Seediscussions,stats,andauthorprofilesforthispublicationat:https://www.researchgate.net/publication/264427608

CriticalFactorsinCantileverNear-FieldScanningOpticalMicroscopy

ARTICLEinIEEESENSORSJOURNAL·SEPTEMBER2014

ImpactFactor:1.76·DOI:10.1109/JSEN.2014.2325817

READS

63

6AUTHORS,INCLUDING:

FrancescoMarinello

UniversityofPadova

52PUBLICATIONS457CITATIONS

SEEPROFILE

PieroSchiavuta

VenetoNanotech

31PUBLICATIONS314CITATIONS

SEEPROFILE

RaffaeleCavalli

UniversityofPadova

62PUBLICATIONS106CITATIONS

SEEPROFILE

AndreaPezzuolo

UniversityofPadova

9PUBLICATIONS7CITATIONS

SEEPROFILE

Allin-textreferencesunderlinedinbluearelinkedtopublicationsonResearchGate,

lettingyouaccessandreadthemimmediately.

Availablefrom:RaffaeleCavalli

Retrievedon:08February2016

3236 IEEE SENSORS JOURNAL, VOL. 14, NO. 9, SEPTEMBER 2014

Critical Factors in Cantilever Near-FieldScanning Optical Microscopy

Francesco Marinello, Piero Schiavuta, Raffaele Cavalli, Andrea Pezzuolo, Simone Carmignato, and Enrico Savio

Abstract— An important technique for high resolution opticalimaging, beyond the diffraction limit, of nanostructured surfacesis aperture near-field scanning optical microscopy (NSOM). Eventhough NSOM has already demonstrated its good performancein a number of different applications, its quantitative appli-cation is still a challenge, due to a number of factors, whichcommonly influence the quality of the measurement output andconsequently extrapolation of quantitative parameters. In thispaper, a systematic study is reported, analyzing the effect ofthe most critical factors in cantilever NSOM measurements,with particular attention to tip geometry and aperture, scanningconfiguration, and scan mode. Investigations have been carriedout on a commercial instrument, in combination with referencestandard for NSOM calibration (as for instance the Fisherpattern) and other samples opportunely produced for this paper.

Index Terms— Error analysis, optical microscopy, scanningprobe microscopy.

I. INTRODUCTION

CONVENTIONAL microscopy is among the most populartechniques for laboratory and industrial investigations,

basically for the possibility of carrying out simple, fast andnon-invasive analysis of different samples. Limitations foroptical microscopy are mainly connected with the possibilityof achieving three dimensional reconstructions of surfacetopography and with the limitations connected with the lateralresolution, due to the diffraction limit.

Considering the visible light (a wavelength λ of 550 nm)propagating in air with the maximum achievable numericalaperture for a magnification lens NA = 0.95, and consideringnegligible other optical aberrations and distortions, the mini-mum achievable resolution is about d ≈ 200 nm, calculatedthrough d = λ/(2 π NA).

Near-field scanning optical microscopy (NSOM) firstly pro-posed in 1984 by Pohl et al. [1], was introduced to partiallyovercome such limitations. Producing high-resolution optical

Manuscript received February 4, 2014; revised March 28, 2014; acceptedMay 3, 2014. Date of publication May 30, 2014; date of current versionJuly 29, 2014. The associate editor coordinating the review of this paper andapproving it for publication was Dr. Shoushun Chen.

F. Marinello, R. Cavalli, and A. Pezzuolo are with the Department of Land,Environment, Agriculture and Forestry, University of Padova, Legnaro (PD)35020, Italy (e-mail francesco.marinello@unipd.it; raffaele.cavalli@unipd.it;andrea.pezzuolo@unipd.it).

P. Schiavuta is with the Veneto Nanotech Società Consortile per Azioni,Venezia 30170, Italy (e-mail: piero.schiavuta@venetonanotech.it).

S. Carmignato is with the Department of Management andEngineering, University of Padova, Vicenza, 36100, Italy (e-mail:simone.carmignato@unipd.it).

E. Savio is with the Department of Industrial Engineering, University ofPadova, Padova 35131, Italy (e-mail: enrico.savio@unipd.it).

Color versions of one or more of the figures in this paper are availableonline at http://ieeexplore.ieee.org.

Digital Object Identifier 10.1109/JSEN.2014.2325817

images and providing information on surface topography withnanometer resolution, NSOM has opened up for importantadvances and initiation of new research branches in opticalsciences. In particular in the last years, overcoming of Abbe’sdiffraction limits has grown the interest and brought importantresults in the characterization of nanoscale optical fields,with important applications in different research domains suchas lifescience (aggregation of molecules, markers) [2], [3],material science (absorbance, transmission, light emission,fluorescent nanoparticles) [4], sensors (plasmonic structures,resonant optical antennas) [5], [6].

Even though NSOM has already demonstrated its good per-formance in a number of different applications, its quantitativeapplication is still difficult, due to a number of factors whichcommonly influence the quality of the measurement outputand consequently extrapolation of quantitative parameters [7].In the present paper a systematic study is reported, analysingeffect of the most important influencing factors, mainly arisingfrom tip geometry and aperture, scanning configuration andscan mode.

II. NSOM CONFIGURATIONS

The quality of near-field scanning optical microscopy mea-surements depend very much on the specific configurationused. For this reason in the present section a review of mostcommon NSOM configurations is reported.

A. NSOM Working Principle

As mentioned, near-field scanning optical microscopyenables non-diffraction limited imaging and spectroscopy ofsurfaces. The resolution limit breaking is allowed by use of asmall aperture between the light source and the sample surface(Fig. 1). An evanescent field emanates from the aperture andis not diffraction limited within the near-field region, i.e.in a region above the aperture with dimensions less thana single wavelength of the incident light. In practice, near-field measurements are performed using an aperture that ismuch smaller than the incident light wavelength and thenscanning the aperture relatively to the sample at a distancemuch smaller than the same wavelength. This is possiblethrough implementation of a sub-micron optical probe havinga small aperture at the tip end, and scanning the sample at avery short distance from such aperture. Within the near-fieldregion evanescent light is not diffraction limited and nanometrespatial resolution is possible.

The scanning of the aperture relatively to the sample is doneusing well established technologies developed for the family

1530-437X © 2014 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

MARINELLO et al.: CRITICAL FACTORS IN CANTILEVER NSOM 3237

Fig. 1. Schematic representation of near-field microscopy principle.

of Scanning Probe Microscopes (SPMs) to which NSOMbelongs. A probe with a sharp tip is brought in close proximityof the surface, where weak interaction force (in the order of10−9 N) act on probe dynamics. The probe is scanned over thesurface, keeping the tip at a constant height or more often at aconstant distance from the surface (constant force mode): dataare then collected in a raster fashion, point by point, profileby profile.

B. Configuration Modes

Since near-field scanning optical microscopy earlyintroduction, several acquisition techniques have beendeveloped and proposed: after the illumination mode [1],the collection mode was introduced [8] and scatteringmode for apertureless NSOM [5]. Main NSOM operationalconfigurations are reported in Fig. 2.

In Fig. 2.a the so called illumination configuration is shown.An aperture probe is used to generate a near-field emission(“illuminates”) on a small area of a sample surface. Theinteraction of the emission (evanescent field) with the sampleresults in scattered light (propagating wave): the signal istransmitted through the sample and is eventually collected inthe far-field by an inverted microscope. In case of samples nottransparent to the used wavelength, the signal can be collectedalso after reflection in the far-field through an opportunelytilted objective positioned aside the generated far-field. Suchconfiguration can be exploited for analysis of light absorbance,or excitation and detection of photoluminescent signals.

The so called collection configuration is reported in Fig. 2.b.In this case the sample is irradiated by far-field as in classicalmicroscopy: the irradiation is brought to the sample surfacethrough an inverted microscope or, in case of a sample nottransparent to the used wavelength, through an optical systemput aside the sample. The evanescent wave generated on thesample surface is picked up by the probe located within thenear-field region. Such measuring technique is commonly usedfor the detection of localized field on the sample allowinganalysis of light absorbance or accumulation, excitation of

photoluminescent signals, excitation and detection of surfaceplasmons.

In Fig. 2.c a recently introduced configuration, the aperture-less NSOM [9], is also reported. In such case, the techniquerelies on confining and enhancing the field generated by anexternal far-field illumination at the apex of a sharply pointedprobe tip. The enhanced near-field signal therefore has to beisolated and extracted from a relatively intense backgrounddue to the far-field scattered radiation, nevertheless resolutionsdown to less than 10 nm can be achieved.

C. Probe Types

Different probes have been proposed for NSOM signalcollection or transmission, as cleaved crystals, glass pipettes,tapered optical fibers, atomic force microscope like cantilevertips. Tapered optical fibers have been widely implementedsince the early introduction of NSOM technique, but morerecently great efforts have been spent in implementing aper-ture or apertureless cantilever like probes. Indeed, this notonly allows exploitation of NSOM measurement with thelarge majority of commercial and laboratory SPMs, but alsomay open the way to achieve higher speed measurements inrelatively large area. Indeed cantilever type probes normallyperform higher Q-factor and higher resonant frequency withthe consequent possibility of higher data acquisition rates andhigher scanning speed.

Other advantages are also introduced by use of such probes,as for instance higher power throughputs, lower thermal dis-tortions and no pulse chirping or depolarization.

D. Interaction Modes

Two additional considerations can be done on NSOMconfigurations, mainly related with the interaction occurringbetween the tip and the surface, and the signal detection.

Firstly, as any SPM, the instrument can be operated in dif-ferent interaction modes: contact, non contact and intermittentcontact, depending on the interaction force:

- in the contact mode, cantilever deflection under scanningreflects repulsive force acting upon the tip;

- the non contact mode works via the principle of amplitudemodulation detection in absence of repulsive forces;

- in the intermittent contact mode an oscillating cantileverensures high resolution imaging, with minimized interactionbetween the tip and the sample surface.

Secondly, as for traditional optical microscopy, NSOMsignal can be processed in two different modes: standardimaging and spectroscopy. In the standard imaging modethe signal is collected and quantified through a detector;conversely in the spectroscopy mode the illuminating beam orthe collected light is spectrally resolved, in order to achieve aspectroscopic information.

E. The Investigated Instrument

For the present work a commercial NSOM was implemented(a WITec alpha300 S, by WITec GmbH). The instrumentoperates using a micro-fabricated sensor mounted at the end

3238 IEEE SENSORS JOURNAL, VOL. 14, NO. 9, SEPTEMBER 2014

Fig. 2. Different NSOM acquisition techniques: (a) illumination and (b) collection configuration (the dotted sketch resembles the reflection configurationfor measurements in case of not transparent samples). In (c) a representation of the apertureless NSOM mode. Number indicate respectively: 1- light source,2- detector, 3- focusing objective, 4- collection objective.

Fig. 3. Photo and schematic representation of the used NSOM (a) and itsaperture probe (b).

of an arm attached in the proximity of the focus position ofa microscope objective. The probe is a silicon cantilever witha hollow metal-coated pyramid, operating as aperture tip: theSNOM aperture is at the apex of the pyramid with a diameterranging from 50 to 100 nm. The silicon cantilevers have anaverage size of 140×700 μm. The laser light used for opticalimaging is focused into (or is collected from) the backside ofthe hollow tip. Due to the wide opening angle of the hollowpyramid (close to 90 degrees) the transmission coefficient isvery high.

Two step motors allow lateral and vertical positioning of themicroscope objective and of the inverted microscope objective,while an attached inertial drive allows separate vertical andlateral repositioning of the probe: overall the system allowsfocusing and alignment of the optical path, with micrometeraccuracy, fully adequate for ideal NSOM working.

The microscope objective has a double role: focuses (or col-lects) the excitation laser beam and also focuses the beam-deflection laser for probe distance control, without interferencefrom the two laser systems. A schematic of the NSOM systemis reported in Fig. 3.

The instrument can be operated either in illumination or incollection mode, the probe can be controlled both in contactand in non contact mode, while signal can be collectedin standard imaging mode and in spectroscopy mode. Theinstrument is coupled with two excitation laser sources, anArgon ion laser (with emission λ modulable between 457 nmand 514 nm) and a Helium-Neon laser (λ = 633 nm).

III. CRITICAL FACTORS

In the following section the most critical factors affectingthe quality of near-field scanning optical microscopes measure-ments and the possibility of extrapolating quantitative data arediscussed. These are mainly related with the instrument con-figuration, the scan mode, the aperture and the tip geometry,and the sample. Factors which have not a strict connectionwith NSOM and are common to the whole family of scanningprobe microscopes, such as scan speed, thermal drift, creepand hysteresis, interaction force, etc. are not directly analysedhere [10], [11].

A. Effect of Configuration Mode

It was described how the two main configurations arenormally used for optical characterization. In one case thelight excites the sample and the signal is collected by theprobe, in the other the probe aperture is used to generate anear-field which “illuminates” the probe. The configuration isoften determined by the specific experiment to be carried out.For instance the collection mode is needed whenever surface

MARINELLO et al.: CRITICAL FACTORS IN CANTILEVER NSOM 3239

Fig. 4. Microscope view of the probe in two different configurations:(a) collection mode, with the probe seen from the top, and the laser(λ = 633 nm) focused from the bottom through an inverted microscope and(b) illumination mode, with the probe imaged from the bottom (through atransparent sample) by the inverted microscope, and the laser confined withinthe tip.

plasmons or accumulation phenomena have to be characterizedin a functionalized surface, and a surface region has to beexcited through a given wavelength.

In general, when implementing cantilever NSOM, thequality of the signal and of the scanned maps is influencedby the specific used configuration, and in particular by theexcitation light path.

Let’s consider the collection mode first with the excitationlight illuminating the sample. As depicted in Fig. 4.a, whilethe light spot (using high magnification power objectives) hasa size of a few squared microns, due to light scattering a largerarea is excited, often in the order of hundreds of squaredmicrons (then larger than the cantilever). In the collectionphase, the cantilever acts as a mask for blocking spurioussignals and light scatter arising from the sample: neverthelessif light signal is coming from outside the probe, this iscollected by the detector concurrently with the signal fromthe tip. The result is a worsening of the signal to noise rate.

This phenomenon was studied using a transparent glassslide (with a 300 μm thickness). A He-Ne laser (wavelengthλ = 514 nm) was focused at different heights, so that the laserspot was ranged from a minimum diameter of 30 μm (in focusposition) up to 300 μm (defocused position), through a 60×magnification objection.

The test was repeated implementing a 20× magnificationobjection to focus the laser light. For each position, the signalto noise rate was checked through the photodetector: resultsare reported in Fig. 5. As shown by the graph, independentlyfrom the used objective, the signal to noise rate worsens as the

Fig. 5. Effect of the spot size diameter on the signal to noise rate, for twodifferent objectives (20× and 60× magnification).

spot size gets larger. A sudden worsen in particular occurs if,due to inappropriate defocusing, the laser spot is larger thanthe cantilever width (140 μm).

A different situation occurs when measurements are per-formed in illumination mode. Indeed in such case the presenceof a large hollow tip allows for confinement of the lightexcitation. As a consequence the presence of disturbance aris-ing from light scattering or spurious reflections is drasticallyminimized, with the final results of highly better signal to noiserates: in the same analysis conditions, the signal to noise rateexhibited values higher than 150.

The implemented configuration plays a role not only forthe quality of the revealed signal, but also for the quantity ofenergy brought to the sample.

Indeed, while in the collection mode the excitation intensityis spread over the focused spot area of the sample, in theillumination mode the whole laser intensity is confined withinthe probe tip. The intensity of the laser sources most oftenimplemented as excitation light are normally in the rangebetween 10−3 and 10−6 W: considering that the tip has avolume of a few hundreds of cubic microns (corresponding toa few nanograms), the relative heating of the tip can be verysignificant.

This is particularly evident when the signal is collected inspectroscopy mode, and an acceptable signal to noise rate isachieved only using very low scan rates (normally a few tenthsof seconds per point).

In such condition the overheating of the tip causes anoverheating of the measured area, and as a consequenceevident defects on the sample surface (plastic deformationor burning). Additionally the overheating of the tip can alsodamage the tip coating (with a consequent loss in signalquality) or induce distortions of the cantilever, with frequentloss of contact or relevant drift phenomena.

To verify this phenomenon, an experimental test was carriedon a polymer blend deposited on a glass slide trough spincoating, in order to obtain a thin film deposition (thicknessless than 10 μm). The polymer blend had a melting temper-ature Tm higher than 200 °C. Repeated NSOM measurementwere done on the same sample position, over a range of10 × 10 μm, with a sampling of 256 lines and 256 points

3240 IEEE SENSORS JOURNAL, VOL. 14, NO. 9, SEPTEMBER 2014

Fig. 6. Four 10 × 10 μm NSOM images, in illumination mode, with differentscan rates, from 0.004 seconds per point (i.e. 10 μm/s, corresponding to about9 minutes for the full scan) down to 0.08 seconds per point (i.e. 0.5 μm/sduring the forward scan 10 μm/s during the backward scan, corresponding toabout 90 minutes for the full scan).

per line. Acquisition was carried on at different scan rates,in contact mode, with speeds during the forward scanningrespectively at 10 μm/s, 2 μm/s, 1 μm/s and 0.5 μm/s. Signalwas revealed in illumination mode, through a photodetector.Results are reported in Fig. 6, where lighter regions correspondto a higher signal revealed by the photodetector. The exampleshow clearly how at slow scan rates, heat accumulated on thetip causes a blurring of the image, due most probably to adistortion of the sample (the two polymer phases are no moreclearly separated). It is worth noting that in contact mode theinteraction force between the probe and the surface is relativelyhigh (much higher than in non-contact mode).

Even though in the case of NSOM such force is distributedover a large interaction area on the tip, friction phenomenashould be taken into account, to avoid both surface and probedistortion. This is substantiated also by an overall reduction ofthe signal at slow scan rates, which is most probably due topolymer contamination attaching to the tip apex, in proximityor in correspondence of the tip aperture (see Fig. 8.c).

A possible solution is of course the choice of lower laserintensities, but this is possible only within certain limits, inorder not to cause a worsening of the signal to noise rate.

B. Effect of Scan Mode

As in the large majority of scanning probe microscopes,cantilever NSOMs can be actuated both in the so calledcontact and non-contact modes. In near-field scanning opticalmicroscopy the scanning mode is particularly important for thecommonly large size of implemented cantilever. The cantileverlarge size is due to two main reasons: firstly the cantileverhas to be big enough to locate the relatively large hollowtip; secondly the cantilever acts as a protecting mask duringscanning, limiting the signal coming from aside (i.e. not fromtip aperture) both in the illumination and collection modes.

Fig. 7. Four 10 × 10 μm NSOM images, taken with an illumination con-figuration in non-contact mode, with different scan rates, from 0.004 secondsper point down to 0.08 seconds per point, as in Fig. 6. Images are renderedwith the same colour scale, as reported on the right.

As a results, NSOM cantilevers have a width normally largerthan 100 μm and a length often larger than 500 μm. Thedrawback of such a large size is an anomalous dynamicalbehaviour when actuated in non-contact mode. Indeed, whenthe probe is vibrated in the close proximity of the surface,noticeable viscous damping forces act along the cantileverbeam and near the tip end: such forces are proportional tothe volume of air pushed in or out during every oscillationcycle, and finally depends on cantilever area.

Images are rendered with the same colour scale, as reportedon the right.

For this reasons very high set points, often larger than 95%of the free oscillation amplitude, have to be set. The set pointand the oscillation amplitude have then to be finely controlled:too low set points can indeed lead to loss of interaction duringscanning. Conversely too high set points can introduce noiseand spurious vibrations during scanning.

From several investigations conducted on different samplesand at different scanning conditions, it was seen that anadvantage in implementation of non-contact scan mode is notjust the higher resolution but the fact that the configurationis less sensitive to temperature variations. In other words,the cyclic movements of air most probably reduce the heataccumulated at the tip apex, allowing better results when scansare to be performed at very low scan rates. An example isreported in Fig. 7, with a clear improvement with respectto the same polymer blend measured at the same scanningconditions, as reported in Fig. 6.

C. Effect of Tip Geometry

The basic concept for a near-field scanning optical micro-scope consists of a sub-wavelength aperture to be scannedover an extremely flat object (see Fig. 1). In practice, mostspecimens to be investigated deviate from such ideal condition,

MARINELLO et al.: CRITICAL FACTORS IN CANTILEVER NSOM 3241

Fig. 8. Scanning electron microscope images of three NSOM probes:(a) a fresh probe with a small aperture of 60 nm; (b) an used probe, withcontaminations and an enlarged aperture size (diameter larger than 100 nm);(c) an obstructed aperture probe.

having not perfectly flat surfaces. As a consequence, there aretwo fundamental physical conditions for high quality NSOMprobes:

1) the tip should terminate in a well-defined aperture ofsub-wavelength size;

2) the probe shape must allow constant and stable access tothe specimen surface.

Unfortunately most times none of the requirements isfulfilled.

1) Aperture Deviations: In the last years one of the mostimportant issues in NSOM research has been the realizationof small aperture size and reproducible tips. Indeed the finaloptical resolution allowed by a NSOM is a function of the

Fig. 9. (a) The simulated sample, resembling three nanoparticles withdiameters respectively of 25 nm, 100 nm and 400 nm. (b) in an idealNSOM scan the nanoparticles are revealed as dark regions where light isnot transmitted: the size of the dark areas ideally has the same size of thenanoparticles.

aperture size. In particular the detected signal is the resultof the actual surface convoluted with the probe aperture: theresult is clearly a distorted mapping, with a dilation effectproportional to the size of the aperture itself. Normally theaperture has a circular shape, but tip wear or contaminationpick up during measurements can deviate the shape of theaperture or sometime even obstruct it, influencing, limiting orimpeding signal detection.

In Fig. 8 three examples are reported: of a fresh probe (in thezoom box of Fig. 8.a, a clean aperture of about 60 nm in diam-eter can be seen), of a used probe (with contaminations lying inthe proximity of the aperture, and enlarged hole size, Fig. 8.b)and an obstructed aperture probe (with contaminations lyingover the aperture, Fig. 8.c). Simulations on effect of differentaperture sizes and geometries were carried out considering asample with nanoparticles of different sizes, respectively of25 nm, 100 nm and 400 nm. Simulations were done applyingconvolution and dilation algorithms, ass implemented in thecommercial software package SPIP (version 4.6.1, [12]).

The test simulated the effect of probes with different aper-tures on the final sensed map, in particular considering:

- 4 different geometries (circular, square, triangular andirregular);

- three different sizes (50 nm, 100 nm and 200 nm maximumaperture size).

Ideally, the signal detected by a NSOM with a probe havingan infinitesimal aperture should be the one shown in Fig. 9.The actual result of NSOM measurement is the one reportedby Table I.

As evidenced by the simulations, the convolution dilatesthe geometries, with an influence which is more evident whensmall details are to be analysed, or when relatively largeapertures are used. A second effect is the blurring of thefeatures edges, more evident not only when apertures arerelatively large, but also when the probe exhibits an irregularaperture shape.

2) Tip Apex Geometry: The use of a probe with a sharp tipand an aperture at the very end of it should guarantee that theaperture is constantly scanned in close proximity to surfacefeatures of varying geometry and height. Unfortunately therelatively large sizes of NSOM apertures prevent the use ofvery sharp tips. A typical situation is the one reported in thethree dimensional rendering of an inverted scan over a NSOMtip (Fig. 10).

The borders of the aperture are normally not regular. Thismean that during scanning the different protrusion of the tip

3242 IEEE SENSORS JOURNAL, VOL. 14, NO. 9, SEPTEMBER 2014

TABLE I

SIMULATIONS ON EFFECT OF DIFFERENT APERTURE SIZES AND GEOMETRIES ON A NSOM MEASUREMENT

Fig. 10. (a) Three-dimensional rendering of a NSOM tip. (b) Close-up viewof the aperture: arrows indicate the presence of peaks in the irregular crown.

apex (indicated by the arrows in Fig. 10.b) can come intointeraction with the sample surface. More specifically, notalways the same protrusion interacts with the surface, butdepending on the sample topography and on scan direction,during the same scan different protrusions can interact withthe surface. This becomes a critical factors mainly for two

reasons: the variation of the distance of the aperture relativelyto the surface and the misalignment between optical andtopographical signal.

In particular with used probes, wear phenomena, dust par-ticle pick-up and presence of contaminations can alter thecrown around the aperture, generating an irregular circularprofile. Peaks and valleys can be easily recognized with asize of several nanometers: sometimes, when severe wearphenomena occur, new asperities and peaks as big as fewtens of nanometers may be generated. The clear consequenceis that during scanning the relative distance between the tipaperture and the surface may vary by an amount rangingfrom few nanometers to few tens of nanometers. The effectof such tip apex modification reflects on the result on thescanning operation. An example is reported in Fig. 11, wherea measurement on sub-micrometer luminescent monodispersesilica nanospheres is proposed [3]. While Fig. 11.a exhibitsa clean signal revealed by nanoparticles, Fig. 11.b clearlyevidences artifacts due to a worn tip. In particular arrows indi-cate presence of anomalous corners and borders inconsistentlydoubled (respectively on the top left and on the top right ofnanoparticles).

Theoretical models agree in describing the exponentialdecay of near-fields with the distance. As reported byC. Girard et al. [13], when the observation point R graduallymoves away from the surface, the signal decays with anexponential decay low proportional to e−Rη, where η−1 ≈200 nm is the typical decay length for optical surface evanes-cent waves.

Accordingly, if presence of peaks causes a variation inrelative position of the aperture, by way of example from 5 nmto 15 nm, the consequent intensity loss is larger than 5%, whileif the distance variation is from 5 nm to 50 nm, the consequentintensity loss results larger than 25%. Changes in relative

MARINELLO et al.: CRITICAL FACTORS IN CANTILEVER NSOM 3243

Fig. 11. NSOM measurement on luminescent monodisperse silicananospheres, by means of a fresh (a) and worn (b) probe.

aperture height not only lead to an intensity reduction: indeed,the near-field confinement of the light emerging the probe islost with increasing distance, with the result of a proportionalloss in lateral resolution.

It has been said that one of the peculiarities of the NSOMis that it can provide high-resolution optical images coupledwith nanometre-scale topography information. This means thatinvestigations of local transmission or reflection properties,fluorescence or luminescence phenomena at sub-wavelengthresolution can be coupled with three-dimensional imaging ofsurface topography.

This is true with some limitations: in fact, the positionof the aperture and the tip apex which actually interactswith the surface are not coincident. The minimum deviationbetween the aperture centre (where ideally the optical signalis collected) and the aperture crown is normally at least a fewtens of nanometers (as big as half aperture). But as underlinedin the previous paragraph, different peaks or asperities comeinto contact with the surface during scanning, due not onlyto the evolution of the tip shape, but also to presence ofpeaks and corrugations on the analysed surface. Deviationsbetween the actual interaction point and the localization of thedetected optical signal can arise up to 100 - 200 nm. Theseare huge values if compared with the expected resolution,and are particularly negative whenever the exact localizationof intensity fields is wanted, as in the study of plasmons or

Fig. 12. (a) Interaction of a NSOM tip with a rough and structured surfaceand (b) with a planar surface.

accumulation phenomena linked to a particular topography,or in the study of biological samples treated with fluorescentmarkers.

D. Effect of Surface Topography

The effect of surface topography deals with phenomenaalready discussed in the previous paragraph, but since theentity of distortions very often is even one order of magnitudehigher than the effect of tip asperities, a different paragraphwas dedicated. Near-field information can be considered com-plete when we deal with idealized plane samples. In realmeasurements, surface present a topography which normallydeviate from ideal planarity: micrometer sized features withsteep slopes and roughness can interact not only with differentpositions of the aperture crown, but also of the tip flank. Thisresults in large variations of the tip height position, whichcan also bring the aperture out of the near-field region. Twoopposite conditions are reported in Fig. 12.

The presence of peaks and features which cause modifica-tions of the height of the aperture relatively to the surfaceshould be in general avoided. From a geometrical point ofview this can be expressed through two general rules.

The characteristic wavelength of the analysed surface willhave to be at least one order of magnitude higher than theaperture size. The Root Mean Square Gradient, Sdq, whichis the RMS value of the surface slope within the samplingarea has to be lower (better if at least one order of magnitudelower) than the tip flank slopes. In those regions wherelocally the Sdq is higher than the tip slope, the achievedoptical signal will be difficultly analysable in a quantitativeway.

This two rules are normally fulfilled whenever low rough-ness (in terms of Sa or Sq) surfaces are to be analysed: but in

3244 IEEE SENSORS JOURNAL, VOL. 14, NO. 9, SEPTEMBER 2014

general this is not a strict condition. Of course it will be a safecondition if such parameters locally keep constant values, sayin an area as big as 2-4 times the aperture size.

IV. CONCLUSION

Near-field scanning optical microscopy is an interestingscanning probe technique for non-destructive mapping of sur-face optical properties. Exploitation of quantitative measure-ments is strictly linked to the control of several critical factorsaffecting the quality and the repeatability of measurements.The systematic study reported in the present work showshow particular attention has to be paid to the tip geometryand aperture: slight roughness of the tip aperture or of themeasured sample can induce large deviations in the acquiredsignal. Experimental tests also evidenced how in generalthe illumination mode can limit scattering disturbances andprovide better signal to noise rates, but its implementation islimited by heating phenomena which income especially whenslow scan operations are needed.

REFERENCES

[1] D. W. Pohl, W. Denk, and M. Lanz, “Optical stethoscopy: Imagerecording with resolution λ/20,” Appl. Phys. Lett., vol. 44, no. 7,pp. 651–653, 1984.

[2] H. K. Park, A. Gokarna, J. P. Hulme, H. G. Park, and B. H. Chung,“Fluorescent nanoscale detection of biotin-streptavidin interaction usingnear-field scanning optical microscopy,” Nanotechnology, vol. 19, no. 23,p. 235103, 2008.

[3] F. Enrichi et al., “Investigation of luminescent dye-doped orrare-earth-doped monodisperse silica nanospheres for DNA microarraylabeling,” Opt. Mater., vol. 32, no. 12, pp. 1652–1658, 2010.

[4] G. Canton, R. Riccò, F. Marinello, S. Carmignato, and F. Enrichi,“Modified Stöber synthesis of highly luminescent dye-doped silicananoparticles,” J. Nanoparticle Res., vol. 13, no. 9, pp. 4349–4356, 2011.

[5] U. C. Fischer and D. W. Pohl, “Observation of single-particle plasmonsby near-field optical microscopy,” Phys. Rev. Lett., vol. 62, no. 4,pp. 458–461, 1989.

[6] H.-H. Lu, T.-C. Hsiao, S.-M. Hsu, and C.-W. Lin, “Opticalcharacterization of a 1-D nanostructure by dark-field microscopy andsurface plasmon resonance to determine biomolecular interactions,”IEEE Sensors J., vol. 10, no. 3, pp. 712–719, Mar. 2010.

[7] G. Longo, M. Girasole, G. Pompeo, R. Generosi, M. Luce, andA. Cricenti, “A multipurpose hybrid conventional/scanning near-fieldoptical microscope for applications in materials science and biology,”Meas. Sci. Technol., vol. 21, no. 4, p. 045502, 2010.

[8] E. Betzig, M. Isaacson, and A. Lewis, “Collection mode near-fieldscanning optical microscopy,” Appl. Phys. Lett., vol. 51, no. 25,pp. 2088–2090, 1987.

[9] H. G. Frey, C. Bolwien, A. Brandenburg, R. Ros, and D. Anselmetti,“Optimized apertureless optical near-field probes with 15 nm opticalresolution,” Nanotechnology, vol. 17, no. 13, p. 3105, 2006.

[10] F. Marinello, S. Carmignato, A. Voltan, E. Savio, andL. De Chiffre, “Error sources in atomic force microscopy fordimensional measurements: Taxonomy and modeling,” J. Manuf.Sci. Eng., vol. 132, no. 3, pp. 031003-1–031003-8, 2010.

[11] F. Marinello and E. Savio, “Use of cylindrical artefacts for AFM verticalcalibration,” Meas. Sci. Technol., vol. 18, no. 2, pp. 462–468, 2007.

[12] SPIP, Scanning Probe Image Processor, Image Metrology A/S,Hørsholm, Denmark [Online]. Available: http://www.imagemet.com(accessed May 2014).

[13] C. Girard, C. Joachim, and S. Gauthier, “The physics of the near-field,”Rep. Progr. Phys., vol. 63, no. 6, pp. 893–938, 2000.

[14] M. M. Rahman and M. L. Norton, “Hierarchical lithography forgenerating molecular testbeds,” IEEE Sensors J., vol. 10, no. 3,pp. 498–502, Mar. 2010.

Francesco Marinello, photograph and biography not available at the time ofpublication.

Piero Schiavuta, photograph and biography not available at the time ofpublication.

Raffaele Cavalli, photograph and biography not available at the time ofpublication.

Andrea Pezzuolo, photograph and biography not available at the time ofpublication.

Simone Carmignato, photograph and biography not available at the time ofpublication.

Enrico Savio, photograph and biography not available at the time ofpublication.