Single Defect Center Scanning Near-Field Optical Microscopy onGrapheneJulia Tisler,* Thomas Oeckinghaus, Rainer J. Stohr, Roman Kolesov, Rolf Reuter, Friedemann Reinhard,and Jorg Wrachtrup

3. Institute of Physics, Stuttgart University, 70550 Stuttgart, Germany

ABSTRACT: We present a scanning-probe microscope based on an atomic-size emitter, asingle nitrogen-vacancy center in a nanodiamond. We employ this tool to quantitativelymap the near-field coupling between the NV center and a flake of graphene in threedimensions with nanoscale resolution. Further we demonstrate universal energy transferdistance scaling between a point-like atomic emitter and a two-dimensional acceptor. Ourstudy paves the way toward a versatile single emitter scanning microscope, which couldimage and excite molecular-scale light fields in photonic nanostructures or singlefluorescent molecules.

KEYWORDS: Scanning near-field optical microscopy, scanning probe, FRET, graphene, nitrogen-vacancy center, nanodiamond

Imaging the optical near field of nanosized structures is offundamental interest to various areas of photonic, material,

and biological sciences. This need has driven the developmentof several versatile near-field imaging methods, based oncathodoluminescence1 or scanning near-field optical micros-copy,2 which have evolved into a routine technique with aresolution in the 10 nm range. A near field probe made from asingle quantum emitter could enable qualitatively newmeasurements. Simultaneous measurements of its fluorescencelifetime and intensity would allow quantifying couplingefficiencies and near-field intensities,3 and its atomic sizewould push resolution by an order of magnitude. This wouldenable imaging even of extremely short-ranged interactionssuch as fluorescence resonance energy transfer (FRET), anonradiative dipole−dipole interaction between transitiondipole moments of a donor and an acceptor, which typicallyhas a range of only a few nm. To achieve this elusive goal,different donors and acceptor systems have been used likesingle dye molecules4−7 or quantum dots,8,9 but most of thesesystems suffer from blinking or limited photostability, whichhas rendered their use cumbersome in practice and so farprecluded the stable and quantitative observation of FRET witha scanning probe microscope.10 The only remedy to theseproblems so far has been operation at low temperature wheresingle molecular fluorescence usually is stable.11 Recently,however, the nitrogen-vacancy (NV) center, an atom-likephoton emitter in diamond, has emerged as a candidate for amore robust probe for near-field microscopy. It proves to bephotostable even at room temperature,12 can be brought intonm proximity of any photonic system,13,14 and has beenproposed as a stable emitter for near-field microscopy.15 As aresult, FRET has been demonstrated between a stationary NVand a nearby single molecule with high transfer efficiency.16,17

Here, we demonstrate a scanning probe near field micro-scope based on NV centers and use this tool to investigate theFRET interaction of a single NV emitter with graphene.18−21

We demonstrate that a scanning NV center can be reliablynavigated into the short-range of the FRET interaction (∼15nm) and image FRET efficiency in all three dimensions with aresolution on the order of the FRET interaction range. Thisresult opens a new way to probe and image relevant materialproperties like the local Fermi energy, which are predicted tomodify the FRET intensity. It could equally yield a flexiblesource to deterministically inject plasmons at specific locationsinto graphene nanostructures.All experiments use a home-built scanning FRET microscope

based on an atomic force microscope (AFM) with an opticallyaccessible tip. Single NV centers inside nanodiamonds on thetip of the AFM are used as FRET donors. A laser beam excitingthe NV center is focused through the sample onto the NV.Fluorescence of the defect center is collected through the samechannel (see Figure 1). To allow for close proximity betweenthe defect and graphene, small nanodiamonds with diametersaround ∼25 nm containing single NV centers were used. Whenthe NV center is placed in close proximity to the graphenesample, Forster energy transfer quenches its fluorescence with amuch higher efficiency22 than the 2.3% absorption observed fora graphene monolayer in the far field.23 In this process, anelectronic excitation of the NV center is nonradiativelytransferred into an exciton in graphene (Figure 1b), whichquickly dissipates excitation energy mostly by internalradiationless decay.24 Occurring at a rate γnr, this process

Received: March 28, 2013Revised: May 28, 2013Published: June 24, 2013

Letter

pubs.acs.org/NanoLett

© 2013 American Chemical Society 3152 dx.doi.org/10.1021/nl401129m | Nano Lett. 2013, 13, 3152−3156

competes with radiative emission of the NV center and reducesthe fluorescence intensity to a value

γ γγ γ

=+

I ex r

r nr

where γex and γr are the excitation rate and radiative rate,respectively, and we assume γex to be much below saturation.The quenching rate γnr can be computed from Forster’s law25

with the additional assumption that a sheet of graphene can beapproximated by a two-dimensional array of infinitesimalflakes.26

∬

∬μ

γ μ

μ μ

= | |

=

EA s

A sf

r

r

d

d( , )

p

eg

nr graphene

sheet

2 2g

2

graphene

sheet

2 eg2

g2

6(1)

Here, A is a proportionality constant and Ep = E − (ez·E)ezdenotes the in-plane component of E = (3r(μeg·r) − μeg)/(4πε0r

3), the near-field of the transition dipole μeg = −e⟨e|r|g⟩ ofthe NV center’s optical transition between states |g⟩ and |e⟩(red shade in Figure 1b). The graphene transition dipolemoment μg is a scalar, reflecting the fact that graphene is anisotropic material. Hence, exciton transitions can be driven byany in-plane driving field Ep = Eep, regardless of its polarizationep . Precisely, ds

2μg2 = e2∑ki,kf,ω=vF(kf+ki)|⟨kf |x|ki⟩| where |ki⟩,|kf⟩

denote plane-wave electron states in graphene and ω is the NVtransition’s frequency.Integrating eq 1 over an infinitely large graphene surface

yields a modified Forster type law

γ γ=zznr r0

4

4 (2)

with a Forster distance z0 and a quenching rate rising with thefourth power of distance, differing markedly from the z−6 lawcommonly known for point-like objects such as molecules or

atoms. This distance dependence agrees with more elaboratetheories, provided that z is restricted to the near-field range,where the distance to graphene is larger than the lattice spacing,but still much smaller than the optical wavelength.27,28

Recently, it has been confirmed experimentally using quenchingof fluorescent molecules separated from graphene by variousspacers.29

Quantitatively, the Forster distance can be calculated from eq2 and the expressions of γnr and γr stated in previous work19 tobe

ω= ℏ

ℏ ϵ=z

e c94.512( )

15.3 nm0

2 3 3

40

4

(3)

In the following, we discuss our scanning FRET experimentson layers of different graphene thickness. We acquired imagesin the lateral (xy) dimensions (Figure 2c) as well as in a vertical(xz) plane (Figure 2b). For the latter measurement, the AFMwas operated in AC (tapping) mode, where the cantilevercontinuously oscillates and repeatedly touches the samplesurface. A three-dimensional image was reconstructed bycorrelating the emission times of the recorded fluorescencephotons with the phase of the oscillating cantilever.30 Tocompare eq 1 with results from Figure 2c, we transform thefluorescence I(x,y,z) observed in measurements into aquenching rate γnr(x,y,z) by the relation

γ γ=−

−⎛⎝⎜

⎞⎠⎟

I II x y z( , , )

1nr0 B

r(4)

where I0 is the unquenched NV fluorescence intensity and IBthe fluorescence of the background. In all the following, wederive a radiative rate γr = 1/τNV based on the measuredfluorescence lifetime τNV = 8.5 ns of the NV on the tip. Notethat this conversion implicitly assumes unity quantumefficiency.In the case of Figure 2, the monolayer quenching rate was

78.4 × 106 s−1 for a double layer 101.2 × 106 s−1, and for severallayers, 185.9 × 106 s−1.Thanks to their atomic size, scanning single emitters are able

to map near-field couplings with molecular (nm) resolution inall three dimensions. As a first application of this remarkableproperty, we experimentally confirmed the theoretical model ofeq 1. The result is shown in Figure 3. In the lateral dimensions(Figure 3a−c), a high-resolution scan of a graphene edge(Figure 3b, methods) reveals that fluorescence drops smoothlywhen the NV center is moved over the flake, with 68% of thedecay occurring within a length of ∼24 nm centered on theedge. This resolution of our imaging method is limited by thevertical separation between the NV center and the grapheneflake. This distance is nonzero, even though the image wasacquired in contact mode since the NV-center is located insidethe nanodiamond at some residual height above its surface.We quantitatively extract this distance by fitting the observed

profile to theory (solid line in Figure 3c). This latter curve wasobtained by numerically integrating eq 1, taking into accountthe fact that the NV transition has two orthogonal transitiondipoles μ1 and μ2

γ γ γ= +μ μx x x( ) ( ( ) ( ))/21 2nr nr nr

∫ ∫γ μ= | |μ μ

−∞

∞

−∞

∞Ex A x y x y( ) d d ( , )z

pnr g2 , 2NV

Figure 1. Schematic representation of the experimental arrangements(a). Nanodiamonds are glued onto the apex of a Si tip. A highnumerical aperture objective (NA 1.35) is used to excite a single NV inthe nanodiamond and collect NV fluorescence through the graphenelayer. (b) The excitation of the NV center’s atomic dipole is convertedinto an exciton in graphene. Upper half: Sketch of the dipole’s nearfield. Lower half: Energy levels of the lowest optical transition relevantfor energy transfer and graphene band structure near the K point.

Nano Letters Letter

dx.doi.org/10.1021/nl401129m | Nano Lett. 2013, 13, 3152−31563153

The remaining free parameters A, μg2, and zNV are fit to the

data, and the dipole orientation is chosen as μ1∥ez,μ2∥ex to bestfit the observed behavior. The achieved resolution is limited bythe spatial extent of the near-field Ep, which is of the order ofthe NV−graphene distance. Reverting the argument, we canextract this distance from the fit, finding that zNV = 15.9 ± 0.4nm. As discussed above, we interpret this value as the distancebetween the NV center and the surface of the nanodiamond.We also measured the vertical dependence γnr(z) by

repeatedly approaching and retracting the tip on a largegraphene surface (Figure 3d). At a tip−sample spacing of about8 nm (corresponding to zNV = 24 nm), the tip rapidly snaps tothe surface, so that no data could be acquired between thispoint and contact (15 nm). As for Figure 3c, we find that thedata is well described by eq 1 and in particular agrees with thepredicted z−4 law of eq 2. Using the value zNV obtained fromthe lateral scan, we can quantitatively fit the data and infer aForster distance of z0 = 14.5 ± 0.4 nm, in reasonable agreementwith the theory prediction (z0 = 15.3 nm, eq 3).Summarizing, we have demonstrated optical scanning

fluorescence resonance energy transfer microscopy with asingle NV center in nanodiamond as emitter. We reliably placedthis local probe sufficiently close to graphene to observe energytransfer as large as 40%. Our results confirm theoretical

predictions and recent experiments,29 in particular the z−4

dependence of the energy transfer rate between a point-likeemitter and a graphene monolayer. Employing a singlescanning emitter enabled us to map the transfer efficiency inall three dimensions without the need for any specialpreparation of the sample. With this technique, we haveobtained nanoscale resolution SNOM images of arbitrarilyshaped graphene flakes. This opens the way to image energytransfer into arbitrary nanostructures, such as structuredgraphene or carbon nanotubes. Furthermore, our methodpromises to map changes of material properties such as theFermi energy in a contactless and noninvasive way bymonitoring concomitant changes in the FRET efficiency. Inanother extension of the experiment, the scanning emittercould be used to inject single plasmons instead of excitons inother materials, e.g., silver nanowires31 or doped graphene.27,28

While graphene is an interesting photonic material in its ownright, application of the technique to other nanophotonicstructures and acceptors, e.g., single molecules, certainly wouldbe of great interest. Applications in biological sciences are easilyenvisioned. Here, scanning FRET might become a valuableaddition to other FRET based techniques for, e.g., imaginglarger protein structures of cellular surfaces. Such methods may

Figure 2. Scanning FRET on graphene mono- and multiple layers. (a) Near-field images in three dimensions (b,c). Detailed plot of the NVfluorescence intensity as a function of axial (b) as well as lateral position (c) of the nanodiamond over a multilayer (b) and single layer (c) graphenesample. Close to the graphene, we observe a quenching of fluorescence (z < 5 nm in left half of panel b), which is stronger than the attenuation ofthe excitation laser caused by the absorption of the multilayer (z > 5 nm in panel b). (d) Fluorescence decay measurements of the NV center next tothe graphene and on graphene. (e) Detailed plot of the NV lifetime next to the graphene and on graphene.

Nano Letters Letter

dx.doi.org/10.1021/nl401129m | Nano Lett. 2013, 13, 3152−31563154

be combined with the magnetic field sensing capabilities of theNV center to yield a truly multifunctional local probe.Methods. The setup used for the experiments is a home-

built scanning confocal microscope combined with an atomicforce microscope (AFM) (MFP-3D Asylum Research). Weused a frequency doubled cw Nd:YAG laser (CoherentCompass) and, for the lifetime measurements, a pulsed fiberlaser (Fianium). The excited and fluorescent light is collectedby a high NA objective (Olympus PlanAPO, NA = 1.35), andthe fluorescent light beam passes a dichroic beamsplitter (640DCXR, Chroma) and a bandpass filter (650−750 nm) until it isdetected in an avalanche photodiode (SPQR-14, Perkin-Elmer). The lifetime is measured by TCSPC electronics(PicoHarp 300, PicoQuant GmbH) and by SymPhoTime V4.0software.The nanodiamonds are brought onto commercially available

cantilevers (NanoAndMore, PL2-NCHR) by an UV-curableglue (Thorlabs, NOA81).To detect an image in the xz-plane, we used a field-

programmable gate array (FPGA, OpalKelly XEM3005) thatbins the arriving photons corresponding to the different heightsof the oscillating tip. A custom software is used to generate athree-dimensional image.The edge profile (Figure 3c) was obtained by averaging a

two-dimensional scan (150 scan lines of Figure 3b) along thedirection of the graphene edge.

■ AUTHOR INFORMATION

Corresponding Author*(J.T.) E-mail: j.tisler@physik.uni-stuttgart.de.

NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTS

This work was supported by the EU via projects SQUTEC andDINAMO, the German Science foundation (DFG) via researchgroup 1493 and SFB/TR 21, the Max-Planck society, theVolkswagenstiftung, the BMBF (CHIST-ERA QINVC), andcontract research ‘Methoden fur die Lebenswissenschaften’ ofthe Baden-Wurttemberg Stiftung.

■ REFERENCES(1) Vesseur, E. J. R.; de Waele, R.; Kuttge, M.; Polman, A. Nano Lett.2007, 7, 2843−2846.(2) Esteban, R.; Vogelgesang, R.; Dorfmueller, J.; Dmitriev, A.;Rockstuhl, C.; Etrich, C.; Kern, K. Nano Lett. 2008, 8, 3155−3159.(3) Frimmer, M.; Chen, Y.; Koenderink, A. F. Phys. Rev. Lett. 2011,107, 123602.(4) Vickery, S. A.; Dunn, R. C. Biophys. J. 1999, 76, 1812−1818.(5) Mivelle, M.; van Zanten, T. S.; Neumann, L.; van Hulst, N. F.;Garcia-Parajo, M. F. Nano Lett. 2012, 12, 5972−5978.(6) Bian, R. X.; Dunn, R. C.; Xie, X. S.; Leung, P. T. Phys. Rev. Lett.1995, 75, 4772−4775.(7) Anger, P.; Bharadwaj, P.; Novotny, L. Phys. Rev. Lett. 2006, 96,113002.(8) Shubeita, G. T.; Sekatskii, S. K.; Dietler, G.; Potapova, I.; Mews,A.; Basche, T. J. Microsc. 2003, 210, 274−278.(9) Farahani, J. N.; Pohl, D. W.; Eisler, H.-J.; Hecht, B. Phys. Rev. Lett.2005, 95, 017402.(10) Sekatskii, S. K.; Dietler, G.; Letokhov, V. S. Chem. Phys. Lett.2008, 452, 220−224.(11) Michaelis, J.; Hettich, C.; Mlynek, J.; Sandoghdar, V. Nature2000, 405, 325−328.(12) Michalet, X.; Pinaud, F. F.; Bentolila, L. A.; Tsay, J. M.; Doose,S.; Li, J. J.; Sundaresan, G.; Wu, A. M.; Gambhir, S. S.; Weiss, S. Science2005, 307, 538−544.(13) Cuche, A.; Mollet, O.; Drezet, A.; Huant, S. Nano Lett. 2010, 10,4566−4570.

Figure 3. Quantitative comparison of scanning fluorescence resonance energy transfer microscopy images to theory. A high-resolution contact-modescan of a graphene edge (a,b) is used to measure the step response function of the single NV center scanning over the graphene monolayer. (c) Thesolid line is a fit to the data using eq 1, parametrized by the NV-graphene distance of zNV = 15.9 nm. (d) Vertical dependence γNV(z). A fit of the data(red line) agrees with the predicted z−4 law (eq 2) for a Forster distance of z0 = 14.5 ± 0.4 nm.

Nano Letters Letter

dx.doi.org/10.1021/nl401129m | Nano Lett. 2013, 13, 3152−31563155

(14) Kurtsiefer, C.; Mayer, S.; Zarda, P.; Weinfurter, H. Phys. Rev.Lett. 2000, 85, 290−293.(15) Gruber, A.; Drabenstedt, A.; Tietz, C.; Fleury, L.; Wrachtrup, J.;von Borczyskowski, C. Science 1997, 276, 2012−2014.(16) Tisler, J.; Reuter, R.; Lammle, A.; Jelezko, F.; Balasubramanian,G.; Hemmer, P. R.; Reinhard, F.; Wrachtrup, J. ACS Nano 2011, 5,7893−7898.(17) Kuhn, S.; Hettich, C.; Schmitt, C.; Poizat, J.-P.; Sandoghdar, V.J. Microsc. 2001, 202, 2−6.(18) Swathi, R. S.; Sebastian, K. L. J. Chem. Phys. 2008, 129, 054703.(19) Swathi, R. S.; Sebastian, K. L. J. Chem. Phys. 2009, 130, 086101.(20) Kim, J.; Cote, L. J.; Kim, F.; Huang, J. J. Am. Chem. Soc. 2009,132, 260−267.(21) Chen, Z.; Berciaud, S.; Nuckolls, C.; Heinz, T. F.; Brus, L. E.ACS Nano 2010, 4, 2964−2968.(22) Koppens, F. H. L.; Chang, D. E.; García de Abajo, F. J. NanoLett. 2011, 11, 3370−3377.(23) Nair, R. R.; Blake, P.; Grigorenko, A. N.; Novoselov, K. S.;Booth, T. J.; Stauber, T.; Peres, N. M. R.; Geim, A. K. Science 2008,320, 1308−1308.(24) Stohr, R. J.; Kolesov, R.; Pflaum, J.; Wrachtrup, J. Phys. Rev. B2010, 82, 121408.(25) Forster, T. Ann. Phys. 1948, 437, 55−75.(26) Stohr, R. J.; Kolesov, R.; Xia, K.; Reuter, R.; Meijer, J.;Logvenov, G.; Wrachtrup, J. ACS Nano 2012, 6, 9175−9181.(27) Gomez-Santos, G.; Stauber, T. Phys. Rev. B 2011, 84, 165438.(28) Velizhanin, K. A.; Efimov, A. Phys. Rev. B 2011, 84, 085401.(29) Gaudreau, L.; Tielrooij, K. J.; Prawiroatmodjo, G. E. D. K.;Osmond, J.; Garcia de Abajo, F. J.; Koppens, F. H. L. ArXiv E-Prints2013, 1301.4681.(30) Mangum, B. D.; Shafran, E.; Mu, C.; Gerton, J. M. Nano Lett.2009, 9, 3440−3446.(31) Kolesov, R.; Grotz, B.; Balasubramanian, G.; Stohr, R. J.;Nicolet, A. A. L.; Hemmer, P. R.; Jelezko, F.; Wrachtrup, J. Nat. Phys.2009, 5, 470−474.

Nano Letters Letter

dx.doi.org/10.1021/nl401129m | Nano Lett. 2013, 13, 3152−31563156