Diffractive optical elements (DOEs) offer significant advantages over conventional refractive optics, particularly in non-visible spectral regions such as ultraviolet, gamma rays, and X-rays, where material limitations restrict traditional optical components. Owing to their design flexibility, DOEs enable the generation of complex beam profiles—including circular, vortex, and Airy beams—across a wide range of wavelengths. Despite their structural simplicity and compatibility with micro- and nanoscale fabrication, conventional DOEs often suffer from limited focusing efficiency, frequently requiring additional refractive lenses that introduce optical aberrations, increased system complexity, and higher cost. In this work, we present an integrated design and fabrication approach for micro-scale diffractive optical elements capable of achieving high focusing performance without reliance on supplementary optical components. A machine learning-based decision tree method is employed to generate optimized writing paths, which are subsequently fabricated using direct laser lithography. The proposed integrated DOE structures enable efficient focusing of multiple customized beam profiles within a compact and standalone optical element. This approach improves optical efficiency while maintaining low fabrication cost and system simplicity. The demonstrated integrated micro-DOEs provide a scalable and versatile platform for advanced beam shaping and focusing applications in photonics, particularly where compactness and performance are critical.
Trung et al. (2026) studied this question.