We describe how several optimization problems can be rapidly solved by highly interconnected networks of simple analog processors. Analog-to-digital (A/D) conversion was considered as a simple optimization problem, and an A/D converter of novel architecture was designed. A/D conversion is a simple example of a more general class of signal-decision problems which we show could also be solved by appropriately constructed networks. Circuits to solve these problems were designed using general principles which result from an understanding of the basic collective computational properties of a specific class of analog-processor networks. We also show that a network which solves linear programming problems can be understood from the same concepts.

Simple 'neural' optimization networks: An A/D converter, signal decision circuit, and a linear programming circuit

A new conceptual framework and a minimization principle together provide an understanding of computation in model neural circuits. The circuits consist of nonlinear graded-response model neurons organized into networks with effectively symmetric synaptic connections. The neurons represent an approximation to biological neurons in which a simplified set of important computational properties is retained. Complex circuits solving problems similar to those essential in biology can be analyzed and understood without the need to follow the circuit dynamics in detail. Implementation of the model with electronic devices will provide a class of electronic circuits of novel form and function.

https://science.sciencemag.org/content/sci/233/4764/625.full.pdf

Computing with neural circuits: a model

Anti-reflective coatings (ARCs) have evolved into highly effective reflectance and glare reducing components for various optical and opto-electrical equipments. Extensive research in optical and biological reflectance minimization as well as the emergence of nanotechnology over the years has contributed to the enhancement of ARCs in a major way. In this study the prime objective is to give a comprehensive idea of the ARCs right from their inception, as they were originally conceptualized by the pioneers and lay down the basic concepts and strategies adopted to minimize reflectance. The different types of ARCs are also described in greater detail and the state-of-the-art fabrication techniques have been fully illustrated. The inspiration that ARCs derive from nature (‘biomimetics’) has been an area of major research and is discussed at length. The various materials that have been reportedly used in fabricating the ARCs have also been brought into sharp focus. An account of application of ARCs on solar cells and modules, contemporary research and associated challenges are presented in the end to facilitate a universal understanding of the ARCs and encourage future research.

Anti-reflective coatings: A critical, in-depth review

This review presents an overview of the thermal properties of mesoscopic structures. The discussion is based on the concept of electron energy distribution, and, in particular, on controlling and probing it. The temperature of an electron gas is determined by this distribution: refrigeration is equivalent to narrowing it, and thermometry is probing its convolution with a function characterizing the measuring device. Temperature exists, strictly speaking, only in quasiequilibrium in which the distribution follows the Fermi-Dirac form. Interesting nonequilibrium deviations can occur due to slow relaxation rates of the electrons, e.g., among themselves or with lattice phonons. Observation and applications of nonequilibrium phenomena are also discussed. The focus in this paper is at low temperatures, primarily below $4\phantom{\rule{0.3em}{0ex}}\mathrm{K}$, where physical phenomena on mesoscopic scales and hybrid combinations of various types of materials, e.g., superconductors, normal metals, insulators, and doped semiconductors, open up a rich variety of device concepts. This review starts with an introduction to theoretical concepts and experimental results on thermal properties of mesoscopic structures. Then thermometry and refrigeration are examined with an emphasis on experiments. An immediate application of solid-state refrigeration and thermometry is in ultrasensitive radiation detection, which is discussed in depth. This review concludes with a summary of pertinent fabrication methods of presented devices.

/pdf/opportunities-for-mesoscopics-in-thermometry-and-10z63yzsh2.pdf

Opportunities for mesoscopics in thermometry and refrigeration: Physics and applications

I. Introduction (Preface, Nanostructures in Si Inversion Layers, Nanostructures in GaAs-AlGaAs Heterostructures, Basic Properties). 
II. Diffusive and Quasi-Ballistic Transport (Classical Size Effects, Weak Localization, Conductance Fluctuations, Aharonov-Bohm Effect, Electron-Electron Interactions, Quantum Size Effects, Periodic Potential). 
III. Ballistic Transport (Conduction as a Transmission Problem, Quantum Point Contacts, Coherent Electron Focusing, Collimation, Junction Scattering, Tunneling). 
IV. Adiabatic Transport (Edge Channels and the Quantum Hall Effect, Selective Population and Detection of Edge Channels, Fractional Quantum Hall Effect, Aharonov-Bohm Effect in Strong Magnetic Fields, Magnetically Induced Band Structure).

/pdf/quantum-transport-in-semiconductor-nanostructures-1n3dkljf6x.pdf

Quantum Transport in Semiconductor Nanostructures

Recent proposals for neural network models indicate that an array of amplifiers coupled to a lattice of wires with resistive components at the crosspoints can perform calculations using collective properties similar to those observed in biological systems. Such a network can perform both memory and processing functions. The promise of the connection matrix processor lies in its very high density, fault tolerance, and massively parallel operation. This paper describes the operation of a neural network and exploratory fabrication techniques for its implementation.

Artificial neural networks for computing

Textured surfaces consisting of submicron columns or cones have been formed in a variety of surfaces using reactive ion etching. On metals, insulators, and semiconductors this texturing forms a graded index matching layer which reduces reflection of visible light to as low as 0.1%. The optical properties can be permanently altered on a ∠l μm size scale by irradiation with a low power laser beam and this has been explored as a new optical storage medium. The light energy transforms the optically absorbing textured surface to a nonabsorbing smooth one by melting. Reflectance changes of greater than a factor of 100 have been demonstrated at ∠10 mW light power. The textured surfaces have also been used with dielectric materials to create high transparency optical elements. Highly absorbing films of Ge and Si, that were earlier studied as solar photothermal absorbers, have been investigated as optical storage media. The production, morphology and optical properties of these surfaces are described.

Textured surfaces: Optical storage and other applications

We have constructed electronic neural networks with 22×22 arrays of microfabricated resistive synapses. The purpose of this work is both to develop a VLSI‐compatible process for making large networks, and to extend our understanding of electronic neural network dynamics.

Electronic neural networks

We have performed electron beam lithography studies on thick substrates using beam energies of 20–120 keV and a nominal beam diameter of 2 nm in a Philips 400 electron microscope with scanning capability. Metal lines as narrow as 10 nm were fabricated on Si and GaAs substrates using liftoff of a single thin layer of resist. High resolution (approximately 10 nm) patterns could be written at all beam energies with an exposure latitude that remained approximately constant up to energies for which the range of the backscattered electrons became significantly larger than the pattern area. For large area patterns written with the small beam, the proximity effect is greatly reduced, even at 20 keV, because of the sharp edge of the exposure profile. At high beam energies, the range of backscattered electrons is large enough that they contribute only a slowly varying background dose, leading to a relatively simple proximity correction even for complex patterns.

ELECTRON BEAM LITHOGRAPHY FROM 20 TO 120 kev WITH A HIGH QUALITY BEAM.

We describe a trilevel e‐beam resist process based on the high resolution resist PMMA. This process has been used to form metal features 25 nm wide on thick Si substrates by liftoff. Ni–Cr films patterned in this way have been used as masks for reactive ion etching of Si. The resulting structures were arrays of fins ∠ 30 nm wide and 330 nm high on 95 nm centers with a total length of 3 μm. We also discuss the resolution improvement obtained in a trilevel process based on a higher speed resist, P (MMA/MAA).

R. E. Howard

Papers

Artificial neural networks for computing

Textured surfaces: Optical storage and other applications

Electronic neural networks

ELECTRON BEAM LITHOGRAPHY FROM 20 TO 120 kev WITH A HIGH QUALITY BEAM.

Twenty-five nm features patterned with trilevel e-beam resist.