The local electric field distribution and the effect of surface-enhanced Raman spectroscopy (SERS) were investigated on the quasi-3D (Q3D) plasmonic nanostructures formed by gold nanohole and nanodisc array layers physically separated by a dielectric medium. The local electric fields at the top gold nanoholes and bottom gold nanodiscs as a function of the dielectric medium, substrate, and depth of Q3D plasmonic nanostructures upon the irradiation of a 785 nm laser were calculated using the three-dimensional finite-difference time-domain (3D-FDTD) method. The intensity of the maximum local electric fields was shown to oscillate with the depth and the stronger local electric fields occurring at the top or bottom gold layer strongly depend on the dielectric medium, substrate, and depth of the nanostructure. This phenomenon was determined to be related to the Fabry-Pérot interference effect and the interaction of localized surface plasmons (LSPs). The enhancement factors (EFs) of SERS obtained from the 3D-FDTD simulations were compared to those calculated from the SERS experiments conducted on the Q3D plasmonic nanostructures fabricated on silicon and ITO coated glass substrates with different depths. The same trend was obtained from both methods. The capabilities of tuning not only the intensity but also the location of the maximum local electric fields by varying the depth, dielectric medium, and substrate make Q3D plasmonic nanostructures well suited for highly sensitive and reproducible SERS detection and analysis.
© 2011 OSA
Quasi-3D (Q3D) plasmonic nanostructures are defined as those having physically separated metallic thin film with square arrays of nanoholes on top and metallic nanodiscs at the bottom of each well mediated by a dielectric spacer . Localized surface plasmons (LSPs) can be excited at both 2D nanoholes [2, 3] and 0D nanodots . By engineering different plasmonic elements in a single Q3D nanostructure, additional freedoms can be achieved in tuning plasmonic properties. The coupling of localized surface plasmon resonance (LSPR) at the top gold nanoholes and the bottom gold nanodiscs resulted in new and complex plasmonic modes that have been used for multispectral biosensing and 1D imaging [1, 5]. In addition, the upper surface of the gold nanodiscs at the bottom and the bottom surfaces of the nanoholes on the top create a Fabry-Pérot (FP) resonant nanocavity with strong field confinement. It was demonstrated that the shift of the FP resonance in response to the bulk refractive index change is even more sensitive than SP modes, which makes the Q3D plasmonic nanostructures good candidates for enhanced biosensing . Our group reported that Q3D plasmonic nanostructures have strong enhancement to Raman scattering and exhibit inverted diameter dependence comparing to gold nanodot arrays . By simply varying the diameter of the nanoholes, not only the intensity but also the location of strong electric fields can be varied for optimal surface-enhanced Raman spectroscopy (SERS) detection of small molecules and large microorganisms . The transmittance of light can also be manipulated by varying the diameter and depth , indicating that Q3D plasmonic nanostructures could also be applied for enhancing light harvesting in photovoltaic devices. Furthermore, the large dimension of nanoholes makes it possible to fabricate Q3D plasmonic nanostructures on a large area using low-cost and high-throughput methods such as soft lithography [1, 10].
It is well known that SPR is very sensitive to the dielectric medium at the surface of metal thin film and SPR biosensors are based on the tracking of SPR wavelength shift in response to the refractive index change due to the binding of biomolecules . Similarly, LSPR associated with the metallic nanoparticles or nanostructures are very sensitive to the surrounding dielectric medium . Q3D plasmonic nanostructures can be conveniently fabricated by electron beam lithography (EBL) [7–9], focused ion beam (FIB) , or soft-lithography [5, 10]. Each method offers the possibility to include different dielectric spacer and substrate to make Q3D plasmonic nanostructures. For example, Q3D plasmonic nanostructures can be made by EBL in the heterogeneous dielectric media like polymethyl methacrylate (PMMA) on silicon or ITO coated glass. They can also be made via soft lithography or FIB in a homogeneous dielectric medium, such as polydimethylsiloxane (PDMS), polyurethane (PU), silicon, or glass. The different optical properties, including refractive index and extinction and absorption coefficients, of the dielectric spacer and supporting substrate could dramatically affect the LSPR and thus local electric fields [13, 14]. In addition, the depth of Q3D plasmonic nanostructures, controlled by the thickness of the dielectric spacer, also plays a significant role in manipulating the LSPR and local electric fields. The influence of the depth on the transmittance, FP cavity resonance and SERS has been demonstrated [6, 9].
In this study, we systematically investigated the effects of dielectric spacer, substrate, and depth of Q3D plasmonic nanostructures on the local electric field distribution and SERS using three-dimensional finite-difference time-domain (3D-FDTD) simulations and experimental SERS measurements. Q3D plasmonic nanostructures in heterogeneous dielectric media, such as PMMA on silicon, ITO, glass, and ITO coated glass, as well as in homogeneous dielectric medium, such as PDMS, PU, and silicon, were studied. The depth was varied from 100 to 1200 nm while the diameter and spacing of nanoholes were maintained at 400 and 100 nm, respectively. To verify the trend of electric fields and SERS effect as a function of depth and dielectric medium predicted by 3D-FDTD simulations, SERS spectra were taken for 4-mercaptopyridine (4-MP) molecules adsorbed on the Q3D plasmonic nanostructures fabricated on silicon and ITO coated glass substrates via EBL. The enhancement factors (EFs) obtained from simulations and experiments were compared.
2.1. 3D-FDTD calculations
The 3D-FDTD method  was used to calculate the electric field distribution of Q3D plasmonic nanostructures. The models were conducted using a single hole as the unit cell, with the periodic boundary conditions applied to the x- and y-directions to describe an infinite square array and uniaxial perfectly matched layers along the z-direction. The z-boundaries were placed 400 nm above the top gold layer and 300 nm below the bottom gold layer. The computational domain consists of a square of D/Δx × D/Δy grid points centered on the nanohole (D is the grating periodicity in nm) in the x-y plane with a resolution of Δx = Δy = 2 nm. A non-uniform grid size was used along the z-direction for the overall simulation volume, which varies with the refractive index of each material, having minimum and maximum sizes set at 0.5 and 5 nm, respectively. In order to best capture the electric field in the vicinity of gold-dielectric material interfaces, a small region of 10 nm above and below the gold slabs, extending over the entire x- and y-span of the simulation boundary, was set with a grid size of 0.5 nm in the z-direction. Test calculations with smaller grid sizes indicated that the results converged. The source of the input plane wave pulse was placed 400 nm above the top gold layer and polarized to the x-axis. To obtain the electric field distribution along the z-direction, the x-y plane monitors were placed between the interfaces of the top gold-air and the bottom gold-substrate with an interval of 2 nm. Additional two monitors were placed at the locations that are 2 and 10 nm above the top gold-air interface. The monitors for the y-z and x-z planes were placed crossing the center of the nanohole. The simulation time was set at 150 fs for the input pulse duration of 14.70 fs, ensuring the fields to decay completely before termination of the simulation. The reported electric field intensity was normalized to the intensity of the incident wave, i.e., 1 W/m2. The refractive indices of the materials used in the simulations were taken from Ref. 16. The electric field distribution was reported as the square of the electric field intensity (|E|2).
2.2. SERS substrate preparation
Q3D plasmonic nanostructures were fabricated via EBL using an FEI Sirion scanning electron microscope (SEM) with Nabity NPGS software. A layer of PMMA electron-sensitive resist was spin-coated on a silicon (100) wafer or an ITO coated (ITO thickness ~140 nm) glass substrate. Using different PMMA concentrations (2-6% in anisole) or spin rates (1500-4000 rpm), the PMMA thickness was controlled between 300 and 800 nm. The PMMA coated substrate was exposed to a ~20 pA electron-beam with a line dose varying from 0.34 to 0.40 nC/cm, depending on the thickness of the PMMA layer, to create 50 μm x 50 μm nanohole arrays with a fixed diameter of 400 nm and spacing of 100 nm. Holes were generated after development in 1:3 methyl isobutyl ketone/isopropanol (MIBK/IPA) PMMA developer for 70 seconds followed by an IPA rinse and a post-bake at 95°C for 30 minutes. The Q3D plasmonic nanostructures were obtained by evaporating a 50 nm thick gold film onto the patterned substrate. The lateral and vertical dimensions of nanostructures were characterized by FEI Sirion SEM and tapping mode AFM using a Vecco Dimension 3100 AFM equipped with a Nanoscope IVa controller.
2.3. SERS spectra collection and processing
SERS substrates with Q3D plasmonic nanostructures were cleaned in a UV ozone cleaner for 20 minutes, rinsed with 18.2 MΩ·cm deionized (DI) water, dried with a stream of air, and immersed in a 3 mM 4-MP aqueous solution for 3 hours to ensure the formation of a self-assembled monolayer. Raman spectroscopy was taken on a Renishaw InVia Raman spectrometer attached to a Leica DMLM upright microscope. A 50 × (N.A. = 0.8) objective was used to focus a 785 nm laser on the nanohole arrays to form an approximately 2 μm x 25 μm rectangular spot and collect the 180° scattering light from the sample surface. Spectral resolution of 1.1 cm−1 can be achieved and the spectra ranging from 400 to 2000 cm−1 were collected. All SERS spectra were corrected using linear-baseline between 960 cm−1 and 1170 cm−1 for calculating the peak intensity in order to calculate the enhancement factor (EF) .
2.4. Enhancement factor calculation
The EF was calculated using two different methods. One is using the relationship of the fourth power of electric field, i.e., EF = |Etop/E0|4 + |Ebottom/E0|4, where E0 is the incident field intensity and Etop and Ebottom are the maximum local electric fields at the gold/air interfaces of top holes and bottom discs, which were obtained from the 3D-FDTD simulations. It has to be pointed out that EF calculated in this method counts only the electromagnetic enhancement of Raman scattering, which is usually 2-3 orders of magnitude lower than that obtained from experiments due to not including the chemical enhancement. The other method is using EF = IsurfNbulk/IbulkNsurf, where Isurf and Ibulk are the intensities of the band at 1094 cm−1 for 4-MP molecules adsorbed on the Q3D nanostructure arrays and 4-MP powder, respectively; Nsurf and Nbulk are the number of molecules on the surface of the nanostructures and in the bulk of powder illuminated by the laser, respectively. The SERS active surface area of Q3D nanostructures is calculated as the sum of three contributions. The first two contributions consist of the area bordering the circular belt along the top edge of the nanoholes, one part extending out from the top edge of the hole in the x-y plane, and the other extending down into the hole in the z-direction, both with a width of 2 nm. The third contribution is the area of the circular belt around the edge of the bottom disc with a width of 2 nm on the x-y plane. FDTD calculations show that the electric fields decrease dramatically beyond 2 nm from the edges of gold nanoholes and nanodiscs. Therefore, the width of 2 nm was used in the EF calculations for the SERS active surface area .
3. Results and Discussion
Q3D plasmonic nanostructures with different dielectric spacer and supporting substructures investigated in this work are shown in Fig. 1 in a side-view. All the arrays were maintained the same diameter (400 nm) and spacing (100 nm) while the depth was varied from 100 to 1200 nm (Fig. 1 e)). These nanostructure arrays can be easily fabricated via either EBL (Fig. 1 a) and b)) or soft-lithography (Fig. 1 c)) or dry etching (Fig. 1 d)). The Q3D plasmonic nanostructures were realized by thermally evaporating 50 nm gold on the nanofabricated dielectric nanostructures. The Q3D plasmonic nanostructures shown in Fig. 1 a) and b) could be considered as made in heterogeneous dielectric media because of the existence of an optical interface between the dielectric spacer (PMMA) and the supporting substrate. In contrast, the nanostructures shown in Fig. 1 c) and d) could be considered as made in a homogeneous dielectric medium because both the dielectric spacer and the supporting substrate are the same dielectric material. The optical properties at 785 nm wavelength, including refractive index and extinction and absorption coefficients, of all materials involved in this work are listed in Table 1 .
3.1. Fabry-Pérot interference in Q3D plasmonic nanostructures
In order to understand the effect of the depth of the Q3D plasmonic nanostructure on the optical behavior of the nanostructure, we first neglected the subwavelength dielectric structures and approximated the Q3D nanostructure by two reflecting homogenous planes representing the hole and disc arrays, surrounded by a homogeneous dielectric (inset of Fig. 2 a )). This makes it possible to treat the Q3D nanostructure as a multilayer structure exhibiting the Fabry-Pérot thin-film interference. In the thin-film interference model, the top and bottom planes are characterized by the complex amplitude coefficients for reflection , and transmission , respectively. The reflection and transmission coefficients r, t for the whole structure are:
The interference of waves propagating back and forth between the two planes can be either constructive or destructive, which affects the electromagnetic fields the nanoholes and nanodiscs are subjected to. Stronger exciting fields cause stronger plasmonic oscillations and produce stronger fields at the metal surface. The complex exponential term in the denominator of Eq. (1) describes oscillations of r and t with the depth of the structure. Similar equations can be derived for the fields exciting the holes and discs. The period of the oscillations, p, with depth is
The reflection coefficients , and transmission coefficients , of the free-standing hole and disc layers can be obtained numerically by 3D-FDTD simulations containing only the hole layer or only the disc layer. Using these coefficients in Eq. (1), one can obtain R and T spectra (Fig. 2 a)). In Fig. 2 a), the results are compared with the 3D-FDTD simulation of the structure containing both the disc and the hole layers spaced at 800 nm. The simulation shows that for the depth of 800 nm, the Fabry-Pérot thin-film interference model is an accurate representation of the whole structure. Figure 2 b) shows the spatial distribution of the fields exciting the nanoholes and nanodiscs, clearly displaying the oscillations.
The magnitude of the coefficients , , , and resulting r, t varies widely in the range of almost 0 to 1 (Fig. 2 a)) for wavelengths between 700 and 1200 nm due to variation in permittivity of gold and due to the excitation of LSPs around λ = 780 nm. The changing values of these coefficients influence the depth of the oscillations of reflectance R, transmittance T and the LSP fields in the whole structure. The presence of a dielectric spacer between the planes or a substrate can also change holes’ and discs’ reflection and transmission coefficients.
3.2. Q3D plasmonic nanostructures built in heterogeneous dielectric media
We performed 3D-FDTD simulations of the whole Q3D plasmonic nanostructures to calculate the local electric field distribution upon the irradiation of a 785 nm laser. Q3D plasmonic nanostructures built in heterogeneous dielectric media, PMMA/silicon and PMMA/pure ITO were investigated. The maximum local electric field intensity at the air/gold interfaces of the bottom gold nanodiscs Ebottom and the top gold nanoholes Etop were collected and the fourth power of the intensities was plotted as a function of the depth separating the top gold nanoholes and the bottom gold nanodiscs from 100 to 1200 nm (Fig. 3 , blue: Etop, red: Ebottom, green: Etop + Ebottom). Interestingly, Etop and Ebottom vary periodically in opposite phase with the depth of Q3D. As shown in Fig. 3 a) and b), for the Q3D on PMMA/silicon and PMMA/pure ITO substrates, the intensity and the periodicity of Etop and Ebottom vary in a similar pattern except that |Ebottom|4 is about one order of magnitude lower for Q3D on PMMA/silicon than that on PMMA/ ITO. The Etop are always larger than the Ebottom when the depth is beyond ~300 nm except at several points where both have similar intensities. Only when the depth is larger than ~300 nm, the Ebottom is stronger than the Etop. However, when the substrate was changed to pure glass, Fig. 3 c) shows that the |Ebottom|4 was increased dramatically (by about two orders of magnitudes) compared to those on the silicon or ITO substrate. The Etop remained similar to those on the silicon or ITO substrate except the lowest electric fields are about two orders of magnitudes lower than those on the silicon or ITO substrate. ITO coated glass substrates are commonly used in EBL as they are conductive and transparent. Figure 3 d) shows that for the Q3D made on ITO coated glass substrate, strong oscillation of Etop and Ebottom was observed when the depth is shallower than ~600 nm. The ratio of intensities Etop and Ebottom is reversed around the depth of ~300 nm. However, when the depth is deeper than ~600 nm, there are only weak oscillations of Etop and Ebottom and the intensities are similar.
The periodicity of the oscillations shown in Fig. 3 is about 315 nm. This is in good agreement with Eq. (2), when using the volume average of refractive index n = 1.25 (volume fraction of PMMA for 400 nm hole diameter and 100 nm spacing is 0.50 and its refractive index at 785 nm is 1.49). For the depths of Q3D nanostructures less than ~300 nm, the depth- dependence departs from the periodic behavior displayed by the thin-film interference model across all depths (Fig. 2 b)). This can be attributed to the hole and disc arrays interacting also through their near-fields in addition to far-field interaction described by the model. Apparently, the depth of oscillations is different for different substrate materials. This difference is caused by different reflection coefficient of the disc layer when fabricated onsubstrates of a different refractive index. A lower reflectivity results in lower depth of the oscillation, as can be seen by analyzing the denominator of Eq. (1).
To further study the local electric field distribution varying along the depth of Q3D plasmonic nanostructures, the electric fields at the x-z plane and the x-y plane of top and bottom gold/air interfaces in four typical depths are plotted in Fig. 4 for Q3D on silicon and ITO coated glass substrates. With the depth of 200 nm, it is clearly shown that the electric fields of the bottom gold/air interfaces are stronger than those at the top for both silicon and ITO coated glass substrates. The electric field displays a strong dipole mode on the silicon substrate while a strong quadruple on the ITO coated glass substrates. As the depth is increased to 300 nm, both the top and bottom gold/air interfaces have similar electric fields. Just increasing the depth by 60 nm, makes the electric fields at the top gold/air interfaces much stronger than those at the bottom gold/air interfaces. Further increase in the depth to 600 nm, makes the electric fields at both the top and bottom gold/air interfaces similar again. However, the electric fields on the ITO coated glass substrate are slightly stronger than those on the silicon substrate as shown in Fig. 4 and Fig. 3 a) and d).
3.3. Quasi-3D plasmonic nanostructures built in homogeneous dielectric media
The Q3D plasmonic nanostructures shown in Fig. 1 c) can be easily fabricated via soft-lithography by using either PDMS  or PU  for large areas with low-cost and high-throughput. PDMS and PU are optically clear, and, in general, are considered to be inert, which make them good candidates for biosensor applications The main difference between the Q3D plasmonic nanostructures shown in Fig. 1 c) and Fig. 1 a) and b) is the lack of optical interface between the dielectric spacer and the supporting substrate for the one shown in Fig. 1 c). The Q3D plasmonic nanostructures are completely supported by the same dielectric material. Similar to the Q3D plasmonic nanostructures built in the heterogeneous dielectric medium, Fig. 5 a ) and b) also show the oscillation of Etop and Ebottom. From the trend observed in the heterogeneous dielectric media aforementioned, stronger electric fields at the bottom gold/air interfaces would be observed. Indeed, as shown in Fig. 5 a) and b), |Ebottom|4 are about one to two orders of magnitude higher than |Etop|4. As the refractive index is increased from 1.4 of PDMS to 1.8 of PU, the differences between |Etop|4 and |Ebottom|4 are smaller. Moreover, |Ebottom|4 of Q3D plasmonic nanostructures made in PU are about one order of magnitude lower than those made in PDMS. Therefore, the Q3D plasmonic nanostructures made in homogeneous dielectric medium with small refractive index tend to have stronger electric fields at the bottom gold/air interfaces.
The periodicity of oscillations shown in Fig. 5 is about 320 nm for homogeneous PDMS and about 260 nm for homogeneous PU. This agrees well with the values calculated from Eq. (2), when using the volume average of refractive index n = 1.2 and 1.4, respectively (volume fraction of PDMS and PU is 0.50 and their refractive indices at 785 nm are 1.4 and 1.8, respectively).
The Q3D plasmonic nanostructures could also be made out of silicon via dry etching as shown in Fig. 1 d). In this way the Q3D plasmonic nanostructures are completely supported by Si. These plasmonic nanostructures would have high chemical and physical stability. In addition, silicon and gold surfaces can be functionalized separately via silane and thiol chemistries for the detection of specific binding analytes . Despite of the lack of the optical interface as shown in Fig. 1 c), the behavior of electric fields varying with the depth of Q3D plasmonic nanostructures is quite different from those in PDMS and PU. The electric fields at the top gold/air interfaces do not show obvious regular oscillations with the depth as shown in Fig. 5 c). Instead, |Ebottom|4 decreases as the depth is increased. Such behavior could be attributed to the high refractive index and non-zero absorption coefficient of silicon. As a semiconductor, silicon exhibits different optical properties compared to polymer dielectric media like PMMA, PDMS, and PU. The absorption and extinction coefficients of Si at 785 nm are 0.11 µm−1 and 0.0055 . Clearly, the attenuation prevents the reflected wave from interfering strongly with the incident wave and from forming the oscillations in Etop and Ebottom.
3.4. Comparison of experimental and FDTD simulated EFs as a function of depth
One of the exciting applications of Q3D plasmonic nanostructures is to be used as SERS-active substrates with high reproducibility and wide tunability . The EF due to the electromagnetic enhancement at the top gold nanoholes and the bottom gold nanodiscs can be estimated from the relationship of EFtop = |Etop/E0|4 and EFbottom = |Ebottom/E0|4, where E0 is the incident field intensity . The oscillations of |Etop|4 and |Ebottom|4 with the depth of Q3D plasmonic nanostructures as well as the dependence on spacer and substrate material indicate that these parameters can be used to tailor the Q3D plasmonic nanostructures with different SERS effects. To verify the trends of SERS EFs predicted by the 3D-FDTD simulations, Q3D plasmonic nanostructures were fabricated on Si and ITO coated glass substrates with the same dimensions used in the 3D-FDTD simulations. As shown in Fig. 4, the maximum local electric fields, i.e., the “hot” spots of SERS, in the x-y plane are confined within ~2 nm at the circumference of nanoholes and nanodiscs. Therefore, the experimental EFs were calculated by counting only the area of “hot” spots, which is about 0.35% of the total gold surface area. In order to obtain the accurate experimental EFs, it is desired that analyte molecules cover the full gold surface like 4-MP forming a self-assembled monolayer in this work. For chemical and biological sensing applications, when the analyte concentration is extreme low (e.g., parts per billion (ppb) or parts per trillion (ppt)), only the analytes adsorbed on the “hot” spots exhibit Raman response and can be detected. This phenomenon has been reported in the detection of single molecules using SERS [19, 20]. Time-lapped Raman spectra should be collected in order to obtain reliable detections.
The SERS spectra of 4-MP molecules adsorbed on Q3D plasmonic nanostructures were collected and the EFs were calculated by using the intensity of the 1094 cm−1 band. Figure 6 shows the experimental EFs obtained from the Q3D plasmonic nanostructures with different depth on silicon and ITO coated glass substrates. Clearly, the experimental EFs vary with the depth on both silicon and ITO coated glass substrates. To directly compare the EFs estimated from the 3D-FDTD simulations and calculated from the SERS experiments, the values of |Etop/E0|4 + |Ebottom/E0|4 from the same depth as those in experiments were also plotted in Fig. 6. Using the summation is because 4-MP molecules can access the entire gold surface of the Q3D plasmonic nanostructures and the EFs calculated from the SERS experiments include the signals from 4-MP molecules on both top and bottom gold surfaces . Both the EFs estimated from the electromagnetic simulations and the SERS experiments follow the sametrend with respect to the depth as well as material of the substrate. For example, the 3D-FDTD simulations predict that the maximum SERS response could occur with the depth of ~370 nm on a silicon substrate and the enhancement could be even stronger than the maximum SERS response that could occur on an ITO coated glass substrate with a similar depth (Fig. 3 a) and d)). Indeed, the EFs calculated from the SERS experiments verify the prediction. In addition, the experimental EFs show smaller values with the depth of ~300 and ~580 nm on the silicon substrates, which follow the same trends as shown in the plot of the fourth power of total electric fields vs. the depth in Fig. 3 a) (green line). It was noticed that the EFs estimated from the 3D-FDTD simulations were about two to three orders of magnitude lower than those obtained from SERS experiments. This difference is attributed to the chemical enhancement which is not accounted for in the 3D-FDTD simulations. The chemical enhancement is a resonance-like enhancement between the local electric field and the vibrational modes of 4-MP molecules .
Figure 7 shows the SERS spectra of 4-MP adsorbed on the Q3D plasmonic nanostructures fabricated on Si and ITO coated glass substrates with two different depths. Since all the dimensions are the same, the substrate and the depth effects can be directly compared. Figure 7 a) suggests that with the same depth of 300 nm, the intensity of SERS spectrum of 4-MP on the array on the silicon substrate is weaker than that on the ITO coated glass substrate. By increasing the depth to 370 nm, the trend is reversed. The intensity of the 4-MP SERS spectrum on the Q3D plasmonic nanostructures on the silicon substrate is much stronger than that on the ITO coated glass substrate. In addition, the intensity of SERS spectra of 4-MP on the arrays on both silicon and ITO coated glass substrates are enhanced by a factor of 8 and 1.2, respectively. Obviously, both the nanostructure depth and substrate material can dramatically affect the SERS performance of Q3D plasmonic nanostructures.
The effects of depth, dielectric medium and supporting substrate of Q3D plasmonic nanostructures on the local electric field distribution and SERS were systematically investigated. 3D-FDTD simulations suggest that with respect to the local electric field distribution and the resulting SERS effect, the nanofabricated Q3D plasmonic nanostructures can be split into two groups.
The first type is comprised of the Q3D plasmonic nanostructures fabricated on highly reflective substrates such as silicon, pure ITO, or ITO coated glass. The strong electric fields can be generated at either top gold nanoholes or bottom gold nanodiscs by varying the depth of the nanostructure. At some depths, the similar intensity of electric fields on top and bottom gold layers can be obtained. The change of the refractive index from 3.69 of silicon to 1.78 of pure ITO does not cause significant variations in the total electric fields. However, when the Q3D plasmonic nanostructures are fabricated on ITO coated glass, the oscillation of the maximum local electric fields is damped with the depth beyond ~600 nm and the intensity of both top and bottom gold layers are similar. This type of Q3D plasmonic nanostructures can be used for optimal SERS detection of large microorganisms by creating the strong electric fields at the top gold nanoholes . Silicon substrates provide larger variability and higher sensitivity than the ITO coated glass substrates.
The second type is comprised of the Q3D plasmonic nanostructures fabricated in the homogeneous dielectric medium such as PDMS, PU, and silicon. The stronger electric fields are always located at the bottom gold nanodiscs for all depth investigated. As the refractive index is decreased, the intensity of the maximum local electric fields increase and the difference between the top and bottom gold layers is larger. It is predicted that the EFs of Q3D plasmonic nanostructures made in PDMS could be as high as ~109, one order of magnitude higher than the highest one fabricated on a silicon substrate. The Q3D plasmonic nanostructures made on low reflective substrates such as glass can be assigned to this type because at most depths the stronger electric fields are located at the bottom gold nanodiscs. Even higher EFs of approximately 109 to 1010 could be obtained by fabricating Q3D plasmonic nanostructures on a glass substrate. Q3D nanostructures fabricated in homogeneous PDMS or on a glass substrate could provide very sensitive SERS detections especially for the analytes with the size smaller than subwavelength nanoholes, such as toxins, proteins, and viruses. The behavior of Q3D plasmonic nanostructures fabricated in homogeneous silicon is different from PDMS and PU due to the high refractive index and extinction at 785 nm. The maximum local electric fields decay with the depth for the bottom gold nanodiscs and exhibit lower intensity, lack of oscillation for the top gold nanoholes. Similar EFs as the Q3D plasmonic nanostructures made in heterogeneous dielectric media on silicon substrates could be obtained when the depth is shallower than 400 nm.
The theoretical prediction, that by varying the depth or choosing different substrates SERS EFs can be controlled, was verified by measuring SERS spectra on Q3D plasmonic nanostructures of different depths fabricated on silicon and ITO coated glass. The capability of tuning not only the intensity but also the location of the maximum local electric fields by varying the depth of the nanostructure, dielectric medium, and substrate materials make Q3D plasmonic nanostructures suitable for highly sensitive and reproducible SERS substrates for chemical and biological sensing.
This work was supported in part by the University of Washington (UW) faculty start-up funds and the NOAA Ocean and Human Health Initiative (OHHI) funds. J.J.X. acknowledges a fellowship from the China Scholarship Council. R.W.J acknowledges the NNIN Research Experience for Undergraduates program supported by the NSF. Nanofabrication and SERS studies were performed at the Nanotech User Facility, the UW site of the National Nanotechnology Infrastructure Network (NNIN) supported by the NSF. This research was also supported by research grant KAN200670701 and by Praemium Academiae of the Academy of Sciences of the Czech Republic.
References and links
1. M. E. Stewart, N. H. Mack, V. Malyarchuk, J. A. N. T. Soares, T. W. Lee, S. K. Gray, R. G. Nuzzo, and J. A. Rogers, “Quantitative multispectral biosensing and 1D imaging using quasi-3D plasmonic crystals,” Proc. Natl. Acad. Sci. U.S.A. 103(46), 17143–17148 (2006). [CrossRef] [PubMed]
2. T. W. Ebbesen, H. J. Lezec, H. F. Ghaemi, T. Thio, and P. A. Wolff, “Extraordinary optical transmission through sub-wavelength hole arrays,” Nature 391(6668), 667–669 (1998). [CrossRef]
3. F. J. Garcia-Vidal, L. Martin-Moreno, T. W. Ebbesen, and L. Kuipers, “Light passing through subwavelength apertures,” Rev. Mod. Phys. 82(1), 729–787 (2010). [CrossRef]
5. J. Yao, A. P. Le, S. K. Gray, J. S. Moore, J. A. Rogers, and R. G. Nuzzo, “Functional nanostructured plasmonic materials,” Adv. Mater. (Deerfield Beach Fla.) 22(10), 1102–1110 (2010). [CrossRef] [PubMed]
6. A. Artar, A. A. Yanik, and H. Altug, “Fabry-pérot nanocavities in multilayered plasmonic crystals for enhanced biosensing,” Appl. Phys. Lett. 95(5), 051105 (2009). [CrossRef]
7. Q. M. Yu, P. Guan, D. Qin, G. Golden, and P. M. Wallace, “Inverted size-dependence of surface-enhanced Raman scattering on gold nanohole and nanodisk arrays,” Nano Lett. 8(7), 1923–1928 (2008). [CrossRef] [PubMed]
8. J. J. Xu, L. Zhang, H. Gong, J. Homola, and Q. M. Yu, “Tailoring Plasmonic Nanostructures for Optimal SERS Sensing of Small Molecules and Large Microorganisms,” Small 7(3), 371–376 (2011). [CrossRef] [PubMed]
9. J. J. Xu, P. Guan, P. Kvasnička, H. Gong, J. Homola, and Q. M. Yu, “Light Transmission and Surface-Enhanced Raman Scattering of Quasi-3D Plasmonic Nanostructure Arrays with Deep and Shallow Fabry-Pérot Nanocavities,” J. Phys. Chem. C 115, 10996–11002 (2011).
10. Q. M. Yu, S. Braswell, B. Christin, J. J. Xu, P. M. Wallace, H. Gong, and D. Kaminsky, “Surface-enhanced Raman scattering on gold quasi-3D nanostructure and 2D nanohole arrays,” Nanotechnology 21(35), 355301 (2010). [CrossRef] [PubMed]
13. K. S. Lee and M. A. El-Sayed, “Dependence of the Enhanced Optical Scattering Efficiency Relative to that of Absorption for Gold Metal Nanorods on Aspect Ratio, Size, End-Cap Shape, and Medium Refractive Index,” J. Phys. Chem. B 109(43), 20331–20338 (2005). [CrossRef] [PubMed]
14. A. Baptiste, A. Bulou, J. F. Bardeau, J. Nouet, A. Gibaud, K. Wen, S. Hoeppener, R. Maoz, and J. Sagiv, “Substrate-induced modulation of the Raman scattering signals from self-assembled organic nanometric films,” Langmuir 20(15), 6232–6237 (2004). [CrossRef] [PubMed]
15. Lumerical FDTD Solution, FDTD Solutions 6.5, http://www.lumerical.com/fdtd.php.
16. W. M. Haynes and D. R. Lide, Handbook of chemistry and physics (CRC Press, Boca Raton, 2003).
17. K.-P. S. Dancil, D. P. Greiner, and M. J. Sailor, “A Porous Silicon Optical Biosensor: Detection of Reversible Binding of IgG to a Protein A-Modified Surface,” J. Am. Chem. Soc. 121(34), 7925–7930 (1999). [CrossRef]
18. A. Gopinath, S. V. Boriskina, B. M. Reinhard, and L. Dal Negro, “Deterministic aperiodic arrays of metal nanoparticles for surface-enhanced Raman scattering (SERS),” Opt. Express 17(5), 3741–3753 (2009). [CrossRef] [PubMed]
20. K. Kneipp, Y. Wang, H. Kneipp, L. T. Perelman, I. Itzkan, R. R. Dasari, and M. S. Feld, “Single molecule detection using surface-enhanced Raman scattering (SERS),” Phys. Rev. Lett. 78(9), 1667–1670 (1997). [CrossRef]
21. A. Campion and P. Kambhampati, “Surface-enhanced Raman scattering,” Chem. Soc. Rev. 27(4), 241–250 (1998). [CrossRef]