Surface-enhanced Raman scattering (SERS) is a sensitive spectroscopic method to detect molecular vibrations on or near metallic surfaces supporting plasmonic excitation [1, 2]. At present, it is generally accepted that the SERS spectra can be greatly enhanced, owing to the two mechanisms [3, 4]. Specifically, the electromagnetic mechanism  is related to the local resonant plasmonic fields near metal nanostructures , whereas the so-called chemical contribution  is due to the formation of a charge transfer adsorption band between the Raman scattering molecules and the metallic surface (for the discussion of the well-known publication by Fleischman et al.  and the early history of SERS, see, e.g., ). The electromagnetic mechanism makes the major contribution to the SERS effect because it is both the incident and the Raman emitted field that are enhanced by the plasmonic nanostructures on the surface, thus leading to the well-known fourth-power law .
Since its discovery, the SERS technique has found numerous applications in chemical and biological sensing [8, 9] (including single-molecule detection [10, 11]), molecular and reaction dynamics , and biomedicine . To date, the physical principles of SERS, its experimental implementation, and its applications in fundamental and applied sciences have been extensively reviewed [14–21]; the readers are referred to these reviews and the books [1, 2, 8].
Despite the enormous number of SERS-related publications, all the currently used SERS platforms can be placed into one of the following four broad classes determined according to the underlying fabrication method: (1) regular metal nanolithographic nanostructures [22, 23], (2) metallic nanostructures obtained with the appropriate nanosized templates (‘film-over-spheres’ platforms) [24–30], (3) metal nanoparticles (NPs) assembled on plain substrates (e.g., silicon or glass) [31–34], and (4) ‘SERS tags’ that combine plasmonic NPs and specific Raman reporter organic molecules [15, 21, 35].
The fabricated SERS substrate should ensure several key features [33, 36]: (1) high SERS enhancement and sensitivity, (2) large-scale uniformity, with the integral SERS enhancement variations over the entire substrate surface being less than 10% to 20%, (3) high stability and reproducibility between fabrication runs, and (4) low fabrication costs.
Owing to the presence of electromagnetic ‘hot spots’ near interparticle gaps, local SERS enhancements can be as high as 1011[36, 37], but the surface-averaged enhancement is usually 3 orders of magnitude lower, about 108 in the best experiments . Moreover, these enhancements are unevenly distributed over wide areas. For example, Fang et al.  showed that the enhancement distribution could vary between 2.8 × 104 and 4.1 × 1010, where the hot spots accounted for 0.0063% of the total number of sites examined but contributed about 24% to the average SERS intensity. This means that the major part of the recorded intensity can be due to the negligible percentage of the Raman molecules adsorbed just at these hottest sites. That is why numerous efforts were reported to develop various methods for the nanofabrication of large-scale SERS substrates possessing high and homogeneous electromagnetic enhancement [17, 18].
Although multistage lithographic or patterning techniques produce the most reproducible SERS substrates, these methods are not cost-effective. Moreover, the lithographic SERS substrates can provide only a moderate enhancement as compared with some random assemblies . In common practice, SERS substrates of the second type are fabricated by depositing a thin metal layer onto a self-assembled colloidal crystal. The plasmonic and SERS properties of such substrates are determined by the size of the colloidal templates used and the thickness of the deposited metal film. The film-over-spheres method allows the substrate structure to be precisely controlled, with the number of the necessary fabrication steps being minimal, which makes this technique more cost-effective. Furthermore, these substrates retain their SERS activity for months, even after their being exposed to high temperatures. For example, quite recently, Greeneltch et al. [41, 42] have fabricated a new type of plasmonic SERS substrates in the form of silver or gold nanorods immobilized on silica or polystyrene microspheres covered by thin silver or gold films. This method produces radially oriented SERS-active pillars separated by small gaps. The surface plasmon resonance of such substrates was shown to be capable of being tuned from 330 to 1,840 nm by varying the microsphere diameter. For optimized substrates, the large-scale SERS enhancement was about 108 under near-infrared (NIR) excitation (1,064 nm).
More recently, considerable interest has been aroused in novel nanoprobes named SERS tags [16, 21] that combine plasmonic metal nanoparticles and organic Raman reporter molecules. Such SERS-active nanoprobes produce strong, characteristic Raman signals and can be used as convenient Raman labels for the indirect sensing of the target molecules by various versions of laser microscopic Raman spectrometry. In a sense, these Raman labels can be used in the same way as external chromophores, such as quantum dots or fluorescent dyes.
Perhaps the most simple and cost-effective strategy for the manufacture of SERS substrates is to fabricate self-assembled nanoparticle films (or metal islands [43, 44]) on a plain supporting surface. Owing to the advances in synthesis technologies, there exist a lot of chemical protocols to fabricate metal nanoparticles differing in size, shape, structure, and composition [45–47]. In particular, plasmonic nanopowders [48, 49] seem to be quite suitable for the simple and low-cost fabrication of SERS platforms based on random nanoparticle assemblies . One of the obvious advantages of plasmonic nanopowders is that they retain their plasmonic properties under usual conditions; they can be easily dispersed in water to obtain sols of the desired concentration in a matter of seconds without any sonication, heating, etc. In particular, we have already utilized GNR powders to fabricate monolayer and fractal-like plasmonic films for SERS applications . However, these substrates demonstrated a moderate analytical enhancement  averaged over the probe laser beam spot. One of the possible reasons was too small a number of the analyte molecules in the thin layers probed by the laser light.
In this work, we used gold nanorod (GNR) nanopowders  to prepare concentrated GNR sols that were then employed to deposit GNRs on an opal-like photonic crystal (OPC) film formed on a silicon wafer. Such GNR-OPC substrates combine the increased specific surface, owing to the multilayer nanosphere structure, and various spatial GNR configurations, including those with possible plasmonic hot spots [5, 51]. We demonstrate here the existence of the optimal GNR deposition density for the maximal SERS effect, which turned out to be higher than that for the thick random GNR assemblies  formed directly on a plain silicon wafer.