Unveiling degradation patterns in dye-sensitized solar cells: a machine learning perspective.

Area of Science:

• Materials Science
• Renewable Energy
• Photovoltaics

Background:

• Dye-sensitized solar cells (DSSCs) are a promising renewable energy technology.
• Understanding and mitigating DSSC degradation is crucial for their commercial viability.
• Key factors influencing DSSC longevity require systematic investigation.

Purpose of the Study:

• To systematically investigate time-dependent degradation in DSSCs.
• To develop a novel figure of merit (FOM) for quantifying degradation rates.
• To enhance predictive modeling of DSSC performance decline.

Main Methods:

• Fabrication of 400 DSSC samples with varied TiO2 thickness, porosity, dye, and electrolyte concentrations.
• Utilized finite element modeling (FEM) in COMSOL for performance simulation.
• Employed Long Short-Term Memory (LSTM) models for degradation prediction.

Main Results:

• DSSC power conversion efficiency (PCE) dropped from an average of 4.0% to 0.5% within 360 hours.
• LSTM models achieved a high prediction accuracy with an R² of 0.92.
• Identified complex interactions between parameters affecting device longevity.

Conclusions:

• Optimizing DSSC design parameters can significantly improve device reliability and lifespan.
• The developed degradation model and FOM are applicable to other solar cell technologies like perovskite and organic cells.
• This research provides a pathway for enhancing the efficiency and durability of next-generation solar cells.