ARTICLE pubs.acs.org/JPCC
Effect of Protein Binding Coverage, Location, and Distance on the Localized Surface Plasmon Resonance Response of Purified Au Nanoplates Grown Directly on Surfaces Srinivas R. Beeram† and Francis P. Zamborini* Department of Chemistry, University of Louisville, Louisville, Kentucky 40292, United States
bS Supporting Information ABSTRACT: Here we report the effect of protein binding coverage, location, and binding distance on the localized surface plasmon resonance (LSPR) response of purified Au nanoplates grown directly on surfaces. The response to human anti-IgG binding on the nanoplate surface depends strongly on these variables. The method of anti-IgG attachment controlled the binding location (terrace, edge, or vertex sites) and the linker coverage and chain length controlled the antibody coverage and binding distance, respectively. The average change in λmax (Δλmax) for binding to terraces at medium coverage, edge sites at low coverage and edge sites at medium coverage was 21 ( 5, 44 ( 2, and 53 ( 4 nm on nanoplate samples purified by sonication, for example. The trend was similar for samples purified by taping, except that the shifts were all smaller due to a smaller initial λmax for those samples. Atomic force microscopy (AFM) images reveal the coverage and binding location of anti-IgG on the nanoplates and, when correlated with the LSPR spectra, confirm that the edge sites are more sensitive to protein binding and Δλmax increases with increasing anti-IgG coverage. The LSPR response generally increases as the chain length of the linker decreases and matches quite well with that predicted by theory in most cases, although there are some exceptions. Specifically, the LSPR shift increases in the order of mercaptopropionic acid (MPA) > 11-mercaptoundecanoic acid (MUA) > 16-mercaptohexadecanoic acid (MHDA) as linkers for anti-IgG. The average Δλmax for anti-IgG attached to the edge sites at high coverage on Au nanoplates purified by sonication was 61 ( 5, 53 ( 4, and 42 ( 3 nm, for the three linkers, respectively, for example.
’ INTRODUCTION Some noble metal nanoparticles absorb light in the UVvis region when the frequency of incident photons matches the collective oscillations of the conduction band electrons of the metal, which is known as localized surface plasmon resonance (LSPR).15 The result is a strong absorption band(s) or increased scattering intensity at specific wavelengths for metals like Au and Ag when monitoring the optical properties in transmission mode or reflection/dark-field mode, respectively. It is well-known that the intensity and wavelength of maximum absorbance/scattering (λmax) depends on the size, shape, and composition of the metal nanoparticles.1,2,4,6 It also depends on the refractive index of the environment surrounding the metal.1,4,7 If the size, shape, and composition are constant for a given nanostructure throughout an experiment, then the LSPR peak intensity and λmax are sensitive to changes in the environment, which has been exploited for sensing applications. The optical properties of a metal nanostructure functionalized with a chemical receptor change if a molecule binds to the receptor and significantly alters the refractive index of the medium directly surrounding the metal nanostructure. LSPR spectroscopy has been exploited in this way for sensing a wide variety of analytes, including metal ions,8,9 vapor molecules,10 polymers,11 and biomolecules.1,1224 The method r 2011 American Chemical Society
is especially promising for biosensing applications because it is highly sensitive, simple, low cost, and label-free.1,17,24 Reports of LSPR sensing of biological molecules, such as DNA20,21,23 and proteins,1,1219,22,24 have increased tremendously over the past few years. Englebiene and co-workers first reported on the extinction changes of Au nanoparticles in solution upon antibody binding.25 Other examples of solution-phase measurements include biopolymer adsorption kinetics,26 ligandprotein interactions,27 high throughput screening of proteins,28 pH,29 and ascorbic acid.29 Recently, Yu and Irudayaraj used antibody-functionalized Au nanorods of varying aspect ratio for multiplex biosensing.30 Most of the solution examples detect proteins in the μM to nM range. There are several studies on LSPR sensing with evaporated or chemically synthesized films of metal nanostructures. Chilkoti and co-workers synthesized and assembled Au nanoparticles17,31 and Au nanorods15 for sensing of streptavidin. Rubinstein and co-workers evaporated discontinuous Au films for sensing of avidin and antibodies specific to IgG and hCG antigens.14,32 Received: February 1, 2011 Revised: March 5, 2011 Published: March 28, 2011 7364
dx.doi.org/10.1021/jp2010869 | J. Phys. Chem. C 2011, 115, 7364–7371
The Journal of Physical Chemistry C
ARTICLE
Other examples include films of Au evaporated onto porous anodic alumina,33 fabricated silica/Au core/shell particles,34 Au nanoparticles grown from Au seeds,19 and Ag films sputtered35 or formed by glancing angle deposition36 for the detection of various analytes, such as IgG,36 streptavidin,35 aptamer-protein interactions,33 melittin,34 and BSA.19 Chemically synthesized Au nanoparticles3740 and nanorods16,41 assembled on various surfaces, including smooth Au films,40 SiO2,38 glass,16,39,41 and an optical fiber,37 led to optical sensing of avidin,38 streptavidin,37,41 peptides,40 IgG,16 and cholera toxin.39 Van Duyne and co-workers synthesized arrays of triangular Ag nanostructures by nanosphere lithography for LSPR sensing of streptavidin,42 a biomarker for Alzheimer’s disease,12,24 and a carbohydrate.43 Au nanorods15 and the optical fiber method37 displayed detection limits around 100 pM for streptavidin, whereas the Ag nanosphere lithography approach42 detected streptavidin below 1 pM. There are a few examples of biosensing studies with individual metal nanostructures, including the detection of streptavidin with an Au nanoparticle44 and Au nanorod,18,45 cytokines with an Ag nanoparticle,46 and DNA with an Au nanoparticle coupled to a larger Au nanoparticle.20 One of the major benefits of single nanoparticle biosensing is the narrow bandwidth in the spectrum due to the removal of broadening that occurs when measuring the collective optical signal from many different particles. This improves the ability to detect changes in λmax and increases sensitivity.47 Several factors control the sensitivity of metal nanostructures to a binding analyte. Chilkoti and co-workers recently used the following expression in their work on Au nanorods:45 Δλmax ¼ ½3S0 expð 2r=ld Þ=V S ½ΔRINV A
ð1Þ
Δλmax is the change in the wavelength of maximum extinction for the metal nanostructure, Vs is the sensing volume, which is commonly defined as the volume surrounding the nanostructure that contains 95% of its sensitivity,45 ΔRI is the difference in the refractive index (RI) between the binding molecule and the medium, N is the number of bound molecules, VA is the volume of one individual analyte molecule, r is the distance between the analyte and metal nanostructure, ld is the electromagnetic field decay length, and S0 is the bulk sensitivity to global changes in the RI. S0 has recently been found to be independent of the nanostructure geometry for a given metal; it only depends on the initial λmax.1,4,7 The second term, ΔRINVA, represents the optical mass increase as a result of the bound analyte.45 This term only depends on the properties of the analyte and the medium; not the metal nanostructure. Δλmax decreases exponentially as r increases due to the exponential decrease in the electromagnetic field enhancement as a function of distance from the nanoparticle surface, known as the electric field enhancement decay length, ld. The result is that the LSPR band is not sensitive to analyte molecules greater than ∼4050 nm from the nanostructure surface.17 The equation described above does not take into account the nonhomogeneous distribution of the electric field enhancement across a particular nanostructure. It has been predicted that sharp edges and corners of nanostructures exhibit much larger electric field enhancement compared to smooth or gradually curving regions.48 For example, the calculated electric field enhancement is large at the ends of nanorods, vertices of triangular nanoplates, points of branched structures, or within coupled nanoparticle pairs.48,49 These regions are considered to be “hot spots” that are
highly sensitive to molecules binding in the vicinity. If the analyte binds to a region with a higher electromagnetic field strength, then VA/Vs is effectively larger and ld is larger in that region, leading to a large Δλmax for that binding event. Several groups have tried to place molecules within the “hot spots” of metallic nanostructures for SERS applications. However, surprisingly, there has been no strong effort to do this for applications in LSPR-based sensing. The vast majority of reports simply coat the surface with a receptor and monitor the LSPR shift upon analyte binding without trying to control or characterize the binding location. Van Duyne and co-workers previously showed large shifts in the LSPR band for an array of triangular Ag nanostructures upon nonspecific adsorption of antibody onto the Cr adhesion layer, which localized the antibody near the edges,12 and Sannomiya et al. observed very different changes in the optical properties of an individual large Au nanoparticle upon binding of smaller Au nanoparticles through DNA hybridization, suggesting that the optical shift depends on the binding location.20 We recently demonstrated the controlled binding of human anti-IgG to the edge and vertex sites of Au nanoplates in an attempt to correlate the LSPR response with the binding location.50 Unfortunately, the binding location as determined by atomic force microscopy (AFM) of the Au nanoplates could not be correlated to the LSPR spectrum of Au nanoplates because the sample was dominated by ∼80% nanospheres. In a follow up study, we synthesized samples of >90% Au nanoplates by selectively removing the spherical nanoparticles with sonication or by taping the sample.51 That allowed us to correlate the shift in λmax with the AFM determined binding location of the Au nanoplates. We also determined that the LSPR shift in λmax was significantly larger for anti-IgG bound to the edge sites of Au nanoplates versus Au nanospheres, which is due to a larger initial λmax and therefore greater S0 (see eq 1) for nanoplates. We detected ∼0.7 pM IgG using the edge-exchanged purified Au nanoplates, which is about an order of magnitude better than that with edge-exchanged Au nanospheres. Here we directly compare the LSPR response to anti-IgG binding on edge sites to those bound on terrace sites of purified Au nanoplates and additionally study the effect of anti-IgG binding distance (r) from the Au nanoplates. While others have studied the LSPR shift of metallic nanostructures as a function of the thickness of coatings (usually polymers),52,53 there has been no study on the effect of binding distance for antibodies bound to edge sites. The findings of our work represent an important step in our overall goal to control the synthesis of metallic nanostructures and receptor binding surface chemistry in order to optimize their sensitivity for LSPRbased biological and chemical sensing applications.
’ EXPERIMENTAL SECTION Chemicals. Citric acid trisodium salt was purchased from biorad laboratories. L-Ascorbic acid (99%), sodium borohydride (98.5%), cetyltrimethylammonium bromide (CTAB), 3-mercaptopropionic acid (MPA), 11-mercaptoundecanoic acid (MUA), 16-mercaptohexadecanoic acid (MHDA), 2-mercaptoethanol (ME), and N-hydroxysuccinimide (NHS) were purchased from Sigma-Aldrich. 1-Ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC) was purchased from Pierce, and mercaptopropyltrimethoxysilane, 95% (MPTMS), was purchased from Alfa Aesar. Antibody human anti-IgG was purchased from SigmaAldrich. 7365
dx.doi.org/10.1021/jp2010869 |J. Phys. Chem. C 2011, 115, 7364–7371
The Journal of Physical Chemistry C Synthesis of Au Nanostructures Directly on Surfaces. We synthesized Au nanostructures on glass and Si/SiOx surfaces using a seed-mediated growth procedure as described by our group previously.5458 This method is based on the seedmediated synthesis of Au nanorods in solution described by Murphy and co-workers.59 Glass and silicon slides were first cut and cleaned in piranha solution (1:3 H2O2:H2SO4) for 1015 min. (Caution: this solution, piranha, is a strong oxidizing agent that reacts violently with organics.) After rinsing with water and drying under a stream of N2, the substrates were functionalized with mercaptopropyltrimethoxy silane (MPTMS) by heating them just below boiling in a solution containing 10 mL of 2-propanol, 100 μL of MPTMS, and a few drops of water for about 30 min. After rinsing with 2-propanol and drying under N2, the MPTMS-functionalized silicon and glass slides were placed in an aqueous solution of 35 nm diameter citrate-stabilized Au nanoparticles (“seeds”) for 15 min, which leads to their attachment to the thiol functionality of MPTMS through a strong Authiolate interaction. After rinsing with water, the substrates containing immobilized Au seed nanoparticles were then placed in a freshly prepared growth solution containing 9 mL of 0.1 M cetyltrimethylammonium bromide (CTAB), 450 μL of 0.01 M HAuCl4, and 50 μL of 0.01 M ascorbic acid for 1 h. The samples were removed, rinsed copiously with water, and dried under N2 before further use. As described previously, the source of the CTAB (Aldrich 95%) was critical for synthesizing samples with a large population of nanoplates. Nanoplate Purification. Purification by Tape. We used Scotch brand magic adhesive tape to preferentially remove the spherical nanoparticles from the glass or Si/SiOx surface as described previously.51 The tape was placed on the substrate, pressed gently with one finger and then slowly peeled back at an approximately 90° angle. In the case of glass, the procedure was performed on both sides. All taped samples were placed in a solution of dichloromethane for 67 h, rinsed with isopropyl alcohol and water, and then dried under nitrogen before anti-IgG functionalization as described later to remove any tape residue from the surface. Purification by Sonication. We used a Bransonic ultrasonic cleaner with a 935 W input and 250 W output puissance HF to remove the spherical nanoparticles from substrates containing Au nanostructures.51 The substrate was placed in a glass vial containing 10 mL of nanopure water and then placed in the ultrasonicator for 5 min. The substrate was removed from the vial, washed thoroughly with nanopure water, and dried under N2. Procedure for Functionalizing the Au Nanoplates with Anti-IgG. We functionalized the purified Au nanoplates with human anti-IgG by two methods as described previously.50 In the first method, termed “pure MUA”, we placed the sample in a 1 mM ethanol solution of mercaptoundecanoic acid (MUA) for 1215 h, rinsed thoroughly with ethanol, dried under N2, and then placed the sample in an aqueous solution of 2 mM 1-ethyl3-(3-dimethylaminopropyl) carbodiimide (EDC) and 5 mM N-hydroxy succinimide (NHS) for 1 h. After rinsing with water and drying under N2, we placed the sample in an aqueous pH 7.4 phosphate buffered solution of 0.26 μg/mL human anti-IgG solution for 1215 h in the refrigerator, rinsed with phosphate buffered saline and water, and dried under N2. In the second procedure, termed “place-exchange”, we placed the sample in a 1 mM ethanol solution of mercaptoethanol (ME) overnight and then exchanged the ME monolayer with MUA by placing the sample into a 5 or 6 mM ethanol solution of MUA for 4 h. Finally,
ARTICLE
we attached the human anti-IgG via EDC and NHS coupling as described for the pure MUA strategy. Instrumentation. LSPR extinction spectra were obtained with a Varian CARY 50 BIO UVvisible spectrophotometer. Glass samples containing the Au nanoplates were marked and placed in the optical path before and after attachment of the human anti-IgG. The spectra were measured from 300 to 1000 nm and the noise was reduced by using the smooth operation in the Varian software with a filter size of 101. The changes in wavelength of maximum extinction (λmax) were calculated by allowing the software to chose the λmax of the smoothed spectra. The spectrum of an Au nanoplate sample on a glass slide was monitored several times by repeatedly placing the sample in the spectrophotometer and removing to confirm that the slide could be reproducibly placed in the same location of the spectrophotometer. These spectra showed less than a 1% shift in extinction and taped > nonpurified for all three anti-IgG attachment strategies. The order of the magnitude of Δλmax follows the same order of λmax, initial. The initial average λmax was 704766 nm for sonicated samples, 644675 nm for taped samples, and 539547 nm for nonpurified samples, respectively. As λmax,initial increased, Δλmax increased. The λmax,initial value increases with increasing AR of the nanostructures as shown in Table 1. It is well-known that the bulk RI sensitivity (S0) increases as the AR and λmax, initial increases. This explains the trend in sensitivity for the different purification methods. The large Δλmax in the sonicated samples is promising for increasing the sensitivity to protein binding. Comparing anti-IgG attachment strategies, the trend followed as 6 mM exchange > 5 mM exchange > pure MUA. The average Δλmax was 53 ( 4, 44 ( 2, and 21 ( 5 nm for samples functionalized by the 6 mM place-exchange, 5 mM placeexchange, and pure MUA strategies, respectively on sonicated samples whereas the samples purified by tape showed values of 32 ( 3, 23 ( 1, and 12 ( 1 nm, respectively. Nonpurified samples (77% Au nanospheres) showed Δλmax values of 9 ( 1, 6 ( 3, and 1 ( 1 nm for the three strategies, respectively, as reported previously.50 We conclude that the different Δλmax values for place-exchange and pure MUA samples are due to different anti-IgG binding locations on the Au nanoplates based on the AFM images, which showed anti-IgG bound primarily to edge sites for the place-exchange samples and terrace sites for the pure MUA samples (Figure 1). This indicates that the LSPR band is at least 23 times more sensitive to anti-IgG bound to the edge or vertex sites compared to flat terrace sites as predicted previously48 and shown by our group on spherical particles.50 This was true even though the anti-IgG coverage was apparently as large or larger on the terraces of the Au nanoplates functionalized by the pure MUA strategy compared to the coverage on the edge sites using the 5 mM MUA place-exchange. The larger Δλmax values of the 6 mM place-exchange samples relative to the 5 mM samples were due to the larger coverage of anti-IgG on the edge sites for the former (Figure 1 and Figure S3). In order to better understand the relative sensitivity of the edge sites compared to the terraces, we prepared two samples purified by taping, functionalized with pure MUA, and coated with a very large coverage of anti-IgG from a 250 μg/mL solution. This led to an average Δλmax of 40 nm from the two samples (Figure S13). If we assume the 40 nm shift is the maximum possible shift for the taped samples completely coated with antiIgG, then partial attachment of anti-IgG to the edge sites (by 5 or 6 mM MUA place-exchange) led to a signal that is 5080% of
the maximum possible signal. This is very large considering that all of the edges did not appear covered in the AFM images. This clearly shows that most of the LSPR sensitivity is confined to the edge sites and likely accounts for greater than 80% of the sensitivity. This also means that >80% of the sensing volume, Vs, is at these edge and vertex sites. We note that there was no clear trend in the change in extinction of the LSPR band of the samples using the different attachment strategies (not shown). The results were highly irreproducible for a reason that is unknown at this time, making it difficult to draw any conclusions about the effect of protein binding on the LSPR extinction. This is an area that needs further exploration. The average Δλmax was
