An Open-Source Hardware and Software Toolkit to Enable Agentic RHEED-Guided Thin-Film Synthesis

Le résumé fourni par la source

Supporting Movies/data files Movie M1. STO_stepped_-100_300C_15.8kV_1.4A_5000us.mp4: Rocking curve of TiO2-terminated SrTiO₃ (001) at 300 °C, beam along [100], 15.8 kV, 5000 µs exposure, 1236 frames at 20 fps. Six sinusoidal rocking curve cycles; the dataset of Figure 5. Movie M2. STO N-2_0.mp4: Visualization of the RHEED image stack STO N-2_0.h5 used for the quantitative Auto-RHEED analysis shown in Figure 3. The movie contains 2,923-frame sequence used to extract time-dependent in-plane lattice spacing, coherence length, streak width, and intensity oscillations. Movie M3. STO N-2_1.mp4: Visualization of the RHEED image stack STO N-2_1.h5 used for the agent-driven Auto-RHEED analysis shown in Figure 4. The movie contains 3,827-frame sequence analyzed through the Auto-RHEED MCP workflow. Movie M4. LSMO5_frames0-1700.mp4: Visualization of the RHEED image stack LSMO5_frames0-1700.h5 used for the AI-inference analysis shown in Figure 6. The movie contains 1,700-frame sequence used for DINOv3 embedding, changepoint analysis, clustering, and YOLO-based diffraction-feature segmentation. Data D1. RHEED_alighnment_STO_110_5 deg_images.npy and RHEED_alighnment_STO_110_5 deg_angles.npy: RHEED frames and corresponding azimuthal angles used for automated SrTiO3 [110] substrate alignment analysis in Figure 2.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.