JP3900296B2 - Pattern forming method and integrated circuit manufacturing method - Google Patents

Description

  The present invention relates to a pattern forming method for correcting a light proximity effect of design data used for manufacturing an integrated circuit or the like to form a pattern. The present invention also relates to a method of manufacturing an integrated circuit using such a pattern forming method.

  A conventional LSI manufacturing process will be described with reference to FIG. First, LSI design data as shown in FIG. 29A is created using CAD or the like. Here, a plurality of rectangular patterns 291 are formed as design data. Next, an electron beam (EB) is drawn based on the design data, thereby forming a mask having a plurality of patterns 292 as shown in FIG. The mask pattern 292 is transferred onto the wafer by batch exposure of the mask with ultraviolet rays (UV light). At this time, as shown in FIG. 29C, the transfer pattern 293 on the wafer is deformed from the rectangular mask pattern 292 due to the light diffraction effect, and becomes a shape with rounded corners. When processing such as etching is performed on the wafer based on the transfer pattern 293, the processed pattern 294 shown in FIG. 29D is further deformed by the microloading effect. Thereafter, when an oxidation step used for LOCOS separation, for example, is performed, the deformation of the pattern shape is further accumulated by the so-called bird's beak effect.

  As described above, in the manufacture of an integrated circuit such as an LSI, pattern deformation is accumulated through a number of manufacturing processes, so that the finished dimensions of the pattern often do not match the design data.

  With the recent miniaturization of integrated circuit pattern dimensions, more precise dimension control is required, and the electrical characteristics and various margins of the device are not negligible due to the pattern deformation in the manufacturing process as described above. There was a problem of end.

The present invention has been made to solve such problems, and an object of the present invention is to provide a pattern forming method capable of reducing deformation of a pattern due to the influence of an optical proximity effect in an integrated circuit manufacturing process. .
Another object of the present invention is to provide a method of manufacturing an integrated circuit using such a pattern forming method.

According to the pattern forming method of the present invention, design data composed of a plurality of graphic elements constituting a circuit pattern is created, the design data is divided into a plurality of data blocks, and an optical projection image at the time of pattern transfer for each data block. The pattern size transferred to the wafer is predicted based on the projected image, the difference between the predicted pattern size and the pattern size based on the design data is calculated, and the calculated difference is used as the correction amount. By correcting the design data, correction data that corrects the optical proximity effect that occurs when the pattern is transferred to the wafer is created, a mask pattern is formed based on the correction data, and exposure is performed through the mask pattern. transferred, and the wafer was processed in accordance with the transferred mask patterns, when the correction of the design data, a plurality of lines to each side of the graphic elements to be corrected Divided into, after correcting for each segment is a method of subdividing points for subdividing a plurality of line segments to remove this division points when located in the middle of the straight line representing the respective sides, to this correction data .

The pattern forming method according to claim 2 determines whether or not the calculated difference is within an allowable range in the method according to claim 1, and when the calculated difference is out of the allowable range, the optical projection is performed again based on the generated correction data. Create an image, predict the size of the pattern transferred to the wafer based on this projection image, calculate the difference between the predicted pattern size and the pattern size based on the design data, further correct the correction data, Create an optical projection image based on new correction data until the difference between the predicted pattern size and the design data pattern size is within the allowable range, predict the size of the pattern transferred to the wafer, and This method repeats further correction of correction data .
A pattern forming method according to a third aspect is the method according to the first or second aspect, wherein the correction data is data having a shape obtained by adding an auxiliary pattern to the pattern based on the design data.

The pattern forming method according to claim 4 is the method according to claim 1 , wherein a buffer area is set around each data block, and a graphic element that produces a proximity effect with the graphic element of the data block exists in the buffer area. This is a method of determining whether or not to determine a side to be corrected in the data block according to the determination result.
The pattern forming method according to claim 5 is the method according to claim 1 , wherein it is checked whether or not the graphic elements in each data block have sides facing each other at intervals of a predetermined value or less. Is a correction target, and if not, the data block is excluded from the correction target.

A pattern forming method according to a sixth aspect is the method according to the first aspect, wherein only a data block having a side of a graphic element is to be corrected, and a data block having no side is not to be corrected.
The pattern forming method according to claim 7 is the method according to claim 1 , wherein when a same graphic element exists in a plurality of data blocks, only one of the data blocks is to be corrected, and the correction result is set as another data block. It is a method to divert.
A pattern forming method according to an eighth aspect is the method according to the first aspect, wherein the division period of the data block is matched with the period of the memory cell with respect to the design data of the memory cell array.
Pattern forming method according to claim 9, in the method of claim 1, the polygon is formed by combining the plurality of graphic elements adjacent to each other in each data block is transferred to the wafer the polygon as the corrected The design data is corrected by calculating the difference between the predicted pattern size and the pattern size based on the design data .
The pattern forming method according to claim 10 is the method according to claim 1, wherein a hollow graphic generated by combining a plurality of graphic elements in each data block is divided into a graphic element defining an outer periphery and a graphic element defining a hollow portion. expressed by a, a graphic element that defines the hollowed portion and the graphic element to define these outer circumference and the correction target to predict the size of the pattern to be transferred onto the wafer, due to dimensional and design data of the predicted pattern This is a method of correcting design data by calculating a difference from a pattern size .

Pattern forming method according to claim 11 is the method of claim 1, when applying the modified illumination method in the exposure apparatus, three-beam interference cutoff frequency below periods corresponding magnitude correction target outside the light shielding pattern of the It is a method to do.
The pattern forming method according to claim 12 is the method according to claim 5 , wherein when the modified illumination method is applied to the exposure apparatus, the interval is equal to or smaller than the predetermined value and equal to or larger than a period corresponding to a cutoff frequency of three-beam interference. In this method, a fixed amount of correction is applied to the sides facing each other.

A pattern forming method according to a thirteenth aspect is the method according to the first aspect, wherein an optical projection image is created from a mask pattern based on design data, and then the optical projection image has a light intensity having a predetermined threshold value. In this method, the mask edge is predicted by the method, and the design data is corrected using the distance between the predicted mask edge and the mask edge of the mask pattern based on the design data as a correction amount .

A pattern forming method according to a fourteenth aspect is the method according to the thirteenth aspect , wherein the threshold value is adjusted in accordance with the light intensity at the periphery of the side of the graphic element to be corrected.
A pattern forming method according to a fifteenth aspect is the method according to the thirteenth aspect , wherein the threshold value is adjusted in accordance with the two-dimensional light intensity distribution in the peripheral portion of the side to be corrected.
A pattern forming method according to a sixteenth aspect is the method according to the thirteenth aspect , wherein the threshold value is adjusted according to the presence / absence of a graphic element close to the side to be corrected.
A pattern forming method according to a seventeenth aspect is the method according to the thirteenth aspect , wherein the threshold value is adjusted according to the light intensity of the optical image at the time of defocusing with respect to the light intensity of the optical image at the time of the best focus.

A pattern forming method according to an eighteenth aspect is the method according to the thirteenth aspect , wherein the threshold value is adjusted according to the reflectance of the base substrate to which the pattern is transferred.
A pattern forming method according to a nineteenth aspect is the method according to the thirteenth aspect , wherein the threshold value is adjusted according to the step formed on the base substrate to which the pattern is transferred.
A pattern forming method according to a twentieth aspect is the method according to the thirteenth aspect , wherein the threshold value is adjusted according to the halation generated on the surface of the base substrate to which the pattern is transferred.

A pattern forming method according to a twenty-first aspect is the method according to the thirteenth aspect , based on a correlation between a resist pattern obtained by actually exposing the standard pattern and an optical image of the standard pattern calculated from the standard mask data. This is a method of adjusting the threshold value.
The pattern forming method according to claim 22 is the method according to claim 13 , wherein an optical image of the standard pattern is calculated from the etching pattern obtained by actually exposing and etching the standard pattern and the standard mask data. The threshold value is adjusted based on the correlation with the etching pattern calculated from the above.

A pattern forming method according to a twenty- third aspect is the method according to the thirteenth aspect , wherein a correlation between a standard pattern actually obtained by electron beam drawing of standard mask data and an electron beam drawing pattern calculated from the standard mask data is obtained. This is a method of adjusting the threshold value based on this.
A pattern forming method according to a twenty-fourth aspect is the method according to the thirteenth aspect , based on a correlation between a standard pattern actually obtained by laser drawing of standard mask data and a laser drawing pattern calculated from the standard mask data. This is a method of adjusting the threshold value.

The pattern forming method according to claim 25 is the method according to claim 13 , wherein a resist pattern obtained by actually exposing a standard pattern and an optical image calculated from a pattern dimension obtained by measuring the standard pattern are provided. This is a method of adjusting the threshold based on the correlation.
A pattern forming method according to a twenty-sixth aspect is the method according to the thirteenth aspect , in which the standard pattern is actually exposed to form a resist pattern, and then the resist pattern is etched and the resist pattern is measured. In this method, the threshold value is adjusted based on the correlation with the etching pattern calculated from the obtained pattern size.

A pattern forming method according to a twenty-seventh aspect is the method according to the thirteenth aspect , wherein a resist pattern obtained by exposing the standard pattern after the standard pattern is formed by actually drawing the standard mask data by electron beam drawing, and the standard mask data. Then, after calculating the electron beam drawing pattern, the threshold value is adjusted based on the correlation with the optical image calculated from the electron beam drawing pattern.
A pattern forming method according to a twenty-eighth aspect is the method according to the thirteenth aspect , wherein a standard pattern is actually laser-drawn to form a standard pattern and then the standard pattern is exposed and then the resist pattern obtained by exposing the standard pattern and the standard mask data. In this method, after the laser drawing pattern is calculated, the threshold value is adjusted based on the correlation with the optical image calculated from the laser drawing pattern.

The pattern forming method of claim 29 is the method of claim 13 , wherein an optical image is calculated from an etching pattern obtained by actually exposing and etching a standard pattern and a pattern size obtained by measuring the standard pattern. Then, the threshold value is adjusted based on the correlation with the etching pattern calculated from the optical image.

The pattern forming method of claim 30 is the method of claim 13 , wherein an etching pattern obtained by exposing and etching the standard pattern after actually forming the standard pattern by drawing the standard mask data with an electron beam, In this method, an electron beam drawing pattern is calculated from the standard mask data, an optical image is calculated from the electron beam drawing pattern, and then the threshold value is adjusted based on the correlation with the etching pattern calculated from the optical image.
A pattern forming method according to a thirty-first aspect is the method according to the thirteenth aspect , wherein an etching pattern obtained by exposing and etching the standard pattern after actually forming the standard pattern by laser drawing the standard mask data, In this method, a laser drawing pattern is calculated from mask data, an optical image is calculated from the laser drawing pattern, and then a threshold value is adjusted based on a correlation with an etching pattern calculated from the optical image.

A pattern forming method according to a thirty-second aspect is the method according to the thirteenth aspect , wherein an optical projection image is created from a mask pattern based on design data, and then a mask edge is set with a predetermined light intensity as a threshold value from the optical projection image. After adjusting the threshold from the tilt of the optical projection image at this mask edge, the mask edge is predicted based on the adjusted threshold, and the size of the pattern transferred based on the predicted mask edge It is a method of predicting.
A pattern forming method according to a thirty-third aspect is the method according to the first aspect, wherein a resist development simulation on the wafer surface is performed and the size of the pattern transferred from the edge position is predicted.
A pattern forming method according to a thirty-fourth aspect is the method according to the first aspect, wherein the optical projection image is converted into a resist development time distribution on the wafer surface, and pseudo development is performed by integrating the development time one-dimensionally. In this method, the size of the transfer pattern is predicted from the edge position.

The pattern forming method according to claim 35 is the method according to claim 1, wherein when creating an optical projection image, i = 1 to 4 and a quadrangle surrounded by these four points from the light intensity Ii of four points Pi (xi, yi). In this method, the light intensity I at an internal point P (x, y) is interpolated by I = Σi (Wi · Ii), Wi = (1- | xi-x |) (1- | yi-y |). .

A pattern forming method according to a thirty- sixth aspect is the method according to the first aspect, wherein when the design data is corrected, the side to be corrected is divided into a plurality of line segments, and each division point is divided into both line segments sharing the division point. This is a method of generating two data correspondingly.
A pattern forming method according to a thirty-seventh aspect is the method according to the first aspect, wherein each side is corrected in a direction perpendicular to the side when the design data is corrected.
A pattern forming method according to a thirty- eighth aspect is the method according to the first aspect, wherein when the design data is corrected, the side is excluded from correction when the inclination of the projected image on the side of the graphic element is equal to or smaller than a predetermined value. .
A pattern forming method according to a thirty-ninth aspect is the method according to the first aspect, wherein when the design data is corrected, when the light intensity of the projected image on the side of the graphic element is equal to or less than a predetermined value, the side is excluded from the correction target. is there.
The pattern forming method according to claim 40 is the method according to claim 1, wherein, when the design data is corrected, an upper limit value of the correction amount is provided, and when the calculated correction amount exceeds the upper limit value, the upper limit value is used as the correction amount. Is the method.
Pattern forming method according to claim 41 is the method of claim 1, when the correction of the design data, after correction for each side of the graphic elements to be corrected, then deletes the redundant point located in the middle of the straight line representing the respective sides This is a method of using this as correction data .

The pattern forming method according to claim 41 is the method according to claim 1, wherein when the design data is corrected, all sides to be corrected are corrected independently of each other, and then it is determined whether or not the calculated difference is within an allowable range. When it is out of the allowable range, a projection image is formed again based on the correction data, and the size of the pattern transferred to the wafer is predicted based on the projection image. The predicted pattern size and the pattern size based on the design data The correction data is further corrected by calculating the difference between the predicted pattern and the optical projection image based on the new correction data until the difference between the predicted pattern size and the design pattern size is within the allowable range. This is a method of repeatedly predicting the size of a pattern transferred to a wafer and further correcting new correction data .
The pattern formation method according to claim 42 is the method according to claim 41 , wherein the correction amount is determined for each data block, and the data block whose correction amount is outside the allowable range is extracted into a separate file and based on the corrected data. This is a method of correcting data again.

An integrated circuit manufacturing method according to a 43rd aspect is a method using the pattern forming method according to any one of the 1st to 42nd aspects.

The pattern forming method according to claim 1 creates circuit pattern design data, divides the design data into a plurality of data blocks, and creates an optical projection image for pattern transfer for each data block. The size of the pattern transferred to the wafer is predicted based on the image, the difference between the predicted pattern size and the pattern size based on the design data is calculated, and the design data is corrected using the calculated difference as a correction amount. Creates correction data that corrects the optical proximity effect that occurs during pattern transfer to the wafer, forms a mask pattern based on the correction data, and transfers the mask pattern onto the wafer by exposing through the mask pattern. the wafer is processed in accordance with the mask pattern was, when the correction of the design data, divides each side of the graphic elements to be corrected to a plurality of line segments, to correct for each segment After subdividing points for subdividing a plurality of line segments to remove this division points when located in the middle of the straight line representing the respective sides, since this correction data, deformation due to the influence of the optical proximity effect is mitigated A pattern can be formed.

The pattern forming method according to claim 2 is a method according to claim 1, wherein it is determined whether or not the calculated difference is within an allowable range. If the calculated difference is out of the allowable range, the pattern is formed again based on the created correction data. Create a projected image, predict the size of the pattern transferred to the wafer based on this projected image, calculate the difference between the predicted pattern size and the pattern size based on the design data, and further correct the correction data The optical projection image is created based on new correction data until the difference between the predicted pattern size and the pattern size based on the design data falls within an allowable range, and the size of the pattern transferred to the wafer is predicted. Further , since further correction of new correction data is repeated, the reliability of correction is improved.
The pattern forming method according to claim 3 is the method according to claim 1 or 2, wherein the correction data is data having a shape in which an auxiliary pattern is added to the pattern based on the design data. It becomes a shape with a sharp pattern.

The pattern forming method according to claim 4 is the method according to claim 1 , wherein a buffer area is set around each data block, and a graphic element that produces a proximity effect with the graphic element of the data block exists in the buffer area. Since the side to be corrected in the data block is determined according to the determination result, the independence of each data block is ensured, and the processing speed by parallel calculation or the like can be increased.

The pattern forming method according to claim 5 is the method according to claim 1, wherein when the graphic elements in each data block do not have sides facing each other at an interval of a predetermined value or less, the data block is excluded from correction. The data is compressed, and the processing speed can be increased.
According to the pattern forming method of claim 6, in the method of claim 1, since the data block having no graphic element side is excluded from the correction target, the data is compressed and the processing speed is increased.
The pattern forming method according to claim 7 is the method according to claim 1 , wherein when a plurality of data blocks have the same graphic element, only one of the data blocks is to be corrected, and the correction result is used as another data block. Therefore, the data is compressed and the processing speed can be increased.
According to the pattern forming method of claim 8, in the method of claim 1, since the data block is divided in accordance with the cycle of the memory cell with respect to the design data of the memory cell array, the data can be compressed and the processing speed is increased. Made.
Pattern forming method according to claim 9, in the method of claim 1, the polygon is formed by combining the plurality of graphic elements adjacent to each other, the size of the pattern to be transferred onto the wafer the polygon as the corrected Since the design data is corrected by calculating the difference between the predicted pattern size and the pattern size based on the design data, redundant edges are removed from the inside of the figure, and the processing speed can be increased.
The pattern forming method according to claim 10 is the method according to claim 1, wherein the hollow graphic is expressed by a graphic element defining an outer periphery thereof and a graphic element defining a hollow portion; graphic element that defines the hollowed portion in the correction target to predict the size of the pattern to be transferred onto the wafer, correct the design data by calculating the difference between the dimensions of the pattern due to dimensional and design data of the predicted pattern Therefore, the data is compressed and the processing speed can be increased.

A pattern forming method according to an eleventh aspect is the method according to the first aspect , wherein when the modified illumination method is applied to the exposure apparatus, a light shielding pattern having a size equal to or less than a period corresponding to the cutoff frequency of the three-beam interference is excluded from the correction target. Therefore, the processing speed can be increased.
A pattern forming method according to a twelfth aspect is the method according to the fifth aspect , wherein when the modified illumination method is applied to the exposure apparatus, the pattern forming method has an interval equal to or smaller than a predetermined value and equal to or larger than a period corresponding to the three-beam interference cutoff frequency. Since a fixed amount of correction is applied to the sides facing each other, the correction work is simplified and the processing speed is increased.

A pattern forming method according to a thirteenth aspect is the method according to the first aspect, wherein an optical projection image is created from a mask pattern based on design data, and then the optical projection image has a light intensity having a predetermined threshold value. since predicting the mask edge, it is possible to predict the size of the transfer pattern easily and fast. Further, in the pattern forming method according to claim 14, in the method according to claim 13, the threshold value is adjusted according to the light intensity at the periphery of the side of the graphic element to be corrected, so that more accurate prediction is possible. It becomes. According to the pattern forming method of claim 15, in the method of claim 13, the threshold value is adjusted according to the two-dimensional light intensity distribution around the correction target side, so that highly accurate pattern prediction is possible. It becomes. According to the pattern forming method of the sixteenth aspect, in the method of the thirteenth aspect, the threshold value is adjusted according to the presence / absence of a graphic element close to the side to be corrected, so that highly accurate prediction is possible. According to the pattern forming method of claim 17, in the method of claim 13, the threshold value is adjusted according to the light intensity of the optical image at the time of defocusing with respect to the light intensity of the optical image at the time of best focus. The focus margin can be expanded.

In the pattern forming method according to the eighteenth aspect , the threshold value is adjusted according to the reflectance of the base substrate to which the pattern is transferred in the method according to the thirteenth aspect , so that highly accurate pattern prediction can be performed. According to the pattern forming method of the nineteenth aspect , in the method of the thirteenth aspect , the threshold value is adjusted in accordance with the step formed on the base substrate to which the pattern is transferred, so that highly accurate prediction is possible. In the pattern forming method according to claim 20, in the method according to claim 13 , the threshold value is adjusted according to the halation generated on the surface of the base substrate to which the pattern is transferred, so that the pattern prediction becomes more accurate.

The pattern forming method according to claim 21, The method of claim 13, based on the correlation between the resist pattern obtained by actual exposure standard pattern, the optical image of the reference pattern calculated from the standard mask data Since the threshold value is adjusted, it is possible to flexibly cope with variations accompanying changes in the resist and process, and high-precision correction is possible. The pattern forming method of claim 22 is the method of claim 13 , wherein an optical image of the standard pattern is calculated from the etching pattern obtained by actually exposing and etching the standard pattern and the standard mask data. Since the threshold value is adjusted on the basis of the correlation with the etching pattern calculated from the above, it is possible to flexibly cope with the variation caused by the etching and to perform highly accurate correction.

A pattern forming method according to a twenty- third aspect is the method according to the thirteenth aspect , wherein a correlation between a standard pattern actually obtained by electron beam drawing of standard mask data and an electron beam drawing pattern calculated from the standard mask data is obtained. Since the threshold value is adjusted based on this, the electron beam proximity effect can be taken into the correction.
A pattern forming method according to a twenty-fourth aspect is the method according to the thirteenth aspect , based on a correlation between a standard pattern actually obtained by laser drawing of standard mask data and a laser drawing pattern calculated from the standard mask data. Since the threshold value is adjusted, the laser drawing proximity effect can be taken into the correction.

The pattern forming method according to claim 25 is the method according to claim 13 , wherein a resist pattern obtained by actually exposing a standard pattern and an optical image calculated from a pattern dimension obtained by measuring the standard pattern are provided. Since the threshold value is adjusted based on the correlation, the optical proximity effect generated at the time of optical transfer can be taken into the correction.
A pattern forming method according to a twenty-sixth aspect is the method according to the thirteenth aspect , in which the standard pattern is actually exposed to form a resist pattern, and then the resist pattern is etched and the resist pattern is measured. Since the threshold value is adjusted based on the correlation with the etching pattern calculated from the obtained pattern size, the microloading effect generated during etching can be taken into the correction.

A pattern forming method according to a twenty-seventh aspect is the method according to the thirteenth aspect , wherein a resist pattern obtained by exposing the standard pattern after the standard pattern is formed by actually drawing the standard mask data by electron beam drawing, and the standard mask data. After calculating the electron beam drawing pattern from the electron beam, the threshold value is adjusted based on the correlation with the optical image calculated from the electron beam drawing pattern, so that the electron beam proximity effect at the time of electron beam drawing and the light at the time of light transfer are adjusted. Proximity effects can be incorporated into corrections.
A pattern forming method according to a twenty-eighth aspect is the method according to the thirteenth aspect , wherein a standard pattern is actually laser-drawn to form a standard pattern and then the standard pattern is exposed and then the resist pattern obtained by exposing the standard pattern and the standard mask data. After calculating the laser drawing pattern, the threshold is adjusted based on the correlation with the optical image calculated from this laser drawing pattern, so the laser drawing proximity effect during laser drawing and the optical proximity effect during light transfer are combined. Can be incorporated into the correction.

The pattern forming method of claim 29 is the method of claim 13 , wherein an optical image is calculated from an etching pattern obtained by actually exposing and etching a standard pattern and a pattern size obtained by measuring the standard pattern. After that, the threshold value is adjusted based on the correlation with the etching pattern calculated from this optical image, so that the optical proximity effect at the time of light transfer and the microloading effect at the time of etching can be combined and incorporated into the correction. .

The pattern forming method of claim 30 is the method of claim 13 , wherein an etching pattern obtained by exposing and etching the standard pattern after actually forming the standard pattern by drawing the standard mask data with an electron beam, Since the electron beam drawing pattern is calculated from the standard mask data and the optical image is calculated from the electron beam drawing pattern, the threshold value is adjusted based on the correlation with the etching pattern calculated from the optical image. The electron beam proximity effect at the time of drawing, the optical proximity effect at the time of light transfer, and the microloading effect at the time of etching can be incorporated in the correction.
A pattern forming method according to a thirty-first aspect is the method according to the thirteenth aspect , wherein an etching pattern obtained by exposing and etching the standard pattern after actually forming the standard pattern by laser drawing the standard mask data, Since the laser drawing pattern is calculated from the mask data and the optical image is calculated from the laser drawing pattern, the threshold value is adjusted based on the correlation with the etching pattern calculated from the optical image. The drawing proximity effect, the optical proximity effect during light transfer, and the microloading effect during etching can be combined and incorporated into correction.

A pattern forming method according to a thirty-second aspect is the method according to the first aspect, wherein an optical projection image is created from a mask pattern based on design data, and then a mask edge is set with a predetermined light intensity as a threshold value from the optical projection image. After adjusting the threshold from the tilt of the optical projection image at this mask edge, the mask edge is predicted based on the adjusted threshold, and the size of the pattern transferred based on the predicted mask edge Therefore, it is possible to predict dimensions with higher accuracy.
The pattern forming method of claim 33 is the method of claim 1, wherein the edge position is predicted by performing development simulation of the resist on the wafer surface, so that it can easily cope with changes in process conditions such as exposure amount and development time. Therefore, prediction with higher accuracy becomes possible.
The pattern forming method of claim 34 is the method of claim 1, wherein the edge position is predicted by performing pseudo development by the development time of the resist on the wafer surface, so that the prediction can be performed with high accuracy and at high speed.

The pattern forming method of claim 35 is the method of claim 1, wherein i = 1 to 4 and the light intensity Ii of four points Pi (xi, yi) is set to a point P (x, y inside the rectangle surrounded by these four points. ) Is interpolated by I = Σi (Wi · Ii) and Wi = (1- | xi-x |) (1- | yi-y |), so that a projection image can be obtained with high accuracy. it can.

A pattern forming method according to a thirty- sixth aspect is the method according to the first aspect, wherein the side to be corrected is divided into a plurality of line segments and two data are generated at each division point, so that the adjacent line segments are not affected. Each line segment can be corrected, and the correction becomes easy.
A pattern forming method according to a thirty-seventh aspect is the method according to the first aspect, wherein the correction direction of the side is a direction perpendicular to the side, so that the oblique side is eliminated, the data is compressed, and the electron beam drawing is performed at a high speed. Optical proximity correction data that can be processed is obtained.
The pattern forming method according to claim 38 is the method according to claim 1, wherein when the inclination of the projected image on the side of the graphic element is equal to or less than a predetermined value, the side is excluded from the correction target. Can be prevented, and the reliability of correction is improved.
The pattern forming method according to claim 39 is the method according to claim 1, wherein when the light intensity of the projected image on the side of the graphic element is equal to or less than a predetermined value, the side is excluded from the correction target. Therefore, the reliability of correction is improved.
The pattern forming method according to claim 40 is the method according to claim 1, wherein when the calculated correction amount exceeds the upper limit value, the upper limit value is used as the correction amount, so that correction is performed when abnormality correction occurs. This improves the reliability of the correction.
The pattern forming method according to claim 41 is the method according to claim 1, wherein redundant points located in the middle of the straight line representing each side of the corrected graphic element are deleted and used as correction data. Thus, when iterative calculation is performed, the calculation time is shortened.

The pattern forming method according to claim 41 is the method according to claim 1, wherein after correcting all sides to be corrected, it is determined whether or not the calculated difference is within an allowable range. Based on this projection image, a projection image is formed again, the size of the pattern transferred to the wafer is predicted based on this projection image, and the correction data is calculated by calculating the difference between the predicted pattern size and the pattern size based on the design data. Further correction is made to create an optical projection image based on new correction data until the difference between the predicted pattern size and the pattern size by design data falls within the allowable range, and the size of the pattern transferred to the wafer. Since the prediction and the further correction of new correction data are repeated, the reliability of the correction is improved.
The pattern formation method according to claim 42 is the method according to claim 41 , wherein the correction amount is determined for each data block, and the data block whose correction amount is outside the allowable range is extracted into a separate file and based on the corrected data. Since the data is corrected again, the correction efficiency can be improved.

43. The integrated circuit manufacturing method according to claim 43, wherein the integrated circuit is manufactured using the pattern forming method according to any one of claims 1 to 42 , so that the deformation due to the effect of the optical proximity effect is reduced. Is obtained.

Embodiments of the present invention will be described below with reference to the accompanying drawings.
Embodiment 1. FIG.
FIG. 1 is a diagram showing a pattern forming method according to Embodiment 1 of the present invention in the order of steps. First, LSI design data as shown in FIG. 1A is created using CAD or the like. Here, a plurality of rectangular patterns 11 are formed as design data. Next, the optical proximity correction is performed on the design data by the optical proximity correction method described later to create optical proximity correction data shown in FIG. This optical proximity correction data has a plurality of patterns 12 corresponding to the rectangular pattern 11 of the design data. Each pattern 12 is rectangular in consideration of pattern deformation due to the light diffraction effect during subsequent wafer transfer. The auxiliary figure 121 is added to each of the four corners of the figure.

  By performing electron beam drawing using the optical proximity correction data, a mask having a plurality of patterns 13 is formed as shown in FIG. The mask pattern 13 is transferred to the wafer by exposing the mask together with ultraviolet rays. At this time, as shown in FIG. 1D, the transfer pattern 14 on the wafer has a rectangular shape with the auxiliary figures at the four corners being cut by the diffraction effect of light. When processing such as etching is performed on the wafer based on the transfer pattern 14, a pattern 15 as shown in FIG. 1E is obtained. The processed pattern 15 is deformed from the transfer pattern 14 in FIG. 1D due to the microloading effect, but the deformation amount is small compared to the pattern 294 by the conventional method shown in FIG. 29D. A pattern closer to the design data can be obtained. Thereafter, when an oxidation process or the like used for LOCOS separation is performed, the pattern shape is further deformed, but a pattern with higher dimensional accuracy can be formed as compared with the conventional example.

Embodiment 2. FIG.
FIG. 2 shows the configuration of an optical proximity correction apparatus that implements the optical proximity correction method used in the pattern formation method of Embodiment 1 described above. A data compression unit 2 that compresses data as preprocessing is connected to a design data input unit 1 that takes in design data of an integrated circuit created using CAD or the like, and a projection image at the time of wafer transfer is connected to the data compression unit 2. An optical image calculation unit 3 for calculation is connected. A pattern prediction unit 4 for predicting a resist pattern formed by wafer transfer is connected to the optical image calculation unit 3, and a comparison unit for comparing the predicted pattern and the design data to the pattern prediction unit 4 and the design data input unit 1. 5 is connected. The comparison unit 5 is connected to a correction unit 6 that performs optical proximity correction, and the correction unit 6 is connected to a determination unit 7 that determines whether the correction amount is within an allowable range. A data expansion unit 8 that expands data is connected to the determination unit 7, and a correction data output unit 9 is further connected to the data expansion unit 8. The optical image calculation unit 3 is connected to the determination unit 7. The optical image calculation unit 3 constitutes the optical image forming unit of claim 1.

  A correction method using the optical proximity correction apparatus having such a configuration will be described with reference to the flowchart of FIG. First, when design data created using CAD or the like is fetched from the design data input unit 1, the data compression unit 2 compresses the design data as preprocessing (step S1). Next, a projected image of the pattern transferred to the wafer is calculated by the optical image calculation unit 3 based on the compressed data (step S2). Further, the finished size of the transfer pattern is predicted by the pattern predicting unit 4 based on this projection image (step S3). The finished size of the transfer pattern predicted in this way is compared with the size based on the design data input to the design data input unit 1 in the comparison unit 5, and the difference between these is calculated as a correction amount (step S4). . Based on the correction amount calculated by the comparison unit 5, the correction unit 6 corrects the compressed data (step S5).

  Thereafter, the determination unit 7 determines whether or not the correction amount is within a preset allowable range (step S6). If the correction amount is out of the allowable range, it is determined that sufficient correction has not yet been performed. Returning to step S2, the projection image is calculated again, and the data is corrected according to steps S3 to S5. In this way, steps S2 to S6 are repeated until the correction amount becomes a value within the allowable range. If it is determined in step S6 that the correction amount is within the allowable range, it is determined that the appropriate optical proximity correction has been completed, and after the data decompressed by the data decompressing unit 8 is performed (step S7), the correction is performed. Correction data is output from the data output unit 9 (step S8). This correction data is input to an electron beam lithography apparatus (not shown) for forming a mask.

  As described above, since the data correction is performed based on the comparison between the predicted pattern and the design data after the data compression unit 2 compresses the data, the speed of the correction calculation can be increased. In addition, since the calculation is repeatedly performed until the correction amount is within the allowable range, the accuracy of the finished dimension is improved.

Embodiment 3. FIG.
FIG. 4 shows a block diagram of an optical proximity correction apparatus according to the third embodiment. The apparatus according to the third embodiment is different from the apparatus according to the first embodiment shown in FIG. 2 in that an optical image measuring unit 10 is provided as an optical image forming unit according to claim 1 instead of the optical image calculating unit 3. The optical image measurement unit 10 has an optical system, and measures a projected image of a mask created based on design data. In the second embodiment, the optical image calculation unit 3 calculates the projected image of the pattern transferred to the wafer based on the compressed data in step S2 by software. In the third embodiment, as shown in FIG. In addition, after the data is compressed in step S1, the optical image measuring unit 10 measures the projection image by hardware in step S9. In FIG. 5, steps S1, S3 to S8 are the same as the corresponding steps of the second embodiment shown in FIG.

  As the optical system of the optical image measuring unit 10, an optical system that satisfies the following conditions is used for a stepper that performs wafer transfer. That is, for the use wavelength λ1, the spatial coherency σ1, the magnification m1 and the numerical aperture NA1 of the stepper, the use wavelength λ2, the spatial coherency σ2, the magnification m2 and the numerical aperture NA2 of the optical image measurement unit 10 are λ1 = λ2. It is set so as to satisfy the relationship of σ1 = σ2 and m1 · NA1 = m2 · NA2. By using such an optical system, it is possible to accurately measure the diffraction effect of light at the time of transfer, and it is possible to increase the speed of correction calculation.

Embodiment 4 FIG.
In the data compression process of the second or third embodiment, the data compression unit 2 may divide the design data into a plurality of data blocks 6a to 6i as shown in FIG. Optical proximity correction is performed for each data block. Further, a buffer area is set around each data block. The buffer area is set in order to consider the relevance with the graphic elements in the surrounding data blocks when correcting the graphic elements in the corresponding data block. For example, in FIG. 6, a buffer area 60e is set around the data block 6e, and it is determined whether or not a graphic element that causes a proximity effect with the graphic element of the data block 6e exists in the buffer area 60e. The sides to be corrected in the data block 6e are determined according to the result. By doing so, the independence of each data block is ensured, and it is possible to speed up the correction process such as performing a parallel operation for each data block.

Embodiment 5. FIG.
The data compression unit 2 calculates the distance between each side of the graphic element and the side facing the side in each data block in the fourth embodiment, and the presence of sides facing each other is detected at an interval equal to or less than a predetermined value. In this case, the data block is a correction target. When such a side does not exist, the data block is excluded from the correction target. For example, as shown in FIG. 7A, when figures 7a and 7b exist in the data block, even if the distance A is sufficiently large, the distances B and C are smaller than a predetermined value and it is necessary to perform optical proximity correction. Therefore, this data block is a correction target. On the other hand, in the data block shown in FIG. 7B, since the length A of the figure 7c is sufficiently large and there are no sides facing each other at an interval equal to or smaller than a predetermined value, it is determined that there is no need for correction and the correction target It is assumed to be outside. By determining whether or not correction is necessary for each data block in this way, it is possible to speed up the correction process.

Embodiment 6. FIG.
When the modified illumination method is applied to the exposure apparatus, a light-shielding portion is formed on the secondary light source surface, so that a diffraction image of the light source is formed on the pupil plane of the exposure apparatus. For example, FIGS. 41A and 41B are obtained by modified illumination having linear and cross-shaped light-shielding portions, respectively, and zero-order light source images 411 and 421 and one primary light source image 412 on the pupil plane 410. 4 shows a diffraction pattern at a limit point where two-beam interference with 422 occurs. The period L2 corresponding to such a two-beam interference cutoff frequency is represented by L2 = λ / (σ + 1) NA. Where λ is the wavelength of light, σ is spatial coherency, and NA is the numerical aperture. That is, at a pattern longer than the period L2, interference of at least two light beams is established.

  FIGS. 41C and 41D show modified illuminations having linear and cross-shaped light shielding portions, respectively. Zero-order light source images 411 and 421 and ± primary light source images 412 and The diffraction pattern at the limit point where the three-beam interference between 413, 422 and 423 is established is shown. The period L3 corresponding to such a three-beam interference cutoff frequency is represented by L3 = λ / (1-σb) NA. Where σb is the spatial coherency regarding the light-shielding part of the light source image. That is, three-beam interference is established with a pattern longer than the cycle L3.

  In the normal illumination method, when the finished size of a 0.35 μm wide line for various space widths and the finished size of a 0.35 μm wide space for various line widths were measured, the optical proximity effect is shown in FIGS. It was found that this occurred as in b). The small pattern in the figure is a pattern having a size of the period L2 or less, and condition reading is performed using this small pattern. For this reason, in the small pattern, the dimensions of the finishing line and the finishing space are not affected by the optical proximity effect and are not shifted. The middle pattern is a pattern having a size from the period L2 to L3, and the large pattern is a pattern larger than the period L3, and the dimensions of the finishing line and the finishing space are both shifted. Therefore, as shown by a circle in FIG. 44 (a), it is necessary to correct the optical proximity effect when the line or space is a medium pattern or a large pattern.

  When the same measurement was performed by the modified illumination method, the optical proximity effect was generated as shown in FIGS. 43 (a) and 43 (b). The small pattern, the medium pattern, and the large pattern in the figure are similar to the above-described normal illumination method. Is shown. Since the condition setting is performed using a small pattern, the dimensions of the finishing line and the finishing space are not affected by the optical proximity effect and are not shifted in the small pattern. In the modified illumination method, it was also found that the finished line width with respect to the space of the middle pattern causing the two-beam interference is not affected by the optical proximity effect and does not shift. In other patterns, the finished dimensions are shifted. For this reason, as indicated by a circle in FIG. 44 (b), when the line is a medium pattern or a large pattern and when the space is a large pattern, correction for the optical proximity effect is required. In some cases, no correction is required.

  Therefore, when the modified illumination method is used, in the data compression step, a light shielding pattern having a size equal to or less than the period L3 corresponding to the three-beam interference cutoff frequency can be excluded from the correction target.

  Further, as shown in FIG. 43A, the deviation amount of the finished line width with respect to the large pattern space is substantially constant. Therefore, it is necessary to perform optical proximity correction with a magnitude equal to or smaller than a predetermined value, and a certain correction amount is applied to sides facing each other at intervals of a magnitude greater than or equal to the period L3 corresponding to the cutoff frequency of three-beam interference. By adding, good pattern correction can be easily performed.

Embodiment 7. FIG.
In the data compression process, when processing a huge figure extending over a plurality of data blocks as shown in FIG. 8A, the data blocks 8a to 8B having the sides of the figure as shown in FIG. Only 8s is a correction target, and a data block having no graphic side is not a correction target. This is because the optical proximity effect does not occur if there is no side of the figure. In this way, data blocks that do not need to be corrected can be excluded from the correction calculation, and the correction process can be speeded up.

Embodiment 8. FIG.
In the data compression step, when the same graphic element exists in a plurality of data blocks, one of the data blocks is set as a correction target, and the remaining data blocks are not set as correction targets. For example, the horizontally long graphic shown in the upper part of FIG. 9A extends over the data blocks 9a to 9e. Among these data blocks, the data blocks 9b, 9c and 9d have the same graphic elements. Yes. Therefore, as shown in FIG. 9B, of the data blocks 9b to 9d, only the data block 9b is targeted for correction, and the correction result is diverted to the data blocks 9c and 9d after correction. According to the seventh embodiment, only the data blocks 9a, 9b, and 9e to 9m shown in FIG. 9B need be corrected with respect to the graphic shown in FIG. 9A. Therefore, the correction process can be speeded up.

Embodiment 9. FIG.
In the data compression step, when the figure to be processed has an array portion in which the same cells are continuous, first, the figure 100 of one cell shown in FIG. A figure 101 as shown in FIG. 10B is created, and then the array is developed as shown in FIG. In this way, the correction process can be speeded up.

Embodiment 10 FIG.
As shown in FIG. 11, for the design data of the memory cell array in which the memory cell graphics 110 are continuous, the data block 11a is divided in accordance with the cycle of the memory cells in the data compression step. That is, the division cycle of the data block 11a is matched with the cycle of the memory cell. By doing so, the periodic boundary condition can be used, so that it is not necessary to set the buffer area as shown in FIG. In addition, since the fast Fourier transform (FFT) can be used effectively, the processing speed can be increased.

Embodiment 11. FIG.
As shown in FIG. 12A, when a plurality of graphic elements 121 to 125 are adjacent to each other in the data block, in the data compression process, these graphic elements 121 to 125 are combined to form FIG. ) Is formed. By performing such polygon processing, data is compressed and redundant sides do not exist inside the figure, so that erroneous correction processing is avoided.

Embodiment 12 FIG.
As shown in FIG. 13A, when a hollow graphic is formed by a plurality of graphic elements 131 to 137 in the data block, as shown in FIG. The graphic elements 131 to 137 are expressed by graphic elements 138 that define the outer periphery of the graphic elements 131 and 137, and graphic elements 139a and 139b that define hollow portions. By doing so, it is easy to correct the hollow graphic.

Embodiment 13 FIG.
As shown in FIG. 14, the optical image calculation unit 3 in FIG. 2 can include a plurality of CPUs 141 to 145 connected in parallel to each other. The design data compressed by the data compression unit 2 is divided, and each of the CPUs 141 to 145 performs a parallel operation separately, and then the correction result is created by combining the calculation results. By using such a parallel operation, high-speed processing is possible. For example, as shown in FIG. 14, if five CPUs are used, the processing speed can be increased five times as compared with the case where one CPU is used.

  2 constitutes the data compression unit 2, the data expansion unit 8, and the correction data output unit 9 shown in FIG. 2, and the CPUs 142 to 145 serve as a plurality of child processors, whereby the optical image calculation unit 3 shown in FIG. The pattern prediction unit 4, the comparison unit 5, the correction unit 6, and the determination unit 7 can be configured. The calculation processing method in this case is shown in FIG. First, when CAD data having a mask pattern is input as design data to the parent processor, the parent processor divides the CAD data into a plurality of data blocks, removes duplication between the data, and compresses the data. Set the buffer area around the block. Next, the optical data of the exposure system (not shown) is input to the parent processor and the exposure result of the standard pattern is input. The parent processor obtains the correlation between the standard pattern from the exposure result of the standard pattern and its optical image, in particular, the threshold value of the light intensity set for pattern prediction, and performs common operations for each data block. The correlation and the common calculation result are transmitted to a plurality of child processors.

  After synchronizing with the child processor, the parent processor waits until a signal is received from the child processor. When a processing wait signal is received from the child processor, the data block data corresponding to the child processor is transmitted to the child processor together with the processing start signal. On the other hand, when the calculation end signal is received from the child processor, the calculation result is subsequently received and stored from the child processor, and when the processing of all data is completed, the end signal is transmitted to each child processor to perform all processing. finish. If there is unprocessed data, it waits until the next signal is received from the child processor.

On the other hand, when the child processor receives the processing start signal and the data of the corresponding data block from the parent processor, the child processor performs optical calculation and correction calculation, and compresses the correction data when the correction amount is within the allowable range, to the parent processor. The calculation result is transmitted together with the calculation end signal.
In this way, a significant speed increase is achieved by performing parallel operations on the outermost loop of the flowchart shown in FIG. 3 by a plurality of child processors.

Embodiment 14 FIG.
In the thirteenth embodiment, the correlation between the standard pattern and its optical image can be obtained from the exposure result of the standard pattern as follows. As shown in FIG. 31, first, the dimension of the resist pattern obtained by actually exposing the standard pattern is measured. The parent processor calculates an optical image of the standard pattern calculated from the standard mask data, and searches for the ambient light intensity Ix. Next, a deviation amount Δ between the measured value of the resist pattern and the standard pattern is obtained, and the light intensity Iy at a position shifted by Δ of the optical image of the standard pattern is retrieved. Further, the parent processor obtains a correlation between the light intensities Ix and Iy by least square fitting. The child processor calculates a different threshold according to the ambient light intensity Ix based on the correlation obtained by the parent processor, and performs correction calculation using this threshold.
By doing so, it is possible to flexibly cope with variations accompanying changes in the resist and processes, and to perform highly accurate correction.

  As shown in FIG. 32, the dimension of the etching pattern obtained by etching after exposing the standard pattern can be used instead of the dimension of the resist pattern. In this way, it is possible to cope with fluctuations associated with etching and to perform highly accurate correction.

Embodiment 15. FIG.
In the thirteenth embodiment, instead of obtaining the correlation between the standard pattern and its optical image, the correlation between the standard mask data and the mask pattern after manufacturing the mask can be obtained. As shown in FIG. 45, first, the size of a standard pattern obtained by actually writing standard mask data with an electron beam is measured. On the other hand, the parent processor calculates an electron beam drawing pattern from the standard mask data, and searches for an electron beam drawing pattern Ix around the mask edge. Next, a deviation amount Δ between the measured value of the standard pattern dimension and the dimension of the electron beam drawing pattern is obtained, and the electron beam drawing pattern Iy at a position shifted by Δ is searched. Further, the parent processor obtains the correlation between Ix and Iy by least square fitting. The child processor calculates a different threshold according to the surrounding electron beam drawing pattern Ix based on the correlation obtained by the parent processor, and performs correction calculation using this threshold.
In this way, only the electron beam proximity effect when the mask is drawn with the electron beam can be evaluated and incorporated into the correction.

  Also, as shown in FIG. 46, using laser drawing instead of electron beam drawing, a standard pattern obtained by actually performing laser drawing of standard mask data, and a laser drawing pattern calculated from the standard mask data. A correlation may be obtained. In this way, only the proximity effect when the mask is laser-drawn can be evaluated and incorporated into the correction.

Embodiment 16. FIG.
In the thirteenth embodiment, instead of obtaining the correlation between the standard pattern and its optical image, the correlation between the pattern size of the standard pattern and the pattern size of the resist pattern obtained by exposing the standard pattern can be obtained. As shown in FIG. 47, for example, the standard pattern obtained in the fifteenth embodiment is actually exposed to form a resist pattern, and the dimension of this resist pattern is measured. On the other hand, the parent processor calculates an optical image from the pattern size obtained by measuring the standard pattern, and searches for the optical image Ix around the mask edge. Next, a deviation amount Δ between the measured value of the resist pattern and the standard pattern is obtained, and the optical image Iy at a position shifted by Δ is searched. Further, the parent processor obtains the correlation between Ix and Iy by least square fitting. The child processor calculates a different threshold according to the surrounding optical image Ix based on the correlation obtained by the parent processor, and performs correction calculation using this threshold.
By doing so, it is possible to evaluate and incorporate only the optical proximity effect generated during the optical transfer of the mask into the correction.

Embodiment 17. FIG.
In the thirteenth embodiment, instead of obtaining the correlation between the standard pattern and its optical image, the correlation between the dimension of the resist pattern and the dimension of the etching pattern obtained by etching the resist pattern can be obtained. As shown in FIG. 48, for example, the standard resist pattern obtained in Embodiment 16 is actually etched to form an etching pattern, and the dimension of this etching pattern is measured. On the other hand, the parent processor calculates the etching pattern from the pattern dimension obtained by measuring the standard resist pattern, and searches for the etchant concentration Ix around the resist edge. Next, a deviation amount Δ between the measured value of the etching pattern dimension and the standard resist pattern dimension is obtained, and the etchant concentration Iy at a position shifted by Δ is searched. Further, the parent processor obtains the correlation between Ix and Iy by least square fitting. The child processor calculates a different threshold according to the surrounding etchant concentration Ix based on the correlation obtained by the parent processor, and performs correction calculation using this threshold.
In this way, only the microloading effect that occurs during etching can be evaluated and incorporated into the correction.

Embodiment 18. FIG.
In the thirteenth embodiment, instead of obtaining the correlation between the standard pattern and its optical image, the correlation between the standard mask data and the resist pattern after electron beam drawing and light transfer can be obtained. As shown in FIG. 49, first, standard mask data is actually drawn with an electron beam to form a standard pattern, and the dimension of the resist pattern obtained by exposing the standard pattern is measured. On the other hand, the parent processor calculates an electron beam drawing pattern from the standard mask data, calculates an optical image, and then searches for an optical image Ix around the mask edge. Next, a deviation amount Δ between the measured value of the resist pattern and the standard mask data is obtained, and the optical image Iy at a position shifted by Δ is searched. Further, the parent processor obtains the correlation between Ix and Iy by least square fitting. The child processor calculates a different threshold according to the surrounding optical image Ix based on the correlation obtained by the parent processor, and performs correction calculation using this threshold.
By doing so, the electron beam proximity effect and the optical proximity effect when the electron beam is drawn from the standard mask data and further optically transferred can be evaluated in a composite manner and taken into the correction.

  Further, as shown in FIG. 50, a resist pattern obtained by using laser drawing instead of electron beam drawing, forming a standard pattern by actually drawing the standard mask data by laser drawing, and exposing the standard pattern After calculating the laser drawing pattern from the standard mask data, the correlation with the optical image calculated from the laser drawing pattern may be obtained. In this way, the laser drawing proximity effect and the optical proximity effect when laser drawing is performed from the standard mask data and optical transfer is performed can be combined and evaluated for incorporation.

Embodiment 19. FIG.
In the thirteenth embodiment, instead of obtaining the correlation between the standard pattern and its optical image, the correlation between the size of the standard pattern and the size of the etching pattern obtained by exposing and etching the standard pattern. You can also ask for relationships. As shown in FIG. 51, for example, after the standard pattern obtained in the fifteenth embodiment is actually exposed, an etching pattern is formed by etching, and the dimension of this etching pattern is measured. On the other hand, the parent processor calculates an optical image from the pattern size obtained by measuring the standard pattern, calculates an etching pattern from this optical image, and searches for the etchant concentration Ix around the resist edge. Next, the amount of deviation Δ between the measured value of the etching pattern and the standard pattern is obtained, and the etchant concentration Iy at the position shifted by Δ is searched. Further, the parent processor obtains the correlation between Ix and Iy by least square fitting. The child processor calculates a different threshold according to the surrounding etchant concentration Ix based on the correlation obtained by the parent processor, and performs correction calculation using this threshold.
By doing in this way, the optical proximity effect and the microloading effect generated at the time of optical transfer and etching can be evaluated in a composite manner and taken into correction.

Embodiment 20. FIG.
In the thirteenth embodiment, instead of obtaining the correlation between the standard pattern and its optical image, the correlation between the standard mask data and the etching pattern after electron beam drawing, optical transfer, and etching can be obtained. As shown in FIG. 52, first, standard mask data is actually drawn by an electron beam to form a standard pattern, and after the standard pattern is exposed, the dimension of the etching pattern obtained by etching is measured. On the other hand, the parent processor calculates an electron beam drawing pattern from the standard mask data, calculates an optical image, calculates an etching pattern from the optical image, and searches for an etchant concentration Ix around the resist edge. Next, a deviation amount Δ between the measured value of the etching pattern dimension and the standard mask data is obtained, and the etchant concentration Iy at a position shifted by Δ is searched. Further, the parent processor obtains the correlation between Ix and Iy by least square fitting. The child processor calculates a different threshold according to the surrounding etchant concentration Ix based on the correlation obtained by the parent processor, and performs correction calculation using this threshold.
By doing in this way, the electron beam proximity effect, the optical proximity effect, and the microloading effect that are generated at the time of electron beam drawing, light transfer, and etching can be evaluated in a composite manner and incorporated into correction.

  Further, as shown in FIG. 53, laser drawing was used instead of electron beam drawing, and standard mask data was actually laser drawn to form a standard pattern, which was then exposed and etched. After sequentially calculating the laser drawing pattern and the optical image from the standard mask data, the correlation between the etching pattern and the etching pattern calculated from the optical image may be obtained. In this way, the laser drawing proximity effect, the optical proximity effect, and the microloading effect that occur during laser drawing, light transfer, and etching can be evaluated and incorporated into correction.

Embodiment 21. FIG.
As shown in FIG. 15, the optical image measurement unit 10 in FIG. 4 can be configured from a plurality of optical systems 151 to 155. The design data compressed by the data compression unit 2 is divided and separately measured in parallel by the optical systems 151 to 155, and then correction data is created by combining the measurement results. By using such parallel measurement, high-speed processing becomes possible. For example, if five optical systems are used as shown in FIG. 15, a high-speed processing 5 times faster than that using a single optical system can be achieved.

Embodiment 22. FIG.
As shown in FIG. 16, after the projection image is formed from the mask pattern 161 based on the design data, in the transfer pattern prediction step, the pattern prediction unit 4 performs masking according to the position having the light intensity I of the predetermined threshold value ITH. The edge is predicted, and thereby the size of the pattern 162 transferred to the resist or the like on the wafer surface is predicted. At this time, the distance D between the mask edge of the mask pattern 161 based on the design data and the predicted mask edge is the correction amount. The threshold value ITH is set to a light intensity of about 0.20 to 0.40, where the light intensity in a flat pattern having no boundary portion is 1. Thus, by predicting the mask edge using the threshold value ITH, the resist development calculation can be omitted, and high-speed processing is possible.

Embodiment 23. FIG.
In the twenty-second embodiment, the threshold value ITH can be adjusted according to the light intensity at the periphery of the side of the graphic element to be corrected. For example, when a mask pattern 171 as shown in FIG. 17A is used, the light transmission portion is small, so that the light intensity at the periphery of the correction target side is not sufficiently strong. In this case, the threshold value ITH is set high. On the other hand, when mask patterns 173, 181 and 183 as shown in FIG. 17B, FIG. 18A and FIG. 18B are used, the light transmission part is large. In this case, the threshold value ITH is set low. By doing so, the transfer patterns 172, 174, 182 and 184 can be predicted reflecting the state of the periphery of the side to be corrected, so that high-speed and high-precision pattern prediction is possible.

Embodiment 24. FIG.
In the twenty-second embodiment, the threshold value ITH can be adjusted in accordance with the two-dimensional light intensity distribution around the correction target side. For example, when correcting a rectangular figure as shown in FIG. 33, the point Pa located at the corner of the rectangular pattern 330 and the point Pb located at the side have different light intensity distributions at the periphery thereof, so correction is performed. The amount is also different. Therefore, a plurality of monitor points Pm are provided around each point Pa and Pb to monitor the two-dimensional light intensity distribution, and it is identified whether the points Pa and Pb are located at corners or sides. . If it is located at the corner, as shown in FIG. 34, it is necessary to set a correction amount D4 larger than the side correction amount D3 in consideration of the influence of etching, development, etc. Set low. By doing so, it is possible to reflect the influence of the two-dimensional environment around the pattern, and it is possible to predict the pattern with high accuracy.

Embodiment 25. FIG.
In the twenty-second embodiment, the threshold value ITH can be adjusted according to the presence / absence of a graphic element close to the side to be corrected. For example, in the mask pattern 351 as shown in FIG. 35A, there is no adjacent graphic element in the vicinity of the side 351a to be corrected, but in the mask pattern 353 as shown in FIG. There is an adjacent graphic element in the vicinity of the side 353a. For this reason, when the mask pattern 353 is used, the light intensity decreases on the right side of the side 353a. Therefore, monitor points Pm are provided on the left and right sides of the correction target side to identify whether adjacent graphic elements are close to each other. When the sum of the light intensities at the left and right monitor points Pm is large as in the mask pattern 351, it is determined that there is no adjacent graphic element, and the threshold value ITH is set high. On the other hand, when the sum of the light intensities at the left and right monitor points Pm is small as in the mask pattern 353, it is determined that there is an adjacent graphic element, and the threshold value ITH is set low. In this way, highly accurate pattern prediction is possible.

Embodiment 26. FIG.
In the twenty-second embodiment, the threshold value ITH can be adjusted according to the light intensity of the optical image at the time of defocusing with respect to the light intensity of the optical image at the time of best focus. As shown in FIG. 36, the light intensity Id of the optical image at the time of defocus is calculated, and it is determined whether or not this light intensity Id is larger than the light intensity Ib of the optical image at the time of best focus. When the light intensity Id at the time of defocusing is larger than the light intensity Ib at the time of best focus, the resist is formed small. Therefore, the threshold value ITH is lowered to finish the resist large. Since the resist is formed large, the threshold value ITH is increased to finish the resist small. By doing so, the effective focus margin can be expanded.

Embodiment 27. FIG.
In the twenty-second embodiment, the threshold value ITH may be adjusted based on the tilt of the projected image at the predicted mask edge, and then the mask edge may be predicted again using the adjusted threshold value ITH. As shown in FIGS. 19A and 19B, the threshold value ITH is adjusted in accordance with the inclination of the projected image at the mask edges once predicted for the mask patterns 191 and 193, and the new threshold value ITH is used again. The mask edge in the projected image is predicted, and the transfer patterns 192 and 194 are predicted. According to this method, since the inclination of the projected image, that is, the inclination of the light intensity can be reflected, more accurate pattern prediction can be performed.

Embodiment 28. FIG.
As shown in FIG. 20, after forming a projected image based on the mask pattern 201, the pattern predicting unit 4 performs a resist development simulation on the wafer surface based on the projected image, and predicts the transfer pattern 202 from the result. You may make it do. According to this method, it is possible to easily cope with changes in process conditions such as exposure amount and development time, and it is possible to predict patterns with higher accuracy.

Embodiment 29. FIG.
As shown in FIG. 21, after forming a projection image based on the mask pattern 211, the pattern predicting unit 4 converts the projection image into a distribution 212 of the required development time of the resist on the wafer surface. It is also possible to perform pseudo development by integrating one-dimensionally and predict the transfer pattern 213 from the result. According to this method, pattern prediction is performed more easily and faster than development simulation.

Embodiment 30. FIG.
After the projection image is formed from the mask pattern based on the design data, the threshold value ITH can be adjusted in the transfer pattern prediction step according to the reflectance of the base substrate to which the pattern is transferred. As shown in FIG. 37A, when a substrate 371 having a low reflectance such as WSi is used as the base substrate, the influence of exposure by the reflected light from the substrate 371 is small, so that the resist dimensions of the positive resist are small. Finished greatly. On the other hand, as shown in FIG. 37 (b), when a substrate 372 having a high reflectance such as Al is used as the base substrate, exposure is also performed by reflected light from the substrate 372. Is finished small. Therefore, the threshold ITH is set high for the base substrate having a low reflectance as shown in FIG. 37 (a), and conversely, the threshold value ITH is set for the base substrate having a high reflectance as shown in FIG. 37 (b). Set threshold ITH low. By doing so, the transfer pattern can be predicted in consideration of the influence of the reflectance of the base substrate, and the pattern can be predicted with high speed and high accuracy.

Embodiment 31. FIG.
After the projection image is formed from the mask pattern based on the design data, the threshold value ITH can be adjusted in the transfer pattern prediction step in accordance with the level difference formed on the base substrate to which the pattern is transferred. As shown in FIG. 38A, a resist 383 is formed so as to cover the step of the base substrate 381 or the opening 382, and the resist film thickness of the opening 382 is different from that shown in FIG. If it is thicker than the resist film thickness of the portion, the threshold value ITH is set high. By doing so, the transfer pattern can be predicted in consideration of the level difference of the base substrate and the local variation of the resist film thickness, and the pattern can be predicted with high accuracy.

Embodiment 32. FIG.
After the projection image is formed from the mask pattern based on the design data, the threshold value ITH can be adjusted in the transfer pattern prediction step according to the halation generated on the surface of the base substrate to which the pattern is transferred. For example, as shown in FIG. 39A, a large step 393 is usually formed at the boundary between the memory cell portion 391 and the peripheral circuit portion 392. When the bit line 394 is formed across the step 393 from the memory cell portion 391 to the peripheral circuit portion 392, as shown in FIG. 39B, halation occurs at the portion of the step 393, and the reflection in the lateral direction. Light is generated. Therefore, as shown in FIG. 39C, the threshold value ITH of the step portion is set lower than the other portions. By doing so, the transfer pattern can be predicted in consideration of the influence of halation due to the level difference of the base substrate, and the pattern can be predicted with high accuracy.

Embodiment 33. FIG.
In the calculation of the projection image in the optical image calculation unit 3 according to the second embodiment, first, the light intensity at the intersection point of the preset mesh is calculated, and then already calculated as i = 1 to 4 as shown in FIG. Further, the light intensity I at the point P (x, y) inside the quadrangle surrounded by these four points is interpolated from the light intensity Ii of the four points Pi (xi, yi) according to the equation I = Σi (Wi · Ii). Can do. However, Wi = (1- | xi-x |) (1- | yi-y |). If the projected image is calculated in this way, the light intensity can be calculated even on the boundary line of the mesh, and a highly accurate projected image can be obtained.

Embodiment 34. FIG.
In the data correction process of the second or third embodiment, the correction unit 6 divides each side of the figure 231 before correction shown in FIG. 23 (a) into a plurality of line segments as shown in FIG. 23 (b). After that, each line segment can be corrected. At this time, each division point 232 is caused to generate two data corresponding to both line segments sharing the division point. In this way, when one line segment is corrected, the influence on the adjacent line segment is prevented, and high-accuracy correction is possible.

Embodiment 35. FIG.
In the data correction step, the correction unit 6 can perform correction by limiting the correction direction of each side to be corrected to a direction perpendicular to the side. For example, the side 242 of the figure 241 before correction shown in FIG. 24A is corrected in a direction perpendicular to the side 242 to form a side 244 shown in FIG. 24B, and the figure 243 after correction is obtained. . In this way, the side of the diagonal direction is not formed by the correction, and the data is compressed. For this reason, high-speed processing becomes possible at the time of electron beam drawing.

Embodiment 36. FIG.
In the data correction step, the correction unit 6 can exclude the side from the correction target when the inclination at the pattern edge of the projection image is equal to or less than a predetermined value. For example, as shown in FIG. 25, when the mask pattern 251 is too fine, the transfer pattern 252 becomes unclear and accurate pattern transfer cannot be performed. In such a case, the inclination at the pattern edge of the projected image of the mask pattern 251 is small. Therefore, when the inclination at the pattern edge of the projected image is equal to or less than a predetermined value, the side is excluded from the correction target, so that the occurrence of abnormal correction can be prevented in advance, and highly reliable optical proximity correction processing can be performed. .

Embodiment 37. FIG.
In the data correction process, when the light intensity of the projection image is a predetermined value or less, the correction unit 6 can exclude the side from the correction target. For example, the side 263 inside the figure 261 shown in the portion 262 of FIG. 26A is a side that does not need to be subjected to optical proximity correction and should not be corrected. Since such a side 263 is located inside the figure 261, the light intensity of the projected image has an extremely small value as shown in FIG. Therefore, when the light intensity of the projected image is less than or equal to the predetermined value I〓, the occurrence of abnormality correction can be prevented in advance by excluding that side from the correction target, and highly reliable optical proximity correction processing is performed. It becomes possible.

Embodiment 38. FIG.
In the data correction step, the correction unit 6 sets an upper limit value of the correction amount, and when the calculated correction amount exceeds the upper limit value, determines that the correction is abnormal, and performs correction using the upper limit value as the correction amount. be able to. For example, as illustrated in FIG. 27A, when the correction amount D1 of the side 272 of the graphic 271 exceeds the upper limit value D2, the correction amount is set to the upper limit value as illustrated in the graphic 273 of FIG. A new side 274 is formed by replacing D2. In this way, even if abnormality correction occurs, this can be avoided, and highly reliable optical proximity correction processing can be performed.

Embodiment 39. FIG.
In the data correction step, the correction unit 6 may delete redundant points existing on the same straight line after correcting each side to be corrected. For example, in the corrected figure 281 shown in FIG. 28A, the points 282 to 285 are redundant points located in the middle of the straight line, and are unnecessary as corrected data. Therefore, the correction unit 6 removes these points 282 to 285 to obtain a figure as shown in FIG. By doing so, the data is compressed, and the calculation time can be shortened when the correction calculation is repeated.

Embodiment 40. FIG.
In step S6 of the second or third embodiment, when the correction amount is outside the allowable range, the projection image is formed again to perform correction. At this time, the correction amount is changed every time each side to be corrected is corrected. Instead of determining and repetitively correcting, the correction amount is determined after correcting all sides independently of each other, and iteratively correcting as necessary. In this way, the correction process is speeded up. In addition, since the independence of the correction of each side is maintained, asymmetric correction is unlikely to occur, and the correction reliability is improved.

Embodiment 41. FIG.
In the second or third embodiment, the design data is divided into a plurality of data blocks, a correction amount is calculated for each data block, the correction amount is determined, and a data block whose correction amount is outside the allowable range is stored in a separate file. You may make it extract. Note that correction data is output for a data block whose correction amount is within an allowable range. For example, as shown in FIG. 54, after pre-processing the mask data, optical calculation is performed to predict a resist pattern, and evaluation is performed by comparing the predicted pattern with the mask data. A difference between the predicted pattern dimension and the dimension based on the mask data is calculated as a correction amount. When the correction amount is 10% of the minimum dimension, for example, 0.03 μm or more, the data block is extracted into another file.

  As a result, only an area requiring correction and an area having a small process margin are extracted from an LSI pattern that requires huge data. In the case of random logic, development efficiency is improved by using this function. The correction-required data extracted in this way is preprocessed using the optical parameters, and the mask data is displayed and the projection image is calculated and displayed. Whether correction is necessary or not is determined from the displayed projection image while referring to the correction data previously output for the data block within the allowable range. If correction is necessary, after manual correction, the mask data and the projected image are displayed again to make a determination. Thus, manual correction, mask data, and display of the projected image are repeated until correction is not necessary. If it is determined that correction is no longer necessary, correction data is output.

  In this way, in addition to a fully automatic optical proximity correction system that performs correction for each data block, a semi-auto optical proximity correction system that manually corrects the extracted data is used, so that the fully automatic system is immature In the case of special patterns, flexible correction can be efficiently performed when it is desired to monitor an optical image sequentially, which is effective.

Embodiment 42. FIG.
In the second or third embodiment, as shown in FIG. 40A, the design data is divided into a plurality of data blocks, and the buffer area 402 is set around each data block 401 to perform correction calculation, and then step S7. If the buffer area of each data block 403 is deleted as shown in FIG. 40 (b) in the data development process in FIG. 40 (b), then the correction data is stored and further developed as shown in FIG. 40 (c). Data is compressed.

It is a figure which shows the pattern formation method which concerns on Embodiment 1 of this invention in order of a process. It is a block diagram which shows the optical proximity correction apparatus in Embodiment 2. 10 is a flowchart illustrating the operation of the second embodiment. It is a block diagram which shows the optical proximity correction apparatus in Embodiment 3. 10 is a flowchart showing the operation of the third embodiment. It is a figure which shows the correction method which concerns on Embodiment 4. FIG. It is a figure which shows the correction method which concerns on Embodiment 5. FIG. It is a figure which shows the correction method which concerns on Embodiment 7. FIG. It is a figure which shows the correction method which concerns on Embodiment 8. FIG. It is a figure which shows the correction method which concerns on Embodiment 9. FIG. It is a figure which shows the correction method which concerns on Embodiment 10. FIG. It is a figure which shows the correction method which concerns on Embodiment 11. FIG. FIG. 20 is a diagram illustrating a correction method according to a twelfth embodiment. It is a figure which shows the optical image calculation part of the correction apparatus which concerns on Embodiment 13. FIG. FIG. 22 is a diagram illustrating an optical image measurement unit of a correction apparatus according to a twenty-first embodiment. FIG. 22 is a diagram illustrating a correction method according to a twenty-second embodiment. It is a figure which shows the correction method which concerns on Embodiment 23. FIG. It is a figure which shows the correction method which concerns on Embodiment 23. FIG. FIG. 38 shows a correction method according to Embodiment 27. FIG. 29 is a diagram illustrating a correction method according to Embodiment 28. FIG. 32 shows a correction method according to Embodiment 29. It is a figure which shows the correction method which concerns on Embodiment 33. FIG. It is a figure which shows the correction method which concerns on Embodiment 34. FIG. FIG. 38 shows a correction method according to Embodiment 35. It is a figure which shows the correction method which concerns on Embodiment 36. FIG. FIG. 38 shows a correction method according to Embodiment 37. FIG. 38 shows a correction method according to Embodiment 38. It is a figure which shows the correction method which concerns on Embodiment 39. FIG. It is a figure which shows the conventional pattern formation method in process order. It is a figure which shows the correction method which concerns on Embodiment 13. FIG. It is a figure which shows the correction method which concerns on Embodiment 14. FIG. It is a figure which shows the modification of the correction method which concerns on Embodiment 14. FIG. FIG. 25 is a diagram illustrating a correction method according to a twenty-fourth embodiment. FIG. 25 is a diagram illustrating a correction method according to a twenty-fourth embodiment. FIG. 25 is a diagram illustrating a correction method according to a twenty-fifth embodiment. FIG. 32 is a diagram illustrating a correction method according to a twenty-sixth embodiment. FIG. 22 is a diagram illustrating a correction method according to a thirtieth embodiment. It is a figure which shows the correction method concerning Embodiment 31. It is a figure which shows the correction method which concerns on Embodiment 32. FIG. It is a figure which shows the correction method which concerns on Embodiment 42. FIG. It is a figure which shows the principle of the correction method which concerns on Embodiment 6. FIG. It is a figure which shows the principle of the correction method which concerns on Embodiment 6. FIG. It is a figure which shows the principle of the correction method which concerns on Embodiment 6. FIG. It is a figure which shows the principle of the correction method which concerns on Embodiment 6. FIG. FIG. 18 is a diagram illustrating a correction method according to Embodiment 15. FIG. 22 is a diagram illustrating a modification of the correction method according to the fifteenth embodiment. FIG. 18 is a diagram illustrating a correction method according to Embodiment 16. FIG. 18 is a diagram illustrating a correction method according to Embodiment 17. FIG. 19 is a diagram illustrating a correction method according to an eighteenth embodiment. FIG. 22 is a diagram illustrating a modification of the correction method according to the eighteenth embodiment. FIG. 20 is a diagram illustrating a correction method according to a nineteenth embodiment. It is a figure which shows the correction method which concerns on Embodiment 20. FIG. FIG. 32 is a diagram illustrating a modification of the correction method according to the twentieth embodiment. 42 is a diagram illustrating a correction method according to Embodiment 41. FIG.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 Design data input part, 2 Data compression part, 3 Optical image calculation part, 4 Pattern prediction part, 5 Comparison part, 6 Correction part, 7 Judgment part, 8 Data expansion part, 9 Correction data output part, 10 Optical image measurement part 141-145 CPU, 151-155 optical system, 6a-6i, 8a-8s, 9a-9m, 11a data block, 60e buffer area.

Claims (43)

Create design data consisting of multiple graphic elements that make up the circuit pattern,
The design data is divided into a plurality of data blocks, an optical projection image at the time of pattern transfer is created for each data block, the size of the pattern transferred to the wafer is predicted based on the projection image, and the predicted pattern is determined. Creates correction data that corrects the optical proximity effect that occurs during pattern transfer to the wafer by calculating the difference between the dimensions of the pattern and the pattern dimensions of the design data, and correcting the design data using the calculated difference as a correction amount And
Create a mask pattern based on the correction data,
The mask pattern is transferred to the wafer by exposing through the mask pattern,
Process the wafer according to the transferred mask pattern ,
When correcting the design data, each side of the graphic element to be corrected is divided into a plurality of line segments, corrected for each line segment, and then the division point divided into the plurality of line segments is in the middle of a straight line representing each side. A pattern forming method characterized by deleting the division point when it is located in the position and using it as correction data . Determine whether the calculated difference is within an acceptable range,
When it is out of the allowable range, an optical projection image is created again based on the created correction data, and the size of the pattern transferred to the wafer is predicted based on this projection image, and the predicted pattern size and design data are used. Calculate the difference between the pattern size and further correct the correction data. The optical data based on the new correction data until the difference between the predicted pattern size and the pattern size from the design data is within the allowable range. The pattern forming method according to claim 1, wherein generation of a projection image, prediction of a dimension of a pattern transferred to a wafer, and further correction of new correction data are repeated.   3. The pattern forming method according to claim 1, wherein the correction data is data having a shape obtained by adding an auxiliary pattern to a pattern based on design data. In order to consider the relationship with the graphic elements in the surrounding data blocks around each data block, after correcting the design data by setting a buffer area, the buffer area set around each data block is deleted. 2. The pattern forming method according to claim 1 , wherein this is used as correction data .   Checks whether graphic elements in each data block have sides facing each other at intervals of a predetermined value or less, and if so, sets that data block as a correction target, and if not, sets that data block as a correction target 2. The pattern forming method according to claim 1, wherein the pattern is formed outside.   2. The pattern forming method according to claim 1, wherein only a data block having a side of a graphic element is subjected to correction, and a data block having no side is excluded from correction. 2. The pattern forming method according to claim 1 , wherein when the same graphic element exists in a plurality of data blocks, only one of the data blocks is to be corrected, and the correction result is used for another data block.   2. The pattern forming method according to claim 1, wherein the division period of the data block is matched with the period of the memory cell with respect to the design data of the memory cell array. The combined plurality of graphic elements adjacent to each other in each data block a polygon form, this polygon as a corrected predicted dimension of the pattern to be transferred onto the wafer, the predicted pattern dimension and design The pattern forming method according to claim 1, wherein the design data is corrected by calculating a difference between the size of the pattern based on the data . A hollow graphic generated by combining a plurality of graphic elements in each data block is expressed by a graphic element that defines the outer periphery and a graphic element that defines the hollow part, and these graphic elements and the hollow elements that define the outer periphery. predicting the size of the pattern to be transferred to the wafer graphic element that defines the part as a correction target, correcting the design data by calculating the difference between the dimensions of the pattern due to dimensional and design data of the predicted pattern The pattern forming method according to claim 1.   2. The pattern forming method according to claim 1, wherein when the modified illumination method is applied to the exposure apparatus, a light shielding pattern having a size equal to or smaller than a period corresponding to a cutoff frequency of three-beam interference is excluded from correction.   When the modified illumination method is applied to the exposure apparatus, a certain correction amount is applied to sides facing each other at intervals of a size equal to or smaller than the predetermined value and equal to or larger than a period corresponding to the cutoff frequency of the three-beam interference. The pattern forming method according to claim 5. After creating an optical projection image from a mask pattern based on the design data, a mask edge is predicted by a position having a predetermined threshold light intensity in the optical projection image,
The pattern forming method according to claim 1, wherein the design data is corrected using a distance between the predicted mask edge and the mask edge of the mask pattern based on the design data as a correction amount .   14. The pattern forming method according to claim 13, wherein the threshold value is adjusted according to the light intensity at the periphery of the side of the graphic element to be corrected.   14. The pattern forming method according to claim 13, wherein the threshold value is adjusted in accordance with a two-dimensional light intensity distribution around the side to be corrected.   14. The pattern forming method according to claim 13, wherein the threshold value is adjusted in accordance with the presence / absence of a graphic element adjacent to the side to be corrected.   14. The pattern forming method according to claim 13, wherein the threshold value is adjusted according to the light intensity of the optical image at the time of defocusing with respect to the light intensity of the optical image at the time of best focus.   14. The pattern forming method according to claim 13, wherein the threshold value is adjusted according to the reflectance of the base substrate to which the pattern is transferred.   14. The pattern forming method according to claim 13, wherein the threshold value is adjusted in accordance with a step formed on the base substrate onto which the pattern is transferred.   14. The pattern forming method according to claim 13, wherein the threshold value is adjusted according to halation occurring on the surface of the base substrate to which the pattern is transferred.   14. The pattern according to claim 13, wherein the threshold value is adjusted based on a correlation between a resist pattern obtained by actually exposing the standard pattern and an optical image of the standard pattern calculated from the standard mask data. Forming method.   Based on the correlation between the etching pattern obtained by actually exposing and etching the standard pattern and the optical image of the standard pattern calculated from the standard mask data, and the etching pattern calculated from this optical image. The pattern forming method according to claim 13, wherein the pattern is adjusted.   14. The threshold value is adjusted based on a correlation between a standard pattern actually obtained by electron beam drawing of standard mask data and an electron beam drawing pattern calculated from the standard mask data. Pattern formation method.   14. The pattern according to claim 13, wherein the threshold value is adjusted based on a correlation between a standard pattern actually obtained by laser drawing of the standard mask data and a laser drawing pattern calculated from the standard mask data. Forming method.   The threshold value is adjusted based on a correlation between a resist pattern obtained by actually exposing a standard pattern and an optical image calculated from a pattern size obtained by measuring the standard pattern. The pattern formation method of Claim 13.   There is a correlation between the etching pattern obtained by actually exposing a standard pattern to form a resist pattern and then etching the resist pattern, and the etching pattern calculated from the pattern size obtained by measuring the resist pattern. 14. The pattern forming method according to claim 13, wherein the threshold value is adjusted based on the threshold value.   Calculate the electron beam drawing pattern from the resist pattern obtained by exposing this standard pattern after forming the standard pattern by actually writing the standard mask data with the electron beam, and then calculating from this electron beam drawing pattern. 14. The pattern forming method according to claim 13, wherein the threshold value is adjusted based on the correlation with the optical image.   Standard mask data is actually laser-drawn to form a standard pattern, and then a resist pattern obtained by exposing the standard pattern. After calculating the laser drawing pattern from the standard mask data, the optical calculated from the laser drawing pattern 14. The pattern forming method according to claim 13, wherein the threshold value is adjusted based on the correlation with the image.   There is a correlation between the etching pattern obtained by actually exposing and etching the standard pattern, and the etching pattern calculated from this optical image after calculating the optical image from the pattern size obtained by measuring the standard pattern. 14. The pattern forming method according to claim 13, wherein the threshold value is adjusted based on the threshold value.   The standard mask data is actually drawn with an electron beam to form a standard pattern, and then the standard pattern is exposed and etched, and the electron beam drawing pattern is calculated from the standard mask data and the electron beam drawing. 14. The pattern forming method according to claim 13, wherein after the optical image is calculated from the pattern, the threshold value is adjusted based on the correlation with the etching pattern calculated from the optical image.   The standard mask data is actually laser-drawn to form a standard pattern, and then the standard pattern is exposed and etched, and the laser drawing pattern is calculated from the standard mask data, and the optical pattern is calculated from the laser drawing pattern. 14. The pattern forming method according to claim 13, wherein after the image is calculated, the threshold value is adjusted based on the correlation with the etching pattern calculated from the optical image. After creating an optical projection image from the mask pattern based on the design data, a mask edge is estimated from the optical projection image with a predetermined light intensity as a threshold value,
After adjusting the threshold from the inclination of the optical projection image at this mask edge, predict the mask edge by the adjusted threshold,
The pattern forming method according to claim 13, wherein the size of the pattern transferred to the wafer is predicted based on the predicted mask edge. Perform development simulation of resist on wafer surface,
The pattern forming method according to claim 1, wherein the size of the pattern transferred from the edge position is predicted. Convert the optical projection image into the development time distribution of the resist on the wafer surface,
Perform pseudo-development by integrating this development time one-dimensionally,
The pattern forming method according to claim 1, wherein the size of the pattern transferred from the edge position is predicted.   When creating an optical projection image, the light intensity I at the point P (x, y) inside the rectangle surrounded by the four points Pi (xi, yi) from the light intensity Ii at the four points Pi (xi, yi) is set as I = 1-4. The pattern forming method according to claim 1, wherein interpolation is performed by Σi (Wi · Ii), Wi = (1− | xi−x |) (1− | yi−y |).   The design data is corrected by dividing a side to be corrected into a plurality of line segments, and generating two pieces of data corresponding to both line segments sharing the division point at each division point. 1 pattern formation method.   2. The pattern forming method according to claim 1, wherein when the design data is corrected, each side is corrected in a direction perpendicular to the side.   2. The pattern forming method according to claim 1, wherein when the design data is corrected, when the inclination of the projected image on the side of the graphic element is not more than a predetermined value, the side is excluded from the correction target.   2. The pattern forming method according to claim 1, wherein when the design data is corrected, when the light intensity of the projected image on the side of the graphic element is not more than a predetermined value, the side is excluded from the correction target.   2. The pattern forming method according to claim 1, wherein when the design data is corrected, an upper limit value of the correction amount is provided, and when the calculated correction amount exceeds the upper limit value, correction is performed using the upper limit value as the correction amount. When correcting the design data, after correcting all the sides to be corrected independently of each other, it is determined whether the calculated difference is within the allowable range, and if it is out of the allowable range, a projection image is formed again based on the correction data. Then , the size of the pattern transferred to the wafer is predicted based on the projected image, the difference between the predicted pattern size and the pattern size based on the design data is calculated, and the correction data is further corrected to be predicted. Create an optical projection image based on new correction data, predict the size of the pattern transferred to the wafer, and new correction data until the difference between the pattern size and the pattern size in the design data falls within the allowable range. The pattern forming method according to claim 1, wherein the further correction is repeated. Determine the correction amount for each data block,
42. The pattern forming method according to claim 41 , wherein a data block whose correction amount is outside the allowable range is extracted into another file, and the data is corrected again based on the corrected data. 43. A method of manufacturing an integrated circuit, wherein the pattern forming method according to any one of claims 1 to 42 is used.

Images