FTu4E.3.pdf
CLEO:2015 © OSA 2015
Plasmonic Nanoantennas on Nanopedestals for UltraSensitive Vibrational IR-Spectroscopy Dordaneh Etezadi1,2, Arif E. Cetin1,2 and Hatice Altug1,2,† 2
1 Department of Electrical and Computer Engineering, Boston University, MA 02215, USA Bioengineering Department, Ecole Polytechnique Federale de Lausanne (EPFL), Lausanne CH-1015 SWITZERLAND †
[email protected]
Abstract: We experimentally demonstrate that elevating polarization-insensitive nanoring antennas on nanopedestals enables high surface enhanced infrared absorption (SEIRA) signals. This is due to larger and accessible nearfields offering better overlap with biomolecules.
Recent advancements in Surface Enhanced Infra-Red Absorption (SEIRA) spectroscopy have demonstrated unique opportunities in bioapplications as it provides direct access to rich finger-print vibrational information of most organic molecules. The use of SEIRA for non-destructive, label-free and ultra-sensitive spectroscopy as well as high-resolution microscopy, is still at its infancy. For applications associated with extremely low signal levels, such as biosamples with ultralow analyte concentrations, high signal-to-noise ratio (SNR) is required for reliable spectroscopic analysis within minimum detection time. Under such circumstances, to increase the SEIRA signal levels, it is highly important to (i) maximize the field overlap with the target biosample and (ii) preserve the maximum number of photons interacting with the sample. In this work, we introduce a polarization-insensitive midIR nanoring antenna system fabricated on a dielectric nanopedestal to provide maximum field overlap with the target biomolecules, while effectively leverage the incident light source. Our strategy of exposing antenna tips supporting larger and accessible nearfields can be also important for applications demanding strong light-matter interactions. Figure 1a and 1b show the schematic and scanning electron microcopy (SEM) image of the gold nanoring antenna design on the silicon nitride nanopedestal. The antenna response can be finely tuned by simply changing the circumference, where the plasmonic resonance scales linearly with the ring radius, i.e., Rout shown in the figure inset. More importantly, the symmetric nature of the nanoring antennas eliminates the need for the use of polarizers in the measurements and provides an alignment insensitive design that can utilize the total power of the incident source. Figure 1c is the cross-sectional profile of the nearfield intensity enhancement distribution supported by the nanoring antennas. For the two antennas engineered for the same protein vibrational mode (Amide I band of the protein bilayer consisting of protein A/G and IgG), the antennas on the nanopedestals provide ~1.5 times larger field enhancements. More importantly for the nanoantennas on the substrate, significant portion of the local fields is concentrated within the silicon nitride substrate, thus not accessible for the target biomolecules. In contrast, for the nanoantennas on nanopedestal realized through our isotropic etching technique, the tip ends of the antennas and hence the full mode volume of the hot spots are now available for biodetection (see the illustration in Figure 1d).
Figure 1. (a) Schematic illustration and (b) SEM image of the gold nanoring antenna on the silicon nitride nanopedestal. (c) Electric field intensity profile for the antenna systems on (top) nanopedestal and (bottom) substrate. (d) Schematic cross-sectional profile illustrating the availability of E-field hot spots for biomolecular bindings. For Amide I band, the antenna on the pedestal is designed for larger dimensions (Rin = 750 nm and Rout = 950 nm) compared to the one on the substrate (Rin = 670 nm and Rout = 870 nm). We optimize the silicon nitride thickness and the antenna dimensions to provide a robust pedestal (height = 70 nm). The pedestal is formed on a silicon nitride membrane. After formation of the pedestals by isotropic etching of silicon nitride with hydrofluoric acid, the remaining membrane thickness is 260 nm. Thickness of the gold antenna is 100 nm.
For calculating the absorption signals using these antennas, we numerically model a bilayer molecule consisting of protein A/G and Immunoglobulin (IgG) antibody. The real (black curve) and imaginary (green curve) parts of the modeled protein permittivity is shown in Figure 2a. Our surface chemistry protocols ensure that the protein bilayer only covers the surface of the gold antennas. This protein coverage is modeled by defining an 8 nm thick dielectric layer as illustrated in Figure 2b. Figure 2c shows the response of the ring antenna on the nanopedestal before (black curve) and after (green curve) the protein binding. The vibrational signatures are clearly observed as spectral dips in the reflection response due to the Amide I and II absorption bands. Here, in addition to the spectral dips due to the protein absorption resulting from the imaginary part of the complex protein permittivity, we also observe a red-shift
FTu4E.3.pdf
CLEO:2015 © OSA 2015
in the antenna response which originates from the array sensitivity to the real part of the protein permittivity. In order to take this shift into account, we replace the initial response (black curve) with the reflection spectrum from the antenna array covered with a reference material with the real part of the refractive index of the protein bilayer and zero absorption (no imaginary part). This response (black dashed curve) overlaps with the response of the antennas covered with the protein bilayer model (green curve), important for the absorption signal calculation. Figure 2d shows the reflection difference spectrum calculated from the antennas engineered for Amide I vibrational band, for the antenna on the silicon nitride nanopedestal (green curve) and on the silicon nitride substrate (dashed green curve). The antenna arrays on the nanopedestal yield a reflection difference as large as 0.06254. This value corresponds to an additional enhancement in the reflection difference as large as 2.6 times compared to the ring antennas fabricated directly on the dielectric substrate (0.02397).
Figure 2. (a) Real (black curve) and imaginary (green curve) parts of the modeled permittivity of the protein bilayer. (b) Schematic illustration of the protein bilayer with a thickness of ∼8 nm on the surface of the gold antenna. In the figure, the purple color representing the protein bilayer shows the sensing volume and the black line shows the sensing surface. (c) Calculated bare reflection spectrum of the nanoring antenna on the nanopedestal with no protein layer (black curve) and after the coating of the protein bilayer with real (black dashed curve) and complex (green curve) permittivity. (d) Calculated reflection difference spectra for the nanoring antennas on the silicon nitride nanopedestal (green curve) and directly on the substrate (dashed green curve). In the figures, red and blue vertical lines show the Amide I and II vibrational modes, respectively.
The enhancement in the absorption signal with the use of the nanopedestal can be explained by considering the improved sensing volume and the large local electromagnetic fields associated with it. For a quantitative comparison, we calculate the integral of the local electromagnetic fields over the protein bilayer region. We calculate the sensing surface, S as schematically shown in Figure 2b. The ring antennas on the nanopedestal support ∼1.7 times larger sensing surface compared to the ones on the substrate. Next, we calculate the integral of the enhanced fields over the sensing volume (V) by considering the 8 nm thick bilayer on the gold surface, i.e., V(nm3) = S(nm2) × 8 nm. For Amide I band, the nanopedestal design enables ∼2.6 times larger integral value, which is in good agreement with the previously presented enhancement in the reflection difference (∼2.6 times in Figure 2d). We experimentally demonstrate the advantage of the nanopedestals through SEIRA measurements. Figure 3a and 3b show the experimental spectral response of the nanoring antennas on the silicon nitride substrate and on nanopedestals, respectively. Figure 3c shows the initial response of the nanoring antenna on the silicon nitride pedestal (black curve). After introducing the protein bilayer, in addition to the spectral dips corresponding to the absorption (green curve), we also observe a red-shift in plasmonic resonance, as explained earlier. In order to consider this shift, we utilize a polynomial fitting procedure and obtain a shifted spectrum (black dashed curve) which nicely overlaps with the response corresponding to the protein bilayer. Figure 3d shows the experimental reflection difference spectra of the ring antennas on the silicon nitride nanopedestal (green curve) and substrate (dashed green curve). Interestingly, even though the antenna system on the substrate favors the plasmonic excitations more (by covering Amide I and II better), our antenna system still supports larger absorption signal due to its highly accessible large local electromagnetic fields. We determine that the antennas on the nanopedestals give 30 mOD and 18 mOD absorption signals at the Amide I and II bands, respectively, while the antennas on substrate provide only 3 mOD for Amide I and 7 mOD for Amide II.
Figure 3. Experimental reflection spectra for the gold nanoring antennas on the silicon nitride (a) substrate and (b) nanopedestals. (c) Experimental reflection spectra of the nanoring antenna on the nanopedestal before (black curve) and after (green curve) the coating of the protein bilayer. Dashed black curve shows the antenna response after the correction using the polynomial fit in order to compensate the refractive index shift. (d) Experimental reflection difference spectra for the nanoring antennas on the silicon nitride nanopedestal (green curve) and substrate (dashed green curve) demonstrating the protein vibrational signatures. [1] Arif E. Cetin, Dordaneh Etezadi, and Hatice Altug, “Accessible Nearfields by Nanoantennas on Nanopedestals for Ultrasensitive Vibrational Spectroscopy”, Advanced Optical Materials, 2, 866 (2014).
