Dismiss
InnovationQ will be updated on Sunday, Oct. 22, from 10am ET - noon. You may experience brief service interruptions during that time.
Browse Prior Art Database

Width and Overlay Measuring Algorithm

IP.com Disclosure Number: IPCOM000088055D
Original Publication Date: 1977-Apr-01
Included in the Prior Art Database: 2005-Mar-04
Document File: 4 page(s) / 124K

Publishing Venue

IBM

Related People

Barshay, M: AUTHOR [+4]

Abstract

Described is an algorithm for use with the system of U. S. Patent 3,957,376 (Fig. 1) to measure classical width targets (Fig. 2A) and overlay targets (Fig. 2B).

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 60% of the total text.

Page 1 of 4

Width and Overlay Measuring Algorithm

Described is an algorithm for use with the system of U. S.

Patent 3,957,376 (Fig. 1) to measure classical width targets (Fig. 2A) and overlay targets (Fig. 2B).

The algorithm is a Fourier Transform Algorithm (FFT) modified by a sliding slope method to provide the measurements.

The sliding slope method is a signal preconditioning process that smoothes and differentiates the data. This is especially useful for analysis of diffraction patterns (Fig. 3A) of width geometries. The method consists of fitting five consecutive points of the diffraction pattern to a straight line and calculating a slope. The five-point fit is then moved through the diffraction pattern shifting one point at a time, calculating another slope each time. Fourier transform analysis is then performed on the resultant vector of slopes rather than on the raw data. The raw data has been smoothed and differentiated in a piece-wise-linear fashion. A circuit analogy would be to transmit the signal through a combination low-pass then high-pass filter.

In diffraction analysis, the width of the geometry to be measured is proportional to the frequency of its diffraction pattern.

Fourier analysis gives the power spectrum of the used pattern to determine the frequency of the pattern. The true frequency component is normally the maximum component outside of the DC component (Fig. 3B).

A comparison with actual data of the new sliding slope method plus Fourier analysis ver...