Effect of Cooling Rate on Aluminium-Containing High-Entropy Alloys
Abstract (summary)
In this work, a HEA system is devised based on the outcome of neural network models. The most successful neural network in this work achieves a testing accuracy of 92%. This neural network operates solely on the compositional data of alloys, as opposed to the orthodox approach of using Hume-Rothery (HR) data. Considering that an alloy's composition is always known for certain (unlike HR features that are dependent on estimates), this approach is expected to enable the average researcher to rapidly screen potential HEA compositions. The outcome of the neural network model led to the study of the AlxCrCuFeNi system, whereby x = 1.4 was predicted to be the limit of the alloy's solid-solution window. The x = 1.0, x = 1.3, x = 1.5 and x = 2.0 compositions were manufactured using an arc-melter to confirm the prediction, whereby noticeable microstructural complexities are observed in the x = 1.5 system that are not observed in the x = 1.0 and x = 1.3 systems. 'Chinese Script' and 'Sunflower' structures are observed in the x = 1.5 system, whereas the x = 2.0 system displayed a microstructure dominated by intermetallics and very brittle mechanical behaviour. The x = 1.0 and x = 1.3 alloys showed Al-Ni intermetallic needles in their interdendritic regions which adhere closely to dendrite peripheries. The x = 1.0 alloy was processed for rapid solidification using a 6.5 m long drop-tube facility. This is in order to explore the possibility of suppressing intermetallic growth and achieving a single-phase simple solid solution. The sizes of the retrieved powders ranged from ˃ 850 µm to 38 µm, with a corresponding range of cooling rates from 112 K/s to 1.13×106 K/s. With higher cooling rates, simpler microstructures are obtained and at the highest cooling rate of around 1.13 ×〖10〗^6 K/s, a microstructure free of intermetallics is observed in powders of the 38 - 53 μm size fraction. The effect of rapid cooling is also studied in the eutectic HEA (EHEA) that is AlCoCrFeNi2.1. In equilibrium conditions, AlCoCrFeNi2.1 is dual-phase L12/B2. By processing AlCoCrFeNi2.1 using the drop-tube facility, rapidly-solidified powders were achieved with sizes from 850 µm ≤ d < 1000-38 µm ≤ d < 53 µm with corresponding estimated cooling rates of 114 K/s to 1.75×106K/s respectively. Average interlamellar spacing was found to decrease from 2.10µm in the as-cast alloy to 348nm in the powders of the 38µm
Indexing (details)
Alloys;
Neural networks