Plasmonic resonance enhanced SERS signal

(1)

Plasmonic resonance enhanced SERS signal

Kabusure Mogasa Kabusure

MSc Thesis May 2019

Department of Physics and Mathematics

University of Eastern Finland

(2)

Kabusure M. Kabusure Plasmonic resonance enhanced SERS signal, 107 pages

University of Eastern Finland

Master’s Degree Programme in Photonics

Supervisors Assoc. Prof. Tommi Hakala

Ph.D. Tarmo Nuutinen

Abstract

In the past decades, plasmonics has played an important role in advancement and development of nano-optics, structural materials and nanotechnology. Plasmonic nanostructures have shown their potential in providing the means of light confinement down to the sub-wavelength scale. This unique behaviour has proved their significance in different research areas such as biosensing and information technol- ogy. Particularly, they can also be harnessed for surface enhanced Raman scattering (SERS). In this work, we study surface enhanced Raman scattering in plasmonic nanoparticle lattices. We show that, radiative coupling can be used to engineer and enhance plasmonic effects. We demonstrate tunability of resonance wavelength via periodicity for both gold and silver nanoparticle arrays. We present two dimensional tunability px and py universal method for creating double resonances. We experi- mentally reported periodicity matching such that, x period 500 nm is well coupled in an efficient way to the Raman excitation wavelength (785 nm), while y period 600 nm is coupled to the Raman shift of the Rhodamine 6G at 890 nm (1510cm⁻¹).

Generally, we observe Raman peaks at 1310 cm⁻¹, 1365 cm⁻¹ and 1510 cm⁻¹. We investigate the effect of inter-particle position from an ordered array against the random samples. We find that the SERS enhancement is well provided by an ordered ordered, however, “Random-1” sample is also able to induce plasmonics hotspots due to the presence of arbitrarily small gaps between particles. Despite the challenges in fabrication method, this work can be useful in different applications specifically in fluorescence enhancement, multimode lasing in plasmonic nanoparticle array to name a few.

Keywords: Plasmonics; surface lattice resonance; nanoparticle arrays; surface enhanced Raman scattering.

(3)

Preface

“If you want to accomplish great things, you first need to conquer your worst fear”

First of all, I would like to thank God for all his blessings and grace that gave me strength to be able to complete this Thesis.

I express my sincere gratitude to my supervisor Assoc. Prof. Tommi Hakala for believing in me and gave me an opportunity to work with him in this project. The moment he agreed to work with me, I felt it was like the beginning of the new chapter towards my new career in scientific researches. I am also grateful to my second supervisor Ph.D. Tarmo Nuutinen for his close guidance and support whenever I needed him. I also wish to thank nanophotonic group, my group discussion mates (Kenneth O. Nyave, Oyemakinwa Kehinde, James Amoan and John Massawe) as well as my colleagues in the 2017 year group of the Institute of Photonics at the University of Eastern Finland for their support and never ending arguments. I am also grateful to Msc. Benjamin Asamoah for spending most of his time with me in experimental work. Great thanks to Khairul Alam for working hard in fabrication whenever we needed him.

I would like to thank my family and my Grandmother (Harriet) for praying every time and keep saying “There is nothing you can not do”. I would be selfish if I do not express how grateful I am to my best friend (Salome) for the love and great effort she showed to make sure I always remember my to-do list as well encouraging me everyday.

Finally, I want to take this opportunity to give my out-most gratitude to everyone who has contributed in one way or another for the completion of my Master’s Thesis work.

Joensuu, the 08th of May 2019 Kabusure Mogasa Kabusure

(4)

Chapter I

Introduction

A new type of secondary radiation was first discovered by Sir C. V. Raman and K. S. Krishnan in 1928; which was later termed as Raman scattering [1,2], and was concurrently revealed by Landsberg and Mandelstan [3]. Since then, there has been great development in the instrumentation field using this technique [4]. Unlike Rayleigh scattering, as the form of elastic scattering of light where scattered photons have the same energy as the incident photons, Raman scattering is an inelastic scattering of light such that the scattered photons have different energy compared to that of the incident photon [5]. The strength of the effect depends on both gaining and loosing of energy when the incident photon interacts with the sample. Two cases were observed which are Stokes and anti-Stokes scattering. Stokes scattering occurs when the scattered photon has less energy compared to the incident photon and the latter occurs when the scattered photon has higher energy compared to the incident photon [6].

Raman conducted an experiment of illuminating a liquid sample with the violet light. The scattered light had the same color as the incident light: This is termed as Rayleigh scattering. His associate K. R. Ramanathan observed some other weak scattering in the same radiation process which was associated with fluorescence [2].

Early January 1928, S. Venkatedwaharan made some discoveries on the purified glycerin. The substance does not appear blue when exposed to sunlight but rather radiates very sparkling green light; This was reported in Nature (March 31, 1928)

‘a new type of secondary radiation’ from the dispersed sunlight in both purified liquid and also Raman signal from dust free air [2]. In his report to Nature, Raman pointed out that 60 liquids experimented showed the same attributes of having shift

(8)

of wavelength from the original wavelength of the incident light. “It is thus”, Raman said, “a phenomenon whose universal nature has to be recognized [7].”

Despite being the most common technique, Raman scattering is a weak phenomenon with a cross - section of around 10⁻²⁹cm²/molecule comparing to other optical processes such as fluorescence whose cross - section is around 10⁻¹⁹cm²/molecule [8]. This had proved to be a drawback in applications where the background fluorescence is substantial. With surface enhanced Raman scattering (SERS) employed where molecules are attached to the plasmonic nanostructures, the Raman signal intensity can be greatly enhanced by several orders of magnitude [8,9]. Hence, this paved a way to counter the low sensitivity problems associated with the Raman spectroscopy [9].

SERS was first discovered by Fleischmann et al in 1974 [10]. In his experiment, he observed significant appearances of Raman scattering on pyridine adsorbed onto a roughened silver electrode surface from an aqueous solution. The task was to develop a chemically specific spectroscopic probe which could be employed in elec- trochemical processes studyin site [11]. The interaction of probes with the analytes resulted to change of optical properties of the probes, for instance; absorption and fluorescence, thereby improving analytical sensitivity [12,13]. It does not only of- fer greater spatiotemporal resolution but also enables one to visualize molecular reactions with bare eyes [13]. His approach was to roughen the electrode, thereby increasing the surface area and in turn increasing the number of adsorbed molecules.

His partners, Jeanmaire and Van Duyne, were focused on the sensitivity of the Ra- man spectroscopy for adsorbed molecules and demonstrated for the first time that resonance Raman spectra can be achieved with sub mW laser powers which makes it easier to employ cheap laser systems for the surface Raman study [14].

Tentatively, Jeanmaire and Van Duyne suggested an electric field enhancement mechanism whereas Albercht and Creighton speculated that different resonance Ra- man scattering of molecules associated with the interaction of metals surface might be the case [11]. Electric field enhancement results from the fields that are associated by localized plasmon resonance of metallic nanostructures which in turn tend to increase the Raman signal intensity [8]. Resonance Raman scattering proved to have greater efficiency to the order of 10⁴ to 10⁶ compared to that of normal Raman scattering and tends to occur when the incident photon energy is almost equal to the electronic transition of the molecule [15]. Several experimental results had to

(9)

be addressed, for instance: The intensities of the bands observed decrease signifi- cantly with increasing vibrational frequency and Raman peak varied with different excitation wavelengths (514.5 and 785 nm) [11,16].

The relative SERS enhancement factors were determined as a function of laser excitation wavelength and the electrode potential. In turn, the charge transfer effect was evaluated from the excitation profiles. Excitation profile is simply the relation between the intensity and the energy released by the molecules [17]. The excitation profiles are not to be confused with absorption profile. However, in some circum- stances, when the molecule does not emit photons, i.e., the energy is dissipated inside the molecule, these two profiles are interchangeabe [18]. Increased intensity of Raman signal is realized if electrode potential is also taken into considerations.

Other factors that are necessary to the same applications include the method of surface preparation and nature of electrolyte [19]. Variation of electrode potential induced substantial change of Raman signal intensity. In the experiment, at first -0.2 V was applied and then the pyridine was removed from the electrolyte. When voltage was increased to -0.6 V, significant increase of intensity was observed [20].

This demonstration shows that the intensity increases with the increase of electrode potential.

The enhancement was remarkably some orders of magnitude, extending tens of nanometers from the surface, which was also attributed by the electrode morphology, structure, composition and size [11]. Thereafter, electromagnetic and chemical mechanism were then put forward to describe the overall process study. The former is based on the enhancement of electromagnetic fields from the metal surfaces while the latter focus on changes of electronic structure of molecules that contribute mainly in resonance Raman scattering [8].

Following the development of nanotechnology and emergence of different research fields such as photovoltaics, photocatalysis and biosensing, plasmonic has played great role in confining and guiding of light beyond the diffraction limit [21,22].

It is highly associated with electromagnetic fields on the surface of the metal [23]

which is also referred to as surface plasmon. The term “Plasmons” are therefore the coupled oscillations between the free electrons at the metal-dielectric surface and the electromagnetic field [8].

Before 20^thcentury, metal nanostructures were used for decorations of glasses and church windows due to their spectacular coloration effects. In 1902, Prof. Robert W.

(10)

Wood conducted an experiment on metallic diffraction grating where he measured optical reflection using polychromatic light. He observed dark fringes in the detector, which he could not figure out what it was at that time [24]. In 1904, Maxwell Garnett explained shiny colors [25] revealed by the metal doped glasses with the help of Drude theory and other principles of electromagnetic light put forward by Lord Rayleigh. Later, Mie theory of light scattering was brought up by Gustav Mie when he experimented on microscopic particles [26]. In 1956, David Pines investigated on multiple energy losses associated by the travelling electrons in the proximity of the metal surface [27]. Based on other research work of gas discharge, he came up with the term ‘Plasmons’. A year later Rufus Ritchie published a study that plasmon modes propagate near the metal surfaces [28]. Conjointly, J. J.

Hopfield suggested the term ‘polariton’ for the coupled electrons and light at the metal-dielectric interface [29].

Subsequently, Ritchie and coworkers made further studies of Wood’s experiment on metal gratings in relation to surface plasmon behaviour [30]. Furthermore, in 1968, A. Otto et al. described the behavior of surface plasmons on metal films which in turn brought progress to researchers [31,32]. Until then, there was no relation of surface plasmon and metal nanoparticles. However, in 1970 Uwe Kreibig and Peter Zacharias performed a study on optical properties of gold and silver, and for the very first time they were able to link these properties with the behaviors attributed on surface plasmons. [24]

Plasmonics are categorized into two main ingredients which are surface plasmon polaritons (SPPs) and localized surface plasmons (LSPs). The term ‘surface plasmon polaritons (SPPs)’ was first introduced by Stephen Cunningham and his fellow scientists in 1974. SPPs result when photons couple to the conduction electrons in planar metal -dielectric interface while LSPs involve localized optical fields confined to nanoscale dimension [8]. Localized surface plasmon resonances (LSPRs) occur when the localized surface plasmons oscillate with the same frequency as the incident electromagnetic light. These LSPRs can be utilized in surface enhanced Raman spectroscopy. [33] Periodic arrangements of individual nanoparticles can couple to produce collective surface lattice resonances (SLRs).

Both LSPRs and SLRs depends on the morphology of the particles such as shape and size. In this case, resonance occurs when the localized surface plasmon resonances (LSPRs) are in the same phase for every nanoparticle in an ordered array

(11)

[34]. LSPR has been employed in many applications for instance, nano-photonics and biosensing [8]. Recently, there has been technological advances where LSPR continues to play a greater part in sensing with distinctive applications such as in multiplexing and integration with microfluidics [35].

In this work, we are going to investigate the lattice periodicity matching Px and Py of an ordered array of gold nanoparticles with the wavelength of the Raman excitation and Raman line of the analyzed molecule, respectively. This work is di- vided into seven Chapters. This chapter is followed by aims and hypothesis of the work. Chapter 3 comprises of theoretical background on plasmonic, Raman scattering and surface enhanced Raman scattering. Chapter 4 describes methodology and all related measurements and procedures used to prepare the samples and their characterization. Chapter 5 outlines the results of the work presented. Chapter 6 presents thorough discussion and analysis of the experimental results. Lastly, we conclude in summary and any future prospect of the work is discussed.

(12)

Chapter II

Aim

It has been demonstrated in the article. [36] that the resonance wavelength resulted in the nanoparticle arrays is the product of the periodicity and the refractive index of the substrate. In our work, we are going to fabricate two dimension array of metal nanoparticles with lattice periodicity Px and Py, in the sense that Px and Py

are matched to the laser excitation wavelength and Raman shift of the molecules, respectively.

(13)

Chapter III

Theory

3.1 Plasmonics

Plasmonics is the study of interaction of light and metal nanoparticles [8]. In the past decades, this field has played great role in advancement and development of Nano- optics, structural materials and nanotechnology [34]. Noble metals can be nurtured into structures of nanoscale dimension. In such structures, electromagnetic waves can also be manipulated beyond diffraction limit and down to the sub-wavelength confinement of light [37]. This is possible because the metallic nanostructures are able to localize and guide electromagnetic light with high fidelity [34,37]. One thing about metallic nanostructure is that, you can change their optical properties by varying their sizes, shapes and morphology which in turn makes them the best candidate to a variety of applications such as in optical imaging, telecommunications, biomedicine and photovoltaics [8,34].

3.1.1 Surface plasmon resonance

Surface plasmon (SP) refers to the metal nanoparticles (NP) interaction with light.

Apart from quantum mechanical effects, Maxwell equations also describe plasmons.

A classical driven-damped oscillator model can provide an intuitive picture of plasmon resonances [38]. A metallic nanoparticle can be described as a lattice structure with free electrons that move inside the nanoparticle [39]. When an electromagnetic light impinges on the nanoparticle, charges (both positive and negative) will experience an electric field which in turn, tend to migrate them onto the nanoparticle surface forming an electric dipole. Consequently, an electric field produced by

(14)

these dipoles tend to oppose that of electromagnetic light which results to restoring electrons to their original position as illustrated in Figure3.1 [40].

Electron displacement varies with electric dipole and so is the restoring force.

This dependence is no different from the linear oscillator system [39]. On the other hand, if the field is removed electrons will continue to oscillate with some frequency known as resonant frequency; in surface plasmons, this is referred to as plasmon resonance. At surface plasmon resonance frequency, the maximum amplitude of electron oscillation is observed [41–44].

Figure 3.1: Light interaction with a metallic NP. Adopted from [39].

SPR creates a dip in the transmission profile of the illuminating light. The SPR band intensity depends on material’s morphology, optical environment, size and shape and is well described by the Mie Theory [45]. Mie theory can be employed for all particle sizes. SPR can thus be defined be described by the equation

C_ext= 24π²R³ε^3/2m

λ

εi

(εr+ 2εm)²+ε²_i (3.1) where Cext is the extinction cross section, λ is the wavelength of the illuminating light, ε is the complex permittivity of the metal defined as ε= εr(ω) +iεi(ω),εm is the permittivity of the surrounding medium which is equal to the square of refractive index, η²_m.

The real and imaginary part of the permittivity are practically related to SPR position and bandwidth, respectively. The SPR resonance is observed under the condition when εr(ω) is relatively equal to -2εm. Noble metal nanoparticles show

(15)

significant SPR band in the visible region, however, very rare metals have UV resonances [46,47]. SPR band for gold nanoparticle can be observed around 520 nm in the visible region. However, this SPR wavelength may change as illustrated in Fig 3.2since SPR band is particle size dependent [48]. For instance, particle larger than 10 nm may tend to red shift the SPR wavelength as well as the intensity [45].

Figure 3.2: Absorption spectra of Au NPs with different size. Figure adopted from [45].

3.1.2 Surface plasmon absorption and scattering

Electron amplitude can be observed indirectly by relating electron oscillation with the change of kinetic and electrostatic energy [49]. Based on the energy conservation law, internal change of energy should be equal to the change of external energy. This result to decrease of the reflected light intensity after exciting surface plasmons in the nanoparticles [22]. In other way, light extinction depends on electron oscillation.

Light extinction can be defined as the sum of absorption and scattering cross section. Light absorption results when the photon energy is dissipated as heat into the particle while light scattering occurs when the light interact with electron oscillation

(16)

leading to radiation of photons. The scattered light can either be of the same wavelength (Rayleigh scattering) or of different wavelength (Raman scattering) which is dependent on molecular vibrations and rotations. Altogether, light interaction with nanoparticles can also lead to change of propagation vector, momentum, and energy [39]. Figure 3.3 illustrates the processes.

Figure 3.3: (left) Description of absorption cross section (right). Illustration of optical processes. Adopted from [39].

The surface plasmon absorption, scattering and extinction efficiencies are directly related to their cross sections. The description of these efficiencies can be evaluated using Mie theory [41]. According to El-Sayed and co-workers employing Mie theory [50,51], the optical absorption and scattering are highly dependent on the particle size as elucidated in Fig3.4. Besides the dependence on size, shape, and the dielectric material, extinction depends also on the polarization state [43].

(17)

Figure 3.4: Optical tuning by sizes and shapes. Spectra of the efficiency of extinction Q_ext (green solid), scattering Q_sca (black dotted) and absorption Q_abs (red dashed) for gold nanospheres. (A) For a 20 nm Au NP, the total extinction is dominated by absorption. (B) The scattering is observed when the particle size is 40 nm. (C) When the particle size is 80 nm, both absorption and scattering are observed in the same proportion, and for polystyrene nanospheres in (D) when particle diameter is 300 nm. Adopted from [45].

3.1.3 Models and numerical approach of SP

Classical description provides a better understanding on surface plasmons as it was discussed above. However, numerical analyses are non-trivial as they demand the use of Maxwell’s equations to describe NP with proper boundary conditions. The analytical solution can be provided with the use of Mie theory developed by Gustav Mie [41]. Mathematically, extinction, scattering and absorption cross section of

(18)

material are given by the following equations

C_ext= 2π k²_out

∞

X

l=1

(2l+ 1)ℜe[a_l+b_l], (3.2)

Cscat= 2π k_out²

∞

X

l=1

(2l+ 1)(|al|²+|bl|²), (3.3)

Cabs =Cext−Cscat, (3.4)

where

al = mϕl(mx).ϕ^,_l(x)−ϕ^,_l(mx).ϕl(x)

mϕl(mx).η_l^,(x)−ϕ^,_l(mx).ηL(x); (3.5)

bl= ϕl(mx).ϕ^,_l(x)−mϕ^,_l(mx).ϕl(x)

ϕl(mx).η^,_l(x)−mϕ^,_l(mx).ηl(x) (3.6) k is the propagation vector in the dielectric medium, x =|k| ×R, R is the nanoparticle radius, m =ηin/ηout,ηin and ηout are the complex refractive index of the metal and that of dielectric medium, respectively, ϕl and ηl are the cylindrical Bessel - Ricatti functions. There are several approaches that are employed to determine the light absorption based on the size and shape of NPs. These will be discussed in the following subsections.

Discrete dipole approximation

The fact that NP size is rather small compared to the wavelength justifies using discrete dipole approximation (DDA) in calculating optical properties of a material.

For DDA to be valid, and if SPR band is observed in the visible region, the size of the particle should not exceed 50 nm [48]. The electric field created inside the particle is constant throughout the volume as described in the Fig3.5and it follows quasi-static approximation.

DDA provide information about extinction, absorption and scattering cross sections (C) of the metallic nanostructure and normalized in respect to its geometrical cross section πa²_{ef f} which in turn, give efficiency Q = C/πa²_{ef f} where aef f is the effective radius of the nanoparticle [52].

(19)

Light

Light (a)

(b)

Figure 3.5: Light interaction with NP of different sizes(a) subwavelength NP (b) Large NP comparable to the wavelength of the light. Adopted from [39]

The effective dielectric function theories

This technique aims at changing properties of a material for instance; dielectric constant (εr+iεi) and surrounding dielectric constant εm by replacing heterogeneous medium with homogeneous medium, thus having effective dielectric constant εef f. There have been many theories proposed that depends with homogeneous optical responses [44], but the most common are discussed below. The first approach was explained by Newton, he described effective dielectric as the mean value of NPs and that of medium was associated with volume of the NP (In this case spherical NP) [53]. Unlike Newton approach, Maxwell Garnett enriched his theory by maintaining uniform dielectric polarization using homogeneous material under illuminating light [54–57]. As described in the quasi-static approximation, Eq. (3.7) shows the resulting polarizability which holds for a particle with diameter less than λ/10 [58].

α(ω) = 4πε0a³ ε(ω)−εm

ε(ω) + 2ε_m (3.7)

where a is the particle diameter, ε(ω) and εm are the relative permittivities of the metal and dielectric medium surrounding the particle, respectively [59]. Maxwell Garnett approach is more advantageous than the dipole approximation because inter-particle interaction can be taken into consideration with a significant anal-

(20)

ysis [60,61].

3.2 Types of plasmon modes

There are two categories of plasmon modes namely; Surface plasmon polaritons and localized surface plasmons.

3.2.1 Surface plasmons polaritons

These are propagating plasmons bound to the interface of a metal and dielectric medium [34]. Surface plasmons polaritons (SPPs) can be intuitively understood as in-phase fluctuations of collective electrons in the metal associated with electromagnetic field in both media [62]. They propagate along the surface and their properties vary with the surface environment [63]. Figure 3.6 shows the illustration of a SPPs in a metal-dielectric interface.

Figure 3.6: SPP in a metal-dielectric interface. (a) collective electron fluctuations associated with electromagnetic field. (b) SPP mode shows the expo- nential decay at the metal-dielectric interface. (c) Dispersion relation curve for SPPs. It shows the momentum mismatch as one of the challenges for exciting SPPs. Adopted from [64].

From Figure 3.6(a), assuming that the dielectric medium can either be air or glass, electromagnetic field varies about half of the wavelength of the light for the dielectric medium and it also varies with skin depth of the metal as illustrated in Figure 3.6(b). The momentum mismatch between SPPs and the photons in Figure3.6(c) reduce the efficiency in exciting SPPs, however, with proper correction

(21)

methods such as the Kretschmann [32] technique and Otto configuration [31] or grating coupler [65], the excitation of SPPs can be possible. Excitation of SPPs depends on both incident angle and wavelength [49]. With a proper selection of incident angle, the excitation of SPPs emerge along with the narrow resonance of the spectral width∼50 nm (full width at half maximum) in the reflectance spectrum.

Despite the challenges associated with the SPR excitation, [66,67] SPR has proved to have significant sensitivity and become the major key in the development of biosensing techniques [68,69].

3.2.2 Localized surface plasmons

These are collective oscillations of free electrons in a metallic nanoparticle and associated oscillations of the electromagnetic field. Tuning of nanoparticle properties such as size, shape, morphology and optical environment defines the spectral width and resonance frequency [44,70]. Resonance frequency occurs in the visible to near infrared region for cognaic metallic nanostructures. In contrary to SPPs, LSPs have proven largely in demand for applications that involve light manipulation at nanoscale dimension, [71–76] nanolensing [72,77] and nanolasing, [78,79] nanotweez- ers, [80] and improvement of photovoltaic devices [81]. These unparalleled features that LSPs have over SPPs are simply because momentum matching is not needed in exciting LSPs [82].

Resonance linewidth for a Au nanoparticle is approximately 80–100 nm full width at half maximum for LSPs and that of SPPs is around 50 nm. In that sense, LSPs are limited to some applications due to its broad spectral width resonance [34]. In spite of the fact that aluminium is not one of the cognaic metals, nevertheless, it can be employed to applications that depend on the resonance in the UV-visible region [83–85]. Localized surface plasmon resonance can be described using quasi- static approximation as shown in Equation. (3.7).

3.3 Surface lattice resonances in nanoparticle array

3.3.1 Surface lattice resonance

Extreme narrow resonance is the key and major difference between LSPRs and surface lattice resonances (SLRs). This feature emerge when nanoparticles are arranged in an ordered array whereby, once the periodicity of the array is the same or in some

(22)

order with the resonance wavelength of the particle, the collective resonance is produced from the interaction between the diffractive orders (DOs) and the LSPRs of individual particle. This collective excitation is known as surface lattice resonance (SLR) [86–90]. In addition to having very narrow resonance linewidth down to few nm (3–4 nm), SLRs on metallic nanostructure array exhibit dispersion relations that can be attributed with different polarizations and angle resolved incident light.

These unique characteristics play an important role in tailoring lght fields at the nanoscale [91].

Assume we have two-dimensional square geometry array of nanoparticles and allow an incident light to impinge at an angle of incidence normal to the lattice plane. The diffraction pattern observed is illustrated in Figure 3.7. If the field is y-polarized and the component of the propagation vector of the incident light is defined as k^inc = k^inc_z z, the interference either constructive or destructive resultˆ from the diffracted waves produced in the nanoparticle array. Here, only the first diffractive orders in x direction are considered. The momentum kick provided by the lattice in x direction is given by |kx| = 2π/p, where p is the lattice periodicity.

Consequently, the total wave vector isk =kx+kinc =kinc±2π/px. [92]

k

^inc

k(1,0)

k(0,0)

k(-1,0)

k(1,0) ,

k(-1,0) ,

k(0,0) ,

E

x z

y

Figure 3.7: Diffraction orders up to the first order from two dimensional square geometry array of nanoparticles. Adopted from [92].

There are two factors that affect the performance and the length of the propaga-

(23)

tion distance of SLRs in the periodic structure which are ohmic and radiative losses.

These losses occur in the metal-dielectric medium interface. Intensity of losses can be identified based on the linewidth of the mode. The narrower the line, the lower the losses. At the same time, the sharper the dip in the absorption spectrum is characterized with higher extinction efficiencies. The optical response of a given nanoparticle can be modified by the proximity of other nanoparticles: in this case the particles can be considered coupled via electromagnetic fields [34]. There are two forms of coupling, namely; Near field and far field coupling. Near field coupling comes from the evanescent, non radiative component of the electromagnetic field and becomes dominant when the interparticle spacing is well below the wavelength of the electromagnetic field. It can result in change of spectra due to the hybridization of the plasmonic modes [93–97]. If particles are separated by large distance, then radiation can induce coupling (far field coupling). In an ordered arrays of nanoparticles with periodicities approximately equal to radiation wavelength, the far field coupling can result in hybridization of diffracted order and the single particle resonances termed as SLRs.

SLRs can be excited by direct illumination regardless of its polarization state.

Two modes namely; transverse electric (TE) and transverse magnetic (TM) mode are described with respect to the plane of incidence. In the TE mode, dipoles have fields oriented in a direction normal to the propagation constant ky of the incoming light as illustrated in Figure3.8 (a), while in the TM mode, the induced dipoles oscillate parallel to the direction of ky as elucidated in Figure 3.8 (b). If TM-polarized light of angle other zero angle is used to excite nanoparticle array, dipoles are induced also in the z-direction. Both TE and TM modes exist in the lattice plane simultaneously, can be measured with respective polarization state by considering the direction of the wave vector in the lattice plane [92].

(24)

x y z

x y

z

k

inc

k

y

E

inc

k

^y

k

inc

E

inc

(a) (b)

Figure 3.8: Excitation field with different polarization state(a) TE polarization (b) TM polarization. Adopted from [92].

3.3.2 Empty lattice approximation

In surface lattice resonances, the diffracted orders produced by the ordered periodic arrangement of nanoparticle couple with the localized surface plasmon resonance (LSPRs) [34]. To understand optical modes and the effect of the periodicity, we need to consider the dispersion relation and what kind of treatment is necessary.

Consider a homogeneous medium with refractive indexη, the dispersion relation [98]

is given by

k= ηE

~c (3.8)

whereE is the energy of the photon. If we consider k²₀ =k_x²+k_y²+k_z² andkx orky is greater thank0, then kz has to be imaginary. In space dimension bounded by kx,ky

and E, propagating plane wave form a cone with edges defined byk_x²+k²_y =E²/(~c)² and evanescent wave exist outside in the vicinity of the cone as shown in Fig3.9.

(25)

Figure 3.9: Dispersion relation of light ink_x,k_y andE-space showing modes that propagate inside the cone in xy plane and outside the cone (evanescent modes). Adopted from [99].

Now, if we have a periodic structure with periodicity px and py and there exist wave equation in this system, eventually its solutions will also be periodic [40]. All modes defined by an integer multiple of the reciprocal lattice vectors G~x = 2π/pxxˆ or G~y = 2π/pyyˆ are similar (Bloch waves). The dispersion of light in kx, ky and E-space will have four extra cones displaced from the origin as depicted in Fig3.10.

The periodic structure exhibit complex dispersion relation because there are several points of intersection. To have a meaningful system, we need to have only one scatterer that would couple all modes simultaneously. The next section is going to give fully description on point dipoles. [99]

(26)

Figure 3.10: Dispersion relation of light ink_x, k_y and E-space showing the first order scatterer and other four modes defined by the reciprocal lattice vectors G~_x = 2π/p_xxˆ and G~_y = 2π/p_yy. In this case periodicityˆ p_x and p_y are 570 nm, refractive index is 1.52 and the wavelength is 375 nm by period.

Adopted from [99].

3.3.3 Lattice of point dipoles

When we induce point dipoles as the scatterers and allow the x-polarized light to travel inside the periodic structure of an infinite lattice, the radiative effect will be in y-direction. Y direction radiation results to the first order diffracted order whose dispersion relation is illustrated in Fig 3.11. Two modes are observed; TM and TE modes. TM mode dispersion is at ky = 0 is given by the equation E/~c=

±p

k²_x+G²_y, whereGy = 2π/py is the momentum kick provided by the lattice. TE mode is at kx = 0 is given by the equation E/~c = ky ±Gy. When kx = 0 and ky = 0, this point is referred as the Γ - point where TM and TE mode intersect. [99]

(27)

Figure 3.11: Dispersion relation in a periodic system in k_x, k_y and E-space.

(a) The modes are shown with highlighted dashed lines with either k_x = 0 or k_y = 0. (b) TM modes and (c) TE modes. The crosscut gives us more information about the periodic system. Adopted from [99].

3.4 Optical response of noble metal nanoparticles

In this section, optical response of nanoparticles will be discussed. This will give us the insight on modelling properties of plasmonic lattice analytically. On top of that, two basic properties namely; Permittivity and polarizability with respect to the effect of electric field account to the response of noble metals properties. Only the case of polarizability of spherical nanoparticle will be discussed. [92].

3.4.1 Permittivity

Permittivity describes the response of the electrical charges of the material. Optical properties of the metal are described by using frequency dependent dielectric function. Dielectric function is described by the relative permittivity in respect with that of vacuum as given in the Equation: ε =εrεo. Even though, dielectric function for some dielectric materials such as air or glass are positive and real, other metals for instance, noble metals exhibit dielectric function with negative real part as shown in Equation. (3.9) [40].

(28)

εm(ω) =−εreal(ω) +iεIm(ω) (3.9) Plasmonic properties of noble metals originate from their large negative dielectric function. Dielectric function is related to the refractive index by the equation n =

√η, however, for the strong LSPRs response, the real part of the dielectric function should be large. [92]

In attempting the free electron approximation, we start with the real bulk metals and model the motion of electron by considering the Equation. (3.10) where the dielectric function is contributed by unbounded electrons. The free electrons condition is explained by realizing the Drude free electron model.

mδ²x

δt² +mγ₀dx

dt +kx=eE(t) =−e

mE(t) (3.10)

wheremis the effective mass of an electron,e is an electron charge,E(t) is the time varying electric field, x is the electron displacement andγ0 as a damping constant.

In the free electron gas, the third term from Equation. (3.10) becomes zero and the solution to the second differential equation is given by

x(ω) = eE(ω)

m(iωγ0+ω²). (3.11) In turn the polarization per unit volume can be written in the form

P(ω) = −Nex(ω) = Ne²E(ω)

m(iωγ0+ω²) (3.12) where N is the free electron density. The relation between polarization and the absorption coefficient can be defined in the equation

P(ω) =Nα(ω)E(ω) =ε0χ(ω)E(ω) (3.13) and therefore, the susceptibility is calculated as follows

(29)

χ(ω) = −Ne²

εom(ω²+iΓω) (3.14)

Now, the dielectric function can finally be deduced [100,101] to equation

ε= 1− ωp

iωγ₀+ω², (3.15)

where ω²_p = ^{N e}_ε ²

0m is the plasma frequency of the bulk metal. In this case ωp depends only on the charge density N. From Equation. (3.15), the real and imaginary part of the dielectric function can be represented as

εreal= 1− ω_p² γ₀²+ω² ε_Im= ω_p²γ0

ω(γ₀²+ω²). (3.16)

In high frequency case where γ0 << ω, eq. (3.16) becomes

εreal= 1−ω_p² ω² εIm = ω_p²γ0

ω³ . (3.17)

Since the LSPR occurs when ε_real+ 2ε_m= 0, then ω_{LSP R}= ωp

√2ε_m+ 1, (3.18)

where ωLSP R is the angular resonance frequency of the LSPR.

Even though, gold and silver has almost the same plasma frequency, their LSPR frequency for the spherical shape of size in the order of 50 nm occurs at 530 nm and 420 nm, respectively [48,102]. This difference of resonance wavelength originates from other contributions in the dielectric function from the inter-band electronic transitions. The free electron gas model holds only at lower frequency regime to calculate the resonance wavelength of the metallic nanoparticles; therefore, a slight

(30)

modification is required to impose that drawback. The equation is then expressed as

ε= 1− ωp

iωγ0+ω² +χb, (3.19)

where χb is the complex interband susceptibility, in addition to that, ωLSP R [101]

can be rewritten as follows

ωLSP R= ωp

√2εm+χb+ 1. (3.20)

In the free electron gas model, the bandwidth (Γ) of the plasmonic band is the same asγ0 [101]. However, for the case of noble metal, Γ is expressed as

Γ =γ0

s 1 +

2χ2ωp

(1+χ1+2εm)√ 1+2εm

γ₀ . (3.21)

The ratio of wavelength resonance to the dielectric function of the medium can be referred as the sensitivity S and given by the equation

S = dλLSP R

dεm

= λp

√2εm+χb+ 1. (3.22)

This can also be shown in Figure 3.12, where the contribution of free and bound electrons affect the real and imaginary part of the bulk gold dielectric function is elucidated.

(31)

Figure 3.12: Effect of contribution of free and bound electrons to the real and imaginary part of the gold bulk dielectric function. The experimental results are adopted from Johnson and Christy where they concluded that the inter-band transitions present in metals at some frequencies for instance; in visible and higher frequency regimes limit the validity of the Drude model [103].

Adopted from [101].

(32)

3.4.2 Polarizability

Ability of an atom to induce dipole moment in the presence of electric field is known as polarizability. Practically, polarizability increases as the number of electrons per unit volume increases. In the case of nanoparticles, the size and shape also affect the polarizability. Unlike non-spherical nanoparticle, spherical nanoparticle has polarizability that is equal in every direction (isotropic), in that sense, its effect is trivial to prevail. However, having non-spherical nanoparticles poses more challenge since it is difficult to analytically recover the analytical expression for the polarizability. In ordered lattices of nanoparticles, however, the nanospherical shape can result in for- mation of band gaps [104] . As analyzing non-spherical particles with the Equation.

(3.7) is not feasible, therefore, a new model must be established [92].

Assume we have a rod-shaped nanoparticle, the shape can be regarded as ellipsoid with the semi axis radii such that a₁ < a₂ < a₃. The general formula for ellipse is given by x²/a²₁+y²/a²₂ +z²/a²₃ = 1. Now, if this ellipsoid is placed in the electric field such that the principal axes (i = 1, 2, 3) lie in parallel with axes (i = x, y, z), the static polarizability can be expressed by the equation [43]

α_i = 4πεa₁a₂a₃ εm−εh

3εm+ 3Li(εm−εh), (3.23) where εm and εh are the relative permittivities of the material and isotropic homogeneous medium, respectively. Li is the geometrical factor given by

Li = a1a2a3

2

Z _∞

0

dq (q+a²_i)p

(q+a²₁)(q+a²₂)(q+a²₃). (3.24) The geometrical factors satisfy to the condition of the normalization P

iLi = 1 so that it can be reduced to the sphere if Eq. (3.24) givesL1 =L2 =L3 = 1/3

The Equation. (3.23) holds if the size of the nanoparticle is of order 50 nm.

Beyond that, further modification is required to account radiative damping and dy- namic depolarization [105]. A modified long wavelength approximation (MLWA) is employed by introducing an effective polarizabilityα^{M LW A}=α^static/(1−²₃ik³α^static−

k²

a¹α^static), where a1 is the semi axis radii of the rod shaped nanoparticle,k =ηk0 is the wave vector in the homogeneous surrounding medium of refractive index η[88].

(33)

3.5 Raman scattering

Raman scattering is an inelastic scattering where the molecules of the sample are excited to higher energy levels. Inelastic effect occurs when scattered photons have different energy from the photons interacting with it [106,107]. The effect provide fingerprint of the molecule which in turn can also be used in qualitative and quanti- tative analysis. Qualitative analysis result to sample identification while the latter, gives the content of the analytes in the sample [108].

Raman scattering was discovered by C. V. Raman and coworker K. S. Krishnan in 1928 in India. The idea was first demonstrated theoretically by Smekal in 1921.

Consequently, the same observation was made by G. Landesberg and L. Mandelstan in Russia [109]. The technique that allows one to observe vibrational and rotational mode of the molecular structure of a material is referred to as Raman spectroscopy.

3.5.1 Classical description of Raman scattering

When electromagnetic light interacts with a molecule, the electric field induce dipole moment on the molecules [110]. This induced dipole moment is a frequency dependent and it also creates the field which opposes the electric field producing it leading to electric olarization. This kind of interaction set the molecules into vibrational mode or even higher vibrational mode that leads to significant excitation [111].

A molecule has a very short time period of order of nanosecond before it starts descending to the ground state. Electron relaxation is associated with emission of light of frequency smaller or larger than that of incident light. This change of frequency result to Raman scattering. Point to note, the classical description holds only for vibrational molecules. [111]

3.5.2 Quantum mechanical description of Raman scattering

Unlike classical theory, quantum theory describes the possession of energy such that electron occupies energy levels and later subdivided into vibrational states. In this case, we apply perturbation theory to the Raman scattering [111–113] and electrons follows the Pauli’s exclusion principle where the total spin is zero. When electrons are excited, Pauli’s exclusion principle is affected in turn creates the perturbation into the system. This subsection focuses mostly on the vibrational component of the excitation.

(34)

When electric field interact with electrons, two forms of excitations occur namely;

inter-vibrational states excitation and intra-vibrational states excitation. The former involves exciting electrons to higher energy levels while the latter involves exciting electrons to lower energy levels which in turn involves loss of energy. [113]

Conservation of energy is also accounted during transitions such that, the difference of energy in the initial state and that of final state should correspond to the energy of the incident light [114,115]. Light scattering can be categorized as either elastic or inelastic scattering. In the elastic scattering, there is no change of energy, i.e., scattered photon has the same wavelength as that of the incident photon regardless of its polarization. This form of scattering is called Rayleigh scattering.

In the inelastic scattering, the scattered photon has different energy from that of the incident photon. The energy difference is associated with transition between two states in the molecules. Again, this form of scattering is referred to as Raman scattering. [111]

3.5.3 Raman scattering processes

Raman scattering, not to be confused with fluorescence effect, it occurs instanta- neously while fluorescence is the two-way process (it starts with the excitation of electrons followed by emission with a finite lifetime). Raman scattering falls under two categories; the one with energy less than that of incident photon is known as Stokes Raman scattering and the one with energy higher than that of incident photon is called anti-Stokes Raman scattering. Figure3.13desribes the energy transitions for both Rayleigh and Raman scattering. Basically, Raman effect is observed from Stokes process because the anti-stokes process is much weaker compared to its rival the Stokes process. Energy lost by the photons during scattering process is called Raman shift and it is denoted by ∆ER =EL−ES where EL is the incident photon energy and ES is the energy of the scattered photon. In that case, Stokes process result to positive energy while anti-Stokes process gives negative energy. Ra- man shift is expressed in wavenumber (cm⁻¹) and Raman spectrum is represented as Raman intensity against wavelength. [113,114,116]

(35)

Virtual State

Energy

Rayleigh Scattering

(elastic)

Stokes Scattering

Anti-Stokes Scattering

Raman (Inelastic)

hv₀ hv₀ hv₀ hv₀^-hv_m hv hv₀₊hv_m

0

0₊hv_m E

E0

Figure 3.13: Jablonski diagram showing vibrational Energy Transitions for Rayleigh, Stokes and anti-Stokes processes. Adopted from [117].

3.6 Surface enhanced Raman spectroscopy

Now that we have provided succinct knowledge of Raman scattering and surface plasmon resonance, it is appropriate to introduce the concept of SERS and how they are related to it.

Surface enhanced Raman spectroscopy (SERS) is an optical spectroscopy technique that employs local field enhancement from localized surface plasmon resonances of the metallic nanostructures to increase the Raman signals of the adsorbed molecules [116]. The local field enhancement depends on size, shape, optical environment, and morphology of the nanoparticle [8]. Therefore, one can tune optical properties of nanostructures based on the need of application by altering those parameters [118].

The enhancement in SERS was first observed by Fleischmann et al. [10] in 1974 on the experiment involving roughened silver electrode and they noticed that, the increase of Raman signal was contributed by the significant large number of molecules

(36)

adsorbed on the electrode. In 1977, coincidentally, Jeanmarie et al. [14] and Alber- cht et al. [119] described that an enhancement was caused by localized optical field from metallic nanostructure. From that time, SERS became vast topic in different research areas such as in molecular detection [120] and in biosensing [121–123].

3.6.1 SERS enhancement factor

It is very important to know how greatly the Raman signal can be enhanced. This can be determined by the enhancement factors. There are two categories that falls under enhancement factors namely; Electromagnetic enhancement factor and Chem- ical enhancement factor.

Electromagnetic enhancement: This is the main contributor of the enhancement in SERS. It results from the coupling of both the incident electromagnetic light and radiated signal with the local optical field from the SERS substrate. Electromag- netic (EM) enhancement is then subdivided into incident field and Raman signal enhancement. EM enhancements depends largely on local field enhancements from metallic nanoparticles as depicted in Fig3.14. One more interesting feature is SERS hotspots, [124] which are areas in proximity to the nanoparticles. Raman signal is enhanced largely for molecules around the hotspots. In most cases the analytes are adsorbed on the surface either through physisorption or chemisorption [116].

Incident light (Laser) Rayleigh scattering

man scattering

anoparticle Analyte

Strong EM

Figure 3.14: Schematic showing SERS process. Adopted from [125].

Chemical enhancement: Let alone the existence, the definition of chemical enhancement is still subtle and its contribution to the enhancement factor is rather small compared to the electromagnetic enhancement. It is simply the interaction between an isolated nanoparticle and the analytes on the surface of the metal-dielectric interface [126]. This interaction produces orbital resonance excitation [127] which is

(37)

somehow associated with transfer of charges from analytes to the nanoparticles and vice versa. Furthermore, the transfer of charge induces the change of polarizability which lead to the increase of Raman signal [128].

The enhancement factor for SERS is usually in the order of 10⁸−10¹⁰[129,130]

and ≈ 10⁰ −10³ for the electromagnetic and chemical enhancement mechanisms, respectively. The SERS enhancement factor depends on the following parameters;

SERS substrate, Detection setup (like scattering configuration), analyte adsorption properties (such as adsorption efficiency and analytes concentration on the surface) and laser excitation properties (such as wavelength and angle of incidence). Even though, the large contribution of enhancement factor is led by electromagnetic enhancement mechanism, the fact of having chemical enhancement explores the unique properties of the analyte and nanoparticles interaction from vibrational mode excitation [116].

(38)

Chapter IV

Methodology

To study and describe the behaviors of metal nanoparticles relies on the effective production of nanoparticles into desired shape and size. The production of metal nanoparticles can be categorized into two methods; “top down” or “bottom up” [38].

Top down methods such as electron beam lithography [131], usually starts from bulk metal and go through some processes until nanoscale entities are obtained while bottom up methods include techniques such as chemical synthesis or self- assembly [132]. In this work, we are mainly interested in top down method (electron beam lithography) and further discussion on the processes involved will be provided.

4.1 Samples

The samples used in our work are cylindrical nanoparticles (gold and silver) arranged in a two-dimensional array having a constant inter-spacing distance (periodicity along x and y) and the random set samples, in which the nearest neighbour distance was set larger than 50 nm for random 2 sample and no condition for random 1 sample. This design helps to demonstrate the minimum inter-spacing distance that would give good coupling between the nanoparticles. The sample is comprised of two layouts of nanoparticles array namely; 5 x 11 matrix of arrays, in which px is scanned from 500 to 540 nm in step of 10 nm and py is scanned from 580 to 620 nm in step of 4 nm. In addition, we fabricated 3 arrays of random samples with non-uniform periodicity but with same particle density as the samplespx = 500 nm, py = 580 andpx = 520 nm, py = 600 and px = 540 nm, py = 620, the overall layout is illustrated in Fig 4.1. Nanoparticle diameter is set to 110 nm. The samples are prepared in the cleanroom facilities at the University of Eastern Finland. For the

(39)

sample fabrication, several processes are required. In this section we are going to introduce some basic processes, followed by the whole fabrication process.

Figure 4.1: Sample design layout. The right figure shows one of the nanoparticle array where the first and the second number are px and py period. The text font is bigger for clarity.

4.1.1 Spin coating

This is the process of depositing resist onto the substrates. The device that is used for spinning is known as the spin coater. It has two main parts: a rotating stage which include suction hole that provides fastening mechanism to the sample and a driving unit. Spinning is done such that the viscous liquid (resist) is poured onto the substrate, then the substrate is spun at a rapid speed around 1000 - 5000 RPM for 30 to 60 seconds to produce uniform distribution of the resist onto the substrate. The higher the speed, the smaller the thickness of the layer, however, the temperature is above resist glass transition temperature, where the smaller chains on polymer are bound together. Electron beam lithography then breaks these chains.

(40)

4.1.2 Electron beam lithography

Electron beam lithography (EBL) is the resist lithography method in which the exposure is done by direct electron beam writing. It creates the predefined patterns in the resist which are later transferred to the substrate. Unlike other lithography tools, EBL can produce features beyond diffraction limit of light into the nanoscale regime. The current photo-lithography tools uses Deep Ultraviolet (DUV) lasers of wavelength 248 nm and 193 nm to create features of size down to 50 nm [133,134].

The resolution in EBL depends on the resolution of the resist used, in turn, the best resolution one can get when using polymer resist is around 20 nm and with inorganic resist is around 5 nm [135]. In this work, the photo resists used are PMMA (poly- methyl-methacrylate) and AR-P6200.18.

The EBL exposure can be done through the following steps: Firstly, SEM focus- ing is done. Secondly, beam current is set which varies from 1 pA to 10 nA based on the SEM specifics [136] and the used aperture. Third step is to define the exposure pattern for each layout including their exposure parameters such as step size, dwell time, area dose and the writefield size associated with the magnification order as well as electron beam energy. The dwell timeT_dwell defines how long the beam stays in one spot. Step size defines what is the minimum distance between neighbouring exposure spots. Area dose is then simply the current×T_dwell×number of spots per unit area and it has the units of Coulomb/cm².

The area where the electron probe moves during the exposure is called the writefield. The size of the writefield is set by the maximum deflection provided by the electron beam lenses and the control electrons. The smaller the writefield, the more the writing. Following this, small feature size should be exposed by small writefield and large ones should be exposed by larger writefield. The area of the resist which is chemically affected by the electron probe is known as exposed area [137]. Different procedures such as imaging and measuring dimension of the dose test structures are considered to optimize the final dose value. It is more effective to perform the dose test from time to time when you are working with new resist or the same resist, this helps to overcome any changes made in SEM calibration and those that occur when the resist gets older.

(41)

4.1.3 Electron beam evaporator

Electron beam evaporator (EBE) is the high vacuum equipped system for the thin layer deposition on different substrates. It consists three main parts: a gun, a target and a substrate holder. All components are placed in a vacuum chamber (vacuum bell jar). The electron gun has the same working principle in producing beam of electrons as scanning electron microscope (SEM), see ref. [138]. Electron beam are directed to the target until the target evaporates and fills up the vacuum chamber.

This evaporated material flies inside the chamber and condenses on the surface of the substrate forming a deposited thin film. The deposition rate depends on the electron beam current driven into the system. The thickness of the film depends not only on the rate of evaporation, but also the time the substrate is exposed to the evaporation process. The time of evaporation can be controlled by the shutter which is situated between the target and the substrate.

The electron beam evaporator is a valuable tool for metal film deposition to the substrates. It allows undirectional deposition, since the evaporated material flies in the straight line to the substrates. The evaporator system used in this work is high vacuum experimentation system Univex 300.

4.1.4 Lift off

Lift off is basically a pattern transfer step which is very important in Nanofabrica- tion. It is also considered as one of the unstable and unreliable processes such that, it is not that much implemented in many industrial applications. Be that as it may, it can be employed with significant chance of success if the thickness of the resist is increased by using single- or double-layer stack of resist [139]. The metal that is not in contact with the substrate is removed in the lift off process, usually done using acetone (CH₃COCH₃). Since acetone is a flammable liquid, it should always stay at a temperature under its boiling point of 57^◦C. Point to note, the substrate can be preheated on the hot plate to increase the solubility of the resist. Moreover, lift off process is followed by the cleaning procedure, always using de-ionized water and dry the substrate with the nitrogen pistol.

(42)

4.1.5 Etching and development process

Etching is used to remove layers from the sample by applying specific solution such as acids, bases or other solvents. Developing a substrate is simply removing part of the resist that has been exposed with the electron beam lithography. Development process is done to ensure best possible resolution during the removal of the exposed resist from the unexposed resist. After etching, the resist is developed using methyl- isobutyl ketone and isopropanol (IPA) mixture for 30 seconds. Point to note, the proportion of the solution is very important. Lastly, the substrate should always be cleaned with IPA and blow dried with the nitrogen.

4.1.6 Sample fabrication process

The samples used in the experiment consist of an array of cylindrical nanoparticles (both gold and silver). The fabrication process starts from the initial design that involves planning the dimensions, BEAMER software is used for pattern design which is going to be used later for electron beam lithography. Sample layout file is created and opened by Cjob software.

Fused silica substrate with thickness of 0.5 mm is used in the fabrication. Firstly, the glass substrate is cleaned with acetone and then undergo oxygen-plasma treatment for 1 minute. The device used for treatment is called MARCH CS-1701.

The essence of this procedure is to remove any residue or contamination from the substrate. Secondly, the substrate is coated with the PMMA (poly-methyl- methacrylate) which is the electron sensitive resist. 400µL of PMMA is spin coated at 1500 RPM for 2 minutes. This produces the thickness of the resist down to 350 nm. The substrate is then removed from the spin coater and placed on a hot plate (175^◦C) for 5 minutes to solidify the resist film. The baking time depends on the type of the resist.

Next, copper is deposited on the substrate in the evaporator vacuum. This is done to avoid accumulation of charge electron during EBL process. The evaporation takes some time, since vacuum needs to be established in the chamber, evaporation of the metal and followed by the deposition. The total time is approximately 30 minutes and ends with copper thickness of 30 nm. The device used in the deposition is called High Vacuum Experimentation System Univex 300.

Plasmonic resonance enhanced SERS signal