Method and device for measuring interfaces of an element

Abstract

The invention relates to a method (100) for measuring a composite moulding layer of an element to be measured based on signals measured by an interferometry system, the element to be measured being a semiconductor object comprising one or more chips and at least one composite moulding layer of at least some of the chips, the method being characterised in that it comprises the following steps: a) receiving (102) a plurality of measured peaks relating to beams reflected by the element to be measured when it is illuminated by an optical beam emitted by the interferometry system, b) receiving (104) a model comprising a plurality of modelled positions, each modelled position corresponding to one interface of the element to be measured encountered by the optical beam, c) automatically associating (106), among the measured peaks, so-called useful peaks, with corresponding modelled positions.

Description

DESCRIPTION TITLE: Method and device for measuring interfaces of an element Technical field.

[0001] The present invention relates to a method for measuring the interfaces of an element. It also relates to a measuring device implementing such a method. Prior art

[0002] When manufacturing elements to be measured or characterized, comprising one or more chips and one or more layers of molding composite of at least one of said chips, it may be necessary to control or measure the thicknesses of said layers or the positions of constituent elements, or even the spaces between constituent elements, along a measurement axis, during the molding process and possibly after the molding process.

[0003] One method known for this is the use of low-coherence interferometry techniques. A measurement optical beam from a broad-spectrum optical source is propagated through the surfaces of the chip molding composite layers. The beam reflections from these surfaces are collected and analyzed by interfering with each other and / or with a reference beam to determine the differences in optical path lengths between interfering beams, and from this, the positions and / or distances between corresponding surfaces or interfaces can be deduced. This allows, for example, the determination of the thicknesses of the chip molding composite layers. The interferences result in peaks that include spurious peaks, called ghost peaks, which are inserted between the interface peaks, corresponding to a surface of the element being measured.Indeed, the reflection signals on the interfaces to be observed are likely to be reflected several times between the interfaces, causing additional returns, which are a kind of echo, producing the ghost peaks in the measured signal to be analyzed.

[0004] It is possible to manually select the so-called useful peaks, that is, via a user-controlled software interface. However, this solution is time-consuming, unreliable, and difficult to implement.

[0005] The aim of the present invention is to resolve at least one of the aforementioned drawbacks. Description of the invention

[0006] At least one of these goals is achieved with a method for measuring the characteristics of a molding composite layer of an element to be measured from signals measured by an interferometry system, said element to be measured being a semiconductor object comprising one or more chips and at least one molding composite layer of at least a part of said chips, the method is characterized in that it comprises the following steps: a) receive a plurality of measured peaks relating to beams reflected by the element to be measured when it is illuminated by an optical beam emitted by the interferometry system, b) receive a model comprising a plurality of modeled positions, each modeled position corresponding to an interface of the element to be measured encountered by the optical beam, c) automatically associate, among the measured peaks, so-called useful peaks, with corresponding modeled positions, d) determine one or more thicknesses of the element to be measured from the so-called useful peaks associated with these modeled positions.

[0007] The process allows for the automatic detection of useful peaks and thus distinguishes spurious peaks, due to noise and measurement echoes, from peaks corresponding to an interface measurement.

[0008] According to one embodiment, step c) may include the following steps: cl) extract, for each measured peak, a measured position, c2) determine distances between each measured position and each modeled position, c3) for each of the distances, determine a deviation score, said deviation score decreasing as the distance increases in absolute value, c4) select useful peaks as corresponding to the measured positions exhibiting the highest score in relation to the corresponding modeled position.

[0009] Advantageously, the distance can be, for example, the absolute value of the difference or the gap between two positions.

[0010] According to a particular embodiment, the process may include a step to add a tolerance to each measured position, or to each modeled position.

[0011] Advantageously, step c2) can be performed using the following formula: D[c2(i,j) ] = pos_meas[i] — pos_Mdl[j] with pos_meas[ï] being the measured position of index i between 1 and I the number of measured positions, pos_Mdl[j] being the modeled position of index j between 1 and J the number of positions of the provided model and [c2(i,y)] being the distance for the candidate pair (i,y).

[0012] According to another embodiment, step c2) can be carried out using the following formula: D[c2(i, j, k)] = pos_meas[i] — pos_Mdl[j] — e[k] * TOL[j] with pos_meas[i] being the measured position with index i between 1 and I the number of measured positions, pos_Mdl[j] being the modeled position with index j between 1 and J the number of positions of the provided model, D[c2(i,j)] being the distance for the candidate pair (i,j), TOL[j] being the tolerance of the modeled position and e[k] being a weighting factor, and k a weighting index between 1 and K.

[0013] Thus, it is possible to take into account the errors related to the measurement process of the measured peaks. For example, the tolerance sequence e[k] can vary as follows: for example e[1] = -1 ; e[2]= 0 ; e[3] = +1, or e[1] = -1.2 ; e[2]= -0.6 ; e[3] = 0 ; e[4]= +0.6 ; e[5] = +1.2.

[0014] Tolerances on modeled and / or measured positions may depend on tolerances on the thicknesses of the model layers.

[0015] The process may further include, following step c2), a step to reorder the distances determined in step c2). In particular, the process then generates an ordered, for example, ascending, list of distances D[c2(i,j)] or distances D[c2(i,j,k)], which are renumbered as D[c], with indices c from 1 to C. The number C may correspond to the maximum number of different possible offsets between the measured and modeled positions in order to align the model peaks with the measured peaks. Preferably, for each c, the indices i[c], j[c], (k[c]) at the origin of index c may be stored.

[0016] According to one embodiment, the deviation score can be determined at step c3) by the following formula:SC(c) = P(pos_meas[ic] - pos_Mdl[jc] - D[c]), for (ic,jc) in 1, I x 1, J with SC(c) being the deviation score of index c corresponding to this index c2(ic,jc) after its ordering, and D[c] the distance determined at step c2) and P being a function which has a maximum of 1 when its argument tends towards 0 and a minimum of 0 when its argument is greater than a given threshold, the function P varying inversely to the absolute value of its argument.

[0017] For example, the deviation score can range from 0 to 1. Thus, the deviation score can be at most 1 and at least 0; and can be at its maximum when the value of pos_meas[ic] - pos_Mdl[jc] - D[c] is equal to 0. The deviation score can decrease when the value of pos_meas[ic] - pos_Mdl[jc] - D[c] is close to 0. Finally, the deviation score can be 0 when pos_meas[ic] - pos_Mdl[jc] - D[c] is large or very large, for example, greater than a threshold. In particular, the function P can be a Gaussian function.

[0018] Furthermore, the P-function can depend on other parameters. In particular, the P-function can depend on the amplitudes of the measured peaks. In this case, this allows the deviation score to be higher when the amplitude of the measured peak is greater, which helps to choose a useful peak over another potentially similar one that is due to measurement aberrations such as a ghost or spurious peak. This allows the selection of the most probable peak to associate with the model.

[0019] According to one embodiment, step c4) may include the following substeps: c4-l) compare the gap scores and select one or more indices c0 giving the highest gap score(s), or greater than a threshold; and for each index c0 select the indices cl neighboring or close to c0 whose associated indices il and jl correspond to associations between measured peak and modeled peak complementary to i0 and j0, increasing the number of associated pairs (i, j), c4-2) determine a relevance criterion for each index c0 selected in the previous step, such as the number of pairs selected and an overall score on the selected pairs, and c4-3) select the useful peaks as the measured peaks associated with the index(es) c0 for which the relevance criterion is satisfied.

[0020] These sub-steps allow us to consider cases where an optimized number of modeled peaks are matched with the measured peaks.

[0021] In particular, step d) may include the following steps: dl) determining so-called optical distances calculated as the differences in the positions of each of the previously selected useful peaks, d2) divide optical distances by optical refractive indices in order to obtain so-called physical thicknesses.

[0022] Preferably, the model in step b) can correspond to a model of the thicknesses of the layers to be measured. For example, the model can be decomposed into corresponding interface positions.

[0023] In this document, "item to be measured" means a sample, such as a wafer or semiconductor object or panel, comprising one or more chips and at least one layer of molding composite of at least one of said chips, deposited on said sample.

[0024] The molding composite can be of any shape / composition. Following a non-limiting example, the molding composite can, for example, be in the form of an epoxy filled with thermal expansion control elements, such as silica beads.

[0025] According to another aspect of the invention, a device is proposed for measuring a demolding composite layer of an element to be measured from signals measured by interferometry, said element to be measured being a semiconductor object comprising one or more chips and at least one molding composite layer of at least a part of said chips, the device comprising: a measurement channel configured to produce a measurement beam at a predetermined position relative to the field of view of the element to be measured, and comprising an optical sensor configured to acquire signals reflected by the element, a processing module configured to process the signals acquired by the optical sensor, the processing module being configured to implement the process as described above.

[0026] The present invention is used for the inspection of a composite layer of chip molding of a semiconductor object, or sample.

[0027] In particular, the invention makes it possible to measure the thickness of said composite layer, during the molding of chip(s) of the sample with the molding composite, that is to say after deposition of said molding composite and / or during the thinning of the composite layer.

[0028] In particular, the invention makes it possible to control said molding composite layer after the molding process is completed, during at least one measurement operation applied to said sample.

[0029] Description of the figures and methods of realization

[0030] Other advantages and features of the invention will become apparent upon reading the detailed description of implementations and embodiments, which are by no means limiting, and the following attached drawings: [Fig.1] illustrates a schematic representation of an example of a measured signal used in an example of the method according to the invention. [Fig. 2] illustrates an example of a modeled peak pattern. [Fig. 3] illustrates an example of an embodiment of the method according to the invention, [Fig. 4] illustrates an example of a gap score determination function implemented by the method of Figure 3. [Fig. 5] Figures 5a-5d are schematic representations of an example of a chip molding process of a sample, or of a semiconductor device, with a molding composite, during which the thickness of the molding composite layer can be measured according to the invention.

[0031] These embodiments are not exhaustive; in particular, variants of the invention may be considered that comprise only a selection of features described or illustrated hereafter, isolated from the other described or illustrated features (even if this selection is isolated within a sentence including these other features), provided that this selection of features is sufficient to confer a technical advantage or to differentiate the invention from the prior art. This selection includes at least one preferably functional feature without structural details, and / or with only a portion of the structural details if this portion alone is sufficient to confer a technical advantage or to differentiate the invention from the prior art.

[0032] The method 100 of Figure 4 aims to determine the characteristics of a molding composite layer of a measurement element of the semiconductor object type comprising one or more chips and at least one molding composite layer of at least a part of said chips.

[0033] The method 100 includes a first step 102 of measuring a measurement signal 51, shown in Figure 2. For example, the measurement signal 51 can be obtained by a low-coherence interferometry system based on a double Michelson interferometer using single-mode optical fibers. The interferometry system may include: - a light source capable of emitting a measurement beam to illuminate the element to be measured 5, and - an optical delay line to produce a reference beam.

[0034] The retro-reflections from the element to be measured are collected and analyzed by interfering with each other and / or with the reference beam, which produces the measurement signal 51.

[0035] The measurement signal 51 in Figure 2 is an example of an interferometric measurement for the element to be measured, comprising four layers of chip-molding composite from a semiconductor sample. The composite layers have respective thicknesses d1, d2, d3, and d4 and are separated by spacings 11, 12, and 13. The continuous curve illustrates a measurement signal 51 with useful peaks 52 corresponding to interferogram envelopes and representative of reflections of the measurement beam at interfaces of the composite layers of the element to be measured.

[0036] In the example in Figure 2, the respective positions of the expected composite layer surfaces, as identified on the measurement signal 51, are indicated by crosses, corresponding to the peaks 52. However, the measurement signal 51 also includes so-called ghost peaks 53 which are intercalated between the useful peaks 52.

[0037] To identify the actual interfaces, represented by the useful peaks 52, and distinguish them from the phantom peaks 53, the method 100 comprises steps 104 and 106, which consist of selecting the useful peaks 52 of the measurement signal 51 using a known model of the decomposite layer stacking. An example of model 200 is shown in Figure 3, where the symbols dl' to d4' and el' to e3' are to be associated, respectively, by the method, with the intervals dl to d4 and el to e3 of Figure 2.

[0038] In step 104, the model 200 corresponding to the element to be measured is provided. The model 200 comprises lines 202, where each line corresponds to the position of an interface or surface of one of the composite layers of the element. This correspondence means that the position can be offset by a certain distance (positive or negative), and that since the thicknesses of the measured elements are generally different from the thicknesses of the model, there is no simple correspondence with, for example, a single offset for all the lines of the model and the lines of the measurement.

[0039] Model 200 depends on the composition and arrangement of the element to be measured. For example, process 100 includes a step of communicating with a database containing several models and a step of selecting the model corresponding to the element to be measured.

[0040] The process then includes a step 106 of associating the peaks of the model with corresponding useful peaks in the measurement signal. To this end, step 106 comprises the following substeps: a) extract, for each peak of the measurement signal, a measured position and, for each peak of the model, a modeled position, b) determine distances between each measured position and each modeled position according to the following formula: D[c2(i, f) ] = pos_meas[i] — pos_Mdl[j] with pos_meas[i] being the measured position with index i between 1 and I the number of measured positions, pos_Mdl[j] being the modeled position with index j between 1 and J the number of positions in the provided model, and [c2(i,y)] being the distance for the candidate pair (i, y). The number J corresponds in this case to the number of lines 202. The number I corresponds to all the peaks represented by a cross and a plus sign in the measured signal 51. c) Reorganize the previously determined distances in an ordered manner. At this stage, an ordered list of distances D[c2(i,j)] is generated, for example in ascending order, which are renumbered as D[c], with indices c between 1 and C. For each c, the indices i[c], j[c] at the origin of index c are stored. d) For each distance, determine a gap score, for the plurality of measured and modeled positions, according to the following formula: SC(c) = P(pos_meas[ic] - pos_Mdl[jc] - D[c]), for (ic,jc) in ℝ, I x ℝ, J, where SC(c) is the slack score of index c corresponding to that index c2(ic,jc) after its ordering, and D[c] is the distance determined at step c2. P is a function that has a maximum of 1 when its argument tends towards 0 and a minimum of 0 when its argument is greater than a given threshold, the function P varying inversely with the absolute value of its argument. For example, the slack score can be between 0 and 1. Thus, the slack score can be at most 1 and at least 0; and can be maximum when the value of pos meas [ ic ] - pos Md i[ jc ] - D[c] is equal to 0. The gap score can increase when the value of pos meas [ ici - pos Mdl [ jci -D[c] tends towards 0 (through positive values, or through negative values). Finally, the gap score can be 0 when pos meas [ ic ] - pos Md i[ jc] -£>[c] is large or very large, for example greater than a threshold. Figure 5 represents an example of the function P, which is a Gaussian function. The argument x corresponds to the value of pos_meas[ic] - pos_Mdl[jc] - D[c]. Furthermore, the function α can depend on other parameters. In particular, the function P can also depend on the amplitudes of the measured peaks. In this case, this allows the deviation score to be higher when the amplitude of the measured peak is greater, which helps to choose a useful peak over another potentially similar one that is due to measurement aberrations such as a ghost or spurious peak. This allows the selection of the most probable peak to associate with the model. e) compare the gap scores and select one or more indices c0 giving the highest gap scores; and for each index c0 select the neighboring or close indices cl whose associated indices il and jl correspond to associations between measured peak and modeled peak complementary to i0 and j0, increasing the number of associated (i,j) pairs, f) determine a relevance criterion for each index c0 selected in the previous step, such as the number of pairs selected and an overall score on the selected pairs, This overall score can be the product of the scores for each selected pair.g) select the useful peaks as the measured peaks associated with the c0 index(es) for which the relevance criterion is satisfied.

[0041] Alternatively, in substep v), the distances can be determined taking into account measurement tolerances. The distances can be calculated using the following formula: D[c2(i, j, k)] = pos_meas[i] — pos_Mdl[j] — e[k] * TOL[j] with pos_meas[i] being the measured position with index i between 1 and I the number of measured positions, pos_Mdl[j] being the modeled position with index j between 1 and J the number of positions of the provided model, D[c2(i,j)] being the distance for the candidate pair (i,j), TOL[j] being the tolerance of the modeled position and e[k] being a weighting factor, and k a weighting index between 1 and K.

[0042] Thus, it is possible to take into account errors related to the measurement process of the measured peaks, and especially production deviations which produce a dispersion of the peak positions from the measurements, relative to the model. For example, the tolerance sequence e[k] can vary as follows: for example e[1] = -1; e[2] = 0; e[3] = +1, or e[1] = -1.2; e[2] = -0.6; e[3] = 0; e[4] = +0.6; e[5] = +1.2.

[0043] Tolerances on modeled and / or measured positions may depend on tolerances on the thicknesses of the model layers.

[0044] Process 100 then includes a step 108 of extracting the thicknesses (dl-d4) and spacings (el-e3) from the selected useful peaks. For this, so-called optical distances are calculated as the differences between the positions of each of the previously selected useful peaks. These distances are divided by the optical refractive indices of the composite layers to obtain the thicknesses (dl-d4) and spacings (el-e3).

[0045] The 100 process can be used to control a chip molding composite layer of a sample or object, after the chip molding process is complete, particularly during manufacturing steps that take place after the chip molding process.

[0046] Alternatively, process 100 can be used to control the chip molding composite layer of a sample or object, during the chip molding process, particularly after the molding composite has been deposited on the chips and / or during thinning steps of said molding composite.

[0047] Figures 5a-5d are schematic representations of a non-limiting example embodiment of a chip molding process of a sample, or a semiconductor device, with a molding composite.

[0048] The process shown in figures 5a-5d allows at least one chip of a sample 1100 to be molded with a molding composite, referred to as compound in the following.

[0049] The element to be measured or element to be characterized can be sample 1100.

[0050] Sample 1100, shown in figures 5a-5d, is by no means exhaustive and is given for illustrative purposes only.

[0051] Sample 1100 includes a support 1102, also called a carrier, which may be, for example, in the form of a metal plate or a glass plate.

[0052] Sample 1100 comprises one or more chips 1104 deposited on support 1102. In the example shown, only one chip 1104 is depicted. Of course, the sample may contain several chips arranged one on top of the other, or side by side, or a combination of these two configurations.

[0053] Optionally, an intermediate layer 1106, called an interposer, may be arranged between the support 1102 and the chip 1104.

[0054] Optionally, one or more electrical connections 1108, such as electrical connection pads or electrical disconnection tracks, may be located on the chip 1104, on the side opposite the support 1102. In the following, without loss of generality, it is assumed that the electrical connections are located on a top face of the chip 1104, opposite the support 1102.

[0055] Optionally, an electrical insulation layer 1110 may be deposited on the chip 1104, in particular on the top face of the chip 1104, between or around the electrical connections 1108.

[0056] Again, the sample example given in Figures 5a-5d is by no means limiting and sample 1100 may include only some of the elements described above and / or other element(s) than those described above.

[0057] In the example shown, chip 1104 is located on the side of a first face 1112 of sample 1100. This first face 1112 is referred to as the top face of sample 1100 hereafter. The opposite face 1114 of sample 1100 is referred to as the second face, or bottom face, hereafter.

[0058] The molding process, as known, aims to mold the chip 1104 in a molding composite 1120. The molding composite 1120 can be of any shape / composition. According to one embodiment, the molding composite 1120 can, for example, be in the form of an epoxy filled with thermal expansion control elements, such as silica beads.

[0059] Figure 5a represents sample 1100 before molding chip 1104 in molding composite 1120.

[0060] Figure 5b shows sample 1100 after the molding composite 1120 has been deposited onto chip 1104. The molding composite 1120 can be deposited using any known technique, for example, by depositing a paste. The molding composite 1120 is deposited onto chip 1104 so as to completely cover said chip 1104. Chip 1104 is then embedded in the molding composite 1120, as shown schematically in Figure 5b. In all cases, the molding composite 1120 has a significant thickness, which may be greater than chip 1104, resulting in a non-negligible thickness above chip 1104, for example, on the order of several tens of micrometers, for example, on the order of 50 micrometers or 100 micrometers above the chip.

[0061] Following a non-limiting embodiment example, the thickness of the molding composite 1120 is on the order of 900 pm, in a measurement position located at the periphery of the chip 1104.

[0062] The compound 1120 deposited on the sample has a first interface 1122, also a free interface or upper interface, on the side opposite the support 1102. The compound has a second interface 1124, also a buried interface or lower interface, in contact with the upper face 1112 of the support 1102.

[0063] Next, the molding composite 1120 is thinned to reduce the thickness of the molding composite 1120, until the upper face of the chip 1104, or the electrical connections 1108 located on said chip 1104, is exposed. The thinning of the molding composite 1120 is generally carried out in several passes, each pass removing a fraction of the thickness of the molding composite 1120, thus moving the upper interface 1122 of the compound 1120 towards the sample 1100.

[0064] The thinning, or removal, of the molding composite 1120 above the chip 1104 can be achieved by any known technique, for example by planing, grinding, and / or by a chemical process.

[0065] Figure 5c schematically represents the thinning stage.

[0066] The thinning step ends when all the molding composite 1120 above the chip 1104 has been removed so that the top face of the chip 1104 is apparent, where the electrical connections 1108 on said top face of the chip 1104 are apparent.

[0067] Figure 5d shows sample 1100 when the thinning step is complete. In this example, since chip 1104 has electrical connections 1108, the thinning step is terminated when these connections 1108 are visible. The molding composite 1120 is still present around chip 1104.

[0068] The invention is not limited to the example just described and the molding process may include other step(s) than the one(s) just described.

[0069] Generally, the chip molding process from a sample requires controlling the thickness of the composite 1120 during its thinning for obvious reasons. It is important not to damage the chip 1104, or the electrical connections 1108 if applicable, during the thinning step. Otherwise, the chip 1104 is unusable and the sample 1100 is discarded, resulting in a significant loss.

[0070] The invention makes it possible to measure the thickness of the composite layer 1120 at any time after the composite has been deposited, for example to control and / or guide the thinning.

[0071] The invention makes it possible to measure the thickness of the 1120 composite layer after the end of thinning, and / or when the molding process is complete.

[0072] Of course, the invention is not limited to the examples just described and many modifications can be made to these examples without departing from the scope of the invention.

Claims

DEMANDS 1. A method (100) for measuring the characteristics of a molding composite layer of an element to be measured from signals measured by an interferometry system, said element to be measured being a semiconductor object comprising one or more chips and at least one molding composite layer of at least a portion of said chips, the method is characterized in that it comprises the following steps: a) receive (102) a plurality of measured peaks relating to beams reflected by the element to be measured when it is illuminated by an optical beam emitted by the interferometry system, b) receive (104) a model (200) comprising a plurality of modeled positions, each modeled position corresponding to an interface of the element to be measured encountered by the optical beam, c) automatically associate (106), among the measured peaks, so-called useful peaks, with corresponding modeled positions, d) determine (108) one or more thicknesses of the element to be measured from the so-called useful peaks associated with these modeled positions.

2. A method (100) according to the preceding claim, wherein step c) comprises the following steps: cl) extract, for each measured peak, a measured position, c2) determine distances between each measured position and each modeled position, c3) For each of the distances, determine a gap score, said gap score decreasing as the distance increases in absolute value, c4) select useful peaks as corresponding to measured positions exhibiting the highest score in relation to the corresponding modeled position.

3. Method (100) according to the preceding claim, comprising a step for adding a tolerance to each measured position, or to each modeled position.

4. Method (100) according to claim 2 or 3, wherein step c2) is carried out by the following formula: D[c2 i,;)] = pos_meas[i] — pos_Mdl\f] with pos_meas[ï] being the measured position of index i between 1 and J the number of measured positions, pos_Mdl[j] being the modeled position of index j between 1 and J the number of positions of the provided model and D[c2(i,j)] being the distance for the candidate pair (i,y).

5. Method (100) according to claim 3, wherein step c2) is carried out by the following formula: D[c2 i,j,k) ] = pos_meas[i] — pos_Mdl\f] — e[k] * TOL\f] with pos_meas[i] being the measured position of index i between 1 and J the number of measured positions, pos_Mdl[j] being the modeled position of index j between 1 and J the number of positions of the provided model, D[c2(i,j)] being the distance for the candidate pair (i ), TOL[j] being the tolerance of modeled position and e[ / c] being a weighting factor, and k a weighting index between 1 and K.

6. Method (100) according to any one of claims 2 to 5, wherein the method comprises, following step c2), a step for rearranging the distances determined in step c2) in an orderly manner.

7. Method (100) according to claims 4 and 6, wherein the deviation score is determined in step c3) by the following formula: SC(c) = P(pos_meas[ic] − pos_Mdl[jc] − D[c]), for (ic,jc) in 1, I x 1, J with SC(c) being the index deviation score c corresponding to this index c2(ic,jc) after its ordering, and D[c] the distance determined at step c2) and P being a function which has a maximum of 1 when its argument tends towards 0 and a minimum of 0 when its argument is greater than a given threshold, the function P varying inversely to the absolute value of its argument.

8. Method (100) according to the preceding claim, wherein the function P further depends on the amplitudes of the measured peaks.

9. A method (100) according to any one of the preceding claims, wherein step c4) comprises the following substeps: c4-1) compare the gap scores and select the index(es) c0 giving the highest gap score(s); and for each index c0 select the neighboring or close indices cl whose associated indices il and jl correspond to associations between measured peak and modeled peak complementary to i0 and j0, increasing the number of associated (i, j) pairs, c4-2) determine a relevance criterion for each index c0 selected in the previous step, such as the number of pairs selected and an overall score on the selected pairs, and c4-3) select the useful peaks as the measured peaks associated with the index(es) c0 for which the relevance criterion is satisfied.

10. A method (100) according to any one of the preceding claims, wherein step (d) comprises the following steps: d1) determine so-called optical distances calculated as the differences in the positions of each of the previously selected useful peaks, d2) divide the optical distances by optical refractive indices in order to obtain so-called physical thicknesses.

11. Method (100) according to any one of the preceding claims, wherein in step b) the model corresponds to a thickness model of the layers to be measured, which model is decomposed into corresponding interface positions.

12. Method (100) according to any one of the preceding claims taken in combination with claim 3, wherein the tolerances on the modeled and / or measured positions depend on tolerances on the thicknesses of the model layers.

13. A device for measuring a molding composite layer of an element to be measured from signals measured by interferometry, said element to be measured being a semiconductor object comprising one or more chips and at least one molding composite layer of at least a part of said chips, the device comprising a measurement channel configured to produce a measurement beam at a determined position relative to a field of view of the element and comprising an optical sensor configured to acquire signals reflected by the element to be measured, a processing module configured to process the signals acquired by the optical sensor, the processing module being configured to implement the method (100) according to one of the preceding claims.

Images