Method for determining a thickness map of a layer to be measured

Abstract

The invention relates to a method for determining a thickness map of a layer to be measured, using a camera and a light source. The method comprises the following steps: A. Acquiring a measurement image of the layer illuminated by the light source with the camera; B. Obtaining a reference image of a reference reflector illuminated by the light source with the camera; C. Determining a normalized measurement image, each pixel of the normalized measurement image having at least three coordinates defining a normalized measurement vector; D. Obtaining, for each pixel, a predetermined normalized theoretical vector comprising at least three coordinates, a function of a theoretical thickness of a theoretical layer; E. Identifying, for each pixel, the normalized theoretical vector closest to the normalized measurement vector, the thickness associated with said nearest normalized theoretical vector corresponding to the measured thickness of the layer for the point of the layer imaged on the pixel.Figure to be published: Fig. 2.

Description

Title of the invention: Method for determining a thickness map of a layer to be measured. Technical field

[0001] The invention relates to the field of thin films, and more particularly to optical mapping of the thickness of thin films.

[0002] BACKGROUND

[0003] The interference phenomenon amplifies the variations in the reflectivity of a thin film with wavelength. When a thin film is illuminated with white light, the reflected light is no longer white, and the perceived color will then depend on its refractive index and its thickness. The perceived color varies with thickness. Knowing one, one can deduce the other. The refractive index of a material is a quantity that can be measured precisely and is known and documented for many thin films. In principle, it is therefore possible to use the interference phenomenon to map the thickness of a thin film.

[0004] Color charts exist that list the color perceived by the naked eye as a function of thickness for a given material. The perceived color is an interpretation of the relative intensities perceived by the eye or camera in three wavelength ranges and not a color defined by its wavelength. The color of a layer therefore allows the thickness of a layer to be determined by eye. However, these color charts do not allow for the precise determination of a thickness.

[0005] It is also possible to use a color camera to determine the thickness of a layer by perceiving color through the camera rather than with the eye. A color camera operates on a principle inspired by the human eye. Each pixel of the camera's detector generally comprises sub-pixels equipped with three filters: red, green, and blue. These filters allow the collected light to be filtered, thus enabling coarse spectroscopy over three different wavelength ranges. Figure 1a illustrates an example of the spectral sensitivities of a camera (in arbitrary units) SCR, SCG, SCB associated with three red, green, and blue sub-pixels, respectively, as a function of the X wavelength. The spectral sensitivity associated with each color, typically called "quantum efficiency," is specific to a given camera and is generally provided by the manufacturer.This is typically the product of the spectral sensitivity of the camera without colour filters (black and white camera) and the transmission of the associated colour filter.

[0006] In the example of [Fig.la], the camera also has an ICR infrared filter that cuts wavelengths above 650 nm (also illustrated in [Fig.la]) allowing the infrared light to be cut off, to which the camera sensor is sensitive, unlike the human eye.

[0007] It is also possible to use a detector in which each pixel comprises four sub-pixels equipped respectively with four filters: red, green, blue, and infrared. Figure 1b illustrates an example of the sensitivities (in %) of such an SCR, SCG, SCB, and SCIR camera associated with the four aforementioned filters, as a function of the X wavelength.

[0008] A color camera is therefore at least a "trispectral" camera. Its main advantage is the ability to map an entire sample in a single measurement. However, its low spectral resolution does not allow for the precise measurement of thickness via the color measured by the camera.

[0009] The color measured by the camera is coded in a three-dimensional space (R, G, B), each color being the integral of the intensity of white light by the reflectivity of the thin film by the sensitivity of the camera in the color, over the wavelength range.

[0010] The relative intensity of the three measured colors depends on the camera's relative gain for each of the three colors (called white balance). However, in order to perform a colorimetric measurement that can be used quantitatively, the camera would need to be calibrated on a surface of known reflectivity, and the white balance would need to be applied across the entire field of the acquired image. In practice, this balance is global and cannot be performed pixel by pixel. Consequently, some areas are more or less well balanced, and it is not possible to obtain a white balance across the entire image.

[0011] Because of this impossibility, colorimetry is not used to precisely quantify the thicknesses of thin films at all points of them with a camera.

[0012] For example, document CN107310173 uses colorimetry to control the thickness of thin films during their production. The purpose of this document is to detect thick defects such as hoops, ribs, and grooves during film production. The color of an image is analyzed and compared to a predetermined color to determine if the thickness is defective. However, the method does not require precise mapping of the layer thickness since it is only intended to detect significant defects in the layer.

[0013] Thus, known colorimetry methods do not allow for the accurate mapping of thin film thicknesses.

[0014] One of the aims of the present invention is to overcome the aforementioned drawbacks by proposing a method that uses colorimetric information obtained with a

[0015]

[0016]

[0017]

[0018]

[0019]

[0020]

[0021]

[0022] camera to accurately map the thicknesses of thin layers. SUMMARY The invention relates to a method for determining a thickness map of a layer to be measured, having a refractive index and deposited on a substrate, with: • a camera comprising a detector and configured to image the layer on the detector, the detector comprising a plurality of pixels (Pi) indexed i, each pixel being subdivided into at least three sub-pixels comprising respectively at least three filters (FR, FG, FB), each sub-pixel detecting an intensity, a pixel being characterized by at least three coordinates corresponding to the intensities detected by the at least three sub-pixels, and • a source exhibiting an emission spectrum over a spectral band of interest, the process comprising the steps of: A. Acquire a measurement image of the layer illuminated by the source with the camera; B. Obtain a reference image of a reference reflector illuminated by the source, with the camera, the reference reflector having a predetermined reference reflectivity. C. To determine a normalized measurement image equal, for each pixel, to the ratio of the intensities of the measurement image and the reference image, each pixel of the normalized measurement image having at least three so-called normalized coordinates defining a normalized measurement vector, To have, for each pixel, a predetermined normalized theoretical vector comprising at least three coordinates called normalized theoretical coordinates, a function of a theoretical thickness of a theoretical layer, said normalized theoretical vector having been determined from a predetermined theoretical reflectivity of said theoretical layer and the reference reflectivity, said theoretical layer having a refractive index equal to the refractive index of the layer to be measured, E identify, for each pixel, the theoretical normalized vector closest to the normalized measurement vector, the thickness associated with said theoretical normalized closest vector corresponding to the measured thickness of the layer for the point of the layer imaged on the pixel. According to one embodiment, each of the at least three coordinates of the normalized theoretical vector (y^ ( / ))' is determined by, respectively:

[0023] 1^=.1; SCRU) ■ ^2) • SE(A)dA J, SCRU) ■ RFrefO) ■ SE(^Â

[0024] ■ SEGX2 L SCGÇÏ) • RFre AA) ■ SEGAdÀ

[0025] ^n = "mi L SCB(À) • * SE{A)dA

[0026] with:

[0027] lReMn, IGeMn, ltfMn: coordinates of the normalized vector (i ) ),

[0028] SCR ( A ), SCG ( 2 ) and SCB ( 2 ) camera sensitivity for the at least three sub-pixels respectively,

[0029] Re(ft ( 2 ) predetermined theoretical reflectivity of the theoretical layer of theoretical thickness e

[0030] RFref(^ reflectivity of the reference reflector

[0031] SE ( 2 ) emission spectrum of the source,

[0032] Xmin and Xmax are the limits of the spectral band of interest.

[0033] According to one embodiment the three filters correspond to three colored filters red, green and blue.

[0034] According to one embodiment, step D consists of loading a previously calculated normalized theoretical vector.

[0035] According to another embodiment, step D comprises a first substep consisting of loading refractive index values ​​onto the band of interest and a second substep consisting of determining said normalized theoretical vector.

[0036] According to one embodiment, step B consists of acquiring an image of the reference reflector illuminated by the source with the camera.

[0037] According to another embodiment, step B consists of loading a stored reference image.

[0038] According to one embodiment, the process further comprises a control step F comprising the substeps of:

[0039] Fl identify, where appropriate, pixels exhibiting an abnormal layer thickness by comparison with the thicknesses of at least two nearest neighbors of said pixels,

[0040] F2 for each of the identified pixels, define a range of possible thicknesses based on said thicknesses of said nearest neighbors of said identified pixels,

[0041] F3 determine, for said pixels identified in step Fl, a corrected thickness from the identification of a normalized theoretical vector whose corresponding thickness is within said thickness range.

[0042] According to one embodiment, each pixel is subdivided into four sub-pixels, the first, second and third sub-pixels comprising respectively three red, green and blue filters, the fourth sub-pixel comprising a filter in the infrared spectral band, the normalized measurement vector and the normalized theoretical vector being determined in a four-dimensional space.

[0043] According to one embodiment, the reference reflector is a mirror or a blank substrate identical to the substrate on which the layer is deposited.

[0044] According to another aspect, the invention relates to a system for determining a thickness map of a layer to be measured having a refractive index and disposed on a substrate, the system comprising: • a source exhibiting an emission spectrum over a spectral band of interest, the source being configured to illuminate said layer to be measured; • a camera comprising a detector and configured to image said layer to be measured on the detector, the detector comprising a plurality of pixels indexed i, each pixel being subdivided into at least three sub-pixels, each sub-pixel detecting an intensity, a pixel being characterized by at least three coordinates corresponding to the intensities detected by the at least three sub-pixels, the camera being configured to acquire a measurement image (Im) of said layer; and • a processing unit configured for: - to have a reference image of a reference reflector illuminated by the source with the camera, the reference reflector having a reference reflectivity, - determine a normalized measurement image equal, for each pixel, to the ratio of the intensities of the measurement image and the reference image, each pixel of the normalized measurement image having three so-called normalized coordinates defining a normalized measurement vector, - to have, for each pixel, a normalized theoretical vector} predetermined comprising at least three coordinates called normalized theoretical coordinates, a function of a theoretical thickness (e) of a theoretical layer (CT), said normalized theoretical vector having been determined from a predetermined theoretical reflectivity (Reth(k)) of said theoretical layer and the reference reflectivity, said theoretical layer having a refractive index equal to the refractive index (n(X)) of the layer to be measured; and - identify, for each pixel, the closest normalized theoretical vector (y™ ( / ) j to the normalized measurement vector, the associated thickness (em(i)) of said closest normalized theoretical vector corresponding to the thickness of the layer measured for the point of the layer imaged on the pixel.

[0045] According to one embodiment, the camera is further configured to acquire an image of the reference reflector illuminated by the source and to transmit said image of the reference reflector to the processing unit.

[0046] According to one embodiment the three filters of the camera correspond to three colored filters red, green and blue.

[0047] According to one embodiment, each pixel is subdivided into four sub-pixels, the first, second and third sub-pixels comprising respectively three red, green and blue filters, the fourth sub-pixel comprising a filter in the infrared spectral band, the normalized measurement vector and the normalized theoretical vector being determined in a four-dimensional space.

[0048] According to one embodiment, the source and the camera are arranged symmetrically with respect to a normal to the substrate.

[0049] The following description presents several embodiments of the device of the invention: these examples are not limiting to the scope of the invention. These embodiments present both the essential features of the invention and additional features related to the embodiments considered. Brief description of the drawings

[0050] The invention will be better understood and other advantages will become apparent upon reading the following description, given by way of non-limiting example, and with reference to the figures, among which:

[0051] [Fig. 1a] [Fig. 1a] represents an example of the sensitivity spectrum of a camera comprising three sub-pixels each presenting a filter on the blue, green and red spectral band;

[0052] [Fig.lb] [Fig.lb] represents an example of the sensitivity spectrum of a camera comprising four sub-pixels, each presenting a filter on the blue, green, red and infrared spectral band;

[0053] [Fig.2] [Fig.2] illustrates a method for determining a map in thickness of a layer to be measured according to the invention;

[0054] [Fig.3] [Fig.3] represents an example of theoretical reflectivity spectra of a silicon dioxide layer for different layer thicknesses;

[0055] [Fig.4] [Fig.4] represents an example of the evolution of the theoretical vector predetermined standard (symbolized by a point) as a function of thickness;

[0056] [Fig.5] [Fig.5] illustrates an example of thickness determination with the method according to the invention when the measurement is noisy;

[0057] [Fig. 6a] [Fig. 6a] represents an example of a layer thickness map to be measured in silicon dioxide on a silicon wafer produced using the method according to the invention;

[0058] [Fig.6b] [Fig.6b] represents the thickness measured in [Fig.6a] according to a section along the y-axis;

[0059] [Fig.7a] [Fig.7a] represents an example of a system for determining the thickness of a layer to be measured according to the invention;

[0060] [Fig.7b] [Fig.7b] represents another example of a system for determining the thickness of a layer to be measured according to the invention; and

[0061] [Fig.7c] [Fig.7c] represents another example of a system for determining the thickness of a layer to be measured according to the invention. DETAILED DESCRIPTION

[0062] The method 1000 for determining a thickness map of a layer to be measured CM according to the invention is schematically shown in [Fig.2]. The layer to be measured CM has a refractive index n(X) and is disposed on a substrate Sub.

[0063] The method 1000 is performed with a Cam camera comprising a detector Det, and which is configured to image the CM layer on the detector. The detector comprises a plurality of pixels Pi indexed i. Thus, an area of ​​the layer is associated with a point Pi of the detector, identified by its index i. The measured thickness map is therefore also indexed em(i).

[0064] Each pixel Pi of the camera is subdivided into at least three sub-pixels comprising respectively at least three filters FR, FG, FB. Each sub-pixel detects an intensity, a pixel thus being characterized by at least three coordinates corresponding to the intensities detected by the three sub-pixels.

[0065] For example, each pixel Pi is subdivided into three sub-pixels comprising respectively three associated filters FR, FG, FB, the three filters corresponding to three color filters red, green and blue. In this case, a pixel is characterized by three coordinates corresponding to the intensities detected by the three filters.

[0066] In another example, each pixel is subdivided into four sub-pixels, the first, second, and third sub-pixels comprising three filters: red, green, and blue, respectively, and the fourth sub-pixel having an FIR filter and SCIR sensitivity in the infrared spectral band. In this case, a pixel is characterized by four coordinates corresponding to the intensities detected by the four filters.

[0067] For a camera in which each pixel comprises three sub-pixels, the spectral sensitivity of the first sub-pixel (for example the SCR(X)) is denoted respectively by the red), SCG(X) the spectral sensitivity of the second subpixel (e.g., green), and SCB(X) the spectral sensitivity of the third subpixel (e.g., blue). These spectral sensitivities are defined over a spectral band of interest [Xmin and Xmax]. When the camera has four subpixels, the sensitivity of the fourth subpixel (e.g., an infrared subpixel) is called SCIR(X).

[0068] The method 1000 is carried out with a source S having a known emission spectrum SE(X) over the spectral band of interest. In one embodiment, the source S is a screen emitting white light. For example, the source S is a computer screen.

[0069] In what follows, the method 1000 is described for a pixel Pi subdivided into three sub-pixels. It can be easily adapted according to the number of sub-pixels.

[0070] The process 1000 comprises the steps described below.

[0071] In step A, the method 1000 comprises acquiring a measurement image Im of the CM layer illuminated by the source S, the image being acquired with the Cam camera. The acquisition is carried out so as to obtain, for each pixel Pi of the measurement image, a measurement vector Vm(i) having three coordinates Rm, G^ corresponding respectively to the three measurement intensities IRm, IGm, IBm detected through the colored filters FR, FG, FB. When the Cam camera includes four colored filters, the measurement vector has four coordinates corresponding respectively to the four measurement intensities detected through the four filters. In one example, the measurement image Im includes the entire CM layer to be measured, or in another example only a portion of the CM layer to be measured (for example, an area of ​​interest or a predetermined portion).

[0072] In step B of the process 1000, a reference image Iref of a reference reflector Rref is obtained, illuminated by the source S with the camera Cam. The reference reflector Rref has a predetermined (i.e., known) reference reflectivity RFref(X). For example, the reference reflector Rref is a mirror or a blank substrate identical to the substrate Sub on which the CM layer is deposited. For example, the CM layer is deposited on a silicon substrate Sub, and the reference reflector Rref is a blank silicon substrate.

[0073] For each pixel Pi of the reference image Iref, a reference vector Vref (O ​​cst) is obtained. The reference vector Vref (i) has three coordinates IRq IGq, ibq corresponding respectively to the three reference intensities of the reference image Iref detected through the colored filters of each pixel Pi.

[0074] In step C, the process 1000 comprises determining a normalized measurement image In equal, for each pixel Pi, to the ratio of the intensities of the image of The measurement Im and the intensities of the reference image Iref are measured. Each pixel of the normalized measurement image In has three normalized coordinates IRn, IGn, IBn, defining a normalized measurement vector Vn(i). The normalized measurement vector Vn(i) is defined by the ratio of a measurement vector Vm(i) and the reference vector Vref (J)' that is to say:

[0075]

[0076]

[0077]

[0078]

[0079]

[0080]

[0081]

[0082]

[0083]

[0084]

[0085] Vn(i)(r^, IGn, IBn} = For example, when the Cam camera has three filters FR, FG, FB, the three filters being red, green and blue color filters, the normalized measurement vector Vn(i) has three coordinates IRn, IGn, IBn defined respectively as follows, for each pixel Pi: IRnfy - IGn \i = IRm(i) iR^i) IGm(î) ~ÏG^ With, for a given pixel Pi indexed i: IRm(i) the intensity of the measurement image for the red sub-pixel of pixel Pi; IGm(i) the intensity of the measurement image for the green sub-pixel of pixel Pi; IBm(i) the intensity of the measurement image for the blue sub-pixel of pixel Pi; IRo(i) the intensity of the reference image for the red sub-pixel of pixel Pi; IGo(i) is the intensity of the reference image for the green subpixel of pixel Pi; and

[0086] IBo(i) the intensity of the reference image for the blue subpixel of pixel Pi.

[0087] The intensities detected by each pixel of the camera, respectively for the layer to be measured CM and for the reference reflector Rref, can be mathematically expressed with physical quantities characteristic of the different elements of the system, and therefore their ratios equal to IRn, IGn and IBn, in the following way:

[0088]

[0089] IRn = SCR(X). RFn£À). aSEtAïdl J, SCRiX) RFm( / À SE(A)dÀ .1. SCR(A). RrefQS aSE(À)dA SC Ri a). Rref(k)SE(X)dA

[0090] IGn = J > SCG( / ^ RF^À). aSEitylÀ fAflWX ]^SCG(^ RFM SEW* J. SCG(A) Rre aSE(À)dÀ ] ; SCGW R,.^. SEW A ■mm J

[0091] IBn = L SCB^À) RF^(À\ aSEiWÀ Amin l4 SCB(fy RF„&). SEW A f^MlX L SCB(A) uSEiÀ^dA j “SCBÇ). R^Ct. SEW A

[0092] With:

[0093] SCI? (A), SCG ( A ) and SC B ( A ) the spectral sensitivity of the camera for the red, green and blue sub-pixel, respectively.

[0094] RFm (A) the reflectivity of the layer to be measured CM;

[0095] RFref ( 2 ) the reflectivity of the reference reflector Rref;

[0096] SE (A) the emission spectrum of the source in arbitrary units

[0097] has the actual spectral intensity of the source S, related to the measurement method

[0098] and Xmax lower and upper bounds of the spectral band of interest. It is noted that in the calculation above, the ratio of the measurement vector Vm(i) and the reference vector Vref(i) makes it possible to disregard the real spectral intensity a of the source S, which simplifies for each pixel (i.e. for each sub-pixel of the pixel).

[0099] In another example, the Cam camera has four filters FR, FG, FB, FIR, the four filters being red, green, blue and infrared filters, the normalized vector Vn(i) has four coordinates IRn, IGn, IBn, IIRn, the red, green and blue coordinates being defined as above and the infrared coordinate being defined as follows:

[0100] HRn- ^.SClRiÀ) - RFM • aWW ], SCIR(a) ■ RrrM) • aSEt / MA \f SCIR&) ■ RFM • SE(Â}dÀ C"“sCIR(iï ■ ■ SEQ)dÀ

[0101] With, for a given pixel Pi:

[0102] SC IR (41 ), the spectral sensitivity of the camera for the red, green and blue sub-pixel, respectively;

[0103] IIRm(i) the intensity of the measurement image for the infrared subpixel of pixel Pi; and

[0104] IIR0(i) the intensity of the reference image for the infrared subpixel of pixel Pi.

[0105] In one embodiment, step B consists of acquiring an image of the reflector reference Rref illuminated by the source with the camera Cam. Thus, an image Iref is acquired before or after the acquisition of the image Im, by positioning the reference reflector in place of the layer to be measured.

[0106] In another embodiment, step B consists of loading a stored reference image. For example, a series of measurements, i.e., image acquisitions Im, can be performed for layers to be measured CM on a substrate Sub of the same type. Thus, a stored reference image Iref corresponding to the reference reflector Rref is used for a series of measurements. For example, a reference image Iref acquired with a silicon reference reflector Rref is acquired and stored. Subsequently, when the layer to be measured CM is placed on a silicon substrate Sub, step C is performed with the stored reference image Iref of the silicon reference reflector Rref, thereby reducing the time required to perform a series of measurements.

[0107] In step D, the method 1000 comprises, for each pixel Pi, the arrangement of a predetermined normalized theoretical vector V^ / n(i) comprising at least three coordinates, referred to as normalized theoretical coordinates IRth, IGth, IBth, which are functions of a theoretical thickness e of a theoretical layer CT. The theoretical layer CT has a refractive index equal to the refractive index n(X) of the layer to be measured CM. The normalized theoretical vector V^w(i) was determined from a predetermined theoretical reflectivity R^h(X) of the theoretical layer CT and the reference reflectivity RFref(X).

[0108] Figure 3 illustrates examples of predetermined theoretical reflectivity R^(X) as a function of wavelength X (in nm) for a silicon substrate alone (curve a), and for a theoretical CT layer of silicon dioxide deposited on a silicon substrate, for layer thicknesses of 50 nm (curve b), 500 nm (curve c), and 1000 nm (curve d). It is noted that the reflectivity varies significantly with thickness in the visible spectral range. These calculations are based on classical reflectivity models, such as the transfer matrix method.

[0109] Figure 4 illustrates the evolution of the calculated normalized theoretical vector (i). for a theoretical CT layer of silicon dioxide on a silicon substrate, as a function of the thickness e. In this figure, the thickness e of the theoretical CT layer varies between 0 (point V, with a normalized vector coordinates (1,1,1)) and 1400 nm (point W), and the refractive index n(X) is known. The theoretical reflectivity R^(X) is calculated from known formulas for a plurality of thicknesses, and the reflectivity of the reference reflector is known. The normalized theoretical vector V^Wn(i) is deduced. In the example in Figure 4, the normalized theoretical vector V^n(i) is represented by a point in three-dimensional space since it is obtained from three coordinates associated with a given thickness (when three filters are used).

[0110] In another embodiment, the normalized theoretical vector (i) is calculated with four coordinates (when four filters are determined) and is represented in four dimensions. The three- or four-dimensional representation makes it possible to limit the probability of crossover points.

[0111] In step E, the process 1000 includes identifying, for each pixel Pi, the normalized theoretical vector V^ / B(i) closest to the normalized measurement vector Vn(i). The thickness associated with the nearest normalized theoretical vector corresponds to the measured thickness em(i) of the CM layer for the point (or area) of the CM layer imaged on pixel Pi. In particular, the normalized measurement vector Vn(i) is compared to the normalized theoretical vector V^h / n(i) (for example, illustrated in (Figure 4). Since the normalized theoretical vector (i) is a function of a theoretical thickness e of a theoretical layer CT, by comparing the two vectors, it is possible to determine the thickness corresponding to the normalized theoretical vector closest to the normalized measurement vector Vn(z). Thus, by this method, a thickness is obtained, and it is therefore possible to perform a thickness map with a single image of the layer to be measured CM.

[0112] Advantageously, the invention eliminates the need for white balance by using normalization with a reference reflector Rref. The reference reflector Rref normalizes the image of the layer to be measured CM and thus eliminates the colorimetric non-uniformity of the source and the camera Cam. Therefore, normalization makes it possible to obtain precise thickness values.

[0113] Moreover, this thickness measurement method is as accurate as other methods used, such as spectral reflectivity, X-ray reflectivity, or ellipsometry, but much faster and less expensive. Indeed, the invention allows the thickness of a thin layer to be mapped across the entire field of view of the Cam camera in a single shot, and therefore in a very short time, with a resolution dependent on the field of view, the lens, and the pixel density of the Cam camera used.

[0114] In one embodiment, the coordinates of the normalized theoretical vector correspond, for each subpixel of a given pixel, to the calculated ratio of the intensity detected by the camera when the layer to be measured is replaced by the theoretical CT layer, and the intensity detected by the camera when the layer to be measured is replaced by the reference reflector.

[0115] Typically, the coordinates of the normalized theoretical vector V€th / n(i) determined at step D are determined by, respectively:

[0116] „ ] ; SCR{À) ■ ■ SEi^dÀ T Ty min J, SCÆ(2) • RFref(Â ) • SE{^

[0117] . ^SCGU) • • SEUW _ __i™_________________:___ J. SCGÇl) ■ RFK 4z) ■ SE(À)dÀ

[0118] f'WV , f J j SCB{à) ■ ■ $E(À)dA T Fr . - ----- d™ L SCB(Â) ■ RF^aX) ■ SEU^dX

[0119]

[0120] With: SCR^A) , SCG(À) and SCB(A) the spectral sensitivity of the camera for the red, green and blue subpixels, respectively;

[0121] Reth ( 2 ) the predetermined theoretical reflectivity of the theoretical layer;

[0122] RFref ( A ) the reflectivity of the reference reflector; .

[0123] SE(2 ) the emission spectrum of the source.

[0124] It is noted that the calculation of the normalized theoretical vector corresponds The calculation of the normalized measurement vector Vn(i) was simplified by the actual spectral intensity of the source S, replacing the reflectivity of the layer to be measured CM with the theoretical reflectivity R^A for a given thickness (for example, illustrated in Figure 3). The normalized theoretical vector (i) is therefore comparable to the normalized measurement vector Vn(i). However, it is necessary to know: the reference reflectivity RFref(X) of the reference reflector Rref(A), the emission spectrum of the source (in arbitrary units), and the spectral sensitivities of the sub-pixels (in arbitrary units).

[0125] In one embodiment, step D consists of loading a previously calculated normalized theoretical vector VetMn(i). In particular, the predetermined normalized theoretical vector V^n(i) is determined for a given refractive index n(X) (and therefore a given material). Thus, the normalized theoretical vector (i) is calculated for a given material and can be used for a series of thickness maps em(i) of layers to be measured CM in that same material. For example, a first normalized theoretical vector V1^(i) is calculated for a first material M1 having a first refractive index n1(X) and a second normalized theoretical vector V2^(i) is calculated for a second material M2 having a second refractive index n2(X).The first normalized theoretical vector Vl^ (i) is loaded to perform a series of thickness mapping determinations of layers to be measured CM in the first material ML. Then, the second normalized theoretical vector V2jh / n(i) is loaded to perform a series of thickness mapping determinations em(i) of layers to be measured CM in the second material M2.

[0126] In another embodiment, step D comprises a first substep consisting of loading refractive index values ​​onto the band of interest and a second substep consisting of determining the normalized theoretical vector V^ / h(i). For example, the first substep consists of identifying the material of the layer to be measured CM and loading the refractive index value n(X) corresponding to the material. Then, the second substep consists of determining the normalized theoretical vector V^ / h(i) based on the loaded refractive index value.

[0127] As indicated above, for each pixel, the determination of a thickness is obtained by comparing the normalized theoretical vector (illustrated in the figure 4) and the normalized measure vector Vn(l). To determine the closest vector, typically a Euclidean distance is determined between the two vectors.

[0128] As illustrated in [Fig. 4], the evolution curve of the normalized theoretical vector as a function of thickness comprises several "turns." The thickness corresponding to each pixel Pi therefore corresponds to the point on the turn closest to the point associated with the normalized measurement vector. If the measurement is noisy, and this noise is on the order of magnitude of the difference between two turns, the determined thickness may be erroneous due to the proximity of the incorrect turn.

[0129] Figure 5 illustrates an example of thickness determination when the measurement is noisy. Two turns, Sp1 and Sp2, are shown. When a measurement is noisy, the measured thicknesses may give an erroneous value. The mean plane between the two turns, Sp1 and Sp2, is denoted Pm. For example, point 50 (corresponding to the normalized measurement vector Vn(i) of a pixel Pi) is closer to turn Sp2 than to turn Sp1, thus suggesting that the thickness associated with point 50 is the thickness e2(50). However, the nearest neighbors PPV1 and PPV2 of point 50 both have thicknesses e(PPV1) and e(PPV2) located on turn Sp1. Thus, the thickness associated with point 50 is not the thickness e2(50) but is actually the thickness e1(50).

[0130] In order to correct these errors, in one embodiment, the process 1000 includes a control step F.

[0131] Step F includes a substep Fl for identifying, where appropriate, "abnormal" pixels exhibiting an abnormal layer thickness by comparison with the thicknesses of at least two nearest neighbors of the pixels. For example, the thickness determined by the method according to the invention is compared to thicknesses determined for nearest neighbors of the pixel, and when the difference between the determined thickness and the thicknesses of the nearest neighbors exceeds a threshold, the determined thickness is identified as abnormal.

[0132] Next, step F includes a substep F2 of defining a range of possible thicknesses [emin; emax], for each of the identified "abnormal" pixels, from the thicknesses of at least two nearest neighbors of the identified pixels.

[0133] Thus, the range makes it possible to limit the thickness to a so-called normal value, that is to say without sudden variation.

[0134] Furthermore, step F includes a substep F3 for determining, for the pixels identified in step F1, a corrected thickness based on the identification of a normalized theoretical vector Veiltjn(i) whose corresponding thickness is within the range of possible thicknesses [emin; emax]. Thus, the thickness is determined in the loop within the range of possible thicknesses [emin; emax].

[0135] Advantageously, this method makes it possible to detect and correct erroneous thickness values. Indeed, the thickness of a layer is generally continuous. Abrupt variations (for example, of several tens of nm) between two consecutive pixels are unlikely.

[0136] Figure 6a illustrates an example of thickness maps em(i) of a silicon dioxide CM layer obtained using the method 1000 according to the invention. A point on the layer is located along two axes X and Y (expressed in mm), and the thickness of the layer determined by the method according to the invention is expressed in nm. The pixels of the image have been transformed to their corresponding positions on the layer. Figure 6b illustrates an example of a thickness measurement of a silicon dioxide CM layer taken in cross-section along the Y-axis (line 60 of Figure 6a). A thickness of 590 nm is noted at point O in Figure 6b. The thickness of the CM layer was measured independently with a spectroscopic ellipsometer at the center O of the CM layer, and a value very close to 590 nm was measured.

[0137] Furthermore, the method according to the invention makes it possible to measure the em(i) thicknesses of the CM layer regardless of its surface roughness, whereas it is not possible to measure the thickness of a thin film with known techniques using reflectivity (e.g., spectral reflectivity, X-ray reflectivity, or ellipsometry) if the surface is rough. This is because these reflectivity measurements are based on measuring interference fringes, i.e., oscillations, the amplitude of which decreases sharply with roughness. In contrast, roughness has no impact on colorimetry. This robustness of the method according to the invention stems from the fact that the spectral ranges of the colors and / or infrared are very broad.

[0138] According to another aspect, the invention relates to a Sys system for determining a thickness map em(i) of a layer to be measured CM having a refractive index n(X) and disposed on a substrate Sub, enabling the implementation of the method 1000 described above. Examples of Sys systems are illustrated in Figures 7a, [Fig.7b] and [Fig.7c].

[0139] The Sys system includes a source S exhibiting an emission spectrum SE(X) over a spectral band of interest, the source S being configured to illuminate the layer to be measured CM. In one example, the source S is a screen emitting white light. For example, the source is a computer monitor.

[0140] The Sys system also includes a Cam camera comprising a Det detector and configured to image the CM layer on the Det detector, the detector comprising a plurality of Pi pixels indexed i, each pixel being subdivided into at least three sub-pixels comprising respectively at least three filters FR, FG, FB, each sub-pixel detecting an intensity, a pixel being characterized by at least three coordinates corresponding to the intensities detected by the at least three sub-pixels.

[0141] The Cam camera is configured to acquire a measurement image Im of the CM layer. In one embodiment, the Cam camera comprises three colors: blue, green and red. In particular, each pixel of the camera's detector Det is subdivided into three sub-pixels, each containing three filters: FR, FG, and FB. These three filters correspond to three colored filters: red, green, and blue. In another example, each pixel is subdivided into four sub-pixels. The first, second, and third sub-pixels each contain three filters: red, green, and blue, respectively. The fourth sub-pixel contains an FIR filter in the infrared spectral band. In this case, a pixel is characterized by four coordinates corresponding to the intensities detected by the four filters. In one example, the camera Cam is a "tri-spectral" color camera equipped with a lens. The lens is focused on the surface of the layer to be measured, CM, which is illuminated by the source S.

[0142] In the embodiment illustrated in [Fig. 7a], the source S and the camera Cam are arranged symmetrically with respect to the normal N to the substrate Sub. In this configuration, the source S illuminates the layer to be measured CM and the camera Cam directly images the layer to be measured CM.

[0143] In the example of Figures 7b, the source S is disposed on a normal to the substrate, and the system further includes a semi-reflective device LS configured to send the light reflected by the layer to be measured CM to the camera Cam.

[0144] In the example of [Fig.7c], the Cam camera is arranged on a normal to the substrate, and the system further includes a semi-reflective device LS configured to send the light emitted by the source to the layer to be measured CM.

[0145] For example, the semi-reflective device LS can be a beam splitter or a beam splitter cube. In the configuration of [Fig. 7c], the lens of the camera Cam is focused on the surface of the layer to be measured CM, which is illuminated by the source S via the semi-reflective device LS. Alternatively ([Fig. 7b]), the source S illuminates the layer to be measured CM, which is imaged by the camera Cam via the semi-reflective device LS.

[0146] In addition, the Sys system includes a PU processing unit configured to implement process 1000. The PU processing unit is connected to the Cam camera.

[0147] The processing unit PU is configured to have a reference image Iref of a reference reflector Rref illuminated by the source with the camera Cam, the reference reflector Rref having a reference reflectivity RFref(X).

[0148] In addition, the processing unit PU is configured to determine a normalized measurement image In equal, for each pixel, to the ratio of the intensities of the measurement image and the reference image, each pixel of the normalized measurement image having three so-called normalized coordinates defining a normalized measurement vector Vw(i).

[0149] Furthermore, the processing unit PU is configured to have, for each pixel Pi, a predetermined normalized theoretical vector comprising at least three coordinates called normalized theoretical coordinates, a function of a theoretical thickness e of a theoretical layer CT, the normalized theoretical vector Veth / nd) having been determined from a predetermined theoretical reflectivity Reth(d of the theoretical layer CT and the reference reflectivity, the theoretical layer CT having a refractive index equal to the refractive index n(X) of the layer to be measured CM.

[0150] In one embodiment, each pixel is subdivided into four sub-pixels, the first, second and third sub-pixels comprising three red, green and blue filters respectively, the fourth sub-pixel comprising a filter in the infrared spectral band, the normalized measurement vector Vn(i) and the normalized theoretical vector being determined in a four-dimensional space.

[0151] In addition, the processing unit PU is configured to identify, for each pixel, the normalized theoretical vector V^''n(i) Ie closest to the normalized measurement vector V„(i), the thickness associated with the nearest normalized theoretical vector (i) Ie corresponding to the thickness of the measured layer CM for the point of the layer imaged on the pixel.

[0152] Because the calculation of the Euclidean distance is very simple and each pixel is independent of the others, it is possible to parallelize the calculation of finding the theoretical vector closest to the measured vector of each pixel, using a GPU (Graphical Processing Unit). This greatly reduces the computation time.

[0153] The processing unit (PU) can be implemented using hardware, software, and / or a combination thereof. For example, hardware devices can be implemented using processing circuits such as, but not limited to, a processor, a central processing unit (CPU), a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field-programmable gate array (FPGA), a system-on-a-chip (SoC), a programmable logic unit, a microprocessor, or any other device capable of responding to and executing instructions in a defined manner. The software can include a computer program, program code, instructions, or a combination thereof, to provide instructions or configure, independently or collectively, a hardware device to operate as desired.The computer program and / or program code may include program instructions or computer-readable code, software components, software modules, data files, data structures, and / or the like, capable of being implemented by one or more hardware devices, such as one or more hardware devices. When a hardware device is a device... A computer processing unit (e.g., a CPU, controller, ALU, digital signal processor, microcomputer, microprocessor, etc.) can be configured to execute program code by performing arithmetic, logical, and input / output operations, according to the program code. Each unit may also include one or more storage devices. The storage device(s) may be tangible or non-transient computer-readable storage media, such as random access memory (RAM), read-only memory (ROM), a permanent mass storage device (such as a disk drive), a semiconductor (e.g., NAND flash), and / or any other similar data storage mechanism capable of storing and recording data.The storage device(s) can be configured to store computer programs, program code, instructions, or a combination thereof, for one or more operating systems and / or to implement the example embodiments described herein. The computer programs, program code, instructions, or a combination thereof, can also be loaded from a separate computer-readable storage medium into the storage device(s) and / or one or more computing devices using a drive mechanism. Such a separate computer-readable storage medium may include a USB (Universal Serial Bus) flash drive, a USB flash drive, a Blu-ray / DVD / CD-ROM drive, a memory card, and / or other similar computer-readable storage media.

[0154] In one embodiment, the Cam camera is further configured to acquire an image of the reference reflector Rref illuminated by the source and to transmit the image of the reference reflector Iref to the PU processing unit. The PU processing unit can then use the image of the reference reflector Rref to determine the thickness map em(i) of the layer to be measured CM.

[0155] Although the invention has been illustrated and described in detail using a preferred embodiment, the invention is not limited to the disclosed examples. Other variations can be deduced by a person skilled in the art without departing from the scope of protection of the claimed invention.

Claims

Demands

1. Method (1000) for determining a thickness map (em(i)) of a layer to be measured (CM) having a refractive index (n(X)) and disposed on a substrate (Sub), with: a camera (Cam) comprising a detector (Det) and configured to image the layer on the detector, the detector comprising a plurality of pixels (Pi) indexed i, each pixel being subdivided into at least three sub-pixels comprising respectively at least three filters (FR, FG, FB), each sub-pixel detecting an intensity, a pixel being characterized by at least three coordinates corresponding to the intensities detected by the at least three sub-pixels, and a source (S) exhibiting an emission spectrum (SE(X)) over a spectral band of interest, the process comprising the steps of: To acquire a measurement image (Im) of the layer (CM) illuminated by the source with the camera, B. to have a reference image (Iref) of a reference reflector (Rref) illuminated by the source, with the camera, the reference reflector having a predetermined reference reflectivity (RFref(X)), C determine a normalized measurement image (Inorm) equal, for each pixel, to the ratio of the intensities of the measurement image and the reference image, each pixel of the normalized measurement image having at least three so-called normalized coordinates defining a normalized measurement vector ( / ))' To have, for each pixel, a predetermined normalized theoretical vector (y^') comprising at least three coordinates called normalized theoretical coordinates, a function of a theoretical thickness (e) of a theoretical layer (CT), said normalized theoretical vector having been determined from a predetermined theoretical reflectivity (Reth(X)) of said theoretical layer and the reference reflectivity, said theoretical layer having a refractive index equal to the refractive index (n(X)) of the layer to be measured, E identify, for each pixel, the normalized theoretical vector (y^, Ie Closest to the normalized measurement vector ^y the associated thickness (em(i)) said normalized theoretical vector closest corresponding to the measured thickness of the layer for the point of the layer imaged on the pixel.

2. Method according to claim 1, wherein each of the at least three coordinates of the normalized theoretical vector (y®(i))' is determined by, respectively: I, SCRQ) ■ ■ SEVW. j ni 'Tm_____'_______ thJll fWr J, - RFF€AA) ■ SEîMl ■Amin „ J. -SEWA .... * ^th / n ~ H»»-'- J ■ SCG(â) ■ RFrt âa) ■ SE^A^dÀ p J, SCB(^-Rt^)-SE(À)dÀ lDthIn ~ d— .1, SCBU) • RF„. / (.$ ■ SEU^ÎÀ Anii» with: IRZ t -, IGZ , . ZBf, . : coordinates of the normalized vector 11A x ^th / n ' * ^th / n thm ' \ l ) SCR ( À ), SCG ( À ) and SCB ( À ) camera sensitivity for the at least three sub-pixels respectively, ( z ) predetermined theoretical reflectivity of the layer theoretical thickness e RFref(A) reflectivity of the reference reflector SE (A) emission spectrum of the source, Xmin and Xmax limits of the spectral band of interest.

3. A method according to any one of claims 1 or 2, wherein the three filters correspond to three colored filters: red, green, and blue.

4. A method according to any one of the preceding claims, wherein step D consists of loading a previously calculated normalized theoretical vector.

5. A method according to any one of claims 1 to 3, wherein step D comprises a first substep consisting of loading refractive index values ​​onto the band of interest and a second substep consisting of determining said normalized theoretical vector.

6. A method according to any one of the preceding claims, wherein step A consists of acquiring an image of the source-illuminated reference reflector with the camera.

7. A method according to any one of claims 1 to 5, wherein step B consists of loading a stored reference image.

8. A method according to any one of the preceding claims further comprising a control step F comprising the substeps of: F1 identifying, where appropriate, pixels exhibiting an abnormal layer thickness by comparison with thicknesses of at least two nearest neighbors of said pixels, F2 for each of the identified pixels, defining a range of possible thicknesses ([emin; emax]) from said thicknesses of said nearest neighbors of said identified pixels, F3 determining, for said pixels identified in step F1, a corrected thickness from the identification of a normalized theoretical vector whose corresponding thickness is within said thickness range.

9. A method according to any one of the preceding claims, wherein each pixel is subdivided into four sub-pixels, the first, second and third sub-pixels comprising three red, green and blue filters respectively, the fourth sub-pixel comprising a filter in the infrared spectral band, the normalized measurement vector (V„(i)) and the normalized theoretical vector being determined in a four-dimensional space.

10. A method according to any one of the preceding claims, wherein the reference reflector (Rref) is a mirror or a blank substrate identical to the substrate on which the layer is deposited.

11. A system (Sys) for determining a thickness map (em(i)) of a layer to be measured (CM) having a refractive index (n(X)) and disposed on a substrate (Sub), the system (Sys) comprising: • a source (S) having an emission spectrum (SE(X)) over a spectral band of interest, the source being configured to illuminate said layer to be measured; • a camera (Cam) comprising a detector (Det) and being configured to image said layer to be measured on the detector, the detector comprising a plurality of pixels (Pi) denoted i, each pixel being subdivided into at least three sub-pixels, each sub-pixel detecting an intensity, one pixel being characterized by at least three coordinates corresponding to the intensities detected by the at least three sub-pixels, the camera being configured to acquire a measurement image (Im) of said layer (CM); and • a processing unit (PU) configured to: - have a reference image (Iref) of a reference reflector (Rref) illuminated by the source with the camera, the reference reflector having a reference reflectivity (RFref(X)), - determine a normalized measurement image equal, for each pixel, to the ratio of the intensities of the measurement image and the reference image, each pixel of the normalized measurement image having three so-called normalized coordinates defining a normalized measurement vector (Vn^)) - have, for each pixel, a predetermined normalized theoretical vector (y^ / (j)) comprising at least three coordinates called normalized theoretical coordinates,function of a theoretical thickness (e) of a theoretical layer (CT), said normalized theoretical vector having been determined from a predetermined theoretical reflectivity (7?^(X)) of said theoretical layer and the reference reflectivity, said theoretical layer having a refractive index equal to the refractive index (n(X)) of the layer to be measured; and - identify, for each pixel, the normalized theoretical vector Ie closest to the normalized measurement vector, the associated thickness (em(i)) of said nearest normalized theoretical vector corresponding to the thickness of the layer measured for the point of the layer imaged on the pixel.

12. System according to claim 11, wherein the camera is further configured to acquire an image of the reference reflector illuminated by the source and to transmit said image of the reference reflector to the processing unit.

13. System according to claims 11 or 12, wherein the three filters of the camera correspond to three colored filters red, green and blue.

14. System according to claim 11 or 12 wherein each pixel is subdivided into four sub-pixels, the first, second and third sub-pixels comprising three red, green and blue filters respectively, the fourth sub-pixel comprising a filter in the infrared spectral band, the normalized measurement vector and the normalized theoretical vector being determined in a four-dimensional space.

15. System according to any one of claims 11 to 14, wherein the source and the camera are arranged symmetrically with respect to a normal to the substrate.

16. System according to any one of claims 11 to 14, wherein the source or camera is disposed on a normal to the substrate, and wherein the system further comprises a semi-reflective device (LS) configured to respectively send the light reflected by the layer to be measured to the camera or to send the light emitted by the source to the layer to be measured.

17. System according to any one of claims 11 to 16 wherein the source is a screen emitting white light.

Images