Effect Of Alloying Elements On Microstructure And Hot Tearing Susceptibility In Direct-chill Casting Of 7××× Aluminum Alloys

7xxx aluminum alloys are widely applied in aerospace due to their excellent properties.Nowadays,to attain high performance such as high strength-to-weight ratio,high damage tolerance,and good corrosion resistance,and so on,new 7xxx alloys are being developed towards high alloying content and large ingots size.These tendencies could induce serious hot tearing defects during direct-chill(DC)casting.Among all factors affecting hot tearing of 7xxx alloys,undoubtedly alloying elements play a significant role.However,to date,limited researches have been performed to investigate the influence of alloying elements(including grain refiners,main elements,and minor elements)on the as-cast microstructure and hot tearing susceptibility in DC casting of 7xxx aluminum alloys.In the context,this topic is systemically investigated by some experimental and theoretical approaches.The effects of grain size and morphology on hot tearing of AA7050 alloy were investigated by a constrained rod casting mold and a newly improved T-shaped mold simulating DC casting conditions.With the optimal addition of Al-5Ti-1B,coarse columnar grains are transformed into fine globular equiaxed grains.Meanwhile,the hot tearing susceptibility is decreased remarkably,which is attributed to the lower rigidity temperature,better feeding ability,lower strain and strain rate imposed to mushy zone and more meandering propagation paths of hot tears.The excess Al-5Ti-1B additions do not affect the grain structure but greatly increase the hot tearing susceptibility due to the agglomerations of secondary phase particles.The effects of three main and two minor alloying elements on hot tearing of non-refined and refined Al-Zn-Mg-Cu model alloys were systematically investigated by a set of experimental and theoretical approaches(including the T-shaped load measurement apparatus,microstructure observation,solidification path calculation,and SKK hot tearing criterion).The minimum and maximum crack widths are observed at Zn contents of 4 and 12 wt.%,respectively,in non-refined Al-xZn-2Mg-2Cu alloys.By comparison,the minimum and maximum values occur in 4-6 and 9 wt.%Zn,respectively,in grain-refined alloys.The increase of Mg and Cu contents has positive and negative effects on improving the hot tearing resistance in grain-refined Al-9Zn-yMg-zCu alloys,respectively.The Fe additions have no obvious effect on the hot tearing susceptibility in Al-9Zn-2Mg-2Cu-mFe alloys,while the Si additions obviously decrease the hot tearing susceptibility in Al-9Zn-2Mg-2Cu-nSi alloys.These tendencies can be attributed to the interaction among the formed stresses,melt feeding ability,final eutectics,and solid bridge.Moreover,both the load value at non-equilibrium solidus and the SKK criterion proposed by Suyitno et al.using measured load developments are found to be good indicators in predicting hot tearing susceptibility,although a contrary tendency is observed in terms of Si additions due to the early formation of solid bridges in high-Si alloys.Finally,the obtained influencing rules of main alloying elements as well as proposed experimental and theoretical approaches are further verified in four commercial 7xxx alloys.The grain size prediction in three-dimensional(3D)large ingots is an important input for the prediction of hot tearing susceptibility in DC-casting ingots.Now,it is promising to apply an as-cast Kampmann—Wagner numerical(KWN)model to predict the 3D grain size distribution in DC-casting ingots of 7xxx alloy.However,it is crucial to ensure the data quality of the Gibbs-Thomson phase diagram dataset of multi-component alloys and enable efficient access to this big dataset.This topic is now in the category of materials informatics which is an emerging field with a goal to achieve high-speed and robust acquisition,management,analysis,and dissemination of diverse materials data.Two intelligent techniques are developed,including(1)intelligent detection and repair of noisy data in a big tabulation phase diagram dataset based on an unsupervised learning algorithm;(2)parameterization and compression of the repaired big dataset based on the artificial neural network.The parameterization dataset in Al-Zn-Mg-Cu system,which has been compressed 105 times,is successfully coupled with the KWN model to predict the as-cast grain size in laboratory-scale 7xxx alloys.This not only significantly reduces the required running memory for simulation without sacrificing computational accuracy but also improves the quality of simulation.In the future,the proposed method will be applied to predict the 3D grain size distribution in industrial DC-casting 7xxx ingots.