JP6453421B2 - Data processing method, data processing apparatus, and processing apparatus - Google Patents

Description

  The present invention relates to a data processing method and data processing apparatus for a series of numerical data of a target system (apparatus, analysis data, etc.).

  Various models have been proposed as techniques for performing data smoothing and data prediction of a series of numerical data in a target system. For the purpose of grasping the state of the target system, in order to detect the extreme value (maximum value, minimum value) or inflection point that is the data change point, the first derivative, second derivative, etc. of the data Is often used. In particular, in the case of time-series data such as measurement data including noise, etc., data smoothing processing and differentiation processing are performed with high accuracy, the change point in the target system is detected, and control of the target system is performed. To do is an important technical issue.

  Conventional data smoothing and data prediction techniques include a curve fitting method and a moving average method as described in Non-Patent Document 1. As described in Patent Document 1, the curve fitting method includes a polynomial fitting method (Savitzky-Golay method), and the digital filter includes a Butterworth low-pass filter. In addition, as described in Non-Patent Document 2, the moving average method includes an exponential smoothing method. Non-Patent Document 2 discloses a one-stage exponential smoothing method (one smoothing parameter), but a two-stage smoothing method (two smoothing parameters) is also used for economy such as demand and supply forecasting. Used in related fields.

  As described in Non-Patent Document 3, a difference method is frequently used as a conventional technique for performing data differentiation. Also, as described in Patent Document 1, a polynomial fitting method (Savitzky-Golay method) may be used. Moreover, as an example of controlling time-series data such as measurement data including noise etc., data smoothing processing, detecting data change points by primary differentiation processing and secondary differentiation processing, and controlling a target system Patent Document 2 discloses a method in which in a plasma etching processing apparatus, spectral intensity signal data from plasma emission is subjected to data smoothing processing by moving average processing, and an etching end point is determined by a primary differential value and a secondary differential value. Has been.

JP 2000-228397 A JP-A-61-53728

`` Backside of data processing '' by Kazuhiro Takahashi, Journal of Surface Analysis Vol7.No 1 (2000) p68-p77. "Introduction to Time Series Models" by A.C. Harvey, Naoto Kunito / Takuji Yamamoto, (1985), p173, The University of Tokyo Press, (Original: TIME SERIES MODELS by A.C.Harvey) `` Fluid Computation and Difference Method '' by Kunio Kuwahara, edited by Tetsuya Kawamura, (2005), p1, Asakura Shoten.

  As described in Patent Document 1, when the data differentiation process is performed by the difference method, if the data smoothing process is not sufficient, the output result of the primary differentiation process includes a lot of noise and is smooth (smooth) data. In addition, the S / N ratio (signal / noise ratio) is reduced. When data differentiation is performed again by the difference method using the above data, the output result of the secondary differentiation process is further increased in noise, not smooth data, and the S / N ratio is further reduced. There is.

  When the difference interval is increased, the output result of the first derivative process and the output result of the second derivative process become smooth data and the S / N ratio increases (however, there is an optimum value for the difference interval) Particularly in the case of time-series data, there is a problem that the data delay time due to the data differentiation process increases. For low-pass filters, etc., if the cut-off frequency is decreased, the output data of the primary differentiation process and the output data of the secondary differentiation process become smooth data, and the S / N ratio increases (however, the cut-off) As described above, there is a problem that the data delay time due to the data differentiation process increases.

  In addition, when data differentiation processing is performed by a polynomial fitting method (Savitzky-Golay method), it is common to require a plurality of data and derive a differential value in time of the central data. For this reason, in the case of sequential data processing, there is a problem that, in principle, a time delay occurs at least by the time difference from the latest time data to the central data.

  In general, in the data smoothing process, an error is likely to occur for a certain period immediately after the start of data processing. For example, in the case of time-series data, since the data period is short and the sampling time interval is large, in the case where the number of data is small, the ratio of the period with a large error is large with respect to the entire data period. In addition, there is a problem that the reliability is poor even if the data differentiation process is performed in a period in which the error is large.

  Furthermore, there is a method disclosed in Patent Document 2 as a method for determining the end point of plasma etching in a plasma etching processing apparatus. The following two methods are mainly used for the method for determining the end point of plasma etching disclosed in Patent Document 2. There are three issues.

  Current semiconductor devices have a high step structure such as a Fin Field Effect Transistor (Fin FET) structure due to high performance and high integration. Further, in normal plasma etching, microloading, which is a difference in etching performance between a sparse pattern and a dense pattern, occurs. Furthermore, the film to be etched may be uneven within the wafer surface.

  For these reasons, for example, the time-series data of spectral intensity signal data from plasma emission used for determining the end point of plasma etching may change in two stages. Thus, when the time-series data of spectral intensity signal data from plasma emission changes in two stages and the end point of plasma etching is determined by the second change, the first change and the second change occur in a short period of time. Therefore, the end point of plasma etching cannot be detected because the data processing cannot follow the second change. Here, the time point when the time-series data of the spectral intensity signal data from the plasma emission changes is the end point of plasma etching.

  Further, for example, in the case of determining the end point of the plasma etching by the change in the time series data of the spectral intensity signal data from the plasma emission even in the plasma etching for a short time with respect to the delay time of the primary differential value or the secondary differential value Data processing for calculating the primary differential value or the secondary differential value cannot follow the change in the time series data of the spectral intensity signal data from the plasma emission, which is the end point of the plasma etching. That is, there is a first problem that the responsiveness in detecting the end point of plasma etching is insufficient.

  Next, the plasma etching mask pattern is roughly classified into a groove pattern and a hole pattern. In addition, the aperture ratio of the hole pattern wafer is usually smaller than the aperture ratio of the groove pattern wafer and may be less than 1%. Further, the spectral intensity of plasma emission decreases as the aperture ratio decreases. For this reason, for example, in the case of a wafer having an aperture ratio of less than 1%, it is difficult to detect the end point of plasma etching because the change in time series data of spectral intensity signal data from plasma emission is too small. That is, there is a second problem that it cannot cope with a low S / N ratio. Here, the aperture ratio is the ratio of the area to be etched to the area of the entire wafer.

  From the above, when the target system change point is detected by primary differential data or secondary differential data and the target system is controlled, the control accuracy is not sufficient due to low S / N ratio, time delay, etc. There is a problem.

  Therefore, the present invention provides a data processing method and data processing apparatus that can achieve both a high S / N ratio and a reduced data delay in a data processing method and data processing apparatus for processing data.

The present invention provides a processing chamber to be controlled, a measuring device that acquires data related to the processing chamber, and smoothing based on an error between the data acquired by the measuring device and the predicted value of the smoothed data. Arithmetic processing apparatus using a response type two-step exponential smoothing method for changing the optimization parameter and performing data processing based on a data processing program for processing the data with the lower limit value of the change range of the smoothing parameter larger than 0 And at least one of the data smoothing processing result obtained by the arithmetic processing device, the primary differentiation processing result obtained by the arithmetic processing device, and the secondary differentiation processing result obtained by the arithmetic processing device. A control device for controlling the processing chamber based on a change in the state of the processing chamber detected based on the result or a change in the state of the detected processing chamber. When N is 0 or a natural number, the response coefficient of the smoothing parameter is an N-th power response of the absolute value obtained by dividing the relative error by the absolute error, and the relative error is acquired by the measuring device. The absolute error is an absolute error between the data acquired by the measuring device and the predicted value of the smoothed data. Is a processing device .

  Further, the present invention uses a processing chamber to be controlled, a measuring device for acquiring data related to the processing chamber, and a two-stage exponential smoothing method, and smoothes the data related to the processing chamber as an input value. A first predicted value that is a predicted value of the generated data and a second predicted value that is a predicted value of the slope of the smoothed data as an output value, and the output value using a two-stage exponential smoothing method As the output value, the third predicted value, which is the predicted value of the smoothed data, and the fourth predicted value, which is the predicted value of the slope of the smoothed data, are input as the second predicted value obtained as An arithmetic processing device that performs data processing based on a data processing program having a step of obtaining, a first predicted value obtained by the arithmetic processing device, a second predicted value obtained by the arithmetic processing device, According to the arithmetic processing unit State of the processing chamber detected based on at least one prediction value among the third prediction value obtained and the fourth prediction value obtained by the arithmetic processing unit or the state of the processing chamber detected And a control device for controlling the processing chamber according to the change of the processing chamber.

  Further, the present invention is based on a process chamber to be controlled, a measuring device that acquires data related to the processing chamber, and an error between the data acquired by the measuring device and the predicted value of the smoothed data. An operation that uses a response type two-step exponential smoothing method that changes the smoothing parameter and performs data processing based on a data processing program that processes data by setting the lower limit value of the changing range of the smoothing parameter to be greater than zero. At least one of a processing device, a data smoothing processing result obtained by the arithmetic processing device, a primary differential processing result obtained by the arithmetic processing device, and a secondary differential processing result obtained by the arithmetic processing device. Control device for controlling the processing chamber according to a change in the state of the processing chamber detected based on one processing result or the detected state of the processing chamber A processing unit, characterized in that it comprises a.

  In addition, the present invention provides a processing chamber in which a sample placed on a sample stage is subjected to plasma processing, a measuring device that acquires light emission data during plasma processing of the sample, and data smoothed by the measuring device. For processing data by using a response type two-step exponential smoothing method that changes a smoothing parameter based on an error from a predicted value of the data that has been processed, and by setting the lower limit value of the range in which the smoothing parameter is changed to greater than zero An arithmetic processing unit that performs data processing based on a processing program, a data smoothing processing result obtained by the arithmetic processing unit, a first-order differential processing result obtained by the arithmetic processing unit, and an arithmetic processing unit Among the secondary differential processing results, the plasma processing state of the sample detected based on at least one processing result or the detected sample A processing unit, characterized in that it comprises a control device for controlling the processing chamber in response to changes in the plasma processing conditions.

  Further, the present invention provides a data processing method for processing data using a two-stage exponential smoothing method, wherein the first value is a predicted value of data smoothed using the two-stage exponential smoothing method and using the data as an input value. The second predicted value, which is a predicted value of the slope of the smoothed data and the predicted value of the smoothed data, is obtained as an output value, and the second predicted value is smoothed using the second predicted value smoothing method as an input value. And a third predicted value that is a predicted value of the data and a fourth predicted value that is a predicted value of the slope of the smoothed data are obtained as output values.

  According to the present invention, in a data processing method for processing data using a two-stage exponential smoothing method, a step of obtaining a polynomial approximation of an arbitrary number of data after the start of data input, and data using the polynomial approximation Obtaining a first predicted value that is a predicted value of smoothed data before the start of input and a second predicted value that is a predicted value of a slope of smoothed data before the start of data input; And a step of performing a two-stage exponential smoothing process on the data using the predicted value and the second predicted value as initial values, respectively.

  Further, the present invention provides a data processing method for processing data using a response type two-stage exponential smoothing method that changes a smoothing parameter according to an error, and detects an error between input data and a predicted value of the smoothed data. Accordingly, a predicted value of the smoothed data is obtained while changing the first smoothing parameter so that the first smoothing parameter does not become 0, and the smoothed data is smoothed with the slope of the predicted value of the smoothed data. Obtaining a predicted value of the slope of the smoothed data while changing the second smoothing parameter so that the second smoothing parameter does not become zero according to an error from the predicted value of the slope of the data. Features.

  In the data processing method of processing data using the exponential smoothing method according to the present invention, when the data is processed by an N-stage exponential smoothing process and the N is a natural number of 3 or more, the N-stage exponential smoothing is performed. The process obtains the predicted value of the smoothed data and sequentially obtains the predicted value of the first slope in the smoothed data to the predicted value of the N-1th slope in the smoothed data. It is characterized by that.

  Furthermore, the present invention provides a plasma processing method for plasma processing a sample, measuring light emission data during plasma processing of the sample, and based on an error between the measured light emission data and a predicted value of the smoothed light emission data. The response type two-step exponential smoothing method for changing the smoothing parameter is used, and the emission data is processed by setting the lower limit value of the range for changing the smoothing parameter to be greater than 0, and obtained by the response type two-step exponential smoothing method. At least one of the obtained data smoothing processing result, the first-order differential processing result obtained by the response-type two-step exponential smoothing method, and the second-order differentiation processing result obtained by the response-type two-step exponential smoothing method Control the plasma processing of the sample according to a change in the plasma processing state of the sample detected based on the result or the detected plasma processing state of the sample And wherein the Rukoto.

  According to the present invention, a data processing method and a data processing apparatus for processing data can achieve both a high S / N ratio and a reduction in data delay.

1 is a diagram illustrating an overall configuration of a data processing apparatus 1 according to a first embodiment. FIG. 6 is a diagram illustrating a flow of data processing according to the first embodiment. FIG. 5 is a diagram illustrating a flow for deriving an initial value in the first embodiment. 1 is a longitudinal sectional view of a magnetic field microwave plasma etching apparatus according to Example 1. FIG. It is a flowchart of the typical data processing for an etching end point detection. The data processing result by the data processor of Example 1 is shown. It is the data processing result of the data smoothing by the conventional low-pass filter and the difference method, and a data differentiation process. It is a figure in Example 2 which shows the relationship between the response coefficient and the absolute value of relative error / absolute error when the power N is used as a parameter. In the data processing apparatus of Example 2, it is a figure which shows the data processing result in case the power parameter N = 1. In the data processing apparatus of Example 2, it is a figure which shows the data processing result in case the power parameter N = 5. FIG. 10 is a diagram illustrating a data processing result in a case where there is a data change point immediately after the start of data input when the power parameter N = 1 of the data processing apparatus according to the second embodiment. FIG. 10 is a diagram illustrating a flow of data processing according to a third embodiment.

  Embodiments of the present invention will be described below with reference to the drawings.

  First, a data processing apparatus according to a first embodiment of the present invention will be described with reference to FIGS. Here, an example in which the present invention is applied to etching end point detection by plasma spectroscopy for the purpose of highly accurate etching processing in a magnetic field microwave plasma etching apparatus will be described.

  FIG. 1 shows a configuration diagram of a data processing apparatus 1 according to the first embodiment. In this embodiment, the data processing device 1 is composed of a data input / output device 2, a data storage device 3, which is a storage device, a data processing program storage device 4, and a data arithmetic processing device 5, and moves data between them. Are connected as possible. In addition to the above, a data display device (not shown) is provided as necessary. The data processing apparatus 1 can input / output data to / from the target system 6 (apparatus, analysis data, etc.). Thereby, the target system 6 is controlled with high accuracy. In this embodiment, the target system 6 is a microwave plasma processing apparatus. Further, the data processing apparatus 1 may be used alone, and can be used for data analysis and the like.

  The data input / output device 2 can input / output process data, parameters of a data processing program, and the like. The data input / output device 2 receives data to be processed collectively or sequentially from the target system 6 or the like, stores it in the data storage device 3 such as RAM, and stores the data stored in the data processing program storage device 4 such as RAM. According to the processing program, the data arithmetic processing device 5 performs data smoothing processing and data differentiation processing, and the data input / output device 2 outputs the data smoothing processing result data and data differentiation processing result data to the target system 6 or the like. This is used to control the target system 6.

FIG. 2 shows an overall flowchart of the data processing method stored in the data processing program. Data to be processed in batch or sequentially is received and data is input. In this embodiment, the input data is time-series data Y1 _t : t = 1, 2,. Next, an initial value is derived by a method described later. Next, the first two-stage exponential smoothing process is performed by the following formulas (1) and (2), and the predicted output value S1 _t of the smoothed data and the predicted value B1 of the smoothed data slope are performed. get _t .

Data smoothing: S1 _t = α1 Y _t + (1-α1) (S1 _t-1 + B1 _t-1 ) (1)
Slope of smoothed data: B1 _t = γ1 (S1 _t -S1 _t-1 ) + (1-γ1) B1 _t-1 (2)
Next, the predicted value B1 _t of the smoothed data of the first output is set as the second input data Y2 _t , and the second two-stage exponential smoothing process is performed to smooth the data of the second output. A predicted value S2 _t and a predicted value B2 _t of the slope of the smoothed data are obtained. Next, data smoothing processing result data S1 _t , data primary differentiation processing result data S2 _t , and data secondary differentiation processing result data B2 _t are output collectively or sequentially. Here, in the first two-stage exponential smoothing, the smoothing parameter α1 of the data smoothing, the smoothing parameter γ1 of the smoothed data slope, and the smoothing of the data smoothing in the second two-stage exponential smoothing The smoothing parameter α2 and the smoothing parameter slope smoothing parameter γ2 are set to arbitrary constants in advance.

However, 0 <α1 <1, 0 <γ1 <1, 0 <α2 <1, and 0 <γ2 <1. Since the predicted output value B1 _t of the smoothed data of the first output also corresponds to the result of the first derivative processing, it may be used. However, since the data results vary greatly, the second two-stage exponential smoothing is performed. The data smoothing process is performed.

FIG. 3 shows a flowchart for deriving the initial value of this embodiment. In general, in the two-stage exponential smoothing process, the initial value of the predicted value S1 of data smoothing and the initial value of the predicted value B1 of the slope of smoothed data are derived, for example, by the following method. The initial value of the data smoothing prediction value S1 is S1 = input data Y ₁ (method A1) or S1 = average value of initial N input data ({Y ₁ + Y ₂ +... + Y _N } / N) etc. (Method A2). The initial value of the predicted value B1 of the slope of the smoothed data is B1 = Y ₂ -Y ₁ (method B1), B2 = {(Y ₂ -Y ₁ ) + (Y ₄ -Y ₃ )} / 2, etc. (method Derived in B2).

In general, the two-stage exponential smoothing process has a problem that an error is large immediately after the start of data processing. One of the causes is that the initial value according to the conventional derivation method has a large error between the predicted value of the smoothing of the true initial data and the predicted value of the slope of the true initial smoothed data. . In addition, conventionally, the desired initial N data is used to derive the initial value, and then the N + 1-th and subsequent input data Y _{N + 1} ,. The initial N pieces of input data were not subjected to the two-stage exponential smoothing process, and the data output results for the above period were missing.

  In particular, when the number of input data to be processed is small, the effect of missing data output results becomes significant. From the above, when the state of the target system changes immediately after the start of data processing, the target system is increased due to a large error in data processing immediately after the start of data or a lack of data output immediately after the start of data. There was a problem that it was difficult to control accurately.

As shown in FIG. 3, in this embodiment, a polynomial approximation formula is obtained by the least square method using desired initial data Y1 _t (t = 1, 2,..., N) after the start of input of N pieces of data. Is derived. In this example, ten time-series data at equal time intervals were used. From polynomial approximate expression derived above, we derive the virtual data prediction value S1 ₀ of the smoothing of the data is, the predicted value B1 ₀ of inclination of the smoothed data immediately before t = 0 of the input data. In this embodiment, a linear linear expression is used as a polynomial approximation expression, and the predicted value S1 _{0 of} the smoothing of data and the predicted value B1 ₀ of the slope of the smoothed data are expressed by Expressions (3) and (4), respectively. ).
Predicted value of data smoothing S1 ₀ = {330 Y1 ₁ + 275 Y1 ₂ + 220 Y1 ₃ + 165 Y1 <sub>4
+ 110 Y1 5 + 55 Y1 6</sub> + 0 Y1 7 - 55 Y1 8 - 110 Y1 9 - 165 Y1 10} / 825 Equation (3)
Predicted value of the slope of the smoothed data _{B1 0 = {-45 Y1 1 -} 35 Y1 2 - 25 Y1 3 - 15 Y1 4
-5 Y1 ₅ + 5 Y1 ₆ + 15 Y1 ₇ + 25 Y1 ₈ + 35 Y1 ₉ + 45 Y1 ₁₀ } / 825 Formula (4)
Further, in the second two-stage exponential smoothing process shown in FIG. 2, the initial value S2 ₁ of the predicted value of the smoothing of data and the initial value B2 ₁ of the predicted value of the slope of the smoothed data are respectively S2 ₁ = S1 ₁ and B2 ₁ = 0. In the case of sequential data processing, after initial N pieces of data are input, an initial value is derived, and then the initial N pieces of data are sequentially subjected to a two-stage exponential smoothing process, and data smoothing processing and data differentiation processing results are output. . After that (after t = N + 1), the 2-stage exponential smoothing process is sequentially performed for each data input, and the data smoothing process and the data differentiation process result are output in real time. For this reason, according to the present embodiment, there is an effect that the initial input data after the start of the data input has less errors and can perform the data smoothing process and the data differentiation process with high accuracy.

  FIG. 4 is a longitudinal sectional view of a magnetic field microwave plasma etching apparatus according to the first embodiment of the present invention. In this embodiment, the magnetic field microwave plasma etching apparatus corresponds to the target system 6 (apparatus, analysis data, etc.) in FIG. The inside of the processing chamber 11 defined by the container 7, the discharge tube 8, the quartz plate 9 and the quartz window 10 is decompressed by the vacuum exhaust device 13 with the exhaust opening / closing valve 12 being opened. The etching gas passes through the gas pipe 14 through the mass flow controller (not shown), passes between the quartz plate 9 and the quartz shower plate 15, and is introduced into the processing chamber 11 from the gas hole of the quartz shower plate 15. Is done. The etching gas introduced into the processing chamber 11 adjusts the pressure in the processing chamber 11 to a desired pressure by the exhaust speed variable valve 16.

  Further, the processing chamber 11 is in a magnetic field region generated by the coils 17 and 18 and the yoke 19. In this case, a microwave having a frequency of 2.45 GHz oscillated from the magnetron 20 propagates in the rectangular TE10 mode through the isolator (not shown), the power monitor (not shown), and the matching unit 21 in the rectangular waveguide 22. Then, the light propagates in the circular waveguide 24 in the circular TE11 mode via the circular-rectangular converter 23. Thereafter, the microwave is introduced into the cavity resonator 25, passes through the quartz plate 9 and the quartz shower plate 15, and enters the processing chamber 11. In the processing chamber 11, a magnetic field region having a magnetic flux density of 875 Gauss that generates electron cyclotron resonance with the introduced 2.45 GHz microwave is perpendicular to the central axis of the processing chamber 11 and the direction in which the microwave is introduced, and the processing chamber 11. It is formed on the entire surface with respect to the cross-sectional direction with respect to the central axis.

  The wafer 27 placed on the wafer placement electrode 26 which is a sample stage is etched from the plasma mainly generated by the interaction between the 2.45 GHz microwave and the 875 Gauss magnetic field. In addition, in order to control the etching shape of the wafer 27 as a sample, a high frequency power supply 28 is connected to the wafer mounting electrode 26 via a matching unit (not shown) so that a high frequency voltage can be applied. Yes. Further, a chiller unit (not shown) is connected to the wafer mounting electrode 26 so that the temperature of the wafer 27 can be controlled.

  The processing chamber 11, the wafer 27, and the wafer mounting electrode 26 are coaxially arranged, the gas hole region of the quartz shower plate 15 for introducing the etching gas, the exhaust opening / closing valve 12 that is a vacuum exhaust unit, and the exhaust speed variable valve 16. The vacuum exhaust device 13 is also arranged coaxially with respect to the processing chamber 11. For this reason, the gas flow on the wafer 27 is coaxially symmetric. Since the coils 17 and 18 and the yoke 19 that generate the magnetic field are also arranged coaxially with the processing chamber 11, the magnetic field profile in the processing chamber 11 and the electron cyclotron resonance region of the magnetic flux 875 Gauss are formed coaxially with the processing chamber 11. The Further, since the circular waveguide 24 and the cavity resonator 25 are also arranged coaxially with respect to the processing chamber 11, the microwave introduced into the processing chamber 11 is also introduced coaxially with respect to the processing chamber 11.

  Since the magnetic field is generated coaxially with respect to the processing chamber 11 and the microwave is also introduced coaxially with respect to the processing chamber 11, the plasma formed by the interaction between the magnetic field and the microwave is generated coaxially with respect to the processing chamber 11. Then, electrons and ions in the plasma are transported coaxially with respect to the wafer 27. Further, since the flow of the etching gas is coaxial with respect to the processing chamber 11, radicals generated by plasma and reaction products resulting from etching of the wafer 27 are also introduced and exhausted coaxially with respect to the wafer 27. Therefore, the etching process performance such as the etching rate, the material selection ratio, and the etching shape can be etched with good uniformity within the wafer surface.

  Light from the plasma generated in the processing chamber 11 passes from the side of the processing chamber 11 through the quartz window 10 and the optical fiber 29 and is introduced into the spectroscope 30, and as time-series data on the wavelength dependence of the light intensity. Is output. Further, from above the processing chamber 11, the quartz shower plate 15, the quartz plate 9, the cavity resonator 25, the circular waveguide 24, the circular rectangular converter 23, and the optical fiber 31 are introduced into the spectroscope 32, and the light intensity is increased. Is output as time-series data of the wavelength dependence.

  An etching gas and an etching reaction product from the wafer 27 are introduced into the processing chamber 11, and these are dissociated by the interaction between the microwave and the magnetic field to generate plasma. For this reason, the light from the plasma generated in the processing chamber 11 includes atoms, molecules, radicals, and information on these reactants that constitute the etching gas and the etching reaction product.

For example, in a typical poly-Si etching of a Si substrate in which a poly-Si film and a SiO ₂ film are arranged below the pattern mask, it is required to perform poly-Si etching with a high selectivity with respect to the underlying SiO 2. As the etching gas, a halogen-based gas is used, and the etching reaction product contains Si as an etching target material and halogen. Since the etching reaction product is re-dissociated by the plasma, the light intensity of light emission having a wavelength of 288 nm caused by Si from the plasma is monitored by the spectrometer 30 or the spectrometer 32. In this case, when the etching of the poly-Si film is completed and the underlying SiO2 appears, the etching rate of the underlying SiO2 is small, so the plasma emission intensity at a wavelength of 288 nm due to Si rapidly decreases and finally reaches a constant value. Get closer. This change in plasma emission is monitored to detect the end point of etching.

  The light from the side of the processing chamber 11 includes information on the etching gas and the etching reaction product. In addition to the above information, the light from the upper side of the processing chamber 11 includes plasma light and the film structure and level difference of the wafer 27. In order to cause interference in the structure, information on the film structure and the step structure of the wafer 27 is also included. By analyzing the plasma emission data, the film thickness during etching, the etching depth, and the like can be monitored. In this embodiment, for the sake of simplicity, the plasma emission data from the side of the processing chamber 11 is used for the etching end point monitor.

  FIG. 5 shows a typical data processing flow for detecting the etching end point. The input data was created with an evaluation function of Equation (5) that simulated the change in plasma emission intensity during etching.

Y (t) = H / [1 + exp {-A (t-T)}] + C t + D + F (R-0.5) Equation (5)
Here, H, A, T, C, D, and F are arbitrary constants, and R is a random number from 0 to 1. If the above evaluation function is used, the analytical true value of the data smoothing process, the first derivative process, and the second derivative process is known. Therefore, in various data processing methods, the absolute error from the true value, data processing Data processing performance such as the accompanying delay time and S / N ratio (signal / noise ratio) can be compared and evaluated.

  As shown in FIG. 5, the typical data processing flow for the etching end point is after the input data waveform as shown in FIG. 5 (A) is subjected to the data smoothing process as shown in FIG. 5 (B). First differential processing is performed as shown in FIG. 5 (C), and second differential processing is performed as shown in FIG. 5 (D). In the input data including a lot of noise, the change point becomes clear by the data smoothing process. This change point is detected as a peak value point (time) in the primary differentiation process and a zero-crossing point (time) in the secondary differentiation process. Based on this, the etching end point is determined, the etching apparatus is controlled, and highly accurate etching processing is performed.

  The change point is the peak of the first derivative process and the zero cross of the second derivative process, which can be determined more clearly and more easily in sequence, but the absolute value of the signal intensity decreases sequentially. For this reason, data processing with a high S / N ratio is important. In particular, in the case of etching a mask pattern with a small area to be etched and a low aperture ratio, since the change in plasma emission intensity before and after the etching end point is small, data processing with a higher S / N ratio is required. Generally, in data smoothing processing and data differentiation processing, as the S / N ratio increases, the delay time increases and the absolute error from the true value increases. In other words, the S / N ratio, delay time, and absolute error are in a trade-off relationship, and data smoothing processing and data differentiation processing that simultaneously satisfy the S / N ratio, delay time, and absolute error are required.

  In the present embodiment, the data processing of FIG. 5 is performed using the data smoothing processing and data differentiation processing flow of FIG. 2 and FIG. 5 corresponds to output data from the spectroscope 30 that monitors plasma emission during etching in the magnetic field microwave plasma etching apparatus of FIG. As shown in FIG. 4, a system control device 33 (including a data input / output device, a data processing device, a data display device, etc.) that controls the magnetic field microwave etching device as a system, and the data processing device 1 of this embodiment include Is provided. The data processing device 1 may be incorporated as a part of the system control device 33.

  Output data from the spectroscope 30 and the spectroscope 32 is transmitted to the data processing apparatus 1, and the data smoothing processing result, the data primary differential result, and the data secondary differential result are transmitted to the system control apparatus 33 which is a control apparatus. The Based on the data smoothing processing result, the data primary differential result, and the data secondary differential result, the system control device 33 determines the etching end point and controls the magnetic field microwave etching device as a system. In the etching end point determination, in order to mainly control plasma generation, FIG. 4 shows the system controller 33 connected to the magnetron 20 and the high frequency power supply 28. Moreover, although illustration in FIG. 4 is omitted, the system control device 33 is also connected to other devices constituting the system.

  FIG. 6 shows a data processing result by the data processing apparatus 1 according to the first embodiment of the present invention. 6A shows the input data and data smoothing processing waveform, FIG. 6B shows the primary differentiation processing waveform, and FIG. 6C shows the secondary differentiation processing waveform. Further, FIG. 7 shows a data processing result by the data processing apparatus 1 based on the conventional low-pass filter and difference method. In this case, the cut-off frequency of the low-pass filter was 0.025 Hz. FIG. 7A shows input data and data smoothing processing waveforms, FIG. 7B shows primary differentiation processing waveforms, and FIG. 7C shows secondary differentiation processing waveforms. Here, the same input data is used in FIG. 6 and FIG.

  6 and 7 are compared, as shown in FIG. 6A, in the data smoothing processing and data differentiation processing device of the first embodiment of the present invention, the absolute error between the input data and the data smoothing processing result is It can be seen that the data smoothing process can be carried out satisfactorily. Further, as shown in FIGS. 6B and 6C, the primary differentiation process and the secondary differentiation process can be performed with a high S / N ratio. Therefore, it is possible to clearly detect the peak point (time) of the first-order derivative and the zero-cross point (time) of the second-order derivative, which are references for determining the etching end point. Comparing the data processing performance with numerical data, the absolute error is 22.4 in the data smoothing processing and data differentiation processing device of the first embodiment of the present invention, and 191.2 in the conventional low-pass filter and difference method.

  The S / N ratio is 32 in the data smoothing processing and data differentiation processing device of the first embodiment of the present invention, and 28 in the conventional low-pass filter and the difference method. The zero-crossing time of the second derivative is 5.4 seconds earlier and the delay time is shorter than the conventional low-pass filter and difference method by the data smoothing processing and data differentiation processing device of the first embodiment of the present invention.

  According to the first embodiment of the present invention, there is an effect that data smoothing processing and data differentiation processing can be sequentially performed in real time with a small absolute value error, a high S / N ratio, and a short delay time.

  Next, a data processing apparatus according to the second embodiment of the present invention will be described. In this embodiment, in FIG. 2 showing the entire flow of the data smoothing processing and data differentiation processing method in the first embodiment of the present invention, equations (1) and (1) 2) is replaced by the following formulas (6) to (15).

Data smoothing: S1 _t = α1 _t Y _t + (1-α1 _t ) (S1 _t-1 + B1 _t-1 ) Equation (6)
Slope of smoothed data: B1 _t = γ1 _t (S1 _t -S1 _t-1 ) + (1-γ1 _t ) B1 _t-1 equation (7)
Smoothing coefficient: α1 _t = (K _α -L _α ) F _α + L _α equation (8)
Response coefficient: F _α = (| δα _t / Δα _t | + φ) Formula ^N (9)
Relative error: δα _t = A1 (Y _t -S1 _t ) + (1-A1) δα _t-1 formula (10)
Absolute error: Δα _t = A1 | Y _t -S1 _t | + (1-A1) Δα _t-1 + φ Equation (11)
Smoothing coefficient: γ1 _t = (K _γ -L _γ ) F _γ + L _γ equation (12)
Response coefficient: F _γ = (| δγ _t / Δγ _t | + φ) Formula ^N (13)
Relative error: δγ _t = A2 {(S1 _t -S1 _{t -1} )-B _t } + (1 -A2) δγ _t-1 (14)
Absolute error: Δγ _t = A2 | (S1 _t -S1 _t-1 )-B _t | + (1-A2) Δγ _t-1 + φ Equation (15)
Here, K _α , L _α , K _γ , L _γ , N, A1, A2, and φ are arbitrary constants. _{However, 1> K α> L α} >0,1> K γ> L γ>0,1>A1>0,1>A2> 0. φ is to avoid the absolute errors Δα _t and Δγ _t and the response coefficients F _α and F _γ from becoming zero values. Selected.

The data smoothing equation (6) and the smoothing data gradient equation (7) are the same as in the first embodiment of the present invention shown in FIG. As shown in equation (12), the smoothing coefficient α1 _t represents the range of K _α > α1 _t > L _α , and the smoothing coefficient γ1 _t represents the range of K _γ > γ1 _t > L _γ . It changes sequentially according to the situation. Also, as shown in equations (9) and (13), the response coefficient F _α is a power response of the absolute value of the relative error / absolute error of the data smoothing, and the response coefficient F _γ is the response coefficient F _γ This is the power response of the absolute value of the relative error / absolute error of the slope. In Equation (10), Equation (11), Equation (14), and Equation (15), the influence of the most recent data is larger, and the past data is accumulated so that the influence decreases exponentially. Smoothing is performed, and the constants A1 and A2 are parameters for integrating and smoothing exponentially.

The relative error Δα _t and the absolute error Δα _t in the equations (10) and (11) calculate the relative error and the absolute error between the input data Y _t and the data smoothing prediction value S1 _t , respectively. The equation (14), relative error [Delta] [gamma] _t of formula (15), the absolute error [Delta] [gamma] _t are each slope of the predicted value of the smoothed data (S1 _t -S1 _t-1) and the predicted tilt of the smoothed data The relative error and absolute error with the value B _t are calculated.

  FIG. 8 shows the relationship between the response coefficient and the absolute value of the relative error / absolute error when the power N is a parameter in the second embodiment of the present invention. The absolute value of the relative error / absolute error can be evaluated from the maximum value 1 to the minimum value 0 as the deviation between the input data and the predicted value of the smoothing of the data. When the absolute value of the relative error / absolute error is 0, the input data and the smoothed prediction value of the data match. As shown in FIG. 8, when N = 0, the response coefficient of the smoothing coefficient is 1, and the smoothing coefficient is constant, so that it is substantially the same as the first embodiment of the present invention.

  When N = 1, the response coefficient increases from 0 to 1 in proportion to the absolute value of the relative error / absolute error, so that the smoothing coefficient increases in proportion from the minimum value to the maximum value as well. . When N = 5, the absolute value of the relative error / absolute error is the fifth power of the absolute value, and the response coefficient increases from 0 to 1, so the smoothing coefficient is the fifth power response from the minimum value to the maximum value as well. Increase with. That is, when comparing N = 1 and N = 5, the smoothing coefficient is kept small in the case of N = 5 even if the predicted value of smoothing of the input data and data is slightly shifted, and conversely the input data If the predicted value of the smoothing of the data deviates greatly, there is an effect of increasing the smoothing coefficient rapidly.

  If the smoothing coefficient is small, the S / N ratio of data processing increases, but the absolute error becomes large and the delay time becomes long. On the other hand, when the smoothing coefficient is large, the absolute error becomes small and the delay time becomes short, but the S / N ratio of data processing decreases. It is necessary to select an optimal smoothing coefficient according to input data. In the first embodiment of the present invention, the smoothing coefficient is fixed. However, in the second embodiment of the present invention, the smoothing coefficient is, for example, a difference between the input data and the predicted value of the data smoothing. Adapt and adjust sequentially. For this reason, the trade-off between data processing S / N ratio and absolute error and delay time is avoided, and data smoothing processing and data differentiation processing are performed with high S / N ratio, small absolute error and short delay time. be able to.

  The smoothing coefficient is sequentially changed according to the error between the input data and the predicted value of the smoothing of the data. Once the smoothing coefficient becomes zero, the smoothing coefficient is thereafter No change. If the smoothing coefficient is close to zero, the absolute error and the delay time become long. Therefore, setting the lower limit value within the movable range of the smoothing coefficient is stable and optimal in the adaptive two-stage exponential smoothing process. There is an effect that it is possible to perform the data processing.

  FIG. 9 shows the data processing result when the power parameter N = 1 in the data smoothing processing and data differentiation processing apparatus of the second embodiment of the present invention. FIG. 9A shows the input data and data smoothing processing waveform, FIG. 9B shows the primary differentiation processing waveform, and FIG. 9C shows the secondary differentiation processing waveform. FIG. 10 shows a data processing result when the power parameter N = 5 in the data processing apparatus according to the second embodiment of the present invention. 10A shows input data and data smoothing processing waveforms, FIG. 10B shows primary differentiation processing waveforms, and FIG. 10C shows secondary differentiation processing waveforms. Here, the input data in FIGS. 9 and 10 is the same as the input data in FIGS.

  Comparing FIG. 9 and FIG. 6, in the data processing apparatus according to the second embodiment of the present invention, when the power parameter N = 1, as shown in FIG. 9B and FIG. The second order differential process and the second order differential process can be performed with a higher S / N ratio than in the first embodiment of the present invention. Further, comparing FIG. 9 and FIG. 10, in the data processing apparatus of the second embodiment of the present invention, when the power parameter N = 5, as shown in FIG. 10 (B) and FIG. 10 (C). It can be seen that the primary differential process and the secondary differential process can be performed at a higher S / N ratio than when the power parameter N = 1.

  When the data processing performance is compared with numerical data, the absolute error is 23.6 in the case of the power parameter N = 1 in the data smoothing processing and data differentiation processing device of the second embodiment of the present invention, and the power parameter N = In the case of 5, it is 27.6. The S / N ratio is 226.5 when the power parameter N = 1 and 1607.2 when the power parameter N = 5. The zero-crossing time of the second derivative is 0.4 seconds earlier and the delay time is shorter when the power parameter N = 1 than when the power parameter N = 5. Therefore, the power parameter N is optimally selected according to the priority of the high S / N ratio, absolute error, and delay time in data processing performance, and desired data smoothing processing and data differentiation processing can be performed.

  FIG. 11 shows the data processing result in the case where there is a data change point immediately after the start of data input when the power parameter N = 1 of the data processing apparatus of the second embodiment of the present invention. Even when there is a data change point immediately after the start of data input, the data smoothing process result, the primary differential process result, and the secondary differential process result are not lost and the error is smaller than the data immediately after the start of data input. Output is good.

  For this reason, it is particularly effective for detecting an etching end point in a short etching process. As semiconductor devices are highly integrated and miniaturized, the number of processes for etching a multilayer thin film is increasing in semiconductor etching, and it is important to detect the etching end point in a short etching process and etching step. For this short-time process correspondence, the smoothed prediction value of the data immediately before the start of data input and the predicted value of the slope of the smoothed data are derived by the polynomial approximation described with reference to FIG. 3 of the first embodiment of the present invention. The contribution of the initial value is large. In addition, according to the second embodiment of the present invention, the S / N ratio was improved by the first-order differential process and the second-order differential process, and the etching end point could be determined more clearly.

  In the case of the present embodiment, in addition to the effect of the first embodiment, the smoothing coefficient is sequentially optimized in accordance with the error between the input data and the predicted value of the smoothing of the data, so that the data processing performance is achieved. Overcoming the trade-off relationship between S / N ratio, absolute value error, and delay time, data smoothing processing and data differentiation processing can be performed at a high S / N ratio. For this reason, since the state of the target system (device etc.) and the change of the state can be detected clearly, there is an effect that the target system (device etc.) can be controlled with high accuracy.

  Next, a data processing apparatus 1 according to a third embodiment of the present invention will be described with reference to FIG. In this embodiment, in FIG. 2 showing the overall flow of the data smoothing process and the data differentiation process method in the first embodiment of the present invention, the formulas (1) and (2) of the first two-stage exponential smoothing process are shown. Is replaced by the following formulas (16) to (18) of the three-stage exponential smoothing process.

Predicted value of data smoothing: S1 _t = α1 Y1 _t + (1 -α1) (S1 _t-1 + B1 _t-1 ) Equation (16)
Predicted slope of smoothed data:
B1 _t = γ1 (S1 _t -S1 _t-1 ) + (1 -γ1) (B1 _t-1 + G1 _t-1 ) Equation (17)
Predicted slope of smoothed data slope:
G1 _t = β1 (B1 _t -B1 _t-1 ) + (1-β1) G1 _t-1 equation (18)
Next, the predicted value B1 _t of the smoothed data of the first output is set as the second input data Y2 _t, and the predicted value G1 _t of the smoothed data of the first output is set as the second input data. Set Y3 _t and perform two second-stage exponential smoothing processes. Thereby, predicted values S2 _t and S3 _t of smoothing of the two data of the second output are obtained. Next, data smoothing processing result data S1 _t , data primary differentiation processing result data S2 _t , and data secondary differentiation processing result data S3 _t are output collectively or sequentially.

  Here, in the first three-stage exponential smoothing, the smoothing parameter α1 of the predicted value of the smoothing of the data, the smoothing parameter γ1 of the predicted value of the smoothed data slope, and the predicted value of the slope of the smoothed data slope Smoothing parameter β1 and smoothing parameters α2 and α3 of the predicted value of data smoothing and smoothing parameters γ2 and γ3 of the predicted value of the smoothed data in the second two-stage exponential smoothing of It is set in advance. However, 0 <α1 <1, 0 <γ1 <1, 0 <β1 <1, 0 <α2 <1, 0 <γ2 <1, 0 <α3 <1, 0 <γ3 <1.

The predicted value B1 _t of the slope of the smoothed data output for the first time also corresponds to the result of the first derivative process, and the predicted value G1 _t of the slope of the smoothed data of the first output corresponds to the result of the second derivative process. For this reason, this may be used, but the data smoothing process is performed by the second two-stage exponential smoothing process because the variation of the data result is large.

  In the present embodiment, the smoothing parameter is fixed, but as in the second embodiment of the present invention, the smoothing coefficient is sequentially changed according to the error between the input data and the predicted value of the smoothing of the data, An adaptive three-stage exponential smoothing process may be used.

  In the case of the present embodiment, there are the same effects as the first embodiment. Further, the data smoothing process, the primary differentiation process, and the secondary differentiation process may be performed only by the first three-stage exponential smoothing process. In this case, there is an effect that the data smoothing process, the first-order differential process, and the second-order differential process can be performed easily and with a small data storage capacity by one process. Further, as can be easily guessed, by extending the equations (16) to (18) to the N-stage exponential smoothing process in the same manner as in this embodiment, the data smoothing process and the first-order differential process can be changed to N. Up to -1st order differential processing can be performed.

  As described above, in the above-described embodiments, the case where the etching is applied with high accuracy by applying to the etching end point detection in the microwave plasma etching apparatus has been described in detail. Also in an etching apparatus and a film forming apparatus, or a processing apparatus and other apparatuses in other fields, numerical data obtained from the apparatus or the like is used as an input, and the data processing apparatus and data processing method of the present invention are applied. Since the state change point can be monitored and the change point of the state can be detected with high accuracy, the target device can be controlled with high accuracy.

  Also, in the economic and financial fields such as demand and supply prediction, there is an effect that data can be analyzed with high accuracy by applying the data processing apparatus and data processing method of the present invention.

  According to the present invention, data smoothing processing and data differentiation processing are performed in sequential data processing, with a high S / N ratio, small data delay, and high reliability in the initial period of data processing. Processing can be performed. Further, according to the present invention, the data smoothing value, the first derivative value, and the second derivative value can be obtained sequentially and in real time with high S / N ratio, short delay time, or high reliability at the beginning of data processing. . Furthermore, the present invention can control the target system with high accuracy using the data smoothing value, the first derivative value, and the second derivative value.

  In addition, this invention is not limited to an above-described Example, Various modifications are included. For example, the above-described embodiments have been described in detail for easy understanding of the present invention, and are not necessarily limited to those having all the configurations described. Further, a part of the configuration of a certain embodiment can be replaced with the configuration of another embodiment, and the configuration of another embodiment can be added to the configuration of a certain embodiment. Further, it is possible to add, delete, and replace other configurations for a part of the configuration of each embodiment.

DESCRIPTION OF SYMBOLS 1 ... Data processing device, 2 ... Data input / output device, 3 ... Data storage device, 4 ... Data processing program storage device, 5 ... Data operation processing device, 6 ... Target system, 7 ... Container, 8 ... Discharge tube, DESCRIPTION OF SYMBOLS 9 ... Quartz plate, 10 ... Quartz window, 11 ... Processing chamber, 12 ... Exhaust opening / closing valve, 13 ... Vacuum exhaust apparatus, 14 ... Gas piping, 15 ... Quartz shower plate, 16 ... Exhaust speed variable valve, 17 ... Coil, DESCRIPTION OF SYMBOLS 18 ... Coil, 19 ... Yoke, 20 ... Magnetron, 21 ... Matching device, 22 ... Rectangular waveguide, 23 ... Circular rectangular converter, 24 ... Circular waveguide, 25 ... Cavity resonator, 26 ... For wafer mounting Electrode, 27 ... wafer, 28 ... high frequency power supply, 29 ... optical fiber, 30 ... spectrometer, 31 ... optical fiber, 32 ... spectrometer, 33 ... system controller

Claims (7)

A processing chamber to be controlled;
A measuring device for acquiring data relating to the processing chamber;
A response type two-stage exponential smoothing method that changes the smoothing parameter based on an error between the data acquired by the measuring device and the predicted value of the smoothed data, and the lower limit value of the range in which the smoothing parameter is changed An arithmetic processing unit that performs data processing based on a data processing program that processes data with a value larger than 0;
At least one processing result among the data smoothing processing result obtained by the arithmetic processing device, the primary differentiation processing result obtained by the arithmetic processing device, and the secondary differentiation processing result obtained by the arithmetic processing device. A control device that controls the processing chamber based on a change in the state of the processing chamber detected based on or a change in the state of the processing chamber detected,
When N is 0 or a natural number, the response coefficient of the smoothing parameter is the Nth power response of the absolute value of the value obtained by dividing the relative error by the absolute error,
The relative error is a relative error between the data acquired by the measurement device and the predicted value of the smoothed data,
The processing apparatus according to claim 1, wherein the absolute error is an absolute error between the data acquired by the measuring device and a predicted value of the smoothed data. A processing chamber in which a sample placed on a sample stage is plasma-treated;
A measuring device for acquiring emission data during plasma processing of the sample;
A response type two-stage exponential smoothing method that changes the smoothing parameter based on an error between the data acquired by the measuring device and the predicted value of the smoothed data, and the lower limit value of the range in which the smoothing parameter is changed An arithmetic processing unit that performs data processing based on a data processing program that processes data with a value larger than 0;
At least one processing result among the data smoothing processing result obtained by the arithmetic processing device, the primary differentiation processing result obtained by the arithmetic processing device, and the secondary differentiation processing result obtained by the arithmetic processing device. A control device for controlling the processing chamber based on a change in the plasma processing state of the sample detected based on or a change in the plasma processing state of the detected sample,
When N is 0 or a natural number, the response coefficient of the smoothing parameter is the Nth power response of the absolute value of the value obtained by dividing the relative error by the absolute error,
The relative error is a relative error between the data acquired by the measurement device and the predicted value of the smoothed data,
The processing apparatus according to claim 1, wherein the absolute error is an absolute error between the data acquired by the measuring device and a predicted value of the smoothed data. The processing apparatus according to claim 1 or 2,
The relative error and the absolute error are subjected to exponential smoothing processing. In a plasma processing method for plasma processing a sample,
Measure emission data during plasma treatment of the sample,
Using a response type two-step exponential smoothing method that changes a smoothing parameter based on an error between the measured light emission data and a predicted value of the smoothed light emission data, and a lower limit value of a range in which the smoothing parameter is changed Process the flash data larger than 0,
Data smoothing processing result obtained by the response-type two-stage exponential smoothing method, first-order differential processing result obtained by the response-type two-stage exponential smoothing method, and second-order obtained by the response-type two-stage exponential smoothing method Controlling the plasma processing of the sample based on a change in the plasma processing state of the sample detected based on at least one of the differential processing results or the detected plasma processing state of the sample;
When N is 0 or a natural number, the response coefficient of the smoothing parameter is the Nth power response of the absolute value of the value obtained by dividing the relative error by the absolute error,
The relative error is a relative error between the measured light emission data and the predicted value of the smoothed light emission data .
The plasma processing method, wherein the absolute error is an absolute error between the measured light emission data and a predicted value of the smoothed light emission data . In a data processing method for processing data using exponential smoothing,
Acquire data related to the processing chamber to be controlled,
A response type two-step exponential smoothing method is used in which a smoothing parameter is changed based on an error between the acquired data and a predicted value of the smoothed data, and the lower limit value of the range in which the smoothing parameter is changed is 0. To process data,
Data smoothing processing result obtained by the response-type two-stage exponential smoothing method, first-order differential processing result obtained by the response-type two-stage exponential smoothing method, and second-order obtained by the response-type two-stage exponential smoothing method Controlling the processing chamber based on the state of the processing chamber detected based on at least one processing result among the differential processing results or a change in the state of the processing chamber detected;
When N is 0 or a natural number, the response coefficient of the smoothing parameter is the Nth power response of the absolute value of the value obtained by dividing the relative error by the absolute error,
The relative error is a relative error between the acquired data and a predicted value of the smoothed data ,
The absolute error is an absolute error between the acquired data and a predicted value of smoothed data. The plasma processing method according to claim 4, wherein
The plasma processing method, wherein the relative error and the absolute error are exponentially smoothed. The data processing method according to claim 5, wherein
The relative error and the absolute error are subjected to exponential smoothing processing.

Images