Precision Fabrication of Nanostructures by Optimal Mixed H2/H Control of Microcontact Next Generation Lithography Systems

目前，许多不同的所谓NGL（下一代光刻）技术正在开始探索，以延续迄今为止令人难以置信的成功的光学光刻技术，它可以将PI带到70纳米临界尺寸的设备上，其中一种更有趣的NGL技术，称为微接触或软光刻，即使在制造10纳米特征尺寸和工作集成设备时也表现出卓越的图案保真度。微接触光刻方法包括1:1的共形掩模或模板，通过几种不同的方法直接与基材表面接触，以完成接触后的印刷过程，包括物理，化学和光子方法。最近，PI的斯坦福半导体制造小组开发了一种利用透膜材料（PMM）的新型微接触NCL策略。在这种方法中，模式转移是通过反应物质的分子运输通过可渗透的多孔模板来实现的，从而在衬底表面形成空间选择性抗蚀刻掩膜。在微接触NGL技术设想的尺寸上，在整个印刷轨迹中，包括延伸、保持和收缩，都需要对掩膜和基板进行极其精确的定位和对齐。PI建议为此目的开发一个最优的多变量控制系统，以他们最近开发的PMM系统作为特定的测试工具。初步的探索使他们专注于双伺服六轴压电驱动的纳米定位挠性平台，以及实时掩模-衬底间隙检测和用于定位和对准的激光干涉测量。总体而言，精细级系统将采用6个压电致动器和19个高分辨率定位探测器。虽然控制规律将在他们的PMM技术和相关的近场直写模式系统上开发和演示，但该策略将普遍适用于采用柔性定位方法的其他微接触NCL技术。下面是他们提出的方法的一些细节。首先，将使用由斯坦福大学的Cho和Kailath开发的所谓子空间方法的最新进展来确定弯曲阶段的状态空间模型。在计划的方法中，将生成频域数据并计算线性时不变模型。子空间识别技术提供了一种非迭代的方法来生成多变量状态空间模型。对于在传感器集中具有相当冗余的应用，子空间方法用于模型识别是有用的，因为通过使用位移结构理论的结果，可以获得快速算法。他们还计划研究子空间识别输出的使用，以确定控制单元的最小检测器和执行器集。利用确定的模型，他们将设计一个最优的混合H2/H控制器。H2/H控制目标适用于本项目，因为需要优化定位速度以增加吞吐量，同时防止掩膜在最坏情况下撞击基板表面。由于振动引起的随机和带限干扰，以及诸如非线性梁弯矩、迟滞、可变初始条件和地形效应等内部效应，都会进入装置，并且必须沿着所需的定位轨迹进行补偿。简而言之，本提案设想扩展和应用最新的控制设计理论来设计一个非常高性能的纳米定位控制系统。一个特定的新的所谓的PMM技术将是开发的试验台。然而，这些技术与几种微接触技术有关；它们也应该是有用的特定应用，如制造曲面，制造MEMS（微机电系统），微合成和微流体系统

0000541KailathA number of different so-called NGL (next generation lithography) techniques are currently begin explored for continuation of the so-far incredibly successful optical lithography techniques, which can take the PI's to devices with 70 nm critical dimensions, one of the more interesting classes of NGL techniques, called microcontact or soft lithography, has demonstrated exceptional patterning fidelity in even the fabrication of 10 nm feature sizes and working integrated devices. Microcontact lithography methods involve a 1:1 conformal mask or template, brought into direct contact with the surface of the substrate with several different approaches for completing the printing process subsequent to contact, including physical, chemical, and photonic means. Recently, a novel microcontact NCL strategy utilizing permeable membrane materials (PMM) has been developed in the PI's Stanford semiconductor manufacturing group. In this approach, pattern transfer is achieved by molecular transport of reactive species through a permeable porous template to form a spatially selective etch-resistant mask on the substrate surface.At the dimensions envisaged for microcontact NGL techniques, extremely precise positioning and alignment of mask and substrate will be needed during the entire printing trajectory including extension, hold, and retraction. The PI's propose to develop an optimal multivariable control system for this purpose, with their recently developed PMM system as a specific test vehicle. Preliminary explorations have led them to focus on a dual servo 6-axis piezeo-driven nanopositioning flexure stage along with real-time mask-substrate gap detection and laser interferometry for positioning and alignment. Overall, the fine-stage system will employ six piezoactuators and nineteen high resolution positioning detectors. Although the control laws will be developed and demonstrated on their PMM technology and a related near-field direct write patterning system, the strategy will be generally applicable to other microcontact NCL techniques that employ flexure positioning methods.Some details on their proposed approach follow. First, a state-space model of the flexure stage will be identified using recent advances in the so-called subspace methods developed by Cho and Kailath at Stanford. In the planned approach, frequency domain data will be generated and a linear time invariant model will be computed. Subspace identification techniques offer a non-iterative method to generate multivariable state-space models. For their application which has considerable redundancy in the sensor set, the subspace approach to model identification is useful since, by employing results from displacement structure theory, fast algorithms can be obtained. They also plan to investigate the use of the subspace identification output to determine a minimal set of detectors and actuators to control the unit.Using the identified model, they will design an optimal mixed H2/H controller. The H2/H control objective is applicable to this project because of the need to optimize the positioning speed for an increase in throughput while guarding against worst-case crashes of the mask to the substrate surface. Both stochastic and bandlimited disturbances due to vibrations, as well as internal effects such as nonlinear beam bending moments, hysteresis, and variable initial conditions and topography effects, enter the plant and must be compensated along the desired positioning trajectory.In brief, this proposal envisages the extension and application of recent control design theories to design a very high performance nanopositioning control system. A specific new so-called PMM technology will be the testbed for the development. However the techniques are relevant to several microcontact technologies; they should also be useful for specific applications such as fabrication on curved surfaces, and for manufacturing MEMS (microelectro- mechanical systems), microsynthetic and microfluidic systems.***