Recent developments and case studies regarding VLSI device chip failure analysis are reviewed. The key failure analysis techniques reviewed include EMMS (emission microscopy), OBIC (optical beam induced current), LCM (liquid crystal method), EBP (electron beam probing), and FIB (focused ion beam method). Further, future possibilities in failure analysis, and some promising new tools are introduced.

A novel pulse neural network model for sound localization has been proposed. Our model is based on the physiological auditory nervous system. Human beings can perceive the sound direction using inter-aural time difference (ILD) and inter-aural level difference (ILD) of two sounds. The model extracts these features using only pulse train information. The model is divided roughly into three sections: preprocessing for input signals; transforming continuous signals to pulse trains; and extracting features. The last section consists of two parts: ITD extractor and ILD extractor. Both extractors are implemented using a pulse neuron model. They have the same network structure, differing only in terms of parameters and arrangements of the pulse neuron model. The pulse neuron model receives pulse trains and outputs a pulse train. Because the pulses have only simple informations, their data structures are very simple and clear. Thus, a strict design is not required for the implementation of the model. These advantages are profitable for realizing this model by hardware. A computer simulation has demonstrated that time and level differences between two signals have been successfully extracted by the model.

In this paper, I investigate a property of self-organizing feature map (SOFM) in terms of reference vector density q(x) when probability density function of input signal fed into SOFM is p(x). Difficulty of general analysis on this property is briefly discussed. Then, I employ an assumption (conformal map assumption) to evaluate this property, and it is shown that for equilibrium state, q(x)p(x)s holds. By giving Lyapunov functioin for time evolution of reference vector density q(x) in SOFM, the equilibrium state is proved to be stable in terms of distribution. Comparison of the result with one which is based on different assumption reveals that there is no unique result of a simple form, such as conjectured by Kohonen. However, as there are cases in which these assumptions hold, these results suggest that we can consider a range of the property of SOFM. On the basis of it, we make comparison on this property between SOFM and fundamental adaptive vector quantization algorithm, in terms of the exponent s of the relation q(x)p(x)s. Difference on this property between SOFM and fundamental adaptive vector quantization algorithm, and propriety of mean squared quantization error for a performance measure of SOFM, are discussed.

Optical and electrical measurements of thin film n-channel SOI-MOSFETs reveal that the exponential tail in photon emission spectra originates from electron-hole recombination. Bremsstrahlung radiation model as a physical mechanism of photon emission was experimentally negated. Negative threshold voltage shift at the initial stage of high field stress is found to be caused by hole trapping in buried oxide. Subsequent turnover characteristics is explained by a competing process between electron trapping in the front gate oxide and hole trapping in the buried oxide. As to the degradation of transconductance, generated surface state as well as trapped holes in the buried oxide which reduce vertical electric field in SOI film are involved in the complicate degradation of transconductance.

The dielectric breakdown strength of thermally grown silicon dioxide films was studied for MOS capacitors fabricated on silicon wafers that were intentionally contaminated with magnesium and zinc. Most of magnesium was detected in the oxide film after oxidation. Zinc, some of which evaporated from the surface of wafers, was detected only in the oxide film. The mechanism of the dielectric degradation is dominated by formation of metal silicates, such as Mg2SiO4 (Forsterite) and Zn2SiO4 (Wilemite). The formation of metal silicates has no influence on the generation lifetime of minority carriers, however, it provides the flat-band voltage shift less than 0.3 eV, and forces to increase the density of deep surface states with the zinc contamination.

This paper proposes new algorithms for adaptive FIR filters. The proposed algorithms provide both fast convergence and small final misadjustment with an adaptive step size even under an interference to the error. The basic algorithm pays special attention to the interference which contaminates the error. To enhance robustness to the interference, it imposes a special limit on the increment/decrement of the step-size. The limit itself is also varied according to the step-size. The basic algorithm is extended for application to nonstationary signals. Simulation results with white signals show that the final misadjustment is reduced by up to 22 dB under severe observation noise at a negligible expense of the convergence speed. An echo canceler simulation with a real speech signal exhibits its potential for a nonstationary signal.

This paper investigates the properties of synchronized alternating one-way multicounter machines (lsamcm's) which operate in real time (lsamcm-real's) and whose leaf-sizes are bounded by a constant or some function of the length of an input. Leaf-size reflects the number of processors which run in parallel in scanning a given input. We first consider the hieracrchies of lsamcm-real's based on the number of counters and constant leaf-sizes. We next show that lsamcm-real's are less powerful than lsamcm's which operate in linear time when the leaf-sizes of these machines are bounded by a function L(n) such that limn[L(n) log n/n]0 and L(n)2.

The main problem in statistical pattern recognition is to design a classifier. Many researchers point out that a finite number of training samples causes the practical difficulties and constraints in designing a classifier. However, very little is known about the performance of a classifier in small training sample size situations. In this paper, we compare the classification performance of the well-known classifiers (k-NN, Parzen, Fisher's linear, Quadratic, Modified quadratic, Euclidean distance classifiers) when the number of training samples is small.

This paper deals with the size of switching functions in Exclusive-OR sum-of-products expressions (ESOPs). The size is the number of products in ESOP. There are no good algorithms to find an exact minimum ESOP. Since the exact minimization algorithms take a time in double exponential order, it is almost impossible to minimize ESOPs for an arbitrary n-variable functions with n5. Then,it is necessary to study the size of some concrete functions. These concrete functions are useful for testing heuristic minimization algorithms. In this paper we present the lower bounds on size of periodic functions in ESOPs. A symmetric function is said to be periodic when the vector of weights of inputs X such that f(X)1 is periodic. We show that the size of a 2t+1-periodic function with rank r is proportional to n2t+r, where t0 and 0r2t, i.e., in polynomial order,and thet the size of a (2s+1)2t-periodic function with s0 and t0 is greater than or equal to (3/2)n-(2s+1)2t, i.e., in exponential order. The concrete function the size of which is greater than or equal to 32(3/2)n-8 is presented. This function requires the largest size among the concrete functions the sizes of which are known. Some results for non-periodic symmetric functions are also given.

Thin (Ba0.75Sr0.25)TiO3 (BST) films to be used as dielectric materials in 256 Mbit DRAM capacitors were investigated. These films were deposited by an rf-sputtering method at substrate temperatures of 480 to 750. As substrate temperature increases, the dielectric constant to the films also increases, from 230 to 550. BST films prepared at temperatures higher than 700 show larger current leaks than films prepared at lower temperatures. A dielectric constant of 250, corresponding to a silicon oxide equivalent thickness (teq) of 0.47 nm, and a leak current density about 110-8 A/cm2 were obtained in 30-nm-thick film deposited at 660. Both of these values are sufficient for use in a 256 Mbit DRAM capacitor.

Genetic algorithms have been shown to be very useful in a variety of search and optimization problems. In this paper, we describe the implementation of genetic algorithms for channel routing problems and identify the key points which are essential to making full use of the population of potential solutions, that is one of the characteristics of genetic algorithms. Three efficient crossover techniques which can be divided further into 13 kinds of crossover operators have been compared. We also extend our previous work with ability to deal with dogleg case by simply splitting multi-terminal nets into a series of 2-terminal subnets. It routes the Deutsch's difficult example with 21 tracks without any detours.

This paper discusses a way of identifying the network configuration of ATM-LANs, which are composed of a number of ATM hubs. In general, a Network Management System (NMS) sets and gets the necessary data to and from the network elements. In managing an ATM-LAN, the ATM connection between the NMS and each network element, namely the ATM hub, must be established in order to get and set the necessary data. This forms a remarkable contrast with conventional LANs such as the IEEE802.3 LAN, which is a shared media network and enables broadcast communication without setting up any connection. This paper proposes a new protocol and a procedure that establishes the ATM connection between the NMS and each ATM hub, while identifying the overall network configuration. First, this paper makes clear the peculiarity of the ATM-LAN in terms of automatically identifying the network configuration. Next, the identification protocol that achieves the required properties is precisely explained. Then, the proposed identification protocol is evaluated in terms of required bandwidth and identification time.

Dislocation-free thin silicon layers are created on the three kinds of substrates such as oxide film, synthetic quartz glass and sapphire. They are bonded with silicon wafers using hydrogen bonding at room temperature but without any adhesive, and their bonding are changed into covalent bonding at elevated temperature. Thick (2 µm) silicon layers are first produced by surface grinding and polishing, and then thinned to 0.1 µm by plasma assisted chemical etching (PACE). A multiple repeated process of thinning the silicon layer and annealing the bonded silicon/quartz and silicon/sapphire interface is applied for tight bonding between a silicon wafer and a quartz wafer, and a silicon wafer and a sapphire wafer which have different thermal expansion coefficients. In case of bonding with sapphire, oxide with 200 in thickness plays an important role in the preventions of void formation and diffusion of interface contaminants into the silicon layer.

A 0.25-µm BiCMOS technology has been developed using three sophisticated technologies; the HSST/BiCMOS device, synchrotron orbital radiation (SOR) X-ray lithography, and an advanced two-level metallization. The HSST/BiCMOS provides a 25.4-ps double-poly bipolar device using High-performance Super Self-Aligned Process Technology (HSST), and a 42 ps/2 V CMOS inverter. SOR lithography allows a 0.18 µm gate and 0.2 µm via-hole patternings by using single-level resists. The metallization process features a new planarization technique of the 0.3-µm first wire, and a selective CVD aluminum plug for a 0.25 µm via-hole with contact resistance lower than 1Ω. These 0.25-µm technologies are used to successfully fabricate a 4 KG 0.25 µm CMOS gate-array LSI on a BiCMOS test chip of 12 mm square, which operates at 58 ps/G at 2 V. This result demonstrates that SOR lithography will pave the way for the fabrication of sub-0.25-µm BiCMOS ULSIs.

The theoretical modelling bandgap narrowing and percentage of ionized impurity atoms for uncompensated uniformly doped silicon containing conventional impurities (B, P, As, Sb) under thermodynamic-equilibrium conditions is presented. As distinct from existing approaches, this modelling is valid for impurity concentrations up to electrically-active-impurity-concentration limits and for the temperature range from 40 K up to 400 K. A relevant and efficient calculation software is proposed. The results of the calculations are compared with the results extracted by many authors from measurement data. A good agreement between these results is noted and possible reasons of some discrepancies are pointed out. The present modelling and software can be used for investigation of BJT charge-neutral regions as well as diffused or implanted resistors.

This paper describes a new free-space optical switch structure based on cascaded beam shifters (each consists of a liquid-crystal polarization controller array and a birefringent plate). This structure comprises 2-input, 2-output switching elements that are locally connected by links. It is applicable to a variety of switching networks, such as a Clos network. The switching network based on this structure is an analog switch that is transparent to signal format, bit rate, and modulation type, so it can handle various types of optical signals. Theoretical feasibility studies indicate that compact large-scale switches (i.e., 100-1000 ports) with relay lens systems can be implemented using beam shifters with a 0.4-dB insertion loss and a 30-dB extinction ratio. Experimental feasibility studies indicate that a 1024-cell beam shifter module with a 0.5-dB insertion loss and a 23-dB extinction ratio is possible at present. An alignment-free assembly technique using precise alignment guides is also confirmed. An experimental 8-stage, 1024-input 256-output concentrator shows low insertion loss characteristics (6.8dB on average) owing to the low-loss beam shifters and the alignment-free assembly technique. Practical switching networks mainly require the improvement of the extinction ratio of the beam shifter module and the development of a fiber pig-tailing technique. This switch structure is applicable to transparent switching networks such as subscriber line concentrators and inter-module connectors.

A new mobility model dependent upon electron concentration is presented for studying the screening effect on ionized impurity scattering. By coupling this model with the drift-diffusion and Hartree models, the effects of self-consistent and quasi-equilibrium screening on carrier transport in heavily doped systems are revealed for first time. The transport mechanism is found to be dominated by the electron-concentration-dependent mobility, and transconductance is shown to be determined by effective mobility and changes from degraded to enhanced characteristics with electron concentration modulation.

An eye-contact technique using a blazed half-transparent mirror (BHM) is developed. This half-transparent mirror (HM) consists of an in-line array of many slanting micro-HMs. We fabricated a prototype system and confirmed the principle of this technique. The resolution of an image reflected by a BHM was simulated to determine how to improve the image quality and the factors degrading the resolution were clarified.

This paper describes an implementation of the finite element method to examine the effects of actual lip shape on the sound radiation. A three-dimensional finite element approach by Galerkin method was used. The accuracy of the calculation of finite element method for the sound radiation was tested by comparing it with the exact solutions for a circular piston radiator on an infinite baffle. Using a set of finite element models of the vocal tract, we calculated the responses to a pure tone input and the sound fields over the frequency range of 100 Hz-7 kHz. The transfer functions are examined in detail for vowels /a/ and /i/ when the shape of the actual lips is simplified as a planeradiation surface. The effects of lip shape on the distribution of sound pressures are also shown in both the vocal tract and the surrounding space of the mouth opening.

A photonic network with a space- and frequency switching capability is proposed. It provides point-to-point and point-to-multipoint connections without internal blocking. The switching network exclusively uses frequency switching stages and a shared-medium architecture. Our proposal is compared with similar published networks which are either also constructed solely from frequency switching stages or from frequency and space switching stages. It is shown that the proposed switching network features fewer optical and opto-electronic components, fewer different types of component/module, lower losses, a higher capacity and an easier expansibility.