⚡ electrical engineering

Nanofluidic memristor with an ultra-low operating voltage

This study introduces a Cu–PAF/CuSO₄–Cu nanofluidic memristor that achieves ultra-low voltage (sub-100 mV) switching and energy-efficient synaptic emulation by optimizing electrode-electrolyte pairs and utilizing reversible ion adsorption within nanochannels to overcome previous voltage limitations in neuromorphic computing.

Original authors: Yike Xiao, Yukun Bao, Tianyong Peng, Kangrui Li, Pingyuan Yan, Wei Huang, Chuanxiang Sheng, Qi Wang, Ye Feng Yu

Published 2026-07-10

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Yike Xiao, Yukun Bao, Tianyong Peng, Kangrui Li, Pingyuan Yan, Wei Huang, Chuanxiang Sheng, Qi Wang, Ye Feng Yu

Original paper licensed under CC BY 4.0 (https://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 your brain as a bustling city where billions of tiny messengers (neurons) pass notes to each other. To keep the city running smoothly, these messengers need a way to remember recent conversations without burning out the power grid. In the world of computers, we've been trying to build "artificial synapses" (memory switches) to copy this brain magic, but most of them are like old, clunky lightbulbs: they need a huge jolt of electricity just to flicker on.

A team of researchers from Nanjing University of Science and Technology and their colleagues has built a new kind of switch that is more like a whisper than a shout. They call it a nanofluidic memristor, and it works at an ultra-low voltage that is surprisingly close to how real biological neurons operate.

The "Whispering" Switch

Think of a standard computer switch as a heavy gate that requires a massive team of people to push it open. This new device, however, is like a delicate, self-closing door that only needs a gentle breeze to move.

The secret to this magic lies in two clever tricks:

The Perfect Match: The team carefully picked a specific "dance partner" for their system: copper electrodes and a copper sulfate liquid. In the world of chemistry, these two are so comfortable with each other that they barely need any extra push to start moving. This minimizes the "electromotive force," which is basically the internal resistance the system fights against.
The Spongey Tunnel: Between the copper electrodes sits a special membrane made of porous anodic alumina (PAF). Imagine this membrane as a tunnel filled with millions of tiny, funnel-shaped pores. These pores act like sponges that can grab onto copper ions (charged particles) and let them go again, very quickly.

How It "Remembers"

When you apply a tiny voltage (as low as 20 mV, which is less than one-hundredth of a volt!), copper ions start to swim through the liquid. As they swim through the tiny tunnels, they stick to the walls of the pores (adsorption) and then peel off (desorption).

This sticking and peeling changes how easily electricity flows, creating a "memory" effect.

The Hysteresis Loop: If you plot the current on a graph, it doesn't draw a simple straight line. Instead, it draws a pinched loop. Think of it like a rubber band: if you stretch it (apply voltage), it snaps back, but not exactly to where it started. It remembers how far you stretched it.
The Sweet Spot: The researchers found that if they push too hard (above 100 mV), the ions rush through too fast, and the "sponge" effect gets overwhelmed, ruining the memory. But if they stay in the 20 mV to 50 mV range, the device works perfectly, showing a strong "pinched hysteresis loop" that proves it's acting like a true memory switch.

Mimicking the Brain's Short-Term Memory

Real synapses in your brain have something called "short-term plasticity." This means if you send two signals close together, the connection gets stronger (like a path getting worn down by footsteps) or weaker (like a path getting overgrown).

The team tested their device with pairs of voltage pulses:

Paired-Pulse Facilitation (PPF): When they sent two quick pulses, the second one got a bigger response, just like a brain synapse getting excited.
Paired-Pulse Depression (PPD): Under different conditions, the response got weaker, mimicking a tired synapse.

They even saw a "non-monotonic" effect, where the response went up and then down in a specific pattern, similar to what happens in a frog's optic tectum. The device could hold this memory for about 500 microseconds before fading away, which is exactly the kind of short-term memory needed for processing fast information.

The Energy Breakthrough

Here is the most exciting part: energy.

Old devices: Often need hundreds or thousands of millivolts to work.
This device: Works at 20 mV.
Energy cost: Each "spike" of activity costs only 100 femtojoules (fJ). To put that in perspective, that is a tiny fraction of the energy used by a single photon of light.

The authors measured this directly, calculating that a group of five pulses only used about 500 fJ in total. This is a massive drop compared to existing nanofluidic devices, which usually require much higher voltages to overcome the energy barriers at the electrode surface.

What This Means (and What It Doesn't)

The paper suggests that by carefully choosing materials that naturally want to react with each other (like copper and copper sulfate) and using a membrane that can grab and release ions, we can build computer parts that run on the same tiny energy scale as our brains.

However, the authors are careful to note that this is a specific demonstration. They showed it works with copper and copper sulfate, and they suggest other pairs like iron or silver might work too, but they haven't tested those yet. They also showed the device works consistently across different batches and sizes (from millimeters down to micrometers), proving it's scalable, but this is still a lab experiment, not a product you can buy yet.

In short, this research offers a viable pathway toward energy-efficient computing. It proves that we don't need to force ions to move with a sledgehammer; sometimes, a gentle nudge and a good sponge are all you need to build a brain-like computer.

Technical Summary: Nanofluidic Memristor with an Ultra-Low Operating Voltage

Problem Statement
Nanofluidic memristors, which utilize ion transport for information processing, are promising candidates for neuromorphic computing due to their biological compatibility and high information capacity. However, their practical application has been hindered by high operating voltages, typically ranging from hundreds to thousands of millivolts. This requirement stems from the need to overcome the electromotive force (EMF) and overpotential at the electrode-electrolyte interface to initiate Faradaic reactions. In contrast, biological neurons operate with action potentials in the range of tens of millivolts (-70 mV to +35 mV) and achieve ultra-low energy consumption. Existing nanofluidic devices fail to match this biological efficiency, creating a significant gap between artificial and biological synaptic performance.

Methodology and Framework
To address this challenge, the authors propose a design framework centered on two primary mechanisms:

Minimization of Electromotive Force (EMF): By carefully selecting electrode-electrolyte pairs, the system aims to drive the equilibrium cell potential ( $E_{cell}$ ) toward zero. This allows electrochemical reactions to be initiated with minimal external voltage, confining the operating range within the system's overpotential ( $E_o$ ).
Reversible Ion Adsorption/Desorption in Asymmetric Nanochannels: The device incorporates an intermediate membrane with asymmetric nanochannels. Memristive behavior is generated through the reversible adsorption and desorption of ions on the channel surfaces, coupled with the Ion Current Rectification (ICR) effect. Theoretical analysis suggests that memristive behavior is observable when the number of available adsorption sites ( $N_1$ ) significantly exceeds the number of ions passing through the channel ( $N_2$ ), a condition maintained by operating below the threshold where bulk ion generation dominates.

Device Implementation
Based on this framework, the authors fabricated a Cu–PAF/CuSO₄–Cu nanofluidic memristor (CNM):

Electrode-Electrolyte System: A Copper (Cu) electrode paired with a Copper Sulfate (CuSO₄) electrolyte was chosen due to the low equilibrium potential of the Cu²⁺/Cu redox couple ( $E^0 \approx +0.34$ V vs. SHE).
Membrane: A Porous Anodic Alumina Oxide (PAF) film served as the intermediate membrane. PAF was selected for its high surface-to-volume ratio, abundant surface hydroxyl groups, and proven ability to reversibly adsorb heavy metal ions like Cu²⁺.
Fabrication: The device consists of a PAF film embedded in a Polydimethylsiloxane (PDMS) reservoir filled with 1 M CuSO₄ solution, sandwiched between two copper electrodes.

Key Results

Ultra-Low Operating Voltage: The device successfully operates within a voltage window of 20 mV to 50 mV, with characterization performed at 20 mV. This is an order of magnitude lower than conventional nanofluidic memristors. At 10 mV, the device exhibited linear ohmic behavior, while 20–50 mV induced pronounced bipolar memristive behavior with a pinched hysteresis loop.
Mechanism Verification: The memristive switching is attributed to the adsorption and desorption of copper ions on the PAF surface.
- Negative Bias: Cu²⁺ ions enter the nanochannels and adsorb, increasing conductance.
- Positive Bias: Ions desorb and efflux, decreasing conductance.
- The presence of a non-zero short-circuit current indicates spontaneous electrochemical activity, suggesting the device functions as a memristor in series with a battery.
Synaptic Plasticity Emulation: The CNM successfully emulates Short-Term Plasticity (STP), a critical function of biological synapses:
- Paired-Pulse Facilitation (PPF) and Depression (PPD): The device demonstrated both PPF (current increase) and PPD (current decrease) in response to pulse trains.
- Non-Monotonic PPF (NMPPF): The device replicated complex biological behaviors, including NMPPF, where the facilitation effect rises to a peak and then decays.
- Modeling: The decay profiles of $\Delta I$ versus pulse interval ( $\Delta T$ ) were well-fitted by a biexponential function ( $R^2 > 0.97$ ), corresponding to rapid and slow ion redistribution processes.
Energy Efficiency: The energy consumption per pulse was calculated to be approximately 100 fJ, with a total energy of ~500 fJ for a five-pulse group inducing a clear synaptic weight change.
Stability and Scalability: The device showed robust endurance over 40 cyclic scans with minimal shift in high/low conductance states. It also maintained memristive characteristics across different device batches and when electrode dimensions were scaled from mm² to μm².

Significance and Claims
The authors claim that this work establishes a viable pathway toward energy-efficient neuromorphic computing on nanofluidic platforms. By minimizing the electromotive force through specific electrode-electrolyte selection and leveraging ion adsorption dynamics within asymmetric nanochannels, the study demonstrates that nanofluidic memristors can operate at sub-100 mV voltages comparable to biological action potentials. The device's ability to emulate complex short-term synaptic plasticity with ultra-low energy consumption (100 fJ/spike) positions it as a competitive alternative to other neuromorphic mechanisms, bridging the performance gap between artificial and biological systems.