Manufacturing Bits: Dec. 1

New phase-change materials
The National Institute of Standards and Technology (NIST) has developed an open source machine learning algorithm for use in discovering and developing new materials.

NIST’s technology, called CAMEO, has already been used by researchers to discover a new phase-change memory material. CAMEO, which stands for Closed-Loop Autonomous System for Materials Exploration and Optimization, can be used to develop other materials. The code for CAMEO is open source and will be freely available for use by scientists and researchers.

Typically, to develop new materials and other technologies, researchers follow a trial-and-error methodology. But this is often a complex and time-consuming process.

That’s where machine learning fits in. A subset of artificial intelligence (AI), machine learning, uses advanced algorithms in systems to recognize patterns in data as well as to learn and make predictions about the information.

Credit: N. Hanacek / NIST: CAMEO, short for Closed-Loop Autonomous System for Materials Exploration, is an AI algorithm developed by a multi-institutional team including researchers from NIST.

Using machine learning, CAMEO looks for new materials by operating in a closed loop. “It determines which experiment to run on a material, does the experiment, and collects the data,” according to NIST. “It can also ask for more information, such as the crystal structure of the desired material, from the scientist before running the next experiment, which is informed by all past experiments performed in the loop.”

In the lab, researchers from NIST installed CAMEO in a computer. The system connects to an X-ray diffraction system in a facility over a data network. In this case, the experiments were performed at the Stanford Synchrotron Radiation Lightsource (SSRL).

The SSRL, part of the SLAC National Accelerator Laboratory, is a third-generation storage ring. SSRL’s bright X-rays are a resource for researchers to study technologies at the atomic and molecular level. In X-ray diffraction, X-rays hit a sample at different angles, enabling researchers to figure out the crystal structures of the sample.

In the lab, researchers wanted CAMEO to find the best germanium, antimony and tellurium (Ge-Sb-Te) alloy. The goal was to find the best contrast between the crystalline and amorphous states.

In the experiment, CAMEO was given 177 potential materials to investigate. CAMEO performed 19 different experimental cycles, which took 10 hours, compared with the estimated 90 hours it would have taken a scientist with the full set of 177 materials, according to NIST.

Using this approach, researchers discovered the material Ge4Sb6Te7 or GST467. GST467 has twice the optical contrast of Ge2Sb2Te5, a well-known material used for DVDs. GST467 also has applications for photonic switching devices, memory chips and other applications.

The technology could be used to find other new materials. But needless to say, running experiments at a synchrotron facility is expensive.

So, NIST enables researchers to utilize the technology remotely. “This opens up a wave of scientists to still work and be productive without actually being in the lab,” said Apurva Mehta, a researcher at SLAC.

Metal-insulator transitions
Northwestern University has developed a new computational approach to accelerate the development of materials that exhibit metal-insulator transitions (MIT).

MITs are a class of electronic materials, which can be reversibly switched between electrically conducting and insulating states.

To develop these materials, Northwestern University has bypassed traditional machine learning-based discovery models. Conventional machine learning methods are limited due to the lack of data, making the design of new MIT materials difficult, according to researchers.

Instead of the traditional approach, Northwestern has developed integrated techniques using statistical inference, optimization theory, and computational materials physics. It combines multi-objective Bayesian optimization with latent-variable Gaussian processes.

Using this technology, researchers developed a family of MIT materials called complex lacunar spinels. “This advance overcomes traditional limitations imposed by chemical intuition-based materials designs,” said Wei Chen, the Wilson-Cook Professor in Engineering Design and professor and chair of mechanical engineering at Northwestern. “By reframing functional materials design as an optimization problem, we have not only found a solution to the challenge of working with limited data, but also demonstrated the ability to efficiently discover optimal new materials for future electronics.”

“Our method paves the way forward for optimization of multiple properties and the co-design of complex multifunctional materials where prior data and knowledge is sparse,” added James Rondinelli, professor of materials science and engineering and the Morris E. Fine Professor in Materials and Manufacturing at the McCormick School of Engineering at Northwestern.

AI lab
The Tokyo Institute of Technology has developed a system that combines robotics and artificial intelligence to develop new materials.

This approach can produce and test compounds ten times faster than doing manual work, allowing for the rapid creation of huge shared databases. In turn, the autonomous system and database will be used to discover exotic material properties and new laws of physics.

“The training of future materials scientists must evolve; they will need to understand what machine learning can solve and set the problem accordingly. The strength of human researchers lies in creating concepts or identifying problems in society. Combining those strengths with machine learning and robotics is very important,” said Taro Hitosugi, a professor at Tokyo Institute of Technology.