**Acknowledgements**

*Nanoplasmonics*

while the nanodiscs remained undisturbed. Since our study primarily intended application of these arrays within a SERS-based sensor in liquid media, the nanorod arrays could not be considered further. Despite this instability, the nanorods could still find a use for plasmonic sensors functioning in the gas phase, or alternatively, if their mechanical stability can be improved through deposition of an overcoating of a dielectric, for example, alumina as shown earlier in literature [95]. The nanodisc arrays were evaluated for SERS performance against three different molecules, namely, naphthalene thiol that represented a covalently bound analyte, 2,2′-bipyridyl that represented non-covalently bound analyte, and crystal violet which represented the possibility of resonance Raman effect. The SERS performance was benchmarked against commercially available Klarite substrates and was proved better for all three molecules. The approach demonstrated an inherently reproducible approach to the fabrication of plasmonic arrays, as several copies can be derived starting from the same nanopillars NIL stamp. The performance of these arrays nevertheless has scope for improvements by further EM enhancements by reducing feature separations. The inability to deliver low feature separations is a limitation of the approach, which can be overcome by post-processing of the porous templates

obtained after NIL, for instance, via pore-widening approaches.

Molecular self-assembly using amphiphilic copolymers and colloids derived thereof can deliver nanoplasmonic interfaces with high spatial resolutions, with control over geometric variables in steps of only a few nanometers. The approach enables metal nanoarrays with spatial coherence between features, orthogonal control over the different geometric attributes, and standard deviations below 10%. These characteristics can be leveraged to better understand and predict the optical properties of these arrays, allowing rational routes to maximize plasmonic sensing performance. The self-assembly parameters at both the template production and pattern transfer stages could be rigorously controlled to ensure high uniformity, reproducibility, and scalability of the resulting plasmonic arrays on full wafers. The correlation of the geometry ⇔ optical ⇔ SERS performance was demonstrated with a combination of experiments and numerical simulations. Plasmonic nanoarrays presenting a large number of gap hot spots, with gap distances down to sub-10 nm length scale, are possible to obtain in case of nanoparticle cluster arrays and nanopillar arrays. The homogeneously distributed hot spots over large areas present an opportunity to not only detect but also quantify the concentration of analytes, with large dynamic range with promisingly low limits of detection. Among the key challenges for future developments is to identify configurations that naturally drive the co-localization of analytes with EM hot spots to achieve maximize plasmonic signal enhancements. Further, the efforts to enhance EM fields solve only a part of the sensing challenge. In addition to maximizing the EM enhancements, the surface needs to be tailored to maximize analyte interactions and their concentrations on

Yet another challenge is the application of plasmonic arrays for biosensing. The high spatial resolutions sought for maximizing EM enhancements at gap or curvature hot spots are not compatible with the spatial requirements to accommodate large biomolecules like proteins. Further, the sensitivity of the plasmonic sensor extends typically to only a few nanometers from the surface. This is a challenge considering that the size of biomolecular interactions can already be a few tens of nanometers, for example, for an immunosandwich assay. Further, the plasmonic sensor needs to be adapted to work in complex media, for which the surface

**4. Conclusions and outlook**

**126**

the surface.

Funding received from National Research Fund of Luxembourg (FNR) via the project PLASENS (C15/MS/10459961) and FNR-PRIDE (FNR PRIDE/15/10935404) is gratefully acknowledged.
