1. Introduction

Nanometer-sized gap between metallic structures is able to confine the electromagnetic field into subwavelength volume with much enhanced intensity [1]. It has been the basic platform for many applications in molecular electronics and nanoplasmonics [2–4], including surface-enhanced spectroscopy [5–8], nonlinear optics [9], and high-performance biosensing devices [10–12]. Since the smaller gap size gives rise to better sensitivity, as well as higher level of integration, there have been many efforts devoted to the fabrication of ultra-narrow gaps [13–22]. However, it remains to be a challenge to achieve reproducible and high-throughput fabrication of nanogaps with precisely defined gap size smaller than 5 nm.

It is well known that the conventional direct writing method such as electron beam lithography (EBL) is difficult to produce sub-10 nm gaps. Although several indirect techniques, including electromigration [23], break junction [24, 25], edge lithography [26] and angled deposition methods [27], can produce very narrow nanogaps, they are either unsuitable for the high-throughput production or restricted to limited patterning shapes. Recently a new method based on the atomic layer deposition (ALD) technique and adhesive tape assisted ‘plug-and-peel’ process was proposed [18, 28–31], which makes it possible to realize the wafer-scale production of nanogaps array with gap sizes down to 1-2 nm. However, it was pointed out by Ji et al [32] that the adoption of the ‘peeling off’ step would introduce the difficulties to guarantee the high quality lateral deposition and accurate vertical thickness control of the second metallic layer, which are the two prerequisites for the successful fabrication of the nanogap patterns. Moreover the success rate for obtaining large area nanogaps array with less cracks or adhesive residuals is still unsatisfactory. In this work, we present a simple yet robust method through the combination of the atomic layer deposition (ALD) and lift-off process to fabricate sub-5nm nanogaps array with uniform morphologies over a large area. The potential use of the as-prepared nanogaps as the surface enhanced Raman scattering (SERS) substrates has also been experimentally demonstrated.

2. Fabrication of ultrafine nanogaps array

The fabrication process is schematically displayed in Figs. 1(a)-1(f). First, the substrate was covered with a thin metallic (such as Au, Ag, e.g.) film by E-beam evaporation, followed by a spin-coated thin photoresist. A resist pattern was produced through standard lithography technique (such as EBL or deep UV lithography), which was subsequently used as the etching mask for the ion beam etching (IBE) process (Fig. 1(b)). The as-formed sample was then coated conformally with an ultrathin Al₂O₃ layer via ALD (Fig. 1(c)). During this procedure, the thickness of the Al₂O₃ layer was precisely controlled at sub-nanometer scale as it plays the key role on the determination of the gap size. A second round of IBE process was adopted to remove the Al₂O₃ layer on the topside of resist pattern (Fig. 1(d)), while the Al₂O₃ layers adhered on the vertical sidewall were remained, which can be served as the sacrificial layer for the next step. The sample was processed with metallic evaporation and lift-off procedure again (Fig. 1(e)), and was then immersed into 3% H₂SO₄ solution for 12 hours to etch off the residual Al₂O₃ sacrificial layer. Finally, as shown in Figs. 1(f) and 1(g), the array with nanometers gap formed alongside the contour of the initial metallic pattern can be obtained.

 

Fig. 1 The schematics of the fabrication flow of nanogaps. (a) EBL of the nanopatterns on Si/SiO₂ substrates covered with Au film. (b) Removal of the Au film without the protecting by PMMA through IBE process. (c) ALD of the ultrathin Al₂O₃ film. (d) Removal of Al₂O₃ film by IBE. (e) E-beam deposition of metallic film. (f) Removal of PMMA resist and remained Al₂O₃ films. (g) The schematics of the metallic nanogaps array. (h) Uniform nanogaps of 5 nm width around the gold patterns of high density.

Download Full Size | PDF

This method allows creating sub-nanometer nanogaps regardless of either optical lithography or EBL is used to define the larger pattern. In this case, the gap width is determined only by the thickness of ALD-grown Al₂O₃ layer and the gap depth is determined by the thickness of initial metallic film. Thanks to the atomic layer growth of the ALD technique, the width of the as-prepared nanogaps can be precisely controlled within sub-nanometers. We can conservatively estimate that the gap width would reach the level of sub-5 nm. With the initial metallic patterns defined by the conventional lithography such as EBL, a high density of nanogaps array could be made readily. Figure 1(h) shows a typical nanogaps array with the rectangle pattern and the gap width of 5 nm.

By using standard optical lithography, patterns composed by nanogaps array could be produced over a large area to realize the high throughput production. Figures 2(a)-2(c) exhibit the nanogaps array fabricated covering over the entire 4-inch Si wafer, on which the 4 nm nanogaps are formed along the contour of micro-rectangle unit (4 × 60 μm) defined by UV lithography. From the high resolution transmission electron microscopy (HRTEM) image in Fig. 2(c), we can find that the width of the nanogap is about 4.3 nm and there is no obvious Al₂O₃ remaining in the nanogaps. Other than the sub-5 nm scale size control of the uniformly fabricated nanogaps and high throughput mentioned above, the present method also possesses other advantages. On one hand the contour profile of the nanogaps array is not restricted, which can be produced in various shapes during the preliminary lithography step. The density of the nanogaps array is also only depended on the resolution of the preliminary lithography technology. For example, Figs. 2(d) and 2(f) show the SEM and HRTEM images of gold squares with the edge lengths of 10 μm and 1 μm, outlined by 5.2 nm and 4.1 nm nanogaps respectively (see Figs. 2(e) and 2(g)). And Fig. 2(h) shows the gold triangle patterns with the edge lengths of 1 μm outlined by 3.3 nm nanogaps (see Figs. 2(h) and 2(i)). Moreover, since the method does not involve stripping or peel-off procedure, the as-prepared nanogaps array over the large area (like a wafer-scale) remains undamaged from organic residuals, cracks and tears. On the other hand, the geometry of the nanogaps could be well-controlled since both of the gap sidewalls are vertically oriented with the flat and continuous top surfaces in comparison with the existing method [15]. Also the gap depth, determined by the sidewall heights, could be accurately controlled by the thickness of metallic layers fabricated through the two rounds of metallic E-beam evaporation processes. This method thus offers the potential applications for the fabrication of metamaterials with electromagnetic coupling [6].

 

Fig. 2 SEM and HRTEM images of the nanogaps array with different shapes. (a) Wafer scale fabrication of 4 nm nanogaps array. (b) and (c) Magnified SEM and HRTEM images of the 4 nm nanogaps on the 4 inch silicon wafer. (d) Nanogaps with the width of 5.2 nm along the gold square, and (e) its HRTEM images with high resolutions of the nanogaps. (f) SEM and (g) HRTEM images of the nanogaps array along the side of the nano-squares. (h) SEM and (i) HRTEM images of the nanogaps array along the nano-triangles.

Download Full Size | PDF

3. SERS characterization of the nanogaps array

Metallic nanogaps with such ultra-narrow sizes are capable of generating significant field enhancement caused by gap plasmons. It has been widely used as the substrates for SERS measurements. With our fabricated nanogaps, it is possible to study the gap size dependence of the SERS signals for a given molecule. We have carried out Raman measurements on three gold nanogaps array samples with identical square patterns (10 μm × 10 μm) but different gap sizes ranging from 3 to 10 nm, for which the gap depth is kept constant to be 30 nm. A self-assembled monolayer of 4-aminothiophenol (4-ATP) molecules was chosen to be the probe for SERS measurement [33]. The freshly fabricated nanogaps arrays were incubated in the ATP alcohol solution with a concentration of 10⁻⁵ M for 12 hours to form the molecular monolayer coated all over the sample surfaces, rinsed thoroughly with ethanol to remove the residual molecules, and then blow-dried with a N₂ stream before use. All the Raman spectra were measured by using 663 nm nonpolarized laser, of which the focusing spot is ~1 μm in diameter (it should be noted that when measuring Raman signals from the gap region, the gap slits were positioned at the center of the laser spot). The representative Raman spectra acquired from the three samples are shown in Fig. 3(a). As expected, Raman spectra from the nanogap region exhibit the typical spectral features with three dominate peaks at 1080, 1177 and 1587 cm⁻¹ from 4-ATP molecules, indicating a successful assembly of the 4-ATP molecules. Besides, much stronger intensities from the sub-10 nm nanogaps [yellow, green and red line] were observed, comparing that there was no obvious signal from the unpatterned Au surface [black line]. Taking the Raman peak around 1080 cm⁻¹ as an example, the relative Raman enhanced factor (EF) ratio of 5 nm and 3 nm arrays to 10 nm array is evaluated to be about 11 and 31 respectively, in our system. It is noted that the high uniformity and precision of our fabricated nanogaps result in very clean spectra, avoiding the unexpected signal fluctuations [15] introduced by the wall-surface roughness of the nanogaps produced by other methods. It is interesting to observe the polarization-dependent SERS spectra of ATP molecules by the confocal Raman imaging of the nanogaps. As shown in Figs. 3(b) and 3(c), for the square patterned nanogaps array (10 μm × 10 μm) with a 5 nm gap width and 30 nm gap depth, the SERS intensity is maximized when the polarization of the incident laser is perpendicular to the orientation of the nanogap slits, whereas is minimized when the two are parallel. Such a phenomenon evidently suggests the polarization-dependent distribution of the electronic filed intensity of the nanogaps array, and is consistent with the FDTD simulation results shown in Figs. 3(d) and 3(e). These results provide yet another strong evidence that the detected SERS signals come from the molecules inside the gap, but not the surface.

 

Fig. 3 Raman spectra of ATP molecules in the nanogaps. (a) Raman spectra of ATP molecules enhanced by the metallic nanogaps with the width of 3nm, 5nm and 10nm, respectively, in comparison with the Raman spectrum of the ATP molecules on gold film (black line). (b) and (c) Raman mapping of the ATP molecules in the nanogaps under incident lasers of different polarization directions.(d) and (e) Simulated electric field intensity mapping in the nanogap under the perpendicular and parallel polarization with respect to the nanogap.

Download Full Size | PDF

Besides the capability of enhancement of the substrate, one of the other important factors to evaluate the performance of the SERS substrate is its uniformity. Figure 4(a) shows the enhanced Raman spectra obtained from the ten random positions (The distance between different positions is apart from at least 100 micrometers) in the nanogaps array shown in Fig. 2(f). It is obviously that these Raman spectra have similar profiles, implying the uniformity of the Raman enhancement over the whole substrate. Figure 4(b) gives the distribution of the Raman intensities around 1080 cm⁻¹ and 1587 cm⁻¹ peaks from above ten positions, and the relative standard deviation (RSD) of the intensity is estimated to be about 10.1% and 11.8%, respectively. The results further demonstrate a very uniform Raman enhancement of this as-prepared nanogaps array SERS substrate.

 

Fig. 4 The characteristics of the uniformity of SERS signals on the nanogap array. (a) Enhanced Raman spectra in ten random positions over the whole substrate with dense nanogaps array. (b) The distribution of intensities for the 1080 cm⁻¹ and 1587 cm⁻¹ peaks from different positions.

Download Full Size | PDF

It is clear that our method is particularly useful for making large scale SERS substrate with well-defined spectral features of high signal/noise ratio. We would like to further point out that by using the flexible film assistant hydrolysis of a sacrificial layer and transfer process (HSLT) proposed in our previous report [34], the metallic nanogaps array over a large area could be completely transferred onto flexible film as showing in Fig. 5(a). Firstly, a thin layer of PMMA film was spinning coating on the metallic nanogaps at 2000 rpm and baked at 180 °C for 2 minutes. After that, the whole sample was immersed in 1 M KOH solution for 5 minutes to etching off the underneath SiO₂ layer to release the suspending PMMA/PDMS + Au nanogaps film. Then the flexible film can be transferred to other substrate for further use. Because the metallic layer was tightly supported by the PMMA/PDMS layer, the nanogaps array can retain its origin morphologies without any damage. Moreover, due to the metallic layer attached to the SiO₂ substrate, the exposed surface of the nanogaps array has an ultrasmooth morphologies, which can reduce the loss of the propagation of surface plasmon polaritons. Following above procedure, the produced flexible SERS substrate can be easily attached onto even the nonplanar surfaces. In Fig. 5, we show two different examples, in which a square patterned nanogaps array over an area of 1 cm × 1 cm was transferred to a thin PDMS film (Fig. 5(b)) and a PMMA film, respectively. It can be clearly seen from the corresponding SEM images in Figs. 5(c) and 5(d), the nanogaps remain their gap size and uniformity on both substrates. It is noted that the use of PMMA generates a higher success rate in terms of preserving the shape and size of the nanogap. This fact is also reflected by the measured SERS of 4-ATP molecules as illustrated in Fig. 5(e). The molecular sample was prepared in the same way as described in Fig. 3. The expected large enhancement for the Raman signals is clearly observed. The PMMA substrate can provide much better SERS resolution and it is also much easier to attach on the uneven surface. However, the PDMS has its own advantage, such as better mechanical strength and more stable chemical property. These two complementary substrates can thus be used for different chemical environments.

 

Fig. 5 Nanogaps array on flexible substrate. (a) Schematic fabrication process of nanogaps on PDMS/PMMA. (b) Nanogaps with the width of 5 nm on PDMS film, and (c) the corresponding SEM image of high resolutions. (d) The SEM image of the nanogaps array on PMMA film. (e) Raman spectra of 4-ATP molecules obtained from two flexible substrates in comparison with that from the flat gold film.

Download Full Size | PDF

4. Conclusion

In summary, the atomic layer deposition of the Al₂O₃ film provides a sub-nanometer thickness controllability of the sacrificial layer. Our method has been employed to successfully fabricate high quality metallic nanogaps array with sub-5 nm gap size. The fabricated nanogaps with sub-nanometers width are highly uniform over a large area. It provides an excellent substrate for SERS measurements. Our fabrication procedures can be combined with many high throughput top-down methods such as UV lithography and nanoimprint. It is a low cost and high efficiency technique that can find a variety of applications in SERS, biosensors and molecular electronics.