Improved batch-by-batch controller in process control

Abstract

A method (40) for setting process control for manufacturing a device in production comprises the steps of: obtaining (S1) measurements (MET, ART) characterizing the device; specifying (S2) a process step (14) that should be used to process the component; a parameterization of the process step (14) is determined (S3) for the process step (14) by means of an R2R controller on the basis of the measurement results (MET, ART). The method (40) is characterized in that the measurement result is either a virtual measurement result (ART), or a virtual measurement result (ART) and a physical measurement result (MET).

Description

Technical Field

[0001] The present invention relates to a method for improving a run-to-run (R2R) controller by taking into account human measurement data, and to an apparatus, computer program, and machine-readable storage medium configured to perform the method. Background Technology

[0002] As is known in the prior art, systems used for process control (e.g., R2R controllers, statistical process control (SPC systems)) acquire measurement data of the processed wafer. These measurements are typically physical measurements on the wafer. Based on these physical measurements, product parameters related to process control can be determined. These measurements are, for example, so-called in-line measurements, which are measured on the product using physical measurement equipment. Product parameters include, for example, the thickness, depth, width, and resistivity of the structure. In contrast, there are process or formulation parameters that describe the manufacturing steps on the wafer (exposure, etching, polishing, deposition, annealing, processing time, pressure, gas flow rate, dosage).

[0003] Based on physical measurements, a batch-to-batch (R2R) controller is used to determine formulation recommendations for the next processing step of the same wafer, or for the same processing step of the next wafer, so that appropriate readjustments can be made when measurements are not within the required range.

[0004] The control of the processing steps is performed here in a piece-to-piece control manner; that is, the recipe for each subsequent processing step for a wafer or batch (a batch of wafers) is influenced by the R2R controller. The recipe characterizes the corresponding manufacturing or processing step, such as exposure, etching, deposition, polishing. However, the R2R controller does not intervene during the processing of the wafer / batch. This control method is also known as R2R process control. Batch-to-batch controllers used in semiconductor manufacturing are known, for example, see Moyne, J. (2014). "Run-to-Run Control in Semiconductor Manufacturing (In: Baillieul, J., Samad, T. (eds) Encyclopedia of Systems and Control. Springer, London. https: / / doi.org / 10.1007 / 978-1-4471-5102-9_255-1).

[0005] The R2R controller is primarily used to compensate for disturbances such as long-term drift during manufacturing by making corresponding recipe adjustments in subsequent processing, thereby maintaining the relevant online parameters at the desired target values.

[0006] However, a drawback is that product and process parameters generated during production cannot be immediately used for process control because the data paths are undefined, nonexistent, or difficult to implement, such as incoming material quality parameters, runtime data, and process data (FDC). In other words, product parameters from arbitrary data sources and the quantities derived therefrom cannot be used for process control tasks.

[0007] The process of using feature models for recipe creation and recipe matching is disclosed in US 2023 / 0091058 A1. Summary of the Invention

[0008] Advantages of the present invention The objective of this invention is to provide a virtual measurement device (“artificial, virtual sensor”) that can be attached to or replace the physical measurement process (MET) to provide additional data to the R2R controller, thereby enabling the R2R controller to perform control more reliably and accurately.

[0009] The difference between this and virtual metering (VM) is that in VM, manual measurements are generated based on process parameters. This virtual sensor, however, should feed back any manual measurements—not just process parameters—based on data and software to the process control system, such as an R2R controller.

[0010] Within the factory control system (Manufacturing Execution System, MES), a high degree of transparency is ensured by exposing the data path from the data source to the process control system.

[0011] Another advantage is the flexibility to combine measurement recipes, which can be created arbitrarily, from simple heuristics to complex data science applications such as aggregation, classification, or estimation methods.

[0012] Therefore, in summary, the present invention allows for supplementing or replacing physical measurement processes in R2R controllers using a software-based approach.

[0013] Disclosure of the present invention In a first aspect, the present invention relates to a method for setting process controls used in manufacturing devices to control production machinery during production or manufacturing. The device is preferably a semiconductor device, such as a sensor or chip. It should be noted that the process controls determined according to the method of the first aspect can be used to manufacture one or more devices. For example, semiconductor devices on a wafer or in a batch (Lot / Los) can undergo substantially the same processing according to the determined process controls.

[0014] The method begins by acquiring or receiving measurement results (MET, ART) characterizing the device. The measurement results are either virtual measurement results (ART) or a combination of virtual measurement results (ART) and physical measurement results (MET). Physical measurement results can be understood as those acquired on the device using actual measurement equipment. Virtual measurement results can be understood as those determined based on representative (measured) quantities.

[0015] Subsequently, a pre-defined process step, in particular, can be obtained or received, by which the device should be processed. This process step can be defined or otherwise pre-defined based on the device's manufacturing recipe. In addition to this process step, an optimal process window can be provided, allowing the R2R controller to more reliably parameterize the process step by considering this window. Therefore, this process step describes how the measured device should be further processed in the next step, or with what settings (given by the parameterization determined by the R2R controller) should be performed on which processing step. It is also conceivable that this process step be used for at least one subsequent device in a series of devices to be processed.

[0016] Next, the R2R controller determines the parameterization of the process step based on the measurement results (MET, ART). The R2R controller can be configured to perform wafer-to-wafer, batch-to-batch, or other types of control.

[0017] The proposed method involves acquiring virtual measurement results (ART) via a virtual measurement device that executes the measurement process through software using at least one data query. Therefore, the virtual measurement device performs a software-based measurement process in the form of a measurement recipe or a computer-implemented measurement algorithm. This virtual measurement device can also be referred to as a virtual sensor. The virtual measurement device can be configured to execute one or more different measurement recipes, thereby outputting one or more virtual measurement results.

[0018] Furthermore, it is proposed that the virtual measurement device can output the same device status as the physical measurement device, especially the E10 status. Advantageously, the virtual measurement device can now be easily integrated into existing processes, for example, as a plug-in. Rules applicable to the physical measurement device can therefore remain effective and applicable without fundamental modification. It should be noted that from the R2R controller's perspective, the virtual measurement device is perceived just like the physical measurement device. Especially due to the communication between the R2R controller and the virtual measurement device, the virtual measurement device appears to the R2R controller as a real measurement device. However, the virtual measurement device can be purely a computer program, executed in a factory or server.

[0019] Furthermore, it is proposed that after the process steps are executed using the parameters determined by the R2R controller, at least one downstream measurement process is performed on the processed device. The measurement results of this downstream measurement process are provided to the R2R controller as further measurement results. These downstream measurement results are either Virtual Measurement Results (ART) or a combination of Virtual Measurement Results (ART) and Physical Measurement Results (MET). Advantageously, in this case, providing the R2R controller with more detailed feedback regarding its parameter presets becomes possible.

[0020] In other respects, the present invention relates to an apparatus and a computer program configured to perform the above-described methods, and to a machine-readable storage medium having the computer program stored thereon. Attached Figure Description

[0021] The embodiments of the present invention will now be described in more detail with reference to the accompanying drawings. In the drawings: Figure 1 A known R2R controller is schematically illustrated; Figure 2 An embodiment of the R2R controller of the present invention with virtual sensors is illustrated schematically; Figure 3 The communication between the MES and the virtual sensors and R2R controller is illustrated schematically. Figure 4 A flowchart illustrating an embodiment of the method of the present invention is shown schematically. Detailed Implementation

[0022] Figure 1 A schematic diagram 10 of the information flow of an R2R controller known in the prior art is shown.

[0023] Information flow diagram 10 illustrates the process flow 11 for manufacturing semiconductor devices. In a first step, one or more physical measurements 12 are performed on a wafer. Measurement 12 can be the wafer's thickness, width, depth, etc. The measurement results (MET) are then transmitted and provided to the R2R controller 13.

[0024] Based on the measurement result MET, the R2R controller 13 sets the process step 14 for the wafer. In other words, the R2R controller 13 outputs the setting of the process step 14 based on the measurement result MET, so that after the process step 14 is completed, the processed wafer has product parameters that are as close as possible to the preset numerical range.

[0025] Optionally, after completing process step 14 according to the settings output by R2R controller 13, one or more downstream physical measurements 15 can be performed on the processed wafer. The measurement results obtained during this period (Post-MET) can be provided as feedback to R2R controller 13, enabling the controller to make better settings for the next process step 14 based on the subsequently obtained measurement results.

[0026] Figure 2 A schematic diagram 20 illustrating the information flow of an R2R controller according to an embodiment of the present invention is shown. Figure 2 The implementation form depicted here is related to Figure 1 The main difference in the implementation is that the R2R controller obtains the measurement from the virtual sensor 16 as an additional input.

[0027] The virtual sensor 16 is treated as a virtual measuring device, which generates artificial measurement results (ART) and provides the measurement results to the R2R controller 13. Figure 2 The implementation form drawn by Figure 1 The main difference in the implementation is that the R2R controller 13 uses these additional measurement results ARTs from the virtual sensor 16 to additionally set the process steps 14 based on these artificial measurement results ARTs.

[0028] Therefore, this virtual measurement device can be used to provide software-based measurements to factory control systems (MES), especially R2R controllers, in addition to or as a replacement for physical measurements (MET).

[0029] The advantage of virtual measurement devices (ARTs) is that, from the perspective of the factory control system (MES), it is not necessarily required to identify that no physical measurement has been performed. Therefore, the virtual ART behaves like a physical ART relative to the R2R controller. This can be achieved by having the virtual ART use the same readiness state as the physical ART (e.g., according to the E10 model). In this way, the virtual ART appears to be a real measurement device. Preferably, the virtual ART behaves like a physical ART by communicating like a real measurement device. Thus, virtual ARTs can be easily integrated into existing factory control systems without fundamental modifications while retaining applicable rules.

[0030] Figure 3 The diagram illustrates the communication between the MES and the virtual measurement device / virtual sensor, as well as the R2R controller.

[0031] Figure 4 A flowchart 40 schematically illustrates an implementation of a method for setting process control in semiconductor device manufacturing.

[0032] The method begins with step S1. In this step, measurement results (MET, ART) characterizing the device are provided. The acquisition of the measurement results is based on... Figure 2 The measurement results are either virtual measurement results (ART) or a combination of virtual measurement results (ART) and physical measurement results (MET).

[0033] The following is the specification for process step 14, which stipulates that S2 should be used to process the device. Information about process step 14 can be obtained through a database query.

[0034] Next is step S3, where the parameterization of the process step is determined based on the measurement results (MET, ART) using the R2R controller used to process step 14.

[0035] Optionally, after step S3, S4 can be output as the parameterization determined in process step 14, used to process the device. Optionally, the determined parameterization can be used to control the device in step S5.

[0036] The above is based on Figure 2 The concept and / or based on Figure 4 The method can be used, for example, as shown below. Different applications are available in... Figure 2 The diagram is illustrated using measurement recipes R1-R4. Each measurement recipe defines how the virtual sensor performs measurements (mathematical calculation rules) and where the data originates (data source).

[0037] In the first application example, supplier data is output by a virtual measurement device. This supplier data can be stored in the database of the factory control system (MES). For example, the virtual measurement device retrieves the wafer thickness from the database as the measurement result ART. The wafer thickness is provided to the R2R controller as a measurement value, which then presets the polishing time for the processing steps.

[0038] The goal of this R2R controller is to produce a preset wafer thickness and / or a wafer surface that is as uniform as possible.

[0039] By directly correlating the wafer thickness determined by the virtual measurement device with the desired wafer thickness, the R2R controller can pre-set a proportional increase in the polishing time used for processing.

[0040] The advantage in this situation is that it ensures consistent wafer thickness, allowing each wafer to be used under the highest quality requirements.

[0041] In the second application example, FDC process data is output by a virtual measurement device.

[0042] Virtual sensors can detect anomalies in process parameters or measurement results, based on data set on the production machine or data executed by the formula.

[0043] For example, anomalies can be detected in process data and parameters such as temperature, pressure, or gas flow rate to identify abnormal fluctuations. These anomalies are then transmitted as measurements to the R2R controller. The R2R controller can then make corresponding recipe modifications based on this information, for example, by taking appropriate countermeasures when excessive fluctuations occur.

[0044] Alternatively, it can be envisioned that the degree to which a corresponding production machine can be trusted is determined based on the intensity of the fluctuations. In cases of high fluctuations, it can be inferred that trust in that machine should be reduced. If strong fluctuations occur, the R2R controller accordingly determines a conservative formula adjustment for the next process. In other words, the unreliability of the manufacturing machine can be identified and correspondingly considered in process control.

[0045] In the third application example, geometric data is output by a virtual measurement device. Semiconductor coordinate databases primarily contain the geometric design of a chip or the entire wafer, especially geometry such as open area information. In this case, the background is that the etching process operates with varying intensities depending on the size of the etched surface. Therefore, the etching duration is set based on the size of the open area (large / small). With an R2R controller, the geometric data, especially any intermediate stages of the design, can be continuously used to control the etching process. This method can also be applied correspondingly to other manufacturing steps, such as implantation.

[0046] In the fourth application example, metadata is output by a virtual measuring device. Metadata can be data potentially contained within a database. For this data type, or for datasets derived from metadata, no separate measuring device exists. For example, the database might contain the duration between the end of a manufacturing step and the measurement process, via timestamps of individual data points, and this information is provided to the process control system using a virtual measuring device.

[0047] In the fifth application example, physical measurements are (completely) replaced by virtual measurements. For example, specific product parameters can be predicted based on measurements of process parameters (data sources) using neural networks (or mathematical calculation rules), i.e., the results of virtual measurement are provided to the process control system (e.g., an R2R controller) using virtual measurement devices.

[0048] In the sixth application example, the machine status is output by a virtual measurement device (MET). The virtual MET can predict, for example, the wear of machine parts based on the physical measurement results (MET). The R2R controller can then adjust the formulation to achieve the same process results based on the current wear, or trigger repair / maintenance. Specifically, for example, for epitaxial equipment, lamp runtime can be determined using a virtual MET; in the case of polishing equipment, polishing pad life can be determined; or in the case of deposition equipment, target life can be determined, and this information is provided to process control to execute automated R2R controller actions (e.g., different reset scenarios or formulation modifications).

[0049] In the seventh application example, a machine learning system is used for a virtual measurement device. Data-intensive AI methods (decision-making, classification, clustering, estimation, and learning methods) are typically not performed directly in the production environment, but rather in an exported offline environment. To provide the results of such methods to the production environment, the virtual measurement device can act as a bridge between the AI ​​algorithm and the process control system, or present the results as Virtual Measurement Results ART output. Specifically, for example, clustering analysis can be used to optimally summarize distinguishable product-specific attributes (e.g., rate gradation, product subgroups) and then provide contextual grouping suggestions to the R2R controller.

Claims

1. A method (40) for setting process controls to manufacture one or more devices in production, comprising the following steps: Obtain (S1) measurement results (MET, ART) characterizing the one or more devices; Based on the measurement results (MET, ART), the parameterization of the preset process steps (14) for the one or more devices or subsequent devices is determined (S3) by means of the R2R controller; The measurement results mentioned therein are either virtual measurement results (ART) or virtual measurement results (ART) and physical measurement results (MET). The virtual measurement result (ART) is acquired by a virtual measurement device (16), which determines the measurement result through software using at least one data query. The virtual measuring device (16) is characterized in that it is configured to output device status, especially E10 status, like a physical measuring device.

2. The method according to claim 1, wherein the virtual measuring device (16) performs the measurement process by means of software at least one data query and at least one data processing step using the result of the data query.

3. The method according to any one of the preceding claims, wherein, After the process step (14) is performed using the determined parameters, at least one downstream measurement process is performed on at least one device processed according to the process step (14), wherein the measurement results of the downstream measurement process are provided to the R2R controller as further measurement results, wherein the measurement results of the downstream measurement process are virtual measurement results (ART) and / or physical measurement results (MET).

4. An apparatus configured to perform the method according to any one of the preceding claims.

5. A computer program comprising instructions that, when executed by a computer, cause the computer to perform the method according to claims 1 to 3.

6. A machine-readable storage medium having a computer program as claimed in claim 5 stored thereon.

Images