Summary: As artificial intelligence energy demand is expected to surge, researchers are developing hardware inspired by the human brain’s remarkable energy efficiency. This study focuses on neuromorphic computing: a design approach that merges memory and processing in the same physical location—similar to biological synapses—to dramatically reduce power consumption.
Using organic transistors engineered to emulate synaptic behavior, the team at the University of Missouri is building foundational components for AI systems that can perform complex tasks while using a fraction of the energy required by conventional silicon chips.
Key Research Findings
- The efficiency gap: Modern data centers consume vast amounts of electricity, whereas the human brain completes sophisticated computations on roughly 20 watts of power—an energy target that neuromorphic hardware aims to approach.
- Synaptic architecture: Standard chips separate memory and processing, forcing constant data transfer and high energy use. Mizzou’s work develops organic synaptic transistors that store and process information at the same site to remove this bottleneck.
- Importance of the interface: Device performance depends less on bulk material alone and more on the thin boundary where semiconductor and insulator meet. That interface governs charge trapping, dynamics, and learning behavior.
- Molecular design matters: Materials that appear similar in bulk can behave very differently when integrated into devices. Minor structural differences at the molecular level produced large changes in how a synaptic transistor learns and adapts.
- Targeted AI tasks: This neuromorphic hardware is tailored to excel at energy-sensitive applications such as pattern recognition and decision-making, where reduced power draw and on-device learning are most valuable.
Source: University of Missouri-Columbia
As conventional microelectronics approach physical limits and AI workloads grow, researchers at the University of Missouri are redesigning computing hardware with inspiration from the brain.
The timing of this effort is urgent: energy use tied to AI processing is projected to rise sharply during the coming decade, creating an immediate need for more sustainable architectures.
Neuromorphic computing rethinks hardware so it behaves more like biological neural networks instead of conventional Von Neumann systems. The core idea is to allow a single device to both hold and operate on information—an approach that drastically reduces the energy cost of moving data back and forth.
“One of the brain’s greatest advantages is its efficiency,” said Suchi Guha, professor of physics at Mizzou’s College of Arts and Science. “It performs extraordinarily complex operations on roughly 20 watts of power. In contrast, today’s computing architectures are far more energy intensive.”
To realize brain-like computing, Guha and colleagues are designing electronic components that mimic synapses—the neural connections responsible for learning, adaptation and memory. Their strategy focuses on organic ferroelectric transistors that can emulate synaptic plasticity while remaining compatible with low-cost fabrication approaches.
Rethinking the computer chip
Traditional chips use transistors as switches, but they keep computation and storage separate. That separation forces repeated data transfers between processor and memory, a major source of latency and energy consumption. The brain’s synapses overcome this by combining storage and computation locally, allowing continuous adaptation with minimal energy.
The Mizzou team built and tested several organic semiconducting materials in synaptic transistor configurations. Although some materials looked nearly identical under conventional characterization, their behavior in devices diverged significantly. The decisive factor was the interfacial region where the semiconductor contacts the dielectric.
“Performance isn’t defined solely by the material’s chemical formula,” Guha explained. “It’s also determined by how that material interacts with neighboring layers and the microscopic structure of the boundary. Small differences there can radically change synaptic response.”
Moving toward energy-efficient, brain-like AI
By mapping how molecular design and interface quality control synaptic function, this work supplies practical guidelines for developing more capable neuromorphic hardware. Devices that integrate storage and processing could enable AI systems that learn on-device, consume much less power, and perform tasks like image recognition and decision-making with high efficiency.
Although neuromorphic computing is an evolving field, advances such as these close the gap between biological and artificial computation. “The brain remains the gold standard for energy-efficient intelligence,” Guha said. “To achieve truly intelligent machines, we must start with hardware that learns and adapts as biology does.”
The study, titled “Structure–Function Coupling in Pyridyl Triazole Copolymers for Neuromorphic Synaptic Transistors,” appears in ACS Applied Electronic Materials.
Co-authors include Arash Ghobadi, Abhijeet Abhi, Thomas Kallos, Dillan Gamachchi, Indeewari Karunarathne, Andrew Meng, Joseph Mathai, Shubhra Gangopadhyay and Steven Kelley at Mizzou, with Salahuddin Attar and Mohammed Al-Hashimi at Hamad Bin Khalifa University.
Key Questions Answered:
A: The interface is where signals are exchanged and charge traps form; it acts like the conversation between two components. Even if two materials are chemically similar, variations at that microscopic boundary can change how charges move and are retained, which directly impacts the device’s ability to mimic learning.
A: Potentially. By eliminating the constant transfer of data between separate memory and processing units—the Von Neumann bottleneck—neuromorphic devices could reduce energy consumption across many computing tasks, improving battery life for portable devices over time.
A: No. Organic semiconductors are selected because molecular design and flexibility allow researchers to tune charge transport and interfacial properties in ways that better emulate biological synapses—capabilities that are harder to achieve with rigid, traditional silicon.
Editorial Notes:
- This article was edited by a Neuroscience News editor.
- The full journal paper was reviewed as part of reporting.
- Additional context and explanations were added by the editorial staff.
About this neuromorphic computing and AI research news
Author: Eric Stann
Source: University of Missouri-Columbia
Contact: Eric Stann – University of Missouri-Columbia
Image: The image is credited to Neuroscience News
Original Research: Open access. “Structure–Function Coupling in Pyridyl Triazole Copolymers for Neuromorphic Synaptic Transistors” by Arash Ghobadi, Salahuddin Attar, Abhijeet Abhi, Thomas B. Kallaos, Dilan M. Gamachchi, Indeewari M. Karunarathne, Andrew C. Meng, Joseph C. Mathai, Shubhra Gangopadhyay, Steven P. Kelley, Mohammed Al-Hashimi, and Suchismita Guha. DOI: 10.1021/acsaelm.5c02633
Abstract
Structure–Function Coupling in Pyridyl Triazole Copolymers for Neuromorphic Synaptic Transistors
Organic ferroelectric transistors are promising, low-cost candidates for synaptic devices in neuromorphic systems. Interfaces that combine donor–acceptor semiconducting polymers with poly(vinylidene fluoride) (PVDF) copolymers as the dielectric show particular promise for reproducing synaptic responses.
The research synthesizes three pyridyl triazole (PyTr)–based copolymers by varying the linking unit between PyTr acceptors and thiophene-based donors. The linkers include a selenium-substituted thiophene, a benzothiadiazole unit, and a fluorine-substituted thiophene.
When paired with the hexafluoropropylene copolymer of PVDF (PVDF-HFP) as the dielectric, these PyTr semiconductors function as p-type materials in transistor structures and display carrier mobilities in the range of 0.1–0.2 cm2 V–1 s–1.
Synaptic plasticity was probed by applying long-term voltage pulse protocols at the gate to emulate potentiation and depression. The devices’ synaptic responses were then evaluated for image recognition performance using a multilayer perceptron neural network.
The copolymer featuring the benzothiadiazole linker reached recognition accuracy near 80%, while the fluorine-substituted thiophene linker failed to show synaptic behavior, underscoring the decisive influence of the semiconductor–dielectric interface.
A detailed analysis of interface trap density and morphological characteristics was performed to trace how interfacial properties directly govern synaptic device performance, offering practical design rules for future neuromorphic materials and devices.