Comparing data driven and physics inspired models for hopping transport in organic field effect transistors

Area of Science:

• Organic electronics
• Condensed matter physics
• Materials science

Background:

• Organic semiconductors are increasingly used in devices like transistors and LEDs.
• Disordered organic semiconductors typically exhibit hopping transport.
• Existing models for hopping transport lack consensus and uniform accuracy.

Purpose of the Study:

• To evaluate data-driven symbolic regression for modeling field-effect mobility in organic semiconductors.
• To investigate a physics-inspired renormalization approach for describing charge transport.
• To compare the interpretability and accuracy of different modeling approaches.

Main Methods:

• Symbolic regression to model the relationship between field-effect mobility, temperature, and gate voltage.
• A physics-inspired renormalization approach using a scale-invariant reference temperature.
• Comparison of model accuracy using Mean Absolute Error (MAE).

Main Results:

• Symbolic regression achieved high accuracy (MAE ~ O(10^-2)) but lacked physical interpretability.
• The renormalization approach provided better generality and interpretability (MAE ~ O(10^-1)) but was less accurate than symbolic regression.
• Both approaches outperformed the traditionally used hopping transport model.

Conclusions:

• Physics-based modeling approaches offer superior interpretability and intuitive understanding of experimental data.
• Data-driven methods like symbolic regression excel in accuracy but may sacrifice physical insight.
• Hybrid approaches combining data-driven insights with physics-based frameworks may be beneficial for organic semiconductor research.