August 1, 2005 / Vol. 30, No.

15 / OPTICS LETTERS

1953

Spiral interferometry
Severin Frhapter, Alexander Jesacher, Stefan Bernet, and Monika Ritsch-Marte
Division of Biomedical Physics, Innsbruck Medical University, Mllerstrasse 44, A-6020 Innsbruck, Austria Received March 14, 2005 We present a surprising modication of optical interferometry. A so-called spiral phase element in the beam path of a standard microscope results in an interferogram of phase samples, for which the interference fringes have the shape of spirals instead of closed contour lines as in traditional interferograms. This conguration overrides the basic problem of interferometry, i.e., that elevations and depressions cannot be distinguished. Therefore a complete sample prole can be reconstructed from a single exposure, promising, e.g., high-speed metrology with a single laser pulse. The method is easy to implement, it does not require a spatially separated reference beam, and it is optimally stable against environmental noise. 2005 Optical Society of America OCIS codes: 070.6110, 090.1970, 100.5090.

Optical interferometry measures the phase prole of a sample with subwavelength resolution. Interferograms of sufciently smooth objects consist of closed contour lines of interference maxima and minima. Each contour line is an indicator for the height of the phase prole, measured in units of the optical wavelength. However, there are the problems that the phase level is dened by the contours only modulo one wavelength and that there is an ambiguity whether going to an adjacent contour line is a step up or down. In standard interferometry there exist several methods of fringe demodulation to deal with this problem. Typically, one collects additional information by recording more than one interferogram, e.g., with different phase offsets between object and reference beams (phase stepping), or from different directions (angular multiplexing), or with different illumination wavelengths (wavelength multiplexing) However, in all these cases multiple interferograms have to be recorded, which requires the sample to be mechanically stable between successive exposures. Specialized setups for instantaneous phase measuring based on simultaneous recording of multiple phase-stepped interferograms1 with a corresponding number of image detectors are quite complicated. Here we demonstrate a type of spiral interferometry that allows unambiguous reconstruction in one step. A spiral phase element in the beam path of an optical microscope used as a spatial lter generates spiral-shaped interference fringes that cover the area of the phase sample. The local direction of the fringes is a measure of the phase level and, most notably, the direction of rotation of the spirals allows one to distinguish between depressions and elevations in the sample. In traditional interferometry of a surface or a transmissive phase object, an image-carrying light wave is coherently superposed upon a reference wave, resulting in closed interference fringes that form contour lines. As an example, a simulated surface prole and its corresponding interferogram are sketched in Figs. 1(a) and 1(b), respectively. In such an interferogram the resultant contour fringes [Fig. 1(b)] do not distinguish between troughs and ridges, and it is not clear whether a neighboring contour is
0146-9592/05/151953-3/$15.00

uphill or downhill. Therefore the overall phase prole of a sample cannot be reconstructed uniquely. Our modication of interferometry, which is explained below, overcomes this ambiguity. Figure 1(c) shows simulated interferograms that were obtained by spatial ltering of the image wave with a modied spiral phase hologram.2,3 These interferograms consist of open spirallike interference fringes instead of closed contour lines. In the ideal case a single intensity fringe covers the whole area of the image, splitting at saddle points of the phase topography. The rotational orientation of the interference spirals distinguishes elevations from depressions within the samples phase prole. The local tangential direction

Fig. 1. Reconstruction of a simulated phase prole by the method of spiral interferometry. (a) Simulated sample prole of a surface or a refractive structure. (b) Normal interferogram. The closed-contour lines do not distinguish between elevations and depressions. (c) Spiral interferogram obtained by ltering with the modied spiral phasecontrast method. Depending on the topography, the spirals change their rotational direction. (d) A single contour line of the spiral interferogram. (e) Processed contour line. The local direction of the line is proportional to the surface height, modulo one wavelength, which allows a unique height to be assigned to each single point of the contour line. (f) Reconstructed surface prole by interpolation between the sampling points given by the contour line. 2005 Optical Society of America

1954

OPTICS LETTERS / Vol. 30, No. 15 / August 1, 2005

Fig. 2. Sketch of the experimental setup. A phase sample at a microscope sample stage is illuminated with light from a laser diode. The transmitted light is collected by a microscope objective. A relay lens L1 is used to image the Fourier transform of the image wave at a SLM displaying a phase hologram. The light diffracted from the SLM into the rst diffraction order is imaged through lens L2 at a CCD camera, after the other diffraction orders are blocked. At the left the central section of a spiral phase hologram is sketched, which is displayed at the SLM (gray levels correspond to phase values). The modication necessary to produce the described spiral fringes is done by substituting a blazed grating (indicated above), which diffracts the zeroorder Fourier spot of the incident image wave, for a circular area about the central singularity. At the image detector this zero-order Fourier component evolves into a plane wave, which interferes with the remainder of the image wave, as required for spiral interferometry.

of the spiral fringes is a measure of the phase level of the sample. Thus it is possible to use a standard contouring algorithm to calculate a contour line at an adequate gray level of the image [indicated in Fig. 1(d)]. Then one processes the contour by computing the local tangential direction, which gives the height of the phase prole at the corresponding sample position in units of optical wavelength . Since the local tangential direction repeats after each 2 revolution, the phase level is given only modulo ; i.e., to compute the absolute phase level one has to keep a record of the number of revolutions. The result is a threedimensional spiral curve with a unique phase height level at each position, as shown in Fig. 1(e). To obtain a complete reconstruction of the object phase prole, one can interpolate the data by using the processed data points at the contour line for sampling [Fig. 1(f)]. Our method to create such spiral interferograms is a self-referenced interferometric method; i.e., it works without using a separate reference beam. Instead, it uses the zero-order Fourier component of the incident image wave as a reference by manipulating it in a different way from the remaining wave, similar to the well-known method of phase-contrast microscopy. In our setup the phase of the image wave is spatially ltered in its Fourier plane with a spiral phase element (a spiral phase plate or a phase hologram3), which imprints a vortex-shaped phase prole46 of the form x + iy / x + iy at the incident wave front, where x and y are Cartesian coordinates measured from the center of the spiral phase element. In our application of spiral interferometry, the singularity in the center is replaced by a blazed grating (indicated in Fig. 2), which diffracts the zeroorder Fourier component of the incident image wave

to the same direction as the surrounding spiral phase hologram. Thus the zero-order Fourier component is interferometrically superimposed upon the remaining image wave with a specic relative phase, which can be adjusted by control of the phase of the inner grating. For smooth sample phase topographies, the superposition of the spiral-ltered image wave by the plane wave from the zero-order Fourier component results in the spiral-shaped interferograms reported here. Phase-step discontinuities (except for integer wavelength steps) can be identied as an abrupt fringe displacement. The rotational phase of the interference spirals can be controlled continuously by shifting of the phase of the inner grating. Basically, the appearance of the interference spirals is a consequence of the fact that the spatial ltering operation with a spiral phase element is a two-dimensional analog of the actually one-dimensional Hilbert transform.4,7 Its subsequent interferometric superposition by a plane wave (i.e., by the zero-order Fourier component) then reveals the phase information in the Hilbert-transformed image, which is spiral shaped. The appearance of spiral fringes in interferometric setups has already been reported8 and suggested for interferometric applications.9 An analytical derivation for the shape of similar interference spirals was recently presented and suggested for beam collimation testing.10 A sketch of our setup is displayed in Fig. 2. A transmissive phase sample is placed on a microscope stage (Zeiss Axiovert 135) and illuminated with a plane wave (TEM00 mode) emerging from a 50 mW laser diode at a wavelength of 780 nm. The image wave is collected by a microscope objective (magnication, 20 ; N.A., 0.3) and guided through a relay lens to a reective spatial light modulator (SLM) outside the microscope. The optical setup is designed such that the rear focal plane of the microscope objective (indicated in the gure) is expanded by a factor of 2 and is imaged at the position of the SLM. The SLM is a high-resolution liquid-crystal display that contains 1920 1200 square pixels of 10 m 10 m size, which act as individually addressable phase shifters for a reected light wave. For spiral phase ltering, the SLM displays the hologram sketched at the left in Fig. 2 (not to scale; gray levels correspond to phase values in a range from 0 to 2 ), which consists of a blazed phase grating with a forklike dislocation in its center. Such gratings are known to produce so-called doughnut beams (LaguerreGauss beams) by rstorder diffraction of an incident Gaussian wave. The number of tines of the fork denes the so-called topological charge of the doughnut mode, which has to be set to 1 to allow the two-dimensional Hilbert transform to be performed. The hologram is centered with respect to the incident zero-order Fourier spot of the image wave. The spiral-phase-ltered rst-order diffracted image wave is spatially separated from the other diffraction orders and imaged at a CCD camera. An experimental demonstration of spiral interferometry is shown in Fig. 3. Figure 3(a) displays a normal interferogram of a phase sample that consists of

August 1, 2005 / Vol. 30, No. 15 / OPTICS LETTERS

1955

Fig. 3. Experimentally obtained interferogram of an oil drop smeared on a glass coverslip. (a) Normal contourlike interference fringes. (b) Spiral interferogram of the same sample region obtained after ltering with the modied spiral phase hologram (the blazed grating in a small central area; upper left in Fig. 2). (c) Section of the spiral interferogram to be processed. (d) Single contour line of the spiral interferogram. (e) Surface prole reconstructed by processing the contour line and tting the surface at the obtained sampling points.

oil smears on a glass coverslip, which was recorded by holographic implementation of standard phasecontrast microscopy.5 We obtained the corresponding spiral interferogram [Fig. 3(b)] by switching the SLM to display a hologram like that sketched in Fig. 2, in which the central singularity of a spiral phase hologram is replaced by a blazed grating. The diameter of the central grating (which is typically 10 pixels, i.e., 100 m) can be adjusted interactively (by trial and error) to produce a maximal interference contrast. We processed a part of the interferogram [upper middle oil drop; Fig. 3(c)] with the contour-line method described above. The contour lines obtained at a single, intermediate gray level with a standard numerical contouring algorithm are drawn in Fig. 3(d). After the contour was processed by the method described above, and after subsequent threedimensional interpolation, the reconstructed object surface shown in Fig. 3(e) was obtained. The results demonstrate that one can reconstruct a whole phase prole of a sample landscape from a single interferometric image without facing the wellknown general problem of ambiguities in the interpretation of typical interferograms. This may have practical potential, for instance, for high-speed interferometric imaging of fast processes with pulsed la-

sers. If single-exposure imaging is not necessary, the resolution can be increased by methods analogous to phase stepping in traditional interferometry, in which one can adjust the phase steps continuously by just shifting the phase of the inner grating, resulting in revolving interference spirals. Numerical analysis of a sequence of spiral interferograms recorded with different rotational phases provides an accurate method of phase reconstruction without the requirement of mechanical phase stepping. A further advantage is that the method can easily be implemented with existing standard microscopes. For this purpose it is not necessary to use an electronically controlled SLM; one may merely insert a transparent spiral wave plate into a Fourier plane of the beam path, which can be at the rear objective aperture plane, and thus no further optical components need be inserted. This kind of interferometry has the advantage that it is insensitive to external vibrations because the image wave is used as its own reference. Possible imprecision owing to wave-front and polarization aberrations caused by high-curvature microscope lenses can for the most part be compensated for by subsequent numerical processing, by use of a calibration interferogram recorded without a sample. Whereas interferometry of transmissive samples probably has its major eld of applications such as imaging of cells in biology, a modication of the illumination setup (similar to epiuorescence illumination) makes the method feasible also for surface reection interferometry, which is important for many technical applications, e.g., in the semiconductor industry. S. Bernets e-mail stefan.bernet@uibk.ac.at. References address is

1. R. Smyth and R. Moore, Opt. Eng. 23, 361 (1984). 2. S. N. Khonina, V. V. Kotlyar, M. V. Shinkaryev, V. A. Soifer, and G. V. Uspleniev, J. Mod. Opt. 39, 1147 (1992). 3. N. Heckenberg, R. R. McDuff, C. P. Smith, and A. G. White, Opt. Lett. 17, 221 (1992). 4. J. A. Davis, D. E. McNamara, D. M. Cottrell, and J. Campos, Opt. Lett. 25, 99 (2000). 5. K. Crabtree, J. A. Davis, and I. Moreno, Appl. Opt. 43, 1360 (2004). 6. S. Frhapter, A. Jesacher, S. Bernet, and M. RitschMarte, Opt. Express 13, 689 (2005). 7. R. Bracewell, The Fourier Transform and Its Applications (McGraw-Hill, 1965). 8. R. R. Shannon, R. E. Weekley, and D. Shafer, Appl. Opt. 4, 1193 (1965). 9. C. Guo, X. Cheng, X. Ren, J. Ding, and H. Wang, Opt. Express 12, 5166 (2004). 10. P. Senthilkumaran, Appl. Opt. 42, 6314 (2003).