Heterodimeric Plasmonic Nanogaps for Biosensing

We report the study of heterodimeric plasmonic nanogaps created between gold nanostar (AuNS) tips and gold nanospheres. The selective binding is realized by properly functionalizing the two nanostructures; in particular, the hot electrons injected at the nanostar tips trigger a regio-specific chemical link with the functionalized nanospheres. AuNSs were synthesized in a simple, one-step, surfactant-free, high-yield wet-chemistry method. The high aspect ratio of the sharp nanostar tip collects and concentrates intense electromagnetic fields in ultrasmall surfaces with small curvature radius. The extremities of these surface tips become plasmonic hot spots, allowing significant intensity enhancement of local fields and hot-electron injection. Electron energy-loss spectroscopy (EELS) was performed to spatially map local plasmonic modes of the nanostar. The presence of different kinds of modes at different position of these nanostars makes them one of the most efficient, unique, and smart plasmonic antennas. These modes are harnessed to mediate the formation of heterodimers (nanostar-nanosphere) through hot-electron-induced chemical modification of the tip. For an AuNS-nanosphere heterodimeric gap, the intensity enhancement factor in the hot-spot region was determined to be 106, which is an order of magnitude greater than the single nanostar tip. The intense local electric field within the nanogap results in ultra-high sensitivity for the presence of bioanalytes captured in that region. In case of a single BSA molecule (66.5 KDa), the sensitivity was evaluated to be about 1940 nm/RIU for a single AuNS, but was 5800 nm/RIU for the AuNS-nanosphere heterodimer. This indicates that this heterodimeric nanostructure can be used as an ultrasensitive plasmonic biosensor to detect single protein molecules or nucleic acid fragments of lower molecular weight with high specificity.


Introduction
Real-time and label-free detection of protein molecules at ultralow concentration in their natural state is considered a longstanding need in biomedical research, as it could allow for the identification of the onset of cancers before they become clinically relevant, to monitor disease progression and evaluate therapeutic success, thereby increasing survival rates and quality of life [1]. However, the detection of single protein molecules is extremely challenging because of their acutely small size (<3 nm) [2]. One way to detect these proteins is to use the unusual electromagnetic responses of electron clouds of noble metal nanoparticles (NPs), which leads to the well-known localized surface plasmon resonance damage in the samples. The zero-loss peak for each spectrum was removed using the standard reflected tail method, which reflects the tail on the energy-gain side of the spectrum onto the energy-loss side, typically with a predefined scaling factor, and subtracts it [63][64][65][66].
Numerical Simulation: FEM simulations are used here to study the interaction of a plasmonic biosensor of a single AuNS and a hybrid AuNS-nanosphere with photons. All the data regarding nanostars spikes and their cores are taken from the TEM studies. The nanostructure's surface is discretized with tetrahedral mesh elements with a typical maximum and minimum side lengths of 24.5 and 1.05 nm, respectively. During simulations, the electric field was mapped in the entire space. Photon interaction with single AuNS and AuNS-nanosphere heterodimer with a different nanosphere size and a gap distance were studied. Different cross sections (Scattering, Absorption and Extinction) of those nanoantennas were also examined. During sensitivity calculation, we considered first air (ε = 1.0) and then water (ε = 1.7689) as the homogeneous surrounding media. In all simulations, the wavelength-dependent permittivity of the gold was taken from [67]. We also calculated the shift in LSPR of the single AuNS and the AuNS-nanosphere heterodimer when a BSA molecule (of 0.11 attogram mass) is coupled in the hot spot region of those nanoantennas immersed in water medium. During simulation, a BSA protein molecule was taken as a dielectric particle of cylindrical shape with a permittivity of 2.25, with a characteristic height of 3.4 nm and a 6.8 nm short cylinder diameter [68].

Results and Discussion
To obtain nanostars in aqueous suspension with high yield and stability, there are parameters that need to be precisely controlled during synthesis. These parameters are summarized in this section (see Methods and Supporting Information for more details). One of these parameters is the relative amount of reducing agent to the gold precursor. Here, ascorbic acid (AA, C 6 H 8 O 6 ) was used as the reducing agent, and chloroauric acid (HAuCl 4 ) was used as the gold precursor. The ratio of AA to chloroauric acid was maintained to be 1.5-2.0 to reduce all HAuCl 4 molecules present in the solution completely. Secondly, the presence of Ag + (AgNO 3 ) was necessary in this synthesis process for nanostar formation. During synthesis, the main function of Ag + was to expedite the anisotropic growth of Au branches on certain crystallographic facets. The synthesis procedure will only yield polydispersed nanorods and nanospheres in the absence of Ag + (AgNO 3 ). Increasing the amount of AgNO 3 during the synthesis will increase the length and the number of spikes of the AuNS (see Figure S1). The third factor was the presence of HCl. A small amount of HCl helped in this case to slightly decrease the pH of the solution and a red-shifted plasmon band [33]. The fourth factor was the simultaneous injection of AA and AgNO 3 , in the HAuCl 4 solution (in the presence of HCl), because this will significantly influence the polydispersity, yield, and quality of the nanostars. The fifth factor was the stirring speed. A study to probe the best stirring speed found that 700 rpm was the best for a successful synthesis as the measured extinction cross section for the AuNS solution is highest for that speed for any specific bar (see Figure  S2). The sixth factor was the injection time of polyvinylpyrrolidone (PVP). The best injection time of PVP for a successful synthesis was after 2 min from the time of simultaneous injection of AA and AgNO 3 (see Figure S3), as by that time the reduction process had quite enough time to be completed and was not affected by the PVP addition by any means. Here, PVP concentration was taken lower so that the reduction kinetics was not affected. PVP was added for creating a hindrance to the nucleation process among the nanostars only for a highly stable AuNS aqueous suspension. The nanostars were found to be stable in aqueous solution for more than five months. During AuNS synthesis and the study of different synthesis parameters, the effect of slight variations in temperature was neglected, as the temperature was never above 60 • C, so any observable effect on the synthesis end product due to temperature variation was not expected [51]. Figure 1 shows the X-ray energy dispersive spectroscopy (XEDS) of the synthesized AuNS solution. In the spectrum, the most intense peak corresponds to carbon due to the carbon tape substrate used. The peak corresponding to Au is the second most intense peak in the spectrum, and the weight percentage of the gold present in our synthesized gold nanoparticle solution is 22.51%. Figure 1a  shows a typical lower magnification TEM image where almost all nanoparticles were found to have at least one spike in their surface, confirming the relatively high yield of the synthesis method. Figure 1b shows a randomly selected higher magnification TEM image of a gold nanoparticle.
Micromachines 2018, 9, x 5 of 13 substrate used. The peak corresponding to Au is the second most intense peak in the spectrum, and the weight percentage of the gold present in our synthesized gold nanoparticle solution is 22.51%. Figure 1a shows a typical lower magnification TEM image where almost all nanoparticles were found to have at least one spike in their surface, confirming the relatively high yield of the synthesis method. Figure 1b shows a randomly selected higher magnification TEM image of a gold nanoparticle.   Figure 2b shows the difference between the normalized experimental extinction characteristics of the stabilized AuNS solution and the corresponding theoretical investigations. Here, two modes are distinctive in the experimental extinction spectrum. As per the data collected from the TEM images of several AuNSs, the average spike length (ASL) of the nanostars was about 63 nm, taking into account all possible spike lengths from the smallest spike length (SSL) of 33 nm to the largest spike length (LSL) of 90 nm, and the average diameter of the core was almost 60 nm. Depending on the length of the spike, the tip radius varied from 5 to 1 nm. FEM simulations were performed for two AuNSs with ASL and LSL spike lengths and with the same core diameter of 60 nm to determine their extinction characteristics. The purple and blue curves in Figure  2b represent the normalized extinction spectra of the theoretical AuNS with ASL and LSL spike lengths, respectively. By considering the convolution of both theoretical curves, we can conclude that the resultant theoretical extinction characteristics match well with the experimental curve profile and the LSP resonances.   Figure 2b shows the difference between the normalized experimental extinction characteristics of the stabilized AuNS solution and the corresponding theoretical investigations. Here, two modes are distinctive in the experimental extinction spectrum. As per the data collected from the TEM images of several AuNSs, the average spike length (ASL) of the nanostars was about 63 nm, taking into account all possible spike lengths from the smallest spike length (SSL) of 33 nm to the largest spike length (LSL) of 90 nm, and the average diameter of the core was almost 60 nm. Depending on the length of the spike, the tip radius varied from 5 to 1 nm. FEM simulations were performed for two AuNSs with ASL and LSL spike lengths and with the same core diameter of 60 nm to determine their extinction characteristics. The purple and blue curves in Figure 2b represent the normalized extinction spectra of the theoretical AuNS with ASL and LSL spike lengths, respectively. By considering the convolution of both theoretical curves, we can conclude that the resultant theoretical extinction characteristics match well with the experimental curve profile and the LSP resonances. Micromachines 2018, 9, x 6 of 13 Although experimental and theoretical studies of the extinction properties of the AuNS solution provided relevant information about the ensemble behavior of AuNSs in aqueous solution, to move towards a controlled construction of hybrid plasmonic nanogaps for sensing, local information regarding the plasmonic field distribution and resonance frequencies is necessary. To use the regio-specific surface chemistry method based on hot electron injection introduced by Cortés et al. [56] for the creation of heterodimeric nanogaps between AuNSs and Au nanospheres, STEM-EELS investigations of the nanostructure were performed to gain precise information about single AuNS plasmonic modes. Figure 3 illustrates the EELS analysis of the synthesized AuNS. Figure 3a shows a STEM-HAADF image of the AuNS that was used for EELS analysis. The boxes on the image indicate the regions from which the EELS spectrum images were acquired. Shown in Figures 3c,d are the spectrum images in the center of the nanostar (the black dotted box in Figure 3a) and the spike of the nanostar (colored boxes within the orange box area in Figure 3a).  Although experimental and theoretical studies of the extinction properties of the AuNS solution provided relevant information about the ensemble behavior of AuNSs in aqueous solution, to move towards a controlled construction of hybrid plasmonic nanogaps for sensing, local information regarding the plasmonic field distribution and resonance frequencies is necessary. To use the regio-specific surface chemistry method based on hot electron injection introduced by Cortés et al. [56] for the creation of heterodimeric nanogaps between AuNSs and Au nanospheres, STEM-EELS investigations of the nanostructure were performed to gain precise information about single AuNS plasmonic modes. Figure 3 illustrates the EELS analysis of the synthesized AuNS. Although experimental and theoretical studies of the extinction properties of the AuNS solution provided relevant information about the ensemble behavior of AuNSs in aqueous solution, to move towards a controlled construction of hybrid plasmonic nanogaps for sensing, local information regarding the plasmonic field distribution and resonance frequencies is necessary. To use the regio-specific surface chemistry method based on hot electron injection introduced by Cortés et al. [56] for the creation of heterodimeric nanogaps between AuNSs and Au nanospheres, STEM-EELS investigations of the nanostructure were performed to gain precise information about single AuNS plasmonic modes. Figure 3 illustrates the EELS analysis of the synthesized AuNS. Figure 3a shows a STEM-HAADF image of the AuNS that was used for EELS analysis. The boxes on the image indicate the regions from which the EELS spectrum images were acquired. Shown in Figures 3c,d are the spectrum images in the center of the nanostar (the black dotted box in Figure 3a) and the spike of the nanostar (colored boxes within the orange box area in Figure 3a).   Figure 3a shows a STEM-HAADF image of the AuNS that was used for EELS analysis. The boxes on the image indicate the regions from which the EELS spectrum images were acquired. Shown in Figure 3c,d are the spectrum images in the center of the nanostar (the black dotted box in Figure 3a) and the spike of the nanostar (colored boxes within the orange box area in Figure 3a).
In the low energy-loss region, the spectrum exhibits one major peak at 2.2 eV (~564 nm). Numerical investigation of the Au nanosphere with a 60 nm diameter shows LSPR at 550 nm (see Figure S4), suggesting that the 2.2 eV is associated with the LSPR at the core of AuNS. The spectra from the spike (Figure 3d) exhibit two peaks at 1.2 eV (~1033 nm) and 1.8 eV (~689 nm). The relative intensity of the two peaks varies spatially along the spike as shown in the intensity maps of 1.2 eV mode and 1.8 eV mode in Figure 3b. The peak at 1.2 eV shows a maximum intensity at a position approximately halfway along the length of the spike, but the presence of the 1.8 eV mode can also be seen at that location (see Figure S5). At the pinnacle of the 88 nm spike (the region within the cyan box), the major plasmonic resonance is at 1.2 eV. On the other hand, the 1.8 eV mode exhibits maximum intensity at a location closer to the core of the nanostar. FEM investigation on the extinction properties for the AuNSs with an 88 nm spike length and a 60 nm core shows that the LSPR occurs at 1060 nm (see Figure S6). Both the numerical result and the EELS investigation indicate that the mode that is predominating in an AuNS tip region is 1.2 eV, which is crucial information to design regio-specific interactions that allow selected nanostructures to be bound at the tip of the AuNS via the non-localized surface chemistry method. Figure 4 shows how a regio-specific interaction-driven by hot-electron injection-can promote the formation of a heterodimeric (nanostar-nanosphere) nanostructure for plasmonic biosensing via the method introduced by Cortés et al. [61]. In the low energy-loss region, the spectrum exhibits one major peak at 2.2 eV (~564 nm). Numerical investigation of the Au nanosphere with a 60 nm diameter shows LSPR at 550 nm (see Figure S4), suggesting that the 2.2 eV is associated with the LSPR at the core of AuNS. The spectra from the spike (Figure 3d) exhibit two peaks at 1.2 eV (~1033 nm) and 1.8 eV (~689 nm). The relative intensity of the two peaks varies spatially along the spike as shown in the intensity maps of 1.2 eV mode and 1.8 eV mode in Figure 3b. The peak at 1.2 eV shows a maximum intensity at a position approximately halfway along the length of the spike, but the presence of the 1.8 eV mode can also be seen at that location (see Figure S5). At the pinnacle of the 88 nm spike (the region within the cyan box), the major plasmonic resonance is at 1.2 eV. On the other hand, the 1.8 eV mode exhibits maximum intensity at a location closer to the core of the nanostar. FEM investigation on the extinction properties for the AuNSs with an 88 nm spike length and a 60 nm core shows that the LSPR occurs at 1060 nm (see Figure S6). Both the numerical result and the EELS investigation indicate that the mode that is predominating in an AuNS tip region is 1.2 eV, which is crucial information to design regio-specific interactions that allow selected nanostructures to be bound at the tip of the AuNS via the non-localized surface chemistry method. Figure 4 shows how a regio-specific interaction-driven by hot-electron injection-can promote the formation of a heterodimeric (nanostar-nanosphere) nanostructure for plasmonic biosensing via the method introduced by Cortés et al. [61]. The plasmonic biosensor detection mechanism is based on the measurement of the LSPR shift of a nanoantenna when a protein molecule enters the region of space of its plasmon field. According to the sensing principle, when a protein molecule is nearby a plasmonic antenna, it feels an intense attraction toward the hot-spot region because of the force induced by the strong light confinement [68]. Therefore, when a small protein molecule is adsorbed onto the surface of the nanoparticle; the evanescent field starts to polarize the molecule. As the local field at the hot spot of the nanoantenna has to do some work to polarize the molecule the overall energy of the mode reduces accordingly. Experimentally, this situation can be visualized as a shift of the previously observed plasmonic mode of the nanoantenna, which is actually a narrow dip in the transmission spectrum or a sharp peak in the elastic spectrum [68][69][70]. In this case, the local field of the nanogap behaved similarly in the presence of the bio-nanoparticle. The energy of the hybrid mode of the heterodimeric nanogap reduces in presence of a foreign molecule. This reduction in the energy of the hybrid mode was realized from its shift in the transmission spectrum and thus can be used for molecular detection. Here, because of the high electric field enhancement at the binding site (nanogap) and the optimum distribution of the field, the gap mode shift for a single (or a few) molecule of lower molecular weight (such as the BSA protein molecule with 66.5 kDa) is detectable (numerical investigation supports this claim). This detection could be even more specific by attaching a probe molecule at the gap.
For large protein molecules (>1 MDa), energy loss to polarize the molecule is higher, so a larger LSPR shift that is easy enough to detect occurs; however, for the case of a protein molecule of ultrasmall molecular weight and thus smaller polarizability, the loss of energy of the LSPR shift becomes difficult to detect. Recently, single protein molecule detection has been reported using a plasmonic biosensor based on nanorods [69]. The reported S/N was very low, as the FWHM line width of the LSPR spectrum was larger (~50 nm) than the average wavelength shift (~0.1 nm) produced due to the adsorption of a molecule on the surface of the nanoplasmonic particle [69]. Therefore, real-time (label-free) detection of protein molecules of lower molecular weight (e.g., the Thyroglobulin molecule of a 660 kDa molecular weight and a BSA molecule of a 66.5 kDa molecular weight) in ultralow concentration appears to be forbidden with current technologies.
Our synthesized AuNSs with sharp spikes can create great intensity enhancement at their hot spots after resonant light illumination. A hybrid plasmonic nanogap constructed by coupling Au nanosphere at the tip of the AuNS via the above-mentioned hot electron mediated method realizes an excellent nanoscale detector for single small protein molecules. This hybrid nanogap can be converted to a sensor by immobilizing a capture molecule (probe) at the nanogap to specifically recognize molecules at a point of care. Figure 5 shows the interaction of a single plasmonic nanogap with a BSA protein molecule. Panel (a) reports how a single plasmonic nanostar antenna behaves when it interacts with photons in water. Panel (b) shows how the situation is modified if a BSA molecule is located in close proximity to the tip. Panel (c) is similarly describing the photon interaction of a heterodimer composed of a nanostar with an 88 nm spike length and a 100 nm gold nanosphere separated by a distance of 3.4 nm in water. Panel (d) shows the same interaction in the presence of a single BSA protein molecule situated in the heterodimeric nanogap. For this investigation, the incident light polarization has been assumed to be along the z-direction, whereas the propagation direction has been assumed to be the +y-direction.
All the extinction properties corresponding to a single AuNS and an AuNS-nanosphere heterodimer plasmonic biosensor immersed in water in the presence and absence of a BSA molecule are shown in Figures S6 and S7, respectively. The sensitivity of the heterodimeric nanogap can be easily compared with the single AuNS tip sensitivity based on the FEM simulation results.
All the extinction properties corresponding to a single AuNS and an AuNS-nanosphere heterodimer plasmonic biosensor immersed in water in the presence and absence of a BSA molecule are shown in Figures S6 and S7, respectively. The sensitivity of the heterodimeric nanogap can be easily compared with the single AuNS tip sensitivity based on the FEM simulation results. Here in all cases, the incident light wave has polarization along the z-axis and propagation along the +y direction. The AuNS has a spike length of 88 nm and a 60 nm core, and the size of the Au nanosphere is 100 nm.
In the above FEM simulation results, the enhancement measurement is given in terms of the quantity|E E ⁄ |. However, the shift in the wavelength due to the adsorption of the protein molecule is proportional to the light intensity ( = ) at the binding site. That is why it is better to interpret the results in terms of the light intensity enhancement. The intensity enhancement by a 33 nm AuNS spike (the smallest spike length) in water is approximately 1.6 × 10 3 , whereas the intensity enhancement for 88 nm spikes is approximately 2.5 × 10 5 in water. We have evaluated that a single BSA protein molecule (molecular weight ~ 66.5 kDa) in water can produce a measurable shift (>1 nm) if it is located in close proximity of an 88 nm nanostar spike, having a sensitivity of about 1940 nm/RIU. It is worth noticing that the remarkable tenfold increase of the light intensity enhancement in the absence of protein molecules within the heterodimeric gap with respect to a single nanostar spike. Strikingly, for a 3.4 nm heterodimeric gap (100 nm gold nanosphere-88 nm AuNS spike), the Here in all cases, the incident light wave has polarization along the z-axis and propagation along the +y direction. The AuNS has a spike length of 88 nm and a 60 nm core, and the size of the Au nanosphere is 100 nm.
In the above FEM simulation results, the enhancement measurement is given in terms of the quantity |E/E 0 |. However, the shift in the wavelength due to the adsorption of the protein molecule is proportional to the light intensity ( I I 0 = E E 0 2 ) at the binding site. That is why it is better to interpret the results in terms of the light intensity enhancement. The intensity enhancement by a 33 nm AuNS spike (the smallest spike length) in water is approximately 1.6 × 10 3 , whereas the intensity enhancement for 88 nm spikes is approximately 2.5 × 10 5 in water. We have evaluated that a single BSA protein molecule (molecular weight~66.5 kDa) in water can produce a measurable shift (>1 nm) if it is located in close proximity of an 88 nm nanostar spike, having a sensitivity of about 1940 nm/RIU.
It is worth noticing that the remarkable tenfold increase of the light intensity enhancement in the absence of protein molecules within the heterodimeric gap with respect to a single nanostar spike. Strikingly, for a 3.4 nm heterodimeric gap (100 nm gold nanosphere-88 nm AuNS spike), the intensity enhancement was found to be 2.3 × 10 6 in the absence of a BSA molecule in water. Hence, such a heterodimeric plasmonic sensor can detect a single BSA molecule with a sensitivity of about 5800 nm/RIU, and its sensitivity is expected to increase further by decreasing the size of the asymmetric tip-sphere nanogap.

Conclusions
To summarize, we reported the study of heterodimeric nanostructures for plasmonic sensing based on the formation of nanogaps between nanostar tips and nanospheres. The stable, highly tunable AuNSs were synthesized here via a simple, one-step, surfactant-free, wet-chemistry method with high yield and characterized via optical and EELS spectroscopic investigations. The formation of the nanogap is controlled and triggered by exploiting the light-induced hot-electron injection at the tip of the nanostars through the localized surface chemistry method. These hybrids and asymmetric plasmonic nanogaps can dramatically confine the incident electromagnetic field at their hot spots, irrespective of the incident light polarization. The numerical investigation supports the experimental results and predicted an ultrahigh sensitivity of this single molecule plasmonic biosensor (>5000 nm/RIU). This result will allow for the fabrication of nanogap sensors with high sensitivity and specificity for proteins and nucleic acids tests, which can find large use in point-of-care diagnostic technologies based on an easy and accurate optical readout.