Low Coherence Vibration Insensitive Fizeau Interferometer

Brad Kimbrough, James Millerd, James Wyant, John Hayes

4D Technology Corporation, 3280 E. Hemisphere Loop, Suite 146, Tucson, AZ 85706
(520) 294-5600, (520) 294-5601 Fax, [email protected]

An on-axis, vibration insensitive, polarization Fizeau interferometer is

realized through the use of a novel pixelated mask spatial carrier phase
shifting technique in conjunction with a low coherence source and a
polarization delay-line. In this arrangement, coherence is used to
effectively separate out the orthogonally polarized test and reference beam
components for interference. With both the test and the reference beams
on-axis, the common path cancellation advantages of the Fizeau
interferometer are maintained. The interferometer has the unique ability
to isolate and measure any surface that is substantially normal to the
optical axis of the cavity. Additionally, stray light interference is
substantially reduced due to the source’s short coherence. An expression
for the fringe visibility on-axis is derived and compared with that of a
standard Fizeau. Using a 15 mW source, the maximum camera shutter
speed, used when measuring a 4% reflector, was 150 usec, resulting in
very robust vibration insensitivity. We experimentally demonstrate the
measurement of both sides of a thin glass plate without the need to modify
the plate between measurements. Experimental results show the
performance of this new interferometer to be within the specifications of
commercial phase shifting interferometers.

Introduction

Determination of the spatial variations of optical phase is of primary importance in the

fields of optical testing and metrology, optical information processing, and adaptive
optics. There exist several techniques for the encoding of spatial phase modulation in
fringe patterns and the subsequent analysis of these fringe patterns for a quantitative
determination of phase.1 The two primary groups of fringe pattern analysis techniques
are: Temporal phase measurement, sometimes known as the phase shifting method, and
spatial phase measurement which is also called the spatial carrier method.

Temporal phase measurement is a well-established method for measuring optical

wavefront phase 2. In this technique, three or more interferogram intensity profiles are
recorded. For each recording, there is a different relative phase between the test and
reference beams. The phase distribution of the test wavefront is then calculated using the
recorded interferogram intensities. DeGroot, et. al demonstrated the measurement of
multiple surfaces within a cavity using wavelength tuning and temporal phase-shifting 3.

Interferometry XIII: Techniques and Analysis, edited by Katherine Creath, Joanna Schmit,
Proc. of SPIE Vol. 6292, 62920F, (2006) · 0277-786X/06/$15 · doi: 10.1117/12.682956

Proc. of SPIE Vol. 6292 62920F-1

Spatial phase measurement utilizes a single interferogram to extract phase information.4
In this technique, a spatial carrier, typically in the form of high frequency tilt fringes, is
applied to the interferogram. The intensity profile of the modulated spatial carrier
interferogram is recorded and then analyzed to determine the phase. The primary
advantage of the spatial phase measurement technique over temporal phase measurement
is that only one image is required, allowing acquisition times several orders of magnitude
smaller than in temporal phase shifting. Rapid acquisition offers both significant
vibration immunity, and the ability to measure dynamic events.

A novel approach to spatial phase measurement has been developed by 4D Technology

Corporation called the pixelated mask spatial carrier method.5 In this technique, the
relative phase between carrier and test wavefront is modified on a pixel-by-pixel basis by
a micro-polarizer phase shifting array placed just prior to detection. This technique
requires that the test and reference beams be orthogonally polarized. This is easily
implemented in a Twyman-Green interferometer that uses a polarizing beam splitter to
separate the source wavefront into orthogonally polarized test and reference beams. In
the Fizeau interferometer, the test and reference beams are not separated as in the
Twyman-Green, but travel along a common-path up to the reference surface. Due to this
almost common path arrangement, the configuration of the Fizeau interferometer is
substantially simpler and only the reference surface must be produced with a high optical
precision. The main drawback with this arrangement is that the overlapping test and
reference beams make it difficult to implement a polarization Fizeau interferometer.

One solution to this problem has been implemented by 4D Technologies and is called the
tilted beam Fizeau. 6 In this arrangement, a relatively large tilt angle is placed on the
reference flat giving a large angular offset between the test and reference beams. Prior to
incidence on the pixelated mask sensor, the two beams pass through a Wollaston prism,
producing two pairs of orthogonally polarized test and reference beams. Only one of
these beam pairs is collinear with each other and with the optical axis. The major
drawback with this technique is that with the tilted reference surface, the returning test
and reference beams no longer follow a common path, resulting in retrace errors in the
measurement. For plano reference and test optics the interferometer can be designed to
have very low retrace errors; however, the use of transmission spheres can introduce
substantial errors and calibration techniques must be used.

Kuchel proposed the use of a low temporal coherence source and an optical delay-line
that splits the source beam into two orthogonally polarized components and allows a
controllable phase difference between the two beams.7 In this paper we report on
combining the delay-line Fizeau configuration with 4DTechnology’s pixelated-mask
instantaneous phase sensor. The benefits of the resulting instrument are three fold. First,
the systems permits high spatial resolution instantaneous phase measurement in a single
camera frame. Secondly, the two orthogonally polarized beams are collinear – allowing
true on-axis operation so that transmission flats and spheres can be used without
calibration. Finally, the delay line allows matching of the reference and test beam optical
paths, facilitating the measurement of specific surfaces within an optical cavity. For

Proc. of SPIE Vol. 6292 62920F-2

example, the measurement of transparent surfaces such as thin lenses or plates can be
made by suppressing the interference due to reflection from the back surface.8, 9
Additionally, the use of very low temporal coherence sources reduces coherent system
noise due to spurious reflections. We present the theory of operation of the instrument
and results obtained measuring both sides of an uncoated glass plate using only two
measurements and without the need to modify the plate.

Theory of Operation

The basic layout of the short coherence vibration insensitive Fizeau is shown in figure 1.
The illumination system is composed of a short coherence laser source and an optical
delaying device that splits the source beam into two orthogonally polarized components
and applies a controllable optical path difference between the two beams. This output is
used as the illumination source for a standard Fizeau interferometer arrangement. The
output of the Fizeau is imaged via an afocal imaging system onto the pixelated mask
sensor.

The optical delaying device is essentially a polarization Twyman-Green interferometer.

It is convenient to think of the illumination as being composed of two separate but
overlapping beams, the s-polarized beam and the p-polarized beam. The half wave plate
just in front of the laser source is used to adjust the intensity ratio between the two beams.
As will be shown in the next section, maximum visibility is obtained when the intensities
of the s and p components are matched, regardless of the test and reference mirror
reflectivities. Upon exiting the optical delaying device, there is an optical path delay
between the s-polarized and the p-polarized beams. Mirror M1 is mounted to a stage and
allows the optical path length of the exiting p-polarized beam to be adjusted relative to
that of the s-polarized beam which originates from M2.

Collimator
Short
coherence
source Beam — —
AL Splitter

M2
Test
Zero Path Delay Reference

Pixelated Mask —+
Camera

Figure 1: Short coherence vibration insensitive Fizeau functional diagram.

Proc. of SPIE Vol. 6292 62920F-3

Both the s and p-polarized beams illuminate and are reflected by the test and reference
surfaces. The returning beams undergo reflection at the non polarizing beam splitter and
are imaged via an afocal imaging system onto the pixelated mask sensor. The quarter
wave plate, QWP, just prior to the pixelated mask is used to convert the linearly polarized
test and reference beams into right and left circular polarizations as is necessary for the
operation of the pixelated mask sensor.

Ignoring multiple Fizeau cavity reflections for the moment, there are essentially 4 beams
incident upon the pixelated mask sensor; s and p-polarized test surface reflected beams,
and s and p-polarized reference surface reflected beams. For the pixelated mask sensor
we would like interference to occur between either the s-polarized reference beam and
the p-polarized test beam, or between the p-polarized reference beam and the s-polarized
test beam. Since the source has a very short coherence length, approximately 250 um,
only those beams whose optical paths, from the source to the camera, have been matched
will interfere. No interference will occur between the other beam pair and there
intensities will simply add to the background.

Visibility

Figure 2 shows a Fizeau cavity with flat and parallel test and reference surfaces whose
reflectivities are Rt and Rr respectively. The cavity is illuminated with both s (Is - top)
and p (Ip - bottom) polarized light. Both beams are collinear, but have been separated for
clarification. The s-polarized beam is delayed such that its reflection off of the reference
surface is path matched to the p-polarized beams first reflection off of the test surface.
Both of these beams are shown as solid red lines pointing to the right.

Test Ref Is
Surface Surface
I s Rr
1 Is (Tr )2 Rt
S 2 Is (Tr )2 Rt (Rr Rt)
3
Is (Tr )2 Rt (Rr Rt)2
Rt Rr
Ip
I p Rr
1 Ip (Tr )2 Rt
P
2 Ip (Tr )2 Rt (Rr Rt)
3
Ip (Tr )2 Rt (Rr Rt)2

Figure 2: Multiple test surface reflections in the low coherence Fizeau cavity

Proc. of SPIE Vol. 6292 62920F-4

The bold numbers within the cavity indicates the number of test surface reflections each
beam has undergone prior to exiting the cavity. Path matched, coherent, beams are of
similar color and type, (e.g. red-solid, blue-dashed). The formula to the right of the
beams indicates their intensities. Note that there are more than one set of interfering
beams and that each beam is coherent with only one other beam. Each coherent pair will
interfere, producing a fringe pattern. The primary interference pattern is created by the
interference of the s-polarized beam from the reference surface and the p-polarized beam
that has reflected once from the test surface, solid red lines. The next coherent beam pair
is composed of the s-polarized beam that has reflected once from the test surface, and the
p-polarized beam that has reflected twice from the test surface, blue dashed lines. Each
successive reflection of the s-polarized beam off of the test surface is coherent with the
p-polarized beam that has undergone one additional test surface reflection. The
interferogram detected at the output of the interferometer is composed of the sum of the
primary fringe pattern with the fringe patterns of the other coherent beam pairs. This
fringe pattern can be represented mathematically as:

I = I p r + Is r + I p t[1] + 2 I pt[1] Isr Cos[θ1 ]

+ Is t[1] + I pt[2] + 2 Ist[1] I pt[2] Cos[θ 2 ] +...............

(1)
+ Is t[n-1] + I pt[n] + 2 Ist[n-1] I pt[n] Cos[θ n ] + Is t[n]

where I pr and Isr are the intensities of the p and s- polarized beams reflected off of the
reference surface, I p t[n] and Is t[n] are the intensities of the p and s-polarized beams that
have undergone n reflections off of the test surface, and θn is the phase difference
between the nth pair of interfering beams.

Writing equation 1 in terms of the reference and test surface reflectivity, Rr and Rt, and
separating the constant terms from the interference terms gives:

I = (Is + I p ) (R r + (1-R r )2 R t [1+(R r R t )+.... + (R r R t ) n-1 ])

+ 2 (1-R r ) R t R r Is I p Cos[θ1 ] (2)

n≥2
+2 (1-R r ) 2 R t R t R r Is I p ( Cos[θ ] + (R R )Cos[θ ] + ..... + (R R )
2 t r 3 t r
n− 2
Cos[θ n ])

Simplifying the geometric series in the first term of equation 2 gives:

Proc. of SPIE Vol. 6292 62920F-5

 ⎛ 1-(R r R t ) n ⎞
I = (Is + I p ) ⎜ R r + Tr 2 R t ⎟ + 2 (1-R r ) Is Ip R t R r Cos[θ1 ]
⎝ 1-(R r R t ) ⎠
n≥2 (3)
+2 (1-R r ) 2 R t Is I p R t R r ( Cos[θ ] + (R R )Cos[θ ] + ..... + (R R )
2 t r 3 t r
n− 2
Cos[θ n ])

In order to evaluate equation 3 further, the phases of the interference terms must be
determined. These phase terms are a function of the test surface shape, and the
separation of the test and reference surfaces. In the simplest case, the cavity is nulled and
the test and reference surfaces are close to parallel at every point. In this case it can be
shown that the phase difference between all secondary interfering beam pairs, θn, is equal
to -θ1. The negative sign resulting from the fact that for the secondary interfering beam
pairs, all reflections occur at an air to glass interface, whereas for the primary interfering
beam pair, the reference beam results from a glass to air interface reflection off of the
reference surface. Setting θn = - θ1, and taking the limit as n becomes very large,
equation 3 reduces to:

⎛ (1-R r ) 2 R t ⎞ ⎛ (1-R t ) ⎞
I = (Is + I p ) ⎜ R r + ⎟ + 2 (1-R r ) Is Ip R t R r ⎜ ⎟ Cos[θ1 ] (4)
⎝ 1-R r R t ⎠ ⎝ 1− RtRr ⎠

From equation 4, the fringe visibility is determined to be:

⎛ Is I p ⎞ (1-R ) (1-R ) R R
Vis = ⎜ ⎟ r t t r
(5)
⎜ (Is + I p ) ⎟ (1-R r )R t + (1-R t ) R r
⎝ ⎠

A plot of fringe visibility as a function of test surface reflectivity for three different
values of reference surface reflectivity is shown in figure 3. Visibility curves for a
standard Fizeau arrangement are also shown for comparison.

Proc. of SPIE Vol. 6292 62920F-6

 J_ U U
Reference Surface
Reflectivity
80 / • 4% Path Matched Fizeau
E15%
60 •25% Standard Fizeau
—U—U—
40

20

20 40 60 80 100
Test Surface Reflectivity (%)

Figure 3: Visibility versus test surface reflectivity for the low coherence Fizeau.

For these plots Is = Ip. Unlike a standard polarization interferometer, matching the test
and reference beam intensities at the interferometer output will not maximize fringe
contrast. Beam balance affects the first term in equation 5, and this term maximized
when the illumination beam intensity ratio is equal to one. Fringe visibility is maximized
when test and reference surface reflectivities are nearly matched. As can be seen from
the plot, the maximum fringe visibility is obtained at low reference surface reflectivities.
As the reference surface reflectivity is increased, the maximum fringe visibility is
reduced. Additionally, like the standard Fizeau, multiple cavity reflections results in a
loss of fringe contrast for high test surface reflectivities.

Optical Configuration

A diagram of the low coherence Fizeau prototype that was built is shown in figure 4. The
prototype is composed of three major sections: (1) A custom designed low coherence
source, including spatial filtering and collimation, (2) polarization path matching, and (3)
Fizeau cavity and imaging system. The source filtering and collimation section consists
of the laser diode, a spatial filter, and a collimating lens. The laser diode output is 15
mW at 658 nm. The coherence length of this source is approximately 250 um FWHM.

Proc. of SPIE Vol. 6292 62920F-7

 Polarization Path Matching Low coherence collimated Source
M2
Spatial
Filter
Stage

HWP
QWP
L0
M1 LD
PBS

L1 L2

BS CCD
Stop
QWP Pixelated
Mask
Fizeau
Interferometer

Reference
Test

Figure 4: Low coherence Fizeau prototype.

The polarization path matching section is essentially a polarization Twyman-Green

interferometer and consists of a half wave plate, HWP, a polarizing beam splitter, PBS,
two quarter wave plates, QWP, a fixed mirror, M2, and a mirror mounted to a linear
stage, M1. The two proceeding components constitute the source for the interferometer.
The output of the path matching section is a collimated beam with a diameter of
approximately 14 mm. This beam first passes through a non-polarizing beam splitter,
BS, and then a diverging lens and a collimation lens, producing a collimated beam with a
diameter of approximately 100 mm. This beam is then reflected off of the reference and
test surfaces and returns to the beam splitter. A portion of this beam is then reflected
towards the afocal imaging system composed of L1, L2, and an aperture stop. L1 and L2
have focal lengths of 100 and 200 mm respectively, giving a magnification of 2. Prior to
the pixelated mask and CCD, the input beam passes through a QWP changing the linear
s-polarized and p-polarized components into right and left circular polarizations.

Transparent Glass Disk Measurements

To demonstrate the utility of the interferometer we made measurements on thin glass

plates, a test that is virtually impossible with conventional Fizeau interferometers because
of multiple reflections. Figure 5 shows three possible measurements that can be made on
a transparent glass plate without turning the test piece around. Designate one side of the
plate as side A, and the other as side B. The diagram on the left side of the figure shows

Proc. of SPIE Vol. 6292 62920F-8

the transparent glass plate with a side A height profile, ha, a nominal thickness of to, and a
side B surface height profile of hb. For measurement (1), side A is facing the reference
surface and has been path matched. The measured height profile, H, is equal to the defect
height profile, ha. In measurement (2), side A is facing the reference surface, but now
side B has been path matched. The measured height, H, is now given by:
H = h a -n (h a +h b ) , where n is the glass index of refraction and is assumed to be constant.

to
B A

(1) H = ha

hb ha
Test Ref

B A

(2) H = ha – n (ha+hb)

B A Test Ref

B A
Transparent Glass Disk

(3) H = n (ha+hb)
n = glass index
to = nominal thickness
ha = side A profile
Ref Test
hb = side B profile
H = measurement results

Figure 5: Transparent Glass Plate Test Arrangements. In test (1) side A is path matched
with the reference surface. In test (2) side B is path matched with the reference surface. In
test (3) side A is path matched with side B.

For measurement (3), side A has been path matched to side B and the surface
measurement result is given by: H = n (h a +h b ) , where once again the glass index is
assumed constant. The transmission flat (external reference surface) can be removed for
this measurement.

Experimental Results

Measurements (1)-(3) as outlined above were conducted on a four inch diameter glass
disk having a nominal thickness, to, of 1 mm. All measurements were carried out on a
rigid optical table and no steps were taken to control environmental vibrations. Figure 6
provides the results of measurements (1) and (2) on the left and right sides respectively.
Note that measurement (2) does not directly provide the surface height profile of side B.

Proc. of SPIE Vol. 6292 62920F-9

Letting H1 and H2 be equal to the results of measurements (1) and (2) respectively, the
surface height profile of side B, hb can be calculated as follows:

1
hb = ( H1 − H 2 ) − H1 (6)
n

(1) Side A (2) Side B (through Glass)

53
2o
till,

0.023pnd

PV = 6.93 waves PV = 6.051 waves

= 4.86 um = 3.982 um

H = ha H = h a -n (h a +h b )
Figure 6: Transparent glass disk measurement results for test arrangements (1) and (2).

Using an index of refraction equal to 1.5, the surface height profile of side B was
calculated from the results of measurements (1) and (2). Figure 7 shows this result along
with a direct measurement of the side B profile obtained by turning the test piece around
and path matching side B to the reference surface. Both the direct and calculated
measurements are in close agreement, indicating that both surface profiles, ha and hb, can
be determined without turning the disk around.

Proc. of SPIE Vol. 6292 62920F-10

 Direct Measurement Calculated
Using Measurements (1) and (2)

Figure 7: Comparison of a direct measurement of the side B surface profile with the
surface profile calculated from the results of measurements (1) and (2).

Figure 8 shows the results of measurement (3). This measurement was conducted by
path matching side A to side B. This measurement can be interpreted as the optical
thickness deviation about the nominal disk thickness. Since the reference flat is not used
in this measurement, it may be removed from the system. Because beams reflected off of
the reference flat add to the non-coherent background intensity, its removal will result in
an increase in the fringe contrast.

PV = 1.3 waves
= 0.945 um

H = n (h a +h b )

Figure 8: Transparent glass disk measurement results for test arrangement (3).

Conclusions

An on-axis, vibration insensitive, polarization Fizeau interferometer has been realized

through the use of a novel pixelated mask spatial carrier phase shifting technique in

Proc. of SPIE Vol. 6292 62920F-11

conjunction with a special low coherence source and a polarization path matching
configuration. An expression was derived for the theoretical fringe visibility for on-axis
measurement, and the results compared with that of a standard Fizeau interferometer.
The visibility for the low coherence Fizeau was found to be near the theoretical
maximum for low reference surface reflectivities. Visibility is maximized regardless of
test and reference surface reflectivities when input beam balance ratio is equal to 1.
Similar to a standard Fizeau, visibility is reduced for higher test surface reflectivities due
to multiple cavity reflections. Measurements were made on both sides of an uncoated
glass disk without any special treatment or movement and with only two measurements
were required. In addition, the thickness of the glass was measured. Using a 15 mW
source, the maximum camera exposure time was 150 usec, resulting in very robust
vibration insensitivity. The novel arrangement also permits the use of transmission
spheres without the need for calibration.

References

1. D. Malacara, M. Servin, and Z. Malacara, “Interferogram analysis for optical testing,”

(Marcel Dekker, New York, 1998).
2. D. Malacara, Optical Shop Testing (Wiley, New York, 1992).
3. P. de Groot, Measurement of Transparent Plates with Wavelength-Tuned Phase-Shifting
Interferometry, Applied Optics 39 (16), 2658-2663, June 2000.
3. R. Jozwicki, M. Kujawinska, and L. Salbut, "New contra old wavefront measurement
concepts for interferometric optical testing," Optical Engineering 31, 422 , (1992).
4. J. E. Millerd, N. J. Brock, J. B. Hayes, M. B. North-Morris, M. Novak, and J. C. Wyant,
"Pixelated phase-mask dynamic interferometer," Proc. SPIE 5531, 304-314, (2004).
5. J. E. Millerd in “Fringe 2005,” edited by W. Osten, (Springer, New York, 2005), pg 640.
6. M. Küchel, "Interferometer for measuring optical phase difference," U. S. Patent 4,872,755
(1989).
7. E. Novak, C. Ai, and J. C. Wyant, "Errors caused by nearly parallel optical elements in a laser
Fizeau interferometer utilizing strictly coherent imaging," Proc. SPIE 3134, 456-460 (1997).
8. J. Shiefman, "Using software to model coherent metrology systems," Proc. SPIE 5174, 69-82
(2003).

Proc. of SPIE Vol. 6292 62920F-12