Scott Schrage, July 13, 2022
Team achieves ‘significant breakthrough’ in data-encoding material
Going against the grain — many of them, to be precise — has steered material scientists from the University of Nebraska–Lincoln to a technologically promising discovery.
For years, most in the field believed that a material known as hafnium oxide was most appealing when it resembled a stained-glass window, its crystal riven with nanoscopic boundaries that formed minuscule fragments, or grains.
Only a surplus of small grains, it was thought, would persuade hafnium oxide to yield ferroelectricity: a willingness to reverse the vertical separation, or polarization, of its negative and positive charges — to encode the 1s and 0s of digital data — in exchange for a drink of voltage. The energy-efficiency of that data encoding, and the ability to retain 1s and 0s even when a power source is pulled, have positioned ferroelectrics like hafnium oxide as prime players in next-gen microelectronics.
But in a new Nature Materials study, a team led by Nebraska’s Xiaoshan Xu, Evgeny Tsymbal and Alexei Gruverman has shown that hafnium oxide was misdiagnosed all along — that small grains need not apply for its ferroelectricity to emerge. In fact, the Husker researchers have demonstrated that growing a higher-quality, larger-grained crystal of hafnium oxide can actually generate higher polarization and potentially more reliable ferroelectricity. The quality of the crystal, meanwhile, is offering sharper insights into how and why that ferroelectricity occurs.
“This is a very big step forward,” said Gruverman, Charles Mach University Professor of physics and astronomy at Nebraska.
Many materials, including hafnium oxide, can behave differently depending on the arrangements of their atoms and the types of symmetry their lattice-forming molecules establish across a crystal. In its most common, stable phases, hafnium oxide produces no ferroelectricity at all. Only one — its orthorhombic, or O, phase — can turn hafnium oxide ferroelectric.
Unfortunately for electrical engineers, the O phase is also less stable, meaning that it’s prone to reverting to another, non-ferroelectric one. Based on the results of prior studies, most material scientists were convinced that only an abundance of tiny, disorderly grains could lock in the O phase and, by extension, the material’s ferroelectricity. Even then, the resulting polarization never approached what theory-based simulations predicted.
The Nebraska researchers suspected another way was possible. They began by growing a thin film of hafnium oxide, containing a dash of the element yttrium, atop a base material with a carefully calibrated symmetry of its own. When grown at the proper temperature — roughly 1,600 degrees Fahrenheit — the hafnium oxide film adopted what many thought impossible: a high-quality crystal featuring large grains and a stable, ferroelectricity-friendly O phase.
The team found that the polarization values of its large-grained samples generally sustained even when the finished material was subjected to low temperatures. That distinguishes it from its smaller-grained predecessors, whose polarization has tended to plummet, sometimes as much as three-fold, when moving from room temperatures to freezing.
Xu said such a drop suggests that the ferroelectricity of the lower-quality, smaller-grained hafnium oxide comes not from the material itself, but instead from its defects — potentially explaining the disparity between theoretical predictions and actual readings of its polarization. Tsymbal concurred that its ferroelectricity, or an effect that resembles it, may originate from oxygen atoms vacating their usual places in the material’s molecular lattice.
The fact that the Husker team’s higher-quality hafnium oxide maintained its polarization even amid freezing temperatures is just one of several signs that its ferroelectricity is an intrinsic property of the material itself. And the highly preserved crystallinity of its samples is allowing the team to characterize their structure and behavior — especially their intrinsic ferroelectricity — in ways not possible with lower-quality samples, whose grains and oxygen vacancies can seriously complicate analyses.
“We can start digging out a lot of this intrinsic property, because we can realize the high crystallinity and ferroelectricity at the same time,” said Xu, associate professor of physics and astronomy.
The atom-level specificity of those future investigations should prove especially useful, the team said, in overcoming the challenges that stand between hafnium oxide and a place in microelectronics. As of now, even the thinnest layers of the material demand a fair amount of voltage — more than most electrical engineers would prefer — before agreeing to reverse their polarization.
“In industry terms, that just means higher energy costs, so we want to reduce that,” Xu said. “But previously, all the data were based on low-quality samples. That high voltage issue may or may not be intrinsic, so if we follow the work we have done here, there may be a route to circumvent or solve that problem.
“We want to follow this road to solve a few problems before hafnium oxide really gets applied in devices.”
That the team has gotten as far down the road as it has, the researchers said, is a credit to merging its three lanes of expertise: the thin film-growing prowess of Xu’s group, the theoretical acumen of Tsymbal’s, and the ferroelectricity know-how of Gruverman’s.
“Without our complementary expertise,” Xu said, “this discovery would have been impossible.”
According to Tsymbal, the latest pit stop on the three-year trip has been well worth it.
“There were a lot of interactions and a lot of discussions before we came to this conclusion,” he said. “It was quite a long way to go, but it was really very rewarding, because I think it is a significant breakthrough for the community. It completely changes the vision and understanding of this material.”
Xu, Gruverman and Tsymbal authored the Nature Materials study with Nebraska’s Jeffrey Shield, Yu Yun, Pratyush Buragohain, Ming Li, Zahra Ahmadi, Xin Li, Haohan Wang, Jing Li and Lingling Tao; Purdue University’s Haiyan Wang and Yizhi Zhang; and Ping Lu of Sandia National Laboratories. The team received support from the National Science Foundation and the U.S. Department of Energy.