Exploring Artificial Nanophotonic Nanowire-Based Neuron

Sammanfattning

The combination of photonics and nano-structures offers a multitude of solutions for artificial neural networks. Light is not only fast, but reduces the clutter and energy losses associated with wired connections. Energy demand is further reduced by implementing nano-components, which decreases the size of the footprint. Researchers are developing a nano-chip that combines sensing and neural computations to complete homing navigation, drawing inspiration from the insect's brain. This lightweight and energy efficient device would allow tiny robots or sensors to navigate without relying on external infrastructure like satellites. The artificial neurons of the device consist of III-V semiconducting nanowires that receive overlapping light signals inside a 2D waveguide from external sources and from neighboring neurons. Bees rely on the sky's polarization by aligning internal photodetectors to the light's electric field to track its heading. The device mimics this functionality by employing polarization sensitive nanowires. To enhance this property, nanowires can be coupled to two metal nano-structures in a bowtie configuration that exhibit plasmonic resonances. When the electric field is aligned along the axis of this bowtie at resonant wavelengths, the localized electric field is enhanced. This paper demonstrates this optical enhancement of the nanowire and nano-antenna system. InP nanowires are assembled with gold nano-antennas. To quantify the effect of the enhancement, the photocurrent is measured through the nanowires with excitation wavelengths between 430 nm and 700 nm. The results show a clear amplification of the photocurrent for the nanowire/nano-antenna device in comparison to the single nanowires. The report concludes with recommended actions to further optimize the device for implementation.