Development of Interatomic Potential for Al-Tb Alloy by Deep Neural Network Learning Method

Short- and medium-range orders in Al90Tb10 glass and their relation to the structures of competing crystalline phases

“…The trained neural network model reproduces well the ab initio results of energies and forces in the training data set. For more details of our neural network interatomic potential for Al-Tb alloys, we refer the readers to our previous publication [31]. Our previous study [31] showed that MD simulations using the obtained deep potential reproduce accurately the AIMD results of Al90Tb10 liquid.…”

“…For more details of our neural network interatomic potential for Al-Tb alloys, we refer the readers to our previous publication [31]. Our previous study [31] showed that MD simulations using the obtained deep potential reproduce accurately the AIMD results of Al90Tb10 liquid. The success in the studies of Al90Tb10 liquid provides us with a promising opportunity to study the short-to medium-range orders in Al-Tb metallic glass at an accuracy comparable to AIMD while computationally much faster.…”

“…Recently, we have developed an interatomic potential for Al-Tb system using a neural network deep machine learning method [28][29][30][31]. The data set used in the machine learning to train the interatomic potential is composed of snapshots of liquid Al90Tb10, pure Al and pure Tb liquids, and crystal structures of fcc Al, hcp Tb, Al17Tb2, Al4Tb, Al3Tb, Al2Tb, AlTb, Al2Tb3, Al2Tb and AlTb3.…”