Method of generating a computational model for improving parameter setting of one or more display manufacturing tool, method of setting parameters of one or more display manufacturing tool, and display manufacturing plant equipment
The calculation model generated by the machine learning model improves the equipment conditions of the display manufacturing factory, automated process adaptation and equipment maintenance decisions, solves the problems of complex parameter control and factors in display manufacturing, improves the yield and quality, and reduces costs and Decision-making risks.
Patent Information
- Application Number
- CN202280093954.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2022-03-21
- Publication Date
- 2024-11-08
AI Technical Summary
In the display manufacturing process, complex production parameter control and the mutual influence of multiple factors make it difficult to improve yield and quality, and existing methods have problems of high cost and decision-making risks.
The machine learning model is used to generate a computing model, and by training the input and result data of multiple substrates, it provides parameter recommendations to improve equipment conditions of the display manufacturing factory, automated process adaptation and equipment maintenance decisions.
Improve the yield and quality of display manufacturing, reduce costs, reduce the risk of invalid decision-making caused by unconsidered factors, and improve the understanding and processing ability of influencing factors.
Smart Images

Figure CN118922833A_ABST
Abstract
Description
Technical Field
[0001] Embodiments of the present disclosure relate to display manufacturing, and in particular to display manufacturing fab equipment including, for example, one or more display manufacturing tools. Embodiments of the present disclosure relate to automated display manufacturing including at least partially automated improvements in manufacturing conditions. Embodiments of the present disclosure further relate to a method of generating a computational model for improving the manufacturing of displays. For example, a machine learning model or artificial intelligence may be utilized to provide recommendations for improved display manufacturing. Background Art
[0002] Typically, displays are manufactured on large area substrates. Multiple layers are deposited on, for example, a glass substrate. These layers can be configured to generate electronic circuits including, for example, thin film transistors to control the operation of individual pixels of the substrate. For example, multiple smaller displays (such as mobile phone displays) can be manufactured on a large area substrate, or a smaller number of large substrates (for example, for television screens) can be manufactured on a large area substrate. In view of the pixel density and complexity of modern displays, both cases result in multiple layers and particularly structured layers deposited on the substrate. Each of the manufacturing steps and also the combination of manufacturing steps are constantly improved to obtain the best known methods for manufacturing displays. For example, a vacuum processing system can be used.
[0003] A vacuum processing system is a system comprising at least a vacuum chamber having a processing region, wherein a substrate can be positioned relative to the processing region to process the substrate. Several methods for depositing materials on a substrate are known. For example, a substrate can be coated by using a physical vapor deposition (PVD) process (such as a sputtering process or an evaporation process, a spraying process, etc.) or a chemical vapor deposition (CVD) process. The substrate on which the material is deposited, i.e. the substrate to be coated, is introduced into a vacuum chamber of the vacuum processing system and positioned relative to the processing region of the vacuum chamber of the vacuum processing system. Processing parameters are controlled to generate desired layer properties.
[0004] For example, in display manufacturing technology, large area substrates can consider coating processes (i.e., material deposition processes). For example, in the manufacture of larger displays, the trend towards larger substrates leads to larger vacuum processing systems. In addition, due to the increase in the number of display product types on rigid and flexible substrates with enhanced functionality, such as increased pixel resolution, enhanced color depth and dynamic refresh rate range, limited power consumption (especially related to mobile devices), and sensor capabilities integrated into the display area (such as finger touch, fingerprint sensors, under panel cameras, transparent displays), the display structure becomes more complex, which leads to increasingly complex production parameter control. Cost control through parallel display manufacturing on increasingly larger substrates adds additional complexity to process parameter control.
[0005] A large number of factors may affect particles, defects and yield of display manufacturing. Additionally, the specific coincidence of multiple effects may determine the quality of the produced display or even yield loss. Intensive statistical hypothesis testing on assumptions is a necessary attempt to evaluate the potential root causes of certain problems with a high risk of drawing invalid and / or cost-intensive conclusions on manufacturing improvement decisions.
[0006] It would be beneficial to provide improved methods of controlling factors that influence successful display manufacturing (eg, having improved yield of production and / or improved quality of manufactured displays). Summary of the invention
[0007] In view of the above, a method for generating a computational model for improving manufacturing conditions in one or more display manufacturing tools for manufacturing display devices on multiple substrates in a display manufacturing plant equipment, a method for setting conditions in a display manufacturing plant equipment for manufacturing display devices and multiple substrates, in particular parameters of one or more display manufacturing tools, and a display manufacturing plant equipment are provided. Further features, details, aspects, implementations and embodiments are shown in the dependent claims, the description and the drawings.
[0008] According to an embodiment, a method for generating a computational model for improving manufacturing conditions in one or more display manufacturing tools for manufacturing display devices on multiple substrates in a display manufacturing plant equipment is provided. The method includes: determining multiple input values for one or more display manufacturing tools, each input value is associated with a corresponding substrate identification number in a plurality of substrate identification numbers; determining multiple result input values for substrates having the plurality of substrate identification numbers, the result input values providing quantitative quality values; providing a training data set of multiple input values and multiple result input values, the multiple input values and the multiple result input values being related by corresponding substrate identification numbers; and generating or generating a computational model from the training data set by machine learning.
[0009] According to an embodiment, a method for setting conditions in display manufacturing plant equipment for manufacturing display devices on multiple substrates, in particular, parameters of one or more display manufacturing tools is provided. The method includes: calculating multiple suggested input values mapped by the computational model to one or more quantitative quality values based on a computational model; comparing the multiple suggested input values with multiple input values; and providing a recommendation to change one or more of the input values in the multiple input values to one or more of the suggested input values to generate a new set of input values. For example, the computational model is generated according to the method for generating a computational model of the embodiment described herein.
[0010] According to an embodiment, a display manufacturing plant equipment is provided. The display manufacturing plant equipment includes: at least one first display manufacturing tool, the at least one first display manufacturing tool having: one or more deposition sources, the one or more deposition sources are used to deposit material layers on a substrate in the first display manufacturing tool, and multiple sensors, the multiple sensors are used to determine actual input values; a controller, the controller is used to control setting input values; and a data processing unit, the data processing unit having a memory including instructions and having a processor, wherein the instructions, when executed by the processor, cause the data processing unit to provide a recommendation for changing one or more of the multiple setting input values. BRIEF DESCRIPTION OF THE DRAWINGS
[0011] In order to be able to understand in detail the manner in which the above-mentioned features of the present disclosure are achieved, the present disclosure briefly summarized above may be described in more detail by reference to embodiments, some of which are illustrated in the accompanying drawings. However, it should be noted that the accompanying drawings illustrate only typical embodiments of the present disclosure and therefore should not be considered as limiting the scope, as the present disclosure may admit of other equally effective embodiments.
[0012] Figure 1 A schematic diagram showing a configuration of display manufacturing factory equipment including a display manufacturing tool, an electron beam test tool, an electron beam review tool, and an optical inspection tool according to an embodiment of the present disclosure is shown.
[0013] Figure 2 A diagram illustrating an architecture for generating and / or utilizing values provided by a computational model according to an embodiment of the present disclosure is shown.
[0014] Figure 3 A flow chart depicting a method of generating a computational model for improving manufacturing conditions for display manufacturing according to an embodiment of the present disclosure is shown.
[0015] Figure 4A A graph is shown plotting quantitative quality values as a function of parameters and the improvements made by setting conditions of display manufacturing plant equipment.
[0016] Figure 4B A general model of a neural network that may be used with embodiments of the present disclosure is depicted.
[0017] Figure 5 A flowchart showing a method of setting conditions in display manufacturing factory equipment according to an embodiment of the present disclosure is shown.
[0018] Figure 6 A functional block diagram of data processing and control in manufacturing plant equipment according to an embodiment of the present disclosure is depicted. DETAILED DESCRIPTION
[0019] Reference will now be made in detail to various embodiments of the present disclosure, some examples of which are illustrated in the accompanying drawings. Methods of generating computational models for improving display manufacturing are described, as well as methods of utilizing computational models, for example, by setting conditions in display manufacturing plant equipment, such as parameters of display manufacturing tools. For example, training a machine learning architecture or an artificial intelligence architecture to obtain a computational model, as well as utilizing a computational model are described. It will be apparent to those skilled in the art that embodiments of the present disclosure may be practiced without all of the details provided for the detailed embodiments.
[0020] Embodiments of the present disclosure may combine feedback from various result parameters to utilize available manufacturing equipment status, process data, hardware data, consumables data, and media supply data. Result parameters may include, for example, display tests from electron beam test (EBT) tools, electron beam review (EBR) equipment, and multiple other results as described below. Optionally, input values may be further provided by preventive maintenance data and intermediate products, and the results may include final quality or yield data surveys. The data is processed by a machine learning unit or an artificial intelligence unit to generate or implement recommendations for process adaptation, hardware use, system use, and / or maintenance.
[0021] Embodiments allow for the discovery of new and / or hidden display product improvement factors, faster identification of new and / or hidden factors and known factors, and / or the ability to automatically feedback and verify improvements. Thus, the likelihood of invalid decisions based on missing the right factors to consider is reduced. Processing and detection capabilities can be provided across a large number of complex, interdependent influencing factors. Additionally, the understanding of equipment improvement factors (Pareto chart of influencing factors) can be improved.
[0022] Embodiments of the present disclosure provide concepts for display manufacturing and diagnostic data inputs processed via machine learning or artificial intelligence to recommend or automatically implement process adaptations and provide product sequencing decisions or equipment maintenance or part replacements to optimize production defects, yield, productivity and reduced costs.
[0023] Figure 1 A display manufacturing plant equipment 100 is shown. The display manufacturing plant equipment 100 includes one or more display manufacturing tools. For example, Figure 1 Two in-line deposition tools 110 and two cluster deposition tools 120 are shown. The display manufacturing plant equipment 100 may further include display testing tools, such as an e-beam testing tool 130, an e-beam inspection tool 140, and / or an optical inspection tool 150. The display manufacturing plant equipment further includes a controller 190. According to some embodiments, which may be combined with other embodiments described herein, the data center 102 may be disposed within the display manufacturing plant equipment or outside the display manufacturing plant equipment 100.
[0024] According to some embodiments, the controller 190 may be a system computer or a data processing unit. The controller 190 may include a system computer or a data processing unit. The system computer may process or preprocess the data according to an embodiment of the present disclosure. If the system computer preprocesses the data, the data may be sent to the data center 102. The data center 102 may be a host computer or a cloud computer. The data center 102 may include a big data infrastructure and / or may store incoming data, such as all incoming data. According to some embodiments that may be combined with other embodiments described herein, the data center may, for example, generate or generate a computing model by training a machine learning algorithm, in particular by training an artificial intelligence algorithm. According to some embodiments that may be combined with other embodiments described herein, the data center may include a data processing unit for training a computing model or providing recommendations based on a computing model. Some or all of the tasks that may be associated with the data center 102 may be performed by the controller 190 of the display manufacturing plant equipment 100.
[0025] According to some embodiments, one or more inline deposition tools 110 may be provided. Figure 1 As shown in the exemplary embodiment, the inline deposition tool may include a plurality of vacuum chambers 111. The plurality of vacuum chambers may include at least one production chamber, for example, including a deposition source (such as Figure 1 The rotatable sputtering cathode 112 is shown in Figure 1. The rotatable sputtering cathode can deposit material on a substrate, such as a large area substrate for manufacturing a display by a PVD process (such as a sputtering process). The plurality of vacuum chambers may include at least one of a load lock chamber, a transfer chamber, a pre-processing chamber, and a vacuum spin chamber, wherein the substrate can be rotated under vacuum.
[0026] The inline deposition tool 110 includes a plurality of first components 114 to set parameters and / or set input values. The plurality of first components may include a heater, a cooler, a gas flow setting valve, a gas flow control valve, a power supply, a vacuum pump. For example, the power supply may provide power to the deposition source. In addition, the plurality of first components may additionally or alternatively include deposition materials, consumable components, etc.
[0027] The inline deposition tool includes a plurality of second components 116 to determine or measure actual input values. The plurality of second components may include sensors, such as one or more sensors for one or more temperatures, one or more gas flow rates, one or more currents (such as deposition source currents), one or more voltages (such as deposition source voltages), one or more power values (such as deposition source power); sensors for arc detection; pressure gauges; and sensors for residual gas analysis.
[0028] According to some embodiments, which may be combined with other embodiments described herein, the inline deposition tool 110 may include a controller 119. The controller 119 may set a plurality of setting input values and may determine or measure actual input values. Additionally, according to some embodiments, the controller 119 may communicate with the controller 190 and / or the data center 102.
[0029] Additionally or alternatively, one or more cluster deposition tools 120 may be provided. Figure 1 The cluster deposition tool 120 exemplarily shown in FIG. 1 may have similar components to the inline deposition tool 110. For example, a vacuum chamber 211 is provided. The production chamber, the load lock chamber, and optionally other chambers are connected to a central transfer chamber in the cluster deposition tool. For example, a robot with a processing arm may be disposed in the central transfer chamber to move a substrate from one vacuum chamber to another vacuum chamber for substrate processing.
[0030] According to some embodiments, which can be combined with other embodiments described herein, the production chamber can be a vacuum chamber including a PVD deposition source 122. Alternatively, the production chamber can be a vacuum chamber including a CVD deposition source. According to some embodiments, which can be combined with other embodiments described herein, one or more display manufacturing tools for structuring (e.g., etching, pre-processing or post-processing) can be provided in a display manufacturing factory.
[0031] Figure 1The vacuum pump 125 at one of the vacuum chambers of the cluster deposition tool 120 is exemplarily shown. According to some embodiments, which can be combined with other embodiments described herein, one or more vacuum pumps can be provided at the vacuum chambers of the display manufacturing tool, in particular at each of the display manufacturing tools. Additionally or alternatively, a vacuum pump holder having a plurality of vacuum pumps connected to the vacuum chambers can be provided.
[0032] As described above, the cluster deposition tool may include a transfer chamber. The transfer chamber may be a central transfer chamber. The robot may be at least partially disposed within the transfer chamber. The robot may have a robotic arm. The robot may transfer substrates between chambers coupled to the transfer chamber. At least one load lock chamber may be coupled to the transfer chamber. One or more deposition devices (i.e., deposition chambers) may be coupled to the transfer chamber. The robot may transfer substrates between the load lock chamber and the deposition chamber, and vice versa, or between different deposition chambers attached to the transfer chamber.
[0033] According to yet further embodiments, one or more other processing chambers may be coupled to the vacuum transfer chamber, such as the central transfer chamber. Specifically, one or more other processing chambers may be selected from a heating chamber coupled to the transfer chamber, a cooling chamber coupled to the transfer chamber, a pre-cleaning chamber coupled to the transfer chamber, a storage chamber coupled to the transfer chamber, an inspection chamber coupled to the transfer chamber, and a CVD chamber coupled to the transfer chamber. One or more of the above chambers of the same type and / or different types may be coupled to the central transfer chamber. For example, the inspection chamber may measure the thickness of a layer deposited in a previous deposition process or may control the thickness of one or more layers before the substrate is unloaded from the processing system. Control of layer thickness may be provided. A cleaning or pre-cleaning chamber may remove oxides from, for example, a metal layer or may remove photoresist residues from a previous manufacturing operation.
[0034] As described with respect to the inline deposition tool 110, the cluster deposition tool 120 may also include a plurality of first components 124 and a plurality of second components 126. The plurality of first components 124 set parameters (i.e., operating conditions) and / or set input values. The plurality of second components 126 measure or determine parameters (i.e., operating conditions) and / or actual input values.
[0035] According to some embodiments, which may be combined with other embodiments described herein, the cluster deposition tool 120 may include a controller 129. The controller 129 may set a plurality of setting input values and may determine or measure actual input values. Additionally, according to some embodiments, the controller 129 may communicate with the controller 190 and / or the data center 102.
[0036] The factory equipment, tools, devices and systems described herein can be configured to move and handle, in particular, objects having a size of 1 m 2or a large area substrate or a large area carrier with a surface of or larger. The term "substrate" may particularly include substrates similar to glass substrates, such as glass plates. In addition, the substrate may include a wafer of a transparent crystal such as sapphire. However, the term "substrate" may include other substrates that may be non-flexible or flexible, such as foils or coils. The substrate may be formed of any material suitable for material deposition. According to some embodiments of the present disclosure that may be combined with other embodiments described herein, the substrate is configured for display manufacturing, and the substrate may be particularly a large area substrate. The substrate may be processed in a vertical orientation, particularly in an in-line deposition tool. Alternatively, the substrate may be processed and / or transported in a horizontal orientation (e.g., in a cluster deposition tool). The terms "horizontal" and "vertical" allow deviations from a completely horizontal or vertical orientation, such as deviations of ±15° or less. For example, some applications such as μOLED, μLED, miniLED may utilize smaller substrates. For example, an 8" Si wafer may be used. Such a substrate may be placed on a larger carrier to move and process the substrate, particularly a substrate having a diameter greater than 0.03 m. 2 or larger surface substrate or with a surface of 0.1m 2 or larger surface area substrates. The substrate and carrier can have a surface area of up to 10m 2 and larger sizes.
[0037] Embodiments described herein relate particularly to depositing materials on large area substrates, for example for display manufacturing. According to some embodiments, a large area substrate or carrier supporting one or more substrates may have a thickness of at least 0.1 m 2 For example, the deposition system may be suitable for processing large area substrates, such as substrates with a size of approximately 1.4 m 2 (1.1m×1.3m) substrate, GEN 5 substrate corresponding to approximately 4.29m 2 The GEN 7.5 substrate corresponds to the substrate of (1.95m×2.2m), and the GEN 7.5 substrate corresponds to the substrate of approximately 5.7m 2 (2.2m x 2.5m) substrates corresponding to the GEN 8.5 substrate, or even with approximately 8.7m 2 (2.85m×3.05m) substrates corresponding to GEN 10 substrates. Even larger generations such as GEN 11 and GEN 12 and corresponding substrate areas may be similarly implemented. According to yet further implementations, half the size of the above-mentioned generations of substrates may be processed. Alternatively or additionally, semiconductor wafers may be processed and coated in a deposition system according to the present disclosure.
[0038] According to some embodiments, which may be combined with other embodiments described herein, the display manufacturing plant equipment 100 further includes a display testing tool. Figure 1As shown, the display test tool may be an electron beam test (EBT) tool. Alternatively, a display test tool may be provided that may capacitively couple the optical device to the pixel electrodes of the display.
[0039] For EBT, an electron beam is used to test multiple or each pixel structure including, for example, a thin film transistor (TFT) and a pixel electrode. EBT allows testing of pixels of a display, testing of lines (ie, source and drain lines) and / or display driver circuit testing. Figure 1 An electron beam test tool 130 is shown including a plurality of electron beam columns 132 and a controller 139 for the test tool.
[0040] According to some embodiments that may be combined with other embodiments described herein, the display manufacturing plant equipment further includes an electron beam inspection tool 140. The electron beam inspection tool includes two scanning electron microscopes 142 (SEM) and can provide a focused ion beam (FIB) column 144. The display inspection tool provides SEM images. SEM images can be used for critical dimensions (CD) and / or defect review (DR). In contrast to EBT tools having a field of view of 10 cm or more, EBR tools have a field of view of 1 mm or less. The EBR tool provides high resolution, i.e., SEM images. The electron beam inspection tool 140 includes a controller 149.
[0041] According to some embodiments, which can be combined with other embodiments described herein, for inspecting a display device having at least 1.375 m 2 An EBR tool for large area substrates of a size of may include: a vacuum chamber; a substrate support arranged in the vacuum chamber, wherein the substrate support is configured to support a large area substrate for display manufacturing and provide a substrate receiving area having a first receiving area size along a first direction; and a first imaging charged particle beam microscope, the first imaging charged particle beam microscope being configured to generate a charged particle beam for inspecting a substrate supported by the substrate support and configured to generate an image of a portion of the substrate, the first imaging charged particle beam microscope having a field of view less than 1 mm. Further optionally, the first imaging charged particle beam microscope may include: a retarding field lens component of an objective lens; and a second imaging charged particle beam microscope, the second imaging charged particle beam microscope being adjacent to the first imaging charged particle beam microscope and being 30 cm or more away from the first imaging charged particle beam microscope along the first direction.
[0042] According to some embodiments that may be combined with other embodiments described herein, the display manufacturing plant equipment may further include an optical inspection tool 150. The optical inspection tool 150 may include a light source, such as a laser 152 that emits an optical beam 153. The beam is directed onto the substrate for optical inspection with a beam guiding element 154. The optical inspection tool may be an automated optical inspection tool. In addition, the optical inspection tool may include a controller 159. According to some embodiments that may be combined with other embodiments described herein, the optical inspection may inspect the substrate, in particular the entire substrate. Optical inspection with the optical inspection tool 150 may generate a map of substrate coordinates with potential display defects. The map with potential display defects may be transmitted to the EBR tool 140, where the optical inspection at the relevant substrate coordinates has SEM resolution. Defect review or critical dimensioning may be performed at the desired substrate coordinates. For example, defect classification may be provided.
[0043] The display test tool, EBR tool, and / or optical inspection tool described above can be used to determine a plurality of result values for a previously processed substrate. The result input values provide a quantitative quality value. A training data set can be provided based on the input values and result input values used to generate a computational model.
[0044] According to one embodiment, a display manufacturing plant equipment is provided. The display manufacturing plant equipment includes at least a first display manufacturing tool. At least the first display manufacturing tool includes one or more deposition sources for depositing material layers on a substrate in the first display manufacturing tool and multiple sensors for determining actual input values. At least the first display manufacturing tool includes a controller for controlling setting input values. The display manufacturing plant equipment includes a data processing unit having a memory including instructions and having a processor, wherein the instructions, when executed by the processor, cause the data processing unit to provide a recommendation for changing one or more input values of a plurality of input values.
[0045] According to some embodiments that may be combined with other embodiments described herein, the instructions, when executed by a processor, cause a data processing unit to provide a recommendation for changing one or more of a plurality of input values by executing a method for setting conditions in display manufacturing factory equipment, particularly parameters of one or more display manufacturing tools, for manufacturing display devices on a plurality of substrates according to an embodiment of the present disclosure.
[0046] According to some embodiments that may be combined with other embodiments described herein, the display manufacturing factory equipment may further include a display testing tool (such as an electron beam testing tool) and be configured to provide result input values for substrates having multiple substrate identification numbers, the result input values providing a quantitative quality value including at least one result from a display pixel test, a display line test, and a display driver circuit test.
[0047] According to some embodiments that may be combined with other embodiments described herein, the display manufacturing plant equipment further includes: an electron beam inspection tool configured to generate a SEM image of a portion of one or more layers on a display substrate generated by the display manufacturing plant equipment; and / or an automated optical inspection tool configured to generate a light optical image of a portion of one or more layers on a display substrate generated by the display manufacturing plant equipment. According to still further additional or alternative embodiments, a tool for providing a result input value may be provided. Tools for providing result input values may be selected from metrology for determining physical, electrical and optical film properties (such as profilometry (thickness), resistance measurement methods (4-point probe or eddy current detection), photometry (particularly spectrophotometer), and ellipsometry for measuring optical parameters and film optical and thickness uniformity), automated optical inspection (AOI), MW-PCR / MW-PCD (microwave detected photoconductivity response and decay) for indicating carrier concentration and lifetime (e.g. for amorphous oxide semiconductors such as IZO, IGZO and other metal oxides), AFM (film roughness), and XRD (film crystallinity). Additionally or alternatively, laser particle scanning (LPS) may be provided to determine particle adders that occur during specific manufacturing steps.
[0048] Other tools for determining TFT properties may be provided. Tools for determining TFT properties may be selected from measurements of TFT output properties (drain-source current versus drain-source voltage for various gate voltages), transfer characteristics (drain-source current versus gate-source voltage), including determination of threshold voltage. Other tools may be BTS (bias temperature stress) to measure threshold voltage shifts (e.g., NBTS and PBTS = negative BTS and positive BTS), illumination stress measurements, current leakage measurements, CV measurements (capacitance voltage measurements). In addition, for displays including light-emitting materials, such as LEDs, OLEDs, or quantum dot LEDs, performance and lifetime measurement tools may be provided, such as photoluminescence tools, electroluminescence measurements (such as IVL measurements), and / or OLED lifetime measurement tools. According to some embodiments that may be combined with other embodiments described herein, metrology tools for at least one of display performance and display lifetime may be provided.
[0049] Figure 2 A diagram illustrating an architecture including values provided for generating and / or utilizing a computational model according to an embodiment of the present disclosure is shown. In addition, Figure 3 A flow chart illustrating a method for generating a computational model is shown. Figure 2 and Figure 3Described together to illustrate the improvement of manufacturing conditions in one or more display manufacturing tools in display manufacturing plant equipment, particularly for manufacturing display devices on multiple substrates. According to some embodiments that can be combined with other embodiments described herein, multiple substrates are processed sequentially in the manufacturing plant equipment. Multiple substrates can be provided simultaneously in the display manufacturing plant equipment, wherein different substrates of the multiple substrates are simultaneously subjected to different processing operations, transportation operations, and testing operations. One substrate can be processed in a separate processing station at a time. A separate processing station processes a substrate or a carrier with multiple substrates one by one.
[0050] According to one embodiment, a method for generating a computational model for improving manufacturing conditions in one or more display manufacturing tools for manufacturing display devices on multiple substrates in display manufacturing plant equipment is provided. In operation 302, multiple input values of one or more display manufacturing tools are determined, wherein each input value is associated with a corresponding substrate identification number in a plurality of substrate identification numbers. In operation 304, multiple result input values for substrates having multiple substrate identification numbers are determined, wherein the result input value provides a quantitative quality value. For example, the quantitative quality value may include at least one result from a display pixel test, a display line test, and a display driver circuit test. In operation 306, a training data set of multiple input values and multiple result input values is provided. The multiple input values and the multiple result input values are related by corresponding substrate identification numbers. In operation 308, a computational model is generated or generated from the training data set by machine learning.
[0051] At a first specific time, a first substrate among a plurality of substrates undergoes a specific manufacturing operation. The first substrate has a first substrate identification number. In addition, other substrates have other substrate identification numbers. A plurality of substrate identification numbers are provided. Exemplarily, reference is made to the first substrate and the first substrate identification number. The first specific time at which the first substrate among the plurality of substrates undergoes the specific manufacturing operation is associated with an identifier timestamp among a plurality of identifier timestamps.
[0052] Other identifier timestamps are associated with the first substrate for a subsequence of manufacturing operations occurring after a first specific time. Thus, for some embodiments that may be combined with other embodiments described herein, each of a plurality of input values is associated with a corresponding identifier timestamp in a plurality of identifier timestamps. The first substrate having a first substrate identification number may be subjected to testing, inspection, or other measurement or metrology, such as providing layer characteristics of a previously deposited layer, electrical characteristics of a layer or thin film transistor, optical film properties, yield inspection, and the like. A plurality of input values and a plurality of result input values may be associated by corresponding identifier timestamps and may be associated by, for example, a substrate identification number of the first substrate. Manufacturing conditions may be associated with result input values that provide quantitative quality values. Thus, a training data set for a machine learning architecture may be provided.
[0053] According to some embodiments, which may be combined with other embodiments described herein, multiple identifier timestamps may be used to determine a time period between generation of an input value and generation of a result input value for a substrate having the same substrate identification number. Determining the result input value may vary depending on the time period that has passed between a manufacturing operation (or a processing operation) and a metrology operation. Additionally, multiple identifier timestamps may be utilized to determine a time period between subsequent manufacturing steps for a substrate having the same substrate identification number.
[0054] The manufacture of a display as described herein involves the manufacture of multiple layers of a display. According to some embodiments of the present disclosure, when referring to the manufacture of a display, it should be understood that one or more layers of the display or one or more structures of the display are manufactured. The manufacture of a display is not limited to the manufacture of an entire display. According to an embodiment of the present disclosure, the manufacture of a display may produce an intermediate product, which may ultimately produce a display or be intended to produce a display. According to some embodiments that may be combined with other embodiments described herein, the manufacture of a display may also involve the manufacture of a test display or a dummy display. For example, a large area substrate (glass substrate) may be loaded in a display manufacturing tool, and only one layer or one layer stack may be provided on the substrate. Additionally or alternatively, only one layer or only one layer stack may be processed in a display manufacturing tool. The test display or dummy display may be transferred to one or more metrology tools to provide one or more of the metrology operations described herein to the test display or dummy display.
[0055] like Figure 2 As shown, the input values 210 may include a plurality of set input values 212. In addition, the input values may include actual input values 214. In addition, optionally, the input values 210 may include historical values 216, such as input values or result input values of substrates processed before the first substrate (i.e., in the past). According to some embodiments that may be combined with other embodiments described herein, the plurality of input values may include: a plurality of set input values of parameter settings from one or more display manufacturing tools; and a plurality of actual input values of measured (measurand) inputs from one or more display manufacturing tools. The set input values correspond to parameter settings and predetermined values and manufacturing conditions.
[0056] According to some embodiments that may be combined with other embodiments described herein, the setup input value includes at least one setup input value selected from the group consisting of: temperature setting, gas flow setting, deposition source power source setting, warning threshold setting, deposition material, vacuum pressure setting, display product type, display substrate size, display pixel resolution, consumable part material and consumable composition. Additionally or alternatively, the actual input value includes at least one actual input value selected from the group consisting of: temperature, gas flow, deposition source current, deposition source voltage, deposition source power, deposition source arc detection tracking, vacuum pressure, residual gas analysis and error warning.
[0057] Setting input values and actual input values allows distinguishing between set points or target values and process values of manufacturing conditions. Therefore, the difference between the target value and the process value can be determined. In addition, the time change (i.e., trend) of the difference between the target values in the process value can be evaluated, and the time change can serve as an input to the machine learning algorithm. By making the multiple input values include multiple historical values, the time change or trend of the difference between the target values in the process value can be further supported. According to some embodiments that can be combined with other embodiments described herein, the multiple historical values may include multiple historical input values (set input values or actual input values) of one or more display manufacturing tools and may include multiple historical result values of one or more metrology tools.
[0058] As described above, the plurality of historical values may be associated with a historical timestamp at a time point prior to the timestamp corresponding to the corresponding substrate identification number. For example, the historical timestamp may refer to a substrate that has undergone a manufacturing step prior to the current substrate to be evaluated. According to some embodiments that may be combined with other embodiments described herein, the plurality of historical values are at least one historical value selected from the group consisting of: historical values of set input values, historical values of actual input values, historical values of result input values, production equipment maintenance values, tool component replacement values, and consumable exchange values.
[0059] The result input value 220 may include, for example, a plurality of test result values 222 from an EBT tool. The result input value 220 may further include, for example, a plurality of EBR result values 224 from an EBR tool. In addition, optionally, the result input value 220 may include values 226 from optical inspection (such as automated optical inspection (AOI)), film properties, TFT properties, and / or yield. According to some embodiments that may be combined with other embodiments described herein, the quantitative quality value includes: one or more results selected from display pixel testing, display line testing, and display driver circuit testing; results from an electron beam review tool for display manufacturing; results from an automated optical inspection tool; results of film properties; results of thin film transistor properties; results from intermediate or final displays; yield data analysis; and the probability of non-failing products. In particular, the test result value 222 may be one or more of a display pixel test, a display line test, and a display driver circuit test, which may be provided with an electron beam test tool for display manufacturing.
[0060] According to some embodiments, which may be combined with other embodiments described herein, the quantitative quality value may include a qualitative value or a probability or number of categorical data. For example, the probability or number of yield binning types (e.g., type A, B, C) of defect classification, the probability of open line, short line, gate short, gate open, dark pixel, white pixel, and particle may be used for the training data set. The probability or number of data such as vectors and matrices providing defect locations may also be used.
[0061] The plurality of input values 210 and result input values 220 described above may be combined into a training data set according to operation 306. The training data set may be provided to a machine learning algorithm to generate or produce a computational model 200 (see Figure 2 ). As described below, particularly with respect to Figure 4A and Figure 4B , a computational model can be provided to generate a new recipe 205. Figure 2 As shown, new recipe 205 may be provided to controller 250 to provide a new set of setup input values.
[0062] According to one embodiment, a method for setting conditions in display manufacturing plant equipment for manufacturing display devices on multiple substrates, in particular parameters of one or more display manufacturing tools, may be provided. The method includes: calculating a plurality of suggested input values mapped by the computational model to one or more quantitative quality values based on a computational model; and comparing the plurality of suggested input values with a plurality of input values. Thus, the method may provide a recommendation to change one or more of the plurality of input values to one or more of the suggested input values to generate a new set of input values.
[0063] According to some embodiments that may be combined with other embodiments described herein, improving manufacturing in one or more display manufacturing tools may include one or more of the following: improving parameter settings of one or more manufacturing tools, improving production equipment maintenance of one or more manufacturing tools, improving product type selection, replacing manufacturing consumables (such as evaporation materials, sputtering targets, CVD precursor materials, gas purifiers, or particulate filters).
[0064] In particular, by utilizing the historical values 216, recommendations for recipe changes, i.e., changing the properties or materials of the layers deposited in a particular manufacturing tool, can be provided. For example, the computational model may recommend changing a first material that is not deposited from a deposition tool to a second material that is subsequently deposited in the same deposition tool. For example, the second material may be sensitive to oxygen previously used during the deposition of the first material. As another example, the computational model may recommend changing the recipe or material to be deposited in the tool, or may recommend, for example, pausing manufacturing in the tool when the temperature of the entire system increases over time. In addition, the computational model may recommend using another display manufacturing tool in a display manufacturing plant equipment, wherein a cluster of display manufacturing tools is provided. In addition, the computational model may recommend changing the intervals between production equipment maintenance based on historical values or field service organization staff to perform production equipment maintenance. Production equipment maintenance may include replacing parts of the tool or may include cleaning or other maintenance operations of the tool. For example, cleaning may include cleaning of shields, targets of deposition sources, gas ducts, or other components. For example, maintenance operations may include cleaning operations, target replacement, and / or gas purifier or filter replacement, protective shield replacement. According to yet further embodiments, additionally or alternatively, the computational model may recommend parameter settings for one or more manufacturing tools, such as temperature settings, power settings, voltage settings, current settings, gas flow rates, and composition settings, etc. According to yet further embodiments, additionally or alternatively, the computational model may recommend pausing specific display manufacturing operations, for example, to avoid or reduce idle time between specific substrate processing operations in display manufacturing plant equipment.
[0065] According to some embodiments, which can be combined with other embodiments described herein, the computational model can be generated or produced by any of the embodiments described herein. Figure 2 and Figure 3 Some features, details, and aspects of generative computational models are described.
[0066] According to some embodiments, improved conditions or optimized conditions for setting up equipment in a display manufacturing plant may be provided based on a mathematical function of a multi-dimensional computing model. Figure 4AA graph is shown in which the quantitative quality value (z-axis) is shown as a function of a parameter or condition. The parameter or condition can be any input value, such as a set value, an actual input value, a historical value, or a combination thereof. Exemplarily, a first input value is shown on the X-axis, and a second input value is shown on the Y-axis. The quantitative quality value is shown as a function of two axes. It will be appreciated by those skilled in the art that the number of input values for display manufacturing plant equipment is typically greater than two, i.e., the multidimensional model includes a large number of dimensions.
[0067] The quantitative quality value can be shown as Figure 4A 410 , at least a local maximum or advantageously a global maximum. According to some embodiments, which may be combined with other embodiments described herein, display manufacturing plant equipment is typically run with the best known method (BKM). As a result of the training data set, a computational model may be provided as a mathematical function, such as Figure 4A The function that generates the picture shown. For mathematical functions, the derivative of the function can be provided to find the maximum value of the function. According to some embodiments that can be combined with other embodiments described herein, the operation of optimizing the operating conditions may include determining whether the derivative of the multidimensional function of the calculation model is zero.
[0068] The present disclosure relates to the improvement of quality values. Therefore, reference is made to local maxima or global maxima. In general, optimization may involve finding extreme values, and, for example, production time may be minimized rather than maximized. Since production time may also be related to measures such as reduction of production speed or production time, which will have a maximum value as the desired improvement effect, the quality value is described as a maximum value. Similarly, material consumption will be minimized. On the other hand, the reduction of material consumption may be maximized again. Therefore, reference is made to increased quality values (quality is a positive characteristic), where it should be understood that, depending on the term of the quality value, a minimum value may also be a beneficial effect.
[0069] According to some embodiments, particularly for complex condition landscapes of display manufacturing plant equipment including set input values, actual input values, historical values, or combinations thereof, it may be desirable to determine a maximum accessible value. For example, there may not be a global maximum in the multidimensional input value space (phase space), but rather a tendency to improve conditions in the multidimensional phase space that are favorable toward the corners or edge regions of the training data set. Therefore, the operation of optimizing operating conditions may include determining whether a derivative of a multidimensional function of a computational model is a minimum or determining whether a multidimensional function of a computational model has a maximum value.
[0070] Without limiting the scope of the present disclosure, it should be noted that the tools in the display manufacturing plant equipment can be implemented as display manufacturing tools, test machines, various types of inspection machines, such as optical imaging machines, electron beam inspection machines, etc. For some metrology tools, an entire large area substrate (e.g., an entire glass substrate of a particular generation (GEN X)) can be inspected or tested to detect potential defects, to detect layer characteristics and / or other result input values. In other cases, the metrology tool can be an audit tool, which generally has a higher resolution and is used to find out whether a potential defect is indeed a defect or to which defect type or category it is attributed. Such audit tools (EBR) are generally configured to inspect segments of the substrate one at a time at a higher resolution.
[0071] A general model of an exemplary deep neural network that can be used as DNN 422 is shown in Figure 4B Even if Figure 4B A DNN is shown, i.e., a neural network with more than three layers, and a neural network with fewer layers may also be used. The following description with reference to a DNN may be similarly applied to general neural networks (NNs) and other artificial intelligence (AI) computing models, particularly machine learning (ML) computing models.
[0072] The illustrated exemplary DNN includes a DNN module 424 having an input layer 431, an output layer 433, and one or more hidden layers for the NN (denoted as 432-1 to 432-i) or two or more hidden layers for the DNN disposed between the input layer and the output layer. Optionally, the DNN includes a pre-DNN module 423 and a post-DNN module 425.
[0073] Each layer of the DNN module 424 may include a plurality of basic computing elements (CEs) 434, which are typically referred to in the art as dimensions, neurons, or nodes. The CEs included in the input layer are Figure 4B 435. The computational elements of a given layer are connected to the CEs of the subsequent layer through connections 435. Each connection 435 between a CE of a previous layer and a CE of a subsequent layer is associated with a weighted value (for simplicity, in Figure 4B not shown).
[0074] A given hidden CE may receive input from a CE of a previous layer via a corresponding connection, each given connection being associated with a weighted value that may be applied to the input of the given connection. The weighted value may determine the relative strength of the connection, and thus determine the relative influence of the corresponding input on the output of the given CE. A given hidden CE may be configured to calculate an activation value (e.g., a weighted sum of the inputs) and further derive an output by applying an activation function to the calculated activation. The activation function may be, for example, an identity function, a deterministic function (e.g., linear, sigmoid, threshold, etc.), a random function, or other suitable function. The output from a given hidden CE may be transmitted to a CE of a subsequent layer via a corresponding connection. Similarly, as described above, each connection at the output of a CE may be associated with a weighted value, which may be applied to the output of the CE before being received as an input to a CE of a subsequent layer. In addition to the weighted value, there may also be a threshold value (including a limit function) associated with the connection and the CE.
[0075] The weight values and / or thresholds of the (deep) neural network may be initially selected prior to training and may be further iteratively adjusted or modified during training to achieve an optimal set of weight values and / or thresholds in the trained DNN module. After each iteration, the difference between the actual output produced by the DNN module and the target output associated with the corresponding training data set may be determined. The difference may be referred to as an error value. Training may be determined to be complete when an error function or loss function representing the error value is less than a predetermined value, or when a limited change in performance between iterations is achieved.
[0076] The DNN input data set used to adjust the weights / thresholds of the deep neural network is referred to as a training data set in this article. The input to DNN 422 may be pre-processed by a pre-DNN module 423 before being input to the DNN module 424, and / or the output of the DNN module 424 may be post-processed by a post-DNN module 425 before being output from the DNN 422. In this case, the training of DNN 422 further includes determining the parameters of the pre-DNN module and / or the post-DNN module. While the parameters of the pre-DNN module and / or the post-DNN module can be predefined, and optionally, the parameters of the pre-DNN module and / or the post-DNN module can be adjusted during training, the DNN module can be trained to minimize the loss function of the entire DNN. The training-based parameter set may further include parameters related to pre-DNN and post-DNN processing.
[0077] It should be noted that the teachings of the subject matter of the present disclosure are not constrained by the number of hidden layers and / or DNN architecture. As non-limiting examples, the layers in the DNN can be convolutional, fully connected, locally connected, pooled / sub-sampled, recursive, etc.
[0078] According to some embodiments that can be combined with other embodiments described herein, a computational model for improving manufacturing conditions in display manufacturing plant equipment can be trained by a training data set according to an embodiment of the present disclosure. In order to set the conditions in the display manufacturing plant equipment so as to improve the quantitative quality value, a DNN 422 with at least partially frozen weights can be back-propagated. Thus, optimization or improvement of manufacturing conditions such as parameter settings of one or more display manufacturing tools can be provided by back-propagating the computational model.
[0079] In view of the above, a method of generating a computational model (e.g., training a machine learning algorithm) and a method of using the computational model to set conditions in display manufacturing plant equipment can be combined. Thus, iterative improvements in operating conditions in display manufacturing plant equipment can be provided. Figure 5 As shown, the automatic improvement of manufacturing conditions in display manufacturing plant equipment begins at operation 502. In operation 504, as described with respect to Figure 3 Explanation is made illustratively, generating a computational model. The computational model may be generated according to any of the embodiments described herein and specifically includes a plurality of input values and a plurality of result input values. The plurality of input values may be set input values, actual input values, historical values, or a combination thereof. According to an embodiment of the present disclosure, the plurality of result input values provide a quantitative quality value. In operation 506, the quality value may be increased.
[0080] According to some embodiments that can be combined with other embodiments described herein, in operation 508, a plurality of suggested input values can be calculated based on the computational model. The suggested input values are mapped by the computational model to one or more quantitative quality values with increased quantitative quality values, i.e., the manufacturing conditions are improved or optimized by the computational model. In operation 510, a recommendation can be provided by comparing the plurality of suggested input values with the plurality of input values. In operation 512, a new input value set is generated based on the recommendation. The recommendation provides an improved quantitative quality value for the new input value set compared to the given input value set.
[0081] A new display is at least partially manufactured with a new input value set. A display or a portion or layer of a display that is at least partially manufactured or fully manufactured is tested, inspected, and / or measured with one or more of the metrology tools described herein. For example, a display that is partially manufactured or fully manufactured can be tested with a test tool, such as an electron beam test tool for pixel testing, a line test for dedicated testing. The layer of a partially manufactured display can be inspected by an electron beam review tool. The optical, electrical, or other properties of the layer or component of the display can be tested. The yield of the display manufactured on the substrate can be determined. In operation 514, any one of the result values described herein can be provided. In operation 516, the new input value set and the newly generated result value set can be used for a training data set. The computational model can be adaptive to the adaptive training data. The adaptive computational model can be further used to further improve the condition settings of the display manufacturing plant equipment. After one or more iterations, the process can end at operation 518.
[0082] In view of the above, the process of setting conditions in display manufacturing plant equipment may further include: manufacturing a display with a new set of setting input values; generating at least one result from a display pixel test, a display line test, and a display driver circuit test to obtain a new result input value; and adapting the computational model using adaptive training data including the new set of setting input values, new result input values, and a new action input value set.
[0083] According to some embodiments that may be combined with other embodiments described herein, the provision of the recommendation includes setting manufacturing parameters for one of a plurality of display manufacturing tools and may further include at least one of the following items: the type of display product to be manufactured, the type of display product not to be manufactured, production equipment maintenance, replacement of tool components, replacement of consumables, selection of tool components, and generation of quantitative quality values.
[0084] Figure 1 A display manufacturing plant setup 100 is shown. Figure 1 The display manufacturing plant equipment 100 shown includes controllers 119 and controllers 129 for various manufacturing tools. In addition, controllers 139, 149, and 159 for metrology tools are shown. The display manufacturing plant equipment 100 may further include a controller 190. The data center 102 may be disposed inside or outside the display manufacturing plant equipment. According to some embodiments that may be combined with other embodiments described herein, the data center 102 includes a computer-based system 195, such as a data processing system. The computer-based system 195 may include a processor 191, a memory 193, and a graphical user interface 197. Each of the controllers described above may similarly include one or more processors, one or more memories, a graphical user interface, and / or an input interface and an output interface.
[0085] Figure 6 A functional block diagram of data processing and control in a manufacturing plant equipment according to an embodiment of the present disclosure is depicted. The controller 190 of the display manufacturing plant equipment includes one or more processors 192, one or more memories 194 and may also include a graphical user interface 196. The computer-based system 195 includes an input interface 692 and an output interface 696. In addition, a processor and memory block 694 is provided. Data processing for display manufacturing plant equipment can be used, for example, to manufacture displays on large area substrates. The data processing system can be further used for multiple metrology operations. Metrology as used herein includes display testing, electron beam inspection of displays, and optical inspection (such as automated optical inspection). In addition, metrology tools for determining physical, electrical, and optical film properties; metrology tools for determining TFT properties; and / or metrology tools for OLED performance and OLED lifetime measurement can be provided.
[0086] The computer-based system 195 can use the set input values, actual input values, historical values and result input values to automatically determine manufacturing-related values, metrology-related values and / or manufacturing-related information. The computer-based system 195 is hereinafter referred to as the FPI (manufacturing process information) system. The FPI system can be operably connected to one or more controllers and / or tools, respectively. Figure 6 The manufacturing tool controllers 119 / 129 and the metrology tool controllers 139 / 149 / 159 are shown as examples. The FPI system is further operatively connected to a data repository 622 where information of manufacturing plant equipment may be stored.
[0087] The FPI system includes a processor and memory block (PMB) 694, which is operably connected to an input interface 692 (such as a hardware-based input interface) and an output interface 696 (such as a hardware-based output interface). PMB 694 is configured to provide the processing required to operate the FPI system and includes a processor (not shown separately) and a memory (not shown separately). The processor of PMB 694 can be configured to execute several functional modules according to computer-readable instructions implemented on a non-transitory computer-readable memory included in the PMB. Such functional modules are hereinafter referred to as being included in the PMB. The functional modules included in the processor include an operably connected training set generator and a module for computing a model. For example, the computing model may include a neural network, such as Figure 4B The deep neural network (DNN) shown. The module for computing the model is configured to initiate data processing using, for example, one or more (deep) neural networks to output specific application data (e.g., recommendations for setting manufacturing conditions) based on the input data.
[0088] The FPI system is configured to receive data (and / or their derivatives) generated by manufacturing tools, metrology tools, controller 190 and / or data stored in one or more data repositories via input interface 692. The FPI system is further configured to process the received data and send the results (or a portion of the results) to a storage system 612, a tool equipped in a display manufacturing plant, a computer-based graphical user interface (GUI) 614 for presenting the results, and / or an external system (e.g., a FAB's yield management system (YMS)) via output interface 696. GUT 614 can be further configured to enable user-specified inputs related to operating the FPEI system.
[0089] As a non-limiting example, a display or a portion of a display may be manufactured by one or more display manufacturing tools. Setting input values or actual input values may be provided by a display manufacturing tool. In addition, input values related to maintenance, display product type, display substrate size, display pixel resolution, consumable parts materials and / or consumable compositions may be provided. This may also optionally include historical values. A display or a portion of a display may be subjected to one or more metering operations as disclosed herein. The result data (hereinafter referred to as result input values) of the information providing one or more quantitative mass values may be transmitted to the FPI system directly or via one or more intermediate systems.
[0090] When processing the received data (optionally with other data, such as design data), the FPI system can generate a computational model and / or provide recommendations based on the computational model. The results can be sent to any of the one or more tools; can be stored in the storage system 612; can be presented via the GUI 614 and / or sent to an external system. Those skilled in the art will appreciate that the teachings of the subject matter of the present disclosure are not limited to Figure 6 Constraints of the depicted system. Equivalent and / or modified functionality may be combined or divided in another manner and implemented in any suitable combination of software and firmware and hardware.
[0091] It should be noted that this can be implemented in a distributed computing environment Figure 6 The system depicted, wherein Figure 6 The aforementioned functional modules shown may be distributed on several local and / or remote devices and may be connected via a communication network. Figure 6The functionality described may be variously distributed within various controllers of the manufacturing plant equipment. At least a portion of the tools, data repositories, storage systems, and / or GUIs may operate external to the manufacturing plant equipment and in data communication with the FPI system, for example, via input interface 692 and output interface 696. The FPI system may be implemented as one or more stand-alone computers used in conjunction with one or more of the tools according to embodiments of the present disclosure. The corresponding functions of the FPI system may be at least partially integrated with one or more tool controllers.
[0092] It will also be understood that the systems according to the embodiments described herein may be implemented at least in part on a suitably programmed computer. Likewise, embodiments of the present disclosure contemplate a computer readable computer program for performing the methods of embodiments of the present invention. According to some embodiments of the present invention, a non-transitory computer readable memory may be further provided for performing the methods described herein, the memory tangibly containing a program of instructions executable by a computer. According to other aspects of the subject matter of the present disclosure, a non-transitory computer readable medium comprising instructions is provided, which, when executed by a computer, causes the computer to perform a method of generating a computational model and / or a method of setting conditions in display manufacturing plant equipment according to an embodiment of the present disclosure.
[0093] According to some embodiments that may be combined with other embodiments described herein, display manufacturing plant equipment may include a modeling data processing unit having an additional memory including additional instructions and having an additional processor, wherein the additional instructions, when executed by the additional processor, cause the modeling data processing unit to generate a computational model according to the method described in any one of claims 1 to 11.
[0094] In view of the above, embodiments of the present disclosure may provide one or more of the following advantages: The manufacturing of displays in display manufacturing plant equipment may be improved by generating computational models through machine learning models or artificial intelligence and / or providing recommendations based on computational models for improving display manufacturing.
[0095] The present disclosure provides multiple embodiments, some of which are listed below.
[0096] Embodiment 1: A method for generating a computational model for improving manufacturing conditions in one or more display manufacturing tools in display manufacturing plant equipment, particularly for manufacturing display devices on multiple substrates, the method comprising: determining multiple input values for one or more display manufacturing tools, each input value being associated with a corresponding substrate identification number among multiple substrate identification numbers; determining multiple result input values for substrates having multiple substrate identification numbers, the result input values providing quantitative quality values; providing a training data set of multiple input values and multiple result input values, the multiple input values and the multiple result input values being related by corresponding substrate identification numbers; and generating or producing a computational model from the training data set by machine learning.
[0097] Embodiment 2: The method of Embodiment 1, wherein each of the plurality of input values is further associated with a corresponding identifier timestamp of a plurality of identifier timestamps.
[0098] Embodiment 3: The method as described in Embodiment 2, wherein the multiple input values and the multiple result input values are further related by corresponding identifier timestamps.
[0099] Embodiment 4: A method as described in any of Embodiments 1 to 3, wherein the quantitative quality value includes: one or more results selected from display pixel testing, display line testing, and display driver circuit testing; results from an electron beam inspection tool used for display manufacturing; results from an automated optical inspection tool; results of film properties; results of thin film transistor properties; results from intermediate or final displays; yield data analysis; probability of non-failed products; and the probability or number of yield bin types, defect classifications, line opens, line shorts, gate shorts, gate opens, dark pixels, white pixels, particles, and any of the above items at one or more substrate locations.
[0100] Embodiment 5: The method of Embodiment 4, wherein one or more of display pixel testing, display line testing, and display driver circuit testing is provided with an e-beam testing tool for display manufacturing.
[0101] Embodiment 6: A method as described in any of Embodiments 1 to 5, wherein the plurality of input values include: a plurality of setting input values of parameter settings from one or more display manufacturing tools; and a plurality of actual input values of measured inputs from one or more display manufacturing tools.
[0102] Example 7: A method as described in Example 6, wherein the setting input value includes at least one setting input value selected from the group consisting of: temperature setting, gas flow setting, deposition source power source setting, warning threshold setting, deposition material, vacuum pressure setting, display product type, display substrate size, display pixel resolution, consumable part material and consumable composition.
[0103] Embodiment 8: A method as described in any one of Embodiments 6 to 7, wherein the actual input value includes at least one actual input value selected from the group consisting of: temperature, gas flow rate, deposition source current, deposition source voltage, deposition source power, deposition source arc detection tracking, vacuum pressure, residual gas analysis and error warning.
[0104] Embodiment 9: The method as described in any one of embodiments 6 to 8, wherein the plurality of input values further comprises a plurality of historical values selected from at least one of the following items: a plurality of historical input values of one or more display manufacturing tools, and a plurality of historical result values of one or more metrology tools.
[0105] Embodiment 10: The method of Embodiment 9, wherein the plurality of historical values are associated with historical timestamps of time points prior to the timestamps corresponding to the corresponding substrate identification numbers.
[0106] Embodiment 11: A method as described in Embodiment 10, wherein the multiple historical values are at least one historical value selected from the following items: historical values of setting input values, historical values of actual input values, historical values of result input values, production equipment maintenance values, tool part replacement values and consumable replacement values.
[0107] Embodiment 12: A method as described in any one of Embodiments 1 to 11, wherein improving manufacturing conditions in one or more display manufacturing tools includes one or more of the following items: improving parameter settings of one or more manufacturing tools, improving production equipment maintenance of one or more manufacturing tools, improving product type selection, improving manufacturing scheduling, and improving hardware configuration settings.
[0108] Embodiment 13: A method for setting conditions in display manufacturing factory equipment for manufacturing display devices on multiple substrates, in particular parameters of one or more display manufacturing tools, the method comprising: calculating, based on a computational model, a plurality of recommended input values mapped by the computational model to one or more quantitative quality values; comparing the plurality of recommended input values with a plurality of input values; and providing a recommendation to change one or more of the input values in the plurality of input values to one or more of the recommended input values to generate a new set of input values.
[0109] Embodiment 14: The method of embodiment 13, wherein the computational model is generated according to the method of any one of embodiments 1 to 12.
[0110] Embodiment 15: A method as described in any of Embodiments 13 to 14, wherein setting the condition includes at least one of the following items: determining whether the multidimensional function of the computational model has a minimum value or a maximum value; and backpropagating the computational model with weights that are at least partially frozen.
[0111] Embodiment 16: A method as described in any one of Embodiments 13 to 15, further comprising: manufacturing a display having a new set of setting input values; generating at least one result from a display pixel test, a display line test, and a display driver circuit test to obtain a new result input value; and adapting the computational model using adaptive training data including the new set of setting input values, new result input values, and a new action input value set.
[0112] Embodiment 17: A method as described in any of Embodiments 13 to 16, wherein the recommendation provided further includes at least one of the following items: the type of display product to be manufactured, the type of display product not to be manufactured, production equipment maintenance, replacement of tool parts, replacement of consumables, selection of tool parts, and generation of quantitative quality values.
[0113] Embodiment 18: A display manufacturing factory equipment, comprising: at least one first display manufacturing tool, the at least one first display manufacturing tool comprising: one or more deposition sources, the one or more deposition sources being used to deposit a material layer on a substrate in the first display manufacturing tool, and a plurality of sensors being used to determine actual input values; a controller being used to control setting input values; and a data processing unit having a memory including instructions and having a processor, wherein the instructions, when executed by the processor, cause the data processing unit to provide a recommendation for changing one or more of a plurality of setting input values.
[0114] Embodiment 19: The display manufacturing factory equipment as described in Embodiment 18, wherein the instructions, when executed by the processor, cause the data processing unit to provide a recommendation for changing one or more of the plurality of setting input values by executing the method according to any one of Embodiments 13 to 17.
[0115] Embodiment 20: The display manufacturing factory equipment as described in Embodiment 19, further comprising: a display testing tool, in particular an electron beam testing tool, which is configured to provide result input values for a substrate having a plurality of substrate identification numbers, and the result input values provide a quantitative quality value including at least one result from a display pixel test, a display line test, and a display driver circuit test.
[0116] Embodiment 21: The display manufacturing plant equipment as described in any of Embodiments 18 to 20, the display manufacturing plant equipment further comprising: an electron beam inspection tool, the electron beam inspection tool being configured to generate an SEM image of a portion of one or more layers on a display substrate generated by the display manufacturing plant equipment; an automated optical inspection tool, the automated optical inspection tool being configured to generate an optical image of a portion of one or more layers on a display substrate generated by the display manufacturing plant equipment; a metrology tool for determining at least one of physical, electrical and optical film properties; a metrology tool for determining TFT properties; and a metrology tool for at least one of display performance and display life.
[0117] Embodiment 22: A display manufacturing plant equipment as described in any one of Embodiments 18 to 21, the display manufacturing plant equipment further comprising a modeling data processing unit having an additional memory including additional instructions and having an additional processor, wherein the additional instructions, when executed by the additional processor, cause the modeling data processing unit to generate a computational model according to a method as described in any one of Embodiments 1 to 11.
[0118] While the foregoing is directed to implementations of the present disclosure, other and further embodiments of the disclosure may be envisaged without departing from the basic scope of the disclosure and the scope of the disclosure is determined by the claims that follow.
Claims
1. A method of generating a computational model for improving manufacturing conditions in one or more display manufacturing tools in a display manufacturing plant for manufacturing display devices on a plurality of substrates, the method comprising: determining a plurality of input values for the one or more display manufacturing tools, each input value being associated with a corresponding substrate identification number of a plurality of substrate identification numbers; determining a plurality of result input values for substrates having the plurality of substrate identification numbers, the result input values providing quantitative quality values; providing a training data set of the plurality of input values and the plurality of resulting input values, the plurality of input values and the plurality of resulting input values being related by the corresponding substrate identification numbers; as well as The computational model is produced or generated from the training data set by machine learning. 2 . The method of claim 1 , wherein each of the plurality of input values is further associated with a corresponding identifier timestamp of a plurality of identifier timestamps. 3 . The method of claim 2 , wherein the plurality of input values and the plurality of result input values are further related by the corresponding identifier timestamps.
4. The method of any one of claims 1 to 3, wherein the quantitative quality value comprises: one or more results selected from a display pixel test, a display line test, and a display driver circuit test; Results from an electron beam review tool used in display manufacturing; Results from automated optical inspection tools; results for film properties; results for thin film transistor properties; results from intermediate or final displays; yield data analysis; The probability of non-failed products; and the probability or quantity of yield bin type, defect classification, line open, line short, gate short, gate open, dark pixel, white pixel, particle, and any of the above at one or more substrate locations.
5. The method of claim 4, wherein one or more of the display pixel test, the display line test, and the display driver circuit test is provided with an e-beam test tool used for display manufacturing.
6. The method of any one of claims 1 to 3, wherein the plurality of input values comprises: a plurality of setting input values from parameter settings of the one or more display manufacturing tools; and A plurality of actual input values of measured inputs from the one or more display manufacturing tools.
7. The method of claim 6, wherein the setting input values include at least one setting input value selected from the group consisting of: temperature settings, gas flow settings, deposition source power settings, warning threshold settings, deposition materials, vacuum pressure settings, display product types, display substrate sizes, display pixel resolutions, consumable part materials, and consumable compositions.
8. The method of claim 6, wherein the actual input value comprises at least one actual input value selected from the group consisting of: temperature, gas flow, deposition source current, deposition source voltage, deposition source power, deposition source arc detection tracking, vacuum pressure, residual gas analysis, and error warning.
9. The method of claim 6, wherein the plurality of input values further comprises a plurality of historical values selected from at least one of the following: a plurality of historical input values of the one or more display manufacturing tools, and Multiple historical result values for one or more measuring instruments.
10. The method of claim 9, wherein the plurality of historical values are associated with historical timestamps of time points prior to timestamps corresponding to corresponding substrate identification numbers.
11. The method of claim 10, wherein the plurality of historical values are at least one historical value selected from the following: historical values of setting input values, historical values of actual input values, historical values of result input values, production equipment maintenance values, tool component replacement values, and consumable replacement values.
12. A method as described in any one of claims 1 to 3, wherein the improvement of manufacturing conditions in one or more display manufacturing tools includes one or more of the following items: improving parameter settings of one or more manufacturing tools, improving production equipment maintenance of one or more manufacturing tools, improving product type selection, improving manufacturing scheduling, and improving hardware configuration settings.
13. A method of setting conditions in display manufacturing plant equipment, in particular parameters of one or more display manufacturing tools, for manufacturing display devices on multiple substrates, the method comprising: calculating, based on the computational model, a plurality of suggested input values mapped by the computational model to one or more quantitative quality values; comparing the plurality of suggested input values to a plurality of input values; as well as A recommendation is provided to change one or more of the input values in the plurality of input values to one or more of the suggested input values to generate a new set of input values.
14. The method of claim 13, wherein the computational model is generated according to the method of claim 1.
15. The method according to any one of claims 13 to 14, wherein the setting condition comprises at least one of the following: determining whether the multidimensional function of the calculation model has a minimum value or a maximum value; and Backpropagating the computational model with the at least partially frozen weights.
16. The method of any one of claims 13 to 14, further comprising: manufacturing a display using the new set of setting input values; generating at least one result from a display pixel test, a display line test, and a display driver circuit test to obtain a new result input value; as well as The computational model is adapted using adaptive training data including the new setting input value set, the new result input value set, and the new action input value set.
17. A method as described in any one of claims 13 to 14, wherein the provision of the recommendation further includes at least one of the following items: type of display product to be manufactured, type of display product not to be manufactured, production equipment maintenance, replacement of tool parts, replacement of consumables, selection of tool parts, generation of quantitative quality values.
18. A display manufacturing plant equipment, the display manufacturing plant equipment comprising: At least one first display manufacturing tool, the at least one first display manufacturing tool comprising: one or more deposition sources for depositing a layer of material on a substrate in the first display manufacturing tool; and a plurality of sensors for determining actual input values; a controller for controlling a setting input value; and A data processing unit having a memory including instructions and having a processor, wherein the instructions, when executed by the processor, cause the data processing unit to provide a recommendation to change one or more of the plurality of setting input values.
19. The display manufacturing plant equipment of claim 18, wherein the instructions, when executed by the processor, cause the data processing unit to provide the recommendation to change one or more of the plurality of setting input values by executing the method according to any one of claims 13 to 14.
20. The display manufacturing plant equipment according to claim 19, further comprising: A display testing tool, particularly an electron beam testing tool, is configured to provide result input values for substrates having multiple substrate identification numbers, the result input values providing a quantitative quality value including at least one result from a display pixel test, a display line test, and a display driver circuit test.
21. The display manufacturing plant equipment according to any one of claims 18 to 20, further comprising: an electron beam inspection tool configured to generate an SEM image of a portion of one or more layers on a display substrate generated using the display manufacturing plant equipment; an automated optical inspection tool configured to generate a light optical image of a portion of one or more layers on a display substrate generated by the display manufacturing plant equipment; metrology tools for determining at least one of physical, electrical and optical film properties; Metrology tools for determining TFT properties; and A metrology tool for at least one of display performance and display life.
22. The display manufacturing plant equipment according to any one of claims 18 to 20, further comprising: A modeling data processing unit having a further memory including further instructions and having a further processor, wherein the further instructions, when executed by the further processor, cause the modeling data processing unit to The calculation model is generated according to the method as claimed in any one of claims 1 to 3.
