JPH11211655A - Ellipsometry and measurement equipment - Google Patents

Abstract

(57) [Summary] [PROBLEMS] To overcome the ambiguity of independent determination of the thickness and refractive index of a thin film by ellipsometry and to analyze a film structure with high reliability. SOLUTION: In the ellipsometry for analyzing the film thickness and the refractive index or composition of a film structure in which a thin film is deposited on a substrate, the entire measurement sample is immersed in a liquid, and the refractive index of the liquid is continuously changed. Performs ellipsometry measurements and detects discontinuous changes in the phase shift difference (Δ) of polarized light waves in the ellipsometry measurement data in liquid near the Brewster energy determined by the substrate and the incident angle. Is determined.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to the analysis of surfaces of semiconductors, metals and their compounds, and interfaces of ultra-thin films.
The present invention relates to a method and apparatus for evaluating a thin film structure having excellent sensitivity.

[0002]

2. Description of the Related Art Structures in which one or more ultra-thin insulating films are deposited on a substrate are used in many fields such as electronic devices, decorative products, and metal surface treatments. The deposition technology of various metals, semiconductors and insulator thin films is a key technology in the production of these products. Especially in recent years, especially regarding the structure, further thinning and miniaturization are required,
The influence of the thickness and refractive index of the ultra-thin film, or the interface between the thin films on the characteristics has been increasing. Therefore, it is essential to establish a technology to realize a denser interface with less defects, a technology to control the electronic properties of the interface, and a new analysis technology to examine the geometrical and electronic structure of the interface at the nanometer level. That is widely recognized.

Ellipsometry (ellipsometry) has been widely used as a method for analyzing the thickness and refractive index of a thin film structure, as well as the state of a mixed layer and the state of a transition layer at an interface. This is because the method is non-contact, uses light, does not damage the sample, and has advantages such as rapid analysis that are not found in other methods. In recent years, a so-called spectroscopic ellipsometer capable of continuously changing the energy of incident light has been marketed, and the accuracy of the ellipsometry has been greatly improved.

However, when analyzing the refractive index and the film thickness of a film by ordinary ellipsometry, the larger the film thickness, the higher the accuracy, and conversely, in the analysis of an ultra-thin film structure, sufficient accuracy is often obtained. There was an unfortunate disadvantage. Similarly, in the ordinary ellipsometry, the accuracy is higher as the difference in optical constant between the layers in the multilayer structure and the film thickness are larger, and conversely, the accuracy is often higher in the analysis of an ultrathin multilayer structure having similar optical constants. There was a disadvantage that it could not be obtained sufficiently.

More specifically, the conventional drawbacks will be described. For example, in the case of a structure in which an extremely thin silicon oxide film having a thickness of several tens of mm or less is deposited on a silicon substrate, the conventional ellipsometry uses this structure from the air. Light incident into the film structure passes through the film layer and is reflected on the substrate. Then, the amount of p-polarized light and s-polarized light when the emitted light reaches the detection unit and the change in the phase of the light wave are measured, and then the film thickness and the refractive index are known by calculation. Was difficult.

The main factor affecting the measured ellipsometric data is the product of the physical thickness of each layer and the refractive index (this product is often called the optical thickness). This is because it is difficult to separate and measure the refractive index.
To overcome this difficulty, recently developed spectroscopic ellipsometers measure ellipsometric data at many wavelengths, increase the number of parameters, and determine the absolute value of each, rather than the product of film thickness and refractive index. This became possible in principle. If one of the absolute film thickness and the refractive index can be obtained, the other can be easily obtained. Actually, it is not difficult to determine both the film thickness and the refractive index of a film having a film thickness of about several hundreds of mm or more using a spectroscopic ellipsometer. However, the difference between the ellipsometric data of the sample having the film and the ellipsometric data of the substrate is too small for an ultra-thin film of several tens of mm or less. It is still difficult to find the absolute value of.

[0007] In support of the fact that this difficulty actually occurs, a DR using an extremely thin silicon oxide film or silicon nitride film is used.
In the research and development of electronic devices such as AM, the refractive index of a film is assumed after measuring the value of the capacitance, not the physical thickness (dimension of the absolute length), to determine the film thickness. There is an actual situation in which the film thickness converted by the calculation (so-called converted film thickness) is commonly used. This indicates that it is difficult to measure the absolute film thickness or the refractive index of such an extremely thin film with high reliability.

[0008] Here, if the refractive index of the actual film structure is constant irrespective of the film thickness, no problem occurs, but according to recent research, when the film thickness becomes several tens of mm or less, the refractive index of the film becomes smaller. It has been reported that when the thickness is large or sometimes as large as several tens of percent compared to the bulk value. For example, KJ Hebert, T. Lab
ayen and EA Irene (The Physics and Chemistryof
Si-SiO2 Interface-3, HZ Massoud, EH Poindext
er and CR Helms, Editors, Proc.Vol. 96-1, p. 81
-96, The Electrochemical Society, Pennington, NJ, 1
996) analyzed the film thickness and refractive index of silicon oxide films of various thicknesses formed on a silicon substrate by a thermal oxidation method by combining ellipsometry and tunnel current measurement. The refractive index with respect to the wavelength of light 6328 ° is a constant 1.465 at a film thickness of 300 ° and above, while the film has a refractive index of 50 °.
Å reported 1.865, an increase of about 27%.

As described above, in the case of an extremely thin film having a thickness of several tens of mm or less, a significant change in the refractive index actually occurs. I know that I can't get it. However, as described above, in the conventional ellipsometry, ambiguity has always occurred in the determination of the film thickness and the refractive index in the analysis of the ultrathin film, as described above. Ellipsometric decisions were difficult. This difficulty is similar in principle to other films as well as silicon oxide.

For example, in the case of a structure in which a silicon nitride film is deposited as a first layer on a silicon substrate and a silicon oxide film is further deposited as a second layer on the silicon substrate, the conventional ellipsometry analyzes the multilayer film structure from the air. The incident light passes through both layers and is reflected on the substrate. The amount of p-polarized light and s-polarized light when the emitted light reaches the detection unit and the change in the phase of the light wave are measured,
Thereafter, the thickness and the refractive index of each layer are known by calculation, but accurate analysis was difficult with this.

The main factor affecting the measured ellipsometric data is the product of the physical thickness of each layer and the refractive index (this product is often called the optical thickness). Since the optical properties of silicon nitride are similar, the thicker silicon oxide film and the thinner silicon nitride film contribute almost equally to the ellipsometric data and cannot be distinguished. Therefore, when the thickness of each layer is unknown, it is extremely difficult to separate and analyze the silicon oxide film and the silicon nitride film.

In support of the fact that this difficulty has actually occurred, the D using a stacked structure of a silicon oxide film and a silicon nitride film
In the research and development of electronic devices such as RAM, the value of the capacitance of the laminated structure was measured instead of the physical thickness (dimension of the absolute length) to determine the overall film thickness of the laminated structure. Later, there is an actual situation in which a film thickness converted assuming that the entire laminated structure is a silicon oxide film (so-called converted film thickness) is commonly used. This indicates that it is difficult to measure the absolute film thickness of each layer in such a laminated structure with high reliability.

As described above, in the conventional ellipsometry, in the analysis of the stacked structure of the silicon oxide film / silicon nitride film, ambiguity always arises in the determination of the film thickness of each layer and the absolute value of the refractive index. The same difficulty is also a problem in principle not only in the silicon nitride / silicon oxide system but also in the laminated structure of another inorganic transparent film.

[0014]

SUMMARY OF THE INVENTION The present invention overcomes the ambiguity of the independent determination of the thickness and refractive index of a thin film by the above-mentioned ellipsometry, and makes it possible to analyze the film structure with high reliability. It is to be. Also, the present invention overcomes the ambiguity of independent determination of the thickness and refractive index of each layer of the laminated structure of the inorganic transparent film by the above-mentioned ellipsometry, and makes it possible to separate and analyze each layer with high reliability. Is what you do.

[0015]

SUMMARY OF THE INVENTION The present invention relates to an ellipsometry for analyzing the thickness, refractive index or composition of a thin film structure formed by depositing a thin film on a substrate. Ellipsometric measurement is performed while the refractive index is continuously changed, and the discontinuity of the difference (Δ) in the phase shift of the polarized light wave in the ellipsometric measurement data in the liquid near the Brewster energy determined by the substrate and the incident angle This is an analysis method characterized in that the refractive index of a thin film is determined by detecting a temporal change.

Further, according to the present invention, in this new analysis method, a liquid mixture of two or more liquids is used as a liquid for immersing a measurement sample, and the refractive index of the liquid is adjusted by continuously adjusting the mixing ratio. This is an analysis method characterized by performing an ellipsometric measurement while performing.

The present invention also relates to an ellipsometric method for analyzing the film thickness and the refractive index or composition of each layer of a film structure in which one or more ultra-thin films are deposited on a substrate. Analysis is performed assuming that the thin film structure portion is composed of the components of the uppermost layer, and the reduced film thickness is obtained. Then, the total reduced film thickness of the thin film structure portion that may include an interface transition layer below the outermost layer film When is the reduced film thickness obtained in the air, the incident angle and the energy condition such that Δ in the ellipsometric data in the liquid changes discontinuously depending on the presence or absence of the interface transition layer by simulation calculation,
Finally, the entire measurement sample is immersed in a liquid having a refractive index equal to or higher than the refractive index of the outermost layer film, and ellipsometric measurement is performed under the conditions determined by simulation. This is an analysis method characterized by determining the composition ratio of a small transition layer at an interface with high sensitivity by detecting the composition ratio.

Further, according to the present invention, in this new analysis method, a mixture of two or more liquids is used as a liquid for immersing a measurement sample, and the refractive index is adjusted by continuously adjusting the mixing ratio. This is an analysis method characterized by performing ellipsometric measurement.

Further, according to the present invention, there is provided an apparatus for performing an ellipsometric measurement by immersing an entire measurement sample in a liquid, wherein the mixing ratio of two or more liquids having different refractive indices is continuously adjusted. This is an ellipsometric analyzer having a mechanism for adjusting the refractive index.

[0020]

DESCRIPTION OF THE PREFERRED EMBODIMENTS (Explanation of the Principle of the Present Invention) The present inventor first gave a measurement sample having a structure in which a single layer of film was deposited on a substrate to have a refractive index equal to or lower than that of the film. The simulation calculation of the measured value when the measurement sample was immersed in the liquid having a large refractive index and the polarization analysis was performed was performed. Such a measuring method is called a liquid immersion method. The ellipsometric data is usually represented by two parameters, ψ, which corresponds to the reflectance ratio of s-polarized light and p-polarized light, and Δ, which corresponds to the difference in the phase shift of the light wave when both polarized lights are reflected. Depending on the structure,
The film thickness and the refractive index are determined by fitting with the calculated value of Δ. As described in the above-mentioned "Prior Art", it has been found that in the usual analysis flow of the conventional ellipsometry, the analysis of a multilayer film composed of layers having similar optical properties is inferior in accuracy. The conventional method determines the ultrathin film structure of the model case that is considered to be difficult to analyze, and this structure is observed by the method of the present invention (本,
Δ) Data was calculated by simulation.

That is, the model structure was a structure in which silicon oxide was deposited at 50 ° on a silicon single crystal (100 plane), and the measurement was performed by the liquid immersion method while continuously changing the refractive index of the liquid (ψ, Δ) The data was simulated. In such a structure, it has been extremely difficult to separate and analyze the film thickness and the refractive index by the conventional ellipsometry using no liquid.

As a result of this simulation, it has been found that a large discontinuous change may appear in Δ in the ellipsometric data near the Brewster energy determined by the substrate and the incident angle. FIG. 1 shows a schematic diagram of the model structure used, and FIG. 2 shows a simulation result of the liquid immersion method.

The model structure will be described with reference to FIG. In this structure, a silicon oxide thin film layer having a thickness of 50 ° is deposited on a silicon wafer. The refractive index of silicon oxide is unknown because it is an extremely thin film, and detailed analysis of such a structure is usually extremely difficult and has a large error. This difficulty has evolved in recent years,
The so-called spectroscopic ellipsometry capable of scanning the measurement wavelength is essentially the same. This is because the relative wavelength dependence of the refractive index is considered to be the same regardless of the film thickness. This is because there is no change in the data characterizing the change in the refractive index, although only the tendency of the refractive index is observed.

Therefore, the analysis of the refractive index is not usually performed.
The thickness of the film is assumed to be silicon oxide with a known refractive index (referred to as “converted film thickness”), and is used for quality control of products. Actually, as described above, the refractive index of an ultra-thin film changes, and the determination of the true refractive index and the film thickness is performed by ellipsometry, which is a non-destructive, non-contact, rapid analysis means. It is an object of the present invention.

In general, in the ellipsometry, since it is mathematically impossible to directly calculate the refractive index and the film thickness of each layer from the measured data (）, Δ), a model structure is formed. The refractive index and the film thickness are obtained by fitting (ψ, Δ) obtained by simulation calculation from the model structure to the actually measured values. The model structure in FIG.
This is for a silicon oxide system of 0 °.

FIG. 2 shows an example of a simulation when the measurement is performed by the liquid immersion method. In this simulation, since the change range of the energy was narrow, the dispersion of the refractive index of the liquid was ignored, and the refractive index was constant in this range regardless of the energy. In this simulation, the dielectric constant (refractive index 2)
1 was changed from 2.1 to 2.2, ellipsometry measurement data at an incident angle of 69 ° of the model structure of FIG. 1 was calculated. As a result, as the dielectric constant of the liquid increases, the characteristic Δ in the energy range of 1.8 to 1.9 eV is obtained.
The jump was shown to happen. In FIG. 2, the dielectric constant of the liquid is shown in increments of 0.01, and it can be seen that a jump occurs between 2.12 and 2.13. It turned out to happen at the time. The incident angle and energy at which this “jump of Δ” occurs are related to each other and appear near the Brewster's angle of the substrate (in this case, silicon single crystal) at that energy. On the other hand, a jump appears in a similar area.

An analysis method for accurately determining the refractive index of a thin film utilizing this phenomenon is the present invention. As can be seen from FIG. 2, when measurement is performed in this condition in an energy region around 1.6 to 2.2 eV, a point where the refractive index of the liquid matches the thin film can be detected, and the refractive index of the solution at that point is measured. Then, the refractive index of the thin film is known.

FIG. 3 shows the value of Δ (light energy is fixed at 1.8 eV) when the dielectric constant of the liquid is gradually increased. Δ jumps between the dielectric constants of 2.12 and 2.13. Show the situation. In fact, while the silicon oxide thin film of the model structure used in this simulation has a refractive index of 1.457 at an energy of 1.8 eV, a jump of Δ occurring at the same energy causes a dielectric constant of the liquid of 2.127 (that is, 2.127). This occurs when the refractive index reaches 1.458), and it can be seen that this method is a precise method for determining the absolute value of the refractive index up to three decimal places.

Regarding ψ, which is another actually measured data of the ellipsometry, even if the refractive index and the incident angle of the liquid are adjusted, the phenomenon of a large change as shown in FIG. 3 does not occur.

Further, the present inventor has determined that, for a measurement sample having a structure in which one or more films are deposited on a substrate and an extremely thin composition change layer is provided below the uppermost film, the refractive index of the uppermost film is the same as that of the uppermost film. A simulation calculation of measured values was performed when the sample was immersed in a liquid having a refractive index that was greater than or equal to or greater than that, and the polarization analysis was performed. Ellipsometric data is usually s-polarized and p-polarized
It is represented by two parameters, ψ, which corresponds to the reflectance ratio of polarized light, and す る, which corresponds to the difference between the phase shifts of light waves when both polarized lights are reflected. This is fitted with the calculated values of ψ and Δ by the model structure. By doing so, the thickness and the refractive index of the multilayer structure are determined. As described in the above-mentioned “Prior Art”, it has been found that in the usual analysis flow of the conventional ellipsometry, analysis of a multilayer film having similar optical properties is inferior in accuracy. First, in order to clarify this point quantitatively, the multilayer structure of the model case, which seems to be difficult to analyze with the conventional method, was determined, and the (ψ, Δ) data observed with this structure was calculated by simulation.

As a model structure, a silicon single crystal (100
Basically, silicon oxide is deposited as the first layer and silicon nitride is deposited as the second layer on the surface, and an extremely thin interfacial oxynitride layer is formed at the interface between the silicon oxide layer and the silicon nitride layer by mixing both layers. The difference between the (ψ, Δ) data in the liquid immersion method in the case where it was formed and in the case where it was not formed was simulated. In such a system, the optical characteristics of silicon oxide and silicon nitride are similar in the conventional ellipsometry using no liquid, so the ellipsometry data shows that the thicker silicon oxide film and the thinner silicon nitride film are almost the same. It also contributes and is indistinguishable. Therefore, when the film thickness of each layer is unknown, it has been extremely difficult to separate and analyze the interface transition layer formed by mixing the silicon oxide film, the silicon nitride film, and both films.

As a result of this simulation, it has been found that a large discontinuous change may appear in Δ in the ellipsometric data around Brewster energy determined by the substrate and the incident angle. 4 shows a schematic diagram of the model structure used, FIG. 5 shows a simulation result of measurement in normal air, and FIG. 6 shows a simulation result by an immersion method.

The model structure will be described with reference to FIG. In this structure, first, a thin silicon nitride thin film layer is provided on a silicon wafer, and a silicon oxide thin film layer is stacked thereon. Since the refractive index of silicon nitride can be almost regarded as a constant multiple of that of silicon oxide regardless of energy, detailed analysis of such a structure is usually extremely difficult and has a large error. This difficulty is essentially the same in so-called spectroscopic ellipsometry, which can scan the measurement wavelength developed in recent years. Because the wavelength dependence of the refractive index of silicon nitride is similar to that of silicon oxide, the same tendency can be seen at any wavelength even if measurement and analysis are performed at many wavelengths by spectroscopic ellipsometry. This does not cause a change in data characterizing the multilayer structure.

Therefore, the detailed analysis of each layer is not usually performed, and the film thickness (referred to as “converted film thickness”) assuming that all the layers are made of silicon oxide is obtained and used for quality control of products. ing. In fact, it is known that an oxynitride interfacial transition layer may exist between both layers. This is the aim of the present invention.

In general, in the ellipsometry, since it is mathematically impossible to directly calculate the refractive index and the film thickness of each layer from the measured data (ψ, Δ), a model structure is formed. The refractive index and the film thickness are obtained by fitting (ψ, Δ) obtained by simulation calculation from the model structure to the actually measured values. In the model structure of FIG. 4, a silicon nitride / silicon oxide system having a total equivalent film thickness of 120 ° has an oxynitride interfacial transition layer between both layers (three-layer model) and does not exist (two-layer model). Model) are shown. FIG. 5 explains that the difference between the two models was difficult to detect by the conventional ellipsometry in air.

FIG. 5 is a plot of simulated Δ data for the structure of both the three-layer model and the two-layer model of FIG. Here, the incident angle of light is selected to be 75 °, which is the most sensitive to a silicon wafer in air (refractive index: 1.0). As is clear from the figure,
The values are almost the same for both models, indicating that it is difficult to analyze the interface transition layer with this conventional method.

On the other hand, FIG. 6 shows an example of a simulation when the measurement is performed by the liquid immersion method. As a result of various changes in the refractive index and the incident angle of the liquid, the refractive index was 1.50.
3. It was found that a large discontinuous change as shown in FIG. 6 occurred at an incident angle of 70 °. Such a large discontinuous change of Δ is hereinafter referred to as “jump of Δ”. The refractive index was chosen to match the refractive index of the upper thin film (in this example, silicon oxide) in this energy region. As can be seen from the figure, 2.3 to 2.4 e under this condition.
Measurement in the energy region near V can clarify the existence of the interface transition layer.

Δ depending on the presence or absence of the interface layer as shown in FIG.
Jump is not always observed even when using the immersion method, and occurs at a certain optimum combination of the refractive index of the liquid and the incident angle. For example, even when the incident angle is the same of 70 °, such a phenomenon does not occur when the refractive index of the liquid is 1.483, which is slightly low, and the difference in the Δ data is slight even in the immersion method as in the case of the air in FIG. Therefore, it is difficult to identify the interface transition layer.

Regarding ψ, which is another actually measured data of the ellipsometry, the phenomenon that a large change as shown in FIG. 6 does not occur even if the refractive index and the incident angle of the liquid are adjusted.

Hereinafter, the present invention will be further described. First,
The present invention is an analysis method for characterizing a thin film deposited on a substrate by ellipsometry. Basically, the method of the present invention
This is an analysis method that measures the polarization of a sample in a liquid while changing the refractive index of the liquid, finds a matching point between the refractive index of the liquid and the thin film, and obtains the refractive index of the thin film from the refractive index of the liquid at that time.

More specifically, the sample is immersed in a liquid that is considered to be smaller than the refractive index of the thin film portion, and while gradually increasing the refractive index of the liquid, the incident angle and the angle satisfying the Brewster condition with respect to the substrate are determined. Perform ellipsometric measurements with light energy. As shown in FIG. 2, for example, a certain range may be scanned. If the optimum energy point is already known, the refractive index of the solution is continuously increased by measuring only one point. The efficiency of the measurement will be further improved.

Here, as a solution, at least two kinds of liquids having at least one kind of liquid having a refractive index smaller than the expected refractive index of the thin film and one kind of liquid having a large refractive index are prepared. It is adjusted by mixing a high refractive index liquid with the liquid. It is necessary that these liquids do not chemically react with the thin film or the substrate. As shown in FIG. 2, the jump of Δ near the matching point (point of equality) between the refractive index of the liquid and the thin film is a very remarkable phenomenon.
Focusing on the energy point where the change is large, Δ at that point is 1
It is easy to find a point that jumps from less than 80 ° to more than 180 °. Since this is the matching point, the measurement is ended here, the mixed liquid is collected, and the refractive index of the liquid is measured using an apparatus such as an Abbe refractometer. This is the refractive index of the desired thin film. Once the refractive index is accurately determined, it is easy to accurately determine the absolute value of the thickness of the ultrathin thin film by ordinary ellipsometry measurement.

In the ellipsometric measurement, Δ is often represented by an angle (unit: degree), but the angle is a periodic quantity, and 360 ° and 0 ° have the same physical meaning. So is this jump really discontinuous or 3
Although it looks like a jump because it has crossed the boundary of 60 ° (0 °), it is necessary to pay attention to whether the substance is a continuous change. However, it can be seen from FIG. 2 that this is a truly discontinuous change for our approach.

That is, in FIG. 2, a curve (for example, a dielectric constant of 2.1) corresponding to the dielectric constant of the liquid in the vicinity of where the change occurs.
2 and the curve of 2.13) from the high energy side (the right side of the figure), the energies from 1.82 to 1.8.
At around 84 eV, the value of Δ suddenly changes to the high angle side or the low angle side. This is not a jump at one measurement point but a change in an energy region having a width of about 0.01 eV. Therefore, this jump is 360 ° (0
It can be seen that this is not apparently due to crossing the (°) boundary, but is actually due to a sudden (but not instantaneous) increase or decrease of Δ.

Examples of the liquid used include carbon tetrachloride and benzene having a refractive index close to that of silicon oxide. Of these two, benzene has a higher refractive index, and the refractive index can be adjusted by mixing both liquids. The idea of using a mixed liquid of carbon tetrachloride and benzene in ellipsometry by silicon oxide immersion method is described in the literature (EAI
rene and VA Yakovlev, "A New Ellipsometry Techn
ique for InterfaceAnalysis: Application to SiO2 ",
p. 81 in "The Physics and Chemistry of SiO2 and the
Si-SiO2 Interface 2 ", ed. By CR Helms and B.
E. Deal, Plenum 1993), this document and other documents do not find the “jump of Δ” referred to in the present invention, and therefore, the composition changes continuously even when a mixed solvent is used. There is no idea to seek the optimal point.

Further, the present invention provides an analysis method using the jump of Δ for determining the refractive index of a thin film by performing an ellipsometric measurement while continuously changing the refractive index of a liquid in which a measurement sample is immersed. This is an analysis method characterized by adjusting the refractive index of the liquid by continuously adjusting the mixing ratio of the liquid.

Further, the present invention is an analysis method for characterizing an interface transition layer composed of substances having similar optical properties such as silicon oxide and silicon nitride in a multilayer film structure by ellipsometry. . Basically, in the method of the present invention, the same sample is subjected to polarization measurement twice in air and liquid, and the analysis of the ultra-thin interface transition layer is performed by combining numerical simulations.

More specifically, first, an ellipsometry is performed in air, and the measured data is used to determine that the thin film structure portion that may include the interface transition layer of interest is composed of only the top layer component. Analysis of the film thickness is performed assuming that. The film thickness obtained in this way is the “converted film thickness”. Although the reduced film thickness generally does not match the actual physical film thickness, it can be obtained with very high reproducibility for one sample by ellipsometry.

Next, a model structure for analyzing the interface transition layer is created, and a simulation calculation of the liquid immersion method is performed. Here, the model structure defines the thickness and refractive index of the substrate and each layer. In this simulation, the total equivalent film thickness of the thin film structure part that may include the interface transition layer of interest is set to the equivalent film thickness obtained in the air, and under that condition, the model with the interface transition layer and the model without the interface transition layer are used. Assume that models are constructed and compared. If the incident angle and the refractive index of the liquid are changed, the condition under which the jump of Δ as described in the above “Explanation of the principle of the present invention” appears in the interface transition layer can be obtained. In this case, the refractive index of the liquid is almost the same as or slightly larger than the refractive index of the substance in the uppermost layer of the thin film.
Since the incident angle appears near the Brewster's angle of the substrate in the liquid immersion state, the jump condition can be easily obtained.

Next, under these conditions, the polarization analysis of the sample is carried out by the liquid immersion method, and by observing whether or not a jump of Δ occurs, the presence or absence of the ultra-thin interface transition layer can be first known. Furthermore, since the measured data is in a state of being highly sensitive to the refractive index and the thickness of the interface transition layer, they can be determined with high accuracy using a normal fitting program.

Further, in the present invention, in the above-mentioned analysis method using the jump of Δ, a mixture of two or more liquids is used as the liquid, and the ellipsometry is performed while continuously adjusting the mixing ratio. This is an analysis method characterized by detecting a minute change in composition at an interface with high sensitivity.
That is, in the above description, the optimum refractive index and incident angle of the liquid for the liquid immersion method were obtained by simulation calculation in the middle of the entire analysis process. Conversely, the jump condition of Δ can be found directly by the actual measurement of the polarization analysis of the liquid immersion method. In this case, for example, the incident angle is set to the Brewster angle, and the refractive index of the liquid is continuously changed while measuring Δ near the Brewster energy by the liquid immersion method.
To detect the point where the jump occurs.

More specifically, the measurement cell of the liquid immersion method is partly opened, and the measurement is performed while mixing the second liquid using a chemical instrument such as a burette. If necessary, three or more kinds of liquids can be mixed. As an example of the liquid used, carbon tetrachloride and benzene having a refractive index close to that of silicon oxide can be given. Of these two, benzene has a higher refractive index, and the refractive index can be adjusted by mixing both liquids. When the jump condition of Δ is obtained by the actual measurement of the polarization analysis by the liquid immersion method, the procedure for analyzing the interface transition layer by a simulation calculation is performed.

Further, the present invention relates to an analysis method for determining the jump condition of Δ by the actual measurement of the ellipsometry of the immersion method, wherein an ellipsometric analyzer having a continuous adjusting mechanism for continuously adjusting the mixing ratio of the liquid. It is. FIG. 7 shows the configuration. The cell body is made of a chemically stable material such as glass or Teflon, and the light entrance and exit windows are desirably made of quartz glass for optics so that they can respond to light of various wavelengths. The inside of the cell is filled with a mixed liquid, and a liquid injection part utilizing a burette for a chemical experiment or the like is provided above the cell. There is also a drain tube at the bottom for draining liquid.
A liquid mixture having a refractive index such that a jump of Δ occurs by taking in and out of the liquid using the burette and the drain tube is adjusted.

Using this apparatus, the refractive index (dielectric constant) of the liquid
When the ellipsometry is performed while continuously changing, the measurement cell of the immersion method is partly opened, and the measurement is performed while mixing the second liquid using a chemical instrument such as a burette. If necessary, three or more kinds of liquids can be mixed.

[0055]

The present invention will be described in more detail with reference to the following examples. (Example 1) For the purpose of accurately determining the refractive index of an ultra-thin silicon oxide film by the method of the present invention, a sample was prepared in which an oxide film was deposited on a silicon single crystal. The sample was immersed in a hydrogen fluoride aqueous solution (1%) for 1 minute to remove the surface natural oxide film from the silicon wafer (100 surfaces) in dry oxygen at 900 ° C. for 1 minute.
Oxidized for 0 minutes.

First, a conventional spectroscopic ellipsometry measurement was performed on this sample in air, and the refractive index of the thin film was measured using bulk SiO 2.
Assuming a typical value of 1.465 of 1.465, the film thickness was analyzed by the ordinary method and found to be 78 °. The above is not a method according to the present invention, but results obtained by a conventional method for comparison are obtained for reference.

Next, this sample was analyzed according to the procedure of the present invention. First, the liquid to be used was selected. In the method of the present invention, it is necessary to use both a liquid having a refractive index lower than the refractive index of the thin film and a liquid having a high refractive index as the liquid for the immersion method.
Carbon tetrachloride may be used as a liquid having a refractive index close to that of amorphous silicon oxide. However, in this sample, since the refractive index was expected to increase due to the thin film thickness, benzene ( Refractive index at the sodium D line wavelength 1.498) and diiodomethane (methylene iodide, refractive index 1.749 at the same wavelength) were selected. Since these liquids do not react with silicon oxide, no inconvenience such as a change in the surface state during measurement occurs. Diiodomethane is a yellow colored liquid. This is because there is an absorption band on the high energy side, and it is transparent in the low energy region measured in this embodiment.

As shown in FIG. 7, an ellipsometric analyzer having a continuous adjustment mechanism for continuously adjusting the mixing ratio of the liquid was prepared, and benzene was added to the sample cell portion at 3/3 of the cell height.
The sample was immersed in the cell for about 4 times, and an ellipsometric measurement was performed at an incident angle of 69 ° and an energy of 1.80 eV. These incident angles and energies were determined by simulation calculation to find the Brewster condition under which the jump of Δ occurs, and as points near the Brewster condition that can be measured. As a result of the measurement, the value of Δ was 169.2 °. Since Δ is smaller than 180 °, as described in the description of the principle of the present invention, it was confirmed at this point that the refractive index of the liquid was smaller than that of the thin film as expected.

Subsequently, when diiodomethane was dropped from the burette for mixing of the apparatus shown in FIG. 7 and the measurement of Δ was continued at the same energy while sufficiently stirring, Δ decreased.
Eventually, the liquid mixture in the cell became full, so an appropriate amount of liquid was discharged from the drain tube, and diiodomethane was dropped again from the burette to continue the measurement. Then, at a certain point, Δ suddenly jumped from about 5 ° to about 355 °, so the dropping of diiodomethane was immediately terminated, and the refractive index of the mixed liquid in the cell was determined by an Abbe-type refractometer. As a result, the refractive index was 1.688. As described above, the refractive index of this sample was determined to be 1.46 by ordinary ellipsometry.
Although a film thickness of 78 ° assumed to be 5 was obtained, since the result of the ellipsometry is determined by (refractive index × physical film thickness), conversely, a refractive index of 1.688 obtained by the method of the present invention was used. The true film thickness was determined to be 68 °.

For the purpose of cross-checking the film thickness by another method, the sample was cleaved, the cross section was carefully polished, and the structure near the oxide film was observed with a transmission electron microscope. As a result, the average film thickness was determined to be 65 ° ± 6 °.
It is considered that there is also an error of about 10% in the film thickness measurement by an electron microscope such as deterioration of the sample at the polishing stage due to the extremely thin film, but the value of 68 ° according to the method of the present invention coincided within the error. The film thickness 78 ° when using a thick silicon oxide film of 1.465 as the refractive index greatly deviates therefrom. This example has revealed that the method of the present invention makes it possible to analyze the absolute value of the refractive index of an ultrathin thin film by a nondestructive, noncontact and quick means.

(Example 2) For the purpose of analyzing a very thin composition change layer existing at the interface between the thin film layers of silicon nitride and silicon oxide by the method of the present invention, a laminated sample in which a transition layer may be formed was prepared. did. The sample was prepared by depositing silicon nitride on a silicon wafer (100 surface) and silicon oxide thereon by LPCVD, and then performing heat treatment in a nitrous oxide gas atmosphere. As a material gas for LPCVD, dichlorosilane and ammonia were used for a silicon nitride film, and dichlorosilane and nitrous oxide were used for a silicon oxide film. The subsequent heat treatment with nitrous oxide was performed at 800 ° C. for 20 minutes. Separately, a thin film of silicon nitride alone or silicon oxide alone was deposited under the same conditions, and the respective film thicknesses were measured by ellipsometry to be 21 ° and 78 °, respectively.

Next, this sample was analyzed according to the procedure of the present invention. First, ordinary polarization analysis was performed in air at an incident angle of 75 °, and the film thickness (converted film thickness) assuming that the entire thin film portion of the sample was silicon oxide was 118 °. Based on this, various simulations of ellipsometric data in a liquid were made based on model structures with and without a transition layer (three-layer model) at the interface between silicon nitride and silicon oxide. In the two-layer model, the thickness of the silicon nitride layer, which is the first layer, is considered to be about 21 °. Therefore, from this and the total reduced film thickness of 118 °, the (second layer) 78 ° S
iO2 / (first layer) 21 @ SiN / silicon wafer substrate (all of which have actual film thickness).

Here, the actual film thickness of silicon nitride is 7
It uses the fact that it becomes 0%. Next, in the three-layer model, it is assumed that an interface transition layer is formed between the two layers of silicon nitride and silicon oxide by a contribution of 5 ° in terms of a converted film thickness (third layer) 73 {SiO 2 / (second layer). ) 8.5ÅS
iON / (first layer) 17.5 ° SiN / silicon wafer substrate (all of which have actual film thickness).

As a result of a simulation using these models, it was found that a jump occurs in the delta value in the two-layer and three-layer models when the refractive index of the liquid is 1.503 and the incident angle is 70 °. Simulation is 1.5 eV
To 4.0 eV at 0.05 eV intervals. FIG. 8 shows the simulation result. FIG. 9 shows the result of Δ under the same conditions for reference. ψ indicates that no jump occurs.

Next, an actual measurement of an ellipsometry by a liquid immersion method under these conditions was performed. Carbon tetrachloride and benzene were used as liquids, and the two were mixed to prepare a liquid having a refractive index of 1.503 near an energy of 2.2 eV. For the mixing, a liquid continuous mixing cell for ellipsometry measurement described in the section of the embodiment of the invention was used. The refractive index of the mixed liquid was determined using an Abbe-type refractometer. As a result, when about 35 parts by volume of carbon tetrachloride and 65 parts of benzene were mixed, the refractive index was 1.50.
It was 3. The measurement was performed from 1.5 eV to 4.0 eV.
The measurement was performed at an interval of 025 eV. FIG. 10 shows the delta measurement results. Immediately from the comparison between FIG. 8 and FIG. 10, it was immediately found that an interfacial transition layer was formed in this sample. When the thickness of the transition layer was further optimized by fitting, the structure of this sample was (third layer) 72 ° Si
O2 / (second layer) 10.2ÅSiON / (first layer) 1
6.8 ° SiN / silicon wafer substrate (all in actual film thickness).

As a result of destructive analysis by secondary ion mass spectrometry and photoelectron spectroscopy, the sample was found to be approximately 5 to 15
An oxide / nitride mixed layer of Å was observed, and this example revealed that the method of the present invention enables analysis of an ultra-thin interface transition layer by non-destructive means. .

[0067]

As described above, according to the present invention, it is possible to overcome the ambiguity of independent determination of the film thickness and refractive index of a thin film by an ellipsometric method and to perform separation analysis with high reliability. It is. In addition, the present invention overcomes the ambiguity of independent determination of the thickness and refractive index of each layer of the laminated structure of the transparent film by ellipsometry, and enables analysis of the ultrathin composition change region at the interface with high reliability. Is what you do. This enables nondestructive and precise analysis of the refractive index and the ultra-thin interface transition layer, and speeds up optimization of process conditions and the like.

[Brief description of the drawings]

FIG. 1 is a diagram of a model structure for describing an analysis method for accurately determining a refractive index of an extremely thin film according to the present invention.

FIG. 2 is a diagram showing the results of a simulation calculation of the energy dependence of Δ in liquids having various dielectric constants with respect to the model structure of FIG. 1;

FIG. 3 shows an energy of 1.8 e for the model structure of FIG.
The value of Δ at V changes as the dielectric constant of the liquid increases, and d (2.10) and the like in the figure show the curve of Δ when the dielectric constant of the liquid is a numerical value in parentheses. It is a figure showing the simulation result shown.

FIG. 4 is a view showing a model structure for explaining a method of analyzing an extremely thin interface composition change region in the present invention.

FIG. 5 is a diagram showing simulated Δ data for the structure of both the three-layer model and the two-layer model of FIG. 4 for measurement in air.

FIG. 6 is a diagram showing simulated Δ data for the structure of both the three-layer model and the two-layer model of FIG. 4 for measurement in a liquid having a refractive index of 1.503.

FIG. 7 is a configuration diagram of an ellipsometric analyzer having a continuous adjustment mechanism for continuously adjusting the mixing ratio of liquids.

FIG. 8 is a diagram showing a simulation result of Δ for a laminated sample in which a transition layer is formed at an interface between a thin film layer of silicon nitride and silicon oxide used in Example 2.

FIG. 9 is a diagram illustrating a simulation result of Δ for a stacked sample in which a transition layer is generated at the interface between the silicon nitride and silicon oxide thin film layers used in Example 2.

FIG. 10 is a diagram showing a measurement result by a liquid immersion method of Δ for a laminated sample in which a transition layer is generated at an interface between a silicon nitride and a silicon oxide thin film layer used in Example 2.

Claims (5)

[Claims] In an ellipsometry for analyzing the film thickness and the refractive index or composition of a film structure in which a thin film is deposited on a substrate, the entire measurement sample is immersed in a liquid to continuously change the refractive index of the liquid. The ellipsometric measurement is performed while detecting the point in time at which the discontinuous change of the phase shift difference of the polarized light wave in the ellipsometric measurement data in the liquid near the Brewster energy determined by the substrate and the incident angle occurs. Ellipsometry characterized by determining the refractive index of a thin film by measuring the refractive index of a liquid at a point in time. 2. The ellipsometry according to claim 1, wherein a mixture of two or more liquids is used as a liquid for immersing the measurement sample, and the refractive index is adjusted by continuously adjusting the mixing ratio. Ellipsometry, characterized by performing ellipsometry measurements while performing. 3. An ellipsometric method for analyzing the thickness, refractive index, or composition of each layer of a film structure in which one or more ultra-thin films are deposited on a substrate. Analysis is performed assuming that it is composed of the components of the uppermost layer, and the reduced film thickness is obtained. Then, the total reduced film thickness of the thin film structure portion that may include the interface transition layer under the outermost layer film is calculated in air. The incident angle and energy conditions are calculated by simulation to find that the difference in the phase shift of the polarized light wave in the ellipsometric data in the liquid changes discontinuously depending on the presence or absence of the interface transition layer when the reduced film thickness is obtained in Finally, the entire measurement sample is immersed in a liquid having a refractive index equal to or higher than the refractive index of the outermost layer film, and ellipsometric measurement is performed under the conditions determined by simulation. An ellipsometric method characterized in that the composition ratio of a small transition layer at an interface is determined with high sensitivity by detecting a discontinuous change in a phase shift difference between polarized light waves. 4. The ellipsometric method according to claim 3, wherein a liquid mixture of two or more liquids is used as a liquid for immersing the measurement sample, and the mixture ratio is continuously adjusted to obtain a refractive index of the liquid. Ellipsometry characterized by performing an ellipsometric measurement while adjusting the temperature. 5. An apparatus for performing an ellipsometric measurement by immersing an entire measurement sample in a liquid, wherein a mechanism for continuously adjusting a mixing ratio of two or more liquids having different refractive indexes to adjust the refractive index. An ellipsometric measurement device having: