From carbon electrolysis to catalytic structures: Experimentation and AI "allied" to transform greenhouse gases into fuels and useful chemicals.

In March 2025, EMPA, the prestigious Swiss institute for materials and technologies, awarded the young researcher Carlota Bozal-Ginesta, arrived from Spain, the “Young Scientist Fellowship”. The two-year grant allows her to develop a project that combines machine learning algorithm and high-throughput experimentation to optimize electrode structure in CO2 electrolysis.

Its goal is ambitious and clear: transform the greenhouse gas par excellence into a precious resource, generating synthetic fuels and chemicals usable in industry, through the use of water, renewable electricity and suitable catalysts.

The challenge arises from a known problem. The available catalysts, although efficient, do not guarantee a sufficient selectivityThe electrolysis process produces a complex mixture of gaseous and liquid compounds, often more than twenty, the separation of which is costly and energy-intensive.

Bozal-Ginesta's innovation consists in investigating how the electrode microstructure influence the composition of products, with the help of algorithms capable of recognizing hidden correlations. An approach that, in Switzerland, combines microscopy, spectroscopy, and simulations to propose new electrode geometries.

New scenarios: more materials and advanced configurations

The most recent developments in the field of electrochemical CO2 reduction show an increasing attention to the electrode designInternational research, particularly in Europe, has highlighted how configurations based on gas diffusion electrodes e assembled electrode membranes can improve mass transportation management and increase efficiency.

However, realistic operating conditions remain complex: the presence of impurities, pressure and temperature variations, and the stability of materials over time are factors that hinder the transition from the laboratory to the pilot plant.

In this context, i mathematical models and multiphysics simulations They are proving to be fundamental tools. Analyzing electrode behavior as a function of catalyst thickness, porosity, or electroactive area allows predicting performance even before building a prototype. It is precisely this synergy between rapid experimentation and predictive modeling that Bozal-Ginesta's work finds its innovative strength.

The impact of AI on electrode design

The use of the machine learning algorithm It does not limit itself to processing large amounts of experimental data. Carlota Bozal-Ginesta has set three lines of action.
The first concerns the study of the correlations between the electrode structure and electrochemical performance, with the aim of understanding which parameters are truly decisive for increasing selectivity.

The second is the development of digital tools capable of analyzing microscopy images, spotting patterns and details that the human eye misses.
Finally, the third axis aims at the so-called inverse design: starting from the desired product, the algorithm suggests how to build the electrode with optimal structural characteristics.

The Catalan researcher, however, underlines the need for a responsible approach:

Machine learning is a powerful tool for testing our hypotheses, but it can't replace the critical thinking of scientists. We need curated data, solid models, and informed interpretations, otherwise we risk following false leads.

says.

From the laboratory to the industry: obstacles and opportunities

Despite progress, barriers to industrial application remain substantial. Electrodes must operate with high current densities and maintain stability over long periods, conditions far from laboratory testing.

Corrosion, loss of porosity, and catalyst degradation are problems that compromise the life of the devices. Furthermore, the balance between selectivity and energy efficiency It's delicate: greater precision in producing a compound often means higher voltages and higher energy costs.

The international scientific community is working to standardize methodologies and datasets, so as to facilitate comparison between laboratories and accelerate technology transfer. It is an area in which the European collaboration plays a strategic role, combining expertise in materials, chemical engineering and computational science.

The Italian contribution in the words of Alessandro Bianchi

Even in Italy Interest in these technologies is growing, with the involvement of universities and research centers such as the Polytechnic University of Milan, ENEA, and the CNR. Professor Alessandro Bianchi, head of the electrochemical group at the National Agency for New Technologies, Energy and Sustainable Economic Development, emphasizes:

"If we want Italy to move beyond being a spectator, we need to invest in high-performance experimental platforms and well-organized data infrastructures. Only then can we develop truly competitive prototypes."

The expert's words reflect a shared belief: the energy transition cannot ignore CO2 capture and utilization technologies.

In a country aiming to increase renewable energy production, the ability to convert excess carbon into synthetic fuels offers a way to reduce dependence on imported fossil fuels and open up emerging markets.

Towards 2030: Digital Electrodes for a Circular Economy

As of mid-September 2025, the landscape appears to be rapidly evolving. multiphysics models and machine learning algorithm are transforming research into a field where traditional chemistry meets digital chemistryIt is plausible that by 2030 we will see the first pilot plants capable of producing synthetic fuels or light olefins at levels of efficiency compatible with industrial scale.

Carlota Bozal-Ginesta's work at the Swiss Federal Laboratories for Materials Science and Technology is not only a scientific contribution, but an example of how a interdisciplinary approach can generate value. By integrating data, models and a critical vision, scenarios open up in which CO2 stops being just atmospheric waste and becomes raw material for a circular economy.

The direction is clear: simply seeking better catalysts isn't enough. We need to rethink structures, processes, and design criteria. Those who master this frontier, transforming electrochemistry into an ally of artificial intelligence, will have the potential to transform a global problem into an unprecedented industrial and environmental opportunity.

Here are three insights that might interest you:

Acetaldehyde from CO2: towards a more sustainable industrial chemistry
Gamma-ray lasers: the extreme light revolution is coming
New fuels from artificial photosynthesis on soap bubbles