You are on page 1of 45

8. Testing of Curved Surfaces and/or Lenses 8.1 Radius of Curvature 8.1.1 Spherometer 8.1.2 Autostigmatic Measurement 8.1.3 Newton’s Rings 8.1.

4 Interferometer and Radius Slide 8.2 Surface Figure 8.2.1 Test Plate 8.2.2 Twyman-Green Interferometer (LUPI) 8.2.3 Fizeau (Laser source) 8.2.4 Spherical Wave Multiple Beam Interferometer (SWIM) 8.2.5 Shack Cube Interferometer 8.2.6 Scatterplate Interferometer 8.2.7 Smartt Point Diffraction Interferometer 8.2.8 Sommargren Diffraction Interferometer 8.2.9 Measurement of Cylindrical Surfaces 8.2.10 Star Test 8.2.11 Hartmann Test 8.2.12 Foucault Test 8.2.13 Wire Test 8.2.14 Ronchi Test 8.2.15 Lateral Shear Test 8.2.16 Radial Shear Test

8.0 Curved Surfaces and/or Lenses 8.1 Radius of Curvature 8.1.1 Spherometer A spherometer is an instrument that measures the sag of a surface with great precision. A common spherometer is the Aldis spherometer in which three small balls are arranged to form an equilateral triangle. In the center of the triangle there is a probe mounted on a micrometer. In use, the surface to be measured is placed on the balls, and the probe is brought into contact with the surface. The sag of the surface is measured using a micrometer. If we denote the sag of the surface by h, the distance between the center of the balls d, and the radii of the balls by r, then the radius of curvature of the unknown surface is given by R= d2 h + ± r, 6h 2

Where the positive sign is taken when the unknown surface is concave, and the negative sign when convex. If the balls are arranged to form a triangle of sides d1, d2, and d3, the radius of curvature is given by


(d1 + d 2 + d3 )2 + h ± r.
54h 2

If a ring spherometer of radius r is used, the radius of curvature is given by r2 h + . 2h 2


The major source of inaccuracy in using a spherometer is determining the exact point of contact between the probe and the surface being tested. In some spherometers the point of contact is determined by observing the Newton’s ring interference pattern formed between the test surface and an optical surface mounted on the end of the probe. As the probe is brought up to the surface, the ring pattern expands, but when the point of contact is reached, no further motion occurs.


Autostigmatic Measurement

A travelling microscope with a vertical illuminator is often used to measure the curvature of a surface directly, as shown in Fig. 8.1.2-1. The instrument consists of a normal microscope configuration with a small beamsplitter placed behind the objective. A tungsten bulb is placed off to the side of the beamsplitter by a distance equal to the separation of the eyepiece reticle from the beamsplitter. If the microscope is focused on a surface, an image of the filament of the flashlight bulb will be projected on the surface, and reimaged in the plane of the reticle by the microscope objective. To measure the radius of curvature of a surface, the microscope is first focused on the surface, and then on the center of curvature of the surface as shown in Fig. 8.1.2-1. The separation of the two focus positions is equal to the radius of curvature of the surface. A convex surface can be measured using an auxiliary positive lens as shown in Fig. 8.1.2-2.

Fig. 8.1.2-1 Autostigmatic test for measuring surface curvature.

Fig. 8.1.2-2 Autostigmatic test for measuring surface curvature of convex surfaces.


Newton’s Rings

A Fizeau interferometer can also be used to measure radius of curvature of a surface. A surface having a long radius of curvature can be compared interferometrically with a flat surface to yield Newton’s rings as shown in Fig. 8.1.3-1. When viewed from above we see a series of concentric rings around a central dark spot. The radius ρm of the mth dark ring from the center is given by

ρ m = mλR ,
where R is the radius of curvature of the surface being measured.

ρ2 R= mλ

m = order #

Fig. 8.1.3-1 Radius of curvature measurement using Newton’s rings.

Shorter radius of curvatures can be measured with a Fizeau interferometer if a reference surface of approximately the same radius of curvature as the surface being measured is available. For example, Fig. 8.1.3-2 shows the result of comparing a reference surface having radius of curvature R with a test surface having approximately the same radius of curvature. Naturally, one surface is concave, while the other is convex. If d is the diameter of the piece under test, and m is the number of fringe spacings the fringes depart from straightness, the difference in radius of curvature of the two surfaces is given by 4mλR 2 . d2

∆R =

Helium Lamps


Beamsplitter Part being tested Test Plate

4mλR 2 ∆R = d2

Fig. 8.1.3-2. Radius of curvature measurement using a spherical test plate.


Interferometer and Radius Slide

Measuring Radius of Curvature


Two positions which give null fringe for spherical mirror.
1998 - James C. Wyant Part 3 Page 25 of 28


Surface Figure

Two categories 1) Techniques that measure the surface or wavefront directly. 2) Techniques that measure surface or wavefront slope.

8.2.1 Test Plate We have previously described the Fizeau interferometer for measuring surface flatness. The exact same procedure can be used to measure the surface figure of a curved surface, even an aspheric surface, if a reference surface is available. The interferograms as analyzed in the same manner as described above. It should be noted that the two surfaces must have nearly the same radius of curvature if the resulting interferogram is to have a sufficiently small number of fringes to be analyzed.


Twyman-Green Interferometer

The basic Twyman-Green interferometer or LUPI (Laser Unequal Path Interferometer) described above can be used to test concave or convex spherical mirrors. In the test setup the center of curvature of the surface under test is placed at the focus of the diverger lens so the wavefront is retroreflected back onto itself. In a like manner a lens can be tested. The good features of the LUPI are (1) any size concave optics can be tested; (2) the two paths in the interferometer do not have to be matched as long as the conditions for coherence are satisfied; (3) surface contours are obtained just as for a test plate; and (4) the test is a noncontact test. The disadvantages are (1) the test is sensitive to vibration and turbulence; and (2) expensive. Mounting the reference mirror on the surface being tested can reduce bad effects of vibrations. In this case the vibrations are coupled and the interference pattern will become much more stable. If the mirror is close to the interferometer a pencil or meter stick can be placed on top of the mirror and the interferometer to couple vibrations.

Classical Fizeau Interferometer
Helium Lamps


Ground Glass With Ground Side Toward Lamps

Part to be Tested Test Glass

1998 - James C. Wyant

Part 1 Page 4 of 43

Twyman-Green Interferometer (Spherical Surfaces)
Reference Mirror Diverger Lens Laser Beamsplitter Imaging Lens Test Mirror

Incorrect Spacing
1998 - James C. Wyant

Correct Spacing

Part 1 Page 10 of 43


Fizeau (Laser Source)

Fizeau Interferometer-Laser Source (Spherical Surfaces)
Beam Expander Laser Diverger Lens Reference Surface

Imaging Lens

Test Mirror


1998 - James C. Wyant

Part 1 Page 13 of 43

8.2.4 Ref:

Spherical Wave Multiple Beam Interferometry (SWIM) Heintze, Polster, and Vrabel, Appl. Opt. 6 1924 (1967) Polster, Appl. Opt. 8 522 (1969)

Two surfaces being compared in a Fizeau interferometer can be given a high reflectance coating to yield sharp, high-finesse multiple-beam interference fringes. The surfaces being compared can be either flat or spherical, the only restriction being that the two surfaces under test must have essentially the same radius of curvature and the reflectance surface must be at least as large in diameter as the surface being tested. These restrictions can be eliminated for the testing of concave surfaces by using the setup shown in Fig. 8.2.4-1, where a single reference surface can be used for the testing of any diameter and radius of curvature concave surface.

Fig. 8.2.4-1 Spherical wave interferometer, multiple-beam (SWIM).

For the SWIM shown in Fig. 8.2.4-1, both the reference surface and the surface being tested must be coated. At the common center of curvature of the two surfaces a lens is placed that images one surface on the other. This lens is very important since, without it, any surface deformation in the mirror being tested or any displacement of the two centers of curvature will cause beam walkoff, which greatly reduces the fringe finesse as shown in Figure 8.2.4-2. A laser having a single longitudinal mode should be used with the SWIM. For a given aberration, the shape of the interference fringes obtained using the SWIM is the same as for the LUPI; the difference being the increased finesse of the fringes

obtained using the SWIM. This increased finesse is useful if the interferogram is to be interpreted by eye; however, if the interferogram is to be scanned with a comparator, the gain in accuracy achieved by using multiple beam interferometry to give high-finesse fringes is minimal. The greatest disadvantage of the SWIM is that the surface being tested must be coated to have a high reflectance, which is a problem except for the final testing where the mirror would generally already be coated.

Fig. 8.2.4-2 Reduced finesse fringes due to beam walkoff. (ref: K. Kinosita, “Numerical evaluation of the intensity curve of a multiple-beam Fizeau fringe,” J. Phys. Soc. Jpn. 8, 219-225 (1953), J. R. Rogers, “Fringe shifts in multiple-beam Fizeau interferometry”, J. Opt. Soc. Am., 72, 638-643 (1982).)


J. Opt. Soc. Am./Vol. 72, No. 5/May 1982

John R. Rogers

Fringe shifts in multiple-beam Fizeau interferometry
John R. Rogers
Optical Sciences Center, University of Arizona, Tucson, Arizona 85721 Received August 7, 1981; revised manuscript received January 7, 1982 A computer program was developed to compute the profile of multiple-beam Fizeau fringes and was tested against laboratory observations. By using the program as a tool for studying the behavior of the fringes, a simplified model of the fringe formation process was developed. The shifts of the fringes from their expected locations are caused primarily by a reduction in spatial frequency that is due to the oblique projection of the fringes onto the detector. The angle at which the fringes are inclined to the input beam of the interferometer may be calculated by using a phase law developed by Brossel [Proc. Phys. Soc. 59,224 (1947)] and is independent of the reflectivity of the cavity. Once the orientations of the fringe planes and the detector are known, the fringe shifts may be calculated by using simple trigonometry. Although the absolute shift of a particular fringe may be an appreciable fraction of a fringe spacing, neighboring fringes have nearly identical shifts, making the relative fringe shift inconsequential. The finesse of the fringes may be increased by either tilting the Fizeau cavity or defocusing the detector optics. Unlike the tilt angle required to minimize the fringe shift, the tilt angle that maximizes the finesse is a strong function of the cavity reflectivity.

It has long been known that multiple-beam Fizeau fringes do not exhibit the symmetry and degree of finesse characteristic of Fabry-Perot fringes because of the tendency of the higher-order beams to walk in the direction of the wide end of the wedge. An expression governing the relative phases of the beams was developed by Brossel,1 but computing the fringe profile still required the summation of a great number of beams. Kinosita,2 in 1952, laboriously performed this sum, obtaining a fringe profile for one particular wedge arrangement. More recently, Hall3 constructed a computer model for performing the summation. It is the modern high-speed computer that permits the shape and location of a multiple-beam fringe to be computed and the dependence of these on the various parameters of the system to be studied. In this paper, a model for computing the profile of a transmission Fizeau fringe is described, and its output is verified quantitatively against experimental results. The computer model is then applied as a tool for studying the shifts of the fringes from the traditional expectation that they be located at the positions where the Fizeau wedge is an integral number of half-wavelengths thick. A simplified model of the fringe formation process is developed, and it is shown to predict accurately the locations of the fringes. The concept of fringe sharpening, introduced by Langenbeck,4 is also studied.

where T and R represent the intensity transmission and reflection coefficients, respectively. The error involved in truncating the infinite summation after some finite number of beams may be estimated by making the pessimistic assumption that all the terms left out of the sum are exactly in phase with one another, thereby ensuring that the estimated error exceeds the actual error. The error caused by truncating the series after n terms then becomes an infinite geometric series with the summation index beginning at n + 1. This series may be expressed as the difference between an infinite series beginning at one and a finite series beginning at one and ending at n. By using the analytic solutions for these last two sums, the intensity error may be shown to be

For simplicity, the assumption is made that there is no absorption (a good approximation, if dielectric mirrors are used), 2 so the factor of T2/(1 - R) becomes unity. The above expression may be inverted to produce the number of beams required for a given accuracy in the sum

In this expression, it is convenient to use base-10 logarithms, so that the effect of changing the allowed intensity error AI by an order of magnitude can readily be seen. With the modulus of each beam given by Eq. (l), and the number of beams required given by Eq. (2), only the relative phases of the beams remain to be computed. This will be done according to the method described by Brossel. As shown in Fig. 2, an input wave front drawn through the apex angle of a wedged pair of mirrors will be imaged into a family of wave fronts fanned out about the apex, the angle between any two wave-front images-being twice the angle of the wedge. This is true regardless of the orientation of the wedge relative to

The intensity profile of the fringe is formed point by point in the image plane by adding the complex amplitudes of the beams as they emerge from the wedge, as shown in Fig. 1. The jth-order beam, having experienced two transmissions and 2j reflections, has a magnitude (relative to an input beam of unit amplitude) given by


John R. Rogers

Vol. 72, No. 5/May 1982/J. Opt. Soc. Am.



R=.984 Fig. 1. Interference of three beams emerging from a Fizeau wedge.

4 Fringes/Inch

Fig. 3.

Computed and observed fringe profiles; normal incidence.

Tilt =


Fig. 2. Fan of wave-front images produced by a wedge.

the input wave front. Because these wave fronts are all images of one another, they have the same relative phase, and so their phases at the detector plane may be computed simply by propagating them along their normals to the detector. Simple trigonometry shows that the path length from the jth wave front to the detector at position (z, y) is given by

where z, y, and are as seen in Fig. 2. The phase of the jth beam is computed directly from the path length and makes use of the fact that the beam angle increases by each time one beam number j is increased. Considerable computational savings are gained by computing the sine and cosine of once and then computing the values for the higher-order angles recursively.
Fig. 4. mrad. Computed and observed fringe profiles, wedge tilted -0.7

EXPERIMENTAL VERIFICATION The ability of the program to predict fringe profiles was checked by setting up a Fizeau interferometer and imaging


Shack Cube Interferometer

Only 1 good surface required. Single longitudinal mode laser required. Accessory optics required for testing lenses or convex surfaces. Thin absorber, such as a pellicle, is required for testing coated surfaces. Lower cost, but not as flexible as Twyman Green or regular laser based Fizeau interferometer.

Shack Interferometer

Ref: R. V. Shack and G. W. Hopkins, Opt. Eng. 18, p. 226, March-April 1979
1998 - James C. Wyant

Shack Interferometer

1998 - James C. Wyant

8.2.6 Scatterplate Interferometer The scatterplate interferometer illustrated in Fig. 8.2.6-1 for the testing of a spherical mirror gives fringes of constant contour just as does the LUPI; however, its operation does not depend on the knowledge of the quality of auxiliary optics. Due to the common path feature of the interferometer, the light source can be either a laser, or more commonly, a white light source such as a zirconium arc or tungsten bulb with a Wratten spectral filter. The light source is focused onto a pinhole, which is then reimaged onto the surface under test. A scatterplate a few millimeters in diameter is placed at the center of curvature of the mirror under test. Part of the light illuminating the scatterplate will pass unscattered to focus directly on the mirror surface, while part of the light will be scattered uniformly over the mirror surface. The mirror will reimage the scatterplate back on itself--inverted, of course.



Scatterplate (near center of curvature of mirror being tested) Mirror being Tested

Fig. 8.2.6-1 Scatterplate interferometer for testing concave mirror. Fig. 8.2.6-2 shows a photo of a scatterplate interferometer where the light source is a helium neon laser.

Fig. 8.2.6-2 Scatterplate Interferometer

The light transmitted through the scatterplate the second time can be divided into four categories dependent upon the influence of the scatterplate on the transmitted light: (1) unscattered-unscattered, (2) unscattered-scattered, (3)scattered-unscattered, and (4) scattered-scattered. The unscattered-unscattered light will produce a bright or hot spot in the interferogram, while the scattered-scattered beam will reduce the fringe contrast. If a laser source is used the scattered-scattered beam will give a high-frequency speckle pattern across the interferogram. The interference of the unscattered-scattered and scattered-unscattered beams gives fringes of constant contour that are just the same as those produced by a LUPI. Tilt in the interference fringes is introduced by lateral translation of the scatterplate, while longitudinal translation controls the amount of defocus. Fig. 8.2.6-3 shows 3 typical interferograms obtained using a scatterplate interferometer. The scatterplate interferometer was moved slightly between recording the three interferograms to change the amount of tilt and defocus.

Fig. 8.2.6-3 Scatterplate interferograms of parabolic mirror. The scatterplate interferometer is a very simple device, having a minimum of quality components. The most critical part of the instrument is the scatterplate itself, which can easily be made. The basic procedure for making a scatterplate is to expose a photographic plate to a speckle pattern produced by illuminating a piece of ground glass with a laser beam. Since the scatterplate must have inversion symmetry, two superimposed exposures to the speckle pattern must be made, where the plate is rotated 180o between the exposures. To ensure that the scatterplate illuminates the surface under test as uniformly as possible, during the making of the scatterplate the solid angle subtended by the illuminated piece of ground glass, as viewed from the photographic plate, should be at least as large as the solid angle of the surface under test, as viewed from the scatterplate during the test. After development, the photographic plate should be bleached to yield-a phase scatterplate. The exposure, development, and bleaching should be controlled so that the scatterplate scatters 10 to 20% of the incident light. Fig. 8.2.6-4 shows a high magnification photograph of a scatterplate. This scatterplate was made for operation at a wavelength of 10.6 microns so the detail making up the scatterplate was large enough to easily observe through a microscope. Note the inversion symmetry of the structure making up the scatterplate.

Lateral Displacement Introduces Tilt

Lateral Displacement


• •


1998 - James C. Wyant

Longitudinal Displacement Introduces Defocus

Center of curvature • A

• Á Image of A

1998 - James C. Wyant

Fig. 8.2.6-4 High magnification photograph of a scatterplate.

The advantages of the scatterplate interferometer are that the instrument is simple and inexpensive, requires no accessory optics, and because the interferometer has a common path, it is less sensitive to vibration and turbulence. If an incoherent source is used as the light source, the coherent noise (extraneous fringes), normally associated with using a laser as the light source, is absent. The disadvantages are that the hot spot can cause a loss of fringes for a small portion of the interferogram and the twice-scattered beam will cause the interferogram to have somewhat lower contrast than can be obtained using a LUPI. References J.M. Burch, Nature (London) 171, 889 (1953). J.M. Burch, “Interferometry in Scattered Light,” in Optical Instruments and Techniques, J.H. Dickson, Ed. (Oriel, London, 1970), p. 220. J. M. Burch, “Scatter Plate Interferometry,” J. Opt. Soc. Am. 52, 600 (1962). (Abstract only).

R. M. Scott, “Scatter Plate Interferometry,” Appl. Opt. 8, 8531 (1969). John Strong, “Concepts of Classical Optics”, Appendix B (J. Dyson), (W.H. Freeman, San Francisco 1958), p. 377. L. Rubin and J.C. Wyant, “Energy distribution in a scatterplate interferometer,” J. Opt. Soc. Am. 69, 1305 (1979). L. Rubin and O. Kwon, “Infrared scatterplate interferometry,” Appl. Opt. 19, 3219 (1980). J. Huang, T. Honda, N. Ohyama, J. Tsujiuchi, “Fringe scanning scatter plate interferometer using polarized light,” Optics Comm. 68, 235 (1988). D. Su and L. Shyu, “Phase shifting scatter plate interferometer using a polarization technique,” J. Mod. Opt. 38, 951 (1991).

Scatterplate Interferograms

647.1 nm

520.8 nm

476.2 nm

All λ’s except red

Absorptive scatterplate

568.2 nm, rotating ground glass

520.8 nm, rotating ground glass

476.2 nm, rotating ground glass

1998 - James C. Wyant

Scatterplate Interferometer
J. M. Burch, Nature 171, p. 889, 1953. A.H. Shoemaker and M.V.R.K. Murty, Appl. Opt., 5, p. 603, 1966. Source Mirror Under Test


Focusing Lens

(Near center of curvature of mirror being tested)

Imaging Lens


Phase-Shifting Scatterplate Interferometer


(near center of curvature of mirror being tested)

Mirror being Tested

1998 - James C. Wyant

Phase-Shifting Scatterplate Interferometer
• Direct-scattered beam passes through a 1/4 wave-plate • Phase shift produced by applying voltage to E/O Cell • Must correct for phase error produced by 1/4 wave-plate
Ref: Huang, J., Honda, T., Ohyama, N., and Tsujiuchi, J., 1988, Opt. Commun., 68, 235.

1998 - James C. Wyant

Phase-Shifting Scatterplate Interferometer
• Scattered-direct beam passes through a polarizer • 1/2 wave-plate is rotated to produce phase shift • Must correct for phase error produced by the polarizer
Ref: Su, Der-Chin and Shyu, Lih-Horng, 1991, J. Modern Optics., 38, 5, 951.

1998 - James C. Wyant

Phase-Shifting Scatterplate Interferometer Common Path Interferometers
• Advantages
– Less Sensitive to Vibration – Less Sensitive to Air Turbulence

• Challenge
– Difficult to Separate Test and Reference Beams for Phase-Shifting

Scatterplate Interferometer
Four Terms:
Scattered-Scattered Scattered-Direct Direct-Scattered Direct-Direct


High magnification image . of a scatterplate

Scatterplate Interferogram
• • • (Near center of curvature of mirror being tested)

Mirror Being Tested

Scatterplate has inversion symmetry Scatterplate is located at center of curvature of test mirror Direct-scattered and scattered-direct beams produce fringes

Separating Test and Reference Beams
• Cannot Separate Spatially
– Both Test and Reference Traverse the Same Path

• If Test and Reference Have Orthogonal Polarization then Phase-Shifting is Possible

Birefringent Scatterplate Makes Phase Shifting Possible
• Scatterplate is Made of Calcite • Oil Matches Ordinary Index of Calcite • Scattering is Polarization Dependent
Index Matching Oil Calcite

Glass Slide

Scatterplate Production
Generated using six step process
1. Clean substrate 2. Spin coat with Photoresist 3. Expose » Holographic » Photomask 4. Develop 5. Etch with 37% HCl Diluted 5000/1 in DI 6. Remove Photoresist
1 2



5 6

Photomask Exposure
• Desired pattern is designed in computer • Pattern is written into chrome on a quartz substrate using an electron beam • UV source illuminates photoresist only where chrome has been removed

Incident UV Radiation



} Sample

Phase Shifting Scatterplate Interferometer
Source Rotating Ground Glass Liquid Crystal Retarder (0°)

Polarizer (45°) λ/4 Plate (45°)

Interferogram Analyzer (45°)

Scatterplate (0°)

Mirror Being Tested

Phase-Shifting Scatterplate Interferometer

Phase-Shifted Fringe Pattern

Frame 1

Frame 2

Frame 3

Frame 4

Surface Measurement Using Phase-Shifting Scatterplate Interferometer

Measurement Comparison
Scatterplate WYKO 6000

RMS = 0.008738 Waves PV = 0.03750 Waves

RMS = 0.008738 Waves PV = 0.03405 Waves

Summary Phase-Shifting Scatterplate

• Birefringent scatterplate makes phaseshifting possible • For this example
– Accuracy ≈ 0.035 waves peak-to-valley – Repeatability ≈ 0.003 waves RMS

• Low sensitivity to vibration