Repository hosted by TU Delft Library

Home · Contact · About · Disclaimer ·
 

Stealth fiducial markers: enhancing re-detection efficacy of defects on blank wafers

Author: Bouwens, M.A.J.
Type:Thesis
Date:2015
Publisher: TNO Delft University of Technology
Place: Delft
Identifier: 779158
Keywords: Nanotechnology · High Tech Systems & Materials · Industrial Innovation · Nano Technology · NI - Nano Instrumentation · TS - Technical Sciences

Abstract

In this thesis a stealth fiducial marker system for blank wafers is designed, fabricated and validated. The goal of this marker system it to map the coordinates of TNO’s Rapid Nano (a fast optical inspection tool) to the coordinates of scanning electron microscopes and atomic force microscopes. This way defects, that have been detected in the Rapid Nano, can be re-detected with a higher efficacy in the scanning electron microscope or atomic force microscope for further study. To develop the fiducial marker system, firstly a list of requirements for the markers and the marker system was drafted. To this end, experts on all three platforms were interviewed and previously tried fiducial markers for these platforms were studied. Also a Monte Carlo simulation was performed, to study the error propagation in the positioning accuracy of the markers and defects. The list of requirements led to a design for the fiducial marker system. Multiple fiducial markers are spread across the blank wafer. Each marker has a unique shape, so they can be distinguished from one another. The design of the markers provides information about the position and orientation of the marker at a glance. Once a single marker is found, it is easy to locate the other markers. The material and the height of the markers was optimised with finite element (COMSOL) simulations and experiments. The optimised fiducial markers are very visible in bright field, and almost invisible (“stealth”) in optical dark field, the microscopy mode that is used in TNO’s Rapid Nano. This results in markers that are are well-visible and accurately localisable in (dark-field) optical microscopy, scanning electron microscopy and atomic force microscopy. The fiducial marker system was fabricated with e-beam lithography, evaporation of gold/platinum and lift off. The functioning of the fiducial marker system was validated by re-detecting programmed defects that have been mapped by the Rapid Nano. The fiducial marker system greatly enhances the re-detection rate of defects. Without fiducial marker system a re-detection rate of 50% is considered challenging, but in the experiments with the fiducial marker system 100% of the programmed defects were re-detected on both a scanning electron microscope and the atomic force microscope. In fact, if the fiducial marker system is used, only one standard SEM (7x7 um2) image per particle is needed to comply with the ITRS requirement on re-detection probability for particles down to 70 nm. The experimental results on the scanning electron microscope are in agreement with the results from Monte Carlo simulations on the propagation of inaccuracy in the position of markers/defects. The atomic force microscope performs worse than expected, because the stage precision in the used AFM model is worse than specified by the manufacturer. The requirements on contamination, used materials and lifetime of the markers are not met. In a next design for the fiducial marker system, lift off should be avoided as a fabrication method. It proved very difficult to find a suitable material for the markers that is compliant with standards from the semiconductor industry. Therefore it should also be investigated if the markers are still visible enough in optical and scanning electron microscopes, if the contrast between the marker and the silicon substrate is only provided by topographical contrast. In this case, no “prohibited” material has to be used and the lifetime of the fiducial marker system would improve at the same time. During the finite element analysis in COMSOL, a peculiar property of the scattering in the Rapid Nano was discovered: Two opposing edges of a fiducial marker scatter differently. The asymmetry was confirmed with experimental data. This effect could potentially be used for advanced defect classification in the Rapid Nano.