Threshold plasticity of SOI-GST microring resonators.

Area of Science:

• Neuromorphic Engineering
• Materials Science
• Computational Neuroscience

Background:

• Spiking Neural Networks (SNNs), or third-generation Artificial Neural Networks, offer biological interpretability and hardware efficiency.
• Beyond synaptic plasticity, threshold plasticity offers an alternative unsupervised learning mechanism, exemplified by the Bienenstock, Cooper, and Munro rule.
• This plasticity rule links neuronal thresholds to post-synaptic neuron output.

Purpose of the Study:

• To investigate the threshold characteristics of a microring resonator integrated with Ge2Sb2Te5.
• To explore the device's potential for implementing threshold plasticity in neuromorphic systems.
• To demonstrate how material phase transitions influence the resonator's learning behavior.

Main Methods:

• Simulations were conducted to analyze the nonlinear optical effects within a microring resonator.
• The Ge2Sb2Te5 material's refractive index and attenuation were modeled under varying conditions.
• The impact of wavelength detuning and material phase transitions (amorphous to crystalline) on threshold power was examined.

Main Results:

• The microring resonator exhibited complex dependencies of threshold characteristics on intracavity refractive index, attenuation, and wavelength detuning.
• Class II excitability was observed in the device's nonlinear response.
• The threshold power was successfully modified by altering the Ge2Sb2Te5 refractive index and loss through phase transitions.

Conclusions:

• The simulated microring resonator device shows promise for implementing unsupervised learning via threshold plasticity.
• The ability to tune the threshold power through material phase changes offers a novel approach for neuromorphic hardware.
• The device demonstrates both excitatory and inhibitory learning capabilities, adaptable by modifying material properties.