Surface plasmons in metallic nanoparticles.pdf

This content has been downloaded from IOPscience. Please scroll down to see the full text.

Download details:

IP Address: 140.112.5.2

This content was downloaded on 04/12/2014 at 17:26

Please note that terms and conditions apply.

Surface plasmons in metallic nanoparticles: fundamentals and applications

View the table of contents for this issue, or go to the journal homepage for more

2011 J. Phys. D: Appl. Phys. 44 283001

(http://iopscience.iop.org/0022-3727/44/28/283001)

Home Search Collections Journals About Contact us My IOPscience

iopscience.iop.org/page/termshttp://iopscience.iop.org/0022-3727/44/28http://iopscience.iop.org/0022-3727http://iopscience.iop.org/http://iopscience.iop.org/searchhttp://iopscience.iop.org/collectionshttp://iopscience.iop.org/journalshttp://iopscience.iop.org/page/aboutioppublishinghttp://iopscience.iop.org/contacthttp://iopscience.iop.org/myiopscience

IOP PUBLISHING JOURNAL OF PHYSICS D: APPLIED PHYSICS

J. Phys. D: Appl. Phys. 45 (2012) 389501 (1pp) doi:10.1088/0022-3727/45/38/389501

Corrigendum: Surface plasmons inmetallic nanoparticles: fundamentalsand applications2011 J. Phys. D: Appl. Phys. 44 283001

M A Garcia

Department of Electroceramics, Institute for Ceramic and Glass, CSIC, C/Kelsen 5, 28049 Madrid, Spainand IMDEA Nanociencia, Madrid 28049, Spain

Received 9 August 2012, in final form 13 August 2012Published 4 September 2012Online at stacks.iop.org/JPhysD/45/389501

(Some figures may appear in colour only in the online journal)

In this paper, figure 4 contains two errors. First, lithium (Li)appears twice. The right values for Li should be wavelength450 nm and extinction cross section 170 nm2. The other pointmarked as Li (480 nm wavelength , 680 nm2 cross section)corresponds to sodium (Na).

Second, the extinction cross section for silver was dividedby 20 in order to have it in the same scale as the rest. Thatshould be indicated in the figure and in the caption. Thecorrected figure 4 is displayed below.

For the case of 10 nm Ag nanoparticles, the extinctioncross section is 50 times larger than the geometrical section ofthe nanoparticle.

Acknowledgment

I thank Dr Liu and Professor Liz-Marzan for bringing myattention to these errors.

200 300 400 500 600 7000

100

200

300

400

500

600

700

Ag (20)

K

CuY

Ca

Rb

Eu

Mg

Pt

Pd

Na

Au

ext (nm

2)

Wavelength (nm)

Li

Figure 4. Resonant frequency and extinction cross section formetallic NPs with 10 nm size in air. The extinction cross section ofAg has been divided by 20 for a better presentation (the real value is4080 nm2).

0022-3727/12/389501+01$33.00 1 2012 IOP Publishing Ltd Printed in the UK & the USA

http://dx.doi.org/10.1088/0022-3727/45/38/389501http://dx.doi.org/10.1088/0022-3727/44/28/283001http://stacks.iop.org/JPhysD/45/389501

IOP PUBLISHING JOURNAL OF PHYSICS D: APPLIED PHYSICS

J. Phys. D: Appl. Phys. 44 (2011) 283001 (20pp) doi:10.1088/0022-3727/44/28/283001

TOPICAL REVIEW

Surface plasmons in metallicnanoparticles: fundamentalsand applicationsM A Garcia

Department of Electroceramics, Institute for Ceramic and Glass, CSIC, C/Kelsen 5, 28049 Madrid, Spainand IMDEA Nanociencia, Madrid 28049, Spain

E-mail: [email protected]

Received 21 February 2011, in final form 5 May 2011Published 24 June 2011Online at stacks.iop.org/JPhysD/44/283001

AbstractThe excitation of surface plasmons (SPs) in metallic nanoparticles (NPs) induces opticalproperties hardly achievable in other optical materials, yielding a wide range of applications inmany fields. This review presents an overview of SPs in metallic NPs. The concept of SPs inNPs is qualitatively described using a comparison with simple linear oscillators. Themathematical models to carry on calculations on SPs are presented as well as the mostcommon approximations. The different parameters governing the features of SPs and theireffect on the optical properties of the materials are reviewed. Finally, applications of SPs indifferent fields such as biomedicine, energy, environment protection and informationtechnology are revised.

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

1. Introduction

Surface plasmon (SP) resonance is the most outstandingoptical property of metallic nanostructures. It consists of acollective oscillation of conduction electrons excited by theelectromagnetic field of light. SP resonance is in the originof optical properties hardly achievable with other physicalprocesses [13]. In the case of metallic nanoparticles (NPs),where the electrons are confined in the three dimensions, theelectron oscillations induce an electric field around the NPthat can be much larger than the incident light one. SPsare one of the best examples that things are different at thenanoscale. When the size of a metallic particle is reducedto a few nanometres, the optical properties are dramaticallymodified by the appearance of SPs and its resulting behaviouris completely different from the bulk metal one. SPs openthe possibility to amplify, concentrate and manipulate light atthe nanoscale, overcoming the diffraction limit of traditionaloptics and increasing resolution and sensitivity of opticalprobes [4]. Consequently, SPs can be used in a wide range of

fields, including biomedical [58], energy [911], environmentprotection [1214], sensing [15] and information technology[3, 4, 16] applications. Nowadays, there are well-establishedapplications of SPs that increase rapidly with the developmentof our capabilities to fabricate and manipulate nanomaterials.

In this review the fundamentals of SPs are surveyed.It does not pretend to be a presentation of the proper andexact formalism required to describe SPs but an attempt toexplain what SPs are in an accessible way to most readerswith different background. For this purpose, analogies withsimple systems such as linear oscillators are used. Afterdescribing the physical basis of SPs, the theories that provideexact mathematical solutions and the usual approximations tocarry on proper calculations on SPs are summarized. Themain parameters affecting the SP excitation processes are alsorevised. Finally, some applications of SPs on different fieldsare surveyed. It should be impossible to review here thehuge number of applications of SPs reported in the literature.Thus, when selecting just some of them I do not pretend toneglect the researchers in the field the credit they merit but

0022-3727/11/283001+20$33.00 1 2011 IOP Publishing Ltd Printed in the UK & the USA

http://dx.doi.org/10.1088/0022-3727/44/28/283001mailto: [email protected]://stacks.iop.org/JPhysD/44/283001

J. Phys. D: Appl. Phys. 44 (2011) 283001 Topical Review

Light

E

Light

E

+ + + ++

- - - --

ERES

Figure 1. Scheme of the light interaction with a metallic NP. The electric field of the light induces the movement of conduction electronswhich accumulate at the NP surface creating an electric dipole. This charge accumulation creates an electric field opposite to that of the light.

just to highlight some applications to illustrate the capabilitiesof SPs.

The synthesis of metallic NPs in not addressed here.There are excellent reviews of this topic that the reader isreferred to [1723]. Moreover, NPs are not the only metallicnanostructures exhibiting SPs. Films, wires and patternedmedia exhibit also SPs [2, 24, 25]. In the case of NPs,the confinement of electrons leads to localized SPs, whileother structures with large dimensions compared with lightwavelength (wires and films) hold extended SPs that propagatealong the interface between the metal and the dielectricmedium. There is a third case corresponding to mesoscopicsystems, with dimensions comparable to wavelength. In thiscase SP modes can also be excited [2, 3]. We will only treathere the case of localized SPs in NPs.

2. Fundamentals of SPs in NPs

The scientific study of SPs started in the early 20th centurywhen Gustav Mie published his pioneering work explainingthe surprising optical properties of metallic colloids [26].However, SPs in NPs have been empirically used for a longtime, particularly for glass colouring. Glasses from the latebronze age (10001200 BC) found in Fratessina di Rovigo(Northern Italy) were coloured by the presence of metallic CuNPs [27]. The Lycurgus cup (400 AC) is probably the mostfamous example of the use of SPs in ancient times, exhibitingdifferent coloration when observed upon illumination insideor outside of the cup [28, 29]. Actually many Roman mosaicsused metallic NPs to achieve red coloration [30, 31]. At thattime, the Mayan civilization also used metallic NPs to developblue paints exploding SPs [32]. There is also evidence of theancient use of metallic NPs in glass and pottery from Egyptiandynasties [33], Celtic enamels [34], Japanese [35] and Chineseglasses [36].

In the Middle Ages, the development of glassmanufacturing processes induced outstanding advances instained glass fabrication [37, 38]. The variety of metallic NPsfor glass colouring increased substantially as we can observein many church windows from that period. The developmentof glass chemistry during the Renaissance [39] and modernage [40] provided better tuning of coloration effects based onthe SPs of metallic NPs. Excellent revisions of the historicaluse of SPs in metallic NPs for glass and pottery colouring arepresented in [31, 41].

All these developments were empirically achieved withoutknowing the real origin of the surprising optical effects

appearing when metallic atoms were introduced in the glass.After Mies work in the early 20th century, the origin of theoptical properties of metallic NPs was understood, but furtherexploitation was limited by the capabilities to synthesize andmanipulate NPs in a controlled way. It was not until thedevelopment of nanotechnology at the end of 20th century thatthe applications of SPs spread quickly in many fields.

2.1. SPs in NPs

As indicated above, SPs correspond to an interaction betweenmatter and the electromagnetic field of the light. Thus, theexact analysis of SPs implies solving the Maxwell equationswith the appropriate boundary conditions [1]. The solutionof these equations is only possible for certain conditions andeven in this case, the results are mathematical series that donot explain what are SPs. However, a simplified classicalpicture can be more useful to understand the physical meaningof SPs. A metallic NP can be described as a lattice of ioniccores with conduction electron moving almost freely inside theNP (the Fermi sea) as figure 1 illustrates. When the particle isilluminated, the electromagnetic field of the light exerts a forceon these conduction electrons moving them towards the NPsurface. As these electrons are confined inside the NP, negativecharge will be accumulated on one side and positive chargein the opposite one, creating an electric dipole. This dipolegenerates an electric field inside the NP opposite to that of thelight that will force the electrons to return to the equilibriumposition (figure 1). The larger the electron displacement, thelarger the electric dipole and consequently the restoring force.The situation is similar to a linear oscillator with a restoringforce proportional to the displacement from the equilibriumposition. If the electrons are displaced from the equilibriumposition and the field is removed later, they will oscillate witha certain frequency that is called the resonant frequency; inthe case of SPs it is named the plasmonic frequency. Actually,the electron movement inside the NP exhibits some degree ofdamping. The ionic cores and the NP surface partially dampthe electron oscillations. Thus, the system is similar to a linearoscillator with some damping.

When an alternating force is applied to a linear oscillator,the system oscillates with the same frequency as the externalforce but the amplitude and phase will depend on boththe force and the intrinsic parameters of the oscillator. Inparticular, the oscillating amplitude will be maxima for theresonant frequency (figure 2(a)). It is quite straightforward tounderstand that, if the frequency of the external force is the

2


Am

plit

ud

e

FrequencyR

350 400 450 500 550Ab

so

rption

(a

rb. u

nits)

Wavelength (nm)

a

Am

plit

ud

e

Frequency

R 350 400 450 500 550

Ab

so

rption

(a

rb. u

nits)

Wavelength (nm)

a(a) (b)

Figure 2. (a) Oscillation amplitude for a linear oscillator as a function of the external force frequency. (b) Optical absorption spectrumcorresponding to 10 nm silver NPs embedded in a silica glass.

Figure 3. Left: illustration of absorption cross section concept. Right: picture describing transmission, absorption and scattering processes.

same as the plasmonic frequency of the NP, it will be easy tomake the electrons oscillate, but as we move far way from thisfrequency the movement of electrons will be more difficult,i.e. with reduced amplitude.

We cannot directly observe the movement of electronsto determine their oscillating amplitude. However, we candetermine this amplitude indirectly. The electronic oscillationimplies an increase in kinetic and electrostatic energiesassociated with the electric fields of the dipole. As energy mustbe conserved, this increase in energy must be provided by theilluminating light. Therefore, the light extinguishes partiallywhen exciting SPs inside the NP. The larger the electronoscillations, the larger the light extinction, so the opticalabsorption spectrum allows one to detect the excitation of SPs.The resonant frequency for these oscillations in metallic NPscorresponds typically to UVVis light and consequently, theSPs arise absorption bands in this region of the spectrum asfigure 2(b) illustrates [168].

At this stage, SPs can be considered as another electronicprocess in which light is absorbed to promote electrons fromthe ground level to an excited one. What makes the SPs uniqueare the numbers of these processes. The absorbing efficiency ofa particle is given by its absorption cross section. Classicallyit corresponds to the geometrical section of an ideal opaqueparticle absorbing the same number of photons as the studiedparticle. As figure 3 illustrates, we could replace the absorbingNP by a perfect opaque one (absorbing any photon reachingits surface) that will absorb the same number of photons as our

real particle. The section of this ideal particle represents theabsorption cross section of the NP. For instance, if we havean NP absorbing half of the photons reaching its surface, theabsorption cross section will be half of its geometrical section.In addition to absorption, light interacting with matter can bescattered, changing the propagation direction and eventuallyalso energy and moment. For this process, we can define alsothe scattering cross section as the geometrical section of anideal scattering particle (that scatters any photon reaching itssurface) with the same scattering efficiency as the real particle.

The sum of absorption and scattering cross section isdefined as the extinction cross section that represents theefficiency of the particle to remove photons from an incidentbeam (by both absorption or scattering processes). Themaximum possible value of the extinction cross-section forperfect opaque particles is the particle section ( R2).However, for a few nm NPs, extinction cross sections largerthan 10% of the geometrical section are scarcely found forother processes different from SP excitation [42, 43].

Figure 4 presents the resonant wavelength and theextinction cross section for 10 nm size metallic NPs. It isfound that for noble metal NPs the extinction cross sectioncan be up to 10 times their geometrical section; that is, theNP is capable of absorbing and scattering photons even awayfrom its physical position. Somehow, the excitation of the SP isequivalent to concentrate the light passing by the NP to inducea huge extinction. It is worth noting that the light absorptionhas an exponential dependence on the absorption cross section.

3


200 300 400 500 600 7000

100

200

300

400

500

600

700

Ag

K

CuY

Ca

Rb

Eu

Mg

Pt

Pd

Li

Au

ext (

n

m2)

Wavelength (nm)

Li

Figure 4. Resonant frequency and extinction cross section formetallic NPs with 10 nm size in air.

A light beam propagating across a medium with metallic NPsdecay in intensity as

I (x) = I0 eC x (1)I0 being the initial intensity, C the concentration of NPs perunit volume, their extinction cross section and x the travelleddistance. Therefore, a moderate increase in the extinction crosssection can lead to a huge enhancement of light absorption.As an example, if we have a dielectric matrix containing1020 NPs cm3 with 10 nm size (1% in volume) and theirextinction cross section is equal to the geometrical section( R2), the light transmitted across 1m of the materialwill be 45%; if the extinction cross section of the NPs isincreased to a factor of 10, the transmitted light is reduced to0.04% (i.e. 3 orders of magnitude). These numbers make SPsunique, since the cross section values found here are largerthan other optical processes such as electronic transitions insemiconductors (band to band transitions), rare earths atoms,defect related absorption processes in solids, and electronicexcitations in molecules.

A more direct example of the huge extinction in metallicNPs due to SP excitation is provided by a comparison withinterband transitions. In addition to SPs, there are otherpossible electronic excitations in metallic NPs. In a metallicmaterial, valence and conduction bands overlap forming acontinuous spectrum of available states. However, someinner levels do not split enough to overlap these bands sothe system may exhibit interband transitions similar to thosein semiconductors [44]. Transitions between these innerlevels and the conduction band induce an absorption edgesimilarly to the case of semiconductors. Actually, some metalspresent a weak luminescence emission due to electron decaybetween these bands. For bulk materials these transitions arevery unlikely and optical absorption and emission associatedwith these transitions are very weak. For instance, a weakphotoluminescence associated with interband transitions hasbeen measured for bulk gold [45] corresponding to transitionsbetween the 3d level and the conduction band. However, forNPs with a reduced number of atoms, energy bands are notso well formed because of the limited number of atoms andinterband transitions become more prominent.

Figure 5 shows the optical absorption spectrum of Agand Au NPs. In the case of Ag, both SP band and interbandtransition absorption edges are well resolved. For Au NPs,both absorptions overlap. This overlapping is important whenanalysing the shape of the SP. In particular, the width of the SPband (related to the NP size) must be determined separating thecontribution of the interband transitions. We may observe howthe SP absorption is larger than that of the interband transitionsdespite the fact that the absorbing centre is the same for bothprocesses.

The reason for this surprising behaviour can bequalitatively explained with the same classical picture ofSPs. When the incident light reaches the NP, the conductionelectrons move resulting in a charge accumulation at theNP surface. This charge creates a field inside the NP (therestoring field) but also out of the NP. The large electrondensity and mobility in the metallic NP produces a large chargeaccumulation at the NP surface and consequently intense fieldsin a region larger than the NP size. As figure 6 illustrates, thereare wide regions where the electric field created by the particleis opposite to that of the light, so the interference is destructive,leading to light extinction beyond the NP volume [46]. Forother regions, the result of the interference between bothelectric fields is a net field with other propagating direction,hence inducing light scattering. This mechanism explainsqualitatively the huge extinction cross section of metallic NPswhen SPs are excited. Actually, the excitation of SPs withlaser sources can even yield a modification of the NP shapeby local increase in the temperature associated with the hugeabsorption [4750].

2.2. Models and calculations of SPs

While the above classical description can be useful to illustrateand understand what are SPs, it is too simple to carryon numerical analyses. An accurate calculation of theSP and the associated light absorption requires solving theMaxwell equations at the NP region using the proper boundaryconditions. The analytic solution can be obtained only forcertain geometries and it was developed by Gustav Mie in theearly 1900s [1, 26]. The Mie theory provides an exact solutionfor spherical NPs assuming that they are non-interacting (i.e.the distance between the NPs is large enough so we may assumethat the electric field created by an NP does not affect therest of them). In this case, Maxwell equations can be solvedanalytically and the extinction cross section is given by

ex = 2|k2|L=1

(2L + 1) Re[aL + bL] (2)

where

aL = mL(mx) L(x) L(mx) L(x)

mL(mx) L(x) L(mx) L(x);

bL = L(mx) L(x)m L(mx) L(x)

L(mx) L(x)m L(mx) L(x)where k is the light wavevector in the dielectric medium,x = |k| R, R is the NP radius, m = n/nm, n and nm being

4


200 300 400 500 6000

10

20

30

40

Ab

so

rptio

n (

%)

Wavelength (nm)

Ag NPs

40 nm diameterSP band

Interband

transitions

400 500 600 700 8000

10

20

30

40

50

Ab

so

rptio

n (

%)

Wavelength (nm)

Au NPs

40 nm diameter SP band

Interband

transitions

200 300 400 500 6000

10

20

30

40

Ab

so

rptio

n (

%)

Wavelength (nm)

Ag NPs

40 nm diameterSP band

Interband

transitions

400 500 600 700 8000

10

20

30

40

50

Ab

so

rptio

n (

%)

Wavelength (nm)

Au NPs

40 nm diameter SP band

Interband

transitions(a) (b)

Figure 5. Optical absorption spectra for (a) Ag and (b) Au NPs with 40 nm size (embedded in a silica matrix with = 2.25). For Au NPs,the contributions to the optical absorption of interband transitions and SPs are resolved.

ELIGHTENP

Light

Figure 6. Illustration of electric fields of incident light and thatcreated by the electron oscillations near the NP.

the refraction index of the metal (complex) and that of thesurrounding dielectric medium, respectively, and L and Lare the cylindrical BesselRiccati functions. These equationsare not easy to handle, although there are codes available tocarry on calculations.

There is a large list of theories that allow calculating thelight absorption associated with SPs for specific size rangesor when NPs are not spherical or they are interacting. Anexcellent review of these theories is given in [1]. We surveyhere just the most common ones.

The dipolar approximation. The dipolar approximationoffers simple equations to calculate light absorption of metallicNPs provided that the NP size is smaller than the lightwavelength (in addition to the hypothesis of the Mie theory).As the SP resonance band falls at the visible part of thespectrum (360720 nm), the condition of small NPs isusually achieved for NP diameter below 50 nm. In thiscase, the electric field inside the NP can be considered asuniform (see figure 7) and the particle can be describedby an electric dipole. Mathematically, this approximationis achieved assuming that x is small enough, hence, inequation (2) we can just consider the first term L = 1. (Notethat the term L = 0 corresponding to the total electric chargeof the NPs is zero).

+++

+ ++

---

---

- -

++

Light

Light

Figure 7. Electric field and charge distribution at the surface of NPsfor size (top) much smaller and (bottom) comparable to the lightwavelength.

Within this approximation, the absorption cross section isgiven by

ext = 242R3

3/2m

2

(1 + 2m)2 + 22(3)

being the light wavelength, m the real dielectric function ofthe medium and 1 + i2 the complex dielectric function of themetal. As the size of NPs increases, the approximation is notvalid anymore, since the electric field inside the NP cannot beconsidered uniform. In that case, more multipolar terms (i.e.more terms in equation (2)) must be considered as discussedbelow.

The effective dielectric function theories. Another approachin calculating the optical properties of metallic NPs consistsof replacing the heterogeneous system formed by the NPs andthe surrounding media by a homogeneous medium with aneffective dielectric function (eff1 + ieff2). These theories areknown as effective medium theories. There are a large numberof these theories depending on the exact conditions that areimposed on the homogeneous medium [1]. The first approachwas proposed by Newton [1, 51] defining an effective dielectric

5


function as an average of the NPs and medium dielectricfunction weighted by their volume.

However, the best approach within the framework ofeffective medium theories was developed by Maxwell-Garnett[5255] proposing to replace the NPs dielectric mediumsystem by an homogeneous material exhibiting the samedielectric polarization upon light illumination. With thiscondition, the complex effective dielectric function is given by

eff meff + 2m

= f m + 2m

(4)

and the extinction cross section of the material is then given by

(cm1) = 8.88 107

(nm)

1eff +

21eff +

22eff . (5)

For the case of spherical, small and isolated NPs, this modelleads to the same result as the dipolar approximation. Aninteresting advantage of the Maxwell-Garnett model is that itcan be modified to take into account interparticle interactions[56, 57] or the presence of non-spherical NPs [1, 6, 58].

2.3. Size effects

The size of NPs has a dramatic effect on the SP resonanceprocesses and consequently on the optical properties of theNPs. Actually, there are different mechanisms inducing sizedependence of the SP properties. The predominant one willdepend on the NP size range we consider. As a generalapproach, we may distinguish two regimes corresponding tosmall NPs (smaller than the light wavelength) and large oneswith size comparable to the wavelength.

Small NPs (radius up to 50 nm). For this size range, wemay assume that the NP is properly described by a dielectricdipole. The size dependence of the SP affects mainly the widthand the intensity of the resonance band while the effect on theresonance wavelength is quite reduced [59]. Dealing with sizeeffects in this range, we can distinguish between intrinsic andextrinsic size effects.

The intrinsic effects are related to the damping of theelectron oscillations. When SPs are excited, the electrons aredamped in their movement by the scattering with the ionic coresand with the surface. The damping constant for the electronoscillations is given by

= 0 + AvFR. (6)

The first term (0) describes the damping due the scatteringof the oscillating electrons with the ionic cores. It is sizeindependent and its value is given by 0 = vF/l, vF being thevelocity of the conduction electrons (Fermi velocity) and lthe electron mean free path in the metal. This term depends juston the nature of the metal and the crystal structure. For Au at300 K we havevF = 1.4106 m s1 and l = 30 nm leading to0 = 4.61013 s1 and for Ag we have vF = 1.39106 m s1and l = 52 nm leading to 0 = 2.6 1013 s1.

The second term corresponds to the scattering ofoscillating electrons with the particle surface. A is a material-dependent constant that takes into account the features ofthe surface scattering. It is commonly considered as aphenomenological parameter although it is well establishedthat it depends on the surface features such as crystal quality,facets, capping species, strain, etc [1, 59]. The electronsin a certain shell close to the NP surface will scatter thissurface when they oscillate. The larger the velocity of theoscillating electrons (which is about the Fermi velocity), thelarger the number of electrons scattering the surface duringtheir movement (i.e. the thickness of the shell of electronsscattering the surface during oscillations). As the particle sizeincreases, the fraction of electrons in the shell close to thesurface decreases and therefore the total damping is reduced.Therefore, the surface damping is proportional to the Fermivelocity in the metal and inversely proportional to the particleradius.

It is well known that the resonance of a linear oscillatorreduces its intensity and increases the width with the dampingconstant, while the effect on the resonance frequency is verylimited (a slight shift towards smaller frequencies, i.e. largerwavelengths). Experimentally, it is observed that the full-width-at-half-maximum (FWHM) of the SP resonance bandis given by

= a + bR. (7)

Figure 8 presents the optical absorption spectra of silver NPsembedded in silica (n = 1.5) with different sizes. Theresonance absorption intensity increases with particle sizewhile the resonance peak slightly moves. The dependenceof the FWHM of the resonance peak with the particle size,shown in figure 8, follows the inversely proportional law ofequation (7) for radius below 25 nm (inset in figure 8).

In addition to this intrinsic effect of the size of NPs onthe SP, it has been reported that the dielectric function ofsmall NPs is also size dependent [1, 60]. At this scale, theenergy bands are not so well defined as in a bulk solid, andconsequently contribution of the interband transitions to thedielectric function becomes size dependent. Accordingly thereis an additional dependence of the SP on the size of NPsthrough the dependence of its dielectric function. This effect isalso appreciable for NPs below 5 nm size and the value of thedielectric function can be calculated from the bulk one [60].

Large NPs (radius larger than 50 nm). These NPs cannotbe considered much smaller than the light wavelength. Thus,the particle is no more described by a dipole (figure 7) andfurther multipolar terms are required. As a consequence,the resonance band splits into several peaks: two peaks forquadrupole, three peaks for an octopole, etc [59]. Figure 9shows the contribution of the different terms in equation (2) tothe extinction cross section of Au NPs. For Au NP with radiusbelow 60 nm, the quadrupolar term (L = 2) is negligible andthe multipolar effects must be taken into account.

If the NP size is further increased, NPs cannot beconsidered anymore as multipoles and the SPs become

6


350 400 450 5000.000

0.005

0.010

0.015

0.020

10 nm

20 nm

5 nm

Ab

so

rba

nce

(a

rb. u

nits)

Wavelength (nm)

2 nm

Ag NPs

nm=1.5

1/R (nm-1)

0,25 0,50 0,75 1,00

FW

HM

(eV

)

0,00

0,25

0,50

0,75

0 5 10 15 20 25 3000

0,25

0,50

0,75

0.25

0.50

0.75

0.25 0.50 0.75 1

0.25

0.50

0.75

1/R (nm-1)

0,25 0,50 0,75 1,00

FW

HM

(eV

)

0,00

0,25

0,50

0,75

0 5 10 15 20 25 300.

0,25

0,

0,75

0.25

0.50

0.75

0.25 0.50 0.75 1

0.25

0.50

0.75

1/R (nm-1)

0,25 0,50 0,75 1,00

FW

HM

(eV

)

0,00

0,25

0,50

0,75

0 5 10 15 20 25 30

0,25

0,

0,75

0.25

0.50

0.75

0.25 0.50 0.75 1

0.25

0.50

0.75

350 400 450 5000.000

0.005

0.010

0.015

0.020

10 nm

20 nm

5 nm

Ab

so

rba

nce

(a

rb. u

nits)

Wavelength (nm)

2 nm

Ag NPs

nm=1.5

1/R (nm-1)

0,25 0,50 0,75 1,00

FW

HM

(eV

)

0,00

0,25

0,50

0,75

0 5 10 15 20 25 30

0,25

0,

0,75

0.25

0.50

0.75

0.25 0.50 0.75 1

0.25

0.50

0.75

1/R (nm-1)

0,25 0,50 0,75 1,00

FW

HM

(eV

)

0,00

0,25

0,50

0,75

0 5 10 15 20 25 30

0,25

0,

0,75

0.25

0.50

0.75

0.25 0.50 0.75 1

0.25

0.50

0.75

1/R (nm-1)

0,25 0,50 0,75 1,00

FW

HM

(eV

)

0,00

0,25

0,50

0,75

0 5 10 15 20 25 300.

0,25

0,

0,75

NP Radius (nm)

0.25

0.50

0.75

FW

HM

(eV

)

0.25 0.50 0.75 1

0.25

0.50

0.75(a) (b)

Figure 8. (a) Optical absorption spectrum for Ag NPs with different sizes calculated according to the Mie theory. (b) FWHM as a functionof the particle radius; inset shows the linear relationship between FWHM and the inverse of the radius.

400 500 600 700 800

L=1

400 500 600 700 800

L=2

400 500 600 700 800

L=3

400 500 600 700 800

L=4

400 500 600 700 800

R=100 nm

R=60 nm L=1 L=2 L=3 L=4

R=40 nm L=1 L=2 L=3 L=4

R=80 nm L=2L=1 L=3 L=4

Wavelength (nm)

Ab

so

rptio

n C

oe

ff.

Figure 9. Optical absorption spectrum for Au NP with different sizes calculated according to the Mie theory for nm = 1.5 and thecontribution of the different multipolar terms.

propagating waves with well-defined modes or dispersionrelations [2]. These are named itinerant SPs and representa quite different kind of phenomenology [3].

Size dispersion. Although synthesis methods for metallicNPs improved substantially in the last years, the result is alwaysa set of NPs with a certain size dispersion. Consequently,the experimentally measured optical absorption spectrum of amaterial containing metallic NPs will be a weighted averageof the absorption spectra corresponding to the NPs present inthe sample.

The size dispersion induces a broadening of the absorptionband. As figure 10 illustrates, the larger the size dispersion,

the wider the absorption band. In particular, we found that fora set of Ag NPs with an average size of 10 nm and a dispersion4.2 nm, the FWHM of the SP absorption band is three timessmaller than that for 10 nm Ag NPs with a size dispersion8.5 nm. Thus, it is not possible to determine the size of theNPs just from the FWHM of the absorption band, as it dependson two parameters: average size and size distribution width.

2.4. Shape effects

The resonance of SPs is strongly affected by the particleshape. Since the restoring force for SPs is related to thecharge accumulated at the particle surface, it will be influenced

7


Radius (nm)

0 3 6 9 12 15

Radius (nm)

0 5 10 15 20 25

b

a

300 400 500 600

0.0

0.4

0.8

Ab

so

rptio

n c

oe

ffic

ien

t (a

rb.

un

its)

Wavelength (nm)

a

b

Figure 10. (Left) size distributions and (right) optical absorption spectra for Ag NPs both with 10 nm average size but different sizedispersions calculated according to the Mie theory and using nm = 1.5.

+++

- - -

++

++

++

++

++

+

--

--

--

--

--

-

0 1 2 3 4 5 62

4

6

8

10

12

Aspect Ratio

Re

so

na

nce

In

ten

sity (

arb

. u

nits)

600

800

1000

1200

Re

so

na

nce

Pe

ak (n

m)

(a)(b) (c)

400 600 800 1000 1200 1400

45

21

Ab

so

rptio

n C

oe

f. (a

rb. u

nits)

Wavelength (nm)

3

Figure 11. (a) Illustration of charge accumulation for longitudinal and transversal SPs. (b) Calculated absorption spectra for Au nanorodswith different aspect ratios indicated in the figure. (c) Resonant position and intensity of the longitudinal SP as a function of the nanorodaspect ratio.

by the particle geometry. The most clear example of theseshape effects are the nanorods [6, 7]. As figure 11 illustrates,for a nanorod, the charge accumulation at the NP surfacewill be different for electron oscillations along the rod axis(longitudinal plasmons) and along a perpendicular direction(transversal plasmons). In particular, the charge accumulationwill be maximum for the latter and (minimum for electrondisplacement along the rod axis). The restoring force isproportional to this charge accumulation, and therefore, forelectrons oscillating along the rod axis we should expectsmaller forces and consequently smaller resonance frequencies(i.e. larger resonant wavelengths). For nanorods, the resonantfrequency of transversal plasmons falls at about the sameposition as for spherical NPs (actually, at wavelengths slightlysmaller) while the resonance of longitudinal plasmons shiftstowards larger wavelengths when the nanorod aspect ratioincreases, as illustrated in figure 11(b). This behaviourprovides a method to tune the resonance of the SP at thedesired wavelength by controlling the aspect ratio, whichis especially useful for biomedical purposes as explained insection 3.1.

Using polarized light it is possible to excite separatelythe longitudinal or transversal SP of a nanorod. However,

if the rods are not macroscopically oriented (as it is thecase for colloidal suspensions or solutions) we will alwaysobserve an average over the possible orientation with bothlongitudinal and transversal resonance bands in the spectrumas in figure 11(b).

Other geometries as triangular prisms, nanocubes ornanocages give rise to more complicated effects, but in general[6], deviation from sphericity shifts the resonance towardslarger wavelengths.

Calculations for non-spherical NPs are rather compli-cated. Actually the Maxwell equations can be solved ana-lytically only for spherical nanoshells or ellipsoids. The Ganstheory [1, 6, 58] provides a method to calculate the SP spectraof non-spherical NPs within the dipolar approximation frame-work by approximating the nanostructure to an ellipsoid.

For other geometries, calculation of the SP spectra canbe addressed by the so-called discrete dipole approximation[6, 61]. Within this method, the nanostructure is replaced byan ensemble of nanospheres smaller than the nanostructure.Each nanosphere is described by a polarization factor (that canbe derived from Maxwell equations since they are assumed tobe spherical) and the field at the nanosphere position is thatof the incident light plus the field created by the rest of the

8


450 500 550 600 650 700 750 800

Op

tica

l a

bso

rptio

n (

arb

. u

nits)

Wavelength (nm)

1

1.5

2

2.5

3

3.5

Figure 12. Optical absorption spectra for Au NPs with 10 nm sizein a dielectric medium with different dielectric functions calculatedaccording to the Mie theory.

NPs. With this method, any geometry can in principle beaddressed.

2.5. Surrounding media

The excitation of SPs does not depend only on the featuresof NPs but it is also significantly modified by the surroundingmedium. It has two clear effects on the SP excitation process.On the one hand, the dielectric function of the surroundingmedium determines the light wavelength at the vicinity ofthe NP hence altering the geometry of the electric field at thesurface of NPs.

But the most important effect is related to the polarizationof the medium. During SP excitation, the charge accumulationcreates an electric field in the vicinity of the NPs (in addition tothat of the incident light). This field induces the polarization ofthe dielectric medium, resulting in a charge accumulation at theedges of the medium (i.e. at the interface between the dielectricand the metallic NPs) that will partially compensate the chargeaccumulation due to the movement of conduction electrons inthe NP. This reduction of charge will depend on the dielectricfunction of the media; the larger m, the larger the polarizationcharge, and hence, the larger the effect on the SP. Reducingthe net charge at the NP surface implies a reduction in therestoring force. For an oscillator, it is well known that reducingthe restoring force leads to smaller resonant frequency. Thus,increasing the dielectric constant of the surrounding media willshift the SP resonant band towards larger wavelengths (i.e.smaller frequencies).

Figure 12 shows the absorption spectra for 10 nm size AuNPs embedded in media with different dielectric functions.The surrounding medium has a limited effect on the bandwidth(measured in eV), that is mainly related to the particle size.The resonance intensity also increases with the dielectricfunction of the medium. This is the expected behaviour fora forced oscillator: when the resonance frequency decreases,the amplitude at resonance increases. This is mathematicallyreflected in equation (3), where the absorption cross section isproportional to 3/2m .

400 600 800

Ext. C

oeff. (a

rb. units)

Wavelength (nm)400 600 800

K=100

K=0

Extin

ctio

n c

oe

ff. (a

rb. u

nits)

Wavelength (nm)

K=50

(a) (b)

Ag NPs

R=20 nm

Figure 13. (a) Optical absorption spectra for Au NPs with dipolarinteractions. Parameter K represents the intensity of the electricfield created at a NP position by the rest of the NPs as describedin [57]. (b) Optical absorption spectrum of arrays of 20 nm Ag NPswith different geometries. Adapted from [1].

NPs obtained by chemical routes are commonly cappedwith organic molecules to prevent agglomeration. Thesecapping agents have a significant effect on the SP features sincethey represent the very first environment of the NP. Moreover,the strength of the bonds between the NP and the moleculesinduces an extra effect [1, 62, 63]. The modification of theNP electronic configuration due to surface bonds modifies theconditions for electron oscillation, yielding to variations in theSP spectrum that can be used to investigate the characteristicsof these bonds [64].

2.6. NPs interactions effects

As detailed above (see figure 6), the excitation of SP inducesan electric field in the vicinity of the NPs. If the NPsare close enough, the net applied field at the NPs positionswill be that of the incident light plus the field created bythe rest of NPs. Therefore, the resonant conditions willbe modified. In general, interaction of interparticles red-shifts the resonance band increasing the FWHM, as illustratedin figure 13(a) [56, 57, 65, 66]. When NPs are arrangedin particular geometries it is possible to create controlledinterferences that allow additional tuning of the SP excitationband (figure 13(b)). These effects can be achieved when theNPs are created by film deposition and patterning or by self-assembly processes.

The field created by an NP quickly decays with distanceand therefore, the main effect is due to adjacent NPswhile those far away have a very weak effect (note thatfor a dipole, the electric field decreases as 1/r2). Dueto this fast decay, the electric field created at an NPposition by other NPs close enough will not be uniformbut it will exhibit significant amplitude variations in theNP volume. As SP excitation is a resonant process, it isrequired that all the electrons receive the same electric field;otherwise the movement of electrons will be heterogeneousleading to charge accumulation inside the NP volume thatraise electric forces, resulting in an extra damping of theoscillations.

9


3. Applications

The high absorption coefficient associated with the excitationof SPs opens a wide range of potential applications of metallicNPs in many fields. The possibility to concentrate, amplifyand manipulate light at the nanoscale through SPs provides amethod to improve optical effects or activate processes in acontrolled way. There are hundreds of potential applicationsof SPs in different fields that have been thoroughly reportedin the literature which are impossible to review here. We willjust highlight some of them in order to illustrate capabilities ofSPs in four selected fields: biomedicine, energy, environmentprotection and information technologies.

3.1. Biomedicine

Biomedical applications are certainly the most developed SPpotential uses. In addition to the possibility to explode SPs,noble metal NPs exhibit additional features that make themespecially useful for this field [6, 67, 68]. NPs have thebiological size, comparable to entities such as viruses, DNAchains, cells and bacteria. Hence, it is possible for the NPs tointeract individually with these organisms increasing efficiencyand specificity of medical treatments. Moreover, noblemetal NPs are highly biocompatible and easy to functionalize(mainly through thiol chains that create strong bonds at thesurface of NPs and may act as linkers with larger organicmolecules). Thus, in vitro applications of metallic NPsexploding SPs are well established while in vivo ones (alwaysslower to develop due to the requirements of long term effectessays) show certainly promising results.

DNA sequencing. Noble metal NPs are currently used forDNA sequencing, replacing dye-based technology [5, 69]. Theuse of dyes causes some problems such as photobleaching andinstability in certain environments that limit their applications[5, 70]. Moreover, the spatial resolution and sensitivityrequired to carry out thousands of test compel us to use confocalmicroscopy leading to complex and expensive devices [71].The use of metallic NPs may overcome these limitations asnoble metals are highly stable and the huge absorption crosssection associated with SP resonance allows the easy detectionof small quantities of NPs.

Figure 14 illustrates the strategy for the developmentof DNA sequencers. The target DNA chain to beanalysed is attached to a metallic NP (gold NPs are themost commonly used ones) creating a NP/DNA complex(figure 14(a)) by polymerase chain reaction or throughthiol chains [5, 72]. A sensing surface is functionalizedwith patterns of oligonucleotides (short nucleic acid chain,typically with fifty or fewer bases) to be used as probes(figure 14(b)). These oligonucleotides are able to form self-assembled monolayers prompting the formation of geometricaldense patterns. Finally, the solution containing the targetNP/DNA complex is placed over the sensing surface. DNAstrands spontaneously bind to oligonucleotides containing thecomplementary chain [73]. Therefore, the target DNA/NPcomplexes will accumulate over the surface functionalizedwith oligonucleotides with chains complementary to the target

Figure 14. Scheme of the DNA labelling process with NPs.(a) Target DNA chains are attached to metallic NPs. (b) A sensingsurface is prepared with regions of oligonucleotides with specificsequences to be detected. (c) When the solution containingDNA/NPs complexes is placed on the sensing surface, thesecomplexes will bind to the oligonucleotides with the sequencescomplementary to the target DNA; the SP allows the spatialdetection of the NPs identifying the sequences present in thetarget DNA.

DNA one (figure 14(c)). Illuminating the sensing surface withlight at the plasmonic frequency of the gold NPs, it is possibleto detect the regions where the DNA/NPs complexes areaccumulated. The large scattering due to SP excitation allowsdetecting small amounts of NPs in small regions using anoptical microscope (figure 15). Sensing surfaces with a largenumber of oligonucleotides can be automatically analysedwith a digital camera, allowing a quick identification of thesequences present in the target DNA [5], as illustrated infigure 15.

Cell labelling. Following a very similar strategy, it is possibleto use SPs for cell labelling. For this purpose, NPs arefunctionalized with biomolecules exhibiting an affinity forcertain species and/or particular cell features and introducedinto cell cultures. This technology is particularly usefulto identify cancer cells [7478]. NPs functionalized withmolecules having higher affinity for these cancer cells thanfor normal ones will concentrate around these cancer cells thatcan be then detected by optical microscopy.

Au NPs can be used to label malignant epithelial cellsas demonstrated by El-Sayed et al [77]. Non-functionalizedNPs tend to accumulate within the cell cytoplasm providinganatomic labelling information of the cell. The NPs exhibit thesame affinity for malignant and non-malignant cells. However,if the NPs are functionalized with monoclonal anti-epidermalgrowth factor receptor (anti-EGFR) antibodies, the affinity formalignant cells is 6 times larger than for non-malignant ones.The accumulation of NPs is detectable with optical microscopyat SP wavelength due to the high scattering cross section,

10


Figure 15. Left: optical reflection image of NP-labelled DNA chip; the regions with NPs appear brighter due to the reflection of incidentlight. Right: close-up of one square by scanning force microscopy reveals individual NPs. Reproduced with permission from [5].

Figure 16. Light scattering images of HaCaT non-cancerous cells (left), HOC cancerous cells (middle), and HSC cancerous cells (right)after incubation with anti-EGFR antibody conjugated gold NPs. The conjugated NPs bind specifically to the surface of the cancer cells(right). Reproduced with permission from [77].

providing a method to identify malignant cells specifically(figure 16). In addition, a spectroscopic study has shown thatthe SP absorption band for the NPs accumulated at the surfaceof malignant cells is sharper and red shifted [77], which may bean additional method for cell labelling using smaller quantitiesof NPs and more reliable than intensity contrast.

Similar works have demonstrated selective labelling usingAu NPs functionalized for ductar carcinoma cells (ATCC: Hs578T) [76], human epidermal growth factor receptor 2 (HER2)[79], cervical epithelial cancer cells (SiHa) [80] human oralsquamous carcinoma cells (HOC) [81] and others [82, 83].Outstanding results have also been achieved using combinedtechniques in which the excitation of SPs is used to amplify ortrigger other optical processes such as photoacoustic imaging[84] SERS [85, 86] or dye fluorescence [87].

In vivo labelling and imaging. Development of in vivoimaging and labelling techniques are quite limited becausethe absorption band of SPR is almost identical to that ofhaemoglobin, the main protein of human blood as shown infigure 17. Therefore, it is very difficult to detect or energizegold NPs in the bloodstream.

The strategy to overcome this limitation is the use ofnon-spherical NPs such as nanorods, nanocubes, nanocages[6, 71, 78, 88]. Among these geometries, nanorods are the mostcommonly used ones since controlling their aspect ratio it ispossible to tune the SP resonance position to the biological

500 600 700 800 9000.0

0.5

1.0

1.5

2.0

2.5

Nanoparticles

Nanorods

Blood

Coef. A

bs. (c

m-1)

Wavelength (nm)

10 nm

Figure 17. Optical absorption spectrum of Au NPs and nanorods.Inset shows a TEM image of the nanorods. Absorption spectrum ofhuman blood is also shown for comparison. While the SP band ofNPs and the transversal SPs of nanorods falls in the same region ofthe spectrum as the blood, the longitudinal SP band is located at thenear-IR, where biological tissue scarcely absorbs.

window between 800 and 110 nm where the blood and tissuesscarcely absorb (see figure 17). Nevertheless, even with the useof nanorods, it is very complicated to observe directly scatteredor transmitted light across biological tissue due to the largenumber of scattering bodies in the bloodstream (erythrocytes,leukocytes, thrombocytes, etc). Thus, the best results in this

11


Figure 18. Left: schematic of in vivo experiment to detect injected gold nanorods by optoacoustic imaging. Right: a typical optoacousticimage of a nude mouse before (a) and after (b) subcutaneous injection of Au nanorods in the abdominal area. Injected NPs were brightlyvisible in the optoacoustic image (b). Drawing in (a) depicts the approximate position of the nude mouse during the experiment.Reproduced with permission from [90].

line up to now have been achieved using nanorods combinedwith other techniques such as SERS [89] and photoacousticimaging.

For the combined use of SP resonance and photoacous-tic techniques, gold nanorods functionalized with the appro-priated biomolecules are injected into the tissue. The tissueis later illuminated with pulses of light at the SP resonancewavelength in the NIR. The strong absorption cross section ofthe nanorods leads to a local increase in temperature duringthe laser pulse. This heat is quickly dissipated to the adjacenttissue which undergoes a thermoplastic expansion creating anacoustic wave that propagates through the tissue and can bemechanically detected at the surface. Using this technique,Eghtedari et al recently detected Au nanorods in mice [90].The experiment is illustrated in figure 18. Au nanorods werechemically synthesized controlling the aspect ratio to exhibitthe maximum absorption of longitudinal SPs at about 750 nm.The nanorods were injected into the mouse that was placedonto an acoustic transducer to detect acoustic waves and con-vert them into an electrical signal. The mouse was illumi-nated with an alexandrite laser (757 nm wavelength) emitting75 ns pulses with an average energy per pulse of 6 mJ cm1.The laser beam scanned the mouse recording the signal onthe transducer after each pulse. Regions with high concentra-tion of Au nanorods presented enhanced light absorption andconsequently produced more intense acoustic waves after thelaser pulse that were detected with the acoustic transducer afew microseconds after the laser pulse. The images of pho-toacoustic wave intensity obtained upon scanning the mousewith the laser beam before and after gold nanorod injectionare shown in figure 18. The region where the nanorods wereinjected is brighter demonstrating the capability of the methodto detect gold nanorods in a living organism.

Hyperthermia. In addition to the possibility of detecting,imaging and labelling biological entities, the SPs offer thepossibility to activate processes in a controlled way, so theycan also be used for therapy. The intense light absorption and

electromagnetic field enhancement allow energizing the NPsto trigger events locally as desired.

Hyperthermia is a non-invasive technique for cancertreatment consisting of heating the biological tissues totemperatures slightly (510 C) over the normal one topromote the selective destruction of abnormal cells [91].This increase in temperature may induce denaturalization ofproteins, disruption of the organized biomolecular assembliesin the nucleus and cytoskeleton, and increase the membranepermeability [92]. A further step is the selective hyperthermia,consisting of a controlled heating of malignant cells. Thisstrategy can provide advantages in cancer treatments as thenormal tissue remains unaffected promoting the replacementof malignant cells with healthy ones, although the long termimpact on patients is still undetermined.

The light absorption associated with SP excitationproduces a local heating that can be exploded for selectivehyperthermia. For this purpose, nanorods are functionalizedto bind specific malignant cells (as in the imaging techniquesdescribed above). When the tissues are illuminated with anNIR laser at the SP wavelength, the energy absorbed by thenanorods is partially converted into heat, increasing locallythe temperature. Due to the high absorption efficiency ofnanorods it is possible to increase their temperature severaldegrees. As the functionalization of the nanorods promotetheir accumulation around malignant cells, these later will beselectively heated, favouring their elimination with a reducedeffect on healthy cells.

Recently, Wei co-workers [9294] have shown in vitroselective elimination of adherent human KB cells (a tumourcell line derived from oral epithelium) using gold nanorodsto perform localized hyperthermia. An adherent KB cellculture was exposed to gold nanorods conjugated to promotethe binding to the cell membrane (see figure 19).

A laser working at 815 nm (where the NR exhibited a highabsorption due to excitation of SPs) with femtosecond pulseswas used to irradiate the culture. The irradiation induceda local heating of the nanorods (and consequently of themalignant cells) leading to a severe damage of the membrane.

12


Figure 19. Photothermolysis mediated by folate-conjugatednanorods. (a), (b) KB cells with membrane-bound nanorods (red)exposed to fs-pulsed near-infrared laser irradiation experiencedmembrane damage and blebbing. The loss of membrane integritywas indicated by ethidium bromide nuclear staining (yellow). (c),(d) NIH-3T3 cells were unresponsive to nanorods and did not sufferphotoinduced damage under the same irradiation conditions.Reproduced with permission from [93].

The membrane blebbing was optically confirmed using a dye(ethidium bromide) as indicator of the membrane permeability.As figure 18 illustrates, in the absence of nanorods, the celldid not suffer appreciable effects. Therefore, with the properfunctionalization of the nanorods to bind specific malignantcells, it is possible to perform selective hyperthermia on thesecells, while normal ones do not suffer significant damage.Positive results have been found also functionalizing Aunanorods with CTAB and folate that promotes the uptakeof nanorods and internalization by the KB cells instead ofaccumulating at the membrane [92].

Drug delivery. This heat dissipation and local increase intemperature associated with SP excitation can also be usedfor drug delivery. This is particularly useful for certaindrugs that are highly aggressive and may induce undesiredsecondary effects on normal tissues. Controlled drug deliveryleads to higher efficiency reducing these secondary effects.Commonly, the strategies of drug delivery are based oncovering the drug with a coating that avoids interaction withnon-targeted cells [95]. Once the drug reaches the targetcell, the coating is removed releasing the drug to act at thedesired region. Moreover, release rate can also be controlled toimprove effectiveness. Metallic NPs holding SPs can be usedfor this purpose [9699]. A controlled release of the drug canbe achieved by exciting SPs: the light concentration around theNP can provide the energy required to activate the drug releaseprocess. Figure 20 presents an example of SP controlled drugdelivery [95]. The drug is coated with a shell consisting of a

polymer with embedded gold NPs and functionalized on theouter side with antibodies for specific cells. Once the ensemblereaches the target cells, the region is irradiated with light at theSP wavelength. The excitation of SPs at the NPs induces a localheating and melting of the polymer shell, releasing the drugat the desired region and time. The heating efficiency of themetallic NPs can be optimized with different methods [100].

3.2. Energy

The present-day energy problem impels us to search for energysources different from traditional ones based on fossil fuels.In this search, the Sun appears as a clean source of energywith a lifetime of over 5000 million years releasing daily4000 times the worlds electric energy consumption. Theuse of solar energy can be addressed by different techniquesincluding photovoltaic [101104], photochemical [105107]or photothermal [108110] ones. The main limitation of thesetechnologies relies on the efficiency of the light absorptionand conversion processes that make them hardly viable froman economical point of view. The capabilities to concentrateand locally amplify the light electric field by means of SPsoffer new methods to increase the efficiency of photoenergeticprocesses, overcoming the efficiency limitation required forthe massive implantation of these technologies.

Photovoltaic devices represent the most exploded wayto harvest solar energy. While there are several generationsof photovoltaic devices [103], all of them are based onjunctions where electronhole pairs created by light absorptionare separated, leading to charge accumulation that produceselectric voltage, i.e. electric energy, as figure 21(a) illustrates.The main limitation in the development and implantationof photovoltaic devices is their efficiency. Photovoltaicdevices are economically viable with a cost below 1$ W1[111], a value scarcely achievable by last generation solarcells [103]. A key problem in improving the efficiency ofphotovoltaic devices is that large thicknesses are required toabsorb most of the incoming light, especially for silicon-based solar cells, due to the indirect gap nature of silicon.However, these larger thicknesses enhance the electronholerecombination probability increasing losses and requires alarger amount of material (i.e. larger production costs). Thus,the efficiency of solar cells still is an open question: nowadays,the most efficient cell devices reach 25% energy conversionefficiency [103].

The incorporation of metallic NPs in solar cells canincrease the efficiency of charge separation by several physicalprocesses [10, 112, 113]. As figure 21 illustrates, if we placemetallic NPs at the device surface, upon light illumination, thelarge scattering cross section associated with SPs can scatterthe normal incident light beams. Hence, the effective pathacross the active absorption layer is increased. Moreover,with a proper architecture, this scattering may be used topromote total reflection inside the layer trapping the photonsuntil they are finally absorbed (figure 21(b)). Another methodto improve efficiency exploding SPs consists in placing the NPsat the junction interface (figure 21(c)). The light concentrationand local amplification in the vicinity of the NP when SPs

13


Figure 20. (a) Drug is coated with a polymer shell containing gold NPs and antibodies for specific uptake. (b) When the drug reaches thetarget, SPs are excited with a laser beam, the dissipated heat melts the polymer and (c) releases the drug at the desired region. Reproducedwith permission from [95].

Figure 21. Top: scheme of a photovoltaic cell. Light absorption atthe pn junction induces charge separation. If the absorption takesplace far from the junction, recombination is more likely, reducingthe cell efficiency; bottom: illustration of processes improvingphotovoltaic efficiency by (a) light scattering and (b) localamplification. Reproduced with permission from [10].

are excited increases the absorption efficiency. Stuart andHarris [114] demonstrated an increase up to a factor 20 inthe photocurrent of a silicon photodetector by placing silverNPs on the surface.

Increases of photovoltaic cell efficiency of the order of1015% have been achieved by incorporating Au and Ag NPson the cell surface [115119] (figure 22). The efficiency inthis kind of device has a complex dependence on the size ofNPs, shape and spatial distribution, since interacting effectsplay a different role on the light scattering and absorptionprocesses [112]. The impact of both mechanisms in the finalimprovement is not easy to disentangle although most recentresults point towards the scattering process as the key one.Actually, some outstanding results have been achieved usingnon-resonant silica NPs [11]. The use of SPs has been explored

not only in silicon-based solar cells but also in third generationorganic devices [120122] with efficiency enhancement up toa factor of 2.

Photothermal technology is an alternative to photovoltaicsespecially when the energy is used for heating [123]. Intypical photothermal devices, a fluid absorbing light increasesits temperature and is submitted to a thermal cycle releasingenergy. The efficiency of thermal cycles is limited by thetemperature difference between the hot and cold focus. Thus,reducing the mass that absorbs the solar light will increase itstemperature and consequently improve the efficiency of thedevice. The huge extinction cross section of SPs provides amethod to achieve intense light absorption by small masses.Moreover, the tunability of SP resonance band allows thedesign of selective absorbers that reduces energy losses dueto blackbody radiation [124].

In addition to the pure physical methods, SPs can also beused to catalyse reactions for energy harvesting using chemicalpaths, for instance the water decomposition for hydrogengeneration as described in the next section.

In general, any photoenergetic process addresses theproblem to explode the wide solar spectrum. As most opticaltransitions in bulk materials exhibit well-defined absorptionbands, it is difficult to design materials capable of absorbingefficiently over a large part of the solar spectrum. SPs are notan exception, but, as described above, it is possible to tunethe SP band in a wide range, something not straightforwardwith other absorbing materials such as dyes or semiconductors.This tunability renders SPs especially advantageous for widespectrum photoenergetic processes.

3.3. Environment

The development of technological societies requires aprogressive increase of the industrial activity. Manyindustrial processes result in hazardous products to humansand the environment [125]. Reaction of these productsto be transformed into inert ones is mandatory in order

14


Figure 22. Improvement of efficiency on (a) silicon and (b) organic photovoltaic cells by incorporation of silver NPs. Reproduced withpermission from [119] and [121], respectively.

to ensure a secure and sustainable development of oursociety. This problem has got particular importance withthe recent incorporation to the first world of new regionsthat increased exponentially their industrial activity in thelast years. Catalysis provides a pathway to eliminatethese hazardous products quickly, hence reducing theirpotential danger. Among the different catalytic processes,photocatalysis exhibits particular importance since it is a cleanprocess that can use solar light and mimics other biologicalprocesses [126128].

Metallic NPs exhibit catalytic activity [129] that canbe improved upon light illumination to excite SPs [130].However, the most exploded use of SPs in photocatalysis is theenhancement of photocatalytic efficiency in transition metaloxides by incorporation of metallic NPs. The principle ofphotocatalysis in these materials [131, 132] is illustrated infigure 23. Transition metal oxides (such as TiO2) exhibitan energy gap in the UVVIS part of the spectrum. Thus,upon illumination, electrons can be promoted to the conductionband. The excited electrons are available to participate in theoxidation of the products that are absorbed on the catalyticsurface. Similarly, holes created in the valence band allow thedonation of an electron from the absorbed products yieldingtheir reduction. It is straightforward to understand that theefficiency of the catalytic surface (measured as the numberof reactions that takes place per unit area and time) willdepend critically on the light absorption process that governsthe number of active electrons/holes in the material.

The incorporation of metallic NPs have been used in thepast since they act as electron traps retarding electronholerecombination and therefore increasing the process efficiency[133135]. However, the use of NPs exhibiting SP resonanceby the photocatalytic surface generates an additional increasein light absorption efficiency by the local electric fieldamplification and scattering processes as in the photovoltaicdevices described above.

Awazu et al [13] demonstrated the increase inphotocatalytic efficiency of a TiO2 layer by embedding Ag NPsnear the surface as shown in figure 24(a). The mechanism ofefficiency improvement is basically the same described abovefor photovoltaic devices (figure 21). The light scattering atthe silver NPs increases the effective optical path of photons

O2

O2- M+ (aq.)

H2O- Org. Comp.

H2O

M (s)

H2O CO2

Photon

CB

VB

Egap

h

e

Photo-reduction

Phot

o-ox

idat

ion

Figure 23. Scheme of the photocatalysis processes in transitionmetal oxides.

leading to a higher absorption rate, while the local electricfield enhancement also assists the absorption processes. Withthis arrangement, they found an increase above one orderof magnitude in the decomposition rate of methylene blueunder near-UV irradiation (figure 24(b)). Similarly thecatalysis of carbon monoxide is substantially improved bythe combined use of Au and Fe2O3 NPs [136], as shown infigure 24(d). While separated use of Au or Fe2O3 NPs leadsto almost negligible photocatalytic efficiency, their combineduse shows outstanding activity results as the excitation ofSPs in Au NPs enhances the photocatalysis of Fe2O3. SPenhancement photocatalysis of TiO2 has been demonstratedfor other reactions such as methanol dehydrogenation, aceticacid decomposition, propanol oxidation and others [137140].

The incorporation of metallic NPs holding SPs in pho-tocatalytic materials may have additional advantages. A keyproblem in some photocatalysts such as ZnO is the photo-instability in aqueous solution due to photocorrosion upon UVillumination [141]. It has been reported that incorporation ofAg NPs onto ZnO films leads to an increase in photocatalyticefficiency and higher stability of ZnO preventing photocorro-sion [142]. As figure 24(c) shows, the presence of Ag NPsavoids the photodegradation upon UV illumination.

SPs can also be used to probe catalytic reactions. Larssonet al [14] could monitor CO and H2 oxidation on Pt, and

15


Figure 24. (a) TEM images of TiO2 film on Si with Ag NPs and (b) decomposition rate of methylene blue upon UV irradiation on the TiO2surface without (black) and with (red and blue) Ag NPs with different geometries. (c) Durability of ZnO photocatalysts for thephotodegradation of crystal violet under UV irradiation with and without Ag NPs. (d) Photocatalytic CO decomposition activity usingFe2O3 and Au NPs. (e) Shift in the SPR peak of Au NPs during the photo-oxidation of NOx NO2 on BaO photocatalyst. Reproducedwith permission from [13] (a) and (b), [142] (c), [136] (d) and [14] (e).

NOx conversion to N2 on Pt/BaO reaction by followingchanges in the SP resonance of Au NPs incorporated in thephotocatalytic surface. The high sensitivity of SPs to thelocal environment dielectric properties induces a shift in theSP resonant position when the catalytic process takes placeas illustrated in figure 24(e). With this method, they wereable to detect surfactant coverage with sensitivity below 103

monolayers.A particularly interesting case of photocatalysis is

the photo-assisted hydrogen fuel generation. Waterdecomposition (2H2O 2H2 +O2) provides the possibility tostore hydrogen to be used as fuel [143, 144]. The endothermicdecomposition process (that takes place spontaneously attemperatures of about 2500 K [145]) may be activated byseveral methods, that increase locally the temperature orcatalyse the reaction including chemical, electrical, thermaland optical ones. Thus, solar light can be used to promote water

splitting [143, 145] using a fully clean and environmentallyfriendly method. As in the abovementioned processes, SPshave demonstrated their capabilities to improve the efficiencyof the process [135, 146, 147]. Liu et al [147] recently found anincrease in a factor of 66 on the water splitting photocatalysisby TiO2 when incorporating Au NPs.

It is worth mentioning that an indirect improvement ofcatalytic efficiency using SPs may be due to the increasein temperature induced by light absorption. The localenhancement of the temperature associated with SP excitationmay help the catalysis process that should be a photothermalassisted catalysis rather than a real photocatalytic one.

3.4. Information technology

Optical signals are essential for information technologydue to fast propagation along communication lines and the

16


Figure 25. (a) Illustration of the laser-induced reshaping of nanorods and (b) TEM images and optical absorption spectra of the nanorods.(c) Images recorded at the same position observed with different light wavelengths and polarizations. Reproduced with permissionfrom [158].

absence of ohmic losses. Light manipulation is commonlyachieved by dielectric materials while metals are scarcelyused due the short light penetration depth. However, withdielectric materials, it is impossible to focus a beam below/2 due to the diffraction at the aperture edges [148]. Anadditional key problem of optical information storage is thedifficulty in obtaining re-usable memories since stable andmechanically solid rewritable optical devices are not easyto develop. These restrictions have represented a seriousproblem for the development of information technologywhere miniaturization has achieved the nanometric scale.Magnetic memories use areas below 100 nm size for bitwriting/recording, values that cannot be used in traditionaloptical ones.

Most of the plasmonic applications in this field areachieved with itinerant SPs [2, 3] but there are also someoutstanding examples of localized SP applications forinformation technologies. The use of nanometals holdingSPs arises new methods for light manipulating, reading andwriting information in sizes significantly smaller than the lightwavelength, while the huge absorption and scattering crosssections allows detection of very small signals [149, 150].Thus, SPs can be used to design subwavelength optical

devices as antennas, lenses and resonators for extreme lightconcentration and manipulation [4, 151].

The electric field amplification around a metallic NP whenSPs are excited makes these elements to act as nanoantennaswith very low power consumption and large amplificationfactors [13, 152]. Moreover, patterns of NPs with specificgeometries provide methods for selective frequency antennaswith even larger amplification [2, 3, 153, 154] and to fabricatesubwavelength light waveguides with a decay length of theorder of 500m [16, 155]. Photodetectors with improvedperformance can also be developed exploding SPs [156]similarly to the case of photovoltaic cells. The incorporationof metallic NPs in photodetectors increases their efficiencyand allows you to reduce their size [149, 157] leading to fasterdevices with reduced power consumption.

The dependence of SP resonance on factors such as size,shape and orientation permits the design of new devices thatrepresent a breakthrough with conventional ones. Recently,Zijlstra et al [158] designated a five-dimensional storagedevice based on the use of Au NPs and nanorods. Inaddition to the two dimensions of the storage surface, theyincluded the size of NPs, aspect ratio and orientation asnew ones.

17


In this device, the information is written using thethermal reshaping of the NPs: when SPs are excited witha laser, the local increase in temperature induces a meltingthat allows one to reshape the particles obtaining nanorodswith controlled orientation and aspect ratio that depend onthe laser wavelength, intensity and polarization [4750].Moreover, as the SP band position depends on the NP geometry(figure 25(b)) it is possible to reshape selectively NPs withcertain geometry into other one. Hence, the populationof nanorods and their geometrical features can be tunedoptically (figure 25(a)). The information is read using differentwavelengths and light polarizations to excite selectively somenanorods depending on their morphology. Consequently, itis possible to store several bits of information in the samephysical space. With 5 different wavelengths and 2 differentlight polarizations they were able to write and read 6 differentimages in the same area, as figure 25(c) illustrates with apotential information density of 1 Tby cm2.

Another example of the potential use of SPs for theinformation technology is extraordinary light transmission[159162]. When a uniform light beam reaches an opaquesurface with nanometric holes, the light transmission is limitedby the diffraction laws. Such a limit can be overcome whendealing with metallic films with holes and SPs are excited. Theexcitation of SPs produces a charge accumulation at the edge ofholes. This leads to a non-uniform wavefront that concentratesthe light at the hole region increasing the light transmission.

Excitation of SPs can also be used to enhance thesensitivity of Raman spectroscopy by concentrating and locallyamplifying the light electric field yielding to much bettersensitivity of the technique [7, 163166]. The performanceof scanning near field optical microscopy (SNOM) is alsoimproved by placing metallic NPs at the tip that concentratesthe light [149, 167].

4. Outlook and challenges

The excitation of SPs in metallic NPs allows us to exploreregions prohibited by traditional optics, overcoming thediffraction limit and allowing local light amplification oforders of magnitude. Therefore, potential uses of SPs areas ubiquitous as the optics is. These applications includecompletely new approaches to problems not yet solved andimprovement of well-established optical processes. Futurebasic research on SPs will certainly prompt new potentialapplications. However, the present-day knowledge is solidenough to develop a vast range of applications, some of themalready achieved, others in development and some others yetto be addressed.

The milestone for further development of theseapplications relies on our ability to process and manufacturenanomaterials. In the case of biomedical applications whereeconomical cost is not a key issue, the challenge is aperfect control of NP size and shape dispersion, as well asfunctionalization. The recent advances in colloidal chemistrypushed the field substantially in recent years to be themost developed among SP applications. For energy andenvironmental sectors, processing large amounts of NPs with

a relatively good control of morphology and size dispersion atcompetitive costs is the key to promote a widespread use of SPtechnology. Finally, for information technology, the advanceson designing and understanding SPs of complex structures andNP ensembles as well as improvement of lithography methodswill drive future advances.

Acknowledgments

I would like to thank Prof. J Llopis for lectures and fruitfuldiscussions about surface plasmons over many years. Josede la Venta is thanked for critical reading of the manuscriptand valuable comments. This work has been partiallysupported by Spanish Ministry of Science and innovationthrough the projects FIS-2008-06249, MAT2010-C21088-C03and Comunidad de Madrid, project NANOBIOMAGNET(S2009/MAT-1726).

References

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

[2] Brongersma M L and Kik P G 1988 Surface PlasmonNanophotonics (Berlin: Springer)

[3] Maier S A 2006 Plasmonics (Berlin: Springer)[4] Barnes W L, Dereux A and Ebbesen T W 2003 Nature

424 824[5] Fritzsche W and Taton T A 2003 Nanotechnology 14 R63[6] Hu M, Chen J, Li Z Y, Au L, Hartland G V, Li X, Marquez M

and Xia Y 2006 Chem. Soc. Rev 35 1084[7] Eustis S and El-Sayed M A 2006 Chem Rev. Soc. 35 209[8] Garcia M A 2010 Surface Plamons in biomedicine Recent

Developments in Bio-Nanocomposites for BiomedicalApplications ed A Tiwari and S Pilla (New York:Novascience Publishers) ISBN 978-1-61761-008-0

[9] Pillai S, Catchpole K R, Trupke T and Green M A 2007J. Appl. Phys. 101 093105

[10] Atwater H A and Polman A 2010 Nature Mater. 9 205[11] Matheu P, Lim S H, Derkacs D, McPheeters C and Yu E T

2008 Appl. Phys. Lett. 93 113108[12] Narayanan R and El-Sayed M A 2005 J. Phys. Chem. B

109 12663[13] Awazu K, Fujimaki M, Rockstuhl C, Tominaga J,

Murakami H, Ohki Y, Yoshida N and Watanabe T 2008J. Am. Chem. Soc. 130 1676

[14] Larsson E M, Langhammer C, Zoric I and Kasemo B 2009Science 326 1091

[15] Homola J (ed) 2006 Surface Plasmon Resonance BasedSensors (Springer Series on Chemical Sensors andBiosensors) (Heidelberg: Springer)

[16] Ozbay E 2006 Science 311 189[17] Fedlheim D L and Foss C A 2001 Metal Nanoparticles:

Synthesis, Characterization, and Applications(Boca Raton, FL: CRC Press)

[18] Mariotti D and Sankaran R M 2010 J. Phys. D: Appl. Phys.43 323001

[19] Yong K T, Swihart M T, Ding H and Prasad P N 2009Plasmonics 4 79

[20] Grzelczak M, Perez-Juste J, Mulvaney P andLiz-Marzan L M 2008 Chem. Soc. Rev. 37 1783

[21] Zhanga Y and Erkey C 2006 J. Supercrit. Fluids 38 252[22] Panigrah S, Kundu S, Ghosh S K, Nath S and Pal T 2004

J. Nanopart. Res. 6 411[23] Goia, D V and Matijevic E 1998 New J. Chem. 22 1203

18

http://dx.doi.org/10.1038/nature01937http://dx.doi.org/10.1088/0957-4484/14/12/R01http://dx.doi.org/10.1039/b517615hhttp://dx.doi.org/10.1039/b514191ehttp://dx.doi.org/10.1063/1.2734885http://dx.doi.org/10.1038/nmat2629http://dx.doi.org/10.1063/1.2957980http://dx.doi.org/10.1021/jp051066phttp://dx.doi.org/10.1021/ja076503nhttp://dx.doi.org/10.1126/science.1176593http://dx.doi.org/10.1126/science.1114849http://dx.doi.org/10.1088/0022-3727/43/32/323001http://dx.doi.org/10.1007/s11468-009-9078-2http://dx.doi.org/10.1039/b711490ghttp://dx.doi.org/10.1016/j.supflu.2006.03.021http://dx.doi.org/10.1007/s11051-004-6575-2http://dx.doi.org/10.1039/a709236i


[24] Raether H 1988 Surface Plasmons on Smooth and RougeSurfaces and on Gratings (Berlin: Springer)

[25] Sambles J R, Bradbery G W and Yang F 1991 Contemp.Phys. 32 173

[26] Mie G 1908 Ann. Phys. 25 377[27] Angelini I, Artioli G, Bellintani P, Diella V, Gemmi M,

Polla A and Rossi A 2004 J. Archaeol. Sci. 31 1175[28] Barber D J and Freestone I C 1990 Archaeometry 32 33[29] Freestone I, Meeks N, Sax M and Higgitt C 2007 Gold Bull.

40 270[30] Colomban P, March G, Mazerolles L, Karmous T, Ayed N,

Ennabli A and Slim H 2003 J. Raman Spectrosc. 34 205[31] Colomban P 2009 J. Nano Res. 8 109[32] Jose-Yacaman M, Rendon L, Arenas J and Serra Puche M C

1996 Science 273 223[33] Colomban P and Truong C 2004 J. Raman Spectrosc. 35 195[34] Brun N, Mazerolles L and Pernot M 1991 J. Mater. Sci. Lett.

10 1418[35] Nakai I, Numako C, Hosono H and Yamasaki K 1999 J. Am.

Ceram. Soc. 82 689[36] Wood N 1999 Chinese Glazes (Philadelphia, PA: University

of Pennsylvania Press)[37] Perez-Villar S, Rubio J and Oteo J L 2008 J. Non-Cryst.

Solids 354 1833[38] Rubio F, Perez-Villar S, Garrido M A, Rubio J and Oteo J L

2009 J. Nano Res. 8 89[39] Padovani S, Sada C, Mazzoldi P, Brunetti B, Borgia I,

Sgamellotti A, Giulivi A, DAcapito F and Battaglin G2003 J. Appl. Phys. 93 10058

[40] Vilarigues M, Fernandes P, Alves L C and da Silva R C 2009Nucl. Instrum. Methods Phys. Res. B 267 2260

[41] Bobin O, Schvoerer M, Ney C, Rammah M, Pannequin B,Cilia Platamone E, Daoulatli A and Gayraud R P 2003Color: Res. Appl. 28 352

[42] Leatherdale C A, Woo W K, Mikulec F V and Bawendi M G2002 J. Phys. Chem. B 106 7619

[43] Jia W, Douglas E P, Guo F and Sun W 2004 Appl. Phys. Lett.85 6326

[44] Dresselhaus M S and Dresselhaus G 1962 Phys. Rev.125 499

[45] Mooradian A 1969 Phys. Rev. Lett. 22 185[46] Dynich R A and Ponyavina A N 2008 J. Appl. Spectrosc.

75 832[47] Link S, Burda C, Nikoobakht B and El-Sayed M A 2000

J. Phys. Chem. B 104 6152[48] Huang W, Qian W and El-Sayed M A 1998 J. Appl. Phys. 98

114301[49] Chang S S, Shih C W, Chen C D, Lai W C and Wang C R C

1999 Langmuir 15 701[50] Link S, Burda C, Nikoobakht B and El-Sayed M A 2000

J. Phys. Chem. B 104 6152[51] Lichteneker K 1926 Physik Z. 27 115[52] Maxwell-Garnett J C 1904 Phil. Trans. R. Soc. 203 385[53] Maxwell-Garnett J C 1906 Phil. Trans. R. Soc. 205 237[54] Niklasson G A, Granqvist C G and Hunderi O 1981 Appl.

Opt. 20 26[55] Liebsch A and Persson B N J 1983 J. Phys.C: Solid State

Phys. 16 5375[56] Bergman D J 1978 Phys. Rep. 43 377[57] Garcia M A, Paje S and Llopis J 1999 Chem. Phys. Lett.

315 313[58] Gans R 1912 Ann. Phys. 37 881[59] Kreibig U, Schmitz B and Breuer H D 1987 Phys. Rev. B

36 5027[60] Hovel H, Fritz D, Holger A, Kreibig U and Vollmer M 1993

Phys. Rev. B 48 18178[61] Kelly K L, Coronado E, Zhao L L and Schatz G C 2003

J. Phys. Chem. B 107 668

[62] Hovel H, Fritz S, Holger A, Kreibig U and Vollmer M 1993Phys. Rev. B 21 18178

[63] Zhang P and Sham T K 2002 Appl. Phys. Lett. 81 736[64] Garcia M A, de la Venta J, Crespo P, Lopis J, Penades S,

Fernandez A and Hernando A 2005 Phys. Rev. B72 241403

[65] Quinten M and Kreibig U 1986 Surf. Sci. 172 557[66] Westcott S L, Oldenburg S J, Lee T R and Halas N J 1999

Chem. Phys. Lett. 300 651[67] Garcia M A 2010 Surface plasmons in biomedicine Recent

Developments in Bio-Nanocomposites for BiomedicalApplications (New York: Novascience Publishers)

[68] Liao H, Nehl C L and Hafner J H 2006 Nanomedicine1 201

[69] Csaki A, Maubach G, Born D, Reichert J and Fritzsche W2002 Single Mol. 3 275

[70] Zu L and Soper S A 2006 Multiplexed Fluorescencedetection for DNA Sequencing Reviews in Fluorescencevol 3 (Berlin: Springer) chapter 23

[71] Schena M 2000 Microarray Biochip Technology(Walton-on-Thames, Surrey: EATON Pub)

[72] Hanfiled J F and Powell R D 2000 J. Histochem. Citochem.48 471

[73] Bloomfield V A, Crothers D M and Tinoco I 1974 J. Phys.Chem. Nucleic Acids (New York: Harper and Row)chapter 7

[74] Yezhelyev M V, Gao X, Al-Haji A, Nie S and ORegan S M2006 The Lancet Oncol. 7 657

[75] Brigger I, Dubernet C and Couvreur P 2002 Adv. Drug Deliv.Rev. 54 631

[76] Li J L, Wang L, Liu X Y, Zhang Z P, Guo H C, Liu W M andTang S H 2009 Cancer Lett. 274 319

[77] El-Sayed I H, Huang X and El-Sayed M A 2005 Nano Lett.5 829

[78] Cai W, Gao T, Hong H and Sun J 2008 Nanotechnol. Sci.Appl. 1 17

[79] Loo C, Hirsch L, Lee M H, Chang E, West J, Halas N andDrezek R Opt. Lett. 30 1012

[80] Sokolov K et al 2003 Cancer Res. 63 1999[81] Huang X, El-Sayed I H, Qian W and El-Sayed M A 2006

J. Am. Chem. Soc. 128 2115[82] Cai W, Gao T, Hong H and Sun J 2008 Nanotechnol. Sci.

Appl. 1 17[83] Jain P K, El-Sayed I H and El-Sayed M A 2007 NanoToday 2

18[84] Li P C et al 2007 IEEE Trans. Ultrason. Ferroelectr. Freq.

Documents

Surface plasmons in metallic nanoparticles.pdf