DEVICE AND METHOD FOR MEASURING HEIGHT PROFILES ON AN OBJECT

Description

[0001] The present invention relates to a device, a system, and a method for measuring surface profiles or height profiles on flat objects, in particular on wafers in the semiconductor industry. The device comprises a beam splitter for splitting a monochromatic light beam and forwarding a partial beam to the object. The system comprises the optical device, a detector unit, a light source, and a signal evaluation unit for determining the surface profile.

[0002] In many industries, surface profiles of flat products are optically measured using imaging techniques. For example, in the semiconductor industry, the surfaces of wafers are measured. Wafers are discs made of semiconductor, glass, or ceramic materials. In certain applications, they are typically inspected over their entire surface or at least over large areas. The circuits ("chips") are manufactured on the wafers, which later perform a variety of tasks in electronic devices for computing, sensor technology, control, and monitoring.

[0003] To electrically contact the circuits, conventional processes involved running wires from the chip package to the chip, where they were attached to metallic contact pads. With advancing miniaturization and ever-increasing integration, it has become state of the art to create contacts by applying solder balls to the contact points on the chip. This occurs while the chip is still in the wafer assembly, i.e., the wafer has not yet been sawn to separate the individual chips. The contacts can thus be distributed across the entire chip surface ("ball grid arrays"). Contacting is achieved by inserting the chip into a package equipped with contact pads. Heating and pressure melt the solder, establishing contact. The uniform height of the solder contact balls is crucial for successful contacting using this process.

[0004] Currently, this process is shifting from solder balls to copper cylinders, which are only provided with a very thin solder cap for electrical contact. This allows the contact structures to be made even smaller and placed more densely. Since copper is not ductile like solder, the permissible height tolerances for this newer process are significantly reduced.

[0005] It is therefore necessary to inspect the contact structures (solder balls or copper cylinders) for their height even more precisely than before. This requires fast and accurate measuring methods that enable precise and, if possible, complete measurement of the contact structure heights. Complete measurement of all contacts is considered particularly advantageous because even a single abnormal contact can render the finished component unusable. Therefore, high measurement speed is particularly important. Current contact structures typically have diameters of a few 10 µm, with a sharply decreasing trend. The aspect ratio (ratio of height to diameter) is usually around 1.

[0006] At the same time, processes are currently being developed to stack multiple chips in a single assembly. The goal is to almost completely eliminate the use of raised contact structures. These processes are being investigated both for integration in the wafer assembly and for individual chips, each of which is placed on a chip still in the wafer assembly. These processes also require precise and fast profile measurement to ensure the high requirements for surface planarity in the sub-nanometer range.

[0007] The typical sizes for the contact structures of these processes are diameters in the single-digit µm range and profile heights of a few nm to a few 10 nm.

[0008] Similar tasks can be solved for other structures and in other industries. These include measuring the height of conductor tracks and other structures on electronic boards (printed circuit boards) or measuring a variety of mechanical-electrical system (MEMS) components such as acceleration, pressure, and velocity sensors or gyroscopes.

[0009] All of these applications share a number of similarities. They all require rapid testing of a large number of generally similar test objects, such as circuit boards, wafers, displays, and similar objects. The use of sensors to generate profile and / or height data of the structures under investigation is also common across all applications. Furthermore, the applications considered here do not necessarily require a high-resolution 3D image. In some cases, a single value (such as the cap height of the contact balls and cylinders discussed above) is sufficient for testing. State of the art

[0010] Conventional methods for height measurement use point sensors, line sensors, or area sensors. Area sensors, such as those based on deflectometry methods, or white light interferometers, have the advantage of capturing an entire surface. In deflectometry, the surface to be measured is used as a mirror through which a regularly structured light source is viewed. The height variation of the object is reflected in the distortion of the mirror image and can be extracted from it. The method is very fast, but limited in its lateral resolution and therefore not suitable for the described task, or only to a limited extent.

[0011] White-light interferometry requires a series of images taken at varying distances from the object or interferometer mirror. This is too slow for a full-surface inspection of a wafer. Line sensors use methods such as triangulation, stereometry, or chromatic confocal imaging. Each of these methods exhibits measurement artifacts when moving to small contacts, making precise measurements difficult or impossible, or the methods themselves are too slow. Conventional point sensor applications require very precise positioning on the contact structure. Since the lateral placement of the contacts itself is on the order of magnitude of the sensor's resolution, this is time-consuming and increasingly error-prone.

[0012] US 2011 / 032503, US 2009 / 009773, DE 10 2014 108886, and DE 695 24 298 disclose various methods for improving profile measurements based on the interferometric principle. US 2011 / 032503 explains a method for determining the global inclination of a largely flat surface in a single acquisition step using a tilted reference mirror. In US 2009 / 009773, such a tilted reference mirror is used together with white light interferometry to facilitate the mathematical evaluation of the raw data. DE 10 2014 108886 discloses the use of rapid modulation of the reference arm length to stabilize or accelerate the phase determination in the interferometry measurement. A second beam with a phase shift is also discussed, but only to ensure the correct phase determination beyond doubt. DE 695 24 298 uses a heterodyne process to improve signal quality.What all these publications have in common is that no solution for considering different material reflections is revealed.

[0013] CN 109 387 155 presents a method in which a complete, model-based characterization of the reflection of samples consisting of stacks of one or more layers is performed by sequentially performing multiple interferometry measurements at different wavelengths and reference mirror positions. Although this approach fully accounts for material differences, performing multiple measurements—especially with the switching between different light sources and mechanically moving components during the measurement—is time-consuming and therefore too slow for many applications. Finally, EP 3 477 252 presents a solution for accounting for the reflection differences of different surface materials. However, the method requires spatially separating the light beam and incorporating a perfect light trap, both of which are difficult to implement.In addition, different types of measurement data are recorded, which makes the evaluation more susceptible to errors. Disclosure of the invention

[0014] It is an object of the invention to propose an improved system and an improved device for detecting a surface profile of an object surface of an object by means of interferometric distance measurement, which in particular allow the heights of even small contact structures to be measured in a shorter time with high accuracy.

[0015] The object is achieved with an optical device having the features of claim 1, a system having the features of claim 3 and a method having the features of claim 15.

[0016] According to one aspect of the present invention, an optical device comprises a beam splitter for splitting a light beam from a light source into a first partial beam and a second partial beam, a beam splitter for splitting each partial beam from the light source into a reference beam and a measurement beam, and a mirror for reflecting the two reference beams. Each measurement beam is directed to a measurement area on the object surface for reflection and, after reflection, is returned to the beam splitter as an object beam.

[0017] Each reference beam is reflected by the mirror and directed to the beam splitter as a mirror beam. The mirror is designed such that the two reflected reference beams, i.e., the mirror beams, exhibit a phase difference. The object beam and the mirror beam each interfere after striking the beam splitter and are each fed to a detector unit for analysis as an evaluation beam. The two evaluation beams exhibit a phase difference upon striking the detector unit.

[0018] According to the invention, the device comprises a detector unit for detecting the evaluation beams and a signal evaluation unit for evaluating the detected evaluation beams and determining the surface profile.

[0019] To determine the surface profile, each measurement beam is directed to a measurement area on the object surface for reflection. After reflection, it is directed to the beam splitter as an object beam. Each reference beam is reflected by the mirror and directed to the beam splitter as a mirror beam. The respective object beam and the corresponding mirror beam interfere after striking the beam splitter and are each fed to a detector unit for analysis as an evaluation beam.

[0020] The signal evaluation unit determines the intensities of the two evaluation beams and uses the intensities of the two evaluation beams to determine a height profile of the surface. For the height profile, the influence of the surface's reflectance on the measured intensities is computationally compensated. This allows a correct profile to be determined for object surfaces with less than 100% reflection, as well as for object surfaces with different reflection coefficients within the area to be measured.

[0021] According to another aspect of the present invention, the system for detecting the surface profile of an object surface of an object by means of interferometric distance measurement comprises a light source for generating a monochromatic light beam and the optical device described above.

[0022] According to the invention, the two evaluation beams that strike the detector unit exhibit a phase difference, from which the surface profile of an object's surface can be deduced. In previous arrangements, which are also based on interferometric distance measurement, a so-called optical trap is used to absorb at least one partial beam after splitting a light beam into two partial beams. The known arrangements are designed such that a non-interfering beam and an interfering beam are fed to a detector arrangement. A height profile is then deduced from the evaluation of the two different beam types.

[0023] Within the scope of the invention, however, it was recognized that it is also possible to evaluate two interfering light beams with a detector unit and to record the height profiles from the measurement results using a signal evaluation unit. The detector unit has two detectors, preferably sensors or cameras. Such an inventive arrangement has the advantage that an optical trap (light trap) does not need to be used. An optical trap is an optical element that completely absorbs the incident radiation. In practice, this is very complex and difficult to manufacture. If it is not possible to absorb almost all of the radiation, interference effects occur that can influence the measurement and falsify the result.

[0024] Within the scope of the invention, it was further recognized that when using two interferometric, i.e., interfering, beams, they must exhibit a phase difference to enable evaluation. Preferably, the phase difference is not equal to an integer multiple of the wavelength used.

[0025] The device and system according to the invention have the advantage that, by using two evaluation beams with a phase offset or a phase difference, which is preferably fixed, even objects whose surface does not reflect incoming light 100% can be measured. This makes it possible, in particular, to measure surfaces that have a different degree of reflection in parts of their surface. For example, this is the case with objects that have different thicknesses or different materials. In particular, wafers in the semiconductor industry have a large number of different material combinations locally, some of which also form partially transparent multiple layers. Such thicknesses and material combinations can be measured precisely and quickly using the device and system according to the invention.

[0026] According to the invention, a reference beam is generated from each of the two partial beams. In a preferred embodiment, the phase difference is therefore generated so that each of the reference beams (first reference beam from the first partial beam and second reference beam from the second partial beam) strikes a mirror, e.g. with two mirror segments or partial mirrors or mirrors, and is reflected there, wherein the mirrors or mirror segments have different positions. Preferably, the mirror surfaces are arranged parallel to one another, wherein the position of the mirror surface in the direction of the surface normal is different. The mirrors therefore have an offset in the direction of the surface normal. This results in differences in the travel time between the reference beams and the reflected, so-called mirror beams, which leads to the phase difference.

[0027] In contrast to the optical trap, where most of the radiation is absorbed, the mirror reflects most of the radiation. At least 30% of the radiation is reflected, preferably at least 50%, very preferably at least 80%, particularly preferably at least 95%.

[0028] In a preferred embodiment, a mirror is used that comprises two parallel mirror segments whose mirror surfaces are also offset in the direction of the normal. Alternatively, the mirror segments can be separate mirrors. It is also possible to apply the mirror surface to an optical element, such as a transparent block, a glass block, or a Plexiglas block, e.g., by vapor deposition.

[0029] In a particularly preferred embodiment, the beam splitter and the mirror of the device according to the invention are integrated into one optical element. For example, the optical element can be formed from a transparent material or a glass block. For example, the optical element could consist of two transparent blocks, each having a triangular base, preferably an equilateral and rectangular base.

[0030] One of the blocks can be designed such that the side surface has an offset at one of the legs of the triangular plan, with the outer surfaces being parallel but offset in the direction of their normals. The mirrors or mirror segments can be attached to these two offset surfaces, which are preferably of equal size and preferably aligned parallel. The mirrors are preferably vapor-deposited or applied to the transparent block and their offset surfaces.

[0031] If both transparent blocks are joined together in such a way that their long sides, which correspond to the side surface at the hypotenuse of the triangular base, are brought together and contacted, a beam splitter can be formed. For example, the two transparent blocks can be glued together. One or both of the long sides can be provided with a partially transparent mirror.

[0032] In a preferred embodiment, the light source comprises a laser that generates monochromatic light. If multiple light sources are used, whose light beams are preferably bundled, the light sources each comprise a laser.

[0033] In a preferred embodiment, multiple light sources are used, each generating a monochromatic light beam. The light beams of the individual light sources preferably have different wavelengths. Preferably, none of the light sources has a wavelength that is an integer multiple of the wavelength of another light source. The light beams of the individual light sources therefore have different wavelengths, which are preferably not an integer multiple of the wavelength of another light beam.

[0034] In a preferred embodiment with multiple light sources, the light beams from the individual light sources are bundled in a beam coupler, so that the outgoing light beams comprise the light beams from all light sources. For example, the different light sources can generate light beams in different colors.

[0035] In a preferred embodiment, the beam splitter comprises a semi-transparent mirror that allows part of an incident light beam to pass through as the reference beam and reflects another part of the incident light beam toward the surface of an object as the measurement beam, thus deflecting it. Preferably, the light beam is split 50% into a reference beam and a measurement beam.

[0036] The detector unit of the system according to the invention preferably comprises a time-delayed integration camera, a so-called TDI camera, which is preferably a TDI line scan camera. In a preferred embodiment, a color TDI camera or color TDI line scan camera is used. Particularly preferably, the detector unit has a TDI multi-channel line scan camera, which can be color or monochrome.

[0037] When splitting the light beams from the light source(s) into two partial beams, two cameras or line scan cameras, preferably TDI cameras, are preferably used in the detector unit. Since the arrangement of two partial beams, their respective splitting into reference beams and measurement beams, and the subsequent superposition of the reflected mirror beams or object beams to form evaluation beams, results in two beams, one beam is directed to each camera.

[0038] In a further preferred embodiment, an optical element is provided which separates the bundled light beams, in particular the bundled evaluation beams. The optical element is arranged such that the separation of the light beams occurs before they hit the detector unit and its cameras. The separation of the bundled light beams, for example the bundled evaluation beams, can occur before or after the two evaluation beams have been separated and guided to the respective cameras. It is also possible to separate the bundled light beams after the evaluation beams have been separated and directly before the beams enter the respective camera. Separation is advantageous when cameras are used that are sensitive to the individual light beams and their wavelengths.When using multi-channel line scan cameras or color TDI cameras, such separation and such an optical element for separation can be dispensed with.

[0039] According to the invention, the object is also achieved in that the light source continuously illuminates the object to be measured with light rays having at least one monochromatic wavelength; the relative movement between the combined detector unit (so-called camera / interferometer arrangement) and the object is continuous and uninterrupted; a fixed, suitably structured reference mirror generates a phase shift between two partial beams of the light source; with the detectors of the detector unit (e.g.the camera sensors of the interferometer arrangement) the intensity of both partial beams of the light (evaluation beam) created by the interference of the light reflected from the object (object beam) with the light reflected at the reference mirror (mirror beam) is measured simultaneously at all wavelengths used; the measured intensity of the first partial beam interference signal (first evaluation beam) is corrected at each wavelength by means of the measured intensity of the second partial beam interference signal (second evaluation beam) modified by the fixed reference phase offset; and the distance to the object is calculated from the corrected interference signal intensities of the first partial beam interference signal (first evaluation beam).

[0040] In such an arrangement, line sensors for interferometry measurements are preferably used as detectors. This allows a wide strip of the object to be measured continuously. Another aspect of the invention is the correction of the influence of the different surface materials of the object to be measured. To expand the measurement range to the ranges necessary for the profile heights to be measured, it is proposed to use multiple wavelengths simultaneously in the system according to the invention.

[0041] Within the scope of the invention, it was recognized that the distance to the measurement object can be clearly determined with very high resolution from the intensity measurement of a signal modulated by interferometry. However, this only applies if the surface of the measurement object reflects 100% of the light used or if the reflectance is precisely known for each measurement point. If the measurement object has partially transparent layers, then for an absolute distance measurement, their thicknesses and optical properties (refractive index and absorption coefficient) at the wavelength used must be precisely known for each measurement point. For a relative profile measurement, one could dispense with exact knowledge of these values, but this would be limited to the fact that the thicknesses and materials must be exactly the same for all measurement points, or their variation would limit the measurement accuracy.

[0042] Interferometry is therefore not readily applicable for measuring the surface profile of an object, particularly for measuring the height of a wafer, because the intensity of the measured interferometry signals also depends on the material of the structure being measured and its surroundings. For example, the wafers to be measured locally comprise a multitude of different material combinations, some of which are formed by semi-transparent multilayers. Variations in thickness and material make precise, rapid measurements at one or a few wavelengths impossible. Without knowledge of the local reflectance and the phase shift of the reflected light (evaluation beam) generated by the semi-transparent layers, a sufficiently accurate statement about the distance cannot be made.

[0043] In order to correct the influence of the material-related local variation of the intensity of the reflected light, according to the invention the interferometry measurement is carried out simultaneously with at least two partial beams which differ by a fixed phase shift.

[0044] Using two partial beams, the accuracy of the measurement can be significantly increased, as the influence of the reflectance can be eliminated. This is sufficient for many tasks, for example, in cases where there are no partially transparent layers on the measurement object or their influence can be neglected.

[0045] The use of more than two partial beams can be used to improve the robustness of the measurement by increasing data redundancy.

[0046] The following example illustrates the correction for a single detector element and a single wavelength λ 1 with reference to a height measurement on a wafer, without loss of generality. To realize the profile measurement at high speed, line scan cameras with multiple detector lines are advantageously used. The signal Si arriving at the interferometry detector can be represented as S i = q ∗ R ∗ γ ∗ 1 + c os 2 π λ 1 / 2 z ri − z s − φ

[0047] Here, q denotes the intensity of the light source, R the effective reflection coefficient of the surface of the measuring object at the wavelength λ i , y a device constant resulting from detector sensitivity, absorption and reflection losses at the optics, z ri and zs the distances between the splitter mirror of the interferometer and the reference mirror or the surface of the measuring object and ϕ the phase shift which may be generated by partially transparent layers on the measuring object surface.

[0048] For the case of negligible phase shift ϕ ~ 0, the ratio of the measured intensities is obtained by combining two partial beams i = 1.2: s 2 s 1 = 1 + c os 2 π λ 1 / 2 z r 2 − z s 1 + c os 2 π λ 1 / 2 z r 1 − z s

[0049] This can be done with Δ z r = z r 1 − z r 2 , Δ z = z r 1 − z s und p = 2 π λ 1 / 2 transform into: s 2 s 1 = 1 + c os p Δ z − Δ z r 1 + c os p Δ z

[0050] To determine the desired real distance to the surface of the measuring object, Δz must be determined by solving the transcendental equation. Since Δz r is a known quantity (or can be determined by a calibration measurement using a flat mirror instead of the measuring object), the surface profile of the measuring object can be determined, provided it does not have vertical steps larger than half the wavelength of the light used.

[0051] When using only one wavelength for the interferometry measurement and the inventive solution described so far, the distance working range of the measurement is limited to half a wavelength of the light used. Outside of this range, the relationship between intensity and distance becomes ambiguous, since the signal intensity alone cannot be used to directly determine the number of periods the path difference between the reference beam and the object beam is. Conventional interferometers solve this problem by determining a reference distance during initialization and then continuously counting the measured intensity periods. However, this is unsuitable for the inventive use of a line-type interferometer, since a reference would have to be available that is parallel to the interferometer line arrangement by better than half a wavelength, and there should be no steps on the measurement object that are more than half a wavelength high.At least the latter is not the case, for example, for measuring contact structures on wafers.

[0052] However, if two or more different wavelengths are used, the working range within which the measurement can be performed unambiguously without a starting reference and period counting can be extended, since the different wavelengths have different periodicities, and thus, a unique measurement can be obtained from the combination of the signals of the wavelengths used in a wider range. The fundamentals of such a method are presented in "Multi-Wavelength Interferometry for Length Measurements Using Diode Lasers" by K. Meiners-Hagen et al. The directly detectable working range A for two wavelengths λ 1 and λ 2 is: A = Λ 2 = 1 2 ∗ λ 1 ∗ λ 2 λ 2 − λ 1

[0053] If one considers not only the total intensity profile of the synthesized wavelength A, but also the two wavelengths λ 1 and λ 2 individually, the working range A can be extended to a multiple of the wavelengths λ 1 and λ 2 (method of exact phase fractions).

[0054] However, the method described in the above-mentioned publication cannot be used for height measurement on a wafer, since the intensity of the measured interferometry signals also depends on the material of the structure being measured and its surroundings. The inventive method described above solves this problem.

[0055] Using the two simultaneously measured, phase-shifted interferometry signals, the interferometry measurements at the different wavelengths can be individually corrected. This eliminates material dependence. Subsequently, the distance for each point recorded with the line sensor array can be calculated individually within the unique working range, as shown in the above-mentioned publication. This creates a precise distance image of the surface. By using line sensors with high clock rates, very high measurement speeds can be achieved.

[0056] With the device and system according to the invention, profiles up to several 100 µm in height can be recorded with a resolution of less than 0.1 µm. This is sufficient, in particular, for many requirements when measuring the semiconductor chips and wafers described above. Applications include, for example, measuring the heights of contact structures (solder balls or cylinders), measuring the height of conductor tracks on the topmost wiring level of the chips, or determining the planarity on reconstructed wafers (wafers which are assembled by embedding previously separated chips in a carrier material, usually epoxy). Analogous tasks also exist for other structures and in other industries. Such tasks include, for example, the measurement of structures on electronic boards and printed circuit boards or the measurement of a variety of mechanical-electrical system components (MEMS) such as pressure sensors, speed sensors, or gyroscopes.

[0057] With appropriate illumination of the measurement area on the object surface and the use of suitable line or area detectors, this arrangement enables simultaneous measurements at multiple points. Unlike, for example, white-light interferometry, the measurement result is achieved with a single measurement at different wavelengths, without having to change the distance between the object and detector. This significantly accelerates the measurement process.

[0058] Preferably, a holding device is provided for holding an object in the form of a wafer or another flat object with a flat surface. The flat object to be measured is positioned in the holding device in such a way that the object surface is illuminated with the measuring light in the desired manner.

[0059] Lasers can be used as radiation sources because they provide sufficiently monochromatic light with high radiation intensity. Due to their high radiation intensity within a very narrow wavelength range, lasers are particularly suitable as radiation sources for the present invention. However, one or more spectrally broadband radiation sources equipped with appropriate spectral filters can also be used. The use of multiple spectral filters also makes it possible to generate multiple wavelengths from one radiation source.

[0060] In particular, the invention can provide for the radiation sources to generate light at wavelengths where no wavelength is an integer multiple of the wavelength of any of the other radiation sources. For example, a wavelength is not exactly twice the wavelength of another radiation source. The smaller the difference between the wavelengths, the larger the operating range. However, in practice, it must be considered that the radiation beams are never completely monochromatic and always have a bandwidth, which must be considered when selecting the wavelengths.

[0061] The combination of radiation when using different wavelengths can be achieved using spliced ​​optical fibers or dichroic mirrors. Spliced ​​optical fibers are optical fibers that are fused together at a specific section or are routed closely together, so that the light from one optical fiber passes into the other. With dichroic mirrors, only the light of a specific wavelength range is reflected, while light of at least one other wavelength range is transmitted. Such mirrors can be dimensioned for use at various angles. The most common angle is 45°. With a suitable arrangement, the outgoing reflected light and the transmitted light can have the same direction of propagation.

[0062] The phase difference of beams required for the arrangement (device and system) can be generated by a path difference, the path to be traveled by the light beams, as well as by the passage or reflection on different media.

[0063] One or more prisms, gratings, or other spectrally dispersive elements can be used to separate light beams of different wavelengths. This spatial separation allows the signals to be recorded simultaneously, allowing for high measurement speeds. However, temporal separation can also be achieved by performing the measurements at the different wavelengths one after the other, which requires a longer measurement time.

[0064] A particularly preferred embodiment of the invention provides that each of the detector arrangements comprises a multi-channel time-delayed integration (TDI) line scan camera. These cameras comprise several TDI blocks, each of which is sensitive to a specific wavelength range. Thus, the TDI arrangement can be used to generate a high signal-to-noise ratio, and at the same time, all wavelengths used can be measured simultaneously. This enables particularly fast and very precise measurements. However, single-channel TDI line scan cameras can also be used in the so-called area mode, i.e. like a normal area camera, in that the different wavelengths used are distributed across the sensor surface by a dispersive element, for example a prism. This allows the large width of the TDI detector line of, for example, 16384 pixels to be used for high throughput, and at the same time, the, for example,48, 96, or 256 TDI lines are available for simultaneous, separate acquisition of the different wavelengths. With this configuration, a lower measurement speed is available due to the lower clock rate in this area readout mode. The same applies to the use of standard area scan cameras, which is also possible.

[0065] In particular, the optical device used to generate the interference can comprise a semi-transparent mirror which reflects part of the radiation at a first surface towards the object surface (measurement beam) and transmits another part of the radiation (reference beam). The reflected radiation can then be combined with the transmitted beam. This can be done, for example, by reflecting the transmitted beam (reference beam) back into itself at a mirror (as a mirror beam) and coupling it into the reflected beam (object beam) at the semi-transparent mirror. Such an arrangement is called a Michelson interferometer. However, other arrangements can also be used, such as a so-called Mach-Zehnder interferometer, in which the incident light is split and recombined after reflection of a partial beam at the sample surface.

[0066] The inventive method for detecting the surface profile of an object surface (30) of an object by means of interferometric distance measurement comprises several steps, such as emitting a monochromatic light beam by means of a light source in the direction of an optical device, splitting the light beam of the light source into a first and second partial beam by means of a beam splitter, splitting each partial beam into a reference beam and a measuring beam, preferably by means of a beam splitter, guiding the reference beams to a mirror, reflecting the reference beams off the mirror, and forming mirror beams which have a phase shift. The mirror beams are guided to the beam guide and reflected there. The measuring beams are guided to a measuring region on the object surface and reflected there, thereby forming object beams which are guided to the beam guide.In further steps, the mirror beams and object beams interfere to form evaluation beams that exhibit a phase shift. One step involves detecting the evaluation beams, preferably using a detector unit, evaluating them, and determining the surface profile of the object surface, preferably using a signal evaluation unit. The two measured intensity values ​​of the evaluation beams, both of which were reflected from the same location on the object surface, are used to compensate for the influence of the reflectance of the light beams at the object surface when determining the surface profile. In this way, a correct surface profile is determined, both for object surfaces with less than 100% reflection and for object surfaces with different reflection coefficients within the area to be measured.

[0067] According to the invention, the method creates a monochromatic inspection image for each wavelength from the determined reflection coefficients of the object surface. The reflection coefficients are calculated from the previously determined height profile and the measured signal intensity of at least one of the beams (evaluation beams). However, the signal intensities of both evaluation beams are particularly preferably used.

[0068] The following procedure is explained using wafers as an example, without loss of generality. However, it can be used equally well for other flat objects or test objects for measuring surface profiles and profiles.

[0069] Preferred embodiments of the invention are described in the dependent claims. It is understood that the features mentioned above and those to be explained below can be used not only in the respective combinations specified, but also in other combinations or alone, without departing from the scope of the present invention. In particular, the method for the device and the system for detecting the surface profile of an object surface by means of interferometric distance measurement can be implemented according to the embodiments described in the dependent claims.

[0070] An embodiment of the invention is described below with reference to the accompanying drawings. They show: Fig. 1 is a schematic representation of an inventive system for interferometric distance measurement with three laser light sources; Fig. 2 is a schematic representation of method steps for setting up the system and for interferometric distance measurement. Fig. 3 is a detailed representation of the path lengths traveled by a first partial beam of light and the associated directions in an enlarged section of Figure 1 ; Fig. 4 a detailed representation of the path lengths traveled by a second partial beam of light and the corresponding directions in an enlarged section of Figure 1 ; Fig. 5 a schematic representation of a first embodiment of the system according to the invention according to Figure 1 ; and Fig. 6 a schematic representation of a second embodiment of the system according to the invention according to Figure 1 .

[0071] Fig. 1shows an embodiment of the system 1 according to the invention with an optical device 2, at least one light source 7, a detector unit 8, and a signal evaluation unit 9. The optical device 2 comprises a beam splitter 3, a beam splitter 4, and a mirror 5. The mirror 5 has two mirror segments 6, which in turn are designed as two mirrors 60, 70. The beam splitter 3 is preferably designed as a beam-shaping optic 80. The beam splitter 4 is preferably a partially transparent mirror 90 with preferably a 50% partial transmission. The at least one light source 7 preferably comprises, as shown here, three laser sources 110, 120, 130.

[0072] The structure of system 1 and the beam path in system 1 are described below using a wafer as the measurement object. In system 1, wafers 20 or other flat objects whose surface profile is to be captured are moved one after the other relative to a camera (detector unit 8).

[0073] In the present preferred embodiment, the camera is stationary, and the objects are moved beneath the camera. For this purpose, a holding and transport unit 10 is preferably provided, which supports and moves the wafers 20. The movement preferably occurs in a direction perpendicular to the surface normal of the object surface 30 of the wafer 20. In another embodiment, the camera can be moved. In an alternative embodiment, the relative movement can be split between the camera and the object, so that, for example, the camera preferably executes the movement in one axial direction, while the object is preferably movable in the direction perpendicular thereto. The movement between the camera and the object is preferably continuous.

[0074] The embodiment in Fig. 1uses three lasers as light sources 7 and two TDI (time delayed integration) multi-channel line scan cameras as sensors or detectors of the detector unit 8. However, a different number of light sources can also be used. Alternatively, a broadband light source can be used in combination with narrowband filters.

[0075] Likewise, simple multi-channel line sensors (without TDI method) or area cameras or a set of several line cameras can be used instead of the TDI sensors.

[0076] In Fig.1 To simplify the illustration, the beam offsets that are generated when the beams pass through a plane-parallel plate are omitted. In the example in Fig. 1The light from the three laser sources 110, 120, and 130 is combined by a fiber coupler 100. The three fibers 170, 180, and 190 fed by the laser sources are spliced ​​together in the fiber coupler 100, so that the light from the laser sources emerges in a common, multicolored light beam 84. Transmission via a fiber optic cable merely serves to decouple the lasers from the actual recording system. Instead of fiber optic transmission, the laser beams can also be directed directly into the recording unit via suitable optics, which reduces losses but requires more adjustment effort. The three beams from the three laser modules can also be combined using dichroic mirrors.

[0077] The beam-shaping optics 80 form the light beam 84 into a parallel light bundle consisting of the two bundle halves, so-called partial beams, 410 and 420 with a cross-section adapted to the recording surface. The bundle halves 410 and 420 are arranged around the beam center axis 400. They do not necessarily have to be adjacent to one another, i.e., a central area between them can remain unused. The splitting is preferably, but not necessarily, symmetrical. All wavelength components are evenly distributed over the entire cross-section and in particular between the bundle halves (partial beams) 410 and 420. The light bundle (consisting of the partial beams 410, 420) is directed by the 50% partially transparent mirror 90, partly as a partial bundle (measuring beams 430 and 440), via an objective 50, onto the wafer surface 40 to be measured.The other 50% of the light beam (partial beams 410, 420) pass through the partially transparent mirror 90 as partial beams (reference beams) 470 and 480.

[0078] The reference beams 470 and 480 strike the two reference mirrors 60 and 70, where they are reflected. The resulting partial beams (mirror beams 490 and 500) are again reflected by the partially transmissive mirror 90 at 50%.

[0079] The partial beams directed onto the wafer surface 40, measuring beams 430 and 440, are focused by the lens 50 into a common imaging area. Reflected by the wafer surface 40, they are reflected back by the lens 50 as partial beams (object beams) 450 and 460. The first object beam 450 comprises the reflected light of the first measuring beam 430, and the second object beam 460 comprises the reflected light of the second measuring beam 440. Half of the light from each of the object beams 450 and 460 passes through the partially transmissive mirror 90. The light originating from the first object beam 450 then interferes with the light of the first mirror beam 490 reflected by the partially transmissive mirror 90 to form a light beam forming the first evaluation beam 520.

[0080] The first evaluation beam 520 is now modulated in intensity by the interference of the light waves according to the distance difference between the partially transmissive mirror 90 and mirror 60 on the one hand, and between the partially transmissive mirror 90 and wafer surface 40 on the other. The intensity modulation occurs independently for all three wavelengths included.

[0081] In the same way, an interferometry signal is generated from the light of the second object beam 460. After passing through the partially transparent mirror 90, the light originating from the second object beam 460 interferes with the light of the second mirror beam 500 reflected by the partially transparent mirror 90 in a light bundle, the second evaluation beam 510.

[0082] The second evaluation beam 510 is now modulated in intensity by the interference of the light waves according to the distance difference between the partially transparent mirror 90 and mirror 70 on the one hand, and the partially transparent mirror 90 and wafer surface 40 on the other. The intensity modulation also occurs independently for all three wavelengths included. Fig. 3 and 4 show enlarged sections of Fig. 1 to illustrate the origin of the interference and the path differences to be taken into account.

[0083] The two evaluation beams 510 and 520 differ in their intensity at the three wavelengths used. The intensity difference of each wavelength, considered individually, is determined exclusively by the phase shift generated at mirrors 60 and 70. In the illustrated embodiment, the phase shift is generated by a different distance between the two mirrors 60 and 70 and the partially transmissive mirror 90. However, the phase shift can also be generated in other ways, e.g., by coating one of the mirror segments 6 or the partial mirrors (e.g., mirror 70) with a transparent coating.

[0084] The second evaluation beam 510 is directed by mirror 260 as a second detection beam 540 onto a prism 270. The prism 270 serves to spectrally split the light from the second detection beam 540 into the three wavelength components λ 1 , λ 2 , and λ 3 . Without limiting the generality, another suitable dispersive element, such as a grating, can be used instead of the prism 270. The thus split spectral components from the second detection beam 540 are focused by a first tube optics 200 onto a TDI multi-channel line scan camera 210. When using TDI multi-channel line scan cameras with color filters permanently installed in front of the sensor blocks, the spectral splitting of the beam can also be omitted.

[0085] By recording the light intensity incident on each pixel of the line scan camera 210, three signals i 1,1 , i 2,1 , and i 3,1 are generated. Their intensity is modulated by the profile height at the respective location on the wafer surface 40 and can thus be used for interferometric determination of the surface profile. The first index represents the wavelength, and the second index represents the camera (210) that supplies the signal.

[0086] Since mirror 260 only blocks the path of the second evaluation beam 510, the first evaluation beam 520, intensity-modulated by interference, impinges unhindered on prism 250 as the first detection beam 530. Prism 250 serves to spectrally split the light from the first detection beam 530 into the three wavelength components λ 1 , λ 2 , and λ 3 . Without limiting the generality, another suitable dispersive element, such as a grating, can be used instead of prism 250. The thus split spectral components from the first detection beam 530 are focused by a second tube optics 230 onto a TDI multi-channel line scan camera 220.By recording the light intensity incident on each pixel of the line scan camera 220, three signals i 1,2 , i 2,2 and i 3,2 are generated, the intensity of which is modulated by the surface profile of the wafer surface 40 and can thus be used for the interferometric determination of the surface profile.

[0087] Fig. 5 shows an alternative embodiment of the system according to the invention, in which the splitter mirror 90 and the two mirrors 60, 70 (reference mirrors) are replaced by a prism 600 consisting of two blocks bonded together, with a partially reflecting layer being provided on the bonded 45° surface. This layer forms the partially transparent mirror 610, which splits the two partial beams 410, 420, as the mirror 90 according to the embodiment of Fig. 1 The details and beam paths of the individual rays are shown in Fig. 1 described and are listed for clarity in Fig. 5no longer shown.

[0088] The according to the embodiment of Fig. 1 The two existing mirrors 60, 70, which serve as reference mirrors, are provided on the outer surface of the prism 600. For this purpose, for example, mirror coatings can be vapor-deposited onto the outer surface or mirror surfaces can be applied or glued on.

[0089] By appropriately constructing the prism 600, the necessary offset can be created on the outer surface so that the mirrors 60, 70 have the same offset as in Fig. 1 is shown.

[0090] After the mirror beams 490, 500 and the object beams 450, 460 interfere in the prism 600 to form the evaluation beams 520 and 510, these beams are bundled by a bundle optics 620 and directed onto an optical element 700. The optical element 700 preferably has a rectangular or square base. On the two sides facing the bundle optics 620, mirrors 730, 740 are arranged to reflect incident beams. The mirrors 730, 740 can also be designed as independent mirrors or reflective elements; however, they must then be aligned or adjusted.

[0091] Incident beams are deflected at the optical element 700. The two evaluation beams 520 and 510 are separated at the mirrors 730, 740 and directed to the cameras 210, 220. At the mirror 730, the evaluation beam 520 is reflected into a first detection beam 530, which is directed to the first camera 210 by a deflecting mirror 710. The second evaluation beam 510 is deflected at the mirror 740 as a second detection beam 540 to a deflecting mirror 710, so that the second detection beam 540 impinges on the second camera 210. The signals recorded by the cameras 210 and 220 are processed in the signal evaluation unit 9, so that the profile of the wafer surface 40 of the wafer 20 is recorded.

[0092] In the embodiment shown here, the detection beams 530, 540 are also focused such that they impinge on the cameras 210, 220 at a single point.

[0093] Fig. 6shows an alternative embodiment of the system 1 according to the invention, which corresponds to the structure according to the embodiment according to Fig. 5 The two versions of the Fig. 5 and 6 differ only in the use of the optical element that deflects the two evaluation beams 510, 520 into the detection beams 530, 540. According to Fig. 6 a flatter optical element 750 is used, so that no right-angled beam guidance is achieved, as in the embodiment according to Fig. 5 According to Fig. 5 the evaluation beams are directed by 90° onto the two deflection mirrors 710, 720, which are arranged at a 45° angle, so that the detection beams 530, 540 are directed at a 90° angle to the two cameras 210, 220 after they hit.

[0094] According to Fig. 6A deflection other than 90° is achieved by means of the optical element 750 and the two mirrors 730, 740 due to the arrangement of the mirrors 730, 740 on the non-rectangular side surfaces of the diamond-shaped optical element 750. The two deflection mirrors 710 and 720 are arranged accordingly so that the detection beams are directed, preferably focused, onto a point on the respective camera 210, 220. This allows for a significantly more compact measurement setup, so that the system can be made smaller and more compact, and the accuracy of the measurements can be increased.

[0095] Of course, the Prisma 600 can be Figures 5 and 6 preferably also in an embodiment of the invention according to Figure 1 be used, replacing the partially transparent mirror 90 and the mirrors 60, 70. The optical device 2 can also be made of Figure 1 in the statements according to the Figures 5 , 6be used.

[0096] Likewise, the detection unit can be Figure 1 , which comprises the cameras 210, 220 and the prisms 275, 270, the tube optics 200, 230 and the mirrors 260, by the detection unit according to the Figures 5 , 6 be replaced so that the bundle optics 620, the optical element 700 or 750, the mirrors 710, 720, 730 and 740, and the cameras 210, 220 are used. The same applies vice versa. Signal evaluation

[0097] The following explanation of the calculation of the surface profile from the acquired signals ix, rx, and qx with x = 1, 2, 3 for the three wavelengths λ 1 , λ 2 , and λ 3 is performed only for one pixel of the camera sensors of the line scan cameras 210 and 220, respectively. It is understood that this calculation can be performed for each pixel of the cameras 210, 220. Thus, depending on the sensor size and arrangement, it is possible to determine a plurality of profile elevation points simultaneously.

[0098] For a line scan camera with, for example, the 16,384 points per line available today, this corresponds to 16,384 height values ​​for each camera readout clock cycle. Furthermore, with the arrangement described here, the line scan cameras 210, 220 can be moved continuously relative to the wafer 20. Depending on the clock speed of the cameras, a corresponding number of lines with height information is obtained per unit of time. With a clock speed of the proposed multi-channel TDI line scan cameras of, for example, 100 kHz, more than 1,600 million height values ​​are obtained per second. Such cameras are offered by various manufacturers (e.g., Vieworks and Dalsa Teledyne). The use of such TDI multi-channel line scan cameras is a particularly suitable variant, as it allows very high measurement speeds to be achieved. These cameras contain several (usually four) TDI blocks in one camera, which can be operated and read out simultaneously.When using such cameras, the wavelengths used for a line on the wafer surface are recorded sequentially. This means that while the first TDI block records the line area at λ 1 , the second TDI block detects the signal at λ 2 , and the third TDI block detects the signal at λ 3 .

[0099] For the calculation presented below, this temporal offset is essentially irrelevant. Only the obtained signal patterns need to be assigned and evaluated in a temporally staggered manner (according to the spatial offset of the TDI blocks). This arrangement is particularly well-suited to combining high signal quality (correspondingly high-resolution and robust measurements) with high speed. Alternative arrangements are explained below.

[0100] The measurement procedure and your preparation is described in Fig. 2illustrated. For a correct determination of the height values, the measurement must be prepared by measuring the dark signal and determining the transfer function of the optics and sensors.

[0101] In the dark signal measurement, the signal d is measured at each camera pixel y of the two line scan cameras 210 and 220 with the light source switched off. This determines the so-called dark noise of the camera, which represents an offset for each subsequent measurement and is subtracted from the signal. This is performed for both line scan cameras (sensor arrays) 210 (index z=1) and 220 (index z=2). d xyz ausgelesener Signalwert bei Wellenlänge x λ x am Pixel y des Sensors z

[0102] To determine the optical and electrical transfer function of the arrangement (System 1), a bright signal measurement h is carried out with a known flat object. For this purpose, the wafer surface 40 is replaced by a flat reference piece with known reflection properties. Since the signal h at each sensor pixel is uniquely determined by the intensity of the light source (signal value q), the transfer function M, the reflectance of the reference piece r , the path difference lz of the two interfering light bundles (first object beam 450 and first mirror beam 490) (incident on sensor (line scan camera) 220, z = 2) or second mirror beam 500 and second object beam 460 (incident on sensor (line scan camera) 210, z = 1) and the dark signal d, with known values ​​h, q, r, and d, the transfer function M can be determined for each wavelength x, for each pixel y and for both sensors z as a function of the path difference.The transfer function is generally different for each wavelength x, each pixel y, and each camera z. It is determined by the sensitivity of the individual pixels, the illumination, material properties, coatings, and aberrations of the optics.

[0103] In order to check the output intensity of each laser light source 110, 120, 130 and to include it in the calculation as a correction or reference value, the signals q 1 , q 2 and q 3 of the monitor diodes 140, 150, 160, which are usually built into each laser module (on the side of the laser facing away from the output), can be used directly.

[0104] The signal h of the light reference measurement is: h xyz , href l z = q xy , href ∗ M xyz l z ∗ r x , ref + d xyz

[0105] These include: h xyz, href (lz )Measured value of the reference measurement at wavelength x at pixel y of sensor z (interferometer, line scan cameras 210 [z = 1] and 220 [z = 2]) as a function of the path difference lz . q xyz, href Incident light intensity of the reference measurement at wavelength x at pixel y. M xyz (lz )Transfer function at wavelength x at pixel y of sensor z as a function of the path difference lz . rx, ref Reflection coefficient of the reference measurement (known material) at wavelength x. d xyz Read out dark signal value (no excitation light) at wavelength x at pixel y of sensor z.

[0106] Note: Despite "dark = no light", the dark signal values ​​are "wavelength dependent" because different sensor pixels are used for the different wavelengths - these can have different dark count values.

[0107] The argument lz denotes the path difference between the two interfering beams, i.e. first object beam 450 and first mirror beam 490, incident on line scan camera 220, z = 2, and second object beam 460 and second mirror beam 500, incident on line scan camera 210, z = 1, respectively.

[0108] In Fig. 3 The relationships for the light beam pair, first object beam 450 and first mirror beam 490, are shown. As can be seen from this, the path difference l 2 is exactly 2x (z mirror2 - z sample ), since the path difference z difference cancels out because it is traversed exactly once by each partial beam (first measurement beam 430 and second partial beam 420). The distance z mirror2 of the reference mirror 60 has the index 2 here, since the two reference mirrors 60, 70, without loss of generality, have a different distance from the splitter mirror 90 of the beam splitter 4 to produce the phase shift according to the invention.

[0109] For the pair of light beams second object beam 460 and second mirror beam 500 analogous conditions apply as in Fig. 4 shown. Here, both later interfering partial beams (first partial beam comprising second partial beam 420 and second object beam 460; second partial beam comprising second reference beam 480 and second mirror beam 500) pass through the path difference z difference exactly twice, which cancels it out again and the path difference l 1 to be taken into account is exactly equal to 2x (z mirror1 - z sample ).

[0110] To determine the distance of the wafer surface 40 from the sensor array z sample (which corresponds to the optical device 2), slightly different methods can be used to define and apply the transfer function. The selected method is used equally for both sensors, line scan cameras 210 and 220.

[0111] First, the transfer function M(I) can be decomposed into a non-interfered factor M max< and the interference effect. The factor M max< is obtained by determining the transfer function M(I) only at its maxima. The intensity modulation due to the interference can then be determined directly from the values ​​of the measurement run. This first approach is described in more detail below.

[0112] Secondly, the intensity modulation caused by interference can be included in the transfer function M(I). M(I) is then determined in the reference measurement for the entire working range as a function of the path difference between the two beams (first and second partial beams) and the distance to the wafer surface. The distance difference to the wafer surface is determined during the measurement run by comparing the transfer function values ​​with those of the bright signal measurement performed as a reference. This comparison is performed for the triplet of wavelengths x = 1, 2, 3 and for both sensors z = 1, 2, and searches for the distance that matches all three wavelengths and both sensors.

[0113] For the sensors line scan cameras 210 and 220 the following transfer functions apply (where the / 2 in the denominator results from the 2x run through the difference z mirror -z sample): M xyz l z = M xyz max ∗ 1 + cos 2 π λ x / 2 ∗ z mirror , z − z sample

[0114] Where: M xyz (lz ) transfer function at wavelength x at pixel y of sensor z as a function of the path difference lz . M xyz max maximum value of the transfer function at wavelength x at pixel y of sensor z → constructive interference = path difference lz . = integer multiple of the wavelength λ x z mirror, z distance of the splitter mirror 90 from the mirror 60 or 70 (cf. Fig. 4 ) z sample distance of the splitter mirror 90 from the wafer surface 40 (cf. Fig. 4 )

[0115] If the maximum signal values ​​are determined for each of the three wavelengths x = 1, 2, 3, then M max< can be determined for each wavelength:< M xyz max = h xyz , href l z f ü r h max x − d xyz q xy , href ∗ r x , ref

[0116] This includes: I z for hmax (x) Path difference at which the signal at sensor z is maximum, i.e. I z is an integer multiple of the wavelength λ x.

[0117] The determination of M max< is carried out for each wavelength, for example, by continuously changing the distance z sample of the reference piece, and thus running through a full wavelength period for all three wavelengths x = 1, 2, 3 and both sensors z = 1, 2. The respective z sample,y positions at which the maximum value occurs can also be used to determine the size of the phase shift Δz r = z mirror1,y - z mirror2,y of the two mirrors 60 and 70 at each pixel y.

[0118] During the measurement run with the unknown wafer surface 40 to be examined, the sensor signals i z1 , i z2 and i z3 of the camera sensors 210 [z=1] and 220 [z=2] are now recorded simultaneously, as are the output intensities of the laser light sources 110, 120, 130 by means of the built-in monitor diodes 140, 150, 160.

[0119] The following applies to the signals at the sensors, line scan cameras 210 and 220: i xyz = q xy ∗ M xyz l z ∗ r xy , wafer + d xyz i xyz = q xy ∗ M xyz max ∗ 1 + cos 2 π λ x / 2 ∗ z mirrorz , y − z sample , y ∗ r xy , wafer + d xyz

[0120] Where: i xyz Interferometry measurement value at wavelength x at pixel y of the sensor zq xy irradiated light intensity of the measurement at wavelength x at pixel y M xyz max maximum value of the transfer function at wavelength x at pixel y of sensor z → constructive interference = path difference lz . = integer multiple of the wavelength λ x r xy, wafer reflection coefficient of the wafer 20 at wavelength x at pixel yd xyz read dark signal value (no excitation light) at wavelength x at pixel y of the sensor z

[0121] Using the relation Δz ry = z mirror1,y - z mirror2,y and the transformation Δz y = z mirror1,y - z sample,y, the equations for the two sensors can be simplified to: i xy 1 = q xy ∗ M xy 1 max ∗ 1 + cos 2 π λ x / 2 ∗ Δ z y ∗ r xy , wafer + d xy 1 and i xy 2 = q xy ∗ M xy 2 max ∗ 1 + cos 2 π λ x / 2 ∗ Δ z y − Δ z ry ∗ r xy , wafer + d xy 2

[0122] Since the determination of Δz y is sufficient for the profile measurement of the wafer surface 40, i.e. the detection of the surface profile of the wafer, and the distance differences Δz ry from the preliminary measurement for determining the maxima of the transfer functions M max< are known, this pair of signal equations (11), (12) defines for each pixel y a system of equations consisting of 6 equations (for the arguments x = 1, 2, 3 and z = 1, 2) for the 4 unknown quantities: the three reflection coefficients r xy,wafer the wafer surface 40 the desired distance difference Δz y , which results in the profile of the wafer surface 40

[0123] By suitable choice of the wavelengths λ 1 , λ 2 and λ 3 for x = 1, 2, 3, a working range of 0.5 mm with clear assignment of the two intensity measurement triplets can be achieved without difficulty. i xy 1 and i xy 2 to a path difference Δ zyThis is sufficient for a wide range of profile measurement tasks. This is explained in the aforementioned publication by K. Meiners-Hagen, R. Schrödel, F. Pollinger, and A. Abou-Zeid. In the arrangement disclosed therein, for example, the wavelengths 532 nm, 632 nm, and 780 nm are used, and a working range of 0.6 mm with unambiguous assignment of the distance difference is achieved.

[0124] To simplify the mathematical evaluation, the equations can be summarized: i xy 1 − d xy 1 i xy 2 − d xy 2 = M xy 1 max M xy 2 max ∗ 1 + cos 2 π λ x / 2 ∗ Δ z y 1 + cos 2 π λ x / 2 ∗ Δ z y − Δ z ry and with: P y = i xy 1 − d xy 1 i xy 2 − d xy 2 ∗ M xy 2 max M xy 1 max to: P y ∗ 1 + cos 2 π λ x / 2 ∗ Δ z y − Δ z ry = 1 + cos 2 π λ x / 2 ∗ Δ z y

[0125] This reduces the task for each pixel y to the determination of Δ zy from the remaining 3 equations, each of which must be fulfilled simultaneously. Due to the combination, the intensities q 1y , q 2y and q 3y of the laser sources are no longer directly included in the calculation and the reflection coefficients r xy,waferthe wafer surface 40 do not need to be explicitly determined.

[0126] If Δz y is represented in units of half a wavelength, Δz y can be written for each of the three wavelengths λ x as the sum of integer parts δ x and the rest fx be represented as: Δ z yx = λ x 2 ∗ δ x + f x

[0127] The desired path difference Δ zy is determined from the Δ z yx , by using the triple δ x integer parts, for which the mean deviation of the corresponding Δ z yx from the respective mean value is minimal. Use of a reference profile for the height measurement of recurring structures

[0128] An important application of the height profile measurement presented here is the measurement of the heights of contact structures on wafers. Here, only the height of the structure is of interest. This ensures that all contact structures actually make electrical contact during subsequent insertion and contacting of the chips in the package or during the 3D integration of multiple chips, and that none of the structures is so large that the chip is mechanically damaged during installation. For this purpose, it is necessary to determine whether the zenith points of all contact structures lie in one plane with a sufficiently small tolerance.

[0129] In order to solve this task quickly and reliably, it is proposed to carry out a high-resolution reference measurement on a few examples of the contact structures, to create a reference model of the contact structure from this and to record only as many measuring points for the routine measurement of similar wafers so that some of them are definitely located on the top side of the contact structure.

[0130] Such a reference model can be created using the same profile measurement method described above. For this purpose, in a first embodiment of the invention, the lens 50 can be designed as an interchangeable lens with different magnifications. The reference measurement can then be performed at a high magnification. The reference model can be generated, for example, by averaging or median calculation from several measured contacts.

[0131] The routine measurement can be carried out at a lower magnification so that just enough measuring points are determined on the contact surface to enable a reliable comparison with the reference model.

[0132] In another, alternative method, lens 50 can be a fixed lens with a fixed magnification. The reference measurement can be performed by multiple measurements with a small lateral offset. This, in turn, results in a denser measurement point coverage for the measured contact. The reference model can again be generated by averaging or median calculation from several measured contacts.

[0133] Comparing the zenith points with a reference model has the advantage of allowing a lower measurement point density. This eliminates the need to always locate the zenith point itself as the measurement point. Instead, it is sufficient to use a few points, e.g., 5 to 10 points on the contact surface. These can be located anywhere on the contact. The zenith height is then determined by adjusting the model height to the measured points. This allows for a lower measurement point density overall and thus higher throughput, while still reliably measuring the zenith heights of the contacts. Use for inspection

[0134] After determining the height profile of the surface, the values ​​Δ zy This allows the values ​​for the reflection coefficients r xywafer to be determined from the intensity data i xy1 of one of the channels according to equation (11): r xy , wafer = i xy 1 − d xy 1 q xy ∗ M xy 1 max ∗ 1 + cos 2 π λ x / 2 ∗ Δ z y

[0135] From this, a monochromatic image can be created for each wavelength λ x at the respective wavelength by merging the pixel values ​​calculated in this way for all y-locations.

[0136] To improve image quality, the intensity values ​​of both channels (z = 1, 2) can be used according to Eq. (18) and an averaged image can be created. r xy , wafer = 1 2 ∗ q xy ∗ i xy 1 − d xy 1 M xy 1 max ∗ 1 + cos 2 π λ x / 2 ∗ Δ z y + i xy 2 − d xy 2 M xy 2 max ∗ 1 + cos 2 π λ x / 2 ∗ Δ z y − Δ z ry

[0137] These images can be used for further tasks, such as searching for defects or checking the lateral alignment of the data acquisitions. Versions

[0138] Alternative embodiments are intended to be part of the invention without limiting its generality.

[0139] Thus, the use of TDI technology in the described invention merely serves to improve the signal-to-noise ratio and is not necessary for the inventive principle. For simpler requirements, an arrangement of three independent line-scan cameras or three independent, single-channel TDI line-scan cameras can be selected instead of TDI multi-channel camera line-scan sensors (line-scan cameras 210 and 220). The further spatial splitting required for this design can be achieved by increasing the respective distance between the prisms 250 and 270 and the tube optics 230 and 200, respectively. The tube optics 230 and 200 can also be designed as three individual optics for this purpose.

[0140] Alternatively, conventional area scan cameras or TDI single-channel line scan cameras operated in area readout mode can be used. With such an arrangement, the clock frequency is reduced accordingly, for example, to 1 kHz (for the described area readout mode of a TDI single-channel line scan camera), while still obtaining more than 16 million height values ​​per second.

[0141] It is understood that when using area scan cameras or TDI single-channel line scan cameras in area readout mode, the three wavelengths λ 1 , λ 2 , and λ 3 are imaged on the camera sensors onto different lines x = 1, 2, 3. The assignment of the obtained signal images, which each view the same point on the wafer surface 40 at the three wavelengths, is achieved by spatially assigning the camera sensor areas. For the basic evaluation described above, it is irrelevant whether the assignment is spatial (with area sensors) or temporally staggered (with line sensors).

[0142] It should be noted that for laterally high-resolution applications (e.g., in the single-digit µm range), the use of multiple cameras requires lateral and rotational correction for the recorded signal images anyway, since adjusting the entire array to a camera offset of less than 1 µm is mechanically difficult to achieve. Such a mathematical correction can usually be achieved by recording a reference pattern from which the exact location on the wafer viewed by each pixel can be determined.

[0143] In further design variants, the number of wavelengths used can be adapted to the required working range. For particularly small working ranges, a single-wavelength version is possible, or for smaller areas, two wavelengths. With only one wavelength, the inventive use of the two "sensor arms" shown serves to determine the reflectance of the sample and thus enables—in contrast to conventional multi-wavelength interferometry arrangements—the measurement of profiles with changing or unknown materials. For even larger working ranges or to improve reliability through measurement redundancy, expansion to more than three wavelengths is recommended, which can be achieved in particular with the aforementioned multi-block TDI camera (cameras with seven TDI blocks have already been presented).

[0144] In further design variants, the combination of two interferometry sensors (line scan cameras 210, 220) presented here can also be divided into a version of two consecutive measurements or implemented in two measuring heads that can be used one after the other.

[0145] Illumination can be achieved with continuously emitting, monochromatic light sources. Lasers are just as suitable for this purpose, as are other broadband beam sources, which can be combined with appropriate interference filters. The only requirement is that the coherence length of the light used is sufficiently large for the intended work area.

[0146] For the method according to the invention, it is sufficient that a phase shift is generated between the two beams, i.e., between the first mirror beam 490 and the second mirror beam 500 of the "reference arm," which is different from an integer multiple of the wavelength for all wavelengths used. This can be achieved, as shown in the exemplary embodiment, by a spatial offset of the mirrors 60 and 70. However, a mirror can also be provided with one or two different coatings to generate the phase shift. For example, a phase shift arises on an applied transparent layer due to interference of the beams reflected from the front and back of the layer.

[0147] Likewise, instead of the beam splitter 90, a bonded prism with a square cross-section can be used, which is provided with a partially mirror-coated layer in the bonded 45° area. This creates a partially transparent mirror. When using such a prism, the "reference arm" can be created by vapor-depositing a mirror coating directly onto the outer surface of the prism. By applying two different layer thicknesses or different materials, the required phase shift can be created. Although this arrangement leads to a reduction in the interference contrast due to the significantly different dispersion in the "reference and measuring arm" of the interferometer, it is advantageous for simple requirements due to its robustness. An embodiment with a bonded prism is described in the Figures 5 and 6 shown.

Claims

1. Optical device for detecting a surface profile of an object surface (30) of an object by means of interferometric distance measurement, comprising a beam splitter (3) for splitting a light beam (84) of a light source (7) into a first partial beam (410) and a second partial beam (420), wherein the light beam consists of one or several monochromatic light components; a beam splitter (4) for splitting each partial beam (410, 420) of the light source (7) into a reference beam (470, 480) and a measuring beam (430, 440); a mirror (5) for reflecting the two reference beams (470, 480), which is designed to generate a phase difference between the reflected beams (490, 500); a detector unit (8) for detecting evaluation beams (520, 510); and a signal evaluation unit (9) for evaluating the detected evaluation beams (520, 510) and determining the surface profile; wherein each measuring beam (430, 440) is directed for reflection onto a measuring area on the object surface (30) and, after reflection, is directed to the beam splitter (4) as an object beam (450, 460); each reference beam (470, 480) is reflected at the mirror and directed to the beam splitter (4) as a mirror beam (490, 500); the object beam (450, 460) and the mirror beam (490, 500) each interfere after impinging on the beam splitter (4) and are each fed as an evaluation beam (520, 510) to a detector unit (8) for evaluation; the signal evaluation unit (9) is designed to determine the intensities of the two evaluation beams (520, 510); and the signal evaluation unit (9) is designed to use the intensities of the two evaluation beams (520, 510) to determine a height profile of the surface (30), for which height profile the influence of the degree of reflection of the surface (30) on the measured intensities is computationally compensated, and in order to thereby determine a correct profile for object surfaces (30) with non-100% reflection and for object surfaces (30) with different reflection coefficients within an area to be measured, characterized in that the signal evaluation unit is further designed to generate a monochromatic inspection image for each wavelength from the determined reflection coefficients of the object surface (30) and to calculate the reflection coefficients on the basis of the previously determined height profile and the measured signal intensity of at least one of the beams (510 and 520) so that merely one measurement of the intensities of the two evaluation beams (520, 510) is sufficient for determining the height profile of the measuring area observed and for eliminating the influence of the degree of reflection.

2. Device according to claim 1, characterized in that the signal evaluation unit (9) is designed to determine the phase difference of the two evaluation beams (520, 510) and to determine the height profile of the surface (30) from the intensities and the phase difference of the two evaluation beams (520, 510).

3. System for detecting the surface profile of an object surface (30) of an object by means of interferometric distance measurement, comprising - a light source (7) for generating a monochromatic light beam (84); and - the optical device (2) according to claim 1 or 2.

4. Device or system according to one of the preceding claims, characterized in that the mirror (5) has two parallel mirror segments (60, 70) whose mirror surfaces are offset in the direction of the normal.

5. Device or system according to one of the preceding claims, characterized in that the beam splitter (4) and the mirror (5) are integrated in an optical element, preferably a prism (600).

6. Device or system according to the preceding claim, characterized in that the optical element, preferably the prism (600), is formed from two adjacent glass blocks and two mirror surfaces or mirrors (60, 70) are applied, preferably vapor-deposited, to one side of a glass block.

7. Device or system according to one of the preceding claims, characterized in that the beam splitter (4) comprises a semi-transparent mirror (90) which allows a part of an incident light beam (84, 410, 420) to pass as a reference beam (470, 480) and reflects another part of an incident light beam as a measuring beam (430, 440) towards the object surface.

8. System according to one of the preceding claims, characterized in that the light source (7) is a laser.

9. System according to one of the preceding claims, characterized in that several light sources (7) are provided for generating a monochromatic light beam (84), the light beams (84) of which have different wavelengths and preferably no wavelength of a light beam (84) is an integer multiple of the wavelength of another light beam (84) and the light beams (84) of which are preferably bundled in a beam coupler.

10. Device or system according to one of the preceding claims, characterized in that the detector unit (8) comprises a time-delayed integration camera, preferably a color time-delayed integration camera, particularly preferably a time-delayed integration multi-channel line camera.

11. Device or system according to claim 9, characterized in that it comprises an optical element for separating the bundled light beams, which is arranged such that the separation of the light beams takes place before they impinge on the detector unit (8).

12. Device or system according to one of the preceding claims, characterized in that an objective (50) is arranged between the optical device (2) and the object surface (30) of the object in order to direct the measuring beams (430, 440) onto a measuring region on the object surface (30).

13. Device or system according to the preceding claim, characterized in that the objective (50) is an interchangeable lens with different magnifications or a fixed lens.

14. System according to one of the preceding claims with a drive unit for generating a relative movement between the optical device (2) and the object.

15. Method for detecting the surface profile of an object surface (30) of an object by means of interferometric distance measurement comprising the following steps: Emitting a light beam (84) consisting of one or several monochromatic light components towards an optical device (2) by means of a light source (7); Splitting the light beam (84) of the light source (7) into a first partial beam (410) and a second partial beam (420) by means of a beam splitter (3); Splitting each partial beam (410, 420) of the light source (7) into a reference beam (470, 480) and a measuring beam (430, 440) by means of a beam splitter (4); Directing the reference beams (470, 480) to a mirror (5); Reflecting the reference beams (470, 480) on the mirror (5) and forming mirror beams (490, 500) which have a phase shift; Directing the mirror beams (490, 500) to the beam splitter (4); Reflecting the mirror beams (490, 500) at the beam splitter (4); Directing the measuring beams (430, 440) to a measuring area on the object surface (30); Reflecting the measuring beams (430, 440) on the object surface (30) and forming object beams (450, 460); Directing the object beams (450, 460) to the beam splitter (4); Interfering the mirror beams (490, 500) with the object beams (450, 460) and forming evaluation beams (510, 520), wherein the evaluation beams (510, 520) have a phase difference; Detecting the evaluation beams (510, 520) by means of a detector unit (8); Evaluating the detected evaluation beams (510, 520) and Determining the surface profile of the object surface (30) by means of a signal evaluation unit (9); Using the two measured intensity values of the evaluation beams (510, 520), both of which were reflected from the same location on the object surface (30), to compensate for the influence of the degree of reflection of the light beams on the object surface (30) when determining the surface profile and thereby to determine a correct profile for object surfaces (30) with non-100% reflection as well as for object surfaces with different reflection coefficients within an area to be measured, characterized in that the method further comprises the following steps: Generating a monochromatic inspection image for each wavelength on the basis of the determined reflection coefficients of the object surface (30), wherein the reflection coefficients are calculated on the basis of the previously determined height profile and the measured signal intensity of at least one of the beams (510 and 520) so that the correct profile is determined by just one measurement of the two evaluation beams (510, 520) and without adjusting the distance.

16. Method according to claim 15, characterized by the following steps: Evaluating the detected evaluation beams (510, 520) and determining the phase difference; and compensating for the influence of the degree of reflection of the light beams on the object surface (30) by means of the determined phase difference when determining the surface profile.

17. Method according to claim 15 or 16, characterized in that the signal intensities of both beams (510 and 520) are used.

18. Method for optically inspecting an object surface (30) for defects, wherein an error evaluation is based on an inspection image determined by means of the method according to claim 17 and on a subsequent back calculation of the reflection of the light on the object surface (30), wherein the reflection rxy is preferably obtained by: r xy = i xy − d xy q xy ∗ M xy max ∗ 1 + cos 2 π λ x / 2 ∗ Δ z y with ixy = measured value at wavelength x at a pixel y of a sensor; dxy = dark signal value (without excitation light); qxy = irradiated light intensity of the measurement; M xyz max = maximum value of the transfer function at wavelength x at pixel y of the sensor; and Δzy = zmirror,z - zsample. with zmirror,z = distance beam splitter (4) to reference mirror and zsample = distance beam splitter (4) to object surface.