Discovering the mysteries of thin atomic layers through machine learning

In a groundbreaking development, an interdisciplinary research team at Empa has successfully and efficiently simulated amorphous aluminium oxide on a computer for the first time. This simulation, led by Professor Jean-Luc Guespin, offers unprecedented insights into the atomic arrangement in amorphous Al2O3 layers, including the distribution of hydrogen within the material.

Amorphous materials, such as amorphous aluminium oxide, do not possess a periodic structure. Instead, their atoms are more or less randomly distributed. This randomness makes them challenging to model and simulate, especially when it comes to the smallest element in the periodic table - hydrogen. Due to its size, hydrogen is difficult to measure and model, yet it plays a crucial role in the chemical state of other elements, particularly in the case of hydrogen bindings with oxygen in materials like aluminium oxide.

Empa researchers overcame this challenge by deriving the distribution of hydrogen in aluminum oxide for the first time using a method that combines experimental data, high-performance simulations, and machine learning. They used an innovative spectroscopy method called HAXPES to characterise the chemical state of aluminum in the different thin layers.

The simulation model provides valuable information about the atomic arrangement in amorphous Al2O3 layers, revealing that the inclusion of hydrogen changes the material properties. Specifically, the aluminum oxide becomes "looser," or less dense, which is advantageous for some high-tech applications.

Amorphous aluminium oxide is frequently used in the form of protective thin layers and membranes, and this new understanding of its atomic structure paves the way for new applications, particularly in the production of green hydrogen. The heat efficiently reduces structural defects, trap states, grain boundaries, and phase impurities in amorphous materials, making them ideal for solar hydrogen and photoelectrodes.

Simulating the growth of a thin coating of amorphous aluminium oxide from scratch on an atomic level would take longer than the age of the universe with today's methods. However, the Empa team has achieved a significant milestone by atomically accurately simulating amorphous aluminium oxide with hydrogen inclusions.

This breakthrough in simulating amorphous materials is crucial for effective materials research, as it opens up new possibilities for understanding and manipulating the properties of these materials to create innovative solutions for various industries, including renewable energy.