Device for optically measuring the thickness of an opaque object and method for calibrating such a device

Abstract

Method for calibrating a device (14) for measuring the thickness (d) of an opaque object (10), wherein the device comprises a first OCT sensor (20) having a first optical coherence tomograph (22), a first scanning device (26) and a first collecting optic (38), and a second OCT sensor (20') having a second optical coherence tomograph (22'), a second scanning device (26') and a second collecting optic (38'), wherein the first OCT sensor (20) is arranged along a z-direction relative to the second OCT sensor (20') such that the first collecting optic (38) is arranged coaxially to the second collecting optic (38'), and wherein the method comprises the following steps: a) the first optical coherence tomograph (22) generates a measuring light beam (24); b) the first scanning device (26) is controlled such that the measuring light beam (24) is focused into a first location (70), where there is no opaque object at the location (70); c) the second scanning device (26') is controlled so that it scans an area around the first location (70) so that the measuring light beam (24) generated in step a) is directed at least temporarily by the second scanning device (26') onto a detector (56') of the second OCT sensor (20'); d) taking into account intensity signals measured by the detector (56') in step c), a first coordinate system used by the first OCT sensor (20) is aligned with a second coordinate system used by the second OCT sensor (20'); e) From the comparison in step d), a correction rule is generated, according to which points in the first coordinate system can be uniquely and reversibly assigned to points in the second coordinate system.

Description

BACKGROUND OF THE INVENTION 1. Field of the invention

[0001] The invention relates to a device for optically measuring the thickness of opaque objects using two OCT distance sensors arranged opposite each other and coaxially. The object can be, for example, a wafer, a battery film, or an opaque layer in a multilayer structure that also includes transparent layers. The invention particularly relates to a method for calibrating such a device. 2. Description of the state of the art

[0002] In metrology, a common task is to measure the thickness of opaque objects that have a primarily plate-, disc-, or foil-like geometry. Typical examples of this type are wafers or battery foils. Wafers are circular or square discs approximately one millimeter thick, used as substrates for integrated circuits, micromechanical components, or photoelectric coatings. Battery foils are required for the production of battery cells and usually consist of a thin metal foil coated with an electrolyte paste.

[0003] An object is considered opaque if it is impervious to the measuring light of the optical measuring device. An object is considered opaque if less than 1% of the incident light can pass through it.

[0004] Thickness is preferably measured optically and therefore without contact, since the surfaces of objects are often sensitive. Various methods for optical thickness measurement are known in the prior art.

[0005] CN 108955549 A discloses a method for measuring the thickness of transparent objects, in which two chromatic confocal and coaxially aligned measuring devices measure the distance to the object from opposite sides.

[0006] WO 2022 / 245195 A1 describes a method for thickness measurement in which two measuring devices project a moiré pattern onto opposite sides of the object being measured from different angles.

[0007] US Patent 10,563,975 B1 describes a method for measuring the thickness of transparent objects, in which two OCT sensors take distance measurements from opposite sides.

[0008] DE 10 2019 114 167 A1 describes a similar method in which, for measuring the thickness of an opaque layer supported by a substrate, two coaxially arranged OCT sensors each direct a measuring light beam from opposite sides onto the object being measured.

[0009] WO 2023 / 224222 A1 describes a method for measuring the alignment of battery plates of secondary batteries, in which an OCT sensor is arranged on each side of the secondary battery. Both sensors perform an alignment check simultaneously.

[0010] All the methods and devices described above have in common that flat objects, such as wafers, cannot be measured quickly. To perform measurements at multiple locations, these known methods require relative movement between the device and the object. Such relative movement necessitates highly precise and therefore expensive positioning systems and also requires a relatively long measurement time, as comparatively large masses must be moved. Measurement time is particularly important when the measurement is to be performed as part of quality control during a manufacturing process. For example, in wafer production, it is generally insufficient to perform thickness measurements only on a sample basis. Ideally, each manufactured wafer is checked to ensure it meets the geometric specifications.

[0011] A device that enables the rapid measurement of even large wafers is known from DE 10 2022 104 416 A1 of the applicant. The OCT sensor disclosed therein includes, in addition to the actual optical coherence tomograph, a scanning device that deflects the measuring light beam generated by the coherence tomograph in two spatial directions and directs it onto the wafer via a converging optic. This allows the measuring light beam to scan the entire surface of even large wafers very quickly. If the light propagation in the reference arm of the coherence tomograph is suppressed by means of a glare reduction device, this OCT sensor can also be used to measure the thickness of transparent objects.

[0012] To measure the thickness of opaque objects, two such OCT sensors can be positioned opposite each other along a z-direction so that the optical axes of the collecting lenses are coaxial. For thickness measurement, each sensor measures the distance to the same location (x, y) in an xy-plane. From the distances measured by each sensor, the z-coordinates of the object's surfaces facing the sensors are determined. Since the distance between the sensors is constant, a thickness profile of the object can be determined from these distances—either after a calibration measurement with an object of precisely known thickness or after measuring the distance between the OCT sensors.

[0013] However, it has become apparent that the measurement accuracy achievable with this known device occasionally falls short of expectations. Although the OCT sensors can measure distances to within a few micrometers, the measurement errors are sometimes significantly larger, particularly when the thickness of the object changes considerably over very short distances. SUMMARY OF THE INVENTION

[0014] The object of the invention is to increase the measurement accuracy in such devices when measuring the thickness of flat, opaque objects.

[0015] This problem is solved by a method for calibrating a device for measuring the thickness of an opaque object, wherein the device comprises a first OCT sensor having a first optical coherence tomograph, a first scanning device and a first collecting optic, and a second OCT sensor having a second optical coherence tomograph, a second scanning device and a second collecting optic, wherein the first OCT sensor is arranged along a z-direction relative to the second OCT sensor such that the first collecting optic is arranged coaxially to the second collecting optic, and wherein the method comprises the following steps: a) the first optical coherence tomograph generates a measuring light beam; b) the first scanning device is controlled such that the measuring light beam is focused to a first location, where there is no opaque object at that location; c) the second scanning device is controlled in such a way that it scans an area around the first location, so that the measuring light beam generated in step a) is directed at least temporarily by the second scanning device onto a detector of the second OCT sensor; d) taking into account intensity signals measured by the detector in step c), a first coordinate system used by the first OCT sensor is aligned with a second coordinate system used by the second OCT sensor; e) From the comparison in step d), a correction rule is generated, according to which points in the first coordinate system can be uniquely and reversibly assigned to points in the second coordinate system.

[0016] The invention is based on the understanding that the measurement errors that occur are primarily caused by the fact that the coordinate systems used by the two OCT sensors do not exactly match. This is mainly due to tolerances in the mounting of the OCT sensors. Other contributing factors can include manufacturing tolerances, thermal effects, and even minute changes in position resulting from contact between the OCT sensors. In the most general case, the lack of match consists of the coordinate systems being shifted, rotated, and / or compressed / stretched relative to each other.

[0017] If the coordinate systems do not match, the same coordinates (x, y) describe different locations in the xy-plane where the object is measured. The two OCT sensors then measure the distances from which the thickness is derived at laterally offset locations. In other words, the measurement points in this case are located on the object's surface that are not exactly opposite each other along the z-direction. This leads to a distorted thickness profile. Tests have shown that the lateral offset can be up to one millimeter without calibration.

[0018] According to the invention, the device is therefore calibrated at least before initial commissioning, preferably regularly, and ideally after every contact or vibration. The result of the calibration is a correction formula with which the coordinates of one OCT sensor can be uniquely and reversibly assigned to the coordinates of the other OCT sensor.

[0019] Alternatively, calibration could be performed using a calibration standard. This could be transparent, for example, so that structures within it with a reflective profile that contrasts with the surroundings are visible from both sides. Alternatively, a transparent calibration standard with structures that can be detected from both sides by OCT sensors using thickness or topography measurements can be used. The use of an opaque calibration standard is also an option if the exact position of the structures on one side relative to the structures on the other is known. Using a calibration standard allows for simultaneous relative calibration between the two OCT sensors and absolute calibration in xy and / or z. However, opaque calibration standards, at least, are expensive to manufacture and require careful storage.Above all, however, introducing a calibration standard into the beam path is very time-consuming. Since the calibration according to the invention does not require a calibration standard, it can be carried out with very little effort and in a very short time (usually just a few minutes).

[0020] In the simplest case, both OCT sensors continue to use their coordinate systems unchanged after calibration, but convert the coordinates into each other. If, for example, one OCT sensor performs a distance measurement at the coordinate (x1, y1), the correction formula generated in step e) may result in the corresponding location being described by the coordinate (x2, y2) for the other OCT sensor, where, in general, x1 ≠ x2 and y1 ≠ y2. This ensures that the measurement points are exactly opposite each other along the z-direction.

[0021] The coordinate systems used by the OCT sensors are primarily important for controlling the respective scanning device, as this variably determines where the measurement light beam is deflected. In most cases, the scanning devices contain one or preferably two swiveling mirrors, prisms, or other movable deflection elements. Each coordinate in the xy-plane is then assigned, for example, a pair of swivel angles, whereby this assignment can be defined by lookup tables or polynomials. The correction formula generated in step e) can then alternatively be used not for a coordinate conversion, but for a conversion of the aforementioned assignment. In this case, no coordinate conversion takes place; instead, the coordinates are converted differently into swivel angles of the scanning devices.

[0022] For the measuring light generated by the first OCT sensor to reach the second OCT sensor, no opaque objects may be present in the measuring plane. While transparent objects in the measuring plane could theoretically be tolerated, they can cause undesirable deflections due to light refraction. Therefore, the space between the two OCT sensors is preferably free of any objects. The exception is objects with openings through which the measuring light can pass unhindered.

[0023] For calibration, the first OCT sensor is used solely as a light source to generate a measurement beam, and the second OCT sensor as a detector to measure the intensity of this beam. The interference characteristic of OCT sensors, but unnecessary or even undesirable for calibration, between measurement light reflected from a reference arm and measurement light reflected from the object, could only occur if a transparent object is positioned in the measurement plane during calibration and the light source of the second OCT sensor also generates measurement light. To prevent such interference from disrupting the intensity measurement, the light source of the second OCT sensor is preferably switched off during calibration.If switchable dimming devices are present in the reference arms of the optical coherence tomographs, these can alternatively or additionally prevent the propagation of light in the reference arms during calibration.

[0024] When scanning the environment in step c), the points passively scanned by the second scanning device must be positioned so closely together that, at least in one position of the second scanning device, the measurement light beam generated by the first OCT sensor can reach the detector via the second collecting optics and the second scanning device. Preferably, the maximum distance a between two adjacent points passively scanned by the second OCT sensor is a < d, where d is the diameter of the passive measurement spot, which can be on the order of 10 µm to 60 µm.

[0025] If steps b) to d) are performed for only one location, it is impossible to distinguish whether the coordinate systems are translated, rotated, or scaled relative to each other. Therefore, steps b) to d) are preferably repeated for a second location that differs from the first. Then, in step d), a relative rotation, translation, and linear scaling of the two coordinate systems can be determined. Preferably, the two locations lie in a plane that runs in an xy-plane of one of the two coordinate systems. This simplifies the calculation of the correction formula in step e). However, other locations are also acceptable, as long as they differ with respect to the x- and / or y-coordinates of one of the two coordinate systems.

[0026] Additionally, for more precise calibration, steps b) to d) can be repeated for a third or further locations. Using a third location, for example, allows differentiation between a rotation and different scaling in the x and y directions.

[0027] In order to determine which position of the second scanning device corresponds to the position of the first scanning device for a specific location in the xy-plane, it can be determined in step d) at which control of the second scanning device the intensity signal measured by the detector is at its maximum.

[0028] Since the smallest unit of control for the scanning device's swivel angle results in a displacement of the measurement spot in the focal plane that is smaller than the passive measurement spot itself, it is possible to select the scanning grid such that the passive measurement spots overlap significantly. This enables a very precise determination of the active measurement spot's location, as it does not simply determine a single non-zero intensity value at a single location, but rather an intensity distribution that represents the respective overlap at each grid point between the active measurement spot, formed by the focus of the measuring light beam, and the passive measurement spot.

[0029] The area scanned by the second scanning device in step c) can encompass the entire xy-measurement plane. However, scanning with a fine scanning grid is time-consuming. If the location of the measuring light beam is already known quite accurately, the area to be scanned can be significantly smaller, covering, for example, an area of ​​only a few square centimeters around the expected location. Calibration can then be performed much more quickly.

[0030] If, however, the approximate location of the measuring light beam is unknown, a rough estimate of its location can be obtained by dividing step c) into a first sub-step and a second sub-step. In the first sub-step, the area surrounding the location is scanned with a first sampling grid. In the second sub-step, a smaller area surrounding the location is scanned with a second sampling grid that is finer than the first.

[0031] The first step, in which the surrounding area (e.g., the entire measurement plane or a larger section thereof) is scanned with the coarser scanning grid, serves only to roughly locate the active measurement spot. Once its approximate location is known, the second step is performed, in which a closer area around this approximately determined location is scanned with a scanning grid that is fine enough to determine the location of the first measurement spot with sufficient accuracy.

[0032] This division into two sub-steps with different scanning grids allows for very rapid calibration, even when the focal point of the measuring light beam is not even approximately known, and therefore the entire xy-measuring plane must be scanned in the first sub-step. Such a situation can arise, for example, after initial installation or modification of the measuring device. Once the focal point of the measuring light beam is roughly known, the first sub-step can potentially be omitted in subsequent calibrations.

[0033] The first step with a coarse scanning grid also makes the preceding computational reflection of the coordinate systems depending on the symmetry of the arrangement, mentioned at the end of section 3, unnecessary, since this results directly from the correction rule in step e).

[0034] To align the coordinate systems of the two OCT sensors, an xy-calibration should be performed for both the first and second OCT sensors prior to the calibration described here. This calibration assigns control commands for the first and second scanning devices to coordinates in the first and second coordinate systems, respectively. Only then is it ensured that the correction rule generated in step e) is valid not only for the location(s) of the calibration measurement but across the entire xy-plane. How such an xy-calibration can be performed using a calibration standard is described in detail in the applicant's unpublished German application 10 2024 109 466.4, filed on April 4, 2024.

[0035] The invention also relates to a calibrated device for measuring the thickness of an opaque object, wherein the device comprises: a first OCT sensor, which has a first optical coherence tomograph, a first scanning device and a first collecting optic, a second OCT sensor, which has a second optical coherence tomograph, a second scanning device and a second collecting optic, wherein the first OCT sensor is arranged along a z-direction relative to the second OCT sensor such that the first collecting optic is arranged coaxially to the second collecting optic, the first scanning device can be controlled based on a first coordinate system and the second scanning device can be controlled based on a second coordinate system, and The first coordinate system and the second coordinate system are transformed or can be transformed into each other by a transformation obtained by calibration according to the inventive method described above.

[0036] To measure an opaque object of a given thickness with the device, the first and second OCT sensors are advantageously positioned such that a measurement beam from the first OCT sensor is approximately focused on a first surface of the object facing it, and a measurement beam from the second OCT sensor is approximately focused on a second surface of the object facing it. In this context, "focused" means that the waist of the measurement beam lies approximately on the surface. To ensure this, a distance is set between the two OCT sensors. After each adjustment of the distance, a calibration is advantageously carried out according to the inventive method described above.

[0037] When OCT sensors with a large depth of field are used, the method according to the invention also works well even if the focal planes of the two OCT sensors do not exactly coincide. In general, it still works sufficiently well if the focal planes are less than one Rayleigh length apart.

[0038] To perform a thickness measurement with this device, a trigger signal can be applied to both the first and second OCT sensors, initiating a precisely simultaneous measurement of the distance to the object. This ensures that vibrations of the object or the device do not affect the locations scanned by the two optical coherence tomographs. If the distance measurements for a coordinate (x, y) are performed successively, vibrations could lead to measurements being taken at different coordinates.

[0039] The trigger signal can initiate not only a single measurement, but also a predefined sequence of measurements at different locations. However, the more individual measurements are performed consecutively, the greater the potential for the timing of these measurements to differ, which is why the sequences should not be too long. Therefore, measuring a larger area, such as that required for a wafer, can be divided into several sequences, each triggered individually by its own corresponding signal.

[0040] The trigger signal can be generated by an external control unit and sent to both OCT sensors. Alternatively, one of the two OCT sensors can generate the trigger signal and transmit it to the other OCT sensor.

[0041] The locations that are successively scanned in the xy-plane by the two OCT sensors can, for example, lie on a straight line, a rectangle, a circle or a spiral. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Exemplary embodiments of the invention are explained in more detail below with reference to the drawings. These show: Fig. 1 a wafer to be measured in a perspective and not to scale representation; Fig. 2 a measuring device according to the invention in an exemplary embodiment in a schematic representation; Fig. 3 important parts of a scanning device that are part of the system in the Fig. The measuring device shown in Figure 2 is in a simplified perspective view; Fig. 4. A cross-section through a wafer using measuring light beams for thickness measurement; Fig. 5 that are in the Fig. 2 Measuring device shown during the scanning of a first location during the calibration process; Fig. 6 the xy-plane (measuring plane) with two locations onto which a measuring light beam generated by the first OCT sensor is directed successively; Fig. 7a a section of the measurement plane which is passively scanned in a first step using a coarse scanning grid; Fig. 7b the excerpt from the Fig. 7a, while in a second sub-step passive scanning is performed with a fine scanning grid; Fig. 8 a scan line that leads to the one in the Fig. The pattern of passive sampling points shown in 7a leads to this; and Fig. 9 who are in the Fig. 2 Measuring device shown during the scanning of a second location during the calibration process. DESCRIPTION OF PREFERRED EXAMPLES 1. Surveying of wafers

[0043] The Fig. Figure 1 shows a wafer 10 in a perspective, non-essential view. In the illustrated embodiment, the wafer 10 has the shape of a right circular cylinder, with the thickness exaggerated. Real wafers 10, for example, have a diameter of up to 300 mm, while the thickness is only about 1 mm. Occasionally, wafers 10 with a square, rather than circular, surface are also used.

[0044] It is assumed here that the wafer consists of an opaque material or a material contained within the Fig. 1. It has an opaque coating indicated by hatching.

[0045] The ideal circular cylindrical shape of the wafer 10 is indicated by dashed lines 12. Due to manufacturing tolerances, deviations from this ideal shape may occur, which are shown in the Fig. 1 are exaggerated. To determine these deviations, wafer 10 must be measured. By measuring the topography of both wafer surfaces, all common geometric specifications of the wafer can be derived. These specifications include, among others, the TTV (abbreviation for Total Thickness Variation), which is the maximum difference between the thickest and thinnest points of a wafer.

[0046] The distribution of measurement points at which the topography is measured is adapted to the specific measurement task. In the case of the Fig. In the embodiment shown in Figure 1, it is assumed that measuring points are arranged along two lines 11, 13, which are perpendicular to each other, intersect in the center of the wafer 10, and each extend to the circumferential edge of the wafer 10. Other measuring patterns, e.g., spirals or grid patterns, are also possible. The measuring points can be very close together and, for example, have a distance of a few micrometers. In other measuring patterns, the distances are in the range of 1 mm. 2. Structure and function of the measuring device

[0047] The Fig. Figure 2 shows a schematic representation of a measuring device according to the invention, designated as 14 in general. The measuring device 14 serves to measure a wafer 10, which is supported by a holder 16. The holder 16 can, for example, be designed as a ring holder having a central opening 18 and being provided with guides (not shown) on which the wafer 10 can be precisely aligned. The holder 16 is generally not part of the measuring device 14. When measuring within the production process, the wafer 10 can be fed to the measuring device 14, for example, by a conveying device (not shown).

[0048] The measuring device 14 comprises a first OCT sensor 20, which measures a surface S of the wafer 10 facing it, and a second OCT sensor 20', which measures the opposite surface S' of the wafer 10. The structure of the first OCT sensor 20 is described below. The second OCT sensor 20' has an identical structure and is therefore not described separately. To distinguish the components of the two OCT sensors 20 and 20', reference numerals referring to components of the second OCT sensor 20' are preceded by an apostrophe.

[0049] The first OCT sensor 20 comprises an optical coherence tomograph 22, which generates a measuring light beam 24 and whose structure is explained in more detail below.

[0050] A scanning device, indicated by 26, deflects the measuring light beam 24 variably in two orthogonal scanning directions. For this purpose, the scanning device 26 has a first scanning mirror 28, which is rotatably mounted about a first axis of rotation 30. A second scanning mirror 32 is rotatably mounted about a second axis of rotation 34, which is oriented perpendicular to the first axis of rotation 30. The scanning mirrors 28 and 32 are driven by galvanometer drives (not shown), which are controlled by a control unit 36. Fig. Figure 3 shows the two scan mirrors 28, 32 enlarged and greatly simplified.

[0051] The first OCT sensor 20 also includes an optical system 38, which is located in the Fig. 2 is indicated by three lenses L1, L2 and L3 and can be configured as an F-theta lens. In the illustrated embodiment, the optical system 38 focuses the measuring light beam 24 deflected by the scanning device 26 such that it always strikes the surface S of the wafer 10 facing the optical system 38 at approximately a perpendicular angle.

[0052] The optical coherence tomograph 22 includes a light source 42, a first beam splitter 44 which splits the light generated by the light source into the measuring light beam 24 and a reference light beam 46, a reference arm 48 for guiding the reference light beam 46, and an object arm 50 which uses the optical system 38 and the scanning device 26 and in which the measuring light beam 24 is guided.

[0053] During a measurement, the measuring light beam 24, propagating in the object arm 50, is focused by the optical system 38 onto the surface S of the wafer 10, where it is reflected predominantly specularly and to a lesser extent diffusely, and travels along the same light path back through the object arm 50 to the first beam splitter 44. There, the reflected portion of the measuring light beam 24 is superimposed with the reference light beam 46, which is guided in the reference arm 48 and reflected there by a mirror 52. Both light components are directed by a second beam splitter 54 onto a detector 56, which converts the optical interference signal into an electrical signal.

[0054] In the illustrated embodiment, the optical coherence tomograph 22 is designed as an FD-OCT, where FD stands for Fourier Domain. The detector 56 therefore contains a spectrometer that records the spectral intensity distribution of the interference signal. From this, an evaluation unit 57 connected to the detector 56 can calculate, in a manner known per se, the distance of the surface 40 to the measuring device 14 (for example, the lens L3) at the point of incidence of the measuring light beam 24. For further details on the optical coherence tomograph, reference is made to DE 10 2017 128 158 A1 (corresponding to US 2018 / 0164089 A1).

[0055] In the illustrated embodiment, the wavelength range of the light generated by the light source 42 lies in the infrared or near-infrared range. This corresponds to wavelengths between 950 nm and 2000 nm, with the wavelength range of 1000 nm to 1200 nm being preferred.

[0056] If, in certain situations, the propagation of the reference light beam 46 in the reference arm is undesirable, the light path can be interrupted by a switchable diaphragm 60 arranged in the reference arm, which may, for example, be designed as a central or focal-plane shutter. Such a situation may arise, for example, if the wafer 10 is transparent, so that the thickness can only be measured with one of the two OCT sensors 20 or 20'. Details are explained in the aforementioned DE 10 2022 104 416 A1, to which reference is made in this respect.

[0057] In the Fig. In the embodiment shown in Figure 2, the measuring light 24 propagates completely in free space. In other embodiments, the light is guided partially in optical fibers. If the optical coherence tomograph 22 is housed in its own casing, it is advantageous to also guide the light between the optical coherence tomograph 22 and the scanning unit 26 in an optical fiber. The scanning unit 26 and the optical system 38 can then be housed in a measuring head that can be mounted at a distance from the optical coherence tomograph 22.

[0058] As mentioned previously, the identically constructed second OCT sensor 20' measures the surface S' of the wafer 10 from the opposite side. The optical axes OA, OA' of the two OCT sensors 20, 20' are coaxially aligned. The angle between the two optical axes OA, OA' should not exceed approximately 1°.

[0059] In the illustrated embodiment, the entire setup, including measuring device 14, holder 16, and wafer 10, is oriented such that the wafer 10 extends in a horizontal plane and the optical axes OA, OA' are vertically aligned. Alternatively, the entire setup can be tilted by 90° so that the wafer extends in a vertical plane. Such a tilted arrangement prevents the wafer 10 from sagging under the influence of gravity and thus deforming further. 3. Thickness measurement

[0060] The Fig. Figure 4 shows a schematic, non-scaled section of the opaque wafer 10, whose thickness d is to be measured at several locations in the xy-plane. The thickness d at a coordinate (x, y) is defined as the distance between the surfaces S, S' along the z-direction, as shown on the left in the Fig. 4 is recognizable. In order to measure this distance correctly, the two measuring light beams 24, 24' must scan the surfaces S, S' at measuring points that are exactly opposite each other along the z-direction, i.e. have the same coordinates (x, y).

[0061] If this condition is not met, as shown on the right in the Fig. As indicated in Figure 4 for two measurement light beams 24a, 24a', a measurement error Δd occurs. This is often particularly large when the lateral distance (i.e., the distance in the xy-plane) between the measurement light beams 24a, 24a' is large and / or the thickness d of the wafer 10 varies greatly even over short distances.

[0062] The reason for this lateral offset of the measurement points is that each OCT sensor 20, 20' has its own coordinate system, but the two coordinate systems do not coincide perfectly. If, for example, the origin of each of the two coordinate systems lies on the respective optical axis OA of the optical systems 38, 38', and the measurement light beam 24, 24' is to be directed onto the optical axis, controlling the scanning devices 26, 26' with the target (x, y) = (0, 0) ensures that the respective measurement light beam 24, 24' is focused precisely onto the optical axis OA, OA' of the respective OCT sensor 20 or 20'. This applies at least if a corresponding xy calibration has been carried out beforehand, so that a control of the target (0, 0) leads to swivel angles of the scan mirrors 28, 32 or 28', 32', in which the measuring light rays 24, 24' actually spread concentrically to the respective optical axis OA, OA'.How such an xy calibration can be carried out is described in the above-mentioned, unpublished German application 10 2024 109 466.4 of the applicant.

[0063] However, even the smallest deviations of the optical axes OA, OA' result in the measurement points scanned by the measuring light beams 24, 24' on the surfaces S, S' no longer being exactly opposite each other along the z-direction, leading to the problem on the right in the Fig. 4 indicated lateral offsets occur.

[0064] Besides the optical axes OA, OA' not being perfectly aligned, other causes for misaligned coordinate systems are also possible. Due to manufacturing tolerances of the scanning devices 26, 26' and the optical systems 38, 38', the two coordinate systems can be not only laterally offset from each other, but also rotated or scaled (i.e., compressed or stretched). A rotation of the coordinate systems can, for example, result from the OCT sensors 20, 20' being mounted in a twisted position relative to each other. However, at least after performing the xy-calibration described above in accordance with the aforementioned unpublished German application 10 2024 109 466.4, the two coordinate systems can always be transformed into each other to a very good approximation by a linear transformation.

[0065] In order to perform this transformation, a calibration is carried out, which will be explained in more detail in the next section.

[0066] The foregoing explanations have not taken into account that the coordinate systems in the Fig. The mirror-symmetric arrangement of the two OCT sensors 20, 20' shown in Figure 2 is not truly identical, even with perfectly aligned optical axes and without rotations or manufacturing tolerances, but rather mirror-symmetric with respect to the xz-plane or yz-plane. When "coincident coordinate systems" are mentioned in this context, it is assumed that this mirror symmetry has already been taken into account computationally. 4. Calibration

[0067] To perform the calibration, the wafer 10 is removed from the measurement plane so that the measuring light beam 24 generated by the first OCT sensor 20 can propagate unhindered to the second OCT sensor 20'. The holder 18 does not need to be removed, as its opening 18 is large enough not to obstruct the propagation of the measuring light beam 24.

[0068] For calibration, the light source 42' of the second OCT sensor 20' is switched off, so that only the measuring light beam 24 generated by the first OCT sensor 20 propagates. The scanning device 26 of the first OCT sensor 20 is controlled so that the measuring light beam 24 is directed to a specific xy coordinate.

[0069] The Fig. Figure 5 shows this constellation, assuming by way of example that the xy-coordinate lies on the optical axis OA and the origin of the coordinate system coincides with the optical axis OA, so that for the xy-coordinate (x, y) = (0, 0) holds.

[0070] The detector 56 of the first OCT sensor 20 either remains without voltage or any signals generated by it are not evaluated. Since there is no object in the measuring range of the first OCT sensor, no interference normally occurs on the detector 56.

[0071] The detector 56' of the second OCT sensor 20' is used only for intensity measurement during calibration. Since the light source 42' of the second OCT sensor 20' is switched off and therefore no reference light beam 46' is generated, interference cannot occur there either.

[0072] The detector 56' can only receive a signal if the scan mirrors 28', 32' of the second OCT sensor 20' are in a position where they direct the measuring light beam 48, after passing through the optical system 38', onto the detector 56'. Due to the reversibility of the light path, this is the same position in which a measuring light beam 24' generated by the second OCT sensor 20' would be directed into the detector 56 of the first OCT sensor 20.

[0073] As part of the calibration, it is first necessary to find out how the scan mirrors 28', 32' of the second OCT sensor 20' must be controlled so that a measuring light beam 24' generated by the second OCT sensor 20' is focused exactly at the same place (or at a place only slightly shifted in the z-direction) as the measuring light beam 24 generated by the first OCT sensor 20.

[0074] The starting point for this search can be a corresponding position such as that of the scan mirrors 28, 32 of the first OCT sensor 20, taking into account the reflection of the coordinate systems mentioned above. This position corresponds to the position in the Fig. In the exemplary constellation shown in Figure 2, a location on the optical axis OA is found, but this location does not lie exactly on the optical axis OA' of the second OCT scanner 20'. Therefore, the second OCT scanner 20' passively scans this area to find the location 70 where the measuring light beam 24 crosses the xy-plane. The term "passive" here describes the fact that the second OCT scanner 20' does not emit light, but only passively receives light. During this passive scanning process, the scan mirrors 28', 32' of the second OCT sensor 20' are successively adjusted, similar to active scanning during a normal measurement, as shown in Figure 2. Fig. 5 is indicated by a dashed line representing the beam path.

[0075] The Fig. Figure 6 schematically shows the xy-plane in which the wafer 10 to be measured is normally located, and the location labeled 70 where the measuring light beam 24 passes through the xy-plane. Preferably, the scanning process described above is carried out in two steps. In a first step, the entire area in the Fig. The xy-plane shown in Figure 6, or at least a part of it, in which location 70 is presumed, was scanned with a coarse scanning grid, as is done in the Fig. 7a indicates a section 71 marked with dashed lines. One can see in the Fig. 7a, that the grid of sampling points 74 is chosen so coarsely that each point in the section 71 is passively sampled at least once and at most twice. The grid width, i.e., the smallest distance between two adjacent sampling points 74, can be, for example, 0.5 to 0.7 times the diameter of the passive measurement spot 73. Dashed lines are shown in the Fig. Figure 7a indicates some of the passive measurement spots 73, which are targeted at different times during the first sub-step with the coarse scanning grid. The scanning points 74 can be approached, for example, according to a meandering pattern, as shown in the Fig. 8 is indicated.

[0076] In the Fig. 7a further shows that the active measurement spot 75 sought is located at position 70 and is covered by a total of seven passive measurement spots 73. This allows position 70 to be roughly located.

[0077] In a second step, a smaller sub-area is scanned, using the finest possible scanning grid. The position and, if necessary, the size of this sub-area are determined based on which passive measurement spots 73 detected measurement light during the coarse scanning in the first step. In the illustrated embodiment, the grid width in the second step is only about 1 / 10 of the diameter of the passive measurement spots 73.

[0078] The passive scanning described above yields an intensity distribution in the vicinity of point 70, to which intensity values ​​obtained at several sampling points contribute. By simply determining the centroid of this intensity distribution, it is possible to ascertain where the maximum of the intensity distribution, and thus location 70, lies within the coordinate system of the second OCT sensor 20'.

[0079] Each scanning point 74 can be assigned coordinates in the coordinate system of the second OCT sensor 20' via the control commands of the scan mirrors 28', 32'. By interpolation, the location of point 70 in the coordinate system of the second OCT sensor 20' can thus be determined from the position of the center of gravity of the intensity distribution. A simple computational shift of the two coordinate systems relative to each other can then align their origins.

[0080] In order to also detect rotation and / or scaling, the calibration measurement described above must be repeated for a second location 76, which may be, for example, near the edge of the scannable measurement field, as is the case with the Fig. Figure 6 illustrates this. Choosing a differently located point is also possible. For example, locations 70 and 76 could be at the end of a horizontal or vertical scan line.

[0081] After determining the coordinates of two locations 70 and 76 in the coordinate systems used by the OCT sensors 20 and 20', a correction formula can be generated in a next step. This formula allows any point in the first coordinate system to be uniquely and reversibly assigned to points in the second coordinate system. In the most general case, this correction formula can take the form of an affine transformation, which can be mathematically formulated as multiplication by a 2x2 matrix and the addition of a displacement vector.

[0082] For a more precise determination of the transformation, for example, to distinguish between a rotation and different scaling in the x and y directions, the calibration measurement described above can be repeated for a third location. Using additional locations is also conceivable, but this may lead to an overdetermination of the transformation parameters, necessitating an averaging of the obtained parameters. 5. Triggers

[0083] If, after calibration with the measuring device, 14 thickness measurements are to be performed, it should be ensured that the measurement at each specified xy-coordinate is carried out exactly simultaneously. If a time Δt elapses between a measurement at an xy-coordinate with the first OCT sensor 20 and a measurement with the second OCT sensor 20', vibrations, which are often unavoidable, especially during measurements during ongoing production, can cause the wafer 10 to change its position relative to the OCT sensors 20 and 20' within the time Δt. This would lead to unsystematic measurement errors.

[0084] To ensure perfectly simultaneous measurements, each measurement should be triggered by a trigger signal supplied to both the first and second OCT sensors. This trigger signal can be generated by an external unit and fed to the two OCT sensors 20 and 20'. In the embodiment shown in the figures, it is assumed that the first OCT sensor 20 generates the trigger signal itself and supplies it to the second OCT sensor 20' via a data line 78.

[0085] Since the measurement points are approached by the measuring light beam at a high clock rate (typically several kilohertz) during scanning, it is neither possible nor necessary to synchronize each individual measurement. It is sufficient if the trigger signal initiates a predefined sequence of measurements at different locations. These locations can, for example, lie along a line. The scanning of each line can then be triggered by a separate trigger signal. The more identical the setup of the two OCT scanners (20, 20') is, the longer the time elapses before the individual measurements are no longer perfectly synchronized.

[0086] Synchronization can be verified by evaluating the measured intensity during a scan without an object, similar to the calibration described above. With optimal calibration, the maximum intensity of the measurement spot of the other scanner is measured at each measurement point. Deviations from the maximum intensity can be correlated with temporal / spatial positional errors via the size of the measurement spot.

Claims

[1] Method for calibrating a device (14) for measuring the thickness (d) of an opaque object (10), wherein the device comprises a first OCT sensor (20) having a first optical coherence tomograph (22), a first scanning device (26) and a first collecting optic (38), and a second OCT sensor (20') having a second optical coherence tomograph (22'), a second scanning device (26') and a second collecting optic (38'), wherein the first OCT sensor (20) is arranged along a z-direction relative to the second OCT sensor (20') such that the first collecting optic (38) is arranged coaxially with the second collecting optic (38'), and wherein the method comprises the following steps: a) the first optical coherence tomograph (22) generates a measuring light beam (24); b) the first scanning device (26) is controlled such that the measuring light beam (24) is focused into a first location (70), where there is no opaque object at the location (70); c) the second scanning device (26') is controlled so that it scans an area around the first location (70) so that the measuring light beam (24) generated in step a) is directed at least temporarily by the second scanning device (26') onto a detector (56') of the second OCT sensor (20'); d) taking into account intensity signals measured by the detector (56') in step c), a first coordinate system used by the first OCT sensor (20) is aligned with a second coordinate system used by the second OCT sensor (20'); e) From the comparison in step d), a correction rule is generated, according to which points in the first coordinate system can be uniquely and reversibly assigned to points in the second coordinate system. [2] Method according to claim 1, wherein steps b) to d) are repeated for a second location (76) which is different from the first location (70). [3] Method according to claim 2, wherein in step d) a relative rotation, a translation and a scaling of the two coordinate systems is determined. [4] Method according to one of the preceding claims, wherein in steps c) and d) a two-dimensional intensity distribution is measured by the detector (56') and in step d) it is recorded at which control the intensity distribution has its center of gravity. [5] Method according to one of the preceding claims, wherein step c) comprises a first sub-step and a second sub-step, wherein in the first sub-step the area around the location (70) is scanned with a first scanning grid, and in the second sub-step a smaller area around the location (70) is scanned with a second scanning grid that is finer than the first scanning grid. [6] Method according to one of the preceding claims, wherein an xy calibration is performed for the first OCT sensor (20) and the second OCT sensor (20') by which control commands for the first scanning device (26) and for the second scanning device (26') are assigned coordinates in the first coordinate system and the second coordinate system, respectively. [7] Device (14) for measuring the thickness (d) of an opaque object (10), wherein the device (14) comprises: a first OCT sensor (20) which has a first optical coherence tomograph (22'), a first scanning device (26) and a first collecting optic (38), and a second OCT sensor (20') which has a second optical coherence tomograph (22'), a second scanning device (26') and a second collecting optic (38'), wherein the first OCT sensor (20) is arranged along a z-direction relative to the second OCT sensor (20') such that the first collecting optic (38) is arranged coaxially to the second collecting optic (38'), the first scanning device (26) can be controlled on the basis of a first coordinate system and the second scanning device (26) can be controlled on the basis of a second coordinate system, and the first coordinate system and the second coordinate system are or can be transformed into each other by a transformation obtained by a calibration according to one of the preceding claims. [8] Device according to claim 7, wherein a trigger signal can be supplied to the first OCT sensor (20) and the second OCT sensor (20') which triggers an exactly simultaneous measurement of the distance to the object (10). [9] Device according to claim 8, wherein the trigger signal initiates a predetermined sequence of measurements at different locations. [10] Device according to claim 9, wherein the locations lie on a straight line, a rectangle, a circle or a spiral.

Images