December 15, 1999 / Vol. 24, No. 24 / OPTICS LETTERS 1829

Visualizing the whispering gallery modes in acylindrical optical microcavity

M. L. M. Balistreri

Applied Optics Group, MESA+ Research Institute and Department of Applied Physics, University of Twente, P.O. Box 217,7500 AE Enschede, The Netherlands

D. J. W. Klunder, F. C. Blom,* A. Driessen, and H. W. J. M. Hoekstra

Lightwave Device Group, MESA+ Research Institute and Department of Applied Physics, University of Twente, P.O. Box 217,7500 AE Enschede, The Netherlands

J. P. Korterik, L. Kuipers, and N. F. van Hulst

Applied Optics Group, MESA+ Research Institute and Department of Applied Physics, University of Twente, P.O. Box 217,7500 AE Enschede, The Netherlands

Received August 24, 1999

Whispering gallery modes in cylindrical integrated optics microcavities have, for what is to our knowledgethe first time, been mapped with a photon scanning tunneling microscope. Optical images were obtainedwith a spatial resolution of 50 nm. By combination of information on the spatial optical distributionswith wavelength-dependent measurements, an unexpectedly rich variety of intracavity phenomena, such aspolarization conversion and interference of copropagating and counterpropagating modes, could be directlyobserved. A quantitative comparison of the experimental data with computer simulations results in acomprehensive understanding of the various whispering gallery modes inside the microcavity. 1999 OpticalSociety of America

OCIS codes: 130.0130, 130.3120, 250.5750, 170.5810, 170.0110.

The rapidly increasing bandwidth demand of commu-nication systems has resulted in an immense drivetoward all-optical networks in which, besides datatransport, routing and switching are also performed inthe optical domain. As the trend is to employ a largenumber of densely spaced wavelengths (more than 100)at high speed (tens of gigabits per second),1 the re-alization of the network nodes requires very reliableand complex optical structures. Similar to the case ofminiaturization in electronics, these advanced opticalstructures can be realized only by use of integratedcircuitry. For integrated electronic circuitry, effortsover the past 30–40 years have led to a substantial sci-entific and technological infrastructure, which is stilllargely missing for photonic structures.

In this Letter we report an important step in the de-velopment of optical waveguide technology. For whatis to our knowledge the first time, the optical field in-side a nontrivial integrated optics device, a cylindri-cal microcavity, has been experimentally determinedwith unprecedentedly high spatial and spectral reso-lution. This approach contrasts with conventional ex-perimental methods that mostly have been restrictedeither to a black-box-like characterization with respectto the functional behavior or to coarse, diffraction-limited imaging techniques.2 In our case, looking di-rectly into the microcavity by means of a photonscanning tunneling microscope (PSTM) with a spatialresolution of 50 nm revealed a variety of unexpectedphenomena. In the case of microspheres, the bulk-optics equivalent of the resonators presented here,

0146-9592/99/241829-03$15.00/0

Knight et al.3 have successfully measured the whisper-ing gallery modes (WGM’s) at the surface by photontunneling.

Integrated optical microcavities recently attractedbroad interest, as they can be applied as compactfilter elements in dense wavelength-division multi-plexing networks,4 microlasers,5,6 all-optical switches,7

and more generally as all-optical data processing ele-ments.8 The cylindrical microcavity that we investi-gated in this study7 is shown schematically in Fig. 1.The straight waveguide is present for the excitation ofthe WGM’s. The light in the WGM’s circles around byrepeated total internal ref lection at the cavity bound-ary. As a result, the WGM’s form a collection of con-centric rings close to the rim of the cavity. When thewavelength of the light in a WGM fits an integer num-ber of times in a round trip, the microcavity is in reso-nance for that specific mode.

The principle of photon tunneling is based on thelocal frustration of the evanescent field with a decaylength of 40 nm at the cavity–air interface by a near-field optical fiber probe9 – 11 with a 50-nm aperture12

(Fig. 1). As a result, the evanescent wave is locallyconverted into a propagating wave that is coupled intothe aperture and guided through the fiber toward a de-tector. Scanning the probe over the microcavity sur-face allows one to construct an image of the opticalfield distribution, with a resolution given by the sizeof the aperture rather than by the wavelength of light.For quantitative high-resolution imaging it is essentialto operate the probe at a constant distance above the

1999 Optical Society of America

1830 OPTICS LETTERS / Vol. 24, No. 24 / December 15, 1999

Fig. 1. Schematic overview of WGM’s in a cylindrical mi-crocavity, which are probed with a PSTM. The cylindri-cal Si 3N4 microcavity (radius, 64 mm; height, 115 nm) anda straight Si 3N4 channel waveguide (step height 9 nm)are fabricated in a Si 3N4 SiO2 layer system on a Sisubstrate.13

surface, while staying well within the evanescent field.A height feedback mechanism based on shear force in-teraction is implemented that keeps the probe heightat �10 nm.13 Importantly, the force feedback yields ahigh-resolution topographic image that is simultane-ously obtained with the optical field distribution.

Topographical and optical maps of the microcavityare shown in Figs. 2A and 2B, respectively. In the top-ographical image the cylindrical shape of the microcav-ity, with a radius of 64 mm and a height of 115 nm, canbe clearly seen. The f lat facet on the left-hand sideof Fig. 2A is caused by the processing of the adjacentstraight waveguide that is used to excite the WGM’s.The WGM’s were excited with TE-polarized light from atunable dye laser. The microcavity is brought into res-onance at a wavelength near 674 nm. As expected, theoptical image (Fig. 2B) clearly shows a light distribu-tion that is confined to the outer rim of the microcavity.Actually, one really can see the propagation of the lightboth in the adjacent waveguide and in the microcavity.Unexpectedly, however, an interference pattern, ratherthan the expected perfect rings, is observed in the op-tical field distribution. Such a pattern indicates thatseveral WGM’s are simultaneously excited in the micro-cavity. Moreover, this pattern means that the variousallowed modes interfere with one another. By zoom-ing in (Fig. 2C), we find a period of the order of 8 mmfor the so-called mode-beat pattern. Closer examina-tion (Fig. 2D) reveals yet another interference pattern,with a periodicity near 190 nm. This periodicity issurprisingly small and can only be the result of inter-ference between two modes, one propagating clockwiseand the other counterclockwise. This observation un-mistakably shows the power of the PSTM, as this sub-wavelength feature would have remained hidden fromconventional imaging techniques.2

To unravel the interplay of the various modes in thecavity further and thus find the origin of the observedinterference patterns we performed spectral scans.An attractive method for combining spectral and spa-tial scanning works as follows: While we continuouslyscan one line perpendicular to the rim of the

microcavity, the wavelength of the incoming lightis varied with time. As a result, we build up a quasiimage (Fig. 3A) in which each vertical line representsthe optical field distribution in the radial direction forone particular wavelength. Horizontal cuts representthe variation of the optical field as a function of wave-length; i.e., they give the spectrum for one particularradial position. The wavelength scans have one addi-tional advantage, as they allow the determination ofthe relative phases of the different modes. Figure 3Adepicts a wavelength scan for a wavelength range from648 to 678 nm. Two conclusions are immediatelyobvious. First, the vertical features in the imagethat vary rapidly as a function of wavelength indicate

Fig. 2. PSTM images of the microcavity and the generatedWGM’s. The dashed and the solid lines indicate the cor-responding line profile and cavity edge, respectively. A,Topography of the Si 3N4 cavity detected in shear forcefeedback. The line profile shows the 128-mm diameterand the average height of 115 nm. The straight couplingchannel is positioned at the left-hand side of the cavity.B, Photon tunneling image of a microcavity in resonanceat a wavelength of 674 nm, recorded simultaneously withA. The intensity profile shows the confinement of theWGM field close to the cavity edge. C, A close look at thecavity edge (solid curve) reveals modal fields at different ra-dial distances. Moreover, a beat pattern is observed, witha period of approximately 8 mm (arrows and line profiles).D, High-resolution photon tunneling image of a spatial beatpattern close to the cavity edge. An interference patternwith a 190-nm period is observed. The large modulationdepth of the interference fringes (line profile) indicates theunique spatial sensitivity of the near-field optical probe.

December 15, 1999 / Vol. 24, No. 24 / OPTICS LETTERS 1831

Fig. 3. A, Wavelength scan at the boundary of the mi-crocavity. The white arrows in Fig. 2A indicate the po-sition of the scan. The wavelength was varied from 648to 678 nm in steps of 0.03 nm. B, Two line plots (spec-tra) at the radial positions indicated by the dotted lines inA. Each WGM has a specific FSR.

that the cavity goes in and out of resonance when thewavelength is varied. Second, the optical fields at therim exhibit two distinct bands along the edge of the cav-ity, corresponding to the rings in Fig. 2C. Figure 3Bshows two spectra through the outer bands (i.e., fordifferent radial positions). Both spectra show in-resonance and out-of-resonance behavior. It is clearfrom the dashed vertical lines in Fig. 3B that the reso-nance behavior at the two radial positions is out ofphase as a result of the difference in propagationconstants of the WGM’s. In addition to the rapidlyvarying resonance aspects, the spectra exhibit aslowly varying feature. This feature is attributed tothe beating of the various WGM’s, analogous to theobservations in the spatial maps (Fig. 2C).

A Fourier analysis of the spectra yields the so-calledfree spectral ranges (FSR’s) for the various radialWGM’s and the periods corresponding to the spectralmode beat. A detailed scrutiny of the Fourier analy-sis combined with simulations of the microcavity be-havior yields several surprises. First, in addition tothe expected FSR of 0.54 nm, which corresponds to theTE-polarized WGM, a FSR of 0.51 nm was observed.The calculations show that this FSR can be attributedonly to a TM-polarized mode in the cavity. The exis-tence of this mode is also confirmed by the 8-mm beatlength of the interference pattern in Fig. 2C, whichcan be attributed only to the spatial mode beat of aTE-polarized mode with a TM-polarized mode. Thepresence of TM-polarized light was unexpected, sinceonly TE-polarized light was coupled into the structure.The second outcome from the analysis is that the ob-served spectral periods near 1.0 nm correspond to theinterference between clockwise- and counterclockwise-propagating modes. This outcome further corrobo-rates our earlier conclusion concerning the 190-nmspatial beat patterns, which was based solely on directoptical imaging (Fig. 2D). We attribute both the polar-ization conversion and the change in the propagationdirection to the coupling of the straight waveguide withthe microcavity. Note that, against intuition, TE- as

well as TM-polarized light can be simultaneously de-tected, leading to the observed interference pattern ofthe mutually orthogonal fields.14

In conclusion, it has been demonstrated that aPSTM yields detailed information on the optical fielddistribution of a nontrivial planar waveguide deviceas a function of wavelength and position. The infor-mation is obtained with a spatial resolution that isunattainable with conventional microscopic methods.Moreover, the measurements allow a quantitativecomparison with computer simulations. Includingthe unexpected PSTM observations in the computercalculations that are based exclusively on geometricaland materials properties of the device results in anexcellent simulation of the behavior of the device. Itis anticipated that the PSTM will be used for the devel-opment of novel high-performance planar waveguidedevices, e.g., dense wavelength-division multiplexers,in which demands on the device parameters are ex-tremely high.

This work was supported by the Dutch Foundationfor Fundamental Research on Matter (FOM). Weare grateful to R. Stoffer for fruitful discussions.M. L. M. Balistreri’s e-mail address is m.l.m.balistreri@tn.utwente.nl.

*Present address, Uniphase Netherlands B. V.,Prof. Holstlaan 4, Eindhoven, The Netherlands.

References

1. G. E. Keiser, Opt. Fiber Technol. 5, 3 (1999).2. D. Rafizadeh, J. P. Zhang, R. C. Tiberio, and S. T. Ho,

J. Lightwave Technol. 16, 1308 (1998).3. J. C. Knight, N. Dubreuil, V. Sandoghdar, J. Hare, V.

Lefevre-Seguin, J. M. Raimond, and S. Haroche, Opt.Lett. 21, 698 (1996).

4. B. E. Little, S. T. Chu, H. A. Haus, J. Foresi, and J. P.Laine, J. Lightwave Technol. 16, 998 (1997).

5. C. Gmachl, F. Capasso, E. E. Narimanov, J. U. Nockel,A. D. Douglas Stone, J. Faist, D. L. Sivco, and A. Y.Cho, Science 280, 1557 (1998).

6. A. F. J. Levi, R. E. Slusher, S. L. McCall, J. L. Glass,S. J. Pearton, and R. A. Logan, Appl. Phys. Lett. 62,561 (1993).

7. F. C. Blom, D. R. van Dijk, H. J. W. M. Hoekstra, A.Driessen, and Th. J. A. Popma, Appl. Phys. Lett. 71,747 (1997).

8. P. L. Gourley, Sci. Am. 278, 56 (1998).9. A. G. Choo, H. E. Jackson, U. Thiel, G. N.

De Brabander, and J. T. Boyd, Appl. Phys. Lett.65, 947 (1994).

10. S. Bourzeix, J. M. Moison, F. Mignard, F. Barthe,A. C. Boccara, C. Licoppe, B. Mersali, M. Allovon, andA. Bruno, Appl. Phys. Lett. 73, 1035 (1998).

11. N. F. van Hulst, M. H. P. Moers, O. F. J. Noordman,R. G. Tack, F. B. Segerink, and B. Bolger, Appl. Phys.Lett. 62, 461 (1993).

12. E. Betzig and J. K. Trautman, Science 257, 2484 (1992).13. K. Karrai and R. D. Grober, Appl. Phys. Lett. 66, 1842

(1995).14. E. Betzig and R. J. Chichester, Science 262, 1422

(1993).15. R. M. de Ridder, K. Worhoff, A. Driessen, P. V.

Lambeck, and H. Albers, IEEE J. Sel. Top. QuantumElectron. 4, 930 (1998).