JP2001033226A - Pattern inspecting device and pattern inspecting method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a pattern inspecting device capable of preventing the occurrence of a pseudo defect by complementing the difference between fine design data and sensor data. SOLUTION: This pattern inspecting device is provided with a reference data calculating circuit 12 filtering design data D1 to calculate reference data D2, a feature portion extracting circuit 13 extracting the feature portion of a pattern based on the reference data D2, an edge direction calculating circuit 14 calculating the boundary line direction of the boundary section between different kinds of the pattern based on the feature portion, a corrected reference data calculating circuit 16 filtering the reference data D2 to calculate corrected reference data D3, a sensor 20 detecting a pattern on an inspection object W, and a defect detecting circuit 17 comparing the corrected reference data D3 with the sensor data D4 obtained from the sensor 20 and detecting a defect based on the difference.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a pattern inspection apparatus and a pattern inspection method used for an inspection apparatus for inspecting the shape of a fine pattern formed on a mask or a wafer.

[0002]

2. Description of the Related Art When inspecting a defect of a fine pattern formed on a mask for exposure, a design data of a pattern as shown in FIG. 7A and sensor data actually detected are used. By detecting the difference between In the drawing, P1 and P2 indicate portions having different values constituting the pattern. However, in an actual semiconductor process, a portion where the width of a pattern changes (hereinafter referred to as an “edge portion”).
, Etc.) are not formed according to the design data, and the shape of the corner becomes gentle as shown in FIG. 7C. If there is such a difference, the defect may be recognized as a defect even though there is no defect, that is, a pseudo defect may occur.

For this reason, a filter (hereinafter, referred to as “filtering”) using a predetermined function such as a blur function on the design data is collectively performed on the entire pattern, so that the boundary of the pattern is shown in FIG. As shown in the figure, smooth reference data is created, and the reference data and the sensor data are compared. In a corner where a pattern is bent, another filtering is performed to further adapt the curvature to the process characteristics.

[0004]

The above-described conventional method for inspecting a fine pattern formed on a mask has the following problems. That is, in recent years, the pattern has become finer, and it has been necessary to increase the defect detection sensitivity.
For this reason, there has been a problem that a difference between the reference data and the sensor data, which has not been a problem, is recognized as a defect. In particular, the smaller the edge portion, the larger the difference between the reference data and the sensor data tends to be.

In view of the above, the present invention provides a pattern inspection apparatus and a pattern inspection method capable of preventing generation of a pseudo defect by further correcting reference data which complements a difference between design data and sensor data in a fine pattern. It is intended to provide.

[0006]

In order to solve the above-mentioned problems and to achieve the object, the invention according to the first aspect is based on the presence or absence of a defect in a pattern formed by a plurality of types of portions on an object to be inspected. In a pattern inspection device for detecting the reference data, reference data calculation means for calculating reference data by applying a filter of a predetermined function to the design data of the pattern,
A characteristic portion extraction unit that extracts a characteristic portion of a pattern based on the reference data; a boundary direction calculation unit that calculates a boundary direction of a boundary between different types in the pattern based on the characteristic portion; A function having a predetermined number of pixels arranged along a direction orthogonal to the boundary direction, and changing a value corresponding to a pixel located at the center based on a value corresponding to a pixel in the vicinity thereof. Correction reference data calculation means for calculating correction reference data by applying a filter, a sensor for detecting a pattern on the inspection object, and comparing the correction reference data with sensor data obtained from the sensor; And a defect detecting means for detecting a defect based on the difference.

According to a second aspect of the present invention, in the first aspect, a weighting factor is set for each of the pixels.

According to a third aspect of the present invention, there is provided a pattern inspection method for detecting the presence / absence of a defect in a pattern formed by a plurality of types of portions on an object to be inspected. A reference data calculating step of calculating reference data by applying a function filter, a characteristic part extracting step of extracting a characteristic part of a pattern based on the reference data, and a boundary between different types in the pattern based on the characteristic part A boundary direction calculating step of calculating a boundary line direction of the reference data, the reference data has a predetermined number of pixels arranged along a direction orthogonal to the boundary line direction, and corresponds to a pixel located at the center. A correction reference data calculation step of calculating correction reference data by applying a filter of a function that changes a value based on a value corresponding to a pixel in the vicinity thereof; A detection step of detecting a pattern on the object, and so and a defect detection step of comparing the sensor data obtained in the correction reference data and the detection step, detecting a defect based on the difference.

According to a fourth aspect of the present invention, in the third aspect, a weighting factor is set for each of the pixels.

[0010]

FIG. 1 shows a pattern inspection apparatus 10 for inspecting a pattern formed on a mask M according to an embodiment of the present invention. FIG. 2A shows design data D1. , (B) shows an example of a pattern based on reference data D2, (c) shows an example of a pattern based on corrected reference data D3, and (d) shows an example of a pattern based on sensor data D4. Further, the pattern is formed of a portion P1 and a portion P2 having different values from each other.

In the present embodiment, the inspection of a semiconductor circuit pattern formed on a mask will be described. However, the present invention can be applied to the inspection of a chip pattern printed on a semiconductor wafer.

The pattern inspection apparatus 10 includes a storage unit 11 for storing pattern design data (CAD data) D1,
A reference data generation circuit 12 for generating reference data D2 for inspecting a pattern;
A feature extraction circuit 13 for extracting the second feature portion, an edge direction calculation circuit 14 for calculating an edge direction from the extracted feature portions, a template T storage section 15 for storing a plurality of templates T described later, Direction calculation circuit 1
And a defect for detecting a defect by comparing sensor data D4 from the sensor 20 and the corrected reference data D3, which will be described later, with a corrected reference data calculation circuit 16 for calculating the corrected reference data D3 based on the edge direction obtained by the step 4. And a detection circuit 17.

An output from a sensor 20 for detecting a pattern formed on the mask M is connected to the defect detection circuit 17.

The pattern inspection apparatus 1 configured as described above
At 0, the pattern is inspected as follows. That is, the design data D1 is read from the storage unit 11, filtered by a predetermined blur function, and the reference data D2 is created.

Next, a characteristic portion of the pattern is extracted based on the reference data D2. Here, the characteristic part is
The edge portion Q between the portions P1 and P2, the corner portion where the portion P1 is bent, and the flat portion where the portion P1 is flat are characteristic portions of the pattern. Then, the direction of each characteristic portion is extracted. That is, as shown by an arrow X in FIG.
If so, the direction in which the boundary line K between the portion P1 and the portion P2 in the design data D1 is formed is extracted as the edge direction.

Next, a template T having a predetermined number of pixels along the edge direction is selected. As shown in FIGS. 4A to 4D, several types of templates T are prepared in advance. The template T has an odd number of pixels. The reason why the odd number is set is that an operation is performed on pixels located at the same distance from the center window. Also,
Since pixels are provided in the X direction, the Y direction, and the oblique direction, a template T having three pixels shown in FIG. 4A in which pixels are set in the X direction is used.

Next, as shown in FIG. 5A, a load coefficient (range of 0 to 1) is set for each of the pixels t1 to t3, and a template T is determined.

The reference data D2 is filtered based on the template T. It is to be noted that filtering of the reference data D2 is usually performed on a portion that becomes a problem when the defect detection sensitivity is increased. The filtering is performed as shown in FIGS. That is, as shown in FIG. 6A, the template T is applied near the edge Q, and the average of the results obtained for each pixel is set as the value of the central pixel.

For example, as shown in FIG.
The value of the central pixel α2 is replaced with α2 ′ by filtering in the region of α3. That is, α2 ′ = (t1α1 + t2α2 + t3α3) / 3.

Next, as shown in FIGS. 6B and 6C, the same calculation is performed by moving the template T by one pixel in the X direction. The correction ends when the template T passes through the edge portion Q.

Such correction has the effect of taking the moving average of the value of the designated number of pixels for the pixel of interest.
Even if there is a small step in the edge direction, the values are averaged using the values of the preceding and following pixels, so that smoothing (smoothing) can be applied, and smoothness suitable for process characteristics can be obtained. be able to.

In this manner, the correction reference data D3 is calculated.

Next, the sensor data D4 is detected by the sensor 20 based on the actual pattern. Then, the correction reference data D3 and the sensor data D4 are compared, and a data in which a difference exceeding a predetermined error range is found is detected as a defect.

Here, based on actual data, the correction results when the number of pixels N of the template T is 3 and the weighting factors are all 1 will be described below. The difference means a difference between values of adjacent pixels.

Reference data D2: 143 139 122 89 69 61 Difference: 4 17 33 20 8 Correction reference data D3: 142 134 116 93 93 73 62 Difference: 8 18 23 20 11 Sensor data D4: 145 135 118 95 76 66 • Difference: 10 17 23 19 10 When comparing the difference, that is, the gradient, it can be seen that the corrected reference data D3 is closer to the sensor data D4 than the reference data D2. That is, by correcting the reference data D2 and performing the correction on the corrected reference data D3, it is possible to prevent the occurrence of a pseudo defect based on the difference between the reference data D2 and the sensor data D4.

By setting the weighting factor and the number of pixels, it is possible to finely adjust the correction of the reference data according to the process characteristics. For example, a moving average calculation filter having a simple width N is obtained by setting all the load coefficients to 1. Further, by changing the weight coefficient, it is possible to adjust the influence of a pixel distant from the center pixel.

By changing the number of pixels used for the template T, the amount of the moving average can be adjusted.

In the setting, an actual inspection is performed, the sensor data D4 is compared with the correction reference data D3, and adjustment is performed so as not to be recognized as a pseudo defect.

As described above, according to the pattern inspection apparatus 10 of the present embodiment, the reference data D2 calculated based on the design data D1 is filtered on the basis of the characteristic portion, thereby smoothing (smoothing). Is applied, and smoothness suitable for the process characteristics can be obtained.

Further, by setting the weighting factor and the number of pixels, it is possible to finely adjust the correction of the reference data according to the process characteristics.

Therefore, it is possible to prevent the occurrence of a pseudo defect which becomes a problem when the sensitivity is improved, and to eliminate erroneous defect detection and erroneous determination in inspection.

It should be noted that the present invention is not limited to the above-described embodiment, and it is needless to say that various modifications can be made without departing from the spirit of the present invention.

[0033]

According to the present invention, it is possible to prevent the occurrence of a pseudo defect by further correcting reference data that complements a difference between design data and sensor data in a fine pattern.

[Brief description of the drawings]

FIG. 1 is a view showing a pattern inspection apparatus to which an inspection method according to an embodiment of the present invention is applied.

FIG. 2 is a view showing a specific example of a pattern in the pattern inspection apparatus.

FIG. 3 is a view showing an edge direction in the pattern inspection apparatus.

FIG. 4 is a view showing an example of a template stored in the pattern inspection apparatus.

FIG. 5 is an explanatory diagram showing the principle of reference data correction in the pattern inspection apparatus.

FIG. 6 is an explanatory diagram showing a specific example of reference data correction in the pattern inspection apparatus.

FIG. 7 is an explanatory view showing a conventional pattern inspection method.

[Explanation of symbols]

 REFERENCE SIGNS LIST 10 pattern inspection apparatus 11 storage unit 12 reference data generation circuit 13 feature extraction circuit 14 edge direction calculation circuit 15 template storage unit 16 correction reference data calculation circuit 17 defect detection circuit 20 sensor

 ──────────────────────────────────────────────────続 き Continuing on the front page F term (reference) 2F065 AA31 AA54 AA61 BB13 CC18 DD00 EE08 JJ03 JJ16 JJ26 LL21 QQ24 QQ25 QQ26 QQ32 RR09 2G051 AA51 AA56 AB02 AB20 CA03 CA04 EA14 EB01 EB02 ADB13A01 DB01A04 DJ18 5B057 AA03 CH09 DA03 DA08 DB02 DC08 DC16 DC36

Claims (4)

[Claims] 1. A pattern inspection apparatus for detecting the presence / absence of a defect in a pattern formed by a plurality of types of parts on an object to be inspected, comprising: applying a filter of a predetermined function to the design data of the pattern; Reference data calculating means for calculating a characteristic part of the pattern based on the reference data; and a boundary for calculating a boundary direction of a boundary between different types in the pattern based on the characteristic part. Direction calculating means, the reference data has a predetermined number of pixels arranged along a direction orthogonal to the boundary direction, and a value corresponding to a pixel located at the center corresponds to a pixel in the vicinity thereof Correction reference data calculating means for calculating correction reference data by applying a filter of a function that changes based on a value to be detected, and detecting a pattern on the inspection object. A sensor for, the correction reference data and comparing the sensor data obtained from the sensor, a pattern inspection apparatus characterized by and a defect detection means for detecting a defect based on the difference. 2. The pattern inspection apparatus according to claim 1, wherein a weighting factor is set for each of said pixels. 3. A pattern inspection method for detecting the presence / absence of a defect in a pattern formed by a plurality of types of portions on an object to be inspected, wherein the reference data is obtained by applying a filter of a predetermined function to the design data of the pattern. A reference data calculating step of calculating a pattern part; a characteristic part extracting step of extracting a characteristic part of a pattern based on the reference data; and a boundary calculating a boundary direction of a boundary between different types in the pattern based on the characteristic part. A direction calculating step, wherein the reference data has a predetermined number of pixels arranged along a direction orthogonal to the boundary direction, and a value corresponding to a pixel located at the center corresponds to a pixel in the vicinity thereof A correction reference data calculating step of calculating correction reference data by applying a filter of a function that changes based on a value to be detected, and detecting the pattern on the inspection object. Detecting step and a pattern inspection method for comparing the sensor data obtained in the correction reference data and the detection step, characterized in that it comprises a defect detection step of detecting a defect based on the difference of. 4. The pattern inspection method according to claim 1, wherein a weighting factor is set for each of said pixels.