Ferrodiod memory breakthrough: 1112°F data storage revolution

Standard semiconductor drives are designed to operate under normal conditions. When the temperature significantly rises, the electrons storing data become unstable and begin to dissipate. This leads to device failure and data loss.

To address this issue, scientists from the University of Pennsylvania developed memory based on non-volatile ferroelectric diodes, referred to as ferrodiods, which are exceptionally resistant to high temperatures. Details of this project were described in the article "A scalable ferroelectric non-volatile memory operating at 600 °C," published in the journal "Nature."

The memory uses a metal-insulator-metal design with nickel and platinum electrodes. A key component is a thin layer (45 nm) of synthesized aluminum scandium nitride (AlScN). This material provides outstanding properties due to its ability to maintain an electrical state ("on" or "off," representing the ones and zeros of digital data) after the external electric field is removed, even at very high temperatures.

The thickness of this layer is crucial. Dhiren Pradhan, a doctoral researcher working on the new memory, explained that increased activity leads to diffusion and material degradation if the layer is too thin. Conversely, a too thick layer prevents ferroelectric switching, as the switching voltage scales with thickness, limiting practical application. For this reason, Pradhan's and Roy Olsson's labs worked for many months to find the appropriate layer thickness.

The ferrodiod memory was tested under extreme conditions—it operated for more than 60 hours at 1112°F.

The new technology can be used in devices operating in extreme temperatures. In an interview with Penn Today, Deep Jariwala from the University of Pennsylvania’s School of Engineering noted that memory devices capable of functioning at high temperatures could be applied in various fields, from deep Earth drilling to space exploration. Jariwala emphasized that the new memory enhances existing devices and opens new possibilities in science and technology.

Deep Jariwala also pointed out that ferrodiod memory could improve silicon carbide-based systems, enabling the integration of processing and data storage on a single chip. He referred to this as "memory-assisted computing," which should accelerate processes that involve processing large amounts of data under challenging conditions, supporting the development of artificial intelligence.