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
- observation station
- mask
- regression
- mask graph
- 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
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 present invention relates to a kind of adaptive optics proximity effect correction method of employing kernel regression technology, belong to photoetching and differentiate
Rate strengthens technical field.
Background technology
The commonly used etching system of current large scale integrated circuit is manufactured.The etching system of main flow is at present
The arf DUV etching system of 193nm, enters sub-wavelength and deep sub-wavelength scope with photoetching technique node, the interference of light and
Diffraction phenomena is more significantly, the strong influence image quality of etching system.Must be increased using resolution ratio for this etching system
Strong technology, in order to improve image quality.Optical proximity correction technology (optical proximity correction, letter
Claim opc) it is a kind of important photoetching resolution enhancing technology.Opc technology is broadly divided into two big class: rule-based opc technology
(rule-based opc, abbreviation rbopc) and the opc technology (model-based opc, abbreviation mbopc) based on model.
Rbopc technology is modified to the mask local figure such as lines location, live width, the end of a thread according to the rule pre-establishing.Rbopc institute
The rule of foundation need to be formulated according to engineering experience or be drawn according to experiment and emulation matching.Although rbopc operation efficiency is higher,
But it is only capable of the optical proximity effect of local being compensated it is impossible to obtain the globally optimal solution of photomask optimization problem, limit
Ability in terms of improving etching system resolution ratio for the rbopc technology, is generally used for 180nm or more than 150nm technology node.
Different from rbopc, the physical model based on etching system imaging process for the mbopc technology or Mathematical Modeling, opc is asked
Topic carries out mathematical modeling, and opc problem is converted into mathematical optimization problem.It is above-mentioned that mbopc technology adopts optimization algorithm to solve
Optimization problem, revises mask graph, thus reaching the purpose improving etching system resolution ratio and anti-aliasing degree.According to optimizing
The difference of mask partitioning scheme in journey, mbopc can be divided into again opc (edge-based opc, abbreviation ebopc) based on edge and
Opc (pixel-based opc, abbreviation pbopc) based on pixel.Mask edge is divided into some sections by ebopc, circulation
Optimize the position of each section.And mask is divided into some pixels by pbopc first, pass through afterwards to optimize the saturating of each pixel
Cross rate, mask is integrally optimized.Compared with ebopc, pbopc has the higher optimization free degree, and can be in mask master
Produce necessary secondary graphics around volume graphic, be more beneficial for improving imaging resolution and the anti-aliasing degree of etching system.Cause
This pbopc is usually used in mask key position (hotspot) is finely revised, and receives related scholar and research both at home and abroad
Personnel's is widely studied.
With the continuous extension of photoetching technique node, mask size constantly expands, and mask graph density also improves constantly, because
The simulation number of this pbopc is substantially increased.How to effectively improve optimization efficiency becomes the Important Problems of pbopc method research and development
One of.On the other hand, the manufacture process of mask is the important step of whole IC manufacturing flow process, thus the manufacturing of mask
Property for academic circles at present and industrial quarters of interest.Mask manufacturability herein refers to the manufacturing cost of mask.Excellent in given opc
On the basis of changing result, mask graph is divided into the trapezoidal of some non-overlapping copies.Afterwards, shape-variable light beam (variable
Shaped beam, abbreviation vsb) mask inscribed machine and trapezoidal is burnt on mask plate these one by one using electron beam.Therefore covering
In mould partition graph, trapezoidal number is fewer, and the writing time of mask is shorter, and cost is also lower.Because pbopc is in mask graph
All pixels be optimized, and around main body figure increase secondary graphics, cover after therefore significantly increasing optimization
Trapezoidal sum in the complexity of mould figure and mask partition graph, thus be added significantly to the manufacturing cost of mask.With this phase
Ebopc simply moves ratio to each section at mask graph edge, and the trapezoidal sum in ebopc mask partition graph is relatively
Few, fabrication mask cost is relatively low.
In sum, pbopc has higher compensation precision to optical proximity effect, but its operation efficiency is relatively low, and excellent
Mask after change has higher complexity.The computing of pbopc algorithm therefore how is effectively improved for large area mask graph
Efficiency, and effectively improve while guaranteeing etching system imaging performance optimize mask manufacturability be that current opc method is ground
One of hot issue in studying carefully.
Pertinent literature (a.gu and a.zakhor .ieee trans.semiconductor manufacturing21
(2), 263-271 (2008)) propose a kind of method that employing linear regression technique improves ebopc operation efficiency.But more than
The linear regression technique that method is adopted is only applicable to the relatively low ebopc optimization problem of dimension, is not suitable for pbopc optimization problem
In.Said method only accounts for how improving the speed of ebopc algorithm simultaneously, without considering how to improve further mask
Manufacturability.Therefore existing method cannot significantly more efficient raising pbopc method operation efficiency and its optimize after mask can
Manufacturing.
Content of the invention
It is an object of the invention to provide a kind of adaptive optics proximity effect correction method of employing kernel regression technology.The party
Method can effectively improve the operation efficiency of pbopc algorithm, and effectively improves excellent while guaranteeing etching system imaging performance
Change the manufacturability of mask.
Realize technical scheme as follows:
A kind of adaptive optics proximity effect correction method of employing kernel regression technology, concretely comprises the following steps:
Step 101, set up ebopc database and pbopc database；
Step 102, mask graph to be optimized is divided into some sub- mask graphs, between adjacent described sub- mask graph
Having width is woverlapOverlapping region；
The observation station in every sub- mask graph in step 103, respectively determination step 102, and the observation station that will determine
It is designated as ok, the observation station in its neutron mask graph includes the observation station on salient angle summit, re-entrant angle summit and mask graph edge；
Step 104, for each of step 103 observation station okDistribute sub-regions mapk, in each sub-regions only
Comprise an observation station；
Step 105, for each observation station ok, carry out sampling in region about and take a little, and by each sampled point
Respective pixel value is arranged as a vector in orderWherein,Represent n × 1 real number vector space, n be for
The sampled point number of each observation station；
Step 106, calculate each observation station okAverage distance with mask graph about；IfThen
In step 107 kernel regression is carried out using pbopc database, otherwise in step 107 core is carried out using ebopc database and return
Return, threshold represents predetermined threshold value.Mask graph after the more big then recurrence of threshold is simpler, otherwise after returning
Mask graph is more complicated.
Step 107, be directed to each observation station ok, using kernel regression technology, according to described vectorSelected from step 106
Database in select priori opc optimum results be weighted averagely, generation corresponding to observation station okOpc regression result, and
By observation station okOpc regression result be filled into corresponding subregion mapkIn, thus being spliced into for each sub- mask graph
One opc regression result；
Step 108, in the corresponding opc regression result of each sub- mask graph, removing its outer width is woverlap
Fringe region, and the opc that corresponding for all sub- mask graphs opc regression result is spliced into corresponding to overall mask graph returns
Sum up fruit；
Step 109, the opc regression result to the overall mask graph obtaining in step 108 post-process, and will be final
The opc figure obtaining is as final opc optimum results.
Concretely comprising the following steps of ebopc database and pbopc database is set up in step 101 of the present invention:
Step 201, from full chip mask chosen area as training mask graph；
Step 202, opc optimization is carried out to training mask graph, obtain respectively its corresponding pbopc optimize figure and
Ebopc optimizes figure；
Step 203, the observation station found in this training mask graph, and the observation station searching out is designated as oi, wherein instruct
Practice the observation station in mask graph and include the observation station on salient angle summit, re-entrant angle summit and training mask graph edge；
Step 204, for each observation station oi, carry out sampling in region about and take a little, and by each sampled point
Respective pixel value is arranged as a vector in orderWherein n is the sampled point number for each observation station；
Step 205, to train each observation station o on mask graphiCentered on, corresponding to training mask graph
Ebopc optimize figure in intercept size be m × m figure, be designated asPbopc corresponding in training mask graph optimizes
Intercept the figure that size is m × m in figure, be designated as
Step 206, each observation station being directed on training mask, set up vectorWithOne-to-one relationshipIt is stored in ebopc database, realize the foundation of ebopc database；Set up vectorWithOne-to-one relationshipIt is stored in pbopc database, realize the foundation of pbopc database.
Each observation station o is given in step 104 of the present inventionkDistribute sub-regions mapkConcretely comprise the following steps:
Step 301, distribute one centered on this observation station for each salient angle summit, re-entrant angle summit and edge observation station,
Cd is the initial subregion of square of the length of side, and wherein cd is the minimum feature in objective circuit figure at chip；
Step 302, be directed to each edge observation station, along edge residing for described edge observation station, with identical expansion rate
Extend the length (and both sides are along described edge direction) of its corresponding square initial subregion respectively to both sides, until this son
The subregion of region and other observation stations meets, and the width wherein keeping the corresponding subregion of edge observation station is cd；
Step 303, for each salient angle summit, re-entrant angle summit and edge observation station, with identical expansion rate to surrounding
Its corresponding subregion of all Directional Extensions, until this subregion is met with the subregion of other observation stations, or the distance of extension
Reach predetermined higher limit.
For each observation station o in step 105 of the present invention and step 204kOr oi, carry out in region about
Sampling takes the detailed process to be a little:
Step 401, with observation station okOr oiCentered on, set up with c × αjNm is multiple concentric circles of radius, the plurality of
In concentric circles, maximum diameter of a circle is more than the optical proximity effect distance of this etching system, and wherein c and α is ginseng set in advance
Amount, j=1,2,3 ...；
Step 402, in observation station okOr oiPlace takes 1 sampled point, with okOr oiTake 8 on each concentric circles in the center of circle
Individual sampled point, this 8 sampled points and okOr oiLine and x-axis angle be respectively 0 °, 45 °, 90 °, 135 °, 180 °, 225 °,
270 ° and 315 °；
Step 403, the value of each sampled point is arranged in order as a vector according to the order outside from the center of circleOr, the value of wherein sampled point is as sampled the pixel value in sample point for the figure.
Each observation station o is calculated in step 106 of the present inventionkAverage distance with mask graph aboutConcrete
Step is:
Step 501, with each observation station okFor starting point, respectively to x-axis angle be 0 °, 45 °, 90 °, 135 °, 180 °,
8 direction search of 225 °, 270 ° and 315 ° and okClosest sub- mask graph, if the distance value on this 8 directions divides
Wei not di, i=1,2 ... 8；
If certain direction of step 502 is in okIn the mask graph of place (inclusion coincident), then make the corresponding distance of the direction
di=0, if searching in some directions less than other mask graphs, make the direction corresponding apart from diEqual to this etching system
Optical proximity effect distance；
Step 503, calculating are corresponding to okAverage distanceWherein ndFor non-zero diNumber.The present invention
It is directed to each observation station o in described step 107k, using kernel regression technology, select priori opc optimum results from opc database
It is weighted average, generate corresponding to observation station okThe concretely comprising the following steps of opc regression result:
Step 601, calculating observation point okCorresponding sampling point vectorCorresponding with priori datas all in opc database
Sampling point vectorBetween Euler's distance
Step 602, choose withMinimum p of Euler's distanceCalculate kernel function
Wherein p is predetermined kernel regression candidate samples quantitative value, and h is the bandwidth controlling smoothing range；
Step 603, for p choosing in step 602Calculate kernel regression result
If choosing ebopc database to carry out kernel regression,RepresentIf choosing ebopc database to carry out kernel regression,Represent
Place after the opc regression result of the overall mask graph obtaining in step 108 being carried out in step 109 of the present invention
That manages concretely comprises the following steps:
Step 701, the opc regression result according to the overall mask graph obtaining in step 108, calculate in its photoresist
As z, by z, nonoverlapping part is designated as with targeted graphicalWillOverlapping between the opc regression result obtaining in step 108
Part is designated asRemove in opc regression resultPart, and the opc regression result after processing is designated as opc1；
Step 702, targeted graphical edge is inwardly indented wsNm, and the targeted graphical after reducing is designated as t1, by opc1With
t1The pixel value in institute's hole in intersection is set to 1, wherein wsFor indentation distance set in advance, hole refers to central part
Point pixel value is 0, and be closed have the region that pixel value is 1 around figure.Opc regression result after processing is designated as
opc2；
Step 703, targeted graphical edge is expanded outwardly wd1Edge after nm is designated as contour1, by targeted graphical edge
Expand outwardly wd2Edge after nm is designated as contour2, by contour1And contour2Between region be designated as t2, wherein wd1With
wd2For expansion distance set in advance；Remove opc2In with t2Visuals that is overlapping and being connected with main body figure, will
Opc regression result after process is designated as opc3；
Step 704, using fabrication mask rule detection (mask manufacturing rule check, abbreviation mrc) side
Method is to opc3Carry out process and obtain opc4, make opc4Meet the mask manufacturability condition setting；
Step 705, using ebopc algorithm to opc4It is optimized and obtain opc5So that adopting opc5Obtain as mask
Imaging figure is closer to targeted graphical in edge, meets the imaging requirements at pattern edge；
Step 706, using pbopc algorithm to opc5It is optimized and obtain opc6So that adopting opc6Obtain as mask
Overall imaging figure is closer to targeted graphical, meets the imaging requirements of whole figure；
Step 707, using fabrication mask rule detection method to opc6Carry out process and obtain opc7, make opc7Meet and set
Mask manufacturability condition.
Beneficial effect
First, the present invention utilizes kernel regression technology, effectively increases the operation efficiency of traditional pbopc；
Secondly, the present invention make use of the advantage of pbopc and ebopc using adaptive approach simultaneously, is improving etching system
While imaging performance, effectively reduce the complexity of mask graph, improve the manufacturability of mask, reduce the system of mask
Cause this.
Brief description
The flow chart that Fig. 1 adopts the adaptive optics proximity effect correction method of kernel regression technology for the present invention；
Fig. 2 is the schematic diagram setting up ebopc and pbopc database；
Fig. 3 is to determine mask graph observation station, and determines the schematic diagram of the corresponding subregion of each observation station；
Fig. 4 is calculating observation point okAverage distance with mask graph aboutSchematic diagram；
Fig. 5 is initial mask figure, each step mask post-processes in the mask graph obtaining and last photoresist and is imaged
Schematic diagram.
Specific embodiment
The present invention is described in detail further below in conjunction with the accompanying drawings.
The principle of the present invention: effectively improve the operation efficiency of traditional pbopc method using kernel regression technology；Utilize simultaneously
Low two advantages high with pbopc mask imaging precision of ebopc mask complexity, are used in mixed way two kinds of opc, are improving photoetching system
While system imaging performance, effectively reduce the complexity of mask graph, improve the manufacturability of mask, reduce mask
Manufacturing cost.
As shown in figure 1, the present invention adopts the adaptive optics proximity effect correction method of kernel regression technology, concrete steps
For:
Step 101, set up ebopc database and pbopc database respectively using a large amount of priori opc optimum results；
As shown in Fig. 2 setting up the concrete steps of ebopc database and pbopc database in step 101 of the present invention
For:
Step 201, from full chip mask choose appropriate area as training mask graph；
Step 202, opc optimization is carried out to training mask graph, obtain respectively its corresponding pbopc optimize figure and
Ebopc optimizes figure；
Step 203, the observation station found in this training mask graph, and the observation station searching out is designated as oi, wherein instruct
Practice the observation station in mask graph and include the observation station on salient angle summit, re-entrant angle summit and training mask graph edge；Wherein side
On edge, observation station is exactly to adopt an observation station every a segment distance on the edge of mask graph.
Step 204, for each observation station oi, carry out sampling in region about and take a little, and by each sampled point
Respective pixel value is arranged as a vector in orderWherein n is the sampled point number for each observation station；
Step 205, to train each observation station o on mask graphiCentered on, corresponding to training mask graph
It is the figure of m × m that ebopc optimizes intercepting size in figure, is designated asPbopc corresponding in training mask optimizes in figure
Intercept the figure that size is m × m, be designated as
Step 206, each observation station being directed on training mask, set up vectorWithOne-to-one relationshipIt is stored in ebopc database, realize the foundation of ebopc database；Set up vectorWithOne-to-one relationshipIt is stored in pbopc database, realize the foundation of pbopc database.
As shown in Figure 20 1, for each observation station o in step 204 of the present inventioni, carry out in region about
Sampling takes the detailed process to be a little:
Step 401, with observation station oiCentered on, set up with c × αjNm is a series of concentric circles of radius, and some row are with one heart
In circle, maximum diameter of a circle should be greater than the optical proximity effect distance of this etching system, and wherein c and α can set according to actual conditions
Fixed, αjThe expansion factor of exactly one concentric radius of circle is so that the concentric radius of circle of radius ratio internal layer of the concentric circles of outer layer is to refer to
Number form formula increases, j=1, and 2,3 ....
Step 402, in observation station oiPlace takes 1 sampled point.With oiFor 8 sampled points are taken on each concentric circles in the center of circle,
This 8 sampled points and oiLine and x-axis angle be respectively 0 °, 45 °, 90 °, 135 °, 180 °, 225 °, 270 ° and 315 °, wherein
X-axis direction is for level to the right.
Step 403, the value of each sampled point is arranged in order as a vector according to the order outside from the center of circleIts
The value of middle sampled point is as sampled the pixel value in sample point for the figure.
Step 102, mask graph to be optimized is divided into some sub- mask graphs, between adjacent described sub- mask graph
Having width is woverlapOverlapping region；
Step 103, as shown in Figure 30 2, determine the observation station in the every sub- mask graph in step 102 respectively, including
Observation station on salient angle summit, re-entrant angle summit, and mask graph edge, observation station is designated as ok；
Step 104, for each of step 103 observation station okDistribute sub-regions mapk, in each sub-regions only
Comprise an observation station；
Each observation station o is given in step 104 of the present inventionkDistribute sub-regions mapkConcretely comprise the following steps:
Step 301, distribute one centered on this observation station for each salient angle summit, re-entrant angle summit and edge observation station,
Cd is the initial subregion of square of the length of side, and wherein cd is the minimum feature in objective circuit figure at chip；
Step 302, as shown in Figure 30 3, for each edge observation station along edge residing for this observation station, with identical extension
Speed extends the length of its corresponding square initial subregion respectively to both sides, until the son of this subregion and other observation stations
Meet in region.In this step, the width keeping the corresponding subregion of edge observation station is cd.Figure 30 3 is after step 302
The final subregion partition graph obtaining.
Step 303, as shown in Figure 30 4, for each salient angle summit, re-entrant angle summit and edge observation station, expanded with identical
Exhibition speed is to its corresponding subregion of all Directional Extensions around, until this subregion is met with the subregion of other observation stations,
Or the distance of extension reaches predetermined higher limit.
Step 105, for each observation station ok, carry out sampling in region about and take a little, and by each sampled point
Respective pixel value is arranged as a vector in orderWherein n is the sampled point number for each observation station；
For each observation station o in step 105 of the present inventionk, carry out the tool taking a little of sampling in region about
Body process is with above-mentioned steps 401 to step 403.
Step 106, calculate each observation station okAverage distance with mask graph about；IfThen
In step 107 kernel regression is carried out using pbopc database, otherwise in step 107 core is carried out using ebopc database and return
Return, wherein symbol threshold represents predetermined threshold value, the mask graph after the more big then recurrence of threshold is simpler, otherwise
Mask graph after recurrence is more complicated.
As shown in figure 4, calculating each observation station o in step 106kAverage distance with mask graph aboutConcrete
Step is:
Step 501, with each observation station okFor starting point, respectively to x-axis angle be 0 °, 45 °, 90 °, 135 °, 180 °,
8 direction search of 225 °, 270 ° and 315 ° and observation station okClosest mask graph, if the distance on this 8 directions
Value is respectively di, i=1,2 ... 8；
If certain direction of step 502 and observation station okPlace mask graph coincident, then make the direction corresponding apart from di
=0, if searching in some directions less than other mask graphs, make the direction corresponding apart from diLight equal to this etching system
Learn kindred effect distance；
Step 503, calculating are corresponding to okAverage distanceWherein ndFor non-zero diNumber.
Step 107, be directed to each observation station ok, using kernel regression technology, according to described vectorSelected from step 106
Database in select priori opc optimum results be weighted averagely, generation corresponding to observation station okOpc regression result, and
By observation station okOpc regression result be filled into corresponding subregion mapkIn, thus being spliced into for each sub- mask graph
One opc regression result；
It is directed to each observation station o in step 107 of the present inventionk, using kernel regression technology, select from opc database
Priori opc optimum results are weighted averagely, generating corresponding to observation station okThe concretely comprising the following steps of opc regression result:
Step 601, calculating observation point okCorresponding sampling point vectorCorresponding with priori datas all in opc database
Sampling point vectorBetween Euler's distance
Step 602, choose withMinimum p of Euler's distanceCalculate kernel function
Wherein p is predetermined kernel regression candidate samples quantitative value, and h is the bandwidth controlling smoothing range；
Step 603, for p choosing in step 602Calculate kernel regression result
If choosing ebopc database to carry out kernel regression,RepresentIf choosing ebopc database to carry out kernel regression,Represent
Step 108, in the corresponding opc regression result of each sub- mask graph, removing its outer width is woverlap
Fringe region, and the opc that corresponding for all sub- mask graphs opc regression result is spliced into corresponding to overall mask graph returns
Sum up fruit；
Step 109, the opc regression result to the overall mask graph obtaining in step 108 post-process, excellent further
Change its imaging results, improve mask manufacturability, and using the final opc figure obtaining as final opc optimum results.
After the opc regression result of the overall mask graph obtaining in step 108 being carried out in step 109 of the present invention
Process concretely comprises the following steps:
Step 701, using in step 108 obtain overall mask graph opc regression result as business software input
Data, calculates imaging z in its photoresist, nonoverlapping part is designated as with targeted graphical by zWillObtain with step 108
Opc regression result between overlapping part be designated asRemove in opc regression resultGuo divides, and the opc after processing is returned
Sum up fruit and be designated as opc1；
Step 702, targeted graphical edge is inwardly indented wsNm, and the targeted graphical after reducing is designated as t1, by opc1With
t1The pixel value in institute's hole in intersection is set to 1, wherein wsFor indentation distance set in advance, hole refers to central part
Point pixel value is 0, and be closed have the region that pixel value is 1 around figure.Opc regression result after processing is designated as
opc2.
Step 703, targeted graphical edge is expanded outwardly wd1Edge after nm is designated as contour1, by targeted graphical edge
Expand outwardly wd2Edge after nm is designated as contour2, by contour1And contour2Between region be designated as t2, wherein wd1With
wd2For expansion distance set in advance.Remove opc2In with t2Visuals that is overlapping and being connected with main body figure, will
Opc regression result after process is designated as opc3；
Step 704, using fabrication mask rule detection (mask manufacturing rule check, abbreviation mrc) side
Method is to opc3Carry out process and obtain opc4, make opc4Meet the mask manufacturability condition setting；The mrc herein carrying out just is processed
It is that the minimum widith with mrc module correction mask graph, the minimum spacing between adjacent pattern, and pattern edge projection etc. are several
What structure is it is therefore an objective to make revised mask graph more regular, and can be manufactured by actual mask CD writer.Can make
The property made condition should have industrial quarters to specifically give when mask imprinting, this and mask-making technology, and mask CD writer model is all
There is relation.
Step 705, using ebopc algorithm to opc4It is optimized and obtain opc5So that adopting opc5Obtain as mask
Imaging figure is closer to targeted graphical in edge, meets the imaging requirements at pattern edge；Spendable in this step
Ebopc algorithm has a lot, can reach object above using any ebopc algorithm, but ebopc algorithm is different, tool
The operation of body is different, and this depends on specific algorithm and the software being used.Therefore, then no one of this step is rigid will
Ask, be close to which kind of degree is relevant with specific technique with the layer residing for residing technology node, mask, industrial quarters can provide tool
Body requires)
Step 706, using pbopc algorithm to opc5It is optimized and obtain opc6So that adopting opc6Obtain as mask
Overall imaging figure is closer to targeted graphical, meets the imaging requirements of whole figure；Situation in this step and step 705
Identical.
Step 707, using mrc method to opc6Carry out process and obtain opc7, make opc7Meet the mask manufacturability setting
Condition.Situation in this step is identical with step 704.
The embodiment of the present invention:
It is illustrated in figure 5 initial mask figure, each step mask post-processes the mask graph obtaining and last photoresist
The schematic diagram of middle imaging.501 is initial mask figure, is also the targeted graphical that opc optimizes simultaneously；502 is to obtain in step 108
Mask graph opc regression result；503 is the opc regression result opc after the process obtaining in step 7044；504 is step
Opc regression result opc after the process obtaining in 7055；505 is the opc regression result after the process obtaining in step 707
opc7；506 is to adopt opc7Imaging results in the photoresist obtaining as mask graph.
Table 1 lists this method and initial mask figure and the performance indications of professional software pbopc module contrast, including becoming
The segmentation of mask after the average edge displacement error (edge placement error, abbreviation epe) of picture, operation time and optimization
Trapezoidal sum in figure.From the data in table 1, it can be seen that the present invention utilizes kernel regression technology, effectively increase the computing of traditional pbopc
Efficiency.The present invention make use of the advantage of pbopc and ebopc using adaptive approach simultaneously simultaneously, is improving etching system imaging
While performance, effectively reduce the complexity of mask graph, improve the manufacturability of mask, reduce being manufactured into of mask
This.
Table 1. this method and initial mask and professional software pbopc performance comparison
| Initial mask | Professional software pbopc | This method | |
| Average epe (nm) | 14.6 | 4.0 | 6.4 |
| Operation time (s) | - | 587 | 302 |
| Mask complexity | 1244 | 11400 | 6731 |
Although combining the specific embodiment of the Description of Drawings present invention, it will be apparent to those skilled in the art that
Under the premise without departing from the principles of the invention, some deformation, replacement and improvement can also be made, these also should be regarded as belonging to this
Bright protection domain.
Claims (8)
1. a kind of adaptive optics proximity effect correction method using kernel regression technology is it is characterised in that concretely comprise the following steps:
Step 101, set up ebopc database and pbopc database；
Step 102, mask graph to be optimized is divided into some sub- mask graphs, has between adjacent described sub- mask graph
Width is woverlapOverlapping region；
The observation station in every sub- mask graph in step 103, respectively determination step 102, and the observation station of determination is designated as
ok, the observation station in its neutron mask graph includes the observation station on salient angle summit, re-entrant angle summit and mask graph edge；
Step 104, for each of step 103 observation station okDistribute sub-regions mapk, only comprise in each sub-regions
One observation station；
Step 105, for each observation station ok, carry out sampling in region about and take a little, and each sampled point is corresponded to picture
Plain value is arranged as a vector in order
Step 106, calculate each observation station okAverage distance with mask graph aboutIfThen exist
In step 107, kernel regression is carried out using pbopc database, otherwise in step 107 kernel regression is carried out using ebopc database,
Wherein symbol threshold represents predetermined threshold value；
Step 107, be directed to each observation station ok, using kernel regression technology, according to described vectorThe number selected from step 106
It is weighted averagely, generating corresponding to observation station o according to selecting priori opc optimum results in storehousekOpc regression result, and will see
Measuring point okOpc regression result be filled into corresponding subregion mapkIn, thus being spliced into one for each sub- mask graph
Opc regression result；
Step 108, in the corresponding opc regression result of each sub- mask graph, removing its outer width is woverlapEdge
Region, and the opc that corresponding for all sub- mask graphs opc regression result is spliced into corresponding to overall mask graph returns knot
Really；
Step 109, the opc regression result to the overall mask graph obtaining in step 108 post-process, and will finally obtain
Opc figure as final opc optimum results.
2. according to claim 1 adopt kernel regression technology adaptive optics proximity effect correction method it is characterised in that
Concretely comprising the following steps of ebopc database and pbopc database is set up in described step 101:
Step 201, from full chip mask chosen area as training mask graph；
Step 202, opc optimization is carried out to training mask graph, obtain that its corresponding pbopc optimizes figure and ebopc is excellent respectively
Change figure；
Step 203, the observation station found in this training mask graph, and the observation station searching out is designated as oi, wherein train mask
Observation station in figure includes the observation station on salient angle summit, re-entrant angle summit and training mask graph edge；
Step 204, for each observation station oi, carry out sampling in region about and take a little, and each sampled point is corresponded to picture
Plain value is arranged as a vector in order
Step 205, to train each observation station o on mask graphiCentered on, in the ebopc corresponding to training mask graph
Optimize and in figure, intercept the figure that size is m × m, be designated asPbopc corresponding in training mask graph optimizes in figure
Intercept the figure that size is m × m, be designated as
Step 206, each observation station being directed on training mask, set up vectorWithOne-to-one relationship
It is stored in ebopc database, realize the foundation of ebopc database；Set up vectorWithOne-to-one relationship
It is stored in pbopc database, realize the foundation of pbopc database.
3. according to claim 2 adopt kernel regression technology adaptive optics proximity effect correction method it is characterised in that
Each observation station o is given in described step 104kDistribute sub-regions mapkConcretely comprise the following steps:
Step 301, distribute one centered on this observation station for each salient angle summit, re-entrant angle summit and edge observation station, cd is
The initial subregion of square of the length of side, wherein cd is the minimum feature in objective circuit figure at chip；
Step 302, it is directed to each edge observation station, along edge residing for described edge observation station, with identical expansion rate respectively
Extend the length of its corresponding square initial subregion to both sides, until the subregion phase of this subregion and other observation stations
Meet, the width wherein keeping the corresponding subregion of edge observation station is cd；
Step 303, for each salient angle summit, re-entrant angle summit and edge observation station, with identical expansion rate to around owning
Its corresponding subregion of Directional Extension, until this subregion is met with the subregion of other observation stations, or the distance of extension reaches
Predetermined higher limit.
4. according to claim 3 adopt kernel regression technology adaptive optics proximity effect correction method it is characterised in that
For each observation station o in described step 204i, carrying out in region about samples takes the detailed process to be a little:
S41, with observation station oiCentered on, set up with c × αjNm is multiple concentric circles of radius, maximum in the plurality of concentric circles
Diameter of a circle be more than etching system optical proximity effect distance, wherein c and α be parameter set in advance, j=1,2,3 ...；
S42, in observation station oiPlace takes 1 sampled point, with oiFor 8 sampled points are taken on each concentric circles in the center of circle, adopt for this 8
Sampling point and oiLine and x-axis angle be respectively 0 °, 45 °, 90 °, 135 °, 180 °, 225 °, 270 ° and 315 °；
S43, the value of each sampled point is arranged in order as a vector according to the order outside from the center of circleWherein sampled point
Value be as sampled the pixel value in sample point for the figure.
5. adopt the adaptive optics proximity effect correction method of kernel regression technology according to claim 1 or 4, its feature exists
In for each observation station o in described step 105k, carrying out in region about samples takes the detailed process to be a little:
Step 401, with observation station okCentered on, set up with c × αjNm is multiple concentric circles of radius, in the plurality of concentric circles
Maximum diameter of a circle is more than the optical proximity effect distance of etching system, and wherein c and α is parameter set in advance, j=1, and 2,
3…；
Step 402, in observation station okPlace takes 1 sampled point, with okFor taking 8 sampled points on each concentric circles in the center of circle, this 8
Individual sampled point and okLine and x-axis angle be respectively 0 °, 45 °, 90 °, 135 °, 180 °, 225 °, 270 ° and 315 °；
Step 403, the value of each sampled point is arranged in order as a vector according to the order outside from the center of circleWherein adopt
The value of sampling point is as sampled the pixel value in sample point for the figure.
6. adopt the adaptive optics proximity effect correction method of kernel regression technology according to claim 1 or 4, its feature exists
In calculating each observation station o in described step 106kAverage distance with mask graph aboutConcretely comprise the following steps:
Step 501, with each observation station okFor starting point, respectively to x-axis angle be 0 °, 45 °, 90 °, 135 °, 180 °, 225 °,
270 ° and 315 ° of 8 direction search and okClosest sub- mask graph, if the distance value on this 8 directions is respectively di,
I=1,2 ... 8；
If certain direction of step 502 is in okIn the mask graph of place, then make the direction corresponding apart from di=0, if searching on certain direction
Rope less than other mask graphs, then makes the direction corresponding apart from diOptical proximity effect distance equal to etching system；
Step 503, calculating are corresponding to okAverage distanceWherein ndFor non-zero diNumber.
7. adopt the adaptive optics proximity effect correction method of kernel regression technology according to claim 1 or 4, its feature exists
In for each observation station o in described step 107k, using kernel regression technology, select priori opc to optimize from opc database
Result is weighted averagely, generating corresponding to observation station okThe concretely comprising the following steps of opc regression result:
Step 601, calculating observation point okCorresponding sampling point vectorThe corresponding sampling with priori datas all in opc database
Point vectorBetween Euler's distance
Step 602, choose withMinimum p of Euler's distanceCalculate kernel function
Wherein p is predetermined kernel regression candidate samples quantitative value, and h is the bandwidth controlling smoothing range；
Step 603, for p choosing in step 602Calculate kernel regression resultIf
Choose ebopc database and carry out kernel regression, thenRepresentIf choosing ebopc database to carry out kernel regression,Represent
8. adopt the adaptive optics proximity effect correction method of kernel regression technology according to claim 1 or 4, its feature exists
In the concrete step in described step 109, the opc regression result of the overall mask graph obtaining in step 108 being post-processed
Suddenly it is:
Step 701, the opc regression result according to the overall mask graph obtaining in step 108, calculate imaging z in its photoresist,
By z, nonoverlapping part is designated as with targeted graphicalWillOverlapping portion between the opc regression result obtaining in step 108
Minute mark isRemove in opc regression resultPart, and the opc regression result after processing is designated as opc1；
Step 702, targeted graphical edge is inwardly indented wsNm, and the targeted graphical after reducing is designated as t1, by opc1With t1Weight
The pixel value in institute's hole in conjunction partly is set to 1, wherein wsFor indentation distance set in advance, hole refers to core picture
Element value is 0, and be closed have the region that pixel value is 1 around figure, by process after opc regression result be designated as
opc2；
Step 703, targeted graphical edge is expanded outwardly wd1Edge after nm is designated as contour1, will be outside for targeted graphical edge
Expansion wd2Edge after nm is designated as contour2, by contour1And contour2Between region be designated as t2, wherein wd1And wd2
For expansion distance set in advance；Remove opc2In with t2Visuals that is overlapping and being connected with main body figure, will be processed
Opc regression result afterwards is designated as opc3；
Step 704, using fabrication mask rule detection method to opc3Carry out process and obtain opc4, make opc4Meet covering of setting
Mould manufacturability condition；
Step 705, using ebopc algorithm to opc4It is optimized and obtain opc5So that adopting opc5The imaging obtaining as mask
Figure is closer to targeted graphical in edge, meets the imaging requirements at pattern edge；
Step 706, using pbopc algorithm to opc5It is optimized and obtain opc6So that adopting opc6The entirety obtaining as mask
Imaging figure is closer to targeted graphical, meets the imaging requirements of whole figure；
Step 707, using fabrication mask rule detection method to opc6Carry out process and obtain opc7, make opc7Meet covering of setting
Mould manufacturability condition.
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 |
|---|---|---|
| CN108490735B (en) | Method, apparatus and computer readable medium for full chip mask pattern generation | |
| US20210240907A1 (en) | Method and Apparatus for Integrated Circuit Mask Patterning | |
| US8732625B2 (en) | Methods for performing model-based lithography guided layout design | |
| US10684544B2 (en) | Optical proximity correction (OPC) method and method of manufacturing mask by using the OPC method | |
| CN107844033B (en) | Method for correcting global metal layer process hot spots | |
| CN103901713B (en) | Self-adaption optical proximity effect correction method adopting kernel regression technology | |
| US7882480B2 (en) | System and method for model-based sub-resolution assist feature generation | |
| US8589830B2 (en) | Method and apparatus for enhanced optical proximity correction | |
| US20200166832A1 (en) | Simulating near field image in optical lithography | |
| JP4395047B2 (en) | Renesting into designs for efficient long-range calculation of correlation maps | |
| CN101297390A (en) | Systems, masks, and methods for photolithography | |
| US10747938B2 (en) | Method for integrated circuit manufacturing | |
| CN110426914A (en) | A kind of modification method and electronic equipment of Sub-resolution assist features | |
| US9779186B2 (en) | Methods for performing model-based lithography guided layout design | |
| US8762901B2 (en) | Method for process proximity correction | |
| KR20220050980A (en) | Neural Network-Based Mask Synthesis for Integrated Circuits | |
| CN110597023B (en) | Photoetching process resolution enhancement method and device based on multi-objective optimization | |
| US10656530B2 (en) | Application of FreeForm MRC to SRAF optimization based on ILT mask optimization | |
| US10578963B2 (en) | Mask pattern generation based on fast marching method | |
| CN103376644B (en) | A kind of mask pattern correction method | |
| WO2022086825A1 (en) | Methods and systems to determine shapes for semiconductor or flat panel display fabrication | |
| US20200064732A1 (en) | Hessian-free calculation of product of hessian matrix and vector for lithography optimization | |
| US20170329886A1 (en) | Generating manufacturable sub-resolution assist feature shapes from a usefulness map | |
| CN107037695B (en) | It is a kind of to calculate coherent imaging etching system channel capacity and the method for image error lower limit | |
| Yang et al. | Adaptive GMM Convolution for Point Cloud Learning. |
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 |