MACHINE LEARNING-GUIDED STUDY OF FLUOROSCENCE CARBON DOTS WITH OPTICAL PROPERTIES PREDICTION

• Main Author: Faiz, Abdurrahman
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/71822
• Institution: Institut Teknologi Bandung

Carbon dots (CDs) are a type of carbon nanoparticles that have been extensively studied for their unique properties, such as adjustable fluorescence, stability and biocompatibility. However, obtaining CDs with a certain fluorescence can be challenging due to many factors affecting their synthesis, such as precursor concentration, reaction time, pH and type of solvent. To vary the synthesis factors, repeated experiments with errors are required which require a lot of money. To predict the peak wavelength of fluorescence on CDs, one of the methods that can be used is machine learning. Machine learning (ML) is a technique that allows computers to learn without being given explicit instructions. In this study, ML is used to predict the peak wavelength of CDs fluorescence emissions. CDs were synthesized using ethylenediamine (EDA) and phosphoric acid (PA) precursors with the help of microwave radiation. Synthesis parameters such as precursor volume and irradiation time are varied, along with the peak wavelength of the fluorescence emission (excitation 365 nm) used to build a dataset that is used as input to ML algorithms. The algorithms used are algorithms such as decision tree, random forest, and XGBoost. The algorithm is tested using R2 and RMSE values. Based on the results of the synthesis of CDs, it shows blue fluorescence when irradiated with UV light with different intensities and wavelengths between samples. The results of the optimized XGBoost Model show that R2 is around 0.94 in the test data and 0.99 in the training data.