New developments in Surface Enhanced Raman Spectroscopy – A review

(1)

MSc Chemistry

Track Analytical Sciences

Literature Thesis

New developments in Surface Enhanced Raman

Spectroscopy – A review

by

Amber van Kooten Niekerk BSc

UvA #: 11822279, VU #: 2581588

September 2018

12 ECTS-credits

Period: July 1

th

_{2018 to September 23}

th

₂₀₁₈

Supervisor/Examiner:

Examiner:

Dr. F. Ariese

Prof. Dr. G. W. Somsen

(2)

(3)

3

Abstract

The aim of this review was to provide an overview of the latest developments of novel substrates with respect to Surface-Enhanced Raman Spectroscopy (SERS). The topics discussed in this review are new developments in single-composition colloidal nanomaterials, multiple-composition colloidal nanomaterials, planar substrates, composite-metal substrates, SERS based sensors and hyphenated SERS methods.

Single-composition colloidal nanomaterial developments recently focus on new morphologies to achieve a higher sensitivity, the control of aggregation by adding reducing and stabilizing agents and improving the repeatability of the analysis. Nanorods offer a higher sensitivity compared to spherical nanoparticles. Recently, new morphologies proposed are nanostars and nanourchins. New reducing and stabilizing agents proposed are based on citrate complexes for novel applications and nanocrystalline cellulose (NCC). To achieve a higher repeatability, it is proposed to employ screen printing or inkjet printing, which is particularly attractive to use for industrial purposes.

Multiple-composition colloidal nanomaterials are a novel concept, with its most recent development (2010) being Shell-isolated nanoparticle-enhanced Raman spectroscopy (SHINERS). Compared to single-composition colloidal nanomaterials, SHINERS offers more substrate flexibility and prevents the core from oxidizing, increasing the stability of the substrate. The tuneable core-shell particles have been shown to be particularly useful for drug delivery applications, as the shell may exhibit drug-release features with a lower toxicity than previously used drug-drug-release molecules.

Planar substrates are one of the first SERS substrates and are mainly used today to improve the ease of application. A recyclable method was proposed using photocatalytic degradation of the adsorbed analytes, followed by a self-cleaning method centrifuging the nanocomposite and washing the substrate with deionized water to remove the residual molecules and ions.

Composite-metal substrates are known to be more uniform and reproducible than colloidal solutions. Recent developments concentrate on novel morphologies for improved hot spot detection for multiple purposes. New morphologies proposed are gold hexagonal-packed nanorod arrays, tuneable nanodomes, large surface nanopyramids and gecko-inspired nanotentacles for ‘press and peeled-off’ pesticide detection.

SERS based sensors have been developed to use for more convenient purposes for industry. New affinity-based SERS sensors were evaluated for the detection of mercury and Salmonella Typhimurium. Also, novel drug-delivery colloidal trackers are explored for cancer therapy and the treatment of schizophrenia, mainly consisting of core-shell particles.

Microfluidics is a novel and interesting field for SERS due to effective dissipation of heat and allowing the removal of photodamaged analyte molecules from the detection volume when using a high-power laser excitation. Furthermore, the recent progress of Lab-On-A-Chip (LoC) allows gene distinction for mutation analysis. Combined with portable Raman systems and more sensitive equipment, it may provide a promising development for more accessibility. A LC-SERS application is demonstrated allowing a gradient elution mode for a more selective SERS detection in samples with a higher structural complexity.

Overall, new morphologies for substrates have been employed as well as improvement of enhancement limiting parameters to gain more sensitivity, in some cases almost reaching single-molecule level. Also, more robust composite-metal substrates and several tuneable, facile and re-usable substrates have been employed. These developments promise a wide range of study perspectives for fundamental and industrial purposes.

(4)

4

Abbreviations

AEF Analytical Enhancement Factor

AFM Atomic Force Microscopy

ASA Ascorbic Acid

AgNP Silver nanoparticle

AuNS gold nanostar

AuNU gold nano-urchin

BAuNSP Branched nanoporous gold Nano Shell Particles

CE Chemical enhancement

CIPRO Ciprofloxican

CT Charge-transfer

CV Crystal Violet

Cy5 Cyanine 5

DCEF Direct Current Electric Field

DCF Discarded Cigarette Filter

DDA Discrete dipole approximation

DNA Deoxyribonucleic acid

DoE Design of Experiments

DOX Doxorubicin EF Enhancement Factor EM Electromagnetic ENRO Enrofloxican FDTD Finite-Difference Time-Domain FQ fluoroquinolone

HAuCl4 Chloroauric Acid Tetrahydrate

HAV hepatitis A virus

HOMO Highest occupied molecular orbital

LIL Laser Induced Lithography

LLOQ Lower Limit Of Quantification

LOD Limit Of Detection

LUMO Lowest unoccupied molecular orbital

LSPR Localized surface plasmon resonance

MACE Metal Assisted Chemical etching

MMP Matrix metalloprotease

MPTS 3-mercatopropyltrimethoxysilane

MV Methyl Violet

NCC Nanocrystalline cellulose

NF Notch Filter

NMF-ALS Non-negative Matrix Factorization Alternating Least Squares

NP Nanoparticle

(6)

6

PBS Phosphate Buffered Saline

PC Pheochromocytoma

PCA Principal Component Analysis

PDMS Polydimetylsiloxane

PEG Polyethylene glycol

PS Polystyrene

PVP Polyvinylpyrrolidone

PZC Point of zero charge

R6G Rhodamine 6G

RBF Riboflavin

RCF Relative centrifugal force

(R)GO (Reduced) Graphene Oxide

RIE Reactive Ion Etching

RNA Ribonucleic Acid

RSD Relative Standard Deviation

ROX X-rhodamine

SCS Scattering Correlation Spectroscopy SEM Scanning Electron Microscopy

SERS Surface-enhanced Raman scattering (optical effect) / Surface-enhanced Raman spectroscopy (technique)

SSEF SERS Substrate Enhancement Factor

SSR Surface Selection Rule

SHINERS Shell-isolated nanoparticle-enhanced Raman spectroscopy TEM Transmission Electron Microscopy

TERS Tip-enhanced Raman spectroscopy

TOA Trioctylamine

(7)

7 SERS applications SERS advantages scope objectives

Chapter 1. Introduction

Surface-enhanced Raman spectroscopy (SERS) is a spectroscopy technique that uses metallic surfaces (substrates) to provide amplification of Raman scattering signals by a maximum factor of about 1014

[1], which is typically referred to as the so-called enhancement factor (EF) value. [2, 3]

SERS is a technique widely used in fields such as the pharmaceutical field [4-5], bio-applications [6-7], food industry [8], explosives trace detection [9] and water pollutant trace detection [10], and its fundamental and application wise development are still ongoing. The SERS effect has sparked scientists’ interests for the purpose of achieving higher selectivity, sensitivity and robustness compared to normal Raman spectroscopy methods. [11] Compared to other detection techniques used for ultrasensitive detection, the advantages of SERS are that the technique in most cases is less time-consuming, provides narrow line-widths and has non-destructive testing capabilities. [12] Ever since the discovery of SERS, more than half a million articles (accessed on 13th_{of July 2018) have been}

published by Elsevier about SERS. In this chapter, the scope of the review will be discussed as well as theoretical background of SERS.

1.1 Scope

As discussed in the section above, there are many new developments within the field of SERS. The main areas contributing to SERS are fundamental aspects, theoretical aspects, electro-chemical studies, synthetic chemistry, substrate development and related spectroscopic techniques. [2] In this thesis, the main topics discussed are the development and applications of novel substrates. A few examples will be provided for each novel substrate. Moreover, the advantages and limitations will be highlighted. Some of the other areas will also be briefly included. For more information it is advised to consult resources such as reviews from Song-Yuan Ding et al. [13] (theoretical aspects), Andrew J. Wain et al. [14] (electrochemical studies) and M. Fernanda Cardinal et al. [15] (synthetic chemistry).

The aim of this review is to provide an overview of the latest developments of novel substrates with respect to SERS. The main parameters evaluated in this review are:

• the sensitivity gained by the substrate;

• the uniformity of signal distribution gained from the substrate; • the complexity of the substrate fabrication;

• the stability of the substrate; • the flexibility of the substrate; • and the re-usability of the substrate.

The substrates highlighted in this thesis are single composition colloidal substrates (chapter 2), multiple composition colloidal substrates (chapter 3), planar composite-metal substrates (chapter 4), higher-order composite-metal substrates (chapter 5) and SERS-based sensors (chapter 6). Also, to demonstrate a broader range of applications, the advantages of coupling SERS with novel techniques will be discussed in chapter 7. Finally, the performance of the substrates is compared in chapter 8 and ideas will be proposed for some future study perspectives for new SERS applications in chapter 10.

(8)

8 Raman Spectro-scopy polarizability Stokes Raman Raman setup RRS

1.2 Theoretical Background

To be able to give a clear insight of the new developments within SERS, it is required to understand its theoretical aspects. This section gives an overview of the origin and mechanisms of SERS, which will later be used when discussing the novel substrate developments.

1.2.1 Introduction to Surface-enhanced Raman Spectroscopy

In order to understand the SERS mechanism, it is important to understand the process of Raman scattering itself. Raman scattering was first experimentally observed in 1928 by the Indian Physicist C. V. Raman. Raman scattering is considered to be the inelastic scattering of light, which is far less probable than Rayleigh elastic scattering. When the inelastic scattering occurs, it invokes a change in polarizability in which the polarizability is a measure of the responsiveness of the electrons in a molecule to the presence of an external electric field. This change in polarizability can be visualized by the Jablonski Diagram as a virtual state (which can to some degree be viewed as a mathematical construction of perturbation theory). Physically speaking, even though it is a virtual state between vibrational and excited states, it behaves more like a vibrational state than as an excited state. The Raman signal can further be characterized by the Anti-Stokes and Stokes signal, in which for the Anti-Stokes signal the energy is higher than the energy of the laser and for the Stokes signal the energy is lower than the energy of the laser. [2] The instrumental setup for a typical Raman spectroscopy experiment is illustrated in figure 1. [16] The Stokes and Anti-stokes phenomena are represented in figure 2, [2] in which it can be noted that the peak at a wavenumber of 0 cm-1_{is a result of Rayleigh}

scattering which is typically partially blocked by the notch filter (NF) to prevent overwhelming signal interference from Rayleigh scattering.

If the virtual state matches with a real electronic state of the molecule, the scattering process is said to be resonant. This phenomenon is used in applications based on techniques like Resonant Raman Spectroscopy (RRS) and Surface-Enhancement Resonant Raman Spectroscopy (SERRS). [2]

Figure 1. Schematic diagram of typical Macro-Raman (a) and Micro-Raman (b) configurations. CCD: charge-coupled device, CL: collection lens, FL: focusing lens, NF notch filter.

Reproduced from Ref. [16]

Figure 2. Jablonski diagrams of anti-Stokes (a) and Stokes (b) processes and the representation of both processes in a Raman

(9)

9 Raman

cross-section

SERS brief history

For Raman scattering, the average signal produced by the process is directly proportional to the laser power density and to the Raman cross-section of the molecule, in which the cross-section of a molecule corresponds to the effective area of a homogeneous incoming beam for which the molecule undergoes Raman scattering. This phenomenon is represented in equation 1. [2]

𝑃 = 𝜎𝑆𝐼𝑛𝑐 (1)

Definition Raman cross-section: intensity (P, W), incident power density SInc (W∙m-2), Raman cross-section (𝜎, m2)

For example, double bonds and aromatic groups exhibit reasonable Raman scattering cross sections. [10] A simplistic Raman approach is that the induced Raman dipole 𝑝𝑅 is a response to the incident light and is a product of Raman polarizability tensor 𝛼̂𝑅 and the magnitude of electric field 𝐸𝐼𝑛𝑐 at the molecule position under monochromatic light excitation of frequency 𝜔𝐿. This phenomenon is represented in equation 2. [2]

𝑝𝑅= 𝛼̂𝑅𝐸𝐼𝑛𝑐 (𝜔𝐿) (2)

Magnitude of induced Raman dipole: Raman dipole (𝑝𝑅), Raman polarizability tensor (𝛼̂𝑅), electric field (𝐸𝐼𝑛𝑐), oscillating

frequency 𝜔𝐿

The SERS effect was first observed in 1974 by a group of scientists from the department of Chemistry at the University of South Hampton (M. Fleishmann et al.) working on combining Raman spectroscopy with electrochemistry. It was observed that the bands of pyridine change considerably when pyridine is examined close to the surface of a silver electrode, which did not line up with the knowledge of the Raman theory at that time. It was also observed that the peak intensity decreases with shift of the potential of the electrode away from the point of zero charge (PZC). It was thought that this effect corresponds to adsorption/desorption properties of the molecule, as the adsorbate is replaced by water, anions and cations when adding a charge to the solution, resulting in a reduction of interaction with the surface. [17] The SERS phenomenon was further explored in 1977, proposing enhancement mechanisms based on electromagnetic [18] and electrochemical effects [19]. It is found that the signal enhancement is provided by interaction of electromagnetic radiation (plasmon resonances) in the metal substrate based on certain surface selection rules. Nowadays, it is confirmed that surface-enhanced Raman scattering mainly consists of using the large local field enhancements that can exist at metallic surfaces to boost the Raman scattering signal of molecule close to or at the surface. [2] The sections below highlight what factors actually contribute to enhancing the signal and how to calculate the EF value.

(10)

10 Distance dependency ‘g’ -term dependency Angle dependency E4 enhance-ment FDTD & DDA

1.2.2 Electromagnetic enhancement

As demonstrated in equation 1, the average intensity of a molecule is proportional to the cross-section multiplied by the incident laser intensity. For SERS, the cross-section is multiplied by the EF value. The enhancement factor (EF) is the magnitude of increase in the Raman scattering cross-section when the molecule is adsorbed to a SERS-active substrate. The induced Raman dipole mainly originates from multiplicative contributions consisting of a change in the Raman polarizability tensor and the electric field, as shown in equation 2. According to the electromagnetic theory for SERS, the amplitude of the electric field is dramatically enhanced, so therefore the product Raman dipole will also be enhanced. [2] The electromagnetic enhancement can be intensified up to 1014_times,_[1]_{up to the level of a single}

molecule. The electromagnetic contribution of the EF value can be separated into two multiplicative EF’s, for the incident field and for the re-emitted (Raman) field.

The electromagnetic (EM) enhancement factor (FEM) depends on the large local field enhancements

that can occur close to metallic surfaces when localized surface plasmon resonances (LSPR) are excited. [2] The LSPR occurs when the collective oscillation of valence electrons in a substrate material is in resonance with the frequency of the incident light, which is shown in figure 3a. [16]

The resulting magnitude of the electromagnetic field is illustrated in equation 3.

𝐸𝑜𝑢𝑡(𝑥, 𝑦, 𝑧) = 𝐸0𝑧̂ − 𝛼𝐸0[ 𝑧̂

𝑟3−

3𝑧

𝑟5(𝑥𝑥̂ + 𝑦𝑦̂ + 𝑧𝑧̂)] (3)

Magnitude of the electromagnetic field: Magnitude electromagnetic field outside of the particle (𝐸𝑜𝑢𝑡), Magnitude

electromagnetic field of the particle (𝐸0), metal polarizability 𝛼, radial distance 𝑟, Cartesian coordinates (𝑥, 𝑦, 𝑧), Cartesian

unit vectors (𝑥̂, 𝑦̂, 𝑧̂)

As equation 3 illustrates, the field enhancement decays with 𝑟−3. [16] When a molecule approaches the surface, direct adsorption occurs either through physisorption or chemisorption. This practically comes down to a finite sensing vector of about ~10 nm from the surface. [2] The metal polarizability may be further expressed as demonstrated in equation 4.

Figure 3. Illustration of the localized surface plasmon resonance effect (a) and extinction efficiency (b) of a spherical silver nanoparticle of 35-nm radius with ȁ𝐸𝑚𝑎𝑥ȁ2 = 85. Reproduced from Ref. [16]

𝑧 𝑥 𝑦 EM enhance-ment Distance dependency

(11)

11 ‘g’ -term dependency Angle dependency E4 enhance-ment FDTD & DDA 𝛼 = 𝑔a3, 𝑔 = ( 𝜀𝑖𝑛−𝜀𝑜𝑢𝑡 (𝜀𝑖𝑛+𝜒𝜀𝑜𝑢𝑡)) (4)

Polarizability of a metal nanoparticle: radius of the sphere (𝑎), dielectric constant of nanoparticle 𝜀𝑖𝑛, shape factor 𝜒

(spherical particle geometry = 2), dielectric constant of external environment 𝜀𝑜𝑢𝑡.

As equation 4 illustrates, the maximum enhancement occurs when the denominator of 𝑔 = 0, thus 𝜀𝑖𝑛 ≈ −2𝜀𝑜𝑢𝑡. Equation 3 and 4 can be manipulated into equation 5.

ȁ𝐸_𝑜𝑢𝑡ȁ2_{= 𝐸}

02[1 − 𝑔]2+ 3𝑐𝑜𝑠2𝜃(2𝑅𝑒(𝑔) + ȁ𝑔ȁ2) (5) Manipulation from equation 3 and 4: Magnitude electromagnetic field outside of the particle (𝐸𝑜𝑢𝑡), Magnitude

electromagnetic field of the particle (𝐸0)

As equation 5 illustrates, the peak enhancement occurs when 𝜃 = 0° or 180°, corresponding with Raman back-scattering setup as illustrated in figure 1. In the cases where 𝑔 is large, the maximum enhancement approaches ȁ𝐸𝑜𝑢𝑡ȁ2= 4𝐸02ȁ𝑔ȁ2. Overall, the electromagnetic enhancement is dependent on multiple factors, such as the distance between the surface and the molecules, the dielectric constants of the substrate and the external environment and the angle of incident light.[16] Specifically, for the Raman Stokes-shifted frequency, equation 5 can be reworked into equation 6, which considers the fact that the emission of radiation from the dipole may also be enhanced while equation 3 gives a more general expression for the enhancement of the incident field.

𝐸𝐹 =ȁ𝐸𝑜𝑢𝑡ȁ2|𝐸′𝑜𝑢𝑡|

2

ȁ𝐸₀ȁ4 = 4ȁ𝑔ȁ

2_ȁ𝑔′ȁ2 ₍₆₎

Theoretical E4_{enhancement approximation: Magnitude electromagnetic field upon re-emission outside of the particle}

(𝐸′𝑜𝑢𝑡)

Equation 6 is commonly referred to as E4_{enhancement as a field enhancement to the fourth power}

occurs at the nanoparticle surface. Equation 6 illustrates that the magnitude of the SERS EM enhancement is 104_{to 10}5_{for a small sphere (ȁ𝑔ȁ ≈ 10) and can even approach 10}8_{for higher-order}

silver nanostructures (with a much larger ȁ𝑔ȁ). [16]It is found by E4_{enhancement modelling using}

finite-difference time-domain (FDTD) and Discrete dipole approximation (DDA) simulations that small gaps for SERS substrates further increase the enhancement. [20]

(12)

12 ‘hot spots’

An example of a FDTD simulation is illustrated in figure 3b, which is expressed as the extinction efficiency of in this case a spherical silver nanoparticle in suspension. [16]. In the field of SERS, these gaps are commonly referred to as hot spots, which are defined as highly localized regions of intense local field enhancement caused by surface plasmon resonances. [21] For higher-order structures, it is found that the order of enhancement is determined by the number of intrinsic hot spots. The number of intrinsic hotspots increases as nanospheres < nanotriangles < nanostars in suspension, and so does the enhancement effect. Figure 4 illustrates the comparison of SERS spectra of Rhodamine 6G (R6G) in gold nanosphere, gold triangle and gold nanostar suspensions. Even though nanostars provide large enhancement of the electric field around the nanoparticles, it also exhibits a very high sensitivity to local changes in the dielectic environment. [10]

An example of a FDTD simulation is illustrated in figure 3b, which is expressed as the extinction efficiency of in this case a spherical silver nanoparticle in suspension. [16]. In the field of SERS, these gaps are commonly referred to as hot spots, which are defined as highly localized regions of intense local field enhancement caused by surface plasmon resonances. [21] For higher-order structures, it is found that the order of enhancement is determined by the number of intrinsic hot spots. The number of intrinsic hotspots increases as nanospheres < nanotriangles < nanostars in suspension, and so does the enhancement effect. Figure 4 illustrates the comparison of SERS spectra of R6G in gold nanosphere, gold triangle and gold nanostar suspensions. Even though nanostars provide large enhancement of the electric field around the nanoparticles, it also exhibits a very high sensitivity to local changes in the dielectic environment. [10]

Figure 4. Comparison of SERS spectra of 5 μM Rhodamine 6G in gold nanosphere, nanotriangle and nanostar suspensions. Reproduced from Ref. [10]

(13)

13 Colloidal aggregation aggregation orientation Wavelength dependency Distance dependency spacers Wavelength dependency

When an aggregated state of metal colloidal nanoparticles is considered, the enhancement effect reaches an optimum at a certain gap distance. In general, it is thought that ȁ𝐸ȁ2~ 1

𝑔𝑎𝑝2. At very small

gaps, the effect decreases due to complex repulsion mechanisms. The orientation of the aggregated metal particles is also an important factor. Figure 5 demonstrates that for a silver cylindrical rod in water, the theoretical electric field enhancement is 1.8 x 106_{and for aggregated states: 39x for}

head-to-head, 30x for perpendicular and 0.27x for side-by-side orientation respectively compared to the isolated state. [20]EEM

When considering equation 3, taking the E4_{enhancement approximation in consideration, the overall}

distance dependence should scale with r-12_{. This assumption is particularly useful for applications in}

which direct contact between the adsorbate of interest and the surface is not possible. In practice, when using spacers (such as thin films) as substrates, not only the above parameters contribute to the 𝐹𝐸𝑀 factor but also the chemical uniformity, the conformity of the spacers and whether the spacers are pinhole free. [16]

As mentioned above, the electromagnetic contribution of the EF value can be separated into two multiplicative EF’s, for the incident field and for the re-emitted (Raman) field. In order to achieve the maximum enhancement, both the incident field and the radiated field need to optimally contribute to the electromagnetic enhancement. However, as demonstrated in section 1.2.1 for Raman scattering both fields are irradiated at a different wavelength. When the vibrational energy spacing corresponds to the fingerprint region of the spectrum (500 – 1500 cm-1_{), the small Stokes-shift and E}4_{approximation}

are no longer accurate estimations. This principle is illustrated in figure 6. Recent advances in nanofabrication and better tunability in both the SERS excitation and detection made it possible to consider this effect in the enhancement approximation. [16]

Figure 6. Surface-enhanced Raman extinction spectrum of pyridine with profile fit maximum λex,max = 662 nm (left) and LSPR

λmax = 690 nm (right). Reproduced from Ref. [16]

Figure 5. Contours of EEM for silver plasmon particle cylindrical rod (19 x 52 nm) in water in the presence of optical radiation with (a) isolated state, and dimers of: (b) nm gap head-to-head orientation, (c) nm gap perpendicular orientation, (d)

(14)

14

1.2.3 Electrochemical enhancement

Another multiplicative contribution to the EF value is the chemical enhancement (CE) factor (𝐹𝐶𝐻). The 𝐹𝐶𝐻 is heavily molecule dependent. The impact (and to some scientists even the existence) of this contribution is still a subject of discussion. This is due to the interpretation whether electrochemical enhancement can or cannot be explained by the EM theory. [2] CE provides a smaller contribution to the EF value (𝐹𝐶𝐻 ~ 102) compared to the EM effect contribution (𝐹𝐸𝑀 ~ 1014). [1] The CE factor is currently viewed as a modification of the electronic polarizability of the probe, which can induce resonant-Raman scattering at wavelengths where the non-adsorbed molecule would not be resonant. This phenomenon originates from a modification of 𝛼̂𝑅, as illustrated in equation 2. This is thought to be caused by the charge-transfer (CT) mechanism in which the molecule needs to be chemically adsorbed on the surface (𝑟 approaches zero). [2]

The charge-transfer (CT) mechanism may be viewed as a change of position of the Fermi level (level in which there is a 50% probability of occupied electrons) relative to the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO). This mechanism is illustrated in figure 7. The laser energy can directly be in resonance with as electronic transition of a molecule-metal complex (a), or can profit from an indirect coupling (CT) through the metal (b and c) [2] due to ligand-metal charge-transfer (b) and ligand-metal-ligand transfer (c). [22]

Situation (a) involves the presence of a surface complex which produces a substantial change in intrinsic polarizability of the molecule due to a new overlap structure of the molecular orbitals. Situation (b) and (c) involve the scenario in which the adsorbate does not bind covalently to the metal. The presence of the metal acts as a perturbation to the electronic structure of the analyte, thus causing a mild change in its electronic distribution. There is also an alternative situation which involves the process of photo-driven charge transfer between the analyte and the metal, and can be used to change the difference in energy (and the Fermi level) between the adsorbed analyte and the metal through an external potential. [2] It should be noticed that the charge-transfer transition moment is coupled with the component of the electric field due to the surface plasmon resonance perpendicular to the surface. [22]

Figure 7. Schematic representation of a charge-transfer mechanism in the SERS cross-section, appearing as a resonant contribution to the measured intensity.

Reproduced from Ref. [2]

CE enhance-ment

CT mechanism

(15)

15

1.2.4 Surface Selection Rules

As mentioned above, the magnitude of the EF value is dependent on numerous factors such as: 1) the characteristics of the laser excitation (wavelength, polarization, and angle of incidence); 2) the detection setup (scattering configuration, collection angle and polarization of detection); 3) the SERS substrate (material, geometry, orientation, refractive index of the environment); 4) the intrinsic Raman cross-sections;

5) analyte adsorption properties (efficiency, concentration, distance for the surface, orientation). These factors mainly determine the EM enhancement factor. The undiscussed topics are the particular orientation of the molecule on the surface and the specific Raman mode symmetry. These are called surface selection rules (SSRs). For a fixed molecular adsorption on the surface, there are changes in the relative intensities of Raman modes with different Raman tensor symmetries. The SSR SERS enhancement factor is thought to be much smaller than the EM factor. [2]

1.2.5 Calculation of the Enhancement Factor

The sections above highlighted the main factors that contribute to the EF value. For the EF value, there are different definitions. This section highlights the importance of making a distinction in definitions of the EF value.

The average EF value is most commonly measured according to equation 7a under identical experimental conditions (laser wavelength, laser power, microscope objective or lenses, spectrometer, etc. [23]), and is known as the SERS substrate enhancement factor (SSEF). The number of molecules is determined by evaluating the spot size and probe volume. [16] There are some side notes to this definition, specifically to how to calculate 𝑁𝑆𝑢𝑟𝑓. For a 2D substrate structure, it is questioned whether to only count the molecules adsorbed on the metal or to also count the adsorbed molecules on the non-metallic parts in between. Also, it is questioned whether to count the molecules in the first layer or also in subsequent layers. In most cases the excitation intensity is also not uniform across the scattering volume. Differences in interpretations may lead to differences of measured EF as much as 2 orders of magnitude. [23] For equation 7a to apply, the parameters as stated above have to be assumed. A small modification can be applied to this equation, which is illustrated in equation 7b. [2] Another approach for the determination of the average EF value is the analytical enhancement factor (AEF), presented in equation 7c. [2] This definition is relatively easier to employ in studies compared to the definition for equation 7a and is applied for specific practical applications, for example in case of SERS active liquids (colloidal solutions). However, it depends strongly on the adsorption properties and surface coverage of the probe, and it strongly depends on the sample preparation procedure for 2D planar substrates. Therefore, the experiment needs to state the experimental conditions clearly and ensure that the coverage is sub monolayer, as SERS is a distance dependent phenomenon (equation 3). [23] (a) 𝑆𝑆𝐸𝐹1= [𝐼𝑆𝐸𝑅𝑆 𝑁_{𝑠𝑢𝑟𝑓} ⁄ ] [𝐼𝑁𝑅𝑆 𝑁𝑣𝑜𝑙 ⁄ ] ; (b) 𝑆𝑆𝐸𝐹2 = [𝐼𝑆𝐸𝑅𝑆 𝜇𝑀𝜇𝑆𝐴𝑀𝐴𝐸𝑓𝑓 ⁄ ] [𝐼𝑁𝑅𝑆 𝑐_𝑅𝑆𝐻_𝐸𝑓𝑓𝐴_𝐸𝑓𝑓 ⁄ ]; (c) 𝐴𝐸𝐹 = [𝐼𝑆𝐸𝑅𝑆 _𝑐 𝑆𝐸𝑅𝑆 ⁄ ] [𝐼𝑁𝑅𝑆 𝑐_𝑁𝑅𝑆 ⁄ ] (7) Practical E4_{enhancement approximation: (a) SSEF: SERS Substrate Enhancement Factor, (b) non-rigorous SSEF, (c) AEF:}

Analytical Enhancement Factor; 𝐼𝑆𝐸𝑅𝑆 SERS intensity, 𝑁𝑠𝑢𝑟𝑓 number of molecules bound to substrate, 𝐼𝑁𝑅𝑆 Normal Raman

intensity, 𝑁𝑣𝑜𝑙 number of molecules in excitation volume, 𝑐𝑁𝑅𝑆 Normal Raman sample concentration, 𝑐𝑆𝐸𝑅𝑆 SERS sample

concentration, 𝐴𝐸𝑓𝑓 effective surface area of scattering volume (m2), 𝐻𝐸𝑓𝑓 effective height, 𝜇𝑀 surface density of the

individual nanostructures producing the enhancement (m-2_{), 𝜇}

𝑆 surface density of molecules on the metal (m-2)

SSRs

SSEF

Non-rigorous SSEF

(16)

16 Advantages Study prospects Novel morpho-logies Multi-branched nanostars Growth reaction

Chapter 2. Single composition-based colloidal nanomaterials

Single composition-based colloidal nanomaterials are characterized by plasmon active particles in a solution. [24] The most important advantages of colloids are the simplicity of its fabrication and efficient ligand adsorption in a large specific surface area. Moreover, there are multiple ways to collect information about the SERS process with the use of colloids. In case of metal colloidal suspensions, the size of the metal particles can be determined by the use of Transmission Electron Microscopy (TEM) and will provide information on ligand adsorption. Also, the presence of surface plasmon resonance bands can be monitored using Ultraviolet-Visible (UV-Vis) absorption spectroscopy [25].

The most used materials for metal colloids are silver and gold [25 - 27], as these metals provide the largest enhancement factors. Silver provides the largest enhancement factor, whereas gold has the advantage of being biocompatible and displays a more stable surface chemistry. [27] The performance of colloidal substrates is very much dependent on the nanoparticle morphology, strength wavelength used and the degree of aggregation [21], as mentioned in chapter 1. Thus, most of the colloid methods mainly focus on improving the sensitivity, the control of aggregation of the substrate and improving the repeatability of preparation. [27] In the section below, novel morphologies and ways to control the above-mentioned parameters are discussed.

Increased sensitivity: novel morphologies for colloidal substrates

In chapter 1.2.2, it is explained that the shape of the substrate is an important factor for enhancement. Mostly, shapes as nanospheres, nanotriangles, nanorods and nanostars are employed for the increase of the sensitivity. [10] Recently, a multitude of novel colloidal shapes have been studied such as multibranched (gold) nanostars, (gold) nano-urchins and nanoflowers. Particular interest has been shown for creating colloidal substrates with many sharp tips, as these tips are responsible for a strong LSPR enhancement. Furthermore, these substrates imply a greater surface area-to-volume ratio for nanovectorization compared to spherical-shaped nanoparticles. [28] The section below will highlight the methods used for the preparation and the results.

Branched gold nanostars (AuNSs) with sharp and relatively long tips, have been used for effective SERS enhancement. [29] Jian Zhu et al. (2014) prepared gold nanostars with different branch lengths using various added concentrations of chloroauric acid tetrahydrate (HAuCl4, as surfactant) and ascorbic acid

(ASA, as reducing agent) in the growth solution. The TEM image of the multibranched gold nanostars is illustrated in figure 8a. The gold cores at the particle center had a diameter of about 10 nm and the average aspect ratio of the synthesized branches was about 5.0. A laser with an excitation wavelength of 785 nm was used. A maximum of six-branched nanostars were synthesized, with six-branched nanostar peaks exhibiting increased absorbance as well as a higher wavelength compared to the lower branched nanostars. The synthesized gold nanostars were compared with the performance of gold nanorods. It was found that the SERS ability of gold nanostars is much stronger than the SERS ability of gold nanorods using 2 x 10-7_{M R6G as a Raman reporter, as demonstrated in figure}_8a_{. The authors’}

explanation for this phenomenon is that the gold nanostar substrate exhibits a third LSPR band in the infrared region (caused by the six-branched gold nanostars), which is absent in the gold nanorod substrate. [30]

(17)

17 thiram

Nano-urchins

SCS

Recently, Jian Zhu et al. (2018) used the multi-branched gold nanostars for the SERS detection of the pesticide thiram on apple peels. The gold stars were fabricated using a seed-growth method using Triton X-100 and HAuCl4. The apple peels were cut into 1 x 1 cm2 squares and deposited on glass slide

with double-sided tape. After securing the peels onto the glass slide, 10 𝜇𝐿 of various concentrations of thiram ethanol solutions was dropped onto the surface of the peels, followed by adding 20 𝜇𝐿 of ethanol and 20 𝜇𝐿 of the gold star solution. The results demonstrated that the AEF was 1.04 x 105_for

thiram and that the Raman activity of gold nanostars with fractal features was found to be 1.37 higher compared to gold nanostars without fractal features. The detection limit was found to be 10-10_{M (with}

a sample volume of 15 𝜇𝐿) thiram in solution, which is lower than the limit of detection (LOD) gained from bimetallic Au@Ag nanorods, silver nanoshells, silver nanowires, ag nanoparticles on Au film over a nanosphere substrate and silver nanoparticles-PDMS films. The LOD was found to be higher than using Au@Ag nanocuboids and hierarchical structural Ag nano-crown arrays as substrates. However, it is claimed that the proposed method is easier to implement than the other two methods. The selectivity of the substrate was also tested using thiram with a concentration of 10− 6_{M in a solution}

of chlorpyrifos, ﬂusilazole, methyl parathion and malathion (common pesticides) with concentrations of 10-4_{M. The spectra showed distinct peaks for thiram and almost no peaks in the other spectra,}

which can be explained by the S-S bond cleavage of thiram, making thiram more a more suitable candidate to achieve successful LSPR. [31]

Nano-urchins (AuNUs) are sea urchin-shaped nanoparticles that have randomly grown short tips around the core and a rough surface. Based on the unique shape of AuNUs with sharp tips, a large SERS enhancement is predicted because of the larger surface area compared to simple and smooth structures. Furthermore, hot spots that induce a large SERS enhancement can also be generated at the connection between two adjacent AuNUs. [29]

Dahia Issaad et al. (2017) performed a comparative study between gold nanospheres and gold nano-urchins. The gold nano-urchins with a nominal diameter of 50 and 80 nm were used (purchased from Sigma-Aldrich) and evaluated using a self-assembled monolayer of thiophenol as a Raman reporter. The length of the AuNU branches were determined to be 6 nm, using Scattering Correlation Spectroscopy (SCS). A laser with an excitation wavelength of 660 nm was used. The surface area of the nano-urchins was estimated to be 400 times higher than that of the gold nanospheres when saturated with thiophenol (𝑐 ~1 𝑀), demonstrating a superior performance. [28]

Figure 8. (a)TEM image of six branched gold nanostars, (b) Raman spectrum of 1 mM pure R6G (purple), gold nanorods (blue) and gold nanostars (red). Reproduced from Ref. [30]

(18)

18 Nanoflowers

Growth solution

Even though a large SERS enhancement is predicted and reported, it is a challenge to control the multiple tips of AuNUs due to variation in size, number and distance from the core and it is a challenge to achieve effective adsorption of Raman probe molecules on the rough surfaces of AuNUs. A recent interest is shown to enhancing the performance of AuNUs to increase the stability of the substrate. Minjung Seo et al. (2018) recently investigated and variated various factors to improve the randomly branched gold nano-urchins as a substrate. AuNUs with an average size of 100 nm were obtained from Sigma-Aldrich. The Scanning Electron Microscopy (SEM) image of the nano-urchins is illustrated in figure 9. The performance was evaluated using Rhodamine 6G as a Raman reporter in a range of 1 mM to 0.5 M. The AuNUs were characterized with SEM image evaluation and UV-Vis extinction spectra. The enhancement of SERS was effectively achieved with a set of conditions that included a concentration for the AuNUs of 0,3 mM (using a Rhodamine concentration of 1 mM, weight ratio ≈ 8:1 AuNPs:R6G), a binding time of 6 to 8 hours, centrifuging for 40 minutes at a Relative Centrifugal Force (RCF) of 2400, and no stirring time was required during sample preparation. [29]

Nanoflowers are flower shaped NPs with a high degree of surface roughness. Nanoflowers are synthesized with the requirement of the use of any templates. Furthermore, the process of synthesis is carried out in an aqueous medium, thus preventing pollution from heavy metals or organic matter to reactants. The morphology of nanoflowers is found to be superior to using globular-shaped surface features and smaller (silver) nanoparticles. Silver nanoflowers are fabricated for measuring food colorants amongst other things. [32-33]

Yan-xiong Wu et al. (2017) recently synthesized silver nanoflowers using an aqueous silver nitrate (AgNO3) reduction by ascorbic acid (ASA) as a reducing agent in the presence of a polyvinylpyrrolidone

(PVP) surfactant, with the PVP surfactant controlling the growth direction of the silver nuclei. The diameter of the nanoflowers were adjusted from 450 to 1000 nm with branches of 10 to 25 nm by tuning the reaction time, reaction potential and reagent concentration. The SEM image of the synthesized nanoflowers is illustrated in figure 10. The synthesized nanoflowers were well dispersed and were not fused together. Many ridges exist over the AgNP surfaces, in which the ridge thickness deviated from 50 to 75 nm. A laser with an excitation wavelength of 785 nm was used. [33]

Figure 10. SEM image of 600-800-nm silver nanoflowers. Reproduced from Ref. [33]

Figure 9. SEM image of 100-nm gold nano-urchins (AuNUs). Reproduced from Ref. [29]

(19)

19 carmine dye Au nanoflowers Controlled aggregation NCC Riboflavin DCF

R6G molecules were used as a Raman reporter to evaluate the performance in a range of 10-8 _{to 10}-3

M. The Lower Limit of Quanitification (LLOQ) was found to be 10−8 M R6G, as clear characteristic bands were still identified. The substrate is reported to be more sensitive than silver nanoparticle decorated reduced graphene oxide (rGO) nanosheet substrates. The substrate quantification performance was also evaluated for carmine dye (a bright-red colorant used in various flowers, paints, inks, cosmetics and foods) using Principal Component Analysis (PCA). According to the authors, PCA was used to improve the accuracy and simplicity of the measurement. A PCA model was built from six concentrations based on eight characteristic peak dimensions. The two components with the highest cumulative contribution rate were used. The build was tested using a concentration of 5 * 10-8_{M and}

a logarithmic fitting curve method, in which the value of the fitting curve corresponded with 4.5 * 10 -8_{M, indicating that the used PCA algorithm can be reliably used for regression analysis.}_[33]

Yu-jie Ai et al. (2018) also demonstrated the performance of silver nanoflowers using R6G as a Raman reporter and four food colorants, e.g. food blue, tartrazine, sunset yellow and acid red as probe molecules. The same nanoflower size, branch length and lowest concentration R6G as Yan-xiong Wu et al. (2017) [33] are reported. The LOD of the colorants were determined using PCA, as according to the authors, PCA can be employed to improve the accuracy and simplicity of the measurement. LOD values of 79.285 μg/L (food blue), 5.3436 μg/L (tartrazine), 45.238 μg/L (sunset yellow) and 50.244 μg/L (acid red) are reported with an average LOD of 10-7_M._[32]_{The selectivity of detection of the food}

colorants was not discussed in the article.

Gold nanoflowers have also been employed. However, it is still a challenge to maintain the stability of the nanoflowers. It is found that anisotropic gold nanoflowers may transform into isotropic spherical gold nanoparticles within 1 hour after the reaction due to its unstable character. For gold nanoflowers, it is thought that intraparticle ‘ripening’ induced by chlorine ions (from HAuCl4) is likely to affect both

the formation and stability of the nanoflowers. Increasing the pH of the storing solution or removal of the chloride ions increase the stability of the fabricated nanoflowers. [34] It is also possible to use catanionic vesicles as a reducing agent for the fabrication of gold nanoflowers. [35]

Controlled aggregation: reduction agents

Controlled aggregation is an important aspect for colloid substrate design as it grants uniformity in sensitivity for the analysis.The uniformity of sensitivity in silver nanoparticles is primarily caused by the size and shape of the nanoparticles. Generally, in order to achieve aggregation control, reducing- and stabilizing agents are added. [21] Chemical reduction of silver salts is the common synthetic route adopted for the preparation of AgNPs. This synthesis route requires the use of suitable solvents, reducing agents and stabilizing agents. [36]

Recently, interest has been shown to nanocrystalline cellulose (NCC) being used as a reducing agent in the synthesis of silver nanoparticles. NCC has attracted increasing attention due to its impressive mechanical, thermal and optical properties. During the process of formation, the hydroxyl groups on the cellulose surface form a complex with silver ions through ion-dipole interaction and prevent mass aggregation on reduction by coating the AgNPs. Segun A. Ogundare et al. (2018) developed a method using NCC as a reducing and stabilizing agent in the formation of silver nanoparticles (AgNPs) for the detection of riboflavin (RBF), an organic compound serving as an effective sensitizer to enhance photochemical transformation of many organic compounds). A schematic illustration of NCC-AgNP-RBF is demonstrated in figure 11. NCC was isolated from discarded cigarette filters (DCF), one of the world’s most littered items. [36]

(20)

20 BET TGA Serum aggregation PBS dopamine

Briefly, cellulose was obtained from DCF via ethanolic extraction, NaOCl bleaching and deacetylation with NaOH in ethanol. It was found that NCC is capable of serving as a reducing agent to form AgNP’s across a wide pH range (5-10) at a temperature of 80 °C. The diameter of the silver nanoparticles ranged from 19.93 nm at pH 5 to 4.61 nm at pH 9. A Brunauer-Emmett-Teller (BET) analysis was performed to demonstrate that NCC assists with the prevention of mass aggregation during drying and therefore improve the stability of the substrate. The temperature of thermal decomposition was demonstrated by evaluating thermal gravimetric analysis (TGA) thermograms and it was found that segments of NCC degrade at temperature of 145, 225 and 271 °C, in two stages, namely the dehydration of the anhydroglucose chain segments and the chain decomposition and oxidation degradation of the NCC agent. Furthermore, it was demonstrated that NCC can be used for the detection of riboflavin, achieving an LOD of 3 x 10-7_{M. It should be noted that when saturation of}

riboflavin molecules occurred (5 μM), the fluorescence effect from the free riboflavin molecules interfered with the SERS signal leading to reduction in intensity. [36]

Lately, SERS has been explored for the analysis of biomolecules in complex samples as SERS mainly provides shorter analysis times, measurements are more economical and it often requires little to no sample preparation. Furthermore, the resulting SERS spectra are molecule- and orientation-specific and enable multiplex detection and quantification, making it a suitable technique for the analysis of biomolecules in complex samples. E. Dumont et al. (2018) recently developed a pre-aggregation strategy to overcome the gold nanoparticle stabilization effect in serum-containing samples using a salt solution. Several aggregation agents were tested such as sodium nitrate (NaNO3), potassium

chloride (NaCl), sodium chloride (NaCl) and Phosphate Buffered Saline (PBS), as shown in figure 12. The most optimal aggregation agent was found to be the concentrated PBS solution. Furthermore, the strategy was demonstrated with the quantification of dopamine (DA, a neurotransmitter involved in many important physiological behaviors) in culture media of rat pheochromocytoma (PC-12) cells. Most serum-containing samples contain proteins which can potentially form coronas around the SERS substrate. This prevents controlled aggregation of the substrate and results in a decrease of signal enhancement. A Design of Experiments (DoE) approach was used for the optimization of the sample preparation using various amounts of PBS and dopamine concentrations. [27] The gold nanoparticles were synthesized and coated with trisodium citrate according to the Lee and Meisel method [37]. DA could be quantified from 2.64 to 263.65 μM for the PC-12 cell medium (the DA basal level in biological fluids is 1 nm to 1 μM). A second matrix with a lower concentration of proteins was tested, leading to an improved sensitivity (1.32 μM). [27]

Figure 11. Schematic illustration of NCC-AgNP-RBF. Retrieved from Ref.

[36]

Figure 12. Raman spectra of DA in PC-12 cell culture medium samples using 1 M NaNO3 (blue), 1 M KCl (orange), 1 M NaCl (green), PBS (yellow) and concentrated PBS (purple) solutions. The most optimal solution was found to be

(21)

21

Increased repeatability: (reactive) inkjet printing

Recently, interest has been shown to improve the repeatability of substrate fabrication, most promisingly to be used in industry. An advanced technique for repeatable substrate fabrication is by the use of screen printing. An illustration of screen printing is shown in figure 13. This technique incorporates a cost-effective, simple and reproducible printing approach for the mass-production of SERS substrates and SERS-active chips. [38]

There are two approaches for the production of SERS active patterns, multi-run printing to build up structures until a desired number of hot spots is reached or using highly concentrated inks. However, when using a high concentration, the ink feed systems and nozzles may easily be clogged. A solution for this is by applying reactive printing methods using ink solvents. When applying reactive printing, depending on the duration of UV illumination, the size and thus the color of the colloid substrates change, thus the enhancement of certain molecules is promoted by the Raman Resonance effect. Polychromatic SERS substrates are substrates with multiple plasmon resonances in the visible range and are the most advanced substrates for different kinds of analytes. Reactive printing can be used for polychromatic path formation as it allows simultaneous resonance enhancement effects. The two most common ink solvents used for SERS chip pattern formation are silver nitrate and ammonia-based silver complexes. The drawback for silver nitrate is that it needs specific storage conditions to prevent early formation of silver particles. The drawback for ammonia-based silver complexes is that it requires stabilizers (polymer surfactants) to prevent the formation of silver mirrors that may interact with aldehydes and alcohols and thus prevent the substrate from being used in complex solvents. [38] Mavlavi Dustov et al. (2018) reported a photochemically reduced AgI_{citrate complex as a precursor}

for polycolor nanocoating. Criteria such as thermal stability, kinematic viscosity and surface tension were studied. Furthermore, the substrate was irradiated with UV light of 312 nm for a certain duration to demonstrate the polychromatic properties. Also, the size distribution was determined by TEM to ensure that the evolution of the size of colloidal silver corresponded to irradiation time. A TEM micrograph of silver colloids irradiated for 1 minute is illustrated in figure 14. To evaluate the EF value, SERRS spectra of R6G with a concentration of 10-7_{M and Methylene Blue (MB) dyes were collected by}

the assistance of irradiation using a red and a green laser. The estimated AEF value was found to be 106_{for R6G and 10}4_{for MB. It was also found that the samples provide a more beneficial enhancement}

when using the red laser, as using a higher energy laser led to degradation of the citrate anions. The uniformity and reproducibility of the substrate were determined by full profile mapping of the 1365.1 cm-1 _{characteristic band of R6G. An absolute band intensity dispersion of ~30% and band position of}

1365.1 ± 0.7 cm-1_{(5-95% quantile) were demonstrated.}_[38]

Figure 13. Theoretical scheme for a polycolor SERS platform using photoreactive silver complex inks. Reproduced from Ref. [38], supplement

S1 Screen printing Reactive printing UV light irradiation

Figure 14. TEM micrograph of silver colloids synthesized via UV-induced photochemical reduction of silver citrate

(22)

22

Chapter 3. Multiple composition-based colloidal nanomaterials

SHINERS

Three of the major challenges in substrate applications, especially in life science fields, are tunability, stability and flexibility. Apart from changing the size of colloids in suspension, the tunability and stability of the nanoparticles is generally limited. A novel concept to overcome this limitation is called Shell-isolated nanoparticle-enhanced Raman spectroscopy (SHINERS), which is created originally from the idea of other molecules ‘borrowing’ the SERS effect from the plasmonic metal. The first study of the ‘borrowing’ effect was demonstrated by van Duyne et al. in 1983. In 2010, Tian et al. developed SHINERS to overcome the limitations of materials and morphology generalities in SERS. Core-shell particles consist of gold or silver nanoparticles with a layer of ultrathin, fully enclosed, uniform and optically transparent inert shell, by employing methods such as a sodium citrate reduction method (inert silica) or an atomic layer deposition method (aluminum). The ultrathin coating generally keeps the nanoparticles from agglomeration, separates them from direct contact with the probed material an allows the nanoparticles to conform to different contours of substrates. It increases the stability and dispersibility because the coated shell can protect the nanoparticles from sintering, the effect of other reagents and oxidation. It also increases the biocompatibility as the shell can be coated with silica, polymers and even biomolecules to decrease the toxicity of the substrate. For drug delivery purposes (which are further discussed in chapter 6), the release or leaching of the core can be controlled by changing the environmental pH, ionic strength or temperature. The thickness of the shell nanoparticles is usually designed so that the shell controls the distance of the core from the surface to be studied and so that the core particle still generates a large surface enhancement, and is usually only about 2 nanometers for ultrathin shells. [39-40] Figure 15 illustrates a HRTEM image of core-shell nanoparticles with different thicknesses. [40]

SHINERS substrates are generally classified in two categories, ultrathin and thicker shells. In the ultrathin shell applications several SERS strategies are developed to improve the types of atoms to be used for the SERS effect. For example, Tian et al. (2004) employed a versatile and straight-forward wet-chemical synthesis as a ‘borrowing’ SERS strategy to expand SERS to transition metals such as platinum, palladium, rhodium, cobalt, nickel and ruthenium. The shell can also consist of multiple metals (trimetallic nanostructures) to increase the reactivity for electrocatalytic reactions. In the thicker shell applications, several SERS strategies are developed to 1) improve tunable SPR properties, 2) add markers or tags to the shell consisting of for example silica, polymers, DNA or proteins, 3) allow NIR region enhancement, 4) improve reproducibility by adding self-assembled polystyrene spheres and 5) add magnetic materials to allow suitable bio separation, biological analysis and subsequent immunoassays. [39]

Figure 15. HRTEM images of Au/SiO2 core-shell nanoparticles with different shell thicknesses. Reproduced from ref. [40].

SHINERS Advantages Ultrathin applications Thicker shell applications

(23)

23 Jiao Liu et al. (2018) recently obtained stable core-shell hexagon silver-carbon nanocomposites (Ag@CS’s) using glucose as the carbon source and stabilizer, silver nitrate as the silver source and trioctylamine (TOA) as the soft template and extractant. TOA molecules can interact with each other and congregate as micelles. The outer hydrophilic TOA group react with silver ions to form a silver amine complex at the interface. The silver ions at the interface are reduced to AgNPs by glucose deposited on the surface of the silver core and carbonized to form a carbon layer. The silver core shows the hexagon shape, while the carbon layer exhibits a uniform spherical shape, mainly because the glucose polycondensation products chose the lowest potential energy way to wrapped silver core. A laser was used with an excitation wavelength of 514 nm. The TEM image of the Ag@CS is illustrated in figure 16. The Ag@CS were shown to exhibit a narrow size distribution of around 180 nm and the shell thickness is determined to be about 15 to 20 nm. The hydrothermal carbon shell not only effectively avoids the mass aggregation of Ag nanoparticles core, but also serves as a protective layer from oxidation of the AgNPs of the SERS active nanostructure. As the immersing time lengthened from 1 month to 1 year when stored in an ethanol solution, the thickness of the hydrothermal carbon shell was significantly reduced from 20 nm to 4 nm. [41]

The composition and morphology of the thin carbon layer are similar to graphene oxide, which is proposed by the authors to be of great significance for further study. It is observed that the intensities of signals from the substrate are strongly dependent on the time that the probe molecules are in contact with the Ag@CSs. It is also found that drop casting the probe molecules onto the substrate and pure R6G powder have a SERS activity of three orders of magnitude lower than when the Ag@CS particles are immersed in the probe solution for a period of time. The calculated SSEF1 for this substrate

with R6G as Raman reporter is ~106 _{(using a concentration of 10}-6_{M) and the LLOQ was found to be}

10-9_M._[41]

Figure 16. TEM image of Ag@CS. Reproduced from Ref.

[41]

Hexagon SHINERS

(24)

24

Nanomushrooms

Jing Su et al. (2017) reported deoxyribonucleic acid (DNA)-mediated gold−silver nano-mushrooms with interior nanogaps directly synthesized and used for multiplex and simultaneous SERS detection of various DNA and ribonucleic Acid (RNA) targets. The oligonucleotide-coated AuNPs were fabricated based on the procedure as per S. J. Hurst et al. (2006) [42]. The modified AuNP solution was mixed with PVP, sodium ascorbate and 1 x 10-3_{M AgNO}

3. Lastly the solution was incubated, washed and

suspended in a phosphate buffer. Capture-DNA-coated matrix metalloproteases (MMPs) were prepared and used to capture the RNA and DNA targets. After preparation, the probe-DNA-functionalized Au–Ag SERS probe was added to the solution, incubated and washed. Figure 17 shows a schematic illustration of the multiplex ‘sandwich-type’ nano-mushroom strategy for the detection of virus gene DNA/RNA. The mushroom is shown as a substrate, consisting of the silver head and the oligonucleotide-coated AuNPs bottom with gap DNA strings connecting to the silver and probe DNA used as a spacer. The probe DNA is connected to target RNA/DNA strings with MMPs and labeled with rhodamine (ROX), which are bound to a silicon wafer. The specific oligonucleotide-labeled with X-rhodamine (ROX) was designed to function as a probe DNA for the recognition of hepatitis A virus (HAV) DNA fragments. miRNA-21 fragments (microRNA encoded by the MIR21 gene) were used to demonstrate the multiplicity of the application. In the presence of the target DNA, a large number of Au−Ag nanomushrooms were coupled with the MMPs due to DNA hybridization. [43]

The diameters of the silver head and gold bottom were 100 nm and 50 nm respectively and a highly uniform interior nanogap indicating by time-independent Raman spectra. It was found that the substrate can be stored at 4 °C for 3 months without aggregation. The blank control sample spectra had no bands, indicating that the Au-Ag nanomushrooms were not bound to the MMPs. HAV DNA targets could be detected in the presence of interfering DNA sequences present with a more than 103

-fold greater concentration in the sample. The lowest concentration of miRNA-21 target detected was 10 fM and 1 pM for each DNA target. The enhancement factor was found to be about 109_{. The}

nanomushrooms were able to effectively differentiate miRNA-21 from random RNA at pM level. For the selectivity of the substrate, three miRNA targets (ROX (miRNA-21), 4-ABT (miRNA-31) and CY3 (miRNA-41)) were captured by the nanomushrooms and detected simultaneously in a single sample in a dynamic range of five orders. Additionally, the substrate performance in a buffer solution was compared to the performance in a serum solution, demonstrated no significant differences in the results. [43]

Figure 17. Schematic illustration of Multiplex Analytical Strategy for Virus Gene DNA/RNA with ready-to-use nano-mushrooms. Reproduced from

Ref. [43] Nano- mushrooms PVP MMPs DNA/RNA detection

(25)

25

Chapter 4. Planar composite-metal substrates

The use of colloidal nanoparticles such as silver and gold have the drawback of mass aggregation, in which the substrate may not successfully create the hotspots needed to achieve a higher signal and thus actually decrease the signal. Also, colloidal nanoparticles are randomly distributed which results into irregular substrate enhancement. To overcome these drawbacks, apart from the applications discussed in chapter 2 and 3, spreaders containing a large surface area support can be used for compacting and stabilizing the nanoparticles. [44]

Graphene sheets

Tawfik A. Saleh et al. (2018) investigated Silver loaded graphene as a substrate for sensing 2-thiouracil, a substance used for the treatment of hyperthyroidism, using surface-enhanced Raman scattering. The silver nanoparticles are seeded on the graphene sheets to obtain uniformly distributed silver nanoparticles, as illustrated in Figure 18a. The Ag/G nanoparticles were synthesized using a reduction method with NaBH4 used as a reducing agent. X-ray photoelectron spectroscopy (XPS) was employed

to analyse the interaction between the components of the prepared material. The results indicated that the silver nanoparticles were successfully deposited on the graphene nanosheets.

TEM imaging and mapping tools were used to analyse the uniformity of distribution and to estimate the size of the silver nanoparticles. Amounts of silver were added in repeated steps in five cycles until a uniform loading of silver was obtained. The average size of the silver nanoparticles was estimated to be 15 nm. The maximum analytical enhancement factor (AEF) for 2-thiouracil with a concentration of 1 x 10-2_{M was determined to be 2.9 x 10}3_{(at 815 cm}-1_{). The LOD was found to be 10 nM.}_[44]

A recent trend has emerged to produce self-cleaning, re-usable substrates. Lu-lu Qu et al. prepared a multifunctional TiO2-Au-RGO substrate for recyclable SERS applications. A schematic demonstration of

the recycling procedure is illustrated in figure 18b. For the preparation, a facile two-step hydrothermal route was applied, consisting of the modification of TiO2 nanoparticles on a planar graphene oxide sheet (TiO2-rGO) and depositing gold colloidal nanoparticles on the TiO2-rGO nanocomposites by

in-situ reduction of chloroauric acid in an ethanol-water solution. The application was demonstrated on Rhodamine-6G (R6G) and provided a LOD of 1.2 x 10-10_{M and a LLOQ of 2.0 x 10}-10_{M. For}

concentrations above 5.0 x 10-8_{a non-linear behavior was observed, possibly due to saturation of the}

adsorbed R6G on the nanocomposites. Also, the reproducibility was demonstrated by applying a recycling method using photocatalytic degradation of the adsorbed analytes, followed by a self-cleaning method centrifuging the nanocomposite and washing the substrate with deionized water to remove the residual molecules and ions. The concentration R6G decreased (based on a pseudo-first-order kinetics reaction) with the increase of the irradiation time (with a maximum of 80 minutes) and it was observed that the TiO2-Au-RGO microstructure did not change in the photocatalytic process.

When the TiO2-Au-rGO nanocomposite was immersed in R6G solution after cleaning, the SERS signal was fully recovered. The procedure was repeated four times, observing a good repeatability. [45] Graphene sheets 2-thiouracil Re-usable graphene oxide sheets

Figure 18. (a) TEM image of silver nanoparticles (15 nm) deposited on a graphene sheet and (b) schematic illustration of the re-usable multifunctional TiO2-Au-RGO substrate. Retrieved from Ref. [44] and[45] respectively.

(26)

26

Paper-based SERS

Paper-based SERS is a SERS type that consists of utilizing filter or printing paper with colloidal plasmonic nanoparticles to provide signal amplification. The key features of paper-based SERS mainly are easy to apply and low-cost application. Furthermore, Raman acquisition can be performed immediately without the need for the sample to dry and the species can be analysed in a sensitive and non-destructive manner. [46] An optimized dry substrate and wet analyte configuration may allow paper-based SERS substrates to be stored before use and Raman acquisition could be performed immediately without the need for the sample to dry, making it an interesting target for industrial purposes. Shuai He et al. (2017) recently developed a gold nanostar (AuNS) immobilized paper-based SERS substrate, using crystal violet (CV) as a Raman reporter. As excitation source, a laser with an excitation wavelength of 785 nm was used. Multiple drops of ~100 pM sodium-citrate treated colloidal AuNSs were dropped on paper. The morphology and performance of printing paper and common laboratory filter paper were investigated. The filter paper with high porosity increased the absorbance of the AuNS colloids, while printing paper with lower porosity increased the AuNS holdout in the paper, ensuring that the colloids permeate less easily into the paper. The higher porosity in the filter paper was preferred to be used as it generates a higher SERS EF due to light and colloid penetration. The detection limit was found to be 1 nM, which is lower than two commercial Au/Ag-based SERS chips. The maximum SSEF1 of CV at a low concentration of 1 nM was calculated to be 1.2 x 107. It is

noteworthy that many paper-based SERS studies dry the analytes before Raman acquisition. In this study it was found that the SERS enhancement was stronger when the analyte was still wet. It was found that the citrate-treated AuNS immobilized on the filter paper was less uniformly distributed compared to the non-treated citrate AuNS immobilized on the filter paper. It was also demonstrated and the negative surface charge of the paper provides charge stabilization on colloidal AuNS, although this might hinder their immobilization on the paper substrate subsequently. [46] The demonstration of the repeatability of the substrate were not mentioned in the article.

Paper-based SERS

Printing paper Filter paper