Classification of Semiconductors Using Photoluminescence Spectroscopy and Machine Learning

Area of Science:

• Materials Science
• Spectroscopy
• Artificial Intelligence

Background:

• Photoluminescence spectroscopy is a key technique for semiconductor characterization.
• It relies on the optical emission of light after photon absorption.
• Accurate material identification is crucial for semiconductor applications.

Purpose of the Study:

• To develop a machine learning-based method for identifying semiconductor substances from their photoluminescence spectra.
• To classify six specific semiconductor materials: gallium oxide (Ga2O3), zinc oxide (ZnO), gallium nitride (GaN), cadmium sulfide (CdS), tungsten disulfide (WS2), and cesium lead bromide (CsPbBr3).

Main Methods:

• Utilized simulated photoluminescence spectra as input data for neural network models.
• Trained neural networks to map spectral data to specific semiconductor identities.
• Evaluated the classification accuracy of the developed algorithm.

Main Results:

• The machine learning algorithm achieved high accuracy (>90%) in identifying the six chosen semiconductor materials.
• The model successfully identified a mixed sample containing gallium oxide and zinc oxide.
• Demonstrated the efficacy of photoluminescence spectroscopy combined with machine learning for material identification.

Conclusions:

• Machine learning, specifically neural networks, offers an efficient and accurate approach to identifying semiconductor materials using photoluminescence spectra.
• This technique holds promise for automated material characterization and quality control in semiconductor research and industry.