitates coarsen, eventually developing planar faults parallel to {100} planes inside the L1 2 precipitates. This strain contrast is attributed to antiphase boundaries (APBs) generated in the transition to an imperfect D0 23 structure. As coarsening progresses, the precipitates elongate along these faults into a cigar-shaped morphology, eventually transforming into D0 23 -structured precipitates exhibiting a disc-like morphology with a cube-on-cube orientation relationship with α-al. Examples of spheroidal L1 2 precipitates exhibiting APBs are provided in Figures 5.18(d) and Zedalis and Fine [84] observed similar faults in Al 3 Zr (L1 2 ) precipitates formed af-

165 SECTION 6.5 DISCUSSION 165 [001] Al [100] Al [010] Al Fig. 6.17: Schematic of the L1 2 to D0 23 transformation in Al 3 Zr precipitates. Spheroidal L1 2 precipitates coarsen, eventually developing antiphase boundaries (APBs) parallel to {100} planes. Preferential coarsening occurs along the APBs, resulting in cigarshaped precipitates. These eventually transform into disc-shaped D0 23 -structured precipitates like those in Figure 6.5. ter aging Al-0.22Zr (at.%) at 600 C for 12 h, which they postulated were APBs. This was confirmed by Chen et al. [114, 334] studying Al 3 (Zr 0.25 V 0.75 ) (L1 2 ) precipitates formed in melt-spun Al-0.34Zr-0.98V (at.%) alloys aged at 500 C. Using high resolution electron microscopy, these researchers showed that the APBs are associated with a displacement of a/2 110 on {100} planes [334]. Lee et al. [294] showed similar planar faults in L1 2 precipitates (purportedly Al 3 (Zr 0.40, V 0.40, Ti 0.20 ) 2 ) formed in Al-0.8Zr-0.8V-0.4Ti (at.%) alloys aged at 425 C for h. Ryum [85] demonstrated that the transformation from the cubic L1 2 structure to an imperfect tetragonal structure involves the introduction of an APB with a displacement vector a/2 110 on {100}-type planes in D0 22 -structured Al 3 Hf. The observed APBs in the spheroidal L1 2 -structured precipitates (Figures 5.18(d) and 6.6) therefore represent the early stages of transformation to the equilibrium D0 23 structure. As indicated in Figure 6.17, a preferential growth in the precipitate along the APB represents a later stage of transformation. Figure 6.6 displays examples of such cigar-shaped precipitates, which have also been observed by others [84, 114, 294]. Temperature of the transformation The L1 2 to D0 23 structural transformation seems to occur at approximately 500 C, which corresponds also to the onset of overaging as observed by Vickers microhardness in Figures 6.1 and In Chapter 5, D0 23 precipitates were never observed during extended aging times (1,600 h) at 425 C. In the present isochronally aged alloys, D0 23 precipitates are not observed at 450 C, 2 The precipitate compositions were not measured directly.

166 CHAPTER 6 MICROSTRUCTURAL COARSENING IN AL-ZR AND AL-ZR-TI ALLOYS, T 450 C 166 Figure 6.13, and are observed to a limited extent at 525 C, Figure After extended aging times (100 h) at 500 C, Figures 6.2 and 6.3, the transformation is extensive along dislocations, although most precipitates are still L1 2 -structured. Even at 575 C, L1 2 precipitates are observed in Figure 6.15, indicating that the L1 2 to D0 23 transformation is not spontaneous. Additional factors, including aging time and dislocation density in the specimen, will also naturally influence on the kinetics and the apparent transformation temperature. Nevertheless, the transformation process seems to occur at 500 C within reasonable aging times. This conclusion concurs with results of several studies in the scientific literature. Zedalis and Fine [84] observed transformation to the equilibrium D0 23 structure in Al-0.24Zr (at.%) aged at 450 C. Their specimens, however, were initially cold rolled 95%; the high dislocation density therefore explains why these researchers observed D0 23 precipitates at a relatively low temperature. Izumi and Oelschlägel [76] reported a fine dispersion of small round particles (L1 2 -crystal structure), which are statistically distributed within the dendritic cells of the cast structure in an Al-0.33Zr (at.%) aged at 450 C. Even after extended aging times (218 h), these researchers reported that the Al 3 Zr precipitates had not transformed to the stable D0 23 structure and the L1 2 precipitates were resistant to coarsening with R = 10.9 nm. Nes [79] aged an Al-0.05Zr (at.%) alloy at 460 C up to 700 h and observed small, R < 35 nm Al 3 Zr (L1 2 ) precipitates and, like Izumi and Oelschlägel [76], never observed the equilibrium D0 23 structure at 460 C. Ryum [78] studied precipitation of Al 3 Zr (L1 2 ) in a chill-cast Al-0.15Zr (at.%) alloy aged at 500 C and at this temperature, the equilibrium D0 23 structure was observed only after long aging times (>120 h), primarily at grain boundaries. Rystad and Ryum [190] studied recrystallization in an Al-0.15Zr (at.%) alloy annealed in the range C; L1 2 -structured Al 3 Zr precipitates were observed at all temperatures except at 600 C, when plate-shaped equilibrium D0 23 structured Al 3 Zr precipitates were formed after 8 h. Chaudhury and Suryanarayana [83] observed the precipitation behavior of Al 3 Zr in an Al-3.0Zr (at.%) alloy prepared by RSP. The highly supersaturated solid solution was stable beyond 300 C, with discontinuous precipitation of L1 2 Al 3 Zr observed after aging for 1 h at 320 C. Metastable L1 2 -structured Al 3 Zr was still observed after aging aging for 1 h at 500 C, followed quickly by the formation of the equilibrium D0 23 structure after 3 h at the same temperature. Finally, Srinivasan et al. [94] formulated single-phase Al 3 Zr (L1 2 ) through mechanical alloying of pure elemental powders, and found that the L1 2 to D0 23 transformation occurs at 550 C, as monitored by X-ray diffraction and differential scanning calorimetry.

167 SECTION 6.5 DISCUSSION 167 Stage I Stage II Stage III Stage IV Fig. 6.18: Schematic of four stages of microstructural coarsening occurring on the nanometer-scale of the precipitates and the micrometer-scale of the dendrites Mechanism of microstructural coarsening The mechanism of microstructural coarsening described in this chapter may be summarized schematically in Figure 6.18, in which four stages of coarsening are defined. Stage I coarsening corresponds to the peak-aged condition, where well-defined precipitate-rich dendrites contain a dense distribution of small (R < 5 nm) coherent L1 2 precipitates. Larger interdendritic spheroidal L1 2 precipitates like those in Figures 5.11 and 5.12 are also depicted, many of them heterogeneously nucleated on dislocations.

168 CHAPTER 6 MICROSTRUCTURAL COARSENING IN AL-ZR AND AL-ZR-TI ALLOYS, T 450 C 168 Stage II is the initiation of microstructural coarsening, where dendritic L1 2 precipitates coarsen by Ostwald ripening and the heterogeneously-nucleated precipitates coarsen rapidly by dislocation pipe diffusion. Within the precipitate-rich dendrites, preferential coarsening occurs heterogeneously on dislocations. During Stage II there is a macroscopically-observed decrease in strength. Stage III coarsening signals the beginning of the L1 2 to D0 23 structural transformation, limited to dislocations and probably other heterogeneous nucleation sites such as grain boundaries. The heterogeneously-nucleated L1 2 precipitates grow into elongated cigar-shaped precipitates, exhibiting preferential growth in 100 directions. Antiphase boundaries, characterized by sharp lines of no-contrast parallel to {100} planes inside the L1 2 precipitates, signal the beginning of the transition to the D0 23 structure. The transformation to the disc-like D0 23 phase occurs at the expense of the surrounding L1 2 precipitates, leaving behind precipitate-free cyclinders up to ca. 1 µm in diameter surrounding the dislocation. By Stage IV coarsening, most of the metastable L1 2 precipitates have transformed to the equilibrium D0 23 structure. There are, however, remnants of the precipitate-rich dendrites that still contain coherent, spheroidal L1 2 precipitates. Despite their presence, virtually all strength is lost on the macroscopic scale. The dark-field TEM micrographs in Figure 6.19 displays actual examples of the four stages of coarsening described in Figure Figure 6.19(a) displays Al-0.1Zr-0.1Ti(a) aged at 375 C (0.69T m ) for 1,600 h. The microstructure contains well-defined dendrites full of small ( R < 5 nm) coherent L1 2 precipitates. Stage II coarsening is reached during extended aging at 425 C (0.75T m ), as evidenced by Figure 6.19(b) displaying Al-0.1Zr-0.1Ti(b) aged at 425 C (0.75T m ) for 1,600 h. The microstructure consists of coherent L1 2 -structured precipitates, R ca. 12 nm, but precipitate coalescence in heterogeneous arrays is also widespread. Stage III coarsening is characterized by the beginning of the L1 2 to D0 23 transformation, which is occurring along a dislocation in Figure 6.19(c) for Al-0.1Zr-0.1Ti(b) aged at 500 C (0.83T m ) for 100 h (after aging at 375 C for 100 h). The transformation occurs at the expense of the surrounding L1 2 dendritic precipitates, forming a precipitate-free cylinder 400 nm in radius in Figure 6.19(c). In Stage IV coarsening, the majority of the precipitates have transformed to the D0 23 structure, as shown Figure 6.19(d) for Al-0.1Zr-0.1Ti(d) isochronally aged to 575 C (0.91T m ).

169 SECTION 6.5 DISCUSSION 169 Fig. 6.19: Centered superlattice dark-field TEM micrographs displaying the four stages of microstructural coarsening shown schematically in Figure Panel (a) is Al-0.1Zr-0.1Ti(a) aged at 375 C for 1,600 h. Well-defined dendrites contain small ( R ca. 5 nm) coherent L1 2 -structured precipitates. Panel (b) is Al-0.1Zr-0.1Ti(b) aged at 425 C for 1,600 h. Heterogeneously-nucleated spheroidal L1 2 -structured precipitates, fed by rapid pipe diffusion, grow at the expense of nearby dendritic precipitates. Panel (c) is Al-0.1Zr-0.1Ti(b) aged at 500 C for 100 h (after aging at 375 C for 100 h). Transformed D0 23 precipitates are observed heterogeneously-nucleated on a dislocation, forming a precipitate-free cylinder around the dislocation. Panel (d) is Al-0.1Zr- 0.1Ti(c) isochronally aged to 500 C. The microstructure consists of fully-transformed D0 23 precipitates exhibiting a disc-like morphology (ca. 200 nm in diameter and 50 nm in thickness) and a cube-on-cube orientation relationship with α-al Mechanism of strengthening Theories of precipitation strengthening have been thoroughly reviewed and assessed, originally by Brown and Ham [343] and more recently by Ardell [20, 23]. Precipitate shearing, precipitate bypass by dislocation looping, or a combination of these two mechanisms can generally explain ambient temperature strength in coarse-grained, non-strain-hardened, precipitate-strengthened alloys in the absence of other strengthening mechanisms (e.g., Hall-Petch or solid-solution strength-

170 CHAPTER 6 MICROSTRUCTURAL COARSENING IN AL-ZR AND AL-ZR-TI ALLOYS, T 450 C 170 ening). For large precipitate radii, R, the strength is controlled by Orowan dislocation looping, defined in Eq.(1.1) [25]: σ or = M 0.4 Gb π (1 ν) ( ) ln 2 R b, (1.1) λ where M = 3.06 [26], b = nm [27], G = 25.4 GPa [27], and ν = [26] for Al. The mean planar radius, R, and effective inter-precipitate distance λ, are, for a monodispersed assembly, given by [20, 22, 23]: and λ = R = π R, (1.2) 4 ( ) 2π 3f 2π R, (1.3) 4 where f is the precipitate volume fraction. Equations (1.2) and (1.3) are also good approximations for polydispersed arrays [23]. Deformation by dislocation shearing is expected to occur at small R, as observed experimentally in Al-Li [344, 345] and Al-Sc [115, 180] alloys with shearable, coherent precipitates. For the shearing mechanism, the increase in yield strength results from three contributions: (i) modulus hardening; (ii) coherency strengthening; and (iii) order strengthening. The strengthening due to modulus mismatch, σ ms, results from differences in the shear moduli of the precipitate and matrix phases and is given by [20]: ( f 1/2 ( ) R 3m/2 1 σ cs = M ( G) Γ) 3/2 b, (6.1) b where G is the modulus mismatch between the matrix and precipitate, Γ = 1 2 Gb2 is the line tension of the dislocation, and m = 0.85 is a constant [20]. The elastic constants of Al 3 Zr (L1 2 ) are not clear at present. Nakamura and Kimura [346] measured the elastic constants of Al 3 Zr (D0 23 ) single crystals and report G = 85.1 GPa, which is in good agreement with the extrapolated results of Zedalis et al. [347] for D0 23 -structured Al 3 Zr precipitates in α-al solid-solution. George et al. [348] studied the elastic constants of an Fe-stabilized Al 3 Zr L1 2 trialuminide, (Al 0.67, Fe 0.08 ) 3 Zr, and measured G = 68 GPa as reported also in references [49, 50]. George et al. [49,348] cautioned that this value of G may be affected by Kirkendall voids in their specimen,

171 SECTION 6.5 DISCUSSION 171 perhaps accounting for the differences in the reported values for the D and L1 2 -structured Al 3 Zr trialuminides. Their measured value of 68 GPa for Fe-modified Al 3 Zr (L1 2 ) is, however, identical to that of Al 3 Sc (L1 2 ) [ ]. Taking the smaller value, G = 68 GPa, results in a value of G = 42.6 GPa entering Eq.(6.1); this has the added benefit of providing a conservative estimate for the magnitude of σ ms, thereby over-predicting the critical R at which the transition from shearing to looping occurs. Coherency strengthening, σ cs, arises through the elastic strain-field interactions between a coherent precipitate and a dislocation and is given by [20] 3 : ( ) R fb 1/2 σ cs = M χ(ɛg) 3/2, (6.2) Γ where χ = 2.6 [20], ɛ is a mismatch parameter approximated by 2 3δ; δ = is the lattice parameter mismatch at room temperature defined in Table 1.1 for Al 3 Zr (L1 2 ). Order strengthening is due to the formation of an antiphase boundary (APB), which occurs when a matrix dislocation shears an ordered precipitate. At peak strength, σ os is given by [20, 23]: σ os = M 0.81 γ AP B 2b ( ) 3πf 1/2, (6.3) where γ AP B 0.5 J m -2 is an average value of the Al 3 Sc APB energy for the (111) plane taken from several reported values [49, 348, 349, 351, 353]. A comparable APB energy is assumed for Al 3 Zr (L1 2 ). Equations (1.1), (6.1), (6.2), and (6.3) each depend on the precipitate volume fraction, f. For a homogeneous alloy containing 0.1 at.% Zr, the equilibrium volume fraction of precipitated Al 3 Zr would be f = from the lever rule, assuming negligible solubility at the aging temperature. The solutes in the present alloys are microsegregated, however, with the dendrite centers enriched approximately twofold in Zr based on the measured EDS linescans in Figure 5.5. The equilibrium volume fraction of Al 3 Zr precipitates in the dendrites is therefore approaching 3 There is some confusion regarding the expression for σ cs, Eq.(6.2), used in previous studies on Al-Sc alloys by Marquis et al. [180,352] and Fuller et al. [115,319]. In those studies, the effective value of Γ entering Eq.(6.2) is 0.18Gb 2, attributed to Ardell [20]. Ardell, however, does not use this expression for Γ; indeed, as he discusses in detail, Γ is not constant but instead depends on the character of the dislocation; Γ = 1 2 Gb2 is a reasonable approximation used by Brown and Ham [343]. The inconsequential discrepancy between Γ = 1 2 Gb2 used presently, and Γ = 0.18Gb 2 used previously [115, 180, 319, 352], can easily be accommodated by the value of χ chosen, which varies between 2 and 3 for various theories [20]. 8

172 CHAPTER 6 MICROSTRUCTURAL COARSENING IN AL-ZR AND AL-ZR-TI ALLOYS, T 450 C 172 f = There is also the assumption of negligible solid-solubility. For the isochronally aged specimens considered presently, the aging temperature is progressively raised to 600 C, with a concomitant (and significant) decrease in equilibrium volume fraction of precipitates (assuming equilibrium is reached during the 3 h aging treatment). The volume fraction of Al 3 Zr (L1 2 ) is further reduced during the transformation to the equilibrium D0 23 structure, which commences by 525 C, Figure 6.14(b). Finally, a small amount of Ti is incorporated into the precipitates, shown in Chapter 4, and Ti also influence the metastable Al-Zr solvus, discussed in Figure While these effects are small, as evidenced by the isothermal age hardening curves in Figures where Ti has little effect on the strength, ternary additions will nevertheless influence f. Which volume fraction, f, to use for modeling the theoretical contributions to strength is therefore not trivial. For isochronally aged specimens, the assumption of negligible solidsolubility is reasonably valid, at least in the peak-aged condition, since precipitates are nucleated at the lowest possible temperature. As coarsening commences at higher temperatures, the equilibrium volume fraction of L1 2 precipitates is reduced by increasing solubility of solute(s) and the transformation to D0 23. An advantage of isochronal aging, however, is that the relative fractions of dendritic and interdendritic regions remains relatively constant (this is not true of isothermal aging, where the initial aging temperature has a strong influence on the volume fraction of the precipitate-rich dendrites, as discussed in Figure 6.21). As a point of reference, a value of f = is assumed, which is approximately the total volume fraction (including precipitatefree regions) of Al 3 Zr or Al 3 (Zr 1 x Ti x ) precipitates in the alloy at peak strength on the isochronal aging curve. This f will under-predict the strengthening in the precipitate-rich dendrites, but will over-predict the strengthening at higher temperatures because of increased solid-solubility and L1 2 to D0 23 transformation. Figure 6.20 displays the theoretical and measured increases in strength of isochraonallyaged Al-0.1Zr(c) and Al-0.1Zr-0.1Ti(c) as a function of the precipitate radius, R. The measured strength increment is compared to the value for the as-cast, unaged specimens using a conversion factor of 3 between Vickers microhardness and strength [354]. Calculations for the shearing mechanisms, σ ms + σ cs and σ os, are applicable for R 3 nm, confirming that shearing is not operative for the data in Table 6.3. These conclusions are in agreement with prior results on Al-0.18Sc [180] and several Al-Sc-Zr alloys [115], for which Orowan dislocation looping is the main strengthening mechanism for R 2 nm.

173 SECTION 6.5 DISCUSSION σ m s + σ c s A l-0.1 Z r(c ) A l-0.1 Z r-0.1 T i(c ) f = S tr e n g th e n in g in c r e m e n t (M P a ) σ o s σ o r M e a n p r e c ip ita te r a d iu s (n m ) Fig. 6.20: Vickers microhardness yield stress increment vs. mean precipitate radius R for Al-0.1Zr(c) and Al-0.1Zr-0.1Ti(c) isochronally aged at 450, 525, and 575 C. The curves represent predictions of Eqs.(1.1), (6.1), (6.2), and (6.3) for f = Table 6.4: Mean precipitate radii, R, observed in the dendrite centers after different thermal histories. Alloy Aging treatment Vickers microhardness (MPa) Mean precipitate radius, R (nm) Al-0.1Zr(b) 425 C for 1600 h 319 ± 12 a 10.9 ± 1.9 e Al-0.1Zr-0.1Ti(b) 351 ± 10 b 11.6 ± 2.3 e Al-0.1Zr(b) 500 C for 100 h (after 375 C for 100 h) 325 ± 17 c 12.3 ± 2.9 f Al-0.1Zr-0.1Ti(b) 319 ± 12 c 12.0 ± 4.0 f Al-0.1Zr(c) Isochronally aged to 525 C 327 ± 13 d 6.8 ± 1.8 g Al-0.1Zr-0.1Ti(c) 341 ± 8 d 6.6 ± 1.9 g a Figure 5.7. b Figure 5.9. c Figure 6.1. d Figure e Table 5.3. f Table 6.2. g Table 6.3.

174 CHAPTER 6 MICROSTRUCTURAL COARSENING IN AL-ZR AND AL-ZR-TI ALLOYS, T 450 C 174 For the R values measured in Table 6.3, the experimentally-measured strength is much less than that predicted by Eq.(1.1), indicating that the observed strengthening is not directly controlled by the Orowan bypass mechanism. This is illustrated more dramatically by the data in Table 6.4, which compares experimentally-measured precipitate radii and Vickers microhardness from Al-0.1Zr and Al-0.1Zr-0.1Ti (at.%) alloys reported previously in Tables 5.3, 6.2, and 6.3. The alloys in Table 6.4 exhibit similar hardness values, yet the precipitate radii differ substantially depending on the alloy thermal history. Specimens isochronally aged to 525 C exhibit smaller radii than those aged isothermally at 425 C or in a sequential 375 C C aging treatment, despite the similar strength. The data in Figure 6.20 and Table 6.4 indicate that strengthening cannot be predicted from the measured R, assuming precipitate bypass by Orowan dislocation looping. Precipitation strengthening is instead probably occurring on multiple length scales, as might be anticipated considering the nanometer-scale of the precipitates and the micrometer-scale of the precipitate-rich dendrites. Table 6.4 also demonstrates that thermal history has a strong influence on alloy strength Effect of aging temperature Increasing the aging temperature (decreasing the undercooling) in any precipitation reaction reduces the chemical driving force for nucleation leading to: (i) larger precipitates (because the critical radius for nucleation is larger); and (ii) reduced equilibrium volume fraction of the dispersed phase. Thus, precipitation-strengthening (at ambient temperature) is generally reduced because the precipitates are at a smaller number density and volume fraction. Higher aging temperatures therefore lead to reduced precpitation hardening, as observed in Figures The effect of decreasing strength with decreasing precipitate volume fraction must also be interpreted within the context of the micrometer-scale dendritic distribution of Al 3 Zr or Al 3 (Zr, Ti) precipitates. Figure 6.21(a) shows schematic isoconcentration contours of the as-solidified α- Al solid-solution for a peritectic system (k 0 > 1) in which the cores of the dendritic cells are enriched in solute concentration compared to regions at the periphery (C 1 < C 2 < C 3 ). Figure 6.21(b) shows the expected precipitation behavior for such a cored solid-solution at three temperatures T 1 < T 2 < T 3. At the lowest aging temperature, T 1, all compositions C 1, C 2, and C 3 are supersaturated with respect to the solvus and hence these regions of the original dendritic α-al solid-solution may be expected to exhibit precipitation, as indicated by the dots in Figure

175 SECTION 6.5 DISCUSSION 175 C 1 C 1 < C 2 < C 3 T 3 S o lv u s C 3 C 2 C 2 C 1 C 3 T e m p e r a tu r e T 2 T 1 (a ) (b ) C 1 C 2 C 3 C o m p o s itio n Fig. 6.21: Schematic isoconcentration contours and the expected precipitation behavior for a dendritically-solidified α solidsolution in which k 0 > 1. (a) Isoconcentration contours corresponding to C 1 < C 2 < C 3 within the original α dendrites. (b) The expected precipitation behavior from the α solid-solution in (a) for three aging temperatures T 1 < T 2 < T 3. As aging temperature is increased, fewer isoconcentration contours have sufficient supersaturation to allow precipitation, with those that do indicated by dots. Therefore, for higher aging temperatures, the relative amount of interdendritic precipitate-free zones is expected to increase. 6.21(b). At the higher temperature, T 2, the region corresponding to C 1 is no longer supersaturated with solute and therefore precipitation should not occur. At the highest aging temperature, T 3, only the concentration C 3 is sufficiently rich to effect precipitation. Higher aging temperatures, therefore, not only result in smaller volume fractions of the precipitated phase (typical for any precipitation reaction), but because of the dendritic distribution of precipitates, the volume fraction of precipitate-free interdendritic regions is also increased. This also explains why during isochronal aging, Figure 6.12, strengths of the order of 415 MPa may be obtained for Al-0.1Zr after aging up to 450 C, whereas during isothermal aging, Figure 5.7, the attainable strength at 425 C is much less: ca. 300 MPa. A similar effect was observed by Sato et al. [271] during aging of an Al-0.17Zr (at.%) alloy at 450 C, whereby ramp heating at a controlled rate of 100 C h -1 produced a nearly 200 MPa increase in peak hardness compared to the same alloy aged directly at 450 C (540 MPa vs. 350 MPa) Effect of Ti additions The isothermal (Figures and 6.1) and isochronal (Figure 6.12) hardness curves, as well as the mean precipitate radii in Tables 5.3, 6.2, and 6.4 indicate that there is no benefit, in terms of coarsening resistance, of Al 3 (Zr 1 x Ti x ) (L1 2 ) precipitates as compared to Al 3 Zr (L1 2 ). Even

176 CHAPTER 6 MICROSTRUCTURAL COARSENING IN AL-ZR AND AL-ZR-TI ALLOYS, T 450 C L (A l) L + A l 3 Z r 0.2 a t.% T i 0.1 a t.% T i M e ta s ta b le L 1 2 S o lv u s C (A l) + A l 3 Z r A to m ic P e r c e n t Z r Fig. 6.22: Al-rich equilibrium binary Al-Zr phase diagram (solid lines) [123], and calculated (dotted lines) metastable L1 2 Al 3 Zr and Al 3 (Zr 1 x Ti x) solvus by Murray [313]. Additions of Ti reduce the solid-solubility of Zr. at 425 C (0.75T m ), the Al-Zr or Al-Zr-Ti alloys overage very sluggishly and the precipitates in the dendrites remain small, R 10 nm, after extended aging times (1,600 h or 9.5 weeks, Table 5.3). Only at or above 500 C (0.83T m ) do the alloys significantly overage within a 3 h treatment, Figure 6.12, and after 100 h at this temperature the microstructures are again indistinguishable, R 12 nm (Table 6.2). After 3 h at 575 C (0.91T m ), the Al 3 Zr (L1 2 ) or Al 3 (Zr 1 x Ti x ) (L1 2 ) precipitates still exhibit impressive coarsening resistance, R 17 nm (Table 6.3), with no improvements in coarsening from Ti additions. While Ti has no effect on the long-term stability of the Al 3 (Zr 1 x Ti x ) precipitates, there is a noticeable influence on the incubation time for nucleation, as evidenced in Figures 5.7, 5.9, and Moreover, the peak hardness of Al-0.1Zr-0.1Ti(b) is greater than that of Al-0.1Zr(b) at 425 C (Figures 5.7 and 5.9) and that of Al-0.1Zr-0.1Ti(c) is similarly greater than that of Al- 0.1Zr(c) during isochronal aging (Figures 6.12). The same effect was observed by Sato et al. [271]

177 SECTION 6.6 CHAPTER SUMMARY 177 in chill-cast Al-0.17Zr and Al-0.12Zr-0.15Ti (at.%) alloys aged at 450 C for up to 1,200 h. The Al-0.12Zr-0.15Ti alloy had a reduced incubation time for nucleation and also exhibited a larger peak strength (400 MPa) compared to Al-0.17Zr (350 MPa). Murray [313] has calculated that additions of Ti reduce the solid-solubility of Al 3 Zr (L1 2 ), Figure The resulting increased supersaturation (for a given Zr concentration) accounts for the reduced incubation time for nucleation in the Al-Zr-Ti alloys and also explains the larger peak strength, since the equilibrium volume fraction of precipitate-rich dendrites is greater. Titanium additions thus have a similar effect as isochronal aging in Al-Zr-based alloys. The effective supersaturation of Zr is enhanced during nucleation, increasing the volume fraction of the precipitate-rich dendrites and reducing the width of the precipitate-free interdendritic channels. Unfortunately, the variation in apparent dendritic cell size varies considerably throughout the alloy, Figure 5.6, and there is no practical way to quantify this phenomenon using traditional 2-D microscopy. 6.6 Chapter Summary The transformation of Al 3 Zr (L1 2 ) and Al 3 (Zr 1 x Ti x ) (L1 2 ) precipitates to their respective equilibrium D0 23 structures was investigated in conventionally-solidified Al-0.1Zr and Al-0.1Zr-0.1Ti (at.%) alloys isothermally aged at 500 C or isochronally aged C. The following results were obtained and discussed: There is no benefit from Ti additions, both in terms of coarsening resistance of the metastable L1 2 precipitates, or in delaying the L1 2 to D0 23 transformation. Both alloys overage rapidly at the same rate at or above 500 C, during which metastable L1 2 precipitates transform to their equilibrium D0 23 structure. The L1 2 to D0 23 transformation is limited primarily to dislocations and other defects, where heterogeneously-nucleated L1 2 -structured precipitates, fed by rapid pipe diffusion, coarsen by dissolution of surrounding L1 2 precipitates within a ca. 500 nm radius of the dislocation. The coarsening precipitates often exhibit antiphase boundaries (APBs) parallel to {100} planes, representing a transition to an imperfect equilibrium D0 23 structure. Preferential growth is similarly observed along 100 directions, resulting in elongated spheroids. The cigar-shaped precipitates eventually transform to disc-like D0 23 precipitates, ca. 200 nm in

178 CHAPTER 6 MICROSTRUCTURAL COARSENING IN AL-ZR AND AL-ZR-TI ALLOYS, T 450 C 178 diameter and 50 nm in thickness, exhibiting a cube-on-cube orientation relationship with α-al. The disc-like morphology of the equilibrium phase reflects the tetragonal symmetry of the D0 23 unit cell. The lattice parameter mismatch of Al 3 Zr (D0 23 ) is δ a = δ b = -0.88% and δ c = 6.92% (Table 1.1); the mismatch is most severe in the c-direction, favoring disc-like precipitates that are thin along c. The equilibrium D0 23 -structured precipitates form heterogeneously on dislocations because of a significant lattice parameter mismatch with the α-al solid-solution, δ = 2.89% (Table 1.1). The transformation does not occur spontaneously throughout the alloy, as evidenced in Figure 6.6 displaying spheroidal L1 2 precipitates, partially-transformed imperfect L1 2 precipitates with APBs, and disc-like D0 23 precipitates in close proximity. The transformation is also extraordinarily sluggish; even after aging at 575 C (0.91T m ) for 3 h, L1 2 -structured precipitates are observed, Figure Precipitation strengthening in segregated Al-Zr alloys occurs on multiple length scales: (i) on the nanometer-scale scale by an Orowan strengthening mechanism and (ii) on the micrometer-scale related to the volume fraction of the precipitate-rich dendrites. Microstructural coarsening similarly occurs on several length scales: (i) by Ostwald ripening (volume diffusion-controlled coarsening) of spheroidal nanometer-scale precipitates within the dendrites; and (ii) by local dissolution of the micrometer-scale precipitate-rich dendrites during the D0 23 transformation. Because strengthening depends on the volume fraction of the precipitate-rich dendrites, thermal history has a strong influence on the observed alloy strength. During isochronal aging, Figure 6.12, strengths of the order of 415 MPa may be obtained for Al-0.1Zr (at.%) after aging up to 450 C; during isothermal aging, Figure 5.7, the attainable strength at 425 C is only ca. 300 MPa. Unlike isothermal aging, isochronal aging nucleates precipitates at the lowest possible aging temperature where the supersaturation is greatest. Because of the corresponding large chemical driving force, the critical radius for nucleation is small and the equilibrium volume fraction of the precipitate-rich dendrites is large, promoting a larger number density of fine precipitates.

179 SECTION 6.6 CHAPTER SUMMARY 179 Ternary additions of Ti have no effect on the coarsening resistance of Al 3 (Zr 1 x Ti x ) (L1 2 ) as compared to Al 3 Zr (L1 2 ); Ti does, however, accelerate the incubation time for nucleation and also increases the peak hardness achieved during aging (Figures 5.7, 5.9, and 6.12). Titanium reduces the solid-solubility of Al 3 Zr (L1 2 ), Figure 6.22, and these are effects of solute supersaturation (chemical driving force and precipitate-rich dendrite volume fraction) and is not an indication of coarsening resistance.

180 180 CHAPTER 7 Creep of Al-Zr and Al-Zr-Ti Alloys Creep behavior at 300, 350, or 400 C of Al-0.1Zr and Al-0.1Zr-0.1Ti (at.%) alloys are presented and discussed. At all temperatures investigated, creep threshold stresses are observed and found to be much smaller than the ambient temperature yield stress, indicative of a climb-controlled bypass mechanism. For a given temperature, the ternary Al- Zr-Ti alloy exhibits a smaller threshold stress than the binary Al-0.1Zr alloy. This may be attributable to: (i) differences in texture arising from a reduced grain size in the ternary Al-Zr-Ti alloy; (ii) reduction in the lattice parameter mismatch of the Al 3(Zr 1 xti x) precipitates with the α-al matrix; or (iii) finer precipitates. Comparisons are made with Al-Sc and Al-Sc-M (M = Zr or Ti) alloys with similar radii and volume fractions of precipitates. 7.1 Introduction IN CHAPTER 5 the microstructure and ambient temperature mechanical properties of Al-0.1Zr and Al-0.1Zr-0.1Ti (at.%) alloys, during isochronal aging at 375, 400, or 435 C, are discussed. This chapter investigates the elevated mechanical properties of similar isothermally aged alloys by creep experiments performed at 300, 350, or 400 C. The creep behavior of precipitation- or dispersion-strengthened materials generally follows a power-law equation of the form [355]: ( ɛ = A ap σ nap exp Q ), (7.1) R g T where ɛ is the minimum, secondary strain rate, A ap is a constant, σ is the applied stress, n ap is the apparent stress exponent, Q ap is the apparent activation energy, R g is the universal gas constant, and T the absolute temperature. When the apparent stress exponent significantly exceeds that of the matrix, an athermal threshold stress, σ th, is assumed, below which creep is negligible. This

181 SECTION 7.2 EXPERIMENTAL PROCEDURES 181 Table 7.1: Compositions and aging conditions of the Al-Zr and Al-Zr-Ti alloys investigated by creep. Alloy Nominal comp. (at.%) Verified comp. (at.%) Aging conditions Creep temperature ( C) Zr Ti Zr Ti Al-0.1Zr(d) h at 400 C 300, 350, 400 Al-0.1Zr-0.1Ti(d) h at 400 C 300, 350, 400 leads to a modified power-law equation [356, 357]: ( ɛ = A (σ σ th ) n exp Q ), (7.2) R g T where A is a constant, n is the matrix stress exponent, and Q is the matrix creep activation enthalpy, which is usually equal to the activation enthalpy for self-diffusion. The threshold stress, σ th, arises from the increase in length of the dislocation during the climb bypass process for metals with non-shearable, coherent precipitates [30]. Since creep deformation is negligible for stresses below this threshold, σ th is analogous to the yield stress describing deformation at ambient temperature. It is thus desirable to have large values of σ th. Indeed, an upper bound for σ th is the Orowan stress, σ or, above which dislocations bypass dispersoids by bowing within their glide plane, and hence σ or controls the yield stress. 7.2 Experimental Procedures A series of binary Al-0.1Zr and ternary Al-0.1Zr-0.1Ti alloys were investigated; alloy designations, compositions, aging conditions prior to creep, and creep temperatures are summarized in Table 7.1. The alloys were prepared employing non-consumable electrode arc-melting, as described in Sections 3.2 and 4.2. The verified compositions in Table 7.1 were obtained by bulk chemical analysis performed by direct current plasma emission spectroscopy (ATI Wah Chang, Albany, OR). Creep specimens in the shape of parallelepipeds (ca mm 3 ) were electrode-dischargemachined (EDM) with their axes perpendicular to the surface of the ingot which had been in contact with the water-cooled copper crucible during arc-melting. The height of the creep specimen spans nearly all of the button ingot height. Constant-load compression creep tests, corre-

182 CHAPTER 7 CREEP OF AL-ZR AND AL-ZR-TI ALLOYS 182 Table 7.2: Vickers microhardness of the aged Al-Zr and Al-Zr-Ti alloys before and after creep. Alloy Vickers microhardness (MPa) Creep temperature and time As-aged Post-creep Al-0.1Zr(d) 335 ± ± h at 300 C 366 ± h at 400 C Al-0.1Zr-0.1Ti(d) 401 ± ± h at 400 C sponding to compressive stresses in the range of 4 30 MPa, were performed in air at C. A superalloy creep cage translated tensile loads in the pull-rods to compressive stresses on the specimen. Frictional effects on the end-loaded specimens were minimized by using boron nitride coated alumina platens in the creep cage. Specimen temperature was measured in the three-zone furnace with a temperature stability of ± 1 C after a 1 h soak at the test temperature. Specimen strains were calculated from extensometric displacements of cage plattens measured using a linear variable differential transducer (LVDT) with a resolution of 2.5 µm. Continuous measurements were made on each specimen by monitoring the strain rate during the test; the data were collected in ca s increments depending on the strain rate. Once steady-state deformation was achieved, the load was increased, resulting in three to five data points suitable for determining a stress exponent from one specimen. Total specimen strain never exceeded ɛ = The alloys were age hardened isothermally at 400 C for 100 h prior to creep, which produces a peak-aged hardness of ca. 420 MPa in alloys Al-0.1Zr(b) and Al-0.1Zr(b)-0.1Ti(b) (Figures 5.7 and 5.9). The hardness of the present alloys prior to creep is slightly less, as indicated in Table 7.2: 335 ± 7 MPa for Al-0.1Zr(d) and 401 ± 15 MPa for Al-0.1Zr-0.1Ti(d). The reduced strength is consistent with the slightly smaller solute concentration in Al-0.1Zr(d) and Al-0.1Zr-0.1Ti(d) (Table 7.1) compared to Al-0.1Zr(b) and Al-0.1Zr-0.1Ti(b) (Table 5.1). The greater strength of the ternary alloy is, however, inconsistent with what is observed for Al-0.1Zr(b) and Al-0.1Zr-0.1Ti(b) (Figures 5.7 and 5.9), where the peak strengths of the alloys are indistinguishable at 400 C and below. The higher strength of Al-0.1Zr(d) may indicate: (i) a larger volume fraction of precipitates (or precipitate-rich dendrites); or (ii) smaller precipitates leading to increased Orowan strengthening at ambient temperature. Based on the results in Chapters 5 and 6, however, the latter explanation seems unlikely since at all aging temperatures investigated, Al-Zr and Al-Zr-

183 SECTION 7.3 EXPERIMENTAL RESULTS σ = 1 9 M P a ε = 4.8 x s -1 σ = 1 7 M P a ε = 6.3 x s S tr a in σ = 1 4 M P a ε = 1.8 x s -1 σ = 1 5 M P a ε = 2.1 x s Fig. 7.1: Variation of strain, ɛ, with time for alloy Al-0.1Zr(d) crept at 300 C for four levels of stress. These data correspond to the right-most curve in Figure 7.2. The strain rate measured at steady-state is also indicated for each load. Ti alloys are observed to have statistically identical precipitate radii. Moreover, strength is not controlled directly by the Orowan mechanism, Chapter 6, and therefore hardness is not a direct indication of precipitate radius. 7.3 Experimental Results A primary creep regime, where the strain rate decreases continuously with time, always precedes steady-state creep. As shown in Figure 7.1, the strain associated with primary creep, even near the threshold stress, can be large (of the order ɛ = ). This may represent unimpeded dislocation glide through the precipitate-free interdendritic channels, not unlike the deformation through γ matrix channels between γ precipitates in Ni-based superalloys [28,29]. Primary creep is eventually exhausted as deformation accumulates in the interdendritic regions, at which point steady-state creep ensues and dislocation climb around the precipitates in the dendritic regions controls the deformation. The minimum strain rate from the steady-state regime is plotted as a function of applied

184 CHAPTER 7 CREEP OF AL-ZR AND AL-ZR-TI ALLOYS 184 M in im u m s tr a in r a te (s -1 ) A p p lie d s tr e s s (M P a ) Fig. 7.2: Double logarithmic plot of minimum creep rate at 300, 350, or 400 C vs. applied stress, for Al-0.1Zr(d) and Al-0.1Zr-0.1Ti(d) aged for 100 h at 400 C. Each curve represents data from a single specimen; two specimens each of Al-0.1Zr(d) and Al-0.1Zr-0.1Ti(d) were tested at 400 C. Data for pure Al calculated from reference [27]. Table 7.3: Threshold stress, σ th, of the Al-Zr and Al-Zr-Ti alloys crept at 300, 350, or 400 C. Alloy Creep temperature ( C) σ th (MPa) Al-0.1Zr(d) Al-0.1Zr-0.1Ti(d)

185 SECTION 7.4 DISCUSSION ln (ε) ln (σ-σ th ) Fig. 7.3: ln( ɛ) vs. ln(σ σ th ) for the data in Figure 7.2. stress, Figure 7.2, for tests performed at 300, 350, or 400 C. Despite the non-uniform precipitate distribution and the associated precipitate-free interdendritic channels described in Chapter 5, the Al-Zr and Al-Zr-Ti alloys show remarkably high creep resistance at C. The apparent stress exponent, n, of both alloys is much higher than for pure Al (n = 4.4), indicative of the existence of a threshold stress. The value of this threshold stress is determined from linear fits of ( ɛ) 1/n vs. σ [358] using n = 4.4 [27] for dislocation creep of Al, and are given in Table 7.3 for the alloys studied. The validity of this approach is demonstrated in Figure 7.3, where the data at each temperature fall on straight lines with slopes near n = 4.4, the assumed value of the stress exponent. 7.4 Discussion Transformation to the equilibrium D0 23 structure Chapter 6 discusses that the transformation to the equilibrium D0 23 structure of Al 3 Zr or Al 3 (Zr 1 x Ti x ) occurs on dislocations and heterogeneous nucleation sites. Dislocations gener-

186 CHAPTER 7 CREEP OF AL-ZR AND AL-ZR-TI ALLOYS 186 ated during creep may therefore hasten the transformation of the strengthening precipitates, thereby reducing creep resistance. Post-creep Vickers microhardness measurements for creep experiments performed at 400 C for hundreds of hours, Table 7.2, indicate no loss in strength compared to the as-aged condition prior to creep, suggesting that the mechanism of microstructural coarsening on dislocations is not active at 400 C. This is consistent with the results in Chapter 6, which indicate that the L1 2 to D0 23 transformation occurs at ca. 500 C. Indeed, the hardness is actually increased after creep, which may indicate additional precipitation during the test Differences in creep behavior between Al-Zr and Al-Zr-Ti alloys As discussed in Section 6.5.5, additions of Ti to Al-Zr alloys accelerates the nucleation kinetics and increase the peak strength at aging temperatures greater than 400 C, attributable to a reduction in the solid-solubility of Zr. This decreases the incubation time for nucleation and increases the equilibrium volume fraction of the precipitated Al 3 (Zr 1 x Ti x ) (L1 2 ) phase. There is no improvement, however, in coarsening resistance of Al 3 (Zr 1 x Ti x ) precipitates as compared to Al 3 Zr during isothermal aging at C; even after extending aging times at 425 C (75% of the absolute melting temperature of Al), the Al-Zr or Al-Zr-Ti alloys investigated overage very sluggishly and the precipitates in the dendrites remain small, R 10 nm, Table 5.3. Because of the indistinguishable microstructure, it might be expected that the Al-Zr and Al- Zr-Ti alloys would behave similarly during creep. Figure 7.2, however, indicates that at 300, 350, or 400 C, the ternary Al-Zr-Ti alloy consistently creeps faster than the binary Al-Zr, despite the higher concentration of solutes (Table 7.1). This may be attributable to: (i) the reduced grain size in the ternary Al-Zr-Ti alloy; (ii) reduction in the lattice parameter mismatch of the Al 3 (Zr 1 x Ti x ) precipitates with the α-al matrix; or (iii) smaller precipitates. Each of these effects is discussed in the following. Crystallographic texture Figure 5.1 displays the as-cast macrostructure of alloys Al-0.1Zr(b) and Al-0.1Zr-0.1Ti(b), which are similar to those studied presently. The relative extents of equiaxed and columnar zones in the ingot cross-section varies between the two alloys: Al-0.1Zr(b) is comprised entirely of columnar grains whereas half of the ingot cross-section in Al-0.1Zr-0.1Ti(b) is equiaxed. The grain struc-

187 SECTION 7.4 DISCUSSION 187 ture of these alloys is shown more clearly in Figures 7.4 and 7.5, which display a series of optical micrographs for alloys Al-0.1Zr(b) and Al-0.1Zr-0.1Ti(b), respectively. It is conceivable that the poorer creep performance of the Al-Zr-Ti alloy is attributable to a macroscopic texture effect. It is well-known [ ] that dendrites, in the absence of heterogeneous nucleation sites, grow in crystallographically-preferred 100 directions, explaining the columnar structure shown in Figure 7.4. Because primary precipitates of Al 3 (Zr 1 x Ti x ) act as heterogeneous nucleants of α-al, the as-cast structure for Al-0.1Zr-0.1Ti(b) (Figure 7.5) is predominantly equiaxed and is therefore no longer textured along 100. This difference in crystallographic texture may explain the poorer creep performance of the ternary alloy. The refined grain structure in Figure 7.5 would also be more susceptible to grain boundary sliding, but this deformation mechanism seems unlikely given the existence of a threshold stress in the Al-Zr-Ti alloy. Reduced lattice parameter mismatch Marquis and Dunand [33] showed that elastic interactions due to precipitate-matrix modulus and lattice parameter mismatches can influence significantly the creep threshold stress for coherent precipitates, with σ th increasing with increasing lattice parameter mismatch, δ. This theory has been confirmed by numerous creep experiments on Al-Sc [180, 181], Al-Mg-Sc [182], Al-Sc-Zr [115], Al-Sc-Ti [118], and Al-Sc-RE (RE = Y, Dy, or Er) [359] alloys. It has been shown [289] that the lattice parameter of the metastable L1 2 Al 3 Zr phase may be reduced by additions of Ti, Figure 7.6. The reduced creep strength of the Ti-containing alloys, therefore, may also be attributable to a reduced lattice parameter mismatch between the Al 3 (Zr 1 x Ti x ) precipitates and the α-al matrix. Assuming Vegard s Law holds, the lattice parameter of Al 3 (Zr 1 x Ti x ) (L1 2 ) may be estimated, Figure 7.6, with δ = 0 (at room temperature) occurring at x = In Chapter 4 the compositions of Al 3 (Zr 1 x Ti x ) precipitates formed at 375 or 425 C were measured directly by 3-D atom-probe tomography (3-D APT) to be Al 3 (Zr 0.91 Ti 0.09 ) and Al 3 (Zr 0.83 Ti 0.17 ), respectively, resulting in calculated lattice parameter mismatches with α-al of δ = +0.43% and +0.22%, respectively, which is less than that of Al 3 Zr (L1 2 ), δ = +0.75%. It is important to note, too, that these mismatches are reduced further at elevated temperature due differences in coefficient of thermal expansion for Al 3 M and α-al, as shown experimentally by Harada and Dunand [121] for Al 3 Sc-based tria-

188 CHAPTER 7 CREEP OF AL-ZR AND AL-ZR-TI ALLOYS 188 Fig. 7.4: Series of optical micrographs illustrating the macrostructure of alloy Al-0.1Zr(b). The ingot crosssection is composed entirely of columnar grains (etched using Poultan s reagent). Fig. 7.5: Series of optical micrographs illustrating the macrostructure of alloy Al-0.1Zr-0.1Ti(b). The grain size is more equiaxed and refined than that of Al-0.1Zr(b), Figure 7.4 (etched using Poultan s reagent).

189 SECTION 7.4 DISCUSSION 189 A l 3 (Z r 1 -x T i x ) la ttic e p a r a m e te r (n m ) A l 3 Z r (L 1 2 ) L 1 2 A l 3 T i, A l 3 Z r S rin iv a s a n e t a l. ( ) L 1 2 A l 3 (Z r,t i) M a le k e t a l. ( ) L a ttic e p a ra m e te r o f p u re A l A l 3 T i (L 1 2 ) S to ic h io m e tr ic p a r a m e te r, x Fig. 7.6: Dependence of the lattice parameter (at ambient temperature) of the metastable Al 3 (Zr 1 x Ti x) (L1 2 ) phase on the stoichiometric parameter x. Measured lattice parameters of Al 3 (Zr 1 x Ti x) are from Malek et al. [289]. The lattice parameters of the metastable L1 2 Al 3 Zr (4.077 Å) and Al 3 Ti (3.967 Å) trialuminides are from Srinivasan et al. [94]. luminides. Precipitate size There is also a strong effect of precipitate size on σ th, as predicted by Marquis and Dunand [33] and verified experimentally in numerous studies on Al reinforced with Al 3 Sc (L1 2 ) precipitates [115, 118, , 359]. The threshold stress is essentially a trade-off between the Orowan stress (which decreases with precipitate size) and repulsive interactions from modulus and lattice parameter mismatches (which increase with precipitate size), resulting in an optimum precipitate size for creep. A difference in the Al 3 Zr (L1 2 ) and Al 3 (Zr 1 x Ti x ) (L1 2 ) precipitate radii in the Al- 0.1Zr(d) and Al-0.1Zr-0.1Ti(d) alloys studied presently may account for their disparate behavior in creep. Based on the microstructural study of similar alloys, Chapter 5, this seems unlikely since the microstructures of those alloys are shown to be identical after extended isothermal aging times at 425 C (Table 5.3). Moreover, as discussed in Chapter 6, the strengthening by the dendritically-distributed precipitates cannot be predicted completely by the Orowan dislocation looping mechanism; precipitate radius, therefore, is not indicated by the strength.

190 CHAPTER 7 CREEP OF AL-ZR AND AL-ZR-TI ALLOYS 190 M in im u m s tr a in r a te (s -1 ) A l-0.1 Z r-0.1 T i(d ) n = 4.4 A l-z r A l-z r-t i A l-s c A l-s c -T i A l-s c -Z r A l S c R = 3.0 n m A l S c R = 8.5 n m A l S c Z r R = 8.1 n m A l-0.1 Z r(d ) A l S c Z r R = 4.8 n m A l S c T i R = 5.8 n m A l S c T i A l S c T i R = 7.4 n m R = n m A p p lie d s tr e s s (M P a ) M in im u m s tr a in r a te (s -1 ) A p p lie d s tr e s s (M P a ) Fig. 7.7: Minimum creep rate at 300 C vs. applied stress, comparing data in Figure 7.2 to Al-Sc [180, 181], Al-Sc- Zr [115], and Al-Sc-Ti alloys [118]. Mean precipitate radius, R, is indicated for the Al-Sc alloys. Fig. 7.8: Minimum creep rate at 350 or 400 C vs. applied stress, comparing data in Figure 7.2 to Al-Sc-Ti alloys [360] Comparison to Al-Sc alloys Figures 7.7 and 7.8 compare the present creep results on Al-0.1Zr and Al-0.1Zr-0.1Ti alloys with those secured in the same laboratory on Al-Sc [180, 181], Al-Sc-Zr [115], and Al-Sc-Ti alloys [118] crept at 300 C, and Al-Sc-Ti alloys crept at 350 and 400 C [360], for alloys of comparable precipitate volume fraction and radius ( R ca nm). At 300 C, Figure 7.7, the Al-Sc based alloys generally outperform the Al-Zr based ones studied presently. The poorer performance of the Zr-containg alloys might reflect the inhomogeneous precipitate distributions discussed in Chapter 5. The lattice parameter mismatch with α-al of Al 3 Zr, δ = +0.75%, is also less than that of Al 3 Sc (L1 2 ), δ = +1.32% (Table 1.1), and so mismatch effects might also be contributing to the reduced σ th for the Zr-containing alloys. It is also conceivable that the present Zr-containing alloys are not aged to the optimum precipitate size for creep. Figure 7.7 indicates that Al-0.1Zr is comparable to Al-0.12Sc ( R = 3.0 nm) [180, 181]. Despite having a smaller volume fraction, Al-0.06Sc exhibits a much-improved threshold stress, attributable to the larger precipitate radii ( R = 8.1 nm). The alloys crept presently correspond approximately to the ambient temperature peak-aged strength, which is typically not optimum for creep, as discussed. The present Al-0.1Zr(d) and Al-0.1Zr-0.1Ti(d) alloys were not investigated

191 SECTION 7.5 CHAPTER SUMMARY 191 by TEM; the precipitate sizes are assumed to be similar to those in Table 5.3. Figure 7.8 compares the current results to those of van Dalen et al. [360] on Al-0.06Sc-0.06Ti alloys crept at 350 and 400 C. Here again the Zr-containing alloys exhibit smaller values of σ th ; Al-0.06Sc-0.06Ti at 400 C is comparable to Al-0.1Zr at 350 C. 7.5 Chapter Summary Al-0.1Zr and Al-0.1Zr-0.1Ti (at.%) alloys crept at 300, 350, or 400 C exhibit creep threshold stresses σ th ca MPa, indicative of a climb-controlled bypass mechanism. For a given temperature, the ternary Al-Zr-Ti alloy exhibits a smaller threshold stress than the binary Al-0.1Zr alloy. This disparity is probably attributable to: (i) a different texture due to the reduced grain size in the ternary Al-Zr-Ti alloy; or (ii) reduction in the lattice parameter mismatch of the Al 3 (Zr 1 x Ti x ) precipitates with the α-al solid-solution. Comparisons are made with Al-Sc and Al-Sc-M (M = Zr or Ti) alloys with similar radii and volume fractions of precipitates.

192 192 CHAPTER 8 Final Remarks and Suggestions for Future Work 8.1 Introduction THIS THESIS has demonstrated that Al-Zr based alloys offer tremendous potential for developing novel high-strength, high-temperature Al-based alloys. Coherent nanometer-scale precipitates, Al 3 Zr (L1 2 ), are remarkably stable at temperatures up to 500 C (0.83T m ), well beyond the capabilities of existing creep-resistant Al alloys. There are, however, significant limitations related to the peritectic nature of Al-Zr alloys and the associated: (i) high casting temperatures necessary to melt the alloy above the liquidus; (ii) fast cooling rates necessary to suppress primary Al 3 M; and (iii) inability to homogenize the alloys when the dendrites are supersaturated beyond the maximum solid-solubility. This chapter offers new directions for developing improved creep-resistant alloys by considering: (i) judicious aging treatments for maximizing strength and creep resistance of Al-Zr alloys; (ii) an alternate system to Al-Zr; or (iii) ternary and higher-order alloying additions to Al-Zr alloys. 8.2 Improvements with Judicious Aging Chapter 6 indicates that precipitation strengthening in segregated Al-Zr alloys occurs on multiple length scales: (i) on the nanometer-scale scale by an Orowan strengthening mechanism and (ii) on the micrometer-scale related to the volume fraction of the precipitate-rich dendrites. The thermal aging history consequently has a strong influence on the observed alloy strength. Compared to isothermal aging, isochronal aging nucleates precipitates at the lowest possible aging temperature where the solute supersaturation and chemical driving force is greatest, resulting in: (i) smaller precipitates since the critical radius for nucleation is reduced; and (ii) a larger volume fraction of precipitate-rich dendrites since the supersaturation is greater at lower

193 SECTION 8.3 ALLOYS CONTAINING TA 193 temperatures (Figure 6.21). In Chapter 7, the maximum temperature investigated by creep was 400 C, primarily because beyond this temperature a precipitous drop in hardness was observed in the isothermally aged specimens (Figures ). Isochronal aging, Figure 6.12, indicates that Al-Zr alloys retain strength to ca. 450 C, beyond which precipitate coarsening occurs along with transformation to the equilibrium D0 23 structure. Furthermore, alloys isochronally aged to 450 C (Figure 6.12) are stronger than similar alloys aged isothermally at 400 C (Figures 5.7 and 5.9) at ambient temperature. Because of the larger Orowan stress, isochronally aged creep specimens should also exhibit a larger threshold stress in creep than the alloys studied in Chapter 7. Moreover, isochronally aging creep specimens to 450 C (0.77T m ) permits creep experiments at this temperature, where the microstructures in Al-Zr alloys are still intrinsically stable. 8.3 Alloys Containing Ta Based on the criteria established in Chapter 1, Al-Ta alloys seem particularly intriguing for developing new high-strength high-temperature Al-based alloys. With exception to the Al-Sc system, Al-Ta has the most attractive solvus for precipitation hardening of the systems in Figure 1.4, with a relatively large maximum solid-solubility (0.24 at.% Ta) that decreases significantly with temperature. The maximum solid-solubility of Ta in α-al is approximately three times that of Zr, suggesting that Al-Ta alloys may be more amenable to solutionizing prior to precipitation aging (Al-Zr alloys cannot be homogenized as discussed in Appendix B). This would mitigate many of the problems associated with inhomogeneous precipitate distributions that occur from aging as-cast Al-Zr alloys. As a Group 5 element, Ta should be an extremely slow diffuser in α-al (Figure 1.7), and as with Al 3 V and Al 3 Nb, trialuminides based on the other Group 5 elements, Al 3 Ta could conceivably exist as a metastable L1 2 structure. Table 1.1 indicates, however, that Al 3 Ta (L1 2 ) has not been reported in the scientific literature. The Al-Ta system is peritectic, however, which introduces complications with respect to castability, discussed in Chapters 1 and 2. Nevertheless, peritectic Al-Ta alloys may be conventionally cast, provided one appreciates the necessity for high casting temperatures and relatively fast solidification rates to suppress primary precipitation of Al 3 Ta.

194 CHAPTER 8 FINAL REMARKS AND SUGGESTIONS FOR FUTURE WORK 194 There is much to glean from a study on Al-Ta alloys. Tracer diffusivity studies of Ta in α- Al are unknown to the author. As mentioned above, no information is available concerning the existence of Al 3 Ta (L1 2 ). There are also no reported studies investigating this system with respect to precipitation hardening. Since the lattice parameter of Al 3 Ta (L1 2 ) is unknown and, anticipating extremely sluggish diffusion kinetics, a prudent approach might initially involve Al-Ta alloys produced by rapid solidification processing (RSP), resulting highly-supersaturated solid-solutions and bypassing many of the challenges in conventionally casting peritectic alloys. The concomitant large chemical driving force for nucleation will furthermore help to overcome any significant elastic strain energy hindering nucleation, should one exist if the lattice parameter mismatch of Al 3 Ta (L1 2 ) with α-al is large (this is a limiting factor in developing conventionally-cast alloys based on Al- Ti, Chapter 3). Assuming Al 3 Ta (L1 2 ) precipitates form during aging, coarsening studies may be performed to estimate the diffusivity of Ta in α-al, and transmission electron microscopy (TEM) may elucidate the lattice parameter of Al 3 Ta (L1 2 ), and its coherency with α-al. From these results, the Al-Ta system may then be evaluated from the standpoint of developing a conventionallysolidified alloy. Tantalum may also be beneficial as a ternary addition to Al-Sc and Al-Zr based alloys, where the anticipated slower diffusion of Ta may confer improved coarsening resistance of Al 3 Sc (L1 2 ) or Al 3 Zr (L1 2 ). Partitioning of Ta to these precipitates might also alter the lattice parameter mismatch with α-al, thereby further affecting coarsening resistance as well as influencing the creep threshold stress. 8.4 Alloying Additions to Al-Zr Alloys Inhomogeneously dendritically-distributed Al 3 Zr (L1 2 ) precipitates are a significant problem in commercial wrought alloys, where Zr is added as a recrystallization inhibitor [78, 168, ]. During solutionizing, which is typically performed at ca. 500 C, coherent Al 3 Zr (L1 2 ) precipitates form, which inhibit recrystallization by pinning migrating grain boundaries. Because precipitation of Al 3 Zr (L1 2 ) occurs at such a high temperature, the interdendritic precipitate-free channels are larger than those observed in the present study. As in the present study, the precipitatefree interdendritic regions are deleterious since there is locally a lower resistance to recrystal-

195 SECTION 8.4 ALLOYING ADDITIONS TO AL-ZR ALLOYS 195 lization [361]. As a result, a number of recent studies have emerged investigating the effect of ternary alloying additions in mitigating the effects of an inhomogeneous distribution of Zr in commercial wrought Al alloys Al-Zr-Mg Alloys Calculations by Sigli [362] indicate that the metastable L1 2 Al-Zr solvus is strongly shifted to smaller solubilities by additions of Zn, Cu, Mg, and Li (in increasing order of efficacy). Robson and Prangnell [363] similarly demonstrated that additions of Zn, Cu, and Mg to commercial 7xxx alloys influences the supersaturation of Zr in α-al, thereby accelerating the precipitation kinetics of Al 3 Zr (L1 2 ) and ultimately influencing the recrystallization behavior. By decreasing the solid-solubility of Zr in α-al, these ternary additions should increase the volume fraction of precipitate-rich dendrites during post-solidification aging, as schematically illustrated in Figure Robson and Prangnell [363] predict that Mg additions have the greatest influence on minimizing the precipitate-free interdendritic regions. In addition to the increased volume fraction of preciptiate-rich dendrites created by a decreased solid-solubility of Zr, Mg additions also confer solid-solution strengthening, further improving creep performance as observed in Al-Sc- Mg alloys [182] Al-Zr-Sc Alloys There is currently strong interest in adding Sc to 7xxx and other commercial Al alloys to improve the Al 3 M (L1 2 ) precipitate distribution, thereby inhibiting recrystallization [337, 364, 365]. Robson [337] has shown that by combining a Sc, eutectic-forming solute (k 0 < 1), with Zr, a peritectic-forming solute (k 0 > 1), the precipitate-free regions associated with a Sc-free alloy may be eliminated. His argument, supported by experimentally-measured solute concentration profiles across the dendrites, is that during solidification Zr and Sc solute atoms segregate to the dendrite interiors and exteriors, respectively, effectively filling-in the interdendritic regions with Sc, which forms Al 3 Sc (L1 2 ) precipitates during post-solidification aging. A similar effect was observed by Lieblich and Torralba [366,367] with Al-Li-Ti alloys, where Ti (a peritecticforming solute) segregates to the dendrite interiors and Li (a eutectic-forming solute) segregates to the periphery. A further advantage of combining Zr and Sc is the improved coarsening resistance of Al 3 (Sc 1 x Zr x )compared to Al 3 Sc (L1 2 ) [ ]. Finally, Sc has been calculated to

196 ) ) CHAPTER 8 FINAL REMARKS AND SUGGESTIONS FOR FUTURE WORK V ic k e r s m ic r o h a r d n e s s (M P a ) A l-0.1 Z r-0.1 S c (a ) A l-0.1 S c A l-0.1 Z r A l-0.1 S c (h o m o g e n iz e d ) A l-0.1 Z r-0.1 S c (b ) V ic k e r s m ic r o h a r d n e s s (M P a ) A l Z r S c (a ) A l S c A l S c (h o m o g e n iz e d ) A l Z r S c (b ) A l Z r -1 E le c tr ic a l c o n d u c tiv ity (M S m A l-0.1 S c A l-0.1 Z r-0.1 S c (a ) A l-0.1 S c (h o m o g e n iz e d ) A l-0.1 Z r A l-0.1 Z r-0.1 S c (b ) -1 E le c tr ic a l c o n d u c tiv ity (M S m A l S c A l Z r S c (a ) A l S c (h o m o g e n iz e d ) A l Z r A l Z r S c (b ) A s -c a s t A s -c a s t Fig. 8.1: Vickers microhardness and electrical conductivity evolution during isochronal aging (3 h at each temperature) of Al-0.1Zr, Al-0.1Sc and Al-0.1Zr-0.1Sc (at.%) [369]. Fig. 8.2: Vickers microhardness and electrical conductivity evolution during isochronal aging (3 h at each temperature) of Al-0.06Zr, Al-0.06Sc and Al-0.06Zr-0.06Sc (at.%) [369]. stabilize the L1 2 structure of Al 3 Zr [368]. This thesis is closed by returning to the conclusions posited in Chapter 1. Of all the solutes capable of forming L1 2 -structured Al 3 M precipitates, Sc and Zr were shown to exhibit the most promise for developing castable, creep-resistant Al-based alloys. Scandium is singular among the transition metal solutes in that it forms a thermodynamically stable cubic L1 2 trialuminide, providing the highest increment of strengthening per atomic percent of any alloying element added to Al [178]. Zirconium is an extremely slow diffuser in α-al, Figure 1.7, and forms metastable L1 2 -structured Al 3 Zr precipitates that exhibit impressive thermal stability to ca. 500 C, discussed in Chapters 5 and 6.

197 SECTION 8.4 ALLOYING ADDITIONS TO AL-ZR ALLOYS 197 Figures 8.1 and 8.2 present results of a preliminary study [369] comparing the precipitation behavior in several binary Al-Zr and Al-Sc alloys during 3 h isochronal aging, as monitored by Vickers microhardness and electrical conductivity. Precipitation of Al 3 Sc (L1 2 ) commences between 200 and 250 C, with peak hardness occurring at 325 C; precipitation of Al 3 Zr (L1 2 ) is comparatively delayed, commencing between 350 and 375 C with a peak at C, depending on the supersaturation. Figures 8.1 and 8.2 summarize, in a nutshell, the merits of the two systems: Al-Sc providing improved precipitation hardening and Al-Zr conferring improved thermal stability. Ternary Al-Zr-Sc alloys containing equiatomic fractions of Zr and Sc (2:1 Zr:Sc by weight) are also studied in Figures 8.1 and 8.2, where a pronounced synergistic effect is observed when both Zr and Sc are added to Al alloys. At 325 C, where peak hardness occurs in the binary Al-Sc alloys, Zr additions result in a secondary increase in strength that is attributed to precipitation of Al 3 Zr (L1 2 ) at higher temperatures. The ternary alloy reaches peak strength at 400 C, delaying overaging by over 100 C compared to the Zr-free alloy. These results are preliminary and are not entirely understood at present. The increased strength of the Al-Zr-Sc alloys compared to their respective binaries may be attributable to the phenomenon observed by Robson [337], where the precipitate-free interdendritic channels are filled with Al 3 Sc (L1 2 ) precipitates (the precipitate-free interdendritic regions limit the precipitation strengthening of Al 3 Zr (L1 2 ) in Al-Zr alloys, Chapter 6). Preliminary 3-D atom-probe tomographic analyses [369] on Al-0.1Zr-0.1Sc (at.%) alloys indicate that the increase in strength comes from segregation of Zr to pre-existing Al 3 Sc precipitates, thereby decreasing the interprecipitate distance, λ, and increasing the yield stress of the alloy. Perhaps it is a combination of both phenomena.

198 198 REFERENCES [1] M. E. Fine, Precipitation Hardening of Aluminum Alloys, Metall Trans A, 6(4), pp (1975). [2] C. M. Adam, Structure/Property Relationships and Applications of Rapidly Solidified Aluminum Alloys, in Rapidly Solidified Amorphous and Crystalline Alloys (edited by B. H. Kear, B. C. Giessen, and M. Cohen), pp , Elsevier, Boston (1982). [3] W. M. Griffith, J. Sanders, R E, and G. J. Hildeman, Elevated Temperature Aluminum Alloys for Aerospace Applications, in High-Strength Powder Metallurgy Aluminum Alloys (edited by M. J. Koczak and G. J. Hildeman), pp , TMS AIME, Warrendale (1982). [4] Y. W. Kim and W. M. Griffith (eds.), Dispersion Strengthened Aluminum Alloys, TMS, Warrendale (1988). [5] D. J. Skinner and K. Okazaki, High Strength Al-Fe-V Alloys at Elevated Temperatures Produced by Rapid Quenching from the Melt, Scripta Metall, 18(9), pp (1984). [6] D. J. Skinner, R. L. Bye, D. Raybould, and A. M. Brown, Dispersion Strengthened Al-Fe-V-Si Alloys, Scripta Metall, 20(6), pp (1986). [7] D. J. Skinner, The Physical Metallurgy of Dispersion Strengthened Al-Fe-V-Si Alloys, in Dispersion Strengthened Aluminum Alloys (edited by Y. W. Kim and W. M. Griffith), pp , TMS, Warrendale (1988). [8] D. J. Skinner, K. Okazaki, and C. M. Adam, Physical Metallurgy and Mechanical Properties of Aluminum Alloys Containing Eight to Twelve Weight Percent Iron, in Rapidly Solidified Powder Aluminum Alloys, ASTM STP 890 (edited by M. E. Fine and E. A. Starke), pp , ASTM, Philadelphia (1986). [9] G. Thursfield and M. J. Stowell, Mechanical Properties of Al-8wt.%Fe-Based Alloys Prepared by Rapid Quenching from the Liquid State, J Mater Sci, 9(10), pp (1974). [10] Y. W. Kim and W. M. Griffith, Annealing Behavior and Tensile Properties of Elevated-Temperature PM Aluminum Alloys, in Rapidly Solidified Powder Aluminum Alloys, ASTM STP 890 (edited by M. E. Fine and E. A. Starke), pp , ASTM, Philadelphia (1986).

199 REFERENCES 199 [11] E. Hornbogen and E. A. Starke, Theory Assisted Design of High Strength Low Alloy Aluminum, Acta Metall Mater, 41(1), pp (1993). [12] A. W. Zhu, B. M. Gable, G. J. Shiflet, and E. A. Starke, The Intelligent Design of Age Hardenable Wrought Aluminum Alloys, Adv Eng Mater, 4(11), pp (2002). [13] A. W. Zhu, B. M. Gable, G. J. Shiflet, and E. A. Starke, The Intelligent Design of High Strength, Creep- Resistant Aluminum Alloys, in Aluminum Alloys 2002: Their Physical and Mechanical Properties Pts 1-3, vol of Materials Science Forum, pp (2002). [14] E. Sahin and H. Jones, Extended Solubility, Grain Refinement and Age Hardening in Al - 1 to 13 wt.% Zr Rapidly Quenched from the Melt, in Rapidly Quenched Metals II (edited by B. Cantor), pp , The Metals Society, London (1978). [15] J. E. Hatch, Aluminum: Properties and Physical Metallurgy, American Society for Metals, Metals Park (1984). [16] H. Jones, Phenomenology and Fundamentals of Rapid Solidification of Aluminium Alloys, in Dispersion Strengthened Aluminum Alloys (edited by Y. W. Kim and W. M. Griffith), pp , TMS, Warrendale (1988). [17] F. H. Froes, Y. W. Kim, and S. Krishnamurthy, Rapid Solidification of Lightweight Metal Alloys, Mater Sci Eng A, 117, pp (1989). [18] E. Orowan, Theory of Yield without Particle Shear, in Symposium on Internal Stresses in Metals and Alloys, p. 451, Institute of Metals, London (1948). [19] R. F. Decker, Alloy Design Using Second Phases, Metall Trans, 4, pp (1973). [20] A. J. Ardell, Precipitation Hardening, Metall Trans A, 16(12), pp (1985). [21] J. W. Martin, Micromechanisms in Particle Hardened Alloys, Cambridge University Press, Cambridge (1980). [22] E. Nembach, Particle Strengthening of Metals and Alloys, John Wiley & Sons, New York (1997). [23] A. Ardell, Ch. 12: Intermetallics as Precipitates and Dispersoids in High-Strength Alloys, in Intermetallic Compounds: Principles and Practice, Vol. 2 (edited by J. H. Westbrook and R. L. Fleischer), pp , John Wiley & Sons, Chichester (1994). [24] J. W. Martin, Precipitation Hardening, 2nd edn., Butterworth-Heinemann, Boston (1998). [25] P. Hirsch and F. Humphreys, Plastic Deformation of Two-Phase Alloys Containing Small Nondeformable Particles, in Physics of Strength and Plasticity (edited by A. S. Argon), pp , MIT Press, Cambridge, MA (1969).

200 REFERENCES 200 [26] M. A. Meyers and K. K. Chawla, Mechanical Metallurgy: Principles and Applications, Prentice-Hall, New Jersey (1984). [27] H. J. Frost and M. F. Ashby, Deformation-Mechanism Maps: The Plasticity and Creep of Metals and Ceramics, Pergamon Press, New York (1982). [28] T. M. Pollock and A. S. Argon, Creep Resistance of CMSX-3 Nickel Base Superalloy Single Crystals, Acta Metall Mater, 40(1), pp (1992). [29] J. H. Westbrook, Superalloys (Ni-base) and Dislocations An Introduction, in Vol. 10: L1 2 Ordered Alloys (edited by F. R. N. Nabarro and M. S. Duesberry), Dislocations in Solids, pp. 1 26, Elsevier, Amsterdam (1996). [30] J. Rosler and E. Arzt, The Kinetics of Dislocation Climb over Hard Particles I. Climb without Attractive Particle-Dislocation Interaction, Acta Metall, 36(4), pp (1988). [31] E. Arzt, Creep of Dispersion Strengthened Materials: A Critical Assessment, Res Mech, 31(4), pp (1991). [32] E. Arzt, in Mechanical Properties of Metallic Composites (edited by S. Ochiai), pp , Marcel Dekker, New York (1994). [33] E. A. Marquis and D. C. Dunand, Model for Creep Threshold Stress in Precipitation-Strengthened Alloys with Coherent Particles, Scripta Mater, 47(8), pp (2002). [34] I. M. Lifshitz and V. V. Slyozov, The Kinetics of Precipitation from Supersaturated Solid Solutions, J Phys Chem Solids, 19(1/2), pp (1961). [35] C. Wagner, Theorie der Alterung von Niederschlägen durch Umlösen, Z Elektrochemie, 65(7/8), pp (1961). [36] H. A. Calderon, P. W. Voorhees, J. L. Murray, and G. Kostorz, Ostwald Ripening in Concentrated Alloys, Acta Metall Mater, 42(3), pp (1994). [37] L. M. Angers, Y. C. Chen, M. E. Fine, J. R. Weertman, and M. S. Zedalis, Rational Design of High Temperature Aluminum Alloys, in Aluminum Alloys: Their Physical and Mechanical Properties, vol. 1 (edited by E. A. Starke and T. H. Sanders), pp , EMAS, Warley (1986). [38] M. E. Fine, Stability and Coarsening of Dispersoids in Aluminum Alloys, in Dispersion Strengthened Aluminum Alloys (edited by Y. W. Kim and W. M. Griffith), pp , TMS, Warrendale (1988). [39] S. K. Das, Ch. 8: Al-Rich Intermetallics in Aluminum Alloys, in Intermetallic Compounds: Principles and Practice, Vol. 2 (edited by J. H. Westbrook and R. L. Fleischer), pp , John Wiley & Sons, Chichester (1994).

201 REFERENCES 201 [40] T. B. Massalski, Binary Alloy Phase Diagrams, vol. 1, 2nd edn., ASM International, Materials Park (1990). [41] H. Okamoto, Phase Diagrams for Binary Alloys, ASM International, Materials Park (2000). [42] H. Okamoto, Phase Diagrams of Dilute Binary Alloys, ASM International, Materials Park (2002). [43] J. Royset and N. Ryum, Scandium in Aluminium Alloys, Int Mater Rev, 50(1), pp (2005). [44] J. C. Foley, J. H. Perepezko, and D. J. Skinner, Formation of Metastable L1 2 -Al 3 Y through Rapid Solidification Processing, Mater Sci Eng A, 179, pp (1994). [45] M. Yamaguchi and H. Inui, Ch. 7: Al 3 Ti and its L1 2 Variations, in Intermetallic Compounds: Principles and Practice, Vol. 2 (edited by J. H. Westbrook and R. L. Fleischer), pp , John Wiley & Sons, Chichester (1994). [46] R. L. Fleischer, D. M. Dimiduk, and H. A. Lipsitt, Intermetallic Compounds for Strong High- Temperature Materials: Status and Potential, Annual Review of Materials Science, 19, pp (1989). [47] K. S. Kumar, Ternary Intermetallics in Aluminum Refractory-Metal X Systems (X = V, Cr, Mn, Fe, Co, Ni, Cu, Zn), Int Mater Rev, 35(6), pp (1990). [48] W. S. Chang and B. C. Muddle, Trialuminide Intermetallic Alloys for Elevated Temperature Applications - Overview, Metal Mater Int, 3(1), pp (1997). [49] E. P. George, D. P. Pope, C. L. Fu, and J. H. Schneibel, Deformation and Fracture of L1 2 Trialuminides, ISIJ Int, 31(10), pp (1991). [50] K. S. Kumar, Advanced Intermetallics, in Physical Metallurgy and Processing of Intermetallic Compounds (edited by N. S. Stoloff and V. K. Sikka), pp , Chapman and Hall (1996). [51] V. A. Raman and K. Schubert, Uber den Aufbau einiger zu TiAl 3 verwandter Legierungsreihen, Z Metallkd, 56, pp (1965). [52] S. Mazdiyasni, D. B. Miracle, D. M. Dimiduk, M. G. Mendiratta, and P. R. Subramanian, High Temperature Phase Equilibria of the L1 2 Composition in the Al-Ti-Ni, Al-Ti-Fe, and Al-Ti-Cu Systems, Scripta Metall, 23(3), pp (1989). [53] A. E. Carlsson and P. J. Meschter, Relative Stabilities of L1 2 and D0 22 Structures in Ternary MAl 3 - Base Aluminides, J Mater Res, 5(12), pp (1990). [54] J. P. Nic, S. Zhang, and D. E. Mikkola, Observations on the Systematic Alloying of Al 3 Ti with 4 th Period Elements to Yield Cubic Phases, Scripta Metall Mater, 24(6), pp (1990).

202 REFERENCES 202 [55] S. Zhang, J. P. Nic, and D. E. Mikkola, New Cubic Phases Formed by Alloying Al 3 Ti with Mn and Cr, Scripta Metall Mater, 24(1), pp (1990). [56] D. E. Mikkola, J. P. Nic, S. Zhang, and W. W. Milligan, Alloying of Al 3 Ti to Form Cubic Phases, ISIJ Int, 31(10), pp (1991). [57] Y. Ma, T. Arnesen, J. Gjonnes, and J. Tafto, Laser Processed Al 3 Ti-Based Intermetallics: Al 5±X Ti 2±Y (Fe,Ni,or Cu) 1±Z, J Mater Res, 7(7), pp (1992). [58] C. Amador, J. J. Hoyt, B. C. Chakoumakos, and D. Defontaine, Theoretical and Experimental-Study of Relaxations in Al 3 Ti and Al 3 Zr Ordered Phases, Phys Rev Lett, 74(24), pp (1995). [59] Y. Nakayama and H. Mabuchi, Formation of Ternary L1 2 Compounds in Al 3 Ti-Base Alloys, Intermetallics, 1(1), pp (1993). [60] M. Takeda, T. Kikuchi, and S. Makihara, Stabilizing Effects of Third Elements on an L1 2 -Al 3 Ti Compound, J Mater Sci Lett, 18(8), pp (1999). [61] S. S. Nayak and B. S. Murty, Synthesis and Stability of L1 2 -Al 3 Ti by Mechanical Alloying, Mater Sci Eng A, 367(1-2), pp (2004). [62] P. B. Desch, R. B. Schwarz, and P. Nash, Formation of Metastable L1 2 Phases in Al 3 Zr and Al- 12.5%X-25%Zr (X = Li, Cr, Fe, Ni, Cu), J Less-Common Metals, 168(1), pp (1991). [63] K. I. Moon, K. Y. Chang, and K. S. Lee, The Effect of Ternary Addition on the Formation and the Thermal Stability of L1 2 Al 3 Zr Alloy with Nanocrystalline Structure by Mechanical Alloying, J Alloys Compounds, 312(1-2), pp (2000). [64] K. I. Moon, S. C. Kim, and K. S. Lee, A Study on the Microstructure of D0 23 Al 3 Zr and L1 2 (Al+12.5 at.% Cu) 3 Zr Intermetallic Compounds Synthesized by PBM and SPS, Intermetallics, 10(2), pp (2002). [65] K. I. Moon, S. H. Lee, and J. K. Seon, The Effect of Cu and Zn on the Phase Stability of L1 2 Al 3 Hf Intermetallic Compound Synthesized by Mechanical Alloying, Intermetallics, 10(8), pp (2002). [66] A. E. Carlsson and P. J. Meschter, Relative Stability of L1 2, D0 22, and D0 23 Structures in MAl 3 Compounds, J Mater Res, 4(5), pp (1989). [67] J. H. Xu and A. J. Freeman, Band Filling and Structural Stability of Cubic Trialuminides: YAl 3, ZrAl 3, and NbAl 3, Phys Rev B, 40(17), pp. 11,927 11,930 (1989). [68] J. H. Xu and A. J. Freeman, Bandfilling and Structural Stability of Trialuminides: YAl 3, ZrAl 3, and NbAl 3, J Mater Res, 6(6), pp (1991).

203 REFERENCES 203 [69] P. R. Subramanian, J. P. Simmons, M. G. Mendiratta, and D. M. Dimiduk, Effect of Solutes on Phase Stability in Al 3 Nb, in High-Temperature Ordered Intermetallic Alloys III, vol. 133 (edited by C. T. Liu, A. I. Taub, N. S. Stoloff, and C. C. Koch), pp , MRS, Pittsburgh (1989). [70] T. Ohashi and R. Ichikawa, On Rapid Solidified and Annealed Structures of Al %Ti Alloys, J Japan Inst Light Metals, 27(3), pp (1977). [71] S. Hori, H. Tai, and Y. Narita, Microstructures of Rapidly Solidified Al-Ti Alloys Containing Titanium up to 40% and its Thermal Stability, in Rapidly Quenched Metals (edited by S. Steeb and H. Warlimont), Proceedings of the Fifth International Conference on Rapidly Quenched Metals, pp , Elsevier Science Publishers, Wurzburg (1985). [72] A. Majumdar, R. H. Mair, and B. C. Muddle, Microstructure in Rapidly Quenched Al-Ti, Al-B and Al- Ti-B Alloys, in Science and Technology of Rapidly Quenched Alloys, vol. 80 (edited by M. Tenhover, W. L. Johnson, and L. E. Tanner), pp , Materials Research Society, Boston (1987). [73] J. F. Nie, A. Majumdar, and B. C. Muddle, Development of High-Temperature Dispersion Strengthening in Rapidly Quenched Al-Ti-X Alloys, Mater Sci Eng A, 179, pp (1994). [74] J. F. Nie and B. C. Muddle, High Temperature Precipitation Hardening in a Rapidly Quenched Al- Ti-Ni Alloy: Part 1. Precipitation Hardening Response, Mater Sci Eng A, 221(1-2), pp (1996). [75] J. F. Nie and B. C. Muddle, High Temperature Precipitation Hardening in a Rapidly Quenched Al- Ti-Ni Alloy: Part 2. Precipitate Characterisation, Mater Sci Eng A, 221(1-2), pp (1996). [76] O. Izumi and D. Oelschlägel, On Decomposition of a Highly Supersaturated Al-Zr Solid Solution, Scripta Metall, 3(9), pp (1969). [77] O. Izumi and D. Oelschlägel, Structural Investigation of Precipitation in an Aluminium Alloy Containing 1.1 wt.% Zr, Z Metallkd, 60(11), pp (1969). [78] N. Ryum, Precipitation and Recrystallization in an Al-0.5wt.%Zr Alloy, Acta Metall, 17(3), pp (1969). [79] E. Nes, Precipitation of Metastable Cubic Al 3 Zr-Phase in Subperitectic Al-Zr Alloys, Acta Metall, 20(4), pp (1972). [80] S. Hori, T. Kitagawa, T. Masutani, and A. Takehara, Structure and Phase Decomposition of Supersaturated Al-Zr Solid Solutions Rapidly Solidified, J Japan Inst Light Metals, 27(3), pp (1977). [81] S. Hori, T. Kondo, and S. Ikeno, Effects of Small Addition of Si on the Precipitation of Al 06%Zr Alloys, J Japan Inst Light Metals, 28(2), pp (1978).

204 REFERENCES 204 [82] W. Dahl, W. Gruhl, W. G. Burchard, G. Ibe, and C. Dumitrescu, Solidification and Precipitation in Aluminum-Zirconium Alloys: Part 2. Precipitation Processes in Cast Aluminum-Zirconium Alloys, Z Metallkd, 68(3), pp (1977). [83] Z. A. Chaudhury and C. Suryanarayana, A TEM Study of Decomposition Behavior of a Melt- Quenched Al-Zr Alloy, Metallography, 17(3), pp (1984). [84] M. S. Zedalis and M. E. Fine, Precipitation and Ostwald Ripening in Dilute Al Base-Zr-V Alloys, Metall Trans A, 17(12), pp (1986). [85] N. Ryum, Precipitation in an Al-1.78wt.%Hf Alloy after Rapid Solidification, J Mater Sci, 10(12), pp (1975). [86] S. Hori, N. Furushiro, and W. Fujitani, Solidified Structure, Precipitation and Recrystallization Characteristics of Al-Hf Alloys, Aluminium, 57, pp (1981). [87] S. Hori, N. Furushiro, and W. Fujitani, Solidified Structure, Precipitation and Recrystallization Characteristics of Al-Hf Alloys, J Japan Inst Light Metals, 30(11), pp (1980). [88] S. Hori and N. Furushiro, Structure of Rapidly Solidified Al-Hf Alloys and its Thermal Stability, in Proc. 4th Int. Conf on Rapidly Quenched Metals, vol. 2 (edited by T. Masumoto and K. Suzuki), pp , The Japan Institute of Metals, Sendai, Japan (1981). [89] S. Hori, N. Furushiro, and W. Fujitani, Effects of Small Additions of Si on the Precipitation of Al-Hf Alloys, J Japan Inst Light Metals, 31(16), pp (1981). [90] S. Hori, Y. Unigame, N. Furushiro, and H. Tai, Phase Decomposition in Splat Quenched Al-6% Hf Alloy, J Japan Inst Light Metals, 32(8), pp (1982). [91] N. Furushiro and S. Hori, Phase Decomposition Process of a Rapidly Solidified Al-3wt.%Hf- 0.3wt.%Si Alloy During Aging, in Rapidly Quenched Metals (edited by S. Steeb and H. Warlimont), Proceedings of the Fifth International Conference on Rapidly Quenched Metals, pp , Elsevier Science Publishers, Wurzburg (1985). [92] N. Furushiro and S. Hori, A Possible Mechanism of Phase Transformation of Al- 3 Hf from L1 2 to D0 2 2 During Aging in a Rapidly Solidified Al-3Hf-0.3Si Alloy, Acta Metall, 33(5), pp (1985). [93] S. K. Pandey and C. Suryanarayana, Structure and Transformation Behavior of a Rapidly Solidified Al wt.% Hf Alloy, Mater Sci Eng A, 111, pp (1989). [94] S. Srinivasan, P. B. Desch, and R. B. Schwarz, Metastable Phases in the Al 3 X (X = Ti, Zr, and Hf) Intermetallic System, Scripta Metall Mater, 25(11), pp (1991). [95] K. H. J. Buschow and J. H. Vanvucht, Systematic Arrangement of the Binary Rare-Earth Aluminium Systems, Philips Research Reports, 22(3), pp (1967).

205 REFERENCES 205 [96] J. F. Cannon and H. T. Hall, Effect of High Pressure on the Crystal Structures of Lanthanide Trialuminides, J Less-Common Metals, 40(3), pp (1975). [97] M. X. Quan, P. Haldar, J. Werth, and B. C. Giessen, Metastable Extensions of Intermediate Phases in some Aluminum - Rare-Earth Metal Systems, in Rapidly Solidified Alloys and Their Mechanical and Magnetic Properties, vol. 58 (edited by B. C. Giessen, D. E. Polk, and A. I. Taub), pp , MRS, Pittsburgh (1986). [98] S. J. Savage, D. Eliezer, and F. H. Froes, Microstructural Observations and Thermal Stability of a Rapidly Solidified Aluminum-Gadolinium Alloy, Metall Trans A, 18(8), pp (1987). [99] S. J. Savage, F. H. Froes, and D. Eliezer, Microstructural Characterization of As-Cast Rapidly Solidified Al-Sm, Al-Gd, and Al-Er Binary Alloys, in Rapidly Solidified Materials (edited by P. W. Lee and R. S. Carbonara), pp , ASM, Metals Park (1986). [100] B. Dill, Y. Li, M. Al-Khafaji, W. M. Rainforth, and R. A. Buckley, Structure, Properties and Response to Heat Treatment of Melt-Spun Al-Y and Al-La Alloys, J Mater Sci, 29(15), pp (1994). [101] M. Al-Khafaji, Y. Li, W. M. Rainforth, and H. Jones, Phase Constitution in Melt-Spun Al-10 wt.% Y, Phil Mag B, 70(5), pp (1994). [102] A. Meyer, Cubic and Rhombohedral Forms of Al 3 Er, J Less-Common Metals, 20(4), pp (1970). [103] J. L. Murray, The Al-Sc (Aluminum-Scandium) System, J Phase Equil, 19(4), pp (1998). [104] E. A. Brandes and G. B. Brook (eds.), Smithells Metals Reference Book, 7th edn., Butterworth- Heinemann, Oxford (1992). [105] K. S. Vecchio and D. B. Williams, Convergent Beam Electron-Diffraction Study of Al 3 Zr in Al-Zr and Al-Li-Zr Alloys, Acta Metall, 35(12), pp (1987). [106] J. L. Murray, A. J. McAlister, and D. J. Kahan, The Al-Hf (Aluminum-Hafnium) System, J Phase Equil, 19(4), pp (1998). [107] P. Villars, A. Prince, and H. Okamoto, Handbook of Ternary Alloy Phase Diagrams, vol. 1, ASM International, Materials Park (1995). [108] O. I. Zalutskaya, V. R. Ryabov, and I. I. Zalutsky, The X-Ray Analysis of Aluminum-Rich Ternary Alloys of Aluminium and Rare-Earth Metals of the Yttrium Subgroup and Scandium, Dopovidi Akademii Nauk Ukrainskoi RSR Seriya A-Fiziko-Matematichni Ta Technichni Nauki, (3), pp (1969). [109] G. Petzow and G. Effenberg (eds.), Ternary Alloys: A Comprehensive Compendium of Evaluated Constitutional Data and Phase Diagrams, vol. 5, VCH, Weinheim (1992).

206 REFERENCES 206 [110] S. Tsunekawa and M. E. Fine, Lattice Parameters of Al 3 (Zr x Ti 1 x ) vs x in Al-2at.%(Ti+Zr) Alloys, Scripta Metall, 16(4), pp (1982). [111] M. Zedalis and M. E. Fine, Lattice Parameter Variation of Al 3 (Ti,V,Zr,Hf) in Al - 2 at.% (Ti,V,Zr,Hf) Alloys, Scripta Metall, 17(10), pp (1983). [112] R. E. Lewis, D. D. Crooks, Y. C. Chen, M. E. Fine, and J. R. Weertman, High Temperature Al-Zr-V Alloys using Rapid Solidification Processing, in Proc. 3rd Int. Conf on Creep and Fracture of Engineering Materials and Structures (edited by B. Wilshire and R. W. Evans), pp , The Institute of Metals, London (1987). [113] Y. C. Chen, M. E. Fine, J. R. Weertman, and R. E. Lewis, Coarsening Behavior of Ll 2 Structured Al 3 (Zr x V 1 x ) Precipitates in Rapidly Solidified Al-Zr-V Alloy, Scripta Metall, 21(7), pp (1987). [114] Y. C. Chen, M. E. Fine, and J. R. Weertman, Microstuctural Evolution and Mechanical-Properties of Rapidly Solidified Al-Zr-V Alloys at High-Temperatures, Acta Metall Mater, 38(5), pp (1990). [115] C. B. Fuller, D. N. Seidman, and D. C. Dunand, Mechanical Properties of Al(Sc,Zr) Alloys at Ambient and Elevated Temperatures, Acta Mater, 51(16), pp (2003). [116] C. B. Fuller, J. L. Murray, and D. N. Seidman, Temporal Evolution of the Nanostructure of Al(Sc,Zr) Alloys: Part I Chemical Compositions of Al 3 (Sc 1 x Zr x ) Precipitates, Acta Mater, 53(20), pp (2005). [117] C. B. Fuller and D. N. Seidman, Temporal Evolution of the Nanostructure of Al(Sc,Zr) Alloys: Part II Coarsening of Al 3 (Sc 1 x Zr x ) Precipitates, Acta Mater, 53(20), pp (2005). [118] M. E. van Dalen, D. C. Dunand, and D. N. Seidman, Effects of Ti Additions on the Nanostructure and Creep Properties of Precipitation-Strengthened Al-Sc Alloys, Acta Mater, 53(15), pp (2005). [119] E. A. Marquis and D. N. Seidman, Nanoscale Structural Evolution of Al 3 Sc Precipitates in Al(Sc) Alloys, Acta Mater, 49(11), pp (2001). [120] Y. Harada and D. C. Dunand, Microstructure of Al 3 Sc with Ternary Transition-Metal Additions, Mater Sci Eng A, , pp (2002). [121] Y. Harada and D. C. Dunand, Thermal Expansion of Al 3 Sc and Al 3 (Sc 0.75 X 0.25 ), Scripta Mater, 48(3), pp (2003). [122] J. L. Murray, Personal communication regarding the equilibrium Al-Ti phase diagram. (2005).

207 REFERENCES 207 [123] J. L. Murray, A. Peruzzi, and J. P. Abriata, The Al-Zr (Aluminum-Zirconium) System, J Phase Equil, 13(3), pp (1992). [124] T. W. Clyne and M. H. Robert, Stability of Intermetallic Aluminides in Liquid Aluminium and Implications for Grain Refinement, Met Technol, 7(MAY), pp (1980). [125] P. D. Merica, R. G. Waltenberg, and H. Scott, Scientific Papers of the Bureau of Standards, 347(15), pp (1919). [126] N. Ryum, Physical Metallurgy of Heat Treatable Alloys, in Aluminum Alloys: Their Physical and Mechanical Properties, vol. 3 (edited by E. A. Starke and T. H. Sanders), pp , EMAS, Warley (1986). [127] V. L. Kononenko and S. V. Golubev, Phase Diagrams of Binary Systems of Aluminum with La, Ce, Pr, Nd, Sm, Eu, Yb, Sc and Y, Izv Akad Nauk SSSR, Met, (2), pp (1990). [128] V. L. Kononenko and S. V. Golubev, Phase Diagrams of Binary Systems of Aluminum with La, Ce, Pr, Nd, Sm, Eu, Yb, Sc and Y, Russian Metall, (2), pp (1990). [129] G. Rummel, T. Zumkley, M. Eggersmann, K. Freitag, and H. Mehrer, Diffusion of Implanted 3d- Transition Elements in Aluminum. Part I: Temperature Dependence, Z Metallkd, 86(2), pp (1995). [130] G. Rummel, T. Zumkley, M. Eggersmann, K. Freitag, and H. Mehrer, Diffusion of Implanted 3d- Transition Elements in Aluminum. Part II: Pressure Dependence, Z Metallkd, 86(2), pp (1995). [131] S. I. Fujikawa, Solid State Diffusion in Light Metals, J Japan Inst Light Metals, 46(4), pp (1996). [132] Y. Du, Y. A. Chang, B. Huang, W. Gong, Z. Jin, H. Xu, Z. Yuan, Y. Liu, Y. He, and F. Y. Xie, Diffusion Coefficients of Some Solutes in fcc and Liquid Al: Critical Evaluation and Correlation, Mater Sci Eng A, 363(1-2), pp (2003). [133] J. Grammatikakis, K. Eftaxias, and V. Hadjicontis, Interconnection of the Diffusion-Coefficients of Various Elements in Aluminum, J Phys Chem Solids, 49(10), pp (1988). [134] O. Madelung (ed.), Landolt-Bornstein: Numerical Data and Functional Relationships in Science and Technology, vol. 26 of Diffusion in Solid Metals and Alloys, Springer-Verlag, Berlin (1990). [135] S. Dais, R. Messer, and A. Seeger, Nuclear-Magnetic-Resonance Study of Self-Diffusion in Aluminium, Mater Sci Forum, 15-18, pp (1987). [136] S. I. Fujikawa, Impurity Diffusion of Scandium in Aluminum, Def Diff Forum, , pp (1997).

208 REFERENCES 208 [137] D. Bergner and N. van Chi, Untersuchungen zur Diffusion von 3d-Metallen in Al, Wissenschaftliche Zeitschrift der Pädagogischen Hochschule, 15(3) (1977). [138] G. Erdelyi, D. L. Beke, F. J. Kedves, and I. Godeny, Determination of Diffusion Coefficients of Zn, Co and Ni in Aluminum by a Resistometric Method, Phil Mag B, 38(5), pp (1978). [139] S. Fujikawa and K. Hirano, Impurity-Diffusion of Copper in Aluminum, Def Diff Forum, 66-69, pp (1989). [140] N. L. Peterson and S. J. Rothman, Impurity Diffusion in Aluminum, Phys Rev B, 1(8), pp (1970). [141] T. Marumo, S. Fujikawa, and K. Hirano, Diffusion of Zirconium in Aluminum, J Japan Inst Light Metals, 23(1), pp (1973). [142] K. Hirano and S. I. Fujikawa, Impurity Diffusion in Aluminum, J Nuc Mater, 69-70(1-2), pp (1978). [143] N. van Chi and D. Bergner, Diffusion of Mo and W in Al, in DIMETA-82: Diffusion in Metals and Alloys (edited by F. J. Kedves and D. L. Beke), pp , Trans Tech Publications, Switzerland (1983). [144] S. P. Murarka and R. P. Agarwala, Diffusion of Rare Earth Elements in Aluminum, Tech. Rep. 368, Bhabha Atomic Research Center (Indian Atomic Energy Commision) (1968). [145] L. F. Mondolfo, Aluminum Alloys: Structure and Properties, Butterworths, London (1976). [146] D. Lazarus, Effect of Screening on Solute Diffusion in Metals, Phys Rev, 93(5), pp (1954). [147] A. D. LeClaire, On the Theory of Impurity Diffusion in Metals, Phil Mag, 7, pp (1962). [148] P. Shewmon, Diffusion in Solids, 2nd edn., TMS, Warrendale, PA (1989). [149] J. Philibert, Atomic Movements - Diffusion and Mass Transport in Solids, Monographies de physique, Les Editions de Physique, Les Ulis, France (1991). [150] T. Hoshino, R. Zeller, and P. H. Dederichs, Local-Density-Functional Calculations for Defect Interactions in Al, Phys Rev B, 53(14), pp (1996). [151] N. Sandberg and R. Holmestad, First-Principles Calculations of Impurity Diffusion Activation Energies in Al, Phys Rev B, 73(1) (2006). [152] W. B. Alexander and L. M. Slifkin, Diffusion of Solutes in Aluminum and Dilute Aluminum Alloys, Phys Rev B, 1(8), pp (1970).

209 REFERENCES 209 [153] L. M. Angers, D. G. Konitzer, J. L. Murray, and W. G. Truckner, Microstructures in Rapidly Solidified Aluminum Alloys Containing Ti, Zr, Er, in Dispersion Strengthened Aluminum Alloys (edited by Y. W. Kim and W. M. Griffith), pp , TMS, Warrendale (1988). [154] R. C. Weast, M. J. Astle, and W. H. Beyer (eds.), Handbook of Chemistry and Physics, 66th ed., CRC Press, Boca Raton (1985). [155] F. A. Crossley and L. F. Mondolfo, Mechanism of Grain Refinement in Aluminum, Trans AIME, 191, pp (1951). [156] D. G. McCartney, Grain Refining of Aluminum and Its Alloys Using Inoculants, Int Mater Rev, 34(5), pp (1989). [157] B. S. Murty, S. A. Kori, and M. Chakraborty, Grain Refinement of Aluminium and its Alloys by Heterogeneous Nucleation and Alloying, Int Mater Rev, 47(1), pp (2002). [158] J. A. Marcantonio and L. F. Mondolfo, The Nucleation of Aluminium by Several Intermetallic Compounds, J Inst Metals, 98, pp (1970). [159] J. A. Marcantonio and L. F. Mondolfo, Grain Refinement in Aluminum Alloyed with Titanium and Boron, Metall Trans, 2(2), pp (1971). [160] T. Ohashi and R. Ichikawa, Grain Refinement in Aluminum-Zirconium and Aluminum-Titanium Alloys by Metastable Phases, Z Metallkd, 64(7), pp (1973). [161] G. P. Jones and J. Pearson, Factors Affecting Grain Refinement of Aluminum Using Titanium and Boron Additives, Metall Trans B, 7(2), pp (1976). [162] D. H. St. John and L. M. Hogan, Al 3 Ti and Grain Refinement of Aluminum, J Australasian Inst Met, 22(3-4), pp (1977). [163] L. Arnberg, L. Backerud, and H. Klang, Grain-Refinement of Aluminum: 1. Production and Properties of Master Alloys of Al-Ti-B Type and Their Ability to Grain Refine Aluminum, Met Technol, 9(Jan.), pp. 1 6 (1982). [164] L. Arnberg, L. Backerud, and H. Klang, Grain-Refinement of Aluminum: 2. Intermetallic Particles in Al-Ti-B-Type Master Alloys for Grain-Refinement of Aluminum, Met Technol, 9(Jan.), pp (1982). [165] L. Arnberg, L. Backerud, and H. Klang, Grain-Refinement of Aluminum: 3. Evidence of Metastable Phase in Al-Ti-(B) System, Met Technol, 9(Jan.), pp (1982). [166] G. K. Sigworth, The Grain Refining of Aluminum and Phase-Relationships in the Al-Ti-B System, Metall Trans A, 15(2), pp (1984).

210 REFERENCES 210 [167] L. Bäckerud, G. Chai, and J. Tamminen, Solidification Characteristics of Aluminum Alloys. Vol. 1. Wrought Alloys, American Foundrymen s Society (1990). [168] K. T. Kashyap and T. Chandrashekar, Effects and Mechanisms of Grain Refinement in Aluminium Alloys, Bull Mater Sci, 24(4), pp (2001). [169] M. Easton and D. H. St John, Grain Refinement of Aluminum Alloys: Part I. The Nucleant and Solute Paradigms - A Review of the Literature, Metall Mater Trans A, 30(6), pp (1999). [170] H. W. Kerr, J. Cisse, and G. F. Bolling, Equilibrium and Nonequilibrium Peritectic Transformations, Acta Metall, 22(6), pp (1974). [171] H. W. Kerr and W. Kurz, Solidification of Peritectic Alloys, Int Mater Rev, 41(4), pp (1996). [172] D. H. St. John and L. M. Hogan, Thermal Analysis of Peritectic Alloys, J Mater Sci, 17(8), pp (1982). [173] S. P. Midson and H. Jones, On the Absolute Stability Criterion in Relation to Predicting Extension of Solid Solubility, in Proc. 4th Int. Conf on Rapidly Quenched Metals, vol. 2 (edited by T. Masumoto and K. Suzuki), pp , The Japan Institute of Metals, Sendai, Japan (1981). [174] N. J. E. Adkins, N. Saunders, and P. Tsakiropoulos, Rapid Solidification of Peritectic Aluminum Alloys, Mater Sci Eng, 98, pp (1988). [175] B. Chalmers, Principles of Solidification, John Wiley & Sons, New York (1964). [176] M. C. Flemings, Solidification Processing, McGraw-Hill, New York (1974). [177] W. Kurz and D. J. Fisher, Fundamentals of Solidification, 4th edn., Trans Tech Publications, Switzerland (1998). [178] M. Y. Drits, L. B. Ber, Y. G. Bykov, L. S. Toropova, and G. K. Anastasyeva, Aging of Alloy Al-0.3at.%Sc, Fiz metal metalloved, 57(6), pp (1984). [179] C. B. Fuller, D. N. Seidman, and D. C. Dunand, Creep Properties of Coarse-Grained Al(Sc) Alloys at 300 C, Scripta Mater, 40(6), pp (1999). [180] D. N. Seidman, E. A. Marquis, and D. C. Dunand, Precipitation Strengthening at Ambient and Elevated Temperatures of Heat-Treatable Al(Sc) Alloys, Acta Mater, 50(16), pp (2002). [181] E. A. Marquis, D. N. Seidman, and D. C. Dunand, Precipitation Strengthening at Ambient and Elevated Temperatures of Heat-Treatable Al(Sc) Alloys, Acta Mater, 51(1), pp (2003). [182] E. A. Marquis, D. N. Seidman, and D. C. Dunand, Effect of Mg Addition on the Creep and Yield Behavior of an Al-Sc Alloy, Acta Mater, 51(16), pp (2003).

211 REFERENCES 211 [183] E. A. Marquis and D. N. Seidman, Coarsening Kinetics of Nanoscale Al 3 Sc Precipitates in an Al- Mg-Sc Alloy, Acta Mater, 53(15), pp (2005). [184] E. A. Marquis, D. N. Seidman, M. Asta, and C. Woodward, Composition Evolution of Nanoscale Al 3 Sc Precipitates in an Al-Mg-Sc Alloy: Experiments and Computations, Acta Mater, 54(1), pp (2006). [185] S. Hori, S. Saji, and T. Kobayashi, Prolonged Aging of a Subperitectic Al-Zr Alloy, Technology Reports of the Osaka University, 28, pp (1978). [186] S. Hori, S. Saji, and T. Kobayashi, Prolonged Aging of an Al % Zr Alloy, J Japan Inst Metals, 37(10), pp (1973). [187] R. Ichikawa and T. Ohashi, Age Hardening of Chilled Castings of Al 0.2-2wt.%Zr Alloys, J Japan Inst Light Metals, 18(6), pp (1968). [188] N. Ryum, The Effect of Fine Disperse Al 3 Zr on Recrystallization of Aluminium, J Inst Metals, 94(5), pp (1966). [189] M. Sundberg, R. Sundberg, and B. Jacobson, Recrystallization of fcc Metals Containing Precipitates. Influence of Si and Zr on Recrystallization of Al, Jernkont Ann, 155, pp (1971). [190] S. Rystad and N. Ryum, A Metallographic Investigation of the Precipitation and Recrystallization Process in an Al-Zr Alloy, Aluminium, 53(3), pp (1977). [191] H. Westengen, L. Auran, and O. Reiso, Effect of Minor Additions of Transition Elements on the Recrystallization of Some Commercial Aluminium Alloys, Aluminium, 57(12), pp (1981). [192] A. Ruder and D. Eliezer, Microstructure and the Thermal Stability of Rapidly Solidified Aliminum- Rare Earth Alloys, Israel J Tech, 24, pp (1988). [193] A. Ruder and D. Eliezer, Microstructural Transitions in an RS Al-4La Alloy, J Mater Sci, 24(4), pp (1989). [194] A. Ruder and D. Eliezer, A TEM Study of a Rapidly Solidified Al-4La Alloy, J Mater Sci Lett, 8(6), pp (1989). [195] G. Waterloo and H. Jones, Microstructure and Thermal Stability of Melt-Spun Al-Nd and Al-Ce Alloy Ribbons, J Mater Sci, 31(9), pp (1996). [196] Z. G. Zhang, X. F. Bian, and Y. Wang, Microstructural Characterization and Microhardness of Rapidly Solidified Al-Ce Alloys, Z Metallkd, 93(6), pp (2002). [197] S. J. Savage and F. H. Froes, Rapid Solidification of an Luminum-Neodymium Alloy, in Rapidly Solidified Metastable Materials, vol. 28 (edited by B. H. Kear and B. C. Giessen), pp , North- Holland, New York (1984).

212 REFERENCES 212 [198] D. Eliezer, S. J. Savage, Y. R. Mahajan, and F. H. Froes, Microstructures and Thermal Properties of Aluminum-Rare Earth Alloys, in Rapidly Solidified Alloys and Their Mechanical and Magnetic Properties, vol. 58 (edited by B. C. Giessen, D. E. Polk, and A. I. Taub), pp , MRS, Pittsburgh (1986). [199] S. J. Savage, Y. R. Mahajan, A. G. Jackson, and F. H. Froes, Microstructures and Properties of a Rapidly Solidified Aluminum-Gadolinium Alloy, in Rapidly Quenched Metals (edited by S. Steeb and H. Warlimont), Proceedings of the Fifth International Conference on Rapidly Quenched Metals, pp , Elsevier Science Publishers, Wurzburg (1985). [200] M. Fass, D. Itzhak, D. Eliezer, and F. H. Froes, Corrosion Behavior of Rapidly Solidified Al-Er Binary and Ternary Alloys in NaCl Solution at Room Temperature, J Mater Sci Lett, 6(10), pp (1987). [201] M. Fass, D. Itzhak, D. Eliezer, and F. H. Froes, Effects of Heat Treatment on the Corrosion Behaviour of Rapidly Solidified Al-Er Alloys in NaCl Solution, J Mater Sci Lett, 7(1), pp (1988). [202] A. Ruder and D. Eliezer, Microstructure and Thermal Stability of a Rapidly Solidified Al-4Er Alloy, J Mater Sci, 25(8), pp (1990). [203] Z. Nie, T. Jin, J. Fu, G. Xu, J. Yang, J. Zhou, and T. Zuo, Research on Rare Earth in Aluminum, in Aluminum Alloys 2002: Their Physical and Mechanical Properties Pts 1-3, vol of Materials Science Forum, pp (2002). [204] Z. R. Nie, T. N. Jin, J. X. Zou, J. B. Fu, J. J. Yang, and T. Y. Zuo, Development on Research of Advanced Rare-Earth Aluminum Alloy, Trans Nonferrous Met Soc China, 13(3), pp (2003). [205] Z. R. Nie, J. B. Fu, J. X. Zou, T. N. Jun, J. J. Yang, G. F. Xu, H. Q. Ruan, and T. Y. Zuo, Advanced Aluminum Alloys Containing Rare-Earth Erbium, in Aluminium Alloys - Their Physical and Mechanical Properties, vol. 28 (edited by J. F. Nie, A. J. Morton, and B. C. Muddle), pp , Institute of Materials Engineering Australasia (2004). [206] D. H. St. John and L. M. Hogan, The Peritectic Transformation, Acta Metall, 25(1), pp (1977). [207] D. H. St. John and L. M. Hogan, Solute Distribution at the Solid Liquid Interface During Steady- State Peritectic Solidification, J Mater Sci, 19(3), pp (1984). [208] D. H. St. John and L. M. Hogan, A Simple Prediction of the Rate of the Peritectic Transformation, Acta Metall, 35(1), pp (1987). [209] D. H. St. John, The Peritectic Reaction, Acta Metall Mater, 38(4), pp (1990).

213 REFERENCES 213 [210] J. Cisse, H. W. Kerr, and G. F. Bolling, Nucleation and Solidification of Al-Ti Alloys, Metall Trans, 5(3), pp (1974). [211] I. Maxwell and A. Hellawell, Constitution and Solidification of Peritectic Alloys in System Al-Ti, Acta Metall, 23(8), pp (1975). [212] H. A. F. El-Halfawy, E. S. K. Menon, M. Sundararaman, and P. Mukhopadhyay, Al 3 Ti Precipitates in a Chill-Cast Al-14 wt.% Ti Alloy, Metallography, 12(3), pp (1979). [213] H. J. Schurhoff, W. Gruhl, and W. G. Burchard, Contribution to the Nucleation During Solidification of Aluminum-Titanium Alloys, Z Metallkd, 74(3), pp (1983). [214] A. Majumdar and B. C. Muddle, Microstructure in Rapidly Solidified Al-Ti Alloys, Mater Sci Eng A, 169(1-2), pp (1993). [215] M. G. Chu, Microstructure and Solidification Analysis of Melt-Spun Al-Ti and Al-Ti-B Alloys, Mater Sci Eng A, 179, pp (1994). [216] S. Hori, S. Saji, and A. Takehara, Structure of Rapidly Solidified Al-Zr Alloys and its Thermal Stability, in Proc. 4th Int. Conf on Rapidly Quenched Metals, vol. 2 (edited by T. Masumoto and K. Suzuki), pp , The Japan Institute of Metals, Sendai, Japan (1981). [217] T. Ohashi and R. Ichikawa, New Metastable Phase in Rapidly Solidified Al-Zr Alloys, Metall Trans A, 3(8), pp (1972). [218] I. G. Brodova, D. V. Bashlykov, A. B. Manukhin, V. V. Stolyarov, and E. P. Soshnikova, Formation of Nanostructure in Rapidly Solidified Al-Zr Alloy by Severe Plastic Deformation, Scripta Mater, 44(8-9), pp (2001). [219] S. Hori, H. Tai, and Y. Narita, Rapidly Solidified Structure and Grain Refinement of Aluminum Containing Titanium, J Japan Inst Light Metals, 32(11), pp (1982). [220] S. Hori, S. Saji, and A. Takehara, Metastable Phase and Grain Refinement in Rapidly Solidified Al-Zr Alloys, J Japan Inst Light Metals, 31(12), pp (1981). [221] D. Srinivasan and K. Chattopadhyay, Non-equilibrium Transformations Involving L1 2 -Al 3 Zr in Ternary Al-X-Zr Alloys, Metall Mater Trans A, 36A(2), pp (2005). [222] W. Dahl, W. Gruhl, W. G. Burchard, G. Ibe, and C. Dumitrescu, Solidification and Precipitation in Aluminum-Zirconium Alloys: Part 1. Influence of Zirconium on Cast Structure, Z Metallkd, 68(2), pp (1977). [223] A. F. Norman and P. Tsakiropoulos, Rapid Solidification of Al-Hf Alloys - Solidification Microstructures and Decomposition of Solid-Solutions, Int J Rapid Solid, 6(3-4), pp (1991).

214 REFERENCES 214 [224] J. Cisse, G. F. Bolling, and H. W. Kerr, Crystallographic Orientations between Aluminum Grown from the Melt and Various Titanium Compounds, J Crystal Growth, 13/14, pp (1972). [225] I. G. Brodova, I. V. Polents, V. O. Esin, and E. M. Lobov, On the Formation of the Cast Structure of Supercooled Al-Ti Alloys, Phys Met Metall, 73(1), pp (1992). [226] W. T. Kim, B. Cantor, W. D. Griffith, and M. R. Jolly, TEM Characterization of Melt Spun Al-3Ti-1B and Al-5Ti-1B Alloys, Int J Rapid Solid, 7(4), pp (1993). [227] A. F. Norman, P. B. Prangnell, and R. S. MEwen, The Solidification Behaviour of Dilute Aluminium- Scandium Alloys, Acta Mater, 46(16), pp (1998). [228] K. B. Hyde, A. F. Norman, and P. B. Prangnell, The Growth Morphology and Nucleation Mechanism of Primary L1 2 Al 3 Sc Particles in Al-Sc Alloys, in Aluminium Alloys: Their Physical and Mechanical Properties, Pts 1-3, vol of Materials Science Forum, pp (2000). [229] K. B. Hyde, A. F. Norman, and P. B. Prangnell, The Effect of Cooling Rate on the Morphology of Primary Al 3 Sc Intermetallic Particles in Al-Sc Alloys, Acta Mater, 49(8), pp (2001). [230] K. B. Hyde, A. F. Norman, and P. B. Prangnell, The Effect of Ti on Grain Refinement in Al-Sc Alloys, in Aluminum Alloys 2002: Their Physical and Mechanical Properties Pts 1-3, vol of Materials Science Forum, pp (2002). [231] A. F. Norman, K. Hyde, F. Costello, S. Thompson, S. Birley, and P. B. Prangnell, Examination of the Effect of Sc on 2000 and 7000 Series Aluminium Alloy Castings: For Improvements in Fusion Welding, Mater Sci Eng A, 354(1-2), pp (2003). [232] E. Nes and H. Billdal, Nonequilibrium Solidification of Hyperperitectic Al-Zr Alloys, Acta Metall, 25(9), pp (1977). [233] W. A. Tiller, Ch. 9: Solidification, in Physical Metallurgy (edited by R. W. Cahn), 2nd edn., pp , North-Holland, Amsterdam (1970). [234] M. C. Flemings and R. Mehrabian, Ch. 10: Segregation in Castings and Ingots, in Solidification (edited by T. J. Hughel and G. F. Bolling), pp , American Society for Metals, Metals Park, Ohio (1971). [235] L. Bäckerud, G. Chai, and J. Tamminen, Solidification Characteristics of Aluminum Alloys. Vol. 2. Foundry Alloys, American Foundrymen s Society (1990). [236] K. Asboll and N. Ryum, Precipitation in an Al-0.60wt.%Ti Alloy, J Inst Metals, 101(JUL-A), pp (1973). [237] H. Jones and W. M. Rainforth, The Coarsening of Dispersed Al 3 Ti in Aluminum-Based Matrices, Metall Mater Trans A, 34(2), pp (2003).

215 REFERENCES 215 [238] W. E. Frazier, PM Al Alloys: Hot Prospects for Aerospace Applications, Adv Mater Process, 134(5), pp (1988). [239] J. F. Nie, S. Sridhara, and B. C. Muddle, Characterization of Dispersed Intermetallic Phases in Rapidly Quenched Al-Ti-Ce Alloys, Metall Trans A, 23(12), pp (1992). [240] J. F. Nie and B. C. Muddle, Microstructure in Rapidly Solidified Al-Ti-Ni Alloys, Mater Sci Eng A, 215(1-2), pp (1996). [241] W. E. Frazier and M. J. Koczak, Coarsening Behavior of Rapidly Solidified Aluminum-Titanium Alloys, in High Strength Powder Metallurgy Aluminum Alloys II (edited by G. J. Hildeman and M. J. Koczak), pp , TMS-AIME (1986). [242] J. M. Wu, S. L. Zheng, and Z. Z. Li, Thermal Stability and its Effects on the Mechanical Properties of Rapidly Solidified Al-Ti Alloys, Mater Sci Eng A, 289(1-2), pp (2000). [243] Y. Wang, Z. Zhang, W. Wang, and X. Bian, Microstructural Evolution and Microhardness of a Melt- Spun Al-5Ti-1B Alloy During Annealing, Mater Sci Eng A, 366(1), pp (2004). [244] R. D. Schelleng, Mechanical Property Control of Mechanically Alloyed Aluminum, JOM, 41(1), pp (1989). [245] G. X. Liang, Z. C. Li, and E. Wang, Thermal Stability and Mechanical Properties of Mechanically Alloyed Al-10Ti Alloy, J Mater Sci, 31(4), pp (1996). [246] W. E. Frazier and M. J. Koczak, Mechanical and Thermal-Stability of Powder-Metallurgy Aluminum Titanium-Alloys, Scripta Metall, 21(2), pp (1987). [247] G. S. Murty, M. J. Koczak, and W. E. Frazier, High Temperature Deformation of Rapid Solidification Processed / Mechanically Alloyed Al-Ti Alloys, Scripta Metall, 21(2), pp (1987). [248] J. A. Hawk, P. K. Mirchandani, R. C. Benn, and H. G. F. Wilsdorf, The Effect of Ti Addition on the Elevated Temperature Strength and Microstructural Stability of MA Aluminum Alloys, in Dispersion Strengthened Aluminum Alloys (edited by Y. W. Kim and W. M. Griffith), pp , TMS, Warrendale (1988). [249] J. A. Hawk, K. R. Lawless, and H. G. F. Wilsdorf, Tensile Strength of MA Aluminum Alloys with Titanium Additions, Scripta Metall, 23(1), pp (1989). [250] J. A. Hawk, J. K. Briggs, and H. G. F. Wilsdorf, Creep in Mechanically Alloyed Dispersion Strengthened Alloys, in Advances in Powder Metallurgy, vol. 3 (edited by T. G. Gasbarre and W. F. Jandeska), pp , MPIF, Princeton, NJ (1989).

216 REFERENCES 216 [251] P. K. Mirchandani, R. C. Benn, and K. A. Heck, Structure-Property Relationships in Elevated Temperature High Stiffness Mechanically Alloyed Al-Ti Based Alloys, in Lightweight Alloys for Aerospace Applications (edited by E. W. Lee, E. H. Chia, and N. J. Kim), pp , TMS-AIME (1989). [252] P. K. Mirchandani, D. O. Gothard, and A. I. Kemppinen, Elevated Temperature Al-Ti Alloy by Mechanical Alloying, in Advances in Powder Metallurgy, vol. 3 (edited by T. G. Gasbarre and W. F. Jandeska), pp , MPIF, Princeton, NJ (1989). [253] P. W. Sonawane, W. Krishnaswamy, A. Dutta, and R. Sundaresan, Mechanically Alloyed Al-Ti Alloy for Superplastic Forming Characteristics, Mater Sci Forum, 88-90, pp (1992). [254] K. M. Lee and I. H. Moon, High Temperature Deformation Behavior of Mechanically Alloyed Al-Ti Alloys, Scripta Metall Mater, 26(1), pp (1992). [255] T. D. Bayha, R. J. Kilmer, and F. E. Wawner, The Fracture Characteristics of Al-9Ti/SiC p Metal Matrix Composites, Metall Trans A, 23(6), pp (1992). [256] N. Q. Wu, G. X. Wang, Z. Z. Li, and J. M. Wu, Microstructure and Sliding Wear Behavior of PM Alloy Al-10Ti after Thermal Exposure, Wear, 203, pp (1997). [257] S. H. Wang and P. W. Kao, The Strengthening Effect of Al 3 Ti in High Temperature Deformation of Al-Al 3 Ti Composites, Acta Mater, 46(8), pp (1998). [258] I. C. Barlow, H. Jones, and W. M. Rainforth, Evolution of Microstructure and Hardening, and the Role of Al 3 Ti Coarsening, During Extended Thermal Treatment in Mechanically Alloyed Al-Ti-O Based Materials, Acta Mater, 49(7), pp (2001). [259] W. L. Fink, K. R. van Horn, and P. M. Budge, Constitution of High-Purity Aluminum-Titanium Alloys, AIMME Trans, 93, pp (1931). [260] H. A. F. El-Halfawy, Al 3 Ti Precipitates in Aluminum-Titanium Alloys, in Titanium 80 - Science and Technology 4th International Conference on Titanium (edited by H. Kimura and O. Izumi), pp , TMS-AIME, Kyoto, Japan (1980). [261] D. H. St John and L. M. Hogan, Thermal stability in the Al-Al 3 Ti System, J Mater Sci, 15(9), pp (1980). [262] K. Venkateswarlu, S. K. Das, M. Chakraborty, and B. S. Murty, Influence of Thermo-Mechanical Treatment of Al-5Ti Master Alloy on its Grain Refining Performance on Aluminium, Mater Sci Eng A, 351(1-2), pp (2003). [263] N. F. Levoy and J. B. Vander Sande, Phase-Transformations in the Al-Li-Hf and Al-Li-Ti Systems, Metall Trans A, 20(6), pp (1989).

217 REFERENCES 217 [264] P. Malek, M. Janecek, and B. Smola, Structure and Properties of Melt-Spun Al-Zr-Ti Alloys: Part IV. Microstructure and Microhardness Stability at Elevated Temperatures, Kovovè Mater, 38(3), pp (2000). [265] B. S. You and W. W. Park, Age Hardening Phenomena and Microstructure of Rapidly Solidified Al- Ti-Si and Al-Cr-Y Alloys, Scripta Mater, 34(2), pp (1996). [266] Y. R. Mahajan, S. D. Kirchoff, and F. H. Froes, Thermal Stability of Rapidly Solidified Al-Ti-Gd Alloy, Scripta Metall, 20(5), pp (1986). [267] T. Ohashi and R. Ichikawa, Effect of Fe and Si on Age Hardening Properties of Supersaturated Solid Solutions of Al-Zr Alloy by Rapid Solidification, J Japan Inst Metals, 34(6), pp (1970). [268] V. Dobatkin, V. I. Elagin, V. M. Federov, and R. M. Sizova, Decomposition of Saturated Solid Solutions in Granulated Aluminium Alloys, Russian Metall, (2), pp (1970). [269] W. W. Park and T. H. Kim, The Phase Change due to Aging in Rapidly Solidified Al-X Alloys, J Korean Inst Metals, 3(1), pp (1985). [270] W. W. Park and T. H. Kim, Age Hardening Phenomena in Rapidly Solidified Al-Alloys, Scripta Metall, 22(11), pp (1988). [271] T. Sato, A. Kamio, and G. W. Lorimer, Effects of Si and Ti Additions on the Nucleation and Phase Stability of the L1 2 -type Al 3 Zr Phase in Al-Zr Alloys, in Aluminium Alloys: Their Physical and Mechanical Properties, Pts 1-3, vol , pp , Grenoble, France (1996). [272] E. Babić, E. Girt, R. Krsnik, B. Leontic, M. Ocko, Z. Vucic, and I. Zoric, Microhardness Variation in Al-Based 3d Transition Metal Alloys, Physica Status Solidi A, 16(1), pp. K21 K25 (1973). [273] O. Setiukov and I. Fridlyander, Peculiarities of Ti Dendritic Segregation in Aluminum Alloys, in Aluminium Alloys: Their Physical and Mechanical Properties, Pts 1-3, vol of Materials Science Forum, pp , Grenoble, France (1996). [274] V. L. Nordheim, Zur Elektronentheorie der Matalle. II, Ann d Physik, 9(6), pp (1931). [275] J. Royset and N. Ryum, Kinetics and Mechanisms of Precipitation in an Al-0.2 wt.% Sc Alloy, Mater Sci Eng A, 396(1-2), pp (2005). [276] Z. Liu and M. Asta, In preparation (2006). [277] Z. Liu, Thermodynamics of Nano-Scale Precipitate-Strengthened Fe-Cu and Al-Transition-Metal Systems from First-Principles Calculations, Ph.D. thesis, Northwestern University (2006). [278] G. W. Lorimer and R. B. Nicholson, The Nucleation of Precipitates in Aluminium Alloys, in Mechanism of Phase Transformations in Crystalline Solids, Session II, pp , Institute of Metals, London (1969).

218 REFERENCES 218 [279] K. C. Russell, Ch. 6: Nucleation in Solids, in Phase Transformations (edited by H. I. Aaronson), American Society for Metals, Metals Park, Ohio (1970). [280] R. Wagner, R. Kampmann, and P. W. Voorhees, Ch. 5: Homogeneous Second-Phase Precipitation, in Phase Transformations in Materials (edited by G. Kostorz), pp , Wiley-VCH, New York (2001). [281] J. W. Christian, The Theory of Transformations in Metals and Alloys. Part I: Equilibrium and General Kinetic Theory, 2nd edn., Pergamon Press, Oxford (1975). [282] H. I. Aaronson and F. K. LeGoues, An Assessment of Studies on Homogeneous Diffusional Nucleation Kinetics in Binary Metallic Alloys, Metall Trans A, 23(7), pp (1992). [283] R. D. Doherty, Ch. 14: Diffusive Phase Tranformations in the Solid State, in Physical Metallurgy (edited by R. W. Cahn and P. Haasen), 3rd edn., pp , Elsevier, Amsterdam (1983). [284] T. Ohashi, K. Suzuki, and R. Ichikawa, Mechanical Properties of Al-Ti and Al-V Alloys Solidified Rapidly, Bull Nagoya Inst Tech, 23, pp (1971). [285] V. R. Parameswaran, J. R. Weertman, and M. E. Fine, Coarsening Behavior of L1 2 Phase in an Al-Zr- Ti Alloy, Scripta Metall, 23(1), pp (1989). [286] S. Z. Han, S. C. Chung, and H. M. Lee, Alloy Design and Coarsening Phenomenon of L1 2 Precipitates in High-Temperature Al-2at.%(Ti,V,Zr) Systems, Metall Mater Trans A, 20(7), pp (1995). [287] M. S. Chuang and G. C. Tu, The Effect of Ti-Addition on the L1 2 Precipitates of Rapidly-Solidified Al-Cr-Zr Alloy, Scripta Metall Mater, 31(9), pp (1994). [288] P. Malek, M. Janecek, B. Smola, and P. Bartuska, High Temperature Stability of an Al-Zr-Ti Alloy, Key Eng Mat, 97-98, pp (1994). [289] P. Malek, P. Bartuska, and J. Plestil, Structure and Properties of Melt-Spun Al-Zr-Ti Alloys: Part I. Composition of As-Melt-Spun ribbons, Kovovè Mater, 37(6), pp (1999). [290] P. Malek, M. Janecek, B. Smola, and P. Bartuska, Structure and Properties of Melt-Spun Al-Zr-Ti Alloys: Part II. The Solidification Microstructure and Microhardness, Kovovè Mater, 38(1), pp (2000). [291] P. Malek, B. Chalupa, and J. Plestil, Structure and Properties of Melt-Spun Al-Zr-Ti Alloys: Part III. Phase Transformations at Elevated Temperatures, Kovovè Mater, 38(2), pp (2000). [292] P. Malek, M. Janecek, and P. Bartuska, Structure and Properties of a Powder Metallurgy Al-Zr-Ti Alloy, Kovovè Mater, 40(6), pp (2002).

219 REFERENCES 219 [293] C. H. Tsau and Y. C. Chen, The Coarsening of the Precipitates in Melt-Spun Al-Ti-Zr Ribbons, Mater Chem Phys, 73(2-3), pp (2002). [294] H. S. Lee, S. Z. Han, H. M. Lee, and Z. H. Lee, Coarsening Behavior of L1 2 Precipitates in Melt-Spun Al-Ti-V-Zr Alloys, Mater Sci Eng A, 163(1), pp (1993). [295] S. C. Chung, S. Z. Han, H. M. Lee, and H. S. Kwon, Coarsening Phenomenon of L1 2 Precipitates in Rapidly Solidified Al-3at.%(Ti,V,Zr) System, Scripta Metall Mater, 33(5), pp (1995). [296] M. K. Miller, A. Cerezo, M. G. Hetherington, and G. D. W. Smith, Atom Probe Field Ion Microscopy, vol. 52 of Monographs on the Physics and Chemistry of Materials, Oxford University Press (1996). [297] K. Hono, Atom Probe Microanalysis and Nanoscale Microstructures in Metallic Materials, Acta Mater, 47(11), pp (1999). [298] M. K. Miller, Atom Probe Tomography: Analysis at the Atomic Level, Plenum, New York (2000). [299] A. Cerezo, C. R. M. Grovenor, M. G. Hetherington, W. Sha, B. A. Shollock, and G. D. W. Smith, Analysis of Nanometer-Sized Precipitates Using Atom Probe Techniques, Mater Charact, 25(1), pp (1990). [300] S. P. Ringer and K. Hono, Microstructural Evolution and Age Hardening in Aluminium Alloys: Atom Probe Field-Ion Microscopy and Transmission Electron Microscopy Studies, Mater Charact, 44(1-2), pp (2000). [301] B. W. Krakauer, J. G. Hu, S. M. Kuo, R. L. Mallick, A. Seki, D. N. Seidman, J. P. Baker, and R. J. Loyd, A System for Systematically Preparing Atom-Probe Field-Ion-Microscope Specimens for the Study of Internal Interfaces, Rev Sci Instrum, 61(11), pp (1990). [302] B. W. Krakauer and D. N. Seidman, Systematic Procedures for Atom-Probe Field-Ion Microscopy Studies of Grain-Boundary Segregation, Rev Sci Instrum, 63(9), pp (1992). [303] T. F. Kelly, P. P. Camus, D. J. Larson, L. M. Holzman, and S. S. Bajikar, On the Many Advantages of Local-Electrode Atom Probes, Ultramicroscopy, 62(1-2), pp (1996). [304] T. F. Kelly and D. J. Larson, Local Electrode Atom Probes, Mater Charact, 44(1-2), pp (2000). [305] T. E. Kelly, T. T. Gribb, J. D. Olson, R. L. Martens, J. D. Shepard, S. A. Wiener, T. C. Kunicki, R. M. Ulfig, D. R. Lenz, E. M. Strennen, E. Oltman, J. H. Bunton, and D. R. Strait, First Data from a Commercial Local Electrode Atom Probe (LEAP), Microsc Microanal, 10(3), pp (2004). [306] A. Cerezo, T. J. Godfrey, and G. D. W. Smith, Application of a Position-Sensitive Detector to Atom Probe Microanalysis, Rev Sci Instrum, 59(6), pp (1988).

220 REFERENCES 220 [307] A. Cerezo, T. J. Godfrey, S. J. Sijbrandij, G. D. W. Smith, and P. J. Warren, Performance of an Energy- Compensated Three-Dimensional Atom Probe, Rev Sci Instrum, 69(1), pp (1998). [308] M. K. Miller, The Development of Atom Probe Field-Ion Microscopy, Mater Charact, 44(1-2), pp (2000). [309] O. C. Hellman, J. Vandenbroucke, J. B. du Rivage, and D. N. Seidman, Application Software for Data Analysis for Three-Dimensional Atom Probe Microscopy, Mater Sci Eng A, 327(1), pp (2002). [310] O. C. Hellman, J. A. Vandenbroucke, J. Rusing, D. Isheim, and D. N. Seidman, Analysis of Three- Dimensional Atom-Probe Data by the Proximity Histogram, Microsc Microanal, 6(5), pp (2000). [311] O. C. Hellman and D. N. Seidman, Measurement of the Gibbsian Interfacial Excess of Solute at an Interface of Arbitrary Geometry using Three-Dimensional Atom Probe Microscopy, Mater Sci Eng A, 327(1), pp (2002). [312] G. Sha and A. Cerezo, Field Ion Microscopy and 3-D Atom Probe Analysis of Al 3 Zr Particles in 7050 Al Alloy, Ultramicroscopy, 102(2), pp (2005). [313] J. L. Murray, Personal communication regarding the Al-Zr-Ti phase diagram. (2006). [314] F. Vurpillot, A. Bostel, and D. Blavette, Trajectory Overlaps and Local Magnification in Three- Dimensional Atom Probe, Appl Phys Lett, 76(21), pp (2000). [315] F. Vurpillot, A. Cerezo, D. Blavette, and D. J. Larson, Modeling Image Distortions in 3DAP, Microsc Microanal, 10(3), pp (2004). [316] D. Blavette, F. Vurpillot, P. Pareige, and A. Menand, A Model Accounting for Spatial Overlaps in 3D Atom-Probe Microscopy, Ultramicroscopy, 89(1-3), pp (2001). [317] T. T. Tsong, Field Ion Image Formation, Surface Science, 70(1), pp (1978). [318] M. K. Miller and M. G. Hetherington, Local Magnification Effects in the Atom Probe, Surface Science, 246(1-3), pp (1991). [319] C. B. Fuller, Temporal Evolution of the Microstructures of Al(Sc,Zr) Alloys and Their Influences on Mechanical Properties, Ph.D. thesis, Northwestern University (2003). [320] B. Forbord, W. Lefebvre, F. Danoix, H. Hallem, and K. Marthinsen, Three Dimensional Atom Probe Investigation on the Formation of Al 3 (Sc,Zr)-Dispersoids in Aluminium Alloys, Scripta Mater, 51(4), pp (2004). [321] A. Tolley, V. Radmilovic, and U. Dahmen, Segregation in Al 3 (Sc,Zr) Precipitates in Al-Sc-Zr Alloys, Scripta Mater, 52(7), pp (2005).

221 REFERENCES 221 [322] E. Clouet, M. Nastar, A. Barbu, C. Sigli, and G. Martin, Precipitation in Al-Zr-Sc Alloys: A Comparison between Kinetic Monte Carlo, Cluster Dynamics and Classical Nucleation Theory, in Solidto-Solid Phase Transformations in Inorganic Materials, vol. 2 (edited by J. M. Howe, D. E. Laughlin, J. K. Lee, U. Dahmen, and W. A. Soffa), pp , TMS (2005). [323] J. D. Eshelby, The Determination of the Elastic Field of an Ellipsoidal Inclusion, and Related Problems, Proc Roy Soc (London), A241(1226), pp (1957). [324] J. D. Eshelby, Elastic Inclusions and Inhomogeneities, in Progress in Solid Mechanics, vol. 2 (edited by I. N. Sneddon and R. Hill), pp , North-Holland Pub. Co. ; Interscience Publishers, Amsterdam [Netherlands] New York (1961). [325] K. C. Russell, Linked Flux Analysis of Nucleation in Condensed Phases, Acta Metall, 16(5), pp (1968). [326] J. W. Cahn, Nucleation on Dislocations, Acta Metall, 5(3), pp (1957). [327] R. B. Nicholson, Ch. 7: Nucleation at Imperfections, in Phase Transformations (edited by H. I. Aaronson), pp , American Society for Metals, Metals Park, Ohio (1970). [328] G. M. Novotny and A. J. Ardell, Precipitation of Al 3 Sc in Binary Al-Sc Alloys, Mater Sci Eng A, 318(1-2), pp (2001). [329] S. F. Baumann and D. B. Williams, Experimental Observations on the Nucleation and Growth of δ (Al 3 Li) in Dilute Al-Li Alloys, Metall Trans A, 16(7), pp (1985). [330] M. F. Ashby and L. M. Brown, Diffraction Contrast from Spherically Symmetrical Coherency Strains, Phil Mag, 8(91), pp (1963). [331] S. Iwamura and Y. Miura, Loss in Coherency and Coarsening Behavior of Al 3 Sc Precipitates, Acta Mater, 52(3), pp (2004). [332] J. Royset and N. Ryum, Some Comments on the Misfit and Coherency Loss of Al 3 Sc Particles in Al-Sc Alloys, Scripta Mater, 52(12), pp (2005). [333] M. S. Zedalis, Development of an Elevated Temperature Aluminum Alloy Containing Al 3 X-Type Dispersed Phases, Ph.D. thesis, PhD Dissertation, Northwestern University (1985). [334] Y.-C. Chen, Thermal Stability of L1 2 Structured Al 3 (Zr 1 x V x ) in Rapidly-Solidified Al-Zr-V Alloys, Ph.D. thesis, PhD Dissertation, Northwestern University (1988). [335] A. J. Ardell, The Effect of Volume Fraction on Particle Coarsening: Theoretical Considerations, Acta Metall, 20(1), pp (1972).

222 REFERENCES 222 [336] A. D. Brailsford and P. Wynblatt, The Dependence of Ostwald Ripening Kinetics on Particle Volume Fraction, Acta Metall, 27(3), pp (1979). [337] J. D. Robson, A New Model for Prediction of Dispersoid Precipitation in Aluminium Alloys Containing Zirconium and Scandium, Acta Mater, 52(6), pp (2004). [338] R. Mehrabian, Rapid solidification, Int Met Rev, 27(4), pp (1982). [339] E. J. Lavernia, J. D. Ayers, and T. S. Srivatsan, Rapid Solidification Processing with Specific Application to Aluminum-Alloys, Int Mater Rev, 37(1), pp (1992). [340] H. M. Lee and H. Lee, Site Fraction Behavior of Transition-Elements in Al 3 (Ti,V,Zr) Phase, CAL- PHAD, 16(1), pp (1992). [341] R. W. Baluffi, S. M. Allen, and W. C. Carter, Kinetics of Materials, John Wiley & Sons, Hoboken (2005). [342] J. Y. Barghout, G. W. Lorimer, R. Pilkington, and P. B. Prangnell, The Effects of Second Phase Particles, Dislocation Density and Grain Boundaries on the Electrical Conductivity of Aluminium Alloys, in Aluminium Alloys: Their Physical and Mechanical Properties, Pts 1-3, vol. 217 of Materials Science Forum, pp (1996). [343] L. M. Brown and R. K. Ham, Strengthening Methods in Crystals, Elsevier Publishing Company, London (1971). [344] S. M. Jeon and J. K. Park, Transition Behavior of Deformation Mode from Shearing to Looping in Al-Li Single Crystals, Phil Mag A, 70(3), pp (1994). [345] S. M. Jeon and J. K. Park, Precipitation Strengthening Behavior of Al-Li Single Crystals, Acta Mater, 44(4), pp (1996). [346] M. Nakamura and K. Kimura, Elastic Constants of TiAl 3 and Zr Al 3 Single Crystals, J Mater Sci, 26(8), pp (1991). [347] M. S. Zedalis, M. V. Ghate, and M. E. Fine, Elastic Moduli of Al 3 Zr, Scripta Metall, 19(5), pp (1985). [348] E. P. George, J. A. Horton, W. D. Porter, and J. H. Schneibel, Brittle Cleavage of L1 2 Trialuminides, J Mater Res, 5(8), pp (1990). [349] C. L. Fu, Electronic, Elastic, and Fracture Properties of Trialuminide Alloys: Al 3 Sc and Al 3 Ti, J Mater Res, 5(5), pp (1990). [350] R. W. Hyland, Jr. and R. C. Stiffler, Determination of the Elastic Constants of Polycrystalline Al 3 Sc, Scripta Metall Mater, 25(2), pp (1991).

223 REFERENCES 223 [351] C. L. Fu and M. H. Yoo, Electronic Structure and Mechanical Behavior of Transition-Metal Aluminides: A First-Principles Total-Energy Investigation, Mater Chem Phys, 32(1), pp (1992). [352] E. A. Marquis, Microstructural Evolution and Strengthening Mechanisms in Al-Sc and Al-Mg-Sc Alloys, Ph.D. thesis, Northwestern University (2002). [353] K. Fukunaga, T. Shouji, and Y. Miura, Temperature Dependence of Dislocation Structure of L1 2 Al 3 Sc, Mater Sci Eng A, , pp (1997). [354] D. Tabor, The Physical Meaning of Indentation and Scratch Hardness, Br. J. Appl. Phys., 7(5), pp (1956). [355] A. K. Mukherjee, J. E. Bird, and J. E. Dorn, Experimental Correlations for High-Temperature Creep, Trans ASM, 62, pp (1969). [356] J. H. Gittus, Theoretical Equation for Steady-State Dislocation Creep in a Material Having a Threshold Stress, Proc Roy Soc (London), 342(1629), pp (1975). [357] J. C. Gibeling and W. D. Nix, The Description of Elevated Temperature Deformation in Terms of Threshold Stresses and Back Stresses: A Review, Mater Sci Eng, 45(2), pp (1980). [358] R. Lagneborg and B. Bergman, The Stress/Creep Rate Behavior of Precipitation-Hardened Alloys, Metal Science, 10, pp (1976). [359] R. A. Karnesky, D. N. Seidman, and D. C. Dunand, Creep of Al-Sc Microalloys with Rare-Earth Element Additions, Mater Sci Forum, Accepted (2006). [360] M. van Dalen, D. C. Dunand, and D. N. Seidman, Creep Data for Al-Sc-Ti Alloys, Personal Communication, (2006). [361] J. D. Robson and P. B. Prangnell, Dispersoid Precipitation and Process Modelling in Zirconium Containing Commercial Aluminium Alloys, Acta Mater, 49(4), pp (2001). [362] C. Sigli, Zirconium Solubility in Aluminum Alloys, in International Conference on Aluminum Alloys (ICAA 9), pp , Brisbane, Australia (2004). [363] J. D. Robson and P. B. Prangnell, Modelling Al 3 Zr Dispersoid Precipitation in Multicomponent Aluminium Alloys, Mater Sci Eng A, 352(1-2), pp (2003). [364] Y. W. Riddle and T. H. Sanders, A Study of Coarsening, Recrystallization, and Morphology of Microstructure in Al-Sc-(Zr)-(Mg) Alloys, Metall Mater Trans A, 35, pp (2004). [365] B. Forbord, H. Hallem, N. Ryum, and K. Marthinsen, Precipitation and Recrystallisation in Al-Mn- Zr with and without Sc, Mater Sci Eng A, , pp (2004).

224 REFERENCES 224 [366] M. Lieblich and M. Torralba, Cellular Microstructure and Heterogeneous Coarsening of δ in Rapidly Solidified Al-Li-Ti Alloys, J Mater Sci, 26(16), pp (1991). [367] M. Lieblich and M. Torralba, Aging Characteristics of Rapidly Solidified Al-Li-Ti Alloys at 473 K, J Mater Sci, 27(13), pp (1992). [368] J. H. Xu and A. J. Freeman, Phase Stability and Electronic Structure of ScAl 3 and ZrAl 3 and of Sc- Stabilized Cubic ZrAl 3 Precipitates, Phys Rev B, 41(18), pp. 12,553 12,561 (1990). [369] C. P. Lee, K. E. Knipling, D. C. Dunand, and D. N. Seidman, In preparation (2006). [370] I. G. Brodova, I. V. Polents, D. V. Bashlikov, P. S. Popel, and O. A. Chikova, The Forming Mechanism of Ultradispersed Phases in Rapidly Solidified Aluminium Alloys, Nanostructured Materials, 6(1-4), pp (1995). [371] I. Brodova, V. M. Zamyatin, P. Popel, V. O. Esin, B. A. Baum, A. I. Moiseyev, I. P. Korshunov, A. L. Topchii, Y. G. Tikhomirov, and I. V. Polents, Conditions of Formation of Metastable Phases During Crystallization of Al-Zr Alloys, Rasplavy, 2(6), pp (1988). [372] I. G. Brodova, D. V. Bashlykov, A. B. Manukhin, T. I. Yablonskikh, and N. A. Zolotova, Effect of the Time-Temperature Melt Treatment on the Structure and Phase Composition of a Rapidly Quenched Al-1.4% Hf Alloy, Phys Met Metall, 89(3), pp (2000). [373] I. Brodova, D. Bashlykov, A. Manukhin, E. Rozhicyna, P. Popel, and V. Manov, Disperse Structure Forming in Rapidly Quenched Al-Hf Alloy, Mater Sci Eng A, 304, pp (2001). [374] E. Nes and N. Ryum, On the Formation of Fan-Shaped Precipitates During Decomposition of a Highly Supersaturated Al-Zr Solid Solution, Scripta Metall, 5(11), pp (1971). [375] E. Nes and H. Billdal, The Mechanism of Discontinuous Precipitation of Metastable Al 3 Zr Phase from an Al-Zr Solid-Solution, Acta Metall, 25(9), pp (1977). [376] O. Beeri, D. C. Dunand, and D. N. Seidman, Effect of Fe and Si Contamination on Preciptiation in Dilute Al-Sc Alloys, Personal Communication, (2006).

225 225 APPENDIX A Alloy Preparation and Castability Peritectic alloys are not amenable to conventional casting because of the necessary liquidus elevation and the potential for properitectic precipitation in the melt, as discussed in Chapter 2. The alloys must be superheated well beyond the melting point of Al, and the cooling rate and/or solute concentration must be controlled so as to avoid primary Al 3 M. Initially, however, these difficulties were not appreciated and efforts to conventionally cast 1 Al-Ti and Al-Zr based alloys were invariably met with failure. Large primary Al 3 Ti or Al 3 Zr precipitates formed, Figure A.1, and the alloys exhibited no appreciable precipitation hardening during postsolidification aging. The presence of primary precipitates, like those in Figure A.1, was initially (and erroneously) believed to indicate that the α-al solid-solution surrounding the needle-like primary precipitates is supersaturated in solute. In fact, the exact opposite is true the presence of primary precipitates ensures that the α-al solid-solution has a composition following the Al 3 M liquidus, meaning the last solid to form has a composition C Lp, Figure 2.1, that is necessarily less than maximum solid-solubility. A.1 Advantages of Arc-Melting It was revealed, largely through trial and error, that arc-melted alloys consistently exhibit stronger precipitation hardening than those of similar composition that were conventionally-cast. 1 The reasons for this are manifold (in order of decreasing significance): 1. High melt temperatures The temperature of the arc is much greater than that obtained in a resistively-heated furnace. High temperatures are required to melt into the single-phase melt since the liquidus of a peritectic alloy increases rapidly with solute content. 1 By melting in a resistively-heated furnace (in air), and casting into graphite molds.

226 APPENDIX A ALLOY PREPARATION AND CASTABILITY 226 Fig. A.1: Optical micrograph of a conventionally-solidified hyperperitectic Al-Ti alloy. Numerous plate-like Al 3 Ti primary precipitates are observed. 2. Inert atmosphere The arc-melter chamber has a gettered argon atmosphere, which minimizes oxidation of the solutes and maintains accurate alloy compositions. 3. Accelerated solidification rates Solidification in the arc-melter is effected by an underlying continuously water-cooled copper hearth, which might accelerate the solidification of the alloys. Accelerated solidification cooling rates are required to suppress properitectic precipitation, as discussed in Chapter 2. A.1.1 High melt temperatures The interrelation between solidification rate, solute concentration, and solidified microstructure in peritectic alloys was discussed in Chapter 2. In addition to these factors, Brodova et al. investigated the effect of melt superheat (above the liquidus) on the microstructures of dilute Al-Ti [225, 370], Al-Zr [218, 370, 371], Al-Hf [372, 373] alloys. It was shown that increasing the melt temperature and/or hold time above the liquidus has the same effect as increasing the cooling rate: i.e., primary Al 3 M precipitates become smaller and more equiaxed; the likelihood for primary precipitation of metastable phases is increased; and, for sufficiently large superheating, primary precipitation of Al 3 M may be suppressed completely. Brodova et al. argued that

227 APPENDIX A ALLOY PREPARATION AND CASTABILITY 227 significant superheating is necessary to transform the liquid from a colloidal state (consisting fine Al 3 M primary precipitates dispersed in the melt) to a homogenous single phase. This same phenomenon is responsible for so-called fade in commercial Al casting alloys, where the grain refining efficacy of Ti additions falls off with holding time of the melt [156, 157, 161, 163, 168]. A.1.2 Inert atmosphere Alloys produced by conventional means, melting in air, were generally found to be significantly depleted in solute compared to their nominal initial compositions. This was attributed to oxidation of Zr and/or Ti during melting. Ingot melting in the arc-melter occurs in an inert argon atmosphere, mitigating solute loss from oxidation. A.1.3 Accelerated solidification rates Solidification in the arc-melter is effected by the water-chilled copper hearth, which might result in an enhanced solidification rate since the mold is continuously cooled after melting. In actuality, the effect is probably not significant since, unlike in a conventional mold where heat is removed through multiple surfaces, heat is extracted only through the bottom of the ingot in the arc-melter. Indeed, solidification of the button ingot can be observed by eye to occur over the course of several seconds, indicating that the solidification rate is at most C s -1.

228 228 APPENDIX B Homogenization and Other Prior Annealing Treatments The alloys in this thesis were not homogenized because, empirically, homogenization was found to be deleterious to the attainable peak strength achieved during the ensuing isothermal aging treatment of Al-Zr based alloys. Other annealing treatment (e.g., pre-aging at 500 C) prior to the precipitation aging heat treatment had a similar effect. Figure B.1 shows the observed hardness of an Al 0.2 at.% Zr subjected to various initial annealing treatments, prior to isothermal aging at 375 C for 100 h. There is a systematic decline in peak hardness attained by aging at 375 C with increasing thermal exposure prior to aging. B.1 Homogenization (or Solution) Anneals A homogenization, or solution, heat treatment involves heating the alloy in the single phase α-al solid-solution and allowing diffusion to eliminate the solute microsegregation incurred during solidification, thereby homogenizing the solute distribution in the alloy. For peritectic alloys like Al-Zr with k 0 > 1, however, the dendrites can be supersaturated well beyond the maximum solid-solubility of Zr in α-al. During homogenization, there will be a tendency for precipitation to occur since, locally, the alloy composition does not lie in the single phase α-al solidsolution. Because most of Zr atoms are precipitated into coarse (likely equilibrium D0 23 ) Al 3 Zr precipitates during homogenization, the achievable hardness during the ensuing isothermal aging treatment is reduced dramatically as demonstrated in Figure B.1. Figure B.2 supports this hypothesis of precipitation of Al 3 Zr during homogenization. Displayed are two scanning electron microscope (SEM) micrographs of a conventionally-solidified Al-2.36Zn-0.26Zr (at.%) alloy after homogenization at 640 C for 650 h. As discussed previously, conventional solidification often results in large primary Al 3 Zr or Al 3 Ti precipitates; an example of a primary Al 3 Zr precipitate is visible in Figure B.2(a). Smaller plate-like precipitates, ca. 2

229 APPENDIX B HOMOGENIZATION AND OTHER PRIOR ANNEALING TREATMENTS Vickers microhardness (MPa) Hardness of pure Al: ca. 220 MPa Vickers microhardness (HV 200 ) 100 Fig. B.1: Observed Vickers microhardness of Al-0.2 at.% Zr subjected to various annealing treatments prior to isothermal aging at 375 C for 100 h. Pre-aging was performed at 500 C for 1 h. Homogenization was carried out at 640 C for 240 h. Fig. B.2: SEM micrographs of a conventionally-solidified Al-2.36Zn-0.26Zr (at.%) alloy following homogenization at 640 C for 650 h.

230 APPENDIX B HOMOGENIZATION AND OTHER PRIOR ANNEALING TREATMENTS nm in length, are visible in Figure B.2(b). Unlike the large primary phase, these smaller precipitates exhibit one of three unique orientations in Figure B.2(b), suggesting that they formed in the solid-state and have a cube-on-cube orientation relationship with α-al. These precipitates are probably disc-like equilibrium D0 23 -structured Al 3 Zr precipitates, like those described in Chapter 6, formed in locally supersaturated regions of the alloy during homogenization at 640 C. This conclusion concurs also with that of Dahl et al. [82], who observed orthogonal D structured precipitates after homogenizing alloys containing at.% Zr for 25 h at 640 C. Dahl et al. also concluded that Al-Zr alloys are incapable of precipitation hardening following homogenization treatments, consistent with the experimental evidence in Figure B.1. B.2 Pre-Aging at 500 C Pre-aging at 500 C similarly results in a significant decrease in hardness attained during subsequent aging at 375 C (Figure B.1). The purpose of the 500 C pre-age is to suppress discontinuous precipitation, which can occur in Al-Zr alloys [374, 375]. Pre-aging has been employed in several studies on Al-Zr, Al-Zr-V, Al-Zr-Ti, and Al-Zr-Ti-V alloys as discussed in Table 5.4. It is known that Al 3 Zr (L1 2 ) precipitates are nucleated during pre-aging at 500 C, as shown by Tsau and Chen [293] using transmission electron microscopy (TEM). As discussed in Figure 6.21, higher isothermal aging temperatures yield a greater volume fraction of precipitate-free interdendritic regions because nucleation is limited to the inner-most regions of the solute-rich dendrites. Since precipitates are nucleated at 500 C, the volume fraction of precipitate-free interdendritic regions is greater than if aging had been carried out isothermally at 375 C, thus explaining the reduced peak hardness for the pre-aged specimen. Zedalis and Fine [84] claimed that the necessity of pre-aging is dependent on the temperature of the ensuing aging treatment. For their coarsening studies on Al-Zr and Al-Zr-V alloys at 375, 400, and 425 C, pre-aging was necessary only at lower aging temperatures; the discontinuous precipitation mechanism was not observed at 425 C. Discontinuous precipitation, heterogenous nucleation on a moving grain boundary, is prevalent at lower aging temperatures where sluggish diffusion kinetics may preclude homogenous nucleation. Since their arc-melted alloys were not pre-aged prior to aging at 425 C, the coarsening study of Al 3 Zr (L1 2 ) precipitates by Zedalis and Fine [84] is directly comparable to the data for Al 3 Zr (L1 2 ) and Al 3 (Zr 1 x Ti x ) (L1 2 )

231 APPENDIX B HOMOGENIZATION AND OTHER PRIOR ANNEALING TREATMENTS 231 precipitates presented in Chapter 5.

232 232 APPENDIX C Length Scales of Analytical Techniques and Potential Biases The alloys studied in this thesis are highly segregated, as discussed an Chapter 5 and shown in Figure 5.6. During aging, the initially solute-rich dendrites produce homogeneously-distributed nanometer-scale Al 3 Zr (L1 2 ) or Al 3 (Zr 1 x Ti x ) (L1 2 ) precipitates which demarcate the dendritic and interdendritic regions of the alloy. The precipitate-rich dendritic regions, when observed in 2-D projection by transmission electron microscopy (TEM) or scanning electron microscopy (SEM), are of the order of 1 10 µm across and vary considerably in size and morphology across length scales of the order of 100 µm. The width of the interdendritic precipitate-free channels similarly varies. C.1 Length Scales of Analytical Techniques Figure C.1 is a visual depiction of the primary analytical techniques used in this study. These are (in increasing length scale): 3-D atom-probe tomography (3-D APT), transmission electron microscopy (TEM), scanning electron microscopy (SEM), Vickers microhardness, creep, and electrical conductivity. Because of microsegregation of solutes occurring on the µm scale, what is observed on the sub-micrometer-scale cannot be straightforwardly correlated to what is measured on the macroscopic scale. As discussed in Chapter 4, the measured solute concentrations within a 3-D APT analysis volume varies considerable from analysis to analysis, depending on the proximity to the solute-rich dendrites. The observed precipitate size and morphology by TEM, discussed in Chapter 5, is similarly strongly dependent on where one looks. Figure C.2 is an SEM micrograph displaying several precipitate-rich dendrites in relation to the hole in a TEM foil. The microstructure observable by TEM, which is limited to electron-transparent areas, is a small sample of the actual microstructure varying on the micrometer-scale. Indeed, montages of several micrographs, e.g. Figures 5.21, 6.2, and 6.3, are usually required to study the

233 APPENDIX C LENGTH SCALES OF ANALYTICAL TECHNIQUES AND POTENTIAL BIASES 233 Transmission electron microscopy Dendrites Atom-probe tomography Scanning electron microscopy Vickers microhardness Electrical conductivity Creep Length scale (m) Fig. C.1: Length scales of the analytical techniques used. Those techniques sampling regions on a scale larger than the dendrites (> 10 4 m) are unbiased by the dendritic distribution of precipitates, whereas those techniques limited to the submicron scale (< 10 6 m) are highly biased by the inhomogeneous distribution of solutes and precipitates. micrometer-scale of the dendrites by TEM. Fortunately, the larger-scale techniques in Figure C.1 are largely unbiased by the dendritic distribution of solutes and precipitates. Figure C.3 indicates that the tip-to-tip length of a Vickers microhardness pyramidal indent is of the order of 100 µm for the alloys studied. Considering the scale of the microsegregation, Figure 5.6, a single hardness measurement therefore samples several precipitate-rich dendritic regions and precipitate-free interdendritic channels, meaning that the measured hardness value is an average of the strong and weak regions of the alloy and is generally not subject to sampling biases. This applies also to creep and electrical conductivity. C.2 Biases Associated with Sample Preparation The dendritic distribution of precipitates was unrecognized for some time, primarily due to preferential chemical etching during electropolishing of TEM foils. Figure C.4 displays low magnification TEM micrographs illustrating preferential electropolishing around the precipitaterich dendrites. The small dendritic precipitates are consequently hard to image, and are easily missed, because of absorption from incoherent scattering occurring in the thicker regions of the foil. The most apparent population of precipitates observed by TEM is the larger ( R ca. 20 nm) spheroidal interdendritic precipitates, whereas the most numerous population is the smaller ( R ca nm) dendritic precipitates. The increased foil thickness can also accentuate the apparent volume fraction of precipitates in the dendrites.

234 APPENDIX C LENGTH SCALES OF ANALYTICAL TECHNIQUES AND POTENTIAL BIASES 234 Fig. C.2: SEM micrograph displaying precipitate-rich dendrites around the specimen edge of an electropolished TEM foil V ic k e r s m ic r o h a r d n e s s (M P a ) R a n g e o f ty p ic a l h a rd n e s s v a lu e s d V ic k e r s m ic r o h a r d n e s s (H V ) C o rre s p o n d in g in d e n t s iz e In d e n t d ia g o n a l le n g th (µm ) Fig. C.3: Vickers microhardness value as a function of indent diagonal length for a 200 g load (used on all hardness measurements).

235 APPENDIX C LENGTH SCALES OF ANALYTICAL TECHNIQUES AND POTENTIAL BIASES 235 Fig. C.4: Bright-field and centered superlattice dark-field TEM micrographs of Al-0.1Zr and Al-0.1Zr-0.1Ti (at.%) alloys, respectively, after aging at 375 C. The foils preferentially electropolish around the precipitate-rich dendrites. As in TEM foil preparation, preferential electropolishing in the precipitate-free interdendritic regions probably also occurs during preparation of 3-D APT specimens since the electropolishing procedures are similar. There is therefore a tendency for necking in the interdendritic regions, resulting in 3-D APT specimens whose apexes (the region analyzed) are largely devoid of precipitates. This theory explains why even very large 3-D APT analysis volumes are often precipitate-free, as discussed in Chapter 4.

236 236 APPENDIX D Study on an Al-Zr-Ti Alloy Prepared by Parameswaran et al. A brief study, summarized here, was carried out on an arc-melted Al-Zr-Ti alloy investigated previously by Parameswaran et al. [285]. Prof. M. E. Fine kindly provided as-cast arc-melted buttons from that study, which demonstrated a dramatic improvement in coarsening resistance of Al-Zr-Ti alloys compared to Al-Zr or Al-Zr-V alloys [84] prepared by similar arc-melting procedures. The reported improved coarsening resistance is displayed in Figure D.1, reproduced from reference [285]. The as-cast button provided by Prof. Fine, labeled AZT 2, had a composition of Al-0.19Zr-0.18Ti (at.%) as confirmed presently by Bodycote Materials Testing (Skokie, IL). This is comparable to the alloy composition reported in [285], Al-0.20Zr-0.17Ti (at.%), indicating that the alloy described presently is comparable (if not identical) to that studied in [285]. The present author repeated the aging procedures in [285], which involved pre-aging the alloys at 500 C for 1 h prior to isothermal aging at 425 C. Pre-aging had been reported to suppress discontinuous precipitation, as discussed in Appendix B. Precipitation hardening was monitored by Vickers microhardness, Figure D.2, which was not performed by Parameswaran et al. [285]. During the 500 C pre-age, Al 3 (Zr 1 x Ti x ) (L1 2 ) precipitates are nucleated, as indicated by the increase in hardness and shown elsewhere by Tsau and Chen using transmission electron microscopy (TEM) [293]. During the subsequent aging treatment at 425 C, there is negligible additional strengthening 1 and the L1 2 precipitates overage sluggishly. The microstructure from specimen AZT 2 aged at 425 C for 400 h was investigated using TEM, offering a direct comparison with data in Figure D.1. The dark-field TEM micrograph of the Al 3 (Zr 1 x Ti x ) precipitates observed are displayed in Figure D.3. The mean radius, R, of the 104 precipitates in Figure D.3 is 19.2±3.6 nm, which is inconsistent with the measured R = 6.9 nm 1 After pre-aging, the alloy reaches a peak strength of only ca. 420 MPa during the ensuing isothermal aging treatment, consistent with the results in Figure B.1. Comparable alloys that are not pre-aged are much harder, Figures 5.8 and 5.10.

237 APPENDIX D STUDY ON AN AL-ZR-TI ALLOY PREPARED BY PARAMESWARAN ET AL. 237 Fig. D.1: Variation of spheroidal L1 2 precipitate radius vs. time for Al 3 (Zr 1 x Ti x) [285] (curve A) and Al 3 (Zr 1 x V x) [84] (curve B) precipitates at 425 C. From Parameswaran [285]. reported by Parameswaran et al. [285] (curve A in Figure D.1). The precipitates in Figure D.3 are, however, quite similar to those measured by Zedalis and Fine, R = 21.7 nm, on Al-Zr and Al-Zr-V alloys [84] (curve B in Figure D.1). Because of the disparity in observed precipitate size from what Parameswaran et al. [285] reported, further work on this alloy was curtailed. In hindsight, and armed with the understanding of the various precipitate morphologies discussed in Chapter 5, it is obvious by their ca. 20 nm radius and their slightly lobed morphology that the precipitates observed by the present author in Figure D.3 are interdendritic, with the dendrite center out of view toward the bottom right of the micrograph (see also Figure 5.13). It is emphasized that the alloy whose microstructure is displayed in Figure D.3 was not made by the present author; dendrite formation, and the attendant microsegregation of solutes, is ubiquitous in conventionally-solidified alloys. The difference in the curves of Figure D.1 and hence the apparent improved coarsening resistance of Al 3 (Zr 1 x Ti x ) precipitates compared to Al 3 Zr and Al 3 (Zr 1 x V x ) (L1 2 ) is therefore simply an artifact of two inidividuals viewing the same heterogeneos microstructure and reporting two different precipitate radii. After 400 h at 425 C, Parameswaran et al. [285] reported

238 APPENDIX D STUDY ON AN AL-ZR-TI ALLOY PREPARED BY PARAMESWARAN ET AL V ic k e r s m ic r o h a r d n e s s (M P a ) V ic k e r s m ic r o h a r d n e s s (H V ) Fig. D.2: Vickers microhardness vs. aging time for Al-0.19Zr-0.18Ti (at.%) at 425 C. Fig. D.3: Centered superlattice dark-field TEM micrograph of Al 3 (Zr 1 x Ti x) (L1 2 ) precipitates observed after aging at 425 C for 400 h (preceded by a 500 C 1 h pre-age). The precipitates have a mean radius R of 19.2±3.6 nm.

239 APPENDIX D STUDY ON AN AL-ZR-TI ALLOY PREPARED BY PARAMESWARAN ET AL. 239 R = 6.9 nm precipitates. In Figure D.3, the precipitate radius observed by the present author, on the same alloy, is R = 19.2 nm. It is a classic case of missing the forest for the trees, or in this case, the dendrites for the precipitates.

240 240 APPENDIX E Potential Contamination of Fe and/or Si, and its Effects on Precipitation As discussed in Sections 3.2 and 4.2, one source of Al used in preparing the alloys in this thesis was supplied by Atlantic Equipment Engineers (Bergenfield, NJ). Though nominally 99.99% pure, independent chemical analysis [376] revealed that the Al was contaminated with 260 and 257 at.ppm of Si and Fe, respectively. These are common impurities in Al, that can produce dramatically accelerated nucleation kinetics and enhanced precipitation strengthening in Al- Zr [81, 267], Al-Hf [89, 91, 92], and Al-Sc [376] alloys. Figure E.1 displays scanning electron microscope (SEM) micrographs of as-cast Al-0.2Zr- 0.2Ti (compositions in Table 5.1), displaying numerous petal-like properitectic Al 3 (Zr 1 x Ti x ) precipitates (discussed previously in Figure 5.4 and Table 5.2) causing pronounced grain refinement in this alloy. Decorating the grain boundaries are other precipitates, confirmed to be Fe-rich by energy dispersive X-ray spectroscopy (EDS) in SEM. The precipitates exhibit a lamellar structure, characteristic of a eutectic constituent, and are believed to be eutectic α-al/al 3 Fe formed at the end of solidification. It is unknown as to what degree the various alloys in this thesis are contaminated with Fe and/or Si, and efforts are currently underway to verify the concentrations of these impurities. Nevertheless, Figure E.1 clearly demonstrates that there is a sensible amount of Fe contamination in the richest Al-Zr-Ti alloy studied; it is perhaps most apparent in this alloy because this has the finest grain structure, Figure 5.1. Since all alloys were made by similar methods, it is reasonable to suspect that they are also equally contaminated, meaning that the comparisons between the various alloys are still valid. Moreover, as shown in Figure E.1(a), the interaction of Fe and/or Si with Zr and/or Ti may be limited since the solutes segregate opposite to one another during solidification; Fe and Si are eutectic-forming constituents that segregate to the grain boundaries, whereas Zr and Ti are peritectic-forming constituents that segregate to the dendrite centers.

241 APPENDIX E POTENTIAL CONTAMINATION OF FE AND/OR SI, AND ITS EFFECTS ON PRECIPITATION 241 Fig. E.1: SEM secondary and backscattered electron images displaying Fe-rich precipitates decorating the grain boundaries in assolidified Al-0.2Zr-0.2Ti. Panel (a) is a secondary electron image. Z-contrast delineates solute-rich dendritic cells (lighter contrast). Numerous petal-like Al 3 (Zr 1 x Ti x) primary precipitates are also observed, leading to pronounced grain refinement. Panel (b) is a backscattered electron image displaying a detailed view of the primary Al 3 (Zr 1 x Ti x) precipitates and Fe-rich precipitates at the grain boundaries. Panel (c) is a backscattered electron image, showing a grain boundary triple junction decorated by lamellar α-al/al 3 Fe eutectic.