A new galvanometer-based laser processing system reported in Engineering integrates full-in-situ imaging and machining without requiring coordinate conversion or calibration, offering a streamlined approach for precision tasks in pan-semiconductor manufacturing.

Traditional galvanometer laser systems rely on multiple-step coordinate transformations—from pixel to planar coordinates and then to angular coordinates—to align the laser beam with the target. These steps introduce alignment errors due to mechanical assembly imperfections, field distortion of the f-theta lens, and positional uncertainties between imaging and scanning units. Calibration procedures are typically used to reduce such errors, but they add complexity and still leave residual inaccuracies.

The newly proposed system avoids these issues by capturing images directly in angular coordinates, using the same galvanometer scanners for both imaging and laser processing. This configuration eliminates the need for coordinate conversion and, by extension, the errors associated with it. The imaging subsystem uses a double-telecentric setup, projecting the scanning plane onto a luminance sensor via a measurement focusing lens and an f-theta lens. By scanning the plane point-by-point and recording light intensities in angular coordinates, the system generates grayscale images that directly correspond to the galvanometer’s deflection angles. These same angles are later used to guide the laser beam during processing.

The system was prototyped with a 355 nm Nd:YAG laser and a galvanometer scanner offering a maximum optical angle of 0.26 rad and a minimum resolution of 0.5 μrad. The f-theta lens has a 100 mm focal length, and the luminance sensor module (TOSHIBA TCD1304) provides a 200 μm × 200 μm measurement region. The prototype achieves a scanning range of 27 mm × 27 mm and a minimum scanning step of 0.412 μm, with a spatial resolution (SR) of 93 μm determined via edge spread function analysis.

In one experimental case, the system was used to cut special-shaped holes in flexible printed circuits (FPCs). A pentagram-shaped hole was processed with a laser line width of 30 μm, resulting in a machining precision below 15 μm. In contrast, a conventional vision-based coaxial camera approach required image stitching and calibration, leading to visible distortions and positioning errors of approximately 200 μm at the vertices.

In a second case, the system identified and removed misaligned Micro-LED dies prior to laser-induced forward transfer (LIFT). A coarse scan located the chip array, followed by a fine scan to confirm misaligned dies. Laser scanning was then used to selectively ablate the polymer sacrificial layer beneath the misaligned chips. All three misaligned dies were successfully removed from a total of 115 chips, with no damage to the remaining array.

The authors note that while the system eliminates alignment errors due to coordinate conversion, other error sources—such as galvanometer repeatability and laser spot size—still contribute to the overall processing error, currently under 15 μm. Future improvements in sensor resolution and focal length could enhance image clarity, while higher-bandwidth communication protocols may reduce imaging time.

This approach is particularly suited for small-batch, highly customized and complex processing tasks where flexibility and precision are critical. The system’s ability to perform imaging and laser processing in a single coordinate framework without calibration offers a practical alternative to existing methods in semiconductor microfabrication.

The paper “Galvanometer-Based Alignment-Error-Free Full-in-Situ Imaging and Laser Processing System with Applications to Pan-Semiconductor Manufacturing,” is authored by Yuxuan Cao, Kuai Yang, Yingchun Guan, Zhen Zhang. Full text of the open access paper: https://doi.org/10.1016/j.eng.2025.07.041. For more information about Engineering, visit the website at https://www.sciencedirect.com/journal/engineering.