Raman Nanospectroscopy of a Photo-Catalytic Reaction

Raman Nanospectroscopy of a Photo-Catalytic

Reaction Cover: Photograph by the author, Evelien van Schrojenstein Lantman, of the Sol Lumen project, which was part of the 375th anniversary of Utrecht University. Yvette Mattern is the artist who created Sol Lumen.

Printed by Gildeprint Drukkerijen

ISBN 978-90-393-6260-0 Raman Nanospectroscopy of a Photo-Catalytic

Reaction

Raman Nanospectroscopie van een Fotokatalytische Reactie

(met een samenvatting in het Nederlands)

Proefschrift

ter verkrijging van de graad van doctor aan de

Universiteit Utrecht op gezag van de rector magniﬁcus,

prof. dr. G.J. van der Zwaan, ingevolge het besluit van het

college voor promoties in het openbaar te verdedigen

op woensdag 17 december 2014 des ochtends te 10.30 uur

door

Evelien Martine van Schrojenstein Lantman

geboren op 25 oktober 1986 te Rijswijk Promotor: Prof. dr. ir. B. M. Weckhuysen

Copromotor: Dr. A. J. G. Mank

This thesis is part of NanoNextNL, a micro and nanotechnology innovation consortium of the Government of the Netherlands and 130 partners from academia and industry. More information on www.nanonextnl.nl. This work was also ﬁnancially supported by the Netherlands Research School Combina- tion - Catalysis (NRSCC) and a European Research Council (ERC) Advanced Grant (no. 321140). Contents

Chapter 1 General Introduction 1

Chapter 2 Surface-Enhanced Raman Spectroscopy Studies of the 27 Photo-Catalytic Reaction of Nitro-Aromatics

Chapter 3 Reaction Kinetics of the Photo-Catalytic Reduction of 51 p-Nitrothiophenol

Chapter 4 Separation of Time-Resolved Phenomena in Surface- 69 Enhanced Raman Scattering of the Photo-Catalytic Re- duction of p-Nitrothiophenol

Chapter 5 The Photo-Catalytic Reduction of p-Nitrothiophenol 83 Monitored at the Nanoscale Using Tip-Enhanced Ra- man Spectroscopy

Chapter 6 Summary and Outlook 97

Chapter 7 Samenvatting en Vooruitblik 101

Appendix A Vibrational Assignments 107

Appendix B In-Situ Tip-Enhanced Raman Scattering Monitoring of 111 the Photo-Reduction of p-Nitrothiophenol

List of Abbreviations and Symbols 113

List of Publications and Presentations 115

Acknowledgements 119

Curriculum Vitae 121

CHAPTER 1

General Introduction

1 Chapter 1: General Introduction

1.1. Heterogeneous Catalysts

The field of chemistry is built upon the study of atoms and molecules and their interactions. By creating the right circumstances, it is possible to shuffle the atoms involved and get to the desired end-product. The art of chemistry is in finding those correct circumstances, but also in being able to recognise and characterize each compound. Chemistry is everywhere, and the chemical industry is a major player in the world’s economy. As manufacturers, they try to find the best road to a certain end-product, taking into account energy efficiency, sustainability, and cost efficiency. To make a chemical reaction happen between molecules, a certain amount of energy is required. This energy is called the activation energy of the chemical reaction, and can be seen as a barrier that needs to be crossed, as illustrated in Fig. 1.1a. A catalyst is a compound that lowers this energy barrier by creating a different reaction pathway. The word catalysis comes from the Greek κατα-λυειν, meaning "breaking down" or "dissolving". By definition, a catalyst is not consumed in the process. 1 The typical heterogeneous catalyst is most often a solid on the surface of which gasses or liquids can react. Fig. 1.1b schematically shows the steps involved in a classic catalytic reaction - the so-called Langmuir-Hinshelwood mechanism. Reactant A, which is often a gas, adsorbs onto the catalyst surface. This adsorbed species undergoes a reaction, yielding adsorbed product species B, which can then desorb again. For the actual catalysis to take place, it is crucial that sufficient surface area is available for A to adsorb, yet once

A(g) <-> A(ads) A(ads) <-> B(ads) B(ads) <-> B(g) A(g) <-> B(g)

FIG. 1.1. a) A catalyst provides an alternative pathway to a chemical reaction, such that the energy barrier is lowered. Figure taken from Chorkendorff and Niemantsverdriet. 1 b) Langmuir- Hinshelwood kinetics of a catalytic reaction.

2 Reaction Monitoring In-Situ a) b) c) d)

FIG. 1.2. a) Photograph of a car exhaust catalyst. b) The rigid support is a monolith with a honeycomb-like structure. c) Onto the monolith walls a second, porous support is fixated, which d) contains the actual catalytic active nanoparticles. Figures adapted from Chorkendorff and Niemantsverdriet. 1 reacted, B must also be able to leave the catalyst surface again, to make place for new reactant species A. Well-known is the car exhaust catalyst, shown in Fig. 1.2. This catalyst converts the exhaust gasses to less toxic compounds through a reduction or oxidation reaction, and requires a few steps of magnifications to visualize the actual active catalytic particles. The small particles have a large surface area with respect to the amount of material used, i.e. the largest accessibility for reactants. These particles are of nanometer size and sit on a support framework. In general, this approach is taken, as catalyst materials are often non-abundant metals, e.g. platinum, and therefore need to be used as efficiently as possible. When discussing catalytic properties, usually the properties of a whole assembly of supported nanoparticles are discussed. However, each of those particles can have a slightly different shape or size, each of which can affect the surface properties and thereby the catalytic activity. Identification of the most active catalytic particles and monitoring the chemical reactions over them, would allow a scientist to steer preparation of materials to even better performing catalysts.

1.2. Reaction Monitoring In-Situ

How can a discrimination be made between active and less- or non-active catalysts? Many analytical tools are available, but not all give the information required for in-situ reaction monitoring.

1.2.1. Electron Microscopy and X-Ray Spectroscopy. It is possible to investigate changes in the catalyst structure or morphology on a nanometer scale. Structural changes in the catalyst particles can be investigated in-situ using for example x-ray diffraction, to identify and characterize the crystalline structure of catalyst particles. 2,3 Changes in state

3 Chapter 1: General Introduction

TABLE 1.1. Diffraction limit for optical measurements.

Range Wavelength NA Diffraction limit

IR 2.5-15 m 0.7 2-15 µm Optical 400-750 nm 0.7 (dry) 290-540 nm 1.4 (oil) 140 - 270 nm

or shape can be observed during a reaction, but these cannot directly be linked to the reaction route or reactivity of the molecules involved. The excitation energy used in these techniques is too high to probe molecular-sensitive information.

1.2.2. Optical Spectroscopy. Optical spectroscopy is a strong technique to monitor molecular events at catalyst particles. The energy transitions that are probed are molecule-speciﬁc, and thereby reactants, products and any intermediates can be distin- guished. The maximum resolution of optical spectroscopy is determined by the diffraction limit of light. This was ﬁrst described by Ernst Abbe in 1873, 4 and is dependent on the wavelength of light, λ, and the numerical aperture (NA), the angle of focussing:

λ d (1.1) = 2NA This typically results in a maximum resolution of hundreds of nanometers (see Table 1.1). To bring this in the context of heterogeneous catalysis: a single catalytic particle on a support is in the size range of a few nanometers. This is a large mismatch in length scale for the study of these supported metal(oxide) particles. Optical spectroscopy does yield molecular information, and a schematic energy diagram of the probed transitions is shown in Fig. 1.3. The electronic states (S0,S1,T1) each have their unique set of vibrational levels (ν0, ν1, ν2). Absorption spectroscopy probes the transitions from the electronic ground state S0 to excited states S1 and higher (not shown). After a molecule has been excited to a higher energy level, there are many routes back to the ground state, most of which are non-radiative. There are two exceptions that can be used to probe the state of the molecule: ﬂuorescence and phosphorescence.

4 Reaction Monitoring In-Situ

S ISC 1 T

Energy 1 Fluorescence Absorption Phosphorescence

ν ν2 ν1 0 S0

FIG. 1.3. Optical processes involved in electronic spectroscopy. S0 is the ground electronic state of the molecule, S1 the ﬁrst excited state. Absorption of photons of the right energy can cause this excitation, subsequent radiative decay from this excited state back to S0 is called ﬂuorescence. Alternatively, it is also possible that intersystem crossing (ICS) takes place, through this process triplet states (here T1) can be reached. The resulting radiative decay towards the S0 state is called phosphorescence.

1.2.2.1. Absorption. Measurement of absorption yields information on the electronic levels of the analyte. Typically measurements are conducted between 200 and 800 1 nm (6.2 - 1.55 eV or 50 000 - 12 500 cm− ), spanning both the ultraviolet (UV) spectrum (200 - 400 nm) and visible (VIS) spectrum (400 - 800 nm). Therefore, this methodology is nicknamed UV-VIS (absorption) spectroscopy, or UV-VIS in short. UV-VIS spectroscopy is well-known in the field of process technology, and can be used to track colour changes in material or chemicals in time at a specific point in a production line. This is typically done in either reflectance or transmission mode. 5–7 Alternatively, UV- VIS spectroscopy can be used for microscopy imaging. This is typically used in reflectance mode, with an upright microscope setup. 8–10 Typically two types of materials are studied with UV-VIS in the field of heterogeneous catalysis. Transition metals and their ions are coloured due to clear d-d optical transitions, that are strongly dependent on the direct surroundings of the metal, and especially sensitive to the nature of ligands to a metal ion. 8,11–16 The other group of well-studied materials are molecular aromatic ring structures. Their π-π* transitions are strong optical absorbers and can help to illucidate catalyst deactivation. 17–25 1.2.2.2. Emission. Fluorescence is the radiative emission from excited energy levels

(e.g. S1) to the ground state, and ﬂuoresence spectroscopy is a common characterization technique. However, it does compete with non-radiative decay routes, and commonly,

5 Chapter 1: General Introduction laser excitation is used to obtain sufficient signal. The technique is mostly used in funda- mental studies, making use of dedicated microscope setups. 11,26 Coke formation, a general catalyst deactivation, can be studied using fluorescence. 22,27–30 Alternatively, the addition of higly fluorescent probe-molecules can be utilized to efficiently visualize e.g. catalytic activity of particles, 26,31–37 up to single-molecule levels. 38–47 Recent years have seen the development of various novel fluorescence techniques with super-resolution imaging capabilities, typically in the range of 30-50 nm. Stimulated emission depletion microscopy (STED) is based on non-linear fluorescence spectroscopy, and requires extremely photo-stable fluorescent probes. 48–51 Other techniques, that include stochastical optical reconstruction microscopy (STORM) and photo- activated localization microscopy (PALM), are based on single-event fluorescence. 52–55 The fluorescence of single molecules can be localized to pinpoint its position using point- spread functions. Some of these techniques have already been applied in catalysis. 47,56 However, all these techniques require fluorescent probes, and are thereby inherently limited in the range of reactions that can be studied. The practical limitations of fluorescence spectroscopy to wider use in catalysis are not only due to the effect of higher non-radiative decay at higher temperatures, but also because probe-molecules can interfere with the reactivity. After excitation of electrons to excited states, inter-system crossing (ISC) can occur from an excited singlet to triplet state (T1), involving a spin flip of the excited state. Sub- sequent radiative decay to the ground state (S0) is called phosphoresence, and typically 5 persists much longer than fluoresence due to the longer lifetime of the T1 state. Phos- phorescence is not a commonly used technique, as it is normally only a likely route for a return to the ground state at temperatures below 0 ◦C. At reaction temperatures in catalysis, phosphorescence is an unlikely decay path.

1.2.3. Vibrational Spectroscopy. The sub-levels in both the ground and excited state mentioned in Fig. 1.3 (ν0, ν1, ν2 etc.) depict the vibration modes of the molecule. In contrast to fluorescence and UV-VIS spectroscopies, vibrational spectroscopy is molecular specific, up to a fingerprint-level. The excitation of the molecule results in stretching, bending, of other vibrational deformations. There are a variety of spectroscopic techniques and phenomena related to the characterization of the vibrational modes of a molecule, as shown in Fig. 1.4. Infrared spectroscopy directly probes the vibrational transitions, whereas Raman scattering is an indirect method to probe the vibrational modes of a molecule. 1.2.3.1. Infrared Spectroscopy. Infrared (IR) (absorption) spectroscopy probes the transitions between the vibrational modes in the electronic ground state (illustrated in

6 Reaction Monitoring In-Situ Energy

ν2 ν1 ν0 anti-Stokes resonance Infrared Rayleigh Raman Raman Raman

FIG. 1.4. Schematic version of a Jablonski diagram for the mechanisms of infrared spectroscopy, Rayleigh scattering, Raman scattering, anti-Stokes Raman scattering and resonance Raman scattering.

Fig. 1.4). These vibrational modes are molecule-specific, and the probed transitions give a fingerprint spectrum related to the functionalities of a molecule. The selection rules of IR spectroscopy dictate that vibrational modes require a change in electronic dipole moment for IR-activity. The complete IR spectrum of a molecule is unique to the structure of the molecules. This makes IR spectroscopy a very powerful molecular characterization technique. IR spectroscopy is used in various stages in the catalyst lifetime. Amongst other applications, IR spectroscopy makes it possible to identify the binding sites and reactive gasses, 57–59 to detect reaction product formation 12,60, and to investigate the nature or compositions of the catalytic material. 61–63 Several forms of IR spectroscopy are used, either based on measuring in transmission or reflectance mode. Transmission IR spectroscopy can be used for bulk measurements through a (self-supporting) catalyst material in a reaction cell. In reflection mode, various geometries are used, the most common of which is attenuated total reflectance IR spectroscopy (ATR-IR). In ATR-IR, the evanescent field of a reflected IR beam is used to probe the outer micrometers of a reaction mixture or catalyst material. As the IR beam does not have to pass through the sample, it is easily incorporated in on-line measurements. Diffuse reflectance infrared fourier transform spectroscopy (DRIFTS) can also be used for in-situ measurements of catalyst material, as elliptical mirrors collect all light reflecting

7 Chapter 1: General Introduction off the powdered samples. For flat reflective samples, used in e.g. surface science studies, reflective absorption IR spectroscopy (RAIRS) is used. 7,64,65 Finally, IR spectroscopy is easily incorporated into multi-probe in-situ experiments, combining several techniques for a better understanding of the processes that occur within a reactor. 6,59 All techniques and studies described above, result in bulk characterization, without spatial information. For IR microscopy, typically a brilliant IR source is required, as found in synchrotrons. 66 IR microscopy has been used to study the diffusion of gas through porous structures 67,68, as well as the chemical reactivity over catalytic materials. 69,70 The resolution of IR microscopy as determined by the diffraction limit of light is in the order of 2.5 to 15 µm (see Table 1.1), depending on the exact vibration probed. This is much larger than the size of single catalytic particles. 1.2.3.2. Raman Spectroscopy. Complementary to IR spectroscopy is the technique of Raman spectroscopy. It is an indirect method as it uses inelastic scattering to probe vibrational information. Scattering is by definition an instantaneous technique. A molecule is excited by a laser to a virtual excited state and without delay, the molecule re- turns to its electronic ground state under scattering of a photon. Most frequently, the scattered photon has the same energy as the excitation photon, known as elastic scattering or Rayleigh scattering (illustrated in Fig. 1.4). However, a small fraction of the excited molecules will not return to the ground vibrational level, but to an excited vibrational level within the electronic ground state (see Fig. 1.4 - Raman). In other words, some of the scattered light has undergone an energy loss that the molecule has used for vibrational excitation (similar to what is observed in IR spectroscopy). This inelastic scattering process is called Raman scattering. The selection rules for Raman spectroscopy dictate that only vibrational transitions correlated to a change in polarizability are observed. In practice, the most intense vibrations are those that hardly have a dipole moment, e.g. C=C stretching vibrations. Raman is therefore a complementary technique to IR spectroscopy. A combination of both spectroscopy techniques will show the full set of vibrational modes. There are two types of Raman scattering, Stokes and anti-Stokes Raman scattering. Stokes scattering is red-shifted (lower in energy) with respect to the laser line, whereas anti-Stokes scattering is blue-shifted (higher in energy). The Boltzmann distribution function describes the distribution of molecules in excited (vibrational) states, depending on the temperature. At room temperature, some molecules will be at an excited vibrational state within the electronic ground state. These molecules are further excited by the in- coming excitation photons, but return to the ground electronic (and vibrational) state, as illustrated in Fig. 1.4 - anti-Stokes Raman. In most circumstances, the ground vibrational

8 Reaction Monitoring In-Situ state will be the highest populated state, which makes Stokes Raman the most intense and commonly used mode. 71 Raman measurements in heterogeneous catalysis are usually done in reflection mode, as most samples are non-transparent. Bulk measurements can be executed using e.g. optical probes that excite and collect through the same set of optical fibres, 72 but micro- scopic studies have also been reported. 73,74 Recently, several new methodologies have been developed and applied to catalysis. Coherent anti-stokes Raman scattering (CARS) enables higher spatial resolution and chemical sensitivity with respect to normal Raman microscopy. 69,75 Diagonally offset Raman scattering (DORS) allows Raman tomography (mapping) without the need for sectioning a sample. 76 Catalytic studies with Raman spectroscopy are manyfold and have been thoroughly described in reviews. 77–79 Amongst the materials that are typically followed are metal oxides, 61,80 zeolites, 81 and carbonaceous species formed during catalyst deactivation. 72 1.2.3.3. Challenges for Raman Spectroscopy. Raman spectroscopy is one of the most promising techniques for in-situ characterization of catalytic processes. It is easy to incorporate in in-situ setups, as it can be performed through both glass and water. Re- mote application is possible because of the nature of the optical technique, allowing the use of aggressive atmospheres and high temperatures during a reaction. However, some challenges still remain for Raman spectroscopy for the elucidation of catalytic reactions, which will be addressed in this PhD Thesis. Spatial Resolution - As discussed above, the spatial resolution of optical spectroscopy is dictated by the diffraction limit of light. In Raman spectroscopy, excitation lasers are typically in the visible wavelength range. As indicated in Table 1.1, the spatial resolution for Raman spectroscopy is thus anywhere between 150 and 500 nm. Coupling to in-situ cells typically requires long-distance, low NA objectives, resulting in a lower spatial resolution. The use of shorter wavelength is not a solution, as the related energies (e.g. x-rays) are too high to probe the molecular-sensitive energy transitions. However, compared to IR spectroscopy, the resolution of Raman spectroscopy is more than an order of magnitude closer to the size-range of single-catalytic particles. The only viable alternative to improve this spatial resolution of Raman spectroscopy is via the use of non-linear or near-field optical phenomena. Small Scattering Intensity - The Raman cross-section is a measure for the likelihood that a photon is scattered by the analyte through Raman scattering. This cross-section is rather low, making Raman scattering an informative but weak process. Raman excitation is commonly to a so-called virtual state, which does not correspond to a real excited state of the analyte. In the case of excitation into a real excited electronic state, the Raman

9 Chapter 1: General Introduction cross-section becomes much larger, the so-called resonance-Raman. 78 Resonance Raman usually requires UV-excitation to reach the excited state. This greatly increases the chance of unwanted photo-reactions, resulting in damage of the sample. Fluorescence - Along with Raman scattering, laser excitation to a real excited state can also induce fluorescence. This is a major problem in Raman spectroscopy, as it can result in a broadband interference with the Raman signal. Whether or not fluorescence is a problem in resonance Raman spectroscopy, depends on the optical pathways in the analyte’s electronic structure. The fluorescence is usually caused by impurities in the sample, or side-reactions during catalysis. The two main differences between Raman and fluorescence are the lifetime, and the dependence on the excitation wavelength. Fluorescence is a photonic decay from an excited state to the ground state. Therefore, changing the excitation source to a higher wavelength or lower excitation energy will bypass the fluorescence effect. Alternatively, setting a time-gate with respect to the excitation process will separate these processes. 82,83 To get a decent Raman signal, however, long integration times and high excitation power will be required, as much of the signal will be disregarded.

1.3. Overcoming Challenges in Raman Spectroscopy

Overall, Raman spectroscopy is a promising approach for the in-situ monitoring of catalytic processes. Only three challenges remain to be resolved for Raman spectroscopy to become the ultimate tool for use in heterogeneous catalysis: the spatial resolution, the small scattering intensity, and ﬂuorescence. In this PhD Thesis, a single approach is used in a variety of ways, to tackle the three challenges mentioned above, respectively light on the nanoscale, surface-enhanced Raman scattering, and ﬂuorescence quenching.

1.3.1. Light on the Nanoscale. The diffraction limit of light only describes so-called far-field phenomena. Light can be focussed on a nanoscale, if one steps away from the classical far-field approach, governed by the diffraction limit, and into near-field phenomena. Optical imaging with a nanoscale resolution is possible, if either the light source or detection probe are within a wavelength’s distance from the sample. These kinds of near- field microscopy are typically done with optical fibres, but have a high signal loss in the coupling into or out of these optical fibres. An alternative approach remotely excites a near-field. Light can not only be described by particles (photons), but also by an electromagnetic field. The electrons confined within a metal can ‘feel’ this field, and consecutively act with it in a oscillation or resonance, a so-called plasmon (see Fig. 1.5a). With sufficiently small nanoparticles, these resonances occur in the range of visible light, resulting in gold and nanoparticles of various colours,

10 Overcoming Challenges in Raman Spectroscopy

a b iii iv ii v vi i vii

FIG. 1.5. a) Electrons within a nanoparticle resonate with the oscillating electro-magnetic field of light. This oscillation is called a surface plasmon. Image taken from Kelly et al. 85 b) Several samples of nanoparticles. From left to right: i) Au nanospheres, ii) and iii) Au nanorods of different aspect ratio, iv) and v) Ag nanoprisms of different aspect ratios, vi) and vii) Ag nanospheres of different sizes. as shown in Fig. 1.5b. The exact frequency (or colour) of this plasmon band depends on parameters like the shape, size or material of the particles. Even subtle changes like the chemisorption of molecules onto the metal surface can influence the plasmon resonance, and this in itself can be a method to study a catalytic reaction. 84 The oscillations of the electrons create an intense electromagnetic field at the ‘poles’ of the nanoparticles. This can be calculated using Mie scattering theory (see Fig. 1.6a). When two nanoparticles are close together, their plasmon modes can communicate, and form another mode - the so-called gap mode plasmon (see Fig. 1.6b). This gap-mode plasmon is much more intense than a single-particle plasmon, and is usually at lower energy, red-shifted, to the single nanoparticle plasmon mode. Scaling up to large-scale clusters, the gap-mode in chains of nanoparticles becomes the most-important plasmonic feature, and it is this plasmon band that causes a large Raman enhancement for molecules present in the near-field volume. 86,87 Fig. 1.6b indicates a new measurement volume of 10-100 cubic nanometer, which is the order of magnitude requires for reaction monitoring∼ over single catalytic particles.

1.3.2. Surface-Enhanced Raman Scattering. The nano-scale measurement volume created by the surface plasmon modes between two nanoparticles should still yield sufficient Raman signal. Fortunately, the localization of the electromagnetic field also results in a greatly enhanced Raman signal of molecules present therein. In 1974 Fleischmann et al. 89 first noticed an enhancement effect for Raman spectroscopy of pyridine on rough Ag

11 Chapter 1: General Introduction

a 10 b 100

1 1

FIG. 1.6. a) Electric ﬁeld around a Au nanospheren (30 nm) under 633 nm excitation (Mie scattering theory). b) Electric ﬁeld around a Au nanosphere dimer (both 30 nm) under 633 nm excitation. Both taken from Talley et al. 88

electrodes. Since then this effect, surface-enhanced Raman scattering (SERS), has gained a huge interest as analytical technique in e.g. (bio)medical applications, 90, forensics 91 and art history. 92 Typical SERS substrates include roughened electrodes, (clusters) of nanoparticles and annealed thin films. 93 Though all metal nanoparticles show plasmon excitation, the energy at which this occurs varies much with the actual metal. As Ra- man spectroscopy is typically measured at excitation wavelengths in the visible spectrum, Ag, Au and Cu are the most common SERS materials. Coupling between two different materials is also possible, and shifts the plasmon resonance to a different energy. The following description of the SERS enhancement effect is loosely based on an excellent textbook by Le Ru and Etchegoin. 71 1.3.2.1. Electromagnetic Enhancement. The gap plasmon modes as described above, create a large electromagnetic field at the junction of two nanoparticles. Effect- ively, the nanoparticles act as antennas or lenses to focus impeding light to these small volumes. Molecules present at the junction between nanoparticles will thereby ‘feel’ a much larger electromagnetic field than they would without the adjacent nanoparticles. Even though the molecular Raman absorption cross section will not have changed, the highly increased electromagnetic field causes a greatly increased Raman scattering intensity. The power of the Raman scattering P can be approximated with the following formula:

E2 P (1.2) 4 ∝ λ In practice, the power is inversely proportional to the quadruple power of the excitation wavelength (λ), and quadraticaly proportional to the magnitude or intensity of the

12 Overcoming Challenges in Raman Spectroscopy

FIG. 1.7. The charge-transfer mechanism. 71

electromagnetic field (E). As the SERS-effect locally increases the electromagentic field with a huge factor, the Raman effect is increased even more. Typical Raman enhancement factors can be expected in the range of 106 - 1011. The SERS-effect decays roughly with a factor R for molecules that are a distance x R+x away from the nanoparticle with radius R. 94 Therefore, molecules can only experience the enhanced field if they are within the first few nanometers from a nanoparticle (junction). The electromagnetic enhancement effect is a general effect observed with SERS, and is identical for all analytes. Of course, only those molecules present within the electromagnetic near-field experience the electromagnetic enhancement effect. 1.3.2.2. Chemical Enhancement Effects. A large contribution to the SERS effect is the so-called ‘chemical enhancement’. It is an umbrella term for all effects that are molecule-specific, e.g. the chemical interaction between the analyte and nanoparticles, as well as optical properties of said analyte. The charge-transfer mechanism is similar to resonance Raman, but makes use of the electronic level of the metal nanoparticles. It is shown schematically in Fig. 1.7. The electrons within the metal are excited, or heated up, under laser excitation. If this excitation energy on top of the Fermi-energy of the metal is sufficient to reach the excited state of the analyte present on the metal, charge-transfer occurs to the molecule. Due to the increased Raman absorption cross section at excitation into the lowest unoccupied molecular orbital (LUMO) excited state, the Raman scattering intensity will be even further increased as a result of the electromagnetic enhancement.

13 Chapter 1: General Introduction

Chemical interactions between the SERS-active material and analyte of choice are crucial to yield a good SERS effect. This occurs most effectively if the analyte has a functional group that coordinates or binds to the metal surface, e.g. thiols or carboxylic acids. If this is not the case, one could develop a SERS substrate with a speciﬁc surface functionality, tailored to bind the desired analyte. This technique is inspired by the technique of surface plasmon resonance. 95,96

1.3.3. Fluorescence Quenching. Depending on the optical pathway specific to an analyte, it could happen that the fluorescence is increased in a similar fashion as the Raman signal. This effect is called surface-enhanced fluorescence. 97,98 This is an analyte- specific problem, and can be avoided by avoiding the resonance excitation via e.g. a longer excitation wavelength or a different SERS-active material.

1.4. SERS in Catalysis

SERS has great prospects in the ﬁeld of catalysis, to monitor those processes that take place on catalytic surfaces. Several reviews have covered this topic. 78,99,100 Combining SERS and catalysis in a single experiment has been accomplished in various experimental set-ups. As SERS was originally discovered on Ag electrodes, it is no sur- prise that the earliest reactions studied were electrochemical in nature. These early studies did not use the Ag surface of the electrodes as catalytic surface, but instead thin layers of catalytically active transition metals were deposited onto the electrode surfaces. 101 Thereby the SERS activity and catalytic activity were separated from each other. This approach was later extended to gas-ﬂow systems, for monitoring non-electrochemical re- 102–106 actions at temperatures up to 500 ◦C and pressures up to 1 atm. Others have taken different approaches to combining catalysis and SERS. Nano-roughened transition metal surfaces do also show (weak) SERS-enhancement, and can be used directly as both catalyst and enhancement material. 107–111 Alternatively, small catalytic particles have been deposited on top of SERS-active substrates. 112–115 Similar approaches are used for application in homogeneous catalysis, by immobilization of metal complexes onto SERS substrates. 116 Nanoparticle-based SERS substrates are also used. They can either be attached to surfaces, 117 or brought directly into catalyst support materials. 118,119 Inert SERS particles, through a thin silica layer, can also be used as SERS probes on catalytic materials. 120,121 As with the electrodes, advanced synthesis procedures also allow for more intricate nanoparticles like core-shell geometries, combining SERS-activity and catalysis in a single nanoparticle. 122–127 Even more sophisticated nanoparticles add small satellites of Au on top of that, to incorporate the gap-mode plasmon mode within a single nanoparticle. 128,129

14 SERS in Catalysis a) b)

detector

FIG. 1.8. a) Diagram of a scanning tunneling microscope. A voltage bias between tip and sample is used as feedback mechanism, measuring the electric tunnelling current as function of location on the sample. b) Diagram of an atomic force microscope. A laser feedback system is used to monitor the van der Waals interactions between tip and sample.

Others simply mix SERS-active nanoparticles with (smaller) catalytic particles and thereby follow reactions on the catalytic particles. 130 Many catalytic processes have been investigated with SERS, but in general no step is taken after observation of a reaction. SERS should be applicable for reaction elucidation, more quantitative characterization, and ultimately, the study of individual nanoscale active sites. The studies described in this PhD Thesis aim to show that all of these are possible.

1.4.1. Scanning Probe Microscopy. Reaction monitoring of single catalytic particles can be achieved via SERS, but requires an additional characterization technique to identify, localize, and characterize the morphology features over a surface with catalytic particles. Scanning probe microscopy (SPM) is the method of choice, as it provides nano-scale morphological information. The two most common SPM techniques are atomic force microscopy (AFM) and scanning tunnelling microscopy (STM). Many more variations exist in the family of techniques in SPM, each probing a different surface property. 131 First discovered by IBM in 1981 is STM, 132 based on the electron tunneling between the tip of a conducting (e.g. tungsten) tip and sample when a voltage bias is applied. A schematic diagram of an STM is shown in Fig. 1.8a. As the electron current is measured, samples are required to be(semi-)conductors. Later, the same group invented AFM for imaging non-conductive samples. 133 A schematic diagram of AFM is shown in Fig. 1.8b. The governing interactions between (silicon or silicon nitride) tip and sample are van der Waals forces, so AFM traces the morphology of

15 Chapter 1: General Introduction a surface. A laser-feedback system is used to monitor the tip-sample interactions. In most cases, the tip is excited to oscillate at its resonance frequency, to limit the actual contact between tip and sample. This limits the wear of the tip and avoids damage to the sample. In both AFM and STM imaging of the sample is achieved via piezo-scanners that can be attached to the tip and sample. These allow precise control of the location of the tip with respect to the sample, and ultimately result in the high resolution of either technique. A typical resolution for STM is 0.1 nm, due to the high radial decay of the tunnelling current. For AFM, the x-y resolution is highly dependent on the shape and size of the tip apex (5-30 nm). The height of surface features can typically be resolved to <0.2 nm. However, highly specialized AFM has also shown to be capable of imaging single molecules. 134 AFM and STM require samples to be relatively ﬂat (within 1 µm over a 20x20 µm area), so application in heterogeneous catalysis is normally limited to model samples. However, single catalytic particles ( 1-5 nm) can easily be located and imaged over a ﬂat support material, and coupling to Raman∼ and other optical spectroscopies can easily be integrated through transparent samples, from the side of the tip, or from above, if special tilted tips are used. 135

1.4.2. Tip-Enhanced Raman Scattering. SERS has the enhancement effect to monitor single catalytic particles, if a gap-mode plasmon is present, but requires an additional technique, like SPM, to couple Raman information to morphological information of catalytic particles. Ultimately, it would be best to combine SPM and SERS in a single technique, as done in tip-enhanced Raman scattering (TERS). This approach was first mentioned in 1928, 136 but only developed in 2000, by three individual groups. 137–139 The development of the field since then has been covered in various reviews. 140–144 The most effective enhancement is obtained when the tip and a flat metal (Au of Ag) couple to form a gap-mode. 145 Both AFM and STM are techniques are scanning probe microscopy (SPM) techniques, based on thin needles tracing the surface morphology on the nanoscale. In practice, AFM-tips for TERS are typically coated with a thin Ag or Au layer for TERS measurements, while in STM-TERS, Ag wires are often used. This makes TERS a single-hotspot technique, resulting in a spatial resolution in the order of 10 nm in diameter or smaller, dependent on the tip diameter (Fig. 1.9). 145 This superresolution in combination with the possibility to position on an area of interest makes TERS an ideal technique for molecular characterization in the single-molecule regime, 146,147 even with the intensity and spectral fluctuations characteristic to the single-molecule regime. 148–156

16 Scope of This PhD Thesis

a) b)

FIG. 1.9. a) Electromagnetic ﬁeld simulation of a Au tip over a Au substrate, excitement at 633 nm. M is the maximum ﬁeld enhancement in this simulation. Taken from Yang et al. 145 b) Several possible TERS geometries. Taken from Blum et al. 157

As a surface-science technique, TERS shows great promise in the field of catalysis. 77,78 Metal-ligand interactions have already been studied as a first step to application in catalysis. 158,159 Recently, the first reactions have also been observed with TERS. 160,161 TERS-setups come in various flavours, of which Fig. 1.9 shows a small set of options. The most important differentiation between these different setups is the optical path. The easiest configuration to optimize uses a tip-approach from the top of the sample and optical alignment from below. This bottom-illumination setup can only be applied for transparent samples. In the case of top-illumination setups, the Raman excitation laser does not need to pass through the sample, but optical access to the tip is more difficult to achieve. Differences between experimental setups, like the angle of plasmon excitation, excitation wavelength, or the quality of the TERS-tips used, all influence the outcome of experiments. 157 This means that significant prior knowledge of the catalytic system under study is required to obtain useful results concerning reactivity and pathway on a nanoscale. TERS, but also SERS, is sensitive to a multitude of experimental parameters, which all have to be taken into consideration. Further development of the field will show how the sensitivity to experimental parameters will be handled in the future.

1.5. Scope of This PhD Thesis

Raman has many properties that are of great advantage to reaction monitoring in heterogeneous catalysis. SERS and TERS are especially suited to gain insight into the

17 Chapter 1: General Introduction functioning of single catalytic particles with respect to their bulk properties. In this PhD thesis, the capabilities of SERS and TERS to heterogeneous catalysis are expanded to obtain information on as small a scale as possible. The photo-catalytic reduction of p-nitrothiophenol is chosen as a model reaction to show the capabilities of SERS and TERS in catalytic studies. In Chapter 2, the literature relevant to this reaction is discussed and SERS data are collected for later interpretation of the experiments described in this PhD Thesis. Chapter 3 shows that SERS can be used for quantification of reaction kinetics within a two-dimensional system. Photo-reduction of p-nitrothiophenol is performed within a self-assembled monolayer on the surface of a SERS substrate. Application of a newly developed kinetic model proves that the reduction is a second-order reaction, settling a long-standing debate in literature. Scaling down to single-hotspots, or nano-scale sampling volumes, results in highly fluctuating SERS spectra in time, as a result of the limited number of molecules probed. Chapter 4 describes a chemometric method to separate catalytic conversion from short- term fluctuations. This allows for simultaneous analysis of reaction kinetic and identification of spurious orientation events or the presence of reaction intermediates. Finally, TERS is used in Chapter 5 in the smallest measurement volume possible. The reduction of pNTP in-situ, is measured over a single catalytic particle, the TERS tip. A dual-wavelength approach, for separation of reaction and monitoring, proved to be essential to observe any reaction dynamics. Subsequent chemometric analysis is used to separate the long-term trends from short-term fluctuations, and hints to the identity of reaction intermediates. A summary of this PhD thesis, followed by a translation in Dutch, can be found in Chapters 6 and 7, with an outlook to possible future steps for this line of research.

References

[1] I. Chorkendorff and J. W.Niemantsverdriet, Concepts of Modern Catalysis and Kinetics, Wiley-VCH, Weinheim, Germany, 2003. [2] J. W. Niemantsverdriet, Spectroscopy in Catalysis, Wiley-VCH, Weinheim, Germany, 2007. [3] B. M. Weckhuysen, Chem. Soc. Rev., 2010, 39, 4557–4559. [4] E. Abbe, Arch. für Mikroskopische Anat., 1873, 9, 413–418. [5] Handbook of Spectroscopy, ed. G. Gauglitz and D. S. Moore, Wiley-VCH, Weinheim, Germany, 2014. [6] U. Bentrup, Chem. Soc. Rev., 2010, 39, 4718–4730.

18 References

[7] F. C. Meunier, Chem. Soc. Rev., 2010, 39, 4602–4614. [8] R. A. Schoonheydt, Chem. Soc. Rev., 2010, 39, 5051–5066. [9] L. Espinosa-Alonso, A. M. Beale and B. M. Weckhuysen, Acc. Chem. Res., 2010, 43, 1279–1288. [10] F. C. Jentoft, Adv. Cat., 2009, 52, 129–211. [11] B. M. Weckhuysen, Angew. Chem. Int. Ed., 2009, 48, 4910–4943. [12] T. Burgi, J. Catal., 2005, 229, 55–63. [13] T. A. Nijhuis, S. J. Tinnemans, T. Visser and B. M. Weckhuysen, Phys. Chem. Chem. Phys., 2003, 5, 4361–4365. [14] G. Guilera, M. A. Newton, C. Polli, S. Pascarelli, M. Guinó, K. K. Hii and K. K. Hii, Chem. Commun., 2006, 4306–4308. [15] U. Bentrup, J. Radnik, U. Armbruster, A. Martin, J. Leiterer, F. Emmerling and A. Brückner, Top. Catal., 2009, 52, 1350–1359. [16] A. Brückner and E. Kondratenko, Catal. Today, 2006, 113, 16–24. [17] M. H. F. Kox, E. Stavitski and B. M. Weckhuysen, Angew. Chem. Int. Ed., 2007, 46, 3652–3655. [18] M. H. F. Kox, E. Stavitski, J. C. Groen, J. Pérez-Ramírez, F. Kapteijn and B. M. Weck- huysen, Chem. Eur. J., 2008, 14, 1718–1725. [19] L. R. Aramburo, L. Karwacki, P.Cubillas, S. Asahina, D. a. M. de Winter, M. R. Drury, I. L. C. Buurmans, E. Stavitski, D. Mores, M. Daturi, P.Bazin, P.Dumas, F. Thibault- Starzyk, J. a. Post, M. W. Anderson, O. Terasaki and B. M. Weckhuysen, Chem. Eur. J., 2011, 17, 13773–13781. [20] E. Stavitski, M. H. F.Kox and B. M. Weckhuysen, Chem. Eur. J., 2007, 13, 7057–7065. [21] J. P.Hofmann, D. Mores, L. R. Aramburo, S. Teketel, M. Rohnke, J. Janek, U. Olsbye and B. M. Weckhuysen, Chem. Eur. J., 2013, 19, 8533–8542. [22] D. Mores, J. Kornatowski, U. Olsbye and B. M. Weckhuysen, Chem. Eur. J., 2011, 17, 2874–2884. [23] M. Hunger and W. Wang, Chem. Commun., 2004, 584–585. [24] J. J. H. B. Sattler, I. D. González-Jiménez, A. M. Mens, M. Arias, T. Visser and B. M. Weckhuysen, Chem. Commun., 2013, 49, 1518–1520. [25] Q. Qian, D. Mores, J. Kornatowski and B. M. Weckhuysen, Microporous Mesoporous Mater., 2011, 146, 28–35. [26] G. De Cremer, B. F. Sels, D. E. De Vos, J. Hofkens and M. B. J. Roeffaers, Chem. Soc. Rev., 2010, 39, 4703–4717. [27] D. Mores, E. Stavitski, M. H. F.Kox, J. Kornatowski, U. Olsbye and B. M. Weckhuysen, Chem. Eur. J., 2008, 14, 11320–11327.

19 Chapter 1: General Introduction

[28] Q. Qian, J. Ruiz-Martínez, M. Mokhtar, A. M. Asiri, S. A. Al-Thabaiti, S. N. Basahel, H. E. van der Bij, J. Kornatowski and B. M. Weckhuysen, Chem. Eur. J., 2013, 19, 11204–11215. [29] A. N. Parvulescu, D. Mores, E. Stavitski, C. M. Teodorescu, P.C. A. Bruijnincx, R. J. M. K. Gebbink and B. M. Weckhuysen, J. Am. Chem. Soc., 2010, 132, 10429–10439. [30] L. Karwacki, E. Stavitski, M. H. F.Kox, J. Kornatowski and B. M. Weckhuysen, Angew. Chem. Int. Ed., 2007, 46, 7228–7231. [31] I. L. C. Buurmans and B. M. Weckhuysen, Nat. Chem., 2012, 4, 873–886. [32] I. L. C. Buurmans, J. Ruiz-Martínez, W. V.Knowles, D. van der Beek, J. A. Bergwerff, E. T. C. Vogt and B. M. Weckhuysen, Nat. Chem., 2011, 3, 862–867. [33] M. B. J. Roeffaers, R. Ameloot, M. Baruah, H. Uji-I, M. Bulut, G. De Cremer, U. Müller, P. A. Jacobs, J. Hofkens, B. F. Sels and D. E. De Vos, J. Am. Chem. Soc., 2008, 130, 5763–5772. [34] I. L. C. Buurmans, J. Ruiz-Martínez, S. L. van Leeuwen, D. van der Beek, J. a. Bergwerff, W. V. Knowles, E. T. C. Vogt and B. M. Weckhuysen, Chem. Eur. J., 2012, 18, 1094–1101. [35] M. A. Karreman, I. L. C. Buurmans, J. W. Geus, A. V. Agronskaia, J. Ruiz-Martínez, H. C. Gerritsen and B. M. Weckhuysen, Angew. Chem. Int. Ed., 2012, 51, 1428–1431. [36] M. B. J. Roeffaers, R. Ameloot, A.-J. Bons, W. Mortier, G. De Cremer, R. de Kloe, J. Hofkens, D. E. De Vos and B. F. Sels, J. Am. Chem. Soc., 2008, 130, 13516–13517. [37] M. B. J. Roeffaers, B. F. Sels, H. Uji-i, B. Blanpain, P. L’hoëst, P. a. Jacobs, F. C. De Schryver, J. Hofkens and D. E. De Vos, Angew. Chem. Int. Ed., 2007, 46, 1706–1709. [38] T. Cordes and S. A. Blum, Nat. Chem., 2013, 5, 993–999. [39] M. B. J. Roeffaers, G. De Cremer, H. Uji-i, B. Muls, B. F. Sels, P. A. Jacobs, F. C. De Schryver, D. E. De Vos and J. Hofkens, Proc. Natl. Acad. Sci. U. S. A., 2007, 104, 12603–12609. [40] M. B. J. Roeffaers, B. F. Sels, H. Uji-I, F. C. De Schryver, P. A. Jacobs, D. E. De Vos and J. Hofkens, Nature, 2006, 439, 572–575. [41] T. Tachikawa and T. Majima, Langmuir, 2012, 28, 8933–8943. [42] T. Tachikawa, T. Yonezawa and T. Majima, ACS Nano, 2013, 7, 263–275. [43] J. Kirstein, B. Platschek, C. Jung, R. Brown, T. Bein and C. Bräuchle, Nat. Mater., 2007, 6, 303–310. [44] P.Chen, X. Zhou, H. Shen, N. M. Andoy, E. Choudhary, K.-S. Han, G. Liu and W.Meng, Chem. Soc. Rev., 2010, 39, 4560–4570. [45] W. Xu, J. S. Kong, Y.-T. E. Yeh and P.Chen, Nat. Mater., 2008, 7, 992–996. [46] X. Zhou, W.Xu, G. Liu, D. Panda and P.Chen, J. Am. Chem. Soc., 2010, 132, 138–146.

20 References

[47] X. Zhou, N. M. Andoy, G. Liu, E. Choudhary, K.-S. Han, H. Shen and P. Chen, Nat. Nanotechnol., 2012, 7, 237–241. [48] S. W. Hell, Science, 2007, 316, 1153–1158. [49] T. A. Klar, S. Jakobs, M. Dyba, A. Egner and S. W. Hell, Proc. Natl. Acad. Sci. U. S. A., 2000, 97, 8206–8210. [50] T. A. Klar and S. W. Hell, Opt. Lett., 1999, 24, 954–956. [51] S. W. Hell and J. Wichmann, Opt. Lett., 1994, 19, 780–782. [52] M. Bates, B. Huang, G. T. Dempsey and X. Zhuang, Science, 2007, 317, 1749–1753. [53] M. J. Rust, M. Bates and X. Zhuang, Nat. Methods, 2006, 3, 793–795. [54] S. T. Hess, T. J. Gould, M. V. Gudheti, S. A. Maas, K. D. Mills and J. Zimmerberg, Proc. Natl. Acad. Sci. U. S. A., 2007, 104, 17370–17375. [55] S. T. Hess, T. P.K. Girirajan and M. D. Mason, Biophys. J., 2006, 91, 4258–4272. [56] M. B. J. Roeffaers, G. De Cremer, J. Libeert, R. Ameloot, P. Dedecker, A.-J. Bons, M. Bückins, J. A. Martens, B. F. Sels, D. E. De Vos and J. Hofkens, Angew. Chem. Int. Ed., 2009, 48, 9285–9289. [57] C. Lamberti, A. Zecchina, E. Groppo and S. Bordiga, Chem. Soc. Rev., 2010, 39, 4951–5001. [58] A. Urakawa, N. Maeda and A. Baiker, Angew. Chem. Int. Ed., 2008, 47, 9256–9259. [59] M. A. Newton and W. van Beek, Chem. Soc. Rev., 2010, 39, 4845–4863. [60] L. R. Knöpke, N. Nemati, A. Köckritz, A. Brückner and U. Bentrup, ChemCatChem, 2010, 2, 273–280. [61] I. E. Wachs, Catal. Today, 1996, 27, 437–455. [62] P. Munnik, M. Wolters, A. Gabrielsson, S. D. Pollington, G. Headdock, J. H. Bitter, P.E. de Jongh and K. P.de Jong, J. Phys. Chem. C, 2011, 115, 14698–14706. [63] M. Wolters, P.Munnik, J. H. Bitter, P.E. de Jongh and K. P.de Jong, J. Phys. Chem. C, 2011, 115, 3332–3339. [64] F. Zaera, Chem. Soc. Rev., 2014, doi:10.1039/c3cs60374a. [65] J. Ryczkowski, Catal. Today, 2001, 68, 263–381. [66] A. M. Beale, S. D. M. Jacques and B. M. Weckhuysen, Chem. Soc. Rev., 2010, 39, 4656–4672. [67] C. Chmelik, F.Hibbe, D. Tzoulaki, L. Heinke, J. Caro, J. Li and J. Kärger, Microporous Mesoporous Mater., 2010, 129, 340–344. [68] C. Chmelik, J. Kärger, M. Wiebcke, J. Caro, J. van Baten and R. Krishna, Microporous Mesoporous Mater., 2009, 117, 22–32. [69] M. H. F. Kox, K. F. Domke, J. P. R. Day, G. Rago, E. Stavitski, M. Bonn and B. M. Weckhuysen, Angew. Chem. Int. Ed., 2009, 48, 8990–8994.

21 Chapter 1: General Introduction

[70] E. Stavitski, M. H. F. Kox, I. Swart, F. M. F. de Groot and B. M. Weckhuysen, Angew. Chem. Int. Ed., 2008, 47, 3543–3547. [71] E. C. Le Ru and P.G. Etchegoin, Principles of Surface-Enhanced Raman Spectroscopy and related plasmonic effects, Elsevier, Amsterdam, The Netherlands, 1st edn., 2009. [72] J. J. H. B. Sattler, A. M. Beale and B. M. Weckhuysen, Phys. Chem. Chem. Phys., 2013, 15, 12095–12103. [73] J. A. Bergwerff, T. Visser, B. R. G. Leliveld, B. D. Rossenaar, K. P.de Jong and B. M. Weckhuysen, J. Am. Chem. Soc., 2004, 126, 14548–14556. [74] N. C. Jeong, H. Lim, H. Cheong and K. B. Yoon, Angew. Chem. Int. Ed., 2011, 50, 8697–8701. [75] K. F. Domke, T. A. Riemer, G. Rago, A. N. Parvulescu, P.C. A. Bruijnincx, A. Enejder, B. M. Weckhuysen and M. Bonn, J. Am. Chem. Soc., 2012, 134, 1124–1129. [76] M. W. Zandbergen, S. D. M. Jacques, B. M. Weckhuysen and A. M. Beale, Angew. Chem. Int. Ed., 2012, 51, 957–960. [77] V.Calvino-Casilda and M. A. Ban, Catalysis, 2012, 24, 1–47. [78] H. Kim, K. M. Kosuda, R. P. Van Duyne and P. C. Stair, Chem. Soc. Rev., 2010, 39, 4820–4844. [79] I. E. Wachs and F. D. Hardcastle, Catalysis, 1993, 10, 102–153. [80] I. E. Wachs and C. A. Roberts, Chem. Soc. Rev., 2010, 39, 5002–5017. [81] P.-P. Knops-Gerrits, D. E. De Vos, E. J. Feijen and P. A. Jacobs, Microporous Mater., 1997, 8, 3–17. [82] P. Matousek, M. Towrie, A. Stanley and A. W. Parker, Appl. Spectrosc., 1999, 53, 1485–1489. [83] C. E. Harvey, I. E. Iping Petterson, B. M. Weckhuysen, C. Gooijer, F.Ariese and A. J. G. Mank, Appl. Spectrosc., 2012, 66, 1179–1185. [84] C. Novo, A. M. Funston and P.Mulvaney, Nat. Nanotechnol., 2008, 3, 598–602. [85] K. L. Kelly, E. Coronado, L. L. Zhao and G. C. Schatz, J. Phys. Chem. B, 2003, 107, 668–677. [86] R. W. Taylor, T.-C. Lee, O. A. Scherman, R. Esteban, J. Aizpurua, F. M. Huang, J. J. Baumberg and S. Mahajan, ACS Nano, 2011, 5, 3878–3887. [87] R. Esteban, R. W.Taylor, J. J. Baumberg and J. Aizpurua, Langmuir, 2012, 28, 8881– 8890. [88] C. E. Talley, J. B. Jackson, C. Oubre, N. K. Grady, C. W. Hollars, S. M. Lane, T. R. Huser, P.Nordlander and N. J. Halas, Nano Lett., 2005, 5, 1569–1574. [89] M. Fleischmann, P.Hendra and A. McQuillan, Chem. Phys. Lett., 1974, 26, 163–166. [90] Y. Wang, B. Yan and L. Chen, Chem. Rev., 2013, 113, 1391–1428.

22 References

[91] E. L. Izake, Forensic Sci. Int., 2010, 202, 1–8. [92] F. Casadio, M. Leona, J. R. Lombardi and R. Van Duyne, Acc. Chem. Res., 2010, 43, 782–791. [93] G. McNay, D. Eustace, W. E. Smith, K. Faulds and D. Graham, Appl. Spectrosc., 2011, 65, 825–837. [94] R. Aroca, Surface-Enhanced Vibrational Spectroscopy, John Wiley & Sons, Chichester, UK, 2006. [95] K. M. Mayer and J. H. Hafner, Chem. Rev., 2011, 111, 3828–3857. [96] S. Zeng, D. Baillargeat, H.-P. Ho and K.-T. Yong, Chem. Soc. Rev., 2014, 43, 3426– 3452. [97] E. Fort and S. Grésillon, J. Phys. D. Appl. Phys., 2008, 41, 013001. [98] R. F. Aroca, G. Y. Teo, H. Mohan, A. R. Guerrero, P.Albella and F. Moreno, J. Phys. Chem. C, 2011, 115, 20419–20424. [99] C. E. Harvey, PhD Thesis, Utrecht University, 2013. [100] B. Ren, G.-K. Liu, X.-B. Lian, Z.-L. Yang and Z.-Q. Tian, Anal. Bioanal. Chem., 2007, 388, 29–45. [101] L. W. H. Leung and M. J. Weaver, J. Am. Chem. Soc., 1987, 109, 5113–5119. [102] C. T. Williams, H. Y. H. Chan, A. A. Tolia, M. J. Weaver and C. G. Takoudis, Ind. Eng. Chem. Res., 1998, 37, 2307–2315. [103] C. T. Williams, H. Y. H. Chan, A. A. Tolia, M. J. Weaver and C. G. Takoudis, Ind. Eng. Chem. Res., 1998, 37, 2307–2315. [104] C. T. Williams, K.-Y. Chen, C. G. Takoudis and M. J. Weaver, J. Phys. Chem. B, 1998, 102, 4785–4794. [105] C. T. Williams, C. G. Takoudis and M. J. Weaver, J. Phys. Chem. B, 1998, 102, 406–416. [106] C. T. Williams, C. A. Black, M. J. Weaver and C. G. Takoudis, J. Phys. Chem. B, 1997, 101, 2874–2883. [107] W. Cai, B. Ren, X. Li, C. She, F. Liu, X. Cai and Z.-Q. Tian, Surf. Sci., 1998, 406, 9–22. [108] Z.-Q. Tian, B. Ren and B. W. Mao, J. Phys. Chem. B, 1997, 101, 1338–1346. [109] B. Ren, X. Xu, X. Li, W. Cai and Z.-Q. Tian, Surf. Sci., 1999, 427-428, 157–161. [110] Q. Huang, X. Li, J. Yao, B. Ren, W. Cai, J. Gao, B. Mao and Z.-Q. Tian, Surf. Sci., 1999, 427-428, 162–166. [111] J. Yao, B. Ren, Z. Huang, P.Cao, R. Gu and Z.-Q. Tian, Electrochim. Acta, 2003, 48, 1263–1271.

23 Chapter 1: General Introduction

[112] J. Solla-Gullón, R. Gómez, A. Aldaz and J. Pérez, Electrochem. Commun., 2008, 10, 319–322. [113] F. J. E. Scheijen, G. L. Beltramo, S. Hoeppener, T. H. M. Housmans and M. T. M. Koper, J. Solid State Electrochem., 2007, 12, 483–495. [114] R. Gómez, J. Solla-Gullón, J. M. Pérez and A. Aldaz, J. Raman Spectrosc., 2005, 36, 613–622. [115] R. Gómez, J. M. Pérez, J. Solla-Gullón, V. Montiel and A. Aldaz, J. Phys. Chem. B, 2004, 108, 9943–9949. [116] O. Diaz-Morales, T. J. P.Hersbach, D. G. H. Hetterscheid, J. N. H. Reek and M. T. M. Koper, J. Am. Chem. Soc., 2014, 136, 10432–9. [117] L. Lu, G. Sun, H. Zhang, H. Wang, S. Xi, J. Hu, Z.-Q. Tian and R. Chen, J. Mater. Chem., 2004, 14, 1005–1009. [118] R. Silva, A. V.Biradar, L. Fabris and T. Asefa, J. Phys. Chem. C, 2011, 115, 22810– 22817. [119] Y. Zhu, D. Wang, L. Zhang, F. Sun, J. Xu, S. Jiang and Q. Yu, RSC Adv., 2013, 3, 10154–10157. [120] J.-F. Li, J. R. Anema, Y.-C. Yu, Z.-L. Yang, Y.-F. Huang, X.-S. Zhou, B. Ren and Z.-Q. Tian, Chem. Commun., 2011, 47, 2023–2025. [121] J. F. Li, Y.-F. Huang, Y. Ding, Z.-L. Yang, S. B. Li, X. S. Zhou, F. R. Fan, W. Zhang, Z. Y. Zhou, D.-Y. Wu, B. Ren, Z. L. Wang and Z.-Q. Tian, Nature, 2010, 464, 392–395. [122] R. Liu, J.-F. Liu, Z.-M. Zhang, L.-Q. Zhang, J.-F. Sun, M. Sun and G.-B. Jiang, J. Phys. Chem. Lett., 2014, 5, 969–975. [123] R. J. Taylor, Y. X. Jiang, N. V. Rees, G. A. Attard, E. L. Jeffery and D. J. Willock, J. Phys. Chem. C, 2011, 115, 21363–21372. [124] N. V. Rees, R. J. Taylor, Y. X. Jiang, I. R. Morgan, D. W. Knight and G. A. Attard, J. Phys. Chem. C, 2011, 115, 1163–1170. [125] P.Zhang, Y.-X. Chen, J. Cai, S.-Z. Liang, J.-F. Li, A. Wang, B. Ren and Z.-Q. Tian, J. Phys. Chem. C, 2009, 113, 17518–17526. [126] K. N. Heck, B. G. Janesko, G. E. Scuseria, N. J. Halas and M. S. Wong, J. Am. Chem. Soc., 2008, 130, 16592–16600. [127] J.-F.Li, Z.-L. Yang, B. Ren, G.-K. Liu, P.-P.Fang, Y.-X. Jiang, D.-Y. Wu and Z.-Q. Tian, Langmuir, 2006, 22, 10372–10379. [128] W. Xie, C. Herrmann, K. Kömpe, M. Haase and S. Schlücker, J. Am. Chem. Soc., 2011, 133, 19302–19305. [129] W. Xie, B. Walkenfort and S. Schlücker, J. Am. Chem. Soc., 2013, 135, 1657–1660.

24 References

[130] V. Joseph, C. Engelbrekt, J. Zhang, U. Gernert, J. Ulstrup and J. Kneipp, Angew. Chem. Int. Ed., 2012, 51, 7592–7596. [131] Y. Leng, Materials Characterization, Wiley-VCH, Weinheim, Germany, 2013. [132] G. Binnig, H. Rohrer, C. Gerber and E. Weibel, Phys. Rev. Lett., 1982, 49, 57–61. [133] G. Binnig and C. F. Quate, Phys. Rev. Lett., 1986, 56, 930–933. [134] L. Gross, F. Mohn, N. Moll, P. Liljeroth and G. Meyer, Science, 2009, 325, 1110– 1114. [135] C. E. Harvey, E. M. V. S. Lantman, A. J. G. Mank and B. M. Weckhuysen, Chem. Commun., 2012, 48, 1742–1744. [136] E. H. Synge, Phil. Mag., 1928, 35, 356–362. [137] M. S. Anderson, Appl. Phys. Lett., 2000, 76, 3130–3132. [138] N. Hayazawa, Y. Inouye, Z. Sekkat and S. Kawata, Opt. Commun., 2000, 183, 333– 336. [139] R. M. Stöckle, Y. D. Suh, V. Deckert and R. Zenobi, Chem. Phys. Lett., 2000, 318, 131–136. [140] T. Schmid, L. Opilik, C. Blum and R. Zenobi, Angew. Chem. Int. Ed., 2013, 52, 5940–5954. [141] S. Kawata, Appl. Spectrosc., 2013, 67, 117–125. [142] M. Richter and V. Deckert, in Surf. Thin Film Anal., ed. G. Friedbacher and H. Bubert, Wiley-VCH, Weinheim, Germany, 2nd edn., 2011, ch. 31, pp. 481–497. [143] K. F. Domke and B. Pettinger, ChemPhysChem, 2010, 11, 1365–1373. [144] E. Bailo and V.Deckert, Chem. Soc. Rev., 2008, 37, 921–930. [145] Z.-L. Yang, J. Aizpurua and H. Xu, J. Raman Spectrosc., 2009, 40, 1343–1348. [146] 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, 82–86. [147] B. Pettinger, Mol. Phys., 2010, 108, 2039–2059. [148] A. Kudelski and B. Pettinger, Chem. Phys. Lett., 2004, 383, 76–79. [149] C. Neacsu, J. Dreyer, N. Behr and M. Raschke, Phys. Rev. B, 2006, 73, 193406. [150] K. Domke and B. Pettinger, Phys. Rev. B, 2007, 75, 236401. [151] C. Neacsu, J. Dreyer, N. Behr and M. Raschke, Phys. Rev. B, 2007, 75, 236402. [152] W. Zhang, B.-S. Yeo, T. Schmid and R. Zenobi, J. Phys. Chem. C, 2007, 111, 1733– 1738. [153] T. Ichimura, H. Watanabe, Y. Morita, P. Verma, S. Kawata and Y. Inouye, J. Phys. Chem. C, 2007, 111, 9460–9464. [154] R. L. Agapov, A. V. Malkovskiy, A. P. Sokolov and M. D. Foster, J. Phys. Chem. C, 2011, 115, 8900–8905.

25 Chapter 1: General Introduction

[155] 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, J. Phys. Chem. C, 2012, 116, 478–483. [156] M. D. Sonntag, D. Chulhai, T. Seideman, L. Jensen and R. P. Van Duyne, J. Am. Chem. Soc., 2013, 135, 17187–17192. [157] C. Blum, L. Opilik, J. M. Atkin, K. Braun, S. B. Kämmer, V. Kravtsov, N. Kumar, S. Lemeshko, J.-F. Li, K. Luszcz, T. Maleki, A. J. Meixner, S. Minne, M. B. Raschke, B. Ren, J. Rogalski, D. Roy, B. Stephanidis, X. Wang, D. Zhang, J.-H. Zhong and R. Zenobi, J. Raman Spectrosc., 2014, 45, 22–31. [158] C. Fokas and V.Deckert, Appl. Spectrosc., 2002, 56, 192–199. [159] K. F. Domke and B. Pettinger, ChemPhysChem, 2009, 10, 1794–1798. [160] E. M. van Schrojenstein Lantman, T. Deckert-Gaudig, A. J. G. Mank, V.Deckert and B. M. Weckhuysen, Nat. Nanotechnol., 2012, 7, 583–586. [161] M. Sun, Z. Zhang, H. Zheng and H. Xu, Sci. Rep., 2012, 2, 647.

26 CHAPTER 2

Surface-Enhanced Raman Spectroscopy Studies of the Photo-Catalytic Reaction of Nitro-Aromatics

Abstract

In this Chapter a literature overview is given on the analysis of the reduction of nitro- aromatics and their reduction pathways. These reactions have been examined using Raman spectroscopy, and many details about the reactivity and sensitivity to chemical environment have already been reported. The reduction product of these reactions has been the subject of debate for some time, and therefore a wealth of spectroscopic information is available. The photo-catalytic reduction of p-nitrothiophenol is easily induced during surface-enhanced Raman scattering (SERS) measurements, while the molecules readily adsorb into the hotspots. This reaction is found to be an ideal model reaction for demonstrating the capabilities of SERS and TERS in catalysis.

This chapter is based on: E.M. van Schrojenstein Lantman, A.J.G. Mank, B.M. Weckhuysen, in preparation.

27 Chapter 2: Surface-Enhanced Raman Spectroscopy Studies of the Photo-Catalytic Reaction of Nitro-Aromatics

2.1. Introduction

Surface-enhanced Raman scattering (SERS) is a surface-specific technique. As those molecules adsorbed onto a SERS material are measured most efficiently, a model reaction ideally takes place solely on the surface. To show the applicability of nano-scale analysis via SERS, the reaction must be easy to induce and show sufficient signal to allow time- resolved reaction monitoring. The best documented chemical reactions investigated with surface-enhanced Raman scattering (SERS) are those related to p-nitrothiophenol (pNTP) and p-aminothiophenol (pATP). These molecules have large absorption cross-sections at visible wavelengths, and are thereby easily detected at low concentrations. Especially if SERS-active particles are also involved, this can even lead to surface-enhanced resonance Raman scattering (SERRS) at certain excitation wavelengths. The reduction of pNTP has been studied since the 1990’s, 1 but in the more general terms of the reduction of nitro-aromatics, literature goes back to 1968. 2 There are various mechanisms to reduce nitro-aromatics, including photo-chemical reduction via ultraviolet (UV) excitation, electro-chemical reduction, and photo-catalytic reduction. Understand- ing the various reaction pathways is crucial for interpretation of experimental studies, as many variables can affect the SERS-observations. The photo-catalytic reduction of pNTP is used as a model-reaction in the rest of this PhD Thesis, and the nanoscale SERS and TERS studies require in-depth knowledge of this class of reactions. Therefore, an overview of the literature on the reduction of nitro- aromatics is given in this Chapter, focussing on detection with SERS. The compounds covered are nitrobenzene, nitrophenol, p-nitrobenzoic acid and pNTP,depicted in Fig. 2.1 The thiol functionality of pNTP ensures that the reaction takes place right at the catalyst surface. This makes it ideal for testing novel spectroscopic techniques, like TERS.

2.2. Reduction Mechanisms

The reduction of nitro-aromatics can be achieved in a variety of ways. Most straightforward is chemical reduction, but alternative pathways have received a wealth of interest. These pathways can be divided into three main categories: UV photo-chemistry, electrochemistry, and plasmon-mediated photo-chemistry or photo-catalysis. Chemical Reduction - The reaction from nitro-aromatics to amino-aromatics is a reduction process, and therefore requires a reductant to occur, e.g. NaBH4 or H2. Metal nanoparticles can facilitate this process via electron transfer between reduction agent and reactant.

28 Reduction Mechanisms

Photo-Chemical Reduction - For a molecule in solution, a certain activation energy is required to excite it from the ground state to its electronically excited state. For ‘normal’ photo-chemistry, the excitation wavelength needs to correspond to, or be higher than, the activation energy, as illustrated in Fig. 2.2a. A photochemical process usually requires high energy excitation, in the UV-wavelength range. Electrochemical Reduction - A similar process occurs in electrochemistry. The ground state of an adsorbed molecule on an electrode is not necessarily the same as in solution, but probably similar. Under normal circumstances, the electrode will have an

NO2 NO2 NO2 NO2

OH SH a) b) c) HO O d)

FIG. 2.1. Nitro-aromatics covered in this Chapter - a) nitrobenzene, b) nitrophenol, c) p- nitrobenzoic acid, and d) p-nitrothiophenol.

a) b) c) Molecule Metal Adsorbate Metal Adsorbate Reactive Reactive Reactive excited state excited state excited state

e- hν E EF F hν

Ground state Ground state Ground state

FIG. 2.2. Various reduction mechanisms. a) Photo-catalytic reduction through direct photonic excitation of the reactant molecule into the reactive excited state. b) Electrochemical reduction through increase in electrode potential and subsequent charge-transfer into the reactive excited state of the adsorbed reactant. c) Photo-catalytic reduction through optical excitation within a SERS-active material and subsequent charge-transfer into the reactive excited state of the adsorbed reactant.

29 Chapter 2: Surface-Enhanced Raman Spectroscopy Studies of the Photo-Catalytic Reaction of Nitro-Aromatics

TABLE 2.1. An overview of the literature describing reduction of nitro-aromatics.

Reactant Reduction pathway Electro- Photo- Plasmon- Chemical chemical chemical mediated nitrobenzene - 3–8 2 9 p-nitrophenol 10–13 --- p-nitrobenzoic acid - 14 - 14,15 p-nitrothiophenol 16–21 1,22–27 28 21,29–42 energy level around its Fermi level. Through application of a potential this energy level can be increased. As soon as the bias has reached the level of the reactive excited state of the adsorbed molecule, the electrons will ‘leak’ into that state via the charge-transfer process and start the reaction (Fig. 2.2b). Photo-Catalytic Reduction - Midway between photo-chemistry and electrochemistry is the photo-catalytic reduction. Speciﬁc nanoparticles, usually gold or silver, have a Fermi-level higher than the ground state of the adsorbed molecule. These particles are also capable of coupling to speciﬁc wavelengths of light through their plasmon resonance. As illustrated in Fig. 2.2c, this light effectively heats up the electrons in the metal and increase their energy up to the level of their excited state of the adsorbed molecule. Via a charge transfer process this energy is transferred, so a reaction can be started.

2.3. Reduction of Nitro-Aromatics

Most of the early work in the photo-reduction or electrochemical reduction of nitro- aromatics was performed with nitrobenzene. The later studies with p-nitrobenzoic acid, p-nitrophenol and p-nitrothiophenol based their reaction chemistry on these initial nitrobenzene studies. Table 2.1 gives an overview of the literature describing the reduction of nitro-aromatics, sorted by reaction pathway. The difference between these nitro-aromatic compounds lies in the group in para- position to the nitro-functionality. Where nitrobenzene will coordinate to catalyst surfaces through the nitro-functionality, both p-nitrobenzoic acid and p-nitrothiophenol will coordinate via respectively their carboxylic- and thiol-functionality. SERS was ﬁrst reported by Fleischmann et al. in 1974. 43 It took until 1985 for the ﬁrst SERS study of the reduction of nitrobenzene to be published by Shindo et al. 3 Other common characterization tools for the reduction mechanisms of the nitro-aromatics include cyclic voltammetry and UV-VIS spectroscopy.

30 Reduction of Nitro-Aromatics

2.3.1. Chemical Reduction. The chemical reduction of nitro-aromatics typically uses

NaBH4 as a reducing agent. p-Nitrophenol was found to be converted to p-aminophenol with pseudo ﬁrst-order reaction kinetics at an excess of NaBH4 and in the presence of Au nanoparticles. 10 More advanced temperature-dependent measurements of the reaction rate have been utilized to determine the activation energy for a variety of nanoparticle catalysts. 11 The metal catalyst required for the reaction can also be paired with SERS as a characterization technique. However, these studies typically use UV-VIS spectroscopy for reaction monitoring, rather than SERS. The coordination of p-nitrophenol to the catalytic surface is disputed to be either via the nitro-functionality 13 or via the phenolate-group. 12 Alterations to the material of the metal nanoparticles, e.g. addition of Cu atoms, has shown to change the adsorption energy of the reactant and thereby the reactivity of the catalytic particles. 13 Unlike nitrophenol, the chemical reduction of pNTP cannot be monitored with UV- VIS spectroscopy. The thiol-functionality strongly binds pNTP metal nanoparticles in the reaction mixture, making it unlikely to observe unbound reactants or their products in solution via UV-VIS spectroscopy. The chemical reduction of pNTP has been studied by SERS with a variety of materi- 21 16–18 20 als, including Au, combinations of Au and Pt, Au/Pd nanoparticles, and Au and 19 SiO2. Typically, the reduction occurs via NaBH4, though exceptions are known, in the 17 20 form of KBH4 and H2/N2. The chemical reduction study of pNTP over sophisticated bi-functional Au/Pt/ Au 18 (core/shell/ satellites) nanoraspberries by Xie et al is shown in Fig. 2.3. Pt acts as the catalyst, whereas the combination of a Au core with Au satellites creates SERS hotspots on a single particle. Upon increasing concentration of NaBH4, full reduction to pATP was observed by SERS (633 nm excitation), via p,p-dimercaptoazobisbenzene (DMAB) as a reaction intermediate. In a later variation to the nano-raspberries, the authors used

Au/SiO2/Au (core/shell/ satellites) particles to monitor the milder reaction as catalysed by the Au satellite particles. 19 Joseph et al. used a Au SERS substrate combined with smaller (2 nm) Pt nanoparticles as catalyst, to observe the chemical reduction of pNTP to pATP through SERS monitoring. 16 2-Naphthalenethiol was incorporated in the experiment as an internal standard, for 1 quantiﬁcation of the SERS signal. This led to a reaction constant of 0.011 s− , assuming pseudo-ﬁrst-order reaction kinetics.

An alternative environment for chemical reduction is a gas ﬂow of H2/N2, which was followed over composite AuPd nanoparticles by Huang et al. First-order reaction kinetics were observed, which were similar to those reported by Joseph et al., with a reaction

31 Chapter 2: Surface-Enhanced Raman Spectroscopy Studies of the Photo-Catalytic Reaction of Nitro-Aromatics

FIG. 2.3. a) Three distinct spectral components observed during the chemical reduction of pNTP over Au/Pt/Au nanoraspberries, and b) their contribution to the Raman spectra as function of the 18 amount of NaBH4 added to the reactant solution. Figure taken from Xie et al.

3 1 constant of 21.66 x 10− s− . No signals from DMAB were observed, which indicated a strongly reducing environment. 20

2.3.2. Photo-Chemical Reduction. The photo-chemical reduction of nitro-functionalized aromatic molecules was first described in 1986, and based on nitrobenzene. 2 The reactants were reduced in solution via UV-irradiation, and the detailed experimental study lead the authors to suggest the reaction mechanism shown in Fig. 2.4a. The first step requires photo-excitation of nitrobenzene. The later steps include hydrogen abstraction from solvent molecules. Depending on the excitation wavelength, different reaction products were observed: short-wavelength UV-light (λ<290 nm) was required for full reduction to aniline, while longer excitation wavelengths (λ>290 nm) resulted in dimeric species as major products. This photo-chemical reduction has also been used to photo-pattern SAMs of pNTP on Ag island films, 28 but no other studies have been reported. However, the most recent reported reaction mechanisms for electrochemical and photo-catalytic reduction 14 are based on the photo-chemical reaction mechanism as published by Barltrop et al. 2

2.3.3. Electrochemical Reduction. The reduction of nitrobenzene can be induced via electrochemistry, while SERS was used as a tool to observe the adsorbates present on the surface of the electrode. In 1985 the ﬁrst SERS study of the reduction of a nitro- aromatic was published by Shindo et al. 3 The authors attributed the Raman spectra of the reaction product to trans-azobenzene, adsorbed through the azo-group onto the electrode.

32 Reduction of Nitro-Aromatics

a) hv ArNO2 ArNO2* ArN=NAr

RH ArNO hv o-hydroxyazocompound base ArNO2H hv hv ArN=N(O)Ar RH RH hv or base base

ArN(OH)2 ArNO ArNHOH ArNHOH ArNH2 -H2O ArNO2H RH hv

+ b) ArNO2 ArN=NAr products e-+H+ ArNO2 or ArNO 4e-+4H+ ArNO2H H2O + ArN=N(O)Ar Ar(H)N=N(H)Ar + H2O

- + e-+H+ 2e +2H

- + 2e-+2H+ 2e +2H ArNHOH ArNH ArN(OH)2 ArNO 2 + +

H2O H2O

c) ArNO2 hv hv ArNO2* RH (sensitized) ArN=N(O)Ar ArN=NAr ArNO2H hv alkali RH RH or hv

ArN(OH)2 ArNO ArNOH ArNHOH ArNH2 + H2O

FIG. 2.4. Reaction mechanisms for the reduction of nitro-aromatic molecules. a) Photo-chemical reduction via UV-irradiation and hydrogen-abstraction of solvent molecules. Mechanism as proposed for nitrobenzene by Barltrop et al. 2 b) Reaction mechanism of electrochemical reduction as proposed for p-nitrobenzoic acid by Sun et al. 14 c) Reaction pathway for plasmon-mediated photo- chemical reduction as proposed for p-nitrobenzoic acid by Sun et al. 14

33 Chapter 2: Surface-Enhanced Raman Spectroscopy Studies of the Photo-Catalytic Reaction of Nitro-Aromatics a) b)

FIG. 2.5. a) Relation between reduction potential of pNTP (vs. Ag/AgCl) and the pH in the solution. 25 Taken from Futamata et al. b) Cyclic voltammogram of a pNTP-covered electrode in 0.1 M NaClO4, measured at 50 mV/s. The dashed line shows the second scan. c) Difference in SEIRA spectra at -0.2 and -1.0 V vs. SCE. Figures taken from Matsuda et al. 1

The same research group continued its efforts to elucidate the processes involved in electrochemical analysis of nitrobenzene, azobenzene and related compounds. An initial communication described the different spectra for adsorbed and desorbed transazoben- zene moieties, 4 while later, they concluded that these different spectra could actually be caused by electromagnetic versus chemical enhancement of the azobenzene molecules. 5 More importantly, this study ended with a proposal of an electrochemical reaction mechanism for the reduction of nitrobenzene to aniline. In later publications they showed and assigned the SERS spectrum of aniline. 6,7 An extensive study of the electrochemical reduction of nitrobenzene was published in 1988 by Gao, Gosztola and Weaver, using slightly different reaction conditions to the group of Nishihara and Shindo described above. Through comparison of neat Raman and adsorbate SERS spectra to the spectra observed during the in-situ electrochemical reduction of nitrobenzene, the authors ﬁnd a difference in reactivity and pathway depending on the pH of the reactant solution. 8 For pNTP this pH-dependence was also observed, and studied in greater detail. 25 The potential of the electrochemical reduction was observed to be strongly dependent on the pH of the solution, as shown in Fig. 2.5a.

34 Reduction of Nitro-Aromatics

The combination of cyclic voltammetry and in-situ surface-enhanced infrared absorption (SEIRA) described the full electrochemical reduction of pNTP to pATP. 1 Fig. 2.5b and 1 c show that the spectral features of pNTP (NO2 stretching at 1346 cm− ), present at -0.2 V vs. SCE disappears when the potential was brought down to -1.0 V vs. SCE, corresponding 1 to the formation of pATP (NH2 deformation at 1628 cm− ). Where the partial electrochemical reduction of pNTP has been studied for the formation of SAMs of hydroxylaminethiophenol, 26,27the pure electrochemical reduction of pNTP has not been studied extensively. At the time pNTP became a molecule of interest, combinations with other techniques like SERS were common practice, and exactly this combination leads to a convoluted reduction mechanism between electrochemical and photo-catalytic reduction. For example, the use of 633 nm excitation was shown to shift the redox potential to 0.7 V vs. SCE, 1 and can even occur at 0 V vs. SCE if the sample is left long enough to react. 22 This shift has also been observed for other reduction poten- tials and wavelengths. 23,24 However, the exact potential is much dependent on variables like laser power density, excitation wavelength and the fraction of electrode surface that is excited by the Raman excitation laser.

2.3.4. Photo-Catalytic Reduction. The initial studies of photo-catalytic reduction started with the investigation of p-nitrobenzoic acid (pNBA) on Ag island films, as it readily adsorbes onto Ag surfaces through the carboxylic acid group. 14,15 Roth, Venkatachalam and Boerio addressed why the SERS spectrum of pNBA (at 514.5 nm laser excitation) is so different to its neat Raman spectrum in a review in 1986. 15 The SERS spectrum was identified as the result of the photo-chemical reduction of pNBA to the dimeric azodiben- zoate, but is actually photo-catalysed. This reaction was shown to be photo-activated rather than thermally induced. It shows that it is very important to separate SERS and reaction in this case, to prevent misinterpretation of the SERS spectrum, 44 although both rely on Ag nanoparticles and light. The wavelength- and power-dependence of both phenomena are different and can be separated if chosen wisely. A similar, more thorough study followed soon after by Sun et al. 14 They investigated the photochemical reduction of pNBA on both Ag island films and Ag electrodes for the effects of e.g. laser excitation wavelength and laser power to the reaction rate. The photo-catalytic reduction capability was attributed to the charge-transfer process, similar to electrochemical reduction. The authors came to the reported reaction mechanisms as reproduced in Fig. 2.4b and c, and observed convolution between these mechanisms, as the electrochemical reduction was observed to take place at higher reduction poten- tials under green laser excitation in comparison to dark electrochemical reduction. Also, like in the electrochemical reduction, a pH dependence was reported for the reduction

35 Chapter 2: Surface-Enhanced Raman Spectroscopy Studies of the Photo-Catalytic Reaction of Nitro-Aromatics

a) b) c)

FIG. 2.6. a) Time-resolved photo-catalytic reduction of pNTP adsorbed onto SERS-active Ag nanoparticle, at 514 nm excitation. b) Relative Raman intensities as the reaction progresses, under ambient conditions (as in a), and c) in vacuum. Figure taken from Kim et al. 29

pathways, where alkaline conditions result in dimeric products, and acidic conditions in aminobenzoic acid as the main reaction product. The photo-catalytic reduction of pNTP is expected to follow the same reaction mechanism as reported for pNBA. The ﬁrst description of pNTP was by Han et al. in 2002, 30 since then followed by many others. The reduction readily occurs when adsorbed onto Ag SERS-active substrates and under excitation of a green laser, as illustrated by Fig. 2.6a. Ambient conditions were found to result in a faster reaction than in vacuum: the decay time of pNTP is respectively 3.3 min (Fig. 2.6b) and 19 min (Fig. 2.6c). 29 As with most chemical reactions, cooling the substrate to 77 K greatly reduced the reaction rate, 31 whereas heating of the sample increased the reaction rate. 32 The reaction has generally been observed on Ag, but was also found on Cu 33 and Au. 21 The reaction rate was strongly dependent on the excitation wavelength used, coupled to the laser power density. On Ag, 532 nm readily excites strong surface plasmons, while the same power density at 633 nm has been shown to yield a much lower reaction rate. 34 The reduction has been studied with various SERS geometries and nanoparticle 36 37 materials, including Ag nanoplates, SERS-active Cu2O nanoparticles, Pd SERS- 45 38 40 active substrates, Au/Ag alloy nanoparticles, Pt@Ag nanoparticles, and sandwich- geometries that have Ag nanoparticles over a SERS-active Au substrate, 46 or Au micro- powder over a Ag ﬁlm. 39 Also high vacuum (Au) STM-TERS has been used to study the

36 Points of Debate in Literature reduction of a pNTP-functionalized Ag ﬁlm. The combination with STM allowed the re- searchers to explore both the voltage dependence and laser power dependence of the reduction reaction. 41 The temperature of the SERS hotspot in this geometry was found to be 330 K, 42 proving that thermal heating is not a parameter in the reduction process. Recently, the photo-catalytic reduction of nitrobenzene has been attracting attention as a sustainable reaction pathway to azo-compounds. For example, supported Au nanoparticles were shown to be capable of reducing nitrobenzene to azobisbenzene in isopro- panol and KOH with high conversion and selectivity, under photo-excitation. 9 The reverse pathway has also been investigated, where aniline is oxidized over Au nanoparticle catalysts to azobisbenzenes and nitrobenzene. 47

2.4. Points of Debate in Literature

Care has to be taken when comparing studies of (‘bulk’) solution-based nitrobenzene with self-assembled p-nitrothiophenol or other para-nitro-aromatics with a surface- coordinating functional group. Nitrobenzene reduction occurs via adsorption of the nitro- group onto the metal surfaces. In the case of e.g. pNTP,molecules are adsorbed onto the metal surface via their thiol-functionality and have a ﬁxed orientation with the nitro-group facing away from the metal surface. 48 Also, pNTP molecules in a monolayer are in very close proximity to each other. This can make dimerization reactions more likely than in the case of e.g. diluted nitrobenzene.

2.4.1. Comparing Reaction Conditions. The direct comparison of different reduction studies of pNTP requires the knowledge of a variety of parameters. For chemical reduction, an obvious variable is the catalytic materials (Au, Pt and Pd), but other reaction parameters include the total catalytic surface area, e.g. per unit volume or reactant.

More trivial variations can be found in the NaBH4 concentration or reaction temperature, which will both affect the reaction rate of reduction. The reaction mechanisms of photo-chemical, electrochemical, and photo-catalytic reductions are very similar in nature, and can easily overlap. In all cases, excitation energies, whether optical or electric, must be coupled to the respective plasmon resonances or exciton energies, and consequent excited electrons should be high enough in energy to allow for charge-transfer to the excited reactive state of pNTP,as illustrated in Fig. 2.2. Most commonly probed is the combination between electro-chemical reduction and photo-catalytic reduction, when SERS is used for in-situ reaction monitoring. The electrode potential can give the metal the right push to increase the energy of electrons excited by laser excitation to the surface plasmon, and thereby reach the charge-transfer level into the adsorbate reactive excited state.

37 Chapter 2: Surface-Enhanced Raman Spectroscopy Studies of the Photo-Catalytic Reaction of Nitro-Aromatics

Care has to be taken when comparing SERS data from different studies, as many have not separated conversion from measurement. The parameters that determine whether a measurement and catalysis are independent, are: SERS-active material • Local SERS near-ﬁeld intensity • Excitation wavelength • Laser power density • Electrochemical potential • pH • Surface coverage of reactant • Concentration of reducing agent • Reaction temperature. • Apart from the various reaction parameters, it is also important to be aware of the differences in the analytical tools used for reaction monitoring. Electrochemical characterization techniques, like cyclic voltammetry, are bulk techniques: they analyse the full SAM simultaneously. Coupling of electrochemical reactions to SERS measurements can cause convolution of electrochemical and photo-catalytic reduction. However, SERS will only measure at the so-called ‘hot-spots’ within the laser focus, whereas e.g. cyclic voltammetry will measure the full electrode surface. Thevariations in measurement volume can and will lead to discrepancies between different characterization techniques, if measurement and catalysis are not separated for the SERS measurement.

2.4.2. Reduction Product. SERS spectroscopy is a very informative technique that yields molecule sensitive information in nano-scale measurement volumes. However, there are many factors that inﬂuence the spectrum. Based on the studies described above, reduction of nitro-aromatics can yield either amino-aromatics or dimeric azo- functionalized aromatics. In the case of pNTP, identiﬁcation of the reaction product is hindered by the strong wavelength-, pH, and substrate-dependency of the SERS spectrum of pATP. Its SERS spectrum is usually remarkably different from its normal, neat Raman spectrum. This phenomenon has been the subject of many studies, 49 and has resulted to pATP being exposed to a wide range of SERS-active materials and conditions. An overview of (some) experiments described in literature is given in Table 2.2. The neat Raman spectrum of pATP is shown in Fig. 2.7a (red). The amino- functionality does not result in a characteristic Raman peak, and the most prominent 1 vibrational bands are at 1087 and 1593 cm− , respectively belonging C-S stretching and C-C stretching (see Table A.2 in Appendix A). Adsorbed onto Ag SERS substrates, the

38 Points of Debate in Literature

FIG. 2.7. a) Neat Raman spectra of DMAB (top, black) and pATP (bottom, red), at an excitation wavelength of 488 nm. For clarity, molecule structures are shown to the right. Vibrational assignment of the spectra can be found in Appendix A. Raman spectra on Ag SERS substrates at b) 488 nm excitation, c) 514.5 nm excitation, and d) at 632.8 nm excitation. In all subﬁgures, the top spectrum belongs to DMAB (black), and the bottom spectrum to pATP (red). The ﬁgure was adapted from Kim et al. 50 spectrum is distinctly different, as shown in Figs. 2.7 b-d (black). Vibrations at 1140, 1 1390 and 1435 cm− appear, and as shown here, the intensity of these peaks depends on the excitation wavelength. At 488 nm excitation, these are the most prominent peaks in the spectrum. Resonance Raman enhancement occurs only for those bands linked to absorption at this wavelength, and likely plays a role in the changed ratio. Historically, these new spectral features were interpreted as the chemical enhancement effect of SERS, due to a charge-transfer process. 51 However, since an article by Huang et al., the formation of DMAB on SERS substrates is now a plausible explanation for these intense SERS bands. 52 In Fig. 2.7a (black), the normal spectrum of DMAB is

39 Chapter 2: Surface-Enhanced Raman Spectroscopy Studies of the Photo-Catalytic Reaction of Nitro-Aromatics

FIG. 2.8. Energy diagram for the electrochemical and/or photo-chemical and -catalytic reaction from pATP to DMAB. Reproduced from Zhang et al. 53

shown alongside that of pATP. As azo-functionalities are easily polarized, they result in 1 strong bands in a Raman spectrum, in this case at 1140, 1393 and 1442 cm− . In the corresponding SERS spectra in Fig. 2.7b-d (black), the spectral features closely resemble that of the normal Raman spectrum for pATP.The corresponding energy diagram is shown in Fig. 2.8. Wavelength-dependent UV-Raman spectroscopy failed to produce the SERS- speciﬁc bands for pATP, which means that different excited states are probed, or that resonance enhancement for speciﬁc bands is not present. 51 Since the start of this debate, DMAB has gained interest as a SERS analyte, including e.g. pH-dependence and potential-dependence of its SERS spectrum in comparison with pATP. 52,54 The discussion is further supported by a variety of theoretical calculations, including DFT-calculations of the Raman spectra, 55–59 and calculations of the possible reaction pathway between pATP and DMAB. 60,61 It is interesting to note that the spec- 1 tral features at 1140, 1390 and 1440 cm− have only been observed for molecules with nitrogen-functionality. 62

2.4.3. Variations in SERS of p-Aminothiophenol. pATP adsorbed onto SERS-active substrates causes spectral features that much resemble peaks in the neat Raman and SERS spectra of DMAB. These spectral features have long been ascribed to the chemical enhancement effect, and have frequently been used to estimate (favourable) enhancement

40 Points of Debate in Literature

1 FIG. 2.9. Potential-dependence of the 1440 cm− peak intensity for various Raman excitation wavelengths as measured with a Ag electrode. a) 632.8 nm (1.96 eV), b) 514.5 nm (2.41 eV), and c) 488.0 nm (2.54 eV). Taken from 51.

factors for various SERS substrates. As described above, they are more likely to represent the catalytic efﬁciency of the reaction from pATP to DMAB. Literature contains a multitude of examples in which authors have not separated conversion and analysis, as they used substrates that are both catalytically active and Raman enhancing. The convolution between the photo-catalysis and electrochemistry of pATP was ﬁrst observed by Osawa et al (shown in Fig. 2.9). 51 The electrochemical reduction potential was shown to change, based on the Raman excitation wavelength used. Similar experiments were repeated later, and show the same behaviour. 63,64 For aniline, the electrochemical dimerization to azobenzene has been observed under alkaline conditions. 65 Though other oxidation reactions for aniline are possible and prominent under acidic conditions, these are not expected to be readily accessible to pATP in monolayers, nor have they been observed. In fact, various SERS studies of pATP observe a pH-dependence. At acidic conditions, the reaction to DMAB is not possible, 53 but DMAB is observed at alkaline conditions. 59,63,66–68 Groups disagree on the reversibility of the pH-dependence. 59,68 The conversion of pATP to DMAB has been investigated at various temperatures. 31 At 77 K DMAB was still observed, though this could possibly be due to local heating at SERS hotspots. Other experiments, studying the SERS spectra from room temperature up 32 to 90 ◦C show an expected increase in conversion.

41 Chapter 2: Surface-Enhanced Raman Spectroscopy Studies of the Photo-Catalytic Reaction of Nitro-Aromatics

A number of recent studies described the activation of oxygen over Au an Ag surfaces, as a contributor to the redox-reaction between pATP and DMAB. 53,69,70 The reaction of pATP to DMAB is catalysed via the charge-transfer mechanism over Au or Ag nanoparticles, whereas the reverse reaction is postulated to be through hot-electron injection from the metal into the azo-functionality. 69 The surface plasmons generated via the catalyst particles are strong enough to activate oxygen on Au and Ag surfaces to produce O−2 , to then form metal oxides or hydroxides. 70 These surface species act as oxidation agent to facilitate the oxidation of pATP to DMAB. For the reverse reaction, a hydrogen source 69 is required. Typically, H2O or OH− will act as such, unless hindered by an abundance of H+ under acidic conditions. 53 As with more photo-chemical reactions, many parameters inﬂuence the reaction rate. For pATP they are very similar to those described for the reduction of nitro-aromatics. As DMAB forms even more readily from pATP than from pNTP, the parameters chosen for SERS of pATP are extremely critical. The choice of catalytic activity of the substrate, combined with the excitation wavelength and laser power density are the critical to the conversion measured by the SERS spectrum. Not all reports in literature state these parameters, and especially the laser power density is crucial when comparing spectra at different wavelengths 52 - this can only be done if they are observing similar laser power densities between the various excitation sources.

2.5. Concluding Remarks

The reduction of nitro-aromatics is much described for analysis with SERS. The four reaction mechanisms are: chemical, photo-chemical, electrochemical, and photo-catalytic reduction, which have all been examined for various nitro-aromatics, being nitrobenzene, nitrophenol, p-nitrobenzoic acid and p-nitrothiophenol. However, the use of SERS as a characterization technique for these reduction mechanisms must be done with care, to either avoid or quantify the convolution between reaction and measurement. Using SERS to monitor the reduction of nitro-aromatics, can also induce an additional reduction via photo-catalysis. Though it requires extra care in the use of SERS as a characterization technique, but simultaneously provides an excellent model-reaction. The easy photo-catalytic reduction of nitro-aromatics over typical SERS substrates makes it an ideal model-reaction to explore the capabilities of nanoscale analysis with SERS in heterogeneous catalysis. However, many factors inﬂuence the reaction, including the SERS-active material, excitation wavelength and laser power density, surface coverage of

42 Concluding Remarks the reactant, along with a range of external chemical effects like pH and reaction temperature. For in-situ monitoring, a long excitation wavelength and/or low power density can be used to slow down photo-catalytic reduction. pNTP is the ideal reactant for application in SERS, as it readily adsorbs onto metal surfaces, and thereby places itself directly into the hotspots. The photo-catalytic reduction can easily be induced during SERS measurements, and is well-described in literature. After careful examination of the literature describing the SERS characteristics of p-aminothiophenol, the photo-catalytic reduction product as observed with SERS is most likely p,p’-dimercaptoazobisbenzene. Based on all this, we propose the photo-catalytic reduction of p-nitrothiophenol as a model reaction in the rest of this PhD Thesis. This photo-catalytic reduction is used to explore both surface- and tip-enhanced Raman scattering for reaction monitoring.

43 Chapter 2: Surface-Enhanced Raman Spectroscopy Studies of the Photo-Catalytic Reaction of Nitro-Aromatics , ), 64 64 64 31 31 69 MeOH: / acetone: 633 nm, 2 488 nm 514 nm, / ammonia: 64 N 2 514 nm, N ), 568 nm, 64 31 785 nm (solvent ), Ice: 64 70 86 64 64 514 nm, O: 85 2 64 31 633 nm (laser power : : 488 nm, : 638 nm 2 2 2 O N 633488 nm, 633 nm, nm, 488 nm, MeOH:H evaporation 790 nm, 514 nm633 (pH nm N Other ), ), 83 82 52 63,68 ) 51,64 poten- 85 pH, pH 63 , 67,68 4 85 ), 785 nm, ), 568 nm, (pH, 70 power density (NaBH Medium 51,63,64 83,84 82,83 51,52,63,64 Aqueous 488514 nm nmpotential (potential 633 nm tial, 638 nm790 (pH nm (potential, ), 72 72 71 36 36 66 50 79,87,88 40,72,87 50,68,74,75 50,68,74,75 36,38,87,88 (temper- 568 nm, 72,87 DMHAB 66 66 488 nm 794 nm, 794 nm, DMAB, 633 nm, 514 nm, (DMAB, 514 nm, 633 nm 71 71 70 647 nm 514 nm, 23,36,71–73 ), 568), nm, 642 nm, 87 87,88 72,87 36,80,81 36,81,89 66 72,87 66 50 50 50 73,76–79 63 458 nm, Air (DMAB, 633 nm DMHAB 7851064 nm nm, DMHAB 514ature, nm 638 nm, 633 nm, 568 nm, 568 nm, 1064 nm SERS experiments of DMAB pATP, and p,p’-dimercaptohydrazobisbenzene (DMHAB) in literature. Excitation wavelengths as Material Ag Ag & Au 488 nm, Ag & Pt 488 nm, Ag & Cu 488 nm, 2.2. BLE A T stated for the SERS materials used. Any additional parameters are indicated in brackets.

44 Concluding Remarks 70 638 nm : 2 N 785 nm (solvent ) 99 O: 2 MeOH:H evaporation ), 100 68 82,98 DMAB potential, 67,68 ), 785 nm ), 790 nm 100 70 32 (pH, ), 633 nm (poten- 93,97 100 ) ) 85 100 tial 633 nm 638 nm (pH (temperature (pH DMAB 514 nm (pH, ), ), 94 90 66 91 59 70,71,95 pH 647 nm 70,71,90–94 568 nm, 71 100 potential 66 66 66 830 nm, 59 633 nm 785 nm, 94 surface-coverage, 96 633 nm 642 nm, 647 nm 647 nm 514 nm, 71 81,89,90 70 66 66 100 66 70,71,90 68 37 532 nm, laser power, 638 nm, (laser658 power, nm, 1064 nm (DMAB, 798 nm, 633 nm, O 488 nm 2 Au CuCu 488 nm, Au & Cu 488 nm, ZnO 514 nm,

45 Chapter 2: Surface-Enhanced Raman Spectroscopy Studies of the Photo-Catalytic Reaction of Nitro-Aromatics

References

[1] N. Matsuda, K. Yoshii, K.-i. Ataka, M. Osawa, T. Matsue and I. Uchida, Chem. Lett., 1992, 1385–1388. [2] J. A. Barltrop and N. J. Bunce, J. Chem. Soc. C Org., 1968, 1467–1474. [3] H. Shindo and C. Nishihara, Surf. Sci., 1985, 158, 393–397. [4] C. Nishihara, H. Shindo and J. Hiraishi, J. Electroanal. Chem. Interfacial Electrochem., 1985, 191, 425–429. [5] C. Nishihara and H. Shindo, J. Electroanal. Chem. Interfacial Electrochem., 1986, 202, 231–239. [6] H. Shindo, J. Chem. Soc., Faraday Trans. 1, 1986, 82, 45–51. [7] H. Shindo and C. Nishihara, J. Chem. Soc., Faraday Trans. 1, 1988, 84, 433–439. [8] P.Gao, D. Gosztola and M. J. Weaver, J. Phys. Chem., 1988, 92, 7122–7130. [9] H. Zhu, X. Ke, X. Yang, S. Sarina and H. Liu, Angew. Chem. Int. Ed., 2010, 49, 9657– 9661. [10] M. H. Rashid and T. K. Mandal, Adv. Funct. Mater., 2008, 18, 2261–2271. [11] J. Zeng, Q. Zhang, J. Chen and Y. Xia, Nano Lett., 2010, 10, 30–35. [12] S. Praharaj, S. Nath, S. K. Ghosh, S. Kundu and T. Pal, Langmuir, 2004, 20, 9889– 9892. [13] Z. D. Pozun, S. E. Rodenbusch, E. Keller, K. Tran, W. Tang, K. J. Stevenson and G. Henkelman, J. Phys. Chem. C, 2013, 117, 7598–7604. [14] S. Sun, R. L. Birke, J. R. Lombardi, K. P. Leung and A. Z. Genack, J. Phys. Chem., 1988, 92, 5965–5972. [15] P. G. Roth, R. S. Venkatachalam and F. J. Boerio, J. Chem. Phys., 1986, 85, 1150– 1155. [16] V. Joseph, C. Engelbrekt, J. Zhang, U. Gernert, J. Ulstrup and J. Kneipp, Angew. Chem. Int. Ed., 2012, 51, 7592–7596. [17] R. Liu, J.-F.Liu, Z.-M. Zhang, L.-Q. Zhang, J.-F.Sun, M. Sun and G.-B. Jiang, J. Phys. Chem. Lett., 2014, 5, 969–975. [18] W.Xie, C. Herrmann, K. Kömpe, M. Haase and S. Schlücker, J. Am. Chem. Soc., 2011, 133, 19302–19305. [19] W. Xie, B. Walkenfort and S. Schlücker, J. Am. Chem. Soc., 2013, 135, 1657–1660. [20] J. Huang, Y. Zhu, M. Lin, Q. Wang, L. Zhao, Y. Yang, K. X. Yao and Y. Han, J. Am. Chem. Soc., 2013, 135, 8552–8561. [21] X. Ren, E. Tan, X. Lang, T.-T. You, L. Jiang, H. Zhang, P.Yin and L. Guo, Phys. Chem. Chem. Phys., 2013, 15, 14196–14201. [22] N. Matsuda, T. Sawaguchi, M. Osawa and I. Uchida, Chem. Lett., 1995, 145–146.

46 References

[23] M. Futamata, J. Phys. Chem., 1995, 99, 11901–11908. [24] T. Zhu, H. Z. Yu, Y. C. Wang and Z. F. Liu, Mol. Cryst. Liq. Cryst. Sci. Technol., Sect. A, 1999, 337, 241–244. [25] M. Futamata, C. Nishihara and N. Goutev, Surf. Sci., 2002, 514, 241–248. [26] S. Schwamborn, L. Stoica, S. Neugebauer, T. Reda, H.-L. Schmidt and W. Schuh- mann, ChemPhysChem, 2009, 10, 1066–1070. [27] J. U. Nielsen, M. J. Esplandiu and D. M. Kolb, Langmuir, 2001, 17, 3454–3459. [28] X. M. Yang, D. A. Tryk, K. Ajito, K. Hashimoto and A. Fujishima, Langmuir, 1996, 12, 5525–5527. [29] K. Kim, Y. M. Lee, H. B. Lee, Y. Park, T. Y. Bae, Y. M. Jung, C. H. Choi and K. S. Shin, J. Raman Spectrosc., 2010, 41, 187–192. [30] S. W. Han, I. Lee and K. Kim, Langmuir, 2002, 18, 182–187. [31] K. Kim, J.-y. Choi and K. S. Shin, J. Phys. Chem. C, 2014, 118, 11397–11403. [32] V.Canpean and S. Astilean, Spectrochim. Acta A, 2012, 96, 862–867. [33] B. Dong, Y. Fang, X. Chen, H. Xu and M. Sun, Langmuir, 2011, 27, 10677–10682. [34] L. Kang, P. Xu, B. Zhang, H. Tsai, X. Han and H.-L. Wang, Chem. Commun., 2013, 49, 3389–3391. [35] S. J. Lee and K. Kim, Chem. Phys. Lett., 2003, 378, 122–127. [36] Y. Wang, X. Zou, W. Ren, W. Wang and E. Wang, J. Phys. Chem. C, 2007, 111, 3259– 3265. [37] T. You, L. Jiang, P.Yin, Y. Shang, D. Zhang, L. Guo and S. Yang, J. Raman Spectrosc., 2014, 45, 7–14. [38] K. Kim, K. L. Kim and K. S. Shin, J. Raman Spectrosc., 2012, 43, 228–236. [39] K. S. Shin, C. S. Park, W. Kang and K. Kim, Chem. Lett., 2008, 37, 180–181. [40] K. Kim, K. L. Kim and K. S. Shin, J. Phys. Chem. C, 2011, 115, 23374–23380. [41] M. Sun, Z. Zhang, H. Zheng and H. Xu, Sci. Rep., 2012, 2, 647. [42] Z. Zhang, L. Chen, M. Sun, P.Ruan, H. Zheng and H. Xu, Nanoscale, 2013, 5, 3249– 3252. [43] M. Fleischmann, P.Hendra and A. McQuillan, Chem. Phys. Lett., 1974, 26, 163–166. [44] B. O. Skadtchenko and R. F. Aroca, Spectrochim. Acta A, 2001, 57, 1009–1016. [45] K. S. Shin, Y. K. Cho and K. Kim, Vib. Spectrosc., 2014, 70, 120–124. [46] K. Kim, I. Lee and S. J. Lee, Chem. Phys. Lett., 2003, 377, 201–204. [47] A. Grirrane, A. Corma and H. García, Science, 2008, 322, 1661–1664. [48] F.Cecchet, D. Lis, J. Guthmuller, B. Champagne, G. Fonder, Z. Mekhalif, Y. Caudano, A. A. Mani, P.A. Thiry and A. Peremans, J. Phys. Chem. C, 2010, 114, 4106–4113.

47 Chapter 2: Surface-Enhanced Raman Spectroscopy Studies of the Photo-Catalytic Reaction of Nitro-Aromatics

[49] Y.-F. Huang, D.-Y. Wu, H.-P.Zhu, L.-B. Zhao, G.-K. Liu, B. Ren and Z.-Q. Tian, Phys. Chem. Chem. Phys., 2012, 14, 8485–8497. [50] K. Kim, K. L. Kim, H. B. Lee and K. S. Shin, J. Phys. Chem. C, 2012, 116, 11635– 11642. [51] M. Osawa, N. Matsuda, K. Yoshii and I. Uchida, J. Phys. Chem., 1994, 98, 12702– 12707. [52] Y.-F. Huang, H.-P.Zhu, G.-K. Liu, D.-Y. Wu, B. Ren and Z.-Q. Tian, J. Am. Chem. Soc., 2010, 132, 9244–9246. [53] X. Zhang, P.Wang, Z. Zhang, Y. Fang and M. Sun, Sci. Rep., 2014, 4, 5407. [54] K. Kim, D. Shin, K. L. Kim and K. S. Shin, Phys. Chem. Chem. Phys., 2012, 14, 4095– 4100. [55] J. W. Gibson and B. R. Johnson, J. Chem. Phys., 2006, 124, 64701. [56] D.-Y. Wu, X.-M. Liu, Y.-F. Huang, B. Ren, X. Xu and Z.-Q. Tian, J. Phys. Chem. C, 2009, 113, 18212–18222. [57] M. Sun and H. Xu, ChemPhysChem, 2009, 10, 392–399. [58] S. Liu, X. Zhao, Y. Li, X. Zhao and M. Chen, J. Chem. Phys., 2009, 130, 234509. [59] X.-F. Lang, P. Yin, E.-Z. Tan, T.-T. You and L. Guo, J. Raman Spectrosc., 2013, 44, 425–432. [60] L.-B. Zhao, M. Zhang, Y.-F. Huang, C. T. Williams, D.-Y. Wu, B. Ren and Z.-Q. Tian, J. Phys. Chem. Lett., 2014, 5, 1259–1266. [61] S. Duan, Y.-J. Ai, W. Hu and Y. Luo, J. Phys. Chem. C, 2014, 118, 6893–6902. [62] K. Kim, D. Shin, J.-Y. Choi, K. L. Kim and K. S. Shin, J. Phys. Chem. C, 2011, 115, 24960–24966. [63] K. Kim, H. B. Lee, D. Shin, H. Ryoo, J. W. Lee and K. S. Shin, J. Raman Spectrosc., 2011, 42, 2112–2118. [64] K. Kim, K. L. Kim, J.-Y. Choi, D. Shin and K. S. Shin, Phys. Chem. Chem. Phys., 2011, 13, 15603–15609. [65] P.Gao, D. Gosztola and M. J. Weaver, J. Phys. Chem., 1989, 93, 3753–3760. [66] K. Kim and H. S. Lee, J. Phys. Chem. B, 2005, 109, 18929–18934. [67] M. Sun, Y. Huang, L. Xia, X. Chen and H. Xu, J. Phys. Chem. C, 2011, 115, 9629– 9636. [68] K. Kim, K. L. Kim, D. Shin, J.-Y. Choi and K. S. Shin, J. Phys. Chem. C, 2012, 116, 4774–4779. [69] P.Xu, L. Kang, N. H. Mack, K. S. Schanze, X. Han and H.-L. Wang, Sci. Rep., 2013, 3, 2997.

48 References

[70] Y.-F. Huang, M. Zhang, L.-B. Zhao, J.-M. Feng, D.-Y. Wu, B. Ren and Z.-Q. Tian, Angew. Chem. Int. Ed., 2014, 53, 2353–2357. [71] K. Uetsuki, P.Verma, T. Yano, Y. Saito, T. Ichimura and S. Kawata, J. Phys. Chem. C, 2010, 114, 7515–7520. [72] K. Kim, H. B. Lee, J.-Y. Choi, K. L. Kim and K. S. Shin, J. Phys. Chem. C, 2011, 115, 13223–13231. [73] K. Kim, S. H. Lee, K. L. Kim and K. S. Shin, J. Raman Spectrosc., 2013, 44, 518–524. [74] K. Kim, K. L. Kim and K. S. Shin, Langmuir, 2013, 29, 183–190. [75] K. Kim, K. L. Kim and K. S. Shin, J. Phys. Chem. C, 2013, 117, 5975–5981. [76] L. Kang, P. Xu, D. Chen, B. Zhang, Y. Du, X. Han, Q. Li and H.-L. Wang, J. Phys. Chem. C, 2013, 117, 10007–10012. [77] X. Tian, L. Chen, H. Xu and M. Sun, RSC Adv., 2012, 2, 8289–8289. [78] M. Sun, Y. Hou, Z. Li, L. Liu and H. Xu, Plasmonics, 2011, 6, 681–687. [79] Y.Huang, Y.Fang, Z.-L. Yang and M. Sun, J. Phys. Chem. C, 2010, 114, 18263–18269. [80] J. Zheng, Y. Zhou, X. Li, Y. Ji, T. Lu and R. Gu, Langmuir, 2003, 19, 632–636. [81] Q. Zhou, X. Li, Q. Fan, X. Zhang and J. Zheng, Angew. Chem. Int. Ed., 2006, 45, 3970–3973. [82] G. Merga, N. Saucedo, L. C. Cass, J. Puthussery and D. Meisel, J. Phys. Chem. C, 2010, 114, 14811–14818. [83] K. Kim, J. K. Yoon, H. B. Lee, D. Shin and K. S. Shin, Langmuir, 2011, 27, 4526– 4531. [84] Y. Fang, Y. Li, H. Xu and M. Sun, Langmuir, 2010, 26, 7737–7746. [85] W. Hill and B. Wehling, J. Phys. Chem., 1993, 97, 9451–9455. [86] N.-J. Kim, M. Lin, Z. Hu and H. Li, Chem. Commun., 2009, 6246–6248. [87] K. Kim, H. B. Lee, J. K. Yoon, D. Shin and K. S. Shin, J. Phys. Chem. C, 2010, 114, 13589–13595. [88] K. Kim, J.-Y. Choi and K. S. Shin, J. Phys. Chem. C, 2013, 117, 11421–11427. [89] X. Hu, T. Wang, L. Wang and S. Dong, J. Phys. Chem. C, 2007, 111, 6962–6969. [90] M. Baia, F. Toderas, L. Baia, J. Popp and S. Astilean, Chem. Phys. Lett., 2006, 422, 127–132. [91] K. Ikeda, S. Suzuki and K. Uosaki, Nano Lett., 2011, 11, 1716–1722. [92] W.-H. Park and Z. H. Kim, Nano Lett., 2010, 10, 4040–4048. [93] K. S. Shin, J. Raman Spectrosc., 2008, 39, 468–473. [94] G.-K. Liu, J. Hu, P.-C.Zheng, G.-L. Shen, J.-H. Jiang, R.-Q. Yu, Y. Cui and B. Ren, J. Phys. Chem. C, 2008, 112, 6499–6508. [95] B. Cao, B. Liu and J. Yang, CrystEngComm, 2013, 15, 5735–5738.

49 Chapter 2: Surface-Enhanced Raman Spectroscopy Studies of the Photo-Catalytic Reaction of Nitro-Aromatics

[96] L.-S. Jiao, L. Niu, J. Shen, T. You, S. Dong and A. Ivaska, Electrochem. Commun., 2005, 7, 219–222. [97] V.Canpean, M. Iosin and S. Astilean, Chem. Phys. Lett., 2010, 500, 277–282. [98] J.-H. Yoon, J. S. Park and S. Yoon, Langmuir, 2009, 25, 12475–12480. [99] N.-J. Kim, J. Phys. Chem. C, 2010, 114, 13979–13984. [100] K. Kim, K. L. Kim and K. S. Shin, Phys. Chem. Chem. Phys., 2013, 15, 9288–9294.

50 CHAPTER 3

Reaction Kinetics of the Photo-Catalytic Reduction of p-Nitrothiophenol

Abstract

The phenomenon of surface-enhanced Raman spectroscopy (SERS) is ideal for the study of heterogeneous catalysis. The experiments in this Chapter show the capability of SERS as a tool to extract information on the reaction kinetics of a photo-catalytic process. The reduction of p-nitrothiophenol has been selected as a model system using green laser excitation and catalytic active Ag island films. Reaction kinetics were derived to describe this two-dimensional surface reaction, currently under debate if a dimerization can take place. Second-order reaction kinetics are shown to best fit the trend that the peak areas of both reactant and product follow in time. This was further substantiated with a dilution study, where a mixed self-assembled monolayer of 1% p-nitrothiophenol (pNTP) and 99% thiophenol were exposed to the same reaction conditions as for a pure pNTP monolayer. Though the pNTP signal was clearly observed, no conversion could be measured. Therefore, the photo-catalytic reduction is confirmed to be of the second-order, likely with p,p’-dimercaptoazobisbenzene (DMAB) as an intermediate, if not the reaction product.

This chapter is based on: E. M. van Schrojenstein Lantman, O. L. J. Gijzeman, A. J. G. Mank, B. M. Weckhuysen, ChemCatChem, DOI: 10.1002/cctc.201402647.

51 Chapter 3: Reaction Kinetics of the Photo-Catalytic Reduction of p-Nitrothiophenol

3.1. Introduction

The interaction between reactants at catalytic surfaces has been the subject of study for many decades. 1–5 Novel techniques are regularly introduced to gain more insight into chemical processes at surfaces, while they happen. 6–11 SERS is increasingly gaining pop- ularity as a powerful characterization tool in this field. 12 Typically, silver or gold nano- structured surfaces are used to enhance Raman signals of molecules, to a sensitivity that reaches single molecule levels. 13 This high sensitivity also makes SERS applicable in e.g. forensic trace detection 14 or biomolecular essays. 15 As the enhancement effect is confined to the metal surface, it allows for the selective study of those molecules that are in close proximity to the surface. This high spatial confinement makes SERS ideal to follow reactions on a catalytic surface. We have chosen a photo-reduction reaction as model system. This reaction can be performed in a self-assembled monolayer (SAM), which allows us to create a single reactant layer at the catalytic surface. Such monolayers are commonly characterized by grazing incidence angle infrared (IR) spectroscopy, X-ray photo-electron spectroscopy (XPS), scanning tunneling microscopy (STM) or atomic force microscopy (AFM). 16 SERS gives more localized, molecular information than grazing angle IR spectroscopy, while the combination of Raman spectroscopy with AFM is expected to allow nano-scale resolution. 17–22 Up to now, the few reactions in SAMs reported in literature are monitored using AFM, 23,24 STM, 25,26 or SERS 26–28 and are nearly all photo-reactions, e.g. ultraviolet (UV) induced photo-isomerizations, 25 or dimerizations, either within a SAM of the pure reactant, 28 or in a diluted form. 26 Here we show the possibilities in catalytic studies have not been used to their fullest. A typical heterogeneous catalytic reaction at a surface involves three separate steps, as shown in Fig. 3.1a. A reaction within a SAM is specifically the description of the two- dimensional reaction kinetics on the catalyst surface (Fig. 3.1b). We will take the study of reactions in a SAM to the next level by quantifying the specific two-dimensional reaction kinetics. One of the most described reactions in a SAM, as discussed in Chapter 2, is the photo-catalytic reduction of p-nitrothiophenol (pNTP). A large discussion point is the interpretation of the reaction product surface-enhanced Raman scattering (SERS) spectrum, which could be either p-aminothiophenol (pATP) 29–37 or p,p’-dimercaptoazobisbenzene (DMAB) 38–42 see also Fig. 3.1c). The challenge in the identification of the reaction product through other characterization techniques is twofold: an extremely low number of molecules that are converted during the reaction, and at the same time these few molecules are chemically bound to the surface. The latest developments indicate that the oxidation

52 Experimental

a) b) 5 x NA 0.12

Time hν

A(g) <-> A(ads) A(ads) <-> B(ads) B(ads) <-> B(g) A(g) <-> B(g) A(ads) <-> B(ads) c)

NO2 NH2 N N ?

S S S S S

Reactant Solvent Product

FIG. 3.1. a) Reaction steps in a typical heterogeneous catalytic reaction from A(g) to B(g) involve adsorption of reactants (A) onto the catalytic surface, a surface-reaction yielding a reaction product (B) and subsequent desorption of the reaction products into the gas environment. b) Here, a photo-catalytic surface reaction is studied. Novel with respect to a) is that only the surface-reaction is monitored, as function of irradiation at 532 nm (5x objective, NA 0.12). c) The reaction under study is the photo-reduction of p-nitrothiophenol to either p-aminothiophenol or p,p’-dimercaptoazobisbenzene. For dilution experiments, thiophenol is used as the 2D-equivalent to a solvent.

of pATP to DMAB is possible through a simultaneous plasmonic activation of oxygen. 43,44 This is further supported by a thorough theoretical study into the reaction pathway. 45 Others are still convinced that the reduction product of pNTP is pATP, with a deviation of the neat Raman spectrum of pATP due to chemical enhancement. 46,47 The goal of this Chapter is to show how SERS can help to elucidate the nature of the photo-catalytic reduction process of pNTP and its related kinetics.

3.2. Experimental

Silver island ﬁlms were prepared by vacuum deposition of 7 nm silver onto glass cov- erslips. Directly before use, ﬁlms were annealed for 1 min at 300 ◦C under argon. The

53 Chapter 3: Reaction Kinetics of the Photo-Catalytic Reduction of p-Nitrothiophenol

FIG. 3.2. The setup used for this study. Both 532 nm and 785 nm lasers are coupled to a Renishaw InVia spectrometer, that is combined with a Leica microscope. colour change from blue to red/yellow indicates a nanoscale reorganisation of the silver, creating a SERS-active substrate. The SERS substrates were immediately immersed in fresh 2 mM ethanolic thiol solutions for 15-18 h. To control the pNTP surface concentration, the ratio between p-nitrothiophenol (Fluka, technical grade) and thiophenol (Merck, for synthesis) was adjusted to keep the total thiol concentration constant. XPS measurements indicate a similar thiol concentration adsorbed onto the silver island film, which has been reported earlier for related SAMs. 48 SERS spectra were measured through a 5x objective (NA 0.12), to limit the effect of any local heterogeneities in the catalytic activity. At 532 nm laser excitation with a power 3 2 density of 6x10 W cm− using a Renishaw InVia Raman microscope (shown in Fig. 3.2), the SERS enhancement was sufficiently high while the catalytic activity was low enough at this power density to be able to follow the reaction in a realistic time-frame. Data processing was done with Matlab 2013a. Normalization of the total SERS intens- 1 ity was performed on the basis of the total SERS intensity between 1020 and 1625 cm− . Baseline-corrected peak integration was performed using trapezoidal numerical integration. The resulting time-dependent signal was smoothed using a 0th order Savitsky-Golay filter of 7 pixels. The signals for reactant and product was simultaneously fitted using the NonlinearModelFitting function in Mathematica 9.0. The quality of fit (R2) was calculated by addition of the sums of residuals and the sum of squares for both reactant and product.

54 Results and Discussion

3.3. Results and Discussion

3.3.1. SERS Reaction in a Self-Assembled Monolayer. SERS has the capability to surface dynamics of only few molecules. Time-resolved SERS spectra provide the information required to investigate the two-dimensional reaction kinetics. We also show here the added value of dilution of the reactant on the surface with a similar yet inert thiol. This dilution approach is the two-dimensional equivalent of a solvent and will provide the ultimate evidence whether a dimerization process is involved in the reaction mechanism of the photo-catalytic reduction of pNTP. pNTP will be reduced if it is adsorbed onto a Ag substrate that can be plasmonic- ally excited with green laser light, 49 as illustrated in Fig. 3.3a. The two Raman peaks 1 50 for tracking the reaction are 1335 cm− (reactant pNTP,asymmetric NO2-stretch) and 1 51 38 1440 cm− (product, assigned to pATP as b2 mode, or to DMAB as a N=N stretch), as indicated in Fig. 3.3b. See also Appendix A for more information on the vibrational assignment of these Raman spectra. Tracking both peak areas gives an idea of the surface concentration of both species (Fig. 3.3c). The postulated reaction mechanism 52 is expected to have a single rate-determining step. We chose to approximate the reaction as a one-step reaction and thereby identify the order of reaction of this rate-limiting step. The ratio between product and reactant (Fig. 3.3d) flattens off at a specific value rather than continuing to infinity. The maximum conversion is dependent on the exact reaction conditions. Fig. 3.4 shows a higher final conversion than Fig. 3.3, indicating the variability of this parameter. In terms of reaction rate, a factor has to be present to set a limit to complete conversion of the full SAM through the forward reaction. Likely candidates are e.g. a backward reaction or diffusion processes within the SAM. To test for these contributions to backward reaction, sample illumination was blocked for significant periods of time. The results of this experiment are shown in Fig. 3.4. It is clear that these dark periods do not have any effect on the signal of either reactant or product (Fig. 3.4b). This indicates that a backward reaction or diffusion does not occur on this timescale. A realistic explanation is a photo-catalytically inert part of the silver surface. Differ- ences in conversion rate therefore reflect the variation in photo-catalytic activity of the SERS substrate. This variation in photo-catalytic activity has therefore also to be taken into account when describing the reaction rate using the (surface-enhanced) Raman intensity. The Ag island films that are used as SERS substrate, are rough on the nano-scale (Fig. 3.5a), but homogeneous in morphology on the scale of the measurements described here. The SERS enhancement (Fig. 3.5c) effect over this substrate shows a homogeneous distribution, as does the maximum catalytic conversion of the same area (Fig. 3.5d). Small

55 Chapter 3: Reaction Kinetics of the Photo-Catalytic Reduction of p-Nitrothiophenol

FIG. 3.3. a) Time-dependent SERS measurements at 532 nm from a pNTP-coated Ag island film, each horizontal line represents a (colour-coded) spectrum. b) The first (top) and last (bottom) spectrum of a). The coloured bands depict the peak integration area for quantification of reactant and product in time. c) Time traces for reactant (red) and product (blue) as taken from peak areas shown in b). d) Relative peak area as function of time. Spectra were taken with 5 s integration 3 2 time at 6x10 W cm− . deviations between positions are likely due to noise in the Raman spectra. We therefore assume a linear relation between enhancement effect and photo-catalytic activity of the surface.

3.3.2. Reaction Kinetics - Derivation. For the surface dynamics, a full surface coverage is assumed at all times during the reaction. As described earlier, we have simpli- ﬁed the reaction to a one-step reaction and therefore assume only reactant and product molecules to be present at the surface. The derivation of both ﬁrst- and second-order reaction kinetics can be found below. In short, the conversion of reactant A to product B is described through the reaction constant k. The surface coverage of reactant, here denoted as A, is described as the total number of surface sites in the measurement area

(Ncat + Nncat ) multiplied by the relevant fraction of sites occupied by A (θA). The original

56 Results and Discussion

FIG. 3.4. a) SERS peak areas for pNTP and product in time, with intermittent illumination. The Raman excitation laser was blocked from the sample between 500 and 1000 s and 1500 and 2000 s. b) Shows the same data without the dark periods. Spectra were taken with 5 s integration time 3 2 at 6x10 W cm− .

surface coverage is set to θA,0. The correction form surface coverage to Raman intensity is made through the Raman cross sections σA and σB. 3.3.2.1. Assumptions. For a surface reaction, taking place in a two-dimensional space, the concentration of reactant and product is easiest expressed as function of surface fraction (respectively θA and θB) which depends on the surface concentration through a normalization factor NT of the total number of surface sites per measured surface area. We assume that the total surface coverage is constant, i.e.:

θB = θA,0 θA (3.1) − 3.3.2.2. Forward reaction - first order. Assuming a first order reaction of reactant A to product B, we only take the forward reaction into account for the reaction rate. This rate is defined as the amount of product molecules formed as function of time, or the amount of reactant consumed by the reaction as function of time:

dθA r = dθB d t = = kθA (3.2) − d t Which leads to

k θA = θA,0e− t (3.3)

57 Chapter 3: Reaction Kinetics of the Photo-Catalytic Reduction of p-Nitrothiophenol

a) 53 nm b) Intensity (a.u.)

1600 1400 1200 1000 Raman shift (cm−1)

20x20 µm c) d)

FIG. 3.5. a) Atomic Force Microscope (AFM) image (Dimension FastScan, Bruker) of a typical Ag SERS substrate as used in this Chapter. Over the same substrate, SERS mapping of pNTP SAMs (15 by 15 points, 200x200 µm) was performed through a 5x objective (NA 0.15) at 514 nm with 5 2 30 s integration time at 5x10 W cm− . The data shown here were at maximum conversion. b) Average spectrum over a 200x200 m measurement area. c) The SERS intensity over this area is 1 best depicted through by the peak area over 1545 - 1595 cm− (blue in b), and homogeneous over this area. d) The catalytic conversion over this same area is calculated by dividing the peak area 1 1 over 1415 - 1463 cm− by the peak area over 1309 - 1357 cm− (both red in b). The resulting catalytic conversion is homogeneous over the surface.

58 Results and Discussion

Assuming full surface coverage, θb = θA,0 θA, means the rate can similarly be expressed − for θB:

kt θB = θA, 0(1 e− ) (3.4) − 3.3.2.3. Forward reaction - second order. In the possible second order reaction, two reactant molecules A react together to form a single product B. Taking only the forward reaction into account, the reaction rate becomes:

dθb dθA 2 r = 2 = = kθA (3.5) d t − d t

This differential equation is solved for θA as:

θA,0 θA = (3.6) kθA,0 t + 1

1 Assuming full surface coverage, θB = 2 (θA,0 θA), means the rate can similarly be ex- − pressed for θB: 1 1 θB = θA,0(1 ) (3.7) 2 − kθA,0 t + 1 3.3.2.4. Measurement ‘volume’ and catalytic area. The Raman intensity of a surface species is directly related to the surface fraction of that species. The required correction factors are the measurement volume, or total number of sites in the measurement area, NT , and the Raman cross-section speciﬁc to that molecule, σ. This corresponds to the following relations for species A and B:

IA = σANT θA (3.8)

IB = σB NT θB (3.9)

Experiments (Fig. 3.4) indicated that incomplete conversion can best be explained by a photo-catalytically inert part of the Ag surface. This is likely based on a distribution of catalytic activity over the surface, but is here assumed to split into a discrete set of either the photo-catalytic active or inactive surface sites:

NT = Ncat + Nncat (3.10)

59 Chapter 3: Reaction Kinetics of the Photo-Catalytic Reduction of p-Nitrothiophenol

3.3.2.5. Reaction rates. The above equations can be combined to describe the Ra- man intensity of both A and B for a ﬁrst order reaction:

kt IA = σANcat θA,0e− + σANncat θA,0 (3.11)

kt IB = σB Ncat θA,0(1 e− ) (3.12) − In the case of a second-order reaction, or a dimerization, the reaction kinetics become:

1 IA = σANcat θA,0 + σANncat θA,0 (3.13) kθA,0 t + 1

1 1 IB = σB Ncat θA,0(1 ) (3.14) 2 − kθA,0 t + 1 These relations have been visualized in Fig. 3.6, at an arbitrarily chosen time-frame and set of reaction constants. The shape of the ﬁrst- and second-order reaction rates differ sufﬁciently to allow for discrimination at low noise levels.

3.3.3. Reaction Kinetics - Application. Application of Eq. 3.11-3.14 enables a quantitative description of the reaction (see Fig. 3.7 and Table 3.1). The values presented are representative for similar experiments, though a distribution in both reaction rates and the fraction of catalytically active sites has been observed (see Fig. 3.8 and Tables 3.2 and 3.3). It is important to note this has no influence on the conclusion. As all are reproducible in their favour for second order reaction kinetics Despite the low magnification objective used for these experiments, this distribution is probably still caused by variations in the SERS activity (e.g. oxidation degree, roughness, etc.) of the silver island films. The fraction of catalytically active surface sites σANcat θA,0/(σANcat θA,0 + σANncat θA,0) is roughly two thirds of the total SERS intensity. This could be a large effect at low laser power density. A high laser power density (see Fig. 3.9) does result in full conversion. Based on the R2 values for the various fits in Table 3.1 the reaction rate is better described with second-order reaction kinetics. In addition, fit residuals (see Fig. 3.7) also show a better fit for second-order reaction kinetics, especially at the start of the reaction. It is therefore likely that the rate-limiting step of the reaction is of the second 3 1 order. The reaction constant is in the order of magnitude of 1x10− s− , similar to a 49 previously reported value. The relative Raman cross section (σB Ncat θA,0/σANcat θA,0) for reactant and product are similar. the reaction product as for the reaction. These results

60 Results and Discussion a) b)

σANcatθA,0 σANcatθA,0

-1 -1 k=0.001 s kθA,0=0.001 s A A I I

-1 -1 k=0.01 s kθA,0=0.01 s

-1 -1 k=0.1 s kθA,0=0.1 s σ N θ σ N θ A ncat A,00 200 400 600 800 1000 A ncat A,00 200 400 600 800 1000 Time (s) Time (s) c) d) σ N θ ½σ N θ B cat A,0 k=0.1 s-1 B cat A,0 kθ =0.1 s-1 k=0.01 s-1 A,0

-1 kθA,0=0.01 s B B I I

k=0.001 s-1 -1 kθA,0=0.001 s

0 0 0 200 400 600 800 1000 0 200 400 600 800 1000 Time (s) Time (s)

FIG. 3.6. a) Two-dimensional, ﬁrst-order reaction kinetics (equation 3.11) describing the decay of 1 the reactant for reaction rates k=0.1, 0.001, and 0.0001 s− . b) Second-order reaction kinetics

(equation 3.13) describing the decay of reactant for the combined parameter kθA,0 at the same values as in a). c) and d) show respectively the ﬁrst- and second-order reaction rates for the formation of product, under the same circumstances as in a) and b), respectively. are reproducible for these reaction conditions, as also shown in Fig. 3.8 and Tables 3.2 and 3.3.

3.3.4. Reactant Dilution in Self-Assembled Monolayers. Besides describing reactions quantitatively, reaction kinetics in combination with SERS can also be employed in other ways. To discriminate between ﬁrst- and second-order reactions, it is insight- ful to examine the reaction rate of diluted SAMs. In this particular reaction, we have studied a mixed SAM under exactly the same reaction conditions as in Fig. 3.3. The vast difference is in the composition of the SAM: roughly 1% of the SAM consists of pNTP, the rest is thiophenol, which acts as a two-dimensional ‘solvent’. The reaction rate of a

61 Chapter 3: Reaction Kinetics of the Photo-Catalytic Reduction of p-Nitrothiophenol

TABLE 3.1. Fit parameters from Eq. 3.11-3.14 for Fig. 3.3b after a Savitsky-Golay ﬁlter. Fits are shown in Fig. 3.7.

1st order 2nd order

σANcat θA,0 0.129 0.003 0.1658 0.0029 ± ± σB Ncat θA,0 0.0691 0.0007 0.0752 0.0007 1 ± ± k(s− ) 0.003 0.007 1 ± − kθA,0(s− ) 0.00584 0.00026 − ± σANncat θA,0 0.0782 0.00012 0.0632 0.0008 ± ± R2 0.90 0.95

FIG. 3.7. a) Time traces of pNTP (red) and product (blue) fitted for a first-order reaction with corresponding fit residuals (data from Fig. 3.3b). c) Corresponding fit residuals. b) Same time traces fitted for second-order reaction kinetics with d) corresponding fit residuals. Fitted values can be found in Table 3.1. The solid black line is the fitted trend, dashed black lines show the 95% confidence intervals. pure first-order reaction would be independent of the initial reactant concentration (Eq. 3.11), whereas a hundredfold dilution is expected to result in a similar decrease in reaction rate (Eq. 3.13). Exactly this effect is observed in practice, as shown in Fig. 3.10. The small quantity of pNTP in the SAM is still sufficient to observe an identifiable SERS

62 Results and Discussion

FIG. 3.8. a) SERS spectra from a pNTP-coated Ag island ﬁlm at 532 nm as function of time. Spectra 3 2 taken with 5 s integration time at 6x10 W cm− . b) Time traces for reactant (red) and product (blue) as taken from peak areas shown in a). c) and d) show duplicate measurements at a different location on the Ag island ﬁlm.

TABLE 3.2. Fit parameters from Eq. 3.11-3.14 for the data presented in Fig. 3.8b.

1st order 2nd order

σANcat θA,0 0.114 0.003 0.147 0.003 ± ± σB Ncat θA,0 0.0649 0.0007 0.0701 0.0007 1 ± ± k(s− ) 0.00377 0.00016 1 ± − kθA,0(s− ) 0.0070 0.0004 − ± σANncat θA,0 0.0975 0.0007 0.08854 0.0008 ± ± R2 0.87 0.93 signal. However, no reaction whatsoever is observed on this time-scale, as no product peaks are visible at the end of the measurement. This provides the ﬁnal proof that the ambient photo-reduction of pNTP involves a dimerization step. The formation of DMAB

63 Chapter 3: Reaction Kinetics of the Photo-Catalytic Reduction of p-Nitrothiophenol

TABLE 3.3. Fit parameters from Eq. 3.11-3.14 for the data presented in Fig. 3.8d.

1st order 2nd order

σANcat θA,0 0.1234 0.0027 0.1626 0.0027 ± ± σB Ncat θA,0 0.0720 0.0006 0.0774 0.0006 1 ± ± k(s− ) 0.00158 0.00005 1 ± − kθA,0(s− ) 0.00305 0.00012 − ± σANncat θA,0 0.0846 0.0007 0.0706 0.0007 ± ± R2 0.84 0.92

FIG. 3.9. a) SERS spectra from a pNTP-coated Ag island ﬁlm at 532 nm as function of time show near-instantaneous complete conversion. Spectra are taken with 5 s integration time at 2x106 W 2 cm− . b) Time traces for reactant (red) and product (blue) as taken as taken from peak areas shown in a). is at least the rate limiting step and an intermediate in the reaction, if not simply the reaction product.

3.4. Conclusions

We have described the possibility to study reaction kinetics via SERS, using the showcase photo-catalytic reduction of p-nitrothiophenol. A novel two-dimensional approach in terms of reaction kinetics was derived for use with SERS. By following the character- 1 1 istic Raman peaks, i.e. 1335 cm− for reactant and 1440 cm− for the product, it was shown in two distinctly different ways that this photo-catalytic reduction at the very least involves a dimerization reaction. The rate-limiting reaction of pure SAMs of pNTP of silver SERS substrates was shown to be of the second order, and is only inﬂuenced by a forward reaction. Complete conversion was held back by a contribution of the SERS

64 References

FIG. 3.10. a) Time-resolved spectra for a sample corresponding to a 1% surface coverage of pNTP, b) shows the ﬁrst (top) and last (bottom) spectrum of this timeseries. Spectra are taken with 50 s integration time. signal from a non-catalytically active SERS surface. The most compelling evidence for a dimerization was obtained through incorporation of an inert thiol into the SAMs of pNTP. Thiophenol was thus used as the 2-dimensional equivalent of a solvent. The reaction rate was extremely slowed down by this dilution effect, up to the point that no reaction was observed under the standard reaction conditions - hence proving that a second-order reaction is involved. This shows that the sensitivity and stability of SERS is now acceptable for quantiﬁcation of reaction kinetics, even in a two-dimensional system.

3.5. Acknowledgements

Onno Gijzeman (Utrecht University) is thanked for his help with the mathematical derivation of the reaction kinetics as described in this Chapter.

References

[1] G. A. Somorjai and Y. Li, Introduction to Surface Chemistry and Catalysis, Wiley-VCH, Hoboken, New Jersey, 2nd edn., 2010. [2] G. A. Somorjai, H. Frei and J. Y. Park, J. Am. Chem. Soc., 2009, 131, 16589–16605. [3] Handbook of Heterogeneous Catalysis, ed. G. Ertl, H. Knözinger, F. Schüth and J. Weitkamp, Wiley-VCH, Weinheim, Germany, 2nd edn., 2008. [4] G. Ertl, Angew. Chem. Int. Ed., 2008, 47, 3524–3535. [5] J. Hagen, Industrial Catalysis, Wiley-VCH, Weinheim, Germany, 2005. [6] I. L. C. Buurmans and B. M. Weckhuysen, Nat. Chem., 2012, 4, 873–886. [7] M. A. Bañares, Adv. Mater., 2011, 23, 5293–5301. [8] F. Tao and M. Salmeron, Science, 2011, 331, 171–174.

65 Chapter 3: Reaction Kinetics of the Photo-Catalytic Reduction of p-Nitrothiophenol

[9] B. M. Weckhuysen, Chem. Soc. Rev., 2010, 39, 4557–4559. [10] In-situ spectroscopy of catalysts, ed. B. M. Weckhuysen, American Scientiﬁc Publish- ers, Stevenson Ranch, CA, USA, 1st edn., 2004. [11] H. Topsø e, J. Catal., 2003, 216, 155–164. [12] H. Kim, K. M. Kosuda, R. P. Van Duyne and P. C. Stair, Chem. Soc. Rev., 2010, 39, 4820–4844. [13] E. C. Le Ru and P.G. Etchegoin, Annu. Rev. Phys. Chem., 2012, 63, 65–87. [14] F. Inscore, C. Shende, A. Sengupta, H. Huang and S. Farquharson, Appl. Spectrosc., 2011, 65, 1004–1008. [15] M. M. Harper, K. S. McKeating and K. Faulds, Phys. Chem. Chem. Phys., 2013, 15, 5312–5328. [16] C. Vericat, M. E. Vela, G. Benitez, P. Carro and R. C. Salvarezza, Chem. Soc. Rev., 2010, 39, 1805–1834. [17] Z. Zhang, S. Sheng, H. Zheng, H. Xu and M. Sun, Nanoscale, 2014, 6, 4903–4908. [18] 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, 82–86. [19] M. Sun, Z. Zhang, Z. H. Kim, H. Zheng and H. Xu, Chem. Eur. J., 2013, 19, 14958– 14962. [20] M. Sun, Z. Zhang, H. Zheng and H. Xu, Sci. Rep., 2012, 2, 647. [21] E. M. van Schrojenstein Lantman, T. Deckert-Gaudig, A. J. G. Mank, V.Deckert and B. M. Weckhuysen, Nat. Nanotechnol., 2012, 7, 583–586. [22] K. F. Domke and B. Pettinger, ChemPhysChem, 2010, 11, 1365–1373. [23] W. T. Müller, D. L. Klein, T. Lee, J. D. Clarke, P.L. McEuen and P.G. Schultz, Science, 1995, 268, 272–273. [24] C. Blackledge, D. A. Engebretson and J. D. McDonald, Langmuir, 2000, 16, 8317– 8323. [25] B. K. Pathem, Y. B. Zheng, S. Morton, M. A. Petersen, Y. Zhao, C.-H. Chung, Y. Yang, L. Jensen, M. B. Nielsen and P.S. Weiss, Nano Lett., 2013, 13, 337–343. [26] M. Kim, J. N. Hohman, Y. Cao, K. N. Houk, H. Ma, A. K.-Y. Jen and P.S. Weiss, Science, 2011, 331, 1312–1315. [27] Y. B. Zheng, J. L. Payton, C.-H. Chung, R. Liu, S. Cheunkar, B. K. Pathem, Y. Yang, L. Jensen and P.S. Weiss, Nano Lett., 2011, 11, 3447–3452. [28] Y. B. Zheng, J. L. Payton, T.-B. Song, B. K. Pathem, Y. Zhao, H. Ma, Y. Yang, L. Jensen, A. K.-Y. Jen and P.S. Weiss, Nano Lett., 2012, 12, 5362–5368. [29] K. Kim, J. K. Yoon, H. B. Lee, D. Shin and K. S. Shin, Langmuir, 2011, 27, 4526– 4531.

66 References

[30] K. Kim, H. B. Lee, D. Shin, H. Ryoo, J. W. Lee and K. S. Shin, J. Raman Spectrosc., 2011, 42, 2112–2118. [31] K. Kim, K. L. Kim, J.-Y. Choi, D. Shin and K. S. Shin, Phys. Chem. Chem. Phys., 2011, 13, 15603–15609. [32] K. Kim, D. Shin, K. L. Kim and K. S. Shin, Phys. Chem. Chem. Phys., 2012, 14, 4095– 4100. [33] K. Kim, D. Shin, J.-Y. Choi, K. L. Kim and K. S. Shin, J. Phys. Chem. C, 2011, 115, 24960–24966. [34] K. Kim, K. L. Kim, D. Shin, J.-Y. Choi and K. S. Shin, J. Phys. Chem. C, 2012, 116, 4774–4779. [35] K. Kim, K. L. Kim, H. B. Lee and K. S. Shin, J. Phys. Chem. C, 2012, 116, 11635– 11642. [36] K. Kim, K. L. Kim and K. S. Shin, Langmuir, 2013, 29, 183–190. [37] K. Kim, K. L. Kim and K. S. Shin, J. Phys. Chem. C, 2013, 117, 5975–5981. [38] Y.-F. Huang, H.-P.Zhu, G.-K. Liu, D.-Y. Wu, B. Ren and Z.-Q. Tian, J. Am. Chem. Soc., 2010, 132, 9244–9246. [39] Y.Huang, Y.Fang, Z.-L. Yang and M. Sun, J. Phys. Chem. C, 2010, 114, 18263–18269. [40] M. Sun, Y. Huang, L. Xia, X. Chen and H. Xu, J. Phys. Chem. C, 2011, 115, 9629– 9636. [41] Y.-F. Huang, D.-Y. Wu, H.-P.Zhu, L.-B. Zhao, G.-K. Liu, B. Ren and Z.-Q. Tian, Phys. Chem. Chem. Phys., 2012, 14, 8485–8497. [42] V.Canpean and S. Astilean, Spectrochim. Acta A, 2012, 96, 862–867. [43] Y.-F. Huang, M. Zhang, L.-B. Zhao, J.-M. Feng, D.-Y. Wu, B. Ren and Z.-Q. Tian, Angew. Chem. Int. Ed., 2014, 53, 2353–2357. [44] P.Xu, L. Kang, N. H. Mack, K. S. Schanze, X. Han and H.-L. Wang, Sci. Rep., 2013, 3, 2997. [45] L.-B. Zhao, M. Zhang, Y.-F. Huang, C. T. Williams, D.-Y. Wu, B. Ren and Z.-Q. Tian, J. Phys. Chem. Lett., 2014, 5, 1259–1266. [46] K. Kim, J.-Y. Choi and K. S. Shin, J. Phys. Chem. C, 2013, 117, 11421–11427. [47] K. S. Shin, Y. K. Cho and K. Kim, Vib. Spectrosc., 2014, 70, 120–124. [48] W. A. Hayes and C. Shannon, Langmuir, 1996, 12, 3688–3694. [49] K. Kim, Y. M. Lee, H. B. Lee, Y. Park, T. Y. Bae, Y. M. Jung, C. H. Choi and K. S. Shin, J. Raman Spectrosc., 2010, 41, 187–192. [50] B. O. Skadtchenko and R. F. Aroca, Spectrochim. Acta A, 2001, 57, 1009–1016. [51] M. Osawa, N. Matsuda, K. Yoshii and I. Uchida, J. Phys. Chem., 1994, 98, 12702– 12707.

67 Chapter 3: Reaction Kinetics of the Photo-Catalytic Reduction of p-Nitrothiophenol

[52] S. Sun, R. L. Birke, J. R. Lombardi, K. P. Leung and A. Z. Genack, J. Phys. Chem., 1988, 92, 5965–5972.

68 CHAPTER 4

Separation of Time-Resolved Phenomena in Surface-Enhanced Raman Scattering of the Photo-Catalytic Reduction of p-Nitrothiophenol

Abstract

Nano-scale Raman spectroscopy tends to result in large variations in intensity and spectral features as a function of time. The main reason for these ﬂuctuations are the molecular dynamics on the nano-scale, like diffusion and rotation processes, that hamper the straightforward analysis of chemical processes on this scale. The measurement volume is linked to a discrete number of molecules, and thereby any ensemble-averaging is ruled out. This so-called blinking frustrates interpretation of the spectra and thereby limits the applicability of surface-enhanced Raman spectroscopy (SERS) for studying reaction kinetics when a high spatial resolution is of interest. This Chapter deals with a chemometrics method to separate long-term effects from short-term effects, based on the lack of ﬁt of a principal component analysis. The long-term irreversible changes in spectrum due to the reactivity of the molecules can hereby be separated from short-lived reversible effects, caused by e.g. reorientation or diffusion of molecules over a SERS substrate, or even related to short-lived chemical events like reaction intermediates. With this method, it is possible to study effects that occur on different time-scales in separate ways by taking the photo-catalytic reduction of p-nitrothiophenol as a showcase system.

This chapter is based on: E. M. van Schrojenstein Lantman, P.de Peinder, A. J. G. Mank, B. M. Weckhuysen, submitted

69 Chapter 4: Separation of Time-Resolved Phenomena in Surface-Enhanced Raman Scattering of the Photo-Catalytic Reduction of p-Nitrothiophenol

4.1. Introduction

Surface-enhanced Raman scattering (SERS) is a novel and powerful characterisation technique and slowly developing toward maturity as a tool for heterogeneous catalysis. Many effects play a role in the actual SERS spectra that are observed, and full understanding of the reasons for spectral fluctuations is important for practical application of reactivity studies on the scale of single catalytic particles. The large enhancement effects that are related with the phenomenon opens doors to nano-scale molecular characterization, that is otherwise impossible. This resolution is present both in quantity, or the number of molecules required for characterization, and in spatial resolution, especially with techniques like tip-enhanced Raman scattering (TERS). However, single hot-spots measurements require consideration in their interpretation, due to the small measurement volume, but also in their time-dependent variability. 1 Spectral fluctuations are a common feature in these time-resolved measurements. 2–4 Due to the small number of molecules and extreme enhancement effect involved in these measurements, rare events are more readily observed where they are not (or less) visible in long time-scale or bulk measurements, due to averaging over many more modes. The interpretation of short-term variations in the spectra has been the subject of debate, as the high enhancement field at a hotspot can also lead to photo-decomposition of molecules within that hotspot. 5–7 However, the spectral fluctuations are commonly explained through local changes in chemical environment around the sensed molecules, or orientation effects. 4,8 The effect of blinking is at any rate a dynamic process, dependent on temperature and mobility of the molecules on the surface. 9,10 In the case of heterogeneous catalysis, short-term fluctuations might actually give valuable information about e.g. the orientations of molecules or the presence of reaction intermediates. To study these phenomena in detail, the short-lived spectral variations must be analysed separately from long-term effects. The near-field intensity over a SERS substrate is far from homogeneous. A few select locations will have a very intense near-field, where other locations on the surface will hardly cause a Raman enhancement effect. This makes it an ideal technique to study any short-lived reaction intermediates that are otherwise insignificant in a bulk observation, which is best studied in few-hotspot samples to limit ensemble-averaging. Sufficient long measurements will eventually capture the signature spectrum of short-lived species, as it arises at the location of an intense hot-spot. All the above effects result in spectral differences between the spectral fluctuations from the regular (bulk) spectra. Therefore, the use of chemometrics can help to separate spectra of large datasets in an effective manner. Various groups have already linked SERS

70 Introduction measurements to multivariate analysis techniques. 11–14 But where signal intensity fluctuations are no problem for multivariate analysis, spectral fluctuations on various timescales can give problems in these methods. In this Chapter, the use of multivariate analysis tools is presented as a method to separate reactivity, related to bulk changes, from short-term blinking, related to orientations, intermediate states and temporary conditions. Both are important to understand the dynamics at the catalyst particle and the developed methodology allows analysis of both effects without removal of any spectra. The chemometric method used here is a combination of principal component analysis (PCA) and multivariate curve resolution (MCR). In PCA, the variance in a dataset is decomposed in a set of orthogonal principle components. Using only a limited number of these principle components, a large amount of variance of the input spectra can be described. 15. With MCR it is possible to decompose a mixture of spectra into their individual component spectra. 16 The main difference between PCA and MCR lies in the components they calculate. PCA components describe variance and can thereby also result in components with both negative and positive peaks. It is typically used as a tool to gain insight into the variance within a dataset. MCR decomposes a set of spectra into a pre-defined number of spectral components, which relate to the actual spectra of the related compound. This description of a dataset with a limited number of components usually works well for the description of bulk changes. However, short-term fluctuations, like reaction intermediates, will be ignored as they are not prominent enough in the full dataset. This does not mean they are not observed. We use PCA to analyse the full dataset and separate reaction kinetics from short-term variations. These separated datasets are subsequently examined with MCR, for direct comparison to normal Raman spectra. We present here a novel chemometric method based on PCA for time-filtering (see Fig. 4.1). Chemical reactions can hereby be separated from short-term variations, enabling better analysis of either dataset. To present this analysis method, we use a self-assembled monolayer (SAM) of p-nitrothiophenol (pNTP) on a flat Au substrate, as illustrated in Fig. 4.2. SERS-activity, and related reactivity, is introduced by deposition of Ag nanoparticles, similar to other experiments described in literature. 17 Rather than using green laser excitation, an excitation wavelength of 785 nm is chosen. This excitation excites the coupled plasmons of Au and Ag and gives sufficient enhancement effect to observe dynamics over single nanoparticle hotspots on the flat Au substrate. The choice of this low-energy excitation wavelength also ensures that the catalytic reaction is sufficiently slow to observe reactivity during the acquisition of 15 000 consecutive 1 s SERS spectra.

71 Chapter 4: Separation of Time-Resolved Phenomena in Surface-Enhanced Raman Scattering of the Photo-Catalytic Reduction of p-Nitrothiophenol

one-component PCA

Q-residuals < mean+ 0.1 x st. dev. spectral blinking For t =200 s

two-component MCR MCR

reaction kinetics component identiﬁcation

FIG. 4.1. Workﬂow for the chemometric analysis of the time-resolved SERS spectra measured. A one-component principle component analysis (PCA) is used to create a ﬁlter, separating short-term spectral blinking from reaction kinetics. Both sets of spectra can subsequently be analysed via multivariate curve resolution (MCR).

FIG. 4.2. The experimental setup consists of single Ag nanoparticles deposited on a SAM of pNTP assembled on a ﬂat gold substrate. Raman spectra have been measured at 785 nm excitation.

4.2. Experimental

4.2.1. Silver Nanoparticle Synthesis. Ag nanoparticles were synthesized according to Lee and Meisel. 18 In short, 0.04908 g silver nitrate (Sigma Aldrich, ACS reagent, 99.0%) was dissolved in 10 ml mQ water. 3.667 ml of this stock solution was added≥ to 100 ml mW water in a round-bottomed flask equipped with a Teflon stirrer bar and reflux column. After heating the solution to reflux temperature, 2 ml of 0.09 M sodium citrate tribasic dihydrate (Sigma, reagent grade, 99%) was added. After 5 min at reflux

72 Results and Discussion temperature, the sol had turned a yellow/brownish colour with silver reﬂectance. This sol was left to cool and stored in a refrigerator until required for experiments.

4.2.2. Sample Preparation. A ﬂat Au substrate of 100 nm Au over a 0.5 x 1.0 cm Si wafer (Philips Innovation Services) was cleaned with UV/ozone and anhydrous ethanol. Subsequently, the substrate was immersed into a 10 mM ethanolic solution of pNTP (Fluka, technical grade) for 24 h, followed by threefold rinsing in 10 ml ethanol. A ten- fold dilution of Ag nanoparticle sol was directly dropcasted onto the substrate and left to dry in ambient conditions.

4.2.3. Raman Spectroscopy. Raman measurements were performed on a Renishaw InVia microscope, using 785 nm diode laser excitation, through a 50x long working distance objective. The described experiment was done using 0.25 mW laser excitation power, for 15 000 consecutive spectra at a 1 s integration interval.

4.2.4. Chemometrics. Chemometric analysis was performed using the PLS Toolbox 6.71 (Eigenvector Co.) in combination with Matlab 2012a (Mathworks). Unless mentioned otherwise, raw spectra were loaded directly into Matlab and smoothed with a second-order (9 pixel) Savitsky-Golay filter. A quick spike removal was done through a home-written routine: a one-component PCA (mean centre pre-processing) was analysed on the basis of the q-residuals in time. Any spectrum with a q-residual value higher than the mean + 0.1 x standard deviation of all q-residuals was removed if the datapoint was three times larger than its direct neighbours. A five times higher filter was used for the first 100 spectra, to cope with any Q-residuals higher at the start of the reaction. The result of this filtering is the separation of the total dataset in categories that relate to spectral variations on different time-scales that van be linked to different physical and chemical changes on the surface.

4.3. Results and Discussion

4.3.1. Single Hot-Spot SERS. Obtaining SERS over single hot-spots is achieved by deposition of single nanoparticles over a pNTP functionalized ﬂat Au surface, as shown in Fig. 4.2. This is, incidentally, also a study over single-catalytic nanoparticles. A set of 15 000 spectra were recorded at 785 nm excitation wavelength, with 1 s integration time. Pre-processing was done via a cosmic-ray remover (WiRe 3.2, Renishaw) to remove the spikes in individual spectra. These narrow peaks occur only in single spectra and their removal has no effect on the time-resolved scale. Spectra are analysed for the spectral 1 region of 1000 - 1635 cm− , where most vibrations characteristic for both reactant and

73 Chapter 4: Separation of Time-Resolved Phenomena in Surface-Enhanced Raman Scattering of the Photo-Catalytic Reduction of p-Nitrothiophenol product are found. A weighted-least squares (WLS) baseline correction is calculated, after which the absolute value of this dataset is taken. The main peaks of pNTP and DMAB (see Appendix A) are clearly visible in the colour plot (Fig. 4.3a), as well as in the first and last spectrum of the series (Fig. 4.3b). A simple method for tracking the reaction progress is to plot peak areas of reactant and product as a function of time, as was also used in Chapter 3. Fig. 4.3c shows the results of this approach. The main trends are visible: the DMAB comes up with time, whereas the pNTP signal appears to be less affected by the reaction and suffers more from SERS- intensity variations, as observed previously in Chapter 3. The noise level in the signals is rather high, most likely due to the noise in the spectra themselves, though peak shifts also contribute to the noise level. This method is sufficient for monitoring the (irreversible) reaction in time, but shows quite some noise in the signal. MCR is a more reliable method, as it takes the full spectrum into account, rather than just a single peak. For this dataset, an unrestrained two-component MCR analysis results in a model that describes 94% of the total signal variance. Component 1 (62%, see Fig. 4.3e) resembles the reactant very well, while the second component (32%) comprises a mixture of DMAB with a smaller contribution of pNTP.The corresponding scores in time (Fig. 4.3f) show more detail and less noise respect to the peak area analysis method (Fig. 4.3c). MCR can therefore be concluded to be a satisfactory method to track this reaction. Nonetheless, the spectral residuals after MCR (Fig. 4.3d) still show many spectral features. Blinking is a prominent feature in this dataset and occurs in spectral patterns that do not overlap with either pNTP and DMAB. As this photo-reaction encompasses multiple reaction steps, 19 these spectra can contain valuable information regarding the mechanism. In fact, when time- and spatial-resolution of SERS are high enough, these spectral fluctuations are expected to be the way to identify possible reaction intermediates.

4.3.2. Separation of Signal Variations on Different Time-Scales. To separate short- term (reversible) fluctuations from the long-term reaction progress, a new method of data-processing is introduced. The photo-reduction of pNTP to DMAB can largely be characterized by the disappearance of the pNTP spectrum, and the rise of the DMAB spectrum. After a simple spike removal, PCA is used to separate long-term trends from short-term trends, using a 200 s time-filter. The exact window required for the time-filter depends on the dynamics of the system. A one-component PCA on the time-series (after spike removal) results in the component and score shown in Figs. 4.4a and 4.4b. The spectral pattern of the component shows great resemblance to the spectrum of DMAB, and describes 56% of the variance in the dataset. The score plot of this component in time is similar to that expected for a reaction,

74 Results and Discussion

FIG. 4.3. a) Contour plot of the Raman spectrum in time. b) First and last spectrum. c) pNTP (green) and DMAB (blue) peak area as a function of time. d) Residuals after two-component MCR. e) Calculated MCR components, component 1 (green) describes 62% of the variance in the dataset, component 2 (blue) 32%, resulting in a total of 94% of the variance explained by the two- component MCR model. f) MCR scores in time.

75 Chapter 4: Separation of Time-Resolved Phenomena in Surface-Enhanced Raman Scattering of the Photo-Catalytic Reduction of p-Nitrothiophenol

FIG. 4.4. One-component PCA calculation of the full dataset. a) Loading of component 1 (56% variance captured), b) score of component 1 in time, c) separation of spectra into the three categories, black and blue spectra describe the bulk reaction, red spectra are spectral fluctuations and will be analysed separately later. d) Q-residuals in time with 200 s filter. with a slow increase in loading as time progresses. Hardly any short-term fluctuations are visible. The q-residuals in time (Fig. 4.4d) are a way of plotting lack-of-fit for all spectra. These show a stable (low) baseline, but with quite some short-term deviation from this baseline. The PCA component does not describe those spectra very well as they are present only for short periods during the total 15 000 s. In other words, they show some form of spectral deviation from the bulk reaction. To separate these spectral fluctuations for closer examination, any q-residuals with a value higher than the mean value + 0.1 x the standard deviation of all q-residuals are selected, unless they consist of a series of spectra that are above this norm for 200 s consecutively. The height of this q-residual criterion is

76 Results and Discussion depicted by the blue line in Fig. 4.4d. At the start of the reaction, the PCA does not always fit the full trend of reactivity, therefore the q-residual criterion of the first 100 spectra is a five-fold higher value to ensure no reactivity information is lost. In Fig. 4.4d, all black data points represent spectra that have a sufficiently low q- residual value, and are taken for further analysis of bulk reaction. In this first separation, the green data points are also included. They are above the set q-residual criterion, but persistent enough to be considered as bulk processes. Red data points are below the q-residual criterion as the corresponding spectra are only present during short time- intervals. They are set aside and can thus be analysed separately from the bulk reaction. The colour coding in Fig. 4.4d is also taken to Figs. 4.4b and c, to indicate the relation between these different parameters. Fig. 4.4c shows the breakdown of the full set into the different sets of spectra. Black spectra fall within the Q-residual criterion and describe the reaction. Green spectra do not meet up to the Q-residual criterion, but make it through the time-filter. They are likely a orientation-effect due to changes in the monolayer, and are included in the analysis of the bulk reaction in Fig. 4.5. Red spectra are set aside and will be analysed in Fig. 4.6. The spectral pattern of these three sets is quite clear. With respect to the black spectra, the green spectra are only small deviations. In contrast, the red spectra show much more vibrational bands than observed in the black spectra.

4.3.3. Analysis of Bulk Surface Reaction. Having separated the photo-catalytic reduction from short-term variations, each dataset can now be analysed separately. The reaction spectra are taken through a two-component MCR, after pre-processing of the spectra by a WLS baseline correction and taking the absolute value of the resulting spectra. Other than assuming two components to describe the reaction, no other constraints are put onto the MCR. The calculated MCR components are shown in Fig. 4.5a and depict pNTP (in green, 27% variance) and a mixture of pNTP and DMAB (in blue, 70% variance). The score as a function of time is shown in Fig. 4.5b, and, like the components, rather similar to the results obtained in Fig. 4.3e and f. The main difference is the level of noise in the score, which is lower in Fig. 4.5b. The removal of spectral fluctuations and subsequent analysis of the pure reaction has led to better description of this reaction via MCR. As can be expected, the variance explained by two MCR components within this filtered dataset is slightly higher (97%) than in the non-filtered dataset (94%). This is further emphasised by the input spectra (Fig. 4.5c) and spectral residuals (Fig. 4.5d). Less spectral fluctuations are put into the model, so less are found in the spectral residuals, when comparing Fig. 4.5d to 4.3d.

77 Chapter 4: Separation of Time-Resolved Phenomena in Surface-Enhanced Raman Scattering of the Photo-Catalytic Reduction of p-Nitrothiophenol

FIG. 4.5. MCR model of all spectra belonging to the bulk reaction. In a) the two calculated components are shown. Component 1 (green) represents 27% of the calculated model, component 2 (blue) represents 70% of the calculated model, with a total of 97% variance explained. b) Shows the score of component 1 and 2 in time. c) All input spectra for the MCR model and d) the spectral residuals that are not explained by the MCR components. These two ﬁgures are shown on the same intensity scale for comparison.

Overall, the time-filter benefits the MCR model. But for application of this model to the description of reaction kinetics as in Chapter 3, some additional steps will have to be taken. The mixture of pNTP and DMAB in component 2 makes it difficult to determine the fraction of pNTP in the total signal. It will likely be a linear combination of both component 1 and 2. DMAB is simpler, as that is only expressed in component 2. Though the ‘actual’ score of DMAB will only be a fraction of the score for component 2, this will be a matter of scaling the intensity and will not affect the calculation of reaction rate.

78 Results and Discussion

FIG. 4.6. MCR model of all short-lived spectral components. a) The four calculated MCR components. The variance capture with the components is for black 52%, for red 24%, for blue 13%, and for green 5%. b) Score plot in time for these four components, colour-coded as for a). c) All input spectra for this MCR analysis d) Residuals after MCR analysis.

4.3.4. Analysis of Spectral Fluctuations. Most important in this novel chemometric method, is its ability to analyse short-term spectral fluctuations in large dataset. These fluctuations are expected to give information on short-lived events on the surface of the Ag substrate, which could be either due to reorientation of pNTP or DMAB, or a reaction intermediate. Fig. 4.4 showed how variations present for less than 200 s were separated from the long-term effects of the bulk reaction. The separated spectra that exhibit these short-term fluctuations are shown in Fig. 4.6c. Many spectral features are present, and at varying intensities. A four-component MCR analysis was found to describe the majority (94%) of the spectral components in this dataset. The largest component (Fig. 4.6a and b, black, 52%))

79 Chapter 4: Separation of Time-Resolved Phenomena in Surface-Enhanced Raman Scattering of the Photo-Catalytic Reduction of p-Nitrothiophenol has a maximum contribution at around 4 000 s in the dataset. It is very similar to component 1 in Fig. 4.5a, and is also most prominent in the beginning of the reaction. In comparison to the reactant component in Fig. 4.5, a slight variation in peak ratios is observed, which is most likely caused by a different orientation or local environment for the reactant. Similarly, the second component (Fig. 4.6a and b, red, 24%) has similarities to component 2 in Fig. 4.5a, and is present at the end of the reaction. Again, the differences are in peak ratios, likely caused by molecular orientation or variations in the local environment. The third (Fig. 4.6a and b, blue, 13%) and fourth (Fig. 4.6a and b, green, 5%) component show a different in their behaviour in the score in time. Both are rather low in intensity, apart from one instance - at 11 000 for component 3, and 12 500 for component 1 4. The most characteristic peak in the third component is the vibration at 1380 cm− . This is a known vibration of DMAB (see Appendix A), and appears to be strongly enhanced at that specific moment. The fourth component is hard to place in the context of either pNTP or DMAB, and could be a reaction intermediate. Reaction intermediates will have a similar molecular structure as pNTP and DMAB, but with a different functional group. The characteristic 1 vibrations at e.g. 1335 and 1440 cm− would not be expected at exactly that location in a intermediate spectrum. Also, reorientation of either pNTP and DMAB should not result in new vibrations with respect to the normal Raman spectrum However, density functional theory (DFT) calculations should be able to shed more light on the chemical information in these spectral components. The novelty of this chemometric method is its capability of analysing short-lived spectra. Though most deviating spectra will probably be explained via orientation effects, there is a small chance that intermediates will also be present at the intense enhancement field in a SERS hotspot, if the measurement runs long enough. Normally short-lived spectra are disregarded in the analysis of a 15 000 spectra dataset, but now these spectra can be identified for the study of the surface dynamics.

4.4. Conclusions

We have shown a new approach to chemometrics of SERS spectra. With a one-step time-ﬁlter with a PCA approach, the photo-catalytic reduction of pNTP monitored by SERS was analysed. With this PCA analysis, the reaction (> 200 s) were separated from strong spectral deviations occurring over shorter time intervals (< 200 s). Both sets of data were then separately analysed.

80 References

Unfiltered MCR analysis on the reaction data yields less noise, compared to simple peak area analysis and to unfiltered MCR. Additionally, the blinking spectra separated from the bulk reaction can now be analysed separately. Here, a four-component MCR analysis has shown the main components of these deviating spectra. Where the bulk reaction data cleans up nicely with the use of this filter, the real advantage is in the blinking spectra. Here, we have shown that this approach is able to isolate these events. If SERS is going to reveal reaction intermediates, it will be during sparse events. It is important to note that SERS has this inherent capability due to the inhomogeneous enhancement effect over the SERS-active substrate. Short-lived events in an intense hotspot will show up in SERS-spectra, if the single-hotspot approach is used.

4.5. Acknowledgements

Arjan Mank (Philips) is acknowledged for the Raman experiments, while Peter de Peinder (VibSpec) is thanked for the development of the chemometric method.

References

[1] J. Stadler, T. Schmid and R. Zenobi, Nanoscale, 2012, 4, 1856–1870. [2] T. Schmid, L. Opilik, C. Blum and R. Zenobi, Angew. Chem. Int. Ed., 2013, 52, 5940– 5954. [3] W. Zhang, B.-S. Yeo, T. Schmid and R. Zenobi, J. Phys. Chem. C, 2007, 111, 1733– 1738. [4] D. R. Ward, N. K. Grady, C. S. Levin, N. J. Halas, Y. Wu, P.Nordlander and D. Natelson, Nano Lett., 2007, 7, 1396–1400. [5] C. Neacsu, J. Dreyer, N. Behr and M. Raschke, Phys. Rev. B, 2006, 73, 193406. [6] K. Domke and B. Pettinger, Phys. Rev. B, 2007, 75, 236401. [7] C. Neacsu, J. Dreyer, N. Behr and M. Raschke, Phys. Rev. B, 2007, 75, 236402. [8] T. Ichimura, H. Watanabe, Y. Morita, P.Verma, S. Kawata and Y. Inouye, J. Phys. Chem. C, 2007, 111, 9460–9464. [9] Z. Wang and L. J. Rothberg, J. Phys. Chem. B, 2005, 109, 3387–3391. [10] Y. Maruyama, M. Ishikawa and M. Futamata, J. Phys. Chem. B, 2004, 108, 673–678. [11] B. R. Wood, E. Bailo, M. A. Khiavi, L. Tilley, S. Deed, T. Deckert-Gaudig, D. McNaughton and V.Deckert, Nano Lett., 2011, 11, 1868–1873. [12] M. Richter, M. Hedegaard, T.Deckert-Gaudig, P.Lampen and V.Deckert, Small, 2011, 7, 209–214. [13] W. Cheung, I. T. Shadi, Y. Xu and R. Goodacre, J. Phys. Chem. C, 2010, 114, 7285– 7290.

81 Chapter 4: Separation of Time-Resolved Phenomena in Surface-Enhanced Raman Scattering of the Photo-Catalytic Reduction of p-Nitrothiophenol

[14] W.Xie, C. Herrmann, K. Kömpe, M. Haase and S. Schlücker, J. Am. Chem. Soc., 2011, 133, 19302–19305. [15] R. Bro and A. K. Smilde, Anal. Methods, 2014, 6, 2812–2831. [16] A. de Juan, J. Jaumot and R. Tauler, Anal. Methods, 2014, 6, 4964–4976. [17] W.-H. Park and Z. H. Kim, Nano Lett., 2010, 10, 4040–4048. [18] P.C. Lee and D. Meisel, J. Phys. Chem., 1982, 86, 3391–3395. [19] S. Sun, R. L. Birke, J. R. Lombardi, K. P. Leung and A. Z. Genack, J. Phys. Chem., 1988, 92, 5965–5972.

82 CHAPTER 5

The Photo-Catalytic Reduction of p-Nitrothiophenol Monitored at the Nanoscale Using Tip-Enhanced Raman Spectroscopy

Abstract

Recent advances in optical spectroscopic approaches allow spatio-temporal information to be obtained while still providing molecular information. However, these methods do not have the nanometre-scale spatial resolution required for the study of molecular dynamics over single catalytic particles and/or require the use of ﬂuorescent labels. In this Chapter, it is shown that time-resolved tip-enhanced Raman scattering (TERS) can monitor photo- catalytic reactions over a single catalytic particle (< 10 nm). A silver-coated atomic force microscope tip is used to both enhance the Raman signal and to act as the catalyst. The tip is placed in contact with a self-assembled monolayer (SAM) of p-nitrothiophenol (pNTP) molecules adsorbed on gold nanoplates. The photo-catalytic reduction process of pNTP is induced at the apex of the tip with green laser light, while red laser light is used to monitor the conversion. This dual-wavelength approach can also be used to observe other molecular effects, such as diffusion within SAMs. Finally, the chemometric method developed in Chapter 4 is applied for improved analysis of the TERS data. TERS is shown to be capable of following reactions on the level of a single catalytic particle, while the application of advanced chemometric tools separates the reaction from short-lived species or orientations.

This chapter is based on: E. M. van Schrojenstein Lantman, T. Deckert-Gaudig, A. J. G. Mank, V. Deckert, B. M. Weckhuysen, Nature Nanotech. 2012, 7, 773-781.

83 Chapter 5: The Photo-Catalytic Reduction of p-Nitrothiophenol Monitored at the Nanoscale Using Tip-Enhanced Raman Spectroscopy

5.1. Introduction

Various spectroscopic techniques are used to study heterogeneous catalytic materials at work, including infrared, ultraviolet-visible, Raman or fluorescence spectroscopy. 1–3 At these wavelengths molecular senstive information can be obtained, which is very informative for reaction studies. However, these techniques are usually limited in terms of their spatial resolution to the micrometre scale because of the diffraction limit of light. The techniques that do have the spatial resolution required to look at individual catalytic particles often only image dense elements and therefore only the catalytic particle itself. 1,4 Fortunately, recent advances have opened up possibilities to combine the spatial resolution with the chemical sensitivity. 5–7 Where Chapter 3 has shown that SERS measurements can yield valuable information on a ‘bulk’ scale, the resolution of the measurement was still limited by the diffraction limit of light, and covered a large ensemble of catalytic particles. To go down to single catalytic particles, one needs to ensure that only one catalytic particle is present within the observation volume. Chapter 4 uses sparsely distributed nanoparticles over a Au surface, yet no control of location of nanoparticles can be obtained with this method. TERS is a promising nano-spectroscopic technique that combines atomic force microscopy (AFM) with the phenomenon of surface-enhanced Raman scattering (SERS). Through a thin coating of silver or gold on an AFM tip, a dramatic intensity boost of the Raman signal can be obtained at the end of the tip, while maintaining the spatial resolution of the AFM 8–12 (see Fig. 5.1). Following excitation with an appropriate wavelength, a strong field enhancement is induced at the apex of the coated tip. This results in highly enhanced Raman scattering signals in molecules in the direct vicinity of the tip. The observation area has been estimated to have a diameter of 10 nm 13, well below the diffraction limit, and as a near-field effect independent of the≤ wavelength used. Although the combination of TERS and (heterogeneous) catalysis is very promising, only a limited number of preliminary studies can be found in the literature to date. 14,15 None of these studies goes as far as studying a catalytic reaction in real time. In the present work, a photo-catalytic process, the reduction of pNTP, was selected where the silver-coated TERS tip itself acts as the catalyst. Self-assembly of the molecules on gold nanoplates ensures that no more than a monolayer of reactant is studied. These nanoplates also create a larger TERS enhancement by the formation of a gap-plasmon mode when the TERS-tip is in close contact. 13 As the full SERS activity is localized at the tip- apex, a nano-scale measurement volume is created, with full flexibility as to the material under study and position on the sample. Only the utmost point of the tip is in contact with the reactant, making it is a first-of-a-kind study of a single catalytic ‘particle’ in action.

84 Experimental

FIG. 5.1. (a) Schematic overview of the experimental setup. A monolayer of pNTP is assembled on a sample of gold nanoplates on glass. This sample is placed on a combination of an inverted Raman microscope and an upright AFM. Optical excitation and collection of the spectra occurs from below, with the facility to switch between green (532 nm) and red (633 nm) excitation. A silver-coated AFM tip is used as a catalyst for the reduction of pNTP to DMAB at 532 nm. It also provides the electromagnetic ﬁeld enhancement required for TERS at 633 nm. In b) the actual TERS setup is shown. The sample is sandwiched between the AFM on top and the optical microscope below. Photograph taken with permission from Volker Deckert, Institute of photonic Technology and Physical Chemistry, Jena, Germany.

The combination of two laser sources is crucial for controlled observation of the reaction at the catalytic site. On excitation with a short wavelength, a reaction can be temporarily activated. A longer-wavelength excitation is dedicated to monitoring. Thus it is possible to differentiate between the processes that occur naturally on the sample (such as diffusion and orientation effects) and the actual photo-catalytic reduction of pNTP. It is important to note that the measurement volume in TERS is such that no ensemble averaging takes place and spectral variations due to individual events are much more prominent.

5.2. Experimental

TERS measurements were performed as previously described in literature. 16 The extreme signal-enhancement required for the present study can only be obtained if the combination and alignment of the TERS-active AFM tip and the optical microscope is at its best. TERS-active tips are prepared in-house, and stored under Ar until use, but for one week at most. An example of a TERS-tip is shown in Fig. 5.2a. A high numerical aperture (NA 1.4), oil-immersion objective is used to focus the light. It also creates some

85 Chapter 5: The Photo-Catalytic Reduction of p-Nitrothiophenol Monitored at the Nanoscale Using Tip-Enhanced Raman Spectroscopy

FIG. 5.2. a) A Scanning electron microscopy image of a typical (20 nm) silver-coated TERS tip. b) A typical Au nanoplate for SAM assembly, visualized through AFM. z-polarized light to excite the most efﬁcient (gap-)plasmon mode between the TERS tip and the sample. 13 As laser alignment within the microscope cannot be altered, an AFM tip-scanner is used to align the TERS tip to the laser focus. Then an AFM-sample scanner will scan the sample between the TERS tip and optical microscope, this means that the optical alignment is kept in place, while the structural and Raman information is obtained by manoeuvring the sample only. For excitation, either a 532 nm or 633 nm is used. As the TERS tip is on the top-side of the sample, the samples need to be transparent to allow both laser excitation and signal collection through the sample. To ensure only a monolayer coverage of pNTP and a ﬁxed starting orientation of the analyte, a gold substrate is required. The use of a single monolayer of reactants prevents interference from other molecules. Gold nanoplates were synthesized with a thickness of maximum 60 nm, following a previously reported synthesis procedure. Full self-assembled monolayers (SAMs, 100% surface coverage) were grown on the gold nanoplates by immersing the substrates for 24 h in a 9 mM ethanolic pNTP solution, followed by drying under vacuum. Samples were subsequently stored under argon. To obtain a sample with partial surface coverage, the substrate was immersed for a shorter period of time: 50% coverage is observed after 12 h in 3 mM ethanolic pNTP solution, as demonstrated in Fig. 5.3 for SAMs of pNTP. 17 1 All spectra were preprocessed by normalization on the basis of the 520 cm− Si-Si resonance of the AFM tip and subsequent linear baseline subtraction. Chemometric analysis was performed using the PLS Toolbox 6.71 (Eigenvector Co.) in combination with Matlab 2012a (Mathworks). In this case, raw spectra were processed with a quick spike-removal routine before a fourth-order weighted-least-squares baseline

86 Results and Discussion

FIG. 5.3. AFM height (a, colour scale: 0 - 45 nm) and phase (b, colour scale: 0 - 14 ◦) images of a Au nanoplate with partial surface coverage of pNTP (12 h immersion, 3.6 mM pNTP). The same measurement for a Au nanoplate (c, colour scale: 0 - 120 nm, and d, colour scale: -27 - 29 ◦) with a full surface coverage (12 h immersion, 7.0 mM pNTP). The measurements were performed with a regular Si AFM-tip in tapping mode on a Bruker Nano Dimension 3100. correction, taking the absolute value for further analysis. A one-component principle component analysis (PCA) with mean-centre preprocessing was performed on the basis of q-residuals in time. Any spectrum with a q-residual value higher than the mean + 0.1 x the standard deviation of all q-residuals in time was set aside as a short-term phenomenon if present for less than 30 s (time ﬁlter). The other spectra, described by a long-term trend, were analysed via multivariate curve resolution (MCR) with preprocessing via 9- pixel second-order Savitsky-Golay ﬁlter and subsequent normalization on the basis of the full spectrum area.

5.3. Results and Discussion

5.3.1. Monitoring SAMs with TERS. For TERS studies of these samples, the AFM tip was positioned in the focus of the excitation laser by means of tip scanning. Subsequently, a sizeable gold nanoplate with a SAM was selected using the sample scanning mode of the AFM to give the alignment schematically shown in Fig. 5.1. Both laser excitation and collection of the TERS spectra were performed along the same optical path, accessing the

87 Chapter 5: The Photo-Catalytic Reduction of p-Nitrothiophenol Monitored at the Nanoscale Using Tip-Enhanced Raman Spectroscopy sample from below. 16 All (time-resolved) measurements were carried out by positioning the tip on a gold nanoplate and subsequent monitoring the Raman signal over time at that location. There is a risk that molecule are transferred from the SAM to the clean Ag surface of the tip. This would result in contamination of the Raman signal and prevent nanoscale imaging of the SAM. Reference measurements were also taken next to the gold nanoplates, to ensure a clean tip. These were carried out on the glass substrate in close proximity to the nanoplate under study. To use a SAM in a reaction, a crucial prerequisite is a stable starting point, that is, a complete SAM. A full surface coverage leads to a constant TERS signal with 633 nm excitation, as shown in Fig. 5.4a. The same signal is observed for at least an hour. A partial, 50% surface coverage, as in Fig. 5.4b typically results in blinking or ﬂuctuations in the intensity on a timescale of several seconds. However, the presence of pNTP can still be recognized in the time-averaged spectrum in Fig. 5.4d. Usually, dynamic effects are associated with the single-molecule regime. 18–20 Here, the small sampling volume of TERS probably leads to the observation of orientation and diffusion dynamics within SAMs (as previously described); these are much more pronounced in an incomplete SAM than in a close-packed SAM.

5.3.2. Dual-Wavelength TERS. To study the actual reaction on a nanometre scale, excitation and analysis is conducted using just one wavelength, 532 nm. In that case, all molecules in the observation area are immediately converted to p,p’-dimercaptoazobisbenzene (DMAB) (Fig. 5.4c) and no information can be obtained about the reaction rate. To circumvent the problem above, we have studied the conversion using a combination of two wavelengths: 633 nm excitation to monitor the molecules on the sample, and short periods of 532 nm excitation to activate the reaction. To control the rate of reaction, low laser power (0.7 µW) was used at 532 nm. To be able to link any observations on a molecular scale to this photo-reaction, it is crucial to have a stable starting point. This means that a continuous monolayer is required. Before starting the reaction, the TERS signal was monitored at 633 nm excitation for at least 1 min to ensure the signal is stable. Subsequently, the irradiation source was switched to 532 nm for a given time interval (for example, 30 s), then immediately switched back to 633 nm to follow the reaction as function of time. The results of this novel TERS mode are shown in Fig. 5.5a. Irradiating with green light changes the stable and complete pNTP monolayer (top) into a monolayer in which multiple states are present. The main components are pNTP and DMAB (Fig. 5.5b) and 1 a plot of the relative peak areas of bands at 1335 cm− (pNTP,see Appendix A) and 1440 1 cm− (DMAB, see Appendix A), shown in Fig. 5.5c, conﬁrms the catalytic activity by the

88 Results and Discussion

FIG. 5.4. Time-dependent TERS measurements of pNTP SAM on a gold nanoplate substrate. a) A total of 150 spectra with 2 s integration time and 633 nm excitation (436 µW) with a full surface coverage. b) Sixty spectra with 5 s integration time and 633 nm excitation (530 µW) with partial surface coverage (50%). c) Sixty spectra with 5 s integration time and 532 nm excitation (0.80 mW) with full surface coverage. d) Time-averaged spectra and reference spectra of the series in a) 1 (i and ii), b) (iii and iv) and c) (v and vi). Asterisks in a-d indicate the location of the 950 cm− band of the SiO2 signal of the glass substrate.

appearance of the product peaks after exposure to the green light. The simultaneously induced intensity ﬂuctuations, similar in behaviour to that in Fig. 5.4b, show that the photo-reaction upsets the monolayer stability. The change in monolayer structure can be explained directly by the reaction itself - which is a dimerization and thereby likely reorganization on a molecular level. 21 Both the ﬂuctuations and increase in DMAB con- tent are reproducible observations, and are found to be typical for time-resolved TERS measurements. Similar experiments can be found in Appendix B.1.

89 Chapter 5: The Photo-Catalytic Reduction of p-Nitrothiophenol Monitored at the Nanoscale Using Tip-Enhanced Raman Spectroscopy

FIG. 5.5. Time-dependent TERS measurement before and after 532 nm exposure. a) Time- dependent TERS spectra at 633 nm excitation (5 s integration time, 380 µW) shown before (top) and after (below, white band) illumination. b) Two spectra from a) are shown: spectrum (i) is taken at 90 s and spectrum (ii) at 265 s, respectively before and after 532 nm exposure. Spectrum (iii) is the reference spectrum taken on glass. Asterisks in a) and b) indicate the location of the 950 1 cm− band of the SiO2-glass signal of the glass substrate. c) peak areas as function of time for the 1 1 pNTP band at 1335 cm− (i) and for the band at 1440 cm− (ii), belonging to DMAB. The period of green illumination between 100 and 130 s is indicated by the shaded band.

5.3.3. Chemometrics for TERS. In Fig. 5.5, the catalytic conversion is tracked via peak area calculations. This is a valuable method of reaction tracking, and has also been used in Chapter 3 to quantify reaction kinetics. However, the small ensemble of molecules in single-hot-spot measurements typically results in spectral blinking, including slight peak shifts. In the photo-reduction of pNTP,especially the area between 1300 and 1 1500 cm− shows much variation in peak position and superposition of spectral features on top of long-term trends. This can be caused by one of many variable, that are introduced by the small ensemble of molecules under study. These include reorganisation of the SAM during the reaction, diffusion of molecules over the surface, and the presence of reaction intermediates in the hotspot. A more accurate analysis would include the full spectrum in quantification of the contribution of a specific compound, but also requires separation of short-term variations from the reaction. A method for this separation was described in Chapter 4 and is here applied to TERS spectra. After this separation, chemometric analysis will result in better quantification of the reduction of pNTP,and separate analysis of spectral fluctuations. Slight alterations to the original method are made. As only 70 spectra are recorded in the measurement in Fig. 5.5, and at a longer time-interval, fewer short-term fluctuations

90 Results and Discussion a) b) 0.25 0.5

0.2

0.15

0.1

0.05 0 PCA score

PCA loading 0

−0.05

−0.1

−0.5 1600 1500 1400 1300 1200 1100 1000 0 100 200 300 400 Raman shift (cm−1) Time (s) c) d) 0.5

0.25 0.4 0.2 0.3 0.15

0.2 0.1 Q residuals Intensity (a.u.)

0.05 0.1

0 0 1600 1500 1400 1300 1200 1100 1000 0 100 200 300 400 Raman shift (cm−1) Time (s)

FIG. 5.6. One-component PCA calculation of the TERS dataset as presented in Fig. 5.5 in Chapter 5. a) First principle component (62% variance captured), and b) corresponding score as function of time. c) Shows the separation of spectra into two categories: black spectra describe the bulk reaction, while red spectra are spectral fluctuations. d) Q-residuals in time with 30 s filter. will thus expected to be less frequent. The results of the initial one-component PCA are shown in Fig. 5.6. The calculated component shows spectral features related to the reduction of pNTP and formation of DMAB. Its score in time also shows an evolution expected of a reaction - before illumination ( 100 s) hardly any variation is visibly, afterwards the score rises and stabilizes at a higher∼ value. Q-residuals are again used as selection criterium, with their mean value + the standard deviation as the boundary condition (illustrated in Fig. 5.6d). No long-term spectral fluctuations are observed above this selection criterium. The time-filter was adjusted to 30 s accordingly. Five spectra are hereby excluded from the long-term trend. As clearly

91 Chapter 5: The Photo-Catalytic Reduction of p-Nitrothiophenol Monitored at the Nanoscale Using Tip-Enhanced Raman Spectroscopy observable in Fig. 5.6c, these are strong spectral variations. The spectra of the reduction will be analysed via MCR, while the ‘blinking’ spectra will be examined separately.

5.3.4. Analysis of Photo-Catalytic Reduction Reaction. The long-term phenomenon describe the ‘bulk’ reduction of pNTP.The conversion of pNTP to DMAB is described via a two-component MCR model, shown in Fig. 5.7. The two components are similar to those found in Chapter 4, and describe pNTP (blue, 48%) and a combination of pNTP and DMAB (green, 49%). They show a clear conversion pattern in time, as the scores of the blue component (reactant) go down in value after reaction. At the same time, the green component (product) is not visible before reaction, but present as the main feature after the 532 nm illumination. The input spectra (Fig. 5.7c) are rather well described by the MCR model, though a few spectral features are present in the residual spectra (Fig. 5.7d). These non-explained 1 features are present at 1140, 1360 and 1440 cm− , and indicate either an extra component or an orientation state. A three component MCR analysis of the same dataset describes this additional com- 1 ponent, as shown in Fig. 5.8. Most characteristic is the vibration at 1360 cm− . This vibration has previously been observed in chemisorbed pATP molecules, 22 but not in 1 combination with the vibrations at 1140 and 1440 cm− . Are more likely explanation is chemisorption of the azo-group of DMAB onto the TERS-tip, which can shift the related resonances. 23

5.3.5. Analysis of Short-Term Events. The original TERS dataset contains information on the conversion of pNTP,as well as other sources of short-term variation: ‘blinking’. These spectra are only present for a short period of time and have too much spectral variation to be explained by the pNTP reduction process. Fig. 5.9 shows these six spectra, and highlights their main spectral deviations. Taking them out of the reaction evaluation enhanced the data on reaction kinetics which already yields an improvement over con- ventional chemometrics. Interpretation of these spectra is not possible at this moment, and likely requires in- depth density functional theory (DFT) calculations for assignment of these vibrations to specific molecular orientations or configurations. Because of the enhanced analysis of the spectra, shorter integration times for TERS could be used. This results in even better defined reaction kinetics and identification of short-term events. Some of the separated 1 spectra observed here, including those with a vibration at 1500 cm− (ring vibration) 1 might be due to orientation effects. Other vibrations, e.g. between 1200 and 1300 cm−

92 Conclusions a) b) 1 0.25

0.8 0.2

0.15 0.6 Score 0.1 0.4 Intensity (a.u.)

0.05 0.2

0 0 1600 1500 1400 1300 1200 1100 1000 0 100 200 300 400 Raman shift (cm−1) Time (s) c) d)

0.25 0.25

0.2 0.2

0.15 0.15

0.1 0.1

Intensity (a.u.) 0.05 0.05

0 Residual intensity (a.u.) 0

−0.05 −0.05 1600 1500 1400 1300 1200 1100 1000 1600 1500 1400 1300 1200 1100 1000 Raman shift (cm−1) Raman shift (cm−1)

FIG. 5.7. Two-component MCR calculation of the bulk reaction spectra in the TERS dataset. a) Two components describing the reaction. Component 1 (blue) represents 48% of the variance in the dataset, component 2 (green) represents 49% of the variance, with a total of 97% variance explained. b) Shows the corresponding scores in time. c) All input spectra, after preprocessing, and d) the spectral residuals that are not explained by this MCR model. Please note that c and d are shown on the same intensity scale for comparison. cannot be assigned to either pNTP,DMAB, or even p-aminothiophenol, and are most probably related to reaction intermediates.

5.4. Conclusions

TERS is unique in enabling the study of molecule dynamics and chemical reactions on a nanometre scale, far from the averaging effects of standard experiments. The dual- wavelength approach discussed in this Chapter makes it possible to probe the dynamic

93 Chapter 5: The Photo-Catalytic Reduction of p-Nitrothiophenol Monitored at the Nanoscale Using Tip-Enhanced Raman Spectroscopy a) b) 1

0.25 0.8

0.2 0.6 0.15 Score 0.4

Intensity (a.u.) 0.1

0.2 0.05

0 0 1600 1500 1400 1300 1200 1100 1000 0 100 200 300 400 Raman shift (cm−1) Time (s) c) d)

0.25 0.25

0.2 0.2

0.15 0.15

0.1 0.1

Intensity (a.u.) 0.05 0.05 Residual intensity (a.u.) 0 0

−0.05 −0.05 1600 1500 1400 1300 1200 1100 1000 1600 1500 1400 1300 1200 1100 1000 Raman shift (cm−1) Raman shift (cm−1)

FIG. 5.8. Three-component MCR calculation of the bulk reaction in the TERS dataset. a) The three components describing the reaction. Component 1 (blue) represents 53% of variance in the dataset, while component 2 (green) represents 14% and component 3 (red) represents 31% of the total variance in the dataset. In total 98% of the variance is explained in this model. c) shows all input spectra, and d) the residuals that could not be explained via this MCR. c and d are shown on the same intensity scale for comparison.

behaviour of the reactants and the photo-reaction of pNTP separately. Molecular effects such as monolayer instability have been observed, as well as a reaction on a single catalytic particle. For these dynamic studies, the high spatial resolution of TERS is highly advantageous. Particularly for heterogeneous catalysis, its use for the study of single catalytic particle shows much promise. Combination of TERS with advanced chemometrics has shown to require three components to describe the reduction process. Additional analysis of short-lived species provides a wealth of spectral vibrations that allude to local

94 References

0.8

0.7 220 s 0.6 215 s 0.5

0.4 165 s

0.3 Intensity (a.u.) 160 s 0.2

0.1 150 s 0 1600 1500 1400 1300 1200 1100 1000 Raman shift (cm−1)

FIG. 5.9. All short-lived spectra that were separated from the long-term reaction in Fig. 5.6. . The grey bands highlight vibrations that are different to the normal spectra related with the reduction of pNTP.

ﬂuctuations in the state of the surface species, being it either reactants, reaction intermediates or reaction products.

5.5. Acknowledgements

Tanja Deckert-Gaudig (Institute of Photonic Technology, Jena, Germany) and Volker Deckert (Institute of Photonic Technology and Physical Chemistry, Friedrich-Schiller Uni- versity Jena, Jena, Germany) are thanked for their help with the TERS measurements, and for their ﬂexibility regarding the use of the experimental setup.

References

[1] B. M. Weckhuysen, Angew. Chem. Int. Ed., 2009, 48, 4910–4943. [2] C. Lamberti, A. Zecchina, E. Groppo and S. Bordiga, Chem. Soc. Rev., 2010, 39, 4951– 5001. [3] I. L. C. Buurmans and B. M. Weckhuysen, Nat. Chem., 2012, 4, 873–886. [4] J. Singh, C. Lamberti and J. A. van Bokhoven, Chem. Soc. Rev., 2010, 39, 4754–4566. [5] M. B. J. Roeffaers, B. F. Sels, H. Uji-I, F. C. De Schryver, P.A. Jacobs, D. E. De Vos and J. Hofkens, Nature, 2006, 439, 572–575. [6] W. Xu, J. S. Kong, Y.-T. E. Yeh and P.Chen, Nat. Mater., 2008, 7, 992–996. [7] S. Kawata, Y. Inouye and P.Verma, Nat. Photonics, 2009, 3, 388–394.

95 Chapter 5: The Photo-Catalytic Reduction of p-Nitrothiophenol Monitored at the Nanoscale Using Tip-Enhanced Raman Spectroscopy

[8] R. M. Stöckle, Y. D. Suh, V. Deckert and R. Zenobi, Chem. Phys. Lett., 2000, 318, 131–136. [9] B. Pettinger, B. Ren, G. Picardi, R. Schuster and G. Ertl, Phys. Rev. Lett., 2004, 92, 096101. [10] E. Bailo and V.Deckert, Chem. Soc. Rev., 2008, 37, 921–930. [11] B.-S. Yeo, J. Stadler, T. Schmid, R. Zenobi and W. Zhang, Chem. Phys. Lett., 2009, 472, 1–13. [12] K. F. Domke and B. Pettinger, ChemPhysChem, 2010, 11, 1365–1373. [13] Z.-L. Yang, J. Aizpurua and H. Xu, J. Raman Spectrosc., 2009, 40, 1343–1348. [14] C. Fokas and V.Deckert, Appl. Spectrosc., 2002, 56, 192–199. [15] K. F. Domke and B. Pettinger, ChemPhysChem, 2009, 10, 1794–1798. [16] A. Rasmussen and V.Deckert, J. Raman Spectrosc., 2006, 37, 311–317. [17] C. Vericat, M. E. Vela, G. Benitez, P. Carro and R. C. Salvarezza, Chem. Soc. Rev., 2010, 39, 1805–1834. [18] A. Kudelski and B. Pettinger, Chem. Phys. Lett., 2004, 383, 76–79. [19] C. Neacsu, J. Dreyer, N. Behr and M. Raschke, Phys. Rev. B, 2006, 73, 193406. [20] R. L. Agapov, A. V.Malkovskiy, A. P.Sokolov and M. D. Foster, J. Phys. Chem. C, 2011, 115, 8900–8905. [21] S. Sun, R. L. Birke, J. R. Lombardi, K. P. Leung and A. Z. Genack, J. Phys. Chem., 1988, 92, 5965–5972. [22] K. Uetsuki, P.Verma, T. Yano, Y. Saito, T. Ichimura and S. Kawata, J. Phys. Chem. C, 2010, 114, 7515–7520. [23] J.-H. Yoon and S. Yoon, Phys. Chem. Chem. Phys., 2011, 13, 12900–12905.

96 CHAPTER 6

Summary and Outlook

6.1. Summary

A chemical reaction between molecules requires a certain amount of energy to occur. A catalyst can lower this activation energy, so less heat is required for the reaction, resulting in a greener and more sustainable process. Heterogeneous catalysts are typically made up of tiny metal nanoparticles, embedded in the larger framework of a porous support material. The surface of the metal nanoparticles is active catalytic material, where reactions take place. Ideally, one would like to follow a reaction over a single catalytic nanoparticle, to fully understand the dynamics involved in a reaction for development of even more efficient catalysts. Imaging at such a high resolution typically requires high-energy sources, and yields no molecular information. On the other hand, optical spectroscopy techniques are well-capable of following reactions occurring over catalysts, and they are especially suitable for monitoring the state of molecules during a reaction. Due to diffraction limitations, however, these techniques are limited in spatial resolution to hundreds of nanometers. Raman spectroscopy is one of these techniques, and is selective, highly informative and easily incorporated under catalytic working conditions. This makes it worthwhile to tackle the last challenges in Raman spectroscopy that hamper analysis on the scale of single nanoparticles - being the spatial resolution, the low scattering intensity and interference by fluorescence. In this PhD Thesis it is shown that the solution lies in the use of the nanophotonic properties of metal nanoparticles. At specific excitation wavelengths, electrons in metal nanoparticles can be excited to a higher energy level, called the surface plasmon resonance. The local electromagnetic field linked to this resonance can be extremely intense at the close contact between two nanoparticles, and when the resonances of the single particles combine, they form a gap-mode plasmon. This near-field is localized within a nano-scale volume and is used to create a nanometer-range spatial resolution for Raman spectroscopy. In gold or silver nanoparticles, this high intensity near-field also creates a large enhancement of the Raman scattering for molecules present within this field. This effect is called surface-enhanced Raman scattering (SERS), and can be used to overcome

97 Chapter 6: Summary and Outlook the low scattering intensity of normal Raman spectroscopy. Finally, the SERS-effect inherently limits the interference by fluorescence, tackling the final challenge for Raman spectroscopy as a nano-scale analytic tool for in-situ reaction monitoring. Here, SERS is optimized for the investigation of heterogeneous catalytic reactions on a nano-scale. As a model-reaction, the photo-catalytic reduction of p-nitrothiophenol (pNTP) is chosen. The reduction of nitro-aromatics is a class of well-studied reactions in literature, and Chapter 2 gives an overview of the various reduction pathways and relevant SERS-experiments that show why the reaction is worth studying. For application with SERS, pNTP is the most ideal reactant candidate. Its thiol-functionality ensures the self-assembly onto silver and gold surfaces, thereby positioning itself directly into the nano-scale measurement volume on SERS substrates. This ensures the detectability of the reagent and subsequent reaction. The photo-catalytic reduction of pNTP is easily triggered via the combination of green laser excitation and a silver SERS substrate, but is influenced by many parameters. This reaction serves as a model reaction in the rest of this PhD Thesis to probe and optimize the capabilities of SERS as reaction monitoring technique. The kinetics for this photo-catalytic reduction of pNTP are analysed in Chapter 3, via monitoring over a large number of catalytic active sites. The reduction is performed within a self-assembled monolayer and is thus confined in a two-dimensional space. The reaction rate is most likely limited by an uneven catalytic activity over the full SERS-enhancing area. The rate was found to resemble second-order reaction kinetics. Experiments with an effective dilution of the reactant via the assembly of a mixed monolayer with 1% pNTP and 99% thiophenol did not show any reaction under identical reaction conditions. This two-dimensional catalytic approach is shown to yield new interesting possibilities in the characterization of heterogeneous catalysts. In combination with the recent literature, discussed in Chapter 2, it also proves that p,p’-dimercaptoazobisbenzene is the reaction product in the photo-catalytic reduction of pNTP, settling a long-standing discussion in literature. Where large-scale SERS substrates show clear reaction kinetics, single-hotspot SERS measurements can show strong spectral and intensity-fluctuations due to the small ensemble of molecules in the measurement volume. These spectral fluctuations are very interesting in the study of heterogeneous catalytic reactions, as reaction intermediates are expected to have short life-spans and might only be observed in the most intense hotspots. Chapter 4 describes a novel chemometric method to separate short-term fluctuations from bulk reactivity in time-resolved spectra on the basis of a time-dependent one-component principal component analysis. With this method, the reaction catalysed

98 Outlook by a single silver particle on a monolayer of pNTP adsorbed on Au was analysed in time via two-component multivariate curve resolution, while a set of spectral deviations were set aside and analysed separately for their spectral features with multi-component multivariate curve resolution. The fit of reaction kinetics is thereby improved while also allowing investigation of components with limited lifetime. An imaging variation to SERS comes in the form of tip-enhanced Raman scattering (TERS), where a gold- or silver-coated needle of an atomic force microscope is used to create the SERS-enhancement effect. Only when and where the needle is in contact with the sample, will an enhanced Raman signal be obtained from the sample. The incorporation in an atomic force microscope allows local analysis with nanometer precision. Chapter 5 shows how this technique is used to study the reduction of pNTP over a single catalytic particle: the TERS-tip itself. A dual wavelength approach was found to be crucial to follow reaction dynamics, with 633 nm laser excitation for monitoring purposes, and 532 nm laser excitation to induce the reduction. Using the chemometric method described in Chapter 4, the data can be separated into reaction kinetics and short-term fluctuations, that could belong to reaction intermediates. This is the first description of a catalytic reaction over a single catalytic particle, as observed by TERS. Herewith, TERS has extended the possibilities to have both the time-resolution and sufficient enhancement effect to follow a chemical reaction over a single catalytic particle.

6.2. Outlook

With the use of SERS and TERS, Raman spectroscopy can now be scaled down to monitor few molecules or a single catalytic particle. The further application of these techniques to heterogeneous catalysis, beyond the studied model reaction, might be achieved in a several ways. A first approach makes use of the high sensitivity of SERS and TERS to identify intermediates in reactions. This will require the measurement of many spectra at short integration times, to capture as many single-event fluctuations as possible. This will be most effective with single-hotspot samples. Chemometric analysis will indicate the most common spectral features. These can subsequently be interpreted with the help of e.g. density functional theory calculations. The SERS-activity is most prominent in Ag, Au and Cu nanoparticles, any reaction catalysed over these particles can therefore be measured directly with SERS. For reactions with other catalysts, the method can still be applied efficiently by application of the catalytic material directly onto SERS-active substrates, given that the gap-mode plasmon is still accessible. The SERS material can even be made inert via a thin layer (< 3 nm)

99 Chapter 6: Summary and Outlook of e.g. silica. This coating might even increase the temperature-resistance of the SERS substrates and chemical inertness. The high sensitivity of SERS can also be used in a different approach. SERS substrates can be used to detect low concentrations of molecules in a gasflow. A sophisticated setup would involve a series of SERS-substrates each functionalized differently to specifically bind or attract a specific group of molecules. Setting up these SERS-cells in-line, after a reactor, can thereby be used to identify reaction products. The same possibilities exist for TERS. TERS acts on a smaller volume, creating even more localized analysis. Catalytic conversion over a single particle can be identified by comparing reactant and product ratios, whereas short-term fluctuations can pinpoint even more clearly reaction intermediates. It will be crucial to ensure that the TERS tip is inert. As with SERS, a thin (silica) coating will be required to limit side-reactions occurring over the TERS tip. With a distribution of catalytic particles over a surface, e.g. in a reaction cell, the TERS tip can be used to locate the catalytic particles, and subsequently monitor in-situ the reaction over the selected particle. Finally, besides near-field Raman several other varieties of near-field microscopy are being developed, e.g. near-field infrared (IR) microscopy. It has, like Raman, high molecular sensitivity and because it combines IR and atomic force microscopy to yield a nano-scale spatial resolution. However, it is still hampered by the need for brilliant IR light sources, with broadband-tunable excitation wavelengths. These are not readily available yet. Nonetheless, SERS and TERS, as well as the near-field techniques described above, open up a wide field of options for nano-scale studies of catalytic reactions. They offer chemical information with the required resolution for true single catalytic particle analysis.

100 CHAPTER 7

Samenvatting en Vooruitblik

7.1. Samenvatting

Het laten plaatsvinden van een chemische reactie tussen moleculen vereist een be- paalde hoeveelheid energie. Een katalysator kan deze activatie-energie verlagen, zodat er minder warmte nodig is bij het proces, met als resultaat een groener en duurzamer proces. Heterogene katalysatoren bestaan normaliter uit kleine metaal nanodeeltjes, in- gebed in het grotere raamwerk van een poreus dragermateriaal. Het oppervlak van de metaal nanodeeltjes is het katalytisch actief materiaal waar de reacties plaatsvinden. Ide- aliter zou men een reactie willen volgen over één enkel katalytisch nanodeeltje, om de dynamiek in de reactie volledig te begrijpen en om daarmee efficiëntere katalysatoren te ontwikkelen. Het in beeld brengen van deze materialen met zulk een hoge resolutie vereist wel lichtbronnen die hoogenergetisch licht uitzenden, maar geeft geen moleculaire informatie. Daarentegen zijn optisch spectroscopische technieken zeer goed in staat om reacties over katalysatoren te volgen, en dan met name om de verschillende moleculaire toestanden tijdens de reactie te meten. Vanwege het diffractie-limiet van licht zijn deze technieken wel gelimiteerd in hun ruimtelijke resolutie tot een paar honderd nanometer. Raman spectroscopie is één van deze technieken met een hoge moleculaire gevoeligheid, en is ook gemakkelijk te integreren onder katalytische omstandigheden. Deze twee voordelen maken het zeer interessant om de laatste paar uitdagingen aan te pakken voor Raman spectroscopie als techniek voor enkele katalytische nanodeeltjes - namelijk de ruimtelijke resolutie, de lage verstrooiingsintensiteit en interferentie vanwege fluorescentie processen. In dit proefschrift wordt aangetoond dat de oplossing gevonden kan worden in de na- nofotonische eigenschappen van metaal nanodeeltjes. Bij specifieke excitatiegolflengtes worden de elektronen in metaal nanodeeltjes geëxciteerd naar een hoger energieniveau, de oppervlakte-plasmon resonantie. Het lokale elektromagnetisch veld dat hierbij hoort kan uitermate intens zijn wanneer twee nanodeeltjes elkaar nog net niet raken, wanneer de resonanties van de enkele nanodeeltjes koppelen tot een overbruggende resonantie. Het nabij-veld wordt hiermee gelokaliseerd op de nanoschaal en kan daarom gebruikt

101 Hoofdstuk 7: Samenvatting en Vooruitblik worden voor een zeer hoge ruimtelijke resolutie voor Raman spectroscopie, op de nanoschaal. Met name bij nanodeeltjes van goud of zilver veroorzaakt dit intense nabij-veld ook een hoge versterking van de Raman verstrooiing van de moleculen die aanwezig zijn in dit nabij-veld. Dit effect wordt daarom oppervlakte-versterkt Raman verstrooiing (SERS) genoemd, en kan gebruikt worden om de lage Raman verstrooiingsintensiteit te overwinnen. Tenslotte zorgt het SERS-effect gelijktijdig met de Raman versterking ook voor minder ﬂuorescentie, waardoor de laatste uitdaging voor Raman spectroscopie op de nanometerschaal opgelost is voor de toepassing als analytische toepassing om reacties in-situ te volgen. SERS is geoptimaliseerd voor de studie van heterogene katalytische reacties op een nanometerschaal. Als model reactie is de fotokatalytische reductie van p-nitrothiophenol (pNTP) gekozen. De reductie van nitro-aromaten is een klasse van reacties die uitgebreid bestudeerd is in de literatuur. Een overzicht van de verschillende reductie-mechanismen en de relevante SERS-experimenten wordt gegeven in Hoofdstuk 2. Voor de toepassing met SERS is pNTP de meest ideale kandidaat als reagens, omdat de thiol-functionaliteit ervoor zorgt dat de moleculen zichzelf vastzetten en ordenen op zilver en goud opper- vlakken. Ze positioneren zichzelf daarmee direct in het meetvolume op SERS substraten. De foto-katalytische reductie van pNTP is gemakkelijk te veroorzaken bij de combinatie van excitatie met een groene laser en een zilver SERS substraat. In dit proefschrift wordt deze fotokatalytische reductie van pNTP gebruikt als modelreactie om de mogelijkheden van SERS zowel uit te proberen als te optimaliseren om reacties te volgen. De reactiekinetiek voor deze fotokatalytische reductie van pNTP wordt in Hoofdstuk 3 bestudeerd door te meten over een groot aantal katalytisch actieve gebieden. De reductie wordt uitgevoerd binnen een zelf-georganiseerde monolaag, en is dus beperkt tot een tweedimensionale ruimte. Er wordt aangetoond dat de reactiesnelheid voorname- lijk beschreven wordt door de voorwaartse reactie, maar een volledige omzetting van de monolaag wordt hoogstwaarschijnlijk verhinderd door een ongelijke katalytische activiteit over het volledige SERS-actieve oppervlak. De reactiesnelheid blijkt het best te beschrijven met tweede-orde reactiekinetiek. In een tweede experiment is effectief een verdunning van het reagens bereikt door het laten formeren van een gemengde monolaag met daarin 1% pNTP en 99% thiophenol. Deze monolaag vertoonde geen reactie bij blootstelling aan identieke reactieomstandigheden. Hiermee is aangetoond dat deze tweedimensionale SERS aanpak interessante mogelijkheden geeft voor de karakterisatie van heterogene katalysatoren. In combinatie met de recente literatuur, zoals behandeld in Hoofdstuk 2, bewijzen deze experimenten ook dat p,p’-dimercaptoazobisbenzeen het

102 Samenvatting reactieproduct is in de fotokatalytische reductie van pNTP,onderwerp van een langdurig debat in de literatuur. Waar SERS substraten met veel hotspots duidelijke reactiekinetiek laten zien, kunnen metingen aan enkele hotspots grote fluctuaties in intensiteit en spectraal karakter vertonen, vanwege het kleine aantal moleculen in het meetvolume. Deze spectrale fluctuaties zijn zeer interessant in het onderzoek naar heterogeen katalytische reacties, omdat reactie-intermediairen maar een korte levensduur hebben en vermoedelijk alleen in deze fluctuaties zichtbaar zijn. Hoofdstuk 4 beschrijft een nieuwe chemometrische methode waar korte-termijn fluctuaties gescheiden kunnen worden van de bulk reactiviteit in tijdsafhankelijke spectra. Deze methode is gebaseerd op de tijdsafhankelijke fit-kwaliteit van een principale componenten analyse met één component. De bulk reactie van één enkel zilver katalytisch deeltje op een monolaag van pNTP op goud wordt op deze wijze geanalyseerd via een twee-componenten multivariate lijn resolutie. Daarnaast worden de spectrale afwijkingen apart geanalyseerd door middel van meervoudig-componenten multivariate lijn resolutie. Een flexibele variatie op SERS komt in de vorm van tip-versterkte Raman verstrooiing (TERS), waar een goud- of zilver-gecoate naald van een atoomkrachtmicroscoop gebruikt wordt om het SERS-versterkingseffect op te wekken. Alleen wanneer en precies daar waar de naald in contact is met het monster, zal er een versterkt Raman signaal gemeten worden van het monster. De combinatie met de atoomkrachtmicroscoop geeft de mogelijkheid om met nanometer-precisie de versterkingslocatie te kiezen. Hoofdstuk 5 laat zien hoe deze techniek gebruikt kan worden om de reductie van pNTP te bestuderen over één enkel katalytisch deeltje: de TERS-naald zelf. Een aanpak met twee excitatiegolflengtes bleek cruciaal om de reactie-dynamiek te kunnen volgen. Hierbij is een 633 nm laser excitatie gebruikt om de processen op het oppervlak te volgen, terwijl een 532 nm laser excitatie de reactie induceert. Met behulp van de chemometrische methode uit Hoofdstuk 4 kunnen de spectra gescheiden worden tot reactiekinetiek en korte-termijn fluctuaties met mogelijk reactie-intermediairen. Dit is de eerste beschrijving van een katalytische reactie over één enkel katalysator deeltje, zoals geobserveerd met TERS. Samenvattend is SERS geschikt om zowel reacties te volgen als de reactiekinetiek te kwantificeren. Karakteristieke spectrale en intensiteitsfluctuaties bemoeilijken normaliter het volgen van een reactie, maar kunnen nu met behulp van een nieuwe chemometrische methode gescheiden worden van bulk veranderingen en apart geanalyseerd worden, om bijvoorbeeld reactie-intermediairen te identificeren. Tenslotte heeft TERS de mogelijkheden uitgebreid om met de tijdsresolutie en een voldoende versterkingseffect een chemische reactie te volgen over één enkel katalysatordeeltje.

103 Chapter 7: Samenvatting en Vooruitblik

7.2. Vooruitblik

Met het gebruik van SERS en TERS zijn noch de resolutie van Raman spectroscopie noch de interpretatie van de bijbehorende spectra en fluctuaties een beperking tegenwoor- dig. Raman spectroscopie kan nu verkleind worden om enkele moleculen of één enkel katalysator deeltje te bestuderen. Buiten de bestudeerde modelreactie, kan de praktische toepassing van deze technieken in heterogene katalyse op verscheidene manieren bereikt worden. SERS en TERS zijn vanwege hun hoge gevoeligheid ideale kandidaten om reactie- intermediairen te identificeren. Omdat de SERS-activiteit het sterkst aanwezig is in Ag, Au en Cu nanodeeltjes, kunnen de reacties die door deze materialen gekatalyseerd worden, direct bestudeerd worden met SERS. Maar ook voor reacties met andere katalysatoren kan de methode gemakkelijk toegepast worden door het katalysator materiaal direct op SERS-actieve substraten aan te brengen, zolang de overbruggende plasmon nog steeds toegankelijk is. Het SERS materiaal kan zelf inert gemaakt worden door het aanbrengen van een dunne (< 3 nm) laag van bijvoorbeeld silica. Deze laag zou zelfs de hittebesten- digheid van de SERS substraten en chemische inertheid kunnen verbeteren. De hoge gevoeligheid van SERS kan ook gebruikt worden met een andere aanpak. SERS substraten kunnen namelijk gebruikt worden om lage hoeveelheden moleculen in een gasstroom te meten. Een geavanceerde opstelling zou een aantal SERS-substraten bevatten, die elk gefunctionaliseerd zijn om een specifieke groep moleculen te binden. Door deze SERS-substraten in serie achter een reactor te plaatsen, kunnen bijvoorbeeld reactieproducten geïdentificeerd worden. TERS geeft de gebruiker meer flexibiliteit ten opzichte van de locatie van het versterkingseffect. Met een verdeling aan katalysator deeltjes op een oppervlak, bijvoorbeeld in een reactie-cel, kan de TERS tip gebruikt worden om de deeltjes te lokaliseren en ver- volgens in-situ de reactie over het gekozen katalysator deeltje te volgen. De katalytische omzetting over één enkel deeltje kan bepaald worden door de verhouding van reactant en product te bepalen, terwijl korte-termijn fluctuaties intermediairen kunnen aanwijzen. Het zal belangrijk zijn om te zorgen dat de TERS-naald inert is. Net als bij SERS kan een dunne (silica) laag mogelijke nevenreacties op de TERS-naald beperken. Naast TERS, of nabij-veld Raman, zijn er nog een aantal variaties op nabij-veld mi- croscopie ontwikkeld, met name nabij-veld fluorescentie- en nabij-veld infraroodmicroscopie. Vooral nabij-veld infraroodmicroscopie zou zeer interessant kunnen zijn om toe te passen in heterogene katalyse, omdat het, net als Raman, een hoge moleculaire gevoeligheid heeft. Helaas wordt deze toepassing momenteel nog gehinderd door de noodza- kelijke sterke infrarood bronnen, met verstelbare breedband-excitatiegolflengtes. Deze

104 Vooruitblik zijn nog niet vrij verkrijgbaar. Daarentegen brengen SERS en TERS, met de hierboven beschreven nabij-veld technieken, een breed scala aan mogelijkheden voor de bestudering van katalytische reacties op de nanometer schaal. Ze bieden chemische informatie aan met de benodigde ruimtelijke resolutie voor werkelijke analyse op de schaal van enkele katalysatordeeltjes.

105

APPENDIX A

Vibrational Assignments

TABLE A.1. Neat Raman vibrations of p-nitrothiophenol at 633 nm laser excitation. As taken from 1 Skadtchenko et al. vs = very strong, s = strong, m = medium, w = weak, vw = very weak, sh = shoulder.

Raman wavenumbers Assignment 1 (cm− ) 3099 vw C-H stretch. 3072 w C-H stretch. 3054 vw C-H stretch. 2550 w, 2589 vw S-H stretch. 1595 vw C=C stretch. 1577 s C=C stretch. 1505 vw NO2 antisym. 1477 w Ring stretch. 1422 w Ring stretch. 1389 w Ring stretch. 1362 w C-H bend.

1335 vs NO2 sym. 1181 m C-H bend. 1100 s C-H bend. 1079 m C-H bend. 1062 vw C-H bend. 1010 vw Ring breath. 965 vw C-H wag. 931 vw C-H wag. 856 m C-H wag. 840 vw C-N stretch. 820 sh C-H wag.

107 Appendix A: Vibrational Assignments

740 vw C-S stretch. 681 vw C-H wag. 627 w Ring deform. 552 w C-N bend. 531 vw Ring deform. 470 vw C-H wag. 406 vw Ring deform. 317 vw Ring deform. 294 vw C-C-S deform. 273 w Ring wag.

161 vw NO2,C-S rock. 194 vw S-H wag.

TABLE A.2. Neat Raman vibrations of p-aminothiophenol at 514.5 nm laser excitation. As taken 2 from Osawa et al. vs = very strong, s = strong, m = medium, w = weak, vw = very weak, sh = shoulder.

Raman wavenumbers Assignment 1 (cm− ) 3060 w C-H stretch. 2565 w S-H stretch. 1620 vw NH bend. 1595 s C-C stretch. 1490 w C-C stretch., C-H bend. 1480 w 1403 w 1310 w C-C stretch., C-H bend. 1266 w 1206 m 1173 m C-H bend. 1118 w C-H bend. 1089 vs C-S stretch. 1011 w C-C bend., C-C-C bend. 960 vw C-H wag. 820 w C-H wag.

108 References

799 m C-H stretch., C-S stretch., C-C stretch. 699 vw C-H wag., C-S wag., C-C wag. 634 m C-C-C bend. 521 w C-C-C bend. 463 m C-C-C bend. 396 w C-C torsion 321 w C-H bend., C-S bend. 256 w C-N bend., C-S bend. 196 m C-N wag., C-S wag., C-C torsion 154 w

TABLE A.3. Neat Raman vibrations of p,p’-dimercaptoazobisbenzene at 633 nm laser excitation. As taken from Huang et al. 3

Raman wavenumbers Assignment 1 (cm− ) 1588 C-C stretch. 1481 C-H out-of-plane bend., N-N stretch., C-C stretch. 1456 N-N stretch., C-C stretch., C-H out-of-plane bend. 1395 N-N stretch., C-C stretch., C-H out-of-plane bend. 1306 C-C stretch. 1183 C-N stretch., C-H out-of-plane bend., C-C stretch. 1147 C-H out-of-plane bend., C-N stretch. 1064 C-C stretch., C-S stretch. 1002 C-C-C ring bend., C-C stretch.

References

[1] B. O. Skadtchenko and R. F. Aroca, Spectrochim. Acta A, 2001, 57, 1009–1016. [2] M. Osawa, N. Matsuda, K. Yoshii and I. Uchida, J. Phys. Chem., 1994, 98, 12702– 12707. [3] Y.-F. Huang, H.-P.Zhu, G.-K. Liu, D.-Y. Wu, B. Ren and Z.-Q. Tian, J. Am. Chem. Soc., 2010, 132, 9244–9246.

109

APPENDIX B

In-Situ Tip-Enhanced Raman Scattering Monitoring of the Photo-Reduction of p-Nitrothiophenol CH CH NO2 NO2 NN NN NN NN ν β ν ν β ν ν a) ν b) * *

0.8 0.8 40 100 0.6 0.6 80 Time (s) 200 0.4 Time (s) 0.4 120 0.2 0.2 300

160 1600 1400 1200 1000 1600 1400 1200 1000 Raman shift (cm-1) Raman shift (cm-1)

FIG. B.1. Time-dependent tip-enhanced Raman scattering-spectra at 633 nm excitation before and after illumination with a 532 nm laser are shown in a) at 392 µW 633 nm excitation, with a 10 s, 66 µW 532 nm excitation, 2 s integration time for each spectrum, and in b) with 380 µW 633 nm excitation and two 30 s 70 µW 532 nm excitations, 5 s integration time for each recorded 1 spectrum. The asterisk in a and b shows the location of the 950 cm− band of the SiO2 signal of the glass substrate.

111

List of Abbreviations and Symbols

A reactant AFM atomic force microscopy ATR-IR attenuated total reﬂectance infrared spectroscopy B product CARS coherent anti-Stokes Raman scattering DMAB p,p’-dimercaptoazobisbenzene DMHAB p,p’-dimercaptohydrazobisbenzene DORS diagonally offset Raman scattering DFT density functional theory DRIFTS diffuse reﬂectance infrared fourier transform spectroscopy IR infrared ISC inter-system crossing k reaction constant λ wavelength of light LUMO lowest unoccupied molecular orbital MCR multivariate curve resolution

ν0, ν1, ν2 vibrational states NA numerical aperture

Ncat fraction of surface sites with photo-catalytic activity

Nncat fraction of surface sites not photo-catalytically active

NT total number of surface sites per measured surface area PALM photo-activated localization microscopy pATP p-aminothiophenol PCA principle component analysis pNTP p-nitrothiophenol r reaction rate RAIRS reﬂective absorption infrared spectroscopy

σA Raman cross section of A

113 List of Abbreviations and Symbols

σB Raman cross section of B

S0 electronic ground state

S1 ﬁrst excited singlet state SCE standard calomel electrode SEIRA surface-enhanced infrared absorption SERS surface-enhanced Raman scattering SERRS surface-enhanced resonance Raman scattering SPM scanning probe microscopy STED stimulated emission depletion microscopy STM scanning tunnelling microscopy STORM stochastical optical reconstruction microscopy

θA fraction of surface sites occupied by A

θA,0 fraction of surface sites occupied by A at the start of the reaction

θB fraction of surface sites occupied by B t time

T1 (ﬁrst) excited triplet state TERS tip-enhanced Raman scattering UV-VIS ultraviolet-visible absorption spectroscopy UV ultraviolet VIS visible WLS weighted least squares XPS x-ray photo-electron spectroscopy

114 List of Publications and Presentations

Publications

E.M. van Schrojenstein Lantman, A.J.G. Mank, B.M. Weckhuysen, in preparation. (Chapter 2)

E.M. van Schrojenstein Lantman, O.L.J. Gijzeman, A.J.G. Mank, B.M. Weckhuysen, Chem- CatChem, DOI 10.1002/cctc.201402647. (Chapter 3)

E.M. van Schrojenstein Lantman, P.de Peinder, A.J.G. Mank, B.M. Weckhuysen, submitted. (Chapter 4)

E.M. van Schrojenstein Lantman, T. Deckert-Gaudig, A.J.G. Mank, V.Deckert, B.M. Weck- huysen, Nature Nanotechnology 2012, 7, 773-781. (Chapter 5)

A.M. Beale, J.P.Hofmann, M. Sankar, E.M. van Schrojenstein Lantman, B.M. Weckhuysen, Recent trends in Operando and in-situ characterization: Techniques of Rational Design of Catalysts in Heterogeneous Catalysts for Clean Technology: Spectroscopy, Design, and Mon- itoring 2013 (Eds. K. Wilson, A.F. Lee), Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, Germany.

C.E. Harvey, E.M. van Schrojenstein Lantman, A.J.G. Mank, B.M. Weckhuysen, Chem. Com- mun. 2012, 48, 1742-1744.

M. Diskus, O. Nilsen, H. Fjellvåg, S. Diplas, P. Beato, C. Harvey, E. van Schrojenstein Lantman, B.M. Weckhuysen, J. Vac. Sci. Technol. A 2012, 30, 01A107.

Oral Presentations

Evelien van Schrojenstein Lantman, Tanja Deckert-Gaudig, Arjan Mank, Volker Deckert, Bert Weckhuysen, Tip-enhanced Raman spectroscopy for nanoscale monitoring of catalytic processes, Frontures of NanoScopy, Delft, The Netherlands, October 28, 2013.

115 List of Publications and Presentations

Evelien van Schrojenstein Lantman, Onno Gijzeman, Arjan Mank, Bert Weckhuysen, Ob- servation of reaction kinetics in self-assembled monolayers: surface- and tip-enhanced Raman scattering, SciX2013, Milwaukee (MI) USA, October 2013.

Evelien van Schrojenstein Lantman, Tanja Deckert-Gaudig, Arjan Mank, Volker Deckert, Bert Weckhuysen, Tip-enhanced Raman spectroscopy:a novel approach to study a catalytic reaction while it occurs, IRDG, Océ, Arnhem, The Netherlands, May 2012.

Evelien van Schrojenstein Lantman, Tanja Deckert-Gaudig, Arjan Mank, Volker Deckert, Bert Weckhuysen, Catalytic processes monitored at the nanoscale with tip-enhanced Raman spectroscopy, NWO analytische scheikunde, Lunteren, The Netherlands, November 2012.

Evelien van Schrojenstein Lantman, Wat kun je doen met nanoscience?, Beta Excellent, Utrecht, The Netherlands, April 23, 2012.

Evelien van Schrojenstein Lantman, Wat kun je doen met nanoscience?, Beta Excellent, Utrecht, The Netherlands, April 17, 2012.

Evelien van Schrojenstein Lantman, Wat kun je doen met nanoscience?, Eureka!Cup, Utrecht, The Netherlands, February 1, 2012.

Evelien van Schrojenstein Lantman, Tanja Deckert-Gaudig, Arjan Mank, Volker Deckert, Bert Weckhuysen, Tip-Enhanced Raman Spectroscopy; a novel approach to study a catalytic reaction in-situ, IRDG, Venlo, The Netherlands, May 10, 2012.

Evelien van Schrojenstein Lantman, Tanja Deckert-Gaudig, Arjan Mank, Volker Deckert, Bert Weckhuysen, Using Tip Enhanced Raman Spectroscopy to monitor a heterogeneous catalytic reaction in-situ, company visit to Dow Chemicals, Terneuzen, The Netherlands, August 2011.

116 Poster Presentations

Evelien van Schrojenstein Lantman, Combining AFM and Raman for the characterization of heterogeneous catalysis, NRSC-Catalysis workshop, Utrecht, The Netherlands, January 2011.

Clare Harvey, Evelien van Schrojenstein Lantman, Arjan Mank, Bert Weckhuysen, An integrated AFM-Raman study of surface enhanced Raman spectroscopy (SERS) over supported silver nanoparticle catalysts, MICROSCIENCE2010, London, UK, May/June 2010.

Poster Presentations

Evelien van Schrojenstein Lantman, Tanja Deckert-Gaudig, Arjan Mank, Volker Deckert, Bert Weckhuysen, Catalytic processes monitored at the nanoscale with tip-enhanced Raman spectroscopy, NCCC XIV,Noordwijkerhout, The Netherlands, March 2013.

Evelien van Schrojenstein Lantman, Tanja Deckert-Gaudig, Arjan Mank, Volker Deckert, Bert Weckhuysen, Catalytic processes monitored at the nanoscale with tip-enhanced Raman spectroscopy, MicroNanoConference, Ede, The Netherlands, December 2012.

Evelien van Schrojenstein Lantman, Tanja Deckert-Gaudig, Arjan Mank, Volker Deckert, Bert Weckhuysen, Using Tip Enhanced Raman Spectroscopy to monitor a catalytic reaction in-situ, 4th International Congress on Operando Spectroscopy, Brookhaven (NY), USA, April-May 2012.

Evelien van Schrojenstein Lantman, Tanja Deckert-Gaudig, Arjan Mank, Volker Deckert, Bert Weckhuysen, Using Tip Enhanced Raman Spectroscopy to monitor a catalytic reaction in-situ, Summer School on Plasmonics 2, Hyères, France, October 2011.

Evelien van Schrojenstein Lantman, Tanja Deckert-Gaudig, Arjan Mank, Volker Deckert, Bert Weckhuysen, Using Tip Enhanced Raman Spectroscopy to monitor a heterogeneous catalytic reaction in-situ, NCCC XII, Noordwijkerhout, The Netherlands, March 2011.

117 List of Publications and Presentations

Evelien van Schrojenstein Lantman, Tanja Deckert-Gaudig, Arjan Mank, Volker Deckert, Bert Weckhuysen, Using Tip Enhanced Raman Spectroscopy to monitor a heterogeneous catalytic reaction in-situ, The Analytical Challenge 2010, Lunteren, The Netherlands, Novem- ber 2010.

Evelien van Schrojenstein Lantman, Tanja Deckert-Gaudig, Arjan Mank, Volker Deckert, Bert Weckhuysen, Using Tip Enhanced Raman Spectroscopy to monitor a heterogeneous catalytic reaction in-situ, MicroNanoConference 2010, Enschede, The Netherlands, November 2010.

118 Acknowledgements

Dit proefschrift en het onderzoek dat eraan vooraf ging, zouden niet tot stand gekomen zijn zonder de bijdragen van anderen. Ten eerste Bert, bedankt voor de steun en begeleiding, maar ook de vrijheid om mijn eigen weg te zoeken in het onderzoek. Jij hebt mij elke keer opnieuw uitgedaagd om een extra stap te zetten. Eén van jouw ideeën, TERS voor katalyse, heeft uiteindelijk het volledige thema van mijn proefschrift bepaald, en daarbij een mooi artikel in Nature Nanotechnology opgeleverd. Ook bij de rest van de projecten heb jij mij scherp gehouden en ervoor gewaakt dat deze interessant genoeg bleven voor katalytisch onderzoek. Heel erg bedankt voor de afgelopen 5 jaar. Arjan, als ‘dagelijkse’ begeleider, hebben wij vele donderdagen en vrijdagen doorge- bracht in Utrecht of Eindhoven. Maar ook daarbuiten kon ik je altijd snel bereiken om advies te vragen, onderzoek-gerelateerd of breder. Ik wil je hier graag voor bedanken. Verder waren er steeds weer nieuwe ideeën voor geniale experimenten en ‘snelle’ papers. Jammer dat er maar zo weinig in een promotie-onderzoek passen. I would also like to thank Volker Deckert and Tanja Deckert-Gaudig for the collaboration and access to TERS setups. Volker, thank you for the mentoring you gave me regarding TERS and related topics, but also for inviting me for a presentation at the SciX conference in Kansas City, 2012. I greatly enjoyed our collaboration. Tanja, we have spent many hours in the dark and quiet of the TERS labs, working as long as the TERS tip would be active, or as long as we would be awake enough. I hope the magic sample keeps doing its magic and helps you with many projects to come. I would also like to thank you, Volker, and the other members of the lab in Jena for the warm welcome I received in Jena at each visit. Onno, het uiteindelijke kinetiek-probleem bleek wat minder ingewikkeld dan ver- wacht, maar nog steeds interessant. Bedankt voor het stellen van de goede en lastige vragen, en voor jouw hulp bij het aﬂeiden van de reactiekinetiek. Peter, bedankt voor jouw bijdrage aan dit proefschrift. Chemometrie is een lastig onderwerp, en nog lastiger wanneer het gecombineerd wordt met SERS. Samen zijn we

119 Acknowledgements tot een werkende methode gekomen, en ik heb gaandeweg veel geleerd. Ik hoop dat onze nieuwe methode ook voor jou in de toekomst nog toepasbaar is op andere experimenten. Dear Clare, partner in science. Together we started the AFM-Raman lab, had many tea-breaks, made up cunning plans both with and without the supervision of Arjan, took early trains to Eindhoven, and went to several conferences. I greatly enjoyed working together with you. Thanks for the great time! It was fun working in the AFM-Raman lab, and Jan Philipp, Rogier, and Zafer all contributed to that. I hope we keep in contact, and I wish you all the best in you future careers. Inge, na een half jaar master-onderzoek is het je uiteindelijk toch gelukt om gecoate SERS deeltjes te maken. Dit ging natuurlijk niet zonder horten en stoten, maar je mag trots zijn op wat je hebt bereikt. Apparaten en experimenten werken nooit zomaar. Fouad, Ad, Ad, Vincent, maar ook Marjan en Rien, bedankt voor jullie ondersteuning en hulp. Uiteraard kan ik Dymph en Monique niet vergeten, bedankt voor jullie hulp bij allerhande onderwerpen. The group of Inorganic Chemistry and Catalysis has grown rather large, and I do not dare try to name you all, as I might forget some... However, I do thank everyone of you for the great time. Coffee-breaks, lunch-time, barbecues, or just short chats by the coffee machine, I value them all. Thank you! Buiten werktijd heb ik de steun van vrienden gehad, voor aﬂeiding van het onderzoek, of juist door ervaringen uit te wisselen. Nynke, Johannes, Mathilde, Bram, Hanneke, Hans, Simone, Ruben en Anne Maaike, bedankt voor de vele gezellige etentjes en alle andere dingen die we gedaan hebben. Marnix, Irene en Hugo, ik ben trots op wat jullie bereikt hebben, en wat jullie vast en zeker nog in de toekomst gaan doen. Marnix en Irene, en wie weet ook Hugo, heel veel succes en plezier met jullie eigen promotie. Mama, papa en opa Cossee, bedankt voor jullie onvoorwaardelijke interesse, steun en alle gevraagde en ongevraagde adviezen. Lieve Martijn, met jou is alles leuker.

Evelien

120 Curriculum Vitae

Evelien van Schrojenstein Lantman was born on October 25th 1986 in Rijswijk, The Netherlands. She graduated from secondary school at the Nieuwe Veste, Coevorden, in 2004, and continued her education at Utrecht University. She obtained a Bachelor of Science degree in chemistry in 2007, and a Master in Science degree (cum laude) in 2009. This included a master thesis "Synthesis of Anisotropic CdSe/CdTe Nanoparticles", under the supervision of dr. Celso de Mello Donega and prof. dr. Daniël Vanmaekelbergh, and a second research thesis, "Waveguiding in CdS and ZnSe Nanowires" under the supervision of dr. Bert van Vugt and prof. dr. Ritesh Agarwal (University of Pennsylvania). In November 2009, Evelien started her PhD degree in the group of Inorganic Chem- istry and Catalysis, under the supervision of prof. dr. ir. Bert Weckhuysen and in collaboration with dr. Arjan Mank (Philips Innovation Services). The results of this work are described in this PhD thesis and were published in peer-reviewed journals and presented at national and international conferences. Within the group of Inorganic Chemistry and Catalysis, she was PhD and postdoc representative for a year in the group management team. From 2010 to 2012, she was also involved as treasurer to the PhD Platform of the nationwide graduate school Netherlands Institute for Catalysis Research (NIOK). Evelien twice taught the ﬁrst-year chemistry students’ lab introduction project, supervised a master student (Inge van Elteren, Forensic Sciences), and gave various lab demonstrations on the topics of AFM and SERS. Evelien contributed to the science outreach programmes of Eureka Cup and Beta Excellent in 2013. She attended the Summerschool in Plasmonics 2, Hyères, France, and a Risk analysis and technology assessment by NanoNextNL, Amers- foort, The Netherlands, and was invited to attend the Lindau Nobel Laureate Meeting in 2013.

121