《自然》(20240822出版）一周论文导读

编译｜冯维维

Nature, Volume 632 Issue 8026, 22 August 2024《自然》，第632卷，8026期，2024年8月22日

物理学Physics

Attosecond delays in X-ray molecular ionization

X射线分子电离的阿秒延迟

▲ 作者：Taran Driver, Miles Mountney, Jun Wang, Lisa Ortmann, Andre Al-Haddad, Nora Berrah, Christoph Bostedt, Elio G. Champenois, Louis F. DiMauro, Joseph Duris, Douglas Garratt, James M. Glownia, Zhaoheng Guo, Daniel Haxton, Erik Isele, Igor Ivanov, Jiabao Ji, Andrei Kamalov, Siqi Li, Ming-Fu Lin, Jon P. Marangos, Razib Obaid, Jordan T. O’Neal, Philipp Rosenberger, James P. Cryan Show authors

▲ 链接：https://www.nature.com/articles/s41586-024-07771-9

▲ 摘要：光电效应并非真正瞬时的，而是表现出阿秒的延迟，可以揭示复杂的分子动力学。亚飞秒持续时间的光脉冲为解决光电离动力学提供了必要的工具。过去十年，人们对单光子吸收极紫外光子后的光电离延迟进行了大量的研究。然而，测量时间分辨核能级的光离仍然遥不可及。所需的X射线光子能量所需的核心级光电离是不可用阿秒桌面源。

研究者报告了对核心能级电子的X射线光发射延迟的测量，其延迟出乎意料地大，在氧K壳层阈值附近的NO中高达700阿秒（1阿秒=10-18秒）。这些测量利用来自自由电子激光器的阿秒软X射线脉冲扫描K壳层阈值附近的整个区域。

研究者还发现延迟谱是丰富调制的，这表明有几个贡献，包括由于形状共振引起的光电子的瞬态捕获，在分子的快速非辐射弛豫中发射的与奥格-迈特纳电子的碰撞以及多电子散射效应。结果表明，在综合理论模型的支持下，X射线阿秒实验可以揭示核能级光电离复杂的相关动力学。

▲ Abstract：The photoelectric effect is not truly instantaneous but exhibits attosecond delays that can reveal complex molecular dynamics. Sub-femtosecond-duration light pulses provide the requisite tools to resolve the dynamics of photoionization. Accordingly, the past decade has produced a large volume of work on photoionization delays following single-photon absorption of an extreme ultraviolet photon. However, the measurement of time-resolved core-level photoionization remained out of reach. The required X-ray photon energies needed for core-level photoionization were not available with attosecond tabletop sources. Here we report measurements of the X-ray photoemission delay of core-level electrons, with unexpectedly large delays, ranging up to 700?as in NO near the oxygen K-shell threshold. These measurements exploit attosecond soft X-ray pulses from a free-electron laser to scan across the entire region near the K-shell threshold. Furthermore, we find that the delay spectrum is richly modulated, suggesting several contributions, including transient trapping of the photoelectron owing to shape resonances, collisions with the Auger–Meitner electron that is emitted in the rapid non-radiative relaxation of the molecule and multi-electron scattering effects. The results demonstrate how X-ray attosecond experiments, supported by comprehensive theoretical modelling, can unravel the complex correlated dynamics of core-level photoionization.

A hot-emitter transistor based on stimulated emission of heated carriers

基于受热载流子受激发射的热发射极晶体管

▲ 作者：Chi Liu, Xin-Zhe Wang, Cong Shen, Lai-Peng Ma, Xu-Qi Yang, Yue Kong, Wei Ma, Yan Liang, Shun Feng, Xiao-Yue Wang, Yu-Ning Wei, Xi Zhu, Bo Li, Chang-Ze Li, Shi-Chao Dong, Li-Ning Zhang, Wen-Cai Ren, Dong-Ming Sun & Hui-Ming Cheng▲ 链接：https://www.nature.com/articles/s41586-024-07785-3

▲摘要：热载流子晶体管是一类利用载流子的多余动能的器件。与依赖稳态载流子传输的常规晶体管不同，热载流子晶体管将载流子调制到高能状态，从而提高了器件的速度和功能。这些特性对于需要快速切换和高频操作的应用至关重要，例如先进的电信和尖端的计算技术。

然而，传统的热载流子产生机制要么是载流子注入，要么是加速，这在功耗和负差分电阻方面限制了器件性能。混合维器件结合了大块材料和低维材料，可以通过利用由能带组合形成的各种势垒，为热载流子产生提供不同的机制。

研究者报告了一种基于双混合维石墨烯/锗肖特基结的热发射极晶体管，该晶体管利用受热载流子的受激发射，实现了超过玻尔兹曼极限的每十年低于1毫伏的亚阈值摆幅，以及在室温下峰谷电流比大于100的负差分电阻。进一步演示了具有高逆变增益和可重构逻辑状态的多值逻辑。

这项工作报告了一种多功能热发射极晶体管，具有低功耗和负差分电阻应用的巨大潜力，标志着后摩尔时代的一个有希望的进步。

▲ Abstract：Hot-carrier transistors are a class of devices that leverage the excess kinetic energy of carriers. Unlike regular transistors, which rely on steady-state carrier transport, hot-carrier transistors modulate carriers to high-energy states, resulting in enhanced device speed and functionality. These characteristics are essential for applications that demand rapid switching and high-frequency operations, such as advanced telecommunications and cutting-edge computing technologies. However, the traditional mechanisms of hot-carrier generation are either carrier injection or acceleration, which limit device performance in terms of power consumption and negative differential resistance. Mixed-dimensional devices, which combine bulk and low-dimensional materials, can offer different mechanisms for hot-carrier generation by leveraging the diverse potential barriers formed by energy-band combinations. Here we report a hot-emitter transistor based on double mixed-dimensional graphene/germanium Schottky junctions that uses stimulated emission of heated carriers to achieve a subthreshold swing lower than 1?millivolt per decade beyond the Boltzmann limit and a negative differential resistance with a peak-to-valley current ratio greater than 100 at room temperature. Multi-valued logic with a high inverter gain and reconfigurable logic states are further demonstrated. This work reports a multifunctional hot-emitter transistor with significant potential for low-power and negative-differential-resistance applications, marking a promising advancement for the post-Moore era.

生态学Ecology

Unforeseen plant phenotypic diversity in a dry and grazed world

干旱和放牧世界中不可预见的植物表型多样性

▲ 作者：Nicolas Gross, Fernando T. Maestre, Pierre Liancourt, Miguel Berdugo, Rapha?l Martin, Beatriz Gozalo, Victoria Ochoa, Manuel Delgado-Baquerizo, Vincent Maire, Hugo Saiz, Santiago Soliveres, Enrique Valencia, David J. Eldridge, Emilio Guirado, Franck Jabot, Sergio Asensio, Juan J. Gaitán, Miguel García-Gómez, Paloma Martínez, Jaime Martínez-Valderrama, Betty J. Mendoza, Eduardo Moreno-Jiménez, David S. Pescador, César Plaza,Yoann Le Bagousse-Pinguet▲ 链接：https://www.nature.com/articles/s41586-024-07731-3

▲摘要：地球拥有非凡的植物表型多样性，但正受到持续的全球变化的威胁。然而，目前尚不清楚日益增加的干旱和牲畜放牧压力——这两个全球变化的主要驱动因素——如何塑造植物表型多样性基础上的性状共变。研究者评估了全球干旱地区20种化学和形态特征的共变对干旱和放牧压力的响应。

他们的分析涉及6大洲326个样地调查的301种多年生植物的133769个性状测量值，1347个观测值。超过约0.7的干旱阈值(接近半干旱区和干旱区之间的过渡)导致性状多样性意外增加88%。这个阈值出现在放牧动物存在的情况下，并随着放牧压力的增加向较低的干旱水平移动。

此外，57%观察到的性状多样性仅发生在最干旱和放牧的旱地，突出了这些极端环境的表型独特性。研究表明，旱地是全球植物表型多样性的储存库，挑战了恶劣环境条件降低植物性状多样性的普遍观点。他们还强调，许多替代策略可能使植物能够应对气候变化和土地利用集约化引起的环境压力增加。

▲ Abstract：Earth harbours an extraordinary plant phenotypic diversity1 that is at risk from ongoing global changes. However, it remains unknown how increasing aridity and livestock grazing pressure—two major drivers of global change—shape the trait covariation that underlies plant phenotypic diversity. Here we assessed how covariation among 20 chemical and morphological traits responds to aridity and grazing pressure within global drylands. Our analysis involved 133,769 trait measurements spanning 1,347 observations of 301 perennial plant species surveyed across 326 plots from 6 continents. Crossing an aridity threshold of approximately 0.7 (close to the transition between semi-arid and arid zones) led to an unexpected 88% increase in trait diversity. This threshold appeared in the presence of grazers, and moved toward lower aridity levels with increasing grazing pressure. Moreover, 57% of observed trait diversity occurred only in the most arid and grazed drylands, highlighting the phenotypic uniqueness of these extreme environments. Our work indicates that drylands act as a global reservoir of plant phenotypic diversity and challenge the pervasive view that harsh environmental conditions reduce plant trait diversity. They also highlight that many alternative strategies may enable plants to cope with increases in environmental stress induced by climate change and land-use intensification.

Low-latitude mesopelagic nutrient recycling controls productivity and export

低纬度中上层的养分循环控制着生产力和出口

▲ 作者：Keith B. Rodgers, Olivier Aumont, Katsuya Toyama, Laure Resplandy, Masao Ishii, Toshiya Nakano, Daisuke Sasano, Daniele Bianchi & Ryohei Yamaguchi

▲ 链接：https://www.nature.com/articles/s41586-024-07779-1

▲ 摘要：低纬度海洋（LL）占全球净初级生产和出口的一半。研究认为，南大洋主导着海洋初级生产和出口，这对全球初级生产和出口对气候变化的响应具有启示意义。新研究对单个模型进行了观察分析和敏感性研究，结果表明，72%的LL初级生产和55%的出口受当地中上层宏量营养素循环的控制。

总体而言，34%的LL输出是由南大洋提供的预形成的大量营养物质维持的，这些营养物质通过较深的翻覆单元提供，而浅层预形成的向北供应，穿过南纬30度，通过亚极和温跃层水团，仅维持7%的LL输出。

对1850-2300年高排放低缓解[共享社会经济路径（SSP5-8.5）]和低排放高缓解（SSP1-2.6）气候情景下运行的5个耦合模式比对项目第6阶段（CMIP6）模式的分析表明，它们对LL初级生产的预估不仅在幅度上，而且在标志上都存在显著差异。

在更强的SSP5-8.5强迫下，随着上层海洋变暖更加明显，考虑温度依赖的再矿化作用的CMIP6模式促进了增温条件下LL中上层营养物质潴留的增强，这对稳定或增加（而不是减少）高排放低减缓条件下LL产量提供了一级贡献。这强调了对中远洋再矿化及其对海洋变暖敏感性的机制理解对于预测未来生态系统变化的重要性。

▲ Abstract：Low-latitude (LL) oceans account for up to half of global net primary production and export. It has been argued that the Southern Ocean dominates LL primary production and export, with implications for the response of global primary production and export to climate change. Here we applied observational analyses and sensitivity studies to an individual model to show, instead, that 72% of LL primary production and 55% of export is controlled by local mesopelagic macronutrient cycling. A total of 34% of the LL export is sustained by preformed macronutrients supplied from the Southern Ocean via a deeper overturning cell, with a shallow preformed northward supply, crossing 30°?S through subpolar and thermocline water masses, sustaining only 7% of the LL export. Analyses of five Coupled Model Intercomparison Project Phase 6 (CMIP6) models, run under both high-emissions low-mitigation (shared socioeconomic pathway (SSP5-8.5)) and low-emissions high-mitigation (SSP1-2.6) climate scenarios for 1850–2300, revealed significant across-model disparities in their projections of not only the amplitude, but also the sign, of LL primary production. Under the stronger SSP5-8.5 forcing, with more substantial upper-ocean warming, the CMIP6 models that account for temperature-dependent remineralization promoted enhanced LL mesopelagic nutrient retention under warming, with this providing a first-order contribution to stabilizing or increasing, rather than decreasing, LL production under high emissions and low mitigation. This underscores the importance of a mechanistic understanding of mesopelagic remineralization and its sensitivity to ocean warming for predicting future ecosystem changes.

化学Chemistry

The catalytic asymmetric polyene cyclization of homofarnesol to ambrox

同型芳尼醇催化不对称多烯环化制氨溴

▲ 作者：Na Luo, Mathias Turberg, Markus Leutzsch, Benjamin Mitschke, Sebastian Brunen, Vijay N. Wakchaure, Nils N?thling, Mathias Schelwies, Ralf Pelzer & Benjamin List

▲ 链接：https://www.nature.com/articles/s41586-024-07757-7

▲摘要：多烯环化是生物学中最复杂和最具挑战性的转化之一。在一个反应步骤中，多个碳碳键、环体系和立体中心由简单的无环前体组成。同时实现这种对产品分布和立体化学的精确控制对化学家来说是一项艰巨的任务。

特别是，(3E,7E)-同型芳醇的多烯环化成天然存在的有价值的龙柏香气味剂(?)-降龙涏香醚被认为是化学合成中长期存在的挑战。

研究者报告了在氟化醇存在下，使用高布朗斯特酸性和限制性咪胺二磷酰化催化剂，通过催化不对称多烯环化合成(?)-降龙涏香醚和倍半萜内酯天然产物(+)-香紫苏内酯的非对映选择性和对映选择性。

包括氘标记研究在内的几个实验表明，该反应主要通过与Stork-Eschenmoser假说一致的途径进行。机理研究表明，咪胺二磷酰化催化剂的酶样微环境对于获得异常高的选择性具有重要意义，而以前认为只有在酶催化的多烯环化中才能实现。

▲ Abstract：Polyene cyclizations are among the most complex and challenging transformations in biology. In a single reaction step, multiple carbon–carbon bonds, ring systems and stereogenic centres are constituted from simple, acyclic precursors. Simultaneously achieving this kind of precise control over product distribution and stereochemistry poses a formidable task for chemists. In particular, the polyene cyclization of (3E,7E)-homofarnesol to the valuable naturally occurring ambergris odorant (?)-ambrox is recognized as a longstanding challenge in chemical synthesis. Here we report a diastereoselective and enantioselective synthesis of (?)-ambrox and the sesquiterpene lactone natural product (+)-sclareolide by a catalytic asymmetric polyene cyclization by using a highly Br?nsted-acidic and confined imidodiphosphorimidate catalyst in the presence of fluorinated alcohols. Several experiments, including deuterium-labelling studies, suggest that the reaction predominantly proceeds through a concerted pathway in line with the Stork–Eschenmoser hypothesis. Mechanistic studies show the importance of the enzyme-like microenvironment of the imidodiphosphorimidate catalyst for attaining exceptionally high selectivities, previously thought to be achievable only in enzyme-catalysed polyene cyclizations.

人工智能Artificial Intelligence

Loss of plasticity in deep continual learning

在深度持续学习中丧失可塑性

▲ 作者：Shibhansh Dohare, J. Fernando Hernandez-Garcia, Qingfeng Lan, Parash Rahman, A. Rupam Mahmood & Richard S. Sutton

▲ 链接：https://www.nature.com/articles/s41586-024-07711-7

▲摘要：人工神经网络、深度学习方法和反向传播算法构成了现代机器学习和人工智能的基础。这些方法几乎总是在两个阶段使用，一个阶段是更新网络的权重，另一个阶段是在使用或评估网络时保持权重不变。这与自然学习和许多需要持续学习的应用形成对比。

目前还不清楚深度学习方法在持续学习环境中是否有效。研究者证明了它们并非如此——标准的深度学习方法在持续学习环境中逐渐失去可塑性，直到它们的学习效果不如浅网络。

研究者使用经典的ImageNet数据集和跨网络和学习算法的广泛变化的强化学习问题来显示这种可塑性的丧失。只有通过不断向网络中注入多样性的算法，才能无限期地保持可塑性，例如持续反向传播算法，这是反向传播的一种变体，其中一小部分较少使用的单元不断地随机重新初始化。

研究结果表明，基于梯度下降的方法是不够的——持续的深度学习需要一个随机的、非梯度的成分来保持可变性和可塑性。

▲ Abstract：Artificial neural networks, deep-learning methods and the backpropagation algorithm1 form the foundation of modern machine learning and artificial intelligence. These methods are almost always used in two phases, one in which the weights of the network are updated and one in which the weights are held constant while the network is used or evaluated. This contrasts with natural learning and many applications, which require continual learning. It has been unclear whether or not deep learning methods work in continual learning settings. Here we show that they do not—that standard deep-learning methods gradually lose plasticity in continual-learning settings until they learn no better than a shallow network. We show such loss of plasticity using the classic ImageNet dataset and reinforcement-learning problems across a wide range of variations in the network and the learning algorithm. Plasticity is maintained indefinitely only by algorithms that continually inject diversity into the network, such as our continual backpropagation algorithm, a variation of backpropagation in which a small fraction of less-used units are continually and randomly reinitialized. Our results indicate that methods based on gradient descent are not enough—that sustained deep learning requires a random, non-gradient component to maintain variability and plasticity.