Data calculation method and substrate processing apparatus

Abstract

To provide a data calculation method and a substrate processing apparatus capable of obtaining levels and ratios for each level of a pulse waveform having a plurality of levels.SOLUTION: A data calculation method includes obtaining a first set of data over a period of time, dividing the obtained first data group into a plurality of groups according to a value range of each data included in the first data group, extracting a second data group included in a valid group from among the divided groups, and outputting statistics for each of the groups on the basis of the extracted second data set.SELECTED DRAWING: Figure 9

Description

Technical Field

[0001] The present disclosure relates to a data calculation method and a substrate processing apparatus.

Background Art

[0002] In a plasma processing apparatus, for example, by supplying an RF (Radio Frequency) voltage to a substrate to be processed, ions and radicals generated in the plasma are drawn into the substrate, and a process such as etching is executed. At this time, the voltage (Vpp) on the substrate is monitored and recorded as an index of the process state, and is used for predicting the process result, state monitoring, abnormality detection, etc. In addition, it has been proposed to monitor the peak voltage value of the pulsed RF bias voltage and perform feedback control (Patent Document 1).

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] The present disclosure provides a data calculation method and a substrate processing apparatus capable of obtaining the levels and ratios for each level of a pulse waveform having a plurality of levels.

Means for Solving the Problems

[0005] A data calculation method according to an aspect of the present disclosure includes: obtaining a first data group in a predetermined period; dividing the obtained first data group into a plurality of groups according to the value range of each data included in the first data group; extracting a second data group included in a valid group from the plurality of divided groups; and outputting a statistical value for each group based on the extracted second data group.

Effects of the Invention

[0006] According to this disclosure, the levels and ratios of each level of a pulse waveform having multiple levels can be determined. [Brief explanation of the drawing]

[0007] [Figure 1] Figure 1 shows an example of a plasma processing apparatus in one embodiment of the present disclosure. [Figure 2] Figure 2 is a block diagram showing an example of the functional configuration of the control unit in this embodiment. [Figure 3] Figure 3 shows an example of the bias voltage waveform when multiple RF signals are supplied. [Figure 4] Figure 4 is a diagram illustrating the general grouping in this embodiment. [Figure 5] Figure 5 shows an example of grouping in this embodiment. [Figure 6] Figure 6 shows an example of how to perform grouping using the average value in this embodiment. [Figure 7] Figure 7 shows an example of effective group extraction in this embodiment. [Figure 8] Figure 8 shows an example of the level-specific energy levels calculated in this embodiment. [Figure 9] Figure 9 is a flowchart showing an example of the data calculation process in this embodiment. [Figure 10] Figure 10 is a flowchart showing an example of the extraction process in this embodiment. [Figure 11] Figure 11 shows an example of the results of one-dimensional cluster analysis in Example 1 according to this embodiment. [Figure 12] Figure 12 shows an example of the results obtained when the one-dimensional cluster analysis of Example 1 is repeated a predetermined number of times. [Figure 13] Figure 13 shows an example of the ratio of data points for each level in Example 1. [Figure 14]FIG. 14 is a diagram showing an example of the relationship between the dataset of Example 2 and the grouping. [Figure 15] FIG. 15 is a diagram showing an example of the relationship between the dataset of Example 3 and the grouping. [Figure 16] FIG. 16 is a diagram showing an example of the calculated levels for each level in Example 3. [Embodiments for Carrying Out the Invention]

[0008] Hereinafter, embodiments of the disclosed data calculation method and substrate processing apparatus will be described in detail based on the drawings. Note that the disclosed technology is not limited by the following embodiments.

[0009] In recent years, in a plasma processing apparatus, in some cases, fine process control is performed by supplying an RF signal in pulses that are turned on and off at high speed. When the RF signal is supplied in pulses, the voltage (Vpp) on the substrate to be monitored also becomes pulsed, so it is required to monitor the pulsed voltage (Vpp). However, when monitoring the peak voltage value of the pulsed RF bias voltage, it is difficult to monitor the level for each level for a pulse waveform having a plurality of levels such as superimposing RF signals of different frequencies and supplying them.

[0010] Also, as another method, it is conceivable to detect the level based on the timing of the pulses of the bias RF signal, but it is difficult to detect the level when the output of the source RF signal changes. Furthermore, in the method of differentiating the pulse waveform to extract the change points, it is easily affected by noise, and also, a mechanism for performing level extraction is separately required, which is difficult to use for control in a plasma processing apparatus. Also, in general cluster analysis, since it targets classification in a plurality of dimensions, the amount of calculation increases, which is difficult to use for control in a plasma processing apparatus. Therefore, it is expected to obtain the level and ratio for each level of a pulse waveform having a plurality of levels.

[0011] [Configuration of Plasma Processing Apparatus] The following describes an example configuration of a capacitively coupled plasma processing apparatus as an example of a plasma processing apparatus 1. Figure 1 is a diagram showing an example of a plasma processing apparatus in one embodiment of the present disclosure. As shown in Figure 1, the capacitively coupled plasma processing apparatus 1 includes a plasma processing chamber 10, a gas supply unit 20, a power supply 30, and an exhaust system 40. The plasma processing apparatus 1 also includes a substrate support unit 11 and a gas introduction unit. The gas introduction unit is configured to introduce at least one processing gas into the plasma processing chamber 10. The gas introduction unit includes a shower head 13. The substrate support unit 11 is located inside the plasma processing chamber 10. The shower head 13 is located above the substrate support unit 11. In one embodiment, the shower head 13 constitutes at least a part of the ceiling of the plasma processing chamber 10. The plasma processing chamber 10 has a plasma processing space 10s defined by the shower head 13, the side wall 10a of the plasma processing chamber 10, and the substrate support unit 11. The side wall 10a is grounded. The shower head 13 and the substrate support portion 11 are electrically insulated from the plasma processing chamber 10 housing.

[0012] The substrate support portion 11 includes a main body portion 111 and a ring assembly 112. The main body portion 111 has a central region (substrate support surface) 111a for supporting a substrate (wafer) W and an annular region (ring support surface) 111b for supporting the ring assembly 112. The annular region 111b of the main body portion 111 surrounds the central region 111a of the main body portion 111 in a plan view. The substrate W is disposed on the central region 111a of the main body portion 111, and the ring assembly 112 is disposed on the annular region 111b of the main body portion 111 so as to surround the substrate W on the central region 111a of the main body portion 111. In one embodiment, the main body portion 111 includes a base and an electrostatic chuck. The base includes a conductive member. The conductive member of the base functions as a lower electrode. The electrostatic chuck is disposed on the base. The upper surface of the electrostatic chuck has the substrate support surface 111a. The ring assembly 112 includes one or more annular members. At least one of the one or more annular members is an edge ring. Also, although not shown, the substrate support portion 11 may include a temperature control module configured to adjust at least one of the electrostatic chuck, the ring assembly 112, and the substrate W to a target temperature. The temperature control module may include a heater, a heat transfer medium, a flow path, or a combination thereof. A heat transfer fluid such as brine or gas flows through the flow path. Further, the substrate support portion 11 may include a heat transfer gas supply portion configured to supply a heat transfer gas between the back surface of the substrate W and the substrate support surface 111a.

[0013] The shower head 13 is configured to introduce at least one processing gas from the gas supply unit 20 into the plasma processing space 10s. The shower head 13 has at least one gas supply port 13a, at least one gas diffusion chamber 13b, and a plurality of gas inlet ports 13c. The processing gas supplied to the gas supply port 13a passes through the gas diffusion chamber 13b and is introduced into the plasma processing space 10s through the plurality of gas inlet ports 13c. The shower head 13 also includes a conductive member. The conductive member of the shower head 13 functions as an upper electrode. In addition to the shower head 13, the gas introduction unit may also include one or more side gas injectors (SGIs) attached to one or more openings formed in the side wall 10a.

[0014] The gas supply unit 20 may include at least one gas source 21 and at least one flow controller 22. In one embodiment, the gas supply unit 20 is configured to supply at least one processing gas to the shower head 13 from a corresponding gas source 21 via a corresponding flow controller 22. Each flow controller 22 may include, for example, a mass flow controller or a pressure-controlled flow controller. Furthermore, the gas supply unit 20 may include at least one flow modulation device that modulates or pulses the flow rate of at least one processing gas.

[0015] The power supply 30 includes an RF power supply 31 coupled to the plasma processing chamber 10 via at least one impedance matching circuit. The RF power supply 31 is configured to supply at least one RF signal (RF power), such as a source RF signal and a bias RF signal, to the conductive members of the substrate support 11 and / or the showerhead 13. This causes plasma to be formed from at least one processing gas supplied to the plasma processing space 10s. Thus, the RF power supply 31 can function as at least part of the plasma generation unit. Furthermore, by supplying the bias RF signal to the conductive members of the substrate support 11, a bias potential is generated on the substrate W, and ionic components in the formed plasma can be drawn into the substrate W.

[0016] In one embodiment, the RF power supply 31 includes a first RF generation unit 31a and a second RF generation unit 31b. The first RF generation unit 31a is coupled to a conductive member of the substrate support unit 11 and / or a conductive member of the shower head 13 via at least one impedance matching circuit and is configured to generate a source RF signal (source RF power) for plasma generation. In one embodiment, the source RF signal has a frequency in the range of 13 MHz to 150 MHz. In one embodiment, the first RF generation unit 31a may be configured to generate a plurality of source RF signals having different frequencies. One or more generated source RF signals are supplied to the conductive member of the substrate support unit 11 and / or a conductive member of the shower head 13. The second RF generation unit 31b is coupled to a conductive member of the substrate support unit 11 via at least one impedance matching circuit and is configured to generate a bias RF signal (bias RF power). In one embodiment, the bias RF signal has a lower frequency than the source RF signal. In one embodiment, the bias RF signal has a frequency in the range of 400 kHz to 13.56 MHz. In one embodiment, the second RF generation unit 31b may be configured to generate a plurality of bias RF signals having different frequencies. The generated bias RF signals are supplied to the conductive member of the substrate support unit 11. In various embodiments, at least one of the source RF signal and the bias RF signal may be pulsed.

[0017] For example, the first RF generation unit 31a is electrically connected to the conductive member of the shower head 13 via a conductive part 33a such as wiring. An impedance matching circuit 34a is provided in the conductive part 33a. The impedance matching circuit 34a matches the output impedance of the first RF generation unit 31a with the input impedance of the load side (shower head 13 side). The first RF generation unit 31a supplies a source RF signal for generating plasma to the conductive member of the shower head 13.

[0018] Furthermore, for example, the second RF generation unit 31b is electrically connected to the conductive member of the base of the substrate support unit 11 via a conductive part 33b such as wiring. An impedance matching circuit 34b is provided on the conductive part 33b. The impedance matching circuit 34b matches the output impedance of the second RF generation unit 31b with the input impedance on the load side (substrate support unit 11 side). The second RF generation unit 31b supplies a bias RF signal to the conductive member of the substrate support unit 11 to draw ion components in the plasma into the substrate W.

[0019] The plasma processing apparatus 1 is equipped with a measurement unit 35 that measures either voltage or current on an electrode placed in the plasma processing chamber 10 or on wiring connected to the electrode. In this embodiment, the measurement unit 35 is provided on a conductive part 33b connected to a conductive member of the substrate support part 11. The measurement unit 35 is configured to include a probe that detects current and voltage, and measures voltage and current. The measurement unit 35 measures the voltage and current of the conductive part 33b through which the bias RF signal flows, and outputs signals indicating the measured voltage and current to a control unit 100, which will be described later.

[0020] The power supply 30 may also include a DC power supply 32 coupled to the plasma processing chamber 10. The DC power supply 32 includes a first DC generation unit 32a and a second DC generation unit 32b. In one embodiment, the first DC generation unit 32a is connected to a conductive member of the substrate support unit 11 and configured to generate a first DC signal. The generated first DC signal is applied to the conductive member of the substrate support unit 11. In one embodiment, the first DC signal may be applied to other electrodes, such as electrodes in an electrostatic chuck. In one embodiment, the second DC generation unit 32b is connected to a conductive member of the shower head 13 and configured to generate a second DC signal. The generated second DC signal is applied to the conductive member of the shower head 13. In various embodiments, the first and second DC signals may be pulsed. The first and second DC generation units 32a and 32b may be provided in addition to the RF power supply 31, and the first DC generation unit 32a may be provided in place of the second RF generation unit 31b.

[0021] The exhaust system 40 may be connected to, for example, a gas outlet 10e located at the bottom of the plasma processing chamber 10. The exhaust system 40 may include a pressure regulating valve and a vacuum pump. The pressure regulating valve regulates the pressure in the plasma processing space 10s. The vacuum pump may include a turbomolecular pump, a dry pump, or a combination thereof.

[0022] The plasma processing apparatus 1 configured as described above further includes a control unit 100, which will be described later. Figure 2 is a block diagram showing an example of the functional configuration of the control unit in this embodiment. The plasma processing apparatus 1 shown in Figure 1 is comprehensively controlled by the control unit 100.

[0023] The control unit 100 is, for example, a computer and controls each part of the plasma processing apparatus 1. The operation of the plasma processing apparatus 1 is comprehensively controlled by the control unit 100. The control unit 100 controls the plasma processing apparatus 1 to perform the various processes described in this disclosure. The control unit 100 is provided with an external interface 101, a process controller 102, a user interface 103, and a storage unit 104.

[0024] The external interface 101 is capable of communicating with various parts of the plasma processing apparatus 1 and inputs and outputs various types of data. For example, signals indicating voltage and current measured by the measurement unit 35 are input to the external interface 101.

[0025] The process controller 102 is equipped with a CPU (Central Processing Unit) and controls each part of the plasma processing apparatus 1.

[0026] The user interface 103 consists of a keyboard for the process manager to input commands to manage the plasma processing apparatus 1, and a display that visualizes and displays the operating status of the plasma processing apparatus 1.

[0027] The memory unit 104 stores control programs (software) and processing condition data, such as recipes, for implementing various processes performed by the plasma processing apparatus 1 under the control of the process controller 102. The control programs and recipes may be stored on computer-readable storage media (e.g., hard disks, optical discs such as DVDs, flexible disks, semiconductor memory, etc.). Furthermore, the control programs and recipes can be transmitted online from other devices, for example, via a dedicated line.

[0028] The process controller 102 has internal memory for storing programs and data, reads control programs stored in the storage unit 104, and executes the processing of the read control programs. The process controller 102 functions as various processing units when the control programs are running. For example, the process controller 102 has the functions of a plasma control unit 102a and a calculation unit 102b. In this embodiment, the case in which the process controller 102 has the functions of a plasma control unit 102a and a calculation unit 102b is described as an example. However, the functions of the plasma control unit 102a and the calculation unit 102b may be implemented in a distributed manner by multiple controllers.

[0029] The plasma control unit 102a controls the plasma processing. For example, the plasma control unit 102a controls the exhaust system 40 to evacuate the plasma processing chamber 10 to a predetermined vacuum level. The plasma control unit 102a controls the gas supply unit 20 to introduce processing gas from the gas supply unit 20 into the plasma processing space 10s. The plasma control unit 102a controls the power supply 30 and, in conjunction with the introduction of processing gas, supplies source RF signals and bias RF signals from the first RF generation unit 31a and the second RF generation unit 31b to generate plasma in the plasma processing chamber 10.

[0030] The plasma processing apparatus 1 according to this embodiment performs, for example, cycle etching. The plasma control unit 102a controls the RF power supply 31 and supplies high-frequency power from the RF power supply 31 in pulse form. The RF power supply 31 supplies at least one of the source RF signal and the bias RF signal in pulse form. For example, the plasma control unit 102a controls the RF power supply 31 and supplies the source RF signal and the bias RF signal in pulse form from the first RF generation unit 31a and the second RF generation unit 31b, respectively. The pulse frequency for turning the supply of the source RF signal and the bias RF signal on and off is, for example, 100 Hz to 10 kHz. Hereinafter, the source RF signal with a higher frequency will also be referred to as HF (High Frequency), and the bias RF signal with a lower frequency will also be referred to as LF (Low Frequency).

[0031] The calculation unit 102b calculates statistical values ​​from the voltage and current of the signal input from the measurement unit 35 to be used for monitoring and other purposes as an indicator of the process state. Examples of statistical values ​​used include the mean and variance. The calculated statistical values ​​can be used for endpoint detection, feedback control to the RF power supply 31, anomaly detection, etc. The calculation unit 102b includes an acquisition unit 102c, a splitting unit 102d, an extraction unit 102e, and an output control unit 102f.

[0032] The acquisition unit 102c acquires a first data set over a predetermined period based on the signal input from the measurement unit 35 via the external interface 101. In the following description, as the first data set, for example, a predetermined period of 20ms, a predetermined cycle of 100ms, and a data acquisition period of 20s are used, and a data set of 2000 points obtained by sampling the voltage pulse waveform of the signal input from the measurement unit 35 at a sampling frequency of 100kHz is used. Alternatively, a data set consisting of multiple data points, such as voltage and current, may be used as the first data set. The acquisition unit 102c outputs the acquired first data set to the splitting unit 102d.

[0033] Here, we will explain the pulse waveform of the voltage of the signal input from the measurement unit 35 using Figure 3. Figure 3 is a diagram showing an example of the bias voltage waveform when multiple RF signals are supplied. As shown in Figure 3, when the source RF signal and the bias RF signal are supplied as pulse waveforms having multiple levels, the voltage (Vpp) at the monitored substrate W, that is, the bias Vpp which is the voltage of the signal measured by the measurement unit 35, will also have multiple levels. In the example in Figure 3, the source RF signal rises to H level at timing 50, falls to M level at timing 52, and falls to L level at timing 54. The bias RF signal rises to H level at timing 51, falls to M level at timing 53, and falls to L level at timing 55. Since the bias Vpp is affected not only by the bias RF signal but also by the source RF signal, it rises to M1 level at timing 51, falls to M2 level just before timing 52, and then rises to H level at timing 52. Furthermore, the bias Vpp overshoots to the M3 level and falls at timing 53, then rises to the M2 level at timing 54. Subsequently, the bias Vpp falls to the L level at timing 55. Note that, as shown at timing 56, if neither the source RF signal nor the bias RF signal is supplied, the bias Vpp becomes the L level. In this embodiment, the energy levels for each level of a pulse waveform with multiple levels, such as the bias Vpp described above, are determined. In the following description, the values ​​calculated at each level, for example, the voltage values ​​at each level, are referred to as energy levels.

[0034] Returning to the explanation of Figure 2, when the division unit 102d receives the first data group from the acquisition unit 102c, it divides the first data group into groups according to the voltage (Vpp) value of each data. For example, the division unit 102d calculates the average value of the first data group and divides the first data group into multiple groups based on the calculated average value. The division unit 102d also calculates an average value for each divided group and further divides each group into multiple groups based on the calculated average value. The division unit 102d repeats this group division until it obtains more than the desired number of groups.

[0035] Here, we will explain the grouping of the first data group using Figures 4 to 6. Figure 4 is a diagram illustrating the general outline of the grouping in this embodiment. As shown in Figure 4, for example, the number of data points in the pulse waveform of bias Vpp, that is, the 2000 points sampled by the acquisition unit 102c, can be seen from the frequency distribution to have five levels: H, M1 to M3, and L. The number of levels can also be determined in advance based on the pulse waveforms of the supplied source RF signal and bias RF signal. When the splitting unit 102d divides the original data group (group) into two in one grouping, in order to cover five levels, 2 n We only need to divide the data into 3 groups (n=3), which means dividing it into 8 groups G0 to G7, so the grouping process will be performed n=3 times.

[0036] Figure 5 shows an example of grouping in this embodiment. As shown in Figure 5, the splitting unit 102d first calculates the mean (Ave.) of the first data group 57a. As shown in Figure 5, the splitting unit 102d may also calculate the variance (Var.) at this time, or detect the maximum (Max.) and minimum (Min.) values. At this time, the splitting unit 102d calculates the mean and variance by loop calculation, with N being the number of data points in the first data group 57a, Sum being the sum, and SumSq being the sum of squares. The splitting unit 102d uses the mean of the first data group 57a as a threshold to divide the first data group 57a into a group 57b that is above the mean and a group 57c that is below the mean.

[0037] Next, the splitting unit 102d calculates the mean (Ave.) of group 57b. The splitting unit 102d may also calculate the variance (Var.) in the same way as in the case of the first data group 57a. For group 57b, the splitting unit 102d calculates the mean and variance by loop calculation, with NH being the number of data points, SumH being the sum, and SumSqH being the sum of squares.

[0038] On the other hand, the splitting unit 102d calculates the number of data points NL, the sum SumL, and the sum of squares SumSqL for group 57c from the results of loop calculations in the first data group 57a and group 57b. In other words, the splitting unit 102d calculates the number of data points NL, the sum SumL, and the sum of squares SumSqL using an equation that subtracts the number of data points NH, the sum SumH, and the sum of squares SumSqH of group 57b from the number of data points N, the sum Sum, and the sum of squares SumSq of the first data group 57a, respectively. Based on the number of data points NL, the sum SumL, and the sum of squares SumSqL calculated by the equation, the splitting unit 102d calculates the mean (Ave.) of group 57c. The splitting unit 102d may also calculate the variance (Var.) in the same way as in the case of group 57b. These can be expressed as equations (1) to (5) below. Alternatively, the mean and variance of group 57c may be calculated using loop calculations, and the mean and variance of group 57b may be calculated using formulas. In this case, formulas (1) to (5) should be used with L and H swapped.

[0039] Number of data points NL = N - NH ... (1) Sum SumL = Sum - SumH ... (2) Sum of squares SumSqL = SumSq - SumSqH ... (3) Average value AveL = SumH / NH ···(4) Variance VarL =SumSqL / NL-AveL×AveL (5)

[0040] Next, the splitting unit 102d uses the mean value of group 57b as a threshold to divide group 57b into group 57d (above the mean) and group 57e (below the mean). The splitting unit 102d also uses the mean value of group 57c as a threshold to divide group 57c into group 57f (above the mean) and group 57g (below the mean). In other words, the first data set 57a is divided into four groups at this point. The splitting unit 102d calculates the mean and variance for each of groups 57d through 57g, similar to how it did for groups 57b and 57c. The calculation of the mean and variance is the same as for groups 57b and 57c described above, so the explanation is omitted.

[0041] Similarly, the splitting unit 102d divides the four groups 57d to 57g into eight groups G0 to G7. The splitting unit 102d calculates the mean and variance for groups G0 to G7 in the same way as for groups 57b to 57g. Note that the calculation of the mean and variance is the same as for groups 57d to 57g described above, so the explanation is omitted.

[0042] Figure 6 shows an example of grouping using the average value in this embodiment. Figure 6 shows the average value calculated at each stage up to the division into the four groups 57d to 57g shown in Figure 5, as a graph. Note that the pulse waveforms are different from those in Figures 3 and 4. Graph 58a shows the state where the first data group 57a is divided into two groups based on the average value. In graph 58a, the data group of pulse waveforms in the H-side region 59b is classified into group 57b, and the data group of pulse waveforms in the L-side region 59c is classified into group 57c.

[0043] Graph 58b shows the average values ​​of groups 57b and 57c, respectively, and divides groups 57b and 57c into two groups. In Graph 58b, the pulse waveform data set in region 59d on the HH side is classified into group 57d, and the pulse waveform data set in region 59e on the HL side is classified into group 57e. In addition, in Graph 58b, the pulse waveform data set in region 59f on the LH side is classified into group 57f, and the pulse waveform data set in region 59g on the LL side is classified into group 57g.

[0044] Graph 58c shows the mean values ​​of groups 57d to 57g in Graph 58b. Although not shown in the illustration, when the mean values ​​of groups 57d to 57g are used to divide the data in Graph 58c, the groups G0 to G7 (HHH to LLL) are obtained. Alternatively, the division unit 102d may not calculate the variance for the first data group 57a and groups 57b to 57g, but instead calculate the variance for groups G0 to G7 after the data has been divided into groups G0 to G7. The division unit 102d outputs each data group of the divided groups G0 to G7, along with the mean value and variance for each group, to the extraction unit 102e.

[0045] Returning to Figure 2, when the extraction unit 102e receives data sets from groups G0 to G7 from the splitting unit 102d, it extracts a second data set from groups G0 to G7 that is included in a valid group. For example, the extraction unit 102e determines whether each group is in a transient state based on the CV (Coefficient of Variation) value, which is derived from the variances and mean values ​​of each group G0 to G7, and the number of data points in each group G0 to G7. In other words, the extraction unit 102e, Groups G0 to G7 are extracted as valid groups if the number of data points in groups G0 to G7 is greater than a predetermined threshold for data points, and the CV value representing the rate of change in groups G0 to G7 is smaller than a predetermined threshold for CV value. The number of data points From a predetermined baseline of data points Few, and Groups G0 to G7 The CV value represents the rate of change. From the predetermined reference value of the CV value Larger groups are considered transitional groups and therefore invalid groups. and do not extract Furthermore, group G0, which contains the highest value, is considered invalid even if it has a small number of data points. Do not extract . The predetermined reference values ​​for the number of data points and the predetermined reference values ​​for the CV values ​​may be input from the user interface 103, or they may be stored in the control program (software) stored in the storage unit 104, or they may be stored as part of the recipe containing stored processing condition data, etc.

[0046] Furthermore, the extraction unit 102e compares the difference in means and variances of multiple adjacent groups from groups G0 to G7 to determine whether or not to aggregate them into the same group. If the extraction unit 102e determines that the groups should be aggregated into the same group, it aggregates the multiple adjacent groups into one group. In other words, the extraction unit 102e removes invalid groups from each data set of groups G0 to G7 and extracts each data set included in the valid group after aggregating adjacent groups as a second data set in which the group level classification is complete. The extraction unit 102e outputs the extracted second data set to the output control unit 102f.

[0047] Here, the extraction of valid groups will be explained using Figure 7. Figure 7 shows an example of the extraction of valid groups in this embodiment. In the example in Figure 7, groups G2, G5, and G6 are considered invalid groups among groups G0 to G7, as they are transitional groups. Also, groups G0 and G1 are aggregated at level Le0. Similarly, groups G3 and G4 are aggregated at level Le1. Group G7 is classified at level Le4 on its own because the adjacent group G6 is considered an invalid group. Here, level Le0 corresponds to the highest H level, and level Le4 corresponds to the lowest L level. Also, levels Le1 to Le3 correspond to the intermediate M1 to M3 levels. In the example in Figure 7, there were no groups classified at levels Le2 and Le3, but depending on the original pulse waveform, there may be groups classified at levels Le2 and Le3. Also, in the example in Figure 7, the levels were set to a maximum of 5 levels, from Le0 to Le4, but this is not limited to this, and the number of levels may be 4 or less, or 5 or more.

[0048] Returning to Figure 2, when the second data group extracted from the extraction unit 102e is input to the output control unit 102f, it outputs statistical values ​​for each group after level classification to the plasma control unit 102a and the user interface 103 based on the second data group. As statistical values ​​for each group, the output control unit 102f outputs, for example, the average value of the second data group for each group, as the level of the value of each data included in the first data group over a predetermined period. The output control unit 102f also outputs the ratio of the number of data points for each level as a duty cycle to the plasma control unit 102a and the user interface 103. Furthermore, the output control unit 102f may detect endpoints and anomalies based on these levels and duty cycles, and output the detection results to the plasma control unit 102a and the user interface 103.

[0049] Here, the output statistics will be explained using Figure 8. Figure 8 is a diagram showing an example of the level-specific energy levels calculated in this embodiment. As shown in Figure 8, the output control unit 102f outputs, for example, the H level of level Le0 including groups G0 and G1, the M1 level of level Le1 including groups G3 and G4, and the L level of level Le4 including group G7, for the pulse waveform of the voltage of the signal input from the measurement unit 35. The output control unit 102f also outputs values ​​such as 14% for the H level, 40% for the M1 level, and 38% for the L level as duty cycles for each level. Groups G2, G5, and G6 are treated as invalid groups, but as shown in Figure 8, when correlated with the pulse waveform, it can be seen that they represent the transient period.

[0050] The plasma control unit 102a controls the plasma processing based on the endpoint detection results obtained from statistical values ​​calculated by the calculation unit 102b. For example, when the calculation unit 102b detects a process endpoint, the plasma control unit 102a terminates the plasma processing.

[0051] This method uses the average, allowing for robust data extraction. However, if the number of points in each group is small (i.e., the pulse-on duty cycle is low), the values ​​may fluctuate. For example, in groups with only about 5 points per pulse, the average value may fluctuate due to subtle variations in the timing of pulse on and data acquisition. In such cases, increasing the sampling frequency to increase the number of points is the best solution, but this may not always be possible due to hardware limitations.

[0052] In such cases, an interpolation unit (not shown) can be provided between the acquisition unit 102c and the splitting unit 102d to interpolate the data. For example, adding a point consisting of the average value of two points as an intermediate point between data points (linear interpolation) can produce the same effect as doubling the sampling frequency. Also, adding four points at equal intervals using linear interpolation can produce the same effect as quintuple the sampling frequency. Note that the interpolation method is not limited to linear interpolation; polynomial interpolation, exponential interpolation, logarithmic interpolation, etc., may also be used.

[0053] Table 1 below shows an example of the magnitude of the mean variation when pulses with a pulse frequency of 400 Hz and a duty cycle of 5% are acquired using 100 kHz sampling. In Table 1, the magnitude of the mean variation is expressed by dividing the mean by four times the standard deviation. Without interpolation, one group per acquired pulse period can only have about 7 points. Therefore, the mean of the group without interpolation shows a variation of about 1.6%. On the other hand, in the group where points are added by interpolation, for example, if four or more points are added, the magnitude of the variation can be improved to less than one-third compared to the case without interpolation.

[0054] [Table 1]

[0055] [Data calculation method] Next, the data calculation process according to this embodiment will be described. Figure 9 is a flowchart showing an example of the data calculation process in this embodiment.

[0056] In the data calculation process according to this embodiment, the case in which plasma processing is performed on a substrate W placed on the central region 111a of the main body 111 of the substrate support part 11 (mounting table) in the plasma processing apparatus 1 will be explained as an example, but the operation of the plasma processing by the plasma control unit 102a will be omitted from the explanation.

[0057] The acquisition unit 102c of the calculation unit 102b acquires a first data set for a predetermined period based on the signal input from the measurement unit 35 (step S1). The acquisition unit 102c outputs the acquired first data set to the splitting unit 102d.

[0058] When the first data group is input to the splitting unit 102d from the acquisition unit 102c, the splitting unit 102d uses the average value to split the first data group into two parts. n The data is divided into groups (for example, n=3) (step S2). The division unit 102d divides the first data set into groups G0 to G7, for example. The division unit 102d calculates the mean and variance of each group G0 to G7. The division unit 102d outputs each of the divided data sets G0 to G7, along with the mean and variance of each group, to the extraction unit 102e.

[0059] When the extraction unit 102e receives the data sets for groups G0 to G7 from the splitting unit 102d, along with the mean and variance of each group, it performs the extraction process (step S3). The extraction process will now be explained using Figure 10. Figure 10 is a flowchart showing an example of the extraction process in this embodiment.

[0060] The extraction unit 102e initializes the variables i, j, and k to i=0, j=0, and k=m (step S31). Here, m is the number of divisions of the first data group, which is 2. nCorresponding to each individual, if n=3, then m=8. The extraction unit 102e classifies group Gi into level Lej (step S32). That is, it classifies group G0, which contains the highest value, into level Le0. The extraction unit 102e increments the variable i (step S33).

[0061] The extraction unit 102e determines whether group Gi is in a transitional period based on the CV value of group Gi and the number of data points in group Gi (step S34). If the extraction unit 102e determines that group Gi is in a transitional period (step S34: Yes), it increments the variable i (step S35) and returns to step S34.

[0062] On the other hand, if the extraction unit 102e determines that group Gi is not in a transitional period (step S34: No), it determines whether group Gi is a neighboring group or not (step S36). If the extraction unit 102e determines that group Gi is not a neighboring group (step S36: No), it increments the variable j (step S37) and generates level Lej (step S38). The extraction unit 102e classifies group Gi into the generated level Lej (step S39). On the other hand, if the extraction unit 102e determines that group Gi is a neighboring group (step S36: Yes), it proceeds to step S39 without generating level Lej and classifies group Gi into an existing level Lej.

[0063] The extraction unit 102e determines whether or not the variable i = k (step S40). If the variable i is not k (step S40: No), the extraction unit 102e proceeds to step S35. On the other hand, if the variable i is k (step S40: Yes), the extraction unit 102e outputs the extracted second data group to the output control unit 102f, terminates the extraction process, and returns to the original process.

[0064] Returning to the explanation of Figure 9, when the output control unit 102f receives the second data group extracted from the extraction unit 102e, it outputs statistical values ​​for each group after level classification, such as the energy levels and ratios for each level, to the plasma control unit 102a, etc., based on the second data group (step S4). After outputting the statistical values, the output control unit 102f determines whether or not to terminate the data calculation process (step S5). If the output control unit 102f determines not to terminate the data calculation process (step S5: No), it returns to step S1 and repeats the data calculation process for a predetermined period in the next cycle. On the other hand, if the output control unit 102f determines to terminate the data calculation process (step S5: Yes), it terminates the data calculation process. This makes it possible to determine the energy levels and ratios for each level of a pulse waveform having multiple levels.

[0065] [Example 1] Next, as Example 1, we will describe a case where the voltage (Vpp) on the substrate W to be processed is monitored by the measurement unit 35 during a 20-second plasma treatment. Figure 11 is a diagram showing an example of the results of a one-dimensional cluster analysis in Example 1 according to this embodiment. As shown in graph 60 of Figure 11, in Example 1, an RF signal obtained by pulse-modulating a 13MHz RF signal in four steps at 500Hz is used as the bias RF signal. Graph 60 represents the voltage (Vpp) of the bias RF signal measured by the measurement unit 35. The voltage (Vpp) applied to the substrate W to be processed is correlated with the acceleration voltage of the ions being processed and is an important indicator of process stability. In Example 1, a data group of 2000 points sampled at a sampling frequency of 100kHz every 100ms, which is a predetermined period, is used as the first data group, that is, the data group corresponding to the first 20ms (predetermined period) of 100ms. Note that the data sampling is performed without synchronization with the pulse waveform of the bias RF signal.

[0066] Table 61 in Figure 11 shows the results of a one-dimensional cluster analysis obtained by performing the data calculation process described above on the bias RF signal in Graph 60. Table 61 also includes statistical values ​​for each division up to the third division from the first data group. Table 61 has items such as "Division Count", "Group", "Ave.", "Var.", "Count", "Sum", "SumSq", "sigma", "CV", "Duty", and "Level".

[0067] "Number of divisions" indicates the number of times the division unit 102d performed group division. "Group" indicates the label of each group in the data set that was grouped by the division. "Ave." indicates the mean value of each group. "Var." indicates the variance of each group. "Count" indicates the number of data points belonging to each group. "Sum" indicates the sum of the data belonging to each group. "SumSq" indicates the sum of the squares of the data belonging to each group. "sigma" indicates the standard deviation σ in each group. "CV" indicates the CV value in each group. "Duty" indicates the ratio of each group to the first data set. "Level" indicates the level (H, M1, M2, L, etc.) for the second data set whose level division was completed by the extraction unit 102e, if it corresponds to each level, and indicates that it is transient if it is a transient period. Note that the shaded area in Table 61 corresponds to the value output as a statistical value in the output control unit 102f.

[0068] Figure 12 shows an example of the results when the one-dimensional cluster analysis of Example 1 is repeated a predetermined number of times. Figure 13 shows an example of the ratio of data points for each level in Example 1. Graph 62 shown in Figure 12 is obtained by repeating the data calculation process of Graph 60, which represents one predetermined period (20 ms), at predetermined cycles (100 ms) for more than 200 times corresponding to a 20-second plasma treatment, and plotting the average value of each group classified into each level as the level. As shown in Graph 62, the plotted levels are the H level, M1 level, M2 level, and L level, which are four levels. Alternatively, as shown in Figure 13, the duty cycle, which is the ratio of data points corresponding to each level in Graph 62, may be plotted and graphed. The output control unit 102f can detect endpoints and anomalies based on the changes in the plotted values ​​at each level. In this way, in Example 1, the levels and duty cycles for each level of the pulse waveform can be determined. Furthermore, by continuing the cluster analysis of Example 1 during the process, the changes in the levels of each level at each step of the process can be determined. Furthermore, the duty cycle can be calculated from the number of data points at each level.

[0069] [Example 2] Next, as Example 2, we will describe a case in which, in addition to the conditions of Example 1, a 1 MHz RF signal is supplied as a bias RF signal. In Example 2, the measurement unit 35 monitors the voltage V, current I, and phase difference P, and the plasma emission intensity is monitored using a high-speed photodiode or the like in an observation window (not shown) provided in the plasma processing chamber 10. Note that the voltage V corresponds to the voltage (Vpp) in Example 1.

[0070] Figure 14 shows an example of the relationship between the dataset and grouping in Example 2. As shown in Figure 14, in Example 2, the calculation unit 102b treats the voltage V, current I, and phase difference P of the 13MHz bias RF signal, which are input parameters for channels Ch.0 to 2, as a dataset DS1, and divides the dataset DS1 into multiple groups based on the value of the voltage V of the 13MHz bias RF signal. At this time, the current I and phase difference P are divided into multiple groups together with the corresponding voltage V. In other words, dataset DS1 is the first data group of a dataset that includes multiple types of related data. Furthermore, the calculation unit 102b divides the dataset DS1 into multiple groups based on a specific type of data (voltage V).

[0071] Similarly, the calculation unit 102b treats the voltage V, current I, and phase difference P of the bias RF signal, which are the 1MHz input parameters for channels Ch.3 to 5, as a data set DS2, and divides the data set DS2 into multiple groups based on the value of the voltage V of the 1MHz bias RF signal. At this time, the current I and phase difference P are divided into multiple groups together with the corresponding voltage V. In other words, data set DS2 is the first data group of a data set containing multiple related types of data. Furthermore, the calculation unit 102b divides the data set DS2 into multiple groups based on a specific type of data (voltage V). In addition, for the emission intensity, which is the input parameter for channel Ch.6, the calculation unit 102b groups the data group D3 alone as the first data group. In other words, the calculation unit 102b divides the first data group into multiple groups for each type of input parameter.

[0072] The calculation unit 102b calculates, for example, impedance Z as a parameter based on the second data group into which the datasets DS1 and DS2 are grouped, and outputs the calculated impedance Z as a statistical value. In other words, the calculation unit 102b calculates impedance Z based on the simultaneously acquired voltage V, current I, and phase difference P. To put it another way, the calculation unit 102b levels the second input parameters, current I and phase difference P, using the first input parameter, and then calculates the third parameter, impedance Z, using the first and second parameters. Furthermore, the calculation unit 102b calculates, for example, the emission intensity level as a parameter based on the second data group into which the data group D3 is grouped. The calculation unit 102b outputs statistical data that allows for the creation of graphs corresponding to graph 62 in Figure 2, for example, graphs representing the voltage V, current I, and impedance Z levels corresponding to a 13MHz bias RF signal, graphs representing the voltage V, current I, and impedance Z levels corresponding to a 1MHz bias RF signal, and a graph representing the emission intensity level. In this way, in Example 2, cluster analysis can be easily performed even for data of two or more dimensions by combining it with one-dimensional cluster analysis.

[0073] [Example 3] Next, as Example 3, we will describe a case in which two types of grouping are performed based on the luminescence intensity, using the same configuration as in Example 2.

[0074] Figure 15 shows an example of the relationship between the dataset and grouping in Example 3. As shown in Figure 15, in Example 3, the calculation unit 102b divides the light emission intensity, which is an input parameter for channel Ch.6, into multiple groups as datasets DS1-3 based on the voltage V of the 13MHz bias RF signal. Similarly, the calculation unit 102b divides the light emission intensity into multiple groups as datasets DS2-3 based on the voltage V of the 1MHz bias RF signal. In other words, the first data group is a data group corresponding to three types of input parameters: 13MHz, 1MHz, and light emission intensity. Furthermore, the calculation unit 102b divides the data group corresponding to one type of input parameter (light emission intensity) from the first data group into multiple groups for each of the other two types of input parameters (13MHz, 1MHz).

[0075] The calculation unit 102b calculates the parameters based on the second data group, in which the datasets DS1, DS2, DS1-3, and DS2-3 are grouped, and outputs the calculated parameters as statistical values.

[0076] Figure 16 shows an example of the calculated level-specific energy levels in Example 3. In Figure 16, the energy levels of voltage V in dataset DS1, the energy levels of voltage V in dataset DS2, and the energy levels of emission intensity and emission intensity based on datasets DS1-3 and DS2-3 are shown. In the emission intensity graph in Figure 16, DS1-H represents the H level based on dataset DS1-3, DS1-M1 represents the M1 level based on dataset DS1-3, and DS2-H represents the H level based on dataset DS2-3. In other words, the calculation unit 102b outputs statistical data that allows for the creation of graphs representing the levels of voltage V, current I, and impedance Z corresponding to a 13MHz bias RF signal, and graphs representing the levels of voltage V, current I, and impedance Z corresponding to a 1MHz bias RF signal, as output corresponding to graph 62 in Figure 2. Furthermore, the calculation unit 102b outputs statistical data that allows for the creation of, for example, a graph representing the emission intensity levels corresponding to the voltage V levels of a 13 MHz bias RF signal, and a graph representing the emission intensity levels corresponding to the voltage V levels of a 1 MHz bias RF signal, as output corresponding to graph 62 in Figure 2. In this way, in Example 3, similar to Example 2, cluster analysis can be easily performed on data of two or more dimensions by combining it with one-dimensional cluster analysis.

[0077] As described above, according to this embodiment, the calculation unit 102b acquires a first data group over a predetermined period, divides the acquired first data group into multiple groups according to the range of values ​​of each data included in the first data group, extracts a second data group that is included in a valid group from among the divided multiple groups, and outputs statistical values ​​for each group based on the extracted second data group. As a result, the levels and ratios for each level of a pulse waveform having multiple levels can be determined.

[0078] Furthermore, according to this embodiment, the calculation unit 102b, after acquiring the first data group, calculates data to fill in the gaps between adjacent data points in the acquired first data group, and adds the calculated data to the first data group. As a result, the variability of the group's average value can be suppressed.

[0079] Furthermore, according to this embodiment, the calculation unit 102b calculates the average value of the first data group and divides the first data group into multiple groups based on the calculated average value. As a result, the levels and ratios for each level of a pulse waveform having multiple levels can be determined.

[0080] Furthermore, according to this embodiment, the calculation unit 102b calculates an average value for each group and further divides the group into multiple groups based on the calculated average value. As a result, the levels and ratios for each level of a pulse waveform having multiple levels can be determined.

[0081] Furthermore, according to this embodiment, the calculation unit 102b divides the first data group or group into two groups, calculates the average value of one group by loop calculation, and calculates the average value of the other group based on the average value before division and the average value calculated by the other group. As a result, the computational load can be reduced.

[0082] Furthermore, according to this embodiment, the calculation unit 102b calculates the variance for each of the divided groups, and based on the calculated variance for each group, the CV value based on the mean value for each group, and the number of data points for each group, groups that are determined to be in a transient state are not extracted as invalid groups. As a result, false detection of levels can be suppressed.

[0083] Furthermore, according to this embodiment, the calculation unit 102b compares the difference in mean values ​​and variances among multiple adjacent groups, and if it determines that they should be grouped together, it combines the multiple adjacent groups into a single group. As a result, groups belonging to the same level can be grouped together.

[0084] Furthermore, according to this embodiment, the calculation unit 102b outputs the average value of the second data group for each group as the level of the value of each data included in the first data group over a predetermined period. As a result, the level values ​​at each level can be determined.

[0085] Furthermore, according to this embodiment, the calculation unit 102b outputs the ratio of data points for each level. As a result, the ratio (duty cycle) for each level in the first data group can be determined.

[0086] Furthermore, according to this embodiment, the calculation unit 102b calculates 2 when the number of levels is n. n The data is divided into groups of one or more. As a result, all levels in the first data set can be determined without omission. The number of levels, n, may be input from the user interface 103, or it may be one of the control programs (software) stored in the memory unit 104, or one of the stored processing condition data, etc., that are stored as part of the recipe. Alternatively, the number of levels, n, may be automatically calculated from the processing condition data, based on the timing of the bias RF signal pulses and the source RF signal.

[0087] Furthermore, according to this embodiment, the calculation unit 102b acquires a first data set over a predetermined period that is repeated at a predetermined cycle. As a result, the levels and ratios of each level of a pulse waveform having multiple levels can be continuously determined.

[0088] Furthermore, according to this embodiment, the first data group is a dataset containing multiple related types of data. The calculation unit 102b divides the dataset into multiple groups based on a specific type of data, calculates parameters for each dataset based on the second data group corresponding to the dataset, and outputs the calculated parameters for each group as statistical values. As a result, cluster analysis can be easily performed on data of two or more dimensions by combining it with one-dimensional cluster analysis.

[0089] Furthermore, according to this embodiment, the relevant multiple types of data are the voltage, current, and phase difference of the high-frequency power supplied to the electrodes in the plasma processing vessel. As a result, the impedance of the plasma in the substrate W during processing can be calculated.

[0090] Furthermore, according to this embodiment, the first data group is a data group corresponding to multiple types of input parameters. The calculation unit 102b divides the first data group into multiple groups according to the type of input parameter. As a result, the levels and ratios for each level can be determined for each type of input parameter.

[0091] Furthermore, according to this embodiment, the first data group is a data group corresponding to three types of input parameters. The calculation unit 102b divides the data group corresponding to one type of input parameter from the first data group into multiple groups for each of the other two types of input parameters. As a result, it is possible to determine the levels and ratios for each level according to the relationship between the input parameters.

[0092] Furthermore, according to this embodiment, the input parameters are the frequency of the high-frequency power supplied to the electrodes in the plasma processing vessel, and the emission intensity of the plasma in the plasma processing vessel. As a result, the levels and ratios of emission intensity for each frequency of high-frequency power can be determined.

[0093] The embodiments disclosed herein should be considered in all respects to be illustrative and not restrictive. The above embodiments may be omitted, replaced, or modified in various forms without departing from the scope and spirit of the appended claims.

[0094] Furthermore, the above-described embodiment explained an example in which the measurement unit 35 is provided on the conductive part 33b connected to the substrate support part 11. However, it is not limited to this. The measurement unit 35 can be provided on an electrode placed inside the plasma processing chamber 10 or on wiring connected to the electrode in order to measure the state of the plasma inside the plasma processing chamber 10. For example, the measurement unit 35 may be provided on the conductive part 33a connected to the conductive member of the shower head 13. Alternatively, a measuring electrode may be placed inside the plasma processing chamber 10, and the measurement unit 35 may be provided on the electrode or on wiring connected to the electrode. In this embodiment, the measurement unit 35 is provided on the substrate support part 11 side of the impedance matching circuit 34b of the conductive part 33b. This allows the measurement unit 35 to measure the state of the plasma inside the plasma processing chamber 10.

[0095] Furthermore, although the above-described embodiment uses a capacitively coupled plasma processing apparatus 1 that performs etching and other processing on a substrate W using a capacitively coupled plasma as the plasma source as an example, the disclosed technology is not limited to this. Any plasma source can be used as long as it is a device that performs processing on a substrate W using plasma; for example, inductively coupled plasma, microwave plasma, magnetron plasma, etc.

[0096] Furthermore, in the embodiment described above, a data set of 2000 points sampled from the pulse waveform of the voltage of the signal input from the measurement unit 35 is used as the first data set, but the invention is not limited to this. For example, the processing results (dimensions, etching depth, etc.) of substrates etched by the plasma processing apparatus 1 may be sampled for each of the multiple substrates. By dividing the acquired first data set into multiple groups, extracting a second data set, and outputting statistical values ​​for each group based on the extracted second data set, the variability in the processing results can be divided into groups, and the cause of the variability can be investigated based on the output statistical values ​​for each group.

[0097] Furthermore, although the above-described embodiment uses a data calculation method executed by a computer included in the control unit 100, it is not limited to this. For example, an FPGA (Field Programmable Gate Array) or a specially designed gate array programmed specifically for data calculation may be used. Alternatively, the acquired data set may be processed using spreadsheet software such as Microsoft Excel®, or the calculations may be performed manually.

[0098] It should be noted that the components of each illustrated device do not necessarily have to be physically configured as shown. In other words, the specific forms of distribution and integration of each device are not limited to those shown, and all or part of them can be functionally or physically distributed and integrated in any unit according to various loads and usage conditions.

[0099] Furthermore, the various processing functions performed by each device may be executed in whole or in part on a CPU (or a microcomputer such as an MPU (Micro Processing Unit) or MCU (Micro Controller Unit)). It goes without saying that the various processing functions may also be executed in whole or in part on a program analyzed and executed by the CPU (or a microcomputer such as an MPU or MCU), or on wired logic hardware. [Explanation of Symbols]

[0100] 1. Plasma processing equipment 10 Plasma processing chamber 11. Substrate support section 13 Shower head 20 Gas Supply Department 31 RF power supply 31a First RF generation unit 31b Second RF generation unit 33a,33b Conductive part 34a, 34b Impedance matching circuit 35 Measurement section 40 Exhaust System 100 Control Unit 101 External Interface 102 Process Controller 102a Plasma control unit 102b Calculation Unit 102c Acquisition Department 102d Divided section 102e Extraction part 102f Output Control Unit 103 User Interface 104 Storage section W board

Claims

1. To acquire the first set of data over a predetermined period, The acquired first data set is divided into multiple groups according to the range of each data value included in the first data set, Extracting a second set of data that is included in a valid group from among the multiple groups that have been divided, Based on the extracted second data set, statistical values ​​for each group are output, It has, The first data set is a data set of measurement signals corresponding to pulse waveforms having multiple levels, which are measured when processing a substrate using a substrate processing apparatus. Data calculation method.

2. Furthermore, after acquiring the first data set, data is calculated to fill in the gaps between adjacent data points in the acquired first data set, and the calculated data is added to the first data set. A data calculation method according to claim 1, comprising:

3. The division process involves calculating the average value of the first data group and dividing the first data group into a plurality of groups based on the calculated average value. The data calculation method according to claim 1 or 2.

4. The division process involves calculating an average value for each group and, based on the calculated average value, further dividing the group into multiple groups. The data calculation method according to claim 3.

5. The division process involves dividing the first data group or group into two groups, calculating the average value of one group by loop calculation, and calculating the average value of the other group based on the average value before division and the average value calculated by the first group. The data calculation method according to claim 3 or 4.

6. The aforementioned division process involves calculating the variance for each of the multiple groups that have been divided, The extraction process involves determining the CV value based on the calculated variance and mean value for each group, and the number of data points for each group. Groups whose CV value is greater than a predetermined threshold and whose number of data points is less than a predetermined threshold are not extracted as invalid groups. A data calculation method according to any one of claims 3 to 5.

7. The extraction process involves extracting the group containing the highest value from among the groups as a valid group. The data calculation method according to claim 6.

8. The extraction process involves comparing the difference in the mean values ​​and the variances of multiple adjacent groups, and if it determines that they should be grouped together, it aggregates the multiple adjacent groups into a single group. The data calculation method according to claim 6 or 7.

9. The output process outputs the average value of the second data group for each group as the level of the value of each data included in the first data group during the predetermined period. A data calculation method according to any one of claims 1 to 8.

10. The output process outputs the ratio of data points for each level. The data calculation method according to claim 9.

11. The above division process is performed when the number of levels is n, and 2 ｎ Divide into groups of one or more. The data calculation method according to claim 9 or 10.

12. The acquisition process involves acquiring the first data set during the predetermined period, which is repeated at predetermined intervals. A data calculation method according to any one of claims 1 to 11.

13. The first data set is a data set containing multiple related types of data, The aforementioned splitting process divides the dataset into a plurality of groups based on a specific type of data, The output process involves calculating parameters for each dataset based on the second data group corresponding to the dataset, and outputting the calculated parameters for each group as statistical values. A data calculation method according to any one of claims 1 to 12.

14. The aforementioned related data of multiple types are the voltage, current, and phase difference of the high-frequency power supplied to the electrodes in the plasma processing vessel. The data calculation method according to claim 13.

15. The first data set is a data set corresponding to multiple types of input parameters, The division process involves dividing the first data set into multiple groups according to the type of input parameter. A data calculation method according to any one of claims 1 to 14.

16. The first data set mentioned above is a data set corresponding to three types of input parameters, The division process involves dividing the data group corresponding to one type of input parameter from the first data group into multiple groups, according to the other two types of input parameters. A data calculation method according to any one of claims 1 to 15.

17. The input parameters are the frequency of the high-frequency power supplied to the electrodes in the plasma processing vessel, and the emission intensity of the plasma in the plasma processing vessel. The data calculation method according to claim 15 or 16.

18. A substrate processing apparatus, A measurement unit that measures data related to processing on the substrate, It has a control unit and The control unit is configured to control the substrate processing apparatus to acquire a first data set from the measurement unit over a predetermined period of time. The control unit is configured to control the substrate processing apparatus to divide the acquired first data group into a plurality of groups according to the range of values ​​of each data included in the first data group. The control unit is configured to control the substrate processing apparatus to extract a second data set that is included in a valid group from among the divided plurality of groups. The control unit is configured to control the substrate processing apparatus to output statistical values ​​for each group based on the extracted second data set. The first data set is a data set of measurement signals corresponding to pulse waveforms having multiple levels, which are measured when the substrate processing apparatus processes the substrate. Circuit board processing equipment.

Images