Chip transfer method

Abstract

The invention relates to a method for transferring dies from a die block arranged on a support substrate onto a recipient substrate, the method in fact comprising: - forming a substrate (10), called pseudo-donor substrate, comprising a support substrate (2) and a die block (1), in which two adjacent dies are spaced apart by a first distance (d1), - chemical-mechanical polishing of the die block, - forming a fragilization zone in at least one part of the die block to delimit a corresponding die, - bonding the pseudo-donor substrate to the recipient substrate via the die block, - separating the die block along the fragilization zone to transfer the corresponding dies onto the recipient substrate, two adjacent dies being spaced apart by a second distance greater than the first distance (d1), - prior to bonding, locally roughening the surface of the die block and / or of the recipient substrate to render regions of said surface unsuitable for bonding, thereby preventing the transfer of dies to said regions.

Description

Technical Field

[0001] The present invention relates to a method for transferring a chip from a substrate known as a pseudo-donor substrate to a recipient substrate. Background Technology

[0002] In the fields of microelectronics, optics, or optoelectronics, the design of multilayer structures sometimes requires transferring tiles, which are part of a layer, from a donor substrate to a support substrate or acceptor substrate.

[0003] This type of process is often referred to as a piling process, which involves the partial transfer of layers taken from a donor substrate to form one or more pilings arranged in a predetermined pattern or position on a support substrate.

[0004] The dimensional difference between the donor substrate and the support substrate may necessitate this padding process. Specifically, due to this dimensional difference, it is not possible to transfer the donor substrate layer that covers the entire surface of the support substrate.

[0005] One well-known layer transfer method is Smart Cut. TM A method in which a brittle region defining the layer to be transferred is formed by implanting atomic species into a donor substrate, the donor substrate is bonded to a support substrate, and the donor substrate is separated along the brittle region to transfer the layer from the donor substrate to the support substrate. However, this method assumes that the donor substrate and the support substrate have the same dimensions.

[0006] Currently, while silicon substrates with relatively large dimensions (typically 300 mm in diameter) are available, other materials of interest are currently only available in the form of smaller bulk substrates, such as 10 or 15 cm in diameter. Furthermore, these materials of interest are sometimes particularly expensive, so it is desirable to minimize any waste incurred during transfer. This is especially true for III-V semiconductor materials, including nitrides (e.g., regarding binary compounds, indium nitride (InN), gallium nitride (GaN), and aluminum nitride (AlN)), arsenides (e.g., regarding binary compounds, indium arsenide (InAs), gallium arsenide (GaAs), and aluminum arsenide (AlAs)), and phosphides (e.g., regarding binary compounds, indium phosphide (InP), gallium phosphide (GaP), and aluminum phosphide (AlP)).

[0007] A Smart Cut-based TM The solution is to remove one or more pieces from at least one donor substrate instead of transferring the entire layer of the donor substrate and transfer the pieces to an intermediate support to form a “pseudo-donor substrate”. An embrittlement region is formed in each piece by implanting atomic species. The pseudo-donor substrate is then bonded to the acceptor substrate via the pieces, and each piece is separated along the embrittlement region to transfer a portion of each piece to the acceptor substrate.

[0008] For dummy donor substrates to bond to acceptor substrates, the free surfaces of all sheets must lie in the same plane. For this purpose, CMP (chemical mechanical polishing) can be performed prior to bonding, and preferably before grinding the dummy donor substrate. CMP combines the mechanical action of a polishing pad and abrasive powder with the chemical action of a polishing solution containing abrasive powder (the abrasive powder and the solution together are referred to as a slurry) to planarize the surface of the sheet assembly.

[0009] However, if the distance between the wafers is large, typically greater than or equal to 250 µm, the flexible polishing pad will partially insert into the gap between the wafers, thereby eroding the edges of the wafers. This edge rounding reduces the usable surface area of ​​the chip transferred onto the acceptor substrate.

[0010] Conversely, the edges of the polishing pads will damage the polishing pads very quickly, requiring frequent replacement of the polishing pads and increasing the cost of chemical mechanical polishing operations.

[0011] To avoid this edge rounding of the sheets, it is desirable to form a pseudo-donor substrate in which the distance between the sheets is less than or equal to 250 µm.

[0012] However, depending on the application of the final structure including the acceptor substrate and the chip, the distance between the chips may need to be greater than 250 µm, for example greater than 1 mm, or even larger.

[0013] One solution is to reduce the size of the chip transferred to the donor substrate by using a photolithographic mask to locally protect a portion of the chip surface while exposing another portion. The mask-exposed portion of the chip can then be locally etched to further reduce chip size and increase the distance between adjacent chips.

[0014] However, this method is not entirely satisfactory because it requires very precise mask placement and significantly increases the manufacturing time of the final structure. Furthermore, it consumes a large amount of chip material, which is typically expensive. Therefore, this method is not economically feasible on an industrial scale. Summary of the Invention

[0015] Therefore, one object of the present invention is to design a chip transfer method that avoids chip edge rounding even when chips are far apart, a method that is easy to use in industry and can minimize chip material consumption.

[0016] Therefore, the present invention provides a method for transferring a chip from a chip stack disposed on a support substrate to a receiver substrate, comprising: - A substrate called a pseudo-donor substrate is formed, which includes a support substrate and sheets, wherein two adjacent sheets are spaced apart by a first distance; - Perform chemical mechanical polishing on the sheet; - A brittle zone is formed in at least a portion of the sheet to define the corresponding chip; - The dummy donor substrate is bonded to the acceptor substrate via the sheet; - Separate the pieces along the brittle region to transfer the corresponding chips onto the acceptor substrate, with a second distance between two adjacent chips greater than the first distance; The method is characterized in that it includes locally roughening the surface of the wafer and / or the acceptor substrate prior to bonding, so that regions of the surface cannot be bonded, thereby preventing the chip from shifting to the regions.

[0017] Localized roughening can create regions at the bonding interface where adhesion between the wafer and the acceptor substrate is insufficient to allow chip transfer. Therefore, it allows for selective transfer of the chip to the acceptor substrate and selective transfer from a high-density wafer in a dummy donor substrate to a lower-density chip in the final structure. Since only the desired chip is transferred to the acceptor substrate, no chip material is wasted.

[0018] Depending on other advantageous but optional features, these features may be used individually or in combination where technically feasible: - Roughening is performed by localized laser irradiation of the surface of the receptor substrate; - Laser irradiation is performed using a laser with a wavelength between 100 nm and 550 nm, preferably between 250 nm and 400 nm, in pulses with a duration between 1 ns and 10 µs, preferably between 10 ns and 500 ns; - The method includes applying a pattern-defining mask and irradiating the surface through the mask; - The pattern is selected to divide the acceptor substrate into multiple regions, which are separated by roughened regions facing the same piece, thereby allowing the piece to be divided into multiple chips according to the pattern; - The first distance is less than or equal to 250 µm, preferably less than or equal to 100 µm; - The second distance is greater than or equal to 1 mm, preferably greater than or equal to 2 mm; - The formation of the embrittled region is achieved by implanting ionic species; - The pseudo-donor substrate includes a first set of wafers and a second set of wafers; - The method includes sequentially transferring pieces from a first group and pieces from a second group onto at least one receptor substrate; - The method includes: - Selectively implant ion species into the first set of wafers to form brittle regions, thereby defining the chip within the first set of wafers. - Local roughening of the first acceptor substrate facing the second set of sheets. - Bond the dummy donor substrate to the first acceptor substrate. - Separate the first set of wafers along the brittle region to transfer the corresponding chip onto the first acceptor substrate. Starting from the pseudo-donor substrate obtained by separating the first set of wafers, ion species are selectively implanted into the second set of wafers to form embrittlement regions, thereby defining the chip within the second set of wafers. - Local roughening of the second acceptor substrate facing the first set of sheets. - Bond the pseudo-donor substrate to the second acceptor substrate. - Separate the second set of wafers along the brittle region to transfer the corresponding chip onto the second acceptor substrate; - Only the chips of the first set of wafers are transferred onto the recipient substrate, and the pseudo-donor substrate is recycled for a new round of chip transfer of the first set of wafers; - The first group of panels and the second group of panels are formed from different materials; - The crystallinity of the first group of flakes is higher than that of the second group of flakes; - The materials of the first and second groups of plates have essentially the same material removal rate and chemical reactivity for chemical mechanical polishing; - The piece includes: - Semiconductor materials, such as group III-V materials, especially indium nitride (InN), gallium nitride (GaN), aluminum nitride (AlN), indium arsenide (InAs), gallium arsenide (GaAs), aluminum arsenide (AlAs), indium phosphide (InP), gallium phosphide (GaP), or aluminum phosphide (AlP), or group IV or group IV-IV materials, especially germanium or silicon carbide (SiC), - Piezoelectric materials, such as lithium tantalate (LiTaO3), lithium niobate (LiNbO3), and potassium sodium niobate. Barium titanate (BaTiO3), quartz, lead zirconate titanate (PZT), lead-magnesium niobate and compounds of lead titanate (PMN-PT), zinc oxide (ZnO), aluminum nitride (AlN) or scandium aluminum nitride (AlScN), and / or - Electrically insulating materials, such as diamond, strontium titanate, yttriated zirconia, or sapphire. Attached Figure Description

[0019] Other features and advantages of the invention will become apparent from the following detailed description taken in conjunction with the accompanying drawings, in which: - Figure 1 A top view and a cross-sectional view of a pseudo-donor substrate according to an embodiment of the present invention are shown; - Figure 2 It shows Figure 1 Top view and cross-sectional view of the pseudo-donor substrate and acceptor substrate before bonding; - Figure 3 The cross-sectional and bottom views of the acceptor substrate are shown after the chip has been transferred from the dummy donor substrate. - Figure 4 A cross-sectional view and a top view of a pseudo-donor substrate according to one embodiment are shown, wherein the sheet is divided into two groups; - Figures 5A to 5F Schematic illustration of from Figure 4 The steps involved in performing two consecutive chip transfers on a pseudo-donor substrate; - Figure 6 A cross-sectional view and a top view of a dummy donor substrate including a target piece and a sacrificial piece according to one embodiment are shown; - Figure 7 It shows Figure 1 The top and cross-sectional views of the spurious donor and acceptor substrates prior to bonding are shown. In one embodiment, a region of the acceptor substrate has been roughened to form a pattern of a sheet facing the spurious donor substrate.

[0020] For readability, the components are not necessarily drawn to scale. Components represented by the same reference symbol in different figures represent similar components and will not be described repeatedly. Detailed Implementation

[0021] Figures 1 to 3 The general principles of the invention are illustrated.

[0022] refer to Figure 1 The pseudo-donor substrate 10 is formed by arranging the sheet 1 on the support substrate 2.

[0023] The support substrate serves as a mechanical support for the sheet. The support substrate can be made of silicon, glass, or sapphire (non-limiting list). Figure 1 The layer indicated around the supporting substrate 2 is an oxide layer, which is typically present around the silicon substrate, but this layer is optional.

[0024] The diameter of the support substrate is advantageously greater than or equal to 150 mm, preferably greater than or equal to 200 mm, and more preferably greater than or equal to 300 mm.

[0025] Preferably, at least some of the sheets are formed from an expensive material available only in small sizes. Alternatively, the sheets may be formed by stacking such materials.

[0026] The sheet may advantageously include at least one of the materials: - Semiconductor materials, such as group III-V materials, especially indium nitride (InN), gallium nitride (GaN), aluminum nitride (AlN), indium arsenide (InAs), gallium arsenide (GaAs), aluminum arsenide (AlAs), indium phosphide (InP), gallium phosphide (GaP), or aluminum phosphide (AlP), or group IV or group IV-IV materials, especially germanium or silicon carbide (SiC). - Piezoelectric materials, such as lithium tantalate (LiTaO3), lithium niobate (LiNbO3), and potassium sodium niobate. Barium titanate (BaTiO3), quartz, lead zirconate titanate (PZT), lead magnesium niobate and compounds of lead titanate (PMN-PT), zinc oxide (ZnO), aluminum nitride (AlN) or scandium aluminum nitride (AlScN), and / or - Electrically insulating materials, such as diamond, strontium titanate, yttrium oxide zirconium oxide, or sapphire.

[0027] As described below, the pieces can be the same or made of different materials.

[0028] In the illustration, the pieces have four straight edges and are arranged at regular intervals to form a grid consisting of a set of parallel rows and columns. However, it should be noted that this arrangement of pieces is for illustrative purposes only; therefore, the pieces may optionally have different shapes or be arranged at different distances from each other. Furthermore, the pieces are not necessarily square, but can have any other shape suitable for their intended use, whether or not they have straight edges. For example, the pieces can be rectangular, circular, or any other shape composed of straight lines and / or curves.

[0029] The pieces are separated from each other, and the distance between the opposite edges of two adjacent pieces is d1. The distance d1 between all pieces can be the same or different.

[0030] The distance d1 is less than or equal to 250 µm, preferably less than or equal to 100 µm.

[0031] To form the dummy donor substrate, a sheet can be cut from one or more corresponding donor substrates and placed on a temporary support using a robot (a technique known as "pick-and-place"). Advantageously, the sheet has the same thickness as the corresponding donor substrate. Typically, the sheet thickness is between 50 µm and 1 mm, for example on the order of 300 to 650 µm, depending on the thickness of the donor material. The diameter of the donor substrate is typically smaller than the diameters of the temporary and support substrates.

[0032] Temporary supports are formed of one of the following materials, for example: tape secured to the frame, silicone, or glass (not a limiting list).

[0033] The sheet is then bonded to a support substrate, and the temporary support is removed to expose the opposite surface of the sheet. Advantageously, the sheet bonding to the support substrate is direct, but a bonding layer can also be used between the sheet and the support substrate.

[0034] Alternatively, robots and alignment marks set on the support substrate can be used to assemble the sheets directly onto the support substrate with sufficient precision. This avoids the use of temporary supports and the need to transfer the sheets from temporary supports to the support to form a dummy donor substrate.

[0035] Then, the free surface of the sheet is chemically and mechanically polished.

[0036] Optionally, a planarization step by mechanical abrasion (“grinding”) can precede the chemical mechanical polishing. This planarization may be particularly advantageous when the sheets have different thicknesses, to bring their free surfaces substantially in the same plane, or to reduce the thickness of the sheets. However, if the sheets have coplanar free surfaces, this planarization can be omitted and chemical mechanical polishing can proceed directly.

[0037] Chemical mechanical polishing (CMP) can smooth the surface of a sheet to achieve a roughness compatible with bonding to the acceptor substrate. This roughness is typically less than 0.5 nm RMS. This roughness is usually measured using atomic force microscopy (AFM) over a 1 µm x 1 µm area. Other measurement methods, such as optical interferometry, can also be used.

[0038] When the distance between the wafers is small enough, the polishing pad operates essentially within the plane of the wafer's free surface, without interfering with the wafers. Therefore, the edges of the wafers remain straight (unrounded). Otherwise, the edges of the wafers might be rounded.

[0039] A brittle region is formed in at least a portion of the wafer to define the chip to be transferred onto the acceptor substrate.

[0040] The embrittlement region is typically formed by ion implantation, preferably with hydrogen and / or helium. Those skilled in the art can define the implantation operating conditions based on the wafer material and the thickness of the chip to be transferred.

[0041] The formation of the embrittled zone is preferably performed after chemical mechanical polishing, but it can also be performed optionally beforehand.

[0042] The acceptor substrate can be a silicon substrate, but other materials are also considered. In particular, the choice of material can be guided by the intended application (e.g., to address electrical and / or thermal conductivity issues in the final structure). The material of the acceptor substrate can also be selected based on its compatibility with the sheet material, such as having a coefficient of thermal expansion close to that of the sheet. For example, the acceptor substrate can include gallium arsenide, sapphire, or glass (a non-limiting list).

[0043] The acceptor substrate advantageously has the same diameter as the spurious donor substrate support, but may optionally have a larger diameter.

[0044] To ensure that the chips on the acceptor substrate are aligned with a greater inter-chip distance than those on the support substrate, local bonding is performed between the dummy donor substrate and the acceptor substrate. The term "local bonding" refers to adhesion between the wafer and the acceptor substrate that does not occur across the entire surface of the wafer in contact with the acceptor substrate, but only on a given portion of that surface. Chip transfer to the acceptor substrate occurs only in the area where adhesion is sufficiently achieved.

[0045] To control the placement of the transferred chip, the surface of the wafer or acceptor substrate is roughened in areas where the chip should not be transferred. The roughness of the roughened surface is typically greater than 0.5 nm RMS, preferably greater than 2 nm RMS, making bonding impossible. This allows the wafer to be selectively bonded only in the unroughened regions.

[0046] Roughening is advantageously performed by laser irradiation. The wavelength of the laser used can be selected between 100 nm and 550 nm, preferably between 250 nm and 400 nm. Irradiation is performed in pulses, with a duration typically between 1 ns and 10 µs, more advantageously between 10 ns and 500 ns. For a silicon substrate with a wavelength of 308 nm and a pulse duration of 160 ns at room temperature, measuring 15 x 15 mm... 2 The energy density provided in the region is approximately 1.9 J / cm². 2 The energy density corresponds to the melting threshold of the irradiated material. Those skilled in the art can adjust the energy density according to the laser used and the material to be treated.

[0047] Preferably, the irradiation is applied to the acceptor substrate rather than the sheet to avoid affecting the properties of the sheet, especially when the sheet is made of a III-V material.

[0048] The roughening region can be defined solely by controlled scanning of a laser beam on the surface of the acceptor substrate, particularly when the size of the region to be roughened is approximately a few millimeters. Alternatively, particularly for more precise control over the size and / or shape of the region to be roughened, irradiation can be performed by applying a mask to the acceptor substrate, the mask comprising apertures facing the region to be roughened.

[0049] Figure 2 The acceptor substrate 3 is shown, on which roughened regions 30 have been formed relative to certain pieces of the spurious donor substrate.

[0050] It should be noted that the unroughened region does not necessarily have the same size or shape as the wafer. Specifically, a portion of the surface of the acceptor substrate facing the wafer (e.g., corresponding to the periphery of the wafer) can be roughened to allow the transfer of chips smaller than the wafer (the portion of the wafer in contact with the roughened region will not be transferred due to lack of adhesion).

[0051] exist Figure 2 In the cross-sectional view, for ease of visualization, the roughened regions 30 protrude slightly relative to the rest of the surface of the acceptor substrate 3, but they are generally coplanar with the rest of the surface.

[0052] Bonding involves aligning the dummy donor substrate and the acceptor substrate to ensure that the roughened region faces the sheet, and then bringing the acceptor substrate and the sheet into contact.

[0053] For this type of alignment, a plane or notch around the substrate is usually sufficient. Alternatively, the bonding device can consider the edges of the wafer for alignment, or use an alignment pattern set on the acceptor substrate.

[0054] After the dummy donor substrate and the acceptor substrate are brought into contact, the chip is separated along the brittle region. This is done in a manner known per se (Smart Cut). TM The method can initiate separation through heat treatment, mechanical and / or chemical action in the embrittled zone.

[0055] This separation and transfer of the chip to the acceptor substrate only occurs in regions where there is sufficient adhesion between the chip and the acceptor substrate, i.e., the non-roughened regions. In roughened regions, transfer does not occur, and the chip remains completely on the support substrate.

[0056] Therefore, as Figure 3 As shown, only a portion of the chips 1' are transferred onto the acceptor substrate 3. As a result, the distance d2 between two adjacent chips 1' on the acceptor substrate 3 is greater than the initial distance d1 between the chips on the dummy donor substrate 10.

[0057] The distance d2 can be greater than or equal to 1 mm, preferably greater than or equal to 2 mm.

[0058] The remaining portion of the dummy donor substrate (including the remaining portion of chip-separated dies and chip-not-separated dies) can optionally be recycled into a new round of chip transfer processes.

[0059] Various embodiments of the present invention will now be described.

[0060] exist Figure 2 and Figure 3 In the case shown, only a portion of the chip is transferred. Therefore, the remaining portion of the dummy donor substrate comprises two pieces of different thicknesses.

[0061] To recover the dummy donor substrate, mechanical abrasion is required to restore all pieces to the same thickness, followed by chemical mechanical polishing to obtain a surface smooth enough for bonding.

[0062] However, this recycling process means that a material thickness corresponding to the chip thickness is consumed on the unbonded dies.

[0063] Various solutions can be envisioned to optimize the use of paving materials and / or avoid wasting paving materials, especially when paving materials are expensive.

[0064] The first solution is to use dummy donor substrates multiple times to transfer chips from one or more recipient substrates.

[0065] like Figure 4 As shown, the sheets are divided into two groups and distributed on the surface of the supporting substrate. For example, the dummy donor substrate includes alternating first group of sheets 11 and second group of sheets 12. The distance d1 between two adjacent sheets (particularly sheets from two different groups) is less than or equal to 250 µm. On the other hand, the distance between two adjacent sheets in the same group is greater than 250 µm, preferably greater than 1 mm.

[0066] The two sets of wafers can be formed from the same material or by stacking the same materials. Alternatively, the two sets of wafers can be formed from two different materials or by stacking the same materials. For example, the first set of wafers may include a semiconductor material, while the second set of wafers may include another semiconductor material or a piezoelectric material.

[0067] The chip transfer is carried out in two steps, depending on the group to which the chip belongs.

[0068] In the first step, embrittlement zones 110 are selectively formed in the first set of sheets 11. For this purpose, a mask 4, for example made of graphite, is placed opposite the surface of the sheets, the mask including apertures facing the first set of sheets 11. Figure 5AThen, ion implantation is performed through a mask to form embrittlement regions 110 in the first set of sheets 11, but not in the second set of sheets 12.

[0069] Elsewhere, the surface of the receptor substrate is locally roughened in areas designed to contact the second set of sheets.

[0070] refer to Figure 5B The dummy donor substrate is bonded to the acceptor substrate. Due to roughening, adhesion occurs only between the first set of sheets 11 and the acceptor substrate 3.

[0071] The first set of wafers 11 separates along the corresponding brittle region 110, allowing the chip 11' to be transferred onto the acceptor substrate 3. Figure 5C Therefore, the acceptor substrate 3 includes chips 11' spaced apart by a distance d2, which is greater than d1.

[0072] The dummy donor substrate can be directly reused for the second-step chip transfer.

[0073] Embrittled regions 120 are selectively formed in the second set of sheets 12. For this purpose, a mask 4, for example made of graphite, is placed opposite the surface of the sheets, the mask including apertures facing the second set of sheets 12. Figure 5D Then, ion implantation is performed through a mask to form embrittlement regions 120 in the second set of wafers, but not in the remainder of the first set of wafers.

[0074] Furthermore, the surface of the acceptor substrate is locally roughened in the region intended to contact the first set of wafers. This acceptor substrate can be the same as the acceptor substrate used in the first step, for example, when the first and second sets of wafers are intended to be transferred to different regions of the acceptor substrate. In this case, the relative orientation of the dummy donor substrate and the acceptor substrate may differ between the two wafer transfer steps, depending on the desired arrangement in the final structure. Alternatively, for the second step, a different acceptor substrate can be used than that used in the first step.

[0075] refer to Figure 5E The dummy donor substrate is bonded to the acceptor substrate. Due to roughening, adhesion occurs only between the second set of sheets 12 and the acceptor substrate 3.

[0076] The second set of wafers 12 separates along the corresponding brittle region 120, allowing the chip 12' to be transferred onto the acceptor substrate. Figure 5F Therefore, the acceptor substrate 3 includes chips 12' spaced apart by a distance d2, which is greater than d1.

[0077] Since the surface of the second set of wafers has been polished before the first chip transfer step, they can be bonded, thus eliminating the need for chemical mechanical polishing between the two chip transfer steps.

[0078] At the end of the second step, the dummy donor substrate can be recycled for the two new chip transfer steps described above, particularly by chemical mechanical polishing.

[0079] Undoubtedly, more than two sets of chips can be envisioned, and therefore more than two chip transfer steps can be envisioned.

[0080] A second way to save tiling material is to use a different material for dies that are not intended for chip transfer.

[0081] like Figure 6 As shown, the dummy donor substrate 10 is formed by arranging target pieces 1 (intended for transferring a chip onto a recipient substrate) and pieces 5 of another material (typically cheaper than the target pieces) on a support substrate 2. These other pieces 5 are sacrificial pieces, used only to fill the gaps between the target pieces to ensure that the distance d1 between two adjacent pieces is less than 250 µm, thereby avoiding rounding of the edges of the target pieces during chemical mechanical polishing and avoiding damage to the polishing pad. Preferably, each target piece is surrounded by sacrificial pieces at each of its edges to allow for uniform polishing of the target pieces.

[0082] The sacrificial piece can be made of the same material as the target piece, but may have lower crystallinity. For example, if the target piece is made of single-crystal InP, the sacrificial piece can be made of polycrystalline InP, or of single-crystal InP with a high dislocation density. The advantage of using the same material for both the target and sacrificial pieces is that all pieces exhibit similar performance in chemical mechanical polishing, particularly in terms of hardness and material removal rate. Therefore, all pieces can be polished uniformly.

[0083] Alternatively, the sacrificial piece can be formed from a material with a different composition than the target piece. In this case, the material of the sacrificial piece should preferably be selected to have behavior similar to that of the target piece material in terms of chemical mechanical polishing, particularly in terms of material removal rate (the ratio of removal rates typically needs to be less than 2) and chemical reactivity with the slurry. In practice, those skilled in the art know how to find suitable material pairs for the target and sacrificial pieces, either by selecting the slurry and adjusting the material of the sacrificial piece accordingly, or by selecting the material of the sacrificial piece and conducting research to determine or even customize the slurry to have similar material removal rates and chemical reactivity.

[0084] Ion implantation is performed on the dummy donor at least within the target wafer to define the chip to be transferred.

[0085] Depending on the behavior of the sacrificial block material towards injection, especially in terms of foaming, injection can be performed simultaneously within the sacrificial block.

[0086] Alternatively, especially if the material of the sacrificial sheet produces sputtering-related defects, the mask as described in the previous embodiments is used to inject only into the target sheet.

[0087] Bubbling is a phenomenon known to those skilled in the art, in which ionic species injected into the material deform the free surface of the material under heat treatment. The resulting bubbles may therefore impair bonding.

[0088] The acceptor substrate is locally roughened to roughen the surface facing the sacrificial piece.

[0089] The dummy donor substrate is bonded to the acceptor substrate. Due to roughening, adhesion occurs only between the target sheet and the acceptor substrate.

[0090] When the wafers separate along the embrittlement zone, regardless of whether the sacrificial wafer 5 has an embrittlement zone, only the chip from the target wafer 1 is transferred to the acceptor substrate.

[0091] Therefore, on the acceptor substrate 3, the chip 1' is separated by a distance d2 (greater than d1).

[0092] The dummy donor substrate can be recycled for new uses on the target wafer. To do this, if necessary, mechanical abrasion is performed to flush the sacrificial wafer with the target wafer, followed by chemical mechanical polishing to allow the surfaces of the target wafer to bond.

[0093] Another application of roughening is the production of chips with shapes and / or sizes different from those of the tessellation plate. Specifically, since the transfer only occurs in sufficiently smooth areas, a tessellation plate can be broken down into multiple smaller chips by bonding it to surfaces with alternating smooth and roughened regions.

[0094] Figure 7 The principle is schematically illustrated by an enlarged view of two opposing regions on the acceptor substrate and the dummy donor substrate. The region on the dummy donor substrate corresponds to a single piece 1. On the other hand, region 32 of the acceptor substrate 3 includes several regions with different roughnesses: a smooth region 31 capable of bonding the piece, and roughened regions 30 extending between the smooth regions that cannot be bonded. In the illustrated embodiment, these regions are arranged in a checkerboard pattern, but undoubtedly they can take on any other shape depending on the desired shape of the chip.

[0095] Once the wafer 1 is bonded to the region 32 of the acceptor substrate, and the wafer separates along the brittle region, only the portion of the wafer actually bonded to the acceptor substrate is transferred onto the acceptor substrate to form a chip.

[0096] The minimum size of the roughened region depends on the laser used.

[0097] Using a mask can be useful for more precise control over the size and shape of the region to be roughened, especially when its size is less than 100 µm. The mask includes apertures corresponding to the region to be roughened, allowing illumination of the acceptor substrate surface only at these apertures while protecting the rest of the acceptor substrate surface.

[0098] Of course, the above implementation methods can be combined.

[0099] Regarding the bonding conditions between the sheet and the acceptor substrate, those skilled in the art can use any known technique applicable to the material under consideration. For example, a dielectric layer, such as SiO2, Al2O3, AlN, or SiN, can be formed on the surface of the sheet and / or the acceptor substrate to enable oxide-to-oxide bonding. Alternatively, those skilled in the art can use an intermediate bonding layer, such as a metal or polymer. Bonding is preferably performed at atmospheric pressure, but can also be performed in a vacuum if necessary. Annealing can also be performed to enhance the bonding energy.

Claims

1. A method for transferring a chip from a wafer disposed on a support substrate to a recipient substrate, the method comprising: - A substrate (10) called a pseudo-donor substrate is formed, which includes the support substrate (2) and sheets (1, 11, 12), wherein two adjacent sheets are spaced apart by a first distance (d1); - Perform chemical mechanical polishing on the sheet (1); - A brittle region (110, 120) is formed in at least a portion of the pieces (1, 11, 12) to define the corresponding chip (1', 11', 12'); - The pseudo-donor substrate (10) is bonded to the acceptor substrate (3) via the pieces (1, 11, 12); - Separate the pieces (1, 11, 12) along the brittle regions (110, 120) to transfer the corresponding chips (1', 11', 12') onto the acceptor substrate (3), with the spacing between two adjacent chips being a second distance (d2) greater than the first distance (d1); The method is characterized in that it includes locally roughening the surface of the pieces (1, 11, 12) and / or the acceptor substrate (3) before bonding, so that regions (30) of the surface cannot be bonded, thereby preventing the chip from being transferred to the regions (30).

2. The method according to claim 1, wherein roughening is performed by local laser irradiation of the surface of the receptor substrate (3).

3. The method of claim 2, wherein the laser irradiation is performed using a laser with a wavelength between 100 nm and 550 nm, preferably between 250 nm and 400 nm, in pulses with a duration between 1 ns and 10 µs, preferably between 10 ns and 500 ns.

4. The method of claim 2, further comprising applying a pattern-defining mask and irradiating the surface through the mask.

5. The method of claim 4, wherein the pattern is selected to divide a plurality of regions of the acceptor substrate, the regions being separated by roughened regions facing the same piece, thereby dividing the piece into a plurality of chips according to the pattern.

6. The method according to any one of claims 1 to 5, wherein the first distance (d1) is less than or equal to 250 µm, preferably less than or equal to 100 µm.

7. The method according to any one of claims 1 to 6, wherein the second distance (d2) is greater than or equal to 1 mm, preferably greater than or equal to 2 mm.

8. The method according to any one of claims 1 to 7, wherein the formation of the embrittled region is performed by implanting ionic species.

9. The method according to any one of claims 1 to 8, wherein the dummy donor substrate comprises a first set of sheets (11) and a second set of sheets (12).

10. The method according to claim 9, wherein the pieces (11) from the first group and the pieces (12) from the second group are successively transferred onto at least one receptor substrate.

11. The method of claim 10 in conjunction with claim 8, further comprising: - Selectively implant ion species into the first set of wafers (11) to form a brittle region (110), thereby defining a chip (11') in the first set of wafers; - Local roughening of the first acceptor substrate (3) facing the second set of pieces (12); - The pseudo-donor substrate is bonded to the first acceptor substrate (3); - Separate the first set of pieces (11) along the brittle region (110) to transfer the corresponding chip (11') onto the first acceptor substrate (3); - Starting from the pseudo-donor substrate obtained by separating the first set of wafers, ion species are selectively implanted into the second set of wafers (12) to form brittle regions (120), thereby defining the chip (12') in the second set of wafers; - Local roughening of the second acceptor substrate facing the first set of sheets; - Bond the pseudo-donor substrate to the second acceptor substrate; - Separate the second set of wafers along the brittle region to transfer the corresponding chip onto the second acceptor substrate.

12. The method of claim 9, wherein only chips (11') from the first set of wafers (11) are transferred onto the recipient substrate (3), and the pseudo-donor substrate (10) is recycled for a new round of chip transfer from the first set of wafers (11).

13. The method according to any one of claims 9 to 12, wherein the first set of sheets (11) and the second set of sheets (12) are formed of different materials.

14. The method according to claim 12, wherein the crystallinity of the first set of sheets (11) is higher than that of the second set of sheets (12).

15. The method according to claim 13 or 14, wherein the materials of the first set of plates (11) and the second set of plates (12) have the same material removal rate and chemical reactivity for chemical mechanical polishing.

16. The method according to any one of claims 1 to 15, wherein the pieces (1, 11, 12) comprise: - Semiconductor materials, such as group III-V materials, especially indium nitride (InN), gallium nitride (GaN), aluminum nitride (AlN), indium arsenide (InAs), gallium arsenide (GaAs), aluminum arsenide (AlAs), indium phosphide (InP), gallium phosphide (GaP) or aluminum phosphide (AlP), or group IV or group IV-IV materials, especially germanium or silicon carbide (SiC); - Piezoelectric materials, such as lithium tantalate (LiTaO3), lithium niobate (LiNbO3), and potassium sodium niobate. Barium titanate (BaTiO3), quartz, lead zirconate titanate (PZT), lead magnesium niobate and compounds of lead titanate (PMN-PT), zinc oxide (ZnO), aluminum nitride (AlN) or scandium aluminum nitride (AlScN); and / or - Electrically insulating materials, such as diamond, strontium titanate, yttrium oxide zirconium oxide, or sapphire.

Images