Method and assembly for direct copper-to-copper joining

Abstract

The present invention relates to a method for copper-copper direct bonding, comprising the steps of: a) providing a first substrate with a first pure copper deposit having a bonding surface; b) providing a second substrate with a second pure copper deposit having a bonding surface; c) connecting the bonding surface of the first deposit with the bonding surface of the second deposit to obtain a connected deposit; d) converting the first and second deposits of the connected deposit to a connected and converted deposit, wherein: - the first and second deposits are formed by an electrochemical copper deposition process and have copper grains with a smaller grain size than the grain size after conversion in step d); - the connected and converted deposits have grains with a larger grain size than the grain size of the first and second deposits before conversion in step d), as well as assemblies and devices manufactured by this method.

Description

Technical Field

[0001] The present invention relates to a novel method for copper-to-copper direct bonding in the manufacture of assemblies. In particular, the present invention relates to the manufacture of assemblies that interconnect conductive copper structures in wafer-to-wafer processes, die-to-wafer processes, or die-to-die processes in the semiconductor industry.

Background Art

[0002] Modern electronic device manufacturers are faced with the requirement of further miniaturization and must pursue the trend of higher density electrical interconnects. Heterogeneous integration is considered an important technology to meet current and future requirements in the semiconductor industry. Heterogeneous integration involves integrating separately manufactured components into an assembly, thereby enhancing functionality and improving operating characteristics. Copper is a major interconnect material for individual components and their packaging. In this context, 3D stacking of components within a package can increase interconnect density, reduce form factor, and enhance efficiency.

[0003] The bonding of individual components is required in various processes such as wafer-to-wafer processes, die-to-wafer processes, and die-to-die processes. Conventional bonding techniques include copper bumps and copper pillars combined with solder materials, which are widely used in current packaging applications. Such copper pillars are usually manufactured by electrolytic copper deposition. However, in general processes, they are usually troubled by relatively large non-uniformities in the range of several micrometers, especially with regard to coplanarity and thickness variations. Solder materials, usually tin or tin-silver alloys, can correct these non-uniformities to some extent.

[0004] To increase interconnect density, the distance between copper bumps and pillars needs to be reduced. Conventional soldering techniques are sometimes unsuitable for fine-pitch applications because solder material can ooze out from the structure during joining, potentially causing electrical shorts. These problems can sometimes be overcome by directly forming copper-copper interconnects, eliminating the need for conventional solder caps at microbumps, and clearly reducing the pitch between bumps to values ​​of 20 μm or less. Therefore, several attempts have been proposed to improve joint formation, including ion beam surface activation, chemical mechanical polishing (CMP), surface passivation by alloying copper with other elements, metallic and organic finishes, and structural design.

[0005] Many applications involve temperature-sensitive devices, necessitating alternative bonding technologies for low temperatures. Hybrid bonding, involving direct copper-copper interconnects encapsulated in silicon dioxide, has been proposed as a promising alternative.

[0006] WO2020046677A1 describes a hybrid bonding process that, by its structure, provides a vertical stack of conductive layers encapsulated by a surrounding dielectric, without the use of solder or adhesives, wherein the dielectric has a shape and thermal expansion capability designed to expand the stack vertically over a precise recess distance at low temperatures to create a direct copper-copper bond at the bonding interface. In the first stage, the oxide surfaces of the inorganic dielectric are directly bonded, and in the second stage, the metal encapsulated by the surrounding dielectric is directly bonded. The copper-copper bond is based on surface diffusion of copper atoms that diffuse into the interface and form a permanent bond.

[0007] US9881888B2 discloses a Cu-Cu interconnection structure in which the average particle size of the copper post body is coarse, and the average particle size of the outer surface of the copper post is small. Then, a portion of the outer surface of the copper post is brought into contact with another copper surface, pressurized, and heated to achieve copper-to-copper fusion.

[0008] Chuan Seng Tan: "Recent progress in copper-based wafer bonding for 3-D ICs application," 2008 INTERNATIONAL CONFERENCE ON ELECTRONIC MATERIALS AND PACKAGING: (EMAP 2008), IEEE, October 22, 2008 (2008-10-22), pp. 45-48, discloses thermocompression bonding of metallic copper (also known as diffusion bonding) and its use in 3D stacking of ICs. In this bonding, blanket Cu films deposited on two oxide wafers fuse under appropriate bonding conditions, i.e., a temperature range of 300-400°C and a contact pressure of 226 kPa, to form a homogeneous layer.

[0009] However, the aforementioned methods cannot meet the future demands associated with miniaturization. Improvements in bond formation, bond strength, and ductility are still required. Since many applications involve temperature-sensitive devices, low annealing temperatures are still necessary to avoid component damage and loss of conductivity. [Prior art documents] [Patent Documents]

[0010] [Patent Document 1] WO2020046677A1 [Patent Document 2] US9881888B2 [Patent Document 3] EP3286358B1 [Non-patent literature]

[0011] [Non-Patent Document 1] Chuan Seng Tan: "Recent progress in copper-based wafer bonding for 3-D ICs application", 2008 INTERNATIONAL CONFERENCE ON ELECTRONIC MATERIALS AND PACKAGING: (EMAP 2008) IEEE, October 22, 2008 (2008-10-22), pp. 45-48 [Overview of the Initiative] [Problems that the invention aims to solve]

[0012] Therefore, the object of the present invention is to overcome the shortcomings of the prior art and to provide a means for improving joint formation and joint strength.

[0013] A further object of the present invention is to provide a means for lowering the annealing temperature during copper-to-copper direct bonding.

[0014] Another object of the present invention is to reduce the number of method steps and provide an improved bonding surface for the deposits to be bonded.

[0015] Another object of the present invention is to improve electromigration within the formed copper-copper junction. [Means for solving the problem]

[0016] These objectives are addressed by the present invention.

[0017] In one embodiment of the present invention, a method for copper-copper direct joining is: a) A step of preparing a first substrate comprising a first pure copper deposit having a bonding surface, preferably the first substrate being a wafer-shaped or die-shaped substrate. b) A step of preparing a second substrate comprising a second pure copper deposit having a bonding surface, preferably the second substrate being a wafer-shaped substrate or a die-shaped substrate. c) Connecting the bonding surface of the first deposit and the bonding surface of the second deposit to obtain a connected deposit; d) Converting the first deposit and the second deposit of the connected deposit to obtain a connected and converted deposit; comprising: - The first deposit and the second deposit are formed by an electrodeposited copper deposition process, and have copper grains with a particle size smaller than the particle size after conversion in step d), preferably, the grain size is a nanocrystalline size, more preferably, the particle size is less than 0.8 μm on average, more preferably, 0.01 μm to 0.70 μm, most preferably, 0.01 μm to 0.3 μm of copper grains; - The connected and converted deposit has grains with a particle size larger than the particle size before conversion in step d), preferably, the particle size is 0.1 μm to 13 μm on average, more preferably 1 to 10 μm; Step d) is carried out by applying an annealing process having an annealing temperature of 200 °C or lower, preferably 150 °C to 200 °C.

[0018] In another aspect of the present invention, the assembly i) - Connecting the bonding surface of the first pure copper deposit of the first substrate and the bonding surface of the second pure copper deposit of the second substrate; - Converting the first deposit and the second deposit of the connected deposit to obtain a connected and converted deposit; and comprises the connected and converted deposit obtained thereby; - The first deposit and the second deposit are formed by an electrodeposited copper deposition process, and have copper grains with a smaller particle size compared to the copper grains of the connected and converted deposit having a larger particle size after applying an annealing process having an annealing temperature of 200 °C or lower, preferably 150 °C to 200 °C. Preferably, the grains of the first deposit spread through the bonding surface of the first and second deposits to the second deposit, and vice versa.

[0019] In yet another aspect of the present invention, the device i) - Connecting the bonding surface of the first pure copper deposit of the first substrate to the bonding surface of the second pure copper deposit of the second substrate; - Converting the first deposit and the second deposit of the connected deposits into connected and converted deposits; Comprising the connected and converted deposits obtained thereby; - The first deposit and the second deposit are formed by an electrodeposited copper deposition process and have smaller grain sizes of copper grains compared to the copper grains of the connected and converted deposits having larger grain sizes after applying an annealing process having an annealing temperature of 200 °C or lower, preferably 150 °C to 200 °C; The first and second substrates are part of a microelectronic device, preferably a wafer or a die.

[0020] In particular, according to the present invention, additional method steps are avoided. For example, changing the bonding surface structure before the bonding process to prepare a smooth bonding surface, and additional steps for reducing the bonding temperature such as surface-activated bonding using ultra-high vacuum (UHV) conditions or the Cu nanorod method using a copper nanorod array as a bonding intermediate layer are prevented.

[0021] Since the complex step of inserting a specific structure into the bonding surface is avoided, the present invention helps to further smooth the bonding surface and reduce the need for a CMP process to prepare a uniform / smooth surface with a very small surface roughness. In this context, a smooth surface means that the surface roughness Ra is less than 300 nm, preferably less than 50 nm. Preferably, the surface roughness Ra is in the range of 5 nm to 200 nm, more preferably in the range of 10 nm to 50 nm, still more preferably in the range of 〖10 nm〗^40 nm, and / or the deviation of the average thickness of the deposit is 5 to 10%.

[0022] Since the metal deposits were formed by an electrochemical copper deposition process, the bonding surface of the metal deposits is already very smooth. However, if a surface roughness Ra of less than 10 nm is required, a short CMP process may be useful, and / or CMP may be used as a short cleaning process to prepare the bonding surface of the substrate. In particular, short CMP is used for, for example, the removal of copper oxide and / or fine dust-like particles.

[0023] Furthermore, since no clear interface (the area where the bonded surfaces come into contact with each other) is detected after the transformation process, the bond strength, bond formation, and electromigration performance are greatly improved. In other words, the smooth surface of the original bonded surface (which formed a clear interface before transformation) becomes indistinct or almost undetectable after transformation. Rather, because the sediment grains pass through the surface from both sides and grow to the other side, the original bonded surface becomes indistinct and fuses (at least some of it is interpenetrating with each other).

[0024] Further aspects of the present invention can be found in the dependent claims or in the following description.

[0025] The features of exemplary embodiments will become apparent to those skilled in the art by describing them in detail with reference to the attached drawings. [Brief explanation of the drawing]

[0026] [Figure 1] This diagram shows the formation of a direct copper-to-copper bond. [Figure 2] This is a diagram showing the simulation of conversion step d). [Figure 3] This figure shows the normalized particle sizes of copper alloys and pure copper deposits. [Figure 4] These are FIB SEM microscope images of pure copper deposited using different time and temperature parameters. [Figure 5a] This is a FIB SEM micrograph of the formation of a copper-copper direct bond between two wafers according to the present invention. [Figure 5b]This is a FIB SEM micrograph of the formation of a copper-copper direct bond between two wafers according to the present invention. [Figure 5c] This is a FIB SEM micrograph of the formation of a copper-copper direct bond between two wafers according to the present invention. [Figure 6a] This is a FIB SEM micrograph of the formation of a copper-copper direct bond between two wafers in a comparative example. [Figure 6b] This is a FIB SEM micrograph of the formation of a copper-copper direct bond between two wafers in a comparative example. [Figure 6c] This is a FIB SEM micrograph of the formation of a copper-copper direct bond between two wafers in a comparative example. [Modes for carrying out the invention]

[0027] Here, embodiments illustrated in the accompanying drawings will be given in detail. The effects and features of the exemplary embodiments, as well as methods for carrying them out, will be described with reference to the accompanying drawings. In the drawings, the same reference numerals represent the same elements, and redundant descriptions are omitted. As used herein, the terms "and / or" include any and all combinations of one or more of the enumerated items associated with the term. Furthermore, the use of "may" when describing embodiments of the present invention refers to "one or more embodiments of the present invention."

[0028] In the following description relating to embodiments of the present invention, singular terms may include plural forms unless the context explicitly indicates otherwise. For example, when “deposit” is used in the following description, it includes “multiple deposits.” Furthermore, “on average” means that the number is the average of several numbers, and a particular number may be higher or lower than the average.

[0029] In the context of the present invention, pure copper refers to the purity of a copper deposit, meaning that the copper content of the copper deposit is 99% by mass, preferably 99.9% by mass, and more preferably 99.99% by mass. That is, each pure copper deposit means a copper deposit with a copper content of at least 99% by mass, preferably at least 99.9% by mass, and more preferably at least 99.99% by mass. In particular, pure copper does not contain alloying metals or organic residues in the aforementioned mass percent.

[0030] The first and second substrates can be part of an assembly after the method of the present invention has been carried out, and / or can be part of a microelectronic device as wafer-like or die-like substrates processed by the present invention. The bonding of the bonded surfaces by the method of the present invention can be a single step in a wafer-to-wafer (W2W), die-to-wafer (D2W), or die-to-die (D2D) process.

[0031] The first and second substrates may independently further comprise additional metal deposits having different metallic compositions, such as copper alloys, to form a vertical stack. The additional metal deposits of the first substrate are connected to the first deposits, and the additional metal deposits of the second substrate are connected to the second deposits. Each additional metal deposit has its own coefficient of thermal expansion (COE). The first and second deposits, as well as the additional metal deposits, may be encapsulated by a dielectric material. The manufacture of these substrates is known in principle to those skilled in the art.

[0032] The first substrate may further comprise a first nonconductive material, for example, a dielectric material having a bonding surface and at least partially containing the first deposit, and the second substrate may further comprise a second nonconductive material, for example, a dielectric material having a bonding surface and at least partially containing the second deposit. Here, the bonding surface of the first and second deposits is below or at the same height as the bonding surface of the first and second nonconductive materials.

[0033] The first deposit of the first substrate and the second deposit of the second substrate can be independently selected from the group consisting of vias, bumps, pillars, and / or pads.

[0034] Prior art has focused on optimizing CMP, oxide-oxide bonding processes for connecting dielectric materials that contain copper vias, and improving the diffusion of copper atoms on the copper surface between the bonding surfaces of the substrates to be bonded. However, the present invention addresses different aspects of copper deposits before and after bonding, and improved conditions for the subsequent copper-copper bonding process in a hybrid bonding process sequence related to time and annealing temperature, from the perspective of copper particle size and purity of copper deposits in the electrolytic deposition process.

[0035] This invention is based primarily on two findings. First, the crystalline structure of the plated deposits should remain unchanged and preserved until the copper-copper bonding process is initiated. Second, the growth of grains initiated within the copper-copper bonding process (primarily step d)) results in grains penetrating the original uniform bonding surface in one deposit and into the other deposit. This causes the original uniform bonding surface to disappear or largely disappear. In these bonded and transformed deposits, the bonding strength is improved, and the reliability of the bonded (and transformed) deposits is greatly improved.

[0036] Our own experiments have shown that the timescale for maintaining small grains at room temperature depends on the electrolytic copper process and can be controlled by appropriately adjusting the deposition parameters. In particular, the purity of the copper deposit and the temperature used for copper deposition make it possible to maintain a microstructure similar to the initial morphology for a long period of time at room temperature. As a result, a timescale sufficient for introduction into industrial hybrid bonding process sequences should be obtained. Although not bound by theory, grain growth on the interface (the bonding surface of the deposits) preferably starts from the initial structure when a thermal load is applied, for example, during a transformation relaxed by an annealing process.

[0037] In particular, it was found that the annealing temperature of copper electrolytically deposited films may be affected by co-deposited impurities, especially depending on organic plating additives and plating conditions, and that the annealing temperature can be lowered considering the requirements of the bonding process flow and the resulting package quality.

[0038] The conversion step of the present invention is carried out particularly after an electrochemical copper deposition step for preparing a pure copper deposit, and especially after preparing a conductive copper structure as vias on a substrate. Preferably, the conversion step is carried out after the electrochemical copper deposition step without further heat treatment, for example, without an annealing step before applying the present invention. That is, the copper deposit is exposed only to ambient temperature (room temperature) or below after the electrolytic process.

[0039] Preferably, the period after steps a) and b) until steps c) and / or d) is carried out shall not exceed three weeks, more preferably not exceeding 120 hours, and even more preferably not exceeding 72 hours. Preferably, the substrates from steps a) and b) are stored at room temperature (in the range of 18°C ​​to 25°C), preferably 25°C or lower, and more preferably in the range of 5°C to 25°C during the above period.

[0040] According to the present invention, step c)—the step of connecting the bonding surface of the first deposit and the bonding surface of the second deposit to obtain a bonded deposit—is understood to mean, for example, placing the substrates facing each other and clamping the substrates together on a bonding chuck, thereby bringing at least a portion of the bonding surfaces of the deposits on the substrates into direct contact with each other.

[0041] If the first substrate comprises a first nonconductive material having a bonding surface and at least partially containing a first deposit, and the second substrate further comprises a second nonconductive material having a bonding surface and at least partially containing a second deposit, and the bonding surfaces of the first and second deposits are below (or partially below) or at the same height as the bonding surfaces of the first and second nonconductive materials, then step c) includes, for example, placing the substrates facing each other and clamping the substrates together on a bonding chuck to bring at least a portion of the bonding surfaces of the nonconductive materials of the substrates into direct contact with each other. This connection can be achieved by pressing the first and second bonding surfaces in the nonconductive materials of the first and second substrates with constant pressure. In this embodiment, the bonding surfaces of the first and second deposits on the first and second substrates are brought into contact with each other simultaneously with the bonding surfaces of the nonconductive materials, or at least partially into contact after the bonding surfaces of the nonconductive materials have been bonded.

[0042] The pressure can be applied with a contact force of 1 to 5,000 N. In one embodiment, the contact force in step c) can be 3,500 to 4,500 N when the substrate is in complete contact.

[0043] Step c) can be carried out without heating to a higher temperature, and is particularly carried out at a temperature below 100°C, preferably at room temperature (preferably in the range of 18 to 25°C).

[0044] The method according to the present invention can be used in the manufacture of assemblies and / or devices in packaging applications.

[0045] The present invention makes it possible to improve the bonding formation and bonding strength of copper deposits between substrates to be joined.

[0046] Grain growth was achieved at the bonding interface (the region where the bonding surfaces of the first and second deposits come into contact with each other). Grain growth is crucial for good and sufficient copper-copper bonding formation and high reliability. The growth of particle size (small particles resulting from electrochemical deposition as a starting point) is prevented, or at least delayed or slowed, until the conversion process is initiated.

[0047] The copper particles in the first and second deposits after the electrochemical copper deposition process are preferably nanocrystalline, meaning that the particle size is less than 1 μm, more preferably less than 0.8 μm on average, more preferably 0.01 μm to 0.70 μm, and most preferably 0.01 to 0.3 μm.

[0048] The copper particles in the connected and transformed sediments have a particle size of an average of 0.1 μm to 13 μm, more preferably 1 to 10 μm, and most preferably 1 μm to 5 μm. In principle, the particle size of the copper particles in the connected and transformed sediments is limited only by the thickness of the connected and transformed sediments (the length of the combined region of the first and second sediments).

[0049] After the consolidation step c), the particle sizes of the first and second sediments become part of the consolidated sediments, and during the transformation step d), the particle sizes grow, and the formed copper grains spread through the joining surface of the first and second sediments (also called the interface where both joining surfaces are in contact with each other) into the other sediment. That is, the grains of the first sediment spread into the second sediment, and vice versa. In this context, "vicinally" means that the particle size of the second sediment also increases, and the formed copper grains of the second sediment spread through the joining surface of the second sediment (again, also called the interface where both joining surfaces are in contact with each other) into the first sediment. As a result, the joining surface (forming the interface) which had a uniform tendency of the sediments before the transformation disappears, and the copper grains extend through the consolidated and transformed sediments (region), preferably at least 75% of the consolidated and transformed sediments (region), more preferably the entire consolidated and transformed sediments (region). For example, if the region of connected and transformed sediments has a maximum length of 5 μm (the combined length of the first and second sediments), the copper grains may preferably have a maximum length of 5 μm, for example, 4 to 5 μm.

[0050] Preferably, the smaller particle sizes of the first and second sediments prior to the conversion step are almost completely converted (meaning more than 90% of the smaller particle sizes) to the larger particle sizes of the connected sediments, while the larger particle sizes pass through the connected sediments.

[0051] Preferably, the connected and transformed sediments do not have, or at least partially have, a detectable joining surface (interface) between the original first sediment and the original second sediment. The original uniform and smooth surface becomes, for example, at least partially invisible under a FIB SEM microscope after transformation.

[0052] In one embodiment, the smaller particle sizes of the first and second sediments are almost completely converted (meaning more than 90% of the smaller particle sizes) to the larger particle sizes of the connected sediments, the larger particle sizes pass through the connected sediments, and the connected sediments have no detectable bonding surface (interface) between the original first and original second sediments after conversion step d).

[0053] In the context of the present invention, the absence of a detectable bonding surface (interface of both bonded surfaces after transformation) between the original first deposit and the original second deposit means that the bonding surface that existed in the first and second deposits after step a), b), or c) and before step d) can no longer be detected as a smooth / uniform surface. This is because the particle size of each deposit grows from nanocrystal size to an average particle size of 0.1 μm to 13 μm, causing each to penetrate into the other deposit. In other words, the bonding surface, or at least a portion of the bonding surfaces (also called the interface between these bonding surfaces), can no longer be detected as belonging to the original deposit. Instead, the original distinct surface, or at least a portion of the surface, has fused and penetrated into the other. This is in contrast to the prior art in which deposits are used in which the particle size cannot grow or whose particle size has already been transformed before the bonding or transformation process (e.g., as shown in the comparative example).

[0054] In this context, the absence of a “detectable” bonding surface (interface) means that the detection of a bonding surface, or at least a portion of the surface (having a homogeneous tendency), of the original copper deposit in step d) is not visible, or partially not visible / observable, as belonging to the original deposit after the transformation in step d) by known methods such as EBSD (electron backscatter diffraction) or FIB SEM microscopy.

[0055] The conversion step d) is carried out by applying an annealing step having an annealing temperature of 200°C or less, preferably 150°C to 200°C, and more preferably 150°C to less than 200°C. Preferably, the annealing time is 10 to 90 minutes.

[0056] For the D2W or D2D process, the annealing time is preferably 45 to 75 minutes, most preferably 60 minutes, and the annealing temperature is preferably 150°C to 200°C, more preferably less than 150°C to 200°C. For the W2W process, the annealing time is preferably 10 to 20 minutes, and the annealing temperature is preferably 150°C to 200°C, more preferably less than 150°C to 200°C.

[0057] In another embodiment, step c) is carried out together with step d), and the temperature used in step c) is the annealing temperature of step d). In this embodiment, the joining and transformation steps are carried out simultaneously by applying an annealing step having an annealing temperature of 200°C or less, more preferably 150°C to 200°C, and most preferably 150°C to less than 200°C. In this embodiment, the first and second deposits undergo thermal expansion, and the copper grains of the first and second deposits also begin to transform and spread, and the joining surfaces of the first and second deposits come into contact with each other, and as a result, the grown copper grains of the first and second deposits finally come into contact with each other, penetrate, and form a joined and transformed deposit. In a special embodiment, the temperature is gradually increased up to the above annealing temperature.

[0058] Thermal expansion may be supported by additional metallic deposits in the first and second substrates that are attached to the first and second deposits on the opposite side of the joint surface (but not to the joint surface). The additional metallic deposits may have two or more metallic compositions, for example, a copper alloy. Each additional metallic deposit also has its own coefficient of thermal expansion (COE). This also eases the joining step c), resulting in contact between the joint surfaces of the first and second deposits.

[0059] Preferably, the method of the present invention is a direct hybrid bonding method wherein a first substrate further comprises a first nonconductive material, for example, a dielectric material having a bonding surface and at least partially containing a first deposit, and a second substrate further comprises a second nonconductive material, for example, a dielectric material having a bonding surface and at least partially containing a second deposit, wherein the bonding surfaces of the first and second deposits are below or equivalent to the plane of the bonding surfaces of the first and second nonconductive materials. In this embodiment, the bonding surfaces of the nonconductive materials are first connected (when the bonding surfaces of the first and second deposits are below) by, for example, arranging the substrates facing each other and clamping the substrates together on a bonding chuck, thereby bringing at least a portion of the bonding surfaces of the nonconductive materials of the substrates into direct contact with each other, and then the bonding surfaces of the deposits and step d) are performed. When the bonding surfaces of the first and second deposits are at the same height as the bonding surfaces of the nonconductive materials, both bonding steps (connection steps) are performed simultaneously. Preferably, steps c) and d) are carried out together, where, during bonding, the bonding surface of the substrate (of the nonconductive material and deposits) is heated from room temperature to the annealing temperature described above. Due to the heating, step d) is carried out almost simultaneously or is started when the annealing temperature is reached. In principle, the process flow of direct hybrid bonding includes electrolytic copper deposition, CMP, and two bonding steps in step c). Thus, a first bonding of the dielectric material and a second or simultaneous bonding of the first and second deposits are performed. Such hybrid bonding methods are known in principle to those skilled in the art and are described, for example, in WO2020046677A1. The nonconductive material can be a material mainly composed of polymers such as SiO2, SiCN, SiN, or polyimide (PI). Since the copper deposits are formed in the electrochemical copper deposition step, the bonding surface of the metal deposits is already very smooth (preferably having a surface roughness Ra of 10 nm to 30 nm). However, if a surface roughness Ra of less than 10 nm is required, and / or if cleaning is necessary, a short CMP process may be effective.

[0060] Alternatively, step c) can be carried out without heating to a higher temperature, particularly at a temperature below 100°C, preferably at room temperature (preferably in the range of 18-25°C), and step d) is preferably carried out as the step immediately following.

[0061] Any aqueous electrochemical copper deposition bath can be used, yielding first and second pure copper deposits having small-particle, preferably nanocrystalline, copper grains. The aqueous acidic copper plating bath is operated in the method according to the present invention by applying an electric current to the substrate and at least one anode, preferably in a temperature range of 15°C to 40°C, more preferably in a temperature range of 20°C to 35°C.

[0062] As explained at the beginning, the most important aspect of the present invention is that the small particle size of the first and second deposits must be maintained until step d) of the method of the present invention is commenced. Preferably, after the electrochemical copper deposition step of preparing the first and second copper deposits, and before the transformation in step d), no other transformation steps that may change the particle size of the copper grains in the first and second deposits are performed, preferably no annealing steps or any other steps including heating are performed, for example, no annealing steps with an annealing temperature above 100°C are performed, preferably no annealing steps of 100°C to 200°C for 1 hour are performed. Preferably, the period after steps a) and b) and before steps c) and / or d) are performed is not more than three weeks, more preferably not more than 120 hours, and even more preferably not more than 72 hours. Preferably, the substrates from steps a) and b) are stored at room temperature (in the range of 18°C ​​to 25°C), preferably 25°C or lower, more preferably in the range of 5°C to 25°C during the above period.

[0063] Furthermore, surface modification of the joint surface of the first and second deposits by an additional treatment step to roughen the joint surface after the electrochemical copper deposition step and / or before the bonding step c) is not required in the present invention, and therefore is not applied to the method of the present invention and is excluded from the method of the present invention. In particular, the joint surface of the first deposit and the joint surface of the second deposit have a smooth surface without having any generated nanotexture structure. These additional structures result in a surface roughness Ra of several hundred nanometers. In other words, nanostructured surfaces as nanotexture surfaces within the copper surface, for example, by pickling with very dilute acid, etching, polishing, or other methods, are not formed on the joint surface of the first and second deposits.

[0064] However, if necessary, the surface roughness of the bonded surface after steps a) and / or b) can be further reduced by CMP (chemical mechanical polishing). A shorter CMP time can be used for cleaning the bonded surface.

[0065] The electrochemical copper deposition process can be part of a process to prepare structures on first and second substrates in the form of first and second deposits, and the first and second deposits can be independently selected from the group consisting of vias, bumps, pillars, and / or pads. Thus, in principle, any electrochemical copper deposition process can be used as long as it results in good filling performance of the structure to be filled and preferably a smooth bonding surface of the first and second deposits. The process of preparing the structure is known in principle to those skilled in the art and may be, for example, a damascene process or a process of filling a structured resist with copper.

[0066] The electrochemical copper deposition process preferably includes using an aqueous acidic copper electrolytic bath containing a copper ion source, a sulfur-containing brightener compound, a suppressor compound, a leveler selected from the group consisting of a guanidine-containing compound, a urea-containing compound, an imidazole-containing compound, and a pyridine-containing compound, and a halide ion. The bath preferably has a pH value of less than 2, more preferably less than 1. All compounds can be purchased as products from Atotech Deutschland GmbH.

[0067] The copper ion source is preferably selected from the group consisting of copper sulfate and alkylsulfonates such as copper methanesulfonate. The copper ion concentration in the aqueous acidic copper plating bath is preferably in the range of 4 g / L to 90 g / L, more preferably 10 g / L to 70 g / L, and even more preferably 30 g / L to 65 g / L.

[0068] The acid is preferably selected from the group including sulfuric acid, fluoroboric acid, phosphoric acid, and methanesulfonic acid, and is added at a concentration of preferably 10 g / L to 400 g / L, more preferably 20 g / L to 300 g / L, and even more preferably 30 g / L to 300 g / L.

[0069] The sulfur-containing brightener compound is preferably selected from the group consisting of 3-(benzthiazolyl-2-thio)-propylsulfonic acid, 3-mercaptopropane-1-sulfonic acid, ethylenedithiodipropylsulfonic acid, 3-N,N-dimethylaminodithiocarbamoyl-1-propanesulfonic acid, and 3,3'-dithiobis-1-propanesulfonic acid (SPS) or its salts, with dithiobis-1-propanesulfonic acid (SPS) being preferred. The (total) concentration of all optionally present accelerator-brightener additives in the aqueous acidic copper bath composition is preferably in the range of 0.01 mg / L to 100 mg / L, more preferably in the range of 0.05 mg / L to 20 mg / L, and even more preferably in the range of 0.1 to 10 mg / L.

[0070] The suppressor compound is preferably selected from the group consisting of polyethylene glycol (PEG), polypropylene glycol, poly(ethylene glycol-lane-propylene glycol), polyethylene glycol / polypropylene glycol copolymer, poly(ethylene glycol)-block-poly(propylene glycol)-block-poly-(ethylene glycol), and poly(propylene glycol)-block-poly(ethylene glycol)-block-poly(propylene glycol), preferably polyethylene glycol or polyethylene glycol / polypropylene glycol copolymer (PEG / PPG). Preferably, the molecular weight (MW [g / mol]) of the suppressor is 4,000 to 8,000. The (total) concentration of the arbitrary carrier-suppressor additive is preferably in the range of 0.005 g / L to 20 g / L, more preferably 0.01 g / L to 20 g / L, and even more preferably 0.01 g / L to 5 g / L.

[0071] The guanidine-containing compounds are polymeric compounds having repeating units of guanidine residues and divalent residues. Examples are disclosed in EP3286358B1, incorporated by reference, and can be purchased from Atotech Deutschland GmbH. The concentration of the guanidine compound in the aqueous acidic copper plating bath is in the range of 0.01 mg / L to 1000 mg / L, more preferably 0.1 mg / L to 100 mg / L, and even more preferably 0.5 mg / L to 50 mg / L.

[0072] The halide ions are preferably selected from Cl, Br, and / or I. In the case of chlorides, the halide ions are preferably added in a concentration range of 10 to 200 mg / L, more preferably 20 to 80 mg / L, and even more preferably 30 to 60 mg / L. If another type of halide ion is selected, the molar amount of each is preferably selected, and if multiple different halides are selected, the total molar amount of each is selected.

[0073] Preferably, the electrochemical copper deposition process is carried out by applying a current density of 1 to 2 ASD to obtain a first deposit and a second deposit, and direct current (DC) plating and pulsed plating can be used. In one embodiment, the electrochemical copper deposition process is carried out by pulsed plating, for example, by forward and reverse pulses, or forward pulses and off-pulses with varying currents and durations.

[0074] Next, the present invention will be described with reference to the following drawings and non-limiting embodiments. [Examples]

[0075] Figure 1 shows the formation of a copper-copper direct bond according to the prior art (a) and a copper-copper direct bond according to the present invention (b).

[0076] It is known that electrochemically deposited copper using prior art also initially consists of fine grains after deposition, but readily recrystallizes with temperature and time to form larger grains. Grains that do not grow further are thought to lead to the formation of interfaces between the bonded components (Figure 1a). In contrast, when bonding is performed with grain sizes close to the initial state after deposition, the interface disappears, while grain growth is observed on the bonding surface (interface) of the first and second deposits (Figure 1b).

[0077] Figures 2 to 4 below support the findings in Figure 1, which can be applied to direct copper-copper bonding and lead to improved bonding strength and conductivity.

[0078] Figure 2 simulates conversion step d) of the present invention, in which one copper layer (first deposit) is annealed. Figure 2 shows the effect of annealing temperature over time on the average particle size of a 4.5 μm thick copper deposit (having approximately 2,500 ppm of organic impurities on the substrate (comparative example) and a pure copper deposit on the substrate (example of the present invention)) after applying annealing steps of 1 hour at different temperatures.

[0079] The annealing temperature of copper deposits can be adjusted by co-deposition of other elements, both inorganic and organic. Co-deposition of inorganic elements, particularly other metals, forms alloys, resulting in a decrease in conductivity compared to pure copper. Organic impurities are usually due to the co-deposition of organic plating additives. It has been shown that deposits of considerably high purity can be obtained by appropriately designing the additives. This method was used to lower the temperature required for annealing copper materials. The particle size of various layers prepared by electrolytic deposition was measured by EBSD after annealing at different temperatures for 1 hour (Figure 2a). Deposits prepared from electrolyte 1 required significantly higher temperatures to obtain the final particle size compared to electrolyte 2. Both electrolytes showed similar particle sizes after high-temperature annealing on the order of deposit thickness (4.5 μm). Dynamic secondary ion mass spectrometry (DSIMS) measurements of the amount of each co-deposited impurity showed higher values ​​in electrolyte 1 (see Table 1 below). Differential scanning calorimetry (DSC) revealed an exothermic peak, further confirming that copper with high impurity levels requires high temperatures for crystallization (see Table 1 below). Electrolyte 1 was based on conventional plating organic additives that tend to cause significant co-deposition, while electrolyte 2 was composed of optimized additives, which made it possible to obtain a high-purity final layer.

[0080] It was argued that the upper limit for copper-copper junction formation in hybrid junction applications is 200°C. Therefore, copper materials should reach their maximum particle size at such temperatures. Figure 2a shows that deposits from both electrolytes eventually reach a particle size of approximately 3.5 μm, close to the thickness of the deposit. However, electrolyte 1 requires temperatures above 300°C. In contrast, electrolyte 2 produces copper that reaches its maximum particle size at approximately 200°C. The average particle size, including twins, reached its maximum within the margin of error in electrolyte 1 at 200°C, but remained at 1.1 μm even at 230°C. The corresponding EBDS image is shown in Figure 2b. The comparison of particle sizes of electrolytes 1 and 2 in this temperature range confirms the potential suitability of electrolyte 2.

[0081] Figure 2a directly shows the results of EBSD measurements of the average particle size, including twins, of 4.5 μm thick copper deposits obtained from electrolyte 1 (gray) and electrolyte 2 (black), after annealing at different temperatures for 1 hour (top). Examples of corresponding EBSD images for electrolyte 2 after annealing at 110°C and 400°C for 1 hour, respectively (bottom).

[0082] Figure 2b directly shows EBSD images of 4.5 μm thick copper deposits obtained from electrolyte 1 (top) and electrolyte 2 (bottom) after annealing at 230°C for 1 hour.

[0083] The deposit prepared from electrolyte 1 (Comparative Example - (Comp.Ex)) required significantly higher temperatures to obtain the final particle size compared to electrolyte 2 (Example 1 of the present invention - (Inv.Ex1)). Both electrolytes produced similar particle sizes after high-temperature annealing, on the order of deposit thickness (4.5 μm). Dynamic secondary ion mass spectrometry (DSIMS) measurements of the amounts of co-deposited impurities revealed higher values ​​in electrolyte 1 (Table 1). Differential scanning calorimetry (DSC) confirmed exothermic peaks, further supporting the idea that copper with high impurity levels requires high temperatures for crystallization (Table 1). Electrolyte 1 was based on conventional plating organic additives prone to significant co-deposition, while electrolyte 2 was composed of optimized additives, enabling the acquisition of a high-purity final layer.

[0084] [Table 1]

[0085] Electrolyte 1 (Comp.Ex): Sulfur-containing Brightener SPS 2mg / L Suppressor compound PEG (MW 6000) 300 mg / L Leveler PEI (MW 600) 0.1 mg / L Deposition using 5ASD, temperature 25℃ Annealed at over 200°C after deposition. Electrolyte 2 (Inv.Ex): Sulfur-containing Brightener SPS 4mg / L Suppressor compound PEG / PPG copolymer (MW 6000) 10 mL / L Guanidine-containing compound: Atotech® Spherolyte Leveler DB 20 mL / L Deposition using 1ASD, temperature 25℃ After deposition in step d), anneal at 200°C for 1 hour.

[0086] Figure 3 shows the normalized grain size of a copper alloy (comparative example) and a pure copper deposit (example of the present invention), obtained by investigating one copper layer (first deposit) over time at room temperature. Figure 3 shows the normalized grain size of deposits prepared from electrolyte 2 using parameter sets 1 (5ASD DC, 25°C) (gray) and 2 (1ASD DC, 25°C) (black) as a function of time at room temperature (top). Examples of corresponding microstructures immediately after deposition (left), 12 hours later (middle), and 24 hours later (right).

[0087] All process steps prior to step d), particularly the deposition process itself, the joining process c), and, for example, the CMP process, are carried out below room temperature (ambient temperature). Therefore, it is necessary to maintain a morphology similar to the initial structure after deposition for as long as possible and to ensure sufficient time to carry out all necessary process steps before the copper-copper joining in step d) at high temperatures.

[0088] To investigate the timescale of microstructural changes, the particle size of sediments prepared from electrolyte 2 (described above) was measured at different time intervals at room temperature (Figure 3). Grain growth at room temperature is commonly referred to as self-annealing. The time to obtain the maximum particle size can be varied by the process parameters of electrodeposition. Using parameter 1, the maximum particle size was obtained after approximately 9 hours. This time may not be sufficient for implementation in production sequences. In contrast, sediments prepared using parameter 2 required more than 24 hours at room temperature before no further grain growth was observed. This time is much longer and therefore more suitable for current industrial production process flows.

[0089] Figure 4 simulates the transformation of sediments during the joining process and shows different sediments investigated over time at different temperatures. Figure 4 shows FIB SEM micrographs of pure copper deposits deposited under different time and temperature parameters, and here, the FIB SEM micrograph of copper deposited from electrolyte 2 under parameter 2 (1ASD DC, 25°C) is shown below. Figure 4a) A single 4.5 μm layer after deposition (simulating the first deposit on the first substrate). Figure 4b) A single 4.5 μm layer after 24 hours at room temperature (simulating the first deposit on the first substrate). Figure 4c) Two consecutively deposited 4.5 μm layers after annealing at 200°C for 1 hour, wherein the second layer is a layer deposited after self-annealing of the first layer at room temperature for 24 hours (simulating the transformation of the first deposit on the first substrate and the second deposit on the second substrate according to a non-inventive example). Figure 4d) Two consecutively deposited 4.5 μm layers after annealing at 200°C for 1 hour, wherein the second layer was deposited directly after the deposition of the first layer (simulating the transformation of the first deposit on the first substrate and the second deposit on the second substrate according to the present invention).

[0090] The ability of a deposit with a small initial microstructure to grow further through the interface (the bonding surface where the first and second deposits are attached) was investigated using two consecutive electrochemical deposition processes with different annealing conditions in between. A first deposit with a thickness of 4.5 μm was prepared from electrolyte 2 with process parameter 2. Initially, small particles were observed (Figure 4a), and significant growth occurred within 24 hours at room temperature as self-annealing (Figure 4b). A second electrochemical deposition process was performed under the same conditions on the initial and self-annealed microstructures. After the deposition of the two layers, the entire stack was annealed at 200°C for 1 hour to approximate the thermal load conditions during copper-copper junction. When the second layer was deposited on the self-annealed first layer, an interface was observed after annealing (Figure 4c - non-inventive). In contrast, when the second layer was deposited on the first layer, which retained its initial microstructure, no interface or grain growth was detected throughout the entire deposit consisting of the two consecutively deposited layers (Figure 4d).

[0091] Figure 5 shows FIB SEM micrographs of the formation of copper-copper direct junctions in wafer pairs according to the present invention and comparative examples.

[0092] Four 300 mm DBI5 wafers are prepared as two first substrates and two second substrates according to steps a) and b) of the present invention, wherein the first substrates have a bonding surface and further comprise a first nonconductive material for containing the first deposit, and the second substrates have a bonding surface and further comprise a second nonconductive material for containing the second deposit. Both substrates are made from the same nonconductive material, and the first and second copper deposits are formed by the same electrochemical copper deposition process. The electrochemical copper deposition process was carried out using electrolyte 2 (Example 1 of the present invention). All steps were performed within 24 hours.

[0093] Examples of the present invention Sixty minutes after electrochemical copper deposition, the two wafers (the first and second substrates) were treated with a short CMP process. As described in the overview, the CMP process was performed for a short time primarily to clean the surface of the copper deposits. However, this reduced polishing time and waste. Subsequently, the two wafers were joined at ambient temperature (step c) and annealed at below 200°C for 60 minutes according to step d).

[0094] Figure 5a shows a FIB SEM micrograph of the obtained copper-copper junction wafer pair, with a magnitude of 2k.

[0095] This photograph shows that connected and transformed sediments are formed by grains passing through the first sediment and entering the second sediment, and vice versa, and that a clear interface does not disappear between the first and second sediments, but rather at least a portion of it becomes invisible to the naked eye.

[0096] Figure 5b shows another FIB SEM micrograph of the copper-copper junction wafer pair at a size of 20k, rotated 90° from Figure 5a, and Figure 5c shows a FIB SEM micrograph of the same wafer pair at a size of 50k.

[0097] The transformed sediments do not have a clearly detectable interface between the original first sediment and the original second sediment. This means that during transformation step d), the particle size grows, and the formed copper grains pass through the interface between the first and second sediments and spread into the other sediment.

[0098] Comparative Example Sixty minutes after electrolytic copper deposition, the wafers (first and second substrates) were treated in an annealing process at 200°C for 60 minutes to induce grain growth through heat treatment. Following this process, the two wafers (first and second substrates) were treated in a short CMP process. Subsequently, the two wafers were joined together and annealed again at 200°C for 60 minutes according to steps c) and d).

[0099] Figure 6a shows a FIB SEM micrograph of the obtained copper-copper junction wafer pair, which has a size of 2k.

[0100] The formed connected sediments exhibit a clearly visible interface between the first and second sediments, and it can be seen that there are no areas on the surface of one sediment that are fused to the other surface.

[0101] Figure 6b shows another FIB SEM micrograph of the copper-copper junction wafer pair at a size of 20k, viewed by rotating Figure 6a by 90°, and Figure 6c shows a FIB SEM micrograph of the same wafer pair at 50k.

[0102] The formed sediments have a visible bonding surface (interface) between the original first sediment and the original second sediment, and it can be seen that the grown grains have not penetrated the bonding surface of the sediments; in other words, there are no grains that have extended across the entire area of ​​the first and second sediments.

[0103] In other words, during the first heat treatment (the first annealing step before step c), the transformation occurred and was completed within each deposit (the first and second deposits). The further annealing step d) could not initiate further grain size growth. In particular, the grain size did not increase further, and the copper grains did not spread through the joining surface of the first and second deposits into the other deposit. This is because grain size growth had already been completed in the first annealing step.

Claims

1. A method for direct copper-to-copper joining, a) A step of preparing a first substrate having a bonding surface and comprising a first copper deposit having a copper content of at least 99.9% by mass, b) A step of preparing a second substrate having a bonding surface and comprising a second copper deposit having a copper content of at least 99.9% by mass, c) A step of applying pressure with a contact force of 1 to 5,000 N to connect the joining surface of the first sediment and the joining surface of the second sediment, thereby obtaining a joined sediment, d) A step of converting the first and second sediments of the connected sediments to form a connected and converted sediment. Includes, - The first and second deposits are formed by an electrochemical copper deposition process and have copper grains with a particle size smaller than the particle size after conversion in step d), - The connected and transformed sediments have particles with a particle size larger than the particle size of the first and second sediments before transformation in step d), and step d) is carried out by applying an annealing step having an annealing temperature of 200°C or less. The particles of the first and second sediments having smaller particle sizes are converted to larger particle sizes of the connected sediments by more than 90% of the smaller particle sizes, and the larger particle sizes pass through the connected sediments. A method wherein the connected sediments do not have, or at least partially have, a detectable joining surface (interface) between the original first sediment and the original second sediment after conversion step d) in a 2k, 20k, or 50k FIB SEM microscope.

2. The method according to claim 1, wherein the annealing process is carried out for a period of 10 to 90 minutes.

3. The method according to claim 1 or 2, wherein the surface roughness Ra of the bonding surface of the first and second deposits is less than 200 nm.

4. The method according to any one of claims 1 to 3, wherein, after the electrochemical copper deposition step and before the connection of step c), a conversion step to change the particle size of the copper particles in the first deposit and the second deposit is not performed.

5. The method according to any one of claims 1 to 4, wherein the method is a direct hybrid bonding method, the first substrate further comprises a first nonconductive material having a bonding surface and at least partially containing the first deposit, the second substrate further comprises a second nonconductive material having a bonding surface and at least partially containing the second deposit, and the bonding surface of the first and second deposits is below the surface of the bonding surface of the first and second nonconductive materials.

6. The method according to any one of claims 1 to 5, wherein the electrochemical copper deposition step includes the use of an aqueous acidic copper deposition bath operated in a temperature range of 15°C to 40°C.

7. The method according to any one of claims 1 to 6, wherein the electrochemical copper deposition step includes using a copper ion source, a leveler selected from the group consisting of a sulfur-containing brightener compound, a suppressor compound, a guanidine-containing compound, a urea-containing compound, an imidazole-containing compound, and a pyridine-containing compound, and an acidic copper deposition bath containing halide ions.

8. The method according to any one of claims 1 to 7, wherein the electrochemical copper deposition step is carried out by applying a current density of 1 to 2 ASD to obtain the first deposit and / or the second deposit.

9. The method according to any one of claims 1 to 8, wherein the joining surface of the first deposit and the joining surface of the second deposit have smooth surfaces, and the smooth surfaces have a surface roughness Ra of less than 200 nm.

10. The method according to any one of claims 1 to 9, wherein no surface modification of the joining surface of the first deposit and the second deposit is applied after the electrochemical copper deposition step and / or before the joining step c).

11. i) An assembly comprising connected and transformed sediments, wherein the sediments are - By applying pressure with a contact force of 1 to 5,000 N, the bonding surface of the first copper deposit having at least 99.9 mass% copper content on the first substrate and the second bonding surface of the second copper deposit having at least 99.9 mass% copper content on the second substrate, and - A process of transforming the first and second sediments of the connected sediments into connected and transformed sediments. Obtained by, - The first and second deposits are formed by an electrochemical copper deposition process, and after applying an annealing process having an annealing temperature of 200°C or less, they have copper particles with smaller particle sizes compared to the copper particles of the connected and converted deposits which have larger particle sizes, and more than 90% of the particles of the first and second deposits which have smaller particle sizes are converted to the larger particle sizes of the connected deposits, and the larger particle sizes pass through the connected deposits. The connected deposits, in a 2k, 20k, or 50k FIB SEM microscope, do not have, or at least partially have, a detectable joining surface (interface) between the original first deposit and the original second deposit after conversion step d). assembly.

12. i) A device comprising an assembly having connected and transformed deposits, wherein the deposits are, - By applying pressure with a contact force of 1 to 5,000 N, the bonding surface of the first copper deposit having at least 99.9 mass% copper content on the first substrate and the second bonding surface of the second copper deposit having at least 99.9 mass% copper content on the second substrate, and - A process of transforming the first and second sediments of the connected sediments into connected and transformed sediments. Obtained by, - The first and second deposits are formed by an electrochemical copper deposition process and have copper particles having a smaller particle size compared to the copper particles of the connected and transformed deposits having a larger particle size after applying an annealing process having an annealing temperature of 200°C or less. - The particles of the first and second sediments having smaller particle sizes are converted to larger particle sizes of the connected sediments by more than 90% of the smaller particle sizes, and the larger particle sizes pass through the connected sediments. The first and second substrates are microelectronic devices, A device wherein the connected deposits do not, or at least partially, have a detectable bonding surface (interface) between the original first deposit and the original second deposit after conversion step d) in a 2k, 20k, or 50k FIB SEM microscope.

Images