·Home ·Table of Contents ·Methods and Instrumentation

The Develop of Microscope Image Processing System Fu Ping Wang Peihong (Daqing Manage Office of Oil China 163453) Chen Hexin Ma Yukuan (Jilin University of Technology Changchun China 130025) Contact

Abstract:

In this paper, we introduce a microscope processing system and its characteristic. The system includes PC computer acting as master controller, MCS-51 single-chip microcomputer which is used to collect and memorize data of image and acting as vice-controller as well, X-Y precise automatic (or hand control) worktable of microscope shoot acts as input device, CRT and printer act as output device. Some key problem met in research is discussed on detail on this paper. Several new measures had been taken in some segments, such as hand control for pre-location, automatic precise location and realization of hardware etc. At last we take application of automatic testing for CMOS mask defect in order to introduce the practical application of the system.
key words: edge detection; precise microscope worktable; NLLAP operator

1. Introduction

The microscope processing technique is so important in scientific research that many people have been involved in its application. It has been widely applied to lots of scientific fields, such as medicine, biology, metallic material science and other corresponding industrial fields. Since so many fields request microscope processing technique, the emergence of microscope processing system with high accuracy is imperative. The system proposed in the paper had been developed for all kinds of needs.
Especially during the VLSI producing, graph quality of mask must accord with the size of CAD designing graph. As we known, the lines of LSI are usually less than several fraction of one mm. Any defect on mask will influence the properties of circuit chip. Then it is urgent to develop an automatic testing system for mask defect. The main function of the system is designed for this purpose, i.e. auto-testing function for CMOS mask defect.

2. Components of System

As shown in Fig.1, the microscope processing system consists of following units: the left part with a camera in the picture is worktable of precision microscope; memory and controller of image collection are in the box with a black handle below the worktable; PC master computer and CRT are on the right; then a high resolution image monitor and a printer are on the right two.

Fig 1: the microscope processing system

The performance of system are as follows:


1. Total length of travel of worktable in one direction (X direction or Y direction):80mm;
2. Every step: 1mm
3. Action error: £1mm;
4. Flatness of track:
   X direction: left right 10²
   up down 1¢54²
   Y direction: left right 24²
   Up down 1¢28²
5. Can check out defect of 1/5 line width
6. Location accuracy in X,Y direction£ one pixel(1/10 line width)
7. Total accumulated error (80mm length of travel accumulated):£2mm

3. Mathematical method of CMOS mask defect testing

Suppose microscope image shot by worktable of precision microscope is f(x, y),0£x, y£511, 0£f(x, y)£255. Because optical field formed by transmitted-light source on CMOS mask is not very uniform, pretreatment is very important.
However if we consider edge detection and filter of noise (due to uneven optical field) at the same time, applying some quick algorithm, the processing time can be reduced. Thus we apply non-linear method for CMOS mask edge detection.
Define input digital image f (x, y) as an set {w_{x, y}} on a pane, here {x, y} is all over z² and its subset. Let working section window be nxn and expressed by A. For two-dimensional median filter, it can be defined on condition {w_{x, y}(x, y)Îz² } as followed :

(1)

Where med means median.
As we know, the median filter can filter impulse noise and exponential noise in signal. While signal often contains these noise in when CMOS mask is on transmitted-light and during photoelectric transfer process.
Beckers proposed a non-linear Laplace operator - NLLAP, i.e.

NLLAP(x, y)={max(A)-b(x, y)}-{-min(A)+b(x, y)}=max(A)-2b(x, y)+min(A)

(2)


Where max (A) is the maximum grey value of window A, min(A) is the minimal grey value of window A, b(x, y) is center position(x, y) grey value of window A.
Let O_{x, y} in formula (1) represent b(x, y) in formula(2),we can get a non-linear Laplace operator as follows:

NLLAP'(x, y)={max(A)-Ox, y}-{-min(A)+ Ox, y }= max(A)-2 Ox, y +min(A)

(3)


Concerning the cross zero testing we adopt threshold method. Then we can chose threshold value D by testing . If the formula below accord to:

÷ NLLAP'(x, y)÷ £D

(4)


Then pixel at (x, y) is edge pixel. Detailed discussing see reference [2].
For mask CAD graph is frame graph, edge detected by the method above will match original design frame. We can judge whether defect exist in CMOS mask or not.

4. Software of System

Software of the System consists of two parts: control software and operational testing software. The former includes PC computer control software and vice-controller control software.

4.1 Control software
PC computer and vice-controller (consists of MCS-51 single-chip microcomputer) share 62 bus. Communication between them adopts inquiry mode. Fig.2 is shown PC computer control software,,Fig.3 is control software of vice-controller. PC control software accomplish such functions as:

1. Order sub-controller to generate corresponding number of step and signal of direction to make worktable move.
2. Set up the status word of image collection and memory unit and control the image collecting time.
3. Control the display mode of image monitor.
4. Judge whether the mask defect are finished testing or not, then output a distribution table of defect.

Fig 2: PC computer control software

Fig 3: Sub-controller control software


As shown in fig.2, according to the request, PC computer output a series of orders to sub-controller, then let sub-controller's =0 and apply interruption. After sub-controller respond the interrupt requisition, it send "busy" signal immediately (as shown in fig.3) and take control command or data from data port of PC computer. After that it recover to =1 and send "free" signal. The keyboard function in this figure refer to the four control buttons on a small box in front of controller in fig.1. They can control the worktable moving in any direction independently. When the CMOS mask defect automatic testing system starts to work,"original point"(i.e. the first chip mask on left-up corner) is fixed position at first. Then trying several graphs (X-direction and Y direction), the sub-controller counts parameters. Then these parameters are divided by the number of graphs given by PC computer to work out the step number of every graph, then it is saved in sub-controller. When the sub-controller receive order of moving in X-direction or Y-direction, the worktable will move according this step number. That is calculation function in Fig.3.

4.2 Defect testing software
As shown in fig.4, the software mainly has functions as follows:

1. Using no-linear edge detection method mentioned above to detect edge.
2. Divide full image into blocks in proportion according to CAD design graph to determine which block exist defect.
3. Determine "true" or "false" of defect in block according to the given threshold. If it is "true", then record the position of the corresponding block.
4. After all finished, output the distribution table of mask defect.

Fig 4: Arithmetic testing software

5. Discussion

We introduce the hardware components of microscope processing image system and the mathematical method and software of CMOS mask defect automatic testing. It is obviously, the system can be applied to medicine, biology, metallic material science and so on, if work with other software units. Thus its application is so widely that it is an important device in scientific research and industry.

Reference

1. Beckers A L D. Ingenieuv;s' thesis, dept of applied physics, Delft University of Technology, 1986.
2. Fu Ping A new non-linear edge detection filter. Robots, 1991, 13(1):32-35.
3. Chen Hexin. Non-linear filter and digital image processing. Beijing: National Defense Industry Publishing House, 1997.

© AIPnD , created by NDT.net

|Home|    |Top|