Method for permanently bonding wafers by a connecting layer by means of solid state diffusion or phase transformation

Abstract

A method for bonding of a first solid substrate to a second solid substrate which contains a first material with the following steps, especially the following sequence: formation or application of a function layer which contains a second material to the second solid substrate, making contact of the first solid substrate with the second solid substrate on the function layer, pressing together the solid substrates for forming a permanent bond between the first and second solid substrate, at least partially reinforced by solid diffusion and/or phase transformation of the first material with the second material, an increase of volume on the function layer being caused.

Claims

1. A method for bonding of a first solid substrate to a second solid substrate, the method comprising: forming or applying a function layer on the second solid substrate, the function layer containing a second material, contacting the first solid substrate with the function layer on the second solid substrate, the first solid substrate containing a first material, and pressing together the first and second solid substrates to form a permanent bond between the first solid substrate and the second solid substrate and at least partially reinforce the permanent bond by solid diffusion of the second material contained in the function layer into the first material contained in the first solid substrate such that the function layer is consumed by the first solid substrate, the second solid substrate, or a combination thereof, wherein the function layer has an average thickness in a range between 0.1 nm and 25 nm, before formation of the permanent bond between the first solid substrate and the second solid substrate, wherein the solid diffusion is induced on respective contact surfaces of the first and second solid substrates at an interface between the first and second solid substrates, and wherein gaps between the respective contact surfaces at the interface are closed by volume expansion of at least one of the respective contact surfaces caused by the solid diffusion and the step of pressing together the first and second solid substrates.

2. The method as claimed in claim 1, wherein the formation of the permanent bond takes place at a temperature between room temperature and 500 C.

3. The method as claimed in claim 2, wherein said temperature is between room temperature and 200 C.

4. The method as claimed in claim 2, wherein said temperature is between room temperature and 150 C.

5. The method as claimed in claim 2, wherein said temperature is between room temperature and 100 C.

6. The method as claimed in claim 2, wherein said temperature is between room temperature and 50 C.

7. The method as claimed in claim 2, wherein the formation of the permanent bond takes place during a maximum 12 days.

8. The method as claimed in claim 2, wherein the formation of the permanent bond takes place during a maximum 1 day.

9. The method as claimed in claim 2, wherein the formation of the permanent bond takes place during a maximum 1 hour.

10. The method as claimed in claim 2, wherein the formation of the permanent bond takes place during a maximum 15 minutes.

11. The method as claimed in claim 1, wherein the permanent bond has a bond strength of greater than 1.5 J/m.sup.2.

12. The method as claimed in claim 11, wherein the bond strength is greater than 2 J/m.sup.2.

13. The method as claimed in claim 11, wherein the bond strength is greater than 2.5 J/m.sup.2.

14. The method as claimed in claim 1, wherein, during the solid diffusion, a mixed material is formed having a molar volume that is greater than a molar volume of the second material and of the first material.

15. The method as claimed in claim 1, further comprising plasma activating surfaces of the solid substrates prior to and/or after the application/formation of the function layer.

16. The method as claimed in claim 1, wherein the solid diffusion is limited to a first surface layer of the first solid substrate having a maximum initial thickness less than 1 m.

17. The method as claimed in claim 16, wherein the maximum initial thickness is smaller than 100 nm.

18. The method as claimed in claim 16, wherein the maximum initial thickness is smaller than 10 nm.

19. The method as claimed in claim 16, wherein the maximum initial thickness is smaller than 1 nm.

20. The method as claimed in claim 1, wherein the step of pressing together the first and second solid substrates takes place at a pressure between 0.1 and 10 MPa.

21. The method as claimed in claim 1, wherein during forming of the permanent bond, a solubility boundary of the first material contained in the first solid substrate for the second material contained in the function layer is exceeded only slightly.

22. The method as claimed in claim 21, wherein, during formation of the permanent bond, a solubility boundary of the first material contained in the first solid substrate for the second material contained in the function layer is exceeded at no site of solid diffusion.

23. The method as claimed in claim 1, wherein the solid diffusion takes place at least predominantly as grain boundary diffusion.

24. The method as claimed in claim 1, wherein the second solid substrate contains the first material, and wherein the the function layer is consumer by the combination of the first solid substrate and the second solid substrate.

25. The method as claimed in claim 1, wherein the first material is a first metal and the second material is a second metal.

26. A method for bonding of a first solid substrate to a second solid substrate, the method comprising: forming or applying a function layer on the second solid substrate, the function layer containing a second material, contacting the first solid substrate with the function layer on the second solid substrate, the first solid substrate containing a first material, and pressing together the first and second solid substrates to form a permanent bond between the first solid substrate and the second solid substrate and at least partially reinforce the permanent bond by phase transformation of the first material contained in the first solid substrate with the second material contained in the function layer such that the function layer is consumed by the first solid substrate, the second solid substrate, or a combination thereof, wherein the function layer has an average thickness in a range between 0.1 nm and 25 nm, before formation of the permanent bond between the first solid substrate and the second solid substrate, wherein the phase transformation is induced on respective contact surfaces of the first and second solid substrates at an interface between the first and second solid substrates, and wherein gaps between the respective contact surfaces at the interface are closed by volume expansion of at least one of the respective contact surfaces caused by the phase transformation and the step of pressing together the first and second solid substrates.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) FIG. 1a shows a first and a second solid substrate of the present invention at the instant of the method of the present invention immediately after contact of the first solid substrate is made with the second solid substrate,

(2) FIG. 1b shows permanently joined solid substrates after carrying out the method of the present invention,

(3) FIG. 2a shows one step of the method of the present invention for formation/application of one function layer,

(4) FIGS. 2b to 2d show enlargements of the boundary surface between the second solid substrate and the function layer according to FIG. 2a,

(5) FIG. 3 shows one alternative step of the method of the present invention for formation/application of one function layer,

(6) FIGS. 3b to 3d show enlargements of the boundary surface between the second solid substrate and the function layer according to FIG. 3a,

(7) FIG. 4 shows one alternative step of the method of the present invention for formation/application of one function layer,

(8) FIGS. 4b to 4d show enlargements of the boundary surface between the second solid substrate and the function layer according to FIG. 4a,

(9) FIG. 5 shows one alternative step of the method of the present invention for formation/application of one function layer,

(10) FIGS. 5b to 5d show enlargements of the boundary surface between the second solid substrate and the function layer according to FIG. 5a.

(11) The same or equivalent features are identified with the same reference numbers in the figures.

DETAILED DESCRIPTION OF THE INVENTION

(12) The invention describes a method for a volume expansion of regions near the surface (proceeding from an effective contact surface 6 between the solid substrates 1, 2 which are to be bonded), specifically of a first surface layer 3 of the first solid substrate 1 and/or a second surface layer 4 of the second solid substrate 2 and/or a function layer 5 which is provided on a surface layer 3, 4.

(13) The volume expansion takes place by a solid reaction between the first material A and a second material B such that gaps 10 are closed between the contacted solid substrates 1, 2 along the effective contact surface 6 (interface 11) during an additional application of pressure in a bond process. The gaps 10 immediately after the solid substrates 1, 2 that have made contact according to FIG. 1a are still comparatively large due to their unevenness. In this way the effective contact surface 6 is accordingly small. In FIG. 1b a volume expansion (growth) has been caused by the solid reaction so that the gaps 10 have been distinctly reduced in size, in particular in the volume at least by a factor of 2, preferably at least by a factor of 5, even more preferably at least by a factor of 10. Accordingly the effective contact surface 6 has become larger.

(14) The solid state reaction takes place here between the second material B which is preferably present in the function layer 5 (>50% by volume) and the first material A of one of the solid substrates 1 and/or 2, preferably at least of the first, especially upper solid substrate 1.

(15) The function layer 5 can be produced using various methods which are described below, as a result of which an altered second surface side 4 (since it additionally has the function layer 5) with a new surface 4o is formed.

(16) According to the present invention it is also conceivable for the two solid substrates 1, 2 to consist of two different materials A.sub.1 and A.sub.2. If the second material B is chosen such that a volume expansion in a reaction/mixture with the two first materials A.sub.1 and A.sub.2 takes place, this description can be applied thereto and is covered by the invention. In one preferred case the materials A.sub.1, A.sub.2 and B are chosen to be single-phase, single-component systems. Alternatively, first materials A.sub.1 and A.sub.2 are identical, especially materials A and B being chosen from the group of metals.

(17) The surface layers 3 and/or 4 are those volumetric regions under the surfaces 3o, 4o of the substrates 1, 2 in which the solid reaction of the present invention at least largely proceeds. The surface layers 3, 4 have especially an average thickness D smaller than 1 m, preferably smaller than 100 nm, even more preferably smaller than 10 nm, most preferably smaller than 1 nm. If the second material B is made as polycrystalline material, the average thickness D of the surface layers 3, 4 is especially a maximum 50 times as large as the average grain diameter of the second material B (especially of the single crystals of second material B), more preferably a maximum 20 times as large, even more preferably a maximum 10 times as large, most preferably a maximum 5 times as large, most preferably of all a maximum twice as large. FIGS. 1a and 1b are thus shown highly enlarged.

(18) FIG. 1a shows the state of contact-making of the surfaces 3o, 4o before the solid reaction between the two materials A and B, while FIG. 1b shows the surface layers 3, 4 after completed volume expansion and the successful bonding process. The figures show the change of the volume and the gap closure which occurs with it along the interface 11 or along the effective contact surface 6 which becomes larger due to the gap closure. By enlarging the effective contact surface 6 and by distinctly reducing the average distance between the surfaces 3o, 4o, the bond force is greatly increased. Even more ideally the holes are completely closed so that at least later the bond interface can no longer be recognized.

(19) The second material B of the function layer is at least partially consumed by the solid reaction in the surface layer 3 and/or 4. The average thickness R of the function layer 5 is dependent on the most varied parameters (temperature, choice of materials A and B, bond pressure, time progression, diffusion rates). The materials A and B can be metals, plastics, ceramics or semiconductors, metals being preferred. The solid substrates 1, 2 can be made especially as wafers.

(20) The different embodiments of the present invention are now detailed. Here the materials A and B are selected as single-phase, single-component materials. The material A and or the material B each consist preferably of a single or uniform material. In the illustrated embodiments the second material B is applied only to the second solid substrate 2.

(21) One embodiment of the present invention consists in expanding the volume by dissolving the second material B in the first material A. By way of example, for this embodiment of the present invention the copper-tin system has been named. The substance for the first material A is the metal copper and the second material B is the metal tin.

(22) The expansion of the copper which contributes to closing of the gaps 10 in the interface 11 takes place by the formation of a copper mixed crystal C.

(23) A mixed crystal C is a crystalline phase which consists of at least two different materials, here the materials A and B which are completely miscible with one another within a concentration range. According to CuSn phase diagrams, copper at room temperature has a solubility for tin. The solubility increases as temperature rises and has a peak at roughly 850K. Conversely tin has a negligibly low solubility for copper up to the melting point. Based on the relationship between the volume and the concentration of each mixed crystal (Vegard's rule), in the simplest embodiment tin as the second material B is deposited on the surface layer 4 of the second solid substrate 2 (at least surface layer 4 of copper as the first material A) (FIG. 2a). The method parameters are chosen such that the second material B (tin) at this instant does not join the first material A (copper). The amount of tin is such that in a diffusion of the tin into the copper which takes place later (FIG. 2c) an intermetallic phase preferably never forms. In other words: The surface layer 4 is not saturated with tin at any point such that intermetallic phases can form. For the binary phase system CuSn according to the phase diagram at room temperature a molar concentration for Sn in Cu of roughly 0.01 (corresponds to roughly 14% by weight Sn) may not be exceeded. Thus the formation of the Cu.sub.3Sn phase is suppressed since the solubility limit of copper for the Sn has not yet been exceeded. For rising temperature the solubility of Sn in Cu becomes accordingly greater.

(24) The volume of the copper mixed crystal C very probably changes however by the absorption of the second material B (tin) into the first material A (copper). Since tin has a larger atomic radius than copper, the volume of the copper mixed crystal C rises with increasing tin content (FIG. 2d). The start of the diffusion process of tin into copper is controlled preferably based on process parameters, especially by a temperature increase since the diffusion constants depend explicitly on the temperature.

(25) In the embodiment presented here the permanent bonding takes place below 200 C. The temperature at which a noticeable diffusion of the tin into the copper begins, with the other process parameters which have been chosen in the present invention, is between room temperature (RT) and 200 C., more preferably between RT and 150 C., even more preferably between RT and 100 C., most preferably between RT and 50 C. For one skilled in this art it is clear that any parameter which can control the desired diffusion can be used to achieve the desired effect.

(26) Copper accepts tin by the controlled use of diffusion, thus increases its volume and can thus close the gaps 10 in the interface 11. For metals the high plasticity additionally promotes the process of closing of the gaps 10.

(27) According to the present invention the intention is moreover to prevent the solubility boundary of the first material A (copper) for the second material B (tin) from being exceeded so that the separation of intermetallic phases is prevented as extensively as possible, preferably completely. To the extent materials A and B are chosen which are completely miscible in the solid state, the solubility boundary of the present invention can remain ignored.

(28) The deposition of the second material B (tin) on the first material A (copper) is carried out in the present invention such that the solubility boundary of the copper for tin at the corresponding temperature is exceeded at as few sites as possible, more preferably at no site in the surface layer 4 (see FIG. 2b). The components (single crystals or multicrystals) of the first material A are shown schematically.

(29) The growth of the first material A in the form of the mixed material C is shown schematically in FIG. 2d. Intermediate spaces 7 (if present) between components of the first material A become smaller, so that the components try to expand in the direction of the contact surface 6. In doing so the mixed material C expands primarily in the region of the gaps 10 due to the pressure of the opposing solid substrate 1, which pressure is prevailing on the active contact surface 6.

(30) In order to largely prevent the formation of intermetallic phases, according to another version it is provided that the second material B (tin) as a function layer 5 is deposited not only on the surface 4o of the first material A (copper) (FIG. 2), before the actual dissolution process starts, but to be introduced into the surface layer 4 within the layer thickness d (FIG. 3) without allowing the tin to pass into solution with the copper.

(31) For this purpose the second material B (tin) will travel especially via grain boundary diffusion processes to greater depths of the polycrystalline first material A (copper), preferably will not yet pass into the volume of the grains, at most penetrate on the outer edge of the grains slightly into the depth and only in a decisive bonding process actually penetrate into the volume in order to cause the increase of the volume (FIG. 3). The polycrystalline microstructure of the first material A is polycrystalline, therefore consists of individual grains which are separated from one another by intermediate spaces 7 (here: grain boundaries). The intermediate spaces 7 for a polycrystalline microstructure are two-dimensional lattice structural defects into which atoms of different species can penetrate. Preferably the microstructure of at least one of the solid substrates 1, 2 which are to be bonded is produced such that the second material B (tin) is located not only on the surface 4o, but also in the surface layer 4 without dissolving in the copper. Thus the function layer 5 in this embodiment is at least partially identical to the surface layer 4.

(32) One version of the present invention therefore consists in using the difference between grain boundary diffusion and volumetric diffusion to convey the tin into the volumetric depth of the copper layer without obtaining concentration elevations in the copper grains (see especially FIG. 3). The process parameters here are chosen such that the grain boundary diffusion takes place before a volumetric diffusion since for the diffusing speciesin the exemplary embodiment tinit is much easier to advance into the extensive grain boundaries than through the narrow lattice of the bulk (of the copper grains). Here the consideration of the diffusion coefficient is decisive. The diffusion coefficient for the grain boundary diffusion for an intended temperature is larger than the diffusion coefficient for the corresponding volumetric diffusion. The grain boundary surface to grain volume ratio must also be considered here since at a higher ratio per unit of volume there are accordingly more grain boundaries. Especially preferably the aggregate state of the phase which is moving along the grain boundaries is liquid. Therefore according to the present invention materials are recommended which have a very low melting point. The diffusion rate of the liquid phase along the grain boundary is accordingly high.

(33) Thus the second material B here diffuses into the first material A not only on the boundary surface between the function layer 5 and the surface 4o before application of the function layer 5, due to the penetration of the surface layer 4 from more or less all sides of each component (copper grain) of the first material A.

(34) Another version consists in depositing a layer structure (FIG. 4). The components copper and tin are deposited in layers, preferably in alternation material A and B by conventional deposition methods. In this way there is an intermediate solution to the pure surface solution according to FIG. 2 and the mixed solution according to FIG. 3 in which there are several boundary surfaces for diffusion of the second material B into the first material A.

(35) According to another embodiment of the invention it is provided that the first material A (copper) and the second material B (tin) as microparticles and/or nanoparticles 8, 9 will be deposited from a solution on the surface 4o, therefore a mechanical alloy will be applied to the second solid substrate 2 (FIG. 5). For spherical particles at the known density of the copper and of the tin and the known average spherical radii of the copper and tin particles the required mixing ratio for the copper-tin mixed crystal can be exactly computed. Formulas which have been adapted accordingly can be used for particles with a different shape.

(36) In the bonding process in this embodiment preferably sinter bridges arise which weld the microparticles and/or nanoparticles 9 to one another into a sinter matrix 12. At the same time diffusion of the second material B into the sinter matrix 12 takes place. Since the microparticles and/or nanoparticles 8 of the second material B are present statistically uniformly distributed through the mechanical alloy in the sinter matrix 12 of the first material A, an optimum uniform distribution of the second material B over a volumetric region which can be sharply delineated is possible. The materials A and B are chosen according to the present invention such that the volume of the newly formed mixed crystal C (via the sinter process) is larger than the volume of the mechanical alloy prior to the bonding process.

(37) It must be considered here that mechanical alloys due to the microparticles and/or nanoparticles 8, 9 have a generally lower density than bulk materials since between the microparticles and/or nanoparticles 8, 9 there is a large amount of empty space which is closed only after by the sinter process. In the most optimum case the empty space is completely broken down. Preferably the structure after the bonding process is again a polycrystalline structure with a mixed crystal.

(38) According to the present invention there is also a preferred version in which copper and tin are directly deposited with the intended concentration onto the second solid substrate by PVD and/or CVD processes. In this way the mixed material C (mixed crystal) is directly produced. An overly large tin concentration would lead to intermetallic phases.

(39) In another embodiment the increase of the volume is accomplished by the formation of a completely new phase, therefore a phase transformation instead of diffusion (also conceivable in combination according to the present invention). The reaction of the second material B with the first material A leads to any compound AxBy whose molar volume is greater than the sum of the two molar volumes of the materials A and B. The phase transformation will proceed in a controlled manner by the choice of corresponding process parameters. Therefore the deposited second material B should only then preferably react only with the first material A if this is desirable, therefore during the bond process. The increase of the volume is therefore caused by a phase transformation.

(40) In one specific embodiment of the phase transformation the increase of volume is produced by a martensitic transformation. A martensitic transformation is a diffusion-less phase transformation which takes place by a shear mechanism process of the lattice. The shear mechanism process takes place only by a very rapid temperature drop. Another advantage of a martensite consists in the often extremely high dislocation densities which are caused by the shear processes of martensite formation. In the bond process the pores would be closed by means of the volume expansion during the martensitic transformation, but at the same time the structure would be extremely consolidated, i.e. its dislocation density would rise. The dislocation density can possibly be used in some martensites as an aid for a later recrystallization process.

(41) In another embodiment the increase of the volume during the bond process in the regions near the surface is caused by the spinodal decomposition of an alloy. A spinodal decomposition is a spontaneous separation of a homogeneous structure into at least two phases by a critical parameter being exceeded, preferably the critical temperature. The homogeneous structure was produced by a cooling process of a multicomponent system whose concentration is within the spinodals. It is known to one skilled in the art which alloys can decompose spinodally, and how these alloys have to be produced and heat-treated. In particular alloys with spinodal decomposition whose decomposition process leads to an expansion of volume are chosen.

REFERENCE NUMBER LIST

(42) 1 first solid substrate 2 second solid substrate 3 first surface layer 3o surface 4, 4 second surface layer 4o, 4o surface 5, 5 function layer 6 effective contact surface 7 intermediate spaces 8 microparticles and/or nanoparticles 9 microparticles and/or nanoparticles 10 gap 11 interface 12 sinter matrix A first material B second material C mixed material