High Entropy Alloys Mined From Binary Phase Diagrams

Abstract

High entropy alloys (HEA) are a new type of high-performance structural material. Their vast degrees of compositional freedom provide for extensive opportunities to design alloys with tailored properties. However, compositional complexities present challenges for alloy design. Current approaches have shown limited reliability in accounting for the compositional regions of single solid solution and composite phases. For the first time, a phenomenological method analysing binary phase diagrams to predict HEA phases is presented. The hypothesis is that the HEA structural stability is encoded within the phase diagrams. Accordingly, we introduce several phase-diagram inspired parameters and employ machine learning (ML) to classify 600+ reported HEAs based on these parameters. Compared to other large database statistical prediction models, this model gives more detailed and accurate phase predictions. Both the overall HEA prediction and specifically single-phase HEA prediction rate are above 80%. To validate our method, we demonstrated its capability in predicting HEA solid solution phases with or without intermetallics in 42 randomly selected complex compositions, with a success rate of 81%. The presented search approach with high predictive capability can be exploited to interact with and complement other computation-intense methods such as CALPHAD in providing an accelerated and precise HEA design.

Accelerated crystal structure prediction of multi-elements random alloy using expandable features

“…As expected, the accuracy for binary dataset increases with a large number of features, and it is up to 91.78%. This behavior is well known and shown in most machine learning works 22 , 23 when the training set and test set are divided from the same data set.…”

“…The accuracy of HEA is comparable with the previous works that classify the phases of HEA with machine learning. For classification of bcc , fcc, and NSP (non-single phase) of HEA with support vector machine (SVM), it has 90.69% accuracy 22 and classification of bcc , fcc, and hcp phase of HEA 87 ~ 89% accuracy 23 .…”

“…This work has an advantage of saving costs for generating training set data. The existing work based on raw features and neural networks used at least 80% data of HEA in the training process to classify the phases of HEA 23 , 55 , 56 . The HEA data for training is limited because it is based on the experimental results, and it is hard to get calculation data for its substantial computational cost.…”

“…The structural phase calculation from AKAI-KKR-CPA showed consistency with 165 kinds of compositions among the 212 compositions from the experiment. For predicting the structural phase of multi-element alloy in experimental data, we make the multi-element dataset with 611 binary alloys and 106 ternary alloys from NIMS material database 43 , and 259 HEA data 23 , 25 , 44 was used. Since our attention is restricted to the d valence element, we excluded the alloy with p , s, or f valence elements in multi-element alloy.…”

“…In addition, the possible compositional number of HEA is more than 10 6 , so preparing training data set of HEA using DFT calculation like other crystal system 19 , 20 is infeasible. Although some machine learning-based approach shows accurate performance 21 , 22 , the most approaches for predicting phases of unexplored HEA are restricted to nearly equiatomic cases 23 , 24 . It is because the calculation of the non-equiatomic HEA dataset requires huge computation due to its vast compositional space 25 .…”