Long-Distance Resonant Energy Transfer Mediated by HybridPlasmonic−Photonic ModesKatherine Akulov, Dan Bochman, Adina Golombek, and Tal Schwartz*

School of Chemistry, Raymond and Beverly Sackler Faculty of Exact Sciences and Tel Aviv University Center for Light-MatterInteraction, Tel Aviv University, Tel Aviv 6997801, Israel

ABSTRACT: Metallic nanostructures are well known for their potential toenhance resonant energy transfer between chromophores, mediated by theirplasmonic field. In most cases, the distances for efficient energy transfer aredetermined by the dimensions of the structure, making dissipation in themetal dominant when long-range energy transfer is desired. Here, wepropose and study a new mechanism for long-distance energy transfer, whichcan lead to highly efficient energy transfer over distances comparable to theoptical wavelength, based on hybrid photonic−plasmonic modes in metal-dielectric structures. We study such structures theoretically and characterizethe energy-transfer processes in them, showing that within these structures,energy can be funneled with ∼25% efficiency over a range of about 150 nm,with minimal dependence on the location of the acceptor molecules insidethe structure. Finally, we demonstrate this new mechanism experimentally,providing a proof-of-concept for long-distance energy transfer in suchstructures. Our multilayer system is compatible with standard photovoltaic and organic light-emitting diodes and therefore themechanism presented can be readily employed in such organic optoelectronic devices to improve their performance.

■ INTRODUCTIONThe potential for metallic nanostructures to improve energytransfer between chromophores has long been recognized.1,2

Since this early work, many different geometries have beenproposed and studied, relying on either surface plasmons inplanar structures3−9 or localized plasmons on metallicnanostructures10−18 to mediate long-range energy transfer.Motivated by the promising opportunities for solar en-ergy,5,19,20 light-emitting devices,7,21,22 and other applications,enhancing the energy-transfer range and its efficiency usingplasmonic systems is still a highly desirable goal, attractingconsiderable effort even today.23 In plasmon-mediated energytransfer, the excitation energy of an emitter is coupled to theconfined mode of a resonant plasmonic structure, causing asignificant enhancement of the near field generated by thedonor. Moreover, the spatial extent of the field generated bythe donor is then determined by the plasmonic structure,providing a mechanism for overcoming the short-range nature(with a 1/R6 distance dependence) of the usual Forsterresonance energy transfer (FRET) process. Indeed, usingplasmonic structures, energy transfer over distances 1−2 ordersof magnitude larger than the nanometric scale typical of FREThas been reported.3,9,14,24 However, since the length scale forefficient energy transfer is usually set by the dimensions of thestructure, carrying the energy over long distances requireslarger structures, which unavoidably makes the losses in themetal dominant and limits the energy-transfer efficiency.9

Recently, strong light-molecule coupling has been demon-strated to provide an alternative route for enhanced energytransfer,25−28 taking advantage of the collective nature of

resulting polaritonic states. However, reaching the strongcoupling regime in such systems typically requires highmolecular densities, which is not always permissible in practicalapplications.Here, we explore a new mechanism for long-range energy

transfer between chromophores, using a metal-dielectricmicrocavity and strong coupling between surface plasmonsand cavity photons. Taking advantage of the delocalized natureof the hybrid photonic−plasmonic modes, we achieve energytransfer with an efficiency of ∼25% over distances of 100−200nm. Moreover, we show that the spectral response of thestructure and the wavelength dependence of the energy-transfer efficiency can be conveniently tailored. This propertyprovides superior flexibility in the selection of the donor−acceptor pair compared to previous methods.

■ HYBRID PLASMONIC−PHOTONIC MODES IN AMETALLIC MICROCAVITY

In their pioneering work,3 Andrew and Barnes showed thatnormal-mode coupling between two surface plasmons residingon either side of a metal layer can be used for transferringenergy from donor molecules deposited on one side toacceptor molecules on the opposite side. Using the coupledplasmonic modes as an “optical bridge” between the molecules,these experiments demonstrated energy transfer across metal

Received: March 30, 2018Revised: June 14, 2018Published: June 18, 2018

Article

pubs.acs.org/JPCCCite This: J. Phys. Chem. C 2018, 122, 15853−15860

© 2018 American Chemical Society 15853 DOI: 10.1021/acs.jpcc.8b03030J. Phys. Chem. C 2018, 122, 15853−15860

Dow

nloa

ded

via

TE

L A

VIV

UN

IV o

n Ju

ly 3

1, 2

018

at 1

3:50

:25

(UT

C).

Se

e ht

tps:

//pub

s.ac

s.or

g/sh

arin

ggui

delin

es f

or o

ptio

ns o

n ho

w to

legi

timat

ely

shar

e pu

blis

hed

artic

les.

films as thick as ∼100 nm. In our work, we rely on a similaroptical-mode coupling to mediate energy transfer betweenremote molecular layers. However, here, we strongly couple asurface plasmon mode with a microcavity mode, takingadvantage of the properties of these two different electro-magnetic excitations. On the one hand, the plasmonic modeacts as a very efficient channel to capture the donor emissiondue to its evanescent nature and the increased density of statesin its vicinity.29 On the other hand, the cavity mode primarilyresides in the dielectric medium within the cavity, minimizinglosses in the metal. As we show, the mixing between thesemodes can provide a new mechanism for long-distance energytransfer, which maintains its high efficiency even over distancescomparable to the optical wavelength.As illustrated in Figure 1a, our system is a half-wavelength

metallic Fabry−Perot cavity consisting of two Ag mirrorsseparated by a ∼160 nm thick dielectric layer (e.g., transparentpolymer), taken with a refractive index of 1.5. In addition to

the standing-wave mode residing between the cavity mirrors,this structure also supports surface plasmon modes at the topmetal−air interface. By making the top silver mirror sufficientlythin (<70 nm), we allow a small leakage of the plasmonic fieldthrough the metal film and into the cavity, where it overlapswith the intracavity photonic field. As a result, the plasmonicand photonic modes can interact resonantly, resulting in theformation of hybrid modes, which are linear combinations ofthe photonic and plasmonic modes.30 As we demonstrate here,these hybrid modes can mediate energy transfer betweendonor molecules placed on top of the structure and acceptormolecules residing inside the cavity region. In our calculations,we assume that the acceptor molecules are homogeneously andisotropically distributed in the cavity. Furthermore, the donorlayer, which is assumed to be infinitely thin, is separated fromthe top Ag mirror by a thin (9 nm) polymer spacer to preventdissipation into lossy surface waves.29,31

The resonant plasmon−photon coupling is demonstrated inFigure 1b, showing the simulated dispersion of the structure(energy vs in-plane momentum, kx), as well as the dispersioncurves of the uncoupled cavity photons (white solid line) andsurface plasmons at the metal−air interface (white dashedline). The dispersion was obtained by calculating thereflectivity of the structure, which was carried out using thetransfer-matrix formalism,32 and expressing the Fresnelcoefficient in terms of the in-plane momentum. As seen, atthe crossing point of the cavity and the plasmon lines, thedispersion of the composite structure splits into two branches,which are separated by a distinct energy gap. This splittingsignifies the resonant coupling between the two electro-magnetic modes, and the energy gap corresponds to thecoupling strength, which is determined by the thickness of thetop mirror. As shown in Figure 1c, the mixing between thephotonic and plasmonic modes varies with the in-planemomentum, giving rise to different spatial distributions ofthe electromagnetic field for different locations on thedispersion curve. Far from resonance, the electric field showsthe characteristics of a standing-wave mode (Figure 1c,i) or anevanescent plasmonic mode (Figure 1c,ii). However, aroundthe resonance, the modes are a symmetric (Figure 1c,iii) andan antisymmetric (Figure 1c,iv) combination of the plasmonand the cavity modes, exhibiting both exponential decay intoair and a standing-wave pattern in the dielectric layer betweenthe mirrors. For the specific choice of parameters for thestructure, the wavelengths of these mixed modes are 575 and535 nm. This normal-mode splitting is similar to the splittingoccurring for surface plasmons in thin metal films3,33,34 orlocalized plasmons in nanoparticles.35,36 However, here, wecouple electromagnetic modes with different natures, givingrise to hybrid photonic−plasmonic modes. These modes,which reside both outside and inside the metallic cavity, can beused to efficiently funnel energy from a donor overlayerlocated above the top mirror to acceptor molecules locateddeep within the cavity region. The precise location of theresonance can be controlled by varying the distance betweenthe cavity mirrors. Moreover, since the plasmon−photoncoupling is determined by the leakage of the plasmonic tail intothe cavity and its modal overlap with the cavity mode, we caneasily control the coupling strength and the spectral splitting inthe dispersion by changing the thickness of the Ag mirror.These properties make this structure very versatile, as theenergies of the coupled modes can be conveniently tuned tomatch the spectra of a desired donor−acceptor pair. We note

Figure 1. (a) Schematic sketch of the metallic microcavity structure.(b) Reflectivity map, calculated by the T-matrix formalism, showingthe dispersion of the hybrid photonic−plasmonic cavity with anormal-mode splitting into two branches (false-color image). Thewhite lines mark the uncoupled dispersion of the cavity (solid line)and the plasmon on the Ag−air interface (dashed line). Also markedare the air light line (black solid line) and the light line for a mediumwith a refractive index of 1.5 (black dashed line). (c) Calculated modeprofiles for several representative points on the dispersion diagram(marked (i−iv) in (b)), showing the in-plane component of theelectric field Ex for a photonic-like mode (i), the plasmonic-like mode(ii), and the hybrid symmetric (iii) and antisymmetric (iv) modes.The shaded regions mark the location of the different layers in thestructure (gray for Ag mirrors; pink for dielectric layers).

The Journal of Physical Chemistry C Article

DOI: 10.1021/acs.jpcc.8b03030J. Phys. Chem. C 2018, 122, 15853−15860

15854

that the observed line-crossing and normal-mode splitting arepossible only due to the higher refractive index of the mediuminside the cavity (compared to air on the plasmon side), whichallows the cavity dispersion to extend beyond the air light lineand to cross the plasmon dispersion.

■ SIMULATIONS OF ENERGY TRANSFER INMETAL-DIELECTRIC CAVITIES

A convenient theoretical approach for studying energytransport in photonic structures is the widely used dyadicGreen’s function formalism.37−40 To study the energy transferin our structures, we follow Celebi et al.,4 who showed that, fora layered system, simple expressions for the energy-transferefficiency can be obtained from the dyadic Green’s function.We characterize each layer j with a homogeneous and isotropicdielectric function ϵ0ϵj (with ϵ0 the dielectric constant ofvacuum) and define the normalized in-plane momentum as

u kk

x

d= , with k

cd d= ϵω being the wavenumber in the vicinity

of the donor, where ω is the donor oscillation frequency and cis the speed of light in vacuum. Furthermore, for each layer, wedefine the normalized longitudinal wavenumber as

h uj2j

d= −

ϵϵ . Using these definitions, the total flux through

a horizontal plane at a height z can be calculated using thenormal component of the Poynting vector S, giving

S Aq u h

hF z B z

F z B z u

Re d34

Re( )

( ) ( )

( ) ( ) d

zj j

j0

3

d2

p p

p p

∫ ∫=ϵ *

| | ϵ[ − ]

[ + ]*

⊥∞ l

mooonooo

|}ooo~ooo (1)

for a vertical dipole and

S Aq u h

F z B z

F z B zuh

hF z B z

F z B z u

Re d38

Re( )

( ) ( )

( ) ( ) ( ) ( )

( ) ( ) d

zj j

j

j

s

0

3p p

p p

d

s s

s

∫ ∫=ϵ *

ϵ[ − ]

[ + ]* +*

[ + ]

[ − ]*

∞ lmooonooo

|}ooo~ooo (2)

for a horizontal dipole. Here, q is the emission quantum yieldof the donor and Fp/s(z) and Bp/s(z) describe forward- andbackward-propagating planewaves for p and s polarizations,respectively. Within the jth layer, these fields can be written asFp/s(z) = f j

p/s eikdhj(z − zj) and Bp/s(z) = bjp/s e−ikdhj(z − zj), where zj

represents the location of the top interface of each layer andthe field amplitudes f j

p/s and bjp/s are related by the Fresnel

coefficients for transmission and reflection between adjacentlayers, respectively. For the case in which the donor is locatedjust above the structure (taken as z = 0), we numericallycalculate these amplitudes for each value of u, taking f 0

p/s = 1and bN+1

p/s = 0. Once the energy flux in each plane is known, thefraction of donor energy dissipated in the jth layer is thencalculated by taking the difference in the net flux going throughthe bottom and top interfaces of the layer and thennormalizing by the enhancement factor of the donor decay

rate (due to the modified local density of states near thestructure), which is given by

fq u

hr u1

32

Re d0

3

d

p∫= +⊥∞

(3)

for a vertical dipole and

fq u

hr h r u1

34

Re ( )d0 d

pd2 s∫= − −

∞

(4)

with rp = b0p and rs = b0

s being the reflection coefficients of thecomplete structure, for light coming from the donor side. 1−4can be used to calculate the yield of energy transfer from thedonor to the whole acceptor layer under the assumption thatthe acceptor molecules are isotropically and homogeneouslydistributed within the host layer and can thus be described by acomplex dielectric function ϵa(ω). Moreover, by taking the realpart of the integrands in eqs 1 and 2, calculating the derivativeswith respect to z and normalizing by f⊥/∥, we obtain thedistribution of energy transfer ϕ(kx, z). This quantity describesthe efficiency (fraction of the donor energy) at which energy istransferred through modes with a particular in-planemomentum value kx and dissipated at height z within thestructure.4 Generally, it is possible to represent the energy-transfer efficiency as a spectral integral over a product of threecomponents, which are the (normalized) fluorescencespectrum of the donor; the absorption cross section of theacceptor; and a coupling factor, which describes the effect ofthe structure and generally may depend on the intermoleculardistance, molecular orientation, and wavelength.23 To under-stand the contribution of the hybrid plasmonic−photonicmodes to the energy-transfer process, we focus here on thestructural coupling factor. Therefore, in the following, we willassume a wavelength-independent emission for the donor anduse a constant imaginary part for the refractive index in theacceptor layer, taken here as Im 0.025a{ ϵ } = (which is thehighest absorption of the host/acceptor layer, which does notaffect the dispersion of the structure).Figure 2 shows the energy-transfer distribution ϕ(kx, z)

(logarithmic scale) in the vicinity of the hybrid plasmonic−photonic modes (points (iii) and (iv) in Figure 1b,corresponding to the wavelengths 535 and 575 nm) for avertical dipole (Figure 2a,c) and for a horizontal dipole (Figure2b,d). As can be seen, the energy dissipation occurs both in themetal mirrors and in the acceptor layer, and it is concentratedaround several distinct values of the in-plane momentum kx, asexpected for a system having a resonant response. In all fourcases, the most dominant channel for energy dissipation in theacceptor layer resides around kx = 12.2 μm−1, corresponding tothe avoided crossing point on the dispersion curve. This resultclearly demonstrates that the coupling between the plasmonicand photonic cavity modes can indeed induce an efficientchannel for energy transfer between the donor and theacceptor within the cavity. In addition to this channel, we alsoobserve several other and less efficient channels for energytransfer. For 535 nm (Figure 2a,b), a weaker peak can be seenat kx ∼ 15 μm−1, originating from a mode that is primarilyconcentrated in the cavity, as it resides in a photonic-likeregion of the dispersion curve (see Figure 1b). Furthermore,we also observed a small contribution from plasmonic modesat higher values of kx, which lie beyond the light line of thehost polymer (having a refractive index of 1.5) and correspondto surface plasmon modes at the Ag−polymer interfaces. All of

The Journal of Physical Chemistry C Article

DOI: 10.1021/acs.jpcc.8b03030J. Phys. Chem. C 2018, 122, 15853−15860

15855

these energy-transfer paths are similarly seen in the case of ahorizontal donor (Figure 2b,d), but here, their magnitudes arelower. In addition, the horizontal donor also transmits some ofits energy through low kx modes, which have a predominantlyphotonic-like nature. Interestingly, in comparison to all otherchannels, for the hybrid modes, the dissipated power densityinside the acceptor layer hardly shows any variation withheight, meaning that the energy-transfer efficiency in thesestructures is practically independent of the location of theacceptor molecule within the cavity. This unique behavior isdistinctively different from the usual behavior seen in Forster-type energy transfer, which decreases as the sixth power of thedistance, or plasmon-mediated energy transfer, which decaysexponentially with distance. To highlight the effect of thisbehavior, we calculate the energy-transfer efficiency profileacross the structure, which represents the fraction of the donorenergy (at a given wavelength) that is transferred to structureat a given height z. This profile is calculated by numericallyintegrating the energy-transfer distribution ϕ(kx, z) (Figure2a,c) over the in-plane momentum. The results (for a verticaldonor) are shown in Figure 3a for 535 nm and in Figure 3b for575 nm, where the solid lines correspond to a full integrationover all kx values and the dashed line is the result of theintegration over a narrow region around kx ∼ 12.2 μm−1, whichtakes into account the contribution of the hybrid modes only.In both cases, we observe narrow and exponentially decaying

peaks in the metal mirrors, whereas the energy-transfer profilewithin the acceptor layer is almost flat, with only weak spatialvariations coming from the nonhybrid modes. These resultsalso show that energy transfer occurs predominantly throughthe hybrid modes (dashed lines in Figure 3a,b), which isresponsible for 60−80% of the donor−acceptor energy-transferyield. The invariance of the energy-transfer profile within thecavity may seem counterintuitive at first, but it can beunderstood in the following way: For surface plasmons at themetal−air interface, the field transmitted into the polymerpropagates at an angle very close to 45° (with in-planemomentum just outside the air light line). In the cavity region,this field is composed of two counterpropagating planewaves ofsimilar amplitudes, which are given by F(z) ∼ eikdhaz and B(z)∼ e−ikdhaz. At each point, the energy absorbed by the acceptorwill be proportional to the local intensity, which is taken as |Ex|

2 + |Ez|2, with Ex(Ez) being the in-plane (normal)

component of the electric field. For p-polarization and apropagation angle of 45°, these projections are givenapprox ima te l y a s E e e sinx

z L z L zL

i / i /∼ − ∼π π π+ − and

E e e coszz L z L z

Li / i /∼ + ∼π π π+ − (where we have used the

cavity resonance condition given by 2kdhaL = 2π, i.e.,constructive interference for a complete round-trip, with Lbeing the cavity thickness). Therefore, the sum of theirabsolute square values is constant, resulting in a nearly uniformabsorption profile for the acceptors inside the cavity. In Figure3c,d we plot the exact distributions of |Ex|

2 (blue line) and |Ez|2

Figure 2. Simulation results showing the energy-transfer distributionϕ(kx, z) in the structure as a function of in-plane momentum andheight, calculated at wavelengths of 535 nm (a, b) and 575 nm (c, d)for a vertical donor (left) and a horizontal donor (right). The whitehorizontal lines mark the interfaces between the different layers, andthe emitting dipole is located at z = 0. The vertical dashed line in allpanels corresponds to the air light line.

Figure 3. Distribution of energy-transfer efficiency as a function ofheight for 535 nm (a) and 575 nm (b), calculated by integrating ϕ(kx,z) over the entire in-plane momentum axis (solid line) and over anarrow momentum interval around the hybrid mode (dashed line).Calculated intensity distribution of the in-plane (blue) and normal(red) components of the antisymmetric mode at 535 nm (c) and thesymmetric mode at 575 nm (d). The shaded areas in all panescorrespond to the Ag mirrors.

The Journal of Physical Chemistry C Article

DOI: 10.1021/acs.jpcc.8b03030J. Phys. Chem. C 2018, 122, 15853−15860

15856

(red line) components for the symmetric (Figure 3c) andantisymmetric (Figure 3d) hybrid modes, as obtained from thenumerical simulations. As can be seen, the field intensityprofiles indeed match our qualitative description above. Thisresult also implies that if the acceptor molecules areconcentrated in a thin layer inside the cavity (rather thanhomogeneously distributed over the entire cavity region),similar energy-transfer efficiency will be obtained, irrespectiveof the location of the layer within the cavity and its distancefrom the donor layer.To fully characterize the different energy-transfer pathways

to the acceptor, we integrated the energy-transfer distribution(see Figure 2) over the entire height of the acceptor layer,which gives the total energy dissipation in the acceptor layer asa function of in-plane momentum. We repeated the calculationfor a series of wavelengths in the range of 400−800 nm, andthe results are shown in Figure 4a,b, for a vertical and

horizontal donor, respectively. When displaying the energy-transfer efficiency in momentum-energy coordinates, we find astriking resemblance between these results and the dispersionof the hybrid structure (shown in Figure 1b). Once again, onecan see that modes throughout the whole dispersion contributeto the donor−acceptor coupling process, but the mostdominant contribution comes from the region around thecrossing point between the plasmon and cavity dispersionlines, confirming that the mixing between these modes is thedominant channel for energy transfer in our system. One can

also note the two weaker curves lying beyond the light line ofthe polymer (marked by the white dashed lines), which areattributed to surface plasmon modes residing on the interfacesbetween the metal mirrors and the high-refractive-index layers.Finally, we calculate the total energy-transfer efficiency,

given by ∫ ϕdz dkx, where the integration is carried out overthe entire acceptor layer and over the entire kx axis. We plotthe results as a function of wavelength in Figure 4c for avertical dipole (blue curve) and for a horizontal dipole (redcurve). The two spectra exhibit broad distributions, withmaxima close to the hybrid mode wavelengths. Specifically, forthe vertical dipole, the spectrum is composed of two, partiallyoverlapping contributions: one at the higher-energy coupledmode (535 nm) and a slightly weaker one at the lower-energycoupled mode (575 nm). The higher energy-transfer efficiencyof the high-energy mode could be related to its asymmetricspatial distribution (see Figure 1c,iii), which leads to a smallerfraction of energy being absorbed by the metal mirror, as canalso be seen when comparing the energy-dissipationdistributions in Figure 2a,c for these two wavelengths. Mostimportantly, the results show that the hybrid plasmonic−photonic structure leads to a donor−acceptor couplingefficiency larger than 20% over separation distances in therange of 100−200 nm.To assess the contribution of the resonant plasmon−photon

coupling, we remove the bottom mirror such that the cavitymode does not exist and energy transfer can only occurthrough the plasmonic modes residing on both sides of the topAg mirror. This configuration is similar to the one studied inref 3; however, due to the asymmetry between the upper andlower sides of the mirror, the coupling between the twoplasmonic modes is very weak. Figure 5a shows the calculated

energy-transfer distribution for this structure, with a verticaldonor emitting at 535 nm. As can be seen, the energy transfercan still occur across the metal film via these plasmonic modes;however, in comparison to Figure 2a, we find that a muchlarger fraction of the energy is dissipated in the metal layer,primarily in the plasmonic mode at the metal−air interface.When integrating over the whole in-plane momentum axis andover the entire acceptor layer, we find that the total energy-transfer efficiency is around 6%, which is about 4 times smallerthan in the hybrid plasmonic−photonic structure. Next, we

Figure 4. Calculated energy-transfer distribution (integrated over theentire acceptor layer) as a function of in-plane momentum and energyfor a vertical (a) and a horizontal (b) dipole. The solid linecorresponds to the air light-line, and the dashed line marks the lightline for a medium with a refractive index of n = 1.5. (c) Donor−acceptor coupling efficiency as a function of emission wavelength,calculated as the fraction of donor emission absorbed in the acceptorlayer, for a vertical (blue line) and a horizontal (red line) dipole.

Figure 5. Calculated energy-transfer distribution for a structurewithout the bottom mirror and no cavity (a) and for a structure withboth Ag mirrors removed (b). We note that in (b), the separationbetween the donor and acceptor layers was kept to be 52 nm, which isidentical to the separation in the hybrid cavity structure, as well as in(a).

The Journal of Physical Chemistry C Article

DOI: 10.1021/acs.jpcc.8b03030J. Phys. Chem. C 2018, 122, 15853−15860

15857

repeat the same calculations with both Ag layers removed. Inthese calculations, we keep the distance between the donor andacceptor layers fixed (i.e., 52 nm) and replace the top Agmirror by a poly(vinyl alcohol) (PVA) spacer. Since thisdistance is much larger than the FRET radius and the plasmon-mediated energy transfer is absent, energy transfer can onlyoccur through a trivial radiative process. As seen in Figure 5b,this radiative energy transfer is uniformly distributed within theregion between the air light line and the glass light line,corresponding to free photons propagating at angles of 42−90°through the polymer−acceptor layer. In this configuration, wefind that the fraction of the donor energy absorbed by theacceptor molecules is around 15%. Therefore, we concludethat even though the incorporation of metal in our plasmonic−photonic system inevitably introduces losses, its effect on theenergy flow between the donor and acceptor still results inhigher energy-transfer efficiency, and that this enhancement isdue to the resonant coupling between the plasmonic andphotonic modes.

■ EXPERIMENTAL METHODS AND RESULTSTo demonstrate the long-distance energy transfer in the hybridcavity, we fabricated three types of cavity samples: one withboth the donor and acceptor molecules, and two referencesamples with either the donor alone or the acceptor alone. Wechose to work with fluorescein as the donor molecule andrhodamine B (RhB) as the acceptor molecule. The micro-cavities were prepared by sputtering a thin (35 nm) silver layeronto a glass substrate, followed by the deposition of a 164 nmlayer of poly(vinyl alcohol) (PVA, MW 205 000; Sigma-Aldrich). This layer was deposited by spin-coating a 3 wt%PVA/H2O solution at 1800 rpm. For the samples containingthe acceptor molecules, the polymer was doped withrhodamine B molecules (RhB, Sigma-Aldrich), which wereadded to the PVA solution before spin-coating in a 1:4700mass ratio. Next, a second silver layer (43 nm thick) wassputtered on top of the polymer to complete the cavity andcovered with a very thin PVA spacer (∼9 nm). The samespacer was deposited above all of the samples to avoid anydifferences in the dispersion of the three samples. Finally, forthe samples containing the donor molecules, we deposited thefluorescein molecules (Sigma-Aldrich) by spin-casting a 2.7 ×10−7 M fluorescein/EtOH solution at 1000 rpm. In ourexperiments, we used low dye concentrations for both thedonor and acceptor, to avoid strong light-molecule couplingand its contribution to the energy-transfer process.25−28

Figure 6a shows the excitation (dashed lines) and emission(solid lines) spectra for fluorescein (in blue) and RhB (red),measured using a HORIBA Jobin Yvon Flurolog FL-3spectrofluorometer. As seen in Figure 6a, the fluoresceinexcitation and emission peaks are around 503 and 523 nm,respectively, whereas the rhodamine B (RhB) has an excitationpeak at 555 nm and an emission peak at 577 nm. Next, weused a dedicated single-shot Fourier plane imaging system41 tomeasure the angle-resolved reflection of the samples, fromwhich the dispersion of the hybrid cavity can be obtained.Figure 6b shows the dispersion of the sample containing boththe donor and acceptor molecules (color image), on which thesimulated dispersion for a structure with the same layerthicknesses is superimposed (extracted from the T-matrixcalculations presented in Figure 1b). The splitting of thedispersion curve into two branches is clearly visible in themeasured dispersion and fits the simulation results. This

confirms that the plasmonic and the photonic cavity modeshybridize to form the composite modes in the sample, with themode profiles shown in Figure 1c. Furthermore, the coupled-mode energies are close to the emission peak of fluorescein(blue solid line) and the excitation peak of RhB (red dashedline), ensuring efficient coupling of the molecules to the hybridmodes. We verified that the two other samples (i.e., containingeither the donor or the acceptor alone) have the samecharacteristic dispersion (up to deviations of 30 meV, which issmaller than half-width at half-maximum of the absorptionspectrum of the molecules, being ∼70 meV).Finally, we proceeded to characterize the photolumines-

cence properties of the samples. The emission is collected fromthe cavity samples at 45°, whereas the excitation beam islaunched at 62.5° (with respect to the cavity normal). Asshown in Figure 6c,d, we performed measurements comparingthe donor/acceptor sample (D + A) and the reference sampleswith donor alone (D) and acceptor alone (A). When excitingthe sample containing the acceptor alone (red curve) at 470nm, we observe an emission peak around 540−550 nm. Wenote that this emission is slightly blue-shifted with respect tothe emission of the bare molecular film molecules, whichresults from the different optical environment and the angle-dependent out-coupling for the emission when exiting thecavity. This emission peak is not observed for the samplecontaining the donor alone (blue curve). In the mixed sample(black curve), we find a pronounced 3- to 4-fold emissionenhancement, suggesting increased efficiency of energy

Figure 6. (a) Excitation (dashed lines) and emission (solid lines)spectra measured for fluorescein (blue) and RhB (red) molecules. (b)Dispersion diagram of the fabricated sample, measured by angle-resolved reflection spectroscopy. The white dotted lines show thesimulated dispersion calculated for the same structure. The horizontalwhite lines correspond to the emission peak of fluorescein (solid line)and the excitation peak of RhB (dashed line). Emission (c) andexcitation (d) measurements performed on the samples containingdonor molecules (blue curve), acceptor molecules (red curve), andboth donor and acceptor molecules (black curve).

The Journal of Physical Chemistry C Article

DOI: 10.1021/acs.jpcc.8b03030J. Phys. Chem. C 2018, 122, 15853−15860

15858

injection into the RhB molecules in the cavity. To reveal theorigin for this enhancement, we performed excitation measure-ments on the samples, probing the acceptor emission (at 600nm) as a function of the excitation wavelength. As seen inFigure 6d, for the mixed sample (black curve), we find a clearexcitation peak at the donor absorption wavelength (∼500nm). This excitation peak does not appear in any of thereference samples (acceptor or donor alone; red and bluecurves, respectively), indicating that this signal is indeedtriggered by absorption into the donor molecules, whichsubsequently transfers their excitation energy to the acceptormolecules. Therefore, our result provides a proof-of-conceptthat the hybrid plasmon−photonic cavity can indeed mediateenergy transfer over a distance of ∼150 nm, as predicted byour simulations.

■ CONCLUSIONSIn summary, we studied energy transfer in hybrid structuressupporting both photonic cavity modes and surface plasmonmodes, both theoretically and experimentally. We showed thatwith an appropriate design these two different types of wavescan strongly couple to form composite modes, whichsimultaneously extend over the interior part of the structureas well its outer side. Most importantly, we demonstrated thatthese mixed modes can give rise to a new energy-transfermechanism and efficiently mediate energy transfer over largedistances. In a manner similar to earlier research, whichharnessed the plasmonic modes of metallic structures, ourmechanism gives rise to nonradiative energy transfer overdistances much larger than the typical FRET radius andcomparable to the optical wavelength. However, since ourmethod relies on the coupling between surface plasmons andphotonic modes, the length scale for energy transfer is nolonger set by the dimensions of the metallic part of the system.Therefore, the amount of metal can be significantly reduced,which lowers parasitic losses in the metal and increases theoverall energy-transfer efficiency. Our simulations show thatthe coupling efficiency provided by the structure is around 20−25%, and further study is required to understand the influenceof the structure parameters on the energy-transfer process andto optimize it, potentially obtaining even higher couplingefficiencies. Another advantage of our new mechanism is thatthe spectral properties of the structure can be easily tuned tomatch a large variety of different donor−acceptor pairs. Theplanar architecture of the hybrid structure naturally lends itselfto different organic electronic devices and may prove to bebeneficial in light-harvesting applications, in which solar energyneeds to be absorbed and then efficiently transported into theactive region inside the device, where it is converted toelectrical energy.

■ AUTHOR INFORMATIONCorresponding Author*E-mail: talschwartz@tau.ac.il.ORCIDTal Schwartz: 0000-0002-5022-450XAuthor ContributionsThe manuscript was written through contributions of allauthors. All authors have given approval to the final version ofthe manuscript.NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSThis study was supported by the Israel Science Foundation(Grant no. 1993/13) and the Marie Curie Career IntegrationGrant (no. PCIG12-GA-2012-618921). K.A. acknowledgesfinancial support from the Ministry of Science and Technology,Israel. A.G. acknowledges financial support from the Center forAbsorption in Science, Ministry of Absorption, Israel. Wethank P. Ginzburg for the useful discussions.

■ REFERENCES(1) Gersten, J. I.; Nitzan, A. Accelerated Energy Transfer BetweenMolecules Near a Solid Particle. Chem. Phys. Lett. 1984, 104, 31−37.(2) Hua, X. M.; Gersten, J. I.; Nitzan, A. Theory of Energy TransferBetween Molecules Near Solid State Particles. J. Chem. Phys. 1985, 83,3650−3659.(3) Andrew, P.; Barnes, W. L. Energy Transfer Across a Metal FilmMediated by Surface Plasmon Polaritons. Science 2004, 306, 1002−1005.(4) Celebi, K.; Heidel, T. D.; Baldo, M. A. Simplified Calculation ofDipole Energy Transport in a Multilayer Stack Using Dyadic Green’sFunctions. Opt. Express 2007, 15, 1762−1772.(5) Heidel, T. D.; Mapel, J. K.; Singh, M.; Celebi, K.; Baldo, M. A.Surface Plasmon Polariton Mediated Energy Transfer in OrganicPhotovoltaic Devices. Appl. Phys. Lett. 2007, 91, No. 093506.(6) Marocico, C. A.; Knoester, J. Effect of Surface-PlasmonPolaritons on Spontaneous Emission and Intermolecular Energy-Transfer Rates in Multilayered Geometries. Phys. Rev. A 2011, 84,No. 053824.(7) Collini, E.; Todescato, F.; Ferrante, C.; Bozio, R.; Scholes, G. D.Photophysics and Dynamics of Surface Plasmon Polaritons-MediatedEnergy Transfer in the Presence of an Applied Electric Field. J. Am.Chem. Soc. 2012, 134, 10061−10070.(8) Poudel, A.; Chen, X.; Ratner, M. A. Enhancement of ResonantEnergy Transfer Due to an Evanescent Wave from the Metal. J. Phys.Chem. Lett. 2016, 7, 955−960.(9) Golmakaniyoon, S.; Hernandez-Martinez, P. L.; Demir, H. V.;Sun, X. W. Cascaded Plasmon-Plasmon Coupling Mediated EnergyTransfer Across Stratified Metal-Dielectric Nanostructures. Sci. Rep.2016, 6, No. 34086.(10) Druger, S. D.; Arnold, S.; Folan, L. M. Theory of EnhancedEnergy Transfer Between Molecules Embedded in SphericalDielectric Particles. J. Chem. Phys. 1987, 87, 2649−2659.(11) Lakowicz, J. R.; Kusba, J.; Shen, Y.; Malicka, J.; D’Auria, S.;Gryczynski, Z.; Gryczynski, I. Effects of Metallic Silver Particles onResonance Energy Transfer Between Fluorophores Bound to DNA. J.Fluoresc. 2003, 13, 69−77.(12) Reil, F.; Hohenester, U.; Krenn, J. R.; Leitner, A. Forster-TypeResonant Energy Transfer Influenced by Metal Nanoparticles. NanoLett. 2008, 8, 4128−4133.(13) Lessard-Viger, M.; Rioux, M.; Rainville, L.; Boudreau, D. FRETEnhancement in Multilayer Core−Shell Nanoparticles. Nano Lett.2009, 9, 3066−3071.(14) Lunz, M.; Gerard, V. A.; Gun’ko, Y. K.; Lesnyak, V.; Gaponik,N.; Susha, A. S.; Rogach, A. L.; Bradley, A. L. Surface PlasmonEnhanced Energy Transfer between Donor and Acceptor CdTeNanocrystal Quantum Dot Monolayers. Nano Lett. 2011, 11, 3341−3345.(15) Asselin, J.; Viger, M. L.; Boudreau, D. Metal-EnhancedFluorescence and FRET in Multilayer Core-Shell Nanoparticles.Adv. Chem. 2014, 2014, 1−16.(16) Pustovit, V. N.; Urbas, A. M.; Shahbazyan, T. V. EnergyTransfer in Plasmonic Systems. J. Opt. 2014, 16, No. 114015.(17) de Torres, J.; Ferrand, P.; Colas des Francs, G.; Wenger, J.Coupling Emitters and Silver Nanowires to Achieve Long-RangePlasmon-Mediated Fluorescence Energy Transfer. ACS Nano 2016,10, 3968−3976.(18) Marocico, C. A.; Zhang, X.; Bradley, A. L. A TheoreticalInvestigation of the Influence of Gold Nanosphere Size on the Decay

The Journal of Physical Chemistry C Article

DOI: 10.1021/acs.jpcc.8b03030J. Phys. Chem. C 2018, 122, 15853−15860

15859

and Energy Transfer Rates and Efficiencies of Quantum Emitters. J.Chem. Phys. 2016, 144, No. 024108.(19) Atwater, H. A.; Polman, A. Plasmonics for ImprovedPhotovoltaic Devices. Nat. Mater. 2010, 9, 205−213.(20) Li, J.; Cushing, S. K.; Meng, F.; Senty, T. R.; Bristow, A. D.;Wu, N. Plasmon-Induced Resonance Energy Transfer for SolarEnergy Conversion. Nat. Photonics 2015, 9, 601−607.(21) Hobson, P. A.; Wedge, S.; Wasey, J. A. E.; Sage, I.; Barnes, W.L. Surface Plasmon Mediated Emission from Organic Light-EmittingDiodes. Adv. Mater. 2002, 14, 1393−1396.(22) An, K. H.; Shtein, M.; Pipe, K. P. Surface Plasmon MediatedEnergy Transfer of Electrically-Pumped Excitons. Opt. Express 2010,18, 4041−4048.(23) Hsu, L.-Y.; Ding, W.; Schatz, G. C. Plasmon-CoupledResonance Energy Transfer. J. Phys. Chem. Lett. 2017, 8, 2357−2367.(24) Bouchet, D.; Cao, D.; Carminati, R.; De Wilde, Y.;Krachmalnicoff, V. Long-Range Plasmon-Assisted Energy Transferbetween Fluorescent Emitters. Phys. Rev. Lett. 2016, 116, No. 037401.(25) Coles, D. M.; Somaschi, N.; Michetti, P.; Clark, C.; Lagoudakis,P. G.; Savvidis, P. G.; Lidzey, D. G. Polariton-Mediated EnergyTransfer Between Organic Dyes in a Strongly Coupled OpticalMicrocavity. Nat. Mater. 2014, 13, 712−719.(26) Zhong, X.; Chervy, T.; Wang, S.; George, J.; Thomas, A.;Hutchison, J. A.; Devaux, E.; Genet, C.; Ebbesen, T. W. Non-Radiative Energy Transfer Mediated by Hybrid Light-Matter States.Angew. Chem., Int. Ed. 2016, 55, 6202−6206.(27) Zhong, X.; Chervy, T.; Zhang, L.; Thomas, A.; George, J.;Genet, C.; Hutchison, J. A.; Ebbesen, T. W. Energy Transfer BetweenSpatially Separated Entangled Molecules. Angew. Chem., Int. Ed. 2017,56, 9034−9038.(28) Du, M.; Martínez-Martínez, L. A.; Ribeiro, R. F.; Hu, Z.;Menon, V. M.; Yuen-Zhou, J. Theory for Polariton-Assisted RemoteEnergy Transfer. 2017, arXiv:1711.11576. arXiv.org e-Print archive.https://arxiv.org/abs/1711.11576 (accessed Jan 1, 2018).(29) Barnes, W. L. Fluorescence near interfaces: The Role ofPhotonic Mode Density. J. Mod. Opt. 1998, 45, 661−699.(30) Jiang, J.-J.; Xie, Y.-B.; Liu, Z.-Y.; Tang, X.-M.; Zhang, X.-J.; Zhu,Y.-Y. Amplified Spontaneous Emission via the Coupling BetweenFabry-Perot Cavity and Surface Plasmon Polariton Modes. Opt. Lett.2014, 39, 2378−2381.(31) Ford, G. W.; Weber, W. H. Electromagnetic Interactions ofMolecules with Metal Surfaces. Phys. Rep. 1984, 113, 195−287.(32) Yeh, P. Optical Waves in Layered Media; Wiley, 2005.(33) Sarid, D. Long-Range Surface-Plasma Waves on Very ThinMetal Films. Phys. Rev. Lett. 1981, 47, 1927−1930.(34) Berini, P. Long-Range Surface Plasmon Polaritons. Adv. Opt.Photon. 2009, 1, No. 484.(35) Rechberger, W.; Hohenau, a; Leitner, a; Krenn, J. R.;Lamprecht, B.; Aussenegg, F. R. Optical Properties of Two InteractingGold Nanoparticles. Opt. Commun. 2003, 220, 137−141.(36) Nordlander, P.; Oubre, C.; Prodan, E.; Li, K.; Stockman, M. I.Plasmon Hybridization in Nanoparticle Dimers. Nano Lett. 2004, 4,899−903.(37) Chance, R. R.; Prock, A.; Silbey, R. Adv. Chem. Phys. 2007, 1−65.(38) Hartman, R. L. Green Dyadic Calculations for InhomogeneousOptical Media. J. Opt. Soc. Am. A 2000, 17, 1067−1076.(39) Dung, H. T.; Knoll, L.; Welsch, D.-G. Intermolecular EnergyTransfer in the Presence of Dispersing and Absorbing Media. Phys.Rev. A 2002, 65, No. 043813.(40) Wubs, M.; Vos, W. L. Forster Resonance Energy Transfer Ratein Any Dielectric Nanophotonic Medium With Weak Dispersion. NewJ. Phys. 2016, 18, No. 053037.(41) Ohad, A.; Akulov, K.; Granot, E.; Rossman, U.; Patolsky, F.;Schwartz, T. Spatially-Resolved Measurement of Plasmon DispersionUsing Fourier-Plane Spectral Imaging. Photon. Res. 2018, 6, 653−658.

The Journal of Physical Chemistry C Article

DOI: 10.1021/acs.jpcc.8b03030J. Phys. Chem. C 2018, 122, 15853−15860

15860