Refractory complex concentrated alloys (RCCAs) are a new class of materials with an enormous potential for high-temperature structural applications. These alloys exhibit high-temperature strength surpassing Ni superalloys, the current state-of-the-art, but, unfortunately, their corrosion resistance is far from ideal. This Designing Materials to Revolutionize and Engineer our Future (DMREF) project seeks to optimize the composition of RCCAs to achieve an unsurpassed combination of strength and oxidation resistance at high-temperatures. These properties would enable the realization of rotation detonation engines for hypersonic vehicles of interest in national defense and a significant reduction in fuel consumption and pollution over the lifetime of a land-based gas turbines that power the electric grid. In addition to providing hands-on training to graduate students, this program will support undergraduate students who will be exposed to cutting edge research tools in materials science, computer simulations and machine learning. The research team will partner with existing programs at Purdue with a track record of attracting a diverse and talented cadre of students, including underrepresented populations. To encourage widespread use of the technology and data developed, the products of this project will be made available via the nanoHUB open platform, where students, educators, and researchers can explore data and perform simulations online, using a web-browser.
The design and optimization of RCCAs with the combination of properties sought after for high temperature structural applications is a daunting technical task due to the extremely large number of potential alloys, and because the oxidation behavior of these complex alloys is not fully understood. Adding oxidation testing variables (temperature, partial pressure of O2) to the compositional ones, the space to be explored is 17 dimensional, which is clearly out of reach to brute force approaches given the time and cost involved in high-temperature oxidation experiments. Physics-based modeling could, in principle, help reduce the number of experimental trials, however, the ability to predict oxidation in complex alloys is limited. Thus, the team will develop an iterative approach that combines multi-fidelity and multi-cost experiments and physics-based modeling within a machine learning for accelerated materials discovery (ML-AMD) framework. ML-AMD will use sequential learning with deep neural networks (DNNs) to develop models based on disparate sources of information (accounting for uncertainties) and identify simulations and experiments to carry out in order to maximize information gain towards the design goal.
This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.
