Browse Prior Art Database

Registration by Digital Cross Correlation

IP.com Disclosure Number: IPCOM000088643D
Original Publication Date: 1977-Jul-01
Included in the Prior Art Database: 2005-Mar-04
Document File: 3 page(s) / 57K

Publishing Venue

IBM

Related People

Cardenta, PA: AUTHOR [+2]

Abstract

Accurate overlay of patterns written by electron(E-) beam systems is obtained by accurately locating registration marks in the written field. This article describes a method of determining registration mark location by utilizing digital cross correlation without the use of digital (or analog) multipliers, but with the use of the speed and resolution advantages of digital adders.

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 55% of the total text.

Page 1 of 3

Registration by Digital Cross Correlation

Accurate overlay of patterns written by electron(E-) beam systems is obtained by accurately locating registration marks in the written field. This article describes a method of determining registration mark location by utilizing digital cross correlation without the use of digital (or analog) multipliers, but with the use of the speed and resolution advantages of digital adders.

Assume F(t) to be the detected registration signal and C(t) to be a fixed function similar to the registration signal. Let R(tau) be their cross correlation function defined as:

(Image Omitted)

choose C(t+tau) to be as shown in Fig. 1, that is, equal to +/- K(j) at times regions, and zero elsewhere. In this case the correlation equation may be defined as:

(Image Omitted)

where f(i) is the i/th/ quantized digital value of the registration signal at time t(i) and K(j) is defined as follows: K(j) = 1/16 for a </- i </- b and g </- i </- h

K(j) = 1/4 for b </- i </- c and f </- i </- g
(3) K(j) = 1/2 for c </- i </- d and e </- i </- f K(j) = 1 for d </- i </- e. Note that K(j) is a power 2 function which makes it simple to generate the product K(j)f(i) by "shifting to right" of f(i) digital words.

A simple way of generating equation (2) at several times tau(1), Tau(2), .... is by sampling the registration signal by an analog-to-digital (A/D) converter at a high rate and storing the sampled data in a memory. By hardwiring adders to match equation (2), the output of each adder represents the correlation function at a given time tau(1), tau(2) ... etc. The adder whose output is greatest represents the location of the peak of the correlation function. For greater resolution as many adders as the number of sample points could be used. However, this may have a practical limit.

Another method of generating equation (2) is by utilizing several scan lines of information (e.g., 20 scans per mark) to determine the correlation. The first scan data is stored in the memory, then it is shifted by one word and added to the second scan line data. This operation is repeated for consecutive scans. At the end of the final scan, the memory data represents the correlation function R(tau). The peak location (memory address) determines the registration mark location in time.

Fig. 2 shows the hardware implementation of such method. The registration signal from the...