For decades, silicon computer chips have upheld Moore’s Law, delivering exponential growth in computing power. But this relentless miniaturization faces physical limits, as components approach the quantum realm. One futuristic idea attempts to mimic biology by using ions instead of electrons to process information.
The brain’s incredible energy efficiency has inspired this radical departure from conventional electronics. Using ions rather than electrons, the brain processes information and performs computations with minimal power consumption. Harnessing aqueous ions to compute represents an attempt to mimic the brain’s biological ionic signaling.
Riding the Ion Flows: How Brain Cells Communicate
The brain contains billions of neurons that communicate with each other to process information. Neurons are cells that have a central body and long extensions called axons and dendrites.
Neurons communicate by sending electrical signals along their axons. These signals are generated by the movement of charged atoms called ions, such as sodium, potassium, and calcium. At rest, the neuron has different concentrations of these ions inside and outside its cell membrane.
When a neuron needs to signal, tiny pore-like structures in the cell membrane open, allowing specific ions to rush in or out of the cell. This movement of ions creates a voltage or signal that travels down the axon.
The axon is connected to the dendrites of other neurons. When the electrical signal reaches the end of the axon, it triggers the release of chemical messenger molecules called neurotransmitters.
The neurotransmitters cross the small gaps, called synapses, between axons and dendrites. They bind to receptors on the receiving neuron’s dendrites, which causes ion channels to open there as well.
This allows ions to flow into the receiving neuron, creating another electrical signal. In this way, the signal gets passed from neuron to neuron through coordinated ion flows along chains of axons and dendrites.
This ionic signaling allows neurons to rapidly communicate signals over long distances and form circuits and pathways in the brain. The complex patterns of neuronal signaling underlie information processing and computation in the brain.
By mimicking aspects of these biological ion flows in engineered aqueous solutions, researchers hope to recreate some of the brain’s incredible computational abilities in new ionic computing devices.
Ionic and Electronic Circuits Have Fundamental Differences
While ions move slower than electrons, researchers hope engineered ionic circuits can come close to the extreme energy efficiency of their biological counterparts. If successful, aqueous ionic computing could enable a new class of ultra-low power devices by drawing inspiration from the brain.
Ionic circuits operate on a fundamentally different principle than electronic circuits, relying on the movement of charged atoms or molecules dissolved in liquid solutions rather than the flow of electrons in solid semiconductor materials. This underlying distinction leads to a number of key contrasts between the two approaches.
Ions move at a much slower pace through aqueous solutions compared to rapidly flowing electrons in electronics, imposing limits on the speed at which ionic circuits can perform computations. The sluggish mobility of ions is a major constraint ionic systems face versus the blazing speeds of modern electronic processors. However, the chemical interactions possible in ionic liquids allow for tuning component properties in ways rigid solid-state materials do not easily permit. For example, ionic computing aims to achieve the extremely low power consumption seen in biological ion signaling, while electronics are inherently more energy intensive.
Another contrast arises in scalability, as miniaturizing ionic circuits faces challenges arising from fluid constraints that electronics do not encounter. Microfabrication techniques have enabled electronics to shrink to tiny scales. On the other hand, the diversity of ionic species presents opportunities to encode richer information representations compared to simple electronic signals.
Despite their fundamental differences, each approach offers distinct advantages that suggest combining ionic and electronic circuits could yield hybrid architectures possessing novel capabilities. Ionic components might provide new flexible, low power computational mechanisms, while electronics offer efficient interconnectivity and interfacing. Diverse ionic signals could facilitate adaptive algorithms and cognition, complemented by the robust digital logic of electronics.
Rather than competing, these divergent computing substrates may find integrative synergy through hybridization. Ionic/electronic combinations remain an exciting, unexplored frontier promising applications difficult to reach with either technology alone.
Engineering Ionic Circuits
Up to now though, aqueous ionics has focused on individual components like diodes and transistors. But taking a huge leap, researchers have recently interconnected 256 new electrochemical ionic transistors into a fully functioning 16×16 array circuit. This pioneering work, reported in Advanced Materials, provides an exciting glimpse of the potential for ionic computing architectures.
At the heart of this achievement is a clever ionic transistor design using a center disk electrode surrounded by two ring electrodes. By running opposite polarity currents through the rings, the team creates local increases in the concentrations of two charged molecule types around the disk. This tunes the electrochemical reaction rate at the disk electrode, allowing the ring current to control the disk’s output current.
To scale this up, the researchers fabricated an array of 256 ionic transistors on a computer chip. In a first, they performed analog multiply-accumulate (MAC) computations on an ionic circuit. MAC involves multiplying an input by a weight and accumulating the result — a basic operation for machine learning. In the ionic system, the ring current sets the weight, the disk voltage is the input, and the disk currents are accumulated.
This technique could substantially boost the energy efficiency of MAC relative to digital chips, bringing us closer to bio-inspired computing. While limited to binary weights/inputs, it matched theoretical predictions, proving the concept. Most ambitiously, the team did matrix vector multiplication using MAC across all 256 transistors.
The researchers note challenges like slow transistor response and lack of independent ionic pathways. However, chip engineering advances could address these. Fundamentally though, ion mobility restricts speed versus electronics. Instead, the focus is on new capabilities, like using diverse ion types to encode rich information.
This pioneering work suggests that aqueous ionics could complement solid-state electronics for specialized ultra-low power computing applications. With more development, ionic circuits may yet catch up on speed and scale. But their unconventional advantages will likely open novel applications not easily accessible to standard silicon. This discovery ushers in an exciting new experimental era in ionic computing architectures.
