Science & Technology

Accelerate the discovery of new materials for 3D printing

The growing popularity of 3D printing for manufacturing all kinds of items, from customized medical devices to affordable homes, is driving demand for new 3D printing materials designed for very specific applications. ..

To reduce the time it takes to discover these new materials, MIT researchers use machine learning to optimize new 3D printing materials with multiple properties such as toughness and compressive strength. We have developed a driven process.

By streamlining material development and reducing the amount of chemical waste, the system reduces costs and reduces its impact on the environment. Machine learning algorithms can also drive innovation by proposing unique chemicals that human intuition may miss.

“Material development is still a very manual process. Chemists enter the lab, manually mix the ingredients, make samples, test them and reach the final formulation, but. Our system can perform hundreds of iterations in the same period, rather than having a chemist who can only iterate a few times in a few days, “says Mike Foshey, mechanical engineer and project manager for computational design and manufacturing. Mr. says. He is a group (CDFG) of the Institute for Computer Science and Artificial Intelligence (CSAIL) and is the co-author of this paper.

Additional authors include Timothy Erps, co-author of CDFG’s Technical Associate. Minako Nakobitchirukovich, CSAIL Postdoc. Wan Shou, a former MIT postdoc and now an assistant professor at the University of Arkansas. Senior author Wojciech Matusik, professor of electrical engineering and computer science at MIT. These are Hanns Hagen Geotzke, Herve Dietsch and Klaus Stoll from BASF.Research today Science Advances..

Discovery optimization

In a system developed by researchers, optimization algorithms perform much of the trial-and-error discovery process.

The material developer selects several components and inputs the details of their chemical composition into the algorithm to define the mechanical properties that the new material should have. The algorithm then increases or decreases the amount of these components (such as turning the knobs on the amplifier) ​​to see how each equation affects the properties of the material before reaching the ideal combination.

The developer then mixes, processes, and tests the sample to see how the material actually works. The developer reports the result to the algorithm. The algorithm automatically learns from the experiment and uses the new information to determine another formulation to test.

“I think this is better than traditional methods because many applications rely heavily on optimization algorithms to find the best solution. Specializing in pre-selecting material formulations. You don’t have to have a chemist at hand, “says Foshey.

Researchers have created a free open source material optimization platform called AutoOED It incorporates the same optimization algorithm. AutoOED is a complete software package that allows researchers to perform their own optimizations.

Make material

Researchers have used the system to test the system by optimizing the formulation of a new 3D printing ink that cures when exposed to UV light.

They set the objective of the algorithm to identify the six chemicals used in the formulation and to identify the material with the best performance in terms of toughness, compressive modulus (rigidity), and strength.

Manually maximizing these three properties is especially difficult because they can conflict. For example, the strongest material may not be the hardest. Chemists typically use a manual process to try to maximize one property at a time, resulting in a lot of experimentation and a lot of waste.

After testing only 120 samples, the algorithm came up with 12 best performing materials with optimal trade-offs for 3 different properties.

Foshey and his collaborators are amazed at the wide variety of materials that the algorithm can produce, and state that the results are far more diverse than expected based on the six components. This system facilitates exploration. This can be especially useful in situations where certain material properties cannot be found intuitively and easily.

Faster in the future

You can further accelerate the process by using additional automation. Researchers have manually mixed and tested each sample, but robots may operate the dispensing and mixing system in future versions of the system, says Fossy.

In the future, researchers want to test this data-driven discovery process for applications other than developing new 3D printing inks.

“This generally has a wide range of applications throughout materials science. For example, if you want to design a new type of battery with higher efficiency and lower cost, you can do that with such a system. If you want to optimize the painting of a car that has good performance and is environmentally friendly, you can do that with this system as well, “he says.

Keith A. Brown, an assistant professor of mechanical engineering at Boston University, presents a systematic approach to identifying the best materials, which could be a major step towards achieving high-performance structures. It states that there is.

“It is especially encouraging to focus on new material formulations, as this is a factor that is often overlooked by researchers who are constrained by commercial materials. Also, data-driven methods and experimental science. The combination allows the team to identify the material in an efficient way. Since the efficiency of the experiment is identifiable by all experimenters, the method here is more data-driven by the community. It can motivate you to do it, “he says.

The study was supported by BASF

Accelerate the discovery of new materials for 3D printing Accelerate the discovery of new materials for 3D printing

Back to top button