JP2005072568A - Method and device for cleaning microstructure, semiconductor device and manufacturing method therefor, and mictostructure and manufacturing method therefor - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a method for cleaning fine particles adhered to a microstructure, using a supercritical fluid as the cleaning solvent. <P>SOLUTION: This method for cleaning a microstructure, using supercritical fluid as a cleaning solvent, contains the microstructure within a cleaning chamber 24 with the distance between the inner wall of the cleaning chamber and the surface of the microstructure on which the fine particles to be removed are adhered being kept equal to 3 mm or smaller, and rotates the microstructure at revolutions 400 per minute of or higher, while introducing the supercritical fluid into the cleaning chamber. The cleaning is performed, with the flow rate introduced into the cleaning chamber which is kept at 1 L/min or larger. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

The present invention relates to a method for cleaning (/ drying) a microstructure using a supercritical fluid as a cleaning medium, and a cleaning apparatus for performing the method, and more particularly, using a supercritical fluid as a cleaning medium. The present invention relates to a cleaning method and a cleaning apparatus that can efficiently clean fine particles adhering to a microstructure.
Furthermore, the present invention relates to a semiconductor device obtained using the cleaning method and a manufacturing method thereof, and a microstructure obtained using the cleaning method and a manufacturing method thereof.

In this specification, the microstructure means a structure having a microstructure, and the microstructure is separated from the supporting solid substrate as a separation portion, and approaches or is separated from the substrate as necessary or supported. Although it does not include a separation portion that is separated from a solid substrate, it is a structure having a microstructure having a large ratio of pattern height and width called aspect ratio.
A typical example of a microstructure having a separation part is a microstructure called a micromachine having a minute drive component as a separation part, and a representative example of a microstructure having no separation part is, for example, an LSI micropattern. It is a mask having an opening pattern for forming itself or its fine pattern.

  As shown in FIG. 23A, a microstructure 50 called a micromachine having a minute movable part is configured as a laminated film of a supporting solid substrate 52, a structure film 54, and an optical film 56, and a gap 58 Through the support solid substrate 52 at both ends, and has a movable separating portion, that is, a beam structure 60. In the manufacturing process of the micromachine, a sacrificial layer etching method is applied when the beam structure 60 is formed, that is, when the hollow structure (gap portion 58) on the supporting solid substrate 52 is formed. FIG. 23A is a cross-sectional view of a hollow structure optical micromachine having a doubly supported beam structure.

In the sacrificial layer etching method, first, a sacrificial layer (not shown) having the same dimensions as the gap 58 is formed on the supporting solid substrate 52, and then the sacrificial layer is patterned into a predetermined shape, and then the structure film 54 and the optical layer A film 56 is laminated on the entire surface of the supporting solid substrate 52. For example, a metal film is used as the optical film 56.
After the laminated film of the structural film 54 and the optical film 56 is patterned into a desired shape of the beam structure 60, the sacrificial layer is selectively etched away, and the beam supported on the supporting solid substrate 52 via the gap 58. A structure 60 is formed. The beam structure 60 composed of the laminated film of the structural film 54 and the optical film 56 formed on the sacrificial layer is held from the supporting solid substrate 52 via the gap 58, and is supported with respect to the supporting solid substrate 52. Function as a movable separation part.

  When the laminated film of the structure film 54 and the optical film 56 is dry-etched, the etching gas, the resist, the structure film 54, the optical film 56, and the like react with each other to form a reaction product and adhere to the bottom surface and the side wall. It becomes a particulate residue. Next, when the sacrificial layer is etched, the residue remaining in the etching process of the laminated film remains as it is without being removed, or is further changed into a residue that is difficult to be removed. Further, during the sacrificial layer etching, a reaction product of the etching gas and the sacrificial layer material is generated and new fine particles are generated, which is important for the function of the micromachine as shown in FIG. It also adheres on the optical film 56. For this reason, it is necessary to remove these particulate contamination.

  By the way, when these fine structures are cleaned to remove contaminating fine particles, if they are washed and dried with an aqueous or organic liquid as used in the normal semiconductor manufacturing process, they are supported through the voids. There was a problem that the laminated film of the optical material layer and the structural material layer held on the solid substrate was damaged or fixed to the supporting solid substrate.

This phenomenon occurs when the cleaning liquid in the drying process or the liquid used in the drying process eventually evaporates, leaving the liquid remaining in the minute gap between the separation part consisting of the laminated film and the supporting solid substrate. When the volume of the voids is reduced by evaporation, suction force is generated between the supporting solid substrate and the separation part due to the surface tension of the cleaning liquid or drying liquid, and the separation part has insufficient rigidity. This is a phenomenon in which a part is damaged or fixed to a supporting solid substrate.
In addition, destruction may occur during cleaning and rinsing. This is a phenomenon in which a microstructure including a separation portion formed on a supporting solid substrate is extremely small and mechanically fragile, so that it is destroyed by water pressure due to stirring in a solution.

  Because of such a problem, a cleaning liquid used in a normal semiconductor process cannot be used for cleaning a microstructure such as a micromachine. For this reason, there was no choice but to send to the next step without washing. As a result, the yield, reliability, and device characteristics deteriorate due to the contamination of fine particles.

In recent years, with the increase in scale of MOS LSIs, pattern miniaturization of LSI wiring structures and the like has progressed. Now, a pattern of less than 100 nm is formed.
If it is less than 100 nm, it is no longer than the wavelength of a laser used as a light source in photolithography, so we tried to form a pattern by photolithography using various methods such as exposure methods and masks, such as using a halftone phase shift mask. Is approaching the limit.

Therefore, in devices of 70 nm and after, lithography by electron beam exposure is being studied for practical use.
Unlike a light exposure mask, an electron beam exposure mask 70 used in electron beam exposure lithography is an opening made of a pattern having a large aspect ratio (length / width) through which an electron beam passes, as shown in FIG. This is a stencil membrane structure in which a membrane 74 having 72 is supported by a support frame 76.
When a pattern with a high aspect ratio is formed by dry etching, fine foreign matter remains on the side wall of the opening 72 of the mask 70 and on the front and back surfaces of the membrane 74 as shown in FIG. . Further, when using the mask, that is, in the process of mask transport, mask mounting, exposure work, etc. in the exposure apparatus, fine particles may adhere to the mask. FIG. 23B is a cross-sectional view of the electron beam exposure mask 70.

Further, in order to increase the speed of LSI, it is indispensable to reduce the capacitance between wires, and in order to reduce the capacitance between wires, a low dielectric constant film is used as an interlayer insulating film between wires. In order to produce the ultimate low dielectric constant film, it is required not only to select a film material but also to make the structure of the low dielectric constant film porous.
In the damascene process for forming the buried wiring structure, as shown in FIG. 23C, an etching stop layer 80 and a low dielectric constant film 82 are formed on the base film 78, and then the low dielectric constant film 82 is patterned. After the wiring groove 84 is formed, a wiring material (not shown) is formed by embedding a wiring material in the wiring groove 84. After the low dielectric constant film 82 is etched, as shown in FIG. Foreign matter due to the reaction product of the etching gas and the low dielectric constant film 82 remains. These residues cause wiring defects and need to be removed and cleaned. FIG. 23C is a cross-sectional view of the wiring groove.

  In addition, it is indispensable to reduce the resistance of the source / drain in order to reduce the power consumption of the LSI. In order to reduce the resistance, a method of implanting ions at a high concentration of 1 × 10 15 ions / cm 2 or more into the source / drain portion is used. When performing the ion implantation, as shown in FIG. 23D, the gate electrode 88 and the side wall 89 other than the source / drain 87 formed on the semiconductor substrate 86 are masked with a photoresist 90. . The photoresist 90 must be stripped after ion implantation. 90a is a normal photoresist portion, and 90b is a photoresist portion cured by ion implantation.

As described above, in the LSI device and mask manufacturing process, these fine particles must be removed. However, there are the following problems.
Usually, in the manufacturing process of a semiconductor device, when removing foreign matters, wet cleaning using water as a solvent is performed. However, when cleaning the mask 70 and the wiring groove 84 described above, when a liquid solvent such as water is used as the cleaning liquid, the cleaning liquid does not penetrate into the pattern grooves having a high aspect ratio because the surface tension of the cleaning liquid is large. Even if the cleaning can be performed, it is difficult to discharge the cleaning liquid out of the pattern groove and dry it when rinsing the cleaning liquid.

Another major problem in the drying process when forming such a fine pattern is pattern collapse and sticking of adjacent patterns. This occurs when the rinsing liquid is dried, and becomes more noticeable in a high aspect ratio pattern.
This phenomenon is due to a bending force (surface tension or capillary force) that works due to a pressure difference between the rinse liquid remaining between the patterns when the substrate is dried and external air. This capillary force is a large force that distorts the silicon pattern depending on the surface tension of the rinse liquid generated at the gas-liquid interface between the patterns, so the problem of the surface tension of the rinse liquid becomes important. Yes.

  In the case of a porous low dielectric constant film, when wet cleaning is performed, when a solvent such as water enters and exits the pores of the low dielectric constant film, a pressure difference due to the generation of a gas-liquid interface as described above occurs. In many cases, the holes in the film are crushed and the dielectric constant increases.

  In order to solve the problems described above, it is considered that cleaning and drying may be performed using a fluid having a small surface tension. For example, the surface tension of water is about 72 dyn / cm, whereas the surface tension of methanol is about 23 dyn / cm. Therefore, drying after replacing water with methanol is more preferable than drying from water. Sticking and breaking of microstructures can be suppressed. However, since the surface tension is still very high, it is not possible to completely prevent the destruction of the microstructure, and the problem cannot be solved completely.

In order to solve the problem of microstructural destruction due to surface tension, it is necessary to use a fluid with zero surface tension as the rinse liquid, or to replace the rinse liquid with a fluid with zero surface tension and dry it. is there.
A fluid with zero surface tension is a supercritical fluid that is called a critical temperature and critical pressure, and is a fluid in a state phase where each substance exists at a temperature and pressure that are equal to or higher than the values inherent to each substance. is there.
In the supercritical state, a substance has a significantly lower viscosity and density than the liquid state of the substance, even though the dissolving power in other liquids and solids is almost the same as the liquid state of the substance, and a diffusion coefficient. Has a unique property that is extremely large. That is, it can be said that the liquid has a gaseous state. As a result, since the gas-liquid interface is not formed, the surface tension becomes zero. Therefore, if the drying is performed in a fluid in a supercritical state, the effect of the surface tension is eliminated, so that pattern collapse or the like does not occur at all.

  A supercritical fluid has the ability to dissolve a liquid such as a cleaning liquid, and rapidly gasifies by reducing the surrounding pressure to a critical pressure or less. Drying after washing with the supercritical fluid can be easily performed by first releasing the supercritical fluid in the supercritical state and then reducing the pressure to vaporize the supercritical fluid.

  In the cleaning process of the microstructure with the supercritical fluid, the microstructure is separated from the support substrate surface by using the sacrificial layer as a whole or part of the sacrificial layer, or the microfabrication pattern with a large aspect ratio obtained by etching. Directly after the etching solution, after passing through the cleaning solution, or after being replaced with another liquid, the liquid is attached to the surface and brought into contact with the supercritical fluid contained in the pressure vessel, These liquids are removed by dissolving them in a supercritical fluid.

  Next, the pressure inside the container is reduced below the critical pressure while keeping the container temperature at or above the critical temperature, the supercritical fluid is gasified and removed, and the microstructure is taken out into the atmosphere. Since the surface tension of the supercritical fluid is extremely small, when it is removed from the surface of the microstructure, the stress applied to the microstructure by the surface tension is negligibly small. Therefore, the microstructure is not deformed or broken.

For example, JP 2000-91180 A introduces a supercritical fluid while removing moisture in a reaction chamber, or introduces a supercritical fluid after removing moisture in a reaction chamber, Proposed means for drying (see page 4).
Japanese Patent Application Laid-Open No. 9-139374 discloses that the liquid adhering to the microstructure is removed by dissolving it in the supercritical fluid in the pressure vessel, and then the supercritical fluid is gasified by reducing the vessel pressure below the critical pressure. A method and apparatus for removing a dried microstructure into the atmosphere has been proposed (see page 5).
JP 2000-91180 A JP-A-9-139374

  As the supercritical fluid, it is possible to use many substances that have been confirmed to be in a supercritical state such as carbon dioxide, ammonia, water, alcohols, low molecular weight aliphatic saturated hydrocarbons, benzene, diethyl ether and the like. it can. Among these, carbon dioxide having a supercritical temperature of 31.3 ° C., which is close to room temperature, is one of the preferable substances because it is easy to handle and can be cleaned without exposing the object to be cleaned to high temperature. One.

  By the way, as a method for efficiently removing fine particles from the solid surface, a fine etching process is performed on the surface of the part where the fine particles adhere to separate the fine particles from the surface, and then the separated fine particles are flowed into a liquid such as a cleaning liquid. There is a cleaning method in which it is transported and removed along with it. Even when supercritical carbon dioxide is used as the cleaning liquid, the adhesion force of the fine particles to the surface can be reduced by performing the treatment by adding an etching agent.

However, when the supercritical fluid is used as the cleaning liquid, the supercritical fluid, for example, supercritical carbon dioxide, has a gas-like property that its viscosity and density are extremely small, so that it is separated by etching. When the fine particles are transferred by the flow of the supercritical fluid, there is a problem that the transfer force of the fine particles by the flow is remarkably small and it is difficult to remove the fine particles effectively.
Accordingly, the problem of the present invention is that when the supercritical fluid is used as a cleaning liquid, the viscosity and density of the supercritical fluid are remarkably smaller than those of general liquids, and therefore, the fine particles cannot be efficiently removed from the microstructure. That's what it means. Accordingly, an object of the present invention is to provide a cleaning method and a cleaning apparatus capable of efficiently cleaning fine particles adhering to a microstructure using a supercritical fluid as a cleaning medium.
A further object of the present invention is to provide a highly reliable semiconductor device obtained by using such a cleaning method, a manufacturing method thereof, a microstructure, and a manufacturing method thereof.

  In order to solve the above problems, the present inventor increases the transport force of fine particles by the flow of supercritical carbon dioxide by increasing the fluid resistance and shear stress of supercritical carbon dioxide when cleaning the microstructure. With this in mind, the flow of supercritical fluid, for example, the flow of supercritical carbon dioxide, was analyzed as described below.

式(１)で、Ｃ0は係数、ρは超臨界二酸化炭素の流体密度、及びｕは超臨界二酸化炭素の流速である。

式(２)で、μは超臨界二酸化炭素の粘度、ｕは超臨界二酸化炭素の流速、及びｙは表面からの高さである。 The fluid resistance R and shear stress τ of supercritical carbon dioxide can be easily expressed in the following two equations (1) and (2).

In Equation (1), C0 is a coefficient, ρ is the fluid density of supercritical carbon dioxide, and u is the flow rate of supercritical carbon dioxide.

In equation (2), μ is the viscosity of supercritical carbon dioxide, u is the flow rate of supercritical carbon dioxide, and y is the height from the surface.

As can be seen from the equation (1), the fluid resistance is proportional to the density of the fluid. Therefore, in order to increase the fluid resistance, it is only necessary to lower the temperature and increase the pressure. When the temperature is low and the pressure is high, the density of supercritical carbon dioxide becomes close to the density of the liquid and increases, so that the fluid resistance of supercritical carbon dioxide increases. Specifically, at a temperature of 35 ° C. and a pressure of 25 MPa, the fluid resistance of supercritical carbon dioxide is about 70% of the fluid resistance of the liquid.
However, this still lacks particulate transport. This is because the shear stress is small.

As can be seen from Equation (2), the shear stress of the supercritical fluid is proportional to the viscosity of the fluid. The viscosity of supercritical carbon dioxide increases with decreasing temperature and increasing pressure, so lower the temperature, increase the pressure, increase the viscosity, and thereby increase the shear stress of the supercritical fluid. I can. On the other hand, the lower limit of the temperature is 31 ° C. which can maintain the supercritical state of supercritical carbon dioxide, and the upper limit of the pressure is practically about 35 MPa in consideration of the specifications of the apparatus and piping.
However, under these temperature and pressure conditions, the viscosity does not increase dramatically, the shear stress acting on the adhered microparticles with a diameter of 0.1 micron is 0.1 N / m 2 or less, and the shear stress of the liquid is 1 N / m 2. Is less than 1/10, preferably 1/20 of 2 N / m2.

ナビエストークスの式は、次の式（６）及び（７）で表される。シリコンウエハの中心から流体を供給した場合、式（７）で、第１項は円周方向の項であり、第２項は半径方向の項である。

That is, it is difficult to increase the shear stress by increasing the viscosity of the supercritical fluid to a desired viscosity. Therefore, we considered increasing the shear stress based on equation (2) by increasing the surface flow velocity by rotating the substrate in supercritical carbon dioxide.
First, the Couette-Poise flow between parallel flat plates shown in FIG. 5 was expressed by the Naviestokes equation, and the surface flow velocity in three dimensions of the fluid when the substrate was cleaned was analyzed. Here, assuming the equations (3) to (5),

Navi Estoke's equation is expressed by the following equations (6) and (7). When fluid is supplied from the center of the silicon wafer, the first term is a circumferential term and the second term is a radial term in Equation (7).

Here, when the wall surface velocity u is u = 0 and the volume flow rate q is q = u0h = constant, the following equations (8) to (10) are established.

Therefore, based on the equations (8) to (10), the surface flow velocity and shear stress of the three-dimensional supercritical carbon dioxide were simulated in consideration of the flow caused by the rotation of the substrate in the cleaning chamber. As shown in FIG. 6, the cleaning apparatus targeted in the simulation accommodates a substrate W on which fine particles are adhered, rotates the substrate W while introducing supercritical carbon dioxide as a cleaning medium, and cleans the substrate W. 86 is a cleaning device that cleans the substrate W and removes fine particles adhering to the substrate W. FIG. 6 is a schematic cross-sectional view showing the configuration of a conventional cleaning chamber, in which 88 is a supercritical carbon dioxide supply pipe, 90 is a pump for injecting supercritical carbon dioxide, and 92 is a rotating shaft for rotating the substrate. It is.
FIG. 7 shows, as an example of simulation, the distribution of shear stress on the surface of a wafer having a diameter of 200 mm when the temperature, pressure, flow rate, distance between the substrate surface and the inner wall of the chamber are set to optimum values, and the rotation speed is changed.

When the rotation speed per minute is 400 rotations or more, the shear stress is 1 N / m 2 or more in almost all areas of the entire wafer surface as long as it is 3 mm or more away from the center, and the rotation speed per minute is 500. Above the rotation, if the region is 3 mm or more away from the center, the shear stress is 2 N / m 2 or more in almost all regions of the entire wafer surface, which is the same level as the immersion cleaning with liquid. As described above, the shear stress can be increased by increasing the number of rotations of the substrate as a parameter other than temperature and pressure. If the shear stress is less than 1 N / m 2, there is no cleaning effect. From the viewpoint of cleaning effect, the shear stress may be 1 N / m 2 or more, preferably 2 N / m 2 or more, and there is no upper limit.
When the entire flow rate is the same, as shown in the example of the distribution in the height direction of the surface flow velocity in FIG. The flow rate increases and the shear stress increases. FIG. 8A shows the distribution in the height direction of the surface flow velocity when the distance between the substrate surface and the chamber inner wall is long, and FIG. 8B shows the height of the surface flow velocity when the distance between the substrate surface and the chamber inner wall is short. The direction distribution is shown.

Increasing the flow rate of the fluid itself is also effective. By increasing the number of rotations of the substrate and decreasing the distance between the substrate surface and the inner wall of the device facing the substrate, the shear stress of the supercritical fluid is reduced. It was found that the level could be equivalent to the stress.
That is, according to the above analysis, the rotation speed of the substrate is 400 rpm or more, preferably 500 rpm or more, the distance between the substrate surface to which the fine particles adhere and the inner wall of the chamber facing is 5 mm or less, preferably 3 mm or less, and the fluid flow rate is converted into liquid It was found that by setting the pressure at 1 L / min or more, the shear stress of the fluid can be increased, the transport force of the fine particles can be increased, and the fine particles can be effectively removed. When the rotational speed is less than 400 rpm, there is no cleaning effect. The upper limit of the rotational speed can be set to about 3000 rpm in consideration of the limit of the high pressure device. When the distance between the substrate surface and the inner wall of the chamber exceeds 5 mm, there is no cleaning effect. The lower limit of the distance can be set to about 0.3 mm in consideration of the production limit of the cleaning device.

  The simulation results also show that when the number of rotations is further increased, the surface flow velocity of the fluid becomes very fast, the pressure at the supply port decreases, and the pressure needs to be replenished to balance the pressure. It was. That is, the surface flow velocity can be remarkably improved by high-speed rotation and fluid supply with a large flow rate.

However, the flow rate flowing from the supply port is determined by the capacity of the pump of the apparatus and has a limit. In reality, only a flow rate of about 3 L / min can be flowed in terms of liquid. Therefore, in order to increase the shear stress, the flow rate is replenished from another system.
Therefore, the fluid that has already passed through the substrate surface is circulated again without attenuating the speed, joined to the newly supplied fluid, and the fluid is supplied to the substrate, thereby increasing the flow rate. A large surface flow rate can be maintained. As a result, the shear stress becomes larger and the removal rate of the fine particles can be further increased.

  In order to achieve the above object, based on the above knowledge, the cleaning method according to the present invention adheres to the surface of the microstructure in the cleaning method of the microstructure using the supercritical fluid as the cleaning medium. The relative speed of the supercritical fluid with respect to the particles to be removed is maintained so that the shear stress of the supercritical fluid in the vicinity of the particles to be removed is 1 N / m 2 or more, preferably 2 N / m 2 or more.

Moreover, the cleaning method according to the present invention is a method for cleaning a microstructure using a supercritical fluid as a cleaning medium.
The distance between the inner wall of the cleaning chamber and the adhesion surface of the fine particles to be removed is maintained at 5 mm or less, preferably 3 mm or less, the microstructure is accommodated in the cleaning chamber, and a supercritical fluid is introduced into the cleaning chamber. In the supercritical fluid, the microstructure is rotated at 400 revolutions per minute or more, preferably at 500 revolutions or more.

  In the present invention, the supercritical fluid includes, in addition to the supercritical fluid itself, a supercritical fluid obtained by adding an additive such as a cleaning agent or an etching agent to the supercritical fluid. In other words, the etching agent, the cleaning agent, both the etching agent and the compatibilizer, both the cleaning agent and the compatibilizing agent, and the etching agent, the cleaning agent, and the compatibilizing agent are added to the supercritical fluid as additives. To do. Thereby, the cleaning effect by the supercritical fluid can be enhanced. The microstructure means a structure having a microstructure, and there is no restriction as to whether the microstructure is a movable part.

Preferably, cleaning is performed while maintaining the flow rate of the supercritical fluid introduced into the cleaning chamber at 1 L / min or more.
In a preferred embodiment of the method of the present invention, supercritical carbon dioxide is used as the supercritical fluid.

A cleaning apparatus according to the present invention includes a cleaning chamber that houses a microstructure, introduces a supercritical fluid as a cleaning medium, and cleans the microstructure, and cleans the microstructure and adheres to the microstructure. In a single wafer cleaning device that removes fine particles
The cleaning chamber is configured such that the distance between the inner wall of the cleaning chamber and the adhered surface of the fine particles to be removed is 5 mm or less, preferably 3 mm or less, and the microstructure is contained in the supercritical fluid in the containing chamber. It is characterized by having a rotation mechanism that rotates the body at a rotational speed of 400 revolutions per minute or more, preferably 500 revolutions or more.

  Preferably, the cleaning apparatus includes circulation means for circulating at least a part of the supercritical fluid discharged from the cleaning chamber to the cleaning chamber. Thereby, the flow rate of the supercritical fluid is increased, the shear stress is increased, and the particulate removal performance is improved.

  On the other hand, when the substrate is not rotated, it is advantageous to supply the fluid in parallel to the substrate because the flow rate does not decrease. In that case, the shear stress in the vicinity of the surface can be obtained simply by the above-described equation (2). That is, it is a function of the distance between the substrate surface and the inner wall of the cleaning tank and the linear velocity. The linear velocity is determined by the distance between the substrate surface and the inner wall of the cleaning tank and the flow rate. Therefore, when the entire flow rate is the same, the distance between the substrate surface and the inner wall of the cleaning tank facing the substrate surface is set to a minimum level that can be further processed as shown in the example of distribution in the height direction of the surface flow velocity in FIG. By doing so, the surface velocity at the position of the adhering fine particles can be increased, and the shear stress can be increased. For example, when the linear velocity is 3 m / sec, if the distance between the substrate surface and the inner wall of the cleaning tank facing the substrate is 1 mm or less, the shear stress is 1 N / m 2 or more, preferably 0.5 mm or less. The stress is 2 N / m 2 or more. When the distance between the substrate surface and the inner wall of the cleaning tank facing the substrate is larger than 0.5 mm, the linear velocity must be increased, and an expensive large flow pump is required to secure the necessary flow rate. It becomes unrealistic.

Based on the above-mentioned knowledge, the cleaning method according to the present invention is a method for cleaning a microstructure using a supercritical fluid as a cleaning medium, and includes a cleaning chamber inner wall and an adhesion surface of a fine particle to be removed. The distance is maintained at 1 mm or less, preferably 0.5 mm or less, the microstructure is accommodated in the cleaning chamber, and a supercritical fluid is introduced into the cleaning chamber at a linear velocity of 3 m / sec.
When the distance is larger than 1 mm, the shear stress used for cleaning cannot be obtained, and there is no cleaning effect. The lower limit of the distance can be set to about 0.3 mm in consideration of the cleaning device manufacturing limit.

  As described above, in the present invention, the supercritical fluid includes a supercritical fluid obtained by adding an additive such as a cleaning agent or an etching agent to the supercritical fluid in addition to the supercritical fluid itself. In other words, the etching agent, the cleaning agent, both the etching agent and the compatibilizer, both the cleaning agent and the compatibilizing agent, and the etching agent, the cleaning agent, and the compatibilizing agent are added to the supercritical fluid as additives. To do. Thereby, the cleaning effect by the supercritical fluid can be enhanced. The microstructure means a structure having a microstructure, and there is no restriction as to whether the microstructure is a movable part.

Preferably, cleaning is performed while maintaining the flow rate of the supercritical fluid introduced into the cleaning chamber at 1 L / min or more.
In a preferred embodiment of the method of the present invention, supercritical carbon dioxide is used as the supercritical fluid.

A cleaning apparatus according to the present invention includes a cleaning chamber that houses a microstructure, introduces a supercritical fluid as a cleaning medium, and cleans the microstructure, and cleans the microstructure and adheres to the microstructure. In a single wafer cleaning device that removes fine particles
The cleaning chamber is configured so that the linear velocity of the supercritical fluid is 3 m / sec and the distance between the inner wall of the cleaning chamber and the adhesion surface of the fine particles to be removed is 1 mm or less, preferably 0.5 mm or less. It is characterized by having.

Preferably, the distance from the cleaning chamber inner wall surface facing the substrate surface is 1 mm or less, preferably 0.5 mm or less at a liquid conversion of 1 L / min or more.
Further, the cleaning device is formed obliquely so that the distance from the surface on the microstructure installation side gradually increases as the inner wall surface of the cleaning chamber moves away from the microstructure. The cleaning chamber is configured to flow the supercritical fluid in one direction from the inlet toward the outlet.

  On the other hand, the above-described cleaning method of the present invention is suitable for application to semiconductor devices and microstructures.

The semiconductor device according to the present invention is characterized in that a required region of the semiconductor device is cleaned by the above-described cleaning method using the supercritical fluid of the present invention.
In addition, the method for manufacturing a semiconductor device according to the present invention includes a step of cleaning the processed substrate by the above-described cleaning method using the supercritical fluid of the present invention after selecting the required aspect ratio. It is said.

The microstructure according to the present invention is characterized in that a required region of the microstructure is cleaned by the above-described cleaning method using the supercritical fluid of the present invention.
In the method for manufacturing a microstructure according to the present invention, after the sacrificial layer is selectively etched, the microstructure substrate in which the hollow portion is formed is cleaned by the above-described cleaning method using the supercritical fluid of the present invention. It has the process.

As described above, in the cleaning method and apparatus using the supercritical fluid according to the present invention, the particle transfer ability of the supercritical fluid is effectively increased by the effect of increasing the shear stress accompanying the increase in the surface flow velocity. In addition, the particle removal performance is improved as compared with the conventional method and apparatus.
In particular, the cleaning effect is increased by maintaining the relative velocity of the supercritical fluid with respect to the fine particles to be removed so that the shear stress of the supercritical fluid is 1 N / m 2 or more. When the introduction flow rate of the supercritical fluid is constant, the distance between the inner wall of the cleaning chamber and the adhesion surface of the fine particles to be removed is maintained at 5 mm or less, and the microstructure is accommodated in the cleaning chamber and cleaned. By introducing the supercritical fluid into the chamber and rotating the microstructure in the supercritical fluid at a rotational speed of 400 revolutions or more per minute, the shear stress of the supercritical fluid becomes 1 N / m ² or more. The relative velocity of the supercritical fluid with respect to the removed fine particles is increased, and the cleaning effect is increased.

  Further, with respect to the separation of the fine particles from the surface by the etching process, excessive etching is not required, so that the amount of chemical substances such as an etchant can be minimized. Therefore, unlike the conventional cleaning method, the optical film made of a metal film is not eroded by an etching agent or the like, and optical characteristics such as reflectance do not change, and the dimensional change of the mask pattern does not occur. .

Since a supercritical fluid such as supercritical carbon dioxide is used as a cleaning medium, surface tension does not occur. Therefore, the microstructure of the beam structure is not destroyed by the influence of surface tension that occurs at the gas-liquid interface, the pores of the porous low dielectric constant film are not crushed, the removal of fine particles, and the microstructure The body can be efficiently dried.
In addition, by immediately reusing the discharged fluid, the target pollutant can be removed without requiring a large-capacity pump.

  Further, in the cleaning method and the cleaning apparatus of the present invention, the microstructure is maintained in the cleaning chamber by keeping the distance between the inner wall of the cleaning chamber and the non-adhering surface of the fine particles to be removed within 1 mm without rotating the substrate. The relative velocity of the supercritical fluid with respect to the particles to be removed is maintained so that the shear stress of the supercritical fluid is 1 N / m 2 or more by introducing the supercritical fluid into the cleaning chamber at a linear velocity of 3 m / sec. This increases the cleaning effect. In other words, the shear stress of the fluid can be increased without rotating the substrate, and the transport force of the particles to be removed can be increased to effectively remove the particles to be removed.

  By forming the inner wall surface of the cleaning chamber obliquely so that the distance from the surface on the microstructure body side gradually increases as the distance from the microstructure body increases, the fluid is not decelerated on the microstructure body in the cleaning chamber. Therefore, it can flow smoothly at a required flow rate.

Also, as described above, since excessive etching is not required, the amount of chemical substances added can be minimized, so that the film of metal material is eroded and optical characteristics such as reflectance do not change, Also, there is no dimensional change of the mask pattern.
Because supercritical carbon dioxide is used as a cleaning medium, the beam-shaped microstructure is not destroyed by the effect of surface tension generated at the gas-liquid interface, and the holes in the low dielectric film are crushed in the LSI manufacturing process. In addition, fine particles and residues can be removed, and the photoresist can be removed and dried.

  In this cleaning method and cleaning apparatus, for example, even when a small-diameter chip is cleaned, the surface flow velocity can be increased and the cleaning effect can be increased. Incidentally, in the case of a small-diameter chip, since the distance from the center to the edge of the substrate is not sufficient, the above-described cleaning method performed without rotating the substrate is advantageous when there is a possibility that the surface flow velocity increase due to rotation may not be sufficiently obtained. . In addition, since the cleaning chamber for supercritical fluid is exposed to high pressure, safety measures such as leakage prevention are easy because of the absence of a rotating mechanism, the cleaning chamber can be simplified, and the manufacturing cost of the cleaning apparatus can be reduced. Can also be reduced.

  As described above, it is possible to prevent breakage due to the weakness of the microstructure, sticking of the microstructure to the substrate, and sticking, which have occurred in the drying process with conventional cleaning methods, and high safety. A microstructure having a movable portion having a desired microstructure can be manufactured with a high yield by a simple process at a low cost. Therefore, by applying the cleaning method and the cleaning apparatus according to the present invention, it is possible to maintain the quality of electrical characteristics and the like of electronic devices and electronic parts when manufacturing electronic devices and electronic parts using the microstructure as parts. , Leading to improved yield.

According to the semiconductor device and the manufacturing method thereof according to the present invention, since the semiconductor device is manufactured by cleaning using the above-described cleaning method of the present invention, it is possible to provide a semiconductor device with high reliability claims and high yield. A semiconductor device can be manufactured.
According to the microstructure and the manufacturing method thereof according to the present invention, since it is manufactured by cleaning using the above-described cleaning method of the present invention, it is possible to provide a highly reliable microstructure. A microstructure can be manufactured with high yield.

  The best mode for carrying out the present invention will be described below with reference to the drawings.

Embodiment 1 of cleaning apparatus
First, with reference to FIG. 1 to FIG. 3, the configuration of an embodiment of a cleaning apparatus according to the present invention will be described. 1 is a block diagram showing the configuration of the cleaning device, FIG. 2 is a flow sheet showing the configuration of the cleaning device main body, FIG. 3A is a schematic cross-sectional view showing the configuration of the cleaning chamber, and FIG. It is typical sectional drawing which shows the structure of this cleaning chamber.
The cleaning device 10 according to the present embodiment is a single wafer cleaning device, and is supplied from a cleaning device body 12, a carbon dioxide supply system 14 that supplies carbon dioxide to the cleaning device body 12, and a carbon dioxide supply system 14. A booster pump 16 for increasing the pressure of the liquid carbon dioxide, a cleaning agent supply unit 18 for supplying an etching agent, a compatibilizer, a cleaning agent and the like (hereinafter collectively referred to as a cleaning agent) to the cleaning device body 12, and a cleaning device. The supercritical carbon dioxide discharged from the main body 12 was separated into carbon dioxide gas and a cleaning agent by gas-liquid separation to recover the cleaning agent, and separated by a gas-liquid separation / cleaning liquid recovery unit 20. And a carbon dioxide recovery system 22 that recovers carbon dioxide.

As shown in FIG. 2, the cleaning apparatus main body 12 is configured as a flat-plate container for automatic opening and closing high-pressure processing with a lid, and includes a cleaning chamber 24 that accommodates and cleans a substrate (object to be cleaned) W, and a pressure booster. A heater 26 that heats high-pressure carbon dioxide supplied from the pump 16 to supercritical carbon dioxide, a supply pipe 28 that supplies supercritical carbon dioxide from the heater 26 to the cleaning chamber 24, and a substrate in the cleaning chamber 24 And a cleaning object transport mechanism 29 for transporting the cleaning object such as the above.
The cleaning agent is supplied from the cleaning agent supply unit 18 to the supply pipe 28 and flows into the cleaning chamber 24 together with the supercritical carbon dioxide.
Waste supercritical carbon dioxide flows out from the discharge pipe 30 to the gas-liquid separation / cleaning liquid recovery unit 20 via the pressure adjustment valve 32 and the heater 34. A pressure regulating valve 32 provided in the supercritical carbon dioxide discharge pipe 30 controls the pressure in the cleaning chamber 24, and a heater 34 removes waste supercritical carbon dioxide whose temperature has been lowered by adiabatic expansion in the pressure regulating valve 32. Heat to gasify. On the other hand, the cleaning agent is liquid.

The cleaning chamber 24 constitutes a characteristic part of the present invention. As shown in FIG. 3, the cleaning chamber 24 has a distance (indicated as S in FIG. 3) between the substrate surface on which the particles to be removed adhere and the chamber inner wall facing the substrate surface. ) Is 5 mm or less, preferably 3 mm or less, a disk-shaped chamber chamber 35 that can accommodate the substrate, and a rotation mechanism that rotates the stage supporting the substrate at a rotational speed of 400 rpm or more, preferably 500 rpm or more (in FIG. 3). Only the rotating shaft 31 is shown), and heating means 36 with temperature control devices (not shown) provided above and below the chamber 35 are provided.
The supercritical carbon dioxide containing the cleaning agent is supplied to the chamber chamber 35 from the supply pipe 28 connected to the upper center of the chamber chamber 35, and the waste supercritical carbon dioxide is discharged to the lower center of the chamber chamber 35. 30 (see FIG. 2). Although not described in detail, the rotating shaft 31 of the rotating mechanism also serves as a supporting shaft of a stage that supports the substrate.

In the cleaning apparatus 10, the substrate is cleaned by flowing supercritical carbon dioxide at a flow rate of 1 L / min or more in liquid conversion while rotating the substrate W at a rotation speed of 400 rpm or more, preferably 500 rpm or more.
Preferably, as shown in FIG. 3 (b), the supercritical carbon dioxide that has passed through the substrate surface is circulated again without attenuating the velocity by the fluid circulation path 38, so that the newly supplied supercritical carbon dioxide is obtained. Combine and increase the flow rate for cleaning. The fluid circulation path 38 is provided with a filter 40 for removing fine particles in the circulating supercritical carbon dioxide.

As shown in FIG. 1, the carbon dioxide supply system 14 includes a cooling device 42 that cools and liquefies carbon dioxide gas, and supplies the liquefied carbon dioxide to the booster pump 16. The cleaning agent supply unit 18 includes a booster pump 43, and pressurizes the substrate etching agent and the substrate protective agent (hereinafter collectively referred to as “cleaning agent”) by the booster pump 43, and sends it to the supply pipe 28.
The gas-liquid separation / cleaning liquid recovery unit 20 includes a gas-liquid separation tank 44 and a cleaning agent recovery tank 46. The gas-liquid separation tank 44 separates carbon dioxide gas and liquid cleaning agent into the carbon dioxide gas and the liquid cleaning agent. The gas is sent to the carbon dioxide recovery system 22 and the cleaning agent is recovered in the cleaning agent recovery tank 46.

Embodiment 1 of cleaning method
A cleaning method for cleaning using the above-described cleaning apparatus 10 will be described.
First, an object to be cleaned (or a drying process), in this embodiment, a substrate W on which the micromachine 50 shown in FIG. 23A is placed, is stored in the cleaning chamber 24, the lid is closed, and the cleaning chamber 24 is closed. Is sealed.
Carbon dioxide supplied from the carbon dioxide supply system 14 is pressurized to 7.3 MPa or more by the booster pump 16, heated to 31.1 ° C. or more by the heater 26, and converted to supercritical carbon dioxide.
Further, the cleaning agent is supplied from the cleaning agent supply unit 18, is pressurized to 7.3 MPa or more by the booster pump 43, is press-fitted into the supply pipe 28, and is mixed with supercritical carbon dioxide at a ratio of 0.5 to 20% by weight, for example. To do.

  Since these detergents generally have a higher critical temperature and critical pressure than carbon dioxide, the critical temperature and critical pressure of a fluid mixture with supercritical carbon dioxide are higher than the critical temperature and critical pressure of carbon dioxide. For this reason, it is desirable to maintain the temperature and pressure of the cleaning chamber 24 at, for example, 40 ° C. and 10 MPa or higher so that the cleaning agent dissolves well in supercritical carbon dioxide.

The supercritical carbon dioxide mixed with the cleaning agent in this way is supplied from the upper part of the cleaning chamber 24 to the central part of the substrate W and discharged from the lower part of the substrate W, as shown in FIG. The temperature of the processing fluid is controlled by the heating means 36 with a temperature control device.
When the internal pressure of the cleaning chamber 24 becomes equal to or higher than a predetermined pressure, the pressure regulating valve 32 is opened, and supercritical carbon dioxide containing the cleaning agent is discharged to the gas-liquid separation and cleaning liquid recovery unit 20 via the heater 34. In this way, by appropriately discharging the supercritical carbon dioxide filled in the cleaning chamber 24, the pressure and temperature in the cleaning chamber 24 can be kept constant.

The supercritical carbon dioxide adiabatically expanded by the pressure regulating valve 32 returns to the atmospheric pressure in the gas-liquid separation tank 44, and the cleaning liquid is recovered as the discharged liquid. The removed fine particles are accumulated in the cleaning agent recovery tank 46 after being dissolved or accompanied by the cleaning liquid. On the other hand, the carbon dioxide gas discharged as gas is recovered by the carbon dioxide recovery system 22.
The recovered effluent and carbon dioxide gas can be reclaimed and reused.

Here, the feature of the method of the present embodiment is the configuration and function of the cleaning chamber 24. Unlike the conventional cleaning chamber as shown in FIG. 6, the cleaning chamber 24 of the present embodiment shown in FIG. Cleaning is performed at a substrate rotation speed of 400 rpm or more, preferably 500 rpm or more, a distance from the inner wall of the apparatus facing the substrate surface of 5 mm or less, preferably 3 mm or less, and a supercritical carbon dioxide flow rate of 1 L / min or more in liquid conversion. . Furthermore, as shown in FIG. 3B, the supercritical carbon dioxide that has already flowed along the substrate surface once again is removed while the fine particles are removed by the filter 40 without attenuating the velocity via the fluid circulation path 38. Circulate, join the newly supplied supercritical carbon dioxide, and increase the flow rate of supercritical carbon dioxide to perform cleaning.
With the above-described configuration and function of the cleaning chamber 24, a shearing stress sufficient for removing the fine particles can be obtained, so that the transfer force of the removed fine particles can be increased to effectively remove the fine particles.

  After the fine particles are removed by the flow of supercritical carbon dioxide in the cleaning chamber 24, only the supercritical carbon dioxide not containing the cleaning agent is supplied as a rinse liquid while the substrate W is immersed in the supercritical carbon dioxide. The cleaning agent concentration of critical carbon dioxide is gradually diluted to discharge supercritical carbon dioxide containing cleaning liquid.

  After substituting with supercritical carbon dioxide, the chamber chamber 35 is depressurized to discharge the carbon dioxide, and when cooled, the substrate is filled with gaseous carbon dioxide and dried. In supercritical drying, for example, when carbon dioxide is used as the supercritical fluid, the pressure is maintained while maintaining the temperature at 31.1 ° C. or higher and 7.38 MPa or higher, and then maintaining the temperature at 31.1 ° C. or higher. Is reduced to atmospheric pressure. Thereafter, the temperature is lowered from 31.1 ° C. or higher to room temperature (for example, 20 ° C.). As a result, the cleaning chamber becomes dry.

  In this way, by performing supercritical drying, it is possible to dry the microstructure body layer having a beam shape without destroying it. When other fluid is used as the supercritical fluid, it may be washed and dried at a pressure and temperature suitable for the fluid to be used.

In this embodiment, supercritical carbon dioxide is used as the supercritical fluid, hydrogen fluoride and methanol are used as the cleaning agents, and the addition rates are 0.5% by weight and 5% by volume, respectively.
FIG. 4A shows a graph of the microparticle removal rate of the micromachine processed by the method of the embodiment described above. A fine particle removal rate of 80 to 90% was obtained on wafers of various devices. Moreover, it confirmed that there was no sticking to the movable part (beam structure) of a micromachine after washing | cleaning. That is, in this embodiment method, since the beam structure of the micromachine does not pass through the gas-liquid interface, the particle structure can be removed and dried without destroying the beam structure.

Embodiment 2 of cleaning method
In the present embodiment, when the electron beam exposure mask 70 shown in FIG. 23 (b) is the object to be cleaned and cleaned in the same manner as in Example 1 of the cleaning method, as shown in the graph of FIG. 4 (b), A high fine particle removal rate similar to that of Embodiment 1 was obtained. Further, when the low dielectric constant film 82 shown in FIG. 23 (c) was used as an object to be cleaned and cleaned in the same manner as in Embodiment 1 of the cleaning method, Example 1 was obtained as shown in the graph of FIG. 4 (c). The same high particle removal rate was obtained. Further, it was confirmed that the size and interval of the openings 72 are not changed and the surface is not eroded by cleaning the electron beam exposure mask 70.

  After cleaning by the present embodiment method, the structural change of the low dielectric constant film 82 before filling the wiring trench was examined using infrared absorption analysis. As a result, Si—O, which is a bond of the low dielectric constant film before cleaning, No change was observed in Si-H, Si-OH, Si-CH3, C-H and the like. That is, the structure of the low dielectric constant film does not change even when the cleaning according to the present embodiment is performed, so that the dielectric constant of the low dielectric constant film 82 does not change.

  In the cleaning method of the present embodiment, since the microstructure does not pass through the gas-liquid interface, the microstructure is not destroyed, the surface is not eroded, the mask has no dimensional change, and the low dielectric constant film Fine pores could be removed and dried without crushing the pores.

Embodiment 2 of cleaning apparatus
With reference to FIG.9 and FIG.10, the structure of Embodiment 2 of the washing | cleaning apparatus based on this invention is demonstrated. 9 is a conceptual diagram showing the configuration of the cleaning apparatus, FIG. 10A is a cross-sectional view showing the configuration of the cleaning chamber, and FIG. 10B is a top view of the cleaning chamber. This example is a cleaning apparatus applied to a single wafer type small-diameter substrate cleaning apparatus.

  The cleaning apparatus 101 of this embodiment is roughly divided into a carbon dioxide supply means 102, a pressure increase means 103, a temperature rise means 104, an additive supply means 105 and 106, a dissolution tank 107, and a cleaning tank (hereinafter referred to as a cleaning chamber). 108, gas-liquid separation / additive recovery means 109, and carbon dioxide recovery means (not shown). The cleaning chamber 108 is composed of a cleaning chamber main body 133 containing an embedded heater 132 on which the substrate W to be cleaned is placed, and a chamber lid 134 for forming a sealed space between the cleaning chamber main body 133. . The dissolution tank 107 has an embedded heater 136 and a compatibility confirmation window 137 for confirming whether carbon dioxide and the additive are mixed. The carbon dioxide supply means 102 is composed of a CO2 cylinder containing carbon dioxide (CO2) liquefied by pressurization. The cylinder pressure is, for example, about 15 MPa. The pressure raising means 103 is constituted by a pressure raising pump, and the temperature raising means 104 is constituted by a heater means. A path from the carbon dioxide supply means 102 to the dissolution tank 107 through the pressure raising means 103 and the temperature raising means 104 is connected to each other by a first pipe 111. A cooler 112 is disposed between the carbon dioxide supply means 102 and the pressure raising means 103, and a first switching valve 113 for switching the flow of fluid is provided in the first pipe 111 between the temperature raising means 104 and the dissolution tank 107. It is done. The paths from the additive supply means 105 and 106 to the dissolution tank 107 through the booster pumps 114 and 115 and the line heaters 116 and 117 are connected by a second pipe 118 and a third pipe 119.

  From the dissolution tank 107 to the cleaning chamber 108 via the second switching valve 120 and the line heater 121 for switching the fluid flow, and further from the cleaning chamber 108 to the line heater 122 and the third switching valve 123 for switching the fluid flow. The path leading to the gas-liquid separation / additive recovery means 109 is connected by the main pipe 124. The first switching valve 113 and the second switching valve 120 are connected by a first bypass pipe 125. Further, the second switching valve 120 and the third switching valve 123 are connected by a second bypass pipe 126. A pressure adjusting valve 127 for controlling the fluid pressure is connected to the third switching valve 123. Each cylinder 102, 105, 106 is provided with an open / close cock 128, 129, 130, respectively.

Embodiment 3 of cleaning method
Next, a cleaning method for cleaning using the above-described cleaning apparatus 101 will be described. That is, a series of cleaning processes in the supercritical fluid will be described.
First, an object to be cleaned (for example, a substrate W) to be cleaned (or a drying process) is accommodated in the cleaning chamber 108, and the lid 134 is closed to make the cleaning chamber 108 sealed.
The carbon dioxide from the carbon dioxide supply means 102 is cooled by the cooler 112, pressurized in the liquid state to 3 MPa or more by the pressure raising means 103, and further heated to 31.1 ° C. or more by the temperature raising means 104, To do. This supercritical carbon dioxide is sent to the dissolution tank 107 through the first switching valve 113. On the other hand, an additive such as an etchant from the additive supply means 105, 106 is boosted by the booster pumps 114, 115, heated to a required temperature by the line heaters 116, 117, and supplied to the dissolution tank 107. The additive is mixed with carbon dioxide that has already been heated and pressurized at a rate of 0.5 to 20%, for example. Since these additives generally have a supercritical temperature and a critical pressure higher than those of carbon dioxide, the critical temperature and critical pressure of the mixed fluid with supercritical carbon dioxide are higher than those of carbon dioxide. In the confirmation window 137, it is confirmed whether or not the additive is well dissolved in carbon dioxide to form a homogeneous phase. Until it is confirmed, the second switching valve 120 is switched to the second bypass pipe 126 side, and the fluid in the dissolution tank 107 passes through the second bypass pipe 126 to the gas-liquid separation / additive recovery means 109 side. Shed.

  Initially, when supercritical carbon dioxide is flowing through the first bypass pipe 125 and the second bypass pipe 126, the pressure is controlled to a desired pressure, for example, 25 MPa, by the pressure control valve 127. Control. Thereafter, the first switching valve 113 is switched to supply supercritical carbon dioxide to the dissolution tank 107 and to supply the additive to the dissolution tank 107.

If the additive is not dissolved well in carbon dioxide, the additive supply amount is controlled so as to be dissolved. If dissolved, the second switching valve 120 is switched to the main pipe 124 side to supply the supercritical fluid to the cleaning chamber 108. It is desirable to keep the temperature and pressure of the cleaning chamber 108 high, for example, at 40 ° C. and 10 MPa or more.
In the cleaning chamber 108, the microscopic etching of the substrate W weakens the adhesion force between the fine particles and the substrate, and then the fine particles are separated into the supercritical fluid according to the flow force of the supercritical fluid and discharged. . As a result, the substrate W is cleaned.

  When the fine particles are removed in the cleaning chamber 108, only the supercritical carbon dioxide is supplied as a rinse while the substrate W is immersed in the processing fluid (supercritical fluid + etching agent). Drain the fluid. In order to perform rinsing effectively, the first switching valve 113 is switched to the first bypass piping 125 side, the second switching valve 120 is switched to the cleaning chamber 108 side, and the dissolution tank 107 and the second bypass are switched. Only supercritical carbon dioxide obtained from the carbon dioxide supply means 102 through the pressure raising means 103 and the temperature raising means 104 is passed through the first bypass pipe 125 without passing through the dissolution tank 107 so as not to flow into the pipe 126. It is desirable to supply the cleaning chamber 108 directly. Even if only supercritical carbon dioxide is supplied through the dissolution tank 107, since the etching agent remains in the dissolution tank 107 at the initial stage of rinsing, the fluid supplied from the dissolution tank 107 to the cleaning chamber 108 includes: This is because an etching agent is included.

  The temperature of the processing fluid is controlled by a temperature control device such as an embedded heater 132 and a line heater 121 in the cleaning chamber 108. At this time, when the internal pressure of the cleaning chamber 108 becomes equal to or higher than a certain pressure, the pressure regulating valve 14 is opened, and a processing fluid (supercritical fluid and an additive such as an etching agent) passes through the gas-liquid separation / additive recovery means 109. It is discharged out of the system. In this way, by appropriately discharging the processing fluid filled in the cleaning chamber 108, the pressure and temperature in the cleaning chamber 108 can be kept constant. When the pressure of the solvent or other substance discharged downstream of the pressure regulating valve 127 returns to the atmospheric pressure, a medium (for example, an etching agent) separated as a fluid is recovered as the discharged liquid. The removed fine particles are also accumulated in the cleaning agent recovery means 109 while being dissolved in the liquid. On the other hand, what is discharged as gas (for example, carbon dioxide) is recovered as exhaust gas. The recovered effluent and exhaust gas can be reused in a usable state.

  After the cleaning, the mixed waste liquid of carbon dioxide and the etching agent is separated by the gas-liquid separation / additive recovery means 109 and is recovered and regenerated as necessary. Thereafter, when the pressure of the cleaning chamber 108 is lowered to discharge carbon dioxide and cooled, the substrate W is filled with gaseous carbon dioxide and dried. In the supercritical drying, for example, when carbon dioxide is used as the supercritical fluid, for example, after maintaining the state at 31.1 ° C. or higher and 7.38 MPa or higher, the temperature is maintained at 31.1 ° C. or higher. Reduce the pressure to atmospheric pressure. Thereafter, the temperature is lowered from 31.1 ° C. or higher to room temperature (for example, 20 ° C.). As a result, the inside of the cleaning chamber 108 becomes dry. In this way, by performing supercritical drying, for example, a microstructure layer having a beam shape can be dried without being destroyed. When other fluid is used as the supercritical fluid, it may be washed and dried at a pressure and temperature suitable for the fluid to be used.

  Here, the feature of the present embodiment is the configuration of the cleaning chamber 108. FIG. 10 shows a configuration of the cleaning chamber 108, that is, a configuration of the cleaning chamber 108 for realizing a distance d1 between the surface of the substrate W and the inner wall of the cleaning chamber 108 facing the substrate W of 1 mm or less, preferably 0.5 mm or less. Show. As shown in the cross section of FIG. 10A, the cleaning chamber 108 of the present embodiment is provided with a recess 141 for installing the substrate W at the center of the cleaning chamber main body 133 on the substrate W installation side. The depth d2 of the recess 141 is the same as the thickness of the substrate W. In the case of a silicon substrate, since the substrate thickness is about 0.7 mm, the depth d2 of the recess 141 is about 0.7 mm. At this time, the variation of the depth dimension needs to be 0.1 mm or less. Further, the flatness and roughness of the surface of the recess 141 at the center of the cleaning chamber body 133 and the inner wall surface of the chamber lid 134 must be 0.01 mm or less. When the chamber lid 134 is closed, the distance d1 from the inner wall surface of the chamber lid 134 facing the substrate W to the surface of the substrate W is 1 mm or less, preferably 0.5 mm or less, and in this example about 0.5 mm. As described above, the chamber lid 134 is processed. That is, the distance d3 from the inner wall surface of the chamber lid 134 facing the substrate to the bottom surface of the recess 141 at the center of the cleaning chamber main body 133 is maintained at 1.2 mm or less.

  Further, as the shape of the chamber lid 134, as shown in the cross-sectional view of FIG. 10A, from the portion corresponding to the installation portion of the substrate W toward the supply port 143 side and the discharge port 144 side of the supercritical fluid, Each is a sealed space of a predetermined distance L1 and is formed so as to be inclined so that the distance between the inner wall surface of the chamber lid 134 gradually increases from the surface of the cleaning chamber body 133 where the recess 141 is not formed. Incidentally, when the inner wall surface of the chamber lid 134 is not inclined in this way and a step is formed in the cleaning chamber, the flow velocity of the fluid is reduced at the step portion. However, since the inner wall surface of the chamber lid 134 is inclined to the installation portion on the substrate W as in this example, the flow rate of the fluid flows smoothly on the substrate W without being reduced, and the substrate W Can be washed.

  Further, as shown in FIG. 10B, the chamber lid 134 as viewed in plan view from the portion corresponding to the installation portion of the substrate W toward the supply port 143 side and the discharge port 144 side of the supercritical fluid, Each is formed in a sealed space of a predetermined distance L1 so that the width gradually becomes narrower. The sealed space on the fluid supply port 143 side and the discharge port 144 side in the cleaning chamber 108 has a diameter corresponding to the diameter of the pipe. Further, the cleaning chamber body 133 is formed such that a flow path 145 connected to the pipe is formed on the fluid supply port side and the discharge port side, and a central portion including the concave portion 141 is exposed. The chamber lid 134 is attached to the cleaning chamber main body 133 with bolts 146 at portions corresponding to the fluid supply port side and the discharge port side. As described above, as shown in FIG. 10, the shape of the chamber lid 134 is processed so as to spread obliquely in the cross section and in the lateral direction from the substrate installation portion, so that the cleaning chamber 108 can be easily provided without decelerating the fluid. A flow path capable of flowing a large volume of supercritical fluid can be formed.

A method for loading the substrate W will be described with reference to FIG.
First, the substrate W to be cleaned is stored in the recess 141 of the cleaning chamber body 133. When the chamber lid 134 is closed with the bolt 146 after the substrate W is set, the distance d1 between the surface of the substrate W and the inner wall surface of the chamber lid 134 is held to 1 mm or less, in this example about 0.5 mm. Fluid is supplied in one direction from the supply port (left in the figure) 143 to the discharge port (right in the figure). The fluid flows from the supply port of the cleaning tank, passes between the surface of the substrate W and the inner wall surface of the chamber lid 134 facing the substrate W, and is discharged from the discharge port 144. It is desirable to perform cleaning with a fluid flow rate of 1 L / min or more in liquid conversion.

According to the third embodiment of the cleaning method described above, a shear stress sufficient for removing the fine particles can be obtained, and the fine particles can be effectively removed by increasing the transport force of the fine particles.
As a result of examining the fine particle removal effect of the substrate processed by the cleaning method according to this embodiment, a fine particle removal rate of 80 to 90% was obtained in wafers of various devices. Further, it was confirmed that the micromachine did not stick after cleaning and that the electron beam exposure mask had no change in the hole interval.

  In other words, since the beam structure of the microstructure does not pass through the gas-liquid interface, the cleaning of the present embodiment does not destroy the microstructure, and the surface is not eroded, and the mask is subject to dimensional change. Without removing fine particles, it can be dried. The structural change of the low dielectric film before the formation of the wiring material was examined using infrared absorption analysis. As a result, Si—O, Si—H, Si—CH 3, C—H, etc., which are bonds of the low dielectric film before the cleaning There was no change. That is, since the structure of the low dielectric film does not change even after the cleaning of the present embodiment, the dielectric constant does not change.

  In addition, the photoresist that has been ion-implanted and hardened by the cleaning of the present embodiment is lifted off, and then quickly transported and discharged. That is, since the beam structure of the microstructure does not pass through the gas-liquid interface, the cleaning of this embodiment does not destroy the microstructure, erode the surface, and change the dimensions of the mask. In addition, the pores of the low dielectric film were not crushed, and the fine particle removal, residue removal, photoresist removal, and drying could be performed.

  Here, the embodiment of cleaning and drying of the microstructure described above has been described by taking, as an example, a fine movable element called a micromachine, a mask for electron beam lithography, and wiring using an LSI low dielectric film. It can be applied to many other microstructures in the manufacturing process.

  Next, an embodiment of a microstructure including a semiconductor device obtained by using the above-described cleaning method of the present invention will be described together with its manufacturing method.

11 and 12 show a semiconductor device having a trench capacitor and a manufacturing method thereof.
In the present embodiment, first, as shown in FIG. 11A, a thermal oxide film 152 is grown on one main surface of a silicon semiconductor substrate 151, and a silicon nitride (SiN) film 153, silicon oxide ( A SiO2) film 154 is formed by a CVD (chemical vapor deposition) method.
Next, as shown in FIG. 11B, the silicon oxide film 154 is selectively dry-etched through a photoresist mask (not shown), and the silicon nitride film 153 and heat are further masked using the silicon oxide film 154 as a mask. The oxide film 152 is selectively dry etched to form an opening 155. This opening 155 corresponds to the portion of the trench capacitor to be formed.

Next, as shown in FIG. 11C, the silicon substrate 151 is selectively dry-etched using the three-layer films 152, 153, and 154 as masks to form deep trenches (grooves), that is, trenches 156 having a high aspect ratio. Form. Etching residue removal cleaning is performed.
Next, the inside of the high aspect ratio trench 156 is cleaned by the above-described cleaning method of the present invention before film formation. Next, as shown in FIG. 11D, a high dielectric constant film 157 such as HfSiOx, HfSiOxN, HfAlOx or the like is formed on the three silicon oxide film 154 on the substrate surface including the inner surface of the trench 156, for example, by the CVD method or Ald (Ald). The film is formed by either an atomic layer deposition (MD) method or an MD (Mist deposition) method.

Next, the inside of the high aspect ratio trench 156 is cleaned by the above-described cleaning method of the present invention before film formation. Next, as shown in FIG. 12E, a polysilicon film 158 is formed on the entire surface by a CVD method so as to be embedded in the trench 1567.
Next, as shown in FIG. 12F, all the films 158, 157, 154, 153, and 152 on the silicon substrate 151 are removed by a CMP (chemical mechanical polishing) method, and the substrate surface is planarized.

Next, as shown in FIG. 12G, a polysilicon film 159 and a silicon nitride film 160 are formed on the surface of the substrate so as to be connected to the polysilicon layer 18 by the buried CVD method in the trench 156. Next, as shown in FIG. . Next, the silicon nitride film 160 is selectively dry-etched through a resist mask, and the polysilicon film 159 and the silicon substrate 151 are selectively dry-etched using the silicon nitride film 160 as a mask to form a shallow trench 161 for element isolation. Then, a high-density silicon oxide film 162 is buried in the trench 161 by plasma CVD to obtain a semiconductor device 164 having a target trench capacitor 163.
Since this semiconductor device 164 is cleaned well, this kind of highly reliable semiconductor device can be obtained with a high yield.

13 and 14 show a semiconductor device having a MOS transistor IC and a manufacturing method thereof.
In the present embodiment, first, as shown in FIG. 13A, an element isolation region 172 by a trench is formed on one main surface of a silicon semiconductor substrate 171. That is, after forming a shallow trench 173 in the semiconductor substrate 171, a high-density silicon oxide (SiO 2) layer 174 is buried in the trench 173 by plasma CVD to form an element isolation region 172.
Next, as shown in FIG. 13B, a chemical oxide film 175 is formed on the surface of the semiconductor substrate 171 with thin ozone water.

Next, as shown in FIG. 13C, a high dielectric constant film 176 such as HfSiOx, HfSiOxN, HfAlOx or the like, which becomes a gate insulating film, is formed on the entire surface of the substrate 171 by, for example, a CVD method or an Ald (atomic layer deposition) method. The film is formed by any one of MD (Mist deposition) methods.
Next, as shown in FIG. 13D, an electrode material layer 177 made of, for example, TiN, Ru, Pt or the like, which becomes a gate electrode, is formed on the high dielectric constant film 176 by a CVD method.

Next, as shown in FIG. 14E, the electrode material layer 177 is selectively patterned by dry etching through a resist mask (not shown) to form a gate electrode 177G, and the high dielectric constant film 176 is further formed. The underlying chemical oxide film 175 is similarly patterned to form a gate insulating film 178. This selective etching is a high aspect ratio etching process. Next, so-called etching residue removal cleaning is performed to remove residues generated by etching. This cleaning is performed by the above-described cleaning method of the present invention.
Next, impurities are ion-implanted into the semiconductor substrate 171 by self-alignment using the resist mask and the gate electrode 177G as masks to form the source / drain regions 180 having a low impurity concentration. Next, a silicon nitride (SiN) film is formed on the entire surface by a CVD method, and then etched back to form a sidewall spacer 182 made of a silicon nitride film on the sidewall of the gate electrode 177G. Next, impurities are ion-implanted into the semiconductor substrate 171 by self-alignment using the resist mask and the side wall spacers 182 as masks to form source / drain regions 181 having a high impurity concentration.
Thereafter, the resist mask is peeled off. That is, resist peeling cleaning is performed using the above-described cleaning method of the present invention.

  Next, as shown in FIG. 14F, after pre-deposition cleaning, a Ni sputtered film is formed and annealed in a nitrogen atmosphere, and the surface of the gate electrode 177G and the surface of the source / drain region 181 Then, a nickel silicide layer 183 is formed.

Next, as shown in FIG. 14G, an interlayer insulating film 184 is formed by CVD of PSG (phosphorus silicate glass) or BPSG (boron phosphorus silicate glass). Next, a contact hole 185 is formed in the interlayer insulating film 184 by dry etching through a resist mask. After the contact hole 185 is formed, cleaning before film formation is performed by the above-described cleaning method of the present invention. Next, for example, tungsten is formed by a CVD method so as to be embedded in the contact hole 185, and a contact plug 186 connected to the nickel silicide layer 183 of the required source / drain region 181 is formed. Next, excess tungsten is removed so as to be flush with the surface of the interlayer insulating film 184 by CMP (Chemical Mechanical Polishing), and after planarization, a copper (Cu) wiring 187 is formed by plating. In this way, the semiconductor device 188 having the target MOS transistor IC is obtained.
Since this semiconductor device 188 is well cleaned, this kind of highly reliable semiconductor device can be obtained with a high yield.

15 and 16 show a semiconductor device having damascene wiring and a manufacturing method thereof. In the figure, transistors are not shown, and only damascene wiring and interlayer insulating film portions are shown.
In this embodiment, as shown in FIG. 15A, an interlayer insulating film 191 is formed on a semiconductor substrate (not shown) on which a semiconductor element such as a transistor is formed, and dry etching is performed through a resist mask. A wiring groove 192 is formed, and then a wiring Cu film is formed by plating, and an unnecessary portion of the Cu plating film is removed by a CMP method, and a Cu wiring 193 is embedded in the groove 192.

  Next, a diffusion preventing film 194, a low dielectric constant insulating film 195, an etching stopper film 196, a low dielectric constant insulating film 195, and a diffusion preventing film 194 are sequentially formed. The etching stopper film 196 is formed of a SiC film by CVD, for example. The low dielectric constant film 195 is formed of, for example, a film of SiC, porous silica, polyaryl ether, nanocrystalline silica, or the like by CVD or coating. Next, a trench resist mask 198 having openings 197 having a predetermined pattern on the upper surface is formed.

  Next, as shown in FIG. 15B, the diffusion prevention film 194 and the low dielectric constant film 195 above the etching stopper film 196 are dry-etched to form a trench 199. After etching, etching residue removal cleaning is performed using the above-described cleaning method of the present invention. Next, a new photoresist is formed, and resist patterning is performed to form a via hole mask 200 that covers the inner wall of the trench 199 and forms a trench having a width smaller than that of the trench 199.

  Next, as shown in FIG. 15C, the etching stopper film 196, the low dielectric constant film 195 and the diffusion prevention film 194 are dry-etched to form a via hole 201 reaching the Cu wiring 193. After etching, etching residue removal cleaning is performed using the above-described cleaning method of the present invention.

Next, as shown in FIG. 16D, a diffusion prevention film 202 is formed on the inner surfaces of the trench 199 and the via hole 201 by Ta sputtering, CoWB, or CoWP plating, for example.
Next, as shown in FIG. 16 (e), a wiring Cu film 203 ′ is formed so as to fill the trench 199 and the via hole 201 so as to be connected to the underlying Cu wiring 193.

Next, as shown in FIG. 16F, an unnecessary wiring Cu film 203 ′ is removed by CMP to form a buried Cu wiring 203. Next, using the above-described cleaning method of the present invention, the slurry particles and the polishing powder after the removal by CMP are cleaned. In this way, a semiconductor device 204 having damascene wiring is obtained.
Since this semiconductor device 204 is well cleaned, this kind of highly reliable semiconductor device can be obtained with a high yield.

FIG. 17 shows a semiconductor device manufactured using the trench capacitor shown in FIGS. 11 and 12, the MOS transistor shown in FIGS. 13 and 14, and the damascene wiring forming method shown in FIGS. In FIG. 17, parts corresponding to those in FIGS. 11 and 12, FIG. 13 and FIG. 14, FIG. 15 and FIG.
The semiconductor device 206 is configured by forming a MOS transistor Tr and a trench capacitor 163 on a silicon semiconductor substrate 171 and connecting the MOS transistor Tr and the trench capacitor 163 to a damascene wiring. A SiC film 212, a CVD-SiO 2 film 213, and a passivation film 214 are laminated on the surface of the intermediate wiring layer 215. Of course, the semiconductor device 206 can be obtained with high reliability and high yield.

18, 19, and 20 show a microstructure called a micromachine having a minute movable portion, in particular, a diaphragm structure and a manufacturing method thereof.
In the present embodiment, first, a silicon single crystal substrate 231 is prepared as shown in FIG. Next, as shown in FIG. 18B, a silicon oxide film to be the sacrificial layer 232 is formed on one main surface of the substrate 231 by a CVD method.

Next, as shown in FIG. 18C, a resist mask 233 having a predetermined pattern is formed. Next, as shown in FIG. 18D, the sacrificial layer 232 is patterned by etching through the resist mask 233.
Next, as shown in FIG. 18E, a silicon nitride film 234, a polycrystalline silicon film 235, and a silicon nitride film 236 are sequentially laminated by a CVD method to form a laminated film 237.

  Next, as shown in FIG. 19F, a resist mask 238 having a predetermined pattern is formed on the laminated film 237, and then the laminated film 237 is patterned through the resist mask 238 as shown in FIG. To do. In this patterning, for example, sacrificial layer etching openings 239 reaching the four corners of the sacrificial layer 232a corresponding to the movable part and electrode contact openings 240 reaching the sacrificial layer 232b corresponding to one of the electrode extraction regions are formed. To be done. Here, the sacrificial layer 232b is a film that is used as an insulating film and is not removed later.

Next, as shown in FIG. 19H, for example, an aluminum electrode 241 is selectively formed in the electrode contact opening 240.
Next, as shown in FIG. 19 (i), the sacrificial layer 232a is removed by etching using, for example, a hydrofluoric acid aqueous solution or hydrogen fluoride vapor. Note that in the case where the sacrificial layer 232a is formed of polysilicon, etching is performed using a potassium hydroxide aqueous solution or a xenon fluoride gas. By removing the sacrificial layer 232a, a diaphragm 243 serving as a movable portion is formed with the space 242 interposed therebetween. Thereafter, the residue removal cleaning after the sacrificial layer etching is performed using the above-described cleaning method of the present invention. Thus, the target diaphragm structure 244 is obtained.
FIG. 20 is a completed diaphragm structure 244. The diaphragm 243 is supported by support portions 245 at four corners.
Since this diaphragm structure 244 is cleaned well, this type of diaphragm structure with high reliability can be obtained with a good yield.

When the diaphragm structure 244 is configured as a pressure sensor, the diaphragm 243 of the movable part is formed of a laminated film of silicon nitride film / polycrystalline silicon film / silicon nitride film as described above.
When the diaphragm structure 244 is configured as an infrared sensor, the diaphragm 243 of the movable part is formed of a laminated film of a gold upper electrode / PVDF pyroelectric film / titanium lower electrode.
When the diaphragm structure 244 is configured as a thin film bulk acoustic resonator (FBAR), the diaphragm 243 of the movable part is formed of a laminated film of molybdenum upper electrode / AIN piezoelectric layer / molybdenum lower electrode.

  21 and 22 show a microstructure (a so-called beam type element) having a beam (movable part) and a manufacturing method thereof.

As shown in FIG. 21, a microstructure 251 which is a beam-type element of this embodiment includes a common lower electrode 253 formed on one main surface of a substrate 252 and is opposed to the lower electrode 253 with a space 257 interposed therebetween. In this way, a plurality of both-end supported beam-type beams 254 [254a, 254b, 254c, 254d, 254e, 254f] are arranged. Each beam 254 is formed of a laminated film of an insulating film 255 and an upper electrode 256 that also serves as a reflective film thereon.
The microstructure 251 is electrostatically driven by a voltage applied between the upper electrode 256 and the lower electrode 243 so that the beam 254 approaches and separates from the lower electrode 253 by electrostatic force.

FIG. 22 shows a manufacturing process of the microstructure 251 of FIG. 21, and shows a cross section taken along the line AA ′ and a cross section taken along the line BB ′ of FIG.
First, as shown in FIG. 22A, a sacrificial layer 261 is formed in a predetermined shape on a substrate 252 on which a lower electrode 253 (not shown) is formed. The sacrificial layer 261 may be any material that can be selectively etched with respect to the beam formed on the substrate 252 and the sacrificial layer 261. For example, when a silicon substrate is used as the substrate 252, a SiO 2 film or PSG (phosphorus-doped glass) is generally used as the sacrificial layer 261, and when a SiO 2 substrate is used as the substrate 252, polycrystalline silicon is generally used as the sacrificial layer 261. .
On the sacrificial layer 261, a laminated film 262 including an insulating film 255 to be a beam and an upper electrode film 256 is formed. Further, a resist mask 263 having a pattern corresponding to the beam pattern is formed on the laminated film 262.

  Next, as shown in FIG. 22B, the stacked film 262 and the sacrificial layer 261 are patterned through the resist mask 263 to form a plurality of beams 254 (254a to 254f) by the stacked film 261 together with the sacrificial layer 261. In this example, each beam 254 is formed in a doubly-supported beam structure in which both end portions are directly formed on the substrate 261 as support portions and the remaining movable portion overlaps the sacrificial layer 261. The electrode film 256 is formed of a metal film.

  Next, as shown in FIG. 22C, the sacrificial layer 261 is removed by selective etching. By this selective etching of the sacrificial layer 261, a space 257 is formed below the beam 254 so that the beam 254 can move with respect to the substrate 252. In this selective etching of the sacrificial layer 261, an etching residue 264 is attached.

  Thereafter, the residue removal cleaning after the sacrificial layer etching is performed using the above-described cleaning method of the present invention. As a result, a microstructure 251 having the target beam shown in FIG.

The above-described microstructure 251 can be applied to a sensor element, a resonator, an element having a minute spring, an optical element, and the like. That is, the beam 254 is used for sensors, vibrators, micro springs, optical elements, and the like. The stacked film 262 may be anything as long as the sacrifice layer 261 can be selectively etched later, and a metal film, an oxide film, a semiconductor film, or the like can be used depending on the purpose of the microstructure. A SiN film or a polycrystalline silicon film formed by using a low pressure CVD method is commonly used because of its good mechanical properties and ease of processing.
Since this microstructure 264 is well cleaned, this type of highly reliable microstructure can be obtained with high yield.

The smile structure 251 shown in FIG. 21 can be applied to an optical element using diffracted light, for example.
In this case, among the plurality of beams 254, one other beam 254b, 254d, 254f is a fixed beam, and the other beam 254a, 254c, 254e is a movable beam. The upper electrode 256 of each beam 254 also serves as a reflection film. This optical microstructure 251 is a voltage applied between the upper electrode 256 and the lower electrode 243, and every other beam 254a, 254c, 254e is lowered by an electrostatic force. By approaching / separating from the electrode 253, the diffraction intensity of the diffracted light reflected by the beam 254 is modulated. This optical microstructure 251 can be used as a light modulation element.

In this specification, as an embodiment of cleaning and drying of a microstructure, a low dielectric constant film provided with a minute movable element called a micromachine, a mask for electron beam lithography, and a wiring groove for LSI wiring is exemplified. As described above, the present invention can be similarly applied to many microstructures encountered in the LSI manufacturing process and other microfabrication processes.

It is a block diagram which shows the structure of one Embodiment of the washing | cleaning apparatus of this invention. It is a flow sheet which shows the composition of the main part of a washing device. FIG. 3A is a schematic cross-sectional view showing the configuration of the cleaning chamber, and FIG. 3B is a schematic cross-sectional view showing the configuration of another cleaning chamber. It is a bar graph which shows a particulate removal rate. It is a schematic diagram of Couette-Poise flow between parallel plates. It is typical sectional drawing which shows the structure of the conventional washing | cleaning chamber. The distribution of shear stress on the surface of a 200 mm diameter wafer when the substrate rotation speed is changed is shown. FIG. 8A shows the distribution in the height direction of the surface flow velocity when the distance between the substrate surface and the chamber inner wall is long, and FIG. 8B shows the height of the surface flow velocity when the distance between the substrate surface and the chamber inner wall is short. The direction distribution is shown. It is a schematic block diagram which shows other embodiment of the washing | cleaning apparatus of this invention. 10A and 10B are a cross-sectional view and a top view, respectively, showing the configuration of the cleaning chamber. 11A to 11C are manufacturing process diagrams (part 1) showing a semiconductor device having a trench capacitor using the cleaning method of the present invention and a manufacturing method thereof. FIGS. 12E to 12G are manufacturing process diagrams (part 2) showing a semiconductor device having a trench capacitor using the cleaning method of the present invention and a manufacturing method thereof. FIGS. 13A to 13D are manufacturing process diagrams (part 1) showing a semiconductor device having a MOS transistor using the cleaning method of the present invention and a manufacturing method thereof. FIGS. 14E to 14G are manufacturing process diagrams (part 2) showing a semiconductor device having a MOS transistor using the cleaning method of the present invention and a manufacturing method thereof. 15A to 15C are manufacturing process diagrams (part 1) showing a semiconductor device having a damascene wiring using the cleaning method of the present invention and a manufacturing method thereof. FIGS. 16D to 16F are manufacturing process diagrams (part 1) showing a semiconductor device having damascene wiring using the cleaning method of the present invention and a manufacturing method thereof. It is a block diagram which shows embodiment of the semiconductor device which concerns on this invention. 18 (a) to 18 (e) are manufacturing process diagrams (part 1) showing the diaphragm structure using the cleaning method of the present invention and the manufacturing method thereof. FIGS. 19F to 19I are manufacturing process diagrams (part 2) showing the diaphragm structure and the manufacturing method thereof using the cleaning method of the present invention. FIG. 20 is a schematic perspective view of a diaphragm structure obtained by the manufacturing method of FIGS. 18 and 19. It is a schematic block diagram of the microstructure which has the beam applied as a beam type element. 22 (a) to 22 (d) are cross-sectional views taken along lines AA 'and BB' in FIG. 21, and show a microstructure having a beam using the cleaning method of the present invention and a method for manufacturing the same. It is a manufacturing process figure shown. FIGS. 23A to 23D are a cross-sectional view of a hollow optical micromachine having a doubly-supported beam structure, a cross-sectional view of an electron beam exposure mask 70, a cross-sectional view of a wiring groove, and a diffusion layer of a transistor, respectively. It is sectional drawing of the ion implantation process to.

Explanation of symbols

  10 .... Cleaning device of embodiment, 12 .... Cleaning device body, 14 .... Carbon dioxide supply system, 16 .... Booster pump, 18 .... Cleaning agent supply unit, 20 .... Gas-liquid separation / cleaning liquid recovery unit, 22..CO2 recovery system, 24..Cleaning chamber, 26..Heater, 29..Subject to be cleaned transport mechanism, 28..Supply pipe, 30..Drain pipe, 32..Pressure adjustment valve, 34. · Heater, 35 · · Chamber chamber, 36 · · Heating means with temperature control device (not shown), 38 · · Fluid circulation path, 40 · · Filter, 42 · · Cooling device, · · · Booster pump 44 ..Gas-liquid separation tank 46 ..Cleaning agent recovery tank 50 ..Micromachine 52 ..Supported solid substrate 54 ..Structural film 56 ..Optical film 58. Body, 70 .. mask for electron beam exposure, 7 ..Opening 74..Membrane 76..Support frame 78..Undercoat film 80..Etching stop layer 82..Low dielectric constant film 84..Wiring groove 101..Cleaning device 102. · Carbon dioxide supply means 103 ·· Pressure increase means 104 ·· Temperature increase means 105 · 106 · · Additive supply means 107 · · Dissolution tank 108 · · Cleaning tank (cleaning chamber) 109 · · Gas-liquid Separation / additive recovery means, 113, 120, 123 .. switching valve, 125, 126 .. bypass piping

Claims (34)

In a method for cleaning a microstructure using a supercritical fluid as a cleaning medium,
Cleaning characterized by maintaining the relative velocity of the supercritical fluid with respect to the fine particles to be removed so that the shear stress of the supercritical fluid in the vicinity of the fine particles to be removed attached to the surface of the microstructure is 1 N / m 2 or more. Method. In a method for cleaning a microstructure using a supercritical fluid as a cleaning medium,
The distance between the inner wall of the cleaning chamber and the adhesion surface of the fine particles to be removed is maintained at 5 mm or less, the microstructure is accommodated in the cleaning chamber, and the supercritical fluid is introduced into the cleaning chamber while introducing the supercritical fluid. The cleaning method is characterized in that the microstructure is rotated at 400 revolutions per minute or more. The cleaning method according to claim 2, wherein cleaning is performed while maintaining a flow rate of the supercritical fluid introduced into the cleaning chamber at 1 L / min or more.   The cleaning method according to any one of claims 1 to 3, wherein supercritical carbon dioxide is used as the supercritical fluid.   Add any one of the etching agent, cleaning agent, both of the etching agent and the compatibilizer, both cleaning agent and compatibilizer, and the etching agent, the cleaning agent, and the compatibilizing agent to the supercritical fluid as additives. The cleaning method according to any one of claims 1 to 4, wherein: Single wafer type that houses a microstructure, has a cleaning chamber that cleans the microstructure by introducing a supercritical fluid as a cleaning medium, and cleans the microstructure to remove fine particles adhering to the microstructure In the cleaning device of
The cleaning chamber is configured such that the distance between the inner wall of the cleaning chamber and the adhesion surface of the particles to be removed of the accommodated microstructure is 5 mm or less, and the microstructure is per minute in the supercritical fluid in the accommodation chamber. A cleaning apparatus comprising a rotation mechanism that rotates at a rotation speed of 400 rotations or more. The cleaning apparatus according to claim 6, further comprising a circulation unit that circulates at least a part of the supercritical fluid discharged from the cleaning chamber to the cleaning chamber. In a method for cleaning a microstructure using a supercritical fluid as a cleaning medium,
The distance between the inner wall of the cleaning chamber and the adhesion surface of the fine particles to be removed is maintained at 1 mm or less, the microstructure is accommodated in the cleaning chamber, and a supercritical fluid is introduced into the cleaning chamber at a linear velocity of 3 m / sec. A cleaning method characterized by: The cleaning method according to claim 8, wherein the cleaning is performed while maintaining a flow rate of the supercritical fluid introduced into the cleaning chamber at 1 L / min or more. The cleaning method according to claim 8 or 9, wherein supercritical carbon dioxide is used as the supercritical fluid. Etching agent, cleaning agent, both etching agent and compatibilizer, both cleaning agent and compatibilizing agent, and etching agent, cleaning agent, and compatibilizing agent should be added to the supercritical fluid. The cleaning method according to any one of claims 8 to 10, wherein: Single wafer type that houses a microstructure, has a cleaning chamber that cleans the microstructure by introducing a supercritical fluid as a cleaning medium, and cleans the microstructure to remove fine particles adhering to the microstructure In the cleaning device of
The cleaning chamber is configured such that the linear velocity of the supercritical fluid is 3 m / sec, and the distance between the inner wall of the cleaning chamber and the adhesion surface of the fine particles to be removed is 1 mm or less. Cleaning device. The cleaning apparatus according to claim 12, wherein an inner wall surface of the cleaning chamber is formed obliquely so as to gradually increase a distance from the surface on the microstructure installation side as the distance from the microstructure is increased. The cleaning apparatus according to claim 12 or 13, wherein the cleaning chamber is configured to flow the supercritical fluid in one direction from the inlet toward the outlet. A required region of a semiconductor device is cleaned by a cleaning method using a supercritical fluid having a high surface flow velocity and a shear stress of 1 N / m 2 or more. The required area of the semiconductor device is
A cleaning method using a supercritical fluid in which the distance between the inner wall of the cleaning chamber and the surface to be cleaned is 5 mm or less, the rotation speed of the substrate to be cleaned is 400 rpm or more, and the flow rate of the supercritical fluid is 1 L / min or more. A semiconductor device characterized by being washed. The required area of the semiconductor device is
The semiconductor device is characterized by being cleaned by a cleaning method using a supercritical fluid in which the distance between the inner wall of the cleaning chamber and the surface to be cleaned is 1 mm or less and the flow rate of the supercritical fluid is 1 L / min or more. The semiconductor device according to claim 16 or 17, wherein the semiconductor device is cleaned by the cleaning method using supercritical carbon dioxide as a supercritical fluid. The supercritical fluid is added with any one of the etching agent, the cleaning agent, both the etching agent and the compatibilizer, both the cleaning agent and the compatibilizing solvent, and the etching agent, the cleaning agent, and the compatibilizing solvent. The semiconductor device according to claim 16 or 17, wherein the semiconductor device is cleaned by a cleaning method. After selecting and processing the required aspect ratio,
A method of manufacturing a semiconductor device, comprising: cleaning the processed substrate by a cleaning method using a supercritical fluid having a high surface flow rate and a shear stress of 1 N / m 2 or more. After selecting and processing the required aspect ratio,
The processed substrate is supercritical with a distance between the inner wall of the cleaning chamber and the surface to be cleaned of 5 mm or less, a rotation speed of the substrate to be cleaned of 400 rpm or more, and a supercritical fluid flow rate of 1 L / min or more. A method for manufacturing a semiconductor device, comprising: a step of cleaning with a cleaning method using a fluid. After selecting and processing the required aspect ratio,
Cleaning the processed substrate with a cleaning method using a supercritical fluid in which the distance between the inner wall of the cleaning chamber and the surface to be cleaned is 1 mm or less and the flow rate of the supercritical fluid is 1 L / min or more. A method of manufacturing a semiconductor device. 23. The method of manufacturing a semiconductor device according to claim 21, wherein cleaning is performed by the cleaning method using supercritical carbon dioxide as a supercritical fluid. The supercritical fluid is added with any one of the etching agent, the cleaning agent, both the etching agent and the compatibilizer, both the cleaning agent and the compatibilizing solvent, and the etching agent, the cleaning agent, and the compatibilizing solvent. The method for manufacturing a semiconductor device according to claim 21 or 22, wherein the semiconductor device is cleaned by a cleaning method. A microstructure having a required region of the microstructure is cleaned by a cleaning method using a supercritical fluid having a high surface flow velocity and a shear stress of 1 N / m 2 or more. The required area of the microstructure is
A cleaning method using a supercritical fluid in which the distance between the inner wall of the cleaning chamber and the surface to be cleaned is 5 mm or less, the rotation speed of the substrate to be cleaned is 400 rpm or more, and the flow rate of the supercritical fluid is 1 L / min or more. A microstructure characterized by being washed. The required area of the microstructure is
The microstructure is cleaned by a cleaning method using a supercritical fluid in which the distance between the inner wall of the cleaning chamber and the surface to be cleaned is 1 mm or less and the flow rate of the supercritical fluid is 1 L / min or more. The microstructure according to claim 26 or 27, wherein the microstructure is cleaned by the cleaning method using supercritical carbon dioxide as a supercritical fluid. The supercritical fluid is added with any one of the etching agent, the cleaning agent, both the etching agent and the compatibilizer, both the cleaning agent and the compatibilizing solvent, and the etching agent, the cleaning agent, and the compatibilizing solvent. The microstructure according to claim 26 or 27, wherein the microstructure is cleaned by a cleaning method. After selectively etching the sacrificial layer,
Manufacturing of a microstructure having a step of cleaning a microstructure substrate formed with a hollow portion by a cleaning method using a supercritical fluid having a high surface flow velocity and a shear stress of 1 N / m 2 or more. Method. After selectively etching the sacrificial layer,
For a microstructure substrate having a hollow portion formed therein, the distance between the inner wall of the cleaning chamber and the surface to be cleaned is 5 mm or less, the rotation speed of the substrate to be cleaned is 400 rpm or more, and the flow rate of the supercritical fluid is 1 L / min or more. A method for manufacturing a microstructure, comprising: a step of cleaning by a cleaning method using a supercritical fluid. After selectively etching the sacrificial layer,
A step of cleaning the microstructure substrate in which the hollow portion is formed by a cleaning method using a supercritical fluid in which the distance between the inner wall of the cleaning chamber and the surface to be cleaned is 1 mm or less and the flow rate of the supercritical fluid is 1 L / min or more. A method for producing a microstructure characterized by comprising: The method for producing a microstructure according to claim 31 or 32, wherein the cleaning is performed by the cleaning method using supercritical carbon dioxide as a supercritical fluid. The supercritical fluid is added with any one of the etching agent, the cleaning agent, both the etching agent and the compatibilizer, both the cleaning agent and the compatibilizing solvent, and the etching agent, the cleaning agent, and the compatibilizing solvent. The method for manufacturing a microstructure according to claim 31 or 32, wherein the cleaning is performed by a cleaning method.

Images