INTERFEROMETRIC MEASURING DEVICE AND INTERFEROMETRIC METHOD FOR DETERMINING THE SURFACE TOPOGRAPHY OF A MEASURING OBJECT

DE502020009504D1Active Publication Date: 2024-11-07POLYTEC GMBH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
DE502020009504
Authority / Receiving Office
DE · DE
Patent Type
Patents
Current Assignee / Owner
Priority Date
2019-05-29
Filing Date
2020-05-26
Publication Date
2024-11-07
Estimated Expiration
2040-05-26

AI Technical Summary

Technical Problem

Interferometric measuring devices face challenges in accurately determining surface topography due to sensitivity to optical path length difference (OPD) fluctuations, which can be caused by vibrations, refractive index changes, and other uncontrolled influences, requiring complex and expensive stabilization measures.

Method used

A method and device that use a multi-element detector to capture light intensities at various OPD changes, incorporating additional detector elements to create a phase-shifted synthetic correlogram, allowing for robust evaluation of surface topography without precise knowledge of actual OPD changes, thereby reducing sensitivity to deviations in OPD fluctuations.

Benefits of technology

The approach significantly enhances the robustness and accuracy of surface topography determination, making it less dependent on precise OPD control and reducing the need for complex stabilization measures, leading to more reliable and cost-effective measurements.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader
Need to check novelty before this filing date? Find Prior Art

Description

[0001] The invention relates to an interferometric measuring device and an interferometric method for determining the surface topography of a measuring object.

[0002] Interferometric measuring devices are known for determining the surface topography of measurement objects. These devices focus a portion of the light from a light source on the surface of the measurement object, reflect it back as measurement light, and then combine it with another portion of the light source, serving as reference light, using suitable interferometer optics. The interference of the measurement and reference light creates an interferogram whose fringe structure allows conclusions to be drawn about the surface topography of the measurement object. In metrology, surface topography refers to the description of the geometric shape of the surface being examined, particularly in the form of one- or two-dimensional surface profiles.

[0003] There are numerous embodiments of such interferometric measurement methods in the prior art. In many of these, the optical path length difference (OPD) between the optical path of the measurement light and the optical path of the reference light is changed when recording the interferograms. This allows separate interferograms to be recorded for different path length differences (OPD). This results in an intensity curve for each individual point in the interferogram depending on the set OPD. Such an intensity curve, sampled for different OPD values, is usually called a "correlogram" for the corresponding point.Typically, each point in the interferogram is assigned a corresponding point on the surface of the measurement object, for example, through an optical image, so that the corresponding correlogram can then be assigned to a specific measurement point on the measurement object. Using the correlogram associated with the measurement point, the respective difference between the optical path lengths of the measurement and reference light is then determined in the subsequent evaluation, thereby determining the surface topography of the measurement object.

[0004] The surface topography of a measurement object is usually specified by the respective height values ​​hp< of a plurality of surface points p, thus their spatial position relative to a known reference surface, usually a plane. If it is available in this form, it can easily be specified relative to any other reference surfaces, provided that these other reference surfaces have a known spatial relationship to the first reference surface.

[0005] Therefore, for the sake of simplicity, the reference surface in interferometric measuring devices is often assumed to be the surface where, in a given reference state of the interferometer, the optical path length difference OPD between the measuring and reference arms is zero. This reference surface is often an optically conjugate plane to a reference surface of the interferometer, usually a mirror, and is often, but not always, represented by a plane in the object space of the measurement object. Based on the above, this surface can be used as the reference surface for the surface topography for all further considerations without loss of generality, since conversion to other reference surfaces is easily possible.

[0006] In the reference state of the interferometer, the OPD for each measuring point p on the measuring object is usually -2*n*hp< , i.e. it corresponds exactly to the negative height value of the corresponding measuring point p, which is doubled due to the forward and return travel of the light, if necessary corrected by the refractive index n of the material between the reference and measuring surface, which in the following, however, is assumed to be 1 for the sake of simplicity without loss of generality, since the slightly modified description for other values ​​is obvious to the person skilled in the art.

[0007] Common interferometric measurement methods for determining the surface topography of measurement objects are, on the one hand, phase-shifting interferometry (PSI) and, on the other hand, short-coherence interferometry, especially scanning white light interferometry (WLI).

[0008] In any case, for all relevant measuring points p on the measuring object for different OPOs changed by z 1 , z 2 , ... , zn compared to the reference state of the interferometer, the corresponding intensities I p< (z 1 ) , I p< (z 2 ), ... I p< (zn ) are recorded in the interferograms, so that for each measuring point p the corresponding correlogram c P< = (cp< (1), cp< (2), ..., cp< (n)) is obtained, where the measured light intensity at the i-th point of the correlogram is cp< (i) = lp< (zi ).

[0009] Preferably, the OPD is changed in a predetermined manner during the recording of the correlograms, usually such that z 1 , z 2 , ... , zn are equidistant with a known distance Δz. For this purpose, for example, either the length of the measuring arm or the reference arm of the interferometer is changed mechanically or by introducing material with a different refractive index. In some cases, the measurement object is moved relative to the interferometer or the interferometer is moved relative to the measurement object, which essentially also amounts to a change in the measuring arm length. The length change occurs either continuously or stepwise; it is only important that the respective OPD changes are known as precisely as possible, since the accuracy of the further evaluation and thus of the determined surface topography depends significantly on the exact knowledge of the respective OPD changes. This is also advantageous in the invention described below.

[0010] In phase-shifting interferometry (PSI), the phase difference PD between the measurement light and the reference light is varied, and the correlograms are recorded depending on the set phase difference. Typically, the phase variation is achieved, as described, by changing the optical path lengths. However, it can also be achieved, as described in US Pat. No. 4,594,003, by changing the wavelength of the illumination light used. All considerations made below—including the subsequent description of the invention—can be easily applied to the latter case. Therefore, for the sake of simplicity, the further description is limited, without loss of generality, to the more common case of phase variation by changing the optical path lengths.By using illumination light with a sufficiently long coherence length or by sampling only a few periods of the interference signal, the intensity of the measurement light changes sinusoidally with the set phase. By determining the phase position of the recorded sinusoidal signal as a function of the set path length for different measurement points, the differences in the OPOs between different measurement points and thus ultimately the surface topography of the measurement object compared to the reference surface are determined.

[0011] In short-coherence interferometry, especially scanning white-light interferometry, the OPD is also varied during the measurement. However, since short-coherence, i.e., spectrally broad, light is used here, the correlograms only show modulation when the measurement and reference light traverse approximately the same optical path length, i.e., the path length difference is small. However, if the OPD is larger than the coherence length of the light used, the modulation disappears.

[0012] Typically, the general functional correlogram curve c(OPD) can be described as a function of the set OPD by a sinusoidal modulation with a slowly changing envelope, which approaches zero for large OPD values: c OPD = A + B * env OPD * cos φ OPD

[0013] Here, env(OPD) is the slowly changing envelope, which can, for example, have the shape of a Gaussian curve env(OPD) = exp(-a*OPD 2< ), and ϕ(OPD) is a phase that changes depending on the OPD, in the simplest case ϕ(OPD) = 2π / λ*OPD, where λ is the effective wavelength of the illumination light and 2π / λ can be interpreted as the carrier frequency of the modulation. The constants A and B describe the offset of the signal and the associated modulation amplitude, respectively, whereby both are influenced, among other things, by the reflectivity of the measurement object at the measurement point. Both constants are not particularly relevant for the following considerations - including the following description of the invention - but they complicate the description considerably. Therefore, for the sake of simplicity of presentation, A=0 and B=1 are assumed below without loss of generality. The measurement signals can easily be converted into this form if necessary, e.g.by subtracting the mean value and appropriately normalizing the signal.

[0014] In this respect, the general functional correlogram curve c(OPD) is described below without loss of generality with c OPD = env OPD * cos φ OPD assumed.

[0015] Accordingly, to simplify the presentation and also without loss of generality, the light intensities lp< (zi ) = cp< (i) = c(zi -2*hp< ) are assumed below to be corrected for their offset, which is achieved by simply subtracting the offset of the respective correlogram, whereby this offset is most easily taken from the edge areas of the correlogram, where no or hardly any interference occurs, or is determined in another simple way, e.g. by subtracting the mean value (lp< (z 1 ) + lp< (z 2 ) + ... + lp< (zn )) / n of the light intensities of the correlogram, although more sophisticated options can be used if required, which, for example, take the phase position of the individual points into account. It should be noted that the correction of the light intensities for their offset, as well as their rescaling orNormalization represents a transformation of the n-dimensional light intensity vectors (lp< (z 1 ), lp< (z 2 ), ... , lp< (zn )) given by lp< (zi ) and is also to be understood as such in the sense of the following description and in particular the patent claims.

[0016] Typically, it is assumed, to a good approximation, that the correlogram curve for all measurement points, after deducting the offset and normalization, has the specified general functional form. In practice, there may be minor deviations from this, for example, due to different material dispersion on the path of the light to the various measurement points and the associated reference points. Relevant procedures are known for accounting for these effects and correcting their impact on the measurement results. For example, measuring these effects or effects resulting from them using a known measurement object, e.g., a precise plane mirror, and appropriately considering them when evaluating the measurements on the measurement objects to be measured. Such measures can, of course, also be carried out in conjunction with the inventive procedure described below, if necessary.

[0017] Typically, in the so-called "envelope evaluation," it is assumed that the envelope maximum is reached when the optical paths in the measurement and reference arms are of equal length, i.e., the OPD assumes the value zero. Accordingly, the envelope maximum is determined for each correlogram by determining the value zp< for the OPD change compared to the reference state of the interferometer at which the correlogram envelope maximum lies. Based on what has been said so far, the corresponding OPD is zero there, i.e., zp< - 2*hp< = 0, from which the height value hp< of the measurement point p can easily be determined. Should, in a situation deviating from the usual case, the OPD be different from zero at the location of the envelope maximum, the procedure would still be the same; only the right-hand side of the above equation would contain a value different from zero, but the height value hp< could still be determined as before.In any case, from the multitude of determined height values ​​hp< one ultimately obtains the spatial position of the measuring points p and thus the surface topography of the measuring object in comparison to the reference surface.

[0018] There are various approaches to determining the positions of the envelope maxima. A frequently used approach is to first determine the envelope itself from the correlograms, whose envelopes are initially modulated sinusoidally. To do this, for example, using a Hilbert transformation, one generates a synthetic correlogram s(OPD) = env(OPD) * sin(ϕ(OPD)) from the correlogram c(OPD) = env(OPD) * cos(ϕ(OPD)) that is phase-shifted by 90° in the carrier frequency. The two correlograms c(OPD) and s(OPD) are then squared and added together. After taking the square root, the desired envelope env(OPD) is obtained. Alternatively, the envelope env(OPD) can also be determined by piecewise fitting sine functions to the correlogram and point-by-point evaluation of the corresponding amplitudes.The position of the maximum of the envelope is then usually determined by fitting a suitable curve to the previously determined envelope env(OPD).

[0019] An alternative approach to determine the position of the envelope maximum first uses a Fourier transform of the correlogram and then determines the linear slope of the phase plotted against the spatial frequency at the location of the carrier frequency in Fourier space, from which the position of the envelope maximum can also be determined.

[0020] Once the positions of the respective envelope maxima have been determined, a fairly accurate determination of the surface topography of the measuring object can often be made, which is particularly useful for measuring objects with a rather rough surface.

[0021] In certain cases, especially for test objects with a relatively smooth surface, the accuracy of determining the surface topography can be significantly increased using so-called "phase evaluation." In addition to evaluating the position of the envelope maximum, the sinusoidal modulation of the correlogram near the envelope maximum is also used for evaluation. In this case, the first determined position of the envelope maximum is no longer used as the reference point with an explicitly and / or implicitly specified value for the OPD or PD. Instead, the much more precisely determined position closest to the envelope maximum with a specific phase position of the sinusoidal oscillation, for example, the position of its nearest descending zero crossing, is used.Comparing the OPD differences of the positions determined in this way from different measurement points yields a significantly more precise surface topography than using only the positions of the envelope maxima. For practical implementation, it is usually sufficient to simply determine the phase position of the sinusoidal modulation at the determined position of the envelope maximum. This then yields the corresponding position with sufficient accuracy for any phase position of the sinusoidal modulation near the envelope maximum, assuming the change in OPD across the correlogram measurements is known.

[0022] At this point, it is useful to point out that the evaluation methods of correlograms in short-coherent interferometry and phase-shifting interferometry are closely related. While the general functional course of a correlogram in short-coherent interferometry is of the type c OPD = env OPD * cos φ OPD , the corresponding correlograms in phase-shifting interferometry have a constant envelope env(OPD) = const. and a phase proportional to the OPD ϕ(OPD) = 2π / λ * OPD, where λ is again the wavelength of the illumination light. They are therefore of the type c OPD = const . * cos 2 π / λ * OPD , This ultimately represents only a special case of the correlograms of short-coherence interferometers, namely the special case in which the finite coherence length for the measurement performed is so large that it no longer needs to be taken into account. This allows most of the considerations or statements made below in connection with short-coherence interferometers to be immediately transferred to phase-shift interferometers, without the need for detailed reference.

[0023] The described procedures and considerations are well known and are used in many ways in many types of interferometers to determine surface topographies.

[0024] Finally, all types of known interferometers can be used for this purpose and in particular for the realization of the invention described below, particularly Michelson interferometers, Mach-Zehnder interferometers, Mirau interferometers, Linnik interferometers and Fizeau interferometers, but other types of interferometers are also used.

[0025] In order to accurately determine surface topographies using such interferometers and the described procedures, it is essential, according to the current state of the art, to control the OPD changes set or occurring during the acquisition of correlograms as precisely as possible. Even small deviations of the actual OPD changes from the expected or set OPD changes regularly lead to large deviations in the determined surface topographies compared to the actual surface topographies.

[0026] This fact is well known; therefore, when using interferometers for this purpose, appropriate measures are regularly taken to avoid unwanted and uncontrolled influences on the OPD during measurements, and if this is not possible, at least to determine them as best as possible and to take them into account in the evaluation.

[0027] Accordingly, the interferometers used are generally designed and constructed to be as little affected as possible by ambient vibrations, since such vibrations generally affect the measuring and reference arms of the interferometer differently, resulting in unknown and uncontrolled OPD changes. Vibration-isolated or even actively vibration-compensated tables are usually used as the base platform for the construction of such interferometers. For high-precision measurements, even air turbulence with the associated refractive index fluctuations in the measuring and reference arms is so disruptive that it must be prevented, e.g. by keeping the air temperature in the interferometer as constant as possible and / or by guiding the air flow as laminarly as possible. In extreme cases, the interferometer is sometimes even operated under vacuum conditions. The positioning andDisplacement units, which are used to adjust the OPD during the measurement, as well as the associated guide and fixation elements. Since inaccuracies at this point have a significant impact on the measurement accuracy of the overall system, complex and expensive components are typically used here.

[0028] However, in some cases, even these measures are not sufficient to sufficiently stabilize the OPD. In this case, in addition to the actual interferometer, additional measuring devices are integrated into the overall system. Their primary task is to measure the OPD fluctuations actually occurring during the measurement as precisely as possible so that their influences can then be taken into account when evaluating the correlograms. A comprehensive description of the measuring devices that can be used for this purpose and possible procedures for appropriately incorporating the determined results into the evaluation of the correlograms can be found, for example, in patent specification US 8,902,431. Various methods for evaluating the correlograms are known from documents US 2019 / 049236 A1, US 5,953,124 A, or E. Garbusi et al.: "Single frame interferogram evaluation," Applied Optics, Vol. 47, No. 12 (2008-04-20), page 2046.

[0029] All the measures described to prevent and / or determine undesired or uncontrolled influences on the OPD during interferometric measurements are complex, time-consuming and / or expensive.

[0030] The present invention is therefore based on the object of specifying an interferometric measuring device and an associated method which are sufficiently robust against OPD fluctuations so that the described complex measures for avoiding disturbing influences on the OPD can be largely dispensed with, whereby neither the disturbing OPD fluctuations have to be determined nor taken into account in the evaluation method.

[0031] This object is achieved by a method according to claim 1 and a device according to claim 15. Advantageous embodiments can be found in the dependent subclaims.

[0032] The method according to the invention for the interferometric determination of the surface topography of a measurement object comprises the following method steps: Generating illumination light and reference light by means of at least one light source and illuminating the measurement object with the illumination light, combining the illumination light reflected by the measurement object as measurement light with the reference light, whereby an interference pattern is generated in a detection area, changing the optical path length difference and / or the phase difference between the measurement light and the reference light, detecting light intensities lp< (zi ) of the interference pattern on a plurality of detector elements p of a multi-element detector in the detection area for at least two optical path length differences or phase differences changed by different amounts zi, determining that amount zp< for the change in the optical path length difference or phase difference for a plurality of detector elements p of the multi-element detector, for which in each case the optical path length difference orthe phase difference between the measuring light and the reference light reaches a value explicitly and / or implicitly specified for this detector element p, and determining the surface topography of the measuring object from the amounts zp< . . determined for different elements p of the multi-element detector.

[0033] It is essential that, in addition to the light intensities lp< (zi ), the light intensities lq< (zi ) of at least one other detector element q of the multi-element detector are also used to determine the value zp< assigned to the detector element p.

[0034] The light intensities Ip(zi) and Iq(zi) are thus determined using at least two different detector elements p and q. These detector elements p and q are assigned, via an optical image, points at different locations on the measurement object and / or an optical surface conjugated to the multi-element detector, preferably an optically conjugated plane. This optically conjugated surface to the multi-element detector is preferably a reference surface of the interferometer, a surface in the space of the measurement object, an intermediate image surface of the optical image, a pupil, and / or an aperture plane of the optical image.

[0035] In the method according to the invention, the light intensities lp(zi) of the interference pattern are detected on a large number of detector elements p of a multi-element detector in the detection area for at least two optical path length differences or phase differences changed by different amounts zi. It is advantageous to carry out the detection for a larger number of optical path length differences or phase differences changed by different amounts zi, in particular for at least 3, preferably at least 10, particularly preferably at least 20 different zi. Typically, a smaller number of measuring points is sufficient for PSI methods compared to WLI methods. It is therefore advantageous, when designing the method according to the invention as a WLI method, to carry out the detection for a larger number of optical path length differences or phase differences changed by different amounts zi.To carry out phase differences, in particular for at least 5, preferably at least 20, particularly preferably at least 50 different zi.

[0036] It is within the scope of the invention that analog variables are used for the variables used to describe the method according to the invention, which - optionally together with other variables or parameters - have the information content of the named variable and can thus be converted if necessary. This applies in particular to the variable z and the absolute value zp< assigned to a detector element. For example, if a constant rate of change of the change in the optical path length difference and / or phase difference is known, a time of measurement tp< can also be specified, which can be converted into the absolute value zp< if necessary. This also applies to the device according to the invention described below and to the advantageous embodiments of the method according to the invention and the device according to the invention.

[0037] The device according to the invention described below is preferably designed to carry out the method according to the invention, in particular a preferred embodiment thereof. The method according to the invention is preferably designed to be carried out using the device according to the invention described below, in particular a preferred embodiment thereof, such as the advantageous configurations described below: The interferometric measuring device according to the invention for determining the surface topography of a measurement object has the following features: The interferometric measuring device according to the invention comprises at least one light source, for example a laser, an LED, a superluminescent diode, a light bulb, a gas discharge lamp, an arc lamp, a plasma radiation source, etc.The measurement object is illuminated with a portion of the light from the light source, preferably using suitable illumination optics, most preferably using illumination optics that image the light source into an aperture plane of an objective, for example, a Köhler illumination optic. Reference light is generated with a portion of the light from the light source, preferably by splitting the light from the light source into a first portion, which serves to illuminate the measurement object, and a second portion, which serves as reference light, using a beam splitter, e.g., a neutral splitter or a polarization splitter.

[0038] With the aid of interferometer optics, the illumination light reflected by the measurement object as measurement light through reflection or scattering is combined with the reference light, preferably by combining the measurement light and reference light using a beam splitter, but at least by directing the measurement and reference light to a common detection area, where an interference pattern is created by the interference of the measurement and reference light. Such interferometric measuring devices are well known and used in a variety of ways. Thus, ultimately, all types of known interferometers can be used to implement the present invention; particularly noteworthy are Michelson interferometers, Mach-Zehnder interferometers, Mirau interferometers, Linnik interferometers, and Fizeau interferometers.However, the present invention is not limited to these interferometer types; any other interferometer types can also be advantageously used within the scope of the present invention without limitation. The interferometer optics thus preferably comprise one or more optical elements from the list of optical lenses, beam splitters, filters, apertures, mirrors, polarization filters, retardation plates, and other elements that influence the polarization of light.

[0039] The interferometric measuring device according to the invention further comprises at least one adjustment unit, by means of which the optical path length difference OPD or the phase difference PD between the measuring light and the reference light can be changed by an amount z. For this purpose, either the optical path length of the measuring light or the reference light or the optical path length of both is changed. To change the optical path length of the measuring light, for example, the measurement object can be moved relative to the interferometric measuring device or, with the same effect, the interferometric measuring device can be moved relative to the measurement object. It is also possible to change the optical paths of the illumination light and / or the measuring light such that the OPD and / or the PD changes. For this purpose, movable mirrors, movable lenses, or other movable beam guidance elements can be used, for example.It is also possible to insert elements into the beam path of the measuring light that change the optical path length by altering the optical refractive index. Such elements can be, for example, different flat plates made of optical material, or, if a continuous change in the optical path is desired, wedge plates made of optical material that can be moved relative to one another. Optical elements such as LCD elements, Pockels cells, etc., in which the refractive index changes when electrical voltages are applied, are also conceivable.

[0040] The same variety of possibilities exists and is preferably used to change the optical path length of the reference light. Particularly preferred is an arrangement in which the reference light is reflected or scattered by a reference surface, most preferably by a reference mirror, whereby this reference surface can be moved, thereby changing the optical path length of the reference light.

[0041] To change the optical path length of the measurement and / or reference light, mechanical adjustment units for moving components can be used particularly advantageously within the scope of the description. Motor-driven spindles with stepper or DC motors with or without encoders and tachometers, linear units, piezo-driven stages or stepping motors, and many other drive units are particularly suitable for this purpose. The adjustment units used particularly preferably comprise their own measuring system, or a measuring system is provided so that the respective position of the adjustment unit and thus the change in the optical path length difference OPD and / or the phase difference PD can be monitored or determined as precisely as possible. Control systems, such as PID controls for precise positioning or for achieving the best possible synchronization of the adjustment unit, are also extremely advantageous.In simple embodiments, it is also sufficient to know the start and end points of the adjustment unit in order to be able to infer those points in time at which the OPD and / or the PD change reaches a certain desired, explicitly and / or implicitly specified value, or to be able to infer the amount z of change in OPD or PD present at certain points in time during the adjustment, given a reasonably uniform adjustment speed.

[0042] If only the phase difference PD between the measurement and reference light is to be changed, but not the optical path length difference OPD, then the wavelength of the light from the light source is preferably changed. This can be achieved, for example, with some light sources such as laser diodes or LEDs by changing the operating temperature and / or by changing the applied current, or with lasers, for example, by changes in the laser cavity, such as by changing the length or by influencing it with spectrally filtering elements. Such light sources with variable wavelengths are commercially available in many different designs. With broadband light sources, the wavelength can ultimately also be influenced by using variable spectral filters.

[0043] Regardless of which of the many options is used to adjust the optical path length difference and / or the phase difference between the measurement light and the reference light, it is important to control or at least know the amount z by which the OPD or PD is changed. To achieve this, a relationship is advantageously established between the adjustment parameter(s) of the adjustment unit and the corresponding change in the OPD or PD.If the system parameters are known, this is usually very simple. For example, when using a piezo-driven retro-reflecting mirror in the reference arm as an adjustment element for the optical path length of the reference light, this reference is given on the one hand by the calibration of the piezo actuator, with the aid of which the applied voltage U is preferably converted into a displacement distance w(U) of the mirror, and on the other hand by the knowledge that with a displacement w(U) of the mirror, the optical path of the reference light changes by 2*w(U) due to the influence of both the outward and return paths, or by 2*n*w(U) if the refractive index n of the medium through which it passes, such as air, cannot be neglected. Since such considerations are very easy to implement in most other cases and are also very familiar to those skilled in the art, we will refrain from further discussion here without loss of generality.In cases where the relationship between the adjustment parameter or the adjustment parameters of the adjustment unit and the amount z of change in the OPD or PD cannot be established so easily or not with sufficient accuracy, an additional measuring system is preferably provided, with the aid of which the amount z of change in OPD or PD is then directly measured and / or controlled, for example by means of a suitable length or speed measuring system.

[0044] In all cases, it is assumed that the amount z by which the optical path length difference OPD or the phase difference PD between the measuring light and the reference light is changed by the adjustment unit is known with a certain, in many cases sufficient, accuracy.

[0045] However, it should be noted that the accuracy with which the amount z of the OPD or PD change is known in real measurements is always finite. This means that although knowledge of this amount z is available at any given time, the known value never fully corresponds to the actual value. Nevertheless, within the scope of the present invention, just as in the prior art, this only finitely precise value of z is used for the evaluation, at least as long as no better value is known. However, the device and method according to the invention differ from the prior art in that the evaluation according to the invention is significantly less sensitive and more robust with regard to the unavoidable deviations of the real values ​​of z compared to the known values ​​of z, as will be explained in more detail below.

[0046] The interferometric measuring device according to the invention further comprises at least one multi-element detector with a plurality of detector elements p, which is designed to detect the light intensities lp< (zi ) of the interference pattern on the detector elements p in the detection area for at least two optical path length differences or phase differences changed by different amounts zi . Cameras, preferably CCD or CMOS cameras, are particularly suitable for this purpose. Photodiode arrays or lines can also be used advantageously, particularly in cases where high speed is more important than a large number of detector elements. It is important that the detector elements detect the light intensities incident on them at times at which the respective amount zi , by which the optical path length difference or phase difference has changed compared to the reference state of the interferometer, is known as precisely as possible.For this purpose, the adjustment unit is preferably positioned first, and then a trigger signal is provided to the multi-element detector, for example, from the adjustment unit or a control unit controlling it. However, it is also possible, for example, in an advantageous embodiment, for the adjustment unit to move continuously, and at times when it passes through defined points, the adjustment unit, its control unit, or a measuring system associated with the movement of the adjustment unit sends trigger signals to the multi-element detector that initiate image acquisition.Alternatively, it is also conceivable for the multi-element detector to control its image acquisition independently or for this to be controlled by a third unit, in which case the signals from the multi-element detector or the third unit are used to read out the positions of the adjustment unit at the appropriate times, either directly or via a suitable measuring system. There are also a multitude of other ways in which synchronization of image acquisition and adjustment position can be achieved, all of which can be used within the framework of the device according to the invention. The important thing is that the light intensities detected by the multi-element detector can be assigned as precisely as possible to the respective corresponding amount zi by which the optical path length difference or the phase difference was changed compared to the reference state of the interferometer.In order to achieve the best possible resolution in determining the surface topography of the measurement object, the multi-element detector preferably has a large number of detector elements, particularly advantageously more than 100, preferably more than 10,000, most preferably more than 100,000 detector elements.

[0047] At this point, it should be noted that in most cases, each detector element p of the multi-element detector is assigned a corresponding point on the surface of the measurement object through optical imaging by means of the interferometer optics. For the sake of simplicity, this point is also called p, which generally does not lead to confusion. Given this assignment, the height value hp< of point p on the surface of the measurement object can be determined from the absolute value zp< for the change in the optical path length difference or phase difference compared to the reference state of the interferometer, at which the optical path length difference or the phase difference between the measurement light and the reference light for the detector element p reaches the specified value.If such a direct assignment between detector element p of the multi-element detector and a corresponding point on the measurement object surface does not exist because the detector element is not imaged onto the measurement object by the interferometer optics, but rather onto another location, for example into the pupil of the imaging optics, then the procedure according to the invention can still be used in the same way, but in this case the surface topography is not determined by directly assigning the value zp< for the detector element p to a height value hp< of the corresponding measurement object point p. Instead, the surface topography in this case is carried out by transforming the determined height or phase profile hp< in the surface optically conjugated to the detector elements onto the measurement object, the corresponding procedure being known to the person skilled in the art and therefore not explained further here.The procedure according to the invention can also be used in this case in the same way as in the more clearly illustrated case in which each detector element p is directly assigned a measurement object point p.

[0048] The interferometric measuring device according to the invention further comprises at least one evaluation unit which is connected to the multi-element detector. Particularly advantageously, the evaluation unit is additionally connected to the adjustment unit, its control unit and / or a measuring system assigned to the adjustment unit or a movement of the adjustment unit, whereby the evaluation unit can be provided with information about the changes carried out by the adjustment unit. The evaluation unit is preferably a computer, a compute module, a microprocessor, a DSP, an FPGA, etc. It is essential that the evaluation unit is designed to determine, for a large number of detector elements p of the multi-element detector, the amount zp< for the change in the optical path length difference or phase difference compared to the reference state of the interferometer at which the optical path length difference or phase differencethe phase difference between the measuring light and the reference light reaches a specified value, and from this the surface topography of the measuring object is determined.

[0049] For this purpose, the light intensities lp< (zi ) of the interference pattern detected by the detector elements p of the multi-element detector are transmitted to the evaluation unit for a large number of adjustment positions of the adjustment unit, but at least for two adjustment positions. This provides the evaluation unit with the corresponding correlograms cp< = (cp< (1), cp< (2), ..., cp< (n)) = (lp< (z 1 ), lp< (z 2 ), ..., lp< (zn )) for a large number of optical path length differences or phase differences between the measuring light and the reference light that have changed by various amounts z 1 , z 2 , ..., zn compared to the reference state of the interferometer.

[0050] Typically, from these correlograms, according to the described procedure according to the state of the art, those amounts zp< for the change in the optical path length difference or the phase difference compared to the reference state of the interferometer are determined at which the optical path length difference or the phase difference between the measuring light and the reference light reaches a predetermined value - in particular the preferably predetermined value zero - and from this the surface topography of the measuring object is determined.

[0051] The specified value can be specified explicitly or implicitly, as described below.

[0052] The predetermined value is preferably specified as a constant value; particularly preferably, as described above, the value zero is specified. It is within the scope of the invention for a different value to be specified for each detector element p. Preferably, the same value, in particular the value zero, is specified for a plurality of detector elements, particularly preferably for all detector elements.

[0053] It is therefore within the scope of the invention that a constant value such as "zero" (for the path length difference) or "0°" (for the phase difference) is explicitly specified, whereby zp< is determined such that at z = zp< precisely this value is reached. It is also within the scope of the invention for the value to be specified implicitly, in particular by specifying it such that at z = zp< a specific point in the correlogram is reached, preferably an envelope maximum or another prominent point in the correlogram and / or a specific phase position of the correlogram near the envelope maximum or another prominent point in the correlogram. Such an implicit specification also specifies a value for the optical path length difference or phase position - even if this is not explicitly known at the time of specification or is subsequently determined.

[0054] The evaluation unit requires information about the amounts zi by which the optical path length difference or the phase difference was changed compared to the reference state of the interferometer during the acquisition of the light intensities lp< (zi). This information is usually available to the evaluation unit, for example, either because the amounts zi are known through a predetermined control of the adjustment unit of the interferometric device, or because the adjustment unit and / or a measuring system associated with it transmits this information to the evaluation unit via a suitable data or trigger line.Even in the event that the adjustment unit and / or a measuring system assigned to it triggers the data acquisition of the multi-element detector by means of a trigger line, this information can be made available to the evaluation unit, for example by making the configuration of the trigger times dependent on the adjustment position of the adjustment unit known to the evaluation unit or by defining it. In addition to the implementations described, there are a multitude of further possibilities for how the evaluation unit can obtain the information about the amounts z 1 , z 2 , ..., zn associated with the correlograms cp< = (cp< (1), cp< (2), ..., cp< (n)) = (lp< (z 1 ), lp< (z 2 ), ..., lp< (zn )), all of which can also be advantageously used within the scope of the present invention.

[0055] It should be noted that, at least when using state-of-the-art procedures for further evaluation, the most precise knowledge possible of the magnitudes z 1 , z 2 , ..., zn is essential for the quality of the results. The known procedures assume that the magnitudes z 1 , z 2 , ..., zn known to the evaluation unit are also their real values ​​and use these for further calculations. This leads to problems, for example, particularly when a Fourier or Hilbert transform is used in connection with the determination of the envelope and / or the envelope maximum and / or the phase of a correlogram.Although there are mathematical methods that can be used to perform such transformations for signal waveforms sampled at non-equidistant sampling points, these, like conventional transformations, lead to erroneous results if the assumed sampling points of the signal waveforms to be transformed do not correspond to the actual sampling points. Similar problems also exist with all other known procedures for determining envelopes, envelope maxima and / or phase positions of correlograms if the assumed sampling points do not correspond to the actual ones, regardless of whether they are equidistant or not. For example, fitting curves, especially sinusoidal curves, to the signal waveforms of correlograms leads to large deviations if the sampling points at which the fitting function and its deviation from the measured data are calculated are not the correct ones.

[0056] Of course, it is also advantageous for the procedure according to the invention if the amounts z 1 , z 2 , ..., zn are known as precisely as possible, but the procedure according to the invention is considerably more robust against deviations of the known values ​​for the zi from the corresponding real, but unknown values ​​of the zi .

[0057] Such deviations can occur, for example, because (i) due to vibrations or movements, the length of the measuring and / or reference arm of the interferometer is influenced in an uncontrolled manner during the measurements, (ii) changes in the density and / or the refractive index of at least one of the media traversed by the light in the measuring and / or reference arm influence the measurements in an uncontrolled manner, or (iii) inaccuracies and / or errors occur when determining the amounts zi.

[0058] In the interferometric measuring device according to the invention, the evaluation unit is now designed such that, in order to determine the value zp assigned to the detector element p, it uses not only the light intensities lp(zi) on the detector element p but also the light intensities lq(zi) of at least one other detector element q of the multi-element detector.

[0059] The advantage of this becomes very clear from a simple but relevant consideration: In most cases, the height values ​​hp and hq of directly adjacent surface points p and q of the measuring object are not exactly the same, but very similar.

[0060] Accordingly, the corresponding correlograms cp< = (lp< (z 1 ), lp< (z 2 ), ..., lp< (zn )) and cq< = (lq< (z 1 ), lq< (z 2 ), ..., lq< (zn )) differ only slightly. The individual values ​​cp< (i) and cq< (i) can be represented using the general functional correlogram curve c(OPD) as cp< (i) = c(zi -2*hp< ) and cq< (i) = c(zi -2*hq< ).

[0061] Since hp< and hq< differ only slightly, hq< can be written as hq< = hp< + Δ, where Δ is very small. c q i = c z i − 2 * h q = c z i − 2 * h p − 2 * Δ ≈ c z i − 2 * h p − 2 * Δ ∗ c ′ z i − 2 * h p , where c'(zi -2*hp< ) is the derivative of the general functional correlogram c(OPD) at the point zi -2*hp<.

[0062] This means that by taking the difference between the two correlograms cq< and cp<, up to a simple scaling factor, one can immediately determine the derivative of the correlogram for the measuring point p at exactly the points zi -2*hp< at which the correlogram itself was recorded.

[0063] Now, for the typical general functional correlogram curve c(OPD) = env(OPD) * cos (ϕ(OPD)) of short-coherent interferometry, one can assume in good approximation that the envelope changes much more slowly than the sinusoidal modulation, which means that for the derivative formed according to the product and chain rule c ′ OPD = env ′ OPD * cos φ OPD − env OPD * sin φ OPD * φ ′ OPD the first term can be neglected, which is why, under the realistic assumption that ϕ(OPD) depends approximately linearly on the OPD, i.e. ϕ'(OPD) is constant, the derivative c'(OPD) is proportional to the synthetic correlogram s(OPD) = env(OPD) * sin (cp(OPD)) which is phase-shifted by 90° in the carrier frequency.

[0064] The same applies, of course, not only to short-coherent interferometry, but also in particular to phase-shifting interferometry, since here the conditions env'(OPD)=0 and the constancy of ϕ'(OPD), which are only approximately valid for short-coherent interferometry, even apply exactly.

[0065] In summary, this means that by simply calculating the difference between the correlograms cp< and cq< up to a scaling factor f, the synthetic correlogram, which is phase-shifted by 90° in the carrier frequency compared to the correlogram cp<, can be obtained. s p = ( s p ( 1 ) , s p ( 2 ) , … , s p ( n ) ) = f * ( Q p ( z 1 ) , Q p ( z 2 ) , … , Q p ( z n ) ) = f * ( l q ( z 1 ) − l p ( z 1 ) , q ( z 2 ) − l p ( z 2 ) , … , l q ( z n ) − l p ( z n ) ) where Q p< (zi ) = lq< (zi )-lp< (zi ) represents a signal which is phase-shifted in its modulation compared to the light intensities lp< (zi ), as described. This synthetic correlogram sp< shifted in the carrier frequency is sampled at exactly the same points zi at which the original correlogram cp< was sampled, whereby no knowledge of the values ​​zi is used to generate this synthetic correlogram. As a result, the generation of the synthetic correlogram sp< corresponding to the correlogram cp< using the procedure according to the invention is independent of the exact knowledge of the real values ​​of the parameters zi , which makes the evaluation significantly more robust against all types of interference compared to previously known methods.

[0066] In particular, it is advantageous to use a direction or tangential vector determined as described below to determine a phase-shifted signal Q p< (z 1 ), Q p< (z 2 ), ... , Q p< (zn ) for the light intensities lp< (z 1 ), lp< (z 2 ), ... , Q p< (zn ) or to determine a corresponding associated phase-shifted signal for the vectors lp< (z 1 ), lp< (z 2 ), ... , lp< (zn ) resulting from projection and / or transformation.

[0067] By means of the light intensities lp< (z 1 ), lp< (z 2 ), ... , lp< (zn ) and the phase-shifted signals Q p< (z 1 ), Q p< (z 2 ), ... , Q p< (zn ) it is thus possible to carry out further steps known from the prior art, in particular to determine an envelope and / or phase position of the signal curve given by the light intensities. It is therefore advantageous to use the light intensities lp< (z 1 ), lp< (z 2 ), ... , lp< (zn ) together with the phase-shifted signals Q p< (z 1 ), Q p< (z 2 ), ... , Q p< (zn ) to determine an envelope and / or a phase position of the signal curve given by the light intensities lp< (z 1 ), lp< (z 2 ), ... , lp< (zn ) or to carry out the same for vectors resulting from the lp< (z 1 ), lp< (z 2 ), ... , lp< (zn ) by projection and / or transformation and the associated phase-shifted signals.In particular, it is advantageous to use the determined phase position and / or envelope to determine for the element p of the multi-element detector the amount zp< for the change in the optical path length difference or phase difference at which the optical path length difference or the phase difference between the measuring light and the reference light reaches an explicitly and / or implicitly predetermined value.

[0068] It is therefore advantageous that the light intensities lp< (z 1 ), lp< (z 2 ), ... , lp< (zn ) together with the phase-shifted signals Q p< (z 1 ), Q p< (z 2 ), ... , Q p< (zn ) are used to determine an envelope and / or a phase position of the signal curve given by the light intensities lp< (z 1 ), lp< (z 2 ), ... , lp< (zn ) or to carry out the same for vectors resulting from the lp< (z 1 ), lp< (z 2 ), ... , lp< (zn ) by projection and / or transformation and the associated phase-shifted signals.

[0069] For further evaluation, the scaling factor f is preferably determined so that the correlogram sp< has the same scaling as the correlogram cp<. There are a number of ways to do this; a relatively simple one is to optimize the factor f so that (cp< ) 2< + (sp< ) 2< , which is the square of the envelope of the correlogram, has as flat a curve as possible - for this, cp< must of course be the correlogram corrected for its offset, whereby in the simplest case the offset is taken from the edge regions of the correlogram where little or no interference occurs, or alternatively the offset is determined by suitable averaging of the individual correlogram values.Instead of optimising the factor f by making the square of the correlogram envelope as flat as possible, the same can of course also be done for the square root of this, which results in the envelope itself, or any other suitable function thereof. Unless the case is so simple that the optimum value of f can be determined directly from the available data, e.g. using a formula, it is advantageous to use any mathematical optimisation method such as the gradient method to determine the parameter f in such a way that a given evaluation function b(f), which appropriately evaluates the deviation from a flat curve, is minimal. An advantageous example of such an evaluation function is the sum . b f = ∑ i = 2 n l p z i 2 + f * Q p z i 2 − l p z i − 1 2 + f * Q p z i − 1 2 2 , With the help of which the deviation of the squared envelope for each point i of the correlogram from that of the previous point i-1 is determined by difference, these deviations are then squared and subsequently summed over all points. For the optimal factor f, this b(f) is minimal.

[0070] It should be noted that the lp< (zi ) for this are the light intensities of the correlogram at point p corrected by their offset as described above.

[0071] It is even simpler to use only the points in the vicinity of the correlogram maximum to determine the factor f and to assume that for these points (cp< ) 2< + (sp< ) 2< must be almost constant, so that the standard deviation of the values ​​lp< (zi ) 2< +(f*Q p< (zi )) 2< determined from the points in the vicinity of the envelope maximum or the square root thereof is preferably used as the evaluation function.

[0072] A preferred procedure for determining the scaling factor f, particularly in phase-shifting interferometry, is to determine not only the value for f but also a value for the offset o of the correlogram cp< . To do this, one uses one of the known mathematical optimization methods to simultaneously optimize the two parameters f and o such that the envelope square (lp< (zi ) -o) 2< + (f*Q(zi )) 2< or a function thereof shows the flattest possible curve as a function of i or, in the case of phase-shifting interferometry, is even as constant as possible. For this purpose, a suitable evaluation function is preferably used, as described above, which evaluates the deviations from the desired flat curve or a constant value.

[0073] Another preferred approach to determining the factor f is to fit a realistic envelope function to the envelope determined as a function of the factor f, for example, a theoretically known envelope function or one determined and / or approximated from measured data, and to use the sum of the corresponding squared deviations ("chi-square") as an evaluation function as a function of f.

[0074] In addition to the described possibilities for determining the scaling factor f, there are any other possibilities, and it is within the scope of the present invention to use such possibilities as well, provided that they provide a sufficiently accurate value for the required scaling factor f.

[0075] The Q p< (zi ) thus correspond to the synthetic correlograms sp< except for the scaling factor f.

[0076] If, after determining the appropriate scaling factor f for each correlogram cp<, the corresponding and correctly scaled synthetic correlogram sp< shifted in the carrier frequency has also been determined, the further evaluation is preferably carried out according to or based on the known procedures according to the state of the art.

[0077] It is therefore advantageous to scale the direction vector, tangential vector and / or the determined phase-shifted signal Q p< (z 1 ), Q p< (z 2 ), ... , Q p< (zn ) determined as described above in such a way that the amplitude and / or the envelope of the carrier frequency thereof corresponds to the amplitude or envelope of the carrier frequency of the light intensities lp< (z 1 ), lp< (z 2 ), ... , lp< (zn ).

[0078] For example, in a preferred embodiment, the two correlograms cp< and sp< are squared and added together, and the square root is taken to determine the envelope of the correlogram. The position of the envelope maximum can then be determined, for example, by fitting a suitable curve to it. The inventors' investigations have shown that, particularly in the presence of one of the aforementioned disturbances, the envelope determined using the inventive method is significantly smoother and more realistic than with previously known methods, and the position of the envelope maximum is significantly more precise.

[0079] Even the more precise phase evaluation of the correlograms, in which in addition to the evaluation of the position of the envelope maximum, the sinusoidal modulation of the correlogram in the vicinity of the envelope maximum is also used for the evaluation, produces considerably better results using the procedure according to the invention. Because for each correlogram point cp< (i) the corresponding point sp< (i) of the correlogram shifted by 90° in the carrier frequency is determined independently of the knowledge of the amount of the OPD or phase change zi set by the adjustment unit, but nevertheless at exactly the same point zi, a precise phase value ϕ p< (zi ) for each OPD or phase change is obtained by means of ϕ p< (zi ) = ϕ p< (i) = unwrap( arctan2(sp< (i),cp< (i)) ).PD change zi, where arctan2 is the variant of the arctangent function that covers a value range of 360°, i.e. extends over all four quadrants of the unit circle, and unwrap() denotes a phase unwrapping carried out after the arctangent formation, which is preferably chosen so that at the position of the previously determined envelope maximum ϕ p< (z) assumes a value as close as possible to 0°.

[0080] If one now knows the phase curve ϕ p< (zi ), which as a function of the zi increases or decreases monotonically with a relatively constant gradient, one can quite easily determine the value zp< at which the OPD or PD reaches an explicitly and / or implicitly specified value using linear regression (or if necessary in the case of possible deviations from linearity, for example due to dispersion etc., by curve fitting), e.g. by ϕ p< (z) assuming an explicitly and / or implicitly specified value there, e.g. 0° for the maximum of the sinusoidal modulation closest to the envelope maximum or 90° for the descending zero crossing of the sinusoidal modulation closest to the envelope maximum.

[0081] Advantageously, in order to determine the value zp< assigned to the multi-element detector element p, information about an envelope and / or a phase position of the function lp< (z) which is dependent on the change z in the optical path length difference and / or the phase difference between the measuring light and the reference light is thus determined, in particular by using the light intensities lq< (zi ) of at least one other detector element q of the multi-element detector in addition to the light intensities lp< (zi ) to determine this information.

[0082] In this determination of zp<, possible inaccuracies in the values ​​zi used for the evaluation are only insignificant, which makes the method very robust against disturbances and ultimately leads to excellent results.

[0083] For the sake of completeness, it should be pointed out that the deviations of the points with the coordinates (zi , ϕ p< (zi )) from the course of the regression line or the fit curve are advantageously used to determine corrections Δz i for the zi and thus ultimately to obtain more precise zi values, e.g. in that in an advantageous embodiment the correction values ​​Δz i are determined such that the points (zi +Δz i , ϕ p< (zi )) lie exactly on the regression line or the fit curve. In particular by combining the correction values ​​Δz i obtained for different detector elements p, e.g. by means of suitable averaging methods, possibly including suitable weighting, the available zi values, which may originally deviate significantly from the real zi values ​​due to interference, can be corrected in such a way that the known evaluation methods also lead to better results.In this respect, the procedure according to the invention can be combined with the previously known procedure in order to achieve better results there as well.

[0084] In principle, the now corrected zi values ​​can, of course, also be used as a new starting point for the inventive evaluation, which can even be developed into an iterative process through repetition. However, the inventors' investigations show that calculating and using corrected zi values ​​in conjunction with the inventive procedure offers hardly any significant advantages, since the inventive procedure is already very robust against disturbances, so that it can typically be dispensed with.

[0085] At this point, it should be emphasized again that, although the procedure described so far has been explained using the example of short-coherent interferometry, it can easily be transferred to phase-shifting interferometry. Of course, in phase-shifting interferometry, there is no envelope of the correlogram, or rather, this is a constant, but this eliminates only the step of determining the envelope maximum, and all approximations that were necessary in the context of short-coherent interferometry are even valid exactly. The actual procedure, in which the correlograms of different detector elements p and q are combined, remains completely intact in all details and ramifications, so that the procedure according to the invention, in particular according to the described preferred embodiments, fully covers not only short-coherent interferometry but also phase-shifting interferometry.

[0086] At this point, it should be emphasized again that the synthetic correlogram sp< = (sp< (1), sp< (2), ..., sp< (n)) determined as described above and below in the case of short-coherence interferometry represents a correlogram with the same envelope and phase shifted by 90° in the carrier frequency compared to the correlogram cp<, although this is a very good and, in most cases, useful approximation. However, this is not exactly correct. As previously shown, this only applies if the envelope env(OPD) - as in phase-shifting interferometry - changes infinitely slowly over the general functional correlogram curve c(OPD) and the phase derivative ϕ'(OPD) is constant.If this deviation between the synthetic correlogram sp< determined as described and the actually required correlogram with the same envelope, which is phase-shifted by 90° compared to the correlogram cp< in the carrier frequency, is not negligible, this can be remedied by determining an appropriately corrected synthetic correlogram instead of the previously determined synthetic correlogram sp<, which is then used instead of sp<.

[0087] To do this, form the equation c ′ OPD = env ′ OPD * cos φ OPD − env OPD * sin φ OPD * φ ′ OPD using s(OPD) = env(OPD) * sin (ϕ(OPD)) to s OPD = − c ′ OPD − env ′ OPD * cos φ OPD / φ ′ OPD and using c(OPD) = env(OPD) * cos (ϕ(OPD)) then s OPD = − c ′ OPD − env ′ OPD / env OPD * c OPD ) / φ ′ OPD .

[0088] This means that a corrected synthetic correlogram can be generated quite easily instead of sp< by determining the derivative vector c'(OPD) (or (cp< )' for the concrete point p) and then subtracting the original correlogram c(OPD) (or cp< ) from it, after multiplying it by the factor env'(OPD) / env(OPD), and dividing the difference by ϕ'(OPD).

[0089] The quantities env(OPD), env'(OPD), and ϕ'(OPD) are preferably determined from a model of the general functional correlogram curve, which is known from the properties of the measuring device or determined from measured data. The position of the model relative to the measured data is easily determined, e.g., from the position of the correlogram envelope determined using the original correlogram cp< and the uncorrected synthetic correlogram sp<. Experience has shown that particularly high accuracy is not required for either the model or its exact position relative to the specific correlograms—the procedure is extremely robust against deviations of any kind, so that the conceivable iterative procedure for determining the correct synthetic correlogram will rarely be needed in practice.

[0090] This provides an advantageous procedure for obtaining a correct synthetic correlogram that is phase-shifted by 90° in the carrier frequency compared to the correlogram cp<, if the approximations used in the simplest case to determine the synthetic correlogram sp< are ever too rough, using the original correlogram cp< and a relatively simple model, which synthetic correlogram is then preferably used instead of the insufficiently accurate synthetic correlogram sp< everywhere within the scope of the inventive procedure described above and below without further restrictions.

[0091] In the foregoing, only a very simple example has been used to illustrate how the inventive design of the evaluation unit can be used to achieve more robust results in determining the surface topography of measurement objects compared to the previously known state of the art.

[0092] The essential inventive property of the method according to the invention and of the evaluation unit of the device according to the invention, namely that it uses not only the light intensities lp< (zi ) on the detector element p but also the light intensities lq< (zi ) of at least one other detector element q of the multi-element detector to determine the value zp< assigned to the detector element p, can, however, be advantageously designed and used in a much more far-reaching manner.

[0093] First of all, according to what has been said so far, it is relatively clear that when determining the value zp< assigned to the detector element p, one does not have to limit oneself to the additional information of a single further detector element q, but it is advantageous to use at least two, advantageously at least three, preferably at least ten and most preferably at least twenty detector elements q of the multi-element detector.

[0094] Advantageously, the at least one other detector element q, whose light intensities lq< (zi ) are used to determine the value zp< assigned to the detector element p, is selected such that it has predetermined neighborhood properties with respect to the detector element p.

[0095] This is because it is advantageous not to simply use any detector elements q and the associated correlograms to determine the value zp< assigned to the detector element p, but rather to select detector elements q that exhibit certain advantageous neighborhood properties to the detector element p. Such an advantageous neighborhood property is, for example, in many cases, of course, a spatial proximity between the detector elements p and q or the associated object points, but the similarity of the respective associated correlograms can also define a neighborhood on the basis of which the points q can be advantageously selected.In this respect, it is advantageous if the at least one detector element q, which is used to determine the value zp< assigned to the detector element p, is selected such that it has predetermined neighborhood properties with respect to the detector element p with regard to at least one predetermined metric.

[0096] In particular, it is therefore advantageous that the neighborhood properties are defined at least partially with the aid of a metric, wherein this metric particularly advantageously comprises a spatial distance between the detector elements p and q and / or a spatial distance between points assigned to them on the measurement object and / or an optical surface conjugated to the multi-element detector.

[0097] Such a metric for determining the proximity of detector elements p and q can include determining a spatial distance between the detector elements p and q and / or a spatial distance between points assigned to them on the measurement object and / or an optical surface conjugated to the multi-element detector. This is advantageous because spatially adjacent detector elements are usually also assigned to object points with similar height values. Furthermore, disturbances affecting spatially adjacent detector elements are also very similar in many cases.In particular, in the case of disturbances caused by changes in the density and / or the refractive index of at least one of the media traversed by the light in the measuring and / or reference arm, for example by turbulence of air at different temperatures, the disturbances are usually locally different, so that they can best be dealt with by a correspondingly local evaluation using detector elements q that are spatially adjacent to the detector element p.

[0098] It is advantageous if the correlograms cp< = (lp< (z 1 ), lp< (z 2 ), ..., lp< (zn )) and cq< = (lq< (z 1 ), lq< (z 2 ), ..., lq< (zn )) are themselves as similar as possible, but of course not the same. Accordingly, it is advantageous if the metric for determining the neighborhood of detector elements p and q comprises a metric in the n-dimensional vector space of the vectors (lp< (z 1 ), lp< (z 2 ), ... , lp< (zn )) and (lq< (z 1 ), lq< (z 2 ), ... , lq< (zn )) or a metric in a vector space of the vectors resulting from them by projection and / or transformation. Such a metric can, for example, include the determination of the length of the difference vector cq< -cp< or of the corresponding vectors projected onto subspaces, whereby the latter case also includes using only those correlogram points that are close to the respective envelope maximum for the neighborhood determination.

[0099] Advantageously, the metric for determining the proximity of detector elements p and q also uses a distance metric based on height values ​​hp< and hq< determined by means of a first rough determination.

[0100] In principle, when selecting the detector elements q used to determine the value zp<, it is in most cases very advantageous to use detector elements q for which there is a high degree of similarity, for example between the correlograms cp< and cq<, which are therefore adjacent in the sense of a neighborhood metric. However, it can be disadvantageous if, for example, the correlograms cp< and cq< are too similar or even identical, since in this case little or no additional information can be obtained. Therefore, in some cases it can even be advantageous to exclude detector elements q that are considered too adjacent in terms of their neighborhood properties when selecting the detector elements q to be used.

[0101] In principle, there are a variety of ways to use the correlogram information of several additional detector elements q 1 , q 2 , ... , qm to determine the value zp< assigned to the detector element p. The most obvious possibility is simply to determine individual values ​​zp< sequentially as described using the correlograms cp< and c qj< of the detector elements p and qj and to average these values ​​appropriately, possibly weighted, to obtain the final value for zp<.

[0102] However, it is more advantageous not to average the values ​​for zp< obtained by means of the different detector elements qj, but instead to appropriately average the different synthetic correlograms sp< determined with the aid of the detector elements qj in order to then determine the value for zp< from the resulting averaged synthetic correlogram in combination with the correlogram cp< belonging to the point p, for example as described.

[0103] However, it is even more advantageous to imagine the correlograms cp< = (cp< (1), cp< (2), ..., cp< (n)) and c qj< = (c qj< (1), c qj< (2), ..., c qj< (n)) as points or vectors in the n-dimensional vector space and to consider the derivative sp< = (sp< (1), sp< (2), ..., sp< (n)) given up to the factor f as a direction vector which represents the main direction of the vectors cp< = (cp< (1), cp< (2), ..., cp< (n)) and c qj< = (c qj< (1), c qj< (2), ..., c qj< (n)) or (lp< (z 1 ), lp< (z 2 ), ... , lp< (zn )) and (lq< (z 1 ), lq< (z 2 ), ... , lq< (zn )) is at least locally approximated.

[0104] The previously described determination of the synthetic correlogram sp< from the correlogram cp< belonging to the detector element p and only a single further correlogram cq< belonging to the detector element q, when viewed in the n-dimensional vector space, already represents in this sense at least a local approximation of a direction vector to a point distribution of correlogram vectors (here, however, consisting of only two points).

[0105] It is therefore advantageous that the approximation used in determining the value zp< assigned to the detector element p comprises the determination of a direction vector which, at least locally in an environment of the light intensity vector (lp< (z 1 ), lp< (z 2 ), ..., lp< (zn )) or of a vector resulting from it by projection and / or transformation, approximates the main direction of the point distribution given by the vectors (lp< (z 1 ), lp< (z 2 ), ..., lp< (zn )) and (lq< (z 1 ), lq< (z 2 ), ..., lq< (zn )) or of the vectors resulting from them by projection and / or transformation.

[0106] In order to be able to carry out the evaluation as discussed in the n-dimensional vector space, the actually measured light intensity vectors (lp< (z 1 ), lp< (z 2 ), ..., lp< (zn )) and (lq< (z 1 ), lq< (z 2 ), ..., lq< (zn )) are preferably used, or they are advantageously adjusted for offset and / or amplitude scaling factors as already described several times. If in the previous or following or in the patent claims light intensity vectors of the type (lp< (z 1 ), lp< (z 2 ), ..., lp< (zn )) or correlograms or correlogram vectors of the type cp< = (cp< (1), cp< (2), ..., cp< (n)) - which are otherwise identical - were or are used, both are preferably meant - unless explicitly stated otherwise or obviously nonsensical. In addition, it is preferable to limit all observations and considerations to a subset of the correlogram points of the type lp< (z 1 ), lp< (z 2 ), ... , lp< (zn )) orcp< (1), cp< (2), ..., cp< (n), which is advantageous, for example, in terms of computational effort. This is also explicitly covered by the description.

[0107] All this is summarized by the reference used, inter alia, in the patent claims to the "light intensity vector (lp< (z 1 ), lp< (z 2 ), ..., lp< (zn )) or a vector resulting from it by projection and / or transformation", whereby the terms "projection and / or transformation" may additionally include further aspects.

[0108] There are again a multitude of possibilities for determining the synthetic correlogram sp< = (sp< (1), sp< (2), ..., sp< (n)), which is given as a direction vector up to the described factor f, which at least locally approximates the main direction of the point distribution given by the vectors cp< = (cp< (1), cp< (2), ..., cp< (n)) and c qj< = (c qj< (1), c qj< (2), ..., c qj< (n)) or (lp< (z 1 ), lp< (z 2 ), ..., lp< (zn )) and (l qj< (z 1 ), l qj< (z 2 ), ..., l qj< (zn )). The simplest of these is simply to carry out a fit, advantageously a least-squares fit or a regression of a straight line in n-dimensional space to the point distribution given by the given vectors using known mathematical methods and then to determine the direction vector and thus the synthetic correlogram sp< up to the factor f.The factor f can then be determined as already described, and from the two correlograms cp< and sp< belonging to the detector element p, the desired value zp< is then preferably determined as previously described.

[0109] Another, particularly advantageous way to determine the synthetic correlogram sp< as a direction vector that at least locally approximates the point distribution given by the correlogram vectors cp< and c qj< is to use the well-known mathematical method of principal component analysis (PCA). This involves determining the principal axes of the point distribution given by the vectors cp< and c qj<, with the direction of the eigenvector corresponding to the largest eigenvalue of the so-called covariance matrix being the desired direction vector. Once this direction vector is available, the process continues as described above.It should be noted that the principal component analysis does not necessarily have to be carried out completely, it only has to be carried out - and is therefore preferably only carried out - until the desired eigenvector is available, which is the case very early in many of the known numerical methods for PCA, since it belongs to the largest eigenvalue that is usually determined first.

[0110] In the method according to the invention, the value zp< assigned to the detector element p is determined with the aid of an approximation to at least part of the n-dimensional light intensity vectors (lp< (z 1 ), lp< (z 2 ), ... , lp< (zn )) and (lq< (z 1 ), lq< (z 2 ), ... , lq< (zn )) given via lp< (zi ) and lq< (zi ) or the vectors resulting from them by projection and / or transformation.

[0111] The evaluation unit of the device according to the invention is designed to determine the value zp< assigned to the detector element p with the aid of an approximation to at least part of the n-dimensional light intensity vectors (lp< (z 1 ), lp< (z 2 ), ... , lp< (zn )) and (lq< (z 1 ), lq< (z 2 ), ... , lq< (zn )) given via lp< (zi ) and lq< (zi ) or the vectors resulting from them by projection and / or transformation.

[0112] However, it is particularly advantageous not to restrict oneself to the at least local approximation of a direction vector to the point distribution given by the vectors cp< = (cp< (1), cp< (2), ..., cp< (n)) and c qj< = (c qj< (1), c qj< (2), ..., c qj< (n)) or (lp< (z 1 ), lp< (z 2 ), ... , lp< (zn )) and (l qj< (z 1 ), l qj< (z 2 ), ... , l qj< (zn )), but instead to at least locally carry out the approximation of a suitable low-dimensional submanifold of the n-dimensional vector space to this point distribution, for example a maximally three-dimensional submanifold, preferably a maximally two-dimensional submanifold and very particularly preferably a one-dimensional submanifold, ie a one-dimensional parameterized curve in the corresponding high-dimensional vector space.

[0113] That this is possible is particularly easy to see using the example of phase-shifting interferometry, if one additionally assumes, for once, that the correlograms to be evaluated cp< = (cp< (1), cp< (2), ..., cp< (n)) and c qj< = (c qj< (1), c qj< (2), ..., c qj< (n)) are offset-corrected and normalized as described above.

[0114] Then, for example, the correlogram cp< = (cp< (1), cp< (2), ..., cp< (n)) is given by cp< (i) = cos( 2π / λ *(zi -2*hp< ) ) which, according to the addition theorems, is cp< (i) = cos(2π / λ *zi )*cos(2π / λ*2* hp< ) + sin(2π / λ *zi )*sin(2π / λ*2* hp< ). If we now represent this in vector notation in the form

[0115] cp< = (cp< (1), cp< (2), ..., cp< (n)) = cos(2π / λ *2*hp< ) * v + sin(2π / λ*2* hp< ) * w with the vectors v = cos 2 π / λ * z 1 , cos 2 π / λ * z 2 , … , cos 2 π / λ * z n and w = (sin(2π / λ *z 1 ), sin(2π / λ*z 2 ), ..., sin(2π / λ *zn )), it is easy to see that all possible cp< - and c qj< -vectors always lie on an ellipse, i.e., on a one-dimensional parameterized curve in n-dimensional space, which itself lies entirely on a two-dimensional submanifold, in this case, on a plane. This applies regardless of the values ​​of zi , although the exact position and shape of this ellipse certainly depend on the values ​​of zi .

[0116] This provides various evaluation options: The simplest option is to actually fit an ellipse in n-dimensional space to the points defined by the measured correlograms cp< and c qj<, project the correlogram vector cp< onto this ellipse, and obtain the synthetic correlogram vector sp< by differentiating the one-dimensional differentiable submanifold of the n-dimensional space given by the ellipse as a tangent vector to this submanifold at the location of the projected correlogram vector cp<. Differentiation is very easy to perform in this case, since the ellipse, as a one-dimensional submanifold of the n-dimensional space, can also be represented as a one-dimensional parameterized curve in this vector space.After determining the tangential vector and its scaling factor as described, further evaluation is preferably carried out using sp< as previously stated.

[0117] It is therefore advantageous to determine a tangent vector to the differentiable manifold M by differentiating it.

[0118] If one omits, as previously assumed, the correlograms cp< = (cp< (1), cp< (2), ..., cp< (n)) and c qj< = (c qj< (1), c qj< (2), ..., c qj< (n)) to be evaluated in order to correct their offset and / or their scaling, which is less advantageous but still possible, one obtains a two- or even three-dimensional differentiable submanifold of the n-dimensional space instead of the ellipse. Here, too, the correlogram vector cp< is preferably projected onto this submanifold and the synthetic correlogram vector sp< is obtained by differentiation at the position of the projected correlogram vector cp<, so that the procedure then continues as described.

[0119] It is therefore particularly advantageous that the approximation comprises the adaptation of a differentiable submanifold M, in particular a low-dimensional differentiable submanifold, at least to the point distribution from the points (lp< (z 1 ), lp< (z 2 ), ... , lp< (zn )) and (l qj< (z 1 ), l qj< (z 2 ), ... , l qj< (zn )) in the n-dimensional vector space or in a vector space of the vectors resulting from them by projection and / or transformation, in particular that the differentiable submanifold is a maximally three-dimensional, in particular maximally two-dimensional submanifold of the associated vector space.

[0120] The described procedures can also be carried out for correlograms cp< and c qj< obtained using short-coherence interferometry. If offset-corrected and normalized correlograms are used for evaluation, as previously with phase-shifting interferometry, an ellipse is no longer obtained. Instead, due to the changing envelope, another one-dimensional parameterized curve or one-dimensional differentiable submanifold is obtained, which resembles an elliptical spiral but, unlike an ellipse, no longer closes. Nevertheless, the procedure can be continued here as with phase-shifting interferometry, although determining the curve shape to be fitted is more complex and may require a model of the envelope and the relative position of the modulation with respect to the envelope; this information can also be determined from measured data.If, on the other hand, one omits to correct the offset and / or scaling of the correlograms cp< and c qj< to be evaluated, the procedure becomes even more complex, but is still feasible in principle.

[0121] In an advantageous embodiment, the approximation to the n-dimensional light intensity vectors (lp< (z 1 ), lp< (z 2 ), ..., lp< (zn )) and (lq< (z 1 ), lq< (z 2 ), ..., lq< (zn )) or the vectors resulting from them by projection and / or transformation is therefore described by a one-dimensionally parameterized curve in the associated vector space.

[0122] In particular, it is advantageous that the approximation to the light intensity vectors (lp< (z 1 ), lp< (z 2 ), ..., lp< (zn )) and (lq< (z 1 ), lq< (z 2 ), ..., lq< (zn )) or the vectors resulting from them by projection and / or transformation is at least locally a linear approximation. This is therefore an approximation to a straight line, plane or another linear subspace of the vector space that has a lower dimension than the vector space itself.

[0123] However, there is a significantly more advantageous procedure for evaluating the points in n-dimensional space defined by the measured, advantageously offset-corrected and normalized correlograms cp< and c qj<, which is applicable to both short-coherent interferometry and phase-shifting interferometry. This procedure will be described below only for short-coherent interferometry, since, as can be easily seen, the application in the special case of phase-shifting interferometry is just as feasible. For this evaluation, it is again assumed that the envelope of the correlograms changes significantly more slowly than their modulation. Under this assumption, the spiral closes to a first approximation also in the short-coherent case, and the one-dimensional curve in n-dimensional space defined by the correlograms lies to a first approximation in a two-dimensional subspace, i.e.a plane, where it is approximately represented by an ellipse (for PSI all of this applies exactly anyway). This two-dimensional plane is then determined and parameterized in n-dimensional space by a plane fit or a two-dimensional regression, by a principal component analysis, or by parts thereof, etc. The points defined by the correlograms cp< and c qj< are then projected onto this plane, where they are still described by an ellipse. This ellipse is then determined within this projection plane by a two-dimensional fit or comparable method, so that an associated affine mapping can then be defined which maps the ellipse onto an associated circle. If the projection plane was determined with the help of a principal component analysis, in an alternative embodiment the corresponding affine mapping is advantageously carried out directly using the two largest eigenvalues ​​orthe corresponding eigenvectors of the PCA covariance matrix are determined. Using the resulting affine mapping, all points belonging to the projected correlogram vectors cp< and c qj< can also be transformed. After this transformation has been performed, the spatial angle between two projected and affinely mapped points belonging to different correlograms cp< and c qj< corresponds to the phase difference between the sinusoidal modulations of the corresponding correlograms and can thus be used to evaluate the phase of the correlograms.

[0124] It is therefore advantageous that the determination of the direction vector comprises a principal component analysis (PCA) or parts thereof, an eigenvalue determination, a linear regression or another optimization method for minimizing the deviation of a straight line defined by the direction vector and / or a base point suitable for it from the vectors (lp< (z 1 ), lp< (z 2 ), ..., lp< (zn )) and (lq< (z 1 ), lq< (z 2 ), ..., lq< (zn )) or the point distribution given from them by projection and / or transformation of the vectors.

[0125] For envelope evaluation, the distance of the points belonging to the correlogram vectors cp< and c qj< to the projection plane is preferably used. This distance assumes more or less discrete, but clearly distinguishable values ​​for all correlograms, depending on whether a correlogram has the identical phase position or a phase shifted by k * 360° (k integer). This yields results comparable to the evaluation described so far, both for the envelope evaluation and for the phase evaluation, which are also correspondingly robust against disturbances. The advantage of the latter approach, however, is that the desired values ​​zp< are obtained for a large number of correlograms cp< simultaneously, which significantly reduces the computational effort.

[0126] It is advantageous that the amount zp< determined for the element p of the multi-element detector (6) for the change in the optical path length difference or phase difference, at which the optical path length difference or the phase difference between the measuring light and the reference light reaches an explicitly and / or implicitly predetermined value, is used to obtain information about the spatial position of a surface point of the measuring object (1) assigned to the detector element p (6b), in particular a height hp< relative to a reference surface, in order to determine the surface topography of the measuring object (1).

[0127] Further advantageous features and embodiments are explained below using exemplary embodiments and the figures. Herein: Figure 1 shows a first embodiment of a device according to the invention with a Michelson interferometer; Figures 2 to 11 show representations for explaining embodiments of methods according to the invention for implementation by means of the device according to Figure 1 ; Figure 12 shows a second embodiment of a device according to the invention with a Michelson interferometer; Figures 13 to 14 show representations for explaining embodiments of methods according to the invention for implementation by means of the device according to Figure 12 ; Figure 15 shows a third embodiment of a device according to the invention with a Mirau lens; Figure 16 shows a representation to explain an embodiment of a method according to the invention for implementation by means of the device according to Figure 15; Figure 17 shows a fourth embodiment of a device according to the invention with a Michelson interferometer; Figures 18 to 22 show representations for explaining embodiments of methods according to the invention for implementation by means of the device according to Figure 17 .

[0128] The figures show schematic representations, not to scale. In the figures, identical reference symbols denote identical or equivalent elements.

[0129] Figure 1 shows a first embodiment of a device according to the invention with a Michelson interferometer. The device serves for the interferometric determination of the surface topography of a measuring surface 1a of a measurement object 1.

[0130] The device has a light source 2, which in this case is designed as an LED, with a broadband light spectrum, in this case in the wavelength range 490 nm to 550 nm with a maximum at 520 nm.

[0131] The light from the light source passes through a condenser 3 onto a semi-transparent mirror 4, by means of which approximately 50% of the light intensity is directed as illumination light onto the measurement surface 1a of the measurement object 1. The illumination light reflected as measurement light from the measurement surface 1a partially passes through the semi-transparent mirror 4 and is imaged by an imaging optics 5, which in this case is designed as a telecentric imaging optics, into a detection region of the device in which a multi-element detector 6 is arranged. The imaging optics 5 in this case has optical lenses 5a and 5b and an optical aperture 5c for the telecentric configuration.

[0132] The multi-element detector 6 is embodied as a CCD camera and has a CCD sensor configured as an area sensor 6a with 1024 x 1024 sensor elements arranged as a sensor array at the intersection points of a square grid. Each sensor element represents a detector element of the multi-element detector 6.

[0133] Approximately 50% of the light generated by light source 2 passes through semi-transparent mirror 4 as reference light and reaches a reference mirror 8 via an optical filter 7. The reference light reflected by reference mirror 8 in turn passes through optical filter 7 and is partially deflected by semi-transparent mirror 4 to also be imaged into the detection area by imaging optics 5. An interference pattern is thus generated in the detection area by the superposition of measurement light and reference light, which can be evaluated by multi-element detector 6. Filter 7 serves to attenuate the light intensity of the reference light in order to achieve approximately equal light intensities of measurement light and reference light in the detection area.

[0134] Using an adjustment unit 9, the distance of the device from the measurement object 1 and thus the optical path length difference OPD between the measurement light and the reference light can be changed. In the present embodiment, the optical path length of the measurement light is thus changed for this purpose.

[0135] By means of the multi-element detector 6, the light intensities lp< (zi ) of the interference pattern on the detector elements of the multi-element detector 6 in the detection area can thus be detected for at least two optical path length differences OPD changed by different amounts zi.

[0136] The device thus has an interferometer optics for combining the measuring light with the reference light and generating the interference pattern in the detection area, wherein the interferometer optics in this case has the elements condenser 3, semi-transparent mirror 4, imaging optics 5, optical filter 7 and reference mirror 8.

[0137] The device further comprises an evaluation unit 10, which is designed as a commercially available computer with a processor, data storage, screen display unit, and keyboard as an input unit, in this case in the form of a laptop. The evaluation unit 10 is connected to the multi-element detector 6 and the adjustment unit 9, so that measurement signals from the detector elements of the multi-element detector can be read out by the evaluation unit, and the optical path length difference OPD can be changed by means of control signals via the adjustment unit 9.

[0138] The evaluation unit is designed to determine, for a plurality of detector elements p of the multi-element detector 6, the amount zp< for the change in the optical path length difference at which the optical path length difference between the measuring light and the reference light reaches an explicitly and / or implicitly predetermined value, in this case 0, in order to determine therefrom the surface topography of the measuring surface 1a of the measuring object 1.

[0139] It is essential that the evaluation unit is designed to use not only the light intensities lp< (zi ) but also the light intensities lq< (zi ) of at least one other detector element q of the multi-element detector to determine the value zp< assigned to the detector element p, as will be explained in more detail below using an embodiment of the method according to the invention: The first embodiment of a method according to the invention is designed as scanning white light interferometry (WLI).

[0140] First, the device is moved into a reference state by means of the adjustment unit 9. The virtual reference surface 11, which in the reference state corresponds to an optical path length difference OPD = 0, is in Figure 1 marked by a dash-dotted line and is perpendicular to the plane of the drawing in Figure 1 The exact location of this reference surface 11 is not important for the measurement. Thus, the reference surface 11 can penetrate the object, as in the present example. However, it is also possible for the reference surface to be located above the object, for example. It is essential that height information hp< is determined for a large number of measurement points p, so that the relative height difference between the measurement points can be calculated, thus obtaining a surface topography of the measurement surface 1a.

[0141] Starting from the previously described reference state, the adjustment unit 9 is now controlled by means of the evaluation unit 10 in such a way that the device is moved away from the measuring object 1 in equidistant predetermined steps, that is to say in the illustration according to Figure 1 is moved upwards. This increases the optical path length of the measuring light, but the optical path length of the reference light remains unchanged, resulting in a change in the OPD. In this case, measured values ​​of all detector elements p of the multi-element detector 6 are recorded at 64 positions of the device. Figure 1 In the device shown, each of the 1024 x 1024 (= 1048576) detector elements p of the multi-element detector 6 is assigned a measuring point p on the measuring surface 1a of the measuring object.

[0142] By means of the evaluation unit, the corresponding intensities lp< (z 1 ), lp< (z 2 ), ..., lp< (zn ) are recorded in interferograms for a total of 64 OPOs changed by z 1 , z 2 , ..., zn (n = 64) compared to the reference state, so that for each measuring point p the corresponding correlogram cp< = (cp< (1), cp< (2), ..., cp< (n)) is obtained, where the measured light intensity at the i-th point of the correlogram is cp< (i) = lp< (zi ).

[0143] The zi values ​​are specified equidistantly, which means - at least according to the specification - the specified distance Δz between two consecutive zi values ​​is constant.

[0144] After performing this measurement, a correlogram with 64 measured values ​​is available for each detector element p.

[0145] In Figure 2aAn example of such a correlogram of a detector element p is shown. The x-axis represents the specified z-values ​​in [µm], and the y-axis represents the light intensity in arbitrarily chosen units, preferably normalized to the light intensity outside the interference region of the correlogram, i.e., normalized to the offset of the correlogram. This also applies to the other correlograms shown in the figures, unless otherwise stated.

[0146] In principle, as explained at the beginning when describing the state of the art, the correlogram according to Figure 2a An envelope must be determined, and the value for zp< must be determined that corresponds to the maximum of the envelope (in this case, approximately at a z-position of 5 µm). For this measurement point, hp< would thus be determined at half the determined z-value, i.e., 2.5 µm.

[0147] However, if disturbances occur during the measurement, for example, in the form of vibrations, the measurements will not be taken exactly at the specified z-values. Rather, the z-values ​​of the actual measurements will be shifted to the right or left in the correlogram display, depending on the effect of the vibrations.

[0148] In Figure 2b A correlogram with noise is shown as an example. The measurement conditions are essentially identical to those shown in Figure 2a, but noise has been introduced, which, as previously described, leads to shifts in the zi values, so that they are no longer exactly equidistant.

[0149] In the following, the effect of such disturbances on the evaluation according to the state of the art will be described: In Figure 3 A state-of-the-art evaluation in the undisturbed case, i.e. with equidistant zi values, is shown: Figure 3ashows the unperturbed correlogram from Figure 2a .

[0150] Using Hilbert transformation, the Figure 3b As described above, the envelope env p< (z) can be calculated from the original correlogram cp< (z) and the synthetic correlogram sp< (z) obtained via the Hilbert transformation according to Figure 3c be obtained.

[0151] In the undisturbed case, a determination of the z-value of the maximum of this envelope already allows a good approximation to determine the desired value for zp< and thus twice the value of hp< at this measuring point p (about 5 µm).

[0152] It should be noted that during the evaluation an offset adjustment of the original correlogram cp< (z) was carried out, so that the values ​​of cp< (z), as in the Figure 3c shown, are now distributed around the zero value.

[0153] The correlogram according to Figure 4a corresponds to the correlogram according to Figure 2b with the non-equidistant zi values ​​due to the disturbance of the measurement caused by vibrations.

[0154] If this correlogram is now calculated according to Figure 4a the Hilbert transformation is carried out, the correlogram sp< (z) results according to Figure 4b . The image shown in Figure 4c together with the original ( Figure 4a The envelope env p< (z), plotted against the offset-adjusted correlogram cp< (z) and determined in the same way as before, exhibits an inconsistent profile due to the disturbances. In particular, no clear maximum can be determined. Thus, with the state-of-the-art evaluation method, the desired value of 5 µm for zp< could not be determined, or only very inaccurately, with such a disturbance.

[0155] The special feature of the method according to the invention, which enables an evaluation with great accuracy despite the presence of such disturbances, is explained below with reference to the first embodiment and further embodiments of the method according to the invention with reference to the Figures 5 to 11 explains: Figure 5 shows a schematic view of a section of the area sensor 6a of the multi-element detector 6. As already described, the CCD chip has 1024 x 1024 detector elements, with one detector element p being marked with reference numeral 6b in black. It is important that, in addition to the light intensities lp< (zi ), the light intensities lq< (zi ) of at least one other detector element q of the multi-element detector are also used to determine the value zp< assigned to the detector element p.

[0156] For this purpose, in the present embodiment, the neighboring elements are considered in a square of 3 × 3 pixels: The detector element 6b has eight neighboring elements qj, which are Figure 5 are highlighted by a thick black border. From these eight neighboring elements, one neighboring element is now selected as follows: For each neighboring element, the difference between the correlogram of the neighboring element and the correlogram of the detector element p is calculated. For the neighboring elements q 1 , q 2 , and q 3 , for example, the correlogram of this neighboring point is shown in a diagram above, and the difference between the correlogram of this neighboring point and the correlogram of the detector element p is shown in a diagram below (6b).

[0157] Using a vector norm, a measure is assigned to each difference correlogram, which thus represents a difference vector c qj< - cp<. In this case, the measure is determined based on the standard deviation formed from the elements c qj< (i) - cp< (i) of the difference vector. If the correlograms of a neighboring element qj and the detector element p deviate only slightly according to this measure, the correlograms of these two detector elements are very similar, which is disadvantageous for the evaluation according to the present embodiment, since signal noise may make exact difference calculation difficult. In principle, a large deviation according to the previously described measure can also be disadvantageous, since a linear approximation as described below is then no longer permissible.In the present measurement situation, however, a deviation that is too large is unlikely, since only nearest neighbors are considered, for which the correlograms usually differ only slightly.

[0158] In the present embodiment of the method according to the invention, the eight neighboring elements qj are selected, the one for which the difference vector has the largest value according to the previously described measure. In the example according to Figure 5 this is the detector element q 3 .

[0159] To determine the value zp< assigned to the multi-detector element p, the light intensities of the detector element q 3 are also used, as explained in more detail below: In Figure 6a are the correlograms of the detector elements p and q = q 3 with introduced disturbances of the measurements as well as in Figure 2bshown. Since the adjacent detector elements p and q are assigned to spatially adjacent measuring points p and q on the measuring surface 1a of the measuring object 1, the correlograms differ only slightly.

[0160] In Figure 6b Therefore, a section magnification of the z-value 5 µm is shown. Figure 6c shows the difference correlogram cq< (z) - cp< (z). This difference correlogram can also be interpreted as a 64-dimensional difference vector. In the present embodiment, further evaluation is performed in 64 dimensions, i.e., the measurement points of all zi values ​​are used.

[0161] It is also within the scope of the invention to use a subset, i.e. a difference vector of lower dimension, such as a selection of points which are arranged approximately symmetrically around a maximum of the correlogram cp<, for example ± 16 points, i.e. a total of 33 points. In an alternative embodiment, points are selected symmetrically around the point at which the modulation is strongest. This is recognizable by the fact that the absolute value of the difference between two consecutive points is greatest there. In a further modification of the exemplary embodiment, a sine fit is performed on the correlograms, and the value of the maximum amplitude of this fit is determined as the central value for a symmetric selection of a subset of points for the subsequent evaluation.

[0162] In Figure 7a is the correlogram of the detector element p and the difference correlogram according to Figure 6cAs explained before, the difference correlogram Q p< can be plotted according to Figure 6c can be regarded as a synthetic correlogram sp< = f * Q p< with a scaling factor of f=1.

[0163] However, to determine the correlogram envelope, a synthetic correlogram sp< is necessary, which has been scaled with the scaling factor f so that it has the same amplitude as the original correlogram cp<.

[0164] As in Figure 7b and Figure 7d As shown, the determination of the correlogram envelope env p< (zi ) via the square root of (cp< (zi )) 2< + (sp< (zi )) 2< without a correct scaling of the sp< (zi ) would lead to a wavy result for the envelope env p< (zi ), which would not allow the determination of a maximum or at least would lead to a considerable risk of error. Figure 7b shows the case where according to Figure 7a was scaled with a factor f that was too small.

[0165] However, if according to Figure 7c scaled with a factor f that is too large, the resulting envelope is also wavy and cannot be evaluated or can only be evaluated with a considerable risk of error, as in Figure 7d shown.

[0166] In the present embodiment, the correct scaling factor f is determined by first determining a rough value im for the index i at which the envelope maximum lies by determining the maximum of the correlogram cp<. Then, to determine the scaling factor f, 21 points located symmetrically around the index im of the offset-corrected original correlogram cp< and the synthetic correlogram sp< = f * Q p< scaled with the factor f are used. For these points, the standard deviation σ(f) of the 21 values ​​(cp< (zi )) 2< + (sp< (zi ,f)) 2< with i= im -10 , i= im -9, ..., i= im +10 can be specified for each possible value of f. The gradient method is then used to determine the value for f at which the said standard deviation is minimal. Using this value for f, the correctly scaled synthetic correlogram sp< = f * Q p< is determined for further evaluation.

[0167] Figure 8a now shows the correlogram cp< (zi ) from Figure 6a or Figure 2b together with the correctly scaled synthetic correlogram sp< (zi ) = f * Q p< (zi ) determined as described above. It can be seen that the amplitude of the synthetic correlogram sp< (zi ) scaled by this factor f corresponds to that of the original correlogram cp< (zi ).

[0168] In Figure 8b The envelope env p< (z) is shown, which results from the correlogram cp< and the correctly scaled synthetic correlogram sp< determined as described above. The envelope env p< (zi ) was determined as described by taking the square root of (cp< (zi )) 2< + (sp< (zi )) 2< .

[0169] Figure 9In summary, panel a shows the correlogram of detector element p, which is affected by disturbances in the z-values. Panel b shows the synthetic correlogram sp< , which was scaled with the determined correct factor f as described above. This synthetic correlogram was thus determined using the correlogram of detector element q 3 as well.

[0170] Part c shows the correlogram cp< according to part a together with the envelope env p< obtained as described above. A comparison of this envelope determined with the method according to the invention with the envelope from Figure 4c , which was determined using one of the state-of-the-art methods, clearly shows that considerable robustness was achieved against disturbances leading to non-equidistant zi values, such as vibrations.

[0171] Now the maximum of the envelope env p< is determined according to Figure 9c to determine the zp< value for the detector element p and, accordingly, the measurement point p. In this specific case, this is done via a parabolic fit to the envelope env p< (zi ) at the previously selected 21 points zi .

[0172] However, the measurement data allow a further increase in the accuracy when determining the zp< value, as shown below using Figure 10 explained in an advantageous development of the described embodiment.

[0173] Figure 10a shows the correlogram cp< and the envelope obtained by means of the synthetic correlogram according to the representation in Figure 9c .

[0174] In Figure 10bin the abscissa direction z, an enlargement of a section around the previously determined maximum of the envelope env p< (z) is shown. On the ordinate axis, however, it is not the intensity that is plotted against the corresponding z-value, but a phase ϕ p< (z) assigned to the z-value. As described above, for each correlogram point cp< (i) an associated point sp< (i) was determined using the synthetic correlogram, whereby the synthetic correlogram sp< can be viewed as a correlogram shifted by 90° in the carrier frequency to the correlogram cp<. From the two correlograms, a phase can therefore be determined for each position zi, for example by forming the extended arctangent arctan2(sp< (zi ),cp< (zi )) for each point zi at the corresponding indices i and performing phase unwrapping.

[0175] As described above, in this embodiment, the reference state was selected by adjusting the optical path lengths (OPD = 0). The specified value for the optical path length difference in this embodiment is therefore 0. Accordingly, the specified phase value for determining the magnitude zp< is also 0°. As shown in Figure 10b As shown, a straight line is fitted using linear regression and the z-value for zp< is determined at which the phase value corresponds to the specified value (in this case 0°).

[0176] This two-step procedure, in which an approximate value zp< was first obtained by determining the maximum of the envelope and then the zp< value with phase 0° was determined using a regression line, enables a significant improvement in measurement accuracy and a further increased robustness against disturbances such as vibrations.

[0177] In Figure 10cFinally, it is shown how the method according to the invention can also fundamentally provide information about the amount Δz by which a z-value zi has been changed due to the disturbance: In the ideal, disturbance-free case, the values ​​zi in the representation according to Figure 10b or Figure 10c on the dotted regression line. For each value zi, a value Δz i can be determined by calculating the difference between the value zi and the z-value that corresponds to the assigned phase value on the regression line. These correction values ​​Δz are shown in Figure 10c for the zi values ​​with i = 30, 31, ..., 37 are indicated by horizontal arrows.

[0178] It is therefore also within the scope of the invention to correct the zi values ​​by the Figure 10cobtained correction values ​​Δz i and to carry out an evaluation, in particular in a manner known per se, using the corrected zi values.

[0179] Furthermore, a more precise correction of the zi values ​​can be achieved by averaging the correction value for all or a selection of detector elements for each zi. This is because, for a fixed value of i, which according to the present embodiment corresponds to a measurement time, the disturbance Δz i is, to a good approximation, the same for all measurement points and thus also for all detector elements. Advantageously, a correction value Δz i is therefore determined for a subset, preferably for all detector elements, and the correction value Δz i is calculated by averaging the correction values ​​over the subset of measurement points, preferably over all measurement points, in particular by weighted averaging, in order to correct all zi values.

[0180] In Figure 11It is schematically shown that in the present embodiment, the unscaled synthetic correlogram Q p< = sp< / f can be regarded as a difference vector of the correlograms cq< - cp< and thus represents a direction vector in the n-dimensional vector space, which represents the light intensity vectors c p 1 , c p 2 , … , c p n = l p z 1 , l p z 2 , … , l p z n and (cq< (1), cq< (2), ..., cq< (n)) = (lq< (z 1 ), lq< (z 2 ), ..., lq< (zn )) approximates the given point distribution, which in this embodiment, however, only consists of two points.

[0181] For practical reasons, only three dimensions were selected for representation in the figure instead of the 64 dimensions actually available: In the representation according to Figure 11 Of the 64-dimensional vector space, only dimensions 31, 32 and 33 are shown.

[0182] In particular, a restriction to a subset of zi -values, in this case 21 of the possible 64 zi -values, corresponds to Figure 6 As described above, a projection in the vector space onto a subspace whose dimension corresponds to the number of selected zi values. Thus, even vectors resulting from projection and / or transformation can be evaluated using the method according to the invention.

[0183] An example of such a transformation is the first step of subtracting the offset from each correlogram, for example, by determining the offset based on the correlogram points outside the interference region or by averaging the correlogram values ​​and then subtracting the offset. Scaling or other changes by normalizing the correlogram also represent such a transformation. Rotations and other transformations can also be performed.

[0184] The measuring device according to the first embodiment and the method according to the first embodiment can also be used for measuring objects which partially reflect the measuring light at several levels: In the case of such a measuring object with several partially reflecting surfaces for measuring by means of the device according to Figure 1 For example, as an alternative to the previous description of the measurement object 1, the measurement surface 1a of the measurement object 1 on the side facing the device can be a coated glass surface, which only partially reflects the measurement light. Furthermore, the measurement object has a further interface in the volume of the measurement object on a metallically coated rear side (the side facing away from the device, in Figure 1bottom side) at which that portion of the measurement light which penetrates into the measurement object is reflected within the measurement object. By means of the methods known from the prior art and by means of the method according to the invention, two or more values ​​zp< can be assigned to a location point p, since a (partial) reflection of measurement light occurs at two or more height positions hp<. If the difference between the different heights hp< is greater than the coherence length of the detected measurement and reference light, the interferences belonging to the different reflective layers do not overlap in the correlograms, so that the partial correlograms belonging to the different layers can be separated within the measurement data and thus evaluated separately, and the different zp< values ​​can be determined separately from one another.However, the method according to the invention can also be used particularly advantageously when the various partial correlograms of the points zp< overlap, since in this case, too, the synthetic correlograms can be correctly determined using the method according to the invention. This makes it possible, on the one hand, to use the known methods for evaluating overlapping correlograms; on the other hand, additional advantageous evaluation options are available, since, for example, the common envelope and the common phase of the various partial correlograms can be very easily determined using the described procedure.

[0185] In Figure 12 A second embodiment of a device according to the invention is shown schematically.

[0186] This setup is also based on the principle of the Michelson interferometer. To avoid repetition, only the essential differences of the Figure 12 shown second embodiment compared to the one in Figure 1 The first exemplary embodiment shown will be discussed below: The measurement object 1 is to be measured at a predetermined elevated temperature and is therefore arranged on a heating plate 1b. This is relevant, for example, if the curvature of the measurement surface 1a of electronic components is to be examined due to the effects of heat.

[0187] The problem here is that, due to the heating by the heating plate 1b, air turbulence 1c occurs between the measuring surface 1a and the device in the beam path of the measuring light. Such air turbulence leads to density changes and, accordingly, to an inhomogeneous refractive index. Fluctuations in the refractive index of the air due to the density fluctuations caused by heating can lead to significant measurement errors. This is the case even in the absence of additional disturbances, such as vibrations.

[0188] The arrangement of the optical components of the device is identical to that of the device according to the Figure 1 shown first embodiment.

[0189] In the present case, the evaluation unit 10 is connected to the multi-element detector 6 as described above. The multi-element detector 6, embodied here as a CMOS camera, is connected to the adjustment unit 9 via a trigger line. This is intended to schematically illustrate that, in the second embodiment, equidistant zi values ​​are specified for the adjustment unit, and the adjustment unit 9 sends a trigger signal to the multi-element detector 6 upon reaching a specified zi value, so that a camera image is recorded.

[0190] The reference surface 11 is shown above the measurement object 1, which represents the reference state OPD = 0. In this exemplary embodiment, the reference surface 11 is thus located above the measurement object, so that negative height values ​​hp< result for measurement points p on the measurement surface 1a of the measurement object 1.

[0191] The evaluation unit 10 is designed to perform an evaluation according to a second embodiment of the method according to the invention. This essentially corresponds to the first embodiment, so that, to avoid repetition, only the essential differences will be discussed below: Figure 13 The area sensor 6a of the multi-element detector 6 is again shown schematically. A comparison of the Figures 13 and 5 shows that in this second embodiment of the method according to the invention a larger number of detector elements is used.

[0192] Also in the representation according to Figure 13A detector element p is marked with reference numeral 6a in black. The neighboring detector elements under consideration are marked by a thick black border: In this case, the 48 neighboring elements qj of the detector element p (6a), located in a square of 7 x 7 detector elements, are used for the evaluation. In this case, too, desired neighborhood properties are defined in order to evaluate the suitability of a neighboring element for the evaluation. As a measure for evaluating the neighborhood properties, the length norm of the difference vector cq< - cp< for the detector element q is formed as a metric. For each of the 48 neighboring elements, there is thus a value that is Figure 13for example, for five neighboring elements q, this is represented by a corresponding difference correlogram cq< - cp< and by the numerical value resulting from the metric length norm for this difference correlogram. In this case, the selection is made in such a way that those ten neighboring elements from the 48 neighboring elements are chosen which have the most similar correlograms. Thus, values ​​are calculated for all 48 neighboring elements according to the metric length norm of the difference vector, and the ten neighboring elements with the ten lowest values ​​are determined. This makes sense in the present measurement situation, since it can be assumed that any difference in the correlograms will be primarily caused by the air turbulence, and thus neighboring elements with only minor differences in terms of the metric length norm of the difference vector will have similar measurement conditions.

[0193] In the exemplary representation according to Figure 13The detector elements with a difference vector length norm of 1.2153 and 0.8715 are not included in the group of detector elements with the ten lowest values. However, the detector elements with the values ​​of 0.4293, 0.3916, and 0.3444 are included in this group and are therefore included in the evaluation (along with seven other detector elements not shown in detail).

[0194] The second embodiment thus differs in particular from the first embodiment in that several adjacent detector elements qj (ten in the present case) are used to determine the value zp< assigned to the detector element p by using not only the light intensities lp< (zi ) but also the light intensities l qj< (zi ) of the ten adjacent elements q 1 , ..., q 10 determined as described above.

[0195] In alternative embodiments of the second exemplary embodiment, not a fixed number of detector elements is selected, for example in the spatial vicinity of the detector element p, but rather all those detector elements for which certain predetermined neighborhood properties are present are selected from a specific preselection of detector elements. For example, it may be advantageous to select all those detector elements qj for which the correlogram c qj< deviates by a suitable amount from the correlogram cp< at the detector element p, e.g., so that the standard deviation of the values ​​c qj< (i) - cp< (i) determined from the difference vectors corresponds approximately to twice the quantization and camera noise. Here, an upper and lower limit is advantageously defined for the values ​​determined with the corresponding metric, based on which the detector elements to be used are selected.This takes into account that the correlograms c qj< and cp< (i) should not be too different on the one hand, since otherwise a linear approximation would be difficult, but on the other hand they should not be too similar, since otherwise the difference correlogram may be very small and disturbed or even dominated by possible noise.

[0196] In Figure 14 It is shown schematically how the value zp< assigned to the detector element p is determined based on the several neighboring elements qj: For reasons of representation, analogous to Figure 11 also at Figure 14For the representation, a projection onto a three-dimensional subspace is carried out by representing only the dimensions 31, 32 and 33 of the 64-dimensional vector space. As previously described, in this exemplary embodiment, ten neighboring elements were selected, so that in the 64-dimensional vector space, in addition to the correlogram vector cp< belonging to the detector element p, there are ten correlogram vectors cq< for the neighboring elements q, which, for the sake of simplicity, are represented only by their endpoints marked as points.

[0197] In contrast to the representation according to Figure 11and the formation of a difference vector described for this purpose, the determination of the direction vector in the 64-dimensional space is not carried out directly by determining a difference vector between two correlograms cq< and cp< . Instead, the direction vector 15 is determined, which belongs to the straight line that best approximates the point cloud given by the end points of the correlogram vectors c qj< and cp< (also denoted by qj and p for the sake of simplicity), and which also passes through the end point p of the correlogram vector cp<, also denoted by p. The determination of this direction vector can be carried out with the aid of a straight line fit and subsequent determination of the direction vector from this, in this case in the 64-dimensional vector space.In an alternative embodiment, a principal component analysis (PCA) is performed on the point cloud in the 64-dimensional vector space defined by the vector endpoints qj and p, where the eigenvector corresponding to the largest eigenvalue of the covariance matrix is ​​also the desired direction vector. Since many PCA algorithms first determine the largest eigenvalue of the covariance matrix and the corresponding eigenvector, it is particularly advantageous to abort the principal component analysis once the corresponding eigenvector is available. This allows, in this specific embodiment, the full principal component analysis to be dispensed with and only a partial principal component analysis needs to be performed.

[0198] The found direction vector corresponds again to the unscaled synthetic correlogram Q p< (zi ) = sp< / f , as described in the first embodiment, and is therefore also in Figure 14referred to as sp< / f.

[0199] According to this second embodiment, an unscaled synthetic correlogram Q p< = sp< / f is determined using a plurality of detector elements q. Further evaluation is carried out as described in the first embodiment: The scaling factor f for scaling the synthetic correlogram is again determined using one of the described procedures. Subsequently, the envelope is determined from the correlogram cp< and the associated, correctly scaled synthetic correlogram sp<, and the z-value of the maximum of this envelope is determined as the value for the desired zp<. If necessary, a phase evaluation can also be carried out in addition to the envelope evaluation, and all resulting results can be used in the manner described or in accordance with known possibilities.

[0200] Figure 15shows a third embodiment of a device according to the invention: In this embodiment, the generation of the interference pattern is carried out by means of a Mirau interferometer: By means of the light source 2, illumination light is generated, which passes through a condenser lens 3 to a semi-transparent mirror 4. The illumination light is thus partially reflected in the illustration according to Figure 15 It is reflected downwards and impinges upon a Mirau lens, which is constructed in a conventional manner. In particular, the Mirau lens 12 comprises a front lens 12a with a central reference mirror and a beam splitter 12b.

[0201] The measurement light emerging from the Mirau lens toward the measurement object 1 is at least partially reflected by the measurement surface 1a of the measurement object and re-enters the beam path of the Mirau lens 12. At the same time, reference light is separated from the illumination light by the beam splitter 12b of the Mirau lens and directed toward the central reference mirror. There, this reference light is reflected and returned to the beam splitter 12b. A portion of the reference light is reflected again at the beam splitter 12b, so that it overlaps with the measurement light reflected by the measurement object, passes together with it at least partially through the semi-transparent mirror 4, and is imaged by a tube lens 13 onto the multi-element detector 6, which in this case is designed as a CMOS camera.The multi-element detector 6 accordingly has a CMOS image sensor as area sensor 6a, in this case an array of 512 x 512 detectors, i.e. a total of 262144 detectors.

[0202] In this embodiment, the adjustment unit 9 is designed as a piezo adjuster with a piezo controller 9a. By means of the piezo controller 9a, the adjustment unit 9 designed as a piezo adjuster can be controlled in such a way that the Mirau lens is adjusted according to the Figure 15 shown double arrow can be moved towards and away from the measuring object 1.

[0203] The remaining device, however, is not moved by the adjustment unit 9.

[0204] In this case, too, the device comprises an evaluation unit 10 designed as a laptop, which is connected to the multi-element detector 6. Analogous to Figure 12and the second embodiment, in this third embodiment the adjustment unit 9, in this case the piezo controller 9a, is also connected to the multi-element detector 6 via a trigger line.

[0205] In this case, the interferometer optics comprises the condenser lens 3, the beam splitter 4, the Mirau lens 12 and the tube lens 13.

[0206] The measurement object 1 is in this case a MEMS element, which has a slightly movable, pressure-sensitive membrane as the measurement surface 1a.

[0207] The measurement procedure is basically analogous to the measurement procedure in Figure 12In the second embodiment described above, the Mirau lens is continuously moved at an approximately constant speed by means of the piezo controller 9a and the adjustment unit 9. At equidistant time intervals, a trigger signal is sent via the trigger line to the multi-element detector 6 to record a camera image. As in the previously described embodiments, the camera images are stored and evaluated by the evaluation unit 10.

[0208] The evaluation is carried out according to a third embodiment of a method according to the invention, which is designed similarly to the previously described Figure 14 described second embodiment.

[0209] In the third embodiment, however, a large number of detector elements q are used to determine the value zp< associated with the detector element p:

[0210] The selection of the detector elements qj is carried out by using all 48 neighboring elements of the detector element p in a 7 x 7 array for the evaluation. Figure 16 For reasons of clarity, only a view of dimensions 31, 32 and 33 of the n-dimensional vector space of the correlogram vectors is shown. As in Figure 16As can be seen, the point cloud represented by the correlograms of the detector elements q 1 to q 48 can be better approximated by a curve than by a straight line. In this third exemplary embodiment, a curve 15a, in this case an ellipse segment, in an alternative embodiment a parabolic segment, is fitted to the point cloud. This results in an approximation by adapting a low-dimensional differentiable submanifold to the point distribution, in this case a one-dimensional differentiable submanifold. The approximation is carried out in this case by determining a squared deviation and optimizing the free parameters of the curve using one of the common mathematical methods so that the squared deviation is minimized.

[0211] As in Figure 16As can be seen, after determining the curve 15a for the correlogram cp<, the unscaled synthetic correlogram Q p< = sp< / fp< can be determined as a tangential vector 16a to the curve at the point p.

[0212] The further evaluation is now carried out as described above, by first determining the scaling factor fp< for the synthetic correlogram at point p and then determining the value zp< based on the correlogram cp< and the correctly scaled synthetic correlogram sp<.

[0213] However, the advantage of this third embodiment is that the determined curve can be used not only to determine the tangential vector 16a at point p, but also to determine tangential vectors to points qj (see, for example, tangential vector 16b to point q1 and tangential vector 16c to point q2). A single determination of the curve thus allows the determination of the value zp< not only for point p, but also the corresponding values ​​zq< for all or at least a plurality of points qj (j = 1, ..., 48).

[0214] In Figure 17 A fourth embodiment of a device according to the invention is shown. This largely corresponds to the first embodiment according to Figure 1However, in the fourth embodiment, the adjustment unit 9 is arranged on the reference mirror 8 in order to move it toward or away from the light source in the direction of the double arrow. The light path of the reference light is thus influenced by the adjustment unit 9. However, the position of the device relative to the measurement object 1, and in particular the measurement surface 1a, remains constant, meaning that the optical path length of the measurement light also remains unchanged.

[0215] Apart from this, however, the mode of operation is the same as that of the first embodiment. In particular, the method described for the first embodiment can also be used with the device according to Figure 17 be performed.

[0216] Based on the Figure 17illustrated fourth embodiment of a device according to the invention, a fourth embodiment of the method according to the invention is shown below: For this purpose, the device according to Figure 17 In a modification, it is designed to perform a PSI measurement. In this case, the light source 2 is a monochromatic laser with a wavelength of 520 nm.

[0217] In this embodiment, the adjustment unit 9 is used to vary the phase difference zi between the measurement light and the reference light. The light intensities lp< (zi ) are thus recorded as a function of the respective amount zi of the phase difference changed compared to the reference state of the interferometer and again result in the correlograms cp< = (cp< (1), cp< (2), ..., cp< (n)) = (lp< (z 1 ), lp< (z 2 ), ..., lp< (zn )), where the zi are phase differences, measured, for example, in degrees or radians.

[0218] In the present specific embodiment, the measurements are performed at six different values ​​for zi. The correlogram vectors cp< are thus given as vectors in the six-dimensional vector space, and the same considerations can be made as in the previous embodiments, except that the values ​​of zi are given here not as OPD differences, but as phase differences.

[0219] In this fourth embodiment of a method according to the invention, all detector elements are used for evaluation. This is shown in Figure 18 schematically shown when reduced to dimensions 3, 4 and 5: As in Figure 18As can be seen, the point cloud results in an ellipse in n-dimensional space, in this case in 6-dimensional space, of which only three dimensions are shown here. As mentioned at the beginning, phase-shifting interferometry can be considered a special case of white-light interferometry with regard to the correlograms, in which a constant envelope env(OPD) = const is given.

[0220] As explained in more detail above, the point cloud in PSI can be described as a closed ellipse. Figure 21 The case of white light interferometry will also be discussed later.

[0221] A disturbance that affects the zi values, in particular leading to non-equidistant zi, does indeed change the position of the Figure 18sketched ellipse in space, but the elliptical shape is retained. An evaluation based on parameters of the elliptical shape is thus not or only slightly affected by the described disturbances in the zi values.

[0222] According to the fourth embodiment of the method according to the invention, the Figure 18An ellipse 15a is fitted to the point cloud shown. For this purpose, the ellipse is parameterized in n-dimensional space, in this case in 6-dimensional space, using n+5 free parameters, in this case 11 free parameters, namely with n parameters to define the normal vector of the plane in which the ellipse lies, including the distance of this plane from the coordinate origin, 2 parameters to define the displacement of the ellipse in the plane relative to the point of impact of the normal vector on the plane, and 3 parameters to define the length and rotation of the principal axes of the ellipse. For each correlogram vector cp< orcq< a distance to the thus parameterized ellipse can now be specified as a function of these parameters, whereby the sum of the squared distances over all correlogram vectors cp< and cq< is minimized using one of the known mathematical minimization methods, and in this way the corresponding parameters defining the ellipse are obtained. For a point cp< of the detector element p, the synthetic (unscaled) correlogram Q p< = sp< / fp< is determined by determining the tangent vector 16a to the ellipse at point p, which is now easily possible because the ellipse is parameterized as a one-dimensional differentiable manifold and the tangent vector can be determined by differentiation. Subsequently, the scaling vector fp< is determined as described above, and the associated zp< value is determined from the correlogram cp< and the associated synthetic correlogram sp<. As in . Figure 18As is immediately apparent, the determination of the ellipse enables the evaluation not only with regard to the detector element p, but also with regard to all other detector elements, such as the illustrated detector elements q 1 , q 2 , q 3 and q 4 with associated correlograms c q1< , c q2< , c q3< and c q4< , whereby the respective associated tangential vectors 16b, 16c, 16d, 16e are also shown. With the help of these tangential vectors Q qj< = s qj< / f qj<, the associated scaling factors f qj< , the associated correctly scaled synthetic correlograms s qj< and the associated values ​​z qj< can be determined as described, which saves a great deal of computational effort compared to determining the values ​​zp< and z qj< individually.

[0223] In a modification of the fourth embodiment, a principal component analysis is first performed, thereby determining the plane in which the ellipse lies. For the points to be evaluated, for example, cp<, the point is projected into this plane to minimize noise or interference, among other things. This means shifting the point cp< perpendicular to the plane so that the point lies within the plane. After all points have been projected into the previously determined plane, the ellipse is determined. This achieves greater accuracy.

[0224] In Figure 19The point cloud is shown schematically - again for representation by projection into a three-dimensional subspace of the six-dimensional vector space, given by the light intensities measured at z 3 , z 4 and z 5 - which is obtained by determining the plane 14 using PCA and projecting the individual points into this plane. It should be noted here that the plane shown is generally not a plane in the three-dimensional space spanned by the three spatial directions shown. Although the plane is certainly a two-dimensional plane in n-dimensional space, it is generally not a plane that lies in the three-dimensional vector space spanned by the axes shown. Accordingly, the representation according to Figure 19 to be understood as a simplified illustration for better understanding.

[0225] For the further evaluation according to this embodiment of the method according to the invention, only the Figure 19 shown plane with the projected points: Figure 20 shows in part a the level 14 according to Figure 19 This was determined, as described, using principal component analysis to determine the two largest eigenvalues ​​and corresponding eigenvectors of the covariance matrix. A partial principal component analysis is also possible, which is performed until the aforementioned values ​​are determined.

[0226] The two eigenvectors 17a, 17b define the plane in which the ellipse lies. The points given by the correlogram vectors are then projected into this plane.

[0227] In Figure 20aFor simplicity, the orientation of the plane is chosen such that the x-axis of the plane is determined by the eigenvector 17b to the largest eigenvalue, and the y-axis by the eigenvector 17a to the second largest eigenvalue. The length of the semiaxes of the ellipse is proportional to the two eigenvalues. Therefore, an affine mapping can be performed that transforms the ellipse into a circle, for example, by stretching the y-axis with the ratio of the two eigenvalues.

[0228] The result is in Figure 20b shown. In the circle shown here, the relative phase angles 18 required for the evaluation can be easily determined or read. Using the phase angles 18, the zp< and z qj< values ​​for the individual detector elements and thus the individual measurement points can be determined in a known manner.

[0229] The to Figure 18The method explained can also be used in white light interferometry (WLI). Therefore, a further embodiment of a method according to the invention is presented below, in which WLI is carried out using the method described in Figure 17 The fourth embodiment of a device according to the invention is shown, wherein the light source 2 is a broadband light source such as an LED or a superluminescent diode instead of a laser. The method for recording the intensity values ​​for different zi is carried out analogously to the Figure 1described embodiment. In this case, however, the optical path length of the measuring light is not changed. The change in the OPD is achieved by changing the optical path length of the reference light by moving the reference surface 8a of the reference mirror 8 using the adjustment unit 9. In this case, too, intensities are recorded for each detector element p of the multi-element detector 6 for 64 zi values, so that a correlogram is available for each detector element, which can again be represented as a point in 64-dimensional space.

[0230] Figure 21 shows analogous to Figure 19The schematic representation of the point cloud in 64-dimensional space, although for practical reasons, only three dimensions of this space are shown. Here, too, an approximately elliptical course is evident, although it no longer forms a closed ellipse. Rather, the course resembles a helix around a cylinder with an elliptical cross-sectional area. The fact that no closed ellipse forms is due to the fact that after each orbit, the maximum of the envelope has moved further, whereas with the PSI method, the envelope can be described as a constant function, as previously explained.

[0231] But even if there is no closed ellipse, a two-dimensional plane in n-dimensional space can still be determined, in this case by principal component analysis, in which the points are approximately located, as in Figure 21 is schematically represented by level 14.

[0232] For display in Figure 21 For better clarity, points (and thus detector elements) qj were selected whose correlogram c qj< is shifted by a maximum of slightly more than half a period relative to the correlogram cp< on detector element p. This clearly shows that the ellipse does not close.

[0233] In the actual application of this embodiment of the method according to the invention, correlograms c qj< with a shift of approximately ±2 periods relative to the correlogram cp< associated with the detector element p are permitted. It is sufficient to determine the corresponding shift using one of the known prior art methods that are not as immune to interference as the method according to the invention, thereby also using a metric for determining neighborhood properties. Likewise, in alternative embodiments, a selection of points using other values ​​and / or in combination with a spatial distance metric of the detector elements is possible.

[0234] For each point p or qj, the distance vector to the specific plane is now determined (see exemplary distance vectors 19 in Figure 21). Points belonging to a correlogram shifted by one or more periods have distance vectors of different lengths relative to this plane. These distances assume relatively discrete values ​​at the respective locations belonging to the same projected point, since the ellipse winds out of the plane and emerges "one floor higher" with each orbit.

[0235] Therefore, the length of the distance vectors can be used to determine how many periods the correlogram is shifted. The length of the distance vector thus yields a result comparable to the envelope evaluation in conventional correlogram evaluation. The phase evaluation is performed as described in the previous figures.

[0236] The combination of this phase evaluation with the information on the shift of the correlogram envelope obtained from the distance vector allows the shift of the individual correlograms to be determined relatively precisely (and independently of disturbances such as vibrations), without the shift being given only modulo one period or 360° as in phase-shifting interferometry, but in this case completely.

[0237] Using the described procedure, the evaluation can be performed for a large number of points at once, saving considerable computing time. The topography of the measured object is determined from the determined correlogram shifts.

[0238] If, in a modification of the embodiment, the evaluation is carried out only for a part of the points (e.g. as described for correlograms shifted by a maximum of ±2 periods compared to cp<), the method is preferably carried out for several, preferably overlapping, sets of points so that the determined results can be combined.

[0239] A further modification of the exemplary embodiment is based on the fact that knowledge of the previously described distance vector means that the position of the envelope maximum is also known. Instead of simply projecting the point qj into the specific plane 14 as described above, with knowledge of the functional course of the envelope, only the envelope of the correlogram is shifted, for example by dividing the correlogram at point qj by the functional envelope at point qj and multiplying it by the functional envelope at point p, so that point qj is mapped into the same plane as point p. If this mapping is carried out for all points, preferably iteratively, a closed ellipse is again obtained as in the previous figures, which can also be evaluated accordingly. This increases the quality of the evaluation results and the topography determined from them.

[0240] In practical application, WLI and PSI also typically differ in that in phase-shifting interferometry, measurements are performed at significantly fewer locations, meaning the number of zi is typically smaller than in WLI measurements. In PSI measurements, the zi are phase differences relative to the reference state of the interferometer, in contrast to WLI measurements, where the zi are OPD differences relative to the reference state of the interferometer. However, phase and OPD are usually proportional, i.e., they can be converted using a conversion factor of 2π / λ (in radians) or 360° / λ (in degrees) according to Phase = 360° / λ * OPD, where λ is the effective wavelength of light.

[0241] In the previously described embodiments, the WLI measurements were performed at 64 locations (z 1 , z 2 , ..., z 64 ), so that each correlogram has 64 values ​​and the complete vector space is 64 dimensional. For PSI measurements, however, just 3 or 4 locations may be sufficient, so that for each detector element, only three or four measured values ​​are available at the three or four z-points z 1 , z 2 , z 3 or z 1 , z 2 , z 3 , z 4 .

[0242] In a variation of the previously described PSI method, only four values ​​are used. Figure 22 The influence of disturbances in such a measurement and a procedure for implementing the method according to the invention as well as the advantage thereof are explained below: Figure 22shows in sub-image a) an ideal, undisturbed signal curve as a solid line with four measured values ​​cp< (1) to cp< (4), each phase-shifted by 90°. In sub-image b), the ideal, undisturbed signal curve is again shown as a solid line, but the measured values ​​have simulated disturbances that lead to a shift in the abscissa direction z, so that the phase shift between the measured values ​​is no longer exactly 90°. The deviations from the ideal values ​​(not known in the actual measuring operation of the present embodiment) are illustrated by the differences Δz 1 to Δz 4. For the value cp< (3), the ideal, undisturbed value c id p< (3) is also marked, which thus corresponds to the value cp< (3) in sub-image a).

[0243] In this variation of the example, four values ​​of zi are used, each of which is phase-shifted by 90° in the ideal, interference-free case. Then: c p i = c OPD = A + B * cos ψ + z i

[0244] In this case, we are looking for the value ψ, which corresponds to the desired zp<, where the offset A and the amplitude B are constants of unknown value.

[0245] In conventional, state-of-the-art evaluation, the desired ψ is determined from the cp< (i) of the corresponding correlogram, which is prone to errors due to perturbations. For example, this is done with four zi values ​​shifted by 90°: ψ = arctan 2 c p 4 − c p 2 , c p 1 − c p 3 = arctan 2 cos ψ + z 4 − cos ψ + z 2 , cos ψ + z 1 − cos ψ + z 3

[0246] If actually z 2 = z 1 +90°, z 3 = z 1 +180°, z 4 = z 1 +270°, then is ψ = arctan 2 sin ψ , cos ψ = ψ , where phase unwrapping may need to be performed.

[0247] However, if the zi are subject to disturbances, such an evaluation becomes incorrect and no longer allows an undisturbed determination of ψ.

[0248] The procedure according to the invention preferably determines the associated synthetic correlogram sp< (here sp< = (sp< (1), sp< (2), sp< (3), sp< (4)) ) for the correlogram cp< (here cp< = (cp< (1), cp< (2), cp< (3), cp< (4)) ) in accordance with the procedures described in the description and in the preceding exemplary embodiments, wherein, as before, the information of other, possibly suitably selected correlograms c qj< of other detector elements qj is used.

[0249] To do this, one first determines the direction vector Q p< (zi ) = 1 / fp< * sp< (i) as described above by subtraction, approximation, fitting, principal component analysis, etc. and then determines the scaling factor fp< using one of the described methods.

[0250] A particular advantage of phase-shifting interferometry is that for all i here (cp< (i)) 2< + (sp< (i)) 2< must result in the same constant value wp<, where cp< (i) here are the light intensities lp< (zi ) of the correlograms corrected by their offset op<, i.e. cp< (i) = lp< (zi ) - op<. The offset op< can be determined, for example, (a) by averaging the lp< (zi ) for suitable zi (e.g., for four zi shifted by 90°) or (b) particularly advantageously, by determining the offset op< at the same time when determining the scaling factor f, by determining fp< and op< such that l p z i − o p 2 + f p * Q p z i 2 is constant.

[0251] In the procedure (a), op< is already present as described, and only the factor fp< has to be determined, at which (lp< (zi ) -op< ) 2< + (fp< ) 2< *(Q p< (zi )) 2< = wp< for all i yields the constant value wp<, which is preferably achieved by a linear regression of the type m * x i + b = y i with xi = (Q p< (zi )) 2< , yi = - (lp< (zi ) -op< ) 2<, where the desired value for fp< can be immediately derived from the gradient m = (fp< ) 2<.

[0252] In procedure (b), fp< and op< must be determined such that for all l p z i − o p 2 + f * Q p z i 2 = w p which again leads to a minimization problem - in this case, for example, fp< , op< and wp< must be determined so that the sum l p z i − o p 2 + f p * Q p z i 2 − w p 2 becomes minimal. This minimization problem can be solved using well-known mathematical methods and yields the desired values ​​for fp< and op< .

[0253] In each case, one obtains the synthetic correlogram sp< (here sp< = (sp< (1), sp< (2), sp< (3), sp< (4)) ) and the original diagram corrected for its offset cp< (zi ) = lp< (zi ) -op< (here: cp< = (cp< (1), cp< (2), cp< (3), cp< (4)) ).

[0254] Once the synthetic correlogram sp< has been determined, in combination with the original diagram cp< corrected for its offset, the phase position at each point zi can be correctly determined without having to use the other zi: ψ z i = arctan 2 s p i , c p i , where phase unwrapping is performed again if necessary.

[0255] Since only the differences in the phase positions at different points pj are relevant for phase-shifting interferometry to determine topography, the corresponding phase differences ψ p2< - ψ p1< can then be determined for any value of i, i.e.: ψ p 2 − ψ p 1 = arctan 2 s p2 i , c p2 i − arctan2 s p1 i , c p1 i , where, if necessary, a value is determined for different or all i, which is then averaged appropriately for the final result.

[0256] It should be noted that in the evaluation according to the invention, disturbances in the zi , for example due to vibrations, have hardly any influence on the results of the evaluation, whereby the method according to the invention generally delivers significantly better results than the conventional method according to the state of the art in the presence of disturbances. List of reference symbols

[0257] 1Measurement object 1aMeasurement surface 1bHeating plate 1cAir turbulence 2Light source 3Condenser 4Semi-transparent mirror 5Imaging optics 5a, 5bOptical lenses 5Coptical aperture 6Multi-element detector 6aArea sensor 6bDetector element 7Optical filter 8Reference mirror 8aReference surface 9Adjustment unit 9aPiezo controller 10Evaluation unit 11Reference surface 12Mirau lens 12aFront lens 12bBeam splitter 13Tube lens 14Plane determined by PCA 15Direction vector 15aCurve 16aTangential vector at p 16bTangential vector at q 1 16cTangential vector at q 2 16dTangential vector at q 3 16eTangential vector at q 4 17a, 17bEigenvectors 18relative phase angles 19distance vectors

Claims

1. Method for interferometric determination of the surface topography of a measurement object (1) with the method steps of - Generating illumination light and reference light by means of at least one light source (2) and illuminating the measurement object (1) with the illumination light, - Merging the illumination light reflected by the measurement object (1) as measurement light with the reference light, wherein an interference pattern is generated in a detection region, - Modifying the optical path length difference and / or phase difference between the measurement light and the reference light, - Detecting light intensities lp(zi) of the interference pattern on a plurality of detector elements p (6b) of a multi-element detector (6) in the detection region for at least two optical path length differences and / or phase differences modified by different amounts zi, - Determining the amount zp for the modification of the optical path length difference and / or phase difference for a plurality of detector elements p (6b) of the multielement detector (6), wherein, in each case, the optical path length difference and / or phase difference between the measurement light and the reference light reaches an explicitly and / or implicitly predetermined value for this detector element p (6b), and - Determining the surface topography of the measurement object (1) from the amounts zp determined for the different elements p of the multi-element detector (6), wherein, in addition to the light intensities lp(zi), the light intensities lq(zi) of at least one other detector element q of the multi-element detector (6) are also used to determine the value zp assigned to the detector element p (6b), characterized in that the value zp assigned to the detector element p (6b) is determined with the aid of an approximation of at least a part of the n-dimensional light intensity vectors (lp(z1), lp(z2) to lp(zn)) and lq(z1), lq(z2) to lq(zn)) provided by lp(zi) and lq(zi) or the vectors arising therefrom by projection and / or transformation.

2. Method according to the preceding claim, wherein to determine the value zp assigned to the multi-element detector element p, information is determined for an envelope and / or a phase position of the function lp(z) dependent on the modification z of the optical path length difference and / or phase difference between the measurement light and the reference light.

3. Method according to any of the preceding claims, wherein the at least one other detector element q, the light intensities lq(zi) of which are used to determine the value zp assigned to the detector element p (6b), is selected such that it has predetermined proximity properties as regards the detector element p (6b).

4. Method according to the preceding claim, wherein the proximity properties are defined at least partially with the aid of a metric, wherein this metric comprises a spatial distance of the detector elements p (6b) and q and / or a spatial distance from points assigned to them to the measurement object (1) and / or an optical surface conjugated to the multi-element detector (6).

5. Method according to any of the preceding claims, wherein the at least one metric for determining the proximity of detector elements p (6b) and q comprises a metric in the vector space of the n-dimensional light intensity vectors (lp(z1), lp(z2) to lp(zn)) and lq(z1), lq(z2) to lq(zn)) provided by lp(zi) and lq(zi) or a metric in a vector space of the vectors arising therefrom by projection and / or transformation.

6. Method according to any of the preceding claims, wherein the approximation comprises the adjustment of a differentiable submanifold M to the point distribution from the points (lp(z1), lp(z2) to lp(zn)) and lq(z1), lq(z2) to lq(zn)) in the n-dimensional vector space or in a vector space of the vectors arising therefrom by projection and / or transformation.

7. Method according to the preceding claim, wherein the differentiable submanifold M is a maximum three-dimensional submanifold of the corresponding vector space.

8. Method according to any of the preceding claims, wherein the approximation of the light intensity vectors (lp(z1), lp(z2) to lp(zn)) and lq(z1), lq(z2) to lq(zn)) or the vectors arising therefrom by projection and / or transformation is, at least locally, a linear approximation.

9. Method according to any of the preceding claims, wherein the approximation used when determining the value zp assigned to the detector element p (6b) comprises the determination of a direction vector (15), which, at least locally, in a vicinity of the light intensity vector (lp(z1), lp(z2) to lp(zn)) and / or a vector arising therefrom by projection and / or transformation, approximates to the principal direction of the point distribution provided by the vectors (lp(z1), lp(z2) to lp(zn)) and lq(z1), lq(z2) to lq(zn)) and / or the vectors arising therefrom by projection and / or transformation.

10. Method according to the preceding claim, wherein the determination of the direction vector (15) comprises a principal component analysis (PCA) or parts thereof, an intrinsic value determination, a linear regression or an other optimization method for minimizing the deviation of a straight line defined by the direction vector (15) and / or a suitable base point for it from the point distribution provided by the vectors (lp(z1), lp(z2) to lp(zn)) and lq(z1), lq(z2) to lq(zn)) and / or the vectors arising therefrom by projection and / or transformation.

11. Method according to any one of claims 9 to 10, wherein the determined direction vector or a tangent vector, wherein the tangent vector is determined for this by differentiation of the differentiable submanifold M, is used to determine a phase-shifted signal (Qp(z1), Qp(z2), ... , Qp(zn)) to the light intensities (lp(z1), lp(z2) to lp(zn) or to determine a corresponding related phase-shifted signal from the vectors (lp(z1), lp(z2) to lp(zn) arising from projection and / or transformation.

12. Method according to claim 2, wherein the determined phase position and / or envelope is used to determine the amount zp for the modification of the optical path length difference and / or phase difference for the element p of the multi-element detector (6), wherein the optical path length difference and / or phase difference between the measurement light and the reference light reaches an explicitly and / or implicitly predetermined value13. Method according to any of the preceding claims, wherein the determined amount zp for the element p of the multi-element detector (6) for the modification of the optical path length difference and / or phase difference, wherein the optical path length difference and / or phase difference between the measurement light and the reference light reaches an explicitly and / or implicitly predetermined value, is used to obtain information on the spatial position of a surface point of the measurement object (1) assigned to the detector element p (6b), in order to determine the surface topography of the measurement object (1).

14. Method according to any of the preceding claims, wherein the method is designed to determine the surface topography of measurement objects (1) in the presence of at least one of the following disturbances: (i) Vibrations or movements, which affect the length of the measurement and / or reference arm of the interferometer during measurements (ii) Modifications in the density and / or refractive index of at least one of the media through which light passes in the measurement and / or reference arm (iii) Inaccuracies and / or errors when determining the amount z of the modification of the optical path length difference and / or phase difference between the measurement light and the reference light by the adjustment unit (9)15. Device for interferometric determination of the surface topography of a measurement object (1) with at least one light source (2) for illuminating the measurement object (1) and for generating reference light, with an interferometer optic for merging the illumination light reflected by the measurement object (1) as measurement light with the reference light, wherein an interference pattern is generated in a detection region, with at least one adjustment unit (9), to modify the optical path length difference and / or phase difference between the measurement light and the reference light by an amount z, with a multi-element detector (6) with a plurality of detector elements p, which is designed to detect the light intensities lp(zi) of the interference pattern on the detector elements p in the detection region for at least two optical path length differences and / or phase differences modified by different amounts zi, with an evaluation unit (10), which is connected to the multi-element detector (6) and is designed to determine the amount zp for the modification of the optical path length difference and / or phase difference for a plurality of detector elements p of the multielement detector (6), wherein the optical path length difference and / or phase difference between the measurement light and the reference light reaches an explicitly or implicitly predetermined value for this detector element p (6b), and, from this, to determine the surface topography of the measurement object (1), wherein the evaluation unit (10) is designed, in addition to the light intensities lp(zi), to also use the light intensities lq(zi) of at least one other detector element q of the multielement detector (6) to determine the value zp assigned to the detector element p (6b) characterized in that the evaluation unit (10) is designed to determine the value zp assigned to the detector element p (6b) with the aid of an approximation of at least a part of the n-dimensional light intensity vectors (lp(z1), lp(z2) to lp(zn)) and lq(z1), lq(z2) to lq(zn)) provided by lp(zi) and lq(zi) or the vectors arising therefrom by projection and / or transformation.