Coverage management method, computing system, and non-volatile computer-readable storage medium

Abstract

A coverage management method, coverage management system, and computing system are provided that set forth techniques for using machine learning to manage vertical alignment or coverage in semiconductor manufacturing. Alignment of interconnect features in fan-out wafer level packaging processes is evaluated and managed by the disclosed techniques. Big data and neural network systems are used to correlate coverage error source factors to coverage metrology categories. The coverage error source factors include tool-related coverage source factors, wafer or die-related coverage source factors, and process context-related coverage error source factors.

Description

[0001] This invention is a divisional application of the invention patent application filed on June 27, 2019, with application number 201910567654.5 and invention title "Coverage Management Method, Coverage Management System and Computing System". Technical Field

[0002] The embodiments of the present invention relate to semiconductor technology, and more particularly to an overlay management method, an overlay management system, and a computing system. Background Technology

[0003] As semiconductor technology evolves, semiconductor dies are becoming increasingly smaller, while more and more functions are being integrated into a single die. Therefore, there is a need to incorporate an ever-increasing number of input / output (I / O) pads with a smaller die surface area into integrated circuit (IC) packages. Fan-out wafer-level packaging (WLP) has emerged as a promising packaging technology to address this challenging situation. In fan-out WLP, the die is diced from the original front-end wafer and then positioned on a carrier wafer to be packaged together with interconnects and I / O pads. An advantage of the fan-out WLP process is that the I / O pads associated with the die can be redistributed to an area larger than the die's own surface. Therefore, the number of I / O pads packed with the die can be increased.

[0004] Fan-out WLP packages can be used to pack a single die, multiple dies side-by-side, or multiple dies in a vertically configured package-on-package (POP) configuration. The POP configuration in a fan-out WLP is achieved by vertically connecting interconnect features (e.g., vias) between multiple dies.

[0005] Overlay metrology processes are used in various semiconductor manufacturing processes to monitor and control vertical alignment. Overlay metrology typically indicates the alignment accuracy of a first patterned layer or features thereon relative to a second patterned layer located at a different vertical level than the first patterned layer, e.g., vertical alignment. Overlay error refers to the misalignment between a first portion on the first patterned layer and a second portion on the second patterned layer. Overlay error metrology (e.g., a measured value) can be measured based on the offset between the first and second portions or between the actual position and the target position of the first portion. In wafer processing, target positions can be determined based on advanced process control (APC).

[0006] In fan-out WLP (Wafer-Loop Layered Components), the tested and approved dies are positioned onto the carrier wafer. Multilayer interconnects are formed, connecting the dies to associated I / O pads and within the interconnects themselves. The interconnects are formed using wafer-level processes, employing photoresist and photolithography similar to those used in front-end wafer fabrication. Therefore, vertical alignment between or within consecutive interconnect layers needs to be managed. Summary of the Invention

[0007] This invention provides an overlay management method, comprising: determining a tool position relative to a wafer processing tool forming a first feature on a first wafer; determining a project position of a first portion of the first wafer corresponding to the first feature; determining an overlay metric relative to the first feature; generating a dataset including the tool position, the project position, and the overlay metric; and generating a function based on the dataset using machine learning, the function relating one or more of the tool position or the project position on a first side of the function to the overlay metric on a second side of the function.

[0008] This invention provides an overlay management system, comprising: a wafer processing tool configured to form features on a wafer; a metrology tool configured to measure the overlay alignment of the features on the wafer; an overlay modeling tool configured to generate an estimated overlay alignment metrology of the features based on one or more of alignment parameters of the wafer processing tool or alignment parameters of the wafer; and a process control tool configured to adjust one or more of the alignment parameters of the wafer processing tool or alignment parameters of the wafer based on the estimated overlay alignment metrology.

[0009] This invention provides a computing system, including: a processor; and a storage unit storing executable instructions, which, when executed by the processor, configure the processor to perform actions including: receiving data relative to a tool position of a wafer processing tool forming a first feature on a first wafer; receiving data relative to a position of the first wafer on a wafer holding tool forming the first feature on the first wafer; receiving context data relative to the formation of the first feature on the first wafer; and generating data for adjusting one or more of the tool position or the project position based on applying at least one of the tool position data, the project position data, and the context data to a regression model function, the regression model function relating at least one of the tool position data, the project position data, and the context data to a coverage metric of the first feature on the first wafer. Attached Figure Description

[0010] The best understanding of various aspects of the invention will be achieved by reading the following detailed description in conjunction with the accompanying drawings. In the drawings, unless the context otherwise indicates, the same reference numerals identify similar elements or actions. The dimensions and relative positions of the elements in the drawings are not necessarily drawn to scale. In fact, the dimensions of various features may be arbitrarily increased or decreased for clarity of discussion.

[0011] Figure 1 This is an exemplary system for managing overlay alignment;

[0012] Figure 2 This is an example dataset of error sources;

[0013] Figure 3 This is an example coverage error dataset;

[0014] Figure 4 This is an exemplary neural network system;

[0015] Figure 5 This is an exemplary operation of a neural network system;

[0016] Figure 6 This is an exemplary process for overlay alignment management;

[0017] Figure 7 It is an exemplary carrier wafer with a die positioned on it in a fan-out wafer-level packaging process;

[0018] Figure 8 yes Figure 7 Exemplary overlay metering on the wafer shown;

[0019] Explanation of icon numbers:

[0020] 100: Coverage Management System / System;

[0021] 110: On-site chip manufacturing system;

[0022] 112: Chip processing toolset / lithography toolset / toolset;

[0023] 114: Process Log;

[0024] 116: Measurement toolset;

[0025] 120: Input dataset;

[0026] 122: Coverage error data;

[0027] 124: Error source data / Coverage source data;

[0028] 126: Other data;

[0029] 130: Big Data Unit;

[0030] 140: Neural Network Unit / Machine Learning Unit;

[0031] 142: Alignment control unit;

[0032] 144: Verification Unit / Verification Module;

[0033] 150: Output dataset / alignment control for output data;

[0034] 152: Error prediction data;

[0035] 154: Corrective adjustment data;

[0036] 210: Subset of tool alignment error source factors / tool alignment subset;

[0037] 212: Previous tools aligned the data;

[0038] 214: Wafer stage position data / Wafer stage position / Tool alignment factor;

[0039] 216: Step position data / step position / tool ​​alignment factor;

[0040] 218: Photomask bond position data / Photomask bond position / Tool alignment factor;

[0041] 220: Exposure position data / exposure position / tool ​​alignment factor;

[0042] 250: Subset of wafer / die alignment error source factors / wafer / die alignment subset;

[0043] 252: Chip offset data;

[0044] 254: Chip rotation data;

[0045] 256: Die offset data;

[0046] 258: Diode rotation data;

[0047] 280: Subset of context error source factors / Subset of context error sources;

[0048] 282: Depth of focus data;

[0049] 284: Exposure duration data;

[0050] 286: Step speed data;

[0051] 288: Irradiation setting data;

[0052] 290: Irradiation source / Irradiation source data;

[0053] 292: Enhanced global alignment / EGA position data;

[0054] 294: Field location data;

[0055] 296: Measurement location data / measurement location;

[0056] 310: Coverage of metering location data;

[0057] 320: x-axis coverage error data;

[0058] 330: Y-axis coverage error data;

[0059] 340: Covers rotation (angle) data;

[0060] 410: Processing unit;

[0061] 420: Storage unit;

[0062] 430: Applications of Neural Networks;

[0063] 432: Training set generation module;

[0064] 434: Machine Learning Module;

[0065] 436: Verification module;

[0066] 438: Prediction module;

[0067] 440: Communication unit;

[0068] 450: Interface unit;

[0069] 460: Other components;

[0070] 500: Operational structure;

[0071] 510: Operation;

[0072] 520: Machine learning operation / machine learning process;

[0073] 530: Predictive operation;

[0074] 600: Operation process;

[0075] 610, 620, 630, 640, 650, 660, 670: Operation;

[0076] 700: Carrier chip / Chip / Substrate chip;

[0077] 702: Target chip location;

[0078] 710, 710A: Core;

[0079] 712, 712A: Target location;

[0080] 720: Part of a chip;

[0081] 722, 724: Alignment markers;

[0082] 730: Chip stage;

[0083] 810: Measurement data of interconnected features;

[0084] X, Y, Z: Axes. Detailed Implementation

[0085] This invention describes techniques for using machine learning to manage vertical alignment or overlay in semiconductor manufacturing. For example, the disclosed techniques are used to evaluate and manage the alignment of interconnect features in a fan-out WLP process.

[0086] In this embodiment, big data and neural network systems are used to correlate coverage error source factors with coverage measurement categories. These coverage error source factors include tool-related coverage source factors, wafer or die-related coverage source factors, and processing context-related coverage error source factors. Tool-related coverage source factors include, but are not limited to, light source position, photomask position, lens position, scanning direction, stepper position during exposure, or wafer stage position. Wafer or die-related coverage error source factors include, but are not limited to, wafer position offset or rotation, die position offset or rotation on the wafer, wafer shape characteristics (such as dimensional deviations or thickness deviations), and die shape characteristics (such as shape deviations or thickness deviations). The processing context coverage error source factors include, but are not limited to, global alignment regions (e.g., regions on the wafer processed under global alignment control), global alignment locations (i.e., measurement locations on the wafer used to determine global alignment adjustments in enhanced global alignment (EGA) control), field alignment locations (e.g., locations on the wafer processed under point-by-point or field-by-field alignment control), tool characteristics (such as wafer stage shape), exposure field locations (e.g., the vertical position of a photomask, which translates to exposure size, depth of focus, and exposure duration), stepper tool stepping speed, illumination source (white light, green light, ivory light, or yellow light), other illumination settings, or metrological locations used to measure layer coverage. Coverage metrology information is also obtained. For example, categories of coverage errors include the magnitude of the coverage error (e.g., on the x and y axes), rotational offsets between features, or the location of the coverage error (e.g., relative to the wafer or relative to the die). These coverage metrology data are collected and fed into machine learning.

[0087] Neural network-based machine learning techniques are used to correlate coverage error source factors with coverage error metrics. A regression model function is obtained as a result of the machine learning. The data is continuously updated and enhanced to continuously train the machine learning process and thus the regression model function. The regression model function can be used to predict or estimate coverage metrics on interconnects formed in fan-out WLP packaging processes. Corrective adjustments to tool alignment settings and / or wafer alignment settings can be obtained based on the regression model function.

[0088] Various methods can be used to validate the regression model function. For example, the estimated coverage measurement value of a reference wafer can be compared with the actual measurement value. The reference wafer can be a newly processed wafer or a previously processed wafer. Coverage error source data and actual measurement data of the reference wafer are identified or retrieved from a database. The coverage error source data of the reference wafer are applied to the regression model function to estimate the coverage measurement value. The estimated coverage measurement value is compared with the actual measurement data to determine if the estimate is sufficiently accurate. A threshold is used for the comparison. The threshold can be determined based on the characteristics of the wafer processing procedure or the coverage tolerance requirements of the layer. For example, in a fan-out WLP, a threshold of approximately 0.1 μm can be selected to determine whether the estimated coverage measurement value meets the actual measurement data of the interconnect features above the die packaged in the fan-out WLP.

[0089] The regression model function can also be validated relative to the estimated corrective adjustments. For example, the estimated corrective adjustments for tool alignment or wafer / die alignment can be implemented using a newly processed reference wafer. Coverage metrology is performed on the reference wafer to determine whether the coverage error has been reduced or eliminated as estimated by the regression model function. The threshold can be reused.

[0090] Using the disclosed techniques, the effective capacity of metrology tools is improved due to AI-based coverage metrology. In embodiments, the effective capacity of physical metrology tools is increased by approximately 40 times. For example, using the disclosed virtual coverage metrology technology, the processing capacity of the coverage metrology team has increased from 292 WLP / day to approximately 11,400 WLP / day. Coverage metrology measurement accuracy has been improved to less than 0.1 μm. The processing cycle for coverage-related quality control has been substantially reduced from approximately 4 hours to substantially real-time. Consequently, the overall process cycle time for fan-out WLPs has also been improved, for example, to 0.1 days.

[0091] Because the estimated coverage value is compared with the actual coverage value in the cross-validation of the artificial intelligence function, the disclosed technique avoids the problem of model overfitting.

[0092] In this disclosure, the cover measurement process associated with a stepper machine is used as an illustrative example to illustrate the disclosed technique, and this is not intended to limit the scope of the invention. The disclosed technique can also be used to predict cover measurement relative to other bumping alignment tools, such as die-to-die shifting measurement or ball mount template alignment measurement. Furthermore, the disclosed virtual cover measurement technique can be applied relative to front-end semiconductor process alignment tools or processes.

[0093] The following disclosure provides numerous different embodiments or instances for implementing various features of the described subject matter. Specific examples of components and constructions are illustrated below to simplify this description. Of course, these are merely examples and are not intended to be limiting. For example, the following description of forming a first feature on or on a second feature may include embodiments in which the first and second features are formed in direct contact, and may also include embodiments in which additional features may be formed between the first and second features, such that the first and second features may not be in direct contact. Furthermore, reference numerals and / or letters may be repeated in various instances of the invention. Such repetition is for the purpose of brevity and clarity, and not by virtue of indicating a relationship between the various embodiments and / or configurations discussed.

[0094] Furthermore, for ease of explanation, this document may use spatially relative terms such as "beneath," "below," "lower," "above," and "upper" to describe the relationship between one element or feature shown in the figures and another element or feature(s). These spatially relative terms are intended to encompass different orientations of the device in use or operation, in addition to those shown in the figures. The device may have other orientations (rotated 90 degrees or other orientations), and the spatially relative descriptive terms used herein may be interpreted accordingly.

[0095] In the following description, certain specific details are set forth to provide a thorough understanding of various embodiments of the invention. However, those skilled in the art will understand that the invention can be practiced without these specific details. In other instances, well-known structures associated with electronic components and manufacturing techniques have not been described in detail to avoid unnecessarily obscuring the description of embodiments of the invention.

[0096] Unless the context otherwise requires, throughout the specification and the above claims, the term "comprise (and its variations, such as comprises and comprising)" shall be interpreted as having an open and inclusive meaning, that is, "including but not limited to".

[0097] The ordinal numbers used, such as first, second, and third, do not necessarily imply a ranked order, but can simply distinguish between multiple instances of an action or structure.

[0098] Throughout this specification, the terms "an embodiment" or "an embodiment" refer to a particular feature, structure, or characteristic described in connection with that embodiment being included in at least one embodiment. Therefore, the phrases "in one embodiment" or "in an embodiment" appearing in various places throughout this specification do not necessarily all refer to the same embodiment. Furthermore, in one or more embodiments, the particular features, structures, or characteristics may be combined in any suitable manner.

[0099] Unless otherwise clearly indicated, the singular forms “a” and “the” as used in this specification and the appended claims include all referents. It should also be noted that unless otherwise clearly indicated, the term “or” is generally used in its meaning, including “and / or”.

[0100] Figure 1 This is an example of an overlay management system 100. For example... Figure 1 As shown, system 100 includes a wafer field manufacturing system 110, an input dataset 120, a big data unit 130, a neural network unit 140, and an output dataset 150. The wafer field manufacturing system 110 includes a wafer processing toolset 112 (e.g., a lithography toolset 112), a process log 114, and a metrology toolset 116. The input dataset 120 includes coverage error data 122, error source data 124, and other data 126. The other data 126 may be historical data about wafer processing, including historical coverage error data and historical error source data as supplement to the wafer processing data obtained and maintained by the process log 114, or other data about wafer processing besides the aforementioned wafer processing data. The neural network unit 140 may include an alignment control unit 142 and a verification unit 144, or work in conjunction with the alignment control unit 142 and the verification unit 144. The output dataset 150 includes error prediction data 152, corrective adjustment data 154, and other output data.

[0101] In operation, wafer processing toolkit 112 is configured to process a wafer (e.g., a carrier wafer on which a die is positioned) in an exemplary fan-out WLP process to form interconnect layers on the die. Wafer processing toolkit 112 may be a photolithography tool and includes, for example, a stepper tool (e.g., a step-repeat camera), a wafer stage or chuck, and a gantry tool and other suitable tools. The stepper passes light through a photomask, thereby forming an image of the photomask pattern. The image is focused and reduced by a lens and projected onto the photoresist-coated surface of the wafer. The stepper operates in a step-repeat manner, wherein the pattern on the photomask is repeatedly exposed across the surface of the wafer in a grid. As a step from one shot position to another, the stepper moves the wafer via a wafer stage. The wafer can be stepped back and forth and left and right using the grid under the lens of the stepper. In fan-out WLP, a gantry picks up the die and positions it onto a designated grid point or area on the carrier wafer. In some cases, more than one gantry tool is used to position the die onto the carrier wafer.

[0102] Process log 114 is configured to monitor, measure, determine, and record wafer processing parameters related to toolset 112 and the wafer located on toolset 112 during processing. Some or all of the wafer processing parameter data are identified as contributing factors to or related to coverage error and are incorporated into error source data 124.

[0103] After the lithography toolset 112 has characterized the wafer on a layer or on the layer, the metrology toolset 116 is configured to measure the wafer to determine the presence of overlay errors relative to the layer or features on the layer. The metrology data also includes details of the overlay errors. These metrology details are categorized as the location of the overlay error on the wafer or a die, the offset size of the overlay error on the x-axis and / or y-axis, or the rotation angle of the overlay. Some of the measurement results from the metrology toolset 116 are incorporated into the overlay error data 122. For the purpose of modeling the correlation between error source factors and overlay error categories, the zero-overlay error scenario is also useful and can be collected as part of the overlay error data 122.

[0104] Big Data Unit 130 is configured to collect coverage error data 122, error source data 124, and other data 126, and to combine or coordinate various datasets for further analysis. For example, Big Data Unit 130 links coverage error data 122, coverage source data 124, and other data 126 of different data categories together in various ways or by other suitable linking methods, relative to a wafer layer, a location on the wafer layer, a structure on the wafer layer, or multiple vertical structures designed to overlap vertically. In this description, the term "feature" on the wafer is used to refer to any part of the wafer that links the coverage error data 122 with the error source data 124. For example, a feature on the wafer could be an interconnect structure. The coverage error data 122 entries are coverage errors of the interconnect structure, and the linked error source data 124 entries are wafer processing parameters recorded relative to the interconnect structure.

[0105] The big data unit 130 also tunes the collected data for measurement scales, units of measurement, time scales, tolerance thresholds, etc., so that each data entry can be used in the same analysis process. The big data unit 130 also enhances the collected data by addressing data entry omissions and data entry interpolation. Other data processing, tuning, or combination techniques may also be present and included within the big data unit 130.

[0106] Other data 126 may be historical data related to coverage error data or error source data relative to previously processed wafers. This historical data is processed in the same or similar manner as coverage error data 122 and error source data 124 within the big data unit 130.

[0107] Neural network unit 140 (or machine learning unit 140) is an artificial intelligence machine learning unit configured to perform a supervised learning process using data provided by big data unit 130 to infer regression model functions that correlate coverage error source factors with coverage error categories. The "supervision" component of machine learning can be set relatively loosely. For example, instead of assigning each factor to an "input" category (e.g., error source factor) and an "output" category (e.g., coverage error category), neural network unit 140 allows the neural network engine to operate "blindly" to maximize the advantages of machine learning in handling the data. Furthermore, neural network unit 140 can be configured to perform multiple different supervised learning tasks, where different groups of factors are classified as "inputs" or "outputs." For example, the coverage error measurement positions in the coverage error data 122 may be classified as outputs in some learning tasks or as inputs in others. Different learning tasks can produce different regression model functions that correlate "input" factors with "output" factors.

[0108] During supervised learning, neural network unit 140 can also distinguish between systematic and random error source factors. Systematic error source factors are those that tend to repeat the same data values ​​or patterns in similar situations. Random error source factors are those that tend to have random data values ​​in similar situations, or data values ​​that cannot be controlled by system 100. For example, die positioning offset can be a systematic factor caused by a specific gantry tool. Stepper machine position offset can also be a systematic factor for a specific stepper machine. As another example, die thickness variation or die size variation can be a random factor that occurs randomly or is uncontrollable in fan-out WLP processes.

[0109] In cases where one or more regression model functions are inferred through a supervised machine learning process, the regression model functions are used by neural network unit 140 to estimate or predict coverage metrics on the wafer. In an embodiment, a backpropagation network is used to determine the regression model functions that correlate input factors with output factors. For example, when computing the regression model functions, input factors are "processed" by "neurons" to contribute to the output, and a backpropagation algorithm is used to determine the weight of the contribution of each of the "neurons".

[0110] Alignment control unit 142 is configured to control coverage alignment in wafer processing (e.g., wafer processing of interconnects on a fan-out WLP) using a regression model function or coverage estimation / prediction results. Specifically, alignment control unit 142 may apply processing log information of wafer processing operations as input to the regression model function to generate alignment control output data 150. This alignment control output data can then be used to control or adjust wafer processing. For example, alignment control output data 150 particularly includes error prediction data 152 and corrective adjustment data 154. Error prediction data 152 indicates the predicted / estimated coverage measurement in one or more wafer locations. Corrective adjustment data 154 indicates a change in one or more error source factors in error source data 124 to eliminate the estimated / predicted coverage error.

[0111] In one embodiment, the alignment control unit 142 is part of the neural network unit 140. In other embodiments, the alignment control unit 142 is a separate unit from the neural network unit 140 and includes mechanisms for automatically adjusting or causing adjustments to one or more of the wafer processing tool settings or wafer position to eliminate overlay errors.

[0112] The validation module 144 of the neural network unit 140 is configured to validate the regression model function based on metrological data obtained from a relevant reference wafer. The reference wafer can be a previously processed wafer for which coverage error data and error source data are readily available. The reference wafer can also be a new wafer that has undergone wafer processing operations and is particularly used for validation purposes. For example, the estimated coverage metrology (e.g., the location and magnitude of the estimated coverage error) of the reference wafer is compared with the actual metrological measurements of the reference wafer. A threshold can be used to assess whether the estimated / predicted coverage metrology is sufficiently accurate or meets the actual metrological measurements. The comparison results can be used to further or continuously train the corresponding regression model function and / or to enable the corresponding regression model function to learn. For example, the training dataset, consisting of actual measurements of coverage error source factors and coverage errors, is continuously updated to reflect the latest state of the tools and processes. In an embodiment, a fixed number of data entries in the training dataset are used to train the regression model function. New measured data entries replace the oldest data entries to maintain the fixed number. Similarly, the corrective adjustment data is essentially a prediction of the coverage error-free value of the reference wafer processed based on the corrective adjustment data. Actual metrological data from a reference wafer can be used to evaluate whether corrective adjustment data effectively eliminates coverage errors. Again, a threshold can be used in the evaluation. The threshold can be selected based on the wafer design or wafer processing technology. For fan-out WLP processes, the threshold can be selected, for example, based on the size of the interconnect features.

[0113] Figure 2Exemplary error source data 124 is shown. Error source data 124 includes a subset 210 of tool alignment error source factors, a subset 250 of wafer / die alignment error source factors, and a subset 280 of context error source factors. Tool alignment subset 210 includes data about tool alignment parameters or settings that affect overlay alignment. In an embodiment, tool alignment subset 210 includes previous tool alignment data 212, wafer stage position data 214, step position data 216, photomask key position data 218, exposure position data 220, or other suitable tool alignment parameter data. Wafer / die alignment subset 250 includes data about the position of the wafer on the wafer stage or the die on the wafer (in the case of fan-out WLP) that affects overlay alignment. In an embodiment, wafer / die alignment subset 250 includes wafer offset data 252, wafer rotation data 254, die offset data 256, die rotation data 258, or other wafer / die alignment data. The context error source subset 280 includes contextual information about coverage metrology measurements. Using this included contextual information, different wafer types and different wafer processing parameters can be identified and considered in neural network processing. In an embodiment, regression model functions are determined or trained separately for different wafer types and / or different wafer processing parameters. For example, different training datasets formed by coverage error source factors and coverage errors are maintained for each wafer type and / or each set of wafer processing parameters. The different training datasets are updated separately to maintain a fixed number of data entries in each training dataset. In an embodiment, the context error source subset 280 includes depth-of-focus data 282, exposure duration data 284, step speed data 286, illumination setting data 288, illumination source 290, enhanced global alignment (“EGA”) position 292, field position data 294, metrology position data 296, or other suitable contextual data.

[0114] Previous tool alignment data 212 refers to the alignment position of the wafer processing toolset 112 used by a previous wafer that underwent the same wafer processing process as the current wafer (e.g., the wafer of the current data entry). Previous tool alignment data 212 is measured relative to the same features or process steps as the current wafer and is measured at the same scale as the current wafer. In other words, previous tool alignment data 212 is equivalent to the current data entry. For example, in the case where the tool alignment data of the current wafer includes wafer stage position data 214, step position data 216, photomask bond position data 218, and exposure position data 220, previous tool alignment data 212 includes all the same data categories from the previous wafer.

[0115] The wafer stage position data 214 can be measured by the position of the actuators used to anchor the wafer stage. For example, the wafer stage may include three actuators for anchoring the wafer stage's x-axis, y-axis, and z-axis positions. The positions of the x-axis, y-axis, and z-axis actuators are recorded and logged to indicate the wafer stage position.

[0116] Step position data 216 is the position of the stepper.

[0117] Photomask key position data 218 indicates the position of the photomask anchored by the photomask key.

[0118] Exposure location data 220 indicates the location or area on the chip exposed to illumination light.

[0119] The wafer stage position 214, step position 216, photomask key position 218, or exposure position 220 can be represented as positions on the x-axis, y-axis in the lateral plane, and / or the z-axis in the vertical plane. In some embodiments, the wafer stage position 214, step position 216, photomask key position 218, or exposure position 220 can each be represented as a positional error relative to the corresponding target position. In measuring the positional error, x-axis offset, y-axis offset, z-axis offset, and rotational error (angle) can be used.

[0120] It should be noted that for each of the tool alignment factors 214, 216, 218, and 220, one or more of three different data types can be recorded and logged. The first type is position data measured by a measuring device (e.g., a laser device), referred to as the "measured position". The second type is the position determined by the Enhanced Global Alignment ("EGA") system of Advanced Process Control ("APC") before applying EGA adjustments, referred to as the "pre-alignment position" or "pre-EGA position". The third type is the position determined after applying EGA adjustments, referred to as the "post-alignment position" or "post-EGA position". It should be understood that each of the three position data types may deviate from the actual position of toolset 112. Such deviations in toolset position determination can contribute to coverage errors. It is not necessary to use all three types of toolset alignment position data in the operation of neural network unit 140. Neural network unit 140 may selectively use some types of data in some regression models and other types of data in others. Furthermore, it is not necessary to include all three types of data in the error source data 124. In some operating scenarios, EGA is not performed, and the position data type before and after EGA is unavailable. In other scenarios, the measured position data type is unavailable or not included in the error source data 124.

[0121] These three types of tool alignment data (e.g., measured position, pre-EGA position, post-EGA position) can be correlated with each other. For example, the post-EGA position is obtained based on the pre-EGA position. However, obtaining each of these three types introduces additional or different factors that can contribute to potential biases. For example, the EGA can be affected by the EGA measurement point (referred to as the "EGA position") selected on the wafer to determine EGA adjustment. Therefore, by including one or more of the three position data types in the tool alignment subset 210, the neural network unit 140 can implement a regression model function that more effectively reflects or accounts for the correlation between the tool alignment factor and the coverage measurement category.

[0122] Other methods for determining tool alignment may also exist and be included in this invention.

[0123] In the wafer / die alignment subset 250, wafer offset data 252 indicates the offset of the wafer on the wafer stage. The wafer offset includes both x-axis and y-axis offsets. In embodiments, the wafer offset is determined by wafer alignment marks included on the wafer or other suitable mechanisms.

[0124] The wafer rotation data 254 indicates the rotation angle of the wafer on the wafer stage. In an embodiment, the wafer rotation is determined by a wafer alignment mark or other suitable mechanism.

[0125] Die offset data 256 indicates the amount of offset of the die on the carrier wafer. Die offset includes x-axis offset and y-axis offset. In an embodiment, die offset is determined by die alignment marks included on the die or other suitable mechanisms.

[0126] Die rotation data 258 indicates the angle of rotation of the die on the carrier wafer. In an embodiment, die rotation is determined by die alignment marks or other suitable mechanisms.

[0127] Similar to tool alignment subset 210, the data categories in wafer / die alignment subset 250 may also include three data types, such as measured location, pre-EGA location, and post-EGA location. It should be noted that the EGA system used for die alignment may differ from the EGA system used for wafer alignment.

[0128] The measured position, pre-EGA position, and post-EGA position are provided as examples of data types within the alignment position data category. These examples are not intended to limit the scope of the invention. Other data types, all included in this invention, may also exist for the alignment position data category. For example, field alignment of a wafer or die may also exist, and the tool / wafer / die position before and after the field alignment operation can be used in the error source data 124.

[0129] The depth-of-focus data 282, exposure duration data 284, illumination setting data 288, and illumination source data 290 are self-descriptive. The stepper speed data 286 indicates the stepper's movement speed. These data are essentially parameters in the process plan and are used to identify the context of coverage measurement and coverage error (if any). By incorporating this contextual data, the regression model function generated by the neural network unit 140 can be further enhanced, for example, by eliminating the complexity caused by various contextual factors. That is, the regression model function can be trained or determined for each of the contextual variables.

[0130] EGA location data 292 indicates multiple location points on the wafer / die identified for EGA alignment purposes. Various features on the wafer typically include varying shapes and profiles. Therefore, the selection of EGA locations (e.g., features at EGA locations) results in varying EGA evaluations and adjustments.

[0131] Field position data 294 indicates the size of the wafer / die region illuminated during the photolithography process. In this embodiment, field position data 294 is measured by the size of the photomask opening and the distance between the photomask and the wafer surface.

[0132] Metrological location 296 indicates the location on the wafer or die where a coverage measurement is performed. As described herein, a metrological location on the wafer can act as one or more of an error source factor or a coverage error category. As an error source factor, the metrological location on the wafer affects the presence or detectability of coverage errors.

[0133] It should be understood that Figure 2 The exemplary error source data categories listed herein are merely examples and do not limit the scope of the invention. Other factors that may cause coverage errors and / or contribute to the magnitude of coverage errors may also be used and included in the error source data 124, and these other factors are all included in this invention.

[0134] Figure 3 Example coverage error data 122 is shown. Figure 3 As shown, the coverage error data 122 includes four coverage error categories: coverage measurement location data 310, x-axis coverage error data 320 and y-axis coverage error data 330, and coverage rotation (angle) data 340. Coverage measurement location data 310 indicates the location where a coverage error occurs / is detected on the wafer or die. x-axis coverage error data 320 and y-axis coverage error data 330 indicate the magnitude of the coverage error on the x-axis or y-axis, respectively. Coverage rotation (angle) data 340 indicates a coverage error where the actual measurement of a feature deviates angularly from the target measurement.

[0135] Various methods can be used to define coverage error. In one embodiment, coverage error is determined based on the actual measurement and target measurement of the feature. In another embodiment, coverage error is determined based on the alignment between the upper and lower features. Other methods for determining coverage alignment accuracy or coverage error may also exist and are included in this invention.

[0136] Figure 4 An exemplary neural network unit 140 is shown. (Refer to...) Figure 4 The neural network unit 140 includes: a processing unit 410, such as a computer processor or processing capacity allocated to the neural network unit 140 in a virtual machine application; a storage unit 420 storing the neural network application 430; a communication unit 440 configured to communicate with other computers or machines linked to the neural network unit 140 in a distributed computing environment; an interface unit 450 configured for input, output, and user interaction; and other components 460.

[0137] The neural network application 430 includes: executable instructions that, when executed by the processing unit 410, configure the processing unit 410 to implement a training set generation module 432; a machine learning module 434, including a validation module 436; and a prediction module 438. In an embodiment, the executable instructions dedicated to implementing the training set generation module 432, the machine learning module 434, and the prediction module 438 are stored on the storage unit 420 in a separate dedicated space of the storage unit 420, or stored in a separable / analyzable manner so that the executable instructions for these modules can be easily identified through analysis or indexing.

[0138] One or more components of the neural network unit 140 can be implemented in a distributed computing environment via physical devices (e.g., server computers) or via virtual devices. For example, multiple host servers can be linked to chip processing sites / operations. The host servers can work together in a distributed computing scheme to support one or more virtual layers in which the neural network unit 140 and / or the big data unit 130 reside. The virtual layer can be any virtualization level, such as full virtualization, operating system (OS) level virtualization, application level virtualization, or partial virtualization at some other level.

[0139] Furthermore, it is not necessary for the components of the neural network unit 140 to reside in the same virtualization layer. Some components of the neural network unit 140 can be implemented through physical layers, while others can be implemented through virtual layers of various levels, as is included in this invention.

[0140] Figure 5The operational structure 500 of the neural network application 430 is illustrated. The training set generation module 432 is configured to receive data from the big data unit 130 and generate a training dataset in operation 510. For example, it receives and processes error source data 124 and overlay error data 122 to generate a training dataset for training the machine learning operation 520. The training dataset is continuously updated by adding new data entries and eliminating old ones. In an embodiment of operation 510, the training set generation module 432 processes the received dataset through a pre-training process to optimize the dataset for the purposes of the machine learning process 520.

[0141] In this embodiment, during machine learning operation 520, machine learning module 434 uses a training dataset to perform a supervised training process. Specifically, supervised training defines data objects in the training dataset that are classified as inputs and data objects that are classified as outputs, and pairs input data objects with corresponding output data objects. Then, the paired input and output data entries are analyzed through training connections between or within neurons to generate a regression model function. In this embodiment, coverage source factors are classified as inputs (“x”) and coverage measurement errors are classified as outputs (“y”). The resulting regression model function each links one or more of the error source factors to a coverage error category, such as coverage measurement position, x-axis coverage error, y-axis coverage error, and coverage rotation.

[0142] In prediction operation 530, prediction module 438 uses the resulting regression model function to predict or estimate coverage metrics on the wafer undergoing wafer processing operations. Specifically, wafer processing parameters are obtained from process log 114 as input and applied to the regression model function to produce coverage metrics information as output. Prediction module 438 may also predict corrective adjustments to wafer processing parameters (e.g., tool alignment settings and / or wafer / die alignment factors) to eliminate or mitigate estimated / predicted coverage errors.

[0143] like Figure 5 As shown, coverage errors can be predicted for the entire wafer. Within the wafer, individual coverage errors may follow a consistent pattern or vary between / among different fields on the wafer. Using a stepper device used in wafer processing, wafer processing settings can be adjusted to specifically correct coverage errors in individual fields on the wafer. However, there are situations where coverage errors in two or more fields on the wafer cannot be corrected individually and are interrelated during wafer processing operations. Decision-making operations can be performed to determine overall corrective adjustments to the wafer processing operations, thereby balancing coverage alignment between / among said two or more fields.

[0144] The results of the prediction operation 530 are output to the alignment control unit 142 to control the wafer processing operation accordingly. For example, error prediction data 152 can be used to manually adjust wafer processing parameter settings to avoid or mitigate coverage error problems. Operators or machines can easily use corrective adjustment data 154 to automatically adjust wafer processing parameter settings to avoid or mitigate coverage alignment problems. The adjusted processing parameters and associated coverage measurement data are then fed back into the neural network unit 140 to further train or refine the regression model function, as the adjusted processing parameters, together with the associated coverage measurement data, represent new data entries. In other words, the machine learning process in the neural network unit 140 can be configured as a dynamic process that continuously updates the regression model function that correlates error source factors with coverage error categories.

[0145] The validation module 436 is configured to evaluate whether the generated regression model function accurately represents the correlation between the error source factors and the coverage error category. For example, the predicted coverage measurement value can be compared with the actual measurement value. A threshold can be used to determine whether the estimated / predicted coverage measurement meets the actual coverage measurement data. For example, in fan-out WLP wafer processing operations, a threshold of 0.1 μm can be used to determine whether the estimated coverage error meets the actual coverage measurement.

[0146] Figure 6 This is a flowchart of an exemplary operation process 600. In exemplary operation 610, process log 114 determines the tool position relative to a wafer processing tool (e.g., a stepper) that forms interconnect features on the wafer. The tool position may be one or more of a measured position, a pre-alignment position, or a post-alignment position.

[0147] In exemplary operation 620, process log 114 determines the location of a portion of the wafer corresponding to a feature on the wafer; for illustrative purposes, this location is referred to as a "project location". The project location may be one or more of a measured location, a pre-alignment location, or a post-alignment location.

[0148] Figure 7 The example shown is a carrier wafer 700, on which multiple dies 710 are positioned. The process log identifies the location of a portion 720 (shown as a dashed circle) of the wafer 700 corresponding to interconnect features. For example... Figure 7As illustrated, as an illustrative example in a fan-out WLP process, portion 720 includes the die 710A position on wafer 700. The item position of portion 720 can be measured in various ways. For example, the item position of portion 720 may include x-axis wafer offset and / or y-axis wafer offset, indicated by the misalignment between alignment mark 722 on wafer 700 and alignment mark 724 on wafer stage 730. The item position may also include rotation between wafer 700 and a target wafer position 702 (shown as a dashed circle) on wafer stage 730.

[0149] In addition, the project location may also include x-axis core offset and / or y-axis core offset. For example... Figure 7 As illustrated in the illustration, the dies 710 positioned on the lower portion of the carrier wafer 700 all include an alignment offset relative to a target position 712. This alignment offset may be systematically caused by a faulty gantry tooling system specifically designed to pick up and position the dies 710 on the lower portion of the wafer 700. As illustrated with respect to section 720, the corresponding die 710A may also include a rotational error relative to the target position 712A.

[0150] The wafer alignment position and the die alignment position can be determined together by process log 114.

[0151] In exemplary operation 630, measurement toolset 116 determines coverage measurement relative to the inline feature. Figure 8 An illustrative example of coverage metrology measurement of wafer 700 is shown. For example... Figure 8 As shown, the coverage alignment measurements on different portions of the wafer 700 may be different and not necessarily consistent. Measurement data 810 of the interconnect features on portion 720 are identified as being related to tool location information and item location information. That is, that data can be linked together through the interconnect features.

[0152] In exemplary operation 640, training set generation module 432 generates a dataset containing tool location data, item location data, and overlay measurement data of interconnected features on a portion 720 of wafer 700.

[0153] In exemplary operation 650, machine learning module 434 generates a regression model function by training machine learning using a training dataset. In this embodiment, machine learning module 434 uses the training dataset to set the parameters of a regression function that links input data (e.g., wafer alignment data and die alignment data) with output data (e.g., overlay metrology data). Training can be continuously repeated and updated as new datasets are added to the training dataset to replace older datasets. The new datasets will reflect the updated state of the manufacturing processes in which overlay metrology errors occur. It should be understood that the resulting datasets may be one of many dataset entries used to train the machine learning process. The regression model function may already exist in machine learning module 434 and can be continuously trained and / or enhanced using new training datasets. New regression models can also be created by machine learning module 434.

[0154] In exemplary operation 660, prediction module 438 estimates coverage metrics on the second wafer based on a trained regression model function. For example, tool position and item position relative to the second wafer can be applied to the regression model function to estimate coverage metrics.

[0155] In exemplary operation 670, verification module 436 evaluates the regression model function by comparing the estimated metric with the metric actually measured on the second wafer.

[0156] The technology is illustrated using a fan-out WLP process as an example. It should be understood that the technology is similarly applicable to front-end wafer processing operations and other back-end wafer processing operations.

[0157] The invention can be further understood through the following description of embodiments:

[0158] In a method embodiment, a tool position of a wafer processing tool is determined. The wafer processing tool is used to form a first feature on a first wafer. A project position of a first portion of the first wafer is determined. The first portion corresponds to the first feature. A coverage metric relative to the first feature is determined. A dataset containing the tool position, the project position, and the coverage metric is generated. A function is generated that correlates one or more of the tool position or the project position on a first side of the function with the coverage metric on a second side of the function. The function is generated using the dataset via machine learning.

[0159] In related embodiments, the tool position includes one or more of the following: wafer stage position, stepper position, photomask key position, or exposure position.

[0160] In related embodiments, the project location includes one or more of a wafer offset or a wafer rotation of the first wafer positioned on a wafer stage.

[0161] In related embodiments, the project location further includes one or more of a die offset or die rotation located on the first portion of the first wafer.

[0162] In related embodiments, the tool position includes one or more of the measured position, the position before alignment, and the position after alignment.

[0163] In related embodiments, the project location includes one or more of the measured location, the pre-alignment location, and the post-alignment location.

[0164] In a related embodiment, the method further includes: determining the context in which the first feature is formed on the first wafer; and generating the dataset containing the context.

[0165] In related embodiments, the context includes one or more of the following: the measurement location of the enhanced global alignment operation on the first wafer, relative to the exposure field forming the first feature, and the measurement location on the first wafer for measuring the coverage measurement.

[0166] In a related embodiment, the method further includes: determining a previous tool position relative to the wafer processing tool that forms a first feature on a previous first wafer; and generating the dataset containing the previous tool position.

[0167] In related embodiments, the machine learning is performed using a neural network system.

[0168] In a related embodiment, the method further includes generating an estimated alignment measurement of the second wafer based on the function.

[0169] In a related embodiment, the method further includes evaluating the function by comparing the estimated alignment metric of the second wafer with the measured alignment metric of the second wafer.

[0170] In a related embodiment, the method further includes generating alignment correction data relative to the second wafer based on the function, the alignment correction data specifying an adjustment for at least one of the tool position or the project position.

[0171] In another embodiment, a system includes: a wafer processing tool configured to form features on a wafer; a metrology tool configured to measure the overlay alignment of the features on the wafer; an overlay modeling tool configured to generate an estimated overlay alignment metrology of the features based on one or more of alignment parameters of the wafer processing tool or alignment parameters of the wafer; and a process control tool configured to adjust one or more of the alignment parameters of the wafer processing tool or alignment parameters of the wafer based on the estimated overlay alignment metrology.

[0172] In a related embodiment, the coverage modeling tool generates a function that relates one or more of the alignment parameters of the wafer processing tool or the alignment parameters of the wafer to the coverage metric of the feature.

[0173] In a related embodiment, the system further includes a big data unit configured to combine the alignment parameters of the wafer processing tool, the alignment parameters of the wafer, and the coverage measurement values ​​of the features on the wafer into a dataset.

[0174] A computing system includes a processor and a storage unit. When the executable instructions are executed by the processor, the executable instructions configure the processor to perform various actions including: receiving data relative to a tool position of a wafer processing tool forming a first feature on a first wafer; receiving data relative to a position of the first wafer on a wafer holding tool forming the first feature on the first wafer; receiving context data relative to the formation of the first feature on the first wafer; and generating data for adjusting one or more of the tool position or the project position based on applying at least one of the tool position data, the project position data, and the context data to a regression model function, the regression model function relating at least one of the tool position data, the project position data, and the context data to a coverage metric of the first feature on the first wafer.

[0175] In related embodiments, the tool position data includes one or more of the following: x-axis tool offset, y-axis tool offset, and tool rotation.

[0176] In related embodiments, the data for the project location includes one or more of the following: x-axis wafer offset, y-axis wafer offset, wafer rotation, x-axis die offset, y-axis die offset, or die rotation.

[0177] In related embodiments, one or more of the tool position data or the project position data include measured data, pre-alignment data, and post-alignment data.

[0178] The various embodiments described above can be combined to provide other embodiments. All U.S. patents, U.S. patent application publications, U.S. patent applications, foreign patents, foreign patent applications, and non-patent publications mentioned in this specification and / or listed in the application data sheets are fully incorporated herein by reference. If necessary, aspects of the embodiments may be modified to employ various patent, application, and publication concepts to provide yet another set of embodiments.

[0179] Based on the detailed description above, these and other changes can be made to the embodiments. Generally, the terminology used in the above claims should not be construed as limiting the claims to the specific embodiments disclosed in the specification and claims, but rather should be interpreted to include all possible embodiments and the full scope of any equivalents to which these claims are given. Therefore, the claims are not limited by this disclosure.

Claims

1. A coverage management method, comprising: Receive a dataset, the dataset including a first tool position relative to a tool forming a first feature on a first wafer, a first item position of the first wafer corresponding to a first portion forming the first feature, and a first coverage measure of the first feature formed on the first wafer; Based on the dataset, a function is generated through machine learning, which correlates the tool position and the first item position on the first side of the function with the coverage metric on the second side of the function; Receive a second tool position relative to the tool forming the second feature on the second wafer and a second item position of the second wafer corresponding to the second portion forming the second feature; Based on the function, the second tool position, and the second item position, estimate the second coverage measurement of the second feature on the second wafer; as well as Based on the estimated second coverage metric, adjust the location of the second tool and the location of the second project. Each of the first item position or the second item position includes one or more of the wafer offset or wafer rotation of the first wafer positioned on the wafer stage.

2. The coverage management method according to claim 1, wherein each of the first tool position or the second tool position includes one or more of a wafer stage position, a stepper position, a photomask key position, or an exposure position.

3. The coverage management method of claim 1, wherein each of the first item location or the second item location further includes one or more of a die offset or die rotation of a die positioned on the first portion of the first wafer.

4. The coverage management method according to claim 1, wherein each of the first tool position or the second tool position includes one or more of the measured position, the pre-alignment position, and the post-alignment position.

5. The coverage management method according to claim 1, wherein each of the first item location or the second item location includes one or more of the measured location, the pre-alignment location, and the post-alignment location.

6. The coverage management method according to claim 1, further comprising: Determine the context in which the first feature is formed on the first wafer; as well as Generate the dataset containing the aforementioned context.

7. The coverage management method of claim 6, wherein the context includes one or more of the following: a measurement location for an enhanced global alignment operation on the first wafer, a measurement location relative to an exposure field forming the first feature, and a measurement location on the first wafer for measuring the coverage measurement.

8. The coverage management method according to claim 1, further comprising: Determine the position of a previous tool relative to the tool that forms the first feature on the previous first wafer; as well as Generate the dataset containing the locations of the previously used tools.

9. The coverage management method of claim 1, wherein generating the function comprises generating the function using the machine learning through a neural network system.

10. The coverage management method of claim 1, further comprising evaluating the function by comparing the estimated second coverage metric of the second wafer with the measured coverage metric of the second wafer.

11. A non-volatile computer-readable storage medium having executable instructions stored thereon, wherein when the executable instructions are executed by a processor, the executable instructions cause the processor to perform various actions including: Receive a dataset, the dataset including a first tool position relative to a tool forming a first feature on a first wafer, a first item position of the first wafer corresponding to a first portion forming the first feature, and a first coverage measure of the first feature formed on the first wafer; Based on the dataset, a function is generated through machine learning, which correlates the tool position and the first item position on the first side of the function with the coverage metric on the second side of the function; The second tool position is received relative to the tool forming the second feature on the second wafer, and the second item position is received relative to the second portion of the second wafer corresponding to the second feature; Based on the function, the second tool position, and the second item position, estimate the second coverage measurement of the second feature on the second wafer; as well as Based on the estimated second coverage metric, adjust the location of the second tool and the location of the second project. Each of the first item position or the second item position includes one or more of the wafer offset or wafer rotation of the first wafer positioned on the wafer stage.

12. The non-volatile computer-readable storage medium of claim 11, wherein each of the first tool position or the second tool position comprises one or more of a wafer stage position, a stepper position, a photomask key position, or an exposure position.

13. The non-volatile computer-readable storage medium of claim 11, wherein each of the first item location or the second item location further comprises one or more of a die offset or die rotation of a die positioned on the first portion of the first wafer.

14. The non-volatile computer-readable storage medium of claim 11, wherein the action comprises: Determine the position of a previous tool relative to the tool that forms the first feature on the previous first wafer; as well as Generate the dataset containing the locations of the previously used tools.

15. A computing system, comprising: processor; as well as A storage unit storing executable instructions that, when executed by the processor, configure the processor to perform actions including: Receive a dataset, the dataset including a first tool position relative to a tool forming a first feature on a first wafer, a first item position of the first wafer corresponding to a first portion forming the first feature, and a first coverage measure of the first feature formed on the first wafer; Based on the dataset, a function is generated through machine learning, which correlates the tool position and item first position on the first side of the function with the coverage metric on the second side of the function; The second tool position is received relative to the tool forming the second feature on the second wafer, and the second item position is received relative to the second portion of the second wafer corresponding to the second feature; Based on the function, the second tool position, and the second item position, estimate the second coverage measurement of the second feature on the second wafer; as well as Based on the estimated second coverage metric, adjust the location of the second tool and the location of the second project. Each of the first item position or the second item position includes one or more of the wafer offset or wafer rotation of the first wafer positioned on the wafer stage.

16. The computing system of claim 15, wherein each of the first tool position or the second tool position comprises one or more of a wafer stage position, a stepper position, a photomask key position, or an exposure position.

17. The computing system of claim 15, wherein each of the first tool position or the second tool position includes one or more of the measured position, the pre-alignment position, and the post-alignment position.

Images