Low Energy Transistors
Computing is mostly performed by silicon transistors. They are the essential component in most of today’s computers, but they also consume a lot of energy when operating.
Another type of semiconductor component is FET: field-effect transistor. This type of transistor uses an electric field to control the current through a semiconductor and consumes a lot less power than a silicon transistor. Currently, the most widely used field-effect transistor is the MOSFET (metal–oxide–semiconductor field-effect transistor).
Scientists at Penn State University and the University of Minnesota are exploring a previously overlooked property of FETs: incipient ferroelectricity.
Incipient ferroelectricity is the characteristic of temporary, scattered polarization, as if the material has the potential to become ferroelectricity, but it needs a little push.
This could be very useful for neuromorphic computing, a technique that tries to replicate the way a brain works instead of using traditional binary computing methods. Such a method could store data with a lot less energy, as well as handle harsh conditions like outer space.
The researchers published their results in Nature Communication, titled “Multifunctional 2D FETs exploiting incipient ferroelectricity in freestanding SrTiO3 nanomembranes at sub-ambient temperatures”1.
What is Incipient Ferroelectricity?
Ferroelectricity is the characteristic of certain materials that have a spontaneous electric polarization that can be reversed by the application of an external electric field.
This is commonly used for building ferroelectric capacitors smaller in physical size compared to dielectric capacitors. It is a physical characteristic also used for building ferroelectric RAM memory for computers and RFID cards.
Incipient ferroelectricity is when the ferroelectric characteristic only occurs in specific conditions: it can hold an electrical charge but needs certain conditions to achieve an electrical charge.
“Incipient ferroelectricity means there’s no stable ferroelectric order at room temperature. Instead, there are small, scattered clusters of polar domains. It’s a more flexible structure compared to traditional ferroelectric materials.”
Dipanjan Sen – Doctoral candidate in engineering science and mechanics
Usually, this trait has been considered a limitation for a material, not an advantage, as it can lead to short memory retention from these transistors.
This was until the researchers’ team found that incipient ferroelectricity became less incipient and more traditional at colder temperatures.
Using Incipient Ferroelectricity
The more flexible nature of incipient ferroelectricity would be a problem for traditional computing methods. But this can be turned into an asset.
“In cryogenic conditions, this material exhibited traditional ferroelectric-like behavior suitable for memory applications. But at room temperature, this property behaved differently. It had this relaxor nature: a more disordered, short-range polarization response.”
Saptarshi Das – Professor of Engineering at Penn State.
The incipient ferroelectricity can be used to create neuron-like computing systems, called neuromorphic computers.
These systems imitate how the human brain processes information using neurons and use much less energy than traditional computers. Like our brain, it saves energy by only using power when needed, like flipping a light switch on and off, instead of staying on all the time like traditional computers.
To leverage this into useful applications, FETs with enhanced incipient ferroelectric behaviors needed to be built.
Incipient Ferroelectric FETs
Strontium Titanate
This was the task of the researchers from the University of Minnesota, who developed the FETs by depositing a layer of atoms on a substrate to form a thin film. These films, made of strontium titanate (STO), were then combined with molybdenum disulfide, a two-dimensional material.
Source: Nature Communication
Normally, strontium titanate is typically non-ferroelectric, meaning it does not have a permanent electric field.
But at very low temperatures, it exhibits ferroelectric-like behavior. Strontium titanate thin films are also a perovskite material.
“We were surprised to see that these well-known perovskite materials could exhibit exotic ferroelectric properties at the device level.
It wasn’t something we anticipated, but once we started fabricating the devices, we saw behaviors that could really redefine advanced electronics.”
Perovskite materials have a specific type of crystal structure, which are used to produce solar panels, but might also be key in developing photonics computing.
FETs
Using strontium titanate thin films, the researchers built three artificial neurons and used them to perform a classification task using a grid of three-by-three-pixel images, as a proof-of-concept.
Source: Nature Communication
The devices were able to classify each image into different categories. This learning method could eventually be used for image identification and classification or pattern recognition. Importantly, it works at room temperature, reducing energy costs.
Mayukh Das – Doctoral candidate in engineering science and mechanics
These FET-based artificial neurons, using incipient ferroelectricity, could create a low-cost, efficient computing system that uses a lot less energy.
This joins other new applications of FETs for advanced computing systems, notably the Ferroelectric High Electron Mobility Transistor (FeHEMT) for next-gen communication devices.
Future Applications
This is still, for now, very much an experimental device, exploring the possibilities of a new application for incipient ferroelectricity.
It will likely have the most important application in AI applications, as LLMs and other AI systems require massive memory to store all the data required, which is very energy-intensive.
Some improvements in neural network architecture and software, like DeepSeek, can help. Ultimately, a new type of hardware more fit to match AI requirements will be needed to keep the AI industry from becoming one of the largest consumers of energy in the world.
“Perfecting these materials and integrating them into everyday devices like smartphones or laptops will take time, so there’s so much more to explore.
In addition, we’re examining other materials, like barium titanate, to uncover their potential. The opportunities for growth are immense, both in materials and device applications.”
Dipanjan Sen – Doctoral candidate in engineering science and mechanics
Neuromorphic Computing Companies
1. Intel
Intel Corporation (INTC -4.43%)
Intel is a giant in the semiconductor sector and has evolved over the years from a founder of the industry to a scientific and innovation leader, losing the top spot of manufacturing volume to companies like Taiwan’s TSMC.
Intel is a leader in neuromorphic computing, in part thanks to its Loihi 2 chip.
Source: Intel
It also created the Intel Neuromorphic Research Community, which includes Pennsylvania State University, involved in recent vanadium dioxide research, as well as 75+ other research groups.
Source: Intel
Intel is also very active in mimicking biological sense through replicating the way our brain works (itself a branch of neuromorphic computing), something we discussed further in our article “Biomimetic Olfactory Chips: Are Artificial Intelligence and E-Noses the Next Canary in a Coal Mine?”
Overall, research from Intel Lab is at the forefront of semiconductor innovation, including AI, quantum computing, neuromorphic computing, etc.
We discussed Intel’s advances in quantum computing in our articles “The Current State of Quantum Computing” and in September 2024 “Stock Of The Week: Intel (INTC)”.
2. IBM
International Business Machines Corporation (IBM -1.26%)
Another historical pioneer in computing, semiconductors, and chip design, International Business Machines Corporation (IBM) is also investigating neuromorphic computing.
It is developing SyNAPSE: Scalable energy-efficient neuro synaptic computing, supported by Defense Advanced Research Programs Agency (DARPA), to combine “nanoscience, neuroscience, and supercomputing to simulate and emulate the brain’s abilities for sensation, perception, action, interaction and cognition “.
It is equally at the forefront of the development of quantum computers. For example, it developed its 127-qubit “Eagle” quantum computer, which was followed by a 433-qubit system known as “Osprey and the 1,121 superconducting qubit quantum processor ‘Condor’.
Source: All About Circuits
Together with Intel, IBM is among the companies most aggressively pushing for new forms of computing technologies, like quantum and neuromorphic computing. It is likely to benefit from progress in better understanding materials used in neuromorphic transistors.
Studies Referenced:
1. Sen, D., Ravichandran, H., Das, M. et al. Multifunctional 2D FETs exploiting incipient ferroelectricity in freestanding SrTiO3 nanomembranes at sub-ambient temperatures. Nat Commun 15, 10739 (2024). https://doi.org/10.1038/s41467-024-54231-z