KR100733120B1 - Method and apparatus for detecting processing of semiconductor waper - Google Patents

Abstract

In the present invention, the light from the light source unit is irradiated onto the semiconductor wafer surface during the treatment using plasma. The interference light is received through a port through which the interference light of a plurality of wavelengths obtained from the surface of the semiconductor wafer passes through the plasma in the vacuum processing chamber. The etching process state of the film on the surface of the semiconductor wafer is detected based on the information obtained by using the interference light of the plurality of wavelengths received in the light receiving step.

Description

METHOD AND APPARATUS FOR DETECTING PROCESSING OF SEMICONDUCTOR WAPER}

1 is a view showing the overall configuration of an etching apparatus for a semiconductor wafer with a film thickness measuring apparatus according to an embodiment of the present invention;

2A and 2B are diagrams showing examples of longitudinal cross-sectional shapes of materials to be processed and actual patterns of wavelengths of interference light during etching;

3A and 3B are diagrams in which the wavelength of the differential coefficient value time series data of the interference light corresponding to the respective film thicknesses (distance from the boundary plane) shown in A, B, and C of FIGS. 2A and 2B is a parameter;

4 is a flowchart showing a procedure for obtaining a film thickness of a material to be treated when etching is performed by the film thickness measuring apparatus of FIG. 1;

5 is a diagram showing the overall configuration of an etching apparatus for a semiconductor wafer with a film thickness measuring apparatus according to another embodiment of the present invention;

6 is a flowchart showing the operation of the embodiment of FIG. 5;

7 is an operation explanatory diagram of the embodiment of FIG. 5;

8 is a view showing the overall configuration of an etching apparatus for a semiconductor wafer with a film thickness measuring apparatus according to another embodiment of the present invention;

9 is an operation explanatory diagram of the embodiment of FIG. 8;

10 is a flowchart illustrating operation of the embodiment of FIG. 8;

11 is a view showing the overall configuration of an etching apparatus for a semiconductor wafer with a film thickness measuring apparatus according to another embodiment of the present invention;

12 is an operation explanatory diagram of the embodiment of FIG. 11;

13 is an operation explanatory diagram of the embodiment of FIG. 11;

14 is a flowchart illustrating operation of the embodiment of FIG. 11;

15 is an operation explanatory diagram of the embodiment of FIG. 11;

16 is a view showing the overall configuration of an etching apparatus for a semiconductor wafer with a film thickness measuring apparatus according to another embodiment of the present invention;

17A and 17B are diagrams illustrating the operation of the embodiment of FIG. 16;

18 is a view showing the overall configuration of an etching apparatus for a semiconductor wafer with a film thickness measuring apparatus according to another embodiment of the present invention;

19A to 19C are diagrams illustrating the operation of the embodiment of FIG. 18;

20 is a diagram showing the overall configuration of an etching apparatus for a semiconductor wafer with a film thickness measuring apparatus according to another embodiment of the present invention;

21 is a view showing a film thickness change for explaining another embodiment of the present invention.

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a film thickness measurement method for detecting a film thickness of a target material by a light emission spectroscopy method in the manufacture of a semiconductor integrated circuit, and a method of processing a target material using the same. The present invention relates to a method for measuring a film thickness of a workpiece to be precisely measured to obtain a desired thickness by measuring the film thicknesses of various layers provided, and to a method of treating a workpiece using the same.

BACKGROUND OF THE INVENTION In the manufacture of semiconductor wafers, dry etching is widely used for the removal or pattern formation of layers of various materials, especially layers of dielectric materials, formed on the surface of the wafer. The most important point in the control of process parameters is to accurately determine the etch end point for stopping the etch at the desired thickness during processing of such a layer.

During the dry etching process of a semiconductor wafer, the light emission intensity of a specific wavelength in plasma light changes with the etching progress of a specific film. Thus, as one of the etching end point detection methods of a semiconductor wafer, there is conventionally a method of detecting a change in the emission intensity of a specific wavelength from plasma during a dry etching process and detecting the end point of etching of a specific film based on the detection result. At that time, it is necessary to prevent erroneous detection based on the shaking of the detection waveform due to noise. As a method for accurately detecting a change in luminescence intensity, Japanese Patent Laid-Open No. 61-53728, Japanese Patent Laid-Open No. 63-200533 and the like are known. In Japanese Patent Laid-Open No. 61-53728, noise reduction is performed by a moving average method and Japanese Patent Laid-Open No. 63-200533 by approximation processing by a first least square method.

With the recent miniaturization and high integration of semiconductors, the aperture ratio (the etching target area of a semiconductor wafer) is decreasing, and the light emission intensity of a specific wavelength accepted by the photodetector from the optical sensor is weakened. As a result, the level of the sampling signal from the photodetector decreases, and it is difficult for the end point determining unit to reliably detect the end point of etching based on the sampling signal from the photodetector.

Moreover, when detecting the end point of etching and stopping a process, it is important that the thickness of the remainder of a dielectric layer is actually equal to a predetermined value. In the conventional process, the whole process is monitored using the time-thickness control method based on the premise that the etching rate of each layer is constant. The value of the etching rate is obtained by, for example, processing the sample wafer in advance. In this method, the etching process is stopped at the same time as the time corresponding to the predetermined etching film thickness has elapsed by the time monitoring method.

However, it is known that actual films such as SiO ₂ layers formed by low pressure chemical vapor deposition (LPCVD) techniques have low thickness reproducibility. The tolerance of the thickness due to process variations during LPCVD corresponds to about 10% of the initial thickness of the SiO ₂ layer. Therefore, the time monitoring method cannot accurately measure the actual final thickness of the SiO ₂ layer remaining on the silicon substrate. The actual thickness of the remaining layer is finally measured by a technique using a standard spectroscopic interferometer, and when found to be overetched, the wafer is rejected and discarded.

In the insulating film etching apparatus, a change over time is known such that the etching rate decreases as the etching is repeated. In some cases, the etching may be stopped in the middle, and the solution is essential. In addition, it is important to monitor the fluctuation of the etching rate over time for the stable operation of the process. In the conventional method, only the time monitoring of the end point determination is available, and there is no appropriate method to cope with the temporal change or fluctuation of the etching rate. Further, the end point determination when the etching time is as short as about 10 seconds should be made as an end point determination method for shortening the determination preparation time and the time of the determination time needs to be short enough, but it is not necessarily sufficient. Further, in the insulating film, the etching target area is often 1% or less, and the change in plasma emission intensity from the reaction product generated by the etching is small. Thus, there is a need for an endpoint determination system that can detect even minor changes, but no practical and inexpensive system has been found.

On the other hand, as another method of detecting an etching end point of a semiconductor wafer, Japanese Patent Application Laid-Open No. 5-179467, Japanese Patent Application Laid-Open No. 8-274082, Japanese Patent Application Laid-Open No. 2000-97648, Japanese Patent Application Laid-Open No. 2000-106356 A method of using the interferometer disclosed in the publication or the like is also known. In the use of this interferometer, the monochromatic radiation emitted from the laser hits a wafer having a vertical incident angle on a wafer including a laminated structure of different materials. For example, in what is a SiO ₂ layer laminate stacked on the Si ₃ N ₄ layer, the reflected radiation on the boundary surface formed between the reflection from the upper surface of the SiO2 layer emitting light and the, SiO ₂ layer and a Si ₃ N ₄ layer By this, an interference fringe is formed. The reflected radiation is irradiated to a suitable detector, which generates a signal whose intensity varies with the thickness of the SiO ₂ layer during etching. If the upper surface of the SiO ₂ layer is exposed during the etching process, the etching rate and the current etching thickness can be continuously and accurately monitored immediately. Instead of the laser, a method of measuring a predetermined emission light emitted by the plasma by a spectrometer is also known.

According to the method using an interferometer, the position of the boundary surface of the laminated structure is accurately measured. However, in order for the interference pattern to be formed by the radiation light reflected from the upper surface of a certain layer and the radiation light reflected from the boundary surface, the process reaches the boundary surface and cannot be measured at that point in time. Therefore, in the actual etching process, even if the processing reaches the boundary surface by measuring the thickness online from the interference fringes of the reflected light, the processed layer is overetched even if it is fed back to the process control. In order to avoid excessive etching, it is necessary to use a combination with the above-described time monitoring method. Since this requires a premise of a film thickness value, etc., it is difficult to perform an appropriate etching under the recent semiconductor miniaturization request for the above reason.

The content of each of the above-mentioned conventional documents is summarized as follows.

In Japanese Patent Laid-Open No. 5-179467, interference light (plasma light) is detected using three kinds of color filters of red, green, and blue to detect the end point of etching.

In addition, Japanese Patent Application Laid-Open No. 8-274082 (USP 5658418) uses the time variation of two wavelengths of interference wave and the differential wave form to obtain the extreme value of the wave form (maximum and minimum of waveforms: Counts zero. By measuring the time until the count reaches a predetermined value, the etching rate is calculated. Based on the calculated etching rate, the remaining etching time until the predetermined film thickness is obtained is obtained. Stop.

In addition, Japanese Patent Laid-Open No. 2000-97648 discloses a waveform (wavelength is used as a parameter) of the difference between the light intensity pattern (the wavelength is used as a parameter) of the interference light before processing and the light intensity pattern of the interference light after the treatment or processing. Then, the step (film thickness) is measured by comparison between the difference waveform and the difference waveform that is databased.

In addition, Japanese Patent Laid-Open No. 2000-106356 relates to a rotating coating apparatus, and obtains a film thickness by measuring a time variation of interference light over multiple wavelengths.

In USP 6081334, the characteristic behavior of the temporal change of the interference light is obtained by measurement, which is databased, and the termination of etching is determined by comparing the interference waveform measured with the database. This determination prompts the change of the etching process conditions.

In the above known examples, the following problems occur.

① The thinner the material to be etched, the smaller the interference light intensity and the smaller the number of interference fringes.

(2) When etching is performed using a mask material (for example, a resist), the interference light from the mask material is superimposed on the interference light from the material to be etched.

③ If the etching rate changes during the process, the interference waveform is distorted.

Because of the above, it was difficult to accurately measure and control the thickness of the layer to be treated, particularly the layer to be treated in the plasma etching treatment, with the required measurement accuracy.

SUMMARY OF THE INVENTION An object of the present invention is to provide a method for measuring a film thickness of a workpiece and a method of treating a workpiece using the same, which can accurately measure the actual thickness of the target layer online, particularly in plasma etching. have.

It is another object of the present invention to provide an etching process that can control each layer of a semiconductor device with high precision online at a predetermined thickness.

Another object of the present invention is to provide an apparatus for measuring the thickness of a workpiece, which can accurately measure the actual thickness of the layer to be processed online.

In order to solve the above problems of the prior art and to attain the object of the present invention, the present inventors obtain waveforms of time derivatives of the interference waveforms for each of a plurality of wavelengths, and based on the derivatives of the interference waveforms. The pattern showing the wavelength dependence of the value (that is, the pattern of the differential value of the interference waveform whose wavelength is a parameter) was obtained, and the film thickness was measured using the pattern.

In the present invention, the reason for using the pattern representing the wavelength dependency of the time derivative of the interference waveform is as follows.

Since the measurement is based on the in-situ measurement during etching, the film thickness of the film to be processed is changed every time. Therefore, time differential processing of the interference waveform is possible. In addition, the differential processing can remove the noise of the interference waveform.

In addition, the refractive index of the material to be etched (for example, polysilicon) varies greatly with respect to the wavelength. Therefore, the characteristic change (film thickness dependence) of the substance can be detected by the interference light measurement over multiple wavelengths.

The present invention relates to a method of detecting a semiconductor wafer, wherein the etching process of the semiconductor wafer is performed using a plasma formed in the vacuum chamber with respect to a film on the surface of the semiconductor wafer to be disposed in the vacuum chamber. Irradiating light from the light source unit to the surface of the semiconductor wafer during the treatment using plasma; and passing the light through a port through which a plurality of wavelengths of interference light is transmitted from the surface of the semiconductor wafer to face the plasma in the vacuum processing chamber. Receiving light and detecting an etching process state of the film on the surface of the semiconductor wafer on the basis of information obtained by using the interference light of the plurality of wavelengths received at the light receiving step. Detection method of semiconductor wafer processing to provide.

And placing a semiconductor wafer having a configuration substantially the same as that of the semiconductor wafer to be processed in the vacuum processing chamber, irradiating light from the light source unit to the semiconductor wafer to receive light obtained from the semiconductor wafer; Detecting the etching treatment state of the film on the surface of the semiconductor wafer based on the information obtained by using the light received in the step and the information obtained by receiving the light during the processing of the semiconductor wafer to be processed. A method of detecting a semiconductor wafer process is provided.

In addition, the present invention is a semiconductor wafer processing detecting apparatus for detecting an etching process performed by using a plasma formed in the vacuum container with respect to a film on the surface of a semiconductor wafer to be disposed on a sample table in a vacuum container. A light emitting unit that irradiates light onto the surface of the semiconductor wafer during the treatment using plasma, a light receiving device that receives interference light having a plurality of wavelengths from the surface of the semiconductor wafer through a port facing the plasma in the vacuum processing chamber, and And a detection device for detecting an etching process state of the film on the surface of the semiconductor wafer based on information obtained by using the interference light of the plurality of wavelengths received by the light receiving device. do.

Further, information obtained by receiving light from the semiconductor wafer by irradiating light onto the semiconductor wafer from the light source unit with respect to the semiconductor wafer disposed in the vacuum processing chamber and having a configuration substantially the same as that of the semiconductor wafer to be processed; An apparatus for detecting a semiconductor wafer is provided, wherein the etching treatment of the film on the surface of the semiconductor wafer is detected on the basis of the information obtained by receiving the light during the processing of the semiconductor wafer to be processed.

According to one aspect of the present invention, the film thickness measuring method for measuring the film thickness of the workpiece,

a) setting a standard pattern whose wavelength is a parameter of the derivative value of the interference light with respect to the predetermined film thickness of the first (sample) to-be-processed material;

b) measuring the intensity of the interfering light with respect to the second to-be-processed material of the same structure as the said 1st to-be-processed material with respect to a some wavelength, respectively, and obtaining the actual pattern which makes the wavelength of the derivative value of this measured interference light intensity a parameter. Steps;

c) obtaining a film thickness of the second to-be-processed material based on the standard pattern and the actual pattern of the derivative value.

In the present invention, the following aspects can be considered.

First, when the film thickness of the material to be etched is thick, the periodicity of the interference fringe becomes remarkable. In this case, the absolute film thickness is obtained by using interference light of three wavelengths or more.

On the other hand, when the film thickness of the material to be etched is thin, the periodicity of the interference fringe does not appear. Therefore, in this case, the absolute film thickness is obtained by using two wavelengths of interference light.

According to the present invention, it is possible to provide a method for measuring a film thickness of a target material and a method for processing a sample of the target material using the same, in which the actual thickness of the layer to be processed can be accurately measured online in a plasma treatment, particularly a plasma etching treatment.

Moreover, the etching process which can control each layer of a semiconductor device online with high precision with a predetermined thickness can be provided. In addition, it is possible to provide an apparatus for measuring the film thickness of a material to be able to accurately measure the actual thickness of the layer to be processed online.

Hereinafter, each Example of this invention is described. In each of the following embodiments, those having the same function as the first embodiment will be denoted by the same reference numerals as those in the first embodiment, and detailed description thereof will be omitted.

1 to 4, a first embodiment of the present invention will be described. This embodiment sets a standard pattern which shows the wavelength dependence of the derivative value of the interference light (with wavelength as a parameter) to the predetermined film thickness of the sample to be processed when plasma etching a to-be-processed material such as a semiconductor wafer. Next, the intensity | strength of the several wavelength of the interference light in the actual process with respect to the to-be-processed material of the same structure as the sample to-be-processed material is measured, respectively, and shows the wavelength dependency of the derivative value of this measured interference light intensity (wavelength is a parameter The actual pattern is obtained, and the film thickness of the processing target material is obtained by comparing the standard pattern of the differential value with the actual pattern.

First, the whole structure of the etching apparatus of the semiconductor wafer provided with the film thickness measuring apparatus of this invention is demonstrated using FIG. The etching apparatus 1 includes a vacuum vessel 2, and the etching gas introduced therein is decomposed by microwave power or the like to become a plasma, and the plasma 3 causes the semiconductor wafer on the sample table 5 to be disposed. The material 4 to be treated is etched. Multiwavelength radiation light from the measuring light source (for example, halogen light source) of the spectrometer 11 of the film thickness measuring apparatus 10 is guide | induced to the vacuum container 2 by the optical fiber 8, and a to-be-processed material (4) is a vertical incidence angle. The material to be treated 4 contains a polysilicon layer, and the emitted light is formed by interference light reflected from the upper surface of the polysilicon layer and radiation light reflected from the boundary surface formed between the polysilicon layer and the base layer. Interfering light is guided to the spectrometer 11 of the film thickness measuring apparatus 10 via the optical fiber 8, and the film thickness measurement and the endpoint determination process are performed based on the state.

The film thickness measuring apparatus 10 includes a spectrometer 11, a first digital filter circuit 12, a differentiator 13, a second digital filter circuit 14, a differential waveform pattern database 15, and a differential waveform comparator 16. And an indicator 17 for displaying the result of the comparator. 1 shows the functional configuration of the film thickness measuring apparatus 10. The actual configuration of the film thickness measuring apparatus 10 except for the indicator 17 and the spectrometer 11 is a CPU and a film thickness measuring process. A storage device comprising a ROM holding various data such as a program and a differential wave pattern database of interference light, a RAM for holding measurement data, an external storage device, and the like, an input / output device for data, and a communication control device.

The spectral intensities of the multiple wavelengths introduced by the spectrometer 11 become current detection signals corresponding to the light emission intensities, and are converted into voltage signals. The signals of a plurality of specific wavelengths output by the spectroscope 11 as sampling signals are stored in a storage device such as RAM as time series data y _ij . This time series data y _ij is then smoothed by the first digital filter circuit 12 and stored in a storage device such as RAM as smoothed time series data Y _ij . Based on this smoothed time series data Y _ij , the time series data d _ij of the derivative coefficient value (primary derivative value or secondary derivative value) is calculated by the differentiator 13 and stored in a storage device such as RAM. . The time series data d _ij of the differential coefficient value is smoothed by the second digital filter circuit 14 and stored in a storage device such as RAM as the smoothed differential coefficient time series data D _ij . From this smoothed differential coefficient time series data _{Di j} , an actual pattern representing the wavelength dependence of the derivative value of the interference light intensity (wavelength is used as a parameter) is obtained.

On the other hand, in the differential wave pattern database 15, the differential wave pattern data value Pj of the interference light intensity corresponding to each of the wavelengths corresponding to the material of the target material to be measured, for example, polysilicon, is measured in advance. It is set. In the differential waveform comparator 16, the actual pattern and the differential waveform pattern data value Pj are compared to obtain a film thickness of the workpiece. The result is displayed by the result indicator 17.

In addition, although only one spectrometer 11 is shown in the Example, when the inside of a to-be-processed material is to be measured and controlled, what is necessary is just to provide a plurality of spectrometers 11.

2A and 2B show examples of longitudinal cross-sectional shapes of the workpiece 4 during etching and actual wavelength patterns of interference light. In FIG. 2A, the base member 41, the etching target material 42, and the mask material 43 are laminated on the substrate 40 on the substrate 40. For example, when etching the gate film, the substrate of the workpiece 4 is an insulating film of SiO ₂ , and a polysilicon gate layer is formed on the polycrystalline base material corresponding to the source and the drain.

The multi-wavelength light emitted from the spectrometer 11 is in contact with the material 4 to be processed including the laminated structure of the material to be etched and the base material at a normal incident angle. The radiation light 9 guided to the etched portion instead of the mask material 43 is formed between the radiation light 9A reflected from the upper surface of the etching target material 42 and the etching target material 42 and the base material 41. Interference light is formed by the emitted light 9B reflected at the formed boundary surface. The position of reflection of the emitted light 9A as the etching process proceeds is changed as A, B and C. The reflected light is guided to the spectroscope 11 to generate a signal whose intensity varies depending on the thickness of the layer of etching target material 42 during etching.

As shown in Fig. 2B, the smoothed time series data Y _ij of the live waveform (multi-wavelength) of the interference light maintains a relatively large value until the distance from the boundary surface is near zero, and is rapidly reduced near zero. In addition, the right side shows the over-etching process rather than the point where the distance from the boundary surface is zero. Based on the smoothed time series data Y _ij , the derivative coefficient time series data d _ij of the first derivative or the second derivative is calculated. 2B, the first derivative and the second derivative of the interference light having a wavelength of 475 nm are shown. The first and second derivatives traverse a value of zero in a plurality of parts within a certain range of distance from the boundary surface. Hereinafter, the point crossing this zero value is called a zero passing point.

As is apparent from Fig. 2B, the zero passing point is also shown at a distance from the interface, in other words, at a relatively large value. This is a large difference compared to the decrease in the value and rapid reduction near zero until the live waveform reaches the boundary. In view of this fact, the present invention is characterized in that the film thickness can be accurately measured even in a state where the film thickness is relatively large. Further, even if the output of the plasma decreases, the first derivative and the second derivative of the interference light maintain a large value, so that accurate film thickness measurement can be performed.

3A and 3B are standard patterns each having a wavelength of a derivative value of interference light with respect to a predetermined film thickness of the material (polysilicon) shown in Figs. The differential coefficient value time series data d _{ij of} the interference light corresponding to the film thickness (distance from the boundary surface) is shown as a pattern. FIG. 3A shows the first differential wave pattern data of the interference light, and FIG. 3B likewise shows the second differential wave pattern data. A, B, and C in the figure show differential wave form pattern data at each film thickness of A (= 30 nm), B (= 20 nm) and C (= 10 nm) in FIG. 2A.

As is apparent from Figs. 3A and 3B, the first differential wave pattern and the second differential wave pattern of the interference light have a unique pattern for each material and film thickness of the target material, and at a specific wavelength, i. It can be seen that the first derivative or the second derivative is zero. For example, at the film thickness C, the wavelength 500 nm is zero passing point. If the materials of the materials to be treated are different, these patterns are also changed. Therefore, data are obtained in advance for various materials and film thickness ranges required for processing, and are stored in the recording device as a first differential pattern or a second differential pattern. It is good.

Next, the procedure of obtaining the film thickness of the to-be-processed material at the time of performing an etching process with the film thickness measuring apparatus 10 of FIG. 1 by the flowchart of FIG. 4 is demonstrated.

First, the extracted differential pattern Pi and the determination value sigma _{0 in the} wavelength region (at least three wavelength regions) are set from the target film thickness value and the film thickness pattern database (step 400). That is, among the standard patterns of differential values for the plurality of wavelengths shown in FIGS. 3A and 3B, which are held in the differential wave pattern database 15 in advance, at least corresponding to the film thickness required by the processing conditions of the material to be processed. Set three standard patterns.

In the next step, sampling of the interference light (for example, every 0.25 to 0.4 seconds) is started (step 402). That is, a sampling start command is issued in accordance with the start of the etching process. The light emission intensity of multiple wavelengths changed with the progress of etching is detected by the photodetector as the light detection signal of the voltage according to the light emission intensity. The light detection signal of the spectroscope 11 is digitally converted to obtain sampling signals y _{i and j} .

Next, the multi-wavelength output signal y _{i , j} from the spectrometer is smoothed by the first stage digital filter 12 to calculate time series data Y _{i , j} (step 404). That is, noise is reduced by the digital filter of the 1st stage, and smoothing time-series data y _i is calculated | required.

Next, the differential coefficients d _{i , j} are calculated by the SG method (step 406). I.e., obtain a coefficient of a signal waveform (primary or secondary) (d _i) by a differential process (SG) method. Further, the smoothed differential coefficient time series data D _{i , j} are calculated by the second stage digital filter 14 (step 408). Then, σ = Σ (D _{i , j} -p _j ) ² value is calculated (step 410). Next, in the differential waveform comparator 16, σ ≦ σ ₀ is determined (step 412), and when σ ≦ σ _O , the film thickness of the workpiece is a predetermined value, and the result is displayed on the display unit 17. ). If σ ≦ σ _O is not found, step 404 is returned. Finally, sampling end is set (step 414).

Here, the calculation of the smoothed non-coefficient time series data D _i will be described. As the digital filter circuit, for example, a second-order Butterworth type low pass filter is used. The smoothed time series data (Y _i ) is obtained by Equation (1) using a second-order Butterworth type low pass filter.

Here, the coefficients b and a differ in numerical values depending on the sampling frequency and the cutoff frequency. For example, at a sampling frequency of 10 Hz and a cutoff frequency of 1 Hz, a2 = -1.143, a3 = 0.4128, b1 = 0.067455, b2 = 0.13491, and b3 = 0.067455.

The time series data d _i of the second derivative value is calculated by the differential coefficient calculating circuit 6 from the equation (2) as follows using the polynomial fitted smoothing differential method of the five points of time series data Y _i .

Here, w-2 = 2, w-1 = -1, w0 = -2, w1 = -1, w2 = 2.

By using the time series data d _i of the differential coefficient values, the smoothed differential time series data D _i is a digital filter circuit (a low-pass filter of the secondary Butterworth type, except that the a and b coefficients of the digital filter circuit are obtained. May be different from).

Thus, according to the film thickness measuring apparatus of FIG. 1, at least one standard pattern of differential values with respect to a plurality of wavelengths as shown as A, B, C in FIG. 3A, 3B is set, and the interference light of the to-be-processed material The film thickness of the workpiece can be obtained by measuring the intensity of the wavelengths respectively, obtaining the actual pattern of the derivative value of each wavelength of the measured interference light intensity, and comparing the standard pattern with the actual pattern of the derivative value. For example, in the case where it is desired to detect a film thickness of 30 nm, that is, A in FIG. 2, a standard pattern of differential values for a plurality of wavelengths corresponding to the film thickness A is set in advance, When the coincidence ratio reaches within the determination value (σ ₀ ), it can be detected that the film thickness of the workpiece is 30 nm. As the standard pattern, one or both of the first derivative pattern and the second derivative pattern may be used.

According to this embodiment, even if the distance from the interface is a relatively large value, for example, 30 nm, the film thickness of the workpiece can be accurately measured.

Next, another embodiment of the present invention will be described with reference to Figs. In this embodiment, the wavelength (λ ₀ ) of the _zero passing point of one of the standard patterns is matched based on a standard pattern of differential values for a plurality of wavelengths corresponding to a predetermined film thickness in advance. The film thickness of the processing target material became a predetermined value according to two conditions in which the coincidence ratio of the actual value of the derivative value at the other wavelength λ _p reached within the determination value σ ₀ . Can be detected.

In Fig. 5, signals of two specific wavelengths output by the spectroscope 11 as sampling signals are storage devices such as RAM (not shown) as time series data y _{i ,} λ ₀ and y _{i ,} λ _p . Is housed in. These time series data are then smoothed by the first digital filter circuit 12 and stored in the storage device as smoothed time series data (Y _{i ,} λ _o and Y _{i ,} λ _p ). Based on these smoothed time series data (Y _{i ,} λ _o and Y _{i ,} λ _p ), the time series data (d _{i ,} λ _o ) of the derivative coefficient value (first derivative or second derivative) by the differentiator 13 is obtained. And d _{i ,} λ _p ) are calculated and stored in the storage device. The time series data of these differential coefficient values are smoothed by the second digital filter circuit 14 and stored in the storage device as smoothed differential coefficient time series data D _{i ,} λ _o and D _{i ,} λ _p . Then, from these smoothed differential coefficient time series data D _{i ,} λ _o and D _i, λ _p , the actual pattern of the derivative value for each wavelength of the interference light intensity is obtained.

On the other hand, in the differential waveform pattern database 15, the standard pattern of the derivative value at the wavelength (λ ₀ ) of one zero (0) passing point and the other wavelength (λ _p ) of the standard pattern is set in advance. . In the differential wave comparator 16, these are compared and the film thickness of the to-be-processed material is calculated | required.

For example, in the case where it is desired to detect a film thickness of 30 nm, that is, A in FIG. 2A, as shown in FIG. 7, the first derivative corresponding to the wavelength of _zero passing point λ ₀ and another wavelength λ _p = 450 nm Set the value P _p . In this embodiment, the components 12 to 16 may be configured by a computer having a CPU, a memory, and the like.

The operation of this embodiment will be described with the flowchart of FIG. First, the differential value of the target film from the thickness values and the film thickness of the pattern database zero pass wavelength (λ _o) and at least one different wavelength (λ _p) with a wavelength (λ _p) (P _p) and the decision value (σ _p ) is set (step 600).

Next, sampling of the interference light of the target material is started (step 602), and the output signals of wavelengths λ _O and λ _p from the spectrometer are smoothed by the first stage digital filter (Y _{i ,} o and Y). _{i ,} p) is calculated (step 604).

Next, the differential coefficients d _{i , o} and d _{i , p} are calculated by the SG method (step 606). Further, the smoothed non-coefficient time series data (D _{i , o} and D _{i , p} ) are calculated using the second stage digital filter (step 608). Further, σ = Σ (D _{i , p} -P _p ) ² is calculated (step 610).

Next, D _{i -1, o} * D _{i , o} ≦ ₀ and also determine σ ≦ σ ₀ (step 612).

In the positive (+) determination of the sign of D _{i -1, o} * D _{i , o} , it is determined to be true if it is negative, and if σ ≦ σ _{0, the} film thickness determination is finished (step 614). If the sign of D _{i -1, o} * D _{i , o} is positive (+) or σ> σ _0, then step 604 is returned.

According to this embodiment, only by focusing on two specific wavelengths, that is, the differential value pattern shown in FIG. 7 has passed across λ ₀ to 0 (X-axis) and the derivative value of the other wavelength λ _p ( When P _p ) becomes the determination value σ _O , the film thickness of the workpiece can be measured. In particular, even if the distance from the interface is a relatively large value, for example, 30 nm, the film thickness of the workpiece can be accurately measured.

Another embodiment of the present invention will be described with reference to FIGS. 8 to 10. In this embodiment, the zero passing pattern P _j of the derivative value of the target wavelength (target wavelength) λ _T is set among the interference light with respect to the predetermined film thickness of the target material, and the actual The zero pass pattern of the derivative value of the interference light intensity is obtained, and the film thickness of the workpiece is determined from the zero pass number n.

In Fig. 8, the signal of the target wavelength λ _T output as the sampling signal by the spectroscope 11 is stored in a storage device (not shown) such as RAM as time series data y _{i and} λ _T. . This time series data is then smoothed by the first digital filter circuit 12 and stored in the storage device as smoothed time series data (Y _{i ,} λ _T ). Based on this smoothed time series data, the time difference data d _{i and} λ _T of the derivative coefficient value (primary derivative value or secondary derivative value) are calculated by the differentiator 13 and stored in the storage device. The time series data of the differential coefficient value is smoothed by the second digital filter circuit 14 and stored in the storage device as the smoothed differential coefficient time series data D _{i and} λ _T. On the other hand, in the differential waveform pattern database 15, data of the zero passing pattern P _j of the standard pattern is set in advance. In the differential waveform comparator 16, the smoothed differential coefficient time series data is compared with the zero pass pattern P _j of the derivative value to obtain the film thickness of the workpiece from the zero pass count.

As shown in Fig. 9, for example, when three zero passing points of the target wavelength λ _T respectively correspond to the film thicknesses A, B, and C, the derivative value passes through these zeros. By passing through the point, for example, the film thickness of 1 Onm can be measured.

In the present embodiment, the components 12 to 16 may be configured by a computer having a CPU, a memory, or the like. The operation of this embodiment is explained by the flowchart of FIG.

First, the wavelength λ _T and the target zero pass frequency N _T of the spectrometer are first set from the target film thickness value and the film thickness pattern database (step 100), and then sampling is started (step 1002). ). Next, the output signal from the spectrometer (wavelength λ _T ) is smoothed by the first stage digital filter to calculate time series data Y _{i ,} λ _T (step 1004). Next, the calculating the differential coefficient (d _i, λ _T) by the SG method (step 1006). Further, the smoothed non-coefficient time series data (D _{i ,} λ _T ) are calculated by the second stage digital filter (step 1008).

Next, a positive (±) judgment of the value sign of (D _{i -1,} λ _T ) * (D _{i ,} λ _T ) is performed, and the zero pass of the derivative is detected by negative (−) = true. (Step 1010). Further, the zero pass count of the differential coefficient is added (n = n + 1) (step 1012), and the target zero pass count N _T is compared with n (step 1014). If the target zero pass number N _T has not been reached, the process returns to step 1004. If the target zero pass number N _T has been reached, it is determined that the predetermined film thickness has been reached and the sampling ends. Set the following.

In this embodiment, the zero-pass pattern P _j of the derivative waveform of the specific wavelength λ _T is set, and the film thickness of the workpiece is determined from the zero-pass frequency of the actual pattern. Even if the distance is relatively large, the film thickness of the workpiece can be measured accurately.

Next, another Example of the film thickness measuring method by this invention is described with reference to FIGS. In this embodiment, the specific wavelength in the interference light of the target material is selected as the guide wavelength λ _G and the target wavelength λ _T , and from the zero-pass pattern of the derivative value of the guide wavelength λ _G , The film thickness range of the target material is obtained, and the film thickness of the target material within the film thickness range is obtained from the zero pass pattern of the derivative value of the target wavelength λ _T.

In Fig. 11, the signals of two specific wavelengths output by the spectroscope 11 as sampling signals are stored in a storage device (not shown) as time series data y _{i ,} λ _G and y _{i ,} λ _T. . These time series data are then smoothed by two first digital filter circuits 12 (12A, 12B) and stored in the storage device as smoothed time series data (Y _{i ,} λ _G and Y _{i ,} λ _T ). Based on these smoothed time series data, the time series data (d _{i ,} λ _G and d _i, λ _T ) of the differential coefficient values (first derivative or second derivative) are determined by two differentiators 13 (13A, 13B). ) Is calculated and stored in the storage device. The time series data of these differential coefficient values are smoothed by two second digital filter circuits 14 (14A, 14B) and stored in the storage device as smoothed differential coefficient time series data (D _{i ,} λ _G and D _{i ,} λ _T ). It is stored. On the other hand, in the differential wave form pattern database 15, the data of the zero passing pattern P _{j of the} wavelengths λ _G and λ _T are set in advance. In the two differential wave comparators 16 (16A, 16B), these smoothed differential coefficient time series data are compared with the zero pass pattern P _j of the derivative value to obtain the film thickness of the workpiece. Also in this embodiment, the components 12A, 12B to 16A, 16B may be configured by a computer having a CPU, a memory, and the like.

Here, the relationship between the data of the zero passing pattern P _{j of the} wavelengths λ _G and λ _T will be described with reference to FIGS. 12 and 13. In the figure, four zero passing points of the target wavelength λ _T respectively correspond to the film thicknesses A, B, C, and D, and three zero passing points of the guide wavelength λ _G respectively. Corresponding to the film thicknesses a, b, and c, respectively. Then, three zero (0) passing points of the guide wavelengths (λ _G ) corresponding to the film thicknesses (a, b, c) and the film thicknesses (A, B, C, D) of each target, that is, the target wavelengths The relationship of the film thickness of four zero passing points of (λ _T ) is as shown in FIG. 13.

Therefore, for example, considering the case where the film thickness D is measured as the target film thickness, the zero passing point of the guide wavelength λ _G corresponding to the film thickness c corresponds to the film thickness D. It can be seen that it appears before the zero passing point of the target wavelength λ _T. Thus, when three zero pass points are detected at the guide wavelength λ _G and four zero pass points are detected at the target wavelength λ _T , it can be seen that the target film thickness D is obtained. have.

The operation of this embodiment will be explained by the flowchart of FIG. First, from the target film thickness value and the film thickness pattern database, the guide wavelength λ _G and the target wavelength λ _{T of the} spectrometer and the target zero pass times N _G and N _T of each wavelength are set ( Step 1400).

Next, in order to know the target zero passage number (m) of the guide wavelength (λ _G), spectroscopy smoothed time series by the output signal from the (wavelength λ _G), the digital filter of the first stage (Y _i, λ _G ) is calculated (step 1402). Further, the differential coefficients d _{i and} λ _G are calculated by the SG method (step 1404). Next, the smoothed differential coefficient time series data _{Di and} lambda _G are calculated by the digital filter of the second stage (step 1406). In addition, a zero pass of the differential coefficient is detected by a positive (±) determination (negative (-) = true) of the (D _{i -1,} λ _G ) * (D _i , λ _G ) value codes (step 1410). If zero pass is detected, the zero pass count of the derivative is added (m = m + 1) (step 1412) and compared with the target zero pass count (N _G ) ( In step 1414, if the target zero pass count m is reached, the process proceeds to the process of obtaining the target zero pass count n of the target wavelength λ _T.

In order to obtain the target zero passage number (n) of the target wavelength (λ _T), the first spectroscope smoothed by an output signal from the (wavelength λ _T) in the digital filter of the first level time-series data (Y _i, λ _T (Step 1416). Next, the calculating the differential coefficient (d _i, λ _T) by the SG method (step 1418). In addition, the smoothed differential coefficient time series data _{Di and} lambda _T are calculated by the second stage digital filter (step 1420). Next, the zero passage of the derivative is detected by the positive (±) determination (negative (-) = true) of the sign (D _{i -1,} λ _T ) * (D _{i ,} λ _T ). Step 1422). If zero pass is detected, the zero pass count of the differential coefficient is added (n = n + 1) (step 1424), and the comparison with the target zero pass count (N _T ) is performed ( Step 1426) If the target zero pass number n is reached, the target film thickness is achieved. Therefore, the result is recorded and output, and the sampling end is set.

In this embodiment, since the film thickness is determined by the zero pass times of the guide wavelength λ _G and the target wavelength λ _T , even if the distance from the interface is relatively large, the film thickness of the workpiece can be accurately measured. Can be.

According to the experiments of the inventors, taking the processing of the insulating film as an example, the zero passing point of the first differential wave pattern and the second differential wave pattern for multiple wavelengths has the characteristics as shown in FIG. That is, in the range where the film thickness is large, a zero passing point appears in either the first differential waveform pattern or the second differential waveform pattern. However, when the film thickness becomes thin, the zero passing point does not appear in the first differential wave pattern. Therefore, in the range where the film thickness is large, the wavelength of either the primary differential waveform pattern or the secondary differential waveform pattern may be the guide wavelength λ _G or the target wavelength λ _T. However, in the range where the film thickness is thin, it is preferable that the wavelength of the primary differential wave form pattern be the guide wavelength λ _G , and the wavelength of the secondary differential wave form pattern be the target wavelength λ _T. Using the characteristics of FIG. 15, for example, when the insulating film is to be processed, the guide wavelength λ _G is set to 475 nm, and when the remaining film thickness is 50 nm, the target wavelength λ _T is the second differential wave pattern. It is set as the wavelength of 455nm, or the wavelength of 475nm of a 1st order differential waveform pattern. When the remaining film thickness is 15 nm or less, the target wavelength lambda _T uses a second order differential waveform pattern. When the remaining film thickness is 15 nm to 35 nm, the target wavelength lambda _T adopts m = 1 of the first differential waveform pattern, and m = 2 of the primary differential waveform pattern when 35 nm to 100 nm.

According to the film thickness measuring apparatus of this invention demonstrated above, the film thickness of the to-be-processed material in the manufacturing process of a semiconductor device, etc. can be measured correctly. Therefore, by using this system, it is possible to provide a method of performing the etching of the workpiece with high accuracy. Hereinafter, the manufacturing process of such a semiconductor device is demonstrated.

First, FIG. 16 is a structural example of an etching apparatus employing the first embodiment of the present invention described with reference to FIGS. Data of the film thickness of the target material of the indicator 17 is sent to the plasma generating apparatus 20, and used to control the conditions for generating plasma in the vacuum vessel. For example, with respect to the material to be treated as shown in Fig. 17A, by changing the generation conditions of plasma in the vacuum vessel according to the film thickness determined by the film thickness measuring apparatus of the present invention, that is, the etching progress of the etching target material, Fig. 17B. It is possible to perform an etching process of an appropriate shape as shown in FIG. In the present embodiment, the components 12 to 16 may be configured by a computer having a CPU, a memory, and the like.

The procedure of the etching process is briefly described below.

Initially, the etching process conditions regarding a to-be-processed material are set. Among the etching treatment target material differential pattern extraction of a predetermined wavelength range (at least three different wavelength regions) of the target for each film thickness values and the film thickness according to a processing pattern for each layer in each layer (P _i) and the decision value (σ ₀ ) is included. Next, the material to be processed is carried in on the electrode to evacuate the vacuum vessel under reduced pressure. Then, a predetermined processing gas is introduced into the vacuum vessel to generate a plasma to start etching of the workpiece. At the same time, the light emission intensity of the multi-wavelength that changes with the progress of etching to start sampling the interference light is detected by the photodetector as the light detection signal of the voltage according to the light emission intensity. The light detection signal of the spectroscope 11 is digitally converted to obtain a sampling signal y _{i , j} . Next, the multi-wavelength output signals y _{i , j} from the spectrometer are smoothed to calculate time series data Y _{i , j} . Next, coefficients (primary or secondary) d _i of the signal waveform are obtained by differential processing (SG method), and smoothed differential coefficient time series data D _{i , j} are further calculated. Then, σ = Σ (D _{i , j} -p _j ) ² value is calculated and σ ≤ σ ₀ is determined, and when σ σ σ ₀ , the film thickness of the material to be treated is etched as a predetermined value. The process ends, the process gas is exhausted, and the material to be processed is taken out of the vacuum vessel.

For example, when the film thickness is intended to be the C value of Fig. 2A, a standard pattern indicating the wavelength dependence of the derivative value corresponding to each film thickness (A, B, C) is set in advance, and the actual pattern at a plurality of wavelengths is set. When the matching ratio with respect to the standard pattern of? Reaches within the determination value (σ ₀ ), it is detected that the remaining film thicknesses of the material to be processed are sequentially A and B, and the treatment is performed while appropriately controlling the processing conditions by plasma such as processing gas supply. Is advanced to control the film thickness to be exactly C, thereby completing the etching process.

In some cases, the etching process may be stopped once in a state where a predetermined film thickness, for example, the film thickness A has been detected, the necessary processing and operations are performed, and then the etching process may be resumed. Alternatively, the treatment may be performed to continuously change the etching treatment conditions in accordance with the film thickness while continuously and accurately measuring the film thickness.

In addition, the same plasma etching control may be performed by employing the measuring method of another embodiment of the present invention. The present invention can also be applied to processes such as plasma CVD, sputtering, chemical mechanical polishing (CMP) or thermal CVD.

Next, FIG. 18 is another structural example of the etching apparatus which employ | adopted 1st Embodiment of this invention. The data of the film thickness of the target material of the indicator 17 is processed by the control device 18 and sent to the plasma generating device 20, the gas supply device 21, and the wafer bias power supply 22, and the inside of the vacuum container. It is used to control the plasma generating conditions. For example, when the film thickness of the material to be etched is considerably thick, such as the hole processing of the insulating film, as shown in Figs. 19A, 19B and 19C, the film is measured by the film thickness measuring apparatus of the present invention by dividing the etching into two steps. By varying the processing conditions in the vacuum vessel according to the thickness, that is, the progress of etching of the etching target material (FIG. 19B), a suitable shape is made faster as shown in FIG. 19C, and there is no overetching on the base material. Can be etched. Also in this case, the same control may be performed by employing another embodiment of the present invention. In the present embodiment, the components 12 to 16 may be configured by a computer having a CPU, a memory, or the like.

In each of the embodiments described above, a multi-wavelength radiated light is emitted from a spectrometer having a light source, and the film thickness is measured by using interference light of reflected light from an object to be processed. On the other hand, you may use the multi-wavelength emission light emitted by a plasma as a light source using the spectroscope which does not have a light source. For example, as shown in FIG. 20, in order to observe the interference light on the target material by the plasma light from the upper surface, the interference light is intercepted by the optical fiber from the port provided on the upper wall of the vacuum chamber 2 by the first spectrometer 11A. In order to observe the state of the plasma light by the optical fiber from the other port provided on the side wall of the spectrometer vacuum vessel 2, and guide the light of both spectrometers to the divider 19. May be induced to the differentiator and treated by the method already described below. In addition, the spectrometers 11A and 11B do not have a light source. 20, the components 13, 15, 16, and 19 may be configured by a computer having a CPU, a memory, and the like.

According to the above method, an independent light source is not necessary, and stable and accurate measurement of the film thickness can be always performed even if plasma light changes. Further, as the number of treatment targets increases, the plasma processing apparatus can be changed over time to increase the processing time, thereby instructing maintenance.

Next, the Example which improves the film thickness precision in this invention is described. In the actual etching treatment, the plasma may change during the etching treatment. Therefore, the precision of the film thickness value measured may worsen. Fig. 21 shows the change in film thickness during polysilicon etching of 2.5 nm of the base oxide film, where? Indicates the film thickness measured value at each time obtained by the pattern comparison of the present invention. When the thickness of the polysilicon is thick at the beginning of etching and the interference light intensity from the polysilicon is weak, or when there is a plasma variation during the etching process, non-uniformity of the film thickness measurement value often occurs. In such a case, the following processing is performed to improve the film thickness precision. The following processing is performed by the software of the computer.

(1) The film thickness calculation start is performed from the time when the thickness of the polysilicon is thin (for example, 175 nm or less) and the plasma variation is small (T1), for example, 30 seconds have passed since the start of the etching process. After time T1, regression line is calculated by using the past film thickness measurement value (Exp. Value). (3) The film thickness measurement value which is out of the regression line shall be noise (for example, the allowable value of the film thickness measurement value not treated as noise shall be ± 1 Onm). (4) Recalculate the regression line from the measured film thickness excluding noise. (5) The current film thickness value is calculated from the regression line, which is calculated again, to be the filmed value (Fitted Value). Each time a predetermined time elapses, the above processes (2) to (5) are repeated to obtain a film thickness calculated value at each time, until the film thickness calculated value matches the final target film thickness value (Thf). The process is performed. In addition, when the film thickness calculated value reaches the final target film thickness value Thf, the etching process is terminated.

Next, the Example regarding the monitoring of the etching process state in this invention is demonstrated. The data string of the film thickness calculated value at each time described above is stored in a memory or an external recorder in the computer, and the received data string is databased in association with the wafer processing number. In this database, the calculated value of the film thickness when the end time of the etching process exceeds, for example, ± 5% with respect to the scheduled end time of etching, or at a suitable time (for example, time T2) during the etching process. When the target film thickness value (Th2 in Fig. 21) is exceeded, for example, by ± 5% at time T2, a warning is issued indicating that the etching process of this wafer process number is abnormal.

According to this method, even if there is a variation in the film thickness measurement value, the film thickness can be calculated with high accuracy, and the etching process for accurately determining the target film thickness can be performed. In addition, the state of the etching process can be monitored, and the number of defects processed on the wafer can be minimized.

According to the present invention, in the plasma treatment, in particular, the plasma etching treatment, it is possible to provide a method for measuring the film thickness of the target material which can accurately measure the actual thickness of the target layer online, and a method of treating the target material using the same.

Moreover, the etching process which can control each layer of a semiconductor device online with high precision with a predetermined thickness can be provided.

In addition, it is possible to provide an apparatus for measuring the film thickness of a material to be able to accurately measure the actual thickness of the layer to be processed online.

Claims (4)

A method of detecting a semiconductor wafer process, wherein the etching process status of the semiconductor wafer is detected by using a plasma formed in the vacuum process chamber with respect to a film on the surface of a semiconductor wafer to be disposed in the vacuum process chamber. Irradiating the surface of the semiconductor wafer with light from a light source unit during the treatment using the plasma; Receiving the interference light through a port through which the interference light of a plurality of wavelengths obtained from the surface of the semiconductor wafer passes through the plasma in the vacuum processing chamber; And detecting an etching process state of the film on the surface of the semiconductor wafer based on information obtained by using the interference light of the plurality of wavelengths received in the light receiving step. The method of claim 1, Arranging a semiconductor wafer having a configuration substantially the same as that of the semiconductor wafer to be processed in the vacuum processing chamber and irradiating light to the semiconductor wafer from the light source unit to receive light obtained from the semiconductor wafer; Detecting the etching treatment state of the film on the surface of the semiconductor wafer based on the information obtained by using the light received in the light receiving step and the information obtained by receiving the light during the processing of the semiconductor wafer to be processed. And a semiconductor wafer process detecting method. A semiconductor wafer processing detecting apparatus for detecting an etching process performed by using a plasma formed in a vacuum container with respect to a film on a surface of a semiconductor wafer to be disposed on a sample table in a vacuum container. A light emitting unit for irradiating light onto the surface of the semiconductor wafer during the processing using the plasma; A light receiving device for receiving interference light of a plurality of wavelengths from the surface of the semiconductor wafer through a port facing the plasma in the vacuum processing chamber; And a detection device for detecting an etching process state of the film on the surface of the semiconductor wafer on the basis of information obtained by using the interference light of the plurality of wavelengths received by the light receiving device. The method of claim 3, wherein Information obtained by receiving light obtained from the semiconductor wafer by irradiating light onto the semiconductor wafer from the light source unit with respect to the semiconductor wafer disposed in the vacuum processing chamber and having a configuration substantially the same as that of the semiconductor wafer to be processed; And detecting an etching process state of the film on the surface of the semiconductor wafer on the basis of the information obtained by receiving the light during the processing of the semiconductor wafer.

Images