JPH11118432A - Film thickness gauging device using ft-ir - Google Patents

Abstract

PROBLEM TO BE SOLVED: To prevent gauging precision from degrading even if the film thickness of an epitaxial layer becomes small. SOLUTION: This film thickness gauging device using an FT-IR, which detects interferogram and gauges its film thickness by means of a reflection method by radiating infrared light to a film formed on a substrate using a Fourier transform infrared spectrophotometer 1, is provided with channel spectrum taking-out means 2, 3 for conducting Fourier transformation for the interferogram obtained by gauging a sample to take out a channel spectrum, a resolving power improvement calculation means 4 which conducts spectral estimation from the channel spectrum, and conducts resolving power improvement calculation to obtain a septrum, and film thickness calculation means 5, 6 for detecting a peak from the septrum and for calculating the film thickness from the peak position, and the resolving power improvement calculation means 4 conducts calculation by means of a maximum entropy method or an autoregression model.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an FT for irradiating a thin film formed on a substrate with infrared light using a Fourier transform infrared spectrophotometer, detecting an interferogram by a reflection method, and measuring the film thickness. The present invention relates to a film thickness measuring device using IR.

[0002]

2. Description of the Related Art FIG. 3 is a diagram showing a conventional example of a film thickness measuring apparatus using FT-IR, FIG. 4 is a diagram for explaining an interferogram and a channel spectrum used for film thickness measurement, and FIG. FIG. 9 is a diagram for explaining a conventional thickness calculation algorithm. As one means for measuring the film thickness of an epitaxial layer formed on a silicon wafer, FT-I
R (Fourier transform infrared spectrophotometer) is used (for example, JP-A-61-140806, JP-B-4-496).
No. 42). FIG. 3 shows a configuration example of this apparatus.
The outline is described below. In FIG. 3, an infrared light source 21 generates infrared light and transmits the infrared light to the reflecting mirror 2.
The light reflected by the light sources 2 and 23 irradiates a sample 25 placed on a sample stage 24. This infrared light is reflected by the front and back surfaces of the sample, and the reflected infrared light is further reflected by the reflecting mirrors 26 and 27 to form a Michelson comprising a beam splitter 28, a fixed mirror 29, and a moving mirror 30. Incident on the interferometer 31 and cause interference. On the other hand, the moving mechanism 32
By moving the movable mirror 30 by a predetermined distance,
The detector 33 detects the interferogram of the interfering infrared light. After the detected interferogram is amplified by the amplifier 34, the A / D converter 3
5 and converted into a digital signal and stored in the memory 37 via the computer 36. Then, the computer 36 performs an arithmetic process to calculate the film thickness of the sample 25 and records it on the recorder 38.

When an epitaxial layer on a silicon wafer is measured by the reflection method using FT-IR as described above, two bursts appear in the interferogram as shown in FIG. As shown in FIG. 4A, one of them is a center burst due to light α1 reflected on the surface of the epitaxial layer 42, and the other is light α2 reflected on the interface between the wafer substrate 41 and the epitaxial layer 42. Is a side burst.

[0004] A spectrum obtained by Fourier-transforming the spectrum includes:
As shown in FIG. 4C, a channel spectrum due to the side burst is seen. In order to extract only the channel spectrum as shown in FIG. 4F, for example, only the wafer without the epitaxial layer 42 is used in FIG.
A reference spectrum as shown in (E) is measured. Then, only the channel spectrum as shown in FIG. 4 (F) is extracted by dividing the sample spectrum as shown in FIG. 4 (D) and the reference spectrum as shown in FIG. 4 (E).

The channel spectrum obtained as described above appears as a triangular wave, and its period varies depending on the thickness of the epitaxial layer 42. When this channel spectrum is subjected to Fourier transform to obtain a septum, a peak appears at a position corresponding to the cycle of the channel spectrum as shown in FIG. The thickness d of the epitaxial layer 42 can be calculated from the position of this peak.

[0006]

FIG. 6 is a diagram for explaining a problem that occurs when the conventional film thickness measurement is thin. However, the conventional film thickness measuring method using the FT-IR has the following problems. In other words, the channel spectrum attenuates as the wave number increases due to the influence of surface scattering and the like, so that the peak of the septram has a width due to this. In addition, the measurable range of the channel spectrum is limited (restriction of spectroscope and sample). This is considered as a window function in the Fourier transform as shown in FIG. 5, and there is a problem that the peak width of the septum increases.

Further, as the thickness of the epitaxial layer becomes thinner, the peak position of the septum becomes closer to zero as shown in FIG. 6, whereby the shape of the peak is disturbed and the error of the peak position becomes larger. . Also,
The relative error to the measurement of the peak position due to the width increases. When the film thickness is small, only a part of one cycle of the triangular wave can be measured in the channel spectrum, and a peak may not be formed in the septram.

[0008]

SUMMARY OF THE INVENTION The present invention has been made to solve the above-mentioned problems, and it is an object of the present invention to prevent the measurement accuracy from being reduced even when the thickness of an epitaxial layer is reduced.

For this purpose, the present invention uses a Fourier transform infrared spectrophotometer, irradiates a thin film formed on a substrate with infrared light, detects an interferogram by a reflection method, and obtains a channel spectrum by a Fourier transform. F for measuring film thickness
What is claimed is: 1. A film thickness measuring apparatus using T-IR, comprising: a channel spectrum extracting means for performing a Fourier transform on an interferogram obtained by measuring a sample to extract a channel spectrum; and a resolution estimating a spectrum from the channel spectrum. Resolution improvement calculation means for performing an improvement calculation to obtain a septram, and a film thickness calculation means for detecting a peak from the septram and calculating a film thickness from the peak position, wherein the resolution improvement calculation means, It is characterized by performing calculation by a maximum entropy method or an autoregressive model.

[0010]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 shows an FT-IR according to the present invention.
FIG. 2 is a diagram showing an embodiment of a film thickness measuring device using the present invention, and FIG. 2 is a diagram showing a comparative example of a peak of a septram by a conventional process and a peak by a process of the present invention. Reference numeral 2 denotes a reference storage unit, 3 denotes a channel spectrum extraction unit, 4 denotes a resolution improvement calculation unit, 5 denotes a peak detection unit, and 6 denotes a film thickness calculation unit.

In FIG. 1, an FT-IR reflection measuring device 1
In order to measure the film thickness of the epitaxial layer formed on the silicon wafer, the interferogram was measured by the reflection method using FT-IR as described above with reference to FIG. It outputs a channel spectrum. The reference storage unit 2 stores a reference spectrum measured using only a wafer without an epitaxial layer, and the channel spectrum extracting unit 3 stores the channel spectrum output from the FT-IR reflection measurement device 1 and the reference storage unit. The reference spectrum stored in 2 is divided, and only a channel spectrum necessary for film thickness measurement is extracted as a measured spectrum. The resolution improvement calculation unit 4 performs a resolution improvement operation so as not to increase the peak width of the septum to be obtained, and includes, for example, calculation using a maximum entropy method (MEM) or an autoregressive model. The peak detecting section 5 detects the peak of the septum obtained by the resolution improving calculating section 4, and the film thickness calculating section 6
Calculates the film thickness of a sample whose film thickness is determined to be unknown from the peak position of the septum.

The maximum entropy method (MEM) as a spectrum estimating method performed by the resolution improving calculation unit 4 is 1967.
Burg's proposal, and based on the idea of maximizing information entropy, spectrum estimation of the entire signal is performed from a finite section. The autoregressive model is based on the proposal of Akaike in 1969. Assuming that a stochastic process can be expressed by an autoregressive model, spectrum estimation is performed by applying an observed waveform to the autoregressive model. Here, the autoregressive model assumed for the observation data is a model of a stochastic process given by the following equation (Equation 1).

[0013]

(Equation 1) Here, x _k (= x (kΔt) is observed time-series data, _nk is stationary white noise independent of x _l (l <k), m is the order of the autoregressive model, and a _mi is the self in the order m. It is a regression coefficient.

The autocorrelation function of the time series data x _k is

[0015]

[Expression 2] R _i = R (iΔt) E ｛x _k x _ki ｝ where E る こ と is an expected value, and the expected value is _obtained by multiplying both sides of Expression 1 by x _k. By

[0016]

(Equation 3) Similarly, x _k−1 , x _k−2 ,..., X
_The expected value is multiplied by _km to obtain a matrix equation (Equation 4).

[0017]

(Equation 4) Here, P _m is the variance of the stationary white noise (E { _nk ² } = σ
² ) Also, by using the Wiener-Khinchine formula, a relational expression [Equation 5] between the autoregressive model {a _k } and the power spectrum S (ω) can be obtained.

[0018]

(Equation 5) The peak width of the septum is caused by the attenuation of the channel spectrum and the window function caused by the range of the measurement region. Therefore, when obtaining the septram, if the resolution improvement calculation is performed by the maximum entropy method (MEM) or the calculation using the autoregressive model as in the present invention instead of the Fourier transform, the influence of the window function by the Fourier transform can be removed. Therefore, the peak width of the septum becomes narrower than the conventional peak width obtained by Fourier transform as shown in FIG. Therefore, measurement errors can be reduced, and accurate measurement results can be obtained even when the thickness of the epitaxial layer is small as shown in FIG.

The present invention is not limited to the above embodiment, and various modifications are possible. For example, in the above embodiment, the calculation using the maximum entropy method (MEM) or the auto-regression model has been described as an example of the resolution improvement calculation. However, another resolution improvement calculation may be performed. Although the thickness of the epitaxial layer formed on the silicon wafer is measured, the present invention may be similarly applied to the measurement of the thickness of another thin film formed on the substrate.

[0020]

As is apparent from the above description, according to the present invention, the resolution improvement calculation by the maximum entropy method (MEM) or the calculation using the autoregressive model is performed from the channel spectrum obtained by measuring the sample. , A septram having a narrow peak width can be obtained, and a measurement error can be reduced. Further, by reducing the measurement error, an accurate measurement result can be obtained even when the thickness of the epitaxial layer is small.

[Brief description of the drawings]

FIG. 1 is a diagram showing an embodiment of a film thickness measuring apparatus using FT-IR according to the present invention.

FIG. 2 is a diagram showing a comparative example of a peak of a septum by a conventional process and a peak by a process of the present invention.

FIG. 3 is a diagram showing a conventional example of a film thickness measuring apparatus using FT-IR.

FIG. 4 is a diagram for explaining an interferogram and a channel spectrum used for film thickness measurement.

FIG. 5 is a diagram for explaining a conventional algorithm for calculating a film thickness.

FIG. 6 is a diagram for explaining a problem that occurs when the conventional film thickness measurement is thin.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... FT-IR reflection measuring device, 2 ... Reference storage part, 3 ... Channel spectrum extraction part, 4 ... Resolution improvement calculation part, 5 ... Peak detection part, 6 ... Film thickness calculation part

Claims (2)

[Claims] 1. A thin film formed on a substrate is irradiated with infrared light by using a Fourier transform infrared spectrophotometer, an interferogram is detected by a reflection method, and a channel spectrum is obtained by a Fourier transform to measure a film thickness. A film thickness measuring apparatus using an FT-IR, which performs a Fourier transform on an interferogram obtained by measuring a sample to obtain a channel spectrum, and performs a spectrum estimation from the channel spectrum. FT-IR comprising: a resolution improvement calculating means for performing a resolution improvement calculation to obtain a septram; and a film thickness calculating means for detecting a peak from the septram and calculating a film thickness from the peak position. Film thickness measuring device. 2. A film thickness measuring apparatus using FT-IR according to claim 1, wherein said resolution improving calculation means performs calculation by a maximum entropy method or an autoregressive model.