Sl

1

Chemically etched fiber tips for near-fieldoptical microscopy: a process for smoother tips

Patrick Lambelet, Abdeljalil Sayah, Michael Pfeffer, Claude Philipona, andFabienne Marquis-Weible

An improved method for producing fiber tips for scanning near-field optical microscopy is presented.The improvement consists of chemically etching quartz optical fibers through their acrylate jacket. Thisnew method is compared with the previous one in which bare fibers were etched. With the new processthe meniscus formed by the acid along the fiber does not move during etching, leading to a much smoothersurface of the tip cone. Subsequent metallization is thus improved, resulting in better coverage of thetip with an aluminum opaque layer. Our results show that leakage can be avoided along the cone, andlight transmission through the tip is spatially limited to an optical aperture of a 100-nm dimension.© 1998 Optical Society of America

OCIS codes: 050.1220, 060.2370, 180.5810, 240.6700, 350.5730.

8

fwct

Cpmttf

In scanning near-field optical microscopy ~SNOM! asharp probe tip of submicrometer dimension picks upoptical information in the near field by sending, col-lecting, or diffracting light at the surface of a sample.1The near field that is the source of subwavelengthoptical information is influenced as much by theshape and optical properties of the sample surface asit is by the tip itself.2 Reliable results thus criticallydepend on the ability to work with well-defined tipswhose geometrical and optical properties are charac-terized and controlled.

Most of the SNOM probes used today are based ontapered optical fibers. Tips are obtained either byheating and pulling3,4 or by chemical etching5,6,7 inaqueous solution of fluorhydric acid ~HF!, producing aconical tip. To yield a subwavelength optical probe,the conical tip is coated with an aluminum layer,leaving a nanometer-size aperture at the tip apex.The ideal tip is characterized by high optical trans-mission through a single hole with dimension of a fewtens of nanometers. The optical resolution is of the

When this research was performed, the authors were with theInstitut d’Optique Appliquee, Ecole Polytechnique Federale, 1015Lausanne, Switzerland. A. Sayah is now with the Institut deMicrosystemes, Ecole Polytechnique Federale, 1015 Lausanne,

witzerland. The e-mail address for P. Lambelet is patrick.ambelet@epfl.ch.

Received 20 April 1998; revised manuscript received 3 August998.0003-6935y98y317289-04$15.00y0© 1998 Optical Society of America

order of the size of the aperture. Furthermore,sharp tips characterized by a small radius of curva-ture are more appropriate, since they allow the tip tocome quite close to the sample surface when thissurface is not perfectly flat.

The heating and pulling process produces long tips~1 mm!, with smooth surfaces, allowing good and uni-orm metallization with aluminum and leading toell-defined apertures. Compromises between opti-

al transmission and aperture diameter lead, for thatype of tip, to typical transmissions of 1028–1024 for

aperture sizes between 30 and 100 nm.4 The ratherlow transmission is due to the strong attenuation ofthe light along the taper as well as to the very smallcone angle of the tip ~typically a few degrees!.

hemical etching, on the other hand, allows one toroduce fiber tips with much shorter cones ~;200m! and thus much larger cone angles. This leads

o higher transmission, with the light being guided inhe intact core of the fiber down to a few micrometersrom the tip apex. Typical transmissions of 1023

have been reported.9 The main drawback of chem-ical etching is the roughness of the surface obtainedafter etching with HF, which decreases the quality ofthe aluminum coating.10 Such tips often show leak-age of light along the conical taper. In this paper wepropose a new etching technique that allows one toproduce much smoother surfaces and thus tips ofhigher quality. Applying this new technique, wehave been able to obtain tips with a smooth metalcoating and no leakage along the taper. The hightransmission ratio of chemically etched tips can then

1 November 1998 y Vol. 37, No. 31 y APPLIED OPTICS 7289

tnu@tpswt1itbttpesrtaifi

ls

7

be combined with some of the advantages of pulledtips, such as smooth cone surfaces and opaque metalcoatings.

The new mechanism for fabrication of the tips isbased on the chemical etching of a glass fiber throughits acrylate jacket, as opposed to the standard chem-ical etching that is usually performed on bare fibersafter removal of the jacket. The etching setup isshown in Fig. 1. An optical fiber, monomode at 633nm ~FS-SN-3221 from 3M!, is dipped with its acrylatejacket into an aqueous 40% HF solution. The HFsolution is covered by oil to protect the fiber againstacid vapor and is heated to 60 °C to accelerate theprocedure. The acid does not dissolve the acrylatejacket but rather diffuses through it to etch thequartz fiber. After 35 min the fiber is removed andrinsed successively with water, trichlorethylene, andacetone. At this point a tip has been formed insidethe acrylate jacket. To remove the jacket, an inci-sion is made a few millimeters above the tip, and thejacket, softened by acetone, can then be pulled bybeing seized in front of the tip. The jacket is thusremoved without damage to the fiber tip after the tipis formed.

This method is similar to a previous technique usedto make fiber tips by chemical etching.7 However,although the mechanism of chemical etching of theglass is the same, etching through the jacket leads toa different tip formation process. When etching isperformed on a bare fiber, the meniscus formed by theacid along the fiber is determined by the acid–oil–quartz interface. As the fiber is etched, its diameterdecreases and following the laws of superficial ten-sion; the meniscus height decreases until the tip isfully formed. In the present method acid does notetch the jacket but rather diffuses through it to etchthe glass. The meniscus height is determined by theacid–oil–acrylate interface and remains constantduring tip formation. The formation of the tip isthus essentially due to a diffusion process. As theions of acid react with the quartz, new ions have todiffuse from the liquid to the surface of the quartz.In the upper part of the meniscus the layer of liquidis very thin, and thus the ions saturate more rapidly,

Fig. 1. An optical fiber is dipped, with its acrylate jacket, inaqueous hydrofluorydric acid covered by oil as a protective layer.The acid diffuses through the acrylate and etches the quartz fiber.After 35 min a sharp tip is formed.

290 APPLIED OPTICS y Vol. 37, No. 31 y 1 November 1998

slowing down the etching rate. In the deeper liquidthe concentration of ions close to the fiber remainshigher owing to more efficient diffusion from thelarge reservoir. This explains why a tip is formed,although the height of the meniscus does not change.Note that the diffusion through the acrylate is veryefficient. With the same acid concentration andtemperature, forming a tip across the acrylate jackettakes only 5 min longer than on bare fibers.

In Fig. 2 scanning electron microscopy with a fieldemission electron microscope ~JEOL, F 6300! is usedo give an image of a tip obtained by this new tech-ique @Fig. 2~a!# and to compare it with a tip producednder the same conditions but by etching a bare fiberFig. 2~b!#. These figures show that the surface ofhe tip etched through the jacket is very smooth ~com-arable to pulled fibers! and does not suffer from theurface irregularities clearly visible in Fig. 2~b!,hich are usually encountered in chemically etched

ips. Here the radius of curvature of the tip apex is5 nm in Fig. 2~a! and 30 nm in Fig. 2~b!. Thismprovement of the quality of the surface shows thathe movement of the meniscus along the tip when aare fiber is etched is the main effect responsible forhe irregularities observed on the surface. Keepinghe jacket on the fiber stabilizes the meniscus androduces a smoother surface, with etching being gov-rned essentially by diffusion. Although the globalhape of the tip, viewed on a larger scale, is symmet-ic, asymmetries can occur in the last few microme-ers of the tip in the form of elongated depressionslong the axial direction. This effect, as can be seenn Figs. 2~a! and 2~b!, however, is less pronounced inbers etched through the jacket.The fiber tips are metallized with a 100-nm-thick

ayer of aluminum ~deposited at 10 nmys, at a pres-ure P , 5 3 1026 mbar!. Axial views of the tip apex

after metallization are displayed in Figs. 2~c! and2~d!. A smoother surface is observed on the tipetched through the jacket, although aluminum grains

Fig. 2. SNOM tips obtained by chemical etching of ~a! a fiber withits acrylate jacket, ~b! a bare fiber; ~c! and ~d!, corresponding axialviews after metallization.

nIebltsm1

t

w

are still present. The aperture is not clearly definedfor either type of tip. We attribute the difficulty ofobtaining clear circular apertures in the metal layerto the very small radius of curvature ~typically 15

m! characterizing the quartz tip of etched fibers.n contrast, pulled tips are characterized by a flatnd, for which much better defined apertures haveeen observed. We measured a transmission ofog~T! 5 23.3 6 0.8 ~26 samples! for the tips etchedhrough the jacket and log~T! 5 22.8 6 0.8 ~13amples! for the tips made from bare fibers. Thiseans that the transmission varies between 8 3

025 , T , 3 3 1022 for tips made with the newetching process and 2 3 1024 , T , 1022 for theother kind of tip. The higher transmission of thelatter tips is attributed to leakage of light along thecone resulting from the greater roughness of thesurface. To check the effect of roughness, we com-pletely closed the aperture at the tip apex by suc-cessively evaporating the aluminum from the sideand from the front of the tip. The transmission ofthe tips etched with the jacket decreased to ourdetection limit of Tclosed 5 1027, but the transmis-sion of the other tips decreased only by a factor of 10and was still Tclosed 5 7 3 1024. This result showsthe importance of the quality of the surface of thequartz tip for guaranteeing efficient metallizationwithout leakage of light along the cone of the tip.Note that quantitative information on the trans-mission of the tip is not sufficient to determine itsquality; information on leakage along the cone andaperture size are important parameters as well.

To further analyze light leakage along the cone, wemeasured the angular emission of both types of tip inthe far field. Figure 3 displays this far-field profilefor tips with a 100-nm-thick coating of aluminum,obtained with a water-immersion microscope objec-tive ~numerical aperture NA 5 1.2! by placing the tipapex at the object focal plane of the objective andimaging its back pupil on a CCD camera. Figure3~a! corresponds to the fiber etched with the jacketand shows a uniform and wide light distribution,which is the fingerprint of a single, small ~,300 nm!aperture. The low-contrast features appearing onthis image are due to leaking light along the cone butwith a total power much lower than the light emittedby the small aperture. On the other hand, Fig. 3~b!,

Fig. 3. Angular light distribution from a tip chemically etched ~a!ith the acrylate jacket, ~b! without the acrylate jacket. The scale

is in numerical apertures, NA.

which corresponds to etching a bare fiber, shows astructure containing many speckles, which is due es-sentially to the light leaking along the cone of the tip.Note that this far-field pattern is highly sensitive toleaking light because the interference contrast of theleaking light with the light coming from the tip de-pends on their relative total power, not on their ac-tual intensity at the tip surface. Fifty percent of thefibers etched with the acrylate jacket show a far-fielddistribution similar to Fig. 3~a!, but none of the fibersetched without the jacket can produce such a uniformprofile. These results show that the problem of lightleakage encountered until now with most etched fi-bers can be solved by etching through the jacket.The resulting smoother surface of the tip allows oneto obtain better aluminum coverage without leakage.This observation also shows that the difficulty en-countered until now in obtaining a good coating onchemically etched tips in comparison to that onpulled tips is due not to a different chemical state ofthe surface but to a higher surface roughness.

The problem of light leakage along the cone of thetip is not necessarily critical for imaging in the near-field, since this leaking light has a poor lateral con-finement and just adds a background to the near-fieldimage. But for applications based on photochemicalreactions11 this leaking light must absolutely beavoided, since it induces a reaction far from the tar-geted positions. For such applications it is critical tohave only one optical aperture on the SNOM tip.

To test the performance of the tip in a near-fieldoptical measurement, a transmission SNOM image ofa 1:1 chromium grating ~period 372 nm, chromiumhickness 21 nm! deposited on glass is taken with a

tip etched through the acrylate jacket. Figure 4shows a cut through such an image. The profile ofthe light transmission shows a contrast of h 5 0.28.Measured on the same grating with four tips of eachtype, the contrast in the optical profile is improved bya factor of 2 ~hjacket 5 0.17 6 0.1 versus hbare 5 0.08 6

Fig. 4. Near-field line scan of a 1:1 chromium grating of period372 nm. Solid curve, topography; dotted curve, near-field opticaltransmission. Optical resolution, Dx 5 100 6 10 nm.

1 November 1998 y Vol. 37, No. 31 y APPLIED OPTICS 7291

2. B. Hecht, H. Bielefeldt, Y. Inouye, D. W. Pohl, and L. Novotny,

7

0.07! for tips etched through the jacket. This differ-ence can be explained by the more prominent back-ground signal resulting from light leakage alongimperfectly coated tips obtained by etching bare fi-bers. From the slope of the sides of the grating, inFig. 4, a resolution of Dx 5 100 6 10 nm is measured~10%–90% of the intensity!.

In conclusion, we have presented a modified etch-ing technique for producing SNOM tips. The fiber isdipped with its acrylate jacket in an HF solutioncovered by a protective oil layer. In this way theformation of the tip is governed only by diffusion andresults in smoother tips compared with those onchemically etched bare fibers. This smoother sur-face, comparable with the surface of a pulled fiber,allows one to obtain a high-quality aluminum coatingon the tip, leading to minimum light leakage alongthe cone.

The authors are grateful to B. Senior ~Centre In-terdepartmental de Microscopie Electronique, EcolePolytechnique Federal Lausanne! for taking the elec-tron micrographs and to M. Gale ~Centre Suissed’Electronique et de Microtechnologie SA, Zurich! forfabricating the test grating. Financial support fromthe Swiss National Fund for Scientific Research isacknowledged.

References1. M. A. Paesler and P. J. Moyer, eds. Near Field Optics, ~J. Wiley,

New York, 1996!.

292 APPLIED OPTICS y Vol. 37, No. 31 y 1 November 1998

“Facts and artifacts in near-field optical microscopy,” J. Appl.Phys. 81, 2492–2498, ~1997!.

3. E. Betzig and J. K. Trautman, “Near-field optics: microscopy,spectroscopy, and surface modification beyond the diffractionlimit,” Science 257, 189–195 ~1992!.

4. G. A. Valaskovic, M. Holton, and G. H. Morrison, “Parametercontrol, characterization, and optimization in the fabrication ofoptical-fiber near-field probes,” Appl. Opt. 34, 1215–1228 ~1995!.

5. D. R. Turner, “Etch procedure for optical fibers,” U.S. Patent.4,469,554 ~5 April 1983!.

6. T. Pangaribuan, K. Yamada, S. Jiang, H. Ohsawa, and M.Ohtsu, “Reproducible fabrication technique of nanometric tipdiameter fiber probe for photon scanning tunneling micro-scope,” Jpn. J. Appl. Phys. 31, L1302–L1304 ~1992!.

7. P. Hoffman, B. Dutoit, and R.-P. Salathe, “Comparison of me-chanically drawn and protection layer chemically etchedoptical-fiber tips,” Ultramicroscopy 61, 165–170 ~1995!.

8. L. Novotny, D. W. Pohl, and B. Hect, “Scanning near-fieldoptical probe with ultrasmall spot size,” Opt. Lett. 20, 970–971~1995!.

9. D. Zeisel, S. Nettesheim, B. Dutoit, and R. Zenobi, “Pulsedlaser-induced desorption and optical imaging on a nanometerscale with scanning near-field microscopy using chemicallyetched fiber tips,” Appl. Phys. Lett. 68, 2491–2492 ~1996!.

10. A. Sayah, C. Philipona, P. Lambelet, M. Pfeffer, and F.Marquis-Weible, “Fiber tips for scanning near-field optical mi-croscopy fabricated by normal and reverse etching,” Ultrami-croscopy 71, 59–63 ~1998!.

11. J. Aizenberg, J. A. Rogers, K. E. Paul, and G. M. Whitesides,“Imaging the irradiance distribution in the optical near-field,”Appl. Phys. Lett. 71, 3773–3775 ~1997!.