Photo credit: www.sciencedaily.com
Researchers at Penn State have made significant strides in the development of a novel type of computer memory, leveraging a phenomenon known as incipient ferroelectricity. This innovation has the potential to greatly improve the efficiency of electronic devices, enabling them to function with reduced energy consumption and in challenging environments, such as outer space.
The findings have been detailed in the journal Nature Communications, focusing on the capabilities of multifunctional two-dimensional field-effect transistors (FETs). These advanced electronic components utilize ultra-thin material layers to regulate electrical signals, facilitating various functions—including switching, sensing, and memory—within a compact design. Their ferroelectric-like characteristics are crucial, as they allow for reversible electric conduction upon application of an external electric field, an essential attribute for modern computing.
Conventional computing systems, particularly in artificial intelligence applications like image recognition, tend to require substantial amounts of energy. However, the energy-efficient design of ferroelectric transistors offers a more sustainable solution.
“AI accelerators are known for their high energy demands,” explained Harikrishnan Ravichandran, a doctoral student in engineering science and mechanics and a co-author of the research. “Our devices are capable of rapid switching while consuming significantly less power, paving the way for a new era of faster and more environmentally friendly computing technologies.”
Incipient ferroelectricity—a property previously seen as marginal—may play a critical role in enabling these quicker, more efficient devices. This form of ferroelectricity occurs in materials that display temporary, scattered polarization, allowing them to function somewhat like tiny dipoles. These materials can maintain an electrical charge under specific conditions without reaching a stable ferroelectric state.
To illustrate, incipient ferroelectricity can be likened to a material poised to achieve ferroelectricity but requiring certain conditions to manifest the property. “Incipient ferroelectricity indicates a lack of stable ferroelectric order at room temperature,” noted Dipanjan Sen, the lead author of the study and a doctoral candidate in engineering science and mechanics. “Instead, it features small, random clusters of polar domains, making it structurally more adaptable than conventional ferroelectric materials.”
This characteristic, often viewed as a limitation, appears to become more pronounced as temperatures decrease. Ravichandran indicated that the devices exhibited distinctive behaviors across different temperature ranges, suggesting a level of flexibility that could open doors to new applications.
“Our research aimed to investigate whether incipient ferroelectricity, typically regarded as a disadvantage due to its association with short memory retention, could actually offer benefits,” stated Saptarshi Das, the corresponding author and a professor of engineering science and mechanics at Penn State. “In cryogenic environments, this material displayed traditional ferroelectric behavior suitable for memory functions, but its characteristics shifted at room temperature, revealing a relaxor nature.”
Relaxor behavior involves a disordered, short-range response in polarization. This more chaotic behavior contrasts sharply with the stable, long-range ordering characteristic of traditional ferroelectric materials, resulting in weaker or less stable properties at room temperature. The researchers posit that this phenomenon could be harnessed for neuromorphic computing, which seeks to emulate the brain’s information processing capabilities, offering substantial energy savings compared to conventional computing systems. This approach mimics the brain’s energy efficiency by limiting power usage to when it is needed, rather than maintaining constant power as seen in typical computers.
“These devices function similarly to neurons, mirroring biological neural activities,” described Mayukh Das, a doctoral candidate and co-author. “To assess this capability, we conducted a classification task using a grid of three-by-three pixel images processed by three artificial neurons. The devices successfully classified the images into various categories, a methodology that could be adapted for tasks like image recognition and pattern detection. Crucially, this operation occurs at room temperature, significantly lowering energy expenditures. Such devices emulate the nervous system’s functionality, creating a cost-effective and energy-efficient computing framework.”
Collaborators from the University of Minnesota contributed to the project by fabricating the FETs, which involved layering atoms on a substrate to produce a thin film. This film comprised strontium titanate, paired with molybdenum disulfide, a two-dimensional compound.
Strontium titanate generally exhibits non-ferroelectric properties, lacking a permanent electric charge. However, when formed into freestanding nanomembranes, it can display polar order that enables ferroelectric-like behavior at very low temperatures.
The thin films of strontium titanate, along with their incipient ferroelectric properties, belong to a class of materials known as perovskites, recognized for their outstanding electronic attributes.
“We were taken aback to find that these renowned perovskite materials could showcase exotic ferroelectric properties at the device scale,” remarked Sen. “This was an unexpected outcome, but the behaviors we observed during device fabrication have the potential to redefine advanced electronic functionalities.”
The research team acknowledged that ongoing studies will address existing challenges such as scalability and market readiness, while also investigating additional materials for potential applications.
“Currently, we’re still in the research and development phase,” Sen added. “Enhancing these materials and integrating them into everyday devices—like smartphones and laptops—will require time and further exploration. We are also looking into other materials, such as barium titanate, to discover new possibilities. The potential for advancement is substantial, spanning both materials and application development.”
The study includes contributions from various authors at Penn State, including Pranavram Venkatram, Zhiyu Zhang, Yongwen Sun, Shiva Subbulakshmi Radhakrishnan, Akash Saha, Sankalpa Hazra, Chen Chen, Joan Redwing, Venkat Gopalan, and Yang Yang. From the University of Minnesota, co-authors include Sooho Choo, Shivasheesh Varshney, Jay Shah, K. Andre Mkhoyan, and Bharat Jalan.
This research received support from the U.S. National Science Foundation and the Army Research Office.
Source
www.sciencedaily.com