《自然》（20240725出版）一周论文导读
编译|未玖
Nature, 25 July 2024, VOL 631, ISSUE 8022
《自然》2024年7月25日，第631卷，8022期
材料科学Materials Science
Plasticity in single-crystalline Mg3Bi2 thermoelectric material
单晶Mg3Bi2热电材料的塑性
▲ 作者：Peng Zhao, Wenhua Xue, Yue Zhang, Shizhen Zhi, Xiaojing Ma, Jiamin Qiu, et al.
▲链接：
https://www.nature.com/articles/s41586-024-07621-8
▲摘要：
大多数最先进的热电材料是无机半导体。由于定向共价键，其通常在室温下表现出有限的塑性，例如，拉伸应变小于5%。
研究组发现单晶Mg3Bi2在沿（0001）平面（即ab平面）施加张力时显示出高达100%的室温拉伸应变。该值比传统热电材料至少高出一个数量级，且优于许多以类似结构结晶的金属。
实验结果表明，变形的Mg3Bi2中存在滑移带和位错，表明位错滑动是塑性变形的微观机制。化学键分析揭示了多个低滑移势垒能面，表明Mg3Bi2中存在多种滑移体系。此外，滑移过程中连续的动态键合防止了原子面解理，从而保持了较大的塑性变形。
重要的是，碲掺杂的单晶Mg3Bi2在室温下沿ab平面显示出约55μW/cm·K2平方的功率因数和约0.65的品质系数，优于现有的延展性热电材料。
▲ Abstract：
Most of the state-of-the-art thermoelectric materials are inorganic semiconductors. Owing to the directional covalent bonding, they usually show limited plasticity at room temperature, for example, with a tensile strain of less than five per cent. Here we discover that single-crystalline Mg3Bi2 shows a room-temperature tensile strain of up to 100 per cent when the tension is applied along the (0001) plane (that is, the ab plane). Such a value is at least one order of magnitude higher than that of traditional thermoelectric materials and outperforms many metals that crystallize in a similar structure. Experimentally, slip bands and dislocations are identified in the deformed Mg3Bi2, indicating the gliding of dislocations as the microscopic mechanism of plastic deformation. Analysis of chemical bonding reveals multiple planes with low slipping barrier energy, suggesting the existence of several slip systems in Mg3Bi2. In addition, continuous dynamic bonding during the slipping process prevents the cleavage of the atomic plane, thus sustaining a large plastic deformation. Importantly, the tellurium-doped single-crystalline Mg3Bi2 shows a power factor of about 55 microwatts per centimetre per kelvin squared and a figure of merit of about 0.65 at room temperature along the ab plane, which outperforms the existing ductile thermoelectric materials.
工程学Engineering
3D printable elastomers with exceptional strength and toughness
具有非凡强度和韧性的3D打印弹性体
▲ 作者：Zizheng Fang, Hongfeng Mu, Zhuo Sun, Kaihang Zhang, Anyang Zhang, Jiada Chen, et al.
▲链接：
https://www.nature.com/articles/s41586-024-07588-6
▲摘要：
三维（3D）打印已成为一种颇具吸引力的制造技术，因其在实现几何复杂的可定制产品方面具有非凡的自由度。然而，其大规模生产的潜力受到低制造效率（打印速度）和产品质量不足（机械性能）的阻碍。
光聚合物超快速3D打印技术的最新进展缓解了制造效率的问题，但典型打印聚合物的机械性能仍远远落后于传统加工技术。这是因为打印要求限制了分子设计实现高机械性能的潜力。
研究组报道了一种3D光打印树脂化学策略，其产生的弹性体抗拉强度为94.6 MPa，韧性为310.4 MJ m-3，两者均远远超过任何3D打印弹性体。从机理上讲，这通过打印聚合物中的动态共价键来实现，同时还实现了网络拓扑结构重构。
这有助于形成分层氢键（特别是酰胺氢键）、微相分离和互穿结构，从而协同促进卓越的机械性能。该工作为使用3D打印进行大规模制造提供了一个更光明的前景。
▲ Abstract：
Three-dimensional (3D) printing has emerged as an attractive manufacturing technique because of its exceptional freedom in accessing geometrically complex customizable products. Its potential for mass manufacturing, however, is hampered by its low manufacturing efficiency (print speed) and insufficient product quality (mechanical properties). Recent progresses in ultra-fast 3D printing of photo-polymers have alleviated the issue of manufacturing efficiency, but the mechanical performance of typical printed polymers still falls far behind what is achievable with conventional processing techniques. This is because of the printing requirements that restrict the molecular design towards achieving high mechanical performance. Here we report a 3D photo-printable resin chemistry that yields an elastomer with tensile strength of 94.6 MPa and toughness of 310.4 MJ m-3, both of which far exceed that of any 3D printed elastomer. Mechanistically, this is achieved by the dynamic covalent bonds in the printed polymer that allow network topological reconfiguration. This facilitates the formation of hierarchical hydrogen bonds (in particular, amide hydrogen bonds), micro-phase separation and interpenetration architecture, which contribute synergistically to superior mechanical performance. Our work suggests a brighter future for mass manufacturing using 3D printing.
人工智能Artificial Intelligence
AI models collapse when trained on recursively generated data
AI模型在使用递归生成数据进行训练时会崩溃
▲ 作者：Ilia Shumailov, Zakhar Shumaylov, Yiren Zhao, Nicolas Papernot, Ross Anderson & Yarin Gal
▲链接：
https://www.nature.com/articles/s41586-024-07566-y
▲摘要：
