ARTICLES The mechanism of morphogenesis in a phase-separating concentrated multicomponent alloy ZUGANG MAO 1 , CHANTAL K. SUDBRACK 1 , KEVIN E. YOON 1 , GEORGES MARTIN 1,2 AND DAVID N. SEIDMAN 1,3 * 1 Department of Materials Science and Engineering, Northwestern University, Evanston, Illinois 60208, USA 2 Commissariat ` a l’ ´ Energie Atomique, Cabinet du Haut Commissaire, B ˆ atiment Si ` ege, 91191 Gif sur Yvette Cedex, France 3 Northwestern University Centre for Atom-Probe Tomography, Northwestern University, Cook Hall, 2220 Campus Drive, Evanston, Illinois 60208, USA *e-mail: d-seidman@northwestern.edu Published online: 25 February 2007; doi:10.1038/nmat1845 What determines the morphology of a decomposing alloy? Besides the well-established effect of the nucleation barrier, we demonstrate that, in a concentrated multicomponent Ni(Al,Cr) alloy, the details of the diffusion mechanism strongly affect the kinetic pathway of precipitation. Our argument is based on the combined use of atomic-scale observations, using three-dimensional atom-probe tomography (3D APT), lattice kinetic Monte Carlo simulations and the theory of diffusion. By an optimized choice of thermodynamic and kinetic parameters, we first reproduce the 3D APT observations, in particular the early-stage transient occurrence of coagulated precipitates. We then modify the kinetic correlations among the atomic fluxes in the simulation, without altering the thermodynamic driving force for phase separation, by changing the vacancy–solute interactions, resulting in a suppression of coagulation. Such changes can only be quantitatively accounted for with non-zero values for the off-diagonal terms of the Onsager matrix, at variance with classical models. Controlling the precipitation microstructure in alloys, by appropriate heat treatments, is at the core of metallurgical skill. With the achievement of atomic-scale resolution with direct observations, using three-dimensional atom-probe tomography (3D APT) 1–3 , and with simulation, using lattice kinetic Monte Carlo (LKMC) techniques 4 , it is now possible to study quantitatively the formation mechanism of the earliest stages of the precipitation microstructure. Provided that the diffusion mechanism is fully taken into account, LKMC enables an excellent simulation of the kinetic pathway for nucleation, growth and coarsening observed at the atomic scale by 3D APT in real alloys. This has been demonstrated 5–8 for the decomposition of a ternary Ni(Al,Cr) supersaturated solution, which is a model for more complex nickel- based superalloys. Recently, the combined use of 3D APT and LKMC has elucidated the origin of the ‘core shell’ structure of Al 3 (Sc,Zr) precipitates in Al, which is attributed to the details of the diffusion mechanism both in the supersaturated Al(Sc,Zr) solid solution and in Al 3 (Sc,Zr) (ref. 9). On the basis of these successes, we address the still unanswered question of the influence of diffusional correlation effects 10 and vacancy properties on the nucleation, growth and coarsening pathway in concentrated alloys. In contrast to LKMC, the phenomenological descriptions of precipitation kinetics, for example, as used in DICTRA-type modelling 11 , rely on two important simplifying assumptions that ignore the latter potential effects: the Onsager matrix is diagonal and vacancies are everywhere at equilibrium 12 . As shown by LKMC studies of dilute alloys, depending on the vacancy properties (that is, both the magnitude of vacancy–solute binding energies and the local composition dependence of the energy of the saddle-point configuration), solute atoms diffuse individually or as small clusters with various degrees of kinetic correlation. The potential importance of cluster migration was revealed by simulations of the ferromagnetic kinetic Ising model with conserved total spin, either with direct or vacancy- mediated exchange dynamics 13–15 . On the basis of a more realistic expression for the activation barrier for atom–vacancy exchange 16 , several LKMC studies have revealed the impact of cluster migration on the kinetic pathway (for example, the existence of an incubation period) in phase-separating alloys 4,17–20 . How do the above conclusions relate to multicomponent alloys with high solubility limits, a situation where the concept of migrating clusters loses its physical significance? In concentrated solutions, we are left with phenomenological parameters (the matrix of Onsager coefficients, chemical potentials, diffusivity matrix), which encompass the atomistics of the alloy. We demonstrate how combined atomic-scale studies of the decomposition of a supersaturated Ni(Al,Cr) solution, using 3D APT observations and LKMC simulations, reveal the effects of flux coupling and correlation on the kinetic pathway for nucleation, growth and coarsening in a concentrated ternary Ni-base alloy, which is a model for complex superalloys. DECOMPOSITION OF Ni(Al,Cr) SOLID SOLUTIONS Solution-treated Ni-5.2 Al-14.2 Cr at.% alloys were quenched and then aged at 873 K for increasing amounts of time (2 min nature materials ADVANCE ONLINE PUBLICATION www.nature.com/naturematerials 1 © 2007 Nature Publishing Group