🔬 applied physics

SOMA: A Single-Material Organic Multivibrator Adaptive Neuron for Fully Integrated PEDOT:PSS Neuromorphic Systems

This paper presents SOMA, a voltage-driven, single-material PEDOT:PSS organic multivibrator neuron that enables fully integrated, scalable, and adaptable neuromorphic systems with rich dynamics and compatibility for on-chip synaptic integration.

Original authors: Nikita Prudnikov, Hans Kleemann

Published 2026-03-24

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Nikita Prudnikov, Hans Kleemann

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are trying to build a computer that thinks more like a human brain and less like a calculator. Current computers are incredibly fast at crunching numbers, but they are terrible at handling messy, real-world data (like recognizing a face in a crowd or reacting to a sudden noise) without using massive amounts of electricity.

Scientists are trying to build "neuromorphic" computers—machines that mimic the brain's structure. The brain works using neurons (cells that fire electrical signals called "spikes") and synapses (connections between them). The challenge is building artificial neurons that are cheap, flexible, and can talk to living tissue.

This paper introduces a new invention called SOMA (Single-Material Organic Multivibrator Adaptive Neuron). Here is the story of how it works, explained simply.

1. The Problem: The "Two-Polymer" Mess

Previously, scientists tried to build these artificial neurons using organic materials (plastics that conduct electricity). However, most designs were like trying to build a house using two different types of bricks that don't stick together well. They needed two different materials to make the transistors work, which made the manufacturing process complicated, slow, and prone to errors. It was like trying to bake a cake where you have to switch ovens halfway through.

2. The Solution: The "One-Brick" Wonder

The team at TU Dresden decided to use just one material for everything: a special conductive plastic called PEDOT:PSS.

The Analogy: Imagine building a Lego castle where every single piece is the exact same red brick. You don't need to sort through boxes of different shapes; you just snap them together.
Why it matters: PEDOT:PSS is naturally sensitive to ions (charged particles), just like the fluids in your body. This makes it perfect for talking to biological systems. Because they only use one material, the manufacturing process is simple, cheap, and scalable.

3. How the SOMA Neuron Works: The "Flashing Light"

The core of this neuron is a circuit called a Multivibrator.

The Analogy: Think of a classic "flip-flop" toy or a blinking Christmas light. It has two states: ON and OFF. When you push a button, it flips. If you leave it alone, it keeps flipping back and forth on its own.
The Innovation: The SOMA neuron is a "smart" blinking light. It has two controls:
1. The "Sensitivity Knob" (Control Input): This sets how easily the neuron gets excited. Turn it up, and the neuron is hyper-alert; turn it down, and it's lazy.
2. The "Trigger" (Spike Input): This is the signal from other neurons.

When you give it a trigger, it doesn't just blink once. It can go into a burst of rapid blinking.

Encoding Information: The neuron can send messages in two ways:
- Latency Coding: "How fast do I blink?" (If the signal is strong, I blink immediately. If it's weak, I wait a bit.)
- Burst Length Coding: "How long do I keep blinking?" (If the signal is strong, I blink for a long time. If it's weak, I blink just once.)

4. The "Inhibitory" Friend: Learning to Say "No"

Brains aren't just about excitement; they need to know when to stop. The team built a small network with two neurons: one that says "Go!" (Excitatory) and one that says "Stop!" (Inhibitory).

The Analogy: Imagine a party. The Excitatory neuron is the DJ playing loud music to get people dancing. The Inhibitory neuron is the bouncer.
The Result: If the bouncer (Inhibitory neuron) arrives before or during the music, the party is shut down, and no one dances. If the bouncer arrives after the party is already over, it doesn't matter. This timing is crucial for the brain to process information correctly, and the SOMA neuron handles this perfectly.

5. Short-Term Memory: The "Hangover" Effect

One of the coolest features is that these neurons have a form of short-term memory.

The Analogy: Imagine you just finished a heavy workout. If someone asks you to run a sprint immediately after, you might be slower or less responsive because your body is still recovering.
The Science: Because the plastic material relies on ions moving around (which is slower than electricity moving through copper), the neuron "remembers" recent activity. If it just fired a huge burst of signals, it takes a moment to "cool down" before it can fire again. This helps the network filter out noise and focus on important patterns.

6. Growing "Wires" on the Chip

Finally, the team showed that they can grow tiny, tree-like conductive structures (dendrites) directly on the chip using electricity.

The Analogy: Imagine if you could grow the wires connecting your computer parts after you built the computer, just by sending electricity through them.
Why it matters: This mimics how brains grow new connections when you learn something new ("fire together, wire together"). It means this system could physically rewire itself to learn, all on a single piece of glass.

The Big Picture

This paper presents a simpler, cheaper, and more flexible way to build brain-like computers. By using just one material (PEDOT:PSS) and a clever circuit design, they created a neuron that can:

Talk to biological tissue.
Encode information in time and duration.
Work with inhibitory signals to stop chaos.
Remember recent events.
Grow its own connections.

It's a major step toward creating smart, low-power devices that can live alongside us, perhaps helping to treat diseases, monitor our health, or act as the "brain" for soft, flexible robots.

1. Problem Statement

Current neuromorphic systems face significant challenges in bridging the gap between electronic processing and biological interfaces, particularly for real-time and edge-computing applications.

Material Complexity: Existing Organic Electrochemical Transistor (OECT) neuron implementations often rely on multiple distinct polymers or ambipolar materials. This complicates fabrication (requiring sequential deposition steps that risk damaging previous layers) and hinders scalability due to a lack of standardized materials.
Operational Limitations: Most reported OECT neurons rely on current-driven operation, which necessitates precise matching of operating currents and bias voltages, making large-scale integration and neuron-to-neuron coupling difficult.
Lack of Adaptability: Many designs lack dynamic modulation mechanisms, limiting their ability to emulate the rich, adaptive dynamics of biological neurons (e.g., variable spiking patterns, short-term memory).
PEDOT:PSS Challenge: While Poly(3,4-ethylenedioxythiophene):polystyrene sulfonate (PEDOT:PSS) is the most commercially available and stable p-type OECT material, its "normally-on" behavior (inherently doped) makes direct translation of standard CMOS-inspired circuit designs impractical.

2. Methodology

The authors propose SOMA (Single-Material Organic Multivibrator Adaptive neuron), a voltage-driven spiking neuron circuit fabricated entirely from PEDOT:PSS.

Circuit Architecture:
- Core: An astable multivibrator (MV) architecture consisting of two inverters connected via an RC feedback loop. Each inverter comprises a single PEDOT:PSS OECT and a 100 kΩ resistor.
- Inputs: The circuit features two distinct inputs:
  1. $V_{control}$ : A constant voltage bias that modulates the neuron's overall responsiveness and sensitivity (analogous to ionic channel density).
  2. $V_{spike}$ : A pulsed input receiving signals from other neurons (analogous to synaptic input strength).
- Spike Transformer: A secondary functional block that converts the MV output into positive voltage pulses compatible with the input, ensuring voltage-level matching and decoupling connected neurons to prevent synchronous oscillation.
Fabrication Process:
- A two-step process on a glass substrate: (1) Spin-coating and photolithographic patterning of PEDOT:PSS for active components (OECTs) and passive components (resistors and capacitors); (2) Inkjet printing of a solid-state electrolyte.
- Single-Material Integration: All active and passive components, including capacitors (interdigitated PEDOT:PSS electrodes) and resistors (patterned PEDOT:PSS lines), are made from PEDOT:PSS.
Simulation: SPICE simulations using a standard PMOS model were employed to validate circuit behavior and optimize resistor ratios.

3. Key Contributions

First Fully Integrated PEDOT:PSS Neuron: Demonstrated a complete spiking neuron circuit using only one material (PEDOT:PSS) for all active and passive elements, eliminating the need for heterogeneous material integration.
Voltage-Driven Operation: Shifted from complex current-driven designs to a robust voltage-driven multivibrator architecture, simplifying network coupling and biasing.
Dual Encoding Schemes: The neuron supports two distinct temporal information encoding modes, selectable via operating voltage:
1. Latency Encoding: The time delay to the first spike correlates with stimulus strength.
2. Burst Length Encoding: The number of spikes in a burst correlates with stimulus intensity.
Inhibitory Functionality: The same circuit architecture was adapted to function as an inhibitory neuron by adjusting the input bias, enabling the construction of excitatory-inhibitory pairs.
Synaptic Compatibility: Demonstrated compatibility with electropolymerized PEDOT:PF₆ dendritic structures, allowing for on-chip growth of synaptic interconnects ("fire together, wire together").

4. Key Results

Dynamic Response: The SOMA neuron exhibits rich dynamics including self-sustained oscillations, stimulus-driven bursting, and tunable excitability.
- Latency: Increasing $V_{spike}$ amplitude reduces the time-to-first-spike.
- Burst Length: Higher $V_{spike}$ amplitudes (at specific $V_{control}$ levels) generate longer spike trains.
- Modes: The circuit can operate in suppression, propagation, or amplification modes based on the combination of $V_{control}$ and $V_{spike}$ .
Inhibitory Coupling: In a two-neuron unit (one inhibitory, one readout), the inhibitory neuron successfully suppressed the readout neuron's activity. The degree of suppression was dependent on the time delay ( $\Delta t$ ) between excitatory and inhibitory inputs, transitioning from amplification to full suppression as the delay decreased.
Short-Term Memory: The neuron demonstrated intrinsic short-term memory (seconds scale). After a period of strong stimulation, the neuron's response to subsequent weak stimuli was gradually suppressed and eventually silenced. This is attributed to the ionic time constants of the OECTs (transient ionic redistribution) rather than circuit design, mimicking biological adaptation.
Scalability: A 25×25 mm² glass substrate successfully hosted 63 fully integrated neurons. The effective footprint per neuron is approximately 2 mm², indicating high potential for dense integration.
Dendrite Integration: Electropolymerized PEDOT:PF₆ dendrites were successfully grown on the same substrate, bridging gaps between neurons and modulating conductivity via electrical stimulation, serving as a physical basis for synaptic plasticity.

5. Significance

Scalability and Accessibility: By utilizing a single, commercially available material (PEDOT:PSS) and a simple two-step fabrication process, the SOMA neuron lowers the barrier to entry for organic neuromorphic research and facilitates large-scale manufacturing.
Bio-Interface Compatibility: The use of OECTs provides inherent ionic-electronic coupling, making the system naturally compatible with biological environments (soft, aqueous, biocompatible), which is crucial for brain-machine interfaces and bio-hybrid systems.
Functional Versatility: The ability to switch between encoding modes and reconfigure the same hardware as excitatory or inhibitory neurons offers a flexible platform for building complex Spiking Neural Networks (SNNs).
Future Pathway: The successful integration of dendritic growth suggests a roadmap for fully adaptive, self-organizing neuromorphic systems where synaptic connections can be formed and tuned in situ, moving closer to the complexity of biological neural networks.

In summary, the SOMA neuron represents a significant step forward in organic neuromorphic electronics, offering a scalable, single-material, and biologically inspired solution for energy-efficient, real-time data processing.