Near-Field Interactions between Metal Nanoparticle … Interactions between Metal Nanoparticle...
Embed Size (px)
Transcript of Near-Field Interactions between Metal Nanoparticle … Interactions between Metal Nanoparticle...
Near-Field Interactions between Metal Nanoparticle SurfacePlasmons and Molecular Excitons in Thin-Films. Part I: AbsorptionBjoern Niesen,, Barry P. Rand,*, Pol Van Dorpe,, David Cheyns, Honghui Shen, Bjorn Maes,,
and Paul Heremans,
Imec, Kapeldreef 75, B-3001 Leuven, BelgiumDepartment of Electrical Engineering (ESAT), Katholieke Universiteit Leuven, Kasteelpark Arenberg 10, B-3001 Leuven, BelgiumPhotonics Research Group (INTEC), Ghent University-Imec, Sint-Pietersnieuwstraat 41, B-9000 Gent, BelgiumMicro- and Nanophotonic Materials Group, Faculty of Science, University of Mons, Avenue Maistriau 19, B-7000 Mons, Belgium
*S Supporting Information
ABSTRACT: In this and the following paper (parts I and II,respectively), we systematically study the interactions between surfaceplasmons of metal nanoparticles (NPs) with excitons in thin-films oforganic media. In an effort to exclusively probe near-field interactions,we utilize spherical Ag NPs in a size-regime where far-field lightscattering is negligibly small compared to absorption. In part I, wediscuss the effect of the presence of these Ag NPs on the absorptionof the embedding medium by means of experiment, numericalsimulations, and analytical calculations, all shown to be in goodagreement. We observe absorption enhancement in the embeddingmedium due to the Ag NPs with a strong dependence on the mediumpermittivity, the spectral position relative to the surface plasmon resonance frequency, and the thickness of the organic layer. Byintroducing a low-index spacer layer between the NPs and the organic medium, this absorption enhancement is experimentallyconfirmed to be a near-field effect. In part II, we probe the impact of the Ag NPs on the emission of organic molecules by time-resolved and steady-state photoluminescence measurements.
When excited by light of appropriate wavelength (), metalnanoparticles (NPs) exhibit so-called localized surface plasmonresonances (LSPRs) due to the collective oscillation of theirconduction electrons.1 The resonance wavelength (LSPR) andtherefore also light absorption and scattering due to LSPRs arehighly dependent on the composition, size, shape, and dielectricenvironment of the NPs.28 Owing to the strong near-fieldenhancement originating from the excitation of LSPRs,9,10
metal NPs can locally increase the absorption in the embeddingmedium. In molecular media, the absorption of a photon resultsin the generation of an exciton due to the transition of anelectron from the highest occupied to the lowest unoccupiedmolecular orbital. The absorption enhancement due to thepresence of metal NPs thus increases the amount of excitonsgenerated in the molecular medium. If not dissociated, theexcitons will eventually decay either radiatively by emittingphotons or nonradiatively by heat dissipation to the environ-ment. The presence of a nearby metal NP can strongly affectthe emission quantum yield by introducing additional non-radiative decay channels and enhancing the radiative decayrate.1113 These plasmonexciton interactions, together withthe sensitivity of the LSPR to the dielectric environment, givemetal nanoparticleorganic molecular systems potentialapplications in organic and dye-sensitized photovoltaic
cells,1419 organic light emitting devices,20,21 chemical andbiological sensors,2224 surface enhanced spectroscopy,25,26
photochemistry,27 and lasers.28,29
Recently, there has been a significant effort dedicated tosurface plasmonexciton interactions. Especially, their effect onthe absorption3033 and photoluminescence3336 of organicmolecules has been extensively studied. Despite the substantialresearch activity on this topic, a systematic study is still missingthat combines experiments, simulations, and analyticalcalculations to individually examine the effect of near-fieldLSPRexciton interactions on the absorption and emission ofmolecular thin-films. In particular, the influence of thedispersive permittivity of strongly absorbing molecular thin-films has not been studied thoroughly. In addition, only a fewstudies exclusively characterize the near-field plasmonexcitoninteractions using metal NPs that exhibit negligible far-fieldscattering.In this and the subsequent paper (parts I and II,
respectively), we utilize spherical metal NPs in a size-regimewhere far-field scattering is negligibly small compared toabsorption, to systematically probe near-field plasmonexciton
Received: June 15, 2012Revised: September 20, 2012Published: September 24, 2012
2012 American Chemical Society 24206 dx.doi.org/10.1021/jp305892e | J. Phys. Chem. C 2012, 116, 2420624214
interactions by means of absorption as well as steady-state andtime-resolved photoluminescence spectroscopy. We employ amodel system of a Ag NP layer on a transparent substrate thatis covered by a thin-film of a molecular medium either in directcontact with the Ag NPs or separated from them by atransparent, nonemissive spacer layer. In part I, we examine theeffect of the Ag NP layer on the absorption of organicmolecules. We observe absorption enhancement in thepresence of the NPs with a strong dependence on thethickness of the organic layer and the wavelength of incidentlight. Numerical simulations show good quantitative agreementwith experiment, which we could also reproduce qualitativelyby an analytical model. The near-field nature of the absorptionenhancement is experimentally confirmed by introducing aspacer layer between the Ag NP layer and the organic medium.Finally, we observe a strong dependence of the absorptionenhancement on the dielectric environment of the Ag NPs. Inpart II,37 we analyze the impact of the Ag NP layer on theemission of organic molecules by steady-state and time-resolvedphotoluminescence measurements. These experimental resultsare then discussed utilizing exciton decay rates obtained by ananalytical model.
2. EXPERIMENTAL METHODS
Copper(II) phthalocyanine (CuPc) was purchased fromSigmaAldrich, and chloroboron subphtalocyanine (SubPc)was purchased from the Luminescence Technology Corp. Bothmaterials were purified twice by thermal gradient sublimationprior to use. Hyflon AD60X, a copolymer of 2,2,4-trifluoro-5-trifluorometoxy-1,3-dioxole and tetra-fluoro-ethylene with amolar ratio of 0.6:0.4, was obtained from SolvaySolexis andused as received. Silver with a purity of 99.99% was purchasedfrom the Kurt. J. Lesker Company. Eagle XG glass (purchasedfrom Corning Inc.) and Si/SiO2 substrates were cleaned bysubsequent sonication in soapy water, deionized water, acetone,and isoproyl alcohol, followed by exposure to UVozone for 15min. Silver, CuPc, and SubPc were deposited by thermalevaporation at a chamber pressure below 2 106 Torr. Silverwas deposited at a rate of 0.01 nm/s; CuPc and SubPc weredeposited at a rate of 0.1 nm/s. Hyflon AD60X thin-films weredeposited in the same chamber by means of vacuumpyrolysis,38 at a pressure below 5 106 Torr and a rate of0.05 nm/s. During all depositions, the rate and film thicknesswere monitored by a quartz-crystal oscillator. The substrate waskept at room temperature and was rotating during thedeposition, leading to a thickness inhomogeneity of 350 nm and 2% for < 350 nm (see the SupportingInformation, Figure S2). The morphology of the Ag NP layerwas determined by a Hitachi SU8000 scanning electronmicroscope. The size-distribution of the Ag NPs wasdetermined from scanning electron micrographs usingGwyddion 2.18. The relative permittivity, = 1 + i2, of allmaterials was determined from their thin-films on Si/SiO2substrates, using a SOPRA GESP-5 spectroscopic ellipsometer.
3. RESULTS AND DISCUSSION
3.1. Ag NP Layer. The deposition of a Ag layer with anominal thickness of 1 nm on a glass substrate by means ofthermal evaporation spontaneously leads to the formation of adense Ag NP layer. As shown by the top-view scanning electronmicrograph in Figure 1a, the Ag NPs approximately have acircular shape. The particle diameter distribution obtained fromthis micrograph has average and maximum values of 6.6 and 12nm, respectively, with a standard deviation of 1.4 nm (Figure1b). From atomic force microscopy measurements,39 anaverage height of 5 nm was obtained. The absorption spectrumof the bare Ag NP layer, shown as a black dotted curve inFigure 2a, exhibits a single absorption band centered at = 425nm, which originates from a dipole LSPR.39 As expected forNPs of this size, there are no additional absorption bands dueto higher order modes.40 The absorption band extending from = 300320 nm originates from Ag interband transitions.41In general, when describing the light extinction of metal NPs,
both absorption as well as light scattering have to be taken intoaccount. The extinction is then given by the sum of absorptionand scattering. According to electrodynamic Mie theory,42 for
Figure 1. (a) Top-view scanning electron micrograph of the Ag NP layer on a Si/SiO2 substrate. (b) Histogram of the Ag NP size-distributionobtained from 474 particles found in the micrograph. The average diameter is 6.6 nm, with a standard deviation of 1.4 nm.
The Journal of Physical Chemistry C Article
dx.doi.org/10.1021/jp305892e | J. Phys. Chem. C 2012, 116, 242062421424207
spherical Ag NPs with a diameter of less than 12 nm in air, thescattering-to-extinction ratio remains below 7 103 (see theSupporting Information, Figure S3). This ratio increases withincreasing relative permittivity of the (nonabsorbing) embed-ding medium (m) and reaches values of up to 0.16 for m = 10.Light scattering thus only becomes significant for the extremecases of the largest particles in the NP layer and very high m.Recently, a strong influence of light scattering in absorbingmedia even for small spherical Ag NPs was predicted from anextended Mie theory.43 However, far-field light scattering wasfound to be negligible for the Ag NP layer employed here, evenwhen it was coated with absorbing organic media (see theSupporting Information, Figure S1). Throughout this part ofthe study, the light extinction will therefore be approximated bythe absorption, and far-field scattering will be neglected. As wewill see later, this approximation does not hold when discussingthe effect of the Ag NPs on light emission in part II.37
3.2. Absorption Enhancement in CuPc. We chose CuPc,a well-characterized organic semiconductor, as a representativematerial to study the effect of the Ag NP layer on theabsorption of organic molecules. A major advantage of thismaterial choice is the complementary nature of the absorptionspectra of CuPc and the Ag NP layer. The absorption spectrumof a 10 nm thick CuPc layer on glass is shown as a red dashedcurve in Figure 2a. It features two absorption bands: the Soretband centered at = 345 nm and the low energy Q band withpeaks at = 625 and 690 nm. The absorption spectrum of 10nm CuPc deposited on the Ag NP layer (black solid curve)exhibits the same features as well as the LSPR absorption bandof the Ag NPs. The LSPR is slightly red-shifted to = 450 nmcompared to that of the bare Ag NP layer due to the change inthe dielectric environment.44,45 The absorption differencebetween the sample with and without Ag NPs is shown as ablue curve. It features a band at 550 nm < < 800 nm that isnot present in the absorption spectrum of the uncovered AgNP layer. This increase in absorption has been attributed to anabsorption enhancement in the CuPc medium caused by the AgNP LSPR.30,43 In the following paragraphs, this absorptionenhancement will be characterized in detail, and it will bediscussed whether it exclusively originates from the Ag NPLSPR.To this end, three-dimensional (3D) finite-element numer-
ical simulations were made. The geometry used for thesesimulations was motivated by our microscopy study (see Figure1). Its cross section is shown as an inset to Figure 2b. The AgNPs are represented by truncated spheroids with a height of 5nm and a lateral dimension of 7 nm. Periodic boundaryconditions in the lateral directions lead to a Ag NP array with acenter-to-center spacing denoted by d. An organic layer with athickness t and a flat surface covers the Ag NPs. The light issimulated by a plane wave with the propagation vector, k,parallel to the y axis and the electric field vector, E0, parallel tothe x axis. It is incident through the semi-infinite glass substratewith = 2.25. The values used for these simulations areshown in the Supporting Information, Figure S4. As areference, neat organic films without Ag NPs were simulated.When the simulation results from such a reference film arecompared to those of an organic film on Ag NPs, the thicknessof the film on Ag NPs is chosen slightly larger than that of theneat reference film such that both organic layers have the samevolume, compensating for the volume occupied by the Ag NPs.The software package used for these simulations, COMSOL
Multiphysics 3.5a, allowed us to monitor the absorption in eachcomponent individually and therefore to deconstruct theabsorption spectra of composite layers as shown in Figure 2b.The simulated absorption of a pure 10 nm thick CuPc layer andthat of a CuPc layer with equivalent t covering the Ag NPs areshown as dashed red and solid black curves, respectively. Theabsorption in the Q band is significantly enhanced in thepresence of the Ag NPs, whereas the Soret band remainsunchanged. The simulated absorption in the Ag NPs whencovered by CuPc (green dash-dot-dot curve) exhibits a sharppeak at = 460 nm, red-shifted compared to the absorptionpeak of the uncovered NPs at = 390 nm (dotted black curve).In addition, it also features a very weak absorption band in theCuPc Q band spectral region, which is absent for the uncoveredNPs.Combined, these simulated spectra of the individual
components accurately describe the experimentally obtainedspectra shown in Figure 2a, including the absence of absorption
Figure 2. (a) Absorption spectra of the bare Ag NP layer (dotted blackcurve), the Ag NP layer coated by 10 nm of CuPc (solid black curve),and a 10 nm thick CuPc layer on glass (dashed red curve). Theabsorption difference between the CuPc layers with and without AgNPs is shown as a blue dash-dot-dot curve. (b) Results obtained fromthree-dimensional numerical simulations using a geometry with a crosssection as shown in the inset. Periodic boundaries in both lateraldimensions result in an infinite two-dimensional array of truncated Agspheroids with center-to-center spacing d. The light wave is incidentthrough the glass substrate with the propagation vector, k, and electricfield vector, E0, as indicated by arrows. Further details of thesimulation are given in the text. This graph shows the simulatedabsorption in the bare Ag NPs (dotted black curve), a pure 10 nmthick CuPc layer (dashed red curve), a CuPc layer covering the AgNPs with t equivalent to the pure CuPc layer (solid black curve), andthe Ag NPs when covered by that layer (green dash-dot-dot curve).For this simulation, d was set to 18 nm.
The Journal of Physical Chemistry C Article
dx.doi.org/10.1021/jp305892e | J. Phys. Chem. C 2012, 116, 242062421424208
enhancement at < LSPR. Moreover, these results suggest thatmost of the absorption in the spectral region of the CuPc Qband occurs within the organic medium and not in the NPs,which is in line with previously reported results from analyticaland numerical calculations.43 The slight mismatch of LSPR andthe narrower absorption band compared to the experimentallyobserved results can be explained by the uniform size and shapeof the simulated particles and the uncertainty about the exactNP shape.In order to determine the range of the absorption
enhancement, samples with various CuPc layer thicknesseswere fabricated. The absorption spectra of the samples with andwithout Ag NPs, shown in Figure 3a as solid and dashed curves,
respectively, clearly show a decreasing influence of the Ag NPswith increasing CuPc layer thickness. The absorption enhance-ment at = 690 nm obtained from these measurements isshown in Figure 3b as black squares. It decreases from 1.44 0.08 for a 8 nm thick CuPc layer to 1.06 0.03 for a 50 nmthick layer. This experimentally measured thickness depend-ence could be reproduced by numerical simulations (gray areain Figure 3b). Quantitative agreement was obtained when d wasset to 18 nm for the thinnest and to 14 nm for the thickest
CuPc layer. Both these values are in line with typical center-to-center distances observed in Figure 1a. A possible cause for thedeviation between experiment and simulation could be theroughness of the CuPc layer surface, which is not taken intoaccount in the simulations and becomes increasingly importantfor thinner layers.The absorption enhancement obtained by these simulations
is equivalent to the enhancement of the E field intensity due tothe Ag NPs, = |ENP|
2/|E0|2, where E0 and ENP denote the E
field without NPs and in their presence, respectively. Thecontour-plot of at = 690 nm in the xy plane cuttingthrough the center of the simulated Ag NP therefore yieldsdirect information about the range of the absorption enhance-ment (see Figure 4a). The absorption is strongly enhanced
between the NPs, whereas it is reduced above the NPs asevidenced by < 1. The border between enhanced and reducedabsorption (the = 1 contour) extends from the top of the NPto the border of the simulation halfway to the neighboring NPwhere it levels off at a height of 12 nm. The very large fieldenhancement close to the Ag NP surface observed in this graphmight possibly not contribute to absorption enhancement inreal-world experiments as the distance between a Ag surfaceand the first CuPc monolayer is about 0.3 nm.46 However, asthe exact morphology of the Ag NPCuPc interface is notknown and the volume of the void is very small compared tothe total CuPc layer volume, we did not take this metalmolecule separation into account, which means that thesimulated absorption enhancement will be slightly over-estimated. Figure 4b shows the average value integratedover a slice of CuPc medium in the xz plane versus the heightalong the y axis for CuPc layer thicknesses of 10 nm (solid redcurve) and 50 nm (dashed black curve). is nearly identical forboth layer thicknesses, with > 1 up to a height of 7 nm. Abovethis height, converges to a value of 0.94.These simulation results allow us to explain the exper-
imentally observed CuPc layer thickness dependence shown inFigure 3. As only the first 7 nm of the CuPc medium benefitfrom the absorption enhancement, increasing the CuPc layerthickness beyond that height will lead to a reduction inabsorption enhancement relative to the total absorption of thelayer. Moreover, |E0|
2 decreases with increasing CuPc layer
Figure 3. (a) Absorption spectra of CuPc layers deposited on glass(dashed curves) or on the Ag NP layer (solid curves) for CuPc layerthicknesses of 10, 20, 30, and 50 nm. (b) Absorption enhancementdue to the Ag NP layer at = 690 nm determined by dividing theabsorption of the Ag NP layer coated with CuPc by that of pure CuPclayers for CuPc layer thicknesses between 8 and 50 nm (blacksquares). The gray area shows the absorption enhancement at thesame wavelength and for the same CuPc thickness range obtained bydividing the simulated absorption in the CuPc layer covering the AgNPs by that of a CuPc layer on glass with equivalent t. Center-to-center spacing (d) values of 14 and 18 nm were used for thesesimulations.
Figure 4. (a) Simulated around a Ag NP when covered by a 50 nmthick CuPc layer. (b) Simulated average value integrated over a sliceof CuPc medium in the xz plane versus the height along the y axis fora 10 nm (red solid curve) and 50 nm (dashed black curve) thick layer.The solid blue curves show the simulated |E0|
2 versus the y axis in neatCuPc layers with thicknesses of 10, 15, 20, 30, and 50 nm. d was set to18 nm for all simulations shown in this figure. The simulation resultsshown here were obtained at = 690 nm.
The Journal of Physical Chemistry C Article
dx.doi.org/10.1021/jp305892e | J. Phys. Chem. C 2012, 116, 242062421424209
thickness (blue curves in Figure 4b), which means that, for thesame value, |ENP|
2 |E0|2 and thus also the absorptiondifference due to the Ag NPs will be smaller for thicker layers.A rather subtle feature of the absorption enhancement can be
observed when analyzing the CuPc Q band spectral region (550nm < < 800 nm) of the absorption spectra shown in Figure 2.The normalized experimentally measured absorption spectra ofthe samples with and without Ag NPs are shown in Figure 5 as
black squares and red triangles, respectively. The normalizedsimulated spectra are shown as solid lines. The relativeintensities of the peaks at = 690 nm in the experimentaland simulated spectra are enhanced by a similar amount in thepresence of the Ag NPs. Considering the experimental resultsonly, this change in relative peak intensity might be interpretedas a change in the CuPc layer morphology when grown on theAg NP substrate, as this ratio is an indicator of the degree ofmolecular aggregation within the CuPc layer.47 An increasedrelative intensity of the peak at = 690 nm might indicate adecrease in CuPc aggregation. However, the numericalsimulations, that do not take into account any change in theinteractions between the CuPc molecules or between themolecules and the substrate when introducing the NPs,accurately reproduce this change in relative peak intensities.Therefore, we conclude that the spectral shape of theabsorption enhancement can be reproduced by only takinginto account the wavelength-dependent enhancement of theelectromagnetic field intensity around the NPs.In addition to numerical simulations, the absorption
enhancement due to the Ag NPs can also be describedanalytically. Because of the small Ag NP size relative to thewavelengths of the measured spectra, the E field across anilluminated NP is approximately constant and the so-calledquasi-static approximation can be employed. According to thisapproximation, the dipole polarizability, (), of a sphericalmetal NP with radius R in a homogeneous medium is given by1
R( ) 4( ) ( )
( ) 2 ( )3 NP m
NP m (1)
where denotes the frequency of the incident electromagneticwave and NP is the relative permittivity of the metal NP. In a
Figure 5. Normalized measured absorption spectra of 10 nm CuPcdeposited on glass (solid red triangles) and on the Ag NP layer (openblack squares), normalized simulated absorption in a 10 nm thickCuPc layer on glass (solid red curve) and in a CuPc layer withequivalent t covering the Ag NPs (solid black curve), and theabsorption spectrum obtained by normalizing the product of theexperimentally measured absorption of 10 nm CuPc on glass and maxfor a Ag NP in CuPc obtained by eq 2 (dashed blue curve). The valuesof r and R used to calculate max are not important as the results werefinally normalized.
Figure 6. (a, b) Simulated absorption in the Ag NPs when covered by a layer with a volume identical to that of a 10 nm thick neat layer for various(a) 1 and (b) 2. (c, d) Simulated absorption enhancement in the layer covering the Ag NPs compared to a 10 nm thick plain layer. d was set to 18nm for all these simulations.
The Journal of Physical Chemistry C Article
dx.doi.org/10.1021/jp305892e | J. Phys. Chem. C 2012, 116, 242062421424210
nonabsorbing medium, the maximal E field intensity enhance-ment, max(, r), at the distance r to the center of the NP isgiven by10,48
( , ) 1( )
As eq 2 is only exact for nonabsorbing media, max can merelybe considered as an upper limit for the E field enhancement inabsorbing media, where an accurate treatment requiresemploying an electrodynamic formalism.43,49 For this reason,and because it assumes a spherical geometry, eq 2 will onlyyield qualitative information when applied to the thin-filmsystem studied here. We calculated max for Ag NPs embeddedin CuPc by inserting the values of Ag and CuPc shown inFigure S4 of the Supporting Information into eqs 1 and 2. Theresulting value was then multiplied with the experimentallymeasured absorption spectrum of the neat 10 nm thick CuPclayer and finally normalized. This calculated absorptionspectrum is shown as a dashed blue curve in Figure 5. It isnearly identical to the simulated spectrum of the compositelayer and thus also well in agreement with the experimentaldata. This means that the spectral shape of the absorptionenhancement in the CuPc Q band is directly related to NP andm. This also suggests that the influence of small deviations ofthe NP shape from a perfect sphere as well as particlesubstrate and particleparticle interactions are negligible. Thelimited influence of the substrate can be explained by its low compared to that of CuPc in this spectral region, such that theLPSR is more strongly affected by the organic medium. Thenegligible contribution of particleparticle interactions isexpected for a Ag NP layer where the distance between theNPs is similar to their diameter, as such interactions would onlybecome relevant for denser NP layers.50
3.3. Spectral Dependence and Influence of . Afterhaving confirmed the validity of numerical simulations todescribe the absorption enhancement in CuPc, we can extendthese simulations to other embedding materials with spectrallyindependent = 1 + i2, employing the same geometry as usedfor CuPc. Figure 6a and b shows the simulated absorption inthe Ag NPs when covered by these various materials. When 2of the material is set to 1 and 1 is altered from 1 to 8, the LSPRabsorption peak of the Ag NPs red-shifts from = 390 to 660nm while its intensity increases by almost 1 order of magnitude(see Figure 6a). For constant 1 = 3, a typical value for manyorganic materials, an increase in 2 dampens the resonanceintensity such that the absorption peak completely vanishes for2 4, whereas the resonance frequency is not affected (seeFigure 6b). The simulated absorption enhancement in theseembedding media due to the Ag NPs is shown in Figure 6c andd. It scales with the intensity of the LSPR and is strongest at slightly larger than LSPR. For this set of materials, values of upto 3.2 are reached for = 3 + 0.5i and = 8 + 1i. For all ofthese materials, the enhancement steeply decreases at < LSPR,whereas at > LSPR, it exhibits an extended tail with valuesabove unity up to = 800 nm. For highly absorbing media (2 2), the absorption enhancement only reaches values below1.5, with an off-resonance enhancement that is comparable oreven stronger than that at LSPR. This strong dependence of theabsorption enhancement on 2 is consistent with previouslyreported results obtained using an analytical model.51 Insummary, the strongest absorption enhancement is thusreached for materials with a large 1 and a weak absorption,
resulting in a small 2. Typically, the enhancement is strongestslightly red-shifted to LSPR. However, exploiting the off-resonance enhancement rather than the stronger enhancementat LSPR can still be beneficial as the absorption in the Ag NPs ismuch smaller far away from the resonance. This effect isespecially relevant for applications where reducing parasiticabsorption of the metal NPs is advantageous. The spatialdistribution of the absorption enhancement was found to onlyweakly depend on the spectral position, as shown for = 3 +0.5i in the Supporting Information, Figure S5. Off-resonance aswell as at LSPR, the absorption enhancement is limited to aheight of 7 nm, such that the stronger enhancement at LSPR ismainly due to higher E field intensities in the interstices of theNPs.We can apply the knowledge gained from these simulation
results to explain the spectral shape of the CuPc Q bandabsorption enhancement. For a material with constant 3 +2i, the absorption enhancement is expected to slightly decreasewith increasing for > LSPR. However, in the case of CuPc,this decrease is outweighed by a larger 1 and smaller 2 at =690 nm compared to the at = 625 nm, leading to a strongerabsorption enhancement at = 690 nm (see the SupportingInformation, Figure S4). This effect is even more pronouncedfor SubPc, a material with a highly dispersive . Whendeposited on the Ag NP layer, the relatively low value of 1 at400 nm < < 550 nm results in the excitation of a LSPR at =445 nm. Due to the narrow and intense absorption band ofSubPc at = 590 nm (dashed black curve in Figure 7), 1
steeply increases at = 570600 nm and reaches values of upto 10.5. This leads to the excitation of a second dipole LSPR at = 610 nm, as reported previously.39 These two LSPR areevidenced by an additional absorption band peaking at = 445nm, as well as an extension of the red tail of the SubPcabsorption band centered at = 590 nm (solid red curve).
3.4. Effect of a Spacer Layer. We studied the effect of aspacer layer on the plasmonexciton interactions for bothCuPc and SubPc by introducing a layer of Hyflon AD60Xbetween the Ag NP layer and the organic medium. Thin-filmsof Hyflon AD60X were found to be completely transparent inthe measured spectral range. Moreover, because of the small 1of Hyflon AD60X, the LSPR of the Ag NPs was only slightlyaltered when covered with it (see the Supporting Information,Figures S4 and S6). We fabricated samples without a spacer andones that had spacer layers with thicknesses of 120 nm. Foreach spacer layer thickness, we fabricated a sample with and
Figure 7. Absorption spectra of 10 nm thick layers of SubPc depositedon glass (dashed black curve) and on the Ag NP layer (solid redcurve).
The Journal of Physical Chemistry C Article
dx.doi.org/10.1021/jp305892e | J. Phys. Chem. C 2012, 116, 242062421424211
one without Ag NPs. For the reference samples without AgNPs, Hyflon AD60X was directly deposited on the glasssubstrate. The spectra of the absorption differences between thesamples with and without Ag NPs are shown in Figure 8a and bfor 10 nm thick layers of CuPc and SubPc, respectively.
The absorption difference in the CuPc Q band rapidlydecreases with increasing spacer layer thickness and finallyvanishes for a 5 nm thick spacer, as shown by Figure 8a. Forspacer layer thicknesses of 520 nm, the absorption differencesthus resemble the absorption curve of the Ag NP layer coveredonly by Hyflon AD60X. The absorption enhancement at =690 nm shown in the inset quantifies this rapid decrease, whichis much steeper than that observed for an increasing CuPc layerthickness without spacer layer (see Figure 3b). The SubPcsamples (see Figure 8b) also show a vanishing absorptiondifference for spacer layers thicknesses beyond 3 nm but with anotably different behavior for the three thinnest spacer layerscompared to the CuPc samples. In particular, the peak observedat = 610 nm without a spacer layer gives way to a shoulder at = 600 nm and a peak at = 595 nm for Hyflon AD60X layerthicknesses of 1 and 3 nm, respectively.In order to gain a better understanding of these results, we
conducted numerical simulations using the same geometry asshown in the inset to Figure 2b but with an additionalconformal Hyflon AD60X layer separating the Ag NPs from theSubPc layer. The simulated absorption in the Ag NPs and theincrease in absorption in the SubPc layer caused by the NPs are
shown in Figure 9a and b, respectively. Without the spacerlayer, the absorption in the Ag NPs shows two bands centered
at = 445 and 630 nm, respectively, due to the two dipoleLSPR. These two bands lead to strong absorption enhancementin the SubPc layer at these wavelengths. The intensity of the AgNP absorption band at = 630 nm rapidly decreases withincreasing spacer layer thickness and completely vanishes at aspacer thickness of 1 nm. The increase in absorption in theSubPc layer, on the other hand, remains for all simulated spacerlayer thicknesses and features the shoulder-to-peak transitionaround = 600 nm observed experimentally (see Figure 8b).For both CuPc and SubPc, the threshold spacer thickness for
absorption enhancement is between 3 and 5 nm, which is aboutthe same as the Ag NP height. These results confirm that theabsorption enhancement occurs within a few nanometers fromthe Ag NP surface independent of the embedding medium. Inaddition, the effect of m on the absorption enhancement canindirectly be observed. By increasing the thickness of a spacerlayer with a low , the effective m gradually decreases from thatof the initial organic medium to that of the spacer layer. ForCuPc with 1 < 4.5, this transition is not too dramatic. ForSubPc with 1 values of up to 10.5, on the other hand, it leadsto large changes in the spectral shape. Especially, the excitationof the LSPR at = 610 nm is extremely sensitive to theembedding medium and is suppressed by a spacer layer of only1 nm. The LSPR at = 445 nm on the other hand is much lesseffected by the spacer layer as at this wavelength the of SubPcis similar to that of Hyflon AD60X. It is noteworthy that thesimulated absorption in the Ag NPs for a spacer layer thicknessof 0.2 nm describes the experimentally obtained absorption
Figure 8. Absorption differences between samples with and withoutAg NPs. These samples were made by depositing Hyflon AD60Xlayers with thicknesses of 1, 3, 5, 10, and 20 nm either on the Ag NPlayer or on glass, followed by a 10 nm thick layer of either (a) CuPc or(b) SubPc. For both CuPc and SubPc, the absorption differences ofsamples without a Hyflon AD60X layer are shown as black curves. Theinset to (a) shows the absorption enhancement ratio at = 690 nmversus the Hyflon AD60X layer thickness for the CuPc samples.
Figure 9. (a) Simulated absorption spectra in the Ag NPs whencovered by a SubPc layer with t equivalent to a 10 nm thick neat layer,with and without a conformal spacer layer of Hyflon AD60Xseparating the NPs from the SubPc layer. (b) Difference betweenthe absorption in the SubPc layer covering the Ag NPs and a 10 nmthick neat SubPc layer, with and without Hyflon AD60X spacer layer.
The Journal of Physical Chemistry C Article
dx.doi.org/10.1021/jp305892e | J. Phys. Chem. C 2012, 116, 242062421424212
difference for SubPc without a spacer layer much moreaccurately than the simulation without a spacer layer. As alreadydiscussed for the numerical simulation of the Ag NPsembedded in CuPc, this distance roughly corresponds to thedistance between the Ag surface and the first molecular adlayer(see Section 3.2). For SubPc, where the low energy LSPRstrongly depends on m, the presence of this small gap makes aconsiderable difference.
We studied the absorption of thin-films comprising Ag NPsthat show negligible far-field scattering embedded in molecularthin-films by means of experiment, numerical simulations, andanalytical calculations. For CuPc, we observed an increasedintensity of the absorption Q band due to the Ag NP layer withabsorption enhancement values of up to 1.44 0.08 at = 690nm for an 8 nm thick CuPc layer. A strong dependence of thisabsorption enhancement on the CuPc layer thickness wasfound. We could quantitatively reproduce these experimentalresults by numerical simulations. From these simulations, theabsorption enhancement was found to mainly occur in theCuPc medium, caused by an enhanced E field intensity in theinterstices of the Ag NPs. The short range of this enhancementwas confirmed experimentally by introducing a transparentspacer layer between the NPs and the CuPc layer, which led toa rapid decrease of the absorption enhancement with increasingspacer thickness. By employing an analytical model, the spectralshape of the CuPc absorption enhancement could be directlylinked to the values of the NP material and the embeddingmedium, while neglecting particleparticle and particlesurface interactions. We studied the effect of m on theplasmonexciton interactions by means of numerical simu-lations of materials with spectrally independent and foundabsorption enhancement at > LSPR, which extends to thenear-infrared for values typical of molecular media. Whenintroducing a low-index spacer between the Ag NPs and a layerof SubPc, we observed a complex behavior of the spectral shapeof the absorption enhancement due to the Ag NPs, which weattribute to the decrease of the effective m with increasingspacer thickness.Our results show that care has to be taken on both the
geometrical features and the material choice when designingmetal NPorganic molecule structures with the aim to enhancethe absorption in the organic medium. For example, by coatingmetal NPs with a transparent layer of only a few nanometers,the absorption enhancement in an absorbing embeddingmolecular medium due to the NPs can be strongly affectedand perhaps eliminated completely. In spite of the complexbehavior of the plasmonexciton interactions, we show that, atleast in the case of CuPc and SubPc, the absorptionenhancement in the embedding molecular medium due tometal NPs can be predicted accurately by numerical simulationsthat do not take the organic layer morphology into account.Furthermore, if the NPs are well separated, a simple analyticalformula provides quantitative information about the spectralshape of the absorption enhancement. In part II,37 we will buildupon the knowledge gained in this study and extend our workon plasmonexciton interactions to the effect of the Ag NPlayer on the emission of organic molecules.
ASSOCIATED CONTENT*S Supporting InformationComparison between the total and specular transmissionspectra, absorption spectra of several organic layers withidentical nominal thickness deposited during a single run,scattering and absorption cross section of a spherical Ag NPaccording to Mie theory, values of all material used in thisstudy, and absorption spectra of Hyflon AD60X thin-filmsdeposited on glass and on the Ag NP layer. This material isavailable free of charge via the Internet at http://pubs.acs.org.
AUTHOR INFORMATIONCorresponding Author*E-mail: [email protected] authors declare no competing financial interest.
ACKNOWLEDGMENTSWe gratefully acknowledge Josine Loo for obtaining scanningelectron micrographs. We thank Robert Gehlhaar for helpfuldiscussions, and SolvaySolexis for providing Hyflon AD60X.This work was supported by the European CommunitysSeventh Framework Programme under grant no. FP7-ICT-2009-4-248154 (PRIMA), the Institute for the Promotion ofInnovation by Science and Technology in Flanders (IWT) viathe SBO project no. 060843 (PolySpec), the BelgianInteruniversity Attraction Poles program under grant no. IAPP6-10 ([email protected]), and COST action MP 0702. P.V.D.thanks the Fonds Wetenschappelijk Onderzoek Vlaanderen(FWO) for financial support.
REFERENCES(1) Kreibig, U.; Vollmer, M. Optical Properties of Metal Clusters;Springer: Berlin, Germany, 1995.(2) Henry, A. I.; Bingham, J. M.; Ringe, E.; Marks, L. D.; Schatz, G.C.; Van Duyne, R. P. J. Phys. Chem. C 2011, 115, 92919305.(3) Rycenga, M.; Cobley, C. M.; Zeng, J.; Li, W.; Moran, C. H.;Zhang, Q.; Qin, D.; Xia, Y. Chem. Rev. 2011, 111, 36693712.(4) Tcherniak, A.; Ha, J. W.; Dominguez-Medina, S.; Slaughter, L. S.;Link, S. Nano Lett. 2010, 10, 13981404.(5) Noguez, C. J. Phys. Chem. C 2007, 111, 38063819.(6) Jain, P. K.; Lee, K. S.; El-Sayed, I. H.; El-Sayed, M. A. J. Phys.Chem. B 2006, 110, 72387248.(7) Muskens, O.; Billaud, P.; Broyer, M.; Del Fatti, N.; Vallee, F.Phys. Rev. B 2008, 78, 205410.(8) Miller, M. M.; Lazarides, A. A. J. Opt. A: Pure Appl. Opt. 2006, 8,S239249.(9) Maier, S. A. Plasmonics: Fundamentals and Applications; Springer:New York, 2007.(10) Tanabe, K. J. Phys. Chem. C 2008, 112, 1572115728.(11) Fort, E.; Gresillon, S. J. Phys. D: Appl. Phys. 2008, 41, 013001.(12) Gersten, J.; Nitzan, A. J. Chem. Phys. 1981, 75, 11391152.(13) Hartling, T.; Reichenbach, P.; Eng, L. M. Opt. Express 2007, 15,1280612817.(14) Atwater, H. A.; Polman, A. Nat. Mater. 2010, 9, 205213.(15) Dunbar, R. B.; Pfadler, T.; Schmidt-Mende, L. Opt. Express2012, 20, A177A189.(16) Niesen, B.; Rand, B. P.; Van Dorpe, P.; Cheyns, D.; Tong, L.;Dmitriev, A.; Heremans, P. Adv. Energy Mater. 2012, DOI: 10.1002/aenm.201200289.(17) Diukman, I.; Tzabari, L.; Berkovitch, N.; Tessler, N.; Orenstein,M. Opt. Express 2011, 19, A6471.(18) Qi, J.; Dang, X.; Hammond, P. T.; Belcher, A. M. ACS Nano2011, 5, 71087116.
The Journal of Physical Chemistry C Article
dx.doi.org/10.1021/jp305892e | J. Phys. Chem. C 2012, 116, 242062421424213
(19) Brown, M. D.; Suteewong, T.; Kumar, R. S. S.; DInnocenzo, V.;Petrozza, A.; Lee, M. M.; Wiesner, U.; Snaith, H. J. Nano Lett. 2011,11, 438445.(20) Choulis, S. A.; Mathai, M. K.; Choong, V.-E. Appl. Phys. Lett.2006, 88, 213503.(21) Fujiki, A.; Uemura, T.; Zettsu, N.; Akai-Kasaya, M.; Saito, A.;Kuwahara, Y. Appl. Phys. Lett. 2010, 96, 043307.(22) Mayer, K. M.; Hafner, J. H. Chem. Rev. 2011, 111, 38283857.(23) Haick, H. J. Phys. D: Appl. Phys. 2007, 40, 71737186.(24) De, M.; Ghosh, P. S.; Rotello, V. M. Adv. Mater. 2008, 20,42254241.(25) Schlucker, S. ChemPhysChem 2009, 10, 13441354.(26) Han, X. X.; Zhao, B.; Ozaki, Y. Anal. Bioanal. Chem. 2009, 394,17191727.(27) Mori, K.; Kawashima, M.; Che, M.; Yamashita, H. Angew. Chem.,Int. Ed. 2010, 49, 85988601.(28) Noginov, M. A.; Zhu, G.; Belgrave, A. M.; Bakker, R.; Shalaev,V. M.; Narimanov, E. E.; Stout, S.; Herz, E.; Suteewong, T.; Wiesner,U. Nature 2009, 460, 11101112.(29) Kena-Cohen, S.; Wiener, A.; Sivan, Y.; Stavrinou, P. N.; Bradley,D. D. C.; Horsfield, A.; Maier, S. A. ACS Nano 2011, 5, 99589965.(30) Rand, B. P.; Peumans, P.; Forrest, S. R. J. Appl. Phys. 2004, 96,7519.(31) Kulkarni, A. P.; Noone, K. M.; Munechika, K.; Guyer, S. R.;Ginger, D. S. Nano Lett. 2010, 10, 15011505.(32) Shen, H.; Bienstman, P.; Maes, B. J. Appl. Phys. 2009, 106,073109.(33) Khurgin, J. B.; Sun, G. J. Opt. Soc. Am. B 2009, 26, B8395.(34) Chen, Y.; Munechika, K.; Ginger, D. S. Nano Lett. 2007, 7,690696.(35) Bharadwaj, P.; Novotny, L. Opt. Express 2007, 15, 1426614274.(36) Cheng, Y.; Stakenborg, T.; Van Dorpe, P.; Lagae, L.; Wang, M.;Chen, H.; Borghs, G. Anal. Chem. 2011, 83, 13071314.(37) Niesen, B.; Rand, B. P.; Van Dorpe, P.; Cheyns, D.; Fron, E.;Van der Auweraer, M.; Heremans, P. J. Phys. Chem. C 2012,DOI: 10.1021/jp3058965.(38) Nason, T. C.; Moore, J. A.; Lu, T.-M. Appl. Phys. Lett. 1992, 60,18661868.(39) Niesen, B.; Rand, B. P.; Van Dorpe, P.; Shen, H.; Maes, B.;Genoe, J.; Heremans, P. Opt. Express 2010, 18, 1903219038.(40) Kreibig, U.; Schmitz, B.; Breuer, H. D. Phys. Rev. B 1987, 36,50275030.(41) See, K. C.; Spicer, J. B.; Brupbacher, J.; Zhang, D.; Vargo, T. G.J. Phys. Chem. B 2005, 109, 26932698.(42) Bohren, C. F.; Huffman, D. R. Absorption and Scattering of Lightby Small Particles; Wiley Interscience: New York, 1983.(43) Lee, J.-Y.; Peumans, P. Opt. Express 2010, 18, 1007810087.(44) Malinsky, M. D.; Kelly, K. L.; Schatz, G. C.; Van Duyne, R. P. J.Am. Chem. Soc. 2001, 123, 14711482.(45) Mock, J. J.; Smith, D. R.; Schultz, S. Nano Lett. 2003, 3, 485491.(46) Song, F.; Huang, H.; Dou, W.; Zhang, H.; Hu, Y.; Qian, H.; Li,H.; He, P.; Bao, S.; Chen, Q.; Zhou, W. J. Phys.: Condens. Matter 2007,19, 136002.(47) Rand, B. P.; Xue, J.; Uchida, S.; Forrest, S. R. J. Appl. Phys. 2005,98, 124902.(48) depends on the angle between the E and r vectors. It ismaximized when these vectors are parallel (Hence, the reduction of rto a scalar variable in eq 2).(49) Lagos, N.; Sigalas, M. M.; Lidorikis, E. Appl. Phys. Lett. 2011, 99,063304.(50) Romero, I.; Aizpurua, J.; Bryant, G. W.; Garca De Abajo, F. J.Opt. Express 2006, 14, 99889999.(51) Khurgin, J. B.; Sun, G.; Soref, R. A. Appl. Phys. Lett. 2009, 94,071103.
The Journal of Physical Chemistry C Article
dx.doi.org/10.1021/jp305892e | J. Phys. Chem. C 2012, 116, 242062421424214