(Nanowerk Spotlight) Scientists have developed a new type of electronic component that demonstrates learning and memory functions like the human brain. The novel “photoferroelectric synapse” device, created by an international team of researchers from China, mimics the biological connections between neurons that allow the nervous system to acquire and store knowledge.
The work, published in Advanced Functional Materials (“Photoferroelectric Perovskite Synapses for Neuromorphic Computing”), offers a potential pathway for building intelligent machines and computers that learn from experience. It also provides insights into how synapses operate in the brain.
a) Schematic diagram of one-step synthesis of perovskite films with P(VDF-TrFE)-containing anti-solvents. b) Schematic configuration of PFEP device, showing PFEP/Spiro-OMeTAD between the top Au electrode and the bottom ITO electrode. (Reprinted with permission by Wiley-VCH Verlag)
Neuromorphic engineering is an emerging field that seeks to replicate brain functions in electronic circuits and devices. A core goal is to develop artificial neural networks composed of synthetic neurons and synapses that demonstrate the adaptability and low power consumption of biological systems. This could lead to transformative computing technologies like self-driving cars, voice assistants, and disease-diagnosing AI that operate more like brains than traditional computers.
A major challenge is creating artificial synapses – the junctions between neurons where signals are transmitted – that display the plasticity seen in biology. Strengthening and weakening of synaptic connections, known as synaptic plasticity, allows new memories to form and old ones to fade in the brain. One approach to achieve this is using multifunctional materials that combine properties like photosensitivity, magnetism, and ferroelectricity (the ability to flip electric polarizations) to mimic nervous system dynamics.
The new photoferroelectric synapse consists of a thin film made from a lead-based perovskite material – a class of materials gaining popularity in solar cells – combined with an organic ferroelectric polymer called P(VDF-TrFE). Perovskites have shown excellent photoelectric properties but typically contain lead, which raises environmental and health concerns. The researchers used a low lead concentration and thin film form to minimize exposure. However, for real-world applications, further work would be needed to replace lead with a non-toxic metal while retaining the multifunctionality demonstrated here.
The perovskite provides photoelectric behavior, generating electric current when exposed to light, while the polymer brings ferroelectricity, meaning it can be electrically polarized, with positive and negative poles analogous to tiny bar magnets that can be flipped by applying voltage.
This combination of properties allows the synapse to mimic several types of synaptic plasticity observed in biological synapses, including paired-pulse facilitation/depression and spike timing-dependent plasticity. These effects tune the strength of connections between neurons in response to patterns of neural activity and are believed to be the cellular basis for learning.
A key innovation was the discovery that polarizing the device – ‘training’ it by applying voltage – both strengthens the material’s ferroelectric properties and enhances its photoelectric response. The researchers showed this photoferroelectric effect could be leveraged to simulate Pavlovian conditioning: the famous experiments where dogs learned to associate a ringing bell with being fed. This demonstration of learning was achieved in the device by combining light and voltage pulses.
According to the authors, the work provides a new paradigm for multifunctional materials for neuromorphic computing and nerve-inspired electronics. The photoferroelectric synapses could pave the way for low-power, self-learning computing systems that operate more like brains than traditional computers, leading to transformative technologies such as self-driving cars, voice assistants, and disease-diagnosing AI. Neuromorphic systems mimicking biology could be orders of magnitude more energy efficient than conventional AI hardware, allowing advanced intelligence on small devices.
By further understanding synaptic function, the research may also shed light on mysteries of biological intelligence.
Get our Nanotechnology Spotlight updates to your inbox!
Thank you!
You have successfully joined our subscriber list.
Become a Spotlight guest author! Join our large and growing group of guest contributors. Have you just published a scientific paper or have other exciting developments to share with the nanotechnology community? Here is how to publish on nanowerk.com.
