Charge Separation on Localized Surface Plasmon and Hot Carrier Transfer to Semiconductors

(1)

UNIVERSITATISACTA UPSALIENSIS

UPPSALA

Digital Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology 2003

Charge Separation on Localized Surface Plasmon and Hot Carrier Transfer to Semiconductors

YOCEFU HATTORI

ISSN 1651-6214 ISBN 978-91-513-1111-1

(2)

Dissertation presented at Uppsala University to be publicly examined in Polhemsalen, Ångströmlaboratoriet, Lägerhyddsvägen 1, Uppsala, Friday, 26 February 2021 at 15:00 for the degree of Doctor of Philosophy. The examination will be conducted in English. Faculty examiner: Professor Stephan Link (Rice University, Houston, Texas).

Abstract

Hattori, Y. 2021. Charge Separation on Localized Surface Plasmon and Hot Carrier Transfer to Semiconductors. Digital Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology 2003. 76 pp. Uppsala: Acta Universitatis Upsaliensis.

ISBN 978-91-513-1111-1.

The relatively recent discovery that plasmonic nanoparticles generate energetic electron-hole pairs known as hot carriers has been the source of interest from many scientific groups.

The capability to extract these short-lived hot carriers from metal nanoparticles (NPs) might potentially lead to applications in solar cells, photodetection, and photocatalysis. However, a better understanding of the hot carrier dynamics, starting from the formation process, is required.

This thesis seeks to elucidate some aspects of charge formation, extraction, and hot carriers' recombination in plasmonic composite systems.

First, two systems based on Ag and Au NPs were designed and studied to elucidate charge carriers' dynamics. The studies revealed that electrons and holes were effectively extracted and injected into suitable acceptors. Additionally, the electron injection and back transfer on TiO2

was significantly affected by the interface's status. The result motivated the following study that consisted of Au plasmonic NPs supported on different metal oxides, namely TiO2, ZnO, SnO2, and Al-ZnO (AZO). The electron dynamics on these systems were widely different. They could not be attributed solely to differences in the Schottky barrier height values, which suggested that interface status, electron bulk mobility, and oxide conduction band density of states are relevant factors to explain electron dynamics. The insertion of an insulator layer between the Au NPs and the metal oxides improved charge separation, which could be further explored to improve device efficiencies.

In situ measurements on Au NPs/TiO2 samples were performed to investigate the effect of an increase of temperature in the range expected for device applications. This increase resulted in a higher number of electrons injected, which was attributed to the enhancement of plasmon decay by phonons.

The last chapter investigates the change in the electron-phonon relaxation upon electron and hole injection, separately. Ab initio methods allowed theoretical investigation of this process and were used to predict the hole injection efficiency.

Keywords: Plasmonics, hot carrier, metal nanoparticles, semiconductors, ultrafast transient absorption spectroscopy

Yocefu Hattori, Department of Chemistry - Ångström, Physical Chemistry, Box 523, Uppsala University, SE-75120 Uppsala, Sweden.

urn:nbn:se:uu:diva-430177 (http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-430177)

(3)

To my mother, Miyako.

(4)

(5)

List of papers

This thesis is based on the following papers, which are referred to in the text by their Roman numerals.

I Light-induced Ultrafast Proton-coupled Electron Transfer Responsible for H₂Evolution on Silver Plasmonics

Yocefu Hattori, Mohamed Abdellah, Igor Rocha, Mariia V. Pavliuk, Daniel L.A. Fernandes, Jacinto Sá*

Materials Today, 2018, 21, 590-593.

II Simultaneous Hot Electron and Hole Injection upon Excitation of Gold Surface Plasmon

Yocefu Hattori, Mohamed Abdellah, Jie Meng, Kaibo Zheng, Jacinto Sá*

J. Phys. Chem. Lett., 2019, 10, 3140-3146.

III Role of the Metal Oxide Electron Acceptor on Au-plasmon Hot Carrier Dynamics and its Implication to Photocatalysis and Photovoltaics

Yocefu Hattori, Sol A. Gutierrez, Jie Meng, Kaibo Zheng, Jacinto Sá*

Manuscript submitted

IV Phonon-assisted Hot Electron Generation in Plasmonic Semiconductor Systems

Yocefu Hattori, Jie Meng, Kaibo Zheng, Ageo Meier de Andrade, Jolla Kullgren, Peter Broqvist, Peter Nordlander, Jacinto Sá*

Accepted Manuscript - Nano Lett., 2021.

V Ultrafast Hot-hole Injection Modiﬁes Hot-electron Dynamics in Au/p-GaN Heterostructures

Giulia Tagliabue, Joseph S. DuChene, Mohamed Abdellah, Adela Habib, David J. Gosztola, Yocefu Hattori, Wen-Hui Cheng, Kaibo Zheng, Sophie E. Canton, Ravishankar Sundararaman, Jacinto Sá*, and Harry A. Atwater*

Nature Materials, 2020, 19, 1312-1318.

Reprints were made with permission from the publishers.

(6)

Papers not included in this thesis:

VI Nano-hybrid Plasmonic Photocatalyst for Hydrogen Production at 20% Efﬁciency

Mariia V. Pavliuk , Arthur B. Fernandes , Mohamed Abdellah, Daniel L.

Fernandes, Caroline O. Machado, Igor Rocha, Yocefu Hattori, Cristina Paun, Erick L. Bastos, Jacinto Sá*

Scientifc Reports 2017, 7, 8670.

VII Hydrated Electron Generation by Excitation of Copper Localized Surface Plasmon Resonance

Mariia Pavliuk, Sol Gutierrez, Yocefu Hattori, Maria E. Messing, Joanna Czapla-Masztaﬁak, Jakub Szlachetko, Jose L. Silva, Carlos Moyses Araujo, Daniel L. A. Fernandes*, Li Lu, Christopher J. Kiely, Mohamed Abdel- lah*, Peter Nordlander, and Jacinto Sá*

J. Phys. Chem. Lett. 2019, 10, 8, 1743-1749.

VIII Direct Observation of a Plasmon-Induced Hot Electron Flow in a Multimetallic Nanostructure

Lars van Turnhout, Yocefu Hattori, Jie Meng, Kaibo Zheng, and Jacinto Sá*

Nano Lett. 2020, 20, 11, 8220-8228.

(7)

Contribution Report:

I - Prepared and carried out most of the characterization of the samples;

performed the transient absorption measurements along with Mohamed Abdellah; supported in the revising process.

II - Prepared and carried out most of the characterization of the samples;

performed all the transient absorption measurements; analyzed and interpreted the results; wrote the manuscript with the support from Jacinto Sa.

III - Planned the work; prepared and carried out most of the characterization of the samples; performed all the transient absorption measurements; analyzed and interpreted the results; wrote the manuscript.

IV - Planned the work along with Jacinto Sa; prepared all the samples and mount the setup for the in situ measurements; performed all the transient absorption measurements; analyzed and interpreted the experimental results; wrote the manuscript with the support from Jacinto Sa and co- authors.

V - Performed the transient absorption measurements along with Mohamed Abdellah.

(8)

(9)

1. Introduction

Figure 1.1. Lycurgus cup, fourth century CE, illuminated from inside (left) and outside (right).

Long time before scientists have started studying the optical properties of metal nanoparticles, artists were using gold and silver nanoparticles to make red-colored glasses.

The ﬁrst milestone in the history of gold ruby glass is a Roman opaque glass cup dated to the fourth century, the Lycurgus cup, which is exhib- ited at the British Museum in Lon- don. The carved decoration depicts a mythological scene that is the tri- umph of Dionysus over Lycurgus, a king of the Thracians (ca. 800 BCE).

Later studies on the Lycurgus cup re-

vealed the presence of silver-gold alloy nanoparticles of 50-100 nm in diameter, which gives the green coloration when shining light from the outside and red when illuminated from inside the cup. Despite the long history of applica- tion, although only applied for artistic purposes, the ﬁeld of plasmonics only emerged in 1990, becoming a promising domain in science and technology.

The research in plasmonics stems from exploiting the functionalities of certain metal nanostructures that can concentrate incoming light flux to volumes much smaller than the diffraction limit. This outstanding phenomenon is a consequence of partially coherent oscillations of free electrons, denominated as surface plasmons, in a metal nanoparticle driven by the external electromagnetic waves commonly referred to as localized surface plasmon resonance (LSPR) or localized surface plasmon polariton (SPP) resonance. The excitation of surface plasmons results in a strong enhancement of the electric field in the nanostructure vicinity, which also is sensitive to the structure morphology and properties of the local environment. The local strong enhancement of the electric field is one of the key points in plasmonics that led to the discovery of surface-enhanced Raman spectroscopy (SERS) technique in 1973 by Mar- tin Fleischmann [1], which allows molecules to undergo much higher scattering efficiencies when adsorbed on metal colloidal nanoparticles or rough metal surfaces. Later, similar techniques that exploit this plasmonic property were also developed, such as surface-enhanced infrared spectroscopy [2], surface-enhanced fluorescence [3] and surface-enhanced hyper Raman scattering (SEHRS) [4]. Conversely, the sensitivity of the LSPR spectral peak profile

(12)

and position with the local environment resulted in using metallic nanostructures as optical sensors, also known as plasmon-enhanced optical sensors [5].

For instance, functionalized Au nanoparticles have been used in colorimetric detection of heavy metals, biological small molecules and biomacromolecules [6–9]. Another important property is related to the process following the surface plasmon excitation in metal nanoparticles, in which collective oscillation of electrons eventually dephases, thermalize and transfer their energy to the lattice, thus generating local heat [10]. This inevitable process on plasmonics is being applied on photothermal cancer therapy which involves the intra- venous or intratumoral injection to introduce gold nanoparticles to cancerous cells and the subsequent exposure to heat-generating near-infrared light [11].

These examples already illustrate the broad range of applications provided by the surface plasmons which extends even further in non-linear optics [12], photodetection [13] and solar energy harvesting [14]. Even more exciting is the discovery of novel phenomena in quantum plasmonics [15, 16].

1.1 Plasmonic Hot Carriers

The rapid expansion of the applications provided by plasmonics eventually reached the domain of dielectric- or semiconductor-based optics and photonic technologies [17]. However, it did not take long until the expected revolution in communication components, such as plasmonic waveguides, resonators and other functional circuit elements, became dampened by the hard reality of fast decay and energy dissipation of SPP. For instance, the fast energy losses re- duce the signal propagation in plasmonic waveguides and lead to the distortion of ultrafast pulses [18]. While research aimed at suppressing loss mechanisms is still pursued [19], another research direction emerged that stem from har- nessing rather than ﬁghting material dissipative losses. In other words, losses in plasmonics also provide unique opportunities. A relevant one, which is related to the main content of this thesis, is the utilization of highly energetic charges (electrons and holes) that are generated when SPP decays.

Following light excitation, the SPP decay transferring the energy to form energetic electron-hole pairs in the femtosecond timescale known as hot carriers, which can have enough energy to be collected by semiconductors in contact or transferred to adsorbed molecules. The signiﬁcant experimental effort in plasmonic hot-driven processes and devices has been the focus of several reviews [20, 21]. Indeed, the recent discovery that metal nanoparticles can also generate hot carriers upon light excitation is seen as a breakthrough in the ﬁeld of plasmonics due to their well-known extraordinary optical properties.

Nevertheless, despite all the excitement, there are still several challenges that hamper the theoretical understanding of the microscopic mechanisms underlying the process of hot carrier generation and their utilization.

(13)

1.2 Challenges

In this section, I would like to highlight and comment my personal opinions on the main current challenges that hinder the understanding and development of plasmonic hot carrier based devices. It is important to mention that the points listed below might be incomplete and matter of debate. Nevertheless, it can hopefully shed some light on the situation of the current stage of this ﬁeld.

Time scale. SPP decay happen in few femtoseconds and hot carrier life time is commensurate with the decay event. The ultrafast nature of these events put a big obstacle for experimentalists, since typical laser pump-probe spectroscopy techniques have temporal resolution longer than 10 fs. In the future, an attosecond or single-cycle probing pulse could reveal the plasmons excitation and de-excitation process. In addition, the energy distribution of the initial hot carriers for different excitation energies might be ﬁnally quantiﬁed.

Plasmons are fundamentally quantum mechanical. The optical response of metal nanoparticles can be well described by classical electromagnetic theory. However, the dynamics triggered by light excitation of plasmons need to be treated in the quantum framework. In semiconductors, the properties can be readily predicted using ab initio methods since it only requires the calculation of the structure unit-cell under periodic boundary condition. For nanoparticles, the calculations becomes computationally very expensive. To put in perspective, the simulation of a nanoparticle with 4 nm diameter contain around 1500 atoms which would require the computation of more than 16000 electrons, which is unprecedentedly large. Nevertheless simulation of few nanometers particle is becoming feasible and can explicitly account for the effects of nanoparticle shape with speciﬁc facets and surface states on the optical response and carrier generation [22, 23].

Nanoparticle Shape. With the advances of ion-beam litography, which of- fers high resolution patterning, the fabrication of nanostructures with different shapes became possible. Moreover, there is an extensive list of bottom-up methods in the literature that takes the advantages of speciﬁc surface stabilizers to promote or suppress growth in speciﬁc crystal facets, allowing synthesis of nanoparticles with different shapes in a controlled way. The sharp edges of metal nanoparticles are favored to give rise to hot spots, which can enhance the generation of hot carriers due to increase in the Landau damping. Unfor- tunatelly, the instability of nanoparticles increases with asymmetry due to the higher surface energy and reactivity. Therefore, asymmetrical particles are al- ways prone to change their shape to quasi-spherical shape with time since it

(14)

possess the lower surface energy between all particle shapes. This process can be even further accelerated by light excitation and charge transfer process.

Beyond noble metals. Gold and silver are almost exclusively employed in hot carrier plasmonic devices due to their chemical stability and well studied properties. But due to cost, they are not considered suitable for wide applications. Nevertheless, copper and aluminum are alternative much cheaper materials with plasmonic behavior that have been pursued over the last years [24, 25]. In addition, certain nonmetallic materials, such as transition metal nitrides, transition metal carbides, and metal oxides have shown to display dielectric functions that are requisite for plasmonic behavior. Although research on nonmetallic materials for hot carrier generation is at an early stage, recent progress have shown that nonmetallic materials can be used for plasmonic photoelectric and photothermal conversions [26].

Hot carriers or just heating? In plasmon-assisted photocatalysis, it is assumed that hot carriers tunnel out of the metal into orbitals of the surrounding molecules and then catalyse the chemical reaction, where thermal effects are considered negligible. This picture has been contested by the work of Dubi et al [27] published in 2020, where is argued that what appears to be photocatalysis is much more likely thermo-catalysis. In their previous paper [28], they have developed a theory that takes into account all channels of energy ﬂow in the electronic system and revisited the main papers in the ﬁeld, showing that it can be used to explain the experimental data observed in those publications.

This debate highlights the complexity of events that are triggered by plasmon excitation and might make it prone to different mechanistical interpretations.

1.3 Aims and Scope

The ﬁeld of plasmonics is relatively young, and even more so is the recent interest in plasmonic hot carriers. As such, several open questions has yet to be clariﬁed, which are mainly related to the challenges aforementioned. The absence of a band gap restricts the lifetime of the electron-hole pair generated through plasmon decay to only about a few femtoseconds, which is at least one million times shorter than electron-hole pairs in semiconductors like silicon.

This known hard fact along with the complex and incomplete understanding of the microscopic mechanisms underlying different plasmon dephasings make the prediction of the prospects and limitations of plasmonic hot carriers devices difﬁcult.

This thesis attempts to address and elucidate the process of generation and extraction of the hot carriers. Chapter 2 introduces the classical theory to describe optical properties of localized surface plasmon along with a conceptual description of plasmon dynamics processes. Chapter 3 brieﬂy describes the

(15)

sample preparation and characterization methods used. The following chap- ters are dedicated to the results and discussions related to the papers attached to this thesis. In chapter 4, the process of both electron and hole injection from silver and gold nanoparticles is investigated using different hole accepting materials and in different physical states (liquid and solid). The next chapter explores the interface properties that dictate injection efﬁciency and electron recombination by using different metal oxides. Despite the existence of a potential barrier (Schottky barrier) between the metal and the semiconductor, the recombination process was shown to depend on other properties, of which the electron bulk mobility was suggested to also play an important role. Chapter 6 was focused on the indirect investigation of plasmon decays by enhancing one of these mechanisms by increasing the temperature. Thus, the rate of plasmon decay through electron-phonon scattering is also increased and the effect on hot electron injection was investigated. This study relevance also stems from the fact that heat generation in plasmonics is an inevitable event and might be naturally part of plasmonic device conditions. In the last chapter, the electron-phonon process that predominantly occurs following the hot electron thermalization, is brought up to discuss its dynamics change upon electron and hole injection. Moreover, this was revelead to be a potential methodology to theoretically obtain the charge injection efﬁciency values.

(16)

2. Theory

The majority of materials that possess plasmonic properties are metals and they are characterized by their quasi-free electrons, i.e., weakly interacting electrons with the nucleus that can move through the crystalline structure of the solid. These free electrons are also called electron gas and they are responsible for the main properties of metals: high conductivity and reﬂectivity. This is the opposite of insulating materials where electrons can only slightly shift from their average equilibrium position.

In 1953, Pines and Bohm [29] published a paper about their studies in- volving the collective behavior of electrons in a dense electron gas to explain the energy losses of electrons passing through metal foils. In their theoretical work, it was found that the electron gas displays both individual particle and collective aspects. The latter component includes the effect of the long-range Coulomb force, which leads to the simultaneous interaction of many particles, resulting in an organized oscillation of the system as a whole denominated the plasma oscillation. The quantization of the plasma oscillation is referred to as plasmon or bulk plasmon, in the same way phonons are described as the quantum of a collective mechanical vibration arising in a solid lattice. Rufus Ritchie [30] extended the work by Pines and Bohm to include the interaction of plasma oscillations at the surface of metals where the term surface plasmon was ﬁrst used. In other words, when a bulk metal is terminated by a surface, new plasmons arise that are strongly localized to the surface. When an electromagnetic wave travels along with a metal-dielectric interface a surface plasmon polariton (SPP) is formed, where the term polariton is used to indicate that a plasmon is coupled with the electromagnetic wave.

The main subject of this thesis involves the investigation of metal nanoparticles that can be categorized in the third subset of plasmons, known as localized surface plasmon. If a macroscopic metal particle is subject to light no unique physical phenomena occur. However, if the piece of metal is reduced to the nanoscale dimensions, the resulting metallic nanoparticles can start resonating with the electromagnetic wave becoming a powerful source of optical material in the nanoscale dimension. This striking effect gives rise to a drastic alteration in the incident radiation, increasing their optical cross-section by few orders of magnitude in respect to the nanoparticle size. The resonating property of metal nanoparticles with light is commonly referred to as localized surface plasmon resonance (LSPR) or localized SPP resonance.

The key aspect of LSPR, as the name suggests, is the resonating property.

It is widely known experimentally that gold and silver nanoparticles exhibit

(17)

this behavior in the visible light range, which are the result of two conditions being simultaneously satisﬁed:

I The permittivity (ε) of the material is negative.

II The electromagnetic wavelength (λ) is large in comparison with the nanoparticle dimensions (d), i.e.,λ d.

The permittivity is a measure of the electric polarizability of the medium.

Therefore, the higher its value, the larger will be the induced electric dipole.

Light can propagate in materials with positive permittivity, albeit the electric ﬁeld is decreased. However, macroscopic materials with negative permittivity do not allow the electromagnetic waves to travel deep from the surface and are scattered, i.e., they are absorbed and reemitted back. This is the case for silver, gold, and some other metals where the negative permittivity extends from the ultraviolet to the infrared frequency and is what gives their known reﬂective property.

The second condition of the smallness of particle dimensions compared to the incident light wavelength allows all the electrons to move with the same phase (ﬁgure 2.1). If the particle size is commensurate with the wavelength of light, some of the electrons will move in opposite directions, and the collective behavior would be lost. In addition, this condition permits the existence of an almost uniform electric ﬁeld inside the particle forλ in the visible frequency since d would be lower than the penetration depth of the incident electromag- netic wave. This is directly related to the ability of metal nanoparticles to absorb light.

Figure 2.1. Illustration of a nanoparticle interacting with an electromagnetic wave whereλ d.

Once an electromagnetic wave impinges on a particle that fulﬁlls these two requirements, the electrons will start oscillating collectively. A maximum amplitude can be achieved for a speciﬁc wavelength referred to as resonance frequency, which occurs in the range where the permittivity of certain metals has negative values.

(18)

The description mentioned above, although simplistic, grasp fundamental aspects of resonance in plasmonic materials that results in electric ﬁeld enhancement in the nanoparticle vicinity. The following sections in this chapter will be dedicated to explaining the development of the theories that allowed the mathematical description and understanding of plasmonic nanoparticles’

optical response. Section 2.5 will discuss the different mechanisms that can lead to the formation of hot carriers and the dynamics that are triggered upon light excitation. In the last one, a basic theory of metal-semiconductor interface will be introduced.

2.1 Permittivity

Dielectric constant, dielectric function, relative permittivity and permittivity are terms that are often seen when studying the optical response of materials but they can easily lead to confusions and misuses. This section has the aim to clarify these concepts since they will be used on the following ones.

It was previously stated that the permittivity is a measure of the ability of a material to be polarized by an electric ﬁeld, which is represented by the greek letterε and the unit is given by F·m⁻¹(farads per meter). Some textbooks also use the term absolute permittivity or dielectric permittivity but are often just called permittivity. Nevertheless, permittivity is not a quantity but a function that depends on the frequency. Naturally, it also depends on the region of the material, direction and intensity of the incident ﬁeld, and other parameters, but here the simple linear, homogeneous and isotropic case is assumed. Besides, the permittivity is usually represented by the relative permittivity (εr) which is the ratio between the permittivity of the material or medium (ε) and the vacuum permittivity (ε0≈ 8.85 × 10⁻¹²F·m⁻¹):

εr(ω) =ε(ω) ε0

(2.1)

Figure 2.2. Conceptual illustration of intraband and interband transition in a solid which contribute toε2.

The relative permittivity is also referred to as dielectric function, perhaps to emphasize the dependence with frequency.

Forεr(0), which is the electrostatic case, the value is denominated dielectric constant or static relative permittivity. Di- electric constant and static relative permittivity are often used terms in the study and design of capacitors since they operate in the low-frequency regime (ω → 0).

In the high frequency or optical frequency regime, the permittivity is repre-

(19)

sented by a complex function: εr(ω) = ε1(ω) + iε2(ω). The imaginary part (ε2) is related to the ability of a material to absorb electromagnetic energy.

In the case of solids, the imaginary part (ε2) is proportional to the probability that a photon can be absorbed to promote an electron to higher energy by intraband or interband transition as is illustrated in the ﬁgure 2.2. The permittivity function or dielectric function of a solid is intimately connected to the band structure and hence, is of extreme importance to describe its optical properties.

2.2 Light-matter Interaction

In the work published by James Clerk Maxwell in 1864 "A Dynamical The- ory of the Electromagnetic Field" [31], he stated the following: "The agree- ment of the results seems to show that light and magnetism are affections of the same substance, and that light is an electromagnetic disturbance propagated through the field according to electromagnetic laws". Back then, Maxwell was following the path to connect all the known electromagnetic laws in a set of twenty equations. It was Oliver Heaviside, an autodidactic engineer, mathe- matician and physicist who borrowed vector calculus notation from fluid me- chanics and condensed the twenty equations in four partial differential equations. This turning point in the classical physics that allowed to describe the interaction between electromagnetic fields and materials which is given by the following equations:

∇ ·D = ρ

∇ ·B = 0

∇ ×E = −∂B

∂t (2.2)

∇ × H = ∂D

∂t + J

The set of four equations connect the macroscopic fields, i.e., dielectric displacement D, electric field E, magnetic field H and magnetic induction B with the free charge density ρ and current density J. There are additionally two relations that describe how the electromagnetic field interact with matter, denominated as constitutive relations. For dielectric materials the expression is:

D = ε0εr(ω)E (2.3)

whereεris the relative permittivity function or dielectric function as was dis- cussed in the previous section. The applied electric ﬁeld (E) induces charge polarization (P) in the medium, which is expressed by P = ε0χEE, where χE

(20)

is the electron susceptibility and is related toεrthroughεr(ω) = 1 + χE(ω).

This allows us to rewrite equation 2.3 as:

D = ε0E +P (2.4)

One of the main breakthroughs of Maxwell was to show that electromagnetic ﬁelds could propagate as traveling waves by assembling all the four equations listed above. The incident monochromatic plane wave propagating in a solid with the speed of light (c) is:

∇²E + ω²

c² εr(ω)E = 0 (2.5)

The solution of the above equation assuming the wave propagating in the z- direction is given by:

E(z,t) = E0exp

iω

εr(ω) c z−t

(2.6) where a negative εr leads to an imaginary quantity. This is mathematically facilitated by writing ˜n(ω) = n(ω) + iκ(ω) and deﬁning the following:

εr(ω) = ˜n(ω)²→ ε1(ω) + iε2(ω) = (n(ω) + iκ(ω))² (2.7) yielding the important relations:

ε1(ω) = n(ω)²− κ(ω)²

ε2(ω) = 2n(ω)κ(ω) (2.8)

The n is the well-known refractive index andκ accounts for attenuation of the wave inside the solid. The intensity of the electric ﬁeld being proportional to |E|², an exponential decay of the intensity inside the medium is obtained I(z) = I0exp(−αz), with α = 2ωκ/c its absorption coefﬁcient. It is interesting to point out that indeed is only the imaginary part (κ) that leads to a decrease in the light intensity.

The parameters n andκ are physical observables which are carried by the electromagnetic waves. This implies that they can be obtained through re- ﬂectivity measurements, and hence be used to calculate the dielectric function through the equation 2.8. This was relevant for the development of theoretical models describing the electronic structure of a solid, since conﬁrmation through experimental results became possible.

2.3 Bulk Plasmons and the Dielectric Function of metals

The Drude model is a simple but powerful model to describe the dielectric response of a metal. In this model, the conduction electrons behave as a gas of free, non-interacting electrons, which relax through electron-electron

(21)

and electron-ion scattering being introduced in a phenomenological way as a scattering with characteristic time (τ, where typically τ ≈ 10^-14 s at room temperature), or its counterpart γ = 1/τ known as the damping factor. De- spite its simplicity, the Drude model corresponds well with the experimental results, particularly at lower frequencies. The model consist in describing the motion of electrons with an effective mass (m^∗) subject to an electric ﬁeld

E(t) = Eoexp(−iωt) and is written by using Newton’s second law:

m^∗dr²

dt² = −γm^∗dr

dt − eE(t) (2.9)

with the solution:

r(t) = e

m^∗(ω²+ iγω)E(t) (2.10) The displacement of the electrons (r(t)) induced by the electric ﬁeld E gen- erates a macroscopic polarization P= −ner for a given electron density n.

Inserting the equation 2.10 in the polarization expression (P) and substituting in the equation 2.4 one obtains:

D = ε0

1− ωp²

ω²+ iγω

E(t)

=⇒ ε^D= 1 − ω_p²

ω²+ iγω = 1 − ω²_pτ²

ω²τ²+ iωτ (2.11) withωp=

ne²/(ε0m^∗) being the bulk plasma frequency. Assuming ω γ (valid at the optical frequency range) and isolating the real and imaginary part allows to obtain the following expressions:

ε1^D(ω) = 1 − ωp²

ω²+ γ² ≈ 1 −ωp²

ω² (2.12)

ε2^D(ω) = ωp²γ

ω(ω²+ γ²) ≈ωp²

ω³γ (2.13)

The expression 2.12 and 2.13 correspond to the real and imaginary part of the dielectric function obtained from the Drude model, respectively. Note that ε₁^Dbecomes negative whenω < ωp, where the electromagnetic wave is rapidly screened and reﬂected. The wave penetration distance is described by the skin depth, which is the distance a wave travels until the initial amplitude decays by factor e⁻¹. In the case of light withλ = 500 nm impinging on a gold surface, the skin depth is approximately 20 nm. This implies that metals can be used as mirrors in the frequency range below the bulk plasma frequency (ωp) if the

(22)

Figure 2.3. a) Imaginary and b) real part of the dielectric function ﬁtted using the Drude model, where the interband transition are not accounted.

thickness is much greater than the skin depth. In contrast, forω > ωptheε₁^D is positive and the metal becomes transparent.

In most metals, the bulk plasma frequency (ωp) is in the ultraviolet regime with energies within 5-15 eV, depending on the metal band structure. The figure 2.3 shows the example for Au, where the experimental data [32] was fitted using the equations 2.12 and 2.13. From the fitting, the values for the bulk plasma frequency and the damping factor can be obtained, where ¯hωp= 8.97 eV and ¯hγ = 66 meV. Naturally, the Drude model does not account for inter- band transitions, which cause the well-known yellow color in gold materials.

2.3.1 The Damping Factor ( γ)

The photon absorption by the conduction electrons are described by the imaginary part ofε^Dthat is proportional to the damping factor (γ) as is shown in the equation 2.13. Any intraband transition of the conduction electrons require change in momentum. Since the photon momentum in the optical frequency range is negligibly small, the absorption of a photon by an electron has to be assisted by a third particle (such as an auxiliary electron, phonons, defects) in order to satisfy energy and momentum conservation. This is deter- mined by the factor γ, which can be splitted in mainly three contributions if electron-defect scattering is neglected:

γ(ω,T) = γe−ph(ω,T) + γe−e(ω,T) + γe−S(d,vF) (2.14) In the expression aboveω is the photon frequency and T is the temperature.

A detailed description and deduction of the theoretical model that resulted in the expression for the termsγe−phandγe−e are beyond the scope of this thesis. Nevertheless, it is worth to emphasize that they are expressions validated through experimental studies [33, 34]. The relevant point here is to explicitly present their equations and understand their dependence with temperature and frequency.

(23)

The expression for the temperature dependence of the collision process between electrons (γe−e) is obtained by employing Born approximation and the Thomas-Fermi screening of Coulomb interaction [35, 36]:

γe−e= πΔΓ 12hE_F

(kBT)²+

hc λ

2

(2.15) whereΓ = 0.55, Δ = 0.77 and EF = 5.5 eV are the average scattering proba- bility over the Fermi surface, the fractional Umklapp scattering and the Fermi energy of free electrons. Holstein [37, 38] derived the expression for electron- phonon (e-ph) scattering by assuming free electrons without Umklapp collisions and single Debye model phonon spectrum:

γe−ph= Γ0

2 5+4T⁵

θ_D⁵ ^θ

T

0

z⁴ e^z− 1dz

(2.16) For Au,θD= 170 K is the Debye’s temperature and Γ0= 0.07 eV, which can obtained by ﬁtting Au bulk permittivity at frequencies below the interband transition onset. The ﬁgure 2.4 depicts the γe−e andγe−phin function of temperature. It is clear that whileγe−e is frequency dependent, its change with temperature is negligible at optical region, i.e., near UV to near IR range.

The scattering rate plays a fundamental role in the direct current (DC) regime (ω = 0), but it is about three order of magnitude lower than the frequency dependent counterpart. In the other hand,γe−phcan increase the rate by three times from the room temperature to 750 K (pre-melting point of Au).

Figure 2.4. Temperature dependence of the rate for electron-electron (γe−e) and electron-phonon (γe−ph) scatterings.

(24)

The last term in the equation 2.14 arises from the conﬁnement effect that is present when particles are smaller than the electron mean free path ( below 30 nm for noble metals) and thus depend on the particle size. The phenomenological expression was initially considered by Kreibg and Vollmer [39] and for spherical nanoparticle has the following relation:

γe−S(d,vF) ∝vF

d (2.17)

where d is the diameter and vFis the Fermi velocity. Classically, this term describes the absorption of a photon by conduction electrons assisted by electron- surface collisions in order to ensure momentum conservation. In the quantum picture, this process is known as Landau damping and, naturaly, is described very differently. This phenomenum is a process of energy transfer between free electrons and electromagnetic waves occurring when their velocities are matched within a certain range. This implies that the wavevector of the electromagnetic ﬁeld has to be much larger than the free space wavevector. While propagating SPP cannot have such a large wavevector along its direction of propagation, the lateral conﬁnment allows the presence of large wavectors.

In short, Landau damping effectively represents a plasmon-electron scattering process, in which a plasmon wave loses a single quantum to generate an electron-hole pair that scales with the conﬁnement length.

The last process leading to photon absorption, that is not present in the equation 2.14, is the interband transition from low energy bands (d bands) to bands of higher energy that is accounted by adding the termεîbin the dielectric function. Rosei et al. derived a model in order to calculate the interband component of the imaginary dielectric functionε₂îb for incoming photon with energy close to the interband transition threshold for Ag [40] and Au [41]. The derived expression contain factors that represents the possible transition prob- abilities, e.g. d→ p, p→s, that are empirically obtained by simply fitting it to the experimentally obtained dielectric functions. The Ag interband transition threshold is ca. 3.95 eV and for Au is ca 2.4 eV [32].

2.4 Localized Surface Plasmon Resonance

In the beginning of this chapter, a qualitative discussion about the physical behavior of LSPR was brieﬂy introduced. Naturally, for a further understanding it is necessary to treat mathematically the interaction of metal nanoparticles with an electromagnetic wave in order to arrive at the resonance condition.

The Mie theory [42] is a theory that describes the absorption and scattering of plane electromagnetic waves by spherical particles of any size which are in a uniform and isotropic dielectric medium. This theory was named after the german physicist Gustav Mie and was developed in order to understand the

(25)

color of colloidal Au NPs in a solution. This theory expands the electromagnetic fields in multipole contributions and the expansion coefficients are found by applying the boundary conditions for the fields at the interface between the particle and the surroundings. The mathematical derivation of the fields can become extraordinarily long for big particles ( λ d ), however, for small particles (λ d ) it is sufficient to consider the first term of the multipole expansion, which is dipolar. This is known as the dipolar approximation, quasistatic or Rayleigh limit. In this regime, the electronic polarization is in phase with the excitation field, so that one can calculate the spatial field distribution by assuming the simplified problem of a particle in an electrostatic field. In other words, one needs to find the Laplace equation’s solution for the potential∇²Φ = 0, which can allow the calculation of the electric field E = −∇Φ.

The optical response of the metal nanoparticle can thus be approximated as a simple dipole in the quasistatic approximation:

p(ω) = ε0εmα(ω)E(ω) (2.18) whereεmis the static relative permittivity or dielectric constant of the medium surrounding the metal and α is the linear polarizability, which its expression for a particle with volume V_NPis given by [39]:

α = 3VNP εr(ω) − εm

εr(ω) + 2εm

(2.19) It is apparent thatα experiences a resonance enhacement when |εr(ω) + 2εm| is a minimum, which for the case of small or slowly varyingε2(valid for noble metals) around the resonance simpliﬁes to:

ε1(ω) = −2εm (2.20)

This condition is the Fröhlich condition that leads to the following expression for surface plasmon frequency (ωsp):

ωsp=

ω²_p

1+ 2εm+ ε^ib1− γ² (2.21) whereε^ib1is the real part of the contribution of interband transitions. Fromα, the scattering and extinction cross-sections (σext= σabs+ σsca) can be calcu- lated [39]:

σsca=

ω c

4εm²

6π|α(ω)|²=3ω⁴V_NP² εm²

2πc⁴

εr(ω) − εm

εr(ω) + 2εm

(2.22)

(26)

σext=ω c

√εmIm[α(ω)] = 9VNPεm³^/2ω c

ε2(ω)

[ε1(ω) + 2εm]²+ ε₂²(ω) (2.23) Two important points can be noticed from the previous equations. First, the scattering cross-section (σsca) depends quadratically with the volume while σext has a linear dependency. The ratio between both leads to:

σsca

σext ∝

d λ

3

(2.24) Therefore, as the particle size decreases considerably in respect to the incoming photon (d λ), the scattering component becomes negligible and the extinction cross-section is dominated by the absorption component (σext≈ σabs, ﬁgure 2.5a).

Second, σext and σsca depend on εm, implying that the light absorption cross-section can be enhanced when the nanoparticles are immersed in a medium with high dielectric constant or refractive index (ﬁgure 2.5b). Moreover, it also shifts the resonance frequency of the surface plasmon to lower frequency according to equation 2.21.

Figure 2.5. a)σsca /σext in function of spherical nanoparticle diameter for different incident wavelengths. b) Theoretical σext of Au nanoparticles (Au NPs) in different environments.

2.5 Hot Carrier Generation and Relaxation Dynamics

Light excitation with wavelength near the LSPR peak can result in the cre- ation of electron-hole pairs. This process starts with the plasmon excitation which decays within its lifetime τSPP≈ γ⁻¹, whereγ contains the contributions from three different processes: e-ph scattering, e-e scattering and surface collision-assisted decay (or Landau damping) as was discussed in the section 2.3.1. Naturally, interband transitions also can take place which is taken into

(27)

Figure 2.6. Mechanisms of electron-hole pair generation in metals. a) Direct vertical interband transition. b) Phonon (or impurity) assisted transition. c) e-e Umklapp scattering assisted transition. d) Landau damping or surface collision assisted transition.

account by adding theεîb term in the dielectric function. The description of these four mechanisms is illustrated in the figure 2.6. Note that the bands shown in this figure does not represent real band structures of Ag or Au, but they can be found in other publications [43, 44]. The interband transition is a vertical transition of the electron from one band to another (figure 2.6a). In the cases of Au and Ag, it is dominated by the d→ p transitions (not shown) that results in a hole with high effective mass (low mobility) and electron with low energy with respect to the Fermi level. All other mechanisms are intraband transitions, i.e., they involve absorption between two states with different wavevectors (momentum) within the same band that needs to be somehow compensated. In the mechanism illustrated in the figure 2.6b, the momentum mismatch is provided by phonons with momentumq. The third mechanism (figure 2.6c), involves electrons undergoing scattering where two electrons and two holes share the energy of the decayed SPP, i.e., E₁ + E₂ + ¯hω = E1’ + E2’. Since this is a elastic scattering the momentum of electrons is also conservedk₁+k₂=k₁’ +k₂’ + G, where G is the reciprocal lattice vector.

The last mechanism (ﬁgure 2.6d) is classically referred to surface collision- assisted decay. This happens when an electron collides with the surface and the momentum is transferred with the entire metal lattice. In the quantum

(28)

pictures this process is known as Landau damping and for particles smaller than the mean free path is the most favorable mechanism of carrier generation for their ejection from the metal [45].

Figure 2.7. Conventionally assumed picture of hot carrier generation and relaxation in metal nanoparticles. a) Plasmon is excited while the carriers are distributed according to the Fermi-Dirac statistics in equilibrium with the lattice at temperature TL0. b) Plasmon decay leads to the formation of electron-hole pairs with energies ranging from E_F - hω to EF + hω. c) The electrons thermalize after several e-e scatterings reaching once again a Fermi-Dirac distribution but with Te> TL0. d) After the electron- phonon relaxation time (τe−ph), the electron and the lattice are at equilibrium with a new lattice temperature TL1 > TL0. This temperature will eventually decrease by heating the nanoparticle environment.

The relaxation dynamics of photoexcited metals (films and nanoparticles) has been extensively studied using time-resolved spectroscopy techniques in the last decades and can be roughly represented as is in the figure 2.7. The time for plasmon decay depends on the mechanism involved, nevertheless,γ is typ- ically ≈ 10⁻¹⁴ s⁻¹ which implies that the plasmon decoherence takes place in few fs that results in the formation of hot carriers (fig.2.7b). The highly energetic electrons quickly relaxes through e-e scattering (≈ 10 fs [46]) estab- lishing a Fermi-Dirac distribution with temperature Te TL0(fig. 2.7c) within few hundred of fs. Then, electrons transfer their energy to the lattice through e-ph scattering with a characteristic timeτe−ph≈ 1 ps that is a couple of orders of magnitude longer thanτe−e. The following process involves phonon- phonon scattering within the metal until it reaches a new lattice temperature TL1in the ps to ns time scale. This temperature will eventually decrease back to the equilibrium lattice temperature T_L0as it releases thermal energy to the environment.

(29)

2.6 Schottky Barrier

The Schottky barrier refers to the potential barrier height formed when a metal and a semiconductor with different Fermi energy levels (EF) are put in contact.

Figure 2.8. Energy diagram of metal and n-type semiconductor before contact (a) and after contact for a n-type semiconductor (b) and p-type semiconductor (c).

The figure 2.8a shows the energy band diagram for the case where the metal EF is lower than the n-type semiconductor. The work function is defined as the energy difference between the vacuum level E_vacand E_F, while the electron affinity (χ) is the change in energy when moving an electron from Evacto the conduction band. When both materials with differentΦM andΦsc are put in contact, charge transfer between metal and semiconductor will occur until E_F is aligned across the interface, resulting in an electric field at the interface. For instance, electrons will flow from the semiconductor to the metal ifΦM >Φsc

(figure 2.8b) simply because available energy states for electrons in the metal are of lower energy. The electrons will move not only from the semiconductor’s surface but also from a specific depth called the depletion region. This charge separation creates an electrostatic field pointing from the semiconductor to the metal, leading to a positive potential in the semiconductor’s surface region, and thus the band will bend upwards. TheΦM of the metal remains the same due to the high free electron density that can effectively screen the electrostatic field. The same reasoning can be applied for the case when ΦM

< Φsc (ﬁgure 2.8c). The resulting surface energy barrier (ΦB) formed in the conduction band of the semiconductor is given by:

(30)

ΦB= ΦM− χ , for n-type (2.25)

ΦB=E_g

q + χ − ΦM, for p-type (2.26) Naturally, an ohmic contact is formed whenΦM < Φsc for a n-type semiconductor andΦM > Φsc for a p-type semiconductor. In the work presented in this thesis, the metals used were Ag and Au with ΦM higher than theΦsc

of the n-type semiconductors (TiO₂,SnO₂,ZnO and Al doped ZnO(AZO)) and ΦMlower than theΦscof the p-type semiconductors (GaN and PEDOT:PSS).

TheΦMof Ag is 4.26-4.74 eV [47], while for Au is 5.1 eV [48]. Nevertheless, it is important to emphasize that these are values representing ﬁlms, therefore they might be overestimated values compared to small nanoparticles. The table 2.1 below contains the Schottky barrier valuesΦBfor different metal/metal oxides junctions obtained from the literature.

Table 2.1. ΦB of different metal/semiconductor composites obtained from other works.

Ag / n-TiO2 Au / n-TiO2 Au / n-ZnO Au / n-SnO2 Au / p-GaN ΦB(eV) 1.0(1)[49] 0.9-1.2 [49–52] 0.62 [53], 0.67 [54] < 0.33 [55] 1.1 [56]

The above values were obtained from metal ﬁlms with thickness varying from 2-60 nm, except for the work done by Hwang et al. [54] where the Au was in the nanoparticle form. TheΦBobtained for Ag / TiO2is unexpectedly high when considering theΦM of Ag and χ ≈ 4 eV of TiO2. However, the above equations are only valid for an ideal metal-semiconductor contact. It is well known that Ag surface can form a oxide layer in the surface, in contrast to the more inert Au surface, that can lead to discrepancies in the expected theoretical Schottky barrier value.

(31)

3. Materials and Methods

In this chapter the sample preparation methods will be brieﬂy presented together with the main techniques used to investigate the photomechanism of plasmonic systems. Further information about the synthesis procedure can be found in the supporting information of the papers presented in this thesis.

Ultrafast transient absorption spectroscopies were the main techniques used and are of great relevance to understand plasmonics, since most of the main processes happen in the femtosecond to picosecond time scale.

3.1 Synthesis of Metal Nanoparticles

3.1.1 Bottom-up Method

The synthesis of metal NPs (Ag and Au) is a relatively simple procedure that generally are synthesized via the reduction of metal precursors in aqueous or organic media with the presence of surface stabilizers, commonly referred as capping ligands. However, if the production of NPs with narrow size distribution and speciﬁc size is desired, it is important to ﬁnd the appropri- ate synthetic procedure that is vastly provided in the literature. For example, the synthesis of sub-10 nm Au NPs is based on strong reducing agents (e.g.

NaBH4), in the presence of strong capping ligands that quench particle growth.

Ag NPs was synthesized by slightly modifying Ajitha et al. [57] proto- col where polyvinylpyrrolidone (PVP) was used as the capping ligand and NaBH4as reducing agent. The only modiﬁcation done in their procedure was using 10% ethylene glycol (EG) in water due to its stabilizing properties, thus improving in the growth conﬁnement and preventing agglomeration. The obtained Ag NPs have size distribution of d = 19±7 nm. The great advantage of using PVP as the capping ligand is that the Ag NPs are stable in different organic solvents. Furthermore, they can be precipitated by adding acetone which facilitates the removal of excess of PVP in the solution. The versatility of using PVP became relevant for the experiments carried in the Paper I since it required using the solvent isopropanol free of PVP, which can be mostly removed by adding acetone followed by centrifugation several times.

Au NPs were synthesized according to the modiﬁed Turkevich method re- ported by Piella et al. [58]. The Turkevich method [59] is based on the single- phase aqueous reduction of tetracholoauric acid (HAuCl₄) by sodium citrate at 100^oC. All other similar citrate-based methods lead to the formation of fairly

(32)

monodisperse quasi-spherical particles larger than 10 nm by varying the synthetic parameters such as pH, reducing agent and solvent. The synthesis of sub-10 nm AuNPs was achieved by Piella et al. with the addition of traces of tannic acid, which can lead to fast production of narrowly dispersed 3 - 10 nm NPs. This procedure was applied to synthesize the Au NPs in the Paper III and Paper IV leading to d = 4-5 nm and 7 nm, respectively.

3.1.2 Top-down Method

An alternative simple method for the preparation of metal NPs involves the evaporation in a vacuum chamber at ca. 10⁻³Torr of few nanometers of Ag or Au followed by annealing at high temperatures. The energy provided by the heat source allows the transition from ﬁlm to an ensemble of quasi-spherical nanoparticles since it possess the lower surface energy of all particle shapes.

This method was applied in the Paper II by depositing 2 nm of Au on top of a film of sintered TiO2and ZrO2 nanoparticles (figure 3.1), resulting in ca. 6 nm Au NPs. The size of the NPs can be controlled by varying the evaporated film thickness.

Figure 3.1. TEM images and size distribution of AuNPs attached to TiO₂(a,c) and ZrO₂(b,d). e) Illustration of the steps involved in the top-down approach to prepare metal NPs. Figure reprinted from paper II.

3.2 Semiconductors

The main semiconductor used in this thesis is the ubiquitous TiO2 in the metastable anatase form. It was only in the Paper III that a comparative investigation was performed using other metal-oxides, namely, aluminum doped zinc oxide (AZO), zinc oxide (ZnO) and stannic oxide (SnO₂). All the aforementioned metal oxides are intrinsic n-type and possess high direct band gap.

(33)

TiO₂, ZnO and AZO have similar values of ca. 3.2 eV [60, 61], while SnO₂ displays band gap higher than 3.6 eV [62]. Nevertheless, TiO2has the highest effective mass of about 5-10 m_e [63] and bulk mobility of ca. 1 cm²/(V·s) [64], while ZnO and SnO₂have much lower effective mass of 0.3 m_e [65, 66]

and thus high bulk mobility of 205 and 200 cm²/(V·s) [67, 68], respectively.

These properties are related to different density of states in the conduction band (CB) region, which is two order of magnitude higher for TiO2 than for ZnO and SnO₂ [69]. The high density of states in the CB of TiO₂ is due to the presence of 3d-orbitals, whereas the CB of other metal oxides are mainly derived from s- and sp-orbitals of metal atoms. An schematic diagram illus- trating the energy levels of the CB and valence band (VB) of different metal oxides, along with the Fermi level of Ag and Au is presented in ﬁgure 3.2 below:

Figure 3.2. Energy diagram of the CB and VB energy levels of TiO₂,ZnO, SnO2and p-GaN taken from other works [56, 60, 70–72] along with the bulk Fermi level of Ag and Au.

TiO₂samples were prepared in two different ways. The first was using the commercial anatase TiO2 purchased from Solaronix (15-20 nm NPs) which was diluted with ethanol and spin-coated in the substrate followed by annealing at 475ôC for 15 minutes. The other method was based on spray-pyrolysis technique using titanium (IV) isopropoxide and acetyl acetonate as complex agent dissolved in t-butanol. ZnO and AZO NPs solutions were purchased from Sigma-Aldrich and SnO2from Alfa Aesar. The purchased solutions were also spin-coated and annealed at 200ôC and 500ôC, respectively for 30 minutes.

(34)

3.3 Transient (NUV-NIR/mid-IR) Absorption Spectroscopy

Transient absorption spectroscopy is a powerful tool for the investigation of the dynamics of of ultrafast photophysical and photochemical phenomena in picosecond-nanosecond time range. In this technique two femtosecond pulses are incident and spatially overlapped in the sample. One consists of an intense quasi-monochromatic pulse that is used to excite (pump) the sample. For each incoming pump pulse the relative change in absorption (eq. 3.1) is recorded by monitoring the variation of the second weaker probe pulse.

ΔA(λ,t) = Ap(λ,t) − Aup(λ,t) (3.1) A_p and A_upare the absorption of the sample upon excitation (p) and without excitation (up). Aup can be measured by using a chopper with half of the repetition rate of the pulses, therefore removing every other incoming pump pulse. The time between the pulses are controlled by using a mechanical delay stage that can be varied up to few nanoseconds, thereby the temporal evolution is obtained by measuring the absorption at different delay times in respect to the ﬁrst pump pulse recorded through an spectrometer (Newport MS260i spectrograph). A schematic illustration of this technique is shown in the ﬁgure 3.3.

Figure 3.3. Illustration of the main components involved during transient absorption measurement. OPA stands for Optical Parametric Ampliﬁcation.

The probe consist of a broad in spectrum pulsed light. For the probing in the visible region, the fundamental beam (795 nm, FWHM≈ 40 fs, generated by the laser from Libra Ultrafast Ampliﬁer System designed by Coherent) passes through the delay stage and is focused on Sapphire or CaF₂crystals for white light continuum generation. In order to probe in the mid-IR region, an

Charge Separation on Localized Surface Plasmon and Hot Carrier Transfer to Semiconductors

Charge Separation on Localized Surface Plasmon and Hot Carrier Transfer to Semiconductors

List of papers

Papers not included in this thesis:

Contribution Report:

Contents

1. Introduction

1.1 Plasmonic Hot Carriers

1.2 Challenges

1.3 Aims and Scope

2. Theory

2.1 Permittivity

2.2 Light-matter Interaction

2.3 Bulk Plasmons and the Dielectric Function of metals

2.3.1 The Damping Factor ( γ)

2.4 Localized Surface Plasmon Resonance

2.5 Hot Carrier Generation and Relaxation Dynamics

2.6 Schottky Barrier

3. Materials and Methods

3.1 Synthesis of Metal Nanoparticles

3.1.1 Bottom-up Method

3.1.2 Top-down Method

3.2 Semiconductors

3.3 Transient (NUV-NIR/mid-IR) Absorption Spectroscopy