The SEMulator3D Automation package enables automated parameter variation studies and related data analysis, and includes both SEMulator3D Metrology and SEMulator3D Expeditor. SEMulator3D Metrology mimics actual metrology operations, providing in-line measurements of device structure. SEMulator3D Expeditor performs parameter variation studies, collecting related data for analysis. Expeditor and Metrology work together to leverage the capabilities of the SEMulator3D modeling engine and predict complex, multivariate process phenomena and design/technology interactions.
SEMulator3D “Virtual” Metrology
Metrology operations perform automated measurements of critical structural attributes during a virtual fabrication run, similar to in-line metrology in the fab. Metrology measures things like film thickness, critical dimension (width), step height, sidewall angle, contact area and more. Metrology operations can be inserted after any step in a SEMulator3D process, and can be applied anywhere on the 3D model. Each operation outputs data in a compact, tabular format suitable for analysis in Excel and other data analysis tools.
In contrast to physical measurements, SEMulator3D Metrology is not limited by physics and can measure parameters that could only be determined destructively in the fab (see Figure 1). For instance: film thickness can be measured on individual features in product layout; sidewall angles and critical dimensions can be measured anywhere, even under overlying films; and contact area can be directly measured. Since Metrology operations are mask-driven, they can be performed on any device type and don’t require special metrology structures.
Metrology output is a useful way to calibrate a SEMulator3D process flow against existing fab data. It can also be used to validate process assumptions, and to understand the full-flow ramifications of process changes on design rules.
Most technology decisions require information about device structure and function under target (centered) process conditions, as well as actual conditions. Since the SEMulator3D modeling engine is physics-driven, it correctly predicts device structure over a range of process conditions. This makes SEMulator3D an excellent tool for investigating and understanding the structural ramifications of process variations.
Expeditor enables automatic, spreadsheet-driven process variation studies. Starting from a calibrated SEMulator3D process description, one or more process parameters of interest may be varied through a specified range in a sequence of model builds. SEMulator3D Metrology data are automatically collected and assembled in spreadsheet form, enabling detailed analysis of the results.
Figures 2 shows a fin spacer patterning flow for a hypothetical sub-22nm technology (click here for more information on this FEOL process flow). Figure 3 shows data from a multivariate process variation study, performed in Expeditor and covering the full factorial process space for five critical process parameters in the spacer patterning flow. The data show the effects of variation in spacer conformality and spacer thickness on fin critical dimension (CD) and height. While lower conformality appears to provide for less top CD variation, it also shows a severe instability in fin height. This instability is due to the thinner sidewall coverage being more prone to break-down during the deep silicon etch, thereby eroding the STI CMP-stop feature, resulting in deep STI recess. This subtle effect arises through a combination of process variations that would have been difficult to foresee without a predictive, 3D, integrated modeling platform.
SEMulator3D’s Structure Search feature automates the identification of possible failure locations, based on criteria such as minimum line width and dielectric thickness, minimum contact areas, or electrical shorts/opens. The entire 3D model can be inspected and checked against user-defined geometric criteria, and failure locations are marked with visual markers, and their location coordinates are recorded. The following checks are supported:
- Minimum Gap
- Minimum Layer Thickness
- Minimum Line Width
- Min/Max Contact Area
- Number of Components – Electrical or Structural
(4) Structure Search is used to search a BEOL Metal1/Via1/Metal2 braid for: (a) separate electrical nets, (b) insulation less than minimum thickness, and (c) metal lines less than minimum width.
Enabling Massively Parallel Variation Studies
The SEMulator3D Automation package adds value by enabling massively parallel variation studies. Multivariate process behavior studies that formerly required multiple wafers, dozens of FA tickets and weeks or months of waiting for fab cycles can now be performed on a workstation class machine in a few hours.