Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “deep”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 1,225 records · Page 68

Deep level transient spectroscopy in plasma-assisted molecular beam epitaxy grown Al{sub 0.2}Ga{sub 0.8}N/GaN interface and the rapid thermal annealing effect

We investigated deep-level traps formed in Al{sub 0.2}Ga{sub 0.8}N/GaN heterostructures grown using plasma-assisted molecular beam epitaxy and by performing deep level transient spectroscopy (DLTS). Two electron traps with activation energies of E{sub c}-150 meV and E{sub c}-250 meV were observed, and their capture cross-sections ({sigma}{sub T}) were estimated to be 2.0x10{sup -18} cm{sup 2} and 1.1x10{sup -17} cm{sup 2}, respectively. Different behaviors in the dependence of DLTS on filling pulse length confirm that the traps originated from N vacancies and dislocations. The amplitude of the dislocation-induced DLTS signal was reduced significantly by high-temperature rapid thermal annealing under N{sub 2} ambient after hydrogen treatment due to the reduction in dislocation density.

36 MATERIALS SCIENCE↗

Analysis of leakage current mechanisms in Pt/Au Schottky contact on Ga-polarity GaN by Frenkel-Poole emission and deep level studies

We report the Frenkel-Poole emission in Pt/Au Schottky contact on Ga-polarity GaN grown by molecular beam epitaxy using current-voltage-temperature (I-V-T) characteristics in the temperature ranging from 200 K to 375 K. Using thermionic emission model, the estimated Schottky barrier height is 0.49 eV at 200 K and 0.83 eV at 375 K, respectively, and it is observed that the barrier height increases with increase in temperature. The extracted emission barrier height ({phi}{sub t}) for Ga-polarity GaN Schottky diode by Frenkel-Poole theory is about 0.15 eV. Deep level transient spectroscopy study shows a deep level with activation energy of 0.44 eV, having capture cross-section 6.09 x 10{sup -14} cm{sup 2}, which is located between the metal and semiconductor interface, and trap nature is most probably associated with dislocations in Ga-polarity GaN. The analysis of I-V-T characteristics represents that the leakage current is due to effects of electrical field and temperature on the emission of electron from a trap state near the metal-semiconductor interface into continuum states associated with conductive dislocations in Ga-polarity GaN Schottky diode.

36 MATERIALS SCIENCE↗

Deep Inelastic Electron Scattering Off the Helium and Tritium Mirror Nuclei

We discuss a possible measurement of the ratio of nucleon structure functions, F{sub 2}{sup n}/F{sub 2}{sup p}, and the ratio of up to down quark distributions, u/d, at large Bjorken x, by performing deep inelastic electron scattering from the {sup 3}H and {sup 3}He mirror nuclei with the 11 GeV upgraded beam of Jefferson Lab. The measurement is expected to be almost free of nuclear effects, which introduce a significant uncertainty in the extraction of these two ratios from deep inelastic scattering off the proton and deuteron. The results are expected to test perturbative and non-perturbative mechanisms of spin-flavor symmetry breaking in the nucleon, and constrain the structure function parametrizations needed for the interpretation of high energy collider and neutrino oscillations data. The precision of the expected data can also test models of the nuclear EMC effect and provide valuable input for its full explanation.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Two Photon Exchange in Quasi-elastic and Deep-inelastic Scattering

In this paper, I present an overview and preliminary results from three experiments at Jefferson Lab that were recently completed using a {sup 3}He gas target with polarization oriented normal to the scattering plane of unpolarized incident electrons. A target single spin asymmetry was formed by periodically flipping the direction of the target spin. In the reaction {up_arrow}{sup 3}He(e,e'), the Born contribution is expected to be zero, giving direct sensitivity to two photon exchange. This asymmetry was measured in the quasi-elastic and deep-inelastic regimes with 0.1 < Q{sup 2} < 1.0 GeV{sup 2}. The asymmetry is predicted to decrease by two-orders of magnitude for deep-inelastic versus quasi-elastic scattering. Preliminary results from these experiments will be presented.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Temperature Dependent Capacitance-Voltage And Deep Level Transient Spectroscopy Study Of Self-Assembled Ge Quantum Dots Embedded In P-type Silicon

Temperature dependent Capacitance-Voltage (TCV) and Deep Level Transient Spectroscopy (DLTS) techniques were used to study how Ge Quantum Dots (QDs) embedded in Silicon trap charge. Atomic Force Microscopy (AFM) is used to obtain the density of QDs, which is in the order of 3x10{sup 11} cm{sup -2}. Three shallow levels, with activation energies of 40, 65 and 90 meV, and densities around 10{sup 16} cm{sup -3}, are found and are related to Boron. Four deep levels, with activation energies of 110, 150, 330 and 380 meV, and densities between 2x10{sup 15} cm{sup -3} and 5x10{sup 15} cm{sup -3}, are also found. TCV results suggest they are related to the Ge QDs.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

Deep level defects in n-type GaAsBi and GaAs grown at low temperatures

Deep level defects in n-type GaAs{sub 1-x}Bi{sub x} having 0 < x < 0.012 and GaAs grown by molecular beam epitaxy (MBE) at substrate temperatures between 300 and 400 Degree-Sign C have been investigated by Deep Level Capacitance Spectroscopy. Incorporating Bi suppresses the formation of an electron trap with activation energy 0.40 eV, thus reducing the total trap concentration in dilute GaAsBi layers by more than a factor of 20 compared to GaAs grown under the same conditions. We find that the dominant traps in dilute GaAsBi layers are defect complexes involving As{sub Ga}, as expected for MBE growth at these temperatures.

36 MATERIALS SCIENCE↗

Deep-level transient spectroscopy on an amorphous InGaZnO{sub 4} Schottky diode

The first direct measurement is reported of the bulk density of deep states in amorphous IGZO (indium-gallium-zinc oxide) semiconductor by means of deep-level transient spectroscopy (DLTS). The device under test is a Schottky diode of amorphous IGZO semiconductor on a palladium (Pd) Schottky-barrier electrode and with a molybdenum (Mo) Ohmic contact at the top. The DLTS technique allows to independently measure the energy and spatial distribution of subgap states in the IGZO thin film. The subgap trap concentration has a double exponential distribution as a function energy, with a value of ∼10{sup 19} cm{sup −3} eV{sup −1} at the conduction band edge and a value of ∼10{sup 17} cm{sup −3} eV{sup −1} at an energy of 0.55 eV below the conduction band. Such spectral distribution, however, is not uniform through the semiconductor film. The spatial distribution of subgap states correlates well with the background doping density distribution in the semiconductor, which increases towards the Ohmic Mo contact, suggesting that these two properties share the same physical origin.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

Identification of nitrogen- and host-related deep-level traps in n-type GaNAs and their evolution upon annealing

Deep level traps in as-grown and annealed n-GaNAs layers (doped with Si) of various nitrogen concentrations (N=0.2%, 0.4%, 0.8%, and 1.2%) were investigated by deep level transient spectroscopy. In addition, optical properties of GaNAs layers were studied by photoluminescence and contactless electroreflectance. The identification of N- and host-related traps has been performed on the basis of band gap diagram [Kudrawiec, Appl. Phys. Lett. 101, 082109 (2012)], which assumes that the activation energy of electron traps of the same microscopic nature decreases with the rise of nitrogen concentration in accordance with the N-related shift of the conduction band towards trap levels. The application of this diagram has allowed to investigate the evolution of donor traps in GaNAs upon annealing. In general, it was observed that the concentration of N- and host-related traps decreases after annealing and PL improves very significantly. However, it was also observed that some traps are generated due to annealing. It explains why the annealing conditions have to be carefully optimized for this material system.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

Identification of the defect dominating high temperature reverse leakage current in vertical GaN power diodes through deep level transient spectroscopy

Deep level defects in wide bandgap semiconductors, whose response times are in the range of power converter switching times, can have a significant effect on converter efficiency. We use Deep Level Transient Spectroscopy (DLTS) to evaluate such defect levels in the n- drift layer of vertical GaN (v-GaN) power diodes with VBD ~ 1500 V. DLTS reveals three energy levels that are at ~0.6 eV (highest density), ~0.27 eV (lowest density) and ~ 45 meV (a dopant level) from the conduction band. Dopant extraction from Capacitance-Voltage measurement test (C-V) at multiple temperatures enables trap density evaluation, and the ~0.6 eV trap has a density of 1.2 × 10 15 cm -3 . Here, the 0.6 eV energy level and its density are similar to a defect that is known to cause current collapse in GaN based surface conducting devices (like HEMTs). Analysis of reverse bias currents over temperature in the v-GaN diodes indicates a predominant role of the same defect in determining reverse leakage current at high temperatures, reducing switching efficiency.

42 ENGINEERING↗

Deep learning-based predictive models for laser direct drive at the Omega Laser Facility

The rich and complex physics of inertial confinement fusion provides a unique and challenging space for high-fidelity first-principles modeling. Consequently, simulation codes that are used to design experiments are computationally expensive and lack the predictive capability required for extensive parameter exploration in search of a high-performing design for laser direct drive. In this article, we present two deep-learning-based predictive models intended to address these difficulties. The first model (TL DNN) acts as a fast emulator of simulations as well as experiments at the Omega Laser Facility. This model is trained on a simulation database and subsequently calibrated on experimental data using transfer learning. To facilitate the development of this model, an autoencoder is developed to reduce the dimensionality of the input space by compressing the laser pulse input. The model predicts key experimental scalar observables of Omega experiments with high accuracy and minimal computational cost. This deep neural net enables rapid exploration of a high-dimensional input parameter space for an optimal implosion design. The second model (DNN SM+) aims to extend the statistical modeling work of Lees et al. [Phys. Rev. Lett. 127, 105001 (2021)], by increasing the complexity of the model space and allowing for coupling between degradation terms. Since the model capacity of DNN SM+ is higher than the model of Lees et al., DNN SM+ can potentially provide an improvement in predictive capability, and we use this model to provide insight into complicated degradation dependencies.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

A physics-constrained deep learning treatment of runaway electron dynamics

An adjoint formulation leveraging a physics-informed neural network (PINN) is employed to advance the density moment of a runaway electron (RE) distribution forward in time. A distinguishing feature of this approach is that once the adjoint problem is solved, its solution can be used to project the RE density forward in time for an arbitrary initial momentum space distribution of REs. Furthermore, by employing a PINN, a parametric solution to the adjoint problem can be learned. Thus, once trained, this adjoint-deep learning framework is able to efficiently project the RE density forward in time across various plasma conditions while still including a fully kinetic description of RE dynamics. As an example application, the temporal evolution of the density of primary electrons is studied, with particular emphasis on evaluating the decay of a RE population when below threshold. Predictions from the adjoint-deep learning framework are found to be in good agreement with a traditional relativistic electron Fokker–Planck solver, for several distinct initial conditions, and across an array of physics parameters. Once trained, the PINN thus provides a means of generating RE density time histories with exceptionally low online execution time.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Solving high-dimensional partial differential equations using deep learning

Developing algorithms for solving high-dimensional partial differential equations (PDEs) has been an exceedingly difficult task for a long time, due to the notoriously difficult problem known as the “curse of dimensionality.” This paper introduces a deep learning-based approach that can handle general high-dimensional parabolic PDEs. To this end, the PDEs are reformulated using backward stochastic differential equations and the gradient of the unknown solution is approximated by neural networks, very much in the spirit of deep reinforcement learning with the gradient acting as the policy function. Numerical results on examples including the nonlinear Black–Scholes equation, the Hamilton–Jacobi–Bellman equation, and the Allen–Cahn equation suggest that the proposed algorithm is quite effective in high dimensions, in terms of both accuracy and cost. Furthermore, this opens up possibilities in economics, finance, operational research, and physics, by considering all participating agents, assets, resources, or particles together at the same time, instead of making ad hoc assumptions on their interrelationships.

97 MATHEMATICS AND COMPUTING↗

Depth of nutrient uptake by deep-rooted plants is regulated by water availability

The capacity of some plants to access water and nutrients at depths greater than one meter is a critical functional trait that confers resistance to drought and impacts both belowground and shallow soil processes. Here, we report water and strontium isotopic data from an alpine meadow transect showing the correlation between water and nutrient acquisition depths. The isotopic compositions of Sr ( 87 Sr/ 86 Sr ratio) and water in rock and soil, and in plant leaf tissues, reveal that deeper-rooted plants acquire a higher proportion of water, Sr, and cation nutrients that are derived from the saprolite, a zone of silicate weathering, than shallow-rooted grass. A three-decade dendrochemical record reveals that reductions of wet precipitation drive deep-rooted plants to acquire cation nutrients from deeper saprolite or bedrock regions. Thus, the depth of cation nutrient acquisition by deep-rooted plant species at this site is tightly coupled with, and likely determined by, water availability in soil, saprolite, and bedrock. The enhanced uptake of cations as well as water from deeper saprolite zones could impact the rate of bedrock weathering and watershed chemistry during drought.

biogeochemistry↗

Reynolds-Averaged Turbulence Modeling Using Deep Learning with Local Flow Features: An Empirical Approach

Reynolds-Averaged Navier-Stoke (RANS) models offer an alternative avenue in predicting flow characteristics when the corresponding experiments are difficult to achieve due to geometry complexity, limited budget, or knowledge. RANS models require the knowledge of subgrid scale physics to solve conservation equations for mass, energy, and momentum. Mechanistic turbulence models, such as k-epsilon, are generally evaluated and calibrated for specific flow conditions with various degrees of uncertainty. These models have limited capability to assimilate a substantial amount of data due to model form constraints. Meanwhile, deep learning (DL) has been proven to be universal approximators with the potential to assimilate available, relevant, and adequately evaluated data. Moreover, deep neural networks (DNNs) can create surrogate models without knowing function forms. Such a data-driven approach can be used in updating fluid models based on observations as opposed to hard-wiring models with precalibrated correlations. The paper presents progress in applying DNNs to model Reynolds stress using two machine learning (ML) frameworks. A novel flow feature coverage mapping is proposed to quantify the physics coverage of DL-based closures. It can be used to examine the sufficiency of training data and input flow features for data-driven turbulence models. The case of a backward-facing step is formulated to demonstrate that not only can DNNs discover underlying correlation behind fluid data but also they can be implemented in RANS to predict flow characteristics without numerical stability issues. Finally, the presented research is a crucial stepping-stone toward the data-driven turbulence modeling, which potentially benefits the design of data-driven experiments that can be used to validate fluid models with ML-based fluid closures.

42 ENGINEERING↗

Popnet : computer vision based deep learning model for forecasting gridded population

Here, this study introduces Popnet, a deep learning model for forecasting 1 km-gridded populations, integrating U-Net, ConvLSTM, a Spatial Autocorrelation module and deep ensemble methods. Using spatial variables and population data from 2000 to 2020, Popnet predicts South Korea’s population trends by age groups (under 14, 15-64 and over 65) up to 2040. In validation, it outperforms traditional machine learning and state-of-the-art computer vision models. The output of this model discovered significant polarisation: population growth in urban areas, especially the capital region, and severe depopulation in rural areas. Popnet is a robust tool for offering significant insights to policymakers and related stakeholders about the detailed future population, which allows them to establish detailed, localised planning and resource allocations.

computer vision↗

Alpert multi-wavelets for functional inverse problems: direct optimization and deep learning

Computational engineering models often contain unknown entities (e.g. parameters, initial and boundary conditions) that require estimation from other measured observable data. Estimating such unknown entities is challenging when they involve spatio-temporal fields because such functional variables often require an infinite-dimensional representation. Here, we address this problem by transforming an unknown functional field using Alpert wavelet bases and truncating the resulting spectrum. Hence the problem reduces to the estimation of few coefficients that can be performed using common optimization methods. We apply this method on a one-dimensional heat transfer problem where we estimate the heat source field varying in both time and space. The observable data is comprised of temperature measured at several thermocouples in the domain. This latter is composed of either copper or stainless steel. The optimization using our method based on wavelets is able to estimate the heat source with an error between 5% and 7%. We analyze the effect of the domain material and number of thermocouples as well as the sensitivity to the initial guess of the heat source. Finally, we estimate the unknown heat source using a different approach based on deep learning techniques where we consider the input and output of a multi-layer perceptron in wavelet form. We find that this deep learning approach is more accurate than the optimization approach with errors below 4%.

97 MATHEMATICS AND COMPUTING↗

Image-based novel fault detection with deep learning classifiers using hierarchical labels

One important characteristic of modern fault classification systems is the ability to flag the system when faced with previously unseen fault types. This work considers the unknown fault detection capabilities of deep neural network-based fault classifiers. Specifically, we propose a methodology on how, when available, labels regarding the fault taxonomy can be used to increase unknown fault detection performance without sacrificing model performance. To achieve this, we propose to utilize soft label techniques to improve the state-of-the-art deep novel fault detection techniques during the training process and novel hierarchically consistent detection statistics for online novel fault detection. Lastly, we demonstrated increased detection performance on novel fault detection in inspection images from the hot steel rolling process, with results well replicated across multiple scenarios and baseline detection methods.

42 ENGINEERING↗

Tracking atomic structure evolution during directed electron beam induced Si-atom motion in graphene via deep machine learning

Using electron beam manipulation, we enable deterministic motion of individual Si atoms in graphene along predefined trajectories. Structural evolution during the dopant motion was explored, providing information on changes of the Si atom neighborhood during atomic motion and providing statistical information of possible defect configurations. The combination of a Gaussian mixture model and principal component analysis applied to the deep learning-processed experimental data allowed disentangling of the atomic distortions for two different graphene sublattices. This approach demonstrates the potential of e-beam manipulation to create defect libraries of multiple realizations of the same defect and explore the potential of symmetry breaking physics. The rapid image analytics enabled via a deep learning network further empowers instrumentation for e-beam controlled atom-by-atom fabrication. Here, the analysis described in the paper can be reproduced via an interactive Jupyter notebook at https://git.io/JJ3Bx.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗