JPH01293478A - Pattern checking method - Google Patents

Abstract

PURPOSE:To accurately measure a pattern to be checked by respectively measuring a position on a picture at the same position as the pattern to be checked for plural picture obtaining areas, obtaining the relation between the pattern position and the pattern position on the picture, and measuring the pattern whose picture distortion is corrected with the use of the relation. CONSTITUTION:Calibration to automatically presume the dynamic deflecting distortion of beam deflection scanning as the relation with the pattern position fluctuated by the real pattern position and the distortion is executed before the checking. As a method to obtain the relation, the reference positions of the pattern to be checked are detected at plural parts in the input picture, and the relation is calculated with the use of picture position data and pattern position data. In the checking, image pickup conditions at the time of the checking and the relation corresponding to the existing partial areas of the measured position are selected out of the distortion relation obtained from the calibration, with the use of it, the measured data are corrected, and the accurate measured result are obtained. Thus, the checking at high accuracy can be realized in the various image pickup conditions.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a method for inspecting the appearance of semiconductor patterns, and particularly to a pattern inspection method suitable for performing highly accurate pattern measurement.

[Conventional technology]

Conventionally, as a method of inspecting the dimensions of a pattern to be inspected using a scanning electron microscope (hereinafter referred to as SEM), an electron beam is scanned over the pattern to be inspected, and a reflected electron or secondary electron detection signal generated from the pattern is detected. Display it as an image on the display device, and use the cursor to specify the positions on the display screen that correspond to the two points of the pattern you want to measure.
By analyzing the signal level sequence of a line including two points and deriving the positions of the two points, the distance between the two points is determined based on the screen position or the signal sequence position. For example, the length of the display screen is equal to the number of pixels on two screens N.

If the magnification is M and the number of pixels between two measured points is n, then
The hole ML between two points is determined by L=nXL/(NXM).

[Problem to be solved by the invention]

However, there is distortion in the scanning of the electron beam, and the pattern to be inspected is deformed on the image obtained by scanning.

That is, the actual pixel size of a scanned image made up of pixels differs depending on the image position. Therefore, in order to avoid this problem and maintain high length measurement accuracy, measures are taken such as limiting the length measurement area within the scanning screen and not allowing length measurement in surrounding areas where distortion is large. There is. However, in actual pattern measurement, it is important to measure the pattern displayed on the screen at multiple positions, and there is also a strong demand for pattern length measurement at the periphery of the screen. Further, as with pattern length measurement, highly accurate measurement methods are required for various shape measurements such as pattern area measurement and size measurement.

Note that image distortion correction is performed in various fields such as remote sensing. However, especially when the theoretical characteristics of distortion are not clear, it is difficult to automatically perform correction calibration without using a specific pattern.

An object of the present invention is to provide a pattern inspection method that does not have the above-mentioned problems and can accurately measure a pattern to be inspected without being affected by distortion caused by imaging the pattern to be inspected.

[Means to solve the problem]

In order to achieve this objective, in the present invention, in pattern inspection using SUN, dynamic deflection distortion of beam deflection scanning is calculated prior to inspection to create a difference between the true pattern position and the pattern position changed due to the distortion. Calibration is performed to automatically estimate the relationship.

As a method for determining this relationship, reference positions of the pattern to be inspected are detected at a plurality of locations within the input image, and the above relationship is calculated using image position data and pattern position data. In particular, in order to improve distortion correction accuracy, the beam scanning area is divided and the above relationship is determined for each partial area. Furthermore, this relationship is determined in accordance with imaging conditions such as beam scanning speed, beam acceleration voltage, and pattern enlargement ratio.

Furthermore, in order to make this calibration possible for any pattern to be inspected, a simple pattern such as a chip corner pattern of a wafer pattern is employed as the calibration pattern. And, as a method to automatically perform calibration,
The corner position is automatically detected as a reference position by processing the corner pattern data stored in the image memory. Furthermore, the beam deflection offset amount is automatically set using a computer,
By evenly distributing the corner positions within the beam scanning area, pattern position data and corresponding image position data are collected.

During inspection, a relationship corresponding to the imaging conditions at the time of inspection and the partial area where the measurement position exists is selected from among the distortion relationships obtained during calibration, and this is used to correct the measurement data to obtain correct measurement results. .

[Embodiments of the invention]

Hereinafter, the present invention will be explained by examples. FIG. 1 is an overall configuration diagram of an appearance inspection apparatus for a pattern inspection method using the present invention. In FIG. 1, 1 is an electron source that generates an electron beam, 2 is a converging lens that converges the electron beam, 3 is a deflection coil for the electron beam, 4 is an objective lens, 5 is a sample on the surface of which a pattern to be inspected is formed, 6 1 is a detector that detects secondary electrons generated by electron lens irradiation, 7 is a deflection signal setting device that sets the deflection offset amount, 8 is a deflection generator that generates a deflection signal for the deflection coil, 9 is a secondary electron detection circuit , 10
1 is an image memory for storing a secondary electron signal train in accordance with the synchronization signal of the deflection coil; 11 is an electronic computer for setting the deflection reference amount and processing the image memory data; and 12 is a CRT display for displaying the image memory data. It is.

An electron beam generated from an electron source 1 is deflected by a deflection coil 3 and scans the surface of a sample 5 two-dimensionally. The secondary electrons generated from the sample surface at this time are converted into electrical signals by the detector 6 and stored in the image memory. By displaying the data in the image memory on a display, an enlarged image of the surface of the pattern to be inspected can be observed.

At this time, distortion associated with scanning occurs in the image obtained by deflection scanning of the beam. In other words, one image position (x+y
) and the position of the pattern to be inspected (U, V), there is a functional relationship of V=F'y (x', y').

Therefore, this function is clarified in advance, and in actual pattern inspection, correct inspection results are obtained by performing the above conversion on the measurement data.

First, a method for deriving the relationship between the image position and the position of the pattern to be inspected will be explained.

Measured image position (x' + y') r Display screen size (L
x p L y) + display screen prime number (nx, ny) as magnification M.

The measured image position is expressed in the dimension of the pattern position by y=Ly/(M-nx)·y'. At this time, (xt y) and the pattern position (U.

■) between Here, mk, nk≧O, mkr≠mkz.

nkx≠nkxp k, kl, 2E [1+...
+Nl holds, and the conversion coefficient s=[bi
Find j]. This can be obtained by measuring the screen position and pattern position of N points on one image.

However, for this purpose, it is necessary to use a pattern having a plurality of points with known dimensions, and the process of automatically detecting the position of each point becomes extremely complicated. Furthermore, since the measurement points must be appropriately distributed within the screen, the pattern to be inspected must be changed depending on the magnification of the microscope. Therefore, in the present invention, the amount of offset of the beam deflection signal is calculated using a computer. By setting from , the function to move the entire beam scanning area is used.

(Hereinafter, this function will be referred to as a field of view movement function.) For this beam scanning area, a value corresponding to the desired movement amount is automatically set in the deflection signal setter by a computer.

The processing procedure will be explained below with reference to FIG.

201... A reference point is set on the pattern to be inspected, and by changing the visual field movement amount, the self-position of the reference point on the screen is moved, and by detecting the reference point, a different screen position is detected. That is, N of (Rs, St) ~ (RN tSs)
The amount of visual field movement is set, and the screen position (xzt yt) to (XN+yll) of the reference point for each set value is detected.

202...Assuming equation (3) and a linear relationship between the visual field movement amount and the pattern reference position on the screen, between the visual field movement amount and the screen position, mk, nk≧Q, mk1≠mkz.

nk1≠nkz, k, kl, k26:
A relationship of [1, -, N holds true. In this formula, (Rt, St) ~ (R+ S)w <xxt yz obtained in 201
) ~ (xt y), the transformation coefficient (,
Calculate iii[cij].

203...Visual field movement: f! ! : The relationship between (R, S) and the inspected pattern position (U, V) is as follows: 'However, the pattern position when the visual field movement amount is (0, O) is (0, 0).

year.

Then, the conversion coefficient [bij] in equation (3) is calculated.

In particular, the vector related to the image position in equation (3) is expressed as t [xyt xt ys 11 (t
[:mon] is.

[*] transposed matrix @) Then, this transformation becomes a geometric transformation between general quadrilaterals.

FIG. 3 shows an example of a pattern used to determine the relationship between the position of the pattern to be inspected and the position of the screen and an example of setting the amount of visual field movement. As a pattern to be used, a pattern having one corner as shown in FIG. 3(a) may be used. At this time, corner positions are used as pattern reference points. Furthermore, the amount of visual field movement is (a-1) to (a
-9), set the corner positions so that they are evenly distributed within the screen. The screen positions of the corners detected for each visual field movement amount are shown in FIG. 3(b). The conversion coefficient is calculated from the nine screen positions obtained in this manner.

Note that the corner position is detected by processing the image data stored in the image memory with a computer to detect the edge positions of the horizontal edges and vertical edges that form the corner.
Find the corner position. To detect the edge position, for example, the image data is differentiated in the horizontal and vertical directions, and then
As shown in FIG. 3(c), the projection distribution of the differentially processed image is taken in the horizontal and vertical directions, and the peak position is taken as the edge position.

Here, an example has been described in which a corner pattern is used and the reference point is set as a corner position. As another example, if a rectangular pattern is used and four apex positions are used as reference points, the correspondence between four types of pattern positions and image positions can be measured with one image input, and the number of visual field movements can be can be reduced.

FIG. 4 shows a procedure for measuring a pattern with accurate dimensions by finding the true pattern position corresponding to the image position specified on the input image.

401...I want to measure the length in the pattern input image2
Image position of point 'ei (Xll yt) t (X21
yz) is detected.

402 Substitute the screen coordinates of the above two points into equation (3) to find the corresponding pattern positions (Ul, Vl) and (U2°V2).

403...The distance between two points of the pattern is v'(Ul-U
2) Find mel at "+v'(Vl-V2)"). Nao, horizontal and vertical distance is IUI, U21゜l
Vl, V21 Nite-seeking Mel.

Furthermore, when measuring the area of a region surrounded by a plurality of positions on the pattern to be inspected, highly accurate measurement is possible by determining the accurate pattern position using equation (3). The pattern position of N points obtained by measurement and correction is (Ut+
Vz) ~ (Un, Vn), the area S of the closed region surrounded by the line segment connecting N points is, however, (Uzt Vz) E (Un+ty Vn
+t).

On the other hand, the actual deflection distortion is between the center of the screen and the periphery of the screen.
The trends are different. Therefore, next, a more precise distortion correction method will be explained.

As a method, the input image is divided and a distortion correction coefficient is determined for each partial region. That is, the correction coefficient calculation process shown in FIG. 2 is performed for each partial area. To explain this in more detail using Figure 3, (a-1), (a-2),
A correction coefficient for the first partial area is determined using the four corner positions detected in (a-3) and (a-4).

Also, (a-2).

Using the four corner positions detected in (a-3), (a6), and (a-5), the correction coefficient of the second partial area is determined as (a-5), (a-6), (a-9
).

A correction coefficient for the third partial area is determined using the four corner positions detected in (a-8). Furthermore, (a-4)
, (a-5) , (a-8) , (a-9)
A correction coefficient for the third partial area is determined using the four corner positions detected in . In this way, the distortion transformation matrix [b
If i jl is determined, even stricter distortion correction can be achieved.

〔Effect of the invention〕

As explained above, according to the present invention, in pattern inspection using a scanning electron microscope (SEM), by using a simple and common pattern such as a chip corner pattern of a wafer pattern, beam deflection scanning movement can be improved. deflection distortion can be automatically estimated as the relationship between the true pattern position and the pattern position changed by the distortion.

Furthermore, highly accurate distortion correction can be realized by dividing the beam scanning region and finding the above relationship for each partial region. In addition, the above relationship can be expressed as beam scanning speed, beam acceleration voltage,
By determining the pattern magnification according to imaging conditions such as pattern magnification, highly accurate inspection can be achieved under various imaging conditions.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of a pattern inspection apparatus adopting the present invention, FIG. 2 is a diagram explaining the image distortion correction coefficient procedure, and FIG. 3 is a diagram explaining how to use the field of view movement function. FIG. 4 is a diagram illustrating a pattern measurement procedure for correcting image distortion. DESCRIPTION OF SYMBOLS 1... Electron source, 2... Converging lens, 3... Deflection coil, 4... Objective lens, 5... Sample, 6... 2
Secondary electron detector. 7... Deflection signal setter, 8... Deflection signal generator, 9
...Secondary electron detection circuit, 10...Image memory, 11
・・・Electric diagram

Claims (1)

[Claims] 1. A pattern that obtains an image of a pattern to be inspected using an imaging device, further measures and corrects image distortion in the obtained image, and obtains dimensions, areas, etc. of the pattern to be inspected without distortion errors. In the inspection method, an image storage unit that stores the video signal of the pattern to be inspected as digitized image data;
and an arithmetic processing unit that calculates a specific position of the pattern to be inspected by arithmetic processing of the image data, and by measuring the position on the image of the same position of the pattern to be inspected for a plurality of image acquisition areas, A pattern inspection method characterized by determining a relationship between a pattern position and a pattern position on an image, and using this relationship to perform pattern measurement with image distortion corrected. 2. Divide the image area into multiple partial areas, find the relationship between the pattern position and the pattern position on the image for each partial area, and perform pattern measurement using these multiple relationships to correct image distortion. The pattern inspection method according to claim 1, characterized in that: 3. The pattern inspection method according to claim 1, wherein a pattern having at least one corner is used as the pattern to be inspected for strain measurement. 4. Irradiate the pattern to be inspected with a charged particle beam according to the scanning signal, use the secondary electron detection signal generated from the pattern as a video signal, and change the amount of offset of the beam scanning area to change the image acquisition area. The pattern inspection method according to claim 1, characterized in that the pattern inspection method is varied. 5. Irradiate the pattern to be inspected with a charged particle beam according to the scanning signal, use the secondary electron detection signal generated from the pattern as a video signal, and determine the scanning speed, accelerating voltage, and size of the scanning area of the charged particle beam. The pattern inspection according to claim 1, characterized in that the relationship between the pattern position and the pattern position on the image is determined by the method according to claim 1, and image distortion is determined and corrected using this relationship. Method. 6. An imaging unit that irradiates the pattern to be inspected with a charged particle beam in accordance with a scanning signal and detects secondary electrons and reflected electrons generated from the pattern, a unit that sets the amount of offset of the beam scanning area, and an image of the pattern to be inspected. It consists of an image storage unit that stores signals as digitized surface image data, an image data arithmetic processing unit, and a computer that controls the device. A pattern inspection device comprising: means for determining the relationship between the position of the pattern and the position of the pattern; and means for correcting image distortion based on this relationship.