Lamiae
Sommaire: Extraction des paramètres de semiconducteur en utilisant les méthodes EBIC et cathodoluminescence
Abstract
Table of Contents
List of Figures
List of Tables
Chapter 1 Introduction
1.1 Background
1.2 Aim of this Work
1.3 Thesis Organization
1.4 Thesis Contribution
Chapter references
Chapter 2 Scanning Electron Microscopy fundamentals
2.1 Introduction
2.2 Interaction of the incident electrons with the irradiated matter
2.2.1 Physical process
2.2.2 function of energy dissipation
2.2.3 electron range
2.2.4 generation of pairs
2.2.5 Generation volume
2.3 Cathodoluminescence (CL)
2.3.1 Physical process
2.3.2 Formation of the CL signal
2.3.3 Calculation of the CL signal using Hergert et al model
2.4 Electron beam induced current (EBIC)
2.4.1 Physical phenomena
2.4.2 Calculation of the EBIC in a normal collector p-n junction configuration
2.4.2.1 The charge collection probability of Donolato
2.4.2.2 Calculation of the EBIC
Chapter references
Chapter 3 Artificial Neural Networks
3.1 Introduction
3.2 The neuron
3.3 Neural networks topology
3.4 Feedforward networks
3.4.1 Single layer feedforward networks
3.4.2 Multilayer feedforward networks
3.5 The learning process
3.6 The learning algorithm
3.6 The backpropagation algorithm
3.6 The Levenberg Marquardt algorithm
3.7 ANN for function approximation
3.7.1 System identification
3.7.2 Inverse modeling
Chapter references
Chapter 4 Genetic algorithms
4.1 Introduction
4.2 The GA operators
4.2.1 Selection
4.2.2 Crossover
4.2.3 Mutation
4.3 The GA parameters
4.3.1 Population options
4.3.2 Fitness scaling options
4.3.3 Selection options
4.3.4 Crossover options
4.3.5 Mutation options
4.3.6 Stopping criteria options
4.4 The continuous GA algorithm
Chapter references
Chapter 5 Semiconductor parameter extraction using artificial neural networks and exhaustive search
5.1 Introduction
5.2 Parameter extraction based on ANN and exhaustive search
5.2.1 Preparation of the training and test data sets
5.2.2 Training the ANN algorithm
5.2.3 Testing the ANN algorithm
5.2.4 Oversampling of the signal using ANN
5.2.5 Parameter extraction through exhaustive search
5.3 Application to cathodoluminescence
5.3.1 Training the ANN
5.3.2 Testing the algorithm
5.3.3 Oversampling of the CL signal
5.3.4 Exhaustive search
5.3.5 Effect of measurment noise
5.4 Application to EBIC
5.4.1 Training the ANN
5.4.2 Testing the algorithm
5.4.3 Oversampling of EBIC
5.4.4 Exhaustive search
5.5 Conclusion
Chapter references
Chapter 6 Semiconductor parameter extraction using artificial neural networks and inverse modeling
6.1 Introduction
6.2 Parameter extraction based on ANN and inverse modeling
6.2.1 Preparation of the training and test data sets
6.2.2 Training the ANN algorithm
6.2.3 Testing the ANN algorithm
6.3 Application to cathodoluminescence
6.3.1 Training the algorithm
6.3.2 Testing the algorithm
6.4 Application to EBIC
6.4.1 Training the algorithm
6.4.2 Testing the algorithm
6.5 Conclusion
Chapter references
Chapter 7 Semiconductor parameter extraction using genetic algorithms
7.1 Introduction
7.2 Parameter extraction using genetic algorithms
7.2.1 Initialize the parameters
7.2.2 Define the objective function
7.2.3 Apply the genetic algorithm
7.2.4 Extract the solution
7.3 Application to cathodoluminescence Effect of the initial population size
7.4 Application to EBIC
7.5 Conclusion
Chapter references
Chapter 8 Conclusion and future work
8.1 Conclusion
8.2 Future work
Author publications
Appendix A
Extrait du mémoire extraction des paramètres de semiconducteur en utilisant les méthodes EBIC et cathodoluminescence
Chapter 1: Introduction
1.1 Background
The investigation of electronic and optical properties of semiconductors is of fundamental importance as the performance of electronic devices is dictated by the properties of the carriers inside the materials. The scanning electron microscopy has proven to be an effective, useful and nondestructive technique in the characterization of semiconductor materials. In scanning electron microscopy, Electron Beam Induced current (EBIC) and Cathodoluminescence (CL) have been extensively used as an analytical tool in the characterization of semiconductor materials. They have been widely used for the investigation of crystal defects in semiconductors [1-7] and for the extraction of defect free region semiconductor parameters such as: absorption coefficient (α), diffusion length (L), dead layer thickness (Zt), relative quantum efficiency (Q) and the normalized surface recombination velocity (S) [8-14]. The extraction of these parameters is difficult because of there nonlinear and complex effect on the CL/EBIC signal, and the high interaction between theme.
The parameter extraction problem is a multi-minimum optimization problem [15,16] and it can be solved using several optimization techniques, which can be roughly classified into two categories: deterministic optimization algorithms (pseudo-objective function substitution method (POSM)[17], Levenberg-Marquardt [18], the gradient descent methods [19],… ) and stochastic optimization algorithms ( simulated annealing method (SA) [20], simulated 1-diffusion method (SD)[21], genetic algorithms (GA) [22, 23]…). These methods were applied to parameter extraction of semiconductor devices and integrated circuits.
……….
Si le lien ne fonctionne pas correctement, veuillez nous contacter (mentionner le lien dans votre message)Mémoire Online: Extraction des paramètres de semiconducteur en utilisant les méthodes EBIC et cathodoluminescence (3.12 MB) (Cours PDF)
