Without a map, it is virtually impossible to determine not only where you are, but also where you are going, especially the properties of materials.
For decades, scientists have understood that while bulk materials behave in certain ways, these rules can break down for materials on the micro- and nano-scale, often in surprising ways. One of those surprises was the discovery that for some materials, applying even compact amounts of stress—a concept known as elastic strain engineering—to the materials can dramatically improve certain properties, provided that the strains remain elastic and do not relax through plasticity, fracture, or phase transformations. Micro- and nano-scale materials are particularly good at holding applied stresses in an elastic form.
Exactly how these elastic deformations (or, by extension, residual stresses) are applied to obtain specific material properties, however, was less clear – until recently.
Using a combination of first-principles computation and machine learning, a team of MIT researchers has produced the world’s first roadmap showing how to tune crystalline materials to achieve specific thermal and electronic properties.
Conducted by Yu LiBattelle Energy Alliance professor of nuclear engineering and professor of materials science and engineering, the team described a framework for understanding exactly how changing a material’s elastic deformation can fine-tune properties such as thermal and electrical conductivity. The work is described in an open-access article published in .
“For the first time, we’ve been able to use machine learning to fully map out the six-dimensional ideal strength limit, which is the upper limit of elastic deformation engineering, and map out these electronic and phononic properties,” Li says. “Now we can use this approach to explore many other materials. Traditionally, people create new materials by changing their chemistry.”
“For example, in a ternary alloy, you can change the percentage of two elements, so you have two degrees of freedom,” he continues. “We showed that diamond, with only one element, is equivalent to a six-element alloy because it has six degrees of freedom for elastic deformation that can be independently tuned.”
Miniature burdens, huge material benefits
The paper builds on foundations laid in the 1980s, when scientists first discovered that the performance of semiconductor materials doubles when a diminutive amount of elastic deformation—just 1 percent—is applied to the material.
The discovery was quickly commercialized by the semiconductor industry and is now being used to boost the performance of microprocessors in devices ranging from laptops to cell phones, says Subra Suresh, emeritus professor of engineering at the Vannevar Bush School of Engineering. But the level of overhead is diminutive compared to what we can achieve now.
In 2018 In their paper, Suresh, Dao, and colleagues showed that 1 percent strain is only the tip of the iceberg.
In a 2018 study, Suresh and colleagues first showed that diamond nanoneedles could withstand elastic deformation of up to 9 percent and still return to their original state. Several groups have since independently confirmed that microscale diamond can indeed elastically deform by about 7 percent in tension reversibly.
“Once we showed that we could bend diamonds on the nanoscale and create stresses of 9 or 10 percent, the question was what to do with that,” Suresh says. “It turns out that diamond is a very good semiconductor material… and one of our questions was, if we could mechanically stress diamond, could we reduce the band gap from 5.6 electron volts to two or three? Could we reduce it to zero, where it would start conducting like a metal?”
To answer these questions, the team first turned to machine learning to get a more exact picture of how deformation changes material properties.
“Strain is a big space,” Li explains. “You can have tensile strain, you can have shear strain in many directions, so it’s a six-dimensional space, and the phonon band is three-dimensional, so there are nine tunable parameters in total. So for the first time, we’re using machine learning to create a complete map to navigate electronic and phonon properties and identify boundaries.”
Armed with this map, the team then showed how stress could be used to radically change the semiconducting properties of diamond.
“Diamond is like the Mount Everest of electronic materials,” Li says, “because it has very high thermal conductivity, very high dielectric strength, very high carrier mobility. We showed that we can control the flattening of Mount Everest… so we show that by engineering stress, you can either improve the thermal conductivity of diamond by a factor of two or significantly worsen it by a factor of 20.”
Up-to-date map, recent apps
According to Li, in the future, the research results could be used to investigate a range of exotic material properties, from drastically reduced thermal conductivity to superconductivity.
“Experimentally, these properties are already available for nanoneedles and even microbridges,” he says. “And we’ve seen exotic properties like reducing (the thermal conductivity of) diamond to just a few hundred watts per meter-kelvin. Recently, people have shown that you can make room-temperature superconductors with hydrides if you squeeze them down to a few hundred gigapascals, so we’ve discovered all sorts of exotic behaviors once we have the map.”
The results could also influence the design of next-generation computer chips, which could run much faster and cooler than today’s processors, as well as quantum sensors and communications devices. As the semiconductor industry moves toward denser architectures, Suresh says the ability to tune a material’s thermal conductivity will be especially essential for heat dissipation.
While the paper could provide insight into the design of future generations of microprocessors, Zhe Shi, an assistant professor in Li’s lab and the paper’s lead author, says more work is needed before such chips make it into the average laptop or cellphone.
“We know that a 1 percent load can give you an order of magnitude increase in CPU clock speed,” Shi says. “There are a lot of manufacturing and device issues that need to be addressed to make that realistic, but I think this is definitely a great start. It’s an exciting start to what could lead to significant advances in technology.”
This work was funded by the Defense Threat Reduction Agency, an NSF Graduate Research Fellowship, the Department of Biological Sciences at Nanyang Technological University, the National Science Foundation (NSF), the Vannevar Bush Chair at MIT, and the Nanyang Technological University Distinguished University Endowed Chair.