S Yoshitake (Mitsubishi Materials) and V Narayan,
Phase Transformations Group,
Department of Materials Science and Metallurgy,
University of Cambridge,
Cambridge, U.K.
Provides data necessary to create a neural network model to predict lattice mismatch in nickel-based superalloys.
Superalloys based on nickel have superior performance at high temperatures (e.g around 1000°C) and may be used to to the manufacture of high-performance creep-resistant turbine blades.
The standard heat treatment of nickel superalloys results in the formation of Ni3Al cuboidal precipitates (gamma-prime phase) in the nickel alloy matrix of fcc structure (gamma phase). The superior hgh temperature performance of the superalloy is attributed to the formation of these gamma-prime phase structures, which can form more than 60% by volume of the material [1]. During service, the crystal structure may undergo anisotropic coarsening, where the gamma-prime phase precipitate is 'rafted' in various directions [2]; the gamma-prime phase precipitates that have grown in the tensile direction hinder the movement of creep dislocations and thus contribute to creep performance.
The nickel-aluminium system is the simplest nickel-based superalloy. In addition, the alloy performance may be enhanced by alloying with elements which substitute for nickel and/or aluminium atomic sites and thus change change the lattice mismatch at the gamma/gamma-prime interfaces [3]. The activation energy of rafting has been found to be related to the lattice misfit of the two phases [4], defined by the equation:-
delta = -(alpha(gamma-prime)-alpha(gamma)) / alpha(gamma)
where alpha(gamma-prime) and alpha(gamma) are the lattice constants of the gamma-prime and gamma phases respectively.
A negative misfit has been found, experimentally, to be favourable for creep resistance [5,6,7,8]. We may predict, using a neural network model, which alloying elements favour a negative misfit, so that new alloys may be designed at minimal cost.
The database used with the neural network consists of:-
Two alloy groups are used:-
The TAR file lattmisfit.tar contains five files. Nickel.misfit.info is an ASCII text version of this document; the other four files make up the neural network database as follows:-
The data in Nickel.misfit.data.ga consist of 17 columns as follows:-
Column 1 is the temperature at which the experiment was conducted, in °C.
Columns 2-17 are compositions (in weight%) of the following elements:-
The data in Nickel.misfit.data.gpa consists of 15 column as follows:-
Column 1 is the temperature at which the experiment was conducted, in °C.
Columns 2-15 are compositions (in weight%) of the following elements:-
The data in Nickel.misfit.data.gp and Nickel.misfit.data.gpb consist of 1 column, giving the is the lattice constant in Ångstroms for the gamma and gamm-prime phases respectively of each alloy.
None.
materials, data, neural, network, lattice, mismatch, nickel, superalloy, alloy, lattice parameter
MAP originated from a joint project of the National Physical Laboratory and the University of Cambridge.
MAP Website administration / map@msm.cam.ac.uk