RapidNano: Towards 20nm Particle Detection on EUV Mask Blanks
- PDF / 827,228 Bytes
- 12 Pages / 432 x 648 pts Page_size
- 94 Downloads / 168 Views
RapidNano: Towards 20nm Particle Detection on EUV Mask Blanks Jacques van der Donck, Peter Bussink, Erik Fritz, and Peter van der Walle TNO, P.O. Box 155, 2600 AD, Delft, The Netherlands. ABSTRACT Cleanliness is a prerequisite for obtaining economically feasible yield levels in the semiconductor industry. For the next generation of lithographic equipment, EUV lithography, the size of yield-loss inducing particles for the masks will be smaller than 20 nm. Consequently, equipment for handling EUV masks should not add particles larger than 20 nm. Detection methods for 20 nm particles on large area surfaces are needed to qualify the equipment for cleanliness. Detection of 20 nm particles is extremely challenging, not only because of the particle size, but also because of the large surface area and limited available time. In 2002 TNO developed the RapidNano, a platform that is capable of detecting nanoparticles on flat substrates. Over the last decade, the smallest detectable particle size was decreased while the inspection rate was increased. This effort has led to a stable and affordable detection platform that is capable of inspecting the full surface of a mask blank. The core of RapidNano is a dark-field imaging technique. Every substrate type has a typical background characteristic, which strongly affects the size of the smallest detectable particle. The noise level is induced by the speckle generated by the surface roughness of the mask. The signal-to-noise ratio can be improved by illuminating the inspection area from nine different angles. This improvement was first shown on test bench level and then applied in the RapidNano3. The RapidNano3 is capable of detecting 42nm latex sphere equivalents (and larger) on silicon surfaces. RapidNano4, the next generation, will use 193 nm light and the same nine angle illumination mode. Camera sensitivity and available laser power determine the achievable throughput. Therefore, special care was given to the optical design, particularly the optical path. With RapidNano4, TNO will push the detection limit of defects on EUV blanks to below 20nm. INTRODUCTION In the semiconductor industry devices with very small features are produced. In current devices, features of 20 nm and smaller are present. Over four decades device densities could be described by Moore’s law, predicting that the number of transistors doubles every two years [1, 2]. As a consequence, the feature size also diminished at a near-constant rate. Producing these devices in an economically feasible way is only possible at high throughput and high yield levels. Contamination by particles of products and equipment is a major contributor to yield loss. In lithographic processes, particle contamination can lead to yield loss in two ways: particle deposition on the wafer resulting in a single defective die and particle deposition on the mask resulting in a defective die in every printed field on the wafer. Because of the diminishing feature size, defect sizes allowed by particles will decrease accordingly. Particles as
Data Loading...
