Pub Date : 2025-01-03DOI: 10.1038/s41566-024-01568-y
David Schmitt, Jan Philipp Bange, Wiebke Bennecke, Giuseppe Meneghini, AbdulAziz AlMutairi, Marco Merboldt, Jonas Pöhls, Kenji Watanabe, Takashi Taniguchi, Sabine Steil, Daniel Steil, R. Thomas Weitz, Stephan Hofmann, Samuel Brem, G. S. Matthijs Jansen, Ermin Malic, Stefan Mathias, Marcel Reutzel
Understanding the impact of spatial heterogeneity on the behaviour of two-dimensional materials represents one of the grand challenges in applying these materials in optoelectronics and quantum information science. For transition metal dichalcogenide heterostructures in particular, direct access to heterogeneities in the dark-exciton landscape with nanometre spatial and ultrafast time resolution is highly desired but remains largely elusive. Here we report how ultrafast dark-field momentum microscopy can spatio-temporally resolve dark-exciton formation dynamics in a twisted WSe2/MoS2 heterostructure with a time resolution of 55 fs and a spatial resolution of 480 nm. This enables us to directly map spatial heterogeneity in the electronic and excitonic structure, and to correlate this with the dark-exciton formation and relaxation dynamics. The advantage of the simultaneous ultrafast nanoscale dark-field momentum microscopy and spectroscopy reported here is that it enables spatio-temporal imaging of the photoemission spectral function that carries energy- and momentum-resolved information on the single-particle band structure, many-body interactions and correlation phenomena.
{"title":"Ultrafast nano-imaging of dark excitons","authors":"David Schmitt, Jan Philipp Bange, Wiebke Bennecke, Giuseppe Meneghini, AbdulAziz AlMutairi, Marco Merboldt, Jonas Pöhls, Kenji Watanabe, Takashi Taniguchi, Sabine Steil, Daniel Steil, R. Thomas Weitz, Stephan Hofmann, Samuel Brem, G. S. Matthijs Jansen, Ermin Malic, Stefan Mathias, Marcel Reutzel","doi":"10.1038/s41566-024-01568-y","DOIUrl":"https://doi.org/10.1038/s41566-024-01568-y","url":null,"abstract":"<p>Understanding the impact of spatial heterogeneity on the behaviour of two-dimensional materials represents one of the grand challenges in applying these materials in optoelectronics and quantum information science. For transition metal dichalcogenide heterostructures in particular, direct access to heterogeneities in the dark-exciton landscape with nanometre spatial and ultrafast time resolution is highly desired but remains largely elusive. Here we report how ultrafast dark-field momentum microscopy can spatio-temporally resolve dark-exciton formation dynamics in a twisted WSe<sub>2</sub>/MoS<sub>2</sub> heterostructure with a time resolution of 55 fs and a spatial resolution of 480 nm. This enables us to directly map spatial heterogeneity in the electronic and excitonic structure, and to correlate this with the dark-exciton formation and relaxation dynamics. The advantage of the simultaneous ultrafast nanoscale dark-field momentum microscopy and spectroscopy reported here is that it enables spatio-temporal imaging of the photoemission spectral function that carries energy- and momentum-resolved information on the single-particle band structure, many-body interactions and correlation phenomena.</p>","PeriodicalId":18926,"journal":{"name":"Nature Photonics","volume":"11 1","pages":""},"PeriodicalIF":35.0,"publicationDate":"2025-01-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142917086","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"物理与天体物理","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-03DOI: 10.1038/s41566-024-01577-x
Artiom Skripka, Zhuolei Zhang, Xiao Qi, Benedikt Ursprung, Peter Ercius, Bruce E. Cohen, P. James Schuck, Daniel Jaque, Emory M. Chan
Optically bistable materials respond to a single input with two possible optical outputs, contingent on excitation history. Such materials would be ideal for optical switching and memory, but the limited understanding of intrinsic optical bistability (IOB) prevents the development of nanoscale IOB materials suitable for devices. Here we demonstrate IOB in Nd3+-doped KPb2Cl5 avalanching nanoparticles, which switch with high contrast between luminescent and non-luminescent states, with hysteresis characteristic of bistability. We elucidate a non-thermal mechanism in which IOB originates from suppressed non-radiative relaxation in Nd3+ ions and from the positive feedback of photon avalanching, resulting in extreme, >200th-order optical nonlinearities. The modulation of laser pulsing tunes the hysteresis widths, and dual-laser excitation enables transistor-like optical switching. This control over nanoscale IOB establishes avalanching nanoparticles for photonic devices in which light is used to manipulate light.
{"title":"Intrinsic optical bistability of photon avalanching nanocrystals","authors":"Artiom Skripka, Zhuolei Zhang, Xiao Qi, Benedikt Ursprung, Peter Ercius, Bruce E. Cohen, P. James Schuck, Daniel Jaque, Emory M. Chan","doi":"10.1038/s41566-024-01577-x","DOIUrl":"https://doi.org/10.1038/s41566-024-01577-x","url":null,"abstract":"<p>Optically bistable materials respond to a single input with two possible optical outputs, contingent on excitation history. Such materials would be ideal for optical switching and memory, but the limited understanding of intrinsic optical bistability (IOB) prevents the development of nanoscale IOB materials suitable for devices. Here we demonstrate IOB in Nd<sup>3+</sup>-doped KPb<sub>2</sub>Cl<sub>5</sub> avalanching nanoparticles, which switch with high contrast between luminescent and non-luminescent states, with hysteresis characteristic of bistability. We elucidate a non-thermal mechanism in which IOB originates from suppressed non-radiative relaxation in Nd<sup>3+</sup> ions and from the positive feedback of photon avalanching, resulting in extreme, >200th-order optical nonlinearities. The modulation of laser pulsing tunes the hysteresis widths, and dual-laser excitation enables transistor-like optical switching. This control over nanoscale IOB establishes avalanching nanoparticles for photonic devices in which light is used to manipulate light.</p>","PeriodicalId":18926,"journal":{"name":"Nature Photonics","volume":"27 1","pages":""},"PeriodicalIF":35.0,"publicationDate":"2025-01-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142917016","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"物理与天体物理","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-02DOI: 10.1038/s41566-024-01603-y
Kun Liao, Yaxiao Lian, Maotao Yu, Zhuochen Du, Tianxiang Dai, Yaxin Wang, Haoming Yan, Shufang Wang, Cuicui Lu, C. T. Chan, Rui Zhu, Dawei Di, Xiaoyong Hu, Qihuang Gong
Integrated photonic chips hold substantial potential in optical communications, computing, light detection and ranging, sensing, and imaging, offering exceptional data throughput and low power consumption. A key objective is to build a monolithic on-chip photonic system that integrates light sources, processors and photodetectors on a single chip. However, this remains challenging due to limitations in materials engineering, chip integration techniques and design methods. Perovskites offer simple fabrication, tolerance to lattice mismatch, flexible bandgap tunability and low cost, making them promising for hetero-integration with silicon photonics. Here we propose and experimentally realize a near-infrared monolithic on-chip photonic system based on a perovskite/silicon nitride photonic platform, developing nano-hetero-integration technology to integrate efficient light-emitting diodes, high-performance processors and sensitive photodetectors. Photonic neural networks are implemented to perform photonic simulations and computer vision tasks. Our network efficiently predicts the topological invariant in a two-dimensional disordered Su–Schrieffer–Heeger model and simulates nonlinear topological models with an average fidelity of 87%. In addition, we achieve a test accuracy of over 85% in edge detection and 56% on the CIFAR-10 dataset using a scaled-up architecture. This work addresses the challenge of integrating diverse nanophotonic components on a chip, offering a promising solution for chip-integrated multifunctional photonic information processing.
{"title":"Hetero-integrated perovskite/Si3N4 on-chip photonic system","authors":"Kun Liao, Yaxiao Lian, Maotao Yu, Zhuochen Du, Tianxiang Dai, Yaxin Wang, Haoming Yan, Shufang Wang, Cuicui Lu, C. T. Chan, Rui Zhu, Dawei Di, Xiaoyong Hu, Qihuang Gong","doi":"10.1038/s41566-024-01603-y","DOIUrl":"https://doi.org/10.1038/s41566-024-01603-y","url":null,"abstract":"<p>Integrated photonic chips hold substantial potential in optical communications, computing, light detection and ranging, sensing, and imaging, offering exceptional data throughput and low power consumption. A key objective is to build a monolithic on-chip photonic system that integrates light sources, processors and photodetectors on a single chip. However, this remains challenging due to limitations in materials engineering, chip integration techniques and design methods. Perovskites offer simple fabrication, tolerance to lattice mismatch, flexible bandgap tunability and low cost, making them promising for hetero-integration with silicon photonics. Here we propose and experimentally realize a near-infrared monolithic on-chip photonic system based on a perovskite/silicon nitride photonic platform, developing nano-hetero-integration technology to integrate efficient light-emitting diodes, high-performance processors and sensitive photodetectors. Photonic neural networks are implemented to perform photonic simulations and computer vision tasks. Our network efficiently predicts the topological invariant in a two-dimensional disordered Su–Schrieffer–Heeger model and simulates nonlinear topological models with an average fidelity of 87%. In addition, we achieve a test accuracy of over 85% in edge detection and 56% on the CIFAR-10 dataset using a scaled-up architecture. This work addresses the challenge of integrating diverse nanophotonic components on a chip, offering a promising solution for chip-integrated multifunctional photonic information processing.</p>","PeriodicalId":18926,"journal":{"name":"Nature Photonics","volume":"115 1","pages":""},"PeriodicalIF":35.0,"publicationDate":"2025-01-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142912075","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"物理与天体物理","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Leveraging the entire space of complex dielectric permittivity, non-Hermitian photonics has fundamentally altered wave propagation with complex optical potentials and has ushered in a host of new photonic applications. Through parity–time symmetry and its breaking—a delicate interplay between gain and loss—even the interaction between just two entities becomes counter-intuitive and intriguing. Here we realize, through hybrid III–V/Si integration, a scalable non-Hermitian switching network on a two-layer integrated photonic chip. Our platform is a hybrid, with a bottom silicon layer and a top InGaAsP layer that provides optical gain. By tuning the gain level in the top layer, vertically coupled waveguides operate below or above the exceptional point, where light is switched across two layers, among different input–output ports. For a single switching unit, the switching dynamics are ultrafast, on the order of 100 ps. In a large switching network, non-blocking and other diverse connectivities are established in single-wavelength and wavelength-selective switching, with high extinction ratios. Our approach adds scalable non-Hermitian switching to photonic design toolkits to simultaneously boost the switching time and bandwidth density to cutting-edge levels, therefore paving the way for compact and ultrafast monolithic integrated silicon photonics in next-generation optical information networks.
{"title":"Non-Hermitian hybrid silicon photonic switching","authors":"Xilin Feng, Tianwei Wu, Zihe Gao, Haoqi Zhao, Shuang Wu, Yichi Zhang, Li Ge, Liang Feng","doi":"10.1038/s41566-024-01579-9","DOIUrl":"https://doi.org/10.1038/s41566-024-01579-9","url":null,"abstract":"<p>Leveraging the entire space of complex dielectric permittivity, non-Hermitian photonics has fundamentally altered wave propagation with complex optical potentials and has ushered in a host of new photonic applications. Through parity–time symmetry and its breaking—a delicate interplay between gain and loss—even the interaction between just two entities becomes counter-intuitive and intriguing. Here we realize, through hybrid III–V/Si integration, a scalable non-Hermitian switching network on a two-layer integrated photonic chip. Our platform is a hybrid, with a bottom silicon layer and a top InGaAsP layer that provides optical gain. By tuning the gain level in the top layer, vertically coupled waveguides operate below or above the exceptional point, where light is switched across two layers, among different input–output ports. For a single switching unit, the switching dynamics are ultrafast, on the order of 100 ps. In a large switching network, non-blocking and other diverse connectivities are established in single-wavelength and wavelength-selective switching, with high extinction ratios. Our approach adds scalable non-Hermitian switching to photonic design toolkits to simultaneously boost the switching time and bandwidth density to cutting-edge levels, therefore paving the way for compact and ultrafast monolithic integrated silicon photonics in next-generation optical information networks.</p>","PeriodicalId":18926,"journal":{"name":"Nature Photonics","volume":"39 1","pages":""},"PeriodicalIF":35.0,"publicationDate":"2025-01-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142912070","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"物理与天体物理","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Magneto-optical (MO) effects have a pivotal role in modern photonic devices for light manipulation and sensing, but the study of these effects has so far been limited to the MO Faraday and Kerr effects. Conventional MO systems encounter considerable intrinsic losses, markedly hampering their ability to amplify the MO effects. Here we introduce a loss-enhanced MO effect to sublinearly amplify the frequency response of a non-Hermitian optical cavity under different background magnetic fields. This exceptional MO effect relies on an architecture of MO material embedded in a Fabry–Pérot cavity, accompanied by a polarization-dependent optical absorption, that is, linear dichroism, to construct a reconfigurable exceptional point. The experimental results show that two eigenmodes of the Fabry–Pérot cavity exhibit sublinear frequency splitting. By electrically reconfiguring the absorber, the eigenfrequency shift can be adaptively enhanced under different background magnetic fields. Using this effect, we demonstrate the detection of subtle magnetic field variations in a strong background, with the system’s response magnified by a factor exceeding 10 and sensitivity increased threefold compared with its conventional Hermitian counterpart. Our study leverages exceptional physics to study the MO effect and develops a new class of reconfigurable MO devices equipped with enhanced sensitivity for potential integration with photonic systems. The authors introduce a loss-enhanced magneto-optical effect and sublinearly amplify the frequency response of a non-Hermitian optical cavity under different background magnetic fields. This effect is exploited to detect subtle magnetic field variations against a strong background with enhanced system response and sensitivity.
{"title":"Observation of loss-enhanced magneto-optical effect","authors":"Ya-Ping Ruan, Jiang-Shan Tang, Zhipeng Li, Haodong Wu, Wenpeng Zhou, Longqi Xiao, Jianfeng Chen, Shi-Jun Ge, Wei Hu, Han Zhang, Cheng-Wei Qiu, Wuming Liu, Hui Jing, Yan-Qing Lu, Keyu Xia","doi":"10.1038/s41566-024-01592-y","DOIUrl":"10.1038/s41566-024-01592-y","url":null,"abstract":"Magneto-optical (MO) effects have a pivotal role in modern photonic devices for light manipulation and sensing, but the study of these effects has so far been limited to the MO Faraday and Kerr effects. Conventional MO systems encounter considerable intrinsic losses, markedly hampering their ability to amplify the MO effects. Here we introduce a loss-enhanced MO effect to sublinearly amplify the frequency response of a non-Hermitian optical cavity under different background magnetic fields. This exceptional MO effect relies on an architecture of MO material embedded in a Fabry–Pérot cavity, accompanied by a polarization-dependent optical absorption, that is, linear dichroism, to construct a reconfigurable exceptional point. The experimental results show that two eigenmodes of the Fabry–Pérot cavity exhibit sublinear frequency splitting. By electrically reconfiguring the absorber, the eigenfrequency shift can be adaptively enhanced under different background magnetic fields. Using this effect, we demonstrate the detection of subtle magnetic field variations in a strong background, with the system’s response magnified by a factor exceeding 10 and sensitivity increased threefold compared with its conventional Hermitian counterpart. Our study leverages exceptional physics to study the MO effect and develops a new class of reconfigurable MO devices equipped with enhanced sensitivity for potential integration with photonic systems. The authors introduce a loss-enhanced magneto-optical effect and sublinearly amplify the frequency response of a non-Hermitian optical cavity under different background magnetic fields. This effect is exploited to detect subtle magnetic field variations against a strong background with enhanced system response and sensitivity.","PeriodicalId":18926,"journal":{"name":"Nature Photonics","volume":"19 1","pages":"109-115"},"PeriodicalIF":32.3,"publicationDate":"2024-12-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142848982","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"物理与天体物理","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-12-13DOI: 10.1038/s41566-024-01586-w
Jiahui Xu, Rui Luo, Zichao Luo, Jun Xu, Zhen Mu, Hongyu Bian, Siew Yin Chan, Benjamin Yue Hao Tan, Dongzhi Chi, Zhongfu An, Guichuan Xing, Xian Qin, Changyang Gong, Yiming Wu, Xiaogang Liu
The secondary X-rays generated by the interaction of high-energy particles with scintillators can be converted into lower-energy excitons through thermalization, emitting light in the process. Capturing these secondary X-ray quanta efficiently is key to enhancing scintillation performance and boosting radiation detector sensitivity. Here we report a molecular design strategy using organic ligands to reclaim energy lost during the relaxation of secondary X-rays. This approach results in an enhancement in radioluminescence within lanthanide metal complexes by more than three orders of magnitude. By controlling the triplet energy of these ligands, we enable lanthanide centres to capture dark triplet excitons with near-unity extraction efficiency. These excitons arise from the absorption of secondary X-rays and transferred to the lanthanide centres through resonance energy transfer. This process delivers radioluminescence with orders of magnitude higher efficiency than existing organic or commercial inorganic scintillators. Tailoring metal centres and their coordination environments allows these organolanthanide scintillators to tune their spectra from ultraviolet to near-infrared, with lifetimes adjustable from tens of nanoseconds to hundreds of microseconds. These molecular scintillators enable high-resolution radiographic imaging and X-ray-mediated photodynamic therapy. Our findings not only unravel the link between scintillation performance and triplet exciton recycling but also lay the foundation for designing highly efficient organic scintillators that could revolutionize various fields. Researchers use organic ligands to reclaim energy lost during the relaxation of secondary X-rays generated by the interaction of high-energy particles with scintillators. Enhanced radioluminescence within lanthanide metal complexes and capture of dark triplet excitons with near-unity extraction efficiency are achieved.
高能粒子与闪烁体相互作用产生的二次 X 射线可通过热化转化为低能激子,并在此过程中发光。有效捕捉这些次级 X 射线量子是提高闪烁性能和辐射探测器灵敏度的关键。在此,我们报告了一种利用有机配体回收二次 X 射线弛豫过程中损失的能量的分子设计策略。这种方法使镧系元素金属复合物的放射发光增强了三个数量级以上。通过控制这些配体的三重态能量,我们使镧系元素中心能够以接近统一的萃取效率捕获暗三重态激子。这些激子产生于对二次 X 射线的吸收,并通过共振能量转移转移到镧系元素中心。与现有的有机或商用无机闪烁体相比,这一过程的辐射效率要高出几个数量级。对金属中心及其配位环境进行定制,可使这些有机镧系闪烁体的光谱范围从紫外到近红外,寿命可从几十纳秒到几百微秒不等。这些分子闪烁体可实现高分辨率放射成像和 X 射线介导的光动力疗法。我们的发现不仅揭示了闪烁性能与三重激子再循环之间的联系,还为设计高效有机闪烁体奠定了基础,从而为各个领域带来革命性的变化。
{"title":"Ultrabright molecular scintillators enabled by lanthanide-assisted near-unity triplet exciton recycling","authors":"Jiahui Xu, Rui Luo, Zichao Luo, Jun Xu, Zhen Mu, Hongyu Bian, Siew Yin Chan, Benjamin Yue Hao Tan, Dongzhi Chi, Zhongfu An, Guichuan Xing, Xian Qin, Changyang Gong, Yiming Wu, Xiaogang Liu","doi":"10.1038/s41566-024-01586-w","DOIUrl":"10.1038/s41566-024-01586-w","url":null,"abstract":"The secondary X-rays generated by the interaction of high-energy particles with scintillators can be converted into lower-energy excitons through thermalization, emitting light in the process. Capturing these secondary X-ray quanta efficiently is key to enhancing scintillation performance and boosting radiation detector sensitivity. Here we report a molecular design strategy using organic ligands to reclaim energy lost during the relaxation of secondary X-rays. This approach results in an enhancement in radioluminescence within lanthanide metal complexes by more than three orders of magnitude. By controlling the triplet energy of these ligands, we enable lanthanide centres to capture dark triplet excitons with near-unity extraction efficiency. These excitons arise from the absorption of secondary X-rays and transferred to the lanthanide centres through resonance energy transfer. This process delivers radioluminescence with orders of magnitude higher efficiency than existing organic or commercial inorganic scintillators. Tailoring metal centres and their coordination environments allows these organolanthanide scintillators to tune their spectra from ultraviolet to near-infrared, with lifetimes adjustable from tens of nanoseconds to hundreds of microseconds. These molecular scintillators enable high-resolution radiographic imaging and X-ray-mediated photodynamic therapy. Our findings not only unravel the link between scintillation performance and triplet exciton recycling but also lay the foundation for designing highly efficient organic scintillators that could revolutionize various fields. Researchers use organic ligands to reclaim energy lost during the relaxation of secondary X-rays generated by the interaction of high-energy particles with scintillators. Enhanced radioluminescence within lanthanide metal complexes and capture of dark triplet excitons with near-unity extraction efficiency are achieved.","PeriodicalId":18926,"journal":{"name":"Nature Photonics","volume":"19 1","pages":"71-78"},"PeriodicalIF":32.3,"publicationDate":"2024-12-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142815691","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"物理与天体物理","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-12-03DOI: 10.1038/s41566-024-01578-w
David A. B. Miller, Zeyu Kuang, Owen D. Miller
Applications of waves in communications, information processing and sensing need a clear understanding of how many strongly coupled channels or degrees of freedom exist in and out of volumes of space and how the coupling falls off for larger numbers. Numerical results are possible, and some heuristics exist, but there has been no simple physical picture and explanation for arbitrary volumes. By considering waves from a bounding spherical volume, we show a clear onset of a tunnelling escape of waves that both defines a limiting number of well-coupled channels for any volume and explains the subsequent rapid fall-off of coupling strengths. The approach works over all size scales, from nanophotonics and small radiofrequency antennas up to imaging optics. It gives a unified view from the multipole expansions common for antennas and small objects to the limiting plane and evanescent waves of large optics, showing that all such waves can escape to propagation to some degree, by tunnelling if necessary, and gives a precise diffraction limit.
{"title":"Tunnelling escape of waves","authors":"David A. B. Miller, Zeyu Kuang, Owen D. Miller","doi":"10.1038/s41566-024-01578-w","DOIUrl":"https://doi.org/10.1038/s41566-024-01578-w","url":null,"abstract":"<p>Applications of waves in communications, information processing and sensing need a clear understanding of how many strongly coupled channels or degrees of freedom exist in and out of volumes of space and how the coupling falls off for larger numbers. Numerical results are possible, and some heuristics exist, but there has been no simple physical picture and explanation for arbitrary volumes. By considering waves from a bounding spherical volume, we show a clear onset of a tunnelling escape of waves that both defines a limiting number of well-coupled channels for any volume and explains the subsequent rapid fall-off of coupling strengths. The approach works over all size scales, from nanophotonics and small radiofrequency antennas up to imaging optics. It gives a unified view from the multipole expansions common for antennas and small objects to the limiting plane and evanescent waves of large optics, showing that all such waves can escape to propagation to some degree, by tunnelling if necessary, and gives a precise diffraction limit.</p>","PeriodicalId":18926,"journal":{"name":"Nature Photonics","volume":"8 1","pages":""},"PeriodicalIF":35.0,"publicationDate":"2024-12-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142760353","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"物理与天体物理","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-12-02DOI: 10.1038/s41566-024-01567-z
Saumil Bandyopadhyay, Alexander Sludds, Stefan Krastanov, Ryan Hamerly, Nicholas Harris, Darius Bunandar, Matthew Streshinsky, Michael Hochberg, Dirk Englund
As deep neural networks revolutionize machine learning, energy consumption and throughput are emerging as fundamental limitations of complementary metal–oxide–semiconductor (CMOS) electronics. This has motivated a search for new hardware architectures optimized for artificial intelligence, such as electronic systolic arrays, memristor crossbar arrays and optical accelerators. Optical systems can perform linear matrix operations at an exceptionally high rate and efficiency, motivating recent demonstrations of low-latency matrix accelerators and optoelectronic image classifiers. However, demonstrating coherent, ultralow-latency optical processing of deep neural networks has remained an outstanding challenge. Here we realize such a system in a scalable photonic integrated circuit that monolithically integrates multiple coherent optical processor units for matrix algebra and nonlinear activation functions into a single chip. We experimentally demonstrate this fully integrated coherent optical neural network architecture for a deep neural network with six neurons and three layers that optically computes both linear and nonlinear functions with a latency of 410 ps, unlocking new applications that require ultrafast, direct processing of optical signals. We implement backpropagation-free in situ training on this system, achieving 92.5% accuracy on a six-class vowel classification task, which is comparable to the accuracy obtained on a digital computer. This work lends experimental evidence to theoretical proposals for in situ training, enabling orders of magnitude improvements in the throughput of training data. Moreover, the fully integrated coherent optical neural network opens the path to inference at nanosecond latency and femtojoule per operation energy efficiency. Researchers experimentally demonstrate a fully integrated coherent optical neural network. The system, with six neurons and three layers, operates with a latency of 410 ps.
{"title":"Single-chip photonic deep neural network with forward-only training","authors":"Saumil Bandyopadhyay, Alexander Sludds, Stefan Krastanov, Ryan Hamerly, Nicholas Harris, Darius Bunandar, Matthew Streshinsky, Michael Hochberg, Dirk Englund","doi":"10.1038/s41566-024-01567-z","DOIUrl":"10.1038/s41566-024-01567-z","url":null,"abstract":"As deep neural networks revolutionize machine learning, energy consumption and throughput are emerging as fundamental limitations of complementary metal–oxide–semiconductor (CMOS) electronics. This has motivated a search for new hardware architectures optimized for artificial intelligence, such as electronic systolic arrays, memristor crossbar arrays and optical accelerators. Optical systems can perform linear matrix operations at an exceptionally high rate and efficiency, motivating recent demonstrations of low-latency matrix accelerators and optoelectronic image classifiers. However, demonstrating coherent, ultralow-latency optical processing of deep neural networks has remained an outstanding challenge. Here we realize such a system in a scalable photonic integrated circuit that monolithically integrates multiple coherent optical processor units for matrix algebra and nonlinear activation functions into a single chip. We experimentally demonstrate this fully integrated coherent optical neural network architecture for a deep neural network with six neurons and three layers that optically computes both linear and nonlinear functions with a latency of 410 ps, unlocking new applications that require ultrafast, direct processing of optical signals. We implement backpropagation-free in situ training on this system, achieving 92.5% accuracy on a six-class vowel classification task, which is comparable to the accuracy obtained on a digital computer. This work lends experimental evidence to theoretical proposals for in situ training, enabling orders of magnitude improvements in the throughput of training data. Moreover, the fully integrated coherent optical neural network opens the path to inference at nanosecond latency and femtojoule per operation energy efficiency. Researchers experimentally demonstrate a fully integrated coherent optical neural network. The system, with six neurons and three layers, operates with a latency of 410 ps.","PeriodicalId":18926,"journal":{"name":"Nature Photonics","volume":"18 12","pages":"1335-1343"},"PeriodicalIF":32.3,"publicationDate":"2024-12-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142760380","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"物理与天体物理","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-12-02DOI: 10.1038/s41566-024-01583-z
Thomas Schneider
Oscillators for tunable terahertz waves with ultra-high spectral purity may pave the way for precise molecular clocks and extremely high-data-rate wireless communications.
{"title":"Ultra-low-noise terahertz sources","authors":"Thomas Schneider","doi":"10.1038/s41566-024-01583-z","DOIUrl":"10.1038/s41566-024-01583-z","url":null,"abstract":"Oscillators for tunable terahertz waves with ultra-high spectral purity may pave the way for precise molecular clocks and extremely high-data-rate wireless communications.","PeriodicalId":18926,"journal":{"name":"Nature Photonics","volume":"18 12","pages":"1230-1231"},"PeriodicalIF":32.3,"publicationDate":"2024-12-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142760357","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"物理与天体物理","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-12-02DOI: 10.1038/s41566-024-01582-0
David Pile
Light-based processing and machine learning featured heavily in San Diego at the 2024 SPIE Optics + Photonics conference. Enthusiasm was coupled with questions related to the real-world applicability and the merits of linear vs non-linear, and all-optical vs hybrid, approaches.
{"title":"Optical computing and artificial intelligence","authors":"David Pile","doi":"10.1038/s41566-024-01582-0","DOIUrl":"10.1038/s41566-024-01582-0","url":null,"abstract":"Light-based processing and machine learning featured heavily in San Diego at the 2024 SPIE Optics + Photonics conference. Enthusiasm was coupled with questions related to the real-world applicability and the merits of linear vs non-linear, and all-optical vs hybrid, approaches.","PeriodicalId":18926,"journal":{"name":"Nature Photonics","volume":"18 12","pages":"1240-1242"},"PeriodicalIF":32.3,"publicationDate":"2024-12-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142760456","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"物理与天体物理","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}