SEOUL, November 12 (AJP) - Tiny bumps on an electrode, far too small to see, can decide where electricity flows inside a next-generation memory chip, South Korean researchers found after reconstructing the chip's inner current path in three dimensions.
KAIST said Thursday that a team led by Hong Seung-bum, a professor in its Department of Materials Science and Engineering, working with fellow materials science professor Park Sang-hee's group, observed the current path inside a resistive memory device in full and traced where it forms and connects to the shape of the electrode surface beneath it.
The device they studied, called resistive random access memory (RRAM), stores information by creating or breaking a path for electricity inside a thin insulating material. When a fine conductive thread, known as a conductive filament, bridges the layer, current flows. When the thread breaks, current stops, and the two states store the digital values of one and zero.
RRAM keeps data even when the power is off, has a simple structure and switches quickly, which makes it a leading candidate for next-generation storage. Researchers are also studying it for hardware that processes information the way the human brain does, a direction tied to artificial intelligence chips.
The trouble is that the filament does not form in the same place or shape every time. When the path changes, the voltage needed to write or erase data changes with it, and the memory becomes harder to operate reliably. Knowing how the path takes shape is the starting point for fixing that.
The team approached the question the way a CT scanner builds an image of the body, by stacking many cross-sections. Using the hard diamond tip of a conductive atomic force microscope, the researchers shaved a titanium dioxide layer 10 nanometers thick, removing about 0.1 nanometer per pass over 100 passes and mapping the current at each depth. One nanometer is one-billionth of a meter.
Stacked together, the current maps revealed the filament's full shape. It was wide near the probe at the top and narrowed toward the platinum electrode at the bottom, like a funnel, with a sudden pinch just below the probe.
The most striking result was where the filament ended. Its tip tended to land on nanoscale protrusions on the platinum electrode rather than on flat areas, a pattern the team confirmed across five filaments formed at different locations. Electrodes look flat to the eye, but at that scale their surfaces rise and fall.
The researchers attributed the pattern to the electric field, which concentrates around pointed features in the same way it gathers at the tip of a lightning rod. Computer simulations backed the reading, showing the field strengthening around the protrusions on the bottom electrode as well as under the probe above, a double concentration that steers the filament's growth.
The team also examined how current flowed within the filament. Titanium dioxide can lose oxygen atoms, leaving vacant sites that help carry current, and the researchers interpreted the current data as showing these vacancies gathering toward the filament's core near the bottom electrode. The pinch near the top, they said, could arise as new vacancies form while existing ones migrate downward.
The work remains basic research. The team has not yet demonstrated that deliberately shaping an electrode can place filaments at chosen spots in a working device, and the behavior of the oxygen vacancies was inferred from current measurements rather than observed directly.
If the bumps can eventually be controlled in manufacturing, filaments could be guided to form where designers want them, making write and erase operations more uniform. The team plans to test how the paths change across different electrode materials, surface shapes, oxide compositions and operating conditions.
"This study goes beyond simply observing the three-dimensional shape of conductive filaments. Its significance lies in connecting their shape and end positions to the actual nanostructure of the electrode," Hong said. "If the surface structure of the electrode is considered as a design variable going forward, we expect this to become an important foundation for developing next-generation resistive memories that operate more reliably."
Park Kun-woo, a student in KAIST's integrated master's and doctoral program, was the first author of the paper, and Hong was the corresponding author. The findings were published online in Nano Letters on Aug. 11. The research was supported by a National Research Foundation of Korea grant and an industry-academia project with Samsung Electronics' Device Solutions Division.
[Reference Information]
Journal/Source: Nano Letters
Title: Three-Dimensional Visualization of Conductive Filaments in TiO2 via Conductive Atomic Force Microscopy Tomography
Link/DOI: https://doi.org/10.1021/acs.nanolett.6c02423
Copyright ⓒ Aju Press All rights reserved.