MIT engineers have developed a printable aluminum alloy that can withstand high temperatures and is five times stronger than traditionally produced aluminum.
The up-to-date printable metal is made from a mixture of aluminum and other elements identified by the team using a combination of simulation and machine learning, which significantly reduced the number of possible material combinations to search through. While classic methods would require simulating more than 1 million possible material combinations, the team’s up-to-date machine learning approach only required evaluating 40 possible compositions before identifying the ideal mix for the high-strength, printable aluminum alloy.
After printing the alloy and testing the resulting material, the team confirmed that, as predicted, the aluminum alloy was as robust as the strongest aluminum alloys currently produced by classic casting methods.
Scientists predict that the up-to-date printable aluminum could be used to make stronger, lighter and temperature-resistant products, such as fan blades in jet engines. Fan blades are traditionally cast from titanium – a material that is more than 50 percent heavier and up to 10 times more costly than aluminum – or made from advanced composites.
“If we could use a lighter, high-strength material, we would save a significant amount of energy in the transportation industry,” says Mohadeseh Taheri-Mousavi, who led the work as a postdoc at MIT and is now an assistant professor at Carnegie Mellon University.
“Because 3D printing can produce complex geometries, save material, and enable unique designs, we see this printable alloy as something that could be used in advanced vacuum pumps, high-end automobiles, and cooling equipment for data centers,” adds John Hart, Class of 1922 professor and chair of the Department of Mechanical Engineering at MIT.
Hart and Taheri-Mousavi provide details on up-to-date a-printable aluminum design article published in the magazine . Co-authors of the MIT paper include Michael Xu, Clay Houser, Shaolou Wei, James LeBeau and Greg Olson, as well as Florian Hengsbach and Mirko Schaper of the University of Paderborn in Germany and Zhaoxuan Ge and Benjamin Glaser of Carnegie Mellon University.
Micro size
The up-to-date work is based on a class at MIT that Taheri-Mousavi took in 2020 and taught by Greg Olson, professor of practice in the Department of Materials Science and Engineering. During the classes, students learned to exploit computational simulations to design alloys with high performance parameters. Alloys are materials made from a mixture of various elements, the combination of which gives the material as a whole exceptional strength and other unique properties.
Olson challenged the class to design an aluminum alloy that was stronger than the strongest printable aluminum alloy ever designed. As with most materials, the strength of aluminum depends largely on its microstructure: the smaller and denser its microscopic, or “precipitable” components, the stronger the alloy will be.
With this in mind, the class used computer simulations to methodically combine aluminum with various types and concentrations of elements to simulate and predict the strength of the resulting alloy. However, the exercise did not produce stronger results. At the end of the class, Taheri-Mousavi wondered: Could machine learning do better?
“At some point, many factors influence the properties of the material non-linearly and you get lost,” says Taheri-Mousavi. “With machine learning tools, they can tell you what you need to focus on and say, for example, these two elements control that feature. This allows you to explore the design space more efficiently.”
Layer by layer
In the up-to-date study, Taheri-Mousavi picked up where Olson left off, this time trying to find a stronger recipe for an aluminum alloy. This time, it applied machine learning techniques designed to efficiently sift through data such as item properties to identify key connections and correlations that should lead to a more desirable outcome or product.
She found that using just 40 compositions combining aluminum with various elements, the machine learning approach quickly led to the development of a recipe for an aluminum alloy with a higher volume fraction of diminutive precipitates and therefore greater strength than what had been found in previous studies. The alloy’s strength was even higher than what was identified after simulating over 1 million possibilities without the exploit of machine learning.
The team realized that to physically produce this up-to-date, robust, low-precipitation alloy, the best solution would be 3D printing instead of classic metal casting, in which molten liquid aluminum is poured into a mold and then allowed to chilly and harden. The longer this cooling time is, the greater the likelihood of individual precipitate growth.
Scientists have shown that 3D printing, also commonly known as additive manufacturing, can be a faster way to chilly and solidify an aluminum alloy. Specifically, they considered laser powder fusion (LBPF), a technique by which powder is deposited layer by layer on a surface in a desired pattern and then rapidly fused using a pattern-tracking laser. The fused pattern is slim enough to harden quickly before another layer is applied and similarly “printed.” The team found that the inherently rapid cooling and solidification of LBPF produced an aluminum alloy with low precipitates and high strength, as predicted by the machine learning method.
“Sometimes we need to think about how to make a material compatible with 3D printing,” says study co-author John Hart. “In this case, 3D printing opens new doors due to the unique features of the process – in particular, the high cooling rate. Freezing the alloy very quickly after melting it with a laser creates this special set of properties.”
To bring their idea to life, the researchers ordered a printable powder formula based on a up-to-date aluminum alloy recipe. They sent the powder — a mixture of aluminum and five other elements — to collaborators in Germany, who printed diminutive samples of the alloy using their own LPBF system. The samples were then sent to MIT, where the team performed multiple tests to measure the alloy’s strength and visualize the samples’ microstructure.
The results confirmed predictions made from initial machine learning explorations: the printed alloy was five times stronger than its cast counterpart and 50 percent stronger than alloys designed using conventional simulations without machine learning. The microstructure of the up-to-date alloy also consisted of a larger volume fraction of diminutive precipitates and was stable at high temperatures up to 400 degrees Celsius – a very high temperature for aluminum alloys.
Scientists are using similar machine learning techniques to further optimize other properties of the alloy.
“Our methodology opens new doors for anyone who wants to design feet in 3D printing,” says Taheri-Mousavi. “My dream is that one day, passengers looking out of an airplane window will see engine fan blades made of our aluminum alloys.”
This work was performed in part using MIT.nano characterization tools.