Method for manufacturing timepiece components

Abstract

The present invention relates to a method for manufacturing watch components (110), in which, in a silicon-based wafer (100), watch components (110) are formed according to target dimensions, as well as at least one test specimen (120) of geometry different from that of the components (110), at least one characterization parameter of the test specimen (120) representative of a characteristic of the components is determined, and from said parameter, said characteristic of the components is determined and / or it is determined whether a dimensional correction should be applied to the components (110).

Description

Technical field of the invention

[0001] The present invention relates to the field of manufacturing watch components, such as watch balance springs, escape wheels, anchors, etc. State of the art

[0002] A watch movement consists of a plurality of components that interact closely with each other. These components include, for example, a pivoting balance wheel and a balance spring designed to produce a restoring torque on the balance wheel, which together determine the reference frequency from which the measurement of the passage of time is taken.

[0003] To guarantee the precision of the movement, it is necessary to precisely control the characteristics of each of the components used. For example, the aforementioned reference frequency depends directly on the moment of inertia of the balance wheel and the stiffness of the balance spring, which must therefore be perfectly controlled.

[0004] A few years ago, new materials began to be used for the manufacture of watch components, which allowed new manufacturing techniques such as deep reactive ion etching (also called DRIE etching, which is an acronym for Deep Reaction Ion Etching) to be employed.

[0005] However, significant geometric variation has been observed between identical components from the same substrate, and even greater variation between identical components from different substrates. If this variation is not controlled, it can impact the accuracy of the watch movement.

[0006] For example, the figure 1 schematically illustrates the cross-sections of the blades of three spirals S1, S2, S3 formed according to identical target dimensions, on different substrates P1, P2, P3.

[0007] The heights h1, h2, h3 of the spiral blades (measured in the direction of the engraving) are identical. But their widths l1, l2, l3 are all different from the target width lc, and are, moreover, different from each other.

[0008] However, the stiffness Rs of a spiral of constant cross-section depends on its material and dimensions, and is defined by the formula [1] below: R s = E × l s 3 × h s 12 L s

[0009] With Rs = spiral stiffness; E = modulus of elasticity; ls = blade width; hs = blade height; Ls = blade length

[0010] The dimensional differences between the spirals therefore also lead to a dispersion of their stiffness. This is evident from the figure 2 which schematically represents, for each batch of spirals made on each plate P1, P2, P3, the dispersion of stiffness R which follows a Gaussian distribution (curves C1, C2, C3).

[0011] Several processes have been developed in recent years with the aim of centering the mean Gaussian distribution of the spirals on a nominal stiffness value and reducing the standard deviation of the Gaussian.

[0012] Document EP 3181938 proposes, for this purpose, to create a balance spring with dimensions larger than those required to obtain a spring with a predetermined stiffness. This is achieved by measuring the stiffness of the spring by coupling it to a balance wheel with a predetermined inertia, calculating the thickness of material to be removed to obtain the necessary dimensions for the desired stiffness, and then removing this thickness from the spring. This approach requires high precision in measuring the spring's frequency to determine its stiffness. Furthermore, this frequency measurement is a meticulous and time-consuming operation. It also carries a risk of contamination of the springs.

[0013] Document EP 3 845 770 A1 proposes using a corrected mask for the photolithographic structuring of a photosensitive resin layer supported by a substrate layer. The mask has windows, each defining the outline of a spiral, with at least some windows having different dimensions to compensate for the stiffness variation among the substrate spirals. However, this method has not proven sufficiently effective in centering the Gaussian stiffness distributions of different substrates to a single nominal stiffness. Finally, document EP3915788A1 proposes a method for manufacturing spirals in a silicon-based wafer, in which some spirals are detached and measured to correct the dimensions of the spirals remaining on the wafer.

[0014] Therefore, there is still a need to improve the manufacturing process of such spirals, particularly batches of spirals from different substrates, in order to control their characteristics and especially their stiffness, while avoiding the aforementioned disadvantages.

[0015] The problem arises in a similar way for the manufacture of other watch components, in particular the balance wheels mentioned above or the anchors or escape wheels, which are subject to the same effects of dimensional dispersion during manufacture. Summary of the invention

[0016] One aim of the present invention is to propose a method for manufacturing watch components that meets the aforementioned needs.

[0017] According to the invention, this goal is achieved through a process for manufacturing components according to the terms of claim 1 as annexed.

[0018] Within the scope of the present invention, the components may include (but are not limited to) watch balance springs or anchors or escape wheels for watch movement.

[0019] Watch components are generally manufactured in a substrate comprising a first silicon-based layer in which the components are to be formed, a second layer called "support" also silicon-based and used to stiffen the substrate and, between these two layers, an intermediate layer also called "stop layer", made of silicon oxide.

[0020] The formation of watch components is typically carried out by deep ion reactive etching (DRIE) through a photosensitive resin mask formed by photolithography on the first layer of the substrate.

[0021] At the end of the forming process, the components were cut from a wafer comprising at least a useful portion of the first layer of the substrate, said useful portion being freed from the support and the arrest layer.

[0022] As explained previously, it is established that illumination and development processes of photosensitive resin such as DRIE etching result in dimensional dispersion between components from different substrates of the same series, which can affect the characteristics of the components.

[0023] The process allows for the deduction, from the characterization of a test structure (specimen) formed on the same wafer as a plurality of components, of at least one characteristic of these components. As an alternative or complement, the process can also identify a deviation between the target dimensions (those theoretically desired) and the actual dimensions of the components (obtained after etching) on ​​the same wafer, and, if necessary, quantify said deviation to allow for subsequent correction.

[0024] Using a test specimen whose geometry and dimensions can be chosen and adapted as needed allows for the systematization of characterization measurements, which can also be quick and simple. The speed and reproducibility of the manufacturing process are improved, notwithstanding variations in the geometry of the components being manufactured from one batch to another.

[0025] Contact with the components to be manufactured is also avoided, eliminating the risks of pollution or damage to them.

[0026] By allowing for better characterization of the manufactured parts, the process also ultimately allows for optimization of the pairing between the different components of the watch movement and for improvement of its precision.

[0027] In the context of the present invention, a test specimen is understood as any element, made in the same wafer as the components to be manufactured, and having a target geometry (i.e. reference geometry) different from the target geometry of said components.

[0028] The target dimension(s) or geometry of a test specimen or component refers to its dimension(s) or geometry as viewed in orthogonal projection onto the overall plane of the wafer (i.e., projected along the height or smallest dimension of the wafer) and as targeted during manufacturing. The target dimension or geometry during manufacturing refers to the dimension or geometry that is desired after etching for said components, without taking into account any minor deviations resulting from the manufacturing process, and in particular the etching operation.

[0029] Similarly, in this application, and unless otherwise specified, a shape of an element refers to said target geometry

[0030] Furthermore, in this application, width is always measured in projection onto the plane of the plate. Height is measured orthogonally to the general plane of the plate.

[0031] Finally, in the present application, a section of an element is a cross-section of said element in a plane orthogonal to the general plane of the plate.

[0032] In some cases, the target geometry of components may be intentionally chosen to differ from a theoretical geometry that would achieve the desired final characteristics for those components. For example, the target dimensions, and in particular the target width, may be chosen to be larger than the theoretical dimensions, and especially the theoretical width, that would achieve the desired final characteristics for the components. For instance, spirals may be formed on a wafer with target dimensions larger than those theoretically required to achieve the desired final stiffness value, to allow for subsequent adjustment by material removal.

[0033] The actual geometry at the end of the engraving process generally diverges from this target geometry due to variations in the process.

[0034] Otherwise, the test specimen has the same height as the manufactured components and is made of the same material. It can even sometimes have the same cross-section as the components, if its target width is the same.

[0035] A test tube used in the context of the present invention is a sacrificial element, which serves exclusively for the characterization of the components to be manufactured and is not intended to be used in a watchmaking mechanism.

[0036] In one method, the test specimen does not have the same general shape as the components to be manufactured. In a case where the components to be manufactured are spirals, anchors or wheels, the test specimen may, for example, have the shape of a beam or a straight blade.

[0037] In a second method, the test specimen can also have the same general shape as the components to be manufactured, but with a target geometry different from that of said components. For example, if the components to be manufactured are spirals, at least one test specimen can be a reference spiral with a target geometry different from that of the spirals to be manufactured.

[0038] A different geometry will for example be characterized by a different general / median line (e.g. a different curvature, a different number of turns, a different number of teeth, etc.) and / or different dimensions and / or a different external profile (e.g. a different flank profile).

[0039] A single test specimen can be formed on the plate. Alternatively, a plurality of test specimens can be formed in the plate, and step b) may in this case include calculating an average of at least one parameter over the plurality of test specimens.

[0040] According to an advantageous example, at least one test specimen can be a blade extending in a straight line, in particular between its two ends, at least one of which is fixed to the plate and the other may or may not be free.

[0041] The blade may include at least one portion of constant cross-section, and in particular, it may have a constant cross-section along its entire length. The blade may also include a portion of enlarged cross-section at its distal end.

[0042] In step b), the specimen can be characterized directly on the plate. In this case, the opening made in the plate around the specimen can advantageously be chosen to be sufficiently large to allow it to be deformed in the plane of the plate.

[0043] As a less advantageous variant, the test specimen can be characterized after having been previously detached from the plate and fixed on a test support.

[0044] Step b) includes, for example, determining a specimen characterization parameter from among: a dimension of the specimen, in particular its width, a deformation value of the specimen, a velocity or acceleration of displacement of the specimen, a natural vibration frequency of the specimen, one or more resonance frequency(ies) of the specimen, an effective stiffness of the specimen, or a plurality of these parameters.

[0045] In this application, the natural frequency of a system is the frequency at which that system oscillates when it is freely evolving, that is, without an excitatory force. A resonance frequency of a system subjected to an excitatory force is a frequency at which a local maximum of displacement amplitude can be measured for a given point of the system.

[0046] A characterization parameter can be measured directly or can be obtained by deduction from another previously measured parameter.

[0047] Step b) may in particular include a measurement and / or determination of a frequency, including a natural frequency or one or more resonance frequencies, and / or an amplitude and / or a velocity and / or an acceleration of displacement of at least one point of the specimen.

[0048] A particularly advantageous feature is that the specimen is defined in a base plane corresponding to the general plane of the insert during its formation, and the displacement is along a direction contained within said base plane. The plane of displacement of the specimen during characterization then coincides with the plane of displacement of the components in operation, making the characterization more precise and direct. The characterization parameter expresses the properties, particularly mechanical and / or geometric, of the specimen and the components with the same weighting of the different value(s) considered.

[0049] According to an example, in step b), a vibratory excitation is applied to the test specimen, and at least one vibratory characteristic of the test specimen is determined during the vibration.

[0050] Vibratory excitation here means any excitation that causes the test tube to vibrate.

[0051] According to one example, the vibratory excitation is maintained.

[0052] According to another example, the vibratory excitation is punctual and not sustained.

[0053] Advantageously, vibratory excitation is variable (or modified) over time. In other words, vibratory excitation involves exciting the specimen at different frequencies over time. For example, several successive excitation phases can be carried out, with the specimen being excited at a given frequency (sustained single-frequency excitation) during each excitation phase.

[0054] In this case, at least one resonance characteristic of the specimen can be identified during vibration, for example a resonance frequency or a resonance peak.

[0055] Advantageously, the vibratory excitation covers a predetermined frequency range, preferably located in the vicinity of the resonance frequency.

[0056] Vibrational excitation can be imposed by a source of acoustic vibrations, for example a piezoelectric source.

[0057] As an alternative, excitation can be imposed by a mechanical excitation source.

[0058] As an example, vibratory excitation is an impulsive excitation produced, for instance, by a shock applied directly to the specimen or to an element attached to the specimen at the time of excitation, for example, to the plate when the specimen is still attached to the plate. Such an impulse or shock can, for example, be induced using an excitation hammer.

[0059] In another example, the test specimen could be loaded by an end force to be deformed, and then released.

[0060] The vibrational characteristic of the specimen determined during or after excitation can, for example, be a characteristic of a natural frequency or a resonance frequency of the specimen, or it can be a velocity or an acceleration or an amplitude of displacement allowing possibly to deduce a natural frequency or a characteristic of a resonance frequency of the specimen.

[0061] According to another example, in step b), a predetermined bending force is applied to the specimen and at least one amplitude of displacement of the specimen is identified under the effect of said bending force.

[0062] According to yet another example of characterization, step b) includes a measurement of at least one dimension of the specimen (for example its width), including an optical measurement, including a laser measurement.

[0063] According to yet another example, step b) includes measuring the stiffness of at least one test specimen in the form of a reference spiral.

[0064] A characteristic of the components that can be deduced from a specimen characterization parameter determined, for example, according to one of the aforementioned methods, may be in particular: a critical dimension of the components, or a mechanical property of the components, for example a stiffness or an inertia, or even a characteristic related to the future operation of the components, in other words a characteristic representative of the way in which said components will interact with other components of watch movement in their movement, in particular an efficiency or a frequency or an amplitude of movement (at iso parameters).

[0065] Critical dimension means, for example, a height, an inclination (especially a flank inclination), a radius of curvature.

[0066] The process makes it possible in particular to characterize dimensions that are difficult to measure optically, especially dimensions that are difficult to access with optical measurement systems.

[0067] For example, in step c) it is determined whether a dimensional correction should be applied to the components. The method according to the invention makes it possible to detect the dimensional error related to photolithography and etching. DRIE thanks to the test tube(s) provided on each plate.

[0068] For example, the process may include a step of comparing the specimen characterization parameter or the corresponding component characteristic with a reference value, to deduce the need for dimensional correction.

[0069] A characteristic of the components and / or the need for dimensional correction of the components can, for example, be deduced from the characterization parameter(s) of the specimen. by means of a predictive model, for example implementing a machine learning algorithm or a polynomial formula, and / or through a lookup table.

[0070] According to one provision of the invention, in step c) a dimensional correction value to be applied to the components is further determined.

[0071] In other words, the process not only allows the dimensional error to be identified, but also allows the dimensional correction value to be quantified.

[0072] The dimensional correction value typically corresponds to a material thickness to be removed from the components to achieve the desired dimensions. In this case, the components are initially manufactured to target dimensions, including a target width, that are larger than the theoretical dimensions, including the theoretical width, required to achieve the desired stiffness.

[0073] The correction is usually—though not necessarily—applied to each lateral face of a component (for example, to vary its width) and generally also to its top and bottom faces (to vary its height). Very often, the correction more broadly consists of a reduction in the entire thickness of the wafer (across all its surfaces), including the components and the test specimen(s).

[0074] The dimensional correction value can also be determined by a predictive model, for example implementing a machine learning algorithm or a polynomial formula.

[0075] As an alternative, the determination of the dimensional correction value is carried out via a lookup table.

[0076] According to one example, the process further includes a step d) in which the dimensions of the components are actually corrected according to the correction value determined in step c).

[0077] Step d) advantageously includes at least one oxidation step and one deoxidation step of the components.

[0078] According to a particular example, several or all of the watch components may be subjected to thermal oxidation prior to step b). In this case, the dimensional correction step d) may include at least one partial deoxidation step and one component reoxidation step.

[0079] According to a specific provision, step a) includes at least the following substeps: a1) provide a substrate comprising a first silicon-based layer, a second silicon-based layer and an intermediate silicon oxide-based layer, a2) cover the first layer with a layer of photosensitive resin, a3) by photolithography, form in said resin layer openings corresponding to the contours outside the attachment of the watch components and at least one test specimen, a4) etch the first layer over its entire height under said openings to form the watch components and at least one test specimen.

[0080] By contour outside attachment of a part we mean here the contour of the part except for a portion of its external contour intended to form an attachment holding it to the rest of the substrate / wafer.

[0081] Substep a3 may, in particular, include the following steps: a31) expose the photosensitive resin layer using a light source, through a photolithographic mask with windows defining the unattached contours of the watch components and the test specimen, a32) locally remove the photosensitive resin using a solvent to form the openings.

[0082] According to an advantageous arrangement, the photolithographic mask is a corrected mask, at least some windows of the mask being of different dimensions so as to reduce the extent of a dimensional dispersion between at least some of the watch components. Brief description of the drawings

[0083] The features and advantages of the present invention will become apparent in greater detail in the following description of illustrative and non-limiting examples of embodiments, with reference to the attached drawings, in which: There figure 1 illustrates the cross-sections of the blades of three spirals formed, according to identical target dimensions, on different plates, The figure 2 illustrates the dispersion of spiral stiffnesses of the same series, formed on three different plates. figure 3 represents a plate in which a plurality of watch spirals and a test tube have been formed, The figure 4 is an enlarged view of detail A of the figure 3 , THE Figures 5A to 5D These are schematic and partial views illustrating the manufacturing process of the plate. figure 3 , THE figures 6 to 9illustrate four other examples of test tubes that can be used in the process of the invention, The Figure 10 illustrates a method for characterizing a test tube, The figure 11 illustrates a second method for characterizing a test tube, The figure 12 illustrates a third method for characterizing a test tube, The figure 13 illustrates a lookup table that can be used to determine the dimensional correction value. figures 14A to 14C illustrate the dimensional correction of a spiral according to one embodiment of the invention, The figure 15 illustrates a variant of cutting a test specimen from a silicon substrate. Detailed description

[0084] The invention is illustrated below by examples relating to the manufacture of balance springs. However, the same principles are applicable to any other type of watch component.

[0085] There figure 3The diagram shows a top view of a silicon wafer 100 in which several watch balance springs 110 have been formed with the same target geometry, and in which a test specimen 120 has also been formed with a target geometry different from that of the components. The target geometry of the components is the geometry that is intended to be obtained after etching and before any subsequent dimensional correction. It may be identical to, or different from, a theoretical geometry that would allow the desired characteristics of the balance springs to be achieved, in particular a specific stiffness. The actual geometry after etching may be slightly different from this target geometry due to variations in the process and, in particular, geometric dispersion within the wafer.

[0086] Such a 100-page plate can be produced by implementing the following process steps, illustrated in the Figures 5A to 5DA substrate 10 is provided, typically a silicon-on-insulator (SOI) substrate, comprising a first layer of silicon 11 in which the spirals are to be formed, a second layer of silicon 12, and, between the two, an intermediate or arrest layer 13 of silicon oxide. The silicon of the first and / or second layers 11, 12 may be polycrystalline, but is preferably monocrystalline. It may be undoped or doped, for example, doped with phosphorus. Doped silicon, being electrically conductive, is more dimensionally stable and has better mechanical strength. Preferably, the substrate 10 is made of silicon that is oriented {1,1,1} and / or is doped, advantageously so as to have a resistivity less than or equal to 0.1 Ω·cm⁻¹, for example, a resistivity of 0.05 Ω·cm⁻¹.

[0087] A layer of photosensitive resin 20 is deposited on the first layer 11 of the substrate 10.

[0088] Through a photolithographic mask 30 equipped with windows 31 whose shapes correspond to the contours of the spirals 110 and the test tube 120, the resin layer 20 is exposed to the light radiation R from a light source (not shown). The radiation R may, in particular, be radiation including UV, i.e., ultraviolet radiation, or even consist of UV radiation.

[0089] According to an advantageous arrangement, the photolithographic mask 30 is a mask corrected according to the teaching of document EP 3 845 770 A1, at least some windows 31 of the mask being of different dimensions so as to aim for the same target geometry for all components, compensating for the dispersion of stiffness within the substrate.

[0090] The irradiated photosensitive resin is then locally removed by dissolving it in a suitable chemical bath to form openings 21 corresponding to the desired contours of the spirals and the test tube. The resin is said to be "developed": in the case of a positive resin as illustrated, it is the portion of the resin exposed to the R radiation from the light source that is removed. The photosensitive resin that has not been irradiated and transformed by the R radiation resists the chemical bath, thus remaining in place on the substrate and forming a sacrificial mask.

[0091] In an implementation variant with a so-called negative resin, the photosensitive resin is chosen so that its irradiation by radiation analogous or identical to R radiation renders this resin insoluble in a suitable chemical bath, in which case the mask is transformed accordingly.

[0092] The silicon of the first layer 11, which is no longer protected, is then etched along its entire height through the openings 21 of the resin, in a direction perpendicular to the general plane of the substrate 10 ( figure 5C ). Several watch spirals 110 as well as the test tube(s) 120 are cut collectively, together and at the same time, all in the first layer 11.

[0093] The etching process is a deep reaction ion etching (DRIE) or Bosch etching. Behind the mask, the first layer 11 is not etched. Furthermore, the stop layer 13 is less sensitive to the etching, which therefore does not affect it, or only to a very small extent.

[0094] Since the inclination of the etching flanks depends on the width of the etched surface, it is preferable, to ensure a correspondence between the inclination of the specimen flanks and that of the component flanks, that the width of an etching border around the specimen be approximately equal to the width of the etching borders around the components. The widths of the resin openings 21 are therefore configured accordingly.

[0095] According to one possible implementation method, for each component as for the test specimen, both an engraving line corresponding to the external profile of the element and a peripheral line around this external line can be engraved. When the wafer is released, the portion located between the external and peripheral lines is detached, freeing up an empty space 130 around the element as illustrated in the figure 4 allowing the element to be moved around its attachment in order to break the latter.

[0096] As an alternative, one could also, as on the figure 15 , engrave a single external engraving line 130' delimiting the test specimen, said line having a width substantially identical to the width of the engraving lines delimiting the components.

[0097] In a further step, the support 12 and the stop layer 13 are then locally removed, or completely eliminated, by chemical etching, so that the plate 100 bearing the watch balance springs illustrated on the figure 1 ( figure 5D ).

[0098] In this application, a wafer is generally understood to mean any silicon-based structure comprising a useful portion (where appropriate freed from the support and intermediate layer of the substrate from which it originates) in which the components to be manufactured are cut, and at least one test specimen with a target geometry different from that of the spirals.

[0099] The plate ultimately holds the watch balance springs 110 and the test specimen 120, so that these can easily be moved together and processed collectively and simultaneously. As illustrated on the figure 3 For example, the test specimen 120 can be made in the useful portion of the plate 100 where the spirals 110 are formed and be surrounded by spirals. Alternatively, the test specimen 120 could be placed anywhere on the plate 100.

[0100] The 120 test tube of the figure 3 appears in more detail on the enlarged view of the figure 4 In this example, the test specimen 120 is connected to the rest of the plate 100 by a proximal end 121 and is free at its distal end 122. The opening 130 made in the plate 100 around the test specimen 120 allows, if necessary, for it to be deformed in the plane of the plate 100.

[0101] In the example of the figure 3The specimen 120 is a blade extending straight along a principal direction Z between the proximal end 121 and the distal end 122. Here, the blade has a constant rectangular cross-section and, in particular, a constant width It along its entire length Lt measured in its principal direction Z. This embodiment is not, however, limiting, and the figure 6 illustrates another example in which the blade 120 has a variable cross-section along its main rectilinear direction Z and terminates, at its distal end 122, in a locally enlarged portion 123 forming a mass. According to yet another example illustrated on the figure 7 The test specimen can be in the form of a blade, for example with a constant cross-section as on the figure 3 , but fixed to the rest of the plate at both ends. According to yet another example illustrated on the figure 8, the test piece 120 can be a spiral with a different profile from that of the spirals to be manufactured 110, with a constant or non-constant cross-section over its entire length Lt measured between its proximal end 121 attached to the plate and its free distal end 122.

[0102] These examples are given for illustrative purposes only and are not exhaustive. Any other suitable geometry, different from that of the spirals to be manufactured, can be considered for the test specimen(s).

[0103] The processes of illumination and development of photosensitive resin, as well as engraving DRIE, result in dimensional differences between the widths l of elements supposed to be identical on different plates.

[0104] As explained previously, in the field of spirals, we are interested in compensating for this dimensional dispersion in order to make the average stiffness of the spirals of the same plate tend towards a nominal value.

[0105] The method according to the invention makes it possible to detect the dimensional error related to the engraving by characterizing the test specimen(s) and to quantify the dimensional correction value to be applied to the balance springs to achieve the desired average stiffness. In the context of balance springs, the dimensional correction value refers to the thickness e of material to be removed from each spring 110 to achieve this stiffness.

[0106] Since the characterization is performed indirectly, via the test specimen, the spirals are preserved. Furthermore, as the 120 test specimen can take on very varied geometries, as mentioned previously, the characterization methods can be quite diverse, and some of these are described below as non-exhaustive examples.

[0107] In these examples, the test specimen 120 is characterized directly on the plate 100. In other words, at least one parameter of the test specimen 120, representative of the difference between the effective dimensions of the spirals and their target dimension or a theoretical dimension allowing to obtain the desired stiffness, is determined while the test specimen is still attached to the plate 100.

[0108] Alternatively, the test specimen 120 can also be characterized after being detached from the plate 100 and fixed to a test support. In this case, as illustrated in the figure 9The test specimen 120 is attached to the plate 100 by one or more release tabs 140. Advantageously, such a release tab 140 has a tapered section 141 at its end located near the test specimen 120. This easily breakable section 141 allows for easy detachment of the test specimen 120. When the test specimen 120 is to be characterized on a test support, it is advantageously attached, at least at one end, to a mounting base 124, preferably wider than the test specimen itself, for fixing it to the support. In this case, the mounting base 124 and the test specimen form a detachable assembly 125 from the plate. The mounting base 124 is preferably located between the release tab 140 and the test specimen 120.

[0109] According to a first characterization method, the specimen 120 is subjected to vibratory excitation, and a vibratory response of the specimen is measured to deduce a characterization parameter such as a resonance frequency.

[0110] In the example shown on the Figure 10 A piezoelectric source 300, or any other suitable source, is used to induce acoustic excitation on the specimen 120 within a frequency range chosen to include at least one resonance frequency of the specimen. By covering or sweeping a wide frequency range, several resonance peaks (or resonance frequencies) can be measured, which can provide greater accuracy.

[0111] For example, one or more frequencies are applied at a time, which are varied over time, in the vicinity of one or more resonance frequencies.

[0112] The excitation is preferably maintained for a sufficient duration to deduce vibrational spectra of the test tube.

[0113] During excitation, for example, an amplitude, velocity, acceleration, or frequency of displacement of at least one point of the specimen 120 can be recorded via a suitable optical measuring means 310 such as a laser vibrometer by Doppler effect.

[0114] The test specimen 120 is generally contained in a base plane of the plate 100. The displacement whose characteristics are measured is along a direction contained in this base plane or, alternatively, along a direction transverse to said base plane.

[0115] The signal thus obtained is then processed, usually by a computer program 320, to deduce whether a dimensional correction is necessary and which one.

[0116] Signal processing typically includes a step of identifying the resonance frequency as a function of the measured amplitude, velocity, or acceleration, for example by Fourier transform frequency analysis.

[0117] It can be noted that the process may or may not include an intermediate step of determining the stiffness of the specimen 120, allowing then to calculate / deduce a level of dimensional correction to be applied to obtain the desired final stiffness.

[0118] The dimensional correction value of the 110 spirals is determined, from the determined resonance frequency or where applicable from the calculated stiffness value, by a predictive model or by means of a lookup table.

[0119] The predictive mathematical model, for example, implements a machine learning algorithm, previously trained during test measurements, and can implement a classification carried out by a neural network, or by partitioning.

[0120] As an alternative, the predictive model can implement a polynomial formula, for example by performing linear regression modeling.

[0121] As an alternative, a lookup table can make it easier, although less precise, to determine the dimensional correction to be made.

[0122] Such a correspondence table can be obtained experimentally, for example by experimentally determining how much a removal of a thickness of one micron over the entire surface of the plate 100 changes the resonance frequency of the specimen 120. By proportionality, one can then determine the number of microns of thickness to be removed to obtain a predetermined resonance frequency, corresponding to a desired stiffness.

[0123] Alternatively, a lookup table can be used that directly links a variation in material thickness to a variation in specimen stiffness.

[0124] Excitement can be sustained as in the example above, but it can also be occasional and not sustained.

[0125] As an example, one could thus use an excitation hammer or any other source allowing to induce a pulsed vibratory excitation on the test specimen.

[0126] There figure 11 This illustrates an example of characterization in which specimen 120 is loaded, near its distal end, by a force F directed approximately orthogonally to its principal direction Z. Once the specimen is deformed, it is released (F=0) and then oscillates on either side of its equilibrium position. Its natural frequency can then be measured.

[0127] In this scenario, a 120 test tube, as illustrated in the..., can also be used to advantage. figure 6 , in the form of a blade ending at its distal end with an enlarged portion 123 forming a mass. This mass 123 amplifies the movement and facilitates measurement.

[0128] Regardless of the vibratory excitation method, the principle can remain the same as described previously, and consist of recording an amplitude or a speed or an acceleration of displacement of the specimen, then processing the signal thus recorded to extract a natural frequency or one or more resonance frequencies of the specimen, to finally deduce the correction value to be applied.

[0129] However, the vibrational response to acoustic or mechanical excitation is only one parameter among others allowing the characterization of the specimen 120 in order to determine the need for a correction and, if necessary, the level of dimensional correction to be applied to the spirals 110 of the plate 100.

[0130] According to another characterization method illustrated on the figure 12, the specimen 120 can be subjected to a predetermined bending force F, and its deformation amplitude (deflection f) can be measured by optical means as previously mentioned.

[0131] Specimen 120 then behaves like a free-fixed beam. Its deflection f allows, for example, the dimensional correction to be applied to be deduced directly or indirectly, using one or the other method mentioned previously.

[0132] There figure 13 illustrates an example of a correspondence table that can be created experimentally by measuring, for a given deflection f of the specimen, what is the average stiffness R s of the spirals, and then proceeding to several correction iterations, for example by oxidation and deoxidation or by partial deoxidation and reoxidation in the case of a spiral already oxidized.

[0133] The table is used as follows: Suppose we measure a deflection f with a value corresponding to the value f1 and we want a stiffness of value R s5, we will apply to the plate a dimensional correction value equal to the sum of the correction values ​​e2, e3, e4 and e5. When the deflection and stiffness values ​​are not exactly shown in the table, the dimensional correction value can be obtained by simple interpolation.

[0134] According to another characterization method, specimen 120 could also be simply measured using a suitable optical device. The dimensional correction to be applied could then be determined from the dimension(s) thus measured.

[0135] According to yet another characterization method, in a case where the test specimen 120 is a characterization spiral with a geometry different from that of the spirals to be manufactured 110, as illustrated for example on the figure 8 , we can measure the stiffness Rt of this characterization spiral 120, then by one or the other method mentioned previously, determine the dimensional correction to be applied to the spirals 110.

[0136] For example, a so-called dynamic stiffness measurement can be carried out, as described for example in document EP 2 423 764, by coupling the characterization balance spring 120 to a reference balance wheel with a predetermined moment of inertia I, by measuring the frequency of the resonator formed by the balance wheel / spring assembly and by deducing the stiffness Rt of the characterization balance spring 120 from the formula: f = 1 2 π Rt I

[0137] As an alternative, a so-called static measurement can also be carried out, as described for example in document EP 2 128 723, by coupling the characterization spiral to a reference spiral via an axis of a rotation system, applying a rotation to the reference spiral using the rotation system, the torque being transmitted to the characterization spiral via the axis, performing a differential measurement of the angle of rotation obtained on the characterization spiral and the angle imposed on the reference spiral, and using this measurement to calculate the stiffness of the characterization spiral using the formula below, in which Rt is the stiffness of the characterization spiral, Rr is the stiffness of the reference spiral, a is the angle of rotation of the reference spiral and a' is the angle measured on the characterization spiral. R t = R r × a a ′

[0138] Dimensional correction is typically achieved by removing material from the components.

[0139] In this case, the target dimensions of the components (in particular their width) are initially chosen to be larger than theoretical dimensions that would allow the desired final stiffness to be achieved for these components.

[0140] Dimensional correction can, for example, be achieved by oxidation and then deoxidation of the spirals, as illustrated in the... figures 14A to 14C .

[0141] There figure 14A This illustrates the cross-section of a blade of a spiral 110 after its formation. Its height is ha and its width is . The dimensional correction value to be applied to the spiral is denoted here as e.

[0142] There figure 14BThis illustrates the same spiral cross-section after a layer of silicon dioxide has formed on its outer surface, consuming some of the silicon. The oxidation is carried out in the humid atmosphere of a thermal oxidation furnace heated to a temperature typically around 1000°C. In the figure, reference numeral 112 designates the unoxidized silicon, and reference numeral 114 designates the silicon dioxide that has formed. The thickness e' of the silicon dioxide layer resulting from the oxidation is greater than the thickness of the unoxidized silicon 112, the oxidation of which led to this silicon dioxide 114.

[0143] A layer of silicon dioxide, preferably all of it, is then removed by etching or chemical attack, for example with hydrofluoric acid, to adjust the dimensions of the spiral. figure 14Crepresents the same section of the spiral, free of silicon dioxide. It has a height hb less than the height ha and a width lb less than the width la.

[0144] Alternatively, several or all of the watch balance springs may have been oxidized to an oxide thickness w after their formation. This oxide thickness w corresponds to the final thermocompensation thickness or a chosen thickness. In this case, the dimensional correction step d) may include at least one step of partial or total deoxidation and reoxidation of the balance springs to the value w.

[0145] The examples mentioned above illustrate the interest of the invention for the determination and / or control of mechanical properties of manufactured components, in particular stiffness.

[0146] As an alternative or complement, the characterization of specimen(s) according to the invention can also be used to obtain information on critical component dimensions, particularly dimensions that are difficult to measure using instruments. Specifically, specimen shapes and geometries can be simulated and optimized to obtain a vibration signature, as mentioned in connection with the Figure 10 allowing, through causal links, to trace back to any dimension of the components.

[0147] Characterizing the 120 test specimen can also allow us to determine the functional characteristics of the components, in other words, to predict how these components will interact with other watch movement components during operation. For example, the vibrational signature of a test specimen can be used to deduce its efficiency, frequency, or amplitude of movement (with identical parameters) during operation.

Claims

1. Method for manufacturing timepiece components (110), characterized in that: a) in a silicon-based wafer (100), timepiece components (110) and at least one test piece (120) with a target geometry different from that of the components (110) are formed according to target dimensions, b) at least one parameter characterizing the test piece (120) and representative of a characteristic of the components is determined, c) based on said parameter, said characteristic of the components is determined and / or it is determined whether a dimensional correction must be applied to the components (110), and in that, in step b), a vibratory excitation is applied to the test piece, chosen from: - an excitation imposed by an acoustic vibration source, - an impulse excitation produced by an impact applied directly to the test piece or to an element integral with the test piece, - an excitation by loading the test piece with an end force to deform it and then releasing the test piece, and at least one vibration characteristic of the test piece is determined during the vibration.

2. Method according to claim 1, wherein said at least one vibratory characteristic of the test piece is a resonance frequency or a natural vibration frequency.

3. Manufacturing method according to claim 1 or 2, wherein the vibratory excitation varies over time, and at least one resonance characteristic of the test piece is identified during the vibration.

4. Manufacturing method according to any one of the claims 1 to 3, wherein step (b) comprises measuring and / or determining a frequency and / or amplitude and / or velocity and / or acceleration of movement of at least one point on the test piece.

5. Manufacturing method according to claim 4, wherein the test piece (120) is defined in a base plane corresponding to the general plane of the wafer (100) during its formation, and the displacement is in a direction contained within said base plane.

6. Manufacturing method according to any one of the claims 1 to 5, wherein in step b), one dimension of the test piece is measured.

7. Manufacturing method according to any one of the claims 1 to 6, wherein in step c), a dimensional correction value (e) to be applied to the components (110) is also determined.

8. Manufacturing method according to claim 7, further comprising a step d) in which the dimensions of the components (110) are corrected according to the correction value determined in step c).

9. Method according to claim 8, wherein step (d) comprises at least one oxidation step and one deoxidation step of the components (110).

10. Manufacturing method according to any one of the claims 1 to 9, wherein the characteristic of the components and / or the need for dimensional correction and / or a dimensional correction value (e) to be applied to the components (110) is determined by means of a predictive model, for example using a machine learning algorithm or a polynomial formula.

11. Manufacturing method according to any one of the claims 1 to 10, wherein the characteristic of the components and / or the need for dimensional correction and / or a dimensional correction value (e) to be applied to the components (110) is determined by means of a correspondence table.

12. Manufacturing method according to any one of the claims 1 to 11, wherein the components (110) are timepiece hairsprings.

13. Manufacturing method according to any one of the claims 1 to 12, wherein at least one test piece (120) is a reference hairspring, and in step b), the effective stiffness (Rt) of said reference hairspring is measured.

14. Manufacturing method according to any one of the claims 1 to 13, wherein at least one test piece (120) is a tongue extending in a straight direction (Z).

15. Method according to claim 14, wherein the tongue has at least one portion of constant cross-section, preferably has a constant cross-section over its entire length.

16. Method according to claim 14 or 15, wherein the tongue has an enlarged section (123) at its distal end (122).

17. Method according to any one of the claims 1 to 16, wherein a plurality of test pieces (120) are formed in the wafer (100), and in step b) an average of the primary parameters from the plurality of test pieces is calculated.

18. Method according to any one of the claims 1 to 17, wherein the step a) comprises at least the following sub-steps: a1) providing a substrate (10) comprising a first silicon-based layer (11), a second silicon-based layer (12) and an intermediate silicon oxide-based layer (13), a2) covering the first layer (11) with a layer of photosensitive resin (20), a31) exposing the photosensitive resin layer (20) using a light source, through a photolithographic mask (30) provided with windows (31) defining the contours without attachment of the timepiece components (110) and the at least one test piece (120), a32) locally removing the photosensitive resin using a solvent to form openings (21) corresponding to the contours without attachment of the timepiece components (110) and of the test piece (120), a4) etching the first layer (11) over its entire height beneath said openings (21) to form the timepiece components (110) and the at least one test piece (120).

19. Method according to claim 18, wherein the photolithographic mask (30) is a corrected mask, at least some windows (31) of the mask being of different dimensions so as to reduce the extent of dimensional dispersion between at least some of the timepiece components (110).

Images