A research team, including scientists from MAX IV have reported in Nature Communications that the quest for atomic perfection in semiconductor devices was based on an oversimplified model.
Semiconductors are the fundamental building blocks of all modern electronics. Improvements to these materials could affect everything from the clock on our microwave to supercomputers used to crunch big data. The search to make them better involves looking at atomic level changes in semiconductor materials in order to understand how they could be improved, and even made perfect.
The problem with semiconductors and the way they are manufactured is that they need to be processed with metal contacts and thin insulating layers, and the interface between the semiconductor and these contacts contains a lot of defects which hamper device performance. If scientists can find a way to reduce the defects or eliminate them completely, then semiconductors could conceivably become faster and smaller. The problem is, these defects occur on the atomic scale and are very difficult to measure.
Scientists working at Max Lab, the predecessor to the newly built MAX IV, together with physicists from Lund University used the SPECIES beamline to investigate a common semiconductor synthesis method. Hafnium dioxide was deposited on the surface of indium arsenide and monitored using ambient pressure X-ray photoelectron spectroscopy (APXPS). The scientists were able to monitor the very first atomic layer that was deposited, and monitor the chemical reactions that were occurring as the process was underway.
They were expecting to see the confirmation of a model that is typically used to explain how atoms deposit on a surface called the ligand exchange model, but what they saw was different. Instead of one chemical reaction occurring at the deposition surface, there were two which had never been seen before. This new insight could explain why atomic perfection has been so difficult to achieve. Furthermore, the new model suggests a path to optimise the deposition process to get closer to atomic perfection.
These experiments were done at Max Lab and as new beamlines become commissioned at MAX IV, the question is, what does the future hold?
When moving SPECIES from MAX II to the 1.5 GeV ring at MAX IV a three-fold increase in the photon flux at the sample Is expected. Together with some recent upgrades, including a new lens system and a delay line detector, at the SPECIES APXPS end station the number of electrons arriving at the detector should show close to an order of magnitude increase.
This will allow experimental time scales to be further shortened from seconds to milliseconds, typical to chemical reactions allowing the monitoring of these reactions as they take place on the surfaces.
Another APXPS beamline, HIPPIE, on the large 3 GeV ring also provides similar capabilities to the SPECIES beamline, but with extended photon energy range. Together these beamlines offer experimental platforms for in situ and operando studies, not only in the field of atomic layer deposition, but also, for example, in the fields of catalysis, electrochemistry and environmental sciences, where the reactions on surfaces can now be studied not only in realistic conditions, but also at relevant time scales.
Self-cleaning and surface chemical reactions during hafnium dioxide atomic layer deposition on indium arsenide
Rainer Timm, Ashley R. Head, Sofie Yngman, Johan V. Knutsson, Martin Hjort, Sarah R. McKibbin, Andrea Troian, Olof Persson, Samuli Urpelainen, Jan Knudsen, Joachim Schnadt & Anders Mikkelsen
volume 9, Article number: 1412 (2018)