New technique makes smaller, faster chips possible for all manufacturers

Advancing semiconductor chip technology hinges on a crucial obstacle: crafting smaller, more efficient electronic components. Nowhere is this challenge more apparent than in lithography, the process central to creating intricate patterns on for chip production.

Lithography relies on photomasks, templates that imprint patterns onto semiconductor wafers. Industry constantly seeks methods to enhance resolution and manufacturability for both masks and wafers, aiming for faster chips and a higher yield of properly functioning ones.

Computational like optical proximity correction (OPC) have made significant headway in improving resolution and pattern fidelity. Inverse lithography technology (ILT) particularly stands out as a promising solution, determining mask shapes to achieve desired on- results. Curvilinear ILT mask shapes, in particular, have shown promise in delivering superior wafer results.

However, until recently, ILT's runtimes limited its application to critical chip “hotspots.” A new system proposed in 2019, featuring GPU-acceleration, revolutionized ILT, making full-chip ILT practical for production.

Yet, this breakthrough relied on multi-beam mask writing, which is pixel-based and shape-agnostic. The challenge was extending full-chip, curvilinear ILT benefits to variable shaped beam (VSB) mask writers, the predominant technology globally.

Addressing this, D2S, Inc. devised mask wafer co-optimization (MWCO). Leveraging insights that both the mask writer and wafer scanner act as low-pass filters, MWCO uses overlapping shots to create curvilinear shapes with fewer shots. By targeting the wafer pattern rather than the mask pattern, simpler shots are generated, ensuring compatibility with VSB writers.

D2S and Micron Technology demonstrated MWCO's prowess, reducing wafer variation by 3x and enhancing the wafer process window by 2x compared to Micron OPC. This marks a significant leap in lithography precision and reliability. Moreover, full curvilinear ILT masks can be written in less than 12 hours, meeting high-volume production demands.

This breakthrough democratizes chip manufacturing, enabling all semiconductor manufacturers to produce smaller, higher-performance chips with , irrespective of access to multi-beam mask writers.

Source: SPIE

Leave a Reply