Method of thinning surface layer of silicon-on-insulator substrate

Abstract

The invention relates to a method for thinning a surface layer of a silicon-on-insulator substrate, comprising: a) a step of measuring the thickness of the surface layer in order to obtain its average thickness (Eaverage) and its thickness unevenness (U), followed by performing the following steps in a single wafer chemical treatment apparatus: b) performing n oxidation and etching cycles on the front side of the surface layer, wherein n is an integer greater than or equal to 1, where each cycle involves dispensing ozonized water followed by dispensing hydrofluoric acid, c) performing a non-uniform etching on the front side of the surface layer, which involves dispensing a silicon etching solution in the central region for a period of time, referred to as a non-uniform etching time (tnu), d) performing a uniform etching on the front side of the surface layer, this involves dispensing the silicon etching solution with a sweeping motion from the edge to the center and again back of the surface layer for a period of time, referred to as a uniform etching time (tu). As the average thickness (Eaverage) and the thickness non-uniformity (U) are known, before steps b), c) and d) are carried out, the average thickness (Eaverage) and the thickness non-uniformity (U) are determined. Determining the number of cycles (n), the non-uniform etch time (tnu) and the uniform etch time (tu) based on a model correlating the number of cycles (n), the non-uniform etch time (tnu) and the uniform etch time (tu) with an average etch thickness and an average etch non-uniformity of the surface layer, the average etch non-uniformity is defined as a difference between average etch thicknesses in a central region and a peripheral region of the surface layer.

Description

Technical Field

[0001] This invention relates to the field of semiconductors, and more particularly to the field of silicon-on-insulator (SOI) substrates. The invention relates to a method for thinning the surface layer of a silicon-on-insulator substrate in a single-wafer chemical processing tool, the method being optimized to achieve a surface layer inhomogeneity of less than or equal to + / - 0.4 nm. Background Technology

[0002] The increasing number of applications based on silicon-on-insulator (SiI) substrates demands extremely high uniformity in the thickness of the silicon surface layer (hereinafter referred to as the active layer). For example, in digital applications, the active layer of a fully depleted silicon-on-insulator (FDSOI) substrate must exhibit minimal thickness variation, as this variation affects the threshold voltage of the transistors generated in and / or on the active layer; in photonics applications, the performance of filters or modulators is also strongly affected by the non-uniformity of the active layer thickness on the SiI substrate.

[0003] Therefore, specifications regarding thickness and uniformity become extremely stringent: for active layers typically less than 50 nm thick, WiW and WtW inhomogeneities (WiW stands for within-wafer and WtW for wafer-to-wafer) are expected to be typically less than + / - 0.4 nm, meaning that every measurement point of the layer must be within + / - 0.4 nm of its average thickness (excluding the peripheral area of ​​a few millimeters). Such uniformity is difficult to achieve because the series of steps used to fabricate the silicon-on-insulator substrate contributes cumulatively to the inhomogeneities of the surface layer.

[0004] One known solution for correcting thickness inhomogeneities in the active layer is to use, for example, the plasma etching method described in document US20140234992, or the cluster ion beam etching method described, in particular, in document WO2013003745, to locally etch the layer. However, this type of solution has a drawback: etching the surface of the active layer creates an amorphous silicon surface layer, which is prone to causing electrical problems and therefore must be removed. The removal of the amorphous layer leads to increased surface roughness, which negatively impacts the performance of devices generated on thin layers.

[0005] Reference WO2004015759 proposes an alternative solution: localized sacrificial thermal oxidation, which locally consumes a larger or smaller thickness of the active layer to correct its thickness inhomogeneity. A drawback of this method is that localized temperature gradients are not easily introduced into the silicon layer; therefore, the resolution of inhomogeneity correction may be limited.

[0006] Document WO2013175278 also aims to improve the thickness uniformity of the surface layer of silicon substrate on insulator by implementing SC1 chemical etching solution: adjusting the amount of solution, application time and / or temperature according to the thickness of the layer to be etched.

[0007] Document EP3200219 describes a solution for thinning the active layer of a silicon-on-insulator substrate in a single-wafer etching tool, achieving a WiW inhomogeneity of approximately + / -0.5 nm.

[0008] It remains advantageous to improve the method of thinning the active layer of silicon-on-insulator substrate to ensure excellent thickness uniformity (better than + / -0.5 nm, or even better than + / -0.4 nm) for very small thicknesses and for silicon-on-insulator substrate diameters of at least 200 mm or even 300 mm.

[0009] Invention Objective

[0010] This invention aims to improve the uniformity of the surface layer of a silicon-on-insulator substrate without degrading the layer's performance (roughness and defect number), and to provide an industrial solution in terms of reliability and cost. Specifically, this invention relates to a method for thinning the surface layer of a silicon-on-insulator substrate in a single-wafer chemical processing tool, optimized to achieve a surface layer non-uniformity of less than or equal to + / - 0.4 nm, wherein the yield is at least 90%. Summary of the Invention

[0011] This invention relates to a method for thinning the surface layer of a silicon-on-insulator substrate, comprising: a) The step of measuring the thickness of the surface layer to obtain the average thickness and thickness non-uniformity of the layer. The following steps are then performed in a single-wafer chemical processing tool, during which the silicon-on-insulator substrate is rotated: b) Perform n oxidation and etching cycles on the front side of the surface layer, where n is an integer greater than or equal to 1. Each cycle involves dispensing ozonated water, followed by dispensing hydrofluoric acid. c) Performing non-uniform etching on the front side of the surface layer, which involves distributing a silicon etching solution in the central region of the surface layer at a temperature between 20°C and 80°C for a given time, referred to as the non-uniform etching time. d) Uniform etching is performed on the front side of the surface layer, which involves distributing the silicon etch solution for a given time at a temperature between 20°C and 80°C using a sweeping motion from the edge of the surface layer to the center and back. This time is called the uniform etching time.

[0012] Given the average thickness and thickness non-uniformity of the surface layer Before proceeding to steps b), c), and d), the number of cycles, the non-uniform etching time, and the uniform etching time are determined based on a model that correlates the number of cycles, the non-uniform etching time, and the uniform etching time with the average etching thickness and the average etching non-uniformity of the surface layer. The average etching non-uniformity is defined as the difference between the average etching thickness in the central region of the surface layer and the average etching thickness in the peripheral region of the surface layer.

[0013] Other advantageous and non-limiting features of the invention, individually or in any technically feasible combination: ●The average etching non-uniformity is positive; ●The central area and the surrounding area correspond to 30% and 70% of the surface layer area, respectively; ● The average etch thickness and average etch non-uniformity are each expressed in quadratic form as a function of the following variables: cycle number, non-uniform etch time, and uniform etch time. ● The model used to determine the number of cycles, non-uniform etching time, and uniform etching time is a three-objective optimization model with lexicographic optimality. ●The objective is: (i) Identify the first set of solutions (non-uniform etch time; uniform etch time) such that for any number of cycles, the average etch thickness of the surface layer can be equal to the target average etch thickness, which is defined as the difference between the target average thickness of the surface layer and the average thickness measured in step a). (ii) Restrict the first set of solutions (non-uniform etch time; uniform etch time) identified in (i) to a second set of solutions (non-uniform etch time; uniform etch time) such that for any number of cycles, the difference between the average etch non-uniformity and the target average etch non-uniformity is minimized. The target average etch non-uniformity is defined as the difference between the target thickness non-uniformity of the surface layer and the thickness non-uniformity measured in step a). (iii) Choose a solution from the second set of solutions defined in (ii) such that the number of cycles is maximized; ●Steps b), c), and d) should be performed sequentially; ● The temperature in steps c) and d) is set to 65℃; ● In step c), the silicon etching solution is dispensed only at the center, or in a sweeping motion from the center to 50% of the surface layer radius and back; ● Average etch non-uniformity depends on the etch rate gradient defined between the center and edge of the surface layer, which in turn depends on: - The flow rate of the silicon etching solution in step c), and / or -Whether there is a solution on the back side of the silicon substrate on the insulator during all or part of step c), and / or - The temperature applied in step c); ●The silicon etching solution in step c) and / or step d) is SC1. Attached Figure Description

[0014] Other features and advantages of the invention will become apparent from the following detailed description given with reference to the accompanying drawings, in which: [ Figure 1 ] Figure 1 Three examples of surface layer thickness maps are shown after a conventional thinning sequence and before the thinning method according to the invention; the gray levels in the figures are in angstroms. [ Figure 2a ][ Figure 2b ] Figure 2a and Figure 2b The non-uniformity of the surface layer of 25 silicon substrates on insulators is illustrated before (Av) and after (Ap) the thinning method according to the invention; Figure 2a This shows the deviation (in angstroms) of the measured thickness relative to the average thickness as a function of the substrate radius, and Figure 2b The diagram shows a stacked view of the surface layers of the silicon substrate on insulator before (Av) and after (Ap) steps b), c), and d) of the method according to the invention. These diagrams are derived from thickness measurements at 625 points. Detailed Implementation

[0015] The present invention relates to a method for thinning a surface layer on a silicon-on-insulator substrate, the surface layer being formed of an active layer made of high-quality monocrystalline silicon and placed on a silicon oxide layer, the silicon oxide layer itself being placed on a carrier substrate conventionally made of monocrystalline silicon.

[0016] Many methods for manufacturing such substrates are known, including Smart Cut. TM The method is based on implanting a lightweight material to create a buried, fragile plane in a donor substrate, and using direct bonding to join the donor substrate and a carrier substrate, wherein a dielectric layer is placed between the two substrates. Transferring the thin layer from the donor substrate to the carrier substrate yields a silicon-on-insulator substrate, and the remaining portion of the donor substrate can be reused for multiple subsequent transfers.

[0017] The donor substrate and the carrier substrate are preferably each in the form of a circular wafer, which has a diameter of 200 mm or 300 mm or even 450 mm and a thickness typically between 500 micrometers and 900 micrometers.

[0018] Following the transfer, for a layer thickness of, for example, approximately 250 nm, the free surface of the thin surface layer exhibits a relatively high level of roughness (approximately 4 to 8 nm RMS measured using atomic force microscopy (AFM) in a 30 × 30 μm scanning area). Surface finishing techniques must then be performed to repair this layer and thin it to a target thickness of less than or equal to 50 nm (e.g., 20 nm, 15 nm, or even 10 nm). In finishing techniques, sacrificial oxidation thermal treatment, etching, and / or surface reconstruction (smoothing) can be performed, particularly in a neutral or reducing atmosphere. These processes are generally performed at high temperatures; for example, oxidation can be performed between approximately 750 °C and 1100 °C, and smoothing can be performed between 950 °C and 1250 °C.

[0019] An advantageous thinning and finishing sequence particularly includes: - Stabilization involves surface oxidation of the surface layer; the sacrificial oxide layer is then removed via chemical etching. - Perform thermal smoothing on the surface under a neutral or reducing atmosphere. - Thinning is achieved by utilizing the new sacrificial oxidation of the surface layer.

[0020] These steps are performed in a batch furnace, and despite continuous improvements in the method, the various contributions to inhomogeneity accumulate, tending to produce non-standard active layers.

[0021] Therefore, the present invention provides a thinning method that at least partially compensates for thickness inhomogeneities in the surface layer of a silicon substrate on an insulator, the initial inhomogeneity typically being less than 15 nm, or even less than 10 nm. The overall thinning resulting from this method is low (generally less than 5 nm, or even less than 3 nm): therefore, the provided method is advantageously applicable after a thinning and finishing sequence such as those described above.

[0022] This method involves measuring the thickness of the surface layer to obtain the average thickness E of the layer. 平均 The first step a) to address thickness non-uniformity U is to measure the thickness at multiple points i distributed throughout the layer surface. The average thickness E is then determined. 平均 The thickness non-uniformity U corresponds to the average thickness measured at multiple points i; the thickness non-uniformity U represents the thickness distribution of the layer at multiple points i. Thickness non-uniformity is expressed as the thickness (at point i) relative to the average thickness E. 平均 The deviation value (e.g., + / - 0.4 nm) indicates that the surface layer thickness at multiple points i falls within the range [E]. 平均 -0.4 nm; E 平均 [+0.4 nm] within.

[0023] The number of measurement points i can be approximately 20, 100, 500, or even 1000, and the measurements are preferably performed at a distance of no more than 3 mm from the peripheral edge of the silicon-on-insulator substrate. It should be remembered that the surface layer generally extends at least 0.5 mm from this edge. Points i may be distributed in radial geometry, star-shaped geometry, or spiral geometry starting from the center of the layer, or in any other geometry that can effectively represent the thickness non-uniformity of the measured layer.

[0024] Without limitation, the thickness measurement in step a) can be performed on the silicon surface layer with or without a sacrificial oxide layer using known methods based on elliptic polarization or reflectance measurement techniques.

[0025] Figure 1 Three examples of surface layer thickness maps shown after a series of standard thinning and finishing processes using various tools. As can be seen, this generally results in radial inhomogeneities of varying magnitudes (grayscale in angstroms) and excessive center thickness.

[0026] Therefore, the steps of the method according to the invention advantageously enable compensation of layer profiles with excessive center thickness. These steps are performed in a single-wafer chemical processing tool. Such tools are well known and, in particular, include arms for holding a silicon-on-insulator substrate, and one or more nozzles for dispensing various types of solutions (chemical solutions or water) onto the front side of the substrate (the side dispensing to the surface layer) and / or onto its back side. During solution dispensing, the substrate is rotated to distribute the solution over the entire surface in question (typically at 10 to 1000 revolutions per minute).

[0027] Steps b), c) and d) described herein may be performed sequentially (preferably) or not sequentially, but in all cases, they are performed one after another without being removed from the tool, since the method according to the invention does not require intermediate thickness measurements, which is particularly advantageous.

[0028] It should be noted that if the silicon-on-insulator substrate contains a sacrificial oxide layer on the silicon surface layer, this oxide layer should be removed before steps b), c), and d) of the method.

[0029] Step b) involves performing n oxidation and etching cycles on the (free) front side of the surface layer, where n is an integer greater than or equal to 1. Each cycle involves dispensing ozonated water, followed by dispensing hydrofluoric acid (HF), to sequentially oxidize the surface layer and etch the resulting oxide layer in each cycle. Each cycle thins the surface layer by approximately 0.4 to 0.5 nm. While the number of cycles that can be performed is not limited, it is preferable to perform 1 to 10 cycles, or even advantageously 1 to 5 cycles, to remain industrially feasible.

[0030] In substrate cleaning, such cycles are known as the acronym "SCROD" (single-wafer spin cleaning with repetitive use of ozonated water and dilute HF).

[0031] Typically, the ozone concentration in the ozonated water is between 20 ppm and 40 ppm, and the mass concentration of the HF solution is between 0.5% and 4%; the dispensing flow rate is 2 liters per minute. Step b) is preferably performed at room temperature. Dispensing can be carried out via a fixed or moving nozzle (i.e., a nozzle that performs a sweeping motion).

[0032] The step of rinsing the back side of the silicon substrate on the insulator can be performed in parallel with the aforementioned cycle performed on the front side.

[0033] Step c) corresponds to etching on the front side of the surface layer, which involves distributing a silicon etching solution (e.g., SC1 (representing standard clean standard 1), TMAH, HF / H2O2, HF / O3, HF / HNO3+CH3COOH, KOH, etc.) in the central region of the surface layer. The fact that the solution is distributed only in the central region implies uneven etching of the surface layer, resulting in more etching in the central region than in the peripheral region.

[0034] At the end of steps b), c), and d) of this method, the average etch non-uniformity is obtained. 蚀刻 This non-uniformity is defined as the average etch thickness e in the central region of the surface layer. C 蚀刻 The average etch thickness e in the peripheral region of the surface layer P 蚀刻 The difference. This etching non-uniformity u 蚀刻 This is positive to compensate for the excessive center thickness of the surface layer. It is mainly related to step c), but other steps may sometimes contribute a small amount.

[0035] In step c), the application can be performed only at the center, or it can be performed using a sweeping motion from the center to 30%, 50%, 80%, or even 90% of the surface layer radius and then back. This allows for adjustment of the profile of the etching non-uniformity between the central and peripheral areas.

[0036] Average etch non-uniformity u 蚀刻The etching rate gradient depends on the etch rate gradient defined between the center and edge of the surface layer, or between the central region and the peripheral region. This gradient depends on the flow rate of the etch solution in step c). This gradient can also be adjusted by distributing the solution on the back side of the silicon-on-insulator substrate during all or part of step c); this distribution can particularly locally alter the temperature of the substrate. Finally, the etch rate gradient depends on the temperature applied in step c).

[0037] Step c) is performed at a temperature between 20°C and 80°C, for example at 65°C for a given duration, referred to as the non-uniform etching time t. nu Uneven etching time t nu It can vary from a few seconds to hundreds of seconds; in practice, it is less than 300 s, 250 s or even 200 s to meet the requirements of industrial efficiency.

[0038] Step d) corresponds to etching on the front side of the surface layer, which involves distributing the silicon etching solution (e.g., SCl, TMAH, HF / H2O2, HF / O3, HF / HNO3+CH3COOH, KOH, etc.) with a sweeping motion from the edge of the surface layer to the center and back; this sweeping ensures that the etching has a certain uniformity across the entire surface of the layer.

[0039] Step d) is performed at a temperature between 20°C and 80°C, for example, at 65°C for a given duration, referred to as the uniform etching time t. u Uniform etching time t u It can vary from a few seconds to hundreds of seconds; in practice, it is less than 400 s or even 300 s, also for industrial efficiency reasons.

[0040] When used as a silicon etching solution in steps c) and d), the ratio (volume ratio) of ammonium hydroxide (NH4OH), hydrogen peroxide (H2O2), and deionized water in SC1 can be selected from, for example, 1 / 1 / 40 to 1 / 1 / 10. The dispensing flow rate is typically between 1 and 2 liters per minute.

[0041] In the method of the present invention, before performing steps b), c), and d), based on the cycle number n and the uneven etching time t nu and uniform etching time t u With the average etch thickness e of the surface layer 蚀刻 and average etching non-uniformity u 蚀刻 The associated empirical model determines the cycle number n and the non-uniform etching time t. nu and uniform etching time t u .

[0042] For example, the central region (uniformly distributed around the center of the substrate) and the peripheral region (concentric, surrounding the central region) can be defined to correspond to 30% and 70% of the surface layer area, respectively.

[0043] Therefore, the average thickness E of the surface layer after step a) 平均 and thickness non-uniformity U, and given the target average thickness E of the surface layer. 目标 and target thickness non-uniformity (U 目标 In the case of ), this model can be used to determine the number of cycles n and the non-uniform etching time t. nu and uniform etching time t u The optimal value.

[0044] Advantageously, the average etch thickness e is expressed in quadratic form in this model. 蚀刻 and average etching non-uniformity u 蚀刻 , as a function of the following variables: cycle number n, non-uniform etching time t nu and uniform etching time t u As shown below: [Equ1]

[0045] [Equ2]

[0046] Compared to the linear form, the quadratic form allows for the determination of the three variables with good accuracy.

[0047] The coefficients α1 to α9 and β1 to β9 were determined empirically based on experimental design and experimental measurements.

[0048] Three-objective optimization models with lexicographical optimality are particularly well-suited for determining the variables sequentially. The objective can be defined as follows: (i) Identifying uneven etching time t nu and uniform etching time t u The first solution S n + (t nu , t u This allows for achieving an average etch thickness e of the surface layer for any number of cycles n. 蚀刻 With the target average etch thickness e 目标 The average thickness e to be etched between them is equal. 目标 The target average thickness E of the surface layer is defined as follows: 目标 The average thickness E measured in step a) 平均 difference; (ii) The set of solutions S identified in (i) n+ (t nu , t u The time limit is t for non-uniform etching. nu and uniform etching time t u The second set of solutions S(t) nu , t u This allows for an average etch non-uniformity u for any number of cycles n. 蚀刻 Target average etch non-uniformity u 目标 Minimize the difference, target average etch non-uniformity u 目标 The target thickness non-uniformity U of the surface layer is defined as 目标 The difference between the thickness non-uniformity U measured in step a) and the difference between the thickness non-uniformity U and the thickness non-uniformity U measured in step a). (iii) The second set of solutions S(t) defined in (ii) nu , t u Choose a recipe that maximizes the number of cycles (n).

[0049] The first objective (i) is to solve for the average etch thickness e of the surface layer. 蚀刻 With the target average etch thickness e 目标 The quadratic equation Equ3, derived from the equality between them, represents the average thickness e of the target to be etched. 目标 The target average thickness E of the surface layer is defined as follows: 目标 The average thickness E measured in step a) 平均 The difference. Equation Equ3 with uniform etching time t u (The time applied in step d) is represented as a variable: [Equ3]

[0050] Due to uniform etching time t u It must be positive, therefore the solution to this equation can be expressed as the non-uniform etching time t. nu The functions for (step c) and the loop number (step b) are as follows: [Equ4] in [Equ5]

[0051] Therefore, at this stage, for each integer n (cycle number), the first group S can be defined. n Possible solution t u (uniform etching time) and t nu (Uneven etching time): [Equ6]

[0052] t u and t nu The first set of possible solutions is further filtered to include only positive values ​​and reflect the technical feasibility of the tool, i.e., the accuracy within 1 second. Thus, at the end of the first step, the following set of solutions S is given. n + (First objective): [Equ7]

[0053] The second objective is to reduce the average etch non-uniformity u represented in Equ2. 蚀刻 Target average etch non-uniformity u 目标 Minimize the difference, target average etch non-uniformity u 目标 The target thickness non-uniformity U of the surface layer is defined as 目标 The difference between the thickness non-uniformity U measured in step a) and the thickness non-uniformity U.

[0054] A set of S retained from the first step n + The solution begins by retaining only those solutions that allow for etch non-uniformity u. 蚀刻 Formula r that minimizes error n : [Equ8]

[0055] Therefore, at the end of the second step, the second solution S (the second objective) is expressed as: [Equ9]

[0056] The third objective is to minimize the cost of the method, which is equivalent to maximizing the number of loops n in step b), since step b) is most efficient in terms of thinning.

[0057] Therefore, the solution with the highest cycle number in the second set of solutions S is selected.

[0058] Figure 2a The trend curve Av in the figure illustrates the initial non-uniformity of the surface layers of the silicon substrate on 25 insulators (i.e., before applying the method according to the invention). Specifically, the deviation of the thickness relative to the average thickness at any measured point i on the 25 surface layers is edited into... Figure 2a In the middle: Note the excessive center thickness (radius 0) that is 0.5 nm greater than the edge thickness on average.

[0059] Figure 2aThe trend curve Ap illustrates the final inhomogeneity of the same 25 surface layers after implementing steps b), c), and d) of the method according to the invention. The deviation of the thickness from the average thickness at any measured point i on the 25 surface layers is edited, and it can be seen that the final inhomogeneity is better, with an average deviation of less than 0.2 nm between the center and the edges. The vast majority of the measured points satisfy the target thickness inhomogeneity U with respect to + / - 0.4 nm. 目标 Requirements.

[0060] In this example, the average thickness is evaluated based on 625 measurement points. (See the stacked diagram of surface layers on a silicon substrate with 25 insulators.) Figure 2b As can be seen in the figure, after implementing the thinning method according to the invention, the thickness non-uniformity profile is modified, and excessive center thickness can be eliminated and the overall uniformity of the surface layer is improved.

[0061] Undoubtedly, the present invention is not limited to the described embodiments and examples, and variations of the embodiments may be applied to the present invention without departing from the scope of the invention as defined in the claims.

Claims

1. A method of thinning a surface layer of a silicon-on-insulator substrate, comprising: a) a step of measuring the thickness of the surface layer to obtain the average thickness (E 平均 ) of said layer and the thickness unevenness (U), Subsequently, the following steps are performed in a single-wafer chemical processing tool, and during this time the silicon-on-insulator substrate is rotated: b) performing n oxidation and etching cycles on the front side of the surface layer, where n is an integer greater than or equal to 1, each cycle involving dispensing ozonated water, followed by dispensing hydrofluoric acid, c) performing a non-uniform etching on the front side of the surface layer, which involves dispensing a silicon etching solution in the central region of the surface layer at a temperature between 20°C and 80°C for a given time, referred to as the non-uniform etching time (t nu ), d) performing a uniform etch on the front side of the surface layer, which involves dispensing a silicon etch solution with a sweeping motion from the edge to the center and back of the surface layer at a temperature between 20°C and 80°C for a given time, which is called the uniform etch time (t u ), wherein, given the average thickness (E 平均 ) and the thickness non-uniformity (U) of the surface layer, prior to carrying out steps b), c) and d), the number of cycles (n), the non-uniform etching time (t nu ) and the uniform etching time (t u ) are determined based on a model relating the number of cycles (n), the non-uniform etching time (t 蚀刻 ) and the uniform etching time (t 蚀刻 ) to the average etching thickness (e nu ) and the average etching non-uniformity (u u ) of the surface layer, the average etching non-uniformity (u 蚀刻 ) being defined as the difference between the average etching thickness in the central region of the surface layer and the average etching thickness in the peripheral region of the surface layer.

2. The thinning method of the preceding claim, wherein the average etch non-uniformity (u 蚀刻 ) is positive.

3. The thinning method of any one of the preceding claims, wherein the central zone and the peripheral zone respectively correspond to 30% and 70% of the area of the surface layer.

4. The thinning method of any of the preceding claims, wherein the average etch thickness (e 蚀刻 ) and the average etch non-uniformity (u 蚀刻 ) are each expressed as a function of the cycle number (n), the non-uniform etch time (t nu ) and the uniform etch time (t u ) in a quadratic form.

5. The thinning method according to the preceding claim, wherein the model for determining the number of cycles (n), the non-uniform etching time (t nu ) and the uniform etching time (t u ) is a three-objective optimization model with lexicographic optimality.

6. The thinning method of the preceding claim, wherein the objective is: (i) identifying a first set of solutions (S nu ) of the non-uniform etching time (t u ) and the uniform etching time (t n + (t nu , t u )) such that for any cycle number (n) an equality between the average etching thickness (e 蚀刻 ) of the surface layer and a target average to be etched thickness (e 目标 ) is achieved, the target average to be etched thickness (e 目标 ) being defined as the difference between a target average thickness (E 目标 ) of the surface layer and the average thickness (E 平均 ) measured in step a); (ii) The first set of solutions (S) identified in (i) n + (t nu , t u The time limit is limited to the non-uniform etching time (t). nu ) and the uniform etching time (t) u The second set of solutions (S(t)) nu , t u This allows for the average etch non-uniformity (u) to be achieved for any number of cycles (n). 蚀刻 ) and target average etching non-uniformity (u 目标 Minimize the difference between the target average etch non-uniformity (u) and the target average etch non-uniformity (u) 目标 The target thickness non-uniformity (U) of the surface layer is defined as this. 目标 The difference between the thickness non-uniformity (U) measured in step a) and the thickness non-uniformity (U) measured in step a). (iii) selecting a recipe from the second set of solutions (S(t nu , t u )) defined in (ii) such that the number of cycles (n) is maximized.

7. The thinning method of any one of the preceding claims, wherein steps b), c) and d) are performed sequentially.

8. The thinning method of any one of the preceding claims, wherein the temperature in steps c) and d) is set to 65°C.

9. The thinning method of any one of the preceding claims, wherein in step c) the silicon etching solution is dispensed only at the center, or in a sweeping motion from the center to 50% of the radius of the surface layer and back.

10. The thinning method of any one of the preceding claims, wherein the average etch non-uniformity (u 蚀刻 ) depends on an etch rate gradient defined between a center and an edge of the surface layer, the gradient depending on: - the flow rate of the silicon etching solution in step c), and / or - whether a solution is dispensed on the back side of the silicon-on-insulator substrate during all or part of step c), and / or - the temperature applied in step c).

11. The thinning method of any one of the preceding claims, wherein the silicon etching solution in step c) and / or in step d) is SC1.

Images