JP4710191B2 - Photomask pattern shape automatic measurement method and apparatus, program therefor, photomask manufacturing method and semiconductor device manufacturing method - Google Patents

Description

[0001]
[Industrial application fields]
The present invention relates to a photomask pattern shape automatic measurement method and a photomask pattern shape automatic measurement program for automatically classifying a desired pattern portion of a mask pattern image of a photomask used in a lithography process of semiconductor manufacturing and automatically measuring the pattern shape. The present invention also relates to a photomask pattern shape automatic measurement apparatus, a method for manufacturing a photomask using the apparatus, and a method for manufacturing a semiconductor device using the photomask.
[0002]
[Prior art]
With the recent miniaturization of semiconductor LSI patterns, photomasks 4 as pattern masters are also required to cope with miniaturization, and at the same time, the demand for higher accuracy is extremely severe. Conventionally, three items of defects, dimensional accuracy, and alignment have been particularly emphasized as important items in the quality of the photomask 4, and the shape characteristic value 10 of the mask pattern 1 of each item is measured at present when semiconductor miniaturization advances. For this purpose, a highly accurate inspection apparatus dedicated to the photomask 4 has been developed and used. However, the demand for higher precision by miniaturization of the photomask pattern 1 is becoming the same in all quality items (pattern shape, pattern data guarantee, durability, cleanliness, etc.) other than the above three items. The accuracy of the pattern shape characteristic value has been considerably emphasized because it directly relates to the accuracy and performance of the LSI circuit.
[0003]
As for the pattern shape 7 of the photomask 4, it is naturally desirable that the pattern 8 according to the design drawing is accurately reproduced on the mask 4 in the mask layout design 8 of the semiconductor circuit 16. However, since a fine pattern is actually processed on a metal thin film on glass using a lithography technique, the mask pattern 1 and the design patterns 8 and 9 are not completely the same shape. There are minor differences 18 such as roundness. This difference is almost on the order of several tens to several hundreds of nanometers on the mask 4, but this may affect the characteristics of the semiconductor circuit 16 due to the recent progress in miniaturization of VLSI. It is beginning to be a concern. That is, the finer the pattern, the larger the pattern shape difference 18 with respect to the pattern 1 itself, and the more it affects the characteristic value 17 of the semiconductor device.
[0004]
In addition, with the recent rapid miniaturization, attempts to improve the pattern shape problem by actively utilizing the optical principle in the projection exposure technique have become active. A typical example is an optical proximity correction mask (hereinafter referred to as an OPC mask). Here, the OPC mask will be described. The OPC mask is added with a fine optical proximity correction pattern (hereinafter referred to as an OPC pattern) so as to be close to or in contact with the original circuit pattern 16 so that the shape of the circuit pattern 16 can be accurately transferred during wafer exposure transfer. It is a mask. The OPC pattern corrects the shape of the transferred pattern due to the optical proximity effect at the time of projection exposure transfer using the optical interference effect between adjacent patterns, and the original design pattern 8 is accurately obtained. The pattern is intended to be transferred to the semiconductor device substrate 19 and is often arranged at the four corners of the original circuit pattern 16 or at a portion closest to the adjacent pattern. Recently, an OPC pattern of a type that can deform the entire circuit pattern 16 in a complicated manner has also been proposed. However, since the original circuit pattern 16 is unnecessary, the OPC pattern itself must be fine enough not to be transferred. Therefore, since the OPC pattern is considerably finer than the conventional pattern, the dimensional rule of the mask pattern 1 is remarkably finer than that of the conventional mask, and very fine processing in terms of mask manufacturing technology. Requires technology. Of course, in terms of miniaturization, it is certain that the conventional photomask 4 will progress in the same manner, and advanced microfabrication techniques are still required. Therefore, the problem of accuracy of the pattern shape characteristic value 10 has come to be emphasized as a problem of photomask manufacturing and inspection technology.
[0005]
When the pattern shape 7 is represented as an inspection item on the mask 4 quality, there are various items. For example, pattern corner roundness (= corner round), straight line pattern edge roughness (= edge roughness), pattern misalignment during drawing (= batting error), shape distortion, taper shape, etc. There are items to check. The items in parentheses are item names that are usually used in the photomask inspection process. More recently, there are cases where roundness of the shape of the OPC pattern or the like becomes a problem when transferring a fine pattern onto the wafer 19 with high accuracy, and it is necessary to consider the rectangularity of the pattern shape.
[0006]
As a means for confirming the accuracy of these pattern shapes 7, conventionally, the shape 7 of the mask pattern 1 is observed using an optical microscope or SEM, and the edges of the pattern 1 are notched or the corners of the pattern 1 are rounded. It was determined by visual inspection by the inspector whether there was any abnormality in the item. As another means, the observation image 15 of the mask pattern 1 obtained by the above-described optical microscope or SEM is read as image data into a computer, and then the pattern 1 shape 7 and the design pattern 8 are compared on the computer screen. Check the accuracy of the shape by measuring the size and angle of the pattern corner, the pattern width, etc. with the mouse using image processing software tools etc. after judging the pattern from the density of the image data It was.
[0007]
[Problems to be solved by the invention]
In the means for confirming the accuracy of the pattern shape 7 as described above, in an optical microscope (including a pattern observation apparatus for the purpose of observation at a high magnification, such as a laser microscope or a confocal microscope as well) or SEM, Since the determination is based on the subjectivity of the inspector, there is a possibility that the inspector may make a different determination. Further, when compared with the design pattern 8 on the screen 1 of the computer, there is a similar problem because of the subjective determination by the inspector. The fundamental problem of the pattern 1 shape 7 confirmation method is that numerical expression for objective evaluation is impossible, that is, the shape characteristic value 10 of the pattern 1 shape 7 is quantitatively measured and accurately evaluated. It is not possible.
[0008]
In this sense, the conventional means using the image processing software tool can be said to be superior to other means because the shape 7 can be expressed numerically by length, angle, and the like. However, in the above-mentioned means, since the contour of the pattern is determined by the contrast of the image data 7 and the measurement is performed manually with the mouse, the measurement point when measuring the size of the corner of the pattern 1, the pattern width, and the like is the measurer. There was a problem that it could not be measured accurately. Further, the characteristic value 10 of the pattern 1 shape 7 can only be measured by the above-mentioned means, but the length and angle can be measured directly, and the curve shape cannot be measured. Further, since the characteristic value of the pattern 1 shape 7 is not limited to one place for one pattern 2, there is a problem that it is troublesome to manually measure a number of patterns.
[0009]
As described above, the conventional means is insufficient to automatically and accurately evaluate the characteristic value 10 of the pattern 1 shape 7. The present invention has been made in view of the above problems, and automatically classifies the pattern 1 shape of the photomask 4 including a fine pattern 12 and automatically and quickly and accurately measures each classified shape 14. It is an object to provide methods and software and apparatus that make it possible to do so.
[0010]
[Means for solving problems]
In order to solve the above-mentioned problem, in the method of automatically measuring the shape of the photomask pattern according to the invention of claim 1, in the method of automatically measuring the shape of the photomask pattern, (a) measure Obtaining a mask pattern image; and (b) Said Cutting out an arbitrary pattern portion from the mask pattern image, and (c) obtaining a normal pattern portion image from design data at the same position as the pattern portion or a normal mask pattern, and (d) Said (E) extracting a contour of a normal pattern partial image; Of the contour Generating vector data from the contour line; (f) Said From vector data Determine a reference vector and use the relative angle between the reference vector and each vector in the vector data. A step of converting to a pattern signal; (g) Said A pattern shape recognition process for determining a pattern shape type from a pattern signal; and (h) Said The process of measuring the shape characteristic value of the pattern part for each type of shape determined by pattern shape recognition, and performing the process consisting of automatic classification and automatic measurement of the shape characteristic value for the desired pattern part of the mask pattern image It is a photomask pattern shape automatic measurement method characterized by enabling.
[0011]
An apparatus for automatically measuring the shape of a photomask pattern according to the invention of claim 2, wherein (a) measure Means for obtaining a mask pattern image for obtaining an image of the mask pattern; and (b) Said Processing means for cutting out an arbitrary pattern portion from the mask pattern image; and (c) At the same position as the pattern Means for acquiring and processing a normal pattern portion image for acquiring a normal pattern portion image from design data and a normal mask pattern; and (d) Said Means for extracting a contour of a normal pattern partial image; and (e) Of the contour Means for generating vector data from contour lines, and (f) Said Means for generating a pattern signal from vector data, and (g) Said Means for recognizing and processing a pattern shape for determining the type of pattern shape from the pattern signal; (h) Said An automatic photomask pattern shape measuring apparatus comprising: means for measuring and processing a shape value for measuring a characteristic value of a pattern portion for each type of shape determined by pattern shape recognition processing.
[0012]
A program for automatically measuring the shape of a photomask pattern according to the invention of claim 2, wherein: (a) measure A mask pattern image acquisition processing program for acquiring an image of a mask pattern; and (b) Said A pattern part cutout program for cutting out an arbitrary pattern part from a mask pattern image; and (c) At the same position as the pattern A normal pattern portion image acquisition processing program for acquiring an image of a normal pattern portion from design data and a normal mask pattern; (d) a contour extraction processing program for extracting the contour of a normal pattern portion image;
(E) Of the contour A vectorization processing program for generating vector data from an outline; (f) Said From vector data Determine a reference vector and use the relative angle between the reference vector and each vector in the vector data. A pattern signal processing program for generating a pattern signal; (g) Said Pattern shape recognition program processing for determining the type of pattern shape from the pattern signal; (h) Said A shape measurement program that measures the characteristic value of the pattern part for each type of shape determined by the pattern shape recognition process, and performs automatic classification and automatic measurement of the shape characteristic value for the desired pattern part of the mask pattern image A photomask pattern shape automatic measurement program characterized by:
[0013]
The invention according to claim 4 is an automatic photomask pattern shape measuring apparatus equipped with the photomask pattern shape automatic measuring program according to claim 3.
[0014]
The invention according to claim 5 described above manufactures a photomask by measuring a photomask pattern shape using the method according to claim 1 or the apparatus according to claim 2 or the apparatus according to claim 4. This is a method for manufacturing a photomask.
[0015]
The invention according to claim 6 is a method for manufacturing a semiconductor device, wherein a semiconductor device is manufactured using the photomask manufactured by the method for manufacturing a photomask according to claim 5.
[0016]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, the contents of the present invention will be described in detail with reference to the accompanying drawings. FIG. 1 is a flowchart of a photomask pattern shape automatic measurement method. A photomask pattern shape automatic measurement method will be described with reference to FIG.
[0017]
S1 is a means for acquiring the mask pattern 1 image 15 to be measured, and an inspection machine, SEM, optical microscope or the like is used. The acquired mask pattern 1 image is taken into a computer as data of a bitmap 11 image.
[0018]
S2 is a means for cutting out the desired pattern portion 2 from the acquired mask pattern 1 image, and cuts out the arbitrarily selected pattern portion 2 rectangular area using a pointing device such as a mouse.
[0019]
S3 is data of the bitmap 11 image of the extracted pattern portion 2. The data of the bitmap 11 image can also be referred to as data obtained by digitizing the shade of the image representing the light transmittance. At this time, the difference between the numerical values of the light-shielding portion and the transmissive portion may be determined as appropriate. Calculated in minutes.
[0020]
S4 is a means for acquiring a normal pattern portion 9 image. The mask pattern 1 (defect pattern image) of S1 is obtained by a method using an inspection machine, an SEM, and an optical microscope in the same manner as the acquisition process, and a design as shown in S5. There is a method of using data 8. When an inspection machine, SEM, or optical microscope is used for acquisition from the mask pattern 1, it can be used and effective only when the same pattern as the pattern portion 2 cut out from the mask pattern 1 exists in the normal state 7. When the design data 8 is used, a process for converting the design data 8 corresponding to the pattern part 2 into a pattern image is performed based on the position coordinates of the cut pattern part and the information on the size of the area.
[0021]
S6 is a means for performing contour extraction processing, which is a process for extracting the contour line 3 of the pattern from the normal pattern portion 9 image, and is performed by applying image processing to the data of the bitmap 11 image of the normal pattern 9. Image processing includes smoothing processing to remove image noise, binarization processing to convert multi-gradation data into binary data consisting of 0 or 1, edge extraction processing to extract the boundary between the pattern and the background, and thickness There is a thinning process that shrinks an image of a line having a width of 1, that is, shrinks to the minimum unit of image data, and by combining these processes as appropriate, an outline that connects the boundary between the light shielding part and the transmission part of the pattern Line 3 can be extracted.
[0022]
In S 7, vector data conversion processing is performed for converting the contour 3 of the normal pattern portion 9 into vector data 5. From the start point (start point) 23 of the contour line 3, vectors having the same magnitude component are arranged in the direction along the contour line, and the vector 5 having the amount and the direction is connected, and the end point of the contour line 3 Vector data processing is performed up to 24. The vector direction component data of these vectors is referred to as vector data 5.
[0023]
In S8, pattern signal processing is performed to convert the vector data into numerical data of a one-dimensional array. The pattern signal 6 is generated based on the vector data 5. Here, the vector at the starting point 23 of the contour line is set as the reference vector 21, Cos θ is calculated with respect to the relative angle θ of each vector 22 with respect to the reference vector 21, and this value is used as the data of the pattern signal 6.
[0024]
In S9, a pattern shape 14 recognition process for specifying the shape type 12 of the normal pattern portion 9 is performed. The shape recognition 14 is performed using data obtained by differentiating the pattern signal 6 to obtain an absolute value.
[0025]
In S10, a shape measurement process is performed on the partial image 2 of the pattern 1 for each pattern 9 shape classified by the pattern shape recognition process 14. By the shape measurement process, automatic measurement is performed individually for each type of pattern shape 14.
[0026]
In S11, the data of the shape characteristic value that is the measurement result is output.
[0027]
Next, an example of the implementation will be described. (1) to (5) in FIG. 2 show the procedure of data processing, and as shown in (1) in FIG. 2, the design data is composed of a collection of coordinate data of the vertices of the rectangular figure. Therefore, using a conversion algorithm such as that used in a die-to-database inspection machine, a normal pattern portion 9 image is obtained by converting into a bitmap 11 image shown in FIG.
[0028]
FIG. 2 (3) shows the result of extracting the contour line 3 from the normal pattern image 9.
[0029]
FIG. 2 (4) shows the result of data converted from the contour 3 into vector data 5. FIG.
[0030]
In FIG. 2 (5), the relative angle θ between the reference vector 21 and each vector 22 is displayed. In FIG. 3 (1), the horizontal axis represents vector data No. For example, the vertical axis represents Cos θ. The graph of the pattern signal at the time of taking is shown. The pattern signal 6 represents the feature of the pattern shape by one-dimensional numerical data. Since each data is obtained from a relative angle with respect to the reference vector 21, each vector 22 is determined for each pattern regardless of the direction of the pattern. Determined uniquely. The data of the pattern signal 6 is not limited to Cos θ, and Sin θ, θ itself, or other numerical values generated from the vector data 5 may be used depending on the case.
[0031]
(2) in FIG. 3 shows data obtained by differentiating the pattern signal 6 in (1) in FIG. 3 to obtain an absolute value.
[0032]
Since the number of peaks in the data of (2) in FIG. 3 is 6, this pattern can be recognized as a hammer pattern.
[0033]
Here, as shown in FIGS. 4 to 6, the types of shapes classified according to the data processing procedure are as follows: pattern edge, pattern corner, line end, hole pattern, serif pattern, hammer There are 6 types of heads.
[0034]
Next, FIGS. 4 to 6 show the pattern signal 6 of each of the six types of shapes and data obtained by differentiating the pattern signal 6. 4 to 6, it can be seen that the number of peaks varies depending on the type of shape, and the number of peaks is the same as the number of corners of the pattern. That is, by counting the number of peaks, the pattern shape type 12 can be recognized and classified. For example, FIG. 5A is a graph of the pattern signal 6 when (a) represents the type of pattern by the data figure, the line end, and (b), the relative angle θ of each vector 22 is Cos θ. (C) shows data obtained by differentiating the pattern signal 6 to obtain an absolute value. The number of peaks is shown by adding (1), (2), and (3) to the peak.
[0035]
The shape measurement processing shown in FIG. 7 is performed according to the procedure of Japanese Patent Application No. 11-367669, a “mask pattern shape measurement method” invented and applied by the author. FIG. 7 shows a flow of the processing procedure. SS1 in FIG. 7 is a process for extracting the contour line 3 of the extracted pattern part 9 image, and is performed by performing image processing on the pattern part image in (1) of FIG. The image processing of the procedure for extracting the contour line 3 of the pattern portion 9 image includes a smoothing process for removing image noise and a binarization process for converting multi-gradation data into binary data consisting of 0 or 1. There are edge extraction processing that extracts the boundary between the pattern and the background portion, thinning processing that shrinks the image of a line having a thickness until the width becomes 1, that is, the minimum unit of image data, etc., and these processing are combined as appropriate Thus, it is possible to extract the outline 3 connecting the boundary between the light shielding part and the transmission part of the pattern.
[0036]
(2) of FIG. 7 is an output result of data obtained by extracting the contour line 3 by the procedure of extracting the contour line 3 from the pattern portion 9 image.
[0037]
Next, a vector data conversion process SS2 for converting the contour 3 of the pattern 9 into vector data is performed. Vector vectors having the same magnitude component are sequentially arranged from the starting point 23 of the contour line 3 in the direction along the contour line, and the vector 5 having the quantity and the direction is connected to the end point 24 of the contour line. Process. Data of the direction component of these vectors is set as vector data.
[0038]
FIG. 7 (3) shows the result of conversion from the contour line 3 data to the vector data 5. FIG.
[0039]
Next, a concatenation / synthesis process for converting the vector data 5 into the line figure data 20 is performed. This is a process of calculating the coordinate value of each pixel of the pattern contour 3 by sequentially adding the coordinates of the starting point 23 of the first vector data and the individual x and y components of the vector data. The coordinate value data of each pixel calculated in this way is arranged in order along the contour of the pattern and constitutes a line figure, and hence is called line figure data 20. The converted line figure data 20 is shown in (4) of FIG.
[0040]
Finally, a coordinate measurement process SS4 for measuring the characteristic value of the pattern shape is performed using the line figure data 20. The algorithm of the coordinate measurement process SS4 differs for each pattern shape type 12, and different shape characteristic values 10 can be measured according to the pattern shape 9. Details of the pattern shape type 12 and the measured shape characteristic value 10 will be described below. In addition to the following, the pattern shape characteristic value 10 can be variously considered depending on the pattern characteristics. Based on the line figure data 20 obtained here, the algorithm of the measurement process is changed or added. Thus, a desired shape characteristic value 10 can be obtained.
[0041]
Next, FIGS. 8 to 13 will be described in order with respect to the six types of pattern portions 14 according to type.
[0042]
A pattern edge measuring method shown in FIG. 8 will be described. The pattern edge measures the roughness of the edge. The characteristic values to be quantified are the average roughness and roughness width: a shown in FIG. Statistical calculation is indispensable for obtaining the average roughness, and it has been difficult to calculate the average roughness of the pattern edge by the conventional image processing method. However, in the present invention, line figure data 20 as shown in FIG. 8B is generated and calculation is performed based on the xy coordinate values, which is possible. Specifically, the average line of the edge is obtained based on the coordinate value of the line figure data 20 between the start point 23 and the start point 24 of the edge portion of the pattern, and the absolute value of the difference f (x) between the average line and each coordinate value. Is divided by the distance from the end point 24 to the start point 23. The roughness width a may be obtained by examining the coordinate values of the line figure data 20 in order of magnitude and determining the difference between the minimum value and the maximum value in the width direction.
[0043]
The pattern corner measurement method shown in FIG. 9 will be described. As the shape characteristic value 10 for quantifying the pattern corner, as shown in FIG. 9A, the horizontal and vertical lengths a and b of the corner portion, the length c of 45 degrees oblique direction, and roundness are used. There are an area s of the portion lacking from the original pattern, an erosion degree s / S representing the degree of chipping depending on a ratio of the area S of the original pattern and the area s of the missing portion. In order to measure these characteristic values, it is necessary to detect the part where the contour line 3 of the pattern shape changes in a curved manner, that is, the start and end of corner bending. As shown in FIG. 9B, at the start and end of corner bending, the slope of the straight line is obtained from the amount of movement of the xy component between two points separated from each other in order from the starting point 23 of the line figure data 20, and the slope is predetermined. It is possible to detect by examining a place where the threshold value is greater than (or less than) the threshold. If noise is present on the contour 3 of the pattern, it may be possible to specify that the noise portion starts to bend at a corner. Therefore, in the present invention, the threshold value can be adjusted according to the image quality of the mask pattern image. It is. The horizontal and vertical lengths a and b of the corner portion can be obtained by calculating the absolute value of the difference between the x and y components between the two points at the corner start and end. The length c of the corner in the diagonal direction can be obtained by calculating the distance from the point where the straight lines extend from the start and end points to the contour line in the direction of 45 degrees diagonally. Further, the area s of the missing portion can be obtained by subtracting the area of the corner portion from a rectangle (= original pattern) having two sides a and b. The area of the corner portion is calculated by summing all the numbers of pixels for each column while tracing the line figure data 20 from the start point 23 to the end point 24. For the degree of erosion representing the degree of corner chipping, the ratio of the area S of the original pattern to the area s of the chipped portion, that is, s / S may be simply calculated.
[0044]
The line end measurement method shown in FIG. 10 will be described. Since there are two corners at the end of the line, as shown in FIG. 10A as characteristic values for quantification, the horizontal and vertical lengths a1, a2, a1, b1, b2, lengths c1 and c2 in a 45-degree oblique direction, areas s1 and s2 of a portion missing from the original pattern, and line width d. Items other than d in the characteristic value can be measured in the same manner as the pattern corner measurement method. The method for detecting the start and end of corner bending is the same, as shown in FIG. 10B, in order from the starting point 23 of the line figure data 20, the xy component movement amount between two points separated by a certain number, and the slope of the straight line. This is done by examining where the slope is above (or below) a predetermined threshold. The line width d is obtained by measuring the absolute value of the difference between the x or y coordinates (in the figure, the x coordinate) between the start point 23 and the end point 24.
[0045]
The hole pattern measurement method shown in FIG. 11 will be described. As shown in FIG. 11A, the characteristic values for quantifying the hole pattern are the horizontal and vertical diameters a and b, the hole area s, the indentation length of the four corners (length in a 45-degree oblique direction). L1-4, and fidelity s / S representing the degree of pattern completion depending on the ratio of the area S of the original pattern and the area s of the actual hole pattern. The horizontal diameter of the hole pattern is obtained by calculating the difference between the maximum value and the minimum value in the x direction of the line figure data, and the vertical diameter is obtained by calculating the difference between the maximum value and the minimum value in the Y direction. Further, the area is obtained by calculating the number of pixels included in the line figure data. The indentation lengths L1 to L4 of the four corners generate a circumscribed rectangle (= original pattern) of the hole pattern from the maximum and minimum points of the line figure data 20 as shown in FIG. It is obtained by calculating the distance from the contour line of the hole pattern. The fidelity representing the degree of pattern completion may be calculated by calculating the ratio of the original pattern area S (a × b) to the hole pattern area s, that is, s / S.
[0046]
A method for measuring the serif pattern shown in FIG. 12 will be described. As the characteristic values for quantifying the serif pattern, as shown in FIG. 12A, the horizontal diameter and the vertical diameter a, b, the slant diameter c, the bulge horizontal and vertical lengths d and e. , The area S of the serif pattern, the indentation length (length in a 45-degree direction) L1 to L3 of each corner portion, and the like. The characteristic value of the serif pattern is measured based on the predicted pattern shown in FIG. This predicted pattern is a predicted pattern predicted from the shape of the serif pattern. At the intersection of the serif pattern and the base pattern, the angle between the serif pattern side and the base pattern side is a right angle in the design data. Therefore, by detecting the point where the slope of the straight line between two points that are a certain distance apart from the line figure data is the maximum with respect to the side of the base pattern (the point closest to the right angle) as the reference point, Patterns can be predicted. A circumscribed rectangle of the serif pattern that passes through these two reference points is taken as a predicted pattern. The horizontal diameter a and vertical diameter b of the serif pattern are obtained by calculating the horizontal diameter and vertical diameter of the predicted pattern, respectively. The length d in the horizontal direction of the bulge is calculated by calculating the distance between the vertical side of the base pattern and the vertical side of the predicted pattern, and the length e in the vertical direction of the bulge is the distance between the horizontal side of the base pattern and the horizontal side of the predicted pattern. It is calculated by calculating. The indentation lengths L1 to L3 of the three corners are obtained by calculating the distance from the vertex of the predicted pattern to the contour line of the serif pattern. The oblique diameter c is obtained by subtracting L1 and L3 from the length of the diagonal line calculated by the square theorem from the vertical and horizontal diameters of the predicted pattern. The line pattern data is traced between two reference points, and the number of pixels for each column is subtracted when progressing in the -x direction, and added when progressing in the + x direction. The area is calculated. Therefore, by subtracting the area [(ab) × (be)] of the overlapping portion from here, the serif pattern area of the hatched portion is obtained.
[0047]
The hammer pattern measuring method shown in FIG. 13 will be described. As shown in FIG. 13 (a), the characteristic values for quantifying the hammer pattern include the horizontal and vertical diameters a and b, the lengths c and d of the protrusions of the hammer, the area S of the hammer, and each corner portion. Indentation length (length in a 45-degree oblique direction) L1-4 and the like. The characteristic value of the hammer pattern is measured based on the predicted pattern shown in FIG. This prediction pattern is a prediction of the original pattern from the shape of the hammer pattern. At the intersection of the hammer pattern and the base pattern, the angle formed by the side of the hammer pattern and the side of the base pattern is a right angle in the design data. Therefore, the original hammer pattern can be detected by detecting the point where the slope of the straight line between two points separated by a certain number of line drawing data is the maximum with respect to the side of the base pattern (closest to the right angle) as the reference point. Can be predicted. When the position coordinates of the two reference points (y coordinates in the case of FIG. 10) are different, the two average positions are calculated and fine-adjusted to be used as a new reference point. A circumscribed rectangle of the hammer pattern that passes through these two reference points is used as a predicted pattern. The horizontal diameter a and vertical diameter b of the hammer pattern can be obtained by calculating the horizontal diameter and vertical diameter of the predicted pattern, respectively. The lengths c and d of the hammer projections are obtained by calculating the distance between the side of the base pattern and the side of the predicted pattern at both ends. The indentation lengths L1 to L4 of the four corners are obtained by calculating the distance from the vertex of the predicted pattern to the contour line of the hammer pattern. The area S of the hammer pattern is obtained by tracing the line figure data between two reference points, subtracting the number of pixels for each column in the -x direction, and adding the number in the + x direction. It is done.
[0048]
As for the pattern shape 7 of the photomask 4, it is naturally desirable that the pattern 8 according to the design drawing is accurately reproduced on the mask 4 in the mask layout design 8 of the semiconductor circuit 16. However, since a fine pattern is actually processed on a metal thin film on glass using a lithography technique, the mask pattern 1 and the design patterns 8 and 9 are not completely the same shape. There are minor differences 18 such as roundness. The problem is that the method and the processing procedure of the present invention are used to perform inspection using the image processing software tool of the present invention on a system equipped with the hardware environment required by the software of the present invention. In addition, by providing and developing automatic measurement equipment for pattern shapes, the minute differences 18 were inspected and automatically measured.
[0049]
Next, the demand for high accuracy by miniaturization of the photomask pattern 1 is that all quality items (pattern shape, pattern data guarantee) other than the items of dimensional accuracy of defect patterns and overlay accuracy between photomasks , Durability, cleanliness, etc.) are becoming demands of strict quality standards, and in particular, the accuracy of mask pattern shape characteristic values is directly related to the accuracy and performance of LSI circuits (semiconductor circuits), so it will be emphasized. It has become. At the time of manufacturing a semiconductor device, a semiconductor circuit formed on the semiconductor device substrate 19 is formed by applying a photosensitive resist on the semiconductor device substrate 19 and a circuit pattern formed on a photomask using optical process means. After the exposure transfer and the development of the resist layer by the development process, a thin film conductive film formation, a pattern layer formation by corrosion, or a process process such as ion irradiation using the resist pattern as a mask layer is sequentially repeated to form a hierarchical structure. In order to form a circuit, it is common to perform a process several times. In addition, a plurality of photomasks are used to form the semiconductor circuit. In the above process, the pattern of the semiconductor circuit is greatly dependent on the accuracy of the photomask, so that the photomask manufactured according to the present invention is used. The semiconductor device was manufactured and the accuracy of exposure transfer was verified.
[0050]
【The invention's effect】
As described above, the photomask pattern automatic measurement method of the present invention is used to automatically classify a desired pattern portion in the measurement of the photomask pattern shape, and automatically and quickly measure the shape for each pattern shape. In addition, since various calculation methods such as statistical methods can be used, it becomes possible to accurately quantify the shape characteristics, and a high-precision photomask is manufactured using the apparatus of the present invention. High-precision products can be supplied even in the manufacture of semiconductor devices using the photomask, and the shape characteristics can be obtained simply by changing the coordinate measurement processing algorithm using the same shape data for other photomask pattern shapes. Automatic shape measurement for quantifying can be performed.
[Brief description of the drawings]
FIG. 1 is a flowchart showing steps S1 to S11 of an embodiment of a photomask pattern shape automatic measuring method of the present invention.
FIG. 2 is a schematic diagram showing steps S4 to S8 for converting a normal pattern partial image into a contour image, vector data, and a pattern signal, where (1) is design data, (2) is a normal mask pattern, and (3) Is the contour line, (4) is the vector data, and (5) is the vector data relative angle θ.
FIG. 3 is a graph showing a procedure subsequent to FIG. 2, wherein (1) is a pattern signal graph and (2) is differentiated data.
FIG. 4 is an example of six types of pattern shapes and pattern signals, (1) is a pattern edge (a) is a normal mask pattern, and (b) is a Cosθ Graph (c) is an absolute value graph of the integral value. (2) Pattern corner
FIG. 5 shows an example of six types of pattern shapes and pattern signals. (1) is a line end (a) is a normal mask pattern, (b) is a graph of Cos θ, and (c) is a graph. It is an absolute value graph of an integral value. (2) A hole pattern. (A) (1) and (2) represent numbers.
6 is an example of six types of pattern shapes and pattern signals, (1) is a serif pattern (a) is a normal mask pattern, (b) is a graph of Cos θ, and (c) is a graph. It is an absolute value graph of an integral value. (2) A hammer head.
FIG. 7 is a flowchart showing a procedure S10 of shape measurement processing.
FIG. 8 is a characteristic value of a pattern edge in an embodiment of the six types.
FIG. 9 is a characteristic value of a pattern corner in an embodiment of the six types.
FIG. 10 is a line end characteristic value in one embodiment of the six types.
FIG. 11 shows the characteristic value of the hole pattern in one of the six types.
FIG. 12 is a characteristic value of a serif pattern in one embodiment of the six types.
FIG. 13 is a characteristic value of a hammer pattern in one of the six types.
[Explanation of symbols]
S1 ... Mask pattern image acquisition process
S2 ... Pattern part cut-out process
S3 ... Pattern partial image
S4: Normal pattern partial image acquisition processing
S5: Design data
S6: contour extraction processing
S7: Vector data processing
S8: Pattern signal processing
S9: Pattern shape recognition processing
S10 ... Shape measurement processing
S11: Shape characteristic value data
SS1 ... contour extraction processing
SS2 ... Vector data processing
SS3: Concatenation / synthesis processing
SS4: Coordinate measurement process
1 ... (Photo) Mask pattern
2 ... Arbitrary pattern part
3 ... (image) outline (extraction)
4 ... Photomask
5 ... Vector data
6 ... Pattern signal
7 ... Photomask pattern shape
8 ... Design data
9: Normal mask pattern (mask data)
10 ... Shape characteristic value
11 ... Bitmap image
12 ... Types of pattern shapes
13: Normal pattern part
14 ... Pattern section by type
15 ... Image
16 ... Semiconductor circuit
17: Characteristic value of semiconductor device
18: Difference in pattern shape (difference in shape between normal pattern 9 and mask pattern 1)
19 ... Substrate for semiconductor device
20 ... Line figure data
21 ... Reference vector
22: Each vector
23 ... Starting point
24 ... End point

Claims (6)

In the method of automatically measuring the shape of the photomask pattern,
(A) obtaining an image of a mask pattern to be measured ;
A step of cutting an arbitrary pattern portion from the (b) the mask pattern image,
(C) obtaining a normal pattern portion image from design data at the same position as the pattern portion or a normal mask pattern;
And (d) a step of extracting the contour of the normal pattern portion image,
(E) generating vector data from the contour line of the contour;
(F) determining a reference vector from the vector data , and performing conversion into a pattern signal using a relative angle between the reference vector and each vector in the vector data ;
(G) a step of the pattern recognition to determine the type of pattern from the pattern signal,
A step of measuring the shape characteristic value of the pattern portion (h) for each type of shape that is determined by the pattern shape recognition,
A photomask pattern shape automatic measurement method, which enables automatic classification and automatic measurement of shape characteristic values for a desired pattern portion of a mask pattern image by executing a process comprising: In an apparatus that automatically measures the shape of a photomask pattern,
(A) means for the mask pattern image acquisition processing to acquire the image of a mask pattern to be measured,
Processing means for pattern portion cut out to cut out any pattern portion from (b) wherein the mask pattern image,
(C) means for acquiring and processing a normal pattern portion image for acquiring an image of a normal pattern portion from design data and a normal mask pattern at the same position as the pattern portion;
(D) means for outline extraction process extracts the outline of the normal pattern portion image,
(E) means for generating vector data from the contour line of the contour;
(F) means for processing the pattern signal of generating a pattern signal from the vector data,
(G) the pattern signal by the pattern shape recognition to determine the type of pattern from, means for processing,
(H) means for the pattern recognition by shape measuring measures a characteristic value of each type in the pattern portion of the shape determined by processing,
An automatic photomask pattern shape measuring device comprising: In a program that automatically measures the shape of a photomask pattern,
(A) a mask pattern image acquisition processing program for acquiring an image of a mask pattern to be measured ;
A pattern portion cutout program for cutting out an arbitrary pattern portion from the (b) the mask pattern image,
(C) a normal pattern portion image acquisition processing program for acquiring images of the normal pattern portion from the design data and the normal mask pattern at the same position as the pattern portion,
(D) a contour extraction processing program for extracting a contour of a normal pattern partial image;
(E) a vector data processing program for generating vector data from the contour line of the contour;
Determining a reference vector from (f) said vector data, and the pattern signal processing program for generating a pattern signal using the relative angle of each vector between the reference vector said in the vector data,
(G) a pattern recognition program processing to determine the type of pattern from the pattern signal,
(H) the shape measuring program for measuring the characteristic values of the pattern shape recognition for each kind in the pattern portion of the shape determined by processing,
A photomask pattern shape automatic measurement program characterized in that automatic classification and shape characteristic value automatic measurement are executed for a desired pattern portion of a mask pattern image.   A photomask pattern shape automatic measurement apparatus, comprising the photomask pattern shape automatic measurement program according to claim 3.   A method of manufacturing a photomask, comprising: measuring a photomask pattern shape using the method according to claim 1, the apparatus according to claim 2, or the apparatus according to claim 4.   6. A semiconductor device manufacturing method, wherein a semiconductor device is manufactured using the photomask manufactured by the photomask manufacturing method according to claim 5.

Images