From Trial and Error to Generative Design: The New Paradigm in Energy Materials
AI and the Energy Transition
Today, it is hard to deny that the 21st century is defined by a major climate crisis. Even if it is not a comfortable reality to acknowledge, many countries have committed to achieving climate neutrality by 2050. Initiatives such as the European Green Deal and the Paris Agreement are clear examples of this commitment. However, current technologies alone are not enough to achieve full decarbonization. We need to design new solutions that enable efficient energy storage, long-lasting batteries, and high-performance catalysts capable of delivering high yields.
Developing new materials has traditionally been a slow and costly process, grounded in scientific knowledge and largely driven by trial and error. This meant that bringing a new material from concept to application could take decades. Today, however, advances in artificial intelligence (AI) are changing this landscape. By combining data science with physics and chemistry, AI can explore material designs and combinations at a speed that was unimaginable just a few years ago. And this transformation extends across virtually every field of knowledge.
AI’s New Mission: Clean Energy
The application of AI in clean energy research is no longer a future prospect. it is already happening. From optimizing photocatalysts for solar energy to designing advanced batteries and hydrogen storage systems, machine learning techniques are helping improve materials in meaningful ways.
- Costs and development time are reduced, accelerating innovation by as much as 15 years.
- AI enables the integration of life-cycle thinking, improving overall sustainability.
- Multiple material properties can be optimized simultaneously, streamlining the development process.
However, for AI to become a truly effective tool in materials discovery, it must be carefully designed and trained.
AI Must Think Like a Scientist
AI systems need to learn the criteria scientists use when evaluating materials. They must correlate atomic structure with macroscopic properties and understand the physical constraints that govern real-world behaviours. Physical laws are often embedded into algorithms to prevent the proposal of materials that would be impossible to realize in practice.
- Machine Learning (ML) and Deep Learning: Algorithms are trained on vast datasets to predict material behaviours and performance.
- Generative Models: These models go beyond analysing known materials; they can propose entirely new structures, effectively designing novel materials from scratch.
- The Closed-Loop Approach: A specific material is proposed and tested through simulations. The goal is to validate its physical behaviours before moving on to experimental testing.
The Diversity of AI-Driven Materials for the Energy Transition
AI is accelerating the discovery of critical materials across multiple clean energy sectors. In battery research, it is used to study electrolyte degradation mechanisms. Deep learning helps optimize the hardness and durability of complex alloys, while researchers also design new biodegradable materials derived from cellulose. These are just a few examples of the vast potential unlocked by these technological tools.
Predicting the Perfect Hydrogen Storage Solution
A particularly relevant case study is the search for materials capable of storing hydrogen safely and efficiently. Hydrogen carries large amounts of clean energy, yet its storage remains a significant challenge.
This is where materials such as metal hydrides and Metal-Organic Frameworks (MOFs) come into play. AI is helping identify complex configurations that can absorb hydrogen like a sponge within the pores of their structure.
The MAST3RBOOST project applies these AI-driven techniques to develop ultra-porous materials derived from biomass waste and municipal solid waste. Through digital twins and multiscale simulations, the project aims to achieve the ambitious goal of storing 5 kg of hydrogen in a tank comparable in size to a conventional gasoline tank.
- AI-driven unsupervised algorithms accelerate the development of porous materials.
- Open data is provided under FAIR principles (Findable, Accessible, Interoperable, and Reusable).
- Manufacturing innovation through Wire-Arc Additive Manufacturing (WAAM) enables the production of customized storage tanks better adapted to vehicle design.
Ultimately, the convergence of artificial intelligence and materials science is not just about creating better products, it is about securing a future for the next generations.
Why This Matters for Climate Goals
The impact extends far beyond the laboratory; it is a vital socioeconomic driver. The ability to discover more affordable and efficient materials directly supports the goals of the 2030 Agenda and the development of sustainable cities.
Optimizing materials that avoid rare earths and other critical elements enhances competitiveness and strengthens energy resilience. Moreover, designing more sustainable materials contributes directly to the decarbonization of transport by enabling advanced battery technologies.
Towards an AI-Enhanced Materials Science
Scientific research is inherently interdisciplinary, driven by the convergence of fields that feed into one another to achieve tangible results. The use of AI in materials discovery is a clear example: computation merges with physics and chemistry to dramatically accelerate innovation.
In this new paradigm, laboratories are no longer limited to physical experimentation alone. Virtual screening, high-throughput simulations, and predictive modelling allow researchers to test thousands of potential materials before selecting the most promising candidates for experimental validation.
Author: Noemi Alonso