JP2008277501A - Manufacturing method of laminated wafer - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a method of manufacturing a high-quality laminated wafer having a proper thin film, while suppressing damages to the wafer. <P>SOLUTION: The manufacturing method of a laminated wafer includes a step of executing ion implantation into a first wafer; a surface activation processing step; a step of bringing the ion implantation surface of the first wafer into close contact with a second wafer; a step of bringing a pulling member into close contact with one end of the rear surfaces of the first wafer; a peeling step of giving external impact, from one end of the ion implantation layer of the first waver and pulling the rear surface of the first wafer by the pulling member, to make the first wafer and the second wafer sequentially separate from the one end subjected to the external impact toward the other end in the ion implantation layer, thereby thinning the first wafer. In the manufacturing method, the peeling step is executed, in a state where the width of a direction perpendicular to a direction connecting the one end to the other end in the region that is in close contact with the first wafer by the pulling member is set at 40% or higher of the diameter of the first wafer. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a method for manufacturing a bonded wafer, and more particularly to a method for manufacturing a bonded wafer in which a thin film is formed by mechanical peeling.

  In order to further improve the performance of semiconductor devices, a silicon on insulator (SOI) wafer has attracted attention in recent years. Further, a silicon on quartz (SOQ) wafer, a silicon on sapphire (SOS) wafer, or the like, in which the support wafer (handle wafer) is not silicon, is also used in the fields of TFT-LCD and high frequency (RF) devices, respectively.

  There are several methods for manufacturing such bonded wafers, and a typical example is the Smartcut method (registered trademark). In this method, hydrogen ions are implanted into a single crystal silicon wafer (a donor wafer (referred to as a first wafer in this specification)) on which an oxide film is formed, and a support wafer (a handle wafer (in this specification, a second wafer)). After that, the silicon wafer is peeled along the hydrogen ion implantation interface, and the single crystal silicon thin film is transferred to the handle wafer. At this time, a hydrogen micro-cavity called a microcavity is formed at the hydrogen injection interface, thereby enabling separation at the interface. Thereafter, in order to increase the bonding strength between the single crystal silicon thin film and the handle wafer, a final surface treatment (CMP, heat treatment, etc.) is performed after heat treatment at a high temperature of 1000 ° C. or higher (for example, Patent Document 1). 2 and non-patent document 1).

  However, this method has several problems. In the case of SOI, it is reported that the surface roughness (RMS) after peeling is very rough because peeling depends on a mechanism by microcavity generation. Moreover, it is reported that not only RMS but Peak to Valley (PV) is very high. According to Non-Patent Document 2, even in a very narrow region of 1 × 1 μm, a height difference of about 65 nm occurs in Peak to Valley (P-V). Considering the whole area of the wafer, it is considered that a height difference of 100 nm or more occurs. For this reason, it is necessary to polish 100 nm or more as a surface treatment after peeling, and even when the surface is flattened by heat treatment, a very high temperature treatment such as 1150 ° C. is necessary, and there is a problem of concern about contamination and cost increase To face.

  Further, the problem is more serious in bonding different types of wafers represented by SOQ and SOS. Different types of wafers usually have different thermal expansion coefficients, and there is a problem that the bonded wafers are broken based on the difference in thermal expansion coefficient in a high temperature process of 500 ° C. The same problem occurs when the handle wafer is not quartz or sapphire, but SiC or the like is used as the handle wafer.

  On the other hand, in the SiGen method, before bonding a silicon wafer in which hydrogen ions or the like are implanted on the bonding surface side and a wafer of silicon or another material, the surface of both or one of these wafer bonding surfaces is subjected to plasma. After processing and bonding the two wafers with the surface activated, heat treatment is performed at a low temperature such as 350 ° C. to increase the bonding strength (bonding strength), and then mechanically peeled off at room temperature for bonding. This is a method for obtaining a laminated SOI wafer (for example, see Patent Documents 3 to 5).

  The difference between these two methods is mainly in the silicon thin film peeling process. The Smartcut method requires high-temperature treatment for peeling the silicon thin film, but the SiGen method can be peeled off at room temperature.

  In particular, when a bonded wafer is manufactured by bonding a semiconductor wafer such as a silicon wafer to another material wafer, a difference in thermal expansion coefficient or a specific heat-resistant temperature occurs between different materials. Therefore, it is desirable to perform the process up to the peeling process at as low a temperature as possible. For this reason, it is considered that the SiGen method capable of low-temperature peeling is preferable as a method for producing a bonded wafer by bonding different types of wafers.

By the way, in the case of the SiGen method, when two wafers are bonded together and a semiconductor wafer (for example, a silicon wafer) is thinned to form a semiconductor thin film (for example, a silicon thin film), hydrogen is injected into the donor wafer to be a thin film. Then, the bonding surface of the donor wafer and the handle wafer as a support substrate is activated by plasma treatment or the like, and after bonding, a low temperature (400 ° C. or lower) heat treatment is applied as necessary, and a mechanical shock is applied to the bonding interface. It is necessary to perform peeling.
Patent Document 6 describes that the donor wafer is peeled after the backing portion is integrated with the back surface of the donor wafer in order to give rigidity to the donor wafer.

  However, when the thin film is mechanically peeled by such a method, there has been a problem that the wafer may be broken or broken at the time of peeling.

Japanese Patent No. 3048201 JP-A-11-145438 US Pat. No. 6,263,941 US Pat. No. 6,513,564 US Pat. No. 6,582,999 US Patent Application Publication No. 2006/0211219 A. J. et al. Auberton-Herve et al. , "SMART CUT TECHNOLOGY: INDUSTRIAL STATUS OF SOI WAFER PRODUCTION AND NEW MATERIAL DEVELOPMENTS" (Electrochemical Society Proceedings Volume 99. SOI Science Chapter 2, Realize

  The present invention has been made in order to solve such problems, and an object of the present invention is to provide a method for manufacturing a high-quality bonded wafer having a good thin film while preventing the wafer from being damaged. And

  The present invention has been made in order to solve the above-described problem. The first wafer and the second wafer are bonded together, the first wafer is thinned, and a thin film is formed on the second wafer. A method of manufacturing a laminated wafer, comprising: implanting hydrogen ions and / or rare gas ions from at least a surface of the first wafer, which is a semiconductor wafer, to form an ion implantation layer; and A surface activation process on at least one of a wafer-implanted surface and a second wafer bonding surface, an ion-implanted surface of the first wafer, and a second wafer bonding surface. Holding the back surface of the first wafer by bringing a pulling member into close contact with one end of the back surface of the first wafer, and A wafer holding step of holding the back surface of the wafer with a holder, and applying an external impact from one end of the first wafer on the side where the pulling member is in close contact with the ion-implanted layer. By pulling the back surface of the first wafer by a pulling member, the first wafer and the second wafer are moved from the one end portion to which the external impact is applied to the ion implantation layer toward the other end portion. And a separation step of thinning the first wafer, and perpendicular to a direction connecting the one end and the other end of the region to be in close contact with the first wafer by the pulling member A method for manufacturing a bonded wafer, wherein the peeling step is performed with a width in a forming direction being 40% or more of a diameter of the first wafer (Claim 1). .

  The width in the direction perpendicular to the direction connecting the one end and the other end where the external impact is applied and the other end of the region where the pulling member is brought into close contact with the first wafer is defined as In the case of a method for manufacturing a bonded wafer in which a peeling process is performed with a diameter of 40% or more of one wafer, the angle between the direction in which the wafer is pulled and the direction in which peeling actually proceeds is relatively small around the wafer. Since stress concentrated on the periphery of the wafer can be relaxed, damage to the wafer can be effectively prevented. As a result, a high-quality bonded wafer having a good thin film can be manufactured while preventing the wafer from being damaged.

In this case, the pulling member may be composed of one or more suction cups or a porous material (Claim 2).
Thus, if the pulling member is made of one or more suction cups or a porous material, the first wafer can be reliably held and pulled by a simple mechanism.

The first wafer may be any of a single crystal silicon wafer, a single crystal silicon wafer having an oxide film formed on the surface, and a compound semiconductor wafer.
According to the present invention, a high-quality bonded wafer having a good silicon thin film that has been particularly demanded in recent years, for example, an SOI wafer can be more reliably manufactured.
Further, according to the present invention, not only a silicon thin film but also a bonded wafer having a good thin film made of a compound semiconductor such as GaN can be more reliably manufactured.

In the method for producing a bonded wafer according to the present invention, the second wafer is a quartz wafer, a sapphire (alumina) wafer, a SiC wafer, a borosilicate glass wafer, a crystallized glass wafer, an aluminum nitride wafer, or a single crystal silicon wafer. It can be either a single crystal silicon wafer having a surface formed with an oxide film or a SiGe wafer.
The second wafer used in the bonded wafer of the present invention can be appropriately selected from these according to the purpose of the semiconductor device to be manufactured.

In addition, it is preferable that the holder for holding the second wafer is a support plate that is in close contact with the back surface of the second wafer.
In this way, if the holder for holding the second wafer is a support plate that is in close contact with the back surface of the second wafer, the wafer may be damaged or stuck due to excessive bending of the second wafer during the peeling process. Local separation of the mating surfaces, generation of peeling marks, untransferred thin film, and the like can be prevented more reliably.

In this case, it is preferable that the support plate has a Young's modulus of 3 GPa or more and a thickness of 1 mm or more.
As described above, the support plate to be in close contact with the second wafer has a Young's modulus of 3 GPa or more and a thickness of 1 mm or more, whereby the support plate and the second wafer in which the support plate is in close contact are obtained. It can be rigid and more reliably prevent wafer breakage due to excessive bending of the second wafer during the peeling process, local separation of the bonding surface, occurrence of peeling marks, and untransferred thin film. be able to.

In particular, the material of the support plate can be silicon.
As described above, if the material of the support plate is silicon, the support plate is suitable particularly when the second wafer is silicon.

Moreover, in the manufacturing method of the bonding wafer of this invention, it is preferable to make the surface roughness of the said support plate into the range of RMS [nm] = 1-100 (Claim 8).
As described above, by setting the surface roughness of the support plate to a range of RMS [nm] = 1 to 100, the surface of the support plate can be easily adhered to the back surface of the second wafer. And the wafer can be supported by a support plate. If the surface roughness of the support plate is too rough, the surface state is reflected on the wafer surface after peeling, and appears as uneven film thickness. On the other hand, if the surface roughness of the support plate is too smooth, it is too close to the wafer and it becomes difficult to remove the wafer from the support plate. Therefore, the above range is preferable.

Further, it is preferable that the support plate is adhered to the back surface of the second wafer by vacuum suction or electrostatic chuck.
As described above, if the support plate is adhered to the back surface of the second wafer by vacuum suction or electrostatic chucking, it can be easily and reliably adhered to the second wafer and stiffened. .

In the method for producing a bonded wafer according to the present invention, after the step of bringing the first wafer and the second wafer into close contact, a heat treatment step of heat-treating the attached wafer at 100 to 400 ° C. is performed, It is preferable to perform the wafer holding step (claim 10).
As described above, after the first wafer and the second wafer are brought into close contact with each other, the attached wafer is heat-treated at 100 to 400 ° C., thereby further increasing the bonding strength between the first wafer and the second wafer. be able to. In particular, when the heat treatment temperature is 100 to 300 ° C., it is possible to further reduce the possibility of thermal distortion, cracking, peeling, and the like due to differences in thermal expansion coefficients even when wafers of different materials are bonded. On the other hand, when the same material is bonded to each other as in the case where both the first wafer and the second wafer are silicon wafers, heat treatment can be performed at a temperature up to 400 ° C., and the bonding strength can be further increased. .

In the method for manufacturing a bonded wafer according to the present invention, the tip of a member having a wedge-shaped cross section is brought into contact with one end of the ion implantation layer during the peeling step, so that the ion implantation layer is externally provided. An impact can be applied (claim 11).
Thus, for example, the starting point of cleavage can be formed by bringing the tip of a member having a wedge-shaped cross section into contact with the one end and applying an external impact to the ion implantation layer of the first wafer. it can. And a favorable thin film can be formed by advancing cleavage from this starting point.

In the method for producing a bonded wafer according to the present invention, it is preferable that the surface activation treatment is performed by at least one of plasma treatment and ozone treatment.
As described above, when the surface activation treatment is performed by at least one of the plasma treatment and the ozone treatment, the surface of the wafer subjected to the surface activation treatment is activated by increasing OH groups. Therefore, in this state, if the surface of the first wafer into which ions are implanted and the surface of the second wafer to be bonded are brought into close contact with each other, the wafer can be bonded more firmly by hydrogen bonding or the like.

  As described above, according to the method for manufacturing a bonded wafer of the present invention, the angle between the direction in which the first wafer is pulled and the direction in which peeling proceeds can be made relatively small even in the periphery of the wafer. Since stress concentrated on the periphery of the wafer can be relieved, damage to the wafer can be effectively prevented. In addition, since a high-temperature heat treatment exceeding 400 ° C., for example, is not required until the first wafer is peeled off, a good thin film can be obtained. As a result, a high-quality bonded wafer having a good thin film can be manufactured while preventing the wafer from being damaged.

Hereinafter, the present invention will be described in more detail.
As described above, when a bonded wafer is manufactured without performing high-temperature heat treatment, if the thin film is mechanically peeled off, there is a problem that the wafer breaks and breaks at the time of peeling. .

The present inventors have intensively investigated the cause of such a problem. As a result, particularly when the substantially circular wafers are bonded to each other, if the width of the portion where the first wafer is pulled is small relative to the wafer, the wafer is frequently damaged, that is, if the width of the pulled portion is narrow, peeling occurs. A traveling wave (a curve connecting points where the peeled state is the same within the wafer surface being peeled) has an arc shape around the start point of the peeling, and the direction of pulling the first wafer and the traveling direction of the surrounding peeling traveling wave The present inventors have found that a large angle is formed between them, the peeling does not proceed, stress concentrates on the periphery, and the first wafer is considered to be damaged.
The inventors of the present invention make the curvature of the separation traveling wave smooth by increasing the width of the tensile portion, and the direction in which the first wafer is pulled also in the peripheral portion of the wafer and the traveling direction of the separation traveling wave in the peripheral portion. It was conceived that the wafer can be prevented from being damaged by preventing a large angle from being formed between the wafers.
Furthermore, the inventors have further conducted experiments and studies, and found that the yield is greatly improved by setting the adhesion region width of the pulling member to about 40% or more of the wafer diameter, thereby completing the present invention.
In the following, “wafer” refers to a substantially circular substrate (including a substrate having a notch, an orientation flat, etc.).

Hereinafter, embodiments of the present invention will be described with reference to the drawings, but the present invention is not limited thereto.
FIG. 3 shows a method for manufacturing a bonded wafer to which the present invention can be applied.

First, as shown in FIG. 3A, a first wafer 10 and a second wafer 20 which are semiconductor wafers are prepared (step a).
At this time, the first wafer 10 can be a single crystal silicon wafer, and in particular, a single crystal silicon wafer having an oxide film formed on the surface thereof. If these materials are selected as the first wafer, a bonded wafer having a silicon thin film can be manufactured. If a single crystal silicon wafer having an oxide film formed on the surface is used, it is convenient for manufacturing an SOI wafer. Further, in order to manufacture a bonded wafer having a compound semiconductor thin film such as GaN instead of a silicon thin film, the first wafer 10 can be a compound semiconductor wafer such as GaN.

  Further, the second wafer 20 can be an insulating wafer of any of a quartz wafer, a sapphire (alumina) wafer, a SiC wafer, a borosilicate glass wafer, a crystallized glass wafer, and an aluminum nitride wafer, or A single crystal silicon wafer, a single crystal silicon wafer having an oxide film formed on its surface, or a SiGe wafer can be used. The second wafer 20 may be appropriately selected from these according to the purpose of the semiconductor device to be manufactured. Of course, other materials may be used.

Next, as shown in FIG. 3B, hydrogen ions are implanted from the surface (ion implantation surface) 12 of the first wafer 10 to form the ion implantation layer 11 (step b).
The ion-implanted layer 11 may be formed by implanting not only hydrogen ions but also rare gas ions or both hydrogen ions and rare gas ions. Other ion implantation conditions such as implantation energy, implantation dose, and implantation temperature may be appropriately selected so that a thin film having a predetermined thickness can be obtained. As a specific example, the wafer temperature at the time of implantation is 250 to 400 ° C., the ion implantation depth is 0.5 μm, the implantation energy is 20 to 100 keV, and the implantation dose is 1 × 10 ¹⁶ to 1 × 10 ¹⁷ / cm. It includes be ^two, but not limited thereto.
Note that if ion implantation is performed through an oxide film using a single crystal silicon wafer having an oxide film formed on the surface, an effect of suppressing channeling of implanted ions can be obtained, and variations in ion implantation depth can be further suppressed. it can. Thereby, a thin film with higher film thickness uniformity can be formed.

Next, as shown in FIG. 3C, a surface activation process is performed on the surface 12 of the first wafer 10 on which ions are implanted and the surface 22 on which the second wafer 20 is bonded (step c). The bonding surface 22 of the second wafer 20 is a surface to be bonded to the first wafer in the bonding process of the next step d.
Of course, the surface activation process may be performed only on either one of the surface 12 into which the first wafer 10 is ion-implanted and the surface 22 on which the second wafer 20 is bonded.
At this time, the surface activation treatment is preferably performed by at least one of plasma treatment and ozone treatment. As described above, when the surface activation treatment is performed by at least one of the plasma treatment and the ozone treatment, the surface of the wafer subjected to the surface activation treatment is activated by increasing OH groups. Therefore, in this state, if the surface 12 into which the first wafer is ion-implanted and the bonding surface 22 of the second wafer are brought into close contact with each other, the wafer can be bonded more firmly by hydrogen bonding or the like.

  When processing with plasma, a wafer cleaned with RCA cleaning or the like is placed in a vacuum chamber, and after introducing a plasma gas, it is exposed to high-frequency plasma of about 100 W for about 5 to 30 seconds, and the surface is subjected to plasma processing. To do. As the plasma gas, for example, when processing a single crystal silicon wafer having an oxide film formed on the surface, oxygen gas plasma, and when processing a single crystal silicon wafer having no oxide film formed on the surface, Gas, argon gas, a mixed gas thereof, or a mixed gas of hydrogen gas and helium gas can be used. Further, an inert gas such as nitrogen gas may be used.

  When processing with ozone, a wafer subjected to cleaning such as RCA cleaning is placed in a chamber into which air is introduced, and after introducing a plasma gas such as nitrogen gas or argon gas, high-frequency plasma is generated, The surface is treated with ozone by converting the oxygen in it into ozone.

Next, as shown in FIG. 3 (d), the ion-implanted surface 12 of the first wafer and the surface 22 to which the second wafer is bonded are brought into close contact (step d).
In this way, if the surfaces subjected to the surface activation treatment are used as the bonding surface, for example, if the wafers are brought into close contact at room temperature under reduced pressure or normal pressure, both wafers can withstand subsequent mechanical peeling without being subjected to high temperature treatment. It can be bonded firmly enough to obtain.

Note that after the step of bringing the first wafer and the second wafer into close contact, a heat treatment step of heat-treating the attached wafer at 100 to 400 ° C. may be performed.
As described above, after the first wafer and the second wafer are brought into close contact with each other, the attached wafer is heat-treated at 100 to 400 ° C., thereby increasing the bonding strength between the first wafer and the second wafer. Can be increased. In particular, when the heat treatment temperature is 100 to 300 ° C., there is little risk of thermal distortion, cracking, peeling, etc. due to differences in thermal expansion coefficient even when wafers of different materials are bonded. If the bonding strength is increased, the occurrence of defects in the peeling process can be reduced.

Next, the pulling member 40 is brought into close contact with one end of the back surface of the first wafer 10 to hold the back surface of the first wafer 10, and the back surface of the second wafer 20 is held by a holder (wafer holding). Step: Step e). The “back surface” is a surface on the opposite side to the ion implantation surface 12 that is the bonding surface for the first wafer 10 and a surface on the opposite side to the bonding surface 22 for the second wafer 20. Point to.
FIG. 3E shows a mode in which the back surface of the first wafer 10 and the back surface of the second wafer 20 are both held by the pulling members 40 and 50 as an example of the wafer holding step.
As the pulling member, one composed of one or more suction cups, one composed of a porous material and adsorbed to the first wafer by vacuum adsorption, etc. can be used. It is not limited.

Further, in this wafer holding step, as shown in FIG. 3 (e ′), the second wafer 20 and the holder can be closely attached using a support plate 52 as a holder. If the second wafer 20 and the support plate 52 are brought into close contact with each other by vacuum chucking or electrostatic chucking, the second wafer 20 is supported by the support plate 52 with a simple mechanism and more reliably. This is preferable.
In addition, the support plate 52 and the second wafer 20 can be securely adhered to each other by bonding with an adhesive. However, it is necessary to remove the adhesive later. On the other hand, the adsorption and electrostatic chuck can be easily detached.

Next, an external impact is applied from one end of the ion implantation layer 11 of the first wafer 10 on the side where the pulling member 40 is in close contact, and at least the back surface of the first wafer 10 is moved by the pulling member 40. By pulling, the first wafer 10 and the second wafer 20 are sequentially separated from the one end portion (exfoliation start point) to which the external impact is applied toward the other end portion by the ion implantation layer 11, The wafer 10 is thinned (peeling step: step f). FIG. 3F shows a mode in which the pulling member 40 holding the back surface of the first wafer 10 and the pulling member 50 as a holding tool holding the back surface of the second wafer 20 are pulled together. Show.
In this way, by separating the first wafer 10 and the second wafer 20 at the ion implantation layer 11 by cleaving from one end to the other end, the cleavage is directed in one direction. Therefore, cleavage control is relatively easy, and a thin film with high film thickness uniformity can be obtained.

  At this time, the tip of the member having a wedge-shaped cross section is brought into contact with one end of the first wafer 10 and an external impact is applied to the ion implantation layer 11 of the first wafer 10 by the wedge-shaped member. A starting point can be formed. The means for forming the starting point of cleavage is not limited to the wedge-shaped member, and for example, an external impact may be applied to the ion implantation layer 11 with a high-pressure fluid such as air, nitrogen gas, or pure water.

  Further, when the above-described wafer holding step is performed using the support plate 52 as a holder for the adhesion between the second wafer 20 and the holder as shown in FIG. As shown in FIG. 3F ′, only the pulling member 40 in close contact with the first wafer 10 may be pulled, and the support plate 52 may be fixed.

In the present invention, as schematically shown in FIG. 1, in the wafer holding step, one end portion that gives an external impact in the peeling step (start of peeling) in the region that is brought into close contact with the first wafer 10 by the pulling member 40. Point) and a width in a direction perpendicular to the direction connecting the separation start point and the opposite point (the other end) with respect to the wafer center in the plane of the wafer (hereinafter referred to as an adhesion region width d). May be 40% or more of the diameter of the first wafer 10. FIG. 1A shows an embodiment in which a suction pad made of a porous material is used as the pulling member 40 to be in close contact with the back surface of the first wafer 10, and FIG. The aspect at the time of using a some suction cup as the member 40 for a tension | pulling is shown.
In addition, when there are a plurality of regions where the pulling member 40 is in close contact with the wafer as shown in FIG. 1B, it is required that the tensile pressure is uniform to some extent around the close contact region. Do not overly sparse the area to be. Further, as shown in FIG. 1B, when there are a plurality of regions where the pulling member 40 is in close contact with the wafer, the close contact region width d is the largest of the plurality of regions in which the pulling member 40 is in close contact with the wafer. Defined on the outside.

In general, the diameters of the first wafer and the second wafer are the same, but the present invention is not limited to this, and the contact area width d described above is only for the diameter of the first wafer. It only has to be decided.
In the following, a case where the diameters of the first wafer and the second wafer are the same will be described, and the contact area width d will describe the ratio of both to the wafer diameter.

In this way, the adhesion region width d is set to 40% or more of the wafer diameter, and then the peeling process (FIGS. 3F and 3F ′) is performed.
The peeling start point is set near the center of the adhesion region width of the pulling member 40 (see FIGS. 1A and 1B).
As described above, by setting the adhesion region width d to a large value such as 40% or more of the wafer diameter, the curvature of the peeling traveling wave is made smooth, and the direction in which the first wafer is pulled also in the peripheral portion of the wafer and the periphery of the wafer. A large angle (angle θ in FIG. 2A) is not formed between the traveling direction of the peeling traveling wave (the peeling traveling direction). For this reason, stress concentrated on the periphery of the wafer can be relieved, so that breakage of the wafer can be effectively prevented.

  On the other hand, as shown in FIG. 2B, when the adhesion region width d is a small value such as less than 40% of the wafer diameter, the separation traveling wave becomes a circle around the vicinity of the separation start point, and the first A large angle θ is formed between the direction of pulling the wafer and the traveling direction of the separation traveling wave in the peripheral portion, separation does not proceed, stress concentrates on the periphery, and the wafer is damaged.

  Then, through the above steps (FIGS. 3A to 3F), high-quality bonding having a good thin film 61 on the second wafer 20 as shown in FIG. 3G. The wafer 60 can be more reliably manufactured by preventing the wafer from being damaged.

  When the wafer holding step and the peeling step are carried out using the support plate 52 as shown in FIGS. 3E 'and 3F', the support plate 52 is attached to the Young plate. It is desirable that the rate is 3 GPa or more and the thickness is 1 mm or more. If the support plate to which the second wafer is brought into close contact is made as described above, it has sufficient rigidity, and the second wafer to which the support plate is brought into close contact also has sufficient rigidity. Note that the higher the rigidity of the support plate 52 that is in close contact with the second wafer, the better. As a result, it is possible to more reliably prevent wafer breakage, local separation of the bonding surface, occurrence of peeling marks, untransferred thin film, and the like due to excessive bending of the second wafer during the peeling process.

  Specifically, the support plate 52 can be a silicon plate, and in particular, when the second wafer is silicon, or a surface on which the support plate 52 is in close contact with silicon in which an oxide film is formed. A case of silicon is preferable. In addition, any one of a stainless steel plate, a ceramic plate such as alumina, an aluminum plate, a glass plate, and a plastic plate such as vinyl chloride can be used. The material of the support plate 52 can be appropriately selected from these depending on the type of the second wafer to be adhered.

  Moreover, it is preferable to make the surface roughness of the support plate 52 into the range of RMS [nm] = 1-100. As described above, the surface roughness of the support plate is in the range of RMS [nm] = 1 to 100, and the surface to be in close contact with the second wafer is smoothed so that the surface of the support plate is the second wafer. It becomes easy to adhere to the back surface, and therefore the second wafer can be supported more firmly. If the surface roughness of the support plate is too rough, the surface state is reflected on the second wafer surface after peeling, and appears as film thickness unevenness. In addition, if the surface roughness of the support plate is too smooth, it is too close to the second wafer and it is difficult to remove the second wafer from the support plate.

  Hereinafter, the reason for setting the contact region width of the pulling member to be in close contact with the first wafer to 40% or more of the first wafer diameter will be described with reference to experimental examples.

(Experimental example 1: Wafer diameter 150 mm)
In the following manner, a bonded wafer was manufactured according to the method for manufacturing a bonded wafer as shown in FIG.
As the first wafer 10, a mirror-polished single crystal silicon wafer with a diameter of 150 mm was prepared. Then, a 100 nm silicon oxide film layer was formed on the surface of the first wafer by thermal oxidation. A synthetic quartz wafer having a diameter of 150 mm was prepared as the second wafer 20 (step a).
Next, hydrogen ions are implanted into the first wafer 10 through the formed silicon oxide film layer to form a microbubble layer (ion implantation layer) 11 parallel to the surface at an average depth of ions (process). b). The ion implantation conditions are an implantation energy of 35 keV, an implantation dose of 9 × 10 ¹⁶ / cm ² , and an implantation depth of 0.3 μm.

Next, the first wafer 10 ion-implanted in the plasma processing apparatus is placed, nitrogen is introduced as a plasma gas, and then a 13.56 MHz high frequency wave having a diameter of 300 mm is parallel under a reduced pressure of 2 Torr (270 Pa). By applying a high frequency power of 50 W between the plate electrodes, the high frequency plasma treatment was performed for 10 seconds on the ion-implanted surface. In this manner, the surface activation treatment was performed on the ion implantation surface of the first wafer 10.
On the other hand, the second wafer 20 is placed in a plasma processing apparatus, nitrogen gas is introduced as a plasma gas between narrow electrodes, and then a high frequency is applied between the electrodes to generate a plasma. Processing was carried out for 10 seconds. Thus, the surface activation process was performed also on the surface bonded in the next bonding step of the second wafer 20 (step c).

The first wafer 10 and the second wafer 20 that have been subjected to the surface activation treatment as described above are brought into close contact with each other at the room temperature using the surface that has undergone the surface activation treatment as a bonding surface, and then the back surfaces of both wafers are adhered to each other. Pressing strongly in the thickness direction (step d).
Next, in order to increase the bonding strength, the wafer in which the first wafer 10 and the second wafer 20 were in close contact was heat-treated at 300 ° C. for 30 minutes.

Next, a suction pad was brought into close contact with the back surface of the first wafer 10 as the pulling member 40 (step e). The adhesion area width d of the suction pad at this time was 30% of the first wafer diameter, that is, 45 mm.
Further, the back surface of the second wafer 20 was held by the support plate 52.
Next, in order to form the starting point of cleavage, an external impact was applied to the ion implantation layer 11 of the first wafer from one end thereof with a blade of a paper cutting scissors. Thereafter, the pulling member 40 that holds the first wafer 10 and the holder that holds the support plate 52 in close contact with the second wafer 20 are relatively separated from each other, whereby the first wafer 10 and the first wafer 10 are separated from each other. The two wafers 20 were sequentially separated by the ion implantation layer 11 from one end portion to which the external impact was applied to the other end portion (step f).

In this way, a total of 20 bonded SOI wafers were manufactured.
As a result, in the peeling step (step f), the wafer was partially damaged at 10 out of 20 times.

(Experimental examples 2 to 6: wafer diameter 150 mm)
Similarly to Experimental Example 1, a 150 mm diameter single crystal silicon wafer having a silicon oxide film layer of 100 nm formed on the surface by thermal oxidation was used as the first wafer 10 and a 150 mm diameter synthetic quartz wafer was used as the second wafer 20. The production of bonded wafers (SOI wafers) was attempted 20 sheets in each case. However, the adhesion region width d of the suction pad (the pulling member 40) was performed as shown in Table 1.

(Experimental examples 7 to 12: wafer diameter 200 mm)
Similarly to Experimental Example 1, a 200 mm diameter single crystal silicon wafer having a silicon oxide film layer of 100 nm formed on the surface by thermal oxidation is used as the first wafer 10, and a 200 mm diameter synthetic quartz wafer is used as the second wafer 20. The production of bonded wafers (SOI wafers) was attempted 20 sheets in each case. However, the adhesion region width d of the suction pad (the pulling member 40) was performed as shown in Table 1.

(Experimental examples 13 to 18: wafer diameter 300 mm)
Similarly to Experimental Example 1, a 300 mm diameter single crystal silicon wafer having a silicon oxide film layer of 100 nm formed on the surface by thermal oxidation was used as the first wafer 10, and a 300 mm diameter synthetic quartz wafer was used as the second wafer 20. The production of bonded wafers (SOI wafers) was attempted 20 sheets in each case. However, the adhesion region width d of the suction pad (the pulling member 40) was performed as shown in Table 1.

  The production yield of bonded wafers of Experimental Examples 1 to 6 (wafer diameter 150 mm), Experimental Examples 7 to 12 (wafer diameter 200 mm), and Experimental Examples 13 to 18 (wafer diameter 300 mm) is shown in FIG. Note that the yield here indicates the ratio at which the bonded wafer can be manufactured without damaging the first wafer and the second wafer.

  In the case of a wafer diameter of 150 mm (Experimental Examples 1 to 6), a yield of 90% can be achieved if the suction pad contact area width d is 40%, and if the suction pad contact area width d is 45%, the wafer can be achieved. No damage occurred.

  In the case of a wafer diameter of 200 mm (Experimental Examples 7 to 12), a yield of 85% can be achieved if the adhesion area width d of the suction pad is 40%, and a wafer can be achieved if the adhesion area width d of the adsorption pad is 45%. No damage occurred.

  Even in the manufacture of bonded wafers (Experimental Examples 13 to 18) in which wafers having a diameter of 300 mm are bonded to each other, the yield of 80% can be achieved if the adhesion area width d of the suction pad is 40%. When the adhesion area width d of the suction pad was 45%, a yield of 90% could be achieved. Further, when the adhesion area width d of the suction pad was 50%, the wafer was not damaged.

From the above experimental results, it can be seen that if the adhesion region width d of the pulling member is 40% or more, preferably 45% or more of the wafer diameter, the wafer can be effectively prevented from being damaged and a bonded wafer can be manufactured. It was.
In particular, it has been found that the effect of the present invention is great when the diameter of the wafer to be bonded is as large as 300 mm.

  The present invention is not limited to the above embodiment. The above-described embodiment is an exemplification, and the present invention has substantially the same configuration as the technical idea described in the claims of the present invention, and any device that exhibits the same function and effect is the present invention. It is included in the technical scope of the invention.

It is a top view which shows typically the area | region which closely_contact | adheres a tension | pulling member to the 1st wafer in the manufacturing method of the bonding wafer of this invention, (a) is a case where a suction pad is used as a tension | pulling member, (b) Indicates a case where a plurality of suction cups are used as the pulling member. In the manufacturing method of a bonded wafer, it is a top view which shows typically a mode that exfoliation of the 1st wafer advances, (a) is when the adhesion region width of a tension member is large, (b) is a tension member This shows a case where the width of the contact area is small. It is a flowchart which shows an example of the manufacturing method of the bonded wafer to which the manufacturing method of the bonded wafer of this invention is applied. It is a graph which shows the relationship between the contact | adherence area | region width | variety of a tension member, and a yield.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 ... 1st wafer, 11 ... Ion implantation layer, 12 ... Ion implantation surface,
20 ... second wafer, 22 ... surface to be bonded,
40 ... Tensile member, 50 ... Tensile member (holding tool), 52 ... Support plate,
60 ... Laminated wafer, 61 ... Thin film.

Claims (12)

A method of manufacturing a bonded wafer having a first wafer and a second wafer bonded together, thinning the first wafer, and having a thin film on the second wafer, comprising:
A step of implanting hydrogen ions or rare gas ions or both from the surface of the first wafer, which is a semiconductor wafer, to form an ion implantation layer;
Applying a surface activation treatment to at least one of the ion-implanted surface of the first wafer and the surface to be bonded to the second wafer;
Adhering the ion-implanted surface of the first wafer and the surface to be bonded of the second wafer;
A wafer holding step in which a pulling member is brought into close contact with one end of the back surface of the first wafer to hold the back surface of the first wafer, and the back surface of the second wafer is held by a holder;
By applying an external impact from one end of the ion implantation layer of the first wafer on the side where the pulling member is in close contact, and at least pulling the back surface of the first wafer by the pulling member The first wafer and the second wafer are sequentially separated from one end portion to which the external impact is applied from the one end portion to the other end portion by the ion implantation layer, and the first wafer is peeled off to form a thin film. A width of a region that is in close contact with the first wafer by the pulling member in a direction perpendicular to the direction connecting the one end and the other end is 40 of the diameter of the first wafer. % Or more, the method for producing a bonded wafer, wherein the peeling step is performed.   The method for producing a bonded wafer according to claim 1, wherein the pulling member is made of one or more suction cups or a porous material.   3. The bonded wafer according to claim 1, wherein the first wafer is any one of a single crystal silicon wafer, a single crystal silicon wafer having an oxide film formed on a surface thereof, and a compound semiconductor wafer. Production method.   The second wafer is a quartz wafer, a sapphire (alumina) wafer, a SiC wafer, a borosilicate glass wafer, a crystallized glass wafer, an aluminum nitride wafer, a single crystal silicon wafer, a single crystal silicon wafer having an oxide film formed on the surface, The method for manufacturing a bonded wafer according to any one of claims 1 to 3, wherein the method is any one of SiGe wafers.   5. The bonded wafer according to claim 1, wherein the holding tool that holds the second wafer is a support plate that is in close contact with a back surface of the second wafer. 6. Production method.   The method for producing a bonded wafer according to claim 5, wherein the support plate has a Young's modulus of 3 GPa or more and a thickness of 1 mm or more.   The method for manufacturing a bonded wafer according to claim 5 or 6, wherein a material of the support plate is silicon.   The method for producing a bonded wafer according to claim 5, wherein the surface roughness of the support plate is in the range of RMS [nm] = 1 to 100.   The bonded wafer according to any one of claims 5 to 8, wherein the support plate is adhered to the back surface of the second wafer by vacuum suction or an electrostatic chuck. Production method.   2. The method according to claim 1, wherein after the step of bringing the first wafer and the second wafer into close contact with each other, a heat treatment step of heat-treating the attached wafer at 100 to 400 ° C. is performed, and then the wafer holding step is performed. The manufacturing method of the bonding wafer as described in any one of Claim 1 thru | or 9.   2. The external impact is applied to the ion implantation layer by bringing a tip portion of a wedge-shaped member into contact with one end portion of the ion implantation layer during the peeling step. The method for producing a bonded wafer according to claim 10.   The method for producing a bonded wafer according to any one of claims 1 to 11, wherein the surface activation treatment is performed by at least one of a plasma treatment and an ozone treatment.

Images