Map partitioning methods and computer program products, computer-readable media
By modeling the layout as a conflict structure graph and using a policy network to select the optimal coloring, the new conflict problem introduced by the DSA template is solved, the generalization and efficiency of the layout splitting algorithm are improved, and the manufacturing cost is reduced.
Patent Information
- Application Number
- CN202411639992.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-15
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2044-11-15
AI Technical Summary
Existing layout splitting strategies cannot be applied to the new conflicts introduced by Directed Self-Assembly (DSA) templates, and require retraining the model for each new DSA process, which increases complexity and cannot be simplified to an accurate coverage problem.
By modeling the original map as a conflict structure graph, the policy network selects the optimal coloring based on the feature information of the node to be colored and the predetermined DSA rules, thereby realizing coloring uniformity feedback and policy update, decoupling the DSA rules from the learning model, and simplifying the map splitting process.
The elimination of the need to retrain the model for each new DSA process improves the generalization and efficiency of the layout splitting algorithm and reduces manufacturing costs.
Smart Images

Figure CN119598945B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to a layout splitting method, and more particularly to a layout splitting method for a mask used in guided self-assembly technology. Background Technology
[0002] With the rapid rise of emerging technology industries such as 5G, artificial intelligence, and the metaverse, massive amounts of data are generating enormous demand for computing power, which in turn is driving high-performance, high-computing-power chips to become a new trend and a driving force for industry development. The market's increasing demand for high-performance chips is also further accelerating the chip iteration cycle.
[0003] For example, in logic chip process nodes of 7nm and below, the minimum design rules for contact / via layers have exceeded the resolution limit of mainstream 193nm immersion lithography machines. To achieve small-cycle contact / via patterns, DUV 193i multi-pattern lithography technology is required.
[0004] Multiple Patterning Lithography (MPL) is a technique that decomposes conflicting lithographic patterns onto different photomasks, subjecting each mask to multiple exposures and processing steps to achieve higher lithographic resolution. However, this technique requires multiple photomasks, and increasing the number of photomasks leads to higher manufacturing costs and increased manufacturing complexity.
[0005] Directed self-assembly (DSA) lithography is a promising new patterning process capable of fabricating periodic nanopatterns. Compared to traditional lithography, it offers advantages such as high resolution and low cost. Based on the molecular weight and microphase separation strength of block copolymers (BCPs), DSA lithography, after annealing, can form high-resolution, high-density patterns and has been successfully applied to the research of various devices. With its low cost and high throughput, DSA technology reduces the number of mask splits, providing a cost-effective solution for pattern splitting. Summary of the Invention
[0006] However, current layout splitting strategies are generally only applicable to layouts that can be composed of standard cells or sparse and near-planar layouts, and they do not consider the new conflicts introduced by the DSA template, thus failing to simplify the problem into an accurate coverage issue. Furthermore, current DSA layout splitting models require retraining for each new DSA process, needing to learn specific DSA rules, such as whether the current node can be combined with other nodes to form a DSA point pair, making the DSA layout splitting model extremely complex.
[0007] This invention was developed to address the aforementioned problems. Its purpose is to provide a layout splitting method that utilizes a policy network to select the optimal color for coloring based on the feature information of the current node to be colored, obtained from the coloring information surrounding the current node, and the feasible coloring types of the current node to be colored, obtained from predetermined DSA rules. This eliminates the need to retrain the model for each new DSA process, decoupling the DSA rules from the learning model and greatly enhancing the generalization of the layout splitting algorithm.
[0008] Technical solutions to solve technical problems
[0009] To address the aforementioned problems, the layout splitting method according to the first aspect of the present invention includes the following steps:
[0010] (1) Model the original layout as a conflict structure graph with vias as nodes and the conflict relationship between the vias as edges;
[0011] (2) Based on the coloring information around the current node to be colored, obtain the feature information of the current node to be colored;
[0012] (3) Based on predetermined rules, determine the feasible coloring types for the current node to be colored;
[0013] (4) Using a policy network, the optimal color is selected for coloring based on the feature information of the current node to be colored and the feasible coloring types of the current node to be colored.
[0014] Furthermore, it also includes the following steps:
[0015] (5) For all nodes to be colored, repeat steps (2), (3), and (4);
[0016] (6) The structure diagram after coloring all nodes to be colored is split into multiple split structure diagrams so that nodes of the same color are located on the same split structure diagram, thereby obtaining multiple split layout diagrams corresponding to the multiple split structure diagrams.
[0017] Further, in step (4), the color uniformity of the current structure graph is determined based on the color selected according to the feature information of the current node to be colored and the feasible coloring types of the current node to be colored, and the evaluation information is fed back to the policy network according to the color uniformity. The policy used in the policy network is updated based on the evaluation information, so as to select the optimal color for coloring.
[0018] Furthermore, the evaluation information is information showing the difference between the staining uniformity of the current structure diagram and the average staining uniformity.
[0019] Furthermore, in step (4), if the color selected based on the feature information of the current node to be stained and the feasible staining types of the current node to be stained is not included in the feasible staining types, evaluation information is fed back to the policy network, and the policy used in the policy network is updated based on the evaluation information, so as to select the optimal color for staining.
[0020] Furthermore, the predetermined rules are rules that guide the self-assembly process.
[0021] The rules in the guided self-assembly process include the number of split layouts, the conflict distance between multiple nodes, and the conditions for multiple nodes to be combined into point pairs.
[0022] Furthermore, the coloring information around the current node to be colored includes the coloring information of nodes within the conflict range around the current node to be colored, as well as the information of the edges corresponding to the conflict relationship with the via.
[0023] Furthermore, the feature information is a high-dimensional feature vector.
[0024] Furthermore, the layout is a layout used in a guided self-assembly process.
[0025] The computer program product according to the second aspect of the present invention includes a computer program that, when executed by a processor, implements the above-described layout splitting method.
[0026] Invention Effects
[0027] According to the layout splitting method of the present invention, by utilizing a policy network, the optimal color is selected for coloring based on the feature information of the current node to be colored obtained from the coloring information around the current node to be colored, and the feasible coloring types of the current node to be colored obtained from the predetermined DSA rules. Therefore, it is not necessary to retrain the model for each new DSA process, and the DSA rules can be decoupled from the learning model, which greatly enhances the generalization of the layout splitting algorithm. Attached Figure Description
[0028] Therefore, in order to understand the above-described features of this case in detail, a more detailed description of this case can be made by referring to the embodiments, and a more specific description of the disclosure as briefly described above can be made by referring to the embodiments, some embodiments of which are illustrated in the accompanying drawings. However, it should be noted that the drawings only illustrate exemplary embodiments and should not be considered as limiting its scope, and other equivalent embodiments are permissible.
[0029] Figure 1 This is a schematic diagram illustrating a specific framework example of the layout splitting method involved in the embodiments of the present invention.
[0030] Figure 2 This is a flowchart illustrating specific steps of a layout splitting method according to an embodiment of the present invention.
[0031] Figure 3A This is a flowchart illustrating an example of the feedback process implemented in the step of selecting the optimal color for coloring based on the feature information of the current node to be colored and the feasible coloring types of the current node to be colored, according to the implementation of the layout splitting method of the present invention.
[0032] Figure 3B This is a flowchart illustrating another example of the feedback process implemented in the step of selecting the optimal color for coloring based on the feature information of the current node to be colored and the feasible coloring types of the current node to be colored, according to the implementation of the layout splitting method of the present invention.
[0033] Figure 4 This is a schematic diagram illustrating an example of a split structure diagram obtained by splitting a structure diagram after coloring all nodes to be colored using the layout splitting method according to an embodiment of the present invention.
[0034] Figure 5 This is a schematic diagram illustrating an example of a structure diagram obtained by coloring a structure diagram, including two nodes (including diagonal ones) constituting a point pair and three nodes (including diagonal ones), using the layout splitting method according to an embodiment of the present invention.
[0035] Label Explanation
[0036] 11-Figure Neural Network
[0037] 12 First Environment Module
[0038] 13 Feature Storage Modules
[0039] 14-Policy Network
[0040] 15 Second Environment Module Detailed Implementation
[0041] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. These embodiments are implemented based on the technical solution of the present invention, providing detailed implementation methods and specific operating procedures. However, the scope of protection of the present invention is not limited to the following embodiments.
[0042] For ease of description, spatial relative terms such as “below,” “under,” “down,” “above,” “up,” etc., may be used herein to describe the relationship of one element or feature relative to another element or feature as shown in the figure. It should be understood that spatial relative terms are intended to include different orientations of the device used or operated in addition to those shown in the figure. For example, if the device in the figure were flipped, an element described as “below” or “under” other elements or features would be oriented as “above” other elements or features.
[0043] Unless otherwise specified, the terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. Terms should be understood to have the meaning consistent with their meaning in the context of the relevant art and should not be interpreted in an idealized or overly formalized manner, unless explicitly stated otherwise herein.
[0044] <Methods of Map Division>
[0045] Hereinafter, with reference to the accompanying drawings, a detailed description will be given of an example of the layout splitting method involved in the embodiments of the present invention.
[0046] As an example, the layout splitting method according to the first aspect of the present invention includes the following steps:
[0047] (1) Model the original layout as a conflict structure graph with vias as nodes and the conflict relationship between the vias as edges;
[0048] (2) Based on the coloring information around the current node to be colored, obtain the feature information of the current node to be colored;
[0049] (3) Based on predetermined rules, determine the feasible coloring types for the current node to be colored;
[0050] (4) Using a policy network, the optimal color is selected for coloring based on the feature information of the current node to be colored and the feasible coloring types of the current node to be colored.
[0051] Figure 1 This is a schematic diagram illustrating a specific framework example of the layout splitting method involved in the embodiments of the present invention. Figure 2 This is a flowchart illustrating specific steps of a layout splitting method according to an embodiment of the present invention.
[0052] like Figure 2As shown, firstly, in step ST101, the original layout is modeled as a conflict structure graph with vias as nodes and the conflict relationships between vias as edges. Specifically, for example, the vias in the original layout can be modeled as nodes in the conflict structure graph, so that there is a one-to-one correspondence between the vias in the original layout and the nodes in the conflict structure graph. Furthermore, this can be achieved using... Figure 1 The edges in the diagram represent the conflict relationship between vias. For example, when there is a gray edge between two nodes, it means that there is a conflict relationship between the vias, and these two nodes need to be assigned to different split structure diagrams; when there is no gray edge between two nodes, it means that there is no conflict relationship between the vias, and these two nodes can be assigned to the same split structure diagram.
[0053] Furthermore, as an example, in the initial state, the nodes in the conflict structure graph are, for example, grayed out. This is first determined by the number of split structure graphs to be split (e.g., ...). Figure 1 The original conflict structure graph (containing three graphs: red, blue, and yellow) is initialized and colored using a predetermined initial algorithm, containing nodes to be colored. Here, red, yellow, and blue do not mean that the nodes in the conflict structure graph will be colored in their visual appearance (red, yellow, and blue), but rather to distinguish the different split structure graphs to be assigned later. For example, all red nodes can be assigned to the first split structure graph, all yellow nodes to the second split structure graph, and all blue nodes to the third split structure graph.
[0054] Then, in step ST102, the feature information of the current node to be colored is obtained based on the coloring information surrounding the current node to be colored. Specifically, for example, a certain node to be colored in the original conflict structure graph can be selected as the current node to be colored. The graph neural network 11 obtains the feature information of the current node to be colored based on the coloring information surrounding the current node to be colored, and inputs the feature information into the feature storage module 13.
[0055] As an example, the coloring information around the node to be colored includes the coloring information of nodes within the conflict range around the node to be colored, as well as the information of the edges corresponding to the conflict relationship with the via. For example, when the conflict distance between two nodes is 108nm in the DSA rule, the aforementioned conflict range is a circular range with a radius of 108nm centered on the node to be colored.
[0056] Furthermore, as an example, the feature information of the current node to be colored may be, for example, a high-dimensional feature vector (see reference). Figure 1 (Below the "+" in feature storage module 13).
[0057] Then, in step ST103, the first environment module 12 determines the feasible dyeing types for the current node to be dyed based on predetermined rules. These predetermined rules can be, for example, rules from the DSA process, which may include, for example, the number of split layouts, the conflict distance between multiple nodes, and the conditions for multiple nodes to form a point pair.
[0058] As an example, such as Figure 1 As shown, based on the rules in the DSA process, it can be determined whether the current node to be stained can be stained red or blue, but not yellow (refer to...). Figure 1 (Above the "+" in feature storage module 13). Specifically, when the current node to be colored is colored red, it is determined that it can be colored red because it meets the rules in the DSA process. Furthermore, when the current node to be colored is colored blue, it is determined that it can be colored blue because it can form a point pair with adjacent blue nodes, satisfying the rules in the DSA process. However, when the current node to be colored is colored yellow, it is determined that it cannot be colored yellow because the distance between it and adjacent yellow nodes is less than the conflict distance, not satisfying the rules in the DSA process.
[0059] Then, in step ST104, the policy network 14 is used to select the optimal color for coloring based on the feature information of the current node to be colored and the feasible coloring types of the current node to be colored.
[0060] As an example, the policy network 14 selects red as the optimal color for coloring based on the feature information (high-dimensional feature vector) of the current node to be colored and the feasible coloring types (red or blue) of the current node to be colored, and inputs it into the second environment module 15.
[0061] Next, use Figure 3A and Figure 3B The two feedback processes that can be used in step ST104 are explained. Figure 3A This is a flowchart illustrating an example of the feedback process implemented in the step of selecting the optimal color for coloring based on the feature information of the current node to be colored and the feasible coloring types of the current node to be colored, according to the implementation of the layout splitting method of the present invention. Figure 3B This is a flowchart illustrating another example of the feedback process implemented in the step of selecting the optimal color for coloring based on the feature information of the current node to be colored and the feasible coloring types of the current node to be colored, according to the layout splitting method according to the embodiments of the present invention. Furthermore, at least one of the following first and second feedback processes may be used in step ST104.
[0062] As the first type of feedback process, such as Figure 3AAs shown, in step ST1041a, the second environment module 15 determines the color uniformity of the current structure graph based on the color selected according to the feature information of the current node to be colored and the feasible coloring types of the current node to be colored. Here, color uniformity refers to, for example, whether the proportions of red, blue, and yellow are uniform (i.e., whether the number of each color is the same).
[0063] Then, in step ST1042a, the second environment module 15 feeds back evaluation information to the policy network based on the coloring uniformity. As an example, if the proportion of red is large when the current node to be colored red (i.e., the number of red elements is high), the evaluation information fed back to the policy network 14 shows a negative reward. If the proportion of red is still small when the current node to be colored red (i.e., the number of red elements is low), the evaluation information fed back to the policy network 14 shows a positive reward. Furthermore, as an example, the evaluation information could also be information showing the difference between the coloring uniformity of the current structure graph and the average coloring uniformity. Thus, the impact of the uniformity (density) of the DSA layout on manufacturability is also considered during the layout splitting process.
[0064] Then, in step ST1043a, the policy used in the policy network is updated based on the evaluation information to select the optimal color for coloring. The purpose of updating the policy is to increase the probability of actions with high rewards, so that the rewards obtained by the actions of the policy network become higher and higher, thereby selecting the optimal color for coloring.
[0065] As a second type of feedback process, such as Figure 3B As shown, in step ST1041b, the second environment module 15 feeds back evaluation information to the policy network when the color selected based on the feature information of the current node to be colored and the feasible coloring types of the current node to be colored is not included in the feasible coloring types. As an example, if the policy network 14 selects yellow based on the feature information of the current node to be colored and the feasible coloring types of the current node to be colored, the evaluation information fed back to the policy network 14 shows a large penalty because yellow is not included in the feasible coloring types (red or blue).
[0066] Then, in step ST1042b, the policy used in the policy network 14 is updated based on the evaluation information to select the optimal color for coloring. The purpose of updating the policy is to reduce the probability of actions with high penalties and increase the probability of actions with high rewards, so that the rewards obtained by the actions of the policy network become higher and higher, thereby selecting the optimal color for coloring.
[0067] Next, return to Figure 2Step ST105 will continue to be explained.
[0068] In step ST105, it is determined whether coloring has been performed on all nodes to be colored. If coloring has not been performed on all nodes to be colored, steps ST102, ST103, and ST104 are repeated for all nodes to be colored. If coloring has been performed on all nodes to be colored, proceed to step ST106.
[0069] In step ST106, the structure diagram after coloring all nodes to be colored is split into multiple split structure diagrams so that nodes of the same color are located on the same split structure diagram, thereby obtaining multiple split layout diagrams corresponding to the multiple split structure diagrams.
[0070] Figure 4 This is a schematic diagram illustrating an example of multiple split structure diagrams obtained by splitting a structure diagram after coloring all nodes to be colored using the layout splitting method according to an embodiment of the present invention. Figure 4 As shown in the figure, as an example, the structure graph after coloring all the nodes to be colored is split into multiple split structure graphs, namely purple, black, green, red, and yellow. The time complexity of coloring all the nodes to be colored is O(N) (i.e., linear), and the time taken to color each node is, for example, 0.0008 seconds. Furthermore, Figure 4 In this system, the collision distance between nodes is 108nm, and each node is fitted with a circular hole with a diameter of 50nm. There are no two nodes with a minimum distance of less than 50nm on the same split layout.
[0071] also, Figure 5 This is a schematic diagram illustrating an example of a structure diagram obtained by coloring a structure diagram comprising two nodes (including diagonal) and three nodes (including diagonal) constituting a point pair using the layout splitting method according to an embodiment of the present invention. Figure 5 As shown, as an example, using the layout splitting method involved in the embodiments of the present invention, it is possible to color two or three nodes of horizontal, vertical, and diagonal constituent point pairs with the same color. Figure 5 (The colors in the middle are blue, green, and red).
[0072] Therefore, according to the layout splitting method of the present invention, by utilizing a policy network, the optimal color is selected for coloring based on the feature information of the current node to be colored obtained from the coloring information around the current node to be colored, and the feasible coloring types of the current node to be colored obtained from the predetermined DSA rules. Thus, it is not necessary to retrain the model for each new DSA process, and the DSA rules can be decoupled from the learning model, which greatly enhances the generalization of the layout splitting algorithm.
[0073] Furthermore, a second aspect of the present invention provides a computer program product, including a computer program that, when executed by a processor, implements the above-described layout splitting method.
[0074] It should be understood that the above description is illustrative and not restrictive. For example, the above embodiments (and / or aspects thereof) can be used in combination with each other. Furthermore, many modifications can be made to adapt particular conditions or materials to the teachings of the various embodiments of the invention without departing from the scope of the invention. While the dimensions and types of materials described herein are used to define parameters of the various embodiments of the invention, the embodiments are not intended to be restrictive but are exemplary. Many other embodiments will become apparent to those skilled in the art upon reading the above description. Therefore, the scope of the various embodiments of the invention should be determined by reference to the appended claims and the full scope of their equivalents.
[0075] Industrial practicality
[0076] The layout splitting method of the present invention can be widely applied to fields such as semiconductor process and chip manufacturing, and has broad research and application value.
Claims
1. A method for partitioning a map, characterized in that, Includes the following steps: (1) Model the original layout as a conflict structure graph with vias as nodes and the conflict relationship between the vias as edges; (2) Based on the coloring information around the current node to be colored, obtain the feature information of the current node to be colored; (3) Based on predetermined rules, determine the feasible coloring types for the current node to be colored; (4) Using a policy network, the optimal color is selected for coloring based on the feature information of the current node to be colored and the feasible coloring types of the current node to be colored; (5) Based on the color selected in step (4), obtain evaluation information and feed it back to the strategy network. The strategy network updates the strategy used in the strategy network based on the evaluation information. The evaluation information includes at least one of the following: information related to the color uniformity of the current structure diagram and information indicating that the selected color is not included in the feasible color types. (6) Repeat steps (2) to (5) until all nodes are colored; as well as (7) The structure diagram after coloring all nodes to be colored is split into multiple split structure diagrams so that nodes of the same color are located on the same split structure diagram, thereby obtaining multiple split layout diagrams corresponding to the multiple split structure diagrams.
2. The layout splitting method as described in claim 1, characterized in that, The evaluation information is information showing the difference between the staining uniformity of the current structure diagram and the average staining uniformity.
3. The layout splitting method as described in claim 1 or 2, characterized in that, The predetermined rules are rules that guide the self-assembly process. The rules in the guided self-assembly process include the number of split layouts, the conflict distance between multiple nodes, and the conditions for multiple nodes to be combined into point pairs.
4. The layout splitting method as described in claim 1 or 2, characterized in that, The coloring information around the current node to be colored includes the coloring information of nodes within the conflict range around the current node to be colored, as well as the information of the edges corresponding to the conflict relationship with the via.
5. The layout splitting method as described in claim 1 or 2, characterized in that, The feature information is a high-dimensional feature vector.
6. The layout splitting method as described in claim 1 or 2, characterized in that, The layout is the layout used in the guided self-assembly process.
7. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the layout splitting method as described in any one of claims 1 to 6.
