Designing Efficient Localized Surface Plasmon Resonance-Based

Sep 25, 2012 - Label-Free Nanoplasmonic-Based Short Noncoding RNA Sensing at ... The Journal of Physical Chemistry C 2013 117 (23), 12325-12336...
0 downloads 0 Views 4MB Size
Article pubs.acs.org/JPCC

Designing Efficient Localized Surface Plasmon Resonance-Based Sensing Platforms: Optimization of Sensor Response by Controlling the Edge Length of Gold Nanoprisms Gayatri K. Joshi,† Phillip J. McClory,†,§ Barry B. Muhoberac,† Amar Kumbhar,‡ Kimberly A. Smith,† and Rajesh Sardar*,† †

Department of Chemistry and Chemical Biology, Indiana University Purdue University Indianapolis, 402 North Blackford Street, Indianapolis, Indiana 46202, United States ‡ Chapel Hill Analytical and Nanofabrication Laboratory, University of North Carolina, Chapel Hill, North Carolina 27599, United States S Supporting Information *

ABSTRACT: Over the past few years, the unique localized surface plasmon resonance properties of plasmonic nanostructures have been used to design label-free biosensors. In this article, we demonstrate that it is the difference in edge length of gold nanoprisms that significantly influences their bulk refractive index sensitivity and local sensing efficiency. Nanoprisms with edge lengths in the range of 28−51 nm were synthesized by the chemical-reduction method and sensing platforms were fabricated by chemisorption of these nanoprisms onto silanized glass substrates. The plasmonic nanosensors fabricated from 28 nm edge length nanoprisms exhibited the largest sensitivity to change in bulk refractive index with a value of 647 nm/RIU. The refractive index sensitivity decreased with increasing edge length, with nanoprisms of 51 nm edge length, displaying a sensitivity of 384 nm/RIU. In contrast, we found that the biosensing efficiency of sensing platforms modified with biotin increased with increasing edge length, with sensing platforms fabricated from 51 nm edge length nanoprisms displaying the highest local sensing efficiency. The lowest concentration of streptavidin that could be measured reliably with this edge length nanoprism was 1.0 pM and the limit of detection was 0.5 pM, which is much lower than found with gold bipyramids, nanostars, and nanorods. In addition, the electromagnetic-field decay length of the sensing platforms was substantially influenced by the edge-length of the nanoprisms.



INTRODUCTION

Common methods of LSPR-based plasmonic nanosensor fabrication involve lithographic techniques and immobilization of metallic nanostructures on a chemically modified support substrate. Such supporting substrate-bound nanostructures provide a major advantage in LSPR-based chemical or biological sensing, which is ease in performing nanostructure surface modification with a variety of ligands or receptors. It has also been shown that the LSPR properties of nanostructures with sharp tips, e.g., rods,3,4,23 bipyramids,6 stars,2,31,32 and prisms,27,33−35 are very sensitive to the change in their local dielectric environment and display high refractive index sensitivity. Considering nanostructures of different shape, the lithographically fabricated silver nanoprisms have been the focus of

Noble metal nanostructures of different sizes and shapes have been the focus of plasmonic-based research, mainly in the design of efficient chemical and biological sensors.1−15 These nanostructures display localized surface plasmon resonance (LSPR) properties, which originate from the collective oscillation of free electron in the vicinity of the metal-dielectric layer interface upon irradiation by light. It has been shown that the LSPR properties of nanostructures are very sensitive to their size,10,16−19 shape,6,19−26 and surrounding dielectric environment.27−30 The effects of the dielectric environment on the LSPR properties, such as absorption and scattering of light by nanostructures, have been studied with great interest. More importantly, the refractive index near surface of the nanostructure is readily influenced by analyte binding, and this process can be monitored in real-time by LSPR peak shifts with optical spectroscopy. © 2012 American Chemical Society

Received: March 20, 2012 Revised: September 5, 2012 Published: September 25, 2012 20990

dx.doi.org/10.1021/jp302674h | J. Phys. Chem. C 2012, 116, 20990−21000

The Journal of Physical Chemistry C

Article

substantial study as LSPR-based sensors.9,12,13,15,29,36−41 However, their tendency toward aerobic oxidation reduces their sensing efficiency and requires all surface modifications and measurements be performed under inert atmosphere. In addition, conventional lithographic approaches involve deposition of the desired noble metals on Cr- or Ti-coated supporting substrates to promote noble metal adhesion. However, the Cr or Ti layer reduces the sensing efficiency of the nanostructure.11,12 Another commonly used LSPR-based sensing platform fabrication method includes immobilization of nanostructures on amine- or thiol-terminated silane-modified supporting substrates.3,23,33,42−44 Importantly, most studies on developing LSPR-based sensors have been performed utilizing chemically synthesized gold nanorods, which possess significantly lower bulk and surface sensitivities compared to other nanostructures such as nanostars or nanoprisms.3,4,6,10,16,23,31,33,35 Although gold nanostars provide comparatively high refractive index sensitivity when compared with either nanorods and nanoprisms, they also display pronounced LSPR peaks in the near to far-infrared region, which could potentially hinder sensitivity measurements due to background absorption and scattering from water.3 In a recent study we have shown that the supporting substrate-bound nanoprisms with average edge length of 22 nm were very sensitive to changes in bulk refractive index and displayed a 583 nm/RIU refractive index sensitivity (RIU = refractive index unit).45 In addition, the utility of substrate-bound nanoprisms was further demonstrated for LSPR-based biosensing by detecting as low as 50 pM of streptavidin (SA) in a model receptor−analyte (biotin-SA) binding assay. Zamborini and co-workers have shown that nanoprisms with average edge length of 100−200 nm could be used to detect 7 pM of human IgG.34 There are many reasons to choose nanoprisms for plasmonic nanosensor fabrication. Indeed, they provide distinctive optical and physicochemical properties in comparison to other nanostructures as follows: (1) localized electromagnetic field (EM-field) enhancement at their sharp tips, (2) large “sensing volume”, which can be defined as the penetration depth within which changes of the refractive can be detected, (3) accurate quantification of analyte due to flat surfaces, and (4) ability to tune the LSPR dipole peak position (λLSPR) between 650 and 900 nm by controlling their edge length. This spectral region provides the additional advantage of avoiding background absorption and scattering from water and endogeneous chromophores, which is extremely important for biosensing applications. Previous reports were mainly focused on the EMfield enhancement at the sharp tips and their influence on biosensing.35,46 However, investigations of the effects of edge length of the nanoprism on refractive index sensitivity and on label-free biosensing in general have not yet been performed. Thus our experimental findings will allow optimization of the edge length and spectral region of nanoprisms for increased sensitivity and response in sensor design. In this report, we show that the refractive index sensitivity of sensing platforms that are fabricated from nanoprisms bound to silanized glass substrates are strongly dependent on their edge length, where the shortest edge length nanoprisms (28 nm) display the highest sensitivity. More importantly, a reverse trend is observed for LSPR-based biosensing with the 51 nm edge length nanoprisms allowing us to reproducibly detect 1.0 pM of SA, which is much lower than the sensitivity reported for spherical nanoparticles, nanorods, bipyramids, and nanostars.2−4,6,10,16,23,47,48 Finally, the long-range distance depend-

ence of the LSPR properties reveals that the longest edge length nanoprisms display the greatest peak shifts. To the best of our knowledge, this is the first example where LSPR-based refractive index sensitivity and label-free biosensing ability of different edge length nanoprisms have been reported.



EXPERIMENTAL SECTION Materials. Chloro(triethyphosphine)gold(I) (Et3PAuCl), poly(methylhydrosiloxane) (PMHS, Mn = 1700−3300), trioctylamine (TOA), triethylamine (TEA), N,N-diisopropylethylamine (DIEA), Streptavidin from Streptomyces avidnii (SA), Bovine Serum Albumin (BSA), N-biotinyl-3,6,9-trioxaundecane-1 (Biotin), 11-mercaptoundecyltetraethylene glycol) (MUTEG), 16-mercaptohexadecanoic acid (MHA), (3mercaptopropyl)triethoxysilane (MPTES), 3-mercaptopropionic acid (MPA), poly(allylamine hydrochloride) (PAH), sodium salt of poly(acrylic acid) (PAA), 1-ethyl-3-(3(dimethylamino)propyl)carbodiimide (EDC), N-hydroxysuccinimide (NHS), anhydrous acetonitrile (CH3CN), ethanol, and carbon tetrachloride (CCl4) were purchased from Sigma Aldrich and were used as received. Hydrochloric acid and 28−30% ammonium hydroxide were obtained from Acros Organics and used without additional purification. Sodium phosphate salts and RBS 35 Detergent were obtained from Thermo Fisher Scientific and used as received. All water was purified using a Thermo Scientific Barnstead Nanopure system. Glass coverslips were obtained from Fisher Scientific. Microscopy and Spectroscopy Measurements. Scanning electron microscopy (SEM) micrographs were accrued using a Hitachi S-4700 FESEM at 20 kV. Absorption and extinction spectra in the range of 300−1000 nm were collected with a Varian Cary 50 Scan UV−visible spectrophotometer using 1 cm quartz cuvettes. All solution spectra were collected using 0.3 mL of nanoprism reaction mixture diluted to 2.0 mL in acetonitrile. All extinction spectra of substrate-bound gold nanoprisms were collected at room temperature and in air by positioning the silanized glass sensing platforms in the cuvette holder using a cuvette for support. The average edge lengths of the nanoprisms were calculated from the SEM images using Image J software. All XPS data were collected on a Kratos Axis Ultra DLD system with a monochromatic Al source with an energy of 1486.6 eV and an X-ray power of 150 W. Band-pass energies of 80 and 20 eV were used for survey scans and high resolution scans, respectively. All data were corrected so that the C 1s line was shifted to 285 eV. For XPS analysis, the nanoprism sample was drop-cast on a clean silicon wafer and dried under high vacuum. Synthesis of Gold Nanoprisms. Gold nanoprisms were synthesized according to our previously published procedure with minor modifications.45 PMHS was used as a reducing agent and TOA as a stabilizing ligand. In a typical synthesis, 0.02 mmol of Et3PAuCl was dissolved in 20 mL of acetonitrile and the solution was stirred for 30 min at room temperature. Then 0.06 mL of TOA was injected followed by another 30 min of stirring. Next, 0.3 mL of PMHS was added and the reaction mixture was stirred at room temperature for 30 min followed by heating to 40 °C. After 220 min at 40 °C, the solution became dark blue and displayed a stable absorption maximum at 700 nm. The solution was then removed from heat, centrifuged, and used for plasmonic nanosensor fabrication. The SEM analysis provided the average edge length of the nanoprisms as 28 ± 2.8 nm. Nanoprisms with larger edge lengths, i.e., 35 ± 3.6, 42 ± 4.5, and 51 ± 5.6 nm, 20991

dx.doi.org/10.1021/jp302674h | J. Phys. Chem. C 2012, 116, 20990−21000

The Journal of Physical Chemistry C

Article

MUTEG helps to form a better SAM, where the glycol units reduce the nonspecific adsorption of proteins.54 Finally, the 1:1 molar thiol ratio also helps to avoid steric hindrance by creating a dispersed surface density of MHA and hence biotin.3,23 Next, the sensing platform was rinsed with a copious amount of ethanol and dried with nitrogen, followed by submersion in a 1:1 (v/v) solution of aqueous EDC and NHS (0.2 M each) for 1 h. The sensing platform was removed, rinsed with water, and dried with nitrogen. It was then incubated in a 1 mM Nbiotinyl-3,6,9-trioxaundecane-1 in ethanol for 12 h. The sensing platform was then rinsed with ethanol and dried with nitrogen. For biosensing studies, different concentrations of SA, i.e., 1 μM, 100 nM, 10 nM, 1 nM, 100 pM, 10 pM, and 1 pM, were prepared in PBS buffer (pH 7.2) by serial dilution. The sensing platform was incubated separately in 1 μM or 100 nM SA solution for two hours or in 1 nM, 100 pM, 10 pM, or 1 pM SA solution for 6 h, which were the times necessary to reach equilibrium.45 The substrates were rinsed with water to remove unattached SA and the extinction spectra were recorded. A schematic representation of the sensing platform construction for label-free biosensing is given in our previous work.45 The limits of detection (LOD) for biosensors fabricated with the three different edge length gold nanoprisms were calculated by measuring the λLSPR of blank samples (without SA bound) six times and subtracting three times the standard deviation from the average of the relative responses of the SA-attached sensing platform. By determining a linear relationship between the shift in the λLSPR and the SA concentration, the minimum amount of SA detectable was calculated from the minimum detectable λLSPR shift.

were prepared under similar reaction conditions with the identical mole ratio of gold salt to PMHS by using 0.019 mL of TEA, 0.032 mL of DIEA, or 0.024 mL of DIEA, as a stabilizing ligand instead of TOA, respectively. We found that the above procedure controls the edge length of nanoprisms but the detailed mechanism is not understood and is currently under our investigation. The XPS analysis (data not shown) documented complete reduction of gold salt to neutral gold nanostructures. Therefore, from the initial gold salt concentration, we calculated the solution gold concentration as 1.14 mM for all four different edge length nanoprisms. Functionalization of Glass Coverslips with MPTES and Construction of Sensing Platforms. The glass coverslips (substrate) were functionalized with MPTES using literature procedures.49,50 Briefly, coverslips were immersed in a 20% (v/ v) aqueous RBS 35 detergent solution at 90 °C for 30 min and then sonicated for 5 min. The coverslips were rinsed with a copious amount of water and immersed in a solution of hydrochloric acid and methanol (1:1 v/v) for 30 min. After rinsing with nanopure water multiple times, coverslips were dried in a vacuum oven at 60 °C overnight. The coverslips were then placed in a solution of 10% MPTES in ethanol for 30 min, followed by 5 min of sonication and rinsing with anhydrous ethanol. The ethanol rinse and sonication steps were repeated at least 5 times. The coverslips were then placed in a vacuum oven for 3 h at 120 °C. The MPTES-functionalized coverslips were then immersed in freshly prepared nanoprism solution for 30 min. After removal, they were rinsed with ethanol and dried under nitrogen. The nanoprisms bound to coverslip formed the sensing platforms and were stored under nitrogen at 4 °C. Refractive Index Sensitivity Studies. The refractive index sensitivity of the substrate-bound nanoprisms was determined by incubating the sensing platform in various solvents with different refractive indices: CCl4 (n = 1.458), C2H5OH (n = 1.362), CH3CN (n = 1.343), and H2O (n = 1.333). Extinction spectra (absorption + scattering) were collected in each solvent after 10 min of incubation. Layer-by-Layer (LBL) Deposition of Polyelectrolyte onto Sensing Platforms. The LBL deposition of polyelectrolyte onto sensing platforms was done according to a routinely used, previously published procedure.3,23 The nanoprisms bound to supporting substrates were functionalized by incubating overnight in 0.1 mM MPA in ethanol. After washing with a copious amount of ethanol and drying under nitrogen, the coverslips were immersed in 0.1 mM aqueous solution of NaOH for 15 min. The coverslips were then rinsed with water and incubated in PAH solution (0.5 mg/mL) in 1.0 M of NaCl for 30 min. The coverslips were rinsed with 1.0 M NaCl and then immersed in PAA (0.5 mg/mL) in 1 M NaCl for 30 min. This LBL deposition procedure was repeated until no apparent shift in λLSPR was observed. The extinction spectrum was collected between each polyelectrolyte deposition step. Each polyelectrolyte layer absorbed was approximately 2.0 nm thick.51−53 The attachment of PAH and PAA were considered as two separate layers. Bioconjugation of Sensing Platforms for Label-Free Biosensing. The supporting substrate-bound nanoprisms were incubated in a 1:1 (v/v) solution of ethanolic MUTEG and MHA (1 mM each) for 4 h. The function of MHA is not only to attach biotin to the nanoprism but also to increase the distance between the nanoprism and biotin, which helps to avoid nonspecific adsorption of proteins onto the sensing platform. It has also been shown that the alkyl-chain part of



RESULTS AND DISCUSSIONS LSPR-Based Sensitivity Studies of Sensing Platforms Fabricated from Different Edge Length Nanoprisms. Recently, we have shown that the LSPR properties of nanoprisms with 22 nm average edge length attached onto MPTES-functionalized glass substrate were very sensitive to changes in the bulk refractive index of their surroundings.45 More importantly, the dipole peak was more sensitive compared to the quadrupole peak with corresponding sensitivities of 583 and 287 nm/RIU, respectively. Therefore, all sensitivity data reported in this manuscript will be based on the LSPR dipole peak shift. To further examine the effects of the edge length of the nanoprism on refractive index sensitivity, we investigated nanoprisms with four different edge lengths larger than 22 nm. The dipole peaks at 700, 750, 800, and 850 nm in the UV−visible absorption spectra in acetonitrile (Figure 1) were produced by nanoprisms with average edge lengths of 28 ± 2.8, 35 ± 3.6, 42 ± 4.5, and 51 ± 5.6 nm, respectively. Previously published methods had only produced gold nanoprisms with edge lengths between 100 and 300 nm, which corresponded to dipole peaks in the range of 1100−1400 nm.55 The nanoprisms we synthesized displayed quadrupole peaks at 550, 560, 575, and 580 nm, respectively, which were significantly blue-shifted in comparison to 100−300 nm gold nanoprisms in which quadrupole peaks were observed in the range of 800−1000 nm. The SEM analysis (Figure 2A) of the nanoprisms with 28 nm of edge length immobilized onto MPTES-functionalized glass substrates showed 81% of the nanostructures were nanoprisms and the remaining were nanospheres or truncated tetrahedrons. We examined the refractive index sensitivity of the sensing platform by measuring the λLSPR in various solvents and found it to be 529 nm/RIU 20992

dx.doi.org/10.1021/jp302674h | J. Phys. Chem. C 2012, 116, 20990−21000

The Journal of Physical Chemistry C

Article

Figure 1. UV−visible absorption spectra of as synthesized gold nanoprisms of different edge length in acetonitrile. The LSPR dipole peak positions that correspond to nanoprisms with average edge lengths of 28, 35, 42, and 51 nm are 700 (blue), 750 (red), 800 (black), and 850 (purple) nm, and the quadrupole peak positions are 550, 560, 575, and 580 nm, respectively. Solutions containing different edge length nanoprisms were used for LSPR-based plasmonic nanosensors fabrication.

(Figure 3A). This sensitivity value is lower than our previously reported value (583 nm/RIU), in which sensing platforms were fabricated using 22 nm edge length nanoprisms.45 We believe that the lower sensitivity for larger edge length nanoprisms involves tip sharpness.35 Later, we will discuss in more detail the refractive index sensitivity studies of different edge length nanoprisms. Recent studies by Beeram et al.33 have shown that the presence of some fraction of nonprismatic nanostructures in a plasmonic sensor reduced the LSPR refractive index sensitivity, which potentially could be enhanced by removing them from the supporting substrate. The authors found that the sensitivity increased from 89 to 302 nm/RIU after removal of the nonprismatic nanostructures. It is known that the refractive index sensitivities of both spherical and tetrahedron-shaped nanostructures are much lower than that of nanoprisms.19 Therefore, removal of these nanostructures should increase the refractive index sensitivity of sensing platforms due to an increase in the relative content of the nanoprisms on the supporting substrates. To investigate this effect, we first prepared sensing platforms using 28 nm edge length nanoprisms. To remove the nonprismatic nanostructures, adhesive tape (Scotch) was applied to the surface of the sensing platform, pressed gently with a finger, and slowly removed at an

Figure 3. (A) Relationship between LSPR dipole peak position of 28 nm edge length nanoprisms bound to silanized glass and the refractive index of the bulk solutions before (red triangles) and after (blue diamonds) tape cleaning. The corresponding sensitivities are 529 and 647 nm/RIU, respectively. (B) Ensemble extinction spectra of supporting substrate-bound nanoprisms before (red) and after (blue) tape cleaning in air. (C) Extinction spectra of the tape-cleaned sensing platform collected in four different solvents: H2O (black), CH3CN (green), C2H5OH (blue), and CCl4 (purple).

Figure 2. SEM images of nanoprisms bound on silanized glass substrate before (A) and after (B) tape cleaning. Scale bars are 100 nm. The blue arrow points to a sphere, the red arrow points to a truncated structure, and the green arrow points to a prism. 20993

dx.doi.org/10.1021/jp302674h | J. Phys. Chem. C 2012, 116, 20990−21000

The Journal of Physical Chemistry C

Article

Importantly, the refractive index sensitivity of sensing platforms that were cleaned with tape was higher than the uncleaned ones, which is in good agreement with the literature report that the presence of additional nonprismatic nanostructures on the substrate reduces LSPR sensitivity.33 Because of these findings, all remaining sensitivity studies for nanoprisms of different edge lengths were performed after sensing platforms were cleaned with adhesive tape. However, it is also clear that the peak width is important in the actual measurements of an induced peak shift because the broader the width, the more difficult the determination of the exact peak position. Thus, the ability to detect a very small shift depends upon the width of the peak. Additionally, it has been shown that larger nanostructures display higher sensitivity, however, due to multipolar excitation and radiative damping, the width of their spectral peak is significantly broader compared to that found with smaller nanostructures. Therefore, figure of merit (FOM) was introduced to quantify the sensing efficiency of the nanostructures.10,56 We calculated the FOM by dividing refractive index sensitivity by the full-width at half maxima (fwhm) of the corresponding peak using the following equation:4,10,35,59

approximately 90° angle. This cleaning procedure was adopted from published literature.27,33,34 The sensing platform was then rinsed with a copious amount of dichloromethane (CH2Cl2) to remove any remaining adhesive material sticking onto the surface of the nanoprisms. Both UV−visible spectroscopy and XPS analyses were performed to determine whether there was any residual tape material on the surface of the nanoprisms. SIFigures 1 and 2 (Supporting Information) represent the extinction spectra and XPS analysis of supporting substratebound gold nanoprisms without tape cleaning, after tape cleaning and before CH2Cl2 wash, and after CH2Cl2 wash. The UV−visible spectra showed no appreciable change of the λLSPR. The XPS analysis demonstrated no residual material sticking on the surface of the nanoprisms even before CH2Cl2 wash. After drying with nitrogen flow, the size and dispersity of the nanoprisms were analyzed by SEM. Figure 2 illustrates the sensing platforms before and after tape cleaning, respectively. Before tape cleaning, ∼81% of the nanostructures attached to the supporting substrate were nanoprisms. After cleaning, the population of nanoprisms significantly increased such that ∼98% of the nanostructures were nanoprisms and the remaining were truncated tetrahedrons. Although we have found that the tape cleaning process was very robust and reproducible, multiple tape cleanings eventually removed the nanoprisms from the supporting substrate as well. We can suggest two possible mechanisms for the preferential removal of the nonprismatic nanostructures observed with tape cleaning. First, we have determined that the height of the nonprismatic nanotructures (spheres and truncated tetrahedrons) was generally higher than nanoprisms. Therefore, with tape cleaning the adhesive part of the tape could make better contact with nonprismatic structures, which would be preferentially removed by removal of the tape. Second, the curvature of a nanosphere of linear dimensions similar to those of a flat nanoprism would likely have less contact area and uniformity of contact with the glass surface, enhancing the ability of tape to remove it. Even after tape cleaning the average edge length of the nanoprisms remained at 28 nm, which indicated that this procedure and multiple washings with CH2Cl2 did not affect the edge length. Importantly, the LSPR peak intensity at 533 nm was reduced significantly after tape cleaning (Figure 3B). Previously, Mirkin and co-workers reported that the intensity of the LSPR quadrupole peak of metallic nanoprisms was much lower in comparison to the dipole peak, and more importantly that peak was assigned to only the quadrupole resonance.25,56−58 However, our work suggests that some contribution to the LSPR peak at 533 nm belonged to either spherical or truncated nanostructures. After cleaning, the low intensity peak at 533 nm previously assigned to the LSPR quadrupole peak of the nanoprisms nearly disappeared, indicating successful removal of nonprismatic nanostructures from the supporting substrate. The LSPR refractive index sensitivity of the tape-cleaned sensing platform constructed with 28 nm edge length nanoprisms was investigated by measuring the shift in the λLSPR in various solvents (see Experimental Section). Figure 3C shows the extinction spectra of the sensing platform and as expected, the λLSPR red-shifted with increasing refractive index of the bulk medium. As illustrated in Figure 3A, the λLSPR responded linearly to changes in the refractive index of the surrounding medium and the refractive index sensitivity was 647 nm/RIU for the 28 nm edge length nanoprisms.

FOM =

sensitivity (nm RIU−1) fwhm (nm)

(1)

The LSPR peak width is strongly related to the size, shape, and the dispersity of the metallic nanostructures and we found that the larger the nanostructures, the wider the peak (see Table 1). Table 1. Comparison of LSPR Dipole Peak Positions in Acetonitrile, Refractive Index Sensitivities, and FOMs of Different Edge Length Gold Nanoprisms Bound to Supporting Substrates λmax (nm) 700 750 800 850

edge length (nm)a,b,c 28 35 42 51

(2.8) (3.6) (4.5) (5.6)

refractive index sensitivity (nm/RIU)

fwhm (nm)

FOM

647 538 447 384

127 135 139 152

5.1 4.0 3.2 2.5

a

The number in parentheses indicates the standard deviation. bIn each case, 1000 nanostructures were counted to determine the percentage of nanoprisms. Six hundred edge lengths were measured to determine the average edge length. cThe truncated particles were distinguishable by their lack of flat tops in the SEM images and were not counted in the edge length and standard deviation calculations.

In this present work, the sensing platform fabricated from 28 nm edge length nanoprisms had 647 nm/RIU sensitivity and a fwhm of 127 nm, which resulted in a FOM of 5.1. The FOM value we achieved from ensembled measurements for nanoprisms of 28 nm edge length is significantly higher than previously reported for metallic nanoprisms, as well as for chemically synthesized gold nanorods, nanocubes, and bipyramids, and is comparable to nanostars where FOM values were determined by single particle scattering measurements.4,10,19,24,35,59 Table 2 summarizes the dipole peak refractive index sensitivities and FOMs of the different shapes of gold nanostructures attached to the surface. The influence of edge length on refractive index sensitivity was also investigated for substrate-bound nanoprisms. Parts A− C of Figure 4 illustrate the SEM images of nanoprisms with average edge lengths of 35, 42, and 51 nm, which displayed 20994

dx.doi.org/10.1021/jp302674h | J. Phys. Chem. C 2012, 116, 20990−21000

The Journal of Physical Chemistry C

Article

nanostructures, the free electrons experience a change in Coulombic interaction with the positive lattice, which leads to a change in the LSPR absorption wavelength. There are many factors, which affect the bulk refractive index sensitivity such as having tips, edge sharpness, thickness of the nanoprisms, and their polarizability.9,10,15,16,35 Additionally, binding to the glass substrate provides an inhomogeneous dielectric environment around the nanoprism. All these components significantly affect the EM-field enhancement. The nanoprisms with shorter edge length have sharper tip curvature resulting in strong EM-field enhancement at their tips compare to nanoprisms with larger edge lengths, which have broader curvature of their tips.16,35,46 Additionally, sharp tips have more free electrons available compared to the other areas in the nanostructure. As a result, the free electrons at the sharp tips can be more easily influenced by the change in the surrounding medium.16 It is also possible that as the surface area of the nanoprisms decreases, the distribution of EM-field enhancement on them becomes more homogeneous. On the other hand, EM-field enhancement would become uneven as the surface areas of the nanoprisms increase. Clearly, the relationship between refractive index sensitivity, the shape of nanostructure, and its supporting substrate is very complex and requires further study. Nevertheless, our investigation of refractive index sensitivity of nanoprisms of different edge length found a reverse in order compared to LSPR-based sensing platforms constructed with gold bipyramids by Hafner and co-workers, where the largest

Table 2. Comparison of LSPR Dipole Peak Positions, Refractive Index Sensitivities, and FOMs for Different Shapes of Gold Nanostructures shape of Au nanostructures 45

prism sphere60 cube19 branch19 rod19 bipyramid6,19 star2,10,32

λmax (nm)

highest refractive index sensitivity (nm/RIU)

650 527− 530 538 1141 653−720 1096 650−750

583 90 83 703 195 540 665

highest FOM 4.9 1.5 1.5 0.8 2.6 4.5 5 and 5.4

measurement ensemble ensemble ensemble ensemble ensemble ensemble single

dipole peaks at 750, 800, and 850 nm respectively (Figure 1). The corresponding refractive index sensitivities of these sensing platforms were 538, 447, and 384 nm/RIU (Figure 4D) and their FOM values were 4.0, 3.2, and 2.5, respectively. Table 1 summarizes the refractive index sensitivities and FOMs of sensing-platforms containing the different edge length nanoprisms and clearly shows that shorter edge length nanoprisms displayed higher refractive index sensitivities and FOMs. The bulk refractive index sensitivity is a measure of the extent to which the solvent affects the displacement of the free electrons of noble metal nanoparticles relative to the positive atomic lattice (core electrons + nuclei) by interaction with light.16,61,62 By changing the dielectric medium surrounding the

Figure 4. SEM images of nanoprisms bound to silanized glass substrate after tape cleaning with average edge lengths of (A) 35, (B) 42, and (C) 51 nm. Scale bars are 100 nm. (D) Relationship between LSPR dipole peak position of the nanoprisms of different edge lengths and refractive index of the bulk solutions. The truncated nanostructures (see red arrow for example) were not counted in the edge length and standard deviation calculations. 20995

dx.doi.org/10.1021/jp302674h | J. Phys. Chem. C 2012, 116, 20990−21000

The Journal of Physical Chemistry C

Article

Figure 5. (A) Extinction spectra of sensing platform fabricated from 28 nm edge length nanoprisms with LBL assemblies of PEs. (B) LSPR dipole peak wavelength shifts versus number of PE layers for nanoprisms with edge lengths 28 nm (blue triangle), 42 nm (red dot), and 51 nm (black diamond). (C) Maximum peak wavelength shift versus edge length of nanoprism. The data were fitted using linear regression yielding the equation y = 12.14(x) + 3.82. (D) PE layer thickness for loss of response versus edge length of nanoprism. The data were fitted by linear regression using the equation y = 0.26(x) + 11.19. The thickness of the PE layer was estimated as described in the text.

bipyramids displayed the highest sensitivity.6 A similar refractive index sensitivity trend was also observed for gold nanorods, where the longest nanorods displayed the highest sensitivity.16 To the best of our knowledge, there is no literature report available on LSPR-based refractive index sensitivity studies of plasmonic nanosensors fabricated from different edge length gold nanoprisms. The geometry of nanoprisms is significantly different than either nanorods or bipyramids. As a consequence, the EM-field enhancement and its distribution on the nanoprisms will be different than that on either the nanorods or bipyramids and difficult to compare. However, our experimental results are also opposite to the theoretical prediction made by Lazarides and co-workers, who reported that RIU sensitivity of unattached nanostructures is independent of size and shape but increases as the dipole peak red shifts and is only dependent on the position of the dipole peak.63 Experimental studies of nanostructures with different shapes and sizes have shown that the larger the dimension of the nanostructure, the more the LSPR peak is red-shifted and the higher is the refractive index sensitivity.9−11,15 In our case we have found the opposite behavior in refractive index sensitivity. The exact reason for the reverse trend of refractive index sensitivity of our nanoprisms is not known and requires further study. Determination of the EM-Field Decay Length of Sensing Platforms. It is important to differentiate between the bulk refractive index sensitivity of metallic nanostructures and their EM-field decay length, which both depend on the geometry, surface structure, and surrounding environment of

the nanostructure.1,9−11,15,29 The EM-field decay length is a measure of the maximum allowable thickness of the adsorbate surrounding the nanostructure surface that still allows observation of a λLSPR shift due to the binding of the analytes, or alternatively, the maximum allowable thickeness of the adsorbate layer that still allows sensing of an increase in the thickness of that layer.3,9,10,15,64 It has been reported for metallic nanostructures, e.g., anisotropic silver nanoparticles (nanoprisms) and gold nanorods and nanospheres, that the EM-field decays exponentially from the surface of the nanostructure and the overall sensor responses can be described by the following equation:36,37,65 ΔλLSPR = ΔnSe−2d / L(1 − e−2d / L)

(2)

where ΔλLSPR is the LSPR dipole peak wavelength shift, Δn is the difference in the refractive index between adsorbate and surrounding medium, S is the refractive index sensitivity, d is the thickness of the adsorbate layer, and L is the EM-field decay length. Further approximation of the equation predicts that the magnitude of the ΔλLSPR decreases exponentially as the thickness of the adsorbate layer increases. Recently, this concept was further verified for gold nanoprisms in LSPR based sensing studies, where the surface of the nanoprisms was functionalized with different chain length alkylthiols containing a terminal carboxylic acid group, which was further modified for biosensing.27 The authors mentioned that the ΔλLSPR due to analyte interaction with the surface of the nanoprism was higher when its surface was modified with shorter carbon chain 20996

dx.doi.org/10.1021/jp302674h | J. Phys. Chem. C 2012, 116, 20990−21000

The Journal of Physical Chemistry C

Article

surface of the nanoprisms is modified for an attachment to the silanized glass substrate but the top surface is not. On the other hand, with the LBL studies, the top surface of the nanoprisms is functionalized with MPA, which has a similar attachment through sulfur and chain length as MPTES. The similar versus asymmetric nanoprism boundary conditions could provide different responses to increasing dielectric constant. Another explanation would be the percentage of bottom surface of the nanoprism that could be exposed to solvents by penetration underneath, which is a function of the area of the nanoprisms. To further confirm our hypothesis, we performed a controlled experiment in which the RIU sensitivities of different edgelength nanoparisms that were dispersed in solution were investigated. We have found that the 28, 35, and 42 nm edge length nanoprisms displayed sensitivities of 370, 397, and 578 nm/RIU, respectively (SI-Figure 3). This is the similar trend to that reported in the literature.10,16 Therefore, the attachment to the surface has a substantial effects on the RIU sensitivity of metallic nanostructures. Nevertheless, we believe further investigation is required to differentiate between these and other possible explanation. Receptor-Analyte Binding Study of Sensing Platforms Fabricated from Different Edge Length Nanoprisms. Metallic nanostructures have shown promise as sensing platforms for label-free biosensing.2,3,6,9,10,15,23,34 Our recent work has demonstrated that nanoprisms bound to a silanized glass surface can be used as label-free biosensors. We used the biotin-SA interaction as a model receptor-analyte system, and reproducibly measured 50 pM of SA by monitoring LSPR spectral shifts.45 The interaction of SA with the sensing platform was highly specific and SA selectively attached to biotin molecules. Van Duyne and co-workers suggested that the sensitivity of plasmonic nanostructures could be enhanced by increasing their sensing volume.35 Furthermore, in a typical receptor-analyte binding event, the LSPR-based sensing response depends on changes in local refractive index at the nanostructure−solution interface. Therefore, the sensing volume will be an important parameter in determining the sensitivity of a sensor. Nath et al.8 showed that the LSPR-based biosensing ability of spherical gold nanoparticles was strongly influenced by their size. It was found that the largest spherical nanoparticles were able to detect the lowest concentration of SA (∼1.0 nM). The nanoprism will have higher sensing volume in comparison to spherical nanoparticles because of the presence of their sharp tips and flat surface, which together will produce a strong EM-field enhancement.34,35,46 However, to the best of our knowledge, LSPR biosensing studies that have focused on the sensing volume of anisotropic metallic nanostructure have not yet been done and require in-depth investigation. Our initial investigation was focused on nanoprisms with 28 nm average edge length. The sensing platforms were functionalized by a mixture of MHA and MPTEG, as previously described.20,23,45 After thiol functionalization, the λLSPR red-shifted ∼15 nm. The amide coupling between the acid group of MHA and the amine group of biotin was performed using EDC and NHS, which resulted in an additional ∼6 nm red shift. The biotin-functionalized sensing platform was then incubated in 1 μM SA in PBS buffer (pH 7.2) resulting an additional ∼28 nm shift. Figure 6 summarizes the ΔλLSPR after each functionalization step and then after SA binding with associated error intervals. Clearly, this binding was easily distinguishable from the surface modification and biotin

thiols compared to the longer ones. We have shown that the bulk refractive index sensitivity of nanoprisms is strongly dependent on their edge length. However, to the best of our knowledge, no reports are available where the EM-field decay length of chemically synthesized gold nanoprisms of varying edge length have been investigated. Such studies are important to optimize the sensitivity of nanoprisms for their potential applications in chemical and biological sensing.3,23,27,37,66 The EM-field decay lengths of three different edge length nanoprisms (28, 42, and 51 nm) were determined by measuring the ΔλLSPR associated with a particlular thickness of the multilayer polyelectrolyte film prepared as described in the literature via the commonly used LBL deposition technique in which oppositely charged polymers were adsorbed onto the surface of the MPA functionalized nanoprisms.3,4,53,67,68 This entropically driven process allows deposition of multiple layers to form a LBL assembly. Following the literature, we estimated the total thickness of the PE layer by assuming uniform deposition of maximal layer thickness and numerically adding the layer components. We take the thickness of the MPA layer as 0.85 nm (Chem3D, Cambridge software) and each polyelectrolyte layer as uniform in nature providing ∼2.0 nm thickness to the PE film.3,51,69 The extinction spectrum was collected after each polyelectrolyte layer deposition onto MPA-functionalized 28 nm edge length nanoprisms (Figure 5A). Figure 5B illustrates the wavelength shift of three different edge length nanoprisms versus the number of polyelectrolyte layers. Red shifts of the λLSPR were observed as the thickness of the polyelectrolyte layers increased, which is due to the increase of the local dielectric environment around the nanostructure. Nanoprisms with 28, 42, and 51 nm edge-lengths detected changes associated with the refractive index of their surrounding medium to distance of 21, 25, and 27 nm, respectively. The trend of wavelength shifts with respect to thickness of PE layers is clearly in good agreement with eq 2. Figure 5C shows the ΔλLSPR with respect to the edge length of nanoprisms. The λLSPR red-shifted ∼12 nm for a 10 nm increased in edge length. These data indicate that the EM-field decay length is larger for longer edge length nanoprisms than for shorter edge length nanoprisms, and this trend is in good agreement with the literature reports, where larger spherical nanoparticles,48 NSLfabricated silver nanoprisms,70 and longer gold nanorods3,23 displayed larger EM-field decay lengths than their smaller counterparts. Figure 5D illustrates the relationship between the edge length of the nanoprisms and the limiting LSPR peak shifts. As the edge length of the nanoprisms increased, the limiting peak shifts increased by ∼2.6 nm for every 10 nm increase in the edge length. Previously, Van Duyne and coworkers reported a similar behavior for NSL-fabricated truncated silver nanostructure.37 However, the EM-field decay lengths of these nanostructures are more than 2-fold lower than the chemically synthesized nanoprisms reported here. Interestingly, our EM-field decay lengths and bulk refractive index sensitivities for different edge length nanoprisms attached to silanized glass surfaces demonstrated opposite trends. On the basis of the literature, one would expect that larger nanostructures would display both higher bulk sensitivities and longer EM-field decay lengths; however, we found a reverse trend of refractive index sensitivity, where shorter edge-length nanoprisms displayed higher sensitivity. A potential explanation is that for our sensitivity measurements, the nanoprisms are in an inhomogeneous dielectric environment because the bottom 20997

dx.doi.org/10.1021/jp302674h | J. Phys. Chem. C 2012, 116, 20990−21000

The Journal of Physical Chemistry C

Article

was 10-fold lower than our previous report in which sensors were fabricated using 22 nm edge length nanoprisms.45 Figure 7 is very significant and indicates that the biosensing ability of plasmonic nanostructures can be greatly enhanced by increasing their sensing volume. It is known that changes in the local dielectric environment around the metallic nanostructures significantly affect their LSPR properties. Additionally, local sensitivity is a measure of the ΔλLSPR with respect to the penetration depth (“sensing volume”) of nanostructure within which changes of the refractive index can be detected. In effect, the larger edge length nanoprisms have higher sensing volume in comparison to the shorter length ones. Thus, with the increase of sensing volume, the LSPR sensitivity of the nanostructure is significantly enhanced and more strongly influenced by minute changes of their local dielectric environment due to the adsorption of a higher amount of analyte molecules onto the surface. We also investigated potential nonspecific interactions between sensing platforms and the biomolecule of interest. Three different control experiments were conducted to evaluate the nonspecific adsorption of SA to the surface of the 28 nm edge length nanoprisms. In the first experiment, a freshly prepared nonfunctionalized sensing platform was incubated in a 100 nM SA solution for 6 h in PBS buffer. After thorough washing with water, the extinction spectrum was recorded and no noticeable ΔλLSPR was observed (data not shown). In the second experiment, a mixed thiol-functionalized sensing platform was exposed to 100 nM of SA for 6 h, washed with water, and the extinction spectrum measured. In this case the largest shift observed after SA incubation (considering 3 trials) was 0.3 nm (data not shown), which could be due to instrumental noise or near negligible interaction between SA and the SAM surface. In the third control experiment, the biotin-functionalized sensing platform was incubated in 100 nM BSA solution in PBS buffer for 6 h. Figure 8 shows the

Figure 6. Shifts of LSPR dipole peak due to different stages of surface modification of sensing platforms fabricated with 28 nm edge length nanoprisms bound to silanized glass substrate. SA represents noncovalent binding of streptavidin to the sensing platform. The peak shift after modification with biotin was induced by a concentration of 1 μM SA.

functionalization. SI-Figure 4 provides UV−vis spectra at different functionalization steps and after 1 μM SA binding. The equilibrium binding kinetics of a biotin-functionalized sensing platform with different concentrations of SA was characterized in our previous report.45 The binding of SA to biotin was monitored by measuring the limiting ΔλLSPR for different SA concentrations ranging from 1 μM to 10 pM in combination with appropriate incubation times. Following this approach, Figure 7 represents the maximum ΔλLSPR (nm)

Figure 7. Average LSPR dipole peak shift as a function of SA concentration for sensing platforms (wetted surfaces) with biotinmodified nanoprisms. Nanoprisms of 28 (light blue), 42 (red), and 51 (dark blue) nm edge lengths are shown. To determine the average LSPR peak shifts, five different batches of nanoprisms were used to fabricate the sensing-platforms. Figure 8. Extinction spectra of biotin functionalized sensing platform fabricated with 28 nm nanoprisms before (blue) and after (red) incubation with 100 nM BSA.

obtained for different concentrations of SA using sensing platforms fabricated with nanoprisms of different edge length. The relationship between ΔλLSPR and concentration was linear for 10 pM to 1 nM SA. However, the overall trend was sigmoidal, which has been reported in the literature.2,3,13,23 The sensing platform constructed with 28 nm edge length nanoprisms yielded values of 28.5, 16.8, 10.6, 7.4, 5.5, 3.8, and 2.4 nm for ΔλLSPR with SA concentrations of 1 μM, 100 nM, 10 nM, 1 nM, 100 pM, 50 pM, and 10 pM, respectively. The lowest concentration of SA that we could reliably measure (10 pM) was much lower than LSPR sensing platforms based on gold nanorods (94 pM) or nanostars (100 pM). The LOD was found to be 5.0 pM. More importantly, this concentration

extinction spectrum as collected after washing with water. We observed an average 1.0 nm dipole peak shift (5 trials), which is over 15-fold lower than that measured with the same SA concentration. These control experiments indicate that the gold nanoprism-based plasmonic nanosensors are highly specific and that analyte molecules selectively bind to the attached receptors and not nonspecifically to the surface of the nanoprisms. To further explore the dependency of the sensing volume of nanoprisms on LSPR biosensing sensitivity, sensing platforms 20998

dx.doi.org/10.1021/jp302674h | J. Phys. Chem. C 2012, 116, 20990−21000

The Journal of Physical Chemistry C

Article

demonstrated that the removal of nonprismatic nanostructures from the sensing platforms increased their sensing efficiency. Furthermore, the shortest edge length nanoprisms displayed the highest refractive index sensitivity of 647 nm/RIU for 28 nm edge length nanoprisms. We have also shown that the EMfield decay length of nanoprisms was strongly edge length dependent with the longest edge length nanoprisms (51 nm) displaying the highest EM-field decay length. Interestingly, it was observed that as the edge length of the nanoprisms increased, their biosensing sensitivity also increased. The plasmonic nanosensors fabricated from 51 nm edge length nanoprisms were able to detect a SA concentration as low as 1 pM reproducibly with a LOD of 0.5 pM. To the best of our knowledge, this is the first example in which systematic studies were performed to optimize the refractive index sensitivity and biosensing sensitivity of anisotropically shaped, metallic nanostructure. However, we believe a more complete investigation and comparison of the refractive index sensitivities and FOM values for different edge length nanoprisms is needed, which could be performed through single particle scattering measurements. Such studies would be well complemented by theoretical calculations and provide a deeper understanding of this complex process at the atomic level.

containing two additional edge lengths were also prepared. As shown in Figure 7, nanoprisms with an edge length of 42 nm displayed average dipole peak shifts of 41.4, 30.5, 23.8, 15.3, 6.0, 4.2, and 3.7 nm for SA concentrations of 1 μM, 100 nM, 10 nM, 1 nM, 100 pM, 50 pM, and 10 pM, respectively. The LOD was found to be 3 pM. The dipole peak shifts for 51 nm edge length nanoprisms were 53.4, 39.6, 29.4, 17.5, 10.2, 9.0, and 8.0 nm, respectively, for the above-mentioned SA concentrations. Additionally, a 4.3 nm shift was observed for 1 pM SA and the LOD was determined to be 0.5 pM. These experimental results clearly suggest that as the sensing volume of a nanoprism increases, its ability for analyte detection is substantially enhanced. To the best of our knowledge, this is the first study of the effect of varying sensing volume on the LSPRbased biosensing ability of anisotropically shaped metallic nanostructure. More importantly, the lowest concentration of SA we could repeatedly measure (1 pM) is many orders of magnitude lower than that measured with chemically synthesized gold nanostars (100 pM) and nanorods (94 pM).2,6,23 It is important to mention that the sensitivities for either nanostarts or nanorods were determined in PBS buffer and in this present investigation the sensitivities were measured for wetted sensing platforms. In this context, PBS buffer has almost the same RI as water and therefore we can compare our LOD with literature reports. Our experimental finding is in good agreement with the previous report on truncated silver nanoprisms by the Van Duyne group, in which the authors were able to detect down to 1 pM of SA, which is the lowest concentration of analyte that could reliably be measured from the ensemble measurement.13 However, the measurements were conducted under N2 atmosphere (dry surface) instead of either in water or in buffer. This limits their potential applications for biological samples. Moreover, we believe that our current findings will lead to plasmonic sensors with advantages over those based on truncated silver nanoprisms. Metallic silver is prone toward atmospheric oxidation, and therefore all surface functionalization steps need to be performed under inert atmosphere. Fabrication of gold nanoprism-based sensors does not require a special experimental setup, and thus makes the process simple and more reliable. With respect to the future of gold nanoprism-based LSPR biosensors, we believe it is still possible to further improve sensitivity from that reported here. First, synthetic methods should be developed that produce ultrathin nanoprisms with an average edge length between 60 and 80 nm. Presumably, these nanostructures would have large sensing volume and their λLSPR would be found at