Reservoir computing is a computational framework suited for temporal/sequential data processing. It is derived from several recurrent neural network models, including echo state networks and liquid state machines. A reservoir computing system consists of a reservoir for mapping inputs into a high-dimensional space and a readout for pattern analysis from the high-dimensional states in the reservoir. The reservoir is fixed and only the readout is trained with a simple method such as linear regression and classification. Thus, the major advantage of reservoir computing compared to other recurrent neural networks is fast learning, resulting in low training cost. Another advantage is that the reservoir without adaptive updating is amenable to hardware implementation using a variety of physical systems, substrates, and devices. In fact, such physical reservoir computing has attracted increasing attention in diverse fields of research. The purpose of this review is to provide an overview of recent advances in physical reservoir computing by classifying them according to the type of the reservoir. We discuss the current issues and perspectives related to physical reservoir computing, in order to further expand its practical applications and develop next-generation machine learning systems.

We have first observed the nucleation and growth process of carbon nanotubes (CNTs) from iron carbide (Fe 3C) nanoparticles in chemical vapor deposition with C 2H 2 by in situ environmental transmission electron microscopy. Graphitic networks are formed on the fluctuating iron carbide nanoparticles, and subsequently CNTs are expelled from them. Our atomic scale observations suggest that carbon atoms diffuse through the bulk of iron carbide nanoparticles during the growth of CNTs.

Understanding how molecules can restructure the surfaces of heterogeneous catalysts under reaction conditions requires methods that can visualize atoms in real space and time. We applied a newly developed aberration-corrected environmental transmission electron microscopy to show that adsorbed carbon monoxide (CO) molecules caused the {100} facets of a gold nanoparticle to reconstruct during CO oxidation at room temperature. The CO molecules adsorbed at the on-top sites of gold atoms in the reconstructed surface, and the energetic favorability of this reconstructed structure was confirmed by ab initio calculations and image simulations. This atomic-scale visualizing method can be applied to help elucidate reaction mechanisms in heterogeneous catalysis.

This review article discusses the current and future possibilities for the application of in situ transmission electron microscopy to reveal synthesis pathways and functional mechanisms in complex and nanoscale materials. The findings of a group of scientists, representing academia, government labs and private sector entities (predominantly commercial vendors) during a workshop, held at the Center for Nanoscale Science and Technology- National Institute of Science and Technology (CNST-NIST), are discussed. We provide a comprehensive review of the scientific needs and future instrument and technique developments required to meet them.

on the occasion of its 100th anniversary Gold, the most stable metallic element, shows remarkable catalytic activity for CO oxidation even at room temperature.[1] Unlike platinum and palladium, [2] gold must be supported in the form of nanoparticles on crystalline metal oxides such as TiO 2[1] and CeO 2 .[3] Despite extensive studies, [4][5][6][7][8][9][10][11][12][13] the mechanism of catalysis by gold nanoparticles (GNPs) is still unclear, in particular in relation to CO oxidation at room temperature. In the present study we observed a real Au/CeO 2 catalyst in CO/air mixtures by means of in situ environmental transmission electron microscopy (ETEM). [14][15][16][17][18][19][20][21][22][23][24] The catalyst was also characterized by catalytic chemical analyses. In real GNP catalysts, the structures of the GNPs are not identical at the atomic scale. Hence, we examined a large number of GNPs in the Au/CeO 2 catalyst using ETEM, and found that the majority of the GNPs behaved systematically, depending on the partial pressures of CO and O 2 at room temperature. GNPs remained faceted during CO oxidation in CO/air and became rounded, or fluctuating multifaceted with decrease of the partial pressure of CO relative to air. We also examined GNPs supported on a non-oxide crystal (TiC) with ETEM. In contrast to GNPs supported on CeO 2 , switching the gases did not induce any morphology change of GNPs supported on TiC. These experimental results have provided a clue toward elucidation of the peculiar catalytic mechanism of supported GNPs. The interface between GNPs and CeO 2 support most likely plays an important role in the catalytic activity, especially the dissociation of O 2 molecules at room temperature. This work thus contributes to improving and developing real catalysts.The Au/CeO 2 catalyst was prepared by the deposition precipitation method.[1] The conversion of CO to CO 2 reached 100 % at room temperature, and the turnover frequency (TOF) of the catalyst was measured as 0.24 mol CO (mol Ausur )À1 s À1 at 303 K. The catalyst sample was examined in vacuum by conventional transmission electron microscopy before and after the oxidation of CO at atmospheric pressure and at 303 K for 5 h. As shown in Figure S1, it was confirmed that the average size and morphology of the GNPs remained unchanged after the oxidation of CO at atmospheric pressure. A detailed description of the catalyst is given in the Supporting Information.First, we summarize the typical morphology of a GNP supported on CeO 2 in various environments at room temperature. During CO oxidation in 1 vol % CO/air gas mixture (1 vol % CO, 21 vol % O 2 , 78 vol % N 2 ) at 1 mbar pressure, the GNP appeared to be faceted in the form of a stable polyhedron enclosed by the major {111} and {100} facets, as shown by Figure 1 a. Unexpectedly, the GNP behaved differently, and became rounded in pure O 2 gas. The GNP exhibited major facets in both inactive N 2 gas at 1 mbar and in vacuum (Figure 1 a). In N 2 gas, N 2 molecules collided with the surface of the GNP at a ra...

Efficient use of precious metal atoms in heterogeneous catalysis is important in chemical transformation and environmental remediation. Co3O4 with singly dispersed Rh atoms, Rh1/Co3O4, was synthesized for reduction of nitric oxide with hydrogen. Studies using extended X-ray absorption fine structure (EXAFS) showed that the singly dispersed Rh atoms are bonded to surface oxygen atoms before catalysis. In situ studies using ambient pressure X-ray photoelectron spectroscopy (AP-XPS), EXAFS, and X-ray Absorption Near Edge Structure (XANES) suggested that the surface of Rh1/Co3O4 with singly dispersed Rh atoms is restructured into a new geometry at 220 °C in the mixture of reactant gases (NO and H2). It forms RhCo n nanoclusters singly dispersed in the surface layer of Co3O4. The restructured catalyst, RhCo n /Co3O4 exhibits a much better catalytic performance in contrast to Rh1/Co3O4 without a restructuring. RhCo n /Co3O4 is highly active for reduction of nitric oxide with hydrogen. Selectivity to the production of N2 at 220 °C is 87% and reaches 97% at 300 °C. In situ studies showed this catalyst maintains its single dispersion of Rh atoms up to 300 °C during catalysis.

Despite the fragility of TiO(2) under electron irradiation, the intrinsic structure of Au/TiO(2) catalysts can be observed by environmental transmission electron microscopy. Under reaction conditions (CO/air 100 Pa), the major {111} and {100} facets of the gold nanoparticles are exposed and the particles display a polygonal interface with the TiO(2) support bounded by sharp edges parallel to the 〈110〉 directions.

We have measured the growth rate of silicon nanowires (SiNWs), which were grown at temperatures between 365 and 495 °C via the vapor-liquid-solid (VLS) mechanism. We grew SiNWs using gold as catalysts and monosilane (SiH4) as a vapor phase reactant. Observing SiNWs by means of transmission electron microscopy, we have found that SiNWs with smaller diameters grow slower than those with larger ones, and the critical diameter at which growth stops completely exists. We have estimated the critical diameter of SiNWs to be about 2 nm. We have also measured the temperature dependence of the growth rate of SiNWs and estimated the activation energy of the growth of SiNWs to be 230kJ∕mol.