Artificial intelligence inspired design of non-isothermal aging for γ-γ' two-phase, Ni-Al alloys

Abstract

In this paper, a state-of-the-art Artificial Intelligence (AI) technique is used for a precipitation hardening of Ni-based alloy to predict more flexible non-isothermal aging (NIA) and to examine the possible routes for the enhancement in strength that may be practically achieved. Additionally, AI is used to integrate with Materials Integration by Network Technology, which is a computational workflow utilized to model the microstructure evolution and evaluate the 0.2% proof stress for isothermal aging and NIA. As a result, it is possible to find enhanced 0.2% proof stress for NIA for a fixed time of 10 min compared to the isothermal aging benchmark. The entire search space for aging scheduling was ~ 3 billion. Out of 1620 NIA schedules, we succeeded in designing the 110 NIA schedules that outperformed the isothermal aging benchmark. Interestingly, it is found that early-stage high-temperature aging for a shorter time increases the γ' precipitate size up to the critical size and later aging at lower temperature increases the γ' fraction with no anomalous change in γ' size. Therefore, employing this essence from AI, we designed an optimum aging route in which we attained an outperformed 0.2% proof stress to AI-designed NIA routes.

Figure 1

The plots show the digitalized parameters for the (a) coarse-tuned and (b) fine-tuned MCTS NIA schedules.

Figure 2

Monte Carlo tree search (MCTS) for a Ni–Al binary alloy system. The different types of NIA space are represented as a shallow tree where each node represents a possible end temperature at a certain step in the NIA route assignment. A route from root to a leaf in the maximum depth represents a full NIA schedule. A full schedule can be obtained from such a tree by using the random rollout technique. The tree is expanded iteratively towards the promising area of the search space. Each iteration consists of 4 steps: selection, expansion, simulation, and backpropagation.

Figure 3

Outline the MCTS-designed NIA routes by computational workflow management.

Figure 4

Initial microstructure (prior to NIA scheduling) of Ni-19.11 at. % Al alloy, annealed at 1000 °C for 32 s.

Figure 5

Variation of the (a) 0.2% proof stress at 725 °C as a function of aging temperature in the range of 500–900 °C and (b) fine-tuned investigation to obtain benchmark 0.2% proof stress in the range of 625–675 °C (0.2% PS: 0.2% proof stress).

Figure 6

Simulated phase-field microstructure showing the evolution of γ′ precipitates at 10 min in Ni–Al binary alloy at different isothermal aging temperatures of (a) 500 °C, (b) 525 °C, (c) 550 °C, (d) 575 °C, (e) 600 °C, (f) 625 °C, (g) 642 °C, (h) 650 °C, (i) 675 °C, (j) 700 °C, (k) 725 °C, (l) 750 °C, (m) 775 °C, (n) 800 °C, (o) 825 °C, (p) 850 °C. The microstructures are all zoomed-in images of simulated $2\mu \mathrm{m}\times 2\mu \mathrm{m}$ phase-field microstructure, as illustrated, for instance, in (a).

Figure 7

Variation of maximum 0.2% proof stress with the number of iterations as a function of different starting temperatures for a fixed time of 10 min.

Figure 8

Fine-tuned NIA scheduling for 0.2% proof stress versus the number of iterations at starting temperature of 700 °C for 10 min.

Figure 9

(a) Sketch map of NIA scheduling processes in this study, (b) γ′ phase fraction, (c) γ′ size and (d) 0.2% proof stress as a function of aging time (the reader is referred to the web version of this article for interpretation of the color references in this figure legend).

Figure 10

Simulated phase-field microstructure of best performed NIA (i.e., NIA 1) route proposed by MCTS at different stages (interval of 1 min time frame) of scheduling at (a) 700 °C, (b) 550 °C, (c) 500 °C, (d) 500 °C, (e) 550 °C, (f) 600 °C, (g) 525 °C, (h) 575 °C, (i) 600 °C, (j) 500 °C.

Figure 11

Variation of γ′ phase fraction with γ′ size of isothermal aging benchmark and outperformed NIA schedules. A black dotted box highlights the zoomed-in image (IAB isothermal aging benchmark).

Figure 12

(a) Scheduling of isothermal aging benchmark, AI-assisted maximum 0.2% proof stress NIA and AI-inspired expert-designed NIA in this study and (b) variation of 0.2% proof stress as a function of second step temperature.

Figure 13

Comparison of optimum AI-Inspired expert-designed NIA with the isothermal aging benchmark and maximum NIA by AI in terms of (a) γ′ phase fraction, (b) γ′ size and (c) 0.2% proof stress as a function of aging time. Images of the tenth minute are zoomed-in in the insets (the reader is referred to the web version of this article for interpretation of the color references in this figure legend).