Recent advances in machine learning and deep learning have provided an opportunity for improvement in the field of lithography. Compared with the numerical simulation, machine learning/deep learning may provide much faster and more efficient performance, but they provide less accurate solution due to the limitation of the statistical approach. Typical machine learning models cannot take into account complex multiple processes such as UV exposure, photoreaction and photo-resist development in lithography. In this work, we developed a newly designed deep learning algorithm not only to improve the model accuracy but also to overcome data limitation. By combining a physics-driven machine learning model and a complex-valued neural network (CVNN), we designed a novel machine learning model structure. Applying CVNN to phase shift mask analysis could improve the model performance dramatically. As such, this work opens up a new class of photo-lithography analysis by using a novel neural network model.
It is well known in the industry that the technology nodes from 30nm and below will require model based SRAF / OPC
for critical layers to meet production required process windows. Since the seminal paper by Saleh and Sayegh
thirty years ago, the idea of using inverse methods to solve mask layout problems has been receiving increasing attention
as design sizes have been steadily shrinking. ILT in its present form represents an attempt to construct the inverse
solution to a constrained problem where the constraints are all possible phenomena which can be simulated, including:
DOF, sidelobes, MRC, MEEF, EL, shot-count, and other effects. Given current manufacturing constraints and process
window requirements, inverse solutions must use all possible degrees of freedom to synthesize a mask.
Various forms of inverse solutions differ greatly with respect to lithographic performance and mask complexity. Factors
responsible for their differences include composition of the cost function that is minimized, constraints applied during
optimization to ensure MRC compliance and limit complexity, and the data structure used to represent mask patterns. In
this paper we describe the level set method to represent mask patterns, which allows the necessary degrees of freedom
for required lithographic performance, and show how to derive Manhattan mask patterns from it, which can be
manufactured with controllable complexity and limited shot-counts. We will demonstrate how full chip ILT masks can
control e-beam write-time to the level comparable to traditional OPC masks, providing a solution with maximized
lithographic performance and manageable cost of ownership that is vital to sub-30nm node IC manufacturing.
Although the mask pattern created by fine ebeam writing is four times larger than the wafer pattern, the mask
proximity effect from ebeam scattering and etch is not negligible. This mask proximity effect causes mask-CD errors and consequently wafer-CD errors after the lithographic process. It is therefore necessary to include
the mask proximity effect in optical proximity correction (OPC). Without this, an OPC model can not predict
the entire lithography process correctly even using advanced optical and resist models. In order to compensate
for the mask proximity effect within OPC a special model is required along with changes to the OPC flow.
This article presents a method for producing such a model and OPC flow and shows the difference in results
when they are used.
As the design rule of device shrinks below 0.14 micrometer, the higher resolution is required for real device application. With smaller feature size below 0.14 micrometer, the lower coating thickness of resist is essential because of the pattern collapse issue at the high aspect ratio. However, the lower resist thickness induces the problem of etch selectivity due to the limited etch resistance of resist. In this study, the method of electron beam stabilization has been applied for improving the etch selectivity of resist patterns having an aspect ratio less than 3:1. With applying the electron beam stabilization, the Deep-UV photoresists based on the chemical structures of Acetal (AS106) and Escap (UV82) types have been evaluated in the respect of etch selectivity as the functions of an electron beam dose and etch condition. The metal etch rate reductions of 20 percent and 26 percent have been occurred for the resists of Acetal and Escap type, respectively, at 2000 (mu) C/cm2. And the thermal and chemical properties were characterized before and after electron beam stabilization using DSC, TGA, and FT-IR. The cross-sectional views of resist pattern after electron beam processing were also investigated to know the chemical stability of resist during the electron beam process. Based on the experimental results, the application possibility of electron beam stabilization for real device fabrication below 0.14 micrometer has been presented in this paper.