d.a. stuart, a.j. haes, c.r. yonzon, e.m. hicks and r.p. van duyne- biological applications of...

8/3/2019 D.A. Stuart, A.J. Haes, C.R. Yonzon, E.M. Hicks and R.P. Van Duyne- Biological applications of localised surface plas

1/21

Biological applications of localised surface plasmonicphenomenae

D.A. Stuart, A.J. Haes, C.R. Yonzon, E.M. Hicks and R.P. Van Duyne

Abstract: Researchers and industrialists have taken advantage of the unusual optical, magnetic,electronic, catalytic, and mechanical properties of nanomaterials. Nanoparticles and nanoscalematerials have proven to be useful for biological uses. Nanoscale materials hold a particularinterest to those in the biological sciences because they are on the same size scale as biologicalmacromolecules, proteins and nucleic acids. The interactions between biomolecules andnanomaterials have formed the basis for a number of applications including detection, biosensing,cellular and in situ hybridisation labelling, cell tagging and sorting, point-of-care diagnostics, kineticand binding studies, imaging enhancers, and even as potential therapeutic agents. Noble metalnanoparticles are especially interesting because of their unusual optical properties which arise fromtheir ability to support surface plasmons. In this review the authors focus on biological applicationsand technologies that utilise two types of related plasmonic phenomonae: localised surface plasmonresonance (LSPR) spectroscopy and surface-enhanced Raman spectroscopy (SERS). Thebackground necessary to understand the application of LSPR and SERS to biological problems

is presented and illustrative examples of resonant Rayleigh scattering, refractive index sensing, andSERS-based detection and labelling are discussed.

1 Introduction

The intense scattering and absorption of light from noblemetal nanoparticles is the source of some of the beautifulcolours in stained glass windows and has attracted theinterest of scientists for generations. These nanoparticlesexhibit a strong UV-vis absorption band that is not present

in the spectrum of the bulk metal [18]. Although scientistshave learned that the characteristic hues of these noblemetal nanoparticle suspensions arise from their stronginteraction with light, the advent of the field of nanoparticleoptics has allowed for a deeper understanding of therelationship between material properties such as composi-tion, size, shape, and local dielectric environment and theobserved colour of a metal suspension. An understanding ofthe optical properties of noble metal nanoparticles holdsboth fundamental and practical significance. Fundamen-tally, it is important to systematically explore the nanoscalestructural and local environmental factors that cause opticalproperty variation, as well as provide access to regimes ofpredictable behaviour. Practically, the tunable opticalproperties of nanostructures can be applied as materialsfor surface-enhanced spectroscopy [913], optical filters [14,15], plasmonic devices [1619] and sensors [2034].

One of the most interesting properties of noble metalnanoparticles arises from their ability to support a localisedsurface plasmon resonance (LSPR). The LSPR results whenthe incident photon frequency is resonant with the collective

oscillation of the conduction electrons of the nanoparticle.The simplest theoretical approach available for modellingthe optical properties of nanoparticles is the Mie theoryestimation of the extinction of a metallic sphere in the longwavelength, electrostatic dipole limit. In the followingequation [35]:

El 24pNAa3e3=2ml ln10

ei

er 2em2 e2

i

" #1

E(l) is the extinction which is, in turn, equal to the sum ofabsorption and Rayleigh scattering, NA is the areal densityof nanoparticles, a is the radius of the metallic nanosphere,em is the dielectric constant of the medium surrounding themetallic nanosphere (assumed to be a positive, real numberand wavelength independent), l is the wavelength of theabsorbing radiation, ei is the imaginary portion of themetallic nanospheres dielectric function, and er is the realportion of the metallic nanospheres dielectric function. TheLSPR condition is met when the resonance term in the

denominator ((er+2em)

2

) approaches zero. Even in thismost primitive model, it is abundantly clear that the LSPRspectrum of an isolated metallic nanosphere embedded inan external dielectric medium will depend on the nanopar-ticle radius a, the nanoparticle material (ei and er), and thenanoenvironments dielectric constant (em). Furthermore,when the nanoparticles are not spherical, as is always thecase in real samples, the extinction spectrum will depend onthe nanoparticles in-plane diameter, out-of-plane height,and shape. In this case the resonance term from thedenominator of (1) is replaced with:

er wem2 2

where w, a shape factor term [11] is a term that describes the

nanoparticles aspect ratio. The values for w increase fromtwo (for a sphere) up to, and beyond, values of 17 for a 5:1aspect ratio nanoparticle. In addition, many of the samples

The authors are with the Department of Chemistry, Northwestern University,2145 Sheridan Road, Evanston, IL 60208-3113, USA

E-mail: [email protected]

r IEE, 2005

IEE Proceedings online no. 20045012

doi:10.1049/ip-nbt:20045012

Paper first received 22nd December 2004
http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


2/21

considered in this work contain an ensemble of nanopar-ticles that are supported on a substrate. Thus, the LSPR willalso depend on interparticle spacing and the substratedielectric constant.

LSPR excitation results in wavelength selective absorp-tion with extremely large molar extinction coefficientsB3 10111/M cm [36], resonant Rayleigh scattering [37,38] with an efficiency equivalent to that of 106 fluorophors[39] and the enhanced local electromagnetic fields near thesurface of the nanoparticle which are responsible for theintense signals observed in all surface-enhanced spectro-scopies, e.g. surface-enhanced Raman scattering (SERS)and surface-enhanced fluorescence [11, 26, 27].

2 Synthetic methods to produce nanoparticles

One of the aims of any application, particularly for sensingresearch, is to be able to generate reproducible results. Innanobiological applications, this translates directly into acritical need to understand and control the syntheticmethods used to generate novel nanoscale materials. Atthe nanoscale, there are many options available rangingfrom nanoparticles in solution, to surface-bound nanopar-ticles, to nanoparticles entrapped in a three-dimensional

matrix. With this plethora of potentially useful nanomater-ial classes, there exist a large number of fabrication methodsto controllably and reproducibly produce these samples. Inthis Section, we will discuss the most pertinent syntheticmethods used in biosensing application: chemical synthesisand lithographic techniques.

2.1 Nanoparticle synthesis by chemicalreductionThe synthesis of inorganic nanocrystals with controllableshapes and sizes has been a driving force in much ofmaterials research. This has been characterised by rapidprogress in the past few years with synthesis of spheres [40,41], rods [4244], triangular prisms [4547], disks [4850],cubes [51] and branched nanocrystals [52]. These structuresare compelling for their use in biolabelling and biosensing.In this review, we will concentrate on the synthesis and useof noble metal nanoparticles for biological sensing plat-forms. The most commonly prepared shapes are spheroids,triangular prisms, rods and cubes (Fig. 1). All of theseshapes are based on the reduction of a metal salt to producenanoparticles with varied shapes and sizes. A different butrelated synthetic technique is the electrochemical reductionof metal salts in the presence of a surfactant or template[5357]. These methods are most effective at producinglarge aspect ratio particles. The effect of nanoparticle shapeon the optical properties is clearly illustrated in Fig. 2, which

shows the scattering spectra from four different morphol-ogies [58].

2.2 Core-shell morphologiesCore-shell particles encompass a range of sample types frommetal surrounding metal [59, 60] to dielectric surroundingmetal. Most notably is the work of Ung et al. [61]. Theseparticles are produced by first creating the metal cores byone of the standard reduction reactions described above. Tostabilise the colloids and remove the citrate ions, the colloidsare mixed and allowed to react with an aminoalkane orthioalkane. Once the citrate has been completely displaced,the colloidal solution is allowed to react with a solution ofactive silica to form the shell (Fig. 3a). After 24h, the

particles can be used for further chemical modification or toincrease the shell thickness [6163]. Transmission electronmicroscopy (TEM) and dark-field optical micrographs of

this type of core-shell structure are shown in Fig. 3b and c.Recently, Yin et al. [64] have developed a simpler and fastermethod to silica coat particles. Their method involves

suspension of aqueous phase particles within an isopropa-nol/water mixture followed by the direct addition of asmall fraction of tetraethyl orthosilicate (TEOS). The TEOS

a

166 nm

c d

b

83 nm 100 nm

125 nm

Fig. 1 Bright field TEM imagesa Gold nanorods (sample provided by Catherine Murphy)b Gold colloidsc Silver triangular prismsdSilver nanocubes (Reproduced with permission from [51]. Copyright2002)

0

1000

800

600

400

200scatteringintensity

400 500 600 700

= 52 nm

(220 meV)

800wavelength, nm

900

a

1000

800

600

400


0400 500 600 700 800

wavelength, nm900

b

1000

800

600

400


0400 500 600 700 800

wavelength, nm900

d

= 100 nm

(271 meV)

900

1000

800

600

400


0400 500 600 700 800

wavelength, nm

c

Fig. 2 Resonant Rayleigh scattering spectra of four individual Agnanoparticlesa The single resonance near 540 nm signifies a spherical nanoparticlewith a diameter near 50 nm

b and c The broad peaks near 650 nm with shoulders are indicative ofplatelet-shaped nanoparticles having a polygonal cross-section

d Two symmetric peaks, an intense one near 700 nm and weaker onenear 450nm, is suggestive of a rod-shaped nanoparticle that is

approximately 40 nm in diameter and 300nm long
http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


3/21

quickly forms a silica shell around the nanoparticleindistinguishable from those created in the Ung et al.method. The Yin et al. method is rapid, but unsuitable forcertain applications, notably the co-encapsulation of Ra-man-active species within the silica shell (see below), and thecoating of particles formed in the presence of surfactantshas met with limited success.

Another form of core-shell particles has a metal shellsurrounding a dielectric centre. The dielectric core, usuallysilica, is produced by the Sto ber method to producemonodispersed silica cores ranging in size from 80500 nm[65, 66]. These particles are then functionalised with3-aminopropyltrimethoxysilane. These functionalised parti-cles are then reacted with small gold nanoparticles. The gold

particles provide nucleation sites for the growth of themetallic shell from metal salts in solution. NH4OH is addedto produce preferential deposition onto the surface, rather

than initiate new particle formation in solution (Fig. 3d)[6569]. The TEM and dark-field optical micrographs ofthis type of nanoshell are shown in Figs. 3e and 3f.

2.3 Lithographic techniquesAll nanoparticle fabrication methods discussed thus far areused to create suspensions of nanoparticles in solution.Another class of techniques addresses substrate-boundnanostructure fabrication. The standard approach formaking substrate-bound nanostructures is electron beam

lithography (EBL). In EBL, the desired pattern is seriallyproduced by exposing a thin layer of photo-resist to highenergy electrons, followed by chemical development anddeposition of the noble metal, Fig. 4a. Several researchgroups have focused on the optical properties of two-dimensional arrays in order to utilise these nanoparticleassemblies as surface-enhanced spectroscopy substrates[7072]. While EBL provides exquisite control overnanoscale morphology, it is an expensive and time-consu-ming technique. An alternative method for the large-scaleproduction of surface-bound nanoparticle arrays is nano-sphere lithography (NSL).

NSL is a powerful fabrication technique to inexpensivelyproduce nanoparticle arrays with controlled shape, size, andinterparticle spacing [73]. Every NSL structure begins withthe self-assembly of size-monodispersed nanospheres ofdiameter D to form a two-dimensional colloidal crystaldeposition mask [8, 7375]. As the solvent evaporates,capillary forces draw the nanospheres together, and the

nanospheres crystallise into a hexagonally close-packedpattern on the substrate. As in all naturally occurringcrystals, nanosphere masks include a variety of defects thatarise as a result of nanosphere polydispersity, site random-ness, point defects, line defects, and polycrystalline domains.Typical defect-free domain sizes are in the 10100 mm range.Following self-assembly of the nanosphere mask, a metal orother material is then deposited by thermal evaporation,electron beam deposition, or pulsed laser deposition,Fig. 4b. After metal deposition, the nanosphere mask isremoved, leaving behind surface-confined nanoparticleswith triangular footprints.

3 Sensing with noble metal nanostructures

It is apparent from (1) that the location of the extinctionmaximum of noble metal nanoparticles is highly dependent

b c

e f

d

a

TEOSthiosilane

60 nm

100 nm

Fig. 3 Schematic illustration and micrographs of core-shellnanoparticlesa Metal nanoparticles are functionalized with mercapto-silanes, whichserve as nucleation sites of the growth of the silica shell

b TEM micrograph of silica encapsulated metal nanoparticlesc Dark-field optical micrograph of the silica-encapsulated metalnanoparticles

dA dielectic core is decorated with small gold nanoparticles, which arethen grown via electroless plating until a continuous shell is formed

e TEM micrograph of the dielectric core-metal shell nanoparticlesf Dark-field optical micrograph of the dielectric core-metal shellnanoparticles

a

1 m

b

500 nm

Fig. 4a Scanning electron micrograph image of electron beam lithographyarrays. Cylindrical nanoparticles with a diameter of 200nm and

heights of 40 nm Ag in square arrangement with a lattice spacing of350 nm. Reproduced with permission from [13]. Copyright 2003, The

American Chemical Society

b Atomic force microscopy micrograph of the nanosphere lithographyarrays. Triangular nanoparticles with perpendicular bisectors of

100nm and heights of 50 nm Ag
http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


4/21

on the dielectric properties of the surrounding environmentand that wavelength shifts in the extinction maximum ofnanoparticles can be used to detect molecule-inducedchanges surrounding the nanoparticle. As a result, thereare at least four different nanoparticle-based sensingmechanisms that enable the transduction of macromolecu-lar or chemical-binding events into optical signals based onchanges in the LSPR extinction or scattering intensity shiftsin LSPR lmax, or both. These mechanisms are: (i) resonantRayleigh scattering from nanoparticle labels in a manneranalogous to fluorescent dye labels [38, 39, 7682]; (ii)nanoparticle aggregation [83]; (iii) charge-transfer interac-tions at nanoparticle surfaces [35, 8488]; and (iv) localrefractive index changes [20, 2325, 27, 28, 84, 8993].

3.1 Solution-phase nanoparticle sensingSolution-phase nanoparticle-based sensing is a simple, yetpowerful detection modality. Because many molecules ofinterest, particularly biological molecules, are in the aqueousphase, it is desirable to have a sensitive and specificdetection system that is homogenous with the phase of thetarget molecule, thereby decreasing the need for extendedsample preparation. Aggregation-based detection has be-come a mainstay in the clinical community since the

development of the latex agglutination test, (LAT) in 1956[94]. In the LAT and similar tests, biomolecular-specificantibodies are conjugated to latex microspheres which,when mixed with a solution (e.g. blood or urine) containingthe target antigen, cause the latex spheres to form visibleaggregates. Whereas LATs are effective and quite rapid(15min to 1 h), they are inherently insensitive, relying onhigh concentrations of analytes and on the human eye as adetector.

In contrast, for solution-phase LSPR-based sensing,signal transduction depends on the sensitivity of the surfaceplasmon to interparticle coupling. When multiple particlesin solution that support a localised surface plasmon are inclose proximity (i.e. interparticle spacings less than thenanoparticle diameter), they are able to interact electro-magnetically through a dipole coupling mechanism. Thisbroadens and red shifts the LSPR, and small clusters ofparticles possess LSPR properties similar to those of alarger single particle. Two methods of detection readily lendthemselves to monitoring these changes in the position ofthe LSPR: (i) UV-visible (UV-vis) extinction (absorptionplus scattering); and (ii) resonant Rayleigh scatteringspectroscopy.

Several papers have been published on a gold nanopar-ticle-based UV-vis technique for the detection of DNA.This colourimetric detection method is based on the changein absorbance spectra (i.e. colour) as particles are brought

together by the hybridisation of complementary DNAstrands [95, 96]. The limits of detection (LOD) reported arein the range of tens of femtomoles of target oligonucleotide.These nanoparticle aggregation assays represent a 100-foldincrease in sensitivity over conventional fluorescence-basedassays [77]. Recently, gold nanoshells, that is, silica beadswith a thin gold coating, have been used to detect antigensin whole blood [22]. In these studies, gold nanoshells werefunctionalised with a specific immunoglobulin. The nano-shells were designed so that they exhibit plasmon resonancesin the near-infrared between the water absorption band andthe absorption of hemoglobin. Upon addition of thenanoshells to whole blood that contains the appropriateantigen, the plasmon resonance broadens, the intensity

decreases, and a slight red shift in the plasmon resonanceoccurs. This immunoassay, which occurs in less than30 min, is capable of detecting picogram/millilitre quantities

of antigens. Sensing based on resonant Rayleigh scatteringdetection is a potentially more powerful, yet less wellpublished, method than the colourimetric technique men-tioned above [38, 97, 98]. Figure 5 demonstrates a simplesensing experiment, wherein the observed scattering colourchanges as gold nanoparticles are drawn together byprotein-protein interactions [99]. The disadvantages of thistype of experiment is that the aggregation of the particles,necessary to induce interparticle coupling and generate asignal, is often irreversible, difficult to quantitate, and canlead to sufficient aggregation of the nanoparticles that theysettle out of solution.

3.2 Nanoparticle labellingWhereas the application of noble metal nanostructures asoptical data storage elements is still under development,companies have already been formed to exploit noble metalnanostructures as biological labels. Nanoscale cylinderswith stripes of different metals have been used asimmunoassay tags; readout mechanisms have been basedon both the extinction [98] and reflectivity [100]. In anotherexample, Siiman and Burshteyn [101] have applied thincoatings of Ag or Au to latex microspheres that then act asflow cytometry labels. The noble metal island films make it

easy to identify different subpopulations of white blood cellswithout significantly changing the beads flow character-istics. When the labelled strand of DNA is exposed to acomplimentary single strand of DNA immobilised on aglass substrate, the Au nanoparticle is bound to the surface.A flatbed scanner can be used to identify DNA matches.

Noble metal nanoparticles can go beyond acting aslabels; recent advances show that changes in nanoparticleoptical properties can act as the signal transductionmechanism in chemosensing and biosensing events. Haeset al. [26] and Malinsky et al. [84] have exploited the extremeLSPR sensitivity of NSL-fabricated nanoparticles tochanges in local refractive index in order to sense smallmolecules, amino acids, proteins, and antibodies. The

LSPR shifts systematically to lower energies as the localdielectric constant increases; accordingly, chemisorption ofalkanethiols of increasing length cause a systematic red-shift

a b c

Fig. 5 Time-lapse photographs of the rapid colourimetric testa 60 nm gold conjugatesb Stirring a 1:1 solution of IgG/gold conjugates and anti-human IgG/gold conjugates after approximately 3 min of stirring

c The same solution as (Fig. 5b), approximately 10 min later. Note the

readily observed change in colour from green to orange, which iscaused by interparticle coupling
http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


5/21

in the LSPR (3 nm/methylene unit) [84]. These nanoparticlesensors take advantage of tunable noble metal nanostruc-ture optical properties in order to sensitively and selectivelydetect trace levels of the target molecule. These experimentsare further detailed below.

3.3 Sensing on surfacesA possible method for circumventing the difficulty ofirreversible aggregate formation is to synthesise nanoparti-cles bound to substrates. Using this method, gold or silvernanoparticles can be chemically attached in a randomfashion to a transparent substrate to detect proteins [32, 89].In this format, signal transduction depends on changes inthe nanoparticles dielectric environment induced by solventor target molecules (not via nanoparticle coupling). Usingthis chip-based approach, a solvent refractive indexsensitivity of 76.4 nm/RIU has been found and a detectionof 16nM streptavidin can be detected [32, 89]. Thisapproach has many advantages including: (i) a simplefabrication technique that can be performed in most labs;(ii) real-time biomolecule detection using UV-vis spectro-scopy; and (iii) a chip-based design that allows formultiplexed analysis.

3.4 Sensing with nanoparticle arraysThe development of biosensors for the diagnosis andmonitoring of diseases, drug discovery, proteomics, and theenvironmental detection of biological agents is an extremelysignificant problem [102]. Fundamentally, a biosensor isderived from the coupling of a ligand-receptor bindingreaction [103] to a signal transducer. Much biosensorresearch has been devoted to the evaluation of the relativemerits of various signal transduction methods includingoptical [104, 105], radioactive [106, 107], electrochemical[108, 109], piezoelectric [110, 111] magnetic [112, 113],micromechanical [114, 115] and mass spectrometric [116,117]. The development of large-scale biosensor arrayscomposed of highly miniaturised signal transducer elementsthat enable the real-time, parallel monitoring of multiplespecies is an important driving force in biosensor research.Recently, several research groups have begun to explorealternative strategies for the development of opticalbiosensors and chemosensors based on the extraordinaryoptical properties of noble metal nanoparticles. Recently, itwas demonstrated that nanoscale chemosensors andbiosensors can be realised through shifts in the LSPR lmaxof triangular silver nanoparticles. These wavelength shiftsare caused by adsorbate-induced local refractive indexchanges in competition with charge-transfer interactions atthe nanoparticle surface. In this review, a detailed study ispresented demonstrating that triangular silver nanoparticles

fabricated by NSL function as extremely sensitive andselective nanoscale affinity chemosensors and biosensors. Itwill be shown that these nanoscale biosensors based onLSPR spectroscopy operate in a manner totally analogousto propagating surface plasmon resonance (SPR) sensors bytransducing small changes in refractive index near the noblemetal surface into a measurable wavelength shift response.

3.5 SPR sensors and their relationship toLSPR sensorsThe potential of propagating SPR biosensors was recog-nised in early 1980s by Liedberg et al. [118], who were ableto detect immunoglobulin antibodies by observing thechange in critical angle when the antibodies bind selectively

on a Au film. Propagating surface plasmons are evanescentelectromagnetic waves that travel along the flat smoothmetal-dielectric interfaces and arise from oscillations of the

conduction electrons in the metal [119]. SPR sensors detectthe local refractive index changes that occur when the targetanalyte binds to the metal film or nanoparticles. Surfacerefractive index sensors have an inherent advantage overoptical biosensors that require a chromophoric group orother label to transduce the binding event, as in the labellingexperiments described above.

It was realised that the sensor transduction mechanism ofthis LSPR-based nanosensor is analogous to that of SPRsensors (Table 1). Since their original discovery, SPRchanges have been used in refractive-index-based sensingto detect analyte binding at or near a metal surface and hasbeen widely used to monitor a broad range of analyte-surface binding interactions including the adsorption ofsmall molecules [120122], ligand-receptor binding [123126], protein adsorption on self-assembled monolayers[127129], antibody-antigen binding [130], DNA andRNA hybridisation [131134] and protein-DNA interac-tions [135].

Important differences to appreciate between the SPR andLSPR sensors are the comparative refractive indexsensitivities and the characteristic electromagnetic fielddecay lengths. SPR sensors exhibit large refractive indexsensitivities (B2 106nm/RIU) [120]. For this reason, the

SPR response is often reported as a change in refractiveindex units. The LSPR nanosensor, on the other hand, hasa modest refractive index sensitivity (B2 102nm/RIU)[84]. Given that this number is four orders of magnitudesmaller for the LSPR nanosensor in comparison to the SPRsensor, initial assumptions were made that the LSPRnanosensor would be 10 000 times less sensitive than theSPR sensor. This, however, is not the case. In fact, the twosensors are very competitive in their sensitivities. The short(and tunable) characteristic electromagnetic field decaylength, ld, provides the LSPR nanosensor with its enhancedsensitivity [26, 27]. These LSPR nanosensor results indicatethat the decay length, ld, is B515nm or B13% of thelights wavelength and depends on the size, shape, andcomposition of the nanoparticles. This differs greatly fromthe 200300 nm decay length or B1525% of the lightswavelength for the SPR sensor [120]. Also, the smallestfootprint of the SPR and LSPR sensors differ. In practice,SPR sensors require at least a 10 10mm area for sensingexperiments. For LSPR sensing, this spot size can beminimised to a large number of individual sensing elements(1 1010 nanoparticles for a 2 mm spot size, nanospherediameter 400 nm) down to a single nanoparticle (with anin-plane width of B20nm) using single nanoparticlemeasurement techniques [81]. The nanoparticle approachcan deliver the same information as the SPR sensor, therebyminimising its effective pixel size to the sub-100nm regime.

Because of the lower refractive index sensitivity, the LSPRnanosensor requires no temperature control whereas theSPR sensor (with a large refractive index sensitivity)requires temperature control. The final and most dramaticdifference between the LSPR and SPR sensors is cost.Commercialised SPR instruments can vary between$150000$300000, whereas the prototype and portableLSPR system costs less than $5000.

There is, however, a unifying relationship between thesetwo seemingly different sensors. Both sensors overallresponse can be described using the following equation[120]:

Dlmax mDn1exp2d=ld 3

where Dlmax is the wavelength shift response, m is therefractive index sensitivity, Dn is the change in refractiveindex induced by an adsorbate, d is the effective adsorbate
http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


6/21

layer thickness, and ld is the characteristic electromagneticfield decay length. It is important to note that for planarSPR sensors, this equation quantitatively predicts anadsorbates effect on the sensor. When applied to the LSPRnanosensor, this exponential equation approximates theresponse for adsorbate layers but does provide a fullyquantitative explanation of its response [26, 27]. Similar tothe SPR sensor, the LSPR nanosensors sensitivity wasrealised to arise from the distance dependence of theaverage induced square of the electric fields that extendfrom the nanoparticles surfaces.

3.6 Real-time binding studies of SPR andLSPR sensorsTo experimentally evaluate the relative performance of thetwo sensors, side by side comparisons were conducted. Theexact procedures followed the standard experimentalprotocols established for the two sensing modalities [145].

For the SPR sensor, Au (50nm) was evaporated ontoglass coverslips with a thin Ti underlayer (10 nm). ABIACore 1000 (Neuch#atel, Switzerland) was used for all

propagating SPR measurements reported. The mannose-functionalised substrate was incorporated into BIAcorecassettes by gluing the chip into the cassettes, andmeasurements were reported as changes in resonance angle(Dy), where 11 10000RU. The LSPR sensors werefabricated using NSL to create Ag nanoparticle arrays(50 nm). LSPR extinction measurements were taken using afibre optically-coupled spectrometer. All spectra in thisstudy were from macroscopic measurements obtained intransmission mode using unpolarised white light. A home-built flow cell was used to control the surroundingenvironment of the Ag nanoparticles and the introductionof analytes [84].

The self-assembled monolayer (SAM) was prepared by

immersing the coverslips in an ethanolic solution containingmaleimide-terminated disulfide (1mM) and 11-mercaptoun-decyl tri(ethylene glycol) disulfide (EG3, 1 mM). After 12 h,

the coverslips were rinsed with ethanol and dried under astream of nitrogen. The substrates presenting maleimide-functional groups were immersed in methanolic solutions of5 mM mannose thiol for 40 min, providing B5% sugar-immobilised surfaces. The mannose-functionalised sensorswere then exposed to 19 mM concanavalin A (Con A) inPBS buffer for 20 min (Fig. 6).

3.7 Real-time comparison of LSPR and SPRdetection of Con ATo directly compare the sensing capabilities of SPR andLSPR sensors, the real-time response of Con A binding to amannose-functionalised flat surface SPR sensor (Fig. 6a)and the LSPR Ag nanosensor (Fig. 6b) were investigated.Con A is a 104 kDa mannose-specific plant lectin comprisedof a tetramer with dimensions of 6.32, 8.69, and 8.93 nmand has four binding sites [146]. The surface bindingconstant for Con A to a mannose-functionalised SAM wasfound to be 5.6 1061/M by SPR imaging studies [147].After the baseline SPR Dy response of the mannose-functionalised Au surface in a running buffer environment

was recorded, 19mM Con A in buffer was injected. Thesensor was then flushed with buffer to remove both non-specifically bound and Con A bound as the 1:1 mannosecomplex. Figure 6a illustrates the real-time monitoring of19mM Con A by the SPR Dy shift. Similarly, the real-timeLSPR response of Con A binding to the mannose-functionalised Ag nanonsensor was also probed. After theLSPR lmax of the mannose-functionalised surface wasrecorded, 19mM Con A was injected into the flow cell, thenthe sample was flushed with PBS buffer, and during thisprocess the LSPR lmax was measured in 5 s intervals for20 min (Fig. 6b).

During the association phase, both the SPR and LSPRsensor showed a rapid response when Con A was exposed

to the surface, which indicates strong Con A binding usingtwo binding sites on the surface followed by weak 1:1binding Con A binding as well as non-specific binding.

Table 1: Comparison between SPR and LSPR sensors

Feature/characteristic SPR LSPR

Label-free detection yes [122, 124, 131, 136] yes [20, 28, 81, 84]

Distance dependence B1000nm [120] B30 nm (size tunable) [26, 27]

Refractive index sensitivity 2 106nm/RIU [105, 120, 121, 123] 2 102nm/RIU [26, 84]

Modes angle shift [137] extinction [20]

wavelength shift scattering [81, 82]

imaging imaging [81, 82]

Requires temperature control yes no

Chemical identification SPR-Raman LSPR-SERS

Field portability no yes

Commercially available yes no

Cost $150 000$300 000 $5000 (multiple particles)

$50 000 (single nanoparticle)

Spatial resolution B10 10mm [137, 138] nanoparticle [81, 82, 97]

Nonspecific binding minimal (determined by surface chem-

istry and rinsing) [105, 136, 137, 139, 140]

minimal (determined by surface

chemistry and rinsing) [20]

Real-time detection time scale 101 103s, planar diffusion time scale 101 103s,[121, 122, 139, 141, 142] radial diffusion [81]

Multiplexed capabilities yes [146, 147] [143, 144] yes-possible

Small molecule sensitivity good [121] better [26]

Microfluidics compatibility yes possible
http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


7/21

However, during the dissociation phase, when the Con A-bound SPR sensor surface was flushed with PBS buffer, theresponse decreased by 60% whereas the response decreased

only by 14% for the LSPR sensor. It is hypothesised thatthis difference is the long-range decay length of the SPRsensor electromagnetic field (ldE200nm) [120], compared tothat of the LSPR sensor (ldE56nm) [26, 27].

3.8 LSPR streptavidin and immunoglobulinsensing on nanoparticle arraysAs described above, nanoscale chemosensing and biosen-sing could be realised through shifts in the LSPR extinctionmaximum (lmax) of these triangular silver nanoparticles[20, 23, 28, 84]. Signal transduction is an effect caused byshifts in wavelength produced by adsorbate-induced localrefractive index changes in competition with charge-transferinteractions at the surfaces of nanoparticles. For theseexperiments, NSL was used to fabricate 25 and 50 nm Agnanoparticles on glass or mica. The peak-to-peak wave-length shift noise of the baseline in repetitive experimentsfrom this spectrometer is B0.5 nm. Taking the limit ofdetection as three times this value, one can conservativelyestimate the limit of detection of an assay will arise from awavelength shift of 1.5 nm.

3.9 Streptavidin sensing using LSPRspectroscopyThe well-studied biotin-streptavidin system with its extre-mely high binding affinity (KaB10

131/M) is chosen toillustrate the ultra-sensitive attributes of these LSPR-based

nanoscale affinity biosensors. The biotin-streptavidin systemhas been studied in great detail by SPR spectroscopy andserves as an excellent model system for the LSPRnanosensor [20]. Streptavidin, a tetrameric protein, canbind up to four biotinylated molecules (i.e. antibodies,inhibitors, nucleic acids, etc.) with minimal impact on itsbiological activity and therefore will provide a readypathway for extending the analyte accessibility of the LSPRnanobiosensor.

In the streptavidin detection scheme, the lmax of the Agnanoparticles were monitored during each surface functio-nalisation step (Fig. 7) [20]. First, the LSPR lmax of thebare Ag nanoparticles was measured to be 561.4nm (step Ain Fig. 7). To ensure a well-ordered SAM on the Ag

nanoparticles, the sample was incubated in a 3:1 Octa-nethiol : mercaptoundecanoic acid (OT/MUA) for 24 h.The LSPR lmax (step B in Fig. 7) was measured to be

598.6 nm, a 38nm red shift. Next, biotin was covalentlyattached via amide bond formation to carboxylated surfacesites (step C in Fig. 7). The LSPR lmax after biotinattachment (step C in Fig. 7) was measured to be

609.6 nm, corresponding to an additional +11nm shift.Exposure to 100 nM streptavidin, resulted in LSPRlmax 636.6 nm (step D in Fig. 7) corresponding to anadditional +27 nm shift.

3.10 Anti-biotin sensing using LSPRspectroscopyA field of particular interest is the study of the interactionbetween antigens and antibodies. For these reasons we havechosen to focus the present LSPR nanobiosensor study onthe prototypical immunoassay involving biotin and anti-biotin, an IgG antibody. Ag nanotriangles were synthesisedusing NSL to develop a LSPR biosensor that monitors theinteraction between a biotinylated surface and free anti-

biotin in solution [28]. The importance of this study is that itdemonstrates the feasibility of LSPR biosensing with abiological couple whose binding affinity is significantly

0.3

buffer buffer bufferCon A

0.2

0.1

0

0 200 400 600 800 400 800 1200

time, s

b

time, s

,

deg

max,nm

a

0

2

4

6

8

bufferCon A

experimental point

line drawn toguide the eye

Fig. 6 The real-time response of a sugar-functionalised sensor as 19 mM of Con A was injected in the cell following buffer injectiona Mannose-functionalised SPR sensorb Mannose-functionalised LSPR nanosensor

0.28

0.24

0.20

0.16

300 500

561.4

598.6 609.6

636.6A

B C

D

700

wavelength, nm

extinction

Fig. 7 LSPR spectra of each step in the surface modification ofNSL-derived Ag nanoparticles to form a biotinylated Ag nanobio-sensor and the specific binding of streptavidin. Step A: Ag

nanoparticles before chemical modification lmax 561.4nm; StepB: Ag nanoparticles after modification with 1 mM 1:3 11-MUA: I-OT, lmax 598.6nm; Step C: Ag nanoparticles after modificationwith 1 mM biotin lmax 609.6nm. Step D: Ag nanoparticles aftermodification with 100 nm streptavidin lmax 636.8nm. All extinc-tion measurements were collected in a N2 environment. (Reproducedwith permission from [20]. Copyright 2002, The American ChemicalSociety)
http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


8/21

lower (1.9 106

4.98 108

1/M) than in the biotin/streptavidin model [20].

Each step of the functionalisation of the samples wasmonitored using UV-vis spectroscopy, as shown in Fig. 8[28]. After a 24 h incubation in SAM, the LSPR extinctionwavelength of the Ag nanoparticles was measured to be670.3nm (step A in Fig. 8). The LSPR wavelength shift dueto biotin binding was measured to be +12.7 nm, resultingin a LSPR extinction wavelength of 683.0 nm (step B inFig. 8). At this stage, the nanosensor was ready to detect thespecific binding of anti-biotin. Incubation in 700 nM anti-biotin resulted in a LSPR wavelength shift of +42.6 nm,giving a lmax of 725.6 nm (step C in Fig. 8).

3.11 Concentration-dependent responseIt was revealed that if the analyte solution was decreased,the LSPR wavelength shift response also decreased.Subsequently, the full response of the sensor was deter-mined over a wide concentration range. For this reason,variable analyte concentrations were exposed to a biotiny-lated LSPR chip to test the sensitivity of the systemto different molecules. Specifically, the LSPR lmax shift,DR, against analyte concentration response curve was

measured over the concentration range 1 1015Mo[streptavidin]o1 106M and 7 1010Mo[anti-biotin]o7 106M (Fig. 9) [20, 28]. The lines seen in Fig. 9 arenot a fit to the data. Instead, the line was computed from aresponse model (a complete analysis of this model isdescribed in [28]). It was found that this response couldbe interpreted quantitatively in terms of a model involving;(i) 1:1 binding of a ligand to a multivalent receptorwith different sites but invariant affinities; and (ii) theassumption that only adsorbate-induced local refractiveindex changes were responsible for the operation of theLSPR nanosensor.

The binding curve provides three important character-istics regarding the system being studied. First, the mass anddimensions of the molecules affect the magnitude of theLSPR shift response. Comparison of the data withtheoretical expectations yielded a saturation response,DRmax 26.5 nm for streptavidin, a 60 kDa molecule, and38.0 nm for anti-biotin antibody, a 150 kDa molecule.Clearly, a larger mass density at the surface of thenanoparticle results in a larger LSPR response. Next, thesurface-confined thermodynamic binding constant Ka,surfcan be calculated from the binding curve and is estimated tobe 1 10111/M for streptavidin and 4.5 1071/M for anti-

biotin antibody. These numbers are directly correlated tothe third important characteristic of the system, the LOD.The LOD is less than 1 pM for streptavidin and 100pM foranti-biotin antibody. As predicted, the LOD of thenanobiosensor studied is lower for systems with higherbinding affinities such as for the well-studied biotin-streptavidin couple and higher for systems with lowerbinding affinities as seen in the anti-biotin antibody system.Given this information and analysis, a similar treatment canbe made for virtually any ligand receptor system. It shouldbe noted that the LOD of the system corresponds to thesmallest reliable wavelength shift response induced by agiven solution concentration. These real LODs are oftentranslated to the commonly reported surface coverage interms of molecules. For the LSPR nanosensor, the surfacecoverage detection corresponds to B25 streptavidin mole-cules per nanoparticle. In the previously described experi-ments, B3 108 nanoparticles were probed (nanoparticledensityB1 1010 nanoparticles per square centimetre witha spot size with a 1 mm radius), and this corresponds to thedetection of a 75 108 streptavidin molecules. A clearmethod to decrease the number of molecules detectedwould be to decrease the number of nanoparticles probed.

500 600

0.08

0.10

0.12

0.14

0.16

670.3

A683.0

725.6

B

C

700 800

wavelength, nm

extinction

Fig. 8 LSPR spectra for each step of the preparation of the Agnanobiosensor, and the specific binding of anti-biotin to biotin. StepA: Ag nanoparticles after modification with 1 mM 3:1 1-OT/11-MUA, lmax 670.3 nm; Step B: Ag nanoparticles after modificationwith 1 mM biotin, lmax 683.0nm; and Step C: Ag nanoparticlesafter modification with 700nM anti-biotin, lmax 725.6 nm. Allspectra were collected in a N2 environment. Reproduced with

permission from [28]. Copyright 2003, The American ChemicalSociety

40

30

20

10

0

LSPR

response,nm

1016 1015 1014 1013 1012 1011 1010 109 108 107 106 105 104 103

concentration, M

streptavidin

Rmax = 26.5 nm

Ka,surf = 11011 1/M

LOD < 1 pManti-biotin

Rmax = 38.0 nm

Ka, surf = 4.5107 1/M

LOD < 700 pM

Fig. 9 The specific binding of streptavidin (left) and anti-biotin (right) to a biotinylated Ag nanobiosensor is shown in the response curves. Allmeasurements were collected in a N2 environment. The solid line is the calculated value of the nanosensors response. Reproduced with permission

from [20]. Copyright 2002, The American Chemical Society. Reproduced with permission from [28]. Copyright 2003, The American ChemicalSociety
http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


9/21

This has been recently demonstrated as described in greaterdetail in a later Section.

3.12 Detection of disease markers using theLSPR sensor chipAlzheimers disease is the leading cause of dementia inpeople over age 65 and affects an estimated 4000 000Americans [148]. Although first characterised almost 100years ago by Alois Alzheimer, who found brain lesions nowcalled plaques and tangles in the brain of a middle-agedwoman who died with dementia in her early fifties [149], themolecular cause of the disease is not understood; and anaccurate diagnostic test has yet to be developed. However,two inter-related theories for Alzheimers disease haveemerged that focus on the putative involvement ofneurotoxic assemblies of a small 42-amino acid peptideknown as amyloid beta (Ab) [150, 151]. The widelyinvestigated amyloid cascade hypothesis suggests that theamyloid plaques cause neuronal degeneration and, conse-quently, memory loss and further progressive dementia. Inthis theory, the Ab protein monomers, present in all normalindividuals, do not exhibit toxicity until they assemble intoamyloid fibrils [152]. The other toxins are known as Ab-derived diffusible ligands (ADDLs). ADDLs are small,

globular, and readily soluble, 3-24mers of the Ab monomer[153] and are potent and selective central nervous systemneurotoxins which possibly inhibit mechanisms of synapticinformation storage with great rapidity [153]. ADDLs nowhave been confirmed to be greatly elevated in autopsiedbrains of Alzheimers disease subjects [154]. An ultra-sensitive method for ADDLs/anti-ADDLs antibody detec-tion potentially could emerge from LSPR nanosensortechnology, providing an opportunity to develop the firstclinical lab diagnostic for Alzheimers disease. Preliminaryresults indicate that the LSPR nanosensor can be used toaid in the diagnosis of Alzheimers disease [155, 156].

3.13 Sensing with single nanoparticlesThe extension of the LSPR sensing technique to the singlenanoparticle limit provides several improvements overexisting array- or cluster-based techniques [157, 158]. First,absolute detection limits are dramatically reduced. Thesurface area of chemically prepared Ag nanoparticles istypically less than 20000nm2, which requires that acomplete monolayer of adsorbate must constitute fewerthan approximately 100 zeptomol. The formation ofalkanethiol monolayers on Ag nanoparticles can result ina LSPR lmax shift of greater than 40 nm, a change that isover 100 times larger than the resolution of convention UV-visible spectrometers [159]. This suggests that the limit ofdetection for single nanoparticle-based LSPR sensing will

be well below 1000 molecules for small molecule adsorbates.For larger molecules such as antibodies and proteins thatresult a greater change in the local dielectric environmentupon surface adsorption, the single molecule detection limitmay be achievable [28]. Second, the extreme sensitivity ofsingle nanoparticle sensors dictates that only very smallsample volumes (i.e. attolitres) are necessary to induce ameasurable response. This characteristic could eliminate theneed for analyte amplification techniques (e.g., the poly-merase chain reaction) required by other analyticalmethods. Third, single nanoparticle sensing platforms arereadily applicable to multiplexed detection schemes. Bycontrolling the size, shape, and chemical modification ofindividual nanoparticles, multiple sensing platforms can be

generated in which each unique nanoparticle can bedistinguished from the others based on the spectral locationof its LSPR [81, 157, 158]. Several of these unique

nanoparticles may then be incorporated into a one device,allowing for the rapid, simultaneous detection of thousandsof different chemical or biological species.

The key to exploiting single nanoparticles as sensingplatforms is developing a technique to monitor the LSPR ofindividual nanoparticles with a reasonable signal-to-noiseratio [81]. UV-visible absorption spectroscopy does notprovide a practical means of accomplishing this task. Evenunder the most favourable experimental conditions, theabsorbance of a single nanoparticle is very close to the shotnoise-governed limit of detection. Instead, resonant Ray-leigh scattering spectroscopy is the most straightforwardmeans of characterising the optical properties of individualmetallic nanoparticles.

3.14 Experimental procedure for singlenanoparticlesColloidal Ag nanoparticles were prepared by reducing silvernitrate with sodium citrate in aqueous solution according tothe procedure referenced above [40]. Immobilised particleson cover slips were inserted into a flow cell and wereexposed to various dielectric environments or molecular

adsorbates [81]. Prior to all experiments, the nanoparticlesin the flow cell were repeatedly rinsed with methanol anddried under nitrogen. All optical measurements wereperformed using an inverted dark-field microscopeequipped with an imaging spectrograph [81, 82]. Theapparatus used in these experiments is shown in Fig. 10.

3.15 Single nanoparticle refractive indexsensitivityThe local refractive index sensitivity of the LSPR of a singleAg nanoparticle was measured by recording the resonantRayleigh scattering spectrum of the nanoparticle as it wasexposed to various solvent environments inside the flow cell.As illustrated in Fig. 11, the LSPR lmax systematically shiftsto longer wavelength as the solvent RIU is increased. Linearregression analysis for this nanoparticle yielded a refractiveindex sensitivity of 203.1 nm RIU1. The refractive indexsensitivity of several individual Ag nanoparticles wasmeasured and typical values were determined to be 170

235nm RIU1 [81]. These are similar to the values obtainedfrom experiments utilising arrays of NSL fabricatedtriangular nanoparticles [26, 84].

condenser

flow cell

video

imaging

spectrograph

CCD

NA = 0.95

outletinlet

NA = 0.5 1.3

Ag nanoparticles are immobilised

on the inside of the bottom window

a b

Fig. 10a Diagram of the instrumentation used for single nanoparticlespectroscopy

b Close-up of the flow cell to show illumination and collectiongeometry
http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


10/21

3.16 Streptavidin sensing with singlenanoparticles

In nanoparticle array sensing, it has been demonstrated thatit is possible to detecto100 streptavidin molecules pernanoparticle (in a solution concentration of 1 pM). Thebiology community would like to reduce the amount ofbiological sample needed for an assay without amplifica-tion. For this reason, experiments have been conducted todetect streptavidin on single Ag nanoparticles [82]. Afterfunctionalisation with a capture biomolecule, the LSPR ofan individual nanoparticle was measured to be 508.0 nm(curve (i) in Fig. 12). Next, 10 nM streptavidin was injectedinto the flow cell, and the lmax of the nanoparticle wasmeasured at 520.7 nm, curve (ii) in Fig. 12. This +12.7 nmshift is estimated to arise from the detection of less than 700streptavidin molecules. It is hypothesised that just as in the

array format, as the streptavidin concentration decreases, afewer number of streptavidin molecules will bind to thesurface thereby causing smaller wavelength shifts.

3.17 Theoretical explanation of LSPRsensingTo model the experiment results presented above, thediscrete dipole approximation (DDA) method [160, 161] afinite element-based approach to solving Maxwells equa-tions for light interacting with an arbitrary shape/composi-tion nanoparticle was used to calculate the plasmonwavelength in the presence or absence of an adsorbate witha wavelength-dependent refractive index layer thickness.Bare silver nanoparticles with a truncated tetrahedral shapewere first constructed from cubic elements and one to twolayers of the adsorbate were added to the exposed surfacesof the nanoparticle to define the presence of the adsorbate.All calculations refer to silver nanoparticles with a dielectricconstant taken from Lynch and Hunter [162].

In this treatment the dielectric constant is taken to be alocal function, as there is no capability for a non-localdescription within the DDA approach. There have beenseveral earlier studies in which the DDA method has beencalibrated by comparison with experiment for truncatedtetrahedral particles, including studies of external dielectriceffects and substrate effects [11, 71, 163] and based on this itis expected that DDA analysis will provide a usefulqualitative description of the results. In particular, it hasbeen demonstrated that the plasmon resonance shift close tothe nanoparticle surface is dominated by hot spots while

that farther away arises from colder regions around thenanoparticle surface. By comparing the overall maximumLSPR shifts of the nanoparticles, it was shown thatincreasing the aspect ratio of Ag nanotriangles produceslarger plasmon resonances shifts and shorter-ranged inter-actions.

This behaviour can be easily compared to lightning rods(invented by Ben Franklin) which are placed at the top of abuilding, and are designed to provide a location wherelightning will strike and from which the energy is carriedharmlessly to the ground. A needle-like metal tip is used in alightning rod, since this leads to high electric fields justabove the tip when the building is charged relative to theatmosphere above it, providing a more likely location for

lightning to strike. A similar description can be used todescribe lights interaction with tips of nanoscale silver andgold nanotriangles [26, 27]. This property is best displayed

100

80

60

40

20

0

450 500 550 600 650 700

Wavelength, nm

intensity

588.0

510.2 574.2 611.9

600.8

640

600

560

520

480

1.0 1.2 1.4 1.6

refractive index

max = 203.1Rl+306.5

max,

nm

a

b

Fig. 11a Resonant Rayleigh scattering spectra of an individual silvernanoparticle in various dielectric environments (N2, methanol,

1-propanol, chloroform, and benzene)

b A plot depicting the linear relationship between the solvent refractiveindex and the nanoparticles lmax. Reproduced with permission from

[81]. Copyright 2003, The American Chemical Society

1200

800

400

0

450 500 550 600

s

catteringintensity

wavelength, nm

(ii)

(i)

Fig. 12 Individual Ag nanoparticle sensor before and afterexposure to 10 nM streptavidin. All measurements were collectedin a nitrogen environment(i) Biotinylated Ag nanoparticle, lmax 508.0 nm.(ii) After streptavidin incubation, lmax 520.7 nm.
http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


11/21

in Fig. 13. This Figure was generated using the DDAmethod. This leads to fields around the nanoparticles whosemagnitudes relative to the incident light are plotted in theFigure. The results show that the electromagnetic fields aregreatly amplified (up to 65 000 times) in the region near thenanoparticle tips, leading to enhanced sensitivity tomolecules that might be located there. This informationcan be used as a basis for constructing nanoparticlestructures and understanding the sensing mechanism forsurface-enhanced spectroscopies.

4 Biological applications of surface-enhancedRaman scattering

Inelastic photon scattering was first witnessed by Ramanand Krishnan in 1928 [164] a process later to bear the nameRaman scattering and to garner the authors a Nobel Prizein 1930. In the original Nature paper [164] the inelasticscattering effect was recognised by its feebleness incomparison with ordinary (Rayleigh) scattering, and by itspolarisation. Raman was able to observe this phenomenonby focusing a beam of sunlight through a telescope and agreen filter, and observing the wavelength-shifted yellowlight scattered from chloroform. The development of thelaser, as a source of high intensity and nearly monochro-matic light, enabled the more practical detection of the oneinelastically scattered photon from the 108 elasticallyscattered photons [165]. Further developments in the fieldof photonics, such as high quality interference and notch

filters, and ultra-sensitive photodetectors and CCD cam-eras, have enabled the routine use of Raman spectroscopyas a complement to conventional vibrational techniquessuch as near-infrared absorbance spectroscopy. As avibrational spectroscopy, Raman scattering generates in-formation rich spectra unique to a given molecule [165]. Thepotential to use the unique molecular information toperform optical fingerprinting capable of distinguishingalmost any analyte by its spectroscopic signature alone hasinterested analytical chemists and biologists for years.Researchers have striven to overcome the low scatteringcross-sections inherent in normal Raman spectroscopy(NRS) which is a process approximately a million-fold lessefficient than fluorescence, and have developed a number of

impressive biological applications. For example, NRS hasbeen used in the detection of glucose, in proteomics, thediagnosis of cancer and cardiovascular health, the identifi-

cation of small molecule drugs, and so forth. Reviews ofthese applications have recently been written by Cottonet al. [166], Nabiev et al. [167], and Petry et al. [168].However, the low signal intensities and complications withspectral identification caused by the generation of a spectralresponse all species in a complex mixture and theconcomitant confusion generated by closely spaced oroverlapping Raman bands unfortunately hinder applicationof NRS. SERS can increase the scattering cross-section of amolecule in proximity to a nanoscale roughened surface by106 to 1014 times, potentially mitigating the two aforemen-tioned disadvantages of NRS [169, 170].

SERS is a plasmonic phenomenon whereby specificmolecules at or near a nanoscale roughed metal surfaceundergo a dramatic increase in the intensity of the observedinelastically scattered light. Since the discovery of SERS byJeanmaire and Van Duyne [171] the exact mechanism ofenhancement has remained controversial [172, 173]. How-ever, it is generally agreed that two mechanisms contributeto the overall enhancement, a chemical effect (CHEM, a)and an electromagnetic (EM, E) enhancement. The intensityof Raman scattering is proportional to the square of P, theelectric field-induced dipole moment. The P term is relatedto the molecular polarisablity (a) and the electric field (E)

experienced by the molecule. These relationships aremathematically expressed thus:

G/ P2 4

and

P aE 5

The electromagnetic effect gives rise to an enhanced Ramanyield for a given molecular polarisability, whereas thechemical effect refers to a change in the polarisability itself.The theory of chemical enhancement (a) was invoked toaccount for a discrepancy between the enhancementpredicted the electrical model (E) (B104) and the empiri-cally determined enhancement (B106). It is thought thatactive sites, either crystal defects or adatoms on the surfaceof the metal surface undergo a charge-transfer processenabling the CHEM (a) enhancement [170]. The depen-dence of optimal SERS excitation frequency on theelectrode potential suggests that new electronic states mayarise upon adsorption of the analyte, which leads toresonant intermediate states for Raman scattering. TheCHEM enhancement has been investigated, particularly theeffect of halide ions on increasing SERS [173176]. Therecent discovery of single molecule SERS (SMSERS) byNie and Emory [177] and Kneipp et al. [178] in 1997, hasnecessitated further refinements to the EM mechanism.

It is well established that the EM mechanism arises from

optical excitation of surface plasmon resonances in smallmetal particles, which leads to a significant increase in theelectromagnetic field strength at the particle surface.Initially, Moskovits [179] developed a resonant fractaltheory to explain surface Raman enhancement in exten-sively aggregated colloids. For single particles and smallaggregates, however, the case can be dramatically differentthan for an ensemble average of particles. Electromagneticenhancement is attributable to the formation a LSPR, asdescribed above. This polarises the particle with anelectromagnetic field significantly higher than the appliedfield. Recent theoretical modelling, has greatly extended thenature of the plasmonic material (e.g. particle morphology,spacing, and coupling) beyond the simple spherical model

developed by Mie, and is more useful in describing actualexperimental conditions. The intensified field at rough orhighly-curved surfaces enhances both the incident laser light

0

65 000

x

0

2

4

6

8

10

z

y

100

50

0

50

100

0

5

10

15

20

25

30

35

Fig. 13 Local E-field (plotted as contours of log 7E72) for aAg nanoparticle (in-plane width 100nm, out-of-plane height 30 nm). Reproduced with permission from [26]. Copyright

2004, The American Chemical Society
http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


12/21

and the scattered Raman light. Due to the dependence ofthe SERS signal intensity on the fourth-power of the EMfield, the signal drops by a factor of ten for each nanometreseparation between the molecule and the surface [180182].This electromagnetic enhancement can be qualitativelyconceptualised as an antenna effect, gathering andstrengthening local electric fields.

4.1 Intrinsic SERS of target moleculesSERS provides not only enhanced signal intensities but alsospecies specific spectra, so it is therefore unsurprising thatmany SERS experiments have been conducted that attemptto directly address molecules of biological interest and theirchemical interactions. The first such experiment wasconducted in 1980 by Cotton et al. on cytochrome c andhemoglobin, molecules with strong electronic absorbancesin the visible. These intense absorbances were matched withthe laser excitation wavelength, lex, to eke out a few ordersof magnitude extra enhancement by exploiting surface-enhanced resonance Raman spectroscopy (SERRS). Theresonance effect has continued to make hemoglobin andcytochrome c popular proteins for study, particularly sincethe advent of SMSERS [183, 184]. In the same year asCottons pioneering experiments, Billman and Otto [185]

also demonstrated the successful detection of cyanide onroughened silver electrodes [188]. Cyanide, while not abiomolecule as such, is nonetheless of great interest in bothbiological, environmental, and defence circles in thedetection of cyanides and similar molecules which have adirect impact on biological systems. As another example,SERS has been used on the bio-terror front for thedetection of anthrax [186]. SERS has been demonstrated tosuccessfully detect dipicolinic acid, the principle chemicalcomponent ofbacillus anthracis spores (15% wt) [187]. Sincethe 1980s, the number of biological applications of SERS(and NRS) has skyrocketed [166168, 188194]. Indeedsignificant strides have been made in the Raman spectro-scopy of proteins and peptides using NRS, and progresscontinues as the use of SERS expands. SERS techniqueshave been applied to an impressive variety of systems [193,195207]. Initially, most SERS experiments were conductedon electrochemically roughened surfaces. However, sincethe discovery of SMSERS, researchers have increasinglytended to use colloidal systems as they are easy to preparechemically and to model theoretically. Exceptions arecommon, however, Maxwell et al. [208] employed size-selected nanoparticle films to study amoxicillin and adenine,and the metal film-over-nanosphere (MFON) surfacesremain popular for their mechanical stability [209].Remarkably, Kneipp et al. [210] have been able to obtainSERS spectra from gold nanoparticles sited within single

living cells. Recently, Jarvis and Goodacre [211] have usedSERS on aggregated silver colloids in combination withchemometric techniques to classify a collection of clinicalbacterial isolates. SERS can also be used to study molecularinteractions and dynamics, such a protein interactions [212214] and pharmacokinetics [162, 215, 216]. Naturally, bothSERS and NRS lend themselves to sensitive detectionmodalities for separation science [216219].

4.2 SMSERSDespite its tremendous power and potential the capriciousnature of SMSERS has resulted in only a few publicationson its biological application. In 1998, Kneipp et al. [220]were the first to successfully obtain SERS spectra from

single nucleoside bases building on their experience with notonly SMSRS [178] but also SERS detection of DNA [221,222]. Shortley afterwards, Keller et al. [223] and Emory and

Keller [224] brought SMSERS to bear on nucleosides.Recently, Xu et al. [225] have returned to the use ofhemoglobin as the SERS analyte, but at the single moleculelevel. This 1999 study was augmented a year later by anexcellent manuscript covering the contribution of the EMmechanism of SMSERS [226]. Since then, these researchershave combined SMSERS with fluorescence correlationspectroscopy to study HRP: a highly useful component inthe bioanalytical toolkit [227]. SMSERS on myoglobin hasreported by Bizzarri and Cannistraro [228] while Habuchi

et al.[229] obtained SERRS spectra from the green

fluorescent protein.

4.3 Limitations to intrinsic SERSThere are two considerable difficulties in applying SERSdirectly to biological problems. First, without resonanceenhancement (as in the case of cytochrome c) it is still oftendifficult to obtain quality Raman spectra of biologicallyinteresting molecules (e.g. drugs, protein, nucleic acids, etc.),particularly at the relatively low concentrations oftennecessitated by the expense or insolubility of the analyte.Even although SERS is potentially sensitive for singlemolecule spectroscopy, not all molecules, such as water, areefficient SERS scatterers. Also many molecules neither

adsorb nor bind to the SERS active surface, making themdifficult to detect, given the consequent lack of chemical(CHEM) enhancement and greatly diminished electromag-netic enhancement (EM). Approximately one-order ofmagnitude in SERS intensity is lost per nanometre distancethe target molecule is from the enhancing substrate, makingthe detection of certain molecules, such as some ionicspecies, carbohydrates, etc., particularly problematic. Sec-ond, large biomacromolecules and numerous small mole-cules of biological importance have complex Ramanspectra, and often have a number of overlapping peaksdue to structural similarities in the molecule. For exampleall biomacromolecules are built from a limited set ofbuilding blocks; the approximately 20 amino acids forproteins, and eight nucleic acids for DNA and RNA.Teasing out specific information about a given biomacro-molecule and its composition, conformation, or interactionsis a particularly daunting, but not impossible task. Theproblem becomes particularly difficult in complex mixtures,which are the norm for biologically interesting systems suchas environmental samples from contaminated waste sites,and the protein jungle of the cytosol.

4.4 Extrinsic SERS labellingSeveral approaches have been reported to address thedifficulties with direct SERS-based detection of biologicallyrelevant targets. An attractive option is to indirectly detect

the presence of interaction of a complex and/or weaklyscattering molecule by labelling it with a small moleculewith high intensity, well defined, and characteristic spectralfeatures [230, 231]. The extrinsic labelling schemes describedbelow are all variations on the ELISA (enzyme-linkedimmunosorbent assay). A graphical summary of the variousextrinsic SERS methods is shown in Fig. 14. In a standardELISA, molecular binding events are detected usingantibodies (Abs) possessing a high affinity for the moleculeof interest, and which are covalently bound to an enzyme(e.g. HRP). Once the target molecule is immobilised on asurface, the target is exposed to the detection AB/enzymecomplex. The substrate for the enzyme is then introduced,and the enzymatic reaction allowed to occur. Typically the

product of the enzymatic reaction is then detected bya change in a physical property, such a colour for HRP/OPD (ortho-phenylenediamine). The amount of target is
http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


13/21

therefore proportional to the intensity of the colourproduced after a set period of time. Dou et al. [232] werethe first to adapt an ELISA-type assay by replacingchromogenic read out with SERS detection. Vo-Dinh andco-workers deposited silver metal on the surface of DNAchips and demonstrated that SERS spectra could beobtained from dye-labelled gene probes [233235]. Usingcolloidal nanoparticles, Graham et al. [236, 237] reportedmultiplexed SERS detection of labelled polymerase chainreaction products, and Ni et al. [238] and Pham et al. [239]showed that colloidal gold with co-adsorbed organicmolecules and Ab could be used for immunoassay readout.

Improvements were realised for this method by using larger,but more optimally enhancing 63 nm gold nanoparticles,which had stronger spectral intensities and were intense

enough to permit single particle imaging [240]. Innovativeresults have also been reported by Mulvaney et al. [241] andDoering and Nie [175] who prepared silica encapsulatednanoparticles for use in multiplexed bioassays. Theseencapsulated particles are stable in the presence of highsalt (common in most biological buffers), and the silica shellprotects not only the particle from the environment, butalso the environment from the potentially toxic Ramanreporter. In a significant departure from traditional SERSsubstrates, Cao et al. [242] prepared SERS-active nano-structures by seeded growth of silver on gold anddemonstrated that these structures are useful for multi-plexed nucleic acid detection. The seeded growth methodwas extended to protein detection and labelling in [243] andcell and tissue labelling in lab [99]. Both methods use smallgold nanoparticles associated with both a reporter moleculefor signal specificity and protein to confer biological activity[244]. After binding of the bioconjugates to the target,electroless deposition is employed to develop a shell of silvermetal around the nanoparticle seed, in a process confusinglyreferred to as silver enhancement. The silver layer forms acomplex fractal surface, providing the nanoscale roughnessnecessary for efficient SERS. Whereas Cao et al. [242]synthesised reporter-labelled DNA/protein conjugates, the

method developed by Stuart et al. [99] functionalises aminereactive dye molecules directly onto existing protein goldbioconjugates. This later method not only is simpler andsubstantially less expensive, but also leads to higher reporterand protein levels which increases both the conjugatesbinding capacity and observed signal intensity.

4.5 Spectroscopic labelling of biologicalsamples with enhanced probesSince Raman scattering can be semi-quantitative, and doesnot photobleach, this method may prove useful to manybiomedical researchers, for example, to those studyingcancer. As a test bed, the technology was applied to tissuesections in a microarray format. Colon cancer/normal tissuemicroarrays were purchased from Zymed Laboratories Inc.,and subsequently deparaffinised and re-hydrated accordingto the manufacturers protocol. Primary antibody (mouseanti-carcinoembryonic antigen, CEA) conjugation wasachieved by following the manufacturers protocol, whichcould readily be automated. Secondary labelling withmalachite green/anti-mouse gold particles and subsequentsilver enhancement was performed as per standardimmunogold and silver staining practice [245]. The limitingfactor is not the spectroscopy, which is capable of singlemolecule sensitivity and large number of strongly SERS-active reporters for multiplexing, but the availability ofsuitably specific biomolecules. Fortunately, the CEA anti-

body binds to a wide range of colon cancer types, but hasminimal interaction with normal tissue. Figure 15 shows theability of the technology to discriminate between thecancerous tissues (red line) and tissues taken near (adjacentmucosa, blue line) and far from the tumor site (remotemucosa, black line). By recording the spectrum of themarking tissue, the hump was eliminated by backgroundsubtraction. Also, the silver staining itself provides a rapidmethod of determining if particles were successfully bound,and serves as a secondary method of staining tissues [246].Figure 15 shows, however, that the level of staining requiredto generate a SERS signal is quite low. To the unaided eye,there is no difference between the normal (Fig. 15b) and thecancer tissue sections (Fig. 15c). Even under microscopic

examination, correctly identifying the silver-enhanced sec-tions of the tissue can require skill and training. Spectro-scopic labelling provides a rapid and unambiguous

enzyme

OPD

azo-compound

antigen

antibody

Ag

colloid SERS

spectrum

laser

excitation

a

reporter

molecule

laser

excitation

Au

colloid

antigen

antibody

SERS

spectrum

c

Ag

colloid laser

excitation

antigen

antibody

SERS

spectrum

silica shell

d

laser

excitation

SERS

spectrum

Au

colloid

Au

colloid

antigenantibody

silver

enhancement

e

SSDNA

Agsubstrate

labelled

ssDNA

laser

excitation

SERS

spectrum

b

Fig. 14 Schematic of extrinsic SERS labeling methodsa SERS used to detect enzyme reaction product.b Detection of a labelled DNA after hybridisation to capture ssDNAbound to a silver substrate

c Extrinsic Raman Labels (ERLs) formed by the co-addition of aRaman reporter and protein to gold colloid

d Composite nanospheres with encapsulated Raman labels andprotective silica shell

e Silver-enhanced SERS detection with protein-modified gold probesacting as nucleation sites for electroless silver deposition
http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


14/21

differentiation, and the operator familiarisation time for theinstrument would be minimal.

4.6 SERS using functionalised substratesA secondary method of overcoming the difficulty in thedirect SERS detection of problematic analytes is to employa functionalised SERS surface. By coating the SERS activesubstrate with a SAM [247], hydro-gel [248] or polymer

[249] substantial gains in signal can be realised fromanalytes whose SERS spectra are otherwise difficult orimpossible to obtain. Whereas the specifics of the interac-tions between the functionalised surface and the complexmixture are inherently unique to each particular combina-tion of analyte and surface, two general mechanismsoperate simultaneously to provide the improved signal.

First, the material coating the SERS substrate can act asa barrier to undesired analytes. In this sense, the coatinglayer acts as a chemical and physical filter. For example,large species may be effectively excluded by controlling thepore size on a polymer substrate coating, or cations may berepelled by using a charged SAM bound to a SERS activesurface. Stokes et al. [249] recognised the potential of

combining polymers with SERS. The technique of SAMfunctionalisation has been largely pioneered by [203, 218,250252] who have used the selectivity of the SAM in the

analysis of bilirubins and salicalate from whole blood, anddrug compounds amongst other targets. From a biomedicaland device perspective, perhaps the most interestingapplication of this selective SAM approach has been thedetection glucose by the collaborative efforts at North-western University [253, 254].

4.7 SERS detection of glucoseThe NIH estimates on the number of diabetics in the US isstaggering. An estimated 17 000 000 people, or about 7% of

the US population suffer from diabetes. Current treatmentof diabetes consists of self-regulation of blood glucose levelsthrough frequent finger-stick monitoring. A faster, easier,automated, and less painful method for frequently measur-ing glucose levels would be of great individual, clinical, andsocietal benefit.

AgFONs with a monolayer of EG3 (Fig. 16a) wereincubated for 1 h in glucose solutions (025 mM) of clini-cally relevant concentrations. SERS spectra were thenmeasured from each sample. The chemometric method ofpartial least squares (PLS) is used when the spectrum of ananalyte of interest is embedded within a complex back-ground spectrum [255]. The result of PLS analysis is a plot

30 000

35 000

25 000

20 000

15 000

10 000

5000

countspersecond

600 800 1000

raman shift, cm1

1200 1400 1600 1800

cancer tissueadjacent mucosaremote mucosa

marker tissue (kidney)

a

b c

Fig. 15a Spectroscopic identification of cancer tissue. Spectra from canceroustissue (red), adjacent mucosa (blue), remote mucosa (46 cm distant)

(black), after secondary labelling with nanoparticle/dye conjugates and

subsequent silver enhancement. The cancerous tissue display clear and

distinct peaks indicating positive labelling with the nanoparticle

probes, whereas the control tissues show broad, featureless spectra

b Photomicrograph of cancer tissue before the silver enhancementprocess

c Photomicrograph of normal tissue after the silver enhancementprocess

600 nm

200

400

440

400

360

320

280

240

200

160

120

80

0

0

40

40 80 120 160 200 240 280 320 360 400 440

actual concentration, mg/dl

predictedconcentration,mg/dl E

E

D

D

B

B

C

C

A

A

a

b

Fig. 16a Schematic showing hypothetical glucose concentration gradientcreated by a partition layer.

b Plot of PLS predicted physiologically relevant glucose concentrationsvs against actual glucose concentrations using leave-one-out cross-

validation projected on a Clark Error grid. AgFON samples were

made (sphere diameter 390nm, metal thickness 200nm), incu-bated for 19 hs in 1 mM EG3, and dosed in glucose solution (range:

025mM) for 10min. Each SERS measurement was made while

samples were in an environmental control cell filled with glucose

solution, using lex 632.8 nm (3.25 mW, 30 s). Dashed line is not a fit,but rather represents a perfect prediction
http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


15/21

of predicted analyte concentration against actual analyteconcentration. As shown in Fig. 16b, the SERS transduc-tion mechanism successfully predicts glucose concentrationquantitatively, with the vast majority of the data pointsfalling within the A or B areas of the Clark-error grid [256].

5 Conclusions

We have highlighted representative research accomplish-ments in the area of the tunable LSPR of silver and gold

nanoparticles in solution and on surfaces and the applica-tion of two unique plasmonic phenomenonae (i.e. theLSPR and SERS) to biological applications and systems.We have provided the reader with a sufficiently deep, yetaccessible, coverage of the underlying principles and theorybehind surface plasmonics and its application to nanobioa-nalysis. We have highlighted work in the field that is notonly relevant and innovative, but also shows immediatepromise for near-term implementation in real-world appli-cations such as point-of-care diagnostics, environmentalmonitoring, proteomics, drug discovery, and fundamentalbiological research. Future commercialisation of nanopar-ticle devices relies on a better understanding of theinteraction between nanoparticles and their immediateenvironment, as well as the development of techniques thatwill preserve nanoparticle optical activity under challengingbiological conditions.

6 Acknowledgments

We gratefully acknowledge financial support from theNanoscale Science and Engineering Initiative of theNational Science Foundation under NSF award EEC-0118025. Any opinions, findings, and conclusion orrecommendations expressed in this material are those ofthe authors and do not necessarily reflect those of the

National Science Foundation. A. Haes acknowledges theACS division of Analytical Chemistry sponsored byDuPont for fellowship support. This project was alsosupported by the Institute for Bioengineering and Na-noscience in Advanced Medicine at Northwestern Uni-versity, and the National Institutes of Health (EY13002 andEY13015). The authors wish to thank the followingindividuals for their experimental assistance, materials, orexpertise: L. Chang, W. E. Doering, M. Glucksberg,N. Halas, W. P. Hall, C. L. Haynes, E. Jeoung, M. Ka ll,W. L. Klein, O. Lyandres, M. Mrksich, C. Murphy, A. D.McFarland, S. Nie, J. C. Riboh, G. C. Schatz, K. Shafer-Peltier, J. Walsh, Y. Xia, and S. Zou.

7 References

1 Haynes, C.L., and Van Duyne, R.P.: Nanosphere Lithography: AVersatile Nanofabrication Tool for Studies of Size-DependentNanoparticle Optics, J. Phys. Chem. B, 2001, 105, pp. 55995611

2 Mulvaney, P.: Not all thats gold does glitter, MRS Bull., 2001, 26,pp. 10091014

3 El-Sayed, M.A.: Some Interesting Properties of Metals Confined inTime and Nanometer Space of Different Shapes, Accounts Chem.Res., 2001, 34, pp. 257264

4 Link, S., and El-Sayed, M.A.: Spectral Properties and RelaxationDynamics of Surface Plasmon Electronic Oscillations in Gold andSilver Nano-dots and Nano-rods, J. Phys. Chem. B, 1999, 103,pp. 84108426

5 Kreibig, U., Gartz, M., Hilger, A., and Hovel, H.: Opticalinvestigations of surfaces and interfaces of metal clusters, in Duncan,

M.A. (ed.): Advances in Metal and Semiconductor Clusters (JAIPress Inc., Stamford, , 1998), Vol. 4, pp. 345393

6 Mulvaney, P.: Surface Plasmon Spectroscopy of Nanosized MetalParticles, Langmuir, 1996, 12, pp. 788800

7 Kreibig, U.: Optics of nanosized metals, in Hummel, R. E. andWissmann, P. (Eds.) Handbook of Optical Properties (CRC Press,Boca Raton FL, 1997), Vol. II, pp. 145190

8 Hulteen, J.C., Treichel, D.A., Smith, M.T., Duval, M.L., Jensen,T.R., and Van Duyne, R.P.: Nanosphere Lithography: Size-TunableSilver Nanoparticle and Surface Cluster Arrays, J. Phys. Chem. B,1999, 103, pp. 38543863

9 Freeman, R.G., Grabar, K.C., Allison, K.J., Bright, R.M., Davis,J.A., Guthrie, A.P., Hommer, M.B., Jackson, M.A., Smith, P.C.,Walter, D.G., and Natan, M.J.: Self-Assembled Metal ColloidMonolayers: An Approach to SERS Substrates, Science, 1995, 267,pp. 16291632

10 Kahl, M., Voges, E., Kostrewa, S., Viets, C., and Hill, W.:Periodically structured metallic substrates for SERS, Sens. Actuators

B, Chem., 1998,51

, pp. 28529111 Schatz, G.C., and Van Duyne, R.P.: Electromagnetic Mechanism ofSurface-Enhanced Spectroscopy, (Wiley, New York, , 2002), Vol. 1

12 Haynes, C.L., and Van Duyne, R.P.: Plasmon-Sampled Surface-Enhanced Raman Excitation Spectroscopy, J

d.a. stuart, a.j. haes, c.r. yonzon, e.m. hicks and r.p. van duyne- biological applications of...

Documents