JPH06104320A - Pattern matching method - Google Patents

Abstract

PURPOSE:To prevent occurrence of insufficient pattern matching by discarding any fraction less than the boundary value of edge data values by edge filtering. CONSTITUTION:The title method uses an electron beam device. At the time of performing alignment between LSI design data and an SEM image, the alignment between the design drawing of a mask obtained from an LSI design data storing means 3 and the SEM image obtained from an SEM image storing means 2 is performed by using a wiring edge. The filtering 5 of the wiring edge is performed by providing a boundary value for separating the edge data row of the SEM image prepared at the time of pattern matching into effective data and noneffective data and setting a prescribed value to the noneffective data. Therefore, the S/N of the edge data row can be improved and the occurrence of insufficient pattern matching can be prevented.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION The present invention utilizes LS using design data.
In the electron beam system that analyzes the inside of the I-chip,
The present invention relates to a pattern matching method for aligning a design mask diagram of I and an SEM (scanning electron microscope) image.

In recent years, in order to miniaturize and increase the density of LSI, SE
The time to search for the target wiring from the narrow field of view of M is getting longer, and the wiring search is a bottleneck in LSI analysis.
On the other hand, attempts have been made to reduce the wiring search time by using LSI design data and to improve the efficiency of LSI analysis.

[0003]

2. Description of the Related Art In the first stage of a wiring search attempt using LSI design data, the positions of the design mask diagram and the LSI mounting stage are registered so that the stage is automatically set at a point specified in the mask diagram. The navigation function of moving was achieved. This allows the CAD processing software that processes the design data on the computer to be used to arbitrarily determine the correspondence between the design data such as the netlist, logic circuit diagram, and mask diagram, and the position of the LSI mounted stage. The wiring search efficiency has improved.

Next, as the second step, the mask diagram and SE of the almost same region displayed by the above navigation function are shown.
Development of an automatic measurement function that corrects the positional deviation from the M image and irradiates the electron beam to the points specified in the mask diagram to perform the measurement is underway.

The pattern matching for correcting the positional deviation between the mask diagram and the SEM image is performed, for example, by the method disclosed in the specification of the present application, Japanese Patent Application No. 03-258140 (in FIG. (Omitted configuration)
Was done in.

[0006]

In the conventional example, when the LSI chip to be measured is covered with an insulating protective film, the quality of the SEM image of the chip is deteriorated and clear contrast is often not obtained. . In this case, there was a problem that the S / N ratio of the edge data string used for pattern matching was particularly insufficient and the pattern matching failed.

An object of the present invention is to improve the S / N ratio of an edge data string and prevent failure of pattern matching.

[0008]

[Means for Solving the Problems] To solve the above problems, 1) a design mask diagram obtained from the LSI design data storage means when the LSI design data and the SEM image are aligned using an electron beam apparatus. And SEM image storage means
A pattern matching method for performing alignment with a SEM image by using a wiring edge, wherein the wiring edge is filtered, and the filtering is performed on the SEM image edge data string created at the time of pattern matching. A pattern matching method in which a boundary value is provided to separate valid data and non-valid data and a predetermined value is set for the non-valid data, or 2) the boundary value is detected by detecting the maximum value in the SEM image edge data sequence, The pattern matching method according to 1) above, wherein the SEM image edge data below the boundary value is set to a predetermined value, and the boundary value is the SEM image edge data string. And make the minimum value in the upper data of the frequency distribution,
1) The pattern matching method described in 1) above, wherein the SEM image edge data below the boundary value is set to a predetermined value, or 4) The pattern described in 1) above, wherein the predetermined value set to the ineffective data is a negative value.・ Achieved by the matching method.

FIG. 1 is a block diagram of the present invention. The function of each unit will be described below. The electron beam lens barrel unit 1 mounts an LSI chip to be inspected on the stage inside the apparatus, irradiates the LSI chip to be inspected with an electron beam, and acquires an image of this LSI chip. Alternatively, measure the wiring voltage.

The SEM image storage means 2 stores the SEM image acquired by the electron beam lens barrel. The LSI design data storage means 3 stores the design data of the LSI to be inspected. The mask figure data reading means 4 reads the mask figure data from the LSI design data storage means.

The edge filtering means 5 uses the SEM image and the mask diagram to detect the edge filter used for pattern matching.
Data is extracted and filtering processing is performed. The pattern matching means 6 is
Pattern matching is performed using data.

The electron beam deflection control unit 7 determines the deflection amount of the electron beam with which the LSI chip to be inspected in the electron beam barrel is irradiated. The control computer 8 controls the electron beam device.

The first form of the edge filtering means, which is a feature of the present invention, has the following means. Maximum value detection method in SEM image edge data sequence: SEM
The maximum value is detected from the image edge data string.

Threshold value determining means: A predetermined ratio of the maximum value is determined as a threshold value. Edge data processing means: Sets edge data below a threshold value to a predetermined value. Next, in the second mode, the following means are provided.

Upper edge data number determining means: Determines the number of edges included in the edge data string. Edge data frequency distribution creating means: Creates a frequency distribution of edge data values from an SEM image edge data string.

Lower edge data processing means: Sets edge data other than the number of upper edge data in the frequency distribution to a predetermined value.

[0017]

In the present invention, an edge
By truncating below the boundary value (threshold value) of the edge data value by filtering, the edge data string
We are trying to improve the S / N ratio. The operation of each unit will be described below.

In the electron beam column, the electron beam is deflected and irradiated at a predetermined position of the LSI chip to be inspected based on a signal from the electron beam deflection control unit, and the LSI chip to be inspected is scanned and generated. Secondary electrons are detected and SEM images are constructed.

The SEM image is sent to the SEM image storage means and stored therein. In response to pattern matching execution request
The SEM image is read from the SEM image storage means, and the SEM image is sent to the edge filtering means.

Further, in response to the pattern matching execution request, the mask figure data reading means reads the mask figure data of the designated area from the LSI design data storage means and sends it to the edge filtering means.

In the edge filtering means,
For the pair of SEM image and mask diagram sent,
Data is extracted and edge data is filtered.

In the pattern matching means, pattern matching is performed using the edge data subjected to the filtering process, and the conversion formula between the mask figure coordinates and the electron beam deflection amount is calculated.

The mask diagram coordinate / electron beam deflection amount conversion formula is sent to the control computer, and the electron beam deflection amount to the designated measurement point is set in the electron beam deflection controller. An electron beam is irradiated to a designated measurement point based on the set electron beam deflection amount, and the voltage waveform is measured.

FIGS. 2A and 2B are diagrams showing an implementation method of the edge filtering process. FIG. 2A shows the processing in the first form of the edge filtering means. First, an edge data string is created from the SEM image. SE created by the maximum value detection means in the SEM image edge data sequence
Detect the maximum value in the M image edge data string. Next, in the threshold value determining means, α% of the detected maximum value is determined as the threshold value. In the edge data threshold processing means, a value equal to or less than the threshold in the edge data string is set to a predetermined value.

FIG. 2B shows the processing in the second form of the edge filtering means. First, an edge data string is created from the SEM image. The edge data frequency distribution creating means creates a frequency distribution of edge data values from the SEM image edge data string. Further, the upper edge data number determining means determines the number of edges included in the edge data string. Next, in the lower edge data processing means, counting is performed from the one having the largest edge data value in the frequency distribution, and the edge data other than those up to the number of upper edge data is set to a predetermined value.

[0026]

EXAMPLE FIGS. 3A and 3B are diagrams showing an SEM image and an edge data sequence created from the SEM image, respectively.

FIG. 3A is an SEM image in which a rectangular area is an edge.
This is a data creation area. FIG. 3B shows an edge data string projected in the horizontal axis direction. The edge data string is created as follows.

An area (edge data creation area) focused on the edge (target edge) of a certain wiring in the SEM image is set. In that area, the luminance is integrated in the vertical or horizontal axis direction to create a projected distribution of luminance. An edge data string is created by performing a difference operation on the created projection distribution. In the pattern matching process, the edge data string created for all edges is used.

Next, in the first form of the edge filtering means, the maximum value in the edge data string of this SEM image is detected. Next, α% of the detected maximum value is determined as a threshold value, and a value equal to or less than the threshold value in the edge data string is set to a predetermined value.

As an example of determining the α value, several sheets were sampled.
Using the SEM image, the target edge is in the edge data string,
There is a method of examining the statistical distribution of how large the value is with respect to the maximum value and using the result.

In the second form of the edge filtering means, the SEM image edge data string to the edge
Create a frequency distribution of data values. FIG. 4 shows a schematic diagram of the frequency distribution of edge data values. This is a frequency distribution showing how many certain edge data values are in the edge data string. The horizontal axis shows the edge data value, the vertical axis shows the frequency, and the total of the shaded areas is the number of upper edge data. Next, the number of edges included in the edge data string is determined. This number is the edge
It is determined from the number of edges existing in the data acquisition area and the edge width displayed in the SEM image. The flow of this processing is shown in FIG.

Next, the frequency distribution is counted from the one having the largest edge data value, the ones up to the upper edge data number are validated, and the other edge data are set to a predetermined value.

Although the edges of adjacent wirings other than the target edge are also included in the edge data string, the edge data value of the edge part including these target edges is smaller than that of the non-edge part. large. Therefore, if the number of edges is known, the second mode is that the ones with a large value in the frequency distribution are counted from those with a large number to reach the number of edges, and the others with smaller edge data values are discarded. Is. In reality, in an SEM image, edges are displayed with a width of several pixels, so the number of edges is calculated by multiplying the number of edges by the edge width according to the SEM image acquisition magnification.

When the edge data below the threshold value is set to a predetermined value by the above edge filtering processing, this predetermined value is set to a negative value. As a result, even if the pattern matching is highly likely to deviate from the original matching state, a large penalty is imposed on the evaluation amount, and there is an effect that failure of the pattern matching can be suppressed.

Next, a concrete example showing the effect of the embodiment will be described in comparison with a conventional example. Result of pattern matching between a SEM image of a certain pattern and a mask diagram by the conventional method (SEM image after being aligned by pattern matching and a mask diagram are overlaid), Y direction (vertical direction) However, when the processing of the first form of the edge filing means of the embodiment was performed, it no longer failed. In this example, 5% of the maximum detection value was used as a threshold value, and edge data below the threshold value was set to a large negative value of -100,000.

[0036]

According to the present invention, S / of the edge data string is
It is possible to improve the N ratio and prevent failure of pattern matching.

[Brief description of drawings]

FIG. 1 is a block diagram of the present invention

FIG. 2 is a diagram showing an implementation method of edge filtering processing.

[Figure 3] Diagram showing SEM images and edge data sequences created from SEM images

FIG. 4 is a diagram showing a frequency distribution of edge data values.

FIG. 5 is a flowchart showing a method of determining the number of upper edge data.

[Explanation of symbols]

 1 electron beam lens barrel 2 SEM image storage 3 LSI design data storage 4 mask diagram data readout 5 edge filtering 6 pattern matching 7 electron beam deflection controller 8 control computer

Claims (4)

[Claims] 1. When aligning the LSI design data with the SEM image using an electron beam apparatus, a design mask diagram obtained from the LSI design data storage means and an SEM image obtained from the SEM image storage means This is a pattern matching method for performing alignment between wiring edges by using the wiring edges, and the wiring edges are filtered, and the SEM image edge data string created at the time of pattern matching is not treated as valid data. A pattern matching method characterized by setting a boundary value for dividing into valid data and setting a predetermined value for non-valid data. 2. The boundary value is the maximum value in the SEM image edge data sequence detected as a predetermined ratio of the maximum value, and the SEM image edge data below the boundary value is set to a predetermined value. The pattern matching method according to claim 1, wherein 3. The boundary value creates a frequency distribution of the SEM image edge data sequence and sets it as the minimum value in the upper data of the frequency distribution, and sets the SEM image edge data below the boundary value to a predetermined value. The pattern matching method according to claim 1, wherein 4. The pattern according to claim 1, wherein the predetermined value set in the invalid data is a negative value.
Matching method.