JP2007273993A - Method of fabricating semiconductor substrate - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To form an SOI structure without causing an increase in cost and a lowering of reliability. <P>SOLUTION: The method includes steps of: forming a plurality of first trenches on a surface of a semiconductor substrate; and changing the plurality of first trenches into one planar cavity by subjecting the semiconductor substrate to heat treatment. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a method for manufacturing a semiconductor substrate having the same effect as an SOI structure.

  In recent years, electronic devices such as DRAMs are required to have higher speed and power saving. One means for realizing high speed and low power consumption is to use an SOI (Silicon On Insulator) substrate instead of the normally used silicon substrate (bulk silicon substrate).

  An SOI substrate is a substrate having a structure in which a silicon region exists on an insulating region, and there are various types of formation methods, such as a bonding method, a SIMOX (Separation by IMplanted OXygen) method, an ELTRAN (Epitaxial). Layer TRANsfer) method.

  However, since the conventional method for forming an SOI substrate is expensive, there is a problem that it is not suitable for a consumer electronic device such as a DRAM. Furthermore, since it is difficult to form a silicon region (element formation region) with few defects, there is a problem that sufficient reliability cannot be obtained as compared with the case of using a bulk silicon substrate.

  As described above, higher performance of electronic devices can be realized by using an SOI substrate, but there are problems in terms of cost and reliability.

  The present invention has been made in view of the above circumstances, and a representative object thereof is a method of manufacturing a semiconductor substrate having the same effect as an SOI structure that can be formed without causing an increase in cost or a decrease in reliability. It is intended to provide.

  Of the inventions disclosed in this application, the outline of typical ones will be briefly described as follows.

  The method of manufacturing a semiconductor substrate according to the present invention includes a step of forming a plurality of first trenches on a surface of the semiconductor substrate, and heat-treating the semiconductor substrate, thereby forming the plurality of first trenches into one flat plate shape. And a step of converting into a cavity.

  In another semiconductor substrate manufacturing method according to the present invention, a plurality of first trenches are formed on a surface of the semiconductor substrate, and a third trench having a wider opening surface than the first trench is formed. And a step of applying heat treatment to the semiconductor substrate, thereby closing the plurality of first trenches and the third trenches so as to have a flat space region and an opening surface on the surface of the semiconductor substrate. And a step of changing the cavity into a single cavity, and a step of filling the inside of the cavity with an insulating film.

  The preferable form of the manufacturing method of these semiconductor substrates is as follows.

(1) A step of forming a second trench reaching the plate-like cavity on the surface of the semiconductor substrate after forming the plate-like cavity, and a step of filling the inside of the second trench and the plate-like cavity with an insulating film And further.

(2) After forming the flat cavity, an oxide film is formed on the inner surface of the flat cavity by thermal oxidation. Thereafter, the step (1) is performed as necessary.

(3) When the shortest interval between the first trenches is D, and the radius of a circle having the same area as the opening surface of the first trench is R, a plurality of the first trenches are set so that D <4R. An array of trenches is formed.

(4) A silicon substrate is used as the semiconductor substrate.

(5) In the above (4), heat treatment is performed to form a cavity in an atmosphere in which SiO ₂ is reduced under reduced pressure.

(6) In the above (4), heat treatment is performed to form a cavity under reduced pressure and in a hydrogen atmosphere.

(7) In the above (4), heat treatment is performed to form a cavity under reduced pressure and at 1000 ° C. to 1200 ° C.

  With the semiconductor substrate having the structure as in the present invention, a structure having the same function as that of SOI can be formed by the semiconductor substrate manufacturing method of the present invention without causing an increase in cost and a decrease in reliability.

  The reason why the increase in cost can be prevented is that the insulating region of the SOI structure is formed by a simple process of changing a plurality of trenches formed in the semiconductor substrate into one cavity by heat treatment.

  As described above, since the single crystal region can be formed by utilizing surface migration by heat treatment, a silicon substrate including some defects can be used as an initial substrate. As a result, the wafer cost can be reduced. That is, there is a possibility that the cost can be reduced as compared with the conventional transistor formed on the bulk substrate as well as the conventional SOI substrate.

  Further, in this method, since a region where a plurality of trenches are formed has an SOI structure, only a desired region can have an SOI structure. Therefore, by forming the SOI structure only in the region where the SOI structure is required, an increase in cost can be further suppressed and the degree of freedom in device design is increased.

  The reason why the reduction in reliability can be prevented is that the change in shape from the plurality of trenches to one cavity is due to the surface migration of the semiconductor so as to minimize the surface energy of the semiconductor substrate. This is because the crystallinity of the semiconductor region is comparable to that of a normal single crystal semiconductor.

  The above and other objects and novel features of the present invention will become apparent from the description of the present specification and the accompanying drawings.

  According to the present invention, it is possible to realize a semiconductor substrate having the same effect as the SOI structure without causing an increase in cost or a decrease in reliability.

  Hereinafter, embodiments of the present invention (hereinafter referred to as embodiments) will be described with reference to the drawings.

(First embodiment)
FIG. 1 shows a method of forming a silicon substrate having a flat cavity (ESS: Empty Space in Silicon), that is, an SON (Silicon On Nothing) substrate that can be said to be the ultimate SOI substrate, according to the first embodiment of the present invention. It is sectional drawing.

  First, as shown in FIG. 1A, a mask material 2 is formed on a single crystal silicon substrate 1, and a photoresist pattern 3 is formed thereon. The mask material 2 will be described later.

  Next, as shown in FIG. 1B, the mask material 2 is patterned by anisotropic etching such as RIE using the photoresist pattern 3 as a mask, and the pattern of the photoresist pattern 3 is transferred to the mask material 2.

  Next, as shown in FIG. 1C, after the photoresist pattern 3 is carbonized and peeled off, the silicon substrate is patterned by anisotropic etching, for example, RIE, using the mask material 2 as a mask, and a plurality of layers are formed on the surface of the silicon substrate. The trenches 4 are two-dimensionally arranged.

Here, the radius of the trench 4 is 0.2 μm, the depth is 2 μm, and the shortest interval between the trenches 4 (see FIG. 3 described later) is 0.8 μm. The layout of the trench 4 will be described later. The mask material 2 is preferably made of a material having an etching rate sufficiently slower than that of silicon when the silicon substrate 1 is patterned by anisotropic etching. When is used, a silicon oxide film or a laminated film of a silicon nitride film and a silicon oxide film is suitable.

Next, after removing the mask material 2, high-temperature annealing is performed in a non-oxidizing atmosphere under reduced pressure (pressure lower than atmospheric pressure), preferably in an atmosphere for reducing SiO ₂ , for example, 1100 ° C., 10 Torr 100% hydrogen atmosphere. 1D through FIG. 1D, the opening surface of each trench 4 is closed to form a cavity, and the cavities formed in each trench 4 are integrated together. By converting into silicon substrate 1
A flat plate-like cavity 5 is formed in the inside. Although the heat treatment temperature is 1100 ° C. here, it may be higher.

  This shape change is due to the surface migration of silicon that occurs to minimize the surface energy after the silicon oxide film on the surface of the silicon substrate 1 is removed.

  Here, whether or not a flat cavity is formed depends on the initial layout of the trench 4. When the shortest interval between the trenches 4 is 0.8 μm as in this embodiment, the cavities formed at the bottoms of the respective trenches 4 are integrated as shown in FIG. A large flat cavity is formed. However, when the shortest interval between the trenches 4 is 0.9 μm, only spherical cavities 6 are formed in the respective trenches 4 as shown in FIG.

  The layout of the trench 4 will be described in more detail using a plan view. FIG. 3 is a plan view showing the layout of the trench 4. Also shown on the right of the layout of each trench 4 in FIG. 3 is a plan view of a flat cavity 5 formed therefrom. The sectional view taken along the line WW ′ of the layout of each trench 4 corresponds to the sectional view of FIG. 1C, and the sectional view taken along the line WW ′ of the flat plate-like cavity 5 is shown in FIG. It corresponds to a cross-sectional view.

  In the figure, D indicates the interval between the trenches 4 and R indicates the radius of the trench 4. The dimension in the short side direction of the cavity 5 is, for example, about 100 μm. The maximum dimension of the cavity 5 in the long side direction is about the same as that of the chip, while the minimum dimension is about the same as that of the MOS transistor region of the logic portion.

  According to the study by the present inventors, when D> 4.5R, a flat plate-like cavity cannot be formed, and only a spherical cavity is formed below each trench, and D < In the case of 4R, it was found that a flat cavity can be formed. In the case of 4R ≦ D ≦ 4.5R, a flat plate-like cavity may or may not be formed.

  Therefore, in the layout of each trench shown in FIG. 3, by setting D <4R, the cavities formed at the bottom of each trench 4 are integrated, and a flat plate is formed under the region where the trench 4 was initially formed. The hollow 5 can be selectively formed.

  That is, according to the present embodiment, the trench 4 is laid out so as to satisfy D <4R only in the region where the flat cavity 5 is to be formed, so that the flat cavity 5 is formed only under that region. A silicon substrate having a partially flat cavity (dielectric region) in the wafer surface can be formed.

  This means that only a desired region in the wafer surface can be made into the SOI structure, and in this region, the advantages of the SOI substrate such as high speed and low power consumption can be enjoyed. Therefore, the advantages of the SOI substrate can be enjoyed without using an expensive SOI substrate.

  In addition, unlike SOI substrates such as SIMOX and ELTRAN, no defects are caused in the silicon region forming the element. Because the cavity is formed by the surface migration of silicon that minimizes the surface energy of the trench, the crystallinity of the silicon region that forms the device is comparable to that of normal single crystal silicon. is there.

  As a part where such a flat cavity is provided, for example, as shown in FIG. 4, there is a substrate in a LOGIC part of a DRAM / LOGIC mixed load which requires high speed and low power consumption.

  When the plurality of trenches 4 are formed by RIE, thermal oxidation with a thickness of about 10 nm is performed on the inner surfaces of the plurality of trenches 4 immediately before the heat treatment for changing the shape of the plurality of trenches 4 into flat cavities is performed. It is desirable to remove the thermal oxide film after the film is formed. By forming and removing such a thermal oxide film, damage to the silicon substrate 1 caused by RIE can be sufficiently removed.

  In the present embodiment, the case where the shape of the opening surface of the trench 4 is a circle has been described, but the same result can be obtained even in the case of a rectangle. In this case, R is the radius of a circle having the same area as that of the rectangle. The same applies to shapes other than the rectangle.

  Further, even if heat treatment is performed without removing the mask material 2, the flat plate-like cavity 5 can be formed similarly. However, in order to use the flattened surface of the silicon substrate 1, it is preferable to perform a heat treatment after removing the mask material 2 that can simultaneously flatten the substrate surface. Even if the heat treatment is performed without removing the mask material 2, the surface can be flattened by adding a CMP (Chemical Mechanical Polishing) process thereafter.

  The substrate surface on the flat plate-like cavity is slightly lowered with respect to the other substrate surfaces. The reason is that the volume of the cavity formed at the bottom of each trench is smaller than the volume of the initial trench, and the volume of the plate-shaped cavity formed is subtracted from the volume of the plurality of trenches formed earlier. This is thought to be because the substrate surface is lowered by that much. The substrate surface on the flat plate-like cavity is flat.

  This means that there is a step at the boundary between the DRAM portion and the LOGIC portion, considering that the flat cavity is applied to the LOGIC portion of the DRAM / LOGIC mixed mounting. That is, whether the DRAM / LOGIC mixed application to which the present invention is applied or not can be understood whether there is a step at the boundary between the DRAM portion and the LOGIC portion. Similar steps occur in other devices.

  The step is 0.1 μm or less when R = 0.2 μm and D = 0.8 μm. Such a level difference allows exposure without problems. With the current technology, exposure can be performed without problems if it is 0.2 μm or less.

  A specific method for reducing the effect of the step will be described. In the case of light exposure, a pattern that is thinner than the mask (reticle) pattern is transferred onto the resist on the step, so that the portion of the pattern corresponding to the step on the mask (reticle) has a wider width in anticipation of being thinner. It is better to use this pattern. Another method is to use electron beam exposure. This is because electron beam exposure is less susceptible to steps than light exposure.

  As described above, if there is a certain level of difference in level, there is no problem if it is left as it is. However, if the influence cannot be ignored, an area other than the cavity formation area should be preliminarily formed before the flat cavity is formed. It suffices to dig down to the extent that it is lowered, or to lift only the area where the cavities are formed after the flat cavity is formed, or to polish the entire surface by CMP to flatten the surface.

  In the case of digging up the amount of decrease in advance, for example, in a state where the flat cavity formation region is covered with a mask, for example, an oxide film, the surface where the flat cavity is not formed is selectively etched by RIE. Retreat.

  On the other hand, in the case of lifting by the lowered amount, for example, selective epitaxial growth of Si using dichlorosilane and hydrochloric acid may be performed in a state in which a region other than a flat cavity formation region is covered with a mask.

  Further, if a flat cavity is formed by heat treatment for a long time at a high temperature, the entire surface can be flattened.

  As described above, according to the present embodiment, an SOI structure in which the dielectric region is hollow is realized by a simple and damage-free process of changing a plurality of trenches into one flat cavity by silicon surface migration. it can. Therefore, according to the present embodiment, a silicon substrate having an SOI structure can be provided without causing an increase in cost or a decrease in reliability.

  Further, since the position and size of the flat cavity can be controlled by the position and size of the plurality of trenches, an SOI structure having a desired size can be easily introduced into a desired region in the silicon substrate.

  In this embodiment, an example in which one flat cavity is formed in a silicon substrate has been described, but a plurality of flat cavities may be formed in a silicon substrate.

(Second Embodiment)
5 to 7 are sectional views showing a method for manufacturing a MOS transistor according to the second embodiment of the present invention. In the following drawings, the same reference numerals as those in the previous drawings indicate the same or corresponding parts, and detailed description thereof will be omitted.

  In this embodiment, a case where a flat cavity is formed in a silicon substrate and a MOS transistor is manufactured on the flat cavity will be described.

  First, as shown in FIG. 5A, a flat cavity 5 is formed in the silicon substrate 1 by the same method as in the first embodiment shown in FIGS. 1A to 1E. .

  Next, as shown in FIG. 5B, a silicon oxide film 7, a silicon nitride film 8, and a photoresist pattern 9 are sequentially formed on the silicon substrate 1.

  Here, the photoresist pattern 9 is laid out so that at least a part of the opening is on the cavity forming region. The figure shows an example in which the entire opening is laid out on the cavity forming region.

  Next, as shown in FIG. 5C, using the photoresist pattern 9 as a mask, the silicon nitride film 8 and the silicon oxide film 7 are sequentially patterned by anisotropic etching such as RIE, and the pattern of the photoresist pattern 9 is silicon nitrided. Transfer to the film 8 and the silicon oxide film 7.

  Next, as shown in FIG. 5D, after the photoresist pattern 9 is carbonized and peeled off, the silicon substrate 1 is patterned by anisotropic etching, for example, RIE, using the silicon nitride film 8 and the silicon oxide film 7 as a mask. Then, the trench 10 connected to the flat cavity 5 is formed.

  Next, as shown in FIG. 6E, a silicon thermal oxide film 11 is formed on the inner surface of the flat cavity 5 by thermal oxidation. Next, as shown in FIG. 4E, after the silicon oxide film 12 is deposited on the entire surface so as to fill the flat cavity 5 and the trench 10, unnecessary portions outside the flat cavity 5 and the trench 10 are unnecessary. The silicon oxide film is removed by CMP to planarize the surface. At this time, it is not necessary to completely fill the inside of the flat cavity 5 with the silicon oxide film 12, and it is sufficient to at least completely fill the trench 10.

  Next, as shown in FIG. 6F, after forming a photoresist pattern 13 for forming element isolation (STI), the silicon nitride film 8 and the silicon oxide film 7 are anisotropically etched using this as a mask. Patterning is sequentially performed by RIE, and the pattern of the photoresist pattern 13 is transferred to the silicon nitride film 8 and the silicon oxide film 7.

  Next, as shown in FIG. 6G, after the photoresist pattern 13 is carbonized and peeled off, the silicon substrate 1 is patterned by anisotropic etching, for example, RIE, using the silicon nitride film 8 and the silicon oxide film 7 as a mask. Thus, the element isolation trench 14 is formed. At this time, the thermal oxide film 11 formed on the inner surface of the flat cavity 4 functions as an RIE stopper.

  Next, as shown in FIG. 6H, after a silicon thermal oxide film 15 is formed on the side surface of the element isolation trench 14 by thermal oxidation, a silicon oxide film 16 is embedded in the element isolation trench 14 to flatten the surface. To.

  For example, after the silicon oxide film 16 is deposited on the entire surface by CVD so as to fill the inside of the element isolation trench 14, unnecessary silicon oxide film 16 outside the element isolation trench 14 is removed by CMP. By doing.

Next, as shown in FIG. 7I, the silicon nitride film 8 and the silicon oxide film 7 are removed. The silicon nitride film 8 is removed using a heated H ₃ PO ₄ solution, and the silicon oxide film 7 is removed using a hydrofluoric acid solution.

  Next, as shown in FIG. 7J, the surface of the silicon substrate 1 is thermally oxidized to form a gate oxide film 17 on the surface. The thermal oxidation is performed, for example, at 900 ° C. in a mixed gas atmosphere of oxygen and HCl. Although an oxide film is used as the gate insulating film here, other insulating films such as a tantalum oxide film and an oxynitride film may be used.

  Next, as shown in FIG. 7K, a conductive film is formed on the entire surface of the substrate, and this is patterned to form the gate electrode 18.

  Examples of the conductive film include a polycrystalline silicon film, a laminated film of a polycrystalline silicon film and a metal silicide film, and a metal film. Each of the polycrystalline silicon films contains impurities and has a lower resistance than that of an undoped polycrystalline silicon film.

  A polycrystalline silicon gate when a polycrystalline silicon film is used, a polycide gate when a laminated film of a polycrystalline silicon film and a metal silicide film is used, and a metal gate MOS transistor when a metal film is used Will be formed respectively. In the case of a metal gate, a so-called damascene gate should be adopted (A. Yagishita et al. IEDM1998 p.785).

  Next, as shown in FIG. 7 (k), impurity ions are implanted into the silicon substrate 1 using the gate electrode 15 as a mask, and then annealing for activating the impurity ions is performed to form a shallow, low-concentration diffusion layer. (Extensions) 19 and 20 are formed.

  Finally, as shown in FIG. 7L, a gate side wall insulating film 21 is formed by a known technique (side wall remaining), and impurity ions are formed on the silicon substrate 1 using the gate side wall insulating film 21 and the gate electrode 18 as a mask. Then, annealing for activating the impurity ions is performed to form the source diffusion layer 22 and the drain diffusion layer 23, thereby completing the MOS transistor having the LDD structure.

  Note that the annealing in the step of FIG. 7 (k) may be omitted, and the activation of impurity ions may be performed collectively by the annealing in the step of FIG. 7 (l).

  Furthermore, in the present embodiment, when forming the trench 10 and the element isolation trench 14, a mask material made of the silicon nitride film 8 and the silicon oxide film 7 is used. However, in consideration of the selection ratio in etching with silicon, It is desirable to use a mask material made of the silicon oxide film 7 ′, the silicon nitride film 8, and the silicon oxide film 7.

  The MOS transistor described in the above embodiment is preferably used for, for example, a MOS transistor constituting a DRAM / LOGIC mixed LOGIC. In this case, the advantages of SOI such as high speed and low power consumption can be enjoyed in the LOGIC region.

  Here, the manufacturing process of the MOS transistor in the LOGIC region includes an etching step for forming a plurality of trenches and a heat treatment step for changing the plurality of trenches into one flat cavity as compared with that of the MOS transistor in the DRAM region. However, the manufacturing process is basically the same.

  Therefore, since the conventional DRAM / LOGIC mixed mounting process can be followed almost as it is, the DRAM / LOGIC mixed mounting that can enjoy the advantages of SOI such as high speed and low power consumption can be easily realized in the LOGIC region.

(Third embodiment)
FIG. 8 is a cross-sectional view showing a MOS transistor manufacturing method according to the third embodiment of the present invention. In the second embodiment, a method of filling a flat cavity with a silicon oxide film has been described, but in this embodiment, a method of leaving a flat cavity without filling it with a silicon oxide film will be described.

  First, as shown in FIG. 8A, a plate-like cavity 5 is formed in the silicon substrate 1 by the same method as in the first embodiment shown in FIGS. 1A to 1E. .

  Next, as shown in FIG. 8B, a silicon thermal oxide film 24 is formed on the inner surface of the flat cavity 5 and the surface of the silicon substrate by thermal oxidation. The thermal oxidation is performed, for example, at 900 ° C. in a mixed gas atmosphere of oxygen and HCl. The silicon thermal oxide film 22 serves as a stopper at the time of RIE as shown in FIG.

  Next, as shown in FIG. 8C, after a silicon nitride film 25 is formed on the silicon substrate 1 via the silicon thermal oxide film 24, a photoresist pattern for forming element isolation (STI) thereon is formed. 26 is formed.

  Next, as shown in FIG. 8D, using the photoresist pattern 26 as a mask, the silicon nitride film 25 and the silicon thermal oxide film 24 are sequentially patterned by anisotropic etching such as RIE, and the pattern of the photoresist pattern 26 is changed to silicon. Transferred to the nitride film 25 and the silicon thermal oxide film 24.

  Next, after removing the photoresist pattern 21, the MOS transistor having the LDD structure shown in FIG. 9 is completed through steps similar to the steps after FIG. 6F shown in the second embodiment.

  In this embodiment, the same effect as that of the second embodiment can be obtained. Further, in this embodiment, there is no step of filling the flat cavity 5 with a silicon oxide film, so that the process can be simplified. It is done.

(Fourth embodiment)
FIG. 10 is a process cross-sectional view illustrating a MOS transistor manufacturing method according to the fourth embodiment of the present invention.

  First, as shown in FIG. 10A, a mask material 2 and a photoresist pattern 27 are sequentially formed on a silicon substrate 1.

  Here, the photoresist pattern 27 is different from the photoresist pattern 3 of FIG. 1A of the first embodiment in that it is in the vicinity of the pattern in addition to the pattern (opening) corresponding to the plurality of trenches 4. In addition, it has a pattern (opening) corresponding to a trench whose opening area is wider than that of the trench 4.

  Next, using the photoresist pattern 27 as a mask, the mask material 2 is patterned by anisotropic etching, for example, RIE, the pattern of the photoresist pattern 27 is transferred to the mask material 2, and then the photoresist pattern 27 is carbonized and peeled off.

  Next, as shown in FIG. 10B, the silicon substrate is patterned by anisotropic etching, for example, RIE, using the mask material 2 as a mask, and a plurality of trenches 4 are formed on the surface of the silicon substrate and these trenches 4 are adjacent to the trenches 4. A trench 28 having a larger opening area is formed.

  Next, as shown in FIG. 10C, after the mask material 2 is peeled off, a plurality of high temperature annealing is performed in a non-oxidizing atmosphere under reduced pressure, for example, 1100 ° C., 10 Torr 100% hydrogen atmosphere. The trench 4 and the trench 28 are changed into one unclosed cavity 5 ′ having a flat space area and having an opening surface on the substrate surface.

  Here, with respect to the plurality of trenches 4, as shown in the first embodiment, since the shape change due to the surface migration of silicon is used, a spherical cavity is formed at the bottom of each trench 4, and as a result, A flat plate-like cavity is formed, but only the corner is rounded below the large trench 28.

  FIG. 12 shows a layout of the trench 4 and a plan view of the cavity. This corresponds to FIG. 3, and the left plan view (trench layout) in FIG. 12 corresponds to the left plan view (trench layout) in FIG. 3, and the right plan view in FIG. ) Corresponds to the plan view (flat cavity) on the right side of FIG.

  Here, since the large trench 28 is a trench for oxidizing the inner surface of the cavity 5 as shown below, the number of trenches may be one or more, and the position thereof is obtained by the shape change of the plurality of trenches 4. The position is not limited to the position shown in FIG. 12 and may be any as long as it is in the vicinity of the plurality of trenches 4. The cross-sectional shape of the large trench 28 is also arbitrary.

  Next, as shown in FIG. 10D, after the silicon thermal oxide film 11 is formed on the inner surface of the cavity 5 ', a silicon oxide film 12 is deposited on the entire surface so as to fill the cavity 5'.

  Next, as shown in FIG. 11E, the unnecessary silicon oxide film 12 outside the cavity 5 'is removed by CMP to flatten the surface.

  Next, as shown in FIG. 11F, a photoresist pattern 31 for forming a silicon oxide film 29, a silicon nitride film 30, and an element isolation trench (STI) is sequentially formed on the substrate.

  Next, as shown in FIG. 11G, using the photoresist pattern 31 as a mask, the silicon nitride film 30 and the silicon oxide film 29 are sequentially patterned by anisotropic etching, for example, RIE, and the pattern of the photoresist pattern 31 is silicon nitrided. The film 30 and the silicon oxide film 29 are transferred.

  Next, after the photoresist pattern 31 is carbonized and peeled off, the MOS transistor having the LDD structure shown in FIG. 13 is completed through steps similar to those shown in FIG. 6F and subsequent steps shown in the second embodiment.

(Fifth embodiment)
In the present embodiment, improvement techniques applicable to the first to fourth embodiments will be described. In the above-described method for forming a silicon substrate (SON substrate) having a flat cavity, a step is inevitably generated at the end of the formation region of the cavity 5 (see FIG. 14).

  The level difference becomes a problem when an attempt is made to produce a device on the silicon substrate 1 on the cavity 5. For example, when a metal film to be an electrode is patterned across a step, patterning cannot be performed as designed, resulting in problems such as shorting or opening of wiring. Furthermore, when the oxidation treatment is performed, stress is generated in the substrate near the step, causing problems such as crystal defects.

  As a method of eliminating this type of level difference, for example, a method of flattening the surface using a CMP method or an RIE method can be considered. The latter method is a method in which a region having a lower surface is covered with a mask film such as an oxide film, and a region having a higher surface is etched by RIE to eliminate the step. However, in any of the methods, it is necessary to add one or more processes separately in order to eliminate the level difference, resulting in an increase in the number of processes and a complicated manufacturing process.

  Therefore, in the present invention, a plurality of trenches having a small aspect ratio are arranged in advance in a region where the cavity 5 is not formed. The trench formed at this time is a trench (dummy trench) having a small aspect ratio so that a cavity cannot be formed in the lower portion of the trench, and the density is designed so as to eliminate a predicted step. By forming the trench designed in this way in advance, the step at the end of the formation region of the cavity 5 can be easily eliminated.

  Hereinafter, a method for forming a SON substrate using the above improvement technique will be described with reference to FIG.

  First, as shown in FIG. 15A, as in the first embodiment, a mask material 2 and a photoresist pattern 3 are formed on a silicon substrate 1, and the mask material 2 is etched using the photoresist pattern 3 as a mask. Then, the pattern of the photoresist pattern 3 is transferred to the mask material 2.

  Next, as shown in FIG. 15B, after the photoresist pattern 3 is peeled off, the silicon substrate 1 is patterned using the mask material 2 as a mask to form trenches 4 and 4 '. Here, the aspect ratios of the trenches 4 and 4 ′ are different from each other, and the densities are also different from each other. The aspect ratio and density will be described later.

  Next, as shown in FIG. 15C, the silicon oxide film 2 is removed with an aqueous hydrogen fluoride solution.

  Next, the silicon substrate 1 in this state is heat-treated in a reducing atmosphere. This heat treatment causes silicon surface migration so that the surface energy of the silicon substrate 1 is minimized.

  As a result, the shape of the region where the trench 4 is formed changes as shown in FIGS. 15D and 15E, and a plate-like cavity 5 is formed in the silicon substrate 1. At this time, the substrate surface on the area where the cavity is formed is lower than that in the step of FIG.

  On the other hand, the shape of the region where the trench 4 'is formed changes as shown in FIGS. 15D and 15E, and the trench 4' disappears but the cavity 5 is not formed. At this time, the substrate surface on the region where the trench 4 ′ has disappeared becomes as low as the substrate surface on the region where the cavity is formed. As a result, the cavity 4 can be formed in the silicon substrate 1 without causing a step as shown in FIG.

  Hereinafter, each process will be described in detail.

First, the shape and number of cavities obtained with respect to the initial trench shape will be described with reference to FIGS. 16 and 17. As shown in FIG. 16, when the initial trench shape is cylindrical, the resulting cavity has a spherical shape. If the initial radius of the cylindrical trenches and R _R, the radius R _S of the spherical cavity is 1.88R _R, is Toi隔λ between cavities of the two spherical adjacent vertical become 8.89R _R.

Therefore, as shown in FIG. 17, the number of cavities obtained can be estimated by dividing the depth L of the initial cylindrical trench by the cavity interval λ. The inventors of the present invention formed a trench with a radius R _R = 0.2 μm, and examined it by changing the depth L to 1 μm and 2 μm.

  As a result, when the depth was 1 μm for heat treatment under the same conditions, for example, 1100 ° C., 10 Torr, 10 min in a hydrogen atmosphere, the trench disappeared and the substrate surface was simply flattened. . On the other hand, when the depth was 2 μm, one spherical cavity was formed. This result was consistent with the number of cavities estimated from the graph shown in FIG. 17, and it was confirmed that the number of cavities could be estimated using FIG.

  Next, the aspect ratio and density of the trench to be formed will be described. The trench 4 is for forming the cavity 5 in the substrate 1. For that purpose, the aspect ratio of the trench 4 needs to be 5 or more. Further, in order to form the tubular or plate-like cavity 5, it is necessary to arrange the trenches 4 in advance in a linear or lattice shape. In this case, the interval D between the trenches 4 needs to be set so that D <4R with respect to the radius R of the trench 4.

On the other hand, the trench 4 ′ is for eliminating a step generated when the cavity 5 is formed. For this purpose, the aspect ratio of the trench 4 ′ needs to be 3 or less so as not to cause a cavity in the silicon substrate 1. The density of the trench 4 ′ is determined by the size of the step. For example, when trenches 4 ′ having a radius of 0.2 μm and a depth of 2 μm are formed with a density of 1.6 per unit area (/ μm ² ), the level difference after forming the cavity 5 is 0.12 μm. Met. In this case, for example, trenches 4 ′ having a radius of 0.5 μm and a depth of 2 μm may be formed with a density of 0.76 (/ μm ² ).

  As described above, according to the present embodiment, when forming trenches to be cavities, a plurality of dummy trenches whose aspect ratio and density are designed so as not to be cavities are simultaneously formed, thereby forming a process. The step generated at the end of the cavity formation region in the silicon substrate can be easily eliminated without increasing the number and complicating the manufacturing process. Here, the case where the shape of the cavity is particularly plate-like has been described, but other shapes may be used. That is, the method described here is effective regardless of its shape as long as it is a cavity where a step is generated.

(Sixth embodiment)
In the present embodiment, other improvement techniques applicable to the first to fourth embodiments will be described. In the above-described method for forming a SON substrate having a flat ESS, when a large-area ESS is formed, the flat ESS is crushed.

  Specifically, when the ESS width is as small as 20 μm, the flat ESS is not crushed as shown in FIG. 18A, but when the ESS width is as large as 180 μm, FIG. As shown in FIG. 18C, which is an enlarged view, the flat ESS is crushed. In FIG. 15, the heat treatment for changing the trench to ESS was 1100 ° C., 10 Torr, 10 min in a 100% hydrogen atmosphere.

  According to the earnest research by the present inventors, as will be described in detail below, by finding an effective calculation formula for obtaining an ESS of a size that does not collapse, and by devising a heat treatment for changing the trench to ESS, ESS It became clear that even if the width was increased, the ESS could not be crushed.

  First, the results of calculating the strength of the ESS structure will be described. FIG. 19 shows a model of the ESS structure used for the above calculation. Assume that the ESS width is a (μm), the ESS depth is b (μm), and the thickness of the silicon layer on the ESS is t (μm). At this time, the deflection δ (μm) of the silicon layer is expressed by the equation (1).

δ = αPa ⁴ / Et ³ (1)
Here, P represents a load applied to the silicon layer. E represents Young's modulus, and in the case of silicon, E = 0.13 (N / μm ² ). α is a dimensionless coefficient that varies depending on the ESS structure (= b / a). When the ESS structure is rectangular and b / a ≧ 2, it is 0.0284, and the ESS is square and b / a = 1. In this case, it is given by 0.0138. The following calculation shows the case of b / a ≧ 2.

First, the self-weight was considered as a load applied to the silicon layer. As a result of calculating the deflection due to its own weight with respect to the ESS structure with t = 1 μm and a = 180 μm, it was found that δ = 5.2 × 10 ⁻⁶ (μm) was very small and negligible. Furthermore, when a larger structure was estimated as a = 1 mm, it was found that the deflection due to its own weight was sufficiently small even in the case of an ESS structure having a large area of δ = 5 × 10 ⁻³ (μm). From the above calculation results, it was found that the shape change due to its own weight has little effect.

Next, the load due to the difference between the pressure inside the ESS and the atmospheric pressure was considered. The pressure inside the ESS is equal to or less than the pressure during the heat treatment during the ESS formation. Therefore, for example, when the heat treatment pressure is 10 Torr, a load of almost atmospheric pressure (1.013 × 10 ⁻⁷ (N / μm ² )) is applied.

Therefore, similarly to the calculation of the dead weight, the bending due to the atmospheric pressure load was calculated for the ESS of t = 1 μm and a = 180 μm. As a result, it was found that δ was as large as 23.2 μm and the ESS was crushed. On the other hand, when a = 20 μm and the ESS width was reduced, it was found that δ = 3.5 × 10 ⁻³ μm and the shape change due to the pressure load could be ignored. This is in good agreement with the result shown in FIG. 18, and means that an ESS having a size capable of avoiding the collapse can be designed using Equation (1).

  Next, using formula (1), it was estimated how much ESS can actually be realized. FIG. 20 shows a calculation result of how much the silicon layer is bent with respect to the plate width (ESS width) when the thickness t of the silicon layer is 0.1 μm and 1 μm.

  FIG. 20 shows that when the thickness t of the silicon layer is as thick as 1 μm, the deflection δ of the silicon layer is sufficiently small even when the ESS width is 20 μm. On the other hand, when the thickness t of the silicon layer is as thin as 0.1 μm, it can be seen that even when the ESS width is 10 μm, the silicon layer is bent by 0.1 μm or more. Since the thickness of the ESS is approximately the same as the thickness t of the silicon layer, it can be expected that the ESS will be crushed. That is, it was found that when the thickness t of the silicon layer is 0.1 μm, an ESS having an ESS width of about 8 μm or more cannot be realized.

  The present inventors have found that the process sequence shown in FIG. 21 is effective as a method for forming a large-area ESS. That is, after performing the first heat treatment for forming the ESS structure, the second heat treatment is continuously performed without opening the chamber, and the pressure inside the ESS is adjusted.

  The first heat treatment is a process for forming an ESS. Therefore, the first heat treatment is desirably performed under conditions of high temperature and reduced pressure at which Si surface migration easily occurs on the surface of the silicon substrate. For example, the first heat treatment may be performed under conditions of 1100 ° C., 10 Torr, and 10 minutes. The atmosphere for the heat treatment may be a non-oxidizing atmosphere, and for example, a 100% hydrogen atmosphere is desirable.

The second heat treatment is a process for adjusting the pressure inside the ESS. Therefore, the second heat treatment is desirably performed under conditions of low temperature and high pressure. The atmosphere for the heat treatment is preferably an atmosphere containing an element having a large diffusion coefficient in silicon, for example, an atmosphere containing hydrogen or a 100% hydrogen atmosphere. The hydrogen diffusion coefficient D (cm ² / s) is given by equation (2).

D = 4.2 × 10 −5 exp (−0.56 / kT) (2)
k is the Boltzmann constant and T is the absolute temperature (K). From equation (2), the diffusion length of hydrogen at 200 ° C. is estimated to be 1 μm in 60 seconds. Therefore, hydrogen can diffuse to the inside of the ESS even at a low temperature of 200 ° C. As a result, the pressure inside the ESS can be effectively varied. That is, by performing the second heat treatment in a hydrogen atmosphere, the pressure inside the ESS can be changed to a pressure equivalent to the pressure during the heat treatment.

  Further, considering that the pressure also decreases in proportion to the temperature from the ideal gas law (PV = nRT), the pressure is lowered during the temperature lowering process during the second heat treatment. Therefore, it is desirable to perform the second heat treatment under pressure in advance. For example, when the temperature of the second heat treatment is 600 ° C., the heat treatment pressure at 600 ° C. may be 3 atm.

  As described above, by adjusting the pressure inside the ESS by the second heat treatment, the load due to the pressure difference between the pressure inside the ESS and the atmospheric pressure can be reduced or eliminated. Can be formed. Further, even if the SON layer is thinned for device fabrication, an element can be formed on the SON layer while maintaining its shape without being crushed.

(Seventh embodiment)
In the case where a transistor is formed on a silicon layer (SON layer) on the ESS of the SON substrate, the thickness of the SON layer needs to be 0.1 μm or less in order to fully exploit the advantages of the SON substrate. However, when the thickness of the large-area SON layer is reduced, as described above, the SON layer is greatly bent by the pressure load.

  FIG. 22 shows the relationship between the thickness of the SON layer and the amount of deflection obtained by calculation using the formula (1). The ESS width of the SON layer was 20 μm. From FIG. 22, after the SON layer thickness is 1 μm, the deflection is negligibly small, whereas when the SON layer thickness is reduced to 0.1 μm, the deflection amount is 1 μm or more. It can be seen that the ESS structure is crushed.

  Considering the above results, it can be said that the second heat treatment is effective to be performed after forming the ESS structure by the first heat treatment and before the step of thinning the SON layer at the time of device fabrication. In the second heat treatment, by increasing the pressure inside the ESS to near atmospheric pressure, a thin SON layer can be formed without being crushed.

  The technique for preventing the flat-area ESS from being crushed will be further described in the fifteenth embodiment. However, as shown in FIG. 10, when the cavity 5 ′ partially opened at the time of forming the cavity is formed, it is not necessary to consider the load due to the pressure difference. An ESS having a large area can be formed.

(Eighth embodiment)
In the present embodiment, an SON substrate is described which has the same effect as a silicon substrate (strained substrate) in which a SiGe layer or the like is buried directly under a channel and can solve the problems of the strained substrate.

  First, a conventional strained substrate will be described. One of the main objectives of transistor miniaturization in LSI is to realize a high-performance LSI by increasing the speed of transistors. However, in recent years, the transistor has entered a region where the gate length is 0.1 μm or less, and its miniaturization is becoming increasingly difficult.

  Against this background, a strained substrate in which a heterogeneous composition layer such as a SiGe layer is buried immediately below the channel near the surface of the silicon substrate has been proposed as a method for realizing high speed without relying on miniaturization. ing.

  According to this type of strained substrate, Si in the vicinity of the substrate surface is distorted by the heterogeneous composition layer, thereby improving the mobility of carriers (electrons or holes) and realizing high performance of the transistor. It becomes.

  However, embedding a different composition layer such as a SiGe layer causes a problem of generation of crystal defects due to lattice distortion. This problem becomes more prominent as the Ge concentration of the SiGe layer is increased in order to increase the lattice distortion. That is, in the conventional strained substrate, how to form the SiGe layer containing Ge at a high concentration inside the substrate without causing crystal defects has been a big problem in the process.

  Hereinafter, a method for forming a SON substrate according to the eighth embodiment of the present invention, which can solve the above problem, will be described with reference to FIG.

  First, as shown in FIG. 23A, a plurality of trenches 4 are formed in an array on the surface of a single crystal silicon substrate 1 having a (100) plane orientation by using a well-known lithography method and RIE method.

  Next, as shown in FIG. 23B, silicon on the surface of the silicon substrate 1 is caused to flow by heat treatment at a pressure of 10 Torr and 1100 ° C. for 3 minutes in a mixed atmosphere of hydrogen and argon, thereby forming a cavity 3. . The thickness (size in the substrate depth direction) of the cavity 3 formed by such heat treatment was 1.2 μm, and the thickness of the silicon layer (SON layer) 33 on the cavity 3 was 0.6 μm.

  Next, as shown in FIG. 23C, a trench 10 reaching the cavity 5 is formed by using well-known photolithography and etching. The opening surface of the trench 10 is a rectangle of 0.3 μm × 0.5 μm, and the depth of the trench 10 is 2.5 μm.

  Next, as shown in FIG. 23D, the surface of the silicon substrate 1 is thermally oxidized to form a silicon oxide film 32 having a thickness of 0.4 μm. As a result of such thermal oxidation, the thickness of the SON layer 33 decreased from 0.6 μm to 0.4 μm.

  Finally, as shown in FIG. 23E, the silicon oxide film 32 on the silicon substrate 1 is selectively removed using the RIE method, and the silicon oxide film 32 is selectively left in the cavity 5 and the trench 10. As a result, the SON substrate is completed.

  When the internal stress in the SON layer 33 of the SON substrate thus obtained was measured by Raman spectroscopy, it was confirmed that a tensile stress of 250 MPa was present.

  The cause of such tensile stress is that the silicon substrate 1 has a larger thermal expansion coefficient than the silicon oxide film 32. When the silicon substrate 1 is oxidized at a high temperature, the strain is relaxed. On the other hand, when the high temperature silicon substrate 1 is cooled to room temperature, the strain is not relaxed. As a result, a tensile stress is generated on the silicon substrate 1 side having a relatively larger thermal expansion coefficient than that of the silicon oxide film 32.

  When the internal stress in the SON layer was measured in the same manner for the SON substrate on which the silicon oxide film 32 formed for comparison was not formed, a significant stress value was not found. This is because the structure obtained in the thermal oxidation process for forming the silicon oxide film 32 and the subsequent etching process for selectively leaving the silicon oxide film in the cavity 5 and the trench 10 is the SON layer 33. It is useful as a method for intentionally forming a stress field inside the film.

  Furthermore, since the SON substrate of this embodiment does not embed a different composition layer such as a SiGe layer, the problem of generation of crystal defects due to lattice distortion does not occur in principle.

  Furthermore, it has been found that the SON substrate of this embodiment has an advantageous structure as compared with a conventional oxide film embedded substrate (SOI substrate). Even in the conventional SOI substrate, since an oxide film exists under the SOI layer, the same effect as that of the SON substrate of this embodiment can be expected in principle.

  However, in the case of a conventional SOI substrate, since the oxide film is too thin compared to the SOI layer, for example, the oxide film is 1 μm or less and the SOI layer is 1 mm. Therefore, a large stress is generated in the SOI layer by the oxide film. Can not.

  On the other hand, in the case of the SON substrate of the present embodiment, the thickness of the SON layer 33 corresponding to the SOI layer of the conventional SOI substrate is 0.6 μm, that is, the SON layer 33 and the silicon oxide film 32 are of the same level. Therefore, a large stress can be generated in the SON layer 33.

  FIG. 24 shows a cross-sectional view of a MOS transistor manufactured using the SON substrate of this embodiment. When the mobility of this MOS transistor was measured, an increase of 35% was observed compared to that produced on a conventional ordinary bulk substrate. Furthermore, the mobility was higher than that of the MOS transistor formed on the SOI substrate or the MOS transistor formed on the SON substrate in which the inner surface of the cavity was not oxidized.

  The reason why the mobility of the MOS transistor formed on the SON substrate of this embodiment is higher than that of the MOS transistor formed on the conventional SOI substrate is that the cavity 5 exists in the substrate, so that the mobility is higher than that of the conventional SOI substrate. Further, it is considered that this is due to a synergistic effect that the parasitic capacitance can be further reduced and that the silicon oxide film 32 can realize a state having a high stress in the SON layer.

  In this embodiment, in order to oxidize the inside of the cavity 5, the trench 10 is formed after the cavity 5 is formed. However, the method shown in FIG. 25 is also possible. In this method, first, as shown in FIG. 22A, a plurality of trenches 4 and one trench 10 having a larger opening diameter and deeper than that are simultaneously formed. Thereafter, heat treatment is performed to change the plurality of trenches 4 into cavities. However, as shown in FIG. 25B, since the upper portion of the large trench 10 is not blocked, a cavity having an opening structure as shown in FIG. 23C is formed. The subsequent steps are the same as those after FIG. The layout of the trenches 4 and 10 is not limited to that shown in FIG. 22A, and various layouts can be adopted.

  In the present embodiment, in order to selectively form the silicon oxide film 32 only on the inner surfaces of the cavity 5 and the trench 10, the silicon oxide film 32 is formed on the entire surface including the substrate surface, and then the silicon oxide film on the substrate surface is oxidized. Although the film 32 has been selectively removed, it may be as follows. That is, after selectively forming an antioxidant film such as a silicon nitride film on the surface of the substrate, only the inner surface of the cavity may be oxidized by an oxidation process.

  In this embodiment, the silicon oxide film 32 is formed inside the cavity 5 or the like in order to generate a tensile stress in the SON layer. However, another film may be formed. That is, any film (different material film) formed of a material having a different thermal expansion coefficient from that of single crystal silicon can be used. Further, even a dissimilar material film formed of a material whose thermal expansion coefficient is not significantly different from that of single crystal silicon can be used as long as distortion can be generated on the semiconductor film side. If the above conditions are satisfied, the film (stress generating film) formed inside the cavity 5 may be an insulating film or a metal film.

  Furthermore, in this embodiment, the case where the thickness of the SON layer 33 and the silicon oxide film 32 is substantially the same has been described. In order to increase the amount of distortion generated in the SON layer 33 by the silicon oxide film 32, the ratio of the thickness of the silicon oxide film 32 to the thickness of the SON layer 33 is better. However, if this ratio is too large, a problem arises in terms of substrate strength.

  From various experiments by the present inventors, the relationship between the thickness of the semiconductor layer such as the SON layer 33 and the thickness of the dissimilar material film such as a silicon oxide film is as follows: (thickness of semiconductor layer) / (thickness of semiconductor layer) It has become clear that the ratio of (the thickness of the different material film) may be in the range of 0.1 to 0.9.

  In this embodiment, the silicon oxide film 32 is formed on the entire inner wall of the cavity. However, if a tensile stress can be generated in the SON layer 33, the stress of the silicon oxide film 32 or the like on a part of the cavity. A generation film may be formed.

(Ninth embodiment)
In the present embodiment, an SON substrate is described which has the same effect as a silicon substrate (strained substrate) in which a SiGe layer or the like is buried directly under a channel and can solve the problems of the strained substrate.

  FIG. 26 is a cross-sectional view showing a SON substrate forming method according to the ninth embodiment of the present invention.

  First, as shown in FIG. 26A, a plurality of trenches 4 are formed on the surface of the silicon substrate 1 by using a well-known lithography method and RIE method.

  Next, as shown in FIG. 26B, a 100 nm thick SiGe layer 41 containing 30% of Ge by atomic density ratio is epitaxially grown on the entire surface so as to cover the inner surface of the trench 4.

Next, as shown in FIG. 26 (c), the surface of the silicon substrate 1 is made to flow by heat treatment at 1050 ° C. for 5 minutes in a vacuum at a pressure of 10 ⁻⁷ Pa. The cavity 5 in which the layer (buried SiGe layer) 41a exists is formed. At this time, a SiGe layer (residential SiGe layer) 41 b is also formed on the surface of the silicon substrate 1.

  Next, a silicon oxide film (not shown) is formed on the substrate surface by thermal oxidation to increase the Ge concentration in the embedded SiGe layer 41a, and then the silicon oxide film and the resident SiGe layer 41b are removed. Thereby, the Ge composition ratio of the embedded SiGe layer 41a can be increased.

  Finally, as shown in FIG. 26D, a silicon layer 42 not containing Ge is epitaxially grown on the surface of the silicon substrate 1 to complete the SON substrate.

  When the stress of the silicon substrate 1 on the cavity 5 of the SON substrate thus obtained and the silicon layer 42 thereon was measured, the value was 80 MPa. From this result, it was found that forming the embedded SiGe layer 41a inside the substrate is effective as a method for intentionally generating stress in the SON layer.

  In this embodiment, the SiGe layer 41 is epitaxially grown after the trench 4 is formed. However, the trench 4 may be formed after the SiGe layer 41 is epitaxially grown on the entire surface of the substrate. In this case, after the trench 4 is formed, the substrate surface is flowed by heat treatment to form the cavity 5 and the embedded SiGe layer 41a.

  In addition, forming a silicon oxide film by thermal oxidation after flowing on the substrate surface is an effective method for increasing the Ge composition ratio of the embedded SiGe layer 41a, but it is not always necessary.

  Further, forming the Si layer 42 by epitaxial growth after flowing on the surface of the substrate is an effective method for forming a SON layer not containing Ge. However, if it is not necessary for device application, the Si layer 42 is formed. unnecessary.

  The SON substrate of this embodiment has the following advantages over the substrate having the conventional SiGe layer 41c shown in FIG.

  In the prior art, in order to form the SiGe layer 41c with few defects and a high Ge composition ratio on the silicon substrate 1, the Si composition of the SiGe layer 41c is increased from a low concentration state using the silicon substrate 41 as a seed. The method of changing continuously in the film thickness direction until the state was taken. Therefore, the thickness of the SiGe layer 41c is about several hundred nm. That is, the SiGe layer 42 needs to be formed thick.

  On the other hand, in this embodiment, since the SiGe layer 41a corresponding to the conventional SiGe layer 41c is formed by surface migration of Si and SiGe (FIG. 26 (c)), the SiGe layer 41a on the cavity 5 is formed on the SiGe layer 41a. Will not cause any defects. Therefore, it is not necessary to form the SiGe layer 41a thick, and the thickness can be reduced to several tens of nm. This is shown in FIG. Many defects are generated in the silicon substrate 1 and the silicon layer 42 in the region 43 where the cavity 5 is not formed below, and the defect density increases. On the other hand, substantially no defects are generated in the silicon substrate 1 and the silicon layer 42 in the region 44 in which the cavity 5 is formed under the device forming region, and the defect density is sufficiently low.

  In the present embodiment, SiGe is used as the material of the dissimilar material film (SiGe layer 41a), but other materials different from the substrate material (Si) can be used as in the eighth embodiment.

  Further, as in the eighth embodiment, the relationship between the thickness of the semiconductor layer such as the Si layer 42 and the thickness of the dissimilar material film such as the SiGe layer 41 is (thickness of the semiconductor layer) / (thickness of the semiconductor layer). It was confirmed that the effect of the present invention was realized when the ratio of the thickness + thickness of the different material film was in the range of 0.1 to 0.9. Furthermore, if a tensile stress can be generated in the SON layer, the SiGe layer 41 may be formed in a part of the cavity.

(Tenth embodiment)
In this embodiment, an example in which the ESS technique of the present invention is applied to the production of a photonic crystal will be described.

  A photonic crystal can be formed by periodically forming materials having different refractive indexes. Photonic crystals are attracting attention as new optical materials for realizing ultra-small optical integrated circuits.

  In addition, since the photonic crystal can be formed on silicon, it is possible to avoid the mounting problems so far and to realize a future optoelectronic integrated circuit fused with the CMOS process.

  Up to now, many methods for producing photonic crystals have been proposed, but in particular, the method for producing three-dimensional photonic crystals has been difficult. Further, a combination of materials having a large difference in refractive index is desirable. For example, a combination of silicon and air is ideal, but its formation method is very difficult.

  FIG. 29 is a schematic diagram of a three-dimensional periodic structure (photonic crystal) according to a tenth embodiment of the present invention that can solve the above problem. In the figure, reference numeral 51 denotes a silicon substrate. Within this silicon substrate 51, spherical cavities 52 of the same size (colors are shown in dark order in the depth direction) are periodically arranged three-dimensionally. ing.

  Next, a manufacturing method of the three-dimensional periodic structure according to the present embodiment will be described with reference to FIG.

  First, as shown in FIGS. 30A to 30C, a mask pattern (not shown) made of an oxide film or the like is formed on the silicon substrate 51, and this mask pattern is used as a mask by a reactive ion etching method. The silicon substrate 51 is etched to form two-dimensionally array of trenches 52 having the same depth and the same opening diameter, and then the mask pattern is removed.

  Next, as shown in FIGS. 30D to 30F, the silicon substrate 51 on which the trench 52 is formed is subjected to a heat treatment under a high temperature and reduced pressure in a non-oxidizing atmosphere to thereby obtain the silicon substrate 51. A plurality of spherical cavities (ESS) 53 of uniform size are formed in a cavity pattern. Specifically, a cavity pattern is formed in which cavities are arranged at equal intervals on the same line in the depth direction of the substrate, and cavities are arranged in a lattice pattern in the same plane in the substrate.

  The heat treatment for forming the cavities 53 is for causing surface migration of silicon. Therefore, it is desirable to completely remove the natural oxide film on the substrate surface before the heat treatment. In order to sufficiently remove the natural oxide film, it is effective to keep the heat treatment atmosphere non-oxidizing. In order to realize this easily, it is desirable that the atmosphere of the heat treatment is an atmosphere of 100% hydrogen, for example. In order to promote the surface migration of silicon, it is desirable to perform heat treatment at a pressure of 10 Torr or less. Typical heat treatment conditions include a 100% hydrogen atmosphere, a temperature of 1100 ° C., a pressure of 10 Torr, and a time of 10 min.

  Although the case where the heat treatment is performed after removing the mask pattern is shown here, the heat treatment may be performed without removing the mask pattern. However, in this case, it is necessary to remove the mask pattern after the heat treatment and perform the heat treatment again to planarize the substrate surface.

  The three-dimensional periodic structure according to the present embodiment is a photonic crystal having a forbidden band with respect to light since materials (silicon / cavity or air) having different refractive indexes are periodically arranged. The wavelength dependence which is one of the characteristics of the photonic crystal is scaled by all (period / wavelength of the cavity 5). Therefore, a photonic crystal that operates at a desired wavelength can be produced by setting the period of the cavity 5 according to the wavelength used.

  A specific method for controlling the period of the cavity 5 is to change the size and depth of the diameter of the trench 52 with respect to the period in the depth direction. On the other hand, regarding the period in the direction perpendicular to the depth direction, the arrangement period of the trenches 52 can be changed.

  As described above, according to the present embodiment, by using the surface migration of silicon, a three-dimensional periodic structure composed of a combination of materials having a large refractive index difference (silicon: 3.6 / air: 1) is obtained. It can be easily realized. This three-dimensional periodic structure operates as a photonic crystal capable of controlling light. Therefore, the three-dimensional periodic structure of the present embodiment can be operated as an optical element such as an optical waveguide, a polarizer, or a prism.

  Furthermore, according to the above method, the period of the cavity 5 can be reduced to about 1 μm or less. That is, a fine optical element can be formed in a silicon substrate. Thereby, an optoelectronic circuit in which an optical element and a CMOS process are fused can be easily manufactured.

(Eleventh embodiment)
FIG. 31 is a schematic diagram of a three-dimensional periodic structure (photonic crystal) according to an eleventh embodiment of the present invention. This embodiment is different from the tenth embodiment in that cavities 53 s and cavities 53 l having different sizes (diameters) are periodically arranged in the silicon substrate 51.

  Specifically, with respect to the depth direction of the substrate, a plurality of spherical cavities 53s or cavities 53l having the same size (colors are shown in order deeper in the depth direction) are arranged at equal intervals on the same line. In the same plane, cavities 53s and cavities 53l having different sizes are arranged in a lattice pattern.

  Next, the manufacturing method of the three-dimensional periodic structure of this embodiment is demonstrated using FIG.

  First, as shown in FIGS. 32A to 32C, a mask pattern (not shown) made of an oxide film or the like is formed on the silicon substrate 51, and this mask pattern is used as a mask by reactive ion etching. The silicon substrate 51 is etched to form trenches 52s and trenches 52l having the same depth and different opening diameters in a grid pattern. Thereafter, the mask pattern is removed.

  Next, as shown in FIGS. 32D to 32F, the silicon substrate 51 in which the trenches 52s and the trenches 52l are formed is subjected to heat treatment under high temperature and reduced pressure in a non-oxidizing atmosphere. Spherical cavities 53s or cavities 53l having the same size are periodically arranged in the depth direction in the silicon substrate 51, and cavities 53s and cavities 53l having different sizes are alternately and periodically arranged in a direction perpendicular to the depth direction. Cavity patterns arranged in the above are formed. Note that as described in the tenth embodiment, heat treatment may be performed without removing the mask pattern.

  The silicon substrate 51 having the cavity pattern thus obtained can be regarded as a photonic crystal capable of controlling light as in the tenth embodiment, and can be operated as an optical element.

  Also in this embodiment, the cavity period, that is, the operating wavelength can be controlled by the same method as in the tenth embodiment. Furthermore, according to the embodiment, since the cavities 52s and 52l having different sizes are used, the operating wavelength can be controlled in a wider range by utilizing the difference in size.

  In the tenth and eleventh embodiments, when the cavities 52, 52s, and 52l are formed by heat treatment in an atmosphere containing hydrogen, hydrogen remains in these. Furthermore, according to the research by the present inventors, it has been confirmed that the cavities 52, 52s, and 52l are formed of polygons with rounded corners. More precisely, it is composed of a polyhedron having a predetermined plane orientation.

  Furthermore, the following was found by examining the angle formed by the plane orientation of the faces constituting the polyhedron with the (100) plane which is the main surface of the silicon substrate. That is, the faces constituting the polyhedron are composed of {100} plane group, {110} plane group, {111} plane group, {311} plane group, {531} plane group, and {541} plane group. Became clear. Since these surface groups have low surface energy, it can be said that the cavity is thermally stable.

(Twelfth embodiment)
Here, an embodiment in which the ESS technology of the present invention is applied to an optical integrated circuit, particularly an embodiment in which the ESS technology is applied to an optical waveguide will be described.

In the optical integrated circuit technology, optical elements such as optical passive elements and light emitting elements are formed on a semiconductor substrate such as a Si substrate or a GaAs substrate, and an optical waveguide is formed mainly of quartz (SiO ₂ ) separately from the optical elements. Is done. Therefore, in the connection part between the optical waveguide and the optical element, it is inevitably necessary to propagate light into the semiconductor region.

One method for propagating light in the semiconductor region is to use the fact that Si has a higher refractive index than SiO ₂ . In this method, the diameter of the optical waveguide made of Si is set to about 5 μm or less, which is several times the wavelength of the light, and the light is totally reflected at the interface between the optical waveguide and the surrounding Si region (Si / SiO ₂ interface). By doing so, light is confined in the Si region.

In an optical waveguide mainly composed of Si, the lower the refractive index of the surrounding material relative to Si, the better, in order to increase the confinement property. The refractive index of Si is 3.4, whereas the refractive index of SiO ₂ is 1.5.

A medium having a lower refractive index than SiO ₂ is naturally a vacuum (refractive index = 1). In reality, air is used as a medium instead of a vacuum. For example, an SOI substrate can be used as a method of making the surroundings of the Si region used as an optical waveguide air, but it is difficult to realize this.

The reason is that by etching the Si region of the SOI substrate, a pattern having a top surface and a side surface where Si is exposed can be easily formed, but the SiO ₂ region of the SOI substrate is etched to This is because it is difficult to selectively remove only the SiO ₂ region.

  FIG. 33 is a perspective view showing an optical waveguide according to the twelfth embodiment of the present invention. In the figure, reference numeral 61 denotes a single crystal silicon substrate having a (100) plane orientation. On this silicon substrate 61, an Si pattern 62 having air around the top, side and bottom surfaces is formed.

  The Si pattern 62 and the surrounding air constitute an optical waveguide. For example, a wavelength of 1.4 μm propagates in the optical waveguide. In an actual optical circuit, one end of the Si pattern 62 is connected to a light emitting part of an optical functional element (not shown), and the other end is connected to a light receiving part of an optical functional element (not shown).

  Such an optical waveguide can be easily formed by using the ESS technology described so far. First, a plurality of trenches are formed on the surface of the silicon substrate 61 using a known lithography method and RIE method. Next, surface migration of silicon is caused by high-temperature heat treatment in a reducing atmosphere to form a large area cavity (ESS) in the silicon substrate 61. Then, using a known lithography method and RIE method, a portion of the silicon region (SON layer) on the cavity of the silicon substrate that is not used as the Si pattern 62 is selectively removed.

FIG. 34 is a perspective view of an optical waveguide using a conventional SOI substrate. In the figure, 61 is a silicon substrate, 63 is a SiO ₂ layer, and 64 is a Si pattern formed by processing a silicon substrate. Unlike the Si pattern 62 of the present invention, the periphery of the upper surface and side surface of the conventional Si pattern 64 is air, but the bottom surface is the SiO ₂ layer 63 unlike the Si pattern 62 of the present invention. The refractive index of SiO ₂ (= 1.5) is larger than the refractive index of air (= 1.0).

  For this reason, the optical waveguide of the present invention shown in FIG. 33 has much less light leaking to the outside than the conventional optical waveguide shown in FIG. 34, and has excellent characteristics as an optical waveguide (optical confinement characteristics). It can be said that it has something.

  As described above, according to the present embodiment, an optical waveguide having a good optical confinement characteristic can be realized, and as a result, an optical integrated circuit with little optical loss can be realized.

(13th Embodiment)
Passive elements such as inductors and capacitors are formed on a semiconductor substrate in the same manner as active elements such as transistors. The parasitic capacitance and parasitic resistance (eddy-current loss) between the passive element and the semiconductor substrate are large.

  For this reason, conventional inductors and capacitors have the following problems when the frequency of a signal flowing through them becomes 1 GHz or higher. That is, the Q value is lowered for the inductor, and it is difficult to obtain a highly accurate capacitance for the capacitor.

  In order to solve the above problem, the present invention uses a silicon substrate having a flat cavity as a semiconductor substrate, and forms a passive element on the silicon substrate on the flat cavity. With such a configuration, the parasitic capacitance and parasitic resistance between the passive element and the semiconductor substrate can be effectively reduced, and the above-described problems can be solved.

  FIG. 35 shows a plan view and a cross-sectional view of a semiconductor device having an inductor to which the present invention is applied. FIG. 36 shows a cross-sectional view of a semiconductor device having an MIM capacitor to which the present invention is applied. In the figure, 70 is a silicon substrate, 71 is a flat cavity (ESS), 72 is a spiral inductor, 73 is a metal electrode, 74 is an insulating film, and 75 is a metal electrode. Both the inductor and the capacitor may be formed on the silicon substrate 70.

  As a method of forming the silicon substrate 70 having the flat cavity 71, any of the formation methods of the above-described embodiments may be used. After such a silicon substrate 70 is formed, a passive element such as an inductor, an active element such as a transistor, and a wiring layer are formed as usual. The reason for forming a passive element or the like after the formation of the cavity 71 is that the formation of the cavity 71 requires heat treatment at a high temperature.

(Fourteenth embodiment)
In recent years, in the field of semiconductors, devices and modules have been increased in density and functionality. With such higher density and higher functionality, the amount of heat generated by devices and the like has increased, and heat dissipation has become very difficult.

  As one of conventional heat dissipation methods, a method is known in which a heat dissipation fin is attached to a device or a package, heat from the device or the like is transmitted to the fin by heat conduction, and heat is released into the air by heat conduction from the fin. However, when the amount of heat generation increases as described above, a sufficient heat dissipation effect cannot be obtained. Therefore, in recent years, heat radiation by downsizing of the entire device and forced air cooling (fan) has become mainstream. However, it is still difficult to obtain the necessary heat dissipation effect.

  In mainframes such as supercomputers, cooling with a refrigerant such as liquid nitrogen or chlorofluorocarbon is the mainstream. It is conceivable to apply this cooling method to a semiconductor device or the like. However, problems such as corrosion of terminals and wirings occur due to impurities present in the refrigerant.

  In order to solve the above problem, the present invention uses a silicon substrate including a plurality of cooling pipes for flowing a coolant as a semiconductor substrate. With such a configuration, by flowing a coolant through the cooling pipe, the silicon substrate can be effectively cooled even if the amount of heat generated by a device or the like due to higher density and higher functionality is increased. Can be solved. Furthermore, since the refrigerant flows through the inside of the substrate where terminals and the like do not exist, the problem of corrosion does not occur.

  FIG. 37 shows a perspective view of a silicon substrate having a cooling pipe (cooling structure) according to a fourteenth embodiment of the present invention. In the figure, reference numeral 81 denotes a silicon substrate, and 82 denotes a cooling pipe. When cooling the silicon substrate, a refrigerant supply mechanism (not shown) is prepared.

  Next, a method of manufacturing a semiconductor device using a silicon substrate having a cooling pipe according to the present embodiment will be described with reference to FIG.

  First, a Si wafer 83 is prepared. In the figure, 84 indicates a scribe line.

  Next, a plurality of flat cavities (hollow structures) 85 are formed so as to be orthogonal to the scribe line 84 using the ESS technique of the present invention. As a method of forming the flat cavity 85, any method of the above-described embodiment may be used. Preferably, a plurality of trench patterns are designed so that a cylindrical cavity 85 is formed.

  Thereafter, necessary elements, wirings, and the like are formed on the silicon region on the cavity 85 of the Si wafer according to a known method, and a plurality of semiconductor devices (not shown) having a desired function are formed on the Si wafer 83.

  Finally, the Si wafer is cut along a scribe line 84 by a known method, and a plurality of chips are taken out from one Si wafer 83. At this time, since the cavity 85 is cut, the cooling pipe is completed at the same time.

(Fifteenth embodiment)
In the present embodiment, a technique for preventing the flat ESS from being crushed, which is different from the sixth and seventh embodiments, will be described. The essence of this embodiment is to form Si pillars for preventing crushing inside the hollow region. Such Si pillar can be formed by the following method.

  First, a mask material made of an oxide film or the like is formed on a silicon substrate, and a photoresist pattern is formed thereon. The same mask material as that described in the first embodiment can be used.

  Next, using the photoresist pattern as a mask, the mask material is patterned by anisotropic etching, for example, RIE, and the pattern of the photoresist pattern is transferred to the mask material.

  Next, after carbonizing and peeling the photoresist pattern, the silicon substrate is patterned by anisotropic etching, for example, RIE, using the patterned mask material as a mask, and a plurality of trenches are two-dimensionally formed on the surface of the silicon substrate. To do. Here, as shown in FIG. 39A, the trench 4 is not formed in the region where the Si pillar is formed.

  Although the figure shows an example in which one trench is removed, a plurality of trenches may be removed. The size of the Si pillar can be changed depending on the number of trenches to be removed.

  Finally, after removing the mask material 2, high-temperature annealing is performed in a reducing atmosphere under reduced pressure, so that one flat cavity 5 is formed inside the silicon substrate 1 as shown in FIG. 39B. And two Si pillars 1p are formed inside the cavity 5.

  Next, an arrangement of Si pillars effective for preventing ESS from collapsing will be described. The Si pillar is provided to prevent the cavity 5 from being crushed due to a pressure difference between the external pressure of the cavity 5 and the internal pressure of the cavity 5 at the time of or after the formation of the cavity 5.

  Therefore, the relationship between the thickness t (= 0.1 μm, 1 μm) of the silicon substrate (hereinafter referred to as the silicon layer) on the cavity 5 and the deflection amount δ of the silicon layer was examined. The result is shown in FIG. From the figure, it can be seen that the amount of deflection δ is larger when the silicon layer is thin, regardless of the width of the cavity.

  In order to reduce the bending amount δ, for example, when the thickness of the silicon layer is 0.1 μm, the width W of the cavity 5 may be set to 5 μm or less. In this case, the amount of deflection δ is a size without a problem of 0.02 μm or less.

  In order to estimate the distance more accurately, it was investigated how far the Si pillars should be arranged with respect to the thickness of the silicon layer by using a deflection calculation formula of the silicon layer. If the deflection amount δ is less than half of the thickness of the silicon layer, it will not be greatly affected. Therefore, the Si pillars are arranged so as to satisfy the following inequality (3) regarding the width w of the thickness of the silicon layer. It was found that the ESS can be formed without any problems.

w ≦ t (E / 0.0568P) ^{1/4 (3)}
Here, E represents the Young's modulus of silicon (= 0.13 (N / μm ² )), and P represents the load (pressure) applied to the silicon layer (N / μm ² ).

  When the thickness of the silicon layer is 0.1 μm, the distance between the Si pillars necessary for preventing the collapse of the ESS is calculated based on the formula (1), and is 6.9 μm or less.

  As described above, even when the silicon layer is thin, the collapse of the cavity 5 due to the pressure difference between the external pressure of the cavity 5 and the internal pressure of the cavity 5 is effectively suppressed by forming the Si pillar 1p in the cavity 5. become able to. As a result, a SON substrate having a cavity 5 with a larger area can be realized. Furthermore, the degree of freedom in designing the SON substrate is increased.

  As shown in FIG. 41, the present inventors have estimated the amount of deflection of the silicon layer with respect to the SON substrate having a flat cavity 5 having a circular shape as viewed from above.

  In this case, the maximum deflection occurs at the center of the circle, and the deflection amount δ of the silicon layer is given by the following equation (4).

δ = 0.0108 Pa ⁴ / (Et ³ ) (4)
Here, a represents the diameter (μm), and t represents the thickness of the silicon layer (μm).

  The amount of bending of the silicon layer of the SON substrate shown in FIG. 41 will be compared with that of a SON substrate having a flat plate-like cavity whose shape viewed from above is rectangular.

  The maximum amount of deflection in the case of a disc having the same diameter as the short side of the rectangle is 3/8 times the maximum amount of deflection in the case of a rectangle. That is, in the case of a circular shape, if the diameter is increased by 1.27 times, the same amount of bending as in the case of a rectangular shape occurs. However, in the case of a rectangle, even if the length of the long side is increased, the maximum amount of bending does not increase, so that a cavity with a larger area can be formed in the rectangle.

(Sixteenth embodiment)
FIG. 42 is a diagram showing a pressure sensor according to a sixteenth embodiment of the present invention.

In the figure, 91 is n-type SON substrate main surface is {100}, 92 rectangular cavity in the n-type SON substrate 91, 93 ₁ to 93 ₄ are formed on the substrate surface on the peripheral portion of the cavity 92 , A p-type diffusion layer as a gauge resistor constituting a bridge circuit, 94 a high impurity concentration p ⁺ -type diffusion layer formed on the substrate surface as a wiring, and 95 a metal wiring made of a metal such as Al. Yes. The metal wiring 95 is connected to the p ⁺ -type diffusion layer 94 through a connection hole opened in an insulating film (not shown) formed on the n-type SON substrate 91.

The pressure sensor of the present embodiment is a diaphragm type semiconductor pressure sensor that utilizes the bending of the SON substrate 91 (silicon layer) on the cavity 92 due to the pressure difference between the external pressure of the cavity 92 and the internal pressure of the cavity 92. is there. When the silicon layer is deflected by a pressure difference, the value of the resistor (gauge resistors) of p-type diffusion layer 93 _{1-93 4} is changed by the piezoelectric resistance effect. This change in resistance value can be detected as an electrical signal by a bridge circuit. This makes it possible to measure the pressure applied to the silicon layer.

  Since the cavity 92 is a vacuum, the measured pressure is an absolute pressure. When the pressure applied to the silicon layer is measured with reference to the atmospheric pressure, an opening 96 connected to the cavity 92 may be provided on the back surface of the n-type SON substrate 91 as shown in FIG.

  The degree of deflection of the silicon layer can be varied depending on the thickness of the silicon layer and its size. Therefore, the pressure range that can be measured by the pressure sensor of the present embodiment can be controlled by the thickness of the silicon layer and its size. Therefore, a pressure sensor capable of measuring a desired pressure range can be realized by appropriately selecting the thickness and size of the silicon layer.

FIG. 44 shows a pressure sensor according to a modification. This pressure sensor is manufactured using an n-type substrate 91 having a main surface of {110}. The SON substrate having a {100} principal surface and the SON substrate having a {110} principal surface have different amounts of change in resistance due to the piezoresistive effect even if the amount of silicon deflection is the same. The pressure sensor shown in FIG. 43, as the sensitivity (variation in resistance due to the piezoresistive effect) increases, but chose a pattern of p-type diffusion layer 93 _{1-93 4.} FIG. 45 shows a pressure sensor corresponding to FIG.

  The present invention is not limited to the above embodiment. For example, although the case where a silicon substrate is used has been described in the above embodiment, the present invention is also effective for other semiconductor substrates such as a silicon germanium substrate. That is, according to the present invention, it is possible to provide an inexpensive and highly reliable SOI (Semiconductor On Insulator) structure that is not limited to silicon.

  Further, in the above-described embodiment, the plurality of trenches 2 that are two-dimensionally arranged are changed to one flat cavity by heat treatment. It can also be obtained by changing the trench into one flat cavity by heat treatment.

  Further, by introducing a Cu wiring in addition to the SOI structure of the present invention, further higher speed and power saving can be realized.

  Furthermore, in the above-described embodiment, the case where the initial trench 4 is formed as a straight trench having the same size in the depth direction is shown, but a bottle shape trench having a constriction in the depth direction may be formed. That is, you may form the trench characterized by the plane which has the minimum cross-sectional area with respect to the depth direction of a trench not being the bottom of a trench. Even if the trench having such a shape is formed, a flat cavity can be effectively formed as in the case where the trench 4 is used.

  Furthermore, the above embodiments include inventions at various stages, and various inventions can be extracted by appropriately combining a plurality of disclosed constituent elements. For example, even if some constituent requirements are deleted from all the constituent requirements shown in the embodiment, if the problem to be solved by the invention can be solved, the configuration from which the constituent requirements are deleted can be extracted as an invention. . In addition, various modifications can be made without departing from the scope of the present invention.

Process sectional drawing which shows the formation method of the flat cavity which concerns on the 1st Embodiment of this invention Sectional drawing for demonstrating the example in which the shape change from several groove | channels to one flat cavity does not occur FIG. 1 is a plan view of a groove layout example shown in FIG. 1 and a flat plate-like cavity formed therefrom. Sectional drawing which shows the example which applied this invention to DRAM / LOGIC mixed mounting Process sectional drawing which shows the first half of the manufacturing method of the MOS transistor which concerns on the 2nd Embodiment of this invention Process sectional drawing which shows the middle half of the manufacturing method of the MOS transistor which concerns on the 2nd Embodiment of this invention Process sectional drawing which shows the second half of the manufacturing method of the MOS transistor which concerns on the 2nd Embodiment of this invention Process sectional drawing which shows the manufacturing method of the MOS transistor which concerns on the 3rd Embodiment of this invention Sectional drawing which shows the MOS transistor which concerns on the 3rd Embodiment of this invention Process sectional drawing which shows the first half of the manufacturing method of the MOS transistor which concerns on the 4th Embodiment of this invention Process sectional drawing which shows the second half of the manufacturing method of the MOS transistor which concerns on the 4th Embodiment of this invention FIG. 10 is a plan view of a groove layout example shown in FIG. 10 and a flat plate-like cavity formed therefrom. Sectional drawing which shows the MOS transistor which concerns on the 4th Embodiment of this invention Sectional drawing for demonstrating the point which should improve the formation method of the SON substrate demonstrated in the 1st-4th Example. Sectional drawing which shows the formation method of the SON substrate concerning the 5th Example of this invention Diagram for explaining the relationship between the initial trench shape and the resulting cavity Illustration to explain the number of cavities obtained for the initial trench shape Photomicrograph showing that flat ESS collapses as ESS width increases ESS structure model used to calculate the strength of the ESS structure The figure which shows the relationship between the plate width calculated about the ESS structure from which the thickness of a silicon layer differs (0.1 micrometer, 1 micrometer), and bending The figure which shows the sequence of the heat processing effective in forming ESS of a large area The figure which shows the relationship between the thickness of the SON layer calculated | required by calculation, and the amount of bending Sectional drawing which shows the formation method of the SON board | substrate concerning the 8th Example of this invention. Cross-sectional view of MOS transistor fabricated using SON substrate Sectional drawing for demonstrating the modification of the formation method of the SON board | substrate of 8th Example. Sectional drawing which shows the formation method of the SON substrate based on 9th Example of this invention Sectional view showing a substrate having a conventional SiGe layer Sectional drawing which shows the SON board | substrate concerning 9th Example of this invention. Schematic diagram of a three-dimensional periodic structure according to a tenth embodiment of the present invention Sectional drawing for demonstrating the manufacturing method of the three-dimensional periodic structure of FIG. Schematic diagram of a three-dimensional periodic structure according to an eleventh embodiment of the present invention Sectional drawing for demonstrating the manufacturing method of the three-dimensional periodic structure of FIG. The perspective view which shows the optical waveguide based on the 12th Example of this invention. A perspective view showing a conventional optical waveguide The top view and sectional drawing of a semiconductor device which have an inductor concerning a 13th example of the present invention Sectional drawing of the semiconductor device which has a capacitor based on 13th Example of this invention 14 is a perspective view of a silicon substrate having a cooling pipe according to a fourteenth embodiment of the present invention. FIG. 14 is a plan view of a silicon substrate having a cooling pipe according to a fourteenth embodiment. Sectional drawing which shows the formation method of the SON substrate based on 15th Example of this invention. The figure which shows the result of having investigated the relationship between the thickness and the deflection | deviation amount (delta) about the silicon substrate on ESS The figure which shows the SON board | substrate which has the flat-plate-shaped cavity whose shape seen from the top is circular The figure which shows the pressure sensor based on the 16th Example of this invention. The figure which shows the modification of the pressure sensor which concerns on the Example The figure which shows the other modification of the pressure sensor which concerns on the same Example The figure which shows another modification of the pressure sensor which concerns on the same Example

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Silicon substrate, 2 ... Mask material, 3 ... Photoresist pattern, 4 ... Groove (1st groove | channel), 5 ... Flat plate-shaped cavity, 5 '... Unclosed cavity, 6 ... Spherical cavity, 7 ... Silicon Oxide film, 8 ... Silicon nitride film, 9 ... Photoresist pattern, 10 ... Groove (second groove), 11 ... Silicon thermal oxide film, 12 ... Silicon oxide film, 13 ... Photoresist pattern, 14 ... Element isolation groove, DESCRIPTION OF SYMBOLS 15 ... Silicon thermal oxide film, 16 ... Silicon oxide film, 17 ... Gate oxide film, 18 ... Gate electrode, 19, 20 ... Extension, 21 ... Gate side wall insulating film, 22 ... Source diffusion layer 23 ... Drain layer, 24 ... Silicon Thermal oxide film, 25 ... Silicon nitride film, 26, 27 ... Photoresist pattern, 28 ... Groove (third groove), 29 ... Silicon oxide film, 30 ... Silicon nitride film, 31 ... Photoresist pattern 32 ... silicon oxide film, 33 ... SON layer, 41 ... SiGe layer, 42 ... silicon layer, 43 ... Si region where no cavity 5 is formed below, 44 ... Si where cavity 5 is formed below Region 51 ... silicon substrate 52 ... trench 53 ... spherical cavity 61 ... silicon substrate 62 ... Si pattern 63 ... SiO ₂ layer 64 ... Si pattern 70 ... silicon substrate 71 ... flat plate cavity 72 ... spiral inductor, 73 ... metal electrode, 74 ... insulating film, 75 ... metal electrode, 81 ... silicon substrate, 82 ... cooling pipe, 83 ... Si wafer, 84 ... scribe line, 85 ... flat cavity (hollow structure) 91 ... SON substrate, 92 ... rectangular cavity, 93 _{1 to} 93 ₄ ... p-type diffusion layer (gauge resistance), 94 ... p ⁺ -type diffusion layer (wiring), 95 ... metal wiring, 96 ... opening.

Claims (16)

Forming a plurality of first trenches on a surface of a semiconductor substrate;
And a step of changing the plurality of first trenches into one flat cavity by performing a heat treatment on the semiconductor substrate. Forming a second trench reaching the flat cavity on the surface of the semiconductor substrate after forming the flat cavity;
The method for manufacturing a semiconductor substrate according to claim 1, further comprising: filling the second trench and the flat cavity with an insulating film. 2. The method of manufacturing a semiconductor substrate according to claim 1, wherein after forming the flat cavity, an oxide film is formed on an inner surface of the flat cavity by thermal oxidation. 3. 3. The method of manufacturing a semiconductor substrate according to claim 2, wherein after forming the second trench, an oxide film is formed on the inner surface of the flat plate-like cavity by thermal oxidation. Forming a plurality of first trenches on the surface of the semiconductor substrate and forming a third trench having a wider opening than the first trench;
By applying heat treatment to the semiconductor substrate, the plurality of first trenches and the third trench are not closed 1 having a flat space region and having an opening surface on the surface of the semiconductor substrate. The process of turning it into two cavities,
And filling the inside of the cavity with an insulating film. When the distance between the first trenches is D, and the radius of a circle having the same area as the opening surface of the first trench is R, the plurality of first trenches is such that D <4R. The method of manufacturing a semiconductor substrate according to claim 1, wherein the semiconductor substrate is formed in an array. 6. The method of manufacturing a semiconductor substrate according to claim 1, wherein an aspect ratio of the first trench is 2.5 or more. Forming a plurality of first trenches having an aspect ratio of 5 or more and a plurality of fourth trenches having an aspect ratio of 4 or less on the surface of the semiconductor substrate;
By applying heat treatment to the semiconductor substrate, the plurality of first trenches are changed into one cavity, and the plurality of fourth trenches are extinguished, and the fourth trench and the cavity are formed. And a step of flattening the surface of the semiconductor substrate comprising: a method of manufacturing a semiconductor substrate. Forming a plurality of trenches on the surface of the semiconductor substrate;
Changing the plurality of first trenches into one cavity by applying a first heat treatment to the semiconductor substrate;
Applying a second heat treatment to the semiconductor substrate, and changing a pressure inside the cavity to reduce a difference between an atmospheric pressure in which the semiconductor substrate exists and a pressure inside the cavity. A method of manufacturing a semiconductor substrate. 10. The method of manufacturing a semiconductor substrate according to claim 9, wherein the first heat treatment is performed at a high temperature and a reduced pressure, and the second heat treatment is performed at a low temperature and a high pressure. The method for manufacturing a semiconductor substrate according to claim 9, wherein the first heat treatment is performed at a high temperature of 1100 ° C. or higher. The method for manufacturing a semiconductor substrate according to claim 9, wherein the second heat treatment is performed in an atmosphere having a hydrogen concentration of 100%. The method for manufacturing a semiconductor substrate according to claim 9, wherein the second heat treatment is performed under a high pressure equal to or higher than atmospheric pressure. The method for manufacturing a semiconductor substrate according to claim 9, wherein the first heat treatment and the second heat treatment are continuous steps. Forming a plurality of trenches on the surface of the semiconductor substrate;
Changing the plurality of trenches into one flat cavity by applying heat treatment to the semiconductor substrate;
Etching the semiconductor substrate, selectively leaving a part of the semiconductor substrate on the cavity, and forming a semiconductor region as a waveguide through which light passes through the space around the top surface, the side surface, and the bottom surface of the semiconductor substrate. And a step of forming the semiconductor substrate so as to be incorporated. 6. The cross-sectional area of the first trench due to a plane perpendicular to the depth direction of the first trench is minimized except at the bottom surface of the first trench. 6. Semiconductor substrate manufacturing method.

Images