Synaptic κ-Ga₂O₃ Photodetectors for Privacy-Enhancing Neuromorphic Computing

Area of Science:

• Optoelectronics
• Materials Science
• Artificial Intelligence

Background:

• Intelligent edge systems require integrated sensing, memory, and computation.
• Persistent photoconductivity (PPC) in materials offers potential for neuromorphic applications.
• Gallium oxide (Ga2O3) is a promising semiconductor for advanced electronic devices.

Purpose of the Study:

• To develop a multifunctional neuromorphic photodetector using κ-phase gallium oxide (κ-Ga2O3).
• To demonstrate in-sensor information processing, including authentication and artificial neural network (ANN) inference.
• To explore the potential of κ-Ga2O3 for compact, intelligent, and privacy-enhancing edge systems.

Main Methods:

• Fabrication of metal-semiconductor-metal photodetectors based on κ-Ga2O3 exhibiting persistent photoconductivity (PPC).
• Exploitation of device-distinguishable photocurrent responses for hardware-level authentication using a 1D deep embedding network.
• Evaluation of neuromorphic inference using a hardware-aware simulation framework, mapping experimental conductance states to a quantization-aware trained ANN.
• Conversion of the ANN to a leaky integrate-and-fire (LIF) spiking neural network (SNN) for device-constrained operation.

Main Results:

• The κ-Ga2O3 photodetectors demonstrated reproducible yet device-distinguishable temporal photocurrent responses due to the pronounced PPC effect.
• Hardware-level authentication achieved an Area Under the Curve (AUC) of approximately 0.97 and an Equal Error Rate (EER) of about 9%.
• The quantized ANN achieved 98.17% accuracy, and the converted LIF SNN retained 96.80% accuracy.

Conclusions:

• The developed κ-Ga2O3 synaptic photodetectors enable integrated sensing, authentication, and inference at the device level.
• This materials-enabled approach offers a pathway toward compact, intelligent, and privacy-enhancing optoelectronic hardware for edge systems.
• The study highlights the potential of persistent photoconductivity in κ-Ga2O3 for advanced neuromorphic computing applications.