On-chip linear and nonlinear control of single molecules coupled to a nanoguide
Pierre Türschmann, Nir Rotenberg, Jan Renger, Irina Harder, Olga Lohse, Tobias Utikal, Stephan Götzinger, Vahid Sandoghdar
OOn-chip linear and nonlinear control of single molecules coupled to a nanoguide
Pierre T¨urschmann, Nir Rotenberg, Jan Renger, Irina Harder, OlgaLohse, Tobias Utikal, Stephan G¨otzinger,
2, 1 and Vahid Sandoghdar
1, 2 Max Planck Institute for the Science of Light, Staudtstr. 2, D-91058 Erlangen, Germany Friedrich Alexander University Erlangen-Nuremberg, D-91058 Erlangen, Germany (Dated: August 13, 2018)While experiments with one or two quantum emitters have become routine in various laborato-ries, scalable platforms for efficient optical coupling of many quantum systems remain elusive. Toaddress this issue, we report on chip-based systems made of one-dimensional subwavelength dielec-tric waveguides (nanoguides) and polycyclic aromatic hydrocarbon molecules. After discussing thedesign and fabrication requirements, we present data on coherent linear and nonlinear spectroscopyof single molecules coupled to a nanoguide mode. Our results show that external microelectrodesas well as optical beams can be used to switch the propagation of light in a nanoguide via the Starkeffect and a nonlinear optical process, respectively. The presented nanoguide architecture pavesthe way for the investigation of many-body phenomena and polaritonic states and can be readilyextended to more complex geometries for the realization of quantum integrated photonic circuits.
In the past three decades, optical studies of singlequantum systems have matured to become commonplacein many laboratories. A next grand challenge in quan-tum nano-optics will be to control mesoscopic assembliesof individual quantum systems, where only a few parti-cles of light and matter interact, possibly in an entangledfashion. The very first steps in this direction, involvingtwo quantum emitters have already been taken in varioussystems [1–6], but the low efficiencies in such experimentshamper their scaling prospects.To achieve significant correlated dynamics and polari-tonic effects, it is desirable to couple many quantumemitters with large scattering cross sections to a com-mon spatial photonic mode, all at the same transitionwavelength λ [7–10]. One particularly attractive ap-proach for realizing this scenario is to couple the emittersto a one-dimensional subwavelength waveguide (nanogu-ide), which can act as an optical bus for communicationamong emitters at distances much larger than λ . Severalgroups have recently taken pioneering steps in this direc-tion by coupling atoms [11, 12], semiconductor quantumdots [13, 14], and molecules [15–17] to nanoguides. Eachsystem has some merits and confronts some challenges.In particular, reaching high densities for the realizationof complex cooperative effects [8–10] remains a nontrivialtask. Here, organic molecules offer a unique advantagebecause they can be embedded in a solid matrix at den-sities of several thousands per cubic wavelength [18].Organic dye molecules are used in a large number ofapplications in physics, chemistry and technology withrecent important contributions to the development offluorescence nanoscopy [19]. Although the widespreaddissemination of results from single-molecule biophysicshas left most scientists with the impression that organicmolecules photobleach and exhibit broad spectra, it turnsout that polycyclic aromatic hydrocarbons (PAH) suchas pentacene or dibenzoterrylene (DBT) can be nearlyindefinitely photostable when embedded in organic crys-tals (see Fig. 1a) [18, 20]. Furthermore, PAHs can havevery stable and narrow resonances at superfluid helium ITO digit electrode(b)(c) Upper substrate with integrated reservoirFilled reservoirLower substrateTiO waveguide with grating coupler Upper substrateDBT p DCBAC(a)
ClCl
Filled reservoir Organic waveguideLower substrateITO counter electrode
Figure 1. a) Molecular structures of anthracene (AC), para -dichlorobenzene ( p DCB) and dibenzoterrylene (DBT). Thefirst two were used to embed DBT. b) Schematics of ananoguide architecture, where p DCB (green) doped withDBT molecules guides light while being surrounded by fusedsilica on all sides. In this structure, we also integrated indiumtin oxide (ITO) microelectrodes (brown) for applying DC elec-tric fields. The structure is cut in half along the nanoguidefor illustration purposes. c) In this case, molecules embeddedin AC (green) are evanescently coupled to a TiO nanogu-ide (red), which is terminated by integrated grating couplers.The upper substrates in (b) and (c) are offset for ease of il-lustration. See Suppl. Info. for fabrication details. a r X i v : . [ qu a n t - ph ] F e b temperature, giving access to scattering cross sectionsclose to the ideal value of 3 λ / π [21].The coupling between the guided mode of a nanogu-ide and PAHs can be achieved in two different strate-gies. First, one can fabricate the nanoguide from thesame organic matrix that carries the PAHs, as we re-cently showed in a nanocapillary geometry [16]. This ar-rangement is facilitated by the moderately high refractiveindex ( n ) of such matrices. In the second approach, onecan place the organic matrix around a nanoguide madeof material with even higher n such that the PAHs wouldcouple to its mode evanescently. In either case, it is de-sirable to pursue chip architectures [22–24] because com-pared to fiber-based systems [15–17], these allow for theincorporation of micro- and nanoelectrodes to tune indi-vidual molecules to the same resonance frequency via theStark effect and can host feedback microstructures suchas photonic crystals. To this end, nanoguides on chipsprovide an ideal platform for scalable quantum opticalnetworks.In both fabrication strategies, molecules have to beplaced very close to crystal boundaries. However, oneshould keep in mind that PAH spectra might degrade inthe close vicinity to interfaces, where the quality of thecrystal is typically compromised in its last molecular lay-ers [25]. This behavior is also known in other solid-statesystems, where the presence of amorphous media and de-fects causes spectral instability of guest quantum emit-ters [26], making it challenging to obtain Fourier-limitedspectra close to structures such as plasmonic antennas ordielectric waveguides.To address all the design and fabrication issues dis-cussed above, we exploit a method based on the capil-lary flow of an organic matrix in the molten liquid phaseand its subsequent crystallization upon controlled cool-ing [16, 25]. This fabrication strategy allows us to achievea uniform coverage of organic molecular crystals dopedwith DBT molecules [27] in and around dielectric nanogu-ides on a chip with negligible crystal defects (see Fig.1b, c). Figure 1b shows the schematics of a nanoguideconcept, where the organic crystal para -dichlorobenzene( p DCB) [28] with n = 1 .
54 defines the guiding mediumfor light surrounded by fused silica ( n = 1 . β factor of 12%, where β is defined as the fraction of the power radiated by theemitter into the nanoguide mode. The advantage of thisapproach is that the molecules lie within the nanoguidewhere its radial mode profile is maximal. Figure 1b alsosketches an arrangement of microelectrodes close to thenanoguide. Fabrication details are described in the On-line Supplementary Information.Figure 1c illustrates the schematics of another archi-tecture, where the nanoguide is made of TiO ( n = 2 . n = 1 . β up to 28% at the inter-face between the nanoguide and AC. The nanoguide crosssection is adiabatically tapered on each end to a grating for facile interfacing of the guided mode with free-spacebeams normal to the substrate plane (see Suppl. Info). Figure 2. a) Schematic of the optical setup; S: sample, TCA:telecentric lens assembly, AL: aspheric lens, SM: scanningmirror, BS: beam splitter, F: spectral filter, APD: avalanchephotodiode. b) Each spectrum shows a fluorescence excita-tion spectrum recorded from molecules in the same excitationspot (see inset). Spectra recorded at various Stark voltagesare offset in the vertical direction. c) Extinction (green) andStokes-shifted fluorescence (red) spectra recorded through theoutput and input grating ports, respectively, (see inset lowerright) as a function of the excitation laser frequency. Thecentral inset shows a zoom of the extinction spectrum.
Figure 2a shows the schematics of the experimentalsetup for spectroscopy and microscopy of single moleculescoupled to nanoguides. The chip samples were mountedinside a helium bath cryostat operating at 1.7 K andcould be positioned in the substrate (x-y) plane usingslip-stick piezo sliders. Two aspheric lenses with numer-ical aperture NA=0.77 were used to access the samplefrom the opposite sides, whereby one of them could bepositioned in all three dimensions, and the other one wasadjustable along z. The sample could be illuminated bynarrow-band (∆ ν < | g, v = 0 (cid:105) ) andexcited ( | e, v = 0 (cid:105) ) states. Upon excitation, the upperstate can decay via broad transitions to | g, v (cid:54) = 0 (cid:105) levels,leading to red-shifted fluorescence. The nanoguide ge-ometry provides convenient simultaneous access to boththe incoherent red-shifted fluorescence and the coherentsignal, which can be detected through the two gratingports or from the side.Each spectrum in Fig. 2b displays fluorescence excita-tion spectra recorded from a few DBT molecules embed-ded in p DCB within the focal spot of the laser beam(see inset). The different horizontally offset spectra wereregistered at different voltages applied to one of the mi-croelectrodes shown in Fig. 1b, while the others weregrounded. The data clearly show that the 00ZPLs ofmolecules within an area of diameter less than a microm-eter can be tuned at about 0.5 GHz/V over tens of GHz,which is a substantial fraction of the inhomogeneous dis-tribution of DBT resonances [28]. We note that individ-ual molecules respond differently to the external electricfield because each experiences a different local residualmatrix field [18, 29]. Considering the very large numberof molecules along the nanoguide, local Stark shifts willallow one to tune several molecules to the same frequency,realizing a mesoscopic ensemble.Next, we turn to the architecture of Fig. 1c, wheremolecules are evanescently coupled to the nanoguide(cross section of 200 nm ×
200 nm). The green curve inFig. 2c shows a spectrum recorded by detecting the lightintensity in the nanoguide through one grating while itwas excited through the other. The inset displays a high-resolution extinction spectrum, revealing sharp spectralfeatures with Lorentzian line shapes. The observed fullwidth at half-maximum (FWHM) of Γ = 30 MHz corre-sponds to the natural linewidth of the 00ZPL in DBT andis a robust indication for the coupling of single moleculesto the nanoguide mode. Furthermore, the narrow res-onances make it possible to modulate the attenuationof the light beam in the nanoguide by a small voltageapplied to microelectrodes, even at MHz speeds [30].The extinction dips in the transmission signal reach upto 7.2%, corresponding to β = 7 .
4% if we assume acombined Franck-Condon/Debye-Waller factor of 0.5 (seeSuppl. Info.).The red spectrum in Fig. 2c presents the Stokes-shiftedfluorescence recorded from the input grating port simul-taneously to the extinction spectrum. While all the fea-tures in the extinction spectrum are also represented bythe corresponding lines in the fluorescence signal, someresonances of the latter do not appear in the former. Weattribute these to the contribution of molecules that areclose to the input grating but with low coupling to the nanoguide mode. (a)(c)(b) Grating GratingMolecule g ( ) ( τ ) Time delay τ (ns)TiO nanoguideGrating GratingMolecule TiO nanoguide Figure 3. a) Wide-field fluorescence image when a singlemolecule was excited by a focused laser beam from the side(see upper schematics). b) Geometry of the nanoguide struc-ture imaged by iSCAT (see text for details). A cross showsthe position of the molecule imaged in (a). c) Second-ordercross-correlation measurement of the red shifted fluorescencedetected through the two grating ports while the moleculewas excited as in (a).
To examine the propagation of light along the nanogu-ide, in Fig. 3a we show a wide-field fluorescence image.Here, a single molecule was first identified in frequencyspace and then excited from the side in the focus ofa laser beam (see inset). The three bright regions de-note the Stokes-shifted fluorescence of the molecule de-tected directly at its location and through the two gratingports. The absence of fluorescence along the nanoguidein Fig. 3a verifies both a very low scattering loss and ahigh spectral selectivity to a single molecule. The imageobtained via interference scattering microscopy (iSCAT)[31] in Fig. 3b helps visualize the chip geometry. By cor-relating the images in (a) and (b), we found the moleculeto lie at about 360 nm from the nanoguide edge.Figure 3a nicely illustrates an advantage of the nanogu-ide architecture as an ideal integrated beam splitter fordetecting the emission of a molecule in opposite direc-tions. In this structure, we found the ratio of the fluo-rescence signal measured from the two gratings to yielda splitting efficiency of 57:43. By recording the two grat-ing signals in a start-stop coincidence configuration, weobtained the second-order cross-correlation function g (2) shown in Fig. 3c. A very strong antibunching at zerodelay confirms that the fluorescence stems from a singlemolecule. Moreover, the theoretical fit at a Rabi fre-quency of 0.9Γ lets us determine an excited-state life-time of 5 ns. We remark in passing that cross-correlationmeasurements in a one-dimensional nanoguide provide anideal platform for revisiting the particle nature of spon-taneous emission since a detector click either occurs onthe right or the left channel [32].A single PAH molecule can act as a highly nonlinearmedium for single photons if it is efficiently coupled to apropagating mode. Such a nonlinearity can be exploitedto switch the propagation of photons by using an externalcontrol laser field which dresses the molecular electronicstates [33]. Our current nanoguide geometry providesa particularly convenient configuration for such studiesbecause by focusing the stronger pump beam from theside and detecting the weaker probe beam through thenanoguide, one can easily separate the two in the detec-tion channel (see inset in Fig. 4a). Figure 4a shows a mapof the nanoguide transmission for various frequency de-tunings between the pump laser and the molecule as wellas between the pump and probe lasers. Here, the pumpstrength was chosen to correspond to a moderate Rabifrequency of Ω pmp = Γ while the probe was kept weakat Ω prb = Γ /
10. The map displays regions where theprobe signal is attenuated by about 7% and other partswhere one detects about 5% more signal than the probeintensity. It is important to note, however, that the lat-ter is predominantly due to the direct resonant scatteringof the pump beam by the molecule into the nanoguidemode. To assess the amplification of the probe beam, wenormalized the signal to its value far from the molecu-lar resonance while keeping the pump-molecule frequencydetuning constant. In Fig. 4b, we display examples ofthe probe transmission versus pump-probe detuning forthree different pump Rabi frequencies of 0 . (blue),1 . (green), and 2 . (red). The vertical dashed linespoint to two frequency regions, where the probe attenu-ation and amplification are switched most sensitively asthe pump power is increased. We find the net coherentamplification of the probe beam to reach 0.3%.The symbols in Fig. 4c summarize the degrees of probeextinction and amplification measured for many pumppowers. The shadowed regions of the solid curves showthe corresponding calculated quantities, allowing for asmall spectral instability less than 0 . , which occurredover several hours during these measurements (see Suppl.Info.). These data show that in addition to a static Starkeffect discussed above, one can also use an external opti-cal field to control the propagation of a light beam in thenanoguide. The latter option is particularly attractivebecause it can be implemented with a diffraction-limitedspatial resolution and at speeds up to hundreds of MHz.In this Letter, we presented linear and nonlinear mea-surements on individual molecules coupled to waveguideson a chip and showed that the propagation of light in ananoguide can be switched by external electrostatic oroptical fields. The demonstration that nanofabricated (a)(b)(c) pump probe Figure 4. a) Nanoguide transmission as a function of pump-molecule and pump-probe frequency detunings for fixed probeand pump Rabi frequencies of Ω prb = 0 . and Ω pmp = Γ ,respectively. b) Exemplary nanoguide transmission spectrafor three different pump intensities Ω pmp =0.5 MHz (blue),Ω pmp =43 MHz (green), Ω pmp =77 MHz (red). The verticaldashed lines point to the strongest attenuations and amplifi-cations of the probe beam. c) Extinction signal (black) andcoherent amplification (red) of the probe beam as a functionof Rabi frequency and incoming photon number of the pumpbeam. The grey regions correspond to expected variationscaused by a very small spectral diffusion less than 0 . structures can be overlapped with molecular crystals overlarge areas opens realistic prospects for investigating po-laritonic and quantum many-body phenomena while con-trolling the contribution of each single emitter. Futureemployment of optical circuit elements such as beamsplitters, routers and interferometers [34] together withnovel devices such as on-chip superconducting detectors[35] will also allow the realization of integrated quantumcircuits. A particular advantage of chip architectures istheir ease of interfacing with other chip-based materialssuch as color centers [6] or rare-earth ions [36, 37] to re-alize hybrid quantum systems. We thank Harald Haakh for fruitful discussions andhelp with theoretical design and Maksim Schwab for hiscontributions to the mechanical components. This workwas supported by an Alexander von Humboldt professor-ship, the Max Planck Society and the European ResearchCouncil (Advanced Grant SINGLEION). [1] P. Maunz, D. L. Moehring, S. Olmschenk, K. C. Younge,D. N. Matsukevich, and C. Monroe, Nature Phys. , 538(2007).[2] Y. L. A. Rezus, S. G. Walt, R. Lettow, A. Renn, G. Zu-mofen, S. G¨otzinger, and V. Sandoghdar, Phys. Rev.Lett. , 093601 (2012).[3] S. Ritter, C. N¨olleke, C. Hahn, A. Reiserer, A. Neuzner,M. Uphoff, M. Mcke, E. Figueroa, J. Bochmann, andG. Rempe, Nature , 195 (2014).[4] W. Pfaff, B. Hensen, H. Bernien, S. B. van Dam, M. S.Blok, T. H. Taminiau, M. J. Tiggelman, S. R. N.,M. Markham, D. J. Twitchen, and R. Hanson, Science , 532 (2014).[5] J. D. Hood, A. Goban, A. Asenjo-Garcia, M. Lu, S.-P.Yu, D. E. Chang, and H. J. Kimble, Proc. Nat. Acad.Sci. , 10507 (2016).[6] E. R. Sipahigil, A. and, D. Sukachev, M. Burek, J. Bor-regaard, M. Bhaskar, C. Nguyen, J. Pacheco, H. Atikian,C. Meuwly, R. Camacho, F. Jelezko, E. Bielejec, H. Park,M. Loncar, and M. Lukin, Science , 847 (2016).[7] A. D. Greentree, C. Tahan, J. H. Cole, and L. C. L.Hollenberg, Nature Phys. , 856 (2006).[8] I. Carusotto and C. Ciuti, Rev. Mod. Phys. , 299(2013).[9] H. Haakh, S. Faez, and V. Sandoghdar, Phys. Rev. A , 053840 (2016).[10] C. Noh and D. Angelakis, arXiv:1604.04433 [quant-ph](2016).[11] E. Vetsch, D. Reitz, G. Sagu´e, R. Schmidt, S. T.Dawkins, and A. Rauschenbeutel, Phys. Rev. Lett. ,203603 (2010).[12] A. Goban, C.-L. Hung, S.-P. Yu, J. Hood, J. Muniz,J. Lee, M. Martin, A. McClung, K. Choi, D. Chang,O. Painter, and H. Kimble, Nat. Commun. , 3808(2014).[13] R. Yalla, F. Le Kien, M. Morinaga, and K. Hakuta, Phy.Rev. Lett. , 063602 (2012).[14] M. Arcari, I. S¨ollner, A. Javadi, S. Lindskov Hansen,S. Mahmoodian, J. Liu, H. Thyrrestrup, E. H. Lee, J. D.Song, S. Stobbe, and P. Lodahl, Phys. Rev. Lett. ,093603 (2014).[15] A. Stiebeiner, O. Rehband, G.-F. R., and A. Rauschen-beutel, Opt. Exp. , 21704 (2009).[16] S. Faez, P. T¨urschmann, H. R. Haakh, S. G¨otzinger, andV. Sandoghdar, Phys. Rev. Lett. , 213601 (2014).[17] S. M. Skoff, D. Papencordt, H. Schauffert, andA. Rauschenbeutel, arxiv:1604.04259 (2016).[18] T. Basche, W. E. Moerner, M. Orrit, and U. Wild, eds., Single Molecule Optical Detection, Imaging and Spec-troscopy (VCH Verlagsgesellschaft mbH, 1996).[19] S. Hell, Science , 1153 (2007).[20] R. J. Pfab, J. Zimmermann, C. Hettich, I. Gerhardt,A. Renn, and V. Sandoghdar, Chem. Phys. Lett. , 490 (2004).[21] I. Gerhardt, G. Wrigge, P. Bushev, G. Zumofen, M. Agio,R. Pfab, and V. Sandoghdar, Phys. Rev. Lett. ,033601 (2007).[22] J. Hwang and E. A. Hinds, New J. Phys. , 085009(2011).[23] G. Kewes, M. Schoengen, O. Neitzke, P. Lombardi, R.-S. Sch¨onfeld, G. Mazzamuto, A. W. Schell, J. Probst,J. Wolters, B. L¨ochel, C. Toninelli, and O. Benson, Sci.Rep. , 28877 (2016).[24] P. E. Lombardi, A. P. Ovvyan, S. Pazzagli, G. Maz-zamuto, G. Kewes, O. Neitzke, N. Gruhler, O. Ben-son, W. Pernice, F. S. Cataliotti, and C. Toninelli,arXiv:1701.00459v1 (2017).[25] B. Gmeiner, A. Maser, T. Utikal, S. G¨otzinger, andV. Sandoghdar, Phys. Chem. Chem. Phys. , 19588(2016).[26] M. Orrit, editor, Mol. Phys.; Special issue on SpectralDiffusion and Related Topics (2009).[27] F. Jelezko, P. Tamarat, B. Lounis, and M. Orrit, J. Phys.Chem. , 13892 (1996).[28] N. Verhart, M. M¨uller, and M. Orrit, Chem. Phys.Chem. , 1524 (2016).[29] T. Y. Latychevskaia, A. Renn, and U. Wild, Chem. Phys. , 109 (2002).[30] C. Brunel, B. Lounis, P. Tamarat, and M. Orrit, Phys.Rev. Lett. , 2722 (1999).[31] K. Lindfors, T. Kalkbrenner, P. Stoller, and V. Sandogh-dar, Phys. Rev. Lett. , 037401 (2004).[32] P. Grangier, G. Roger, and A. Aspect, Euorphys. Lett. , 173 (1986).[33] A. Maser, B. Gmeiner, T. Utikal, S. G¨otzinger, andV. Sandoghdar, Nat. Photon. , 450 (2016).[34] M. Lipson, J. Lightwave Tech. , 4222 (2006).[35] M. K. Akhlaghi, E. Schelew, and J. F. Young, Nat. Com-mun. , 8233 (2015).[36] T. Utikal, E. Eichhammer, L. Petersen, A. Renn,S. G¨otzinger, and V. Sandoghdar, Nat. Commun. , 3627(2014).[37] S. Marzban, J. G. Bartholomew, S. Madden, K. Vu, andM. J. Sellars, Phys. Rev. Lett.115