Simulated Annealing: Rigorous finite-time guarantees for optimization on continuous domains

Simulated annealing is a popular method for approaching the solution of a global optimization problem. Existing results on its performance apply to discrete combinatorial optimization where the optimization variables can assume only a finite set of possible values. We introduce a new general formulation of simulated annealing which allows one to guarantee finite-time performance in the optimization of functions of continuous variables. The results hold universally for any optimization problem on a bounded domain and establish a connection between simulated annealing and up-to-date theory of convergence of Markov chain Monte Carlo methods on continuous domains. This work is inspired by the concept of finite-time learning with known accuracy and confidence developed in statistical learning theory.

TRANSIENT SCANNING ELECTRON BEAM ANNEALING METHODS USED TO STUDY DIFFUSION AND DEFECTS IN IMPLANTED SILICON.

Rapid thermal annealing of arsenic and boron difluoride implants, such as those used for source/drain regions in CMOS, has been carried out using a scanning electron beam annealer, as part of a study of transient diffusion effects. Three types of e-beam anneal have been performed, with peak temperatures in the range 900 -1200 degree C; the normal isothermal e-beam anneals, together with sub-second fast anneals and 'dual-pulse' anneals, in which the sample undergoes an isothermal pre-anneal followed by rapid heating to the required anneal temperature is less than 0. 5s. The diffusion occuring during these anneal cycles has been modelled using SPS-1D, an implant and diffusion modelling program developed by one of the authors. This has been modified to incorporate simulated temperature vs. time cycles for the anneals. Results are presented applying the usual equilibrium clustering model, a transient point-defect enhancement to the diffusivity proposed recently by Fair and a new dynamic clustering model for arsenic. Good agreement with SIMS measurements is obtained using the dynamic clustering model, without recourse to a transient defect model.

Homogeneous pinhole free 1 nm Al2O3 tunnel barriers on graphene

We report on the topographical and electrical characterisations of 1 nm thick Al2O3 dielectric films on graphene. The Al 2O3 is grown by sputtering a 0.6 nm Al layer on graphene and subsequentially oxidizing it in an O2 atmosphere. The Al 2O3 layer presents no pinholes and is homogeneous enough to act as a tunnel barrier. A resistance-area product in the mega-ohm micrometer-square range is found. Comparatively, the growth of Al 2O3 by evaporation does not lead to well-wetted films on graphene. Application of this high quality sputtered tunnel barrier to efficient spin injection in graphene is discussed. © 2012 American Institute of Physics.

Neutronic optimization in high conversion Th-233U fuel assembly with simulated annealing

This paper reports on fuel design optimization of a PWR operating in a self sustainable Th-233U fuel cycle. Monte Carlo simulated annealing method was used in order to identify the fuel assembly configuration with the most attractive breeding performance. In previous studies, it was shown that breeding may be achieved by employing heterogeneous Seed-Blanket fuel geometry. The arrangement of seed and blanket pins within the assemblies may be determined by varying the designed parameters based on basic reactor physics phenomena which affect breeding. However, the amount of free parameters may still prove to be prohibitively large in order to systematically explore the design space for optimal solution. Therefore, the Monte Carlo annealing algorithm for neutronic optimization is applied in order to identify the most favorable design. The objective of simulated annealing optimization is to find a set of design parameters, which maximizes some given performance function (such as relative period of net breeding) under specified constraints (such as fuel cycle length). The first objective of the study was to demonstrate that the simulated annealing optimization algorithm will lead to the same fuel pins arrangement as was obtained in the previous studies which used only basic physics phenomena as guidance for optimization. In the second part of this work, the simulated annealing method was used to optimize fuel pins arrangement in much larger fuel assembly, where the basic physics intuition does not yield clearly optimal configuration. The simulated annealing method was found to be very efficient in selecting the optimal design in both cases. In the future, this method will be used for optimization of fuel assembly design with larger number of free parameters in order to determine the most favorable trade-off between the breeding performance and core average power density.

Changes in the Mg profile and in dislocations induced by high temperature annealing of blue LEDs

The efficiency of the injection and recombination processes in InGaN/GaN LEDs is governed by the properties of the active region of the devices, which strongly depend on the conditions used for the growth of the epitaxial material. To improve device quality, it is very important to understand how the high temperatures used during the growth process can modify the quality of the epitaxial material. With this paper we present a study of the modifications in the properties of InGaN/GaN LED structures induced by high temperature annealing: thermal stress tests were carried out at 900 °C, in nitrogen atmosphere, on selected samples. The efficiency and the recombination dynamics were evaluated by photoluminescence measurements (both integrated and time-resolved), while the properties of the epitaxial material were studied by Secondary Ion Mass Spectroscopy (SIMS) and Rutherford Backscattering (RBS) channeling measurements. Results indicate that exposure to high temperatures may lead to: (i) a significant increase in the photoluminescence efficiency of the devices; (ii) a decrease in the parasitic emission bands located between 380 nm and 400 nm; (iii) an increase in carrier lifetime, as detected by time-resolved photoluminescence measurements. The increase in device efficiency is tentatively ascribed to an improvement in the crystallographic quality of the samples. © 2013 SPIE.

Effect of high temperature post-annealing on sidewalls of GaAs NWs grown by MOCVD

The sidewall facets of GaAs nanowires (NWs) were studied. It has been found that the sidewalls of GaAs NWs grown at 450 °C are {112} facets. However, the sidewalls of GaAs NWs start to become {110} during the postannealing at 650 °C for 30 min. © 2010 IEEE.

Dry oxidation and vacuum annealing treatments for tuning the wetting properties of carbon nanotube arrays.

In this article, we describe a simple method to reversibly tune the wetting properties of vertically aligned carbon nanotube (CNT) arrays. Here, CNT arrays are defined as densely packed multi-walled carbon nanotubes oriented perpendicular to the growth substrate as a result of a growth process by the standard thermal chemical vapor deposition (CVD) technique.(1,2) These CNT arrays are then exposed to vacuum annealing treatment to make them more hydrophobic or to dry oxidation treatment to render them more hydrophilic. The hydrophobic CNT arrays can be turned hydrophilic by exposing them to dry oxidation treatment, while the hydrophilic CNT arrays can be turned hydrophobic by exposing them to vacuum annealing treatment. Using a combination of both treatments, CNT arrays can be repeatedly switched between hydrophilic and hydrophobic.(2) Therefore, such combination show a very high potential in many industrial and consumer applications, including drug delivery system and high power density supercapacitors.(3-5) The key to vary the wettability of CNT arrays is to control the surface concentration of oxygen adsorbates. Basically oxygen adsorbates can be introduced by exposing the CNT arrays to any oxidation treatment. Here we use dry oxidation treatments, such as oxygen plasma and UV/ozone, to functionalize the surface of CNT with oxygenated functional groups. These oxygenated functional groups allow hydrogen bond between the surface of CNT and water molecules to form, rendering the CNT hydrophilic. To turn them hydrophobic, adsorbed oxygen must be removed from the surface of CNT. Here we employ vacuum annealing treatment to induce oxygen desorption process. CNT arrays with extremely low surface concentration of oxygen adsorbates exhibit a superhydrophobic behavior.