JP2002340789A - Method for determining composition of compound semiconductor layer on substrate - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method for determining a composition of a compound semiconductor layer on a substrate whereby a thin film structure, a wavelength dependence of a permittivity and a composition ratio are accurately and correctly determined by setting a combination model of a film thickness, a complex index of refraction and the like and finitely repeating fitting the model to a measured spectrum while changing an angle of incidence. SOLUTION: The method for determining the composition of the compound semiconductor layer on a surface of the substrate of a measuring object with the use of a spectroscopic ellipsometer includes the following steps. The ΨE and ΔE spectrum measurement and data formation steps 10 and 20 obtain measured spectra ΨE(λi ) and ΔE(λi ) for each wavelength λi by changing the wavelength of an incident light. The ΨMk and ΔMk modeling spectrum calculation steps 21 and 22 suppose (dj , Nj , (nj , kj )) for the j-th layer of the compound semiconductor layer and obtain modeling spectra ΨMk (λj ) and ΔMk (λj ). Comparison and evaluation steps 23 and 24 determine a structure reaching an evaluation criterion as the measured result.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a method of measuring a thin film using a spectroscopic ellipsometer, and more particularly to a method of determining the composition of a compound semiconductor layer formed on a substrate.

[0002]

2. Description of the Related Art The amount of polarization change between incident light and reflected light is measured using a spectroscopic ellipsometer, and the film thickness (d),
The complex refractive index N (N = n-ik) can be calculated. The amount of change in polarization (ρ) is represented by ρ = tanψexp (iΔ) and depends on parameters such as wavelength (λ), incident angle (φ), film thickness, and complex refractive index. . (D, n, k) = f (Ψ, Δ, λ, φ)

When the incident angle is fixed, the single-wavelength ellipsometer uses three unknowns of (d, n, k) as follows:
Since only two independent variables can be measured, it is necessary to fix any one of d, n, and k as known. Changing the angle of incidence, even at a single wavelength, increases the measured variables. However, due to the difference in the incident angle (φ) (Ψφ ₁ , Δφ
_{Since 1} ) and (Ψφ ₂ , Δφ ₂ ) have a strong correlation, it is difficult to accurately determine d, n, and k.

[0004]

The information (Ψ _E , Δ _E ) spectrum of the amount of polarization change of the multilayer thin film formed on the substrate, which is measured using a spectroscopic ellipsometer, is obtained from n, k information of the substrate and each layer. N, k, and d. However, thin film analysis is not possible for the following reasons.

The amount of polarization change can be represented by the volume through which light passes (phase angle (β) × area of beam diameter). The phase angle (β) is represented by the following equation. If the beam diameter is constant, the amount of polarization change is as follows. Polarization change amount 変 化 film thickness (d) × complex refractive index N × φ where φ is an incident angle. Therefore, the value of the amount of polarization change also changes according to the correctness of the incident angle. By correctly obtaining the incident angle, the value of the polarization change amount can be also correctly obtained.

As described above, the information (Ψ _E , Δ _E ) spectrum of the measured polarization change of the multilayer thin film includes all of the n, k information of the substrate and the n, k, d information of each layer. From now on, n, k information of the substrate, n, k of each layer,
A unique combination of k and d information cannot be calculated. Therefore, an optimal model is determined using the known permittivity.

Next, a multilayer structure can be determined for a layer formed on the substrate by setting a model and fitting the model to a simulation model.
Further, in the case of a compound semiconductor layer, there is a strong demand to know the composition ratio accurately or to keep it within a certain range. The present inventors have determined that a compound semiconductor A of substance A and substance B
_{When the (1-x)} B _x layer is formed on the substrate A, the complex refractive index N of the semiconductor layer is determined according to the value of x. When the composition ratio x is extremely low, there is almost no difference between n ₀ and k ₀ of the substrate and n _j and k _j of the compound semiconductor layer, and the amount of polarization change is small. Is important (see FIGS. 10 and 11).

An object of the present invention is to set a combination model of the compound semiconductor layer thickness, complex refractive index, composition ratio, and the like, calculate a simulation spectrum thereof, and fit a fitting between the simulation spectrum and the measurement spectrum to an incident angle. Another object of the present invention is to provide a method for determining the composition of a compound semiconductor layer on a substrate, which determines the thin film structure and the composition ratio by changing the composition.

[0009]

[MEANS FOR SOLVING THE PROBLEMS] To achieve the above object
The method according to claim 1 according to the present invention further comprises
Table of compound semiconductor layers formed on the substrate using a meter
A set of compound semiconductor layers formed on the substrate by measuring the surface
Method for determining composition of compound semiconductor layer for determining composition ratio, x, y
The thin film on the surface of the substrate to be measured
To change each wavelength λ_i Change of incident light and reflected light
Measurement spectrum あ る_E (λ_i ) And Δ _E (λ_i ) Get
Ψ_E , Δ_E A spectrum measuring step;
(N₀ (N _0,k₀ )), The compound semiconductor layer on the substrate
Of the j-th layer (d_j,N_j (N_j,k_j))
Determine Dell and Modeling Spectrum Ψ_M (λ_i ) And Δ
_M (λ _i Get_M,Δ_M Modeling spectrum calculation
Tep and the Ψ_E , Δ_E Spectrum and Ψ_M,Δ_M Mo
The Deling spectra were compared, and
_M,Δ_M And determined the structure of (d, n, k and composition
Ratio is obtained), and the model is
If it does not meet the above criteria, select the next modified model
And said Ψ_M,Δ_M Modeling spectrum calculation step
And a correcting step of performing the comparative evaluation step.
It is.

According to a second aspect of the present invention, there is provided the method for determining the composition of a compound semiconductor layer on a substrate according to the first aspect, wherein the evaluation criteria of the model are Ψ _E (λ _i ), Δ
_E (λ _i ) and Ψ _M (λ _i ), Δ _M (λ _i ) in a finite set
And determine the mean square error having the smallest mean square error.

The method according to claim 3 according to the present invention is characterized in that
Item 1. The method for determining the composition of a compound semiconductor layer on a substrate according to Item 1.
And said Ψ_M,Δ_M Model simulation spectrum
The calculating step is as follows:_E , Δ_E Spectrum measurement step
The nominal incident angle of the₀ When the nominal incident angle is
φ₀ Φ near_k Simulation spec
Ψ_Mk (λ_i ) And Δ_Mk (λ_i ), And
The value step is as described above. _E , Δ_E Spectrum and Ψ_MkAnd Δ_Mk
Comparing the modeling spectrum, the evaluation criteria
Ψ_MkAnd Δ_MkComparative evaluation step to determine the structure of the
And is composed of Claim 4 of the present invention
A method of forming a compound semiconductor layer on a substrate according to claim 1.
In the determining method, the compound semiconductor layer is formed of SiGe, AlGa
As, InGaAsP, InGaAs, InAlAs, InGaP, AlGaInP, AlGaI
nAs, AlGaAsSb, InAsSb, HgCdTe, ZnMgSSe, ZnSSe, ZnC
dSe, ZnMnSe, ZnFeSe, ZnCoSe
is there.

[0012]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the method according to the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing a configuration of an ellipsometer used in the method of the present invention. The spectroscopic ellipsometer shown in this block diagram executes a spectrometric data acquisition step 10 in a method described later.

The Xe lamp 1 includes a number of wavelength components.
This is a so-called white light source. Light emitted from the Xe lamp 1 is guided to the polarizer 3 via the optical fiber 2. The light polarized by the polarizer 3 is incident on the surface of the sample 4 to be measured at a specific incident angle (for example, φ = 75 °).
The reflection from the sample 4 is guided to an analyzer 6 via a photoelastic modulator (PEM) 5. Photoelastic modulator (PEM) 5
Is phase-modulated to a frequency of 50 kHz to produce linear to elliptically polarized light. Therefore, Ψ and Δ can be determined with a resolution of several milliseconds. The output of the analyzer 6 is connected to a spectroscope 8 via an optical fiber 7. The output data of the spectroscope 8 is taken into the data acquisition unit 9, and the spectroscopic data acquisition step 10 ends. The position of the PEM 5 can be either after the polarizer 3 or before the analyzer 6.

FIG. 2 is a flowchart of a method for determining the composition of a compound semiconductor layer on a substrate using a spectroscopic ellipsometer according to the present invention. (Step 20) This step is a step of converting spectroscopic measurement data into comparison data. Comparing data in the form of spectral measurement data acquired in acquisition step 10 spectrometric data described above Ψ _E (λ) and Δ _{E (λ).} FIG. 3 is a graph showing the spectroscopic measurement data of Step 20. The vertical axis shows the measured spectra Ψ _E (λ _i ) Psi and Δ _E (λ _i ) Delta, which are changes in the polarization of the reflected light.

(Step 21) This step 21 is a modeling step of a spectroscopic measurement object. FIG. 4 shows step 2
3 is a chart for explaining model data set in FIG. A model is created in consideration of the manufacturing process or the like of the measurement object converted into the comparison data in step 20. The optical constant and composition of the substrate and each layer, and the thickness (d) of each layer are set. In this embodiment, a first layer SiGe (x = 0.1
Assume the compound semiconductor layer of 5), d = 800 °. It is assumed that a second layer which is a natural oxide layer of the compound semiconductor layer is present thereon. The thickness d of the second layer is set to 20 °, and the optical constants (n, k or ε _r, ε _i ) and the composition ratio of the substrate and the first and second layers are set. A known numerical value is used for the optical constant, and a numerical value is sequentially corrected using past accumulated data as needed to prepare a database.

(Step 22) In this step 22,
A modeling spectrum is created from the simulation model set in the step 21 and converted into comparison data.
FIG. 5 is a graph showing data of the model in step 22. The vertical and horizontal axes are as described in FIG. A modeling spectrum is created from the model adopted in step 21. Assuming that n, k or ε _r, ε _{i at} each wavelength is known, from these and the thickness d of the first and second layers,
Making modeling spectrum by calculating [psi _M a (lambda) and Δ _M (λ).

(Step 23) This step is a step of comparing the spectrometry comparison data with the model comparison data. FIG. 6 is a graph in which the spectral measurement data compared in step 23 and the model data are superimposed. The modeling spectrum Ψ _M (λ), calculated in step 22,
Δ _M (λ), and Ψ _E (λ),
Compare Δ _E (λ).

(Step 24) This step is for evaluating the result of the comparison. FIG. 7 is a chart for explaining the fitting performed in step 24. Using the method of least squares (Ψ _E (λ), Δ _E (λ))
And (Ψ _M (λ), Δ _M (λ)) are fitted with parameters so as to minimize the difference. As a result, it is determined whether the measured data matches the model. here,
N measurement data pairs Exp (i = 1, 2,..., N) and N corresponding calculation data pairs Mo of the model
Assuming that d (i = 1, 2,..., N), the measurement error has a normal distribution, and the standard deviation is σ _i , the mean square error (χ ² ) is given as follows. Here, P is the number of parameters.

The evaluation is performed when the mean square error (χ ² ) is within a predetermined range or a loop including step 25 described later (step 22 → step 23 → step 24).
→ selecting step 25 → inner smallest model giving the chi ² finite chi ² values obtained in step 24 in the finite iteration of step 22) as measurement data and model fit.

(Step 25) FIG. 8 is a chart for explaining the change of the model in step 25. In this step, when it is determined in step 24 that the model does not match the spectroscopic data, the model is changed and the next model is set. The first set in step 21
The thickness of the layer of SiGe, 800 °, is changed to 2000 °, and the composition ratio is changed from x = 0.15 to x = 0.2.
The following model is determined by appropriately changing the optical constants of each layer, the composition of each layer, and the like as necessary.

[0021] (Step 22) Step 22 is a model that has been set in step 25, theoretically next [psi _M (
λ) and Δ _M (λ). Step 23 → Step 2
4 → Step 26 → Step 22 is repeatedly executed.

(Step 26) FIG. 9 is a chart showing a graph of a model determined for explaining step 26 and a determined structure. This is a step in which the data of the model determined to be suitable in the evaluation step of step 24 is adopted as a measured value, and the measurement is terminated. In this embodiment, the thickness of SiGe (x = 0.18), which is the first layer, is set to 180
Of the compound semiconductor layer corresponding to 8.4Å and the composition ratio (x = 0.18) (n 1 , k 1), the thickness of the native oxide layer is the second layer of 20.8Å and the natural oxide film layer (n _2, k ₂₎
Those with is are adopted as those given minimum chi ^2.

There are two methods for determining the composition ratio. The inventor of the present invention has determined that the composition ratio is determined at the same time as the film thickness and the optical constant as in step 26 in the example, but the dielectric constant (optical constant) is obtained from the film thickness and the optical constant obtained as the measurement result. There is also a method of determining the composition based on a functional relationship established between the constant and the composition. The optical constants are:
In addition to a known numerical value (reference), a dispersion equation (an equation indicating the wavelength dependence of the dielectric constant of a substance) or the like can be used.

Next, a description will be given of a case where the incident angle near the nominal incident angle φ ₀ is measured as a parameter. As described above, the polarization change amount (ρ) is given by ρ = tanψexp (i
Δ), which depends on parameters such as wavelength (λ), incident angle (φ), film thickness, complex refractive index, and the relationship is as follows. (D, n, k, composition ratio) = F (Ψ, Δ, λ,
φ)

Even if a model is set based on the nominal incident angle φ ₀ shown in FIG.
The incident angle phi ₀ is anticipated that it is better to slightly increase or decrease, the above-mentioned [psi _E, delta _E is also better to consider it reasonable better and was measured data by the angle obtained by correcting the phi ₀ is good.

That is, in the thin film measurement method using the spectroscopic ellipsometer, the nominal incident angle in the Ψ _E , _ΔE spectrum measurement step is φ ₀ , and the Ψ _M, Δ _M model simulation spectrum calculation step is the φ _0,関 数_M0 (λ
_i ), Δ _M0 (λ _i ) and the simulation spectrum Ψ _{Mk where the} nominal incident angle is a function of φ _k near φ _0.
(λ _i ) and Δ _Mk (λ _i ). This model simulation spectrum is calculated in step and 21 to obtain Ψ _E (λ
_i), compared with the Δ _E (λ _i).

Here, FIGS. 10 and 11 are given as examples. The nominal angle of incidence is set at an angle φ _kmin = 74.85 ° near φ ₀
From, in each angle Yuki increases by [Delta] [phi _k = 0.01 °, was the best combination of thickness and composition ratio and its grounds, indicating the chi ^2. In this example, φ _kMax = 75.00 °
And This means that in a step corresponding to the above-described 21 steps, a model corresponding to d and the composition ratio x is sequentially created corresponding to the incident angle φ _k and fitting is performed. In FIG. 10, the uppermost row is φ _kmin = 74.8.
In 5 °, the film thickness d _kmin = 1100.4Å composition ratio x _kmin = 7.7876 (atom%) is the value of chi ² at that time in the optimal combination is 0.6292. FIG.
1 is a graph showing a composition ratio x and chi ² at each incident angle. Considering FIG. 10, when φ _kbest = 74.97, the film thickness d _kbest = 1100.3 ° and the composition ratio x
_kbest = 8.9277 (atom%) is the smallest χ ²
The optimal film thickness and composition ratio are determined using this model. In FIG. 11, XRD
The line shown in FIG. 1 is a result of measurement using another measuring device using X-rays, which is written for reference. According to this measurement, the Ge concentration is x = 9.00 (atom%), which is almost the same value as in the case of the present invention, and a value closer to the true value can be obtained by adjusting the incident angle. Is shown.

[0028]

As described above in detail, according to the present invention, the composition of a thin film, which was previously difficult, can be measured accurately and accurately by using various models and fitting the incident angle. it can.

Various modifications can be made to the embodiment described in detail above within the scope of the present invention. For ease of understanding, explanations have been made using Ψ and Δ consistently in relation to data acquisition and model setting. Similar measurements and fittings are possible using the following pairs of data that are well known to those skilled in the art and are within the scope of the present invention. (N, k), (ε _i , ε _r ), (tan Ψ, cos Δ),
(I _s, I _c)

In the method for determining the composition of the compound semiconductor layer on the substrate, the compound semiconductor layer has been described in detail with reference to the example of SiGe, but other compound semiconductor layers AlGaAs,
InGaAsP, InGaAs, InAlAs, InGaP, AlGaInP, AlGaInAs,
AlGaAsSb, InAsSb, HgCdTe, ZnMgSSe, ZnSSe, ZnCdSe,
It can also be used to determine the composition of any of ZnMnSe, ZnFeSe, and ZnCoSe. Also, an example in which one SiGe layer is formed on a substrate has been described, but the present invention can be similarly used for measurement of a different multilayer structure and measurement of a wide range of film thickness. Although the substrate is shown as an example of Si, other materials (glass, quartz, compound semiconductor, etc.) can be used as well.

The inventor of the present invention has determined that the composition ratio of the compound semiconductor is determined simultaneously with the film thickness and the optical constant as in step 26 in the example, but from the film thickness and the optical constant obtained as the measurement result, There is also a method of determining the composition based on a functional relationship established between the dielectric constant (optical constant) and the composition, which is also included in the technical scope of the present invention.

Although a known numerical value (reference) is used as the optical constant, a dispersion formula or the like showing the wavelength dependence of the dielectric constant of a substance can also be used, and is included in the technical scope of the present invention. .

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a spectroscopic ellipsometer used in step 10 of acquiring spectroscopic measurement data according to the method of the present invention.

FIG. 2 is a flowchart for explaining a thin film measurement method according to the present invention.

FIG. 3 is a graph showing spectroscopic measurement data in Step 20;

FIG. 4 is a table for explaining model data set in step 21;

FIG. 5 is a graph showing data of a model in step 22;

FIG. 6 shows spectrometry data compared in step 23;
It is the graph which superimposedly showed the data of the model.

FIG. 7 is a table for explaining fitting in step 24;

FIG. 8 is a table for explaining a model change in step 25;

FIG. 9 is a chart showing a graph of a determined model and a determined structure for explaining step 26;

FIG. 10 is a graph for determining an optimal model of a film thickness and a composition ratio by performing fitting at each angle while changing an incident angle.

FIG. 11 is a graph of data obtained by the fitting.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Xe lamp 2 Optical fiber 3 Polarizer 4 Sample 5 Photoelastic modulator (PEM) 6 Analyzer 7 Optical fiber 8 Spectrometer 9 Data acquisition part 10 Spectroscopic measurement step 20 Spectroscopic measurement comparison data step 21 Model setting step 22 Model Comparison data conversion step 23 Comparison step 24 Evaluation step 25 Data resetting step 26 End (selection of compatible model) step

Continuation of the front page (72) Inventor Nabatova Gabain, Natalia 4-13-4 Kitakasai, Edogawa-ku, Tokyo Atago Bussan Co., Ltd. F-term (reference) 2G020 AA04 AA05 BA05 BA18 CA15 CB32 CB43 CD04 CD15 CD22 CD36 2G059 AA01 BB10 BB16 CC03 EE05 EE12 HH02 HH03 JJ01 JJ17 JJ19 KK01 MM01 MM02 MM03 MM05 MM09 MM10 4M106 AA01 AA10 BA06 CB21

Claims (4)

[Claims] 1. A method for determining the composition of a compound semiconductor layer formed on a substrate by measuring the surface of the compound semiconductor layer formed on the substrate using a spectroscopic ellipsometer and determining a composition ratio x, y of the compound semiconductor layer formed on the substrate. The measurement spectrum Ψ _E (λ _i ) and Δ _E (λ) are obtained by changing the wavelength of the incident light and changing the polarization of the incident light and the reflected light for each wavelength λ _i by changing the wavelength of the incident light. _i ) Ψ _E , Δ
_E spectrum measuring step, (N ₀ (n _0, k ₀ )) of the substrate, and ( _dj, N _j (n _j, k _j )) of the _j- th layer of the compound semiconductor layer on the substrate. Assuming the model is determined, the modeling spectrum Ψ _M (λ
_i) and obtaining a _{Δ M (λ i) Ψ M} , and delta _M modeling spectrum calculation step, the [psi _E, wherein [psi _M and delta _{E spectra,} delta _M compares modeling spectrum, the [psi _M reaching the criteria _, a comparative evaluation determining the measurement results the structure of Δ _M (d, n, k and the composition ratio is determined), when the model does not match the criteria is to select the next correction model, the Ψ _M, performs delta _M modeling spectrum calculation step, the composition determining method of a compound semiconductor layer on a substrate comprising, a modification step of performing the comparison evaluation step. 2. The method for determining the composition of a compound semiconductor layer on a substrate according to claim 1, wherein the evaluation criterion of the model is Ψ _E (λ _i ), Δ _E (λ _i ) and Ψ in a finite set. _M (λ _i
), Δ _M (λ _i ), and determine the composition of the compound semiconductor layer on the substrate by determining the one with the smallest mean square error. 3. The compound semiconductor layer on the substrate according to claim 1.
In the method for determining the composition of the above,_M,Δ_M Model simulation spectrum calculation
Tep said Ψ_E , Δ_E Public spectrum measurement step
Nominal incident angle φ₀ When the nominal incident angle is φ₀ of
Nearby φ_k Simulation spectrum を
_Mk (λ_i ) And Δ_Mk (λ_i ) Is calculated._E , Δ_E Spectrum and
Ψ_MkAnd Δ_MkCompare modeling spectra and use them as evaluation criteria
Said Ψ reached_MkAnd Δ_MkTo determine the structure of a sample with the measurement results
Determination of composition of compound semiconductor layer on substrate as evaluation step
Method. 4. The method for determining the composition of a compound semiconductor layer on a substrate according to claim 1, wherein said compound semiconductor layer comprises SiG.
e, AlGaAs, InGaAsP, InGaAs, InAlAs, InGaP, AlGaIn
P, AlGaInAs, AlGaAsSb, InAsSb, HgCdTe, ZnMgSSe, Zn
A method for determining the composition of a compound semiconductor layer that is one of SSe, ZnCdSe, ZnMnSe, ZnFeSe, and ZnCoSe.