Journal of Microscopy, Vol. 202, Pt 1, April 2001, pp. 7±11. Received 28 August 2000; accepted 1 December 2000 Near-field optical microscopy based on microfabricated probes R. ECKERT* 1, J. M. FREYLAND*, H. GERSEN* 2, H. HEINZELMANN*, G. SCHUÈ RMANN², W. NOELL², U. STAUFER² & N. F. DE ROOIJ² *Centre Suisse d'electronique et de Microtechnique CSEM S.A., Rue Jaquet-Droz 1, 2007 NeuchaÃtel, Switzerland ²Institut de Microtechnique, Universite de NeuchaÃtel, Rue Jaquet-Droz 1, 2007 NeuchaÃtel, Switzerland Key words. Fluorescence microscopy, microfabricated cantilevers, optical nearfield, NSOM, single molecule, SNOM. Summary We demonstrate high resolution imaging with microfabricated, cantilevered probes, consisting of solid quartz tips on silicon levers. The tips are covered by a 60-nm thick layer of aluminium, which appears to be closed at the apex when investigated by transmission electron microscopy. An instrument specifically built for cantilever probes was used to record images of latex bead projection patterns in transmission as well as single molecule fluorescence. All images were recorded in constant height mode and show optical resolutions down to 32 nm. Introduction In contrast to scanning force microscopy (SFM), which to date is a routine tool for research and development applications as well as in quality control, scanning near-field optical microscopy (SNOM) is not a widespread technique. The main reasons are the still lacking high quality, reproducible and inexpensive probes. Today's probes are mostly manufactured from optical fibres by either heat-pulling or etching (StoÈckle et al., 1999). These are serial processes which are expensive and not suited for reproducible production, and that lead to probes with aperture shapes and sizes that are hard to control. Focused ion-beam milling techniques have improved the aperture characteristics (Lacoste et al., 1997; Veerman et al., 1998), but they are expensive and do not overcome the problem of having to produce and treat the probes one by one. Correspondence to: R. Eckert. Tel.: 141 32 7205 332; fax: 141 32 7205 700; e-mail: rolf.eckert@csem.ch 1 Also at: Institute of Physics, University of Basel, Klingelbergstr. 8, 4056 Basel, Switzerland 2 Permanent address: Faculty of Applied Physics & MESA 1 Research Institute, University of Twente, PO Box 217, 7500 AE Enschede, The Netherlands. Microfabrication processes solve some of these problems and would allow the production of cheap and highly reproducible probes. Ideally, they would be based on cantilevers so all the force imaging modes established in SFM can be utilized, in particular allowing imaging in liquids. This would make SNOM more accessible to non-specialists. In this contribution we report on a microfabrication process for the production of near-field optical probes. The probes are silicon cantilevers with aluminium-coated solid amorphous quartz tips. In order to utilize them a new instrument had been designed which is similar to common AFM set-ups with a feedback based on the beam deflection scheme. Although the probes show no apparent aperture, a full width at half maximum (FWHM) resolution of below 32 nm has been obtained. Microfabricated probes This design of the microfabricated probes allows all common imaging modes of force microscopy to be used. The probes were produced using a microfabrication process, which is reproducible and highly parallel and therefore inexpensive. In contrast to similar work (Noell et al., 1998; Zhou et al., 1998; Vollkopf et al., 1999) the tips used in this work were made of solid amorphous quartz with an index of refraction of 1.51. The resulting reduction of the cut-off radius for non-dissipative light propagation leads to higher transmission. In Fig. 1 an example of such a rectangular-shaped silicon cantilever with a solid quartz tip coated with aluminium is shown. As the cantilevers were of very high stiffness (spring constants deduced from the geometry of the levers were in the range from 1 to 7 N m 21 ) no snap-in events could be observed upon approaching the sample and constant height mode scans could be performed. Rectangular windows q 2001 The Royal Microscopical Society 7
8 R. ECKERT ET AL. a) Al coated quartz tip Si-cantilever 100 µ m b) 60 nm etched into the levers at the position of the tips were used for light coupling. A more detailed description of the tip production process is given elsewhere (to be published). The applicability of the probes for polarization and lowlight level fluorescence measurements was investigated by measuring their polarization and auto-fluorescence properties. Figures 2(a) and (b) show the results of an experiment determining the degree of polarization (DOP) defined by (I max 2 I min )/(I max 1 I min ) of the light transmitted through an aperture of a four-sided pyramidical and conical tip, respectively. Even though the maximum value of the DOP is,0.6 for both cases, the conical tips are more suited for polarization experiments because of their constant DOP for all angles (Lacoste et al., 1998). A comparison of the autofluorescence spectra for a microfabricated tip (solid line) and a conventional 20 cm long tapered optical fibre (dashed line) is shown in Fig. 2(c). The high autofluorescence and Raman scattering background of the fibre in comparison to the microfabricated probe clearly illustrates the advantage of microfabricated probes for low light level applications. Experimental set-up 200 nm Fig. 1. (a) SEM image of a microfabricated optical near-field probe with a rectangular shaped silicon cantilever. The tip is made of amorphous quartz. (b) TEM image of an unused aluminium coated quartz tip. A 60-nm thick aluminium film can be clearly distinguished from the quartz core. The film is also covering the apex of the tip. In order to perform the cantilever-based SNOM work reported here, we developed the dedicated microscope shown schematically in Fig. 3. A home-built tripod microscope head holds the cantilever at an angle of 158 with respect to the sample. An inverted optical microscope (Zeiss Fig. 2. Degree of polarization for (a) a four-sided pyramidal and (b) a conical tip measured for varying launching angles of the linear polarized light. (c) Spectra of the autofluorescence of a microfabricated probe with solid quartz tip (solid line) and of a conventional 20 cm long optical fibre probe (dashed line). Axiovert 135, Oberkochen, Germany) serves as a base for the microscope head. The cantilever deflection is measured by an optical beam deflection scheme (Meyer & Amer, 1988). Scanning of the sample in the x-y plane is provided by a closed-loop piezoelectric scanner (scan range 100 mm 100 mm, Physik Instrumente, Waldbrom, Germany). Three coupled piezoelectric actuators move the sample up to 15 mm in the z-direction. An apochromatic lens is used to focus both the excitation (488 nm) and feedback (670 nm) light beams onto the cantilever, after combining them by a non-polarizing beam splitter. A lateral separation of the foci on the cantilever results from different incident angles of the two beams onto the lens. Manipulating the incident angles independently allows for individual positioning of the feedback and excitation focus on the lever, ensuring that no red light from the feedback beam is directly coupled into the tip. A microscope objective (100, NA ˆ 1.25) is used to collect the transmitted excitation and fluorescence light. The signals are separated by a dichroic mirror (488/514 nm) and coupled into multimode fibres with core diameters of 62.5 mm. As well as collecting the signal, these fibres serve as spatial filters in the confocal arrangement, rejecting any light not originating from the apex of the probe. The transmitted signal is recorded using a photomultiplier tube (PMT). The much weaker fluorescence signal is first cleaned from residual transmitted excitation and feedback laser light using a holographic notch and a bandpass filter and then detected by a single photon counting avalanche photodiode (APD). To monitor tip and
NFO MICROSCOPY BASED ON MICROFABRICATED PROBES 9 Fig. 3. Set-up of the combined optical near-field and force microscope. sample during approach a small fraction of the light can be split off and directed towards a CCD camera. Commercial control electronics (STM 1000, RHK, Michigan, U.S.A.) are used for data acquisition and controlling the instrument. Results The first sample studied to determine the optical resolution achievable with the microfabricated probes was a latex bead projection pattern (Kentax, MuÈ nster, Germany). The regular pattern with hexagonal symmetry was produced by evaporating a thin aluminium film (35 nm) through a shadow mask formed by a closed packed layer of 22 nm latex spheres. Dissolving the spheres after evaporation results in a regular arrangement of metal islands. The experiments on these samples were performed in constant height mode. Figure 4(a) shows a near-field transmission image recorded on this sample. The hexagonal periodicity of the structure is clearly visible with the metal islands appearing as black spots and lattice defects as slightly larger black areas. A first estimate of the optical resolution can be deduced from the width of the measured edges of the islands if one assumes them to have perfectly steep edges. To this purpose, a cross-section through the image was taken (Fig. 4b), which suggests a resolution of,31 nm. This is not a conclusive estimate, however, firstly because the exact edge geometry of the metal islands is not known and secondly some influence of the sample topography on the Fig. 4. (a) Near-field light transmission through a metal island film. (b) Smoothed intensity profile along the white line. optical signal cannot be excluded (Betzig & Trautman, 1992). Even though the scans were performed in constant height mode, bending and friction images (not shown) acquired simultaneously with the optical data show small signals. Because the extent of these interactions and consequently their influence on the optical channel cannot be quantified, the high resolution could partially be due to topography artefacts. An ideal test sample for determining the optical resolution is a point-like source, which avoids issues of edge sharpness and large topographical features (Veerman et al., 1999). An excellent approximation of such a point-like source is a single fluorescently labelled biomolecule. In our experiments we therefore used goat-anti-rat antigens labelled with AlexaFluor 488 (495/519 nm absorption/ emission maximum) randomly distributed on a glass cover slip as a test sample. The sample was prepared by spincoating a 10 26 m solution of the molecules onto the glass, which had previously been cleaned in 2% hydrofluoric acid and high purity water. Figure 5(a) shows a near-field fluorescence image (256 256 pixels) recorded on such a sample recorded
10 R. ECKERT ET AL. Fig. 5. (a) Fluorescence image of labelled biomolecules. (b) Intensity profile along the white line. The optical resolution deduced from the FWHM value of a multipeak fit (solid line) to the data (dotted line) is better than 32 nm. The average FWHM of 47 molecules in the image yields an optical resolution of 31.7 ^ 3.6 nm. in constant height mode with a tip sample gap of,10 nm (neither topography, nor bending or friction signals, which were all simultaneously acquired, show any variations over the whole scan). Roughly 85 bright spots corresponding to single molecules can be identified, the maximum fluorescence count rate is,16 kcps with a binning time of 4 ms per pixel. Gaussian fits of the cross-sections of the 47 brightest molecules yield a FWHM of 31.7 ^ 3.6 nm. Figure 5(b) shows an example of such a cross-section, where the four individual peaks were fitted by a multipeak fit, yielding resolutions better than 32 nm. The small standard variation of the resolution across the whole image is a sign that there was very little wear of the probe during the scan. Conclusion We have demonstrated high-resolution near-field optical imaging with microfabricated, cantilevered probes. The resolution achieved was 32 nm. A closed aluminium layer at the apex of the solid quartz tips could be observed in transmission electron microscope experiments. The probes proved to be especially well suited for fluorescence contrast imaging and polarization contrast applications, owing to their low auto-fluorescence and polarization conservation properties, respectively. The mechanism of light coupling through the metalized tip and light confinement at the tip apex is not yet understood. The far-field transmission through the thin metal layer could be because of the finite conductivity of aluminium. However, a different effect must be responsible for the light confinement necessary for the subdiffraction limited resolution. A potential explanation relies on modes propagating along the air±metal interface in combination with field localization at the tip apex. The possibility that the tip is modified during approach is exceedingly small, given the small and defined aperture that had to be produced in a reproducible way. This is supported by the fact that the tip does not alter during scanning. Additionally, from other experiments we would expect that rather high forces would be necessary to break up a continuous metal layer as it is found at the apex. We consider it to be very likely that microfabricated probes similar to the type used for the work reported here will play an increasingly important role in future near-field optical microscopy. In particular, their usability in many environments and experimental configurations, alongside with their ease of use, makes them attractive. This work was partially funded by the Swiss Priority Program MINAST. References Betzig, E. & Trautman, J.K. (1992) Near-field optics: microscopy, spectroscopy, and surface modification beyond the diffraction limit. Science, 257, 189±195. Lacoste, T., Huser, T. & Heinzelmann, H. (1997) Faraday-rotation imaging by near-field optical microscopy. Z. Phys. B, Condens. Matter, 104, 183±184. Lacoste, T., Huser, T., Prioli, R. & Heinzelmann, H. (1998) Contrast enhancement using polarization-modulation scanning near-field optical microscopy (PM-SNOM). Ultramicroscopy, 71, 333±340. Meyer, G. & Amer, N.M. (1988) Novel optical approach to atomic force microscopy. Appl. Phys. Lett. 53, 1045±1047. Noell, W., Abraham, M., Ehrfeld, W., Lacher, M. & Mayr, K. (1998) Microfabrication of new sensors for scanning probe microscopy. J. Micromech. Microeng. 8, 111±113. StoÈckle, R.M., Fokas, C., Deckert, V., Zenobi, R., Sick, B., Hecht, B. & Wild, U.P. (1999) High-quality near-field optical probes by tube etching. Appl. Phys. Lett. 75, 160±162. Veerman, J.A., Otter, A.M., Kuipers, L. & van Hulst, N.F. (1998) High definition aperture probes for near-field optical microscopy fabricated by focused ion beam milling. Appl. Phys. Lett. 72, 3115±3117. Veerman, J.A., Garcia-Parajo, M.F., Kuipers, L. & van Hulst, N.F.
NFO MICROSCOPY BASED ON MICROFABRICATED PROBES 11 (1999) Single molecule mapping of the optical field distribution of probes for near-field microscopy. J. Microsc. 194, 477±482. Vollkopf, A., Rudow, O., Leinhos, T., Mihakcea, C. & Oesterschulze, E. (1999) Modified fabrication process for aperture probe cantilevers. J. Microsc. 194, 344±348. Zhou, H., Midha, A., Mills, G., Thoms, S., Murad, S.K. & Weaver, J.M. (1998) Generic scanned-probe microscope sensors by combined micromachining and electron-beam lithography. J. Vac. Sci. Technol. B, 16, 54±58.