Current flash memory technologies require relatively large currents to read or write data. Mobile cloud-enabled devices and the Internet of Things will require memory that is energy-efficient and small in size.
The University of Tokyo’s Institute of Industrial Science used a ferroelectric gate insulator and an atomic-layer-deposited oxide semiconductor channel to construct three-dimensional vertically formed field-effect transistors for high-density data storage devices. Furthermore, they discovered that by employing antiferroelectric instead of ferroelectric, they were able to erase data with only a small net charge, resulting in more efficient write operations. This research could lead to new data storage memory that is even smaller and more environmentally friendly.
The researchers produced a proof-of-concept 3D stacked memory cell based on ferroelectric and antiferroelectric field-effect transistors (FETs) with an atomic-layer-deposited oxide semiconductor channel. These FETs can store ones and zeros in a non-volatile way, which means they don’t need power all of the time. The vertical device structure boosts data density while lowering operating energy requirements. In a vertical trench structure, hafnium oxide and indium oxide layers were formed.
Electric dipoles in ferroelectric materials are most stable when aligned in the same direction. The vertical alignment of the dipoles is spontaneously enabled by ferroelectric Hafnium Oxide. The degree of polarisation in the ferroelectric layer stores information, which can be read by the system due to changes in electrical resistance. In the erased state, antiferroelectrics, on the other hand, prefer to alternate the dipoles up and down, allowing for efficient erasure operations within the oxide semiconductor channel.
“We showed that our device was stable for at least 1,000 cycles,” first author Zhuo Li says. The researchers experimented with different indium oxide layer thicknesses. They discovered that tweaking this parameter can result in significant performance gains. The researchers also plotted the most stable surface states using first-principles computer simulations. “Our approach has the potential to greatly improve the field of non-volatile memory,” senior author Masaharu Kobayashi says.