JP3967327B2 - Mask defect inspection method - Google Patents

Description

  The present invention relates to a defect inspection method for a mask used for manufacturing a device such as a semiconductor integrated circuit or a liquid crystal panel.

  In manufacturing an LSI, mask design data is created based on LSI design data obtained by converting a desired LSI design pattern into data, a mask is manufactured based on the mask design data, and a reduced pattern of the mask is formed in a lithography process. By faithfully transferring the wafer onto the wafer, an LSI according to the original LSI design pattern is obtained.

  In recent years, with the progress of higher integration of LSIs and the miniaturization of element sizes built into LSIs, the fidelity of pattern transfer in the lithography process has begun to become a problem. Specifically, the reduced pattern of the mask needs to be faithfully transferred onto the wafer. For example, the pattern deterioration such as a corner that should be 90 ° rounded, a line end shortened, a line width thickened / thinned, etc. As a result, the same pattern as the LSI design pattern cannot be formed on the wafer. Such a phenomenon is referred to as an optical proximity effect (Optical Proximity effect).

  The optical proximity effect was originally intended for a phenomenon caused by optical factors at the time of transfer, but at present, it generally refers to the above phenomenon that occurs throughout the entire wafer process.

  As a matter of course, in manufacturing an LSI, in order to achieve a desired device performance, it is necessary to form a pattern having a size and shape according to the LSI design pattern on the wafer.

  For this reason, in recent years, various methods have been actively studied on the assumption that a method for applying a mask pattern correction that anticipates pattern degradation due to the wafer process, that is, optical proximity effect correction (OPC), is effective. Has been proposed and implemented.

  However, when the optical proximity effect correction is performed on the mask design data, the following phenomenon occurs.

  That is, the mask design data before performing the optical proximity effect correction satisfies a predetermined design rule based on a distance capable of inspecting a mask defect with respect to a pattern size, an inter-pattern distance, and the like. However, by applying optical proximity effect correction to the mask design data, the pattern size, inter-pattern distance, etc. are deformed, and as a result, mask design data after correction of optical proximity effect (hereinafter referred to as corrected mask design data) and obtained. On the actual mask, there may be a region including a portion that violates the design rule originally planned.

  The above phenomenon will be specifically described below with reference to FIG.

  FIG. 15A shows a design pattern based on LSI design data and a pattern obtained on an actual wafer.

  The LSI design data is designed so that a rectangular contact hole (design) 152 is formed under the wiring 151. In particular, when focusing on the contact hole (design) 152, the design rule is observed in the LSI design data, and the gap 154 between the contact holes (design) 152 is kept at a sufficient distance. Mask design data and a mask created based on such LSI design data also satisfy the design rule.

  However, in reality, when a lithography process is performed on the wafer using the mask data and the mask created using the LSI design data, the contact hole (actual) 153 cannot obtain a desired shape due to the optical proximity effect. . Therefore, it is necessary to perform optical proximity effect correction on the mask design data based on the LSI design data and manufacture a mask from the corrected mask design data.

  FIG. 15B uses a design pattern based on LSI design data similar to FIG. 15A, a mask pattern based on corrected mask design data, and a mask created based on this corrected mask design data. The pattern actually obtained on the wafer is shown.

  Similar to FIG. 15A, the LSI design data is designed so that a rectangular contact hole (design) 152 is formed under the wiring 151. On the other hand, the contact hole (after correction) 155 is elongated in the corrected mask design data, and the gap 156 between the contact holes (after correction) 155 is reduced. A contact hole (actual) 157 on the actual wafer obtained by forming a pattern on the wafer using this mask has a shape close to the contact hole (design) 152 in the LSI design.

  However, the interval 156 between the contact holes (after correction) 155 on the corrected mask design data is narrowed, and a phenomenon that the design rule based on the mask defect inspection distance is violated has occurred.

  When the optical proximity effect correction is performed on the mask design data in this way, a phenomenon may occur in which the mask design data and the mask have a region including a portion that violates a design rule based on a distance capable of inspecting the mask defect. is there. When conventional mask defect inspection is performed on such mask design data and mask, a defect that does not actually exist at a location that violates the design rule is detected as a defect (that is, a pseudo defect is detected), and the verification is performed. There is a problem that it takes a lot of time and the productivity of the mask is remarkably lowered.

  That is, the mask defect inspection is performed as follows. FIG. 16 shows a mask defect inspection flow. In the mask defect inspection, an optical sensor is used by using a simulator in a reference data generation circuit 25 for sensor data 23 obtained by reading the shape of the mask 21 actually obtained by the optical sensor circuit 22 and the corrected mask design data 24. The comparison circuit 27 compares the function taking into account the sensitivity characteristics with reference data 26 created by performing convolution integration, and detects the presence or absence of a defect from the difference between the two patterns (28).

  However, as a result of the optical proximity effect correction, the figure on the pattern based on the mask design data after correction becomes smaller than a predetermined distance that can be inspected accurately, or an area where the figures are closer than the predetermined distance. When it occurs, the difference between the reference data 26 created from the corrected mask design data 24 and the sensor data 23 becomes larger than the originally expected value, and this difference is detected as a pseudo defect. This is because the area on the actual mask 21 cannot be accurately detected by the optical sensor circuit 22, and the difference between the actual mask 21 shape and the shape indicated by the obtained sensor data 23 is enlarged. This is because when the reference data 26 is calculated by the data generation circuit 25, such an increase in difference cannot be simulated correctly.

  Therefore, conventionally, when performing mask defect inspection on corrected mask design data and a mask having a region including a portion that violates a design rule based on a distance that can be subjected to mask defect inspection by performing optical proximity effect correction, There has been a demand for a mask defect inspection method capable of eliminating such pseudo defects resulting from optical proximity effect correction.

  In addition, there has been a demand for a method for creating mask design data in which the occurrence of a portion that violates a design rule based on a distance capable of inspecting a mask defect even when optical proximity effect correction is performed is small.

  The present invention has been made in view of the above problems, and in the mask defect inspection process, a mask defect that is caused by performing optical proximity effect correction on mask design data based on design data such as LSI is reduced. It is an object of the present invention to provide a mask defect inspection method capable of improving the productivity by simplifying the defect inspection process and contributing to improving the precision of a fine pattern formed on a wafer or the like.

(First invention)
According to a first aspect of the present invention, there is provided a mask defect inspection mask from corrected mask design data obtained by performing correction for improving the fidelity of a pattern formed on a sample with respect to mask design data obtained based on device design data. A reference data forming step for obtaining reference data;
A sensor data forming step for obtaining sensor data obtained by actually measuring the shape of the mask created based on the corrected mask design data;
In the mask defect inspection method comprising the defect detection step of comparing the reference data and the sensor data and detecting the presence or absence of defects on the mask,
The reference data forming step includes a step of extracting a region including a gap between each graphic on the corrected mask design data or a region where the size of each graphic is smaller than a predetermined value set in advance;
The extracted region includes a step of creating the reference data using a method of simulating a mask shape with higher accuracy than other regions.

(Operation of the first invention)
According to the first aspect of the present invention, when performing correction for improving the fidelity of a pattern formed on a sample, that is, when performing a defect inspection of a mask created based on corrected mask design data subjected to optical proximity effect correction. Compared with other areas, the area including the gap between the figures on the mask design data after correction or the size of each figure set in advance is smaller than a predetermined value S1 at which a mask defect can be inspected, for example. Analysis is performed using a high-precision simulator, and reference data with very little difference from the sensor data is created for this area. Therefore, it is possible to avoid the problem of detecting a pseudo defect in the region and reduce the burden of the mask defect inspection process.

(Second invention)
According to a second aspect of the present invention, the mask design data obtained from the corrected mask design data obtained by correcting the mask design data obtained based on the device design data for improving the fidelity of the pattern formed on the sample is used for mask defect inspection. A reference data forming step for obtaining reference data;
A sensor data forming step for obtaining sensor data obtained by actually measuring the shape of the mask created based on the corrected mask design data;
In the mask defect inspection method comprising the defect detection step of comparing the reference data and the sensor data and detecting the presence or absence of defects on the mask,
The reference data forming step includes a step of extracting a region including a gap between each graphic on the corrected mask design data or a region where the size of each graphic is smaller than a predetermined value set in advance;
The extracted region is excluded from the mask defect inspection target, and the reference data is created only for the region other than the extracted region.

(Operation of the second invention)
According to the second aspect of the invention, when performing the defect inspection of the mask created based on the corrected mask design data subjected to the optical proximity effect correction, the correction data is generated when the reference data created from the corrected mask design data is created. Only reference data other than an area including a portion where the gap between figures on the post-mask design data or the size of each figure is smaller than a predetermined value S1 that can be inspected for mask defect is created. Since it is excluded from the target, it is possible to avoid the problem of detecting pseudo defects in the region, and to reduce the burden of the mask defect inspection process.

(Third invention)
According to a third aspect of the present invention, there is provided a mask defect inspection data from a corrected mask design data obtained by performing correction for improving the fidelity of a pattern formed on a sample with respect to mask design data obtained based on device design data. A reference data forming step for obtaining reference data;
A sensor data forming step for obtaining sensor data obtained by actually measuring the shape of the mask created based on the corrected mask design data;
In the mask defect inspection method comprising the defect detection step of comparing the reference data and the sensor data and detecting the presence or absence of defects on the mask,
The reference data forming step includes a step of extracting a region including a portion where a gap between each figure on the corrected mask design data is smaller than a predetermined value S1 set in advance;
Auxiliary pattern additional data obtained by adding auxiliary pattern data having a size near or smaller than the minimum size W1 capable of defect inspection between adjacent figures in the extracted area to the corrected mask design data. Obtaining a step;
And a step of creating the reference data from the auxiliary pattern additional data.

(Operation of the third invention)
According to the third aspect of the present invention, when performing the defect inspection of the mask created based on the corrected mask design data subjected to the optical proximity effect correction, the reference data is generated after the correction when creating the reference data from the design corrected mask design data. An area including a portion where the gap between each figure on the mask design data is smaller than a predetermined value S1 that can be inspected for a mask defect is extracted, and the difference between the actual mask and the sensor data is extracted for the area. In view of this, auxiliary pattern additional data is created by adding virtual auxiliary pattern data between figures whose gap is smaller than a predetermined value S1 set in advance with respect to the corrected mask design data. Create reference data from Thereby, reference data with very little difference from the sensor data is created for the area. Therefore, it is possible to avoid the problem of detecting a pseudo defect in the region and reduce the burden of the mask defect inspection process.

  According to the mask defect inspection method of the present invention, in the mask defect inspection process, the pseudo defect caused by performing optical proximity effect correction on the mask design data based on the design data such as LSI is reduced, and the mask defect inspection process is simplified. Thus, the productivity can be improved and the accuracy of the fine pattern formed on the wafer can be improved.

  Further, the mask design data creation method applied to the mask defect inspection method performs optical proximity effect correction in the mask defect inspection process even if optical proximity effect correction is applied to mask design data based on design data such as LSI. Can create mask design data with few pseudo defects caused by the process, simplify the mask defect inspection process and improve productivity, and contribute to improving the precision of fine patterns formed on wafers and the like. .

  Hereinafter, details of the present invention will be described with reference to examples.

Example 1
The Example of 1st invention is shown.

  Mask preparation and mask defect inspection were performed as described below.

<Creating a mask>
First, device design data was created. Next, mask design data was obtained based on the device design data, optical proximity effect correction was performed on the mask design data, and corrected mask design data was obtained.

  FIG. 1 shows a mask design pattern 10 based on the mask design data and a mask design pattern 11 (shaded portion) after correction based on the mask design data after correction.

  In the mask design pattern 10 based on the mask design data, there was no region that violated the design rule that would be a pseudo defect at the time of mask defect inspection. On the other hand, the corrected mask design pattern 11 based on the corrected mask design data has a region 12 including a portion that does not maintain the predetermined inter-graphic distance S1 that can be inspected for a mask defect.

  On the other hand, a mask was actually created according to the corrected mask design data.

<Mask defect inspection>
Next, mask defect inspection was performed on the created mask using a mask defect inspection apparatus. A flowchart of the mask defect inspection is shown in FIG.

  First, the corrected mask design data 24 is input to the reference data generation circuit 25 by the corrected mask design data input means (not shown), and the mask is generated from the corrected mask design data 24 by the reference data generation circuit 25 as described later. A reference data forming step for obtaining the reference data 26 for defect inspection was performed.

  On the other hand, for the created mask 21, a sensor data formation process was performed in which the optical sensor circuit 22 measured the shape of the mask to obtain sensor data 23 of the mask. Finally, the defect extraction 28 is performed by comparing the sensor data 23 and the reference data 26 by the comparison circuit 27.

  The mask defect inspection apparatus used includes at least the optical sensor circuit 22 having the above-described operation, the mask design data input means, the reference data generation circuit 25, the comparison circuit 27, and the extracted defect output means.

  The reference data generation process for obtaining the reference data 26 from the corrected mask design data 24 in the reference data generation circuit 25 will be described in more detail.

  FIG. 3 shows a flowchart of the reference data forming process. First, the corrected mask design data is input (31), and an area including a portion where the gap between each figure or the size of each figure is smaller than a predetermined value S1 that can be inspected for a mask defect is extracted ( 32). For regions other than the extracted region, mask pattern simulation is performed using a normal simulator, and reference data is obtained by performing convolution integration with a function considering the sensitivity characteristics of the optical sensor (33).

  For the extracted area, it is first determined whether or not the arrangement of the pattern is listed in the rule table for generating reference data (34). This reference data generation rule table can search for existing reference data using the pattern arrangement as an index. For those listed in the rule table, the existing reference data is generated by referring to the rule table (35). For those not listed in the rule table, reference data is generated by performing a mask pattern simulation with higher accuracy than the normal simulator (36). The obtained mask pattern simulation result is additionally recorded in the rule table (37).

  The reference data 38 was obtained from the corrected mask design data by performing such operations.

  Thus, the sensor data and the region including the portion that does not maintain the predetermined inter-graphic distance S1 that can be inspected by performing the mask defect inspection are subjected to a higher-precision simulation than the other regions. Reference data with very little difference can be created. Therefore, it is possible to avoid the problem of detecting a pseudo defect in the region and reduce the burden of the mask defect inspection process.

  As a reference data forming step, in addition to the above method, reference data is sequentially obtained with respect to the corrected mask design data using a normal simulator, and a gap between each figure or a size of each figure is set in advance. Reference data may be obtained by performing high-precision mask pattern simulation each time a region smaller than the value S1 is reached. At this time, it is desirable to accumulate the pattern arrangement and the obtained high-precision simulation result as a rule table so that the high-precision simulation result can be referred to from the rule table every time the same arrangement is reached.

  As the high-accuracy mask pattern simulation for the extracted region, a method for describing a fine gap with higher accuracy than the simulation for other than the extracted region is used. Specifically, for example, a method of making the size of a pixel for creating reference data finer to 1/2, 1/4, or the like, a method of using a mask drawing simulator and an optical simulator together, and the like can be mentioned.

  The reference data forming step in the mask defect inspection method of the present invention may be performed in real time in the mask defect inspection apparatus or may be performed off-line.

(Example 2)
The Example of 2nd invention is shown.

  <Mask creation> was performed in the same manner as in Example 1, and <Mask defect inspection> was performed in the same manner as in Example 1 except that the following reference data formation step was performed.

  FIG. 4 shows a flowchart of the reference data forming process.

  First, the corrected mask design data is input (41), and an area including a portion where the gap between the figures or the size of each figure is smaller than a predetermined value S1 that can be inspected for a mask defect is extracted ( 42). For regions other than the extracted region, mask pattern simulation is performed using a normal simulator, and convolution integration is performed with a function considering the sensitivity characteristics of the optical sensor to obtain reference data (43).

  On the other hand, no reference data is created for the extracted area (44).

  The reference data 45 was obtained from the corrected mask design data by performing the above operations.

  By performing the mask defect inspection in this way, the mask defect inspection is performed only on the region excluding the region including the portion that does not maintain the predetermined inter-graphic distance S1 that can be inspected by the mask defect. Therefore, it is possible to avoid the problem of detecting a pseudo defect in the region and reduce the burden of the mask defect inspection process.

  The reference data forming step in the mask defect inspection method of the present invention may be performed in real time in the mask defect inspection apparatus or may be performed off-line.

(Example 3)
The Example of 3rd invention is shown.

  Mask preparation and mask defect inspection were performed as described below.

<Creating a mask>
First, device design data was created. Next, mask design data obtained based on the device design data was obtained, optical proximity effect correction was performed on the mask design data, and corrected mask design data was obtained.

  FIG. 5A shows a part of the corrected mask design pattern based on the corrected mask design data.

  In the mask design pattern based on the mask design data, there was no region that violated the design rule that would be a pseudo defect at the time of mask defect inspection. On the other hand, the post-correction mask design pattern 51 based on the post-correction mask design data includes a portion where a predetermined inter-graphic distance that can be inspected for mask defects indicated by inter-graphic distances a1, a2, and a3 is S1 or less.

  On the other hand, a mask was actually created based on the corrected mask design data.

<Mask defect inspection>
Next, mask defect inspection was performed on the created mask using a mask defect inspection apparatus. A flowchart of the mask defect inspection is shown in FIG.

  First, the corrected mask design data 24 is input to the reference data generation circuit 25 by the corrected mask design data input means (not shown), and the corrected mask design data 24 is input to the reference data generation circuit 25 as described later. A reference data forming step for obtaining reference data 26 for mask defect inspection was performed.

  On the other hand, for the created mask 21, a sensor data formation process was performed in which the optical sensor circuit 22 measured the shape of the mask to obtain sensor data 23 of the mask. Finally, the defect extraction 28 is performed by comparing the sensor data 23 and the reference data 26 by the comparison circuit 27.

  The used mask defect inspection apparatus includes at least the optical sensor circuit 22, the mask design data input means, the reference data generation circuit 25, the comparison circuit 27, and the extracted defect output means having at least the above-described functions.

  The reference data generation process for obtaining the reference data 26 from the corrected mask design data 24 in the reference data generation circuit 25 will be described in more detail.

  FIG. 6 shows a flowchart of the reference data forming process. First, the corrected mask design data is inputted (61), and an area including a portion where the gap between the figures or the size of each figure is smaller than a predetermined value S1 that can be inspected for a mask defect is extracted ( 62). For regions other than the extracted region, mask pattern simulation is performed using a normal simulator, and convolution integration is performed with a function considering the sensitivity characteristics of the optical sensor to obtain reference data (63).

  For the extracted region, for example, auxiliary pattern additional data corrected so as to arrange a small auxiliary pattern in the vicinity of or smaller than the minimum size W1 that can be mask-drawn between adjacent sides is created ( 64).

  FIG. 5B shows an auxiliary pattern addition pattern based on auxiliary pattern addition data obtained by adding an auxiliary pattern to the corrected mask design data. An auxiliary pattern 52 is added to the corrected mask design pattern 51. The auxiliary pattern 52 may change the shape and size of the pattern according to the inter-pattern distance (a1, a2, a3) and the length of the opposite sides.

  A mask pattern simulation is performed on the auxiliary pattern additional data in which such auxiliary patterns are arranged by using a normal simulator, and a reference data is obtained by performing a convolution integration with a function considering the sensitivity characteristic of the optical sensor (65).

  By performing such an operation, reference data 66 was obtained from the corrected mask design data.

  If a reference pattern is created from corrected mask design data that does not place an auxiliary pattern, the difference between the reference pattern and the sensor pattern will increase, and a pseudo defect will occur. Since the reference pattern is created from the auxiliary pattern addition data with the mask design data auxiliary pattern added, comparing the sensor patterns obtained from the actual masks where the auxiliary pattern is not placed, both are judged to be the same figure, and the pseudo defect The inspection can be continued without generating any problems.

    For the mask defect inspection, enlarged mask design data was created as follows.

(1-1) <Creation of enlarged mask design data>
FIG. 7 shows a flowchart for creating enlarged mask design data in this example.

  First, device design data was created (not shown). Next, mask design data was obtained based on the device design data (71). Next, optical proximity effect correction was performed on the mask design data to obtain corrected mask design data (72).

  FIG. 8A shows a mask design pattern 81 based on the mask design data and a mask design pattern 82 (shaded portion) based on the corrected mask design data.

  In the shape of the contact hole 81 based on the mask design data, there was no region that violated the design rule that would be a pseudo defect at the time of mask defect inspection. However, the shape of the contact hole 82 based on the mask design data after correction is smaller than a predetermined value S1 at which the distance between the figures approaches by correction and the mask defect inspection can be performed without any problem.

  Next, from the corrected mask design data, an area including a gap between each figure on the corrected mask design data or a region including a part where the size of the figure is smaller than a predetermined value S1 that can be inspected for a mask defect is extracted. (73).

  Next, as shown below, with respect to the mask design data after correction, a figure is connected, a figure is enlarged, or a figure at a location where the distance between figures in the extracted area is smaller than a predetermined value S1. A process (enlargement correction process) (74) for correcting the gap between them was performed, and enlarged mask design data 75 was obtained.

  Examples of the enlargement correction processing include the following (i) to (iv), and (i) to (iv) can be performed alone or in combination.

(I) A method of connecting a figure by arranging a pattern having a length in the gap direction of S1 or more at a position where the gap between the figures in the extracted region is smaller than S1.
(Ii) A method of creating a connection pattern in which the periphery of the figure in the extracted region is thickened by a predetermined value S2 of 0.5 × S1 or more and connecting the figure by narrowing the periphery of the connection pattern by S2.
(Iii) A figure in the extracted area is A, and a connection pattern is created by thickening A by a predetermined value S2 of 0.5 × S1 or more, and a pattern obtained by narrowing the periphery of this connection pattern by S2 is then created. A connection pattern in which B is thinned by a predetermined distance S3 larger than 0.5 × S1, and a pattern obtained by thickening the next S3 by S3 is defined as C, and a pattern obtained by subtracting C from B is defined as D. A pattern in which D is thickened by a predetermined value S4 is E, and a figure is connected by adding E to C.
(Iv) When a location where the size of the figure in the extracted area is smaller than a predetermined value W1 at which mask defect inspection is possible occurs, a pattern having a size greater than or equal to W1 is arranged at the location and the size of the figure How to enlarge,
And the like.

  A specific example of the enlargement correction process will be described with reference to a pattern diagram 8.

  FIG. 8 is a diagram showing a change in pattern shape due to enlargement correction processing.

  First, according to the above (ii), the periphery of each figure of the mask design pattern 82 based on the corrected mask design data shown in FIG. 8A is thickened by a predetermined value S2 to create a connection pattern, and then the periphery of this connection pattern Is reduced by S2 to obtain a first enlarged pattern 83 shown in FIG. At this time, for example, S2 takes a value satisfying 0.5 × S1 ≦ S2. By the processing so far, the figures arranged close to each other in a straight line are connected without any problem, but the figures arranged close to each other at an angle are thinly bridged as shown at the right end of FIG. If this bridging part does not violate the design rules for drawing or mask defect inspection, the mask may be manufactured using the enlarged mask design data corresponding to this first enlarged pattern. Therefore, the following processing is further performed according to (iii) and (iv).

  Next, a reduced pattern is formed by narrowing the periphery of the first enlarged pattern 83 by a predetermined value S3, and the periphery of the reduced pattern is further thickened by S3 to obtain a second enlarged pattern 84 shown in FIG. At this time, for example, S3 takes a value more than half of the lower limit of the pattern thickness allowed in the data. Next, the second enlarged pattern 84 is subtracted from the first enlarged pattern 83 to obtain a patch pattern 85 shown in FIG. Finally, an enlarged patch pattern obtained by thickening the patch pattern 85 by a predetermined size S4 is added to the second enlarged pattern 84, whereby the third enlarged pattern 86 with the patch pattern added as shown in FIG. 8E can be obtained. At this time, for example, S4 takes a value about half the lower limit of the pattern thickness allowed in the data.

  Enlarged mask design data corresponding to the third enlarged pattern 86 thus obtained was obtained.

  A mask was created based on the enlarged mask design data. FIG. 8F shows a pattern 87 (hatched line) transferred onto the wafer from the obtained actual mask, a mask design pattern 81 (solid line) based on the original mask design data, and an enlarged mask based on the enlarged mask design data. A design pattern 86 (dotted line) is shown superimposed. Since the area of the pattern added later is small in this way, there is little influence on the pattern 87 transferred from the actual mask onto the wafer, and the desired dimension close to the mask design pattern 81 based on the original mask design data is solved. I was able to image.

  When the enlarged mask design data is used in this way, a portion where the gap and size between figures are smaller than the predetermined distance S1 can be cleared from the data, and mask drawing can be facilitated.

  Further, reference data is created based on the enlarged mask design data, while sensor data obtained from a mask created based on the enlarged mask design data is created, and the reference data and the sensor data are compared. As a result, detection of pseudo defects in mask defect inspection can be suppressed.

  Further, for example, FIGS. 8G and 8H show the drawing data of FIGS. 8A and 8C. Before performing the step of obtaining enlarged mask design data in (g), this pattern is composed of nine drawing figures. After this step, it is composed of ten drawing figures. In this way, there is little risk of explosion of the drawing data amount even when the enlarged mask design data is used. Further, the six rectangles of numbers 1-6 shown in FIG. 8G become two rectangles of numbers 1 and 2 shown in FIG. 8H by this processing, so that mask drawing becomes easy.

  As described above, even if the enlarged mask design pattern 86 is used, the actual mask pattern 87 is hardly affected and can be resolved to a desired dimension close to the mask design pattern 81 based on the original mask design data. As shown in FIG. 8 (f), in the actual mask pattern 87, a part of the pattern is connected. Therefore, even if a connecting part is generated on such a mask, only the part that does not affect the device performance is enlarged. It is necessary to use mask design data.

  For example, FIG. 9 shows a diagram in which the region 88 in FIG. Referring to FIG. 9, it will be described that there are cases where there is no adverse effect on device performance even if a device is created using a mask in which the enlarged portion is generated using enlarged mask design data.

  FIG. 9A shows a wiring layer 91 formed on the wafer and a contact hole 94 (solid line) formed in the wiring layer 91. Further, a mask design pattern 92 (shaded portion) after correction based on mask design data after optical proximity effect projection projected onto the wafer in order to form the contact hole 94-no connection portion, and an enlarged mask based on the enlarged mask design data The design pattern 93 (dotted line) —with a connecting portion is shown.

  FIGS. 9B to 9G are process diagrams showing the process of forming the contact hole 94 on the wafer using the mask having the pattern 92 or the mask having the pattern 93 by the damascene process. FIGS. 9B to 9G are cross-sectional views on the wafer in the direction indicated by the arrow 95 shown in FIG.

  In FIG. 9 (b), a TEOS (tetraethoxysilane tetrasilane) layer 96 and an amorphous silicon layer 97 are formed on the wafer and subjected to an etching process. A resist 98 is applied to this, and the result of exposure / development using the pattern 92-no connection portion as a mask is shown in FIG. 9C, and the result of exposure / development using the pattern 93-connection portion as a mask is shown in FIG. 9C '. Shown in In FIG. 9C, the resist 98 ′ remains between the contact hole patterns, but in FIG. 9C ′, no resist remains between the contact hole patterns.

  Next, the results of etching FIG. 9C and FIG. 9C ′ are shown in FIG. 9D and FIG. 9D ′, respectively. In either case, the TEOS layer 96 is etched using the resist layer 98 and the amorphous silicon layer 97 as a mask.

  Next, when the resist 98 is removed in FIGS. 9D and 9D ′, the pattern shown in FIG. 9E is obtained on the wafer. The TEOS layer 97 is etched from FIG. 9 (e) to obtain the shape of FIG. 9 (f), the amorphous silicon layer 96 is peeled off, tungsten 99 is embedded in the contact hole, and chemical mechanical polishing is performed, so that FIG. ) Is obtained.

  Thus, for example, the use of a mask having an enlarged mask design pattern in the contact hole forming step does not affect the wafer shape, and does not adversely affect the device characteristics.

(1-2) <Creation of enlarged mask design data>
Enlarged mask design data was created in the same manner as in the example (1-1) except that the enlargement correction process was performed as follows.

  However, the following is a region including a portion where the gap between each figure extracted from the mask design pattern after correction based on the mask design data after optical proximity effect correction is smaller than a predetermined value S1 set in advance. there were.

  FIG. 10A shows a part of the corrected mask design pattern based on the corrected mask design data used in this example. In the mask design pattern based on the mask design data, there was no region that violated the design rule that would be a pseudo defect at the time of mask defect inspection. On the other hand, the corrected mask design pattern 101 based on the corrected mask design data includes a portion where a predetermined inter-graphic distance that can be inspected for mask defects indicated by inter-graphic distances a1, a2, and a3 is S1 or less. However, it is assumed that a1 <a2 <a3 = S1.

  The corrected mask design data was subjected to enlargement correction processing to obtain enlarged mask design data. The enlargement correction processing was corrected using the values shown in the pattern correction table shown in FIG. That is, in each figure, if the gap between figures is a1 or less, it is 1/2 of a1, if the gap is greater than a1 and less than a2, it is 1/2 of a2, and if it is greater than a3 and less than a3, it is 1/2 of a3. Enlarged mask design data was obtained by correcting the opposite sides to be thickened by the width. An enlarged mask design pattern based on the enlarged mask design data is shown in FIG. The fattened side is shown at 102.

  Next, when a mask was created based on this enlarged mask design data, it was possible to resolve to a desired dimension close to the mask design pattern based on the original mask design data.

  As described above, in this example, with respect to the corrected mask design data, a portion where the gap between the figures on the corrected mask design data is smaller than the predetermined distance S1 at which the mask defect can be inspected is classified into several stages according to the distance between the figures. Then, a process for connecting the figures was performed by correcting the sides facing the corresponding places according to each classification by increasing the thickness by half or more of the distance between the figures, thereby obtaining enlarged mask design data.

  When the enlarged mask design data obtained in this way is used, portions where the gap and size between figures are smaller than the predetermined distance S1 can be cleared from the data, and mask drawing can be facilitated.

  Further, reference data is created based on the enlarged mask design data, while sensor data obtained from a mask created based on the enlarged mask design data is created, and the reference data and the sensor data are compared. As a result, detection of pseudo defects in mask defect inspection can be suppressed.

  Further, according to the correction process for connecting the figures in this example, it is possible to connect the figures without increasing the number of vertices and the amount of pattern data for patterns arranged obliquely close to each other.

  When the enlarged mask design data as in this example is used, there is little influence on the actual mask pattern, and it can be resolved to a desired dimension close to the mask design pattern based on the original mask design data. In the pattern formed on the mask, a part of the pattern is connected. Therefore, it is necessary to use the enlarged mask design data only for a portion that does not affect the device performance even if the connection occurs on the mask.

  Enlarged mask design data was created as shown below.

(1-3) <Creation of enlarged mask design data>
Enlarged mask design data was created in the same manner as in the example (1-1) except that the enlargement correction process was performed as follows.

  However, the design pattern based on the mask design data after optical proximity effect correction has the following areas including a portion where the gap between each figure extracted from the figure becomes smaller than a predetermined value S1 set in advance. .

  FIG. 11A shows a part of the corrected mask design pattern based on the corrected mask design data used in this example. In the mask design pattern based on the mask design data, there was no region that violated the design rule that would be a pseudo defect at the time of mask defect inspection. On the other hand, the corrected mask design pattern 111 based on the corrected mask design data includes a portion where a predetermined inter-graphic distance that can be inspected for a mask defect indicated by inter-graphic distances a1, a2, and a3 is S1 or less. However, it is assumed that a1 <a2 <a3 = S1.

  The corrected mask design data was subjected to enlargement correction processing to obtain enlarged mask design data. The enlargement correction process was performed as follows. First, as shown in FIG. 11 (c), the values shown in the pattern correction table shown in FIG. 11 (b) for the opposite side 112 in the figure 111 on the corrected mask design data whose inter-graphic gap is S1 or less. Is used to make a figure 113 with the sides thickened vertically and horizontally. Next, from the corrected mask design pattern figure 111, an overlapping part 114 (shaded part) with the figure 113 in which the side is thickened vertically and horizontally was subtracted. As a result, enlarged mask design data shown in the graphic 115 was obtained.

  The amount of thickening the side is the value shown in the pattern correction table shown in FIG. 11B. If the gap between figures is a1 or less, it is 1/2 of a3, and if the gap is larger than a1 and less than a2 ( If it is 1/2 of a3-a1) and greater than a2 and less than or equal to a3, it is set to 1/2 the size of (a3-a2).

  When a mask was created based on the enlarged mask design data, it was possible to resolve to a desired dimension close to the mask design pattern based on the original mask design data.

  In this way, in this example, the figure is deleted so that the gap between the figures on the corrected mask design data is smaller than the predetermined value S1 at which the mask defect can be inspected is more than the distance S1 with respect to the corrected mask design data. The enlarged mask design data was obtained by correcting the gap between them.

  When the enlarged mask design data obtained in this way is used, portions where the gap and size between figures are smaller than the predetermined distance S1 can be cleared from the data, and mask drawing can be facilitated.

  Further, reference data is created based on the enlarged mask design data, while sensor data obtained from a mask created based on the enlarged mask design data is created, and the reference data and the sensor data are compared. As a result, detection of pseudo defects in mask defect inspection can be suppressed.

  Further, according to the correction process for connecting the figures in this example, it is possible to connect the figures without increasing the number of vertices and the amount of pattern data for patterns arranged obliquely close to each other.

  The corrected enlarged mask design data was created as follows.

(2-1) <Creation of enlarged mask design data after correction>
FIG. 12 shows a flowchart for creating the corrected enlarged mask design data in this example. First, device design data was created (not shown). Next, mask design data was obtained based on the device design data (121).

  FIG. 13A shows a mask design pattern 131 based on this mask design data. In the mask design pattern 131, there was a portion of a distance R1 or less where the distance between each figure is predicted to be closer to a predetermined distance S1 that can be inspected for a mask defect by performing the optical proximity effect correction. FIG. 13A shows this location as a location where the inter-pattern distances are a1, a2, and a3. However, it is assumed that a1 <a2 <a3 = R1.

  Next, if it is assumed from the mask design data that the optical proximity effect correction is performed on the mask design data, it is predicted that the distance between the figures will be equal to or less than a predetermined distance S1 that can be inspected for a mask defect after the correction. A region including a portion having a distance R1 or less was extracted (122).

  Next, as shown below, with respect to the mask design data, a figure is connected, a figure is enlarged, or a figure between the figures at a location where the distance between the figures in the extracted region is smaller than a predetermined value R1. Processing for correcting the gap to be widened (enlargement correction processing) (123) was performed to obtain enlarged mask design data.

  In the enlargement correction process, the mask design data was corrected using the values shown in the pattern correction table shown in FIG. That is, if the interval between figures is a1 or less, it is 1/2 of a1, if the space is greater than a1 and less than a2, it is 1/2 of a2, and if it is greater than a3 and less than a3, it is a width of 1/2 of a3. Thicken the opposite sides at a distance of. An enlarged mask design pattern based on the enlarged mask design data is shown in FIG. The fattened side is shown at 132.

  Next, optical proximity effect correction was performed on the enlarged mask design data (124), and corrected enlarged mask design data was obtained (125).

  FIG. 13D shows an enlarged mask design pattern 133 (dotted line) based on the enlarged mask design data and a pattern 134 (hatched portion) based on the corrected enlarged mask design data.

  When a mask was created on the basis of the corrected enlarged mask design data, it was possible to resolve to a desired dimension close to the mask design pattern based on the original mask design data.

  As described above, according to this example, the mask design data obtained based on the device design data includes a portion where the inter-graphic distance or size is predicted to be smaller than the predetermined distance S1 by the optical proximity effect correction. Prior to the optical proximity correction, correction (enlargement correction processing) is performed to connect the graphic to the graphic in the area, or to expand the size of the graphic, or to expand the gap between the graphic. In other words, since the mask design data is subjected to enlargement correction processing, optical proximity correction is performed, and no additional figures are added after that, in addition to the advantage of preventing an increase in the number of vertices and the amount of data, corrected post-correction enlarged mask design data Can be obtained. In addition, it is possible to connect figures without increasing the number of vertices and the amount of pattern data for figures arranged in close proximity to each other.

  Further, reference data is created based on the corrected enlarged mask design data, while sensor data obtained from a mask created based on the enlarged mask design data is created, and the reference data and the sensor data are By comparing, it is possible to suppress detection of pseudo defects in mask defect inspection. In addition to mask defect inspection, drawing can be facilitated.

  In this example, the enlargement correction processing applied to the mask design data classifies the portions where the gap between the figures on the mask design data is smaller than the predetermined distance R1 into several stages according to the distance between the figures. Then, a process for connecting the figures was performed by correcting each side facing the corresponding part to be thickened by more than one half of the distance between the figures to obtain enlarged mask design data.

  The enlargement correction further includes the following (v) and (vi), and (v) and (vi) can be performed alone or in combination.

  (V) A method of connecting figures by arranging a pattern having a length in the gap direction of R1 or more at a position where the gap between figures in the extracted area is smaller than R1.

  (Vi) A method of creating a connection pattern in which the periphery of the figure in the extracted region is thickened by a predetermined value R1 of 0.5 × R1 or more and connecting the figure by narrowing the periphery of the connection pattern by R1.

  Note that the use of the corrected enlarged mask design pattern as in this example has little effect on the actual mask pattern and can be resolved to a desired dimension close to the mask design pattern based on the original mask design data. In the actual mask pattern, since a part of the pattern is connected, it is necessary to use the corrected enlarged mask design data only for a portion that does not affect the device performance even if the connection occurs on the mask.

(2-2) <Creation of enlarged mask design data after correction>
FIG. 12 shows a flowchart for creating the corrected enlarged mask design data in this example. First, device design data was created (not shown). Next, mask design data was obtained based on the device design data (121).

  In FIG. 14A, a mask design pattern 141 based on the mask design data and a mask design pattern 142 based on the corrected mask design data obtained when optical proximity effect correction is performed on the mask design data are superimposed. Show. In the shape of the contact hole 141 based on the mask design data, there was no region that violated the design rule that would be a pseudo defect at the time of mask defect inspection. However, the shape of the contact hole 142 based on the mask design data after correction is predicted to be smaller than a predetermined value S1 at which the distance between the figures approaches by correction and the mask defect inspection can be performed without any problem.

  Therefore, an area including a portion where a gap between figures on the corrected mask design data or a figure size is predicted to be smaller than a predetermined value S1 that can be inspected for a mask defect is extracted from the mask design data. (122).

  Next, as shown below, with respect to the mask design data, a figure is connected or a figure is enlarged for a portion where a distance between figures in the extracted region is predicted to be smaller than a predetermined value S1, or A process (enlargement correction process) (123) for correcting to widen the gap between figures was performed to obtain enlarged mask design data.

  A specific example of the enlargement correction process will be described with reference to a pattern diagram 14.

  FIG. 14 is a diagram showing a change in pattern shape by the enlargement correction processing.

  First, a mask pattern of the mask design pattern 141 based on the mask design data shown in FIG. 14A is thickened by a predetermined value S2 to create a connection pattern, and then the connection pattern is thinned by S2 to reduce the pattern of FIG. The first enlarged pattern 143 shown in FIG. At this time, for example, S2 takes a value satisfying 0.5 × S1 ≦ S2. Although the figures arranged close to each other in a straight line are connected without any problem by the processing so far, the figures arranged close to each other are slantly bridged as shown in the rightmost figure in FIG. If the bridge part does not violate the design rules for drawing and defect inspection, the mask may be manufactured using the enlarged mask design data corresponding to the first enlarged pattern. However, depending on the design rule, the bridge part violates the rule. Therefore, the following processing is further performed.

  Next, as shown in FIG. 14C, the mask design pattern 141 is subtracted from the first enlarged pattern 143 to obtain a patch pattern 144. If there is a portion of the patch pattern 144 whose width direction or length direction is smaller than the predetermined value S3, the patch pattern 144 is thickened by a predetermined value S4 in the corresponding small direction, and the auxiliary pattern 145 is obtained as shown in FIG. At this time, S3 may be a predetermined size that can be inspected for a mask defect, and S4 may be a value that is half or more of a predetermined size that can be inspected for a mask defect. Further, S3 may be classified into several stages, and a value S4 that is thickened for each stage may be set.

  The auxiliary pattern 145 thus obtained is added to the original mask design pattern 141 to obtain enlarged mask design data. FIG. 14D shows an enlarged mask design pattern based on the enlarged mask design data.

  Next, optical proximity effect correction 124 was performed on the enlarged mask design data to obtain corrected enlarged mask design data 125. FIG. 14E shows the corrected enlarged mask design pattern 146.

  When a mask was created on the basis of the corrected enlarged mask design data, it was possible to resolve to a desired dimension close to the mask design pattern based on the original mask design data. FIG. 14F shows a pattern 147 (dotted line) based on the corrected enlarged mask design data, an initial mask design pattern 141 (solid line), and a pattern 148 (on the wafer formed using the actually obtained mask). Write a diagonal line.

  Further, for example, FIG. 14G shows the drawing data of FIG. After performing this process, it is composed of 15 drawing figures. In this way, there is little risk of the drawing data amount exploding even with the corrected enlarged mask design data.

  As described above, according to this example, the mask design data obtained based on the device design data includes a portion where the inter-graphic distance or size is predicted to be smaller than the predetermined distance S1 by the optical proximity effect correction. Prior to the optical proximity effect correction, correction (enlargement correction processing) is performed to connect the graphic to the graphic in the area, or to expand the size of the graphic, or to expand the gap between the graphic. In other words, the optical proximity effect correction is performed after the mask design data is enlarged, and the number of figures increases by the number of auxiliary patterns. However, since minute figures are unlikely to be generated at the boundary between patterns, drawing and inspection are performed with high accuracy. I can do it.

  In mask defect inspection, reference data is created based on this corrected enlarged mask design data, while sensor data obtained from a mask created based on this corrected enlarged mask design data is created, By comparing the reference data and the sensor data, it is possible to suppress false defect detection in mask defect inspection.

  Note that the use of the corrected enlarged mask design pattern as in this example has little effect on the actual mask pattern and can be resolved to a desired dimension close to the mask design pattern based on the original mask design data. In the actual mask pattern, since a part of the pattern is connected, it is necessary to use the corrected enlarged mask design data only for a portion that does not affect the device performance even if the connection occurs on the mask.

FIG. 3 is a plan view showing a mask design pattern based on mask design data according to the first embodiment. 5 is a flowchart of mask defect inspection according to the first embodiment. 5 is a flowchart of a reference data forming process according to the first embodiment. 10 is a flowchart of a reference data forming process according to the second embodiment. 10 is a plan view showing a mask design pattern based on mask design data according to Embodiment 3. FIG. 10 is a flowchart of a reference data forming process according to the third embodiment. The flowchart of creation of enlarged mask design data. The top view which shows the mask design pattern based on mask design data. Process sectional drawing which shows a contact hole formation process. The top view which shows the mask design pattern based on mask design data. The top view which shows the mask design pattern based on mask design data. The flowchart of preparation of the expansion mask design data after correction | amendment. The top view which shows the mask design pattern based on mask design data. The top view which shows the mask design pattern based on mask design data. The top view which shows the pattern on the design based on LSI design data, and the pattern obtained on an actual wafer. The flowchart of a mask defect inspection.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 ... Mask design pattern 11 ... Mask design pattern 12 after correction | amendment ... Area | region 21 including the location which does not maintain predetermined | prescribed figure distance S1 in which mask defect inspection is possible ... Mask 22 ... Optical sensor circuit 23 ... Sensor data 24 ... Mask after correction | amendment Design data 25 ... reference data generation circuit 26 ... reference data 27 ... comparison circuit 28 ... defect extraction

Claims (3)

Reference for obtaining mask defect inspection reference data from corrected mask design data obtained by performing correction to improve the fidelity of a pattern formed on a sample with respect to mask design data obtained based on device design data A data formation process;
A sensor data forming step for obtaining sensor data obtained by actually measuring the shape of the mask created based on the corrected mask design data;
In the mask defect inspection method comprising the defect detection step of comparing the reference data and the sensor data and detecting the presence or absence of defects on the mask,
The reference data forming step includes a step of extracting a region including a gap between each graphic on the corrected mask design data or a region where the size of each graphic is smaller than a predetermined value set in advance;
And a step of creating the reference data for the extracted region using a method of simulating a mask shape with higher accuracy than other regions. Reference for obtaining mask defect inspection reference data from corrected mask design data obtained by performing correction to improve the fidelity of a pattern formed on a sample with respect to mask design data obtained based on device design data A data formation process;
A sensor data forming step for obtaining sensor data obtained by actually measuring the shape of the mask created based on the corrected mask design data;
In the mask defect inspection method comprising the defect detection step of comparing the reference data and the sensor data and detecting the presence or absence of defects on the mask,
The reference data forming step includes a step of extracting a region including a gap between each graphic on the corrected mask design data or a region where the size of each graphic is smaller than a predetermined value set in advance;
And a step of creating the reference data only for regions other than the extracted region by excluding the extracted region from a mask defect inspection target. Reference for obtaining mask defect inspection reference data from corrected mask design data obtained by performing correction to improve the fidelity of a pattern formed on a sample with respect to mask design data obtained based on device design data A data formation process;
A sensor data forming step for obtaining sensor data obtained by actually measuring the shape of the mask created based on the corrected mask design data;
In the mask defect inspection method comprising the defect detection step of comparing the reference data and the sensor data and detecting the presence or absence of defects on the mask,
The reference data forming step is a step of extracting a region including a portion where a gap between each figure on the corrected mask design data is smaller than a predetermined value S1 that can be inspected for a mask defect ;
The corrected mask design data, and obtaining the auxiliary pattern adding data obtained by adding data of the portion in the auxiliary pattern between FIG shape adjacent the position of the extracted region,
And a step of creating the reference data from the auxiliary pattern additional data.

Images