Analyst

PAPER

Dow

nloa

ded

by U

nive

rsity

of

Min

neso

ta -

Tw

in C

ities

on

19/0

4/20

13 1

1:56

:22.

Pu

blis

hed

on 0

7 Fe

brua

ry 2

013

on h

ttp://

pubs

.rsc

.org

| do

i:10.

1039

/C3A

N36

792D

View Article OnlineView Journal | View Issue

aCollege of Chemistry, Chemical Enginee

University, Suzhou 215123, P. R. China. E-

edu.cnbSchool of Chemistry and Chemical Engineer

P. R. China

† Electronic supplementary information (the procedures for fabricating BSSC thtrajectory of PATP on BSSC thin lm wispectra of PATP, SERS of PATP on at Agof PATP on BSSC thin lm with 514 nm e

Cite this: Analyst, 2013, 138, 2604

Received 2nd December 2012Accepted 6th February 2013

DOI: 10.1039/c3an36792d

www.rsc.org/analyst

2604 | Analyst, 2013, 138, 2604–261

Fabrication of a bowl-shaped silver cavity substrate forSERS-based immunoassay†

Shu Tian,ab Qun Zhou,*a Zhuomin Gu,a Xuefang Gub and Junwei Zheng*a

In this study, a metal sandwich substrate bridged by an immunocomplex has been created for a surface

enhanced Raman scattering (SERS)-based immunoassay. The bottom bowl-shaped silver cavity thin film

layer was prepared by electrodeposition using a closely packed monolayer of 700 nm diameter

polystyrene spheres as a template. The reflection spectra of the films were recorded as a function of

film thickness, and then correlated with SERS enhancement using p-aminothiophenol as the probe

molecule. The results demonstrate that SERS enhancement can be maximized when both the frequency

of the incident laser and Raman scattering approach the resonance frequency of the localized surface

plasmon resonance, providing a guideline for the fabrication and further application of these

nanocavity arrays. The second layer of silver was introduced by the interactions between the

immunocomplexes in the middle layer of the sandwich architecture and the silver nanoparticles. The

proposed structure was used to perform the SERS-based immunoassay. The labeled protein can be

detected over a wide concentration range and the detection limit of TRITC and Atto610 labeled

proteins were 50 and 5 pg mL�1, respectively. The results demonstrate that the new SERS substrate is

suitable for the quantitative identification of biomolecules.

Introduction

The development of new, simple, and sensitive methods for thedetermination of proteins based on the specic recognition ofan antigen with a corresponding antibody has gained increasedinterest.1,2 Enzyme-linked immunosorbent assay (ELISA)3,4 andsurface enhanced uorescence (SEF)5,6 are the most commonlyused immunoassay methods. However, for ELISA, cumbersomeexperimental steps, the large consumption of reagents and arelatively narrow detection range are the stumbling blocks forhigh-efficiency detection.7 For uorescence-based methods,broad emission spectra from molecular uorophores makemultiplex immunoassay impossible, and the susceptibility tophotobleaching may considerably weaken their detectionlimits. A large number of studies have shown that surfaceenhanced Raman scattering (SERS) is an alternative rapid andsensitive technique to immunoassays as it is non-invasive,involves narrow bandwidths, has uorescence-quenching

ring and Materials Science, Soochow

mail: zhq@suda.edu.cn; jwzheng@suda.

ing, Nantong University, Nantong 226007,

ESI) available: Schematic illustration ofin lm; thickness-dependent spectralth 633 nm excitation; normal Ramanand BSSC thin lm; spectral trajectoryxcitation. See DOI: 10.1039/c3an36792d

2

properties, an adjustable excitation wavelength (UV to NIRregion), and the SERS signal from water is extremely weak.8–18

SERS is based on the considerable enhancement of theRaman cross-section of molecules when they are placed in theproximity of a roughened metal surface, due to the contributionof the electromagnetic (EM) and chemical effect.19 The rstobstacle that needs to be solved for SERS detection purposesis the fabrication of a suitable substrate. Gold and silvernanoparticles are commonly used as SERS-active substratesbecause of their simple and convenient preparation andstrong SERS enhancement. The shortcomings of a dispersedmetal sol are equally conspicuous. One of the most obstinateproblems is poor colloidal stability, resulting in decreasedreproducibility and reliability. Periodic metal structure can beused to overcome the problems of poor control over particleaggregation and to generate a huge enhancement at the sametime.20 Moreover, it can precisely conne the local EM eld atthe surface of the metallic structure and optimize the localizedsurface plasmon resonance (LSPR) to a specic wavelength,even to the region of NIR, which is a spectral regionfree from unwanted uorescence or photoinduced sampledegradation.21–23

A metal nanocavity array can be a good candidate as a SERS-active substrate because of its specic surface area morphology,high enhancement, reproducibility, highly homogeneouscharacteristics and stability. The method for fabricating such ananocavity array was rst introduced by Velev and coworkers in1998,24,25 and has been well developed by the same group and

This journal is ª The Royal Society of Chemistry 2013

Paper Analyst

Dow

nloa

ded

by U

nive

rsity

of

Min

neso

ta -

Tw

in C

ities

on

19/0

4/20

13 1

1:56

:22.

Pu

blis

hed

on 0

7 Fe

brua

ry 2

013

on h

ttp://

pubs

.rsc

.org

| do

i:10.

1039

/C3A

N36

792D

View Article Online

other groups.23,26–31 In these studies, polystyrene (PS) or silicaspheres were rst self-assembled onto a glass or a metalsubstrate. Different kinds of nanoparticles then inltrated thetemplate to form the nal structure. For example, Jiang et al.used silica spheres as the template and successfully inltratedthe interspace of the template with different metals.32 Li et al.fabricated a hierarchical silver bowl-like array by thermaldecomposition of silver acetate solution inltrated within theinterstice of the PS template.33 Johnson and Walsh reported avery simple approach which requires no additional reducingagents or specialist instrumentation.34 In this study, thetemplate was made up of a kind of hydroxyl-terminated PSnanosphere, acting as the reducing agent as well as the struc-ture-directing material. Silver nitrate solution within the inter-stice of the PS template was kept at 80 �C for a period of time inthe dark to generate the nal structure. We have previouslyprepared a silver cavity array on the ordered structure of PSparticles via the reduction reaction of silver ions with glucose.27

Despite the effective preparation of the nanocavity arrays, thelack of regulation of the morphology restricts their furtherapplication. Bartlett and co-workers developed another power-ful technique for producing nanostructuredmetal lms from PStemplates by electrodeposition (originally Co, Pt, and Pd).30 Byaltering the amount of charge passed during deposition and thesize of the PS spheres, the authors fabricated Au and Pt nano-void arrays with varying thicknesses and diameters.29 These Aunanovoid arrays have been used as reproducible and tailorablesubstrates for several SERS applications.35–37 Recently, the samegroup reported a theoretical study on the plasmonic propertiesof a gold sphere segment void (SSV) surface,38,39 and they furtherdemonstrated that placing silver nanoparticles inside suchcavities markedly improves their plasmonic response by twoorders of magnitude.

In addition, to detect the Raman signal of an antigen orantibody at very low concentrations, a large Raman cross-section molecule (such as resonant dyes40–42 or aromatic thiolcompounds43) labeled metallic nanoparticles with correspond-ing antibody or antigen adsorbed on its surface are frequentlyused. That is, SERS spectra or the surface enhanced resonanceRaman scattering (SERRS) spectra of the label are detectedrather than those of the biomolecules themselves.9 However,proteins compete with the target molecules for the metallicnanoparticle surfaces, potentially reducing the interactionsbetween the target molecules and nanoparticles, therebyleading to a reduced dye SERS signal. Furthermore, the immunerecognition capabilities between the antigen and correspondingantibody are signicantly reduced aer attaching them to themetal surface, thereby reducing the limit of detection. Cova-lently binding labeled molecules to the antibody can be analternative method.11,12,41 However, the EM enhancement of theanalyte exponentially decays with its distance from the surfaceof the SERS-active substrate.44 In this labeled system, thedistance between the labeled molecules and the surface ofenhancement substrate is noticeably enlarged because of thelarge hydrodynamic radius of the adsorbed antigen and anti-body, thereby reducing the magnitude of the EM enhancementof the dye molecules.

This journal is ª The Royal Society of Chemistry 2013

Herein, we report a SERS-based immunoassay on a bowl-shaped silver cavity (BSSC) thin lm with the assembly of anAg/immunocomplex/BSSC sandwich structure. Compared withother reported protein detection methods based on a BSSC-likesubstrate,23,35,45 the proposed method has several novel andimportant features. First, the cavity structure can localize theEM eld into a small volume; thus, the EM energy density canbe signicantly increased. Silver nanoparticles are introducedinto the cavity array by silver staining. Such a structure allowsthe entire immunocomplexes to be in the coupled EM eldbetween the silver cavity and silver nanoparticles, therebygenerating a strong EM eld and a signicantly enhancedRaman signal. Second, the upper silver nanoparticles show nonegative effects on the specic recognition reaction between theantigen and corresponding antibody. Lastly, direct contactbetween the silver nanoparticles and the uorescent dye mole-cules can enhance the Raman signal while quenching theuorescence signal. The plasmon resonances of such substrateswere observed by reection spectroscopy. The signicantenhancement, reproducibility, and application of such bowl-shaped substrates on protein detection were evaluated by SERS.

Experimental sectionChemicals and materials

Rabbit immunoglobulin G (IgG), TRITC–antirabbit IgG, bovineserum albumin (BSA), avidin, Atto610-biotin and p-amino-thiophenol (PATP) were obtained from Sigma Co., Ltd. and wereused without further purication. Silver nitrate, trisodiumcitrate and all other chemicals were purchased from ShanghaiChemical Reagent Company and were used as received. Ultra-pure water used in washings and all solution preparation wasproduced using a Milli-Q system with resistivity greater than18.2 MU cm.

Assembly of the colloidal templates

The monolayer of PS particles was generated by self-assembly atthe air/water interface in a 5 cm diameter Petri dish (moredetails can be found in an earlier publication),27 then the PSparticles were transferred onto the gold coated slides. The goldcoated slides used in this work were prepared by EM sputteringof a 10 nm thick chromium layer, followed by a 200 nm thickgold layer onto ITO glass slides.

Bowl-shaped silver cavity thin lm preparation

Electrochemical deposition was performed in a thermostatedcell at room temperature, using a conventional three-electrodeconguration controlled by a CHI 660D electrochemical station.The template-coated substrate was the working electrode, aplatinum and anHg/Hg2SO4 electrode were used as the auxiliaryand reference electrodes, respectively. Silver was deposited froma cyanide free plating bath (0.1 M AgNO3 + 0.1 M EDTA + 0.05 MNH4NO3 and several milliliters of concentrated ammonia wereadded to ensure the pH valuewas 9–10).46The bowl-shaped silverlms were produced using multi-current pulse plating with therst pulse at a current density of 30 mA cm�2 for 100 ms (this is

Analyst, 2013, 138, 2604–2612 | 2605

Fig. 1 SEM images of Ag cavities (a–f). The thicknesses of the structured silvercavities are (c) 0.3R, (d) 0.9R, (e) 1.2R and (f) 1.4R, respectively.

Analyst Paper

Dow

nloa

ded

by U

nive

rsity

of

Min

neso

ta -

Tw

in C

ities

on

19/0

4/20

13 1

1:56

:22.

Pu

blis

hed

on 0

7 Fe

brua

ry 2

013

on h

ttp://

pubs

.rsc

.org

| do

i:10.

1039

/C3A

N36

792D

View Article Online

important to ensure good adhesion) followed by a train of pulsesof 5 mA cm�2 for 60 ms separated by a rest time of 1 s (zerocurrent).

Adsorption of the PATP molecules

The adsorption of the PATP molecules on the BSSC thin lmwas carried out by immersion of the different substrates into1 mM PATP ethanol solution for 12 h. The slides were thor-oughly rinsed with ethanol to remove the physically adsorbedPATP molecules and then dried in a jet of nitrogen before SERSmeasurement.

Immunoreaction protocol

Rabbit IgG and avidin (different concentrations in 0.01 M PBSbuffer) were immobilized on the silver cavity thin lm byimmersing the substrates (3 mm� 3 mm square pieces) in 1mLprotein solutions at different concentrations for 2 h at 37 �C,respectively. Aer being rinsed three times with a washingbuffer (0.01 M PBS with 0.05% tween 20), the substrates weresoaked in a blocking buffer (0.1% BSA in PBS solution) for 2 h at37 �C, and then rinsed again three times with the washingbuffer. The substrates coated with rabbit IgG or avidin wereimmersed in 1 mL solutions of their corresponding ligands100 mg mL�1 TRITC–goat antirabbit IgG or 40 mg mL�1 Atto610-biotin for 2 h at 37 �C, respectively. Then all the substrates wererinsed three times by the washing buffer. Aer that, the silversubstrates with adsorbed target immunocomplexes wereimmersed in silver colloid prepared as detailed below for 3 h at37 �C and washed with ultrapure water three times. Colloidalsilver used in this work was prepared by the aqueous reductionof silver nitrate with trisodium citrate according to Lee'sprotocols.47 The concentration of the silver nanoparticles wasestimated to be 1.6 � 10�10 M (given that the silver nano-particles had a spherical-like shape with 60 nm diameter, esti-mated from the TEM image).

SERS measurement

SERS spectra were measured using a Jobin Yvon/HORIBA Lab-Ram ARAMIS Raman spectrometer equipped with an integralBX 41 confocal microscope. The excitation sources included anair-cooled argon ion laser (514 nm, 20 mW) and an HeNe laser(633 nm, 17 mW). For the detection of protein in this work, a D2

attenuation lter was used and the accumulation time was 30 s.Raman scattering was detected with 180 geometry using amultichannel air-cooled (�70 �C) charge-coupled device (CCD)camera.

Results and discussionPreparation and characterization of the BSSC thin lm

In the present study, we fabricated the two-dimensional BSSCthin lm by electrochemical deposition as illustrated in Fig. S1(see ESI†). For the fabrication of the highly ordered PS spheretemplate, water/ethanol dispersion containing monodisperse700 nm diameter PS spheres (5 wt%, also can be differentdiameter) was rst injected slowly onto the surface of water by a

2606 | Analyst, 2013, 138, 2604–2612

microsyringe, the PS spheres spread freely over the top surfaceof water until they nearly covered the whole surface area. Then,a few drops of 2 wt% SDS solution were added onto the watersurface to lower the surface tension and to make the PS spheresclosely packed. The whole system was kept in an environmentfree from outside disturbances in order to sediment the sus-pended PS particles in the water for 24 h. The monolayer of thespheres was then transferred onto the surface of a gold coatedITO glass.

Despite the high quality of the deposits obtained from thealkaline cyanide solutions, we chose a cyanide free plating bathdue to its low toxicity. Multi-current pulse plating wasemployed, the rst pulse was used to initiate the formation ofthe nuclei, the second pulse was used to control the growth ofthe nuclei formed during the previous pulse, the introduction ofthe rest time ensured that the silver ion complex in the solutionhad enough time to spread to the deposition interface. Aerelectrodeposition, the PS spheres were dissolved in toluene,leaving an ordered array of interconnected sphere segmentvoids. The obtained structured silver cavities were robust andadhered well to the ITO electrode.

The inltration and sedimentation of the silver nanoparticlesinto the void space among the PS spheres resulted in theformation of a silver framework directed by the orderly assem-bled PS spheres. Typical scanning electron microscopy (SEM)images of the silver cavity array are shown in Fig. 1. The top viewof the obverse side shows highly ordered hexagonally arrangedcircular pores (each cavity touching six others in one layer) with along-rangeperiodicity (Fig. 1a). Thedistance between the centersof two neighboring pores is ca. 700 nm, identical to the diameter

This journal is ª The Royal Society of Chemistry 2013

Paper Analyst

Dow

nloa

ded

by U

nive

rsity

of

Min

neso

ta -

Tw

in C

ities

on

19/0

4/20

13 1

1:56

:22.

Pu

blis

hed

on 0

7 Fe

brua

ry 2

013

on h

ttp://

pubs

.rsc

.org

| do

i:10.

1039

/C3A

N36

792D

View Article Online

of the colloidal PS spheres, indicating that the Ag deposition haslittle effect on the close packing of the spheres. The image of abending surface clearly shows the three-dimensional structure ofthe cavities and further veries the ordered cavity structures(Fig. 1b). The further magnied image shows that the as-prepared silver cavities mainly consist of nanoparticles in thesize range of 50–80 nm, these silver nanoparticles contact witheach other, resulting in the nal structure.

Moreover, the depth of the cavities, the periodic spacing andpore diameter of the Ag cavities can be easily regulated bychanging the times of the second pulse. The SEM images of thesilver cavity with different thicknesses are shown in Fig. 1c–f. Inthese four different situations, the numbers of the second pulseused were 300, 650, 1100, and 1400, respectively. According tothe pore diameter obtained from the SEM images, the depth ofthese four different cavities were calculated to be 0.3R, 0.7R,1.2R, and 1.4R (where R is the radius of the template PSspheres), respectively.29 The aperture of the silver cavitiesincreased with an increase in the number of the second pulse,and reached amaximum at a pore diameter of 700 nm. A furtherincrease in the number of small pulses led to a continuousincrease in the thickness until the PS template was completelycovered. The spacing between the touching cavities decreasedas the pore diameter of the silver cavity increased.

Fig. 2 (a) UV-vis reflection spectra of BSSC thin film with different thicknessesand (b) thickness-dependent spectral trajectory of PATP on BSSC thin film with514 nm excitation. The vertical axis of b is different points on 6 differentsubstrates, 1–20 represent 0.3R thickness, 21–40 for 0.7R, 41–60 for 0.9R, 61–80for 1.2R, 81–100 for 1.4R and 101–120 for 1.6R. Also shown as a red curve is theSERS spectrum of PATP. A 514 nm excitation line with laser power of 20 mWand a D1 attenuation filter were used. The acquisition time was 5 s with oneaccumulation.

Tunable LSPR and SERS properties

It is well known that the plasmonic properties of metal nano-structures are critically dependent on their shape and size (voiddiameter, lm thickness in this case), and the plasmon reso-nances can be observed in the reection spectra as dips becausethe energy of surface plasmon polariton (SPP) is dissipated asohmic loss in the metals.22 Bartlett and co-workers systemati-cally researched the SPP properties of gold cavity arrays.Although the present study focused on silver cavity arrays, themechanism in the two cases can bear some similarity. In Fig. 2a,we present typical reection spectra of the as-preparedsubstrates at thicknesses of 0.3R, 0.7R, 0.9R, 1.2R 1.4R and 1.6R.Each spectrum shows some dips whose locations change withthe thickness, indicating the alterations of the SPP modes. Ascan be seen in Fig. 2a, when the cavity depth is small, the onlydip that can be observed is at ca. 410 nm, similar to thereective curve of at Ag (gure not shown), which originatesfrom the well-known surface plasmon resonance of silvernanoparticles. Compared with the spectrum obtained fromcavities with 0.3R thickness, new dips appear in curves forcavities with bigger thickness, which can be denoted as B, 1P+and 1P�. According to theoretical and experimental resultspresented by Bartlett and co-workers,48–50 B can be assigned tothe q11 Bragg mode, modes 1P+ and 1P� are the bonding andantibonding states resulting from the coupling between the 1Ppure dipolar Mie mode and the cavity rimmode. As the depth ofthe cavity continues to increase, the Mie modes change theirenergies as a function of the depth and the B mode disappeareduntil the cavity depth was greater than 1.4R. This variation ofthe SPP of the silver cavity is in good agreement with the resultsobtained for the gold SSV.

This journal is ª The Royal Society of Chemistry 2013

A set of cavity arrays with different depths was fabricated toobtain a deeper insight into the correlation between the LSPRand the SERS intensities on this BSSC thin lm. PATP was usedas a probe molecule for SERS study because it has a largescattering cross-section and forms a compact monolayer onmetal surfaces. SERS spectrum of PATP is shown as a red curvein Fig. 2b, the predominant bands are located at 1003, 1077,1178, and 1576 cm�1, which belong to the a1 modes (totallysymmetric, 18a, 7a, 9a, and 8a, respectively); the other set at1140, 1390, 1435 cm�1, which are assigned to the b2 modes(asymmetric, 9b, 3 and 19b, respectively).51 According to thestudy of Osawa et al., the band at 1077 cm�1 was suggested to beenhanced by the EM mechanism only. Therefore, the intensityof the band at 1077 cm�1 was used as an indicator of theenhancement on different nanostructured surfaces.

It has been well-documented that SERS enhancement can bemaximized when both the frequency of the incident laser andRaman-scattered photons approach the resonance frequency ofLSPR.52 As shown in Fig. 2a, two dashed lines (in) imply theincident light used in our present experiments, 514 and 633 nm,respectively. Two corresponding solid lines represent theRaman scattering wavelength corresponding to the 1077 cm�1

band of the PATP molecules, which are 544 and 679 nm (out),respectively. In Fig. 2b, 120 Raman spectra of PATP under the

Analyst, 2013, 138, 2604–2612 | 2607

Fig. 3 (a) Schematic illustration of the procedures for immunoreaction protocoland (b) SEM images of the silver cavity after silver staining. Inset is a furthermagnified figure of a single cavity.

Analyst Paper

Dow

nloa

ded

by U

nive

rsity

of

Min

neso

ta -

Tw

in C

ities

on

19/0

4/20

13 1

1:56

:22.

Pu

blis

hed

on 0

7 Fe

brua

ry 2

013

on h

ttp://

pubs

.rsc

.org

| do

i:10.

1039

/C3A

N36

792D

View Article Online

514 nm laser excitation were randomly collected from sixdifferent substrates corresponding to Fig. 2a (20 points fromeach substrate), and plotted as two-dimensional Raman maps(two-dimensional Raman maps of the PATP under 633 nm laserexcitation are shown in the Fig. S2†). Direct correlation can beobserved between the Raman intensities shown in Fig. 2b andplasmonic resonances in Fig. 2a. That is, when the lLSPR wasclose to the lin or the lout (1.2R and 1.4R for 514 nm excitation,0.9R and 1.6R for 633 nm excitation), corresponding to thedesired resonance coupling between ingoing and outcomingradiation with the plasmons generated on the BSSC surface,large signal enhancement occurred. The result provides guide-lines for the design and fabrication of nanocavity arrays.

To elucidate the SERS activity of the silver cavities preparedin the present case, the SERS spectrum of the PATP moleculesadsorbed in the cavities was measured and compared with thatobtained on the at region (the region without PS template) ofthe same substrate; the results are shown in Fig. S3A (see ESI†).Although the at silver thin lm yielded weak SERS (spectrumb), the absolute SERS intensity of the PATP molecules adsorbedinside the silver cavities (spectrum c) was ca. 20 times greaterthan that of spectrum b, indicating that the Raman scattering ofthe adsorbed PATP molecules gained greater enhancement inthe silver cavities. However, the surface area of the BSSC thinlm increased by only 2 times compared with that of the atone, that is, the higher SERS enhancement of the BSSC thin lmmay not be explained by the increase in surface area. As the onlydifference between the two substrates is the structure of thedeposited silver, it is reasonable to conclude that the particularcavity structure contributes to the additional enhancement byan order of magnitude.

The surface enhancement factor is an important indicator ofthe SERS activity of a substrate. The averaged surfaceenhancement factor (ASEF) of the BSSC thin lms can becalculated using the following equation:53

ASEF ¼ NbulkIsurf

NsurfIbulkR

where Isurf and Ibulk denote the integrated intensities for thestrongest band of the PATP adsorbed on the surface of BSSCthin lms and the solid PATP, respectively. Nsurf and Nbulk

represent the numbers of the corresponding surface and solidmolecules effectively excited by the laser beam, respectively. R isthe surface roughness which can be obtained from AFM andshould be taken into account during the calculation (in thisstudy, R was 1.3). Raman intensity at 1576 cm�1 and 1435 cm�1

were used to calculate the ASEFs for a1 and b2 vibrationalmodes, respectively. The calculated results were 1.0 � 106 forthe a1 mode and 1.1 � 107 for the b2 mode (for calculationdetails, see ESI†).

In addition to the high-intensity enhancement, the SERSspectra were also found to be highly reproducible on this BSSCthin lm. The reproducibility of the surface enhancement wasassessed by randomly recording 100 SERS spectra of PATP onve BSSC thin lms with nominally the same lm thickness.The deviation of the peak intensity at 1077 cm�1 was calculatedas ca. 8.4% (Fig. S3B, ESI†). The results demonstrate that BSSC

2608 | Analyst, 2013, 138, 2604–2612

thin lms exhibit excellent performance in the Ramanenhancement and can be potentially applied in the detection oftrace amounts of target analytes.

Labeled protein detection

Protein molecules always show weak SERS signals; thus, Ramandye and uorescent molecules with SERS activity are usuallyused. TRITC is a derivative of rhodamine and a commonly useduorescent dye with a maximum absorption of 555 nm (thechemical structure of TRITC can be seen in the ESI†). Thus, theSERRS spectra of TRITC can be obtained using 514 nm excita-tion. Fig. 3a illustrates the different stages during the formationof the Ag/Raman reporter labeled immunocomplex/BSSC thinlm sandwich structure. First, a BSSC thin lm was coated withthe antigen, the area where no antigen adsorbed was blockedwith BSA. Secondly, a Raman reporter labeled-antibody solutionwas added for specically recognizing the antigen. Thesubstrate was then washed to remove the nonspecically boundantibody. Finally, the whole substrate was immersed into the Agsol to bring the Ag nanoparticles into the cavity and form theAg/immunocomplex/BSSC thin lm sandwich structure. Fig. 3bshows the SEM image of themorphology of the silver cavity aersilver staining. The silver nanoparticles were uniformly adsor-bed inside the cavity, indicating the uniform adsorption of the

This journal is ª The Royal Society of Chemistry 2013

Paper Analyst

Dow

nloa

ded

by U

nive

rsity

of

Min

neso

ta -

Tw

in C

ities

on

19/0

4/20

13 1

1:56

:22.

Pu

blis

hed

on 0

7 Fe

brua

ry 2

013

on h

ttp://

pubs

.rsc

.org

| do

i:10.

1039

/C3A

N36

792D

View Article Online

immunocomplex inside the cavity. The BSA used for blockingcan also adsorb silver nanoparticles through electrostaticinteraction; however, the effect of these silver nanoparticles onthe Raman signal of TRITC is almost negligible because nosandwich structure is formed. It should be noted that in thetraditional uorescence detection, one would rather label thereagent that is coated to the substrate, since labeling antigensor antibodies in real samples at realistically low concentrationsis usually slow and problematic. In our present study, it seemsinappropriate to label the antigen that adsorbed on the silvercavity, due to the distance-dependent enhancement of SERS.That is, efficient SERS requires close contact of the studiedmolecules with the metallic surface, a distance below 2–3 nm.Taking into account that the hydrodynamic radius of adsorbedrabbit IgG is larger than 10 nm, the EM enhancement of theanalyte exponentially decays with its distance from the surfaceof the SERS-active substrate.

A typical SERRS spectrum of the TRITC labeled immuno-complex obtained on this proposed Ag/immunocomplex/BSSCthin lm sandwich structure is shown in Fig. 4a. The bands at1650, 1543, 1509, 1361 and 1284 cm�1 can be attributed to thearomatic C]C stretching vibrations.45 In the present study, a1650 cm�1 characteristic Raman band was chosen as the indi-cator for quantitative immunoassay (as discussed below). TheRaman scattering wavelength was then calculated as 562 nm;thus, 1.2R thickness BSSC thin lms were used as the SERS-active substrate to generate an optimized SERS signal.

Before the SERRS-based quantitative immunoassay, fourSERS spectra were rst measured independently to explore thecontributions of the bowl-shaped cavity and the second layermetallic nanoparticles to the SERRS spectrum of the TRITClabeled immunocomplex, as shown in Fig. 4. The SERRSintensity of the TRITC labeled complexes from the Ag/BSSCsandwich (Fig. 4a) was about 6 times stronger than those fromthe Au/BSSC sandwich (Fig. 4b) and about 50 times strongerthan those from BSSC thin lms without metallic staining(Fig. 4c). These results indicated that the adsorption of metalnanoparticles remarkably contributed to the high SERS

Fig. 4 SERS spectra of TRITC-labeled immunocomplexes adsorbed on BSSC thinfilm substrate (a) with Ag staining, (b) with Au staining, (c) without Ag staining,and (d) SERS spectra of TRITC-labeled immunocomplexes adsorbed on flat Agwith Ag staining. The concentration of rabbit IgG is 20 ng mL�1.

This journal is ª The Royal Society of Chemistry 2013

enhancement. The role of the BSSC structure was experimen-tally examined by comparing the spectrum from the Ag/BSSCsandwich with a spectrum from the Ag/at Ag sandwich struc-ture (Fig. 4d). The immune complex inside the Ag/at Agsandwich structure yielded weak SERS; the intensities wereabout 20 times smaller than those from the Ag/BSSC sandwichstructure, which is approximately the same as the enhancementeffect of the BSSC lms on PATP molecules. The results suggestthat both the bottom Ag cavity and the upper Ag or Au nano-particles contributed to the SERRS signals observed in thisstudy. The silver spheroidal particles can also generate largerEM enhancement than the gold ones; thus, BSSC thin lm andAg nanoparticles were used to form an Ag/immunocomplex/BSSC sandwich structure for further SERRS detection in thisstudy.

The considerable enhancement of SERRS can be explainedas follows. First, the BSSC thin lms mainly consist of Agnanoparticles in the size range from 50 nm to 80 nm, such aparticle is particularly suited for SERS enhancement,54 thesenanoparticles further interconnect to form many aggregates,accordingly, a large SERS enhancement is achieved at the gap oftwo adjacent nanoparticles because of the enormous EM eld.8

Secondly, we believe that the EM eld is conned in a cavitywith very small volume and could gain further enhancement. Asdemonstrated by Netti and Coyle,55,56 a metal cavity couples veryeffectively to incident light and completely laterally connes thesurface plasmons; multiple plasmon modes covering a widespectral range may be excited. In contrast to spherical metalparticles, electric eld overlap due to negative curvatureconnement further increases EM energy density inside thecavity. Thirdly, as demonstrated by Huang et al., the negativecurvature can serve as a near-eld collimation for chromo-phores arranged at the optical focus of the nanocavities, reso-nantly concentrating light near the bottom of the void surface,therefore signicantly enhancing the coupling between parti-cles and particles inside voids.39 In our present case, an addi-tional enhancement of the EM eld is obtained with the cavityacting to harvest the light, which is then focused in the presenceof the silver nanoparticle introduced by Ag staining. Theintroduction of silver nanoparticles produces strong plasmoncoupling between the underlying thin lm and the uppernanoparticles, thus generating large SERS enhancement acrossthe cavity array. Last but not least, the direct contact betweensilver nanoparticles and the uorescent dye molecules plays animportant role in quenching the uorescence signal, reducingthe limit of detection of the immunocomplexes.

Quantitative immunoassay

Rabbit IgG solutions of different concentrations from 20 ngmL�1 to 100 pg mL�1 were examined; their SERRS spectraare shown in Fig. 5a. A higher rabbit IgG concentration resultedin more intense SERRS signals. The intensity of a peak at1650 cm�1 versus the logarithm of the concentration of rabbitIgG was used for Gaussian curve tting, and plotted in Fig. 5b.Each point represents an average of eight randomly selectedpoints; each error bar represents the sample-to-sample

Analyst, 2013, 138, 2604–2612 | 2609

Fig. 5 SERS spectra of (a) TRITC-labeled immunocomplexes with different concentrations of rabbit IgG concentrations, (b) SERS intensity of band at 1650 cm�1 asa function of concentration using Gaussian curve fitting. The incident laser wavelength is 514 nm.

Analyst Paper

Dow

nloa

ded

by U

nive

rsity

of

Min

neso

ta -

Tw

in C

ities

on

19/0

4/20

13 1

1:56

:22.

Pu

blis

hed

on 0

7 Fe

brua

ry 2

013

on h

ttp://

pubs

.rsc

.org

| do

i:10.

1039

/C3A

N36

792D

View Article Online

variability in the SERRS intensities. As can be seen, theremarkable increase in the SERRS intensities of TRITC covers alarge range of protein concentrations. A linear relationshipexists between the logarithm of the rabbit IgG concentrationand the SERRS peak intensity within the range of 0.2 ng mL�1 to2 ng mL�1 (inset of Fig. 5b). The detection limit of the presentmethod was found to be as low as 50 pg mL�1. The resultssuggested that a SERS-based immunoassay on such a bowl-shaped silver substrate could be a new SERS-based method forqualitative and quantitative immunoassays. In comparison withuorescence detection, the proposed SERS-based immunoassayexhibited advantages in sensitivity and selectivity. In thepresent study, TRITC molecules served as Raman reporters.When the excitation laser was selected properly to match theelectronic energy level of the dye molecules, additionalenhancement could be achieved due to a resonance effect. Inaddition, the relative narrow bandwidth of the SERS signals mayalso provide a better selectivity for the detection.

Atto610 was chosen as another SERRS probe protein–ligandcomplex to further conrm whether the proposed sandwichstructure is suitable for immunoassay. Atto-610 was chosenbecause it belongs to a new generation of uorescent labels forthe red spectral region. Moreover, the interaction betweenAtto610-biotin and avidin can also be attributed to protein–small molecule interactions. The immunoreaction protocol of

Fig. 6 SERRS spectra of Atto610-biotin–avidin complex with different concentratioGaussian curve fitting (b). The incident laser wavelength is 633 nm.

2610 | Analyst, 2013, 138, 2604–2612

avidin and Atto610-biotin is almost the same as that illustratedin Fig. 3a, where avidin is the antigen and Atto610-biotin servesas the Raman reporter labeled antibody. As shown in Fig. 6,under SERS and resonance conditions with 633 nm laser exci-tation, the bands at 1634, 864, 801, and 538 cm�1 can beassigned to vibrational modes of Atto610;41 a strong peak at538 cm�1 of Atto610 was chosen to plot the calibration curve.According to the correlation between SERS and LSPR, a 0.9Rthickness BSSC thin lm was chosen as the bottom substrate tofabricate the sandwich structure. Fig. 6a exhibits the SERRSspectra of Atto610 with a concentration of avidin from 10 ngmL�1 to 10 pg mL�1. Over a large range, the SERRS intensitiesincrease as the concentration of avidin increases. When theconcentration was 10 pg mL�1, a weak SERRS signal of Atto-610was observed (inset of Fig. 6a). The detection limit of the BSSCsubstrates for Atto610-biotin was also investigated, and theresult was estimated to be �5 pg mL�1 (S/N ¼ 3). The intensityof a peak at 538 cm�1 versus the logarithm of the concentrationof avidin was used for Gaussian curve tting and plotted inFig. 6b. A linear relationship also exists between the logarithmof the avidin concentration and the SERRS peak intensity withinthe range of 30 pg mL�1 to 0.5 ng mL�1 (inset of Fig. 6b). Thesedata further indicate that the proposed method can serve as auniversal, highly sensitive and powerful technique to determineprotein concentrations.

ns of avidin (a) and concentration-dependent SERRS intensities at 538 cm�1 using

This journal is ª The Royal Society of Chemistry 2013

Paper Analyst

Dow

nloa

ded

by U

nive

rsity

of

Min

neso

ta -

Tw

in C

ities

on

19/0

4/20

13 1

1:56

:22.

Pu

blis

hed

on 0

7 Fe

brua

ry 2

013

on h

ttp://

pubs

.rsc

.org

| do

i:10.

1039

/C3A

N36

792D

View Article Online

Conclusion

In summary, a bowl-shaped silver cavity thin lm was fabri-cated via multi-current pulse electrodeposition using closelypacked PS spheres as a template. The thickness of the as-prepared silver cavity can be conveniently modulated byaltering the number of the second pulse. Through reectionspectroscopy, the LSPR property of the thin lm was studiedand correlated with the SERS intensities of PATP molecules inthe cavity. The results show that the LSPR wavelength changeswith the variation in the cavity thickness and the largestenhancement can be obtained when the position of LSPRapproachs the range of the excitation laser and the Ramanscattered light. Accordingly, the conditions for the next proteindetection were optimized. An Ag/immunocomplex/BSSC wassubsequently constructed and applied to a SERS-basedimmunoassay. Using this immunoassay method, labeledproteins can be detected within a wide concentration range,and the detection limit of TRITC and Atto610 labeled proteinwere 50 and 5 pg mL�1. This preliminary study shows that thenew SERS-based method is very promising for immunoassay.Higher sensitivity and a wider linear range might be achievedby employing the dye molecules with electronic energy levelsmore efficiently matching the excitation lasers, or with a largerscattering cross-section.

Acknowledgements

Financial support from the Nature Science Foundation of China(no. 20873089, 20975073, 21177067, 21173122), Nature ScienceFoundation of Jiangsu Province ( no. BY2011130, BK2010034)and Scientic and Technological Innovation Projects of Nan-tong City (HS2012006) are gratefully acknowledged. We wouldlike to thank Dr Lei Chen for his assistance in this work.

References

1 S. F. Kingsmore, Nat. Rev. Drug Discovery, 2006, 5, 310–321.2 H. T. Maecker, G. P. Nolan and C. G. Fathman, Curr. Opin.Endocrinol., Diabetes Obes., 2010, 17, 322–328.

3 E. Engvall and P. Perlmann, Immunochemistry, 1971, 8, 871–874.

4 E. Engvall and P. Perlmann, J. Immunol., 1972, 109, 129–135.5 J. Lakowicz, C. Geddes, I. Gryczynski, J. Malicka,Z. Gryczynski, K. Aslan, J. Lukomska, E. Matveeva,J. Zhang, R. Badugu and J. Huang, J. Fluoresc., 2004, 14,425–441.

6 K. Aslan, I. Gryczynski, J. Malicka, E. Matveeva, J. R. Lakowiczand C. D. Geddes, Curr. Opin. Biotechnol., 2005, 16,55–62.

7 H. Zhu and M. Snyder, Curr. Opin. Chem. Biol., 2003, 7, 55–63.

8 H. Xu, E. J. Bjerneld, M. Kall and L. Borjesson, Phys. Rev.Lett., 1999, 83, 4357.

9 W. E. Doering and S. Nie, Anal. Chem., 2003, 75, 6171–6176.10 M. Manimaran and N. R. Jana, J. Raman Spectrosc., 2007, 38,

1326–1331.

This journal is ª The Royal Society of Chemistry 2013

11 X. X. Han, L. J. Cai, J. Guo, C. X. Wang, W. D. Ruan,W. Y. Han, W. Q. Xu, B. Zhao and Y. Ozaki, Anal. Chem.,2008, 80, 3020–3024.

12 X. X. Han, Y. Kitahama, Y. Tanaka, J. Guo, W. Q. Xu, B. Zhaoand Y. Ozaki, Anal. Chem., 2008, 80, 6567–6572.

13 K. Hering, D. Cialla, K. Ackermann, T. Dorfer, R. Moller,H. Schneidewind, R. Mattheis, W. Fritzsche, P. Rosch andJ. Popp, Anal. Bioanal. Chem., 2007, 390, 113–124.

14 L. Chen, X. Han, J. Yang, J. Zhou, W. Song, B. Zhao, W. Xuand Y. Ozaki, J. Colloid Interface Sci., 2011, 360, 482–487.

15 X. X. Han, G. G. Huang, B. Zhao and Y. Ozaki, Anal. Chem.,2009, 81, 3329–3333.

16 X. X. Han, B. Zhao and Y. Ozaki, Anal. Bioanal. Chem., 2009,394, 1719–1727.

17 S. Keskin and M. Culha, Analyst, 2012, 137, 2651–2657.18 M. E. Pekdemir, D. Erturkan, H. Kulah, I. H. Boyaci, C. Ozgen

and U. Tamer, Analyst, 2012, 137, 4834–4840.19 K. Kneipp, H. Kneipp, I. Itzkan, R. R. Dasari and M. S. Feld,

Chem. Rev., 1999, 99, 2957–2976.20 M. R. Jones, K. D. Osberg, R. J. Macfarlane, M. R. Langille

and C. A. Mirkin, Chem. Rev., 2011, 111, 3736–3827.21 L. A. Dick, A. D. McFarland, C. L. Haynes and R. P. Van

Duyne, J. Phys. Chem. B, 2002, 106, 853–860.22 Q. Yu, P. Guan, D. Qin, G. Golden and P. M. Wallace, Nano

Lett., 2008, 8, 1923–1928.23 G. Hong, C. Li and Q. Limin, Adv. Funct. Mater., 2010, 20,

3774–3783.24 P. M. T. O. D. Velev, A. M. Lenhoff and E. W. Kaler, Nature,

1999, 401, 548–548.25 O. D. Velev, T. A. Jede, R. F. Lobo and A. M. Lenhoff, Chem.

Mater., 1998, 10, 3597–3602.26 H. F. Ghaemi, T. Thio, D. E. Grupp, T. W. Ebbesen and

H. J. Lezec, Phys. Rev. B: Condens. Matter, 1998, 58,6779.

27 Q. Zhou, J. J. Zhao, W. W. Xu, H. Zhao, Y. Wu andJ. W. Zheng, J. Phys. Chem. C, 2008, 112, 2378–2381.

28 M. Xu, N. Lu, H. Xu, D. Qi, Y. Wang and L. Chi, Langmuir,2009, 25, 11216–11220.

29 P. N. Bartlett, J. J. Baumberg, S. Coyle and M. E. Abdelsalam,Faraday Discuss., 2004, 125, 117–132.

30 P. N. Bartlett, J. J. Baumberg, P. R. Birkin, M. A. Ghanem andM. C. Netti, Chem. Mater., 2002, 14, 2199–2208.

31 Y. Lei, S. Yang, M. Wu and G. Wilde, Chem. Soc. Rev., 2011,40, 1247–1258.

32 P. Jiang, J. Cizeron, J. F. Bertone and V. L. Colvin, J. Am.Chem. Soc., 1999, 121, 7957–7958.

33 Y. Li, C. Li, S. O. Cho, G. Duan and W. Cai, Langmuir, 2007,23, 9802–9807.

34 L. Johnson and D. A. Walsh, J. Mater. Chem., 2011, 21,7555.

35 M. Abdelsalam, P. N. Bartlett, A. E. Russell, J. J. Baumberg,E. J. Calvo, N. S. G. Tognalli and A. Fainstein, Langmuir,2008, 24, 7018–7023.

36 R. P. Johnson, J. A. Richardson, T. Brown and P. N. Bartlett,J. Am. Chem. Soc., 2012, 134, 14099–14107.

37 S. Mahajan, J. Richardson, T. Brown and P. N. Bartlett, J. Am.Chem. Soc., 2008, 130, 15589–15601.

Analyst, 2013, 138, 2604–2612 | 2611

Analyst Paper

Dow

nloa

ded

by U

nive

rsity

of

Min

neso

ta -

Tw

in C

ities

on

19/0

4/20

13 1

1:56

:22.

Pu

blis

hed

on 0

7 Fe

brua

ry 2

013

on h

ttp://

pubs

.rsc

.org

| do

i:10.

1039

/C3A

N36

792D

View Article Online

38 J. D. Speed, R. P. Johnson, J. T. Hugall, N. N. Lal,P. N. Bartlett, J. J. Baumberg and A. E. Russell, Chem.Commun., 2011, 47, 6335–6337.

39 F. M. Huang, D. Wilding, J. D. Speed, A. E. Russell,P. N. Bartlett and J. J. Baumberg, Nano Lett., 2011, 11,1221–1226.

40 X. X. Han, L. Chen, J. Guo, B. Zhao and Y. Ozaki, Anal. Chem.,2010, 82, 4102–4106.

41 S. Chen, Y. Yuan, J. Yao, S. Han and R. Gu, Chem. Commun.,2011, 47, 4225–4227.

42 M. M. Harper, J. A. Dougan, N. C. Shand, D. Graham andK. Faulds, Analyst, 2012, 137, 2063–2068.

43 Y. Cui, B. Ren, J.-L. Yao, R.-A. Gu and Z.-Q. Tian, J. RamanSpectrosc., 2007, 38, 896–902.

44 A. Otto, et al., J. Phys.: Condens. Matter, 1992, 4, 1143.45 X. X. Han, Y. Kitahama, T. Itoh, C. X. Wang, B. Zhao and

Y. Ozaki, Anal. Chem., 2009, 81, 3350–3355.46 A. Altube, A. Blanco and C. Lopez, Mater. Lett., 2008, 62,

2677–2680.47 P. C. Lee and D. Meisel, J. Phys. Chem., 1982, 86, 3391–3395.48 N. G. Tognalli, A. Fainstein, E. J. Calvo, M. Abdelsalarn and

P. N. Bartlett, J. Phys. Chem. C, 2012, 116, 3414–3420.

2612 | Analyst, 2013, 138, 2604–2612

49 T. A. Kelf, Y. Sugawara, R. M. Cole, J. J. Baumberg,M. E. Abdelsalam, S. Cintra, S. Mahajan, A. E. Russell andP. N. Bartlett, Phys. Rev. B: Condens. Matter Mater. Phys.,2006, 74.

50 S. Cintra, M. E. Abdelsalam, P. N. Bartlett, J. J. Baumberg,T. A. Kelf, Y. Sugawara and A. E. Russell, Faraday Discuss.,2006, 132, 191–199.

51 M. Osawa, N. Matsuda, K. Yoshii and I. Uchida, J. Phys.Chem., 1994, 98, 12702–12707.

52 C. L. Haynes and R. P. Van Duyne, J. Phys. Chem. B, 2003, 107,7426–7433.

53 X.-M. Lin, Y. Cui, Y.-H. Xu, B. Ren and Z.-Q. Tian, Anal.Bioanal. Chem., 2009, 394, 1729–1745.

54 L. H. Lu, A. Eychmuller, A. Kobayashi, Y. Hirano, K. Yoshida,Y. Kikkawa, K. Tawa and Y. Ozaki, Langmuir, 2006, 22, 2605–2609.

55 M. C. Netti, S. Coyle, J. J. Baumberg, M. A. Ghanem,P. R. Birkin, P. N. Bartlett and D. M. Whittaker, Adv.Mater., 2001, 13, 1368–1370.

56 S. Coyle, M. C. Netti, J. J. Baumberg, M. A. Ghanem,P. R. Birkin, P. N. Bartlett and D. M. Whittaker, Phys. Rev.Lett., 2001, 87.

This journal is ª The Royal Society of Chemistry 2013