Roadmap on emerging hardware and technology for machine learning.

Recent progress in artificial intelligence is largely attributed to the rapid development of machine learning, especially in the algorithm and neural network models. However, it is the performance of the hardware, in particular the energy efficiency of a computing system that sets the fundamental limit of the capability of machine learning. Data-centric computing requires a revolution in hardware systems, since traditional digital computers based on transistors and the von Neumann architecture were not purposely designed for neuromorphic computing. A hardware platform based on emerging devices and new architecture is the hope for future computing with dramatically improved throughput and energy efficiency. Building such a system, nevertheless, faces a number of challenges, ranging from materials selection, device optimization, circuit fabrication and system integration, to name a few. The aim of this Roadmap is to present a snapshot of emerging hardware technologies that are potentially beneficial for machine learning, providing the Nanotechnology readers with a perspective of challenges and opportunities in this burgeoning field.

High-Resolution Imaging and Spectroscopy at High Pressure: A Novel Liquid Cell for the Transmission Electron Microscope.

We demonstrate quantitative core-loss electron energy-loss spectroscopy of iron oxide nanoparticles and imaging resolution of Ag nanoparticles in liquid down to 0.24 nm, in both transmission and scanning transmission modes, in a novel, monolithic liquid cell developed for the transmission electron microscope (TEM). At typical SiN membrane thicknesses of 50 nm the liquid-layer thickness has a maximum change of only 30 nm for the entire TEM viewing area of 200×200 µm.

Optical Fluorescence Microscopy for Spatially Characterizing Electron Transfer across a Solid-Liquid Interface on Heterogeneous Electrodes.

Heterogeneous catalytic materials and electrodes are used for (electro)chemical transformations, including those important for energy storage and utilization.1, 2 Due to the heterogeneous nature of these materials, activity measurements with sufficient spatial resolution are needed to obtain structure/activity correlations across the different surface features (exposed facets, step edges, lattice defects, grain boundaries, etc.). These measurements will help lead to an understanding of the underlying reaction mechanisms and enable engineering of more active materials. Because (electro)catalytic surfaces restructure with changing environments,1 it is important to perform measurements in operando. Sub-diffraction fluorescence microscopy is well suited for these requirements because it can operate in solution with resolution down to a few nm. We have applied sub-diffraction fluorescence microscopy to a thin cell containing an electrocatalyst and a solution containing the redox sensitive dye p-aminophenyl fluorescein to characterize reaction at the solid-liquid interface. Our chosen dye switches between a nonfluorescent reduced state and a one-electron oxidized bright state, a process that occurs at the electrode surface. This scheme is used to investigate the activity differences on the surface of polycrystalline Pt, in particular to differentiate reactivity at grain faces and grain boundaries. Ultimately, this method will be extended to study other dye systems and electrode materials.