Related Experiment Videos
A new approach for semiconductor parameter extraction using cathodoluminescence and artificial neural networks
S Soualmia1, A Bouldjedri, A Benhaya
1Department of Physics, Batna University, Algeria. soualmia_s@yahoo.fr
Scanning
|June 4, 2011
Summary
This study introduces a novel method using artificial neural networks (ANN) to extract four key semiconductor parameters from cathodoluminescence data. The technique accurately determines absorption coefficient, diffusion length, dead layer thickness, and quantum efficiency.
Area of Science:
- Semiconductor Physics
- Materials Science
- Computational Physics
Background:
- Cathodoluminescence (CL) is a crucial technique for characterizing semiconductor materials.
- Accurate extraction of semiconductor parameters like absorption coefficient, diffusion length, dead layer thickness, and quantum efficiency is vital for device performance.
- Existing methods may face challenges in simultaneously determining multiple interdependent parameters.
Purpose of the Study:
- To develop a novel parameter extraction technique for simultaneously determining four semiconductor-related parameters from CL data.
- To leverage artificial neural networks (ANN) for learning the complex relationship between CL intensity and fundamental material properties.
- To validate the technique's accuracy and efficiency using experimental data.
Main Methods:
- A feed-forward artificial neural network (ANN) was trained to model the relationship between input semiconductor parameters (absorption coefficient α, diffusion length L, dead layer thickness Zt, relative quantum efficiency Q) and CL intensity versus electron beam energy.
- An exhaustive search method was employed to perform the reverse process, extracting the four parameters from CL curves after ANN training.
- The method optimizes all four parameters simultaneously within a comprehensive search space.
Main Results:
- The ANN-based technique successfully extracts four critical semiconductor parameters (α, L, Zt, Q) from cathodoluminescence data.
- An exhaustive search within the ANN framework ensures the determination of an optimum set of parameter values.
- Computational results on an n-type GaAs sample demonstrated the ability to obtain unique parameter values with errors below 5.5%.
Conclusions:
- The proposed ANN-based parameter extraction method offers an accurate and efficient approach for analyzing cathodoluminescence data.
- This technique enables the simultaneous determination of multiple interdependent semiconductor parameters, improving material characterization.
- The method holds significant potential for advancing the understanding and optimization of semiconductor devices.