CN103901713B - Self-adaption optical proximity effect correction method adopting kernel regression technology - Google Patents
Self-adaption optical proximity effect correction method adopting kernel regression technology Download PDFInfo
- Publication number
- CN103901713B CN103901713B CN201410090470.1A CN201410090470A CN103901713B CN 103901713 B CN103901713 B CN 103901713B CN 201410090470 A CN201410090470 A CN 201410090470A CN 103901713 B CN103901713 B CN 103901713B
- Authority
- CN
- China
- Prior art keywords
- opc
- mask
- regression
- observation point
- pattern
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 63
- 230000003287 optical effect Effects 0.000 title claims abstract description 29
- 230000000694 effects Effects 0.000 title claims abstract description 20
- 238000012937 correction Methods 0.000 title claims abstract description 19
- 238000005070 sampling Methods 0.000 claims abstract description 48
- 238000012549 training Methods 0.000 claims description 27
- 238000003384 imaging method Methods 0.000 claims description 26
- 239000013598 vector Substances 0.000 claims description 26
- 238000005457 optimization Methods 0.000 claims description 25
- 238000004519 manufacturing process Methods 0.000 claims description 15
- 238000001259 photo etching Methods 0.000 claims description 15
- 230000003044 adaptive effect Effects 0.000 claims description 9
- 238000001459 lithography Methods 0.000 claims description 9
- 238000012805 post-processing Methods 0.000 claims description 8
- 229920002120 photoresistant polymer Polymers 0.000 claims description 5
- 238000001514 detection method Methods 0.000 claims description 4
- 238000012935 Averaging Methods 0.000 claims description 3
- 238000009499 grossing Methods 0.000 claims description 2
- 230000006978 adaptation Effects 0.000 claims 1
- 230000002093 peripheral effect Effects 0.000 abstract 2
- 238000010586 diagram Methods 0.000 description 4
- 238000012417 linear regression Methods 0.000 description 2
- 238000004088 simulation Methods 0.000 description 2
- 239000002131 composite material Substances 0.000 description 1
- 238000006073 displacement reaction Methods 0.000 description 1
- 238000010894 electron beam technology Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000002349 favourable effect Effects 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012827 research and development Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000002834 transmittance Methods 0.000 description 1
Landscapes
- Preparing Plates And Mask In Photomechanical Process (AREA)
Abstract
The invention provides a self-adaption optical proximity effect correction method adopting a kernel regression technology. The specific process comprises the following steps: establishing an EBOPC (Edge-Based Optical Proximity Effect) database and a PBOPC (Pixel-Based Optical Proximity Effect) database; dividing a mask pattern to be optimized into a plurality of sub mask patterns; determining an observation point in each sub mask pattern; distributing a sub-region for each observation point; carrying out sampling and point taking in a peripheral region of each observation point; calculating an average distance as shown the specification between each observation point and the peripheral mask pattern; generating OPC regression results by using the kernel regression technology; splicing the OPC regression results corresponding to all the sub mask patterns into the OPC regression result corresponding to the whole mask pattern; carrying out post-treatment on the OPC regression result of the whole mask pattern to obtain a final OPC optimized result. According to the self-adaption optical proximity effect correction method, the kernel regression technology is utilized so that the operation efficiency of traditional PBOPC is effectively improved.
Description
Technical Field
The invention relates to a self-adaptive optical proximity effect correction method adopting a kernel regression technology, and belongs to the technical field of photoetching resolution enhancement.
Background
Current large scale integrated circuits are commonly manufactured using photolithographic systems. The mainstream photoetching system at present is a 193nm ArF deep ultraviolet photoetching system, and as a photoetching technology node enters a sub-wavelength range and a deep sub-wavelength range, the interference and diffraction phenomena of light are more obvious, and the imaging quality of the photoetching system is greatly influenced. For this reason, the lithography system must employ resolution enhancement techniques to improve the imaging quality. Optical Proximity Correction (OPC) is an important lithography resolution enhancement technique. OPC techniques are mainly divided into two main categories: rule-based OPC (Rule-based OPC, abbreviated as RBOPC) and Model-based OPC (Model-based OPC, abbreviated as MBOPC). The RBOPC technology corrects mask local patterns such as line positions, line widths, line heads and the like according to a preset rule. Rules according to which the RBOPC is based need to be established according to engineering experience or obtained through experiment and simulation fitting. Although the RBOPC operation efficiency is high, the RBOPC operation efficiency can only compensate the local optical proximity effect, the global optimal solution of the mask optimization problem cannot be obtained, the capability of the RBOPC technology in improving the resolution of a photoetching system is limited, and the RBOPC operation method is generally used for technical nodes of 180nm or more than 150 nm.
Unlike RBOPC, the MBOPC technology performs mathematical modeling on OPC problems based on a physical model or a mathematical model of the imaging process of a lithography system, and converts the OPC problems into mathematical optimization problems. The MBOPC technology adopts a mathematical optimization algorithm to solve the optimization problem and correct a mask pattern, thereby achieving the purpose of improving the resolution and the pattern fidelity of a photoetching system. Based on the different mask dividing methods in the optimization process, the MBOPC can be further divided into Edge-based OPC (Edge-based OPC, abbreviated as EBOPC) and Pixel-based OPC (Pixel-based OPC, abbreviated as PBOPC). EBOPC divides the mask edge into several segments, and cycles through the optimization of the location of each segment. The PBOPC firstly divides the mask into a plurality of pixels, and then optimizes the whole mask by optimizing the transmittance of each pixel. Compared with EBOPC, PBOPC has higher optimization freedom, can generate necessary auxiliary patterns around the mask main body patterns, and is more favorable for improving the imaging resolution and the pattern fidelity of a photoetching system. PBOPC is therefore commonly used for fine correction of mask critical areas (Hotspot) and has been extensively studied by related researchers and researchers at home and abroad.
With the continuous extension of lithography nodes, the mask size is continuously enlarged, and the mask pattern density is also continuously improved, so that the simulation data volume of PBOPC is greatly increased. How to effectively improve the optimization efficiency becomes one of the key problems in the research and development of the PBOPC method. On the other hand, the mask manufacturing process is an important part of the whole integrated circuit manufacturing process, and thus the mask manufacturability is of interest to the academic and industrial circles at present. Mask manufacturability in this context refers to the cost of manufacture of the mask. Based on the given OPC optimization results, the mask pattern is divided into several non-overlapping trapezoids. Then, a Variable Shaped Beam (VSB) mask writer writes the trapezoids one by one on the mask plate using an electron Beam. Therefore, in the mask division pattern, the smaller the number of trapezoids, the shorter the mask writing time, and the lower the cost. Since PBOPC optimizes all pixels in the mask pattern and adds an auxiliary pattern around the mask body pattern, the complexity of the optimized mask pattern and the total number of trapezoids in the mask division pattern are greatly increased, thereby greatly increasing the manufacturing cost of the mask. In contrast, the EBOPC mask is configured to move only the respective segments of the mask pattern edge, and the total number of trapezoids in the EBOPC mask divided pattern is small, and the mask manufacturing cost is low.
In summary, PBOPC has higher compensation accuracy for the optical proximity effect, but the operation efficiency is low, and the optimized mask has higher complexity. Therefore, how to effectively improve the operation efficiency of the PBOPC algorithm for large-area mask patterns and effectively improve the manufacturability of optimized masks while ensuring the imaging performance of the lithography system is one of the hot problems in the current OPC method research.
The related literature (a.gu and a.zakhor, IEEE trans.semiconductor Manufacturing21(2), 263-271(2008)) proposes a method for improving the operational efficiency of the EBOPC by using a linear regression technique. However, the linear regression technique adopted by the method is only suitable for the EBOPC optimization problem with a low dimension, and is not suitable for the PBOPC optimization problem. While the above method only considers how to increase the speed of the EBOPC algorithm and does not consider how to further improve the manufacturability of the mask. Therefore, the existing method cannot more effectively improve the operation efficiency of the PBOPC method and the manufacturability of the mask after optimization.
Disclosure of Invention
The invention aims to provide an adaptive optical proximity effect correction method adopting a kernel regression technology. The method can effectively improve the operation efficiency of the PBOPC algorithm, and effectively improve the manufacturability of the optimized mask while ensuring the imaging performance of the photoetching system.
The technical scheme for realizing the invention is as follows:
a self-adaptive optical proximity effect correction method adopting a kernel regression technology comprises the following specific steps:
step 101, establishing an EBOPC database and a PBOPC database;
step 102, dividing a mask pattern to be optimized into a plurality of sub-mask patterns, wherein the width w is formed between every two adjacent sub-mask patternsoverlapThe overlapping area of (a);
step 103, respectively determining the observation points in each sub-mask pattern in step 102, and marking the determined observation points as OkWherein the observation points in the sub-mask patterns comprise convex angle vertexes, concave angle vertexes and observation points on the edges of the mask patterns;
step 104, for each observation point O in step 103kAllocating a sub-region MapkEach subarea only contains one observation point;
step 105, for each observation point OkSampling points in the surrounding area, and arranging the corresponding pixel values of each sampling point into a vector in sequenceWherein,a real vector space representing N × 1, N being the number of sample points for each observation point;
step 106, calculating each observation point OkAverage distance from its surrounding mask pattern(ii) a If it isThen in step 107 a kernel regression is performed using the PBOPC database, otherwise in step 107 a kernel regression is performed using the EBOPC database, with threshold representing a predetermined threshold. the larger the threshold is, the simpler the regressed mask pattern is, whereas the more complicated the regressed mask pattern is.
Step 107, for each observation point OkUsing kernel regression techniques, based on said vectorsSelecting a priori OPC optimization results from the database selected in step 106 for weighted averaging to generate a weighted average corresponding to observation point OkAnd (4) OPC regression result of (1), and finding point OkThe OPC regression result is filled into the corresponding sub-region MapkThereby stitching one OPC regression result for each sub-mask pattern;
step 108, removing the OPC regression result corresponding to each sub-mask pattern, wherein the periphery width of the OPC regression result is woverlapAnd stitching the OPC regression results corresponding to all the sub-mask patterns to correspond to the whole maskOPC regression results of the model graphs;
and step 109, performing post-processing on the OPC regression result of the whole mask pattern obtained in step 108, and taking the finally obtained OPC pattern as a final OPC optimization result.
The specific steps of establishing the EBOPC database and the PBOPC database in the step 101 of the invention are as follows:
step 201, selecting a region from a full-chip mask as a training mask pattern;
step 202, performing OPC optimization on the training mask pattern to respectively obtain a PBOPC optimized pattern and an EBOPC optimized pattern corresponding to the training mask pattern;
step 203, finding the observation point in the training mask pattern, and marking the found observation point as OiWherein the observation points within the training mask pattern include convex vertices, concave vertices, and observation points on the edges of the training mask pattern;
step 204, for each observation point OiSampling points in the surrounding area, and arranging the corresponding pixel values of each sampling point into a vector in sequenceWherein N is the number of sampling points for each observation point;
step 205, train each observation point O on the mask patterniFor the center, a pattern with the size of M × M is cut out from the EBOPC optimized pattern corresponding to the training mask pattern and is marked asIntercepting a pattern with a size of M × M in the PBOPC optimized pattern corresponding to the training mask pattern, and recording the intercepted pattern as the pattern
Step 206, for each view on the training maskMeasure points, establish vectorsAndin a one-to-one correspondence relationship ofStoring the data into an EBOPC database to realize the establishment of the EBOPC database; establishing vectorsAndin a one-to-one correspondence relationship ofAnd storing the data into a PBOPC database to realize the establishment of the PBOPC database.
Step 104 of the present invention provides each observation point OkAllocating a sub-region MapkThe method comprises the following specific steps:
step 301, allocating a square initial sub-region which takes the observation point as the center and CD as the side length to each convex angle vertex, concave angle vertex and edge observation point, wherein CD is the minimum line width in the target circuit graph at the wafer;
step 302, for each edge observation point, respectively expanding the length of the corresponding square initial sub-region to both sides (and both sides are along the edge direction) at the same expansion speed along the edge where the edge observation point is located until the sub-region meets the sub-regions of other observation points, wherein the width of the sub-region corresponding to the edge observation point is kept as CD;
and step 303, expanding the corresponding sub-region of each convex angle vertex, each concave angle vertex and each edge observation point in all surrounding directions at the same expansion speed until the sub-region meets the sub-regions of other observation points or the expansion distance reaches a preset upper limit value.
For each observation point O in steps 105 and 204 described in the present inventionkOr OiThe specific process of sampling and point taking in the surrounding area comprises the following steps:
step 401, observe point OkOr OiCentered on c ×αjnm is a plurality of concentric circles with the radius, the diameter of the largest circle in the plurality of concentric circles is larger than the optical proximity effect distance of the photoetching system, wherein c and α are preset parameters, and j is 1,2,3 …;
step 402, at observation point OkOr OiTaking 1 sampling point at OkOr OiTaking 8 sampling points on each concentric circle as the circle center, wherein the 8 sampling points and OkOr OiThe included angles between the connecting line and the x axis are respectively 0 degree, 45 degrees, 90 degrees, 135 degrees, 180 degrees, 225 degrees, 270 degrees and 315 degrees;
step 403, arranging the values of each sampling point into a vector in sequence from the center of the circle to the outsideOrWherein the value of the sampling point is the pixel value of the sampled image at the sampling point.
In step 106 of the present invention, each observation point O is calculatedkAverage distance from its surrounding mask patternThe method comprises the following specific steps:
step 501, with each observation point OkAs a starting point, searching for and O in 8 directions with included angles of 0 degrees, 45 degrees, 90 degrees, 135 degrees, 180 degrees, 225 degrees, 270 degrees and 315 degrees with the x axis respectivelykThe nearest sub-mask pattern is set to have a distance in 8 directionsThe values of separation are respectively di,i=1,2,...8;
Step 502, if a certain direction is OkWithin the mask pattern (including edge coincidence), the corresponding distance d in the direction is setiWhen no other mask pattern is found in a certain direction, the distance d corresponding to the direction is set to 0iEqual to an optical proximity effect distance of the lithography system;
step 503, calculate corresponding to OkAverage distance ofWherein N isdIs non-zero diThe number of (2). In step 107 of the present invention, for each observation point OkSelecting prior OPC optimization results from an OPC database to perform weighted average by adopting a kernel regression technology to generate a result corresponding to an observation point OkThe specific steps of the OPC regression result are as follows:
step 601, calculating observation point OkCorresponding sampling point vectorSample point vector corresponding to all prior data in OPC databaseEuler distance therebetween
Step 602, selecting andp of the smallest Euler distanceComputing kernel functionsWherein P is a predetermined coreRegression candidate sample number value, h is the bandwidth of the control smooth range;
step 603, aiming at the P pieces selected in the step 602Computing kernel regression resultsIf the EBOPC database is selected for kernel regression, thenRepresentsIf the EBOPC database is selected for kernel regression, thenRepresents
The specific steps of post-processing the OPC regression result of the whole mask pattern obtained in step 108 in step 109 of the present invention are as follows:
step 701, calculating an image Z in the photoresist according to the OPC regression result of the whole mask pattern obtained in step 108, and recording a portion where Z does not overlap with the target pattern asWill be provided withThe portion of overlap with the OPC regression results obtained in step 108 is noted asRemoving from OPC regression resultsAnd recording the processed OPC regression result as OPC1;
Step 702, indent the edge of the target graph inwards by wsnm, and the reduced target pattern is denoted as T1Subjecting OPC to1And T1The pixel values in all holes in the coincident portion are set to 1, where wsFor the predetermined setback distance, the hole refers to a figure whose central portion has a pixel value of 0 and is surrounded by a closed region having a pixel value of 1. Recording the treated OPC regression result as OPC2;
Step 703, expanding the edge of the target graph outwards by wd1The edge after nm is denoted as contourr1Expanding the edge of the target graph outwards by wd2The edge after nm is denoted as contourr2Will contour1And contourer2The region in between is denoted as T2Wherein w isd1And wd2Is a preset expansion distance; OPC removal2Neutral and T2The pattern portions overlapped and connected with the mask main body pattern are recorded as OPC regression results after the treatment3;
Step 704, adopt mask manufacturing rule detection (MRC) method to OPC3Processed to obtain OPC4Make OPC4The set mask manufacturability conditions are met;
step 705, adopting EBOPC algorithm to carry out OPC4Optimizing to obtain OPC5So that OPC is adopted5The imaging graph obtained by using the mask is closer to the target graph at the edge, so that the imaging requirement at the edge of the graph is met;
step 706, adopting PBOPC algorithm to perform OPC5Optimizing to obtain OPC6So that OPC is adopted6The whole imaging graph obtained by using the mask is closer to the target graph, so that the imaging requirement of the whole graph is met;
step 707 for OPC using mask manufacturing rule detection6To carry outProcessed to obtain OPC7Make OPC7The set mask manufacturability conditions are met.
Advantageous effects
Firstly, the invention effectively improves the operation efficiency of the traditional PBOPC by utilizing the kernel regression technology;
secondly, the invention adopts the self-adaptive method and simultaneously utilizes the advantages of PBOPC and EBOPC, effectively reduces the complexity of a mask pattern, improves the manufacturability of the mask and reduces the manufacturing cost of the mask while improving the imaging performance of a photoetching system.
Drawings
FIG. 1 is a flow chart of a method for adaptive optical proximity correction using kernel regression in accordance with the present invention;
FIG. 2 is a schematic diagram of building EBOPC and PBOPC databases;
FIG. 3 is a schematic diagram of determining observation points of a mask pattern and determining sub-regions corresponding to each observation point;
FIG. 4 is a view for calculating an observation point OkAverage distance from its surrounding mask patternA schematic diagram of (a);
FIG. 5 is a schematic illustration of the initial mask pattern, the resulting mask patterns from each step of mask post-processing, and the final in-resist imaging.
Detailed Description
The present invention will be further described in detail with reference to the accompanying drawings.
The principle of the invention is as follows: the operation efficiency of the traditional PBOPC method is effectively improved by adopting a kernel regression technology; meanwhile, two advantages of low complexity of the EBOPC mask and high imaging precision of the PBOPC mask are utilized, and the two OPCs are mixed, so that the imaging performance of a photoetching system is improved, the complexity of a mask pattern is effectively reduced, the manufacturability of the mask is improved, and the manufacturing cost of the mask is reduced.
As shown in fig. 1, the adaptive optical proximity correction method using kernel regression of the present invention specifically includes the following steps:
step 101, respectively establishing an EBOPC database and a PBOPC database by utilizing a large number of prior OPC optimization results;
as shown in fig. 2, the specific steps of establishing the EBOPC database and the PBOPC database in step 101 of the present invention are as follows:
step 201, selecting a proper area from a full-chip mask as a training mask pattern;
step 202, performing OPC optimization on the training mask pattern to respectively obtain a PBOPC optimized pattern and an EBOPC optimized pattern corresponding to the training mask pattern;
step 203, finding the observation point in the training mask pattern, and marking the found observation point as OiWherein the observation points within the training mask pattern include convex vertices, concave vertices, and observation points on the edges of the training mask pattern; wherein the observation points on the edge are taken at intervals on the edge of the mask pattern.
Step 204, for each observation point OiSampling points in the surrounding area, and arranging the corresponding pixel values of each sampling point into a vector in sequenceWherein N is the number of sampling points for each observation point;
step 205, train each observation point O on the mask patterniFor the center, a pattern with the size of M × M is cut out from the EBOPC optimized pattern corresponding to the training mask pattern and is marked asThe pattern with size M × M is cut out from the PBOPC optimized pattern corresponding to the training mask and recorded as
Step 206, for each observation point on the training mask, a vector is establishedAndin a one-to-one correspondence relationship ofStoring the data into an EBOPC database to realize the establishment of the EBOPC database; establishing vectorsAndin a one-to-one correspondence relationship ofAnd storing the data into a PBOPC database to realize the establishment of the PBOPC database.
As shown in FIG. 201, for each observation point O in step 204 described in the present inventioniThe specific process of sampling and point taking in the surrounding area comprises the following steps:
step 401, observe point OiCentered on c ×αjnm is a series of concentric circles with radius, the diameter of the largest circle in the series should be larger than the optical proximity effect distance of the lithography system, wherein c and α can be set according to practical situations, αjIs an expansion factor of the radius of the concentric circle, so that the radius of the concentric circle of the outer layer is exponentially increased compared with the radius of the concentric circle of the inner layer, and j is 1,2,3…。
Step 402, at observation point OiTake 1 sample point. In the presence of OiTaking 8 sampling points on each concentric circle as the center of circle, wherein the 8 sampling points and OiThe included angles between the connecting line and the x axis are respectively 0 degrees, 45 degrees, 90 degrees, 135 degrees, 180 degrees, 225 degrees, 270 degrees and 315 degrees, wherein the direction of the x axis is horizontal to the right.
Step 403, arranging the values of each sampling point into a vector in sequence from the center of the circle to the outsideWherein the values of the sampling points are the pixel values of the sampled pattern at the sampling points.
Step 102, dividing a mask pattern to be optimized into a plurality of sub-mask patterns, wherein the width w is formed between every two adjacent sub-mask patternsoverlapThe overlapping area of (a);
step 103, as shown in fig. 302, separately determining observation points in each sub-mask pattern in step 102, including convex angle vertex, concave angle vertex, and observation points on the edge of the mask pattern, and marking the observation points as Ok;
Step 104, for each observation point O in step 103kAllocating a sub-region MapkEach subarea only contains one observation point;
step 104 of the present invention provides each observation point OkAllocating a sub-region MapkThe method comprises the following specific steps:
step 301, allocating a square initial sub-region which takes the observation point as the center and CD as the side length to each convex angle vertex, concave angle vertex and edge observation point, wherein CD is the minimum line width in the target circuit graph at the wafer;
step 302, as shown in fig. 303, for each edge observation point, the length of the corresponding square initial sub-region is respectively expanded to both sides at the same expansion speed along the edge where the observation point is located until the sub-region meets the sub-regions of other observation points. In this step, the width of the sub-region corresponding to the edge observation point is kept as CD. The graph 303 is the final subregion segmentation map obtained after the step 302.
Step 303, as shown in fig. 304, for each convex angle vertex, concave angle vertex and edge observation point, the corresponding sub-region is expanded to all the surrounding directions at the same expansion speed until the sub-region meets the sub-regions of other observation points or the expansion distance reaches a predetermined upper limit value.
Step 105, for each observation point OkSampling points in the surrounding area, and arranging the corresponding pixel values of each sampling point into a vector in sequenceWherein N is the number of sampling points for each observation point;
for each observation point O in step 105 of the inventionkThe specific process of sampling and sampling points in the surrounding area is the same as the above steps 401 to 403.
Step 106, calculating each observation point OkAverage distance from its surrounding mask pattern(ii) a If it isThen the PBOPC database is used for kernel regression in step 107, otherwise the EBOPC database is used for kernel regression in step 107, where the symbol threshold represents a predetermined threshold, and the larger the threshold, the simpler the mask pattern after regression, and vice versa, the more complex the mask pattern after regression.
As shown in FIG. 4, each observation point O is calculated in step 106kAverage distance from its surrounding mask patternThe method comprises the following specific steps:
step 501, with each observation point OkAs a starting point, searching and observing points O in 8 directions with included angles of 0 degrees, 45 degrees, 90 degrees, 135 degrees, 180 degrees, 225 degrees, 270 degrees and 315 degrees with the x axis respectivelykThe nearest mask pattern is set to have a distance value d in each of the 8 directionsi,i=1,2,...8;
Step 502, if a certain direction is associated with the observation point OkThe edges of the mask patterns are overlapped, and the distance d corresponding to the direction is setiWhen no other mask pattern is found in a certain direction, the distance d corresponding to the direction is set to 0iEqual to an optical proximity effect distance of the lithography system;
step 503, calculate corresponding to OkAverage distance ofWherein N isdIs non-zero diThe number of (2).
Step 107, for each observation point OkUsing kernel regression techniques, based on said vectorsSelecting a priori OPC optimization results from the database selected in step 106 for weighted averaging to generate a weighted average corresponding to observation point OkAnd (4) OPC regression result of (1), and finding point OkThe OPC regression result is filled into the corresponding sub-region MapkThereby stitching one OPC regression result for each sub-mask pattern;
in step 107 of the present invention, for each observation point OkSelecting prior OPC optimization results from an OPC database to perform weighted average by adopting a kernel regression technology to generate a result corresponding to an observation point OkThe specific steps of the OPC regression result are as follows:
step 601, calculating observation point OkCorresponding sampling point vectorSample point vector corresponding to all prior data in OPC databaseEuler distance therebetween
Step 602, selecting andp of the smallest Euler distanceComputing kernel functionsWherein P is a predetermined kernel regression candidate sample number value, and h is a bandwidth for controlling a smoothing range;
step 603, aiming at the P pieces selected in the step 602Computing kernel regression resultsIf the EBOPC database is selected for kernel regression, thenRepresentsIf the EBOPC database is selected for kernel regression, thenRepresents
Step 108, removing the OPC regression result corresponding to each sub-mask pattern, wherein the periphery width of the OPC regression result is woverlapAnd splicing the OPC regression results corresponding to all the sub-mask patterns into OPC regression results corresponding to the whole mask pattern;
step 109, post-processing the OPC regression result of the whole mask pattern obtained in step 108, further optimizing the imaging result, improving mask manufacturability, and using the finally obtained OPC pattern as the final OPC optimization result.
The specific steps of post-processing the OPC regression result of the whole mask pattern obtained in step 108 in step 109 of the present invention are as follows:
step 701, calculating an image Z in the photoresist by using the OPC regression result of the whole mask pattern obtained in step 108 as input data of commercial software, and recording a portion where Z does not overlap with the target pattern asWill be provided withThe portion of overlap with the OPC regression results obtained in step 108 is noted asRemoving from OPC regression resultsGuo is divided, and the processed OPC regression result is recorded as OPC1;
Step 702, indent the edge of the target graph inwards by wsnm, and the reduced target pattern is denoted as T1Subjecting OPC to1And T1The pixel values in all holes in the coincident portion are set to 1, where wsFor the preset retraction distance, the hole refers to the pixel value of the central portionA figure of 0 and surrounded by a closed region having a pixel value of 1. Recording the treated OPC regression result as OPC2。
Step 703, expanding the edge of the target graph outwards by wd1The edge after nm is denoted as contourr1Expanding the edge of the target graph outwards by wd2The edge after nm is denoted as contourr2Will contour1And contourer2The region in between is denoted as T2Wherein w isd1And wd2Is a preset expansion distance. OPC removal2Neutral and T2The pattern portions overlapped and connected with the mask main body pattern are recorded as OPC regression results after the treatment3;
Step 704, adopt mask manufacturing rule detection (MRC) method to OPC3Processed to obtain OPC4Make OPC4The set mask manufacturability conditions are met; the MRC processing performed here is to correct geometric structures such as the minimum width of the mask pattern, the minimum pitch between adjacent patterns, and the pattern edge protrusion by the MRC module, in order to make the corrected mask pattern more regular and able to be manufactured by an actual mask writer. The manufacturability conditions should be specified in the industry during mask writing, which is related to the mask manufacturing process and the model of the mask writer.
Step 705, adopting EBOPC algorithm to carry out OPC4Optimizing to obtain OPC5So that OPC is adopted5The imaging graph obtained by using the mask is closer to the target graph at the edge, so that the imaging requirement at the edge of the graph is met; there are many EBOPC algorithms that can be used in this step, and any one of them can be used to achieve the above purpose, but the EBOPC algorithm is different, and the specific operation is different, depending on the specific algorithm and the software used. Therefore, there is no rigid requirement in this step, to what extent it is close to depending on the technology node, the layer on which the mask is placed and the specific process, which specific requirements are given by the industry
Step 706, adopting PBOPC algorithm to perform OPC5Optimizing to obtain OPC6So that OPC is adopted6The whole imaging graph obtained by using the mask is closer to the target graph, so that the imaging requirement of the whole graph is met; the situation in this step is the same as step 705.
Step 707, adopt MRC method to OPC6Processed to obtain OPC7Make OPC7The set mask manufacturability conditions are met. The situation in this step is the same as in step 704.
Example of implementation of the invention:
fig. 5 is a schematic diagram showing the initial mask pattern, the resulting mask patterns from each mask post-processing step, and the final imaging in the photoresist. 501 is an initial mask pattern and is also an OPC optimized target pattern; 502 is the OPC regression results of the mask patterns obtained in step 108; 503 is the processed OPC regression result obtained in step 704, OPC4(ii) a 504 is the processed OPC regression result obtained in step 7055(ii) a 505 is the processed OPC regression result obtained in step 7077(ii) a 506 is by OPC7The result of imaging in the photoresist obtained as a mask pattern.
Table 1 lists the performance indicators of the method compared with the initial mask pattern and the professional software PBOPC module, including the average edge displacement error (EPE) of the image, the operation time and the total number of trapezoids in the segmented pattern of the optimized mask. As can be seen from the data in Table 1, the present invention utilizes the kernel regression technique to effectively improve the operation efficiency of the conventional PBOPC. Meanwhile, the invention adopts the self-adaptive method and simultaneously utilizes the advantages of PBOPC and EBOPC, effectively reduces the complexity of a mask pattern, improves the manufacturability of the mask and reduces the manufacturing cost of the mask while improving the imaging performance of a photoetching system.
TABLE 1 comparison of method with initial mask and professional software PBOPC Performance
Initial mask | Professional software PBOPC | Method for producing a composite material | |
Average EPE (nm) | 14.6 | 4.0 | 6.4 |
Operation time(s) | - | 587 | 302 |
Complexity of mask | 1244 | 11400 | 6731 |
Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, it will be understood that many variations, substitutions and modifications may be made by those skilled in the art without departing from the principles of the invention and these are intended to be within the scope of the invention.
Claims (8)
1. A self-adaptive optical proximity effect correction method adopting a kernel regression technology is characterized by comprising the following specific steps:
step 101, establishing an EBOPC database and a PBOPC database;
step 102, dividing a mask pattern to be optimized into a plurality of sub-mask patterns, wherein the width w is formed between every two adjacent sub-mask patternsoverlapThe overlapping area of (a);
step 103, respectively determining the observation points in each sub-mask pattern in step 102, and marking the determined observation points as OkWherein the observation points in the sub-mask patterns comprise convex angle vertexes, concave angle vertexes and observation points on the edges of the mask patterns;
step 104, for each observation point O in step 103kAllocating a sub-region MapkEach subarea only contains one observation point;
step 105, for each observation point OkSampling points in the surrounding area, and arranging the corresponding pixel values of each sampling point into a vector in sequence
Step 106, calculating each observation point OkAverage distance from its surrounding mask patternIf it isPerforming kernel regression by using a PBOPC database in step 107, or performing kernel regression by using an EBOPC database in step 107, wherein a symbol threshold represents a predetermined threshold;
step 107, for each observation point OkUsing kernel regression techniques, based on said vectorsSelecting a priori OPC optimization results from the database selected in step 106 for weighted averaging to generate a weighted average corresponding to observation point OkAnd (4) OPC regression result of (1), and finding point OkThe OPC regression result is filled into the corresponding sub-region MapkThereby stitching one OPC regression result for each sub-mask pattern;
step 108, removing the OPC regression result corresponding to each sub-mask pattern, wherein the periphery width of the OPC regression result is woverlapAnd splicing the OPC regression results corresponding to all the sub-mask patterns into OPC regression results corresponding to the whole mask pattern;
and step 109, performing post-processing on the OPC regression result of the whole mask pattern obtained in step 108, and taking the finally obtained OPC pattern as a final OPC optimization result.
2. The adaptive optical proximity correction method using kernel regression technology according to claim 1, wherein the specific steps of establishing the EBOPC database and the PBOPC database in step 101 are as follows:
step 201, selecting a region from a full-chip mask as a training mask pattern;
step 202, performing OPC optimization on the training mask pattern to respectively obtain a PBOPC optimized pattern and an EBOPC optimized pattern corresponding to the training mask pattern;
step 203, finding the observation point in the training mask pattern, and marking the found observation point as OiWherein the observation points within the training mask pattern include convex vertices, concave vertices, and observation points on the edges of the training mask pattern;
step 204, for each observation point OiSampling points in the surrounding area, and arranging the corresponding pixel values of each sampling point into a vector in sequence
Step 205, train each observation point O on the mask patterniFor the center, a pattern with the size of M × M is cut out from the EBOPC optimized pattern corresponding to the training mask pattern and is marked asIntercepting a pattern with a size of M × M in the PBOPC optimized pattern corresponding to the training mask pattern, and recording the intercepted pattern as the pattern
Step 206, for each observation point on the training mask, a vector is establishedAndin a one-to-one correspondence relationship ofStoring the data into an EBOPC database to realize the establishment of the EBOPC database; establishing vectorsAndin a one-to-one correspondence relationship ofAnd storing the data into a PBOPC database to realize the establishment of the PBOPC database.
3. The adaptive optical proximity correction method using kernel regression as claimed in claim 2 wherein step 104 is performed for each observation point OkAllocating a sub-region MapkThe method comprises the following specific steps:
step 301, allocating a square initial sub-region which takes the observation point as the center and CD as the side length to each convex angle vertex, concave angle vertex and edge observation point, wherein CD is the minimum line width in the target circuit graph at the wafer;
step 302, for each edge observation point, respectively expanding the length of the corresponding square initial sub-region to both sides along the edge of the edge observation point at the same expansion speed until the sub-region meets the sub-regions of other observation points, wherein the width of the sub-region corresponding to the edge observation point is kept as CD;
and step 303, expanding the corresponding sub-region of each convex angle vertex, each concave angle vertex and each edge observation point in all surrounding directions at the same expansion speed until the sub-region meets the sub-regions of other observation points or the expansion distance reaches a preset upper limit value.
4. The adaptive optical proximity correction method using kernel regression technique as claimed in claim 3 wherein said step 204 is performed for each observation point OiThe specific process of sampling and point taking in the surrounding area comprises the following steps:
s41 finding the point OiCentered on c ×αjnm is a plurality of concentric circles with the radius, the diameter of the largest circle in the plurality of concentric circles is larger than the optical proximity effect distance of the photoetching system, wherein c and α are preset parameters, and j is 1,2,3 …;
s42 at observation point OiTaking 1 sampling point at OiTaking 8 sampling points on each concentric circle as the center of circle, wherein the 8 sampling points and OiThe included angles between the connecting line and the x axis are respectively 0 degree, 45 degrees, 90 degrees, 135 degrees, 180 degrees, 225 degrees, 270 degrees and 315 degrees;
s43, arranging the values of each sampling point into a vector in sequence from the center of a circle to the outsideWherein the values of the sampling points are the pixel values of the sampled pattern at the sampling points.
5. The adaptive optical proximity correction method using kernel regression technology according to claim 1 or 4, wherein step 105 is performed for each observation point OkThe specific process of sampling and point taking in the surrounding area comprises the following steps:
step 401, observe point OkCentered on c ×αjnm is a plurality of concentric circles with the radius, the diameter of the largest circle in the plurality of concentric circles is larger than the optical proximity effect distance of the photoetching system, wherein c and α are preset parameters, and j is 1,2,3 …;
step 402, at observation point OkTaking 1 sampling point at OkTaking 8 sampling points on each concentric circle as the center of circle, wherein the 8 sampling points and OkThe included angles between the connecting line and the x axis are respectively 0 degree, 45 degrees, 90 degrees, 135 degrees, 180 degrees, 225 degrees, 270 degrees and 315 degrees;
step 403, arranging the values of each sampling point into a vector in sequence from the center of the circle to the outsideWherein the values of the sampling points are the pixel values of the sampled pattern at the sampling points.
6. The adaptive optical proximity correction method using kernel regression technique according to claim 1 or 4, wherein each observation point O is calculated in step 106kAverage distance from its surrounding mask patternThe method comprises the following specific steps:
step 501, with each observation point OkAs a starting point, searching for and O in 8 directions with included angles of 0 degrees, 45 degrees, 90 degrees, 135 degrees, 180 degrees, 225 degrees, 270 degrees and 315 degrees with the x axis respectivelykThe sub-mask patterns closest to each other are set to have respective distances d in the 8 directionsi,i=1,2,...8;
Step 502, if a certain direction is OkWithin the mask pattern, let the direction correspond to the distance diWhen no other mask pattern can be searched in a certain direction, the distance d corresponding to the direction is set to 0iEqual to the optical proximity effect distance of the lithography system;
step 503, calculate corresponding to OkAverage distance ofWherein N isdIs non-zero diThe number of (2).
7. The adaptation using kernel regression technique according to claim 1 or 4The optical proximity correction method is characterized in that, in the step 107, for each observation point OkSelecting prior OPC optimization results from an OPC database to perform weighted average by adopting a kernel regression technology to generate a result corresponding to an observation point OkThe specific steps of the OPC regression result are as follows:
step 601, calculating observation point OkCorresponding sampling point vectorSample point vector corresponding to all prior data in OPC databaseEuler distance therebetween
Step 602, selecting andp of the smallest Euler distanceComputing kernel functionsWherein P is a predetermined kernel regression candidate sample number value, and h is a bandwidth for controlling a smoothing range;
step 603, aiming at the P pieces selected in the step 602Computing kernel regression resultsIf the EBOPC database is selected for kernel regression, thenRepresentsIf the EBOPC database is selected for kernel regression, thenRepresents
8. The adaptive optical proximity correction method using kernel regression technology as claimed in claim 1 or 4, wherein the step 109 of post-processing the OPC regression result of the whole mask pattern obtained in step 108 comprises the following steps:
step 701, calculating an image Z in the photoresist according to the OPC regression result of the whole mask pattern obtained in step 108, and recording a portion where Z does not overlap with the target pattern asWill be provided withThe portion of overlap with the OPC regression results obtained in step 108 is noted asRemoving from OPC regression resultsAnd recording the processed OPC regression result as OPC1;
Step 702, indent the edge of the target graph inwards by wsnm, and the reduced target pattern is denoted as T1Subjecting OPC to1And T1The pixel values in all holes in the coincident portion are set to 1, where wsIs a preset setbackDistance, hole refers to a graph whose central portion has a pixel value of 0 and is surrounded by a closed region having a pixel value of 1, and the processed OPC regression result is recorded as OPC2;
Step 703, expanding the edge of the target graph outwards by wd1The edge after nm is denoted as contourr1Expanding the edge of the target graph outwards by wd2The edge after nm is denoted as contourr2Will contour1And contourer2The region in between is denoted as T2Wherein w isd1And wd2Is a preset expansion distance; OPC removal2Neutral and T2The pattern portions overlapped and connected with the mask main body pattern are recorded as OPC regression results after the treatment3;
Step 704, OPC is performed using a mask manufacturing rule checking method3Processed to obtain OPC4Make OPC4The set mask manufacturability conditions are met;
step 705, adopting EBOPC algorithm to carry out OPC4Optimizing to obtain OPC5So that OPC is adopted5The imaging graph obtained by using the mask is closer to the target graph at the edge, so that the imaging requirement at the edge of the graph is met;
step 706, adopting PBOPC algorithm to perform OPC5Optimizing to obtain OPC6So that OPC is adopted6The whole imaging graph obtained by using the mask is closer to the target graph, so that the imaging requirement of the whole graph is met;
step 707 for OPC using mask manufacturing rule detection6Processed to obtain OPC7Make OPC7The set mask manufacturability conditions are met.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410090470.1A CN103901713B (en) | 2014-03-12 | 2014-03-12 | Self-adaption optical proximity effect correction method adopting kernel regression technology |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410090470.1A CN103901713B (en) | 2014-03-12 | 2014-03-12 | Self-adaption optical proximity effect correction method adopting kernel regression technology |
Publications (2)
Publication Number | Publication Date |
---|---|
CN103901713A CN103901713A (en) | 2014-07-02 |
CN103901713B true CN103901713B (en) | 2017-01-18 |
Family
ID=50993127
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201410090470.1A Active CN103901713B (en) | 2014-03-12 | 2014-03-12 | Self-adaption optical proximity effect correction method adopting kernel regression technology |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN103901713B (en) |
Families Citing this family (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107844033B (en) * | 2017-09-30 | 2020-02-21 | 上海华力微电子有限公司 | Method for correcting global metal layer process hot spots |
CN108875141B (en) * | 2018-05-24 | 2022-08-19 | 上海集成电路研发中心有限公司 | Method for determining chip full-mask focusing parameters based on neural network model |
CN109709762B (en) * | 2018-12-29 | 2022-05-27 | 上海华力集成电路制造有限公司 | OPC correction method of through hole layer |
CN114485916B (en) * | 2022-01-12 | 2023-01-17 | 广州声博士声学技术有限公司 | Environmental noise monitoring method and system, computer equipment and storage medium |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1904726A (en) * | 2005-07-29 | 2007-01-31 | 台湾积体电路制造股份有限公司 | Method and system for designing optical shield layout and producing optical shield pattern |
CN101681093A (en) * | 2007-06-04 | 2010-03-24 | 睿初科技公司 | Methods for performing model-based lithography guided layout design |
CN102122111A (en) * | 2011-03-20 | 2011-07-13 | 北京理工大学 | Pixel-based optimization method for optical proximity correction |
CN103163727A (en) * | 2011-12-12 | 2013-06-19 | 无锡华润上华科技有限公司 | Mask pattern correction method |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7934172B2 (en) * | 2005-08-08 | 2011-04-26 | Micronic Laser Systems Ab | SLM lithography: printing to below K1=.30 without previous OPC processing |
US7975244B2 (en) * | 2008-01-24 | 2011-07-05 | International Business Machines Corporation | Methodology and system for determining numerical errors in pixel-based imaging simulation in designing lithographic masks |
-
2014
- 2014-03-12 CN CN201410090470.1A patent/CN103901713B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1904726A (en) * | 2005-07-29 | 2007-01-31 | 台湾积体电路制造股份有限公司 | Method and system for designing optical shield layout and producing optical shield pattern |
CN101681093A (en) * | 2007-06-04 | 2010-03-24 | 睿初科技公司 | Methods for performing model-based lithography guided layout design |
CN102122111A (en) * | 2011-03-20 | 2011-07-13 | 北京理工大学 | Pixel-based optimization method for optical proximity correction |
CN103163727A (en) * | 2011-12-12 | 2013-06-19 | 无锡华润上华科技有限公司 | Mask pattern correction method |
Also Published As
Publication number | Publication date |
---|---|
CN103901713A (en) | 2014-07-02 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107908071B (en) | Optical proximity correction method based on neural network model | |
US8631361B2 (en) | Integrated circuit design method with dynamic target point | |
US7343582B2 (en) | Optical proximity correction using progressively smoothed mask shapes | |
US6453457B1 (en) | Selection of evaluation point locations based on proximity effects model amplitudes for correcting proximity effects in a fabrication layout | |
CN107844033B (en) | Method for correcting global metal layer process hot spots | |
CN108490735B (en) | Method, apparatus and computer readable medium for full chip mask pattern generation | |
US11726402B2 (en) | Method and system for layout enhancement based on inter-cell correlation | |
US20190187552A1 (en) | Optical proximity correction (opc) method and method of manufacturing mask by using the opc method | |
CN108121151B (en) | Mask modeling method for pattern having arbitrary angle | |
CN103901713B (en) | Self-adaption optical proximity effect correction method adopting kernel regression technology | |
TWI747983B (en) | Method of modeling a mask | |
JP5704808B2 (en) | Lithographic mask manufacturability calculation method and computer readable medium | |
US6853743B2 (en) | Mask pattern correction method, mask pattern creation system using the correction method, and computer-readable recording medium | |
US11972187B2 (en) | Methods for modeling of a design in reticle enhancement technology | |
CN110426914A (en) | A kind of modification method and electronic equipment of Sub-resolution assist features | |
US20220128899A1 (en) | Methods and systems to determine shapes for semiconductor or flat panel display fabrication | |
US9978134B2 (en) | Sampling method and apparatus applied to OPC of lithography layout | |
US6350977B2 (en) | Pattern distortion detecting method and apparatus and recording medium for pattern distortion detection | |
JPH11168065A (en) | Mask data design method | |
US20170168385A1 (en) | Layout Hierachical Structure Defined in Polar Coordinate | |
JP4791020B2 (en) | Simultaneous calculation of multiple points on one or more cut lines | |
US10255397B2 (en) | Methods of rasterizing mask layout and methods of fabricating photomask using the same | |
JP2004128108A (en) | Optimizing method of aperture shape of projection aligner | |
KR20160095284A (en) | method for exposing a electron beam and substrate manufacturing method including the same | |
US20130125070A1 (en) | OPC Checking and Classification |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant |