Effects of Size, Shape, and Adsorbates on Ag Nanoparticles' LSPR, Papers of Philosophy

Download Effects of Size, Shape, and Adsorbates on Ag Nanoparticles' LSPR and more Papers Philosophy in PDF only on Docsity! 368 MRS BULLETIN • VOLUME 30 • MAY 2005 Introduction Plasmonics is an emerging branch of nanophotonics that examines the properties of the collective electronic excitations in noble metal films or nanoparticles known colloquially as surface plasmons. The excite- ment of plasmonics lies in its potential to achieve highly miniaturized and sensitive photonic devices by controlling, manipulat- ing, and amplifying light on the nanometer length scale.1–3 To date, a variety of passive plasmonic devices have been demonstrated, including filters,1 waveguides,1,3 polarizers,4 Bragg reflectors,1 and nanoscopic light sources.5 On the horizon are active plasmonic de- vices, such as light-output enhancers for organic light-emitting diodes6,7 as well as switches and modulators.8 Furthermore, our rapidly improving understanding of the interactions between adsorbed mole- cules and plasmonic nanostructures (i.e., molecular plasmonics)9 is having a signifi- cant impact on a broad spectrum of other applications, including nanoscale optical spectroscopy,10 surface-enhanced Raman spectroscopy,11 surface plasmon resonance sensing,12,13 and nanolithography.14 There are two types of surface plasmon resonance—localized and propagating. This article will mostly be concerned with the former, which we term localized surface plasmon resonance (LSPR). This occurs in silver and gold nanoparticles in the 10–200 nm size range and results in am- plification of the electric field E near the particle surfaces such that |E|2 can be 100–10,000 times greater in intensity than the incident field. The field has a spatial range on the order of 10–50 nm and is strongly dependent on nanoparticle size, shape, and local dielectric environment. Propagating plasmons, which are often called surface plasmon polaritons (SPPs), are associated with smooth, thin films of silver and gold with thicknesses in the 10–200 nm range. Propagating plasmons lead to smaller field enhancements (10–100 times) and a larger spatial range (
1000 nm). This article will focus on the fabrication and characterization of plasmonic materials that show promise in chemical/biological sensing and surface-enhanced spectroscopy applications. In the first part, the simple, massively parallel method of nanosphere lithography (NSL) and its use in the fabri- cation of size- and shape-controlled nano- structures is briefly reviewed. Also, the essential physics of LSPR and the theoreti- cal methods used to understand it are de- scribed, and key results concerning the short- and long-range distance depen- dences of the electromagnetic fields sur- rounding the nanoparticles are summarized. In the second part of this article, we focus on the relationship between LSPR spec- troscopy and surface-enhanced Raman spectroscopy (SERS), as revealed by surface- enhanced Raman excitation spectroscopy (SERES). SERES provides a systematic, re- producible way to optimize the signal in- tensity in SERS experiments. Nanosphere Lithography Nanosphere lithography (NSL)15 is a sur- prisingly powerful yet simple approach to the fabrication of nanoparticle arrays with precisely controlled shape, size, and interparticle spacing. Nanosphere lithography (Figure 1) be- gins with the self-assembly of monodis- perse polystyrene or SiO2 nanospheres of diameter D to form a single- or double- layer colloidal crystal mask for material deposition. A substrate (Figure 1a) is pre- pared so that the nanospheres can freely move until they reach their lowest energy configuration. This is achieved by chemi- cally modifying the nanosphere surface with a negative charge that is electrostati- cally repelled by a negatively charged substrate such as mica or chemically treated glass. As the solvent (water) evap- orates, capillary forces draw the nano- spheres together, and they crystallize into an hcp pattern on the substrate. As in all naturally occurring crystals, nanosphere masks include a variety of defects that arise as a result of nanosphere polydispersity, site randomness, point defects (vacancies), line defects (slip dislocations), and poly- crystalline domains. Typical defect-free do- main sizes are in the 10–100
m range. Following self-assembly of the nanosphere mask, a metal or other material is then de- posited by physical vapor deposition from a collimated source normal to the substrate through the nanosphere mask to a con- trolled thickness. The resulting surface is referred to as a metal (e.g., Ag) “film over nanosphere” (FON) surface. Ag FON sur- faces are robust plasmonic materials for Plasmonic Materials for Surface-Enhanced Sensing and Spectroscopy Amanda J. Haes, Christy L. Haynes, Adam D. McFarland, George C. Schatz, Richard P.Van Duyne, and Shengli Zou Abstract Localized surface plasmon resonance (LSPR) excitation in silver and gold nanoparticles produces strong extinction and scattering spectra that in recent years have been used for important sensing and spectroscopy applications.This article describes the fabrication, characterization, and computational electrodynamics of plasmonic materials that take advantage of this concept.Two applications of these plasmonic materials are presented: (1) the development of an ultrasensitive nanoscale optical biosensor based on LSPR wavelength-shift spectroscopy and (2) the use of plasmon-sampled and wavelength-scanned surface-enhanced Raman excitation spectroscopy (SERES) to provide new insight into the electromagnetic-field enhancement mechanism. Keywords: localized surface plasmon resonance spectroscopy, nanosensing, plasmonic materials, surface-enhanced Raman spectroscopy. www.mrs.org/publications/bulletin Plasmonic Materials for Surface-Enhanced Sensing and Spectrosopy MRS BULLETIN • VOLUME 30 • MAY 2005 369 SERS applications.16,17 If the nanosphere mask is removed, typically by sonicating the entire sample in a solvent, surface- confined nanoparticles are left behind that have a triangular footprint. In a typical NSL process, the deposition of 50 nm of Ag over a single-layer mask self-assembled from nanospheres with D
400 nm pro- duces nanotriangles with an in-plane width a 100 nm, height b 50 nm, and interpar- ticle separation distance dip 230 nm. Size- and Shape-Tunable Localized Surface Plasmon Resonance Spectra NSL-derived nanoparticles exhibit in- tense UV–visible extinction (i.e., the sum of absorption and scattering) bands that are   not present in the spectrum of the bulk metal. Figure 1b shows that the LSPR spectra can easily be tuned all the way from the near-UV through the visible spec- trum18 and even into the mid-IR19 by chang- ing the size or shape (triangle or hemisphere) of the nanoparticles. Note that the LSPR bandwidth does not change significantly as the wavelength at peak maximum, max, is tuned. Additionally, several other sur- prising LSPR optical properties have been discovered for NSL-derived Ag nano- particles: (1) max shifts by 2–6 nm per 1 nm variation in nanoparticle width or height,18 (2) the molar decadic (tenfold) extinction coefficient is
3  1011 M1 cm1,18 (3) the LSPR oscillator strength per atom is equivalent to that of atomic sil- ver in gas or liquid phases,18 (4) resonant Rayleigh scattering20,21 occurs with an effi- ciency equivalent to that of 106 fluo- rophors,22 and (5) local electromagnetic fields are amplified by factors of |E|2 104, leading to intense signals in all surface- enhanced spectroscopies.11 Fundamentals of Localized Surface Plasmon Resonance Spectroscopy The simplest theoretical approach avail- able for modeling the optical properties of nanoparticles is classical electrodynamics (i.e., solving Maxwell’s equations with the metal dielectric constant taken from bulk measurements). For spherical particles, this leads to the following (Mie theory) expres- sion for the extinction coefficient E() in the long-wavelength limit:23 . (1) Here, NA is the areal density of the nano- particles, a is the radius of the metallic nano- sphere, m is the dielectric constant of the medium surrounding the nanosphere (as- sumed to be a positive, real number),  is the wavelength, and r and i are the real and imaginary parts of the metal dielectric func- tion. This formula predicts a resonant peak when r  2 m, which for silver and gold occurs in the visible portion of the spectrum. In addition, any change in the dielectric constant of the medium (e.g., when mole- cules adsorb on the particle) leads to a change in the resonance wavelength. When one considers spheroidally shaped particles, the term r  2 m in the denomi- nator in Equation 1 is replaced by r   m , where  is a parameter that depends on the shape of the spheroid, increasing from 2 for a sphere to 17 for a spheroid with an aspect ratio of 5:1. This leads to strong de- εε εε εε εε ε   εiεr  2εm2  εi 2 Eλ
24πNAa3εm32 λln10  ε Figure 1. (a) Schematic representation of the nanosphere lithography (NSL) fabrication process.The AFM image in step 3 is 5
m  5
m. (b) Size- and shape-tunable localized surface plasmon resonance spectra of various Ag nanoparticles (labeled A–H) fabricated by NSL.The wavelength of maximum extinction, max, is changed by varying the in-plane width a and out-of-plane height b of the nanoparticles. 372 MRS BULLETIN • VOLUME 30 • MAY 2005 Plasmonic Materials for Surface-Enhanced Sensing and Spectrosopy provides a straightforward platform to extend the LSPR nanosensor technology to immunoassay technology. The max for biotin-functionalized Ag NSL nanopar- ticle arrays was measured as a function of the concentration of streptavidin (SA) over the concentration range 1015 M [SA] 106 M. Fitting the data to the theoretical normalized response expected for 1:1 binding of a ligand to a multivalent re- ceptor with different sites but invariant affinities yielded the following values: saturation response Rmax
26.5 nm, surface-confined thermodynamic binding constant Ka,surf
1011 M1, and limit of detection LOD 1 pM SA.34 Similarly, the LSPR response curve for the binding of anti-biotin (AB) to biotin-functionalized Ag NSL nanoparticles was measured, and the data analysis yielded Rmax
38.0 nm, Ka,surf
4.5  107 M1, and LOD 7  1010 M AB.35 As predicted, the LOD of the nanobiosensor studied is lower for systems with higher binding affinities, such as for the well-studied biotin–streptavidin couple, and higher for systems with lower binding affinities, as seen in the anti-biotin system. A comparative analysis of real-time re- sponses of a commercial propagating SPR sensor (planar thin film of gold) and the LSPR sensor (Ag NSL nanoparticle array) was carried out using the binding of Con- canavalin A (ConA), a mannose-specific plant lectin, to mannose-functionalized self- assembled monolayers (SAMs).36 During the association phase in the real-time bind- ing studies, both sensors exhibited qualita- tively similar signal-versus-time curves. However, in the dissociation phase, the SPR sensor showed an approximately five times greater loss of signal than the LSPR sensor. A comprehensive set of nonspecific binding studies demonstrated that this signal difference was not the consequence of greater nonspecific binding to the LSPR sensor, but rather a systematic function of nanoparticle structure. Ag nanoparticles with larger aspect ratios showed larger dissociation phase responses than those with smaller aspect ratios. DDA calcula- tions demonstrated that this response is a consequence of the similarity in length scale between the electromagnetic-field decay length and the physical size of ConA.36 Recently, the LSPR sensor has been successfully used for the detection of an Alzheimer’s disease biomarker from both synthetic37,38 and human patient38 samples (see also the article by Thaxton et al. in this issue). In this work, the interactions between the biomarker (antigen), amyloid- derived diffusible ligands (ADDLs), and specific anti-ADDL antibodies were studied. Using the sandwich assay format, the LSPR sen- sor provided quantitative binding informa- tion for both antigen and second antibody detection that permits the determination of ADDL concentration. This unique capa- bility offers the possibility of analyzing the aggregation mechanisms of this putative Alzheimer’s disease pathogen at physio- logically relevant monomer concentrations. Monitoring the LSPR-induced shifts from both ADDLs and a second polyclonal anti- ADDL antibody as a function of ADDL con- centration reveals two ADDL epitopes (the exact binding site on an antigen that binds to an antibody) that have binding constants to the specific anti-ADDL antibodies of 7.3  1012 M1 and 9.5  108 M1. Further- more, this study demonstrated for the first time that the LSPR nanosensor was suc- cessful at analyzing human brain extract and cerebrospinal fluid (CSF) samples. Examination of these results from both Alzheimer’s disease and control patients reveals that the LSPR nanosensor provides new information relevant to the understand- ing and possible diagnosis of Alzheimer’s disease. This exciting advance is one of the first examples in which nanotechnology has been applied to clinical materials for biomolecular diagnostics.38 Figure 4. Localized surface plasmon resonance (LSPR) spectra for Ag and Au nanoparticles fabricated by nanosphere lithography. Long-range distance dependence: (a) Ag nanoparticles (in-plane width a  100 nm, out-of-plane height b  50.0 nm) for 0–20 layers of Cu2/HS–(CH2)10COOH; (b) Au nanoparticles (a  70 nm, b  50.0 nm) for 0–14 layers of Cu2/HS–(CH2)10COOH. (c) Shape dependence for Ag nanoparticles: (solid triangles) LSPR shift versus number of layers in the self-assembled monolayer (SAM) thickness for solvent-annealed Ag nanoparticles; (blue circles) LSPR shift versus number of layers in the SAM thickness for thermally annealed (600 C for 1 h) Ag nanoparticles. (inset, upper left) Atomic force microscopy (AFM) image of solvent-annealed nanoparticles (a  114 nm, b  54 nm). (inset, lower right) AFM image of thermally annealed nanoparticles (a  110 nm, b  61 nm). (d) Composition dependence: (solid circles) LSPR shift versus layer thickness for Ag nanoparticles (a  70 nm, b  50.0 nm); (red triangles) LSPR shift versus layer thickness for Au nanoparticles (a  70 nm, b  50.0 nm). Plasmonic Materials for Surface-Enhanced Sensing and Spectrosopy MRS BULLETIN • VOLUME 30 • MAY 2005 373 with SERS. Consequently, there is renewed experimental effort in SERES seeking to verify various aspects of the EM theory. We will now discuss some recent results from our laboratory obtained using two ap- proaches—plasmon-sampled (PS)42 and wavelength-scanned (WS) SERES. WS- SERES requires both a broadly tunable laser and detection system. This is not commonly available; however, with the size- and shape-tunable LSPR spectra ob- tainable with NSL- or EBL-derived plas- monic materials, this problem can be circumvented. NSL and EBL are used to prepare many samples with different LSPR max values. Each sample is thor- oughly characterized structurally by atomic force microscopy or scanning electron microscopy. Correlated, spatially resolved LSPR and SER spectra are measured with a single excitation wavelength at multiple locations on each sample using a Raman microscope. Figure 5 shows representative LSPR spectra (Figures 5a–5c) and SERS spectra (Figures 5d–5f) for benzenethiol adsorbed on NSL-fabricated Ag nanopar- ticles excited at three common, fixed, exci- tation wavelengths. The SERS enhancement factors for Figures 5d–5f are 7.6  107, 6.3  107, and 9.0  107, respectively,42 which are results that roughly agree with theoreti- cal estimates (the electrodynamic calcula- tions described earlier). PS-SERES plots (enhancement factor versus LSPR max) for the 1575 cm1 band of benzenethiol ex- cited at three different laser wavelengths each showed a well-defined maximum It is important to note that this technol- ogy is in its infancy. There are at least two main research objectives that must be met before it becomes available. First, the sensor must undergo rigorous testing using CSF samples from many more patients. In order for this objective to be met, we must inte- grate this technology with a redesigned chip. Currently, approximately 250
l of CSF is required for each assay. Because spinal taps are painful, it is ideal to minimize the amount of sample needed. By incorporat- ing microfluidics and miniaturizing the sample cell, this objective will be met. The examples just given all involved NSL-derivatized Ag nanoparticle arrays as the sensor platform. It has now been demon- strated that all of the outstanding attributes of the propagating SPR sensor, the “gold standard” in optical biosensing, are retained or exceeded in a single-nanoparticle sen- sor.32,39 Dark-field LSPR scattering spec- troscopy and microscopy were used to demonstrate zeptomole (1021) sensitivity coupled with real-time kinetic analysis.32 Streptavidin biosensing has also been demonstrated on single Ag nanoparticles. A 12.7 nm redshift in the LSPR max arises from the detection of 700 streptavidin molecules.40 Surface-Enhanced Raman Spectroscopy The local electromagnetic fields that ac- company photon excitation of the LSPR are a key factor leading to the intense signals observed in all surface-enhanced spectro- scopies.11 Surface-enhanced Raman spec- troscopy (SERS) is characterized by an ensemble-averaged intensity enhancement factor, EF, of
106, for analytes bound to noble metal surfaces that possess random roughness,41 or EF 107–108, for surfaces with intentionally nanofabricated feature sizes in the
100 nm range.42 Recent reports of single-molecule detection43,44 using SERS on Ag nanoparticle clusters have rejuve- nated interest in this widely used analyti- cal technique. A clear understanding of the mechanism responsible for the enormous enhancement factors (
1014–1015) observed in single-molecule SERS remains elusive. The electromagnetic (EM) mechanism of SERS mentioned earlier predicts that there is a well-defined relationship between the LSPR spectrum and the SERES spectrum of a SERS-active surface. In particular, very specific quantitative predictions for the magnitude of the enhancement factor on nanoparticle size, shape, and local dielectric environment are made. Wavelength- scanned SERES experiments on microfabri- cated surfaces were carried out in the early 1980s at Bell Laboratories,45,46 with a view toward verifying these predictions of the EM mechanism. Both as a consequence of experimental difficulty and the dissolution of the Bell group, many EM predictions re- main purely in the domain of theory. The plasmonic materials that are now readily available from NSL and electron-beam li- thography (EBL) provide a new platform for the detailed study of the electromagnetic- field enhancement mechanism associated  Figure 5. (a)–(f) Correlated, spatially resolved, localized surface plasmon resonance (LSPR) and surface-enhanced Raman spectroscopy (SERS) results for benzenethiol adsorbed on Ag nanoparticle arrays fabricated by nanosphere lithography. (a), (d) Ag nanoparticles fabricated with nanosphere diameter D
280 nm and deposited mass thickness dm
36 nm, probed with an excitation wavelength ex
514.5 nm, power
0.7 mW. A representative atomic force micrograph of the substrate is shown in the inset. (b), (e) Ag nanoparticles fabricated with D
280 nm, dm
36 nm, probed with ex
532.0 nm, power
0.7 mW. (c), (f) Ag nanoparticles fabricated with D
400 nm, dm
56 nm, probed with ex
632.8 nm, power
1.2 mW. (g)–(i) Plasmon-sampled surface-enhanced Raman excitation spectroscopy (PS-SERES) results for the 1575 cm1 band of benzenethiol with three different excitation wavelengths: (g) ex
514.5 nm, (h) ex
532.0 nm, and (i) ex
632.8 nm. For each ex, both the wavelength location of the excitation (solid line) and the scattering (dashed line) are marked.The overlaid curves represent the bin-averaged values of the LSPR max and the enhancement factor. Bin widths are (g) 24 nm, (h) 16 nm, and (i) 16 nm. 374 MRS BULLETIN • VOLUME 30 • MAY 2005 Plasmonic Materials for Surface-Enhanced Sensing and Spectrosopy representing the highest intensity SERS signal (Figures 5g–5i).42 In each case, these PS SERES spectra follow the behavior pre- dicted by the EM theory, that is, the largest enhancement factor occurs when the en- ergy corresponding to the LSPR maximum is located near the midpoint between the energy of laser excitation and the energy of the Raman photons. These results un- ambiguously demonstrate (1) a systematic approach to the optimization of SERS spec- tra on nanoparticle substrates and (2) that large, ensemble-averaged, SERS enhance- ment factors (
1  108) are readily obtain- able from Ag NSL-derived nanoparticle array surfaces. Recently, these results have been corroborated using Au nanoparticle array surfaces fabricated by EBL.47 Figure 6a shows the SERS spectrum of benzenethiol adsorbed on a Ag FON sur- face. The nanostructure of Ag FON surfaces can be quantitatively characterized by atomic force microscopy (Figures 6b and 6c). Ag FON surfaces have a nanostruc- ture size distribution that is less broad than that of a randomly roughened sur- face but not as narrow as a NSL-derived surface. Consequently, Ag FON surfaces have well-defined LSPR spectra (Figures 6d and 6e, solid curves), albeit broader than the LSPR spectra for Ag NSL-derived nanoparticle arrays. WS SERES spectra for the 1081 cm1 band of benzenethiol are shown in Figures 6d and 6e (data points). Using neat liquid benzenethiol as the nor- mal Raman standard, the peak enhance- ment factor values were calculated to be 2.7  106 and 1.9  106 for Figure 6d and 6e, respectively. Two different behaviors are observed in the WS-SERES spectra for these surfaces. In one case, the WS-SERES spectrum peaks to the red of the LSPR spectrum (Figure 6d, data points) and in the other, it tracks the LSPR spectrum (Figure 6e, data points). This outcome is attributed to the different nanostructure size distributions for each surface. From a practical perspective, we point out that it is not yet possible to determine the optimum laser excitation wavelength for an Ag FON without doing SERES. Figure 6f shows the SERS spectrum of benzenethiol adsorbed on a Ag NSL- derived nanoparticle array surface. A typi- cal atomic force micrograph of a Ag NSL surface is shown in the inset in Figure 6g. The LSPR spectrum of a Ag NSL array is shown in Figure 6g (solid red line), along with the corresponding WS-SERES spec- trum for the 1081 cm1 band of benzenethiol adsorbed on the same Ag NSL array (data points). The peak enhancement factor for the 1081 cm1 band of benzenethiol ad- sorbed on this surface is 1.9  107, again using neat liquid benzenethiol as the nor- mal Raman standard. Note that the peak Figure 6. Localized surface plasmon resonance (LSPR), surface-enhanced Raman spectroscopy (SERS), and wavelength-scanned surface-enhanced Raman excitation spectroscopy (WS-SERES) results for benzenethiol adsorbed on Ag film-over-nanosphere (Ag FON) surfaces and nanoparticle arrays fabricated by nanosphere lithography (NSL). (a) SERS spectrum measured from Ag FON surface with excitation wavelength ex
532 nm, power
3.0 mW, and 100 s data acquisition time. (b) Contact-mode atomic force microscopy (AFM) image of Ag FON surface (nanosphere diameter D
410 nm, deposited mass thickness dm
200 nm) used for SERS in (a). (c) Contact-mode AFM image of Ag FON surface (D
500 nm, dm
250 nm) used for WS-SERES in (e). (d) LSPR spectrum (solid line, max
562 nm, FWHM
144 nm) and WS-SERES spectra (data points) for the 1081 cm1 band of benzenethiol measured from the Ag FON surface in (b). (e) LSPR spectrum (solid line, max
638 nm, FWHM
131 nm) and WS-SERES spectra (data points) for the 1081 cm1 band of benzenethiol measured from Ag FON surface in (c). (f) SERS spectrum measured from Ag nanoparticle array surface (ex
532 nm, power
3.0 mW, 100 s data acquisition time). (g) LSPR spectrum (solid line, max
688 nm, FWHM
95 nm) and WS-SERES spectra (data points) for the 1081 cm1 band of benzenethiol measured from a Ag nanoparticle array surface. (inset) Tapping-mode AFM image of a representative array surface.