Thin Solid Films 462 – 463 (2004) 257 – 262

www.elsevier.com/locate/tsf

C–DIC: a new microscopy method for rational study of

phase structures in incident light arrangement
Rainer Danz *, Peter Gretscher
Carl Zeiss, Goettingen, Germany

Available online 24 July 2004

Abstract

Circular polarized light – differential interference contrast (C – DIC) is a new polarization – optical differential interference contrast method
where, unlike conventional DIC according to Nomarski [Interféromètre à polarisation, French Patent 1.059.123, 1952.], the DIC prism (DP)
is arranged in circular, not linear, polarized light. Consequently, the interference contrast generated is invariant in relation to the oscillation
orientation of the DIC prism, and so the latter can be rotated directionally in accordance with the characteristics of the object. This means that
the stage does not need to be rotated while the relationship with the object is preserved. For the user, this means more information and an
increase in sample throughput.
D 2004 Elsevier B.V. All rights reserved.

Keywords: Differential interference contrast; Circular polarized light; Azimuth effect

1. Introduction: the basic principles of the DIC method where / represents the relative object phase difference,
x Vthe location coordinates in the direction of the image
Differential interference contrast (DIC) according to split, s the size of the image split (‘‘shear‘‘) and w the phase
Nomarski [1] is currently the most popular interference difference between the object wave and the laterally dis-
microscopy method for high-contrast depiction of geometric placed reference wave. The phase difference w can be
and physical phase objects (differences in height/difference altered continuously by moving the Wollaston prism per-
in phase shifts of adjacent structures; Fig. 1). pendicular to the optical axis.
The Nomarski method uses a modified Wollaston prism By multiplying Eq. (1) by the complex conjugate func-
(Fig. 2), which affects an angular split of the incident tion, the intensity distribution in the intermediate image
linear polarized light beams. This means that completely plane is obtained, in accordance with I = uu* [3]
identical twin images of every object structure are pro-
duced, as in Figs. 3 and 4, which, laterally offset by a IðxVÞ ¼ 1 þ cos½/ðxVÞ  /ðxV sÞ  W ð2aÞ
small (‘‘differential’’) amount sb object field, interfere If w = p/2 and small /bp, Eq. (2a) can be simplified as
with one another. follows without loss of generality:
The standardized light generation u(x V) (or rather the
standardized amplitude of the electrical or magnetic field IðxVÞ ¼ 1 þ /ðxVÞ  /ðxV sÞ;
vector) in the intermediate image plane is described by or
d/ðxVÞ
uðxVÞ ¼ ei/ðxVÞ þ ei½Wþ/ðxVsÞ ð1Þ IðxVÞ ¼ 1 þ s ð2bÞ
dxV
In accordance with Eq. (2b), the intensity in the
intermediate image plane is proportional to the first deriv-
* Corresponding author. Materials Microscopy, ative of /(x V) with respect to x V, i.e., to the change in the
Goettingen 37081, Germany. Tel.: +49-551-5060-368;
fax: +49-551-5060-464.
relative phase difference of the image and twin image in
E-mail address: danz@zeiss.de (R. Danz). the direction of the image split (Figs. 3b and 4b). This
URL: http://www.zeiss.de. creates the impression of the image in relief that is

0040-6090/$ - see front matter D 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.tsf.2004.05.124
258 R. Danz, P. Gretscher / Thin Solid Films 462 – 463 (2004) 257–262

Fig. 1. Geometric (left) and physical (right) incident light phase object: a geometric phase object superimposes the phase difference U on the reflected light
wave front. This phase difference is only dependent on the step height d. If the phase object demonstrates no differences in height but different materials, the
light wave front is deformed by the difference in the material-specific phase shifts c1 and c2.

characteristic of the DIC method (Figs. 3a and 4a). It is perpendicular to the direction of the split x V(orientation
also evident that the relief contrast is not rotationally effect). When the direction of the split is determined,
symmetric, for example, in phase contrast or dark field. straight structures are therefore selected that are preferen-
Maximum contrast is only to be expected in a preferential tially contrasted. Line structures that lie in the direction of
direction, in other words, at phase boundaries that lie the split (Fig. 5, first and last images of the bottom row)
are not contrasted or only partially contrasted, while phase
structures arranged orthogonally can only be depicted
simultaneously if they are diagonal to the image split
(Fig. 5, middle image of the bottom row).
On the other hand, circular objects can be depicted in
high contrast whatever the direction of the split, though the
nature of the image varies (Fig. 5, top row).

2. The limitations of the conventional DIC setup

according to Nomarski

In materials microscopy, both circular and straight

phase objects occur, which means that the sample and
the direction of the split have to be rotated relative to one
another to avoid misinterpretations due to the orientation
effect. Naturally, it makes sense to rotate the direction of
the split only, while keeping the object fixed, because
otherwise, highly precise and expensive rotary stages
would have to be used; in addition, the relationship
between the object, the orientation and the overall image
of the sample examined would be lost if the object was
rotated. Moreover, linear scanning stages are prescribed for
a large number of applications such as in microelectronics
and microsystems engineering. However, the DIC prism

1
In the modification according to Nomarski [1], the crystallographic
axis of one subprism does not lie parallel to the surface. This means that the
Fig. 2. Optical diagram of differential interference contrast (DIC) according ordinary and extraordinary beams are reunited (interference plane), not at
to Nomarski [1]: 1, light source; 2, collector; IE, input intensity; 3, polarizer; the interface between the individual prisms (DIC concept according to
4, flat glass; 5, DIC prism (modified Wollaston prism1); 6, objective exit Smith, 1947 [2]), but outside the Wollaston prism, and may coincide with
pupil; 7, objective; ‘‘blue’’ and ‘‘green’’, image and twin image of the the objective exit pupil which generally lies inside the lens system. This
reflected wave front and the wave front deformed by the object; 8, means that special objectives with an integrated Wollaston prism are no
specimen; 9, analyzer; IA, output intensity; 10, tube lens; 11, intermediate longer needed to perform the DIC method; furthermore, the interference
image plane. For the interpretation of the references to colour in this figure contrast can be varied (w-variation) simply by moving the prism
legend, the reader is referred to the web version of this article.1 perpendicular to the optical axis.
 R. Danz, P. Gretscher / Thin Solid Films 462 – 463 (2004) 257–262 259

Fig. 6. Conventional arrangement of the DIC prism (DP), in which it must

always be arranged diagonally to the polarizer and analyzer so that the
interfering waves of ordinary and extraordinary light have equal
amplitudes. If this is not the case, the interference contrast is reduced or

polarizer and analyzer, as illustrated in Fig. 6. The polar-

izer and analyzer have to be in the extinction position
aAn = p/2. In this case, the output intensity IA is observed
behind the analyzer, according to Fresnel [2], if there is no
phase object in the object plane:
Figs. 3. and 4. (a) The resulting phase differences (red) represent the first
derivation of the two identical light wave fronts displaced laterally by the
amount s (green and blue). (b) The microscopic image gives the impression
of a relief, although it is not directly possible to infer whether the object is IA ¼ f ðaDP ; WÞ ¼ IE sin2 2aDP sin2 1=2W ð3Þ
raised or depressed (model of the gradient object in Fig. 4b: Dr. Jan Thirase,
Göttingen). For the interpretation of the references to colour in this figure
legend, the reader is referred to the web version of this article. This therefore means that maximum intensities or
equal amplitudes of interfering component waves are only
(DP) according to Nomarski [1] cannot be rotated direc- to be expected if the DIC prism is arranged diagonal to
tionally because it only allows for the effective black/white the polarizer and analyzer; that is, if its oscillation
or gray contrast in defined fixed positions relative to the orientations aDP = (2n + 1)p/4. This happens precisely four

Fig. 5. Top row: circular phase objects can be depicted in high contrast whatever the direction of the split (arrows) of the DIC prism. Bottom row: straight
structures show a significant azimuth effect that leads to misinterpretations.
260 R. Danz, P. Gretscher / Thin Solid Films 462 – 463 (2004) 257–262

Fig. 10. Images of the same phase object in which the directions of the split
are perpendicular to each other. The difference in information is obvious
(objective, EC Epiplan – Neofluar 10  /0.25).
Fig. 7. The generation of circular polarized light: the incident light of input
intensity IE is polarized linearly by the polarizer (Pol) and, after passing
through the first quarter-wave plate (VWP1), is converted into circular
polarized light. Once the light has gone through the DIC prism for the first
time, reflected on the phase object and gone through the DIC prism for the times, at 45j, 135j, 225j and at 315j. In every other
second time, it passes through the second quarter-wave plate (VWP2) which orientation, the intensity is reduced, or even disappears
polarizes it linearly again. The output intensity IAV is independent of the for aDP = np/2 (n = 0, 1, 2, 3, . . .). Therefore, a (isotropic)
oscillation azimuth of the DIC prism because of the circular polarized light.
phase object in the ‘‘prohibited’’ orientation positions of
the DIC prism can hardly be displayed or cannot be
displayed at all.

3. C –DIC: differential interference contrast in circular

polarized light

What, then, should be done in order to be able to depict

all straight object structures with maximum contrast using a
fixed stage? The C –DIC method offers an elegant solution.
Using the approach described below, the oscillation direc-
tionality of the DIC prism can be continuously varied
without any loss of contrast.
In each of the illumination beam path and the image
beam path, a quarter-wave plate [4] (VWP1 or VWP2)
Fig. 8. Diagram illustrating the C – DIC arrangement: unlike in conventional
DIC, two further polarization – optical elements, namely the quarter-wave
is arranged in such a way that the oscillation azimuths
plates VWP1 and VWP2, have been added. Their azimuths a1 and a2 are of the two Plates form the angles a1 and a2 (Figs. 7
crossed and lie diagonal to the polarizer and analyzer. and 8).

Fig. 9. (a) C – DIC prism slider: the C – DIC prism can be rotated azimuthally using the large knurled ring; the image contrast is varied using the knurled
adjusting screw (change to the phase difference w, see text). (b) On both the light entrance and the light exit sides, the C – DIC reflector consists of a polar/
quarter-wave plate combination arranged crosswise.
 R. Danz, P. Gretscher / Thin Solid Films 462 – 463 (2004) 257–262 261

Fig. 11. Synthetically manufactured phase object containing perpendicular phase structures. The arrows indicate the direction of the split (objective, EC
Epiplan – Neofluar 20  /0.50).

The formula for the output intensity IAV, where there is no (orientation of the quarter-wave plates diagonal to the
object, is complicated at first compared to Eq. (3): polars) and d1 = d2 = 1/2p (the phase difference generated
by a quarter-wave plate),
IAV ¼ IE f½cos1=2d2 cos1=2wcos1=2d1 cosaAn

 cos1=2d2 sin1=2wsin1=2d1 cosðaAn  2aDP þ 2a1 Þ IAV ¼ IE ðsin2 1=2w þ cos2 aAn cosw
 cos1=2wsin1=2d2 sin1=2d1 cosðaAn  2a2 þ 2a1 Þ þ 1=2sinwsin2aAn cos2aDP Þ ð5Þ
 cos1=2d1 sin1=2d2 sin1=2wcosðaAn  2a2 þ 2aDP Þ2
þ ½cos1=2d2 cos1=2wsin1=2d1 cosðaAn  2a1 Þ and is finally simplified with crossed polars, i.e., aAn = 1/2p,
þ cos1=2d2 cos1=2d1 sin1=2wcosðaAn  2aDP Þ to
þ cos1=2wcos1=2d1 sin1=2d2 cosðaAn  2a2 Þ
 sin1=2d2 sin1=2wsin1=2d1 cosðaAn  2a2 IAV ¼ f ðwÞ ¼ IE sin2 1=2w ð6Þ
2
þ 2aDP  2a1 Þ g ð4Þ
However, it becomes much simpler once the conditions The intensity is now simply a function of w, no longer
for the two quarter-wave plates are inserted a1 =  a2 = 1/4p of aDP.

Fig. 12. Excimer laser ablation: signature mark on an optical surface (sample was kindly provided by Mr. A. Hanssen, Aalen; objective, EC Epiplan – Neofluar
20  /0.50).
262 R. Danz, P. Gretscher / Thin Solid Films 462 – 463 (2004) 257–262

The conditions are then in place for the successive the three images also show that structures with an alignment
detection and maximum contrast of differently aligned that does not coincide with any of the three directions
phase structures through rotation of the DIC prism (Fig. 5). chosen are always contrasted (pair of parallel structures
SSE –NNW).
Fig. 12 presents an optical surface that has been marked
4. Microscopy practice and examples of applications using an excimer laser. The quality of the ablation points
can be assessed superbly using C – DIC because every single
The C – DIC method was launched on the market in the piece of information about the object can be gathered due to
fall of 2002 in a number of Zeiss microscopes. It is the continuous change in the direction of the split. This is
extremely simple to use with these microscopes: circular absolutely impossible with conventional DIC because the
polarized light is generated and analyzed by a reflector special shape of the object examined means that a rotary
module (Fig. 9b), which is precisely the same size as the stage cannot be used.
polarizer/analyzer reflector for conventional DIC and can
therefore be placed in the same position.
The DIC prism for C – DIC is installed in a special slider 5. Summary
(Fig. 9a) which can be used for reproducible azimuthal
rotation in order to create high-contrast images of differently By using circular polarized light and suitably designed
aligned phase structures. The phase difference w can also be (rotating and sliding) C – DIC prisms, it is possible to depict
adjusted for gray contrast variation by moving the prism phase structures in incident light with high resolution and
perpendicular to the optical axis. high contrast, regardless of their alignment, without having
The first test objects used were MgF2 steps on a glass to rotate the object itself.
substrate (Fig. 10) and phase steps arranged orthogonally on
a front surface mirror (Fig. 11).
For the depiction of the MgF2 steps, where the directions References
of the split are perpendicular to one another, the huge
difference in information is obvious. Fig. 11 clearly dem- [1] G. Nomarski, Interféromètre à polarisation, French Patent 1.059.123,
onstrates that the vertical structures are more likely to be 1952.
depicted where the direction of the split is E – W (left [2] H. Beyer, H. Riesenberg, Handbuch der Mikroskopie, VEB Verlag
Technik, Berlin, 1988.
image); both the vertical and the horizontal structures are
[3] R. Danz, et al., Interferenzkontrastanordnung, Patent Specification DD
depicted where the direction is SE – NW (middle image), 257 888, 1988.
while, finally, the horizontal structures are more likely to be [4] Menzel, et al., Fourieroptik und Holografie, Springer, Vienna,
depicted where the direction of the split is S – N. However, 1973.