Chemical polishing process for silicon wafers

By measuring diffuse reflectivity to control immersion time in anisotropic etching, the chemical polishing process addresses the issues of cost and thickness loss in silicon wafer polishing, ensuring conformal thin film deposition and enhancing photovoltaic cell production efficiency.

FR3168535A1Pending Publication Date: 2026-05-22COMMISSARIAT A LENERGIE ATOMIQUE ET AUX ENERGIES ALTERNATIVES
View PDF 3 Cites 0 Cited by

Patent Information

Authority / Receiving Office
FR · FR
Patent Type
Applications
Current Assignee / Owner
COMMISSARIAT A LENERGIE ATOMIQUE ET AUX ENERGIES ALTERNATIVES
Filing Date
2024-11-19
Publication Date
2026-05-22

AI Technical Summary

Technical Problem

Existing chemical polishing processes for silicon wafers used in photovoltaic cell production are costly and result in excessive reduction of wafer thickness and surface roughness, leading to potential breakage and non-conformal thin film deposition.

Method used

A chemical polishing process that measures the diffuse reflectivity of the cutting face to determine an immersion time for anisotropic etching, ensuring a conformal deposition of a thin layer by controlling the roughness of the cutting face.

Benefits of technology

The process effectively reduces surface roughness while maintaining wafer thickness, enabling conformal deposition of thin films and improving the efficiency and cost-effectiveness of photovoltaic cell production.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 00000000_0000_ABST
    Figure 00000000_0000_ABST
Patent Text Reader

Abstract

The invention relates to a chemical polishing process for obtaining a batch of smoothed wafers. It comprises supplying several silicon wafers, each with a cutting face; measuring the diffuse reflectivity of the cutting face of each silicon wafer; determining an immersion time per wafer for a set of wafers from among those supplied, each immersion time being a function of the diffuse reflectivity of the corresponding cutting face; and anisotropic wet etching of each wafer in the set of wafers along silicon crystal planes by immersion in an etching solution for the corresponding determined immersion time, to obtain the batch of smoothed wafers. Each immersion time is determined to ensure consistent deposition of a thin layer on the cutting face of each wafer in the batch of smoothed wafers. Figure 1
Need to check novelty before this filing date? Find Prior Art

Description

Title of the invention: Chemical polishing process for silicon wafers technical field

[0001] The field of the invention is that of processes for smoothing or polishing silicon wafers, particularly for the manufacture of photovoltaic cells. PRIOR TECHNOLOGY

[0002] Silicon wafers, particularly for the manufacture of photovoltaic cells, are obtained by cutting a silicon ingot. The most common cutting techniques involve sawing the silicon ingot with wires. These techniques allow for the simultaneous cutting of a large number of wafers. Among these, a particularly advantageous method consists of using wires coated with synthetic diamond, called diamond wires. This is a fast method that minimizes the production of silicon waste.

[0003] To minimize the economic and environmental costs of photovoltaic cell production, it is necessary to reduce the amount of silicon waste produced during wafer production and to increase the number of wafers produced per silicon ingot. This latter objective can be achieved by reducing the thickness of the wafers produced after sawing, which can make them more brittle. For example, silicon wafer thicknesses for photovoltaic applications are generally between 150 µm and 200 µm.

[0004] The cutting step generally generates a work-hardened area extending to a shallow depth from each face of the plate created by the cutting, called the cutting face. A work-hardened area is a rough, plastically deformed area of ​​the plate. It contains microcracks and / or crystalline defects, such as dislocations. It may be contaminated by metal atoms. It is therefore necessary to remove it for the manufacture of certain electronic components in and / or on the plate, such as certain photovoltaic cells.

[0005] It is known to remove a work-hardened area by chemical polishing. However, this process is often costly, particularly for the manufacture of photovoltaic cells. Consequently, anisotropic chemical etching of the work-hardened area using an etching solution is often preferred. Such a process is commonly called chemical polishing in the photovoltaic cell industry. The etching solution is generally KOH.

[0006] Removing the work-hardened area results in a reduction of the thickness of the silicon wafer, which is already significantly reduced during the cutting step for the reasons explained above. It is therefore necessary to remove the work-hardened area in such a way as to eliminate it, without reducing the thickness of the silicon wafer too much below a limit for which it might break during subsequent process steps, such as a wafer handling step, or thermal annealing.

[0007] When the work-hardened area is removed by chemical polishing, patterns corresponding to square-based pyramids form on the cutting face. Truncated pyramids, called mesas, are predominant. The shape and density of the patterns depend on the conditions of the chemical etching, such as the KOH concentration of the solution, the temperature of the solution, or the immersion time of the plate in the solution. The collection of patterns, generally larger than 1 micron, generates a roughness on the cutting face that can cause a thin film deposited by liquid on it to break, particularly when one or more patterns have heights greater than the thickness of the thin film. Description of the invention

[0008] The invention aims to remedy at least in part the drawbacks of the prior art, and more particularly to propose a chemical polishing process to obtain a batch of silicon wafers guaranteeing a liquid deposition of a conforming thin layer on a cutting face of each wafer in the batch.

[0009] To this end, the object of the invention is a chemical polishing process for obtaining a batch of smoothed wafers. The process comprises the following steps: supplying several silicon wafers, each comprising a cutting face; measuring the diffuse reflectivity of the cutting face of each silicon wafer; determining an immersion time per wafer for a set of wafers from among the supplied wafers, each immersion time being a function of the diffuse reflectivity of the cutting face of the corresponding wafer; anisotropic wet etching of each wafer in the set of wafers along silicon crystal planes, by immersion in an etching solution for the corresponding determined immersion time, to obtain the batch of smoothed wafers.

[0010] Each immersion time is determined to ensure a conforming deposit of a thin layer on the cutting face of each plate in the batch of smoothed plates.

[0011] Some preferred but not limiting aspects of this chemical polishing process are the following.

[0012] Anisotropic etching can be etching with a solution containing a mass concentration of KOH greater than or equal to 20%. The temperature of the solution can be greater than or equal to 70 °C.

[0013] Each supplied silicon wafer may include a work-hardened area extending in depth from the cutting face. Anisotropic etching can remove the work-hardened area from all the wafers in the assembly.

[0014] The anisotropic etching step can lead to a thinning of each plate in the assembly. The thinning can be less than or equal to 60 µm.

[0015] The step of determining an immersion time may include a substep of grouping the set of plates into subsets of plates with homogeneous diffuse reflectivity. For each subset, the immersion times of all the plates in the subset may be determined so as to be equal to a common immersion time for the subset. All the plates in each subset may be immersed simultaneously in a bath containing the etching solution for a time equal to the common immersion time during the wet etching step.

[0016] The chemical smoothing process may further include selecting the plate set from among the supplied plates having a diffuse reflectivity less than or equal to a predetermined threshold. The determined threshold may guarantee an immersion time less than or equal to a predetermined maximum immersion time.

[0017] The invention also relates to a method for manufacturing an assembly of silicon-based photovoltaic cells, comprising obtaining a batch of wafers smoothed by a chemical polishing process according to any one of the preceding characteristics. The thin film may be a functional layer of one or more photovoltaic cells in the assembly. The manufacturing process may include conformal deposition by liquid or evaporation onto the cutting face of each wafer in the batch of the functional layer.

[0018] The cutting face can be nanostructured between the anisotropic etching step and the deposition step.

[0019] The first cells in the photovoltaic cell assembly can be tandem silicon-perovskite cells. The functional layer can be a perovskite layer or a contact layer for manufacturing the first cells.

[0020] Second cells in the photovoltaic cell assembly may be single-junction silicon-based photovoltaic cells. The manufacturing process may further include manufacturing the second cells in and / or on the wafers of a subsidiary batch selected from among the supplied wafers not selected during the selection step.

[0021] The functional layer may have a thickness less than or equal to 1 pm. Brief description of the drawings

[0022] Other aspects, objects, advantages and features of the invention will become more apparent upon reading the following detailed description of preferred embodiments thereof, given by way of non-limiting example, and made with reference to the accompanying drawings in which:

[0023] [Fig.1] illustrates a method for measuring the diffuse reflectivity of a face that can be used for a selection step of an example of a process according to the invention;

[0024] [Fig.2A] shows measurement results of the diffuse reflectivity of cutting faces as a function of an etching time in a first etching solution;

[0025] [Fig.2B] shows measurement results of the diffuse reflectivity of cutting faces as a function of an etching time in a second etching solution;

[0026] [Fig.3] shows the measurement results of figures 2A and 2B, as a function of average engraving speeds observed;

[0027] [Fig.4] is an optical microscope image of a cut face treated by an example of a chemical polishing process according to the invention.

[0028] DETAILED DESCRIPTION OF SPECIFIC EMBODIMENTS

[0029] In the figures and throughout the description, the same reference numerals represent identical or similar elements. Furthermore, the various elements are not drawn to scale in order to enhance the clarity of the figures. Moreover, the different embodiments and variants are not mutually exclusive and may be combined. Unless otherwise indicated, the terms "approximately," "about," and "in the order of" mean within 10%, and preferably within 5%. Furthermore, the terms "between ... and ..." and equivalents mean that the limits are inclusive, unless otherwise stated.

[0030] The invention relates to a chemical polishing method for silicon wafers cut from a silicon ingot, without involving chemical polishing. Each silicon wafer corresponds to a slice of the silicon ingot, separated from it by a cutting step, for example using a diamond wire. It has a cut face at the level of a surface that held it securely to the silicon ingot.

[0031] The chemical polishing process includes an anisotropic wet etching step of the cutting face, for example in a potassium hydroxide (KOH) solution. The inventors found that a measurement of the diffuse reflectivity of the cutting face after cutting, i.e., on silicon wafers obtained after cutting (wafers commonly called "as eut," using an anglicism), is predictive of the roughness of the cutting face after the wet etching step. Therefore, the chemical polishing process includes a step of determining an immersion time per wafer from a measurement of the diffuse reflectivity of the wafer's cutting face. Each immersion time can be determined to ensure that the roughness of the cutting face is less than or equal to a specification. The specification may be a maximum valley-to-ridge roughness that allows for conformal liquid deposition of a thin layer on the cutting face after the engraving step.

[0032] The term "liquid deposition" encompasses various methods for depositing a layer of a material in liquid form. The liquid may, for example, comprise the material and a solvent. By way of example, deposition may be carried out by centrifugation of the liquid (or "spin-coating"), by immersion in a bath containing the liquid, or by spray coating. In a centrifugal deposition method, the liquid may be dispensed by a nozzle and / or a spray.

[0033] In the description and technical field of photovoltaic cells, a conformal deposition of a layer on a face is such that the layer extends continuously over the face, without any break in the layer. In this case, the face is entirely covered by the layer. The layer resulting from such a deposition is a conformal layer.

[0034] Surface roughness is a quantitative measure of the variations in surface height relative to a reference surface. In the context of the invention, the reference surface is preferably a flat surface. The quantitative measure can be equal to the root mean square of the heights over a spatial frequency range, or to the average of the absolute values ​​of the heights over the spatial frequency range. It can also be equal to the maximum height in absolute value or to a maximum valley-to-peak difference in height over a given distance. It can be measured using an atomic force microscope or a profilometer.

[0035] Figure 1 illustrates a method for measuring the diffuse reflectivity of a surface that can be used for a step in determining the immersion time of an example of a chemical polishing process according to the invention. The measurement method can be implemented in a spectrophotometer.

[0036] An incident beam 10 illuminates a plate 100 on one of its upper faces. The incident beam 10 makes a non-zero angle of incidence with a normal to the upper face. The upper face can be arbitrary. Here, it is a cutting face 101 of a silicon plate 100, for example, a plate for photovoltaic applications. The silicon plate can, for example, have a crystalline orientation (100). The plate 100 is preferably illuminated by the incident beam 10 at a grazing angle.

[0037] The incident beam 10 is partially reflected by specular reflection into a reflected beam 11 propagating along a specular direction in space making an angle θ₀S with the normal to the cutting face 101, equal to 0 in absolute value. The reflectivity specular Rs is equal to the ratio of the intensity of the reflected beam 11 to the intensity of the incident beam 10.

[0038] The incident beam 10 is also partially scattered into a diffuse light 12 by a roughness of the cutting face 101, in spatial directions other than the specular direction. The diffuse reflectivity RD is equal to the ratio of the intensity of the diffuse light 12 to the intensity of the incident beam 10.

[0039] The respective intensities of the incident beam 10, the reflected beam 11, and the scattered light 12 can be measured by one or more photoelectric detectors. Diffuse reflectivity is a strictly increasing function of the roughness of the cutting face 101. It is possible, for example, to empirically establish a correlation law or a nomogram giving the roughness of the cutting face 101 as a function of its diffuse reflectivity. This law or nomogram is likely to vary from one spectrophotometer to another or from one experimental setup to another. They also depend on other parameters, including the wavelength or range of wavelengths of the incident beam 10, the angle of incidence θ, or a dimension of the incident beam 10 at the cutting face 101.

[0040] An example of a chemical polishing process according to the invention will now be described.

[0041] In a first step, silicon wafers 100 are supplied, for example, for applications in the field of photovoltaic cells. The wafers 100 may, for example, all have the same crystal orientation. In this example, they are all silicon wafers oriented along a crystal plane (100). The wafers may, for example, be supplied by a wafer manufacturer. They may be n-type or p-type doped.

[0042] The silicon wafers 100 correspond to slices or portions of slices of one or more silicon ingots, for example, obtained by a Czochralski (CZ) growth process. They can be of any shape. For example, they may all be in the shape of a disk, a parallelepiped, or a right prism. They can have standard thicknesses, for example, between 160 µm and 180 µm. They were obtained by cutting the silicon ingot(s), for example, using one or more diamond wires. Each wafer therefore comprises a cutting face 101 at a cutting plane of the ingot from which it originates. When the wafer 100 has been separated from the ingot by a diamond wire, the cutting face 101 is a surface of the wafer 100 that was in contact with the diamond wire during the cutting process.

[0043] Each silicon plate 100 generally has a work-hardened area extending deep from the cutting face 101, particularly when the plate 100 has been cut using a wire in the presence of an abrasive, or a diamond wire. As As mentioned in the "Prior Art" section, a work-hardened zone typically contains microcracks and / or crystalline defects. When the wafer has been cut with a diamond wire, it generally exhibits striations of varying depths. It may be contaminated by metal atoms. It can extend to a depth of between 5 µm and 12 µm, generally around 8 µm. A work-hardened zone can cause a number of malfunctions in electrical or electronic components fabricated in and / or on the wafers, such as an increased carrier recombination rate, difficulties in collecting electrical charges, altered doping profiles, or premature aging of the components. The work-hardened zone generally has a roughness that can compromise the application of a conformal liquid coating.

[0044] In a second step, the diffuse reflectivity of the cutting face 101 of each silicon wafer 100 is measured by the measurement method illustrated in [Fig. 1]. The diffuse reflectivity can be equal to the average of the diffuse reflectivities obtained over a range of wavelengths and / or at different positions of the cutting face 101. Preferably, the cutting face has not undergone any operations affecting the amplitude or nature of its roughness between the cutting and the measurement step. One or more cleaning steps may be included between the cutting and the measurement step.

[0045] In this example, the average diffuse reflectivity is measured over a range of wavelengths, for example between 300 nm and 1200 nm, or between 350 nm and 1000 nm, here measured at the center of the cutting face 101. The incident beam has, for example, a size of 15 mm by 7 mm, at the cutting face 101. The angle of incidence 0; can be equal to 82 degrees.

[0046] In a third step, an immersion time is determined for each plate 100 based on the corresponding diffuse reflectivity measured in the second step. The immersion time corresponds to an anisotropic wet etching time implemented in a fourth step of the process.

[0047] Figures 2A and 2B illustrate experimental results that can be used to establish the immersion time.

[0048] In [Fig. 2A], the average diffuse reflectivity (y-axis, in percent) measured after different immersion times in a potassium hydroxide (KOH) solution (x-axis, in minutes) for different plates is plotted on a graph (each symbol corresponds to a different plate supplier, each point corresponds to a different plate). The mass concentration of KOH in the solution, here aqueous, is 35%. The immersion temperature is 80 °C. The average diffuse reflectivity is equal to the average of the diffuse reflectivities measured on the cut face of each plate, for wavelengths of the incident beam 10 between 350 nm and 1 pm. Each point obtained for A specific supplier and a zero immersion time correspond to the average diffuse reflectivity measured on the cutting face 101 of each "as eut" plate from that supplier, used in this experiment. It was notably observed during this experiment that the plate-to-plate variability of the average diffuse reflectivity of the "as eut" plates from a single supplier is much lower than the variability of the diffuse reflectivity between two "as eut" plates from two different suppliers.

[0049] The lines represent the linear regressions according to the least squares method obtained from the experimental data. The linear regressions of the plates symbolized by squares, circles, and triangles are respectively represented by a solid line, a dashed line, and a dashed line.For clarity, the linear regression of the plate symbolized by stars has not been shown because it is very close to that of the plate symbolized by triangles.

[0050] It is observed that the average diffuse reflectivity decreases as the immersion time increases, such that immersion in the KOH solution has the effect of smoothing the cutting face. The decrease in average diffuse reflectivity, and therefore in roughness, is substantially linear for all plate suppliers. The decrease in average diffuse reflectivity as a function of immersion time is substantially the same for all plate suppliers and, consequently, differences in roughness between plates are substantially preserved, regardless of the immersion time. Thus, a measurement of the diffuse reflectivity of a cutting face before immersion (zero abscissa) is predictive of the diffuse reflectivity of this cutting face after immersion, and therefore of the roughness of the cutting face, regardless of the immersion time.

[0051] In [Fig. 2B], experimental results similar to those in [Fig. 2A] are plotted on a graph. The mass concentration of KOH in the solution is 30%. The immersion temperature is also 80 °C. The y-axis represents the average diffuse reflectivity in %, equal to the average of the diffuse reflectivities measured for wavelengths between 350 nm and 1 pm. The x-axis represents the immersion time in minutes. Measurements of plates from two different plate suppliers are indicated by squares and circles, respectively.

[0052] Similar to the experiment in [Fig.2A], it was found here that the plate-to-plate variability of the average diffuse reflectivity of the "as eut" plates from one supplier is much lower than the variability of the diffuse reflectivity between two "as eut" plates from the two plate suppliers.

[0053] Just as in [Fig. 2A], it is observed that the decrease in average diffuse reflectivity as a function of immersion time is substantially the same for the Two plate suppliers, up to an immersion time of 15 minutes. It is also observed that below a diffuse reflectivity of approximately 5%, the diffuse reflectivity decreases less rapidly as the immersion time increases. This has the effect of reducing the difference between the diffuse reflectivities of the two plate suppliers at 20 minutes of immersion, while maintaining the same sign. To reduce production costs, it is preferable to decrease the immersion time in the solution.

[0054] By way of comparison, the average diffuse reflectivity of a cut face polished by chemical mechano-polishing (CMP) is typically equal to 1% in the wavelength range between 300 nm and 1.2 pm.

[0055] During the experiments in Figures 2A and 2B, each of the plates was weighed after each immersion time and before immersing them in the solutions. An average etching kinetic was then calculated for each immersion time by dividing the difference in weights before and after immersion by the density of silicon multiplied by the surface area(s) exposed to the solution. The results of this calculation were plotted on the graph in [Fig. 3]. The x-axis shows the value of the average etching kinetic in pm per minute. The y-axis shows the diffuse reflectivities of Figures 2A and 2B, expressed as a percentage. The squares and circles correspond, respectively, to the measurements of the plates in [Fig. 2A] and [Fig. 2B].

[0056] The numerical values ​​corresponding to the experimental points in Figures 2A and 3 are given in Table 1 below. Duration (min) Supplier Engraving speed (pm / min) Rd (%) 5 Circle 1.30 12.08 Triangle 1.34 14.22 Square 1.30 10.04 Star 1.35 14.37 8 Circle 1.28 9.11 Triangle 1.30 10.80 Square 1.21 7.09 Star 1.29 11.09 10 Circle 1.21 7.99 Triangle 1.24 9.21 Square 1.20 6.17 Star 1.26 9.23 Table 1

[0057] The numerical values ​​corresponding to the experimental points in Figures 2B and 3 are given in Table 2 below. Duration (min) Supplier Engraving kinetics (pm / min) Rd (%) 10 Circle 1.29 6.54 Square 1.25 3.79 15 Circle 1.26 4.14 Square 1.23 2.97 20 Circle 1.20 3.2 Square 1.19 2.76 Table 2

[0058] For the implementation of the third step, the immersion time can, for example, be determined from the diffuse reflectivity before immersion, by applying an empirical law that gives the variation of diffuse reflectivity as a function of the immersion time, so as to guarantee that the roughness after immersion of the cutting face is less than a specification. The empirical law can, for example, be a multiplication of the average slope am of the linear regressions obtained in [Fig. 2A] by the immersion time.

[0059] By way of example, the specification may correspond to a target diffuse reflectivity RDC. The immersion time of a plate may then be greater than or equal to (RD,C -RD,o) / am, Rd,o being the diffuse reflectivity of the cutting face 101 measured during the second step.

[0060] The specification may correspond to a maximum roughness level of the cutting face below which it is possible to deposit a thin film of a given thickness by conformal liquid deposition or evaporation. If applicable, the specification is generally a peak-to-valley roughness measured less than or equal to the thickness of the thin film. The specification may incorporate additional roughness provided by an optional nanostructuring step implemented between immersion and thin film deposition. The thin film may, for example, have a thickness less than or equal to 1 µm.

[0061] The third step may be preceded or include a substep of selecting a set of plates such that all the plates in the set have a diffuse reflectivity measured in the second step, less than or equal to a predetermined threshold. The A threshold can be predetermined to ensure that the immersion time of each plate in the assembly is less than a maximum immersion time beyond which its post-immersion thickness would make the plate too brittle, or alternatively, to ensure that the immersion time remains less than a maximum immersion time beyond which the process would be too costly. For photovoltaic cell applications, it is considered that the plate thickness must remain greater than 60 µm, or even 90 µm, to avoid the risk of breakage. Thus, a plate thickness variation during immersion in the solution of 60 µm or less is desirable, preferably 50 µm or less. When each plate includes a work-hardened area, the thickness variation is advantageously sufficient to remove the entire work-hardened area from all the plates.

[0062] In the fourth step of the chemical polishing process example according to the invention, the cutting face 101 is etched by anisotropic wet etching along silicon crystal planes, i.e., the cutting face 101 is brought into contact with an etching solution that etches certain silicon crystal planes more rapidly than others. A tetramethylammonium hydroxide (TMAH) solution or a potassium hydroxide (KOH) solution can, for example, be used for this purpose. The solution can be brought into contact with the cutting face 101 by any means known to those skilled in the art.

[0063] Preferably, each plate in the assembly is immersed in a bath containing the etching solution. Advantageously, several plates are immersed simultaneously. The bath can be maintained at a substantially constant temperature, for example, 70 °C or 80 °C. Preferably, as in this example of a process, the etching solution is a concentrated potassium hydroxide solution, in order to reduce manufacturing costs. The mass concentration of KOH in the solution can, for example, be greater than or equal to 20%, 30%, or 35%.

[0064] It is possible to group the set of plates supplied in the first step into subsets of plates based on the diffuse reflectivity measurements obtained in the second step. The cutting faces 101 of the plates in each subset have homogeneous diffuse reflectivities, for example, within 10%, 5%, or even 1%. The immersion time determined for each plate in a subset can then be equal to a common immersion time so as to guarantee that the post-immersion roughness of the cutting face 101 is less than the specification for all the plates in the subset when they are immersed simultaneously in a bath containing the etching solution. The common immersion time can, for example, be equal to the longest immersion time determined for the subset of plates using the empirical law.

[0065] The fourth step reduces the roughness of the cutting faces 101, thus producing a batch of smoothed plates. Figure 4 shows an optical microscope image of an example of a cutting face obtained after the fourth step, following an immersion of 8 minutes. The roughness is dominated by mesa-shaped crystalline structures, here with horizontal dimensions of approximately 38 pm. It was observed that the width of the mesas increases substantially proportionally with the duration of the immersion.

[0066] The batch of smoothed wafers can be used to manufacture silicon-based photovoltaic cells of all types. It is particularly advantageous for reducing the manufacturing cost of tandem silicon-perovskite photovoltaic cells, notably due to the absence of mechano-chemical polishing and the adjustment of the immersion time.

[0067] In the manufacturing of photovoltaic cells, a layer is often deposited by liquid or evaporation onto the cutting face 101. This layer may be a functional layer, meaning that it plays a role in the operation of the photovoltaic cell. For example, it may be a layer involved in collecting electrical charges.

[0068] In the case of manufacturing silicon-perovskite tandem cells, the layer can be a perovskite layer or a contact layer. A contact layer can be a hole-selective or electron-selective layer. Preferably, nanostructures are formed on a region of the cutting face 101 intended to receive the layer and a light flux to be absorbed by the tandem cell. The nanostructures can be formed after anisotropic wet etching.

[0069] The sizes of the nanostructures and their pairwise separation distances are adapted to improve the absorption efficiency of the light flux by the tandem cell. When the cut face 101 has mesas resulting from the anisotropic etching, these are generally too large relative to the wavelengths of the light flux to achieve this objective, particularly when the light flux is solar flux. By way of example, the nanostructures can have a height less than or equal to 1 pm. They can each have a base with horizontal dimensions between 600 nm and 800 nm. They can be obtained by etching in a solution comprising dilute potassium hydroxide (KOH) and one or more surfactants.

[0070] Thus, it is possible to remove at a lower cost, a work-hardened area created by cutting the plate, and to obtain a cutting face 101 of controlled roughness allowing both to increase the conversion efficiency, or yield, of the tandem cell and to deposit by liquid means in a conforming manner the constituent layers of the tandem cell.

[0071] When the process includes a substep of selecting a set of plates, the unselected plates can advantageously be retained to manufacture standard silicon-based photovoltaic cells, not requiring precise control of the roughness of the cutting face 101. The standard photovoltaic cells can be single-junction photovoltaic cells, for example heterojunction cells or PERC (for "Passivated Emitter and Rear Cell") or Topcon (for "Tunnel Oxide Passivated Contact") type cells.

[0072] Specific embodiments have just been described. Various variants and modifications will be apparent to those skilled in the art.

Claims

Demands

1. A chemical polishing process for obtaining a batch of smoothed wafers, the process comprising the following steps: • supplying several silicon wafers, each comprising a cutting face, • measuring a diffuse reflectivity of the cutting face of each silicon wafer, • determining an immersion time per wafer for a set of wafers from among the supplied wafers, each immersion time being a function of the diffuse reflectivity of the cutting face of the corresponding wafer, • anisotropic wet etching of each wafer in the set of wafers along silicon crystal planes, by immersion in an etching solution for the corresponding determined immersion time, to obtain the batch of smoothed wafers, each immersion time being determined to ensure a conforming deposition of a thin layer on the cutting face of each wafer in the batch of smoothed wafers.

2. Chemical polishing process according to claim 1, wherein the anisotropic etching is etching by a solution containing a mass concentration of KOH greater than or equal to 20%, at a temperature greater than or equal to 70°C.

3. Chemical polishing method according to claims 1 or 2, wherein each silicon wafer comprises a work-hardened area extending in depth from the cutting face, and anisotropic etching removes the work-hardened area from all the wafers in the assembly.

4. Chemical polishing method according to any one of claims 1 to 3, wherein the anisotropic etching step leads to a thinning of each plate of the assembly, less than or equal to 60 pm.

5. A chemical smoothing process according to any one of the preceding claims, wherein the step of determining an immersion time comprises a substep of grouping the set of plates into subsets of plates with homogeneous diffuse reflectivity, wherein, for each sub- together, the immersion times of all the plates in the subset are determined so as to be equal to a common immersion time of the subset, and in which all the plates of each subset are immersed simultaneously in a bath containing the etching solution for a time equal to the common immersion time during the wet etching step.

6. A chemical smoothing process according to any one of the preceding claims, further comprising a selection of the plate set from the plates supplied having their diffuse reflectivity less than or equal to a predetermined threshold guaranteeing an immersion time less than or equal to a predetermined maximum immersion time.

7. A method for manufacturing an assembly of silicon-based photovoltaic cells, comprising the following steps: • obtaining a batch of smoothed wafers by a chemical polishing process according to any one of claims 1 to 6, wherein the thin film is a functional layer of one or more photovoltaic cells of the assembly, • conformal deposition by liquid or evaporation onto the cutting face of each wafer of the batch of the functional layer.

8. A manufacturing method according to claim 7, wherein the cutting face is nanostructured between the anisotropic etching step and the deposition step.

9. A manufacturing method according to claims 7 or 8, wherein the first cells of the photovoltaic cell assembly are silicon-perovskite tandem cells, and wherein the functional layer is a perovskite layer or a contact layer for manufacturing the first cells.

10. A manufacturing method according to claim 9, wherein the second cells of the photovoltaic cell assembly are silicon-based single-junction photovoltaic cells, wherein the batch of smoothed wafers is obtained by a smoothing process according to claim 6, and wherein the manufacturing process further comprises manufacturing the second cells

11. in and / or on the plates of a subsidiary batch chosen from the supplied plates not selected during the selection step. Manufacturing method according to any one of claims 7 to 10, wherein the functional layer has a thickness less than or equal to 1 µm.