TW201335571A - Apparatus and method of simultaneously detecting three dimensional surface skeleton and optical level surface roughness - Google Patents

Abstract

The present invention provides an apparatus and a method of simultaneously detecting the three dimensional surface profile and the optical level surface roughness. The invention is based on differently configured microscope interferometers incorporating techniques of fast Fourier transform, interference phase recovery and digital filtering. Three differently configured microscope interferometers are used, including the improved Michelson, symmetric Linnik and asymmetric Linnik microscope interferometers, and Mirau microscope interferometer, the Fourier transform and Gaussian filter are also used to detect the three dimensional surface morphology and the surface roughness of an micro-optical element. Experimental results verify that the symmetric Linnik microscope interferometer has the highest transversal and longitudinal resolutions and the lowest system measurement error among the microscope interferometers. The system has the same variation trend for detecting roughness as that of the atomic force microscopy in measuring thin film samples. The method has advantages of easy operation, precise measurement, non-destructive measurement, high resolution and wider area detection on roughness.

Description

Device and method for simultaneously detecting three-dimensional surface contour and optical grade surface roughness

The present invention relates to an apparatus and method for simultaneously detecting a three-dimensional surface profile and an optical grade surface roughness.

In recent years, miniaturized components, such as microlenses and optical thin film components, have become more and more widely used in miniaturized optical systems, and their use ranges include miniaturized projection systems or USB 3.0 optical communication transmission systems. All of this depends on the basic characteristics of the focusing and collimation of the beam by the microlens, and with the development of miniaturized manufacturing technology, more micro-lens and optical film components can be further fabricated, but the relative need Accurate measurement techniques provide surface profile and roughness characteristics for microlens and optical film components. In order to be able to determine the geometric parameters of the microlens and optical film elements, the prerequisite is that the three-dimensional surface profile must be reconstructed. Secondly, by reconstructing the surface topography, the radius of curvature of the microlens and the geometric parameters of the optical film element can be further estimated. . In addition, the surface roughness of the microlens and optical film elements affects the surface's scattering loss and microscopic characteristics, which have a significant impact on the performance of the micro-components. In order to meet the practicality, it is proposed to develop a non-contact measuring system capable of simultaneously detecting the three-dimensional surface contour and surface roughness of micro components.

The measurement system commonly used today for three-dimensional surface contours and roughness is an atomic force microscope. Although it has a very high longitudinal resolution, the measurement process is time consuming and expensive. If probe-based measurement techniques are used, the surface of the sample may be destroyed by contact of the probe with the surface of the film.

In view of the limitations of general roughness measurement methods such as substrate size, material, etc., some methods require surface probe height measurement by contact probe scanning or atomic force non-contact scanning. It is more time consuming, and the contact type is easy to scratch the surface of the object to be tested.

Based on the shortcomings of the above-mentioned various current technologies, how to provide a fast, convenient, and accurate measurement method is an important topic in this field.

Accordingly, it is a primary object of the present invention to provide an apparatus and method for simultaneously detecting a three-dimensional surface profile and an optical grade surface roughness.

In order to achieve the object of the present invention, the technical means used in the present invention mainly applies the proposed system architecture to different miniaturized components, based on micro-interferometers of different architectures, combined with fast Fourier transform method, interference. Phase reduction method and digital filtering technology.

The technical means used in the present invention mainly utilizes a microlens array and an optical film to perform topography reduction, axial curvature radius fitting and surface roughness detection of a three-dimensional surface profile. It is proved that the proposed technology can be applied to the characteristic detection of other miniaturized components, and an optical measuring system with simple operation, accurate measurement, non-contact measurement, high resolution and large global area roughness is designed.

The technique used in the present invention is mainly to use three different interferometers: Michelson, symmetrical Linnik and asymmetric Linnik microinterferometer, and Mirau microinterferometer architecture, combined with Fourier transform The method and the Gaussian filter can detect the three-dimensional surface topography and roughness of the micro-optical components.

The system architecture proposed in the present invention can be measured with a microscope objective lens of any magnification and numerical aperture, thereby improving the problem of insufficient lateral resolution of the conventional micro-interferometer system, and the present invention is combined with the Fourier transform method. In the system configuration, the size of the space carrier is artificially applied, and after the equal-strength interference fringes are captured, the three-dimensional surface profile of the sample can be restored by a single interference image, and the axial radius of curvature is fitted and calculated, and finally by digital The Gaussian filter filters out the low-frequency surface profile signal of the sample and retains the sample high-frequency roughness signal to calculate the surface roughness of the sample.

In addition, the system configuration proposed by the present invention can be adjusted with different detection samples for optical path adjustment, and can also be combined with microscope objectives of different magnifications for detection. The designed system architecture is easy to operate, accurate measurement, and non-destructive. The advantages of sex measurement, high resolution and large global area roughness detection.

It has been confirmed by experiments that in the system of symmetric Linnik micro-interferometer, it has the highest lateral and longitudinal resolution and the lowest system measurement error. The roughness measurement results of the film sample are consistent with the atomic force microscope. trend.

The method has the advantages of simple operation, accurate measurement, non-destructive measurement, high resolution and large global area roughness detection.

The invention is based on a micro-interferometer of different architectures, and combines a fast Fourier transform method, an interference phase reduction method and a digital filtering technique. In principle, a two-dimensional Fourier transform method is adopted, and the theory is described as follows.

First consider the carrier intensity, the relationship between the two light interference after the light intensity distribution:

I(x,y) = a(x,y) + b(x,y)cos[2 π f _0,x + 2 π f _0,y + Φ(x,y)] (1)

In (1), a(x, y) is the background intensity signal of the stripe morphology, b(x, y) is the stripe matching coefficient, f _{0, x} and f _{0, x} represent the x direction and y _, respectively. The carrier frequency of the direction, Φ(x, y) is the wavefront phase information. The plural form of its light intensity distribution can be expressed as:

I(x,y) = a(x,y) + c(x,y)exp(2 π jf _0,x x + 2 π f _0,y y) + c ^* (x,y)exp(-2 π jf _0,x x-2 π f _0,y y) (2)

Where * is a conjugate complex number and c(x,y) = b(x, y)exp[jΦ(x, y)] is a complex stripe morphology. Using (2) for fast Fourier transform, you can get:

I _F (x,y) = A(f _x ,f _y ) + C(f _x -f _0,x ,f _y -f _0,y ) + C ^* ( f _x + f _0,x ,f _y + f _0,y ) (3)

After the fast Fourier transform, the original interference signal is converted to the Fourier spectrum, and the Fourier transform plane is observed. It can be seen that the spatial variables a(x, y) , b(x, y) and Φ(x, y) are lower spatial carriers. The frequency f ₀ is to separate the carrier frequency f _{0 of the} equation (3) from the Fourier spectrum. As shown in Fig. 1, C(f _x -f _0,x ,f _y - f _0,y ) is滤波器 by the filter Take the translation to the center spectrum to eliminate the carrier signal, and then perform inverse Fourier transform on it, then extract c(x, y) and calculate the Φ(x, y) phase distribution:

However, the phase after arctangent calculation will be limited to the 2π modulus of π~-π, which is a discontinuous phase distribution. Therefore, phase reduction is required by phase reduction method to increase or decrease the discontinuous phase distribution by 2π. Make it a continuous phase distribution and restore it to the original phase function Φ(x,y) value to reconstruct the sample surface profile.

Secondly, in order to expand the two-dimensional discontinuous phase signal, the phase expansion method of Macy can be used to reduce the adjacent and discontinuous phase of the x and y axes by an integer multiple of 2π to reduce it to two. The correct phase distribution of the dimension.

First, we will first consider the single x direction and expand the phase of adjacent pixels. The calculation formula is as follows:

Φ '( x _i , y )= Φ ( x _i , y )+2 p π (5)

Where Φ '( x _i , y ) is the original phase after phase unwrapping in the x direction, Φ ( x _i , y ) is the discontinuous phase before the phase expansion in the x direction, x _i is the position of any point on the x axis, p is the phase Increase or decrease any integer of 2π. And since the phase of the adjacent pixel is subjected to the arctangent operation, it is limited to the 2π modulus of π~-π. that is

-π< Φ '( x _i , y )- Φ '( x _{i -1} , y )<π (6)

By adding π to equation (5), it can be expanded to

0< Φ ( x _i , y )- Φ '( x _{i -1} , y )+(2 p +1)π<2π (7)

This relationship gives a recursive relationship of phase expansion

Φ '( x _i , y )= remainder [ Φ ( x _i , y )- Φ '( x _{i -1} , y )+(2 p +1)π,2π]+ Φ '( x _{i -1} , y )-π (8)

The remainder is a recursive function by which any point x _{i on the} x-axis can be restored to the correct phase.

After phase unwrapping in the x direction, phase unwrapping is then performed in the y direction.

Φ '( x _c , y _j )= Φ ( x _c , y _j )+2 p _k π (9)

Where Φ '( x _c , y _j ) is the original phase after phase unwrapping in the y direction, Φ ( x _i , y _j ) is the discontinuous phase before the phase expansion in the y direction, x _c is the center position of the x axis, and y _j is The position of any point on the y-axis, p _k is an arbitrary integer whose phase is increased or decreased by 2π. Due to 2π modulus limit, available

-π< Φ '( x _c , y _j )- Φ '( x _c , y _{j -1} )<π (10)

After increasing π and dividing by 2π in equation (5), it can be expanded to

0<{[ Φ '( x _c , y _j )- Φ '( x _c , y _{j -1} )]/2π+0.5}+ p _{k -1} (11)

This relationship gives a recursive relationship of phase expansion

Φ '( x _i , y )= INT {[ Φ '( x _c , y _j )- Φ '( x _c , y _{j -1} )]/2π+0.5}+ p _{k -1} (12)

Where INT is a recursive function and p _k is the number of increments or decrements of 2π by reducing any point y _{j on the} y-axis to the correct phase.

By the above calculation, and by the following equations, the two-dimensional phase can be correctly expanded.

Φ '( x _i , y _j )= Φ ( x _i , y _j )+2 p _k π (13)

After two-dimensional phase unwrapping, as shown in Fig. 2, the two-dimensional discontinuous phase signal can be expanded into a two-dimensional correct phase signal, and finally multiplied by λ/4π to convert the height change of the sample surface to reconstruct the sample. 3D contour of the surface.

Finally, in the measurement of the surface roughness of the film, due to the surface reconstruction of the surface to be measured z(x), including part of the rough surface signal r(x), and part of the wavefront signal ω(x), As shown in Fig. 3, the digital Gaussian filter S(x) must be used as a signal separation tool, and since the Gaussian filter has a linear phase characteristic, it is suitable for one of the most commonly used filters. Used for signal separation of roughness.

When designing a Gaussian filter, the cutoff wavelength λ _{c of} its weight function must be defined, and the reconstructed surface is convolution. The low frequency surface signal of the film can be obtained, and then the Fourier Transform Method (FTM) The reduced reconstructed three-dimensional surface contour signal is subtracted from the low-frequency surface signal to obtain a surface roughness signal of the film, and the three-dimensional surface roughness profile of the film is reconstructed.

Figure 4 shows the penetration characteristics of the Gaussian filter. If the wavelength of a sine wave is equal to the cutoff wavelength of the Gaussian filter, only 50% of the sine wave signal can pass the Gaussian filter. The weight function of the Gaussian filter and the penetration characteristics corresponding to different wavelengths are expressed as follows:

Where S(x) is defined as a Gaussian filter function, A _output / A _input is a Gaussian filter transfer function, and α is a constant equal to , x is the weight function from the origin position, and λ _c is the cutoff wavelength.

In order to simultaneously detect the three-dimensional surface contour and roughness parameters, a film sample is taken as an example, and the software calculation and analysis process is as shown in FIG. Firstly, a micro-interferometer is used with a digital CCD camera to capture an interference image. When the interference image of the appropriate carrier signal is applied by fast Fourier transform, the converted Fourier space spectrum can be observed with three major harmonics. It is the center spectrum of the background signal and the two side waves that are conjugate with the phase information. Here, the side wave containing the phase signal is extracted, and the interference image to which the appropriate carrier frequency is applied must be extracted before the analysis. The wave is separated from the background signal. Secondly, after the side wave is extracted and the fast Fourier inverse conversion is performed, the discontinuous test surface phase information can be obtained, and the discontinuous surface phase signal is phase-expanded. The Fourier transform method is used to reconstruct the three-dimensional surface contour distribution of the film. Finally, the reconstructed three-dimensional surface contour is filtered by setting the cutoff wavelength of the Gaussian filter, and the obtained low-frequency corrugated surface signal of the film is compared with the reconstructed three-dimensional surface contour. Subtraction, obtaining a high-frequency three-dimensional film surface roughness profile signal, and thereby calculating the film surface level Roughness and root mean square roughness value. We use the self-developed MATLAB program to reconstruct the 3D surface profile and roughness profile of the film. The important processing procedure of this program is shown in Figure 6(a)-(h).

In the present invention, in order to apply the proposed experimental architecture to different miniaturized components, it is proposed to verify the spherical microlens array, which includes topographical reduction of the three-dimensional surface profile and axial radius of curvature fitting. Detection. The results of the spherical microlens array measured by an optical image microscope are shown in Fig. 7. This measurement is close to the actual size of the spherical microlens with a diameter of 100 μm, and the radius of curvature of the spherical microlens is 1.51 mm.

In the present invention, the optical path distance between the two ends can be kept under the same length condition by adjusting the system architecture of the interferometer, and the reference objective lens with two specifications of λ/20 is placed on the reference end and the sample end. The axial distance between the reference objective lens and the test objective lens to the reference mirror and the test sample is kept fixed, and the interference image formed under this condition is a straight strip of equal thickness interference fringe, which avoids the objective lens at both ends of the interferometer because of the axis The distance to the distance is different, and the introduced axisymmetric surface error occurs. Secondly, after the reference objective lens of the sample end is replaced by the spherical microlens array, the light passing through the test objective can be concentrated on a single spherical microlens in the microlens array for reflection by adjusting the axial distance between the test sample and the front and back. This condition can introduce the amount of wavefront curvature change caused by the surface topography of the spherical microlens, and the wavefront curvature change caused by the mismatch of the axial distance of the objective lens at both ends of the interferometer. Finally, the equal-thickness interference fringes obtained under the above conditions can be reconstructed by software program analysis, and the axial radius of curvature is fitted.

Fig. 8 is a measurement result of the three-dimensional surface reduction of the spherical microlens and the axial radius of curvature fitting. The single spherical microlens measured by the method is measured within a measurement range of the microlens diameter of 100 μm. The x-axis and y-axis curvature radii are 1.60 mm and 1.51 mm, respectively. The measurement results are similar to the known spherical microlens with a radius of curvature of 1.51 mm. This result can also be used to examine the accuracy of the experimental architecture measurement. In addition, the 3D surface roughness profile reconstructed by Gaussian filtering and the cross-sectional profile of the x-ray y-axis of the microlens, the height distribution of the x, y axial surface, can be numerically calculated to obtain the average roughness of the microlens R _a = 0.869 nm and root mean square roughness R _rms = 1.116 nm.

Micro-interferometers can generally be divided into three structures: Michelson, Mirau and Linnik, as shown in Figure 9. Different types of microstructures can be used to determine the surface roughness of a sample by using several different optical metrology techniques. It has the advantages of accurate, non-contact and large-area measurement.

The first micro-interferometer architecture proposed by the present invention is a modified Michelson micro-interferometer. As shown in FIG. 10, the operation flow of the device is as follows:

First, we use 氦氖 laser (1) as the light source. When the laser light is incident on the spatial filter (2) and the collimating lens (3), a parallel light is formed, which can be divided into two intensity by the beam splitter (4). Equal parallel light, one of which is projected onto a reference mirror (5) mounted on a one-dimensional platform (6), and the second through a microscope objective (7) mounted on a separately controllable one-dimensional translation stage By adjusting the working distance of the microscope objective (7), the light beam is concentrated to form a light cone and projected onto the surface of the microlens sample (8) at the front focal plane position of the objective lens. The microlens sample is mounted on a three-dimensional stage ( 9), the one-dimensional platform (6) and the three-dimensional stage (9) can conveniently perform position sampling and adjustment on the optical path, and the reference mirror (5) is mounted on a tilt platform. In order to apply the required spatial carrier size in a manual intervention. Secondly, when the two beams are respectively reflected by the microlens sample surface (8) and the reference mirror (5), and superposed on the imaging plane via the beam splitter (4), the imaged lens (10) is bunched to the digital CCD camera (11). The equal-strength interference fringes are formed in the image, and finally the CCD camera (11) mounted on the imaging plane is used to capture the interference fringe image, and the analog signal is converted into a digital signal and stored in the computer (12) through the image capture card. The roughness measurement program was used as the analysis.

Compared with the traditional Michelson microinterferometer, the modified Michelson microinterferometer can effectively improve the lateral resolution of the system by using an objective lens of any magnification and numerical aperture, which can effectively analyze the microlens. Surface topography, but this asymmetrical optical path architecture may not only result in the use of objective differences between the sample end and the reference end, but also may introduce optical aberrations after the light passes through the objective lens, and the light to be measured with different detection diameters And the reference light, therefore also causes lateral errors of interference to occur.

The second microinterferometer architecture is a modified Linnik microinterferometer, as shown in Figure 11. In the modified Michelson microinterferometer, the test optical path and the reference optical path may be due to the presence or absence of the installation of the microscope objective (7), resulting in a difference in the introduction of the image between the two optical paths, and the light to be measured and the reference light of different diameters, It will also cause lateral errors of the interferometer to occur. To this end, the present invention proposes an improved Linnik microinterferometer device which can be used with two first objective lenses (26) and a second microscope objective (28) of the same specification as test samples. In theory, the error effects caused by the above reasons can be effectively deducted.

The configuration of the Linnik microinterferometer requires that both the reference mirror (27) and the sample surface (8) be located on the front focal plane of the first microscope objective (26) and the second microscope objective (28), and reference The first microscopic objective lens (26) and the second microscopic objective lens (28) in the end and the sample end are each individually controlled by a single-dimensional translation stage, so as to adjust the working distance of the objective lens. The entire reference end is also constructed on a composite stage (25) which is a combination of a tilt platform and a one-dimensional platform. The stage (25) can be adjusted with the three-dimensional stage (9) at the sample end for optical path adjustment. The amount of spatial carrier applied with human control. The reflected light from the reference mirror (27) and the sample surface (8) is reflected back to the first microscope objective (26) and the second microscope objective (28), and the field of view of the two reflected light is It is necessary to maintain consistency. Under these conditions, straight strip interference fringes can be formed, and the wavefront curvature of the reflected light at the reference end and the test end will also coincide with each other.

The modified Linnik microinterferometer, the operation steps of the interference image capture are as follows:

First, we use 氦氖 laser (1) as the light source. When the laser light enters the spatial filter (2) and the collimating lens (3), it forms a parallel light. When the beam is split into two beams through the beam splitter (4) Equal parallel light, one of which is projected to the reference end, and the other is passed through the test end. When the parallel beam passes through the first microscope objective (26) and the second microscope objective (28), respectively, it is necessary to fine-tune the operation of the two objective lenses simultaneously. The distance, the beam is concentrated and projected onto the reference mirror surface (27) on the front focal plane position of the objective lens and the sample surface (8) of the test, and the test light and reference of the objective lens (27) and the test surface (8) are respectively reflected back to the objective lens. Light, the wavefront curvature of the two reflected lights needs to be expanded to the same angle of view, and secondly by the beam splitter (4) coincident with the imaging lens (10) bunched on a digital CCD camera (21) to form an equal thickness Straight interference fringes, and then adjust the tilt angle of the reference end composite stage (25) to apply the required spatial carrier frequency. Finally, the interference fringe image is captured by the CCD camera (21), and the analog signal is converted to digital. After the signal, save the image capture card to the computer (12), and finally use the thick Measured as a measure of program analysis.

In the present invention, the Linnik micro-architecture can also be combined with a set of first objective lens (26) and second microscope objective (28) with mismatched magnification, but the difference in configuration of the configuration, in addition to the reference The mirror surface (27) and the sample surface (8) need to be located on the front focal planes of the two objective lenses (26) and (28), and the wavefront curvature of the reflected light of the two objective lenses (26) and (28) must also be maintained after being reflected. Consistent, but if a pair of mismatched first microscope objective (26) and second microscope objective (28) are used, the difference between magnification and field of view will result in the first microscope objective during interference imaging. (26) The reflected field of the second microscope objective (28) has the same wavefront curvature, and the field of view of the two beams is different. Therefore, this situation will be limited by the smaller numerical aperture objective. The size of the field of view. Although the asymmetric measurement architecture can also use objective lenses with different magnifications and arbitrary numerical apertures, it can also be observed that the aberrations introduced by the objective lens are different due to different objective magnifications used at both ends of the interferometer, and in addition, due to the light to be measured A slightly different detection diameter than the reference light will also cause lateral errors in the interference, which will also affect the accuracy of the system detection.

The above are only the preferred embodiments of the present invention, and all changes and modifications made to the scope of the present invention should be within the scope of the present invention.

(1). . . Laser

(2). . . Spatial filter

(3). . . Collimating lens

(4). . . Beam splitter

(5). . . Reference mirror

(6). . . One-dimensional platform

(7). . . Microscopic objective

(8). . . Sample surface

(9). . . Three-dimensional stage

(10). . . Imaging lens

(11). . . CCD camera

(12). . . computer

(twenty one). . . CCD camera

(25). . . Composite stage

(26). . . First microscope objective

(27). . . Reference objective

(28). . . Second microscope objective

Figure 1 is a schematic diagram of the side wave interception;

Figure 2 is a schematic diagram of two-dimensional phase expansion;

Figure 3 is a surface contour signal separation diagram;

Figure 4 is a diagram showing the penetration characteristics of a Gaussian filter;

Figure 5 shows the roughness separation and calculation process;

Figure 6 is a measurement of the surface profile and roughness of the film 3D;

Figure 7 is an optical microscopic image of a spherical microlens array;

Figure 8 is a schematic view of a spherical microlens;

Figure 9 is a diagram of three different microscopic interference architectures;

Figure 10 is a schematic diagram of a modified Michelson microinterferometer;

Figure 11 is a schematic diagram of the Linnik microinterferometer.

Claims (12)

A method for simultaneously detecting a three-dimensional surface profile and an optical grade surface roughness, the method comprising: providing a light source to illuminate a surface of the object to be tested; acquiring a discontinuous phase difference signal; and changing the discontinuous phase difference signal to a continuous signal; The continuous phase difference signal is restored to the surface height change of the sample; and the three-dimensional surface contour information of the object to be tested is reconstructed. A method for simultaneously detecting a three-dimensional surface profile and an optical grade surface roughness as described in claim 1, wherein the step of changing the discontinuous phase difference signal to a continuous signal is by adding or subtracting a phase difference by 2π. The method for simultaneously detecting a three-dimensional surface profile and an optical grade surface roughness as described in claim 1, wherein the step of reconstructing the three-dimensional surface profile information of the object to be tested is mainly by multiplying the reduced continuity phase difference by λ/ 4π, where λ is the wavelength of the source. A method for simultaneously detecting a three-dimensional surface profile and an optical grade surface roughness as described in claim 1, wherein the discontinuous phase difference signal is formed by two reflected light beams having uniform viewing angles. A method for simultaneously detecting a three-dimensional surface profile and an optical grade surface roughness as described in claim 1, wherein the light source provided is a neon laser or a single frequency (color) light source. The method for simultaneously detecting the three-dimensional surface contour and the optical grade surface roughness as described in claim 1 of the patent application further includes the step of filtering the three-dimensional surface contour information of the reconstructed object to be tested. The method for simultaneously detecting a three-dimensional surface contour and an optical grade surface roughness as described in claim 6, wherein the filtered three-dimensional surface contour information of the object to be tested is compared with the three-dimensional surface contour information of the object to be tested after reconstruction. Obtain the surface signal of the object to be tested. A device for simultaneously detecting a three-dimensional surface profile and an optical grade surface roughness comprises: a light source; a beam splitter splitting the light from the light source into two beams; and a microscope objective lens respectively focusing the two beams from the beam splitter; the reference mirror a beam that reflects through the microscope objective; a surface of the object to be tested that reflects another beam that passes through the microscope objective; and a camera that captures and records the two beams to form a thick strip of interference fringes. A device for simultaneously detecting a three-dimensional surface profile and an optical grade surface roughness according to the scope of claim 8 wherein the light source is a laser or a single-frequency (color) light source. A device for simultaneously detecting a three-dimensional surface profile and an optical grade surface roughness according to claim 8 of the patent application, which includes a spatial filter and a collimating lens between the light source and the beam splitter. The apparatus for simultaneously detecting a three-dimensional surface profile and an optical grade surface roughness according to claim 8 of the patent application, wherein the microscope objective comprises a first microscope objective and a second microscope objective. The apparatus for simultaneously detecting a three-dimensional surface profile and an optical grade surface roughness according to claim 11, wherein the reference mirror surface and the surface of the object to be tested are both located before the first microscope objective and the second microscope objective lens. On the focal plane.