KR20240095427A - Cleaning method of silicon wafer and manufacturing method of silicon wafer with natural oxide film - Google Patents

Abstract

The present invention includes an SC1 cleaning process of a test wafer in which wafers with different surface roughness are produced by cleaning by changing the SC1 cleaning conditions, a process of removing the SC1 oxide film formed by SC1 cleaning by hydrofluoric acid cleaning, and cleaning with a cleaning liquid having oxidizing power. From the process of forming a natural oxide film, the process of acquiring the correlation between the surface roughness formed by SC1 cleaning and the film thickness of the natural oxide film, and the correlation between the film thickness of the natural oxide film to be formed for the wafer that is the target of natural oxide film formation, A process of determining the surface roughness and the SC1 cleaning conditions that result in this surface roughness, a process of cleaning the SC1 using the determined SC1 cleaning conditions, a process of removing the SC1 oxide film formed by the SC1 cleaning, and using a cleaning liquid with oxidizing power to remove the SC1 oxide film. This is a silicon wafer cleaning method that includes a process of forming a natural oxide film by cleaning the wafer from which the oxide film has been removed. Accordingly, a method for cleaning a silicon wafer is provided in which the film thickness of the natural oxide film is controlled with good precision and reproducibility by adjusting the surface roughness.

Description

Cleaning method of silicon wafer and manufacturing method of silicon wafer with natural oxide film

The present invention relates to a method of cleaning a silicon wafer and a method of manufacturing a silicon wafer with a natural oxide film.

In the manufacturing process of single crystal silicon wafers for semiconductor devices, the main surface is finished in a polishing process. Furthermore, there is a cleaning process to remove abrasives and metal impurities attached to the surface of the silicon wafer during the polishing process. In this cleaning process, a cleaning method called RCA cleaning is used.

This RCA cleaning is a cleaning method that combines SC1 (Standard Cleaning 1) cleaning, SC2 (Standard Cleaning 2) cleaning, and DHF (Diluted Hydrofluoric Acid) cleaning according to the purpose. SC1 cleaning is a cleaning method using an alkaline cleaning solution mixed with ammonia water and hydrogen peroxide in a random ratio. It lifts off attached particles by etching the surface of the silicon wafer and further utilizes the electrostatic repulsion between the silicon wafer and particles. , It is a cleaning method that removes particles while suppressing re-adhesion to the silicon wafer. Additionally, SC2 cleaning is a cleaning method that dissolves and removes metal impurities on the surface of a silicon wafer using a cleaning solution that mixes hydrochloric acid and hydrogen peroxide in an arbitrary ratio. Additionally, DHF cleaning is a cleaning method that removes the natural oxide film on the surface of a silicon wafer using dilute hydrofluoric acid. Furthermore, ozone water cleaning, which has strong oxidizing power, is also sometimes used to remove organic matter adhering to the surface of a silicon wafer or to form a natural oxide film on the surface of the silicon wafer after DHF cleaning. The surface quality of the silicon wafer after cleaning, such as particles and surface roughness, is important, and a combination of these cleaning processes is performed depending on the purpose.

Semiconductor devices such as MOS (Metal Oxide Semiconductor) capacitors and transistors are formed on the surface of a semiconductor silicon wafer. Insulating films such as gate oxide films formed in these semiconductor devices are used under high electric field strengths, and silicon oxide films, which are easy to form, are often used as these insulating films.

A method for evaluating the film thickness of the oxide film on a silicon wafer includes measurement using an ellipsometer. An ellipsometer is used to determine the phase difference (Δdelta) and amplitude ratio (Ψpsi) by injecting polarized light into a substrate sample and measuring the change in polarization state of the incident light and reflected light. Taking a silicon oxide film on a silicon wafer as an example, the polarization state changes as incident light is reflected by the silicon oxide film on the outermost surface and the interface between the silicon oxide film and the silicon wafer. On the other hand, in ellipsometers, there is a short-wavelength type that uses a laser as a light source and a spectral type that includes multiple wavelength components and uses a white light source, and the short-wavelength type uses delta and psi for a specific wavelength (for example, 633 nm). In contrast to measuring, the spectral type can measure delta and psi for each wavelength, and it is known that using the spectral type with a large amount of information allows for more accurate evaluation of film thickness.

As described above, the information obtained by measurement with an ellipsometer is the phase difference and amplitude ratio, and the film thickness cannot be directly obtained. To determine the film thickness, a model based on the substrate sample is created, and the delta and psi theoretically obtained from this model are compared with the delta and psi obtained by measurement with an ellipsometer. On the other hand, the model is created by setting conditions according to the physical properties of the sample, and the items of the conditions to be set include the materials of the substrate and film, the film thickness of each film layer, and the optical constants of the substrate and film. In addition, for setting each item, a known reference according to the sample, a required dispersion equation that shows the wavelength dependence of the dielectric constant and has a plurality of parameters, etc. are usually used.

Furthermore, a process of changing the parameters of the dispersion equation and the film thickness of each film layer of the model is performed so that the degree of difference between the two is minimized for the above comparison (also called fitting). The difference between the two is usually obtained by calculation using the least squares method, and when the result obtained by the least squares method is judged to be somewhat small through fitting, the refractive index and extinction coefficient of the film are calculated from the values of the parameters of the dispersion equation at that time. The film thickness can be obtained by specifying the film thickness at that time as the film thickness of the film of the sample. On the other hand, model creation, fitting, etc. are generally performed manually or automatically using a computer based on the required program.

When irregularities (also referred to as roughness or roughness) exist on the surface of a sample, a method called effective medium approximation may be used (for example, Patent Document 1, etc.). This method improves the calculation results of the least squares method by defining the roughness and voids as one plane layer. In addition, the effective medium approximation may be applied not only to the case where roughness exists on the film surface of the sample, but also to the interface layer in cases where roughness exists at the interface between the substrate and the film or the interface between film layers. Furthermore, the effective medium approximation is sometimes used to lower the value of the refractive index as a technique in performing analysis, regardless of the presence of roughness. Naturally, by using the effective medium approximation, the calculation results of the least squares method also change, and as a result, the value of the film thickness also changes, so the operator determines whether to use the effective medium approximation, for example, from the calculation results of the least squares method. Needs to be.

Patent Document 2 describes that the film thickness of a native oxide film on a silicon wafer obtained with an ellipsometer changes depending on the surface roughness. Specifically, the rougher the surface, the thicker the film thickness, and a method for quantitatively evaluating surface roughness based on the correlation between surface roughness and the film thickness of a natural oxide film is disclosed.

Additionally, AFM (Atomic Force Microscope) is known as a method for evaluating surface roughness on a silicon wafer. As an indicator of surface roughness, arithmetic mean height such as Ra value or Sa value is often used. Ra is the arithmetic mean height at the standard length and is a two-dimensional roughness index, and Sa is a parameter that extends Ra to a surface and is a three-dimensional roughness index. As a method of evaluating roughness in more detail, conversion to the spatial frequency domain may be performed through spectral analysis. This method can extract components at specific wavelengths from the measured surface profile, for example parameters relating to specific spatial wavelengths and amplitude intensity at those wavelengths, such as PSD (Power Spectrum Density). It is expressed as In this way, by performing PSD analysis, it is possible to specify the spatial frequency of the dominantly formed roughness. Additionally, the Haze value obtained by a particle counter using the laser scattering method can be used as an indicator of roughness. Haze is expressed as so-called cloudiness and is widely used as an indicator of the roughness of the silicon surface. A high Haze level indicates that the surface of the wafer is rough.

A dense silicon oxide film with high insulating properties is produced by thermal oxidation of a silicon wafer. Since a natural oxide film formed by cleaning exists on a silicon wafer upon shipment from the viewpoint of particle adhesion, etc., thermal oxidation is performed on a silicon wafer with a natural oxide film formed on it. In many cases, it is processed. At this time, it is known that the thickness of the thermal oxidation film is affected by the film quality (film thickness or structure) of the natural oxide film before thermal oxidation.

Recently, with the miniaturization and multilayering of semiconductor integrated circuits, further thinning of various films including insulating films constituting devices is required. By this thinning, it is necessary to form an ultra-thin insulating film, that is, a silicon oxide film, uniformly and with good reproducibility within the plane or between substrates. For this purpose, it is required to control the film quality, especially the film thickness, of the natural oxide film at the time of shipment of the silicon wafer, which affects the quality of the silicon oxide film. In general, the thicker the natural oxidation film, the thicker the thermal oxidation film. If you want to make the thermal oxidation film thin, it is better for the natural oxide film to be thinner, and if you want to make the thermal oxidation film thick, it is better for the natural oxide film to be thicker. Therefore, controlling the film thickness of the natural oxide film within a certain range with good reproducibility has been particularly desired in recent years.

Patent Document 3 describes the relationship between a silicon wafer cleaned under various conditions and the film thickness of the oxide film after thermal oxidation. Specifically, when the NH ₄ OH concentration of the SC1 cleaning solution is increased, the amount of OH groups contained in the natural oxide film increases and the film thickness after thermal oxidation becomes thicker. The composition (film quality) of the natural oxide film and the film thickness after thermal oxidation A method of controlling the film thickness after thermal oxidation is disclosed by using the correlation with .

Japanese Patent Publication No. 2005-283502 Japanese Patent Publication No. H6-163662 Japanese Patent No. 6791453 Publication

As described above, it is required to control the film thickness of the natural oxide film and the thermal oxide film on the silicon wafer. Generally, in the manufacturing process of silicon wafers, the surface roughness of the wafer is formed through polishing and subsequent cleaning. SC1 cleaning, hydrofluoric acid cleaning, or ozone water cleaning are used to clean the wafer after polishing. It is known that the cleaning process mainly uses SC1 cleaning, which has an etching effect, to roughen the surface.

Patent Document 3 describes the surface roughness Ra after SC1 cleaning or ozone water cleaning, and the value is about 0.06 to 0.12 nm. This Ra value is the roughness value of recently used silicon wafers.

Patent Document 2 discloses that the surface roughness of the wafer affects the thickness of the natural oxide film measured with an ellipsometer, and the surface roughness value at this time is 0.22 to 2.05 nm as the Ra value of AFM, and the above-described It is very high compared to the surface roughness value of 0.06~0.12nm.

In addition, it is generally known that the film thickness of the natural oxide film is about 1 nm. In Patent Document 2, when the Ra value is 0.22 nm, the film thickness of the natural oxide film is 0.097 nm, and when the Ra value is 1.23 nm, the film thickness of the natural oxide film is 1.586 nm. , when the Ra value is 2.05 nm, the film thickness of the natural oxide film is 3.313 nm, which is far from the total film thickness of about 1 nm. In this way, the surface roughness and the film thickness of the natural oxide film of Patent Document 2 are significantly different from the surface roughness and the film thickness of the natural oxide film of recently used silicon wafers. The reason for this is thought to be that, in the invention described in Patent Document 2, treatment to intentionally roughen the surface is performed using a mixed solution of hydrofluoric acid and nitric acid, which is not used as a normal cleaning solution for silicon wafers. That is, using the correlation disclosed in Patent Document 2, it is difficult to discuss, for example, the roughness in the range of Ra of 0.06 to 0.12 nm and the thickness of the natural oxide film, and only, for example, the Ra value of 1 nm. It is assumed that it can be applied in very rough cases exceeding .

Here, paying attention to the Ra value of AFM when assigning the NH ₄ OH concentration of SC1 cleaning described in Patent Document 3 and the thickness of the thermal oxidation film obtained by the spectroscopic ellipso method, the higher the NH ₄ OH concentration, the higher the AFM The Ra value of is high, and the thickness of the thermal oxidation film also tends to be thick (FIG. 9 of Patent Document 3). In Patent Document 3, the composition (film quality) of the natural oxide film (chemical oxide film) formed in the cleaning process, for example, the amount of OH groups measured by ATR (Attenuated Total Reflectance)-FT (Fourier Transform)-IR (Infrared Spectoroscopy) method. It is disclosed that there is a correlation with the thickness of the thermal oxidation film, and that the higher the NH ₄ OH concentration, the greater the amount of OH groups and the thicker the thermal oxidation film.

However, as described above, the result can also be interpreted that there appears to be a correlation between the Ra value obtained by AFM measurement and the film thickness after thermal oxidation obtained by an ellipsometer. As such, there is no known literature that describes the effect of the surface roughness of the wafer formed in the recently used silicon wafer manufacturing process on the thickness of the natural oxide film and thermal oxide film obtained with an ellipsometer. For example, if the wafer surface roughness formed in the manufacturing process of a silicon wafer, such as an Ra value of 0.06 to 0.12 nm, is one of the factors affecting the thickness of the oxide film, in controlling the film thickness of the natural oxide film and the thermal oxide film, It can be considered to be an equally important quality as the composition (film quality) of the natural oxide film described above. Additionally, it is considered useful if the film thickness of the native oxide film formed after cleaning can be controlled by appropriately adjusting the surface roughness of the wafer.

Accordingly, the present invention was made to solve the above problems, and its purpose is to provide a cleaning method for a silicon wafer that can control the film thickness of the native oxide film with good precision and reproducibility by adjusting the surface roughness of the wafer.

The present invention has been made to achieve the above object, and is a method of cleaning a silicon wafer, in which a plurality of test silicon wafers are prepared in advance, and the surface roughness is different by changing the SC1 cleaning conditions and performing SC1 cleaning of the test silicon wafers. An SC1 cleaning process of the test silicon wafer to produce multiple levels of the test silicon wafer, an SC1 oxide film removal process of the test silicon wafer to completely remove the SC1 oxide film formed in the test SC1 cleaning process by hydrofluoric acid cleaning, and , a natural oxide film forming process of the test silicon wafer to form a natural oxide film by cleaning the test silicon wafer from which the SC1 oxide film has been removed using a cleaning solution having oxidizing power, and a surface roughness formed by the SC1 cleaning of the test silicon wafer. A correlation acquisition process for acquiring the correlation between the film thickness of the natural oxide film formed in the natural oxide film formation process of the test silicon wafer, and the natural oxide film formed by cleaning the silicon wafer subject to natural oxide film formation with a cleaning liquid having the oxidizing power. The surface roughness to be formed on the silicon wafer, which is the target of the natural oxide film formation, is determined based on the correlation obtained in the correlation acquisition step so that the film thickness of the oxide film becomes a predetermined thickness, and SC1 becomes the determined surface roughness. An SC1 cleaning condition determination process of determining cleaning conditions, an SC1 cleaning process of the silicon wafer subject to natural oxide film formation, wherein SC1 cleaning of the silicon wafer subject to natural oxide film formation is performed using the SC1 cleaning conditions determined in the SC1 cleaning condition determination process, A SC1 oxide film removal process of the silicon wafer subject to the natural oxide film formation, in which the silicon wafer subject to the natural oxide film formation after the SC1 cleaning process is cleaned with hydrofluoric acid to completely remove the SC1 oxide film formed by the SC1 cleaning, and a cleaning solution having the oxidizing power is used. Thus, by providing an oxide film forming process of the silicon wafer, which is the object of natural oxide film formation, by cleaning the silicon wafer, which is the object of natural oxide film formation, from which the SC1 oxide film has been removed, to form a natural oxide film, the film of the natural oxide film of the silicon wafer formed by cleaning. A method for cleaning silicon wafers that controls thickness is provided.

According to this method of cleaning a silicon wafer, the film thickness of the natural oxide film can be controlled with good precision and reproducibility by using the correlation between the surface roughness of the wafer and the film thickness of the natural oxide film.

At this time, the SC1 cleaning conditions can be a silicon wafer cleaning method that requires one or more of the SC1 chemical solution concentration, cleaning temperature, and cleaning time.

Since these are conditions that are easy to change even in realistic operations, cleaning conditions can be easily set.

At this time, the surface roughness can be a roughness component with a spatial frequency of 60 to 90/μm.

Since these roughness components have a greater influence on the thickness of the oxide film, the stable film thickness can be evaluated and controlled with greater precision.

At this time, the indicator of the surface roughness can be the Haze value of the particle counter.

Since the Haze value can be easily acquired using a particle counter, the throughput is very high and cleaning conditions can be set quickly and easily.

At this time, the surface roughness index can be the average value of the power spectrum density with a spatial frequency of 60 to 90/μm.

Accordingly, since the roughness can be evaluated in detail, the stable film thickness can be evaluated and controlled with greater precision.

At this time, ozone water or hydrogen peroxide water can be used as the cleaning liquid having the oxidizing power.

Accordingly, ozone water and hydrogen peroxide have strong oxidizing power and can uniformly oxidize the wafer surface, allowing an oxide film to be easily and stably formed.

At this time, the method for manufacturing a silicon wafer with a natural oxide film can be used to manufacture a silicon wafer with a natural oxide film using the cleaning method for a silicon wafer according to the present invention.

Accordingly, it is possible to manufacture a silicon wafer with a natural oxide film while controlling the film thickness with high precision and high reproducibility.

As described above, according to the silicon wafer cleaning method of the present invention, it becomes possible to control the film thickness of the natural oxide film with good precision and reproducibility by using the correlation between the surface roughness of the wafer and the film thickness of the natural oxide film.

1 is a flow chart showing an example of a silicon wafer cleaning method according to the present invention.
Figure 2 shows a flow chart of an investigation of the relationship between the surface roughness of a silicon wafer and the film thickness of the oxide film.
Figure 3 shows a graph showing the relationship between the surface roughness (Haze) of a silicon wafer subjected to the roughening process of Figure 2 using CMP and SC1 cleaning, and a natural oxide film and a 5 nm oxide film.
FIG. 4 shows a graph showing the relationship between the surface roughness (Haze) of a silicon wafer subjected to the roughening process of FIG. 2 by single-wafer cleaning, and a natural oxide film and a 5 nm oxide film.
Figure 5 shows the roughening treatment of Figure 2 when cleaning with a liquid composition of NH ₄ OH:H ₂ O ₂ :H ₂ O=1:1:10, a cleaning temperature of 80°C, and a cleaning time of 0, 3, 6, and 12 min. A graph showing the relationship between cleaning time, Haze, and the thickness of the natural oxide film is shown.
Figure 6 shows the AFM measurement results and PSD curves of each sample.
Figure 7 is a diagram showing the relationship between SC1 cleaning conditions and the film thickness of the natural oxide film of the sample after hydrofluoric acid cleaning and ozone water cleaning.
Figure 8 is a diagram showing the relationship between the amount of Haze increase in each sample and the film thickness of the natural oxide film.

Hereinafter, the present invention will be described in detail, but the present invention is not limited thereto.

As described above, there has been a demand for a silicon wafer cleaning method that can control the film thickness of the native oxide film with good precision and reproducibility by adjusting the wafer surface roughness. In order to solve this problem, the present inventors investigated the relationship between the roughness formed in the manufacturing process of silicon wafers, specifically the roughness formed in the polishing process and cleaning process, and the film thickness of the natural oxide film and thermal oxidation film measured with an ellipsometer. We paid attention to this and carefully reviewed it. As a result, when the average value of the power spectrum density (intensity) of a certain frequency band exceeds a predetermined value, the film thickness of the oxide film becomes thick, and the film thickness of the natural oxide film can be controlled by adjusting this specific roughness component. discovered and completed the present invention.

That is, the present inventors, as a cleaning method for silicon wafers, prepare a plurality of test silicon wafers in advance, change the SC1 cleaning conditions, and perform SC1 cleaning of the test silicon wafers, so that the test silicon wafers have multiple levels of different surface roughness. The SC1 cleaning process of the test silicon wafer for producing the test silicon wafer, the SC1 oxide film removal process of the test silicon wafer to completely remove the SC1 oxide film formed in the SC1 cleaning process of the test silicon wafer by hydrofluoric acid cleaning, and using a cleaning solution having oxidizing power. Thus, the natural oxide film forming process of the test silicon wafer to form a natural oxide film by cleaning the test silicon wafer from which the SC1 oxide film has been removed, the surface roughness formed in the SC1 cleaning of the test silicon wafer and the natural oxide film of the test silicon wafer A correlation acquisition process for acquiring the correlation with the film thickness of the natural oxide film formed in the oxide film formation process, and the film thickness of the natural oxide film formed by cleaning the silicon wafer, which is the object of natural oxide film formation, with a cleaning solution having the oxidizing power, is determined by a predetermined amount. SC1 cleaning to determine the surface roughness formed on the silicon wafer, which is the target of the natural oxide film formation, based on the correlation obtained in the correlation acquisition process so that the thickness is obtained, and to determine the SC1 cleaning conditions that become the determined surface roughness. a condition determination process, an SC1 cleaning process of the silicon wafer subject to the natural oxide film formation, in which SC1 cleaning of the silicon wafer subject to the natural oxide film formation is performed under the SC1 cleaning conditions determined in the SC1 cleaning condition determination process, and the natural oxide film forming process is performed after the SC1 cleaning process. A SC1 oxide film removal process of the silicon wafer subject to natural oxide film formation, in which the silicon wafer subject to oxide film formation is cleaned with hydrofluoric acid to completely remove the SC1 oxide film formed by the SC1 cleaning, and the SC1 oxide film is removed using a cleaning liquid having the oxidizing power. A silicon wafer that controls the film thickness of the natural oxide film of the silicon wafer formed by cleaning by providing an oxide film forming process of the silicon wafer, which is the object of natural oxide film formation, by cleaning the silicon wafer, which is the object of natural oxide film formation, to form a natural oxide film. The present invention was completed by discovering that the film thickness of the natural oxide film can be controlled with good precision and reproducibility by using a cleaning method.

Hereinafter, description will be made with reference to the drawings.

First, the relationship between various surface roughnesses formed in the silicon wafer manufacturing process and the film thickness of the oxide film is described. Figure 2 is a flow chart of the investigation. For the prepared silicon wafers, the CMP processing conditions and SC1 cleaning conditions were changed, and CMP processing or SC1 cleaning as a roughening treatment to form roughness was performed to prepare silicon wafers of multiple levels. Next, the oxide film was completely removed by hydrofluoric acid cleaning using a batch scrubber, and then the oxide film was formed by ozone water cleaning. Since the oxide film formed by the SC1 cleaning of the roughening process is completely removed by hydrofluoric acid cleaning at any level, and the oxide film is formed by the subsequent ozone water, it can be interpreted that the oxide film is formed by the same method on multiple levels of silicon wafers. After performing Haze measurement using a particle counter, some of the wafers were thermally oxidized with a film thickness of 5 nm, and the film thicknesses of the natural oxide film and the oxide film formed with a 5 nm target were evaluated by spectroscopic ellipsometry.

FIG. 3 is a graph showing the relationship between the surface roughness (Haze) of a silicon wafer subjected to the roughening process of FIG. 2 using CMP and SC1 cleaning and the film thickness of the native oxide film and the 5 nm target oxide film. At the CMP level (■), even when the Haze value exceeded 10 ppm, the film thicknesses of the natural oxide film and the 5 nm target oxide film were the same, but at the SC1 cleaning level (●), when the Haze increased, the thickness of either the natural oxide film or the 5 nm target oxide film A tendency toward thickening was also obtained. Since the oxide film was formed under the same conditions, the film thickness was assumed to be equivalent to the CMP level, but this was not the case for the SC1 cleaning level.

Furthermore, FIG. 4 shows the relationship between the surface roughness (Haze) of a silicon wafer and the film thickness of the native oxide film and the target oxide film of 5 nm when the roughening process of FIG. 2 was performed by sheet wafer cleaning combining hydrofluoric acid and ozone water cleaning. In addition, hydrofluoric acid cleaning and ozone water cleaning after roughening treatment were performed using a sheet method rather than a batch method. In this case, since the oxide film formation method is different from the batch-type ozonated water, a comparison cannot be made between the SC1 and CMP levels described above and the film thickness of the single-wafer cleaning level, but the effect of Haze within the single-wafer cleaning level is discussed. can do. As a result, the sheet wafer cleaning level was the same as the CMP level, and the film thicknesses of the natural oxide film and the 5nm target oxide film were equivalent even if the Haze changed. Summarizing the above results, it was newly found that the roughness formed by CMP and sheet wafer cleaning does not affect the film thickness of the oxide film, but the surface roughness of the substrate formed by SC1 cleaning influences the film thickness of the oxide film to be thicker. .

Accordingly, the results of additional investigation into the SC1 cleaning level will be explained. The SC1 cleaning of the roughening treatment was performed as a batch cleaning with a liquid composition of NH ₄ OH:H ₂ O ₂ :H ₂ O=1:1:10, a cleaning temperature of 80°C, and a cleaning time of 3, 6, and 12 min, followed by hydrofluoric acid cleaning. Figure 5 shows the results showing the film thickness and Haze value of the natural oxide film after completely removing the oxide film and performing ozone water cleaning. With respect to the film thickness of 1.207 nm of the natural oxide film without cleaning time (set to 0 min), which is Ref., it became 1.258 nm at 3 min, 1.258 nm at 6 min, and 1.261 nm at 12 min. Since the difference between the average value of the film thickness of 1.259 nm for cleaning times of 3, 6, and 12 min and the film thickness of 1.207 nm without cleaning (cleaning time of 0 min) is 0.052 nm, the amount of film thickening for the SC1 cleaner in the roughening treatment is approximately 0.052 nm. It is thought that Haze tends to increase as the cleaning time is longer, but considering that the amount of film thickening at cleaning times of 3, 6, and 12 min is the same, the specific roughness component formed by SC1 cleaning affects the thickening behavior of the film thickness, and 3 The fact that the film thicknesses of , 6, and 12min are equal is thought to mean that the roughness component involved in film thickness (film thickening) is equal.

To verify these, the surface roughness of wafers at representative CMP, SC1 cleaning, and sheet wafer cleaning levels was evaluated using AFM (atomic force microscopy). The observation field of view was 1 μm × 1 μm, and in addition to the three-dimensional arithmetic mean height Sa, a PSD curve was obtained from spectral analysis of the surface profile data. Figure 6 shows the AFM measurement results and PSD curves of each sample.

First, pay attention to the CMP level. At the CMP level, the level with the smallest Haze value (CMP-1) and the level with the largest Haze value (CMP-2) were evaluated in Figure 3. In CMP-2, the power spectrum density (intensity) in the low frequency band (1 to 10/μm) was very high, and the roughness on the low frequency side was dominant. The AFM image was also consistent in that an image resembling a large wave was obtained. As described above, since the roughness formed by CMP does not affect the film thickness of the oxide film, it can be said that this low-frequency component does not affect the film thickness of the oxide film.

Next, pay attention to the SC1 cleaning level. At the SC1 cleaning level, the liquid composition is NH ₄ OH:H ₂ O ₂ :H ₂ O=1:1:10, and the cleaning temperature and cleaning time are set at 3 levels: 60℃/3min, 80℃/3min, and 80℃/12min. evaluated. If these oxide films are formed using the same method, the film thickness becomes thicker by the same amount. As shown in Figure 6, the roughness of all three levels is dominant in the high frequency range (10 to 100/μm) compared to CMP-2, which is consistent with the fact that fine grain-like roughness is obtained in the AFM image. Therefore, it can be said that the roughness component (spatial frequency band) formed in CMP and SC1 cleaning is significantly different.

Lastly, pay attention to the level of sheetfed cleaning. At the sheet-fed cleaning level, in Figure 4, the level with the smallest Haze value (sheet-fed cleaning-1) and the level with the highest (sheet-fed cleaning-2) were evaluated. The power spectrum density (intensity) of the sheet wafer cleaning level was intermediate between CMP and SC1 cleaning in the high frequency range (10~100/μm).

These roughness evaluation results and their influence on the film thickness of the oxide film are considered. In particular, while the Sa value of both SC1 cleaning - 80℃/3min and single wafer cleaning - 2 is 0.108nm, the oxide film thickness becomes thicker in SC1 cleaning - 80℃/3min, and thicker in single wafer cleaning -2. Pay attention to the results that do not give up. Looking at both PSD curves, while the power spectral density (intensity) in the low frequency band (1 to 10/μm) is equal, the power spectral density (intensity) in the high frequency band (especially 50/μm or more) is SC1 cleaning. -80℃/3min is larger than sheet cleaning-2. Therefore, although there is no significant difference between the two on the AFM, it can be said from the PSD curve that the roughness in the high frequency range is more dominant in the SC1 cleaning - 80°C/3min. Furthermore, considering that the three levels of SC1 cleaning shown (60℃/3min, 80℃/3min, 80℃/12min) are all thickened by the same thickness (approximately 0.05nm), the power in the spatial frequency range of 60~90/μm It can be seen that the spectral density (intensity) is the same for all three levels, and is also higher than that of Sheet wafer cleaning-2. Therefore, it became newly clear that the roughness component in the 60 to 90/μm range is a roughness component that affects the film thickness of the natural oxide film and the 5 nm target oxide film. By evaluating and controlling the surface roughness of these roughness components, it is possible to control the film thickness with greater precision and stability.

Meanwhile, the power spectrum density (intensity) of the spatial frequency band of 50/μm or less is within the 3 levels of SC1 cleaning,

80℃/12min > 80℃/3min > 60℃/3min

becomes the magnitude relationship, and the magnitude relationship of the Sa value,

80℃/12min > 80℃/3min > 60℃/3min

This is also consistent with, and the Sa value in this case is dominated by the roughness information on the low-frequency side with high intensity. Despite the above-mentioned sheet wafer cleaning-2 and SC1 cleaning-80°C/3min having the same Sa value of 0.108 nm, the oxide film It is thought that this is the reason for the difference in film thickness.

To summarize the above results, it is believed that if the average value of the power spectrum density (intensity), which is a roughness component in the spatial frequency band of 60 to 90/μm, is above the threshold, the oxide film becomes thick. Here, if we calculate the average value of the power spectrum density (intensity) in the range of 60~90/μm of the three levels of SC1 cleaning, it is 0.16nm ³ at SC1 cleaning-60℃/3min, 0.18nm ³ at SC1 cleaning-80℃/3min, In SC1 cleaning - 80℃/12min, there was 0.17nm ³ . On the other hand, the average value of sheet-fed cleaning-2, in which the oxide film did not thicken as described above, was 0.11 nm ³ . Therefore, the average value of the power spectral density (intensity) of 60 to 90/μm of 0.15 nm ³ is considered to be the critical value, and the film thickness of the silicon oxide film on the silicon wafer where the average value of the power spectral density (intensity) of 0.15 nm ³ or more exists is It can be determined that the surface roughness of the wafer has an influence, and the film thickness resulting from the surface roughness of the substrate is included.

From the above knowledge, it can be interpreted that the surface roughness of a specific wafer is a factor that affects the film thickness of the oxide film (film thickness influencing factor). Here, the factors influencing the film thickness of the natural oxide film and the thermal oxidation film include, for example, the structure of the natural oxide film (chemical oxide film) described in Patent Document 3 and the surface roughness of the silicon wafer described so far. You can.

Here, we consider the surface roughness of the wafer after SC1 cleaning and the structure of the natural oxide film. First of all, SC1 cleaning is a cleaning in which the reaction of oxidizing Si with hydrogen peroxide and the reaction of etching the oxidized SiO ₂ are always in progress. For example, this oxidation and Etching behavior changes. Meanwhile, under general conditions, the etching reaction of SiO ₂ is rate-dependent, so the surface is always covered with a natural oxide film. Since the film thickness of the native oxide film after SC1 cleaning varies depending on the balance of oxidation and etching reactions, it is necessary to control the oxidation and etching reactions to control the film thickness of the native oxide film after SC1 cleaning. This oxidation and etching behavior affects both the wafer surface roughness and the structure of the native oxide film, as described in Patent Document 3. In other words, it can be said that in order to control the thickness of the natural oxide film after SC1 cleaning and the film thickness after thermal oxidation, it is necessary to control both of these factors. In other words, when a minor unintended change in cleaning conditions occurs, the variation in the film thickness of the natural oxide film increases. In particular, the structure of the natural oxide film greatly affects the film thickness after thermal oxidation, so it is an important quality when controlling the film thickness after thermal oxidation.

Here, methods for forming a natural oxide film on the wafer surface include ozone water and hydrogen peroxide water in addition to SC1 cleaning. These do not cause an etching reaction, and only an oxidation reaction occurs. In particular, ozone water has a very strong oxidizing power, so it can form an oxide film with better controllability than with SC1 cleaning.

Accordingly, the present inventors examined whether the film thickness could not be controlled by limiting the variation factor of the natural oxide film to only the wafer surface roughness. That is, SC1 cleaning is performed on the target wafer forming the natural oxide film to form roughness in the range of 60 to 90/μm in the spatial frequency band that thickens the film thickness of the natural oxide film. After that, for example, hydrofluoric acid cleaning is performed to completely peel off the natural oxide film formed by SC1 cleaning, and then the natural oxide film is formed using a cleaning solution with oxidizing power, such as ozone water, which does not involve etching. Accordingly, it was thought that the variable factor could be limited to the wafer surface roughness and the film thickness of the natural oxide film could be controlled.

[Cleaning method of silicon wafer]

In consideration of the above-described content, the cleaning method for a silicon wafer according to the present invention will be described in detail below. 1 is a flow chart showing an example of a silicon wafer cleaning method according to the present invention. First, using a test silicon wafer, the correlation between the surface roughness of the silicon wafer after SC1 cleaning and the film thickness of the native oxide film is obtained. Then, using this correlation, the surface roughness of the silicon wafer, which is the target of natural oxide film formation, is determined, and the silicon wafer is cleaned SC1 according to the SC1 cleaning conditions according to the determined surface roughness of the silicon wafer to form a predetermined surface roughness, SC1 After removing the oxide film, a natural line film is formed. Hereinafter, the cleaning method for a silicon wafer according to the present invention will be described in detail.

(Silicon wafer for testing)

First, prepare multiple test silicon wafers for correlation acquisition (S1 in FIG. 1). There are no restrictions on the conductivity type or diameter of the silicon wafer. Regarding the surface roughness, it is preferable that the surface roughness of the test silicon wafer and the silicon wafer on which the natural oxide film is formed, which will be described later, is 0.5 nm or less in terms of Sa value. This range is because the film thickness of the oxide film calculated by spectroscopic ellipsometry is around 1 nm, and is more suitable for evaluating the film thickness of the natural oxide film of recently used silicon wafers. On the other hand, generally, the Sa value of the surface of a silicon wafer after CMP is 0.1 nm or less, and the Sa value of the back side (DSP surface) is about 0.2 to 0.4 nm, so at least after DSP (double-sided polishing) processing, the Sa value after CMP processing is If it is a wafer, it can be suitably used in the silicon wafer cleaning method according to the present invention.

(SC1 cleaning process of test silicon wafer)

Next, SC1 cleaning is performed on a plurality of test silicon wafers prepared under different cleaning conditions (S2 in FIG. 1). As cleaning conditions, it is desirable to change one or more of the chemical concentration, cleaning temperature, and cleaning time of SC1. This is because these are conditions that are easy to change even in realistic operations, and cleaning conditions can be easily set. For example, if the chemical concentration is NH ₄ OH:H ₂ O ₂ :H ₂ O=1:1:5 to 1:1:100, it may be adjusted. For example, if it is a washing temperature, it may be adjusted within the range of 30 to 90°C. For example, the cleaning time may be adjusted within the range of 0.5 to 10 minutes. One or more of the above conditions can be set, but it is preferable to set more washing conditions, and more than the three conditions above can be set. By performing SC1 cleaning in this way, roughness is formed on the surface of the wafer, and at the same time, an oxide film is formed.

Meanwhile, in general, the oxide film formed by cleaning may be collectively referred to as a natural oxide film or a chemical oxide film. In this specification, in order to distinguish between types of oxide films, the oxide film formed by SC1 cleaning is referred to as “SC1 oxide film.” The oxide film formed by cleaning using a cleaning liquid with oxidizing power other than SC1 cleaning is called a “natural oxidation film.”

(SC1 oxide film removal process of test silicon wafer)

Next, the test silicon wafer after SC1 cleaning is cleaned with hydrofluoric acid to completely remove the SC1 oxide film (S3 in Fig. 1). As long as the SC1 oxide film can be completely removed, there are no restrictions on the hydrofluoric acid cleaning conditions. For example, an example of the conditions is the hydrofluoric acid concentration of 0.3 to 5.0 wt%, temperature of 10 to 30 ° C, and cleaning time of 60 to 360 seconds. am.

(Natural oxide film formation process on silicon wafers for testing)

Subsequently, the test silicon wafer from which the SC1 oxide film was removed is cleaned using a cleaning solution having oxidizing power to form a natural oxide film (S4 in FIG. 1). As a cleaning liquid with oxidizing power, ozone water or hydrogen peroxide water is preferable, and ozone water with stronger oxidizing power is more preferable. Ozone water or hydrogen peroxide water has a strong oxidizing power and can uniformly oxidize the wafer surface, allowing an oxide film to be easily and stably formed. For example, the concentration of ozone water used can be in the range of 3 to 25 ppm, the temperature can be in the range of 10 to 30 degrees Celsius, and the cleaning time can be in the range of 60 to 360 seconds. Also, for example, the concentration of hydrogen peroxide used can be 0.2 to 5.0 wt%, the temperature can be 30 to 90°C, and the cleaning time can be 60 to 360 seconds. On the other hand, when using a batch type washer, the effort is reduced by performing these series of cleanings in one batch.

(Correlation acquisition process)

Subsequently, the correlation between the surface roughness of the test silicon wafer on which the oxide film was formed and the film thickness of the native oxide film was obtained (S5 in Fig. 1). The film thickness of the natural oxide film can be measured using a known measurement method, for example, by spectroscopic ellipsometry. As an indicator of surface roughness, it is easiest to obtain the Haze value with a particle counter. When using the Haze value, roughness can be reflected with high precision by using the difference (Haze increase, ΔHaze) before and after cleaning. Alternatively, in AFM, for example, a roughness index such as Sa may be used, or a PSD curve may be obtained from spectral analysis of AFM profile data, and the average value of the power spectrum density (intensity) in the spatial frequency band of 60 to 90/μm may be used. . On the other hand, roughness evaluation may be performed after SC1 cleaning but before the correlation acquisition process, may be performed before hydrofluoric acid cleaning, or may be performed after SC1 cleaning, hydrofluoric acid cleaning, or cleaning with a cleaning liquid having oxidizing power. This is because the surface roughness component that affects the film thickness of the oxide film hardly changes in hydrofluoric acid cleaning and cleaning with oxidizing power.

In Figure 7, SC1 cleaning is performed at a chemical concentration of NH ₄ OH:H ₂ O ₂ :H ₂ O=1:1:10, cleaning temperature at 40, 50, 55, 60°C, and cleaning time at 3 and 6 min. The film thickness of the natural oxide film under each cleaning condition was shown when hydrofluoric acid cleaning and ozone water cleaning were performed. For both the 3 and 6 min cleaning times, there was a tendency for the film thickness to become thicker as the cleaning temperature increased. This is because the higher the cleaning temperature, the more the roughness component is formed in the spatial frequency range of 60 to 90/μm described above. Since the Haze value of the particle counter reflects the roughness in the range of 60 to 90/μm, the Haze value can be used as an indicator.

In Figure 8, the relationship between the amount of Haze increase before and after cleaning and the film thickness is shown for a cleaning time of 3 minutes. In this way, it can be seen that there is a good correlation between the amount of Haze increase and the film thickness. In this way, the correlation between surface roughness and the film thickness of the natural oxide film can be obtained.

Next, using this correlation, the silicon wafer (S6 in FIG. 1) on which the natural oxide film is to be formed is cleaned to form a natural oxide film of the desired thickness.

(SC1 cleaning conditions determination process)

In the SC1 cleaning condition determination process (S7 in FIG. 1), the film thickness of the natural oxide film to be formed is first set. Next, the required surface roughness is predicted and determined from the correlation, and the SC1 cleaning conditions that become the determined surface roughness are determined. In a specific example, the target film thickness value is set to 1.310 nm. From the correlation in FIG. 8, the Haze increase corresponding to a film thickness of 1.310 nm is 0.020 ppm. Also, from the correlation in Figure 7, to form a roughness of 0.020 ppm, the liquid composition is NH ₄ OH:H ₂ O ₂ :H ₂ O=1:1:10, the cleaning temperature is 55°C, and the cleaning time is 3 min. You can see that it happens. This cleaning condition was determined to be the SC1 cleaning condition.

(SC1 cleaning process of silicon wafers subject to natural oxide film formation)

Next, SC1 cleaning is performed on the silicon wafer on which the natural oxide film is to be formed under the determined cleaning conditions (S8 in FIG. 1).

(SC1 oxide film removal process on silicon wafers subject to natural oxide film formation)

Next, the silicon wafer on which the natural oxide film is to be formed is cleaned with hydrofluoric acid to completely remove the SC1 oxide film formed by the SC1 cleaning (S9 in FIG. 1).

(Oxide film formation process on silicon wafers, which are subject to natural oxide film formation)

Next, the silicon wafer subject to natural oxide film formation from which the SC1 oxide film has been completely removed is cleaned using a cleaning solution having the same oxidizing power as when forming the natural oxide film on the test silicon wafer (oxidation film formation process of the test silicon wafer) to form the natural oxide film. Formed (S10 in Figure 1). In a specific example, a natural oxide film was formed by ozone water cleaning. As a result of evaluating the natural oxide film of the silicon wafer after ozone water cleaning by spectroscopic ellipsometry, it was found that it was about 1.312 nm, and that a natural oxide film equivalent to the set 1.310 nm could be formed.

[Manufacturing method of silicon wafer with natural oxide film]

By using the silicon wafer cleaning method according to the present invention, it is possible to manufacture a silicon wafer with a natural oxide film while controlling the film thickness with high precision.

Hereinafter, the present invention will be described in detail through examples, which do not limit the present invention.

(Example)

As silicon wafers for testing, a plurality of silicon wafers subjected to CMP polishing and single wafer cleaning were prepared. First, the Haze value before SC1 cleaning was acquired using Particle Counter SP3 manufactured by KLA. Next, SC1 cleaning, hydrofluoric acid cleaning, and ozone water cleaning were performed using a batch cleaning machine under the cleaning conditions described later. The SC1 cleaning conditions were liquid composition NH ₄ OH:H ₂ O ₂ :H ₂ O=1:1:10, cleaning time 3 min, and cleaning temperatures were 40, 50, 55, and 60°C. The hydrofluoric acid cleaning was performed at a concentration of 0.5 wt%, a cleaning temperature of 25°C, and a cleaning time of 3 min. The ozone water cleaning was performed at a concentration of 20 ppm, a cleaning temperature of 25°C, and a cleaning time of 3 minutes. Next, the Haze of the test silicon wafer after cleaning was evaluated by SP3. Afterwards, the film thickness of the native oxide film was evaluated using a spectroscopic ellipsometer M-2000V manufactured by JAWoollam. The results are shown in Table 1. The higher the cleaning temperature, the greater the increase in Haze, and the film thickness tends to become thicker, and a correlation between roughness and film thickness was obtained.

Next, the target film thickness of the natural oxide film formed on the silicon wafer was set to 1.330 nm. Using the obtained correlation, the Haze increase when targeting a film thickness of 1.330 nm is approximately 0.028 ppm, and the cleaning conditions of 60°C cleaning temperature/3 min cleaning time are suitable to form this surface roughness, so the SC1 cleaning conditions are used. The liquid composition was determined to be NH ₄ OH:H ₂ O ₂ :H ₂ O=1:1:10, cleaning temperature 60°C/cleaning time 3 min.

Next, the target silicon wafer was subjected to SC1 cleaning at a determined cleaning temperature of 60°C/cleaning time of 3 minutes, and then hydrofluoric acid cleaning and ozone water cleaning were performed under the same conditions as when the correlation described above was obtained. Afterwards, the natural oxide film after cleaning was evaluated with M-2000V, and the thickness was found to be 1.332nm, which was confirmed to be equivalent to the target setting value of 1.330nm. In this way, according to an embodiment of the present invention, the film thickness of the natural oxide film could be controlled by using the relationship between surface roughness and the film thickness of the natural oxide film.

Meanwhile, the present invention is not limited to the above embodiments. The above-described embodiments are examples, and anything that has substantially the same structure as the technical idea described in the claims of the present invention and exhibits the same effects is included in the technical scope of the present invention.

Claims (7)

As a cleaning method for a silicon wafer,
An SC1 cleaning process of the test silicon wafer of preparing a plurality of test silicon wafers in advance and performing SC1 cleaning of the test silicon wafers by changing the SC1 cleaning conditions to produce a plurality of levels of test silicon wafers with different surface roughness;
A SC1 oxide film removal process of the test silicon wafer, in which the SC1 oxide film formed in the SC1 cleaning process of the test silicon wafer is completely removed by hydrofluoric acid cleaning;
A natural oxide film forming process on the test silicon wafer, using a cleaning solution having oxidizing power, to clean the test silicon wafer from which the SC1 oxide film has been removed to form a natural oxide film;
A correlation acquisition process for acquiring a correlation between the surface roughness of the test silicon wafer formed by the SC1 cleaning and the film thickness of the natural oxide film formed in the natural oxide film formation process of the test silicon wafer;
With respect to the silicon wafer, which is the object of natural oxide film formation, the film thickness of the natural oxide film formed by cleaning with the cleaning liquid having the oxidizing power is a predetermined thickness, based on the correlation acquired in the correlation acquisition process. A SC1 cleaning condition determination step of determining the surface roughness formed on the silicon wafer and determining the SC1 cleaning conditions that become the determined surface roughness;
An SC1 cleaning process of the silicon wafer subject to natural oxide film formation, wherein SC1 cleaning of the silicon wafer subject to natural oxide film formation is performed under the SC1 cleaning conditions determined in the SC1 cleaning condition determination process;
A SC1 oxide film removal process of the silicon wafer subject to the natural oxide film formation, wherein the silicon wafer subject to the natural oxide film formation after the SC1 cleaning process is cleaned with hydrofluoric acid to completely remove the SC1 oxide film formed by the SC1 cleaning;
By providing an oxide film forming process on the silicon wafer, which is the natural oxide film formation object, by cleaning the silicon wafer, the natural oxide film object, from which the SC1 oxide film has been removed, using a cleaning liquid having the oxidizing power, to form a natural oxide film, the silicon wafer is formed by cleaning. A cleaning method for a silicon wafer, characterized in that the film thickness of the natural oxide film of the silicon wafer is controlled. According to paragraph 1,
The SC1 cleaning conditions are one or more of SC1 chemical concentration, cleaning temperature, and cleaning time. According to claim 1 or 2,
A method of cleaning a silicon wafer, wherein the surface roughness is a roughness component with a spatial frequency of 60 to 90/μm. According to any one of claims 1 to 3,
A cleaning method for a silicon wafer, characterized in that the indicator of the surface roughness is the Haze value of the particle counter. According to any one of claims 1 to 3,
A method for cleaning a silicon wafer, wherein the surface roughness index is the average value of the power spectrum density with a spatial frequency of 60 to 90/μm. According to any one of claims 1 to 5,
A cleaning method for a silicon wafer, characterized in that ozone water or hydrogen peroxide water is used as a cleaning liquid having the oxidizing power. A method for manufacturing a silicon wafer with a natural oxide film, characterized in that a silicon wafer with a natural oxide film is manufactured by the cleaning method for a silicon wafer according to any one of claims 1 to 6.