A research team led by Prof. HU Weijin from the Institute of Metal Research (IMR) of the Chinese Academy of Sciences has discovered that single-domain ferroelectric thin films can be efficiently achieved by simply elevating the growth temperature. 

Their findings, published in Advanced Functional Materials, offer a straightforward alternative to conventional complex fabrication methods, with significant implications for ferroelectric device performance.

Ferroelectric materials naturally form polydomain structures to minimize electrostatic energy. Nevertheless, single-domain thin films can be achieved through precise control of interfacial atomic layers or strain gradients. The quest for a simple method to obtain single-domain and its impact on the ferroelectric device performance are of great interest.

In this study, the researchers demonstrated that the ferroelectric single-domain, compared to polydomain, can significantly improve the the performance of ferroelectric synaptic devices. To explore this, they utilized BaTiO₃ (BTO) ferroelectric films deposited onto La_0.67Sr_0.33MnO₃ metallic layers at temperatures ranging from 700 ℃ to 850 ℃ using pulsed laser deposition, a technique well-suited for producing high-quality single- crystalline oxide films.

The researchers discovered that when LSMO was grown at temperatures above 800 ℃, the resulting BTO films exhibited a single-domain configuration. They observed that Sr ions tended to diffuse towards the interface, creating a positively charged surface that aligned the polarizations uniformly across the film.

"This approach is much simpler compared to other methods such as atomic layer engineering, which typically requires complex surface treatments and precise growth control," said HU. "It holds the potential for large-scale production of single-domain films beyond the current 5 mm × 5 mm size, possibly through industrial processes like spin coating."

Large-scale single-domain film could not only enhance the ferroelectric synapse performance used for neuromorphic computing, but also have applications in many other fields, including optoelectronics and catalysis.

The study was conducted in collaboration with scientists from IMR and South China Normal University.

The conception of neuromorphic computing (left from Baidu) and pattern recognition accuracy of a single domain ferroelectric synapse (Image by IMR)