Detection of Contaminants in Water Using Surface Enhanced Raman Spectroscopy

(1)

Detection of Contaminants in Water Using Surface-Enhanced Raman Spectroscopy

Freja Hansson

Materials Engineering, master's level (120 credits) 2021

Luleå University of Technology

Department of Engineering Sciences and Mathematics

(2)

SUPERVISORS Dr. Martin Wersäll

Research Institutes of Sweden, RISE

Prof. Timur Shegai

Chalmers University of Technology

EXAMINATOR

Prof. Farid Akhtar

Luleå University of Technology

(3)

Abstract

Due to deteriorating water quality and the world’s increasing demand for clean water, the need for cheap, easy and portable techniques to characterize and quantify pollutants in waters is urgent. Hence, surface-enhanced Raman spectroscopy (SERS) have gained considerable attention in this field. Atrazine and bentazon are two of the most occurring pesticides causing pollution in Sweden, and where therefore examined in this study, along with 4-mercaptopyridine (mpy) as a reference molecule. In this project, silver and gold nanoparticles where synthesised and used as SERS substrates for detection of contaminants in water by using a handheld Raman device provided by Serstech AB. Sodium chloride (NaCl) and magnesium sulfate (MgSO4) where used as aggregation agents allowing the nanoparticles to form hot spots. Mpy was detected to 0.5 nM and an enhancement factor of 10⁸ using silver nanoparticles aggregated with NaCl was obtained. No Raman signal was obtained from atrazine nor bentazon using the handheld Raman device with silver nanoparticles aggregated with NaCl. Therefore the Raman cross-section of the probe molecules where investigated using the handheld Raman device and a conventional Raman device. Bentazon was not detectable using the handheld Raman device but detectable using a conventional Raman device. Atrazine was detectable at high concentrations i.e. atrazine powder using the handheld Raman device and detectable at 100 nM using a conventional Raman device. Since bentazon was not detectable with the handheld Raman device, more focus was put on getting a detectable signal from atrazine using the handheld Raman device. Investigation of the adsorption of atrazine and bentazon to the silver nanoparticle surface was performed. Due to the weaker adsorption to the nanoparticle surface, MgSO4 was used aggregation agent instead of NaCl with mpy, atrazine and bentazon. Mpy was detectable using MgSO4as aggregation agent, atrazine and bentazon was not.

Measurements of mpy, atrazine and bentazon without any salt was performed. For these measurements, no detectable signal from neither molecule was obtained, indicating that the formation of hot spots is necessary to obtained a detectable Raman signal. Measurements of mpy and atrazine with gold nanostars where performed. Enhancement factor using the gold nanostars was calculated to 10⁷, and a detectrable signal from mpy was obtained, not from atrazine. Measurements of atrazine and mpy simultaneously was performed, where mpy peaks was observed but no atrazine peaks. The affinity of the probe molecule and the nanoparticle is crucial to obtain a detectable signal. This study inducates that both the chemical enhancement and electromagnetic enhancement are needed to obtain a detectable signal. For that, strongly binding species is necessary. Consider- ing the simplicity of this method and the limited optimization efforts, there is plenty of room for improvements, including different probe molecules and different SERS substrates. With the right conditions, the evaluated technique reveals a promising and accessible method using a commercially available handheld Raman spectrometer for detection and quantification of contaminants in water.

(4)

Acknowledgements

Firstly, I would like to thank my supervisors Dr. Martin Wersäll from RISE for giving me the opportunity to work with this thesis, Prof. Timur Shegai from Chalmers for always giving me guidance with his knowledge within this field, and Farid Akhtar from Luleå for his support and supervision.

I would also like to thank my colleague Kasper Eliasson, for his very appreciated help during the first months at Chalmers. It has been valuable to have such a professional person to work with.

Also, I would like to thank my colleagues Betül Kücüköz, Adriana Canales Ramos, Aleksandr Po- liakov and Battulga Munkhbat in the Nano and Biophysics division at Chalmers for all their help and supervision.

Finally, my family and friends deserves a special thanks, for your endless support.

(5)

List of Figures

1 Molecular structure of atrazine. . . 5

2 Molecular structure of bentazone. . . 5

3 Molecular structure of 4-mercaptopyridine. . . 6

4 Quantum principle of Raman effect. . . 8

5 Lorentzian function fitted to the 1000 cm ⁻¹ peak of 4- mercaptopyridine in a solution containing AgNPs SERS substrates, 2,5 nM mpy, and 15 nM NaCl. The spectra is an average of 10 spectrum with 5 second integration time. . . 18

6 SEM images of AgNPs with and without aggregation agent . . . 19

7 SEM images of AuNSs . . . 20

8 Extinction spectra of AgNPs and AuNSs. . . 20

9 SERS measurements of AgNPs with 15 mM NaCl and 15 mM MgSO4 . . . 21

10 Mean peak intensity of the peaks at 1006 and 1093 cm⁻¹of mpy with varying NaCl concentration. . . 22

11 Mean peak intensity of the peaks at 1006 and 1093 cm⁻¹ for mpy with varying MgSO₄ concentration. . . 22

12 4-mercaptopyridine spectra of pure mpy crystals and 10 nM mpy concentration in solution containing AgNPs and 10 mM NaCl. . . 24

13 4-mercaptopyridine spectra of 1 M mpy in solution without enhancement and 5 nM in solution containing NP and 15 mM NaCl. . . 24

14 Raman spectra of 0.1, 0.25, 0.5, 0.75 nM mpy with Lorentzian fit to the peak at 1006 and 1093 cm⁻¹. . . 25

15 SERS measurements of atrazine and bentazon. Blank spectra subtracted. . . 26

16 Raman spectra of mpy, atrazine and bentazon powders using the handheld Raman device. . . 27

17 Raman spectra of atrazine powder and 100 nM atrazine dissolved in ethanol with the AgNPs using the conventional Raman device. Lorentzian function fitted to the peaks at 679, 961, 1061 and 1164 cm⁻¹. . . 28

18 Raman spectra of bentazon powder using the conventional Raman device. . . 29

19 Raman spectra of mpy, bentazon, atrazine and blank MgSO4 sample with AgNPs aggregated and MgSO4 obtained using the handheld Raman device. . . 30

20 Raman spectra of atrazine, bentazon and mpy without any agglomeration agent obtained using the handheld device. . . 31

21 Raman spectra of 1 mM mpy and 100 nM mpy with AuNSs obtained using the handheld Raman device. . . 32

(8)

22 Raman spectra of mpy and atrazine with the AUNSs obtained using the handheld Raman device. Lorentzian fit to the peaks at 1000 and 1092 cm⁻¹for mpy, and 997, 1164 and 1588 cm⁻¹ for atrazine. . . 33 23 SERS measurements of atrazine and mpy simultaneously with AgNPs. . . 34 24 Raman spectra of mpy with AgNPs aggregated with (a) NaCl and (b) MgSO4 ob-

tained with the handheld Raman device. . . 34

(9)

List of Abbreviations

SERS Surface Enhanced Raman Spectroscopy Mpy 4-mercaptopyridine

NaCl Sodium chloride MgSO4 Magnesium sulphate IR Infrared spectroscopy NP Nanoparticle

BAM 6-Dichlorobenzamide

Ag Silver

Au Gold

Cu Copper

LSPR Localized surface plasmon resonance EM Electromagnetic enhancement CE Chemical enhancement AgNP Silver nanoparticle AuNS Gold nanostar EF Enhancement factor AgNO3 Silver nitrate NaBH4 Sodium borohydride HAuCl₄ Chloroauric acid HCl Hydrochloric acid

CTAB Hexadecyltrimethylammonium bromide SEM Scanning electron microscope

AgCl Silver chloride

(10)

Nomenclature

α Polarizability

γ Relaxation frequency µ Dipole moment

ν Frequency

Ω Angle

ω Angular frequency σ Raman cross-section ε Relative permittivity

A Average electromagnetic surface intensity enhancement factors a Refractive index

C Concentration

c Speed of light in vacuum E Electromagnetic filed h Planck constant I Intesity

kB Boltzmann constant L Geometrical form m Electron mass N Number of molecules n Electron density q Bond length T Temperature

V Volume

v Potential barrier

(11)

1 Background

This master’s degree project is a collaboration between Research Institutes of Sweden, RISE and Chalmers University of Technology. It has been performed in the Nano and Biophysics division at Chalmers in Gothenburg, lead by Prof. Timur Shegai. A relatively cheap and commercially available handheld Raman spectrometer from Serstech AB will be used for detection of three different molecules; 4-mercaptopyridine, atrazin and bentazon. Silver and gold nanoparticles will be synthesised by following a bottom-up approach and used as the SERS substrates. The aim of this study is to evaluate the system for detection of real-life contaminants found in waters in Sweden and to realize the possibilities and limitations of using this system for real-life environmental analysis.

Recently, water quality has become a critical concern due to the impact on the human health, aquatic environments, wetlands, and agricultural land. Growth of human populations, climate change and industrial and agricultural activities have been identified as the main causes to a de- teriorated water quality [1]. Therefore, development of techniques to characterize and quantify pollutants in waters is urgent. Complex mixtures of chemicals from pharmaceuticals, pesticides from agricultural industry, and industrial chemicals are some of the major sources to pollution in water [1]. Determination and characterization of these pollutants are usually carried out using high-performance liquid chromatography, gas chromatography, or mass spectrometry techniques.

These techniques have have high sensitivity and specificity. However, they usually require complex and expensive equipment and laborious operations [1]. Therefore, simpler and portable techniques with high sensitivity needs to be developed. Portable detection devices would provide the possibility to do in-field analysis. Due to the sensitivity to chemical detection, SERS have gained attention in this field.

Since it’s discovery in the early 1970’s, SERS has gained considerable attention in numerous fields of research. This due to the possibility to overcome some of the issues associated with conventional Raman spectroscopy, including low-sensitivity [2]. Raman spectroscopy is a non-destructive technique that allows the characterization of numerous materials. The Raman spectroscopy technique is based on the observation of the intrinsic vibrational frequency of the studied material.

This vibration frequency of matter depends on elastic constants of the chemical bonds and atom mass. However, the use of Raman spectroscopy is often limited by the weak Raman intensity of molecules. By introducing metallic nanoparticles (NP) , field enhancement on the surface of the particles occurs, which amplifies the Raman intensity of the molecules adsorbed on the surface of the particle [3]. As SERS provides information about the molecular structure and composition of a sample, the development of SERS has begun to emerge a promising tool in the fields of detection of environmental pollutants, medicine, materials science, art conservation, airport security, phar- maceutical analysis, and many others [4–6].

(12)

Despite the complexity of SERS, remarkable analytical achievements in the field of water analysis have been done [7]. In highly controlled environments, contaminants have been detected down to femtomolar concentrations [7]. In addition, SERS can detect multiple contaminants simultaneously, which most existing detection strategies cannot achieve [7]. In this project, SERS measurements of aqueous species will be carried out by dissolving the probe molecule in an aqueous phase containing nanoparticles. The results from SERS measurements depends mainly on three factors: 1) The enhancement of electromagnetic field generated by the plasmonic nanostructure;

2) The intrinsic Raman properties of the probe molecule; and 3) The affinity of the molecule for the plasmonic surface [8].

1.1 Raman Active Vibrations

For a molecule to be Raman active, molecular vibration must cause a change in polarizability of the molecule. For a molecule to be infrared (IR) active, the vibration causes a change in the dipole moment of the molecule. The symmetry of a molecule determines whether a molecule is Raman active or IR active. In general, symmetric or in-plane vibrations are most Raman active, and asymmetric out-of-phase vibrations are IR active [9]. In that regard, Raman and IR spectroscopy are complementary. The symmetry of a molecule include planes of symmetry, center of symmetry and axes of symmetry. A molecule with a center of symmetry structure, vibrations that retain the center of symmetry generate a change in the polarizability but no change in a dipole moment, and are thereby IR inactive and thereby Raman active [9]. Commonly, small molecules without electron-rich atoms have small Raman cross-sections. Consequently, larger molecules with electron- rich atoms, makes the molecule more polarizable and has therefore higher Raman cross-sections [8].

1.2 Molecule - Nanoparticle Interactions

In Raman spectroscopy, high Raman cross-sections is necessary to produce a Raman signal. How- ever, in SERS an additional requirement is needed. The highest SERS signal is obtained when the probe molecule is adsorbed or close to the surface of the plasmonic NP. Maximum SERS intensity would be expected when the polarization vector of the molecule is normal to the surface. If the angle between the polarizing vector and the substrate approaches to 0^◦, the SERS intensity falls off rapidly. The affinity of the molecule and NP can be optimized by chemical bonding. Molecules containing sulfur, nitrogen and oxygen groups show high affinity to gold and silver surfaces and bind well to the metal surface [3]. The Raman intensity peak decreases with exponentially with distance between the molecule and NP. This due to the exponentially decrease in electromagnetic field around the NP [8]. The Raman signal is enhanced only in close vicinity up to roughly 10 nm from the NP [3]. Colloidal silver or gold nanoparticles are commonly used as SERS measure-

(13)

ments. This due to their simple and cost-effective preparation and excellent signal enhancement properties.

1.3 Selection of Probe Molecules

As mentioned, pesticides is a group of chemicals that has been identified as a major source of deteriorating water quality. Along with a growing population comes a growing demand for food, this has increased the need for cultivated areas and an increase in productivity in agriculture. As a consequence, the use of pesticides have increased. Pesticides includes fungicides, herbicides and insecticides. Unfortunately many of the pesticides used are toxic to organisms other than those intended to fight [10]. The use of pesticides leads to reduced biodiversity in the agricultural land- scape. Pesticides are spread further in the ecosystems via water, soil, air and in the food chains and can thus have effects long after the spread at distances far from the fields where they are spread. This can cause serious changes in ecosystems. Some of the pesticides used today are not biodegradable and also have limited solubility in aqueous solution which can lead to environmental contamination, especially contamination of groundwater and drinking water [11].

The presence of pesticides causes different types of problems in different types of water. The presence of pesticides in groundwater is problematic since groundwater is usually used as drinking water, which could affect the human health. In watercourses, aquatic organisms are damaged and biodiversity is affected in contact with pesticides [10].

In 2015, the Swedish government requested a broad screening of environmental toxins in water where the focus was on studies of perfluorinated substances and pesticides. The Competence Cen- ter for Chemical Pesticides (CKB) at the Swedish University of Agricultural Sciences (SLU) was commissioned by The Swedish Environmental Protection Agency to coordinate and implement the sampling campaign regarding pesticides in Sweden [12]. In the same report, chemical analyzes of pesticides were performed by the Laboratory of Organic Environmental Chemistry at SLU. The total number of samples in surface water was 157. The sampling of groundwater was performed in individual drinking water wells and was done in 54 wells and 18 waterworks. Results show that 131 and 108 substances were analyzed and in surface water and in groundwater, respectively. The results were compared with limit values and guideline values for drinking water, and also groundwater for the protection of aquatic organisms in surface waters. Results show that approximately 23 % of all samples in surface water contained pesticides above or equal to the limit value for total pesticide content in drinking water of 0,5 µg/l. For the groundwater samples, approximately 45

% of all samples contained at least one substance above or equal to the limit value for individual pesticide in drinking water 0,1 µg/l [12].

(14)

In surface water, the most frequently found pesticide where herbicides. Bentazone, glyphosate and AMPA were the three most common substances in surface water and were also the most common substances to be present in concentrations above or equal to the limit value for total pesticide content in drinking water of 0,1 µg/l. In groundwater, the three most common substances to be found in individual water wells was BAM (2,6-Dichlorobenzamide), atrazine and atrazindesetyl (atrazine degradation product). These substances has been banned to sell in Sweden since 1989.

The fourth most common substance in individual wells was bentazone which is is still permitted for use. In waterworks, BAM was followed by atrazine, bentazone and atrazindesetyl as the most common substances to be found [12].

1.3.1 Atrazine

As mentioned in the previous section, atrazine is one of the most common contaminants found in ground waters in Sweden in 2015 [12]. Atrazine is a herbicide of the triazine class, which acts as an inhibitor of photosynthesis, and has been found to affect reproduction of aquatic organisms, and may also have endocrine disrupting and carcinogenic effects depending on concentration [11, 13].

Atrazine persists under cool, dry conditions, in a stable pH environments, and therefore, atrazine has been banned in many countries with cool climate with with fine textured soils. Atrazine has been banned in Sweden since 1989, but due to its slow degradation time it was frequently found in ground waters.

Due to its low solubility in water, it is not easily biodegraded [11]. In addition, the degradation of pesticide residues is commonly slower in groundwater than in surface water, that because organic materials and microorganisms are less common below groundwater levels. Previous studies has been done on detection of atrazine using SERS. As mentioned previously, molecules that contain centre of symmetry structure and also contain sulfur, nitrogen and oxygen groups shows a high Raman cross-section and high affinity to gold and silver surfaces, respectively.

Figure 1 shows the molecular structure of atrazine, which contains nitrogen groups and also a benzene ring containing nitrogen. According to literature, nitrogen groups bind well to Ag (silver) and Au (gold) and the symmetric structure of the benzene ring shows a high Raman cross-section, making this contaminant interesting to investigate using SERS. In a study by Rubira et.al. [11], atrazine was detected with SERS using Ag colloids down to 5 · 10⁻¹² M.

(15)

Figure 1: Molecular structure of atrazine.

1.3.2 Bentazon

Bentazon was one of the most frequently found pesticide in surface waters and in ground waters in Sweden in 2015. It was found in 82 % of all samples [12]. BAM, atrazine and bentazone are the three substances that was found to most frequently exceed the limit value of 0,1 µg/l to be used as a source to drinking water [12]. Bentazon is a selective herbicide which interferes with the photosynthesis of the weeds and is allowed to use in Sweden. Since the substance frequently exceeds their respective guideline values it can be considered as a potential problem. Bentazon has high solubility in water (500 mg/L at 20^◦C), it has also low adsorption coefficient values, which leads to high mobile in soil. Because of is high mobility is soil, it tends to leach during extreme rainfalls and is expected to pass through the soil and via cracks to the underlying aquifer, causing contamination of the ground water [14]. The number of studies on detection of bentazone using SERS are very limited. However, as shown in Figure 2, bentazone contains nitrogen groups and also a benzene ring making this molecule interesting for SERS measurements.

Figure 2: Molecular structure of bentazone.

1.3.3 4-mercaptopyridine

4-mercaptopyridine is not a pesticide but was used as a reference probe molecule since it is one of the most commonly studied probes for SERS and because of the thiol group, it adsorbs strongly onto silver. The molecular structure of mpy is shown in Figure 3.

(16)

Figure 3: Molecular structure of 4-mercaptopyridine.

(17)

2 Theroy

2.1 Principles of Raman Spectroscopy

In 1928, the Indian physicist C.V. Raman discovered experimentally that light can change its frequency due to interaction of matter. This phenomena was then termed as the Raman effect or Raman scattering, which is the basic principle in Raman spectroscopy. Due to this discovery, Raman was awarded the Noble Prize in Physics in 1930 "for his work on the scattering of light and for the discovery of the effect named after him." [15].

Raman spectroscopy is used to determine the vibrational modes of molecules and it has its origin in the observation of transitions of molecular vibrational levels. The patterns of these vibrational levels are unique for a specific molecule and this information is therefore sufficient to characterize a specific molecule [16].

Scattering of light can be either elastic or inelastic. In elastic scattering, the incident photon has the same frequency as the scattered one, this is called Rayleigh scattering. Due to the elastic nature of the Rayleigh scattering, this process does not give any information about the molecule it interacts with. In inelastic scattering, the incident photon does not have the same frequency as the scattered one, this process is called Raman scattering. In Raman scattering the change of the photon’s frequency is due to the interactions with the vibrational states of the molecules.

In Raman scattering, two inelastic processes can occur. The incident photon can either excites the molecular vibration and is thus is scattered with the corresponding difference in energy, this is referred to as Stokes scattering. Or, the incident photon acquire vibrational energy and is scattered with a higher energy, this is referred to as Anti-Stokes scattering. This will be described in more detail in the following section.

2.1.1 Quantum theory of Raman effect

This quantum model is based on that the molecules exist in quantized vibrational energy levels corresponding to possible vibrational energy states of the molecule. When a molecule is subject to radiation with energy hνi, the incident photons will collide with the molecule. During collision, the photons can either be scattered elastically or inelastically. During elastic scattering, the incident photon will have the same energy as the scattered one (hνi) and no exchange in energy between photon and molecule will occur (Rayleigh scattering). During inelastic scattering, the incident photons have a different energy than the scattered one, and an exchange in energy between photon and molecule has in this case occurred (Raman scattering). As a result, the molecule gain or lose energy ∆E, which corresponds to the difference in the vibrational or rotational energy levels of the molecule. In quantum mechanical terms, the collision begins with an excitation of the molecule to an virtual state, which is lower in energy than that of electronic states, followed by relaxation by emitting a photon [17]. When the incident light interacts with the molecule, two scenarios can

(18)

occur. Either the molecule will make an upward transition between the molecular energy levels and by that emit an photon with lower frequency than the incident one to be emitted (νf < νi), this is called Stokes Raman scattering. Or, the molecule will make an downward transition between the molecular energy levels and thus emit a photon with higher frequency than the incident photon (νi < νf), this is called Anti-Stokes Raman scattering. The frequencies νf make up the Raman spectrum, see Figure 4 [4, 16].

Figure 4: Quantum principle of Raman effect.

2.1.2 Classical theory of Raman effect

Raman scattering has its origin in the polarizability of the molecule. When a molecule is placed in an electric field, it will suffer from distortion. The electrons and nuclei of the molecule get displaced relative to each other. The positively charged nuclei attracts towards the negative pole of the electromagnetic field and the electrons to the positive pole of the field. Due to the separation of the charges, an electric dipole moment is induced in the molecule, this is called polarization.

The induced dipole moment (µ) depends of the strength of the field (E) and on the polarizability of the molecule (α) and can be described as:

µ = αE (1)

If the molecule is subject to an electromagnetic wave with frequency ν, the electric field experienced by the molecule varies as:

E = E0cos2πνt (2)

Where E0is the amplitude of the electromagnetic wave. Thus, the induced dipole moment under- goes oscillations of frequency ν and can be expressed by combining equation 1 and 2 as:

µ = αE0cos2πνt (3)

Equation 3 implies that interaction of electromagnetic radiation of frequency ν induces a dipole moment that oscillates and emits radiation of the same frequency. Such an oscillating dipole

(19)

moment described in equation 3 emits photons of its own oscillating frequency and this gives the classical explanation of the Rayleigh scattering. The ability to interfere with the local electron cloud depends on the relative location of the individual atoms of the molecule. Thus, the polarizability is a function of the positions of the atoms and therefore the polarizability changes with small displacement from equilibrium position, such as vibrations and rotations of the molecule. The polarizability can thus be described as:

α = α0+ (q − q0) ∂α

∂q

!

(4)

where α0 is the equilibrium polarizability, q0 and q are bond lengths at equilibrium position and any instant, respectively. If a molecule has a harmonic motion, the displacement can be described as:

q − q0= qmaxcos2πνvibt (5)

where q_maxis the maximum distance between atoms relative to their equilibrium position and νvibis the vibrational frequency of the molecule. Combining equation 4 and 5 gives the following:

α = α₀+ ∂α

∂q

!

q_maxcos2πν_vibt (6)

Combining equation 13 and 3 gives the following expression:

µ = E₀cos2πνt

α₀+ ∂α

∂q

!

q_maxcos2πν_vibt

(7)

µ = α0E0cos2πνt + E0cos2πνt ∂α

∂q

!

qmaxcos2πνvibt (8)

µ = α0E0cos2πνt +E0

2 qmax

∂α

∂q

! cosh

2π(ν − νvib)ti +E0

2 qmax

∂α

∂q

! cosh

2π(ν + νvib)ti (9)

The first term in equation 9 corresponds to the Rayleigh scattering which has a excitation frequency ν which is the same frequency as the scattered photon, the second term is the Stokes scattering (ν − νvib) in which the scattered photon has a lower frequency than the incident photon, and the third term is the Anti-Stokes scattering (ν + νvib) in which the scattered photon has a higher frequency than the incident photon. In both of the latter cases, the incident excitation frequency has been modulated by the vibrational frequency of the bonds in the molecule and scattered at a different frequency. For a molecule to be Raman active, the electron density in the molecule must distort from its regular shape, and a molecule must experience a change in polarizability during a vibration [17].

One of the major drawbacks of this conventional Raman spectroscopy is the low Raman signal due to the low probability of the Raman process. The probability depends on the optical cross section of 10⁻¹¹− 10⁻¹⁵ nm⁻² (the area over which incident photons are converted into emitted

(20)

Raman photons), which in turn depends on whether the process is resonant or not with the transitions between ground and excited electronic states of the molecule [4].

The resulting low Raman signal prevents it from expanding in many practical applications. How- ever, there are several processes which can be used to enhance the sensitivity of a Raman signals.

The initial absorption process is proportional to the local electric field at the molecule, by using plasmonic NPs, the local electric field intensity can be greatly amplified relative to the incident light. This results in a million-fold or even higher enhanced Raman signal, which is the principle of SERS [4, 18].

2.2 Plasmas and Plasmonics

When considering the optical properties, metals can be seen as plasmas with positively charged ion cores surrounded with freely mobile conduction electrons. The phenomena of collective oscillations of the electrons caused by the interactions of light has become known by the name of plasmonics. During interactions with a incident electromagnetic filed, collective displacement of surface electrons will occur. This displacement will give rise to a restoring force altered by the change in charge distribution between the displaced electron cloud and positively charged lattice [19]. These forces results in coherent periodic oscillations of the surface charges, which can be greatly enhanced at specific frequencies. This phenomena of oscillations in surface charge densities are referred to as localized surface plasmon resonances (LSPR). The free electrons in a noble metal is free to move trough the material. When the mean free path in the material is smaller than the metal particle, no scattering occurs from the bulk and all interactions with the incoming light occurs on the surface of the particle. Thus, when the free electrons of these particles are coupled to a light source with larger wavelength than the size of the NPs, it can set up standing resonance conditions and they start to oscillate collectively on the surface of metal NPs [20].

When plasmomic materials are coupled with a light source, these forces will give rise to coherent periodic oscillations of the surface conduction electrons which can be enhanced at certain frequencies of the light. For a bulk plasma this plasmon frequency, ωp, is given by:

ω_p²= ne²

ε0m (10)

where n is the electron density, e electron charge, m electron mass and ε₀ the relative permittivity of free space. If the incident light has a lower frequency than the plasma frequency, the electrons act to screen out the incident light, resulting in reflection of the incident light. If the light has a higher frequency than the plasma frequency the electrons do not act to screen out the light, but instead the light will be transmitted. The permittivity of a material consists of an imaginary component that includes the dissipative process of the electrons during their movement within the material. In addition to plasma frequency, a damping term where the frequency dependent relative permittivity in a metal, ε(ω) can be expressed as:

(21)

ε(ω) = 1 − ω²_p

ω²+ iγω (11)

where γ is the relaxation frequency of the metal. The net polarizability, α for spherical metallic nanoparticles can be expressed as following:

α_x,y,z= V ε(ω) − ε_d

εd+ Lx,y,z[ε(ω) − εd] (12)

where V is the geometrical volume of the particle, L_x,y,zis the geometrical form factor in the x,y and z direction, ε(ω) is the relative permittivity of the metal and εd is the relative permittivity of the surrounding dielectric medium. However, for more accurate theoretical estimations, radiation damping and dynamic depolarization needs to be considered. The polarizability expression in equation 12 can be modified according to the modified long-wavelength approximation (MLWA), and can be expressed as:

αM LW A(ω) = α_x,y,z(ω) 4π −_r^k²

nαx,y,z(ω) − i²₃k³αx,y,z(ω) (13) where rn ∈ rx, ry, rz are lengths of each axis of the metallic sphere. The second term in the denominator is the enhancement maximum for particles with small finite volumes and the third term is the radiation damping. From equation 13, scattering, σsca(ω) and extinction, σext(ω) cross-sections can be calculated using the following expressions

σ_sca(ω) = ω⁴a⁴ 6πc⁴

α_{M LW A}(ω)

2 (14)

σext(ω) = ωa

c Im(αM LW A(ω)) (15)

where a is the refractive index of the surrounding medium c is the speed of light in vacuum.

The absorption cross section can thereby be calculated as

σ_abs(ω) = σ_ext(ω) − σ_sca(ω) (16)

These cross-sections reaches a maximum at certain LSPR frequencies

ωLSP R= ωp

q1 + (_L ¹

x,y,z−1)n²

(17)

At this specific frequency, the interactions between the surface charges of the particle and the electromagnetic filed reaches a maximum and the electromagnetic field can be coupled with the oscillating surface charges in the metallic nanoparticles [21].

2.3 Principles of Surface-Enhanced Raman Spectroscopy

The main limitation of Raman spectroscopy is the weak phenomenon resulting in weak Raman signal. About one in 10⁷ photons undergo Raman scattering. In addition, the weak Raman

(22)

signal is easily obscured by fluorescence. The idea of SERS is to amplify the Raman signal from molecules. The field of research on SERS and its applications is growing rapidly. The field ranges from different target molecules to different substrate materials. A large number of variables are involved in SERS measurements. During the SERS process, the molecules undergo much higher scattering efficiencies when they are adsorbed onto the surface of a metal NP. Depending on the SERS substrate, the Raman signal can be amplified by several orders of magnitude [17]. In 1974, the first SERS phenomena was observed. However, at this time, the authors did not know the the origin of the observed phenomena. Since its discovery in 1977, the interest in and the use of SERS has grown exponentially [22]. The field of research on SERS and its applications is growing rapidly. The field ranges from different target molecules to different substrate materials. Nowadays, it is generally agreed that there are two main mechanisms contributing to the enhanced Raman signal, electromagnetic enhancement (EM) and chemical enhancement (CE). The electromagnetic enhancement corresponds to the enhanced electromagnetic field at or near the metal surface. These enhanced fields around the particle arise from coupling of the incident electromagnetic field and the surface plasmons of the particle. The mechanisms of CE depends on the chemical interaction between probe molecules and the noble metal particle. The CE is said to contribute only around two to three orders of magnitude. The EM contribute around six orders of magnitude or higher, and is therefore the most dominating mechanism that contributes to the enhanced Raman signal [17]. The SERS intensity can be expressed as follows

ISERS= IeNsurΩAe(ωe)As(ωs)dσ

dΩ (18)

where Ieis the excitation light intensity, Nsuris the number of adsorbed molecules, Ae(ωexcitation) and As(ωscatter) are average electromagnetic surface intensity enhancement factors and ^dσ

dΩ is the angle of collection optics. From equation 18, SERS intensity enchanement can be done in the following ways: either by increase the number of molecules on or close to the metal surface, and/or by increasing the Raman cross-section and/or by increasing the average electromagnetic surface intensity enhancement factors.

2.3.1 Chemical enhancement mechanism

The criteria for CE is that the probe molecule should be chemically bound to the SERS substrate.

The CE can be divided into three main categories that contributes the the total CE: (i) a resonance Raman effect which occurs when the incident light is matching an electronic transition in the molecule. This leads to resonance Raman scattering. This mechanism is generally thought of as a molecular property, and it has been found that the metal surface affect the resonance. (ii) A charge-transfer effect that occurs when the incident light is in resonance with a metal-molecule transition, and (iii) a non-resonant chemical effect due to ground-state orbital overlap between the metal and the molecule [17].

(23)

2.3.2 Electromagnetic enhancement mechanism

CE requires chemical bonding between the molecule and metal particle. In contrast to CE, EM does not depend on chemical interaction and is in fact analyte independent. EM strongly depend on the size and structure of the nanoparticle. Because of the size of the particle (10-100 nm), it can interact with the incident light. Because the particle is much smaller than the wavelength of light, the particles form a light-induced electric polarization. The alternating electric field of the incident light wave causes collective oscillation of the electrons within the metal particle, called surface plasma oscillations. At a specific frequency, the light and the surface plasma oscillations are in resonance which will strongly enhance the electromagnetic field around the particle, which increases Ae(ωe) and As(ωs) from equation 18. If the molecule is bound or near to the surface of the particle, the Raman signal from the probe molecule is strongly enhanced.

The success of SERS strongly depend on the SERS substrates. Common materials used for SERS are Au, Ag and copper (Cu), whereas Au and Ag is the most common ones due to less reactive than Cu [22]. All these materials have LSPRs in the visible light region, making these materials convenient to use as SERS materials. The LSPR effect strongly depends on NP shape and size. A specific wavelength of incident light can induce collective oscillation of the surface electrons of the particle. The wavelength of the LSPR is dependent on the NP shape, size, and agglomeration state. Due to the LSPR effect, an enhancement in the electromagnetic field is generated around the NP which in turn enhance the Raman signal from a molecule on the surface of the NP. The conditions for resonance are determined by the absorption and scatter of the incident light, which in turn depends on of several factors. Shape, size and dielectric constants of both the metal and the surrounding material is affecting the resonance conditions [20].

In this project, silver nanoparticles (AgNPs) and gold nanostars (AuNSs) will be used as the SERS substrate.

2.3.3 Enhancement factor

One approach to make it easier to compare studies about SERS worldwide is to use an enhancement factor (EF), which measures the enhancement per molecule and cancel out instrumental factors.

EF is defined in equation (19)

EF = I_SERS CSERS

· C_Raman IRaman

(19)

where ISERS is the intensity of the SERS signal, CSERS is the concentration of probe molecules used in SERS, IRaman is the intensity of the Raman signal and CN_Raman is the concentration of the probe molecules used for Raman measurements.

(24)

As mentioned in the previous section, it is generally agreed that the enhancing mechanisms for SERS are chemical enhancement and electromagnetic enhancement, of which the latter is the most dominant. The chemical enhancement results from a charge transfer mechanisms, where the excitation wavelength is resonant with the metal-molecule charge transfer electronic states. This mechanisms can give an enhancement factor to 10³ [22].

The electromagnetic enhancement results from the excitation of the LSPRs as explained in the previous section. Although EM can be obtained from a single NP, it is advantageous to place the molecule in a so-called hot spot. These hot spots appears in nanometer-sized gaps between two NPs. Depending on resonance conditions, the EM of SERS is theoretically calculated to reach enhancement factor of 10⁵− 10⁶ or higher in the hot spot regions [4, 22]. Hot spots can be produced not only in gaps but also in NP junctions, crevices or flat metal surfaces supporting plasmon resonances. The EM enhancement associated with the hot spots strongly depend on the geometry and distance between the NPs.

2.4 Stability and Aggregation in Colloidal Solutions

As mentioned in the previous section, the EM contributes to the enhancement factor by several orders of magnitudes higher than to that of the CE. The EM is due to the excitation of the LSPR on the metal NP surface. The electromagnetic field around the NP is not uniformly amplified, meaning that the EM around a particle strongly depends on the geometry of it [23]. The EM also benefit from plasmon-plasmon interactions among particles and as mentioned, the EM in a hot spot between particles is several orders of magnitude higher than that of the surface of one single particle. The formation of hot spots in a colloidal solution is therefore essential for obtaining a strong Raman signal.

In this study, the NP are synthesized using a self-assembling bottom-up approach which results in a NP colloidal solution. A detailed synthesis description will be carried out in the experimental section. The equilibrium in a colloidal solution is depending on hard-core repulsions, van der Waals attractions, screened Coulomb interactions, and hydrodynamic forces. Colloid stabillity depends on numerous factors, including stabilizing agent, pH, and ionic strength [24]. The stabilizing agent used in the synthesis prevent the NPs to agglomerate by electrostatic repulsion. Depending of which stabilizing agent is used during the synthesis, different shape and sizes of the NPs are obtained.

Citrate is a common stabilizing agent for silver colloids, and will also be used in this study. The long-term stability of silver colloid solutions is usually obtained by the presence of surface charges usually negative, in this case provided by the citrate stabilizing agent. The negative charge of the surface of the NPs provides strong Coulomb repulsion and provides a large potential barrier (vM ax > 15 kBT ) preventing the colloids from agglomerate resulting in a metastable solution of single colloidal particles [23]. By the addition of an electrolyte, this Coulombic repulsion can be screened. Electrolyte concentrations above the critical coagulation concentration, the Coulombic

(25)

force no longer dominates the van der Waals attraction (vM ax < 0) and the colloids will start to agglomerate [23]. In a study by Meyer et.al. [23], the intermediate situation where 0 < vM ax< 15 kBT was investigated. In this case, a barrier still exists but not high enough for a long-lived metastable state of single colloids. In the same study, theoretical and experimental results show that this metastable state exists, where NP aggregates into small clusters and where further agglomeration is prevented by Coulomb blocking [23]. This state of small clusters of aggregated NPs in a solution has gained considerable attention in SERS due to the formation of hot spots within these clusters.

In this project, two different salts (NaCl and MgSO₄) will be used as aggregation agents.

(26)

3 Experimental

3.1 SERSTECH 100 Indicator

The handheld Raman spectrometer 100 indicator from Serstech AB was used to do the SERS measurements. The device is equipped with a 785 nm and 3 levels of laser power, max 300 mW.

For the SERS measurements, laser power of 250 mW was used. Detector type is linear CCD array and the max spectral range from 400 cm⁻¹ to 2 300 cm⁻¹.

3.2 SERS measurements

A 5 ml glass vial was used for the SERS measurements. Stock solutions of the probe molecules and water was mixed with the NP solution followed by incubation time of one hour. Right before the SERS measurements, salt was added and mixed with the solution by shaking.

3.3 Raman measurements

3.3.1 SERSTECH 100 Indicator

To measure the Raman spectra of the pure crystals using the hand-held Raman device, PDMS polymer was added to a glass substrate. The pure mpy, atrazine and bentazon crystals where applied to the polymer. The glass substrate together with the polymer and the crystals was put in a vial and then put in the hand-held Raman device.

3.3.2 Raman microscope - WITec alpha300 R

A convectional Raman device, Raman microscope - WITec alpha300 R, equiped with a 532 nm laser wavelength and 38,2 mW laser power was used to do SERS measurements and Raman measurements.

3.4 Silver nanoparticle synthesis

The AgNP where synthesized by a chemical reduction method according to the method of Lee and Meisel in 1982 [25]. Silver nitrate and trisodium citrate were used for the preparation of AgNPs.

90 mg of AgNO3was dissolved in 500 ml distilled water during heating while stirring until boiling.

100 mg of trisodium citrate was dissolved in 10 ml distilled water and then added to the AgNO3

solution and boiled and stirred for one hour. A change of color was observed from transparent to pale yellow, confirming the formation of AgNPs. Then it was removed from the heating device and cooled to room temperature.

(27)

3.5 Gold nanostar synthesis

For the growth of AuNS, sodium citrate fictionalized gold nanoparticles of 3-5 nm in diameter where used as seeds for the AuNSs to grow from. The gold seed particles were prepared by adding 0.6 mL of ice cold 0.1 M NaBH4 to a 20 mL aqueous solution containing 0.25 mM HAuCl4 and 0.25 mM sodium citrate under vigorous stirring, followed by stirring for 3 h at room temperature.

For the growth of AuNSts from the seeds, 100 µL of the seed solution was added to 10 mL of an aqueous solution of 0,25 mM chloroauric acid (HAuCl₄). 10 µL of 1 N hydrochloric acid (HCl) was added to the solution to adjust the pH. Simultaneously, 50 µL of 0.1 M ascorbic acid and 100 µL of 2 mM silver nitrate (AgNO3) were added to the solution under stirring. The nanostars were functionalized with hexadecyltrimethylammonium bromide (CTAB) by adding 100 µL of 0.1 M CTAB to the solution, followed by stirring for 5 min.

3.6 UV-Vis spectroscopy

The extinction spectra where measured of the synthesized AgNPs and the AuNSs by usig a Varian Cary 500 Scan UV-vis-nir spectrophotometer. The absorbance measurements were made over the wavelength range of 300–1100 nm. A 1 cm path length glass cuvettes where used for the extinction measurements. The AgNP solution where diluted with MilliQ water to 1:4 ratio (NP solution:water).

3.7 Scanning electron microscopy

For the characterization of the morphology of the AgNps and the AuNSs, a Zeiss Smart SEM Supra 55 VP scanning electron microscope (SEM) was used with a electron high tension voltage at 5.0 kV. Polylysine ((C6H12N2O)n) was applied to a glass substrate and then wiped off after 5 minutes allowing it to attach to the glass substrate. Then, NP solution with and without salt was then applied to the polylysine and wiped off after 2 minutes.

3.8 Data treatment

The baseline was subtracted using the software provided by Serstech. A blank spectra was measured and then saved and subtracted from the following measured spectra. The raw data from the spectrometer was files containing an intensity value per CCD pixel. Matlab R2019a was used for data treatment and analysis. The x-axis was transformed to show Raman shift (cm⁻¹) by first record the spectra of an argon lamp with distinct peaks at given wavelengths, followed by fitting a second degree function to the pixel-wavelength pairs corresponding to the peaks. Then the x-axis of the raw spectra was transformed from pixel index to wavelength using the obtained function. Finally, by converting wavelength into cm⁻¹-wavenumbers and subtract the incident laser wavenumber, the wavelength was transformed to Raman shift. The spectrometers built in

(28)

analysis ranges between 450-2200 cm⁻1. Spectra obtained outside this range may not be recorded correctly. The integration time of the measurements was calculated and normalized to five seconds.

A four parameter Lorentzian function was fitted by using Matlab’s built in lsqcurvefit function to the peaks of interest (equation 20). By that, the intensity of the peak was determined.

y = h

(x − p)²+ w+ b (20)

where w is the width parameter determined by the full width half maximum, F W HM = 2∗√ w, h is the height parameter determined by h = _w^h, p is the peak position, b is the baseline. The intensity of the peak is determined as the height of the fitted function determined by its parameters.

In Figure 5, shows an example of the Lorentzian fit to a peak.

Figure 5: Lorentzian function fitted to the 1000 cm⁻¹ peak of 4- mercaptopyridine in a solution containing AgNPs SERS substrates, 2,5 nM mpy, and 15 nM NaCl. The spectra is an average of 10 spectrum with 5 second integration time.

(29)

4 Results and Discussion

4.1 NP Characterization

The prepared NPs where characterized using SEM and UV-Vis spectroscopy. SEM were carried out to characterize the morphology of the AgNPs and AuNSs prepared according to the chemical reduction method. The prepared AgNPs consists of quasi-regular shaped particles in a wide range of shapes. Different shapes of the AgNPs is due to poor control of the morphology when only citrate is used in the synthesis. In addition, as can be seen in Figure 6a, mixtures of spherical and rod-like silver particles are present in the solution. This due to the poor balance of nucleation and growth processes during the synthesis. An estimation of the average AgNP size can be done from Figure 6a. Most of the particles have diameters in the tens of nanometer dimension. However, there is also a large variation in size of the NP resulting from this synthesis method. In Figure 6b, 15 mM NaCl was added to the AgNP solution and it shows the effect salt has on silver colloid solutions. Prior to aggregation, the nanoparticles are isolated. After the salt has been added, the nanoparticles form large clusters of agglomerates made up of hundreds of particles.

(a) AgNPs without aggregation agent (b) AgNPs with aggregation agent

Figure 6: SEM images of AgNPs with and without aggregation agent

The morphology of the AuNSs can be seen in Figure 7. An estimation of the average NS size can be done from Figure 7 to ca 80 nm.

(30)

Figure 7: SEM images of AuNSs

UV-Vis spectroscopy was done to measure the extinction spectra of the prepared AgNPs and AuNSs in solution. In Figure 8 the extinction spectra of the prepared AgNPs and AuNSs are shown in the range from 300 to 1000 nm. The absorption band in visible light region is usually between 350 to 550 nm for silver NPs and the prepared NPs have an absorbance peak around 450 nm.

The broad absorbance peak confirms the large disruption of size and shape of the particles. The AuNSs show a broad absorbance peak due to the multiple plasmon resonances of each individual nanostar. For the AuNSs, an resonance peak at 860 nm is observed.

Figure 8: Extinction spectra of AgNPs and AuNSs.

4.2 Aggregation

Two different salts were used as aggregating agents: NaCl and MgSO4. Both salts cause the silver nanoparticles to agglomerate and form clusters of aggregated NPs. MgSO4 has been reported to

(31)

be adsorbed weakly to the silver surface, which should allow for more target molecules to be adsorb onto the silver surface [26].

As mentioned, for the probe molecule to experience Raman enhancement, it needs to be adsorbed or in the near-field of the nanoparticle. Therefore, species that adsorb on the surface of the NP experience a large enhancement. The solubility of different silver salts can be used to compare the adsorption strengths of its corresponding anions to silver. Silver chloride has a low solubility in water, and the chloride ions adsorbs strongly to silver. By that, the Cl⁻-ions replace other species from the silver surface. Figure 9 shows the SERS spectra of AgNPs aggregated with 15 mM NaCl and AgNPs aggregated with 15 mM MgSO₄. The spectra with AgNPs aggregated with NaCl alters a blank spectra and does not reveal any intensity peaks. According to literature, AgCl has an intensity peak at 245 cm⁻¹[27], which is outside the spectral range of the instrument. This indicated that the citrate on surface of the NP is displaced and replaced with chloride-ions. In contrast, the spectra of AgNPs aggregated with MgSO4 reveals a spectra with multiple intensity peaks, connected to the features of citrate. For example, the peaks at 930, 944 and 1018 are related to citrate [28]. Sulfate have Raman peaks at 982 cm⁻¹ [29]. From Figure 9, no distinct peak at 982 cm⁻¹ that overcomes the noise is observed, indicating that the sulfate ions are not present at the surface of the NP in any significant amount. Also, this confirms the weak adsorption of the sulfate ions on the surface of the NP.

Figure 9: SERS measurements of AgNPs with 15 mM NaCl and 15 mM MgSO₄

The critical coagulation concentration of NaCl and MgSO₄ was investigated experimentally.

Figure 10, shows how the mean peak intensity of the peaks at 1006 and 1093 cm⁻¹ of mpy vary with NaCl concentration. The peaks are detectable from 3 mM NaCl and the mean peak intensity

(32)

increases with increasing NaCl concentration.

Figure 10: Mean peak intensity of the peaks at 1006 and 1093 cm⁻¹ of mpy with varying NaCl concentration.

Figure 19 shows how the mean peak intensity of the peaks at 1006 and 1093 cm⁻¹ of mpy vary with MgSO4 concentration. The critical coagulation concentration is lower than for NaCl, which agrees with literature. The peaks are detectable fron 0,3 mM MgSO4and the mean peak intensity increases with increasing MgSO4 concentration.

Figure 11: Mean peak intensity of the peaks at 1006 and 1093 cm⁻¹ for mpy with varying MgSO4

concentration.

(33)

To measure the critical coagulation concentrations the samples with NPs and probe molecules where incubated for one hour, followed by the adding the salt right before measuring.

4.3 4-mercaptopyridine with NaCl

The SERS spectra of mpy adsorbed on AgNPs in colloidal solution is shown in Figure 12. 10 nM mpy and NP solution was incubated for one hour. Before measuring, 10 mM NaCl was added to the NP and mpy solution as an aggregation agent. There are three possible ways for mpy to be adsorbed onto the surface of the NP, either via the the participation of the lone pair electrons of sulfur or nitrogen making a covalent bond to the NP surface, or via the aromatic π electron system.

Due to the different interactions of mpy and the NP, different orientations of mpy relative to the NP surface is obtained. The different interactions therefore results in different spectra. In Figure 12, the peak at 721 cm⁻¹ shows a downward shift compared to that of mpy crystals to about 707 cm⁻¹ due to the adsorption to the NP surface. The peak at 1090 cm⁻¹ is significantly enhanced upon the adsorption. A a change in a local environment modulates a vibration coupling and if the adsorption would occur via the S atom, the vibration coupling between the substitute and the pyridine ring is modified. This change of the coupling may cause the remarkable enhancement of the enhanced 1090 cm⁻¹ peak and the downward shift of the 707 cm⁻¹ peak. The observation of these shifts of the peaks are understandable since they both are sensitive to the structure and properties of the substrate. In a study by Hu et.al. [30], the pure mpy crystals where studied as well as when adsorbed on a silver mirror and a silver foil was observed. The same shifts of the same peaks where observed. Similar results have been observed for mpy adsorbed via the S atom on a SERS active substrate [30]. In the same study, the orientation of mpy with respect to the the Ag surface was investigated. Because of the bonding of the S atom to the silver surface, the orientation of the flat mpy molecule could be assumed to be in a standing up position from the NP surface. One way to determine the orientation of the planar mpy molecule adsorbed on the NP surface is to observe the weak or strong aromatic C-H stretching peak in the SERS spectra.

Peak intensity would reach a maximum when the C-H stretching band is normal to the surface of the NP. This because of the enhanced electromagnetic filed around the NP also is normal to the surface. If the angle between the C-H stretching mode and the NP surface approaches 0^◦, the peak intensity drops. If the mpy molecule is in a standing up position, the angle between the C-H stretching mode and the NP surface, the angle is 30^◦. Therefore, the C-H stretching mode should experience electromagnetic enhancement. Because of the covalent binding of mpy and AgNP surface, chemical enhancement should also contribute to the enhanced signal. Since silver sulfide is insoluble in water, the bonding of the thiol group to silver can be considered as irreversible. Chloride ions can not displace mpy from the surface of the NP, and the samples where NaCl is used as the aggregation agent therefore results in a spectra containing distinct mpy peaks.

(34)

Figure 12: 4-mercaptopyridine spectra of pure mpy crystals and 10 nM mpy concentration in solution containing AgNPs and 10 mM NaCl.

In Figure 13 the spectra of 1 M mpy in solution without enhancement and 5 nm with enhancement is shown. From this, the enhancement factor could be calculated using equation 19. The enhancement factor was calculated to 10⁸ for mpy with AgNPs and 15 nM NaCl, confirming the contribution of both the chemical enhancement and the electromagnetic enhancement.

Figure 13: 4-mercaptopyridine spectra of 1 M mpy in solution without enhancement and 5 nM in solution containing NP and 15 mM NaCl.

Figure 14 shows the spectra of mpy of varying concentrations with AgNPs and 15 mM NaCl. To

(35)

determine the limit of detection (LOD) of mpy in a sample, the fitting procedure should produce a Lorentzian centered at the correct position, with a height higher than the noise. Using this method, the LOD for mpy is 0.5 nM. However, some of the fits of concentrations below 0.5 nM are at the correct position, though their heights do not overcome the noise and are therefore too low to be considered reliable detection.

Figure 14: Raman spectra of 0.1, 0.25, 0.5, 0.75 nM mpy with Lorentzian fit to the peak at 1006 and 1093 cm⁻¹.

Because of the limited optimization procedures in combination with the simplicity of the system, an enhancement factor of 10⁸ and LOD to sub nanomolar concentrations reveals a promising and powerful tool for environmental analysis.

4.4 Atrazine and Bentazon

After measurements on mpy with AgNPs and NaCl, further investigation of atrazine and bentazon using the handheld device with the AgNPs and NaCl was performed.

In Figure 15 the spectra of atrazine and bentazon is shown. Both samples where incubated for one hour before measuring. Concentration of the probe molecule was 1 µM and both spectra was measured using AgNPs with 15 mM NaCl. In both spectra in Figure 15, the blank spectra without probe molecules have been subtracted in order to remove some of the noise. As can be seen from Figure 15, no intensity peaks from neither molecule are observed. What is clear is that the system used to measure mpy does not work for atrazine nor bentazon.

(36)

Figure 15: SERS measurements of atrazine and bentazon. Blank spectra subtracted.

Some of the problems that can occur in SERS are: (1) Low intrinsic Raman cross-section of the probe molecule. This would mean that the probe molecule is not Raman active and therefore not give any Raman signal. And/or (2) The laser setting in the handheld Raman device is not suitable for the detection of atrazine nor bentazon. And/or (3) Low affinity between the probe molecule and the NP surface. Meaning that there is no contribution of the chemical enhancement factor, and low or no contribution electromagnetic enhancement factor. And/or (4) No hot spot formation. This would mean that there is low contribution of the electromagnetic enhancement factor. It could also mean that the hot spots are not populated with probe molecules. To evaluate why no Raman signal from atrazine or bentazon is observed using the same system as for mpy, I tried methodically to examine and exclude each of these possible problems experimentally.

4.5 Investigation of the Raman Cross-section

4.5.1 Handheld device

To investigate the Raman cross-section of the probe molecules, the handheld Raman device was used to measure the Raman spectra of the pure crystals of the probe molecules. In Figure 16 the Raman spectra of pure mpy, atrazine and bentazon crystals are shown. Here, the atrazine and mpy peaks are shown. For atrazine the peak at 837 cm⁻¹ is connected to the C-C stretching and CH3

wagging, 921 cm⁻¹CH3twisting, 962 cm⁻¹ring breathing mode and C-C stretching, 990 cm⁻¹NH bending, and CN stretching, 1250 cm⁻¹CH2twisting, 1445 cm⁻¹CH2and CH2bending, and 1598 cm⁻¹NH bending. As expected, mpy is detectable with the handheld device. Also, the atrazine is detectable with the handheld device and has comparable Raman cross-section as mpy. However,

(37)

bentazon is not detectable with the handheld device. The spectrum of bentazon in Figure 16 is noise without any detectable peaks. Since bentazon is not detectable with the handheld device, more focus will be put on get a Raman signal from atrazine.

Figure 16: Raman spectra of mpy, atrazine and bentazon powders using the handheld Raman device.

4.5.2 Investigation of laser settings of the handheld device

To evaluate if the laser settings of the handheld device could be the issue of not getting the signal from neither atrazine and bentazon, a conventional Raman microscope was used to do Raman measurements. A 532 nm laser, 50x magnification and 38,2 mW laser power was used for the measurements seen in Figure 17 and 18. In Figure 17, the Raman spectra of atrazine powder and 100 nM atrazine dissolved in ethanol and in solution with AgNPs. Using the conventional Raman device, the peaks at 679, 961, 1061 and 1164 cm⁻¹ are detectable using the conventional Raman device. This indicates that the laser in the conventional device can detect lower concentrations than the handheld. The difference between the two devices are that the laser power of the conventional Raman device is focused to a smaller area than the laser in the handheld device. Also, the laser wavelength in the handheld device is 785 nm with less energy than in the conventional Raman device. This indicates that lasers with lower wavelengths and higher laser powers are needed to detect atrazine in aquatic solutions with lower concentrations than that of high concentrations i.e.

atrazine powder.

(38)

Figure 17: Raman spectra of atrazine powder and 100 nM atrazine dissolved in ethanol with the AgNPs using the conventional Raman device. Lorentzian function fitted to the peaks at 679, 961, 1061 and 1164 cm⁻¹.

In Figure 18 the Raman spectra of bentazon powder is shown. Peaks at 670, 859, 1039, 1243, 1311 cm⁻¹ are observed, though the signal to noise ratio is poor. Due to the limited amount of research on the topic of detecting bentazon with Raman or SERS, it is difficult to distinguish between the bentazon peaks and the noise. The peak at 1039 cm⁻¹ could be connected to the vibrational modes of the benzene ring of bentazon which have been shifted. This indicated that also bentazon is detectable with other laser settings. But since bentazon is not detectable with the handheld device, focus will be put on getting a signal from atrazine using the handheld device.

(39)

Figure 18: Raman spectra of bentazon powder using the conventional Raman device.

4.6 Investigation of the affinity to the silver surface

4.6.1 Atrazine and bentazon with MgSO4

To investigate the affinity between atrazine and the silver surface, measurements using MgSO₄ as the aggregation agent was done. The chloride ions of the NaCl make ionic bonding to the silver surface, if atrazine has weak adsorption to the silver surface, it may be displaced from the silver surface by the chloride ions when adding NaCl. The spectra obtained from the measurements on atrazine and NaCl as the aggregation agent is featureless, like the blank NaCl sample. That could be the reason why the spectra obtained from atrazine using NaCl as the aggregation does not reveal any atrazine peaks. Both NaCl and MgSO4cause the silver NPs to aggregate and form clusters of NPs. MgSO4 has been reported to adsorb weakly on the silver surface, which should allow for more probe molecules to adsorb onto areas with large field enhancements. This allows the probe molecules to adsorb onto the silver NPs, creating the much greater likelihood that when the NPs aggregate, the probe molecules will be situated in the hot spots [26].

In Figure 19, the spectra of mpy, atrazine and bentazon aggregated with MgSO₄together with a blank MgSO4 spectra are shown. For these measurements 1 µM probe molecules was used with 1 mM MgSO4 and AgNPs, the samples where incubation for one hour before measuring. From Figure 19, distinct mpy peaks are observed. The spectra of atrazine, bentazon and the blank follow the same pattern of citrate features, confirming that the citrate is still present on the surface of the NP. No peaks for atrazine nor bentazon are visible. This indicates that atrazine and bentazon is not adsorbed on the surface of the NP. For this reason, measurements of atrazine, bentazon and mpy without any salt was done.

Detection of Contaminants in Water Using Surface Enhanced Raman Spectroscopy