METHOD FOR ETCHING A SUBSTRATE WITH A THIN SURFACE LAYER TO IMPROVE THE THICKNESS UNIFORMITY OF THE THIN SURFACE LAYER - Patent application

Abstract

The present invention relates to a method for etching a major surface (1 a) of a thin film (11) of a substrate (1), the method comprising the step of immersing the substrate (1) in an etching bath, wherein the substrate (1) is oriented relative to the bath (100) such that, as the substrate (1) is introduced into the bath (100), the major surface (1 a) is gradually immersed at an introduction rate from an initial introduction point (PII) to a final introduction point (PFI), and as the substrate (1) exits the bath (100), the major surface (1 a) gradually emerges at an escape rate from the initial escape point (PIS) to a final escape point (PFS), the method being characterized in that the introduction rate is selected to etch the major surface (1 a) according to a first non-uniform profile between the initial introduction point (PII) and the final introduction point (PFI) and / or the escape rate is selected to etch the major surface (1 a) according to a second non-uniform profile between the initial escape point (PIS) and the final escape point (PFS).

Description

[Technical field]

[0001] The present invention relates to the fields of microelectronics, optoelectronics, photonics, etc. In particular, the present invention relates to a method for etching substrates using a chemical bath multi-wafer cleaner, the present invention being particularly suitable for SOI (silicon-on-insulator) substrates with a very thin surface layer of silicon. [Background technology]

[0002] More and more applications based on SOI substrates require a very high uniformity of the thickness of the thin surface layers of silicon (intended to house or support components). For example, in digital applications, the thin layers of FDSOI (fully depleted SOI) substrates must exhibit very low variations in thickness, since they affect the threshold voltage of the transistors produced on said layers. In photonics applications, the performance of filter or modulator devices is also significantly affected by the non-uniformity of the thickness of the thin layers of the SOI substrate.

[0003] Therefore, the specifications in terms of thickness and uniformity have become very demanding, with within-wafer (WiW) and wafer-to-wafer (WtW) non-uniformities of less than a few angstroms, typically less than 4 A, being expected for layers with thicknesses typically less than 50 nm. Such uniformity is difficult to achieve, because the succession of steps when producing SOI substrates results in an accumulation of contributions to the non-uniformity of the thin layers.

[0004] One known solution to correct the non-uniformity of the thickness of the thin layer is to carry out a localized etching of said layer, for example using a plasma etching method as described in US Patent Publication No. 20140234992, or in particular using a cluster ion beam etching method as described in WO 2013003745. However, this type of solution has a drawback: etching the surface of the thin layer creates surface areas of amorphous silicon, which tend to cause electrical problems and therefore must be removed. Removal of the amorphous areas leads to an increase in surface roughness, which negatively affects the performance of devices produced on the thin layer.

[0005] WO2004015759 proposes an alternative solution implementing localized sacrificial thermal oxidation, which locally consumes a greater or lesser thickness of the thin layer to correct thickness non-uniformities. A disadvantage of this approach is that localized temperature gradients are not easy to introduce into silicon wafers, and therefore the solution for non-uniformity correction may be limited. Summary of the Invention [Problem to be solved by the invention]

[0006] The present invention relates to an alternative solution to the prior art solutions, aiming to completely or partially overcome the above mentioned drawbacks. In particular, the present invention relates to a method for preparing a thin layer with the aim of achieving an improved uniformity of the thickness of said thin layer, the implementation of which is simple and reproducible. Such a preparation method may be advantageously implemented to produce SOI structures with a very thin surface layer of silicon. [Means for solving the problem]

[0007] The present invention relates to a method for etching a main surface of a substrate provided with a skin layer, the main surface corresponding to a free face of said skin layer, the method comprising the step of immersing the substrate in an etching bath in order to expose the main surface to an etchant, the substrate being: When the substrate is introduced into the bath, the major surface is gradually immersed at an introduction rate from an initial introduction point to a final introduction point; As the substrate is removed from the bath, the major surface gradually emerges at a removal rate from an initial removal point to a final removal point. and oriented relative to the vessel such that The method includes: the introduction rate is selected to etch the major surface according to a first non-uniform profile between the initial introduction point and the final introduction point; and / or the removal rate being selected to etch the major surface according to a second non-uniform profile between the initial removal point and the final removal point. It is characterized by:

[0008] The present invention has the following other advantageous, non-limiting features, which may be considered alone or in any technically feasible combination:

[0009] the first etching profile defines a variation in etched thickness across the major surface between the initial and final incorporation points of 0.15 nm or more, or even 0.2 nm or more; and / or The second etch profile defines a variation in etched thickness across the major surface between the initial and final removal points of 0.15 nm or greater, or even 0.2 nm or greater.

[0010] The initial introduction point corresponds to the final removal point and the final introduction point corresponds to the initial removal point.

[0011] The introduction and / or removal speed is between 25 cm / s and 0.1 cm / s, preferably between 10 cm / s and 0.5 cm / s.

[0012] When the substrate is introduced into the chamber and removed from the chamber, the introduction rate and / or removal rate, respectively, are not constant.

[0013] The etching bath contains a SC1 (standard clean 1) solution having ammonia, hydrogen peroxide, and deionized water in a ratio of 1 / 2 / 20, respectively.

[0014] The substrate is an SOI substrate, the surface layer of the substrate having a thickness less than 50 nm and arranged on an insulating layer which is arranged on a carrier substrate made from silicon.

[0015] The thin layer is made of single crystal silicon.

[0016] Other features and advantages of the present invention will become apparent from the following detailed description of the invention which is given with reference to the accompanying drawings. [Brief description of the drawings]

[0017] [Figure 1] 1 shows a substrate suitable for being processed using a production method according to the present invention; [Diagram 2] 1A-1D show method steps for transferring a thin layer onto a carrier substrate as known from the prior art; [Diagram 3] FIG. 2 shows a map of the non-uniformity of a thin layer thickness of a substrate suitable for being processed using a production method according to the present invention. [Figure 4a] 2 is a simplified diagram of a substrate as it is introduced into an etching bath in an etching method according to the present invention; FIG. [Figure 4b] 2 is a simplified diagram of a substrate as it is removed from an etching bath in an etching method according to the present invention; FIG. [Figure 5a] FIG. 2 shows a map of a non-uniform etch profile resulting from an etching method according to the present invention across a thin layer of a substrate. [Figure 5b] FIG. 2 shows the normalized etching gradients in north / south profiles along lines numbered 1 to 5 resulting from an etching method according to the invention across a thin layer of a substrate. [Figure 6]FIG. 1 shows a box plot illustrating the etched thickness gradient in an example of a non-uniform etch profile obtained across a plurality of substrates processed using the etching method according to the present invention, the box plot also showing the etched thickness gradient for conventional use of the etching bath. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0018] In the description, the same references in the figures may be used for the same types of elements. The figures are schematic representations and are not drawn to scale for ease of reading. In particular, the thickness of layers along the z-axis is not drawn to scale with respect to the lateral dimensions along the x- and y-axes, and the relative thicknesses of layers with respect to one another are not necessarily taken into account in the figures.

[0019] The present invention relates to a method for etching a main surface 1a of a substrate 1. The main surface 1a is in particular the front surface of the substrate 1, i.e. the front surface whose physical properties (roughness, uniformity, cleanliness, etc.) are closely monitored for the purpose of producing components.

[0020] The etching method is a wet etching method, which involves immersing the substrate 1 in a liquid etching bath 100 to expose the main surface 1a to an etchant (Figure 4a).

[0021] The invention is described in the particular case of a substrate 1 comprising a thin surface layer 11 arranged on a carrier substrate 20 (FIG. 1). The main surface 1a in this case corresponds to the free face of said thin layer 11, on which the components are intended to be produced. The thin layer 11 is advantageously made of a monocrystalline material with a very low defect density.

[0022] In a manner known per se, the substrate 1 may be formed from a thin layer 11 arranged directly on a carrier substrate 20 (FIG. 1(a)), or the substrate 1 may comprise an intermediate layer 30 between the thin layer 11 and the carrier substrate 20 (FIG. 1(b)). This intermediate layer 30 may be insulating, as is the case for an SOI type substrate 1. Alternatively, the intermediate layer 1 may be a semiconducting or conductive substrate, depending on the intended properties of the substrate 1.

[0023] There are many known methods of transferring a thin monocrystalline layer 11 to a carrier substrate 20, including the Smart Cut™ method, which is based on implanting light seeds to create an embedded weakened plane 12 in the donor substrate 10 (FIG. 2(a)). This method is also based on direct bonding by molecular adhesion to form a bonded interface 21 between the two substrates 10 and 20 and bond the two substrates together (FIG. 2(b)).

[0024] Donor substrate 10 and carrier substrate 20 are preferably in the form of circular wafers, each having a diameter of 200 mm or 300 mm, or even 450 mm, and a thickness typically between 500 and 900 microns.

[0025] Of course, either (or both) of the substrates 10, 20 may comprise all or part of the intermediate layer 30 on the face of the substrate to be bonded, when the intermediate layer 30 is required in the final lamination of the substrate 1. Separation along the embedded weakened plane 12 allows the thin layer 11 to be transferred from the donor substrate 10 to the carrier substrate 20 (FIG. 2(c)). Immediately after separation, the free surface of the thin layer 11 shows a high level of roughness (of the order of 40-80 A, measured with an atomic force microscope on a scan of 30×30 microns). During the surface finishing techniques, it is particularly possible to implement sacrificial oxidation heat treatments, etching and / or surface reconstructions in a neutral or reducing atmosphere. These treatments are generally carried out at high temperatures, depending on the material constituting the thin layer 11. For example, in the case of a silicon layer, the oxidation may be carried out between about 750° C. and 1100° C., and the smoothing in a neutral or reducing atmosphere between 950° C. and 1250° C. At the end of these finishing treatments, a substrate 1 with its surface thin layer 11 is obtained. As mentioned in the introduction, the specifications in terms of thickness uniformity of said layer 11 are becoming increasingly demanding, especially as its thickness decreases.

[0026] In particular, in the case of layers 11 having very thin thicknesses of less than 50 nm or even less than 20 nm, the applicant has identified that the non-uniformity in the thickness of the thin layer 11 obtained using the Smart Cut method is due in particular to two non-concentrically symmetric contributions.

[0027] The first contribution to the non-uniformity of the thin layer 11 comes from the inherent variability of the multiple successive steps performed to produce the substrate 1, which results in unpredictable ("random") variations in the thickness of the layer 11 that are therefore difficult to eliminate.

[0028] The second contribution comes from production steps with non-concentric symmetry, such as ion implantation, separation in buried weakening planes, and / or batch heat treatments with supports that locally hold the substrate 1. This contribution produces a more or less noticeable inhomogeneity in the thickness of the thin layer 11 between the cardinal orientations N, S, E, O of the substrate 1. For example, in FIG. 3, a thickness inhomogeneity of the order of 0.6 nm can be seen in the north-south direction between the two cardinal orientations N (north) and S (south). Note that the scale of values ​​in FIG. 3 is in Angstroms. The thin layer 11 is thinner near the orientation S and thicker near the orientation N. This inhomogeneity profile is repeated substantially along all of the imaginary lines that can be drawn parallel to the north-south direction. In other words, the north region of the thin layer 11 (closer to the orientation N) appears thicker than the south region (closer to the orientation S).

[0029] It should be noted that the thickness non-uniformity map in Fig. 3 is established based on measuring the thickness of thin layer 11 by ellipsometry or reflectometry. In this example, thin layer 11 is made of silicon and has an average thickness of 12 nm.

[0030] The above process for generating non-uniformity with non-concentric symmetry is carried out simultaneously on batches of tens or even hundreds of substrates 1 in a multi-wafer apparatus. To ensure traceability and reproducibility of the performance in the different processes across different batches of substrates 1, it is common practice to place all of the substrates 1 in the same position, which is identifiable by the presence of a flat or notch 22 made on the edge of each substrate 1. In practice, the notch will be at the top (North cardinal direction) or at the bottom (South cardinal direction - as in the example of Fig. 3) or at a defined angle between two of the cardinal directions. The non-uniformity profile will therefore appear in a relatively reproducible manner across all of the substrates 1 of the batch that are processed simultaneously or successively in the different processes.

[0031] The etching method according to the present invention specifically addresses the second mentioned contribution to thickness non-uniformity.

[0032] As mentioned above, the etching method aims at exposing the main surface 1a of the substrate 1 to an etchant by immersion in a bath 100 containing a liquid solution.

[0033] The substrate 1 is oriented with respect to the etching bath 100 such that, as the substrate 1 is introduced into the bath 100, the main surface 1a is gradually immersed (FIG. 4a) from an initial introduction point PII to a final introduction point PFI reflecting the total immersion of the substrate 1. In other words, the main surface 1a forms a non-zero angle with the surface of the solution contained in the bath 100. Typically, this angle is of the order of 90°.

[0034] When the substrate 1 is removed from the bath 100 at S, the main surface 1a gradually emerges from an initial removal point PIS to a final removal point PFS reflecting the total removal of the substrate 1 (FIG. 4b).

[0035] According to the etching method of the present invention, the rate of introduction of substrate 1 into bath 100 and / or the rate of removal of substrate 1 from bath 100 are selected to produce a non-uniform etching profile across major surface 1a.

[0036] More specifically, the introduction rate may be selected to etch the major surface 1a according to a first non-uniform profile between the initial introduction point PII and the final introduction point PFI. Alternatively or additionally, the removal rate may be selected to etch the major surface 1a according to a second non-uniform profile between the initial removal point PIS and the final removal point PFS.

[0037] Advantageously, the first etching profile defines a variation (or gradient) in the etched thickness across the main surface 1a between the initial introduction point PII and the final introduction point PFI of 0.15 nm or more, or even 0.2 nm or more.

[0038] Similarly, the second etching profile preferably defines a variation (or gradient) in the etched thickness across the main surface 1a between the initial removal point PIS and the final removal point PFS of 0.15 nm or more, or even 0.2 nm or more.

[0039] The initial introduction point PII may correspond to the final removal point PFS, and the final introduction point PFI may correspond to the initial removal point PIS.

[0040] Optionally, in order to more precisely adjust the non-uniform etching profile to a particular type of non-uniformity in the thickness of the thin layer 11, a rotation by a defined angle can be applied to the substrate 1 so that the introduction points PII, PFI and the removal points PIS, PFS are different from each other.

[0041] According to yet another option, the etching method may be applied multiple successive times, rotating the substrate 1 between each iteration, to precisely tailor the non-uniform etching profile in each iteration to a particular type of non-uniformity in the thickness of the thin layer 11.

[0042] The introduction and removal rates of substrates 1 into and from liquid chemical bath multi-wafer cleaning or etching equipment are customarily fast. For example, they are of the order of 30 cm / s for a substrate with a diameter of 300 mm. Thus, each substrate 1 enters the bath in less than 2 seconds in order to obtain the best possible uniformity of treatment in the process.

[0043] In contrast to this principle, in the etching method of the invention, at least one of the introduction and removal rates of the substrate 1 (or of the substrates 1) is advantageously reduced, which may be between 25 cm / s and 0.1 cm / s, or even more particularly between 10 cm / s and 0.5 cm / s. As will be explained in more detail in a preferred embodiment below, reducing the rate at which the substrate 1 is introduced into and / or removed from the bath 100 produces a non-uniform etching profile and makes it possible to compensate for the non-uniformity in the thickness of the thin layer 11 due to the second contribution mentioned above.

[0044] The introduction rate and / or removal rate may be selected to be non-constant to compensate for the non-linear non-uniformity profile across the diameter of the substrate 1 .

[0045] According to one preferred embodiment, the thin layer 11 is made of single crystal silicon and the solution in the etching bath 100 is an SC1 ("Standard Clean 1") solution based on deionized water, ammonia (NH3), and hydrogen peroxide (H2O2). The SC1 solution, typically used to clean silicon wafers, is particularly effective at removing particulate contaminants from the treated surfaces.

[0046] The etching method according to the invention uses, for example, an SC1 solution with an etching rate of the thin silicon layer 11 of the order of 0.5 nm / min. The volume ratio of each of the three compounds NH3 / H2O2 / H2O of the solution is 1 / 2 / 20 and the temperature of the bath is 70° C. Megasonic agitation of the solution is also preferred.

[0047] By way of example, Figure 5a shows a non-uniform etching profile obtained over a thin layer 11 of a substrate 1 having a diameter of 300 mm in the etching bath SC1 described above, using an introduction speed of the substrate 1 into the bath of 23 cm / s, a residence time in the bath of 90 s, and a removal speed of 0.67 cm / s (i.e. 45 s between the appearance of the initial removal point PIS and the final removal point PFS). In this example, the initial introduction point PII coincides with the final removal point PFS, the final introduction point PFI coincides with the initial removal point PIS, and the notch 22 is at the final introduction point PFI (also at the initial removal point PIS).

[0048] The removal rate, more particularly, is decreased such that when the substrate is removed from the bath, the non-uniform etching profile of Figure 5a (or the second non-uniform etching profile mentioned above) is substantially formed. The etched thickness gradient between the two introduction points PII, PFI (or between the two removal points PIS, PFS) is of the order of 3A, the etched thickness being larger at the initial introduction point PII (or the final removal point PFS) and smaller at the final introduction point PFI (or the initial removal point PIS).

[0049] Figure 5b shows the normalized etching gradient in a North / South profile along the lines numbered 1 to 5 on a substrate 1 subjected to an etching method according to the example described above. It is found that the etched thickness gradient between the two introduction points PII, PFI (or between the two removal points PIS, PFS) is of the order of 3A.

[0050] The reproducibility of this second profile was checked across multiple substrates 1 simultaneously processed in the etching bath 100. The graph in Figure 6 shows in the form of a box plot the results of the thickness gradient (in Angstroms) of the second profile compared to the results obtained when the introduction and removal rates are standard, i.e. greater than 25 cm / s.

[0051] An etching gradient across the substrate 1 between the two introduction points PII, PFI is found statistically clear, equal on average ("average" in FIG. 6) to 2.8 A, with a standard deviation ("standard deviation" in FIG. 6) of the order of 0.3 A.

[0052] Applying a second non-uniform etching profile to a particular substrate 1 exhibiting a thickness non-uniformity, as shown in Figure 3, while taking care to place the notch 22 at the final introduction point PFI (and at the initial removal point PIS), allows the uniformity of the thin layer 11 to be improved. In particular, the second profile will allow a greater thinning of the northern area (initially the thickest area) relative to the southern area (initially the thinnest area) of said substrate 1. Thus, an initial non-uniformity of the thin layer 11 of the order of 0.6 nm is reduced by the etching method according to the invention to less than 0.4 nm.

[0053] Note that the introduction speed could also be limited to between 25 cm / s and 0.1 cm / s to more precisely tune the differential removal between the two introduction points or removal end points.

[0054] The etching method according to this preferred embodiment may be easily integrated into a cleaning sequence in a chemical bath multi-wafer cleaning apparatus (or "wet bench"). For example, the etching method may be integrated into an ozone / SC1 / SC2 type sequence, replacing the SC1 step or inserted between the ozone and SC1 steps. The simple integration of the etching method into the cleaning sequence is particularly advantageous, since the rinsing and loading times of a batch of substrates 1 are shared and are not doubled by separating the stages.

[0055] The etching method according to the invention is particularly suitable for correcting non-concentric symmetrical inhomogeneities in the superficial thin layer 11 of the SOI substrate 1, said thin layer 11 and the underlying insulating layer 30 having a thickness less than 50 nm.

[0056] Needless to say, the invention is not limited to the described embodiments and examples, and modified implementations may be applied to the invention without departing from the scope of the invention as defined by the claims.

[0057] Although the invention has been described and illustrated with particular reference to a preferred embodiment, the invention may of course be transferred to thin layers 11 made of materials other than single crystal silicon (e.g. silicon oxide, etc.) and implemented with etching bath solutions other than SC1 (e.g. hydrofluoric acid HF).

Claims

1. A method for etching a main surface (1a) of a substrate (1) provided with a skin layer (11), said main surface (1a) corresponding to a free face of said skin layer (11), said method comprising the steps of immersing said substrate (1) in an etching bath (100) for exposing said main surface (1a) to an etchant, said substrate (1) being: When the substrate (1) is introduced into the bath (100), the main surface (1a) is gradually immersed at an introduction rate from an initial introduction point (PII) to a final introduction point (PFI); When the substrate (1) is removed from the bath (100), the main surface (1 a) gradually emerges at a removal rate from an initial removal point (PIS) to a final removal point (PFS); and oriented relative to the vessel (100) such that The method further comprises the steps of: the introduction rate is selected to etch the main surface (1 a) according to a first non-uniform profile between the initial introduction point (PII) and the final introduction point (PFI); and / or the removal rate being selected to etch the main surface (1a) according to a second non-uniform profile between the initial removal point (PIS) and the final removal point (PFS). It is characterized by: The method further comprises the steps of: A rotation by a defined angle is applied to the substrate (1) such that the initial introduction point (PII) and the final removal point (PFS) are different from each other and the final introduction point (PFI) and the initial removal point (PIS) are different from each other. Characterized by: How to etch.

2. and / or the first non-uniformity profile defines an etched thickness variation across the main surface (1 a) between the point of initial introduction (PII) and the point of final introduction (PFI) of 0.15 nm or more, or even 0.2 nm or more; the second non-uniformity profile defines an etched thickness variation across the main surface (1 a) between the initial removal point (PIS) and the final removal point (PFS) of 0.15 nm or more, or even 0.2 nm or more; The etching method of claim 1 .

3. 3. An etching method according to claim 1 or 2, wherein the introduction rate and / or the removal rate is between 25 cm / s and 0.1 cm / s, preferably between 10 cm / s and 0.5 cm / s.

4. 4. The etching method according to claim 1, wherein the introduction rate and / or the removal rate, respectively, of the substrate (1) when it is introduced into and removed from the bath (100) are not constant.

5. 5. The method of etching according to any one of claims 1 to 4, wherein the etching bath (100) contains a SC1 (standard clean 1) solution having ammonia, hydrogen peroxide and deionized water in a ratio of 1 / 2 / 20, respectively.

6. 6. The etching method according to claim 1, wherein the substrate (1) is an SOI substrate and the thin surface layer (11) of the substrate (1) is arranged on an insulating layer (30) having a thickness less than 50 nm and arranged on a carrier substrate (20) made of silicon.

7. 7. The etching method according to claim 5 or 6, wherein said thin layer (11) is made of monocrystalline silicon.

Images