Accelerated materials discovery using artificial intelligence, robotics and high performance computing

Traditionally, it takes up to 20 years to bring new materials to market, from ideation. Furthermore, the high cost of clean energy materials technologies is due to their constituent materials, making up more than 50% of the total technology cost.

Digitalizing the lab, using tools such as artificial intelligence, robotics and high performance computing will shorten R&D timelines, drive down material/processing costs and hasten scale-up to accelerate the entire technology development process by a factor of 10. Materials Acceleration Platforms (MAPs), are self-driving laboratories that robotically conduct materials synthesis and characterization testing/data collection. Machine learning algorithms are then used to analyze the data in real time and predict alternative reaction or processing conditions to optimize property outcomes on the fly, guiding the next round of experimentation, or “closing the loop”.

Experts at Natural Resources Canada’s CanmetMATERIALs research centre are currently advancing two MAPs, with a third under development:

1. E-MAP: New catalyst materials for clean hydrogen production and CO2 conversion.
2. TEG-MAP: New thermoelectric material for the conversion of waste heat to electricity, to support energy efficiency in vehicles.
3. 3DP-MAP: Formulation and process development for 3D-printing of metals, to support energy efficiency in vehicles.

Project objectives

Employ ML algorithms for efficient parameter space searches and decision making to recommend conditions for future experiments and random search

Expected results

The development of MAPs is critical to the advancement of clean energy technology worldwide and will provide several benefits, including:

• Opportunities to replicate project outcomes in additional MAPs focused on other subject matter areas
• Enhanced domestic and international collaboration opportunities for hydrogen production, battery technologies, CCUS technology, at the research stage
• Creation of R&D jobs, and later manufacturing.

Data requirements

• Data generated through automated materials synthesis parameters, and characterization results which can often be extremely data-intensive.