Plasma processing method

Abstract

The invention provides a plasma processing method. The thickness of a film to be processed can be detected with good precision even if the fine shape of the surface of a semiconductor wafer is deviated. In the plasma processing apparatus provided with a processing state detection means for detecting the state of a film to be processed of a workpiece to be processed inside a vacuum processing chamber, the processing state detection means is configured by being provided with: a light emission detection unit for detecting light emission of plasma; a calculation unit that obtains differential waveform data of light emission of the plasma; a database unit that stores a plurality of differential waveform pattern data; a film thickness calculation unit that gives a weight on the basis of the difference between the differential waveform data obtained by the calculation unit and the plurality of differential waveform pattern data stored in the database unit, and calculates an estimated value of the film thickness of the film to be processed on the workpiece; and an end point determination unit that determines the end point of the plasma processing on the basis of the estimated value of the film thickness of the film to be processed calculated by the film thickness calculation unit.

Description

[0001] This application is a divisional application of the application No. 202111084850.0 titled "Plasma Processing Apparatus and Plasma Processing Method" filed with the State Intellectual Property Office of China on September 14, 2021. Technical Field

[0002] The present invention relates to a processing apparatus and a processing method for a workpiece in the manufacture of semiconductor integrated circuits and the like, and particularly relates to a plasma processing apparatus and a plasma processing method suitable for accurately detecting the etching amount of various layers provided on a substrate by an etching process using plasma discharge and processing them into a desired film thickness and etching depth. Background Art

[0003] In the manufacture of semiconductor devices, in the removal or patterning of layers of various materials formed on the surface of a semiconductor wafer, and particularly layers of dielectric materials, processing using a dry etching apparatus is widely used.

[0004] In this dry etching apparatus, a processing gas introduced into a vacuum processing chamber is plasmaized to generate ions and radicals, and the semiconductor wafer is etched by reacting the ions and radicals with the layer formed on the surface of the semiconductor wafer.

[0005] In this etching process, it is important to accurately detect the etching end point for stopping the etching process at a desired film thickness and etching depth in the processing of such layers.

[0006] In the dry etching process of a semiconductor wafer, the emission intensity of a specific wavelength in plasma light changes as the etching of a specific film progresses. Therefore, as one of the methods for determining the etching end point of a semiconductor wafer, there has been a method in the past: detecting the change in the emission intensity of a specific wavelength from plasma in the dry etching process and detecting the etching end point of a specific film based on the detection result.

[0007] As an example of such a technique, the scheme described in Japanese Patent Application Laid-Open No. 2007-234666 (Patent Document 1) has been known in the past. In this prior art, a method for performing end point determination based on the temporal change in the amount of reflected light from a semiconductor wafer during etching is described.

[0008] In addition, since the amount of reflected light from the semiconductor wafer as described above also changes depending on the film thickness of layers other than the layer being processed, as a method for accurately detecting the end point even under such conditions, the scheme described in Japanese Patent Application Laid-Open No. 2016-184638 (Patent Document 2) has been known in the past. In this prior art, a high-precision film thickness estimation method in the case where the film thickness of the layer under the film to be processed (base film thickness) is different is disclosed.

[0009] The following method is disclosed in Patent Document 1: By detecting and obtaining the characteristic behavior of the temporal change of the interference light and databaseizing it, the end determination of etching is performed by comparing this database with the detected interference waveform. When performing plasma etching on a workpiece such as a semiconductor wafer, a standard pattern representing the wavelength dependence of the interference light with respect to the etching amount of the workpiece for sampling (sampling semiconductor wafer) is set, and thus a database is created.

[0010] Patent Document 2 describes: Preparing interference spectrum patterns (interference patterns) with different substrate film thicknesses as a database, creating a database generated by synthesizing two databases, calculating the estimated film thickness value at each moment by comparing the synthesized database and the detected interference pattern, and determining the end point.

[0011] Prior Art Documents

[0012] Patent Documents

[0013] Patent Document 1: Japanese Unexamined Patent Application Publication No. 2007 - 234666

[0014] Patent Document 2: Japanese Unexamined Patent Application Publication No. 2016 - 184638

[0015] However, in the above prior art, in the case where there are various deviations in the fine shape of the semiconductor wafer surface, such as not only deviations in the substrate film thickness but also deviations in the mask film thickness, deviations in the width of the mask, and deviations in the film thickness at each position on the semiconductor wafer of the film to be processed, high-precision film thickness estimation cannot be achieved.

[0016] For example, consider the case where the substrate film thickness and the mask film thickness are different, and the fine shape of the semiconductor wafer to be processed is such that the substrate film thickness is slightly thick and the mask film thickness is slightly thick. In this case, it is necessary to prepare a database of interference spectrum patterns obtained from four semiconductor wafers with thick and thin substrate film thicknesses and thick and thin mask film thicknesses, and synthesize them appropriately, but such a method is not disclosed in either Patent Document 1 or 2. Summary of the Invention

[0017] An object of the present invention is to provide a plasma processing apparatus and a plasma processing method that can precisely detect or control the remaining film thickness of the film to be processed even in the case where two or more deviations occur in the fine shape (e.g., substrate film thickness and mask film thickness) of the semiconductor wafer surface, in consideration of the problems of the above prior art.

[0018] The plasma processing method of the present invention uses plasma formed in a processing chamber to process a film to be processed included in a plurality of film layers disposed on the surface of a wafer disposed inside the processing chamber. It is characterized in that light from inside the processing chamber is received at a plurality of times during the processing of the wafer, time-series data related to the light is detected, real pattern data related to the light with wavelength as a parameter obtained from the time-series data is compared with detection pattern data obtained using a plurality of pattern data that are pattern data with wavelength as a parameter obtained in advance, the thickness of the film to be processed is calculated, the plurality of pattern data are a plurality of pattern data with wavelength as a parameter obtained by removing data outside an allowable range from a given reference value in the plurality of pattern data in which the remaining film thickness of the film to be processed during the processing and the time-series data related to the light are correlated, and the end point of the processing using the plasma is determined based on the value of the film thickness.

[0019] To achieve the above object, the plasma processing apparatus of the present invention includes: a vacuum processing chamber that generates plasma in a state where the inside is evacuated to a vacuum to process a workpiece; a processing state detection component that detects the state of a processed film of the workpiece being processed inside the vacuum processing chamber; and a control unit that controls the vacuum processing chamber and the processing state detection component. In the plasma processing apparatus, the processing state detection component includes: a light emission detection unit that detects the light emission of the plasma generated inside the vacuum processing chamber; an arithmetic unit that obtains differential waveform data of the light emission of the plasma detected by the light emission detection unit; a database unit that stores a plurality of differential waveform pattern data in advance; a film thickness calculation unit that calculates an estimated value of the film thickness of the processed film processed on the workpiece by assigning a weight based on the difference between the differential waveform data obtained by the arithmetic unit and the plurality of differential waveform pattern data stored in the database unit; and an end point determination unit that determines the end point of the processing using the plasma based on the estimated value of the film thickness of the processed film calculated by the film thickness calculation unit.

[0020] In addition, to achieve the above object, the plasma processing method uses a plasma processing apparatus that generates plasma in a state where the inside of a vacuum processing chamber is evacuated to vacuum to process a processed film formed on a workpiece. The light emission of the plasma generated inside the vacuum processing chamber is detected by a light emission detection unit, differential waveform data of the light emission of the plasma detected by the light emission detection unit is obtained by an arithmetic unit, and a weight based on the difference between the differential waveform data obtained by the arithmetic unit and a plurality of differential waveform pattern data stored in a database unit is assigned by a film thickness calculation unit to calculate an estimated value of the film thickness of the processed film processed on the workpiece. The end point determination unit determines the end point of the processing using the plasma based on the estimated value of the film thickness of the processed film calculated by the film thickness calculation unit.

[0021] Furthermore, to achieve the above object, the plasma processing method uses a plasma processing apparatus that generates plasma in a state where the inside of a vacuum processing chamber is evacuated to vacuum to process a processed film formed on a workpiece. The light emission of the plasma generated inside the vacuum processing chamber is detected by a light emission detection unit, differential waveform data of the light emission of the plasma detected by the light emission detection unit is obtained by an arithmetic unit, and a weight based on the difference between the differential waveform data obtained by the arithmetic unit and a plurality of differential waveform pattern data stored in a database unit is assigned by a film thickness calculation unit to calculate an estimated value of the film thickness of the processed film processed on the workpiece. The processed film is processed until the estimated value of the film thickness calculated by the film thickness calculation unit reaches a preset remaining film thickness. When the estimated value of the film thickness calculated by the film thickness calculation unit reaches the preset remaining film thickness, the processed film is further processed for a processing time obtained by an additional processing time calculator required to reach a target film thickness from the preset remaining film thickness.

[0022] Effects of the Invention

[0023] According to the present invention, even when various deviations occur in the fine shape of a semiconductor wafer, the processing amount or remaining film thickness of the processed film can be detected with good accuracy.

[0024] In addition, according to the present invention, high-precision film thickness estimation and end point determination can be achieved with respect to various structural deviations between wafers, between batches, etc., and the yield of device manufacturing can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 It is a block diagram showing a schematic structure of a plasma processing apparatus according to Embodiment 1 of the present invention.

[0026] Figure 2AIt is a cross-sectional view of the workpiece involved in Embodiment 1 of the present invention, showing the state before processing.

[0027] Figure 2B It is a cross-sectional view of the workpiece involved in Embodiment 1 of the present invention, showing the state during processing.

[0028] Figure 3A It is an explanatory diagram of the matrix data used in the calculation involved in Embodiment 1 of the present invention.

[0029] Figure 3B It is an explanatory diagram of the matrix data used in the calculation involved in Embodiment 1 of the present invention.

[0030] Figure 4 It is a graph showing the differential waveform pattern used in the calculation involved in Embodiment 1 of the present invention.

[0031] Figure 5A It is a graph showing the deviation of the differential waveform pattern extracted from the database used in the calculation involved in Embodiment 1 of the present invention.

[0032] Figure 5B It is a graph showing the standard deviation obtained based on the deviation of the differential waveform pattern extracted from the database used in the calculation involved in Embodiment 1 of the present invention.

[0033] Figure 6 It is a flowchart showing the order of calculating the remaining film thickness or etching amount of the film to be processed in the etching process involved in Embodiment 1 of the present invention.

[0034] Figure 7 It is a flowchart showing Figure 6 the detailed order of recipe optimization in step S603 of the flowchart.

[0035] Figure 8 It is a graph for explaining the effect of the etching process involved in Embodiment 1 of the present invention, showing the differential waveform pattern obtained by detecting the reflected light from the workpiece wafer and the differential waveform pattern stored in the database, indicating that by weighting the differential waveform pattern database, even when the fine shape of the workpiece deviates, the differential waveform pattern database measured using a test semiconductor wafer with a similar fine shape can be used to accurately detect the film thickness.

[0036] Figure 9 It is a diagram showing the number of interference light pattern data corresponding to each film thickness stored in the differential waveform pattern database set 14 involved in Embodiment 2 of the present invention.

[0037] Figure 10 It is a block diagram showing the schematic structure of the plasma processing apparatus involved in Embodiment 2 of the present invention.

[0038] Figure 11 It is a flowchart showing the processing sequence involved in Embodiment 2 of the present invention.

[0039] Figure 12 It is to show Figure 11 a detailed flowchart of step S1312 in the flowchart in

[0040] Figure 13 It is an explanatory diagram related to the effects of Embodiment 2 of the present invention, and it is a chart showing the deviation of the residual film thickness after processing for each processed wafer.

[0041] Explanation of reference numerals

[0042] 1, 110 Plasma processing device

[0043] 2 Vacuum processing chamber

[0044] 3 Plasma

[0045] 4 Workpiece to be processed

[0046] 5 Sample stage

[0047] 6 Interference light

[0048] 7 Light receiver

[0049] 8, 81 Etch amount measurement component

[0050] 9 Spectrometer

[0051] 10 First digital filter

[0052] 11 Differentiator

[0053] 12 Second digital filter

[0054] 13 Individual film thickness calculator

[0055] 14, 141 Differentiated waveform pattern database set

[0056] 15, 151 Differentiated waveform pattern database

[0057] 16 Weighted film thickness calculator

[0058] 17 Film thickness calculation formula

[0059] 18 Regression analyzer

[0060] 19 Endpoint determiner

[0061] 20 Display

[0062] 21 Formula optimizer

[0063] 22 Light source

[0064] 23 Additional processing time calculator

[0065] 40 Storage unit

[0066] 119 Film thickness estimation end point determiner Detailed implementation mode

[0067] In the present invention, an etching amount measurement component having a plurality of interference spectra with various film thicknesses and structures is equipped in a plasma processing apparatus. Based on the distance between the interference spectrum during etching and each database, the film thicknesses in the database are mixed by this etching amount measurement component to determine the film thickness estimation value. Thus, for deviations in film thickness and structure other than the substrate film thickness, high-precision film thickness estimation can also be achieved.

[0068] The following shows the specific sequence of estimating the film thickness in the present invention.

[0069] (a) Prepare a plurality of databases with different interference spectrum patterns caused by various film thicknesses and structures.

[0070] (b) Calculate the weight of the difference between the pattern following the interference spectrum during wafer processing and the pattern of the interference spectrum in the database, and use the weighted sum of the film thickness values in the database to calculate the film thickness estimation value.

[0071] (c) Use the estimated film thickness to determine whether the target is reached.

[0072] Here, if the calculation of the film thickness estimation value is described in more detail, the wavelength range of the interference spectrum is optimized using the data of the interference spectrum and the film thickness collected in advance. Specifically, the wavelength range with a large deviation in the interference spectrum at the same film thickness is excluded, and the accuracy is evaluated through mutual film thickness estimation between databases. If the accuracy is good, this wavelength range is used in the film thickness estimation.

[0073] In the present invention, in the plasma processing apparatus and the plasma processing method, three or more synthetic interference light patterns weighted corresponding to the difference value from the real interference light pattern during processing are used, and the film thicknesses calculated from the synthetic interference light patterns are synthesized according to the weights, thereby enabling the detection of the film thickness during processing.

[0074] In addition, considering the difference in the fine shape of the semiconductor wafer surface in the present invention, a database of differential waveform patterns of interference light obtained from multiple semiconductor wafers is registered. During the etching process, the time derivative is obtained for each of the multiple wavelengths of the interference light obtained from the surface of the semiconductor wafer, and the pattern of the differential value of the waveform of the interference light is obtained. The weight based on the difference between this pattern and the patterns in the multiple differential waveform pattern databases is calculated for each differential waveform pattern database.

[0075] The present invention relates to a plasma processing apparatus and a plasma processing method. By using this weight to calculate the weighted sum of the film thicknesses calculated from each differential waveform pattern database, it is possible to accurately detect the remaining film thickness of the film to be processed by using a pattern database that is more similar to the fine shape of the semiconductor wafer of the object to be processed.

[0076] Embodiments of the present invention will be described in detail below with reference to the accompanying drawings. In all the drawings used to illustrate the present embodiments, elements having the same function are denoted by the same reference numerals, and duplicate descriptions thereof are omitted in principle.

[0077] However, the present invention is not construed as being limited to the description of the embodiments shown below. Those skilled in the art can easily understand that the specific structure can be changed without departing from the spirit or gist of the present invention.

[0078]

Example 1

[0079] The following uses Figure 1 FIG. 5 to illustrate the overall structure of a plasma processing apparatus for a semiconductor wafer having a unit for detecting the etching amount (here, the etching depth and film thickness of an actual processed part) of the present invention.

[0080] In Figure 1 FIG. 1 shows a plasma processing apparatus according to an embodiment of the present invention. The plasma processing apparatus 1 includes a vacuum processing chamber 2, an etching amount measurement assembly 8, and a control unit 30.

[0081] The vacuum processing chamber 2 includes: a sample stage 5 on which a workpiece 4 such as a semiconductor wafer is placed inside; a light receiver 7 that detects the light emission of the plasma 3 generated inside; an optical fiber 71 that transmits the light emission of the plasma 3 received by the light receiver 7; and a light source 22 that irradiates the workpiece 4 with light. The vacuum processing chamber 2 also includes a gas introduction unit for introducing gas into the interior, a vacuum exhaust unit for evacuating the interior to a vacuum, a power supply unit for supplying power, etc., but their illustrations are omitted for simplicity of explanation.

[0082] The etching amount measurement assembly 8 includes: a spectroscope 9; a first digital filter 10; a differentiator 11; a second digital filter 12; a single film thickness calculator 13; a differential waveform pattern database set 14; a differential waveform pattern database 15; a weighted film thickness calculator 16 that calculates the film thickness 210 of the film to be processed; a film thickness calculation formula 17 used in the calculation of the weighted film thickness calculator 16; a regression analyzer 18; an endpoint determiner 19 that determines the end of etching based on the result of the regression analyzer 18; a display 20 that displays the determination result of the endpoint determiner 19; and a formula optimizer 21 that optimizes the value of the film thickness calculation formula 17.

[0083] In addition, Figure 1The etch amount measurement component 8 in shows a functional structure. The actual structure of the etch amount measurement component 8, excluding the display 20 and the spectroscope 9, can be composed of the following elements: a CPU; a storage device composed of a ROM that holds various data such as an etch depth and film thickness detection processing program, a differential waveform pattern database of the interference light 6, a RAM for holding detection data, and an external storage device; a data input / output device; and a communication control device.

[0084] The control unit 30 receives signals from the etch amount measurement component 8 and external signals, and controls a gas introduction unit (not shown), a vacuum exhaust unit, a power supply unit, etc. that are connected to the vacuum processing chamber 2.

[0085] The etch gas introduced into the interior of the vacuum processing chamber 2 from the gas introduction unit (not shown) is decomposed by microwave power supplied from the power supply unit (not shown), etc., to become plasma 3, and the workpiece 4 such as a semiconductor wafer on the sample stage 5 is etched by this plasma 3.

[0086] The light emission of the plasma 3 is directly received by the light receiver 7 or, like the interference light 6, after being reflected by the workpiece 4 such as a semiconductor wafer, a part of it is received by the light receiver 7 and is introduced into the spectroscope 9 through the optical fiber 71 from the vacuum processing chamber 2. In the spectroscope 9, the incident plasma light emission is spectroscopically analyzed, and the light intensity is converted into a digital signal. Additionally, instead of the light emission of the plasma 3, light can be irradiated onto the workpiece 4 from the light source 22, and its reflected light can be measured in the spectroscope 9.

[0087] Figure 2A represents the fine shape of the surface of the workpiece 4 to be etched. For example, on a silicon substrate 200, a workpiece 4 such as a semiconductor wafer is formed by laminating a processed film 202 made of polysilicon, etc. and a base film 203 made of an oxide film, etc. Additionally, a pattern of a mask 201 made of a resist, etc. is formed on the processed film 202.

[0088] Figure 2B is a schematic diagram when etching the processed film 202. The plasma light 204 incident on the workpiece 4 is reflected on the surface of the workpiece 4. However, among them, first, there is the reflected light 205 at the surface of the mask 201 and the reflected light 206 at the boundary between the mask 201 and the processed film 202. Additionally, at the part where the processed film 202 is exposed without being covered by the mask 201, there are the reflected light 207 at the surface of the processed film 202, the reflected light 208 at the boundary between the processed film 202 and the base film 203, and the reflected light 209 at the boundary between the base film 203 and the silicon substrate 200.

[0089] Among these reflected lights, interference light is formed due to the optical path difference. As the etching progresses, the film thickness of the film to be processed 202 decreases, so the optical path difference of each reflected light changes, and an interference phenomenon with different periods for each wavelength occurs. Among these multi-wavelength interference lights 6, the multi-wavelength interference light 6 received by the light receiver 7 is guided to the spectroscope 9 of the etching amount measurement component 8 via the optical fiber 71, and the etching amount of the film to be processed 202 is measured and the end point of the process (etching here) is determined based on its state.

[0090] In Figure 2B it, the film thickness to be the object of film thickness detection is the film thickness 210 of the film to be processed 202. However, the multi-wavelength interference light 6 received by the light receiver 7 and measured by the spectroscope 9 also varies according to the mask film thickness 211, the substrate film thickness 212, the area ratio of the film to be processed area 213 to the mask area 214, and the deviation of the film thickness 210 of the film to be processed at each position on the semiconductor wafer, such as the deviation of the fine shape of the semiconductor wafer. These deviations become the error factors for the detection of the film thickness 210 of the film to be processed.

[0091] The multi-wavelength interference light 6 related to the workpiece 4 taken into the spectroscope 9 becomes a current detection signal corresponding to the light emission intensity of each of the respective wavelengths, and is converted into a voltage signal. A plurality of signals of specific wavelengths output as sampling signals obtained by the spectroscope 9 at an arbitrary sampling time i are used as time series data yij, and are temporarily stored in a storage device such as a RAM (not shown). Here, j represents the wavelength.

[0092] Next, the time series data yij output from the spectroscope 9 and temporarily stored in a storage device such as a RAM is transmitted to the first digital filter 10, and waveforms above a given frequency that become noise components are removed for smoothing processing, and are temporarily stored in a storage device such as a RAM (not shown) as smoothed time series data Yij.

[0093] The smoothed time series data Yij temporarily stored in this RAM or other storage device is sent to the differentiator 11, and time series data dij of differential values (first-order differential values or second-order differential values) at a given sampling time i is calculated and stored in a storage device such as a RAM (not shown). The time series data dij of the differential values temporarily stored in this RAM or other storage device is sent to the second digital filter 12, and smoothing processing is performed again, and is stored in a storage device such as a RAM (not shown) as smoothed differential coefficient time series data Dij.

[0094] Here, the calculation of the smoothed differential coefficient time series data Di is described. As the first digital filter 10, for example, a second-order Butterworth type low-pass filter is used. With the second-order Butterworth type low-pass filter, the smoothed time series data Yi is obtained by Equation (1).

[0095]

Mathematical Formula 1

[0096]

[0097] Here, the coefficients a and b have different numerical values according to the sampling frequency and the cut-off frequency.

[0098] In addition, the coefficient values of the digital filter are, for example, a2 = -1.143, a3 = 0.4128, b1 = 0.067455, b2 = -0.013491, b3 = 0.067455 (sampling frequency 10 Hz, cut-off frequency 1 Hz).

[0099] Using the polynomial fitting smoothing differentiation method of the 5-point time series data Yi by the differentiator 11, the time series data di of the differential coefficient is calculated as follows according to Equation (2).

[0100]

Mathematical Formula 2

[0101]

[0102] Here, regarding the weight coefficient ω, ω-2 = 2, ω-1 = -1, ω0 = -2, ω1 = -1, ω2 = 2.

[0103] Using the time series data di of the differential coefficient, as the second digital filter 12, the smoothed differential coefficient time series data Di is calculated as follows according to Equation (3) by a second-order Butterworth low-pass filter.

[0104]

Mathematical Formula 3

[0105]

[0106] Performing this calculation for each wavelength j, the smoothed differential coefficient time series data Dij can be obtained. Further, the value obtained by dividing the smoothed differential coefficient time series data Dij by the smoothed time series data Yij is set as Dij and used in subsequent calculations. In addition, the smoothed differential coefficient time series data Dij can also be used as it is.

[0107] On the other hand, the differential waveform pattern database set 14 holds three or more differential waveform pattern databases 15. In the differential waveform pattern database 15, regarding the workpiece 4 to be processed and the film structure on its surface, which are objects to be processed for manufacturing semiconductor devices, data P(m)sj of the interference light pattern obtained in the case of etching a test workpiece with the same material, shape, and structure under the same conditions as the workpiece 4 is stored in advance.

[0108] The differential waveform pattern database 15 stores multiple results measured in different test workpieces to be processed. m represents the ID of the database, s represents the elapsed time at sampling counted from the start of processing, and j represents the emission wavelength. The data P(m)sj of the interference light pattern includes the pattern of the intensity of the interference light from the workpiece to be processed corresponding to the different remaining film thicknesses of the workpiece to be processed or the values representing the same.

[0109] In addition, when etching also progresses laterally, parameters representing the width of the workpiece to be processed region 213 and the width of the mask region 214 can be included instead of the remaining film thickness. The differential waveform pattern database 15 is stored in a storage device such as a RAM or ROM (not shown) inside the etching amount measurement component 8, a hard disk, a DVD disc, or other storage devices.

[0110] The differential waveform pattern database 15 stored in the differential waveform pattern database set 14 is obtained from workpieces to be processed with slightly different micro shapes, such as when the mask 201 is thicker or thinner due to manufacturing deviations, etc. In addition, workpieces to be processed with slightly different micro shapes for testing can be prepared to create the differential waveform pattern database 15.

[0111] The individual film thickness calculator 13 is a process of extracting the remaining film thickness of the workpiece to be processed and the data P(m)sj of the interference light pattern from the previously presented differential waveform pattern database 15. For example, for each of the above differential waveform pattern databases 15, data where s is greater than or equal to a given elapsed time and the data of the remaining film thickness corresponding to that time can be extracted.

[0112] In addition, for each of the above differential waveform pattern databases 15, it can be compared with the true pattern Dij of the interference light corresponding to a given elapsed time, and the elapsed time with the smallest pattern difference and the film thickness value at this time can be detected. That is, the difference between the pattern data P(m)sj stored in the differential waveform pattern database 15 and the true pattern Dij can also be calculated, the pattern data with the smallest difference value can be obtained, and the remaining film thickness corresponding to the pattern data can be extracted.

[0113] The data of the interference light pattern extracted in this way is called Q(m)sj. The remaining film thickness during this elapsed time is associated with this data. The associated remaining film thickness is called r(m)s.

[0114] In the weighted film thickness calculator 16, the data Q(m)sj of the interference light pattern and the data r(m)s of the remaining film thickness extracted from each database are used to calculate the value of the instantaneous film thickness value Zi at time i. To calculate the instantaneous film thickness value Zi, here, the Figure 3A shown matrix R: 301 and Figure 3B shown matrix Q: 302 are created.

[0115] Figure 3A The matrix R: 301 is a matrix obtained by successively combining r(m)s in the row direction starting from m = 1. Hereinafter, the elements in the u-th row are represented by Ru. Figure 3B The matrix Q: 302 is a matrix obtained by successively combining Q(m)sj in the row direction starting from m = 1. Hereinafter, the element in the u-th row and v-th column is represented by Quv. Ru and Quv respectively correspond to the same elapsed time in the same database. In addition, N is used as the number of rows hereinafter.

[0116] The value of the instantaneous film thickness Zi is calculated by the following equations (4) and (5).

[0117]

Mathematical formula 4

[0118]

[0119] The above is the operation formula of the matrix. A and W are the following matrices respectively. T represents transpose.

[0120] A: An N×N matrix for correction. Each element can be an N×N diagonal matrix that is the reciprocal of the sum of the elements of W. Alternatively, it can be the inverse matrix of (K - λI) as in kernel ridge regression. Here, K is an N×N matrix whose element in the u-th row and v-th column is kuv represented by the following equation (5). λ is an arbitrary coefficient, and I is an N×N diagonal matrix with elements of 1.

[0121]

Mathematical formula 5

[0122]

[0123] In the above equation (5), the wavelength range and coefficient σ for taking the sum of j are specified by the values stored in the film thickness calculation recipe 17. exp is the exponential function with the base of the natural logarithm.

[0124] W: Each element represents the weight of the difference between the smoothed differential value time series data Dij following the time i and the pattern of each database. For example, the u-th element Wu is calculated by a function that monotonically decreases according to the magnitude of the pattern difference as shown in the following equation (6).

[0125]

Mathematical formula 6

[0126]

[0127] In the above equation (6), the wavelength range and coefficient σ for taking the sum of j are specified by the values stored in the film thickness calculation recipe 17 in the same way as in equation (5).

[0128] By using the above formula (6), a large value of weight is obtained for the case where the differential waveform pattern of the workpiece 4 at time i is more similar to the differential waveform patterns of each database, and a small weight is obtained for the dissimilar case. By applying such weights to formula (4), the instantaneous film thickness value Zi is calculated based on the remaining film thickness of the database with a similar differential waveform pattern.

[0129] For example, as Figure 4 shown, when the differential waveform pattern 410 of the workpiece 4 and the differential waveform patterns 401 - 403 of each database are obtained, large weighting is applied to the differential waveform pattern 401 of DB1 and the differential waveform pattern 402 of DB2, and small weighting is applied to the differential waveform pattern 403 of DB3. The instantaneous film thickness value Zi can be calculated based on the differential waveform patterns 401 of DB1 and 402 of DB2 that are closer in the differential waveform pattern.

[0130] Since if the fine shape on the surface of the workpiece 4 is similar, the differential waveform pattern of the interference light also takes a similar pattern, the instantaneous film thickness value Zi can be calculated using a database with a fine shape on the surface similar to that of the workpiece 4. In addition, the calculation formula for the weight Wu is not limited to formula (6), as long as it is a function that calculates a small weight when the pattern difference is large.

[0131] The instantaneous film thickness value at this sampling time is detected as Zi, and the value of the instantaneous film thickness value Zi is stored as time - series data in the storage device within the etching amount measurement component 8.

[0132] The film thickness calculation recipe 17 specifies the wavelength range for taking the sum in formulas (5) and (6) and the coefficient σ in the formula. This can be determined by the designer or set by the recipe optimizer 21 described later.

[0133] In the regression analyzer 18, the output from the weighted film thickness calculator 16 is received, or the data of the instantaneous film thickness Zi at this sampling time i stored in the storage device is read out, and the instantaneous film thickness values before time i are read out from the storage device. A regression analysis is performed using these, and the film thickness value at time i is calculated based on the result of the regression line approximation.

[0134] That is, the first - order regression line Y = Xa·t + Xb (Y: remaining film amount, t: etching time, Xa: absolute value is the etching rate, Xb: initial film thickness) is obtained by the regression analyzer 18, and the film thickness Yi (calculated film thickness) value at the sampling time i is calculated based on this regression line. In addition, when the desired remaining film thickness of the workpiece film is smaller than the remaining film thickness in the differential waveform pattern database 15, the calculation of the instantaneous film thickness Zi may not be performed, and only the first - order regression line is used to calculate the film thickness value at time i.

[0135] Next, the data representing the value of the calculated film thickness Yi is sent to the end point determiner 19. In the end point determiner 19, the value of the film thickness Yi is compared with the value of the film thickness (target film thickness) that is the target of the etching process. When it is determined that the film thickness Yi is equal to or less than the target film thickness value, it is assumed that the etching amount of the etched film of the workpiece 4 has reached the target, and the result is displayed on the display 20.

[0136] After that, the generation of the electric field or magnetic field in the plasma formation section is stopped, the plasma 3 disappears, and the etching process of the workpiece 4 ends. Furthermore, the processing conditions such as the gas and pressure for the etching process are changed to perform the processing of the etched film.

[0137] The recipe optimizer 21 performs the process of setting the film thickness calculation recipe 17 described above. This is implemented as a pretreatment before the start of the plasma processing of the workpiece 4. The recipe optimizer 21 extracts the differential waveform patterns in the specified remaining film thickness (for example, the target film thickness) from each database in the differential waveform pattern database 15. The differential waveform pattern 501 of DB1, the differential waveform pattern 502 of DB2, and the differential waveform pattern 503 of DB3 extracted are Figure 5A .

[0138] As Figure 5B shown, the standard deviation 510, which is the deviation of the intensity at each wavelength of the extracted differential waveform pattern, is calculated. For wavelengths with a large deviation, since it is assumed that the intensity changes due to factors different from the remaining film thickness, such as the state of the plasma and the chamber, by excluding them, the accuracy can be improved.

[0139] Therefore, in Figure 5B , the wavelengths with a large standard deviation such as 511 are excluded from the wavelengths in the summation of Equation (4) and Equation (5), and the wavelength range is stored in the film thickness calculation recipe 17 so that only the wavelength range of 512 is used to calculate the instantaneous film thickness Zi.

[0140] In addition, the excluded wavelength range can be determined in terms of relative size, such as the upper 10% or 20% of the wavelengths with the largest deviation among all wavelengths, or a threshold can be determined for the deviation. In addition, when the differential waveform patterns in the database are divided into multiple groups, the standard deviation for each group can also be calculated and its average can be used.

[0141] Furthermore, in order to determine the proportion of the wavelength range to be excluded, the recipe optimizer 21 selects one database (DBp) from the differential waveform pattern database 15 and uses the remaining differential waveform pattern database 15 to perform the process of calculating the instantaneous film thickness using Equation (4).

[0142] For the cases of excluding candidates in multiple wavelength domains, such as excluding the top 10% deviation and excluding the top 20% deviation respectively, the instantaneous film thickness Zs at the moment s of DBp is calculated, the difference from the film thickness r (m) s is calculated, and the wavelength domain with a smaller difference is selected as the optimal wavelength domain.

[0143] In addition, for the coefficient σ in formulas (5) and (6), multiple coefficients σ can also be set and the instantaneous film thickness Zs can be calculated, and the process of determining the combination with a small difference is performed for the combination of the wavelength domain and the coefficient σ. The wavelength domain and the coefficient σ determined in this way are stored in the film thickness calculation recipe 17.

[0144] In this embodiment, by using formula (6), the film thickness is calculated as the weighted sum of multiple databases with a small difference from the differential waveform pattern obtained from the workpiece 4 among multiple databases. Since a more similar differential waveform pattern can be obtained from workpieces with similar micro shapes, even when the micro shape of the workpiece 4 deviates, the database of workpieces with similar micro shapes can be used to perform accurate endpoint determination.

[0145] Next, use Figure 6 The flowchart of Figure 1 is used to illustrate the order of calculating the remaining film thickness or etching amount of the film to be processed by the etching amount measurement component 8 during the etching process. Figure 6 It is a flowchart showing the process of detecting Figure 1 the remaining film thickness etching amount of the plasma processing apparatus according to the embodiment shown. It mainly shows the process of the operation of the etching amount measurement component 8. The process starts from step S601.

[0146] In this embodiment, before processing the workpiece 4, the target remaining film thickness value of the film to be processed and the setting of multiple differential waveform pattern databases used in its detection or determination are performed (step S602).

[0147] In the differential waveform pattern database, the data P (m) sj obtained by collecting data of the interference light pattern for multiple workpieces is used, where the data of the interference light pattern is obtained when the test workpiece with the same material, shape, and structure is etched under the same conditions as the workpiece 4 and its surface film structure, which is the workpiece to be processed for manufacturing semiconductor devices.

[0148] Next, the process of optimizing the wavelength domain and coefficient used in the calculation of the instantaneous film thickness is performed (step S603). This process will be described in the Figure 7 flowchart described later. In addition, this process may not be executed and pre-specified wavelength domain and coefficient may be used.

[0149] Next, a plasma 3 is formed in the vacuum processing chamber 2 to start the processing of the etched film of the workpiece 4 to be processed, and interference light obtained from the etched film is detected at given sampling intervals (e.g., 0.1 to 0.5 seconds) during the etching process (step S604). At this time, a sampling start command is issued along with the start of the etching process.

[0150] During the processing, the intensities of multi-wavelength interference light that change as the etching progresses are transmitted to the spectroscope 9 of the etching amount measurement assembly 8, and the light detector detects, for each given frequency, a light detection signal that is a voltage corresponding to the light intensity and outputs it.

[0151] The light detection signal of the spectroscope 9 is digitally transformed to obtain a sampling signal yij that is a data signal corresponding to an arbitrary time. Next, the multi-wavelength output signal yij from the spectroscope 9 is smoothed by the first-stage first digital filter 10 to calculate the time series data Yij at an arbitrary time (step S605).

[0152] Next, the time series data Yij is transmitted to the differentiator 11, and the differential coefficient dij of the time series is calculated by the polynomial fitting smoothing differentiation method (step S606). That is, the differential coefficient di of the signal waveform is detected by the polynomial fitting smoothing differentiation method.

[0153] The differential coefficient dij is transmitted to the second-stage second digital filter 12 to calculate the smoothed differential coefficient time series data Dij (step S607). The obtained smoothed differential coefficient time series data Dij is divided by the smoothed time series data Yij and transmitted to the individual film thickness calculator 13.

[0154] In addition, here, the smoothed differential coefficient time series data Dij is used, but any value can be used if it is time series data such as Yij itself or a value calculated from Yij by the least squares method that reflects the differences of the workpiece 4 to be processed.

[0155] In the individual film thickness calculator 13, data Q(m)sj of the remaining film thickness of the film to be processed and the pattern of the interference light are respectively extracted from the multiple differential waveform pattern databases 15 in the differential waveform pattern database set 14 (step S608).

[0156] For example, for each of the above differential waveform pattern databases 15, data where s is longer than a given elapsed time and the remaining film thickness data corresponding to that time are extracted.

[0157] Alternatively, the elapsed time from the start of this etching process can be obtained, and data where s is within a given range (e.g., ±10 seconds, etc.) from that elapsed time and the remaining film thickness data corresponding to that time are extracted.

[0158] Alternatively, it is also possible to compare each of the differential waveform pattern databases 15 with the true pattern Dij of the interference light corresponding to a given elapsed time, and extract the elapsed time with the smallest pattern difference and the remaining film thickness at this time.

[0159] Next, in the weighted film thickness calculator 16, the value of the instantaneous film thickness value Zi at time i is calculated using the data Q(m)sj of the pattern of the interference light and the data r(m)s of the remaining film thickness extracted from each database (step S609).

[0160] For calculating the instantaneous film thickness value Zi, for the patterns extracted from each differential waveform pattern database 15, a matrix Q combining the data Q(m)sj and a matrix R combining the similarly extracted remaining film thickness Ru are created.

[0161] The value of the instantaneous film thickness value Zi is calculated by substituting Q, R, and the smoothed differential value time series data Dij at time i into the aforementioned equations (4), (5), and (6). In addition, the wavelength range and coefficient σ for taking the sum in equations (5) and (6) use pre-determined values or values determined by the process shown later Figure 7 The values determined by the process.

[0162] Next, the regression analyzer 18 uses the calculated instantaneous film thickness value Zi and the instantaneous film thickness Zi at the sampling time i stored in the storage device to obtain a first-order regression line, and calculates the calculated film thickness based on this first-order regression line (step S610).

[0163] Furthermore, the calculated film thickness of the currently processed film is compared with the target residual film thickness set in step S302. If it is determined to be below the target residual film thickness, it is determined that the target has been reached, and a signal to end the etching process is sent to the plasma processing apparatus 1 (step S611). In the case where it is determined that the target has not been reached, the process returns to step S305. If it is determined that the target has been reached, the sampling end setting is finally performed (step S612).

[0164] Next, using Figure 7 The flowchart of, to illustrate the process sequence during the recipe optimization process performed by the etching amount measurement component 8 corresponding to S603 of Figure 6 In Figure 1 The process starts from step S701.

[0165] First, the recipe optimizer 21 determines the remaining film thickness (reference film thickness) for comparing the differential waveform patterns. This can be, for example, the target residual film thickness (step S702)

[0166] Next, in each differential waveform pattern database 15, the data P(m)sj of the differential waveform pattern at the reference film thickness is extracted (step S703).

[0167] Next, the standard deviation of P(m)sj for each wavelength j is calculated using the data P(m)sj of the extracted pattern (step S704).

[0168] Next, starting from the wavelength with a relatively large standard deviation, exclusions are made in sequence, and the remaining wavelength range is set as a candidate for the wavelength range to be utilized (step S705). Here, multiple candidates are created, such as the case of excluding 10% and the case of excluding 20%.

[0169] Next, multiple candidates for the coefficient σ of formulas (5) and (6) are created (step S706).

[0170] Then, one database of the differential waveform pattern database 15 is selected, and using the remaining differential waveform pattern databases 15, the instantaneous film thickness is calculated using formula (4), and the process of calculating the error from the remaining film thickness data r(m)s is performed (step S707). This process is performed for the combination of the wavelength range and the coefficient σ, and the combination with a small error is determined.

[0171] As described above, the determined wavelength range and coefficient σ are stored in the film thickness calculation recipe 17 (step S708).

[0172] Through the above, the process is ended (S709).

[0173] Use Figure 8 to illustrate the effects of this embodiment. Here, as an example, the differential waveform patterns 810 and 2 differential waveform pattern databases 15 obtained by detecting the reflected light from the workpiece 4 are cited. Figure 8 The differential waveform pattern 811 of DB4 in is as follows: compared with the workpiece 4, the processed film region 213 of the test semiconductor wafer (the test workpiece) of the differential waveform pattern database 15 is smaller, and the mask film thickness 211 is thicker. The differential waveform pattern 812 of DB5 is as follows: the processed film region 213 of the test semiconductor wafer of the differential waveform pattern database 15 is of the same degree as that of the workpiece 4, and the mask film thickness 211 is also of the same degree.

[0174] In addition, for Figure 8 the wavelength range 802 mainly measures the temporal change of the interference light of the processed film 202. For the wavelength range 801, not only the temporal change of the interference light of the processed film 202 is measured, but also the sum of the temporal change of the interference light of the mask 201 thinner than the processed film is measured.

[0175] Here, consider the case where the differential waveform pattern 811 of the DB4 is compared with the differential waveform pattern 810 of the workpiece 4, and the film thickness when the difference between the two is small is used in the instantaneous film thickness calculation. First, in the DB4, since the processed film region 213 of the workpiece 4 is small, the temporal change of the interference light of the processed film 202 measured is small, and the amplitude of the differential waveform pattern in the wavelength domain 802 becomes smaller than that of the workpiece 4. Therefore, if the differential waveform pattern 811 of the DB4 is used, the sizes of the differential waveform patterns representing the interference light of the processed film 202 do not match, and the calculation accuracy of the difference between the differential waveform patterns is reduced.

[0176] In addition, since a differential waveform pattern with a small difference in the temporal change (wavelength domain 801) of the interference light of the mask film thickness 211 is also selected, a differential waveform pattern 811 in a state where the mask film thickness 211 is close, that is, a state where etching has progressed more in the DB4 (a state where the processed film 202 is thin), is selected, and an error occurs in the instantaneous film thickness calculation.

[0177] In this way, it can be seen that if the differential waveform pattern database 15 of the test semiconductor wafer with different fine shapes on the workpiece 4 is used, the calculation accuracy of the remaining film thickness is reduced. On the other hand, in the workpiece 4 with similar fine shapes, like the differential waveform pattern 812 of the DB5, the difference between the differential waveform patterns becomes smaller.

[0178] Therefore, by weighting the differential waveform pattern database 15 using the difference between the differential waveform patterns, even when the fine shape of the workpiece 4 deviates, the differential waveform pattern database 15 measured with a test semiconductor wafer having a similar fine shape can be used in each workpiece 4 to accurately detect the film thickness.

[0179] According to this embodiment, high-precision film thickness estimation and endpoint determination can be achieved for various structural deviations between wafers, batches, etc., and the yield of device manufacturing can be improved.

[0180]

Embodiment 2

[0181] In this embodiment, consider the following situation: In the past mass production process, regarding the film structure on the surface, data obtained when etching a plurality of workpieces (wafers) having the same material, shape, and structure as the workpiece 4 under the same conditions as the workpiece 4 is used to prepare databases and their sets equivalent to the differential waveform pattern database 15 and the differential waveform pattern database set 14 described in Embodiment 1.

[0182] In this embodiment, for 200 wafers whose film structures on the surface are the same as the workpiece 4 in terms of material, shape, and structure, the target remaining film thickness of the film to be processed (set to 140 nm here) was set, and an etching process was performed. Then, using the data P(m)sj of the interference light pattern obtained through this etching process and the data of the film thickness values of the film to be processed corresponding to the data P(m)sj of each interference light pattern, a differential waveform pattern database set 141 including a differential waveform pattern database 151 of these 200 wafers was created.

[0183] The film to be processed on these 200 wafers was supposed to be etched to a preset target remaining film thickness (140 nm). However, in reality, due to the change over time of the environment in the etching process chamber and the differences in the fine shapes of the surfaces of the workpieces, the etching conditions for the film to be processed on each wafer changed, and the remaining film thickness of the film to be processed after the etching process deviated in each wafer. That is, wafers that ended the etching process with a remaining film thickness thinner than the target and wafers that ended the etching process with a remaining film thickness thicker than the target were mixed.

[0184] Figure 9 In the chart 900, for wafers with the same remaining film thickness after each etching process, the legend patterns are unified, and the number of interference light pattern data P(m)sj corresponding to each film thickness stored in the differential waveform pattern database set 141 created using the data obtained from these 200 wafers is shown in the form of stacked bars.

[0185] It can be seen that until the remaining film thickness (film thickness of the film to be processed) shown in data 901 reaches 145 nm, there are 200 data, the same number as the total number of wafers, for the data of the interference light pattern data P(m)sj. However, if the remaining film thickness (film thickness of the film to be processed) becomes less than 144 nm, it gradually decreases. In the case where the target remaining film thickness shown in data 902 is 140 nm, the number of data is only 108. In the mass production process of etching the actual product wafers, a situation where the number of data near the target remaining film thickness decreases like this is also anticipated.

[0186] Try to consider applying such a situation to the Figure 8 chart described in Embodiment 1. For example, it can be considered that the wafer corresponding to DB5:812 ended the etching with a remaining film thickness thicker than the target, and at the target remaining film thickness, there is no interference light pattern data for DB5. In such a case, when selecting the interference light pattern in a state where the etching has progressed more (when the film to be processed 202 is thinner) in DB4:811, an error occurs in the calculation of the instantaneous film thickness value Zi.

[0187] Thus, it can be considered that the accuracy of determining the end point of the etching process deteriorates when it is determined that there is no data P(m)sj of an interference pattern corresponding to the target remaining film thickness in some or all of the differential waveform pattern databases 15 included in the differential waveform pattern database set 14 described in Embodiment 1.

[0188] In this embodiment, a method for accurately determining the end point of the etching process even when there is a shortage in the interference light pattern data corresponding to the film thickness near the target remaining film thickness as described above is described.

[0189] In Figure 10 FIG. shows a plasma processing apparatus 110 according to Embodiment 2. Different from Figure 1 the plasma processing apparatus 1 according to Embodiment 1 shown, in the etching amount measurement component 81, in addition to the above differential waveform pattern database 151 and differential waveform pattern database set 141, instead of the end point determination unit 19, it has a film thickness estimation end point determination unit 119, a storage unit 40 for storing data of the calculated film thickness values at past times during the process, and an additional processing time calculator 23. Since the other structures are the same as those of Figure 1 the plasma processing apparatus 1 of Embodiment 1 shown and perform the same functions, detailed description thereof is omitted.

[0190] In the plasma processing apparatus 110 of Embodiment 2, in the same manner as in Embodiment 1, the method of calculating the instantaneous film thickness value Zi according to Equations (4), (5), and (6) is used, and the etching process is performed until a given remaining film thickness with an acceptable accuracy is obtained. After detecting that the given remaining film thickness is reached, the additional etching time required to reach the target remaining film thickness is calculated using the etching rate calculated from the remaining film thickness data detected at past times. After continuing the etching process for the calculated time, the etching process is ended.

[0191] Figure 11 represents the detection Figure 10 FIG. shows a flowchart of the operation of determining the remaining film thickness or etching amount of the film to be processed by the plasma processing apparatus 110 according to Embodiment 2 and determining the end point of the etching process. It mainly shows the operation flow of the etching amount measurement component 81. The process starts from step S1301.

[0192] In this embodiment, the following settings are made before processing the workpiece 4: setting the value of the target remaining film thickness of the film to be processed (the remaining film thickness at the end point of etching); setting the value of the remaining film thickness (the remaining film thickness at the end point of film thickness estimation) at which the calculation of the instantaneous film thickness value Zi using the aforementioned equations (4), (5), and (6) is completed; setting a plurality of differential waveform pattern databases used in the calculation of the instantaneous film thickness value Zi using the aforementioned equations (4), (5), and (6); setting the time range extracted from the data of the calculated film thickness values at each moment in the etching process stored in the storage device as the data of the calculated film thickness values used to obtain the first-order regression line in step S1312 (step S1302). Here, the remaining film thickness at the end point of etching is a film thickness thinner than the remaining film thickness at the end point of film thickness estimation.

[0193] In this embodiment, the remaining film thickness at the end point of film thickness estimation is set to the thinnest film thickness within the remaining film thickness range where the calculation accuracy of the instantaneous film thickness value Zi is within the allowable range. In this embodiment, this film thickness is set to 144 nm, and the remaining film thickness value at the end point of film thickness estimation is set to 144 nm.

[0194] In addition, the remaining film thickness value at the end point of etching in this embodiment is set to be equivalent to Figure 9 140 nm of the data 902 shown in the graph 900. In addition, the time range extracted from the data of the calculated film thickness values at each moment in the etching process stored in the storage device as the data of the calculated film thickness values used to obtain the first-order regression line in step S1312 is set to 5.0 seconds from 5.0 seconds before the end point of film thickness estimation is reached to the time when the end point of film thickness estimation is reached.

[0195] In the differential waveform pattern database 151, the interference light pattern data P(m)sj obtained when etching processes are performed on three or more wafers equivalent to the workpiece 4 in terms of the film structure, material, shape, and structure on the surface, and the film thickness value data corresponding to each interference light pattern data P(m)sj are used.

[0196] Next, a process of optimizing the wavelength range and coefficients used in the calculation of the instantaneous film thickness, that is, recipe optimization, is performed (step S1303). Regarding this process, it is the same as the process described in the flowchart in Embodiment 1 using Figure 7 In addition, instead of performing this process, a pre-specified wavelength range and coefficients may be used.

[0197] Next, a plasma 3 is formed in the vacuum processing chamber 2 to start the processing of the etched film of the workpiece 4. In this etching process, the interferometric light obtained from the film to be processed is detected by the light receiver 7 at every given sampling interval (for example, 0.1 to 0.5 seconds) (step S1304). At this time, a sampling start command is issued along with the start of the etching process.

[0198] During the processing, the intensities of the multi-wavelength interferometric light that change as the etching progresses are transmitted to the spectroscope 9 of the etching amount measurement component 81. The spectroscope 9 detects light detection signals corresponding to the light intensities as voltages at each given frequency and outputs them.

[0199] The light detection signals of the spectroscope 9 are digitally transformed to obtain a sampling signal yij that is a data signal corresponding to an arbitrary time. Next, the multi-wavelength output signal yij from the spectroscope 9 is smoothed by the first digital filter 10 to calculate the time series data Yij at an arbitrary time (step S1305).

[0200] Next, the time series data Yij is conveyed to the differentiator 11, and the differential coefficient dij of the time series is calculated by the polynomial fitting smoothing differentiation method (step S1306). That is, the differential coefficient di of the signal waveform is detected by the polynomial fitting smoothing differentiation method.

[0201] The differential coefficient dij is passed to the second digital filter 12 to calculate the smoothed differential coefficient time series data Dij (step S1307). The obtained smoothed differential coefficient time series data Dij is divided by the smoothed time series data Yij and conveyed to the individual film thickness calculator 13.

[0202] In addition, although the smoothed differential coefficient time series data Dij is used here, any value can be used as long as it is the time series data such as Yij itself or the value calculated by using the least squares method for Yij, which reflects the differences in the atmosphere in the processing chamber.

[0203] In the individual film thickness calculator 13, data Q(m)sj of the remaining film thickness of the film to be processed and the pattern of the interferometric light are respectively extracted from the multiple differential waveform pattern databases 151 in the differential waveform pattern database set 141 (step S1308).

[0204] Next, in the weighted film thickness calculator 16, the data Q(m)sj of the pattern of the interferometric light and the data r(m)s of the remaining film thickness extracted for each database are used to calculate the value of the instantaneous film thickness Zi at time i (step S1309).

[0205] In order to calculate the instantaneous film thickness value Zi, for the patterns extracted from each differential waveform pattern database 151, a matrix Q that combines the data Q(m)sj and a matrix R that combines the remaining film thickness Ru extracted in the same way are created.

[0206] The value of the instantaneous film thickness value Zi is calculated by substituting Q, R, and the smoothed differential value time series data Dij at time i into the aforementioned formulas (4), (5), and (6). Additionally, the wavelength range and coefficient σ for taking the sum in formulas (5) and (6) use predetermined values or values determined through the process shown below. Figure 7 The values determined through the process.

[0207] Next, the regression analyzer 18 uses the calculated instantaneous film thickness value Zi and the instantaneous film thickness Zi at the sampling time i stored in the storage device to obtain a first-order regression line, and calculates the calculated film thickness according to this first-order regression line (step S1310).

[0208] Next, in the film thickness estimation end point determiner 119, the calculated film thickness of the currently processed film calculated in step S1310 is compared with the remaining film thickness at the end point of film thickness estimation set in step S1302 (step S1311). If it is determined that the remaining film thickness is less than or equal to the remaining film thickness at the end point of film thickness estimation, it is determined that the end point of film thickness estimation has been reached (step S1311 "Yes"), and the process proceeds to step S1312. In the case where it is determined that the end point has not been reached (step S1311 "No"), the process returns to the process of step S1305.

[0209] Next, Figure 12 Details of step S1312 are shown. If the process proceeds to step S1312, the additional processing time calculator 23 selects data of the calculated film thickness values at multiple times within the time range set in step S1302 from the data of the calculated film thickness values at past times being processed stored in the storage unit 40 (S3121), and uses the selected calculated film thickness value time series data to obtain a first-order regression line: Y = Xc·t + Xd (Y: remaining film thickness, t: etching time, Xc: absolute value is the etching rate, Xd: initial film thickness) (step S3122). Among them, the first-order regression line obtained in this step S3122 is a different regression line from the first-order regression line obtained in step S1310. The etching rate is calculated through this step S3122 (S3123). Furthermore, the difference between the calculated film thickness of the currently processed film for which the end point of film thickness estimation was determined in step S1311 and the remaining film thickness at the etching end point is calculated as the remaining etching amount (S3124), and the additional processing time required to reach the remaining film thickness at the etching end point is calculated by dividing this remaining etching amount by this etching rate (S3125).

[0210] Then, the process proceeds to step S1313, and the etching process is continued for the additional processing time calculated in step S1312 from the time when the end point of the film thickness estimation is determined (S1313 "No"), and if the calculated additional processing time has passed (S1313 "Yes"), a signal for ending the etching process is sent to the plasma processing device 1. Finally, the sampling end setting is performed, and the process ends (step S1314).

[0211] Using the above implementation, Figure 13 The results of performing the etching endpoint determination are shown. Figure 13 In the graph 1400, 1401 represents the remaining film thickness of the processed wafer after processing, and 1402 represents the residual film thickness of the processed wafer after processing. Figure 13 The graph of FIG. 1 shows the deviation of the remaining film thickness after the process. The remaining film thickness after the process was close to the target remaining film thickness of 140 nm in all wafers, and the error was also less than ±1.0 nm, which enabled etching process with high processing accuracy.

[0212] As is clear from the results, according to this embodiment, even when there is interference light pattern data P(m)sj corresponding to the target remaining film thickness stored in part or all of the differential waveform pattern databases 151 included in the differential waveform pattern database set 141 , the etching endpoint can be determined with high accuracy.

[0213] [Example 3]

[0214] In this embodiment, the case where the remaining film thickness at the end point of the film thickness estimation is set to the thinnest film thickness in the remaining film thickness range common to all the differential waveform pattern databases 151 in the differential waveform pattern database set 141 described in Embodiment 2 is described. As for the conditions other than this, since the structure is the same as that of the plasma processing apparatus 110 of Embodiment 2 and the same function is achieved, the detailed description is omitted.

[0215] In this embodiment, the thinnest film thickness in the remaining film thickness range common to the differential waveform pattern databases 151 included in the differential waveform pattern database set 141 is set to be equivalent to Figure 9 145nm of the data 901 in the graph 900. Therefore, Figure 11 The remaining film thickness value at the end point of the film thickness estimation in step S1302 is set to 145 nm. Figure 11 In the process shown, until the remaining film thickness of 145 nm is detected, the remaining film thickness is calculated using the above-mentioned formulas (4), (5) and (6). After the end point of the film thickness estimation is detected, etching processing is additionally performed for the time calculated in step S1312 to determine the end point of the etching processing.

[0216] As a result, the same etch end point determination accuracy as in Example 2 was obtained. Therefore, it is obvious that by the method of the present embodiment, even when there is no data P(m)sj of the interference light pattern corresponding to the remaining film thickness of the target of a part or all of the differential waveform pattern databases 151 stored in the differential waveform pattern database set 141, the etch end point can be determined with good accuracy.

[0217] The invention made by the present inventors has been specifically described based on the embodiments, but the present invention is not limited to the above-described embodiments, and it goes without saying that various modifications can be made without departing from the gist thereof. For example, the above embodiments have been described in detail for easy understanding of the present invention, but are not necessarily limited to having all the structures described. In addition, with respect to a part of the structures of the respective embodiments, addition, deletion, and replacement of other structures can be made.

Claims

1. A plasma processing method, which uses plasma formed in a processing chamber to process a film to be processed included in a plurality of film layers disposed on the surface of a wafer arranged inside the processing chamber. Characterized in that: At a plurality of moments during the processing of the wafer, light from inside the processing chamber is received, and time-series data related to the light is detected. The data of the true pattern related to the light with wavelength as a parameter obtained from the time-series data is compared with the pattern data for detection obtained by using a plurality of pattern data which are pattern data with wavelength as a parameter, and the thickness of the film to be processed is calculated. The plurality of pattern data are pattern data with wavelength as a parameter obtained by removing data outside the allowable range from a given reference value in a plurality of pattern data in which the remaining film thickness of the film to be processed during the processing and the time-series data related to the light are correlated. Based on the value of this film thickness, the end point of the processing using the plasma is determined.

2. The plasma processing method according to claim 1, Characterized in that: The plurality of pattern data have patterns different according to the thickness or structure of the films included in the plurality of film layers.

3. The plasma processing method according to claim 1 or 2, Characterized in that: The true pattern data is compared with the pattern data for detection obtained by using the sum of the plurality of pattern data after removing data outside the allowable range from the given reference value, and the thickness of the film to be processed is calculated.

4. The plasma processing method according to claim 1 or 2, Characterized in that: The true pattern data is compared with the pattern data for detection obtained by using the plurality of pattern data after removing data at wavelengths with a large standard deviation, which is the deviation in each wavelength of the plurality of pattern data from the given reference value, to calculate the thickness of the film to be processed.

5. The plasma processing method according to claim 1 or 2, Characterized in that: The thickness of the film to be processed is calculated by comparing with the pattern data for detection obtained by using the sum of the plurality of pattern data each multiplied by a weight coefficient. The weight coefficient is determined according to the magnitude of the difference from the true pattern data.

6. The plasma processing method according to claim 4, Characterized in that: The thickness of the film to be processed is calculated by comparing with the pattern data for detection obtained by using the sum of the plurality of pattern data each multiplied by a weight coefficient. The weight coefficient is determined according to the magnitude of the difference from the true pattern data.

7. A plasma processing method, which uses plasma formed in a processing chamber to process a film to be processed included in a plurality of film layers disposed on the surface of a wafer arranged inside the processing chamber. Characterized in that: At a plurality of moments during the processing of the wafer, light from inside the processing chamber is received, and time-series data related to the light is detected. Compare the data of the true pattern related to the light with wavelength as a parameter obtained from the data of the time series, and the pattern data for detection obtained using a plurality of pattern data which are pattern data with wavelength as a parameter obtained in advance, to calculate the thickness of the film of the object to be processed. The plurality of pattern data are a plurality of pattern data with wavelength as a parameter obtained by removing the data outside the allowable range from a given reference value from the plurality of pattern data corresponding to the remaining film thickness of the film of the object to be processed during the processing and the time series data related to the light. When it is determined that the calculated film thickness reaches a preset value, further process the film of the object to be processed for the time required for the remaining film thickness to reach the target film thickness from the preset value.

8. The plasma processing method according to claim 7, characterized in that use the calculated film thickness to calculate the processing speed of the film of the object to be processed, and use the processing speed and the preset film thickness value to obtain the time required to reach the target film thickness.

Images