Electron beam exposure write field segmentation method based on big data processing
By adopting a write field segmentation method based on big data processing in the electron beam exposure technology, writing field segmentation is performed on the graphic object, which solves the problems of overcut and repeated calculations in the prior art, and improves processing efficiency and exposure quality.
Patent Information
- Application Number
- CN202411979721.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-31
- Publication Date
- 2025-04-25
AI Technical Summary
When existing electron beam exposure technology processes graphics with huge data volumes, there are problems of overcut and repeated calculations, resulting in inefficient processing.
The electron beam exposure writing field segmentation method based on big data processing is adopted to divide the write field by graphic objects as the target, reducing overcutting, optimizing the correspondence between the graphics and the write field, and improving efficiency. The specific methods include dividing the figure into large and small size figures, performing spatial clustering, using projection dimensionality reduction to determine the sub-layer map range, and performing writing field cutting with the sub-layer map range as the goal.
It effectively reduces the overcut of graphic objects, improves exposure quality and processing efficiency, avoids repeated calculations, and provides higher operational flexibility.
Smart Images
Figure CN119888229A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of precision machining, and in particular relates to an electron beam exposure writing field segmentation method. Background Art
[0002] With the rapid development of the chip industry and precision processing technology, electron beam exposure as a micro-nano processing technology has played a particularly important role.
[0003] The electron beam exposure system sends the geometric pattern to be exposed to the pattern generator, which controls the control circuit to control the scanning of the electron beam and draw the designed geometric pattern on the workpiece. However, due to the limitations of electron optical aberration and distortion, the scanning size of the electron beam cannot be too large each time, and it is necessary to complete the drawing in one scanning field before drawing in the second scanning field.
[0004] With the development of VLSI, the processing accuracy is constantly improving, and the size of workpiece graphics and the scanning field ratio are increasing. In addition, the number and complexity of graphics on each workpiece are also increasing. The processing graphics of a single workpiece may reach 10 8 Even more than 10 9 Densely drawing a large amount of graphics on the same workpiece requires a long time to complete the segmentation of the scanning field and the cutting of the graphics. Therefore, in order to improve the processing efficiency, it is necessary to pre-process the data according to the geometric characteristics, position distribution, granularity, etc. of the drawn graphics.
[0005] The existing technology mainly adopts a method of dividing the writing field from the whole to the part, by counting the drawing range of all graphics to be drawn, evenly dividing the writing field according to the statistical drawing range, arranging it in a two-dimensional matrix, and then polling and drawing the writing field in a scanning movement mode.
[0006] The existing technologies have the following main disadvantages:
[0007] First, the over-segmentation problem:
[0008] The existing technology uses a writing field segmentation method based on the workpiece as a whole, which will cause the problem of over-segmentation of the graphics. For local graphics, due to crossing different writing fields, over-cutting along the boundaries of the writing fields will occur, which will affect the accuracy of graphics drawing and introduce more writing field segmentation errors in subsequent splicing and engraving processes. Figure 1 As shown, the layout size is 1mmX1mm, the write field size is 100μmX100μm, and when 10 write fields are evenly divided in two directions of the layout, over-cutting will occur on the geometric figures.
[0009] Second, the problem of repeated counting
[0010] Electron beam exposure systems are widely used in research institutions. During operation, the exposed graphics need to be modified repeatedly. The above design methods rely heavily on the addition, deletion, and change of graphics. When the amount of data reaches 10 8 Even higher, a simple modification may lead to repeated calculations of a large number of graphics, which is time-consuming and inefficient. Summary of the invention
[0011] In view of this, the present invention proposes an electron beam exposure writing field segmentation method based on big data processing. The present invention realizes writing field segmentation with graphic objects as the target, effectively reduces the occurrence of over-cutting, and improves exposure quality; in addition, the present invention optimizes the correspondence between graphics and writing fields, and improves efficiency.
[0012] The specific technical solutions are as follows:
[0013] A method for electron beam exposure writing field segmentation based on big data processing, which divides the writing field with graphics as the target; specifically, the graphics on the layout are divided into large-size graphics and small-size graphics; multiple small-size graphics are further spatially clustered to form multiple clustering spaces; the sub-layout of each clustering space is determined by using the projection dimensionality reduction method; and the writing field is cut with the sub-layout range as the target.
[0014] Furthermore, the method for dividing large and small-sized graphics is to calculate the span feature of each graphic on the layout. If the span feature is less than the span threshold, the graphic is a small-sized graphic, otherwise it is a large-sized graphic.
[0015] Furthermore, the calculation method of the graphic span feature is to find the minimum rectangular area that completely contains the graphic, the projection of the minimum rectangular area in the x direction is used as the span of the graphic in the x direction, and the projection of the minimum rectangular area in the y direction is used as the span of the graphic in the y direction, and the maximum value of the two spans is the span feature of the graphic.
[0016] Furthermore, the sub-layout is a rectangle parallel to the x-axis of the layout. The method of constructing the sub-layout using the projection dimensionality reduction method is as follows: first, the spans of each small-sized graphic on the x-axis and the y-axis are respectively eliminated; then the partially overlapping spans are merged; next, the distance between adjacent, merged spans is calculated. If the distance is less than a distance threshold, the adjacent spans are merged. At this time, the span of the processed small-sized graphic on the x-axis and the y-axis is the side length of the sub-layout.
[0017] Furthermore, the specific method for performing write field cutting on the sub-layout is: if the range of the sub-layout is larger than the write field size, the write field cutting is performed by average division, or the graphic control downsampling method is used to optimize the write field set in the sub-layout, and the optimized write field set is used to interact with the sub-layout. Otherwise, there is no need to perform write field division on the sub-layout.
[0018] Furthermore, the optimization method of the writing field set specifically includes: uniformly dividing the sub-layout using the writing field size, and numbering each writing field, wherein the writing field number consists of two parts, one part is the serial number of the writing field in the x direction of the sub-layout, and the other part is the serial number of the writing field in the y direction of the sub-layout; using the spatial position feature to determine the writing field range covered by each small-sized graphic, the spatial position feature refers to the position coordinates of the four vertices of the smallest rectangle that completely contains the corresponding small-sized graphic in the corresponding sub-layout; converting the position coordinates of the four vertices into the numbers (Xi, Yi) of the writing fields where the four vertices are located, i =
[0019] 0,1,2,3; the writing field range covered by each small-size graphic in the x direction is [min X i , max X i ], the write field range covered in the y direction is [minY i , max Y i ], thereby obtaining the writing field range covered by each small-sized graphic; the writing field ranges covered by all small-sized graphics in the sub-layout are counted, and the writing field set after the sub-layout optimization is obtained.
[0020] The optimization methods for writing field sets can be used individually.
[0021] Beneficial Effects
[0022] 1. The prior art processes all graphics as a whole, that is, processes according to the writing field, rather than developing with the graphic object as the center, which causes the problem of over-cutting. The present invention starts from the standpoint of large and small graphics, develops with the graphic object as the center, and reduces the problem of over-cutting of graphic objects across the writing field.
[0023] 2. For the graphics to be processed, the area to be drawn is roughly divided into grids by projection dimensionality reduction method to avoid over-segmentation of the image caused by equal division of the writing field.
[0024] 3. For small-sized graphics, a spatial downsampling method is used based on the spatial position and writing field size to simplify the spatial distribution characteristics of the graphics.
[0025] 4. Through the spatial clustering algorithm, find the smallest divisible class, define the graphics set in this class as the smallest graphics set with indivisible writing field size. And use this graphics set as the associated writing field, and perform average writing field segmentation on the writing field in its area.
[0026] 5. This application starts from the spatial characteristic parameters of the graphics itself, avoiding the design method from top to bottom and from the whole to the part. When adding, changing or deleting graphics, it is only necessary to operate on the current class and modify the graphics writing data in the current class, effectively realizing the increment and decrement algorithm with graphics as the object.
[0027] 6. Reduce over-segmentation of graphics through projection dimensionality reduction method, divide the large segmentation of the workpiece space into several small segments of small spaces, and improve the flexibility of graphics drawing.
[0028] 7. By downsampling the spatial feature parameters of the graphics, the spatial size of the processing data is compressed, the algorithm is simplified, and the efficiency of writing field division in large data processing is improved.
[0029] 8. The incremental and decremental algorithms for writing fields of graphics can realize the segmentation of writing fields for different time periods and times of drawing graphics, avoiding repeated calculation of all data, improving the efficiency of writing field segmentation, and providing operators with more operational flexibility. BRIEF DESCRIPTION OF THE DRAWINGS
[0030] Figure 1 , write the field cutting diagram;
[0031] Figure 2 , schematic diagram of the principle of the present invention;
[0032] Figure 3 , Schematic diagram of projection dimension reduction principle;
[0033] Figure 4 , projection sorting diagram;
[0034] Figure 5 , schematic diagram of merging adjacent spans;
[0035] Figure 6 , schematic diagram of sub-layout formation;
[0036] Figure 7 , schematic diagram of sub-layout cutting;
[0037] Figure 8 , write the field number diagram;
[0038] Fig. 9 , schematic diagram of spatial position characteristics;
[0039] Fig.10 , downsampling flow chart. DETAILED DESCRIPTION
[0040] This application mainly adopts projection dimensionality reduction processing, feature screening, graphic downsampling, spatial clustering and other methods to pre-process the drawn graphics, and optimizes the parameters such as electron beam exposure time parameters and exposure measurement according to the characteristic parameters such as the spatial position, layout density, and granularity of the graphics. The main operation process is shown in Figure 2 shown.
[0041] The specific steps of the present invention are as follows:
[0042] Step 1: Load the layout. For each figure on the layout, calculate a minimum rectangular area that can completely contain each figure. The projection of the minimum rectangular area in the x direction is used as the span of the figure in the x direction, and the projection of the minimum rectangular area in the y direction is used as the span of the figure in the y direction. The maximum value of the two spans is the span feature of the figure. Projecting in the x and y directions of the layout is actually projecting the x and y axes on the layout coordinate system. The coordinate system of the small layout is the same as that of the large layout.
[0043] When the span characteristic value is greater than the write field size multiplied by k, the geometric figure is considered to be a large-size figure. When the span is less than the write field size multiplied by k, the figure is considered to be a small-size figure. The value range of K can be set to 0.5 to 1 according to the layout.
[0044] like Figure 2 As shown, write field cutting is performed for large-sized geometric figures, and the following steps are continued for small-sized figures.
[0045] At the same time, the geometric features of the graphics are calculated, including the graphics position, the graphics span, and the distance between two adjacent points in the graphics, which are used as the granularity of the geometric graphics.
[0046] Step 2: Perform spatial clustering on multiple small-sized graphics to form multiple clustering spaces;
[0047] Step 3: First, Figure 3 As shown, the spans of each small-sized graphic on the x and y axes are respectively eliminated, and then the partially overlapping spans are merged;
[0048] Step 4: Figure 4 The projections are sorted as shown;
[0049] Step 5: Figure 5 As shown, the distance between adjacent, merged spans is calculated. If the distance is less than a distance threshold, the adjacent spans are merged. The distance threshold refers to m times the write field size. m is determined according to the geometry and exposure requirements and can be 0.1 to 1.
[0050] Step 6: Figure 6 As shown, the sub-layout is divided according to the merged projection interval.
[0051] Step 7: Figure 7 As shown, the size of the sub-layout is determined, and whether to cut and divide the sub-layout is determined according to the writing field size, so as to complete the writing field cutting and division of all geometric figures.
[0052] In the electron beam exposure system, in order to improve the accuracy of the exposure pattern, a writing field is usually controlled within the range of 10μm to 200μm. For large-area layouts, such as 3mmX3mm layouts, multiple writing fields need to be spliced. For example, when a 3mmX3mm layout is exposed with a writing field size of 20 microns, the writing field data can reach up to 22,500. When writing fields for millions of geometric objects separately for interactive calculations, it often takes a long time and a lot of memory space.
[0053] In order to improve the computing efficiency, for the small-size graphics with large data discussed above, the method of downsampling the graphics control is first used, such as Fig.10 As shown, the write field set it occupies is optimized. The specific implementation method can refer to the following process. Of course, the write field optimization can also be used alone.
[0054] Step 8: Figure 8 As shown, the sub-layout is evenly divided by the writing field size, and each writing field is numbered, wherein the writing field number consists of two parts, one part is the serial number of the writing field in the x direction of the sub-layout, and the other part is the serial number of the writing field in the y direction of the sub-layout;
[0055] Step 9: For each sub-layout, count the spatial position features of each small-sized graphic in the sub-layout. The spatial position features refer to the position coordinates of the four vertices of the smallest rectangle that completely contains the corresponding small-sized graphic in the corresponding sub-layout, such as Fig. 9 As shown;
[0056] Step 10: Use the spatial position feature to obtain the writing field range covered by each small-size graphic. Specifically, calculate the writing field numbers (Xi, Yi) of the four vertices according to the coordinates (xi, yi) of the four vertices of the minimum rectangle. The calculation formula is as follows:
[0057] Xi=Rounddown(xi / Sd)+MOD(xi / Sd)
[0058] Yi=Rounddown(yi / Sd)+MOD(yi / Sd),i=0,1,2,3
[0059] Among them, Rounddown means rounding up, and MOD means remainder;
[0060] The writing field range covered by each small-size graphic in the x direction is [min Xi, max Xi], and the writing field range covered in the y direction is [minYi, max Yi], thereby obtaining the optimized writing field range;
[0061] For example, when the layout size DesignLayoutSize is 3mm×3mm and the write field size FieldSize is 100μm×100μm, the write field set can be expressed as {(Xi,Yi)|Xi is an integer, and 0≤Xi<(n-1), Yi is an integer, and 0≤Yi<(n-1), So the write field set includes elements.
[0062] Step 11: Optimize the writing field set, extract and count the effective writing fields of the sub-layout, and use the optimized writing field range to perform interactive operations with small-size graphics.
[0063] The present invention pre-processes the write field segmentation by adopting analysis methods such as projection dimensionality reduction, graphic downsampling, and spatial clustering, and obtains an indivisible minimum graphic set related to the write field size, thereby improving the flexibility of the write field segmentation and effectively avoiding the recalculation process of the newly added graphics in the write field segmentation from the global consideration of the graphics.
Claims
1. A method for electron beam exposure writing field segmentation based on big data processing, characterized in that: The writing field is divided with graphics as the target; specifically, the graphics on the layout are divided into large-size graphics and small-size graphics; multiple small-size graphics are further spatially clustered to form multiple clustering spaces; the sub-layout of each clustering space is determined by using the projection dimensionality reduction method; Write field cutting is performed with the sub-layout range as the target.
2. The electron beam exposure writing field segmentation method based on big data processing according to claim 1 is characterized in that: The method for dividing large and small-sized graphics is to calculate the span feature of each graphic on the layout. If the span feature is less than the span threshold, the graphic is a small-sized graphic, otherwise it is a large-sized graphic.
3. The electron beam exposure writing field segmentation method based on big data processing according to claim 2 is characterized in that: The calculation method of the graphic span feature is to find the minimum rectangular area that completely contains the graphic, and the projection of the minimum rectangular area in the x direction is used as the span of the graphic in the x direction, and the projection of the minimum rectangular area in the y direction is used as the span of the graphic in the y direction. The maximum value of the two spans is the span feature of the graphic.
4. The electron beam exposure writing field segmentation method based on big data processing according to claim 1 is characterized in that: The sub-layout is a rectangle parallel to the x-axis of the layout. The process of constructing the sub-layout using the projection dimensionality reduction method is as follows: first, the spans of each small-sized graphic on the x and y axes are respectively eliminated; then the partially overlapping spans are merged; next, the distance between adjacent, merged spans is calculated. If the distance is less than the distance threshold, the adjacent spans are merged. At this time, the span of the processed small-sized graphic on the x and y axes is the side length of the sub-layout.
5. The electron beam exposure writing field segmentation method based on big data processing according to any one of claims 1 to 4, characterized in that: The specific method for writing field cutting of a sub-layout is: if the range of the sub-layout is larger than the writing field size, the writing field cutting is performed by average division, or the writing field set in the sub-layout is optimized by downsampling the graphic control, and the optimized writing field set is used to interact with the sub-layout. Otherwise, there is no need to divide the writing field of the sub-layout.
6. The electron beam exposure writing field segmentation method based on big data processing according to claim 5 is characterized in that: The optimization method of the writing field set specifically includes: dividing the sub-layout evenly by the writing field size, and numbering each writing field, wherein the writing field number consists of two parts, one part is the serial number of the writing field in the x direction of the sub-layout, and the other part is the serial number of the writing field in the y direction of the sub-layout; determining the writing field range covered by each small-sized graphic by using the spatial position feature, wherein the spatial position feature refers to the position coordinates of the four vertices of the smallest rectangle that completely contains the corresponding small-sized graphic in the corresponding sub-layout; converting the position coordinates of the four vertices into the numbers (Xi, Yi) of the writing fields where the four vertices are located, i = 0, 1, 2, 3; the writing field range covered by each small-sized graphic in the x direction is [min X i , max X i ], the write field range covered in the y direction is [minY i , max Y i ], thereby obtaining the writing field range covered by each small-sized graphic; the writing field ranges covered by all small-sized graphics in the sub-layout are counted, and the writing field set after the sub-layout optimization is obtained.