Modeling and optimization of semiconductor manufacturing process with neural networks

A neural network-based process model is proposed to optimize the semiconductor manufacturing process. Being different from some works in several research groups which developed neural network-based models to predict process quality with a set of process variables of only single manufacturing step, we applied this model to wafer fabrication parameters control and wafer lot yield optimization. The original data are collected from a wafer fabrication line, including technological parameters and wafer test results. The wafer lot yield is taken as the optimization target. Learning from historical technological records and wafer test results, the model can predict the wafer yield. To eliminate the "bad" or noisy samples from the sample set, an experimental method was used to determine the number of hidden units so that both good learning ability and prediction capability can be obtained.

Fracture properties of PECVD silicon nitride thin films by long rectangular memrane bulge test

Bulge test combined with a refined load-deflection model for long rectangular membrane was applied to determine the mechanical and fracture properties of PECVD silicon nitride (SiNx) thin films. Plane-strain modulus E-ps prestress s(0), and fracture strength s(max) of SiNx thin films deposited both on bare Si substrate and on SiO2-topped Si substrate were extracted. The SiNx thin films on different substrates possess similar values of E-ps and s(0) but quite different values of s(max). The statistical analysis of fracture strengths were performed by Weibull distribution function and the fracture origins were further predicted.

Fracture properties of silicon carbide thin films charcterized by bulge test of long membranes

The mechanical properties and fracture behavior of silicon carbide (3C-SiC) thin films grown on silicon substrates were characterized using bulge testing combined with a refined load-deflection model for long rectangular membranes. Plane-strain modulus E-ps, prestress so, and fracture strength s(max) for 3C-SiC thin films with thickness of 0.40 mu m and 1.42 mu m were extracted. The E, values of SiC are strongly dependent on grain orientation. The thicker SIC film presents lower so than the thinner film due to stress relaxation. The s(max) values decrease with increasing film thickness. The statistical analysis of the fracture strength data were achieved by Weibull distribution function and the fracture origins were predicted.

Dynamic power management approaches based on neural network

Dynamic Power Management (DPM) is a technique to reduce power consumption of electronic system. by selectively shutting down idle components. In this article we try to introduce back propagation network and radial basis network into the research of the system-level policies. We proposed two PAY policies-Back propagation Power Management (BPPM) and Radial Basis Function Power management (RBFPM) which are based on Artificial Neural Networks (ANN). Our experiments show that the two power management policies greatly lowered the system-level power consumption and have higher performance than traditional Power Management(PM) techniques-BPPM is 1.09-competitive and RBFPM is 1.08-competitive vs. 1.79,145,1.18-competitive separately for traditional timeout PM, adaptive predictive PM and stochastic PM.

SOC dynamic power management using artificial neural network

Dynamic Power Management (DPM) is a technique to reduce power consumption of electronic system by selectively shutting down idle components. In this article we try to introduce back propagation network and radial basis network into the research of the system-level power management policies. We proposed two PM policies-Back propagation Power Management (BPPM) and Radial Basis Function Power Management (RBFPM) which are based on Artificial Neural Networks (ANN). Our experiments show that the two power management policies greatly lowered the system-level power consumption and have higher performance than traditional Power Management(PM) techniques-BPPM is 1.09-competitive and RBFPM is 1.08-competitive vs. 1.79 . 1.45 . 1.18-competitive separately for traditional timeout PM . adaptive predictive PM and stochastic PM.

Object-recognition with oblique observation directions based on biomimetic pattern recognition

In this paper, we propose a new scheme for omnidirectional object-recognition in free space. The proposed scheme divides above problem into several onmidirectional object-recognition with different depression angles. An onmidirectional object-recognition system with oblique observation directions based on a new recognition theory-Biomimetic Pattern Recognition (BPR) is discussed in detail. Based on it, we can get the size of training samples in the onmidirectional object-recognition system in free space. Omnidirection ally cognitive tests were done on various kinds of animal models of rather similar shapes. For the total 8400 tests, the correct recognition rate is 99.89%. The rejection rate is 0.11% and on the condition of zero error rates. Experimental results are presented to show that the proposed approach outperforms three types of SVMs with either a three degree polynomial kernel or a radial basis function kernel.

ELECTRONIC AND MAGNETIC-STRUCTURE OF THE TERNARY COMPOUND ND2FE17N

The electronic and the magnetic structure of the Nd2Fe17N1 phase in the family of Nd-Fe-N ternary compounds have been calculated using the first-principles, spin-polarized orthogonalized linear-combination-of-atomic-orbitals method. Results are presented in the form of site-decomposed and spin-projected partial density of states. The occupation sites of the three N atoms are determined by an average radial distribution of all possible N site configurations. Both cases of N occupying the 3b and the 18g sites are studied. The results indicate that the 6c Fe sites have the maximum and the 18h Fe sites have the minimum local moments. This is in good agreement with experiment. It is concluded that the influence on the local moment due to lattice expansion is less important compared to that due to interatomic interaction between the N atom and its neighbors. The results also show the important role of N atoms in raising the Curie temperature of this compound.

Relationship between load/unload response ratio and damage variable and its application

The physics-based parameter: load/unload response ratio (LURR) was proposed to measure the proximity of a strong earthquake, which achieved good results in earthquake prediction. As LURR can be used to describe the damage degree of the focal media qualitatively, there must be a relationship between LURR and damage variable (D) which describes damaged materials quantitatively in damage mechanics. Hence, based on damage mechanics and LURR theory, taking Weibull distribution as the probability distribution function, the relationship between LURR and D is set up and analyzed. This relationship directs LURR applied in damage analysis of materials quantitatively from being qualitative earlier, which not only provides the LURR method with a more solid basis in physics, but may also give a new approach to the damage evaluation of big scale structures and prediction of engineering catastrophic failure. Copyright (c) 2009 John Wiley & Sons, Ltd.

A four-phase confocal elliptical cylinder model for predicting the effective thermal conductivity of coated fibre composites

A four-phase confocal elliptical cylinder model is proposed from which a generalised self-consistent method is developed for predicting the thermal conductivity of coated fibre reinforced composites. The method can account for the influence of the fibre section shape ratio on conductivity, and the physical reasonableness of the model is demonstrated by using the fibre distribution function. An exact solution is obtained for thermal conductivity by applying conformal mapping and Laurent series expansion techniques of the analytic function. The solution to the three-phase confocal elliptical model, which simulates composites with idealised fibre-matrix interfaces, is arrived at as the degenerated case. A comparison with other available micromechanics methods, Hashin and Shtrikman's bounds and experimental data shows that the present method provides convergent and reasonable results for a full range of variations in fibre section shapes and for a complete spectrum of the fibre volume fraction. Numerical results show the dependence of the effective conductivities of composites on the aspect ratio of coated fibres and demonstrate that a coating is effective in enhancing the thermal transport property of a composite. The present solutions are helpful to analysis and design of composites.

Relationship between load/unload response ratio and damage variable and its application

The physics-based parameter: load/unload response ratio (LURR) was proposed to measure the proximity of a strong earthquake, which achieved good results in earthquake prediction. As LURR can be used to describe the damage degree of the focal media qualitatively, there must be a relationship between LURR and damage variable (D) which describes damaged materials quantitatively in damage mechanics. Hence, based on damage mechanics and LURR theory, taking Weibull distribution as the probability distribution function, the relationship between LURR and D is set up and analyzed. This relationship directs LURR applied in damage analysis of materials quantitatively from being qualitative earlier, which not only provides the LURR method with a more solid basis in physics, but may also give a new approach to the damage evaluation of big scale structures and prediction of engineering catastrophic failure. Copyright (c) 2009 John Wiley & Sons, Ltd.

Brownian dynamics simulation of the crystallization dynamics of charged colloidal particles

Crystal formation process of charged colloidal particles is investigated using Brownian dynamics (BD) simulations. The particles are assumed to interact with the pair-additive repulsive Yukawa potential. The time evolution of crystallization process and the crystal structure during the simulation are characterized by means of the radial distribution functions (RDF) and mean square displacement (MSD). The simulations show that when the interaction is featured with long-range, particles can spontaneously assemble into body-centered-cubic (BCC) arrays at relatively low particle number density. When the interaction is short-ranged, with increasing the number density particles become trapped into a stagnant disordered configuration before the crystallization could be actualized. The simulations further show that as long as the trapped configurations are bypassed, the face-centered-cubic (FCC) structures can be achieved and are actually more stable than BCC structures. (C) 2010 Elsevier Inc. All rights reserved.

Numerical investigation of particle velocity distributions in aeolian sand transport

Particle velocity distribution in a blowing sand cloud is a reflection of saltation movement of many particles. Numerical analysis is performed for particle velocity distribution with a discrete particle model. The probability distributions of resultant particle velocity in the impact-entrainment process, particle horizontal and vertical velocities at different heights and the vertical velocity of ascending particles are analyzed. The probability distributions of resultant impact and lift-off velocities of saltating particles can be expressed by a log-normal function, and that of impact angle comply with an exponential function. The probability distribution of particle horizontal and vertical velocities at different heights shows a typical single-peak pattern. In the lower part of saltation layer, the particle horizontal velocity distribution is positively skewed. Further analysis shows that the probability density function of the vertical velocity of ascending particles is similar to the right-hand part of a normal distribution function, and a general equation is acquired for the probability density function of non-dimensional vertical velocity of ascending particles which is independent of diameter of saltating particles, wind strength and height. These distributions in the present numerical analysis are consistent with reported experimental results. The present investigation is important for understanding the saltation state in wind-blown sand movement. (C) 2009 Elsevier B.V. All rights reserved.

Isotopic dependence of production cross sections of superheavy nuclei in hot fusion reactions

The dinuclear system model has been further developed by introducing the barrier distribution function method in the process of heavy-ion capture and fusion to synthesize superheavy nuclei. The capture of two colliding nuclei, formation and de-excitation process of compound nucleus are decribed by using empirical coupled channel model, solving master equation numerically and statistical evaporation model, respectively. Within the framework of the dinuclear system model, the fusion-evaporation excitation functions of the systems Ca-48(Am-243, 3n-5n) (288-286)115 and Ca-48(Cm-248, 3n-5n)(293-291)116 are calculated, which are used for synthesizing new superheavy nuclei at Dubna in recent years. Isotopic dependence of production cross sections with double magic nucleus Ca-48 bombarding actinide targets U, Np, Pu, Am, Cm to synthesize superheavy nuclei with charged numbers Z=112-116 is analyzed systematically. Based on these analysis, the optimal projectile-target combination and the optimal excitation energy are proposed. It is shown that shell correction energy and neutron separation energy will play an important role on the isotopic dependence of production cross sections of superheavy nuclei.