**3.1 Nanoparticle cluster arrays**

Clusters (used interchangeably with "aggregates" in this report) of metal nanoparticles behave differently from their isolated counterparts due to the collective optical behavior arising out of plasmonic coupling between the constituent nanoparticles [45–50]. Clusters of nanoparticles are known to behave as hot particles, with significantly enhanced electromagnetic fields at the inter-particle junctions [51–54]. Consequently, clusters exhibit higher extinction cross sections, with hot spots that can be excited at lower energies than the isolated particles. The aggregation-induced color change of gold nanoparticle suspensions caused by the analyte of interest has been the basis of several biological assays [55]. Such random aggregation typically results in a large distribution in the number of particles per cluster, with a distinct lack of control over those numbers. To achieve clusters with desired optical properties, it is essential to be able to produce them with a narrow distribution in the size, shape, and spatial arrangement between nanoparticles within the cluster and between clusters in an array. Such clusters were demonstrated using templates

fabricated by electron- beam lithography (EBL) [45–47, 53, 56–58] and controlling the size of the template to obtain dimers, trimers, quadrumers, and multimers. Since EBL is time-consuming and expensive to achieve high-resolution patterns spanning large areas, other low-cost means to achieve controlled nanoparticle aggregates using template-assisted means have been reported in the literature. These include the use of DNA, [59] surfactants, [60] block copolymers [60–62], carbon nanotubes [61], cylindrical micelles [62], or microorganisms like bacteria or viruses [63] as templates to attach nanoparticles. While the template-assisted cluster formation allows creating clusters with desired size (or a number of particles per cluster) and shape, they often fall short of abilities to control inter-cluster arrangement within an array. Such an arrangement is important to ensure reproducible inter-cluster plasmonic interactions as well as an ability to engineer them to achieve desired optical properties. In this direction, block copolymers are an excellent solution, as they allow effective control over the spatial arrangement of eventual clusters. Microphase separation of block copolymers (BCP) in thin films, as well as BCP reverse micelles, has been utilized as templates to attain clusters of metal nanoparticles on surfaces. These techniques have exploited the BCP templates either to organize one or more particles from solution phase [64–66] or to drive selective complexation of metal ions and in situ reduction to form clusters [67]. However, these fall short of opportunities for the rational design of the cluster properties. The approach based on copolymer reverse micelle template allows preparing clusters based on electrostatic self-assembly of preformed gold nanoparticles from the solution phase on to the features on the surface. The cluster dimensions (the number of nanoparticles per cluster) and the inter-cluster separation were controlled by control over the template size and their separation. The low standard deviation of the template enables low standard deviation in the geometric attributes of the clusters as well (**Figure 5**).

In a typical experiment, the reverse micelle arrays prepared from PS-b-PVP with a molecular weight of 114 kDa and PDI of 1.1 were spin coated from m-xylene solutions and subsequently exposed to an aqueous suspension of gold nanoparticles. Due to the presence of the pyridyl groups of PVP in the core of the features, the reverse micelle features are positively charged and attract negatively charged citrate-stabilized gold nanoparticles from the suspension. The electrostatic attraction between the nanoparticles occurs locally on the features and not in between them, therefore resulting in a patterned array of nanoparticle clusters. The number of nanoparticles in each feature is a function of its size. The inter-cluster separations are controlled by the control over the pitch of the template. Extinction spectra of the nanoparticle clusters showed a peak at 620 nm, which was over 100 nm red-shifted from the plasmon resonance band of isolated Au nanoparticles. This is attributed to the strong plasmonic coupling between the nanoparticles within the cluster. Further, the absence of a strong contribution at 520 nm would confirm the geometric observation of the absence of isolated nanoparticles (**Figure 6**). Finitedifference time-domain (FDTD) simulations show inter-particle as well as intercluster hot spots. The 3D shape of the clusters was found to be necessary for the inter-cluster plasmonic coupling, and an increase in EM field at inter-particle and inter-cluster hot spots was found to correlate with decreasing inter-cluster separations. The expectations from the geometry and optical properties of the nanoparticle cluster arrays were validated by SERS measurements of a probe molecule. Three different probe molecules were tested, namely, crystal violet, naphthalene thiol, and 2, 2′-bipyridine. In all three cases, the SERS intensities were found to be highest for the largest clusters with the smallest separations. The SERS intensities were higher for crystal violet than for smaller molecules due to resonance Raman effects, enabling SERS enhancement factors over 108 . SERS quantitative assays showed the

**121**

**Figure 6.**

**Figure 5.**

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

*(a) Schematic of the fabrication of gold nanoparticle cluster arrays by electrostatic attraction of negatively charged Au nanoparticles to positively charged templates on surface, (b) zeta potential titration showing the template acquiring positive potential for pH below 8.3, (c) systematic control over the size (pitch kept a constant), and (d) separation (Sc) between the nanoparticle clusters (size kept a constant at N~18) which is* 

*(a) Graph shows the shift in the extinction peak of the nanoparticle clusters with change in separation, Sc, and change in no. of nanoparticles/cluster, N. (b) SERS spectra of CV shows the increase in the peak intensity at 1625 cm−1 with change in the separations, Sc, and no. of nanoparticles/cluster, N. (c) and (d) SERS intensity vs.* 

*concentration showing linear relationship for 1-NT and BPE, respectively [42].*

*shown by the histograms and AFM measurements, respectively [42].*

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

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

#### **Figure 5.**

*Nanoplasmonics*

fabricated by electron- beam lithography (EBL) [45–47, 53, 56–58] and controlling the size of the template to obtain dimers, trimers, quadrumers, and multimers. Since EBL is time-consuming and expensive to achieve high-resolution patterns spanning large areas, other low-cost means to achieve controlled nanoparticle aggregates using template-assisted means have been reported in the literature. These include the use of DNA, [59] surfactants, [60] block copolymers [60–62], carbon nanotubes [61], cylindrical micelles [62], or microorganisms like bacteria or viruses [63] as templates to attach nanoparticles. While the template-assisted cluster formation allows creating clusters with desired size (or a number of particles per cluster) and shape, they often fall short of abilities to control inter-cluster arrangement within an array. Such an arrangement is important to ensure reproducible inter-cluster plasmonic interactions as well as an ability to engineer them to achieve desired optical properties. In this direction, block copolymers are an excellent solution, as they allow effective control over the spatial arrangement of eventual clusters. Microphase separation of block copolymers (BCP) in thin films, as well as BCP reverse micelles, has been utilized as templates to attain clusters of metal nanoparticles on surfaces. These techniques have exploited the BCP templates either to organize one or more particles from solution phase [64–66] or to drive selective complexation of metal ions and in situ reduction to form clusters [67]. However, these fall short of opportunities for the rational design of the cluster properties. The approach based on copolymer reverse micelle template allows preparing clusters based on electrostatic self-assembly of preformed gold nanoparticles from the solution phase on to the features on the surface. The cluster dimensions (the number of nanoparticles per cluster) and the inter-cluster separation were controlled by control over the template size and their separation. The low standard deviation of the template enables low standard deviation in the

In a typical experiment, the reverse micelle arrays prepared from PS-b-PVP with a molecular weight of 114 kDa and PDI of 1.1 were spin coated from m-xylene solutions and subsequently exposed to an aqueous suspension of gold nanoparticles. Due to the presence of the pyridyl groups of PVP in the core of the features, the reverse micelle features are positively charged and attract negatively charged citrate-stabilized gold nanoparticles from the suspension. The electrostatic attraction between the nanoparticles occurs locally on the features and not in between them, therefore resulting in a patterned array of nanoparticle clusters. The number of nanoparticles in each feature is a function of its size. The inter-cluster separations are controlled by the control over the pitch of the template. Extinction spectra of the nanoparticle clusters showed a peak at 620 nm, which was over 100 nm red-shifted from the plasmon resonance band of isolated Au nanoparticles. This is attributed to the strong plasmonic coupling between the nanoparticles within the cluster. Further, the absence of a strong contribution at 520 nm would confirm the geometric observation of the absence of isolated nanoparticles (**Figure 6**). Finitedifference time-domain (FDTD) simulations show inter-particle as well as intercluster hot spots. The 3D shape of the clusters was found to be necessary for the inter-cluster plasmonic coupling, and an increase in EM field at inter-particle and inter-cluster hot spots was found to correlate with decreasing inter-cluster separations. The expectations from the geometry and optical properties of the nanoparticle cluster arrays were validated by SERS measurements of a probe molecule. Three different probe molecules were tested, namely, crystal violet, naphthalene thiol, and 2, 2′-bipyridine. In all three cases, the SERS intensities were found to be highest for the largest clusters with the smallest separations. The SERS intensities were higher for crystal violet than for smaller molecules due to resonance Raman effects,

. SERS quantitative assays showed the

geometric attributes of the clusters as well (**Figure 5**).

enabling SERS enhancement factors over 108

**120**

*(a) Schematic of the fabrication of gold nanoparticle cluster arrays by electrostatic attraction of negatively charged Au nanoparticles to positively charged templates on surface, (b) zeta potential titration showing the template acquiring positive potential for pH below 8.3, (c) systematic control over the size (pitch kept a constant), and (d) separation (Sc) between the nanoparticle clusters (size kept a constant at N~18) which is shown by the histograms and AFM measurements, respectively [42].*

#### **Figure 6.**

*(a) Graph shows the shift in the extinction peak of the nanoparticle clusters with change in separation, Sc, and change in no. of nanoparticles/cluster, N. (b) SERS spectra of CV shows the increase in the peak intensity at 1625 cm−1 with change in the separations, Sc, and no. of nanoparticles/cluster, N. (c) and (d) SERS intensity vs. concentration showing linear relationship for 1-NT and BPE, respectively [42].*

lowest detection limits for naphthalene thiol to be an order of magnitude lower than 2,2′- bipyridine. This is attributed to the higher density of molecules for naphthalene thiol on the surface due to the covalent bonding of the thiols to the gold surface at a given concentration. The low standard deviations of the cluster arrays enabled signal intensity variations below 10% on several square millimeters on the surface.

## **3.2 Metal nanopillars arrays**

Nanopillar shapes offer a unique opportunity to control and enhance EM field profiles and enhancements [68]. Nanopillar arrays for plasmonic sensing have been reported previously in the literature, for detection of disease markers and identification of bacteria as well as environmental pollutants [43, 69–71]. Such configurations can be obtained by metal deposition of optimal thickness on top of high aspect ratio dielectric pillars that can be produced by different approaches, namely, VLS growth [72], pulsed laser deposition [73], black Si production by RIE with gas plasmas [74], and patterned wet or dry etching with templates produced by other means, for example, photolithography, soft-lithography, EBL, NIL, NSL, and molecular self-assembly [75–80]. Nanopattern of pillars offers an advantage over stochastically arranged counterparts due to a better definition of the spatial relationship between the pillars that help rational enhancement to sensing performance. The stochastic arrangements, despite promising results, rely largely on empirical optimization and suffer from difficulty in identifying issues when they arise. Another aspect to consider is the stability of the pillar arrays, which is especially a concern for high aspect ratio structures [81–84]. High aspect ratio pillars can capture analyte between the pillars when they collapse by drying that may offer interesting means to trap analyte potentially at EM hot spots [85, 86]; the approach, however, gives less control over the collapse nor the concentration of analyte that is trapped, thus offering limited opportunity for rational design. The irreversible nature of such collapse would make it difficult to subject the pillars to solvent-based washing and drying steps, limiting them to a single time or single-step usage. Pillars of smaller aspect ratios would solve many of these issues,; however, it will need to be produced with high spatial resolutions to ensure a large number of EM hot spots and preferably with good geometric definition allowing rational optimization toward both enhancing EM hot spots and analyte capture on the surface. An approach based on colloids of amphiphilic copolymers allows the possibility to generate nanopillars that are stable and reusable while allowing significant opportunities to rationally engineer the optical properties toward high plasmonic sensing performance.

Organic reverse micelle templates can be transferred into the underlying Si by adopting nanolithography processes based on reactive ion etching with halogen gas plasmas. Given the low template thickness (typically less than 30 nm) and small widths (sub-100 nm), nanolithography using these templates requires significant process optimization to ensure high selectivity to the underlying substrate, minimum undercut, and high anisotropy. It is possible to substitute the organic templates with a harder inorganic template to improve the selectivity and durability. This is achieved by transferring the organic template pattern into a thin dielectric film to generate harder dielectric masks for the next step of pattern transfer into Si. We had earlier shown a high degree of control over the incorporation of the organometallic precursor by using atomic layer deposition, with the resulting oxide nanostructures enabling nanolithography down to sub-10 nm regime [35]. This was followed by several other investigations in literature focusing on vapor phase incorporation of metal-organic precursors within different block copolymer

**123**

**Figure 7.**

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

domains using ALD processes [87–91]. It is important that the nanolithography process is optimized to not widen standard deviations between the features or across the wafer. The optimization processes across the full wafer level should take into account the "loading effects," which impacts the outcome of the reactive ion etching process depending on the proportion of exposed surface area available for etching. Nanopillar arrays that are shorter and closer allows all advantages of the other reported approaches, while in addition also conserving the metal to be deposited. Nanopillar arrays of silicon obtained by molecular self-assembly approach (**Figure 8**) can subsequently be coated with gold or silver to prepare plasmonic nanopillar arrays. The conditions of the coating, including the choice of evaporation versus sputtering, the respective deposition parameters are critical to the structure and property of the resulting plasmonic arrays. The thickness of the films deposited in relation to the geometry of the underlying silicon pillars would be a key process parameter that determines the aspect ratio and the separation between the metal pillars. Both aspect ratio and the feature diameters increase, while the feature separations decrease systematically as a function of the thickness of metal deposited. The observed film growth is strongly anisotropic with slow growth laterally as compared with vertically. The growth on top of the silicon pillars correlates well with the thickness of the metal deposited, while the increase in diameter occurs much slower. The challenge in the processing includes the conformal deposition of metal, with good step edge coverage, especially when the deposition is made on top

In a specific example, organic templates prepared using PS-b-PVP with a molecular weight of 80.5 kDa was spin-coated on a 100mm Si wafer consisting of a 25 nm film of thermally grown SiO2 layer (**Figure 7a**). The coating is exposed to brief O2 plasma to remove the thin residual layer between the template features and expose the substrate beneath. The resulting organic template is then transferred into an underlying SiO2 thin film using C4F8/CH4 gas plasma. The resulting SiO2 islands provide high selectivity in etching underlying Si using SF6/C4F8 plasma to yield Si nanopillars. The RIE conditions employed resulted in silicon pillars with a positively tapered profile with a feature size of 40 nm (at half-width) and pitch of 78 nm and height of 120 nm, as measured by SEM (**Figure 7b**). The RIE process conditions can be varied to obtain pillars with other shapes as well (**Figure 7c**). The pillar arrays were found to be uniform throughout the wafer, as qualified by uniform color, with a variation of <10% as measured by reflectance spectroscopy [35]. **Figure 8** shows another case of similarly obtained nanopillars arrays prepared using

*(a) Fabrication process involved in pattern transfer from polymer template to fabricate silicon nanopillar arrays. (b) SEM image taken from a random position on the wafer with nanopillar arrays. (c) SEM images of* 

*different shapes of nanopillars obtained by varying etching conditions [35].*

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

of closely separated high aspect ratio nanopillars.

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

domains using ALD processes [87–91]. It is important that the nanolithography process is optimized to not widen standard deviations between the features or across the wafer. The optimization processes across the full wafer level should take into account the "loading effects," which impacts the outcome of the reactive ion etching process depending on the proportion of exposed surface area available for etching.

Nanopillar arrays that are shorter and closer allows all advantages of the other reported approaches, while in addition also conserving the metal to be deposited. Nanopillar arrays of silicon obtained by molecular self-assembly approach (**Figure 8**) can subsequently be coated with gold or silver to prepare plasmonic nanopillar arrays. The conditions of the coating, including the choice of evaporation versus sputtering, the respective deposition parameters are critical to the structure and property of the resulting plasmonic arrays. The thickness of the films deposited in relation to the geometry of the underlying silicon pillars would be a key process parameter that determines the aspect ratio and the separation between the metal pillars. Both aspect ratio and the feature diameters increase, while the feature separations decrease systematically as a function of the thickness of metal deposited. The observed film growth is strongly anisotropic with slow growth laterally as compared with vertically. The growth on top of the silicon pillars correlates well with the thickness of the metal deposited, while the increase in diameter occurs much slower. The challenge in the processing includes the conformal deposition of metal, with good step edge coverage, especially when the deposition is made on top of closely separated high aspect ratio nanopillars.

In a specific example, organic templates prepared using PS-b-PVP with a molecular weight of 80.5 kDa was spin-coated on a 100mm Si wafer consisting of a 25 nm film of thermally grown SiO2 layer (**Figure 7a**). The coating is exposed to brief O2 plasma to remove the thin residual layer between the template features and expose the substrate beneath. The resulting organic template is then transferred into an underlying SiO2 thin film using C4F8/CH4 gas plasma. The resulting SiO2 islands provide high selectivity in etching underlying Si using SF6/C4F8 plasma to yield Si nanopillars. The RIE conditions employed resulted in silicon pillars with a positively tapered profile with a feature size of 40 nm (at half-width) and pitch of 78 nm and height of 120 nm, as measured by SEM (**Figure 7b**). The RIE process conditions can be varied to obtain pillars with other shapes as well (**Figure 7c**). The pillar arrays were found to be uniform throughout the wafer, as qualified by uniform color, with a variation of <10% as measured by reflectance spectroscopy [35]. **Figure 8** shows another case of similarly obtained nanopillars arrays prepared using

#### **Figure 7.**

*Nanoplasmonics*

on the surface.

**3.2 Metal nanopillars arrays**

lowest detection limits for naphthalene thiol to be an order of magnitude lower than 2,2′- bipyridine. This is attributed to the higher density of molecules for naphthalene thiol on the surface due to the covalent bonding of the thiols to the gold surface at a given concentration. The low standard deviations of the cluster arrays enabled signal intensity variations below 10% on several square millimeters

Nanopillar shapes offer a unique opportunity to control and enhance EM field profiles and enhancements [68]. Nanopillar arrays for plasmonic sensing have been reported previously in the literature, for detection of disease markers and identification of bacteria as well as environmental pollutants [43, 69–71]. Such configurations can be obtained by metal deposition of optimal thickness on top of high aspect ratio dielectric pillars that can be produced by different approaches, namely, VLS growth [72], pulsed laser deposition [73], black Si production by RIE with gas plasmas [74], and patterned wet or dry etching with templates produced by other means, for example, photolithography, soft-lithography, EBL, NIL, NSL, and molecular self-assembly [75–80]. Nanopattern of pillars offers an advantage over stochastically arranged counterparts due to a better definition of the spatial relationship between the pillars that help rational enhancement to sensing performance. The stochastic arrangements, despite promising results, rely largely on empirical optimization and suffer from difficulty in identifying issues when they arise. Another aspect to consider is the stability of the pillar arrays, which is especially a concern for high aspect ratio structures [81–84]. High aspect ratio pillars can capture analyte between the pillars when they collapse by drying that may offer interesting means to trap analyte potentially at EM hot spots [85, 86]; the approach, however, gives less control over the collapse nor the concentration of analyte that is trapped, thus offering limited opportunity for rational design. The irreversible nature of such collapse would make it difficult to subject the pillars to solvent-based washing and drying steps, limiting them to a single time or single-step usage. Pillars of smaller aspect ratios would solve many of these issues,; however, it will need to be produced with high spatial resolutions to ensure a large number of EM hot spots and preferably with good geometric definition allowing rational optimization toward both enhancing EM hot spots and analyte capture on the surface. An approach based on colloids of amphiphilic copolymers allows the possibility to generate nanopillars that are stable and reusable while allowing significant opportunities to rationally engineer the optical properties toward high plasmonic sensing

Organic reverse micelle templates can be transferred into the underlying Si by adopting nanolithography processes based on reactive ion etching with halogen gas plasmas. Given the low template thickness (typically less than 30 nm) and small widths (sub-100 nm), nanolithography using these templates requires significant process optimization to ensure high selectivity to the underlying substrate, minimum undercut, and high anisotropy. It is possible to substitute the organic templates with a harder inorganic template to improve the selectivity and durability. This is achieved by transferring the organic template pattern into a thin dielectric film to generate harder dielectric masks for the next step of pattern transfer into Si. We had earlier shown a high degree of control over the incorporation of the organometallic precursor by using atomic layer deposition, with the resulting oxide nanostructures enabling nanolithography down to sub-10 nm regime [35]. This was followed by several other investigations in literature focusing on vapor phase incorporation of metal-organic precursors within different block copolymer

**122**

performance.

*(a) Fabrication process involved in pattern transfer from polymer template to fabricate silicon nanopillar arrays. (b) SEM image taken from a random position on the wafer with nanopillar arrays. (c) SEM images of different shapes of nanopillars obtained by varying etching conditions [35].*

#### **Figure 8.**

*(Top) Silicon nanopillar arrays with increasing thickness of gold (top, insets) show evolution in electromagnetic field profiles (bottom, left) and SERS spectra of 1-naphthalene thiol as function of metal pillar geometry (bottom, right) concentration dependence of intensity of the peak at 1371 cm<sup>−</sup><sup>1</sup> showing quantitative detection with low detection limits [43].*

a different PS-b-P2VP (114 kDa, PDI - 1.1, fPS ~0.5) system. The Si pillar arrays are converted to plasmonic nanoarrays bycoating with thin layer of Cr, followed by systematically increasing thicknesses of gold by electron beam evaporation. The nanopillars coated with 5 nm Cr/120 nm of Au were found to result in gold pillars with separations below 10 nm. The optical modeling of the nanopillars proved EM hot spots with increasing intensity and with increasing metal thickness, as expected. This was correlated well with SERS experiments of 1-naphthalene thiol, where the evolution in SERS intensity was found to saturate at 100 nm. SERS-based plasmonic assays of naphthalene thiol show analytical sensitivity down to 74 ppb, low standard deviations in SERS intensities for different concentrations, and feasibility for quantification with large dynamic range. The approach established clear control over structure, property, and function to optimize the final performance of the plasmonic sensor.
