surface enhanced fluorescence - turro's organic photochemistry

IOP PUBLISHING JOURNAL OF PHYSICS D: APPLIED PHYSICS

J. Phys. D: Appl. Phys. 41 (2008) 013001 (31pp) doi:10.1088/0022-3727/41/1/013001

TOPICAL REVIEW

Surface enhanced fluorescenceEmmanuel Fort1 and Samuel Gresillon2

1 Centre d’Imageries Plasmoniques Appliquees (CIPA) and Laboratoire Materiaux et PhenomenesQuantiques, UMR 7162 CNRS and Universite Paris Diderot - Paris 7, 10 rue Alice Domon et LeonieDuquet, 75205 Paris Cedex 13, France2 Centre d’Imageries Plasmoniques Appliquees (CIPA) and Laboratoire Photons Et Matiere, UPR A0005CNRS and Universite P & M Curie Paris 6, Laboratoire d’Optique Physique, ESPCI, 10 rue Vauquelin,75231 Paris Cedex 5, France

E-mail: [email protected] and [email protected]

Received 18 July 2007Published 17 December 2007Online at stacks.iop.org/JPhysD/41/013001

AbstractFluorescence is widely used in optical devices, microscopy imaging, biology, medical researchand diagnosis. Improving fluorescence sensitivity, all the way to the limit of single-moleculardetection needed in many applications, remains a great challenge. The technique of surfaceenhanced fluorescence (SEF) is based upon the design of surfaces in the vicinity of the emitter.SEF yields an overall improvement in the fluorescence detection efficiency throughmodification and control of the local electromagnetic environment of the emitter. Near-fieldcoupling between the emitter and surface modes plays a crucial role in SEF. In particular,plasmonic surfaces with localized and propagating surface plasmons are efficient SEFsubstrates. Recent progress in tailoring surfaces at the nanometre scale extends greatly therealm of SEF applications. This review focuses on the recent advances in the differentmechanisms involved in SEF, in each case highlighting the most relevant applications.

(Some figures in this article are in colour only in the electronic version)

Introduction

Pushing fluorescence detection to the limit of sensitivity isperformed by controlling the local electromagnetic (EM)environment of the fluorophores. The effect of the surroundingmedia on the emission properties of a molecular emitter(Purcell effect) was first demonstrated using free atoms inhigh-Q cavity in the search for a single photon emitter [1].Taking advantage of the interaction between an emitter and itssurroundings allows one to dramatically improve the detectionefficiency of fluorescence. These improvements in lightemission/detection have been used in various systems, suchas organic molecules [2, 3], semiconductor device structures,micropillars [4], microdiscs [5] and more recently in photoniccrystal defect nanocavities [6–8].

The ability to improve emission of a single emitter ina microcavity, although very interesting from a fundamentalpoint of view, often involves complex geometries and is hardlyapplicable to ensemble measurements. Surface enhancementinvolving a 2D geometry is usually more straightforward. We

show in this paper how surfaces can be designed and tailoredto enhance fluorescence by several orders of magnitude.

The effect of the surroundings changes the emission ofa fluorophore through its spontaneous emission rate and itsangular emission pattern. As we will see, a crucial challengein surface enhanced fluorescence (SEF) is the surface itself,designed to control the fluorophore emission and bring towardsthe detector the maximum amount of light emitted by asingle source located on or near the surface. Usually, thepresence of the surface also modifies the excitation rate ofa fluorophore through the modification of the local EM field atthe fluorophore position. The optimization of this excitationprocess is another crucial aspect of SEF which involves aspecifically designed surface together with a precise excitationand optical geometry.

One of the most promising developments in the fieldof SEF is the mastering of subwavelength optical patternsand nanostructures with plasmonic properties [9]. Surfaceplasmons (SP) are collective charge oscillation modes atthe surfaces of conductors with unique light interaction

0022-3727/08/013001+31$30.00 1 © 2008 IOP Publishing Ltd Printed in the UK

http://dx.doi.org/10.1088/0022-3727/41/1/013001

mailto: [email protected]

mailto: [email protected]

http://stacks.iop.org/JPhysD/41/013001

J. Phys. D: Appl. Phys. 41 (2008) 013001 Topical Review

properties. Due to their high sensitivity to geometry andsurroundings, SP allow one to channel and concentrate lightwithin subwavelength volumes [9]. Plasmonic surfaceshave been used for a variety of optical applications whereenhancement of the EM fields is of importance, such as innon-linear phenomena [10–13].

Very interesting and well documented review articles havebeen published focusing on specific aspects of SEF, for instancethe photonic mode density [14], or radiative decay engineeringin biological applications [15–20]. In this review, we presentthe latest developments of this fast growing field boosted,in particular, by the ability to design nanometric plasmonicstructures and objects. We address the case of surfaces aswell as that of nanoparticles with respect to their enhancementproperties.

This review is divided into three parts. The first partintroduces the principles of SEF and its applications to simplegeometries. The two other parts are devoted to SEF forfluorophores coupled to propagating surface plasmons (part 2),and to SEF for fluorophores coupled to localized enhancementof the EM field (part 3).

1. SEF Principles

1.1. Principles of Fluorescence

Fluorescence is an example of more general luminescencephenomena which are defined as emission of light fromany substance occurring by radiative relaxation from anelectronically excited state. Luminescence is calledfluorescence when excitation is induced by light (also calledphotoluminescence) and when the excited electronic state isa singlet state (compared with phosphorescence for a tripletstate). Fluorescence is thus the property for a material to absorblight at a particular wavelength and to subsequently emit lightof longer wavelength. The processes which occur between theabsorption and the emission of light are usually illustrated bya Jabłonski diagram.

Figure 1 shows a simplified version of the diagram whichillustrates few molecular processes which occur in the excitedstates. When a fluorophore absorbs light energy, it is usuallyexcited to a higher vibrational energy level in the first S1 orsecond S2 electronically excited state, before relaxing rapidlyto the lowest exited energy level in about a picosecond orless (this process is called internal conversion). Fluorescencelifetime τ , defined as the average relaxation time to theelectronic ground state, is typically four orders of magnitudesmaller than the lifetime of internal conversion3, giving thefluorophore enough time to reach the thermally equilibratedlowest vibrational state of S1 [21].

The fluorescence quantum yield or quantum efficiency Q

is a fundamental parameter used to retrieve the fluorescenceefficiency of a given fluorophore [21]. It is commonlydefined as the probability that a given excited fluorophore willproduce a fluorescence photon. In this review, we will use a

3 Internal conversion time τi � few picoseconds whereas fluorescencelifetime τ � few nanoseconds.

S1

S2

S0

Absorption

Internal Conversion

Fluorescence

Figure 1. Simplified Jabłonski diagram illustrating the molecularprocesses involved during a fluorescence excitation and emissioncycle.

more general definition for Q in which the quantum yield isdefined as:

Q = �EM

�EM + �other= τ

τEM, (1)

where �EM represents all the EM relaxation processesand �other are related to non-radiative relaxation processesinduced by direct molecular interaction, essentially molecularcollision [22]. Some EM relaxation processes can be non-radiative. When a fluorophore is surrounded by an absorbingmedium, energy is dissipated into heat through near-fieldcoupling. τEM = 1/�EM is the lifetime of the excited statein the absence of non-EM processes. In the common case of afree fluorophore in an homogeneous non-absorbing medium,the definition 1 of Q is equivalent to the classical one.

The presence of an interface in the vicinity of thefluorophore alters the fluorescence processes through themodification of its local EM environment. As we willshow later on, such changes induce modifications of both theexcitation and the emission processes. The internal conversiondepends mainly on the electronic structure of the fluorophorethrough the overlap of the fluorophore wave functions. Toa first approximation it is insensitive to the modification ofthe local EM environment. This internal process allows oneto separate the influence of the surface on the fluorescencecycle in two independent steps associated with excitation andemission.

1.2. Modification of the molecular detection efficiency

The SEF efficiency can be characterized by the gain in therelative molecular detection efficiency (MDE). The MDE isdefined as the number of detected photons versus the numberof absorbed photons. The associated MDE function, MDE(r0)at a position r0, can be written as follows:

MDE(r0) = �exc(r0)︸︷︷︸excitation

× Q × MCE(r0)︸︷︷︸emission

. (2)

The MDE function is the product of an excitation factorgiven by the excitation rate �exc(r0) and an emission factorwhich is the product of the fluorescence quantum yield Q andthe molecular collection efficiency function MCE(r0) at pointr0. The latter represents the ratio of the detected and emittedphotons.

2


The influence of an interface on the excitation rate �exc(r0)

is straightforward. Since �exc(r0) increases with light intensityIexc(r0) at the fluorophore position, the design of an efficientSEF surface thus consists in increasing the EM field at thefluorophore position, taking into account that fluorophore canbe excited either by a propagating or an evanescent EM field.Similarly, an efficient SEF surface should maximize the MDEthrough the redirection of light towards the detector.

The ability for an interface to modify the relaxationprocesses, i.e. Q, is more complex and less intuitive. Fromsimple inspection of the Jabłonski diagram, one may thinkthat fluorescence emission is an intrinsic property of thefluorophore, while in reality it crucially depends on itssurroundings.

Fluorescence emission is a spontaneous emission processhence its rate of relaxation is given by the coupling of theexcited state of the molecule to the vacuum oscillations ofthe surrounding environment. This can be reformulated inclassical terms where the probability of emitting a photonis related to the photonic mode density (PMD) [14]. Thefluorophore emission can consequently be controlled throughthe modification of the electromagnetic boundary conditionsnear the fluorophore. The alteration of the optical modestructure available to a 3D confined dipole in a small cavity hasbeen studied as early as 1946 by Purcell [1]. First experimentalstudies were carried out by Drexhage in the early 1970s andinvolved changes in the emission characteristics from dyemolecules placed close to reflecting interfaces [23].

The fluorescence relaxation processes which areassociated with the radiative recombination of the excitedelectron–hole pair compete with other non-fluorescent relax-ation processes. SEF partly relies on the ability to tailor thelocal environment of the molecule to maximize the radiativerelaxation rate compared with the one of the same fluorophorein free space. This is done while maintaining the competi-tive non-radiative processes to a low level. In most cases thisresults in the reduction of the fluorescence lifetime and in anincrease of the quantum efficiency of the fluorophore. This iswhy SEF is sometimes incorrectly reduced to radiative decayengineering (RDE) [17]. Since RDE is based only on engi-neering the emission, it applies to every luminescent process(chemiluminescence, electroluminescence, etc) and does nottake advantage of all the other aspects of fluorescence.

1.3. A simple model for the modifications of relaxationprocesses

1.3.1. The classical model. A classical theoretical approachfor the calculation of the relaxation processes induced bythe modification of the environment gives accurate resultsand simple understanding of the underlying principle. Thisclassical model is sometimes called ’CPS model’ after Chance,Prock and Silbey, the authors who introduced 30 years agothe classical approach in the case of a planar interface [24].Neglecting the fluorophore spatial extension, the excitedfluorophore behaviour can be modelled as an oscillating dipole.The fundamental law of dynamics applied to the excited

electron gives the following equation for the dynamics of thedipolar moment p

d2pdt2

+ �0dpdt

+ ω20p = e2

mEloc(r0), (3)

where ω0 and �0 are, respectively, the resonance frequencyand the oscillator damping rate in free space4 and Eloc(r0)

is the electric field at the dipole position r0 resulting fromthe reflection of the emitted electric field off the surroundinginterfaces. Eloc(r0) equals zero in free space. Under thecondition of small damping (�0 < 2ω0), which is usuallyvalid for experimental conditions, the solution is given byp = p0 exp(−�t) exp(iωt) where

� = �0 +e2

mω0|p|2 �m(p∗ · Eloc(r0)),

�ω = ω − ω0 = − �20

8ω0− e2

2mω0|p|2 �e(p∗ · Eloc(r0)), (4)

where the superscript ∗ stands for the complex conjugate.These expressions give the modification of the emission

rate and the frequency shift due to the presence of an interface,as compared with free space emission. The influence of thesurrounding on the relaxation and frequency shift of the dipoleis expressed by the local field reflected by the environmentat the dipole position r0. It can also be written using theelectric field susceptibility S(r, r0, ω0) [24,25] or alternativelythe electric Green function G [26]5. This tensor S is also calledthe linear susceptibility of the electric field due to the presenceof the surrounding. The reflected electric field is given byEloc(r) = S(r, r0, ω0) · p.

For fluorophores with fluorescence quantum efficiencyQ = 1, the damping rate in free space �0 is given by the powerradiated by a dipole in free space and the energy conservation6:

�0 = e2ω20

6πmε0c3. (5)

Using this expression (5), equations (4) can be rewrittenin the following form:

�

�0= 1 +

6πε0

k3�m[u · S(r, r0, ω0)u],

�ω

�0= − �0

8ω0− 3πε0

k30

�e[u · S(r, r0, ω0)u], (6)

where u is the unit vector along the direction of the transitiondipole p. The influence of the environment depends only onthe local surroundings through the tensor S. It is independentof the fluorophore itself.

1.3.2. The quantum approach. Since spontaneous emissionis fundamentally a quantum process, it is interesting to comparethis result with the quantum description. The quantumequivalent of the classical relaxation rate � is the rate of

4 �0 is the damping rate associated with the relaxation of the excitation.5 We choose to define the Green function in such a way that for an electricdipole in r0, E(r) = G(r, r0) · p (see section 3.2.2).6 For Q < 1, �0 is defined by Q�0 = e2ω2

0/6πmε0c3.

3


spontaneous emission. For a two level system coupled to itssurrounding, the rate of spontaneous emission is calculated bythe perturbation method and is given by Fermi’s golden rule:

� = 2π

h2

∑kµ

|〈f, 1kµ|p · Evac(r0)|e, 0〉|2δ(ωkµ − ωeg), (7)

where p is the electric dipole operator and E(r0) the electricfield operator evaluated at the dipole’s position r0. Thisformula explicitly shows that the excited state |e〉 dumps itsenergy hωeg into an empty mode of the vacuum field, creatingone photon |1kµ > with a wavevector k and polarization µ.The decay rate is the sum over all the photonic modes of thesurrounding space. The effect of the nearby surface or objecton the fluorophore damping rate and frequency shift originatesin the quantum model from the alteration of the PMD [14].

For a fluorophore in free space, the spontaneous emissionrate is equal to [27]:

�0 = ω3

3πhε0c3|p0|2|feg|2, (8)

where feg is the Franck–Condon factor between the excited andthe ground state. This result differs from the classical formulagiven in equation (5). However the relative modifications ofthe emission rates due to the presence of a nearby object areidentical. Hence, the classical model gives an exact result forthe normalized emission rate [14]:

�

�0|classical = �

�0|quantum. (9)

The calculated frequency shifts, however, differsignificantly between the classical and the quantumdescription. In the quantum case, they originate from thecoupling between the molecule and the quantum fluctuations.The classical model underestimates the frequency shifts, aquantum analysis is needed to get a precise estimate [14].Experimentally, the measured shifts are rather small. Theyincrease with decreasing lifetime, from a few megahertz forEu3+ with τ ≈ 1 ms to a gigahertz for Na atoms withτ ≈ 1 ps [28].

1.4. SEF engineering

The relation between the maximization of the EM relaxationprocesses and SEF is not straightforward since EM processesinclude both radiative and non-radiative processes. Thus,it is via the modification of the relaxation rate of afluorophore that the contribution to the radiative emission canbe understood. The energy transfer between the fluorophoreand the surface through near-field components can be quiteimportant and sometimes represents the major part of therelaxation processes. Nevertheless the energy flow betweenthe fluorophore and the surface if eventually associated withradiative processes will thus contribute to the enhancementof the fluorescence. From this point of view, it is essentialto understand the fluorophore and its surrounding as a singlesystem to evaluate its global radiating yield. The art of SEFengineering relies in particular on the ability to maximize theradiative relaxation processes from the surface.

The main challenge of RDE for SEF is thus tocontrol and design surface structures which maximize thefluorescence enhancement of a fluorophore by adjustingthe local electric field Eloc (or equivalently the PMD).The modified damping rate of the fluorophore � is obtainedexperimentally using Fluorescence Lifetime Measurementsthrough the relation τ = 1/�.

1.5. Example of simple flat surfaces

The simplest system showing significant modification of thefluorophore excitation and emission processes is a planarinterface below a fluorescent particle.

1.5.1. Relaxation modification on a flat surface. Usingthe theory developed by CPS, it is convenient to evaluatethe power coupled from the source as a function of the in-plane component of the wavevector (u) since the latter is aninvariant quantity in the planar structures considered. The totalmodification of the spontaneous emission rate is evaluatedby integrating the modification to the damping rate over allin-plane wavevectors [24]. The modification to the relativedamping rate is given by

�

�0= (1 − Q) + Q

∫ ∞

0I (u) du, (10)

where the integrand I (u) is the power dissipated by the emitteras a function of the in-plane wavevector u. This integranddepends on the dipole orientation with respect to the surface.

In the case of a dipole positioned at a distance d from aninterface between a medium 1 and 2, the relative damping ratefor a dipole oriented orthogonally and parallel to the surfacerespectively becomes [24]

�⊥�∞

= 1 − 3

2Q.�m

∫ ∞

0rpe−�ϕ1

u3

l1du,

�‖�∞

= 1 +3

4Q.�m

∫ ∞

0[(1 − u2)rp + rs]e

�ϕ1u

l1du, (11)

where rp = (ε1l2−ε2l1)/(ε1l2 +ε2l1) and rs = (l1−l2)/(l1 +l2)

are, respectively, the Fresnel reflection coefficients for the p

polarization and the s polarization; li = −i(εi/ε1 − u2)−1/2 isthe normalized component orthogonal to the surface; �ϕ1 =−2k1l1d is the phase shift associated with the reflection on theinterface 1–2.

1.5.2. Transparent interface. When the fluorophores areimmobilized on a transparent dielectric (glass or fused silica)the presence of the interface induces changes in their emissionproperties. From an electrodynamic point of view, the presenceof the interface has a rather moderate effect on the emissionrate. For example, if molecules are adsorbed on a glass surface(refractive index = 1.5) from a water solution (refractiveindex = 1.33), equation (11) predicts that the relative radiativerate is enhanced by a factor of 1.07 and 1.33, dependingon the molecules emission dipole orientation (parallel andperpendicular, respectively) of the emission dipole of themolecule with respect to the glass surface. This modification

4


Figure 2. Angular distribution of the fluorescence intensity of amolecule positioned at a water–glass interface averaged over allpossible emission dipole orientations. Reprinted with permissionfrom [29]. Copyright 2000 American Chemical Society.

is modest because of the relatively small reflected field atthe fluorophore position. However the alteration of thefluorophore lifetime is in the range of lifetime variationsmeasured in biological studies [30].

The presence of the interface significantly modifies theangular emission of the fluorophores. Figure 2 shows thecalculated angular distributions of the fluorescence intensityof a molecule placed at a water–glass interface averaged overall possible emission dipole orientations. The discontinuityin the refractive index produces a significant maximum ofthe emission in the direction of the critical angle of totalinternal reflection [29]. Between 68% and 77% of thefluorescence is emitted towards the glass side. Thus lightcollection efficiency is enhanced when the detector is placedon the glass side [31]. However a significant part of thefluorescence radiation is emitted into the glass side above thecritical angle. This radiation is sometimes called ‘forbiddenlight’ or ‘supercritical light’. It accounts for 34% of the totalfluorescence emission. The forbidden light is of particularinterest when detecting surface-generated fluorescence dueto the fact that the electromagnetic coupling of the radiatingfluorophore to the glass decays rapidly with distance r fromthe glass surface. The restriction of the detection volumenext to the surface is even stronger than in the case of anevanescent wave excitation [31]. For instance, when dyesemitting at visible wavelengths are used, the forbidden lightsignal stems only from molecules closer than 100 nm to theglass surface. Detection systems for biosensing applicationsbased on objective lenses or substrates with paraboloidalshapes have been developed based on this principle. Thisimaging technique is called ‘supercritical angle fluorescencemicroscopy’ [32].

1.5.3. Reflecting surfaces. Reflecting surfaces provideefficient SEF enhancement. Such substrates are typicallyobtained from semiconductor wafers or by coating standardglass slides with a multilayered dielectric Bragg mirror or ametallic thin film. The presence of the stationary antinode ofthe excitation field in the upper half space above the mirrorimproves the excitation by about a factor of four comparedwith a standard glass slide.

This amplification can be doubled up by a distinctenhancement of the emission light based on the presence ofan interference pattern for a dipole near a surface [33]. Thepresence of the reflecting substrate induces changes in thefluorophore relaxation processes since the emitter interfereswith the reflected EM waves. If the reflected field is in phasewith the emitter, the dipole will be driven harder (called super-radiance), which enhances the emission. In the opposite way,the oscillation strength will be reduced if the reflected waveis out of phase (called sub-radiance). The oscillation of thephase of the reflected light with increasing distance induces asubsequent oscillation of the spontaneous emission rate. Atthe same time, the oscillation strength decreases with distancebecause the emitters are point light sources. The radiation fieldof a dipole weakens with increasing distance and so does thefield reflected at metallic interfaces. This is directly observedthrough lifetime measurements. Drexhage and coworkersconfirmed this classical picture with Eu3+ ions in front of asilver mirror [23, 34].

However, for small emitter–surface distances (<100 nm),it is essential to take into account the near-field componentsof the dipole and calculate the reflected local electric field atthe dipole position. This is in particular the case when onewants to explain the strong quenching of fluorescence neara metallic mirror. In this region the excited molecule maydecay without radiation via coupling to guided waves such assurface plasmons and/or lossy waves. This additional couplingbetween the fluorophore and the substrate plays a major rolein SEF and is addressed in the next section.

The angular emission of the fluorophore is also greatlymodified by the presence of the mirror. When placed at anappropriate distance, the direct and reflected fields add up inthe far field. Depending on the direction and the wavelengthof light, this can give on average a 4-fold improvement of thesignal intensity and a multilobe angular emission pattern [23].

The sum of the various enhancement processes reachesan overall 10- to 15-fold improvement of the signal comparedwith a standard glass slide in the optical configuration ofepifluorescence [35]. Besides, the excitation standing wavemodulation allows one to evaluate the axial position of afluorophore with a nanometre accuracy [36]. This high spatialselectivity has been used to develop a new imaging techniquecalled fluorescence interference contrast (FLIC) microscopy.

These reflecting slides are of particular interest inbiological sensing such as DNA chips, where the fluorophoresare positioned at the very surface. Commercial mirrors arecurrently available for this purpose. These slides also yield anaverage signal amplification over micrometre thick samples,with interesting applications to cell and tissue imaging [37,38].

Fluorescence enhancement with reflecting slides is alsoapplied in fluorescence correlation spectroscopy (FCS). FCS

5


(a) (b) (c)

Figure 3. (a) Schematic representation of an SPP propagating along a metal/dielectric interface. + and − represent the regions with lowerand higher electron density, respectively. SPPs are transverse magnetic modes, and the generation of surface charge requires an electric fieldnormal to the surface. (b) The field decreases exponentially with the distance |z| from the surface. (c) The dispersion curve for an SPP modeshows the momentum mismatch that must be overcome in order to couple light and SPP modes together, with the SPP mode always lyingbeyond the light line, that is, it has greater momentum (kSP) than a free space photon (k0) of the same frequency ω. Reprinted from [9] bypermission from Macmillan Publishers Ltd: Nature, copyright 2003.

is a powerful technique for single-molecule dynamics inbiological systems. One limitation is the collection efficiencyof the fluorescence signal which is in many cases too lowto overcome the background noise and does not allow fastanalysis. Such limitation is reduced using reflecting surfaces.Mirrors with high reflectivity in the fluorophore emissionband allow redirection of the emitted fluorescence towardsthe collection optics. In addition, the standing waves of theexcitation light provide a well-defined length scale which isnecessary for quantitative analysis of the FCS data [39].

2. Coupling with propagating surface plasmon

2.1. Basics on surface plasmons

A plasmon is the quantum of the collective excitation of freeelectrons in solids. Surface plasmons are electron plasmaoscillations near a metal surface that stem from the brokentranslational invariance in the direction perpendicular to thesurface. The charge oscillations are orthogonal to the surfaceplane and induce at the surface an evanescent EM field which istransverse magnetic (generation of surface charge requires anelectric field normal to the surface). The combined surfaceplasmon and induced photon is called a surface plasmonpolariton (SPP). Figure 3(a) shows a schematic of the SPPpropagating at the metal–dielectric interface. The EM fieldin the direction perpendicular to the surface is evanescent,reflecting the non-radiative nature of SPPs which preventsenergy from propagating away from the surface. In thedielectric medium above the metal, typically air or glass, thedecay length of the field is of the order of half the wavelengthof the light, whereas the decay length into the metal is givenby the skin depth (see figure 3(b)) [9].

Much can be understood about the coupling betweenfluorophores and SPPs by looking at their dispersion relationof SPP, i.e. the relationship between the angular frequency(ω) and the in-plane wavevector (k‖) of SPP modes (the in-plane wavevector is the wavevector of the mode in the plane ofthe surface along which it propagates). For light propagating

with an incident angle θ with the normal of the surface in amedium of relative permittivity εd, the dispersion relationshipis simply given by k‖ = nd(ω/c) sin θ , where c is the speed oflight and nd = √

εd is the refractive index of the medium. Thedispersion relationship between the frequency and the in-planewavevector for SPPs propagating along the interface betweena metal and a dielectric can be found by looking for surfacemode solutions of Maxwell’s equations. Under appropriateboundary conditions, the dispersion relation obtained is [40]

kSPP = ω

c

√εdεm

εd + εm. (12)

With such a dispersion curve, the SPP mode alwayslies beyond the light line, see figure 3(c). This means thatSPPs have greater momentum (kSPP) than free space photon(k = ω/c) of the same pulsation ω. Because of this momentummismatch between SPP and incident plane-wave light, thesemodes cannot be coupled simply without providing the missingmomentum.

However, fluorophores on the surface are coupled toSPPs through their near-field components. This may providewavevector, large enough for resonant coupling. The energyflow from the fluorophore to the SPP modes is in most casesdetrimental to the SEF process. After a quantitative evaluationof this relaxation channel, we will show how part of this energycan be ultimately recovered into light and detected.

2.2. Relaxation processes on flat metallic thin films

As already mentioned in the previous section, since the metalsurface acts as a mirror, the emitter interferes with the reflectedelectromagnetic waves and the spontaneous emission rateoscillates with increasing distance. For small emitter–surfacedistances this simple picture needs to be refined. The excitedmolecule may decay non-radiatively through its near-fieldcomponents via coupling to guided waves such as surfaceplasmons and/or lossy waves. These decay channels depend onthe fluorophore–metal distance and on the dipole orientation.The classical model presented in section 1.3.1 allows one to

6


Figure 4. Dissipated power for an isotropic distribution of dipoleorientations as a function of the normalized in-plane wavevector uobtained using equations (11). The figure shows the powerdissipated by the emitter for a range of emitter-surface separations,as indicated. The system consists of an emitter having a wavelengthof emission 614 nm, immersed in vacuum and positioned above aAg surface. The dielectric constant of Ag is −16 + i0.6. From [14],reprinted by permission from Taylor and Francis Ltd(http://www.informaworld.com).

observe and evaluate quantitatively each relaxation channelthrough the amount of energy transfer � versus the in-planewavevector k‖. In the case of a flat surface equations (11)are used.

Figure 4 shows the relative energy flow versus thenormalized in-plane wavevector u = k‖/kd for variousfluorophore–surface distances in the case of a silver/airinterface, kd = ndko being the wavevector and nd the refractionindex in the dielectric.

• Emission of photons. The processes associated with theemission of photons in free space are given by the part ofthe diagram u < 1. Half of the total power is emittedin the upper half space and the other half is emittedtowards the substrate. The latter can be divided into areflected, a transmitted and an absorbed component. Theirrelative amplitudes depend on the reflection properties ofthe substrate. In practice when using an objective lensplaced above the substrate to collect the emitted light, witha numerical aperture NA, the power of the collected lightequals half the power represented by the curve in figure 4,integrated over u < NA/nd, plus the reflected part onthe substrate (readily evaluated using Fresnel formulae).Angular emission pattern of the fluorophore can also becalculated from equations (11) since k‖ = kd sin θ [23].

• Coupling with SPP modes. Excitation of SPP modesis clearly visible in figure 4. It is associated with thesharp peak centred at the in-plane wavevector of the SPPkSPP in the wavevector spectrum. The part of the energyflowing from the fluorophore to the SPP can be evaluatedby integrating the curve around the value kSPP [24]. Asthe fluorophore-surface distance d decreases, its couplingwith the SPP is increasingly more important. SPPcoupling is a prevailing process typically between 20 and200 nm, see figure 5.

Figure 5. Normalized fraction of power dissipated in eachrelaxation channel, i.e. radiation, SPP mode and lossy waves, forvarious fluorophore–surface distances (a) for a dipole orientedperpendicular to the surface, (b) parallel to the surface and (c) for anisotropic distribution. From [14], reprinted by permission fromTaylor and Francis Ltd (http://www.informaworld.com).

• Non-radiative transition and exciton coupling. Forfluorophore–metal distances<20 nm, fluorescence quench-ing originates from the transfer of energy from the exciteddipole to the metallic surface, through coupling with highin-plane wavevector values (cf figure 4). The energy istransferred to electron–hole pairs of the metal (excitons)which act as energy acceptors. This energy is ultimatelydissipated in the metal. This transfer is dipole–dipole innature and yields a d4 dependence for the transfer rate(when considering a plane surface). The relative part ofthis third channel can be evaluated by integration of thecurve over u > 1, while subtracting the SPP channel.

Figure 5 shows the fraction of power dissipated into eachchannel, i.e. radiation, SPP mode and lossy waves, as a functionof fluorophore/surface distances for dipoles oriented paralleland perpendicular to the surface and an isotropic orientation[14]. The dipoles that oscillate perpendicular to the surfacecouple very efficiently to the SPP modes. In that case, up to93% of the total dissipated power is given to the SPP [41].

All relaxation channels are competing with each other.Due to strong near-field coupling, the number of decayingprocesses increases as the fluorophore–surface distance d

decreases. This is observed through lifetime measurements.The fluorescence lifetime drops dramatically for small metal–fluorophore distances. The lossy wave coupling becomes theprevailing process for distances <20 nm even though SPPcoupling is getting stronger [23].

For fluorophores with Q < 1, the non-EM relaxationprocesses remain unaffected by the presence of the surface.

7


Since the number of EM relaxation processes increasesdramatically for small fluorophore-surface distances, thegeneralized quantum yield Q defined in equation (1) increasestowards unity (as the fluorescence lifetime τ tends towardszero). This is of practical importance since the photostabilityis related to the total time spent by the fluorophore in the excitedstate. Hence, for a given fluorophore, decreasing its lifetimeincreases the number of cycles before photodestruction. Thismeans collecting more emitted photons in a given experimentalconfiguration. Alteration in the photobleaching processesas been investigated both theoretically and experimentally[14, 42]. A good agreement between the classical EM modeland the experimental results is found for metal–fluorophoredistances larger than 10 nm. Essentially, for spacer layer below20 nm, because of the strong increase in the relaxation rate,fluorophores undergo more excitation-emission cycles beforephotobleaching. However, one must keep in mind that theseadditional processes coupling through near-field componentsare a priori detrimental to fluorescence detection since theyare related to non-radiative EM processes.

An important aspect of SEF is consequently to useproper surface structures to couple resonantly the SPP modesinto radiative emission and recover the energy. Two mainmomentum-matching techniques are used to couple the SPPmodes to photons. The first one consists of increasing themomentum of the photon by using high index materials andspecific optical geometry. The second technique breaks thetranslational invariance of smooth metallic surface throughcorrugated or roughened surfaces. These techniques aredeveloped in the following sections.

2.3. SPP excitation on flat metallic surfaces

On flat surfaces coupling between propagative light and SPPhelps both excitation of fluorophores and enhancement ofemission processes. The evanescent EM field associated withSPP has its maximum at the surface and is thus efficient forexcitation processes. Subsequently, recovering radiativelythe energy given to the SPP by near-field coupling with thefluorophore plays a crucial role in SEF.

Surface plasmons on the front side of a metallic film(i.e. the fluorophore side) can be excited by the rear sideusing the so-called Kretschmann configuration provided thatthe index of refraction of the material on the rear side is higherthan the one on the front side, as shown in figure 6(a) [40].When impinging on the metallic thin film with an angle θ0, thelight in-plane wavevector is k‖ = ngk0 sin θ0. The dispersionrelation for the SPPs propagating on the opposite side of themetallic thin film can thus be satisfied for a given incident angleθSP as shown in figure 6(b). The excitation of the SPP occursat the minimum of the reflected intensity. At the resonantangle for SPP excitation, the reflected intensity equal zero fora specific thickness dc of the metallic thin film [40]. At thesame time, the EM field reaches its maximum at the metallicsurface on the opposite side. The field can be significantlyenhanced compared with a similar configuration without themetal [43].

The enhancement G is defined as the ratio of thetransmitted intensity of the EM field Tmetal on the opposite side

Figure 6. (a) A schematic representation of the Kretschmannoptical configuration for the three-layer stackingglass/metal/dielectric. The index of refraction of the glass prism ng

is higher than the one of the dielectric nd. (b) Dispersion curves forthe Kretschmann configuration showing that the resonant couplingbetween light and SPP is possible when the light cone crosses theSPP dispersion curve. The in-plane momentum for a given incidentangle θ is increased by �k = ω(ng − nd)/c.

of the metallic thin film and the transmitted intensity withoutthe metal Tw/metal. It is obtained using Fresnel’s equations fora three-layer stacking glass/metal/dielectric [44]:

G = Tmetal

Td/metal= t

gmp tmd

p

1 + rgmp rmd

p exp(2ikzmd), (13)

where tijp and r

ijp are the Fresnel’s transmission and reflection

coefficients in p-polarization, respectively, for the boundarybetween medium i and j and light incident from i to j (g: glass;m: metal; d: dielectric). d is the thickness of the metal thinfilm and kzm the component of the wavevector in the metalnormal to the surface. The resonant excitation of the SPP onone side of the metal thin film induces a charge oscillationwhich results in an enhanced EM field at the metal–dielectricinterface. The EM field increases throughout the film thicknessto reach a maximum at the metal–dielectric interface [40]. Thecalculated enhancement values of the EM field at λ = 634 nmfor an optimized film thickness of about d = 50 nm are 14.3 forsilver and 5.9 for gold. Far from the saturation regime of thefluorophore, this results in large excitation rate enhancements,which is of particular interest in non-linear excitation processessuch as two-photon fluorescence [45].

8


For the Kretschmann excitation configuration, thefluorophore couples to the SPP resonantly through near-fieldcomponents with higher wavevectors. As mentioned before,it is necessary that the fluorophore is far enough from themetallic surface to avoid non-radiative relaxation processes (inorder to find the best compromise between high excitation andquenching). It is obtained with a transparent dielectric spacerlayer, typically a few 10s of nanometres thick (cf figure 5).

A simple enhancement technique for fluorescent signalsis to use simple interference effects induced by the mirrorcoating. The presence of standing waves can enhance theexcitation process by up to a factor of four (for a perfectmirror). The redirection of the emitted light on the mirrorallows a more efficient collection which depends on thelight detection geometry. The modification of the emissionpatterns due to planar boundaries was originally described byDrexhage [23] and more recently exploited for semiconductorcavity design [33]. As mentioned previously, dielectricmirrors can also be used since the SPP plays no role in thissimple enhancement process [35]. The overall enhancementis typically 10- to 15-fold compared with a standard glassslide. The modification of the fluorescence lifetime andintensity of a single molecule as a function of its distancefrom a smooth metallic mirror has been recently investigatedexperimentally [46]. The results agree with classical modelsshowing a clear correlation between intensity and lifetimemeasurements.

The optical configuration using SPP is often called SPfluorescence spectroscopy (SPFS). It has been recently usedin biological applications such as real-time monitoring ofhybridization processes, for trace amounts of PCR productsin biosensors with detection limit of 100 fmol or membraneprocesses [47–49]. Figure 7(a) shows a typical setupfor combined SPFS and surface plasmon resonance (SPR)measurements and figure 7(b) shows the reflectivity R and therelative intensity IS/I0, i.e. the optical intensity at the surface,IS, scaled to the incoming intensity, I0, as a function of theangle of incidence, θ , for SPP excitation at a silver–waterinterface [47].

Using SPP excitation on a flat metallic thin film trough SPresonance fluorescence microscopy, single-molecule real-timeimaging of single fluorophores attached to protein moleculeson metal surfaces in aqueous solution has been reported, seefigure 8 [50]. With a silver thin film on a Cyanine 5 dye marker,a 13-fold signal enhancement compared with a glass substratewas obtained. The enhancement is mainly due to the intensifiedevanescent field associated with the SPP and the redirection ofthe emission signal towards the detector.

Fluorophores can also be used as local probes of theevanescent field associated with the SPP. Surface plasmonfields can thus be imaged by detecting the fluorescence ofmolecular films close to the SPP carrying metal surface.Ditlbacher and co-workers in figure 9 show the field profileof SPP launched at lithographically designed nanoscopicdefects [51].

2.3.1. Surface plasmon cross emission (SPCE). The coreidea of surface plasmon cross emission (SPCE) is the recovery

Figure 7. (a) A schematic diagram of a typical surface plasmonresonance-surface plasmon fluorescence spectroscopy (SPR-SPFS)setup; (b) Reflectivity R and relative intensity IS/I0 as a function ofthe angle of incidence θ for SPP excitation at a silver/waterinterface. IS and I0 are the intensity at the metal/water interface andthe incoming intensity, respectively. Reprinted from [47], Copyright2000, with permission from Elsevier.

of energy transferred by the fluorophore near field to the SPPby coupling of the SPP modes to the radiating EM modes onthe backside of the film.

As we already mentioned, fluorophore in the vicinity of ametallic surface couples resonantly to the SPP through near-field components. The Kretschmann configuration allowsto excite SPP efficiently by the rear side of the metal thinfilm. By reciprocity SPP couples back to the radiating EMfield on the rear side. For a particular emission wavelength,the light emitted is highly directional taking the shape of acone with a half-angle θSPP(ωfluo), which satisfies the SPPmomentum conservation (cf figure 6). First observation of thiscoupling dates back to 1979 by Weber and collaborators [41]and has be more recently extensively studied by Lakowicz andcollaborators [18, 19].

The plasmon-coupled fluorescence emission exhibits acharacteristic angular distribution and polarization (radialpolarization) given by the emission spectrum of thefluorophore and the dispersion relation of the SPPs. Theexcitation of the fluorophores can be performed either bythe evanescent field associated to the SPP (excited in theKretschmann configuration) or by shedding light directly onthe fluorophore [19]. The latter geometry is less favorable since

9


Figure 8. Imaging of a single fluorophore bound to protein molecules on metal by SPP excitation. (a) Schematic drawing of the SPPresonance fluorescence microscopy. (b) Simultaneous fluorescence images of single fluorophores (Cy5) attached to myosin subfragments(lower panel) bound to a tetramethylrhodamine-labelled actin filament (upper panel) on an aluminium surface with a thickness of 30 nm.Reprinted with permission from [50]. Copyright 1998 by the American Physical Society.

Figure 9. Fluorescence images of the field intensity of SPP excitedvia (a) a silver particle (diameter 200 nm, height 60 nm) and (b) asilver wire (width 200 nm, height 60 nm, length 20 mm). Amolecular layer of Rhodamine 6G fluorophore is deposited on amultilayered stack of glass/Ag (70 nm)/SiO2 (10 nm). The dielectriclayer prevents fluorescence quenching. Reprinted with permissionfrom [51]. Copyright 2002, American Institute of Physics.

it does not take advantage of the huge SPP field enhancementand the near null EM field at the metallic surface.

Stefani et al [52] have recently observed the fluorescenceof single molecules in the vicinity of a thin gold film usingan epi-illumination scanning confocal microscope with bothexcitation and emission mediated by the SPP, see figure 10.They showed that the number of detectable photons fromfluophores perpendicular to the interface is enhanced by 140%by the presence of the metal while it is reduced to 26% forparallel molecules. Their experimental data are in a verygood agreement with the CPS model they used to evaluatethe relaxation channels (see figure 10(b)). The point spreadfunction of SPCE microscopy has been recently investigatedin detail by Tang et al [53].

Figure 10. (a) Single-molecule fluorescence image obtained withfull-beam illumination. The white arrow indicates the polarizationdirection of the excitation beam. (b) De-excitation rates of a parallel(left) and a perpendicular (right) dipole as a function of the spacerthickness. The vertical bands show the spacer thicknesses used inthe experiments. Reprinted with permission from [52]. Copyright2005 by the American Physical Society.

It was anticipated that SPCE could increase thefluorescence signals by a factor of up to 1000 [17, 19], whichwould be a revolutionary improvement for analytical assays.Recent experimental [52,54,55] and theoretical results [43,56]contradict this claim. Enderlein and Ruckstuhl in a recenttheoretical paper [43] even claim that significantly less energyis coupled through a silver film into the glass than in the idealcase where no silver film is present. But their theory does nottake into account surface roughness which may influence thefluorescence enhancement.

10


Figure 11. A single skeletal muscle fibre (labelled with 5’IATR)and observed with fluorescence under SPE-TIRFM (right) andstandard TIRFM on a bare glass interface (left). Metal filmthickness is 30 nm < d < 40 nm. Reprinted from [54], withpermission from Biophysical Journal.

In fact, SPCE can be advantageous with respect to moreconventional detection schemes because of the followingaspects:

• high directionality of fluorescence emission in SPCEwhich helps to better discriminate fluorescence,

• strong wavelength dependent angular position, making itpossible to use SPCE as a spectrally resolving technique,

• radial polarization of the emission, in contrast to emissiongenerated at a glass/water interface.

SPCE is of particular interest as an alternative microscopyto total internal reflection fluorescence microscopy (TIRFM).TIRFM is a widely used imaging technique, especially forprobing the structure and dynamics of basal surfaces of cells,due to its high surface selectivity and an ability to suppressbackground fluorescence [57, 58]. Only fluorophores locatedwithin about 100 nm of the glass/sample interface are excitedby the evanescent field.

An imaging technique that combines SPCE within TIRFMconfiguration, adding a thin metallic film onto the glasssubstrate, is called surface plasmon enhanced-TIRFM (SPE-TIRFM) or SPCE microscopy (SPCEM). It has been usedrecently in the imaging of muscle fibrils [55] and of afluorescent protein in situ [54]. Figure 11 compares a standardTIRFM image (left) and a SPE-TIRFM image (right) of a singleskeletal muscle fibre.

SPE-TIRFM offers no significant improvements overTIRF as far as fluorescence collection efficiency and brightnessare concerned [43]. However it has some significantadvantages:

• The detection volume is axially thinner than in standardTIRF [59]. Changes of fluorescence upon axialdisplacement of a molecule are caused not only by thegradient of the evanescent wave intensity as in TIRF, butare also enhanced by quenching of the fluorescence nearthe metal surface.

• The background noise is greatly reduced. Only SPCE istransmitted trough the metal film. Since the fluorophoresare coupled to the SPP by near-field coupling, all otheremissions, in particular from fluorophores further awayfrom the surface, are reflected by the metal.

• Photobleaching is reduced due to the decrease of thefluorescence lifetime.

• Fluorescence is quenched for flurophore–surface separa-tion below 20 nm. This is of interest for biological sys-tems [59].

2.4. Surface plasmon field-enhanced fluorescence oncorrugated surfaces

SPPs are usually non-radiative because their wavevector(momentum) is greater than that of free photons of thesame frequency (see figure 6). This is a direct consequenceof the real part of the SPP effective refractive indexnSPP = √

εdεm/(εd + εm) being slightly larger than therefractive index of the dielectric medium. In the typicalcase of an interface between vacuum (εd = 1) and ametal with low losses (ε

′′m |ε ′

m|) such that |ε ′m| > 1,

typically �e[nSPP] ≈ 1 + 1/(2|ε ′m|) > 1. Instead of a specific

illumination condition, this momentum excess can beovercome using a textured metallic surface.

Patterning the metallic surface with a periodic wavelengthscale corrugation provides an elegant solution. Breaking thetranslational invariance of the thin film allows one to relax themomentum conservation restriction. For a periodic ondulationof period a, the momentum conservation is now given byk′

SP = kSP ±n gB where gB is the grating or Bragg wavevectorgB = 2π/a and n is an integer.

This idea was first demonstrated experimentally for theSPP mode by Knoll et al using dye monolayers deposited in athin film on a silver grating [60], shortly followed by Adamset al [61] confirming the predictions of Aravind et al [62].Adams et al [63] and Sullivan et al [64] have later extended thestudy to thicker films to understand the scattering of waveguidemodes.

Figure 12 shows a typical example of the effectiveenhancement factor plotted as a function of the emission angleθ for a corrugated metal-clad waveguide sample versus anon-corrugated one [64]. With this sample, the directionallyscattered bound-mode features are clearly visible. Theemission features can be attributed to the out-coupling ofthe three waveguide modes (two TM-polarized modes andone TE-polarized mode). The prominent sharp peak at θ ≈15◦ is associated with the SPP mode (i.e. the lowest TM-polarized mode).

The direction of the light beam is highly directionalfollowing the moment conservation relation which can bequite interesting to increase signal to noise ratio. Specificoptical geometries have been proposed for excitation anddetection [64].

The efficiency with which bound light can be recoveredby scattering from a corrugated surface has been shown tobe as high as 80% for the SPP propagating normal to thegrooves [65]. This efficiency reaches an average value of60% independent of the SPP direction using a bi-grating [14].Recovering the energy transferred to SPP modes is essentialin particular to increase the yield of light emitting diode(LED) devices, which necessarily incorporate metal contactsfor efficient electron injection. The advantages of gratingcouplers have being studied in such devices [66, 67].

11


Figure 12. Measured effective enhancement factor versus emission angle θ for a corrugated metal-clad sample. The excitation angle isθexc = 45◦. The metal-clad waveguide is composed of glass/Ag/LiF/dye/air. The LiF thickness is d = 300 nm. (a) On a uncorrugatedsample. (b) On a corrugated sample, the grating period is 640 nm and the grating depth is about 100 nm. Reprinted from [64], withpermission from the Optical Society of America.

The presence of the grating directly affects thespontaneous emission of the emitters themselves, since thephotonic mode density they experience will be necessarilydifferent from the planar case due to the different boundaryconditions. Both the distance dependence of the lifetimeand the spatial distribution of the emitted light aresignificantly changed upon the introduction of a corrugation,quite apart from the Bragg-scattered bound-mode features.One hypothesis is that these perturbations arise from theinterference of the grating scattered dipole fields with the usualupward propagating and reflected fields [68].

The efficiency of the light cross-coupling can also beenhanced by using corrugated thin films. Wedge and Barnesshowed that the fluorescence emission from a structure witha corrugated thin metal film is over 50 times greater thanthat from a similar planar structure [69]. Additionally, theyshowed that samples containing a planar metal film with acorrugated dielectric overlayer have similar properties. Thestrong fluorescence emission is mediated via coupled SPPs.The SPP cross coupling is promising for devices such assurface-emitting LEDs.

2.5. Surface plasmon field-enhanced fluorescence onroughened surfaces

The coupling between a corrugated surface and a fluorophoreis also a simplified approach for the study of the coupling witha roughened surface. To a first approximation, roughenedsurfaces can be considered as multi-corrugated surfacesprovided the roughness is sufficiently small [40].

Okamoto et al showed that the emission of light emittersbased on InGaN quantum wells could be enhanced with useof rough metallic layers [70]. Roughness and imperfectionsin evaporated metal coatings can efficiently scatter SPPs intoradiative modes. The authors measured the topography of anuncoated GaN surface and of a 50 nm thick Ag film evaporatedonto GaN using scanning electron microscopy (SEM), seefigure 13. They measured a modulation depth of the Agsurface of approximately 30–40 nm with a length scale of afew hundred nanometres. They found an enhancement of lightemission using a corrugated grating with similar parameterswhile no enhancement was found for other grating parameters(cf figure 13(d)). This measurement suggests that the size of

(a)

(d)

(b) (c)

Figure 13. (a) SEM image of an uncoated GaN surface, (b) SEMimage of a 50 nm thick Ag film evaporated onto GaN and (c) SEMimage of the grating structure with 33% duty cycle fabricated withina 50 nm thick Ag layer on GaN. (d) Microluminescence imageincluding areas of different nature. All samples used were coatedonto InGaN/GaN QWs with 10 nm GaN spacers. Reprintedfrom [70] by permission from Macmillan Publishers Ltd: NatureMaterials, Copyright 2004.

the metal structures determines the SP-photon coupling andlight extraction.

The presence of metal nanostructures has a dramaticeffect on the redirection of the angular emission of anindividual molecule as well as on the enhancement of theexcitation rate of the fluorophore. The influence of nearbynanosized metal objects on the angular photon emissionhas been observed experimentally on a single molecule forvarious dipole orientations [71]. This property offers thepossibility of enhancing the light collection by tailoring thelocal environment of a fluorophore.

For enhanced roughness, multiple scattering of lightoccurs in nanostructured random media [72]. Light interferesand induces under certain conditions strong localization oflight [73] analogous to Anderson localization of electron wavefunction. This localization is expected in a limited wavelength

12


(a) (c)

(b)

Figure 14. (a) Schematic cross section of the sample. Fluorescence excitation and measurements are made from the same side. Excitationwavelength is 488 nm. (b) SEM image of the nanoporous gold surface showing features. (c) Fluorescence enhancement, measured at780 nm as a function of the etch depth, D (triangles). The solid line is a fit to the data using a model that accounts for the spatial distributionof Si nanocrystals and the enhanced local field. Reprinted with permission from [75]. Copyright 2005, American Chemical Society.

interval. Furthermore, variations in the dielectric constantof the medium should be large enough to realize sufficientlystrong multiple scattering of light. Both phenomena exhibita threshold character with respect to the dielectric contrast.Indeed, SPP are scattered by surface roughness if the scatteringis sufficiently strong [74].

Biteen et al [75] studied the coupling between Sinanocrystals (nc-Si) and a highly roughened surface made ofnanoporous gold thin film. They showed that this couplingcan yield a 4-fold enhancement in the fluorescence intensityat λ = 780 nm, see figure 14, which is related to the4-fold enhancement in radiative decay rate as a result oflocal-field effects. The effective excitation cross section andquantum efficiency are enhanced by a factor of 2. Theauthors successfully modelled the complex geometry of thenanoporous gold film by gold nanoparticles, confirming therole of strong localization of SPP on rough metallic films.Such an increase in the emission rate is promising for the useof nc-Si in LED as opposed to the use of direct band gap lightemitters and should lead to high optical gains in all silicondevice fabrication processes.

This section showed how fluorescence enhancementon flat surface is improved by surface shaping, additionalcorrugation and roughness. Fluorescence is amplified bypropagating surface plasmon modes. In the case of highlyroughened surfaces, multiple electron scattering induceslocalization of surface plasmons. The next section deals withSEF using localized and non-propagating surface plasmons.We will introduce successive enhancement processes startingfrom the simplest nanometric objects (single particle) to morecomplex surfaces.

3. Localized surface enhancements near metalnanoparticles

The localization of light plays a key role in SEF. Earlier studieshave shown how the EM field can be significantly enhancednear metal objects and surfaces when the size of the underlying

object is much smaller than the illumination and/or detectedwavelength. When metal objects are much smaller than thewavelength, the electric and magnetic fields can be localized,giving rise to inhomogeneities in the EM field distribution. Infact, any discontinuity between a metal and a dielectric givesrise to inhomogeneities in the EM field distribution [76] thatcan be exploited for local EM enhancement.

These inhomogeneities can be used to specifically excitea given fluorophore. They are one of the main sources ofenhancements. When coupled to SPP, these inhomogeneitiesare sources of peaks of even higher enhancement. Anylocalization of EM intensity near metal nanostructures thatcouples to propagating modes can increase the fluorescenceby several orders of magnitude. These high EM intensities areapplied to enhance various non-linear effects such as secondharmonic generation [77–80] and Raman scattering [11, 81].

The problem of the spatial localization of light has beenstudied in various metal systems. Each system, from a singlemetal nanoparticle to metal aggregates and inhomogeneousmetal films, has specific enhancement properties linked closelyto its geometry. Single metal ellipsoids for instance aresources of an exceptionally large enhancement. The gapbetween two-metal nanoparticles can be tuned to increase thefluorescence of a given fluorophore by more than 2 ordersof magnitude. However these enhancements are wavelengthspecific. Inhomogeneous metal films, given their randomnature, are expected to enhance the fluorescence of variousfluorophores simultaneously, allowing multiple fluorescenceimaging. At the same time random films are not necessarilyas efficient for a given fluorophore.

The remarkable properties of metal systems arise due tothe strong coupling between the fluorophore and the surfaceplasmon resonance of metals [82]. In section 2.2 we describedhow the distance between the fluorophore and the surfacesdramatically changes the fluorescence rate. In the case ofparticle enhancement, the size of the nanoparticle also playsa role. Most of all, due to the geometry of the problem,characteristic sizes and distances are one order of magnitude

13


smaller than in the case of flat surfaces. Depending on thefluorophore emission frequency ω, on the distance d betweenthe fluorophore and the metal [26] and on the size of the metalnanostructure, different interactions exist with dramaticallydifferent effects.

• For the low frequency regime, ω � ωSP where ωSP isthe surface plasmon frequency and when distance d is ofthe order of 1 nm, the dominating phenomenon is Landaudamping [83], an efficient damping of the fluorescence.

• For ω ∼ ωSP, coherent excitation of the surface plasmontakes place when d � 1 nm and when the metal hasnanoscale structures. The metal may act as a nanoantennaenhancing the strength of the dipole oscillator and thuswe expect enhanced fluorescence. It is worth noticingthat these huge enhancements can be applied to a varietyof optical effects where the signal is usually very low,for instance in surface enhanced raman scattering (SERS)[11, 84], strongly enhanced non-linear effects [85] suchas photoluminescence [86–90] and coherent control [91],as well as surface plasmon amplification by stimulatedemission of radiation (SPASER), predicted by Bergmanand Stockman [92].

• For ω ∼ ωSP , d � 1 nm for a metal with macroscopicextension but no nanoscale structures, besides the surfaceplasmon resonance, the propagating modes of SPPs areexcited. In this case, the metal system plays the role ofan optical antenna that propagates the excitation of thefluorophore to the far field [82, 93, 94]. This leads to anincrease in the fluorescence efficiency and intensity andto a shortening of the fluorescence lifetime.

The present part of our review deals with the thelocalization of light around metal nanostructures whenilluminated by an EM wave and how these localizationsenhance the fluorescence. In the following, we will showhow single particles, particle aggregates and random metalcomposites interact strongly with light. The differencebetween single objects, aggregates and non-continuous filmsand their relation to fluorescence enhancement will bediscussed.

3.1. Near-field images and microscopy

Optical properties of matter are investigated by means offar field (classical) and near-field optical microscopy. Inthe past 20 years, near-field microscopies have proven to bepowerful tools to investigate physical, chemical and biologicalproperties at the nanometre scale. The investigation of the largeenhancement of the electric field exploited in fluorescenceenhancement has been carried out mainly by near-field opticalmicroscopy, also called NSOM. Among all the branches ofnear-field microscopy which exist today, NSOM has beenwidely used in this respect because of its unique capabilityto probe the electromagnetic field that lies in the proximity ofnanometric particles and surfaces.

To properly understand the physical meaning of NSOMmeasurements, we will shortly describe the mechanismresponsible for the high resolution of the NSOM technique.The most basic picture of an optical microscope for the

investigation of the optical properties of a nanoobject is thethought experiment of a nanooptical detector (or nanolightsource) located in the proximity of the nanoobject [95].The resolution of this microscope is then given by thesize of the nanodetector/nanosource itself. To avoid thecomplicated design and fabrication of such a very small opticaldetector/source, two techniques are generally used. Theyare complementary in terms of collection and illuminationefficiency [96, 97].

The first technique, which is also the most commonone, uses a nanohole at the end of an optical waveguide.The nanohole scatters the near-field which is mostly non-propagative and evanescent by nature into propagatingwaves [98]. The waveguide allows transmission of thelight either from the source to the nanohole or fromthe nanohole to the detector. First fluorescence imagesrevealed the ability of near-field optical microscopy to achievehigh resolution [99–101]. Xie and Dunn used near-fieldfluorescence microscopy to reveal the spectral dynamics ofsingle molecules and their quenching by metal coating of thewaveguide [102]. With the work of Novotny, van Hulst andtheir collaborators [71], it was possible to record the effect ofa metal object on the dipolar emission of fluorescence.

The second technique uses a nanoobject in the proximityof the surface. The coupling between the surface and thenanoobject acts as a nanoantenna, which scatters the EMfield located around the surface. This scattered light can beeasily detected by a macroscopic detector. The resolution ofthis technique is then given by the size of the nanoantennawhich can in principle descend down to the size of a singleatom. This theoretical limit of the resolution has not beenreached so far experimentally. However, using a metal ora silicon tip as the nanoobject and conventional microscopeobjective for light collection at the detector, resolution higherthan 20 nm in the visible domain has been demonstrated.This presents an improvement of a factor of more than 10with respect to conventional optical microscopy [103–107].The unbeaten resolution of the antenna-NSOM technique hasbeen used for fluorescence imaging [108, 109] even if theinfluence of the quenching of the fluorescence by the metaltip is still under consideration [110–112]. We shall see in thefollowing sections that even for a metal sphere, the problem ofenhancement is not always clearly understood. Fluorescenceparticles that probe the intensity distribution [113–115] or thetemperature [116] in a near-field optical geometry have alsobeen reported upon recently.

For the moment, relatively few attempts have been done toapply fluorescence near-field optics in biological studies. Thistechnique is invasive due to the fact that the near-field probeneeds to enter a cell in order to image its inside. However someattempts to image membranes [106] and follow the movementof biomolecules [117] have been realized. They show thepower of NSOM to reveal and understand the complexity ofnanoscale biological systems.

3.2. Isolated nanoparticles

Already in the 19th century [118], metal nanoparticles wereknown to have specific optical properties. In his famous 1908

14


(a) (b) (c)

Figure 15. Cartesian coordinates (a) Ex , (b) Ey and (c) Ez of the electric field in a plane located at � 20 nm above the plane where the goldparticle is situated. The incident polarization is along the x-axis, which is horizontal in this figure (ϕ = 0◦).

paper [119], Mie derived the solution for the scattering of lightby a spherical particle. When the particles are far apart anddilute, the formalism given by Mie gives the exact expressionof the EM field around them. When the particles are muchsmaller than the wavelength of the incoming light, approximateexpressions may be derived which allow some physical insightinto the behaviour of light around very small particles. Themain advantages of these approximations lie in the path theyopen to understand more complicated physical problems whereno exact theory exists [120, 121].

3.2.1. Scattering of light by small spherical particles. Fora perfect sphere illuminated by an arbitrarily polarized light,the expression of the electric field can be derived from Miecalculation. This expression can be simplified if the particleis small compared with the incoming wavelength. The keyparameters for Mie calculations are the scattering coefficientsan and bn of the vector spherical harmonics which satisfythe vector wave equation. They can be computed eitherfrom the approximate expressions given in [121] in the caseof particles that are small compared with the wavelength ordirectly computed using the Matlab functions developed byMatzler [122]. For sufficiently small sphere, only a1 is notablydifferent from zero [121]. As a consequence only the firstdipole term subsists in the expression of the electric field.When ρ 1, ρ = k · r , k being the wavevector and r thedistance from the centre of the particle, the scattered field Es

is then purely imaginary:

Es(ρ, θ, ϕ) = i10−4

ρ3

3

2E0[−2 cos(ϕ) sin(θ)er

× cos(ϕ) cos(θ)eθ − sin(ϕ) · eϕ]. (14)

Using equation (14) and a Cartesian coordinate system (x, y, z)

where (x, y) is the plane where the particule is situated and z

is the direction of propagation of the incident light with anincident polarization along the x-axis, we have calculated thespherical components of the scattered field by a single metalparticle in a plane located at an altitude z = 15 nm abovebut close to the plane where the particle is situated, z = 0being the centre of the particle. For the computation, weused gold for the metal, a particle deposited on a substratewith a radius of rp = 6 nm and an illumination wavelength ofλ = 780 nm [123].

In figure 15 the three electric field components are showntogether, depicting the behaviour of the electric field around

the particle. We note that the amplitude Ex is mainly negative,aligned with the direction of the polarization and has two localminima located on each side of the particle. The amplitudeEy presents a quadrangular repartition with two positiveand two negative lobes at 45◦ to the polarization direction.Finally the amplitude Ez is composed of one positive lobeand one negative lobe of equal magnitude. The line joiningthe extrema of the lobes is perpendicular to the direction ofpolarization Ox.

It is clear that in the zone very close to the particlethe scattered field is purely imaginary and differs from thetransverse far field where Ez vanishes. The electric fieldradiated by a dipole source would be aligned along thepolarization direction whatever the distance from the dipole.This is not the case for the three Cartesian components of Ein the near zone of a particle. We note that in the far field thedescription of the particle as an ideal dipole remains valid.

Furthermore, it is clear from the images of the threecomponents of E in figure 15 that the electric field is stronglyinhomogeneous around a metal particle. The relative positionsof the high amplitude peaks (dark and bright spots) arelocalized with respect to the polarization of the incident light.The intensity of fluorescence of a dipole in close proximityto the metal particle depends only on the intensity of theelectric field E2. However, the relative position of the highfluorescence peaks is nevertheless strongly dependent on thelocal distribution of Ei(i = x, y, z) (amplitude of the electricfield components).

Comparing figures 15 and 16, the latter showing NSOMimages of the electric field around a gold sphere, we can seethat the measured NSOM signal does not represent the totalelectric field around the metal particle. Depending on thegeometric characteristics of the probe used to measure thelocal electric field, some components are invisible. Herethe probe reveals mainly the z component of the electricfield. A more detailed analysis of the observed electricfield around metal particles is given by Wannemacher andco-workers [124]. It is clear, however, from the images shownthat high amplitudes (bright and dark spots) of the electricfield around a sphere occur very close to the particle (less than20 nm). As mentioned above, the polarization of the incidentlight is the most important parameter when it comes to thelocalization of the high intensity peaks where enhancementoccurs.

15


Figure 16. Experimental NSOM images of gold spheres with a diameter of 12 nm for 2 orthogonal polarizations. The arrows indicate thedirection of polarization of the incident field.

3.2.2. Absorption and polarizability of a small sphere.

In the quasi-static approximation. The electrostatic approx-imation, also called quasi-static approximation, can be usedwhen the applied field inside the sphere can be consideredconstant. This is the case when rp < λ/10. For a suffi-ciently small sphere (krp 1) of radius rp, in the electrostaticapproximation, the absorption cross section is [120]

Cabs = 4πkr3p Im

(ε − εm

ε + 2εm

), (15)

where ε and εm are the dielectric constants of the particle andof the surrounding medium respectively. When the particle isin air (vaccum), εm = ε0. In a uniform static electric field E0,the potential of an ideal dipole is = p · r/4πεmr3 with adipole moment:

p = 4πεmr3p

ε − εm

ε + 2εmE0.

Thus one introduces the polarizability α to describe how thesphere is polarized by the applied field p = εmαE0. Under theelectrostatic approximation, the quasi-static polarizability α0

of a polarizable sphere is defined by [121]

α0 = 4πr3p

ε − εm

ε + 2εm. (16)

The absorption cross section from equation (15) can then bewritten as follows:

Cabs = kIm(α0). (17)

General expression of the polarizability of a nanoparticle.To be consistent with the law of energy conservation7, theexpression of the polarizability is as follows:

α = α0

1 − i(k2/6π)α0. (18)

It should be pointed out that the polarizability is frequencydependent α = α(ω) through α0 = α0(ω) = 4πr3

p ((ε(ω) −εm)/(ε(ω) + 2εm)). The radiative and non-radiative transfer

7 The optical theorem [44], which accounts for the fact even for a non-absorbing particle, there is an imaginary part of the polarizability for energyloss by scattering.

rates are introduced using the expression of polarizability (18),which is consistent with the law of energy conservation.

Green function G, which is the solution of a givenproblem for a unique point source, satisfies the Helmholtzequation [125, 126]. G connects an electric dipole source p ata position r0 [127] through the relation E(r, ω) = G(r, r0, ω) ·p(ω), where r, r0 and ω are, respectively, the position ofobservation, the position of the dipole and the frequency ofthe incoming radiation (see the introduction). In the presenceof the nanoparticle, Green’s function can be written in the formof a product of the effective polarizability αeff(ω) = α(ω)ε0

and the free space Green’s function [26, 128] G0:

G(r, r0, ω) = G0(r, r0, ω)

+ G0(r, r′, ω) · αeff(ω)G0(r′, r0, ω). (19)

The dyadic S from equations (6), which describes theelectromagnetic response of the environment [24–26], isdefined by G(r, r0, ω) = G0(r, r0, ω)+S(r, r0, ω). S describesthe modification of the free space Green’s function G0 due tothe presence of the nanoparticle and can be identified fromequation (19):

S(r, r0, ω) = G0(r, r′, ω) · αeff(ω)G0(r′, r0, ω). (20)

The fluorescence rates. From G(r, r0, ω) and equations (6)and (5), one can calculate the different fluorescent rates(radiative �r and non-radiative �nr) and the spontaneous decayrate � (or total decay rate) and the free space fluorescence rate�0. The sum of the radiative decay rate �r and the non-radiativedecay rate �nr gives the total decay rate � = �r + �nr. Thefluorescence rate can be written as

�fluo = �exc�r

�= �exc

(1 − �nr/�0

�/�0

), (21)

where �exc ∝ |p · E|2 is the excitation rate dependingon the local excitation field E(r0, ω). Assuming that theenvironment does not affect the polarizability of the molecule,the normalized excitation rate can be expressed as

�exc

�0exc

=∣∣∣∣ u · E(r0)

u · E0(r0)

∣∣∣∣2

, (22)

where �0exc is the excitation rate in free space, u is the unit

vector pointing in the direction of p and E is the total local field

16


(incident electric field plus scattered electric field). Assumingthat the molecule has a high intrinsic quantum yield and that thenon-radiative rate is given by Ohmic losses in the environmentaccording to [24], we have all the ingredients to calculate thedecay rates �r, �nr and �fluo for a fluorescent molecule in thepresence of a nanoparticle. The results depend on the incidentelectric field E0, the unit vector u and the properties of theenvironment encoded in G and αeff .

Fluorescence rate in the presence of a nanoparticle. If wewrite R = |r′ − r0| the distance between the fluorophore andthe particle, it is possible to show how the fluorescence lifetimestrongly decreases when R tends to 0. The variation of thefluorescence lifetime τfluo = 1/�fluo is due to the variationof the radiative �r and non-radiative �nr decay rates of thefluorescence (equation (21)). The decay rates at short distance(kR 1) are as follows.

• The non-radiative rate �nr, proportional to R6 due todipole–dipole interaction. This is the usual dependenceof the non-radiative transfer in the dipolar approximation.The validity of the dipolar model for single-moleculefluorescence has been recently discussed because it doesnot predict the correct fluorescence quenching [129], seebelow. But it is still a good approximation at long distance(R � rp/10).

• The radiative decay rate �r which contains threecontributions, the power directly radiated by thefluorophore, the power radiated by the induced dipole inthe particle proportional to R6 and an interference termproportional to R3. The existence of the R3 term is adirect consequence of the coherence between the radiatedfluorescence emission and emission of the induced dipolein the particle8.

The ratio of the two terms depends on the size of theparticle and on the resonance condition. For a very smallparticle, the R3 term should dominate. But at the plasmonresonance for instance, the R6 term can be much larger [130].This can substantially change the balance between the radiative�r and the non-radiative �nr fluorescence rates.

In the following, we will have a closer look at theresonance condition of a metal nanoparticle and see howthe coupling with a fluorophore dramatically changes theresonance and emission properties of the overall system.

3.2.3. The plasmon resonance for small particles.

The resonance conditions. For a sphere with a diameter muchsmaller than the wavelength, the electric field in equation (14)is mainly localized near its surface, except for the lowest ordermode (n = 1) which is uniform throughout the sphere. Itfollows from the expression of an and bn that a1 is the dominantcoefficient if the complex dielectric function of the particleε = ε′ + iε′′ is

ε = −2εm, (23)

8 This situation is very different from the one of Forster resonant energytransition (FRET) where no coherence exists between the donor and theacceptor and only the R6 dependence exists [26].

where εm is the dielectric function of the medium surroundingthe particle. The frequency ωSP, sometimes called the Frohlichfrequency [131], at which

ε′ = −2εm, ε′′ = 0, (24)

corresponds to a uniform polarization of the sphere (alsocalled dipolar mode of the particle). For a real particle, theabsorptive part of the complex dielectric function is never 0and condition (24) becomes

ε′ = −2εm, ε′′(ωSP) 1. (25)

Unless specified and for simplicity we will always refer to theseconditions as those corresponding to ε = −2εm.

For slightly bigger particles, the condition on thecoefficient a1 gives

ε = −(2 + 125 (krp)

2)εm, (26)

where rp is the particle radius and k is the wavevectork = 2πN/λ, N is the refractive index of the medium andλ is the illumination wavelength. For small particle sizes(rp λ) equation (26) is not different from (23). However,equation (26) gives us an indication of how the frequency ωSP

is shifted towards lower values as the particles size increases(the real part of the dielectric function is almost always anincreasing function of frequency around the frequency forwhich ε = −2εm).

Although any small spherical particles show surfacemodes when their size is much smaller than the wavelength(only the lowest order mode is uniform throughout a sphere),small metal particles have specific properties due to the factthat the negative region of ε′ is relatively broad in a metal.

The Drude model. In the Drude model, which accuratelydescribes the optical properties of simple metals by taking intoaccount the optical response of a collection of free electrons,the dielectric function is

ε = 1 − ω2p

ω2 + iγω, (27)

where γ is the damping constant and ωp, the plasma frequencyof the free electrons, is given by ω2

p = N e2/mε0. N is thedensity of free electrons, m is the effective mass of an electronand ε0 is the dielectric function (or permittivity) of the freespace. Real and imaginary parts of the dielectric functionε = ε′ + iε′′ are

ε′ = 1 − ω2p

ω2 + γ 2,

ε′′ = 1 − γ ω2p

ω(ω2 + γ 2). (28)

The plasmon resonance frequency for a metal sphere. Inthe Drude model described above, there is no maximum inthe imaginary part of the dielectric function. ε′′ decreasesmonotonically with increasing frequency. But for small

17


0.01.5 2.0 2.5 3.53.0

0.5

1.0

1.5

2.0A

bsor

ptio

n

Energy [eV]

Sphere

SALA

Figure 17. Absorption of a silver sphere and an ellipse in a glassmatrix. SA: when the polarization of the incident light is parallel tothe short axis of the ellipse. LA: when the polarization of light isparallel to the long axis.

spherical metal particles9, near the frequency ωSP where ε =−2εm, there is a peak in the absorption cross section 15. Ifω2

p � γ 2 then from the condition (24) and equation (28)

ωSP = ωp√1 + 2εm

. (29)

When the surrounding medium is air (this is still a goodapproximation for particles supported on a flat surface), εm = 1and equation (29) becomes

ωSP = ωp√3. (30)

Equation (30) gives the general expression for the surfaceplasmon10 frequency of a small spherical metal particle in air.The surface plasmon is usually described as the oscillation ofthe free electron in metal. It is sometimes called localizedsurface plasmon (LSP) or surface plasmon resonance (SPR).The LSP should not be confused with the surface plasmonpolariton (SPP, see section 2.1) [9] which is dispersive (ωSPP =ω(kSPP)) whereas LSP are for bound geometries [132].

Higher order modes can be excited for bigger particles.The environment around the particle also changes theabsorption peak [133, 134]. For isolated spherical goldparticles, the absorption peak of the surface plasmon lies inthe range 510 nm < λSP < 570 nm when 5 nm < rp < 50 nm[121,133,135]. For silver, however, the absorption peaks spanalmost the entire visible spectrum depending on the size andthe surrounding medium of the particle [121, 133], see forinstance in figure 17 the absorption spectra for a silver sphere,rp = 25 nm, embedded in a glass matrix. Several studies inthe past decade concentrated on the change of the plasmonresonance shape, linewidth or position due to the surroundingmedium [136–138]. In a recent experiment, Kalkbrenner,Sandoghdar and collaborators show how the sensitivity of theplasmon resonance can be used to probe the electromagneticfield distribution [139].

9 For bigger particles, to a good approximation, the equations given aboveare still valid as long as the size of the particles is smaller than the wavelength.10 A plasmon is a quantized plasma oscillation.

The strong absorption of metal nanoparticles. The absorp-tion of small spheres is their most striking feature. Very smallparticles can absorb more than the light incident on them [121].To understand this statement, a closer look at the absorptioncross section of equation (15) is necessary. If in equation (15)we replace ε by ε′ + iε′′ then

Cabs = 12πkr3p

ε′′εm

(ε′ + 2εm)2 + ε′′2

and for a real metal near the plasmon resonance frequencyat which condition (25) applies, the absorption cross sectionbecomes

Cabs = 12πkr3p

(εm

ε′′(ωSP)

). (31)

The maximum absorption is inversely proportional to theabsorptive part of the complex dielectric function. In fact, theabsorption cross section shown in equation (31) can be morethan 10 times greater than the geometrical cross section forsmall metallic particles. This explains why metal particlesnear the plasmon resonance frequency are potentially veryinteresting when it comes to enhancement properties. It shouldbe pointed out that when plasmon resonance occurs, even ifthe enhancement effects are much stronger, the localization oflight is very similar to the out of resonance situation, at leastfor isolated particles. Careful analysis of the electromagneticfield distribution can be found in the papers by Schatz andcollaborators [134, 140] and Link and El-Sayed [141]. Krennand Aussenegg, in their search for plasmonic devices, were thefirst to measure the fluorescence intensity distribution arounda metal nanoparticle when launching a SPP [142].

The polarizability of a metal spherical particle neara fluorophore. Inserting equations (27) and (30) intoequation (18) yields the following expression for thegeneralized polarizability [26] of a small sphere:

α(ω) = 4πr3p ω2

SP

ω2SP − ω2 − iωγ − i(2/3)k3r3

p ω2SP

, (32)

where the term −iωγ accounts for damping by absorptionand the term −i(2/3)k3r3

p ω2SP accounts for radiative damping.

Equation (32) is consistent with the law of energy conservation.It can be shown from the energy flow distribution of theentire system that one can separate the radiative and the non-radiative contributions [143] in the fluorescence decay rate (seeequation (20)). The variation of the different contributionsto the decay rate with respect to the particle–fluorophoredistance strongly depends on the dipole orientation of thefluorophore [137].

3.3. Fluorescence enhancement by metal particles andaggregates

Recently, various groups have shown how silver particlescan enhance the emission properties of rare-earth fluoro-phores [144–147]. Even if Itoh and collaborators show thatcoupling between the particles is also of importance, allstudies indicate that metal particles by themselves increase thefluorescence and luminescence of fluorophores.

18


3.3.1. Single metal particles.

General description of the resonance effects in metal particles.Most of the times metal particles are not isolated and not purelyspherical. The fact that metal particles aggregate and form non-spherical shapes does not change their huge absorption andscattering properties that we described above. With respect totheir coupling with light, the main difference lies in the numberand position of the resonance absorption peaks in the frequencyrange.

The most general description of a smooth particle witha regular shape11 is an ellipsoid with 3 axes a, b and c

(a � b � c). If b = c the spheroid is prolate (cigar shape) andif b = a the spheroid is oblate (pancake shape). By analogyto a dipole, it is possible to define within the electrostaticapproximation a polarizability αi when the applied electricfield is parallel to the axis i of the ellipsoid. The dipole momentis �p = εmαi

�E0 and the polarizability of the ellipsoid is

αi = 4πabcε − εm

3εm + 3Li(ε − εm),

where Li is a form factor that obeys the following rules0 < Li < 1 and L1 + L2 + L3 = 1 (for a sphere, L1 = L2 =L3 = 1/3). The absorption cross section from equation (17)can then be written as follows in the case of the axis i of theellipsoid:

Cabs = kIm(αi). (33)

There are peaks in the absorption when ε = εm(1 − 1/Li).For an ellipsoid, this means that there are two or even threeresonances. As Li can have any value between 0 and 1, 1/Li

can span a very wide frequency range. Figure 17 shows thetwo resonance peaks in the absorption of an ellipse with along axis c = 100 nm and the short axes a = b = 40 nm.There are two different resonances but only one peak can beseen when the polarization of the incident light is parallel toone of the axes of the ellipse [148–150]. Similar behaviouris seen for nanodiscs or nanorods [141], as these are thelimits of a prolate and an oblate ellipsoid, respectively. It isthen natural to describe nanostructures in terms of ellipsoidsof various shapes. The corners of a triangle for instance,studied intensively at present [134, 151], are never perfectlydefined [152,153]. It is then always possible to find an ellipseof the right dimension and shape that matches the trianglecorner. To a first approximation, the interaction of light witha triangle, a cone or a tip end can be described simply byusing an elongated ellipse. A more detailed description ofthe interaction between light and triangles of various shapescan be found in [151] where Martin and collaborators showthat triangles of specific sizes and shapes are expected toenhance the electric field at different positions when changingthe wavelength or polarization. Light concentrates in differentareas around the triangle depending on the components of theelectric field. When particles are very small compared withthe wavelength, polarization and wavelength act together tocontrol locally the intensity and the localization of the areawhere the light is concentrated. These concentrations of light,for which the word hot-spots is often applied, are due to bothgeometrical and resonance (plasmon) effects.11 By smooth we mean without sharp corners or edges.

Emission coupling between metal particle and fluorophore.In the case of fluorescence, surface plasmon resonancesdo not always enhance the emission properties of thefluorophore [154]. Non-radiative decay rate, or quenching,is also enhanced relatively to the fluorophore–metal particlesdistance [154, 155]. Quenching properties of metal particlesand their sensing ability have attracted much attention inthe past five years but they are beyond the scope of thispaper [156,157]. However the coupling between the resonanceof the metal particle and the emission of the fluorophore (thedipole) are of special importance [137,158]. In the publicationof Thomas et al [137], the authors reveal that:

• close to the plasmon resonance, the energy position of themaximum of the radiative and non-radiative decay ratemoves with respect to the fluorescence emission in freespace,

• the emission rate tends to zero when R, the distancebetween the fluorophore and the metal particle, tends tozero,

• the local-field enhancement and the fluorescenceenhancement are redshifted compared with the plasmonwavelength for the metal particle in free space, and

• the enhancement is much higher when the orientation ofthe fluorescent dipole is along the main axis of the metalparticle (when the particle is not strictly symmetric).

To elaborate on this, consider a fluorescent moleculeinteracting with a single metal particle of diameter d = 2rp.The axis connecting the particle and the fluorescent moleculeis Oz and the distance between the molecule and the metalparticle is z + rp = z + d/2. The local excitation field E(r0)

of the molecule can be expressed in terms of the free spaceGreen’s function:

E(r0) = E0(r0) + G0(r0, r)αeffE0(r0), (34)

where E0 is the incident field. Equation (34) can be introducedinto equation (22) to yield the normalized excitation ratewithin the dipole approximation. Anger et al demonstratedin a recent paper [129] that the dipole approximation, whilegiving a fairly good estimate of the excitation rate �exc (seeequation (22)), strongly overestimates the quantum yield atshort distances and does not predict fluorescence quenching(see figure 18(b)). Higher multipole orders are needed fora more accurate description of the non-radiative rate. Angeret al use a multiple multipole (MMP) method [128] to correctlydescribe the observed variation of the fluorescence rate withrespect to the particle–molecule distance R (see figure 18(a)).They observed that the fluorescence enhancement of nile bluemolecule reaches a maximum at a distance of z = 5 nmfor d = 80 nm gold particle. Their experiment has beencarried out with a laser irradiated particle at λ = 637 nm andthey note that the fluorescence enhancement is strongest forλ = 680 nm and weakest for λ = 485 nm, showing the redshiftof the excitation frequency with respect to the surface plasmonresonance of the gold particle.

The effect of the surrounding medium, substrate orwater for instance, on the fluorescence enhancement has beenstudied experimentally in ensemble measurements with silverparticles [147]. An absorbing substrate can have a dramaticeffect on the enhancement factor [138].

19


Figure 18. Fluorescence rate as a function of goldparticle–fluorophore distance z for a vertically oriented molecule.Inset is the schematic of the experiment. (a) Comparisonexperiment–model for an embedded molecule; the solid curve is theMMP simulation; the dots are for the experiment performed with ad = 80 nm gold sphere on a single nile blue molecule.(b) Calculated fluorescence rate γem of a free fluorophorenormalized to the free space value for different particles of diameterd. The solid curves are the result of MMP calculation whereas thedashed curves correspond to the dipole approximation which failsfor short distance. Reprinted with permission from [129]. Copyright2006 by the American Physical Society.

Tip enhanced fluorescence. One of the clear demonstrationsof enhanced fluorescence near single metal particles ispresently the enhancement of fluorescence in so-calledapertureless near-field optical microscopy where the probe isa metal tip or a dielectric tip covered with metal. Due to thequenching of the fluorescence by the metal, the enhancementproperties of metal near the plasmon resonance are hardto detect [109, 110, 112]. Novotny and collaborators [106]reduced the influence of quenching using J-aggregates andphotosynthetic membranes in which energy transfer are veryfast. This increases the radiative transfer with respect tothe non-radiative transfer via the metal. They were able toinduce 2-photon absorption fluorescence of J-aggregates witha relatively low illumination power when a gold tip is lessthan 30 nm from the fluorophore. The fluorescence increasesdramatically with the tip–fluorophore distance. This is thefirst clear evidence of metal tip enhanced fluorescence. It isworth noting that in this experiment, the geometry of the tiphas been chosen to improve the enhancement of the electricfield [159]. Simulation and theoretical considerations haveshown how polarization direction and incidence strongly affectthe enhancement properties of metal tips [159–162].

The existence of extremely strong field gradients atthe tip apex have been demonstrated in several geometries[161, 163, 164]. The total electric field generated bya plane electromagnetic wave incident on a perfectlyconducting conical tip situated above a dielectric surfacecan be investigated using the above analytical expressionof the electric field at short distance r from the tip apex(kr 1) [127, 160, 165]:

E = k(kr)ν−1 sin β

(ur +

uθ

ν

∂

∂θ

)a(θ0, θ, αc), (35)

where k = ω/c is the wavevector, β is the polarization angle,a is a function of the angle of incidence θ0, the angle ofobservation θ and the half-angle of aperture αc of the conetip only. ur and uθ are the unit vectors in spherical coordinatesand ν depends on the half-angle αc, 0 < ν < 1. The fieldgiven by equation (35) is highly enhanced near the tip apex.The spatial distribution of the electric field does not depend onthe illumination condition except via the polarization angle β.

Using the strong field gradient near a metal tip, Quakeand his group chose to control the light and not thefluorophore [108]. They found an enhancement of thesame order of magnitude as Novotny, typically 20. In thepreviously cited studies, the enhancement due to surfaceplasmon resonance is overshadowed by the strong geometrical(quasi-static) enhancement of the electric field due to theconfinement of the electric lines at the end of the metal cone.Both Quake and Novotny chose to work out of resonance. Butrecently, Richard and co-workers [111] demonstrated that neara gold tip and at a wavelength close to the surface plasmonresonance frequency, the competition between quenchingand plasmonic enhancement of the fluorescence gives anenhancement factor of roughly 3 to 5 when the gold tip is at adistance of 10–30 nm from the fluorescent particle. Althoughtheir model approximates the tip by a simple gold sphere, theyshow promising results for tip induced plasmonic enhancedfluorescence, as in the work by Sugiura and Okada who usedsimilarly the plasmon excitation of a gold sphere to excitespecifically the fluorescence of YOYO-1 in DNA [166].

However the result of the competition between electro-static and plasmonic enhancements is not straightforward. Thetwo processes do not necessarily add up to give a higher flu-orescence output. The term nanoantennas refers to deviceswith extremely high enhancement at the nanoscale. Scalingmicrowave antennas down to the visible frequency has numer-ous potential applications. But the problems of the nanoscaledimension add up here to the peculiar behaviour of the dielec-tric function of metals in the visible domain. However, a superlight emitter [167, 168] has been recently realized which is agreat promise in the field of nanophotonics.

3.3.2. Nanoholes. Another common metal nanoobject is ametal hole [169]. From Babinet’s principle [125], it is easy topicture that interactions of such a hole with light are similar tothose of a nanodisc. Webb and collaborators demonstrated theunique properties of metal subwavelength holes to selectivelyexcite given fluorophores [170,171] due to the non-propagativenature of the electric field inside the holes. The expression forthe diffracted fields in the plane of an aperture is based on theanalytical expression originally proposed by Bethe [172] andcorrected by Bouwkamp [173, 174]. A theoretical formalismof the optical near-field generated by an aperture using Fourieroptics is given in [175]. The above model is based on ainfinitely thin metal layer but nevertheless perfectly describesthe most important feature of a nanohole in a metal film:inside the hole, the main component of the electric field isthe longitudinal Ez component perpendicular to the metalplane [176]. As Ez is non-propagative, these nanoholes aresometimes called zero mode waveguide [170], see figure 19(b).

20


Figure 19. Fluorescence enhancement above a nanohole.(a) Fluorescence decay curves of Rhodamine 6G in open solutionand into a d = 150 nm nanohole. Reprinted with permissionfrom [177]. Copyright 2005 by the American Physical Society.(b) Three-dimensional finite-element time-domain simulation of theintensity distribution for a d = 50 nm nanohole on a silica substratein solution. Reprinted from [170] with permission from AAAS.(Colour online.)

The effect of a nanohole on the fluorescent propertiesof a single fluorophore has been investigated using a SNOMaperture in 2000 [71]. It was shown that the relative positionsof the fluorescent dipole and the hole strongly influence thedirection of emission of fluorescence. Using the multiplemultipole (MMP) method [128], the authors were able toexplain how the metal hole affects the angular emissionproperties of the fluorophore. With a much smaller object,this effect is similar to the one demonstrated in section 2.2,where enhancement can be triggered off by redirecting thefluorescence emission [33, 42]. In a recent experiment, theenhancement effect is proven to be of the order of 15 [177,178].Although the authors are looking for a decrease in theobservation volume, they demonstrate that the small aperture:

• postpones the fluorescence saturation because the excitedvolume is smaller, see figure 19(b), and fewer fluorophoresare exited at the same time,

• increases the local excitation intensity by increasing thefluorescence rate, see figure 19(a).

In the paper on extraordinary optical transmission byEbbesen et al [169], not only does each hole have a peculiarelectromagnetic field distribution [179] but the periodicity ofthe holes also plays a role [180–182] as well as the scatteringof the surface plasmon [183]. Because of the periodicity of theholes on the metal surface, the extension of the confinement

of the electric field perpendicular to the metal surface (inthe direction of Oz) can be several micrometres long, givingrise to multiple high intensity wheels [184]. These canbe used as multiple subwavelength sources to increase thespeed and sensitivity of fluorescence detection. To increasethe transmission, there have been recent propositions to usespecifically shaped holes with rings [185] or coaxial designs[186]. None of these have been used for fluorescence imagingso far. However, the specific properties of a hole plasmon havebeen shown to be efficient for sensing, in an attempt to resolveadsorption of proteins [187].

3.4. Plasmon and electromagnetic coupling

3.4.1. Coupled nanoparticles. As we have already seen,multi featured plasmon resonance can be obtained frommore complex nanoparticles. With spherical symmetry(for instance, in nanocages [188] or nanoshells [189]) orwithout (ellipses, triangle, etc), hybridization of simple particleplasmon resonance takes place [190]. Different resonancemodes appear and are blue- or red-shifted with respect to thesizes and shapes of the particles.

Similarly, when two or more particles are close toeach other, the overall system is similar to an ensemble ofelementary particles. Contrary to a complex object withspherical symmetry, polarization of the incident light is crucialfor it [140, 191]. When two particles are close enough toeach other, the dipole of each particle induces a dipole onthe other. This dipole–dipole interaction ‘hybridizes’ theelementary localized plasmon resonance. To decrease thewidth of the surface plasmon resonance and increase the fieldenhancement, a possible choice is a line of metal spheres in aso-called ‘chain’ of particles. Bergman and Stockmann [192]have demonstrated how a self-similar chain of particles withdecreasing diameters can increase the amplitude of the electricfield by more than 3 orders of magnitude (see figure 20(a).The chain is self-similar when the radii Ri of the particlesi and the distance di,i+1 between the particle i and i + 1follow the relations Ri+1 = κ Ri and di+1,i+2 = κ di,i+1 withκ 1. The condition on the distance ensures that the field ofa given particle is only a weak perturbation of the previous andbigger particle. The high enhancement of the electric field iscalculated in the electrostatic approximation. If each particleenhances its applied field by a factor α, the cumulative effectof the chain of particles results in an enhancement factor ofαn where n is the number of particles in the chain. Thesecalculations were specifically done in order to find the bestsystem for enhancement of Raman scattering but they are easilygeneralized for the enhancement of fluorescence.

Due to the difficulty to control precisely the distancebetween several different particles (metal particles, fluorescentparticles etc), relatively few experimental studies have beendone so far for fluorescence with coupled nanoparticles [194].The use of organic molecules could prove very useful to controlthe distance and position of the nanoparticles. Most of thepublished work deals with SERS [195, 196], even if the lattergroup published a fine theoretical analysis of Raman scatteringcompared with fluorescence enhancement [27]. The high

21


(a)

(b)

Figure 20. Between metal nanoparticles, electric field and opticalforce inside the gap are enhanced. (a) In a self-similar chain of 3particles, in the gap between the smallest silver spheres the electricfield |E| can exceed the applied field by more than 3 orders ofmagnitude. Calculations show the enhancement at λ = 380 nm.Inset: Shape of the self-similar chain of particles. Reprinted withpermission from [192]. Copyright 2003 by the American PhysicalSociety. (b) Conservative force �FC and dissipative force �FD onRhodamine 6G (R6G) at a distance x from the gap of a two-metalparticles system at λ = 550 nm in water. Silver spheres have aradius R = 45 nm and the interparticle distance (the gap) isd = 1 nm. Reprinted with permission from [193]. Copyright 2002by the American Physical Society.

intensity gradient occurring inside the gap leads to opticalforces that can be a source of optical trapping of a singlemolecule. As demonstrated in [193], the time-averaged opticalforce on a fluorophore is

〈 �F 〉 = I0n

cε0(α′ �∇M + α′′ �kM), (36)

where n is the refractive index of the surrounding mediumand I0 is the incident intensity. The isotropic complexpolarizability is α(ω) = α′(ω) + iα′′(ω) and M is the localenhancement evaluated by Xu and Kall using generalized Mietheory. Equation (36) is a sum of two terms. The first oneis a conservative force FC whereas the second one standsfor a dissipative force FD. Because evanescent multipolarcontributions dominate the optical near-field in the gap region,molecular trapping is effective only for very small interparticledistances d (M ∝ d−2). When |〈 �FC〉| � |〈 �FD〉| the opticaltrap is stable, which is the case for silver dimer at λ = 550 nm(see figure 20(b)).

For specific shapes of particles, the electromagneticcoupling inside the gap can lead to huge enhancements[197–199]. Optimized nanostructures are designed to increase

the radiation efficiency of the fluorescent particle, in a similarway as antennas in the radio-wave regime. Hence theemitter couples to the electromagnetic field inside the gapof the antenna in the visible regime. Optimum couplingrequires impedance matching depending on the wavelengthand polarization of light [168]. Because the dielectric functionof metal in the visible regime and in the radio-wave regimediffers significantly (ohmic losses, plasmon resonance), theoptimized antennas are significantly different in the tworegimes.

Hecht et al [168] demonstrate that the use of an antennanear a single emitter helps to tailor the emission and absorptioncross section. Increase in both cross sections allows creationof a superemitter. Under the influence of the antenna, theeffective absorption cross section in the (x, y) plane wherethe emitter is located becomes Cabs(x, y) = |G(x, y)|2C0 ·|G(x, y)| is the field enhancement factor, defined as the ratio ofthe projection of the electric field at the position of the emitteronto the dipole moment with and without the antenna. Givena quantum efficiency in free space Q0 ≈ 1, the normalizedlifetime rate θ of the excited state of a 2-level system and thenormalized count rate γ are

θ = τ(x, y)

τ0, γ = Q(x, y)|G(x, y)|2 = R(x, y)/R0,

where τ(x, y) and R(x, y) are the measured lifetime and countrate, respectively. The subscript 0 denotes free space valuesand is approximated in [168] by value far from the antenna.Using the reciprocity theorem [96] Cabs/Cabs,0 = �r/�r,0 =|G(x, y)|2. We arrived at

γ (x, y)θ(x, y) = Q(x, y)2,

γ (x, y)/θ(x, y) = �2r (x, y)/�2

r,0, (37)

while assuming that the absorption and emission cross sectionsare similar [158, 200], according to [201]. By raster scanninga nanoantenna across a quantum dot, Q(x, y), �r(x, y) and�nr(x, y) = [1/Q(x, y) − 1]�r(x, y) can be determined fromequations (37) and the measurements of τ(x, y) and R(x, y)

(see figure 21). Increase of �nr and simultaneous decrease of Q

outside the quantum dot in the centre of the images indicate thatnon-radiative decays prevail when the quantum dot is beneaththe nanoantenna. This effect is even more pronounced for theperpendicular polarization (compare (c) and (f ) in figure 21),allowing for a reduction of the central spot size along thedirection of the nanoantenna. The difference between theposition of the extrema of Q(x, y) and �nr and the positionof �r demonstrate for [168] that field enhancement and non-radiative decay channels are independent, even if they are bothprominent close to the metal structure. The increase of Q whenthe nanoantenna is in the proximity of the quantum dot is anantenna effect and the system ’nanoantenna + quantum dot’ isan effective quantum emitter.

Using arrays of particles could prove useful tocouple plasmon polariton modes away from the excitationsource [202–204]. Until now, this has been used foroptoeletronic applications, such as beamsplitters [205] andwave-guides [203]. But chains of metal particles could be usedfor adaptative local sources [206] as one can tailor the plasmonresonance of such chains and increase non-linear effects [207].

22


Figure 21. Maps of the quantum efficiency Q(x, y) (left column), the radiative rate �r(x, y)[107 s−1] (centre column) and non-radiative rate�nr(x, y)[107 s−1] (right column) for the 2 orthogonal polarizations, parallel (top) and perpendicular (bottom) to the axis of the nanoantenna.Inset: SEM image of the nanoantenna. Reprinted with permission from [168]. Copyright 2005 by the American Physical Society. (Colouronline.)

3.4.2. Long range coupling, random films and hot spots.In a mix of metal particles of various sizes, localizedsurface plasmon (LSP) resonance occurs at various photonenergies depending on the wavelength and the polarizationof the incident light. Due to LSP resonance different highintensity EM fields couple together only for distances of lessthan 10 nm. Because of their propagative nature, surfaceplasmon polaritons (SPP, see section 2.1) can couple objectsand resonances separated by larges distances, up to severalmicrometres. The coupling of the localized modes of LSPwith long range interactions by means of the propagativemodes of SPP occurs in random films [82, 132], the so-calledsemicontinuous films [72]. In such films the propagationof SPP enhances the resonant effect of LSP which in turngives rise to higher intensity of the propagating SPP wave.Not only does the random nature of the LSP modes broadenthe accessible lauching angles for SPP (see figure 22(a)) butthe various shapes also increase the number of accessibleresonance wavelengths and polarization.

The peculiarity of semicontinuous films lies in the rangeof accessible shapes and sizes of metal particles on the surface.They are usually prepared by thermal evaporation of a metal ona dielectric surface followed by annealing. Metal evaporationends around the percolation threshold, when the firstcontinuous electric path between opposite sides of the substrateappears. Around the percolation threshold (see figure 22(b)),almost every size of metal particles is present in the film,giving rise to resonances all over the film from the visibleto the far-infrared. Since 1969 these films are known to havepeculiar properties [209–211] when the broadband absorptionin the infrared range was first discover. Since then, theyhave been extensively studied both experimentally [209, 210]

and theoretically [81, 85, 212–216], as SERS substrates, forwhich their properties are ideal. These specific propertieshave been linked to the localization of light on roughfilms [217, 218] and Anderson localization [219, 220]. Thefirst experiments demonstrating localization of light have beencarried out on rough gold films [221], aggregates [222, 223]and colloidal films [224] before successive demonstrationsof the localization on semicontinuous films [220]. The sizeof the high intensity peaks, called ‘hot-spots’, where light islocalized, is much smaller than the wavelength of light. Infigure 23, the effect of the change of wavelength predictedin [81, 85] is experimentally shown [225]. A slight changein the wavelength or in the polarization properties of lightallows to change the position of the ‘hot spots’. Becauseof their strong non-linear properties, random metal films areespecially important for non-linear optical application [72,80].The local distribution of the electromagnetic field allows us todramatically increase non-linear effect [77, 79, 226, 227], thatare used for multiphoton enhanced fluorescence [13, 228], asshown in figure 24.

4. Conclusion

Surface enhancement fluorescence (SEF) increases moleculardetection efficiency by tailoring the electromagnetic environ-ment and mastering the local environment of the fluorophore.SEF does not only amplify the fluorescence molecular emis-sion but also modifies fluorescence lifetime, photobleaching,intermolecular coupling and emission patterns. More gener-ally SEF includes the ability to tailor all processes involved influorescence. SEF dramatically opens up the range of possibleapplications based on fluorescence, including emission and

23


Ref

lect

ion

[%]

Incident Angle [degres]

0

100

0 90iSPP

(a)

(b)

Figure 22. (a) A schematic of the specular reflection of glass and thin metal films on glass with respect to the incident angle. Glass (longdashes), continuous metal film (continuous line) and semicontinous film (short dashes). The curves have been horizontally shifted forcomparison. The peak of the plasmon polariton resonance occurs at a precise angle iSPP for continuous films. The plasmon peak is muchbroader for semicontinuous films. (b) TEM images of four gold-dielectric films with different metal thicknesses. From left to right 20, 45,65 and 105 Å equivalent gold thickness. The third image from the left is taken around the percolation threshold (colour online). At thisprecise moment of the deposition, a first electrical path appears along the film. Reprinted from [208] with kind permission from SpringerScience and Business Media.

Figure 23. Calculated (top) and experimental (bottom) SNOM images of localized optical excitations in percolation gold-glass film on thesame spot for two different wavelengths, λ = 720 nm two (left) and λ = 790 nm (right). Reprinted with permission from [225]. Copyright2001 by the American Physical Society.

24


Figure 24. Two-photon laser-scanning fluorescence ofpolymer/silver-cluster composites in false colour. Overlay of threeimages taken at three different excitation wavelengths: λ = 720 nmin blue, λ = 820 nm in green, λ = 890 nm in red (scale bar: 10 µm).Reprinted with permission from [13]. Copyright 2002 AmericanChemical Society. (Colour online.)

detection devices (LED, biosensors, etc). Breakthroughs inboth sensitivity and confinement have also been reported inimaging and microscopy for biology, medical research anddiagnosis.

The fluorophore excitation and emission are stronglymodified by near-field coupling of the fluorophore withevanescent surface modes. Plasmon modes and theirassociated evanescent waves are known to change dramaticallythe electromagnetic field near an interface. Consequently,surfaces with plasmonic properties are the prominent SEFsubstrates. Flat dielectric surfaces are already efficientSEF surfaces; however metallic nanostructured surfaces(corrugated or rough surfaces) reach much higher fluorescenceenhancements. Nanoparticles can be advantageously coupledto fluorophores to locally enhance the fluorescence whendeposited on surfaces or as free probes.

A simple flat surface already gives a typical 10- to 20-fold fluorescence enhancement. More complex surfaces andstructures designed to control plasmon modes and couplethem to radiative light modes, in order to recover theenergy, give access to even higher amplifications. Besides,excitation of the fluorophores through plasmon modes resultsin additional molecular detection efficiency by amplificationof the excitation field. Furthermore, specific illuminationtechniques (i.e. Kretschmann’s configuration) allow betterexcitation of the fluorophore through the excitation of plasmonmodes. For instance, in the case of corrugated surfaces,surface plasmons are efficiently coupled to propagating lightat specific angles and can thus contribute to SEF through boththe excitation and the emission processes.

Localized plasmon resonances in simple and complexstructures provide also very intense electromagnetic enhance-ment (for instance in ‘hot spots’). This is of particularinterest for local probing, confined excitation, non-linearoptical phenomena and intense fluorescence enhancement.

Consequently, these nanostructures are very efficient SEF sur-faces with local enhancements of several orders of magnitude.

The development of plasmonics, together with theability to design surfaces with well-defined patterns andstructures on the nanometric scale, opens up new possibilitiesin SEF. Smart probes are developed which use plasmonproperties and fluorophores, such as the nanoshell filledwith dyes, predicted theoretically [229], or beamingemitted fluorescence using nanoapertures with designedsubwavelength nanopatterns [179]. The recent use of two-dimensional photonic crystal slabs and engineering of theirleaky modes to overlap with the absorption and emission bandof fluorophore demonstrate that the fluorescence intensity canbe enhanced by a factor of up to 100 compared with unpatternedsurfaces [230].

The processes described for emission control and en-hancement of the fluorophores are readily applicable to othersources of molecular luminescence (such as chemilumines-cence, electroluminescence and phospholuminescence) andtheir applications. The popularity of SEF will steadily grow interms of applications in the next few years. In particular in bi-ological imaging, enhancement of fluorescence via plasmonsis a fast growing field. Beyond the basic need for enhance-ment provided by SEF, other aspects brought by the modifiedfluorophore environment appear to be potentially very interest-ing, in particular the channelling of energy via plasmon cou-pling [231,232]. Although the coupling between fluorophoresand nanostructures with simple geometry is well understood,the situation for more complex nanostructures, such as per-colated or roughened metallic thin films, is still being inves-tigated. Alternatively, SEF phenomena can be used to probethe local EM field of complex surfaces, as a new near-fieldimaging technique.

Acknowledgments

The authors wish to thank Remi Carminatti, Sandrine Leveque-Fort and Claude Boccara for fruitful discussions, Eric Le Moal,Yannick Goulam, Sebastien Ducourtieux, Laurent Williameand Romain Lecaque for their experimental efforts, NatalieMalikova for carefully re-reading the manuscript and the editorfor his constant support. SG wishes to thank Jean-ClaudeRivoal, Alexander Kildishev and Vladimir M Shalaev.

References

[1] Purcell E M 1946 Spontaneous emission probabilities atradio frequencies Phys. Rev. 69 681

[2] Rigneault H, Lemarchand F and Sentenac A 2000 Dipoleradiation into grating structures J. Opt. Soc. Am. A17 1048–58

[3] Danz N, Waldhausl R, Brauer A and Kowarschik R 2002Dipole lifetime in stratified media J. Opt. Soc. Am. B 19412–19

[4] Gerard J M, Sermage B, Gayral B, Legrand B, Costard E andThierry-Mieg V 1998 Enhanced spontaneous emission byquantum boxes in a monolithic optical microcavity Phys.Rev. Lett. 81 1110–13

[5] Michler P, Kiraz A, Becher C, Schoenfeld W V, Petroff P M,Zhang L, Hu E and Imamoglu A 2000 A quantum dotsingle-photon turnstile device Science 290 2282–5

25

http://dx.doi.org/10.1364/JOSAA.17.001048

http://dx.doi.org/10.1103/PhysRevLett.81.1110

http://dx.doi.org/10.1126/science.290.5500.2282


[6] Lodahl P, Floris van Driel A, Nikolaev I S, Irman A,Overgaag K, Vanmaekelbergh D and Vos W L 2004Controlling the dynamics of spontaneous emission fromquantum dots by photonic crystals Nature 430 654–7

[7] Kress A, Hofbauer F, Reinelt N, Kaniber M, Krenner H J,Meyer R, Bohm G and Finley J J 2005 Manipulation of thespontaneous emission dynamics of quantum dots intwo-dimensional photonic crystals Phys. Rev. B (Condens.Matter Mater. Phys.) 71 241304

[8] Nomura M, Iwamoto S, Yang T, Ishida S and Arakawa Y2006 Enhancement of light emission from single quantumdot in photonic crystal nanocavity by using cavity resonantexcitation Appl. Phys. Lett. 89 241124

[9] Barnes W L, Dereux A and Ebbesen T W 2003 Surfaceplasmon subwavelength optics Nature 424 824–30

[10] Glass A M, Wokaun A, Heritage J P, Bergman J G, Liao P Fand Olson D H 1981 Enhanced two-photon fluorescenceof molecules adsorbed on silver particle films Phys. Rev. B24 4906–9

[11] Moskovits M 1985 Surface-enhanced spectroscopy Rev. Mod.Phys. 57 783–826

[12] Schatz C C and Van Duyne R P 2002 Handbook ofVibrational Spectroscopy (New York: Wiley)

[13] Wenseleers W, Stellacci F, Meyer-Friedrichsen T, Mangel T,Bauer C A, Pond S J K, Marder S R and Perry J W 2002Five orders-of-magnitude enhancement of two-photonabsorption for dyes on silver nanoparticle fractal clustersJ. Phys. Chem. B 106 6853–63

[14] Barnes W L 1998 Fluorescence near interfaces: the role ofphotonic mode density J. Mod. Opt. 45 661–99

[15] Lakowicz J R 2001 Radiative decay engineering: biophysicaland biomedical applications Anal. Biochem. 298 1–24

[16] Lakowicz J R, Shen Y B, D’Auria S, Malicka J, Fang J Y,Gryczynski Z and Gryczynski I 2002 Radiative decayengineering: II. Effects of silver island films onfluorescence intensity, lifetimes and resonance energytransfer Anal. Biochem. 301 261–77

[17] Lakowicz J R, Malicka J, Gryczynski I, Gryczynski Z andGeddes C D 2003 Radiative decay engineering: the role ofphotonic mode density in biotechnology J. Phys. D: Appl.Phys. 36 R240–9

[18] Lakowicz J R 2004 Radiative decay engineering: III. Surfaceplasmon-coupled directional emission Anal. Biochem.324 153–69

[19] Gryczynski I, Malicka J, Gryczynski Z and Lakowicz J R2004 Radiative decay engineering: IV. Experimentalstudies of surface plasmon-coupled directional emissionAnal. Biochem. 324 170–82

[20] Lakowicz J R 2005 Radiative decay engineering:V. Metal-enhanced fluorescence and plasmon emissionAnal. Biochem. 337 171–94

[21] Lakowicz J R 1999 Principle of Fluorescence Spectroscopy2nd edn (New York: Kluwer/Plenum)

[22] Lichtman J W and Conchello J-A 2005 Fluorescencemicroscopy Nature Methods 2 910–19

[23] Drexhage K H 1974 Interaction of light with monomoleculardye layers Prog. Opt. 12 163–232

[24] Chance R R, Prock A and Sylbey R 1978 Molecularfluorescence and energy transfer near interfaces Advancesin Chemical Physics vol 37, ed I Prigogine and S A Rice(New York: Wiley-Interscience) p 1–65

[25] Wylie J M and Sipe J E 1984 Quantum electrodynamics nearan interface Phys. Rev. A 30 1185–93

[26] Carminati R, Greffet J-J, Henkel C and Vigoureux J M 2006Radiative and non-radiative decay of a single moleculeclose to a metallic nanoparticle Opt. Commun. 261 368–75

[27] Johansson P, Xu H and Kall M 2005 Surface-enhancedRaman scattering and fluorescence near metalnanoparticles Phys. Rev. B 72 035427

[28] Balzer F, Bordo V G and Rubahn H-G 1997 Frequency shiftsand lifetime changes of sodium atoms near rough metalsurfaces Opt. Lett. 22 1262–4

[29] Ruckstuhl T, Enderlein J, Jung S and Seeger S 2000Forbidden light detection from single molecules Anal.Chem. 72 2117–23

[30] Suhling K, Siegel J, Phillips D, French P M W,Leveque-Fort S, Webb S E D and Davis D M 2002Imaging the environment of green fluorescent proteinBiophys. J. 83 3589–95

[31] Enderlein J, Ruckstuhl T and Seeger S 1999 Highly efficientoptical detection of surface-generated fluorescence Appl.Opt. 38 724–32

[32] Ruckstuhl T and Verdes D 2004 Supercritical anglefluorescence (saf) microscopy Opt. Express 12 4246

[33] Benisty H, Stanley R and Mayer M 1998 Method of sourceterms for dipole emission modification in modes ofarbitrary planar structures J. Opt. Soc. Am. A15 1192–201

[34] Amos R M and Barnes W L 1997 Modification of thespontaneous emission rate of Eu3+ ions close to a thinmetal mirror Phys. Rev. B 55 7249–54

[35] Choumane H, Nelep C, Ha N, Chardon A, Reymond G O,Goutel C, Cerovic G, Vallet F, Weisbuch C and Benisty H2005 Double interference fluorescence enhancement fromreflective slides: application to bi-color microarrays Appl.Phys. Lett. 87 031102

[36] Moiseev L, Unlu M S, Swan A K, Goldberg B B andCantor C R 2006 DNA conformation on surfacesmeasured by fluorescence self-interference Proc. NatlAcad. Sci. 103 2623–8

[37] Le Moal E, Fort E, Leveque-Fort S, Janin A, Murata H,Cordelieres F P, Fontaine-Aupart M-P and Ricolleau C2007 Mirror slides for high-sensitivity cell and tissuefluorescence imaging J. Biomed. Opt. 12 024030

[38] Le Moal E, Fort E, Leveque-Fort S, Cordelieres F P,Fontaine-Aupart M-P and Ricolleau C 2007 Enhancedfluorescence cell imaging with metal-coated slidesBiophys. J. 92 2150–61

[39] Lenne P-F, Etienne E and Rigneault H 2002 Subwavelengthpatterns and high detection efficiency in fluorescencecorrelation spectroscopy using photonic structures Appl.Phys. Lett. 80 4106

[40] Raether H 1988 Surfaces Plasmons on Smooth and RoughSurfaces and On Gratings (Springer Tracts in ModernPhysics vol 111) (Berlin: Springer)

[41] Weber W H and Eagen C F 1979 Energy transfer from anexcited dye molecule to the surface plasmons of anadjacent metal Opt. Lett. 4 236

[42] Vasilev K, Knoll W and Kreiter M 2004 Fluorescenceintensities of chromophores in front of a thin metal filmJ. Chem. Phys. 120 3439–45

[43] Enderlein J and Ruckstuhl T 2005 The efficiency ofsurface-plasmon coupled emission for sensitivefluorescence detection Opt. Express 13 8856–65

[44] Born M and Wolf E 1999 Principles of Optics 7th edn(Cambridge: Cambridge University Press)

[45] Goh J Y L, Somekh M G, See C W, Pitter M C, Vere K A andO’Shea P 2005 Two-photon fluorescence surface wavemicroscopy J. Microsc. 220 168–75

[46] Buchler B C, Kalkbrenner T, Hettich C and Sandoghdar V2005 Measuring the quantum efficiency of the opticalemission of single radiating dipoles using a scanningmirror Phys. Rev. Lett. 95 063003

[47] Liebermann T and Knoll W 2000 Surface-plasmonfield-enhanced fluorescence spectroscopy Colloid Surf. A171 115–30

[48] Yao D F, Yu F, Kim J Y, Scholz J, Nielsen P E, Sinner E Kand Knoll W 2004 Surface plasmon field-enhanced

26

http://dx.doi.org/10.1038/nature02772

http://dx.doi.org/10.1063/1.2408636


http://dx.doi.org/10.1103/PhysRevB.24.4906

http://dx.doi.org/10.1103/RevModPhys.57.783

http://dx.doi.org/10.1021/jp014675f

http://dx.doi.org/10.1006/abio.2001.5377

http://dx.doi.org/10.1006/abio.2001.5503

http://dx.doi.org/10.1088/0022-3727/36/14/203

http://dx.doi.org/10.1016/j.ab.2003.09.039



http://dx.doi.org/10.1038/nmeth817

http://dx.doi.org/10.1103/PhysRevA.30.1185

http://dx.doi.org/10.1016/j.optcom.2005.12.009


http://dx.doi.org/10.1021/ac991358k

http://dx.doi.org/10.1364/OPEX.12.004246

http://dx.doi.org/10.1364/JOSAA.15.001192


http://dx.doi.org/10.1063/1.1999018

http://dx.doi.org/10.1073/pnas.0511214103

http://dx.doi.org/10.1117/1.2722730

http://dx.doi.org/10.1063/1.1483116

http://dx.doi.org/10.1063/1.1640341

http://dx.doi.org/10.1111/j.1365-2818.2005.01525.x


http://dx.doi.org/10.1016/S0927-7757(99)00550-6


fluorescence spectroscopy in pcr product analysis bypeptide nucleic acid probes Nucl. Acids Res. 32 177–92

[49] Morigaki K and Tawa K 2006 Vesicle fusion studied bysurface plasmon resonance and surface plasmonfluorescence spectroscopy Biophys. J. 91 1380–7

[50] Yokota H, Saito K and Yanagida T 1998 Single moleculeimaging of fluorescently labeled proteins on metal bysurface plasmons in aqueous solution Phys. Rev. Lett.80 4606–9

[51] Ditlbacher H, Krenn J R, Felidj N, Lamprecht B, Schider G,Salerno M, Leitner A and Aussenegg F R 2002Fluorescence imaging of surface plasmon fields Appl.Phys. Lett. 80 404–6

[52] Stefani F D, Vasilev K, Bocchio N, Stoyanova N andKreiter M 2005 Surface-plasmon-mediatedsingle-molecule fluorescence through a thin metallic filmPhys. Rev. Lett. 94 023005

[53] Tang W T, Chung E, Kim Y-H, So P T C and Sheppard C J R2007 Investigation of the point spread function of surfaceplasmon-coupled emission microscopy Opt. Express15 4634–46

[54] Burghardt T P, Charlesworth J E, Halstead M F, Tarara J Eand Ajtai K 2006 In situ fluorescent protein imaging withmetal film-enhanced total internal reflection microscopyBiophys. J. 90 4662–71

[55] Borejdo J, Gryczynski Z, Calander N, Muthu P andGryczynski I 2006 Application of surface plasmon coupledemission to study of muscle Biophys. J. 91 2626–35

[56] Calander N 2004 Theory and simulation of surfaceplasmon-coupled directional emission from fluorophoresat planar structures Anal. Chem. 76 2168–73

[57] Axelrod D, Thompson N and Burghardt T 1983 Total internalreflection fluorescent microscopy J. Microsc. 129 19–28

[58] Axelrod D 2001 Total internal reflection fluorescentmicroscopy in cell imaging Traffic 2 764–74

[59] Borejdo J, Calander N, Gryczynski Z and Gryczynski I 2006Fluorescence correlation spectroscopy in surface plasmoncoupled emission microscope Opt. Express 14 7878

[60] Knoll W 2003 Surface-plasmon-fluorescence spectroscopyfor DNA hybridization reactions ChemInForm34 306–8

[61] Adams A, Moreland J and Hansma P K 1981 Angularresonances in the light emission from atoms near a gratingSurf. Sci. 111 351

[62] Aravind P K, Hood E and Metiu H 1981 Angular resonancesin the emission from a dipole located near a grating Surf.Sci. 109 95–108

[63] Adams A, Moreland J, Hansma P K and Schlesinger Z 1982Light emission from surface-plasmon and waveguidemodes excited by n atoms near a silver gratingPhys. Rev. B 25 3457–61

[64] Sullivan K G, King O, Sigg C and Hall D G 1994 Directional,enhanced fluorescence from molecules near a periodicsurface Appl. Opt. 33 2447–54

[65] Moreland J, Adams A and Hansma P K 1982 Efficiency oflight emission from surface plasmons Phys. Rev. B25 2297–300

[66] Lupton J M, Matterson B J, Samuel I D W, Jory M J andBarnes W L 2000 Bragg scattering from periodicallymicrostructured light emitting diodes Appl. Phys. Lett.77 3340–2

[67] Kock A, Beinstingl W, Berthold K and Gornik E 1988Surface plasmon polariton enhanced light emission fromSchottky diodes Appl. Phys. Lett. 52 1164–6

[68] Andrew P and Barnes W L 2001 Molecular fluorescenceabove metallic gratings Phys. Rev. B 64 125405

[69] Wedge S and Barnes W L 2004 Surface plasmon-polaritonmediated light emission through thin metal films Opt.Express 12 3673

[70] Okamoto K, Niki I, Shvartser A, Narukawa Y, Mukai T andScherer A 2004 Surface-plasmon-enhanced light emittersbased on InGaN quantum wells Nature Mater. 3 601–5

[71] Gersen H, Garcia-Parajo M F, Novotny L, Veerman J A,Kuipers L and van Hulst N F 2000 Influencing theangular emission of a single molecule Phys. Rev. Lett.85 5312–15

[72] Shalaev V M 2000 Nonlinear Optics of RandomMedia-Fractal Composites and Metal–Dielectric Films(Springer Tracts in Modern Physics vol 158) (Berlin:Springer)

[73] John S 1990 The localization of waves in disordered mediaScattering and Localization of Classical Waves in RandomMedia (Directions in Condensed Matter Physics vol 8)ed Ping Sheng (Singapore: World Scientific) pp 1–96

[74] Bozhevolnyi S I, Volkov V S and Leosson K 2002Localization and waveguiding of surface plasmonpolaritons in random nanostructures Phys. Rev. Lett.89 186801

[75] Biteen J S, Pacifici D, Lewis N S and Atwater H A 2005Enhanced radiative emission rate and quantum efficiencyin coupled silicon nanocrystal-nanostructured goldemitters Nano Lett. 5 1768–73

[76] Bowman J J, Senior T B A and Uslenghi P L E 1987Electromagnetic and Acoustic Scattering by Simple Shapes(Bristol, PA: Hemisphere)

[77] Boyd G T, Rasing Th, Leite J R R and Shen Y R 1984Local-field enhancement on rough surfaces of metals,semi-metals, and semiconductors with the use of opticalsecond-harmonic generation Phys. Rev. B 30 519–26

[78] Smolyaninov I I, Zayats A V and Davis C C 1997 Near-fieldsecond harmonic generation from a rough metal surfacePhys. Rev. B 56 9290–3

[79] Anceau C, Brasselet S, Zyss J and Gadenne P 2003 Localsecond-harmonic generation enhancement on goldnanostructures probed by two-photon microscopy Opt.Lett. 28 713–15

[80] Stockman M I, Bergman D J, Anceau C, Brasselet S andZyss J 2004 Enhanced second-harmonic generation bymetal surfaces with nanoscale roughness: nanoscaledephasing, depolarization and correlations Phys. Rev. Lett.92 057402

[81] Brouers F, Blacher S, Lagarkov A N, Sarichev A K,Gadenne P and Shalaev V M 1997 Theory of giant Ramanscattering from semicontinuous metal films Phys. Rev. B55 13234–45

[82] Stockman M 2005 Semiconductor quantum dots in metalnanostructures MAR05 Meeting of the American PhysicalSociety (Los Angeles, CA, USA)

[83] Larkin I A, Stockman M I, Achermann M and Klimov V I2004 Dipolar emitters at nanoscale proximity of metalsurfaces: giant enhancement of relaxation in microscopictheory Phys. Rev. B 69 121403(R)

[84] Weitz D A, Gramila T J, Genack A Z and Gersten J I 1980Anomalous low-frequency raman scattering from roughmetal surfaces and the origin of surface-enhanced ramanscattering Phys. Rev. Lett. 45 355–8

[85] Shalaev V M and Sarichev A K 1998 Nonlinear optics ofrandom metal–dielectric films Phys. Rev. B 57 13265–88

[86] Lambe J and McCarthy S L 1976 Light emission fromelectron tunneling Phys. Rev. Lett. 37 923–5

[87] Boyd G T, Yu Z H and Shen Y R 1986 Photoinducedluminescence from the noble metals and itsenhancementon roughned surfaces Phys. Rev. B 33 7923

[88] Downes A, Guaino P and Dumas P 2002 Color mapping byscanning tunneling microscopy: chemical analysis ofmetal surfaces Appl. Phys. Lett. 80 380–2

[89] Drachev V P, Khaliullin E N, Kim W, Alzoubi F, Rautian S G,Safonov V P, Armstrong R L and Shalaev V M 2004

27

http://dx.doi.org/10.1093/nar/gnh175

http://dx.doi.org/10.1529/biophysj.106.086074


http://dx.doi.org/10.1063/1.1435410


http://dx.doi.org/10.1364/OE.15.004634



http://dx.doi.org/10.1021/ac049925d

http://dx.doi.org/10.1034/j.1600-0854.2001.21104.x

http://dx.doi.org/10.1364/OE.14.007878

http://dx.doi.org/10.1002/chin.200338272

http://dx.doi.org/10.1016/0039-6028(81)90394-0

http://dx.doi.org/10.1016/0039-6028(81)90514-8



http://dx.doi.org/10.1063/1.1320023

http://dx.doi.org/10.1063/1.99193



http://dx.doi.org/10.1038/nmat1198



http://dx.doi.org/10.1021/nl051207z



http://dx.doi.org/10.1364/OL.28.000713








http://dx.doi.org/10.1063/1.1433918


Quantum size effect in two-photon excited luminescencefrom silver nanoparticles Phys. Rev. B 69 035318

[90] Bouhelier A, Bachelot R, Lerondel G, Kostcheev S, Royer Pand Wiederrecht G P 2005 Surface plasmon charactericticsof tunable nonlinear photoluminescence in single goldnanorods Phys. Rev. Lett. 95 267405

[91] Stockman M I, Bergman D J and Kobayashi T 2004 Coherentcontrol of nanoscale localization of ultrafast opticalexcitation in nanosystems Phys. Rev. B 69 054202

[92] Bergman D J and Stockman M I 2004 Can we make ananoscopic laser? Laser Phys. 14 409–11

[93] Drachev V P, Shalaev V M and Sarychev A K 2006 Ramanimaging and sensing apparatus employing nanoantennasTechnical Report 6985223 B2, US Patent

[94] Roth R M, Panoiu N C, Adams M M, Osgood R M,Neacsu C C and Raschke M B 2006 Resonant-plasmonfield enhancement from asymmetrically illuminatedconical metallic-probe tips Opt. Express 14 2921–31

[95] Synge E H 1928 A suggested method for extendingmicroscopic resolution into the ultra-microscopic regionPhil. Mag. 6 356–62

[96] Greffet J-J and Carminati R 1997 Image formation innear-field optics Prog. Surf. Sci. 56 133–237

[97] Girard C, Joachim C and Gauthier S 2000 The physics of thenear-field Rep. Prog. Phys. 63 893–938

[98] Pohl D W, Denk W and Lanz M 1984 Optical stethoscopy:image recording with resolution λ/20 Appl. Phys. Lett.44 651–3

[99] Harootunian A, Betzig E, Isaacson M and Lewis A 1986Super-resolution fluorescence near-field scanning opticalmicroscopy Appl. Phys. Lett. 49 674–6

[100] Lewis M K, Wolanin P, Gafni A and Steel D G 1998Near-field scanning optical microscopy of singlemolecules by femtosecond two-photon excitation Opt.Lett. 23 1111–13

[101] van Hulst N F, Veerman J-A, Garcia-Parajo M F and (Kobus)Kuipers L 2000 Analysis of individual (macro)moleculesand proteins using nearfield optics J. Chem. Phys.112 7799–810

[102] Sunney X and Dunn R C 1994 Probing single-moleculedynamics Science 265 361–4

[103] Zenhausern F, Martin Y and Wickramasinghe H K 1995Scanning interferomatric aperturless microscopy:optical imaging at 10 Angstrom resolution Science269 1083–5

[104] Bachelot R, Gleyzes P and Boccara A C 1995 Near-fieldoptical microscope based on local perturbation ofdiffraction spot Opt. Lett. 20 1924–6

[105] Inouye Y and Kawata S 1996 Reflection-mode near-fieldoptical microscope with a metallic probe tip for observingfine structures in semiconductor materials Opt. Commun.134 31–5

[106] Sanchez E J, Novotny L and Sunney X 1999 Near-fieldfluorescence microscopy based on two-photon excitationwith metal tips Phys. Rev. Lett. 82 4014

[107] Gresillon S, Ducourtieux S, Lahrech A, Aigouy L,Rivoal J C and Boccara A C 2000 Nanometer scaleapertureless near-field microscopy Appl. Surf. Sci.164 118–23

[108] Gerton J M, Wade L A, Lessard G A and Quake S R 2004Tip-enhanced fluorescence microscopy at 10 nanometerresolution Phys. Rev. Lett. 93 180801

[109] Frey H G, Witt S, Felderer K and Guckenberger R 2004High-resolution imaging of single fluorescent moleculeseith the optical near-field of a metal tip Phys. Rev. Lett.93 200801

[110] Yang T J, Lessard G A and Quake S R 2000 An aperturelessnear-field microscope for fluorescence imaging Appl.Phys. Lett. 76 378–80

[111] Huang F M, Festy F and Richards D 2005 Tip-enhancedfluorescence imaging of quantum dots Appl. Phys. Lett.87 183101

[112] Fragola A, Aigouy L and Boccara A C 2003 Interferenceeffects in apertureless near field fluorescence imagingAppl. Opt. 42 6880

[113] Michaelis J, Hettich C, Mlynek J and Sandoghdar V 2000Optical microscopy using a single-molecule light sourceNature 405 325–8

[114] Aigouy L, De Wilde Y and Mortier M 2003 Local opticalimaging of nanoholes using a single fluorescentrare-earth-doped glass particle as a probe Appl. Phys. Lett.83 147–9

[115] Chevalier N, Nasse M J, Woehl J C, Reiss P, Bleuse J,Chandezon F and Huant S 2005 CdSe single-nanoparticlebased active tips for near-field optical microscopyNanotechnology 16 613–18

[116] Aigouy L, Tessier G, Mortier M and Charlot B 2005Scanning thermal imaging of microelectronic circuitswith a fluorescent nanoprobe Appl. Phys. Lett.87 184105

[117] Yanagida T et al 2002 Near-field microscopy forbiomolecular systems Nano-Optics (Springer Series inOptical Sciences No 84) ed S Kawata et al (Berlin:Springer) pp 191–236, chapter 8

[118] Faraday M 1857 Experimental relations of gold (and othermetals) to light Phil. Trans. 147 145

[119] Mie G 1908 Beitrage zur optik truber speziell kolloidalermetallosungen Ann. Phys. (Lpz.) 25 377–445

[120] Van de Hulst H C 1981 Light Scattering by Small Particles(New York: Dover)

[121] Bohren C F and Huffman D R 1998 Absorption andScattering of Light by Small Particles 2nd edn (New York:Wiley)

[122] Matzler C 2002 Matlab functions for mie scattering andabsorption Research Report 2002-08, Institut furAngewandte Physik, University of Bern - CH,Sidlerstrasse 5 CH-3012 Bern Schweiz

[123] Gresillon S, Lecaque R, Williame L and Rivoal J-C2006 Single metal cluster imaging Appl. Phys. B84 167–73

[124] Pack A, Grill W and Wannemacher R 2003 Aperturelessnear-field optical microscopy of metallic nanoparticlesUltramicroscopy 94 109–23

[125] Jackson J D 1975 Classical Electrodynamics 2nd edn(New York: Wiley)

[126] Mathews J and Walker R L 1970 Mathematical Methods ofPhysics 2nd edn (Reading, MA: Addison-Wesley)

[127] Van Bladel J 1997 Singular Electromagnetic Field andSources (The IEEE/OUP Series on Electromagnetic WaveTheory) (Piscataway, NJ: IEEE Press)

[128] Novotny L and Hecht B 2006 Principles of Nano-Optics(Cambridge: Cambridge University Press)

[129] Anger P, Bharadwaj P and Novotny L 2006 Enhancement andquenching of single-molecule fluorescence Phys. Rev. Lett.96 113002

[130] Chaumet P C, Rahmani A, de Fornel F and Dufour J-P 1998Evanescent light scattering: the validity of the dipoleapproximation Phys. Rev. B 58 2310–15

[131] Frohlich H 1958 Theory of Dielectrics: Dielectric Constantand Dielectric Loss (Monographs on the Physics andChemistry of Materials) 2nd edn (London: OxfordUniversity Press)

[132] Zayats A V and Smolyaninov I I 2003 Near-field photonics:surface plasmon polaritons and localized surface plasmonsJ. Opt. A: Pure Appl. Opt. 5 S16–50

[133] Kreibig U and Vollmer M 1995 Optical Properties of MetalClusters (Series in Material Science vol 25) (Berlin:Springer)

28




http://dx.doi.org/10.1364/OE.14.002921

http://dx.doi.org/10.1016/S0079-6816(98)00004-5

http://dx.doi.org/10.1088/0034-4885/63/6/202

http://dx.doi.org/10.1063/1.94865

http://dx.doi.org/10.1063/1.97565

http://dx.doi.org/10.1063/1.481385



http://dx.doi.org/10.1016/S0030-4018(96)00555-X


http://dx.doi.org/10.1016/S0169-4332(00)00350-0



http://dx.doi.org/10.1063/1.125759

http://dx.doi.org/10.1063/1.2115073

http://dx.doi.org/10.1364/AO.42.006880

http://dx.doi.org/10.1038/35012545

http://dx.doi.org/10.1063/1.1589193

http://dx.doi.org/10.1088/0957-4484/16/4/047

http://dx.doi.org/10.1063/1.2123384

http://dx.doi.org/10.1098/rstl.1857.0011

http://dx.doi.org/10.1007/s00340-006-2232-z

http://dx.doi.org/10.1016/S0304-3991(02)00248-6



http://dx.doi.org/10.1088/1464-4258/5/4/353


[134] Kelly K L, Coronado E, Zhao L L and Schatz G C 2003 Theoptical properties of metal nanoparticles: the influence ofsize, shape and dielectric environment J. Phys. Chem. B107 668–77

[135] Granqvist C G and Hunderi O 1977 Optical properties ofultrafine gold particles Phys. Rev. B 16 3513

[136] Klar T, Perner M, Grosse S, von Plessen G, Spirkl W andFeldmann J 1998 Surface plasmon resonances in singlemetallic nanoparticles Phys. Rev. Lett. 80 4249–52

[137] Thomas M, Greffet J-J, Carminati R and Arias-Gonzalez J R2004 Single-molecule spontaneous emission close toabsorbing nanostructures Appl. Phys. Lett. 85 3863–5

[138] Xu G, Chen Y, Tazawa M and Jin P 2006 Influence ofdielectric properties of a substrate upon plasmonresonance spectrum of supported Ag nanoparticles Appl.Phys. Lett. 88 043114

[139] Kalkbrenner T, Hkanson U, Schadle A, Burger S, Henkel Cand Sandoghdar V 2005 Optical microscopy via spectralmodifications of a nanoantenna Phys. Rev. Lett. 95 200801

[140] Hao E and Schatz G C 2004 Electromagnetic fields aroundsilver nanoparticles and dimers J. Chem. Phys. 120 357

[141] Link S and El-Sayed M A 2000 Shape and size dependenceof radiative, non-radiative and photothermal properties ofgold nanocrystals Int. Rev. Phys. Chem. 19 409–53

[142] Krenn J R, Ditlbacher H, Schider G, Hohenau A, Leitner Aand Aussenegg F R 2003 Surface plasmon micro- andnano-optics J. Microsc. 209 167–72

[143] Das P C and Puri A 2002 Energy flow and fluorescence neara small metal particle Phys. Rev. B 65 155416

[144] Hayakawa T, Selvan S T and Nogami M 1999 Fieldenhancement effect of small Ag particles on thefluorescence from Eu3+-doped SiO2 glass Appl. Phys. Lett.74 1513–15

[145] Strohhofer C and Polman A 2002 Silver as a sensitizer forerbium Appl. Phys. Lett. 81 1414–16

[146] Naranjo L P, de Araujo C B, Malta O L, Santa Cruz P A andKassab L R P 2005 Enhancement of Pr3+ Luminescencein PbO-GeO2 glasses containing silver nanoparticles Appl.Phys. Lett. 87 241914

[147] Itoh T, Ozaki Y, Yoshikawa H, Ihama T and Masuhara H2006 Hyper-Rayleigh scattering and hyper-Ramanscattering of dye-adsorbed silver nanoparticels induced bya focused continuous-wave near-infrared laser Appl. Phys.Lett. 88 084102

[148] Schmitt J, Machtle P, Eck D, Mohwald H and Helm C A1999 Preparation and optical properties of colloidal goldmonolayers Langmuir 15 3256–66

[149] Perner M, Gresillon S, Marz J, von Plessen G, Feldmann J,Porstendorfer J, Berg K-J and Berg G 2000 Observation ofhot-electron pressure in the vibration dynamics of metalnanoparticles Phys. Rev. Lett. 85 792

[150] Bakker R M, Boltasseva A, Liu Z, Pedersen P H, Gresillon S,Kildishev A V, Drachev V P and Shalaev V M 2007 Nanoantenna plasmonics via near-field illumination Opt.Express 15 13682–8

[151] Kottmann J P, Martin O J F, Smith D R and Schultz S 2000Spectral response of plasmon resonant nanoparticles witha non-regular shape Opt. Express 6 213–9

[152] Malikova N, Pastoriza-Santos I, Schierhorn M, Kotov N Aand Liz-Marzan L M 2002 Layer-by-layer assembledmixed spherical and planar gold nanoparticles: control ofinterparticle interactions Langmuir 18 3694–7

[153] Goncalves M R and Marti O 2003 Experimental observationof the scattering of light by planar metallic nanoparticlesNew J. Phys. 5 160

[154] Dulkeith E, Ringler M, Klar T A, Feldmann J, Javier A Mand Parak W J 2005 Gold nanoparticles quenchfluorescence by phase induced radiative rate suppressionNano Lett. 5 585–9

[155] Kuhn S, Høakanson U, Rogobete L and Sandoghdar V 2006Enhancement of single-molecule fluorescence using a goldnanoparticle as an optical nanoantenna Phys. Rev. Lett.97 017402

[156] Dubertret B, Calame M and Libchader A J 2001Single-mismatch detection using gold-quenchedfluorescent oligonucleotides Nature Biotechnol.19 365–70

[157] Dulkeith E, Morteani A C, Niedereichholz T, Klar T A,Feldmann J, Levi S A, van Veggel F C J M,Reinhoudt D N, Moller M and Gittins D I 2002Fluorescence quenching of dye molecules near goldnanoparticles: radiative and nonradiative effects Phys. Rev.Lett. 89 203002

[158] Azoulay J, Debarre A, Richard A and Tchenio P 2000Quenching and enhancement of single-moleculefluorescence under metallic and dielectric tips Europhys.Lett. 51 374

[159] Novotny L, Sanchez E J and Sunney X 1998 Near-fieldoptical imaging using metal tips illuminated byhigher-order Hermite–Gaussian beams Ultramicroscopy71 21–9

[160] Cory H, Boccara A C, Rivoal J-C and Lahrech A 1998Electric field intensity variation in the vicinity of aperfectly conducting conical probe: application tonear-field microscopy Microw. Opt. Technol. Lett.18 120–4

[161] Martin O J F and Girard C 1997 Controlling and tuningstrong optical field gradients at a local probe microscopetip apex Appl. Phys. Lett. 70 705–7

[162] Aigouy L, Lahrech A, Gresillon S, Cory H, Boccara A C andRivoal J-C 1999 Polarization effects in aperturlessscanning near-field microscopy: an experimental studyOpt. Lett. 24 187–9

[163] Denk W and Pohl D W 1991 Near-field optics - microscopywith nanometer size fields J. Vacuum Sci. Technol. B9 510–13

[164] H’dhili F, Bachelot R, Lerondel G, Barchiesi D and Royer P2001 Near-field optics: direct observation of the fieldenhancement below an apertureless probe using aphotosensitive polymer Appl. Phys. Lett. 79 4019–21

[165] Aigouy L, Andreani F X, Boccara A C, Rivoal J C, Porto J A,Carminati R, Greffet J J and Megy R 2000 Near-fieldoptical spectroscopy using an incoherent light source Appl.Phys. Lett. 76 397–9

[166] Kawata S, Ohtsu M and Irie M (ed) 2002 Nano-Optics(Springer Series in Optical Sciences vol 84) (Berlin:Springer)

[167] Muhlschlegel P, Eisler H J, Martin O J F, Hecht B andPohl D W 2005 Resonant optical antennas Science308 1607–9

[168] Farahani J N, Pohl D W, Eisler H J and Hecht B 2005 Singlequantum dot coupled to a scanning optical antenna: atunable superemitter Phys. Rev. Lett. 95 017402

[169] Ebbesen T W, Lezec H J, Ghaemi H F, Thio T and Wolff P A1998 Extraordinary optical transmission throughsub-wavelength hole arrays Nature 391 667–9

[170] Levene M J, Korlach J, Turner S W, Foquet M,Craighead H G and Webb W W 2003 Zero-modewaveguides for single-molecule analysis at highconcentrations Science 299 682

[171] Dahlin A, Zach M, Rindzevicius T, Kall M, Sutherland D Sand Hook F 2005 Localized surface plasmon resonancesensing of lipid-membrane-mediated biorecognition eventsJ. Am. Chem. Soc. 127 5043–8

[172] Bethe H A 1944 Theory of diffraction by small holes Phys.Rev. 66–82 163–82

[173] Bouwkamp C J 1950 On the Bethe’s theory of diffraction bysmall holes Philips Res. Rep. 5 321–32

29

http://dx.doi.org/10.1021/jp026731y



http://dx.doi.org/10.1063/1.1812592

http://dx.doi.org/10.1063/1.2167827


http://dx.doi.org/10.1063/1.1629280

http://dx.doi.org/10.1080/01442350050034180


http://dx.doi.org/10.1063/1.123600

http://dx.doi.org/10.1063/1.1499509

http://dx.doi.org/10.1063/1.2143135

http://dx.doi.org/10.1063/1.2172733

http://dx.doi.org/10.1021/la981078k


http://dx.doi.org/10.1364/OE.15.013682

http://dx.doi.org/10.1021/la025563y

http://dx.doi.org/10.1088/1367-2630/5/1/160


http://dx.doi.org/10.1038/86762


http://dx.doi.org/10.1209/epl/i2000-00504-y

http://dx.doi.org/10.1016/S0304-3991(97)00077-6

http://dx.doi.org/10.1002/(SICI)1098-2760(19980605)18:2<120::AID-MOP10>3.0.CO;2-B

http://dx.doi.org/10.1063/1.118245

http://dx.doi.org/10.1116/1.585558

http://dx.doi.org/10.1063/1.1425083

http://dx.doi.org/10.1063/1.125766

http://dx.doi.org/10.1126/science.1111886


http://dx.doi.org/10.1038/35570


http://dx.doi.org/10.1021/ja043672o

http://dx.doi.org/10.1103/PhysRev.66.163

http://dx.doi.org/10.1021/nl0480969


[174] Bouwkamp C J 1950 On the diffraction of electromagneticwaves by small circular disks and holes Philips Res. Rep. 5401–22

[175] Grober R D, Rutherford T and Harris T D 1996 Modalapproximation for the electromagnetic field of a near-fieldoptical probe Appl. Opt. 35 3488–95

[176] Ducourtieux S, Gresillon S, Rivoal J C, Vannier C, Bainier C,Courjon D and Cory H 2004 Imaging subwavelength holesin a chromium films in scanning near-field opticalmicroscopy: comparison between experiment andcalculation Eur. Phys. J. Appl. Phys. 26 35–43

[177] Rigneault H, Capoulade J, Dintinger J, Wenger J, Bonod N,Popov E, Ebbesen T W and Lenne P-F 2005 Enhancementof single-molecule fluorescence detection insubwavelength apertures Phys. Rev. Lett. 95 117401

[178] Wenger J, Cluzet B, Dintinger J, Bonod N, Fehrembach A-L,Popov E, Lenne P-F, Ebbesen T W and Rigneault H 2007Radiative and nonradiative photokinetics alteration insidea single metallic nanometric aperture J. Phys. Chem. C111 11469–74

[179] Lezec H J, Degiron A, Devaux E, Linke R A,Martin-Moreno L, Garcia-Vidal F J and Ebbesen T W2002 Beaming light from a subwavelength apertureScience 297 820–2

[180] Beruete M, Sorolla M, Campillo I, Dolado J S,Martin-Moreno L, Bravo-Abad J and Garcia-Vidal F J2004 Enhanced millimeter-wave transmission throughsubwavelength hole arrays Opt. Lett. 29 2500–2

[181] Garcia-Vidal F J, Martin-Moreno L and Pendry J B 2005Surfaces with holes in them: new plasmonic metamaterialsJ. Opt. A: Pure Appl. Opt. 7 S97–101

[182] Qiu M 2005 Photonic band structures for surfaces waves onstructured metal surfaces Opt. Express 13 7583

[183] Lalanne P, Hugonin J P and Rodier J C 2005 Theory ofsurface plasmon generation at nanoslit apertures Phys.Rev. Lett. 95 263902

[184] Docter M W, Young I T, Piciu O M, Bossche A,Alkemade P F, van den Berg P M and Garini Y 2006Measuring the wavelength-dependent divergence oftransmission through sub-wavelength hole-arrays byspectral imaging Opt. Express 14 9477–82

[185] Popov E, Neviere M, Fehrembach A-L and Bonod N 2005Enhanced transmission of light trough a circularlystructured aperture Appl. Opt. 44 6898

[186] Van Labeke D, Baida F I and Vigoureux J-M 2004 A newstructures for enhanced transmission through a twodimensional metallic grating J. Microsc. 213 140–3

[187] Rindzevicius T, Alaverdyan Y, Dahlin A, Hook F,Sutherland D S and Kall M 2005 Plasmonic sensingcharacteristics of single nanometric holes Nano Lett.5 2335–9

[188] Chen J, Saeki F, Wiley B J, Cang H, Cobb M J, Li Z-Y, Au L,Kimmey M B, Li X and Xia Y 2005 Gold nanocages:bioconjugation and their potential use as optical imagingcontrast agents Nano Lett. 5 473–77

[189] Pham T, Jackson J B, Halas N J and Lee T R 2002Preparation and characterization of gold nanoshells coatedwith self-assembled monolyaers Langmuir 18 4915–20

[190] Prodan E, Radloff C, Halas N J and Nordlander P 2003 Ahybridization model for the plasmon response of complexnanostructures Science 302 419–22

[191] Kottmann J P and Martin O J F 2001 Plasmon resonantcoupling in metallic nanowires Opt. Express 8 655–63

[192] Li K, Stockman M I and Bergman D J 2003 Self-similarchain of metal nanospheres as an efficient nanolens Phys.Rev. Lett. 91 227402

[193] Xu H and Kall M 2002 Surface-plasmon-enhanced opticalforces in silver nanoaggregates Phys. Rev. Lett.89 246802

[194] Bek A, Klar T and Feldmann J 2007 World of PhotonicsCongress (Munich, Germany: June 2007)

[195] Nie S and Emory S R 1997 Probing single molecules andsingles nanoparticles by surface-enhanced ramanscattering Science 275 1102

[196] Xu H, Bjerneld E J, Kall M and Borjesson L 1999Spectroscopy of single hemoglobin molecules by surfaceenhanced Raman scattering Phys. Rev. Lett. 83 4357–60

[197] Cubukcu E, Fort E A, Crozier K B and Capasso F 2006Plasmonic laser antenna Appl. Phys. Lett. 89 093120

[198] Danckwerts M and Novotny L 2007 Optical frequencymixing at coupled gold nanoparticles Phys. Rev. Lett.98 026104

[199] Sondergraad T and Bozhevolnyi S 2007 Slow-plasmonresonant nanostructures: scattering and field enhancementsPhys. Rev. B 75 073402

[200] Shimitzu K T, Woo W K, Fisher B R, Eisler H J andBawendi M G 2002 Surface-enhanced emission fromsingle semiconductor nanocrystals Phys. Rev. Lett.89 117401

[201] Gersten J and Nitzan A 1980 Electromagnetic theory ofenhanced Raman scattering by molecules adsorbed onrough surfaces J. Chem. Phys. 73 3023

[202] Quinten M, Leitner A, Krenn J R and Aussenegg F R 1998Electromagnetic energy transport via linear chains ofsilver nanoparticles Opt. Lett. 23 1331–3

[203] Maier S A, Kik P G and Atwater H A 2002 Observation ofcoupled plasmon-polariton modes in au nanoparticle chainwaveguides of different lengths: estimation of waveguideloss Appl. Phys. Lett. 81 1714–16

[204] Salerno M, Krenn J R, Hohenau A, Ditlbacher H, Schider G,Leitner A and Aussenegg F R 2005 The optical near-fieldof a gold nanoparticles chains Opt. Commun. 248 543–9

[205] Ditlbacher H, Hohenau A, Wagner D, Kreibig U, Rogers M,Hofer F, Aussenegg F R and Krenn J R 2005 Silvernanowires as surface plasmon resonators Phys. Rev. Lett.95 257403

[206] Nakayama Y, Pauzauskie P J, Radenovic A, Onorato R M,Saykally R J, Liphardt J and Yang P 2007 Tunablenanowire nonlinear optical probe Nature 447 1098–101

[207] Yang Y, Nogami M, Shi J, Chen H, Ma G and Tang S 2006Controlled surface-plasmon coupling in SiO2-coated goldnanochains for tunable nonlinear optical properties Appl.Phys. Lett. 88 081110

[208] Gadenne P and Rivoal J-C 2002 Surface-plasmon-enhancednon linearities in 2d metal-dielectric films NonlinearOptics of Nanostructured Random Media (Applied PhysicsSeries vol 82) ed V M Shalaev (Berlin: Springer)pp 183–211

[209] Marton J P 1969 Optical properties of island films of metalsat long wavelengths J. Appl. Phys. 40 5383–6

[210] Cohen R W, Cody G D, Coutts M D and Abeles B 1973Optical properties of granular silver and gold films Phys.Rev. B 8 3689–701

[211] Gadenne P, Yagil Y and Deutscher G 1989 Transmittance andreflectance in situ measurements of semicontinuous goldfilms during deposition J. Appl. Phys. 66 3019–25

[212] Marton J P and Lemon J R 1971 Optical properties ofaggregated metal systems: I. Theory Phys. Rev. B 4 271–80

[213] Yagil Y and Deutscher G 1988 Scaling and renormalizationin transmittance of thin metal films near the percolationthreshold Appl. Phys. Lett. 52 373–4

[214] Yagil Y, Yosefin M, Bergman D J, Deutscher G andGadenne P 1991 Scaling theory for the optical propertiesof semicontinuous metal films Phys. Rev. B 43 11342–52

[215] Yagil Y, Gadenne P, Julien C and Deutscher G 1992 Opticalproperties of thin semicontinuous gold films over awavelength range of 2.5 to 500 µm Phys. Rev. B46 2503–11

30

http://dx.doi.org/10.1051/epjap:2004014



http://dx.doi.org/10.1364/OL.29.002500

http://dx.doi.org/10.1088/1464-4258/7/2/013



http://dx.doi.org/10.1364/OE.14.009477

http://dx.doi.org/10.1364/AO.44.006898

http://dx.doi.org/10.1111/j.1365-2818.2004.01272.x

http://dx.doi.org/10.1021/nl0516355

http://dx.doi.org/10.1021/nl047950t

http://dx.doi.org/10.1021/la015561y






http://dx.doi.org/10.1063/1.2339286




http://dx.doi.org/10.1063/1.440560

http://dx.doi.org/10.1063/1.1503870

http://dx.doi.org/10.1016/j.optcom.2004.12.023



http://dx.doi.org/10.1063/1.2172022

http://dx.doi.org/10.1063/1.1657398


http://dx.doi.org/10.1063/1.344187


http://dx.doi.org/10.1063/1.99469




[216] Bergman D J and Stroud D 1992 Physical properties ofmacroscopically inhomogeneous media Solid State Phys.46 147–269

[217] Arya K, Su Z B and Birman J L 1985 Localization of thesurface plasmon polariton caused by random roughnessand its role in surface-enhanced optical phenomena Phys.Rev. Lett. 54 1559–62

[218] Arya K, Su Z B and Birman J L 1986 Anderson localizationof electromagnetic waves in a dielectric medium ofrandomly distributed metal particles Phys. Rev. Lett.57 2725–8

[219] Anderson P W 1958 Absence of diffusion in certain randomlattices Phys. Rev. 109 1492–505

[220] Gresillon S, Aigouy L, Boccara A C, Rivoal J C, Quelin X,Desmarest C, Gadenne P, Shubin V A, Sarychev A K andShalaev V M 1999 Experimental observation of localizedoptical excitations in random metal-dielectric films Phys.Rev. Lett. 82 4520–3

[221] Bozhelvolnyi S I, Vohnsen B, Smolyaninov I I andZayats A V 1995 Direct observation of surface polaritonlocalization caused by surface roughness Opt. Commun.117 417–23

[222] Tsai D P, Kovacs J, Wang Z, Moskovits M, Shalaev V M,Suh J S and Botet R 1994 Photon scanning tunnelingmicroscopy images of optical excitations of fractal metalcolloid cluster Phys. Rev. Lett. 72 4149–52

[223] Bozhelvolnyi S N, Markel V A, Coello V, Kim W andShalaev V M 1998 Direct observation of localized dipolarexcitations on rough nanostructured surfaces Phys. Rev. B58 11441–8

[224] Zhang P, Haslett T L, Douketis C and Moskovits M 1998Mode localization in self-affine fractal interfacesobserved by near-field microscopy Phys. Rev. B57 15513–18

[225] Ducourtieux S et al 2001 Near-field optical studies ofsemicontinuous metal films Phys. Rev. B 64 165403

[226] Bozhevolnyi S I, Beermann J and Coello V 2003 Directobservation of localized second-harmonic enhancement inrandom metal nanostructures Phys. Rev. Lett. 90 197403

[227] Williame L, Lecaque R, Le Moal E, Fort E, Gresillon S,Rivoal J-C and Boccara C 2005 Linear and non-linearnear-field behavior on random metal films and surfacesPlasmonic Nanoimaging and Nanofabricationed S Kawata et al, Proc. SPIE 5928 592801

[228] Liu H, Ie Y, Yoshinobu T, Aso Y, Iwasaki H and Nishitani R2006 Plasmon-enhanced molecular fluorescence from anorganic film in a tunnel junction Appl. Phys. Lett.88 061901

[229] Enderlein J 2002 Theoretical study of single moleculefluorescence in a metallic nanocavity Appl. Phys. Lett.80 315–17

[230] Ganesh N, Zhang W, Mathias P C, Chow E, Soares J A N T,Malyarchuk V, Smith A D and Cunningham B T 2007Enhanced fluorescence emission from quantum dots on aphotonic crystal surface Nature Nanotechnol. 2 515–20

[231] Andrew P and Barnes W L 2000 Forster energy transfer in anopical microcavity Science 290 785–8

[232] Andrew P and Barnes W L 2004 Energy transfer across ametal film mediated by surface plasmon polaritons Science306 1002–5

31



http://dx.doi.org/10.1103/PhysRev.109.1492


http://dx.doi.org/10.1016/0030-4018(95)00170-D






http://dx.doi.org/10.1117/12.613884

http://dx.doi.org/10.1063/1.2171795

http://dx.doi.org/10.1063/1.1434314

http://dx.doi.org/10.1038/nnano.2007.216



surface enhanced fluorescence - turro's organic photochemistry

Documents