Analysis device, analysis method, recording medium, and plasma processing control system

Abstract

The present application provides an analysis device, an analysis method, a recording medium, and a plasma processing control system for quantitatively evaluating a state of a processing space using time-series data groups measured in the processing space in which plasma processing is performed. The analysis device has a calculation section that inputs time-series data groups measured in a judgment section that is a prescribed time earlier than a control section to a time-series analysis model to calculate a degree of deviation from a reference state of the processing space, and a determination section that determines a characteristic value for deciding control data for plasma processing of a substrate in the control section based on the degree of deviation obtained by the calculation.

Description

Technical Field

[0001] This disclosure relates to an analytical apparatus, analytical method, analytical procedure, and plasma processing control system. Background Technology

[0002] Generally, in semiconductor manufacturing processes, changes in the state of the plasma processing space affect the quality of the final product when etching the object within that space. Therefore, quantitatively evaluating the state of the processing space during etching is crucial for maintaining the quality of the final product.

[0003] On the other hand, in the processing space of semiconductor manufacturing processes, there are collections of various data acquired before or during etching (hereinafter, datasets of multiple types of time series data are referred to as time series data sets). Additionally, the acquired time series data sets also include time series data related to the state of the processing space.

[0004] Existing technical documents

[0005] Patent documents

[0006] Patent Document 1: Japanese Patent Application Publication No. 2001-060585 Summary of the Invention

[0007] The problem the invention aims to solve

[0008] This disclosure provides an analysis apparatus, analysis method, analysis procedure, and plasma processing control system for quantitatively evaluating the state of a processing space using a set of time-series data measured in a processing space where plasma processing is performed.

[0009] Solution for solving the problem

[0010] One aspect of the analytical apparatus disclosed herein has, for example, the following structure. That is,

[0011] The computing unit inputs a set of time-series data measured in the processing space where plasma processing is performed, specifically within a judgment interval that is a predetermined time preceding the control interval, into a time-series analysis model to calculate the deviation from the reference state of the processing space; and

[0012] The determination unit determines a characteristic value based on the deviation obtained from the calculation, and this characteristic value is used to determine the control data for plasma treatment of the substrate in the control range.

[0013] The effects of the invention

[0014] According to this disclosure, an analysis apparatus, analysis method, analysis procedure, and plasma processing control system can be provided to quantitatively evaluate the state of a processing space using a set of time-series data measured in a processing space where plasma processing is performed. Attached Figure Description

[0015] Figure 1 The first figure shows an example of the system structure of an etching process control system.

[0016] Figure 2 This is the first diagram showing the relationship between the plasma processing flow and the judgment and control intervals in the processing space.

[0017] Figure 3 This is a diagram illustrating an example of the hardware structure of an analysis device.

[0018] Figure 4 This is a graph showing an example of a time series data set.

[0019] Figure 5 The first figure shows a specific example of the processing performed by the learning department.

[0020] Figure 6 The first figure shows a specific example of the processing performed by the determination unit.

[0021] Figure 7 This is a graph showing the correspondence between the count value and the etching rate.

[0022] Figure 8 This is the first flowchart illustrating the process of analysis and control.

[0023] Figure 9 The second figure shows an example of the system structure of the etching process control system.

[0024] Figure 10 This is a diagram illustrating an example of OES data.

[0025] Figure 11 The third figure shows an example of the system structure of the etching process control system.

[0026] Figure 12 This is a diagram illustrating an example of a set of process data.

[0027] Figure 13 This is the second figure, which illustrates the relationship between the plasma processing flow and the judgment and control intervals in the processing space.

[0028] Figure 14 The second figure shows a specific example of the processing performed by the learning department.

[0029] Figure 15The second figure shows a specific example of the processing performed by the determination unit.

[0030] Figure 16 This is a diagram showing specific examples of deviation.

[0031] Figure 17 This is a graph showing the relationship between deviation and etching rate.

[0032] Figure 18 This is the second flowchart representing the process of analysis and control.

[0033] Explanation of reference numerals in the attached figures

[0034] 100, 100', 100”: Etching process control system; 110: Wafer before processing; 120: Chamber; 130: Wafer after processing; 140_1~140_n: Time series data acquisition device; 150: Analysis device; 151: Learning unit; 152: Determination unit; 160: Control device; 510: Time series analysis model generation unit; 610: Time series analysis model execution unit; 620: Count value calculation unit 621: Difference Calculation Unit; 622: Counting Unit; 623: Transformation Unit; 940: Emission Spectroscopy Analysis Device; 1140_1~1140_n: Process Data Acquisition Device; 1400: Learning Unit; 1410_1~1410_401: Model; 1500: Determination Unit; 1510_1~1510_401: Model; 1520_1~1520_401: Deviation Calculation Unit; 1530: Transformation Unit. Detailed Implementation

[0035] Hereinafter, various embodiments will be described with reference to the accompanying drawings. Furthermore, in this specification and the accompanying drawings, structural elements having substantially the same functional structure are labeled with the same reference numerals, and repeated descriptions are omitted.

[0036] [First Implementation Method]

[0037] <System Structure of Etching Process Control System>

[0038] First, the system structure of an etching process control system, which is an example of a plasma processing control system, will be explained. Figure 1 The first figure shows an example of the system structure of an etching process control system. (See figure below.) Figure 1 As shown, the etching process control system 100 includes a semiconductor manufacturing process, time series data acquisition devices 140_1 to 140_n, an analysis device 150, and a control device 160.

[0039] In a semiconductor manufacturing process, an object (pre-processing wafer 110) is etched in a chamber 120, which serves as a processing space for plasma treatment, to generate a result (post-processing wafer 130). Here, pre-processing wafer 110 refers to the wafer (substrate) before it is etched in chamber 120, and post-processing wafer 130 refers to the wafer (substrate) after it is etched in chamber 120.

[0040] Time series data acquisition devices 140_1 to 140_n measure time series data before or during etching of the unprocessed wafer 110 in chamber 120. Assume that time series data acquisition devices 140_1 to 140_n measure different types of measurement items. Furthermore, each of the time series data acquisition devices 140_1 to 140_n can measure either one or multiple measurement items.

[0041] The following time series data sets from the time series data sets measured by the time series data acquisition devices 140_1 to 140_n are stored as learning data in the learning data storage unit 153. These time series data sets are:

[0042] • Under normal conditions, with the state of chamber 120 as the baseline,

[0043] • Use standard processes (pre-determined specific processes),

[0044] The time-series data set measured during the etching process.

[0045] In addition, the time series data set measured in the judgment interval from the time series data set measured by the time series data acquisition devices 140_1 to 140_n is notified to the determination unit 152 as judgment data.

[0046] Furthermore, the judgment interval refers to the interval of time series data used to quantitatively evaluate the state of the chamber 120 during plasma processing in the chamber 120, and this judgment interval corresponds to a specified time preceding the control interval described later.

[0047] An analysis program is installed in the analysis device 150, and the analysis device 150 functions as the learning unit 151 and the determination unit 152 by executing the program.

[0048] The learning unit 151 uses the learning data stored in the learning data storage unit 153 to perform machine learning on the time series analysis model. The result obtained by the learning unit 151 in performing machine learning is notified to the determination unit 152 as the completed learning of the time series analysis model.

[0049] The determination unit 152 calculates the deviation of the state of the chamber 120 in the judgment interval from the reference state by inputting the time series data set measured by the time series data acquisition devices 140_1 to 140_n in the judgment interval into the learned time series analysis model.

[0050] Furthermore, the determination unit 152 determines parameters (e.g., characteristic values ​​such as etching rate indicating process variation) based on the deviation from a reference state. These parameters are used to determine the process control data for etching processing in a control interval later than the judgment interval. In other words, the determination unit 152 can quantitatively evaluate the state of the chamber 120 in the judgment interval. The determination unit 152 then notifies the control device 160 of the determined characteristic values ​​(e.g., etching rate).

[0051] In addition, the control interval refers to the interval in the plasma processing in chamber 120 where the etching process is carried out by controlling the process control data (gas flow rate, temperature, pressure, plasma power supply voltage, processing time, etc.). This control interval is equivalent to the etching process interval that is later than the judgment interval.

[0052] When the control device 160 is notified of a characteristic value (e.g., etching rate) from the determination unit 152 during the determination interval, it determines process control data based on that characteristic value during the control interval and notifies the semiconductor manufacturing process. This enables control of the etching process in the chamber 120 during the semiconductor manufacturing process.

[0053] In this way, in the analysis device 150, the time series data set measured in the judgment interval is input into the time series analysis model to calculate the deviation relative to the reference state, thereby determining the characteristic value of the process control data used to determine the etching process.

[0054] Therefore, the state of the chamber 120 can be quantitatively evaluated using the analysis device 150. Furthermore, by quantitatively evaluating the state of the chamber 120 to determine the process control data during the etching process, the analysis device 150 can control the etching process within a control interval based on a time-series data set of the judgment interval.

[0055] <Judgment Interval and Control Interval>

[0056] Next, the process of plasma treatment in chamber 120 and the relationship between the judgment interval and the control interval will be explained. Figure 2 This is the first diagram showing the relationship between the plasma processing flow and the judgment and control intervals in the processing space.

[0057] like Figure 2As shown, in the plasma treatment performed in chamber 120, multiple aging and etching processes are repeated. Furthermore, in each process, the plasma source is switched between ON and OFF.

[0058] In this embodiment, during the etching process, the time-series data acquisition devices 140_1 to 140_n measure judgment data using a predetermined time interval immediately following the plasma source being turned on (immediately following ignition). The length of the judgment interval can be set to a fixed value, for example, about 1 to 3 seconds, or it can be set to a predetermined ratio relative to the etching process time. Furthermore, the length of the judgment interval can be configured to vary depending on the process or the chamber.

[0059] As described above, by notifying the determination unit 152 of the judgment data measured in the judgment interval and inputting it into the time series analysis model, the deviation from the reference state is calculated, and the characteristic value (e.g., etching rate) is determined.

[0060] Therefore, in the etching process within the control interval immediately following the judgment interval, process control data determined based on the identified characteristic values ​​(e.g., etching rate) is used for control.

[0061] In other words, according to this embodiment, the state of the chamber 120 immediately after the plasma source is turned on can be quantitatively evaluated during the etching process, and the etching process in the subsequent control interval can be controlled.

[0062] <Hardware Structure of the Analysis Device>

[0063] Next, the hardware structure of the analysis device 150 will be explained. Figure 3 This is a diagram illustrating an example of the hardware structure of an analysis device. (As shown...) Figure 3 As shown, the analysis device 150 includes a CPU (Central Processing Unit) 301, a ROM (Read-Only Memory) 302, and a RAM (Random Access Memory) 303. Additionally, the analysis device 150 includes a GPU (Graphics Processing Unit) 304. Furthermore, the processors (processing circuits) such as the CPU 301 and GPU 304, along with the memories such as the ROM 302 and RAM 303, form what is known as a computer.

[0064] Furthermore, the analysis device 150 includes an auxiliary storage device 305, a display device 306, an operation device 307, an I / F (Interface) device 308, and a drive device 309. In addition, the various hardware components of the analysis device 150 are interconnected via a bus 310.

[0065] CPU 301 is a computing device that executes various programs (such as analysis programs) installed in auxiliary storage device 305.

[0066] ROM 302 is a non-volatile memory that functions as the main storage device. ROM 302 stores various programs and data necessary for the CPU 301 to execute various programs installed in the auxiliary storage device 305. Specifically, ROM 302 stores boot programs such as BIOS (Basic Input / Output System) and EFI (Extensible Firmware Interface).

[0067] RAM 303 is a volatile memory such as DRAM (Dynamic Random Access Memory) or SRAM (Static Random Access Memory), which functions as the main storage device. RAM 303 provides a working area for various programs installed in the auxiliary storage device 305 to be expanded when executed by the CPU 301.

[0068] GPU 304 is a computing device for image processing. In this embodiment, when the analysis program is executed by CPU 301, it performs high-speed operations on time series data sets based on parallel processing. In addition, GPU 304 is equipped with internal memory (GPU memory) to temporarily store information required for parallel processing of various time series data sets.

[0069] The auxiliary storage device 305 stores various programs and data used when the CPU 301 executes these programs. For example, a learning data storage unit 153 is implemented in the auxiliary storage device 305.

[0070] Display device 306 is a display device that displays the internal status of analysis device 150. Operating device 307 is an input device used by the administrator of analysis device 150 when inputting various instructions to analysis device 150. I / F device 308 is a connection device for communication with a network (not shown).

[0071] The drive unit 309 is a device for mounting the recording medium 320. The recording medium 320 referred to herein includes media that record information optically, electrically, or magnetically, such as CD-ROMs, floppy disks, and optical discs. Alternatively, the recording medium 320 may also include semiconductor memories that record information electrically, such as ROMs and flash memory.

[0072] Furthermore, various programs installed in the auxiliary storage device 305 can be installed, for example, by installing the allocated recording medium 320 into the drive device 309 and having the drive device 309 read the various programs recorded on the recording medium. Alternatively, the various programs installed in the auxiliary storage device 305 can be downloaded via a network (not shown) and thus installed.

[0073] <Specific examples of time series data sets>

[0074] Next, specific examples of time series data sets measured by time series data acquisition devices 140_1 to 140_n will be described. Figure 4 This is a graph illustrating an example of a time series data set. Furthermore, in Figure 4 In the example, for the sake of simplicity, it is assumed that the time series data acquisition devices 140_1 to 140_n measure one-dimensional data respectively. However, a time series data acquisition device can also measure two-dimensional data (a dataset of one-dimensional data of multiple types).

[0075] Figure 4 (a) shows a time series data set consisting of time series data measured by time series data acquisition devices 140_1 to 140_n within the same time range.

[0076] on the other hand, Figure 4 (b) shows a time series data set consisting of time series data measured by time series data acquisition devices 140_1 to 140_n within a corresponding time range. Figure 4 As shown in (b), the learning data used in machine learning can include not only time series data sets consisting of time series data measured within the same time range, but also time series data sets consisting of time series data measured within corresponding time ranges.

[0077] <Specific examples of the handling conducted by the Learning Department>

[0078] Next, a specific example of the processing performed by the learning unit 151 of the analysis device 150 will be described. Figure 5 The first diagram illustrates a specific example of the processing performed by the learning department. (See diagram for example.) Figure 5 As shown, the learning unit 151 has a time series analysis model generation unit 510.

[0079] The time series analysis model described here is a machine learning model that comprehensively and rapidly extracts relationships between data from multiple time series datasets. This model expresses the relationships between multiple time series data using linear or non-linear regression equations. As an example of a time series analysis model, the cross-correlation model can be cited. In the case of the cross-correlation model, a time delay term that considers the time differences between multiple time series data may also be included.

[0080] In the time series analysis model generation unit 510, the formula represented by symbol 520 is used to define the relationship between the time series data of the measurement items measured by each time series data acquisition device 140_1 to 140_n when etching is performed using a standard process under the reference state.

[0081] Specifically, the time series analysis model generation unit 510 calculates the coefficients (values ​​representing relationships) of the numerical expression represented by the symbol 520 by inputting the time series data of the first node into the numerical expression represented by the symbol 520 in a manner that derives the time series data of the second node.

[0082] In the expression represented by symbol 520, β, α, and C represent specified coefficients, and the following coefficients are represented as follows:

[0083] ·t: time

[0084] • m: Autocorrelation (indicating whether or not there is periodicity)

[0085] • n: Cross-correlation (a coefficient indicating whether they are correlated).

[0086] ·k: Time delay.

[0087] exist Figure 5 In the learning result 530, the coefficients (representing the values ​​of the relationships) of the numerical expression represented by symbol 520, calculated by machine learning on the time series analysis model, are shown. Specifically, the learning result 530 includes "first node", "second node", "autocorrelation", "cross-correlation", and "time delay" as information items.

[0088] In learning result 530, for the "first node" and "second node", the time series data of the time series data group included in the learning data is stored respectively, which is used to derive the formula represented by symbol 520.

[0089] In addition, in learning result 530, the coefficients m, n, and k for "autocorrelation", "cross-correlation", and "time delay" are saved by deriving the time series data of the second node by inputting the time series data of the first node into the formula represented by symbol 520.

[0090] In addition, such as Figure 5 As shown, only one learning result 530 is generated, relative to the time series data set (learning data) measured when etching using a standard process under baseline conditions.

[0091] <Specific examples of the processing performed by the determination department>

[0092] Next, a specific example of the processing performed by the determination unit 152 of the analysis device 150 will be described. Figure 6 The first figure shows a specific example of the processing performed by the determining unit. For example... Figure 6 As shown, the determination unit 152 includes a time series analysis model execution unit 610 and a count value calculation unit 620.

[0093] The time series analysis model execution unit 610 extracts the time series data (measured value 611) of the first node from the time series data set (determination data) measured in the determination interval immediately after the plasma source is turned on during the etching process. Furthermore, the time series analysis model execution unit 610 infers the time series data of the second node (inferred value 612) by inputting the extracted time series data of the first node into the formula represented by symbol 520.

[0094] At this time, the time series analysis model execution unit 610 reads the coefficients m, n, and k corresponding to the time series data input into the formula represented by symbol 520 from the learning result 530, sets them into the formula represented by symbol 520, and then infers the time series data of the second node.

[0095] exist Figure 6 In this context, the measured value 611 represents the time series data of the first node in the formula represented by symbol 520, which is measured in the time series data group (determination data) immediately after the plasma source is turned on during the etching process. Furthermore, the inferred value 612 represents the time series data of the second node inferred by inputting the measured value 611.

[0096] On the other hand, the count value calculation unit 620 has a difference calculation unit 621, a counting unit 622, and a conversion unit 623.

[0097] The difference calculation unit 621 extracts the time series data (measured value 624) of the second node from the time series data group (judgment data) measured immediately after the plasma source is turned on during the etching process. Furthermore, the difference calculation unit 621 obtains the inferred value 612 through the time series analysis model execution unit 610. Finally, the difference calculation unit 621 calculates the difference between the measured value 624 and the inferred value 612.

[0098] The counting unit 622 counts the number of first nodes whose difference calculated by the difference calculation unit 621 is above a predetermined threshold (that is, a predetermined count value). In addition, the counting unit 622 outputs the predetermined count value obtained from the counting as the deviation of the state of the chamber 120 in the judgment interval from the reference state.

[0099] The transformation unit 623 determines a characteristic value (etching rate) based on the deviation of the state of the chamber 120 in the judgment interval from the reference state. This characteristic value is used to determine the process control data for etching processing in the control interval that is later than the judgment interval.

[0100] Furthermore, in order to conduct experiments in advance to determine the correspondence between a predetermined count value representing the deviation from the reference state and the etching rate, the transformation unit 623 determines the characteristic value (e.g., etching rate) based on this correspondence.

[0101] Figure 7 This is a graph showing the correspondence between a specified count value and the etching rate. Figure 7 In (a), the horizontal axis represents the number of wafers that underwent etching, and the vertical axis represents the specified count value. For example... Figure 7 As shown in (a), there is a linear relationship between the number of wafers and the specified count value.

[0102] On the other hand, Figure 7 In (b), the horizontal axis represents the number of wafers that underwent etching, and the vertical axis represents the etching rate. For example... Figure 7 As shown in (b), there is a linear relationship between the number of wafers and the etching rate.

[0103] according to Figure 7 of (a), Figure 7 (b) The etching rate can be determined by counting the specified count values.

[0104] <Analysis and Control Process>

[0105] Next, the process of analysis and control processing in the etching process control system 100 will be explained. Figure 8 This is the first flowchart representing the process of analysis and control.

[0106] In step S801, the time series data acquisition devices 140_1 to 140_n save the time series data set (learning data) measured during etching using a standard process in the reference state to the learning data storage unit 153.

[0107] In step S802, the learning unit 151 of the analysis device 150 uses the time series data group (learning data) stored in the learning data storage unit 153 to perform machine learning on the time series analysis model.

[0108] In step S803, when the etching process in chamber 120 begins, the time series data acquisition devices 140_1 to 140_n measure the time series data set (judgment data) in the judgment interval immediately following the plasma source being turned on.

[0109] In step S804, the determination unit 152 of the analysis device 150 inputs the time series data set (judgment data) measured in step S803 into the time series analysis model to calculate the deviation from the reference state (counting a specified count value).

[0110] In step S805, the determination unit 152 of the analysis device 150 determines a characteristic value (e.g., etching rate) based on the deviation from the reference state. This characteristic value is used to determine the process control data for etching processing in the control interval that is later than the judgment interval.

[0111] In step S806, the control device 160 determines the process control data for the etching process in the control range based on the determined characteristic values.

[0112] Therefore, in semiconductor manufacturing processes, etching processes within a control range can be controlled based on determined process control data.

[0113] Summary

[0114] As can be clearly understood from the above description, in the etching process control system involved in the first embodiment,

[0115] • The deviation from the reference state of the chamber is calculated by inputting the time series data set measured in the plasma processing chamber and the time series data set measured in the judgment interval into the time series analysis model (counting the specified count values).

[0116] • The characteristic value (etching rate) is determined based on the calculated deviation. This characteristic value is used to determine the process control data for etching processes in the control interval that is later than the judgment interval.

[0117] • The etching process within the control range is controlled based on the determined characteristic value (etching rate).

[0118] Thus, according to the first embodiment, an analysis apparatus, an analysis method, an analysis program, and an etching processing control system that can quantitatively evaluate the state of a processing space by using time series data sets measured in the processing space where plasma processing is performed can be provided.

[0119] [Second Embodiment]

[0120] In the above first embodiment, specific examples of the time series data acquisition apparatus and the time series data set are not mentioned. Accordingly, in the second embodiment, the case where the time series data acquisition apparatus is an optical emission spectroscopy apparatus and the time series data set is OES (Optical Emission Spectroscopy) data will be described. In addition, OES data refers to a data set including time series data of luminous intensity corresponding to the number of wavelength types. Hereinafter, the second embodiment will be described focusing on differences from the above first embodiment.

[0121] <System Structure of Etching Processing Control System>

[0122] First, the system structure of the etching processing control system when OES data is used will be described. Figure 9 is the second drawing showing an example of the system structure of the etching processing control system. The difference from Figure 1 is that, in the case of the etching processing control system 100', an optical emission spectroscopy apparatus 940 is provided as the time series data acquisition apparatus. In addition, the difference from Figure 1 is that OES data is stored as a time series data set (learning data) in the learning data storage unit 153, and the OES data is notified to the determination unit 152 as a time series data set (judgment data).

[0123] The optical emission spectroscopy apparatus 940 measures OES data before or during the etching processing of a pre-processing wafer 110 in the chamber 120 by means of optical emission spectroscopy technology. The OES data is, for example, time series data of luminous intensity at each time for each wavelength included in the visible light wavelength range.

[0124] <Specific Example of OES Data>

[0125] Next, a specific example of OES data measured by the optical emission spectroscopy apparatus 940 will be described. Figure 10 is a drawing showing an example of OES data, which shows a luminous intensity data group obtained when measuring each wavelength included in the visible light wavelength range (400 nm to 800 nm) at a 1 nm interval. In addition, in Figure 10 , the horizontal axis represents time, and the vertical axis represents the luminous intensity of each wavelength.

[0126] exist Figure 10 In some cases, the top-level graph shows the luminescence intensity data at various times for a wavelength of 400 nm, while the second-level graph shows the luminescence intensity data at various times for a wavelength of 401 nm. Additionally, Figure 10 The graph of the third layer shows the luminescence intensity data at various times with a wavelength of 402 nm.

[0127] In this way, the etching process control system 100' according to the second embodiment can achieve the same processing as the first embodiment described above by configuring the emission spectrum analysis device 940 as a time series data acquisition device.

[0128] As a result, according to the second embodiment, the same effect as the first embodiment described above can be obtained.

[0129] Furthermore, in the above description, it is assumed that the time series data acquisition device is an emission spectroscopy analysis device and the time series data set is OES data, but the time series data acquisition device can also be a mass spectrometry device (e.g., a quadrupole mass spectrometer). In this case, the time series data set is a dataset containing a number of time series data (mass analysis data) with a corresponding number of detection intensities corresponding to the number of types of mass-related values ​​(m / z values).

[0130] [Third Implementation Method]

[0131] In the second embodiment described above, the case where the time series data set is OES data was explained. However, the time series data set is not limited to OES data; for example, it can also be a set of process data (RF power data, pressure data, temperature data, etc.) measured by various processing sensors.

[0132] The third embodiment will now be described, focusing on the differences from the first or second embodiment described above.

[0133] <System Structure of Etching Process Control System>

[0134] First, the system structure of the etching process control system using process data sets will be explained. Figure 11 The third figure illustrates an example of the system structure of an etching process control system. (Compared to...) Figure 1 The difference lies in the fact that, in the case of the etching process control system 100", process data acquisition devices 1140_1, 1140_2, ... 1140_n are configured as time series data acquisition devices. Additionally, compared to... Figure 1The difference is that the learning data storage unit 153 stores the process data set as a time series data set (learning data) and notifies the determination unit 152 that the process data set is used as a time series data set (determination data).

[0135] Process data acquisition devices 1140_1, 1140_2, ... 1140_n measure a set of process data in chamber 120 before or during etching of the unprocessed wafer 110. The process data set includes, for example, RF power supply data, pressure data, gas flow rate data, current data, voltage data, and temperature data at various times.

[0136] <Specific examples of process data sets>

[0137] Next, a specific example of the process data set measured by the process data acquisition devices 1140_1, 1140_2, ... 1140_n will be described. Figure 12 This is a diagram illustrating an example of a set of process data. Figure 12 The example illustrates a scenario where process data acquisition device 1140_1 measures RF power data as process data 1, and process data acquisition device 1140_2 measures pressure data as process data 2. Additionally, Figure 12 The example shows a case where the process data acquisition device 1140_3 measures gas flow rate data as process data 3.

[0138] Similarly, Figure 12 The example illustrates a scenario where process data acquisition device 1140_n-2 measures current data as process data n-2, and process data acquisition device 1140_n-1 measures voltage data as process data n-1. Additionally, Figure 12 The example illustrates a case where the process data acquisition device 1140_n measures temperature data as process data n.

[0139] In this way, the etching process control system 100 according to the third embodiment can achieve the same processing as the first embodiment by configuring process data acquisition devices 1140_1 to 1140_n as time series data acquisition devices.

[0140] As a result, according to the third embodiment, the same effect as the first embodiment described above can be obtained.

[0141] [Fourth Implementation Method]

[0142] In the first to third embodiments described above, a predetermined time immediately following the switching on of the plasma source during the etching process is defined as the judgment interval. However, the timing of the judgment interval is not limited to this; for example, it can be any predetermined time preceding the control interval. Figure 13This is the second figure, which illustrates the relationship between the plasma processing flow and the judgment and control intervals in the processing space.

[0143] in, Figure 13 (a) shows an example where a predetermined time immediately preceding the cutoff of the plasma source during the aging process is set as the judgment interval. In this case, the control interval can be set from the start to the end of the etching process, and the process control data can be determined using characteristic values ​​based on the time series data set (judgment data) measured within the judgment interval to control the etching process.

[0144] in addition, Figure 13 (b) shows an example of setting up a process (state evaluation process) between aging and etching processes to evaluate the state of chamber 120 and using this process as a judgment interval. In this case, it is also possible to set the etching process from the start to the end as a control interval, and use the characteristic values ​​determined based on the time series data set (judgment data) measured in the judgment interval to determine the process control data and control the etching process.

[0145] In other words, according to this embodiment, the state of the chamber 120 can be quantitatively evaluated based on a set of time-series data over a specified period of time before the etching process begins, thereby controlling the etching process.

[0146] [Fifth Implementation Method]

[0147] In the first to fourth embodiments described above, models that represent the relationship between time series data using linear or nonlinear regression were described as time series analysis models. In contrast, in the fifth embodiment, a deviation value detection model is used instead of a time series analysis model to detect the deviation value of the data based on the data density of each time series data. The fifth embodiment will now be described focusing on the differences from the first to fourth embodiments. Furthermore, in the fifth embodiment, the case where the time series data set is OES data will be described.

[0148] <Specific examples of the handling conducted by the Learning Department>

[0149] First, a specific example of the processing performed by the learning unit of the analysis device 150 will be explained. Figure 14 The second figure shows a specific example of the processing performed in the learning department. For example... Figure 14 As shown, the learning unit 1400 has a number of deviation value detection models (models 1410_1 to 1410_401) corresponding to the number of wavelength types contained in the OES data 1430.

[0150] The time-series luminescence intensity data with a wavelength of 400 nm, measured in OES data 1430 during etching using a standard process under normal conditions (as a reference), is input into model 1410_1. Model 1410_1 then calculates the data density of the luminescence intensity data at each time point. Furthermore, model 1410_1 calculates the range of deviation values ​​under normal conditions (as a reference). Additionally, the range of deviation values ​​calculated by model 1410_1 is set as normal range information in model 1510_1 (described later).

[0151] Similarly, the time-series luminescence intensity data with a wavelength of 401 nm from the OES data 1430, measured during etching using a standard process under normal conditions as a reference, is input into model 1410_2. Model 1410_2 then calculates the data density of the luminescence intensity data at each time point. Furthermore, model 1410_2 calculates the range of deviation values ​​under normal conditions as a reference. In addition, the range of deviation values ​​calculated by model 1410_2 is set as normal range information in model 1510_2 (described later).

[0152] Below, in Figure 14 The function blocks for processing the luminescence intensity data at various times from wavelength = 402nm to wavelength = 799nm are omitted, and therefore the descriptions of these function blocks are also omitted.

[0153] The time-series luminescence intensity data with a wavelength of 800 nm, measured in OES data 1430 during etching using a standard process under normal conditions (as a reference), is input into model 1410_401. Model 1410_401 then calculates the data density of the luminescence intensity data at each time point. Furthermore, model 1410_401 calculates the range of deviation values ​​under normal conditions (as a reference). Additionally, the range of deviation values ​​calculated by model 1410_401 is set as normal range information in model 1510_401 (described later).

[0154] <Specific examples of the processing performed by the determination department>

[0155] Next, a specific example of the processing performed by the determination unit of the analysis device 150 will be described. Figure 15 The second figure shows a specific example of the processing performed by the determining unit. For example... Figure 15 As shown, the determination unit 1500 has a number of learned deviation value detection models (models 1510_1 to 1510_401) corresponding to the number of wavelength types contained in the OES data 1540, and a deviation calculation unit (deviation calculation unit 1520_1 to 1520_401). In addition, the determination unit 1500 has a transformation unit 1530.

[0156] The time-series luminescence intensity data with a wavelength of 400 nm from the OES data 1540 measured by the emission spectroscopy analyzer 940 within the judgment interval is input into the model 1510_1. The model 1510_1 then detects the deviation value of the luminescence intensity data at each time based on the data density of the luminescence intensity data at each time. Furthermore, the model 1510_1 determines whether there is a deviation value in the luminescence intensity data at each time based on the set normal range information and notifies the deviation calculation unit 1520_1.

[0157] The deviation calculation unit 1520_1 calculates the overall deviation of the luminous intensity data with wavelength = 400nm based on binary information indicating whether there is a deviation value notified from model 1510_1, and notifies the transformation unit 1530.

[0158] Similarly, the time-series luminescence intensity data with a wavelength of 401 nm from the OES data 1540 measured by the emission spectroscopy analyzer 940 within the judgment interval is input into the model 1510_2. Thus, the model 1510_2 detects the deviation value of the luminescence intensity data at each time based on the data density of the luminescence intensity data at each time. Furthermore, the model 1510_2 determines whether there is a deviation value in the luminescence intensity data at each time based on the set normal range information and notifies the deviation calculation unit 1520_2.

[0159] The deviation calculation unit 1520_2 calculates the overall deviation of the luminous intensity data with wavelength = 401nm based on binary information indicating whether there is a deviation value notified from model 1510_2, and notifies the transformation unit 1530.

[0160] Below, in Figure 15 The function blocks for processing luminescence intensity data at various times from wavelength 402nm to 799nm are omitted, and therefore the descriptions of these function blocks are also omitted.

[0161] The time-series luminescence intensity data with a wavelength of 800 nm from the OES data 1540 measured by the emission spectroscopy analyzer 940 within the judgment interval is input into the model 1510_401. Therefore, the model 1510_401 detects the deviation value of the luminescence intensity data at each time based on the data density of the luminescence intensity data at each time. Furthermore, the model 1510_401 determines whether there is a deviation value in the luminescence intensity data at each time based on the set normal range information and notifies the deviation calculation unit 1520_401.

[0162] The deviation calculation unit 1520_401 calculates the overall deviation of the luminous intensity data with wavelength = 800nm ​​based on binary information indicating whether there is a deviation value notified from model 1510_401, and notifies the transformation unit 1530.

[0163] The transformation unit 1530 determines the deviation calculation unit corresponding to a specific model in models 1510_1 to 1510_401. Furthermore, based on the deviation notified from the determined deviation calculation unit, the transformation unit 1530 determines the characteristic value (etching rate) of the process control data used to determine the etching process within the control range.

[0164] Furthermore, experiments are conducted beforehand to determine the correspondence between the deviation and the etching rate output by the deviation calculation units 1520_1 to 1520_401 corresponding to a specific model. Then, in the transformation unit 1530, a characteristic value (e.g., etching rate) is determined based on the correspondence obtained through experiments.

[0165] <Specific examples of deviation>

[0166] Next, specific examples of the deviation values ​​output from the deviation calculation units 1520_1 to 1520_401 will be explained. Figure 16 This is a graph showing specific examples of deviation. In Figure 16 In the diagram, the horizontal axis represents the types of wavelengths. The vertical axis represents the luminous intensity data for each wavelength over a specified time period, as well as the overall deviation of the luminous intensity data for each wavelength.

[0167] in, Figure 16 (a) represents the luminous intensity data of each wavelength in chamber state A, and the overall deviation of the luminous intensity data of each wavelength calculated by applying model 1510_1 to 1510_401 to the luminous intensity data of each wavelength.

[0168] in addition, Figure 16 (b) represents the luminous intensity data for each wavelength in chamber state B, and the overall deviation of the luminous intensity data for each wavelength calculated by applying models 1510_1 to 1510_401 to the luminous intensity data for each wavelength. Furthermore, it is known that chamber state A and chamber state B are different.

[0169] exist Figure 16 (a) and Figure 16 In (b), the luminous intensity data for each wavelength are similar, but the deviation of the luminous intensity for a specific wavelength is quite different. That is to say, the deviation output from the specific deviation calculation unit in the deviation calculation unit 1520_1 to 1520_401 can accurately reflect the state of the chamber 120.

[0170] <Correspondence between deviation and etching rate>

[0171] Next, the correspondence between the overall deviation of the luminous intensity data of a specific wavelength output from the deviation calculation units 1520_1 to 1520_401 and the etching rate will be explained. Figure 17 This is a graph showing the relationship between deviation and etching rate. Figure 17 In the diagram, the horizontal axis represents the deviation (the overall deviation of the luminescence intensity data for each wavelength), and the vertical axis represents the etching rate. For example... Figure 17 As shown, the overall deviation of the luminescence intensity data for each wavelength has a roughly linear relationship with the etching rate.

[0172] However, in the transformation section 1530, by referring to Figure 17 It can determine the etching rate based on the deviation output from a specific deviation calculation unit.

[0173] <Analysis and Control Process>

[0174] Next, the analysis and control process in the etching process control system 100' will be explained. Figure 18 This is the second flowchart representing the analysis and control process. (And...) Figure 8 The difference in the first flowchart shown is in steps S1801, S1802, and S1803.

[0175] In step S1801, the learning unit 1400 of the analysis device 150 acquires OES data measured under normal conditions as a reference and uses it as learning data. Furthermore, the learning unit 1400 calculates the data density based on the luminescence intensity data of each wavelength contained in the acquired OES data, calculates the range of deviation values, sets normal range information, and thereby generates a deviation value detection model after learning is complete.

[0176] In step S1802, the determination unit 1500 acquires OES data as judgment data, inputs the luminous intensity data of each wavelength contained in the OES data into the learning completed deviation value detection model, and thereby calculates the overall deviation of the luminous intensity data of each wavelength.

[0177] In step S1803, the determination unit 1500 of the analysis device acquires the deviation with respect to a specific wavelength, and determines the characteristic value (etching rate) of the process control data used to determine the etching process in the control range based on the acquired deviation.

[0178] Summary

[0179] As can be seen from the above description, the etching process control system involved in the fifth embodiment performs the following controls:

[0180] • For each wavelength, the OES data measured in the plasma processing chamber and the OES data measured in the judgment interval are input into the deviation value detection model, thereby calculating the overall deviation of the luminescence intensity data for each wavelength.

[0181] • Based on the overall deviation of the luminescence intensity data calculated for a specific wavelength, determine the characteristic value (etching rate) of the process control data used to determine the etching process in the control interval that is later than the judgment interval.

[0182] • The etching process within the control range is controlled based on the determined characteristic value (etching rate).

[0183] Therefore, according to the fifth embodiment, it is possible to provide an analysis apparatus, analysis method, analysis procedure, and etching process control system that uses OES data measured in a processing space where plasma processing is performed to quantitatively evaluate the state of the processing space.

[0184] [Other Implementation Methods]

[0185] In the first to fourth embodiments described above, the learning unit performed machine learning on a cross-correlation model, which is an example of a time series analysis model. However, the model used by the learning unit for machine learning is not limited to a cross-correlation model; it can be any model capable of calculating the correlation of time series data, or other models.

[0186] Furthermore, in the first to fourth embodiments described above, a predetermined count value was counted by the counting unit 622 counting the number of first nodes whose difference calculated by the difference calculation unit 621 is above a predetermined threshold. However, the method for counting the predetermined count value is not limited to this. For example, the predetermined count value may also be counted by counting a predetermined number of first nodes among the first nodes whose difference calculated by the difference calculation unit 621 is above a predetermined threshold.

[0187] Furthermore, in the above embodiments, the etching rate was described as a characteristic value used to determine the process control data during the etching process. However, the characteristic value determined based on the deviation is not limited to the etching rate. Any characteristic value used to determine the process control data during the etching process (a characteristic value representing process variation) and that is related to the deviation can be determined; other characteristic values ​​may also be determined.

[0188] Furthermore, in the above embodiments, the case of determining characteristic values ​​for determining process control data during etching is described. However, the characteristic values ​​determined based on deviation are not limited to those used to determine process control data during etching; they could also be characteristic values ​​used to determine process control data during plasma treatment of the substrate. Moreover, the plasma treatment of the substrate mentioned here includes not only etching but also film formation and ashing processes.

[0189] Furthermore, in the second and fifth embodiments described above, the case where learning data is generated using luminous intensity data of each wavelength included within the visible light wavelength range was described. However, the luminous intensity data used when generating learning data can also be luminous intensity data of a specific wavelength. Alternatively, it can be luminous intensity data of wavelengths outside the visible light wavelength range.

[0190] Furthermore, while the second embodiment described above uses OES data as a specific example of a time-series data set, and the third embodiment uses process data as a specific example, the time-series data set is not limited to these examples. For example, it could also be a time-series data set representing plasma physical quantities measured by a plasma device. Similarly, while the fifth embodiment describes OES data as a specific example of a time-series data set, the time-series data set is not limited to these examples. For instance, it could be either a process data set or a time-series data set representing plasma physical quantities measured by a plasma device.

[0191] Furthermore, while the fifth embodiment described above illustrates the use of a deviation value detection model, other models that detect data deviation values ​​based on the data density of each time series data can also be used.

[0192] Furthermore, in the above embodiments, the analysis device and the control device are configured separately, but they can also be configured as an integral unit. Additionally, in the above embodiments, the control device and the semiconductor manufacturing process are configured separately, but they can also be configured as an integral unit.

[0193] Furthermore, the present invention is not limited to the structures shown herein obtained by combining other elements such as the structures listed in the above embodiments. Changes can be made to these points without departing from the spirit of the invention, and can be appropriately determined according to their application.

Claims

1. An analytical apparatus for evaluating the state of a processing space where plasma processing is performed, comprising: The learning department uses learning data to perform machine learning on a time series analysis model, the learning data being a set of time series data in the processing space with a baseline state; The computing unit inputs the time series data set measured in the processing space where the plasma processing is performed, specifically the time series data set measured in the judgment interval which is a predetermined time earlier than the control interval, into the learned time series analysis model to calculate the deviation from the reference state of the processing space. as well as The determination unit determines a characteristic value based on the deviation obtained from the calculation. This characteristic value is used to determine the control data for plasma treatment of the substrate in the control interval immediately following the determination interval.

2. The analytical apparatus according to claim 1, characterized in that, The learning unit uses a set of time-series data measured during etching in the processing space under the baseline state to perform machine learning on the time-series analysis model and calculates a value representing the relationship between the time-series data of the first node and the time-series data of the second node. The calculation unit calculates the deviation based on the number of first nodes whose difference is above a predetermined threshold. The difference is the difference between the time series data of the second node inferred by inputting the time series data of the first node extracted from the time series data set measured in the judgment interval into the learned time series analysis model and the time series data of the second node extracted from the time series data set measured in the judgment interval.

3. The analytical apparatus according to claim 1 or 2, characterized in that, The determining unit determines the etching rate based on the deviation obtained from the calculation. This etching rate is used to determine the control data for etching processing in the control interval that is later than the judgment interval.

4. The analytical apparatus according to claim 1 or 2, characterized in that, The judgment interval is a predetermined time after the plasma source is turned on during the etching process, and the control interval is the interval of the etching process that follows the judgment interval.

5. The analytical apparatus according to claim 1 or 2, characterized in that, The judgment interval is a predetermined time before the etching process begins, and the control interval is the interval from the start to the end of the etching process.

6. The analytical apparatus according to claim 1, characterized in that, The time series data set consists of emission spectral data measured by an emission spectroscopy analyzer or mass analysis data measured by a mass spectrometer.

7. The analytical apparatus according to claim 1, characterized in that, The time series data set is the process data set measured by the process data acquisition device.

8. The analytical apparatus according to claim 1, characterized in that, The time series data set is a time series data set of plasma physical quantities measured by a plasma device.

9. The analytical apparatus according to claim 2, characterized in that, The values ​​representing the relationship include autocorrelation, cross-correlation, or time delay calculated by deriving the time series data of the second node by inputting the time series data of the first node into a prescribed formula.

10. The analytical apparatus according to claim 1, characterized in that, The computing unit uses a number of deviation value detection models corresponding to the number of types of time series data included in the time series data group to calculate the deviation, instead of using the learned time series analysis model to calculate the deviation.

11. The analytical apparatus according to claim 10, characterized in that, The determining unit determines the characteristic value based on the deviation calculated based on binary information, wherein the binary information indicates whether there is a deviation value determined by a specific model in the deviation value detection model.

12. A plasma processing control system, comprising: The analytical apparatus according to any one of claims 1 to 11; and A control device that determines control data for plasma treatment of the substrate based on the aforementioned characteristic values.

13. An analytical method for evaluating the state of a processing space where plasma processing is performed, comprising the following steps: The time series analysis model is machine learning using learning data, which is a set of time series data in the processing space with a baseline state. The time series data set measured in the processing space where the plasma treatment is performed, specifically the time series data set measured within a judgment interval that is a predetermined time preceding the control interval, is input into the learned time series analysis model to calculate the deviation relative to the reference state of the processing space; and A characteristic value is determined based on the deviation obtained from the calculation. This characteristic value is used to determine the control data for plasma treatment of the substrate in the control interval immediately following the judgment interval.

14. A non-volatile computer-readable recording medium storing an analysis program for causing a computer to perform the following steps: Machine learning is performed on a time series analysis model using learning data, which is a set of time series data with a processing space having a baseline state. The time series data set measured in the processing space where plasma processing is performed, specifically the time series data set measured within a judgment interval that is a predetermined time preceding the control interval, is input into the learned time series analysis model to calculate the deviation relative to the reference state of the processing space; and A characteristic value is determined based on the calculated deviation, and this characteristic value is used to determine the control data for plasma treatment of the substrate in the control interval immediately following the judgment interval.

15. A computer program product comprising a computer program that, when executed by a processor, performs the analysis method according to claim 13.

Images