导向自组装光刻仿真技术

Simulation Techniques for Directed Self-Assembly Lithography:An Overview

导向自组装(DSA)光刻能够以更低的成本制造集成电路纳米结构,有望将传统光刻拓展应用到更小的工艺节点,是国际器件与系统路线图中列出的主要候选光刻技术之一。嵌段共聚物的导向自组装受引导模板尺寸及形貌、模板对两嵌段亲疏性差异,以及嵌段共聚物材料化学性质、体积分数等多种参数的影响。通过实验遍历整个参数空间寻找最优工艺条件不仅所需周期长而且成本很高。DSA光刻仿真技术能够预测不同参数条件下的DSA结构,研究缺陷形成的机理,反向设计所需的引导模板和材料,已成为DSA光刻工艺研发与应用过程中不可或缺的关键技术。本文综述了自洽场模型、蒙特卡罗方法、动力学模型、简化模型等常见DSA光刻仿真技术的基本原理及其在DSA光刻中的应用。各仿真技术在计算精度、适用范围和计算速度等方面存在较大不同,需要根据不同的应用场景选择不同的仿真技术。

Significance Directed self-assembly(DSA)lithography is a prominent candidate for next-generation lithography techniques.This can extend conventional top-down lithography to advanced technology nodes and enhance the pattern quality of top-down lithography.DSA is driven by interactions between two chemically distinct blocks of block copolymers(BCPs)and between the BCP material and guiding template.This unique capability enables the DSA to form nearly all geometric patterns required in semiconductor manufacturing,offering a cost-effective and efficient patterning approach.The mechanism underlying DSA has gained significant attention in recent years,leading to extensive research and development efforts to incorporate them into semiconductor manufacturing processes.Several companies and institutions,including IMEC,IBM,and CEA-Leti,have established pilot lines for DSA,further driving the implementation of this technology in the semiconductor industry.Simulation techniques play a vital role in the research and applications of DSA lithography.The DSA process can be parameterized using a series of parameters,including the chemical properties of the block copolymer material,the geometry and wetting conditions of the guiding templates,and other relevant factors.DSA can generate a wide range of microphase structures by varying these parameters.Simulation techniques provide an efficient and effective method to explore complex high-dimensional parameter spaces by mapping these parameters to the resulting self-assembled structures.Second,it can be used to investigate the mechanisms underlying defect formation.Addressing the issue of defects in DSA has always been a challenge.DSA patterns may deviate from the target structure,resulting in defects affecting pattern quality.Some defects may even be buried beneath the surface,making their experimental characterization difficult.Simulation techniques play a crucial role in reducing DSA defects to acceptable levels.Furthermore,simulation techniques can be employed to address the inverse DSA problem,which involves deducting the parameters or geometry of the guiding pattern from a target structure.Compared to the role of physical models in computational lithography,simulation techniques have become indispensable in the study and application of DSA,providing valuable insights and fulfilling an irreplaceable role.Many simulation approaches have been developed.These approaches include field-theoretic methods,such as self-consistent field theory and complex Langevin simulation;dynamic models,such as coarse-grained molecular dynamics and dissipative dynamics simulation;and probabilistic simulation methods,such as the Monte Carlo approach and simplified models.Among these methods,the self-consistent field theory(SCFT)is one of the most successful models for studying the phase transitions of BCPs.It can accurately predict self-assembled structures,thereby aiding in understanding the DSA mechanism.The SCFT model has been widely applied in theoretical studies of DSA lithography.However,the SCFT has limitations in that it can only obtain equilibrium structures and lacks information on the system fluctuations and time-dependent behavior of BCPs.More detailed methods,such as Monte Carlo simulation,are required in certain specific applications.Both the SCFT and Monte Carlo approaches are based on the system Hamiltonian to obtain stable phase structures with lower energies,but they cannot provide information about the temporal evolution of the system.Conversly,dynamical models provide accurate predictions of equilibrium structures while also capturing the time evolution of the DSA system based on Newton’s laws of motion.Simplified models are used in applications requiring fast computation,such as the inverse DSA problem,full-chip mask synthesis,and verification.These models are constructed using a phenomenological model calibrated using experimental data or by simplifying intricate physical processes with uncomplicated physical models.Consequently,simplified models offer higher computational efficiency at the expense of accuracy and generalizability.It is important to summarize the existing representative simulation techniques to provide a more rational guide for the future development of this field.Progress Different applications require distinct models.Fredrickson’s research group from the University of California,Santa Barbara,in collaboration with the Intel Corporation,has conducted pioneering studies by applying SCFT to investigate the defect formation mechanism in chemoepitaxy DSA and self-assembled cylindrical morphologies in VIA lithography.Coarse-grained Monte Carlo(MC)simulation,initially developed by Detcheverry’s research group at the University of Wisconsin-Madison,is another powerful tool utilized in the study of DSA lithography.Unlike the SCFT,the MC approach is more accurate and can predict thermal fluctuations because it does not invoke a saddle approximation.Research groups from Tokyo Electron,Global Foundries,and other institutions have widely reported the implementation of the MC method.Dynamical models provide novel insights,particularly regarding their ability to simulate the dynamic pathways through which equilibrium structures are formed.Delony’s research group conducted studies on bridge defects and defect modes induced by underlayer errors in chemoepitaxy DSA,providing a time-evolution analysis of these defects.Additionally,dynamical models have been employed to investigate the shrinking and multiplication of contact holes.These studies contributed to a deeper understanding of the behavior and evolution of DSA systems.The aforementioned rigorous models are accurate to a certain extent.However,they are time-consuming and require several hours to produce reliable results.However,this computational demand may not be feasible for time-sensitive tasks.Many research groups,including IBM,IMEC,and Toshiba,have developed simplified models to accelerate computations significantly by several orders of magnitude to address this issue.These models are constructed using several simple equations or data-driven paradigms.Although these simplified models may have limited accuracy and poor generalizability,they are highly effective in computationally challenging situations where speed is essential.Conclusions and Prospects Simulation techniques are vital tools for developing DSA lithography.Similar to the physical models used in conventional computational lithography,simulation techniques can predict the final structure,interpret the underlying mechanisms,provide a rational basis for the design,and ultimately achieve higher pattern quality.Currently,the simulation of DSA lithography faces several challenges.These challenges include limited scalability for full-chip simulations,loss of accuracy in simplified models,and integration of DSA simulations into traditional electronic design automation(EDA)flows.Furthermore,with the combination of DSA and extreme ultraviolet lithography(EUV),DSA models must simulate the process of utilizing DSA to rectify EUV patterns.With the assistance of simulation techniques,DSA lithography can extend the application of conventional lithography to more advanced nodes.