Northwestern Team Finds Atomic-Level Lever for Metal 3D Printing Control

Metal 3D printing has always struggled with a specific blind spot: what is actually happening inside the melt pool while the laser is moving. A team at Northwestern University has closed that gap with an operando X-ray technique that watches atomic structure form in real time, and the findings rewrite part of the rulebook for how metal parts solidify.

The work, led by associate professor Tao Sun and published in Nature Communications, shows that the arrangement of atoms in liquid metal, not just cooling rate and temperature gradient, determines whether a printed part develops fine, uniform grains or coarse, weak ones. That sounds like a nuance, but it is the difference between a part that passes aerospace qualification and one that does not.

What the X-Rays Saw

Sun's team used operando synchrotron X-ray total scattering combined with rapid pair distribution function analysis at Argonne National Laboratory's Advanced Photon Source. Instead of printing a part and then slicing it open to guess what happened, they captured the melt pool behavior in Inconel 718 as it happened.

The scattering data revealed that molten metal is not the uniform, structureless liquid most models assume. It contains organized atomic clusters with short- and medium-range order, including icosahedral arrangements. As solidification proceeds, some clusters get consumed and reorganize into small, twinned atomic structures. That pathway produces unusually fine grains and distinctive internal boundaries, and the team linked it directly to abnormal columnar-to-equiaxed transition, a phenomenon long observed in printed metals but never traced to a specific atomic-scale mechanism before.

Why This Matters for Production

Today, engineers control metal AM microstructure with two knobs: cooling rate and temperature gradient. That is the classical G-R framework, and it has served the industry well enough for simple alloys. But as parts become more complex and safety requirements tighten, those two variables are not enough to guarantee consistent grain outcomes.

Sun's finding opens a third lever: liquid atomic structure. If you can measure that structure during printing, you can adjust laser parameters mid-process to steer grain formation. The immediate contribution is diagnostic, a clearer explanation of why unusual grain structures form. The longer-term goal is a feedback loop where the printer senses the melt pool state and changes power, scan speed, or hatch spacing on the fly.

The team is already working on two follow-on threads. The first is improving predictive models by building them around measured liquid atomic structure rather than assuming a featureless melt. The second is reconstructing the actual atomic structure directly from X-ray measurements instead of presuming which arrangements exist in the first place.

Where It Fits in the Broader Push

Northwestern is not the only group chasing better microstructure control. Fraunhofer and RMIT recently wrapped the UltraGRAIN project, which tailored grain structure during directed energy deposition builds. A separate University of Manchester study in June 2026 found that just a few degrees of temperature shift at the nozzle can reshape internal microstructure in aluminum parts. Taken together, these efforts point to the same conclusion: geometry in metal AM is largely solved, but microstructure control still depends on physics the field is only beginning to measure directly.

Sun's team has provided one of the clearest measurements yet. The next step is turning that visibility into a control system that works on the shop floor, not just at a synchrotron beamline.