Iontronic Circuits: Building Intelligence in Brine
Imagine a future in which computers process information not with streams of electrons but with hydrated ions flowing through salt water, a system that mimics how the brain itself computes. This emerging field—known as iontronics, a portmanteau of ions and electronics—is rapidly growing as researchers design neuromorphic computing devices, inspired by animal nervous systems and powered by electrolyte solutions at the nanoscale [1–3]. Since Leon Chua introduced the memory resistor or “memristor” in the 1970s [4], these components have been considered revolutionary building blocks for neuromorphic computing. A memristor’s electrical resistance depends on the current that flowed through it before it was powered off, offering a way to store information. Unlike solid-state memristors, fluidic ones still face challenges in terms of scalability and integration with a circuit. A study by Patricio Ramirez of the University of Valencia in Spain and colleagues recently presented a preliminary step toward a scalable version [5]. The researchers demonstrated that nanofluidic memristors, based on membranes containing conical nanopores, can be organized into functional circuits. By arranging membranes in series and in parallel configurations, they showed how synaptic weight—that is, conductance—can be manipulated to represent brain-like properties that could be used in future logic and fluidic neuromorphic systems [6].
In biological systems, neurons communicate via voltage spikes that trigger ion fluxes across channels in the cell membrane. This process inspired previous designs of artificial nanopores that replicate some aspects of biological ion transport. Cone-shaped nanopores enable voltage-driven ionic transport that preferentially moves from the narrow “tip” to the wider “base,” producing a one-way valve effect known as ionic rectification. These conical pores essentially act as nanofluidic diodes [7].
Previous experimental studies [8] and recent theoretical work [9] have shown that current-voltage characteristics in such systems can exhibit hysteresis, a critical feature of memristive behavior. This hysteresis—the conductance depending on the history of the ionic current—enables synaptic functions like potentiation and depression, elements of plasticity vital to brain-like computing. However, fluidic circuits have yet to be demonstrated. Ramirez and colleagues address this with a conceptually simple but effective idea: using identical membranes, connected in different ways, as building blocks for ionic circuits.
The membranes were fabricated with a dense array of conical nanopores, using ion-track etching at the GSI Helmholtz Centre for Heavy Ion Research in Germany. The researchers placed a membrane in an electrochemical cell between two electrodes to measure its conductance. They varied the voltage across the cell—and the rate at which the voltage changed—and measured the resulting current. The current-voltage response developed clear hysteresis loops, meaning that the membrane’s conductance depended not only on the instantaneous voltage but also on the history of the applied bias, a hallmark of memory.
The researchers next explored how these membranes behaved when connected in different configurations. First, they connected two membranes in a series arrangement, with both facing the same direction (pores aligned tip to base). Upon exposure to a sequence of voltage spikes, the system preserved the initial hysteresis and exhibited a gradual increase in conductance, a link’s strengthening known in neurology as potentiation. Next, they flipped one of the membranes, so the pore orientation was opposite (pores aligned tip to tip). This time, upon exposure to the voltage spikes, the rectification vanished—the membrane ceased to act like a one-way valve for ionic flow. The resulting current response was the same for both the “up” and “down” part of the voltage spike, with current flows at zero voltage appearing in both orientations. While these hysteresis loops are different for these membrane configurations, their origin still lies in asymmetric ion transport: Ions always preferentially move from tip to base, even in this series arrangement.
In a parallel configuration things get even more intriguing. When the membranes were arranged in opposite directions—an “antiparallel” setup—the current-voltage relationship developed a rare triple-loop hysteresis. Here, the membrane’s conductance increased for a near-zero voltage, but became more inductive for voltages well above or below zero. This complexity reflects the combined influence of capacitive and inductive memory effects in the nanofluidic system. To investigate such impedance effects, the team used methods common in impedance spectroscopy to analyze the circuit behavior. Capacitance in such systems should originate from the electrical double layers at interfaces. But the observed inductive behavior is more surprising and rarely discussed in fluidic systems [10, 11].
The physical origin of this induction lies in ion accumulation and depletion within the conical nanopores. An applied voltage drives ions to concentrate at the tip, creating a nonequilibrium state. When the voltage is removed, entropy energy drives these ions to again diffuse out of the nanopores, creating a delayed, history-dependent current-voltage response. Upon depletion, surface charges attract ions back into the pore, generating a similar lag. These dynamic, energy-restoring processes mimic inductance in electrical circuits. The ionic induction is inherently volatile: Once the external bias is removed, ions eventually relax to equilibrium. Nonetheless, reducing the ionic diffusion coefficient in confinement offers a tool for extending memory-retention times [12, 13].
Although these parallel nanopore membranes reproduce some of the dynamics thought to underlie biological information processing, the fluidic circuits in a brain are highly integrated across three dimensions. Replicating this complexity with scalable and reliable fluidic neuromorphic systems remains a long-term challenge. Yet Ramirez and colleagues offer a compelling proof of principle: Essential circuit functionalities—rectification, memory, and signal modulation—can be achieved using simple membrane arrangements. Nonetheless, key hurdles remain, including device scalability, integration, and efficient interfacing with electronic components and ensuring the long-term endurance and reproducibility of memristive behavior. Still, this work marks a meaningful step forward, suggesting that fluidic systems may gradually evolve into practical platforms for neuromorphic computing.
References
- P. Robin et al., “Long-term memory and synapse-like dynamics in two-dimensional nanofluidic channels,” Science 379, 161 (2023).
- T. Xiong et al., “Neuromorphic functions with a polyelectrolyte-confined fluidic memristor,” Science 379, 156 (2023).
- A. Noy and S. B. Darling, “Nanofluidic computing makes a splash,” Science 379, 143 (2023).
- L. O. Chua and S. M. Kang, “Memristive devices and systems,” Proc. IEEE 64, 209 (1976).
- P. Ramirez et al., “Memristive arrangements of nanofluidic pores,” Phys. Rev. E 109, 044803 (2024).
- T. Emmerich et al., “Nanofluidic logic with mechano–ionic memristive switches,” Nat. Electron. 7, 271 (2024).
- I. Vlassiouk and Z. S. Siwy, “Nanofluidic diode,” Nano Lett. 7, 552 (2007).
- D. Wang et al., “Transmembrane potential across single conical nanopores and resulting memristive and memcapacitive ion transport,” J. Am. Chem. Soc. 134, 3651 (2012).
- T. M. Kamsma et al., “Iontronic neuromorphic signaling with conical microfluidic memristors,” Phys. Rev. Lett. 130, 268401 (2023).
- P. Ramirez et al., “Neuromorphic responses of nanofluidic memristors in symmetric and asymmetric ionic solutions,” J. Chem. Phys. 160, 044701 (2024).
- J. Bisquert, “Inductive and capacitive hysteresis of current-voltage curves: unified structural dynamics in solar energy devices, memristors, ionic transistors, and bioelectronics,” PRX Energy 3, 011001 (2024).
- P. Robin et al., “Modeling of emergent memory and voltage spiking in ionic transport through angstrom-scale slits,” Science 373, 687 (2021).
- K. Liu et al., “Resistance-restorable nanofluidic memristor and neuromorphic chip,” Nano Lett. 25, 6530 (2025).