Design of Tungsten-free, L12-strengthened Cobalt-based Superalloys

Ohl, Brandon

doi:https://doi.org/10.21985/n2-z7cp-mz79

Work

Design of Tungsten-free, L12-strengthened Cobalt-based Superalloys

Public

Download PDF

Download All Files (.zip)

Recent developments have enabled L12-strengthened Co-based superalloys, which have thepotential to surpass Ni-based superalloys as the material of choice for the hottest sections of turbine blades due to cobalt’s 40 ºC higher melting point. The most-studied branch of Co-based superalloys are based on the L12 phase Co3(Al,W); however, there is interest in replacing W with other γ’-formers to reduce alloy density. Here, mechanical properties and microstructural stability for twenty-one Co-based, W-free, L12-strengthened superalloys are investigated. Six Co-xNi-5Al-yCr-3V-2Ti-1.5Nb-1.5Ta-0.08B, (where x=10, 20, or 30 at.% and y=4 or 8 at.%) found that the γ + γ’ microstructure is stable for all alloys up to 1000 h at 850 ºC, with γ' area fraction increasing from 32 to 49% with increasing Ni in the alloys with 4% Cr and remaining constant at 45% in alloys with 8% Cr; solvus temperatures increase by 15-20 ºC per 10 at.% Ni addition and by 15-20 ºC as Cr doubles from 4 to 8%; average lattice misfits are between 0.6 and 1%, slightly increasing with Cr and Ni content; oxidation resistance improves with Cr, and to a lesser extent, Ni; creep behavior follows the power law for all six alloys at 850 ºC with a stress exponent of 10-12, with improved creep resistance at higher Ni content. Seven quaternary Co-Ni-Ta-Al alloys along the tie line between Ni-12.5Al and Co-12.5Ta (at.%) found that a pure γ + γ’ region exists up to ~50% Co/(Co+Ni) fraction. At 69% Co/(Co+Ni) fraction a needlelike phase precipitates, and above 85% Co/(Co+Ni) fraction, λ3 precipitates (13% in CoNi86, 36% in CoNi100). The γ phase still precipitates cuboidal γ’ at low aging times, but this discontinuously precipitates at higher aging times. Increasing Co:Ni ratios results in decreasing solidus and liquidus temperatures and increased γ/γ’ lattice misfit and γ phase tetragonal distortion. 4 Eight Co-(30-x)Ni-xFe-5Al-4/8Cr-3V-2Ti-1.5Nb-1.5Ta-0.08B (x= 10, 12, 14, 18, at.%) found that Fe substitutions for Ni decreases γ’ volume fraction, and additionally decreases the total γ + γ’ region (due to non-γ’ precipitates and associated depletion zones) with 18 at.% Fe; Ni→Fe substitutions increase solvus temperature by an average of 3 ºC per 1 at.% Fe, with negligible changes to solidus and liquidus temperatures, and doubling Cr from 4 to 8 at.% results in a consistent ~5 ºC increase in solvus and 5-10 ºC decrease in solidus and liquidus; Ni→Fe substitutions reduce ?? ′ and increase ??, resulting in a decreasing lattice misfit of ~0.02% per 1 at.% Fe, but doubling Cr from 4 to 8 at.% reduces both ?? ′ and ?? equally such that the lattice misfit is mostly constant; a marked yield strength anomaly is observed at 800 ºC with yield strength of 530-590 MPa and only weak dependence on Fe and Cr content; Fe substitutions does not affect the stress exponent (n=12) but noticeably decreases creep resistance such that with stress-atconstant-strain-rate decreasing by 5-20 MPa per at.% Fe substitution, while Cr additions improve creep resistance except in the 18Fe case. A machine-learning model was built to predict the strain rate in the steady-state regime of any Co-based superalloy at a particular temperature and stress, given inputs of alloy composition, heat treatment history, and microstructure (γ’ precipitate volume fraction). The model is trained on nearly 1,000 distinct Co-based superalloys with γ/γ’ microstructure reported in the recent literature. Instead of using CALPHAD-predicted inputs which have proven unreliable (especially in newer alloys systems such as these), we have developed additional intermediary machinelearning models for six materials properties. These models require only a compositional input to predict solvus-, solidus-, and liquidus temperatures, peak hardness, and lattice misfit and yield 5 strength (at ambient and elevated temperature). These intermediate materials properties results are fed back into the creep prediction ML model to improve its accuracy. Additionally, we validated results by predicting intermediate- and creep-properties for 16 new alloys, and experimentally determining those values. Six multi-principal-element (“high-entropy”) CoFeNi-based superalloys were produced with (i) various ratios of Co, Fe, and Ni, (ii) a constant concentration (13 at%) of γ′ formers (V, Al, Ti, Nb, and/or Ta, without W), and (iii) up to 8% Cr. The role of different ratios of γ’- to γ-formers on phase stability is investigated via calorimetry and microstructural studies after aging up to 1000 h at 850 °C, culminating in a novel W-free γ’-strengthened superalloy with equiatomic γ- and γ’- forming elements, (Co,Fe,Ni)87(V,Ti,Al)13. We find a stable, continuous γ+γ’ phase field when transitioning from W-free Co-based superalloys to the equiatomic (CoFeNi)87(V,Ti,Al)13 composition, which display a γ/γ’ microstructure with γ’ volume fraction of ~40% and without additional phases.

Creator