Method for manufacturing timepiece hairsprings

Technical field of inventions

0001The present invention relates to the field of watchmaking. More precisely, it relates to a process for manufacturing watch balance springs, which can in particular be watch balance springs with thermocompensation.

State of the art

0002In a clockwork movement, the function of an oscillator is to provide a reference frequency, from which the measurement of the passage of time is made. This oscillator is often called the regulating organ of the clockwork movement. It can be mechanical in particular when it is intended for a mechanical timepiece movement.

0003A mechanical oscillator conventionally used in watchmaking results from the association of a watch spring and a balance, which is mounted to pivot and which plays the role of a flywheel, while the watch balance spring is a spring intended for produce a return torque on this balance.

0004A few years ago, we began to use new materials for the manufacture of watch balance springs, resulting in new manufacturing techniques such as deep reactive ion etching (also called DRIE etching, which stands for Deep Reaction. Ionique Etching) could be used.

0005In the document <patcit id="pcit0001" dnum="EP1422436A"><text>EP 1 422 436</text></patcit>, it is proposed to manufacture a watch balance spring which comprises a silicon core and which has thermal compensation by virtue of a coating made of silicon dioxide.

0006Manufacturing several watch balance springs in the same plate, by engraving, reduces manufacturing costs. However, there is a geometric dispersion between watch balance springs belonging to batches, each of which consists of watch balance springs all cut according to the same pattern, by engraving, in the same plate.

0007In this regard, the document <patcit id="pcit0002" dnum="EP3181938A1"><text>EP 3 181 938 A1</text></patcit> proposes a method of manufacturing a watch balance spring having a predetermined elastic torque. In this process, a watch balance spring blank cut to dimensions greater than the target final dimensions is adjusted after cutting by removing material to a thickness calculated from a measurement on the watch balance spring blank. The solution proposed in this document<patcit id="pcit0003" dnum="EP3181938A"><text>EP 3,181,938</text></patcit> gives sufficiently precise results when it is applied individually, that is to say when a single watch balance spring is processed at a time, which is expensive and complicated to implement. When the solution proposed in<patcit id="pcit0004" dnum="EP3181938A"><text>EP 3,181,938</text></patcit> is applied collectively to a lot of watch balance springs cut by engraving from the same plate, a large number of watch balance springs in this lot have an elastic torque so far from the predetermined elastic torque that they are unusable and constitute a quantitatively very important rebus and little compatible, or even incompatible, with the profitability requirements of industrial production.

Summary description of inventions

0008A first invention has at least the aim of allowing a geometric dispersion between watch springs to be collectively reduced over several watch springs cut by etching from the same wafer.

0009According to the first invention, this object is achieved by virtue of a method of manufacturing watch balance springs in a wafer, comprising a step in which: b) between at least one source emitting radiation and a photosensitive resin carried by one face of the wafer, a reusable mask stops at least part of the radiation except in areas comprising several windows which define the contours of the watch balance springs and which comprise this reusable mask.

0010According to the first invention, the reusable mask is a reusable mask corrected in that at least part of the windows are dimensionally different from one another so as to reduce an extent of a dispersion of the elastic torques of at least part of the watch springs. made in the same wafer.

0011The manufacturing process defined above can incorporate one or more other advantageous characteristics, individually or in combination, in particular among those defined below.

0012Advantageously, the manufacturing process comprises steps which follow step b) and in which:<ul id="ul0001" list-style="none"><li>c) from the photosensitive resin, a sacrificial mask is produced on the wafer by removing from the wafer the photosensitive resin having been irradiated by the radiation in step b) or the photosensitive resin having been subtracted from the radiation in step b) by the corrected reusable mask,</li><li>d) in the plate, the watch balance springs are produced by etching using the sacrificial mask.</li></ul>

0013Advantageously, the manufacturing process is a manufacturing process of a batch of watch balance springs, the elastic torques of which have an average in a predetermined range. Advantageously, in step d), the watch springs are produced according to dimensions greater than the dimensions necessary for obtaining the batch of watch springs, the elastic torques of which have the average in the predetermined range. Advantageously, the manufacturing process comprises steps which follow step d) and in which:<ul id="ul0002" list-style="none"><li>e) determining the quantity of material to be removed from the watch balance springs produced in step d), in order to obtain the batch of watch balance springs having the elastic torques, the average of which is in the predetermined range,</li><li>f) from several or all of the watch balance springs produced in step d), said quantity of material to be removed is removed.</li></ul>

0014Advantageously, step f) is carried out collectively on the watch balance springs still attached to the plate.

0015Advantageously, in step e), the quantity of material to be removed is determined from a measurement carried out on a sacrificial sample taken from among the watch springs of the wafer and which is dissociated from the wafer.

0016Advantageously, in step e), the quantity of material to be removed is determined by means of a correspondence table indicating the thickness of material to be removed for each of several average elastic pairs or for each of several quantities linked to such average elastic torques.

0017Advantageously, the quantity of material to be removed is a thickness of material to be removed from the watch balance springs produced in step d).

0018Advantageously, the wafer is made of a silicon-based material, step f) comprising substeps in which:<ul id="ul0003" list-style="none"><li>f1) several or all of the watch springs produced in step d) are subjected to an oxidation so as to be transformed into watch springs oxidized superficially over a thickness which is the thickness of material to be removed determined in step e),</li><li>f2) superficially oxidized watch springs, the silicon-based material which has been oxidized during sub-step f1) is removed.</li></ul>

0019Advantageously, step d) is carried out by deep reactive ionic etching.

0020Advantageously, the manufacturing process comprises a step in which: g) on ​​each of several or all of the watch balance springs, a thermal compensation portion is formed which modifies the sensitivity to temperature variations of the elastic torque of the corresponding watch balance spring.

0021Advantageously, the manufacturing process comprises a preparatory step which precedes step b) and in which:<ol id="ol0001" compact="compact"><li>a) the corrected reusable mask is produced.</li></ol>

0022Advantageously, the preparatory step comprises sub-steps in which:<ul id="ul0004" list-style="none"><li>a1) a map of the degrees of correction to be applied in different positions on a wafer plane is produced in order to reduce a spatial heterogeneity of an engraving carried out by an engraving device,</li><li>a2) the corrected reusable mask is produced in such a way that the windows are dimensioned according to the mapping carried out in sub-step a1).</li></ul>

0023Advantageously, in the sub-step a1), the following is carried out:<ul id="ul0005" list-style="none"><li>a11) between at least the source and a photosensitive resin carried by one face of a test wafer, an uncorrected mask stops at least part of the radiation from the source except in areas comprising several windows which define the contours of structures test and which are distant from each other within the uncorrected mask,</li><li>a12) the photosensitive resin having been irradiated by the radiation or the photosensitive resin having been subtracted from the radiation by the uncorrected mask is removed from the test plate, a13) in the test plate, the test structures are produced by etching in different positions,</li><li>a14) measurements are taken for correction on the test structures, degrees of correction are determined from these measurements for correction and, to these degrees of correction, locations are assigned on a wafer plane as a function of the positions of the structures test.</li></ul>

0024Advantageously, the test structures are watch springs for testing.

0025Advantageously, the measurements for correction comprise measurements of the frequencies of oscillators each comprising a balance of predetermined inertia and one of the test watch springs.

0026Advantageously, the windows of the uncorrected mask are dimensionally identical to one another.

0027Advantageously, in sub-step a2), the corrected reusable mask is produced in such a way that, in addition to being dimensioned according to the mapping carried out in sub-step a1), the windows are sized so that l step d), the watch balance springs are produced according to dimensions greater than the dimensions necessary for obtaining the batch of watch balance springs, the elastic torques of which have the average in the predetermined range.

0028A second invention is at least aimed at improving the precision of the manufacture of a batch of watch balance springs, the elastic torques of which have an average in a predetermined range.

0029According to the second invention, this object is achieved by means of a method of manufacturing a batch of watch balance springs, the elastic torques of which have an average in a predetermined range.

0030This manufacturing process according to the second invention comprises steps in which:<ul id="ul0006" list-style="none"><li>d) in a wafer made of a silicon-based material, the watch springs are produced according to dimensions greater than the dimensions necessary for obtaining the batch of watch springs whose elastic torques have the average in the predetermined range,</li><li>w) several or all of the watch springs are subjected to a first oxidation so as to be transformed into superficially oxidized watch springs each in an area where, from the silicon-based material, the first oxidation creates a thermal compensation portion (50 ) to modify the sensitivity to temperature variations of the elastic torque of the corresponding watch balance spring, then</li><li>e) from at least one measurement on at least part of the oxidized watch springs, the amount of silicon-based material to be removed from the oxidized watch springs is determined, in order to obtain the batch of watch balance springs having the elastic torques of which the average is within the predetermined range,</li><li>x) from the thermal compensation portion of several or all of the watch balance springs, a residual thickness is removed which corresponds to the quantity of silicon-based material to be removed, then</li><li>y) several or all of the oxidized watch springs are subjected to a second oxidation such that the thermal compensation portion is regenerated.</li></ul>

0031In the manufacturing method according to the second invention, the measurement (s) used in step e) to determine the quantity of silicon-based material to be removed are carried out on one or more watch balance springs already each provided with a compensation portion. thermal which is predefined.

0032In this way, the elastic torques of the watch springs at the end of the manufacturing method according to the second invention have their average in the predetermined range, while these watch springs are each already provided with the predefined thermal compensation portion.

0033In step x), the residual thickness removed from the thermal compensation portion corresponds to the amount of silicon-based material to be removed and it may not be equal to the thickness of this amount of silicon-based material to remove.

0034Indeed, the residual thickness removed in step x) is made of the oxidized silicon-based material, which constitutes the thermal compensation portion and which may be silicon dioxide in the case where the silicon-based material is silicon. In other words, the residual thickness removed from the thermal compensation portion in step x) is not made of the same material as the amount of silicon-based material to be removed. The residual thickness removed from the thermal compensation portion in step x) is made of the material resulting from the oxidation of the silicon-based material to be removed.

0035In the case where the silicon-based material to be removed is silicon (doped or undoped), the residual thickness removed from the thermal compensation portion in step x) is made of silicon dioxide. In the case where the silicon-based material to be removed is silicon (doped or undoped), the residual thickness removed from the thermal compensation portion in step x) is equal to approximately 227% of the thickness of the amount of silicon-based material to be removed.

0036The residual thickness removed in step x) residual is such that the amount of silicon-based material consumed during a second oxidation regenerating the thermal compensation portion in step y) is equal to the amount of material to silicon base to be removed in step x).

0037Advantageously, in step d), the watch springs are produced by etching.

0038Advantageously, step d) is carried out by deep reactive ionic etching.

0039Advantageously, step w) is carried out collectively on the watch balance springs still attached to the plate.

0040Advantageously, step x) is carried out collectively on the watch balance springs still attached to the plate.

0041Advantageously, step y) is carried out collectively on the watch balance springs still attached to the plate.

0042Advantageously, in step e), the quantity of silicon-based material to be removed is determined from a measurement carried out on a sacrificial sample taken from among the watch balance springs of the wafer and which is dissociated from the wafer.

0043Advantageously, in step e), the quantity of silicon-based material to be removed is determined by means of a correspondence table indicating the thickness of silicon-based material to be removed for each of several average elastic pairs or for each of several quantities linked to such average elastic torques.

0044Advantageously, the quantity of silicon-based material to be removed is a thickness of silicon-based material to be removed from the watch balance springs produced in step d).

0045Advantageously, the silicon-based material is silicon, for example doped silicon.

0046Advantageously, the manufacturing process according to the first invention is a manufacturing process of a batch of watch balance springs according to the second invention.

0047Advantageously, the manufacturing process according to the first invention comprises a step which follows step b) and in which: c) from the photosensitive resin, a sacrificial mask is produced on the wafer by removing from the wafer the photosensitive resin having been irradiated by the radiation in step b) or the photosensitive resin having been subtracted from the radiation in step b) by the corrected reusable mask.

0048Advantageously, step d) is carried out using the sacrificial mask.

0049Advantageously, the manufacturing process according to the first invention comprises a preparatory step which precedes step b) and in which:<ol id="ol0002" compact="compact"><li>a) the corrected reusable mask is produced.</li></ol>

0050Advantageously, the preparatory step comprises sub-steps in which:<ul id="ul0007" list-style="none"><li>a1) a map of the degrees of correction to be applied in different positions on a wafer plane is produced in order to reduce a spatial heterogeneity of an engraving carried out by an engraving device,</li><li>a2) the corrected reusable mask is produced in such a way that the windows are dimensioned according to the mapping carried out in sub-step a1).</li></ul>

0051Advantageously, in the sub-step a1), the following is carried out:<ul id="ul0008" list-style="none"><li>a11) between at least the source and a photosensitive resin carried by one face of a test wafer, an uncorrected mask stops at least part of the radiation from the source except in areas comprising several windows which define the contours of structures test and which are distant from each other within the uncorrected mask,</li><li>a12) the photosensitive resin having been irradiated by the radiation or the photosensitive resin having been subtracted from the radiation by the uncorrected mask is removed from the test wafer,</li><li>a13) in the test plate, the test structures are produced by etching in different positions,</li><li>a14) measurements are made for correction on the test structures, degrees of correction are determined from these measurements for correction and, to these degrees of correction, locations are assigned on a wafer plane as a function of the positions of the structures test.</li></ul>

0052Advantageously, the test structures are watch springs for testing.

0053Advantageously, the measurements for correction comprise measurements of the frequencies of oscillators each comprising a balance of predetermined inertia and one of the test watch springs.

0054Advantageously, the windows of the uncorrected mask are dimensionally identical to one another.

0055Advantageously, in sub-step a2), the corrected reusable mask is produced in such a way that, in addition to being dimensioned according to the mapping carried out in sub-step a1), the windows are sized so that l step d), the watch balance springs are produced according to dimensions greater than the dimensions necessary for obtaining the batch of watch balance springs, the elastic torques of which have the average in the predetermined range.

0056The first invention and the second invention can be combined. Despite this, the first invention and the second invention are independent of each other. Each of these first and second inventions may be the subject of a divisional application.

Brief description of the drawings

0057Other advantages and characteristics will emerge more clearly from the following description of particular embodiments of the first invention given by way of non-limiting example and shown in the accompanying drawings, among which:<ul id="ul0009" list-style="dash"><li>the <figref idref="f0001">figure 1</figref> is a perspective view of one of several watch springs produced in a plate by implementing a manufacturing method according to a first embodiment of the first invention,</li><li>the <figref idref="f0001">figure 2</figref> is a mechanical oscillator comprising the watch balance spring shown on <figref idref="f0001">figure 1</figref>,</li><li>the <figref idref="f0002">figure 3</figref> is a block diagram of a manufacturing process according to a first embodiment of the first invention,</li><li>the <figref idref="f0003">figure 4</figref> is a schematic perspective view illustrating a step in the manufacturing process of the <figref idref="f0002">figure 3</figref>,</li><li>the <figref idref="f0004">figure 5</figref> is a schematic and partial view, in section, and illustrates a step of the manufacturing process of the <figref idref="f0002">figure 3</figref>,</li><li>the <figref idref="f0005">figure 6</figref> is a schematic perspective view illustrating an intermediate state between two stages of the manufacturing process of the <figref idref="f0002">figure 3</figref>,</li><li>the <figref idref="f0006">figure 7</figref> is a schematic perspective view of a wafer and a set of watch balance springs having been cut at the same time by etching all in this wafer, during the implementation of the manufacturing process of the <figref idref="f0002">figure 3</figref>,</li><li>the <figref idref="f0007">figure 8</figref> is a schematic view of a map of dimensional correction degrees to be applied at different locations on a wafer plane to reduce the heterogeneity of an engraving performed by an engraving device,</li><li>the <figref idref="f0007">figure 9</figref> is a double graph where a first curve is the graphical representation (distribution of elastic torques) of the number n of watch springs as a function of the elastic torque M, among a set of watch springs collectively cut by engraving in the same plate without implementing the first invention, and where a second curve is the graphic representation (distribution of elastic torques) of the number n of watch balance springs as a function of the elastic torque M, among a set of watch balance springs collectively cut by engraving in the same plate by means of the manufacturing process of the <figref idref="f0002">figure 3</figref>,</li><li>the <figref idref="f0008">figure 10A</figref> is a cross section, schematic, of an elastically flexible strand of a watch balance spring between two stages of the manufacturing process of the <figref idref="f0002">figure 3</figref>,</li><li>the <figref idref="f0008">figure 10B</figref> is a cross section, schematic, of an elastically flexible strand of a watch balance spring between two stages of the manufacturing process of the <figref idref="f0002">figure 3</figref>,</li><li>the <figref idref="f0008">figure 10C</figref> is a cross section, schematic, of an elastically flexible strand of a watch balance spring between two stages of the manufacturing process of the <figref idref="f0002">figure 3</figref>,</li><li>the <figref idref="f0009">figure 11A</figref> is a triple graph where each of three curves is the graphical representation (distribution of elastic couples) of the number n of watch springs as a function of the elastic torque M, among a set of watch springs produced collectively in the same wafer by incompletely implementing the manufacturing process of the <figref idref="f0002">figure 3</figref>,</li><li>the <figref idref="f0009">figure 11B</figref> is a triple graph where each of three curves is the graphical representation (the distribution of elastic couples) of the number n of watch springs as a function of the elastic torque M, among a set of watch springs produced collectively in the same plate by completely implementing the process of <figref idref="f0002">figure 3</figref>, including steps performing a correction by removing material from the measured watch springs to establish the three elastic torque distributions shown in Figure <figref idref="f0009">figure 11A</figref>, and</li><li>the <figref idref="f0010">figure 12</figref> is a cross section, schematic, of an elastically flexible strand of a completely finished watch balance spring, once all the steps of the manufacturing process of the watch have been implemented. <figref idref="f0002">figure 3</figref>,</li><li>the <figref idref="f0011">figure 13</figref> is a block diagram of a manufacturing process which is according to a second embodiment of the first invention and which comprises a manufacturing process according to a second invention,</li><li>the <figref idref="f0012">figure 14A</figref> is a cross section, schematic, of an elastically flexible strand of a watch balance spring between two stages of the manufacturing process of the <figref idref="f0011">figure 13</figref>,</li><li>the <figref idref="f0012">figure 14B</figref> is a cross section, schematic, of an elastically flexible strand of a watch balance spring between two stages of the manufacturing process of the <figref idref="f0011">figure 13</figref>,</li><li>the <figref idref="f0012">figure 14C</figref> is a cross section, schematic, of an elastically flexible strand of a watch balance spring between two stages of the manufacturing process of the <figref idref="f0011">figure 13</figref>,</li><li>the <figref idref="f0012">figure 14D</figref> is a cross section, schematic, of an elastically flexible strand of a completely finished watch balance spring, once all the steps of the manufacturing process of the watch have been implemented. <figref idref="f0011">figure 13</figref>.</li></ul>

Description of preferred embodiments of the inventions

0058On the <figref idref="f0001">figure 1</figref> is shown a watch balance spring 1, which comprises a ferrule 2 intended to be secured to a pivoting shaft. The watch hairspring 1 also comprises an elastically flexible strand 3 connecting at one end to the ferrule 2 and wound in a spiral so as to form several consecutive turns, the last of which is extended by an attachment segment 4 intended to be attached to a fixed balance bridge (not shown), for example by means of a peg (not shown).

0059On the <figref idref="f0001">figure 2</figref> is shown a mechanical oscillator 6, which is of a type called a "sprung balance" and which comprises the watch spring 1, a balance 8 and a shaft 9 carrying this balance 8 as well as the ferrule 2. In the mechanical oscillator 6 , the watch spring 1 is coupled to the balance 8 by the fact that the ferrule 2 and the balance 8 are both integral with the shaft 9.

0060The watch balance spring 1 is part of a batch of watch balance springs 1 which has been produced in the same wafer by implementing a manufacturing method 10 according to a first embodiment of the first invention, which has many advantages. This manufacturing process 10 comprises several successive steps as shown in<figref idref="f0002">figure 3</figref>.

0061In the manufacturing process 10, a corrected reusable mask 11 is used, visible at the <figref idref="f0003">figure 4</figref>. This corrected reusable mask 11 comprises several windows 12 which define the contours of watch balance springs. In the manufacturing process 10, the windows 12 define the outer contours of watch balance springs. As a variant, the windows 12 could define the internal contours of watch balance springs (depending on the type of photosensitive resin used, see below).

0062In the example shown in <figref idref="f0003">figure 4</figref>, windows 12 are 4 in number, thanks to which this <figref idref="f0003">figure 4</figref> is simplified and more details can be included. Of course, the number of windows 12 can be different from 4. The number of windows 12 in the corrected reusable mask 11 can be several tens. Preferably, the number of windows 12 in the corrected reusable mask 11 is greater than 100. Even more preferably, the number of windows 12 in the corrected reusable mask 11 is several hundred.

0063The corrected reusable mask 11 is used in a photolithography forming part of the manufacturing process 10. Except at the level of the windows 12, the corrected reusable mask 11 forms a screen capable of stopping at least part of a radiation intended to transform a photosensitive resin. . In other words, at least part of such radiation is stopped by the corrected reusable mask 11 except at the level of the windows 12.

0064In the first embodiment of the first invention described here, the windows 12 are the only areas where the corrected reusable mask 11 does not form a screen capable of stopping at least part of a radiation intended to transform a photosensitive resin. However, it could be otherwise without departing from the scope of the first invention. For example, in addition to the windows 12, the corrected reusable mask could comprise one or more zones which define the outlines (internal and / or external) of characterization structures and where the screen capable of stopping at least a part of it is absent. 'radiation intended to transform a photosensitive resin. Such characterization structures may be different from the watch springs 1 and could be used to quantify an engraving by being measured and to determine a level of adjustment to be made by removing material from the watch springs 1, in a manner which will be explained in more detail. far.

0065The corrected reusable mask 11 may comprise a glass plate which is transparent so that it can be crossed by the aforementioned radiation and one of the two main faces of which has a chrome-based coating except in certain areas comprising the windows 12. The coating to chromium base is able to stop all or part of the aforementioned radiation. Each window 12 is an area where this chrome-based coating is absent.

0066When the corrected reusable mask 11 is available, the first step of the manufacturing process 10 is step 100, which is that illustrated on <figref idref="f0003">figure 4</figref>.

0067On the <figref idref="f0003">figure 4</figref>, a wafer 15 has two main faces, one of which is covered with a photosensitive resin 16. As will be explained later, the wafer 15 is a multilayer structure, which is not shown in the figure. <figref idref="f0003">figure 4</figref> for the sake of clarity. In the manufacturing process 10, the wafer 15 is a multi-layered structure, but it could be otherwise. In particular, the wafer 15 could be monolayer and consist of the wafer which, below, is referenced 22.

0068Still on the <figref idref="f0003">figure 4</figref>, the corrected reusable mask 11 is placed between a source emitting radiation R and the assembly consisting of the photosensitive resin 16 and the wafer 15. The radiation R can in particular be radiation comprising UV rays, that is to say rays. ultraviolet rays, or even consist of UV.

0069At least part of the radiation R is able to transform the photosensitive resin 16 so as to make it soluble or insoluble in a suitable chemical bath. In the manufacturing process 10, the photosensitive resin 16 is chosen so that its irradiation with the R radiation makes this photosensitive resin 16 soluble in a suitable chemical bath. In a variant, the photosensitive resin is chosen so that its irradiation with radiation similar or identical to the radiation R renders this photosensitive resin insoluble in a suitable chemical bath, in which case the corrected reusable mask 11 is converted accordingly.

0070In step 100, the corrected reusable mask 11 stops the radiation R except at the level of the windows 12. In this way, the photosensitive resin 16 is irradiated in the areas facing the windows 12. Reference 17 designates the photosensitive resin 16 which was irradiated with R-radiation through windows 12 and thus became soluble in a suitable chemical bath.

0071As can be seen at <figref idref="f0002">figure 3</figref>, a step 101 follows step 100. In step 101, the irradiated photosensitive resin 17 is removed by dissolving it in an appropriate chemical bath, which is called development. The photosensitive resin 16 which has not been irradiated and transformed by the radiation R resists the chemical bath, thus remains in place on the wafer 15 and forms a sacrificial mask, which is referenced 18 on<figref idref="f0004">figure 5</figref>. In other words, the sacrificial mask 18 is produced in step 101.

0072The step referenced 102 in <figref idref="f0002">figure 3</figref> follows step 101. Step 102 is illustrated in <figref idref="f0004">figure 5</figref>, on which it can be seen that the multilayer structure of the wafer 15 comprises a support 20, a stop layer 21 and a plate 22. The plate 22 is made of a material capable of being engraved, for example a silicon-based material. Preferably, the wafer 22 is made of silicon. The silicon of the wafer 22 can be polycrystalline. Preferably, the wafer 22 is made of monocrystalline silicon. The silicon in wafer 22 may not be doped. Preferably, the wafer 22 is made of doped silicon, which is electrically conductive, more dimensionally stable and of better mechanical strength. Preferably, wafer 22 is made of silicon doped with phosphorus. Preferably, the wafer 22 is made of silicon which is of orientation {1,1,1} and / or which is doped. Preferably, the wafer 22 is made of silicon which is oriented <1,1,1> and / or which is doped so as to have a resistivity less than or equal to 0.1 Ω.cm^-1, for example a resistivity equal to 0.05 Ω.cm^-1.

0073Preferably, the wafer 11 is of the silicon-silicon-silicon oxide type, that is to say of the type commonly designated by the acronym “SOI”. This is the case in the manufacturing process 10. In the manufacturing process 10, the support 20 is made of silicon, the barrier layer 21 is made of silicon dioxide, and the wafer 22 is made of silicon.

0074In step 102, several watch springs 1 are cut collectively, together and at the same time, all in the wafer 22, by means of an engraving 25, which is a directional engraving occurring where the wafer 22 is not. protected by the sacrificial mask 18. In the manufacturing process 10, the etching 25 is a deep reactive ionic etching, also called DRIE etching (acronym for “deep reaction ion etching”, which is the English designation for deep reactive ionic etching) or etching according to the Bosch process. Behind the sacrificial mask 18, the plate 22 is not engraved. The stop layer 21 is insensitive to etching 25 which therefore does not affect it. The stop layer 21 therefore has the function of stopping the etching 25.

0075The sacrificial mask 18 is sacrificial in that it is ephemeral and is destroyed once the engraving is completed.

0076The <figref idref="f0005">figure 6</figref> shows the state of wafer 15 after burn 25 is complete. On this<figref idref="f0005">figure 6</figref>, the plate 22 and the watch springs 1 cut therein are still linked to the support 20 by the stop layer 21. The support 20 and the stop layer 21 can be removed locally. In the manufacturing process 10, the support 20 and the stop layer 21 are completely removed, for example by etching, so that the watch balance springs 1 and the plate 22 are released, after which the plate 22 and the watch balance springs 1 cut out are as shown on the<figref idref="f0006">figure 7</figref>.

0077On the <figref idref="f0006">figure 7</figref>, the watch balance springs 1 cut from the plate 22 are still attached to this plate 22, so that they can easily be moved together and processed collectively and at the same time.

0078The corrected reusable mask 11 is corrected in that at least part of the windows 12 are dimensionally different from each other so as to reduce an extent of a dispersion of the elastic torques of the watch balance springs 1 cut by etching in the same wafer 12.

0079It has in fact been discovered that, if the reusable mask is not corrected, that is to say if all its windows similar to the windows 12 are dimensionally identical to one another, the elastic torques of the watch balance springs cut from the same wafer 22 by DRIE etching using a sacrificial mask formed by means of this uncorrected reusable mask have very dispersed values, and the extent of the dispersion of these elastic couples is important. This is believed to be due to imperfections in the etching chamber (non-uniform distribution of plasmas / gases).

0080On the graph of the <figref idref="f0007">figure 9</figref>, the elastic couple M is on the abscissa, while the numbers n of watch springs among all the watch springs produced in the same wafer 22 are on the ordinate. Still on the graph of the<figref idref="f0007">figure 9</figref>, curves C1 and C2 are normally histograms but have been smoothed for clarity. Again on the graph of the<figref idref="f0007">figure 9</figref>, increasing elastic torque classes are also indicated along the x-axis. In the example shown in<figref idref="f0007">figure 9</figref>, the number of classes is equal to 80, but it could be other by means of another segmentation into classes. For example, the number of classes could be equal to 20 as was the practice in the past.

0081On the <figref idref="f0007">figure 9</figref>, the curve C1 is the graphic representation of the dispersion of the elastic pairs of watch balance springs produced in the same wafer 22 using an uncorrected reusable mask.

0082Still on the <figref idref="f0007">figure 9</figref>, the curve C2 is the graphic representation of the dispersion of the elastic torques of the watch balance springs 1 produced in the same wafer 22 by using the corrected reusable mask 11, during an implementation of the manufacturing method 10.

0083Said in a simplified manner, the correction incorporated by the corrected reusable mask 11 makes it possible to switch from the curve C1 to the curve C2.

0084On the <figref idref="f0007">figure 9</figref>, we see that the standard deviation associated with the curve C1 is much higher than the standard deviation associated with the curve C2. In other words, the extent D1 of the dispersion of the elastic torques of the watch springs produced in the same wafer by means of an uncorrected reusable mask is large. The extent D2 of the dispersion of the elastic couples of the watch balance springs 1 produced in the same wafer 22 by means of the corrected reusable mask 11 is markedly smaller than the extent D1.

0085In particular, the extent D2 is small enough so that all or practically all the watch springs 1 produced in the same wafer 22 by means of the corrected reusable mask 11 can be classified in a reduced number of classes and matched with balances distributed in a reduced number of classes.

0086A class of watch balance springs is intended to receive all the watch balance springs having an elastic torque situated between two predefined limits. A class of balances is intended to receive all the balances having an inertia situated between two predefined limits. By pairing any horological hairspring and any balance belonging to corresponding classes, a mechanical oscillator is obtained which is ready to be mounted in a horological movement insofar as the final adjustment of this mechanical oscillator can be carried out by means of racket or by playing on the inertia of the balance, while the mechanical oscillator is in place in the watch movement.

0087The corrected reusable mask 11 is performed during a preparatory step 500. As the corrected reusable mask 11 is reusable and can be used a very large number of times, the preparatory step 500 may be performed only once for all, after which the manufacturing process 10 starting directly with step 100 can be carried out a very large number of times without repeating the preparatory step 500.

0088The corrected reusable mask 11 is suitable for a specific engraving installation, namely the engraving installation with and for which it was developed. This engraving installation may change over time. From the measurements taken on the watch balance springs 1 produced by means of the engraving installation, it is possible to detect a change requiring the design and production of a new corrected reusable mask 11.

0089The preparatory step 500 comprises a sub-step 520, in which a mapping of the degrees of correction to be applied in different positions on a wafer plane is carried out in order to reduce a spatial heterogeneity of an etching carried out by the etching device. Such a mapping 30 is shown on the<figref idref="f0007">figure 8</figref>.

0090Sub-step 520 comprises an operation 522, which is identical to step 100 except that the wafer 22 is replaced by a test wafer and except that an uncorrected mask is used instead of the corrected reusable mask 11. This uncorrected mask is used in photolithography. In this uncorrected mask, there are windows which, like windows 12, are areas where there is no screen capable of stopping at least part of the radiation R. The windows of the uncorrected mask define the outlines (internal and / or external) of test watch balance springs or the outlines (internal and / or external) of other test structures. In the manufacturing process 10, the windows of the uncorrected mask are dimensionally identical to each other and define the contours of the test watch balance springs.

0091During operation 522, the uncorrected mask is located between the source emitting the radiation R in the etching installation considered and a photosensitive resin carried by one face of the test wafer. Still in operation 522, the corrected mask stops at least part of the radiation from the source except at the level of the windows, opposite which the photosensitive resin is irradiated and transformed. In the manufacturing process 10, the photosensitive resin is chosen so that its irradiation with the R radiation makes this photosensitive resin soluble in a suitable chemical bath. In a variant, the photosensitive resin is chosen so that its irradiation with radiation similar or identical to the radiation R renders this photosensitive resin insoluble in a suitable chemical bath, in which case the uncorrected reusable mask is converted accordingly.

0092Still in the sub-step 520, an operation 524 follows the operation 522. The operation 524 is identical to the step 101 except that the photosensitive resin which is then removed is that which was irradiated during the operation. 522 and that the test plate carries. In operation 524, a sacrificial mask is therefore formed from the photosensitive resin.

0093In the sub-step 520, an operation 526 follows the operation 524. The operation 526 is identical to the step 102 except that the cutting by engraving in this operation 526 is a cutting of test watch balance springs and that it is performed by means of the sacrificial mask produced during operation 524. In operation 526, test watch balance springs are cut from the same test wafer, by DRIE etching, in the etching installation, in positions distributed over a wafer plan.

0094In the sub-step 520, an operation 528 follows the operation 526. In the operation 528, measurements are taken for correction on the test watch springs produced in the test wafer during the operation 526. Preferably, these measurements for correction are measurements of the frequencies of oscillators each consisting of a predetermined inertia balance and one of the test watch springs coupled to this predetermined inertia oscillator. Preferably, the horological test hairspring has been detached from the test plate before being coupled to the predetermined inertia balance and before the frequency measurement is carried out. The frequency measurement can also be performed while the watchmaker test hairspring is still attached to the test board. In this case, the frequency measurement can be carried out according to the teachings of the document<patcit id="pcit0005" dnum="EP2423764A"><text>EP 2 423 764</text></patcit>.

0095Still in operation 528, the degrees of correction are determined from the measurements for correction made on the test watch balance springs made in the same test plate.

0096To do this, we use the relation which establishes the elastic torque of a hairspring as a function in particular of its dimensions and which is as follows: <maths id="math0001" num="(1)"><math display="block"><mrow><mi>M</mi><mo>=</mo><mfrac><mrow><mi>E</mi><mn>.</mn><mi>h</mi><mn>.</mn><msup><mi>l</mi><mn>3</mn></msup></mrow><mrow><mn>12.</mn><mi>L</mi></mrow></mfrac></mrow></math><img file="EP3845770A1_D0001.tif" /></maths> or :<ul id="ul0010" list-style="dash" compact="compact"><li>M is the elastic torque of the balance spring,</li><li>E is the modulus of elasticity of the elastically flexible strand of the watch balance spring,</li><li>h is the height of the elastically flexible end of the watch balance spring,</li><li><i>l</i> is the thickness of the elastically flexible strand of the watch balance spring, and</li><li>L is the length of the elastically flexible strand of the watch balance spring.</li></ul>

0097On the <figref idref="f0007">figure 8</figref>, a plane of wafer 35 indicates the position of cells 36, each of which is a region in which to cut a single watch hairspring.

0098Each cell 36 is assigned the degree of correction deduced from the measurement for correction carried out on the test watch hairspring cut at the level of this cell 36.

0099For example, to determine the degrees of correction on the basis of the measurements for correction made on the test watch balance springs made in the same test plate, one can proceed as follows.

0100Relation (1) in a specific case allows to choose h₆, L₆ and <i>l</i>₆ to get the value M₆ intended, due to the fact that: <maths id="math0002"><math display="block"><mrow><msub><mi>M</mi><mn>6</mn></msub><mo>=</mo><mfrac><mrow><mi>E</mi><mn>.</mn><msub><mi>h</mi><mn>6</mn></msub><mn>.</mn><msubsup><mi>l</mi><mn>6</mn><mn>3</mn></msubsup></mrow><mrow><mn>12.</mn><msub><mi>L</mi><mn>6</mn></msub></mrow></mfrac></mrow></math><img file="EP3845770A1_D0002.tif" /></maths> or :<ul id="ul0011" list-style="dash" compact="compact"><li>M₆ is the target elastic torque,</li><li>E is the modulus of elasticity of the elastically flexible strand of the watch balance spring,</li><li>h₆ is the target height of the elastically flexible end of the watch balance spring,</li><li><i>l</i>₆ is the target thickness of the elastically flexible strand of the watch balance spring, and</li><li>L₆ is the target length of the elastically flexible strand of the watch balance spring.</li></ul>

0101The shape of a watch hairspring is such that the error (due to the DRIE engraving) which affects the elastic modulus M much more than the other errors is the error on the width / of the watch hairspring. We can therefore neglect the error on the height h of the hairspring, as well as the error on the length L of the hairspring.

0102From an elastic modulus M₇ obtained by measurement on a horological test hairspring in a cell 36, the correction Δ is determined<i>l</i> to be applied in this cell 36 by solving the following equation: <maths id="math0003"><math display="block"><mrow><msub><mi>M</mi><mn>7</mn></msub><mo>=</mo><mfrac><mrow><mi>E</mi><mn>.</mn><msub><mi>h</mi><mn>6</mn></msub><mn>.</mn><msup><mfenced><msub><mi>l</mi><mn>6</mn></msub><mo>+</mo><mi mathvariant="normal">Δ</mi><mi mathvariant="italic">l</mi></mfenced><mn>3</mn></msup></mrow><mrow><mn>12.</mn><msub><mi>L</mi><mn>6</mn></msub></mrow></mfrac></mrow></math><img file="EP3845770A1_D0003.tif" /></maths>

0103In the case where M₇ measured is greater than M₆ referred to, the width of the window 12 at the level of the corresponding elastically flexible strand 3, having to have substantially the width <i>l</i>₆, will be reduced by the magnitude Δ<i>l</i>.

0104In the case where M₇ measured is less than M₆ referred to, the width of the window 12 at the level of the corresponding elastically flexible strand 3, having to have substantially the width <i>l</i>₆, will be increased by the magnitude Δ<i>l</i>.

0105In the case where M₇ measured is equal to or substantially equal to M₆ referred to, the width of the window 12 at the level of the corresponding elastically flexible strand 3, having to have substantially the width <i>l</i>₆, will not be corrected, i.e. it will not be modified.

0106Each cell 36 can be associated with an individual degree of correction.

0107It is also possible to associate an average identical degree of correction with a family of cells 36 at the level of which the individual degrees of correction obtained are close, for example included in a predefined interval which may for example correspond to a class of clock balance springs. This is the case in manufacturing process 11. On the<figref idref="f0007">figure 8</figref>, the same degree of correction is applied to the cells designated by the letter A. The same degree of correction is applied to the cells designated by the letter B. The same degree of correction is applied to the cells designated by the letter C. The same degree of correction is applied to the cells designated by the letter D.

0108The degree of correction applied at the A-cell level is greater than the degree of correction applied at the B-cell level. The degree of correction applied at the B-cell level is greater than the degree of correction applied at the C-cell level. applied at the C-cell level is greater than the amount of correction applied at the D-cell level.

0109On the <figref idref="f0007">figure 8</figref>, the degree of correction increases overall from the inside to the outside (the A cells are rather at the periphery while the C cells are rather at the center of the wafer plane 35). However, the<figref idref="f0007">figure 8</figref> is only an example and is not limiting. Indeed, the distribution of the degrees of correction on the wafer plane 35 can be any. In other words, the distribution of the degrees of correction on the wafer plane 35 may be very different from that shown in the figure.<figref idref="f0007">figure 8</figref> like to be very similar.

0110Each degree of correction is a modification of at least one dimension of the window 12 of the corrected reusable mask 11, with respect to a window of the uncorrected reusable mask employed in operation 522. In particular, a degree of correction can be expressed by increasing or decreasing the width of the portion of the corrected reusable mask 11 corresponding to an elastically flexible strand 3 of a watch balance spring 1. The change in width may be constant over the entire length of the part corresponding to an elastically flexible strand 3. The increase in width or the decrease in width may also not be constant over the entire length of the part corresponding to an elastically flexible strand. flexible 3.

0111Operations 522, 524, 526 and 528 can be repeated several times so that we have several degrees of correction obtained experimentally for each cell 36 and that we can calculate an average degree of correction for each cell 36 and determine a map 30 on the basis of the averages of degrees of correction each obtained for a cell 36 of the wafer plane 35.

0112In the manufacturing process 10, test watch balance springs are cut from the test wafer, during step 526. However, during step 526, test structures can be cut from the test wafer which are not no watch balance springs. These test structures can be configured to allow a measurement of the modulus of elasticity M and so that the variation of the moduli of elasticity between two plates 22 is proportional to the correction to be made on the watch springs 1. The test structures cut out during operation 526 in the test wafer may also be geometric figures, in which case the measurements for correction made during operation 528 may be dimensional measurements on these geometric figures.

0113The preparatory step 500 comprises a sub-step 540 which follows step 520. In the sub-step 540, the corrected reusable mask 11 is produced in such a way that the windows 12 of this corrected reusable mask 11 are dimensioned according to of the mapping 30 carried out in sub-step 520.

0114In the manufacturing method 10 according to a first embodiment of the first invention, the correction integrated into the corrected reusable mask 11 constitutes a first correction, which aims to reduce the dispersion of the elastic pairs of watch balance springs 1 produced together and at the same time. by etching in the same wafer 22, in an etching installation.

0115The manufacturing method 10 according to a first embodiment of the first invention includes a second correction, which aims to reduce the heterogeneity between two engravings carried out in the same engraving installation, by means of the same corrected reusable mask 11, but to two different times and each on a different plate 22. In other words, the second correction aims to reduce the dispersion between the elastic couples of a first set of watch springs 1 and of a second set of watch springs 1, the watch springs 1 of the first set having been engraved in a first plate, while the watch balance springs 1 of the second set have been engraved in a second plate, before or after the watch balance springs 1 of the first set.

0116Because it includes the second correction, the manufacturing process 10 is a manufacturing process for a batch of watch balance springs 1 whose elastic torques have an average in a predetermined range. In order to be able to carry out the second correction, the corrected reusable mask 11 is produced in such a way that, in addition to being dimensioned according to the mapping 30 carried out in sub-step 520, the windows 12 are dimensioned so that step 102, the watch balance springs 1 are cut by etching according to dimensions greater than the dimensions necessary to obtain the batch of watch balance springs 1, the elastic torques of which have the average in the predetermined range.

0117In addition, still to carry out the second correction, the manufacturing method 10 comprises two steps which follow step 102 and which are a step 110 and a step 120 succeeding this step 110.

0118In step 110, a quantity of material to be removed from the watch balance springs 1 produced in step 102 in the same wafer 22. This quantity of material to be removed is more precisely that which must be removed from these watch balance springs. 1, after step 102, to obtain the batch of watch balance springs 1 having the elastic torques whose average is in the predetermined range.

0119In the case of the manufacturing method 10 according to a first embodiment of the first invention, this quantity of material to be removed is a thickness of material to be removed from the watch balance springs 1 produced in step 102.

0120To perform step 110, the procedure is preferably as follows. We begin by detaching from the plate 22 several watch balance springs 1 constituting a sacrificial sample. This sacrificial sample is qualified as "sacrificial" because the watch balance springs 1 constituting it will be used to perform measurements, after which, for the sake of efficiency and profitability, they may not be kept among the watch balance springs 1 to which is applied. step 120. The watch springs of the sacrificial sample can be 5 to 15 in number and are advantageously chosen at locations distributed over the entire plane of the wafer 35.

0121Once dissociated from the plate 22, each of the watch springs 1 of the sacrificial sample is coupled to a predetermined inertia balance in order to constitute with it a mechanical oscillator whose frequency f is measured, still in step 110 After that, there are as many frequency measurements <i>f</i> that there are horological balance springs 1 in the sacrificial sample. We then calculate the average of these frequencies<i>f</i> measured on the mechanical oscillators comprising the watch balance springs 1 of the sacrificial sample.

0122The average elastic torque of the watch springs of the sacrificial sample can be determined from the frequencies <i>f</i> measured and / or from the average of these frequencies <i>f̅</i> measured, using the following relation: <maths id="math0004" num="(2)"><math display="block"><mrow><mi>f</mi><mo>=</mo><mfrac><mn>1</mn><mrow><mn>2</mn><mi>π</mi></mrow></mfrac><msqrt><mrow><mi>M</mi><mo>/</mo><mi>I</mi></mrow></msqrt></mrow></math><img file="EP3845770A1_D0004.tif" /></maths> or :<ul id="ul0012" list-style="dash" compact="compact"><li>f is the (natural) frequency of the sprung balance,</li><li>l is the moment of inertia of the balance around its axis of rotation, and</li><li>M is the elastic torque of the watch balance spring.</li></ul>

0123Then, to determine the thickness of material to be removed, a correspondence table is used giving a thickness of material to be removed for each of several average elastic pairs.

0124The aforementioned correspondence table was obtained experimentally. To establish this aforementioned correspondence table, one can proceed as follows: first, one determines experimentally by how much a shrinkage of a thickness of one micron over the entire surface of a watch balance spring reduces the class attributable to this watch balance spring 1.

0125Then, using a rule of three, that is to say by proportionality, we calculate the number of microns of thickness to be removed for each of several average elastic torques.

0126As a variant, one can use a correspondence table indicating the thickness of material to be removed for each of several average frequencies. <i>f̅</i>.

0127To determine the thickness of material to be removed, one can also use a mathematical model instead of the correspondence table.

0128According to another variant embodiment in accordance with the first invention, it is possible to determine the thickness of material to be removed from the watch balance springs 1 by calculation from the frequencies <i>f</i> measured and / or from the average <i>f̅</i>, using formula (1) explained above.

0129In addition, according to an alternative embodiment, the frequency measurement can be carried out on watch balance springs 1 which have not been detached from the plate 22. In this case, the frequency measurement can be carried out according to the teachings of the document. <patcit id="pcit0006" dnum="EP2423764A"><text>EP 2 423 764</text></patcit>.

0130In addition, the thickness of material to be removed from the watch balance springs 1 produced in step 102 can be determined from a measurement of a magnitude other than the frequency of an oscillator consisting of a watch balance spring 1 and d. 'a balance of predetermined inertia.

0131For example, the thickness of material to be removed from the watch balance springs 1 produced in step 102 can be determined from dimensional measurements of portions cut by etching in the wafer 22, these portions cut by etching by the wafer 22 possibly being portions of the watch balance springs 1 or geometric figures cut from the plate in addition to the watch balance springs 1. The quantity of material to be removed from the watch balance springs 1 produced in step 102 can also be determined from a stiffness measurement carried out on qualification structures which have been cut from the wafer 22 at the same time as the watch balance springs 1. .

0132On the <figref idref="f0008">figure 10A</figref> is shown the cross section of an elastically flexible strand 3 of a watch balance spring 1 which has just been cut by etching in step 102 but which has not yet undergone oxidation.

0133Whatever the method used for determining the thickness of material to be removed e from the watch balance springs 1 at the end of step 102, it can advantageously be taken into account for this determination that the height <i>h₁</i> of the elastically flexible strand 3 of a watch balance spring 1 at the end of step 102 is known with a fairly high precision since it is equal to the thickness of the plate 22. In this way, the inaccuracy of a watch balance spring 1 at the end of step 102 can be considered as being mainly due to steps 100, 101 and 102, and mainly concern the width <i>l₁</i> of the elastically flexible strand 3 of this watch balance spring 1.

0134In step 120 in the case of the manufacturing method 10 according to a first embodiment of the first invention, the thickness of material to be removed e determined in step 110 is removed on the watch balance springs 1 still attached to the plate. 22 and, for the sake of economy, it may not be removed from the watch balance springs 1 belonging to the sacrificial sample.

0135Step 120 can be performed by following the teachings of the document <patcit id="pcit0007" dnum="WO2015113973A1"><text>WO 2015/113973 A1</text></patcit>. More precisely, in the case of the manufacturing method 10 according to a first embodiment of the first invention, the watch balance springs 1 still attached to the plate 22 are collectively subjected, at the same time, to step 120. In step 120 in the case of the manufacturing method 10 according to a first embodiment of the first invention, the thickness of material to be removed e determined in step 110 is removed over the entire surface of the elastically flexible strand 3 of each of the watch balance springs 1 produced in step 102 and remained attached to the plate 22.

0136In the case of the manufacturing method 10 according to a first embodiment of the first invention, step 120 comprises a sub-step 122 and a sub-step 124 succeeding step 122. As a reminder, in the case of the method manufacturing 10 according to a first embodiment of the first invention, the watch balance springs 1 after step 102 are made of silicon, which is the silicon of the wafer 22.

0137In the sub-step 122, at least part of the watch springs 1 produced in the wafer 22 in step 102 are subjected to an oxidation such that the thickness of silicon transformed into silicon oxide is equal to the thickness of material to be removed e determined in step 110. The oxidation can in particular be a thermal oxidation carried out between 800 ° C. and 1200 ° C. under an oxidizing atmosphere, in an oven where silicon dioxide is formed from the silicon. The thermal oxidation can be a wet oxidation obtained in the presence of water vapor in the furnace or else a dry oxidation obtained in the presence of oxygen in the furnace. The oxidation time can be determined by appropriate formulas well known to those skilled in the art or it can be determined from calculators available online, for example at the following address: http://lelandstanfordjunior.com/thermaloxide .html. The thickness of silicon oxide (silicon dioxide) formed varies slightly with the crystallographic orientation of the silicon, but the differences are negligible. The thickness of material to be removed e typically corresponds to approximately 44% of the thickness (referenced e 'on the<figref idref="f0008">figure 10B</figref> mentioned later) of silicon oxide formed.

0138In the manufacturing process 10, the watch balance springs 1 which remain attached to the plate 22 are subjected together and collectively to oxidation, while they are still attached to the plate 22. The watch balance springs 1 of the sacrificial sample cannot be subjected to oxidation. not be subjected to step 120 and the following steps, for the sake of efficiency and cost-effectiveness. In the variant where none of the watch balance springs 1 has been detached from the plate 22, all of the watch balance springs 1 produced in step 102 in the plate 22 are subjected together and collectively to oxidation while they are still attached to the plate 22.

0139The elastically flexible strand 3 shown in <figref idref="f0008">figure 10B</figref> is the same as that shown on the <figref idref="f0008">figure 10A</figref>. On this<figref idref="f0008">figure 10B</figref>, this elastically flexible strand 3 is superficially oxidized, after having undergone the sub-step 122. On the <figref idref="f0008">figure 10B</figref>, the reference 40 designates the unoxidized silicon and the reference 41 designates the silicon dioxide which is formed during the sub-step 122. The thickness e 'of the layer made of the silicon dioxide 41 resulting from the oxidation at the sub-step 122 is greater than the thickness e of the unoxidized silicon 40, the oxidation of which has led to this silicon dioxide 41.

0140In sub-step 124, the silicon dioxide 41 is removed from at least part of the watch balance springs 1 cut from the wafer 22 during step 102. The silicon dioxide can be removed by etching, or by chemical etching. , for example with hydrofluoric acid.

0141In the manufacturing process 10, the watch balance springs 1 remaining attached to the plate 22 are subjected together and collectively to the sub-step 124. In the case where none of the watch balance springs 1 made in the plate 22 is detached from this plate 22 in step 110, the set of watch springs 1 cut by etching in step 102 are submitted together and collectively to sub-step 124.

0142The <figref idref="f0008">figure 10C</figref> represents the same elastically flexible strand 3 as the <figref idref="f0008">figures 10A and 10B</figref>. On the<figref idref="f0008">figure 10C</figref>, this elastically flexible strand 3 is freed from silicon dioxide 41, after having undergone substep 124. The elastically flexible strand 3 of the <figref idref="f0008">figure 10C</figref> at a height <i>h₂</i> less than height <i>h₁</i> and a width <i>l</i>₂ less than width <i>l₁</i>.

0143Step 120 can include substeps 122 and 124 also when wafer 22 is not made of silicon. In particular, step 120 can include substeps 122 and 124 when wafer 22 is made of a silicon-based material.

0144Step 120 can be carried out in a manner other than that described above. The reduction in section of the elastically flexible strand 3 of the watch springs 1 can be achieved by means of a plasma, and this is while the watch springs 1 are in the state of the.<figref idref="f0005">figure 6</figref>, that is to say not yet released from the support 20 and from the stop layer 21, either while the watch balance springs 1 or part of the watch balance springs 1 are in the state shown in figure <figref idref="f0006">figure 7</figref>, that is to say freed from the stop layer 21 and from the support 20, but still attached to the wafer 22. The section reduction by means of a plasma can be carried out even when the sacrificial mask 18 is still present on one side of the watch balance springs 1, in which case only the lateral sides of the watch balance springs 1 are cut.

0145Variations at step 120 are possible. It is possible, for example, to oxidize the hairspring with an oxide thickness w before step 110. This oxide thickness w corresponds to the final thermocompensation thickness or to a chosen thickness. The balance spring is measured after oxidation. The elastic torque is reduced by partially or totally deoxidizing the hairspring and reoxidizing it to the value w.

0146In this regard, the <figref idref="f0011">figure 13</figref> represents the successive steps of a manufacturing process 10a according to a second embodiment of the first invention. The manufacturing process 10a is a manufacturing process for a batch of watch balance springs 1. The manufacturing process 10a comprises a manufacturing process 1000 according to an embodiment of the aforementioned second invention. In the example shown in<figref idref="f0011">figure 13</figref>, the manufacturing process 1000 is part of the manufacturing process 10a. However, it may be otherwise. In other words, the manufacturing process 1000 can be implemented without implementing all or part of the other steps of the manufacturing process 10a.

0147The manufacturing process 1000 is a manufacturing process for a batch of watch balance springs 1, the elastic torques of which have an average in a predetermined range. The manufacturing method 1000 comprises steps 150, 152 and 153, in addition to steps 102 and 110 already described above.

0148Step 150 follows step 102, at the end of which the elastically flexible strand 3 of each of the watch balance springs 1 cut from the plate 22 is as shown in Figure <figref idref="f0012">figure 14A</figref>.

0149At the end of step 150, the elastically flexible strand 3 of each of the watch balance springs 1 cut from the plate 22 is as shown in Figure <figref idref="f0012">figure 14B</figref>. In step 150, several or all of the watch springs 1 are subjected to a first oxidation so as to be transformed into superficially oxidized watch springs each in an area where, from the silicon-based material, the first oxidation creates a portion thermal compensation 50'a provided to modify the sensitivity to temperature variations of the elastic torque of the corresponding watch balance spring 1. On the<figref idref="f0012">figure 14B</figref>, the reference w designates the thickness of the thermal compensation portion 50'a.

0150Step 150 is followed by step 110, in which, from at least one measurement on at least part of the oxidized watch springs 1, the quantity of silicon-based material (referenced w 'on the <figref idref="f0012">figure 14C</figref>) removing the oxidized watch springs 1, to obtain the batch of watch springs 1 having the elastic torques, the average of which is in the predetermined range. Examples of measures that can be employed in step 110 have been described above.

0151Step 152 follows step 110. The resiliently flexible strand 3 of each of the watch balance springs 1 cut from the plate 22 is as shown in FIG. <figref idref="f0012">figure 14C</figref> at the end of step 152, in which, from the thermal compensation portion 50'a of several or all of the watch balance springs 1, a residual thickness which corresponds to the quantity of silicon-based material to be removed is removed. remove w '. In step 152, the residual thickness removed from the thermal compensation portion 50'a corresponds to the amount of silicon-based material to be removed w 'and it may not be equal to the thickness of this amount of material. silicon-based to remove w '.

0152In fact, the residual thickness removed in step 152 is made of the material based on oxidized silicon, which constitutes the thermal compensation portion 50'a and which may be silicon dioxide 41 in the case where the material based on silicon is silicon. In other words, the residual thickness removed from the thermal compensation portion 50'a in step 152 is not made of the same material as the amount of silicon-based material to be removed w '. The residual thickness removed from the thermal compensation portion in step 152 is made of the material resulting from the oxidation of the silicon-based material to be removed w '.

0153In the case where the silicon-based material to be removed is silicon (doped or undoped), the residual thickness removed from the thermal compensation portion 50'a in step 152 is made of silicon dioxide 41. In the case where the silicon-based material to be removed is silicon 40 (doped or undoped), the residual thickness removed from the thermal compensation portion 50'a in step 152 is equal to approximately 227% of l 'thickness of the amount of silicon-based material to be removed w'.

0154The residual thickness removed in residual step 152 is such that the amount of silicon-based material consumed during a second oxidation regenerating the thermal compensation portion in step 153 is equal to the amount of silicon-based material. silicon to be removed in step 152.

0155In the case where the thermal compensation portion 50'a is made of silicon dioxide 41, the residual thickness which corresponds to the amount of silicon-based material to be removed w 'can be removed from the thermal compensation portion 50'. a by means of chemical attack, for example by using hydrochloric acid. The duration of this chemical attack is a controlled duration, determined beforehand as a function of the quantity of silicon-based material to be removed w '.

0156Step 153 follows step 152. The elastically flexible strand 3 of each of the watch balance springs 1 cut from the wafer 22 is as shown in FIG. <figref idref="f0012">figure 14D</figref> at the end of step 153, in which several or all of the oxidized watch springs 1 are subjected to a second oxidation such that the thermal compensation portion is regenerated. On the<figref idref="f0012">figure 14D</figref>, the reference w designates the thickness of the thermal compensation portion 50a as it is after having been regenerated.

0157On the <figref idref="f0012">figures, 14A, 14B, 14C and 14D</figref>, the reference 40 designates the silicon-based material, which can in particular be doped or undoped silicon, which is the case in the manufacturing method 1000 according to an embodiment of the second invention. On the<figref idref="f0012">figures, 14A, 14B, 14C and 14D</figref>, the reference 41 designates the material based on oxidized silicon, which can in particular be silicon dioxide, which is the case in the manufacturing process 1000 according to an embodiment of the second invention.

0158The manufacturing method 10a according to a second embodiment of the first invention differs from method 10 in that it comprises the succession of steps 150, 110, 152 and 153, instead of the succession of steps 110, 120 and 130. For the rest, the manufacturing process 10a can be identical to the manufacturing process 10, which is the case in the second embodiment of the first invention.

0159In the case where it is determined in step 110 that there is no silicon-based material to remove, steps 152 and 153 are not carried out.

0160Preferably, step 150 is carried out collectively on the watch balance springs 1 still attached to the plate 22.

0161Preferably, step 152 is carried out collectively on the watch balance springs 1 still attached to the plate 22.

0162Preferably, step 153 is carried out collectively on the watch balance springs 1 still attached to the plate 22.

0163Preferably, the amount of silicon-based material w ′ to be removed is determined from a measurement carried out on a sacrificial sample taken from among the watch springs 1 of the wafer 22 and which is dissociated from the wafer 22.

0164Preferably, in step 110, the quantity of silicon-based material to be removed w 'is determined by means of a correspondence table indicating the thickness of silicon-based material to be removed w' for each of several pairs. average elastic or for each of several quantities linked to such average elastic torques.

0165On the graph of the <figref idref="f0009">figure 11A</figref> and on the graph of the <figref idref="f0009">figure 11B</figref>, the elastic couple M is on the abscissa, while the number n of watch balance springs 1 among all the watch balance springs 1 produced in the same wafer 22 in step 102 during an implementation of the manufacturing method 10 are on the ordinate .

0166On the graph of the <figref idref="f0009">figure 11A</figref>, C3, C4 and C5 curves are normally histograms but have been smoothed for clarity. On the graph of<figref idref="f0009">figure 11B</figref>, curves C'3, C'4 and C'5 are normally histograms but have been smoothed for clarity.

0167On the graphs of <figref idref="f0009">figures 11A and 11B</figref>, increasing elastic torque classes are also indicated along the x-axis. In the example shown on<figref idref="f0009">figures 11A and 11B</figref>, the number of classes is equal to 80, but it could be other by means of another segmentation into classes. For example, the number of classes could be equal to 20 as was the practice in the past.

0168The C3 curve on the <figref idref="f0009">figure 11A</figref> and the curve C'3 on the <figref idref="f0009">figure 11B</figref> are each a graphic representation of the dispersion of the elastic pairs of watch balance springs 1 produced in the same first wafer 22, during a first implementation of the manufacturing process 10 or 10a in which the corrected reusable mask 11 is used. The measurements used for the curve C3 were produced on the watch balance springs 1 before these watch balance springs 1 were subjected to steps 110 and 120, or to steps 110, 152 and 153. The measurements used for the curve C'3 were carried out on the watch balance springs 1 after these watch balance springs 1 have been subjected to steps 110 and 120, or to steps 110, 152 and 153.

0169The curve C4 on the <figref idref="f0009">figure 11A</figref> and the curve C'4 on the <figref idref="f0009">figure 11B</figref> are each a graphic representation of the dispersion of the elastic pairs of watch balance springs 1 produced in the same second plate 22, during a second implementation of the manufacturing method 10 or 10a in which the same corrected reusable mask 11 is used as in the first implementation of the manufacturing method 10 or 10a. The measurements used for the curve C4 were carried out on the watch balance springs 1 before these watch balance springs 1 were subjected to steps 110 and 120, or to steps 110, 152 and 153. The measurements used for the curve C'4 were produced on the watch balance springs 1 after these watch balance springs 1 have been subjected to steps 110 and 120, or to steps 110, 152 and 153.

0170Curve C5 on the <figref idref="f0009">figure 11A</figref> and the curve C'5 on the <figref idref="f0009">figure 11B</figref> are each a graphic representation of the dispersion of the elastic pairs of watch balance springs 1 produced in the same third plate 22, during a third implementation of the manufacturing method 10 or 10a in which the same corrected reusable mask 11 is used as in the first and second implementation of the manufacturing method 10 or 10a. The measurements used for the curve C5 were carried out on the watch balance springs 1 before these watch balance springs 1 were subjected to steps 110 and 120, or to steps 110, 152 and 153. The measurements used for the curve C'5 were produced on the watch balance springs 1 after these watch balance springs 1 have been subjected to steps 110 and 120, or to steps 110, 152 and 153.

0171The measurements used for curve C3, curve C4 and curve C5 on the <figref idref="f0009">figure 11A</figref> were carried out on watch balance springs 1 having benefited from the first correction incorporated in the corrected reusable mask 11. The measurements used for curve C'3, curve C'4 and curve C'5 on the <figref idref="f0009">figure 11B</figref> were carried out on watch balance springs 1 having benefited both from the first correction incorporated into the corrected reusable mask 11 and from the second correction resulting from steps 110 and 120 or from steps 110, 152 and 153. Consequently, a comparison of the <figref idref="f0009">figures 11A and 11B</figref> makes it easy to understand the effect of the second correction resulting from steps 110 and 120, or steps 110, 152 and 153.

0172The watch springs 1 having been produced in the first plate 22 and having been subjected to steps 110 and 120, or to steps 110, 152 and 153, that is to say the watch springs 1 on which the measurements used were carried out. to produce the curve C'3, constitute a first batch of watch springs 1 whose elastic torques have an average (denoted M₃ on the <figref idref="f0009">figure 11B</figref>) within a predetermined range, which is the predetermined range P over the <figref idref="f0009">figure 11B</figref>.

0173The watch springs 1 having been produced in the second plate 22 and having been subjected to steps 110 and 120, or to steps 110, 152 and 153, that is to say the watch springs 1 on which the measurements used were carried out. to produce the curve C'4, constitute a second batch of watch springs 1 whose elastic torques have an average (denoted M₄ on the <figref idref="f0009">figure 11B</figref>) within the predetermined range P. As can be seen from the <figref idref="f0009">figure 11A</figref>, before having been subjected to steps 110 and 120, or to steps 110, 152 and 153, the watch springs 1 having been produced in the second plate 22, that is to say the watch springs 1 on which have been carried out the measurements used to carry out the curve C4, did not constitute a batch of watch balance springs 1 whose elastic torques have an average in the predetermined range P.

0174The watch springs 1 having been produced in the third plate 22 and having been subjected to steps 110 and 120, or to steps 110, 152 and 153, that is to say the watch springs 1 on which the measurements used were carried out. to produce the curve C'5, constitute a third batch of watch balance springs 1, the elastic torques of which have an average (denoted Ms on the <figref idref="f0009">figure 11B</figref>) within the predetermined range P. As can be seen from the <figref idref="f0009">figure 11A</figref>, before having been subjected to steps 110 and 120, or to steps 110, 152 and 153, the watch springs 1 having been produced in the third plate 22, that is to say the watch springs 1 on which have been carried out the measurements used to carry out the curve C5, did not constitute a batch of watch balance springs 1 whose elastic torques have an average in the predetermined range P.

0175As can be seen on the <figref idref="f0009">figure 11B</figref>, the watch springs 1 having been produced in the first, second and third plates 22 and having been subjected to steps 110 and 120, or to steps 110, 152 and 153, that is to say the watch springs 1 on which have The measurements used to make the curves C'3, C'4 and C'5 have been taken, each of which can be classified in one of classes 0 to 40. Now, when a watch spring is classified in any one of classes 1 to 40, it can be paired with a balance belonging to one of classes 1 to 40, which constitute a reduced number of classes.

0176Steps 110 and 120 can be repeated one or more times, until the average of the elastic torques of the watch springs 1 produced in the same wafer 22 is within the predetermined range.

0177Steps 110, 152 and 153 can be repeated one or more times, until the average of the elastic torques of the watch springs 1 produced in the same wafer 22 is within the predetermined range.

0178As can be seen on the <figref idref="f0002">figure 3</figref>, step 120 can be followed by a step 130 in which, on each of several watch springs 1 constituting at least part of the watch springs 1 produced in the wafer 22 in step 102, a thermal compensation portion is formed . On the<figref idref="f0010">figure 12</figref>, such a thermal compensation portion 50 has the form of a peripheral layer covering the elastically flexible strand 3 of a watch balance spring 1.

0179The thermal compensation portion 50 reduces, cancels or reverses the sensitivity to temperature variations of the elastic torque of the watch balance spring 1. In the case where there are no steps 110 and 120, step 130 can be carried out directly afterwards. step 102.

0180In step 130 in the case of the manufacturing method 10 according to a first embodiment of the first invention, the thermal compensation portion 50 is formed on the watch balance springs 1 still attached to the wafer 22 and, for the sake of economy, it may not be formed on the watch balance springs 1 belonging to the sacrificial sample. In the case of the manufacturing method 10 according to a first embodiment of the first invention, the watch balance springs 1 still attached to the plate 22 are subjected collectively, together and at the same time, to step 130. In step 130 in the case of the manufacturing method 10 according to a first embodiment of the first invention, the thermal compensation portion is formed over the entire length and the entire surface of the elastically flexible strand 3 of each of the watch balance springs 1. carried out in step 102 and remained attached to the wafer 22.

0181As a reminder, in the case of the manufacturing method 10 according to a first embodiment of the first invention, the watch balance springs 1 after step 102 are made of silicon, which is the silicon of the wafer 22.

0182In step 130, at least part of the watch balance springs 1 produced in wafer 22 in step 102 are subjected to oxidation so as to be superficially oxidized to a thickness which is a function of the degree of thermal compensation desired. The oxidation can in particular be a thermal oxidation carried out between 800 ° C. and 1200 ° C. under an oxidizing atmosphere, in a furnace where silicon dioxide is formed from the silicon. The thermal oxidation of step 130 can be a wet oxidation obtained in the presence of water vapor in the furnace or else a dry oxidation obtained in the presence of oxygen in the furnace.

0183In the manufacturing process 10, the watch balance springs 1 remained attached to the plate 22 are subjected together and collectively to the oxidation of step 130, while they are still attached to the plate 22. The watch balance springs 1 of the The sacrificial sample may not be submitted to step 130, for the sake of efficiency and cost-effectiveness. In the variant where none of the watch balance springs 1 has been detached from the plate 22, all of the watch balance springs 1 produced in step 102 in the plate 22 are subjected together and collectively to the oxidation of step 130 while they are still attached to the plate 22.

0184The thermal coefficient of the Young's modulus of a material is also called the thermoelastic coefficient of this material. The<figref idref="f0010">figure 12</figref> is a cross section of an elastically flexible strand 3 of a watch balance spring 1 as it is after step 130. On the <figref idref="f0010">figure 12</figref>, the elastically flexible strand 3 comprises a silicon core 40 and the thermal compensation portion 50 which coats this silicon core 40. The silicon 40 of the core has a thermoelastic coefficient which is negative. The silicon dioxide of the thermal compensation portion 50 has a thermoelastic coefficient which is positive. This is why the thermal compensation portion 50 reduces or cancels or even reverses the sensitivity to temperature variations of the elastic torque of the watch balance spring 1.

0185After having been detached from the plate 22, the watch balance springs 1 are classified, for example each by means of an omega-metric type device (trademark protected) in a step 200, which follows step 130, or step 120 , or else step 110. After having been classified in step 200, the watch springs 1 coming from the same plate 22 are paired with balances 8, in a step 250. In this step 250, the watch springs 1 coming from the same plate 22 are paired with balances 8 chosen from those of a batch that was obtained in a step 210 and which consists of classified balances 8.