Ultramicroscopy 84 (2000) 127±131

Ultramicroscopy Letter

Modulation imaging in re¯ection-mode near-®eld scanningoptical microscopy

Josef Kerimoa, Markus Bu�chlera, William H. Smyrla,*aCorrosion Research Center, Department of Chemical Engineering and Materials Science, University of Minnesota,

221 Church Street S.E., 112 Amundson Hall, Minneapolis, MN 55455, USA

Received 6 December 1999; received in revised form 14 February 2000

Abstract

A simple implementation of modulation of the near-®eld optical signal of near-®eld probes based on the shear-forcefeedback system is demonstrated in a re¯ection-mode near-®eld optical microscope. The modulation exhibits aderivative type of dependence on the near-®eld signal and no sensitivity to topography. It is shown that the modulationimage can be calculated directly from the derivative of the conventional near-®eld scattering image. This type of near-

®eld modulation is an excellent way to reject far-®eld artifacts from the near-®eld signal. # 2000 Elsevier Science B.V.All rights reserved.

Keywords: NSOM; Derivative image; Near ®eld modulation

Near-®eld scanning optical microscopy(NSOM) is currently being used in high-resolutionoptical imaging of surfaces with resolution beyondthe di�raction limit. NSOM has been shown toyield lateral resolution better than 50 nm byscanning a tip across the surface. The mostcommon mode of operation is to use a straight,tapered ®ber probe with a subwavelength apertureto illuminate the sample surface in the near ®eld. Ashear-force feedback system is used to maintain theprobe in the near-®eld by vibrating the probeparallel to the surface [1]. The near-®eld opticalsignal is then detected with conventional far-®eld

optics to yield an optical image that breaks thedi�raction limit.

In near-®eld experiments, it is important todiscriminate the near-®eld from the far-®eld signal.This is important since the components that arenecessary for high resolution reside in the near-®eld [2]. One mode of near-®eld operation that iscurrently of interest, uses apertureless cantileveredAFM probes to generate optical images that breakthe di�raction limit [327]. In these experimentsboth the sample and the perpendicularly vibratingAFM probe are illuminated with far-®eld opticsand the resulting signal is completely dominatedby the far-®eld signal. As the sample-probedistance is oscillated, the scattered near-®eldoptical signal is modulated and detected at thesame frequency with lock-in ampli®cation, whichis an e�ective way to suppress the far-®eld fromthe near-®eld signal. Such ``modulation'' of the

*Corresponding author. Tel.: +1-612-625-0717; fax: +1-

612-626-7246.

E-mail address: smyr1001@maroon.tc.umn.edu

(W.H. Smyrl).

0304-3991/00/$ - see front matter # 2000 Elsevier Science B.V. All rights reserved.

PII: S 0 3 0 4 - 3 9 9 1 ( 0 0 ) 0 0 0 3 0 - 9

near-®eld signal is shown to yield optical resolu-tion between 10 and 100 nm with infrarednear-®eld light [5,7]. The fundamental contrastmechanism in these experiments is still not under-stood but it seems to depend on the verticaltip2sample separation. In contrast to NSOMexperiments with apertureless tips, it was demon-strated that apertured tips with a vertical modula-tion yield a derivative type of signal [8].

Although the perpendicular modulation isrelatively easy to implement for cantileverednear-®eld probes, no straightforward mechanismexists for modulating the near-®eld signal ofstraight ®ber near-®eld probes. The near-®eldsignal in the experiments is in some casescomplicated by a large far-®eld component. Thisincludes near-®eld experiments with small aperturetips that contain a large amount of defects such as``pinholes'' that are located in the far-®eld andwhich degrade the image quality. Further degra-dation of resolution can occur in near-®eldexperiments with thick samples where imaging isdone in the transmission mode. A relatively largefar-®eld component can contribute to the signaland must be suppressed. Furthermore, near-®eldexperiments with illumination and collectionthrough the same aperture contain a large far-®eldbackground that interferes with the signal andmust be suppressed. In this paper we describe asimilar type of ``modulation'' NSOM but with ashear-force feedback system using straight ®berprobes with subwavelength apertures.

The setup used to perform the experimentsconsists of a re¯ection near-®eld microscope and isshown in Fig. 1. The NSOM tips were producedinhouse from single-mode optical ®bers and coatedwith aluminum to produce a subwavelengthaperture at the end. The production of the straightNSOM tips is similar to the procedure describedelsewhere [9]. The NSOM tip was mounted on apiezotube and vibrated parallel to the samplesurface with a sinusoidal voltage at one of theresonance frequencies. The sinusoidal voltage wastaken directly from the output of a commercialanalog lock-in ampli®er (EG&GModel 5208) andthe resonance frequency of the tip ranged from 15to 90 kHz depending on the mounting of the tip.In these experiments, the sinusoidal voltage was

carefully applied to the piezotube to ensure a smalltip vibration amplitude, which was in the range of10 nm.

In these experiments, less than 0.5 mW of488 nm laser light was coupled into the cleavedend of the NSOM tip for imaging. The sample wasthen illuminated with the near-®eld tip, whilethe tip was vibrating, and the modulated re¯ectedlight was collected with a microscope objective(NA 0.3). The objective was located at about 458with respect to the NSOM tip and the light wasdirected to a photomultiplier tube (PMT, Hama-matsu R3788). The modulated output of the PMT,was taken across a 10 kO resistor, and was sent tothe same lock-in ampli®er for demodulation. Thedemodulated output from the lock-in ampli®erwas used to generate the NSOM optical images.Three types of lock-in output signals were used togenerate the NSOM images: ``R'' which is propor-tional to the signal amplitude, ``y'' which is thephase di�erence between the signal and the lock-inreference, and ``X '' which is related to theprevious quantities through X � R cos y.

A sample with a known structure was used toperform the initial experiments. The shearforce-modulation NSOM imaging was carried out on ahighly polished alumina sample with an embedded

Fig. 1. Diagram of the modulation setup and the re¯ection-

mode near-®eld scanning optical microscope (NSOM).

J. Kerimo et al. / Ultramicroscopy 84 (2000) 127±131128

NiFe pattern, see Fig. 1. The sample was ®rstpolished to less than 2 nm roughness and then aprotective �1 nm thick overlayer of diamond-likecarbon was applied onto the surface leaving thesurface roughness at about 1 nm rms. The NiFepattern consists of a 2 mm wide band and is about2 nm tall and was imaged with the NSOM. TheNiFe band is very re¯ective relative to the aluminamatrix and has almost no topography. The samplewas ideal for testing the shearforce-modulationimaging since it contained a very simple structurewith almost no topography. Thus, the shearforce-modulation NSOM images shown below were freefrom complications of topographical artifacts.

The NSOM images of the NiFe alloy band areshown in Fig. 2. The topographical image in (a)was collected concurrently with the optical imagein (b) while the remaining images were collectedsubsequently on the same region. A small drift inthe scanner caused the subsequent images to havea slight change of alignment, but otherwise allimages show the same regions. The direction ofvibration of the NSOM tip is shown in (a) and isapproximately perpendicular to the NiFe band.

The topography image in (a) shows the 2 mmwide NiFe band and numerous contaminants onthe surface that were di�cult to remove. The NiFeband is not distinct in the topography since it is

Fig. 2. NSOM images of the NiFe alloy structure in alumina: (a) shear-force topography, (b) re¯ection NSOM, (c) ``X''-modulation

NSOM, (d) calculated derivative of (b).

J. Kerimo et al. / Ultramicroscopy 84 (2000) 127±131 129

only about 2 nm in height and the surface hasabout 1 nm rms roughness. The roughness of thesample is mainly due to the polishing marks whichare clearly visible. The polishing marks are seen torun from the top right to the bottom left corner ofthe image. In contrast, the corresponding NSOMoptical image (b) contains a clearly visible NiFeband and appears brighter than the surroundingalumina. This image was taken without the lock-indemodulation and simply represents the near-®eldre¯ection signal. The intensity of the re¯ectedsignal is not constant across the NiFe band andadditional fringes are visible at the edges of theNiFe band. The fringes are not visible when thesample is viewed with a far-®eld confocal micro-scope. The NiFe band also appeared uniformwhen imaged with a scanning confocal micro-scope. Apparently, the fringes are caused by thenear-®eld imaging. We have not attempted tostudy the origin of the fringes further but they maybe due to some kind of near-®eld interference or``shadowing'' e�ect of the collection optics by thetip. Similar structures have also been observed byother workers [10,11].

The image corresponding to the demodulated``X '' lock-in output signal is shown in (c). Thefringes and the polishing marks are now morevisible. Interestingly, the mechanism that causesthe fringes also a�ects the near-®eld modulationsignal. Additional ``dark'' and ``bright'' bands arealso visible in the image. The additional ``dark''and ``bright'' bands are shown to be caused by thesensitivity of the modulation to only the``derivative'' of the re¯ection signal. In fact, thederivative of the re¯ection image (b) was calcu-lated and is shown in (d). The derivative of theimage was taken parallel to the direction of the tipvibration, as is indicated in (a). It is clear that allthe bands in the modulation image are reproducedin the calculated derivative image. The calculatedderivative image is more noisy since it is moresensitive to high frequency noise from the originalimage.

To summarize, the shear-force feedback canmodulate the near-®eld signal and the resultingsignal is sensitive to the derivative only. Hence, themodulation image obtained with the lateralmodulation of apertured tips is in agreement with

that reported for vertical modulation [8]. Con-siderable insight into the origin of the modulationsignal and tip2sample interaction for verticalmodulation is also provided by these authors.Furthermore, more complicated structures, includ-ing samples with a topography, were also imagedwith the shear-force modulation. These images doshow a similar behavior but with a more compli-cated derivative signal. The derivative dependson the near-®eld optical signal and orientation ofthe tip vibration and is not a�ected by thetopography.

In conclusion, we have shown that shear-forceimaging can be used to modulate the re¯ectednear-®eld signal to yield a derivative type signal.The derivative depends on the lateral direction ofthe tip vibration and corresponds to the ``X '' lock-in output signal. In particular, the procedureprovides a means of rejecting the backgroundlight re¯ected in the taper of the optical ®ber innear-®eld experiments where the illumination andcollection is achieved through the same probe.Such near-®eld experiments are better suited forre¯ection-NSOM imaging of opaque samplesbecause the experiments are less prone to artifactsintroduced by near-®eld imaging, such as thefringes observed here [10,11]. Furthermore, theuse of uncoated tips in these experiments isreported to yield resolutions as good as 70 nm[12]. This is expected to be additionally increasedusing metal coated ®bers. Coated ®bers, however,have large amounts of light internally re¯ectedfrom the taper back to the detector. It is expectedthat the modulation technique presented inthis paper would allow the near-®eld opticalsignal to be discriminated from such re¯ectedlight. We are currently pursuing this modulationtechnique further in experiments where illumina-tion and collection are done through the sameprobe.

Acknowledgements

This work has been supported by NSF/DMR-9816404 and by Seagate Recording Heads. FrancisGuillaume carried out the confocal imaging of thesamples.

J. Kerimo et al. / Ultramicroscopy 84 (2000) 127±131130

References

[1] E. Betzig, P.L. Finn, J.S. Weiner, Appl. Phys. Lett. 60

(1992) 2484.

[2] D. Courjon, J.-M. Vigoureux, M. Spajer, K. Sarayeddine,

S. Leblanc, Appl. Opt. 29 (1990) 3734.

[3] H.F. Harmann, A. Gallagher, D.J. Nesbitt, Appl. Phys.

Lett. 73 (1998) 1469.

[4] F. Zenhausern, Y. Martin, H.K. Wickramasinge, Science

269 (1995) 1083.

[5] H. Sasaki, Y. Sasaki, J. Appl. Phys. 85 (1999) 2026.

[6] T.C. McGill, P.M. Bridger, Opt. Lett. 24 (1999) 1005.

[7] B. Knoll, F. Keilmann, Nature 399 (1999) 134.

[8] M.A. Paesler, P.J. Moyer, Near-Field Optics, Wiley, New

York, 1996.

[9] D.A. Higgins, J. Kerimo, D.A. Vandenbout, P.F. Barbara,

J. Am. Chem. Soc. 118 (1996) 4049.

[10] D.W. Weston, S.K. Buratto, J. Phys. Chem. 101 B (1997)

5684.

[11] J.A. Cline, M. Isaacson, Appl. Optics 34 (1995) 4869.

[12] G. Krausch, S. Wegscheider, A. Kirsch, H. Bielefeldt, J.C.

Meiners, J. Mlynek, Opt. Commun. 119 (1995) 283.

J. Kerimo et al. / Ultramicroscopy 84 (2000) 127±131 131