From “complex optics” to “intelligent light sources”: A metasurface-based lensless monitoring and uniform illumination paradigm for optical lithography

Background

Optical lithography is a cornerstone of semiconductor manufacturing, and the illumination system acts as its “hidden central hub,” directly determining imaging quality and production efficiency. Within this system, illumination homogenization and monitoring constitute two essential functional modules: the former ensures highly uniform mask illumination, while the latter provides real-time power monitoring and deviation correction, both of which are critical for process stability and reliability. For decades, industrial solutions have largely relied on complex optical architectures based on the Köhler illumination principle. These systems employ integrator rods, microlens arrays, and cascaded lens groups to achieve light-field homogenization. Despite their maturity, a fundamental limitation persists; higher illumination performance typically comes at the cost of increased system complexity, size, and expense. In addition, traditional architectures face several practical challenges, including the stringent alignment requirements introduced by multiple optical components, structural redundancy due to separate monitoring optical paths, and high maintenance costs and downtime risks caused by partial light-source failure. Driven by the demand for large-field exposure, cost-effective manufacturing, and intelligent lithography systems, conventional illumination architectures are approaching their limits in both performance and engineering scalability. Therefore, developing a compact, low-cost, scalable, and intelligently controllable illumination strategy has become a critical challenge for next-generation lithography and advanced manufacturing systems, with significant scientific and industrial implications.

Research Progress and Significance

To address the long-standing trade-off between high uniformity and system complexity in conventional lithographic illumination, a team led by Academician Shaohua Yu at Peng Cheng Laboratory and Prof. Zizheng Cao at Zhejiang University proposes a fundamentally new solution: a lensless monitoring and uniform lithographic illumination system based on asymmetric diffusion metasurfaces. This work achieves several key breakthroughs at both the system architecture and physical mechanism levels.

(1) Dual-function integration within a single device.

By designing a metasurface structure, two essential functions are realized within a single planar device (Fig. 1). The reflection side enables highly uniform illumination, while the transmission side provides real-time light-source monitoring. For the first time in a lensless architecture, homogenization and monitoring are integrated into a single component with the aid of intelligent algorithms, significantly reducing system complexity.

(2) Asymmetric diffusion optical mechanism.

Through precise control of optical path differences introduced by nanoscale structures, the reflection and transmission channels are decoupled and independently engineered (Fig. 2). The reflected light exhibits strong diffusion, enabling uniform illumination, while the transmitted light maintains near-normal propagation for monitoring purposes. This allows two distinct optical responses to coexist within a single device, overcoming the functional limitations of conventional optical components.

(3) Algorithm-driven intelligent light source system.

The system incorporates a genetic algorithm and a gray-level-to-power-level mapping algorithm to enable intelligent control of the LED array. The genetic algorithm optimizes LED configurations, achieving illumination nonuniformity below 3% (Fig. 3). The gray-level-to-power-level mapping algorithm enables real-time identification of LED operating states and accurately detects faulty emitters, dynamically adjusting the illumination configuration (Fig. 4). This enables fault-tolerant, self-adaptive illumination, allowing continuous operation without interrupting the system or replacing defective light sources.

The system was validated on a self-developed large-field lithography platform (Fig. 5). Experimental results show that illumination nonuniformity is less than 3%, enabling clear exposure of multi-scale patterns. The system supports uniform illumination across a large exposure field of 10 cm × 10 cm and remains stable even when some LEDs fail, fully demonstrating the system’s engineering feasibility and application potential.

This work represents not merely a new optical device, but a new paradigm for lithographic illumination systems. From “optical stacking” to “functional integration”, conventional systems rely on cascaded optical modules, whereas this approach achieves multi-functionality within a single metasurface layer. From “passive optics” to “intelligent systems”, traditional designs are static, while the proposed system enables real-time monitoring and adaptive optimization. From “hardware complexity” to “algorithm-driven design”, system complexity is shifted from optical hardware to computational algorithms, enabling a new pathway toward both high performance and low cost.

This work also highlights the strength of industry–academia–research collaboration. The project is supported by Peng Cheng Laboratory and involves close cooperation with Sun Yat-Sen University, Nanjing University, and Zhejiang University. Through joint efforts in device design, algorithm development, and system integration, the collaborative team has established a complete innovation chain spanning “fundamental theory—device design—system integration—platform validation”. This cross-institutional collaboration model provides strong support for rapid engineering translation and industrial deployment of complex optoelectronic systems.

Outlook for Future Applications

This technology demonstrates clear potential for industrial translation across multiple domains. In advanced packaging lithography, it can be directly applied to mask aligners and large-field exposure systems, improving illumination performance while reducing system complexity. In emerging manufacturing fields, it is expected to enable applications in Mini/Micro LED, MEMS and power devices, flexible displays, wearable electronics, and photonic devices. In the context of intelligent manufacturing, the system can evolve into a self-monitoring, self-adaptive intelligent light source, further enhanced by artificial intelligence for dynamic lithography optimization. Moreover, it has the potential to serve as a core functional module in next-generation lithography equipment, supporting technological independence in critical subsystems.

More importantly, this approach inherently possesses the capability of a “programmable optical field”. By integrating with computational lithography techniques (e.g., source–mask co-optimization) and AI-driven design methodologies, it can enable a new class of reconfigurable optical-field control systems. This may ultimately lead to standardized and modular illumination solutions. Supported by sustained collaboration between universities and national research platforms, this work demonstrates strong potential for engineering scalability and industrial incubation, and is expected to contribute to breakthroughs in key components of next-generation lithography systems.

Sources: https://spj.science.org/doi/10.34133/research.1197