Introduction

The ability to create new nanomaterials with precisely customized optical characteristics for numerous optoelectronic and nanophotonic devices based on the control of enhanced light-matter interaction has been of great interest within the scientific community in different fields. The use of such devices for biological sensing [1], nonlinear optics [2], optical sensors employing the phenomena of localized surface plasmon resonance (LSPR) [3, 4], and many others has been reported [5,6,7]. The LSPR are collective oscillations of free electrons in a metallic nanoparticle under light excitation, and they are strongly dependent on the geometric structure of the composite medium and the incident electromagnetic field. Thus, it results in a local field that is greatly enhanced compared to the incident field, and this field enhancement facilitates the observation of the Raman effect. Plasmonic sensors based on ordered arrays of nanoprisms of different materials can be optimized in terms of their geometric properties such as size, shape, and surface/volume ratio. Several techniques have been used to fabricate and tune the size, shape, and arrangement of the nanostructures, such as electron beam lithography (EBL) and focused ion beam (FIB). However, these techniques have limitations because they are quite expensive to implement, and the final samples are very small, with active areas typically less than 1 mm2. On the other hand, conventional nanosphere lithography (NSL) is an affordable method that enables the creation of 2D-ordered arrays of metallic nanostructures on a substrate surface. Monodisperse spherical silica particles are commonly employed to create hexagonally organized, closely packed monolayers of sub-micrometer-sized spheres through self-assembly. These monolayers are subsequently utilized as lithographic masks for NSL to create metallic nanostructures [2, 8,9,10,11]. Some other groups have used self-assembled masks of polystyrene microspheres for the deposition of ordered arrays of metallic posts or other tailor-made periodic surface structures [12, 13]. However, it is difficult to vary the shape and size of the nanostructures once the mask is fixed over a substrate surface. On the other hand, it is well known the use of MeV ion irradiation to modify the shape of silica colloids by the Ion Hammering effect, i.e., ion bombardment transforms spherical silica colloids into ellipsoids [14]. The silica deformation depends on the energy, fluence, and irradiation angle of the beam [14,15,16]. In this work, we take advantage of our previous experience dealing with high-energy MeV ion irradiation to modify the NSL silica particle masks as a function of the ion energy and fluence to decrease the interstice sizes. Therefore, the combined process of ion hammering with NSL allows us to generate ordered arrays of ultrafine Au nanoprisms on Si substrates. Finally, we demonstrate that the ultrafine Au nanoprisms are suitable for surface-enhanced Raman scattering (SERS) substrates for the detection of Rhodamine 6G (R6G) dye as probe molecules. 

Experimental

Sample Preparation

Spherical sub-micrometer-sized colloidal silica particles were prepared by the sol-gel technique according to a modified Stober process, referred to as LIT-synthesis (for low initial temperature of the reactants) [17]. Following this procedure, two batches of silica particles were synthesized with two different average diameters (589 nm and 612 nm) and a quite narrow standard deviation SD (18 nm and 11 nm, respectively). In order to obtain pure silica particles, a variation of the method described in the reference [18] was utilized. First, the original colloidal suspension prepared according to our LIT method [17] was placed into 2-ml Eppendorf tubes and centrifuged at 10,000 rpm. Once centrifuged, the silica particles remained collected at the bottom of the tube, and then, the supernatant was removed and replaced with 2 ml of ethanol. Then, the tubes were placed into an ultrasonic bath for 15 min to make sure that the silica particles were adequately dispersed in the ethanol. This process was repeated three times.

Once obtained, the pure colloidal SiO2 particles were deposited by spin coating onto crystalline Si wafers, previously cleaned by means of a chemical treatment known as the piranha solution [17]. Afterwards, a 10 μl volume of the solution was deposited onto Si substrates (area ≈ 1 cm2) at fixed laboratory room conditions: 60% humidity and 18.8 °C temperature. Then, the samples were allowed to dry in these conditions in order to form a high-quality silica particle monolayer on the silicon substrate.

The 589-nm diameter silica mask samples were irradiated with Si ions at 4 MeV, at fluences of 1.7 × 1015 and 2.2 × 1015 ion/cm2, whereas the 612-nm diameter sample was irradiated with the same energy, but only with 1.7 × 1015 ion/cm2. Thus, three irradiated samples were obtained, whose individual parameters are shown in Table 1. The 4 MeV Si ion irradiation was performed using the 3-MV tandem accelerator (NEC 9SDH2 Pelletron) at Instituto de Física, UNAM, at normal incidence and room temperature. The raster scanner system of the accelerator sweeps the ion beam across the samples and guarantees uniform ion beam irradiation in areas as large as 10 × 10 cm2. The irradiation time depends on the fluence, the current density, and the scanned area. In our case, for 1 cm2 samples and a current density of 500 nA/cm2, the irradiation time was 35 and 60 min for the 1.7 × 1015 and 2.2 × 1015 ion/cm2 samples, respectively.

Table 1 Experimental parameters associated with the 4 MeV Si+ ion irradiation into two different silica mask samples prepared by NSL. The average particle diameter and the standard deviation (SD) determined by SEM measurements are included 
Full size table

After the 4 MeV Si+ ion irradiation, the samples were characterized again by using SEM to determine the shape deformation of the silica particles due to the ion hammering effect. Then, Si-irradiated samples were used as tailored lithographic masks, and a 50-nm-thick Au film was deposited by e-gun thermal evaporation onto these samples, and also on a bare Si substrate for comparison purposes. The film thickness was estimated using a quartz crystal oscillator during the Au evaporation. As the lithographic silica mask must be removed, the usual methods, such as ultrasonic agitation and mechanical removal, result unfeasible in our case, because after the high-energy irradiation, the masks were strongly anchored to the substrate. Instead, the irradiated samples were subjected to a chemical attack with 5 ml of dilute aqueous hydrofluoric acid (HF, 50% in water) for 5 min. In this case, when the silica particles are removed, the gold on the top is released, remaining on the Si substrate only the Au nanostructures located in the interstitial positions. Finally, the samples were carefully washed with ultrapure Milli-Q water and ethanol, leaving the samples drying at room temperature for 24 h. Fig. 1 shows a schematic diagram of the fabrication procedure of the ordered arrays of ultrafine Au nanostructures.

Fig. 1
figure 1

Schematic diagram of the fabrication of ultrafine Au nanoprisms

Full size image

Sample Characterization

The silica mask morphology was characterized by scanning electron microscopy (SEM, JEOL JSM-7800F) before and after the Si+ irradiation to determine the deformation that silica particles suffered due to the ion hammering effect. The diameter size and the standard deviation (SD) of both the as-prepared and the ion-deformed silica particles were determined from the SEM micrographs using the ImageJ processing software [19]. No less than 5000 particles were measured for each sample from 5 or 6 SEM images (×5000 magnification).

Again, SEM images were obtained to study the 50-nm-thick Au film deposited on the as-prepared and on the ion-irradiated silica particles. After the HF attack for the monolayer removal, periodic arrays of ultrafine Au nanoprisms were obtained. The size, shape, and interparticle spacing of the Au-nanostructured features were determined from SEM and AFM micrographs. Finally, Rhodamine 6G dye (C28H31N2O3) was used as a probe molecule for Raman-SERS detection. The samples of Au-nanostructured arrays and a pure Au film were dipped into an aqueous stock solution of R6G at a concentration of 10−5 M for 24 h and then let dry at room temperature. AFM and Raman-SERS measurements were performed by a combined AFM-Raman spectroscopy with a NTEGRA Spectra instrument from NT-MDT. The laser excitation wavelength was 532 nm, and the exposure time was 10 s to record the spectrum in the single acquisition mode.

Results and Discussion

Colloidal silica particles were synthesized using the LIT-synthesis technique and deposited onto Si substrates [17]. The mask morphology was analyzed using SEM, and from Figure 2, it is evident that the silica particles exhibit a uniform spherical shape and are well-formed. The SEM data presented in Table 1 show that the average diameter for the silica particles is 589 nm and 612 nm for both series of samples, with a relatively small standard deviation of 18 nm and 11 nm, respectively. Moreover, the monodisperse silica particles form well-defined two-dimensional honeycomb arrays with regular interparticle distances.

Fig. 2
figure 2

Representative SEM images of as-deposited colloidal silica particles on Si substrates (× 10,000 magnification), together with their corresponding normal size distribution, for two different particle sizes: with mean value μ of 589 nm (left) and 612 nm (right) and a standard deviation of 18 nm and 11 nm, respectively. In addition, the coefficient of variation corresponds to less than 30% in both cases. The scale bar corresponds to 1 μm in both cases

Full size image

These lithographic masks (formed by 589 and 612 nm silica particles on Si substrates) were irradiated at room temperature with 4 MeV Si ions, at fluences of 1.7 × 1015 and 2.2 × 1015 ion/cm2, perpendicularly to the sample surface. According to the SRIM-2013 simulation code, the projected range of Si ions in SiO2/Si is about 2.8 μm at 4 MeV (see Fig. 3) [20]. These values indicate that the incoming ions go completely through the diameter of the silica particles and are finally stopped in the Si substrate. The ion hammering effect induces the shape modification of the original spherical particles into ellipsoidal oblates [2]. Thus, the direct effect of the irradiation allows the closing of the interparticle spacing, i.e., the tuning of the interstice size as a function of the irradiation fluence. In this case, due to the high quality of the ordered arrays of silica particles, the modification leads from spherical to hexagonal-shaped particles. Fig. 4 shows representative SEM images of the 589 nm silica particle array after the Si ion irradiation at two different fluences: 1.7 × 1015 and 2.2 × 1015 ion/cm2. The SEM images were analyzed with the ImageJ software to study the ion-induced deformation, which corresponds to a 30% increase in the projected area for samples 1a and 2a, and 34% for sample 1b.

Fig. 3
figure 3

SRIM-2013 simulated concentration depth profiles for Si and O atoms implanted at 4 MeV through a monolayer of silica particles over a Si substrate

Full size image
Fig. 4
figure 4

Representative SEM images of the modification of 589 nm silica particles (× 10,000) after a 4 MeV Si+ irradiation with a fluence of 1.7 1015 ion/cm2 (sample 1a, left), and 2.2 1015 ion/cm2 (sample 1b, right). The ion-induced deformation is evident as we move from spherical to hexagonal-shaped particles

Full size image

Subsequently, a 50-nm-thick Au film was evaporated onto the silica monolayer. Figure 5 shows the 589 nm as-prepared silica particles and the deformed ones, once covered by the Au film. It can be noticed that due to the interstitial regions being very small, a small amount of Au passes through them during evaporation.

Fig. 5
figure 5

Representative SEM images of the 589 nm as-prepared (left) and ion-irradiated (right) colloidal silica samples after the 50-nm-thick Au film deposition

Full size image

After the HF removal of the silica particle mask, ordered arrays of Au nanoprisms on the Si substrate can be observed. Fig. 6 shows representative SEM and AFM images of the ultrafine Au nanoprisms. By performing a detailed image analysis for samples 1a and 2a, irradiated with the lowest fluence (1.7 × 1015 ions/cm2), the results show that the Au nanoprisms exhibit a narrow average size compared to other NSL samples reported in literature [8, 21,22,23,24,25]. The ultrafine Au nanoprisms exhibit a base area A1a ≈ 3400 ± 600 nm2, with lateral size L1a ≈ 54.2 ± 8.1 nm, and average height h1a ≈ 30.0 ± 5.0 nm, and A2a ≈ 2700 ± 500 nm2, L2a ≈ 43.9 ± 6.9 nm, and h2a ≈ 25.2 ± 4.6 nm, respectively. While for sample 1b, which was irradiated with the highest fluence (2.2 × 1015 ion/cm2), the average base area was A1b ≈ 750 ± 250 nm2, with lateral size as small as L1b ≈ 26.2 ± 7.2 nm, and h1b ≈ 6.3 ± 1.4 nm (see Table 2).

Fig. 6
figure 6

SEM images at × 30,000, together with their corresponding base area size distribution (left column), and 2D and 3D AFM images of ultrafine Au nanoprisms (right column); a sample 1a; b sample 2a; and c sample 1b. The area corresponding to the ultrafine Au nanoprisms bounded with red color in the SEM images was calculated directly using the ImageJ software

Full size image
Table 2 Geometrical characteristics of ultrafine Au nanoprisms fabricated by NSL and Si ion hammering effect
Full size table

On the other hand, an important aspect to highlight is the additional deformation suffered by the Si substrate itself due to the strong anchoring of the silica particles after the 4 MeV Si ion irradiation. From the AFM images, it can be observed that after the silica particle removal, the size of the corresponding crater marks increases with the irradiation fluence. Figure 7 shows the 2D and 3D morphology of the craters with an average depth of 47.2 ± 5.0 nm. We have considered the formation of these craters after the removal of the silica microspheres from the point of view of physical processes leading to such macroscopic deformation of the Si surface. The ion hammering effect on silica microspheres is discussed in the literature [21, 26,27,28,29]. The authors usually consider the formation of tracks of cylindrical shape in a solid by the irradiation of primary ions releasing their energy in the tracks. The temperature in such tracks can reach quite high values, which, among other things, leads to thermal expansion around the track and the thermal stress caused consequently. Several studies have shown that the stress is most pronounced in the direction perpendicular to the im**ing primary ions [27,28,29,30].

Fig. 7
figure 7

Remanent craters on Si substrates due to the ion hammering effect after the removal of the silica microspheres using the HF attack

Full size image

In our case, we should first notice the very limited channels of dissipation of the energy released by the primary ion in the microsphere: the contact with the substrate is minimal, and there is practically no contact with neighboring microspheres. During the ion irradiation, the samples are kept in a vacuum chamber, and thus, it is reasonable to assume that, in our case, the temperature largely affects the shape deformation of the microspheres. As a confirmation of this hypothesis, we note that the deformation of the microsphere is proportional to the ion flux. As it is known, for primary ions with energies of few MeV, the energy loss due to interactions with electrons in a solid must be considered. In this case, the energy loss of the incoming ions due to interactions with nuclei can be neglected to a first approximation. Nevertheless, we cannot exclude the effect of the implantation of recoil atoms during the irradiation of the microspheres. This can be related to the indirect implantation of oxygen atoms belonging to the SiO2 microspheres into the Si substrate. Using the SRIM-2013 code (Monte Carlo-based method), we have estimated the distribution depth profile of the recoil atoms in the near-surface layer of the silicon substrate: the effective penetration depth of oxygen atoms in Si was about 18 nm (see Fig. 3), which is less than half as small as the experimental crater (47-nm depth) that appeared after the etching of the microspheres with HF, which is known to etch only silicon dioxide, but not silicon. We noted above that during the irradiation the microspheres can be heated and may even partially melt. Ion irradiation is carried out in a vacuum chamber with a residual pressure of 10−6 Torr for about 60 min. That is, to explain the obtained craters, in addition to the effect of the indirect implantation of recoil O atoms, we can consider the process of Si oxidation at the area of contact with the “hot” microsphere due to oxygen from the residual atmosphere. Thus, the ion hammering effect and NSL can be used to deform and to nanostructure substrates such as Si, creating surface cavities that can be useful for different applications [31].

Due to the plasmonic properties associated to the LSPR of these ordered arrays of ultrafine Au nanoprisms, together with the chemical effect of the R6G molecules, our samples were tested to be good candidates as high-sensitive SERS substrates. First, the samples were functionalized with R6G at a concentration of 10−5 M and the Raman signals were obtained using a 532-nm laser excitation. Here, it is important to underline that the Raman signals of our samples were measured and averaged from different areas, and correlated in situ with specific AFM regions (as in Fig. 6) to make sure that the R6G signal only comes from the ultrafine Au nanoprisms and not from the residual Au on the substrate. As a reference and for comparative purposes, a 50-nm-thick Au film deposited on a Si substrate was functionalized with R6G in a similar way as the nanostructured samples. Fig. 8 shows the results obtained for our nanostructured samples and the Au film reference.

Fig. 8
figure 8

Characteristic Raman-SERS spectra from the fabricated samples of ultrafine Au nanoprisms and the Au film reference, sensing R6G at a 10−5 M concentration

Full size image

One can observe that all the samples exhibit a large amplification of the R6G Raman-SERS signal due to the presence of the array of ultrafine Au nanoprisms in comparison with the use of the bare Au film. Moreover, the Raman signal shows an intensity variation due to the different values of the active area of the samples. In this case, sample 1b, with the smaller active area, shows the minimum Raman-SERS signal with respect to the other samples, but it remains clearly higher than the one corresponding to the pure Au film. To quantify the enhancement in the Raman intensity for the R6G detection using our nanostructured samples in comparison with the Au film it is necessary to use the enhancement factor (EF) [32]:

$$EF = (ISERS/Ibulk)/(Nbulk/NSERS)$$
(1)

where ISERS is the Raman intensity, and NSERS is the average number of adsorbed probe molecules in the scattering volume excited by the laser beam for the SERS experiments (ultrafine Au nanoprisms), and Ibulk and Nbulk correspond to the non-SERS sample (Au film on Si). The NSERS value can be calculated based on the surface area available for adsorption and the estimated area occupied by the probe molecule. In the case of R6G, this area corresponds to ∼ 4 nm2. The available surface area for a single ultrafine Au nanoprism is in the 250 nm2–2400 nm2 range for all the studied samples. The laser field depth is ∼ 0.38 μm and the laser illuminated spot diameter is D∼2.3 μm. We choose the Raman band at 1650 nm−1, corresponding to the aromatic C-C stretching vibrational modes, to determine the EF in the detection of R6G with a 532-nm laser excitation. This calculation is based on the fact that the SERS intensity is proportional to the number (or concentration) of molecules probed under otherwise equal conditions (same instrument parameter settings) during the same acquisition time. It is important to stress that the optical absorption of R6G is around 525 nm, and therefore, the spectrum corresponding to the 532-nm excitation was obtained after subtracting the fluorescence contribution. For our samples, the calculation results in an EF value in the 1 × 105 – 1 × 106 range.

Conclusions

In summary, the fabrication of honeycomb arrays of ultrafine Au nanoprisms on Si wafers by combining NSL with high-energy MeV Si+ ion irradiation is reported. Characterized by SEM and AFM, the samples show a narrow average size of the base area ranging from 750 to 3400 nm2, with lateral size in the 26–54-nm range, and height from 6 to 30 nm. Rhodamine 6G was chosen as the Raman-SERS probe molecule. The results show that in the case of the nanostructured substrates with the array of ultrafine Au nanoprisms, the Raman signal is highly enhanced compared to the one obtained from the Au film reference, with an EF value in the 1 × 105–1 × 106 range. Therefore, the combination of NSL with MeV ion irradiation results in a suitable nanofabrication technique enabling the ultrafine tailoring of ordered arrays of metallic nanostructures. Moreover, in the process of obtaining these ultrafine ordered structures, we found that the ion hammering effect can modify the silicon substrate to generate ordered arrays of surface cavities. These cavities can be utilized for different applications, as will be discussed in a forthcoming paper.