JP7429368B2 - Device manufacturing method and bonding equipment, semiconductor device - Google Patents

Description

The present invention relates to semiconductor devices.

In the manufacturing process of semiconductor devices, bonding different parts together is an important process. A typical example of bonding is die bonding, in which individual semiconductor chips (dies) are bonded to a heat dissipation substrate or the like. For example, LED (light emitting diode) chips and power devices are mounted on a heat dissipation substrate by die bonding. Bonding technology is also used for packaging (hermetic sealing, vacuum sealing) of MEMS and sensors.

Conventional die bonding has mainly used adhesives such as silver paste, but in recent years, progress has been made in developing die bonding methods that do not use adhesives and do not generate outgas.

For example, the present inventors have proposed a method to perform strong wafer bonding at room temperature, in the air, and without pressure by using a gold (Au) thin film and activating the surface with low-vacuum plasma ( (See Non-Patent Document 1).

E. Higurashi et al., IEICE Transactions on Electronics, vol. E100-C, no. 2, pp. 156-160, 2017 Y. Kurashima et al. Microelectronic Engineering, vol. 189 pp. 1-5, (2018). M. Yamamoto et al. Journal of Automation Technology (IJAT), vol. 13, pp. 254-260, (2019).

Since the method of Non-Patent Document 1 is a method of bonding using the adhesive force between atoms, an extremely smooth bonding surface (rms surface roughness: 0.5 nm or less) is required, and an extremely thin film (thickness: Although this was achieved using ultra-precision polishing such as chemical mechanical polishing (CMP) to obtain a smooth surface, removal processing using ultra-precision polishing such as chemical mechanical polishing (CMP) has generally been used to obtain a smooth surface. In addition, in order to combine dissimilar materials with various functions into a three-dimensional structure, it is necessary not only to bond them at the wafer level, but also to implement high-precision mounting techniques at the chip level of highly functional elements fabricated using optimal processes. must also be used together. In general, the surface of a thick metal thin film electrode with a thickness of 500 nm or more formed by sputtering etc. has an rms surface roughness of several nanometers or more, and the surface roughness increases even more when sintering treatment is performed to form an ohmic contact. Since an ultra-smooth (rms surface roughness: 0.2 nm) surface like that of a Si wafer cannot be obtained, it has been difficult to bond at low temperatures (below 150° C.) and under low loads (below 200 MPa).

Furthermore, it is difficult to apply the CMP process to the inside of a cavity of a Si wafer that has undergone micromachining such as cavity processing, and there is a major problem in that it is difficult to smooth the interior in a post-process.

A template stripping process is known as a method for obtaining a smooth surface. This method is a method of obtaining a smooth surface by peeling a thin film deposited on a smooth template substrate from the template substrate, and research is being conducted using surface plasmon analysis.

The present inventors have studied the use of template stripping for bonding (Non-Patent Documents 2 and 3).

FIGS. 1A to 1D are diagrams illustrating bonding using template stripping. As shown in FIG. 1(a), a template substrate 4 having a metal thin film 6 formed on its surface is prepared. The object 2 whose surface is to be smoothed is then pressed onto it. As a result, as shown in FIG. 1(b), a part of the metal thin film 6 is peeled off and transferred to the surface of the object 2, and a smooth surface is formed on the object 2. The smoothness of the surface of this object 2A reflects the smoothness of the template substrate 4.

As shown in FIG. 1(c), a metal thin film 9 is also formed on the surface of the joining partner 8 of the object 2A, and as shown in FIG. 1(d), the object 2A and the joining partner 8 are joined.

As shown in FIGS. 1A to 1D, in the case of the object 2 having a surface with relatively low surface roughness, the thickness t1 of the metal thin film 6 can be reduced. Therefore, during transfer, excessive pressure is not required and there is no need to heat at high temperatures.

The present inventors have considered bonding objects 2, such as semiconductor chips, having rough surfaces larger than several nanometers by the methods shown in FIGS. 1(a) to 1(d). FIGS. 2A and 2B are diagrams illustrating template stripping on an object 2 having a rough surface.

When the surface of the object 2 is rough, it is necessary to form a metal thin film 6 on the template substrate 4 with a thickness t2 sufficient to compensate for the roughness of the object 2. Here, as the thickness t2 of the metal thin film 6 increases, the surface of the metal thin film 6 becomes rougher (Non-Patent Document 1). In such a case, when transferring the metal thin film 6 from the template substrate 4 to the object 2, it is necessary to apply a large pressure and also to heat it to a high temperature (Non-Patent Document 2). Specifically, in order to obtain a smooth surface using this method, a large pressure of 300 MPa or more or a high temperature heat treatment of 200° C. or more is required. Further, if the surface smoothness obtained by this method is not sufficient, high-temperature heat treatment or large pressure will be required when joining the object 2 and the joining partner 8.

When the target object 2 is a semiconductor chip having an active layer, there is a problem that the semiconductor chip deteriorates when subjected to high pressure treatment and high temperature heat treatment. Due to these circumstances, there is a large practical barrier to applying template stripping to smoothing and bonding objects 2 with rough surfaces.

The present invention has been made in view of the above-mentioned problems, and an exemplary object of one aspect of the present invention is to provide a joining technique that can form a smooth surface at low temperature and/or low load.

One aspect of the present invention relates to a method for manufacturing a semiconductor device. The manufacturing method includes the steps of forming a first metal thin film on the surface of a template substrate having a smooth surface, and applying an activated first metal thin film multiple times on the surface of a member to be smoothed by template stripping. and a step of superimposing the transfer and forming a multilayer metal film on the surface of the member.

By repeatedly performing template stripping, a multilayer metal film having a laminated structure of first metal thin films is formed on the surface of the member. Each layer of the multilayer metal film undergoes plastic deformation on a nanometer scale during the lamination process, and each time the transfer is repeated, the amount of deformation becomes smaller and smoothed, until the final layer becomes extremely smooth (for example, the rms surface roughness is 0.5 nm or less) can be obtained. This manufacturing method has the advantage that high temperature heat treatment is not required and/or large pressure treatment is not required. Immediately after the first metal thin film is deposited (for about a week in the case of a long period of time), the activated state is maintained, so no special surface activation treatment is required, but after a certain amount of time has passed since the deposition. It is preferable to perform template stripping after surface activation treatment.

In one embodiment, the manufacturing method includes the steps of: forming a second metal thin film on the surface of a support substrate to be joined; and contacting the member and the support substrate in a state where the multilayer metal film and the second metal thin film are activated. The method may further include a step of joining the material. When the surface of the supporting substrate is smooth and the bonding surface of the second metal thin film is also smooth, by bringing the first metal thin film, which is also smooth at the atomic level, into contact with the second metal thin film, an adhesion force between atoms is generated. This makes it possible to join the joints in a similar manner. This bonding process also has the advantage of not requiring high-temperature heating or strong pressure between the member and the supporting substrate. Note that the first metal thin film and the second metal thin film may be made of the same material such as Au--Au, or may be made of different materials.

Further, low vacuum plasma or atmospheric pressure plasma may be used for surface activation of the first metal thin film, multilayer metal film, and second metal thin film. When atmospheric pressure plasma is used, processing at atmospheric pressure becomes possible.

In the process of multiple transfers, multiple template substrates having different rigidities may be used. As described above, as the transfer is repeated, the amount of plastic deformation of each layer of the multilayer metal film gradually decreases. By optimizing the rigidity of the template substrate for each transfer, the amount of plastic deformation of each layer can be adjusted, and a smoother surface can be achieved.

Among the plurality of template substrates, the elastic modulus of the template substrate used in the final stage may be relatively larger than the elastic modulus of the template substrate used in the early stage. Thereby, the smoothness of the finally obtained surface can be further improved.

The template substrate used at the beginning may be a flexible substrate. A polymer substrate may be used as the flexible substrate.

The template substrate used in the final stage may be a semiconductor substrate. For example, the semiconductor substrate may be a silicon substrate, or may be a silicon substrate on which a silicon oxide film is formed.

The first metal thin film may be a single metal film such as Au, Au/Ta/Cu/Ta/Au (gold/tantalum/copper/tantalum/gold), or Au/Al/Au (gold/aluminum). /gold) may also be included.

Another aspect of the invention is a bonding device. This bonding apparatus includes a first stage that holds a template substrate on which a first metal thin film is formed, a second stage that holds a support substrate, an activation processing section that activates the first metal thin film, and a bonding device. The chip holder that holds the target chip, the template substrate, and the chip are moved relative to each other, and the first metal thin film is transferred onto the surface of the chip multiple times by template stripping, and the chip is bonded to the support substrate. and a manipulator to do so.

Another aspect of the invention is a semiconductor device. A semiconductor device includes a support substrate, a metal thin film provided on the support substrate, a multilayer metal film provided on the metal thin film, and a chip provided on the multilayer metal film.

The interlayer gap of the multilayer metal film may be larger as it approaches the chip, and may be smaller as it is farther away from the chip.

Note that arbitrary combinations of the above components and expressions of the present invention converted between devices, methods, systems, etc. are also effective as aspects of the present invention.

According to the invention, template stripping can be used to form a smooth surface on a rough member.

FIGS. 1A to 1D are diagrams illustrating bonding using template stripping. FIGS. 2(a) and 2(b) are diagrams illustrating template stripping on an object with a rough surface. FIGS. 3A to 3C are diagrams illustrating some of the steps of the method for manufacturing a semiconductor device according to the embodiment. FIG. 3 is a diagram showing how the surface of the first metal thin film layer is smoothed by repeated transfers. FIG. 3 is a diagram showing the relationship (measurement results) between the number of transfers and surface roughness. FIGS. 6A and 6B are diagrams showing a cross-sectional SEM image and a cross-sectional TEM image after transfer. FIGS. 7(a) to 7(c) are diagrams showing the process of bonding the chip and the support substrate. FIG. 1 is a block diagram showing a joining device according to an embodiment.

(Embodiment)
DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described below based on preferred embodiments with reference to the drawings. Identical or equivalent components, members, and processes shown in each drawing are designated by the same reference numerals, and redundant explanations will be omitted as appropriate. Further, the embodiments are illustrative rather than limiting the invention, and all features and combinations thereof described in the embodiments are not necessarily essential to the invention.

Further, the dimensions of members in each drawing are shown enlarged or reduced as appropriate to facilitate understanding.

FIGS. 3A to 3C are diagrams illustrating some of the steps of the method for manufacturing a semiconductor device according to the embodiment. 3(a) to (c) show a step of smoothing the surface of a member (hereinafter referred to as a chip) 2 having a rough surface 12, specifically, an rms surface roughness of several nm to several hundred nm. has been done.

As shown in FIG. 3A, a first metal thin film 22 is formed on the surface of a template substrate 20 having a smooth surface (for example, rms surface roughness: 0.5 nm or less) (S100). As the template substrate 20, a substrate with weak adhesion to the metal to be transferred can be used, such as a silicon substrate on which an oxide film (SiO ₂ ) is formed or a polymer substrate (flexible substrate) such as polyethylene terephthalate (PET). . In order to reduce the adhesion force with the metal to be transferred, it is also possible to use a polymer substrate with an oxide film formed thereon.

Although the method for forming the first metal thin film 22 is not limited, for example, sputtering can be used. The thickness t1 of the first metal thin film 22 is, for example, about 0.1 nm to 300 nm. For the first metal thin film 22, a single metal film such as Au can be used.

Subsequently, as shown in FIG. 3(b), the surface of the first metal thin film 22 is activated (S102). An ion beam, low vacuum plasma, or atmospheric pressure plasma can be used for surface activation. In addition to activating the first metal thin film 22, the surface 12 of the chip 10 may also be activated. It is preferable to form a thin metal film (also referred to as a buffer layer) 11 with a thickness of several tens to hundreds of nanometers on the surface 12 of the chip 10 and activate it.

Then, as shown in FIG. 3C, the first metal thin film 22 is transferred onto the surface 12 of the chip 10 in a plurality of layers by template stripping, thereby forming a multilayer metal film 16 on the surface of the chip 10. In this example, the transfer is performed three times, but the number of transfers is not limited, and the transfer is performed so that the surface of the chip 10 obtains the required smoothness (for example, rms surface roughness: 0.5 nm or less). All you have to do is decide.

FIG. 4 is a diagram showing how the surface of the first metal thin film layer 14 is smoothed by repeated transfer. As the number of transfers increases, the amount of plastic deformation of the first metal thin film layer 14 becomes smaller.

FIG. 5 is a diagram showing the relationship (measurement results) between the number of transfers and surface roughness. This measurement result confirms that the surface smoothness improves with repeated transfers, but this result is not obvious and was confirmed for the first time through experiments conducted by the present inventors. Please note that.

FIGS. 6A and 6B are diagrams showing a cross-sectional SEM image and a cross-sectional TEM image of the device after transfer. In both figures, the bottom side shows the chip, and the top side shows the multilayer metal film 16 (5 layers). It can be seen that in the multilayer structure of the transferred bonding interface, voids are formed between the first metal thin film layers 14. The presence or absence of this gap and its size can be controlled by the material of the template substrate used and the stripping conditions (pressure, temperature).

Next, a process for bonding the chip 10, which has obtained a smooth surface through the processes shown in FIGS. 3(a) to 3(c), with a support substrate will be described.

FIGS. 7A to 7C are diagrams showing the process of bonding the chip 10 and the support substrate 30. FIG. As shown in FIG. 7A, a second metal thin film 32 made of a material that can be bonded to the first metal thin film 22 is formed on the surface of the support substrate 30 (S200). The material of the supporting substrate 30 is not limited and varies depending on the structure of the semiconductor device, but for example, a metallic heat dissipating substrate with high thermal conductivity or a ceramic substrate can be used. The second metal thin film 32 is preferably made of the same material as the first metal thin film 22 .

As shown in FIG. 7(b), the multilayer metal film 16 of the chip 10 and the second metal thin film 32 of the support substrate 30 are activated (S202). An ion beam or plasma can be used for surface activation.

Then, as shown in FIG. 7C, the chip 10 and the support substrate 30 are bonded together with the multilayer metal film 16 and the second metal thin film 32 in contact with each other, thereby forming the semiconductor device 40 (S204). The above is the joining process.

The semiconductor device 40 manufactured by the manufacturing method according to the embodiment has the following structure.

The semiconductor device 40 includes a support substrate 30, a second metal thin film 32 provided on the support substrate 30, a multilayer metal film 16 provided on the second metal thin film 32, and a chip 10 provided on the multilayer metal film 16. , is provided.

As shown in FIG. 6A, the interlayer gap of the multilayer metal film 16 increases as it approaches the chip 10, and decreases as it moves away from the chip 10.

Next, advantages of manufacturing the semiconductor device according to the embodiment will be explained.
First, by repeating transfer by template stripping, surfaces with large surface roughness can be smoothed by additional processing that actively utilizes plastic deformation of the active Au thin film and surface diffusion and grain boundary diffusion. . Specifically, a smooth surface with an rms surface roughness of 1 nm or less (for example, 0.5 nm) can be obtained. In this process, high temperature heat treatment or large pressure is not necessary.

According to this technique, a smooth surface can be formed without a CMP process, and a new low-temperature bonding process can be established.

Second, the chip 2 with a smooth surface can be bonded at low temperatures (room temperature to 150° C.) and low loads (200 MPa or less). This is because the technology that performs one transfer described with reference to Figure 2 requires large pressure (over 300 MPa) and high-temperature heat treatment (over 200°C). This is advantageous from the point of view.

(Joining device)
FIG. 8 is a block diagram showing a joining device according to an embodiment. The bonding apparatus 100 includes a first stage 102, a second stage 104, an activation processing section 106, a chip holder 108, and a manipulator 110.

The first stage 102 holds a template substrate 20 on which a first metal thin film 22 is formed. The second stage 104 holds the support substrate 30.

The activation processing unit 106 activates the surface of the first metal thin film 22 of the template substrate 20. The activation processing unit 106 may be a plasma generator or an ion beam generation source.

The chip holder 108 holds the chip 10 to be welded. Chip 10 has a rough surface 12, as described above. The activation processing unit 106 can also activate the surface 12 of the chip 10. A buffer layer 11, which is an Au thin film, may be formed on the surface 12 of the chip 10 by pretreatment.

A manipulator (also referred to as a handler) 110 moves the template substrate 20 and the chip 10 relatively, and transfers the first metal thin film 22 onto the surface of the chip 10 multiple times in an overlapping manner by template stripping. After the transfer is completed, the manipulator 110 joins the chip 10 to the support substrate 30. Furthermore, the activation processing unit 106 can also activate the surfaces of the chip 10 and the support substrate 30 before bonding.

The present invention has been described above based on the embodiments. Those skilled in the art will understand that this embodiment is merely an example, and that various modifications can be made to the combinations of these components and processing processes, and that such modifications are also within the scope of the present invention. be. Hereinafter, such modified examples will be explained.

(Modification 1)
The material of the first metal thin film 22 is not limited to Au, and other metals such as Cu (copper) and Ag (silver) may be used. Alternatively, a multilayer metal film such as Au/Ta/Cu/Ta/Au or Au/Al/Au may be used.

(Modification 2)
As described above, in the process of repeating transfer, the first metal thin film layer 14 must be plastically deformed to increase the actual contact area and adhere, so the elastic modulus of the template substrate is an important parameter. Therefore, a plurality of template substrates having different rigidities may be used in the process of multiple transfers. By optimizing the rigidity of the template substrate for each transfer, the amount of plastic deformation of each layer can be adjusted, and a smoother surface can be achieved.

In particular, among the plurality of template substrates, the elastic modulus of the template substrate used in the final stage may be relatively larger than the elastic modulus of the template substrate used in the early stage. For example, a flexible substrate may be used at the beginning (for example, the number of transfers is 1 to 8), and a semiconductor (Si) substrate on which an insulating film (SiO ₂ ) is formed at the end (for example, the last transfer) may be used. Thereby, the smoothness of the finally obtained surface can be further improved.

(Modification 3)
In the embodiment, a case has been described in which the surface of a semiconductor chip is smoothed, but this smoothing method can also be applied to smoothing the support substrate 30 and other members.

(Modification 4)
In FIG. 7, the second metal thin film 32 is formed on the surface of the support substrate 30, but the second metal thin film 32 may not be necessary depending on the type of material of the support substrate 30 (for example, in the case of Au). be.

Although the present invention has been described using specific terms based on the embodiments, the embodiments merely illustrate the principles and applications of the present invention, and the embodiments do not include the scope of the claims. Many modifications and changes in arrangement are possible without departing from the spirit of the present invention.

10 Chip 12 Surface 14 First metal thin film layer 16 Multilayer metal film 20 Template substrate 22 First metal thin film 30 Support substrate 32 Second metal thin film 100 Bonding device 102 First stage 104 Second stage 106 Activation processing section 108 Chip holder 110 manipulator

Claims (11)

forming a first metal thin film on the surface of a template substrate having a smooth surface;
Transferring the activated first metal thin film multiple times onto the surface of the member to be smoothed by template stripping to form a multilayer metal film on the surface of the member;
A manufacturing method characterized by comprising: forming a second metal thin film on the surface of a support substrate to which the member is to be joined;
bonding the member and the supporting substrate by bringing them into contact with each other in a state where the multilayer metal film and the second metal thin film are activated;
The manufacturing method according to claim 1, further comprising:. 2. The manufacturing method according to claim 1, further comprising the step of bringing the member and a supporting substrate to be joined into contact with each other in a state where the multilayer metal film is activated to join them. 4. The manufacturing method according to claim 1, wherein a plurality of template substrates having different rigidities are used in the process of multiple transfers. 5. The manufacturing method according to claim 4, wherein, among the plurality of template substrates, the elastic modulus of the template substrate used in the final stage is relatively larger than the elastic modulus of the template substrate used in the early stage. 6. The manufacturing method according to claim 5, wherein the template substrate used in the initial stage is a flexible substrate. 7. The manufacturing method according to claim 5, wherein the template substrate used in the final stage is a semiconductor substrate. 8. The manufacturing method according to claim 1, wherein the first metal thin film includes a single metal film or a metal multilayer film. 9. The manufacturing method according to claim 1, wherein the activation uses low vacuum plasma, atmospheric pressure plasma, or an ion beam. a first stage that holds a template substrate on which a first metal thin film is formed;
a second stage holding a support substrate;
an activation processing unit that activates the first metal thin film;
a chip holder that holds the chips to be welded;
a manipulator that moves the template substrate and the chip relatively, transfers the first metal thin film onto the surface of the chip multiple times in an overlapping manner by template stripping, and joins the chip to the support substrate;
A joining device comprising: a support substrate;
a metal thin film provided on the support substrate;
a multilayer metal film provided on the metal thin film;
a chip provided on the multilayer metal film;
Equipped with
A semiconductor device characterized in that the interlayer gap of the multilayer metal film is larger as it approaches the chip and becomes smaller as it is farther away from the chip.

Images