Molding is the classic way to shape materials, but can it be done at the nanoscale? Up to now, direct nanomolding has been limited to materials like polymers and amorphous metals that soften and begin to flow when heated above the glass transition temperature, whereas researchers have not succeeded with crystalline metallic materials. A research team led by Ze Liu of Wuhan University in China and Jan Schroers of Yale University has now reported in Physical Review Letters what they call a thermomechanical nanomolding technique that works for a wide range of metals and alloys and explained the diffusion mechanism behind their success.

“The ability to mold crystalline metals and alloys at the nanoscale will open a new era of applications and exploration of fundamental scientific phenomena,” says Golden Kumar, from the University of Texas at Dallas, who was not involved in the study. “The technique is extremely simple, yet versatile, and should be applicable to any metal or alloy as long as sufficiently fast diffusion is achievable.” Possible applications foreseen include high-surface-area metallic nanostructures for a diverse array of applications including catalysts, sensors, photovoltaics, microelectronics, and plasmonics.

Pouring molten metal into a mold to shape it, the initially obvious way to proceed, does not work at the nanoscale, owing both to the reactivity of the melted metal and capillary forces. The alternative, using an applied pressure to force a heated solid material into the mold requires the materials to “flow.” Unfortunately, the dimension of the flow unit in a crystalline metal that is being mechanically deformed is typically the grain size, often in the micrometer range, which is too large to flow into a nanometer-scale mold. As a result, researchers have been discouraged from making attempts to nanomold crystalline metals. The flow units in polymers and amorphous metals are in the nanometer range, accounting for their susceptibility to nanomolding.

According to Schroers, the research group was searching for a versatile technique to produce high-surface-area nanostructures that would be applicable to a broad range of metallic materials, in contrast to current manufacturing that relies on specialized processes for particular materials. Molding is inherently versatile, but to make it work for a wide range of crystalline metals at the nanoscale, the researchers looked for a flow mechanism that would be common to many materials. Diffusion was a possibility, but in the past it was discounted because it was thought to be so much slower than other candidate mechanisms for the deformation required for molding.

It turned out that the key was, in fact, the small dimension, with diffusion becoming dominant as the size decreased to the nanoscale. Diffusion in metals is typically driven by a chemical-potential gradient. For nanomolding, the researchers thought that the pressure gradient between the top and bottom of the mold might be a sufficient driving force.

The first evidence in support of their thinking was in a 2017 publication by Liu, who reported direct molding of nanorods from several metals with dimensions as small as 8 nm and aspect ratios as high as 2000. The report also suggested that diffusion-dominated creep was the enabling mechanism. “But at the time, we didn’t know how small we could go nor how the metals flowed into nanocavities during molding. By designing size-dependent molding experiments and systematically investigating the growth dynamics of metallic nanorods, we confirmed a diffusion-dominated mechanism driven by a pressure gradient,” says Liu, “but displacive deformation can also be involved, especially at high pressure.”

For their experiments, the researchers used a hard Al₂O₃ mold with an array of pits of decreasing diameters where the nanorods would be produced when the target material was placed over the mold, heated to around half the melting temperature or a bit higher, and then subjected to a pressure of several hundred MPa for a few minutes. After molding, the nanorods were extracted by dissolving the mold. The group successfully tried a variety of metals and alloys with different compositions and crystal structures, ranging from single elements to six-element high-entropy alloys. “The only limitation,” Schroers says, “is that the alloy must be a solid solution.”

The aspect ratio, defined as the mold depth, L, divided by the mold diameter, d, was a key parameter in analyzing the results and isolating the operative diffusion mechanism. Paradoxically, the achievable aspect ratio increased as the mold diameter decreased, in contrast to previous experience, where the aspect ratio typically decreased with decreasing mold diameter. For example, the group measured the aspect ratio for thermomechanical molding of a bulk metallic glass and found that it decreased as the diameter dropped below 100 nm.

Could a mechanical deformation mechanism such as dislocation slip, grain-boundary sliding, or grain-boundary rotation be at the root of this behavior? The group ruled these out by considerations such as the decreased dislocation nucleation rate as the diameter decreased and the large grain size. For example, high-resolution transmission electron microscopy showed that there were no grain boundaries on average for grains larger than the mold diameter. Twin boundaries were found but they were too sparse to dominate behavior.

The researchers then turned to modeling the diffusion behavior based on the growth dynamics of the nanorods. They determined that the pressure gradient caused by the high pressure at the mouth of the cavity and the low pressure at the tip of the growing nanorod generated a corresponding vacancy gradient with more vacancies at high pressures than low. This gradient was the driving force for the diffusion-based growth of the nanorods. “Taken together, all the results of our tests, the scaling relationships, and the absolute values of our measurements all point to diffusion,” Schroers says. Kumar agrees, saying, “The experimental observations and the proposed diffusion model are very robust.”

Read the abstract in Physical Review Letters.