Perspective view ofthe STMtopography ofnanocrystallineCufilmthat shows a valley with dissociated dislocations and a ridge with recombined dislocations. The size of the image is 50 nm by 50 nm. Credit: Xiaopu Zhang.
Schematic of a low angle tilt grain boundary.Nanocrystalline films of copper are not flat − and never can be − but possess an unexpected landscape of valleys and ridges, according to researchers [Zhang et al., Science 357 (2017) 397].
Copper and other metals are widely used for electrical contacts and interconnects because of their high electrical and thermal conductivity. But their properties are highly dependent on grain boundaries and dislocations at the surface. Until now, it had been thoughtthatthe surface of nanocrystallinemetals would be smooth on the nanoscale because grain boundaries and individual grains would coalesce.
But the first high-resolution study of a nanocrystalline Cu surface using scanning tunnelling microscopy (STM) carried out by John J. Boland at Trinity College Dublin, together with colleagues at Imperial College London, Intel, and the University of Pennsylvania, has revealed a very different story.
The three-dimensional visualization of the surface of the metal − and the grain boundaries − provided by their STM analysis shows a pattern of tilted grains creating a landscape of valleys and ridges. Far from being flat, the geometry of grain boundaries and dislocations present in the nanocrystalline metal films means that the surface can never be flat.
“The conventional view of small angle grain boundaries is that they are a collection of edge dislocations that are all perpendicular to the surface,” explains Boland. “But we found that the dislocation lines associated with the edge dislocations which comprise the grain boundary do not lie perpendicular to the surface of the material.”
In fact, the dislocation lines rotate in a low energy direction, causing a rotation in adjoining grains, which ultimately leads to roughening of the surface.
“It is likely that there is always some degree of out-of-plane rotation associated with grain boundaries,” continues Boland.
Until recently, scientists had mainly used transmission electron microscopy to analyse dislocations and grain boundaries in nanocrystalline metals, which provides a plan view of the surface and cannot reveal undulations at the surface. Now, with the advent of STM, the surface probe skates over the surface recording the landscape of peaks and troughs. These variations in grain boundary tilt and rotation will affect the thermal, electrical and mechanical properties of nanocrystalline Cu and other similar metals.
“Grain rotation is already contributing to many properties but has never been accounted for in any models,” points out Boland.
He and his team now plan to test out their model with large angle grain boundaries, which could have an even more profound effect on surface roughening and materials properties. But it is not necessarily bad news. Armed with this new knowledge, scientists might be able to control grain rotation and use it to tune materials properties.
“We are exploring approaches to control grain rotation and apply it to other metals and materials,” says Boland.
High-resolution surface topography measurements of defect-dominated materials are notoriously complicated and data interpretation is a delicate business, comments Robert Maass of the University of Illinois at Urbana-Champaign.
“But Boland’s work combines a beautifully conducted experiment with insightful atomistic simulations,” he comments. “They have shown, in unprecedented detail and with the incredible resolution of the order of partial dislocation vectors, how a complicated system can reduce its energy by exploiting the most flexible boundary conditions available, namely free surfaces.”
Maass expects the general implication, that flat nanocrystalline films cannot be made, should be similarly applicable to other nanocrystalline systems with a large surface-to-volume ratio.
“For which technological applications or processes this may be relevant is strongly connected to the magnitude of the topological modulations,” he adds.
This article was originally published in Nano Today (2017), doi: 10.1016/j.nantod.2017.08.003.