METHOD FOR MEASURING THE THICKNESS OF A SURFACE LAYER OF AN SOI SUBSTRATE

Abstract

The invention relates to a method for measuring at least one thickness parameter of a surface layer of an SOI substrate, comprising the following steps: a) measuring the thickness at a predefined number of first points, distributed spatially on the surface, resulting in a first series of values ​​comprising a first maximum and a first minimum; the thickness measurement being carried out by a first technique; b) carrying out a complete mapping of the surface of the surface layer by a second technique, different from the first technique, the mapping translating a physical characteristic of the surface layer correlated with the thickness; c) analyzing the mapping so as to identify whether there is a signature capable of translating a variation in thickness, greater than or equal to the difference between the first maximum and the first minimum, in a localized region of the surface layer;d) if a signature is identified, measuring thickness at a plurality of second points positioned in the localized region, resulting in a second series of values ​​comprising a second maximum and / or a second minimum; the thickness measurement being carried out by the first technique; e) calculating at least one thickness parameter, based on the first series of values ​​and, potentially on the second series of values. Figure to be published with the abstract: No figure;

Description

Title of the invention: METHOD FOR MEASURING THE THICKNESS OF A SURFACE LAYER OF AN SOI SUBSTRATE FIELD OF THE INVENTION

[0001] The present invention relates to the field of semiconductors and in particular SOI (Silicon on Insulator) substrates. The invention relates to a thickness measurement method applying to a surface layer of an SOI substrate, in particular when the surface layer has an average thickness of less than 50nm: the method drastically improves the reliability of the thickness parameters (average, minimum, maximum, etc.) obtained, reliability being understood herein by the fact that these parameters are representative of the surface layer over its entire surface.

[0002] TECHNOLOGICAL BACKGROUND OF THE INVENTION

[0003] More and more applications based on SOI substrates require very good thickness uniformities of the silicon surface layer. For example, in digital applications, the active layer of FDSOI (fully depleted SOI) substrates must have very small thickness variations because these have an impact on the threshold voltage of the transistors produced in and / or on the active layer.

[0004] The specifications in terms of thickness and uniformity therefore become very aggressive: for a layer with a thickness typically less than 50nm, or even 25nm, we expect a uniformity (corresponding to the difference between the maximum thickness and the minimum thickness of the layer) of less than Inm.

[0005] In a manufacturing line, the thickness parameters (average thickness, uniformity, etc.) of a surface layer of FDSOI substrate are extracted from a limited number of measurement points (for example, 41 points), for understandable reasons of production rate. Even if the position of the measurement points on the surface of the surface layer is predefined to try to capture the thickness information most representative of the entirety of said surface, it is not guaranteed to detect the thickness minima and maxima, the location of the latter being likely to vary from one substrate to another.

[0006] SUBJECT OF THE INVENTION

[0007] The present invention aims to improve the reliability of the thickness parameters (average, minimum, maximum, uniformity, etc.) obtained, reliability being understood herein as the fact that these parameters are representative of the surface layer over its entire surface. It relates to a thickness measurement method applying to a surface layer of an SOI substrate, in particular when the surface layer has an average thickness of less than 50nm; the method implements a standard tool thickness measurement at a finite number of points, and uses information from a map to identify signatures of potential thickness variations, and if necessary, carry out additional thickness measurements in the localized regions corresponding to these signatures.

[0008] BRIEF DESCRIPTION OF THE INVENTION

[0009] The invention relates to a method for measuring at least one thickness parameter of a surface layer of an SOI substrate, the surface layer having a surface extending along a main plane, the method comprising the following steps:

[0010] a) measuring thickness at a predefined number of first points, spatially distributed over the surface, resulting in a first series of values ​​comprising a first maximum and a first minimum; the thickness measurement being carried out by a first technique and the first points not covering the surface in its entirety;

[0011] b) carrying out a complete mapping of the surface of the superficial layer by a second technique, different from the first technique, the mapping reflecting a physical characteristic of the superficial layer correlated with the thickness;

[0012] c) analyzing the mapping so as to identify whether there is a signature capable of reflecting a variation in thickness, greater than or equal to the difference between the first maximum and the first minimum, in a localized region of the surface layer;

[0013] d) if a signature is identified, measuring the thickness at a second point or a plurality of second points positioned in the localized region, resulting in a second series of value(s) comprising a second maximum and / or a second minimum; the thickness measurement being carried out by the first technique;

[0014] e) calculating at least one thickness parameter, based on the first series of values ​​and, potentially on the second series of value(s).

[0015] According to other advantageous and non-limiting characteristics of the invention, taken alone or in any technically feasible combination: • the surface layer has an average thickness less than or equal to 50nm, less than or equal to 25nm, less than or equal to 20nm, or less than or equal to 15nm; • the second technique is reflectometry and the physical characteristic is thickness; • the second technique is based on laser scattering surface inspection in which a laser beam scans the surface of the surface layer, and the physical characteristic is the diffuse background noise (“haze”); • a signature corresponds to an abnormal variation in intensity of the diffuse background noise signal, and can appear in the form of a point pattern, a periodic wave, or other patterns; • the signature is identified on the map using an image recognition algorithm or by thresholding the intensity of the diffuse background noise signal; • at the end of step c), coordinates are assigned to the signature, in the main plane, by a piece of equipment - called second equipment - implementing the second technique, and said coordinates are communicated, prior to step d), to a first piece of equipment implementing the first technique; • the second points are positioned in the main plane according to distribution patterns established according to the shape of the signature; • the -at least one- thickness parameter calculated in step e) is uniformity; • several thickness parameters are calculated in step e), including an average, and / or a median, and / or a standard deviation; • the first technique is spectral ellipsometry; • the predefined number of first points is 12, 41 or 625; • the second points are 2, 9, 16 or 25 in number. BRIEF DESCRIPTION OF THE FIGURES

[0016] Other characteristics and advantages of the invention will emerge from the detailed description of the invention which follows with reference to the appended figures in which:

[0017] [Fig.l] [Fig.l] shows the steps of a measuring method according to the present invention;

[0018] [Fig.2]

[0019] [Fig.3] Figures 2 and 3 illustrate two examples of mapping obtained at step b) the measuring method according to the invention ([Fig.2]: by reflectometry, first embodiment; [Fig.3]: by laser scattering inspection, second embodiment);

[0020] [Fig.4a]

[0021] [Fig.4b]

[0022] [Fig.4c]

[0023] [Fig.4d] Figures 4a, 4b, 4c, 4d illustrate four examples of maps obtained according to the second embodiment of step b), having particular signatures; they also illustrate four associated examples of arrangement of the second measurement points in step d), in a measurement method according to the invention;

[0024] [Fig.5a]

[0025] [Fig.5b]

[0026] [Fig.5c]

[0027] [Fig.5d] Figures 5a, 5b, 5c, 5d illustrate an example of results obtained respectively in steps a), b), c), d) when implementing a method according to the invention. DETAILED DESCRIPTION OF THE INVENTION

[0028] The present invention relates to a method for measuring one or more thickness parameters of a silicon surface layer of an SOI (silicon on insulator) substrate.

[0029] By thickness parameters is meant thickness uniformity, mean (average thickness), median, standard deviation, etc. These parameters are usually calculated from the surface layer thickness values ​​obtained at several measurement points. In particular, thickness uniformity can be determined by the difference between the maximum and minimum thickness measured.

[0030] An SOI substrate is usually in the form of a circular wafer, with a diameter ranging from 150mm to 450mm, for example conventionally 300mm for FDSOI substrates. The silicon surface layer is arranged on a dielectric layer, itself arranged on a support substrate. The surface of the surface layer extends in a main plane (x,y).

[0031] Advantageously, this method applies to an SOI substrate having a very thin surface layer, typically with an average thickness less than or equal to 50nm, less than or equal to 25nm, less than or equal to 20nm, or even less than or equal to 15nm. For such substrates, the buried dielectric layer of the SOI substrate is generally between 10nm and 100nm.

[0032] Among the physical characteristics requiring compliance with specifications at the end of the manufacturing line, there are of course the thickness parameters of the surface layer of a SOL substrate. As mentioned in the introduction, it is not easy to guarantee that these thickness parameters are representative of the entire surface layer, because very often, a limited number of measurement points are implemented in the metrology recipes, in order to comply with the production, grading and shipping rates.

[0033] To overcome this problem, the measuring method according to the invention comprises a succession of steps ([Fig.l]) which will now be described.

[0034] Step a) corresponds to the measurement of the thickness of the surface layer of the SOI substrate, at a predefined number of points (called first points), distributed spatially on the surface of said layer. The spatial distribution of the first points in the plane The main (x,y) can take different shapes, more or less classic, for example, a cross, a snail, a star, etc. The most peripheral points preferably extend up to 3 mm from an edge (contour) of the SOI substrate, knowing that the surface layer generally extends at least up to 0.5 mm from said edge.

[0035] This thickness measurement is carried out by a first technique, chosen to be precise and reliable on an SOI substrate. Preferably, the first technique is spectral ellipsometry. We can notably cite the F5X industrial equipment from the company KLA. The wavelength range is typically 240nm - 750nm and the measurement accuracy is 0.03nm for an SOI substrate having a surface layer with an average thickness less than or equal to 50nm.

[0036] The predefined number of first points may be, for example, 12, 41 or 625. A relatively small number of points is preferred to limit the loss of time at this metrology step a): the first points cannot therefore cover the surface of the layer in its entirety.

[0037] Step a) results in a first series of thickness values ​​comprising a maximum (called the first maximum) and a minimum (called the first minimum). At this stage, an intermediate thickness uniformity can be calculated, from the first maximum and minimum.

[0038] Step b) of the method corresponds to the production of a complete mapping of the surface of the superficial layer, said mapping reflecting a physical characteristic of the superficial layer more or less correlated with or representative of the thickness. In other words, the physical characteristic is correlated with the specular reflectivity of the SOI substrate, reflectivity which is itself a function of the thickness of the multilayer according to the Fresnel formalism.

[0039] By complete mapping, we mean that the physical characteristic is measured over the entire surface of the surface layer, with a measurement mesh (spacing between two consecutive measurement points) less than or equal to 5mm, 1mm, or even less than or equal to 500qm (for example 200qm, corresponding to the size of a pixel).

[0040] Step b) implements a second technique, different from the first technique. This second technique is advantageously chosen from among the techniques allowing a high measurement rate.

[0041] According to a first embodiment, the second technique is reflectometry and the physical characteristic is the thickness. Equipment of the ACUMAP type from the company ADE can for example be used. It allows a thickness measurement according to grids whose pitch can go down to 1 mm, or more than 70,000 points, which provides a thickness map of the entire surface of the superficial layer. [Fig. 2] illustrates an example of thickness mapping obtained by reflect- tometry.

[0042] Although a measurement by reflectometry is less precise than a measurement by ellipsometry (as used in step a)), the measurement rate allows a map to be obtained quickly, which can be analyzed in the following step to possibly identify signatures of significant variations in thickness, potentially outside the areas measured in step a).

[0043] According to a second embodiment, the second technique corresponds to a surface inspection by laser scattering, in which a UV laser beam scans the surface of the surface layer.

[0044] It is usual to inspect an SOI substrate by dark-field microscopy using an incident light beam projected onto an inspection point scanning the exposed surface of the substrate, to qualify the level of defectivity. A light-collecting device and a detector make it possible to measure the scattered light at the inspection point. Such an inspection can in particular be carried out using inspection equipment of the Surfscan™ SP1, SP2, SP3 or SP5 type from KLA. When the inspected surface has a surface irregularity or a particle present at the inspection point, this leads to an increase in the scattered light signal. The measurement of this radiation constitutes a signature carrying information on the presence, position and / or nature of a surface defect or a particle at the inspection point.Scanning the entire surface layer with the laser beam provides a complete defect mapping, which is also essential for grading an SOI substrate. The wavelength of the laser beam is between 200nm and 500nm depending on the equipment. Preferably, it is in the ultraviolet (UV) or even deep ultraviolet (DUV) range, typically between 200nm and 280nm. For example, the SP5 equipment offers a 266nm laser.

[0045] After scanning the surface of the superficial layer by the laser beam, in addition to a defect mapping, this type of equipment can also compile a mapping from the diffuse background noise signal (better known as "haze"). The diffuse background noise corresponds to the diffuse reflectance of the measured substrate. An example of a diffuse background noise mapping is illustrated in [Fig.3], for a 300mm diameter SOI substrate, with a superficial layer of 12 nm average thickness and a buried dielectric layer (SiO2) of 25 nm average thickness.

[0046] The diffuse background noise signal is known to be representative of the surface roughness of the surface layer. In the case of a thin surface layer (typically less than or equal to 50nm), in particular when the surface roughness is low and uniform (for example, less than 0.5nm RMS on an AFM scan of 30x30um2, with a uniformity of less than 0.3mn), the diffuse background noise can be correlated with the thickness of the surface layer (Brun et al, “Defect inspection challenges and solutions for ultra-thin SOI” - 2012 SEMI Advanced Semiconductor Manufacturing Conference).

[0047] Indeed, the diffuse reflectance is then very weakly affected by the surface roughness and it is correlated to the specular reflectance, in the particular case where the surface layer is sufficiently transparent to the wavelength of the laser beam, so as to probe the entire thickness of the layer.

[0048] Thus, in the second embodiment, the physical characteristic imaged by the mapping of step b) is the diffuse background noise. This embodiment is particularly advantageous in that, since surface inspection by laser scattering is required to verify the level of defectivity of an SOI substrate, performing the diffuse background noise mapping does not add metrology time to the SOI substrate production process.

[0049] Returning to the general description of the method according to the invention, the following step c) corresponds to the analysis of the mapping obtained in step b), so as to identify whether there is a signature capable of translating a variation in thickness, greater than or equal to the difference between the first maximum and the first minimum, in a localized region of the surface layer.

[0050] Taking as an example the mapping of [Fig.2] (first embodiment of step b), we notice a wave pattern SI which extends globally over the entire surface of the surface layer. This signature SI is visible on the mapping (pattern marked in relation to the surrounding areas), and it is also characterized by a strong variation in thickness.

[0051] Taking as an example the mapping of [Fig.3] (second embodiment of step b), we also note at least one signature S2 corresponding to an abnormal variation in intensity of the diffuse background noise signal, and which is in the form of a periodic wave. Other signatures may be in the form of a point pattern, an aperiodic wave, or other patterns.

[0052] Advantageously, the signature is identified on the map by means of an image recognition algorithm or by thresholding the intensity of the diffuse background noise signal.

[0053] An image recognition algorithm can be fed and trained on different types of signatures likely to be present on a diffuse background noise map; it can thus learn to recognize each of these signatures and to classify them into predefined categories, taking into account criteria of size, shape, contrast, etc.

[0054] Signatures can alternatively be identified by applying a certain threshold (or several thresholds) to the intensity of the diffuse background noise signal, on parameters such as the standard deviation, the relative median or the relative mean of the signal: an intensity detected beyond (or below) the threshold on the map then reflects a signature likely to correspond to a significant variation in thickness, in a localized region of the surface layer.

[0055] Figures 4a, 4b, 4c and 4d show four examples of diffuse background noise maps on which different signatures are noted (surrounded by a white square). In [Fig.4a], the signature S3 corresponds to a point pattern, reflecting a localized excess thickness of the surface layer. In [Fig.4b], the signature S4 corresponds to a scratch starting from the edge of the plate and extending towards the inside of the layer; it can cause a significant variation in the thickness of the surface layer. In [Fig.4c], the signature S5 corresponds to an extended dark area, reflecting an excess thickness of the surface layer. Finally, in [Fig.5d], the signature S6 corresponds to a wave pattern characterized by variations in the thickness of the surface layer.

[0056] For example, when thresholds are applied to detect the signatures: - a point pattern type S3 signature is identified when the relative median of the diffuse background noise signal collected in the “narrow” channel (“haze narrow”) is greater than 0.025; - a stripe-type S4 signature or other fine variations is identified when the standard deviation of the diffuse background noise signal collected in the “narrow” channel (“haze narrow”) is greater than 0.04; - an S5 signature of the extended zone of excess thickness type is identified when the relative median of the diffuse background noise signal collected in the “narrow” channel (“haze narrow”) is less than 0.95; - a wave-type S6 signature is identified when the relative average of the diffuse background noise signal collected in the “wide” channel (“haze wide”) is less than -0.006.

[0057] As a reminder, in SPx type equipment, the diffused light coming from the substrate is collected, either by an elliptical mirror, for the rays of incidence far from the normal (z) to the main plane (x,y) of the layer (collector called "wide" or "wide" channel), or by a lens, for the rays of incidence close to the normal (z) (collector called "narrow" or "narrow" channel).

[0058] Step b) and step c) can be carried out on a single piece of equipment, which implements the second technique and includes an analysis module for identifying potential signatures on the map.

[0059] Alternatively, step c) can be carried out by another piece of equipment comprising the analysis module. The latter receives the diffuse background noise mapping files established in step b), applies signature detection to them (via an image recognition algorithm or by thresholding), and generates, if signatures of interest are present, a file containing the location of the signatures in question.

[0060] If no signature is found on the mapping resulting from step b), the calculation of the thickness parameter(s) (among which, for example, uniformity) can be done from the first series of values ​​obtained in step a). The absence of a signature confirms the fact that these thickness values ​​are representative of the entire surface layer.

[0061] If at least one signature of interest is noted on the map, the method according to the invention provides a step d) of measuring thickness at a second point or at a plurality of second points positioned in the localized region (attached to the signature on the map). This thickness measurement step results in a second value (in the case of a single second measurement point) or in a second series of values, comprising a second maximum and / or a second minimum. The measurement is carried out using the first technique (for example, spectral ellipsometry).

[0062] Preferably, at the end of step c), coordinates are assigned to the signature, in the main plane, by the second device (the one implementing the second technique). These coordinates are communicated automatically, prior to step d), to the first device implementing the first technique, via the generation of a file comprising the location of the signature.

[0063] In the first equipment, the second points are positioned in the main plane according to distribution patterns established according to the shape and type of signature. In the examples illustrated in Figures 4a, 4b, 4c, 4d, it is observed that the distribution patterns of the second points can be in the form of a cross, centered on the coordinates of the center of the signature. The cross has a size greater than the signature so that at least the extreme points of said cross fall outside the signature: this makes it possible to ensure that the signature has been fully evaluated. In the figures, the size of the square surrounding all or part of the distribution pattern of the second points is indicated above said square.

[0064] In the example of [Fig.4d], a star or concentric distribution is preferred to capture possible thickness variations in all directions; a high density of second points in the center favors the detection of the thickness extremum, and the outer points, further away (typically 10 mm) capture the other thickness extremum of the wave pattern.

[0065] Preferably, the second points are 10, 16 or 25 in number, to avoid taking too much measurement time. Of course, any other number of second points is possible, depending on the size, shape and type of signature.

[0066] Of course, if several signatures are identified on the mapping, each of the localized regions of the surface layer corresponding to these signatures will be evaluated during step d), by a local thickness measurement.

[0067] The measuring method according to the invention finally comprises a step e) corresponding to the calculation of at least one parameter of thickness of the surface layer, based on the first series of values ​​and, potentially on the second series of values, if step d) has taken place.

[0068] In the particular case where the -at least one- thickness parameter is uniformity, the first minimum and maximum are compared to the second minimum and maximum: the lowest of the minima and the highest of the maxima is then taken into consideration to determine the thickness uniformity of the layer.

[0069] Step e) may also comprise the calculation of other thickness parameters of interest, namely the mean, the median, the standard deviation, etc. These latter may be established from the first series of values, possibly supplemented by the second series of values. Examples

[0070] - SOI: average thickness of the silicon surface layer 12nm; thickness of the buried oxide layer (SiO2) 25nm; - Step a): measurement on 41 first points ([Fig.5a]) by spectral ellipsometry (first equipment, first technique); obtaining a first series of thickness values: maximum 12.29nm, minimum 11.66nm; - Step b): diffuse background noise mapping (second equipment, second technique) ([Fig.5b]); - Step c): analysis of the mapping (analysis module), identification of a point pattern type signature ([Fig.5c]) and transfer of the signature coordinates to the first device; - Step d): measurement on 9 second points, according to a cross distribution, centered on the signature, by spectral ellipsometry (first equipment, first technique), obtaining a second series of thickness values: maximum 14.47nm, minimum 13.15nm; - Step e): calculation of the thickness parameters to obtain in particular a uniformity, based on the first and second series of values, of 2.8 Inm; the uniformity based on the first series of values ​​was 0.64nm. For a thickness uniformity specification of Inm, the measurement of step d) made it possible to identify that the surface layer of this SOI did not meet the specification, whereas the first series of values ​​(step a)) alone would not have been sufficient to raise it.

[0071] The measurement method according to the present invention drastically improves the reliability of the thickness parameters (average, minimum, maximum, uniformity) taken into account for the gradation of an SOI substrate. These parameters are representative of the surface layer over its entire surface and the expected specifications are thus perfectly respected.

[0072] Of course, the invention is not limited to the embodiments and examples described, and variant embodiments may be made without departing from the scope of the invention as defined by the claims.

Claims

Claims

1. Method for measuring at least one thickness parameter of a surface layer of an SOI substrate, the surface layer having a surface extending along a main plane (x,y), the method comprising the following steps: a) measuring the thickness at a predefined number of first points, distributed spatially on the surface, resulting in a first series of values ​​comprising a first maximum and a first minimum; the thickness measurement being carried out by a first technique and the first points not covering the surface in its entirety; b) carrying out a complete mapping of the surface of the surface layer by a second technique, different from the first technique, the mapping reflecting a physical characteristic of the surface layer correlated with the thickness;c) analyzing the mapping so as to identify whether there is a signature capable of reflecting a variation in thickness, greater than or equal to the difference between the first maximum and the first minimum, in a localized region of the surface layer; d) if a signature is identified, measuring the thickness at a second point or a plurality of second points positioned in the localized region, resulting in a second series of value(s) comprising a second maximum and / or a second minimum; the thickness measurement being carried out by the first technique; e) calculating the -at least one- thickness parameter, based on the first series of values ​​and, potentially on the second series of value(s).;

2. Measuring method according to the preceding claim, in which the surface layer has an average thickness less than or equal to 50nm, less than or equal to 25nm, less than or equal to 20nm, or even less than or equal to 15nm.

3. A measuring method according to one of the preceding claims, wherein the second technique is reflectometry and the physical characteristic is thickness.

4. A measurement method according to claim 2, wherein the second technique is based on laser scattering surface inspection in which a laser beam scans the surface of the surface layer, and the physical characteristic is haze.

5. A measurement method according to the preceding claim, wherein a signature corresponds to an abnormal intensity variation of the diffuse background noise signal, and may be in the form of a point pattern, a periodic wave, or other patterns.

6. A measurement method according to one of the two preceding claims, wherein the signature is identified on the mapping by means of an image recognition algorithm or by thresholding the intensity of the diffuse background noise signal.

7. Measuring method according to one of the preceding claims, in which: - at the end of step c), coordinates are assigned to the signature, in the main plane, by a piece of equipment - called second equipment - implementing the second technique, and - said coordinates are communicated, prior to step d), to a first piece of equipment implementing the first technique.

8. Measuring method according to the preceding claim, in which the second points are positioned in the main plane according to distribution patterns established according to the shape of the signature.

9. Measuring method according to one of the preceding claims, in which the -at least one- thickness parameter calculated in step e) is uniformity.

10. Measuring method according to one of the preceding claims, in which several thickness parameters are calculated in step e), including an average, and / or a median, and / or a standard deviation.

11. A measuring method according to one of the preceding claims, wherein the first technique is spectral ellipsometry.

12. A measuring method according to one of the preceding claims, wherein the predefined number of first points is 12, 41 or 625.

13. A measuring method according to one of the preceding claims, wherein the second points are 2, 9, 16 or 25 in number.

Images