Scalable Dual‑Layer Porous Gold Films Deliver 10⁻¹³ M Detection Limits in Surface‑Enhanced Raman Scattering

Abstract

Surface‑enhanced Raman scattering (SERS) offers unrivaled sensitivity, speed, and molecular fingerprinting for applications in medicine, environmental monitoring, and food safety. Here we present a straightforward, scalable approach to fabricate large‑area SERS substrates that feature spatially stacked plasmonic hotspots. The substrates consist of two consecutive porous gold layers produced by magnetron sputtering, annealing, and hydrofluoric‑acid vapor etching. The resulting dual‑layer films exhibit exceptional Raman enhancement, achieving a limit of detection as low as 10^-13 M for rhodamine 6G (R6G). The high density of three‑dimensional hotspots, combined with the simple, cost‑effective fabrication, positions these substrates as strong candidates for mass‑produced, inexpensive SERS sensors.

Background

SERS is a powerful analytical tool that probes vibrational signatures of molecules with high sensitivity, making it invaluable for bio‑sensing, diagnostics, environmental assessment, and food analysis. The technique relies on the localized surface plasmonic resonance (LSPR) of metal nanostructures, which generates intense electromagnetic “hotspots” at sharp corners, tips, and inter‑nanogaps. These hotspots amplify Raman signals, and the density and distribution of hotspots directly influence SERS performance.

Traditional 2D nanostructures—such as nanoparticles, rough films, and porous arrays—offer limited hotspot density, restricting sensitivity. Three‑dimensional (3D) architectures, by contrast, distribute hotspots throughout the illumination volume, enhancing the probability of molecule‑hotspot interactions. However, many 3D fabrication methods are complex, costly, or lack reproducibility.

Our goal is to create a 3D SERS platform that is simple to fabricate, scalable, and yields a high density of spatially stacked hotspots. By combining magnetron sputtering with thermal annealing and HF vapor treatment, we generate double‑layer porous gold films that meet these criteria.

Methods

Fabrication

Ultra‑thin gold films were deposited on clean SiO₂ substrates via magnetron sputtering at 32 nm min^-1. The deposition time varied from 19 to 75 s, controlling film thickness. Subsequent annealing at 180–220 °C for 30 min (2 °C min^-1 ramp) caused gold to de‑wet and form porous structures. The substrates were then placed in a sealed vessel with 0.5 M HF; vapor etching roughened the SiO₂ surface, facilitating the release of the porous film into water. A second SiO₂ substrate was then slowly introduced to pick up the floating film, producing a dual‑layer stack on a single wafer. All steps are straightforward, require no lithography, and are amenable to large‑area production.

Structural and Optical Characterization

Scanning electron microscopy (SEM, Hitachi S3400) revealed the morphology of the single‑ and double‑layer films. Raman spectra were collected with a Horiba LabRRm 750 using a 633 nm laser (0.061 mW or 0.24 mW) and a ×10 objective (NA = 0.3). Samples were immersed in solutions of R6G, ascorbic acid, or 4‑mercaptobutyrate (4‑MBA) for 24 h before drying.

Results and Discussion

The dual‑layer design creates a 3D network of nanogaps and nanoholes that concentrate electromagnetic fields in all three spatial dimensions. SEM images (Figure 1) show that as sputtering time increases, the pores become smaller and more numerous, while the overall film remains continuous. Raman maps (Figure 11) confirm that the dual‑layer films generate stronger and more uniform signals compared to single‑layer counterparts.

Raman spectra of R6G (10^-8 M) illustrate the dramatic enhancement afforded by the dual‑layer structure. For sputtering times between 19 and 47 s, peak intensities increase by up to 10× relative to a plain Au film, with the 1653 cm^-1 band reaching 3.7× the single‑layer value. At 75 s, hotspot density drops, reducing enhancement—highlighting an optimal thickness window.

Temperature studies (Figure 9) show that annealing at 200–220 °C yields comparable performance, demonstrating fabrication flexibility. Uniformity tests across four random spots and 15 µm × 15 µm mapping confirm consistent enhancement, a prerequisite for reliable sensing.

Detection limits were pushed to 10^-13 M for R6G (Figure 12), surpassing many reported SERS platforms. Similar sensitivity was achieved for ascorbic acid (10^-9 M) and 4‑MBA (10^-10 M) (Figure 13). Linear relationships between Raman intensity at 1368 cm^-1 and the logarithm of R6G concentration were observed, underscoring quantitative capability.

Conclusions

We have demonstrated a low‑cost, scalable method to fabricate dual‑layer porous gold SERS substrates that combine high hotspot density, mechanical flexibility, and excellent reproducibility. The substrates achieve 10^-13 M detection limits for R6G and maintain sensitivity for other analytes, positioning them as promising platforms for biomedical diagnostics, food safety, and environmental monitoring.