Magnetic Recording Media based on High Entropy Alloys

近几十年来，磁记录介质的面密度有了巨大的增长。使数字信息能够存储在不断缩小的纳米级磁铁中的一个关键特性是介质保持其磁化的能力，即磁性各向异性。传统的高磁各向异性材料含有稀土元素，容易出现价格波动和供应链问题。目前，某些四方相的FePt合金正在热辅助磁记录介质中进行研究，但它们含有贵金属。利用富含地球元素的高各向异性材料推进磁记录技术是至关重要的。高熵合金的最新进展为实现新型高各向异性材料提供了一个令人兴奋的新领域。这些材料是具有高构型熵的多元素合金，允许稳定亚稳相。本项目将探索利用高熵合金方法开发具有高磁各向异性的新型多元素相作为传统方法无法实现的磁记录介质。为了有效地探索多元素合金的无限组合，采用了迭代合成、表征、模拟和机器学习方法，以快速识别最有前途的合金相并提高其性能指标。使用高熵合金开发的磁记录介质具有潜在的变革性技术影响。所开发的有效的机器学习方法不仅有利于高熵合金的探索，而且有利于许多其他应用。主要研究人员通过教育和推广工作积极促进更广泛的参与，以吸引各级学生，以及在磁力社区开展广泛的服务活动。研究了具有强磁各向异性的高熵材料作为磁记录介质的应用。采用一种协同迭代的高通量方法，调查了广阔的参数空间，加快了记录媒体开发的研究进展。采用组合制备方法合成了高熵合金薄膜，探索了大的成分变化。通过x射线衍射和电子显微镜对晶体结构、相稳定性和形貌进行了表征。利用磁强计和一阶反转曲线法研究了磁相定量识别和磁化反转特性。用中子散射法探测了磁相关长度尺度，并与样品的微观结构进行了比较。利用机器学习辅助合金成分和生长参数空间的设计与优化，与实验形成迭代反馈回路。这包括使用传统模型和现有数据开发预测机器学习，训练Wasserstein生成对抗网络（WGAN）来解决过拟合问题，搜索最大的磁各向异性，以及实现机器学习辅助的高通量表征。演示了一种实验-机器学习迭代推导的磁记录介质原型。通过构型熵途径克服了传统方法稳定高各向异性相的关键瓶颈。从薄膜到散装材料的最终规模化可能会改变许多其他行业。包含实验和wgan生成的标记数据的大型高熵合金数据库是公开可用的。研究活动与各种各样的教育和推广工作相结合，以培训学生和扩大代表性不足群体的参与。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

Areal density in magnetic recording media has enjoyed tremendous growth in recent decades. A key property that enables digital information to be stored in ever shrinking nanoscale magnets is the ability of the media to maintain its magnetization, known as the magnetic anisotropy. Traditional high magnetic anisotropy materials contain rare-earth elements, which are prone to price fluctuations and supply chain issues. Currently, certain alloys of FePt in a tetragonal phase are being pursued in heat-assisted magnetic recording media, but they contain precious metals. It is critical to advance the magnetic recording technology using high anisotropy materials based on earth-abundant elements. Recent progresses in high entropy alloys offer an exciting new arena to realize novel types of high anisotropy materials. These materials are multi-element alloys with high configurational entropy, allowing stabilization of metastable phases. This project will explore the high entropy alloy approach to develop novel multi-element phases with high magnetic anisotropy as magnetic recording media that cannot be achieved by conventional means. To effectively explore the essentially infinite combinations of the multi-element alloys, an iterative synthesis, characterization, simulation and machine learning approach is adapted, to quickly identify the most promising alloy phases and improve their performance metrics. Magnetic recording media developed using high entropy alloys have potentially transformative technological impacts. The effective machine learning method developed will benefit not only explorations of high entropy alloys, but also numerous other applications. The principal investigators actively promote broader participation through education and outreach efforts to engage students at all levels, as well as extensive service activities in the magnetism community. High entropy materials with strong magnetic anisotropy are investigated for applications as magnetic recording media. A synergetic and iterative high-throughput approach is employed, to survey the vast parameter space and expedite the research progress towards recording media development. Thin films of high entropy alloys are synthesized via combinatorial fabrication, exploring large composition variations. Structural characterizations of crystal structure, phase stability and morphology are performed by x-ray diffraction and electron microscopy. Quantitative magnetic phase identification and magnetization reversal characteristics are investigated using magnetometry and the first-order reversal curve method. Magnetic correlation length scales are probed by neutron scattering and compared with sample microstructures. Machine learning is employed to aid design and optimization of alloy composition and growth parameter space, forming iterative feedback loops with experiments. This includes the development of predictive machine learning using conventional models and existing data, training a Wasserstein Generative Adversarial Network (WGAN) to resolve the overfitting issue, searching for the maximized magnetic anisotropy, and realizing machine-learning-assisted high throughput characterization. An experiment-machine learning iteratively derived prototype magnetic recording media is demonstrated. A key bottleneck of stabilizing high anisotropy phases via the conventional approaches is circumvented via the configurational entropy route. The eventual scaling up from thin films into bulk materials can potentially transform many other industry sectors. A large-size high entropy alloy database containing both the experimental and WGAN-generated labeled data is publicly available. The research activities are integrated with a wide variety of education and outreach efforts for student training and broadening participation from underrepresented groups.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.