(Nanowerk Spotlight) Attempts to engineer artificial retinas with the visual capabilities of the human eye have faced enduring challenges. Though promising in theory, replicated the elaborate structure and energy efficiency of natural vision has not proven straightforward in practice. Past efforts along these lines have fallen short on several key attributes.
Now, drawing upon recent materials innovations, researchers from Sungkyunkwan University in Korea report substantial progress on this persisting problem with their development of flexible, fiber-shaped artificial retina closely exhibiting properties of its biological analog. Leveraging recent breakthroughs in soft electronics and organic-inorganic materials, their photonic device marks major progress toward durable, low-power artificial vision.
The team reports their findings in Advanced Functional Materials ("Bio-Inspired Artificial Retinas Based on a Fibrous Inorganic–Organic Heterostructure for Neuromorphic Vision").
The new research centers on constructing an artificial synaptic device that replicates the form and functions of neurons in the human retina. The retina lines the inside of the eye and contains photoreceptor cells that detect light. These cells connect synaptically to downstream neurons in layers that preprocess visual stimuli into encoded signals. Ganglion cells then transmit these signals via long nerve fibers bundled into the optic nerve for further analysis in the brain’s visual cortex. Emulating this intricate biological process demands advanced materials and careful engineering.
Neuronal structures of retinas and fibrous photonic artificial synapses. a) Schematics of (i) a human eye, (ii) the structure and function of optic neurons in a retina, and (iii) a biological synapse. b) Schematics of (i) the structure of a FPAS array, (ii) the structure and function of each layer in a single FPAS, and (iii) an organic–inorganic heterojunction in an FPAS. c) A schematic of the comparative structure and function of the human retina and FPAS. The light-detection layers (ZnO NRs) correspond to photoreceptor cells. The depletion layer, an intermediate layer formed by a ZnO NRS/PEDOT:PSS heterojunction, is similar to a bipolar cell, which is an intermediate neuron. The PEDOT:PSS, an output layer, is similar to a ganglion cell, which is an output neuron. Gold electrodes on PU fiber correspond to optic nerve fibers. The postsynaptic current, IPSC, is read by the FPAS, and the postsynaptic signal is read by an optic nerve fiber. (Reprinted with permission from Wiley-VCH Verlag)
Previous efforts at engineering artificial retinas have incorporated rigid planar structures unsuitable for integration with the curved rear interior of the eyeball. Flexible devices also faced limitations. For instance, earlier soft electronics still relied on standard field effect transistor architectures that consume substantial power during operation. This contrasted starkly with the retina’s innate ability to convert light into neural signals without an external energy source. Furthermore, mimicking the bundled fibers forming the optic nerve has remained an ongoing difficulty.
The new artificial retina addresses these persistent challenges through an innovative composition and design. It creates a vertical heterostructure by layering a sunlight-responsive zinc oxide nanorod array onto a conductive polymer layer, poly(3,4-ethylenedioxythiophene) polystyrene sulfonate (PEDOT:PSS). This structure sits not on a flat surface but rather wraps around the circumference of a bendable polyurethane fiber coated with gold electrodes.
Remarkably, when exposed to UV light, this fiber-shaped device can modulate its electrical conductivity to emulate essential functions of retinal neurons and synapses without needing any external power. It works by tweaking the capture and release of photon-generated charge carriers at the junction between materials. In the dark, oxygen molecules absorb onto surfaces of the zinc oxide nanorods, capturing free electrons. This makes the nanorods more positively charged, analogous to sodium ions entering biological photoreceptor cells. In contrast, UV exposure generates electron-hole pairs inside the nanorods, fundamentally shifting their electrical state closer to that of retinal cells exposed to visible light.
The research team demonstrated how this mechanism allows their artificial retina fiber to exhibit short-term plasticity and long-term potentiation like a natural synapse when subjected to varying optical stimuli. Their photonic device could also replicate biological functions including paired pulse facilitation—the enhanced response to closely spaced pulses—and spike timing dependent plasticity tuned to pulse duration, intensity and frequency.
Remarkably, the artificial retinal fiber maintained key synaptic characteristics even while severely bent, twisted into coils around tubing, or woven into fabric. Its robustness arises from the soft, durable materials enabling the construction of electronic components on the pliable, fiber-shaped platform. The reliability paves the way for integrating the artificial retina into wearable technologies.
To assess the perceptual capabilities enabled by their synapse design, the researchers manufactured a 3x3 grid of artificial retinal fibers on a flexible substrate. They found the array could successfully detect and memorize visual patterns projected onto it using stencils and ultraviolet light. The stored synaptic strength of these images gradually decayed over time in another striking biomimicry of human memory processes. By monitoring the post-exposure electrical conductance of individual fibers, the imprinted letters remained detectable up to a minute after initial illumination.
The researchers also showed how reading out the array’s synaptic responses could serve as input data for machine learning software to accurately deduce the distance of a UV light source. This demonstration only tapped a small fraction of the potential visual processing power inherent to the artificial retinal grid. It exemplifies how combining next-generation hardware mimicking biological perception with AI algorithms can enable transformative sensory capabilities.
The new artificial retina does face some limitations in its reliance on UV light rather than visible wavelengths detectable by the human eye. However, the fundamental concept could be expanded to visible responsiveness by integrating appropriate photon absorbing nanomaterials with the base zinc oxide nanorod technology.
More pressingly, by closely replicating the functions, connectivity, and durability of natural retinal neurons in a flexible fiber form factor, this advance provides an extraordinarily well-suited platform for further enhancing synthetic vision. Optimizing bio-inspired artificial retinas leveraging similar principles and materials could soon enable wearable visual augmentation, advanced robot sensors, or even prosthetic technologies.
