The phase transformation behavior of an as-cast Ti-42Al-5 Mn(at.%)alloy after subsequent quenching from 1380℃to 1000℃was investigated based on the differential thermal analysis(DTA),electron probe micro analyzer-bac...The phase transformation behavior of an as-cast Ti-42Al-5 Mn(at.%)alloy after subsequent quenching from 1380℃to 1000℃was investigated based on the differential thermal analysis(DTA),electron probe micro analyzer-backscattered electrons(EPMA-BSE),transmission electron microscope(TEM)and X-ray diffraction(XRD).The results show that,the solidification path can be summarized as follows:Liquid→Liquid+β→β→β→α→β+α+γ→βo+α2+γ→βo+γ+α2/γ→βo+γ+α2/γ+βo,sec,with the phase transformationα→βtemperature(Tβ)=1311℃,phase transformationγ→βtemperature of(Tγsolv)=1231℃,phase transformationα2→αorβo→βtemperature(Tα2→α/Tβo→β)=1168 C,eutectoid temperature(Teut)=1132℃and T(α2/γ→βo,sec)≈1120℃.In comparison with Ti-42 Al alloy,the Teut and Tγsolv are slightly increased while both the Tp is decreased obviously by 5%Mn addition.When quenched from the temperature of 1380-1260℃,the martensitic transformationβ→α'could occur to form the needlelike martensite structure inβarea.This kind of martensitic structure is much obvious with the increase of temperature from 1260℃to 1380℃.When the temperature is below Tγsolv(1231℃),theγgrains would nucleate directly from theβphase.For the temperature slightly lower than T(eut)(1132℃),the dottedβ(o,sec)phases could nucleate in the lamellar colonies besides theγlamellae precipitated withinα2 phase.Finally,at room-temperature(RT),the alloy exhibits(po+α2+γ)triple phase with microstructure ofβo+lamellae+γ,of which the lamellar structure consists ofα2,γandβo,sec phases.The phase transformation mechanisms in this alloy,involvingβ→α',β→γ,α2→α2/γandα2→βo,sec were discussed.展开更多
The design and high-throughput screening of materials using machine-learning assisted quantummechanical simulations typically requires the existence of a very large data set,often generated from simulations at a high ...The design and high-throughput screening of materials using machine-learning assisted quantummechanical simulations typically requires the existence of a very large data set,often generated from simulations at a high level of theory or fidelity.Asingle simulation at high fidelity can take on the order of days for a complex molecule.Thus,although machine learning surrogate simulations seem promising at first glance,generation of the training data can defeat the original purpose.For this reason,the use of machine learning to screen or design materials remains elusive for many important applications.In this paper we introduce a new multi-fidelity approach based on a dual graph embedding to extract features that are placed inside a nonlinear multi-step autoregressive model.Experiments on five benchmark problems,with 14 different quantities and 27 different levels of theory,demonstrate the generalizability and high accuracy of the approach.It typically requires a few 10s to a few 1000’s of high-fidelity training points,which is several orders of magnitude lower than direct ML methods,and can be up to two orders of magnitude lower than other multi-fidelity methods.Furthermore,we develop a new benchmark data set for 860 benzoquinone molecules with up to 14 atoms,containing energy,HOMO,LUMO and dipole moment values at four levels of theory,up to coupled cluster with singles and doubles.展开更多
文摘The phase transformation behavior of an as-cast Ti-42Al-5 Mn(at.%)alloy after subsequent quenching from 1380℃to 1000℃was investigated based on the differential thermal analysis(DTA),electron probe micro analyzer-backscattered electrons(EPMA-BSE),transmission electron microscope(TEM)and X-ray diffraction(XRD).The results show that,the solidification path can be summarized as follows:Liquid→Liquid+β→β→β→α→β+α+γ→βo+α2+γ→βo+γ+α2/γ→βo+γ+α2/γ+βo,sec,with the phase transformationα→βtemperature(Tβ)=1311℃,phase transformationγ→βtemperature of(Tγsolv)=1231℃,phase transformationα2→αorβo→βtemperature(Tα2→α/Tβo→β)=1168 C,eutectoid temperature(Teut)=1132℃and T(α2/γ→βo,sec)≈1120℃.In comparison with Ti-42 Al alloy,the Teut and Tγsolv are slightly increased while both the Tp is decreased obviously by 5%Mn addition.When quenched from the temperature of 1380-1260℃,the martensitic transformationβ→α'could occur to form the needlelike martensite structure inβarea.This kind of martensitic structure is much obvious with the increase of temperature from 1260℃to 1380℃.When the temperature is below Tγsolv(1231℃),theγgrains would nucleate directly from theβphase.For the temperature slightly lower than T(eut)(1132℃),the dottedβ(o,sec)phases could nucleate in the lamellar colonies besides theγlamellae precipitated withinα2 phase.Finally,at room-temperature(RT),the alloy exhibits(po+α2+γ)triple phase with microstructure ofβo+lamellae+γ,of which the lamellar structure consists ofα2,γandβo,sec phases.The phase transformation mechanisms in this alloy,involvingβ→α',β→γ,α2→α2/γandα2→βo,sec were discussed.
文摘The design and high-throughput screening of materials using machine-learning assisted quantummechanical simulations typically requires the existence of a very large data set,often generated from simulations at a high level of theory or fidelity.Asingle simulation at high fidelity can take on the order of days for a complex molecule.Thus,although machine learning surrogate simulations seem promising at first glance,generation of the training data can defeat the original purpose.For this reason,the use of machine learning to screen or design materials remains elusive for many important applications.In this paper we introduce a new multi-fidelity approach based on a dual graph embedding to extract features that are placed inside a nonlinear multi-step autoregressive model.Experiments on five benchmark problems,with 14 different quantities and 27 different levels of theory,demonstrate the generalizability and high accuracy of the approach.It typically requires a few 10s to a few 1000’s of high-fidelity training points,which is several orders of magnitude lower than direct ML methods,and can be up to two orders of magnitude lower than other multi-fidelity methods.Furthermore,we develop a new benchmark data set for 860 benzoquinone molecules with up to 14 atoms,containing energy,HOMO,LUMO and dipole moment values at four levels of theory,up to coupled cluster with singles and doubles.
