Micro- and Nano-Raman Characterization of Organic and Inorganic Materials

Micro- and Nano-Raman Characterization of Organic and Inorganic Materials von der Fakultät für Naturwissenschaften der Technischen Universität Chemnit...
Author: Jordan Thomas
2 downloads 0 Views 6MB Size
Micro- and Nano-Raman Characterization of Organic and Inorganic Materials von der Fakultät für Naturwissenschaften der Technischen Universität Chemnitz genehmigte Dissertation zur Erlangung des akademischen Grades

doctor rerum naturalium (Dr. rer. nat.)

vorgelegt von M.Eng. Evgeniya Sheremet geboren am 02. Dezember 1988 in Nowosibirsk, Russland eingereicht am 27. Februar 2015

Gutachter: Prof. Dr. Dr. h.c. Dietrich R.T. Zahn Prof. Dr. phil. II. habil. Lukas Eng

Tag der Verteidigung 07. Oktober 2015 http://nbn-resolving.de/urn:nbn:de:bsz:ch1-qucosa-188175

2

Bibliografische Beschreibung M. Eng. Evgeniya Sheremet Mikro- und Nano-Raman Charakterisierung von organischen und anorganischen Materialien Dissertation (in englischer Sprache), 2015 Betreuer: Dr. Raul D. Rodriguez und Prof. Dietrich R.T. Zahn (Technische Universität Chemnitz, Chemnitz, Deutschland), und Prof. Alexander G. Milekhin (Institut für Halbleiterphysik, Novosibirsk, Russland) 120 Seiten, 45 Abbildungen Die Raman-Spektroskopie ist eine der nützlichsten optischen Methoden zur Untersuchung der Phononen organischer und anorganischer Materialien. Mit der fortschreitenden Miniaturisierung von elektronischen Bauelementen und der damit einhergehenden Verkleinerung der Strukturen von der Mikrometer- zur Nanometerskala nehmen das Streuvolumen und somit auch das Raman-Signal drastisch ab. Daher werden neue Herangehensweisen benötigt um sie mit optischer Spektroskopie zu untersuchen. Ein häufig genutzter Ansatz um die Signalintensität zu erhöhen ist die Verwendung von Resonanz-RamanStreuung, das heißt dass die Anregungsenergie an die Energie eines optischen Überganges in der Struktur angepasst wird. In dieser Arbeit wurden InAs/Al(Ga)As-basierte Multilagen mit einer Periodizität unterhalb des Beugungslimits mittels Resonanz-Mikro-Raman-Spektroskopie und Raster-Kraft-Mikroskopie (AFM) den jeweiligen Schichten zugeordnet. Ein effizienterer Weg um die Raman-Sensitivität zu erhöhen ist die Verwendung der oberflächenverstärkten Raman-Streuung (SERS). Sie beruht hauptsächlich auf der Verstärkung der elektromagnetischen Strahlung aufgrund von lokalisierten Oberflächenplasmonenresonanzen in Metallnanostrukturen. Beide oben genannten Signalverstärkungsmethoden wurden in dieser Arbeit zur oberflächenverstärkten Resonanz-Raman-Streuung kombiniert um geringe Mengen organischer und anorganischer Materialien (ultradünne Cobalt-Phthalocyanin-Schichten (CoPc), CuS und CdSe Nanokristalle) zu untersuchen. Damit wurden in beiden Fällen Verstärkungsfaktoren in der Größenordnung 103 bis 104 erreicht, wobei bewiesen werden konnte, dass der dominante Verstärkungsmechanismus die elektromagnetische Verstärkung ist. Spitzenverstärkte Raman-Spektroskopie (TERS) ist ein Spezialfall von SERS, bei dem das Auflösungsvermögen von Licht unterschritten werden kann, was zu einer drastischen Verbesserung der lateralen Auflösung gegenüber der konventionellen Mikro-RamanSpektroskopie führt. Dies konnte mit Hilfe einer Spitze erreicht werden, die als einzelner plasmonischer Streuer wirkt. Dabei wird die Spitze in einer kontrollierten Weise gegenüber der Probe bewegt. Die Anwendung von TERS erforderte zunächst die Entwicklung und Optimierung eines AFM-basierten TERS-Aufbaus und TERS-aktiver Spitzen, welche Gegenstand dieser Arbeit war. TERS-Bilder mit Auflösungen unter 15 nm konnten auf einer Testprobe mit kohlenstoffhaltigen Verbindungen realisiert werden. Die TERS-Verstärkung und ihre Abhängigkeit vom Substratmaterial, der Substratmorphologie sowie der AFM-Betriebsart wurden anhand der CoPc-Schichten, die auf nanostrukturierten Goldsubstraten abgeschieden wurden, evaluiert. Weiterhin konnte gezeigt werden, dass die hohe örtliche Auflösung der

3

TERS-Verstärkung die selektive Detektion des Signals weniger CdSe-Nanokristalle möglich macht. Schlagwörter Raman-Spektroskopie, oberflächenverstärkte Raman-Streuung (SERS), spitzenverstärkte Raman-Spektroskopie (TERS), Raster-Kraft-Mikroskopie (AFM), lokalisierte Oberflächenplasmonenresonanz, Cobalt-Phthalocyanin, CuS Nanokristalle, CdSe Nanokristalle, InAs/AlAs Nanokristalle, AlAs/InAs Nanokristalle

4

Parts of this work are published in For a complete publication list see page 112 1.

E. Sheremet, R.D. Rodriguez, A.L. Agapov, A.P. Sokolov, M. Hietschold, D.R.T. Zahn,

Nanoscale imaging and identification of a four-component carbon sample. Carbon, 2016, 96, 588-593. 2.

V. Kolchuzhin, J. Mehner, E. Sheremet, K. Bhattacharya, R.D. Rodriguez and D.R.T.

Zahn, Understanding Tip-Enhanced Raman Spectroscopy by Multiphysics Finite Element Simulations. Proceedings of EuroSimE 2015, 2015, 1-5. DOI 10.1109/EuroSimE.2015.7103161. 3.

E. Sheremet, A. Gutakovskii, A. Milekhin, R.D. Rodriguez, D. Dentel, W. Grünewald, D.

Dmitriev, M. Hietschold, A. Toropov and D.R.T. Zahn, Raman, AFM, and TEM profiling of QD multilayer structures. Materials Research Express, 2015, 2, 035003. 4.

A.G. Milekhin, N.A. Yeryukov, L.L. Sveshnikova, T.A. Duda, E.E. Rodyakina,

V.A. Gridchin, E.S. Sheremet and D.R.T. Zahn, Combination of surface- and interferenceenhanced Raman scattering by CuS nanocrystals on nanopatterned Au structures. Beilstein Journal of Nanotechnology, 2015, 6, 749-754. 5.

E. Sheremet, A.G. Milekhin, R.D. Rodriguez, T. Weiss, M. Nesterov,

E.E. Rodyakina, O.D. Gordan, L.L. Sveshnikova, T.A. Duda, V.A. Gridchin, V.M. Dzhagan, M. Hietschold and D.R.T. Zahn, Surface- and Tip-Enhanced Resonant Raman Scattering from CdSe Nanocrystals. Physical Chemistry Chemical Physics, 2015, 17, 21198-21203. 6.

E. Sheremet, R.D. Rodriguez, D.R.T. Zahn, A.G. Milekhin, E.E. Rodyakina and A.V.

Latyshev, Surface-enhanced Raman scattering and gap-mode tip-enhanced Raman scattering investigations of phthalocyanine molecules on gold nanostructured substrates. Journal of Vacuum Science & Technology B, 2014, 32(4). 7.

A.G. Milekhin, N.A. Yeryukov, L.L. Sveshnikova, T.A. Duda, E.E. Rodyakina,

E.S. Sheremet, M. Ludemann, O.D. Gordan, A.V. Latyshev and D.R.T. Zahn, Surface enhanced Raman scattering by organic and inorganic semiconductors formed on laterally ordered arrays of Au nanoclusters. Thin Solid Films, 2013, 543, 35-40. 8.

R.D. Rodriguez, E. Sheremet, S. Mueller, O.D. Gordan, A. Villabona, S. Schulze, M.

Hietschold and D.R.T. Zahn, Compact metal probes: A solution for atomic force microscopy based tip-enhanced Raman spectroscopy. Review of Scientific Instruments, 2012, 83(12).

5

6

Table of contents Bibliografische Beschreibung ..................................................................................................... 3 Parts of this work are published in.............................................................................................. 5 Table of contents ....................................................................................................................... 7 List of abbreviations ..................................................................................................................10 Introduction ...............................................................................................................................11 Chapter 1.

Principles

of

Raman

spectroscopy,

surface-

and

tip-enhanced

Raman

spectroscopies ..........................................................................................................................15 1.1.

Raman spectroscopy: its benefits and limitations .............................................15

1.2.

Electromagnetic enhancement in SERS and TERS..........................................18

1.2.1. Light scattering by a sphere ........................................................................19 1.2.2. Image dipole effect .....................................................................................22 1.3.

Chemical enhancement ....................................................................................23

1.4.

Summary ..........................................................................................................25

Chapter 2. Raman and AFM profiling of nanocrystal multilayer structures.................................27 2.1. Materials and methods ........................................................................................27 2.1.1. Nanocrystal growth .......................................................................................27 2.1.2. Sample preparation ......................................................................................28 2.1.3. TEM, AFM and Raman measurements .........................................................29 2.2. Structure of embedded NCs ................................................................................31 2.2.1. Size and shape of embedded NCs by TEM ..................................................31 2.2.2. Phonon spectra of NCs .................................................................................32 2.3. Profiling on NC multilayers ..................................................................................34 2.3.1. AFM profiling of multilayer NC structures ......................................................34 2.3.2. Raman profiling of NC multilayers.................................................................38 2.4. Summary ............................................................................................................40 Chapter 3.

Surface-enhanced Raman spectroscopy .............................................................43 7

3.1.

Materials and methods .....................................................................................43

3.1.1. SERS substrate preparation .......................................................................43 3.1.2. Organic and inorganic materials .................................................................45 3.1.3. Micro-Raman spectroscopy measurements ................................................46 3.1.4. Micro-ellipsometry .......................................................................................46 3.1.5. Numerical simulations .................................................................................47 3.2. SERS on organic films ........................................................................................47 3.2.1. SERS enhancement of CoPc........................................................................48 3.2.2. Polarization dependence of enhancement in SERS ......................................51 3.3. SERS by nanocrytals ..........................................................................................53 3.4. Summary ............................................................................................................55 Chapter 4.

Implementation of tip-enhanced Raman spectroscopy .........................................57

4.1. TERS enhancement factor ..................................................................................58 4.2.

State of the art of optical systems for TERS .....................................................60

4.3.

Implementation of the optical system................................................................61

4.4.

TERS tips .........................................................................................................64

4.4.1. State of the art of TERS tips .......................................................................64 4.4.2. Fabrication of tips for AFM-based TERS .....................................................66 4.4.3. Mechanical properties of fully metallic TERS tips ........................................68 4.5. Summary ............................................................................................................74 Chapter 5.

Tip-enhanced Raman spectroscopy imaging .......................................................75

5.1. Materials and methods ........................................................................................75 5.1.1. Preparation of multi-component sample........................................................75 5.1.2. TERS experiments .......................................................................................76 5.1.3. Simulations of electric field enhancement .....................................................76 5.2. High resolution discrimination of carbon-containing compounds by TERS ..........78 5.3. Effect of substrate material and morphology on TERS enhancement ..................82 8

5.4. Effect of the AFM imaging mode on TERS enhancement ...................................85 5.5. TERS on free-standing colloidal CdSe NCs ........................................................90 5.6. Summary ............................................................................................................91 Conclusions ..............................................................................................................................93 References ...............................................................................................................................95 List of figures ..........................................................................................................................104 Erklärung ................................................................................................................................109 Lebenslauf ..............................................................................................................................111 Publication list .........................................................................................................................112 Acknowledgements .................................................................................................................117

9

List of abbreviations AFM

Atomic force microscopy

CNT

Carbon nanotube

CoPc

Cobalt phthalocyanine

DFT

Density functional theory

DMF

Dimethylformamide

EBL

Electron beam lithography

EF

Enhancement factor

EM

Electromagnetic

FEM

Finite element method

FWHM

Full width at half maximum

GO

Graphene oxide

IPA

Isopropyl alcohol

LB

Langmuir-Blodgett

LO

Longitudinal optical

LSP(R)

Localized surface plasmon (resonance)

LWD

Long working distance

MEK

Methyl ethyl ketone

NA

Numerical aperture

NC

Nanocrystal

OMBD

Organic molecular beam deposition

PAMAM

Polyamidoamine

PMMA

Poly(methyl methacrylate)

RT

Room temperature

SEM

Scanning electron microscopy

SE(R)RS

Surface-enhanced (resonant) Raman scattering

SFM

Shear force microscopy

SNOM

Scanning near-field optical microscopy

SO

Surface optical

STM

Scanning tunneling microscopy

TE(R)RS

Tip-enhanced (resonant) Raman scattering

TIR

Total internal reflection

TO

Transverse optical

10

Introduction “New Truths become evident when new tools become available.” Rosalyn Yalow, Nobel laureate With the development of nanotechnology, new tools are crucial for gaining novel insights into the details of nanostructures. State-of-the-art label-free nanoscale characterization techniques include electron and scanning probe microscopies. These methods provide rich information on topography, elemental composition, mechanical, electrical and magnetic properties of a sample. However, all these techniques suffer from limited chemical sensitivity that is critical for many cases in materials science, medicine, biology, nanosciences and nanotechnologies. On the other hand, Raman spectroscopy is one of the most information rich techniques. It provides fingerprint information on the sample composition and structure, but has relatively low cross-section, which limits the amount of material that can be analyzed. In addition, the resolution in Raman spectroscopy is defined by the diffraction limit of light (hundreds of nanometers, approximately half of the wavelength of incident light). Many approaches to improve the sensitivity of Raman spectroscopy were developed, and still nowadays this subject is a strong focus of research. Several methods include resonant Raman scattering [1], the use of ingenious micro-cavities or interference-enhanced Raman scattering [2-5], which provide 10 – 105 times Raman signal enhancement. The breakthrough in terms of both sensitivity and spatial resolution in Raman signal detection was brought about by plasmonics. The detection limit can be overcome by the so-called surface-enhanced Raman scattering (SERS), which enhances the Raman signal intensity as much as 105 – 108 times. Such signal amplification is mainly due to the generation of localized surface plasmons (LSPs) on plasmonically active nanoparticles. In 1974 Fleischmann et al. [6] reported unusually high Raman spectra intensity of pyridine adsorbed on an etched silver electrode. The first recognition of SERS as a valid physical phenomenon appeared only three years later by the work of Jeanmarie and Van Duyne [7]; they extended the applicability of SERS to other nitrogen heterocycles and amines. Since then the technique development allowed for the first time single-molecule detection by Raman spectroscopy [8-12]. The spatial resolution limit was for a long time tackled by using nanometer-scale apertures in scanning near-field optical microscopes (SNOM) following the idea of E.H. Synge (1928). Synge’s idea was first realized by D.W. Pohl in 1984 [13]. This approach provided 11

optical imaging with spatial resolution down to 20 – 100 nm. However, there appeared limitations in going beyond this limit since the signal throughput and maximum output power decrease with decreasing the aperture size. Thus, SNOM combined with the inherently low Raman cross-section makes Raman imaging with high spatial resolution very challenging; and the apertures used in SNOM, that consist of a sharpened tapered glass fiber with a metal coating, are not suitable for high resolution topographic imaging. The idea to use a nanometersize plasmonic nanoparticle proposed by J. Wessel in 1985 [14] was experimentally realized in tip-enhanced Raman spectroscopy (TERS), also initially referred to as apertureless SNOM, by three different groups in 2000 [15-17]. This approach is the ultimate case of SERS using a single nanoparticle. Similarly to SERS, TERS enhances weak Raman signals but also spatially confines the enhancement region to the size of the plasmonically active tip apex. In this way the spatial resolution is drastically improved below 10 nm [18, 19]. Recently Zhang et al. demonstrated sub-molecular resolution [20] beyond the boldest expectations. These techniques open perspectives for studying organic and inorganic semiconductors with improved sensitivity and resolution, which is addressed in this PhD work. In the Introduction an overview of the state of the art of Raman spectroscopy, SERS and TERS techniques is given, the theoretical background of this study will be given in Chapter 1, where the mechanisms contributing to SERS/TERS enhancement, namely the electromagnetic and chemical enhancement mechanisms, will be introduced. Chapter 2 exploits the application of micro-Raman spectroscopy and atomic force microscopy for profiling of well-defined nanocluster multilayer structures with the period below diffraction limit of light with embedded InAs/Al(Ga)As and AlAs/InAs nanocrystals (NCs). Chapter 3 is focused on SERS, from the characterization of LSP resonances (LSPRs) of SERS substrates and their comparison with numerical simulations, to the enhancement of cobalt phthalocyanine and CuS and CdSe NC layers deposited on SERS-active substrates prepared by electron beam lithography. The state of the art and the implementation of AFM based TERS are described in Chapter 4, considering such issues as optical coupling and tip-sample distance control. A large part of the chapter is devoted to the “bottleneck” of TERS, which is obtaining TERS active tips. The fully metallic probes for TERS in AFM are introduced, and their mechanical properties crucial for AFM imaging are studied with numerical simulation tools. Chapter 5 demonstrates the application of TERS for chemical identification and analysis of carbon based materials. Then the results on TERS imaging of ultra-thin CoPc films on gold films and nanoclusters are presented. The dependence of TERS enhancement on the substrate material and geometry, as well as imaging mode, is discussed, and compared with theoretical calculations. Finally, TERS is 12

applied for enhancement of the Raman signal from just a few CdSe NCs. At the end the final conclusions of this PhD work are presented.

13

14

Chapter 1.

Principles

of

Raman

spectroscopy,

surface- and tip-enhanced Raman spectroscopies In order to logically and fully describe the background of Raman spectroscopy, as well as surface- and tip-enhanced Raman spectroscopies, in the first section the Raman scattering process is introduced, and approaches for the enhancement of the Raman signal are reviewed. Then the electromagnetic (EM) and chemical enhancement mechanisms present in SERS and TERS are discussed in detail.

1.1. Raman spectroscopy: its benefits and limitations The Raman effect was first observed in 1928 and named after its discoverer C.V. Raman. When a material is illuminated with incident light of energy ℏ𝜔𝑖 , an electron can be excited to a virtual energy state with the annihilation of a photon (Figure 1.1). Then the electron relaxes by creating a photon through one of the following processes: Rayleigh scattering (Figure 1.1a), which is an elastic scattering process; or Raman scattering, which is an inelastic scattering process. Raman scattering can result either in the creation of a photon with a lower energy as compared to the incident photon energy ℏ(𝜔𝑖 − 𝜔𝑣 ) (Stokes scattering, Figure 1.1b), or with the creation of a photon with higher energy ℏ(𝜔𝑖 + 𝜔𝑣 ) (anti-Stokes scattering, Figure 1.1c). Here 𝜔𝑣 corresponds to a vibrational frequency of a molecule, or a phonon frequency of a crystal. Since the phonon is defined as a lattice vibration, it would be appropriate to use this term for crystals, while vibrational energies are reserved for media where a phonon cannot be defined, such as molecules. Providing that the incident light is a monochromatic wave, the energy difference between the incident and the scattered light can be analyzed in order to obtain the vibrational spectrum of a sample. The origin of the scattering radiation is an induced electric dipole in a material. In the classical theory of Raman scattering, it can be described as:

⃗⃗ 𝒊 ⃗ = 𝜶𝑬 𝒑

Eq. 1.1

⃗ 𝑖 – electric field where 𝑝 is an induced electric dipole vector, 𝛼 – polarizability tensor, and 𝐸

15

Figure 1.1. The energy diagram for (a) Rayleigh, (b) Stokes Raman, (c) and anti-Stokes Raman scattering processes. The energy difference ∆𝐸 between incident and scattered photons is equal, above, or below zero, respectively for each scattering process. The upward pointing arrow implies annihilation of a photon with energy ℏ𝜔𝑖 , and the downward pointing arrow implies the creation of a photon with energies indicated. The solid lines indicate real energy levels, while dashed lines indicate virtual energy levels.

vector of the incident radiation. Each component 𝛼𝜌𝜎 of the polarizability tensor can be expanded using Taylor series. By limiting the series to the first order derivate 𝛼𝑘′ , the polarizability tensor can be presented as:

𝜶 = 𝜶𝟎 + 𝜶′𝒗 𝑸𝒗

Eq. 1.2

where 𝑄𝑣 is a normal coordinate of a vibration associated with the molecular vibration frequency

𝜔𝑣 . Assuming both the vibration 𝑄𝑣 and the incident field 𝐸⃗𝑖 to be harmonic, the Eq. 1.1 can be rewritten as:

𝟏 𝟏 ⃗⃗ 𝒊𝟎 𝒄𝒐𝒔𝝎𝒊 𝒕 + 𝜶′𝒗 𝑬 ⃗⃗ 𝒊𝟎 𝑸𝒗𝟎 𝒄𝒐𝒔(𝝎𝒊 + 𝝎𝒗 )𝒕 + 𝜶′𝒗 𝑬 ⃗⃗ 𝒊𝟎 𝑸𝒗𝟎 𝒄𝒐𝒔(𝝎𝒊 − 𝝎𝒗 )𝒕 ⃗ = 𝜶𝟎 𝑬 𝒑 𝟐 𝟐

Eq. 1.3

𝐸⃗𝑖0 and 𝑄𝑣0 here are the amplitudes of the respective harmonic oscillations. In the final form, the first summand corresponds to Rayleigh scattering, the second to anti-Stokes Raman scattering, and the third one to Stokes Raman scattering. The factor

1 2

𝛼𝑣′ 𝑄𝑣0 is referred to as

the Raman tensor and denoted as 𝑅 . In order to derive the Raman tensor, symmetry properties of the vibrational wave function, and of the polarizability tensor components must be known. They depend only on the point group to which the material belongs. Raman tensors for all point 16

groups can be found in the literature [21]. It defines the conditions for the detection of the 𝜔𝑣 vibrational mode called selection rules. The intensity of the Raman mode is proportional to the

⃗ 𝑠 𝑅𝐸⃗𝑖 , where 𝐸⃗𝑠 is a vector corresponding to the polarization of detected square of the product 𝐸 scattered light. If the product has a non-zero value, the mode is allowed, and if the above mentioned product equals zero, the mode is said to be forbidden. Since the vibrational frequency is defined by the atomic masses of the atoms involved, and the interaction forces between them (ionic and covalent bonds, van der Waals forces, electrostatic interactions, etc.), this energy provides a unique fingerprint of the molecule or crystal under investigation. Thus, Raman spectroscopy allows identification of the chemical composition without a priori knowledge of the system. It is possible and routine to use Raman spectroscopy to characterize sample crystallinity, phase, stress, and doping, as well as crystallographic symmetry [22]. Additionally, Raman spectroscopy is a non-destructive technique, which often requires no specific sample preparation, and can be applied to systems in different states: solids, liquids, and gases [23]. Raman spectroscopy is also applied in challenging environments such as deep ocean waters [24] and space exploration. It proved to be useful in the studies of biological systems and medicine [25-27], forensic sciences, works of art and cultural heritage [28], food quality control [29-31], analytical chemistry [32], nanomaterials [33, 34], geological samples [35], crystals, ceramics, glasses, and others [23, 36, 37] .

Despite the power of Raman spectroscopy, it has two main challenges: 1) the Raman intensity is relatively weak (ca. 106 times weaker than the Rayleigh intensity); and 2) the achievable spatial resolution is restricted by the diffraction limit of light. The first challenge can be resolved by enhancing the Raman signal, using different approaches: 

Resonant Raman scattering, where the energy of the incident light approaches the energy of a real electronic transition in the material, thus the probability of the optical transition increases. This leads to much higher scattering cross section. This method allows increase of the signal intensity by 104 times [1].



Raman scattering using micro-cavities (up to 105 signal enhancement) [2, 3] or mirrors to confine and direct incident and scattered light.



Interference-enhanced Raman scattering [4, 5] employs the enhancement of electric field on the surface of a dielectric layer, where interference occurs, creating surface standing waves. Signal enhancement of 10 – 103 times can be

17

achieved by this method. Methods employing surface plasmon polaritons (either in total internal reflection configuration, or using a grating), also rely on the enhanced electric field on the surface [1]. The studied material is then deposited on specially engineered substrates. 

SERS and TERS based on the localized surface plasmons enhance and confine the electric field. With these two methods, the enhancement of the Raman signal can reach values above 107 times (up to 1011 locally) [38]. Chemical enhancement (see Section 1.3) can also contribute to the Raman signal enhancement.

The latter techniques deserve special attention, since they were demonstrated to allow single molecule sensitivity (both in SERS [8, 9] and in TERS [20, 39, 40]). Moreover, TERS allows overcoming the second challenge, drastically improving the spatial resolution down to few nanometers, and even sub-molecular resolution that was recently achieved [20]. These methods are the focus of this work. The theoretical aspects of their enhancement mechanisms are addressed in this chapter.

1.2. Electromagnetic enhancement in SERS and TERS If the electric field is enhanced at the position of the Raman scattering material by an enhancement factor of 𝐸𝐹 the incident radiation is locally enhanced by 𝐸𝐹 2 (𝜔𝑖 ), and the inelastically scattered radiation is enhanced by 𝐸𝐹 2 (𝜔𝑠 ). Since the Raman signal usually has a frequency close to the frequency of the incident light, so that 𝐸𝐹 (𝜔𝑖 ) ≈ 𝐸𝐹 (𝜔𝑠 ), then the full enhancement can be approximated as 𝐸𝐹 4 (𝜔𝑖 ), or simply 𝐸𝐹 4 . Note that this approximation does not take into account the change in the Raman matrix element or change in polarization of electromagnetic radiation [38]. The distribution of electric field can be obtained from the solution of Maxwell’s equations. For an isotropic, non-dispersive, non-magnetic and source-free media the Maxwell’s equations take the following form:

18

⃗𝛁 × ⃗𝑬 ⃗ =−

⃗𝛁 ∙ ⃗𝑬 ⃗ =𝟎

⃗⃗ 𝝏𝑩 𝝏𝒕 Eq. 1.4

⃗⃗ 𝟏 𝝏𝑬 ⃗𝛁 × ⃗𝑩 ⃗ =− 𝒄𝟐 𝝏𝒕

⃗𝛁 ∙ ⃗𝑩 ⃗ =𝟎

⃗ denotes intensity of electric field, 𝐵 ⃗ – flux density of magnetic field, 𝑐 – speed of light. where 𝐸 To determine the time-dependent space distribution of electric and magnetic fields for a particular geometry, we need to solve the equation system including Maxwell’s equations and boundary conditions. The boundary conditions at the interface between two media A and B are:

⃗𝑩 ⃗ 𝒏𝑩 = ⃗𝑩 ⃗ 𝒏𝑨

⃗𝑬 ⃗ 𝒕𝑩 = ⃗𝑬 ⃗ 𝒕𝑨 ⃗𝑬 ⃗ 𝒏𝑩

𝜺𝑨 ⃗𝑬 ⃗ = 𝜺𝑩 𝒏𝑨

⃗𝑩 ⃗ 𝒕𝑩 =

𝝁𝑩 ⃗𝑩 ⃗ 𝝁𝑨 𝒕𝑨

Eq. 1.5

where indices t and n denote components of fields tangential and normal to the interface. Magnetic permeability 𝜇 for non-magnetic materials relevant for this work is equal to one. The dielectric function 𝜀 is a complex number 𝜀 = 𝜀1 + 𝑗𝜀2 .

1.2.1. Light scattering by a sphere Significant electric field enhancement can be achieved in case of scattering of electromagnetic radiation by a sphere. Let us consider a sphere with dielectric function 𝜀 in a dielectric media with dielectric constant1 𝜀𝑚 . As the simplest case, let us assume that the sphere radius 𝑟 is much smaller than the wavelength of the incident light 𝜆, i.e. 2𝜋𝑟 ≪ 𝜆. In that case, the external field 𝐸𝑖 induces a dipole in the sphere, which in turn generates a scattered electric field described by the expression [41]:

𝑬𝒔𝒄𝒂𝒕 = 𝟐 (

𝜺(𝝎) − 𝜺𝒎 )𝑬 𝜺(𝝎) + 𝟐𝜺𝒎 𝒊

Eq. 1.6

To illustrate the spatial distribution of the scattered electric field, as well as its dependence on the optical properties of the sphere defined by 𝜀1 and 𝜀2 , the calculations for a 1

Here and further, dielectric constant implies that the value is considered to be independent of the wavelength.

19

sphere with a 10 nm radius in air (medium with dielectric constant equal to 1) were performed using the finite element method (FEM)1 . The sphere is illuminated by linearly polarized light with wavelength of 633 nm and electric field strength of 1 V/m. The results are presented in Figure 1.2. Generally, the field enhancement occurs for |𝜀| of the sphere higher than the dielectric constant of the surrounding medium, both for spheres with 𝜀1 > 0 and 𝜀1 < 0. Note that the enhanced field rapidly decays from the sphere surface, as visible from Figure 1.2a and b, creating a localized region near the surface with high electric field intensity, also referred to as the near-field region. The same effect can be achieved for a variety of curved surfaces, such as cones [42], tips [43], ellipsoids [44-46], etc., and it is known as field line crowding or lightning rod effect. The latter is associated with conducting materials. The lightning rod effect becomes very important in sharp enhancing structures, such as tips used in TERS, and triangles sometimes used in SERS [47]. The direct measurement of field enhancement at this scale is challenging. The enhancement of the electric field due to this effect was demonstrated to achieve a factor of 6 by electron re-scattering on tungsten and gold tips with tip with radii down to 10 nm [43]. The field concentration achieved at the tip apex due to the lightning rod effect is sufficient for surface structuring under illumination [48, 49], or apertureless SNOM using silicon [50] or tungsten tips [49]. Exceptionally high enhancement is observed for 𝜀 = −2 (Figure 1.2a and c), when the denominator of Eq. 1.6 turns to zero creating a resonance. This effect is referred to as localized surface plasmon resonance (LSPR). Localized surface plasmon (LSP) is the collective oscillation of conduction electrons perturbed by an external field in a metallic nanoparticle. Real metals have non-zero value of 𝜀2 , which leads to damping of the resonance. Thus, the materials with values approaching 𝜀 = −2, would provide maximum enhancement. In the visible spectral range gold, silver, copper, and even aluminum fulfill those conditions. In the case of spheres with radii smaller than 5 nm, which is comparable to the mean free path of conduction electrons, and the dielectric function ought to be modified by introducing additional damping due to collisions of the electrons with the sphere surface. For spheres with sizes comparable to or larger than the wavelength of the incident light, the field cannot be considered as homogeneous over the nanoparticle volume, and higher order poles are induced. A precise solution for such situations was obtained by G. Mie in 1908 [51],

1

More details on the calculations can be found in Section 5.1.3.

20

Figure 1.2. Scattered electric field enhancement around a sphere of 10 nm radius in a dielectric medium with 𝜀𝑚 =1 as a function of the dielectric constant of the sphere. Distribution of scattered electric field around a sphere with 𝜀 = −2 (a) and +2 (b). (c) The dependence of scattered electric field maximum on the dielectric constants of the sphere.

and is referred to as Mie theory. Electric and magnetic fields are expanded in a series, where each term represents a different oscillation mode (pole), which gives rise to a separate LSPR. Each pole is described by coefficients expressed in terms of Bessel and Hankel functions [52, 53]. The main consequence of the Mie theory is the existence of these multiple LSPRs. It also 21

predicts the red shift of the LSPR energy for each of the modes due to the delay in the reaction between the electrons on the two ends of the sphere, a so-called retardation effect. Many other geometries such as spheroid, shell, semi-infinite sharp cone can be solved analytically. But for many complex, more realistic geometries one has to resort to numerical methods such as boundary element method (BEM) [53], finite element method (FEM), the finite difference in the time domain (FTDT) [53], discontinuous Galerkin method [54], etc.

1.2.2. Image dipole effect Even higher electric field enhancement can be achieved for a sphere above a surface, or two spheres close to each other. Let us consider an electric dipole with a dipole moment 𝑝, such as that generated on an illuminated metallic sphere, placed above a semi-infinite plane with dielectric function 𝜀𝑠 . In that case it often convenient to employ the method of image charges rather than solving the full set of Maxwell’s equations with appropriate boundary conditions. In that case the field distribution above the surface is equivalent to the field created by two dipoles: the original dipole 𝑝 at distance 𝑎 above the surface, and its image dipole

𝑝𝑖𝑚 = ±𝐾𝑝 created below the surface (Figure 1.3), where 𝑲=

𝜺𝒔 − 𝜺𝒎 𝜺𝒔 + 𝜺𝒎

Eq. 1.7

Figure 1.3. Dipole above a semi-infinite surface, and its image dipole.

The field on the substrate surface for a sphere of radius 𝑟 can be then calculated as:

𝟐𝜶′ (𝟏 + 𝑲)𝑬 𝟎 𝑬= (𝒓 + 𝒅 )𝟑

Eq. 1.8

where 𝐸0 is the field of the initial dipole induced in the sphere (taken as unity), 𝑟 is the sphere radius, 𝑑 – distance between the tip apex and the surface.

𝜀𝑠 (𝜀𝑚 ) – the dielectric function of the surface (environment), and 𝛼 ′ is a modified polarizability: 22

𝜶′ = (𝜶−𝟏 −

−𝟏 𝒗 ) 𝟖 (𝒓 + 𝒅 )𝟑 𝑲

Eq. 1.9

In other words, the field in the gap between the sphere and the surface may significantly exceed the field generated by a single dipole. The degree of field amplification depends on the optical properties of the plane, and is especially high for a conducting surface. Such regions with extremely high electric field are often referred to as hot spots, and they provide the largest contribution to the detected Raman signal in SERS. Importantly, the presence of the image dipole would modify the resonance conditions shifting the LSPR energy. Analogous cases can be considered for a dipole placed above a hemi-sphere on the semi-infinite surface, or above another sphere. In this case, the radius of curvature has to be taken into account as well. It can produce even higher Raman signal enhancement, further improving the sensitivity of the technique. This effect is often used in SERS, where most of the signal comes from the hot spots in the gaps between enhancing nanoparticles [38], and in gapmode TERS [55-57], when an enhancing tip is placed above a gold or silver surface to improve the sensitivity.

1.3. Chemical enhancement Many effects observed in SERS systems cannot be explained solely by the EM enhancement mechanism. Particularly, there is a controversy on whether or not single-molecule detection achieved with SERS can be attributed to the EM enhancement alone [58, 59]. Secondly, single-molecule SERS is not strongly correlated with LSPR [60, 61], and the much higher enhancement observed for the first monolayer of the material as compared to that found when increasing the number of layers [62] cannot be explained by assuming only EM enhancement. The enhancement is also known to be drastically quenched when depositing only a fraction of a monolayer of oxygen on the SERS substrate [63], which cannot be explained from the EM point of view (if oxidation of the metal substrate is ruled out). In 1979 it was suggested that localized coupling between the metal substrate and the material is responsible for this additional enhancement effect [64]. The alternative enhancement mechanism was named chemical enhancement. There are evidences that the chemical interaction is a prerequisite for SERS enhancement, although it is not sufficient by itself [65]. It makes the separation of the two enhancement mechanisms extremely difficult in real surfaces, and there is still no definite agreement in the literature [59]. 23

Although, the definition from E. Le Ru and Etchegoin [38] gives a very good summary of the subject: chemical enhancement results from the modification of the Raman polarizability tensor upon adsorption of the molecule onto the metal surface. All the mechanisms can be summarized in three cases: •

The metal disturbs the material electronic structure, changing the Raman polarizability of the mode. This mechanism requires only physisorption of the material on the metal surface.



The molecule is covalently bound to the surface, which modifies the existing energy states, or creates new interface states analogous to the interface states in semiconductor physics. Enhancement occurs if the new energy states are in resonance with the incident laser light.



Photo-driven charge transfer that occurs when the metal states provide intermediate energy states for excitation of the material to higher energy levels (Figure 1.4). In this case also resonance enhancement occurs when the energy of the incident laser light is such that ℏ𝜔𝑖 ≈ 𝐸𝐹 − 𝐸𝑀 , where 𝐸𝐹 is the Fermi level of the metal, and 𝐸𝑀 is the energy level of the material, which participate in the transition. Both of the later mechanisms require covalent bonding between the metal and the

material. Generally, such modifications of the Raman polarizability tensor can lead both to enhancement or damping of the signal. They are also likely to lead to the changes in the Raman spectra (change in relative Raman bands intensities and positions) due to perturbations of the

Figure 1.4. Illustration of charge transfer chemical enhancement mechanism.

24

material structure. The charge transfer mechanism is the most studied, since the resonance conditions can be experimentally tuned by creating a potential step on the surface of the metal and adjusting it by applying an external voltage. It is easily achieved in an electrochemical cell. Such experiments showed the dependence of the Raman signal intensity on the electrode voltage and incident wavelength [66-69], which allow analyzing the energy levels of the molecule and the metal. It was also suggested that the atomic roughness is necessary for the observation of chemical enhancement [65], and it needs to be stabilized by halide ions in order to achieve meaningful enhancement factors [58].

1.4. Summary In this chapter Raman spectroscopy and its advantages and limitations were discussed. The Raman spectrum provides rich information on the composition and the structure of a material under study. However, new and emerging technologies rely on small amounts of organic and/or inorganic materials, which are hard to detect by conventional Raman spectroscopy due to low scattering volume. Moreover, the characteristic size of the objects studied is often in the nanometer region, which is below the spatial resolution limit of conventional Raman spectroscopy defined by the diffraction limit of light (hundreds of nanometers). Thus the methods for improving both the sensitivity and the spatial resolution of Raman spectroscopy were here discussed. SERS and TERS are extremely powerful methods for amplifying the Raman signal, since they provide extraordinary sensitivity down to single molecule level. Compared to SERS, TERS provides the additional benefit of spatial resolution below 10 nm for the spectroscopic analysis of nanomaterials. SERS and TERS are based on the electromagnetic and the chemical enhancement mechanisms that were introduced in this chapter. The electromagnetic enhancement mechanism often dominates in SERS and TERS experiments, and is therefore discussed in more detail.

25

26

Chapter 2. Raman and AFM profiling of nanocrystal multilayer structures One of the fascinating systems for Raman investigations are nanocrystals (NCs). Since their optoelectronic properties are strongly size- and shape-dependent, for device fabrication and optimization the key factor is extracting information about the structural and optical properties of the NCs. Through the analysis of phonon response measured by Raman spectroscopy, it is possible to determine not only structural parameters such as superlattice periodicity and layer composition, but also to obtain information on electronic, optical, and phonon properties of the structures [70]. The Raman spectra of NCs give insights into the electronic, optical and phonon properties of NC structures [71, 72], in particular atomic intermixing [73, 74], NC size [75], shape [75], and strain relaxation [76]. The majority of devices are formed of multilayers, namely periodic NC and matrix layers with a period below the diffraction limit of light, and a protective capping layer. However, due to diffraction limited spatial resolution and low scattering volume of single NCs, Raman spectra typically contain the average response of multiple NCs with a distribution of sizes and shapes. In order to localize single NC layers, approaches for atomic force microscopy (AFM), Raman spectroscopy profiling InAs/Al(Ga)As1,2 and AlAs/InAs NC systems were developed. The Raman spectra of InAs/Al(Ga)As and AlAs/InAs NCs are analyzed in detail.

2.1. Materials and methods 2.1.1. Nanocrystal growth The samples under investigation were grown on epitaxy-ready n+ Si-doped (3×1018 cm-3) or semi-insulating (001)-oriented GaAs substrates using a solid-source molecular beam epitaxy system Riber 32P3. The substrates were first degassed for 40 min at 623 K and then loaded directly into the growth chamber. After the native oxide layer was desorbed by annealing the substrates at 883 K for 2 min, a Si-doped GaAs buffer layer of 200-1000 nm was grown at a rate of 1.0 monolayer per second (ML/s) at 853 K under a constant As beam equivalent pressure of 10-5 Torr. Then multilayers of AlAs/Al(Ga)As and InAs/AlAs NCs were grown. The growth of the 1

Al(Ga)As stands for AlAs or AlxGa1-xAs. This notation will be used further. InAs/GaAs stands for InAs NCs embedded in GaAs matrix. This notation will be used further. 3 Sample preparation by Dr. A. Toropov and Dr. D. Dmitriev, Institute of Semiconductor Physics, Novosibirsk, Russia. 2

27

Figure 2.1. Structure of the NC multilayer samples: InAs/AlAs with 50 nm (sample A) and 100 nm (sample B) periods, InAs/Al(Ga)As with 100 nm period (sample C), and AlAs/InAs with 50 nm period (sample D).

NCs was monitored by reflection high energy electron diffraction (RHEED). According to the RHEED data the transition from a two-dimensional to a three-dimensional growth mode (beginning of island formation) for all samples occurs after the deposition of 1.8 MLs of the island material. The total amount of both InAs and AlAs was 2.2 to 2.4 MLs for the NC growth. The substrate temperature was 733 K (773 K) during the growth of AlAs (InAs) NCs in the InAs (AlAs or AlGaAs) matrix at an arsenic BEP of 8×10-6 Torr. The growth interruption after NCs formation ranges from 12 s for the AlAs NCs to 50 s and 150 s for the InAs ones. The first 5 nm of AlAs or AlGaAs spacers were grown at the same temperature as the NCs (773 K); then the temperature was raised to 883 K and the rest of the AlGaAs spacer was deposited. All InAs/Al(Ga)As structures were covered by a 40 nm GaAs capping layer. The exact structure of the samples profiled by TEM and AFM is defined in Figure 2.1.

2.1.2. Sample preparation Sample preparation for AFM measurements included either cleaving (Figure 2.2a) or ion beam slope cutting (Figure 2.2b). Cleaving was performed for InAs/AlAs NC multilayers by pressing the edge of the sample with a diamond cutter from the back side. This process resulted in atomically flat surfaces along the direction.

28

Ion beam slope cutting (ion milling) was implemented for the AlAs/InAs NC system since cleaving was not possible due to defects in the matrix material caused by the large lattice mismatch between the InAs matrix material and the GaAs substrate1. Before ion cutting, the samples were cleaned with ethanol in an ultrasonic bath and then mounted on a sample holder with CrystalbondTM glue by heating up to about 393 K. During the milling process the sample surface is partly protected by a sharp edged mask. This protective mask (thin cover slip) was glued to the sample by heating up to 393 K for about 1 hour and the sample was cut with a diamond wire saw. Finally, the sample was cleaned in ultrasonic bath with acetone to remove any remaining CrystalbondTM glue. The unprotected part of the sample was milled by a highenergy Ar+ ion beam. Ion beam slope cutting was performed with a Leica EM TIC 3X milling system. Instead of one ion beam and sample oscillation the system uses three ion beams hitting the sample from 3 different directions. The result is a flat and polished surface as it is confirmed by AFM. Milling was performed at an ion energy of 6 keV at room temperature as well as with nitrogen cooling at 193 and 143 K. Beveled samples were prepared by mechanical polishing the samples at an angle of 5 to 7° with respect to the plane of the layers2. The sample was first cleaned in acetone and glued on glass with cyanoacrylate adhesive Sicomet8300TM. The glass was displaced from the edge of interest. Then the sample was glued on a sample holder and heated at 353 K for 30 minutes. Subsequent polishing in water was performed with diamond foils (ultraPrepTM Diamond Lapping Films, Buehler) sequentially decreasing the grain size (3, 1, 0.5, and 0.1 µm). The quality of the beveling was controlled with an optical microscope. Finally the glue was dissolved in acetone. The tilt angle was measured with an optical microscope from the width and depth of the beveled structure. For the Raman experiment the sample was glued on a sample holder prepared at a predefined angle (Figure 2.2c). The schematics of AFM and Raman experiments with samples prepared by the methods mentioned above are shown in Figure 2.2.

2.1.3. TEM, AFM and Raman measurements The high resolution TEM (HRTEM) investigation of (110) cross-section samples were made by means of JEOL-4000EX and FEI TITAN 80-300 microscopes operating at 400 and

1

Ion milling by Dr. W. Grünewald, Leica, Vienna, Austria, and T. Jagemann, Solid Surfaces Analysis Group, TU Chemnitz, Germany. 2 Beveling by D. Dentel, Solid Surfaces Analysis Group, TU Chemnitz, Germany.

29

Figure 2.2. Schematics of the AFM (a, b) and Raman (c) experiments. Dotted lines indicate NC layers embedded in the matrix.

300 kV, respectively1. Conventional two-beam {002}-dark-field diffraction contrast was applied to characterize the spatial distribution of NCs and structural defects. The morphology and atomic structure of NCs were investigated in the multi-beam HRTEM mode. AFM measurements were performed in intermittent contact mode with the Agilent 5420 system using NSC14 general purpose Si tips from Mikro-Mash with nominal tip radius and force constant of 8 nm and 5 N/m, respectively. Micro-Raman spectra were acquired with the confocal Horiba JY HR800 spectrometer using 514.5 nm (Ar+ laser) and 514.7 nm (solid state Cobolt laser) excitation wavelengths at a power of 0.1 – 0.5 mW and a spectral resolution of 0.5 cm-1. The laser was focused on the sample via an Olympus BX41 microscope using a 100x objective to the spot of approximately 1 µm2. Micro-Raman spectra from cleaved and ion beam cut samples were recorded from the {110} plane with unpolarized light (in 𝑦′(−, −)𝑦 ̅′ configuration in Porto notations2). Line scans on beveled samples were performed in 𝑧(𝑥, 𝑦)𝑧̅ configuration with a Märzhäuser microscope scanning stage and a step size of 100 nm and spectral resolution of 2.5 cm-1 (Figure 2.2c). Here and further 𝑥, 𝑦, 𝑧, 𝑥 ′ , 𝑦′ refer to directions parallel to the [100], [010], [001], [1-10], [110], respectively. The sign “–“ means that the unpolarized scattered light was analyzed.

1

TEM by Dr. A. Gutakovskii, Institute of Semiconductor Physics, Novosibirsk, Russia. Porto notations describe polarization and direction of the incident and scattered light in Raman ⃗⃗⃗⃗𝑠 , where ⃗⃗⃗ experiment. They are written in the form ⃗⃗⃗ 𝑘𝑖 (𝑒⃗⃗𝑖 , ⃗⃗⃗ 𝑒𝑠 )𝑘 𝑘𝑖 (⃗⃗⃗⃗ 𝑘𝑠 ) is the direction of propagation of incident (scattered) light, ⃗𝑒⃗𝑖 (⃗⃗⃗ 𝑒𝑠 ) is the polarization of the incident (scattered) light. 2

30

2.2. Structure of embedded NCs InAs NCs are currently used in optoelectronic devices such as photodetectors, high temperature lasers, and single photon sources [77, 78] and considered also as a promising system to improve the efficiency of III-V solar cells [79]. Most of these applications use InAs NCs in a GaAs matrix as an active material, which yields interband energy in the range of 1.2 – 1.4 eV (850 – 1000 nm) [80], i.e. in the infrared region. However, using InAs embedded in an AlAs matrix allows the InAs NC interband energy to be tuned in the range of 1.8 – 2 eV (620 – 690 nm) [81], i.e. in the visible range, which is preferable for optoelectronic applications. This shift results from the higher confinement potential of AlAs (band gap of 2.16 eV as compared to 1.42 eV for GaAs) and the smaller size of InAs NCs formed in an AlAs matrix [82, 83]. For InAs/Al(Ga)As based systems, a capping layer is required in order to protect the matrix from oxidation [84, 85]. It was shown, however, that covering the QDs with a capping or matrix layer can modify their morphology and, therefore, optical properties [86-88]. Despite its technological and fundamental potential, the InAs/AlAs system is poorly studied in comparison to the InAs/GaAs one, and its optical absorption in the visible range offers advantage for the resonant Raman investigations. In order to localize single NC layers, cross-sectional TEM and AFM micrographs are acquired, which are further complemented by Raman profiles obtained on beveled samples.

2.2.1. Size and shape of embedded NCs by TEM TEM provides excellent resolution and allows investigation of both planar and crosssectional views of the NC structures. It makes TEM one of the most powerful and well established techniques for investigation of NC size and morphology. The cross-sectional dark field TEM images in Figure 2.3 show a general view of the whole layer sequence of sample C, while HRTEM images demonstrate individual NCs in samples B, C, and D. The layer thicknesses of the periodical structures determined from TEM measurements correspond to the nominal values reported in Figure 2.1. InAs/Al(Ga)As NCs reveal a characteristic lens-like shape and have relatively large base size of about 50 (25-30) nm, and height of 15 (8) nm for InAs NCs in AlAs (Al0.75Ga0.25As) matrix and the presence of a wetting layer is observed in all samples. The inverted structure shows the formation of much smaller NCs of 3 – 5 nm size.

31

Figure 2.3. (a) Dark field TEM image of NC multilayer structures (sample C). HRTEM images of (b) InAs NC in AlGaAs matrix (sample C), (c) InAs NC in AlAs matrix (sample B), and (d) AlAs NC in InAs matrix (sample D).

2.2.2. Phonon spectra of NCs InAs/AlAs (AlAs/InAs) NC superlattices were studied by means of micro-Raman spectroscopy in backscattering geometry with 514.5 nm wavelength in different configurations. As it was mentioned in Section 1.1, the intensity of the Raman mode is proportional to 2

⃗⃗ ) . According to Raman selection rules for zinc-blende crystal structure, the longitudinal (⃗⃗𝐸𝑠 𝑅𝐸 𝑖 optical (LO) phonon is allowed in backscattering geometry from the {100} crystal plane in crossed polarization configuration [89] (for the polarization of the incident light along 𝑥 and polarization of scattered light along 𝑦 as introduced in Section 2.1.3, or 𝑧(𝑥, 𝑦)𝑧̅ in Porto notations). The transverse optical (TO) phonons can be detected from {110} plane in either ̅ or parallel 𝑦′(𝑥′, 𝑥′)𝑦′ ̅ configurations. crossed 𝑦′(𝑥′, 𝑧)𝑦′ 32

The use of different configurations allowed to study both LO and TO phonons (Figure 2.4) in NCs. Additionally, surface optical (SO) phonons with the frequencies between TO and LO frequencies could be detected. For comparison, the Raman spectra of the matrices are overlaid with the NC spectra. The matrix spectra show high intensity of the allowed phonons, and very weak signal from forbidden modes, which confirms that translational symmetry in the matrix layers is largely preserved. On the contrary, weakening of Raman selection rules for InAs NCs, and lifting of the selection rules for AlAs NCs were observed. This result can be a consequence of the breaking of translation symmetry in structures with AlAs nanocrystals (NCs). The NC Raman spectra were fitted with Lorenz functions, and the resulting LO and TO phonon frequencies are indicated in Figure 2.4 with green arrows. Optical phonon frequencies of InAs (AlAs) NCs were shown to be upshifted (downshifted) compared to the ones of the corresponding bulk materials (Figure 2.4). These shifts result from compressive (tensile) built-in strain in the NCs. From the lattice mismatch between the matrix and the NC materials, the expected value of the built-in strain can be estimated [74], and is marked in the Figure 2.4 with

Figure 2.4. Raman spectra of InAs/AlAs and AlAs/InAs NC multilayers. (a) The plot is divided in two spectral regions: of InAs phonons (a), and of AlAs phonons (b). The upper spectra (black) were acquired ̅ geometry (TO in 𝑧(𝑥, 𝑦)𝑧̅ geometry (LO phonon allowed), and the lower spectra (blue) in 𝑦′(𝑥′, 𝑧)𝑦′ phonon allowed). For comparison, the spectra of the matrix (dashed lines) and of the NCs (solid lines) are overlaid. The shift of phonon frequency in NCs is indicated with horizontal black and blue arrows and is induced by built-in strain. Green arrows indicate fitted LO and TO phonon frequencies. The frequencies indicated with italic fonts show calculated phonon frequencies in the absence of intermixing and confinement effect. Red arrow indicates LO phonon frequency measured with 632.8 nm laser.

33

vertical black arrows. The mismatch between the expected and measured frequencies is attributed to material intermixing and phonon confinement effects, both of which would downshift the phonon frequency. Phonon confinement effect for InAs/AlAs NCs could be estimated by performing measurements with a lower energy laser (632.8 nm). Due to the increase of the interband transition energy in InAs NCs with decreasing the NC size, larger NCs will experience resonant Raman scattering when excited with lower energy laser. It allows for size-selective resonant Raman scattering [90, 91]. The LO phonon frequency shifts by 4.7 cm-1 when measured with 632.8 nm laser line, when the incident light is in resonance with larger NCs for which the confinement effect is negligible.

2.3. Profiling on NC multilayers Raman spectroscopy has been extensively applied to the study of multilayer In(Ga)As/Al(Ga)As NC structures, atomic intermixing [74, 92], the QD size [75], shape [75], and strain relaxation [76]. Here Raman spectra of both InAs/AlAs and AlAs/InAs structures are analyzed, and Raman profiles on beveled InAs/Al(Ga)As NC multilayers with 100 nm period are acquired in order to locate single NC layers and to derive its optical and vibrational properties.

2.3.1. AFM profiling of multilayer NC structures Atomic force microscopy (AFM), as compared to TEM, operates at ambient conditions. AFM allows to perform size and morphology analysis of NCs from planar surfaces on freshly grown samples without capping layers [82, 93]. The structural parameters InAs NCs embedded in Al(Ga)As, which are different from those of open NCs [86-88], can be also studied by AFM from a cleaved {110} plane. Cleavage is a simple and well-established method for sample preparation which can provide atomically flat surfaces and precludes unintended diffusion [94]. AFM was already used for the visualization of layers in GaAs/Al(Ga)As multi-quantum well structures using selective etching [95]. It was also established that some III-V materials readily oxidize forming topography pattern on the surface of multilayer structures. For GaAs/A1xGa1-xAs and GaAs/GaInP heterostructures the oxidation rate depends on the layer composition [96, 97]. The drastically different oxidation rates of GaAs and AlAs were applied for a very precise determination of the layer thicknesses in GaAs/AlAs multilayers [98]. The surface profile of unoxidized heterostructures can be probed by AFM since it is determined by the strain relaxation of QDs on a cleaved surface as it was shown for SiGe QDs [99].

34

An alternative way of sample surface preparation for AFM characterization where cleaving is not possible is ion milling. This method has been used successfully for the preparation of SEM [100] and TEM [101] samples. Even brittle materials or very different material combinations can be easily prepared by this method. Ar+ ion milling can achieve sample surfaces of very high quality [102]. Ion beam cutting at an angle in combination with chemical etching was used for observation of AlxGa1-xAs/GaAs layers as thin as 1 nm with AFM [103]. AFM images of cleaved InAs/Al(Ga)As samples presented in Figure 2.5 allow the periodical structure of the samples to be visualized. The dark flat area corresponds to the substrate, which demonstrates very good surface quality obtained by the cleavage technique. The periodic structure seen in the AFM images and, more clearly, in the amplitude images

Figure 2.5. AFM images and corresponding topography profiles of cleaved InAs/Al(Ga)As samples. Each column represents topography (top), amplitude (middle), and topography profile (bottom) along the blue arrow in the topography image. Being sensitive to the topography gradient, the amplitude image highlights surface features. (a) sample A (InAs/AlAs, 50 nm period); (b) B (InAs/AlAs, 100 nm period); (c) C (InAs/AlGaAs, 100 nm period).

35

represents the NC multilayers, with Al(Ga)As matrix rising above the substrate level, and grooves indicating the positions of NC layers (marked by dashed lines) and of a GaAs interlayer (marked by a solid line). The interlayer distances obtained from the AFM images correspond to the nominal thicknesses of the layers given in Figure 2.1 and confirmed by TEM experiments (Figure 2.3). As can be seen in the AFM height profiles shown in Figure 2.5, AlAs rises as much as 15 – 30 nm above the substrate level. Even though AlAs has a smaller lattice constant than InAs, the freshly cleaved AlAs immediately oxidizes in air [97, 104], and the thickness of the aluminum oxide will increase 1.8 times with respect to the initial thickness of AlAs [104]. This leads to the observed rise above the substrate level. The Al0.75Ga0.25As matrix, due to the presence of Ga atoms, is much less prone to oxidation than pure AlAs. This leads to the formation of a 1 nm step at the edge between the substrate and the NC multilayers, which is in excellent agreement with AFM observations reported elsewhere [97]. AFM images of the ion milled AlAs/InAs NC structures prepared at different temperatures are represented in Figure 2.6. As can be seen, after ion milling the sample surface exhibits periodic structures emerging above the matrix, which can be interpreted as oxidized AlAs NC layers since their period is equal to the nominal one (Figure 2.1). Cooling was employed in order to minimize expected sample damage by Ar+ ions. The preparation at 193 K

Figure 2.6. AFM images and topography profiles of ion milled AlAs/InAs samples prepared at 6 keV ion energy for different preparation temperatures: (a) room temperature (RT); (b) 193 K; (c) 143 K. The amplitude or phase image in the middle row is given for a more clear view of the periodic structure of the samples. The bottom row presents the topography profiles along the blue arrows (phase profile for c).

36

gives results similar to those obtained at room temperature, while for the preparation at 143 K the height of AlAs NC layers is not sufficient to be clearly detected over the matrix (surface roughness is 0.1 nm, which is close to the noise level of the AFM). Therefore, it is very likely that ion milling at higher temperatures leads to amorphisation of the sample surface, resulting in intensive subsequent oxidation under ambient conditions. This statement can be verified by the Raman spectroscopy analysis, which is very sensitive to the quality of the crystalline structure and to the presence and amount of defects. Allowed TO phonon mode of the InAs matrix the observed at about 215 cm-1 dominates the Raman spectra recorded from the {110} plane (Figure 2.7), while for the AlAs NCs the intensity of the TO phonon peak observed at 339 cm-1 is weaker compared to the LO phonon mode at 366 cm-1. The latter observation can be explained by the lifting of Raman selection rules in AlAs NCs due to breaking of translational symmetry. Forbidden LO modes of the InAs matrix are observed at 237 cm-1. The frequencies of the AlAs NC phonons are shifted to the lower frequencies with respect to the bulk values (from 402 cm-1 to 367 cm-1 and from 360 cm-1 to 337 cm-1 for LO and TO phonons, respectively). The downshift is attribuited to tensile strain in the NCs due to lattice mismatch [75, 105], and diagonal strain components can be estimated to be equal to 𝜀𝑥𝑥 = 𝜀𝑦𝑦 = 0.09 and 𝜀𝑧𝑧 = −0.44 [105]. In addition to the first order Raman scattering, the second order phonon modes of InAs are observed in the range from 400 to 500 cm-1. The compound overtones arise from the combinations of optical phonons of InAs (denoted in Figure 2.7 as 2TO, 2LO, LO+TO). The nature of the shoulder at 445 cm -1 is not obvious and may originate from third order Raman scattering processes. The overtone frequencies equal a sum of the corresponding values of the vibrational modes of first-order scattering with a good accuracy (better than 3 cm–1). Normalized Raman intensities and full width at half maximum (FWHM) of the observed lines change depending on the preparation method and temperature. As can be seen from Figure 2.7, the intensity of the forbidden InAs LO peak is increasing with preparation temperature for ion milled samples as compared to the cleaved one due to relaxation of selection rules, which may be induced by defects. However, since the laser energy is close to the E1 transition in InAs [106], this effect may be caused by the change in the Fröhlich interaction, or electric field induced Raman scattering due to the Ar + milling process. The intensity increase is accompanied by the apparent asymmetric broadening of the TO mode (from 7.3 cm-1 for a cleaved sample to 10.4 cm-1 for ion milled samples) and a slight downshift of the LO and TO modes of InAs matrix (from 217 cm -1 to 216 cm-1) due to asymmetric phonon state density in partially amorphized InAs matrix layers [107] or/and qvector relaxation [108]. This is confirmed by the decrease in intensity of the InAs combinational 37

Figure 2.7. Normalized Raman spectra of ion milled and cleaved AlAs/InAs samples evidencing sample damage introduced by ion milling. According to Raman selection rules [36], in the backscattering geometry from {110} plane only TO modes are allowed, while LO modes are symmetry forbidden.

modes. Smearing out of AlAs features upon ion cutting is most probably due to stronger oxidation of AlAs as compared to the cleaved sample. It is worth mentioning that the increase of the forbidden LO feature of InAs related to the increased damage does not have to occur simultaneously with decreasing the combinational modes. The reason is varying resonant conditions for observation of the combinational modes in the samples

with ion-induced

damage.

2.3.2. Raman profiling of NC multilayers Complementary to direct structural imaging, micro-Raman scattering was used to extract depth profiles of semiconductor layered structures [109]. Confocal micro-Raman setup allows the choice of the depth from which the signal is acquired with the depth resolution about twice the lateral resolution [29]. The depth profile can be also extracted by scanning over the sample cross-section. The resolution achieved depends on the excitation light wavelength and the objective used and can be as small as 0.3 µm [110, 111]. Much smaller layer thicknesses can be determined by scanning beveled samples [111, 112]. This approach was used to identify the location and the doping profile of δ-doped GaAs with a resolution as good as 10 nm [33]. Profiling of strain, Ge composition, and defects with a resolution of 15 nm was reported for beveled Si/Si1-xGex/Si heterostructures [113]. The space charge width of 100 nm at ZnSe/GaAs

38

interface was determined from the line-shape analysis of the coupled LO phonon-plasmon mode [114]. For our structures, profiling of individual NC layers in the multilayer structure with the period of 50 – 100 nm from the cross-section by means of micro-Raman scattering fails since the period is much smaller than the laser wavelength used in the experiment. Therefore, Raman profiling of NC samples beveled at an angle of 5 – 7° with respect to the sample plane was employed. In this case the effective period of the multilayer structure increased from 100 nm to 1 µm, which is well above the wavelength and enables micro-Raman profiling and even optical imaging. Representative spectra of the beveled NC multilayers are shown in Figure 2.8. In the case of InAs/AlAs NC multilayers (Figure 2.8a) very pronounced peaks of AlAs (AlAs LO and AlAs TO) are detected from the matrix. Note that in this geometry the LO mode is allowed by Raman selection rules. Therefore, the AlAs LO peak intensity is used to obtain the Raman profile. The small intensity of the InAs NC signal and its proximity to that of As clusters appearing due to the polishing procedure obscures the direct observation of the NC layer location. GaAs LO and TO peaks originate from the GaAs interlayers. For InAs/AlGaAs NC multilayers (Figure 2.8b) the matrix signal is represented by AlAs-like and GaAs-like peaks in the region between 350 and 400 cm-1 and between 250 and 275 cm-1, respectively. The AlAslike LO peak position is downshifted by 8 cm-1 with respect to the peak positions of pure

Figure 2.8. Typical Raman spectra of beveled (a) InAs/AlAs (sample B) and (b) InAs/AlGaAs (sample C) NC multilayers. The spectra are averaged over four points.

39

Figure 2.9. Schematic structures, optical images, and Raman intensity profiles of one single stack of (a) InAs/AlAs NC and (b) InAs/AlGaAs NC structures. Schematics of the sample structure: grey areas – GaAs interlayers; white – Al(Ga)As matrix. Vertical arrows indicate the position of InAs NC layers.

material, while GaAs-like LO is downshifted by 27 cm-1, which corresponds to (0.70 ± 0.05) fraction of Al [115]. This value is in good agreement with the nominal alloy composition of 0.75 Al content. The InAs NC signal is relatively strong, but its proximity to the GaAs-like modes makes it unsuited for profiling. The GaAs LO mode originates from the GaAs interlayers. The optical image of one single stack (five layers of NCs separated by 100 nm matrix layers) of the beveled InAs/Al(Ga)As NC structures and their Raman profiles are shown in Figure 2.9. The Raman profiles represent the obvious intensity oscillations of AlAs and AlAs-like phonon modes in InAs/AlAs and InAs/Al(Ga)As structures, respectively. The positions of the maxima of the signal approximately correspond to the center of the matrix layers, while minima indicate the location of individual NC layers.

2.4. Summary Raman spectroscopy provides a wealth of information on strain, size and ordering of the embedded NCs. However, the Raman spectra from such structures always contain averaged Raman response from multiple NCs of different sizes and shapes. In InAs/AlAs NC multilayers size-selective Raman scattering by NCs was achieved with visible excitation wavelengths by employing resonant conditions. Increase of excitation energy leads to resonant conditions with 40

smaller NCs, for which electron interband transition energy is larger. As a result, for smaller NCs the confinement effect of optical phonons becomes significant, what leads to low frequency shift of optical phonon modes due to negative phonon dispersion in studied materials. Apart from that, the effects of mechanical strain and material intermixing lead to the shift of optical phonons of NCs. All three effects were considered for InAs/AlAs and AlAs/InAs NC multilayers in this chapter. The NC layers in the multilayer structure were profiled by cross-sectional AFM and TEM. In the studied NC multilayer structures, the spacing between the NC layers is 100 nm, which is smaller than the resolution provided by micro-Raman spectroscopy. The Raman spectroscopy profiling of NC layers was achieved by increasing effective period of the structure to 1 µm by beveling the structure. The period of Raman phonon intensity variations was found to correspond to the effective period of the beveled NC structures.

41

42

Chapter 3.

Surface-enhanced Raman spectroscopy

SERS is extensively used to improve the sensitivity of Raman spectroscopy. For SERS, the material of interest is typically adsorbed on a SERS substrate, which is capable of supporting LSPs when resonantly illuminated with the excitation light, and therefore enhancing the Raman signal [116, 117]. Surface-enhanced Raman scattering is typically found to provide a Raman enhancement as great as 104 – 106 over the SERS structure [118-120], or up to 1011 locally at the hot spots such that single molecules [8, 9, 58], or a few NCs [58, 121] can be detected. For higher sensitivity, it can be combined with resonant Raman, which is known as surface-enhanced resonant Raman scattering (SERRS). There are multiple approaches for preparing SERS-active substrates: vacuum deposition [122], chemical synthesis [123, 124], nanosphere lithography [125-127], electron beam lithography (EBL) [128, 129], etc. [130-133]. Among the abovementioned methods, EBL allows fabricating very reproducible structures, with desired geometrical shapes [134] and sizes of the plasmonic structures [135]. In turn, the control over the structure geometry allows systematically study the LSPR energy dependence on the substrate parameters, and the corresponding Raman enhancement. In this section, the study of SERRS enhancement of cobalt phthalocyanine (CoPc) as an organic materials, and CuS and CdSe NCs as inorganic Raman scatterers, is realized. The dependence of SERRS enhancement on resonant properties of periodic arrays of gold nanoclusters with varied diameters prepared by EBL is presented.

3.1. Materials and methods 3.1.1. SERS substrate preparation The SERS substrates were prepared by a direct-write electron beam lithography process1. The Si(100) substrate is cleaned in dimethylformamide (DMF) bath for 10 min, rinsed in deionized water, and dried in oxygen plasma for 10 min. Then 130 nm thick polymethyl methacrylate (PMMA 950K) used as an electron resist was spin-coated on the Si substrate, and structured with an electron beam (Raith-150, Germany). The structured resist is then developed in the solution of isopropyl alcohol (IPA) and methyl ethyl ketone (MEK) in the ratio of 3:1, and hardened in IPA for 30 s. The patterned resist is dried under air flow. The developed resist is 1

Substrate preparation by E.E. Rodyakina, Institute of Semiconductor Physics, Novosibirsk,

Russia.

43

Figure 3.1. Typical SEM images (400 × 400 nm 2) of gold nanocluster arrays with periods of 110, 130, and 150 nm (left, middle and right columns, respectively) fabricated on Si surfaces.

removed by a 10 s oxygen plasma treatment. Afterwards, the structure is covered with 5 nm Ti adhesion layer, and a 40 nm gold film. Finally, the remaining resist is removed by a lift-off process in 20 min hot DMF bath and 60 s ultrasound DMF bath. The final structure is rinsed in deionized water and dried in air flow. The resulting SERS substrate contains 10 µm x 10 µm squares, consisting of ordered arrays of gold nanoclusters. Structures with periods from 110 to 150 nm were prepared. The nanocluster diameter was varied from 25 to 135 nm (Figure 3.1) as estimated from scanning electron microscopy (SEM). In order to study the polarization dependence of the SERS

44

enhancement, the gold nanoclusters are arranged in dimers (dimer defined as two closely located gold nanoclusters, for details see Figure 3.4 in Section 3.2.2). The distance between the centers of the nanoclusters in a dimer is fixed to 100 nm, and the spacing between dimers is fixed to 130 nm, for all arrays, while the diameter of the gold nanoclusters changes from array to array. SEM measurements were performed using either the NanoSEM 200 from FEI with 5 kV acceleration voltage and 5 mm working distance, or with a Raith-150 system at 10 kV acceleration voltage, and 6 mm working distance.

3.1.2. Organic and inorganic materials For this study, CoPc films of different thicknesses (0.15 nm CoPc (ca. 0.5 monolayer), 0.5 nm and 5 nm) were deposited by organic molecular beam deposition (OMBD) on samples containing periodic arrays of gold nanoclusters1. The layer thickness was estimated from the calibrated frequency shift of a quartz resonator. The quartz resonator shift was calibrated on reference CoPc films, where the thickness was determined by ellipsometry. CuS NCs were synthesized using the Langmuir-Blodgett-based (LB-based) technique2 [136, 137]. The solution of behenic acid in hexane with the concentration of 3×10-4 M was spread onto the surface of deionized water (filtered through a Vladipore 0.2 μm membrane). The aqueous solution of CuSO4 with the concentration of 4×10-4 M and pH 5.0 was used as a subphase. The obtained copper behenate film with the thickness of only one monolayer was transferred (Y-type deposition) successively 20 times onto the SERS-active substrates at the constant surface pressure of 30 mN/m at room temperature. The nucleation of CuS NCs inside the behenate matrix was performed by sulfidizing copper behenate films at the pressure of 50– 100 Torr at room temperature for 1 – 2 h according to the reaction (C21H43COO)2Cu + H2S = CuS + 2C21H43COOH. Further annealing of the structure at 150°C in Ar atmosphere resulted in the removal of the organic matrix and the formation of freestanding CuS NCs. The absence of organic matrix was verified by infrared (IR) spectroscopy. The deposition of CdSe NCs was performed in similar manner, except that presynthesized CdSe NCs of 5.2 nm diameter (Lumidots) were mixed with behenic acid and deposited using the LB-based approach. After 20 cycles of the LB-based deposition, the sample

1 Thanks to M. Ludemann, E.T. Breyer, Dr. M. Fronk, Dr. D. Lehmann, Dr. F. Haidu (Semiconductor Physics, TU Chemnitz, Germany) for the help with CoPc deposition. 2 LB-based deposition by Dr. L.L. Sveshnikova and Dr. T.A. Duda, Institute of Semiconductor Physics, Novosibirsk, Russia.

45

was annealed in order to remove the organic matrix, and a monolayer of CdSe NCs was formed.

3.1.3. Micro-Raman spectroscopy measurements Raman spectroscopy experiments were performed using Jobin Yvon Dilor XY800, Horiba T64000, and Labram HR800 spectrometers1 equipped with optical microscopes (the laser beam was focused to a spot with a size of approximately 1 μm2 using 100x objective with NA 0.9) in the backscattering geometry at room temperature. Several laser lines 514.5 nm (2.41 eV) of an Ar+ laser (Coherent), 632.8 nm (1.96 eV) of a HeNe laser (Siemens), and 641.7 nm (1.93 eV), and 676.4 nm (1.83 eV) of a Kr+ laser (Coherent) were used as excitation sources. A laser power below 2 mW at the microscope was used in the experiments to avoid local heating effects. For polarization studies, the polarization of the incident light was rotated by a half-wave plate. In parallel (crossed) polarization configuration only light with the polarization parallel (perpendicular) to the polarization of the incident light was detected due to the analyzer in the detection path. The analyzer remained in the same position for each measurement in order to keep the detection path, and therefore the spectrometer sensitivity, constant. The HeNe laser was linearly polarized before the entrance of the Raman spectrometer by a birefringent polarizer. The polarized measurements were performed at laser power below 0.5 mW on the sample.

3.1.4. Micro-ellipsometry The SERS substrates were investigated by micro-ellipsometry aiming to determine the LSPR energy using a Spectroscopic Imaging Ellipsometer (EP3-SE, Accurion GmbH) with the spectral range of 365 – 1000 nm (3.4 – 1.24 eV)2. Compared to classical spectroscopic ellipsometry, this technique has spatial resolution of about 1 μm, allowing for characterization of such small area plasmonic structures. In order to extract the optical properties of gold nanoclusters from ellipsometry results, the measured sample was modeled as a multi-layer system, taking into account the optical properties of the substrate. Here a Maxwell–Garnet effective medium approximation (MG-EMA) was used, with the dielectric functions of silicon, silicon dioxide and gold from a commercially 1 Part of the Raman measurements were performed by Prof. A. Milekhin, Institute of Semiconductor Physics, Novosibirsk, Russia. 2 Micro-ellipsometry measurements and data analysis by Dr. O.D. Gordan, Semiconductor Physics group, TU Chemnitz, Germany.

46

available WVASE32 3.6 database. A silicon oxide thicknesses of 3 nm were used for the models. Using the void fraction as a fitting parameter and a low value of depolarization factor for gold, the data were remarkably well reproduced. The dielectric function obtained for the MGEMA layer is strongly dominated by the metallic nanostructures on top of the substrate, making the determination of the LSPR peak possible even by looking directly at the measured effective optical constants.

3.1.5. Numerical simulations The extinction coefficients and LSP positions were numerically modeled using the Fourier modal method with adaptive spatial resolution and matched coordinates1 [33, 34]. This approach is very efficient for calculating layered periodic systems that are usually fabricated by electron beam lithography. The model volume consists of a semi-infinite silicon substrate (constant refractive index of 𝑛 = 3.68), SiO2 layer (𝑛 = 1.46) and a 40 nm high gold nanocluster. The analytical model of the dielectric function for gold was adapted from the work of Etchegoin et al. [138]. It is important to note that the LSPR position strongly depends on the dielectric function of the gold nanocluster environment. Direct experimental determination of the dielectric function around the gold nanoclusters is challenging, since the optical response is dominated by the signal of the gold nanoclusters. Therefore, the optical response of the silicon substrate covered by a monolayer of CdSe NCs was measured. The dielectric function of the CdSe NC monolayer can be approximated for this thickness range by a Сauchy dispersion. As a result, in the spectral region of interest, the optical response of the system is equivalent to an 11 nm SiO2 layer on Si substrate, which was used in the model.

3.2. SERS on organic films SERS substrates are most often tested using organic materials such as proteins [128], self-assembled monolayers [129], and dyes [134]. Phthalocyanines are one of the common materials (also called Raman probes) for SERS [139, 140], since their Raman spectra are well studied [139, 141], and they can be deposited in very uniform layers by OMBD. Here CoPc was used for the Raman spectroscopy experiments. CoPc, as most phthalocyanines, shows a

1

Numerical simulations were performed by Jun.-Prof. Dr. T. Weiss, 4. Physikalisches Institut, Stuttgart, Germany.

47

distinct absorption peak in red around 650 nm [142, 143]. Therefore, the conditions for resonant SERRS are fulfilled in the red range of the spectrum for metal phthalocyanine thin films deposited on gold nanoclusters.

3.2.1. SERS enhancement of CoPc The Raman spectra of the films deposited on both Si and the SERS substrate were measured with different laser lines ranging from 514.5 nm to 676.4 nm. Figure 3.2a shows typical resonant Raman and SERRS spectra of a 2 nm CoPc film on the Si substrate and on gold nanoclusters measured with the 632.8 nm laser line, exhibiting characteristic Raman modes of CoPc. The mode at 1540 cm-1 (C = N stretch, 𝐵1𝑔 ) [139, 141] is especially sensitive to the nature of the metal atom present in the metal phthalocyanine [144]. In further discussion, two other modes shall be of special interest: the macro breathing mode at 682 cm-1 (𝐴1𝑔 ), and the ring stretching mode (𝐵1𝑔 ) at 750 cm-1 [139, 141]. The assignment of the modes was independently confirmed using density functional theory (DFT) calculations with Gaussian 3.0 software using B3LYP/6-31G hybrid functional approximation.1 The respective vibrations are illustrated in Figure 3.2b. The first-order Si peak originating from the substrate is located at 520 cm-1, and the second order Si Raman peak at about 900 cm-1. It is worth mentioning that the Si phonon peak for the nanostructured area is two-fold weaker than that observed for bare Si due to absorption of incident light by gold nanoclusters. Accordingly, the second order Si feature at 900 cm-1 has lower intensity in the SERS spectra, and is dominated by an enhanced CoPc mode at 960 cm-1. Note that different modes experience different enhancement factors as shown in Figure 3.2a. It is also clear that the selective enhancement does not correlate with the symmetry of the vibrational modes. This suggests that the vibrational or electronic structure of the CoPc molecules might be modified upon deposition on gold, rather than related to the change in molecular orientation. SERS and Raman spectra of the 0.15 nm and 0.5 nm thick CoPc films deposited on gold nanocluster arrays and bare Si substrates are shown in Figure 3.3a. The SERS spectra of CoPc reveal an intense vibrational mode at 683 cm−1, which corresponds to the macro breathing mode mentioned above. The features at 485 and 595 cm−1 can be assigned to ring deformation and benzene radial motion, respectively [139].

1

DFT calculations by Dr. D. Lehmann, Semiconductor Physics, TU Chemnitz, Germany.

48

Figure 3.2. (a) Raman and SERRS spectra of 2 nm CoPc film. Note the difference in scale. For selected peaks the symmetry of the corresponding vibration and the enhancement factor are indicated. (b) The vibrations corresponding to the most intense Raman modes of the CoPc molecule from DFT calculations. Arrows indicate the atoms of the central ring involved in the vibration, and the direction of their displacement.

49

Figure 3.3. a) SERRS and Raman spectra of 0.15 nm and 0.5 nm CoPc films with 676.4 nm wavelength. All spectra were vertically shifted for better representation. b) The dependence of SERS enhancement factor for the 0.15 nm CoPc film of the mode at 682 cm−1 as a function gold nanocluster diameter for the array with period of 150 nm for 632.8, 641.6, and 676.4 nm excitation. The symbols represent the experimental data, solid lines are guides for the eye.

Since no signal of CoPc was observed on the Si substrate for 0.15 nm CoPc film, the enhancement factor was determined from the intensity of the macro breathing mode (685 cm−1) observed in the Raman spectra of 5 nm CoPc film on bare Si and 0.15 nm CoPc film on the nanostructured area by the formula:

𝑰𝟏 ⁄𝒅 𝟏 𝑬𝑭 = 𝑰𝟐 ⁄𝒅 𝟐

Eq. 3.1

where 𝐼1 is the intensity of the SERS signal of the thin CoPc film (thickness 𝑑1 = 0.15 nm) deposited on the nanostructured area; 𝐼2 is the intensity of the Raman signal of the thick CoPc film (thickness 𝑑2 = 0.5 nm) deposited on bare Si. The resulting 𝐸𝐹 versus the gold nanocluster diameter for different excitation wavelengths (632.8, 641.7, and 676.4 nm) is shown in Figure 3.3b. The maximum 𝐸𝐹 of 2×104 was observed for the gold nanoclusters with diameters of 50 – 65 nm. The position of maximum enhancement shifted to larger diameters with increasing excitation wavelength from 632.8 to 676.4 nm. It implies a red shift of the resonance with increasing the gold nanocluster diameter. This is in agreement with the

50

established dependence of the resonance position on size [145], and shows that the observed Raman signal enhancement is dominated by EM enhancement mechanism.

3.2.2. Polarization dependence of enhancement in SERS For the CoPc films deposited on arrays of dimers (Figure 3.4) one can expect that SERS in this case is polarization dependent. To investigate the polarization dependence of such anisotropic structures, Raman measurements were performed with the polarization parallel to and perpendicular to the dimer axis for each structure, as defined in Figure 3.5a. In this case when the polarization of the incident light is chosen along the dimer axis, the electric field behavior is defined by the distance between the gold nanoclusters in the dimer (100 nm); otherwise, it is defined by the distance between neighboring dimers (130 nm). Figure 3.5b and c shows the dependence of SERS enhancement factors on the nanocluster size under the two excitation wavelengths (514.5 and 632.8 nm) obtained for two polarizations of the incident light (Figure 3.5a). The enhancement factor is calculated as the ratio of the intensity of the 𝐵1𝑔 mode of CoPc at 1540 cm-1 on the gold nanostructure versus that on the bare Si substrate under the same measurement conditions.

Figure 3.4. SEM image of one of the dimer structures. The distance between centers of the clusters is kept constant for all arrays (100 nm between the nanoclusters in a dimer and 130 nm between the neighboring dimers), while the gold nanocluster diameter changes from array to array from 25 to 90 nm.

51

Figure 3.5. a) A schematic of a dimer and polarization direction of the incident light with respect to the dimers indicated by arrows; b) and c) are the enhancement factors for the structures with different dimer dimensions measured with 514.5 and 632.8 nm excitation line, respectively, and polarization ( ○) along and (■) across the dimers. Solid lines are guides for the eye. Arrows indicate enhancement factors for a nanostructured gold film, measured with the polarization across (┴) and along (‖) the dimers.

The enhancement factor for the 514.5 nm laser line does not exceed a factor of 10 and is independent of the nanocluster size since it is far from the plasmonic resonance. For the 632.8 nm line, the SERS enhancement factor on the gold nanoclusters increases to 100 with decreasing distance between the nanoclusters (i.e., increasing nanocluster size). This result is consistent with those obtained on similar SERS structures [146], which showed that a smaller distance between gold nanostructures leads to greater enhancement [129]. The results in Figure 3.5b demonstrate that there is no difference when the polarization is parallel to or perpendicular to the dimer axis for the 514.5 nm excitation wavelength, while a significant difference exists between the enhancement factors for the two polarizations with 632.8 nm wavelength (Figure 3.5c). Following some of the reported results [53, 147], it was expected to observe greater enhancement factors when the light is polarized parallel to the dimer axis because the hot spots are confined in a smaller gap and, therefore, the enhancement should be greater. However, as it was shown in few other works [53, 148], the coupling of the LSPs of two closely positioned metallic nanostructures leads to a red shift of the resonance so that at a fixed wavelength the enhancement factor may decrease despite the higher electric field confinement. Therefore, the polarization dependence can be attributed to the shift in LSPR energy for both polarizations.

52

3.3. SERS by nanocrytals While the majority of the SERS and TERS studies are performed on organic materials, the SERS-effect by inorganic semiconductor nanostructures such as CdS, CdSe, GaN, and ZnO was only recently observed [121, 137, 149, 150]. Significant enhancement of Raman scattering by optical and surface optical phonons in NCs and nanocolumns was demonstrated. Here the NCs of two semiconductor materials, copper sulfide (CuxS, x = 1...2) and cadmium selenide (CdSe), are investigated, which are promising for numerous applications in optoelectronics and photovoltaics [151-153]. The band gap energy of CuxS varies from 1.8 eV to 2 eV depending on size and composition [154], and for CdSe NCs it is 2 eV, which is comparable to the LSPR energy of gold clusters. Thus both are also suitable for SERRS observation. Gold nanocluster arrays analogous to those used for SERS on CoPc were used for SERS enhancement. The typical SEM image of the edge of the periodic gold nanocluster array on the Si substrate is presented in Figure 3.6a. It reveals a quite homogeneous distribution of CuS NCs over the surface of bare Si as well as the area with arrays of gold nanoclusters. The Raman spectrum of CuS NCs formed on bare Si shows only one strong Raman line at 521 cm−1 related to the optical phonon of Si ( Figure 3.6b). The Raman spectrum of CuS NCs deposited

Figure 3.6. Typical SEM image of the structure with CuS NCs formed on bare Si (bottom part of the image) and arrays of gold nanoclusters (top part of the image). (b) SERS and Raman spectra of CuS NCs formed on arrays of gold nanoclusters and on bare Si and measured with 514.5 nm laser line.

53

on the array of gold nanoclusters ( Figure 3.6b), on the other hand, reveals in addition to the Si phonon peak a new feature near 474 cm−1, which is attributed to the optical phonon of CuS NCs [155]. The appearance of the new feature manifests the SERS effect by optical phonons in CuS NCs. The determination of the enhancement factor of SERS in CuS NCs is hardly possible because of the NC clustering in the vicinity of the gold structures. Compared to CuS NCs, CdSe NCs form a very homogeneous monolayer coverage, allowing for more quantitative estimation of SERS enhancement by comparing the Raman signal on the SERS substrates and on Si (Figure 3.7a). Figure 3.7b demonstrates the Raman and SERS spectra of the CdSe NC monolayer. While Raman spectra acquired on the CdSe covered Si substrate show no detectable signal of the CdSe NCs, SERS allows the detection of an intense LO feature of the CdSe NCs at 210 cm-1, and higher order features, revealing excellent crystalline quality of the NCs. The dependence of the Raman enhancement on the gold nanocluster size measured with different wavelengths (Figure 3.8a) shows resonant behavior. The SERS signal maximum shifts to the red with increasing the nanocluster size, as already established for organic films. In order to verify that this behavior originates from EM enhancement, the LSPR positions were

Figure 3.7. (a) SEM image of an edge of the nanostructured gold substrate covered with a monolayer of CdSe NCs. Inset shows an example of high magnification image on the silicon substrate covered with CdSe NCs. (b) Raman (black curve) and SERS (red curve) spectra of CdSe NCs measured with 632.8 nm laser line.

54

Figure 3.8. (a) Dependence of Raman intensity of the CdSe LO phonon on the gold nanocluster size for different wavelengths; (b) Comparison of experimental and simulated LSPR positions for different nanocluster diameters with parameters, for which maximum SERS intensities were observed.

characterized by spectroscopic micro-ellipsometry. The LSPR position was determined as the maximum of the extinction coefficient spectrum extracted from ellipsometry measurements. The dependence of LSPR position on the gold nanocluster diameter is shown in Figure 3.8b. The data show an increase of the LSPR position with increasing the gold nanocluster diameter. Additionally, the extinction coefficients and LSPR positions were numerically modeled using the Fourier modal method with adaptive spatial resolution and matched coordinates. The calculated LSPR positions of the gold nanoclusters on the Si substrate are in good agreement with the experiment (Figure 3.8b), showing the red shift of the LSPR position with increase of the nanocluster diameter. When the distance between the neighboring nanoclusters drops below 20 nm, the near-field coupling effect leads to further red shift of the LSPR position. Experimentally determined LSPR positions, SERS enhancement maxima and LSPR positions obtained from the calculations are in a good agreement with each other, and prove the EM nature of the SERS enhancement in CuS and CdSe NCs.

3.4. Summary Ultra-thin organic films and semiconducting nanocrystals deposited on SERS substrates were studied. The SERS substrates were realized as periodic arrays of gold nanoclusters with systematically changing nanocluster size. The LSPR energy position depending on the gold nanocluster size was determined by spectroscopic micro-ellipsometry. It allowed tuning of the

55

LSPR energy position to the energy of electronic transition in organic material, and to the energy of the interband transitions of the semiconducting NCs, and thus to realize SERRS. CoPc films of different thicknesses were used as organic material. The SERS enhancement of up to 2×104 was achieved for CoPc Raman signal. Raman signal intensities obtained under different excitation wavelengths showed a resonant behavior of the SERS enhancement. The polarization dependent measurements with 632.8 nm laser line on gold dimers showed stronger enhancement for polarization across the dimer, which was attributed to the shift of the LSPR position. Remarkable enhancement was demonstrated for the Raman scattering by optical phonons CuS and CdSe NC layers using the same type of SERS substrates, manifesting SERS effect in inorganic NCs. Matching the laser excitation energy to the energy of the interband transition in NCs leads to SERRS by optical phonon modes, which also allowed the detection of high order phonon modes for CdSe NCs. Similarly to organic material CoPc layer, the resonance behavior of the SERS enhancement with changing the excitation wavelength was obtained, which was attributed to LSPR of gold nanoclusters. The observed wavelength and polarization dependences of SERS enhancement both for organic and inorganic materials evidences its EM origin.

56

Chapter 4.

Implementation of tip-enhanced Raman spectroscopy

Tip-enhanced Raman spectroscopy (TERS, nano-Raman) technique provides not only increased sensitivity, but also the additional benefit of sub-diffraction spatial resolution, wherein a plasmonic tip is brought close to the surface and thereby enhances the signal directly under the tip apex. Now, after the first demonstrations of TERS in 2000, the technique is entering the stage when it is the sample, and not the method, which is being studied. But technical optimization and methodological advances are still a matter of research interest. When designing a TERS system, it is important to choose a suitable setup configuration for the intended samples, and simultaneously to maximize the quality of the measurements. In a simplified view, the TERS system can be considered as 

a TERS-active tip required to create enhanced electric field due to LSPR,



the illumination/collection system to focus the incident laser beam on the tip apex and direct the scattered signal to the spectrometer,



tip/sample control to control the tip-sample distance and the tip lateral position on the sample (Figure 4.1).

Both the tip feedback loop and the Raman spectrometer design are well-established, the main challenge is the optical coupling of the two systems, and synchronization of their work. For the sake of clarity, first the concepts of contrast and enhancement factor are introduced, and afterwards the versatility of the setup and its efficiency are considered. Further the chapter is devoted to the implementation of TERS in our lab, both optical coupling, and optimization of all-metal AFM TERS tips.

Figure 4.1. General schematic of a TERS experiment.

57

4.1. TERS enhancement factor The illumination/collection system focuses the laser beam in a spot, which is typically two orders of magnitude larger than the tip radius (Figure 4.2a). Therefore, the signal from all of the illuminated volume will be detected all time. It is referred to as the far-field signal 𝐼𝑓𝑎𝑟 (tip up in Figure 4.2b). When the tip is approached to the sample surface and is in the laser focus, it will create an enhanced electric field only in the small volume underneath the tip apex. The additional Raman signal arising from the enhanced volume is referred to as a near-field signal

𝐼𝑛𝑒𝑎𝑟 (tip down in Figure 4.2b). Thus, when the tip is on the sample, the signal 𝐼𝑡𝑜𝑡𝑎𝑙 constitutes the near and the far field contributions. The contrast 𝐶 is then defined as the ratio between near field and far field signals:

𝑪=

𝑰𝒏𝒆𝒂𝒓 𝑰𝒕𝒐𝒕𝒂𝒍 = −𝟏 𝑰𝒇𝒂𝒓 𝑰𝒇𝒂𝒓

Eq. 4.1

This value is very practical for the experiment, but is not suitable for the interlaboratory comparison, since it carries no information about the signal enhancement provided by the tip. In order to determine the enhancement factor 𝐸𝐹 , the near field and the far field signals have to be normalized by the enhanced 𝑉𝑛𝑒𝑎𝑟 and illuminated 𝑉𝑓𝑎𝑟 volumes, respectively (Figure 4.2a). Clearly, the smaller the illuminated volume, the smaller enhancement factor is needed to achieve a detectable contrast, this is the reason why high NA optics is needed in TERS experiments.

𝑬𝑭 =

𝑽𝒇𝒂𝒓 𝑰𝒏𝒆𝒂𝒓 𝑽𝒇𝒂𝒓 =𝑪 𝑽𝒏𝒆𝒂𝒓 𝑰𝒇𝒂𝒓 𝑽𝒏𝒆𝒂𝒓

Eq. 4.2

Even though this is a commonly used approach to determine the 𝐸𝐹 , and therefore allows comparison with other literature, it is highly imprecise. There are several sources of experimental errors: 

The far-field signal linearly increases when the tip approaches the sample due to

increased scattering by the tip [156, 157]. A simple tip up/tip down comparison cannot account for this change in signal intensity. 

A proper estimation of illuminated volume is crucial for the 𝐸𝐹 estimation. The

area of the laser spot can be either assumed to be diffraction limited, but for more precise

58

Figure 4.2. (a) A schematic of a generic TERS experiment illustrating illuminated area as compared to the tip apex radius; (b) A generic comparison of the far- and near-field signals.

estimation it should be measured using a knife edge test [158]. The depth of the volume is defined by the smallest of the following values: sample thickness, information depth defined by the optical absorption of the sample, or the depth resolution defined by the objective and the laser wavelength when using a confocal aperture at the detector side. 

The estimation of enhanced volume is not straightforward. It is defined by the tip

radius, but the research supporting the ratio between the tip radius and the enhanced volume is scarce. When the sample is well defined (such as isolated CNTs), the enhanced area can be estimated from the achieved resolution. For an estimation of enhanced volume between two nanoparticles the enhanced area diameter was theoretically shown to be 𝑑 = √𝑅𝑡𝑖𝑝 𝑔 [159], where 𝑅𝑡𝑖𝑝 is the tip radius, and 𝑔 – the gap between the tip and the substrate defined by the nanocluster size. When it is not possible, the best approach may be assuming the enhanced volume radius to be 75% of the tip radius based on FEM simulations [160]. The penetration depth of the enhancement may strongly depend on the tip radius and the optical properties of the sample. Recently, in the literature several alternative methods were proposed for the 𝐸𝐹 estimation. Roy et al. [161] proposed using TERS imaging of a single CNT by scanning a tip

59

over the sample. The resulting profile corresponds to the sum of near-field profile (signal obtained when the tip is going above the CNT) and far-field profile (signal obtained due to scattering from the tip shaft when the tip crosses the laser spot). This method allows normalizing near-field to the far-field directly from experimental data. Using the example of a CNT imaging from the work of Hartschuh et al. [162], the authors showed that the 𝐸𝐹 was overestimated by a factor of 50 as compared to tip up / tip down estimation. Kumar et al. [157] suggested to use the tip up / tip down comparison when measuring a two layered sample, where only the signal from the upper layer will experience enhancement due to near-field contribution, while enhancement of the bottom layer can be attributed to the far-field contribution. The benefits of those methods include precise and reproducible 𝐸𝐹 estimation, but they still do not give information about the effect of the optical properties of the substrate on the field enhancement and resolution. Bortchagovsky et al. [163] proposed functionalizing the TERS tip with a thiophenol (PhS) layer, and using the PhS signal as a reference. The amount of material remains constant at each point of the image allowing for local monitoring of the enhancement. The drawback of this method includes possible presence of the Raman signal from the PhS layer, which may interfere with analysis of Raman spectra.

4.2. State of the art of optical systems for TERS The optical coupling systems used for TERS can be divided into two classes. In the transmission mode (bottom illumination or inverted microscope) systems the laser is focused on the tip from the bottom through a transparent sample. The first TERS demonstrations used only the transmission mode systems [15-17]. TERS based on the reflection mode open access to the study of non-transparent systems. These include top and side illumination configurations, as well as parabolic mirror coupling. The bottom illumination configuration allows using oil immersion objectives with NA up to 1.4 [16, 17]. This system provides the best contrast in TERS due to the smallest illumination volume, but it is limited to transparent samples only. It remains the most popular TERS configuration from the first reports till now, also due to its ease of use and straightforward alignment. The beneficial illumination polarization along the tip is realized by higher order laser modes such as a tightly focused Hermite-Gaussian (1,0) mode [164], placing the tip in one of the lobes with the longitudinal field component produced by a tightly focused Gaussian beam [162], or radially polarized light [165]. Although all of the modes do provide the required 60

polarization, the radial polarization gives a sharper TERS image than the linear one [165]. The reason is the stronger 𝑧 component for a radially polarized mode. Application of a mask restricting the illumination angle to total internal reflection (TIR) angles allows exciting only nearfield signal [16, 166]. The parabolic mirror configuration provides the next highest NA of up to 0.99 [167], it is low temperature and UHV compatible [168, 169], with either shear force [167] or STM feedback positioning systems [169]. The drawback is that its collection efficiency and focal fields are very sensitive to the alignment [170]. Even though the sample can be opaque, it has to be small enough to allow focusing the light from the bottom. The top-illumination system was first realized in 2003 by W.X. Sun and Z.X. Shen [171]. Top illumination configuration can use long working distance (LWD) objectives with NA of up to 0.7 [172], but its implementation requires open access for the bulky LWD high NA optics from the top and is mostly available in commercially available TERS-ready systems. It can benefit from higher laser modes the same way the bottom illumination does. The side illumination configuration is one of the most prevalent and versatile configurations along with the bottom illumination due the easy implementation when coupling Raman and any of the available tip-sample distance control systems (AFM [173-175], tuning fork [176], and STM [40]). This combination allows one to study the largest variety of objects, including opaque, large and non-conductive samples. The NAs reported in the literature are typically below 0.5 [177], which is limited by the working distance of the objective. The far field illumination volume is further increased by the illumination angle. The benefit is easy adjustment of polarization along the tip axis using linear polarization. The latter configuration was chosen for implementation in our lab as the most versatile solution.

4.3. Implementation of the optical system When designing our TERS system, we adapted a side illumination/collection setup with Agilent AFM 5420 and LabRam HR800 Raman spectrometer from Horiba. The purpose of optical coupling is focusing the laser spot on the tip apex, and providing their stability. In the first design, a hyper long working distance (HLWD) 20x objective HL20 (Shibuya, Japan) with NA = 0.29 and working distance of 31 mm was used for both focusing the laser beam and collecting

61

the Raman signal1. The illumination angle was 20° with respect to the sample plane, which results in an elliptic shape of the laser spot. The size of the laser spot was experimentally determined by scanning over the edge of a 30 nm layer of cobalt phthalocyanine (CoPc) deposited on a gold substrate2 (Figure 4.3a). The thin films and sharp well-defined edge were achieved by photolithography that allows minimizing the effect of the sample topography on the experimentally determined laser spot size. The sample was scanned in X and Y directions (along the short and the long axes of the laser spot) using 514.5 nm laser line and 100 μm confocal hole aperture. The Raman signal intensity of the CoPc vibration at 1545 cm−1 was followed (Figure 4.3c), and the derivative of the obtained curve was calculated numerically. The laser spot size was determined from the full width at half maximum (FWHM) of the Gaussian peak in the obtained derivative. The resulting values gave a spot size of 4.7x2.5 μm2 with an uncertainty of 8% [158]. This approach can be considered as a modification of a classical scanning knife-edge method adapted for side-illumination TERS. In the later optimized design, it was possible to adapt a 50x HLWD objective with NA of 0.42 (HLB SL50, Shibuya, Japan) with working distance of 20.5 mm. It allowed us to reduce the

Figure 4.3. Spot size determination using the scanning knife-edge method. (a) AFM image of the CoPc/gold interface obtained by photolithography. The sharp interface is evidenced in the cross section shown in (b). A Raman line scan was performed across the interface; the intensity of the CoPc band at 1545 cm−1 is plotted in (c). FWHM of the derivate of the line scan gives a value of 4.7 μm. The approximate size of the laser spot gives a value of 4.7x2.5 μm2 with accuracy 8%. 1 Thanks to A. Villabona (Laboratorio de Fisica Aplicada, Universidad de Los Andes, Merida, Venezuela) and A. Fechner (Semiconductor Physics, TU Chemnitz) for invaluable help in developing the TERS setup. 2 The sample was provided by Dr. Pablo R. F. Siles from IFW Forschungs-Standort Chemnitz, Smart Systems Campus.

62

laser spot size to (1.8±0.6)x(2.8±0.8) µm2 for the 514.7 nm laser line at an incidence angle of 20° as measured using the approach described above. As can be seen in Figure 4.4, the laser beam exits the LabRam objective revolver, and is then reflected from two mirrors. It allows for fast and non-invasive switching between micro-Raman and TERS configurations. The objective laser path is protected by a metallic tube for safety reasons. The holder of the AFM head was redesigned to be able to bring the objective close enough to focus on the tip. For the same purpose the samples have to be lifted from the surface by about 1 cm, so most of the samples can be mounted on specially prepared copper or bronze blocks. For the rough alignment the whole AFM can be moved on a manual XY stage with a 2 cm range in each direction. The optical coupling system can be moved vertically along the rail by a single axis micrometer translation stage. The built-in top-view camera of the AFM system, as well as an independent microscope camera with a 20x HLWD objective, can be used for the rough alignment. Once the rough alignment is performed and the tip apex is close to the focus of the TERS objective, the fine alignment is performed with piezoelectically driven motors controlling mirror and the objective focus. The built-in camera of the LabRam spectrometer is used for the fine alignment. During the imaging, the AFM tip is moved vertically by the AFM head scanner, and the XY movement of the sample is performed with an additional piezoelectric scanner (nPoint, Inc.) with a scanning range of 100x100 µm2 so that the tip remains in focus with the laser during the scan. Alternatively, the tip can be moved in the XY plane by the original AFM scanner in the range of 10x10 µm2.

Figure 4.4. Photo (a) and schematic layout (b) of the TERS setup. Inset in (a) shows an optical image of an AFM TERS tip. Red lines indicate the light path.

63

The TERS scan is performed by acquiring the Raman spectra at preset positions of the plasmonic tip on the sample. The Raman data and the corresponding AFM image can be acquired and saved independently, line by line. The synchronization of the work of AFM and Raman setups is crucially important for such systems, and is achieved by software coupling1.

4.4.

TERS tips

4.4.1. State of the art of TERS tips There is an opinion in the literature regarding what the perfect TERS tip should be like. Since the achievable resolution is defined by the tip apex radius, then the tip size needs to be minimized [160]. Additionally, the small asperity at the tip apex is expected to produce stronger enhancement [178]. On the other hand, there is also an opinion, that the tip radius below 10 nm would lead to damping of enhancement due to too small amount of electrons available there [179]. A large tip radius can be beneficial due to larger decay length of the near-field as the distance from the tip apex increases, which allowed sensing subsurface structures such as CNTs under 5 – 10 nm of SiO2 [180]. Since systematic experimental study of the optimal tip roughness and apex is challenging, so far there are mostly theoretical studies addressing this question [181]. A tip apex with opening angle from 15° to 30° also serves the purpose of higher enhancement, although the effect does not exceed 30% in terms of field enhancement [181]. Finally, the enhancing grains or nanoparticles must be localized at the tip apex, with the tip body being smooth since such tips would supposedly tend to absorb less contaminants from the air [182]. There are several tip preparation techniques described in the literature, including attaching metal nanoparticles at the tip apex [183], welding nanowires [184] or Si nanowires with gold caps [185] to the tip, synthesizing multilayer prism [186], plasmonic silver layer and protective gold layer [186], or applying silicon templates [187] to improve reproducibility. The two most commonly applied tip fabrication approaches include electrochemical etching and metal coating depending on the tip-distance control system used. STM and SFM require an electrochemically etched wire, while AFM uses the deflection of the AFM cantilever, the mechanical behavior of the tip becomes important. This means that the cantilever size and

1

Software was developed by J. Nelson and A. Ueki from Agilent Technologies with important support from E. Leroy of Horiba JY.

64

shape need to be well controlled. This can be achieved with commercial silicon or silicon nitride AFM tips that the TERS user normally coats by a suitable (gold, silver, aluminum) metallic film. Electrochemical etching received considerable attention in the literature. The first tip etching protocol for the purpose of TERS experiments was suggested in 2004 [182]. Etched wire, the cathode, was submerged in a solution of HCl and ethanol. The anode was formed by a gold ring on the surface of the solution. Etching occurs mostly in the meniscus where the intermixing rate is the highest. The etching leads to formation of oxide [182] (or complexes [188]) on the tip surface, which suppresses the etching rate. This in turn leads to the accumulation of Cl-ions at the interface until the etching rate increases again. It results in current oscillations compromising reproducibility. Fast formation of H2 and Cl2, especially at voltages above 1.4 V, leads to bubbling, that makes the tip rough and irreproducible. The bubbling can be suppressed by adding ethanol, since it lowers the surface tension of the solution and allows faster intermixing of the components, improving tip smoothness. Alternatively, isopropanol can be added, since it allows to tune the surface tension in a wider range [188]. Different ways were proposed to improve tip reproducibility: a bottomless beaker producing more stable ion current [179, 189], a reference electrode for stable control of the voltage level [188], and constant current etching producing tips with radius below 20 nm [190]. Silver can be etched with nitric acid [191], but the tip surface tends to be rougher than for gold. The prediction of the tip performance in TERS is not straightforward. Surprisingly, screening the tips based on their geometry by SEM is not sufficient [162, 192, 193]. One possible reason is the multicrystalline structure of the tip. It was suggested that annealing the wires before etching leads to a single crystal tip end, and all sharp tips produced of annealed wires give TERS enhancement [156]. The second preparation procedure by metal coating, widely used in AFM, is simple to realize by metal evaporation. The metal coating can be performed with different noble metals to match the plasmonic resonance with the wavelength used. But metals tend to form grains when being evaporated on the surface [194], and since it is a stochastic process, obtaining grains aligned on the tip such that they provide reasonable enhancement can be hard. At low coverage the probability of obtaining an active nanoparticle at the tip apex is low, while high surface coverage compromise the spatial resolution. Luckily, it is possible to find a reasonable compromise by tuning the thickness of the metal coating [174, 194]. The success rate for metal coated tips often does not exceed 50% [173, 174]. Recently, 100% gain and high enhancement were claimed for silver coated SiO2 tips produced by baking commercial Si tips at 1000°C for 10 hours and subsequent coating with metal [195]. Low refractive index of SiO2 (𝑛 = 1.5) as 65

compared to Si (𝑛 = 4.15 using 532 nm wavelength) shifts the plasmonic resonance of the tip and produces enhancement at 532 nm. Unfortunately, the most important drawback of the metal coated tips is the wear-off of the coating during scanning that seriously compromise stability and reproducibility [158], is not addressed by this method. Tips can be also optimized in order to separate near-field and far-field contribution. One of the possible approaches is spatially separating in-coupling of incident light from the tip apex, where the near field is concentrated. The in-coupling can be performed using a grating on the tip shaft with a period close the laser wavelength formed on the surface of a smooth tip [196] or via a photonic crystal [197], and the energy transfer to the apex is performed by propagating surface plasmons. This method is promising due to the absence of background signal (far field), which can often dominate the TERS spectrum limiting data quality and making data analysis more challenging.

4.4.2. Fabrication of tips for AFM-based TERS As discussed above, electrochemically etched all-metal tips remain the most successful and reproducible solution for TERS experiments. But their application in AFM-based TERS setups is impossible due to the operation principle of the AFM, where the tip is required to have a reflective cantilever. This is needed since the cantilever deflection is monitored using the optical lever method standard in AFM. Fully metallic tips suitable for AFM operation were developed previously1 [158, 198]. Shortly, the tips are prepared from 99.99% pure gold or silver wires of 50 µm diameter (GoodFellow, UK). The wires are flattened forming the reflective cantilever, and bent at a desired angle. By tuning the angle, it is possible to achieve optical access to the tip apex from the top, which is convenient for positioning in AFM with a top camera, or in TERS in the top illumination/collection configuration. The wires are integrated on standard silicon AFM chips. Such tip already often has a sharp edge formed by razor blade cut and can readily be used for AFM and TERS measurements as shown in Figure 4.5a

[158], but this approach lacks

reproducibility. The performance of the tips in AFM imaging in both contact and intermittent contact mode was verified using a TGZ02 calibration grating (Mikro-Masch) consisting of 1D arrays of rectangular SiO2 steps on a Si substrate. The steps have a nominal height of 119 nm and a periodicity of 3000 nm. The structure is coated by a thin layer of amorphous silicon nitride to protect the Si from oxidation. 1

Full-metal TERS tips were developed and prepared by Dr. R.D. Rodriguez, Semiconductor Physics group, TU Chemnitz, Germany.

66

Figure 4.5. (a) SEM image of a mechanically cut custom-made silver AFM cantilever, (b) AFM image in intermittent contact mode of the 1D array of silicon stripes acquired with this cantilever, (c) Height profile along the dashed line in (b). (d-f) Analogous set of data for an electrochemically etched gold tip.

The tip apex can be sharpened in a much more controlled and reproducible way by electrochemical etching (Figure 4.5d). The etching procedure for gold and silver nanowires using saturated water solutions of KNO3 and KBr, respectively. The tip apex is immersed in the etching solution, and etched by applying voltage pulses. Tips with radius below 10 nm can be produced this way (Figure 4.6a). Scanning electron micrographs of the cantilevers were acquired using a Nova NanoSEM 200 from FEI. For comparison, commercial silicon cantilevers (NSC14, Mikro-Masch) were coated with a silver layer by thermal evaporation in vacuum (10−6 mbar) until reaching a nominal thickness of 60 nm. After being used for AFM imaging, the metal coating appeared to wear off (Figure 4.6b, c). The wear-off of the metal coating is a well-known issue in electrical AFM, where metal coatings such as Pt are used to provide electrical conductivity. Additionally, for the tips, which 67

Figure 4.6. (a) SEM image of an electrochemically etched gold tip. (b)-(c) Wear off of the metal coating of a conventional silicon tip with a silver layer. (b) The tip before and (c) after an AFM-TERS experiment.

were subject to illumination, the melting of the metal coating was observed. Those may be the reasons for the reported poor reproducibility of TERS enhancement for metal-coated AFM tips [173, 174].

4.4.3. Mechanical properties of fully metallic TERS tips The mechanical performance of the tips needs to be optimized for the AFM imaging. AFM measurements are most often performed in contact or intermittent contact imaging mode. Contact mode is based on measuring the deflection of the cantilever when it is in close contact to the sample surface. This method is beneficial for TERS imaging due to direct contact between the tip and the sample, leading to high enhancement, but can be detrimental when used for soft sample surfaces. Therefore, in a daily AFM practice a gentler intermittent contact mode is often preferred. It is based on driving an AFM tip to oscillate at a frequency close to its mechanical resonance frequency, and measuring the decrease in the tip oscillation amplitude that depends on the tip-sample distance. Thus it is important that the tips should be available both for contact or intermittent contact modes. There are two crucial values defining the mechanical performance of an AFM tip: the spring constant and the resonance frequency. Most of the commercially available silicon tips have well defined rectangular shaped cantilevers with constant cross-section, and both

68

spring constant 𝑘 and resonant frequency 𝑓0 of a fundamental mode can be easily calculated. For a rectangular cantilever of length 𝐿, width 𝑊 and thickness 𝐻 fixed at one end

𝑯 𝟑 𝒌 ∝ 𝑬𝑾 ( ) 𝑳

𝒇𝟎 ∝ √

𝑬𝒉 𝝆 𝑳𝟐

Eq. 4.3

Eq. 4.4

where 𝐸 and 𝜌 are the Young’s modulus and the density of the cantilever material, respectively. For contact mode soft cantilevers with 𝑘 < 0.1 N/m can be used, while for intermittent contact mode 𝑘 > 1 N/m are needed as they provide high quality factor. When flattening a metallic wire, cantilevers of a complex shape are obtained. In order to tailor mechanical properties of the all-metal cantilevers, FEM analysis was performed with increasing model complexity from simple rectangular cantilevers to realistically shaped cantilevers. The shape was deduced from SEM images of the tips (Figure 4.7). For our cantilevers, which have arbitrary shape, a FEM model is the approach of choice since it allows a more detailed description of the cantilever geometry improving the accuracy of the model [199]. The cantilevers were modeled in ANSYS Mechanical APDL Release V141. The material parameters used in the calculations are given in Table 4.1. The models were developed using

Figure 4.7. The models used for the analysis of the mechanical properties. From left to right (increasing model complexity): (a) rectangular cantilever, (b) single segment cantilever, (c) two segment cantilever, and (d) realistic cantilever. Numbers denote the index of the 𝐻 and 𝑊 parameter of the respective cross-section. Grey planes indicate fixed end of the cantilever.

1

In close collaboration with Dr. V. Kolchuzhin, Microsystems and Biomedical Engineering, TU Chemnitz, Germany.

69

ANSYS element type SOLID45. For extraction of fundamental frequencies, the modal analysis of the cantilevers was performed. This analysis uses the Block-Lanczos method, which is the classical solver for the extraction of modal frequencies and their respective modal shapes that occur naturally in mechanical structures with many degrees of freedom. The static structural analysis was used for determination of the spring constants. The deflection of the cantilever was assumed to be small enough so that the Hooke’s law is satisfied and non-linear effects can be neglected. A point force 𝐹 (1 μN) was applied at the free end along the cantilever middle line that resulted in displacements 𝑢 on the order of nanometers. The spring constant was determined using Hooke’s law:

𝒌=

𝑭 𝒖

Eq. 4.5

Table 4.1. Material parameters for gold and silver used in the simulations.

Young's modulus / GPa Gold

79

Poisson ratio

Density, g/cm3

0.44

19.3

The parameter ranges for the first three models (given in Table 4.2) are based on the size and geometry screening of multiple metal cantilevers by SEM1. For the rectangular cantilever model, the dependence of the spring constant and the fundamental frequency on the geometrical parameters of the cantilever reproduces the analytical dependence described by Eq. 4.3. The results of the modal analysis for the first four modes of the rectangular cantilever are presented in Figure 4.8 (upper row). The single segment model is created by connecting two rectangular cross-sections by straight lines. Due to the fabrication process, the free end of the cantilever is thinner than the fixed end. To understand the effect of this gradual change in cross-section, a single segment model was created (Figure 4.7b). The modal analysis shows (Figure 4.8, lower row) that due to the gradual decrease of the cantilever thickness, only a thin part of the cantilever length is contributing to the oscillation. In other words, the effective cantilever length is smaller. Secondly, the thin and wide free end tends to cause torsional oscillations. In the experiment, such modes

1

In the framework of a research project by Vivek Desale, Semiconductor Physics group, TU Chemnitz, Germany.

70

would induce lateral forces between the tip and the sample that could lead to sample damage, and therefore should be avoided by choosing out-of-plane vibrations. The analysis of the spring constant dependence on the geometrical parameters of the cantilevers (Figure 4.9) shows that the spring constant is inversely proportional to the third power of the length 𝐿, as for the rectangular cantilever. It is also directly proportional to the third power of the fixed end thickness 𝐻1 , and linearly increases with increasing the width of the fixed Table 4.2. Size range and constrains applied for simulations of the simplified cantilever models.

Length / μm

Height / μm

Width / μm

Rectangular cantilever

𝐿=

Single segment

500..3000

model

𝐻 = 5..50

𝐿2 =

model

𝐿1 =1500

Cross-

section 2

section 3

𝐻2 = 𝐻1



𝑊2 = 𝑊1

– –

𝑊= 50..200

Two segment

Cross-

𝐻2 ≤ 𝐻1 𝑊2 ≥ 𝑊1



𝐻3 ≤ 𝐻2 𝑊3 ≥ 𝑊2

Figure 4.8. Modal analysis of rectangular (upper row) and single segment (lower row) beams made of gold. Geometrical parameters used in the calculations are given on the left-hand side. The color scale indicates the normalized displacement (red stands for highest displacement from the original position, blue for lowest). Arrows indicate the direction of the oscillation (out-of-plane (oop), in-plane (ip), or torsional (tors)).

71

end 𝑊1 . The system is much less sensitive to the parameters of the free end. The spring constant is only weakly changing with 𝑊2 (the width of the free end). The dependence on the free end thickness 𝐻2 is linear, which is not expected for the rectangular cantilever model. Also one can see deviation from the linear behavior for the 𝐻2 values below 20 µm, especially prominent at high 𝑊2 . The reason is bending of the thin cantilever end when applying point force. This deviation is not expected for the real cantilevers, since in the real device the force is applied to the tip apex, and then redistributed across a wide area of the cantilever via the etched cone. For a two segment model, the height and width of the intermediated cross-section 𝐻2 and 𝑊2 give dependence similar to that of the fixed end parameters. For the sake of simplicity, the detailed analysis of the simulation results is omitted. As can be seen in Figure 4.9, most of the spring constant values in the typical size range are well above unity, which means that they are suitable only for intermittent contact mode imaging. Based on these results, we can conclude that gold tips for contact mode would be required to have length above 𝐿 > 3000 µm and 𝐻1 < 5 µm. Even though the spring constant can be also tuned by adjusting other geometrical parameters, it is much less sensitive to their changes. Using the current protocol for the cantilever fabrication, most of the tips are suitable for intermittent contact mode due to their high spring constant. In order to test the validity of the model, the tips were simulated with realistic geometries and structural parameters deduced from SEM measurements1. The complex tip geometry was approximated by a five segment model, and a tilted cone in order to simulate the tip. Frequency response of the real cantilever was compared to the results of the harmonic analysis. The harmonic analysis implies driving the cantilever oscillation by an external force modulated at different frequencies. It allows to deduce out-of-plane modes, which will be observed when performing AFM measurements, and estimate their relative intensity. The model was developed with ANSYS element type SOLID95, where the bulk is made of 20-node 3D elements and is most suitable for linear and also nonlinear analysis of structures. The FEM model of this cantilever is shown in Figure 4.10a. Figure 4.10b shows excellent agreement between the experimental and the simulated frequency response, confirming the validity of the model. For

1

In the framework of a research project by Kunal Bhattacharya, Semiconductor Physics group, TU Chemnitz, Germany.

72

Figure 4.9. The dependence of the cantilever spring constant on geometrical parameters of the cantilever for a single segment model.

73

Figure 4.10. a) FEM model of a realistic gold cantilever mounted on a silicon chip; b) comparison of experimental and simulated frequency response of the gold cantilever. Numbers indicate experimental (simulated) frequencies of the respective modes; c) illustration of application of the point force on the tip apex, and the measured displacement for determination of the spring constant.

determination of the spring constant, the point force was applied to the tip apex as shown in Figure 4.10c. The spring constant for such a tip was determined to be around (51 ± 12) N/m.1

4.5. Summary The overview of existing TERS systems, tips and the figures of merit used in TERS experiments such as contrast and enhancement factor is given. A side illumination-collection system TERS based on AFM was chosen as the most universal solution, and developed by coupling commercially available AFM and Raman spectrometer. The AFM feedback loop and requirements on the TERS performance of the tip lead to fabrication of all-metal tips with a reflective cantilever and an apex sharpened by electrochemical etching. Depending on the AFM imaging mode, the tips must have particular spring constant. The spring constant of fabricated all-metal TERS tips was found to be in the range of tens of N/m as a result of simulations using FEM. Optimal geometrical parameters of the tips were found both for contact and intermittent contact AFM imaging modes.

The error was estimated by calculating the spring constant of the tip with (𝐿 + 10% and 𝐻1 −10%), and (𝐿 − 10% and 𝐻1 + 10%), since these parameters were shown to have the highest effect 1

on the spring constant. The deviation of 10% is the error of the size measurements with SEM.

74

Chapter 5.

Tip-enhanced Raman spectroscopy imaging

TERS and its advantage of high spatial resolution was applied for biological studies [19, 200-202], as well as in imaging organic films such as dyes [203, 204] and organic semiconductors [205], localization of defects and strain in carbon nanotubes [206-209]. This technique was also employed to analyze low-dimensional crystalline materials such as nanowires [210], extracting information about phonon confinement along a single nanowire [211], and strain and material distribution in a single quantum dot [212]. This approach allows a spatial resolution in the range 10 – 50 nm [213]; and even submolecular resolution [20], and 1.7 nm resolution for carbon nanotubes [18] was achieved. Several groups have shown that the use of metallic substrate allows to increase the TERS enhancement factor due to image dipole effect (gap-mode TERS) [214, 215], and can be further increased by employing substrates that exhibit a high roughness or curved features (double-tip TERS) [215-217]. The higher electric field confinement in these techniques may ultimately lead to improved sensitivity and spatial resolution [158]. In this chapter, first the application of TERS imaging for the chemical identification of materials on a multi-component sample is demonstrated. Then TERS imaging is employed for understanding the dependence of the enhancement on the substrate material and morphology and on AFM imaging mode. TERS is employed to both organic and inorganic materials.

5.1. Materials and methods 5.1.1. Preparation of multi-component sample A piece of gold coated (50 nm) glass slide (Ted Pella) was functionalized with a monolayer of amino-terminated polyamidoamine (PAMAM) dendrimer (4th generation, Sigma Aldrich) in order to create a charged surface for effective adhesion of GO and CNTs. Next 100 µL of GO suspension was placed on top of the functionalized gold slide for two minutes, and then spun away in a spin coater. Afterwards, 100 µL of CNT suspension (70% metallic, Nanointegris) was placed on top of the sample, held there for 2 minutes, spun away in a spin coater. After that, 100 µL of fresh Piranha solution was placed on the sample for 5 seconds in order to remove the surfactant and the PAMAM dendrimer. Then, the sample was rinsed in

75

copious amount of distilled water and dried at room temperature. As the final step, 50 µL of 10 to 5 M of C60 in toluene was spin-coated on the sample.

5.1.2. TERS experiments TERS measurements were performed using one of the two devices: the customized TERS setup described in detail in Chapter 4, or a commercially available TERS setup coupling an XploRA Raman spectrometer (Horiba JY) and an OmegaScope-R AFM (AIST-NT, Inc.)1. In case of the former system, a gold AFM tip developed in our group was illuminated with a HeNe 632.8 nm laser line using 1 mW laser power. Considering the high spring constants of the custom made cantilevers, for this system the TERS imaging was performed in intermittent contact mode. The latter system (Figure 5.1) is a side illumination/collection setup with a 100x objective (NA 0.7), adjusted by a closed loop piezoelectric scanner in three axes. The objective scanning allows for reproducible alignment of the laser focus on the tip. For excitation, a 637.3 nm laser line of a solid state laser was used. The AFM specially developed for TERS has a thermally compensated design to minimize thermal drift of the tip and/or the sample. The sensitivity of the spectrometer is further improved by using a binning factor of two that allows acquiring spectra at 0.1 s per spectrum and less with incident laser power of 160 μW. Even though binning improves the sensitivity, it also compromises the spectral resolution making it ca. 7 cm-1. TERS measurements were performed both in intermittent contact and contact mode. While intermittent contact mode is preferential to preserve the sample, it provides considerable TERS signal intensities only during a fraction of the tip oscillation cycle, thus when the tip is close the sample surface. In this setup, the contact mode is optimized for TERS imaging providing simultaneously gentle AFM imaging and high TERS enhancement.

5.1.3. Simulations of electric field enhancement The models for simulating the electric field enhancement in TERS include a single metallic tip, which makes the problem non-periodic, and therefore such time efficient methods as the Fourier modal method are not applicable to such geometry. Instead, the system was simulated using FEM with ANSYS 14.0 Multiphysics2. A half/quarter of the desired three-

1 Thanks to Dr. Andrey Krayev (AIST-NT, Inc., Novato, CA, USA) for providing access to their facilities in Novato, California, and the initial training. 2 In close collaboration with Dr. V. Kolchuzhin, Microsystems and Biomedical Engineering, TU Chemnitz, Germany.

76

Figure 5.1. XploRA – Omegascope-R TERS system, the main components indicated. Image courtesy Andrey Krayev, AIST-NT, Inc.

dimensional model was created using Solid Edge ST6 (Siemens Industry Software GmbH & Co. KG) or directly in ANSYS with the dimensions scaled from nanometers to millimeters, since the ANSYS element type HF119 used for high-frequency analysis supports analysis only in MKS (meter, kilogram, and/or second) system of units. An additional cylindrical volume is added around the tip apex for finer meshing. Then the IGES-model is exported to ANSA 13.1.0 (CAE pre-processing software) (BETA CAE Systems S.A.) for meshing. This software allows much better control of the mesh with several mesh elements at the smallest feature (tip-sample gap), and at the same time small total number of nodes (approximately 80 000) allowing for short calculation times of 2.5 min per model1 (Figure 5.2). Finally, the meshed model was exported to ANSYS as a CDB-file. The full 3D model was reconstructed from the half /quarter by mirroring it with respect to the X and Y axes. It makes it possible to guarantee the presence of the mesh nods in the middle plane of the model, and thus to compare different models. The mesh elements are assigned to the HF119 element type, which is a tetrahedral element for high frequency analysis of EM fields. The dielectric functions of the materials [218] were approximated using a rational fit (real and imaginary parts independently) with a very high degree (up to 15) in both the numerator and denominator with a relative error < 1%. Afterwards, the model is rescaled to nanometers, and the problem is solved by SPARSE MATRIX DIRECT SOLVER. The incident light (load) is approximated by a plane EM wave with electric field of 1 V/m is applied by specifying the direction of propagation, polarization and frequency. The These results were computed on a desktop PC with Intel® Core™ i7-3930K CPU @ 3.20GHz and 64.0 GB RAM running Microsoft® Windows® 7 Enterprise. 1

77

Figure 5.2. Example of a mesh for a quarter of a three-dimensional model with a gold tip above half of a gold ellipsoid and gold film. Additional volume introduced for finer meshing is placed at the tip apex.

solution produces values of the electric (𝐸𝑥 , 𝐸𝑦 , 𝐸𝑧 ) and magnetic field (𝐵𝑥 , 𝐵𝑦 , 𝐵𝑧 ) components at each node. Since the incident strength of the electric field is unity, the values obtained from the solution can be directly treated as the enhancement of the electric field.

5.2.

High

resolution

discrimination

of

carbon-containing

compounds by TERS Here a multi-component sample is used to demonstrate the potential of TERS for chemical identification and analysis with nanoscale resolution1. The sample contains graphene oxide (GO), CNTs and fullerene (C60).

The sample preparation procedure is described in

Section 5.1.1. GO is a graphene sheet with attached –OH and –COOH groups, largely maintaining its 2D structure [219]. Therefore, it shows a so-called G band at around 1610 cm-1 arising from C=C 𝐸2𝑔 vibrations in the six-carbon ring [220]. Such high frequency position as compared to graphene, which has the G band at 1588 cm-1, is related to disorder, and was attributed in the literature to doping effects, and to disorder effects due to presence of small isolated double1

In close collaboration with Dr. A. Agapov and Prof. A. Sokolov (Oak Ridge National Laboratory, TN, USA).

78

bond segments [221]. The broken translational symmetry of GO also allows for efficient excitation of the in-plane G mode by the TERS tip, which is not the case for graphene [222]. High concentration of defects in GO results in the activation of the D band at 1350 cm-1 [220], which intensity was chosen for generating the TERS image (green in Figure 5.3). The 2D band (G’) mode at 2700 cm-1 (a second order mode of the D band) is normally very intense in graphene, but inhibited in GO when the electron-phonon coupling is disturbed due to high defect concentration [223]. CNTs are probably the material most extensively studied by TERS [224-227]: local distribution of strain [209] and defects [228, 229], change of the CNT chirality along a single CNT [206, 227], and even defect nature [230] were investigated. Carbon nanotubes (CNTs) are the closest to an ideal test system for TERS imaging. They are commercially available and easily dispersed on a suitable substrate. The TERS spectra of CNTs (Figure 5.3a) exhibit the G band around 1610 cm-1, which is high as compared to the G+ peak position of the reference CNT dispersion on Si without treatment observed at 1588 cm-1. It is attributed to doping [231, 232] due to piranha treatment before the addition of fullerene. Single wall CNTs can be seen as roll ups of single layer graphene sheets, leading to the splitting of the G band into G+ and G– bands for the C=C vibrations along the CNT axis and along the circumferential direction, respectively [220]. Due to the bundling of the CNTs, the presence of the 2D band is a more reliable indicator of the CNT than the G +/G– splitting. Therefore, the 2D band intensity was used for the TERS image of the CNTs (red in Figure 5.3b).

Figure 5.3. The representative spectra (a) and TERS image (b) of the three components. At each pixel of the TERS image full Raman spectrum is acquired. The TERS image is composed from the intensity distributions of the D, C5 and G bands marked in (a) over the image area. Green color follows the intensity of the D band, while red color indicates the intensity of 2D (G’) band. Blue color corresponds to the pentagon ring vibration of C60 (marked with circles in the map). For comparison, the dashed contour indicates the size of the diffraction limited laser spot. The 2D band intensity profile shown with white line is taken along the white arrow.

79

Thanks to the defined geometry of CNTs, they allow for straightforward determination of the spatial resolution. As visible from the 2D band intensity profile across the CNT bundle indicated by the white arrow (Figure 5.3b), the high spatial resolution in this case is limited by the step size of 15 nm. Fullerene is a buckyball formed of five- and six-ring units of sp2 and sp3 hybridized carbons, which was also the molecule used in one of the pioneering TERS experiments [17]. C60 can be identified by the Raman band at 1470 cm-1, arising from the five-ring vibrations [233], and is marked in Figure 5.3a as C5. Its G band is observed at 1585 cm-1. Apart from that, an organic residue is detected all over the substrate. Its Raman spectra reliably show a wide band between 2800 and 3000 cm-1, attributed to –C–H bond vibrations. It allows discrimination of the residue from the other three materials (Figure 5.4). At the same time, the spectra in the fingerprint region (below 2000 cm-1) are varying dramatically in terms of detected bands and their ratios. This behavior is similar to the behavior of carbonaceous materials on SERS substrates [234, 235] attributed to oxidation and the diffusion of molecules in and out of hot spots. Since the signal of the CNTs, C60 and GO dominate the spectra in the respective positions, spectra with negligible organic residue signal could be identified and analyzed in more

Figure 5.4. Selected spectra of organic residue on the substrate.

detail. The resolution achieved by TERS measurements allows analysis of localized changes in defect concentration and the corresponding peak widths and positions, which are not available using conventional micro-Raman spectroscopy. Since the presence of defects is necessary to

80

activate the D band, this information can be retrieved from the D to G intensity ratio (Figure 5.5). While by itself GO is characterized by D to G ratio above one, we observe slightly higher defect concentration at the edges of GO. It is additionally illustrated by a profile (Figure 5.5b), as well as representative spectra along it (Figure 5.5c). Although the difference is very small, it remains detectable when the average spectra over a region are taken. Note higher D to G ratio on the flake edges was not observed for other GO flakes analyzed, and therefore it should not be generalized.

Figure 5.5. (a) and (d) D to G ratio (identical images). The markings indicate the positions relevant for the analysis of the GO (a) and CNT (d) TERS data. (b) D to G ratio along the line marked in (a). (c) Average spectra along the profile. Ellipses in (d) indicate the CNT regions with high defect concentrations. (e) Selected TERS spectra along lines in (d).

Raman spectra of CNTs carry information about the defect concentration in CNTs, as well as bundling and CNT diameter. As an example, the white ellipses in Figure 5.5d circle the high defect concentration regions on the CNT. Selected spectra along the CNTs are presented in Figure 5.5e. The spectra along line 1 show clearly resolved G+ and G– features. The spectra 81

along line 2 show a wide G band, where G+ and G– features are not resolved. It suggests that that region comprises a bundle of the CNTs of different diameters. Finally, the spectra along line 3 show a wide G band as the one discussed above, but closer to the end of the CNT the G band appears to have a smaller width and the G band splitting can be resolved. These results imply that the TERS signal at the end of the bundle originates from fewer CNTs.

5.3. Effect of substrate material and morphology on TERS enhancement It is well established that the use of gold or silver substrates in TERS experiments facilitates the generation of highly confined electric fields under the tip and the sample, which leads to

high enhancement of the Raman signal [55-57] as discussed in Section 1.2.2.

Although, one has to take into account the shift of the LSPR upon approaching the tip to the metallic substrate [236, 237]. Therefore, many of the TERS experiments are performed on thin material layers deposited on gold or silver substrates [207, 238, 239]. In bottom illumination the benefits of gap-mode TERS were achieved using transparent ultra-thin gold platelets [240, 241]. It was reported that not only substrate material, but also substrate morphology have effect on the signal enhancement. Gold nanosteps [217] and individual gold nanoparticles [216] were demonstrated to lead to a significant increase in the observed enhancement at the same material concentration. Thus in order to draw reliable conclusion from the TERS images, the understanding of the effect of substrate material and morphology is crucial. For that purpose different substrate-tip configurations are compared in a single experiment using nanostructures gold films. A nanostructured gold film was prepared on Si substrate using the EBL protocol described in Section 3.1.1, except that some positions were over-exposed to the electron beam, leading to poorer solubility of the respective polymer. These positions formed surface protrusions when later covered with gold. The prepared substrate was covered with 2 nm CoPc film by OMBD. The schematic of the sample cross-section is shown in Figure 5.6a. TERS scanning was performed using an electrochemically etched gold tip and 632.8 nm laser line with 1 mW power in intermittent contact mode AFM. The AFM image (Figure 5.6b) shows 40 nm high gold film corresponding to the nominal thickness, with two 40 nm high gold protrusions filled with over-exposed polymer. The

82

Figure 5.6. (a) Schematic of the sample cross-section. (b) AFM and (c) respective TERS image. (d) Averaged CoPc spectra along the middle line of the TERS image (marked with rectangle in c) from Si to gold.

83

corresponding TERS map shows the intensity of the CoPc mode at 682 cm-1 (Figure 5.6c). The respective spectra are presented in Figure 5.6d. Taking the intensity of the CoPc mode at 682 cm-1 on Si as 100%, the signal intensity on the gold film is 125% and 155% on the gold protrusions. Higher frequency modes above 700 cm-1 also show comparable intensity increase above gold film and gold protrusions, but their lower total intensity makes this change less obvious. Note that there is an additional mode observed in the TERS spectra as compared to the micro-Raman and SERS spectra presented in Section 3.2.1. Its origin is not clear till now. In order to understand the behavior of the TERS signal, numerical simulations of the electric field enhancement in all three configurations were performed (Figure 5.7). In the models shown below the tip radius is 30 nm and tip-sample gap is 2 nm. The simulations also showed that the presence of polymer has a very weak effect on the resulting enhancement, and therefore it can be neglected. The incident electric field of unity is polarized along the tip, and the wavelength is 633 nm. Based on the analysis of multiple models, the electric field was found to increase significantly for a tip above the gold film as compared to silicon for various tip radii, this is well established in the literature and is explained by the image dipole effect. In the case of a tip above a protrusion, the resulting electric field enhancement is very sensitive to the geometry of both the tip and the protrusion. This suggests that the enhancement arises from the tipsubstrate coupling. Note that the contrast between different tip-substrate configurations is much lower in the experiment than in simulations since the imaging was performed in intermittent contact mode.

Figure 5.7. Simulated enhancement of electric field for a gold tip of 30 nm radius above (a) Si substrate, (b) 40 nm thick gold film on Si, (c) gold protrusion of 40 nm height and 200 nm diameter on gold film and Si substrate. Incident field is polarized along the tip, wavelength 633 nm.

84

5.4. Effect of the AFM imaging mode on TERS enhancement As it was discussed in Section 4.4.3, AFM imaging is most often performed in contact and intermittent contact mode. Contact mode provides stronger TERS enhancement due to the direct contact between the tip and the sample. At the same time, intermittent contact mode is often preferred for imaging of soft samples. Here the effect of the two different imaging modes on the TERS enhancement is investigated on an array of gold nanoclusters with 500 nm period covered with 2 nm of CoPc. The TERS experiments were performed using an electrochemically etched gold tip and a 638 nm laser line. As expected from the previous experiment, a 12-fold higher TERS signal intensity is observed above the gold nanoclusters as compared to the Si substrate (Figure 5.8a). In the intermittent contact mode the signal intensities are lower as compared to the contact mode, since the tip is in contact with the surface only a fraction of the time. The intermittent contact mode produces 7fold contrast between the signal on gold nanoclusters and on silicon (Figure 5.8b). The higher NA objective (0.7) of the AIST-NT-XploRA setup and a cantilever oscillation amplitude of only 15 nm produced better contrast than for results discussed in Section 5.3. Another useful parameter is the ratio of the intensities in contact vs. intermittent contact modes 𝑅 . For the mode at 682 cm-1, the ratio for the spectra acquired on Si is 𝑅𝑆𝑖 = 19, and on

Figure 5.8. TERS spectra on gold nanoclusters and silicon covered with 2 nm CoPc measured in contact (a) and intermittent contact (b) modes.

85

gold nanoclusters 𝑅𝐴𝑢 = 29. First of all, it highlights the importance of using contact mode for obtaining a high quality TERS signal. Secondly, the ratio is systematically higher above gold than above silicon 𝑅𝐴𝑢 > 𝑅𝑆𝑖 when analyzing different CoPc modes. Assuming that the tip oscillation amplitude remains constant above gold and above Si, it suggests that the decay length differs depending on the substrate material. In order to clarify this result, calculations based on the image dipole model1 were performed (see Section 1.2.2) for a gold tip approximated as a polarizable sphere with radius 𝑟 at a distance 𝑑 above a semi-infinite silicon or gold substrate, where a single dipole is excited at a distance (𝑟 + 𝑑) from the surface (Figure 5.9a). Figure 5.9b shows the calculated dependence of the field intensity on the surface on the distance between the tip apex and the surface for a 20 nm tip apex radius. It shows that indeed the field above a gold surface decays faster than above a silicon surface. Interestingly, theory predicts that surface roughness such as a 20 nm gold semi-sphere on the gold substrate (gold protrusion in Figure 5.9b), would also reduce the decay length. AFM imaging mode also affects the enhancement pattern. Figure 5.10 shows TERS images in contact and intermittent contact modes acquired with the same tip on a region

Figure 5.9. Image dipole model of a tip above a semi-infinite surface. (a) Parameters of the model. (b) Dependence of the electric field at the surface on the distance between the tip and the sample for a tip of 20 nm radius. 1

Thanks to Prof. E. Bortchagovsky (Institute of High Technologies T.G.Shevchenko Kiev National University, Kiev, Ukraine) for valuable discussions and calculations.

86

containing four gold nanoclusters. The corresponding SEM image is shown in Figure 5.10a. Raman signal is substantially enhanced above the gold nanoclusters as compared to the silicon substrate both in contact and intermittent contact modes due to the gap-mode effect (Figure 5.10b, c). However, the images clearly show different enhancement patterns. To understand this effect, both numerical simulations and complementary AFM images are needed. In the contact mode TERS, maximum Raman signal intensity is observed above the center of the gold nanocluster, as shown in the Raman intensity profile of the mode at 682 cm -1 overlaid with the AFM profile (Figure 5.11b). To adequately simulate the field enhancement, the experiment geometry has to be taken into account. It is schematically shown in (Figure 5.11a). The tilted tip with the tip apex radius of 65 nm as measured by SEM was positioned over a 40 nm high gold nanocluster with rounded edges, nanocluster diameter 200 nm. In order to mimic incidence angle of the laser beam (Figure 5.10a), each model was simulated using incident field projections normal and parallel to the Si surface, and then the full field enhancement was calculated from the simulation results. The enhancement for tip above left and right edges of the gold nanocluster (positions 1 and 3) appears to be comparable, while the enhancement over the center of the nanocluster (position 2) is much larger. This observation is in good agreement with the experimental results, where the maximum Raman signal intensity is also observed for the tip positioned above the gold nanocluster center. The experimental TERS profile also appears to be asymmetric, with a weak tail extending beyond the AFM profile on the left-hand side. This behavior was attributed to the tip tilt, which results in multiple contact points between the tip and the nanocluster on the one side (like the one in position 1), generating hot spots that enhance the Raman signal.

Figure 5.10. (a) SEM image of gold nanoclusters with 500 nm period on Si. TERS images on four gold nanoclusters in contact (b) and intermittent contact (c) modes. Color scale shows Raman intensity in kcts.

87

In intermittent contact the enhancement profile appears to have strong asymmetry and narrow area above the center of the gold nanoclusters with high enhancement (Figure 5.12a). To better understand the pattern, high resolution TERS imaging on a single gold nanocluster was performed (Figure 5.12b-d). For a single nanocluster, the TERS enhancement appears to closely follow the phase image. This result was reproduced for several nanocluster arrays studied. The phase contrast in the intermittent contact mode originates from the phase lag between the driving signal and the cantilever oscillation response, which is determined by the interaction forces between the tip and the sample. The phase contrast is thus often used to discriminate areas with different surface properties, or different materials. To exclude that in this TERS experiment the phase contrast originates from material distribution, the imaging of gold

Figure 5.11. TERS in contact mode. (a) Sketch of the experimental layout. (b) AFM profile of the gold nanocluster (right axis) and the corresponding TERS profile at 682 cm-1 (left axis). (c) Simulated enhancement of the electric field between the tip and the sample for tip positions 1, 2 and 3 with respect to the gold nanocluster, which correspond to the positions 1, 2 and 3 indicated in profile in (b). EF indicates maximum electric field enhancement for the respective position taking into account the illumination angle.

88

Figure 5.12. (a) Height (right axis) and TERS signal (left axis) intensity profiles over four gold nanoclusters. (b-d) TERS imaging of a single gold nanocluster using a gold tip: (b) AFM topography, (c) phase image, (d) TERS intensity of 682 cm-1 mode.

Figure 5.13. AFM topography (a) and phase (b) image of a gold nanocluster acquired with a commercial Si cantilever.

nanoclusters used in the TERS experiment with a commercial tip was performed (Figure 5.13). From comparison of the phase images obtained with gold and commercial tips it can be concluded that there is different interaction between the gold tip and the gold nanocluster at the edges of the gold nanoclusters (Figure 5.12). The detailed explanation of this observation requires further investigations. 89

5.5. TERS on free-standing colloidal CdSe NCs Analogous to SERS, TERS on inorganic materials such as semiconducting NCs can be obtained. As a proof of principle, the TERS experiment on CdSe NC monolayer deposited on aSERS substrate prepared by EBL was performed. The comparison of tip down (near-field) and tip up (far field) spectra is shown in Figure 5.14. It was possible to achieve a 100% increase of the CdSe LO mode intensity at 207 cm-1. The increased background at low frequency can be attributed to enhancement of signal from amorphous Se on the QD surface, and partially to elastic scattering from the tip. The latter could lead to increase of the Raman signal not related to the near-field enhancement [156, 157]. However, since the intensity of the Si peak remains the same, such signal increase can be excluded [157], and the TERS EF can be estimated using Eq. 4.2. Assuming that the CdSe film is thin enough to neglect the drop in field enhancement within the film, we normalize the Raman intensities by area instead of volume. Notice that this assumption leads to an underestimation of enhancement factor, in order to have a conservative evaluation. The area of the illuminated volume is defined by the laser spot size, which is 1.5×2.5 µm2. Then the enhanced area diameter was estimated as 𝑑 = √𝑅𝑡𝑖𝑝 𝑔 [159], where

Rtip is the tip radius (typically below 50 nm among the tips prepared in our lab as confirmed by SEM), and g – the gap between the tip and the substrate defined by the QD size. The TERS EF

Figure 5.14. TERS on CdSe monolayer on SERS substrate. (a) Schematic of the experiment. (b) TERS spectra.

90

was estimated to be 6×104 with accuracy of 14%. Solving the problem of packing smaller circles in a larger circle, we can then expect seven CdSe QDs in the enhanced area.

5.6. Summary TERS by organic and inorganic materials, including carbon-based materials, organic thin films and semiconducting NCs, was realized. The superior spatial resolution was demonstrated for carbon-based materials in gapmode TERS, and the effect of substrate morphology on the TERS enhancement was investigated. The numerical simulations suggest that in this case the enhancement strongly depends on the exact morphology of both the tip and the surface, what was also observed experimentally. This indicates that the enhancement is a result of plasmon coupling between the tip and the substrate. In terms of signal intensity, contact mode was shown to be beneficial for TERS imaging due to the smaller distance between the tip and the sample as compared to intermittent contact mode. Both imaging modes allow visualization of nanometer-scale objects, the spatial distribution of the TERS signal intensity differs. The TERS images obtained in contact mode are consistent with the numerical simulation results. In intermittent mode TERS images are affected by mechanical interaction between the tip and the sample as indicated by a strong correlation between the TERS and AFM phase images. Finally, as a proof of principle, TERS experiments were performed on a CdSe NC monolayer deposited on a SERS substrate. Enhancement of Raman scattering by optical phonons by as few as seven NCs in gap-mode TERS of about 104 was realized.

91

92

Conclusions The PhD work focused on the application of micro- and nano-Raman spectroscopy to the analysis of organic and inorganic materials with improved sensitivity and spatial resolution. Depending on the material nature and structure design, different approaches were realized. In case of well-defined InAs/Al(Ga)As NC multilayers resonant Raman spectroscopy allows size-selective Raman scattering of NCs. The multilayer structure with a layer period below the diffraction limit was also successfully characterized by micro-Raman spectroscopy, and confirmed by AFM and TEM measurements. In case of ultra-thin organic CoPc films and CuxSx and CdSe NCs deposited on arrays of gold nanoclusters, detection of very small amounts of material was possible using surfaceenhanced resonant Raman scattering (SERRS). The enhancement of Raman scattering by vibrational modes in CoPc films and optical phonons in CdSe NCs on the order of 103 was achieved. The electromagnetic enhancement mechanism was shown to dominate the SERRS response of CoPc films and CdSe NCs. The dependence of SERRS enhancement on excitation wavelength was measured, and its maximum agrees very well with the measured LSPR energies of gold nanoclusters. Based on the benefits on SERS, TERS provides an additional gain in the study of nanoobjects by improving the spatial resolution below the diffraction limit. TERS was implemented in side illumination-collection configuration by in-house coupling of a commercial AFM and a Raman spectrometer, and using full-metal TERS tips developed in our lab. Using FEM simulations of the full-metal TERS tips, they were found to be suitable for intermittent contact AFM imaging due to high spring constant. Optimal geometrical parameters of the tips with spring constant below 1 N/m suitable for contact mode imaging were determined. Identification and analysis of carbon-containing compounds was achieved using gapmode TERS with spatial resolution below 15 nm. Further, TERS imaging on CoPc films on nanostructured gold film and on a gold nanocluster array was performed. The TERS enhancement was experimentally demonstrated to depend on the morphology of the substrate, and the complementary FEM simulations show that the electric field enhancement is defined by the tip-sample coupling. The TERS signal intensity was shown to be higher by imaging in contact mode AFM as compared to the intermittent contact mode due to smaller average distance between the tip and the sample. The decay of the Raman signal with the tip-sample distance occurs to be faster above gold than above silicon in accordance with the image dipole model. The approaches developed for TERS of organic materials were successfully applied to

93

obtaining TERS signal from ultra-low quantities of CdSe NCs with enhancement factor on the order of 104.

94

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27.

M.K. Debe. Progress in Surface Science, 1987. 24(1-4), 1-282. A. Fainstein, B. Jusserand, and V. Thierrymieg. Physical Review Letters, 1995. 75(20), 3764-3767. A. Fainstein and B. Jusserand. Semiconductor Science and Technology, 2003. 18(10), S377-S385. G.A.N. Connell, R.J. Nemanich, and C.C. Tsai. Applied Physics Letters, 1980. 36(1), 3133. W.S. Bacsa, E. Pavlenko, and V. Tishkova. Nanomaterials and Nanotechnology, 2013. 3, 1-7. M. Fleischmann, P.J. Hendra, and A.J. McQuilla. Chemical Physics Letters, 1974. 26(2), 163-166. D.L. Jeanmaire and R.P. Vanduyne. Journal of Electroanalytical Chemistry, 1977. 84(1), 1-20. S.M. Nie and S.R. Emery. Science, 1997. 275(5303), 1102-1106. K. Kneipp, Y. Wang, H. Kneipp, L.T. Perelman, I. Itzkan, R. Dasari, and M.S. Feld. Physical Review Letters, 1997. 78(9), 1667-1670. H.X. Xu, E.J. Bjerneld, M. Kall, and L. Borjesson. Physical Review Letters, 1999. 83(21), 4357-4360. A. Weiss and G. Haran. Journal of Physical Chemistry B, 2001. 105(49), 12348-12354. J.A. Dieringer, R.B. Lettan II, K.A. Scheidt, and R.P. Van Duyne. Journal of the American Chemical Society, 2007. 129(51), 16249-16256. L. Novotny, The history of near-field optics, in Progress in Optics, Vol 50. 2007. p. 137184. J. Wessel. Journal of the Optical Society of America B-Optical Physics, 1985. 2(9), 1538-1541. M.S. Anderson. Applied Physics Letters, 2000. 76(21), 3130-3132. N. Hayazawa, Y. Inouye, Z. Sekkat, and S. Kawata. Optics Communications, 2000. 183(1-4), 333-336. R.M. Stockle, Y.D. Suh, V. Deckert, and R. Zenobi. Chemical Physics Letters, 2000. 318(1-3), 131-136. C. Chen, N. Hayazawa, and S. Kawata. Nature Communications, 2014. 5(3312), 1-5. E. Bailo and V. Deckert. Angewandte Chemie-International Edition, 2008. 47(9), 16581661. R. Zhang, Y. Zhang, Z.C. Dong, S. Jiang, C. Zhang, L.G. Chen, L. Zhang, Y. Liao, J. Aizpurua, Y. Luo, J.L. Yang, and J.G. Hou. Nature, 2013. 498(7452), 82-86. D.A. Long, The Raman Effect: A Unified Treatment of the Theory of Raman Scattering by Molecules. Wiley, 2002. P. Yu and M. Cardona, Fundamentals of Semiconductors: Physics and Materials Properties. Springer, 2005. W. Kiefer. Journal of Raman Spectroscopy, 2007. 38(12), 1538-1553. X. Zhang, W.J. Kirkwood, P.M. Walz, E.T. Peltzer, and P.G. Brewer. Applied Spectroscopy, 2012. 66(3), 237-249. Z. Movasaghi, S. Rehman, and I.U. Rehman. Applied Spectroscopy Reviews, 2007. 42(5), 493-541. J. Chan, S. Fore, S. Wachsman-Hogiu, and T. Huser. Laser & Photonics Reviews, 2008. 2(5), 325-349. D. Drescher and J. Kneipp. Chemical Society Reviews, 2012. 41(17), 5780-5799.

95

28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39.

40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53.

54. 55. 56. 57.

H.G.M. Edwards and T. Munshi. Analytical and Bioanalytical Chemistry, 2005. 382(6), 1398-1406. J.H. Cheng, Q. Dai, D.W. Sun, X.A. Zeng, D. Liu, and H.B. Pu. Trends in Food Science & Technology, 2013. 34(1), 18-31. D.I. Ellis, V.L. Brewster, W.B. Dunn, J.W. Allwood, A.P. Golovanov, and R. Goodacre. Chemical Society Reviews, 2012. 41(17), 5706-5727. A.M. Herrero. Food Chemistry, 2008. 107(4), 1642-1651. S. Tsukahara. Analytica Chimica Acta, 2006. 556(1), 16-25. M.S. Dresselhaus, G. Dresselhaus, R. Saito, and A. Jorio. Physics Reports-Review Section of Physics Letters, 2005. 409(2), 47-99. A.K. Arora, M. Rajalakshmi, T.R. Ravindran, and V. Sivasubramanian. Journal of Raman Spectroscopy, 2007. 38(6), 604-617. F. Rauh and B. Mizaikoff. Analytical and Bioanalytical Chemistry, 2012. 402(1), 163-173. R.S. Das and Y.K. Agrawal. Vibrational Spectroscopy, 2011. 57(2), 163-176. E.V. Efremov, F. Ariese, and C. Gooijer. Analytica Chimica Acta, 2008. 606(2), 119-134. E. Le Ru and P. Etchegoin, Principles of Surface-Enhanced Raman Spectroscopy: and related plasmonic effects. Elsevier, 2008. M.D. Sonntag, J.M. Klingsporn, L.K. Garibay, J.M. Roberts, J.A. Dieringer, T. Seideman, K.A. Scheidt, L. Jensen, G.C. Schatz, and R.P. Van Duyne. Journal of Physical Chemistry C, 2012. 116(1), 478-483. W. Zhang, B.S. Yeo, T. Schmid, and R. Zenobi. Journal of Physical Chemistry C, 2007. 111(4), 1733-1738. K.A. Willets and R.P. Van Duyne, Localized surface plasmon resonance spectroscopy and sensing, in Annual Review of Physical Chemistry. 2007. p. 267-297. J.V. Bladel, Singular Electromagnetic Fields and Sources. Wiley, 1996. S. Thomas, M. Krueger, M. Foerster, M. Schenk, and P. Hommelhoff. Nano Letters, 2013. 13(10), 4790-4794. J. Gersten and A. Nitzan. Journal of Chemical Physics, 1981. 75(3), 1139-1152. A.V. Ermushev, B.V. McHedlishvili, V.A. Oleinikov, and A.V. Petukhov. Kvantovaya Elektronika, 1993. 20(5), 503-508. P.F. Liao and A. Wokaun. Journal of Chemical Physics, 1982. 76(1), 751-752. A.B. Zrimsek, A.-I. Henry, and R.P. Van Duyne. Journal of Physical Chemistry Letters, 2013. 4(19), 3206-3210. F. H'Dhili, R. Bachelot, G. Lerondel, D. Barchiesi, and P. Royer. Applied Physics Letters, 2001. 79(24), 4019-4021. J. Jersch, F. Demming, L.J. Hildenhagen, and K. Dickmann. Applied Physics a-Materials Science & Processing, 1998. 66(1), 29-34. D. Haefliger, J.M. Plitzko, and R. Hillenbrand. Applied Physics Letters, 2004. 85(19), 4466-4468. G. Mie. Annalen der Physik, 1908. 330(3), 377-445. M. Born and E. Wolf, Principles of Optics. Cambridge University Press, 2011. V. Myroshnychenko, J. Rodriguez-Fernandez, I. Pastoriza-Santos, A.M. Funston, C. Novo, P. Mulvaney, L.M. Liz-Marzan, and F.J. Garcia de Abajo. Chemical Society Reviews, 2008. 37(9), 1792-1805. A. Hille, R. Kullock, S. Grafstroem, and L.M. Eng. Journal of Computational and Theoretical Nanoscience, 2010. 7(8), 1581-1586. J. Stadler, B. Oswald, T. Schmid, and R. Zenobi. Journal of Raman Spectroscopy, 2013. 44(2), 227-233. R.T. Hill, J.J. Mock, Y. Urzhumov, D.S. Sebba, S.J. Oldenburg, S.-Y. Chen, A.A. Lazarides, A. Chilkoti, and D.R. Smith. Nano Letters, 2010. 10(10), 4150-4154. N. Behr and M.B. Raschke. Journal of Physical Chemistry C, 2008. 112(10), 3766-3773. 96

58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86.

W.E. Doering and S.M. Nie. Journal of Physical Chemistry B, 2002. 106(2), 311-317. M. Moskovits. Physical Chemistry Chemical Physics, 2013. 15(15), 5301-5311. A.M. Michaels, M. Nirmal, and L.E. Brus. Journal of the American Chemical Society, 1999. 121(43), 9932-9939. A.M. Michaels, J. Jiang, and L. Brus. Journal of Physical Chemistry B, 2000. 104(50), 11965-11971. S. Degioanni, A.M. Jurdyc, F. Bessueille, J. Coulm, B. Champagnon, and D. Vouagner. Journal of Applied Physics, 2013. 114, 234307. W. Krasser and A.J. Renouprez. Solid State Communications, 1982. 41(3), 231-235. E. Burstein, Y.J. Chen, C.Y. Chen, S. Lundquist, and E. Tosatti. Solid State Communications, 1979. 29(8), 567-570. R.M. Lazorenko-Manevich. Russian Journal of Electrochemistry, 2005. 41(8), 799-831. T.E. Furtak and S.H. Macomber. Chemical Physics Letters, 1983. 95(4-5), 328-332. A. Otto. Journal of Electron Spectroscopy and Related Phenomena, 1983. 29(JAN), 329342. Y. Xie, D.Y. Wu, G.K. Liu, Z.F. Huang, B. Ren, J.W. Yan, Z.L. Yang, and Z.Q. Tian. Journal of Electroanalytical Chemistry, 2003. 554, 417-425. J.F. Arenas, M.S. Woolley, I.L. Tocon, J.C. Otero, and J.I. Marcos. Journal of Chemical Physics, 2000. 112(17), 7669-7683. M. Cardona, G. Güntherodt, and G. Abstreiter, Light scattering in solids V: superlattices and other microstructures. Springer, 1989. M. Cardona and G. Abstreiter, Light scattering in solids V: superlattices and other microstructures. Springer, 1989. G. Bauer and W. Richter, Optical Characterization of Epitaxial Semiconductor Layers. Springer, 2011. J. Ibanez, R. Cusco, L. Artus, M. Henini, A. Patane, and L. Eaves. Applied Physics Letters, 2006. 88(14), 141905. A. Milekhin, A. Toropov, A. Kalagin, and D.R.T. Zahn. Chinese Journal of Physics, 2011. 49(1), 71-76. D.A. Tenne, A.K. Bakarov, A.I. Toropov, and D.R.T. Zahn. Physica E: Low-dimensional Systems and Nanostructures, 2002. 13(2-4), 199-202. J. Ibanez, A. Patane, M. Henini, L. Eaves, S. Hernandez, R. Cusco, L. Artus, Y.G. Musikhin, and P.N. Brounkov. Applied Physics Letters, 2003. 83(15), 3069-3071. G.-H. Tang, B. Xu, L.-W. Jiang, J.-X. Kong, N. Kong, D.-C. Liang, P. Liang, X.-L. Ye, P. Jin, F.-Q. Liu, Y.-H. Chen, and Z.-G. Wang. Chinese Physics Letters, 2010. 27(4). C. Santori, D. Fattal, J. Vuckovic, G.S. Solomon, and Y. Yamamoto. New Journal of Physics, 2004. 6. D. Guimard, R. Morihara, D. Bordel, K. Tanabe, Y. Wakayama, M. Nishioka, and Y. Arakawa. Applied Physics Letters, 2010. 96(20). S. Buckley, K. Rivoire, and J. Vuckovic. Reports on Progress in Physics, 2012. 75(12). T.S. Shamirzaev, A.V. Nenashev, A.K. Gutakovskii, A.K. Kalagin, K.S. Zhuravlev, M. Larsson, and P.O. Holtz. Physical Review B, 2008. 78, 085323. K. Pierz, Z. Ma, U.F. Keyser, and R.J. Haug. Journal of Crystal Growth, 2003. 249(3-4), 477-482. I. Hapke-Wurst, U. Zeitler, H.W. Schumacher, R.J. Haug, K. Pierz, and F.J. Ahlers. Semiconductor Science and Technology, 1999. 14(11), L41-L43. W.T. Tsang. Applied Physics Letters, 1978. 33(5), 426-429. M. Ochiai, G.E. Giudice, H. Temkin, J.W. Scott, and T.M. Cockerill. Applied Physics Letters, 1996. 68(14), 1898-1900. J. Lee, Z.M. Wang, V.G. Dorogan, Y.I. Mazur, and G.J. Salamo. Ieee Transactions on Nanotechnology, 2010. 9(2), 149-156. 97

87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97. 98. 99. 100.

101. 102. 103. 104. 105. 106. 107. 108. 109. 110. 111. 112. 113. 114.

Q. Gong, P. Offermans, R. Notzel, P.M. Koenraad, and J.H. Wolter. Applied Physics Letters, 2004. 85(23), 5697-5699. F. Ferdos, S.M. Wang, Y.Q. Wei, M. Sadeghi, Q.X. Zhao, and A. Larsson. Journal of Crystal Growth, 2003. 251(1-4), 145-149. R. Schorer, G. Abstreiter, S. De Gironcoli, E. Molinari, H. Kibbel, and H. Presting. Physical Review B, 1994. 49(8), 5406-5414. A.G. Milekhin, A.I. Toropov, A.K. Bakarov, S. Schulze, and D.R.T. Zahn. Jetp Letters, 2006. 83(11), 505-508. A.G. Milekhin, A.I. Nikiforov, O.P. Pchelyakov, S. Schulze, and D.R.T. Zahn. Nanotechnology, 2002. 13(1), 55-58. J. Ibanez, R. Cusco, L. Artus, M. Henini, A. Patane, and L. Eaves. Applied Physics Letters, 2006. 88, 141905. I. Hapke-Wurst, U. Zeitler, H.W. Schumacher, R.J. Haug, K. Pierz, and F.J. Ahlers. Semiconductor Science and Technology, 1999. 14(11), L41. J.K. Garleff, J.M. Ulloa, and P.M. Koenraad, Semiconductors Studied by Cross-sectional Scanning Tunneling Microscopy, in Scanning Probe Microscopy in Nanoscience and Nanotechnology 2, B. Bhushan, Editor. 2011, Springer. p. 321-353. B. Dwir, F. Reinhardt, and E. Kapon. Journal of Applied Physics, 1995. 78(8), 49394942. F. Reinhardt, B. Dwir, G. Biasiol, and E. Kapon. Applied Surface Science, 1996. 104– 105(0), 529-538. F. Reinhardt, B. Dwir, and E. Kapon. Applied Physics Letters, 1996. 68(22), 3168-3170. R.S. Gonnelli. Philosophical Magazine Part B, 2000. 80(4), 599-609. M.S. Dunaevskii, A.N. Titkov, Z.F. Krasilnik, A.V. Novikov, D.N. Lobanov, and R. Laiho. Applied Physics Letters, 2004. 85(11), 1999-2001. O.M. Amirmajdi, R. Ashyer-Soltani, M.P. Clode, S.H. Mannan, W. Yunqi, E. Cabruja, and G. Pellegrini. IEEE Transaction on Electronics Packaging Manufacturing, 2009. 32(4), 265-271. R. Wirth. Chemical Geology, 2009. 261(3–4), 217-229. A. Notargiacomo and L. Maiolo. Microelectronic Engineering, 2013. 110(0), 465-469. A.J. Robson, I. Grishin, R.J. Young, A.M. Sanchez, O.V. Kolosov, and M. Hayne. ACS Applied Materials & Interfaces, 2013. 5(8), 3241-3245. W.T. Tsang. Applied Physics Letters, 1978. 33(5), 426-429. D.A. Tenne, V.A. Haisler, A.I. Toropov, A.K. Bakarov, A.K. Gutakovsky, D.R.T. Zahn, and A.P. Shebanin. Physical Review B, 2000. 61(20), 13785-13790. S. Adachi. Physical Review B, 1987. 35(14), 7454-7463. W. Prettl, N.J. Shevchik, and M. Cardona. physica status solidi (b), 1973. 59(1), 241249. K.K. Tiong, P.M. Amirtharaj, F.H. Pollak, and D.E. Aspnes. Applied Physics Letters, 1984. 44(1), 122-124. S. Nakashima. Journal of Physics: Condensed Matter, 2004. 16(2), S25. A. Feofanov, S. Sharonov, P. Valisa, E. Da Silva, I. Nabiev, and M. Manfait. Review of Scientific Instruments, 1995. 66(5), 3146-3158. M.J. Matthews, J.W.P. Hsu, S. Gu, and T.F. Kuech. Applied Physics Letters, 2001. 79(19), 3086-3088. T. Mitani, S. Nakashima, H. Okumura, and A. Ogura. Journal of Applied Physics, 2006. 100(7). T. Mitani, S. Nakashima, H. Okumura, and A. Ogura. Journal of Applied Physics, 2006. 100(7), 073511. O. Pages, M.A. Renucci, O. Briot, and R.L. Aulombard. Journal of Applied Physics, 1995. 77(3), 1241-1248. 98

115. 116. 117. 118. 119. 120. 121. 122. 123. 124. 125. 126. 127. 128. 129. 130. 131. 132. 133. 134. 135. 136. 137. 138. 139. 140.

D.J. Lockwood and Z.R. Wasilewski. Physical Review B, 2004. 70(15), 155202. P. Stiles, J. Dieringer, N. Shah, and R. Van Duyne. Annual Review of Analytical Chemistry, 2008. 1, 601-626. C.L. Haynes, A.D. McFarland, and R.P. Van Duyne. Analytical Chemistry, 2005. 77(17), 338A-346A. E.C. Le Ru, E. Blackie, M. Meyer, and P.G. Etchegoin. Journal of Physical Chemistry C, 2007. 111(37), 13794-13803. P.G. Etchegoin and E.C. Le Ru. Physical Chemistry Chemical Physics, 2008. 10(40), 6079-6089. M. Moskovits. Reviews of Modern Physics, 1985. 57(3), 783-826. A.G. Milekhin, L.L. Sveshnikova, T.A. Duda, N.V. Surovtsev, S.V. Adichtchev, and D.R.T. Zahn. JETP Letters, 2008. 88(12), 799-801. G. Upender, R. Satyavathi, B. Raju, K.S. Alee, D.N. Rao, and C. Bansal. Chemical Physics Letters, 2011. 511(4-6), 309-314. N.V. Surovtsev, S.V. Adichtchev, T.A. Duda, L.D. Pokrovsky, and L.L. Sveshnikova. Journal of Physical Chemistry C, 2010. 114(11), 4803-4807. X.L. Li, W.Q. Xu, J.H. Zhang, H.Y. Jia, B. Yang, B. Zhao, B.F. Li, and Y. Ozaki. Langmuir, 2004. 20(4), 1298-1304. C.L. Haynes and R.P. Van Duyne. Journal of Physical Chemistry B, 2001. 105(24), 5599-5611. A.B. Zrimsek, A.I. Henry, and R.P. Van Duyne. Journal of Physical Chemistry Letters, 2013. 4(19), 3206-3210. M. Tabatabaei, A. Sangar, N. Kazemi-Zanjani, P. Torchio, A. Merlen, and F. LagugneLabarthet. Journal of Physical Chemistry C, 2013. 117(28), 14778-14786. G. Das, F. Mecarini, F. Gentile, F. De Angelis, M.H.G. Kumar, P. Candeloro, C. Liberale, G. Cuda, and E. Di Fabrizio. Biosensors & Bioelectronics, 2009. 24(6), 1693-1699. M. Kahl, E. Voges, S. Kostrewa, C. Viets, and W. Hill. Sensors and Actuators BChemical, 1998. 51(1-3), 285-291. M. Ranjan and S. Facsko. Nanotechnology, 2012. 23, 485307. K. Chen, K.C. Vo-Dinh, F. Yan, M.B. Wabuyele, and T. Vo-Dinh. Analytica Chimica Acta, 2006. 569(1-2), 234-237. D. Cialla, A. Maerz, R. Boehme, F. Theil, K. Weber, M. Schmitt, and J. Popp. Analytical and Bioanalytical Chemistry, 2012. 403(1), 27-54. C.Y.P. Yang, E.L. Yang, C.A. Steinhaus, C.-C. Liu, P.F. Nealey, and J.L. Skinner. Journal of Vacuum Science & Technology B, 2012. 30, 026801. U. Huebner, R. Boucher, H. Schneidewind, D. Cialla, and J. Popp. Microelectronic Engineering, 2008. 85(8), 1792-1794. L. Gutierrez-Rivera, R.F. Peters, S.K. Dew, and M. Stepanova. Journal of Vacuum Science & Technology B, 2013. 31, 06F901. A.G. Milekhin, L.L. Sveshnikova, S.M. Repinsky, A.K. Gutakovsky, M. Friedrich, and D.R.T. Zahn. Thin Solid Films, 2002. 422(1-2), 200-204. A.G. Milekhin, N.A. Yeryukov, L.L. Sveshnikova, T.A. Duda, S.S. Kosolobov, A.V. Latyshev, N.V. Surovtsev, S.V. Adichtchev, C. Himcinschi, E.I. Zenkevich, W.-B. Jian, and D.R.T. Zahn. Journal of Physical Chemistry C, 2012. 116(32), 17164-17168. P.G. Etchegoin, E.C. Le Ru, and M. Meyer. Journal of Chemical Physics, 2006. 125, 164705. R. Aroca, N. Pieczonka, and A.P. Kam. Journal of Porphyrins and Phthalocyanines, 2001. 5(1), 25-32. V. Linss, O. Stenzel, and D.R.T. Zahn. Journal of Raman Spectroscopy, 1999. 30(7), 531-536.

99

141. 142. 143. 144. 145. 146. 147. 148.

149. 150.

151. 152. 153. 154. 155. 156. 157. 158. 159. 160. 161. 162. 163. 164. 165. 166.

T.V. Basova, V.G. Kiselev, B.-E. Schuster, H. Peisert, and T. Chasse. Journal of Raman Spectroscopy, 2009. 40(12), 2080-2087. C.G. Claessens, U. Hahn, and T. Torres. Chemical Record, 2008. 8(2), 75-97. A.T. Davidson. Journal of Chemical Physics, 1982. 77(1), 168-172. M. Szybowicz, W. Bala, K. Fabisiak, K. Paprocki, and M. Drozdowski. Journal of Materials Science, 2011. 46(20), 6589-6595. W. Kiefer and S. Schlücker, Surface enhanced Raman spectroscopy: analytical, biophysical and life science applications. John Wiley & Sons, 2013. W. Yue, Z. Wang, Y. Yang, L. Chen, A. Syed, K. Wong, and X. Wang. Journal of Micromechanics and Microengineering, 2012. 22, 125007. E.B. Urena, M.P. Kreuzer, S. Itzhakov, H. Rigneault, R. Quidant, D. Oron, and J. Wenger. Advanced Materials, 2012. 24(44), OP314-OP320. C.L. Haynes, A.D. McFarland, L.L. Zhao, R.P. Van Duyne, G.C. Schatz, L. Gunnarsson, J. Prikulis, B. Kasemo, and M. Kall. Journal of Physical Chemistry B, 2003. 107(30), 7337-7342. J.T. Hugall, J.J. Baumberg, and S. Mahajan. Applied Physics Letters, 2009. 95(14), 141111. A.G. Milekhin, N.A. Yeryukov, L.L. Sveshnikova, T.A. Duda, E.I. Zenkevich, S.S. Kosolobov, A.V. Latyshev, C. Himcinski, N.V. Surovtsev, S.V. Adichtchev, Z.C. Feng, C.C. Wu, D.S. Wuu, and D.R.T. Zahn. Journal of Experimental and Theoretical Physics, 2011. 113(6), 983-991. L.J. Zhao, L.F. Hu, and X.S. Fang. Advanced Functional Materials, 2012. 22(8), 15511566. V.I. Klimov, A.A. Mikhailovsky, S. Xu, A. Malko, J.A. Hollingsworth, C.A. Leatherdale, H.J. Eisler, and M.G. Bawendi. Science, 2000. 290(5490), 314-317. R.C. Somers, M.G. Bawendi, and D.G. Nocera. Chemical Society Reviews, 2007. 36(4), 579-591. I. Grozdanov and M. Najdoski. Journal of Solid State Chemistry, 1995. 114(2), 469-475. B. MincevaSukarova, M. Najdoski, I. Grozdanov, and C.J. Chunnilall. Journal of Molecular Structure, 1997. 410-411, 267-270. D. Roy, C.M. Williams, and K. Mingard. Journal of Vacuum Science & Technology B, 2010. 28(3), 631-634. N. Kumar, A. Rae, and D. Roy. Applied Physics Letters, 2014. 104, 123106. R.D. Rodriguez, E. Sheremet, S. Mueller, O.D. Gordan, A. Villabona, S. Schulze, M. Hietschold, and D.R.T. Zahn. Review of Scientific Instruments, 2012. 83, 123708. I. Romero, J. Aizpurua, G.W. Bryant, and F.J. Garcia de Abajo. Optics Express, 2006. 14(21), 9988-9999. A. Downes, D. Salter, and A. Elfick. Journal of Physical Chemistry B, 2006. 110(13), 6692-6698. D. Roy, J. Wang, and C. Williams. Journal of Applied Physics, 2009. 105, 013530. A. Hartschuh, N. Anderson, and L. Novotny. Journal of Microscopy-Oxford, 2003. 210, 234-240. E. Bortchagovsky, T. Schmid, and R. Zenobi. Applied Physics Letters, 2013. 103, 043111. L. Novotny, E.J. Sanchez, and X.S. Xie. Ultramicroscopy, 1998. 71(1-4), 21-29. N. Anderson, A. Bouhelier, and L. Novotny. Journal of Optics a-Pure and Applied Optics, 2006. 8(4), S227-S233. N. Hayazawa, Y. Inouye, Z. Sekkat, and S. Kawata. Chemical Physics Letters, 2001. 335(5-6), 369-374.

100

167.

168. 169. 170. 171. 172. 173. 174. 175. 176. 177. 178. 179. 180. 181. 182. 183. 184. 185. 186. 187. 188. 189. 190. 191. 192. 193. 194.

D. Zhang, X. Wang, K. Braun, H.-J. Egelhaaf, M. Fleischer, L. Hennemann, H. Hintz, C. Stanciu, C.J. Brabec, D.P. Kern, and A.J. Meixner. Journal of Raman Spectroscopy, 2009. 40(10), 1371-1376. P. Anger, A. Feltz, T. Berghaus, and A.J. Meixner. Journal of Microscopy-Oxford, 2003. 209, 162-166. J. Steidtner and B. Pettinger. Review of Scientific Instruments, 2007. 78, 103104. M.A. Lieb and A.J. Meixner. Optics Express, 2001. 8(7), 458-474. W.X. Sun and Z.X. Shen. Ultramicroscopy, 2003. 94(3-4), 237-244. J. Stadler, T. Schmid, and R. Zenobi. Nano Letters, 2010. 10(11), 4514-4520. M.S. Anderson and W.T. Pike. Review of Scientific Instruments, 2002. 73(3), 1198-1203. D. Mehtani, N. Lee, R.D. Hartschuh, A. Kisliuk, M.D. Foster, A.P. Sokolov, and J.F. Maguire. Journal of Raman Spectroscopy, 2005. 36(11), 1068-1075. R. Ramos and M.J. Gordon. Review of Scientific Instruments, 2012. 83, 093706. C.C. Neacsu, J. Dreyer, N. Behr, and M.B. Raschke. Physical Review B, 2006. 73, 193406. B. Pettinger, B. Ren, G. Picardi, R. Schuster, and G. Ertl. Physical Review Letters, 2004. 92, 096101. A. Downes, R. Mouras, M. Mari, and A. Elfick. Journal of Raman Spectroscopy, 2009. 40(10), 1355-1360. S.S. Kharintsev, G.G. Hoffmann, A.I. Fishman, and M.K. Salakhov. Journal of Physics D-Applied Physics, 2013. 46, 145501. N. Anderson, P. Anger, A. Hartschuh, and L. Novotny. Nano Letters, 2006. 6(4), 744749. M. Micic, N. Klymyshyn, Y.D. Suh, and H.P. Lu. Journal of Physical Chemistry B, 2003. 107(7), 1574-1584. B. Ren, G. Picardi, and B. Pettinger. Review of Scientific Instruments, 2004. 75(4), 837841. T. Umakoshi, T.-a. Yano, Y. Saito, and P. Verma. Applied Physics Express, 2012. 5, 052001. P. Brodard, M. Bechelany, L. Philippe, and J. Michler. Journal of Raman Spectroscopy, 2012. 43(6), 745-749. S.H. Christiansen, M. Becker, S. Fahlbusch, J. Michler, V. Sivakov, G. Andra, and R. Geiger. Nanotechnology, 2007. 18, 035503. T. Kim, K.-S. Jeon, K. Heo, H.M. Kim, J. Park, Y.D. Suh, and S. Hong. Analyst, 2013. 138(19), 5588-5593. T.W. Johnson, Z.J. Lapin, R. Beams, N.C. Lindquist, S.G. Rodrigo, L. Novotny, and S.H. Oh. Acs Nano, 2012. 6(10), 9168-9174. S.S. Kharintsev, A.M. Rogov, and S.G. Kazarian. Review of Scientific Instruments, 2013. 84, 093106. L. Eligal, F. Culfaz, V. McCaughan, N.I. Cade, and D. Richards. Review of Scientific Instruments, 2009. 80, 033701. G. Xu, Z. Liu, K. Xu, Y. Zhang, H. Zhong, Y. Fan, and Z. Huang. Review of Scientific Instruments, 2012. 83, 103708. J.S. Lloyd, A. Williams, R.H. Rickman, A. McCowen, and P.R. Dunstan. Applied Physics Letters, 2011. 99, 143108. C.C. Neacsu, G.A. Steudle, and M.B. Raschko. Applied Physics B-Lasers and Optics, 2005. 80(3), 295-300. M. Lopes, T. Toury, M.L. de La Chapelle, F. Bonaccorso, and P.G. Gucciardi. Review of Scientific Instruments, 2013. 84, 073702. M. Asghari-Khiavi, B.R. Wood, P. Hojati-Talemi, A. Downes, D. McNaughton, and A. Mechler. Journal of Raman Spectroscopy, 2012. 43(2), 173-180. 101

195. 196. 197. 198. 199. 200. 201. 202. 203. 204. 205. 206. 207. 208. 209. 210. 211. 212. 213. 214. 215. 216. 217. 218. 219. 220. 221.

N. Hayazawa, T.-a. Yano, and S. Kawata. Journal of Raman Spectroscopy, 2012. 43(9), 1177-1182. C. Ropers, C.C. Neacsu, T. Elsaesser, M. Albrecht, M.B. Raschke, and C. Lienau. Nano Letters, 2007. 7(9), 2784-2788. F. De Angelis, G. Das, P. Candeloro, M. Patrini, M. Galli, A. Bek, M. Lazzarino, I. Maksymov, C. Liberale, L.C. Andreani, and E. Di Fabrizio. Nature Nanotechnology, 2010. 5(1), 67-72. R.D. Rodriguez, A. Anne, E. Cambril, and C. Demaille. Ultramicroscopy, 2011. 111(8), 973-981. F.J. Espinoza-Beltran, K. Geng, J.M. Saldana, U. Rabe, S. Hirsekorn, and W. Arnold. New Journal of Physics, 2009. 11, 083034. U. Neugebauer, P. Roesch, M. Schmitt, J. Popp, C. Julien, A. Rasmussen, C. Budich, and V. Deckert. Chemphyschem, 2006. 7(7), 1428-1430. C. Budich, U. Neugebauer, J. Popp, and V. Deckert. Journal of Microscopy-Oxford, 2008. 229(3), 533-539. B.R. Wood, E. Bailo, M.A. Khiavi, L. Tilley, S. Deed, T. Deckert-Gaudig, D. McNaughton, and V. Deckert. Nano Letters, 2011. 11(5), 1868-1873. J. Steidtner and B. Pettinger. Physical Review Letters, 2008. 100, 236101. N. Hayazawa, Y. Inouye, Z. Sekkat, and S. Kawata. Journal of Chemical Physics, 2002. 117(3), 1296-1301. D. Zhang, U. Heinemeyer, C. Stanciu, M. Sackrow, K. Braun, L.E. Hennemann, X. Wang, R. Scholz, F. Schreiber, and A.J. Meixner. Physical Review Letters, 2010. 104, 056601. N. Anderson, A. Hartschuh, and L. Novotny. Nano Letters, 2007. 7(3), 577-582. G. Picardi, M. Chaigneau, and R. Ossikovski. Chemical Physics Letters, 2009. 469(1-3), 161-165. L.G. Cancado, A. Hartschuh, and L. Novotny. Journal of Raman Spectroscopy, 2009. 40(10), 1420-1426. T.-a. Yano, T. Ichimura, S. Kuwahara, F. H'Dhili, K. Uetsuki, Y. Okuno, P. Verma, and S. Kawata. Nature Communications, 2013. 4, 2592. N. Marquestaut, D. Talaga, L. Servant, P. Yang, P. Pauzauskie, and F. LagugneLabarthet. Journal of Raman Spectroscopy, 2009. 40(10), 1441-1445. Y. Ogawa, Y. Yuasa, F. Minami, and S. Oda. Applied Physics Letters, 2011. 99(5). Y. Ogawa, T. Toizumi, F. Minami, and A.V. Baranov. Physical Review B, 2011. 83, 081302(R). J. Stadler, T. Schmid, and R. Zenobi. Nanoscale, 2012. 4(6), 1856-1870. Z. Yang, J. Aizpurua, and H. Xu. Journal of Raman Spectroscopy, 2009. 40(10), 13431348. M. Sun, Y. Fang, Z. Yang, and H. Xu. Physical Chemistry Chemical Physics, 2009. 11(41), 9412-9419. J. Chen, W. Yang, K. Dick, K. Deppert, H.Q. Xu, L. Samuelson, and H. Xu. Applied Physics Letters, 2008. 92, 093110. W. Zhang, X. Cui, B.-S. Yeo, T. Schmid, C. Hafner, and R. Zenobi. Nano Letters, 2007. 7(5), 1401-1405. E.D. Palik, Handbook of optical constants of solids. Vol. 3. Academic press, 1998. F. Perrozzi, S. Prezioso, and L. Ottaviano. Journal of Physics: Condensed Matter, 2015. 27, 013002. M.S. Dresselhaus, G. Dresselhaus, and M. Hofmann. Vibrational Spectroscopy, 2007. 45(2), 71-81. K.N. Kudin, B. Ozbas, H.C. Schniepp, R.K. Prud'homme, I.A. Aksay, and R. Car. Nano Letters, 2008. 8(1), 36-41. 102

222. 223. 224. 225. 226. 227. 228. 229. 230. 231. 232.

233. 234. 235. 236. 237. 238. 239. 240. 241.

J. Stadler, T. Schmid, and R. Zenobi. Acs Nano, 2011. 5(10), 8442-8448. A.C. Ferrari and D.M. Basko. Nature Nanotechnology, 2013. 8(4), 235-246. S.S. Kharintsev, G.G. Hoffmann, P.S. Dorozhkin, G. de With, and J. Loos. Nanotechnology, 2007. 18, 315502. N. Hayazawa, T. Yano, H. Watanabe, Y. Inouye, and S. Kawata. Chemical Physics Letters, 2003. 376(1-2), 174-180. D. Roy and C. Williams. Journal of Vacuum Science & Technology A, 2010. 28(3), 472475. A. Hartschuh, E.J. Sanchez, X.S. Xie, and L. Novotny. Physical Review Letters, 2003. 90, 095503. C. Georgi and A. Hartschuh. Applied Physics Letters, 2010. 97, 143117. N. Anderson, A. Hartschuh, S. Cronin, and L. Novotny. Journal of the American Chemical Society, 2005. 127(8), 2533-2537. A. Eckmann, A. Felten, A. Mishchenko, L. Britnell, R. Krupke, K.S. Novoselov, and C. Casiraghi. Nano Letters, 2012. 12(8), 3925-3930. P. Corio, P.S. Santos, V.W. Brar, G.G. Samsonidze, S.G. Chou, and M.S. Dresselhaus. Chemical Physics Letters, 2003. 370(5-6), 675-682. K.K. Kim, J.J. Bae, H.K. Park, S.M. Kim, H.-Z. Geng, K.A. Park, H.-J. Shin, S.-M. Yoon, A. Benayad, J.-Y. Choi, and Y.H. Lee. Journal of the American Chemical Society, 2008. 130(38), 12757-12761. M.S. Dresselhaus, G. Dresselhaus, and P.C. Eklund. Journal of Raman Spectroscopy, 1996. 27(3-4), 351-371. A. Kudelski. Journal of Raman Spectroscopy, 2007. 38(11), 1494-1499. K.F. Domke, D. Zhang, and B. Pettinger. Journal of Physical Chemistry C, 2007. 111(24), 8611-8616. B. Pettinger, K.F. Domke, D. Zhang, G. Picardi, and R. Schuster. Surface Science, 2009. 603(10-12), 1335-1341. J.A. Porto, P. Johansson, S.P. Apell, and T. Lopez-Rios. Physical Review B, 2003. 67, 085409. K.F. Domke, D. Zhang, and B. Pettinger. Journal of the American Chemical Society, 2006. 128(45), 14721-14727. G. Picardi, M. Chaigneau, R. Ossikovski, C. Licitra, and G. Delapierre. Journal of Raman Spectroscopy, 2009. 40(10), 1407-1412. T. Deckert-Gaudig and V. Deckert. Small, 2009. 5(4), 432-436. F. Pashaee, R. Hou, P. Gobbo, M.S. Workentin, and F. Lagugne-Labarthet. Journal of Physical Chemistry C, 2013. 117(30), 15639-15646.

103

List of figures Figure 1.1. The energy diagram for (a) Rayleigh, (b) Stokes Raman, (c) and anti-Stokes Raman scattering processes. The energy difference ∆𝐸 between incident and scattered photons is equal, above, or below zero, respectively for each scattering process. The upward pointing arrow implies annihilation of a photon with energy ℏ𝜔𝑖 , and the downward pointing arrow implies the creation of a photon with energies indicated. The solid lines indicate real energy levels, while dashed lines indicate virtual energy levels. ................................................16 Figure 1.2. Scattered electric field enhancement around a sphere of 10 nm radius in a dielectric medium with 𝜀𝑚 =1 as a function of the dielectric constant of the sphere. (a) The dependence of scattered electric field maximum on the dielectric constants of the sphere. Distribution of scattered electric field around a sphere with 𝜀 = −2 (a) and +2 (b). .................21 Figure 1.3. Dipole above a semi-infinite surface, and its image dipole. ..........................22 Figure 1.4. Illustration of charge transfer chemical enhancement mechanism. ..............24 Figure 2.1. Structure of the NC multilayer samples: InAs/AlAs with 50 nm (sample A) and 100 nm (sample B) periods, InAs/Al(Ga)As with 100 nm period (sample C), and AlAs/InAs with 50 nm period (sample D). ..........................................................................................................28 Figure 2.2. Schematics of the AFM (a, b) and Raman (c) experiments. Dotted lines indicate NC layers embedded in the matrix. ..............................................................................30 Figure 2.3. (a) Dark field TEM image of NC multilayer structures (sample C). HRTEM images of (b) InAs NC in AlGaAs matrix (sample C), (c) InAs NC in AlAs matrix (sample B), and (d) AlAs NC in InAs matrix (sample D). .....................................................................................32 Figure 2.4. Raman spectra of InAs/AlAs and AlAs/InAs NC multilayers. (a) The plot is divided in two spectral regions: of InAs phonons (a), and of AlAs phonons (b). The upper spectra (black) were acquired in 𝑧(𝑥, 𝑦)𝑧̅ geometry (LO phonon allowed), and the lower spectra

̅ geometry (TO phonon allowed). For comparison, the spectra of the matrix (blue) in 𝑦′(𝑥′, 𝑧)𝑦′ (dashed lines) and of the NCs (solid lines) are overlaid. The shift of phonon frequency in NCs is indicated with horizontal black and blue arrows and is induced by built-in strain. Green arrows indicate fitted LO and TO phonon frequencies. The frequencies indicated with italic fonts show calculated phonon frequencies in the absence of intermixing and confinement effect. Red arrow indicates LO phonon frequency measured with 632.8 nm laser. ...............................................33 Figure 2.5. AFM images and corresponding topography profiles of

cleaved

InAs/Al(Ga)As samples. Each column represents topography (top), amplitude (middle), and 104

topography profile (bottom) along the blue arrow in the topography image. Being sensitive to the topography gradient, the amplitude image highlights surface features. (a) sample A (InAs/AlAs, 50 nm period); (b) B (InAs/AlAs, 100 nm period); (c) C (InAs/AlGaAs, 100 nm period). .............35 Figure 2.6. AFM images and topography profiles of ion milled AlAs/InAs samples prepared at 6 keV ion energy for different preparation temperatures: (a) 296K (room temperature); (b) 193 K; (c) 143 K. The amplitude or phase image in the middle row is given for a more clear view of the periodic structure of the samples. The bottom row presents the topography profiles along the blue arrows (phase profile for c). ................................................36 Figure 2.7. Normalized Raman spectra of ion milled and cleaved AlAs/InAs samples evidencing sample damage introduced by ion milling. According to Raman selection rules [36], in the backscattering geometry from {110} plane only TO modes are allowed, while LO modes are symmetry forbidden. ...........................................................................................................38 Figure 2.8. Typical Raman spectra of beveled (a) InAs/AlAs (sample B) and (b) InAs/AlGaAs (sample C) NC multilayers. The spectra are averaged over four points. ...............39 Figure 2.9. Schematic structures, optical images, and Raman intensity profiles of one single stack of (a) InAs/AlAs NC and (b) InAs/AlGaAs NC structures. Schematics of the sample structure: grey areas – GaAs interlayers; white – Al(Ga)As matrix. Vertical arrows indicate the position of InAs NC layers. ........................................................................................................40 Figure 3.1. Typical SEM images (400 × 400 nm2) of gold nanocluster arrays with periods of 110, 130, and 150 nm (left, middle and right columns, respectively) fabricated on Si surfaces. .................................................................................................................................................44 Figure 3.2. (a) Raman and SERRS spectra of 2 nm CoPc film. Note the difference in scale. For selected peaks the symmetry of the corresponding vibration and the enhancement factor are indicated. (b) The vibrations corresponding to the most intense Raman modes of the CoPc molecule from DFT calculations. Arrows indicate the atoms of the central ring involved in the vibration, and the direction of their displacement. ................................................................49 Figure 3.3. a) SERRS and Raman spectra of 0.15 nm and 0.5 nm CoPc films with 676.4 nm wavelength. All spectra were vertically shifted for better representation. b) The dependence of SERS enhancement factor for the 0.15 nm CoPc film of the mode at 682 cm−1 as a function gold nanocluster diameter for the array with period of 150 nm for 632.8, 641.6, and 676.4 nm excitation. The symbols represent the experimental data, solid lines are guides for the eye. ....50 Figure 3.4. SEM image of one of the dimer structures. The distance between centers of the clusters is kept constant for all arrays (100 nm between the nanoclusters in a dimer and 130

105

nm between the neighboring dimers), while the gold nanocluster diameter changes from array to array from 25 to 90 nm. .............................................................................................................51 Figure 3.5. a) A schematic of a dimer and polarization direction of the incident light with respect to the dimers indicated by arrows; b) and c) are the enhancement factors for the structures with different dimer dimensions measured with 514.5 and 632.8 nm excitation line, respectively, and polarization (○) along and (■) across the dimers. Solid lines are guides for the eye. Arrows indicate enhancement factors for a nanostructured gold film, measured with the polarization across (┴) and along (‖) the dimers........................................................................52 Figure 3.6. Typical SEM image of the structure with CuS NCs formed on bare Si (bottom part of the image) and arrays of gold nanoclusters (top part of the image). (b) SERS and Raman spectra of CuS NCs formed on arrays of gold nanoclusters and on bare Si and measured with 514.5 nm laser line. ...................................................................................................................53 Figure 3.7. (a) SEM image of an edge of the nanostructured gold substrate covered with a monolayer of CdSe NCs. Inset shows an example of high magnification image on the silicon substrate covered with CdSe NCs. (b) Raman (black curve) and SERS (red curve) spectra of CdSe NCs measured with 632.8 nm laser line. .........................................................................54 Figure 3.8. (a) Dependence of Raman intensity of the CdSe LO phonon on the gold nanocluster size for different wavelengths; (b) Comparison of experimental and simulated LSPR positions for different nanocluster diameters with parameters, for which maximum SERS intensities were observed. .........................................................................................................55 Figure 4.1. General schematic of a TERS experiment. ..................................................57 Figure 4.2. (a) A schematic of a generic TERS experiment illustrating illuminated area as compared to the tip apex radius; (b) A generic comparison of the far- and near-field signals. ...59 Figure 4.3. Spot size determination using the scanning knife-edge method. (a) AFM image of the CoPc/gold interface obtained by photolithography. The sharp interface is evidenced in the cross section shown in (b). A Raman line scan was performed across the interface; the intensity of the CoPc band at 1545 cm−1 is plotted in (c). FWHM of the derivate of the line scan gives a value of 4.7 μm. The approximate size of the laser spot gives a value of 4.7x2.5 μm2 with accuracy 8%. ................................................................................................62 Figure 4.4. Photo (a) and schematic layout (b) of the TERS setup. Inset in (a) shows an optical image of an AFM TERS tip. Red lines indicate the light path. ........................................63 Figure 4.5. (a) SEM image of a mechanically cut custom-made silver AFM cantilever, (b) AFM image in intermittent contact mode of the 1D array of silicon stripes acquired with this

106

cantilever, (c) Height profile along the dashed line in (b). (d-f) Analogous set of data for an electrochemically etched gold tip. .............................................................................................67 Figure 4.6. (a) SEM image of an electrochemically etched gold tip. (b)-(c) Wear off of the metal coating of a conventional silicon tip with a silver layer. (b) The tip before and (c) after an AFM-TERS experiment. ............................................................................................................68 Figure 4.7. The models used for the analysis of the mechanical properties. From left to right (increasing model complexity): (a) rectangular cantilever, (b) single segment cantilever, (c) two segment cantilever, and (d) realistic cantilever. Numbers denote the index of the 𝐻 and 𝑊 parameter of the respective cross-section. Grey planes indicate fixed end of the cantilever......69 Figure 4.8. Modal analysis of rectangular (upper row) and single segment (lower row) beams made of gold. Geometrical parameters used in the calculations are given on the lefthand side. The color scale indicates the normalized displacement (red stands for highest displacement from the original position, blue for lowest). Arrows indicate the direction of the oscillation (out-of-plane (oop), in-plane (ip), or torsional (tors)). ................................................71 Figure 4.9. The dependence of the cantilever spring constant on geometrical parameters of the cantilever for a single segment model. ............................................................................73 Figure 4.10. a) FEM model of a realistic gold cantilever mounted on a silicon chip; b) comparison of experimental and simulated frequency response of the gold cantilever. Numbers indicate experimental (simulated) frequencies of the respective modes; c) illustration of application of the point force on the tip apex, and the measured displacement for determination of the spring constant. ...............................................................................................................74 Figure 5.1. XploRA – Omegascope-R TERS system, the main components indicated. Image courtesy Andrey Krayev, AIST-NT, Inc. ..........................................................................77 Figure 5.2. Example of a mesh for a quarter of a three-dimensional model with a gold tip above half of a gold ellipsoid and gold film. Additional volume introduced for finer meshing is placed at the tip apex. ...............................................................................................................78 Figure 5.3. The representative spectra (a) and TERS image (b) of the three components. At each pixel of the TERS image full Raman spectrum is acquired. The TERS image is composed from the intensity distributions of the D, C5 and G bands marked in (a) over the image area. Green color follows the intensity of the D band, while red color indicates the intensity of 2D (G’) band. Blue color corresponds to the pentagon ring vibration of C60 (marked with circles in the map). For comparison, the dashed contour indicates the size of the diffraction limited laser spot. The 2D band intensity profile shown with white line is taken along the white arrow. ........................................................................................................................................79 107

Figure 5.4. Selected spectra of organic residue on the substrate...................................80 Figure 5.5. (a) and (d) D to G ratio (identical images). The markings indicate the positions relevant for the analysis of the GO (a) and CNT (d) TERS data. (b) D to G ratio along the line marked in (a). (c) Average spectra along the profile. Ellipses in (d) indicate the CNT regions with high defect concentrations. (e) Selected TERS spectra along lines in (d). .............81 Figure 5.6. (a) Schematic of the sample cross-section. (b) AFM and (c) respective TERS image. (d) Averaged CoPc spectra along the middle line of the TERS image (marked with rectangle in c) from Si to gold....................................................................................................83 Figure 5.7. Simulated enhancement of electric field for a gold tip of 30 nm radius above (a) Si substrate, (b) 40 nm thick gold film on Si, (c) gold protrusion of 40 nm height and 200 nm diameter on gold film and Si substrate. Incident field is polarized along the tip, wavelength 633 nm.............................................................................................................................................84 Figure 5.8. TERS spectra on gold nanoclusters and silicon covered with 2 nm CoPc measured in contact (a) and intermittent contact (b) modes. .....................................................85 Figure 5.9. Image dipole model of a tip above a semi-infinite surface. (a) Parameters of the model. (b) Dependence of the electric field at the surface on the distance between the tip and the sample for a tip of 20 nm radius. ..................................................................................86 Figure 5.10. (a) SEM image of gold nanoclusters with 500 nm period on Si. TERS images on four gold nanoclusters in contact (b) and intermittent contact (c) modes. Color scale shows Raman intensity in kcts. .................................................................................................87 Figure 5.11. TERS in contact mode. (a) Sketch of the experimental layout. (b) AFM profile of the gold nanocluster (right axis) and the corresponding TERS profile at 682 cm -1 (left axis). (c) Simulated enhancement of the electric field between the tip and the sample for tip positions 1, 2 and 3 with respect to the gold nanocluster, which correspond to the positions 1, 2 and 3 indicated in profile in (b). EF indicates maximum electric field enhancement for the respective position taking into account the illumination angle. ...................................................88 Figure 5.12. (a) Height (right axis) and TERS signal (left axis) intensity profiles over four gold nanoclusters. (b-d) TERS imaging of a single gold nanocluster using a gold tip: (b) AFM topography, (c) phase image, (d) TERS intensity of 682 cm-1 mode. .........................................89 Figure 5.13. AFM topography (a) and phase (b) image of a gold nanocluster acquired with a commercial Si cantilever. ................................................................................................89 Figure 5.14. TERS on CdSe monolayer on SERS substrate. (a) Schematic of the experiment. (b) TERS spectra. ..................................................................................................90

108

Erklärung Ich erkläre, dass ich die vorliegende Arbeit selbständig und nur unter Verwendung der angegebenen Literatur und Hilfsmittel angefertigt habe. Februar 2015 M.Eng. Evgeniya Sheremet

109

110

Lebenslauf Sheremet Evgeniya Sergeevna Date of birth:

02.12.1988

Place of birth:

Novosibirsk, Russia

Nationality:

Russian

Citizenship:

Russian

Languages:

Russian (native), English (fluent), German (intermediate)

Education From 2014

Researcher at Solid surfaces analysis group, Technische Universität Chemnitz, Chemnitz, Germany

2011 – 2014

Ph.D. study in Semiconductor physics group, Technische Universität Chemnitz, Chemnitz, Germany

2009 – 2011

M.Eng. in Nanotechnology with honors, Novosibirsk State Technical University, Novosibirsk, Russia. Title: “Design of logic modules for quantum cryptography module”

2007 – 2010

Translator

in

Occupational

Field

of

Communication

(English

language) 2005 – 2009

B.Eng. in Nanotechnology with honors, Novosibirsk State Technical University, Novosibirsk, Russia. Title: “Monte-Carlo simulation of Si and Ge nanowhisker growth”

24-26.09.2014

Training School on STM, AFM, and SNOM

Awards and contributions 

First poster prize at Surface Enhanced Spectroscopies conference 2014

(SES 2014), 7-10.08.2014, Chemnitz, Germany.

111



Poster prize at an international conference Frontiers of Plasmonics 3

(FOP 3), 26.03 – 01.04.2014, Xiamen, China. 

The research results obtained during my Ph.D. were presented in 5 invited

talks and 19 international conference contributions.

Publication list 1.

E. Sheremet, R.D. Rodriguez, A.L. Agapov, A.P. Sokolov, M. Hietschold,

D.R.T. Zahn, Nanoscale imaging and identification of a four-component carbon sample. Carbon, 2016, 96, 588-593. 2.

V. Kolchuzhin, J. Mehner, E. Sheremet, K. Bhattacharya, R.D. Rodriguez and

D.R.T. Zahn, Understanding Tip-Enhanced Raman Spectroscopy by Multiphysics Finite Element

Simulations.

Proceedings

of

EuroSimE

2015,

2015,

1-5.

DOI

10.1109/EuroSimE.2015.7103161. 3.

H. Fiedler, M. Toader, S. Hermann, M. Rennau, R.D. Rodriguez, E. Sheremet,

M. Hietschold, D.R.T. Zahn, S.E. Schulz, T. Gessner, Back-end-of-line compatible contact materials for carbon nanotube based interconnects. Microelectronic Engineering, 2015, 137, 130–134. 4.

R.D. Rodriguez, E. Sheremet, T. Deckert-Gaudig, C. Chaneac, M. Hietschold,

V. Deckert and D.R.T. Zahn, Surface- and tip-enhanced Raman spectroscopy reveals spinwaves in iron oxide nanoparticles. Nanoscale, 2015, 7, 9545-9551. 5.

L. Mikoliunaite, R.D. Rodriguez, E. Sheremet, V. Kolchuzhin, J. Mehner, A.

Ramanavicius and D.R.T. Zahn, The substrate matters in the Raman spectroscopy analysis of cells. Scientific Reports, 2015, 5, 13150. 6.

P. Bayat, D. Vogel, R.D. Rodriguez, E. Sheremet, D.R.T. Zahn, S. Rzepka,

B. Michel, Thermo-mechanical characterization of copper through-silicon vias (Cu-TSVs) using micro-Raman spectroscopy and atomic force microscopy. Microelectronic Engineering, 2015, 137, 101-104. 7.

E. Sheremet, A.G. Milekhin, R.D. Rodriguez, T. Weiss, M. Nesterov,

E.E. Rodyakina, O.D. Gordan, L.L. Sveshnikova, T.A. Duda, V.A. Gridchin, V.M. Dzhagan, M. Hietschold and D.R.T. Zahn, Surface- and Tip-Enhanced Resonant Raman Scattering from CdSe Nanocrystals. Physical Chemistry Chemical Physics, 2015, 17, 21198-21203.

112

8.

A.G. Milekhin, N.A. Yeryukov, L.L. Sveshnikova, T.A. Duda, E.E. Rodyakina,

V.A. Gridchin, E.S. Sheremet and D.R.T. Zahn, Combination of surface- and interferenceenhanced Raman scattering by CuS nanocrystals on nanopatterned Au structures, Beilstein Journal of Nanotechnology, 2015, 6, 749-754. 9.

E. Sheremet, A. Gutakovskii, A. Milekhin, R.D. Rodriguez, D. Dentel,

W. Grünewald, D. Dmitriev, M. Hietschold, A. Toropov and D.R.T. Zahn, Raman, AFM, and TEM profiling of QD multilayer structures. Materials Research Express, 2015, 2, 035003. 10.

E. Sheremet, R.D. Rodriguez, D.R.T. Zahn, A.G. Milekhin, E.E. Rodyakina and

A.V. Latyshev, Surface-enhanced Raman scattering and gap-mode tip-enhanced Raman scattering investigations of phthalocyanine molecules on gold nanostructured substrates. Journal of Vacuum Science & Technology B, 2014, 32(4). 11.

E. Sheremet, R.D. Rodriguez, J. Kalbacova, M. Hietschold and D.R.T. Zahn,

Nanoscale Characterization of Carbon Nanotubes from Dispersions to Devices by TipEnhanced Raman Spectroscopy. In Proc. 12. Chemnitzer Fachtagung Mikrosystemtechnik 2014 (MST 2014), Chemnitz, Germany. 2014. 12.

R.D. Rodriguez., S. Mueller, E. Sheremet, D.R.T. Zahn, A. Villabona,

S.A. Lopez-Rivera, P. Tonndorf, S.M. de Vasconcellos and R. Bratschitsch, Selective Raman modes and strong photoluminescence of gallium selenide flakes on sp2 carbon. Journal of Vacuum Science & Technology B, 2014, 32(4). 13.

R. Patra, S. Ghosh, E. Sheremet, M. Jha, R.D. Rodriguez, D. Lehmann,

A.K. Ganguli, O.D. Gordan, H. Schmidt, S. Schulze, D.R.T. Zahn and O.G. Schmidt, Enhanced field emission from cerium hexaboride coated multiwalled carbon nanotube composite films: A potential material for next generation electron sources. Journal of Applied Physics, 2014, 115(9). 14.

R. Patra, S. Ghosh, E. Sheremet, M. Jha, R.D. Rodriguez, D. Lehmann,

A.K. Ganguli, H. Schmidt, S. Schulze, M. Hietschold, D.R.T. Zahn and O.G. Schmidt, Enhanced field emission from lanthanum hexaboride coated multiwalled carbon nanotubes: Correlation with physical properties. Journal of Applied Physics, 2014, 116(16). 15.

J. Luo, D. Billep, T. Blaudeck, E. Sheremet, R.D. Rodriguez, D.R.T. Zahn,

M. Toader, M. Hietschold, T. Otto and T. Gessner, Chemical post-treatment and thermoelectric properties of poly(3,4-ethylenedioxylthiophene):poly(styrenesulfonate) thin films. Journal of Applied Physics, 2014, 115(5). 16.

H. Fiedler, M. Toader, S. Hermann, R.D. Rodriguez, E. Sheremet, M. Rennau,

S. Schulze, T. Waechtler, M. Hietschold, D.R.T. Zahn, S.E. Schulz and T. Gessner, Carbon 113

nanotube based via interconnects: Performance estimation based on the resistance of individual carbon nanotubes. Microelectronic Engineering, 2014, 120: p. 210-215. 17.

A.G. Milekhin, N.A. Yeryukov, L.L. Sveshnikova, T.A. Duda, E.E. Rodyakina,

E.S. Sheremet, M. Ludemann, O.D. Gordan, A.V. Latyshev and D.R.T. Zahn, Surface enhanced Raman scattering by organic and inorganic semiconductors formed on laterally ordered arrays of Au nanoclusters. Thin Solid Films, 2013. 543: p. 35-40. 18.

J. Luo, D. Billep, T. Waechtler, T. Otto, M. Toader, O. Gordan, E. Sheremet,

J. Martin, M. Hietschold, D.R.T. Zahnd and T. Gessner, Enhancement of the thermoelectric properties of PEDOT:PSS thin films by post-treatment. Journal of Materials Chemistry A, 2013, 1(26): p. 7576-7583. 19.

R.D. Rodriguez, M. Toader, S. Hermann, E. Sheremet, S. Mueller, O.D. Gordan,

H. Yu, S.E. Schulz, M. Hietschold and D.R.T. Zahn, Nanoscale optical and electrical characterization of horizontally aligned single-walled carbon nanotubes. Nanoscale Research Letters, 2012, 7. 20.

R.D. Rodriguez, E. Sheremet, D.J. Thurmer, D. Lehmann, O.D. Gordan,

F. Seidel, A. Milekhin, O.G. Schmidt, M. Hietschold and D.R.T. Zahn, Temperature-dependent Raman investigation of rolled up InGaAs/GaAs microtubes. Nanoscale Research Letters, 2012, 7. 21.

R.D. Rodriguez, E. Sheremet, S. Mueller, O.D. Gordan, A. Villabona, S. Schulze,

M. Hietschold and D.R.T. Zahn, Compact metal probes: A solution for atomic force microscopy based tip-enhanced Raman spectroscopy. Review of Scientific Instruments, 2012, 83(12). 22.

A. Milekhin, N. Yeryukov, A. Toropov, D. Dmitriev, E. Sheremet and

D.R.T. Zahn, Raman scattering of InAs/AlAs quantum dot superlattices grown on (001) and (311)B GaAs surfaces. Nanoscale Research Letters, 2012, 7: p. 1-5. 23.

T. Ebert, G. Cox, E. Sheremet, O. Gordan, D.R.T. Zahn, F. Simon and

S. Spange, Carbon/carbon nanocomposites fabricated by base catalyzed twin polymerization of a Si-spiro compound on graphite sheets. Chemical Communications, 2012, 48(79): p. 98679869. 24.

H. Fiedler, M. Toader, S. Hermann, R.D. Rodriguez, E. Sheremet, M. Rennau,

S. Schulze, T. Waechtler, M. Hietschold, D.R.T. Zahn, S.E. Schulz and T. Gessner, Distinguishing between Individual Contributions to the Via Resistance in Carbon Nanotubes Based Interconnects. ECS Journal of Solid State Science and Technology, 2012, 1(6): p. M47M51.

114

25.

A. Nastovjak, I. Neizvestny, N. Shwartz and E. Sheremet, Mechanisms of

nanowhisker formation: Monte Carlo simulation. Optoelectronics, Instrumentation and Data Processing, 2009. 45(4): p. 342-347. 26.

E.S. Sheremet, A.G. Nastovjak, I.G. Neizvestny and N.L. Schwartz, Examination

of Nanotube Growth Conditions by Monte Carlo Simulation. In Proc. 10th Int. Workshops and Tutorials on Electron Devices and Materials (EDM-2009), Erlagol, Altai. 2009. 27.

A.G.

Nastovjak,

I.G.

Neizvestny,

N.L.

Shwartz,

E.S.

Sheremet

and

Z.S. Yanovitskaja. Effect of Substrate-Drop Parameters on Nanowhiskers Morphology. In Proc. 9th Int. Workshops and Tutorials on Electron Devices and Materials (EDM-2008), Erlagol, Altai. 2008.

115

116

Acknowledgements First and foremost, I would like to thank Dr. Raul D. Rodriguez for enthusiasm, energy, encouragement, and unconditional support. I am also very grateful to Prof. Dietrich R.T. Zahn for the opportunity to do this work, and constant guidance. Last but not least, I am indebted to Prof. Alexander Milekhin for finding time and helpful advice at every stage of the work.

Financial support was provided by Deutsche Forschungsgemeinschaft DFG (project ZA146/22-1) and FOR1713 “Sensorical Micro- and Nanosystems”.

Not a single chapter of this work would be possible without help of other group members and collaborators: In Chemnitz Alexander Villabona – electronics of the TERS setup and some essential parts of the TERS system; Axel Fechner – for technical support at each step of development and “Keine Panik”; Jane Eisentraut and Sybille Raschke – for tolerating endless questions about Germany in English; Dr. Vladimir Kolchuzhin – ANSYS simulations and numerous discussions; Lilibeth Leal – first EM field simulations in ANSYS, which served as a basis for all further work; Kunal Bhattacharya – characterization of all-metal TERS tips, and simulations of the realistic tip; Vivek Desale – statistical analysis of all-metal TERS tips; Jana Kalbacova – moral support and helpful discussions; Susanne Müller – initial TERS development, teaching me German and helpful comments; Dr. Daniel Lehmann – DFT calculations; Dr. Ovidiu D. Gordon – micro-ellipsometry measurements; Michael Ludemann, Dr. Francisc Haidu, Peter Robaschik, and Evelyn Brayer for deposition of metal phthalocyanine films; Dr. Steffen Schulze and Prof. Michael Hietschold – introduction to SEM, access to the electron microscopy and fruitful discussions;

117

Doreen Dentel – preparation of beveled AIIIBV NC sample; Dr. Wolfgang Grünewald and Thorsten Jagemann – ion milled samples; Dr. Sascha Hermann – providing CNT dispersion; Michael Ludemann, Dmytro Solonenko and Dr. Volodymyr Dzhagan for valuable discussions and tips.

In Novosibirsk Dr. Ekaterina Rodyakina and Dr. Alexander V.

Latyshev – preparation of SERS

substrates; Dr. Lyudmila L. Sveshnikova, Dr. Tatyana A. Duda – deposition of CuS and CdSe NCs; Dr. Alexander Toropov and Dr. Dmitry Dmitriev – growth of AIIIBV NCs; Dr. A. Gutakovskii – TEM on AIIIBV NCs. Around the world Dr. Andrey Krayev (AIST-NT, Inc., USA) and Dr. Emmanuel Leroy (Horiba Scientific, France) – access to AIST-NT – Horiba TERS setup, and especially Andrey for introduction to the setup, assistance with part of the TERS experiments, and preparation of the 4-component sample; Dr. Alexander Agapov, Prof. Alexey Sokolov (Oak Ridge National Laboratory, USA) – collaboration on TERS imaging of the 4-component sample; Prof. Eugene Bortchagovsky (Institute of High Technologies T.G.Shevchenko Kiev National University; V.Lashkaryov Institute of Semiconductor Physics, National Academy of Sciences of Ukraine) – image dipole model.

Finally, thanks to HLPH group for the critique and questions during the weekly discussions.

118

Suggest Documents