Aluminum oxide or alumina is the fruit fly of materials science: thoroughly researched and well understood. This compound, with the simple chemical formula Al2O3, occurs frequently in the Earth's crust in the form of the mineral corundum and its well-known color variants sapphires and rubies – and is used for a wide variety of purposes, whether in electronics, the chemical industry, or technical ceramics.
A special feature of aluminum oxide is its ability to take on different structures while maintaining the same chemical composition. All of these variants are also well understood – with one exception. In addition to several crystalline forms, aluminum oxide can also exist in an amorphous, i.e., disordered, state. Amorphous alumina has particularly advantageous properties for some high-tech applications, for example, in the form of particularly uniform protective thin-film coatings or ultrathin passivation layers.
Despite its widespread use and the know-how available for processing it, amorphous alumina remains a mystery at the atomic level. "Crystalline materials consist of small, regularly repeating subunits," explains Empa researcher Vladyslav Turlo from the Advanced Materials Processing laboratory in Thun. Thus, examining them down to the level of a single atom is relatively easy – as is modeling them on a computer. After all, if you can calculate the interaction of atoms in a single crystal unit, you can also easily calculate larger crystals consisting of many units.
Amorphous materials have no such periodic structure. The atoms are jumbled together – difficult to examine and even more difficult to model. "If we were to simulate a thin film coating of amorphous alumina grown from scratch at the atomic level, the calculation would take longer than the age of the universe," says Turlo. However, accurate simulations are the key to effective materials research: They help researchers understand materials and optimize their properties.
Experiments meet simulations
Empa researchers led by Turlo have now succeeded for the first time in simulating amorphous alumina quickly, accurately, and efficiently. Their model, which combines experimental data, high-performance simulations, and machine learning, provides information about the atomic arrangement in amorphous Al2O3 layers and is the first of its kind. The researchers have published their results in the journal npj Computational Materials.
The breakthrough was made possible thanks to interdisciplinary collaboration between three Empa laboratories. Turlo and his colleague Simon Gramatte, first author of the publication, based their model on experimental data. Researchers from the Mechanics of Materials and Nanostructures laboratory produced amorphous aluminum oxide thin films using atomic layer deposition and examined them together with colleagues at the Joining Technologies and Corrosion laboratory in Dübendorf.
One of the model's great strengths is that, in addition to the aluminum and oxygen atoms in alumina, it also considers incorporated hydrogen atoms. "Amorphous alumina contains varying amounts of hydrogen depending on the manufacturing method," explains co-author Ivo Utke. Hydrogen, the smallest element in the periodic table, is particularly challenging to measure and model.
Owing to an innovative spectroscopy method called HAXPES, which in Switzerland is only possible at Empa, the researchers were able to characterize the chemical state of aluminum in the different thin films and incorporate it into the simulations to reveal the distribution of hydrogen within alumina for the first time. "We were able to show that, above a certain content, hydrogen binds to the oxygen atoms in the material, affecting the chemical states of the other elements in the material." says co-author Claudia Cancellieri. This changes the material's properties: The aluminum oxide becomes "fluffier" i.e., less dense as a result.
Potential breakthrough for green hydrogen
This understanding of the atomic structure paves the way for new applications of amorphous aluminum oxide. Turlo sees the greatest potential in the production of green hydrogen. Green hydrogen is made by splitting water using renewable energies – or even direct sunlight. To separate hydrogen from oxygen, which is also produced during water splitting, effective filter materials are required that only allow one of the gases to pass through. "Amorphous alumina is one of the most promising materials for such hydrogen membranes," says Turlo. "Thanks to our model, we can gain a much better understanding of how the hydrogen content in the material favors the diffusion of gaseous hydrogen with respect to other larger molecules." In future, the researchers want to use the model to develop better membranes consisting of alumina.
"An understanding of our materials at the atomic level allows us to optimize the material's properties – be it related to mechanics, optics, or permeability – in a much more targeted manner," says materials researcher Utke. The model can now lead to improvements in all application areas of amorphous alumina – and may also be transferred to other amorphous materials over time. "We have shown that it is possible to accurately simulate amorphous materials," summarizes Turlo. And thanks to machine learning, the process now only takes about a day – instead of billions of years.
