A method for wafer transfer and dispatching

Abstract

A silicon wafer transport dispatching method includes the following steps: receiving silicon wafer production tasks, acquiring real-time status data of each production device, and generating an initial task queue; based on the initial task queue and a real-time digital map of the workshop, allocating optimal paths to transport devices and issuing execution instructions; controlling the transport devices to move along the optimal paths, and collecting real-time information on the position, status, and environmental obstacles of the transport devices, and autonomously performing obstacle avoidance; when the silicon wafer arrives at the target device, performing identity verification and equipment handover confirmation; monitoring abnormal events during the transport process in real time, and executing corresponding handling methods according to the abnormality level; generating production reports based on task completion status and yield data, and feeding the data back to the manufacturing execution system. By collecting real-time equipment status and workshop road condition information, combined with intelligent scheduling algorithms and dynamic path planning, the method achieves rapid task allocation and efficient transport, and performs hierarchical response and global optimization when abnormalities occur.

Description

Technical Field

[0001] This application belongs to the field of semiconductor manufacturing automation technology, and in particular relates to a method for intelligent transfer and automatic dispatching of silicon wafers between equipment. Background Technology

[0002] In current automated semiconductor manufacturing workshops, material handling and task allocation between equipment mainly rely on manual scheduling or fixed-program control. Manual scheduling depends on the operator's experience for task allocation, which is susceptible to human error, fatigue, and other factors, resulting in slow task response speeds, with average waiting times exceeding 30 minutes, and uneven equipment load, with some equipment having an idle rate as high as 40%. While fixed-program control can achieve automation, it lacks flexibility and cannot cope with emergencies such as equipment failures and urgent orders, often resulting in some equipment being overloaded while others are idle.

[0003] Furthermore, traditional material handling route planning often adopts a "shortest distance first" strategy, ignoring real-time road conditions such as temporary obstacles and the dynamic impact of other handling equipment. This results in a route conflict rate as high as 20%, which can lead to material blockages in severe cases. Task allocation also lacks a global optimization perspective; for example, urgent tasks are assigned to equipment that is already fully loaded, further causing delivery delays.

[0004] The aforementioned problems collectively hinder the improvement of production efficiency and the rational use of resources, making it difficult to meet the needs of modern flexible manufacturing.

[0005] Therefore, there is an urgent need for a silicon wafer transfer and dispatching method that can achieve intelligent scheduling, dynamic path planning, and anomaly self-healing. Summary of the Invention

[0006] This application provides a silicon wafer transfer and dispatching method to solve the technical problems of slow task response, low equipment utilization, high path conflict rate and imperfect exception handling mechanism in the prior art.

[0007] To solve at least one of the above-mentioned technical problems, the technical solution adopted in this application is:

[0008] A method for dispatching silicon wafers, comprising the following steps:

[0009] S1. Receive silicon wafer production tasks, obtain real-time status data of each production equipment, and generate an initial task queue.

[0010] S2. Based on the initial task queue and the real-time digital map of the workshop, allocate the optimal path to the handling equipment and issue execution instructions;

[0011] S3. Control the transport equipment to move along the optimal path, and collect the position, status and environmental obstacle information of the transport equipment in real time, and autonomously perform obstacle avoidance;

[0012] S4. When the silicon wafer arrives at the target equipment, identity verification and equipment handover confirmation are performed.

[0013] S5. Monitor abnormal events during the handling process in real time and execute corresponding handling methods according to the level of abnormality;

[0014] S6. Generate production reports based on task completion status and yield data, and feed the data back to the manufacturing execution system.

[0015] Furthermore, in S1, production tasks and real-time equipment status data issued by the manufacturing execution system are received, wherein the real-time equipment status data includes at least one of the following: current task volume, remaining capacity, and fault warning information.

[0016] Furthermore, the initial task queue generation process includes: parsing the priority P, process type and process dependency of each task, merging multi-stage related tasks into a task chain, and sorting the task chain using a dynamic priority algorithm, wherein the proportion of P in the priority weight coefficient α of urgent orders is not less than 0.8.

[0017] Furthermore, in S3, the real-time acquisition of the position and status information of the silicon wafer carrier includes: using a combination of UWB ultra-wideband positioning and RFID radio frequency identification to achieve real-time tracking and positioning of the silicon wafer, and simultaneously using LiDAR to sense surrounding obstacles, stopping immediately and triggering local path replanning when an obstacle is encountered.

[0018] Furthermore, in S4, the identity verification is performed by using RFID technology to detect the device status, which is used to confirm the handover conditions;

[0019] The equipment handover confirmation includes: reading the silicon wafer carrier identification code, detecting whether the target equipment door is open, and verifying process compatibility.

[0020] If any verification fails, a retry mechanism is triggered or a handover exception is reported to the scheduling platform.

[0021] Furthermore, in S5, the abnormal event includes at least one of equipment failure, path blockage, and silicon wafer defect.

[0022] Furthermore, the abnormal events are divided into three levels:

[0023] Level 1 exceptions are minor deviations and will be automatically retried by the local execution unit.

[0024] Level 2 anomalies are caused by partial equipment failures or temporary path blockages, triggering nearby AGVs to coordinate avoidance and task redistribution among similar equipment.

[0025] A Level 3 anomaly occurs when critical equipment fails or silicon wafers are damaged. In this case, a global task rescheduling is initiated, and the damaged silicon wafers are guided to an isolation and recycling area. At the same time, the MES production plan is updated.

[0026] Furthermore, when equipment failure occurs, the system queries the list of backup equipment within 5 seconds, selects a backup equipment of the same type and capacity to take over the original task, and updates the transport path to bypass the failure area. When path congestion occurs, the system completes distributed path consistency replanning for multiple AGVs within 3 seconds through V2X vehicle-to-everything communication.

[0027] Furthermore, this also includes replaying transport data based on the digital twin system after each day's production, analyzing bottlenecks, and optimizing equipment layout or task priority rules.

[0028] Furthermore, in S2, optimal route planning is based on real-time traffic information, avoiding congested sections and coordinating the travel routes of multiple handling equipment.

[0029] The silicon wafer transfer and dispatching method designed in this application achieves rapid task allocation and efficient transportation by collecting real-time equipment status and workshop road condition information, combined with intelligent scheduling algorithms and dynamic path planning, and performs hierarchical response and global optimization when anomalies occur.

[0030] This dispatching method significantly improves production efficiency and resource utilization by introducing a dual optimization mechanism of intelligent scheduling and dynamic path planning. It drastically reduces task response time, decreasing emergency task allocation time from 30 minutes to 2 minutes and reducing ordinary task response time by 60%. Equipment idle rate decreases from 40% to 15%, and overall capacity increases by 25%. Path conflict rate drops from 20% to below 3%, material blockage is almost eliminated, and the average single-transfer time is reduced by 40%. Furthermore, the system possesses excellent flexible production capabilities, automatically responding to order fluctuations and unexpected anomalies, increasing on-time delivery rate to 98%, while reducing manual scheduling workload by approximately 70%. Attached Figure Description

[0031] Figure 1 This is a flowchart of a silicon wafer transfer and dispatching method according to an embodiment of this application. Detailed Implementation

[0032] The present application will now be described in detail with reference to the accompanying drawings and specific embodiments.

[0033] This embodiment proposes a silicon wafer transfer and dispatching method, such as... Figure 1 As shown,

[0034] A method for dispatching silicon wafers, comprising the following steps:

[0035] S1. Receive silicon wafer production tasks, obtain real-time status data of each production equipment, and generate an initial task queue.

[0036] It receives production tasks and real-time equipment status data from the Manufacturing Execution System (MES). Basic information about the production tasks includes product model, process requirements, order priority (whether it's an urgent order), quantity, and delivery date. Real-time equipment status data includes at least one of the following: current workload, remaining capacity, and fault warning information.

[0037] The initial task queue generation process includes: parsing the priority P (urgency level), process type, and process dependencies of each task, and merging multi-stage related tasks into a task chain. The parsed tasks are then associated with the real-time device status. For silicon wafers requiring multiple processes to be completed consecutively, the system merges related subtasks into a single task chain, ensuring they are scheduled as a whole and preventing intermediate disconnections.

[0038] A dynamic priority algorithm is used to sort the task chain and generate an initial task queue Q. This algorithm not only considers the urgency (priority) of orders, but also takes into account the principle of equipment load balancing. In particular, the proportion of P (urgency) in the priority weight coefficient α of urgent orders is not less than 0.8, ensuring that system resources can quickly respond to high-priority demands.

[0039] The task queue (Q) contains information such as the target equipment and priority of the silicon wafers to be transported; the real-time digital map is a dynamically updated virtual workshop model that integrates information from positioning base stations, AGV on-board sensors, and obstacle detectors to reflect the static layout and dynamic road conditions of the workshop in real time.

[0040] This step transforms a chaotic and passive task receiving process into an orderly and proactive scheduling starting point through real-time perception, in-depth analysis, and intelligent sorting of production tasks and workshop status, which is the core prerequisite for the realization of subsequent automated operations.

[0041] S2. Based on the initial task queue and the real-time digital map of the workshop, allocate the optimal path to the handling equipment and issue execution instructions.

[0042] Based on the initial task queue Q in S1 and the real-time digital map of the workshop, the system calculates the optimal path from the current location to the target device for each silicon wafer handling task in the task queue. The optimal path here refers not only to the shortest geographical distance, but also to the shortest time or the lowest overall cost.

[0043] Optimal route planning is based on real-time traffic information, avoiding congested sections and coordinating the routes of multiple handling devices. In other words, while planning the route, the system allocates appropriate handling equipment and execution terminals for each task. Allocation principles include equipment proximity, load balancing, and sufficient power.

[0044] After calculation, the system translates the path and allocation scheme into specific, executable instructions, which are then transmitted via a high-speed network (5G). These instructions are sent to the corresponding AGV controllers and robotic arm control systems with millisecond-level latency, ensuring that all devices receive the instructions synchronously and begin collaborative operation. The instructions include the precise coordinate sequence of the target AGV, its travel speed, and its waiting points along the way; as well as the robotic arm's gripping position, gripping force, and placement posture.

[0045] This step uses real-time calculations to allocate appropriate resources to each handling task and plan the globally optimal action route, aiming to solve technical problems such as congestion and conflicts in traditional workshop material flow.

[0046] S3. Control the transport equipment to move along the optimal path, and collect the position, status and environmental obstacle information of the transport equipment in real time, and autonomously perform obstacle avoidance.

[0047] The handling equipment receives the optimal path execution command from S2 and begins to move autonomously along the path. The robotic arm performs precise grasping and placement according to the command.

[0048] During movement, the equipment continuously and in real-time collects the position and status information of the silicon wafer carrier through multiple sensors and feeds it back to the central system or local controller, forming a real-time digital twin image. The collection of the silicon wafer carrier's position and status information includes: using a combination of UWB ultra-wideband positioning technology and RFID radio frequency identification to achieve real-time tracking and positioning of the silicon wafer, and simultaneously using LiDAR to detect surrounding obstacles, stopping immediately and triggering local path replanning when an obstacle is encountered.

[0049] Once the sensors detect a sudden obstacle ahead, the transport equipment will immediately initiate emergency braking to ensure safety and prevent collisions. After stopping, the transport equipment does not passively wait for central commands; instead, the local controller will immediately trigger local path replanning based on the latest environmental perception information, within the framework of the global path. While performing obstacle avoidance and replanning locally, obstacle information and position adjustments are simultaneously reported to the central dispatch platform so that the platform can update the global real-time digital map for reference by other equipment.

[0050] This step, through the combination of UWB and RFID, achieves centimeter-level real-time tracking and end-to-end identity binding of the silicon wafer carrier; and upgrades the traditional mode of running along fixed tracks or relying on ground QR codes to an active safety system based on real-time environmental perception. At the same time, the handling equipment itself can quickly and autonomously solve problems and resume execution, greatly improving the continuity of single-machine operations and the smoothness of the entire system's operation, and reducing the chain of waiting caused by small interruptions.

[0051] S4. When the silicon wafer arrives at the target equipment, verify its identity and confirm the equipment handover.

[0052] Identity verification uses RFID technology to detect the status of equipment and confirm the handover conditions. Specifically, the RFID reader on the transport equipment or the target docking equipment automatically reads the unique identification code in the tag on the carrier to confirm that the target docking equipment is currently in a receptive state.

[0053] During equipment handover confirmation, the system reads the silicon wafer carrier's identification code, checks whether the target equipment door is open, and verifies process compatibility. Once all the above verifications are passed, the system issues the final instruction, and the AGV or robotic arm performs precise alignment, smoothly delivering the silicon wafer carrier into the designated position inside the equipment, or placing it on the equipment's Load Port.

[0054] If any verification fails, a retry mechanism is triggered or a handover exception is reported to the scheduling platform. For minor issues such as momentary communication interruptions, the system will automatically retry reading or verification; for serious issues such as doors not being opened, process mismatches, or identity not being recognized, the system will immediately report the exception event to the central scheduling platform and the MES system, trigger the S5 process, and suspend the handling task to prevent erroneous operations.

[0055] Through multiple automated verifications, human errors such as sending the wrong equipment or executing incorrect processes are fundamentally eliminated, ensuring the accuracy of the production process and the consistency of silicon wafer quality. By pre-verifying equipment status and process compatibility, equipment failures or process contamination that may result from forced handover are avoided, protecting the integrity of production equipment and work-in-process products.

[0056] S5. Monitor abnormal events during the handling process in real time and execute corresponding handling methods according to the level of abnormality.

[0057] The system continuously scans for potential anomalies around the clock through a sensor network distributed throughout the workshop and on various equipment, as well as monitoring data streams from the MES and scheduling platform. These anomalies include at least one of the following: equipment failure, path congestion, and silicon wafer defects.

[0058] Abnormal events are divided into three levels:

[0059] Level 1 anomalies are minor deviations, such as instantaneous errors during RFID reading or brief packet loss in communication; the impact is limited to a single point, and the local execution unit can automatically retry and quickly recover on its own.

[0060] Level 2 anomalies are local equipment malfunctions or temporary path blockages that affect a local area, requiring coordination with surrounding equipment, triggering nearby AGVs to coordinate avoidance and redistributing tasks among similar equipment.

[0061] Level 3 anomalies are caused by critical equipment failure or silicon wafer damage, triggering a global task rescheduling. Because they impact the overall production plan, the highest level of intervention is required. The damaged silicon wafers are guided to an isolation and recycling area, and the MES production plan is updated simultaneously.

[0062] When equipment failure occurs, the system queries the list of backup equipment within 5 seconds, selects a backup equipment of the same type and capacity to take over the original task, and updates the transport path to bypass the failure area. When path congestion occurs, the system completes distributed path consistency replanning for multiple AGVs within 3 seconds through V2X vehicle-to-vehicle communication.

[0063] This step, through an automated closed-loop monitoring-classification-processing mechanism, enables the system to quickly handle the vast majority of sudden anomalies, minimizing the impact of single-point failures within a limited scope and time. This avoids prolonged, full-line shutdowns caused by anomalies in traditional production, significantly improving overall equipment efficiency. Simultaneously, through a tiered response mechanism of local autonomy, partial collaboration, and global coordination, it optimizes processing efficiency: minor issues are quickly resolved locally, while major issues are precisely escalated for handling, greatly reducing unnecessary reliance on central scheduling resources and the need for manual intervention, thereby enhancing production quality and risk control capabilities.

[0064] S6. Generate production reports based on task completion status and yield data, and feed the data back to the manufacturing execution system.

[0065] The system collects a complete data chain generated throughout the entire process from S1 to S5, including task completion status, yield data, and efficiency data. Task completion status includes the start time, end time, actual executing equipment, transport path, any anomalies, and their handling results for each task. Yield data includes quality events recorded based on visual inspection, handover success rate, etc. Efficiency data includes actual equipment utilization, path conflict statistics, average transport time, and task response time.

[0066] The system automatically integrates and analyzes the above data to generate structured production reports and performance analysis reports; and synchronously transmits all key process data and result data back to the Manufacturing Execution System (MES).

[0067] Furthermore, after the daily production is completed, the system uses the full amount of data collected in this step to replay the data transfer based on the digital twin system, analyzes bottlenecks, and optimizes equipment layout or task priority rules to alleviate congestion.

[0068] By introducing a dual optimization mechanism of intelligent scheduling and dynamic path planning, production efficiency and resource utilization are significantly improved, resulting in a substantial reduction in task response time. Emergency task allocation time has decreased from 30 minutes to 2 minutes, and response time for ordinary tasks has been reduced by 60%. Equipment idle rate has decreased from 40% to 15%, and overall capacity has increased by 25%. Path conflict rate has dropped from 20% to below 3%, material blockage is almost eliminated, and the average single-transfer time is reduced by 40%. Furthermore, the system possesses excellent flexible production capabilities, automatically responding to order fluctuations and unexpected anomalies, increasing on-time delivery rate to 98%, while reducing manual scheduling workload by approximately 70%.

[0069] The silicon wafer transfer and dispatching method designed in this application achieves rapid task allocation and efficient transportation by collecting real-time equipment status and workshop road condition information, combined with intelligent scheduling algorithms and dynamic path planning, and performs hierarchical response and global optimization when anomalies occur.

[0070] The embodiments of this application have been described in detail above. These descriptions are merely preferred embodiments and should not be construed as limiting the scope of this application. All equivalent variations and modifications made within the scope of this application should still fall within the patent coverage of this application.

Claims

1. A method for dispatching silicon wafers, characterized in that the steps include... include: S1. Receive silicon wafer production tasks, obtain real-time status data of each production equipment, and generate an initial task queue. S2. Based on the initial task queue and the real-time digital map of the workshop, allocate the optimal path to the handling equipment and issue execution instructions; S3. Control the transport equipment to move along the optimal path, and collect the position, status and environmental obstacle information of the transport equipment in real time, and autonomously perform obstacle avoidance; S4. When the silicon wafer arrives at the target equipment, identity verification and equipment handover confirmation are performed. S5. Monitor abnormal events during the handling process in real time and execute corresponding handling methods according to the level of abnormality; S6. Generate production reports based on task completion status and yield data, and feed the data back to the manufacturing execution system.

2. The silicon wafer transfer and dispatching method according to claim 1, characterized in that, In S1, production tasks and real-time equipment status data issued by the manufacturing execution system are received, wherein the real-time equipment status data includes at least one of the following: current task volume, remaining capacity, and fault warning information.

3. The silicon wafer transfer and dispatching method according to claim 1, characterized in that, The initial task queue generation process includes: parsing the priority P, process type and process dependency of each task, merging multi-stage related tasks into a task chain, and sorting the task chain using a dynamic priority algorithm, wherein the proportion of P in the priority weight coefficient α of urgent orders is not less than 0.

8.

4. A silicon wafer transfer and dispatching method according to any one of claims 1-3, characterized in that, In S3, the real-time acquisition of the position and status information of the silicon wafer carrier includes: using a combination of UWB ultra-wideband positioning and RFID radio frequency identification to achieve real-time tracking and positioning of the silicon wafer, and simultaneously using LiDAR to sense surrounding obstacles, stopping immediately and triggering local path replanning when an obstacle is encountered.

5. A silicon wafer transfer and dispatching method according to claim 1, characterized in that, In S4, the identity verification is performed by using RFID technology to detect the device status, which is used to confirm the handover conditions; The equipment handover confirmation includes: reading the silicon wafer carrier identification code, detecting whether the target equipment door is open, and verifying process compatibility. If any verification fails, a retry mechanism is triggered or a handover exception is reported to the scheduling platform.

6. A silicon wafer transfer and dispatching method according to any one of claims 1-2 and 4-5, characterized in that, In S5, the abnormal event includes at least one of equipment failure, path blockage, and silicon wafer defect.

7. A silicon wafer transfer and dispatching method according to claim 6, characterized in that, The abnormal events are classified into three levels: Level 1 exceptions are minor deviations and will be automatically retried by the local execution unit. Level 2 anomalies are caused by partial equipment failures or temporary path blockages, triggering nearby AGVs to coordinate avoidance and task redistribution among similar equipment. A Level 3 anomaly occurs when critical equipment fails or silicon wafers are damaged. In this case, a global task rescheduling is initiated, and the damaged silicon wafers are guided to an isolation and recycling area. At the same time, the MES production plan is updated.

8. A silicon wafer transfer and dispatching method according to claim 1, characterized in that, When equipment failure occurs, the system queries the list of backup equipment within 5 seconds, selects a backup equipment of the same type and capacity to take over the original task, and updates the transport path to bypass the failure area. When path congestion occurs, the system completes distributed path consistency replanning for multiple AGVs within 3 seconds through V2X vehicle-to-vehicle communication.

9. A silicon wafer transfer and dispatching method according to claim 1, characterized in that, This also includes replaying transport data based on the digital twin system after each day's production, analyzing bottlenecks, and optimizing equipment layout or task priority rules.

10. A silicon wafer transfer and dispatching method according to claim 1, characterized in that, In S2, optimal route planning is based on real-time traffic information, avoiding congested sections and coordinating the routes of multiple transport devices.

Images