To remove the key impurity elements,P and B,from primary Si simultaneously,Sr and Zr co-addition to Al-Si alloy systems during solvent refining has been investigated.Sr reacts with Al,Si,and P in the melt to form a P-...To remove the key impurity elements,P and B,from primary Si simultaneously,Sr and Zr co-addition to Al-Si alloy systems during solvent refining has been investigated.Sr reacts with Al,Si,and P in the melt to form a P-containing Al_(2)Si_(2)Sr phase and Zr reacts with B to form a ZrB_(2) phase.In the Al-Si-Sr-Zr system,high removal fractions of P and B in the primary Si,with 84.8%-98.4%and 90.7%-96.7%,respectively,are achieved at the same time,respectively.The best removal effect is obtained in the sample with the addition of Sr-32000+Zr-3000μg·kg^(-1),and the removal fractions of P and B in the purified Si reach 98.4%and 96.1%.Compared with the Sr/Zr single-addition,the removal effects of Sr and Zr co-addition on P and B do not show a significant downward trend,indicating that the nucleation and growth of the B/P-containing impurity phases are mutually independent.Finally,an evolution model is proposed to describe the nucleation and the growth stages of Sr/Zr-containing compound phases,which reveals the interaction between the impurity phases and the primary Si.展开更多
Recent progress in material data mining has been driven by high-capacity models trained on large datasets.However,collecting experimental data(real data)has been extremely costly owing to the amount of human effort an...Recent progress in material data mining has been driven by high-capacity models trained on large datasets.However,collecting experimental data(real data)has been extremely costly owing to the amount of human effort and expertise required.Here,we develop a novel transfer learning strategy to address problems of small or insufficient data.This strategy realizes the fusion of real and simulated data and the augmentation of training data in a data mining procedure.For a specific task of grain instance image segmentation,this strategy aims to generate synthetic data by fusing the images obtained from simulating the physical mechanism of grain formation and the“image style”information in real images.The results show that the model trained with the acquired synthetic data and only 35%of the real data can already achieve competitive segmentation performance of a model trained on all of the real data.Because the time required to perform grain simulation and to generate synthetic data are almost negligible as compared to the effort for obtaining real data,our proposed strategy is able to exploit the strong prediction power of deep learning without significantly increasing the experimental burden of training data preparation.展开更多
The original version of this article omitted the following from the Acknowledgements:“This work was supported by Beijing Top Discipline for Artificial Intelligent Science and Engineering,University of Science and Tec...The original version of this article omitted the following from the Acknowledgements:“This work was supported by Beijing Top Discipline for Artificial Intelligent Science and Engineering,University of Science and Technology Beijing”.This has now been corrected in both the PDF and HTML versions of the article.展开更多
基金supported by the National Natural Science Foundation of China(Nos.51804294,51874272,52111540265)Anhui Provincial Natural Science Foundation(No.1808085ME121)+4 种基金the Key Laboratory of Photovoltaic and Energy Conservation Materials,Chinese Academy of Science(No.PECL2021QN003)Hefei Institutes of Physical Science,Chinese Academy of Sciences Director’s Fund(No.YZJJZX202018)International Clean Energy Talent Program by China Scholarship CouncilOpen Foundation of State Key Laboratory of Mineral Processing(No.BGRIMM-KJSKL-2022-23)Open Foundation of State Key Laboratory of Complex Nonferrous Metal Resources Clean Utilization(No.CNMRCUKF2205)。
文摘To remove the key impurity elements,P and B,from primary Si simultaneously,Sr and Zr co-addition to Al-Si alloy systems during solvent refining has been investigated.Sr reacts with Al,Si,and P in the melt to form a P-containing Al_(2)Si_(2)Sr phase and Zr reacts with B to form a ZrB_(2) phase.In the Al-Si-Sr-Zr system,high removal fractions of P and B in the primary Si,with 84.8%-98.4%and 90.7%-96.7%,respectively,are achieved at the same time,respectively.The best removal effect is obtained in the sample with the addition of Sr-32000+Zr-3000μg·kg^(-1),and the removal fractions of P and B in the purified Si reach 98.4%and 96.1%.Compared with the Sr/Zr single-addition,the removal effects of Sr and Zr co-addition on P and B do not show a significant downward trend,indicating that the nucleation and growth of the B/P-containing impurity phases are mutually independent.Finally,an evolution model is proposed to describe the nucleation and the growth stages of Sr/Zr-containing compound phases,which reveals the interaction between the impurity phases and the primary Si.
基金The authors acknowledge financial support from the National Key Research and Development Program of China(No.2016YFB0700500)the National Science Foundation of China(No.51574027,No.61572075,No.6170203,No.61873299)+1 种基金the Finance science and technology project of Hainan province(No.ZDYF2019009)the Fundamental Research Funds for the University of Science and Technology Beijing(No.FRF-BD-19-012A,No.FRF-TP-19-043A2).
文摘Recent progress in material data mining has been driven by high-capacity models trained on large datasets.However,collecting experimental data(real data)has been extremely costly owing to the amount of human effort and expertise required.Here,we develop a novel transfer learning strategy to address problems of small or insufficient data.This strategy realizes the fusion of real and simulated data and the augmentation of training data in a data mining procedure.For a specific task of grain instance image segmentation,this strategy aims to generate synthetic data by fusing the images obtained from simulating the physical mechanism of grain formation and the“image style”information in real images.The results show that the model trained with the acquired synthetic data and only 35%of the real data can already achieve competitive segmentation performance of a model trained on all of the real data.Because the time required to perform grain simulation and to generate synthetic data are almost negligible as compared to the effort for obtaining real data,our proposed strategy is able to exploit the strong prediction power of deep learning without significantly increasing the experimental burden of training data preparation.
文摘The original version of this article omitted the following from the Acknowledgements:“This work was supported by Beijing Top Discipline for Artificial Intelligent Science and Engineering,University of Science and Technology Beijing”.This has now been corrected in both the PDF and HTML versions of the article.
