Reticle Inspection Breaks the 100 nm Barrier
Alexander E. Braun, Senior Editor -- Semiconductor International, 11/1/2000
 |
KLA-Tencor (San Jose) has introduced a family of reticle inspection tools, the Tera family, designed to work with sub-wavelength lithography to the 0.13 and 0.10 µm nodes. A principal feature is a proprietary algorithm that provides the capability to inspect any OPC; there are no more OPC rules or limits to impact inspection with these tools (Figure). Inspection is easily done on 0.13 µm design rules with aggressive OPCs, including assist features and sub-resolution lines that parallel primary features, which in the past were difficult to handle for inspection tools.
Previous algorithms were feature-based: They took an exact look at the geometries and identified corners and lines, and performed a reverse alignment feature comparison of those. The new algorithm does not need to identify particular features. It inspects the image as it is, comparing it with an adjacent die or the design database. TeraScan is the first DUV-wavelength database inspection tool. TeraStar is a multibeam wavelength die-to-die and contamination inspection tool. Both platforms can inspect reticles with very small minimum primary feature sizes, combined with the capability of finding very small defects. These are the first systems to break through the 100 nm barrier.
|
|
| The TeraStar and TeraScan are the first systems to break through the 100 nm barrier. They are designed to work with sub-wavelength lithography to the 0.13 and 0.10 µm nodes. (Source: KLA-Tencor) |
These are the first tools with the capability to inspect a terapixel on a reticle - 1 million by 1 million pixels - hence the name Tera. If 0.13 µm design-rule reticles are to be inspected with 100 nm defect sensitivity, this terapixel capability is key.
The TeraScan uses new image-acquisition systems. Previously, a single scanning laser spot system was applied, reflected by acousto-optic elements. The TeraScan uses flood illumination and a parallel pixel imaging system to look at the reticle. This is an intrinsically more parallel system than scanning laser, and enables inspection to take place four times faster than previously, despite the fact that it scans with smaller pixels and higher resolution. TeraStar is an extension of a previous-generation platform, but uses holographic beam replication to generate three scanning beams rather than one, for a threefold increase in throughput.
Requirements for reticle sensitivity are accelerating at twice the rate than for wafers. This is the price for extending the life of 248 nm steppers and scanners. Today, lines that are smaller than the wavelength of light are used. It seems obvious that the industry will continue trying to extend 248 nm. When, eventually, the move is made to 193 nm, history will undoubtedly repeat itself. The Tera platforms are expected to be able to work through both of those extensions. They are already capable of working with attenuated phase shift, the next generation of reticle enhancement technique. .
