Method for transferring a thin film onto a supporting substrate - Patent application
Patent Information
- Application Number
- JP2024547719
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2022-02-14
- Filing Date
- 2022-12-19
- Publication Date
- 2025-10-27
Smart Images

Figure 00000000_0000_ABST
Abstract
Description
[Technical field]
[0001] The present invention relates to the field of microelectronics and semiconductors. In particular, the present invention relates to a method for transferring a thin film onto a supporting substrate based on Smart Cut™ technology, where the thin film exhibits improved roughness after separation. In particular, the transfer method can be used to manufacture SOI structures. [Background technology]
[0002] The Smart Cut™ technology is well known for producing SOI (silicon-on-insulator) structures and more generally for thin film transfer. This technology is based on the formation of a buried brittle surface in a donor substrate by implanting optical species into the substrate. The buried brittle surface, together with the front surface of the donor substrate, defines the thin film to be transferred. The donor substrate and the support substrate are then joined at their respective front surfaces to form a bonded assembly. The assembly is advantageously performed by direct bonding, by molecular adhesion, i.e. without the involvement of any adhesive material, such that a bonded interface is established between the two assembled substrates. Microcrack growth in the buried brittle surface can result in spontaneous separation along a plane through thermal activation, resulting in the transfer of the thin film onto the support substrate (forming a laminate structure). The remaining donor substrate can be reused for subsequent film transfers. After separation, the laminate structure is typically subjected to a finishing treatment to restore the crystal quality and surface roughness of the transferred thin film. In particular, these finishing steps may include oxidizing or smoothing heat treatments (in neutral or reducing atmospheres), chemical cleaning and / or etching and / or chemical-mechanical polishing steps, as known to those skilled in the art. Finally, various tools for inspecting the final structure make it possible to check the entire surface of the thin film.
[0003] If the separation at the embedded brittle surface is spontaneous, significant variability is observed in the surface roughness of the transferred thin film, both at high frequency (micro-roughness) and low frequency (undulations, localized areas of high roughness, mottle, etc.) These variabilities are visible and measurable when inspecting the thin film in the final structure, especially via the inspection tools mentioned above.
[0004] Recall that the surface roughness of the finished thin film can be imaged by mapping obtained using a Surfscan™ inspection tool from KLA-Tencor (Figure 1). The level of roughness and potential patterns (mottle (M), zone density (ZD), etc.) is measured or revealed by measuring the diffuse background noise ("haze"), which corresponds to the intensity of light scattered by the surface of the thin film. The haze signal is expressed as a function of the 0.1 μm -1 ~10μm -1 It varies linearly with the square of the RMS surface roughness (root mean square roughness) in the spatial frequency range of . For more information on this large area roughness inspection and evaluation technique, see the article "Seeing Through the Haze" by F. Holsteyns, Yield Management Solutions, Spring 2004, pp. 50-54.
[0005] The maps in [Figure 1] show the surface roughness of two thin films transferred from two bonded assemblies and processed identically to finish. Map (A) shows a peripheral area of residual roughness (especially the top of the map) known as the "ZD" (Zone Density). Map (B) is completely devoid of this area. A more pronounced mottle (M) is also evident on map (A). The average and maximum roughness (expressed in ppm haze) are also significantly different between the two maps (A) and (B). [Figure 1] shows the variability in the final quality and roughness of the thin films, which is mainly due to the variability in surface roughness (high and low frequencies) after separation.
[0006] It remains important to reduce the surface roughness (regardless of spatial frequency) of these layers after transfer in order to improve the quality of the thin films in the final laminate structure.
[0007] From US 2010 / 330779 it is known to form local unbonded regions at the bond interface bounded by the bonded region in order to constitute a separation trigger and thus limit the roughness of the thin silicon layer transferred onto the glass support substrate. The local unbonded regions are obtained by generating topologies corresponding to cavities and / or domes (or peaks) on the surface of the donor substrate (silicon) or the support substrate (glass) to be bonded. Cavities and / or peaks of the order of 2-3 micrometers are formed on the glass substrate, which are then displaced over a distance of several tens of mm. 2. This creates an unjoined region.
[0008] Object of the invention The present invention proposes a transfer method with a specific fracture initiation point that improves the surface roughness of the thin film after separation and makes it possible to achieve a good surface quality after the finishing step of the stack structure, which is particularly advantageous for the manufacture of SOI structures. Summary of the Invention
[0009] The present invention relates to a method for transferring a thin film onto a support substrate, - providing a bonded assembly comprising a donor substrate and a support substrate assembled by direct bonding at their respective front surfaces following a bonding interface, the bonded assembly having a localized unbonded area within the bonding interface, the donor substrate further comprising a buried brittle surface; - separation along buried brittle planes initiated at local unbonded regions following microcrack growth in the plane by thermal activation, which separation results in transfer of the thin film from the donor substrate to the support substrate.
[0010] The method is characterized in that the localized unbonded regions are generated only by roughened regions intentionally created on at least one of the front surfaces of the donor substrate and the support substrate prior to assembly, the roughened regions being free of topology and having a predetermined roughness with an amplitude of 0.5 nm RMS (0.5 nm RMS exclusive) to 60.0 nm RMS (60.0 nm RMS exclusive).
[0011] According to advantageous characteristics of the invention, taken alone or in any feasible combination, the following is performed: the roughened area is produced by at least one laser shot which produces only a surface melting of the material constituting the donor substrate and / or the support substrate within the first 1 to 30 nanometers on their respective front side, The laser shot has a pulse duration between 1 ns and 1000 ns, preferably between 10 ns and 500 ns. The laser shot is 1.8 J / cm using a 308 nm laser. 2 ~2.5J / cm 2 It is carried out at an energy density of The laser shot irradiates a circular surface with a diameter of less than 200 micrometers. The roughened area is created by multiple laser shots, e.g., 2-15 laser shots, and the adjacent circular surfaces irradiated by successive laser shots are tangential or have an overlap rate of 1%-95%. The front surface, on which the roughened area is formed, is made of monocrystalline silicon and the predetermined roughness has an amplitude of 0.5 nm RMS to 4.0 nm RMS, preferably 1.0 nm RMS to 2.5 nm RMS. The front surface, on which the roughened area is formed, is made of polycrystalline silicon and the predetermined roughness has an amplitude between 0.5 nm RMS and 5.0 nm RMS, preferably between 2.0 nm RMS and 5.0 nm RMS. The front surface on which the roughened region is formed is made of silicon oxide and the predetermined roughness has an amplitude of 1.0 nm RMS to 60.0 nm RMS. A roughened region is formed on the front surface of the donor substrate prior to forming a buried brittle surface in the donor substrate. A donor substrate has a front surface formed from a first material, a pre-roughened region is formed on the surface of the material, and after the formation of the pre-roughened region and before the formation of the buried brittle surface, thermal oxidation of the first material of the donor substrate is performed to form an insulating layer which is assembled on the support substrate in a bonded assembly, and the insulating layer comprises on its free surface a roughened region perpendicular to the pre-roughened region. A support substrate is formed at its front surface from a first material, a pre-roughened region is formed on the surface of the material, and thermal oxidation of the first material of the support substrate is performed after the formation of the pre-roughened region to form an insulating layer which is assembled on the donor substrate in a bonding assembly, the insulating layer comprising on its free surface a roughened region perpendicular to the pre-roughened region. The local unbonded area has, in the plane of the bond interface, a shape whose contour at least in part has a radius of curvature smaller than the radius of a circular bond defect in the same area. the localized unbonded region has at least one lateral dimension, in the plane of the bond interface, of less than 300 micrometers; The local unbonded area is located in the central region of the bonded assembly, in the plane of the bonded interface. A buried brittle plane is formed in the donor substrate by implanting optical species such as hydrogen, helium or a combination of both species. The donor substrate and / or the support substrate have an insulating layer on at least their respective front surfaces, the insulating layer forming a buried insulating layer adjacent the bonding interface in the bonded assembly. To form a stacked SOI structure, the thin film from the donor substrate is single crystal silicon and the support substrate contains single crystal silicon. [Brief description of the drawings]
[0012] Other features and advantages of the present invention will become apparent from the following detailed description of the invention which refers to the accompanying drawings. [Figure 1] Two representative surface roughness maps of two transferred thin films from two bonded assemblies that were identically processed to finish using conventional methods are shown. Both maps were obtained via a Surfscan™ inspection tool. [Diagram 2] 1 is a graph showing the surface roughness of a thin film based on the time to failure for multiple bonded assemblies (different from the bonded assembly shown in FIG. 1) that were identically processed up to the finishing stage using a conventional method. [Figure 3a]FIG. 2 shows a bonding assembly included in an intermediate step of the transfer method according to the present invention. [Figure 3b] Illustrated is an example of a donor or support substrate that includes a roughened area intentionally formed on its front surface in accordance with the transfer method of the present invention, the roughened area having a predetermined roughness. [Figure 4] FIG. 2 shows the laminate structure obtained by the transfer method according to the invention and the remainder of the donor substrate. [Diagram 5] 1 shows the evolution of RMS (root mean square) and PV (maximum peak-to-valley) roughness of a silicon surface irradiated by a laser shot based on the laser energy density. [Figure 6] Two representative maps of surface roughness (after finishing) of two transferred thin films for a first and a second RFSOI structure are shown, the first (110) obtained with the transfer method according to the invention and the second (110') obtained with a conventional transfer method. Both maps were obtained via a Surfscan™ inspection tool. [Figure 7] 1 shows various examples of shapes of roughened areas and associated local unbonded areas implemented in a transfer method according to the present invention. [Figure 8a] 1 shows an image of a non-transferred area corresponding to a local unbonded area of a bonded assembly, the non-transferred area being observed on a laminate structure obtained by a transfer method according to the invention. [Figure 8b] 1 shows an image of a non-transferred area corresponding to a local unbonded area of a bonded assembly, the non-transferred area being observed on a laminate structure obtained by a transfer method according to the invention.
[0013] The figures are schematic and are not drawn to scale for ease of reading, in particular the layer thicknesses along the z-axis are not drawn to scale relative to the lateral dimensions along the x- and y-axes.
[0014] The same reference numbers in the figures and the description may be used for the same type of elements. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0015] The present invention relates to a method for transferring a thin film onto a support substrate to form a laminate structure. As mentioned in the introduction, such a laminate structure may be of the SOI type, comprising a thin silicon surface layer, an intermediate insulating layer of silicon oxide and a silicon support substrate. The support substrate may optionally contain other functional layers, such as, for example, a charge trapping layer for SOI structures designed for radio frequency (RF) applications. However, the transfer method according to the invention is not limited to the manufacture of SOI but can be applied to many other laminate structures in the fields of microelectronics, microsystems and semiconductors.
[0016] The transfer method according to the invention is based on the Smart Cut™ technology. If the separation in the embedded brittle plane is spontaneous, the fracture time (i.e. the time until separation occurs during thermal fracture annealing) may vary between several similarly processed bonded assemblies undergoing the same annealing in the same oven. The fracture time (FT) depends on a number of parameters related to the formation of the embedded brittle plane, the fracture annealing, the nature of the bonded assembly, etc. The applicant has noted that for bonded assemblies prepared in a similar manner and undergoing the same fracture annealing, the separation occurring at a short fracture time (FT) results in a lower high frequency surface roughness (micro-roughness) of the thin film in the final laminate structure (i.e. after transfer and finishing) than the separation occurring at a longer fracture time (FT1), as seen in [Figure 2]. Furthermore, a long fracture time induces localized areas of very high roughness (called dense zones ZD) at the edge of the thin film after fracture, which occurs little or not at all when the fracture time is short. This dense region reduces the quality and roughness of the thin film even after finishing, as can be seen in map (A) of Figure 1.
[0017] The transfer method according to the invention therefore aims to initiate spontaneous separation at the embedded brittle surface in a fast (short time to failure) and reproducible (low variation in time to failure between multiple similar bonded assemblies) manner in order to substantially improve the surface roughness of the transferred thin film.
[0018] To achieve this, the transfer process first involves providing a bonded assembly 100 comprising a donor substrate 1 and a support substrate 2 assembled by direct bonding at their respective front surfaces (1a, 2a) along a bonding interface 3 ([Figure 3a]).
[0019] The donor substrate 1 is preferably in the form of a wafer with a diameter of 100 mm, 150 mm, 200 mm, 300 mm or 450 mm and typically with a thickness of 300 μm to 1 mm. It comprises a front surface 1a and a rear surface 1b. The surface roughness of the front surface 1a is selected to be less than 0 nm RMS, preferentially less than 0.5 nm RMS (measured for example by atomic force microscopy (AFM) with a 20 μm×20 μm scan). The donor substrate 1 can be made of silicon or any other semiconducting or insulating material to which thin film transfer can be targeted (for example SiC, GaN, III-V compounds, piezoelectric materials, etc.). It is also noted that the donor substrate 1 can include, at least on its front surface 1a, one or more additional layers 12, such as an insulating layer. As shown in [Figure 3a], this additional layer 12 becomes an embedded intermediate layer in the bonded assembly 100 after assembly of the donor substrate 1 and the support substrate 2.
[0020] The donor substrate 1 comprises a buried brittle surface 11 that defines the thin film 10 to be transferred. As is well known for the Smart Cut™ technology, such a buried brittle surface 11 can be formed by implanting optical species such as hydrogen, helium or a combination of both. The optical species are implanted to a predefined depth in the donor substrate 1 that corresponds to the thickness of the target thin film 10. These optical species form microcavities around the determined depth distributed in the thin film substantially parallel to the front surface 1a of the donor substrate 1 or parallel to the plane (x,y) in the figure. This thin layer is called buried brittle surface 11 for simplicity.
[0021] The implantation energy of the photospecies is selected to reach a determined depth. For example, hydrogen ions with energies between 10 keV and 210 keV and a density of 5E16 / cm 2 ~1E17 / cm 2to define a thin film 10 having a thickness of the order of 100 nm to 1500 nm. It is recalled that an additional layer may be deposited on the front side 1a of the donor substrate 1 prior to the ion implantation step. This additional layer may consist, for example, of a material such as silicon oxide or silicon nitride. It may be retained for the next assembly step (and form all or part of the intermediate layer of the bonded assembly 100) or it may be removed.
[0022] The support substrate 2 is also preferably in the form of a wafer with a diameter of 100 mm, 150 mm, 200 mm, 300 mm or 450 mm and typically with a thickness of 300 μm to 1 mm. It has a front surface 2a and a rear surface 2b. The surface roughness of the front surface 2a is selected to be less than 1.0 nm RMS, preferably less than 0.5 nm RMS (e.g. measured by AFM on a 20 μm×20 μm scan). The support substrate 2 can be made of silicon or any other semiconducting or insulating material to which thin film transfer can be performed (e.g. SiC, GaN, IILV compounds, piezoelectric materials, insulating materials, etc.). It is also noted that the support substrate 2 can include one or more additional layers, such as an insulating layer and / or a charge trapping layer, at least on its front surface 2a. This (or these) additional layers are embedded in the bonded assembly 100 after assembly of the donor substrate 1 and the support substrate 2.
[0023] The assembly between the donor substrate 1 and the support substrate 2 is based on direct bonding by molecular adhesion. As is known per se, such bonding does not require any adhesive material since an atomic scale bond is established between the joining surfaces forming the bonding interface 3. Several types of molecular adhesion bonds exist, which differ in particular by their temperature, pressure, atmospheric conditions or treatment before contacting the surfaces. One may mention bonding at room temperature with or without prior plasma activation of the surfaces to be assembled, atomic diffusion bonding (ADB), surface activated bonding (SAB), etc.
[0024] The assembly step may include conventional chemical cleaning sequences (e.g. RCA cleaning), surface activation (e.g. oxygen or nitrogen plasma) or other surface treatments (such as cleaning by scrubbing) that can promote the quality of the bonding interface 3 (low defects, strong adhesion energy) before contacting the front faces 1a, 2a to be assembled.
[0025] The bonded assembly 100 according to the invention has the particularity of including within the bonded interface 3 ([Fig. 3a]) a local unbonded area 31. In other words, this local unbonded area 31 is bounded by the bonded interface 3, which is closed reflecting the molecular adhesion forces bonding the front faces 1a, 2a of the assembled substrates 1, 2.
[0026] The local unbonded area 31 is only generated by the presence of a roughened area 31a intentionally formed on at least one of the front surfaces 1a, 2a of the donor substrate 1 and the support substrate 2 before their assembly (FIG. 3b). The roughened area 31a is free of low-frequency topologies or ripples, i.e. with a typical wavelength larger than 100 nm. It has a predetermined high-frequency roughness that is greater than the roughness of the front surfaces 1a, 2a around this roughened area 31a. The roughness of the front surfaces 1a, 2a can typically vary based on the nature of the materials of the surfaces 1a, 2a and on the type of direct bonding performed, but it always allows a bonded (closed) interface to be obtained, while the predetermined roughness in the roughened area 31a prevents local bonding between the two surfaces 1a, 2a.
[0027] The amplitude of the given roughness is strictly greater than 0.5 nm RMS and strictly less than 60.0 nm RMS, with a typical wavelength between 10 nm and 100 nm (corresponding to high frequency micro-roughness). As is well known, the term RMS ("root mean square") corresponds to the root mean square roughness value. The technique used to measure this micro-roughness is the 10×10 μm 2 ~30×30μm 2Atomic Force Microscopy (AFM) with scanning of . It should be recalled that the roughened area 31a does not have a topology corresponding to low frequency ripples, but only a micro-roughness in the spatial frequency range mentioned above. The maximum peak-to-valley (PV) amplitude in the roughened area 31a is typically between 5.0 nm and 300.0 nm, preferentially between 5.0 nm and 60.0 nm.
[0028] In particular, in the example shown in FIG. 3b, the roughened area 31a has a RMS of 2.3 nm and a PV roughness of 25 nm measured by AFM on a 10 μm×10 μm scan (right image in FIG. 3b).
[0029] According to a particular embodiment, particularly used for the manufacture of SOI structures, the front surface 1a, 2a on which the roughened area 31a is produced is made of monocrystalline silicon and the predetermined roughness preferably has an amplitude between 0.5 nm RMS and 4.0 nm RMS (typically between 5.0 nm PV and 40.0 nm PV), even more preferentially this amplitude is between 1.0 nm RMS and 2.5 nm RMS (typically between 10.0 nm PV and 25.0 nm PV), or even between 1.5 nm RMS and 2.5 nm RMS.
[0030] In another advantageous embodiment, the front surface 1a, 2a of the substrate 1, 2, on which the roughened area 31a is formed, is made of polycrystalline silicon and the predetermined roughness has an amplitude between 0.5 nm RMS and 5.0 nm RMS (typically between 5.0 nm PV and 60.0 nm PV). Even more preferably, this amplitude is between 2.0 nm RMS and 5.0 nm RMS (typically between 20.0 nm PV and 60.0 nm PV). This embodiment is particularly useful for manufacturing SOI structures suitable for RF applications, in which the front surface 2a of the support substrate 2 comprises a polycrystalline silicon charge trapping layer.
[0031] According to yet another embodiment, the front surface 1a, 2a on which the roughened area 31a is formed is made of silicon oxide, the predetermined roughness having an amplitude between 1.0 nm RMS and 60.0 nm RMS.
[0032] It should be noted that if the front surface 1a of the donor substrate 1 has a roughened region 31a, the roughened region 31a is preferentially created before forming the buried brittle surface 11 within the substrate 1 in order to limit damage or premature ripening of the microcavities which constitute it.
[0033] It is also possible to form a pre-roughened area on the front surface of the first material forming the donor substrate 1. For example, the first material can be silicon. The first material of the donor substrate 1 is then thermally oxidized to form an insulating layer (for example silicon oxide). The thickness of the insulating layer can be, for example, 200 nm or less. The insulating layer (corresponding to the additional layer 12 shown in [Figure 3a]) comprises on its free surface (front surface 1 of the donor substrate 1) a roughened area 31a perpendicular to the pre-roughened area formed in the first material. In fact, the oxidation protects at least part of the high-frequency roughness of the first material in the pre-roughened area. The aim is to form a micro-roughness in the pre-roughened area that can give rise after oxidation to a predetermined roughness of the roughened area 31a in the amplitude range mentioned above. After forming the buried brittle surface 11 in the donor substrate 1, the insulating layer 12 is assembled on the support substrate 2 in a bonding assembly 100, the roughened area 31a generating a local unbonded area 31 in the bonding interface 3. In this case, the roughened areas 31a on the front surface 1a of the donor substrate 1 are not created directly by "roughening" the insulating layer 12, but rather result from pre-roughened areas present in the underlying material.
[0034] The formation of this pre-roughened area is disclosed with reference to the donor substrate 1, but of course it can also be applied to the support substrate 2.
[0035] Once the bonded assembly 100 has been formed and contains a local unbonded region 31 within its bonded interface 3, the transfer method according to the invention involves applying a thermal anneal thereto, which causes a spontaneous separation along the buried brittle plane 11. The separation results in the transfer of the thin film 10 from the donor substrate 1 to the support substrate 2, forming a laminated structure 110 (FIG. 4). In the vertical direction of the location of the local unbonded region 31, there is an untransferred region 31b where the thin film 10 has not been transferred. The remaining part of the donor substrate 1′ is also obtained.
[0036] The localized unbonded regions 31 act as fracture initiation points within the buried brittle surface 11 and at or near the regions 31 following microcrack growth in the buried brittle surface 11 due to thermal activation. This fracture initiation occurs sooner than in a bonded assembly 100 that does not include the localized unbonded regions 31. This allows for a short fracture time on the surface 10a of the thin film 10 that provides low surface roughness after transfer. The examples detailed below show a clear improvement in the surface roughness after transfer due to the short fracture time provided by the localized unbonded regions 31 acting as fracture initiation points (FIG. 6, laminate structure 110).
[0037] The local unbonded area 31 according to the invention differs from the prior art in that it is due only to the presence of roughened areas 31a on one or the other of the front surfaces 1a, 2a of the donor substrate 1 and the support substrate 2. No low-frequency topology is involved (bumps, holes, cavities, particles). The internal volume of the local unbonded area 31 thus formed is extremely small. Indeed, the accumulation in this internal volume of various gases, resulting in particular from the evaporation of the water monolayer present on the bonding surfaces 1a, 2a or from the outward diffusion of light species by the front surface 1a of the donor substrate 1, allows a rapid pressurization favoring fracture initiation as soon as the level of maturity of the microcracks in the buried brittle surface 11 allows it.
[0038] The local unbonded area 31 can be located at various positions in the plane of the bonded interface 3, in particular in the central area of the bonded assembly 100, or in the peripheral area, or even in an intermediate area between these two extremes. In the peripheral area, the local unbonded area 31 is preferably at least 1 mm from the peripheral (unbonded) crown of the bonded assembly 100. It is to be recalled that the unbonded crown (visible in [Figure 8b]) is due to the presence of recessed edges and chamfers at the edges of the assembled substrates 1, 2.
[0039] The applicant has observed that by locating a local unbonded region 31 at the center or central region of the bonded assembly 100 (e.g., as shown in FIG. 3b), not only is the final roughness ("haze") of the transferred thin film 10 reduced over its entire surface, but the high density region ZD (FIG. 6), which corresponds to a local peripheral region of very high roughness, is significantly limited or even eliminated.
[0040] The local unbonded region 31 can have different shapes in the plane of the bonded interface 3 (i.e., the (x,y) plane). Some examples are shown in FIG.
[0041] Advantageously, the local unbonded region 31 has a shape in the (x,y) plane, where at least a portion of the contour has a radius of curvature smaller than the radius of a circular bond defect in the same region (FIGS. 7(b), (c), (d)). The applicant has demonstrated that a contour with a locally small radius of curvature, straight sections or singular points (such as, for example, cusps) gives the local unbonded region 31 a good fracture initiation efficiency. Thus, short fracture times with little variation between multiple identical bonded assemblies can be obtained, resulting in a low, uniform and reproducible surface roughness of the thin film 10.
[0042] 8a and 8b show optical microscope images of non-transferred areas 31b (created by regions 31) on a laminate structure 110 obtained by the transfer method of the present invention. These regions 31b have contours with specific shapes as previously described with respect to increasing the effectiveness of the local unbonded regions 31 as fracture initiation points.
[0043] To avoid compromising the transfer integrity of the thin film 10, the local unbonded regions 31 advantageously have at least one lateral dimension in the (x,y) plane of less than 300 micrometers. This is typically achieved by forming roughened regions 31a on one of the front faces 1a, 2a of the substrates 1, 2, having an associated lateral dimension of 200 micrometers or less.
[0044] The roughened areas 31a can be produced by various techniques, including chemical etching, wet etching or dry etching, after protecting the relevant front surfaces 1a, 2a, except for the areas to be roughened.
[0045] Nevertheless, it is advantageous to form this roughened area so as to have as little contact as possible with the remainder of the associated front surface 1a, 2a, in order to reduce the risk of contamination or deterioration (such as scratches) of this surface to be assembled.
[0046] A particularly advantageous technique for forming the roughened regions 31a consists in using laser shots that can cause only surface melting of the material of the front faces 1a, 2a, only in the areas to be roughened. This surface melting typically occurs in the first few nanometers (1 nm to 30 nm) of material on the front faces 1a, 2a, and therefore includes the melting limit regime and does not include any ablation regime that removes or hollows out material. The laser shots used here do not generate topologies, do not generate cavities or bumps, but only high frequency roughness in the abovementioned RMS and PV ranges.
[0047] Preferentially, the laser shot is fired over a very short pulse, typically lasting between 1 ns and 1000 ns, more advantageously between 10 ns and 500 ns. The wavelength of the laser used can be chosen between 100 nm and 550 nm, preferably between 250 nm and 400 nm.
[0048] As an example, if the material is silicon, the laser shot will be 1.8 J / cm using a laser with a wavelength of 308 nm. 2 from 2.5 J / cm 2 It can be performed with energy densities between 1.9 + / - 0.1 J / cm2. Figure 5 shows a curve relating the roughness (RMS and PV) of a single crystal silicon surface irradiated by a laser shot (wavelength 308 nm, pulse duration 160 ns) to the energy density of the laser shot. To generate high frequency roughness of the expected amplitude, it is important to be within the melting limit regime (1.9 + / - 0.1 J / cm2 in this example).2 ). This regime corresponds to a "discontinuous" melting of the irradiated surface, which means that molten and non-molten areas are induced on the surface, leading to the formation of the desired high-frequency roughness. If the energy density is too low (<1.8 J / cm 2 ), melting cannot be achieved and little or no roughness is produced. - If the energy density is too high (>2J / cm 2 ), a "homogeneous" melting of the entire irradiated surface occurs, with uniform recrystallization and potentially minimal roughness.
[0049] It should be noted that if the material of the front surface 1a, 2a (receiving the laser shot) is silicon oxide and the underlying material is silicon, the laser shot will penetrate the silicon oxide layer and cause melting of the silicon, which will induce the formation of wrinkles on the oxide surface. These wrinkles will form the expected high frequency roughness on the front surface 1a, 2a in the irradiated areas.
[0050] As already mentioned with regard to the preferential maximum lateral dimension of the local unbonded area 31, the laser beam irradiates a roughly circular surface having a diameter advantageously less than 200 micrometers. The roughened area 31a created after the laser shot is approximately the size of the irradiated surface.
[0051] To generate a favorable shape of the local unbonded region 31, the roughened region 31a can be created using multiple laser shots, for example, 2 (FIG. 7(b)), 3, 4 (FIG. 7(c)), 5, 6 (FIG. 7(d)) or 15 laser shots. To induce a single local unbonded region 31 with a specific shape, rather than multiple circular regions 31, the surfaces irradiated by successive laser shots must be at least tangential. They may also have a certain percentage of overlap (1% to 95%, or 20% to 70%). It is therefore possible to form a contour of the local unbonded region 31 favorable for fracture initiation, since it locally exhibits a singularity or a small radius of curvature, as shown in FIG. 7(b), (c), (d). With an overlap of about 50%, it is possible to form a capsule-shaped roughened region 31a (FIG. 7(e)), and part of the contour of the local unbonded region 31 thus created has straight lines, which are effective for initiating fracture.
[0052] The shapes of the roughened areas 31a and the local unbonded areas 31 described above are of course not exhaustive and any other geometric or other shapes can be envisaged.
[0053] Representative embodiments: In a first example, a transfer method is used to manufacture a FDSOI (fully depleted SOI) structure, ie an FDSOI structure with a thin top film and a thin buried insulating layer.
[0054] The donor substrate 1 is a single crystal silicon wafer of 300 mm diameter and includes on its front surface 1a an insulating layer 12 of silicon oxide of 35 nm thickness. The buried brittle surface 11 is bombarded with helium and hydrogen ions at energies of 40 keV and 25 keV, respectively, at a rate of 1E16 / cm. 2 and 1E16 / cm 2 The ions are formed by co-implanting at a dose of
[0055] The support substrate 2 is a single crystal silicon wafer having a diameter of 300 mm.
[0056] A laser shot is fired at the center of the support substrate 2, on its front side 2a, so as to cause surface-only melting / recrystallization in the melting limit regime. The laser conditions are: wavelength 308 nm, laser pulse duration 160 ns, energy density 1.9 J / cm. 2 , a quasi-circular irradiation surface with a radius of 65 μm. Therefore, the laser shot is 10 × 10 μm 2 This results in the formation of roughened regions on the irradiated surface with an RMS roughness of 2.3 nm + / - 0.5 nm as measured by AFM over a scan of 100 nm (see, for example, the AFM image in Figure 3b).
[0057] Next, conventional surface treatment (cleaning, plasma activation) of the two substrates 1, 2 is performed with the aim of molecular adhesive bonding.
[0058] Assembly based on direct contact between the front surface 1a, 2a of the donor substrate 1 and the support substrate 2 produces a bonded assembly 100. The bonded assembly 100 includes a local unbonded region 31 within the bonded interface 3 that is perpendicular to the roughened region 31a (at the center of the bonded assembly 100). The diameter of the local unbonded region 31 is on the order of 130-250 μm.
[0059] The fracture annealing, which is carried out in a horizontal furnace (suitable for collectively processing multiple bonded assemblies 100), is applied at 200°C to 400°C.
[0060] The localized unbonded regions 31 make it possible to initiate fracture along the buried brittle planes 11 in a short time, i.e., in the case of isothermal annealing, in about 25% of the average fracture time in the absence of the localized unbonded regions 31. By considering multiple collectively processed bonded assemblies 100, the localized unbonded regions 31 further make it possible to obtain less distributed fracture times.
[0061] The resulting SOI structure 110 after separation has, in place of the local unbonded regions 31 of the bonded assembly 100, untransferred regions 31b of approximately the same size as the regions 31.
[0062] The surface quality 10a of the transferred thin film 10 is improved compared to the SOI 110' structure obtained from a conventional bonded assembly, which is free of local unbonded regions 31. This is especially visible after the application of finishing steps (mainly oxidation and smoothing heat treatments) to repair the transferred thin film 10 and smooth its surface. The SOI structure 110 obtained by the method of the invention does not show dense areas of roughness at the periphery of the thin film 10, unlike the particular SOI structure 110' obtained by the conventional method (which is free of local unbonded regions 31).
[0063] In a second example, a transfer method is used to manufacture an RFSOI (SOI for Radio Frequency Applications) structure, ie a thin surface membrane, an insulating layer and a charge trapping layer on a supporting substrate 2.
[0064] The donor substrate 1 is a single crystal silicon wafer of 300 mm diameter and includes on its front surface 1a an insulating layer 12 of silicon oxide of 200 nm thickness. The buried brittle surface 11 is formed by bombarding hydrogen and helium ions with energies of 35 keV and 50 keV, respectively, at a rate of 1.2E16 / cm. 2 and 1.1E16 / cm 2 The ions are formed by co-implanting at a dose of
[0065] The support substrate 2 is a single crystal silicon wafer having a diameter of 300 mm. On the front surface 2a side of the support substrate 2, a polycrystalline silicon charge trap layer having a thickness of about 1 μm is disposed.
[0066] The laser shots are fired at the peripheral area of the support substrate 2, on its front surface 2a side (hence on the polycrystalline silicon) so as to cause melting / recrystallization of the surface only. In particular, the laser shots are fired at a distance of 3 mm to 10 mm, e.g. 5 mm, from the edge of the support substrate 2.
[0067] The laser conditions were as follows: wavelength 308 nm, laser pulse duration 160 ns, energy density 1.9 J / cm 2 , a quasi-circular irradiation surface with a radius of 65 μm. Therefore, the laser shot is 10 × 10 μm 2This results in the formation of roughened areas on the irradiated surface having an RMS roughness of 2.7 nm + / - 0.5 nm as measured by AFM over a scan of 100 nm.
[0068] Next, conventional surface treatment (cleaning, plasma activation) of the two substrates 1, 2 is performed with the aim of molecular adhesive bonding.
[0069] Assembly based on direct contact between the front surface 1a, 2a of the donor substrate 1 and the support substrate 2 produces a bonded assembly 100. The bonded assembly 100 includes local unbonded regions 31 perpendicular to the roughened regions 31a (at the edges of the bonded assembly 100) within the bonded interface 3. The diameter of the local unbonded regions 31 is on the order of 130-250 μm.
[0070] The fracture annealing, which is carried out in a horizontal furnace (suitable for collectively processing multiple bonded assemblies 100), is applied at 200°C to 550°C.
[0071] The localized unbonded regions 31 allow for failure initiation along the buried brittle plane 11 in a short time (approximately 30% of the average failure time in the absence of the localized unbonded regions 31). By considering multiple collectively processed bonded assemblies 100, the localized unbonded regions 31 further allow for a less distributed failure time to be obtained.
[0072] The resulting SOI structure 110 after separation has a non-transferred area in place of the local unbonded area 31 of the bonded assembly 100, the size of which is substantially the same as the size of the area 31, typically less than 250 μm.
[0073] The surface quality 10a of the transferred thin film 10 is improved compared to the SOI 110' structure obtained from a conventional bonded assembly, which does not have local unbonded regions 31. This is especially visible after the application of finishing steps (mainly oxidation and smoothing heat treatments) to repair the transferred thin film 10 and smooth its surface. The map in FIG. 6 of the SOI structure 110 obtained by the method of the invention shows that, unlike the SOI structure 110' obtained by the conventional method (which does not have local unbonded regions 31), there are no dense areas of roughness around the thin film 10. This also indicates a lower overall roughness level over the entire surface of the thin film 10 of the SOI structure 110.
[0074] In a third example, a transfer method is used to manufacture an SOI structure, ie an SOI structure comprising a thin surface film, an insulating layer and a supporting substrate 2 .
[0075] The donor substrate 1 is a single crystal silicon wafer with a diameter of 300 mm.
[0076] Laser shots are fired at the central or peripheral area of the donor substrate 1 on its front side 1a (hence on the silicon) to cause melting / recrystallization of the surface only. The laser conditions are: wavelength 308 nm, laser pulse duration 160 ns, energy density 1.9 J / cm. 2 , a quasi-circular irradiation surface with a radius of 65 μm. Therefore, the laser shot is 10 × 10 μm 2 This results in the formation of a pre-roughened area on the irradiated surface having an RMS roughness of 2.3 nm + / - 0.5 nm as measured by AFM over a scan of 100 nm.
[0077] After a standard cleaning of the donor substrate 1, a thermal oxidation is performed, typically at a temperature in the range of 900°C to 1050°C. After this oxidation, the front surface 1a of the donor substrate 1 comprises an insulating layer 12 of silicon oxide with a thickness of 100 nm. Perpendicular to the pre-roughened region, a roughened region 31a is present on the free surface 1a of the insulating layer 12. This roughened region 31a has an RMS roughness substantially identical to that of the pre-roughened region, i.e. about 2.2 nm + / - 0.5 nm (10 x 10 µm 2(AFM with scanning).
[0078] The buried brittle surface 11 is formed by irradiating hydrogen ions and helium ions through the insulating layer 12 in the donor substrate 1 with energies of 35 keV and 50 keV, respectively, at a rate of 1.2E16 / cm 2 and 1.1E16 / cm 2 The ions are formed by co-implanting at a dose of
[0079] The support substrate 2 is a single crystal silicon wafer having a diameter of 300 mm.
[0080] Next, conventional surface treatment (cleaning, plasma activation) of the two substrates 1, 2 is performed with the aim of molecular adhesive bonding.
[0081] Assembly based on direct contact between the front surface 1a, 2a of the donor substrate 1 and the support substrate 2 produces a bonded assembly 100. The bonded assembly 100 includes local unbonded regions 31 perpendicular to the roughened regions 31a within the bonded interface 3. The diameter of the local unbonded regions 31 is approximately 100 μm.
[0082] The fracture annealing, which is carried out in a horizontal furnace (suitable for collectively processing multiple bonded assemblies 100), is applied at 200°C to 550°C.
[0083] The localized unbonded regions 31 allow for fracture initiation along the buried brittle planes 11 in a short time (approximately 30% of the average fracture time in the absence of the localized unbonded regions 31). By considering multiple collectively processed bonded assemblies 100, the localized unbonded regions 31 allow for a less distributed fracture time to be obtained.
[0084] The resulting SOI structure 110 after separation has a non-transferred area in place of the local unbonded area 31 of the bonded assembly 100, the size of which is substantially the same as the size of the area 31, typically less than 200 μm.
[0085] The surface quality 10a of the transferred thin film 10 is improved compared to the SOI structure 110' obtained from a conventional bonded assembly with no local unbonded regions 31, indicating the absence of high density regions of roughness ZD around the periphery of the thin film 10 and a low overall level of roughness across the entire surface of the thin film 10.
[0086] In a fourth example, a transfer printing method is used to manufacture an RFSOI (SOI for Radio Frequency Applications) structure, ie a thin surface film, an insulating layer and a charge trapping layer on a supporting substrate 2.
[0087] The donor substrate 1 is a monocrystalline silicon wafer of diameter 200 mm, including on its front surface 1a an insulating layer 12 of silicon oxide of thickness 400 nm. The buried embrittled surface 11 is formed by ion bombardment with hydrogen ions of energy 50 keV, 6E16 / cm 2 The ions are formed by implanting at a dose of 100 .mu.m.
[0088] The support substrate 2 is a single crystal silicon wafer having a diameter of 200 mm. On the front surface 2a side of the support substrate 2, a polycrystalline silicon charge trap layer having a thickness of about 2 μm is disposed.
[0089] A number of laser shots are fired at the peripheral area of the support substrate 2, on its front surface 2a side (hence on the polycrystalline silicon), so as to cause melting / recrystallization of only the surface. In particular, the laser shots are fired at a distance of 3 mm to 10 mm, e.g. 5 mm, from the edge of the support substrate 2.
[0090] The laser conditions were as follows: wavelength of each laser shot was 308 nm, laser pulse duration was 160 ns, and energy density was 2 J / cm 2 , a quasi-circular irradiated surface with a radius of 65 μm. A sequence of 11 aligned laser shots with approximately 10% overlap of adjacent irradiated surfaces results in the formation of a roughened region 31a with a shape similar to that shown in FIG. 7(d). The RMS roughness of the roughened region is 10×10 μm 2 It is about 3nm + / - 0.5nm as measured by AFM in a scan.
[0091] Next, conventional surface treatment (cleaning, plasma activation) of the two substrates 1, 2 is performed with the aim of molecular adhesive bonding.
[0092] Assembly based on direct contact between the front surface 1a, 2a of the donor substrate 1 and the support substrate 2 produces a bonded assembly 100. The bonded assembly 100 comprises within the bonded interface 3 a local unbonded region 31 perpendicular to the roughened region 31a. In the plane of the bonded interface (the (x,y) plane), the local unbonded region 31 has a length of about 2 mm and a width of about 100-250 μm.
[0093] The fracture annealing, carried out in a horizontal furnace, is applied at 200°C to 550°C.
[0094] The localized unbonded regions 31 allow for fracture initiation along the buried brittle planes 11 in a short time (approximately 30% of the average fracture time in the absence of the localized unbonded regions 31). By considering multiple collectively processed bonded assemblies 100, the presence of the localized unbonded regions 31 allows for a less distributed fracture time to be obtained.
[0095] The resulting SOI structure 110 after separation has, instead of the local unbonded regions 31 of the bonded assembly 100, untransferred regions 31b of approximately the same size as the regions 31 (FIG. 8b).
[0096] The surface quality 10 a of the transferred thin film 10 is improved compared to the SOI structure resulting from a conventional bonded assembly without local unbonded regions 31 .
[0097] It should be noted that the local unbonded area 31 having the particular shape disclosed in the fourth embodiment also provides the aforementioned advantages when implemented in the bonded assemblies 100 disclosed in the first, second and third embodiments.
[0098] Naturally, the invention is not limited to the embodiments described and variants can be envisaged without departing from the scope of the invention as defined by the claims.
Claims
1. A method for transferring a thin film (10) onto a support substrate (2), comprising the steps of: - providing a bonded assembly (100) comprising a donor substrate (1) and said support substrate (2) assembled by direct bonding at their respective front faces (1 a, 2 a) next to a bonding interface (3), said bonded assembly (100) having local unbonded areas (31) within said bonding interface (3), said donor substrate (1) further comprising a buried brittle surface (11); Separating along the buried brittle plane (11), initiated at the local unbonded region (31), after microcrack growth in the plane (11) by thermal activation, which separation results in the transfer of the thin film (10) from the donor substrate (1) to the support substrate (2), The method is characterized in that the local unbonded regions (31) are generated only by roughened regions (31 a) intentionally created on at least one of the front surfaces (1 a, 2 a) of the donor substrate (1) and the support substrate (2) before assembly, the roughened regions (31 a) being topology-free and having a predetermined roughness with an amplitude between 0.5 nm RMS and 60.0 nm RMS.
2. 2. The transfer method according to claim 1, wherein the roughened areas are produced by at least one laser shot that produces only a surface melting of the material constituting the donor substrate (1) and / or the support substrate (2) on the side of their respective front faces (1 a, 2 a) within the first 1 to 30 nanometers.
3. 3. The transfer method according to claim 2, wherein the laser shot has a pulse duration of 1 ns to 1000 ns, preferentially 10 ns to 500 ns.
4. The laser shot is 1.8 J / cm using a laser with a wavelength of 308 nm. 2 ~2.5 J / cm 2 The transfer method of claim 3 , wherein the transfer method is carried out at an energy density of
5. The transfer method according to any one of claims 2 to 4, wherein the laser shots irradiate a circular surface having a diameter of less than 200 micrometers.
6. 6. The transfer method of claim 5, wherein the roughened area (31 a) is created by a plurality of laser shots, for example 2 to 15 laser shots, and the adjacent circular surfaces irradiated by successive laser shots are tangential or have an overlap rate of 1% to 95%.
7. the front surface (1a, 2a) on which the roughened area (31a) is formed is made of monocrystalline silicon and the predetermined roughness has an amplitude between 0.5 nm RMS and 4.0 nm RMS, preferentially between 1.0 nm RMS and 2.5 nm RMS, or the front surface (1a, 2a) on which the roughened area (31a) is formed is made of polycrystalline silicon and the predetermined roughness has an amplitude between 0.5 nm RMS and 5.0 nm RMS, preferentially between 2.0 nm RMS and 5.0 nm RMS, or The transfer method according to claim 1, wherein the front surface (1a, 2a) on which the roughened area (31a) is formed is made of silicon oxide, and the predetermined roughness has an amplitude between 1.0 nm RMS and 60.0 nm RMS.
8. 2. The transfer method of claim 1, wherein the roughened area (31a) is created on the front surface (1a) of the donor substrate (1) before forming the buried brittle surface (11) in the donor substrate (1).
9. - said donor substrate (1) is made on its front side (1a) from a first material, a pre-roughened area being formed on said surface of said material; thermal oxidation of the first material of the donor substrate (1) is carried out after the formation of the pre-roughened area and before the formation of the buried brittle surface (11) to form an insulating layer that is assembled on the support substrate (2) in the bonding assembly (100), the insulating layer comprising on its free surface the roughened area (31 a) perpendicular to the pre-roughened area; The transfer method according to claim 8 .
10. - said support substrate (2) is made on its front side (2a) from a first material, a pre-roughened area being formed on the surface of said material; - after the formation of the pre-roughened area, a thermal oxidation of the first material of the support substrate (2) is carried out, resulting in an insulating layer that is assembled on the donor substrate (1) in the bonding assembly (100), the insulating layer comprising on its free side a roughened area (31 a) perpendicular to the pre-roughened area, The transfer method according to claim 1 .
11. The transfer method according to claim 1, wherein the local unbonded region (31) has a shape in which at least a portion of its contour has a radius of curvature smaller than the radius of a circular bond defect in the same region on the surface of the bonded interface (3).
12. 2. The transfer method of claim 1, wherein the local unbonded area (31) has at least one lateral dimension at the face of the bond interface (3) that is less than 300 micrometers.
13. 2. The transfer method according to claim 1, wherein the local unbonded area (31) is located in a central area of the bonded assembly (100) at the surface of the bonded interface (3).
14. 2. The transfer method of claim 1, wherein the thin film (10) from the donor substrate (1) is monocrystalline silicon and the support substrate (2) comprises monocrystalline silicon to form a stacked SOI structure (110).