NSF Backs Texas A&M Tool That Cuts Materials Checks to a Day

TL;DR: Texas A&M University reported on Sept. 4, 2026, that professor Raymundo Arróyave received a National Science Foundation award to build PhaseForge, free open software that applies machine learning to CALPHAD thermodynamic modeling and can shrink work once measured in years to one day.

A materials science and engineering professor at Texas A&M University has received a National Science Foundation award to build free, open software for faster materials screening, according to NSF backs Texas A&M tool that speeds materials discovery - Engineering News. The project, called PhaseForge, uses machine learning to answer what the university describes as a question every materials scientist asks: whether a proposed material can actually be made.

The announcement was published Sept. 4, 2026, and written by contributor Leon Contreras for Texas A&M Engineering News. It states that the new award will fund software that compresses materials calculations once measured in years into a single day. That time comparison frames the entire effort, from graduate training to spacecraft parts.

PhaseForge is led by Dr. Raymundo Arróyave, professor in the Department of Materials Science and Engineering at Texas A&M University. The software applies machine learning to a decades-old method called CALPHAD, short for calculation of phase diagrams. The university report explains that CALPHAD tells researchers whether a material they want to build can exist at all.

In the announcement, Arróyave links that existence question directly to thermodynamics. “In materials science, the first thing you need to know is whether a material you want to make indeed can be made,” Arróyave said, according to Engineering News. “The answer to that question is given by thermodynamics.” The university presents that statement as the scientific starting point for PhaseForge.

The project is aimed at inorganic materials. Three early application areas are named: thermal protection systems for spacecraft, shape memory alloys, and ultra-hard coatings that shield other materials from wear. The code is being distributed free and openly online, with the code available on GitHub and tutorials and training materials planned. Anyone can contribute to it, the university states.

What the Sept. 4 announcement confirmed

The Sept. 4 report from Texas A&M Engineering News provides the core verified details of the award. A Texas A&M University materials science and engineering professor received a National Science Foundation award to build free, open software. The project name is PhaseForge. The leader is Dr. Raymundo Arróyave.

The report was issued as a three-minute research item under the categories of artificial intelligence, faculty, materials and manufacturing, and materials science and engineering. The listed media contact is Alyson Chapman of the Engineering News staff. The announcement does not list an award number or dollar figure in the text published by the college.

The stated purpose is narrow and practical. The software will fund free, open tools that use machine learning to model thermodynamics and guide discovery of new materials. An introductory image caption in the same report notes that researchers are using open-source machine learning tools to predict material properties and guide discovery with unprecedented speed.

The Texas AI Docket summary of the same announcement, titled National Science Foundation backs a Texas A&M tool that screens whether an alloy can exist before anyone makes it, likewise records that Texas A&M’s College of Engineering published on Sept. 4, 2026, that the National Science Foundation is funding PhaseForge from the Department of Materials Science and Engineering. That second record notes the software applies machine learning to the CALPHAD phase diagram method and is described by the university as free and openly available.

Why thermodynamics decides what can be built

Materials scientists cannot simulate much about a candidate compound without a thermodynamic foundation, according to the Texas A&M account. The report explains that building those thermodynamic models takes significant time and deep expertise. Most material systems simply do not have one.

Without that foundation, the announcement states, hardly anything about a material can be simulated, leaving researchers to work by experiment alone. That limitation shapes laboratory choices, because trial-and-error synthesis is slower and more constrained than computation first followed by targeted testing.

Arróyave describes the experience of working without a model as navigating without light. He compares it to finding a way through a house in total darkness with a tiny flashlight, according to Engineering News. The comparison underscores why the group treats thermodynamic assessment as the first step rather than a later check.

The CALPHAD approach itself is not new. The university describes it as a decades-old method for calculation of phase diagrams. Its role is to formalize whether a desired composition and structure can persist under given conditions. PhaseForge does not replace that question; it seeks to answer it faster through machine learning.

The broader importance of such groundwork is echoed in a Sept. 2, 2026, NSF explainer titled Materials science: The stuff of the modern world. That explainer states that materials science and engineering explores basic structure, properties and behavior of materials in bulk form, as well as on molecular, atomic and even subatomic levels. It lists semiconductors, metals, ceramics, polymers, composites and more among the materials studied.

The same NSF explainer states that understanding helps explain why materials fail, how they can be made stronger or more durable, how they can exhibit new capabilities, and how entirely new materials can be designed for specific uses. It describes materials as building blocks of modern technology, from smartphone screens and silicon computer chips to artificial joints and aircraft components.

From a full Ph.D. to one day

The most striking claim in the Texas A&M report concerns time. When Arróyave was a graduate student, such an analysis would have taken an entire Ph.D., he said. Now, with the new method, it could be done in one day, according to Engineering News.

The report presents that shift as already demonstrated in early work, not only as a future goal. When his students and a postdoctoral researcher first showed him that the method worked, he said he had trouble sleeping that night. The payoff arrived faster than he expected, the university writes.

The announcement says the project can compress materials calculations once measured in years into a single day. That wording covers thermodynamic assessments that previously required long manual calibration and expert judgment. The university does not state that every materials calculation falls to one day, only that the supported calculations can be compressed on that scale.

The distinction matters for readers tracking laboratory productivity. A model that once demanded a doctoral-length effort blocked simulation of related properties. A one-day foundation would let teams move sooner to property prediction, experimental planning, and iteration.

The university’s image caption reinforces that sequencing, stating that open-source machine learning tools are used to predict material properties and guide discovery of new materials. PhaseForge fits that pattern by targeting the thermodynamic step that precedes detailed simulation.

Three early uses with practical stakes

The Texas A&M announcement names inorganic materials as the focus and lists three areas with immediate practical stakes. The first is thermal protection systems for spacecraft. The second is shape memory alloys. The third is ultra-hard coatings that shield other materials from wear.

Those examples connect thermodynamic screening to hardware that must survive heat, deformation, or abrasion. Thermal protection must remain stable when spacecraft encounter high temperatures. Shape memory alloys must return to a remembered form after deformation. Ultra-hard coatings must protect underlying material from wear.

The announcement does not claim that PhaseForge has already designed a flight-ready tile, a commercial alloy, or a finished coating. It states that the tool targets those classes of inorganic materials because the underlying stability question is central to each. The outcomes known by Sept. 4 are the award, the leadership, the method, the open distribution, and the planned training materials.

That framing aligns with NSF’s Sept. 2 description of remaining opportunities. The agency writes that the next generation of technologies will require materials that do not yet exist. With NSF support, the agency states, researchers are designing materials for new quantum technologies, for making energy-storage systems more efficient and durable, and to maintain performance in environments under extreme heat, pressure and radiation, such as spaceflight, deep-ocean exploration and nuclear applications.

NSF also states that efforts are being made to improve processes for developing materials, such as reducing reliance on finite critical minerals through recyclable and bio-based alternatives and using AI and autonomous laboratories to speed up discovery. PhaseForge belongs to that process-improvement category, using AI to accelerate assessment rather than claiming a new compound.

Who is leading PhaseForge

PhaseForge is led by Dr. Raymundo Arróyave, professor in the Department of Materials Science and Engineering at Texas A&M University, according to Engineering News. Arróyave credits the project’s origins to a hunch and a group culture that rewards experiments.

“We always like to try stuff and to work together,” he said, according to the university report. The report adds that everyone in the group, himself included, contributes to projects on the side. That side-project culture is presented as the setting where the CALPHAD-plus-learning idea was tested.

Department head Dr. Ibrahim Karaman is quoted in support of the approach. “PhaseForge shows how machine learning can help metallurgists answer fundamental questions much faster,” Karaman said, according to Engineering News. The same passage continues that open availability will broaden use.

“By making the tool free and openly available, Dr. Arróyave will help accelerate materials discovery across academia, national laboratories and industry,” Karaman said. The university presents that cross-sector use — campuses, national laboratories, and companies — as the intended audience.

The award itself will support one doctoral student and one postdoctoral researcher, the announcement states. That staffing detail indicates a focused development team rather than a large center. It also matches the origin story involving students and a postdoctoral researcher demonstrating the first result.

Collaborators named in the announcement include Karaman, associate department head and professor Dr. Miladin Radovic, Brandon Bocklund of Lawrence Livermore National Laboratory, and industry partners. The university does not name the industry partners in the published text. It identifies Bocklund’s affiliation as Lawrence Livermore National Laboratory.

Free code, open contributions, planned training

Distribution is a central part of the award. PhaseForge is being distributed free and openly online, and anyone can contribute to it, according to Engineering News. The code is available on GitHub. Tutorials and training materials are planned.

That open model means other teams can inspect, reuse, and extend the software. The university does not describe license terms beyond free and open availability and the ability for anyone to contribute. It states that instructional materials are planned, not already released.

The emphasis on openness is repeated by department leadership. Karaman’s quoted assessment ties acceleration of discovery to making the tool free and openly available. The Texas AI Docket record similarly emphasizes that the university states the tool is free and openly available and that open availability is expected to help accelerate discovery across academia, national laboratories and industry.

For potential users, the practical path described by Sept. 4 is to use the online code and await tutorials and training materials. The announcement lists Alyson Chapman and [email protected] as the media contact for the news item. It does not list a separate help desk or release version in the published text.

How this fits wider U.S. materials efforts

The PhaseForge award arrives during a period of sustained federal attention to materials methods, as documented by NSF’s Sept. 2 explainer. That explainer traces federal investments that helped connect chemistry, physics, engineering and metallurgy into modern materials science. It notes that in the 1950s and 1960s NSF invested in foundational work in areas such as solid-state physics and polymers.

By the early 1970s, the agency had become one of the nation’s top supporters of materials research at universities, assuming leadership of a federal research portfolio previously spearheaded by the Department of War, according to the same NSF page. Through sustained investments including the Materials Research Science and Engineering Centers program, NSF support contributed to breakthroughs the page enumerates.

Those listed contributions include the Nobel Prize-winning discovery of conductive polymers, the Nobel Prize-winning development of blue LEDs, hydrogels and 3D-printed biomaterials used for contact lenses, wound care and artificial organs, and antibacterial coatings that protect crops from disease and drought. The list continues with self-healing materials for medical, construction and aerospace applications and nanoparticle-based materials that can absorb toxic chemicals, harvest water from desert air or enable real-time tracking of biomolecules.

Further items on that NSF list include superalloys that better withstand intense heat of demanding aerospace and energy production applications, transparent conductive ceramics that enable touch-screen devices, stents and other implantable devices made from biodegradable alloys, new methods for stacking and combining materials to create smaller and more powerful microchips, and solar cells that self-heal radiation damage. The page presents those as prior outcomes of long-term support, not as PhaseForge results.

A separate Sept. 3 highlight from Ames Laboratory shows parallel interest in faster alloy software. That item, titled The Alloy Optimization Software technology, states that The Alloy Optimization Software, developed at Lawrence Livermore National Laboratory, was named a 2026 R&D 100 Finalist. It describes the tool as a user-friendly, commercialized tool that enables computational design of optimal alloys in minutes to hours. That time scale is stated by Ames Laboratory for that software, not for PhaseForge.

Taken together, the records available by Sept. 6 describe a shared direction: pair established thermodynamic and design methods with computation to shorten screening. PhaseForge applies machine learning to CALPHAD for inorganic systems at Texas A&M with NSF funding and open code. The award supports continued development, collaboration with Lawrence Livermore National Laboratory and industry partners, and training for a wider community.