4D reconstruction of alumina laser melt pools at 25 kHz via operando X-ray multi-projection imaging

Abstract

Advancing additive manufacturing, e.g., laser powder-bed fusion (LPBF), requires resolving rapid processes such as melt-pool dynamics and keyhole evolution in 4D (3D + time). Operando X-ray tomography is a state-of-the-art approach for 4D characterization, but its temporal resolution is fundamentally constrained by the sample rotation speed, limiting achievable 4D imaging rates and preventing the resolution of these fast phenomena. Here we present rotation-enabled X-ray Multi-Projection Imaging (rotation-XMPI), which captures three angularly resolved projections per time step and thereby decouples temporal resolution from the sample rotation speed. Combined with a self-supervised deep-learning reconstruction framework for multi-angle inputs, rotation-XMPI enables high-fidelity 4D imaging at unprecedented speed. We demonstrate the approach in an operando alumina laser-remelting experiment at MAX IV using three beamlets combined with 25 Hz sample rotation. Rotation-XMPI resolves melt-pool morphology and keyhole evolution; in contrast, conventional and limited-angle tomography remain rotation-limited, and motion blur prevents resolving these dynamics. Overall, rotation-XMPI delivers a 250-fold increase relative to state-of-the-art melt-pool imaging, effectively achieving 25,000 reconstructed volumes per second. This method establishes a practical route to scalable ultrafast 4D imaging for additive manufacturing and other materials processes.

Figures (3)

• Figure 1: Acquisition setup and reconstruction results of laser remelting.a Schematic of the XMPI data-acquisition setup. Three X-ray beamlets are shown in blue and the laser in red. The sample is mounted on a rotating holder. b Photograph of the sample mounted on the holder (background grayed out) and a schematic of the sample during remelting. c Raw projection without flat-field correction, with the melt pool visible. d Simulated melt-pool behavior and morphology, adapted from Refs. luthi2025multiafrasiabi2022effect; unlike the powder-bed simulation, the experiment remelts solid material without a powder layer. e Horizontal orthoslices reconstructed from single-detector data for the static initial sample state (each detector reconstructed independently). f Two time steps from the rotation-XMPI reconstruction, separated by 40 µs. Scale bars: 6 mm in b; 200 µm in c; 200 µm in e; 100 µm in f.
• Figure 2: Comparison of imaging and reconstruction approaches.a,b Schematics of rotation-XMPI and limited-angle tomography in the sample's frame of reference. Circled numbers denote the detector positions, and colors encode acquisition times (red/blue/green); the center depicts schematic melt-pool positions at three times. R1–R3 indicate the corresponding reconstructions. c Comparison of reconstructions using conventional tomography (Gridrec), limited-angle (STRT), and rotation-XMPI algorithms. Reconstructions are shown as horizontal slices and are cropped to the sample for clarity. Scale bars: 100 µm. d Normalized cross-correlation (NCC) between dynamic reconstructions (rotation-XMPI and STRT) and static reference volumes acquired before and after laser remelting. NCC between two independent rotation-XMPI reconstructions (even/odd frame subsets) is also shown as a robustness evaluation.
• Figure 3: Melt pool segmentation.a Sequence of 14 cropped horizontal slices illustrating the evolution of the keyhole and melt pool; the laser scans from bottom-left to top-right in each panel. b Horizontal slices with the segmented melt pool overlaid; the four time points are separated by 160 µs. c, d Vertical slices with the segmented melt pool overlaid. e 3D rendering of the segmented melt pool together with the solid sample. f Side-view rendering of the segmented melt pool. Scale bars: 100 µm in a--d.