Scientists have watched in real time as a promising memory material develops its ability to store information, revealing that tiny gaps left by missing oxygen atoms can have a surprisingly large effect on its performance.
Researchers led by Seoul National University, working with several other South Korean institutions, studied hafnia-zirconia, or HZO.
This material is being explored for next-generation memory chips that could store information while using less power.
The study, published in Advanced Functional Materials, found that controlling microscopic defects called oxygen vacancies can determine when HZO crystallizes and what type of crystal structure it eventually forms.
Ferroelectric materials such as HZO are attractive for nonvolatile memory because they can maintain an electrical state even after power is switched off. Those different electrical states can represent the 0s and 1s used to store digital information.
HZO is particularly promising because it works well with existing silicon semiconductor manufacturing methods. It can also remain ferroelectric when made only a few nanometers thick, potentially allowing manufacturers to pack more memory into smaller devices.
However, HZO does not automatically have the crystal structure needed for ferroelectric behavior. It must undergo heat treatment that transforms an initially disordered film into crystals. One particular structure, known as the orthorhombic phase, gives HZO its useful ferroelectric properties.
The researchers wanted to understand exactly how that structure develops.
They produced HZO films only 10 nanometers thick and changed the number of oxygen vacancies by adjusting the amount of ozone used during manufacturing. An oxygen vacancy occurs when an oxygen atom is missing from the position it would normally occupy in the material.
Using powerful X-rays from the Pohang Accelerator Laboratory, the scientists then watched the films change while they were heated and cooled. This allowed them to follow crystallization as it happened instead of simply comparing the material before and after treatment.
The results showed that having more oxygen vacancies delayed crystallization. The atoms had greater difficulty rearranging themselves and beginning the crystal-forming process. Once crystals started developing, their growth was also restricted, while a nonferroelectric structure remained for longer.
Films containing fewer oxygen vacancies behaved very differently. They began crystallizing at lower temperatures, their crystals grew more easily, and the ferroelectric orthorhombic phase formed more successfully.
The difference was substantial. HZO films with fewer oxygen vacancies started crystallizing at temperatures about 30°C lower than films containing more vacancies. At 400°C, their switchable electrical polarization—a measure closely connected to the ability to store 0 and 1 states—was approximately 11 times greater.
The findings show that oxygen vacancies should not simply be viewed as unwanted defects. Instead, carefully controlling them could provide manufacturers with a powerful way to engineer how memory materials develop during production.
Lowering the temperature needed to create ferroelectric HZO could be particularly valuable. Semiconductor manufacturing involves many delicate components, and excessive heat during later production stages can damage structures that have already been built.
Better control of oxygen vacancies could therefore help engineers create high-density, energy-efficient memory that retains information without continuous power. Such technology could eventually find its way into smartphones, vehicles, Internet of Things devices and advanced 3D memory chips.
