Home / Technology / Electronics & EW / Memristors: Pioneering the Future of Computing from Neuromorphic to Quantum

Memristors: Pioneering the Future of Computing from Neuromorphic to Quantum


In the ever-evolving landscape of computing, breakthrough technologies are reshaping the future. Among these innovations, memristors stand out as a transformative element, propelling us from neuromorphic processors to the realm of quantum computing. Memristors, short for “memory resistors,” are nanoscale devices that hold immense potential for revolutionizing the way we store and process information. This article explores the journey of memristors, from their inception to their pivotal role in advancing computing paradigms, particularly in the transition from neuromorphic to quantum architectures.


In the rapidly advancing landscape of technology, the surge in connected devices within the Internet of Things (IoT) framework is set to unleash an unprecedented amount of data for processing. Traditionally, electronic data processing has relied on integrated circuits, featuring countless transistors that manipulate electrical current. However, the physical limitations of transistors, as exemplified by smartphone chips boasting an average of five billion transistors, pose a significant hurdle. As technology progresses, the demand for smaller, more energy-efficient components becomes paramount, leading to the exploration of a revolutionary alternative: memristors.


The Birth and Evolution of Memristors:

In 1971, the renowned physicist Leon Chua theorized the existence of a fourth fundamental circuit element alongside resistors, capacitors, and inductors — the memristor. It took several decades for technology to catch up and validate Chua’s theory. In 2008, researchers at Hewlett-Packard (HP) Labs successfully developed the first functional memristor, confirming its place in the pantheon of electronic components.

A memristor acts a lot like a resistor but with one big difference: it can change resistance depending on the amount and direction of the voltage applied and can remember its resistance even when the voltage is turned off. When the electric power supply is turned off, the memristor remembers its most recent resistance until it is turned on again.

Memristors do not require a silicon layer, unlike transistors and therefore they are not affected by the same limitations of current microchip manufacturing technology.  Their circuits require fewer transistors, allowing more components (and more computing power) to be packed into the same physical space while also using less power to function.

Memristors, nanoscale devices with unique properties, hold the potential to usher in a new era in electronics. Smaller and simpler than transistors, memristors operate with lower energy consumption and possess the ability to retain data by “remembering” the charge that has passed through them.

Because memristors store and process information in the same location, they can get around the biggest bottleneck for computing speed and power: the connection between memory and processor.  In a conventional computer, logic and memory functions are located at different parts of the circuit.

The advantages of memristors extend beyond size and energy efficiency. These devices can store multiple memory states, with the University of Southampton showcasing a technology that achieves up to 128 discernible memory states per switch. With faster speed, reduced power consumption, and increased information density, memristors surpass traditional transistors, positioning themselves for integration into various everyday items, from textiles to windows and even coffee cups.

Memristors operate by using ions to store data, diverging from the electron flow of conventional transistors. This unique approach has sparked excitement among scientists, engineers, and computer experts who anticipate memristors revolutionizing the field of electronics. The potential transition from electrons to ions could mark the beginning of an era called “ionics.”

  • Beyond Neuromorphic Computing: Memristors are being explored for potential applications in various fields, including robotics, machine learning, image processing, and even bioelectronics.
  • Artificial Synapses: Memristors are being used to create artificial synapses for neuromorphic systems, enabling faster and more efficient learning in these devices.
  • In-memory Computing: Memristors offer the potential for in-memory computing, where data and computations are performed within the same memory space, leading to significant speed and energy efficiency gains.

Novel Memristor Materials and Structures:

In 2020, researchers at the University of Massachusetts Amherst achieved a groundbreaking advancement in neuromorphic computing by developing ultra-low power memristors, or “memory transistors,” using protein nanowires. Led by Ph.D. candidate Tianda Fu, the team addressed the challenge of conventional computers operating at higher voltages than the brain’s action potential signals.

Through experiments with protein nanowires from the bacterium Geobacter, the researchers successfully achieved memristor voltages comparable to neurological levels, demonstrating realistic evidence of ultra-low power computing capabilities. The protein nanowires, derived from the bacterium Geobacter, proved advantageous over silicon nanowires in terms of cost, stability, and compatibility with bodily fluids. The experiments demonstrated a device functioning at the same voltage level as the brain, marking a significant leap in ultra-low power computing.

The device, functioning at the same voltage as the brain, holds promise for exploration in electronics designed for biological voltage regimes, with potential applications in biomedical devices and communication with biological neurons.

The researchers, in collaboration with microbiologist Derek Lovely and electrical and computer engineering researcher Jun Yao, utilized the microbial ability of wild bacterial nanowires to breathe and chemically reduce metals, enabling efficient metal reduction in the memristors. By applying a pulsing on-off pattern of positive-negative charge to a metal thread in the memristor, the team observed changes in metal filaments, creating new connections and branching in the tiny device. This emulation of the learning process in a real brain showcased the device’s unique learning capability, distinct from conventional software-based learning in computers. The team plans to delve deeper into the mechanisms and applications of protein nanowires in memristors, envisioning potential uses such as devices for monitoring heart rate and the feasibility of these devices interacting with biological neurons.

Researchers at the University of Southampton have experimentally demonstrated an artificial neural network (ANN) utilizing memristor synapses, representing a significant advancement in brain-inspired computing. The study, published in Nature Communications, showcases an ANN with memristor synapses capable of reversible learning of noisy input data. This development addresses the need for efficient hardware synapses in large numbers, crucial for practical ANN implementations. Acting like synapses in the brain, the memristor array exhibited learning and re-learning capabilities in an unsupervised manner, presenting potential applications in low-power embedded processors for the Internet of Things. The technology allows for real-time processing of big data without prior knowledge, marking a paradigm shift in in-silico neural circuits and showcasing potential applications in pervasive sensing technologies for the Internet of Things.

Russian Scientists Test New Material for Neurocomputers

Russian scientists have conducted tests on a new material for neurocomputers with potential applications in developing computers based on memristors, designed to store and process data similarly to human brain neurons. The researchers, from the Solid-State Physics and Nanosystems Department at the National Research Nuclear University MEPhI, worked in collaboration with institutions such as the Russian Academy of Sciences’ Solid-State Physics Institute. The material, based on the bipolar effect of resistive switchings (BERS), could serve as a foundation for memristor-based computers, supporting a new approach to data processing known as membrane computing. The team utilized epitaxial fields formed on the surface of a single-crystalline substrate of strontium titanate in their research, demonstrating the potential for creating memristors for next-generation computers.

  • Beyond Metal Oxides: Exploration of new materials like transition metal dichalcogenides and perovskites is underway, offering promising avenues for improved memristor performance and functionality.
  • 3D Memristor Architectures: Researchers are developing 3D memristor architectures for increased density and improved connectivity, enabling the creation of more complex and powerful neuromorphic systems.
  • Hybrid Memristors: Combining memristors with other promising technologies like spintronics is being explored to unlock new functionalities and capabilities in artificial intelligence and computing.

Memristors in Neuromorphic Processors:

This new technology can act like the short-term memory of nerve cells allowing creating computers that operate in a way similar to the synapses in our brains.  Memristor arrays have the capacity to be trained rather than directly programmed, using learning rules.

Inspired by the remarkable capabilities of the human brain, researchers are integrating memristors into various applications. In neuromorphic computing, memristors mimic synaptic functions, enabling energy-efficient and adaptive processors– potentially resulting in computers that switch on and off instantly and never forget.  The potential transition from the von Neumann computer architecture to memristor-based systems is gaining traction, with memristors addressing the challenges posed by the size limit of conventional transistors in neural network-based computing.

One of the most promising applications of memristors lies in the realm of neuromorphic computing, a paradigm inspired by the human brain’s architecture and functionality. The brain’s efficiency in learning and adapting has long been a goal for artificial intelligence, and memristors play a crucial role in mimicking synaptic functions. Memristors can emulate the behavior of synapses, strengthening or weakening connections between artificial neurons based on usage patterns. This ability to mimic synaptic plasticity makes memristors integral to the development of energy-efficient and adaptive neuromorphic processors.

In the quest for efficient learning, memristors have found their place in the development of neuromorphic memristor devices. These devices, inspired by protein nanowires, operate at extremely low power and emulate the efficiency of biological counterparts in the brain. Memristor chips will soon be integrated in textiles, windows, even coffee cups and any imaginable items used in daily life. Researchers are exploring the potential applications of these devices, including monitoring heart rate, with the aim of achieving communication between such devices and actual neurons.

The culmination of these advancements has led to the creation of the first programmable memristor computer at the University of Michigan.

First Programmable Memristor Computer

The University of Michigan has achieved a groundbreaking milestone by developing the first programmable memristor computer, distinct from memristor arrays controlled by external computers. Led by U-M professor Wei Lu, the team integrated the memristor array directly onto a chip, enabling the processing of artificial intelligence (AI) directly on energy-constrained devices like smartphones. This innovation could eliminate the need for cloud processing of voice commands on smartphones, significantly speeding up response times. Memristor AI processors, according to Lu, have the potential to be 10-100 times more efficient than GPUs, which are already superior to CPUs in terms of power and throughput.

The memristor array, with over 5,800 memristors in the experimental-scale computer, allows each memristor to conduct its own calculation, enabling thousands of operations within a core simultaneously. This efficient processing is particularly advantageous for machine learning applications, as memristor arrays can handle complex matrix calculations without the need for external computing elements. The researchers demonstrated the device’s capabilities by applying three machine learning algorithms, achieving high accuracy in tasks such as classification, image reconstruction, and complex data pattern recognition.

Despite challenges in scaling up for commercial use, such as ensuring uniformity among memristors, the breakthrough holds promise for advancing AI capabilities in energy-constrained devices. This breakthrough holds the promise of processing artificial intelligence directly on small, energy-constrained devices like smartphones, eliminating the need for external computing elements and significantly improving response time.

“Memristors offer a possible route towards that end by supporting many fundamental features of learning synapses (memory storage, on-line learning, computationally powerful learning rule implementation, two-terminal structure) in extremely compact volumes and at exceptionally low energy costs. If artificial brains are ever going to become reality, therefore, memristive synapses have to succeed.”


Autonomous Systems

Inspired by how mammals see, a new “memristor” computer circuit prototype at the University of Michigan has the potential to process complex data, such as images and video orders of magnitude, faster and with much less power than today’s most advanced systems. Faster image processing could have big implications for autonomous systems such as self-driving cars, says Wei Lu, UM professor of electrical engineering and computer science.

The application of memristors extends to brain-inspired robot controllers, where researchers have demonstrated the use of memristors in a self-balancing robot. By incorporating memristors into an analog control system, the researchers achieved superior efficiency and performance, reducing cycle time and enhancing the robot’s balance.

Researchers have developed a robot controller using memristors, resulting in a significant reduction in cycle time and smoother balancing. The memristors, which act like synapses in the human brain, contribute to improved analog-digital computation for motion control. The hybrid platform, inspired by the collaboration of the cerebrum and cerebellum in the human brain, integrates digital components for high-level algorithms and analog components for sensor fusion and motion control. This innovation aligns with the efficiency and multitasking capabilities observed in the human brain, showcasing the potential of memristors in enhancing robotic systems.

By adding a memristor to an analog circuit with inputs from a gyroscope and an accelerometer, the researchers, led by Wei Wu, an associate professor of electrical engineering at USC, created a completely analog and completely physical Kalman filter to remove noise from the sensor signal. In addition, they used a second memristor can be used to turn that sensor data into a proportional-derivative (PD) controller. Next they put those two components together to build an analogy system that can do a bunch of the work required to keep an inverted pendulum robot upright far more efficiently than a traditional system. The difference in performance is readily apparent:

China makes breakthrough in system-integrated memristor computing-in-memory chips

China has achieved a significant breakthrough in semiconductor innovation, with researchers at Tsinghua University unveiling the world’s first fully system-integrated memristor computing-in-memory chip designed for efficient on-chip learning. The chip, currently in the laboratory phase, holds promise for advancements in artificial intelligence (AI), autonomous driving, wearable devices, and other high-tech fields. Notably, the chip consumes only 3 percent of the energy compared to an application-specific integrated circuit (ASIC) system for on-chip learning, showcasing its potential for enhancing computing efficiency in the AI era. The research team, part of the School of Integrated Circuits at Tsinghua University, highlighted the chip’s advantages, including lower latency, smaller energy consumption, and improved user privacy and data security. This development reflects China’s ongoing efforts to strengthen its semiconductor research and development amid global challenges and showcases its commitment to achieving self-reliance in chip innovation and technological advancement. The memristor storage and computing chip aligns with China’s broader goals outlined in the 14th Five-Year Plan to accelerate the development of high-end chips.


Recent Breakthroughs in Memristor Research: Ushering in a New Era of Computing

Memristors, revolutionary devices capable of remembering their resistance state even after power is switched off, are rapidly evolving and pushing the boundaries of computing technology. Here’s a look at some of the latest breakthroughs in memristor research:

1. Improved Performance and Endurance:

  • Higher Density: Researchers at MIT have developed memristors with a density of 100 million devices per square centimeter, significantly exceeding previous achievements and paving the way for more powerful and compact neuromorphic systems.
  • Lower Switching Energy: Scientists at IBM have achieved memristor switching with picojoule energy levels, promising significant power savings for future memristor-based devices.
  • Enhanced Endurance: Researchers at Tsinghua University have improved the endurance of memristors, allowing them to withstand millions of switching cycles without significant degradation, making them more reliable for long-term applications.

3. Advances in Neuromorphic Computing:

  • Enhanced Learning Algorithms: Memristor-based neuromorphic systems are being equipped with advanced learning algorithms inspired by the human brain, leading to improved performance in tasks like image recognition and natural language processing.
  • Hardware-Software Co-design: Researchers are focusing on co-designing hardware and software for memristor-based neuromorphic systems, optimizing their performance and efficiency for specific applications.
  • Large-Scale Integration: Efforts are underway to integrate large numbers of memristors into single chips, paving the way for the development of more powerful and commercially viable neuromorphic processors.

Advancing Beyond Neuromorphic: Memristors and Quantum Computing:

As computing ambitions soar, researchers are increasingly turning to quantum computing to solve complex problems beyond the reach of classical computers. Quantum computers harness the principles of quantum mechanics, utilizing qubits to perform computations that would be practically impossible for traditional machines. Memristors are poised to play a pivotal role in the development of quantum computing architectures.

One of the challenges in building practical quantum computers is the need for qubits that are stable and retain their quantum states. Memristors, with their ability to store and recall information based on resistance changes, offer a potential solution. Researchers are exploring the use of memristors to create stable qubits, enabling the development of scalable and error-resistant quantum computers.

Researchers at the University of Vienna, in collaboration with the National Research Council (CNR) and the Polytechnic University of Milan, have made a significant stride in the convergence of artificial intelligence (AI) and quantum computing. They achieved this by creating a photonic quantum memristor, a device that emulates the functionality of a memristor while encoding and transmitting quantum information, effectively interacting with quantum states. This development opens up new possibilities for the application of AI principles within the realm of quantum computing, marking a noteworthy advancement in the synergy between these two cutting-edge fields.

Memristors for Quantum Computing:

  • Qubit Interconnects: Memristors are being investigated as potential interconnects for qubits in quantum computers, enabling efficient communication and control of these delicate quantum systems.
  • Readout and Control: Researchers are exploring ways to use memristors to read out and manipulate the state of qubits, addressing a key challenge in quantum computing.
  • Scalable Quantum Systems: Memristor-based quantum systems have the potential to overcome the scalability limitations of current architectures, leading to the development of more powerful and practical quantum computers.

The Future Landscape:

As researchers continue to explore memristors’ capabilities, from pattern recognition to real-time data processing, their impact on computing paradigms becomes increasingly apparent. Memristors are not just components; they represent a paradigm shift in electronic systems, opening doors to more efficient, compact, and powerful computing solutions.

As memristors continue to evolve, their impact on computing paradigms will only intensify. From enabling brain-inspired neuromorphic processors to facilitating the development of stable qubits for quantum computers, memristors are a linchpin in the technological advancements shaping our future. From their theoretical inception to practical applications in neuromorphic processors and the potential revolution in quantum computing, memristors are carving a path toward unprecedented computational capabilities. As research and development in this field progress, we eagerly anticipate the next chapters in the memristor saga, as they continue to redefine the boundaries of what is possible in the world of computing.


















References and Resources also include




About Rajesh Uppal

Check Also

DARPA’s Next-Generation Microelectronics Manufacturing Program: Shaping the Future of 3DHI Microsystems

Introduction: DARPA, the Defense Advanced Research Projects Agency, is charting new frontiers in microelectronics with …

error: Content is protected !!