Hossein Mosallaei, a professor at Northeastern University, is pioneering a transformative shift in computing. His work, backed by a recent U.S. Department of Energy grant, centers on creating circuits that simultaneously process and store information—a feat traditional computers have yet to achieve. Mosallaei’s research represents a bold effort to close the gap between the human brain and artificial systems.
The grant, part of the DoE's AI Genesis Award, finances Mosallaei’s efforts to develop Self-Driving MXene Memristors. These microscopic devices, crafted from a new material known as MXene, replicate the brain's dual function of processing and retaining data in the same location. The potential of this work is immense, offering a model that could revolutionize the way computing systems operate.
For many years, the brain has stood as the benchmark in computing. It can perform billions of operations each second while using very little energy. Traditional computers, however, divide tasks between a central processing unit and memory, which results in higher energy consumption and slower operation. This fundamental separation between processing and storage has limited the efficiency of computing technology.
Mosallaei’s circuits tackle this division. By merging computation and memory within a single physical framework, his design reduces the energy costs of moving data. This mirrors the brain’s method of using neurons to process and store information simultaneously. The brain is a master of energy efficiency, and Mosallaei’s research aims to replicate that elegance in machines.
MXenes are a category of ultrathin, conductive materials created in 2011. They are about one to two nanometers thick—far thinner than a human hair, which is roughly 50,000 to 100,000 nanometers wide. Mosallaei arranges these materials into complex 3D configurations. Each MXene layer is carefully stacked to form a unique structure that enables the device to perform in ways traditional electronics cannot.
An MXene-based Memristor functions like a memory cell. Applying a voltage forms a conductive link between the material layers. This connection remains even when the power is off, enabling the device to keep a record of its activity. Unlike traditional memory systems, these circuits retain information without the need for constant power, a game-changing feature in the world of computing.
This innovative design allows the device to process information like a brain cell and store the outcome, similar to how repeated neural activity strengthens memory pathways. The MXene Memristor's ability to mimic synaptic behavior is central to the project’s success. It is through this mimicry that Mosallaei believes computers can become as adaptive and efficient as the human brain.
Mosallaei refers to this approach as "modeling the brain." The bridge is activated only when necessary, which conserves energy and enhances performance. This advancement could result in computers executing complex tasks while using significantly less energy than current systems. The implications are vast, ranging from more efficient data centers to smarter edge computing devices.
The origins and potential of MXenes
MXenes originate from the field of metamaterials, a scientific domain that explores materials with properties not found in nature. These materials have the potential to manipulate energy in ways that challenge conventional understanding. MXenes, named for their chemical composition, are constructed from layers of metal atoms and other elements, arranged into nanoscale structures. This structure allows for remarkable flexibility in tailoring their properties.
The breakthrough in MXene development came in 2011, when scientists Yury Gogotsi and Michel Barsoum pioneered their synthesis. Their work opened the door to a new class of materials with unparalleled conductivity and versatility. Since then, researchers have explored MXenes for applications in energy storage, electromagnetic shielding, and now, computing. The potential applications are vast, and the field is gaining momentum in both academic and industrial circles.
A future powered by brain-like computing
Mosallaei’s vision extends beyond the lab. He envisions a future where computers, inspired by the brain’s design, become more intuitive and efficient. The ability to process and store information without the need for constant data transfer could lead to systems that learn and adapt in real-time. These advancements could redefine the role of AI and computing in society, enabling devices that respond to their environment with the intelligence and agility of the human mind.
As Mosallaei continues his work, he is supported by experts like Nathan Kundtz, CEO of Rendered.ai, who sees the transformative potential of metamaterials in computing. Kundtz emphasizes the "tremendous amount of power saving" that could result from adopting these materials. For Mosallaei, the journey is just beginning. With each breakthrough, the dream of brain-like computers moves closer to reality.
Mosallaei’s passion for metamaterials is evident in his work. He describes himself as a "metamaterial person" and has dedicated the last 20 years to creating materials that "you cannot find in nature." This commitment drives his research forward, pushing the boundaries of what is possible in computing. As the world demands more power and efficiency from its technology, Mosallaei's work could lead the way toward a new era of intelligent machines.