Novel Quantum Dot Gate FETs and Nonvolatile Memories Using Lattice-Matched II–VI Gate Insulators

This paper presents the successful use of ZnS/ZnMgS and other II–VI layers (lattice-matched or pseudomorphic) as high-k gate dielectrics in the fabrication of quantum dot (QD) gate Si field-effect transistors (FETs) and nonvolatile memory structures. Quantum dot gate FETs and nonvolatile memories have been fabricated in two basic configurations: (1) monodispersed cladded Ge nanocrystals (e.g., GeO _x -cladded-Ge quantum dots) site-specifically self-assembled over the lattice-matched ZnMgS gate insulator in the channel region, and (2) ZnTe-ZnMgTe quantum dots formed by self-organization, using metalorganic chemical vapor-phase deposition (MOCVD), on ZnS-ZnMgS gate insulator layers grown epitaxially on Si substrates. Self-assembled GeO _x -cladded Ge QD gate FETs, exhibiting three-state behavior, are also described. Preliminary results on InGaAs-on-InP FETs, using ZnMgSeTe/ZnSe gate insulator layers, are presented.     

Compact Harmonic Filter Design and Fabrication Using IPD Technology

An integrated passive device (IPD) technology has been developed to meet the ever increasing needs of size and cost reduction in radio front-end transceiver module applications. Electromagnetic (EM) simulation was used extensively in the design of the process technology and the optimization of inductor and harmonic filter designs and layouts. Parameters such as inductor shape, inner diameter, metal thickness, metal width, and substrate thickness have been optimized to provide inductors with high quality factors. The technology includes 1) a thick plated gold metal process to reduce resistive loss; 2) MIM capacitors using PECVD SiN dielectric layer; 3) airbridges for inductor underpass and capacitor pick-up; and 4) a 10 mil finished GaAs substrate to improve inductor quality factor. Both lumped element circuit simulations and electromagnetic (EM) simulations have been used in the harmonic filter circuit designs for high accuracy and fast design cycle time. This paper will present the EM simulation calibration and demonstrate the importance of using EM simulation in the filter design in order to achieve first-time success in wafer fabrication. The fabricated IPD devices have insertion loss of 0.5 dB and harmonic rejections of 30dB with die size of 1.42 mm for high band (1710 MHz-1910 MHz) and 1.89 mm for low band (824-915 MHz) harmonic filters.     

A comprehensive approach to the analysis of contrast enhanced cardiac MR images

Current magnetic resonance imaging (MRI) technology allows the determination of patient-individual coronary tree structure, detection of infarctions, and assessment of myocardial perfusion. Joint inspection of these three aspects yields valuable information for therapy planning, e.g., through classification of myocardium into healthy tissue, regions showing a reversible hypoperfusion, and infarction with additional information on the corresponding supplying artery. Standard imaging protocols normally provide image data with different orientations, resolutions and coverages for each of the three aspects, which makes a direct comparison of analysis results difficult. The purpose of this work is to develop methods for the alignment and combined analysis of these images. The proposed approach is applied to 21 datasets of healthy and diseased patients from the clinical routine. The evaluation shows that, despite limitations due to typical MRI artifacts, combined inspection is feasible and can yield clinically useful information.     

Electric drive subsystem for a low-storage requirement hybridelectric vehicle

Integrating an electric machine drive system into the powertrain of a hybrid electric vehicle (HEV) represents a challenging exercise in packaging complex electromechanical and power electronic subsystems. The Ford combined alternator starter (FCAS) and its attendant power and control electronics are physically partitioned because power electronics has not yet evolved to the stage in which fully packaged drives can be realized. A similar situation exists for the control and sensor subsystems necessary for a fully functional high-performance drive. Hardware partitioning requires that more attention be given to installation issues and to mitigating system interactions. The FCAS system consists of an integrated starter/alternator (S/A), an S/A module (SAM), and a vehicle electrical infrastructure that can support the power and energy levels demanded. Our field experience with the FCAS system is presented along with test results obtained from vehicle operation     

The Multi-angle Imaging SpectroRadiometer science data system, its products, tools, and performance

Ground processing of data from the Multi-angle Imaging SpectroRadiometer (MISR) instrument, part of NASA's Earth Observing System (EOS), exploits new and unique science algorithms not previously used operationally. A range of data products from Level 1 through Level 3 is being produced. Because of MISR's unprecedented design, extensive prototyping was required from a relatively early stage. The data throughput is large, necessitating an innovative software design approach that maximizes performance. The systematic science processing software was developed at the Jet Propulsion Laboratory, with data processing occurring at the NASA Langley Research Center using the EOS Core System, a collaborative arrangement that works well. With the availability of actual mission data following launch on the Terra spacecraft in December 1999, MISR's computational needs have become better known, and many improvements have been made to both the science software and the production system to achieve a successful overall data processing capability. This paper provides information about MISR data for the science user, and describes the nature and scope of implementation and operations activities.     

Nonvolatile Memory Effect in Indium Gallium Arsenide-Based Metal–Oxide–Semiconductor Devices Using II–VI Tunnel Insulators

This paper reports the successful use of ZnSe/ZnS/ZnMgS/ZnS/ZnSe as a gate insulator stack for an InGaAs-based metal–oxide–semiconductor (MOS) device, and demonstrates the threshold voltage shift required in nonvolatile memory devices using a floating gate quantum dot layer. An InGaAs-based nonvolatile memory MOS device was fabricated using a high-κ II–VI tunnel insulator stack and self-assembled GeO _x -cladded Ge quantum dots as the charge storage units. A Si₃N₄ layer was used as the control gate insulator. Capacitance–voltage data showed that, after applying a positive voltage to the gate of a MOS device, charges were being stored in the quantum dots. This was shown by the shift in the flat-band/threshold voltage, simulating the write process of a nonvolatile memory device.     

The effects of boron penetration on p+ polysilicon gatedPMOS devices

The penetration of boron into and through the gate oxides of PMOS devices which employ p⁺ doped polysilicon gates is studied. Boron penetration results in large positive shifts in V_FB , increased PMOS subthreshold slope and electron trapping rate, and decreased low-field mobility and interface trap density. Fluorine-related effects caused by BF₂ implantations into the polysilicon gate are shown to result in PMOS threshold voltage instabilities. Inclusion of a phosphorus co-implant or TiSi₂ salicide prior to gate implantation is shown to minimize this effect. The boron penetration phenomenon is modeled by a very shallow, fully-depleted p-type layer in the silicon substrate close to the SiO ₂/Si interface     

Dispersion-induced ultrafast pulse reshaping in 1.55-/spl mu/m InGaAs-InGaAsP optical amplifiers

Using the Foreman effective mass Hamiltonian, the electronic structure of the valence band and the interband dipole matrix elements in In/sub x/Ga/sub 1-x/As-In/sub y/Ga/sub 1-y/As/sub z/P/sub 1-z/ quantum-well optical amplifiers are calculated, taking into account the valence band mixing and the biaxial strain. The optical field of the amplified pulse is calculated by solving the wave equation with the computed polarization as a source term. A novel wavelet transform is introduced in analyzing the pulse chirp imposed by the optical amplifier. In the linear propagation regime, the spectrum of the amplified pulse can be either red-shifted or blue-shifted with respect to its initial center frequency, depending on the local gain dispersion spanned by the pulse spectrum. The output pulse shape can be retarded or advanced, depending on the local gain and group velocity dispersion. Furthermore, an initially unchirped pulse centered in the tail of the gain spectrum is significantly reshaped after propagating 600 /spl mu/m, and its spectrum is broadened and distorted considerably. In the spectral region where both gain and group velocity change rapidly, the frequency chirp for a linearly chirped input pulse is significantly weakened after propagation.     

Decoupled multiuser code-timing estimation for code-divisionmultiple-access communication systems

We present herein a decoupled multiuser acquisition (DEMA) algorithm for code-timing estimation in asynchronous code-division multiple-access (CDMA) communication systems. The DEMA estimator is an asymptotic (for large data samples) maximum-likelihood method that models the channel parameters as deterministic unknowns. By evoking the mild assumption that the transmitted data bits for all users are independently and identically distributed, we show that the multiuser timing estimation problem that usually requires a search over a multidimensional parameter space decouples into a set of noniterative one-dimensional problems. Hence, the proposed algorithm is computationally efficient. DEMA has the desired property that, in the absence of noise, it obtains the exact parameter estimates even with a finite number of data samples which can be heavily correlated. Another important feature of DEMA is that it exploits the structure of the receiver vectors and, therefore, is near-far resistant. Numerical examples are included to demonstrate and compare the performances of DEMA and a few other standard code-timing estimators     

Scalable VLSI implementations for neural networks

This paper discusses research on scalable VLSI implementations of feed-forward and recurrent neural networks. These two families of networks are useful in a wide variety of important applications—classification tasks for feed-forward nets and optimization problems for recurrent nets—but their differences affect the way they should be built. We find that analog computation with digitally programmable weights works best for feed-forward networks, while stochastic processing takes advantage of the integrative nature of recurrent networks. We have shown early prototypes of these networks which compute at rates of 1–2 billion connections per second. These general-purpose neural building blocks can be coupled with an overall data transmission framework that is electronically reconfigured in a local manner to produce arbitrarily large, fault-tolerant networks.