High throughput experiments: PVD coating technology

PVD acronym stands for Physical Vapour Deposition. This term describes the various methods of vacuum deposition in which the target material passes from the condensed to the vapour phase and then back to the condensed phase, forming a thin film on the substrate. The most common PVD processes are sputtering and evaporation. The Compositionally Complex Alloys (CCAs) to be synthesized will be selected using a combination of CALPHAD and machine learning (ML) approaches. The FORGE project will use magnetron sputtering to synthesize the most promising CCAs. Magnetron sputtering process has a number of advantages over other PVD techniques such as high deposition rates, ease of sputtering any metal, and production of high-purity films with extremely high adhesion to the substrate, which make the method extremely useful for the efficient implementation of the FORGE assumptions.

Magnetron sputtering

In a conventional magnetron sputtering process, a PVD chamber is first evacuated to a high vacuum to minimize the partial pressures of all background gases and potential contaminants. Next, sputtering gas (e.g. argon or xenon) is introduced into the chamber. To ignite the plasma, high voltage is applied between the cathode, typically located directly behind the sputtering target, and the anode, commonly connected to the chamber as an electrical ground (Figure 1). Electrons that are present in the sputtering gas are accelerated away from the cathode causing collisions with nearby atoms and ionizing them. The positive ions of sputter gas are accelerated towards the negatively charged cathode, leading to high energy collisions with the surface of the target. The target is eroded by high-energy ions within the plasma and the sputtered atoms travel through the chamber. Finally, they deposit onto a substrate to form a thin film. It is important to note that no melting or evaporation occurs during this process, therefore the deposition of refractory metals such as Mo, W, Ta, etc., important to achieve FORGE performance targets (PTs) is not a problem.

Let's synthesize hundreds of alloys in a single process

Three pools of 6-8 elements each were selected under FORGE for different PT, i.e. CO₂ corrosion, H₂ embrittlement, erosive wear and damage. Given the number of elements in the pools and the fact that CCAs should contain at least five of them, the number of possible alloys to be analysed is enormous.

In recent years, significant improvement in the search for new metallic materials has been facilitated by the development of property-prediction models, combinatorial synthesis and high-throughput characterization techniques, e.g. high-speed nanoindentation, automatic chemical analysis and phase analysis with programmable sample-positioning stages. In contrast to conventional methods, synergy of combinatorial deposition of compositionally-focused material libraries and high-throughput characterization approach allows for efficient and cost-effective screening of a broad range of CCAs.

To synthesize hundreds of alloys in a single process we will employ PVD coating technology.  High purity elements will be sputtered and deposited onto an oriented (100) silicon wafer to achieve controlled concentration gradients of all elements in different directions at the same time. In the second stage material libraries will be characterized for chemical composition with X-Ray Fluorescence, structure with X-Ray Diffraction, and for mechanical properties with nanoindentation. The measured PTs will be compared against ML predictions. Low root mean square (RMS) deviation between measurements and ML predictions will indicate higher quality of the trained ML model. Irrespective of the RMS deviation value, we will add these new experimental measurements to the training data set and start training the machine learning model again to improve its predictive power.

Article Courtesy: EMPA