**1. Introduction**

The optical properties of metal nanostructures and nanoarrays have been widely investigated in the context of exploiting particle plasmons to derive high performance within chemical and biosensing devices. These include sensing of range of analytes to diagnose the medical status of an individual [1]; detect the presence of chemical or biowarfare agents [2, 3], toxins, or adulterants in food [4]; or assess/monitor air, water, and soil quality in the environment [2, 5]. Plasmonic sensors offer a range of advantages over other analytical tools including high analytical and calibration sensitivity, quick response times, label-free detection opportunities, ease of integration within different sensor form factors, and the need for simple and portable instrumentation. Plasmonic sensors rely on metal nanostructures that act as nanoantennae to concentrate and enhance the electromagnetic field close to surface [6]. The enhanced EM field can be leveraged within different configurations, namely, localized surface plasmon resonance sensors (LSPR) that follow analyte-induced changes to local refractive index or report vibrational Raman or fluorescence intensities with high signal-to-noise ratios using surface-enhanced Raman spectroscopy (SERS) and metal-enhanced fluorescence (MEF), respectively. In all these cases, the performance of the sensor is critically linked to the optical properties of the plasmonic structures, which in turn correlates with their geometric attributes, namely, size, shape, aspect ratios, separation, distribution, and roughness. Control over geometries is the key to engineering profiles and intensities of the electromagnetic field around plasmonic nanostructures, which in turn determines the performance of the plasmonic transducer. The typical length scales for characteristic dimensions of plasmonic nanostructures correspond to spatial resolutions of few nanometers to few tens of nanometers. Geometric features such as pointed structures or nanoscale gaps have shown to concentrate and enhance EM field, thus acting as EM hot spots, as a function of decreasing radius of curvature or gap distances, with length scales down to sub-10 nm regime [7]. Enhanced EM fields at gap hot spots have shown to result in enhancement factors of the order of a million- to billion-fold in practice. A sizeable contribution to observe SERS signals (24% of the overall intensity) was observed to be rising out of <100 molecules per million when these molecules are positioned in EM hot spots with EM enhancement factors of the order of 10<sup>9</sup> –1010 [8]. The EM hot spots help enhance the sensitivity of different plasmonic sensors, and especially those based on enhanced spectroscopies, namely, SERS and MEF. Gaining high sensitivity would thus be largely determined by the quality and number of EM hot spots [9]. However, addressing the production of EM hot spots with length scales of the order of only a few nanometers pose a profound and non-trivial challenge for nanofabrication. Traditionally, such hot spots have been attained as a natural consequence of stochastic growth or deposition processes [10, 11], including electrochemical growth or roughening, de-wetting of nanoparticles or salts from solution phase [12–14], and de-wetting of thin films on the surface. The stochastic processes, by nature, result in a broad standard deviation in geometries, which reduces the number of most efficient hot spots and also leads to greater spot-to-spot variability in signal intensities [15, 16]. The randomness in geometry makes it particularly hard to predict response, to identify the source of issues, and to adopt a rational approach to optimize performance. Geometries with improved definition have remained forte of top-down lithography tools, for example, E-beam lithography, focused ion-beam milling, and X-ray interference lithography, which are all very time-consuming and also quite expensive [17–19]. The throughput and cost of fabrication is not only an issue for manufacturing but also reduces the efficacy of research due to the limitation in the number of samples available for investigations [20] (**Figure 1**). From the manufacturing perspective, the throughput of the process would be a key determinant of the cost, which is driven to <\$5 per sensor chip for point-of-care applications [28].

Techniques such as nanoimprint lithography and polymeric or colloidal selfassembly techniques allow enhancing throughput, albeit, at the cost of defects. The ability of self-assembly techniques in catering to the parallel fabrication of nanopatterns across arbitrarily large areas at low cost, as well as the several handles it offers toward tunability of structure dimensions down to molecular level, is unmatched by conventional lithography tools. However, self-assemblybased approaches carry limitations that demand careful attention to ensure their

**115**

**Figure 1.**

over metal nanogaps down to sub-10 nm regime.

*Nanoplasmonic Arrays with High Spatial Resolutions, Quality, and Throughput…*

usefulness: (a) Poor uniformity, which is often a result of poor process optimization and in some cases due to the susceptibility of process parameters to environmental variables. This can be addressed by mapping the impact of environmental variables and factoring them within the process optimization. (b) Large standard deviations, either inherent to the primary templates produced by the technique or those that may creep in during different stages of processing. (c) Feature shape: Typically, self-assembled template patterns have a circular feature cross section. Rectangular, triangular, or other feature shapes with lower symmetry are uncommon. (d) In the absence of any external guidance, self-assembly techniques typically lead to a polycrystalline 2D hexagonal order. Long-range ordering, square, or other lattice types besides hexagonal symmetry remain uncommon and can be attained through guidance from a top-down lithographic tool [29–34]. The lack of long-range order and presence of point or line defects are better tolerated by plasmonic sensing applications, so long as the averaged properties are consistent and reproducible with low standard deviations. The chapter will present our approach to plasmonic nanoarrays with high spatial resolutions relying on hierarchical self-assembly of amphiphilic di-block copolymers into soft colloids and their subsequent quasiperiodic organization when deposited on a planar surface [35]. The approach results in well-defined organic templates on the surface with nanometric control over the width, topography, and pitch, realized by control over parameters of molecular self-assembly. By understanding the impact of the different process parameters on the resulting geometric outcome, it is possible to deliver templates with high reproducibility and uniformity on full wafers, with a yield >90% [36]. These templates are translated into highly sensitive SERS-based plasmonic sensors, with control

*Schematic representation of the trade-off between resolution, quality, and throughput in fabricating plasmonic nanoarrays with high spatial resolutions, showing the comparative advantages and limits of (a) nanoparticle assemblies [21] (b) top-down-bottom-up control over nanoparticle assemblies [19], (c) direct-write techniques [22], (d) nanostencil lithography [23], (e) nanoimprint lithography [24], (f) photolithography [25], (g) porous anodized alumina templates [10], and (h) stochastic, random structures, including island films [11, 26],* 

*DOI: http://dx.doi.org/10.5772/intechopen.89064*

*and (i) stochastic chemical growth processes [27].*

*Nanoplasmonic Arrays with High Spatial Resolutions, Quality, and Throughput… DOI: http://dx.doi.org/10.5772/intechopen.89064*

#### **Figure 1.**

*Nanoplasmonics*

ment factors of the order of 10<sup>9</sup>

chip for point-of-care applications [28].

metal nanostructures that act as nanoantennae to concentrate and enhance the electromagnetic field close to surface [6]. The enhanced EM field can be leveraged within different configurations, namely, localized surface plasmon resonance sensors (LSPR) that follow analyte-induced changes to local refractive index or report vibrational Raman or fluorescence intensities with high signal-to-noise ratios using surface-enhanced Raman spectroscopy (SERS) and metal-enhanced fluorescence (MEF), respectively. In all these cases, the performance of the sensor is critically linked to the optical properties of the plasmonic structures, which in turn correlates with their geometric attributes, namely, size, shape, aspect ratios, separation, distribution, and roughness. Control over geometries is the key to engineering profiles and intensities of the electromagnetic field around plasmonic nanostructures, which in turn determines the performance of the plasmonic

transducer. The typical length scales for characteristic dimensions of

plasmonic nanostructures correspond to spatial resolutions of few nanometers to few tens of nanometers. Geometric features such as pointed structures or nanoscale gaps have shown to concentrate and enhance EM field, thus acting as EM hot spots, as a function of decreasing radius of curvature or gap distances, with length scales down to sub-10 nm regime [7]. Enhanced EM fields at gap hot spots have shown to result in enhancement factors of the order of a million- to billion-fold in practice. A sizeable contribution to observe SERS signals (24% of the overall intensity) was observed to be rising out of <100 molecules per million when these molecules are positioned in EM hot spots with EM enhance-

sensitivity of different plasmonic sensors, and especially those based on enhanced spectroscopies, namely, SERS and MEF. Gaining high sensitivity would thus be largely determined by the quality and number of EM hot spots [9]. However, addressing the production of EM hot spots with length scales of the order of only a few nanometers pose a profound and non-trivial challenge for nanofabrication. Traditionally, such hot spots have been attained as a natural consequence of stochastic growth or deposition processes [10, 11], including electrochemical growth or roughening, de-wetting of nanoparticles or salts from solution phase [12–14], and de-wetting of thin films on the surface. The stochastic processes, by nature, result in a broad standard deviation in geometries, which reduces the number of most efficient hot spots and also leads to greater spot-to-spot variability in signal intensities [15, 16]. The randomness in geometry makes it particularly hard to predict response, to identify the source of issues, and to adopt a rational approach to optimize performance. Geometries with improved definition have remained forte of top-down lithography tools, for example, E-beam lithography, focused ion-beam milling, and X-ray interference lithography, which are all very time-consuming and also quite expensive [17–19]. The throughput and cost of fabrication is not only an issue for manufacturing but also reduces the efficacy of research due to the limitation in the number of samples available for investigations [20] (**Figure 1**). From the manufacturing perspective, the throughput of the process would be a key determinant of the cost, which is driven to <\$5 per sensor

Techniques such as nanoimprint lithography and polymeric or colloidal selfassembly techniques allow enhancing throughput, albeit, at the cost of defects. The ability of self-assembly techniques in catering to the parallel fabrication of nanopatterns across arbitrarily large areas at low cost, as well as the several handles it offers toward tunability of structure dimensions down to molecular level, is unmatched by conventional lithography tools. However, self-assemblybased approaches carry limitations that demand careful attention to ensure their

–1010 [8]. The EM hot spots help enhance the

**114**

*Schematic representation of the trade-off between resolution, quality, and throughput in fabricating plasmonic nanoarrays with high spatial resolutions, showing the comparative advantages and limits of (a) nanoparticle assemblies [21] (b) top-down-bottom-up control over nanoparticle assemblies [19], (c) direct-write techniques [22], (d) nanostencil lithography [23], (e) nanoimprint lithography [24], (f) photolithography [25], (g) porous anodized alumina templates [10], and (h) stochastic, random structures, including island films [11, 26], and (i) stochastic chemical growth processes [27].*

usefulness: (a) Poor uniformity, which is often a result of poor process optimization and in some cases due to the susceptibility of process parameters to environmental variables. This can be addressed by mapping the impact of environmental variables and factoring them within the process optimization. (b) Large standard deviations, either inherent to the primary templates produced by the technique or those that may creep in during different stages of processing. (c) Feature shape: Typically, self-assembled template patterns have a circular feature cross section. Rectangular, triangular, or other feature shapes with lower symmetry are uncommon. (d) In the absence of any external guidance, self-assembly techniques typically lead to a polycrystalline 2D hexagonal order. Long-range ordering, square, or other lattice types besides hexagonal symmetry remain uncommon and can be attained through guidance from a top-down lithographic tool [29–34]. The lack of long-range order and presence of point or line defects are better tolerated by plasmonic sensing applications, so long as the averaged properties are consistent and reproducible with low standard deviations. The chapter will present our approach to plasmonic nanoarrays with high spatial resolutions relying on hierarchical self-assembly of amphiphilic di-block copolymers into soft colloids and their subsequent quasiperiodic organization when deposited on a planar surface [35]. The approach results in well-defined organic templates on the surface with nanometric control over the width, topography, and pitch, realized by control over parameters of molecular self-assembly. By understanding the impact of the different process parameters on the resulting geometric outcome, it is possible to deliver templates with high reproducibility and uniformity on full wafers, with a yield >90% [36]. These templates are translated into highly sensitive SERS-based plasmonic sensors, with control over metal nanogaps down to sub-10 nm regime.
